{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 8.651720871487212,
  "eval_steps": 200,
  "global_step": 27400,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00031575623618566466,
      "grad_norm": 0.08905501663684845,
      "learning_rate": 3.156565656565657e-08,
      "loss": 1.1008,
      "step": 1
    },
    {
      "epoch": 0.0006315124723713293,
      "grad_norm": 0.05562233179807663,
      "learning_rate": 6.313131313131314e-08,
      "loss": 1.0995,
      "step": 2
    },
    {
      "epoch": 0.000947268708556994,
      "grad_norm": 0.06451323628425598,
      "learning_rate": 9.46969696969697e-08,
      "loss": 1.0977,
      "step": 3
    },
    {
      "epoch": 0.0012630249447426586,
      "grad_norm": 0.052576255053281784,
      "learning_rate": 1.262626262626263e-07,
      "loss": 1.0996,
      "step": 4
    },
    {
      "epoch": 0.0015787811809283233,
      "grad_norm": 0.0518963597714901,
      "learning_rate": 1.5782828282828283e-07,
      "loss": 1.0986,
      "step": 5
    },
    {
      "epoch": 0.001894537417113988,
      "grad_norm": 0.05337827652692795,
      "learning_rate": 1.893939393939394e-07,
      "loss": 1.0995,
      "step": 6
    },
    {
      "epoch": 0.0022102936532996525,
      "grad_norm": 0.09007469564676285,
      "learning_rate": 2.2095959595959595e-07,
      "loss": 1.0995,
      "step": 7
    },
    {
      "epoch": 0.0025260498894853173,
      "grad_norm": 0.09224730730056763,
      "learning_rate": 2.525252525252526e-07,
      "loss": 1.099,
      "step": 8
    },
    {
      "epoch": 0.002841806125670982,
      "grad_norm": 0.0743400901556015,
      "learning_rate": 2.840909090909091e-07,
      "loss": 1.0993,
      "step": 9
    },
    {
      "epoch": 0.0031575623618566467,
      "grad_norm": 0.04727882519364357,
      "learning_rate": 3.1565656565656567e-07,
      "loss": 1.0989,
      "step": 10
    },
    {
      "epoch": 0.0034733185980423114,
      "grad_norm": 0.06268243491649628,
      "learning_rate": 3.4722222222222224e-07,
      "loss": 1.0986,
      "step": 11
    },
    {
      "epoch": 0.003789074834227976,
      "grad_norm": 0.051946427673101425,
      "learning_rate": 3.787878787878788e-07,
      "loss": 1.098,
      "step": 12
    },
    {
      "epoch": 0.004104831070413641,
      "grad_norm": 0.044534604996442795,
      "learning_rate": 4.1035353535353544e-07,
      "loss": 1.0985,
      "step": 13
    },
    {
      "epoch": 0.004420587306599305,
      "grad_norm": 0.07009590417146683,
      "learning_rate": 4.419191919191919e-07,
      "loss": 1.0994,
      "step": 14
    },
    {
      "epoch": 0.00473634354278497,
      "grad_norm": 0.12264809757471085,
      "learning_rate": 4.7348484848484853e-07,
      "loss": 1.1015,
      "step": 15
    },
    {
      "epoch": 0.0050520997789706345,
      "grad_norm": 0.06465878337621689,
      "learning_rate": 5.050505050505052e-07,
      "loss": 1.0966,
      "step": 16
    },
    {
      "epoch": 0.0053678560151563,
      "grad_norm": 0.045748695731163025,
      "learning_rate": 5.366161616161617e-07,
      "loss": 1.0988,
      "step": 17
    },
    {
      "epoch": 0.005683612251341964,
      "grad_norm": 0.053932223469018936,
      "learning_rate": 5.681818181818182e-07,
      "loss": 1.0982,
      "step": 18
    },
    {
      "epoch": 0.005999368487527629,
      "grad_norm": 0.054532796144485474,
      "learning_rate": 5.997474747474747e-07,
      "loss": 1.0993,
      "step": 19
    },
    {
      "epoch": 0.006315124723713293,
      "grad_norm": 0.08200918138027191,
      "learning_rate": 6.313131313131313e-07,
      "loss": 1.1006,
      "step": 20
    },
    {
      "epoch": 0.006630880959898958,
      "grad_norm": 0.07089132815599442,
      "learning_rate": 6.62878787878788e-07,
      "loss": 1.0987,
      "step": 21
    },
    {
      "epoch": 0.006946637196084623,
      "grad_norm": 0.08691272139549255,
      "learning_rate": 6.944444444444445e-07,
      "loss": 1.0976,
      "step": 22
    },
    {
      "epoch": 0.007262393432270287,
      "grad_norm": 0.056893762201070786,
      "learning_rate": 7.26010101010101e-07,
      "loss": 1.0995,
      "step": 23
    },
    {
      "epoch": 0.007578149668455952,
      "grad_norm": 0.04969458654522896,
      "learning_rate": 7.575757575757576e-07,
      "loss": 1.0999,
      "step": 24
    },
    {
      "epoch": 0.007893905904641616,
      "grad_norm": 0.07230685651302338,
      "learning_rate": 7.891414141414142e-07,
      "loss": 1.0983,
      "step": 25
    },
    {
      "epoch": 0.008209662140827282,
      "grad_norm": 0.05815694108605385,
      "learning_rate": 8.207070707070709e-07,
      "loss": 1.0984,
      "step": 26
    },
    {
      "epoch": 0.008525418377012947,
      "grad_norm": 0.09083133935928345,
      "learning_rate": 8.522727272727273e-07,
      "loss": 1.0959,
      "step": 27
    },
    {
      "epoch": 0.00884117461319861,
      "grad_norm": 0.06982773542404175,
      "learning_rate": 8.838383838383838e-07,
      "loss": 1.0974,
      "step": 28
    },
    {
      "epoch": 0.009156930849384275,
      "grad_norm": 0.05210331082344055,
      "learning_rate": 9.154040404040404e-07,
      "loss": 1.0991,
      "step": 29
    },
    {
      "epoch": 0.00947268708556994,
      "grad_norm": 0.07270777970552444,
      "learning_rate": 9.469696969696971e-07,
      "loss": 1.099,
      "step": 30
    },
    {
      "epoch": 0.009788443321755604,
      "grad_norm": 0.053407203406095505,
      "learning_rate": 9.785353535353537e-07,
      "loss": 1.0984,
      "step": 31
    },
    {
      "epoch": 0.010104199557941269,
      "grad_norm": 0.05233712121844292,
      "learning_rate": 1.0101010101010103e-06,
      "loss": 1.099,
      "step": 32
    },
    {
      "epoch": 0.010419955794126934,
      "grad_norm": 0.09891868382692337,
      "learning_rate": 1.0416666666666667e-06,
      "loss": 1.0968,
      "step": 33
    },
    {
      "epoch": 0.0107357120303126,
      "grad_norm": 0.04894287511706352,
      "learning_rate": 1.0732323232323233e-06,
      "loss": 1.0977,
      "step": 34
    },
    {
      "epoch": 0.011051468266498263,
      "grad_norm": 0.048936471343040466,
      "learning_rate": 1.1047979797979798e-06,
      "loss": 1.0988,
      "step": 35
    },
    {
      "epoch": 0.011367224502683928,
      "grad_norm": 0.07952558249235153,
      "learning_rate": 1.1363636363636364e-06,
      "loss": 1.1004,
      "step": 36
    },
    {
      "epoch": 0.011682980738869593,
      "grad_norm": 0.059824295341968536,
      "learning_rate": 1.167929292929293e-06,
      "loss": 1.0973,
      "step": 37
    },
    {
      "epoch": 0.011998736975055258,
      "grad_norm": 0.051178302615880966,
      "learning_rate": 1.1994949494949494e-06,
      "loss": 1.0982,
      "step": 38
    },
    {
      "epoch": 0.012314493211240922,
      "grad_norm": 0.08401966094970703,
      "learning_rate": 1.231060606060606e-06,
      "loss": 1.0994,
      "step": 39
    },
    {
      "epoch": 0.012630249447426587,
      "grad_norm": 0.04900215193629265,
      "learning_rate": 1.2626262626262627e-06,
      "loss": 1.0985,
      "step": 40
    },
    {
      "epoch": 0.012946005683612252,
      "grad_norm": 0.061783768236637115,
      "learning_rate": 1.2941919191919193e-06,
      "loss": 1.0975,
      "step": 41
    },
    {
      "epoch": 0.013261761919797915,
      "grad_norm": 0.05533769726753235,
      "learning_rate": 1.325757575757576e-06,
      "loss": 1.0974,
      "step": 42
    },
    {
      "epoch": 0.01357751815598358,
      "grad_norm": 0.052419718354940414,
      "learning_rate": 1.3573232323232325e-06,
      "loss": 1.0978,
      "step": 43
    },
    {
      "epoch": 0.013893274392169246,
      "grad_norm": 0.06287167966365814,
      "learning_rate": 1.388888888888889e-06,
      "loss": 1.0979,
      "step": 44
    },
    {
      "epoch": 0.01420903062835491,
      "grad_norm": 0.05939910560846329,
      "learning_rate": 1.4204545454545456e-06,
      "loss": 1.096,
      "step": 45
    },
    {
      "epoch": 0.014524786864540574,
      "grad_norm": 0.07033021003007889,
      "learning_rate": 1.452020202020202e-06,
      "loss": 1.0997,
      "step": 46
    },
    {
      "epoch": 0.01484054310072624,
      "grad_norm": 0.0600593127310276,
      "learning_rate": 1.4835858585858586e-06,
      "loss": 1.0987,
      "step": 47
    },
    {
      "epoch": 0.015156299336911904,
      "grad_norm": 0.051033973693847656,
      "learning_rate": 1.5151515151515152e-06,
      "loss": 1.0977,
      "step": 48
    },
    {
      "epoch": 0.015472055573097568,
      "grad_norm": 0.08523749560117722,
      "learning_rate": 1.5467171717171719e-06,
      "loss": 1.0971,
      "step": 49
    },
    {
      "epoch": 0.015787811809283233,
      "grad_norm": 0.05341352894902229,
      "learning_rate": 1.5782828282828285e-06,
      "loss": 1.0979,
      "step": 50
    },
    {
      "epoch": 0.016103568045468898,
      "grad_norm": 0.05354122444987297,
      "learning_rate": 1.609848484848485e-06,
      "loss": 1.0988,
      "step": 51
    },
    {
      "epoch": 0.016419324281654563,
      "grad_norm": 0.07713225483894348,
      "learning_rate": 1.6414141414141417e-06,
      "loss": 1.0981,
      "step": 52
    },
    {
      "epoch": 0.01673508051784023,
      "grad_norm": 0.0811149850487709,
      "learning_rate": 1.6729797979797982e-06,
      "loss": 1.0963,
      "step": 53
    },
    {
      "epoch": 0.017050836754025894,
      "grad_norm": 0.08243317157030106,
      "learning_rate": 1.7045454545454546e-06,
      "loss": 1.0971,
      "step": 54
    },
    {
      "epoch": 0.017366592990211555,
      "grad_norm": 0.06662411987781525,
      "learning_rate": 1.7361111111111112e-06,
      "loss": 1.0982,
      "step": 55
    },
    {
      "epoch": 0.01768234922639722,
      "grad_norm": 0.09489083290100098,
      "learning_rate": 1.7676767676767676e-06,
      "loss": 1.0987,
      "step": 56
    },
    {
      "epoch": 0.017998105462582886,
      "grad_norm": 0.09841329604387283,
      "learning_rate": 1.7992424242424244e-06,
      "loss": 1.097,
      "step": 57
    },
    {
      "epoch": 0.01831386169876855,
      "grad_norm": 0.04969426244497299,
      "learning_rate": 1.8308080808080809e-06,
      "loss": 1.098,
      "step": 58
    },
    {
      "epoch": 0.018629617934954216,
      "grad_norm": 0.06917382031679153,
      "learning_rate": 1.8623737373737373e-06,
      "loss": 1.0988,
      "step": 59
    },
    {
      "epoch": 0.01894537417113988,
      "grad_norm": 0.06090676411986351,
      "learning_rate": 1.8939393939393941e-06,
      "loss": 1.0983,
      "step": 60
    },
    {
      "epoch": 0.019261130407325546,
      "grad_norm": 0.05249316617846489,
      "learning_rate": 1.9255050505050505e-06,
      "loss": 1.0974,
      "step": 61
    },
    {
      "epoch": 0.019576886643511208,
      "grad_norm": 0.04892574995756149,
      "learning_rate": 1.9570707070707074e-06,
      "loss": 1.0975,
      "step": 62
    },
    {
      "epoch": 0.019892642879696873,
      "grad_norm": 0.08545341342687607,
      "learning_rate": 1.9886363636363638e-06,
      "loss": 1.0994,
      "step": 63
    },
    {
      "epoch": 0.020208399115882538,
      "grad_norm": 0.08052720129489899,
      "learning_rate": 2.0202020202020206e-06,
      "loss": 1.0996,
      "step": 64
    },
    {
      "epoch": 0.020524155352068203,
      "grad_norm": 0.2860078811645508,
      "learning_rate": 2.051767676767677e-06,
      "loss": 1.098,
      "step": 65
    },
    {
      "epoch": 0.02083991158825387,
      "grad_norm": 0.06053997576236725,
      "learning_rate": 2.0833333333333334e-06,
      "loss": 1.0981,
      "step": 66
    },
    {
      "epoch": 0.021155667824439534,
      "grad_norm": 0.05722038820385933,
      "learning_rate": 2.1148989898989903e-06,
      "loss": 1.0996,
      "step": 67
    },
    {
      "epoch": 0.0214714240606252,
      "grad_norm": 0.06237718090415001,
      "learning_rate": 2.1464646464646467e-06,
      "loss": 1.099,
      "step": 68
    },
    {
      "epoch": 0.02178718029681086,
      "grad_norm": 0.04996383190155029,
      "learning_rate": 2.178030303030303e-06,
      "loss": 1.0979,
      "step": 69
    },
    {
      "epoch": 0.022102936532996525,
      "grad_norm": 0.05649196729063988,
      "learning_rate": 2.2095959595959595e-06,
      "loss": 1.0967,
      "step": 70
    },
    {
      "epoch": 0.02241869276918219,
      "grad_norm": 0.05534100532531738,
      "learning_rate": 2.241161616161616e-06,
      "loss": 1.0983,
      "step": 71
    },
    {
      "epoch": 0.022734449005367856,
      "grad_norm": 0.056381888687610626,
      "learning_rate": 2.2727272727272728e-06,
      "loss": 1.098,
      "step": 72
    },
    {
      "epoch": 0.02305020524155352,
      "grad_norm": 0.06978566944599152,
      "learning_rate": 2.304292929292929e-06,
      "loss": 1.0976,
      "step": 73
    },
    {
      "epoch": 0.023365961477739186,
      "grad_norm": 0.0752277821302414,
      "learning_rate": 2.335858585858586e-06,
      "loss": 1.098,
      "step": 74
    },
    {
      "epoch": 0.02368171771392485,
      "grad_norm": 0.07124051451683044,
      "learning_rate": 2.3674242424242424e-06,
      "loss": 1.0986,
      "step": 75
    },
    {
      "epoch": 0.023997473950110516,
      "grad_norm": 0.09954211860895157,
      "learning_rate": 2.398989898989899e-06,
      "loss": 1.1,
      "step": 76
    },
    {
      "epoch": 0.024313230186296178,
      "grad_norm": 0.08303781598806381,
      "learning_rate": 2.4305555555555557e-06,
      "loss": 1.0996,
      "step": 77
    },
    {
      "epoch": 0.024628986422481843,
      "grad_norm": 0.05681541934609413,
      "learning_rate": 2.462121212121212e-06,
      "loss": 1.0972,
      "step": 78
    },
    {
      "epoch": 0.02494474265866751,
      "grad_norm": 0.08923690766096115,
      "learning_rate": 2.493686868686869e-06,
      "loss": 1.0987,
      "step": 79
    },
    {
      "epoch": 0.025260498894853173,
      "grad_norm": 0.0646296888589859,
      "learning_rate": 2.5252525252525253e-06,
      "loss": 1.0983,
      "step": 80
    },
    {
      "epoch": 0.02557625513103884,
      "grad_norm": 0.07599499821662903,
      "learning_rate": 2.556818181818182e-06,
      "loss": 1.0988,
      "step": 81
    },
    {
      "epoch": 0.025892011367224504,
      "grad_norm": 0.054775968194007874,
      "learning_rate": 2.5883838383838386e-06,
      "loss": 1.0963,
      "step": 82
    },
    {
      "epoch": 0.02620776760341017,
      "grad_norm": 0.05625612288713455,
      "learning_rate": 2.619949494949495e-06,
      "loss": 1.0969,
      "step": 83
    },
    {
      "epoch": 0.02652352383959583,
      "grad_norm": 0.06741708517074585,
      "learning_rate": 2.651515151515152e-06,
      "loss": 1.0966,
      "step": 84
    },
    {
      "epoch": 0.026839280075781496,
      "grad_norm": 0.06261222809553146,
      "learning_rate": 2.6830808080808082e-06,
      "loss": 1.0973,
      "step": 85
    },
    {
      "epoch": 0.02715503631196716,
      "grad_norm": 0.05961050093173981,
      "learning_rate": 2.714646464646465e-06,
      "loss": 1.098,
      "step": 86
    },
    {
      "epoch": 0.027470792548152826,
      "grad_norm": 0.05170012265443802,
      "learning_rate": 2.7462121212121215e-06,
      "loss": 1.0964,
      "step": 87
    },
    {
      "epoch": 0.02778654878433849,
      "grad_norm": 0.05003700032830238,
      "learning_rate": 2.777777777777778e-06,
      "loss": 1.0986,
      "step": 88
    },
    {
      "epoch": 0.028102305020524156,
      "grad_norm": 0.05898367613554001,
      "learning_rate": 2.8093434343434343e-06,
      "loss": 1.0979,
      "step": 89
    },
    {
      "epoch": 0.02841806125670982,
      "grad_norm": 0.07057484984397888,
      "learning_rate": 2.840909090909091e-06,
      "loss": 1.0984,
      "step": 90
    },
    {
      "epoch": 0.028733817492895483,
      "grad_norm": 0.04932938143610954,
      "learning_rate": 2.8724747474747476e-06,
      "loss": 1.0971,
      "step": 91
    },
    {
      "epoch": 0.02904957372908115,
      "grad_norm": 0.1260615587234497,
      "learning_rate": 2.904040404040404e-06,
      "loss": 1.096,
      "step": 92
    },
    {
      "epoch": 0.029365329965266813,
      "grad_norm": 0.05851888656616211,
      "learning_rate": 2.935606060606061e-06,
      "loss": 1.0973,
      "step": 93
    },
    {
      "epoch": 0.02968108620145248,
      "grad_norm": 0.07530558854341507,
      "learning_rate": 2.9671717171717172e-06,
      "loss": 1.0953,
      "step": 94
    },
    {
      "epoch": 0.029996842437638144,
      "grad_norm": 0.059522807598114014,
      "learning_rate": 2.9987373737373737e-06,
      "loss": 1.0967,
      "step": 95
    },
    {
      "epoch": 0.03031259867382381,
      "grad_norm": 0.048266660422086716,
      "learning_rate": 3.0303030303030305e-06,
      "loss": 1.0976,
      "step": 96
    },
    {
      "epoch": 0.030628354910009474,
      "grad_norm": 0.07331649214029312,
      "learning_rate": 3.061868686868687e-06,
      "loss": 1.0956,
      "step": 97
    },
    {
      "epoch": 0.030944111146195136,
      "grad_norm": 0.08767957240343094,
      "learning_rate": 3.0934343434343437e-06,
      "loss": 1.0971,
      "step": 98
    },
    {
      "epoch": 0.031259867382380804,
      "grad_norm": 0.0746714249253273,
      "learning_rate": 3.125e-06,
      "loss": 1.0971,
      "step": 99
    },
    {
      "epoch": 0.031575623618566466,
      "grad_norm": 0.05144079029560089,
      "learning_rate": 3.156565656565657e-06,
      "loss": 1.0973,
      "step": 100
    },
    {
      "epoch": 0.031891379854752135,
      "grad_norm": 0.07223725318908691,
      "learning_rate": 3.188131313131313e-06,
      "loss": 1.0977,
      "step": 101
    },
    {
      "epoch": 0.032207136090937796,
      "grad_norm": 0.1083531305193901,
      "learning_rate": 3.21969696969697e-06,
      "loss": 1.0989,
      "step": 102
    },
    {
      "epoch": 0.03252289232712346,
      "grad_norm": 0.0534820556640625,
      "learning_rate": 3.2512626262626266e-06,
      "loss": 1.0969,
      "step": 103
    },
    {
      "epoch": 0.03283864856330913,
      "grad_norm": 0.0541505292057991,
      "learning_rate": 3.2828282828282835e-06,
      "loss": 1.0983,
      "step": 104
    },
    {
      "epoch": 0.03315440479949479,
      "grad_norm": 0.05242721363902092,
      "learning_rate": 3.3143939393939395e-06,
      "loss": 1.0954,
      "step": 105
    },
    {
      "epoch": 0.03347016103568046,
      "grad_norm": 0.04434196650981903,
      "learning_rate": 3.3459595959595963e-06,
      "loss": 1.0981,
      "step": 106
    },
    {
      "epoch": 0.03378591727186612,
      "grad_norm": 0.07222594320774078,
      "learning_rate": 3.3775252525252527e-06,
      "loss": 1.0965,
      "step": 107
    },
    {
      "epoch": 0.03410167350805179,
      "grad_norm": 0.06457417458295822,
      "learning_rate": 3.409090909090909e-06,
      "loss": 1.097,
      "step": 108
    },
    {
      "epoch": 0.03441742974423745,
      "grad_norm": 0.10698195546865463,
      "learning_rate": 3.440656565656566e-06,
      "loss": 1.0973,
      "step": 109
    },
    {
      "epoch": 0.03473318598042311,
      "grad_norm": 0.06179855018854141,
      "learning_rate": 3.4722222222222224e-06,
      "loss": 1.0973,
      "step": 110
    },
    {
      "epoch": 0.03504894221660878,
      "grad_norm": 0.0509493350982666,
      "learning_rate": 3.503787878787879e-06,
      "loss": 1.0982,
      "step": 111
    },
    {
      "epoch": 0.03536469845279444,
      "grad_norm": 0.11875010281801224,
      "learning_rate": 3.5353535353535352e-06,
      "loss": 1.0976,
      "step": 112
    },
    {
      "epoch": 0.03568045468898011,
      "grad_norm": 0.06116895005106926,
      "learning_rate": 3.566919191919192e-06,
      "loss": 1.096,
      "step": 113
    },
    {
      "epoch": 0.03599621092516577,
      "grad_norm": 0.07928094267845154,
      "learning_rate": 3.598484848484849e-06,
      "loss": 1.097,
      "step": 114
    },
    {
      "epoch": 0.03631196716135144,
      "grad_norm": 0.058139409869909286,
      "learning_rate": 3.630050505050505e-06,
      "loss": 1.0959,
      "step": 115
    },
    {
      "epoch": 0.0366277233975371,
      "grad_norm": 0.05641845613718033,
      "learning_rate": 3.6616161616161617e-06,
      "loss": 1.0966,
      "step": 116
    },
    {
      "epoch": 0.03694347963372276,
      "grad_norm": 0.05987712740898132,
      "learning_rate": 3.6931818181818186e-06,
      "loss": 1.0948,
      "step": 117
    },
    {
      "epoch": 0.03725923586990843,
      "grad_norm": 0.048978857696056366,
      "learning_rate": 3.7247474747474745e-06,
      "loss": 1.0967,
      "step": 118
    },
    {
      "epoch": 0.03757499210609409,
      "grad_norm": 0.0873260498046875,
      "learning_rate": 3.7563131313131314e-06,
      "loss": 1.0966,
      "step": 119
    },
    {
      "epoch": 0.03789074834227976,
      "grad_norm": 0.06199382618069649,
      "learning_rate": 3.7878787878787882e-06,
      "loss": 1.0969,
      "step": 120
    },
    {
      "epoch": 0.038206504578465424,
      "grad_norm": 0.08277177065610886,
      "learning_rate": 3.819444444444445e-06,
      "loss": 1.0974,
      "step": 121
    },
    {
      "epoch": 0.03852226081465109,
      "grad_norm": 0.08998177945613861,
      "learning_rate": 3.851010101010101e-06,
      "loss": 1.0967,
      "step": 122
    },
    {
      "epoch": 0.038838017050836754,
      "grad_norm": 0.07284776121377945,
      "learning_rate": 3.882575757575758e-06,
      "loss": 1.0963,
      "step": 123
    },
    {
      "epoch": 0.039153773287022416,
      "grad_norm": 0.09486658871173859,
      "learning_rate": 3.914141414141415e-06,
      "loss": 1.0975,
      "step": 124
    },
    {
      "epoch": 0.039469529523208084,
      "grad_norm": 0.06845968961715698,
      "learning_rate": 3.945707070707071e-06,
      "loss": 1.0961,
      "step": 125
    },
    {
      "epoch": 0.039785285759393746,
      "grad_norm": 0.05178946629166603,
      "learning_rate": 3.9772727272727275e-06,
      "loss": 1.0951,
      "step": 126
    },
    {
      "epoch": 0.040101041995579415,
      "grad_norm": 0.07466963678598404,
      "learning_rate": 4.008838383838384e-06,
      "loss": 1.0971,
      "step": 127
    },
    {
      "epoch": 0.040416798231765076,
      "grad_norm": 0.1045064702630043,
      "learning_rate": 4.040404040404041e-06,
      "loss": 1.0963,
      "step": 128
    },
    {
      "epoch": 0.040732554467950745,
      "grad_norm": 0.08740954846143723,
      "learning_rate": 4.071969696969697e-06,
      "loss": 1.0941,
      "step": 129
    },
    {
      "epoch": 0.041048310704136406,
      "grad_norm": 0.06488921493291855,
      "learning_rate": 4.103535353535354e-06,
      "loss": 1.0947,
      "step": 130
    },
    {
      "epoch": 0.04136406694032207,
      "grad_norm": 0.08195721358060837,
      "learning_rate": 4.135101010101011e-06,
      "loss": 1.0957,
      "step": 131
    },
    {
      "epoch": 0.04167982317650774,
      "grad_norm": 0.06570427119731903,
      "learning_rate": 4.166666666666667e-06,
      "loss": 1.0965,
      "step": 132
    },
    {
      "epoch": 0.0419955794126934,
      "grad_norm": 0.10350573807954788,
      "learning_rate": 4.198232323232324e-06,
      "loss": 1.0944,
      "step": 133
    },
    {
      "epoch": 0.04231133564887907,
      "grad_norm": 0.11973942071199417,
      "learning_rate": 4.2297979797979805e-06,
      "loss": 1.0933,
      "step": 134
    },
    {
      "epoch": 0.04262709188506473,
      "grad_norm": 0.08487990498542786,
      "learning_rate": 4.2613636363636365e-06,
      "loss": 1.0939,
      "step": 135
    },
    {
      "epoch": 0.0429428481212504,
      "grad_norm": 0.061200208961963654,
      "learning_rate": 4.292929292929293e-06,
      "loss": 1.0945,
      "step": 136
    },
    {
      "epoch": 0.04325860435743606,
      "grad_norm": 0.06257519870996475,
      "learning_rate": 4.32449494949495e-06,
      "loss": 1.0961,
      "step": 137
    },
    {
      "epoch": 0.04357436059362172,
      "grad_norm": 0.08100945502519608,
      "learning_rate": 4.356060606060606e-06,
      "loss": 1.0945,
      "step": 138
    },
    {
      "epoch": 0.04389011682980739,
      "grad_norm": 0.10813286155462265,
      "learning_rate": 4.387626262626263e-06,
      "loss": 1.0969,
      "step": 139
    },
    {
      "epoch": 0.04420587306599305,
      "grad_norm": 0.10505937784910202,
      "learning_rate": 4.419191919191919e-06,
      "loss": 1.0926,
      "step": 140
    },
    {
      "epoch": 0.04452162930217872,
      "grad_norm": 0.059682898223400116,
      "learning_rate": 4.450757575757576e-06,
      "loss": 1.0956,
      "step": 141
    },
    {
      "epoch": 0.04483738553836438,
      "grad_norm": 0.0745309591293335,
      "learning_rate": 4.482323232323232e-06,
      "loss": 1.0945,
      "step": 142
    },
    {
      "epoch": 0.04515314177455005,
      "grad_norm": 0.08240137249231339,
      "learning_rate": 4.513888888888889e-06,
      "loss": 1.0953,
      "step": 143
    },
    {
      "epoch": 0.04546889801073571,
      "grad_norm": 0.056432824581861496,
      "learning_rate": 4.5454545454545455e-06,
      "loss": 1.0946,
      "step": 144
    },
    {
      "epoch": 0.04578465424692137,
      "grad_norm": 0.09898233413696289,
      "learning_rate": 4.577020202020202e-06,
      "loss": 1.0947,
      "step": 145
    },
    {
      "epoch": 0.04610041048310704,
      "grad_norm": 0.06089695543050766,
      "learning_rate": 4.608585858585858e-06,
      "loss": 1.0955,
      "step": 146
    },
    {
      "epoch": 0.046416166719292704,
      "grad_norm": 0.09510543942451477,
      "learning_rate": 4.640151515151515e-06,
      "loss": 1.0949,
      "step": 147
    },
    {
      "epoch": 0.04673192295547837,
      "grad_norm": 0.06204484775662422,
      "learning_rate": 4.671717171717172e-06,
      "loss": 1.0935,
      "step": 148
    },
    {
      "epoch": 0.047047679191664034,
      "grad_norm": 0.07339562475681305,
      "learning_rate": 4.703282828282828e-06,
      "loss": 1.094,
      "step": 149
    },
    {
      "epoch": 0.0473634354278497,
      "grad_norm": 0.05891069397330284,
      "learning_rate": 4.734848484848485e-06,
      "loss": 1.0946,
      "step": 150
    },
    {
      "epoch": 0.047679191664035364,
      "grad_norm": 0.06320498138666153,
      "learning_rate": 4.766414141414142e-06,
      "loss": 1.0936,
      "step": 151
    },
    {
      "epoch": 0.04799494790022103,
      "grad_norm": 0.061813872307538986,
      "learning_rate": 4.797979797979798e-06,
      "loss": 1.0943,
      "step": 152
    },
    {
      "epoch": 0.048310704136406694,
      "grad_norm": 0.08990024030208588,
      "learning_rate": 4.8295454545454545e-06,
      "loss": 1.0923,
      "step": 153
    },
    {
      "epoch": 0.048626460372592356,
      "grad_norm": 0.08471833914518356,
      "learning_rate": 4.861111111111111e-06,
      "loss": 1.092,
      "step": 154
    },
    {
      "epoch": 0.048942216608778025,
      "grad_norm": 0.0926666110754013,
      "learning_rate": 4.892676767676768e-06,
      "loss": 1.0897,
      "step": 155
    },
    {
      "epoch": 0.049257972844963686,
      "grad_norm": 0.07611725479364395,
      "learning_rate": 4.924242424242424e-06,
      "loss": 1.0935,
      "step": 156
    },
    {
      "epoch": 0.049573729081149355,
      "grad_norm": 0.09117670357227325,
      "learning_rate": 4.955808080808081e-06,
      "loss": 1.0915,
      "step": 157
    },
    {
      "epoch": 0.04988948531733502,
      "grad_norm": 0.0679323598742485,
      "learning_rate": 4.987373737373738e-06,
      "loss": 1.0957,
      "step": 158
    },
    {
      "epoch": 0.050205241553520685,
      "grad_norm": 0.0844004899263382,
      "learning_rate": 5.018939393939394e-06,
      "loss": 1.0946,
      "step": 159
    },
    {
      "epoch": 0.05052099778970635,
      "grad_norm": 0.06716519594192505,
      "learning_rate": 5.050505050505051e-06,
      "loss": 1.096,
      "step": 160
    },
    {
      "epoch": 0.05083675402589201,
      "grad_norm": 0.07112353295087814,
      "learning_rate": 5.0820707070707075e-06,
      "loss": 1.0929,
      "step": 161
    },
    {
      "epoch": 0.05115251026207768,
      "grad_norm": 0.06587293744087219,
      "learning_rate": 5.113636363636364e-06,
      "loss": 1.0919,
      "step": 162
    },
    {
      "epoch": 0.05146826649826334,
      "grad_norm": 0.06838454306125641,
      "learning_rate": 5.14520202020202e-06,
      "loss": 1.092,
      "step": 163
    },
    {
      "epoch": 0.05178402273444901,
      "grad_norm": 0.07022113353013992,
      "learning_rate": 5.176767676767677e-06,
      "loss": 1.0907,
      "step": 164
    },
    {
      "epoch": 0.05209977897063467,
      "grad_norm": 0.11650644987821579,
      "learning_rate": 5.208333333333334e-06,
      "loss": 1.0886,
      "step": 165
    },
    {
      "epoch": 0.05241553520682034,
      "grad_norm": 0.07217497378587723,
      "learning_rate": 5.23989898989899e-06,
      "loss": 1.0906,
      "step": 166
    },
    {
      "epoch": 0.052731291443006,
      "grad_norm": 0.08375301212072372,
      "learning_rate": 5.271464646464647e-06,
      "loss": 1.0889,
      "step": 167
    },
    {
      "epoch": 0.05304704767919166,
      "grad_norm": 0.08977717906236649,
      "learning_rate": 5.303030303030304e-06,
      "loss": 1.0867,
      "step": 168
    },
    {
      "epoch": 0.05336280391537733,
      "grad_norm": 0.08404728770256042,
      "learning_rate": 5.33459595959596e-06,
      "loss": 1.0934,
      "step": 169
    },
    {
      "epoch": 0.05367856015156299,
      "grad_norm": 0.09051228314638138,
      "learning_rate": 5.3661616161616165e-06,
      "loss": 1.0949,
      "step": 170
    },
    {
      "epoch": 0.05399431638774866,
      "grad_norm": 0.10931982845067978,
      "learning_rate": 5.397727272727273e-06,
      "loss": 1.0891,
      "step": 171
    },
    {
      "epoch": 0.05431007262393432,
      "grad_norm": 0.09252841025590897,
      "learning_rate": 5.42929292929293e-06,
      "loss": 1.0905,
      "step": 172
    },
    {
      "epoch": 0.05462582886011999,
      "grad_norm": 0.0739794671535492,
      "learning_rate": 5.460858585858586e-06,
      "loss": 1.0903,
      "step": 173
    },
    {
      "epoch": 0.05494158509630565,
      "grad_norm": 0.07709450274705887,
      "learning_rate": 5.492424242424243e-06,
      "loss": 1.0928,
      "step": 174
    },
    {
      "epoch": 0.055257341332491314,
      "grad_norm": 0.08967255800962448,
      "learning_rate": 5.5239898989899e-06,
      "loss": 1.0893,
      "step": 175
    },
    {
      "epoch": 0.05557309756867698,
      "grad_norm": 0.06423037499189377,
      "learning_rate": 5.555555555555556e-06,
      "loss": 1.0939,
      "step": 176
    },
    {
      "epoch": 0.055888853804862644,
      "grad_norm": 0.0745222195982933,
      "learning_rate": 5.587121212121213e-06,
      "loss": 1.0885,
      "step": 177
    },
    {
      "epoch": 0.05620461004104831,
      "grad_norm": 0.08286277949810028,
      "learning_rate": 5.618686868686869e-06,
      "loss": 1.0897,
      "step": 178
    },
    {
      "epoch": 0.056520366277233974,
      "grad_norm": 0.08704034239053726,
      "learning_rate": 5.6502525252525255e-06,
      "loss": 1.0891,
      "step": 179
    },
    {
      "epoch": 0.05683612251341964,
      "grad_norm": 0.08263607323169708,
      "learning_rate": 5.681818181818182e-06,
      "loss": 1.0887,
      "step": 180
    },
    {
      "epoch": 0.057151878749605305,
      "grad_norm": 0.09381820261478424,
      "learning_rate": 5.713383838383838e-06,
      "loss": 1.0889,
      "step": 181
    },
    {
      "epoch": 0.057467634985790966,
      "grad_norm": 0.09104549139738083,
      "learning_rate": 5.744949494949495e-06,
      "loss": 1.0871,
      "step": 182
    },
    {
      "epoch": 0.057783391221976635,
      "grad_norm": 0.11508553475141525,
      "learning_rate": 5.776515151515151e-06,
      "loss": 1.0843,
      "step": 183
    },
    {
      "epoch": 0.0580991474581623,
      "grad_norm": 0.1069478988647461,
      "learning_rate": 5.808080808080808e-06,
      "loss": 1.087,
      "step": 184
    },
    {
      "epoch": 0.058414903694347965,
      "grad_norm": 0.07763667404651642,
      "learning_rate": 5.839646464646465e-06,
      "loss": 1.0889,
      "step": 185
    },
    {
      "epoch": 0.05873065993053363,
      "grad_norm": 0.06702792644500732,
      "learning_rate": 5.871212121212122e-06,
      "loss": 1.0927,
      "step": 186
    },
    {
      "epoch": 0.059046416166719295,
      "grad_norm": 0.09636549651622772,
      "learning_rate": 5.902777777777778e-06,
      "loss": 1.0865,
      "step": 187
    },
    {
      "epoch": 0.05936217240290496,
      "grad_norm": 0.0867915078997612,
      "learning_rate": 5.9343434343434345e-06,
      "loss": 1.0882,
      "step": 188
    },
    {
      "epoch": 0.05967792863909062,
      "grad_norm": 0.08751293271780014,
      "learning_rate": 5.965909090909091e-06,
      "loss": 1.0854,
      "step": 189
    },
    {
      "epoch": 0.05999368487527629,
      "grad_norm": 0.09880711138248444,
      "learning_rate": 5.997474747474747e-06,
      "loss": 1.0853,
      "step": 190
    },
    {
      "epoch": 0.06030944111146195,
      "grad_norm": 0.09696859121322632,
      "learning_rate": 6.029040404040404e-06,
      "loss": 1.0885,
      "step": 191
    },
    {
      "epoch": 0.06062519734764762,
      "grad_norm": 0.09753978252410889,
      "learning_rate": 6.060606060606061e-06,
      "loss": 1.083,
      "step": 192
    },
    {
      "epoch": 0.06094095358383328,
      "grad_norm": 0.08978458493947983,
      "learning_rate": 6.092171717171717e-06,
      "loss": 1.0849,
      "step": 193
    },
    {
      "epoch": 0.06125670982001895,
      "grad_norm": 0.11148930341005325,
      "learning_rate": 6.123737373737374e-06,
      "loss": 1.0876,
      "step": 194
    },
    {
      "epoch": 0.06157246605620461,
      "grad_norm": 0.09838321805000305,
      "learning_rate": 6.155303030303031e-06,
      "loss": 1.0851,
      "step": 195
    },
    {
      "epoch": 0.06188822229239027,
      "grad_norm": 0.14109255373477936,
      "learning_rate": 6.1868686868686875e-06,
      "loss": 1.0794,
      "step": 196
    },
    {
      "epoch": 0.06220397852857594,
      "grad_norm": 0.11522294580936432,
      "learning_rate": 6.2184343434343435e-06,
      "loss": 1.08,
      "step": 197
    },
    {
      "epoch": 0.06251973476476161,
      "grad_norm": 0.08193270862102509,
      "learning_rate": 6.25e-06,
      "loss": 1.0882,
      "step": 198
    },
    {
      "epoch": 0.06283549100094726,
      "grad_norm": 0.10392513871192932,
      "learning_rate": 6.281565656565656e-06,
      "loss": 1.0833,
      "step": 199
    },
    {
      "epoch": 0.06315124723713293,
      "grad_norm": 0.10708694159984589,
      "learning_rate": 6.313131313131314e-06,
      "loss": 1.0832,
      "step": 200
    },
    {
      "epoch": 0.06315124723713293,
      "eval_loss": 1.0802605152130127,
      "eval_runtime": 4.5358,
      "eval_samples_per_second": 548.967,
      "eval_steps_per_second": 4.63,
      "step": 200
    },
    {
      "epoch": 0.0634670034733186,
      "grad_norm": 0.13635142147541046,
      "learning_rate": 6.34469696969697e-06,
      "loss": 1.0794,
      "step": 201
    },
    {
      "epoch": 0.06378275970950427,
      "grad_norm": 0.09831973910331726,
      "learning_rate": 6.376262626262626e-06,
      "loss": 1.0807,
      "step": 202
    },
    {
      "epoch": 0.06409851594568992,
      "grad_norm": 0.11791440844535828,
      "learning_rate": 6.407828282828284e-06,
      "loss": 1.0797,
      "step": 203
    },
    {
      "epoch": 0.06441427218187559,
      "grad_norm": 0.13185711205005646,
      "learning_rate": 6.43939393939394e-06,
      "loss": 1.0803,
      "step": 204
    },
    {
      "epoch": 0.06473002841806126,
      "grad_norm": 0.10375262051820755,
      "learning_rate": 6.470959595959596e-06,
      "loss": 1.081,
      "step": 205
    },
    {
      "epoch": 0.06504578465424692,
      "grad_norm": 0.13650067150592804,
      "learning_rate": 6.502525252525253e-06,
      "loss": 1.0781,
      "step": 206
    },
    {
      "epoch": 0.06536154089043258,
      "grad_norm": 0.11903456598520279,
      "learning_rate": 6.534090909090909e-06,
      "loss": 1.0751,
      "step": 207
    },
    {
      "epoch": 0.06567729712661825,
      "grad_norm": 0.1709679216146469,
      "learning_rate": 6.565656565656567e-06,
      "loss": 1.0774,
      "step": 208
    },
    {
      "epoch": 0.06599305336280392,
      "grad_norm": 0.09690535813570023,
      "learning_rate": 6.597222222222223e-06,
      "loss": 1.0865,
      "step": 209
    },
    {
      "epoch": 0.06630880959898958,
      "grad_norm": 0.14622174203395844,
      "learning_rate": 6.628787878787879e-06,
      "loss": 1.082,
      "step": 210
    },
    {
      "epoch": 0.06662456583517525,
      "grad_norm": 0.14259666204452515,
      "learning_rate": 6.660353535353537e-06,
      "loss": 1.078,
      "step": 211
    },
    {
      "epoch": 0.06694032207136091,
      "grad_norm": 0.14993518590927124,
      "learning_rate": 6.691919191919193e-06,
      "loss": 1.08,
      "step": 212
    },
    {
      "epoch": 0.06725607830754657,
      "grad_norm": 0.13724073767662048,
      "learning_rate": 6.723484848484849e-06,
      "loss": 1.0769,
      "step": 213
    },
    {
      "epoch": 0.06757183454373224,
      "grad_norm": 0.1438925862312317,
      "learning_rate": 6.7550505050505055e-06,
      "loss": 1.0801,
      "step": 214
    },
    {
      "epoch": 0.0678875907799179,
      "grad_norm": 0.18603821098804474,
      "learning_rate": 6.786616161616162e-06,
      "loss": 1.0765,
      "step": 215
    },
    {
      "epoch": 0.06820334701610357,
      "grad_norm": 0.15363861620426178,
      "learning_rate": 6.818181818181818e-06,
      "loss": 1.0755,
      "step": 216
    },
    {
      "epoch": 0.06851910325228923,
      "grad_norm": 0.12768477201461792,
      "learning_rate": 6.849747474747475e-06,
      "loss": 1.0797,
      "step": 217
    },
    {
      "epoch": 0.0688348594884749,
      "grad_norm": 0.14207971096038818,
      "learning_rate": 6.881313131313132e-06,
      "loss": 1.075,
      "step": 218
    },
    {
      "epoch": 0.06915061572466057,
      "grad_norm": 0.17873360216617584,
      "learning_rate": 6.912878787878788e-06,
      "loss": 1.0681,
      "step": 219
    },
    {
      "epoch": 0.06946637196084622,
      "grad_norm": 0.1563737392425537,
      "learning_rate": 6.944444444444445e-06,
      "loss": 1.0703,
      "step": 220
    },
    {
      "epoch": 0.06978212819703189,
      "grad_norm": 0.15133796632289886,
      "learning_rate": 6.976010101010101e-06,
      "loss": 1.076,
      "step": 221
    },
    {
      "epoch": 0.07009788443321756,
      "grad_norm": 0.16291847825050354,
      "learning_rate": 7.007575757575758e-06,
      "loss": 1.072,
      "step": 222
    },
    {
      "epoch": 0.07041364066940323,
      "grad_norm": 0.17065799236297607,
      "learning_rate": 7.0391414141414144e-06,
      "loss": 1.0748,
      "step": 223
    },
    {
      "epoch": 0.07072939690558888,
      "grad_norm": 0.183206245303154,
      "learning_rate": 7.0707070707070704e-06,
      "loss": 1.0694,
      "step": 224
    },
    {
      "epoch": 0.07104515314177455,
      "grad_norm": 0.2544916570186615,
      "learning_rate": 7.102272727272728e-06,
      "loss": 1.0714,
      "step": 225
    },
    {
      "epoch": 0.07136090937796022,
      "grad_norm": 0.18675890564918518,
      "learning_rate": 7.133838383838384e-06,
      "loss": 1.067,
      "step": 226
    },
    {
      "epoch": 0.07167666561414587,
      "grad_norm": 0.16745643317699432,
      "learning_rate": 7.16540404040404e-06,
      "loss": 1.0841,
      "step": 227
    },
    {
      "epoch": 0.07199242185033154,
      "grad_norm": 0.16102194786071777,
      "learning_rate": 7.196969696969698e-06,
      "loss": 1.0715,
      "step": 228
    },
    {
      "epoch": 0.07230817808651721,
      "grad_norm": 0.20889601111412048,
      "learning_rate": 7.228535353535354e-06,
      "loss": 1.077,
      "step": 229
    },
    {
      "epoch": 0.07262393432270288,
      "grad_norm": 0.1816893070936203,
      "learning_rate": 7.26010101010101e-06,
      "loss": 1.08,
      "step": 230
    },
    {
      "epoch": 0.07293969055888853,
      "grad_norm": 0.18677875399589539,
      "learning_rate": 7.2916666666666674e-06,
      "loss": 1.0674,
      "step": 231
    },
    {
      "epoch": 0.0732554467950742,
      "grad_norm": 0.12317460030317307,
      "learning_rate": 7.3232323232323234e-06,
      "loss": 1.0764,
      "step": 232
    },
    {
      "epoch": 0.07357120303125987,
      "grad_norm": 0.18762844800949097,
      "learning_rate": 7.354797979797979e-06,
      "loss": 1.0715,
      "step": 233
    },
    {
      "epoch": 0.07388695926744553,
      "grad_norm": 0.16426745057106018,
      "learning_rate": 7.386363636363637e-06,
      "loss": 1.079,
      "step": 234
    },
    {
      "epoch": 0.0742027155036312,
      "grad_norm": 0.1995212435722351,
      "learning_rate": 7.417929292929293e-06,
      "loss": 1.0723,
      "step": 235
    },
    {
      "epoch": 0.07451847173981686,
      "grad_norm": 0.1510445922613144,
      "learning_rate": 7.449494949494949e-06,
      "loss": 1.0647,
      "step": 236
    },
    {
      "epoch": 0.07483422797600253,
      "grad_norm": 0.1497015655040741,
      "learning_rate": 7.481060606060607e-06,
      "loss": 1.0769,
      "step": 237
    },
    {
      "epoch": 0.07514998421218819,
      "grad_norm": 0.16604870557785034,
      "learning_rate": 7.512626262626263e-06,
      "loss": 1.0627,
      "step": 238
    },
    {
      "epoch": 0.07546574044837386,
      "grad_norm": 0.20211978256702423,
      "learning_rate": 7.544191919191919e-06,
      "loss": 1.069,
      "step": 239
    },
    {
      "epoch": 0.07578149668455952,
      "grad_norm": 0.17596371471881866,
      "learning_rate": 7.5757575757575764e-06,
      "loss": 1.0539,
      "step": 240
    },
    {
      "epoch": 0.07609725292074518,
      "grad_norm": 0.15334606170654297,
      "learning_rate": 7.607323232323232e-06,
      "loss": 1.0712,
      "step": 241
    },
    {
      "epoch": 0.07641300915693085,
      "grad_norm": 0.15708066523075104,
      "learning_rate": 7.63888888888889e-06,
      "loss": 1.0635,
      "step": 242
    },
    {
      "epoch": 0.07672876539311652,
      "grad_norm": 0.2821243107318878,
      "learning_rate": 7.670454545454545e-06,
      "loss": 1.0595,
      "step": 243
    },
    {
      "epoch": 0.07704452162930218,
      "grad_norm": 0.20868267118930817,
      "learning_rate": 7.702020202020202e-06,
      "loss": 1.058,
      "step": 244
    },
    {
      "epoch": 0.07736027786548784,
      "grad_norm": 0.1510457992553711,
      "learning_rate": 7.733585858585859e-06,
      "loss": 1.0616,
      "step": 245
    },
    {
      "epoch": 0.07767603410167351,
      "grad_norm": 0.19244422018527985,
      "learning_rate": 7.765151515151516e-06,
      "loss": 1.0651,
      "step": 246
    },
    {
      "epoch": 0.07799179033785918,
      "grad_norm": 0.2643260657787323,
      "learning_rate": 7.796717171717171e-06,
      "loss": 1.067,
      "step": 247
    },
    {
      "epoch": 0.07830754657404483,
      "grad_norm": 0.16809506714344025,
      "learning_rate": 7.82828282828283e-06,
      "loss": 1.0745,
      "step": 248
    },
    {
      "epoch": 0.0786233028102305,
      "grad_norm": 0.15107259154319763,
      "learning_rate": 7.859848484848485e-06,
      "loss": 1.074,
      "step": 249
    },
    {
      "epoch": 0.07893905904641617,
      "grad_norm": 0.3064063787460327,
      "learning_rate": 7.891414141414141e-06,
      "loss": 1.0776,
      "step": 250
    },
    {
      "epoch": 0.07925481528260184,
      "grad_norm": 0.24190372228622437,
      "learning_rate": 7.922979797979798e-06,
      "loss": 1.0493,
      "step": 251
    },
    {
      "epoch": 0.07957057151878749,
      "grad_norm": 0.23116545379161835,
      "learning_rate": 7.954545454545455e-06,
      "loss": 1.0708,
      "step": 252
    },
    {
      "epoch": 0.07988632775497316,
      "grad_norm": 0.1826268583536148,
      "learning_rate": 7.98611111111111e-06,
      "loss": 1.0792,
      "step": 253
    },
    {
      "epoch": 0.08020208399115883,
      "grad_norm": 0.1687881499528885,
      "learning_rate": 8.017676767676769e-06,
      "loss": 1.0689,
      "step": 254
    },
    {
      "epoch": 0.08051784022734448,
      "grad_norm": 0.2598346173763275,
      "learning_rate": 8.049242424242424e-06,
      "loss": 1.0601,
      "step": 255
    },
    {
      "epoch": 0.08083359646353015,
      "grad_norm": 0.1757180094718933,
      "learning_rate": 8.080808080808082e-06,
      "loss": 1.0628,
      "step": 256
    },
    {
      "epoch": 0.08114935269971582,
      "grad_norm": 0.2670367956161499,
      "learning_rate": 8.112373737373738e-06,
      "loss": 1.0554,
      "step": 257
    },
    {
      "epoch": 0.08146510893590149,
      "grad_norm": 0.3284059166908264,
      "learning_rate": 8.143939393939394e-06,
      "loss": 1.0536,
      "step": 258
    },
    {
      "epoch": 0.08178086517208714,
      "grad_norm": 0.2929230034351349,
      "learning_rate": 8.175505050505051e-06,
      "loss": 1.0658,
      "step": 259
    },
    {
      "epoch": 0.08209662140827281,
      "grad_norm": 0.22787392139434814,
      "learning_rate": 8.207070707070708e-06,
      "loss": 1.0667,
      "step": 260
    },
    {
      "epoch": 0.08241237764445848,
      "grad_norm": 0.25668588280677795,
      "learning_rate": 8.238636363636363e-06,
      "loss": 1.063,
      "step": 261
    },
    {
      "epoch": 0.08272813388064414,
      "grad_norm": 0.24845486879348755,
      "learning_rate": 8.270202020202022e-06,
      "loss": 1.0575,
      "step": 262
    },
    {
      "epoch": 0.0830438901168298,
      "grad_norm": 0.21401792764663696,
      "learning_rate": 8.301767676767677e-06,
      "loss": 1.0616,
      "step": 263
    },
    {
      "epoch": 0.08335964635301547,
      "grad_norm": 0.36292022466659546,
      "learning_rate": 8.333333333333334e-06,
      "loss": 1.0376,
      "step": 264
    },
    {
      "epoch": 0.08367540258920114,
      "grad_norm": 0.1590682566165924,
      "learning_rate": 8.36489898989899e-06,
      "loss": 1.0776,
      "step": 265
    },
    {
      "epoch": 0.0839911588253868,
      "grad_norm": 0.2993594706058502,
      "learning_rate": 8.396464646464647e-06,
      "loss": 1.0671,
      "step": 266
    },
    {
      "epoch": 0.08430691506157247,
      "grad_norm": 0.24957239627838135,
      "learning_rate": 8.428030303030303e-06,
      "loss": 1.08,
      "step": 267
    },
    {
      "epoch": 0.08462267129775813,
      "grad_norm": 0.19910024106502533,
      "learning_rate": 8.459595959595961e-06,
      "loss": 1.064,
      "step": 268
    },
    {
      "epoch": 0.08493842753394379,
      "grad_norm": 0.2522094249725342,
      "learning_rate": 8.491161616161616e-06,
      "loss": 1.0659,
      "step": 269
    },
    {
      "epoch": 0.08525418377012946,
      "grad_norm": 0.1643340289592743,
      "learning_rate": 8.522727272727273e-06,
      "loss": 1.0622,
      "step": 270
    },
    {
      "epoch": 0.08556994000631513,
      "grad_norm": 0.20323923230171204,
      "learning_rate": 8.55429292929293e-06,
      "loss": 1.0677,
      "step": 271
    },
    {
      "epoch": 0.0858856962425008,
      "grad_norm": 0.2637196481227875,
      "learning_rate": 8.585858585858587e-06,
      "loss": 1.0704,
      "step": 272
    },
    {
      "epoch": 0.08620145247868645,
      "grad_norm": 0.19106239080429077,
      "learning_rate": 8.617424242424244e-06,
      "loss": 1.0681,
      "step": 273
    },
    {
      "epoch": 0.08651720871487212,
      "grad_norm": 0.29680731892585754,
      "learning_rate": 8.6489898989899e-06,
      "loss": 1.0723,
      "step": 274
    },
    {
      "epoch": 0.08683296495105779,
      "grad_norm": 0.16900919377803802,
      "learning_rate": 8.680555555555556e-06,
      "loss": 1.0638,
      "step": 275
    },
    {
      "epoch": 0.08714872118724344,
      "grad_norm": 0.3333047032356262,
      "learning_rate": 8.712121212121212e-06,
      "loss": 1.07,
      "step": 276
    },
    {
      "epoch": 0.08746447742342911,
      "grad_norm": 0.1957770138978958,
      "learning_rate": 8.74368686868687e-06,
      "loss": 1.0719,
      "step": 277
    },
    {
      "epoch": 0.08778023365961478,
      "grad_norm": 0.17355211079120636,
      "learning_rate": 8.775252525252526e-06,
      "loss": 1.0721,
      "step": 278
    },
    {
      "epoch": 0.08809598989580045,
      "grad_norm": 0.2402275651693344,
      "learning_rate": 8.806818181818183e-06,
      "loss": 1.0618,
      "step": 279
    },
    {
      "epoch": 0.0884117461319861,
      "grad_norm": 0.2632269263267517,
      "learning_rate": 8.838383838383838e-06,
      "loss": 1.0709,
      "step": 280
    },
    {
      "epoch": 0.08872750236817177,
      "grad_norm": 0.2968766689300537,
      "learning_rate": 8.869949494949495e-06,
      "loss": 1.0596,
      "step": 281
    },
    {
      "epoch": 0.08904325860435744,
      "grad_norm": 0.3175920844078064,
      "learning_rate": 8.901515151515152e-06,
      "loss": 1.0479,
      "step": 282
    },
    {
      "epoch": 0.0893590148405431,
      "grad_norm": 0.21412506699562073,
      "learning_rate": 8.933080808080809e-06,
      "loss": 1.0522,
      "step": 283
    },
    {
      "epoch": 0.08967477107672876,
      "grad_norm": 0.23537428677082062,
      "learning_rate": 8.964646464646464e-06,
      "loss": 1.0721,
      "step": 284
    },
    {
      "epoch": 0.08999052731291443,
      "grad_norm": 0.2766004800796509,
      "learning_rate": 8.996212121212122e-06,
      "loss": 1.0611,
      "step": 285
    },
    {
      "epoch": 0.0903062835491001,
      "grad_norm": 0.31695622205734253,
      "learning_rate": 9.027777777777777e-06,
      "loss": 1.0572,
      "step": 286
    },
    {
      "epoch": 0.09062203978528575,
      "grad_norm": 0.2895488440990448,
      "learning_rate": 9.059343434343434e-06,
      "loss": 1.0594,
      "step": 287
    },
    {
      "epoch": 0.09093779602147142,
      "grad_norm": 0.25248533487319946,
      "learning_rate": 9.090909090909091e-06,
      "loss": 1.0496,
      "step": 288
    },
    {
      "epoch": 0.09125355225765709,
      "grad_norm": 0.20282116532325745,
      "learning_rate": 9.122474747474748e-06,
      "loss": 1.0701,
      "step": 289
    },
    {
      "epoch": 0.09156930849384275,
      "grad_norm": 0.42630666494369507,
      "learning_rate": 9.154040404040405e-06,
      "loss": 1.0596,
      "step": 290
    },
    {
      "epoch": 0.09188506473002842,
      "grad_norm": 0.4224831163883209,
      "learning_rate": 9.185606060606062e-06,
      "loss": 1.0809,
      "step": 291
    },
    {
      "epoch": 0.09220082096621408,
      "grad_norm": 0.29358404874801636,
      "learning_rate": 9.217171717171717e-06,
      "loss": 1.0681,
      "step": 292
    },
    {
      "epoch": 0.09251657720239975,
      "grad_norm": 0.34100526571273804,
      "learning_rate": 9.248737373737375e-06,
      "loss": 1.0651,
      "step": 293
    },
    {
      "epoch": 0.09283233343858541,
      "grad_norm": 0.3414398431777954,
      "learning_rate": 9.28030303030303e-06,
      "loss": 1.0584,
      "step": 294
    },
    {
      "epoch": 0.09314808967477108,
      "grad_norm": 0.33828824758529663,
      "learning_rate": 9.311868686868687e-06,
      "loss": 1.0697,
      "step": 295
    },
    {
      "epoch": 0.09346384591095674,
      "grad_norm": 0.3310551047325134,
      "learning_rate": 9.343434343434344e-06,
      "loss": 1.0506,
      "step": 296
    },
    {
      "epoch": 0.09377960214714241,
      "grad_norm": 0.2807527780532837,
      "learning_rate": 9.375000000000001e-06,
      "loss": 1.0633,
      "step": 297
    },
    {
      "epoch": 0.09409535838332807,
      "grad_norm": 0.3926648795604706,
      "learning_rate": 9.406565656565656e-06,
      "loss": 1.049,
      "step": 298
    },
    {
      "epoch": 0.09441111461951374,
      "grad_norm": 0.43429043889045715,
      "learning_rate": 9.438131313131315e-06,
      "loss": 1.0679,
      "step": 299
    },
    {
      "epoch": 0.0947268708556994,
      "grad_norm": 0.5836392045021057,
      "learning_rate": 9.46969696969697e-06,
      "loss": 1.0543,
      "step": 300
    },
    {
      "epoch": 0.09504262709188506,
      "grad_norm": 0.3698714077472687,
      "learning_rate": 9.501262626262627e-06,
      "loss": 1.071,
      "step": 301
    },
    {
      "epoch": 0.09535838332807073,
      "grad_norm": 0.3178062438964844,
      "learning_rate": 9.532828282828283e-06,
      "loss": 1.0468,
      "step": 302
    },
    {
      "epoch": 0.0956741395642564,
      "grad_norm": 0.45679983496665955,
      "learning_rate": 9.56439393939394e-06,
      "loss": 1.0666,
      "step": 303
    },
    {
      "epoch": 0.09598989580044207,
      "grad_norm": 0.3416401445865631,
      "learning_rate": 9.595959595959595e-06,
      "loss": 1.0582,
      "step": 304
    },
    {
      "epoch": 0.09630565203662772,
      "grad_norm": 0.4672188460826874,
      "learning_rate": 9.627525252525254e-06,
      "loss": 1.0586,
      "step": 305
    },
    {
      "epoch": 0.09662140827281339,
      "grad_norm": 0.2530241906642914,
      "learning_rate": 9.659090909090909e-06,
      "loss": 1.0512,
      "step": 306
    },
    {
      "epoch": 0.09693716450899906,
      "grad_norm": 0.3729241192340851,
      "learning_rate": 9.690656565656568e-06,
      "loss": 1.0553,
      "step": 307
    },
    {
      "epoch": 0.09725292074518471,
      "grad_norm": 0.3261946439743042,
      "learning_rate": 9.722222222222223e-06,
      "loss": 1.062,
      "step": 308
    },
    {
      "epoch": 0.09756867698137038,
      "grad_norm": 0.3380749821662903,
      "learning_rate": 9.75378787878788e-06,
      "loss": 1.0735,
      "step": 309
    },
    {
      "epoch": 0.09788443321755605,
      "grad_norm": 0.4099089205265045,
      "learning_rate": 9.785353535353536e-06,
      "loss": 1.0561,
      "step": 310
    },
    {
      "epoch": 0.09820018945374172,
      "grad_norm": 0.31372347474098206,
      "learning_rate": 9.816919191919193e-06,
      "loss": 1.0659,
      "step": 311
    },
    {
      "epoch": 0.09851594568992737,
      "grad_norm": 0.34009605646133423,
      "learning_rate": 9.848484848484848e-06,
      "loss": 1.0486,
      "step": 312
    },
    {
      "epoch": 0.09883170192611304,
      "grad_norm": 0.3971436023712158,
      "learning_rate": 9.880050505050505e-06,
      "loss": 1.0509,
      "step": 313
    },
    {
      "epoch": 0.09914745816229871,
      "grad_norm": 0.3985782563686371,
      "learning_rate": 9.911616161616162e-06,
      "loss": 1.0602,
      "step": 314
    },
    {
      "epoch": 0.09946321439848436,
      "grad_norm": 0.2419920116662979,
      "learning_rate": 9.943181818181819e-06,
      "loss": 1.0531,
      "step": 315
    },
    {
      "epoch": 0.09977897063467003,
      "grad_norm": 0.3071087896823883,
      "learning_rate": 9.974747474747476e-06,
      "loss": 1.0413,
      "step": 316
    },
    {
      "epoch": 0.1000947268708557,
      "grad_norm": 0.26234009861946106,
      "learning_rate": 1.0006313131313133e-05,
      "loss": 1.0514,
      "step": 317
    },
    {
      "epoch": 0.10041048310704137,
      "grad_norm": 0.3232724964618683,
      "learning_rate": 1.0037878787878788e-05,
      "loss": 1.0604,
      "step": 318
    },
    {
      "epoch": 0.10072623934322703,
      "grad_norm": 0.3293020725250244,
      "learning_rate": 1.0069444444444445e-05,
      "loss": 1.0572,
      "step": 319
    },
    {
      "epoch": 0.1010419955794127,
      "grad_norm": 0.29512515664100647,
      "learning_rate": 1.0101010101010101e-05,
      "loss": 1.0454,
      "step": 320
    },
    {
      "epoch": 0.10135775181559836,
      "grad_norm": 0.19367767870426178,
      "learning_rate": 1.0132575757575758e-05,
      "loss": 1.0648,
      "step": 321
    },
    {
      "epoch": 0.10167350805178402,
      "grad_norm": 0.2214515060186386,
      "learning_rate": 1.0164141414141415e-05,
      "loss": 1.0605,
      "step": 322
    },
    {
      "epoch": 0.10198926428796969,
      "grad_norm": 0.22787879407405853,
      "learning_rate": 1.019570707070707e-05,
      "loss": 1.0513,
      "step": 323
    },
    {
      "epoch": 0.10230502052415535,
      "grad_norm": 0.38695091009140015,
      "learning_rate": 1.0227272727272729e-05,
      "loss": 1.0431,
      "step": 324
    },
    {
      "epoch": 0.10262077676034102,
      "grad_norm": 0.31664150953292847,
      "learning_rate": 1.0258838383838384e-05,
      "loss": 1.0736,
      "step": 325
    },
    {
      "epoch": 0.10293653299652668,
      "grad_norm": 0.3188815414905548,
      "learning_rate": 1.029040404040404e-05,
      "loss": 1.0584,
      "step": 326
    },
    {
      "epoch": 0.10325228923271235,
      "grad_norm": 0.422398179769516,
      "learning_rate": 1.0321969696969698e-05,
      "loss": 1.0547,
      "step": 327
    },
    {
      "epoch": 0.10356804546889802,
      "grad_norm": 0.39164063334465027,
      "learning_rate": 1.0353535353535354e-05,
      "loss": 1.0486,
      "step": 328
    },
    {
      "epoch": 0.10388380170508367,
      "grad_norm": 0.4484929144382477,
      "learning_rate": 1.038510101010101e-05,
      "loss": 1.0473,
      "step": 329
    },
    {
      "epoch": 0.10419955794126934,
      "grad_norm": 0.45886120200157166,
      "learning_rate": 1.0416666666666668e-05,
      "loss": 1.0525,
      "step": 330
    },
    {
      "epoch": 0.10451531417745501,
      "grad_norm": 0.38220885396003723,
      "learning_rate": 1.0448232323232323e-05,
      "loss": 1.0746,
      "step": 331
    },
    {
      "epoch": 0.10483107041364068,
      "grad_norm": 0.49964603781700134,
      "learning_rate": 1.047979797979798e-05,
      "loss": 1.0423,
      "step": 332
    },
    {
      "epoch": 0.10514682664982633,
      "grad_norm": 0.4543418884277344,
      "learning_rate": 1.0511363636363637e-05,
      "loss": 1.0534,
      "step": 333
    },
    {
      "epoch": 0.105462582886012,
      "grad_norm": 0.2884381115436554,
      "learning_rate": 1.0542929292929294e-05,
      "loss": 1.0491,
      "step": 334
    },
    {
      "epoch": 0.10577833912219767,
      "grad_norm": 0.43001261353492737,
      "learning_rate": 1.0574494949494949e-05,
      "loss": 1.0531,
      "step": 335
    },
    {
      "epoch": 0.10609409535838332,
      "grad_norm": 0.5152217745780945,
      "learning_rate": 1.0606060606060607e-05,
      "loss": 1.0588,
      "step": 336
    },
    {
      "epoch": 0.10640985159456899,
      "grad_norm": 0.31784680485725403,
      "learning_rate": 1.0637626262626262e-05,
      "loss": 1.0492,
      "step": 337
    },
    {
      "epoch": 0.10672560783075466,
      "grad_norm": 0.672435462474823,
      "learning_rate": 1.066919191919192e-05,
      "loss": 1.0723,
      "step": 338
    },
    {
      "epoch": 0.10704136406694033,
      "grad_norm": 0.4168337881565094,
      "learning_rate": 1.0700757575757576e-05,
      "loss": 1.0642,
      "step": 339
    },
    {
      "epoch": 0.10735712030312598,
      "grad_norm": 0.3587508797645569,
      "learning_rate": 1.0732323232323233e-05,
      "loss": 1.0608,
      "step": 340
    },
    {
      "epoch": 0.10767287653931165,
      "grad_norm": 0.47398513555526733,
      "learning_rate": 1.076388888888889e-05,
      "loss": 1.0536,
      "step": 341
    },
    {
      "epoch": 0.10798863277549732,
      "grad_norm": 0.387988805770874,
      "learning_rate": 1.0795454545454547e-05,
      "loss": 1.0629,
      "step": 342
    },
    {
      "epoch": 0.10830438901168297,
      "grad_norm": 0.37402069568634033,
      "learning_rate": 1.0827020202020202e-05,
      "loss": 1.0536,
      "step": 343
    },
    {
      "epoch": 0.10862014524786864,
      "grad_norm": 0.36246609687805176,
      "learning_rate": 1.085858585858586e-05,
      "loss": 1.0503,
      "step": 344
    },
    {
      "epoch": 0.10893590148405431,
      "grad_norm": 0.2906649708747864,
      "learning_rate": 1.0890151515151515e-05,
      "loss": 1.051,
      "step": 345
    },
    {
      "epoch": 0.10925165772023998,
      "grad_norm": 0.38393262028694153,
      "learning_rate": 1.0921717171717172e-05,
      "loss": 1.0634,
      "step": 346
    },
    {
      "epoch": 0.10956741395642564,
      "grad_norm": 0.4485388398170471,
      "learning_rate": 1.095328282828283e-05,
      "loss": 1.0418,
      "step": 347
    },
    {
      "epoch": 0.1098831701926113,
      "grad_norm": 0.2958104908466339,
      "learning_rate": 1.0984848484848486e-05,
      "loss": 1.0502,
      "step": 348
    },
    {
      "epoch": 0.11019892642879697,
      "grad_norm": 0.3807732164859772,
      "learning_rate": 1.1016414141414141e-05,
      "loss": 1.0396,
      "step": 349
    },
    {
      "epoch": 0.11051468266498263,
      "grad_norm": 0.3919319212436676,
      "learning_rate": 1.10479797979798e-05,
      "loss": 1.0583,
      "step": 350
    },
    {
      "epoch": 0.1108304389011683,
      "grad_norm": 0.21486066281795502,
      "learning_rate": 1.1079545454545455e-05,
      "loss": 1.0566,
      "step": 351
    },
    {
      "epoch": 0.11114619513735396,
      "grad_norm": 0.5642127394676208,
      "learning_rate": 1.1111111111111112e-05,
      "loss": 1.0604,
      "step": 352
    },
    {
      "epoch": 0.11146195137353963,
      "grad_norm": 0.47256481647491455,
      "learning_rate": 1.1142676767676768e-05,
      "loss": 1.0562,
      "step": 353
    },
    {
      "epoch": 0.11177770760972529,
      "grad_norm": 0.3310910761356354,
      "learning_rate": 1.1174242424242425e-05,
      "loss": 1.045,
      "step": 354
    },
    {
      "epoch": 0.11209346384591096,
      "grad_norm": 0.30530717968940735,
      "learning_rate": 1.1205808080808082e-05,
      "loss": 1.0551,
      "step": 355
    },
    {
      "epoch": 0.11240922008209663,
      "grad_norm": 0.5124827027320862,
      "learning_rate": 1.1237373737373737e-05,
      "loss": 1.0616,
      "step": 356
    },
    {
      "epoch": 0.11272497631828228,
      "grad_norm": 0.579778790473938,
      "learning_rate": 1.1268939393939394e-05,
      "loss": 1.0485,
      "step": 357
    },
    {
      "epoch": 0.11304073255446795,
      "grad_norm": 0.35788270831108093,
      "learning_rate": 1.1300505050505051e-05,
      "loss": 1.0654,
      "step": 358
    },
    {
      "epoch": 0.11335648879065362,
      "grad_norm": 0.29750555753707886,
      "learning_rate": 1.1332070707070708e-05,
      "loss": 1.0556,
      "step": 359
    },
    {
      "epoch": 0.11367224502683929,
      "grad_norm": 0.5538851618766785,
      "learning_rate": 1.1363636363636365e-05,
      "loss": 1.0654,
      "step": 360
    },
    {
      "epoch": 0.11398800126302494,
      "grad_norm": 0.39315474033355713,
      "learning_rate": 1.1395202020202021e-05,
      "loss": 1.063,
      "step": 361
    },
    {
      "epoch": 0.11430375749921061,
      "grad_norm": 0.3503830134868622,
      "learning_rate": 1.1426767676767677e-05,
      "loss": 1.0479,
      "step": 362
    },
    {
      "epoch": 0.11461951373539628,
      "grad_norm": 0.3693561553955078,
      "learning_rate": 1.1458333333333333e-05,
      "loss": 1.045,
      "step": 363
    },
    {
      "epoch": 0.11493526997158193,
      "grad_norm": 0.46076834201812744,
      "learning_rate": 1.148989898989899e-05,
      "loss": 1.0546,
      "step": 364
    },
    {
      "epoch": 0.1152510262077676,
      "grad_norm": 0.29400455951690674,
      "learning_rate": 1.1521464646464647e-05,
      "loss": 1.0509,
      "step": 365
    },
    {
      "epoch": 0.11556678244395327,
      "grad_norm": 0.42453163862228394,
      "learning_rate": 1.1553030303030302e-05,
      "loss": 1.053,
      "step": 366
    },
    {
      "epoch": 0.11588253868013894,
      "grad_norm": 0.3482650816440582,
      "learning_rate": 1.158459595959596e-05,
      "loss": 1.0518,
      "step": 367
    },
    {
      "epoch": 0.1161982949163246,
      "grad_norm": 0.37426677346229553,
      "learning_rate": 1.1616161616161616e-05,
      "loss": 1.0525,
      "step": 368
    },
    {
      "epoch": 0.11651405115251026,
      "grad_norm": 0.4312131404876709,
      "learning_rate": 1.1647727272727273e-05,
      "loss": 1.0414,
      "step": 369
    },
    {
      "epoch": 0.11682980738869593,
      "grad_norm": 0.43302181363105774,
      "learning_rate": 1.167929292929293e-05,
      "loss": 1.0583,
      "step": 370
    },
    {
      "epoch": 0.11714556362488159,
      "grad_norm": 0.37830084562301636,
      "learning_rate": 1.1710858585858586e-05,
      "loss": 1.0557,
      "step": 371
    },
    {
      "epoch": 0.11746131986106725,
      "grad_norm": 0.36770448088645935,
      "learning_rate": 1.1742424242424243e-05,
      "loss": 1.0588,
      "step": 372
    },
    {
      "epoch": 0.11777707609725292,
      "grad_norm": 0.5946649312973022,
      "learning_rate": 1.17739898989899e-05,
      "loss": 1.0192,
      "step": 373
    },
    {
      "epoch": 0.11809283233343859,
      "grad_norm": 0.4018798768520355,
      "learning_rate": 1.1805555555555555e-05,
      "loss": 1.0533,
      "step": 374
    },
    {
      "epoch": 0.11840858856962425,
      "grad_norm": 0.4252590835094452,
      "learning_rate": 1.1837121212121214e-05,
      "loss": 1.0453,
      "step": 375
    },
    {
      "epoch": 0.11872434480580991,
      "grad_norm": 0.5019158124923706,
      "learning_rate": 1.1868686868686869e-05,
      "loss": 1.0427,
      "step": 376
    },
    {
      "epoch": 0.11904010104199558,
      "grad_norm": 0.3181716501712799,
      "learning_rate": 1.1900252525252526e-05,
      "loss": 1.0571,
      "step": 377
    },
    {
      "epoch": 0.11935585727818124,
      "grad_norm": 0.3879508674144745,
      "learning_rate": 1.1931818181818183e-05,
      "loss": 1.0579,
      "step": 378
    },
    {
      "epoch": 0.1196716135143669,
      "grad_norm": 0.3472517430782318,
      "learning_rate": 1.196338383838384e-05,
      "loss": 1.0644,
      "step": 379
    },
    {
      "epoch": 0.11998736975055257,
      "grad_norm": 0.3265877962112427,
      "learning_rate": 1.1994949494949495e-05,
      "loss": 1.0452,
      "step": 380
    },
    {
      "epoch": 0.12030312598673824,
      "grad_norm": 0.43720972537994385,
      "learning_rate": 1.2026515151515153e-05,
      "loss": 1.0464,
      "step": 381
    },
    {
      "epoch": 0.1206188822229239,
      "grad_norm": 0.418163537979126,
      "learning_rate": 1.2058080808080808e-05,
      "loss": 1.0697,
      "step": 382
    },
    {
      "epoch": 0.12093463845910957,
      "grad_norm": 0.7270564436912537,
      "learning_rate": 1.2089646464646465e-05,
      "loss": 1.0569,
      "step": 383
    },
    {
      "epoch": 0.12125039469529524,
      "grad_norm": 0.3281802833080292,
      "learning_rate": 1.2121212121212122e-05,
      "loss": 1.0396,
      "step": 384
    },
    {
      "epoch": 0.12156615093148089,
      "grad_norm": 0.5725803971290588,
      "learning_rate": 1.2152777777777779e-05,
      "loss": 1.0206,
      "step": 385
    },
    {
      "epoch": 0.12188190716766656,
      "grad_norm": 0.3543609380722046,
      "learning_rate": 1.2184343434343434e-05,
      "loss": 1.033,
      "step": 386
    },
    {
      "epoch": 0.12219766340385223,
      "grad_norm": 0.42354536056518555,
      "learning_rate": 1.2215909090909092e-05,
      "loss": 1.0498,
      "step": 387
    },
    {
      "epoch": 0.1225134196400379,
      "grad_norm": 0.3691798746585846,
      "learning_rate": 1.2247474747474748e-05,
      "loss": 1.0408,
      "step": 388
    },
    {
      "epoch": 0.12282917587622355,
      "grad_norm": 0.34293627738952637,
      "learning_rate": 1.2279040404040406e-05,
      "loss": 1.0671,
      "step": 389
    },
    {
      "epoch": 0.12314493211240922,
      "grad_norm": 0.33113282918930054,
      "learning_rate": 1.2310606060606061e-05,
      "loss": 1.043,
      "step": 390
    },
    {
      "epoch": 0.12346068834859489,
      "grad_norm": 0.7703113555908203,
      "learning_rate": 1.2342171717171718e-05,
      "loss": 1.0423,
      "step": 391
    },
    {
      "epoch": 0.12377644458478054,
      "grad_norm": 0.57448410987854,
      "learning_rate": 1.2373737373737375e-05,
      "loss": 1.0355,
      "step": 392
    },
    {
      "epoch": 0.12409220082096621,
      "grad_norm": 0.43269726634025574,
      "learning_rate": 1.2405303030303032e-05,
      "loss": 1.0451,
      "step": 393
    },
    {
      "epoch": 0.12440795705715188,
      "grad_norm": 0.32263505458831787,
      "learning_rate": 1.2436868686868687e-05,
      "loss": 1.0482,
      "step": 394
    },
    {
      "epoch": 0.12472371329333755,
      "grad_norm": 0.5135378837585449,
      "learning_rate": 1.2468434343434344e-05,
      "loss": 1.037,
      "step": 395
    },
    {
      "epoch": 0.12503946952952322,
      "grad_norm": 0.7390228509902954,
      "learning_rate": 1.25e-05,
      "loss": 1.0493,
      "step": 396
    },
    {
      "epoch": 0.12535522576570887,
      "grad_norm": 0.5005096793174744,
      "learning_rate": 1.2531565656565657e-05,
      "loss": 1.0334,
      "step": 397
    },
    {
      "epoch": 0.12567098200189453,
      "grad_norm": 0.3473246097564697,
      "learning_rate": 1.2563131313131313e-05,
      "loss": 1.0719,
      "step": 398
    },
    {
      "epoch": 0.1259867382380802,
      "grad_norm": 0.6417760252952576,
      "learning_rate": 1.259469696969697e-05,
      "loss": 1.0536,
      "step": 399
    },
    {
      "epoch": 0.12630249447426586,
      "grad_norm": 0.8617026805877686,
      "learning_rate": 1.2626262626262628e-05,
      "loss": 1.0536,
      "step": 400
    },
    {
      "epoch": 0.12630249447426586,
      "eval_loss": 1.0394976139068604,
      "eval_runtime": 4.5478,
      "eval_samples_per_second": 547.52,
      "eval_steps_per_second": 4.618,
      "step": 400
    },
    {
      "epoch": 0.12661825071045152,
      "grad_norm": 0.36801549792289734,
      "learning_rate": 1.2657828282828283e-05,
      "loss": 1.0335,
      "step": 401
    },
    {
      "epoch": 0.1269340069466372,
      "grad_norm": 0.3151480257511139,
      "learning_rate": 1.268939393939394e-05,
      "loss": 1.0468,
      "step": 402
    },
    {
      "epoch": 0.12724976318282286,
      "grad_norm": 0.4091816842556,
      "learning_rate": 1.2720959595959597e-05,
      "loss": 1.0552,
      "step": 403
    },
    {
      "epoch": 0.12756551941900854,
      "grad_norm": 0.5123893022537231,
      "learning_rate": 1.2752525252525252e-05,
      "loss": 1.0299,
      "step": 404
    },
    {
      "epoch": 0.1278812756551942,
      "grad_norm": 0.3195244371891022,
      "learning_rate": 1.2784090909090909e-05,
      "loss": 1.0485,
      "step": 405
    },
    {
      "epoch": 0.12819703189137985,
      "grad_norm": 0.39171603322029114,
      "learning_rate": 1.2815656565656567e-05,
      "loss": 1.0542,
      "step": 406
    },
    {
      "epoch": 0.12851278812756553,
      "grad_norm": 0.32408589124679565,
      "learning_rate": 1.2847222222222222e-05,
      "loss": 1.0486,
      "step": 407
    },
    {
      "epoch": 0.12882854436375119,
      "grad_norm": 0.30502820014953613,
      "learning_rate": 1.287878787878788e-05,
      "loss": 1.0259,
      "step": 408
    },
    {
      "epoch": 0.12914430059993684,
      "grad_norm": 0.5003529191017151,
      "learning_rate": 1.2910353535353536e-05,
      "loss": 1.0695,
      "step": 409
    },
    {
      "epoch": 0.12946005683612252,
      "grad_norm": 0.40270179510116577,
      "learning_rate": 1.2941919191919191e-05,
      "loss": 1.051,
      "step": 410
    },
    {
      "epoch": 0.12977581307230818,
      "grad_norm": 0.5134631991386414,
      "learning_rate": 1.2973484848484848e-05,
      "loss": 1.0337,
      "step": 411
    },
    {
      "epoch": 0.13009156930849383,
      "grad_norm": 0.34318897128105164,
      "learning_rate": 1.3005050505050507e-05,
      "loss": 1.025,
      "step": 412
    },
    {
      "epoch": 0.13040732554467951,
      "grad_norm": 0.34859728813171387,
      "learning_rate": 1.3036616161616163e-05,
      "loss": 1.0597,
      "step": 413
    },
    {
      "epoch": 0.13072308178086517,
      "grad_norm": 0.2708737552165985,
      "learning_rate": 1.3068181818181819e-05,
      "loss": 1.0542,
      "step": 414
    },
    {
      "epoch": 0.13103883801705082,
      "grad_norm": 0.6029620170593262,
      "learning_rate": 1.3099747474747475e-05,
      "loss": 1.0639,
      "step": 415
    },
    {
      "epoch": 0.1313545942532365,
      "grad_norm": 0.5439067482948303,
      "learning_rate": 1.3131313131313134e-05,
      "loss": 1.0534,
      "step": 416
    },
    {
      "epoch": 0.13167035048942216,
      "grad_norm": 0.4571759104728699,
      "learning_rate": 1.3162878787878787e-05,
      "loss": 1.0368,
      "step": 417
    },
    {
      "epoch": 0.13198610672560784,
      "grad_norm": 0.3394652009010315,
      "learning_rate": 1.3194444444444446e-05,
      "loss": 1.068,
      "step": 418
    },
    {
      "epoch": 0.1323018629617935,
      "grad_norm": 0.7823987603187561,
      "learning_rate": 1.3226010101010103e-05,
      "loss": 1.0451,
      "step": 419
    },
    {
      "epoch": 0.13261761919797915,
      "grad_norm": 0.3505658507347107,
      "learning_rate": 1.3257575757575758e-05,
      "loss": 1.0413,
      "step": 420
    },
    {
      "epoch": 0.13293337543416484,
      "grad_norm": 0.42920446395874023,
      "learning_rate": 1.3289141414141415e-05,
      "loss": 1.0618,
      "step": 421
    },
    {
      "epoch": 0.1332491316703505,
      "grad_norm": 0.4289526641368866,
      "learning_rate": 1.3320707070707073e-05,
      "loss": 1.0566,
      "step": 422
    },
    {
      "epoch": 0.13356488790653614,
      "grad_norm": 0.59524005651474,
      "learning_rate": 1.3352272727272727e-05,
      "loss": 1.0405,
      "step": 423
    },
    {
      "epoch": 0.13388064414272183,
      "grad_norm": 0.33708134293556213,
      "learning_rate": 1.3383838383838385e-05,
      "loss": 1.0562,
      "step": 424
    },
    {
      "epoch": 0.13419640037890748,
      "grad_norm": 0.4750763773918152,
      "learning_rate": 1.3415404040404042e-05,
      "loss": 1.044,
      "step": 425
    },
    {
      "epoch": 0.13451215661509314,
      "grad_norm": 0.7542089819908142,
      "learning_rate": 1.3446969696969697e-05,
      "loss": 1.0441,
      "step": 426
    },
    {
      "epoch": 0.13482791285127882,
      "grad_norm": 0.42671942710876465,
      "learning_rate": 1.3478535353535354e-05,
      "loss": 1.0733,
      "step": 427
    },
    {
      "epoch": 0.13514366908746447,
      "grad_norm": 1.1264578104019165,
      "learning_rate": 1.3510101010101011e-05,
      "loss": 1.0507,
      "step": 428
    },
    {
      "epoch": 0.13545942532365013,
      "grad_norm": 0.6955666542053223,
      "learning_rate": 1.3541666666666666e-05,
      "loss": 1.0379,
      "step": 429
    },
    {
      "epoch": 0.1357751815598358,
      "grad_norm": 0.3626551628112793,
      "learning_rate": 1.3573232323232325e-05,
      "loss": 1.0539,
      "step": 430
    },
    {
      "epoch": 0.13609093779602147,
      "grad_norm": 0.4131193161010742,
      "learning_rate": 1.3604797979797981e-05,
      "loss": 1.0424,
      "step": 431
    },
    {
      "epoch": 0.13640669403220715,
      "grad_norm": 0.4353424310684204,
      "learning_rate": 1.3636363636363637e-05,
      "loss": 1.0487,
      "step": 432
    },
    {
      "epoch": 0.1367224502683928,
      "grad_norm": 0.38123956322669983,
      "learning_rate": 1.3667929292929293e-05,
      "loss": 1.0625,
      "step": 433
    },
    {
      "epoch": 0.13703820650457846,
      "grad_norm": 0.415041983127594,
      "learning_rate": 1.369949494949495e-05,
      "loss": 1.0399,
      "step": 434
    },
    {
      "epoch": 0.13735396274076414,
      "grad_norm": 0.3667745292186737,
      "learning_rate": 1.3731060606060605e-05,
      "loss": 1.0332,
      "step": 435
    },
    {
      "epoch": 0.1376697189769498,
      "grad_norm": 0.31006094813346863,
      "learning_rate": 1.3762626262626264e-05,
      "loss": 1.0481,
      "step": 436
    },
    {
      "epoch": 0.13798547521313545,
      "grad_norm": 0.3369653522968292,
      "learning_rate": 1.379419191919192e-05,
      "loss": 1.0556,
      "step": 437
    },
    {
      "epoch": 0.13830123144932113,
      "grad_norm": 0.5227720141410828,
      "learning_rate": 1.3825757575757576e-05,
      "loss": 1.0567,
      "step": 438
    },
    {
      "epoch": 0.1386169876855068,
      "grad_norm": 0.30393534898757935,
      "learning_rate": 1.3857323232323233e-05,
      "loss": 1.0355,
      "step": 439
    },
    {
      "epoch": 0.13893274392169244,
      "grad_norm": 0.25645992159843445,
      "learning_rate": 1.388888888888889e-05,
      "loss": 1.0518,
      "step": 440
    },
    {
      "epoch": 0.13924850015787812,
      "grad_norm": 0.4034506678581238,
      "learning_rate": 1.3920454545454545e-05,
      "loss": 1.0354,
      "step": 441
    },
    {
      "epoch": 0.13956425639406378,
      "grad_norm": 0.5188763737678528,
      "learning_rate": 1.3952020202020202e-05,
      "loss": 1.0367,
      "step": 442
    },
    {
      "epoch": 0.13988001263024943,
      "grad_norm": 0.4719993770122528,
      "learning_rate": 1.398358585858586e-05,
      "loss": 1.0569,
      "step": 443
    },
    {
      "epoch": 0.14019576886643512,
      "grad_norm": 0.5680590867996216,
      "learning_rate": 1.4015151515151515e-05,
      "loss": 1.0405,
      "step": 444
    },
    {
      "epoch": 0.14051152510262077,
      "grad_norm": 0.4892265796661377,
      "learning_rate": 1.4046717171717172e-05,
      "loss": 1.0549,
      "step": 445
    },
    {
      "epoch": 0.14082728133880645,
      "grad_norm": 0.32585158944129944,
      "learning_rate": 1.4078282828282829e-05,
      "loss": 1.0345,
      "step": 446
    },
    {
      "epoch": 0.1411430375749921,
      "grad_norm": 0.49924561381340027,
      "learning_rate": 1.4109848484848487e-05,
      "loss": 1.051,
      "step": 447
    },
    {
      "epoch": 0.14145879381117776,
      "grad_norm": 0.4428628087043762,
      "learning_rate": 1.4141414141414141e-05,
      "loss": 1.0225,
      "step": 448
    },
    {
      "epoch": 0.14177455004736345,
      "grad_norm": 0.7777857184410095,
      "learning_rate": 1.41729797979798e-05,
      "loss": 1.0354,
      "step": 449
    },
    {
      "epoch": 0.1420903062835491,
      "grad_norm": 0.49991992115974426,
      "learning_rate": 1.4204545454545456e-05,
      "loss": 1.0511,
      "step": 450
    },
    {
      "epoch": 0.14240606251973476,
      "grad_norm": 0.48333653807640076,
      "learning_rate": 1.4236111111111111e-05,
      "loss": 1.0344,
      "step": 451
    },
    {
      "epoch": 0.14272181875592044,
      "grad_norm": 0.5985662937164307,
      "learning_rate": 1.4267676767676768e-05,
      "loss": 1.0316,
      "step": 452
    },
    {
      "epoch": 0.1430375749921061,
      "grad_norm": 0.49024084210395813,
      "learning_rate": 1.4299242424242427e-05,
      "loss": 1.0488,
      "step": 453
    },
    {
      "epoch": 0.14335333122829175,
      "grad_norm": 0.47889405488967896,
      "learning_rate": 1.433080808080808e-05,
      "loss": 1.0783,
      "step": 454
    },
    {
      "epoch": 0.14366908746447743,
      "grad_norm": 0.6387469172477722,
      "learning_rate": 1.4362373737373739e-05,
      "loss": 1.0432,
      "step": 455
    },
    {
      "epoch": 0.14398484370066308,
      "grad_norm": 0.5799915194511414,
      "learning_rate": 1.4393939393939396e-05,
      "loss": 1.0513,
      "step": 456
    },
    {
      "epoch": 0.14430059993684874,
      "grad_norm": 0.5389712452888489,
      "learning_rate": 1.442550505050505e-05,
      "loss": 1.0382,
      "step": 457
    },
    {
      "epoch": 0.14461635617303442,
      "grad_norm": 0.4852917492389679,
      "learning_rate": 1.4457070707070708e-05,
      "loss": 1.0469,
      "step": 458
    },
    {
      "epoch": 0.14493211240922008,
      "grad_norm": 0.44415175914764404,
      "learning_rate": 1.4488636363636366e-05,
      "loss": 1.0403,
      "step": 459
    },
    {
      "epoch": 0.14524786864540576,
      "grad_norm": 0.47853776812553406,
      "learning_rate": 1.452020202020202e-05,
      "loss": 1.0434,
      "step": 460
    },
    {
      "epoch": 0.1455636248815914,
      "grad_norm": 0.4892577528953552,
      "learning_rate": 1.4551767676767678e-05,
      "loss": 1.0216,
      "step": 461
    },
    {
      "epoch": 0.14587938111777707,
      "grad_norm": 0.7873638272285461,
      "learning_rate": 1.4583333333333335e-05,
      "loss": 1.0409,
      "step": 462
    },
    {
      "epoch": 0.14619513735396275,
      "grad_norm": 0.4818909466266632,
      "learning_rate": 1.461489898989899e-05,
      "loss": 1.0626,
      "step": 463
    },
    {
      "epoch": 0.1465108935901484,
      "grad_norm": 0.49495089054107666,
      "learning_rate": 1.4646464646464647e-05,
      "loss": 1.0329,
      "step": 464
    },
    {
      "epoch": 0.14682664982633406,
      "grad_norm": 0.6725126504898071,
      "learning_rate": 1.4678030303030305e-05,
      "loss": 1.0546,
      "step": 465
    },
    {
      "epoch": 0.14714240606251974,
      "grad_norm": 0.4467467665672302,
      "learning_rate": 1.4709595959595959e-05,
      "loss": 1.0398,
      "step": 466
    },
    {
      "epoch": 0.1474581622987054,
      "grad_norm": 0.7707549333572388,
      "learning_rate": 1.4741161616161617e-05,
      "loss": 1.0683,
      "step": 467
    },
    {
      "epoch": 0.14777391853489105,
      "grad_norm": 0.5757113695144653,
      "learning_rate": 1.4772727272727274e-05,
      "loss": 1.028,
      "step": 468
    },
    {
      "epoch": 0.14808967477107673,
      "grad_norm": 0.49514761567115784,
      "learning_rate": 1.480429292929293e-05,
      "loss": 1.0331,
      "step": 469
    },
    {
      "epoch": 0.1484054310072624,
      "grad_norm": 0.4288766384124756,
      "learning_rate": 1.4835858585858586e-05,
      "loss": 1.0357,
      "step": 470
    },
    {
      "epoch": 0.14872118724344807,
      "grad_norm": 0.4515255093574524,
      "learning_rate": 1.4867424242424243e-05,
      "loss": 1.0676,
      "step": 471
    },
    {
      "epoch": 0.14903694347963373,
      "grad_norm": 0.5876258015632629,
      "learning_rate": 1.4898989898989898e-05,
      "loss": 1.0416,
      "step": 472
    },
    {
      "epoch": 0.14935269971581938,
      "grad_norm": 0.3368344008922577,
      "learning_rate": 1.4930555555555557e-05,
      "loss": 1.0427,
      "step": 473
    },
    {
      "epoch": 0.14966845595200506,
      "grad_norm": 0.4572298526763916,
      "learning_rate": 1.4962121212121214e-05,
      "loss": 1.0749,
      "step": 474
    },
    {
      "epoch": 0.14998421218819072,
      "grad_norm": 0.39081454277038574,
      "learning_rate": 1.4993686868686869e-05,
      "loss": 1.0383,
      "step": 475
    },
    {
      "epoch": 0.15029996842437637,
      "grad_norm": 0.5336904525756836,
      "learning_rate": 1.5025252525252526e-05,
      "loss": 1.0463,
      "step": 476
    },
    {
      "epoch": 0.15061572466056206,
      "grad_norm": 0.5760332942008972,
      "learning_rate": 1.5056818181818182e-05,
      "loss": 1.0344,
      "step": 477
    },
    {
      "epoch": 0.1509314808967477,
      "grad_norm": 0.2996172606945038,
      "learning_rate": 1.5088383838383837e-05,
      "loss": 1.0479,
      "step": 478
    },
    {
      "epoch": 0.15124723713293337,
      "grad_norm": 0.30821865797042847,
      "learning_rate": 1.5119949494949496e-05,
      "loss": 1.054,
      "step": 479
    },
    {
      "epoch": 0.15156299336911905,
      "grad_norm": 0.4483844041824341,
      "learning_rate": 1.5151515151515153e-05,
      "loss": 1.0284,
      "step": 480
    },
    {
      "epoch": 0.1518787496053047,
      "grad_norm": 0.49842318892478943,
      "learning_rate": 1.518308080808081e-05,
      "loss": 1.036,
      "step": 481
    },
    {
      "epoch": 0.15219450584149036,
      "grad_norm": 0.48233309388160706,
      "learning_rate": 1.5214646464646465e-05,
      "loss": 1.0642,
      "step": 482
    },
    {
      "epoch": 0.15251026207767604,
      "grad_norm": 0.43348440527915955,
      "learning_rate": 1.5246212121212122e-05,
      "loss": 1.0395,
      "step": 483
    },
    {
      "epoch": 0.1528260183138617,
      "grad_norm": 0.6552442312240601,
      "learning_rate": 1.527777777777778e-05,
      "loss": 1.0267,
      "step": 484
    },
    {
      "epoch": 0.15314177455004738,
      "grad_norm": 0.4281085431575775,
      "learning_rate": 1.5309343434343435e-05,
      "loss": 1.0473,
      "step": 485
    },
    {
      "epoch": 0.15345753078623303,
      "grad_norm": 0.32992470264434814,
      "learning_rate": 1.534090909090909e-05,
      "loss": 1.0577,
      "step": 486
    },
    {
      "epoch": 0.1537732870224187,
      "grad_norm": 0.47376716136932373,
      "learning_rate": 1.537247474747475e-05,
      "loss": 1.0313,
      "step": 487
    },
    {
      "epoch": 0.15408904325860437,
      "grad_norm": 0.3912649154663086,
      "learning_rate": 1.5404040404040404e-05,
      "loss": 1.0418,
      "step": 488
    },
    {
      "epoch": 0.15440479949479002,
      "grad_norm": 0.5113877058029175,
      "learning_rate": 1.5435606060606063e-05,
      "loss": 1.0355,
      "step": 489
    },
    {
      "epoch": 0.15472055573097568,
      "grad_norm": 0.3784676492214203,
      "learning_rate": 1.5467171717171718e-05,
      "loss": 1.0451,
      "step": 490
    },
    {
      "epoch": 0.15503631196716136,
      "grad_norm": 0.2847704291343689,
      "learning_rate": 1.5498737373737373e-05,
      "loss": 1.0681,
      "step": 491
    },
    {
      "epoch": 0.15535206820334702,
      "grad_norm": 0.4957672655582428,
      "learning_rate": 1.553030303030303e-05,
      "loss": 1.0615,
      "step": 492
    },
    {
      "epoch": 0.15566782443953267,
      "grad_norm": 0.785405695438385,
      "learning_rate": 1.556186868686869e-05,
      "loss": 1.0291,
      "step": 493
    },
    {
      "epoch": 0.15598358067571835,
      "grad_norm": 0.49169519543647766,
      "learning_rate": 1.5593434343434342e-05,
      "loss": 1.031,
      "step": 494
    },
    {
      "epoch": 0.156299336911904,
      "grad_norm": 0.7215877771377563,
      "learning_rate": 1.5625e-05,
      "loss": 1.0217,
      "step": 495
    },
    {
      "epoch": 0.15661509314808966,
      "grad_norm": 0.6813948154449463,
      "learning_rate": 1.565656565656566e-05,
      "loss": 1.0433,
      "step": 496
    },
    {
      "epoch": 0.15693084938427534,
      "grad_norm": 0.40172573924064636,
      "learning_rate": 1.5688131313131314e-05,
      "loss": 1.0233,
      "step": 497
    },
    {
      "epoch": 0.157246605620461,
      "grad_norm": 0.44223684072494507,
      "learning_rate": 1.571969696969697e-05,
      "loss": 1.0454,
      "step": 498
    },
    {
      "epoch": 0.15756236185664668,
      "grad_norm": 0.3425401747226715,
      "learning_rate": 1.5751262626262628e-05,
      "loss": 1.0561,
      "step": 499
    },
    {
      "epoch": 0.15787811809283234,
      "grad_norm": 0.3376587927341461,
      "learning_rate": 1.5782828282828283e-05,
      "loss": 1.0665,
      "step": 500
    },
    {
      "epoch": 0.158193874329018,
      "grad_norm": 0.9137601256370544,
      "learning_rate": 1.581439393939394e-05,
      "loss": 1.0518,
      "step": 501
    },
    {
      "epoch": 0.15850963056520367,
      "grad_norm": 0.37523990869522095,
      "learning_rate": 1.5845959595959596e-05,
      "loss": 1.0503,
      "step": 502
    },
    {
      "epoch": 0.15882538680138933,
      "grad_norm": 0.3874257504940033,
      "learning_rate": 1.587752525252525e-05,
      "loss": 1.0336,
      "step": 503
    },
    {
      "epoch": 0.15914114303757498,
      "grad_norm": 0.6542366147041321,
      "learning_rate": 1.590909090909091e-05,
      "loss": 1.0516,
      "step": 504
    },
    {
      "epoch": 0.15945689927376067,
      "grad_norm": 0.6020170450210571,
      "learning_rate": 1.594065656565657e-05,
      "loss": 1.0592,
      "step": 505
    },
    {
      "epoch": 0.15977265550994632,
      "grad_norm": 0.5171272158622742,
      "learning_rate": 1.597222222222222e-05,
      "loss": 1.0523,
      "step": 506
    },
    {
      "epoch": 0.16008841174613198,
      "grad_norm": 0.44275179505348206,
      "learning_rate": 1.600378787878788e-05,
      "loss": 1.038,
      "step": 507
    },
    {
      "epoch": 0.16040416798231766,
      "grad_norm": 0.5569939613342285,
      "learning_rate": 1.6035353535353538e-05,
      "loss": 1.0252,
      "step": 508
    },
    {
      "epoch": 0.1607199242185033,
      "grad_norm": 0.4791787564754486,
      "learning_rate": 1.6066919191919193e-05,
      "loss": 1.0422,
      "step": 509
    },
    {
      "epoch": 0.16103568045468897,
      "grad_norm": 0.42141303420066833,
      "learning_rate": 1.6098484848484848e-05,
      "loss": 1.0374,
      "step": 510
    },
    {
      "epoch": 0.16135143669087465,
      "grad_norm": 0.5101591944694519,
      "learning_rate": 1.6130050505050506e-05,
      "loss": 1.0378,
      "step": 511
    },
    {
      "epoch": 0.1616671929270603,
      "grad_norm": 0.5211359858512878,
      "learning_rate": 1.6161616161616165e-05,
      "loss": 1.0637,
      "step": 512
    },
    {
      "epoch": 0.161982949163246,
      "grad_norm": 0.5075467228889465,
      "learning_rate": 1.619318181818182e-05,
      "loss": 1.0391,
      "step": 513
    },
    {
      "epoch": 0.16229870539943164,
      "grad_norm": 0.476371705532074,
      "learning_rate": 1.6224747474747475e-05,
      "loss": 1.0663,
      "step": 514
    },
    {
      "epoch": 0.1626144616356173,
      "grad_norm": 0.5135627388954163,
      "learning_rate": 1.6256313131313134e-05,
      "loss": 1.0354,
      "step": 515
    },
    {
      "epoch": 0.16293021787180298,
      "grad_norm": 0.4868716895580292,
      "learning_rate": 1.628787878787879e-05,
      "loss": 1.0393,
      "step": 516
    },
    {
      "epoch": 0.16324597410798863,
      "grad_norm": 0.4947311282157898,
      "learning_rate": 1.6319444444444444e-05,
      "loss": 1.0381,
      "step": 517
    },
    {
      "epoch": 0.1635617303441743,
      "grad_norm": 0.36999017000198364,
      "learning_rate": 1.6351010101010102e-05,
      "loss": 1.0403,
      "step": 518
    },
    {
      "epoch": 0.16387748658035997,
      "grad_norm": 0.4358632266521454,
      "learning_rate": 1.6382575757575758e-05,
      "loss": 1.0465,
      "step": 519
    },
    {
      "epoch": 0.16419324281654563,
      "grad_norm": 0.5774652361869812,
      "learning_rate": 1.6414141414141416e-05,
      "loss": 1.0557,
      "step": 520
    },
    {
      "epoch": 0.16450899905273128,
      "grad_norm": 0.4815928339958191,
      "learning_rate": 1.644570707070707e-05,
      "loss": 1.0468,
      "step": 521
    },
    {
      "epoch": 0.16482475528891696,
      "grad_norm": 0.4809669256210327,
      "learning_rate": 1.6477272727272726e-05,
      "loss": 1.0453,
      "step": 522
    },
    {
      "epoch": 0.16514051152510262,
      "grad_norm": 0.6526656746864319,
      "learning_rate": 1.6508838383838385e-05,
      "loss": 1.0309,
      "step": 523
    },
    {
      "epoch": 0.16545626776128827,
      "grad_norm": 0.4233449399471283,
      "learning_rate": 1.6540404040404044e-05,
      "loss": 1.0294,
      "step": 524
    },
    {
      "epoch": 0.16577202399747396,
      "grad_norm": 0.5478879809379578,
      "learning_rate": 1.6571969696969695e-05,
      "loss": 1.0442,
      "step": 525
    },
    {
      "epoch": 0.1660877802336596,
      "grad_norm": 0.36921337246894836,
      "learning_rate": 1.6603535353535354e-05,
      "loss": 1.0326,
      "step": 526
    },
    {
      "epoch": 0.1664035364698453,
      "grad_norm": 0.5152103304862976,
      "learning_rate": 1.6635101010101012e-05,
      "loss": 1.0444,
      "step": 527
    },
    {
      "epoch": 0.16671929270603095,
      "grad_norm": 0.40243425965309143,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 1.0558,
      "step": 528
    },
    {
      "epoch": 0.1670350489422166,
      "grad_norm": 0.42330047488212585,
      "learning_rate": 1.6698232323232323e-05,
      "loss": 1.0546,
      "step": 529
    },
    {
      "epoch": 0.16735080517840228,
      "grad_norm": 0.3860170841217041,
      "learning_rate": 1.672979797979798e-05,
      "loss": 1.0373,
      "step": 530
    },
    {
      "epoch": 0.16766656141458794,
      "grad_norm": 0.5874581933021545,
      "learning_rate": 1.6761363636363636e-05,
      "loss": 1.0257,
      "step": 531
    },
    {
      "epoch": 0.1679823176507736,
      "grad_norm": 0.6289354562759399,
      "learning_rate": 1.6792929292929295e-05,
      "loss": 1.033,
      "step": 532
    },
    {
      "epoch": 0.16829807388695928,
      "grad_norm": 0.527024507522583,
      "learning_rate": 1.682449494949495e-05,
      "loss": 1.0663,
      "step": 533
    },
    {
      "epoch": 0.16861383012314493,
      "grad_norm": 0.5787068009376526,
      "learning_rate": 1.6856060606060605e-05,
      "loss": 1.039,
      "step": 534
    },
    {
      "epoch": 0.16892958635933059,
      "grad_norm": 0.43615061044692993,
      "learning_rate": 1.6887626262626264e-05,
      "loss": 1.0268,
      "step": 535
    },
    {
      "epoch": 0.16924534259551627,
      "grad_norm": 0.6269041895866394,
      "learning_rate": 1.6919191919191922e-05,
      "loss": 1.0344,
      "step": 536
    },
    {
      "epoch": 0.16956109883170192,
      "grad_norm": 0.645095705986023,
      "learning_rate": 1.6950757575757574e-05,
      "loss": 1.0662,
      "step": 537
    },
    {
      "epoch": 0.16987685506788758,
      "grad_norm": 0.7710626125335693,
      "learning_rate": 1.6982323232323232e-05,
      "loss": 1.0542,
      "step": 538
    },
    {
      "epoch": 0.17019261130407326,
      "grad_norm": 0.4514850676059723,
      "learning_rate": 1.701388888888889e-05,
      "loss": 1.038,
      "step": 539
    },
    {
      "epoch": 0.17050836754025891,
      "grad_norm": 0.3876580595970154,
      "learning_rate": 1.7045454545454546e-05,
      "loss": 1.0394,
      "step": 540
    },
    {
      "epoch": 0.1708241237764446,
      "grad_norm": 0.43514904379844666,
      "learning_rate": 1.70770202020202e-05,
      "loss": 1.0605,
      "step": 541
    },
    {
      "epoch": 0.17113988001263025,
      "grad_norm": 0.5225546956062317,
      "learning_rate": 1.710858585858586e-05,
      "loss": 1.0332,
      "step": 542
    },
    {
      "epoch": 0.1714556362488159,
      "grad_norm": 0.367436021566391,
      "learning_rate": 1.7140151515151515e-05,
      "loss": 1.0481,
      "step": 543
    },
    {
      "epoch": 0.1717713924850016,
      "grad_norm": 0.6229331493377686,
      "learning_rate": 1.7171717171717173e-05,
      "loss": 1.0501,
      "step": 544
    },
    {
      "epoch": 0.17208714872118724,
      "grad_norm": 0.25530779361724854,
      "learning_rate": 1.720328282828283e-05,
      "loss": 1.0403,
      "step": 545
    },
    {
      "epoch": 0.1724029049573729,
      "grad_norm": 0.3894161283969879,
      "learning_rate": 1.7234848484848487e-05,
      "loss": 1.0231,
      "step": 546
    },
    {
      "epoch": 0.17271866119355858,
      "grad_norm": 0.5256360173225403,
      "learning_rate": 1.7266414141414142e-05,
      "loss": 1.0182,
      "step": 547
    },
    {
      "epoch": 0.17303441742974424,
      "grad_norm": 0.8339974284172058,
      "learning_rate": 1.72979797979798e-05,
      "loss": 1.0435,
      "step": 548
    },
    {
      "epoch": 0.1733501736659299,
      "grad_norm": 0.5735340714454651,
      "learning_rate": 1.7329545454545456e-05,
      "loss": 1.0427,
      "step": 549
    },
    {
      "epoch": 0.17366592990211557,
      "grad_norm": 0.3366367220878601,
      "learning_rate": 1.736111111111111e-05,
      "loss": 1.0469,
      "step": 550
    },
    {
      "epoch": 0.17398168613830123,
      "grad_norm": 0.6597113609313965,
      "learning_rate": 1.739267676767677e-05,
      "loss": 1.0514,
      "step": 551
    },
    {
      "epoch": 0.17429744237448688,
      "grad_norm": 0.4907837212085724,
      "learning_rate": 1.7424242424242425e-05,
      "loss": 1.0294,
      "step": 552
    },
    {
      "epoch": 0.17461319861067257,
      "grad_norm": 0.5592122673988342,
      "learning_rate": 1.745580808080808e-05,
      "loss": 1.0195,
      "step": 553
    },
    {
      "epoch": 0.17492895484685822,
      "grad_norm": 0.550363302230835,
      "learning_rate": 1.748737373737374e-05,
      "loss": 1.0192,
      "step": 554
    },
    {
      "epoch": 0.1752447110830439,
      "grad_norm": 0.47156625986099243,
      "learning_rate": 1.7518939393939397e-05,
      "loss": 1.0124,
      "step": 555
    },
    {
      "epoch": 0.17556046731922956,
      "grad_norm": 0.5233810544013977,
      "learning_rate": 1.7550505050505052e-05,
      "loss": 1.0337,
      "step": 556
    },
    {
      "epoch": 0.1758762235554152,
      "grad_norm": 0.5525631904602051,
      "learning_rate": 1.7582070707070707e-05,
      "loss": 1.0269,
      "step": 557
    },
    {
      "epoch": 0.1761919797916009,
      "grad_norm": 0.801319420337677,
      "learning_rate": 1.7613636363636366e-05,
      "loss": 1.0594,
      "step": 558
    },
    {
      "epoch": 0.17650773602778655,
      "grad_norm": 0.4921005666255951,
      "learning_rate": 1.764520202020202e-05,
      "loss": 1.0397,
      "step": 559
    },
    {
      "epoch": 0.1768234922639722,
      "grad_norm": 0.4235406517982483,
      "learning_rate": 1.7676767676767676e-05,
      "loss": 1.0505,
      "step": 560
    },
    {
      "epoch": 0.1771392485001579,
      "grad_norm": 0.91356360912323,
      "learning_rate": 1.7708333333333335e-05,
      "loss": 1.029,
      "step": 561
    },
    {
      "epoch": 0.17745500473634354,
      "grad_norm": 0.46292844414711,
      "learning_rate": 1.773989898989899e-05,
      "loss": 1.0493,
      "step": 562
    },
    {
      "epoch": 0.1777707609725292,
      "grad_norm": 0.6037445664405823,
      "learning_rate": 1.7771464646464648e-05,
      "loss": 1.0337,
      "step": 563
    },
    {
      "epoch": 0.17808651720871488,
      "grad_norm": 0.6743975877761841,
      "learning_rate": 1.7803030303030303e-05,
      "loss": 1.0718,
      "step": 564
    },
    {
      "epoch": 0.17840227344490053,
      "grad_norm": 0.5083600282669067,
      "learning_rate": 1.783459595959596e-05,
      "loss": 1.0555,
      "step": 565
    },
    {
      "epoch": 0.1787180296810862,
      "grad_norm": 0.8833627700805664,
      "learning_rate": 1.7866161616161617e-05,
      "loss": 1.0391,
      "step": 566
    },
    {
      "epoch": 0.17903378591727187,
      "grad_norm": 0.49433302879333496,
      "learning_rate": 1.7897727272727276e-05,
      "loss": 1.0437,
      "step": 567
    },
    {
      "epoch": 0.17934954215345753,
      "grad_norm": 0.6566864848136902,
      "learning_rate": 1.7929292929292927e-05,
      "loss": 1.0339,
      "step": 568
    },
    {
      "epoch": 0.1796652983896432,
      "grad_norm": 0.6048679351806641,
      "learning_rate": 1.7960858585858586e-05,
      "loss": 1.0357,
      "step": 569
    },
    {
      "epoch": 0.17998105462582886,
      "grad_norm": 0.7116847634315491,
      "learning_rate": 1.7992424242424244e-05,
      "loss": 1.0354,
      "step": 570
    },
    {
      "epoch": 0.18029681086201452,
      "grad_norm": 0.4730285108089447,
      "learning_rate": 1.80239898989899e-05,
      "loss": 1.0421,
      "step": 571
    },
    {
      "epoch": 0.1806125670982002,
      "grad_norm": 1.093335509300232,
      "learning_rate": 1.8055555555555555e-05,
      "loss": 1.0534,
      "step": 572
    },
    {
      "epoch": 0.18092832333438585,
      "grad_norm": 0.8109234571456909,
      "learning_rate": 1.8087121212121213e-05,
      "loss": 1.0369,
      "step": 573
    },
    {
      "epoch": 0.1812440795705715,
      "grad_norm": 0.40986865758895874,
      "learning_rate": 1.811868686868687e-05,
      "loss": 1.0468,
      "step": 574
    },
    {
      "epoch": 0.1815598358067572,
      "grad_norm": 0.7870145440101624,
      "learning_rate": 1.8150252525252527e-05,
      "loss": 1.037,
      "step": 575
    },
    {
      "epoch": 0.18187559204294285,
      "grad_norm": 0.49582862854003906,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 1.0519,
      "step": 576
    },
    {
      "epoch": 0.1821913482791285,
      "grad_norm": 0.5248562097549438,
      "learning_rate": 1.8213383838383837e-05,
      "loss": 1.0483,
      "step": 577
    },
    {
      "epoch": 0.18250710451531418,
      "grad_norm": 0.6936292052268982,
      "learning_rate": 1.8244949494949496e-05,
      "loss": 1.0369,
      "step": 578
    },
    {
      "epoch": 0.18282286075149984,
      "grad_norm": 0.5838512182235718,
      "learning_rate": 1.8276515151515154e-05,
      "loss": 1.0428,
      "step": 579
    },
    {
      "epoch": 0.1831386169876855,
      "grad_norm": 0.6108795404434204,
      "learning_rate": 1.830808080808081e-05,
      "loss": 1.0327,
      "step": 580
    },
    {
      "epoch": 0.18345437322387118,
      "grad_norm": 0.5305110216140747,
      "learning_rate": 1.8339646464646465e-05,
      "loss": 1.0285,
      "step": 581
    },
    {
      "epoch": 0.18377012946005683,
      "grad_norm": 0.388097882270813,
      "learning_rate": 1.8371212121212123e-05,
      "loss": 1.0526,
      "step": 582
    },
    {
      "epoch": 0.1840858856962425,
      "grad_norm": 0.34786751866340637,
      "learning_rate": 1.8402777777777778e-05,
      "loss": 1.0473,
      "step": 583
    },
    {
      "epoch": 0.18440164193242817,
      "grad_norm": 0.44363564252853394,
      "learning_rate": 1.8434343434343433e-05,
      "loss": 1.0379,
      "step": 584
    },
    {
      "epoch": 0.18471739816861382,
      "grad_norm": 0.5650203227996826,
      "learning_rate": 1.8465909090909092e-05,
      "loss": 1.038,
      "step": 585
    },
    {
      "epoch": 0.1850331544047995,
      "grad_norm": 0.5220074653625488,
      "learning_rate": 1.849747474747475e-05,
      "loss": 1.0353,
      "step": 586
    },
    {
      "epoch": 0.18534891064098516,
      "grad_norm": 0.49310174584388733,
      "learning_rate": 1.8529040404040406e-05,
      "loss": 1.0261,
      "step": 587
    },
    {
      "epoch": 0.18566466687717081,
      "grad_norm": 0.49452316761016846,
      "learning_rate": 1.856060606060606e-05,
      "loss": 1.0279,
      "step": 588
    },
    {
      "epoch": 0.1859804231133565,
      "grad_norm": 0.8313903212547302,
      "learning_rate": 1.859217171717172e-05,
      "loss": 1.0387,
      "step": 589
    },
    {
      "epoch": 0.18629617934954215,
      "grad_norm": 0.4336547553539276,
      "learning_rate": 1.8623737373737374e-05,
      "loss": 1.0368,
      "step": 590
    },
    {
      "epoch": 0.1866119355857278,
      "grad_norm": 0.5167835354804993,
      "learning_rate": 1.8655303030303033e-05,
      "loss": 1.0477,
      "step": 591
    },
    {
      "epoch": 0.1869276918219135,
      "grad_norm": 0.49694541096687317,
      "learning_rate": 1.8686868686868688e-05,
      "loss": 1.0471,
      "step": 592
    },
    {
      "epoch": 0.18724344805809914,
      "grad_norm": 0.43811362981796265,
      "learning_rate": 1.8718434343434343e-05,
      "loss": 1.0416,
      "step": 593
    },
    {
      "epoch": 0.18755920429428483,
      "grad_norm": 0.6405027508735657,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 1.0603,
      "step": 594
    },
    {
      "epoch": 0.18787496053047048,
      "grad_norm": 0.41925665736198425,
      "learning_rate": 1.8781565656565657e-05,
      "loss": 1.0515,
      "step": 595
    },
    {
      "epoch": 0.18819071676665614,
      "grad_norm": 0.9505752325057983,
      "learning_rate": 1.8813131313131312e-05,
      "loss": 1.0285,
      "step": 596
    },
    {
      "epoch": 0.18850647300284182,
      "grad_norm": 0.4320131242275238,
      "learning_rate": 1.884469696969697e-05,
      "loss": 1.0434,
      "step": 597
    },
    {
      "epoch": 0.18882222923902747,
      "grad_norm": 0.4829607307910919,
      "learning_rate": 1.887626262626263e-05,
      "loss": 1.0335,
      "step": 598
    },
    {
      "epoch": 0.18913798547521313,
      "grad_norm": 0.586982250213623,
      "learning_rate": 1.8907828282828284e-05,
      "loss": 1.0425,
      "step": 599
    },
    {
      "epoch": 0.1894537417113988,
      "grad_norm": 0.5257988572120667,
      "learning_rate": 1.893939393939394e-05,
      "loss": 1.0315,
      "step": 600
    },
    {
      "epoch": 0.1894537417113988,
      "eval_loss": 1.0219939947128296,
      "eval_runtime": 4.5669,
      "eval_samples_per_second": 545.225,
      "eval_steps_per_second": 4.598,
      "step": 600
    },
    {
      "epoch": 0.18976949794758446,
      "grad_norm": 0.3960842788219452,
      "learning_rate": 1.8970959595959598e-05,
      "loss": 1.0386,
      "step": 601
    },
    {
      "epoch": 0.19008525418377012,
      "grad_norm": 0.4422132968902588,
      "learning_rate": 1.9002525252525253e-05,
      "loss": 1.0439,
      "step": 602
    },
    {
      "epoch": 0.1904010104199558,
      "grad_norm": 0.8020049333572388,
      "learning_rate": 1.9034090909090908e-05,
      "loss": 1.0253,
      "step": 603
    },
    {
      "epoch": 0.19071676665614146,
      "grad_norm": 0.5259152054786682,
      "learning_rate": 1.9065656565656567e-05,
      "loss": 1.0186,
      "step": 604
    },
    {
      "epoch": 0.1910325228923271,
      "grad_norm": 0.5782701969146729,
      "learning_rate": 1.9097222222222222e-05,
      "loss": 1.0377,
      "step": 605
    },
    {
      "epoch": 0.1913482791285128,
      "grad_norm": 0.5823735594749451,
      "learning_rate": 1.912878787878788e-05,
      "loss": 1.0358,
      "step": 606
    },
    {
      "epoch": 0.19166403536469845,
      "grad_norm": 0.6772051453590393,
      "learning_rate": 1.9160353535353536e-05,
      "loss": 1.0349,
      "step": 607
    },
    {
      "epoch": 0.19197979160088413,
      "grad_norm": 0.42558956146240234,
      "learning_rate": 1.919191919191919e-05,
      "loss": 1.0624,
      "step": 608
    },
    {
      "epoch": 0.19229554783706979,
      "grad_norm": 0.38739314675331116,
      "learning_rate": 1.922348484848485e-05,
      "loss": 1.0219,
      "step": 609
    },
    {
      "epoch": 0.19261130407325544,
      "grad_norm": 0.5514295697212219,
      "learning_rate": 1.9255050505050508e-05,
      "loss": 1.047,
      "step": 610
    },
    {
      "epoch": 0.19292706030944112,
      "grad_norm": 0.3887917101383209,
      "learning_rate": 1.9286616161616163e-05,
      "loss": 1.0037,
      "step": 611
    },
    {
      "epoch": 0.19324281654562678,
      "grad_norm": 0.9426091909408569,
      "learning_rate": 1.9318181818181818e-05,
      "loss": 1.0525,
      "step": 612
    },
    {
      "epoch": 0.19355857278181243,
      "grad_norm": 0.34266209602355957,
      "learning_rate": 1.9349747474747477e-05,
      "loss": 1.0545,
      "step": 613
    },
    {
      "epoch": 0.19387432901799811,
      "grad_norm": 0.7315279245376587,
      "learning_rate": 1.9381313131313135e-05,
      "loss": 1.0263,
      "step": 614
    },
    {
      "epoch": 0.19419008525418377,
      "grad_norm": 0.42028680443763733,
      "learning_rate": 1.9412878787878787e-05,
      "loss": 1.0136,
      "step": 615
    },
    {
      "epoch": 0.19450584149036942,
      "grad_norm": 0.9084534645080566,
      "learning_rate": 1.9444444444444445e-05,
      "loss": 1.0275,
      "step": 616
    },
    {
      "epoch": 0.1948215977265551,
      "grad_norm": 0.5849639773368835,
      "learning_rate": 1.9476010101010104e-05,
      "loss": 1.0366,
      "step": 617
    },
    {
      "epoch": 0.19513735396274076,
      "grad_norm": 0.4220241904258728,
      "learning_rate": 1.950757575757576e-05,
      "loss": 1.0371,
      "step": 618
    },
    {
      "epoch": 0.19545311019892642,
      "grad_norm": 0.48502182960510254,
      "learning_rate": 1.9539141414141414e-05,
      "loss": 1.0393,
      "step": 619
    },
    {
      "epoch": 0.1957688664351121,
      "grad_norm": 0.6528877019882202,
      "learning_rate": 1.9570707070707073e-05,
      "loss": 1.0271,
      "step": 620
    },
    {
      "epoch": 0.19608462267129775,
      "grad_norm": 0.44090506434440613,
      "learning_rate": 1.9602272727272728e-05,
      "loss": 1.046,
      "step": 621
    },
    {
      "epoch": 0.19640037890748344,
      "grad_norm": 0.3660953640937805,
      "learning_rate": 1.9633838383838386e-05,
      "loss": 1.0347,
      "step": 622
    },
    {
      "epoch": 0.1967161351436691,
      "grad_norm": 0.4680846631526947,
      "learning_rate": 1.966540404040404e-05,
      "loss": 1.0323,
      "step": 623
    },
    {
      "epoch": 0.19703189137985475,
      "grad_norm": 0.48509764671325684,
      "learning_rate": 1.9696969696969697e-05,
      "loss": 1.0283,
      "step": 624
    },
    {
      "epoch": 0.19734764761604043,
      "grad_norm": 0.5510538816452026,
      "learning_rate": 1.9728535353535355e-05,
      "loss": 1.0504,
      "step": 625
    },
    {
      "epoch": 0.19766340385222608,
      "grad_norm": 0.5745561122894287,
      "learning_rate": 1.976010101010101e-05,
      "loss": 1.032,
      "step": 626
    },
    {
      "epoch": 0.19797916008841174,
      "grad_norm": 0.5151652693748474,
      "learning_rate": 1.9791666666666665e-05,
      "loss": 1.0255,
      "step": 627
    },
    {
      "epoch": 0.19829491632459742,
      "grad_norm": 0.7137961983680725,
      "learning_rate": 1.9823232323232324e-05,
      "loss": 1.0336,
      "step": 628
    },
    {
      "epoch": 0.19861067256078307,
      "grad_norm": 0.6261966824531555,
      "learning_rate": 1.9854797979797983e-05,
      "loss": 1.0199,
      "step": 629
    },
    {
      "epoch": 0.19892642879696873,
      "grad_norm": 0.47102004289627075,
      "learning_rate": 1.9886363636363638e-05,
      "loss": 1.0486,
      "step": 630
    },
    {
      "epoch": 0.1992421850331544,
      "grad_norm": 0.7463870644569397,
      "learning_rate": 1.9917929292929293e-05,
      "loss": 1.031,
      "step": 631
    },
    {
      "epoch": 0.19955794126934007,
      "grad_norm": 0.5004166960716248,
      "learning_rate": 1.994949494949495e-05,
      "loss": 1.0246,
      "step": 632
    },
    {
      "epoch": 0.19987369750552572,
      "grad_norm": 0.6242977976799011,
      "learning_rate": 1.9981060606060607e-05,
      "loss": 1.0374,
      "step": 633
    },
    {
      "epoch": 0.2001894537417114,
      "grad_norm": 0.6987716555595398,
      "learning_rate": 2.0012626262626265e-05,
      "loss": 1.0227,
      "step": 634
    },
    {
      "epoch": 0.20050520997789706,
      "grad_norm": 0.5326171517372131,
      "learning_rate": 2.004419191919192e-05,
      "loss": 1.039,
      "step": 635
    },
    {
      "epoch": 0.20082096621408274,
      "grad_norm": 0.673435389995575,
      "learning_rate": 2.0075757575757575e-05,
      "loss": 1.0506,
      "step": 636
    },
    {
      "epoch": 0.2011367224502684,
      "grad_norm": 0.9845566153526306,
      "learning_rate": 2.0107323232323234e-05,
      "loss": 1.0308,
      "step": 637
    },
    {
      "epoch": 0.20145247868645405,
      "grad_norm": 0.5673374533653259,
      "learning_rate": 2.013888888888889e-05,
      "loss": 1.0214,
      "step": 638
    },
    {
      "epoch": 0.20176823492263973,
      "grad_norm": 0.5427038073539734,
      "learning_rate": 2.0170454545454544e-05,
      "loss": 1.0157,
      "step": 639
    },
    {
      "epoch": 0.2020839911588254,
      "grad_norm": 0.5208374261856079,
      "learning_rate": 2.0202020202020203e-05,
      "loss": 1.0246,
      "step": 640
    },
    {
      "epoch": 0.20239974739501104,
      "grad_norm": 0.5898987650871277,
      "learning_rate": 2.023358585858586e-05,
      "loss": 1.0308,
      "step": 641
    },
    {
      "epoch": 0.20271550363119673,
      "grad_norm": 0.4361191689968109,
      "learning_rate": 2.0265151515151516e-05,
      "loss": 1.0203,
      "step": 642
    },
    {
      "epoch": 0.20303125986738238,
      "grad_norm": 0.4544173777103424,
      "learning_rate": 2.029671717171717e-05,
      "loss": 1.0372,
      "step": 643
    },
    {
      "epoch": 0.20334701610356803,
      "grad_norm": 0.816819965839386,
      "learning_rate": 2.032828282828283e-05,
      "loss": 1.0248,
      "step": 644
    },
    {
      "epoch": 0.20366277233975372,
      "grad_norm": 0.5031799077987671,
      "learning_rate": 2.035984848484849e-05,
      "loss": 1.0439,
      "step": 645
    },
    {
      "epoch": 0.20397852857593937,
      "grad_norm": 0.5684875249862671,
      "learning_rate": 2.039141414141414e-05,
      "loss": 1.0071,
      "step": 646
    },
    {
      "epoch": 0.20429428481212503,
      "grad_norm": 0.6083799004554749,
      "learning_rate": 2.04229797979798e-05,
      "loss": 1.0274,
      "step": 647
    },
    {
      "epoch": 0.2046100410483107,
      "grad_norm": 0.636752724647522,
      "learning_rate": 2.0454545454545457e-05,
      "loss": 1.0232,
      "step": 648
    },
    {
      "epoch": 0.20492579728449636,
      "grad_norm": 0.4497145116329193,
      "learning_rate": 2.0486111111111113e-05,
      "loss": 1.0019,
      "step": 649
    },
    {
      "epoch": 0.20524155352068205,
      "grad_norm": 0.43697813153266907,
      "learning_rate": 2.0517676767676768e-05,
      "loss": 1.0352,
      "step": 650
    },
    {
      "epoch": 0.2055573097568677,
      "grad_norm": 0.5192165970802307,
      "learning_rate": 2.0549242424242426e-05,
      "loss": 1.0106,
      "step": 651
    },
    {
      "epoch": 0.20587306599305336,
      "grad_norm": 0.38154444098472595,
      "learning_rate": 2.058080808080808e-05,
      "loss": 1.0289,
      "step": 652
    },
    {
      "epoch": 0.20618882222923904,
      "grad_norm": 0.39409616589546204,
      "learning_rate": 2.061237373737374e-05,
      "loss": 1.0296,
      "step": 653
    },
    {
      "epoch": 0.2065045784654247,
      "grad_norm": 0.5476168394088745,
      "learning_rate": 2.0643939393939395e-05,
      "loss": 1.0484,
      "step": 654
    },
    {
      "epoch": 0.20682033470161035,
      "grad_norm": 0.747313916683197,
      "learning_rate": 2.067550505050505e-05,
      "loss": 1.0349,
      "step": 655
    },
    {
      "epoch": 0.20713609093779603,
      "grad_norm": 0.5035751461982727,
      "learning_rate": 2.070707070707071e-05,
      "loss": 1.039,
      "step": 656
    },
    {
      "epoch": 0.20745184717398168,
      "grad_norm": 0.44524678587913513,
      "learning_rate": 2.0738636363636367e-05,
      "loss": 1.0355,
      "step": 657
    },
    {
      "epoch": 0.20776760341016734,
      "grad_norm": 0.5110377669334412,
      "learning_rate": 2.077020202020202e-05,
      "loss": 1.0091,
      "step": 658
    },
    {
      "epoch": 0.20808335964635302,
      "grad_norm": 0.5772188901901245,
      "learning_rate": 2.0801767676767677e-05,
      "loss": 1.0324,
      "step": 659
    },
    {
      "epoch": 0.20839911588253868,
      "grad_norm": 0.5031735897064209,
      "learning_rate": 2.0833333333333336e-05,
      "loss": 1.0143,
      "step": 660
    },
    {
      "epoch": 0.20871487211872433,
      "grad_norm": 0.5328065752983093,
      "learning_rate": 2.086489898989899e-05,
      "loss": 1.0415,
      "step": 661
    },
    {
      "epoch": 0.20903062835491001,
      "grad_norm": 0.5267192125320435,
      "learning_rate": 2.0896464646464646e-05,
      "loss": 1.0053,
      "step": 662
    },
    {
      "epoch": 0.20934638459109567,
      "grad_norm": 0.6140797734260559,
      "learning_rate": 2.0928030303030305e-05,
      "loss": 1.0308,
      "step": 663
    },
    {
      "epoch": 0.20966214082728135,
      "grad_norm": 0.4642013907432556,
      "learning_rate": 2.095959595959596e-05,
      "loss": 1.0251,
      "step": 664
    },
    {
      "epoch": 0.209977897063467,
      "grad_norm": 0.564565122127533,
      "learning_rate": 2.099116161616162e-05,
      "loss": 1.0064,
      "step": 665
    },
    {
      "epoch": 0.21029365329965266,
      "grad_norm": 0.5513519048690796,
      "learning_rate": 2.1022727272727274e-05,
      "loss": 1.029,
      "step": 666
    },
    {
      "epoch": 0.21060940953583834,
      "grad_norm": 0.6336759328842163,
      "learning_rate": 2.105429292929293e-05,
      "loss": 1.0154,
      "step": 667
    },
    {
      "epoch": 0.210925165772024,
      "grad_norm": 0.5286350846290588,
      "learning_rate": 2.1085858585858587e-05,
      "loss": 1.0528,
      "step": 668
    },
    {
      "epoch": 0.21124092200820965,
      "grad_norm": 0.7728225588798523,
      "learning_rate": 2.1117424242424242e-05,
      "loss": 1.0235,
      "step": 669
    },
    {
      "epoch": 0.21155667824439534,
      "grad_norm": 0.6067861914634705,
      "learning_rate": 2.1148989898989898e-05,
      "loss": 1.0149,
      "step": 670
    },
    {
      "epoch": 0.211872434480581,
      "grad_norm": 0.4307912290096283,
      "learning_rate": 2.1180555555555556e-05,
      "loss": 1.0304,
      "step": 671
    },
    {
      "epoch": 0.21218819071676664,
      "grad_norm": 0.4987950623035431,
      "learning_rate": 2.1212121212121215e-05,
      "loss": 1.0528,
      "step": 672
    },
    {
      "epoch": 0.21250394695295233,
      "grad_norm": 0.7484169602394104,
      "learning_rate": 2.124368686868687e-05,
      "loss": 1.029,
      "step": 673
    },
    {
      "epoch": 0.21281970318913798,
      "grad_norm": 0.4005502760410309,
      "learning_rate": 2.1275252525252525e-05,
      "loss": 1.0166,
      "step": 674
    },
    {
      "epoch": 0.21313545942532364,
      "grad_norm": 0.5816409587860107,
      "learning_rate": 2.1306818181818183e-05,
      "loss": 1.0174,
      "step": 675
    },
    {
      "epoch": 0.21345121566150932,
      "grad_norm": 0.6942309141159058,
      "learning_rate": 2.133838383838384e-05,
      "loss": 1.044,
      "step": 676
    },
    {
      "epoch": 0.21376697189769497,
      "grad_norm": 0.5501375794410706,
      "learning_rate": 2.1369949494949497e-05,
      "loss": 1.0116,
      "step": 677
    },
    {
      "epoch": 0.21408272813388066,
      "grad_norm": 0.7036021947860718,
      "learning_rate": 2.1401515151515152e-05,
      "loss": 1.0496,
      "step": 678
    },
    {
      "epoch": 0.2143984843700663,
      "grad_norm": 0.4147731065750122,
      "learning_rate": 2.143308080808081e-05,
      "loss": 0.9978,
      "step": 679
    },
    {
      "epoch": 0.21471424060625197,
      "grad_norm": 0.47014981508255005,
      "learning_rate": 2.1464646464646466e-05,
      "loss": 1.0301,
      "step": 680
    },
    {
      "epoch": 0.21502999684243765,
      "grad_norm": 0.4274318516254425,
      "learning_rate": 2.149621212121212e-05,
      "loss": 1.0308,
      "step": 681
    },
    {
      "epoch": 0.2153457530786233,
      "grad_norm": 0.6518449783325195,
      "learning_rate": 2.152777777777778e-05,
      "loss": 1.0402,
      "step": 682
    },
    {
      "epoch": 0.21566150931480896,
      "grad_norm": 0.5010143518447876,
      "learning_rate": 2.1559343434343435e-05,
      "loss": 0.9916,
      "step": 683
    },
    {
      "epoch": 0.21597726555099464,
      "grad_norm": 0.5423573851585388,
      "learning_rate": 2.1590909090909093e-05,
      "loss": 1.0112,
      "step": 684
    },
    {
      "epoch": 0.2162930217871803,
      "grad_norm": 1.0407977104187012,
      "learning_rate": 2.162247474747475e-05,
      "loss": 1.0172,
      "step": 685
    },
    {
      "epoch": 0.21660877802336595,
      "grad_norm": 0.5896965265274048,
      "learning_rate": 2.1654040404040404e-05,
      "loss": 1.0307,
      "step": 686
    },
    {
      "epoch": 0.21692453425955163,
      "grad_norm": 0.4485927224159241,
      "learning_rate": 2.1685606060606062e-05,
      "loss": 1.0288,
      "step": 687
    },
    {
      "epoch": 0.2172402904957373,
      "grad_norm": 0.930818498134613,
      "learning_rate": 2.171717171717172e-05,
      "loss": 1.0299,
      "step": 688
    },
    {
      "epoch": 0.21755604673192294,
      "grad_norm": 0.740429699420929,
      "learning_rate": 2.1748737373737372e-05,
      "loss": 0.9839,
      "step": 689
    },
    {
      "epoch": 0.21787180296810862,
      "grad_norm": 0.4986327290534973,
      "learning_rate": 2.178030303030303e-05,
      "loss": 1.0399,
      "step": 690
    },
    {
      "epoch": 0.21818755920429428,
      "grad_norm": 0.44794249534606934,
      "learning_rate": 2.181186868686869e-05,
      "loss": 1.0116,
      "step": 691
    },
    {
      "epoch": 0.21850331544047996,
      "grad_norm": 0.6709899306297302,
      "learning_rate": 2.1843434343434345e-05,
      "loss": 1.0324,
      "step": 692
    },
    {
      "epoch": 0.21881907167666562,
      "grad_norm": 0.5644249320030212,
      "learning_rate": 2.1875e-05,
      "loss": 1.0099,
      "step": 693
    },
    {
      "epoch": 0.21913482791285127,
      "grad_norm": 0.6737472414970398,
      "learning_rate": 2.190656565656566e-05,
      "loss": 1.0114,
      "step": 694
    },
    {
      "epoch": 0.21945058414903695,
      "grad_norm": 0.42300182580947876,
      "learning_rate": 2.1938131313131313e-05,
      "loss": 1.0176,
      "step": 695
    },
    {
      "epoch": 0.2197663403852226,
      "grad_norm": 0.6889756321907043,
      "learning_rate": 2.1969696969696972e-05,
      "loss": 1.0388,
      "step": 696
    },
    {
      "epoch": 0.22008209662140826,
      "grad_norm": 0.6835423707962036,
      "learning_rate": 2.2001262626262627e-05,
      "loss": 1.0369,
      "step": 697
    },
    {
      "epoch": 0.22039785285759395,
      "grad_norm": 0.6190686821937561,
      "learning_rate": 2.2032828282828282e-05,
      "loss": 1.0375,
      "step": 698
    },
    {
      "epoch": 0.2207136090937796,
      "grad_norm": 0.6136049628257751,
      "learning_rate": 2.206439393939394e-05,
      "loss": 1.0468,
      "step": 699
    },
    {
      "epoch": 0.22102936532996525,
      "grad_norm": 1.13099205493927,
      "learning_rate": 2.20959595959596e-05,
      "loss": 1.0124,
      "step": 700
    },
    {
      "epoch": 0.22134512156615094,
      "grad_norm": 0.8515063524246216,
      "learning_rate": 2.212752525252525e-05,
      "loss": 1.0199,
      "step": 701
    },
    {
      "epoch": 0.2216608778023366,
      "grad_norm": 0.5937092900276184,
      "learning_rate": 2.215909090909091e-05,
      "loss": 1.0177,
      "step": 702
    },
    {
      "epoch": 0.22197663403852225,
      "grad_norm": 0.6651560068130493,
      "learning_rate": 2.2190656565656568e-05,
      "loss": 0.9954,
      "step": 703
    },
    {
      "epoch": 0.22229239027470793,
      "grad_norm": 0.6118103861808777,
      "learning_rate": 2.2222222222222223e-05,
      "loss": 1.0117,
      "step": 704
    },
    {
      "epoch": 0.22260814651089358,
      "grad_norm": 0.7394393086433411,
      "learning_rate": 2.225378787878788e-05,
      "loss": 1.0384,
      "step": 705
    },
    {
      "epoch": 0.22292390274707927,
      "grad_norm": 0.5154208540916443,
      "learning_rate": 2.2285353535353537e-05,
      "loss": 1.0196,
      "step": 706
    },
    {
      "epoch": 0.22323965898326492,
      "grad_norm": 0.47634661197662354,
      "learning_rate": 2.2316919191919192e-05,
      "loss": 1.027,
      "step": 707
    },
    {
      "epoch": 0.22355541521945058,
      "grad_norm": 0.7508652210235596,
      "learning_rate": 2.234848484848485e-05,
      "loss": 1.0411,
      "step": 708
    },
    {
      "epoch": 0.22387117145563626,
      "grad_norm": 0.6743278503417969,
      "learning_rate": 2.2380050505050506e-05,
      "loss": 1.033,
      "step": 709
    },
    {
      "epoch": 0.2241869276918219,
      "grad_norm": 0.5470077991485596,
      "learning_rate": 2.2411616161616164e-05,
      "loss": 1.0315,
      "step": 710
    },
    {
      "epoch": 0.22450268392800757,
      "grad_norm": 0.529043436050415,
      "learning_rate": 2.244318181818182e-05,
      "loss": 1.0069,
      "step": 711
    },
    {
      "epoch": 0.22481844016419325,
      "grad_norm": 0.5464317202568054,
      "learning_rate": 2.2474747474747475e-05,
      "loss": 1.0488,
      "step": 712
    },
    {
      "epoch": 0.2251341964003789,
      "grad_norm": 0.6560144424438477,
      "learning_rate": 2.2506313131313133e-05,
      "loss": 1.026,
      "step": 713
    },
    {
      "epoch": 0.22544995263656456,
      "grad_norm": 0.5037466883659363,
      "learning_rate": 2.2537878787878788e-05,
      "loss": 1.0182,
      "step": 714
    },
    {
      "epoch": 0.22576570887275024,
      "grad_norm": 0.6642871499061584,
      "learning_rate": 2.2569444444444447e-05,
      "loss": 1.024,
      "step": 715
    },
    {
      "epoch": 0.2260814651089359,
      "grad_norm": 0.5686080455780029,
      "learning_rate": 2.2601010101010102e-05,
      "loss": 1.0164,
      "step": 716
    },
    {
      "epoch": 0.22639722134512158,
      "grad_norm": 0.5014674067497253,
      "learning_rate": 2.2632575757575757e-05,
      "loss": 1.0353,
      "step": 717
    },
    {
      "epoch": 0.22671297758130723,
      "grad_norm": 0.7156490087509155,
      "learning_rate": 2.2664141414141416e-05,
      "loss": 1.0142,
      "step": 718
    },
    {
      "epoch": 0.2270287338174929,
      "grad_norm": 0.6577869057655334,
      "learning_rate": 2.2695707070707074e-05,
      "loss": 1.0185,
      "step": 719
    },
    {
      "epoch": 0.22734449005367857,
      "grad_norm": 0.47454604506492615,
      "learning_rate": 2.272727272727273e-05,
      "loss": 1.0001,
      "step": 720
    },
    {
      "epoch": 0.22766024628986423,
      "grad_norm": 0.6045597791671753,
      "learning_rate": 2.2758838383838384e-05,
      "loss": 1.0182,
      "step": 721
    },
    {
      "epoch": 0.22797600252604988,
      "grad_norm": 0.5124797821044922,
      "learning_rate": 2.2790404040404043e-05,
      "loss": 1.0456,
      "step": 722
    },
    {
      "epoch": 0.22829175876223556,
      "grad_norm": 0.5555612444877625,
      "learning_rate": 2.2821969696969698e-05,
      "loss": 1.0053,
      "step": 723
    },
    {
      "epoch": 0.22860751499842122,
      "grad_norm": 0.6024741530418396,
      "learning_rate": 2.2853535353535353e-05,
      "loss": 1.0447,
      "step": 724
    },
    {
      "epoch": 0.22892327123460687,
      "grad_norm": 0.5767441987991333,
      "learning_rate": 2.2885101010101012e-05,
      "loss": 1.033,
      "step": 725
    },
    {
      "epoch": 0.22923902747079256,
      "grad_norm": 0.5647996068000793,
      "learning_rate": 2.2916666666666667e-05,
      "loss": 1.0332,
      "step": 726
    },
    {
      "epoch": 0.2295547837069782,
      "grad_norm": 0.5806792974472046,
      "learning_rate": 2.2948232323232325e-05,
      "loss": 1.0124,
      "step": 727
    },
    {
      "epoch": 0.22987053994316387,
      "grad_norm": 0.6384309530258179,
      "learning_rate": 2.297979797979798e-05,
      "loss": 1.0399,
      "step": 728
    },
    {
      "epoch": 0.23018629617934955,
      "grad_norm": 0.513745129108429,
      "learning_rate": 2.3011363636363636e-05,
      "loss": 1.0128,
      "step": 729
    },
    {
      "epoch": 0.2305020524155352,
      "grad_norm": 0.6011333465576172,
      "learning_rate": 2.3042929292929294e-05,
      "loss": 1.0204,
      "step": 730
    },
    {
      "epoch": 0.23081780865172088,
      "grad_norm": 0.5065865516662598,
      "learning_rate": 2.3074494949494953e-05,
      "loss": 0.9985,
      "step": 731
    },
    {
      "epoch": 0.23113356488790654,
      "grad_norm": 0.642507016658783,
      "learning_rate": 2.3106060606060605e-05,
      "loss": 1.0253,
      "step": 732
    },
    {
      "epoch": 0.2314493211240922,
      "grad_norm": 0.5689709782600403,
      "learning_rate": 2.3137626262626263e-05,
      "loss": 1.0343,
      "step": 733
    },
    {
      "epoch": 0.23176507736027788,
      "grad_norm": 0.535984218120575,
      "learning_rate": 2.316919191919192e-05,
      "loss": 1.0385,
      "step": 734
    },
    {
      "epoch": 0.23208083359646353,
      "grad_norm": 0.5884150266647339,
      "learning_rate": 2.3200757575757577e-05,
      "loss": 1.023,
      "step": 735
    },
    {
      "epoch": 0.2323965898326492,
      "grad_norm": 0.6700473427772522,
      "learning_rate": 2.3232323232323232e-05,
      "loss": 0.9712,
      "step": 736
    },
    {
      "epoch": 0.23271234606883487,
      "grad_norm": 0.5157325267791748,
      "learning_rate": 2.326388888888889e-05,
      "loss": 1.0033,
      "step": 737
    },
    {
      "epoch": 0.23302810230502052,
      "grad_norm": 0.8712455630302429,
      "learning_rate": 2.3295454545454546e-05,
      "loss": 1.0075,
      "step": 738
    },
    {
      "epoch": 0.23334385854120618,
      "grad_norm": 0.45964473485946655,
      "learning_rate": 2.3327020202020204e-05,
      "loss": 1.0299,
      "step": 739
    },
    {
      "epoch": 0.23365961477739186,
      "grad_norm": 0.434635192155838,
      "learning_rate": 2.335858585858586e-05,
      "loss": 1.0359,
      "step": 740
    },
    {
      "epoch": 0.23397537101357752,
      "grad_norm": 0.46764862537384033,
      "learning_rate": 2.3390151515151514e-05,
      "loss": 1.0292,
      "step": 741
    },
    {
      "epoch": 0.23429112724976317,
      "grad_norm": 0.592023491859436,
      "learning_rate": 2.3421717171717173e-05,
      "loss": 1.0151,
      "step": 742
    },
    {
      "epoch": 0.23460688348594885,
      "grad_norm": 0.6759070754051208,
      "learning_rate": 2.345328282828283e-05,
      "loss": 1.0217,
      "step": 743
    },
    {
      "epoch": 0.2349226397221345,
      "grad_norm": 0.7181152105331421,
      "learning_rate": 2.3484848484848487e-05,
      "loss": 1.0275,
      "step": 744
    },
    {
      "epoch": 0.2352383959583202,
      "grad_norm": 0.5473641753196716,
      "learning_rate": 2.3516414141414142e-05,
      "loss": 1.0083,
      "step": 745
    },
    {
      "epoch": 0.23555415219450584,
      "grad_norm": 0.6031711101531982,
      "learning_rate": 2.35479797979798e-05,
      "loss": 1.013,
      "step": 746
    },
    {
      "epoch": 0.2358699084306915,
      "grad_norm": 0.5421103835105896,
      "learning_rate": 2.3579545454545455e-05,
      "loss": 1.0156,
      "step": 747
    },
    {
      "epoch": 0.23618566466687718,
      "grad_norm": 0.7629731893539429,
      "learning_rate": 2.361111111111111e-05,
      "loss": 1.0096,
      "step": 748
    },
    {
      "epoch": 0.23650142090306284,
      "grad_norm": 0.41731035709381104,
      "learning_rate": 2.364267676767677e-05,
      "loss": 1.0205,
      "step": 749
    },
    {
      "epoch": 0.2368171771392485,
      "grad_norm": 0.5830455422401428,
      "learning_rate": 2.3674242424242428e-05,
      "loss": 1.0439,
      "step": 750
    },
    {
      "epoch": 0.23713293337543417,
      "grad_norm": 0.7724905014038086,
      "learning_rate": 2.3705808080808083e-05,
      "loss": 1.0109,
      "step": 751
    },
    {
      "epoch": 0.23744868961161983,
      "grad_norm": 0.5939716100692749,
      "learning_rate": 2.3737373737373738e-05,
      "loss": 1.0368,
      "step": 752
    },
    {
      "epoch": 0.23776444584780548,
      "grad_norm": 0.8882811069488525,
      "learning_rate": 2.3768939393939396e-05,
      "loss": 1.0083,
      "step": 753
    },
    {
      "epoch": 0.23808020208399117,
      "grad_norm": 0.5719063878059387,
      "learning_rate": 2.380050505050505e-05,
      "loss": 1.0138,
      "step": 754
    },
    {
      "epoch": 0.23839595832017682,
      "grad_norm": 0.7188387513160706,
      "learning_rate": 2.3832070707070707e-05,
      "loss": 1.0131,
      "step": 755
    },
    {
      "epoch": 0.23871171455636248,
      "grad_norm": 0.6570460796356201,
      "learning_rate": 2.3863636363636365e-05,
      "loss": 0.9996,
      "step": 756
    },
    {
      "epoch": 0.23902747079254816,
      "grad_norm": 0.6772882342338562,
      "learning_rate": 2.389520202020202e-05,
      "loss": 1.0543,
      "step": 757
    },
    {
      "epoch": 0.2393432270287338,
      "grad_norm": 0.7315647006034851,
      "learning_rate": 2.392676767676768e-05,
      "loss": 1.0204,
      "step": 758
    },
    {
      "epoch": 0.2396589832649195,
      "grad_norm": 0.6758192777633667,
      "learning_rate": 2.3958333333333334e-05,
      "loss": 1.0076,
      "step": 759
    },
    {
      "epoch": 0.23997473950110515,
      "grad_norm": 0.7392953038215637,
      "learning_rate": 2.398989898989899e-05,
      "loss": 1.0612,
      "step": 760
    },
    {
      "epoch": 0.2402904957372908,
      "grad_norm": 0.4288792610168457,
      "learning_rate": 2.4021464646464648e-05,
      "loss": 1.0449,
      "step": 761
    },
    {
      "epoch": 0.2406062519734765,
      "grad_norm": 0.6269720196723938,
      "learning_rate": 2.4053030303030306e-05,
      "loss": 1.0039,
      "step": 762
    },
    {
      "epoch": 0.24092200820966214,
      "grad_norm": 0.844637930393219,
      "learning_rate": 2.408459595959596e-05,
      "loss": 1.0153,
      "step": 763
    },
    {
      "epoch": 0.2412377644458478,
      "grad_norm": 0.5216944813728333,
      "learning_rate": 2.4116161616161617e-05,
      "loss": 0.9948,
      "step": 764
    },
    {
      "epoch": 0.24155352068203348,
      "grad_norm": 0.6629865765571594,
      "learning_rate": 2.4147727272727275e-05,
      "loss": 1.0309,
      "step": 765
    },
    {
      "epoch": 0.24186927691821913,
      "grad_norm": 0.44199061393737793,
      "learning_rate": 2.417929292929293e-05,
      "loss": 0.982,
      "step": 766
    },
    {
      "epoch": 0.2421850331544048,
      "grad_norm": 0.6618104577064514,
      "learning_rate": 2.4210858585858585e-05,
      "loss": 0.9767,
      "step": 767
    },
    {
      "epoch": 0.24250078939059047,
      "grad_norm": 0.5663085579872131,
      "learning_rate": 2.4242424242424244e-05,
      "loss": 1.0254,
      "step": 768
    },
    {
      "epoch": 0.24281654562677613,
      "grad_norm": 0.6653010249137878,
      "learning_rate": 2.42739898989899e-05,
      "loss": 0.996,
      "step": 769
    },
    {
      "epoch": 0.24313230186296178,
      "grad_norm": 0.48190125823020935,
      "learning_rate": 2.4305555555555558e-05,
      "loss": 1.0172,
      "step": 770
    },
    {
      "epoch": 0.24344805809914746,
      "grad_norm": 0.4528852105140686,
      "learning_rate": 2.4337121212121213e-05,
      "loss": 1.0085,
      "step": 771
    },
    {
      "epoch": 0.24376381433533312,
      "grad_norm": 0.5828052759170532,
      "learning_rate": 2.4368686868686868e-05,
      "loss": 1.0129,
      "step": 772
    },
    {
      "epoch": 0.2440795705715188,
      "grad_norm": 0.9014537334442139,
      "learning_rate": 2.4400252525252526e-05,
      "loss": 1.0143,
      "step": 773
    },
    {
      "epoch": 0.24439532680770445,
      "grad_norm": 0.7106508016586304,
      "learning_rate": 2.4431818181818185e-05,
      "loss": 1.026,
      "step": 774
    },
    {
      "epoch": 0.2447110830438901,
      "grad_norm": 0.5075216889381409,
      "learning_rate": 2.4463383838383837e-05,
      "loss": 1.0179,
      "step": 775
    },
    {
      "epoch": 0.2450268392800758,
      "grad_norm": 0.5805827975273132,
      "learning_rate": 2.4494949494949495e-05,
      "loss": 1.0229,
      "step": 776
    },
    {
      "epoch": 0.24534259551626145,
      "grad_norm": 0.4644012153148651,
      "learning_rate": 2.4526515151515154e-05,
      "loss": 1.0164,
      "step": 777
    },
    {
      "epoch": 0.2456583517524471,
      "grad_norm": 0.637755274772644,
      "learning_rate": 2.4558080808080812e-05,
      "loss": 0.997,
      "step": 778
    },
    {
      "epoch": 0.24597410798863278,
      "grad_norm": 0.5090121030807495,
      "learning_rate": 2.4589646464646464e-05,
      "loss": 0.9957,
      "step": 779
    },
    {
      "epoch": 0.24628986422481844,
      "grad_norm": 0.5575317144393921,
      "learning_rate": 2.4621212121212123e-05,
      "loss": 1.0088,
      "step": 780
    },
    {
      "epoch": 0.2466056204610041,
      "grad_norm": 0.4992276728153229,
      "learning_rate": 2.465277777777778e-05,
      "loss": 1.0536,
      "step": 781
    },
    {
      "epoch": 0.24692137669718978,
      "grad_norm": 0.4862322211265564,
      "learning_rate": 2.4684343434343436e-05,
      "loss": 1.001,
      "step": 782
    },
    {
      "epoch": 0.24723713293337543,
      "grad_norm": 0.5822766423225403,
      "learning_rate": 2.471590909090909e-05,
      "loss": 1.0122,
      "step": 783
    },
    {
      "epoch": 0.24755288916956109,
      "grad_norm": 0.6791715025901794,
      "learning_rate": 2.474747474747475e-05,
      "loss": 1.0121,
      "step": 784
    },
    {
      "epoch": 0.24786864540574677,
      "grad_norm": 0.5054144859313965,
      "learning_rate": 2.4779040404040405e-05,
      "loss": 1.0093,
      "step": 785
    },
    {
      "epoch": 0.24818440164193242,
      "grad_norm": 0.4490484595298767,
      "learning_rate": 2.4810606060606064e-05,
      "loss": 1.0215,
      "step": 786
    },
    {
      "epoch": 0.2485001578781181,
      "grad_norm": 0.5005213618278503,
      "learning_rate": 2.484217171717172e-05,
      "loss": 0.9983,
      "step": 787
    },
    {
      "epoch": 0.24881591411430376,
      "grad_norm": 0.5331529974937439,
      "learning_rate": 2.4873737373737374e-05,
      "loss": 1.007,
      "step": 788
    },
    {
      "epoch": 0.24913167035048941,
      "grad_norm": 0.7384803891181946,
      "learning_rate": 2.4905303030303032e-05,
      "loss": 1.0391,
      "step": 789
    },
    {
      "epoch": 0.2494474265866751,
      "grad_norm": 0.7180820107460022,
      "learning_rate": 2.4936868686868688e-05,
      "loss": 1.0466,
      "step": 790
    },
    {
      "epoch": 0.24976318282286075,
      "grad_norm": 0.46426746249198914,
      "learning_rate": 2.4968434343434343e-05,
      "loss": 1.0119,
      "step": 791
    },
    {
      "epoch": 0.25007893905904643,
      "grad_norm": 0.6153473258018494,
      "learning_rate": 2.5e-05,
      "loss": 1.0231,
      "step": 792
    },
    {
      "epoch": 0.25039469529523206,
      "grad_norm": 0.647190272808075,
      "learning_rate": 2.503156565656566e-05,
      "loss": 1.0219,
      "step": 793
    },
    {
      "epoch": 0.25071045153141774,
      "grad_norm": 0.44715416431427,
      "learning_rate": 2.5063131313131315e-05,
      "loss": 1.0098,
      "step": 794
    },
    {
      "epoch": 0.2510262077676034,
      "grad_norm": 0.509994387626648,
      "learning_rate": 2.5094696969696973e-05,
      "loss": 1.0207,
      "step": 795
    },
    {
      "epoch": 0.25134196400378905,
      "grad_norm": 0.5901062488555908,
      "learning_rate": 2.5126262626262625e-05,
      "loss": 1.0011,
      "step": 796
    },
    {
      "epoch": 0.25165772023997474,
      "grad_norm": 0.7842277884483337,
      "learning_rate": 2.5157828282828284e-05,
      "loss": 0.9971,
      "step": 797
    },
    {
      "epoch": 0.2519734764761604,
      "grad_norm": 0.6934862732887268,
      "learning_rate": 2.518939393939394e-05,
      "loss": 1.0043,
      "step": 798
    },
    {
      "epoch": 0.25228923271234605,
      "grad_norm": 0.5879155397415161,
      "learning_rate": 2.5220959595959597e-05,
      "loss": 1.0279,
      "step": 799
    },
    {
      "epoch": 0.25260498894853173,
      "grad_norm": 0.4626691937446594,
      "learning_rate": 2.5252525252525256e-05,
      "loss": 1.0253,
      "step": 800
    },
    {
      "epoch": 0.25260498894853173,
      "eval_loss": 0.994612455368042,
      "eval_runtime": 4.6394,
      "eval_samples_per_second": 536.707,
      "eval_steps_per_second": 4.526,
      "step": 800
    },
    {
      "epoch": 0.2529207451847174,
      "grad_norm": 0.6353015303611755,
      "learning_rate": 2.5284090909090914e-05,
      "loss": 1.0101,
      "step": 801
    },
    {
      "epoch": 0.25323650142090304,
      "grad_norm": 0.6041882634162903,
      "learning_rate": 2.5315656565656566e-05,
      "loss": 1.0246,
      "step": 802
    },
    {
      "epoch": 0.2535522576570887,
      "grad_norm": 0.5475590229034424,
      "learning_rate": 2.534722222222222e-05,
      "loss": 1.0235,
      "step": 803
    },
    {
      "epoch": 0.2538680138932744,
      "grad_norm": 0.42511406540870667,
      "learning_rate": 2.537878787878788e-05,
      "loss": 1.0176,
      "step": 804
    },
    {
      "epoch": 0.25418377012946003,
      "grad_norm": 0.609727680683136,
      "learning_rate": 2.541035353535354e-05,
      "loss": 1.001,
      "step": 805
    },
    {
      "epoch": 0.2544995263656457,
      "grad_norm": 0.8161380290985107,
      "learning_rate": 2.5441919191919194e-05,
      "loss": 1.0111,
      "step": 806
    },
    {
      "epoch": 0.2548152826018314,
      "grad_norm": 0.6428655982017517,
      "learning_rate": 2.5473484848484852e-05,
      "loss": 1.0086,
      "step": 807
    },
    {
      "epoch": 0.2551310388380171,
      "grad_norm": 0.5883607268333435,
      "learning_rate": 2.5505050505050504e-05,
      "loss": 1.0156,
      "step": 808
    },
    {
      "epoch": 0.2554467950742027,
      "grad_norm": 0.783709704875946,
      "learning_rate": 2.5536616161616162e-05,
      "loss": 0.9937,
      "step": 809
    },
    {
      "epoch": 0.2557625513103884,
      "grad_norm": 0.5270714163780212,
      "learning_rate": 2.5568181818181817e-05,
      "loss": 1.0033,
      "step": 810
    },
    {
      "epoch": 0.25607830754657407,
      "grad_norm": 0.5563300848007202,
      "learning_rate": 2.5599747474747476e-05,
      "loss": 1.0193,
      "step": 811
    },
    {
      "epoch": 0.2563940637827597,
      "grad_norm": 0.6057838201522827,
      "learning_rate": 2.5631313131313135e-05,
      "loss": 1.0281,
      "step": 812
    },
    {
      "epoch": 0.2567098200189454,
      "grad_norm": 0.626346230506897,
      "learning_rate": 2.566287878787879e-05,
      "loss": 1.0198,
      "step": 813
    },
    {
      "epoch": 0.25702557625513106,
      "grad_norm": 0.5775713324546814,
      "learning_rate": 2.5694444444444445e-05,
      "loss": 1.0176,
      "step": 814
    },
    {
      "epoch": 0.2573413324913167,
      "grad_norm": 0.6728518009185791,
      "learning_rate": 2.57260101010101e-05,
      "loss": 1.0388,
      "step": 815
    },
    {
      "epoch": 0.25765708872750237,
      "grad_norm": 0.7749158143997192,
      "learning_rate": 2.575757575757576e-05,
      "loss": 1.044,
      "step": 816
    },
    {
      "epoch": 0.25797284496368805,
      "grad_norm": 0.6062538027763367,
      "learning_rate": 2.5789141414141417e-05,
      "loss": 1.0473,
      "step": 817
    },
    {
      "epoch": 0.2582886011998737,
      "grad_norm": 0.6362780332565308,
      "learning_rate": 2.5820707070707072e-05,
      "loss": 1.0222,
      "step": 818
    },
    {
      "epoch": 0.25860435743605936,
      "grad_norm": 0.657927930355072,
      "learning_rate": 2.585227272727273e-05,
      "loss": 1.0096,
      "step": 819
    },
    {
      "epoch": 0.25892011367224504,
      "grad_norm": 0.9773058891296387,
      "learning_rate": 2.5883838383838382e-05,
      "loss": 1.0348,
      "step": 820
    },
    {
      "epoch": 0.25923586990843067,
      "grad_norm": 0.8367136120796204,
      "learning_rate": 2.591540404040404e-05,
      "loss": 1.0319,
      "step": 821
    },
    {
      "epoch": 0.25955162614461635,
      "grad_norm": 0.6693466901779175,
      "learning_rate": 2.5946969696969696e-05,
      "loss": 1.004,
      "step": 822
    },
    {
      "epoch": 0.25986738238080204,
      "grad_norm": 0.5540122389793396,
      "learning_rate": 2.5978535353535355e-05,
      "loss": 1.0121,
      "step": 823
    },
    {
      "epoch": 0.26018313861698766,
      "grad_norm": 0.6807516813278198,
      "learning_rate": 2.6010101010101013e-05,
      "loss": 0.9939,
      "step": 824
    },
    {
      "epoch": 0.26049889485317335,
      "grad_norm": 0.5016786456108093,
      "learning_rate": 2.604166666666667e-05,
      "loss": 0.9842,
      "step": 825
    },
    {
      "epoch": 0.26081465108935903,
      "grad_norm": 0.6111496686935425,
      "learning_rate": 2.6073232323232327e-05,
      "loss": 1.0407,
      "step": 826
    },
    {
      "epoch": 0.26113040732554466,
      "grad_norm": 0.6065047383308411,
      "learning_rate": 2.610479797979798e-05,
      "loss": 1.0276,
      "step": 827
    },
    {
      "epoch": 0.26144616356173034,
      "grad_norm": 0.6177941560745239,
      "learning_rate": 2.6136363636363637e-05,
      "loss": 1.0074,
      "step": 828
    },
    {
      "epoch": 0.261761919797916,
      "grad_norm": 0.6432917714118958,
      "learning_rate": 2.6167929292929296e-05,
      "loss": 1.026,
      "step": 829
    },
    {
      "epoch": 0.26207767603410165,
      "grad_norm": 0.8014345765113831,
      "learning_rate": 2.619949494949495e-05,
      "loss": 1.0319,
      "step": 830
    },
    {
      "epoch": 0.26239343227028733,
      "grad_norm": 0.8354324698448181,
      "learning_rate": 2.623106060606061e-05,
      "loss": 1.0137,
      "step": 831
    },
    {
      "epoch": 0.262709188506473,
      "grad_norm": 0.49972712993621826,
      "learning_rate": 2.6262626262626268e-05,
      "loss": 0.9999,
      "step": 832
    },
    {
      "epoch": 0.2630249447426587,
      "grad_norm": 0.5812339186668396,
      "learning_rate": 2.629419191919192e-05,
      "loss": 1.0485,
      "step": 833
    },
    {
      "epoch": 0.2633407009788443,
      "grad_norm": 0.528299868106842,
      "learning_rate": 2.6325757575757575e-05,
      "loss": 1.0392,
      "step": 834
    },
    {
      "epoch": 0.26365645721503,
      "grad_norm": 0.817118227481842,
      "learning_rate": 2.6357323232323233e-05,
      "loss": 1.0145,
      "step": 835
    },
    {
      "epoch": 0.2639722134512157,
      "grad_norm": 0.8347872495651245,
      "learning_rate": 2.6388888888888892e-05,
      "loss": 1.0319,
      "step": 836
    },
    {
      "epoch": 0.2642879696874013,
      "grad_norm": 0.4593498706817627,
      "learning_rate": 2.6420454545454547e-05,
      "loss": 0.9936,
      "step": 837
    },
    {
      "epoch": 0.264603725923587,
      "grad_norm": 0.5902599096298218,
      "learning_rate": 2.6452020202020206e-05,
      "loss": 1.0108,
      "step": 838
    },
    {
      "epoch": 0.2649194821597727,
      "grad_norm": 0.6968191862106323,
      "learning_rate": 2.6483585858585857e-05,
      "loss": 1.0233,
      "step": 839
    },
    {
      "epoch": 0.2652352383959583,
      "grad_norm": 0.7899712324142456,
      "learning_rate": 2.6515151515151516e-05,
      "loss": 0.9954,
      "step": 840
    },
    {
      "epoch": 0.265550994632144,
      "grad_norm": 0.6593113541603088,
      "learning_rate": 2.654671717171717e-05,
      "loss": 1.0122,
      "step": 841
    },
    {
      "epoch": 0.26586675086832967,
      "grad_norm": 0.46390917897224426,
      "learning_rate": 2.657828282828283e-05,
      "loss": 0.9901,
      "step": 842
    },
    {
      "epoch": 0.2661825071045153,
      "grad_norm": 0.6570203900337219,
      "learning_rate": 2.6609848484848488e-05,
      "loss": 0.9872,
      "step": 843
    },
    {
      "epoch": 0.266498263340701,
      "grad_norm": 0.6412169933319092,
      "learning_rate": 2.6641414141414147e-05,
      "loss": 1.0136,
      "step": 844
    },
    {
      "epoch": 0.26681401957688666,
      "grad_norm": 0.588208794593811,
      "learning_rate": 2.6672979797979798e-05,
      "loss": 1.0132,
      "step": 845
    },
    {
      "epoch": 0.2671297758130723,
      "grad_norm": 0.6345801949501038,
      "learning_rate": 2.6704545454545453e-05,
      "loss": 0.9922,
      "step": 846
    },
    {
      "epoch": 0.267445532049258,
      "grad_norm": 0.6840760707855225,
      "learning_rate": 2.6736111111111112e-05,
      "loss": 0.98,
      "step": 847
    },
    {
      "epoch": 0.26776128828544365,
      "grad_norm": 0.8888677358627319,
      "learning_rate": 2.676767676767677e-05,
      "loss": 1.0156,
      "step": 848
    },
    {
      "epoch": 0.2680770445216293,
      "grad_norm": 0.878013551235199,
      "learning_rate": 2.6799242424242426e-05,
      "loss": 0.9573,
      "step": 849
    },
    {
      "epoch": 0.26839280075781496,
      "grad_norm": 0.6625912189483643,
      "learning_rate": 2.6830808080808084e-05,
      "loss": 0.9941,
      "step": 850
    },
    {
      "epoch": 0.26870855699400065,
      "grad_norm": 0.7086650729179382,
      "learning_rate": 2.6862373737373736e-05,
      "loss": 1.0036,
      "step": 851
    },
    {
      "epoch": 0.2690243132301863,
      "grad_norm": 0.6387296915054321,
      "learning_rate": 2.6893939393939394e-05,
      "loss": 1.0058,
      "step": 852
    },
    {
      "epoch": 0.26934006946637196,
      "grad_norm": 0.7355128526687622,
      "learning_rate": 2.692550505050505e-05,
      "loss": 1.0322,
      "step": 853
    },
    {
      "epoch": 0.26965582570255764,
      "grad_norm": 0.7605606913566589,
      "learning_rate": 2.6957070707070708e-05,
      "loss": 1.0209,
      "step": 854
    },
    {
      "epoch": 0.26997158193874327,
      "grad_norm": 0.7136550545692444,
      "learning_rate": 2.6988636363636367e-05,
      "loss": 1.0271,
      "step": 855
    },
    {
      "epoch": 0.27028733817492895,
      "grad_norm": 0.7569994926452637,
      "learning_rate": 2.7020202020202022e-05,
      "loss": 1.0243,
      "step": 856
    },
    {
      "epoch": 0.27060309441111463,
      "grad_norm": 0.8083186745643616,
      "learning_rate": 2.7051767676767677e-05,
      "loss": 1.0241,
      "step": 857
    },
    {
      "epoch": 0.27091885064730026,
      "grad_norm": 0.7017297744750977,
      "learning_rate": 2.7083333333333332e-05,
      "loss": 0.9646,
      "step": 858
    },
    {
      "epoch": 0.27123460688348594,
      "grad_norm": 0.7022137641906738,
      "learning_rate": 2.711489898989899e-05,
      "loss": 1.0519,
      "step": 859
    },
    {
      "epoch": 0.2715503631196716,
      "grad_norm": 0.7434066534042358,
      "learning_rate": 2.714646464646465e-05,
      "loss": 1.0369,
      "step": 860
    },
    {
      "epoch": 0.2718661193558573,
      "grad_norm": 0.5491161346435547,
      "learning_rate": 2.7178030303030304e-05,
      "loss": 1.0692,
      "step": 861
    },
    {
      "epoch": 0.27218187559204293,
      "grad_norm": 0.7594206929206848,
      "learning_rate": 2.7209595959595963e-05,
      "loss": 0.9986,
      "step": 862
    },
    {
      "epoch": 0.2724976318282286,
      "grad_norm": 0.7290535569190979,
      "learning_rate": 2.724116161616162e-05,
      "loss": 1.0088,
      "step": 863
    },
    {
      "epoch": 0.2728133880644143,
      "grad_norm": 0.681964099407196,
      "learning_rate": 2.7272727272727273e-05,
      "loss": 1.0083,
      "step": 864
    },
    {
      "epoch": 0.2731291443005999,
      "grad_norm": 0.6063222885131836,
      "learning_rate": 2.7304292929292928e-05,
      "loss": 1.0028,
      "step": 865
    },
    {
      "epoch": 0.2734449005367856,
      "grad_norm": 0.5361270308494568,
      "learning_rate": 2.7335858585858587e-05,
      "loss": 0.9929,
      "step": 866
    },
    {
      "epoch": 0.2737606567729713,
      "grad_norm": 0.6069661378860474,
      "learning_rate": 2.7367424242424245e-05,
      "loss": 1.0146,
      "step": 867
    },
    {
      "epoch": 0.2740764130091569,
      "grad_norm": 0.6231128573417664,
      "learning_rate": 2.73989898989899e-05,
      "loss": 1.0277,
      "step": 868
    },
    {
      "epoch": 0.2743921692453426,
      "grad_norm": 0.6568114161491394,
      "learning_rate": 2.743055555555556e-05,
      "loss": 1.0062,
      "step": 869
    },
    {
      "epoch": 0.2747079254815283,
      "grad_norm": 0.5687716603279114,
      "learning_rate": 2.746212121212121e-05,
      "loss": 0.9735,
      "step": 870
    },
    {
      "epoch": 0.2750236817177139,
      "grad_norm": 0.6050305366516113,
      "learning_rate": 2.749368686868687e-05,
      "loss": 1.036,
      "step": 871
    },
    {
      "epoch": 0.2753394379538996,
      "grad_norm": 0.6430206894874573,
      "learning_rate": 2.7525252525252528e-05,
      "loss": 0.9954,
      "step": 872
    },
    {
      "epoch": 0.2756551941900853,
      "grad_norm": 0.7372235059738159,
      "learning_rate": 2.7556818181818183e-05,
      "loss": 1.0307,
      "step": 873
    },
    {
      "epoch": 0.2759709504262709,
      "grad_norm": 0.6551008224487305,
      "learning_rate": 2.758838383838384e-05,
      "loss": 1.04,
      "step": 874
    },
    {
      "epoch": 0.2762867066624566,
      "grad_norm": 0.6035629510879517,
      "learning_rate": 2.76199494949495e-05,
      "loss": 1.0249,
      "step": 875
    },
    {
      "epoch": 0.27660246289864227,
      "grad_norm": 0.7465946674346924,
      "learning_rate": 2.7651515151515152e-05,
      "loss": 0.9815,
      "step": 876
    },
    {
      "epoch": 0.2769182191348279,
      "grad_norm": 0.5850507020950317,
      "learning_rate": 2.7683080808080807e-05,
      "loss": 1.0312,
      "step": 877
    },
    {
      "epoch": 0.2772339753710136,
      "grad_norm": 0.5737684965133667,
      "learning_rate": 2.7714646464646465e-05,
      "loss": 0.9987,
      "step": 878
    },
    {
      "epoch": 0.27754973160719926,
      "grad_norm": 0.6384199857711792,
      "learning_rate": 2.7746212121212124e-05,
      "loss": 1.0207,
      "step": 879
    },
    {
      "epoch": 0.2778654878433849,
      "grad_norm": 0.7032705545425415,
      "learning_rate": 2.777777777777778e-05,
      "loss": 1.0189,
      "step": 880
    },
    {
      "epoch": 0.27818124407957057,
      "grad_norm": 0.682196319103241,
      "learning_rate": 2.7809343434343438e-05,
      "loss": 1.0147,
      "step": 881
    },
    {
      "epoch": 0.27849700031575625,
      "grad_norm": 0.6537495255470276,
      "learning_rate": 2.784090909090909e-05,
      "loss": 0.9834,
      "step": 882
    },
    {
      "epoch": 0.2788127565519419,
      "grad_norm": 0.686224102973938,
      "learning_rate": 2.7872474747474748e-05,
      "loss": 1.0314,
      "step": 883
    },
    {
      "epoch": 0.27912851278812756,
      "grad_norm": 0.7582664489746094,
      "learning_rate": 2.7904040404040403e-05,
      "loss": 1.0284,
      "step": 884
    },
    {
      "epoch": 0.27944426902431324,
      "grad_norm": 0.7369611263275146,
      "learning_rate": 2.793560606060606e-05,
      "loss": 1.0182,
      "step": 885
    },
    {
      "epoch": 0.27976002526049887,
      "grad_norm": 0.6143050789833069,
      "learning_rate": 2.796717171717172e-05,
      "loss": 1.019,
      "step": 886
    },
    {
      "epoch": 0.28007578149668455,
      "grad_norm": 0.8488231301307678,
      "learning_rate": 2.799873737373738e-05,
      "loss": 1.0,
      "step": 887
    },
    {
      "epoch": 0.28039153773287023,
      "grad_norm": 0.6160892844200134,
      "learning_rate": 2.803030303030303e-05,
      "loss": 1.014,
      "step": 888
    },
    {
      "epoch": 0.2807072939690559,
      "grad_norm": 0.5499072074890137,
      "learning_rate": 2.8061868686868686e-05,
      "loss": 0.9946,
      "step": 889
    },
    {
      "epoch": 0.28102305020524154,
      "grad_norm": 0.48040345311164856,
      "learning_rate": 2.8093434343434344e-05,
      "loss": 1.0231,
      "step": 890
    },
    {
      "epoch": 0.2813388064414272,
      "grad_norm": 0.6918156743049622,
      "learning_rate": 2.8125000000000003e-05,
      "loss": 1.0091,
      "step": 891
    },
    {
      "epoch": 0.2816545626776129,
      "grad_norm": 0.878390371799469,
      "learning_rate": 2.8156565656565658e-05,
      "loss": 1.0349,
      "step": 892
    },
    {
      "epoch": 0.28197031891379853,
      "grad_norm": 0.6275595426559448,
      "learning_rate": 2.8188131313131316e-05,
      "loss": 1.0038,
      "step": 893
    },
    {
      "epoch": 0.2822860751499842,
      "grad_norm": 0.4881672263145447,
      "learning_rate": 2.8219696969696975e-05,
      "loss": 1.0126,
      "step": 894
    },
    {
      "epoch": 0.2826018313861699,
      "grad_norm": 0.7281088829040527,
      "learning_rate": 2.8251262626262627e-05,
      "loss": 0.9885,
      "step": 895
    },
    {
      "epoch": 0.2829175876223555,
      "grad_norm": 0.5502660274505615,
      "learning_rate": 2.8282828282828282e-05,
      "loss": 0.993,
      "step": 896
    },
    {
      "epoch": 0.2832333438585412,
      "grad_norm": 0.5322092771530151,
      "learning_rate": 2.831439393939394e-05,
      "loss": 1.0288,
      "step": 897
    },
    {
      "epoch": 0.2835491000947269,
      "grad_norm": 0.6844850182533264,
      "learning_rate": 2.83459595959596e-05,
      "loss": 1.01,
      "step": 898
    },
    {
      "epoch": 0.2838648563309125,
      "grad_norm": 0.7186392545700073,
      "learning_rate": 2.8377525252525254e-05,
      "loss": 1.014,
      "step": 899
    },
    {
      "epoch": 0.2841806125670982,
      "grad_norm": 0.6419204473495483,
      "learning_rate": 2.8409090909090912e-05,
      "loss": 1.0003,
      "step": 900
    },
    {
      "epoch": 0.2844963688032839,
      "grad_norm": 0.6123802065849304,
      "learning_rate": 2.8440656565656564e-05,
      "loss": 1.042,
      "step": 901
    },
    {
      "epoch": 0.2848121250394695,
      "grad_norm": 0.9632802605628967,
      "learning_rate": 2.8472222222222223e-05,
      "loss": 1.0441,
      "step": 902
    },
    {
      "epoch": 0.2851278812756552,
      "grad_norm": 0.780693769454956,
      "learning_rate": 2.850378787878788e-05,
      "loss": 1.0339,
      "step": 903
    },
    {
      "epoch": 0.2854436375118409,
      "grad_norm": 0.7429075837135315,
      "learning_rate": 2.8535353535353536e-05,
      "loss": 1.0063,
      "step": 904
    },
    {
      "epoch": 0.2857593937480265,
      "grad_norm": 0.6410915851593018,
      "learning_rate": 2.8566919191919195e-05,
      "loss": 1.0015,
      "step": 905
    },
    {
      "epoch": 0.2860751499842122,
      "grad_norm": 0.6330276131629944,
      "learning_rate": 2.8598484848484853e-05,
      "loss": 1.0283,
      "step": 906
    },
    {
      "epoch": 0.28639090622039787,
      "grad_norm": 0.7047046422958374,
      "learning_rate": 2.8630050505050505e-05,
      "loss": 1.0002,
      "step": 907
    },
    {
      "epoch": 0.2867066624565835,
      "grad_norm": 0.8157705068588257,
      "learning_rate": 2.866161616161616e-05,
      "loss": 1.0314,
      "step": 908
    },
    {
      "epoch": 0.2870224186927692,
      "grad_norm": 0.5515238046646118,
      "learning_rate": 2.869318181818182e-05,
      "loss": 1.0113,
      "step": 909
    },
    {
      "epoch": 0.28733817492895486,
      "grad_norm": 0.540562093257904,
      "learning_rate": 2.8724747474747477e-05,
      "loss": 1.0096,
      "step": 910
    },
    {
      "epoch": 0.2876539311651405,
      "grad_norm": 0.6729962229728699,
      "learning_rate": 2.8756313131313133e-05,
      "loss": 1.0095,
      "step": 911
    },
    {
      "epoch": 0.28796968740132617,
      "grad_norm": 0.7190142273902893,
      "learning_rate": 2.878787878787879e-05,
      "loss": 1.019,
      "step": 912
    },
    {
      "epoch": 0.28828544363751185,
      "grad_norm": 0.6106101870536804,
      "learning_rate": 2.8819444444444443e-05,
      "loss": 1.0403,
      "step": 913
    },
    {
      "epoch": 0.2886011998736975,
      "grad_norm": 0.704369306564331,
      "learning_rate": 2.88510101010101e-05,
      "loss": 1.0149,
      "step": 914
    },
    {
      "epoch": 0.28891695610988316,
      "grad_norm": 0.6029422879219055,
      "learning_rate": 2.888257575757576e-05,
      "loss": 0.9777,
      "step": 915
    },
    {
      "epoch": 0.28923271234606884,
      "grad_norm": 0.5662487149238586,
      "learning_rate": 2.8914141414141415e-05,
      "loss": 1.0078,
      "step": 916
    },
    {
      "epoch": 0.2895484685822545,
      "grad_norm": 1.3098922967910767,
      "learning_rate": 2.8945707070707074e-05,
      "loss": 1.0051,
      "step": 917
    },
    {
      "epoch": 0.28986422481844015,
      "grad_norm": 0.558319628238678,
      "learning_rate": 2.8977272727272732e-05,
      "loss": 1.0053,
      "step": 918
    },
    {
      "epoch": 0.29017998105462584,
      "grad_norm": 0.6257441639900208,
      "learning_rate": 2.9008838383838384e-05,
      "loss": 1.0143,
      "step": 919
    },
    {
      "epoch": 0.2904957372908115,
      "grad_norm": 0.7287858724594116,
      "learning_rate": 2.904040404040404e-05,
      "loss": 1.0181,
      "step": 920
    },
    {
      "epoch": 0.29081149352699714,
      "grad_norm": 0.8265687227249146,
      "learning_rate": 2.9071969696969698e-05,
      "loss": 1.0296,
      "step": 921
    },
    {
      "epoch": 0.2911272497631828,
      "grad_norm": 0.7290377020835876,
      "learning_rate": 2.9103535353535356e-05,
      "loss": 1.0017,
      "step": 922
    },
    {
      "epoch": 0.2914430059993685,
      "grad_norm": 0.5102190375328064,
      "learning_rate": 2.913510101010101e-05,
      "loss": 0.9855,
      "step": 923
    },
    {
      "epoch": 0.29175876223555414,
      "grad_norm": 0.6635766625404358,
      "learning_rate": 2.916666666666667e-05,
      "loss": 1.0151,
      "step": 924
    },
    {
      "epoch": 0.2920745184717398,
      "grad_norm": 0.5930847525596619,
      "learning_rate": 2.9198232323232328e-05,
      "loss": 0.9919,
      "step": 925
    },
    {
      "epoch": 0.2923902747079255,
      "grad_norm": 0.6098858118057251,
      "learning_rate": 2.922979797979798e-05,
      "loss": 0.997,
      "step": 926
    },
    {
      "epoch": 0.29270603094411113,
      "grad_norm": 0.7027726769447327,
      "learning_rate": 2.9261363636363635e-05,
      "loss": 1.016,
      "step": 927
    },
    {
      "epoch": 0.2930217871802968,
      "grad_norm": 0.6732422709465027,
      "learning_rate": 2.9292929292929294e-05,
      "loss": 1.0419,
      "step": 928
    },
    {
      "epoch": 0.2933375434164825,
      "grad_norm": 0.7263798117637634,
      "learning_rate": 2.9324494949494952e-05,
      "loss": 1.0054,
      "step": 929
    },
    {
      "epoch": 0.2936532996526681,
      "grad_norm": 0.8164319396018982,
      "learning_rate": 2.935606060606061e-05,
      "loss": 0.9965,
      "step": 930
    },
    {
      "epoch": 0.2939690558888538,
      "grad_norm": 0.5530664920806885,
      "learning_rate": 2.9387626262626266e-05,
      "loss": 1.0186,
      "step": 931
    },
    {
      "epoch": 0.2942848121250395,
      "grad_norm": 0.6751533150672913,
      "learning_rate": 2.9419191919191918e-05,
      "loss": 1.009,
      "step": 932
    },
    {
      "epoch": 0.2946005683612251,
      "grad_norm": 0.9320776462554932,
      "learning_rate": 2.9450757575757576e-05,
      "loss": 0.9959,
      "step": 933
    },
    {
      "epoch": 0.2949163245974108,
      "grad_norm": 0.697204053401947,
      "learning_rate": 2.9482323232323235e-05,
      "loss": 1.0148,
      "step": 934
    },
    {
      "epoch": 0.2952320808335965,
      "grad_norm": 0.824088454246521,
      "learning_rate": 2.951388888888889e-05,
      "loss": 1.0029,
      "step": 935
    },
    {
      "epoch": 0.2955478370697821,
      "grad_norm": 0.5937426686286926,
      "learning_rate": 2.954545454545455e-05,
      "loss": 0.9809,
      "step": 936
    },
    {
      "epoch": 0.2958635933059678,
      "grad_norm": 0.7838863134384155,
      "learning_rate": 2.9577020202020207e-05,
      "loss": 0.9704,
      "step": 937
    },
    {
      "epoch": 0.29617934954215347,
      "grad_norm": 0.7645484209060669,
      "learning_rate": 2.960858585858586e-05,
      "loss": 1.0027,
      "step": 938
    },
    {
      "epoch": 0.2964951057783391,
      "grad_norm": 0.8175382018089294,
      "learning_rate": 2.9640151515151514e-05,
      "loss": 0.9806,
      "step": 939
    },
    {
      "epoch": 0.2968108620145248,
      "grad_norm": 0.6095353364944458,
      "learning_rate": 2.9671717171717172e-05,
      "loss": 0.9957,
      "step": 940
    },
    {
      "epoch": 0.29712661825071046,
      "grad_norm": 0.7754340767860413,
      "learning_rate": 2.970328282828283e-05,
      "loss": 1.0071,
      "step": 941
    },
    {
      "epoch": 0.29744237448689614,
      "grad_norm": 0.5383034348487854,
      "learning_rate": 2.9734848484848486e-05,
      "loss": 0.9876,
      "step": 942
    },
    {
      "epoch": 0.29775813072308177,
      "grad_norm": 0.7458931803703308,
      "learning_rate": 2.9766414141414145e-05,
      "loss": 1.0124,
      "step": 943
    },
    {
      "epoch": 0.29807388695926745,
      "grad_norm": 0.5629051327705383,
      "learning_rate": 2.9797979797979796e-05,
      "loss": 0.9826,
      "step": 944
    },
    {
      "epoch": 0.29838964319545314,
      "grad_norm": 0.7273377180099487,
      "learning_rate": 2.9829545454545455e-05,
      "loss": 0.9766,
      "step": 945
    },
    {
      "epoch": 0.29870539943163876,
      "grad_norm": 0.7134695649147034,
      "learning_rate": 2.9861111111111113e-05,
      "loss": 1.0079,
      "step": 946
    },
    {
      "epoch": 0.29902115566782445,
      "grad_norm": 0.7956920266151428,
      "learning_rate": 2.989267676767677e-05,
      "loss": 1.019,
      "step": 947
    },
    {
      "epoch": 0.29933691190401013,
      "grad_norm": 0.5580201745033264,
      "learning_rate": 2.9924242424242427e-05,
      "loss": 1.0131,
      "step": 948
    },
    {
      "epoch": 0.29965266814019575,
      "grad_norm": 0.6331074237823486,
      "learning_rate": 2.9955808080808086e-05,
      "loss": 1.0084,
      "step": 949
    },
    {
      "epoch": 0.29996842437638144,
      "grad_norm": 0.731529176235199,
      "learning_rate": 2.9987373737373737e-05,
      "loss": 1.0409,
      "step": 950
    },
    {
      "epoch": 0.3002841806125671,
      "grad_norm": 0.6630391478538513,
      "learning_rate": 3.0018939393939392e-05,
      "loss": 0.9862,
      "step": 951
    },
    {
      "epoch": 0.30059993684875275,
      "grad_norm": 0.4715006351470947,
      "learning_rate": 3.005050505050505e-05,
      "loss": 0.996,
      "step": 952
    },
    {
      "epoch": 0.30091569308493843,
      "grad_norm": 0.6286865472793579,
      "learning_rate": 3.008207070707071e-05,
      "loss": 1.0102,
      "step": 953
    },
    {
      "epoch": 0.3012314493211241,
      "grad_norm": 0.6230069994926453,
      "learning_rate": 3.0113636363636365e-05,
      "loss": 1.0318,
      "step": 954
    },
    {
      "epoch": 0.30154720555730974,
      "grad_norm": 0.7811769247055054,
      "learning_rate": 3.0145202020202023e-05,
      "loss": 1.0098,
      "step": 955
    },
    {
      "epoch": 0.3018629617934954,
      "grad_norm": 0.6354441046714783,
      "learning_rate": 3.0176767676767675e-05,
      "loss": 0.9993,
      "step": 956
    },
    {
      "epoch": 0.3021787180296811,
      "grad_norm": 0.5406370162963867,
      "learning_rate": 3.0208333333333334e-05,
      "loss": 1.0649,
      "step": 957
    },
    {
      "epoch": 0.30249447426586673,
      "grad_norm": 0.610536515712738,
      "learning_rate": 3.0239898989898992e-05,
      "loss": 0.9856,
      "step": 958
    },
    {
      "epoch": 0.3028102305020524,
      "grad_norm": 0.6553962826728821,
      "learning_rate": 3.0271464646464647e-05,
      "loss": 1.0014,
      "step": 959
    },
    {
      "epoch": 0.3031259867382381,
      "grad_norm": 0.5936636328697205,
      "learning_rate": 3.0303030303030306e-05,
      "loss": 1.012,
      "step": 960
    },
    {
      "epoch": 0.3034417429744237,
      "grad_norm": 0.658535897731781,
      "learning_rate": 3.0334595959595964e-05,
      "loss": 1.0048,
      "step": 961
    },
    {
      "epoch": 0.3037574992106094,
      "grad_norm": 0.5968596339225769,
      "learning_rate": 3.036616161616162e-05,
      "loss": 1.0061,
      "step": 962
    },
    {
      "epoch": 0.3040732554467951,
      "grad_norm": 0.5465819835662842,
      "learning_rate": 3.039772727272727e-05,
      "loss": 1.0035,
      "step": 963
    },
    {
      "epoch": 0.3043890116829807,
      "grad_norm": 0.518692135810852,
      "learning_rate": 3.042929292929293e-05,
      "loss": 1.0003,
      "step": 964
    },
    {
      "epoch": 0.3047047679191664,
      "grad_norm": 0.6422468423843384,
      "learning_rate": 3.0460858585858588e-05,
      "loss": 1.0292,
      "step": 965
    },
    {
      "epoch": 0.3050205241553521,
      "grad_norm": 0.5907920002937317,
      "learning_rate": 3.0492424242424243e-05,
      "loss": 1.0164,
      "step": 966
    },
    {
      "epoch": 0.3053362803915377,
      "grad_norm": 0.7077932357788086,
      "learning_rate": 3.0523989898989905e-05,
      "loss": 1.001,
      "step": 967
    },
    {
      "epoch": 0.3056520366277234,
      "grad_norm": 0.6101611852645874,
      "learning_rate": 3.055555555555556e-05,
      "loss": 1.0004,
      "step": 968
    },
    {
      "epoch": 0.30596779286390907,
      "grad_norm": 0.7646480798721313,
      "learning_rate": 3.058712121212121e-05,
      "loss": 0.998,
      "step": 969
    },
    {
      "epoch": 0.30628354910009475,
      "grad_norm": 0.6297016143798828,
      "learning_rate": 3.061868686868687e-05,
      "loss": 0.9967,
      "step": 970
    },
    {
      "epoch": 0.3065993053362804,
      "grad_norm": 0.6018003225326538,
      "learning_rate": 3.0650252525252526e-05,
      "loss": 0.9777,
      "step": 971
    },
    {
      "epoch": 0.30691506157246606,
      "grad_norm": 0.6219320893287659,
      "learning_rate": 3.068181818181818e-05,
      "loss": 1.0173,
      "step": 972
    },
    {
      "epoch": 0.30723081780865175,
      "grad_norm": 1.032305121421814,
      "learning_rate": 3.071338383838384e-05,
      "loss": 1.0069,
      "step": 973
    },
    {
      "epoch": 0.3075465740448374,
      "grad_norm": 0.6635480523109436,
      "learning_rate": 3.07449494949495e-05,
      "loss": 1.0135,
      "step": 974
    },
    {
      "epoch": 0.30786233028102306,
      "grad_norm": 0.5410233736038208,
      "learning_rate": 3.077651515151515e-05,
      "loss": 0.9852,
      "step": 975
    },
    {
      "epoch": 0.30817808651720874,
      "grad_norm": 0.5075051188468933,
      "learning_rate": 3.080808080808081e-05,
      "loss": 1.0209,
      "step": 976
    },
    {
      "epoch": 0.30849384275339437,
      "grad_norm": 0.7264770269393921,
      "learning_rate": 3.0839646464646463e-05,
      "loss": 1.0074,
      "step": 977
    },
    {
      "epoch": 0.30880959898958005,
      "grad_norm": 0.5622143745422363,
      "learning_rate": 3.0871212121212125e-05,
      "loss": 1.0214,
      "step": 978
    },
    {
      "epoch": 0.30912535522576573,
      "grad_norm": 0.6878820657730103,
      "learning_rate": 3.090277777777778e-05,
      "loss": 1.0466,
      "step": 979
    },
    {
      "epoch": 0.30944111146195136,
      "grad_norm": 0.6451988220214844,
      "learning_rate": 3.0934343434343436e-05,
      "loss": 0.9862,
      "step": 980
    },
    {
      "epoch": 0.30975686769813704,
      "grad_norm": 0.7151583433151245,
      "learning_rate": 3.096590909090909e-05,
      "loss": 1.0029,
      "step": 981
    },
    {
      "epoch": 0.3100726239343227,
      "grad_norm": 0.6972950100898743,
      "learning_rate": 3.0997474747474746e-05,
      "loss": 0.9775,
      "step": 982
    },
    {
      "epoch": 0.31038838017050835,
      "grad_norm": 0.6440567374229431,
      "learning_rate": 3.102904040404041e-05,
      "loss": 1.0015,
      "step": 983
    },
    {
      "epoch": 0.31070413640669403,
      "grad_norm": 0.5707190632820129,
      "learning_rate": 3.106060606060606e-05,
      "loss": 1.003,
      "step": 984
    },
    {
      "epoch": 0.3110198926428797,
      "grad_norm": 0.6259674429893494,
      "learning_rate": 3.109217171717172e-05,
      "loss": 0.9881,
      "step": 985
    },
    {
      "epoch": 0.31133564887906534,
      "grad_norm": 0.6056646108627319,
      "learning_rate": 3.112373737373738e-05,
      "loss": 0.9846,
      "step": 986
    },
    {
      "epoch": 0.311651405115251,
      "grad_norm": 0.6115797162055969,
      "learning_rate": 3.115530303030303e-05,
      "loss": 0.9947,
      "step": 987
    },
    {
      "epoch": 0.3119671613514367,
      "grad_norm": 0.43531331419944763,
      "learning_rate": 3.1186868686868684e-05,
      "loss": 1.0036,
      "step": 988
    },
    {
      "epoch": 0.31228291758762233,
      "grad_norm": 0.5844328999519348,
      "learning_rate": 3.1218434343434346e-05,
      "loss": 1.0058,
      "step": 989
    },
    {
      "epoch": 0.312598673823808,
      "grad_norm": 0.9022193551063538,
      "learning_rate": 3.125e-05,
      "loss": 1.0154,
      "step": 990
    },
    {
      "epoch": 0.3129144300599937,
      "grad_norm": 0.5474578142166138,
      "learning_rate": 3.1281565656565656e-05,
      "loss": 0.9972,
      "step": 991
    },
    {
      "epoch": 0.3132301862961793,
      "grad_norm": 0.5318132042884827,
      "learning_rate": 3.131313131313132e-05,
      "loss": 1.0069,
      "step": 992
    },
    {
      "epoch": 0.313545942532365,
      "grad_norm": 0.863933801651001,
      "learning_rate": 3.134469696969697e-05,
      "loss": 0.9664,
      "step": 993
    },
    {
      "epoch": 0.3138616987685507,
      "grad_norm": 0.5872600078582764,
      "learning_rate": 3.137626262626263e-05,
      "loss": 1.0304,
      "step": 994
    },
    {
      "epoch": 0.3141774550047363,
      "grad_norm": 0.6628093123435974,
      "learning_rate": 3.140782828282828e-05,
      "loss": 1.0013,
      "step": 995
    },
    {
      "epoch": 0.314493211240922,
      "grad_norm": 0.6687444448471069,
      "learning_rate": 3.143939393939394e-05,
      "loss": 1.02,
      "step": 996
    },
    {
      "epoch": 0.3148089674771077,
      "grad_norm": 0.70506352186203,
      "learning_rate": 3.14709595959596e-05,
      "loss": 1.0218,
      "step": 997
    },
    {
      "epoch": 0.31512472371329336,
      "grad_norm": 0.6913138031959534,
      "learning_rate": 3.1502525252525255e-05,
      "loss": 0.9866,
      "step": 998
    },
    {
      "epoch": 0.315440479949479,
      "grad_norm": 0.5666153430938721,
      "learning_rate": 3.153409090909091e-05,
      "loss": 0.9954,
      "step": 999
    },
    {
      "epoch": 0.3157562361856647,
      "grad_norm": 0.7329150438308716,
      "learning_rate": 3.1565656565656566e-05,
      "loss": 1.0166,
      "step": 1000
    },
    {
      "epoch": 0.3157562361856647,
      "eval_loss": 0.9747978448867798,
      "eval_runtime": 4.5808,
      "eval_samples_per_second": 543.569,
      "eval_steps_per_second": 4.584,
      "step": 1000
    },
    {
      "epoch": 0.31607199242185036,
      "grad_norm": 0.8043568134307861,
      "learning_rate": 3.159722222222222e-05,
      "loss": 1.0071,
      "step": 1001
    },
    {
      "epoch": 0.316387748658036,
      "grad_norm": 0.8169147968292236,
      "learning_rate": 3.162878787878788e-05,
      "loss": 1.0749,
      "step": 1002
    },
    {
      "epoch": 0.31670350489422167,
      "grad_norm": 0.7698376774787903,
      "learning_rate": 3.166035353535354e-05,
      "loss": 0.9902,
      "step": 1003
    },
    {
      "epoch": 0.31701926113040735,
      "grad_norm": 0.5440167188644409,
      "learning_rate": 3.169191919191919e-05,
      "loss": 0.9847,
      "step": 1004
    },
    {
      "epoch": 0.317335017366593,
      "grad_norm": 0.7259222865104675,
      "learning_rate": 3.1723484848484855e-05,
      "loss": 1.012,
      "step": 1005
    },
    {
      "epoch": 0.31765077360277866,
      "grad_norm": 0.6936396360397339,
      "learning_rate": 3.17550505050505e-05,
      "loss": 1.0056,
      "step": 1006
    },
    {
      "epoch": 0.31796652983896434,
      "grad_norm": 0.7094953656196594,
      "learning_rate": 3.178661616161616e-05,
      "loss": 1.0112,
      "step": 1007
    },
    {
      "epoch": 0.31828228607514997,
      "grad_norm": 0.9117681980133057,
      "learning_rate": 3.181818181818182e-05,
      "loss": 0.9644,
      "step": 1008
    },
    {
      "epoch": 0.31859804231133565,
      "grad_norm": 0.8280858397483826,
      "learning_rate": 3.1849747474747475e-05,
      "loss": 1.0058,
      "step": 1009
    },
    {
      "epoch": 0.31891379854752133,
      "grad_norm": 0.7093799710273743,
      "learning_rate": 3.188131313131314e-05,
      "loss": 0.9831,
      "step": 1010
    },
    {
      "epoch": 0.31922955478370696,
      "grad_norm": 0.7082277536392212,
      "learning_rate": 3.191287878787879e-05,
      "loss": 0.9789,
      "step": 1011
    },
    {
      "epoch": 0.31954531101989264,
      "grad_norm": 0.9095261096954346,
      "learning_rate": 3.194444444444444e-05,
      "loss": 1.0024,
      "step": 1012
    },
    {
      "epoch": 0.3198610672560783,
      "grad_norm": 0.7044939994812012,
      "learning_rate": 3.19760101010101e-05,
      "loss": 1.0131,
      "step": 1013
    },
    {
      "epoch": 0.32017682349226395,
      "grad_norm": 0.6312208771705627,
      "learning_rate": 3.200757575757576e-05,
      "loss": 1.0066,
      "step": 1014
    },
    {
      "epoch": 0.32049257972844963,
      "grad_norm": 0.7143269181251526,
      "learning_rate": 3.203914141414141e-05,
      "loss": 0.9893,
      "step": 1015
    },
    {
      "epoch": 0.3208083359646353,
      "grad_norm": 0.6089757680892944,
      "learning_rate": 3.2070707070707075e-05,
      "loss": 1.0168,
      "step": 1016
    },
    {
      "epoch": 0.32112409220082094,
      "grad_norm": 0.6646566987037659,
      "learning_rate": 3.210227272727273e-05,
      "loss": 0.9927,
      "step": 1017
    },
    {
      "epoch": 0.3214398484370066,
      "grad_norm": 0.6993604302406311,
      "learning_rate": 3.2133838383838385e-05,
      "loss": 1.0423,
      "step": 1018
    },
    {
      "epoch": 0.3217556046731923,
      "grad_norm": 0.7348634600639343,
      "learning_rate": 3.216540404040404e-05,
      "loss": 1.019,
      "step": 1019
    },
    {
      "epoch": 0.32207136090937794,
      "grad_norm": 0.7744389772415161,
      "learning_rate": 3.2196969696969696e-05,
      "loss": 1.0012,
      "step": 1020
    },
    {
      "epoch": 0.3223871171455636,
      "grad_norm": 0.6723289489746094,
      "learning_rate": 3.222853535353536e-05,
      "loss": 0.9718,
      "step": 1021
    },
    {
      "epoch": 0.3227028733817493,
      "grad_norm": 0.6799169778823853,
      "learning_rate": 3.226010101010101e-05,
      "loss": 1.0033,
      "step": 1022
    },
    {
      "epoch": 0.3230186296179349,
      "grad_norm": 0.7522501945495605,
      "learning_rate": 3.229166666666667e-05,
      "loss": 1.0205,
      "step": 1023
    },
    {
      "epoch": 0.3233343858541206,
      "grad_norm": 0.8599655032157898,
      "learning_rate": 3.232323232323233e-05,
      "loss": 1.0231,
      "step": 1024
    },
    {
      "epoch": 0.3236501420903063,
      "grad_norm": 0.7063949704170227,
      "learning_rate": 3.235479797979798e-05,
      "loss": 0.9849,
      "step": 1025
    },
    {
      "epoch": 0.323965898326492,
      "grad_norm": 0.8280907869338989,
      "learning_rate": 3.238636363636364e-05,
      "loss": 1.0165,
      "step": 1026
    },
    {
      "epoch": 0.3242816545626776,
      "grad_norm": 0.8098787665367126,
      "learning_rate": 3.2417929292929295e-05,
      "loss": 0.9963,
      "step": 1027
    },
    {
      "epoch": 0.3245974107988633,
      "grad_norm": 0.621744692325592,
      "learning_rate": 3.244949494949495e-05,
      "loss": 1.0054,
      "step": 1028
    },
    {
      "epoch": 0.32491316703504897,
      "grad_norm": 0.5782994031906128,
      "learning_rate": 3.248106060606061e-05,
      "loss": 0.9839,
      "step": 1029
    },
    {
      "epoch": 0.3252289232712346,
      "grad_norm": 0.7808771133422852,
      "learning_rate": 3.251262626262627e-05,
      "loss": 1.0111,
      "step": 1030
    },
    {
      "epoch": 0.3255446795074203,
      "grad_norm": 0.8721826672554016,
      "learning_rate": 3.2544191919191916e-05,
      "loss": 1.0303,
      "step": 1031
    },
    {
      "epoch": 0.32586043574360596,
      "grad_norm": 0.607285737991333,
      "learning_rate": 3.257575757575758e-05,
      "loss": 1.0069,
      "step": 1032
    },
    {
      "epoch": 0.3261761919797916,
      "grad_norm": 0.7039523124694824,
      "learning_rate": 3.260732323232323e-05,
      "loss": 1.0087,
      "step": 1033
    },
    {
      "epoch": 0.32649194821597727,
      "grad_norm": 0.634604811668396,
      "learning_rate": 3.263888888888889e-05,
      "loss": 1.0581,
      "step": 1034
    },
    {
      "epoch": 0.32680770445216295,
      "grad_norm": 0.658629834651947,
      "learning_rate": 3.267045454545455e-05,
      "loss": 0.9961,
      "step": 1035
    },
    {
      "epoch": 0.3271234606883486,
      "grad_norm": 0.6355416178703308,
      "learning_rate": 3.2702020202020205e-05,
      "loss": 1.0089,
      "step": 1036
    },
    {
      "epoch": 0.32743921692453426,
      "grad_norm": 0.7605310678482056,
      "learning_rate": 3.273358585858586e-05,
      "loss": 1.0049,
      "step": 1037
    },
    {
      "epoch": 0.32775497316071994,
      "grad_norm": 0.7607091069221497,
      "learning_rate": 3.2765151515151515e-05,
      "loss": 0.9912,
      "step": 1038
    },
    {
      "epoch": 0.32807072939690557,
      "grad_norm": 0.9227684140205383,
      "learning_rate": 3.279671717171717e-05,
      "loss": 1.0466,
      "step": 1039
    },
    {
      "epoch": 0.32838648563309125,
      "grad_norm": 0.576265275478363,
      "learning_rate": 3.282828282828283e-05,
      "loss": 0.9664,
      "step": 1040
    },
    {
      "epoch": 0.32870224186927693,
      "grad_norm": 0.6430782079696655,
      "learning_rate": 3.285984848484849e-05,
      "loss": 1.001,
      "step": 1041
    },
    {
      "epoch": 0.32901799810546256,
      "grad_norm": 0.7727339863777161,
      "learning_rate": 3.289141414141414e-05,
      "loss": 1.0231,
      "step": 1042
    },
    {
      "epoch": 0.32933375434164824,
      "grad_norm": 0.6146868467330933,
      "learning_rate": 3.29229797979798e-05,
      "loss": 1.0034,
      "step": 1043
    },
    {
      "epoch": 0.3296495105778339,
      "grad_norm": 0.7475928068161011,
      "learning_rate": 3.295454545454545e-05,
      "loss": 0.9384,
      "step": 1044
    },
    {
      "epoch": 0.32996526681401955,
      "grad_norm": 0.7725535035133362,
      "learning_rate": 3.2986111111111115e-05,
      "loss": 0.9973,
      "step": 1045
    },
    {
      "epoch": 0.33028102305020524,
      "grad_norm": 0.8676018714904785,
      "learning_rate": 3.301767676767677e-05,
      "loss": 0.9835,
      "step": 1046
    },
    {
      "epoch": 0.3305967792863909,
      "grad_norm": 0.6378300189971924,
      "learning_rate": 3.3049242424242425e-05,
      "loss": 1.0139,
      "step": 1047
    },
    {
      "epoch": 0.33091253552257655,
      "grad_norm": 0.8098151087760925,
      "learning_rate": 3.308080808080809e-05,
      "loss": 0.9578,
      "step": 1048
    },
    {
      "epoch": 0.3312282917587622,
      "grad_norm": 0.5275644659996033,
      "learning_rate": 3.3112373737373735e-05,
      "loss": 0.9814,
      "step": 1049
    },
    {
      "epoch": 0.3315440479949479,
      "grad_norm": 0.8283154368400574,
      "learning_rate": 3.314393939393939e-05,
      "loss": 1.0027,
      "step": 1050
    },
    {
      "epoch": 0.33185980423113354,
      "grad_norm": 0.5501520037651062,
      "learning_rate": 3.317550505050505e-05,
      "loss": 1.0075,
      "step": 1051
    },
    {
      "epoch": 0.3321755604673192,
      "grad_norm": 0.7567688822746277,
      "learning_rate": 3.320707070707071e-05,
      "loss": 1.0063,
      "step": 1052
    },
    {
      "epoch": 0.3324913167035049,
      "grad_norm": 0.8525197505950928,
      "learning_rate": 3.323863636363637e-05,
      "loss": 1.0101,
      "step": 1053
    },
    {
      "epoch": 0.3328070729396906,
      "grad_norm": 0.49123162031173706,
      "learning_rate": 3.3270202020202025e-05,
      "loss": 1.0197,
      "step": 1054
    },
    {
      "epoch": 0.3331228291758762,
      "grad_norm": 0.6871312856674194,
      "learning_rate": 3.330176767676767e-05,
      "loss": 1.0059,
      "step": 1055
    },
    {
      "epoch": 0.3334385854120619,
      "grad_norm": 0.47814199328422546,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 1.0208,
      "step": 1056
    },
    {
      "epoch": 0.3337543416482476,
      "grad_norm": 0.794310450553894,
      "learning_rate": 3.336489898989899e-05,
      "loss": 0.9798,
      "step": 1057
    },
    {
      "epoch": 0.3340700978844332,
      "grad_norm": 0.5810162425041199,
      "learning_rate": 3.3396464646464645e-05,
      "loss": 0.9934,
      "step": 1058
    },
    {
      "epoch": 0.3343858541206189,
      "grad_norm": 1.346595048904419,
      "learning_rate": 3.342803030303031e-05,
      "loss": 1.0083,
      "step": 1059
    },
    {
      "epoch": 0.33470161035680457,
      "grad_norm": 0.5400522947311401,
      "learning_rate": 3.345959595959596e-05,
      "loss": 0.9941,
      "step": 1060
    },
    {
      "epoch": 0.3350173665929902,
      "grad_norm": 0.6582149863243103,
      "learning_rate": 3.349116161616162e-05,
      "loss": 1.0136,
      "step": 1061
    },
    {
      "epoch": 0.3353331228291759,
      "grad_norm": 0.522890567779541,
      "learning_rate": 3.352272727272727e-05,
      "loss": 0.9642,
      "step": 1062
    },
    {
      "epoch": 0.33564887906536156,
      "grad_norm": 0.5856013298034668,
      "learning_rate": 3.355429292929293e-05,
      "loss": 1.0125,
      "step": 1063
    },
    {
      "epoch": 0.3359646353015472,
      "grad_norm": 0.6147727370262146,
      "learning_rate": 3.358585858585859e-05,
      "loss": 1.0177,
      "step": 1064
    },
    {
      "epoch": 0.33628039153773287,
      "grad_norm": 0.6346240043640137,
      "learning_rate": 3.3617424242424245e-05,
      "loss": 0.9687,
      "step": 1065
    },
    {
      "epoch": 0.33659614777391855,
      "grad_norm": 0.7577140927314758,
      "learning_rate": 3.36489898989899e-05,
      "loss": 0.9888,
      "step": 1066
    },
    {
      "epoch": 0.3369119040101042,
      "grad_norm": 0.5855777263641357,
      "learning_rate": 3.368055555555556e-05,
      "loss": 1.0097,
      "step": 1067
    },
    {
      "epoch": 0.33722766024628986,
      "grad_norm": 0.7088339328765869,
      "learning_rate": 3.371212121212121e-05,
      "loss": 0.9789,
      "step": 1068
    },
    {
      "epoch": 0.33754341648247554,
      "grad_norm": 0.5507346391677856,
      "learning_rate": 3.374368686868687e-05,
      "loss": 1.0057,
      "step": 1069
    },
    {
      "epoch": 0.33785917271866117,
      "grad_norm": 0.5820751190185547,
      "learning_rate": 3.377525252525253e-05,
      "loss": 0.972,
      "step": 1070
    },
    {
      "epoch": 0.33817492895484685,
      "grad_norm": 0.6237749457359314,
      "learning_rate": 3.380681818181818e-05,
      "loss": 0.9718,
      "step": 1071
    },
    {
      "epoch": 0.33849068519103254,
      "grad_norm": 0.6363495588302612,
      "learning_rate": 3.3838383838383844e-05,
      "loss": 1.0099,
      "step": 1072
    },
    {
      "epoch": 0.33880644142721816,
      "grad_norm": 0.6212546825408936,
      "learning_rate": 3.38699494949495e-05,
      "loss": 0.9784,
      "step": 1073
    },
    {
      "epoch": 0.33912219766340385,
      "grad_norm": 0.6509323716163635,
      "learning_rate": 3.390151515151515e-05,
      "loss": 0.9773,
      "step": 1074
    },
    {
      "epoch": 0.33943795389958953,
      "grad_norm": 0.571377694606781,
      "learning_rate": 3.393308080808081e-05,
      "loss": 1.0104,
      "step": 1075
    },
    {
      "epoch": 0.33975371013577516,
      "grad_norm": 0.6466469764709473,
      "learning_rate": 3.3964646464646465e-05,
      "loss": 0.9783,
      "step": 1076
    },
    {
      "epoch": 0.34006946637196084,
      "grad_norm": 0.7700434923171997,
      "learning_rate": 3.399621212121212e-05,
      "loss": 1.035,
      "step": 1077
    },
    {
      "epoch": 0.3403852226081465,
      "grad_norm": 0.6696737408638,
      "learning_rate": 3.402777777777778e-05,
      "loss": 0.9618,
      "step": 1078
    },
    {
      "epoch": 0.3407009788443322,
      "grad_norm": 0.7347307205200195,
      "learning_rate": 3.405934343434344e-05,
      "loss": 0.9884,
      "step": 1079
    },
    {
      "epoch": 0.34101673508051783,
      "grad_norm": 0.646487295627594,
      "learning_rate": 3.409090909090909e-05,
      "loss": 0.983,
      "step": 1080
    },
    {
      "epoch": 0.3413324913167035,
      "grad_norm": 0.7768090963363647,
      "learning_rate": 3.412247474747475e-05,
      "loss": 1.0269,
      "step": 1081
    },
    {
      "epoch": 0.3416482475528892,
      "grad_norm": 0.6206564903259277,
      "learning_rate": 3.41540404040404e-05,
      "loss": 1.0111,
      "step": 1082
    },
    {
      "epoch": 0.3419640037890748,
      "grad_norm": 0.6319180727005005,
      "learning_rate": 3.4185606060606064e-05,
      "loss": 1.0237,
      "step": 1083
    },
    {
      "epoch": 0.3422797600252605,
      "grad_norm": 0.8255574703216553,
      "learning_rate": 3.421717171717172e-05,
      "loss": 1.013,
      "step": 1084
    },
    {
      "epoch": 0.3425955162614462,
      "grad_norm": 0.8304606676101685,
      "learning_rate": 3.4248737373737375e-05,
      "loss": 1.0033,
      "step": 1085
    },
    {
      "epoch": 0.3429112724976318,
      "grad_norm": 0.8529337048530579,
      "learning_rate": 3.428030303030303e-05,
      "loss": 0.9758,
      "step": 1086
    },
    {
      "epoch": 0.3432270287338175,
      "grad_norm": 0.6529088616371155,
      "learning_rate": 3.4311868686868685e-05,
      "loss": 0.9973,
      "step": 1087
    },
    {
      "epoch": 0.3435427849700032,
      "grad_norm": 0.9011030197143555,
      "learning_rate": 3.434343434343435e-05,
      "loss": 1.0159,
      "step": 1088
    },
    {
      "epoch": 0.3438585412061888,
      "grad_norm": 0.7327485084533691,
      "learning_rate": 3.4375e-05,
      "loss": 0.9826,
      "step": 1089
    },
    {
      "epoch": 0.3441742974423745,
      "grad_norm": 0.6820264458656311,
      "learning_rate": 3.440656565656566e-05,
      "loss": 0.9515,
      "step": 1090
    },
    {
      "epoch": 0.34449005367856017,
      "grad_norm": 0.5996601581573486,
      "learning_rate": 3.443813131313132e-05,
      "loss": 0.9828,
      "step": 1091
    },
    {
      "epoch": 0.3448058099147458,
      "grad_norm": 0.8025076389312744,
      "learning_rate": 3.4469696969696974e-05,
      "loss": 1.0023,
      "step": 1092
    },
    {
      "epoch": 0.3451215661509315,
      "grad_norm": 0.5531458258628845,
      "learning_rate": 3.450126262626262e-05,
      "loss": 1.0179,
      "step": 1093
    },
    {
      "epoch": 0.34543732238711716,
      "grad_norm": 0.5715723037719727,
      "learning_rate": 3.4532828282828285e-05,
      "loss": 0.9837,
      "step": 1094
    },
    {
      "epoch": 0.3457530786233028,
      "grad_norm": 0.5960254669189453,
      "learning_rate": 3.456439393939394e-05,
      "loss": 0.9757,
      "step": 1095
    },
    {
      "epoch": 0.34606883485948847,
      "grad_norm": 0.6595295071601868,
      "learning_rate": 3.45959595959596e-05,
      "loss": 0.9932,
      "step": 1096
    },
    {
      "epoch": 0.34638459109567415,
      "grad_norm": 0.5177561640739441,
      "learning_rate": 3.462752525252526e-05,
      "loss": 1.0122,
      "step": 1097
    },
    {
      "epoch": 0.3467003473318598,
      "grad_norm": 0.7754042148590088,
      "learning_rate": 3.465909090909091e-05,
      "loss": 1.018,
      "step": 1098
    },
    {
      "epoch": 0.34701610356804546,
      "grad_norm": 0.6838176846504211,
      "learning_rate": 3.469065656565657e-05,
      "loss": 0.9884,
      "step": 1099
    },
    {
      "epoch": 0.34733185980423115,
      "grad_norm": 0.6153417229652405,
      "learning_rate": 3.472222222222222e-05,
      "loss": 0.996,
      "step": 1100
    },
    {
      "epoch": 0.3476476160404168,
      "grad_norm": 0.9976968169212341,
      "learning_rate": 3.475378787878788e-05,
      "loss": 1.002,
      "step": 1101
    },
    {
      "epoch": 0.34796337227660246,
      "grad_norm": 0.7072298526763916,
      "learning_rate": 3.478535353535354e-05,
      "loss": 1.0244,
      "step": 1102
    },
    {
      "epoch": 0.34827912851278814,
      "grad_norm": 0.8992977738380432,
      "learning_rate": 3.4816919191919194e-05,
      "loss": 0.982,
      "step": 1103
    },
    {
      "epoch": 0.34859488474897377,
      "grad_norm": 0.8760683536529541,
      "learning_rate": 3.484848484848485e-05,
      "loss": 0.978,
      "step": 1104
    },
    {
      "epoch": 0.34891064098515945,
      "grad_norm": 0.9291649460792542,
      "learning_rate": 3.4880050505050505e-05,
      "loss": 1.0268,
      "step": 1105
    },
    {
      "epoch": 0.34922639722134513,
      "grad_norm": 0.9923272728919983,
      "learning_rate": 3.491161616161616e-05,
      "loss": 0.9724,
      "step": 1106
    },
    {
      "epoch": 0.3495421534575308,
      "grad_norm": 0.8213845491409302,
      "learning_rate": 3.494318181818182e-05,
      "loss": 0.9915,
      "step": 1107
    },
    {
      "epoch": 0.34985790969371644,
      "grad_norm": 0.7946368455886841,
      "learning_rate": 3.497474747474748e-05,
      "loss": 0.9905,
      "step": 1108
    },
    {
      "epoch": 0.3501736659299021,
      "grad_norm": 0.8239975571632385,
      "learning_rate": 3.500631313131313e-05,
      "loss": 0.9826,
      "step": 1109
    },
    {
      "epoch": 0.3504894221660878,
      "grad_norm": 0.6889704465866089,
      "learning_rate": 3.5037878787878794e-05,
      "loss": 0.9873,
      "step": 1110
    },
    {
      "epoch": 0.35080517840227343,
      "grad_norm": 0.658951461315155,
      "learning_rate": 3.506944444444444e-05,
      "loss": 0.9992,
      "step": 1111
    },
    {
      "epoch": 0.3511209346384591,
      "grad_norm": 0.7277817130088806,
      "learning_rate": 3.5101010101010104e-05,
      "loss": 0.9918,
      "step": 1112
    },
    {
      "epoch": 0.3514366908746448,
      "grad_norm": 0.7456276416778564,
      "learning_rate": 3.513257575757576e-05,
      "loss": 0.964,
      "step": 1113
    },
    {
      "epoch": 0.3517524471108304,
      "grad_norm": 0.7973557114601135,
      "learning_rate": 3.5164141414141415e-05,
      "loss": 1.0464,
      "step": 1114
    },
    {
      "epoch": 0.3520682033470161,
      "grad_norm": 0.817034900188446,
      "learning_rate": 3.5195707070707076e-05,
      "loss": 0.9894,
      "step": 1115
    },
    {
      "epoch": 0.3523839595832018,
      "grad_norm": 0.9174278378486633,
      "learning_rate": 3.522727272727273e-05,
      "loss": 0.9626,
      "step": 1116
    },
    {
      "epoch": 0.3526997158193874,
      "grad_norm": 0.6597549319267273,
      "learning_rate": 3.525883838383838e-05,
      "loss": 1.0058,
      "step": 1117
    },
    {
      "epoch": 0.3530154720555731,
      "grad_norm": 0.610905110836029,
      "learning_rate": 3.529040404040404e-05,
      "loss": 0.9618,
      "step": 1118
    },
    {
      "epoch": 0.3533312282917588,
      "grad_norm": 0.716684103012085,
      "learning_rate": 3.53219696969697e-05,
      "loss": 0.9722,
      "step": 1119
    },
    {
      "epoch": 0.3536469845279444,
      "grad_norm": 0.6854890584945679,
      "learning_rate": 3.535353535353535e-05,
      "loss": 1.001,
      "step": 1120
    },
    {
      "epoch": 0.3539627407641301,
      "grad_norm": 0.6511158347129822,
      "learning_rate": 3.5385101010101014e-05,
      "loss": 0.9995,
      "step": 1121
    },
    {
      "epoch": 0.3542784970003158,
      "grad_norm": 0.61820387840271,
      "learning_rate": 3.541666666666667e-05,
      "loss": 0.987,
      "step": 1122
    },
    {
      "epoch": 0.3545942532365014,
      "grad_norm": 0.7611803412437439,
      "learning_rate": 3.5448232323232324e-05,
      "loss": 1.0097,
      "step": 1123
    },
    {
      "epoch": 0.3549100094726871,
      "grad_norm": 0.7866639494895935,
      "learning_rate": 3.547979797979798e-05,
      "loss": 1.0068,
      "step": 1124
    },
    {
      "epoch": 0.35522576570887276,
      "grad_norm": 0.7830290794372559,
      "learning_rate": 3.5511363636363635e-05,
      "loss": 0.9963,
      "step": 1125
    },
    {
      "epoch": 0.3555415219450584,
      "grad_norm": 0.6387908458709717,
      "learning_rate": 3.5542929292929297e-05,
      "loss": 1.0074,
      "step": 1126
    },
    {
      "epoch": 0.3558572781812441,
      "grad_norm": 0.6405563950538635,
      "learning_rate": 3.557449494949495e-05,
      "loss": 0.977,
      "step": 1127
    },
    {
      "epoch": 0.35617303441742976,
      "grad_norm": 0.8481563329696655,
      "learning_rate": 3.560606060606061e-05,
      "loss": 1.01,
      "step": 1128
    },
    {
      "epoch": 0.3564887906536154,
      "grad_norm": 0.5605180859565735,
      "learning_rate": 3.563762626262627e-05,
      "loss": 0.9887,
      "step": 1129
    },
    {
      "epoch": 0.35680454688980107,
      "grad_norm": 0.7315924167633057,
      "learning_rate": 3.566919191919192e-05,
      "loss": 0.9931,
      "step": 1130
    },
    {
      "epoch": 0.35712030312598675,
      "grad_norm": 0.6350427269935608,
      "learning_rate": 3.570075757575758e-05,
      "loss": 1.0182,
      "step": 1131
    },
    {
      "epoch": 0.3574360593621724,
      "grad_norm": 0.7591087818145752,
      "learning_rate": 3.5732323232323234e-05,
      "loss": 0.9483,
      "step": 1132
    },
    {
      "epoch": 0.35775181559835806,
      "grad_norm": 0.6540055871009827,
      "learning_rate": 3.576388888888889e-05,
      "loss": 0.9832,
      "step": 1133
    },
    {
      "epoch": 0.35806757183454374,
      "grad_norm": 0.7144197821617126,
      "learning_rate": 3.579545454545455e-05,
      "loss": 1.0246,
      "step": 1134
    },
    {
      "epoch": 0.3583833280707294,
      "grad_norm": 0.7429230213165283,
      "learning_rate": 3.5827020202020206e-05,
      "loss": 0.9715,
      "step": 1135
    },
    {
      "epoch": 0.35869908430691505,
      "grad_norm": 0.6180798411369324,
      "learning_rate": 3.5858585858585855e-05,
      "loss": 1.0085,
      "step": 1136
    },
    {
      "epoch": 0.35901484054310073,
      "grad_norm": 0.7100344300270081,
      "learning_rate": 3.589015151515152e-05,
      "loss": 0.9709,
      "step": 1137
    },
    {
      "epoch": 0.3593305967792864,
      "grad_norm": 0.7373119592666626,
      "learning_rate": 3.592171717171717e-05,
      "loss": 1.0061,
      "step": 1138
    },
    {
      "epoch": 0.35964635301547204,
      "grad_norm": 0.6233521699905396,
      "learning_rate": 3.5953282828282834e-05,
      "loss": 1.0032,
      "step": 1139
    },
    {
      "epoch": 0.3599621092516577,
      "grad_norm": 0.683080792427063,
      "learning_rate": 3.598484848484849e-05,
      "loss": 0.9648,
      "step": 1140
    },
    {
      "epoch": 0.3602778654878434,
      "grad_norm": 0.6820141673088074,
      "learning_rate": 3.6016414141414144e-05,
      "loss": 1.0261,
      "step": 1141
    },
    {
      "epoch": 0.36059362172402903,
      "grad_norm": 0.7919362783432007,
      "learning_rate": 3.60479797979798e-05,
      "loss": 1.002,
      "step": 1142
    },
    {
      "epoch": 0.3609093779602147,
      "grad_norm": 0.7044487595558167,
      "learning_rate": 3.6079545454545454e-05,
      "loss": 0.9714,
      "step": 1143
    },
    {
      "epoch": 0.3612251341964004,
      "grad_norm": 0.794269859790802,
      "learning_rate": 3.611111111111111e-05,
      "loss": 0.9997,
      "step": 1144
    },
    {
      "epoch": 0.361540890432586,
      "grad_norm": 0.6299226880073547,
      "learning_rate": 3.614267676767677e-05,
      "loss": 1.0264,
      "step": 1145
    },
    {
      "epoch": 0.3618566466687717,
      "grad_norm": 0.7166082859039307,
      "learning_rate": 3.6174242424242427e-05,
      "loss": 0.9535,
      "step": 1146
    },
    {
      "epoch": 0.3621724029049574,
      "grad_norm": 0.8513343334197998,
      "learning_rate": 3.620580808080808e-05,
      "loss": 0.9812,
      "step": 1147
    },
    {
      "epoch": 0.362488159141143,
      "grad_norm": 0.7187953591346741,
      "learning_rate": 3.623737373737374e-05,
      "loss": 1.0047,
      "step": 1148
    },
    {
      "epoch": 0.3628039153773287,
      "grad_norm": 0.8213424682617188,
      "learning_rate": 3.626893939393939e-05,
      "loss": 1.0074,
      "step": 1149
    },
    {
      "epoch": 0.3631196716135144,
      "grad_norm": 0.6491833925247192,
      "learning_rate": 3.6300505050505054e-05,
      "loss": 0.9903,
      "step": 1150
    },
    {
      "epoch": 0.3634354278497,
      "grad_norm": 0.6660723686218262,
      "learning_rate": 3.633207070707071e-05,
      "loss": 0.9958,
      "step": 1151
    },
    {
      "epoch": 0.3637511840858857,
      "grad_norm": 0.8317252993583679,
      "learning_rate": 3.6363636363636364e-05,
      "loss": 0.9886,
      "step": 1152
    },
    {
      "epoch": 0.3640669403220714,
      "grad_norm": 0.9798033237457275,
      "learning_rate": 3.6395202020202026e-05,
      "loss": 0.9771,
      "step": 1153
    },
    {
      "epoch": 0.364382696558257,
      "grad_norm": 0.8097819089889526,
      "learning_rate": 3.6426767676767674e-05,
      "loss": 1.0039,
      "step": 1154
    },
    {
      "epoch": 0.3646984527944427,
      "grad_norm": 0.7780023217201233,
      "learning_rate": 3.6458333333333336e-05,
      "loss": 0.9933,
      "step": 1155
    },
    {
      "epoch": 0.36501420903062837,
      "grad_norm": 0.8147579431533813,
      "learning_rate": 3.648989898989899e-05,
      "loss": 0.9828,
      "step": 1156
    },
    {
      "epoch": 0.365329965266814,
      "grad_norm": 0.9642141461372375,
      "learning_rate": 3.652146464646465e-05,
      "loss": 0.9573,
      "step": 1157
    },
    {
      "epoch": 0.3656457215029997,
      "grad_norm": 0.8264437317848206,
      "learning_rate": 3.655303030303031e-05,
      "loss": 0.9823,
      "step": 1158
    },
    {
      "epoch": 0.36596147773918536,
      "grad_norm": 0.6224873661994934,
      "learning_rate": 3.6584595959595964e-05,
      "loss": 0.9761,
      "step": 1159
    },
    {
      "epoch": 0.366277233975371,
      "grad_norm": 0.7242963314056396,
      "learning_rate": 3.661616161616162e-05,
      "loss": 0.9856,
      "step": 1160
    },
    {
      "epoch": 0.36659299021155667,
      "grad_norm": 0.6752322912216187,
      "learning_rate": 3.6647727272727274e-05,
      "loss": 0.9916,
      "step": 1161
    },
    {
      "epoch": 0.36690874644774235,
      "grad_norm": 0.8522183299064636,
      "learning_rate": 3.667929292929293e-05,
      "loss": 0.9651,
      "step": 1162
    },
    {
      "epoch": 0.36722450268392803,
      "grad_norm": 0.9064584970474243,
      "learning_rate": 3.6710858585858584e-05,
      "loss": 1.0081,
      "step": 1163
    },
    {
      "epoch": 0.36754025892011366,
      "grad_norm": 0.831264317035675,
      "learning_rate": 3.6742424242424246e-05,
      "loss": 0.9843,
      "step": 1164
    },
    {
      "epoch": 0.36785601515629934,
      "grad_norm": 0.6225857138633728,
      "learning_rate": 3.67739898989899e-05,
      "loss": 0.9566,
      "step": 1165
    },
    {
      "epoch": 0.368171771392485,
      "grad_norm": 0.8126637935638428,
      "learning_rate": 3.6805555555555556e-05,
      "loss": 0.998,
      "step": 1166
    },
    {
      "epoch": 0.36848752762867065,
      "grad_norm": 0.7728269696235657,
      "learning_rate": 3.683712121212121e-05,
      "loss": 0.9967,
      "step": 1167
    },
    {
      "epoch": 0.36880328386485633,
      "grad_norm": 0.7859524488449097,
      "learning_rate": 3.686868686868687e-05,
      "loss": 0.9606,
      "step": 1168
    },
    {
      "epoch": 0.369119040101042,
      "grad_norm": 0.7982184886932373,
      "learning_rate": 3.690025252525253e-05,
      "loss": 0.9957,
      "step": 1169
    },
    {
      "epoch": 0.36943479633722764,
      "grad_norm": 0.6003834009170532,
      "learning_rate": 3.6931818181818184e-05,
      "loss": 0.9804,
      "step": 1170
    },
    {
      "epoch": 0.3697505525734133,
      "grad_norm": 0.6733149886131287,
      "learning_rate": 3.696338383838384e-05,
      "loss": 0.9982,
      "step": 1171
    },
    {
      "epoch": 0.370066308809599,
      "grad_norm": 0.7363097667694092,
      "learning_rate": 3.69949494949495e-05,
      "loss": 1.011,
      "step": 1172
    },
    {
      "epoch": 0.37038206504578464,
      "grad_norm": 0.6583015322685242,
      "learning_rate": 3.702651515151515e-05,
      "loss": 0.9747,
      "step": 1173
    },
    {
      "epoch": 0.3706978212819703,
      "grad_norm": 0.8060100674629211,
      "learning_rate": 3.705808080808081e-05,
      "loss": 0.987,
      "step": 1174
    },
    {
      "epoch": 0.371013577518156,
      "grad_norm": 0.7698189616203308,
      "learning_rate": 3.7089646464646466e-05,
      "loss": 0.9797,
      "step": 1175
    },
    {
      "epoch": 0.37132933375434163,
      "grad_norm": 0.8423992395401001,
      "learning_rate": 3.712121212121212e-05,
      "loss": 0.9788,
      "step": 1176
    },
    {
      "epoch": 0.3716450899905273,
      "grad_norm": 0.8121135830879211,
      "learning_rate": 3.715277777777778e-05,
      "loss": 1.0034,
      "step": 1177
    },
    {
      "epoch": 0.371960846226713,
      "grad_norm": 0.9089478850364685,
      "learning_rate": 3.718434343434344e-05,
      "loss": 0.975,
      "step": 1178
    },
    {
      "epoch": 0.3722766024628986,
      "grad_norm": 0.740948498249054,
      "learning_rate": 3.721590909090909e-05,
      "loss": 0.9993,
      "step": 1179
    },
    {
      "epoch": 0.3725923586990843,
      "grad_norm": 0.8299227356910706,
      "learning_rate": 3.724747474747475e-05,
      "loss": 0.997,
      "step": 1180
    },
    {
      "epoch": 0.37290811493527,
      "grad_norm": 0.7918025851249695,
      "learning_rate": 3.7279040404040404e-05,
      "loss": 0.9867,
      "step": 1181
    },
    {
      "epoch": 0.3732238711714556,
      "grad_norm": 0.908842921257019,
      "learning_rate": 3.7310606060606066e-05,
      "loss": 1.0175,
      "step": 1182
    },
    {
      "epoch": 0.3735396274076413,
      "grad_norm": 0.7321208119392395,
      "learning_rate": 3.734217171717172e-05,
      "loss": 0.9843,
      "step": 1183
    },
    {
      "epoch": 0.373855383643827,
      "grad_norm": 0.5178143382072449,
      "learning_rate": 3.7373737373737376e-05,
      "loss": 0.9875,
      "step": 1184
    },
    {
      "epoch": 0.3741711398800126,
      "grad_norm": 0.7063398957252502,
      "learning_rate": 3.740530303030303e-05,
      "loss": 0.9945,
      "step": 1185
    },
    {
      "epoch": 0.3744868961161983,
      "grad_norm": 0.7249438166618347,
      "learning_rate": 3.7436868686868686e-05,
      "loss": 0.9976,
      "step": 1186
    },
    {
      "epoch": 0.37480265235238397,
      "grad_norm": 0.7294358015060425,
      "learning_rate": 3.746843434343434e-05,
      "loss": 0.975,
      "step": 1187
    },
    {
      "epoch": 0.37511840858856965,
      "grad_norm": 0.6643713712692261,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.9817,
      "step": 1188
    },
    {
      "epoch": 0.3754341648247553,
      "grad_norm": 0.8091731071472168,
      "learning_rate": 3.753156565656566e-05,
      "loss": 0.9705,
      "step": 1189
    },
    {
      "epoch": 0.37574992106094096,
      "grad_norm": 0.5901147127151489,
      "learning_rate": 3.7563131313131314e-05,
      "loss": 0.996,
      "step": 1190
    },
    {
      "epoch": 0.37606567729712664,
      "grad_norm": 0.6790263652801514,
      "learning_rate": 3.7594696969696976e-05,
      "loss": 1.0019,
      "step": 1191
    },
    {
      "epoch": 0.37638143353331227,
      "grad_norm": 0.6875151991844177,
      "learning_rate": 3.7626262626262624e-05,
      "loss": 0.9857,
      "step": 1192
    },
    {
      "epoch": 0.37669718976949795,
      "grad_norm": 0.7346109747886658,
      "learning_rate": 3.7657828282828286e-05,
      "loss": 0.9878,
      "step": 1193
    },
    {
      "epoch": 0.37701294600568364,
      "grad_norm": 0.7089853882789612,
      "learning_rate": 3.768939393939394e-05,
      "loss": 0.9984,
      "step": 1194
    },
    {
      "epoch": 0.37732870224186926,
      "grad_norm": 0.6922217011451721,
      "learning_rate": 3.7720959595959596e-05,
      "loss": 0.9976,
      "step": 1195
    },
    {
      "epoch": 0.37764445847805495,
      "grad_norm": 0.6682674288749695,
      "learning_rate": 3.775252525252526e-05,
      "loss": 1.0272,
      "step": 1196
    },
    {
      "epoch": 0.3779602147142406,
      "grad_norm": 0.6861808896064758,
      "learning_rate": 3.778409090909091e-05,
      "loss": 0.9747,
      "step": 1197
    },
    {
      "epoch": 0.37827597095042625,
      "grad_norm": 0.8092034459114075,
      "learning_rate": 3.781565656565657e-05,
      "loss": 0.9743,
      "step": 1198
    },
    {
      "epoch": 0.37859172718661194,
      "grad_norm": 0.6695564389228821,
      "learning_rate": 3.7847222222222224e-05,
      "loss": 0.9854,
      "step": 1199
    },
    {
      "epoch": 0.3789074834227976,
      "grad_norm": 0.6866003274917603,
      "learning_rate": 3.787878787878788e-05,
      "loss": 1.0325,
      "step": 1200
    },
    {
      "epoch": 0.3789074834227976,
      "eval_loss": 0.9559548497200012,
      "eval_runtime": 4.5907,
      "eval_samples_per_second": 542.403,
      "eval_steps_per_second": 4.574,
      "step": 1200
    },
    {
      "epoch": 0.37922323965898325,
      "grad_norm": 0.6326848864555359,
      "learning_rate": 3.791035353535354e-05,
      "loss": 0.9831,
      "step": 1201
    },
    {
      "epoch": 0.37953899589516893,
      "grad_norm": 0.6851119995117188,
      "learning_rate": 3.7941919191919196e-05,
      "loss": 0.9733,
      "step": 1202
    },
    {
      "epoch": 0.3798547521313546,
      "grad_norm": 0.6697164177894592,
      "learning_rate": 3.797348484848485e-05,
      "loss": 0.9576,
      "step": 1203
    },
    {
      "epoch": 0.38017050836754024,
      "grad_norm": 0.8336224555969238,
      "learning_rate": 3.8005050505050506e-05,
      "loss": 0.9846,
      "step": 1204
    },
    {
      "epoch": 0.3804862646037259,
      "grad_norm": 0.6200839281082153,
      "learning_rate": 3.803661616161616e-05,
      "loss": 0.9761,
      "step": 1205
    },
    {
      "epoch": 0.3808020208399116,
      "grad_norm": 0.7171930074691772,
      "learning_rate": 3.8068181818181816e-05,
      "loss": 0.9556,
      "step": 1206
    },
    {
      "epoch": 0.38111777707609723,
      "grad_norm": 0.741933286190033,
      "learning_rate": 3.809974747474748e-05,
      "loss": 0.9552,
      "step": 1207
    },
    {
      "epoch": 0.3814335333122829,
      "grad_norm": 0.6994123458862305,
      "learning_rate": 3.8131313131313133e-05,
      "loss": 0.9954,
      "step": 1208
    },
    {
      "epoch": 0.3817492895484686,
      "grad_norm": 0.8379431366920471,
      "learning_rate": 3.816287878787879e-05,
      "loss": 1.0226,
      "step": 1209
    },
    {
      "epoch": 0.3820650457846542,
      "grad_norm": 0.9153494834899902,
      "learning_rate": 3.8194444444444444e-05,
      "loss": 0.9716,
      "step": 1210
    },
    {
      "epoch": 0.3823808020208399,
      "grad_norm": 3.1063318252563477,
      "learning_rate": 3.82260101010101e-05,
      "loss": 0.9883,
      "step": 1211
    },
    {
      "epoch": 0.3826965582570256,
      "grad_norm": 0.8405888676643372,
      "learning_rate": 3.825757575757576e-05,
      "loss": 0.9646,
      "step": 1212
    },
    {
      "epoch": 0.3830123144932112,
      "grad_norm": 0.7677516937255859,
      "learning_rate": 3.8289141414141416e-05,
      "loss": 0.9832,
      "step": 1213
    },
    {
      "epoch": 0.3833280707293969,
      "grad_norm": 0.8398776650428772,
      "learning_rate": 3.832070707070707e-05,
      "loss": 1.0166,
      "step": 1214
    },
    {
      "epoch": 0.3836438269655826,
      "grad_norm": 0.7933024764060974,
      "learning_rate": 3.835227272727273e-05,
      "loss": 0.9675,
      "step": 1215
    },
    {
      "epoch": 0.38395958320176826,
      "grad_norm": 0.6847470998764038,
      "learning_rate": 3.838383838383838e-05,
      "loss": 0.9981,
      "step": 1216
    },
    {
      "epoch": 0.3842753394379539,
      "grad_norm": 0.8840019702911377,
      "learning_rate": 3.841540404040404e-05,
      "loss": 0.987,
      "step": 1217
    },
    {
      "epoch": 0.38459109567413957,
      "grad_norm": 0.8262635469436646,
      "learning_rate": 3.84469696969697e-05,
      "loss": 0.9676,
      "step": 1218
    },
    {
      "epoch": 0.38490685191032525,
      "grad_norm": 0.7011047005653381,
      "learning_rate": 3.8478535353535354e-05,
      "loss": 0.9888,
      "step": 1219
    },
    {
      "epoch": 0.3852226081465109,
      "grad_norm": 0.7658955454826355,
      "learning_rate": 3.8510101010101015e-05,
      "loss": 0.9785,
      "step": 1220
    },
    {
      "epoch": 0.38553836438269656,
      "grad_norm": 0.7751566767692566,
      "learning_rate": 3.854166666666667e-05,
      "loss": 0.9662,
      "step": 1221
    },
    {
      "epoch": 0.38585412061888225,
      "grad_norm": 0.8437785506248474,
      "learning_rate": 3.8573232323232326e-05,
      "loss": 0.9768,
      "step": 1222
    },
    {
      "epoch": 0.3861698768550679,
      "grad_norm": 0.834753155708313,
      "learning_rate": 3.860479797979798e-05,
      "loss": 0.9758,
      "step": 1223
    },
    {
      "epoch": 0.38648563309125356,
      "grad_norm": 0.784618079662323,
      "learning_rate": 3.8636363636363636e-05,
      "loss": 0.9797,
      "step": 1224
    },
    {
      "epoch": 0.38680138932743924,
      "grad_norm": 0.7388161420822144,
      "learning_rate": 3.86679292929293e-05,
      "loss": 0.9887,
      "step": 1225
    },
    {
      "epoch": 0.38711714556362486,
      "grad_norm": 0.7902185320854187,
      "learning_rate": 3.869949494949495e-05,
      "loss": 0.9837,
      "step": 1226
    },
    {
      "epoch": 0.38743290179981055,
      "grad_norm": 0.9412708282470703,
      "learning_rate": 3.873106060606061e-05,
      "loss": 0.9901,
      "step": 1227
    },
    {
      "epoch": 0.38774865803599623,
      "grad_norm": 0.6889628767967224,
      "learning_rate": 3.876262626262627e-05,
      "loss": 1.0025,
      "step": 1228
    },
    {
      "epoch": 0.38806441427218186,
      "grad_norm": 1.0767282247543335,
      "learning_rate": 3.879419191919192e-05,
      "loss": 0.9799,
      "step": 1229
    },
    {
      "epoch": 0.38838017050836754,
      "grad_norm": 0.7968221306800842,
      "learning_rate": 3.8825757575757574e-05,
      "loss": 0.9864,
      "step": 1230
    },
    {
      "epoch": 0.3886959267445532,
      "grad_norm": 0.7500074505805969,
      "learning_rate": 3.8857323232323236e-05,
      "loss": 1.0166,
      "step": 1231
    },
    {
      "epoch": 0.38901168298073885,
      "grad_norm": 0.7204249501228333,
      "learning_rate": 3.888888888888889e-05,
      "loss": 0.9891,
      "step": 1232
    },
    {
      "epoch": 0.38932743921692453,
      "grad_norm": 0.7408673763275146,
      "learning_rate": 3.8920454545454546e-05,
      "loss": 1.0592,
      "step": 1233
    },
    {
      "epoch": 0.3896431954531102,
      "grad_norm": 0.94332355260849,
      "learning_rate": 3.895202020202021e-05,
      "loss": 0.977,
      "step": 1234
    },
    {
      "epoch": 0.38995895168929584,
      "grad_norm": 0.7871600389480591,
      "learning_rate": 3.8983585858585856e-05,
      "loss": 0.9907,
      "step": 1235
    },
    {
      "epoch": 0.3902747079254815,
      "grad_norm": 0.7921582460403442,
      "learning_rate": 3.901515151515152e-05,
      "loss": 1.0264,
      "step": 1236
    },
    {
      "epoch": 0.3905904641616672,
      "grad_norm": 0.8020675182342529,
      "learning_rate": 3.904671717171717e-05,
      "loss": 0.9846,
      "step": 1237
    },
    {
      "epoch": 0.39090622039785283,
      "grad_norm": 0.8091981410980225,
      "learning_rate": 3.907828282828283e-05,
      "loss": 0.961,
      "step": 1238
    },
    {
      "epoch": 0.3912219766340385,
      "grad_norm": 0.5942662358283997,
      "learning_rate": 3.910984848484849e-05,
      "loss": 1.016,
      "step": 1239
    },
    {
      "epoch": 0.3915377328702242,
      "grad_norm": 0.7669585347175598,
      "learning_rate": 3.9141414141414145e-05,
      "loss": 0.9857,
      "step": 1240
    },
    {
      "epoch": 0.3918534891064098,
      "grad_norm": 0.6214404702186584,
      "learning_rate": 3.91729797979798e-05,
      "loss": 0.9579,
      "step": 1241
    },
    {
      "epoch": 0.3921692453425955,
      "grad_norm": 0.9249640703201294,
      "learning_rate": 3.9204545454545456e-05,
      "loss": 0.9606,
      "step": 1242
    },
    {
      "epoch": 0.3924850015787812,
      "grad_norm": 0.660009503364563,
      "learning_rate": 3.923611111111111e-05,
      "loss": 0.9935,
      "step": 1243
    },
    {
      "epoch": 0.39280075781496687,
      "grad_norm": 0.9294548034667969,
      "learning_rate": 3.926767676767677e-05,
      "loss": 1.0259,
      "step": 1244
    },
    {
      "epoch": 0.3931165140511525,
      "grad_norm": 0.6113893389701843,
      "learning_rate": 3.929924242424243e-05,
      "loss": 0.9749,
      "step": 1245
    },
    {
      "epoch": 0.3934322702873382,
      "grad_norm": 0.6945608258247375,
      "learning_rate": 3.933080808080808e-05,
      "loss": 0.9817,
      "step": 1246
    },
    {
      "epoch": 0.39374802652352386,
      "grad_norm": 0.8077424168586731,
      "learning_rate": 3.936237373737374e-05,
      "loss": 1.0045,
      "step": 1247
    },
    {
      "epoch": 0.3940637827597095,
      "grad_norm": 0.807155966758728,
      "learning_rate": 3.939393939393939e-05,
      "loss": 0.9845,
      "step": 1248
    },
    {
      "epoch": 0.3943795389958952,
      "grad_norm": 0.7513564825057983,
      "learning_rate": 3.942550505050505e-05,
      "loss": 0.9575,
      "step": 1249
    },
    {
      "epoch": 0.39469529523208086,
      "grad_norm": 0.769145667552948,
      "learning_rate": 3.945707070707071e-05,
      "loss": 0.9943,
      "step": 1250
    },
    {
      "epoch": 0.3950110514682665,
      "grad_norm": 0.7344409227371216,
      "learning_rate": 3.9488636363636366e-05,
      "loss": 1.0351,
      "step": 1251
    },
    {
      "epoch": 0.39532680770445217,
      "grad_norm": 0.8368252515792847,
      "learning_rate": 3.952020202020202e-05,
      "loss": 1.0032,
      "step": 1252
    },
    {
      "epoch": 0.39564256394063785,
      "grad_norm": 0.6399998664855957,
      "learning_rate": 3.9551767676767676e-05,
      "loss": 0.963,
      "step": 1253
    },
    {
      "epoch": 0.3959583201768235,
      "grad_norm": 0.9310258626937866,
      "learning_rate": 3.958333333333333e-05,
      "loss": 0.9543,
      "step": 1254
    },
    {
      "epoch": 0.39627407641300916,
      "grad_norm": 0.6524232029914856,
      "learning_rate": 3.961489898989899e-05,
      "loss": 0.9349,
      "step": 1255
    },
    {
      "epoch": 0.39658983264919484,
      "grad_norm": 0.6935328841209412,
      "learning_rate": 3.964646464646465e-05,
      "loss": 0.9888,
      "step": 1256
    },
    {
      "epoch": 0.39690558888538047,
      "grad_norm": 0.5882509350776672,
      "learning_rate": 3.96780303030303e-05,
      "loss": 1.0421,
      "step": 1257
    },
    {
      "epoch": 0.39722134512156615,
      "grad_norm": 0.7925480008125305,
      "learning_rate": 3.9709595959595965e-05,
      "loss": 0.9967,
      "step": 1258
    },
    {
      "epoch": 0.39753710135775183,
      "grad_norm": 0.6446495652198792,
      "learning_rate": 3.974116161616162e-05,
      "loss": 1.0041,
      "step": 1259
    },
    {
      "epoch": 0.39785285759393746,
      "grad_norm": 0.8938742280006409,
      "learning_rate": 3.9772727272727275e-05,
      "loss": 0.9817,
      "step": 1260
    },
    {
      "epoch": 0.39816861383012314,
      "grad_norm": 0.7365705966949463,
      "learning_rate": 3.980429292929293e-05,
      "loss": 1.0034,
      "step": 1261
    },
    {
      "epoch": 0.3984843700663088,
      "grad_norm": 0.6947631239891052,
      "learning_rate": 3.9835858585858586e-05,
      "loss": 1.0027,
      "step": 1262
    },
    {
      "epoch": 0.39880012630249445,
      "grad_norm": 0.6903889775276184,
      "learning_rate": 3.986742424242425e-05,
      "loss": 1.0166,
      "step": 1263
    },
    {
      "epoch": 0.39911588253868013,
      "grad_norm": 0.7091154456138611,
      "learning_rate": 3.98989898989899e-05,
      "loss": 0.9815,
      "step": 1264
    },
    {
      "epoch": 0.3994316387748658,
      "grad_norm": 0.7267081141471863,
      "learning_rate": 3.993055555555556e-05,
      "loss": 0.963,
      "step": 1265
    },
    {
      "epoch": 0.39974739501105144,
      "grad_norm": 0.7413873076438904,
      "learning_rate": 3.996212121212121e-05,
      "loss": 0.9839,
      "step": 1266
    },
    {
      "epoch": 0.4000631512472371,
      "grad_norm": 0.6025086045265198,
      "learning_rate": 3.999368686868687e-05,
      "loss": 0.988,
      "step": 1267
    },
    {
      "epoch": 0.4003789074834228,
      "grad_norm": 0.5692706108093262,
      "learning_rate": 4.002525252525253e-05,
      "loss": 0.9817,
      "step": 1268
    },
    {
      "epoch": 0.40069466371960843,
      "grad_norm": 0.691032350063324,
      "learning_rate": 4.0056818181818185e-05,
      "loss": 0.9785,
      "step": 1269
    },
    {
      "epoch": 0.4010104199557941,
      "grad_norm": 0.7036435008049011,
      "learning_rate": 4.008838383838384e-05,
      "loss": 0.9882,
      "step": 1270
    },
    {
      "epoch": 0.4013261761919798,
      "grad_norm": 0.6868932247161865,
      "learning_rate": 4.01199494949495e-05,
      "loss": 0.9906,
      "step": 1271
    },
    {
      "epoch": 0.4016419324281655,
      "grad_norm": 0.8335798382759094,
      "learning_rate": 4.015151515151515e-05,
      "loss": 0.9739,
      "step": 1272
    },
    {
      "epoch": 0.4019576886643511,
      "grad_norm": 0.8513293266296387,
      "learning_rate": 4.0183080808080806e-05,
      "loss": 0.9829,
      "step": 1273
    },
    {
      "epoch": 0.4022734449005368,
      "grad_norm": 0.796454131603241,
      "learning_rate": 4.021464646464647e-05,
      "loss": 0.9502,
      "step": 1274
    },
    {
      "epoch": 0.4025892011367225,
      "grad_norm": 0.841015636920929,
      "learning_rate": 4.024621212121212e-05,
      "loss": 0.9945,
      "step": 1275
    },
    {
      "epoch": 0.4029049573729081,
      "grad_norm": 0.5646660327911377,
      "learning_rate": 4.027777777777778e-05,
      "loss": 0.9937,
      "step": 1276
    },
    {
      "epoch": 0.4032207136090938,
      "grad_norm": 0.8936665058135986,
      "learning_rate": 4.030934343434344e-05,
      "loss": 0.9776,
      "step": 1277
    },
    {
      "epoch": 0.40353646984527947,
      "grad_norm": 0.8422437310218811,
      "learning_rate": 4.034090909090909e-05,
      "loss": 0.9829,
      "step": 1278
    },
    {
      "epoch": 0.4038522260814651,
      "grad_norm": 0.936903178691864,
      "learning_rate": 4.037247474747475e-05,
      "loss": 0.9632,
      "step": 1279
    },
    {
      "epoch": 0.4041679823176508,
      "grad_norm": 0.9644806981086731,
      "learning_rate": 4.0404040404040405e-05,
      "loss": 0.9882,
      "step": 1280
    },
    {
      "epoch": 0.40448373855383646,
      "grad_norm": 0.6993728280067444,
      "learning_rate": 4.043560606060606e-05,
      "loss": 0.9492,
      "step": 1281
    },
    {
      "epoch": 0.4047994947900221,
      "grad_norm": 0.8383119106292725,
      "learning_rate": 4.046717171717172e-05,
      "loss": 0.96,
      "step": 1282
    },
    {
      "epoch": 0.40511525102620777,
      "grad_norm": 0.7728784084320068,
      "learning_rate": 4.049873737373738e-05,
      "loss": 0.988,
      "step": 1283
    },
    {
      "epoch": 0.40543100726239345,
      "grad_norm": 0.7005175948143005,
      "learning_rate": 4.053030303030303e-05,
      "loss": 1.0013,
      "step": 1284
    },
    {
      "epoch": 0.4057467634985791,
      "grad_norm": 0.6723203063011169,
      "learning_rate": 4.056186868686869e-05,
      "loss": 0.9715,
      "step": 1285
    },
    {
      "epoch": 0.40606251973476476,
      "grad_norm": 0.7551887631416321,
      "learning_rate": 4.059343434343434e-05,
      "loss": 0.9709,
      "step": 1286
    },
    {
      "epoch": 0.40637827597095044,
      "grad_norm": 0.7594444751739502,
      "learning_rate": 4.0625000000000005e-05,
      "loss": 0.97,
      "step": 1287
    },
    {
      "epoch": 0.40669403220713607,
      "grad_norm": 0.7541806101799011,
      "learning_rate": 4.065656565656566e-05,
      "loss": 0.9818,
      "step": 1288
    },
    {
      "epoch": 0.40700978844332175,
      "grad_norm": 0.6341336369514465,
      "learning_rate": 4.0688131313131315e-05,
      "loss": 0.9636,
      "step": 1289
    },
    {
      "epoch": 0.40732554467950743,
      "grad_norm": 0.8302831649780273,
      "learning_rate": 4.071969696969698e-05,
      "loss": 0.9933,
      "step": 1290
    },
    {
      "epoch": 0.40764130091569306,
      "grad_norm": 0.5830496549606323,
      "learning_rate": 4.0751262626262625e-05,
      "loss": 0.9937,
      "step": 1291
    },
    {
      "epoch": 0.40795705715187874,
      "grad_norm": 0.5340694189071655,
      "learning_rate": 4.078282828282828e-05,
      "loss": 0.9722,
      "step": 1292
    },
    {
      "epoch": 0.4082728133880644,
      "grad_norm": 0.7567058801651001,
      "learning_rate": 4.081439393939394e-05,
      "loss": 0.9728,
      "step": 1293
    },
    {
      "epoch": 0.40858856962425005,
      "grad_norm": 0.7292104959487915,
      "learning_rate": 4.08459595959596e-05,
      "loss": 1.036,
      "step": 1294
    },
    {
      "epoch": 0.40890432586043574,
      "grad_norm": 0.7865761518478394,
      "learning_rate": 4.087752525252525e-05,
      "loss": 0.9928,
      "step": 1295
    },
    {
      "epoch": 0.4092200820966214,
      "grad_norm": 0.7028623223304749,
      "learning_rate": 4.0909090909090915e-05,
      "loss": 0.9904,
      "step": 1296
    },
    {
      "epoch": 0.4095358383328071,
      "grad_norm": 0.6653420329093933,
      "learning_rate": 4.094065656565656e-05,
      "loss": 0.9498,
      "step": 1297
    },
    {
      "epoch": 0.4098515945689927,
      "grad_norm": 0.7549835443496704,
      "learning_rate": 4.0972222222222225e-05,
      "loss": 0.9839,
      "step": 1298
    },
    {
      "epoch": 0.4101673508051784,
      "grad_norm": 0.7078313231468201,
      "learning_rate": 4.100378787878788e-05,
      "loss": 0.9661,
      "step": 1299
    },
    {
      "epoch": 0.4104831070413641,
      "grad_norm": 0.6736873984336853,
      "learning_rate": 4.1035353535353535e-05,
      "loss": 0.9813,
      "step": 1300
    },
    {
      "epoch": 0.4107988632775497,
      "grad_norm": 0.687944233417511,
      "learning_rate": 4.10669191919192e-05,
      "loss": 0.958,
      "step": 1301
    },
    {
      "epoch": 0.4111146195137354,
      "grad_norm": 0.7532490491867065,
      "learning_rate": 4.109848484848485e-05,
      "loss": 0.9848,
      "step": 1302
    },
    {
      "epoch": 0.4114303757499211,
      "grad_norm": 0.7634623050689697,
      "learning_rate": 4.113005050505051e-05,
      "loss": 0.9446,
      "step": 1303
    },
    {
      "epoch": 0.4117461319861067,
      "grad_norm": 0.6363627314567566,
      "learning_rate": 4.116161616161616e-05,
      "loss": 1.0074,
      "step": 1304
    },
    {
      "epoch": 0.4120618882222924,
      "grad_norm": 0.704836905002594,
      "learning_rate": 4.119318181818182e-05,
      "loss": 0.9881,
      "step": 1305
    },
    {
      "epoch": 0.4123776444584781,
      "grad_norm": 0.8419808149337769,
      "learning_rate": 4.122474747474748e-05,
      "loss": 0.9566,
      "step": 1306
    },
    {
      "epoch": 0.4126934006946637,
      "grad_norm": 0.9840209484100342,
      "learning_rate": 4.1256313131313135e-05,
      "loss": 0.9652,
      "step": 1307
    },
    {
      "epoch": 0.4130091569308494,
      "grad_norm": 1.0579534769058228,
      "learning_rate": 4.128787878787879e-05,
      "loss": 0.9953,
      "step": 1308
    },
    {
      "epoch": 0.41332491316703507,
      "grad_norm": 0.8467779159545898,
      "learning_rate": 4.1319444444444445e-05,
      "loss": 0.9737,
      "step": 1309
    },
    {
      "epoch": 0.4136406694032207,
      "grad_norm": 0.7199661135673523,
      "learning_rate": 4.13510101010101e-05,
      "loss": 0.9796,
      "step": 1310
    },
    {
      "epoch": 0.4139564256394064,
      "grad_norm": 0.838363766670227,
      "learning_rate": 4.138257575757576e-05,
      "loss": 0.9499,
      "step": 1311
    },
    {
      "epoch": 0.41427218187559206,
      "grad_norm": 0.7254040241241455,
      "learning_rate": 4.141414141414142e-05,
      "loss": 0.9812,
      "step": 1312
    },
    {
      "epoch": 0.4145879381117777,
      "grad_norm": 0.75996994972229,
      "learning_rate": 4.144570707070707e-05,
      "loss": 0.9675,
      "step": 1313
    },
    {
      "epoch": 0.41490369434796337,
      "grad_norm": 0.8336907029151917,
      "learning_rate": 4.1477272727272734e-05,
      "loss": 0.9841,
      "step": 1314
    },
    {
      "epoch": 0.41521945058414905,
      "grad_norm": 0.579147458076477,
      "learning_rate": 4.150883838383838e-05,
      "loss": 0.9743,
      "step": 1315
    },
    {
      "epoch": 0.4155352068203347,
      "grad_norm": 0.7560136318206787,
      "learning_rate": 4.154040404040404e-05,
      "loss": 0.986,
      "step": 1316
    },
    {
      "epoch": 0.41585096305652036,
      "grad_norm": 0.7641900181770325,
      "learning_rate": 4.15719696969697e-05,
      "loss": 1.0025,
      "step": 1317
    },
    {
      "epoch": 0.41616671929270604,
      "grad_norm": 0.6760154366493225,
      "learning_rate": 4.1603535353535355e-05,
      "loss": 0.9912,
      "step": 1318
    },
    {
      "epoch": 0.41648247552889167,
      "grad_norm": 0.7439906001091003,
      "learning_rate": 4.163510101010101e-05,
      "loss": 0.9765,
      "step": 1319
    },
    {
      "epoch": 0.41679823176507735,
      "grad_norm": 0.6147472858428955,
      "learning_rate": 4.166666666666667e-05,
      "loss": 0.9668,
      "step": 1320
    },
    {
      "epoch": 0.41711398800126304,
      "grad_norm": 0.7626285552978516,
      "learning_rate": 4.169823232323233e-05,
      "loss": 0.9692,
      "step": 1321
    },
    {
      "epoch": 0.41742974423744866,
      "grad_norm": 0.7155501246452332,
      "learning_rate": 4.172979797979798e-05,
      "loss": 0.9821,
      "step": 1322
    },
    {
      "epoch": 0.41774550047363435,
      "grad_norm": 0.6344617605209351,
      "learning_rate": 4.176136363636364e-05,
      "loss": 1.0041,
      "step": 1323
    },
    {
      "epoch": 0.41806125670982003,
      "grad_norm": 0.6916806101799011,
      "learning_rate": 4.179292929292929e-05,
      "loss": 0.9421,
      "step": 1324
    },
    {
      "epoch": 0.4183770129460057,
      "grad_norm": 0.6895758509635925,
      "learning_rate": 4.1824494949494955e-05,
      "loss": 0.9868,
      "step": 1325
    },
    {
      "epoch": 0.41869276918219134,
      "grad_norm": 0.6114411950111389,
      "learning_rate": 4.185606060606061e-05,
      "loss": 0.9401,
      "step": 1326
    },
    {
      "epoch": 0.419008525418377,
      "grad_norm": 0.6226833462715149,
      "learning_rate": 4.1887626262626265e-05,
      "loss": 0.9452,
      "step": 1327
    },
    {
      "epoch": 0.4193242816545627,
      "grad_norm": 0.6323116421699524,
      "learning_rate": 4.191919191919192e-05,
      "loss": 0.9563,
      "step": 1328
    },
    {
      "epoch": 0.41964003789074833,
      "grad_norm": 0.8543366193771362,
      "learning_rate": 4.1950757575757575e-05,
      "loss": 0.9526,
      "step": 1329
    },
    {
      "epoch": 0.419955794126934,
      "grad_norm": 0.7573553323745728,
      "learning_rate": 4.198232323232324e-05,
      "loss": 0.9576,
      "step": 1330
    },
    {
      "epoch": 0.4202715503631197,
      "grad_norm": 0.7145471572875977,
      "learning_rate": 4.201388888888889e-05,
      "loss": 0.9386,
      "step": 1331
    },
    {
      "epoch": 0.4205873065993053,
      "grad_norm": 0.6834836006164551,
      "learning_rate": 4.204545454545455e-05,
      "loss": 0.9612,
      "step": 1332
    },
    {
      "epoch": 0.420903062835491,
      "grad_norm": 0.8159952759742737,
      "learning_rate": 4.207702020202021e-05,
      "loss": 0.9514,
      "step": 1333
    },
    {
      "epoch": 0.4212188190716767,
      "grad_norm": 0.7535727024078369,
      "learning_rate": 4.210858585858586e-05,
      "loss": 0.9795,
      "step": 1334
    },
    {
      "epoch": 0.4215345753078623,
      "grad_norm": 0.7822113633155823,
      "learning_rate": 4.214015151515151e-05,
      "loss": 0.9231,
      "step": 1335
    },
    {
      "epoch": 0.421850331544048,
      "grad_norm": 0.6913536787033081,
      "learning_rate": 4.2171717171717175e-05,
      "loss": 0.9743,
      "step": 1336
    },
    {
      "epoch": 0.4221660877802337,
      "grad_norm": 0.6682687997817993,
      "learning_rate": 4.220328282828283e-05,
      "loss": 0.9774,
      "step": 1337
    },
    {
      "epoch": 0.4224818440164193,
      "grad_norm": 0.6829761266708374,
      "learning_rate": 4.2234848484848485e-05,
      "loss": 0.9796,
      "step": 1338
    },
    {
      "epoch": 0.422797600252605,
      "grad_norm": 0.7489904761314392,
      "learning_rate": 4.226641414141415e-05,
      "loss": 0.9812,
      "step": 1339
    },
    {
      "epoch": 0.42311335648879067,
      "grad_norm": 0.6775611639022827,
      "learning_rate": 4.2297979797979795e-05,
      "loss": 1.017,
      "step": 1340
    },
    {
      "epoch": 0.4234291127249763,
      "grad_norm": 0.7233102917671204,
      "learning_rate": 4.232954545454546e-05,
      "loss": 1.0236,
      "step": 1341
    },
    {
      "epoch": 0.423744868961162,
      "grad_norm": 0.7234300971031189,
      "learning_rate": 4.236111111111111e-05,
      "loss": 0.9663,
      "step": 1342
    },
    {
      "epoch": 0.42406062519734766,
      "grad_norm": 0.9837728142738342,
      "learning_rate": 4.239267676767677e-05,
      "loss": 0.9823,
      "step": 1343
    },
    {
      "epoch": 0.4243763814335333,
      "grad_norm": 0.6758604645729065,
      "learning_rate": 4.242424242424243e-05,
      "loss": 0.9854,
      "step": 1344
    },
    {
      "epoch": 0.42469213766971897,
      "grad_norm": 0.693645715713501,
      "learning_rate": 4.2455808080808084e-05,
      "loss": 0.9801,
      "step": 1345
    },
    {
      "epoch": 0.42500789390590465,
      "grad_norm": 0.8205156922340393,
      "learning_rate": 4.248737373737374e-05,
      "loss": 0.9769,
      "step": 1346
    },
    {
      "epoch": 0.4253236501420903,
      "grad_norm": 0.8270080089569092,
      "learning_rate": 4.2518939393939395e-05,
      "loss": 1.015,
      "step": 1347
    },
    {
      "epoch": 0.42563940637827596,
      "grad_norm": 0.6055446863174438,
      "learning_rate": 4.255050505050505e-05,
      "loss": 0.9776,
      "step": 1348
    },
    {
      "epoch": 0.42595516261446165,
      "grad_norm": 0.7354779243469238,
      "learning_rate": 4.258207070707071e-05,
      "loss": 1.0052,
      "step": 1349
    },
    {
      "epoch": 0.4262709188506473,
      "grad_norm": 0.6708642244338989,
      "learning_rate": 4.261363636363637e-05,
      "loss": 0.978,
      "step": 1350
    },
    {
      "epoch": 0.42658667508683296,
      "grad_norm": 0.6680448651313782,
      "learning_rate": 4.264520202020202e-05,
      "loss": 0.9707,
      "step": 1351
    },
    {
      "epoch": 0.42690243132301864,
      "grad_norm": 0.5953482985496521,
      "learning_rate": 4.267676767676768e-05,
      "loss": 1.0011,
      "step": 1352
    },
    {
      "epoch": 0.4272181875592043,
      "grad_norm": 0.8412445187568665,
      "learning_rate": 4.270833333333333e-05,
      "loss": 1.023,
      "step": 1353
    },
    {
      "epoch": 0.42753394379538995,
      "grad_norm": 0.6848497986793518,
      "learning_rate": 4.2739898989898994e-05,
      "loss": 0.9689,
      "step": 1354
    },
    {
      "epoch": 0.42784970003157563,
      "grad_norm": 0.778521716594696,
      "learning_rate": 4.277146464646465e-05,
      "loss": 0.968,
      "step": 1355
    },
    {
      "epoch": 0.4281654562677613,
      "grad_norm": 0.6313483715057373,
      "learning_rate": 4.2803030303030305e-05,
      "loss": 0.9911,
      "step": 1356
    },
    {
      "epoch": 0.42848121250394694,
      "grad_norm": 0.8500220775604248,
      "learning_rate": 4.2834595959595967e-05,
      "loss": 0.9833,
      "step": 1357
    },
    {
      "epoch": 0.4287969687401326,
      "grad_norm": 0.7047507166862488,
      "learning_rate": 4.286616161616162e-05,
      "loss": 0.9872,
      "step": 1358
    },
    {
      "epoch": 0.4291127249763183,
      "grad_norm": 0.8311727643013,
      "learning_rate": 4.289772727272727e-05,
      "loss": 1.0104,
      "step": 1359
    },
    {
      "epoch": 0.42942848121250393,
      "grad_norm": 0.6853487491607666,
      "learning_rate": 4.292929292929293e-05,
      "loss": 0.9568,
      "step": 1360
    },
    {
      "epoch": 0.4297442374486896,
      "grad_norm": 0.8327111005783081,
      "learning_rate": 4.296085858585859e-05,
      "loss": 1.0116,
      "step": 1361
    },
    {
      "epoch": 0.4300599936848753,
      "grad_norm": 0.8667533993721008,
      "learning_rate": 4.299242424242424e-05,
      "loss": 0.9788,
      "step": 1362
    },
    {
      "epoch": 0.4303757499210609,
      "grad_norm": 0.733622133731842,
      "learning_rate": 4.3023989898989904e-05,
      "loss": 0.9811,
      "step": 1363
    },
    {
      "epoch": 0.4306915061572466,
      "grad_norm": 0.6578850150108337,
      "learning_rate": 4.305555555555556e-05,
      "loss": 0.9404,
      "step": 1364
    },
    {
      "epoch": 0.4310072623934323,
      "grad_norm": 0.7624351978302002,
      "learning_rate": 4.3087121212121214e-05,
      "loss": 0.9841,
      "step": 1365
    },
    {
      "epoch": 0.4313230186296179,
      "grad_norm": 0.7737752199172974,
      "learning_rate": 4.311868686868687e-05,
      "loss": 0.9369,
      "step": 1366
    },
    {
      "epoch": 0.4316387748658036,
      "grad_norm": 0.6851634979248047,
      "learning_rate": 4.3150252525252525e-05,
      "loss": 1.0141,
      "step": 1367
    },
    {
      "epoch": 0.4319545311019893,
      "grad_norm": 0.922941267490387,
      "learning_rate": 4.318181818181819e-05,
      "loss": 0.9371,
      "step": 1368
    },
    {
      "epoch": 0.4322702873381749,
      "grad_norm": 0.6621057987213135,
      "learning_rate": 4.321338383838384e-05,
      "loss": 0.963,
      "step": 1369
    },
    {
      "epoch": 0.4325860435743606,
      "grad_norm": 0.7303131222724915,
      "learning_rate": 4.32449494949495e-05,
      "loss": 0.9297,
      "step": 1370
    },
    {
      "epoch": 0.4329017998105463,
      "grad_norm": 0.8185912370681763,
      "learning_rate": 4.327651515151515e-05,
      "loss": 0.9284,
      "step": 1371
    },
    {
      "epoch": 0.4332175560467319,
      "grad_norm": 0.8840174078941345,
      "learning_rate": 4.330808080808081e-05,
      "loss": 0.9653,
      "step": 1372
    },
    {
      "epoch": 0.4335333122829176,
      "grad_norm": 1.0140799283981323,
      "learning_rate": 4.333964646464647e-05,
      "loss": 0.9693,
      "step": 1373
    },
    {
      "epoch": 0.43384906851910326,
      "grad_norm": 0.8291605114936829,
      "learning_rate": 4.3371212121212124e-05,
      "loss": 0.9453,
      "step": 1374
    },
    {
      "epoch": 0.4341648247552889,
      "grad_norm": 0.8088762760162354,
      "learning_rate": 4.340277777777778e-05,
      "loss": 0.9802,
      "step": 1375
    },
    {
      "epoch": 0.4344805809914746,
      "grad_norm": 0.7709742784500122,
      "learning_rate": 4.343434343434344e-05,
      "loss": 1.0236,
      "step": 1376
    },
    {
      "epoch": 0.43479633722766026,
      "grad_norm": 0.6838432550430298,
      "learning_rate": 4.346590909090909e-05,
      "loss": 0.9895,
      "step": 1377
    },
    {
      "epoch": 0.4351120934638459,
      "grad_norm": 0.7593288421630859,
      "learning_rate": 4.3497474747474745e-05,
      "loss": 1.0161,
      "step": 1378
    },
    {
      "epoch": 0.43542784970003157,
      "grad_norm": 0.9614288806915283,
      "learning_rate": 4.352904040404041e-05,
      "loss": 0.9664,
      "step": 1379
    },
    {
      "epoch": 0.43574360593621725,
      "grad_norm": 0.7021021246910095,
      "learning_rate": 4.356060606060606e-05,
      "loss": 0.9026,
      "step": 1380
    },
    {
      "epoch": 0.43605936217240293,
      "grad_norm": 0.8286967873573303,
      "learning_rate": 4.359217171717172e-05,
      "loss": 0.9748,
      "step": 1381
    },
    {
      "epoch": 0.43637511840858856,
      "grad_norm": 0.8332382440567017,
      "learning_rate": 4.362373737373738e-05,
      "loss": 0.9515,
      "step": 1382
    },
    {
      "epoch": 0.43669087464477424,
      "grad_norm": 1.1498082876205444,
      "learning_rate": 4.365530303030303e-05,
      "loss": 0.9952,
      "step": 1383
    },
    {
      "epoch": 0.4370066308809599,
      "grad_norm": 1.0594711303710938,
      "learning_rate": 4.368686868686869e-05,
      "loss": 1.0019,
      "step": 1384
    },
    {
      "epoch": 0.43732238711714555,
      "grad_norm": 0.9333885908126831,
      "learning_rate": 4.3718434343434344e-05,
      "loss": 1.0116,
      "step": 1385
    },
    {
      "epoch": 0.43763814335333123,
      "grad_norm": 0.9973906874656677,
      "learning_rate": 4.375e-05,
      "loss": 0.9961,
      "step": 1386
    },
    {
      "epoch": 0.4379538995895169,
      "grad_norm": 0.730192244052887,
      "learning_rate": 4.378156565656566e-05,
      "loss": 0.9268,
      "step": 1387
    },
    {
      "epoch": 0.43826965582570254,
      "grad_norm": 0.8898041844367981,
      "learning_rate": 4.381313131313132e-05,
      "loss": 0.9647,
      "step": 1388
    },
    {
      "epoch": 0.4385854120618882,
      "grad_norm": 0.9104648232460022,
      "learning_rate": 4.384469696969697e-05,
      "loss": 0.9871,
      "step": 1389
    },
    {
      "epoch": 0.4389011682980739,
      "grad_norm": 1.0719940662384033,
      "learning_rate": 4.387626262626263e-05,
      "loss": 0.9426,
      "step": 1390
    },
    {
      "epoch": 0.43921692453425953,
      "grad_norm": 1.0505741834640503,
      "learning_rate": 4.390782828282828e-05,
      "loss": 0.9675,
      "step": 1391
    },
    {
      "epoch": 0.4395326807704452,
      "grad_norm": 1.2109038829803467,
      "learning_rate": 4.3939393939393944e-05,
      "loss": 0.9257,
      "step": 1392
    },
    {
      "epoch": 0.4398484370066309,
      "grad_norm": 0.7733975052833557,
      "learning_rate": 4.39709595959596e-05,
      "loss": 0.977,
      "step": 1393
    },
    {
      "epoch": 0.4401641932428165,
      "grad_norm": 0.858501136302948,
      "learning_rate": 4.4002525252525254e-05,
      "loss": 0.9707,
      "step": 1394
    },
    {
      "epoch": 0.4404799494790022,
      "grad_norm": 1.086286187171936,
      "learning_rate": 4.4034090909090916e-05,
      "loss": 0.9611,
      "step": 1395
    },
    {
      "epoch": 0.4407957057151879,
      "grad_norm": 0.7930727005004883,
      "learning_rate": 4.4065656565656565e-05,
      "loss": 0.9644,
      "step": 1396
    },
    {
      "epoch": 0.4411114619513735,
      "grad_norm": 0.7122814059257507,
      "learning_rate": 4.4097222222222226e-05,
      "loss": 0.9913,
      "step": 1397
    },
    {
      "epoch": 0.4414272181875592,
      "grad_norm": 0.8759936690330505,
      "learning_rate": 4.412878787878788e-05,
      "loss": 0.9641,
      "step": 1398
    },
    {
      "epoch": 0.4417429744237449,
      "grad_norm": 0.8059980273246765,
      "learning_rate": 4.416035353535354e-05,
      "loss": 0.977,
      "step": 1399
    },
    {
      "epoch": 0.4420587306599305,
      "grad_norm": 1.0454977750778198,
      "learning_rate": 4.41919191919192e-05,
      "loss": 0.9671,
      "step": 1400
    },
    {
      "epoch": 0.4420587306599305,
      "eval_loss": 0.9419717788696289,
      "eval_runtime": 4.694,
      "eval_samples_per_second": 530.463,
      "eval_steps_per_second": 4.474,
      "step": 1400
    },
    {
      "epoch": 0.4423744868961162,
      "grad_norm": 0.8167344927787781,
      "learning_rate": 4.4223484848484854e-05,
      "loss": 1.0082,
      "step": 1401
    },
    {
      "epoch": 0.4426902431323019,
      "grad_norm": 0.7355687022209167,
      "learning_rate": 4.42550505050505e-05,
      "loss": 0.912,
      "step": 1402
    },
    {
      "epoch": 0.4430059993684875,
      "grad_norm": 0.779727041721344,
      "learning_rate": 4.4286616161616164e-05,
      "loss": 0.9762,
      "step": 1403
    },
    {
      "epoch": 0.4433217556046732,
      "grad_norm": 1.0250340700149536,
      "learning_rate": 4.431818181818182e-05,
      "loss": 0.9991,
      "step": 1404
    },
    {
      "epoch": 0.44363751184085887,
      "grad_norm": 0.8667997717857361,
      "learning_rate": 4.4349747474747474e-05,
      "loss": 0.9661,
      "step": 1405
    },
    {
      "epoch": 0.4439532680770445,
      "grad_norm": 0.9998833537101746,
      "learning_rate": 4.4381313131313136e-05,
      "loss": 0.9813,
      "step": 1406
    },
    {
      "epoch": 0.4442690243132302,
      "grad_norm": 0.8241989612579346,
      "learning_rate": 4.441287878787879e-05,
      "loss": 0.9787,
      "step": 1407
    },
    {
      "epoch": 0.44458478054941586,
      "grad_norm": 0.763158917427063,
      "learning_rate": 4.4444444444444447e-05,
      "loss": 0.9646,
      "step": 1408
    },
    {
      "epoch": 0.44490053678560154,
      "grad_norm": 0.8473405838012695,
      "learning_rate": 4.44760101010101e-05,
      "loss": 1.013,
      "step": 1409
    },
    {
      "epoch": 0.44521629302178717,
      "grad_norm": 0.9040572643280029,
      "learning_rate": 4.450757575757576e-05,
      "loss": 0.9219,
      "step": 1410
    },
    {
      "epoch": 0.44553204925797285,
      "grad_norm": 0.7456806898117065,
      "learning_rate": 4.453914141414142e-05,
      "loss": 0.9287,
      "step": 1411
    },
    {
      "epoch": 0.44584780549415853,
      "grad_norm": 0.8163866400718689,
      "learning_rate": 4.4570707070707074e-05,
      "loss": 0.9711,
      "step": 1412
    },
    {
      "epoch": 0.44616356173034416,
      "grad_norm": 0.8087484240531921,
      "learning_rate": 4.460227272727273e-05,
      "loss": 0.9924,
      "step": 1413
    },
    {
      "epoch": 0.44647931796652984,
      "grad_norm": 0.8198484778404236,
      "learning_rate": 4.4633838383838384e-05,
      "loss": 0.9708,
      "step": 1414
    },
    {
      "epoch": 0.4467950742027155,
      "grad_norm": 1.214898943901062,
      "learning_rate": 4.466540404040404e-05,
      "loss": 0.9559,
      "step": 1415
    },
    {
      "epoch": 0.44711083043890115,
      "grad_norm": 0.7076473832130432,
      "learning_rate": 4.46969696969697e-05,
      "loss": 0.9976,
      "step": 1416
    },
    {
      "epoch": 0.44742658667508683,
      "grad_norm": 0.837444007396698,
      "learning_rate": 4.4728535353535356e-05,
      "loss": 0.9481,
      "step": 1417
    },
    {
      "epoch": 0.4477423429112725,
      "grad_norm": 0.8801687955856323,
      "learning_rate": 4.476010101010101e-05,
      "loss": 0.9746,
      "step": 1418
    },
    {
      "epoch": 0.44805809914745814,
      "grad_norm": 0.8478079438209534,
      "learning_rate": 4.4791666666666673e-05,
      "loss": 0.9699,
      "step": 1419
    },
    {
      "epoch": 0.4483738553836438,
      "grad_norm": 0.7812259793281555,
      "learning_rate": 4.482323232323233e-05,
      "loss": 0.9601,
      "step": 1420
    },
    {
      "epoch": 0.4486896116198295,
      "grad_norm": 0.8062241077423096,
      "learning_rate": 4.485479797979798e-05,
      "loss": 1.0234,
      "step": 1421
    },
    {
      "epoch": 0.44900536785601514,
      "grad_norm": 0.7889344692230225,
      "learning_rate": 4.488636363636364e-05,
      "loss": 1.0012,
      "step": 1422
    },
    {
      "epoch": 0.4493211240922008,
      "grad_norm": 0.8057528138160706,
      "learning_rate": 4.4917929292929294e-05,
      "loss": 0.9927,
      "step": 1423
    },
    {
      "epoch": 0.4496368803283865,
      "grad_norm": 0.9075132608413696,
      "learning_rate": 4.494949494949495e-05,
      "loss": 0.9724,
      "step": 1424
    },
    {
      "epoch": 0.44995263656457213,
      "grad_norm": 0.7454310059547424,
      "learning_rate": 4.498106060606061e-05,
      "loss": 0.969,
      "step": 1425
    },
    {
      "epoch": 0.4502683928007578,
      "grad_norm": 0.7542299628257751,
      "learning_rate": 4.5012626262626266e-05,
      "loss": 0.9367,
      "step": 1426
    },
    {
      "epoch": 0.4505841490369435,
      "grad_norm": 0.7894321084022522,
      "learning_rate": 4.504419191919192e-05,
      "loss": 1.009,
      "step": 1427
    },
    {
      "epoch": 0.4508999052731291,
      "grad_norm": 0.6177043318748474,
      "learning_rate": 4.5075757575757577e-05,
      "loss": 1.0023,
      "step": 1428
    },
    {
      "epoch": 0.4512156615093148,
      "grad_norm": 0.7028400301933289,
      "learning_rate": 4.510732323232323e-05,
      "loss": 0.9506,
      "step": 1429
    },
    {
      "epoch": 0.4515314177455005,
      "grad_norm": 0.7793817520141602,
      "learning_rate": 4.5138888888888894e-05,
      "loss": 0.9442,
      "step": 1430
    },
    {
      "epoch": 0.4518471739816861,
      "grad_norm": 0.8583313822746277,
      "learning_rate": 4.517045454545455e-05,
      "loss": 0.9558,
      "step": 1431
    },
    {
      "epoch": 0.4521629302178718,
      "grad_norm": 0.7662052512168884,
      "learning_rate": 4.5202020202020204e-05,
      "loss": 0.9801,
      "step": 1432
    },
    {
      "epoch": 0.4524786864540575,
      "grad_norm": 0.7984893918037415,
      "learning_rate": 4.523358585858586e-05,
      "loss": 0.994,
      "step": 1433
    },
    {
      "epoch": 0.45279444269024316,
      "grad_norm": 0.7363730669021606,
      "learning_rate": 4.5265151515151514e-05,
      "loss": 0.9267,
      "step": 1434
    },
    {
      "epoch": 0.4531101989264288,
      "grad_norm": 0.9659639000892639,
      "learning_rate": 4.5296717171717176e-05,
      "loss": 0.9696,
      "step": 1435
    },
    {
      "epoch": 0.45342595516261447,
      "grad_norm": 0.7798213958740234,
      "learning_rate": 4.532828282828283e-05,
      "loss": 1.015,
      "step": 1436
    },
    {
      "epoch": 0.45374171139880015,
      "grad_norm": 0.7812423706054688,
      "learning_rate": 4.5359848484848486e-05,
      "loss": 0.9958,
      "step": 1437
    },
    {
      "epoch": 0.4540574676349858,
      "grad_norm": 0.727967381477356,
      "learning_rate": 4.539141414141415e-05,
      "loss": 0.9508,
      "step": 1438
    },
    {
      "epoch": 0.45437322387117146,
      "grad_norm": 0.7758344411849976,
      "learning_rate": 4.54229797979798e-05,
      "loss": 0.9169,
      "step": 1439
    },
    {
      "epoch": 0.45468898010735714,
      "grad_norm": 0.9576510190963745,
      "learning_rate": 4.545454545454546e-05,
      "loss": 0.9436,
      "step": 1440
    },
    {
      "epoch": 0.45500473634354277,
      "grad_norm": 0.7575417160987854,
      "learning_rate": 4.5486111111111114e-05,
      "loss": 0.9277,
      "step": 1441
    },
    {
      "epoch": 0.45532049257972845,
      "grad_norm": 0.6883498430252075,
      "learning_rate": 4.551767676767677e-05,
      "loss": 0.9813,
      "step": 1442
    },
    {
      "epoch": 0.45563624881591414,
      "grad_norm": 0.7515760064125061,
      "learning_rate": 4.554924242424243e-05,
      "loss": 0.9853,
      "step": 1443
    },
    {
      "epoch": 0.45595200505209976,
      "grad_norm": 1.2316755056381226,
      "learning_rate": 4.5580808080808086e-05,
      "loss": 0.9344,
      "step": 1444
    },
    {
      "epoch": 0.45626776128828544,
      "grad_norm": 0.9252880811691284,
      "learning_rate": 4.5612373737373734e-05,
      "loss": 0.9487,
      "step": 1445
    },
    {
      "epoch": 0.4565835175244711,
      "grad_norm": 0.8156198263168335,
      "learning_rate": 4.5643939393939396e-05,
      "loss": 0.9215,
      "step": 1446
    },
    {
      "epoch": 0.45689927376065675,
      "grad_norm": 0.7672386169433594,
      "learning_rate": 4.567550505050505e-05,
      "loss": 0.9377,
      "step": 1447
    },
    {
      "epoch": 0.45721502999684244,
      "grad_norm": 1.0058461427688599,
      "learning_rate": 4.5707070707070706e-05,
      "loss": 0.9224,
      "step": 1448
    },
    {
      "epoch": 0.4575307862330281,
      "grad_norm": 0.7379342317581177,
      "learning_rate": 4.573863636363637e-05,
      "loss": 1.0489,
      "step": 1449
    },
    {
      "epoch": 0.45784654246921375,
      "grad_norm": 1.0515501499176025,
      "learning_rate": 4.5770202020202024e-05,
      "loss": 0.961,
      "step": 1450
    },
    {
      "epoch": 0.45816229870539943,
      "grad_norm": 0.9027102589607239,
      "learning_rate": 4.580176767676768e-05,
      "loss": 0.9636,
      "step": 1451
    },
    {
      "epoch": 0.4584780549415851,
      "grad_norm": 0.7725859880447388,
      "learning_rate": 4.5833333333333334e-05,
      "loss": 0.9289,
      "step": 1452
    },
    {
      "epoch": 0.45879381117777074,
      "grad_norm": 0.8526875376701355,
      "learning_rate": 4.586489898989899e-05,
      "loss": 0.9654,
      "step": 1453
    },
    {
      "epoch": 0.4591095674139564,
      "grad_norm": 0.7911215424537659,
      "learning_rate": 4.589646464646465e-05,
      "loss": 0.997,
      "step": 1454
    },
    {
      "epoch": 0.4594253236501421,
      "grad_norm": 0.7539464831352234,
      "learning_rate": 4.5928030303030306e-05,
      "loss": 1.0106,
      "step": 1455
    },
    {
      "epoch": 0.45974107988632773,
      "grad_norm": 1.0225934982299805,
      "learning_rate": 4.595959595959596e-05,
      "loss": 0.9435,
      "step": 1456
    },
    {
      "epoch": 0.4600568361225134,
      "grad_norm": 0.7586740255355835,
      "learning_rate": 4.599116161616162e-05,
      "loss": 0.9608,
      "step": 1457
    },
    {
      "epoch": 0.4603725923586991,
      "grad_norm": 0.8932416439056396,
      "learning_rate": 4.602272727272727e-05,
      "loss": 1.0014,
      "step": 1458
    },
    {
      "epoch": 0.4606883485948847,
      "grad_norm": 0.7042852640151978,
      "learning_rate": 4.605429292929293e-05,
      "loss": 0.9924,
      "step": 1459
    },
    {
      "epoch": 0.4610041048310704,
      "grad_norm": 0.8392357230186462,
      "learning_rate": 4.608585858585859e-05,
      "loss": 0.9449,
      "step": 1460
    },
    {
      "epoch": 0.4613198610672561,
      "grad_norm": 1.1916638612747192,
      "learning_rate": 4.6117424242424244e-05,
      "loss": 0.9359,
      "step": 1461
    },
    {
      "epoch": 0.46163561730344177,
      "grad_norm": 0.8546047210693359,
      "learning_rate": 4.6148989898989906e-05,
      "loss": 0.9757,
      "step": 1462
    },
    {
      "epoch": 0.4619513735396274,
      "grad_norm": 0.9135461449623108,
      "learning_rate": 4.618055555555556e-05,
      "loss": 0.9791,
      "step": 1463
    },
    {
      "epoch": 0.4622671297758131,
      "grad_norm": 0.8229580521583557,
      "learning_rate": 4.621212121212121e-05,
      "loss": 0.9469,
      "step": 1464
    },
    {
      "epoch": 0.46258288601199876,
      "grad_norm": 0.6891895532608032,
      "learning_rate": 4.624368686868687e-05,
      "loss": 0.9649,
      "step": 1465
    },
    {
      "epoch": 0.4628986422481844,
      "grad_norm": 0.7803835272789001,
      "learning_rate": 4.6275252525252526e-05,
      "loss": 1.0083,
      "step": 1466
    },
    {
      "epoch": 0.46321439848437007,
      "grad_norm": 0.6858707070350647,
      "learning_rate": 4.630681818181818e-05,
      "loss": 0.9727,
      "step": 1467
    },
    {
      "epoch": 0.46353015472055575,
      "grad_norm": 0.8138331174850464,
      "learning_rate": 4.633838383838384e-05,
      "loss": 0.927,
      "step": 1468
    },
    {
      "epoch": 0.4638459109567414,
      "grad_norm": 0.8001524209976196,
      "learning_rate": 4.63699494949495e-05,
      "loss": 0.9453,
      "step": 1469
    },
    {
      "epoch": 0.46416166719292706,
      "grad_norm": 0.8890300393104553,
      "learning_rate": 4.6401515151515154e-05,
      "loss": 1.0012,
      "step": 1470
    },
    {
      "epoch": 0.46447742342911275,
      "grad_norm": 0.913113534450531,
      "learning_rate": 4.643308080808081e-05,
      "loss": 0.9531,
      "step": 1471
    },
    {
      "epoch": 0.4647931796652984,
      "grad_norm": 0.8897829055786133,
      "learning_rate": 4.6464646464646464e-05,
      "loss": 0.9616,
      "step": 1472
    },
    {
      "epoch": 0.46510893590148406,
      "grad_norm": 0.748453676700592,
      "learning_rate": 4.6496212121212126e-05,
      "loss": 0.9305,
      "step": 1473
    },
    {
      "epoch": 0.46542469213766974,
      "grad_norm": 0.7508452534675598,
      "learning_rate": 4.652777777777778e-05,
      "loss": 0.9566,
      "step": 1474
    },
    {
      "epoch": 0.46574044837385536,
      "grad_norm": 0.805878221988678,
      "learning_rate": 4.6559343434343436e-05,
      "loss": 0.9123,
      "step": 1475
    },
    {
      "epoch": 0.46605620461004105,
      "grad_norm": 1.2982922792434692,
      "learning_rate": 4.659090909090909e-05,
      "loss": 0.9743,
      "step": 1476
    },
    {
      "epoch": 0.46637196084622673,
      "grad_norm": 0.7912974953651428,
      "learning_rate": 4.6622474747474746e-05,
      "loss": 1.0127,
      "step": 1477
    },
    {
      "epoch": 0.46668771708241236,
      "grad_norm": 0.6289088726043701,
      "learning_rate": 4.665404040404041e-05,
      "loss": 0.9361,
      "step": 1478
    },
    {
      "epoch": 0.46700347331859804,
      "grad_norm": 0.728987455368042,
      "learning_rate": 4.668560606060606e-05,
      "loss": 1.0031,
      "step": 1479
    },
    {
      "epoch": 0.4673192295547837,
      "grad_norm": 0.6613839864730835,
      "learning_rate": 4.671717171717172e-05,
      "loss": 0.9893,
      "step": 1480
    },
    {
      "epoch": 0.46763498579096935,
      "grad_norm": 1.0147457122802734,
      "learning_rate": 4.674873737373738e-05,
      "loss": 0.9296,
      "step": 1481
    },
    {
      "epoch": 0.46795074202715503,
      "grad_norm": 0.9073496460914612,
      "learning_rate": 4.678030303030303e-05,
      "loss": 0.977,
      "step": 1482
    },
    {
      "epoch": 0.4682664982633407,
      "grad_norm": 0.8351174592971802,
      "learning_rate": 4.681186868686869e-05,
      "loss": 0.9881,
      "step": 1483
    },
    {
      "epoch": 0.46858225449952634,
      "grad_norm": 0.7132789492607117,
      "learning_rate": 4.6843434343434346e-05,
      "loss": 0.9423,
      "step": 1484
    },
    {
      "epoch": 0.468898010735712,
      "grad_norm": 0.854043185710907,
      "learning_rate": 4.6875e-05,
      "loss": 1.0044,
      "step": 1485
    },
    {
      "epoch": 0.4692137669718977,
      "grad_norm": 1.0409572124481201,
      "learning_rate": 4.690656565656566e-05,
      "loss": 0.9812,
      "step": 1486
    },
    {
      "epoch": 0.46952952320808333,
      "grad_norm": 0.78766268491745,
      "learning_rate": 4.693813131313132e-05,
      "loss": 0.9694,
      "step": 1487
    },
    {
      "epoch": 0.469845279444269,
      "grad_norm": 0.7930814623832703,
      "learning_rate": 4.696969696969697e-05,
      "loss": 0.9173,
      "step": 1488
    },
    {
      "epoch": 0.4701610356804547,
      "grad_norm": 0.747352123260498,
      "learning_rate": 4.700126262626263e-05,
      "loss": 0.9456,
      "step": 1489
    },
    {
      "epoch": 0.4704767919166404,
      "grad_norm": 0.7887730002403259,
      "learning_rate": 4.7032828282828283e-05,
      "loss": 0.9419,
      "step": 1490
    },
    {
      "epoch": 0.470792548152826,
      "grad_norm": 0.742414116859436,
      "learning_rate": 4.706439393939394e-05,
      "loss": 0.9762,
      "step": 1491
    },
    {
      "epoch": 0.4711083043890117,
      "grad_norm": 0.681986927986145,
      "learning_rate": 4.70959595959596e-05,
      "loss": 1.0116,
      "step": 1492
    },
    {
      "epoch": 0.47142406062519737,
      "grad_norm": 0.918107807636261,
      "learning_rate": 4.7127525252525256e-05,
      "loss": 0.9291,
      "step": 1493
    },
    {
      "epoch": 0.471739816861383,
      "grad_norm": 0.8317708373069763,
      "learning_rate": 4.715909090909091e-05,
      "loss": 0.9964,
      "step": 1494
    },
    {
      "epoch": 0.4720555730975687,
      "grad_norm": 1.0338538885116577,
      "learning_rate": 4.7190656565656566e-05,
      "loss": 1.0153,
      "step": 1495
    },
    {
      "epoch": 0.47237132933375436,
      "grad_norm": 0.8151910901069641,
      "learning_rate": 4.722222222222222e-05,
      "loss": 0.978,
      "step": 1496
    },
    {
      "epoch": 0.47268708556994,
      "grad_norm": 0.8474667072296143,
      "learning_rate": 4.725378787878788e-05,
      "loss": 0.9053,
      "step": 1497
    },
    {
      "epoch": 0.4730028418061257,
      "grad_norm": 1.0048943758010864,
      "learning_rate": 4.728535353535354e-05,
      "loss": 0.8985,
      "step": 1498
    },
    {
      "epoch": 0.47331859804231136,
      "grad_norm": 0.9467518925666809,
      "learning_rate": 4.731691919191919e-05,
      "loss": 0.9798,
      "step": 1499
    },
    {
      "epoch": 0.473634354278497,
      "grad_norm": 0.9225822687149048,
      "learning_rate": 4.7348484848484855e-05,
      "loss": 0.9576,
      "step": 1500
    },
    {
      "epoch": 0.47395011051468267,
      "grad_norm": 1.078463077545166,
      "learning_rate": 4.7380050505050504e-05,
      "loss": 0.9724,
      "step": 1501
    },
    {
      "epoch": 0.47426586675086835,
      "grad_norm": 0.9417716860771179,
      "learning_rate": 4.7411616161616166e-05,
      "loss": 0.9516,
      "step": 1502
    },
    {
      "epoch": 0.474581622987054,
      "grad_norm": 0.7968136072158813,
      "learning_rate": 4.744318181818182e-05,
      "loss": 0.9678,
      "step": 1503
    },
    {
      "epoch": 0.47489737922323966,
      "grad_norm": 0.8720269203186035,
      "learning_rate": 4.7474747474747476e-05,
      "loss": 0.9544,
      "step": 1504
    },
    {
      "epoch": 0.47521313545942534,
      "grad_norm": 1.2561118602752686,
      "learning_rate": 4.750631313131314e-05,
      "loss": 0.9857,
      "step": 1505
    },
    {
      "epoch": 0.47552889169561097,
      "grad_norm": 1.1723617315292358,
      "learning_rate": 4.753787878787879e-05,
      "loss": 0.9706,
      "step": 1506
    },
    {
      "epoch": 0.47584464793179665,
      "grad_norm": 0.8848962187767029,
      "learning_rate": 4.756944444444444e-05,
      "loss": 0.9145,
      "step": 1507
    },
    {
      "epoch": 0.47616040416798233,
      "grad_norm": 1.034209966659546,
      "learning_rate": 4.76010101010101e-05,
      "loss": 0.9537,
      "step": 1508
    },
    {
      "epoch": 0.47647616040416796,
      "grad_norm": 1.115201711654663,
      "learning_rate": 4.763257575757576e-05,
      "loss": 0.9719,
      "step": 1509
    },
    {
      "epoch": 0.47679191664035364,
      "grad_norm": 0.8572171926498413,
      "learning_rate": 4.7664141414141413e-05,
      "loss": 0.9923,
      "step": 1510
    },
    {
      "epoch": 0.4771076728765393,
      "grad_norm": 0.8555787801742554,
      "learning_rate": 4.7695707070707075e-05,
      "loss": 0.9557,
      "step": 1511
    },
    {
      "epoch": 0.47742342911272495,
      "grad_norm": 1.4118835926055908,
      "learning_rate": 4.772727272727273e-05,
      "loss": 0.9534,
      "step": 1512
    },
    {
      "epoch": 0.47773918534891063,
      "grad_norm": 0.9165666699409485,
      "learning_rate": 4.7758838383838386e-05,
      "loss": 0.9594,
      "step": 1513
    },
    {
      "epoch": 0.4780549415850963,
      "grad_norm": 0.8107462525367737,
      "learning_rate": 4.779040404040404e-05,
      "loss": 0.9672,
      "step": 1514
    },
    {
      "epoch": 0.47837069782128194,
      "grad_norm": 1.349863052368164,
      "learning_rate": 4.7821969696969696e-05,
      "loss": 0.9738,
      "step": 1515
    },
    {
      "epoch": 0.4786864540574676,
      "grad_norm": 0.7617893815040588,
      "learning_rate": 4.785353535353536e-05,
      "loss": 1.0008,
      "step": 1516
    },
    {
      "epoch": 0.4790022102936533,
      "grad_norm": 0.9133582711219788,
      "learning_rate": 4.788510101010101e-05,
      "loss": 0.9602,
      "step": 1517
    },
    {
      "epoch": 0.479317966529839,
      "grad_norm": 0.7721654772758484,
      "learning_rate": 4.791666666666667e-05,
      "loss": 0.968,
      "step": 1518
    },
    {
      "epoch": 0.4796337227660246,
      "grad_norm": 0.7513347268104553,
      "learning_rate": 4.794823232323233e-05,
      "loss": 0.9577,
      "step": 1519
    },
    {
      "epoch": 0.4799494790022103,
      "grad_norm": 0.8650693297386169,
      "learning_rate": 4.797979797979798e-05,
      "loss": 0.9621,
      "step": 1520
    },
    {
      "epoch": 0.480265235238396,
      "grad_norm": 0.7002776265144348,
      "learning_rate": 4.801136363636364e-05,
      "loss": 0.9503,
      "step": 1521
    },
    {
      "epoch": 0.4805809914745816,
      "grad_norm": 0.9073238372802734,
      "learning_rate": 4.8042929292929295e-05,
      "loss": 0.9741,
      "step": 1522
    },
    {
      "epoch": 0.4808967477107673,
      "grad_norm": 0.9077586531639099,
      "learning_rate": 4.807449494949495e-05,
      "loss": 1.0117,
      "step": 1523
    },
    {
      "epoch": 0.481212503946953,
      "grad_norm": 0.7699394226074219,
      "learning_rate": 4.810606060606061e-05,
      "loss": 0.9301,
      "step": 1524
    },
    {
      "epoch": 0.4815282601831386,
      "grad_norm": 0.9159624576568604,
      "learning_rate": 4.813762626262627e-05,
      "loss": 0.9555,
      "step": 1525
    },
    {
      "epoch": 0.4818440164193243,
      "grad_norm": 0.9476621150970459,
      "learning_rate": 4.816919191919192e-05,
      "loss": 1.0064,
      "step": 1526
    },
    {
      "epoch": 0.48215977265550997,
      "grad_norm": 1.1246193647384644,
      "learning_rate": 4.820075757575758e-05,
      "loss": 0.9001,
      "step": 1527
    },
    {
      "epoch": 0.4824755288916956,
      "grad_norm": 0.9409036040306091,
      "learning_rate": 4.823232323232323e-05,
      "loss": 0.9586,
      "step": 1528
    },
    {
      "epoch": 0.4827912851278813,
      "grad_norm": 0.7821800112724304,
      "learning_rate": 4.8263888888888895e-05,
      "loss": 0.9576,
      "step": 1529
    },
    {
      "epoch": 0.48310704136406696,
      "grad_norm": 0.7174288034439087,
      "learning_rate": 4.829545454545455e-05,
      "loss": 0.9477,
      "step": 1530
    },
    {
      "epoch": 0.4834227976002526,
      "grad_norm": 1.0628429651260376,
      "learning_rate": 4.8327020202020205e-05,
      "loss": 0.9455,
      "step": 1531
    },
    {
      "epoch": 0.48373855383643827,
      "grad_norm": 1.0096452236175537,
      "learning_rate": 4.835858585858586e-05,
      "loss": 0.9767,
      "step": 1532
    },
    {
      "epoch": 0.48405431007262395,
      "grad_norm": 0.8128191828727722,
      "learning_rate": 4.8390151515151516e-05,
      "loss": 0.983,
      "step": 1533
    },
    {
      "epoch": 0.4843700663088096,
      "grad_norm": 0.7873527407646179,
      "learning_rate": 4.842171717171717e-05,
      "loss": 0.9845,
      "step": 1534
    },
    {
      "epoch": 0.48468582254499526,
      "grad_norm": 0.7866812348365784,
      "learning_rate": 4.845328282828283e-05,
      "loss": 0.9849,
      "step": 1535
    },
    {
      "epoch": 0.48500157878118094,
      "grad_norm": 0.720967173576355,
      "learning_rate": 4.848484848484849e-05,
      "loss": 0.965,
      "step": 1536
    },
    {
      "epoch": 0.48531733501736657,
      "grad_norm": 1.3684407472610474,
      "learning_rate": 4.851641414141414e-05,
      "loss": 0.9546,
      "step": 1537
    },
    {
      "epoch": 0.48563309125355225,
      "grad_norm": 0.7669100761413574,
      "learning_rate": 4.85479797979798e-05,
      "loss": 0.9606,
      "step": 1538
    },
    {
      "epoch": 0.48594884748973793,
      "grad_norm": 1.0507805347442627,
      "learning_rate": 4.857954545454545e-05,
      "loss": 0.9544,
      "step": 1539
    },
    {
      "epoch": 0.48626460372592356,
      "grad_norm": 1.0549198389053345,
      "learning_rate": 4.8611111111111115e-05,
      "loss": 0.9353,
      "step": 1540
    },
    {
      "epoch": 0.48658035996210924,
      "grad_norm": 0.8526098728179932,
      "learning_rate": 4.864267676767677e-05,
      "loss": 0.9926,
      "step": 1541
    },
    {
      "epoch": 0.4868961161982949,
      "grad_norm": 0.8626139760017395,
      "learning_rate": 4.8674242424242425e-05,
      "loss": 0.9878,
      "step": 1542
    },
    {
      "epoch": 0.4872118724344806,
      "grad_norm": 0.8979067802429199,
      "learning_rate": 4.870580808080809e-05,
      "loss": 0.9408,
      "step": 1543
    },
    {
      "epoch": 0.48752762867066624,
      "grad_norm": 0.7887360453605652,
      "learning_rate": 4.8737373737373736e-05,
      "loss": 0.913,
      "step": 1544
    },
    {
      "epoch": 0.4878433849068519,
      "grad_norm": 0.878444492816925,
      "learning_rate": 4.87689393939394e-05,
      "loss": 0.9243,
      "step": 1545
    },
    {
      "epoch": 0.4881591411430376,
      "grad_norm": 0.7227399349212646,
      "learning_rate": 4.880050505050505e-05,
      "loss": 0.996,
      "step": 1546
    },
    {
      "epoch": 0.4884748973792232,
      "grad_norm": 0.88020920753479,
      "learning_rate": 4.883207070707071e-05,
      "loss": 0.9625,
      "step": 1547
    },
    {
      "epoch": 0.4887906536154089,
      "grad_norm": 0.7674077749252319,
      "learning_rate": 4.886363636363637e-05,
      "loss": 0.9508,
      "step": 1548
    },
    {
      "epoch": 0.4891064098515946,
      "grad_norm": 0.7766163349151611,
      "learning_rate": 4.8895202020202025e-05,
      "loss": 0.961,
      "step": 1549
    },
    {
      "epoch": 0.4894221660877802,
      "grad_norm": 1.0651202201843262,
      "learning_rate": 4.892676767676767e-05,
      "loss": 0.9359,
      "step": 1550
    },
    {
      "epoch": 0.4897379223239659,
      "grad_norm": 0.8140547871589661,
      "learning_rate": 4.8958333333333335e-05,
      "loss": 0.9347,
      "step": 1551
    },
    {
      "epoch": 0.4900536785601516,
      "grad_norm": 0.7126879096031189,
      "learning_rate": 4.898989898989899e-05,
      "loss": 0.9466,
      "step": 1552
    },
    {
      "epoch": 0.4903694347963372,
      "grad_norm": 0.802253246307373,
      "learning_rate": 4.9021464646464646e-05,
      "loss": 0.9513,
      "step": 1553
    },
    {
      "epoch": 0.4906851910325229,
      "grad_norm": 0.9029814004898071,
      "learning_rate": 4.905303030303031e-05,
      "loss": 0.9482,
      "step": 1554
    },
    {
      "epoch": 0.4910009472687086,
      "grad_norm": 0.8478924632072449,
      "learning_rate": 4.908459595959596e-05,
      "loss": 0.948,
      "step": 1555
    },
    {
      "epoch": 0.4913167035048942,
      "grad_norm": 0.762295126914978,
      "learning_rate": 4.9116161616161625e-05,
      "loss": 0.9367,
      "step": 1556
    },
    {
      "epoch": 0.4916324597410799,
      "grad_norm": 0.9283182621002197,
      "learning_rate": 4.914772727272727e-05,
      "loss": 0.9519,
      "step": 1557
    },
    {
      "epoch": 0.49194821597726557,
      "grad_norm": 0.844781219959259,
      "learning_rate": 4.917929292929293e-05,
      "loss": 0.9361,
      "step": 1558
    },
    {
      "epoch": 0.4922639722134512,
      "grad_norm": 0.7224656939506531,
      "learning_rate": 4.921085858585859e-05,
      "loss": 0.9575,
      "step": 1559
    },
    {
      "epoch": 0.4925797284496369,
      "grad_norm": 0.8484576940536499,
      "learning_rate": 4.9242424242424245e-05,
      "loss": 0.9443,
      "step": 1560
    },
    {
      "epoch": 0.49289548468582256,
      "grad_norm": 0.8423038125038147,
      "learning_rate": 4.92739898989899e-05,
      "loss": 1.008,
      "step": 1561
    },
    {
      "epoch": 0.4932112409220082,
      "grad_norm": 0.9430524110794067,
      "learning_rate": 4.930555555555556e-05,
      "loss": 0.9639,
      "step": 1562
    },
    {
      "epoch": 0.49352699715819387,
      "grad_norm": 1.1608637571334839,
      "learning_rate": 4.933712121212121e-05,
      "loss": 0.961,
      "step": 1563
    },
    {
      "epoch": 0.49384275339437955,
      "grad_norm": 0.869839072227478,
      "learning_rate": 4.936868686868687e-05,
      "loss": 0.96,
      "step": 1564
    },
    {
      "epoch": 0.4941585096305652,
      "grad_norm": 0.8321672081947327,
      "learning_rate": 4.940025252525253e-05,
      "loss": 0.9995,
      "step": 1565
    },
    {
      "epoch": 0.49447426586675086,
      "grad_norm": 0.7560545802116394,
      "learning_rate": 4.943181818181818e-05,
      "loss": 0.9563,
      "step": 1566
    },
    {
      "epoch": 0.49479002210293654,
      "grad_norm": 0.7859839797019958,
      "learning_rate": 4.9463383838383845e-05,
      "loss": 0.9754,
      "step": 1567
    },
    {
      "epoch": 0.49510577833912217,
      "grad_norm": 0.8922802209854126,
      "learning_rate": 4.94949494949495e-05,
      "loss": 0.9769,
      "step": 1568
    },
    {
      "epoch": 0.49542153457530785,
      "grad_norm": 1.0491235256195068,
      "learning_rate": 4.952651515151515e-05,
      "loss": 0.947,
      "step": 1569
    },
    {
      "epoch": 0.49573729081149354,
      "grad_norm": 0.6319274306297302,
      "learning_rate": 4.955808080808081e-05,
      "loss": 0.9699,
      "step": 1570
    },
    {
      "epoch": 0.4960530470476792,
      "grad_norm": 0.8112782835960388,
      "learning_rate": 4.9589646464646465e-05,
      "loss": 0.9614,
      "step": 1571
    },
    {
      "epoch": 0.49636880328386485,
      "grad_norm": 0.6907515525817871,
      "learning_rate": 4.962121212121213e-05,
      "loss": 0.9652,
      "step": 1572
    },
    {
      "epoch": 0.49668455952005053,
      "grad_norm": 0.9472324252128601,
      "learning_rate": 4.965277777777778e-05,
      "loss": 0.9813,
      "step": 1573
    },
    {
      "epoch": 0.4970003157562362,
      "grad_norm": 0.7483730316162109,
      "learning_rate": 4.968434343434344e-05,
      "loss": 0.9454,
      "step": 1574
    },
    {
      "epoch": 0.49731607199242184,
      "grad_norm": 0.9049888849258423,
      "learning_rate": 4.971590909090909e-05,
      "loss": 0.9644,
      "step": 1575
    },
    {
      "epoch": 0.4976318282286075,
      "grad_norm": 0.7648831009864807,
      "learning_rate": 4.974747474747475e-05,
      "loss": 0.9462,
      "step": 1576
    },
    {
      "epoch": 0.4979475844647932,
      "grad_norm": 0.7179455161094666,
      "learning_rate": 4.97790404040404e-05,
      "loss": 0.9417,
      "step": 1577
    },
    {
      "epoch": 0.49826334070097883,
      "grad_norm": 0.9199532270431519,
      "learning_rate": 4.9810606060606065e-05,
      "loss": 0.9983,
      "step": 1578
    },
    {
      "epoch": 0.4985790969371645,
      "grad_norm": 0.9121428728103638,
      "learning_rate": 4.984217171717172e-05,
      "loss": 1.0008,
      "step": 1579
    },
    {
      "epoch": 0.4988948531733502,
      "grad_norm": 0.7058104276657104,
      "learning_rate": 4.9873737373737375e-05,
      "loss": 0.9407,
      "step": 1580
    },
    {
      "epoch": 0.4992106094095358,
      "grad_norm": 0.8348144292831421,
      "learning_rate": 4.990530303030303e-05,
      "loss": 0.9511,
      "step": 1581
    },
    {
      "epoch": 0.4995263656457215,
      "grad_norm": 0.7561733722686768,
      "learning_rate": 4.9936868686868685e-05,
      "loss": 0.9194,
      "step": 1582
    },
    {
      "epoch": 0.4998421218819072,
      "grad_norm": 0.8205838203430176,
      "learning_rate": 4.996843434343435e-05,
      "loss": 0.9585,
      "step": 1583
    },
    {
      "epoch": 0.5001578781180929,
      "grad_norm": 0.7711383104324341,
      "learning_rate": 5e-05,
      "loss": 0.9145,
      "step": 1584
    },
    {
      "epoch": 0.5004736343542785,
      "grad_norm": 0.8265209197998047,
      "learning_rate": 4.999649147428251e-05,
      "loss": 0.9912,
      "step": 1585
    },
    {
      "epoch": 0.5007893905904641,
      "grad_norm": 0.8171094059944153,
      "learning_rate": 4.9992982948565016e-05,
      "loss": 0.9853,
      "step": 1586
    },
    {
      "epoch": 0.5011051468266499,
      "grad_norm": 0.7894766926765442,
      "learning_rate": 4.998947442284752e-05,
      "loss": 0.983,
      "step": 1587
    },
    {
      "epoch": 0.5014209030628355,
      "grad_norm": 0.807807981967926,
      "learning_rate": 4.998596589713003e-05,
      "loss": 0.9445,
      "step": 1588
    },
    {
      "epoch": 0.5017366592990211,
      "grad_norm": 0.6473283171653748,
      "learning_rate": 4.9982457371412536e-05,
      "loss": 0.9691,
      "step": 1589
    },
    {
      "epoch": 0.5020524155352069,
      "grad_norm": 0.8246219158172607,
      "learning_rate": 4.9978948845695036e-05,
      "loss": 0.9557,
      "step": 1590
    },
    {
      "epoch": 0.5023681717713925,
      "grad_norm": 0.8126115798950195,
      "learning_rate": 4.997544031997754e-05,
      "loss": 0.9647,
      "step": 1591
    },
    {
      "epoch": 0.5026839280075781,
      "grad_norm": 1.0541341304779053,
      "learning_rate": 4.997193179426005e-05,
      "loss": 0.9328,
      "step": 1592
    },
    {
      "epoch": 0.5029996842437638,
      "grad_norm": 0.8086041212081909,
      "learning_rate": 4.9968423268542564e-05,
      "loss": 0.9462,
      "step": 1593
    },
    {
      "epoch": 0.5033154404799495,
      "grad_norm": 0.7372657060623169,
      "learning_rate": 4.996491474282507e-05,
      "loss": 0.9193,
      "step": 1594
    },
    {
      "epoch": 0.5036311967161351,
      "grad_norm": 0.9741494655609131,
      "learning_rate": 4.996140621710758e-05,
      "loss": 0.9446,
      "step": 1595
    },
    {
      "epoch": 0.5039469529523208,
      "grad_norm": 0.8159045577049255,
      "learning_rate": 4.9957897691390084e-05,
      "loss": 0.9523,
      "step": 1596
    },
    {
      "epoch": 0.5042627091885065,
      "grad_norm": 0.8295793533325195,
      "learning_rate": 4.995438916567259e-05,
      "loss": 0.9884,
      "step": 1597
    },
    {
      "epoch": 0.5045784654246921,
      "grad_norm": 0.7041229605674744,
      "learning_rate": 4.995088063995509e-05,
      "loss": 0.9298,
      "step": 1598
    },
    {
      "epoch": 0.5048942216608778,
      "grad_norm": 0.783968985080719,
      "learning_rate": 4.99473721142376e-05,
      "loss": 0.9359,
      "step": 1599
    },
    {
      "epoch": 0.5052099778970635,
      "grad_norm": 0.7818081378936768,
      "learning_rate": 4.9943863588520105e-05,
      "loss": 0.9635,
      "step": 1600
    },
    {
      "epoch": 0.5052099778970635,
      "eval_loss": 0.9238277077674866,
      "eval_runtime": 4.5935,
      "eval_samples_per_second": 542.075,
      "eval_steps_per_second": 4.572,
      "step": 1600
    },
    {
      "epoch": 0.5055257341332491,
      "grad_norm": 0.8355650305747986,
      "learning_rate": 4.994035506280261e-05,
      "loss": 0.935,
      "step": 1601
    },
    {
      "epoch": 0.5058414903694348,
      "grad_norm": 0.8562469482421875,
      "learning_rate": 4.993684653708512e-05,
      "loss": 0.9285,
      "step": 1602
    },
    {
      "epoch": 0.5061572466056204,
      "grad_norm": 0.75931715965271,
      "learning_rate": 4.9933338011367625e-05,
      "loss": 0.9164,
      "step": 1603
    },
    {
      "epoch": 0.5064730028418061,
      "grad_norm": 0.8741904497146606,
      "learning_rate": 4.992982948565013e-05,
      "loss": 0.999,
      "step": 1604
    },
    {
      "epoch": 0.5067887590779918,
      "grad_norm": 0.9156028032302856,
      "learning_rate": 4.992632095993264e-05,
      "loss": 0.9935,
      "step": 1605
    },
    {
      "epoch": 0.5071045153141774,
      "grad_norm": 0.9940505027770996,
      "learning_rate": 4.9922812434215146e-05,
      "loss": 0.9209,
      "step": 1606
    },
    {
      "epoch": 0.5074202715503631,
      "grad_norm": 1.0590791702270508,
      "learning_rate": 4.991930390849765e-05,
      "loss": 0.979,
      "step": 1607
    },
    {
      "epoch": 0.5077360277865488,
      "grad_norm": 1.019998550415039,
      "learning_rate": 4.991579538278016e-05,
      "loss": 0.9438,
      "step": 1608
    },
    {
      "epoch": 0.5080517840227344,
      "grad_norm": 0.9930368065834045,
      "learning_rate": 4.991228685706266e-05,
      "loss": 0.9523,
      "step": 1609
    },
    {
      "epoch": 0.5083675402589201,
      "grad_norm": 0.805239737033844,
      "learning_rate": 4.9908778331345166e-05,
      "loss": 0.9449,
      "step": 1610
    },
    {
      "epoch": 0.5086832964951058,
      "grad_norm": 0.9842427968978882,
      "learning_rate": 4.990526980562768e-05,
      "loss": 0.96,
      "step": 1611
    },
    {
      "epoch": 0.5089990527312914,
      "grad_norm": 0.7240799069404602,
      "learning_rate": 4.9901761279910186e-05,
      "loss": 0.9273,
      "step": 1612
    },
    {
      "epoch": 0.5093148089674772,
      "grad_norm": 1.0245256423950195,
      "learning_rate": 4.989825275419269e-05,
      "loss": 0.9585,
      "step": 1613
    },
    {
      "epoch": 0.5096305652036628,
      "grad_norm": 0.8658608198165894,
      "learning_rate": 4.98947442284752e-05,
      "loss": 0.9422,
      "step": 1614
    },
    {
      "epoch": 0.5099463214398484,
      "grad_norm": 0.8498935103416443,
      "learning_rate": 4.989123570275771e-05,
      "loss": 0.9671,
      "step": 1615
    },
    {
      "epoch": 0.5102620776760342,
      "grad_norm": 0.8899562358856201,
      "learning_rate": 4.9887727177040214e-05,
      "loss": 0.9448,
      "step": 1616
    },
    {
      "epoch": 0.5105778339122198,
      "grad_norm": 0.6950564980506897,
      "learning_rate": 4.988421865132272e-05,
      "loss": 0.9084,
      "step": 1617
    },
    {
      "epoch": 0.5108935901484054,
      "grad_norm": 0.6946361064910889,
      "learning_rate": 4.988071012560522e-05,
      "loss": 0.9549,
      "step": 1618
    },
    {
      "epoch": 0.5112093463845911,
      "grad_norm": 0.7454885244369507,
      "learning_rate": 4.987720159988773e-05,
      "loss": 0.9572,
      "step": 1619
    },
    {
      "epoch": 0.5115251026207768,
      "grad_norm": 1.2271912097930908,
      "learning_rate": 4.9873693074170234e-05,
      "loss": 0.9799,
      "step": 1620
    },
    {
      "epoch": 0.5118408588569624,
      "grad_norm": 1.1281588077545166,
      "learning_rate": 4.987018454845274e-05,
      "loss": 1.0013,
      "step": 1621
    },
    {
      "epoch": 0.5121566150931481,
      "grad_norm": 0.8199347257614136,
      "learning_rate": 4.986667602273525e-05,
      "loss": 0.9971,
      "step": 1622
    },
    {
      "epoch": 0.5124723713293338,
      "grad_norm": 0.7636246681213379,
      "learning_rate": 4.9863167497017755e-05,
      "loss": 0.9287,
      "step": 1623
    },
    {
      "epoch": 0.5127881275655194,
      "grad_norm": 0.8881829977035522,
      "learning_rate": 4.985965897130026e-05,
      "loss": 0.9746,
      "step": 1624
    },
    {
      "epoch": 0.5131038838017051,
      "grad_norm": 0.8793246746063232,
      "learning_rate": 4.985615044558277e-05,
      "loss": 0.9623,
      "step": 1625
    },
    {
      "epoch": 0.5134196400378908,
      "grad_norm": 0.9123930335044861,
      "learning_rate": 4.9852641919865275e-05,
      "loss": 0.9918,
      "step": 1626
    },
    {
      "epoch": 0.5137353962740764,
      "grad_norm": 0.7876163125038147,
      "learning_rate": 4.984913339414778e-05,
      "loss": 0.9471,
      "step": 1627
    },
    {
      "epoch": 0.5140511525102621,
      "grad_norm": 0.8838668465614319,
      "learning_rate": 4.984562486843029e-05,
      "loss": 0.9376,
      "step": 1628
    },
    {
      "epoch": 0.5143669087464477,
      "grad_norm": 1.273787260055542,
      "learning_rate": 4.9842116342712795e-05,
      "loss": 0.9555,
      "step": 1629
    },
    {
      "epoch": 0.5146826649826334,
      "grad_norm": 0.8292577862739563,
      "learning_rate": 4.98386078169953e-05,
      "loss": 0.9731,
      "step": 1630
    },
    {
      "epoch": 0.5149984212188191,
      "grad_norm": 0.7570028901100159,
      "learning_rate": 4.983509929127781e-05,
      "loss": 0.9432,
      "step": 1631
    },
    {
      "epoch": 0.5153141774550047,
      "grad_norm": 1.0504679679870605,
      "learning_rate": 4.9831590765560316e-05,
      "loss": 0.9219,
      "step": 1632
    },
    {
      "epoch": 0.5156299336911904,
      "grad_norm": 0.8771341443061829,
      "learning_rate": 4.982808223984282e-05,
      "loss": 0.9251,
      "step": 1633
    },
    {
      "epoch": 0.5159456899273761,
      "grad_norm": 0.9003463387489319,
      "learning_rate": 4.982457371412533e-05,
      "loss": 0.9309,
      "step": 1634
    },
    {
      "epoch": 0.5162614461635617,
      "grad_norm": 0.8879188299179077,
      "learning_rate": 4.9821065188407836e-05,
      "loss": 0.9478,
      "step": 1635
    },
    {
      "epoch": 0.5165772023997474,
      "grad_norm": 1.242435336112976,
      "learning_rate": 4.981755666269034e-05,
      "loss": 0.9206,
      "step": 1636
    },
    {
      "epoch": 0.5168929586359331,
      "grad_norm": 0.9127935171127319,
      "learning_rate": 4.981404813697284e-05,
      "loss": 0.9712,
      "step": 1637
    },
    {
      "epoch": 0.5172087148721187,
      "grad_norm": 0.9411696195602417,
      "learning_rate": 4.981053961125535e-05,
      "loss": 0.9423,
      "step": 1638
    },
    {
      "epoch": 0.5175244711083044,
      "grad_norm": 0.9138306379318237,
      "learning_rate": 4.980703108553786e-05,
      "loss": 0.9314,
      "step": 1639
    },
    {
      "epoch": 0.5178402273444901,
      "grad_norm": 0.9273859262466431,
      "learning_rate": 4.9803522559820364e-05,
      "loss": 0.9317,
      "step": 1640
    },
    {
      "epoch": 0.5181559835806757,
      "grad_norm": 0.8000861406326294,
      "learning_rate": 4.980001403410287e-05,
      "loss": 0.9357,
      "step": 1641
    },
    {
      "epoch": 0.5184717398168613,
      "grad_norm": 0.7868821620941162,
      "learning_rate": 4.979650550838538e-05,
      "loss": 0.9111,
      "step": 1642
    },
    {
      "epoch": 0.5187874960530471,
      "grad_norm": 1.0097359418869019,
      "learning_rate": 4.9792996982667884e-05,
      "loss": 0.9921,
      "step": 1643
    },
    {
      "epoch": 0.5191032522892327,
      "grad_norm": 0.9426999688148499,
      "learning_rate": 4.978948845695039e-05,
      "loss": 0.9659,
      "step": 1644
    },
    {
      "epoch": 0.5194190085254183,
      "grad_norm": 0.8455493450164795,
      "learning_rate": 4.97859799312329e-05,
      "loss": 0.9373,
      "step": 1645
    },
    {
      "epoch": 0.5197347647616041,
      "grad_norm": 0.9038839936256409,
      "learning_rate": 4.9782471405515404e-05,
      "loss": 0.943,
      "step": 1646
    },
    {
      "epoch": 0.5200505209977897,
      "grad_norm": 1.0518511533737183,
      "learning_rate": 4.977896287979791e-05,
      "loss": 0.9766,
      "step": 1647
    },
    {
      "epoch": 0.5203662772339753,
      "grad_norm": 0.934010922908783,
      "learning_rate": 4.977545435408042e-05,
      "loss": 0.9608,
      "step": 1648
    },
    {
      "epoch": 0.5206820334701611,
      "grad_norm": 1.0404542684555054,
      "learning_rate": 4.9771945828362925e-05,
      "loss": 0.9542,
      "step": 1649
    },
    {
      "epoch": 0.5209977897063467,
      "grad_norm": 0.8351597189903259,
      "learning_rate": 4.976843730264543e-05,
      "loss": 0.9602,
      "step": 1650
    },
    {
      "epoch": 0.5213135459425323,
      "grad_norm": 0.9114130735397339,
      "learning_rate": 4.976492877692794e-05,
      "loss": 0.9634,
      "step": 1651
    },
    {
      "epoch": 0.5216293021787181,
      "grad_norm": 0.9023140668869019,
      "learning_rate": 4.9761420251210445e-05,
      "loss": 0.9738,
      "step": 1652
    },
    {
      "epoch": 0.5219450584149037,
      "grad_norm": 1.0707882642745972,
      "learning_rate": 4.975791172549295e-05,
      "loss": 0.8656,
      "step": 1653
    },
    {
      "epoch": 0.5222608146510893,
      "grad_norm": 0.9789724349975586,
      "learning_rate": 4.975440319977546e-05,
      "loss": 0.9855,
      "step": 1654
    },
    {
      "epoch": 0.522576570887275,
      "grad_norm": 1.0362812280654907,
      "learning_rate": 4.9750894674057966e-05,
      "loss": 0.97,
      "step": 1655
    },
    {
      "epoch": 0.5228923271234607,
      "grad_norm": 1.1497483253479004,
      "learning_rate": 4.9747386148340466e-05,
      "loss": 0.9711,
      "step": 1656
    },
    {
      "epoch": 0.5232080833596463,
      "grad_norm": 0.8956688642501831,
      "learning_rate": 4.974387762262297e-05,
      "loss": 0.9517,
      "step": 1657
    },
    {
      "epoch": 0.523523839595832,
      "grad_norm": 0.803141176700592,
      "learning_rate": 4.974036909690548e-05,
      "loss": 0.9429,
      "step": 1658
    },
    {
      "epoch": 0.5238395958320177,
      "grad_norm": 0.9624006152153015,
      "learning_rate": 4.9736860571187986e-05,
      "loss": 0.9584,
      "step": 1659
    },
    {
      "epoch": 0.5241553520682033,
      "grad_norm": 0.9438497424125671,
      "learning_rate": 4.973335204547049e-05,
      "loss": 0.9594,
      "step": 1660
    },
    {
      "epoch": 0.524471108304389,
      "grad_norm": 0.8411515951156616,
      "learning_rate": 4.9729843519753e-05,
      "loss": 0.9833,
      "step": 1661
    },
    {
      "epoch": 0.5247868645405747,
      "grad_norm": 0.7821338176727295,
      "learning_rate": 4.972633499403551e-05,
      "loss": 0.9537,
      "step": 1662
    },
    {
      "epoch": 0.5251026207767603,
      "grad_norm": 0.8652716279029846,
      "learning_rate": 4.972282646831802e-05,
      "loss": 0.9608,
      "step": 1663
    },
    {
      "epoch": 0.525418377012946,
      "grad_norm": 0.948841392993927,
      "learning_rate": 4.971931794260053e-05,
      "loss": 0.9382,
      "step": 1664
    },
    {
      "epoch": 0.5257341332491317,
      "grad_norm": 0.9496994614601135,
      "learning_rate": 4.971580941688303e-05,
      "loss": 0.9914,
      "step": 1665
    },
    {
      "epoch": 0.5260498894853174,
      "grad_norm": 0.8056849241256714,
      "learning_rate": 4.9712300891165534e-05,
      "loss": 0.9353,
      "step": 1666
    },
    {
      "epoch": 0.526365645721503,
      "grad_norm": 0.8934057354927063,
      "learning_rate": 4.970879236544804e-05,
      "loss": 1.0075,
      "step": 1667
    },
    {
      "epoch": 0.5266814019576886,
      "grad_norm": 0.8143153190612793,
      "learning_rate": 4.970528383973055e-05,
      "loss": 0.9637,
      "step": 1668
    },
    {
      "epoch": 0.5269971581938744,
      "grad_norm": 0.9891625642776489,
      "learning_rate": 4.9701775314013054e-05,
      "loss": 0.9728,
      "step": 1669
    },
    {
      "epoch": 0.52731291443006,
      "grad_norm": 0.880293071269989,
      "learning_rate": 4.969826678829556e-05,
      "loss": 0.9382,
      "step": 1670
    },
    {
      "epoch": 0.5276286706662456,
      "grad_norm": 0.7538621425628662,
      "learning_rate": 4.969475826257807e-05,
      "loss": 0.9641,
      "step": 1671
    },
    {
      "epoch": 0.5279444269024314,
      "grad_norm": 0.668260931968689,
      "learning_rate": 4.9691249736860575e-05,
      "loss": 0.9489,
      "step": 1672
    },
    {
      "epoch": 0.528260183138617,
      "grad_norm": 1.0019866228103638,
      "learning_rate": 4.968774121114308e-05,
      "loss": 0.9535,
      "step": 1673
    },
    {
      "epoch": 0.5285759393748026,
      "grad_norm": 0.7422218918800354,
      "learning_rate": 4.968423268542559e-05,
      "loss": 0.9628,
      "step": 1674
    },
    {
      "epoch": 0.5288916956109884,
      "grad_norm": 0.9249774217605591,
      "learning_rate": 4.9680724159708095e-05,
      "loss": 0.9925,
      "step": 1675
    },
    {
      "epoch": 0.529207451847174,
      "grad_norm": 0.6554447412490845,
      "learning_rate": 4.9677215633990595e-05,
      "loss": 0.9816,
      "step": 1676
    },
    {
      "epoch": 0.5295232080833596,
      "grad_norm": 0.9694783687591553,
      "learning_rate": 4.96737071082731e-05,
      "loss": 0.9971,
      "step": 1677
    },
    {
      "epoch": 0.5298389643195454,
      "grad_norm": 0.9638500809669495,
      "learning_rate": 4.967019858255561e-05,
      "loss": 0.9956,
      "step": 1678
    },
    {
      "epoch": 0.530154720555731,
      "grad_norm": 1.2556275129318237,
      "learning_rate": 4.9666690056838116e-05,
      "loss": 0.9473,
      "step": 1679
    },
    {
      "epoch": 0.5304704767919166,
      "grad_norm": 0.9529656171798706,
      "learning_rate": 4.966318153112062e-05,
      "loss": 0.9287,
      "step": 1680
    },
    {
      "epoch": 0.5307862330281023,
      "grad_norm": 0.9732696413993835,
      "learning_rate": 4.9659673005403136e-05,
      "loss": 0.9456,
      "step": 1681
    },
    {
      "epoch": 0.531101989264288,
      "grad_norm": 0.9012371897697449,
      "learning_rate": 4.965616447968564e-05,
      "loss": 0.9572,
      "step": 1682
    },
    {
      "epoch": 0.5314177455004736,
      "grad_norm": 0.7865704894065857,
      "learning_rate": 4.965265595396815e-05,
      "loss": 0.9644,
      "step": 1683
    },
    {
      "epoch": 0.5317335017366593,
      "grad_norm": 1.007348895072937,
      "learning_rate": 4.964914742825065e-05,
      "loss": 0.8999,
      "step": 1684
    },
    {
      "epoch": 0.532049257972845,
      "grad_norm": 0.8959454894065857,
      "learning_rate": 4.964563890253316e-05,
      "loss": 0.9826,
      "step": 1685
    },
    {
      "epoch": 0.5323650142090306,
      "grad_norm": 1.0002062320709229,
      "learning_rate": 4.9642130376815663e-05,
      "loss": 0.9356,
      "step": 1686
    },
    {
      "epoch": 0.5326807704452163,
      "grad_norm": 0.7791248559951782,
      "learning_rate": 4.963862185109817e-05,
      "loss": 0.9431,
      "step": 1687
    },
    {
      "epoch": 0.532996526681402,
      "grad_norm": 0.872742235660553,
      "learning_rate": 4.963511332538068e-05,
      "loss": 0.9496,
      "step": 1688
    },
    {
      "epoch": 0.5333122829175876,
      "grad_norm": 1.0292694568634033,
      "learning_rate": 4.9631604799663184e-05,
      "loss": 0.9195,
      "step": 1689
    },
    {
      "epoch": 0.5336280391537733,
      "grad_norm": 0.8692644834518433,
      "learning_rate": 4.962809627394569e-05,
      "loss": 0.9611,
      "step": 1690
    },
    {
      "epoch": 0.533943795389959,
      "grad_norm": 1.0475646257400513,
      "learning_rate": 4.96245877482282e-05,
      "loss": 1.0083,
      "step": 1691
    },
    {
      "epoch": 0.5342595516261446,
      "grad_norm": 0.7850935459136963,
      "learning_rate": 4.9621079222510704e-05,
      "loss": 0.9484,
      "step": 1692
    },
    {
      "epoch": 0.5345753078623303,
      "grad_norm": 0.9837402701377869,
      "learning_rate": 4.961757069679321e-05,
      "loss": 0.9522,
      "step": 1693
    },
    {
      "epoch": 0.534891064098516,
      "grad_norm": 0.810330331325531,
      "learning_rate": 4.961406217107572e-05,
      "loss": 0.9339,
      "step": 1694
    },
    {
      "epoch": 0.5352068203347016,
      "grad_norm": 0.8215145468711853,
      "learning_rate": 4.961055364535822e-05,
      "loss": 0.9327,
      "step": 1695
    },
    {
      "epoch": 0.5355225765708873,
      "grad_norm": 0.7811933755874634,
      "learning_rate": 4.9607045119640725e-05,
      "loss": 0.9683,
      "step": 1696
    },
    {
      "epoch": 0.5358383328070729,
      "grad_norm": 0.7174308896064758,
      "learning_rate": 4.960353659392323e-05,
      "loss": 0.9572,
      "step": 1697
    },
    {
      "epoch": 0.5361540890432586,
      "grad_norm": 0.9514665603637695,
      "learning_rate": 4.9600028068205745e-05,
      "loss": 0.9176,
      "step": 1698
    },
    {
      "epoch": 0.5364698452794443,
      "grad_norm": 0.6549060344696045,
      "learning_rate": 4.959651954248825e-05,
      "loss": 1.0189,
      "step": 1699
    },
    {
      "epoch": 0.5367856015156299,
      "grad_norm": 0.9747973084449768,
      "learning_rate": 4.959301101677076e-05,
      "loss": 0.9439,
      "step": 1700
    },
    {
      "epoch": 0.5371013577518156,
      "grad_norm": 1.033708930015564,
      "learning_rate": 4.9589502491053266e-05,
      "loss": 0.9593,
      "step": 1701
    },
    {
      "epoch": 0.5374171139880013,
      "grad_norm": 0.8559411764144897,
      "learning_rate": 4.958599396533577e-05,
      "loss": 1.0054,
      "step": 1702
    },
    {
      "epoch": 0.5377328702241869,
      "grad_norm": 0.8490167260169983,
      "learning_rate": 4.958248543961827e-05,
      "loss": 1.0156,
      "step": 1703
    },
    {
      "epoch": 0.5380486264603725,
      "grad_norm": 1.8064954280853271,
      "learning_rate": 4.957897691390078e-05,
      "loss": 0.9679,
      "step": 1704
    },
    {
      "epoch": 0.5383643826965583,
      "grad_norm": 0.7426989078521729,
      "learning_rate": 4.9575468388183286e-05,
      "loss": 0.9458,
      "step": 1705
    },
    {
      "epoch": 0.5386801389327439,
      "grad_norm": 1.1199188232421875,
      "learning_rate": 4.957195986246579e-05,
      "loss": 0.9521,
      "step": 1706
    },
    {
      "epoch": 0.5389958951689295,
      "grad_norm": 1.1293978691101074,
      "learning_rate": 4.95684513367483e-05,
      "loss": 0.9398,
      "step": 1707
    },
    {
      "epoch": 0.5393116514051153,
      "grad_norm": 0.9343318939208984,
      "learning_rate": 4.9564942811030807e-05,
      "loss": 0.9525,
      "step": 1708
    },
    {
      "epoch": 0.5396274076413009,
      "grad_norm": 0.6753599643707275,
      "learning_rate": 4.956143428531331e-05,
      "loss": 0.9272,
      "step": 1709
    },
    {
      "epoch": 0.5399431638774865,
      "grad_norm": 0.6040274500846863,
      "learning_rate": 4.955792575959582e-05,
      "loss": 0.9185,
      "step": 1710
    },
    {
      "epoch": 0.5402589201136723,
      "grad_norm": 0.877693235874176,
      "learning_rate": 4.955441723387833e-05,
      "loss": 0.9534,
      "step": 1711
    },
    {
      "epoch": 0.5405746763498579,
      "grad_norm": 0.878972053527832,
      "learning_rate": 4.9550908708160834e-05,
      "loss": 0.9728,
      "step": 1712
    },
    {
      "epoch": 0.5408904325860435,
      "grad_norm": 0.8091248869895935,
      "learning_rate": 4.954740018244334e-05,
      "loss": 0.9893,
      "step": 1713
    },
    {
      "epoch": 0.5412061888222293,
      "grad_norm": 1.0066298246383667,
      "learning_rate": 4.954389165672584e-05,
      "loss": 0.9483,
      "step": 1714
    },
    {
      "epoch": 0.5415219450584149,
      "grad_norm": 0.8048312067985535,
      "learning_rate": 4.954038313100835e-05,
      "loss": 0.9392,
      "step": 1715
    },
    {
      "epoch": 0.5418377012946005,
      "grad_norm": 0.8592706918716431,
      "learning_rate": 4.953687460529086e-05,
      "loss": 0.9384,
      "step": 1716
    },
    {
      "epoch": 0.5421534575307863,
      "grad_norm": 0.8595340847969055,
      "learning_rate": 4.953336607957337e-05,
      "loss": 0.9682,
      "step": 1717
    },
    {
      "epoch": 0.5424692137669719,
      "grad_norm": 0.8036788702011108,
      "learning_rate": 4.9529857553855875e-05,
      "loss": 0.9693,
      "step": 1718
    },
    {
      "epoch": 0.5427849700031575,
      "grad_norm": 0.8129424452781677,
      "learning_rate": 4.952634902813838e-05,
      "loss": 0.9278,
      "step": 1719
    },
    {
      "epoch": 0.5431007262393432,
      "grad_norm": 0.9042577147483826,
      "learning_rate": 4.952284050242089e-05,
      "loss": 0.9803,
      "step": 1720
    },
    {
      "epoch": 0.5434164824755289,
      "grad_norm": 0.940601646900177,
      "learning_rate": 4.9519331976703395e-05,
      "loss": 0.955,
      "step": 1721
    },
    {
      "epoch": 0.5437322387117146,
      "grad_norm": 0.9299985766410828,
      "learning_rate": 4.95158234509859e-05,
      "loss": 0.9267,
      "step": 1722
    },
    {
      "epoch": 0.5440479949479002,
      "grad_norm": 0.9515458345413208,
      "learning_rate": 4.95123149252684e-05,
      "loss": 0.9759,
      "step": 1723
    },
    {
      "epoch": 0.5443637511840859,
      "grad_norm": 0.8188782334327698,
      "learning_rate": 4.950880639955091e-05,
      "loss": 0.9723,
      "step": 1724
    },
    {
      "epoch": 0.5446795074202716,
      "grad_norm": 0.9041300415992737,
      "learning_rate": 4.9505297873833416e-05,
      "loss": 0.9382,
      "step": 1725
    },
    {
      "epoch": 0.5449952636564572,
      "grad_norm": 0.9017472863197327,
      "learning_rate": 4.950178934811592e-05,
      "loss": 0.9121,
      "step": 1726
    },
    {
      "epoch": 0.5453110198926429,
      "grad_norm": 0.6774863004684448,
      "learning_rate": 4.949828082239843e-05,
      "loss": 0.9278,
      "step": 1727
    },
    {
      "epoch": 0.5456267761288286,
      "grad_norm": 0.9924618005752563,
      "learning_rate": 4.9494772296680936e-05,
      "loss": 0.968,
      "step": 1728
    },
    {
      "epoch": 0.5459425323650142,
      "grad_norm": 0.8920148611068726,
      "learning_rate": 4.949126377096344e-05,
      "loss": 0.935,
      "step": 1729
    },
    {
      "epoch": 0.5462582886011998,
      "grad_norm": 1.00996994972229,
      "learning_rate": 4.948775524524595e-05,
      "loss": 1.0098,
      "step": 1730
    },
    {
      "epoch": 0.5465740448373856,
      "grad_norm": 0.8752623796463013,
      "learning_rate": 4.9484246719528456e-05,
      "loss": 0.9478,
      "step": 1731
    },
    {
      "epoch": 0.5468898010735712,
      "grad_norm": 0.8724526762962341,
      "learning_rate": 4.948073819381096e-05,
      "loss": 0.9025,
      "step": 1732
    },
    {
      "epoch": 0.5472055573097568,
      "grad_norm": 0.8180859088897705,
      "learning_rate": 4.947722966809347e-05,
      "loss": 0.9612,
      "step": 1733
    },
    {
      "epoch": 0.5475213135459426,
      "grad_norm": 0.8301462531089783,
      "learning_rate": 4.947372114237598e-05,
      "loss": 0.9401,
      "step": 1734
    },
    {
      "epoch": 0.5478370697821282,
      "grad_norm": 0.7964144945144653,
      "learning_rate": 4.9470212616658484e-05,
      "loss": 0.9752,
      "step": 1735
    },
    {
      "epoch": 0.5481528260183138,
      "grad_norm": 0.876174807548523,
      "learning_rate": 4.946670409094099e-05,
      "loss": 0.9532,
      "step": 1736
    },
    {
      "epoch": 0.5484685822544996,
      "grad_norm": 0.8040497899055481,
      "learning_rate": 4.94631955652235e-05,
      "loss": 0.9578,
      "step": 1737
    },
    {
      "epoch": 0.5487843384906852,
      "grad_norm": 0.8855234384536743,
      "learning_rate": 4.9459687039506004e-05,
      "loss": 0.9285,
      "step": 1738
    },
    {
      "epoch": 0.5491000947268708,
      "grad_norm": 1.500675082206726,
      "learning_rate": 4.945617851378851e-05,
      "loss": 0.9227,
      "step": 1739
    },
    {
      "epoch": 0.5494158509630566,
      "grad_norm": 0.8640146851539612,
      "learning_rate": 4.945266998807102e-05,
      "loss": 0.9347,
      "step": 1740
    },
    {
      "epoch": 0.5497316071992422,
      "grad_norm": 0.9587616920471191,
      "learning_rate": 4.9449161462353525e-05,
      "loss": 0.9632,
      "step": 1741
    },
    {
      "epoch": 0.5500473634354278,
      "grad_norm": 0.8362916707992554,
      "learning_rate": 4.9445652936636025e-05,
      "loss": 0.9983,
      "step": 1742
    },
    {
      "epoch": 0.5503631196716136,
      "grad_norm": 0.8543821573257446,
      "learning_rate": 4.944214441091853e-05,
      "loss": 0.9216,
      "step": 1743
    },
    {
      "epoch": 0.5506788759077992,
      "grad_norm": 1.3342403173446655,
      "learning_rate": 4.943863588520104e-05,
      "loss": 0.9453,
      "step": 1744
    },
    {
      "epoch": 0.5509946321439848,
      "grad_norm": 0.9361600875854492,
      "learning_rate": 4.9435127359483545e-05,
      "loss": 0.9142,
      "step": 1745
    },
    {
      "epoch": 0.5513103883801705,
      "grad_norm": 0.8977649807929993,
      "learning_rate": 4.943161883376605e-05,
      "loss": 0.9592,
      "step": 1746
    },
    {
      "epoch": 0.5516261446163562,
      "grad_norm": 0.8674793839454651,
      "learning_rate": 4.942811030804856e-05,
      "loss": 0.9,
      "step": 1747
    },
    {
      "epoch": 0.5519419008525418,
      "grad_norm": 0.9525165557861328,
      "learning_rate": 4.9424601782331066e-05,
      "loss": 0.9702,
      "step": 1748
    },
    {
      "epoch": 0.5522576570887275,
      "grad_norm": 0.7808541655540466,
      "learning_rate": 4.942109325661357e-05,
      "loss": 0.954,
      "step": 1749
    },
    {
      "epoch": 0.5525734133249132,
      "grad_norm": 0.7671249508857727,
      "learning_rate": 4.941758473089608e-05,
      "loss": 0.9579,
      "step": 1750
    },
    {
      "epoch": 0.5528891695610988,
      "grad_norm": 0.8305315375328064,
      "learning_rate": 4.9414076205178586e-05,
      "loss": 0.9568,
      "step": 1751
    },
    {
      "epoch": 0.5532049257972845,
      "grad_norm": 0.822963297367096,
      "learning_rate": 4.941056767946109e-05,
      "loss": 0.9588,
      "step": 1752
    },
    {
      "epoch": 0.5535206820334702,
      "grad_norm": 0.976752758026123,
      "learning_rate": 4.94070591537436e-05,
      "loss": 0.913,
      "step": 1753
    },
    {
      "epoch": 0.5538364382696558,
      "grad_norm": 1.0132802724838257,
      "learning_rate": 4.9403550628026106e-05,
      "loss": 1.0011,
      "step": 1754
    },
    {
      "epoch": 0.5541521945058415,
      "grad_norm": 0.6741126775741577,
      "learning_rate": 4.940004210230861e-05,
      "loss": 0.9628,
      "step": 1755
    },
    {
      "epoch": 0.5544679507420271,
      "grad_norm": 0.73548823595047,
      "learning_rate": 4.939653357659112e-05,
      "loss": 0.9324,
      "step": 1756
    },
    {
      "epoch": 0.5547837069782128,
      "grad_norm": 0.8434667587280273,
      "learning_rate": 4.939302505087363e-05,
      "loss": 0.9344,
      "step": 1757
    },
    {
      "epoch": 0.5550994632143985,
      "grad_norm": 0.9624160528182983,
      "learning_rate": 4.9389516525156134e-05,
      "loss": 1.025,
      "step": 1758
    },
    {
      "epoch": 0.5554152194505841,
      "grad_norm": 0.7695242762565613,
      "learning_rate": 4.938600799943864e-05,
      "loss": 0.9487,
      "step": 1759
    },
    {
      "epoch": 0.5557309756867698,
      "grad_norm": 0.7765544652938843,
      "learning_rate": 4.938249947372115e-05,
      "loss": 0.9804,
      "step": 1760
    },
    {
      "epoch": 0.5560467319229555,
      "grad_norm": 0.8511478900909424,
      "learning_rate": 4.937899094800365e-05,
      "loss": 0.9579,
      "step": 1761
    },
    {
      "epoch": 0.5563624881591411,
      "grad_norm": 2.147303819656372,
      "learning_rate": 4.9375482422286154e-05,
      "loss": 0.896,
      "step": 1762
    },
    {
      "epoch": 0.5566782443953268,
      "grad_norm": 0.9119481444358826,
      "learning_rate": 4.937197389656866e-05,
      "loss": 0.9506,
      "step": 1763
    },
    {
      "epoch": 0.5569940006315125,
      "grad_norm": 1.0686912536621094,
      "learning_rate": 4.936846537085117e-05,
      "loss": 0.9748,
      "step": 1764
    },
    {
      "epoch": 0.5573097568676981,
      "grad_norm": 0.9511919617652893,
      "learning_rate": 4.9364956845133675e-05,
      "loss": 0.9716,
      "step": 1765
    },
    {
      "epoch": 0.5576255131038838,
      "grad_norm": 0.8639541268348694,
      "learning_rate": 4.936144831941618e-05,
      "loss": 0.9251,
      "step": 1766
    },
    {
      "epoch": 0.5579412693400695,
      "grad_norm": 0.9162179231643677,
      "learning_rate": 4.935793979369869e-05,
      "loss": 0.9794,
      "step": 1767
    },
    {
      "epoch": 0.5582570255762551,
      "grad_norm": 1.0751301050186157,
      "learning_rate": 4.93544312679812e-05,
      "loss": 0.965,
      "step": 1768
    },
    {
      "epoch": 0.5585727818124407,
      "grad_norm": 0.8162358403205872,
      "learning_rate": 4.935092274226371e-05,
      "loss": 0.9469,
      "step": 1769
    },
    {
      "epoch": 0.5588885380486265,
      "grad_norm": 0.8899351358413696,
      "learning_rate": 4.934741421654621e-05,
      "loss": 0.9488,
      "step": 1770
    },
    {
      "epoch": 0.5592042942848121,
      "grad_norm": 0.8483148217201233,
      "learning_rate": 4.9343905690828715e-05,
      "loss": 0.9783,
      "step": 1771
    },
    {
      "epoch": 0.5595200505209977,
      "grad_norm": 0.8561343550682068,
      "learning_rate": 4.934039716511122e-05,
      "loss": 0.9533,
      "step": 1772
    },
    {
      "epoch": 0.5598358067571835,
      "grad_norm": 0.9675840139389038,
      "learning_rate": 4.933688863939373e-05,
      "loss": 0.9729,
      "step": 1773
    },
    {
      "epoch": 0.5601515629933691,
      "grad_norm": 0.8351843357086182,
      "learning_rate": 4.9333380113676236e-05,
      "loss": 0.9342,
      "step": 1774
    },
    {
      "epoch": 0.5604673192295548,
      "grad_norm": 1.2447516918182373,
      "learning_rate": 4.932987158795874e-05,
      "loss": 0.9706,
      "step": 1775
    },
    {
      "epoch": 0.5607830754657405,
      "grad_norm": 0.7474979162216187,
      "learning_rate": 4.932636306224125e-05,
      "loss": 0.9376,
      "step": 1776
    },
    {
      "epoch": 0.5610988317019261,
      "grad_norm": 0.7896727919578552,
      "learning_rate": 4.9322854536523756e-05,
      "loss": 0.9634,
      "step": 1777
    },
    {
      "epoch": 0.5614145879381118,
      "grad_norm": 0.9381178021430969,
      "learning_rate": 4.931934601080626e-05,
      "loss": 0.9657,
      "step": 1778
    },
    {
      "epoch": 0.5617303441742975,
      "grad_norm": 0.7363710999488831,
      "learning_rate": 4.931583748508877e-05,
      "loss": 0.9214,
      "step": 1779
    },
    {
      "epoch": 0.5620461004104831,
      "grad_norm": 0.8431194424629211,
      "learning_rate": 4.931232895937128e-05,
      "loss": 0.969,
      "step": 1780
    },
    {
      "epoch": 0.5623618566466688,
      "grad_norm": 0.7039254307746887,
      "learning_rate": 4.930882043365378e-05,
      "loss": 0.9509,
      "step": 1781
    },
    {
      "epoch": 0.5626776128828544,
      "grad_norm": 0.6372224688529968,
      "learning_rate": 4.9305311907936284e-05,
      "loss": 0.9265,
      "step": 1782
    },
    {
      "epoch": 0.5629933691190401,
      "grad_norm": 0.9972807168960571,
      "learning_rate": 4.930180338221879e-05,
      "loss": 0.9449,
      "step": 1783
    },
    {
      "epoch": 0.5633091253552258,
      "grad_norm": 0.6911554932594299,
      "learning_rate": 4.92982948565013e-05,
      "loss": 0.9809,
      "step": 1784
    },
    {
      "epoch": 0.5636248815914114,
      "grad_norm": 0.8718883991241455,
      "learning_rate": 4.9294786330783804e-05,
      "loss": 0.9132,
      "step": 1785
    },
    {
      "epoch": 0.5639406378275971,
      "grad_norm": 0.8323532342910767,
      "learning_rate": 4.929127780506632e-05,
      "loss": 0.9291,
      "step": 1786
    },
    {
      "epoch": 0.5642563940637828,
      "grad_norm": 0.9451659917831421,
      "learning_rate": 4.9287769279348824e-05,
      "loss": 0.9108,
      "step": 1787
    },
    {
      "epoch": 0.5645721502999684,
      "grad_norm": 0.9542780518531799,
      "learning_rate": 4.928426075363133e-05,
      "loss": 0.9041,
      "step": 1788
    },
    {
      "epoch": 0.5648879065361541,
      "grad_norm": 0.6628958582878113,
      "learning_rate": 4.928075222791383e-05,
      "loss": 0.9805,
      "step": 1789
    },
    {
      "epoch": 0.5652036627723398,
      "grad_norm": 1.2235008478164673,
      "learning_rate": 4.927724370219634e-05,
      "loss": 0.9607,
      "step": 1790
    },
    {
      "epoch": 0.5655194190085254,
      "grad_norm": 0.9819533824920654,
      "learning_rate": 4.9273735176478845e-05,
      "loss": 0.9343,
      "step": 1791
    },
    {
      "epoch": 0.565835175244711,
      "grad_norm": 1.0759227275848389,
      "learning_rate": 4.927022665076135e-05,
      "loss": 0.9545,
      "step": 1792
    },
    {
      "epoch": 0.5661509314808968,
      "grad_norm": 0.7843246459960938,
      "learning_rate": 4.926671812504386e-05,
      "loss": 0.9032,
      "step": 1793
    },
    {
      "epoch": 0.5664666877170824,
      "grad_norm": 0.678118109703064,
      "learning_rate": 4.9263209599326365e-05,
      "loss": 0.9021,
      "step": 1794
    },
    {
      "epoch": 0.566782443953268,
      "grad_norm": 0.9426901936531067,
      "learning_rate": 4.925970107360887e-05,
      "loss": 0.9738,
      "step": 1795
    },
    {
      "epoch": 0.5670982001894538,
      "grad_norm": 1.2004354000091553,
      "learning_rate": 4.925619254789138e-05,
      "loss": 0.9439,
      "step": 1796
    },
    {
      "epoch": 0.5674139564256394,
      "grad_norm": 0.9385836124420166,
      "learning_rate": 4.9252684022173886e-05,
      "loss": 0.9322,
      "step": 1797
    },
    {
      "epoch": 0.567729712661825,
      "grad_norm": 0.9176542162895203,
      "learning_rate": 4.924917549645639e-05,
      "loss": 0.905,
      "step": 1798
    },
    {
      "epoch": 0.5680454688980108,
      "grad_norm": 0.8118360638618469,
      "learning_rate": 4.92456669707389e-05,
      "loss": 0.9446,
      "step": 1799
    },
    {
      "epoch": 0.5683612251341964,
      "grad_norm": 0.8382562398910522,
      "learning_rate": 4.92421584450214e-05,
      "loss": 0.9467,
      "step": 1800
    },
    {
      "epoch": 0.5683612251341964,
      "eval_loss": 0.9138574600219727,
      "eval_runtime": 4.618,
      "eval_samples_per_second": 539.197,
      "eval_steps_per_second": 4.547,
      "step": 1800
    },
    {
      "epoch": 0.568676981370382,
      "grad_norm": 1.3392270803451538,
      "learning_rate": 4.9238649919303906e-05,
      "loss": 0.9605,
      "step": 1801
    },
    {
      "epoch": 0.5689927376065678,
      "grad_norm": 1.0448668003082275,
      "learning_rate": 4.923514139358641e-05,
      "loss": 0.9051,
      "step": 1802
    },
    {
      "epoch": 0.5693084938427534,
      "grad_norm": 0.9781596064567566,
      "learning_rate": 4.923163286786893e-05,
      "loss": 0.9357,
      "step": 1803
    },
    {
      "epoch": 0.569624250078939,
      "grad_norm": 0.8086451292037964,
      "learning_rate": 4.9228124342151433e-05,
      "loss": 0.96,
      "step": 1804
    },
    {
      "epoch": 0.5699400063151248,
      "grad_norm": 0.8830196857452393,
      "learning_rate": 4.922461581643394e-05,
      "loss": 0.9539,
      "step": 1805
    },
    {
      "epoch": 0.5702557625513104,
      "grad_norm": 0.882318913936615,
      "learning_rate": 4.922110729071645e-05,
      "loss": 0.9515,
      "step": 1806
    },
    {
      "epoch": 0.570571518787496,
      "grad_norm": 0.8065032958984375,
      "learning_rate": 4.9217598764998954e-05,
      "loss": 0.9396,
      "step": 1807
    },
    {
      "epoch": 0.5708872750236818,
      "grad_norm": 0.7637073993682861,
      "learning_rate": 4.9214090239281454e-05,
      "loss": 0.9005,
      "step": 1808
    },
    {
      "epoch": 0.5712030312598674,
      "grad_norm": 0.8187084197998047,
      "learning_rate": 4.921058171356396e-05,
      "loss": 0.9545,
      "step": 1809
    },
    {
      "epoch": 0.571518787496053,
      "grad_norm": 0.8954304456710815,
      "learning_rate": 4.920707318784647e-05,
      "loss": 0.9341,
      "step": 1810
    },
    {
      "epoch": 0.5718345437322387,
      "grad_norm": 0.9159319996833801,
      "learning_rate": 4.9203564662128974e-05,
      "loss": 0.9691,
      "step": 1811
    },
    {
      "epoch": 0.5721502999684244,
      "grad_norm": 1.0181429386138916,
      "learning_rate": 4.920005613641148e-05,
      "loss": 0.931,
      "step": 1812
    },
    {
      "epoch": 0.57246605620461,
      "grad_norm": 0.8370779156684875,
      "learning_rate": 4.919654761069399e-05,
      "loss": 0.9015,
      "step": 1813
    },
    {
      "epoch": 0.5727818124407957,
      "grad_norm": 1.179287075996399,
      "learning_rate": 4.9193039084976495e-05,
      "loss": 0.8979,
      "step": 1814
    },
    {
      "epoch": 0.5730975686769814,
      "grad_norm": 1.10329270362854,
      "learning_rate": 4.9189530559259e-05,
      "loss": 0.9767,
      "step": 1815
    },
    {
      "epoch": 0.573413324913167,
      "grad_norm": 0.8951327800750732,
      "learning_rate": 4.918602203354151e-05,
      "loss": 0.9506,
      "step": 1816
    },
    {
      "epoch": 0.5737290811493527,
      "grad_norm": 1.1991454362869263,
      "learning_rate": 4.9182513507824015e-05,
      "loss": 0.9427,
      "step": 1817
    },
    {
      "epoch": 0.5740448373855384,
      "grad_norm": 0.9961807727813721,
      "learning_rate": 4.917900498210652e-05,
      "loss": 0.9271,
      "step": 1818
    },
    {
      "epoch": 0.574360593621724,
      "grad_norm": 0.7980900406837463,
      "learning_rate": 4.917549645638902e-05,
      "loss": 0.963,
      "step": 1819
    },
    {
      "epoch": 0.5746763498579097,
      "grad_norm": 1.1409136056900024,
      "learning_rate": 4.917198793067153e-05,
      "loss": 1.0371,
      "step": 1820
    },
    {
      "epoch": 0.5749921060940953,
      "grad_norm": 0.8525678515434265,
      "learning_rate": 4.916847940495404e-05,
      "loss": 0.9381,
      "step": 1821
    },
    {
      "epoch": 0.575307862330281,
      "grad_norm": 1.0946546792984009,
      "learning_rate": 4.916497087923655e-05,
      "loss": 0.9602,
      "step": 1822
    },
    {
      "epoch": 0.5756236185664667,
      "grad_norm": 1.2019906044006348,
      "learning_rate": 4.9161462353519056e-05,
      "loss": 0.918,
      "step": 1823
    },
    {
      "epoch": 0.5759393748026523,
      "grad_norm": 1.1223664283752441,
      "learning_rate": 4.915795382780156e-05,
      "loss": 0.9649,
      "step": 1824
    },
    {
      "epoch": 0.576255131038838,
      "grad_norm": 1.0721524953842163,
      "learning_rate": 4.915444530208407e-05,
      "loss": 0.9879,
      "step": 1825
    },
    {
      "epoch": 0.5765708872750237,
      "grad_norm": 0.9468700289726257,
      "learning_rate": 4.9150936776366577e-05,
      "loss": 0.961,
      "step": 1826
    },
    {
      "epoch": 0.5768866435112093,
      "grad_norm": 1.0753976106643677,
      "learning_rate": 4.9147428250649083e-05,
      "loss": 0.9411,
      "step": 1827
    },
    {
      "epoch": 0.577202399747395,
      "grad_norm": 0.8398197889328003,
      "learning_rate": 4.9143919724931583e-05,
      "loss": 0.926,
      "step": 1828
    },
    {
      "epoch": 0.5775181559835807,
      "grad_norm": 0.9491100907325745,
      "learning_rate": 4.914041119921409e-05,
      "loss": 0.9252,
      "step": 1829
    },
    {
      "epoch": 0.5778339122197663,
      "grad_norm": 0.9251866340637207,
      "learning_rate": 4.91369026734966e-05,
      "loss": 0.9259,
      "step": 1830
    },
    {
      "epoch": 0.5781496684559521,
      "grad_norm": 1.114724040031433,
      "learning_rate": 4.9133394147779104e-05,
      "loss": 0.9546,
      "step": 1831
    },
    {
      "epoch": 0.5784654246921377,
      "grad_norm": 0.9273649454116821,
      "learning_rate": 4.912988562206161e-05,
      "loss": 0.9378,
      "step": 1832
    },
    {
      "epoch": 0.5787811809283233,
      "grad_norm": 0.9369655251502991,
      "learning_rate": 4.912637709634412e-05,
      "loss": 0.9728,
      "step": 1833
    },
    {
      "epoch": 0.579096937164509,
      "grad_norm": 1.0078999996185303,
      "learning_rate": 4.9122868570626624e-05,
      "loss": 0.9627,
      "step": 1834
    },
    {
      "epoch": 0.5794126934006947,
      "grad_norm": 1.212428331375122,
      "learning_rate": 4.911936004490913e-05,
      "loss": 0.939,
      "step": 1835
    },
    {
      "epoch": 0.5797284496368803,
      "grad_norm": 1.0436569452285767,
      "learning_rate": 4.911585151919164e-05,
      "loss": 0.9504,
      "step": 1836
    },
    {
      "epoch": 0.580044205873066,
      "grad_norm": 1.089475154876709,
      "learning_rate": 4.9112342993474145e-05,
      "loss": 0.9512,
      "step": 1837
    },
    {
      "epoch": 0.5803599621092517,
      "grad_norm": 1.6419544219970703,
      "learning_rate": 4.910883446775665e-05,
      "loss": 0.9469,
      "step": 1838
    },
    {
      "epoch": 0.5806757183454373,
      "grad_norm": 1.116603136062622,
      "learning_rate": 4.910532594203916e-05,
      "loss": 0.9649,
      "step": 1839
    },
    {
      "epoch": 0.580991474581623,
      "grad_norm": 0.9088723063468933,
      "learning_rate": 4.9101817416321665e-05,
      "loss": 0.9575,
      "step": 1840
    },
    {
      "epoch": 0.5813072308178087,
      "grad_norm": 1.1529728174209595,
      "learning_rate": 4.909830889060417e-05,
      "loss": 1.0202,
      "step": 1841
    },
    {
      "epoch": 0.5816229870539943,
      "grad_norm": 1.2751681804656982,
      "learning_rate": 4.909480036488668e-05,
      "loss": 0.9417,
      "step": 1842
    },
    {
      "epoch": 0.58193874329018,
      "grad_norm": 1.6209897994995117,
      "learning_rate": 4.9091291839169186e-05,
      "loss": 0.9764,
      "step": 1843
    },
    {
      "epoch": 0.5822544995263657,
      "grad_norm": 1.0138822793960571,
      "learning_rate": 4.908778331345169e-05,
      "loss": 0.9951,
      "step": 1844
    },
    {
      "epoch": 0.5825702557625513,
      "grad_norm": 0.8635496497154236,
      "learning_rate": 4.90842747877342e-05,
      "loss": 0.9681,
      "step": 1845
    },
    {
      "epoch": 0.582886011998737,
      "grad_norm": 0.8969125747680664,
      "learning_rate": 4.9080766262016706e-05,
      "loss": 0.9853,
      "step": 1846
    },
    {
      "epoch": 0.5832017682349226,
      "grad_norm": 1.0051966905593872,
      "learning_rate": 4.9077257736299206e-05,
      "loss": 0.9382,
      "step": 1847
    },
    {
      "epoch": 0.5835175244711083,
      "grad_norm": 1.1173367500305176,
      "learning_rate": 4.907374921058171e-05,
      "loss": 0.9623,
      "step": 1848
    },
    {
      "epoch": 0.583833280707294,
      "grad_norm": 0.7901454567909241,
      "learning_rate": 4.907024068486422e-05,
      "loss": 0.9464,
      "step": 1849
    },
    {
      "epoch": 0.5841490369434796,
      "grad_norm": 0.9092691540718079,
      "learning_rate": 4.9066732159146727e-05,
      "loss": 0.9333,
      "step": 1850
    },
    {
      "epoch": 0.5844647931796653,
      "grad_norm": 0.9114457964897156,
      "learning_rate": 4.906322363342923e-05,
      "loss": 0.9435,
      "step": 1851
    },
    {
      "epoch": 0.584780549415851,
      "grad_norm": 1.0674177408218384,
      "learning_rate": 4.905971510771174e-05,
      "loss": 0.8978,
      "step": 1852
    },
    {
      "epoch": 0.5850963056520366,
      "grad_norm": 1.0657317638397217,
      "learning_rate": 4.905620658199425e-05,
      "loss": 0.9716,
      "step": 1853
    },
    {
      "epoch": 0.5854120618882223,
      "grad_norm": 1.0281494855880737,
      "learning_rate": 4.9052698056276754e-05,
      "loss": 0.9494,
      "step": 1854
    },
    {
      "epoch": 0.585727818124408,
      "grad_norm": 0.9179128408432007,
      "learning_rate": 4.904918953055926e-05,
      "loss": 0.9291,
      "step": 1855
    },
    {
      "epoch": 0.5860435743605936,
      "grad_norm": 0.8110784292221069,
      "learning_rate": 4.904568100484177e-05,
      "loss": 0.9778,
      "step": 1856
    },
    {
      "epoch": 0.5863593305967792,
      "grad_norm": 0.8638858199119568,
      "learning_rate": 4.9042172479124274e-05,
      "loss": 0.9281,
      "step": 1857
    },
    {
      "epoch": 0.586675086832965,
      "grad_norm": 0.980686366558075,
      "learning_rate": 4.903866395340678e-05,
      "loss": 0.9387,
      "step": 1858
    },
    {
      "epoch": 0.5869908430691506,
      "grad_norm": 0.9403325319290161,
      "learning_rate": 4.903515542768929e-05,
      "loss": 0.9201,
      "step": 1859
    },
    {
      "epoch": 0.5873065993053362,
      "grad_norm": 0.7984651923179626,
      "learning_rate": 4.9031646901971795e-05,
      "loss": 0.9154,
      "step": 1860
    },
    {
      "epoch": 0.587622355541522,
      "grad_norm": 0.8609490394592285,
      "learning_rate": 4.90281383762543e-05,
      "loss": 0.8806,
      "step": 1861
    },
    {
      "epoch": 0.5879381117777076,
      "grad_norm": 0.8168870806694031,
      "learning_rate": 4.902462985053681e-05,
      "loss": 0.9661,
      "step": 1862
    },
    {
      "epoch": 0.5882538680138932,
      "grad_norm": 0.7330571413040161,
      "learning_rate": 4.9021121324819315e-05,
      "loss": 0.8991,
      "step": 1863
    },
    {
      "epoch": 0.588569624250079,
      "grad_norm": 0.8381677865982056,
      "learning_rate": 4.901761279910182e-05,
      "loss": 0.9466,
      "step": 1864
    },
    {
      "epoch": 0.5888853804862646,
      "grad_norm": 0.7166131734848022,
      "learning_rate": 4.901410427338433e-05,
      "loss": 0.887,
      "step": 1865
    },
    {
      "epoch": 0.5892011367224502,
      "grad_norm": 0.7111034989356995,
      "learning_rate": 4.901059574766683e-05,
      "loss": 0.9314,
      "step": 1866
    },
    {
      "epoch": 0.589516892958636,
      "grad_norm": 0.821064293384552,
      "learning_rate": 4.9007087221949336e-05,
      "loss": 0.9311,
      "step": 1867
    },
    {
      "epoch": 0.5898326491948216,
      "grad_norm": 0.9002666473388672,
      "learning_rate": 4.900357869623184e-05,
      "loss": 0.9353,
      "step": 1868
    },
    {
      "epoch": 0.5901484054310072,
      "grad_norm": 0.9902474284172058,
      "learning_rate": 4.900007017051435e-05,
      "loss": 0.9751,
      "step": 1869
    },
    {
      "epoch": 0.590464161667193,
      "grad_norm": 0.7968148589134216,
      "learning_rate": 4.8996561644796856e-05,
      "loss": 0.8703,
      "step": 1870
    },
    {
      "epoch": 0.5907799179033786,
      "grad_norm": 0.9254236817359924,
      "learning_rate": 4.899305311907936e-05,
      "loss": 0.9062,
      "step": 1871
    },
    {
      "epoch": 0.5910956741395642,
      "grad_norm": 0.8516504764556885,
      "learning_rate": 4.898954459336187e-05,
      "loss": 0.9366,
      "step": 1872
    },
    {
      "epoch": 0.59141143037575,
      "grad_norm": 0.9770042896270752,
      "learning_rate": 4.898603606764438e-05,
      "loss": 0.978,
      "step": 1873
    },
    {
      "epoch": 0.5917271866119356,
      "grad_norm": 1.1613366603851318,
      "learning_rate": 4.898252754192689e-05,
      "loss": 0.9156,
      "step": 1874
    },
    {
      "epoch": 0.5920429428481212,
      "grad_norm": 0.7943533062934875,
      "learning_rate": 4.897901901620939e-05,
      "loss": 0.9188,
      "step": 1875
    },
    {
      "epoch": 0.5923586990843069,
      "grad_norm": 0.8606296181678772,
      "learning_rate": 4.89755104904919e-05,
      "loss": 0.9394,
      "step": 1876
    },
    {
      "epoch": 0.5926744553204926,
      "grad_norm": 1.0939128398895264,
      "learning_rate": 4.8972001964774404e-05,
      "loss": 1.0187,
      "step": 1877
    },
    {
      "epoch": 0.5929902115566782,
      "grad_norm": 1.0984824895858765,
      "learning_rate": 4.896849343905691e-05,
      "loss": 0.9843,
      "step": 1878
    },
    {
      "epoch": 0.5933059677928639,
      "grad_norm": 1.0705106258392334,
      "learning_rate": 4.896498491333942e-05,
      "loss": 0.924,
      "step": 1879
    },
    {
      "epoch": 0.5936217240290496,
      "grad_norm": 0.8987385630607605,
      "learning_rate": 4.8961476387621924e-05,
      "loss": 0.8881,
      "step": 1880
    },
    {
      "epoch": 0.5939374802652352,
      "grad_norm": 1.1135646104812622,
      "learning_rate": 4.895796786190443e-05,
      "loss": 0.9584,
      "step": 1881
    },
    {
      "epoch": 0.5942532365014209,
      "grad_norm": 0.8595818281173706,
      "learning_rate": 4.895445933618694e-05,
      "loss": 0.9422,
      "step": 1882
    },
    {
      "epoch": 0.5945689927376065,
      "grad_norm": 1.4354966878890991,
      "learning_rate": 4.8950950810469445e-05,
      "loss": 0.9733,
      "step": 1883
    },
    {
      "epoch": 0.5948847489737923,
      "grad_norm": 1.1235549449920654,
      "learning_rate": 4.894744228475195e-05,
      "loss": 0.9272,
      "step": 1884
    },
    {
      "epoch": 0.5952005052099779,
      "grad_norm": 0.8777201175689697,
      "learning_rate": 4.894393375903446e-05,
      "loss": 0.9196,
      "step": 1885
    },
    {
      "epoch": 0.5955162614461635,
      "grad_norm": 0.9098970293998718,
      "learning_rate": 4.894042523331696e-05,
      "loss": 0.9653,
      "step": 1886
    },
    {
      "epoch": 0.5958320176823493,
      "grad_norm": 0.8748881220817566,
      "learning_rate": 4.8936916707599465e-05,
      "loss": 0.9315,
      "step": 1887
    },
    {
      "epoch": 0.5961477739185349,
      "grad_norm": 1.0748125314712524,
      "learning_rate": 4.893340818188197e-05,
      "loss": 1.0011,
      "step": 1888
    },
    {
      "epoch": 0.5964635301547205,
      "grad_norm": 0.8721016645431519,
      "learning_rate": 4.892989965616448e-05,
      "loss": 0.946,
      "step": 1889
    },
    {
      "epoch": 0.5967792863909063,
      "grad_norm": 0.7757900357246399,
      "learning_rate": 4.8926391130446986e-05,
      "loss": 0.8949,
      "step": 1890
    },
    {
      "epoch": 0.5970950426270919,
      "grad_norm": 0.835759699344635,
      "learning_rate": 4.89228826047295e-05,
      "loss": 0.9031,
      "step": 1891
    },
    {
      "epoch": 0.5974107988632775,
      "grad_norm": 0.8293968439102173,
      "learning_rate": 4.8919374079012006e-05,
      "loss": 0.9724,
      "step": 1892
    },
    {
      "epoch": 0.5977265550994633,
      "grad_norm": 0.7953991889953613,
      "learning_rate": 4.891586555329451e-05,
      "loss": 0.929,
      "step": 1893
    },
    {
      "epoch": 0.5980423113356489,
      "grad_norm": 1.1218973398208618,
      "learning_rate": 4.891235702757701e-05,
      "loss": 0.966,
      "step": 1894
    },
    {
      "epoch": 0.5983580675718345,
      "grad_norm": 0.9522415995597839,
      "learning_rate": 4.890884850185952e-05,
      "loss": 0.9397,
      "step": 1895
    },
    {
      "epoch": 0.5986738238080203,
      "grad_norm": 0.9793238639831543,
      "learning_rate": 4.8905339976142026e-05,
      "loss": 0.9557,
      "step": 1896
    },
    {
      "epoch": 0.5989895800442059,
      "grad_norm": 1.5518311262130737,
      "learning_rate": 4.890183145042453e-05,
      "loss": 0.9345,
      "step": 1897
    },
    {
      "epoch": 0.5993053362803915,
      "grad_norm": 1.0705715417861938,
      "learning_rate": 4.889832292470704e-05,
      "loss": 0.9283,
      "step": 1898
    },
    {
      "epoch": 0.5996210925165772,
      "grad_norm": 0.9783518314361572,
      "learning_rate": 4.889481439898955e-05,
      "loss": 0.9431,
      "step": 1899
    },
    {
      "epoch": 0.5999368487527629,
      "grad_norm": 0.7436587810516357,
      "learning_rate": 4.8891305873272054e-05,
      "loss": 0.9093,
      "step": 1900
    },
    {
      "epoch": 0.6002526049889485,
      "grad_norm": 1.066175937652588,
      "learning_rate": 4.888779734755456e-05,
      "loss": 0.8827,
      "step": 1901
    },
    {
      "epoch": 0.6005683612251342,
      "grad_norm": 0.9477360844612122,
      "learning_rate": 4.888428882183707e-05,
      "loss": 0.9299,
      "step": 1902
    },
    {
      "epoch": 0.6008841174613199,
      "grad_norm": 1.0538673400878906,
      "learning_rate": 4.8880780296119574e-05,
      "loss": 0.8832,
      "step": 1903
    },
    {
      "epoch": 0.6011998736975055,
      "grad_norm": 0.7275685667991638,
      "learning_rate": 4.887727177040208e-05,
      "loss": 0.9238,
      "step": 1904
    },
    {
      "epoch": 0.6015156299336912,
      "grad_norm": 1.217555046081543,
      "learning_rate": 4.887376324468458e-05,
      "loss": 0.9581,
      "step": 1905
    },
    {
      "epoch": 0.6018313861698769,
      "grad_norm": 0.9281167387962341,
      "learning_rate": 4.887025471896709e-05,
      "loss": 0.9687,
      "step": 1906
    },
    {
      "epoch": 0.6021471424060625,
      "grad_norm": 0.8674626350402832,
      "learning_rate": 4.8866746193249595e-05,
      "loss": 0.9871,
      "step": 1907
    },
    {
      "epoch": 0.6024628986422482,
      "grad_norm": 1.0002055168151855,
      "learning_rate": 4.886323766753211e-05,
      "loss": 0.9876,
      "step": 1908
    },
    {
      "epoch": 0.6027786548784339,
      "grad_norm": 1.0585033893585205,
      "learning_rate": 4.8859729141814615e-05,
      "loss": 1.0027,
      "step": 1909
    },
    {
      "epoch": 0.6030944111146195,
      "grad_norm": 1.1221904754638672,
      "learning_rate": 4.885622061609712e-05,
      "loss": 0.9473,
      "step": 1910
    },
    {
      "epoch": 0.6034101673508052,
      "grad_norm": 1.169028401374817,
      "learning_rate": 4.885271209037963e-05,
      "loss": 0.9531,
      "step": 1911
    },
    {
      "epoch": 0.6037259235869908,
      "grad_norm": 0.8810638785362244,
      "learning_rate": 4.8849203564662135e-05,
      "loss": 0.9115,
      "step": 1912
    },
    {
      "epoch": 0.6040416798231765,
      "grad_norm": 0.8784321546554565,
      "learning_rate": 4.8845695038944635e-05,
      "loss": 0.9421,
      "step": 1913
    },
    {
      "epoch": 0.6043574360593622,
      "grad_norm": 0.964374840259552,
      "learning_rate": 4.884218651322714e-05,
      "loss": 0.8807,
      "step": 1914
    },
    {
      "epoch": 0.6046731922955478,
      "grad_norm": 0.8910267949104309,
      "learning_rate": 4.883867798750965e-05,
      "loss": 0.9232,
      "step": 1915
    },
    {
      "epoch": 0.6049889485317335,
      "grad_norm": 0.7147765755653381,
      "learning_rate": 4.8835169461792156e-05,
      "loss": 0.9651,
      "step": 1916
    },
    {
      "epoch": 0.6053047047679192,
      "grad_norm": 0.8665722012519836,
      "learning_rate": 4.883166093607466e-05,
      "loss": 0.9871,
      "step": 1917
    },
    {
      "epoch": 0.6056204610041048,
      "grad_norm": 0.9806735515594482,
      "learning_rate": 4.882815241035717e-05,
      "loss": 0.9892,
      "step": 1918
    },
    {
      "epoch": 0.6059362172402905,
      "grad_norm": 1.047466516494751,
      "learning_rate": 4.8824643884639676e-05,
      "loss": 0.9522,
      "step": 1919
    },
    {
      "epoch": 0.6062519734764762,
      "grad_norm": 0.8876450061798096,
      "learning_rate": 4.882113535892218e-05,
      "loss": 0.918,
      "step": 1920
    },
    {
      "epoch": 0.6065677297126618,
      "grad_norm": 0.8293878436088562,
      "learning_rate": 4.881762683320469e-05,
      "loss": 0.9278,
      "step": 1921
    },
    {
      "epoch": 0.6068834859488474,
      "grad_norm": 0.8068129420280457,
      "learning_rate": 4.88141183074872e-05,
      "loss": 0.9701,
      "step": 1922
    },
    {
      "epoch": 0.6071992421850332,
      "grad_norm": 0.7386300563812256,
      "learning_rate": 4.8810609781769704e-05,
      "loss": 0.9471,
      "step": 1923
    },
    {
      "epoch": 0.6075149984212188,
      "grad_norm": 0.8953647017478943,
      "learning_rate": 4.8807101256052204e-05,
      "loss": 0.9255,
      "step": 1924
    },
    {
      "epoch": 0.6078307546574044,
      "grad_norm": 0.9020692110061646,
      "learning_rate": 4.880359273033472e-05,
      "loss": 0.9337,
      "step": 1925
    },
    {
      "epoch": 0.6081465108935902,
      "grad_norm": 0.8738405108451843,
      "learning_rate": 4.8800084204617224e-05,
      "loss": 0.9468,
      "step": 1926
    },
    {
      "epoch": 0.6084622671297758,
      "grad_norm": 1.0650477409362793,
      "learning_rate": 4.879657567889973e-05,
      "loss": 0.9396,
      "step": 1927
    },
    {
      "epoch": 0.6087780233659614,
      "grad_norm": 0.8228299021720886,
      "learning_rate": 4.879306715318224e-05,
      "loss": 0.9743,
      "step": 1928
    },
    {
      "epoch": 0.6090937796021472,
      "grad_norm": 1.3267674446105957,
      "learning_rate": 4.8789558627464744e-05,
      "loss": 0.9333,
      "step": 1929
    },
    {
      "epoch": 0.6094095358383328,
      "grad_norm": 1.9717650413513184,
      "learning_rate": 4.878605010174725e-05,
      "loss": 0.9034,
      "step": 1930
    },
    {
      "epoch": 0.6097252920745184,
      "grad_norm": 0.8321818709373474,
      "learning_rate": 4.878254157602976e-05,
      "loss": 0.9527,
      "step": 1931
    },
    {
      "epoch": 0.6100410483107042,
      "grad_norm": 0.8500014543533325,
      "learning_rate": 4.8779033050312265e-05,
      "loss": 0.8909,
      "step": 1932
    },
    {
      "epoch": 0.6103568045468898,
      "grad_norm": 1.0827610492706299,
      "learning_rate": 4.8775524524594765e-05,
      "loss": 0.9043,
      "step": 1933
    },
    {
      "epoch": 0.6106725607830754,
      "grad_norm": 1.1310136318206787,
      "learning_rate": 4.877201599887727e-05,
      "loss": 0.8787,
      "step": 1934
    },
    {
      "epoch": 0.6109883170192612,
      "grad_norm": 1.3738023042678833,
      "learning_rate": 4.876850747315978e-05,
      "loss": 0.9201,
      "step": 1935
    },
    {
      "epoch": 0.6113040732554468,
      "grad_norm": 0.9244831204414368,
      "learning_rate": 4.8764998947442285e-05,
      "loss": 0.9773,
      "step": 1936
    },
    {
      "epoch": 0.6116198294916324,
      "grad_norm": 1.3421738147735596,
      "learning_rate": 4.876149042172479e-05,
      "loss": 0.9493,
      "step": 1937
    },
    {
      "epoch": 0.6119355857278181,
      "grad_norm": 1.1621137857437134,
      "learning_rate": 4.87579818960073e-05,
      "loss": 0.945,
      "step": 1938
    },
    {
      "epoch": 0.6122513419640038,
      "grad_norm": 0.96286541223526,
      "learning_rate": 4.8754473370289806e-05,
      "loss": 0.9507,
      "step": 1939
    },
    {
      "epoch": 0.6125670982001895,
      "grad_norm": 1.0684363842010498,
      "learning_rate": 4.875096484457231e-05,
      "loss": 0.944,
      "step": 1940
    },
    {
      "epoch": 0.6128828544363751,
      "grad_norm": 0.9910898804664612,
      "learning_rate": 4.874745631885482e-05,
      "loss": 0.9523,
      "step": 1941
    },
    {
      "epoch": 0.6131986106725608,
      "grad_norm": 1.0062695741653442,
      "learning_rate": 4.8743947793137326e-05,
      "loss": 0.9339,
      "step": 1942
    },
    {
      "epoch": 0.6135143669087465,
      "grad_norm": 0.872050404548645,
      "learning_rate": 4.874043926741983e-05,
      "loss": 0.9995,
      "step": 1943
    },
    {
      "epoch": 0.6138301231449321,
      "grad_norm": 1.2418783903121948,
      "learning_rate": 4.873693074170234e-05,
      "loss": 0.9792,
      "step": 1944
    },
    {
      "epoch": 0.6141458793811178,
      "grad_norm": 1.1709418296813965,
      "learning_rate": 4.873342221598485e-05,
      "loss": 0.9225,
      "step": 1945
    },
    {
      "epoch": 0.6144616356173035,
      "grad_norm": 0.8964934349060059,
      "learning_rate": 4.8729913690267354e-05,
      "loss": 0.9176,
      "step": 1946
    },
    {
      "epoch": 0.6147773918534891,
      "grad_norm": 1.043475866317749,
      "learning_rate": 4.872640516454986e-05,
      "loss": 0.9382,
      "step": 1947
    },
    {
      "epoch": 0.6150931480896747,
      "grad_norm": 1.0344654321670532,
      "learning_rate": 4.872289663883237e-05,
      "loss": 0.9575,
      "step": 1948
    },
    {
      "epoch": 0.6154089043258605,
      "grad_norm": 1.0773844718933105,
      "learning_rate": 4.8719388113114874e-05,
      "loss": 0.9149,
      "step": 1949
    },
    {
      "epoch": 0.6157246605620461,
      "grad_norm": 1.020099401473999,
      "learning_rate": 4.871587958739738e-05,
      "loss": 0.8964,
      "step": 1950
    },
    {
      "epoch": 0.6160404167982317,
      "grad_norm": 1.064841866493225,
      "learning_rate": 4.871237106167989e-05,
      "loss": 0.9625,
      "step": 1951
    },
    {
      "epoch": 0.6163561730344175,
      "grad_norm": 1.4740124940872192,
      "learning_rate": 4.870886253596239e-05,
      "loss": 0.9545,
      "step": 1952
    },
    {
      "epoch": 0.6166719292706031,
      "grad_norm": 1.2389352321624756,
      "learning_rate": 4.8705354010244894e-05,
      "loss": 0.9883,
      "step": 1953
    },
    {
      "epoch": 0.6169876855067887,
      "grad_norm": 0.9089992046356201,
      "learning_rate": 4.87018454845274e-05,
      "loss": 0.9287,
      "step": 1954
    },
    {
      "epoch": 0.6173034417429745,
      "grad_norm": 0.9316040873527527,
      "learning_rate": 4.869833695880991e-05,
      "loss": 0.9917,
      "step": 1955
    },
    {
      "epoch": 0.6176191979791601,
      "grad_norm": 0.9624812006950378,
      "learning_rate": 4.8694828433092415e-05,
      "loss": 0.8997,
      "step": 1956
    },
    {
      "epoch": 0.6179349542153457,
      "grad_norm": 1.0187394618988037,
      "learning_rate": 4.869131990737492e-05,
      "loss": 0.9568,
      "step": 1957
    },
    {
      "epoch": 0.6182507104515315,
      "grad_norm": 0.898311972618103,
      "learning_rate": 4.868781138165743e-05,
      "loss": 0.934,
      "step": 1958
    },
    {
      "epoch": 0.6185664666877171,
      "grad_norm": 1.0189944505691528,
      "learning_rate": 4.8684302855939935e-05,
      "loss": 0.9567,
      "step": 1959
    },
    {
      "epoch": 0.6188822229239027,
      "grad_norm": 0.9388458132743835,
      "learning_rate": 4.868079433022245e-05,
      "loss": 0.9522,
      "step": 1960
    },
    {
      "epoch": 0.6191979791600885,
      "grad_norm": 1.0711055994033813,
      "learning_rate": 4.867728580450495e-05,
      "loss": 0.9427,
      "step": 1961
    },
    {
      "epoch": 0.6195137353962741,
      "grad_norm": 0.7750597596168518,
      "learning_rate": 4.8673777278787456e-05,
      "loss": 0.9803,
      "step": 1962
    },
    {
      "epoch": 0.6198294916324597,
      "grad_norm": 0.9381986260414124,
      "learning_rate": 4.867026875306996e-05,
      "loss": 0.9417,
      "step": 1963
    },
    {
      "epoch": 0.6201452478686454,
      "grad_norm": 1.076449990272522,
      "learning_rate": 4.866676022735247e-05,
      "loss": 0.948,
      "step": 1964
    },
    {
      "epoch": 0.6204610041048311,
      "grad_norm": 0.7926503419876099,
      "learning_rate": 4.8663251701634976e-05,
      "loss": 0.9547,
      "step": 1965
    },
    {
      "epoch": 0.6207767603410167,
      "grad_norm": 1.0327174663543701,
      "learning_rate": 4.865974317591748e-05,
      "loss": 0.9134,
      "step": 1966
    },
    {
      "epoch": 0.6210925165772024,
      "grad_norm": 0.9219074249267578,
      "learning_rate": 4.865623465019999e-05,
      "loss": 0.9225,
      "step": 1967
    },
    {
      "epoch": 0.6214082728133881,
      "grad_norm": 0.891711413860321,
      "learning_rate": 4.8652726124482497e-05,
      "loss": 0.9584,
      "step": 1968
    },
    {
      "epoch": 0.6217240290495737,
      "grad_norm": 0.9064139127731323,
      "learning_rate": 4.8649217598765003e-05,
      "loss": 0.8774,
      "step": 1969
    },
    {
      "epoch": 0.6220397852857594,
      "grad_norm": 1.2095041275024414,
      "learning_rate": 4.864570907304751e-05,
      "loss": 0.8876,
      "step": 1970
    },
    {
      "epoch": 0.622355541521945,
      "grad_norm": 1.232414960861206,
      "learning_rate": 4.864220054733001e-05,
      "loss": 0.9743,
      "step": 1971
    },
    {
      "epoch": 0.6226712977581307,
      "grad_norm": 1.7205668687820435,
      "learning_rate": 4.863869202161252e-05,
      "loss": 0.9162,
      "step": 1972
    },
    {
      "epoch": 0.6229870539943164,
      "grad_norm": 0.9935349225997925,
      "learning_rate": 4.8635183495895024e-05,
      "loss": 0.935,
      "step": 1973
    },
    {
      "epoch": 0.623302810230502,
      "grad_norm": 0.945504367351532,
      "learning_rate": 4.863167497017753e-05,
      "loss": 0.8814,
      "step": 1974
    },
    {
      "epoch": 0.6236185664666877,
      "grad_norm": 0.9847025275230408,
      "learning_rate": 4.862816644446004e-05,
      "loss": 0.9607,
      "step": 1975
    },
    {
      "epoch": 0.6239343227028734,
      "grad_norm": 1.1909629106521606,
      "learning_rate": 4.8624657918742544e-05,
      "loss": 0.9725,
      "step": 1976
    },
    {
      "epoch": 0.624250078939059,
      "grad_norm": 1.0421901941299438,
      "learning_rate": 4.862114939302505e-05,
      "loss": 0.9382,
      "step": 1977
    },
    {
      "epoch": 0.6245658351752447,
      "grad_norm": 1.2649697065353394,
      "learning_rate": 4.8617640867307565e-05,
      "loss": 0.9575,
      "step": 1978
    },
    {
      "epoch": 0.6248815914114304,
      "grad_norm": 1.1930315494537354,
      "learning_rate": 4.861413234159007e-05,
      "loss": 0.9041,
      "step": 1979
    },
    {
      "epoch": 0.625197347647616,
      "grad_norm": 0.92356276512146,
      "learning_rate": 4.861062381587257e-05,
      "loss": 0.8894,
      "step": 1980
    },
    {
      "epoch": 0.6255131038838017,
      "grad_norm": 1.0690706968307495,
      "learning_rate": 4.860711529015508e-05,
      "loss": 0.9828,
      "step": 1981
    },
    {
      "epoch": 0.6258288601199874,
      "grad_norm": 0.957954466342926,
      "learning_rate": 4.8603606764437585e-05,
      "loss": 0.9235,
      "step": 1982
    },
    {
      "epoch": 0.626144616356173,
      "grad_norm": 1.0745303630828857,
      "learning_rate": 4.860009823872009e-05,
      "loss": 0.9167,
      "step": 1983
    },
    {
      "epoch": 0.6264603725923586,
      "grad_norm": 0.970833957195282,
      "learning_rate": 4.85965897130026e-05,
      "loss": 0.9845,
      "step": 1984
    },
    {
      "epoch": 0.6267761288285444,
      "grad_norm": 0.8796029686927795,
      "learning_rate": 4.8593081187285106e-05,
      "loss": 0.9433,
      "step": 1985
    },
    {
      "epoch": 0.62709188506473,
      "grad_norm": 0.8962977528572083,
      "learning_rate": 4.858957266156761e-05,
      "loss": 0.8834,
      "step": 1986
    },
    {
      "epoch": 0.6274076413009156,
      "grad_norm": 0.9952353835105896,
      "learning_rate": 4.858606413585012e-05,
      "loss": 0.9425,
      "step": 1987
    },
    {
      "epoch": 0.6277233975371014,
      "grad_norm": 0.8955731391906738,
      "learning_rate": 4.8582555610132626e-05,
      "loss": 0.8984,
      "step": 1988
    },
    {
      "epoch": 0.628039153773287,
      "grad_norm": 0.9068630933761597,
      "learning_rate": 4.857904708441513e-05,
      "loss": 0.9612,
      "step": 1989
    },
    {
      "epoch": 0.6283549100094726,
      "grad_norm": 0.8836802840232849,
      "learning_rate": 4.857553855869764e-05,
      "loss": 0.9478,
      "step": 1990
    },
    {
      "epoch": 0.6286706662456584,
      "grad_norm": 1.135219931602478,
      "learning_rate": 4.857203003298014e-05,
      "loss": 0.904,
      "step": 1991
    },
    {
      "epoch": 0.628986422481844,
      "grad_norm": 1.074337363243103,
      "learning_rate": 4.8568521507262647e-05,
      "loss": 0.9265,
      "step": 1992
    },
    {
      "epoch": 0.6293021787180296,
      "grad_norm": 0.9077310562133789,
      "learning_rate": 4.856501298154515e-05,
      "loss": 0.9339,
      "step": 1993
    },
    {
      "epoch": 0.6296179349542154,
      "grad_norm": 0.9551348686218262,
      "learning_rate": 4.856150445582766e-05,
      "loss": 0.8962,
      "step": 1994
    },
    {
      "epoch": 0.629933691190401,
      "grad_norm": 0.769536554813385,
      "learning_rate": 4.8557995930110174e-05,
      "loss": 0.9284,
      "step": 1995
    },
    {
      "epoch": 0.6302494474265867,
      "grad_norm": 0.936859130859375,
      "learning_rate": 4.855448740439268e-05,
      "loss": 0.9187,
      "step": 1996
    },
    {
      "epoch": 0.6305652036627724,
      "grad_norm": 1.351609706878662,
      "learning_rate": 4.855097887867519e-05,
      "loss": 0.9719,
      "step": 1997
    },
    {
      "epoch": 0.630880959898958,
      "grad_norm": 1.1132177114486694,
      "learning_rate": 4.8547470352957694e-05,
      "loss": 0.9235,
      "step": 1998
    },
    {
      "epoch": 0.6311967161351437,
      "grad_norm": 0.9798167943954468,
      "learning_rate": 4.8543961827240194e-05,
      "loss": 0.9303,
      "step": 1999
    },
    {
      "epoch": 0.6315124723713293,
      "grad_norm": 1.0205432176589966,
      "learning_rate": 4.85404533015227e-05,
      "loss": 0.9156,
      "step": 2000
    },
    {
      "epoch": 0.6315124723713293,
      "eval_loss": 0.8948850035667419,
      "eval_runtime": 4.5998,
      "eval_samples_per_second": 541.326,
      "eval_steps_per_second": 4.565,
      "step": 2000
    },
    {
      "epoch": 0.631828228607515,
      "grad_norm": 0.8572655916213989,
      "learning_rate": 4.853694477580521e-05,
      "loss": 0.9523,
      "step": 2001
    },
    {
      "epoch": 0.6321439848437007,
      "grad_norm": 1.1658207178115845,
      "learning_rate": 4.8533436250087715e-05,
      "loss": 0.885,
      "step": 2002
    },
    {
      "epoch": 0.6324597410798863,
      "grad_norm": 0.9058180451393127,
      "learning_rate": 4.852992772437022e-05,
      "loss": 0.9172,
      "step": 2003
    },
    {
      "epoch": 0.632775497316072,
      "grad_norm": 1.004248023033142,
      "learning_rate": 4.852641919865273e-05,
      "loss": 0.9582,
      "step": 2004
    },
    {
      "epoch": 0.6330912535522577,
      "grad_norm": 1.2487596273422241,
      "learning_rate": 4.8522910672935235e-05,
      "loss": 0.9426,
      "step": 2005
    },
    {
      "epoch": 0.6334070097884433,
      "grad_norm": 1.0299949645996094,
      "learning_rate": 4.851940214721774e-05,
      "loss": 0.9855,
      "step": 2006
    },
    {
      "epoch": 0.633722766024629,
      "grad_norm": 1.2060779333114624,
      "learning_rate": 4.851589362150025e-05,
      "loss": 0.9274,
      "step": 2007
    },
    {
      "epoch": 0.6340385222608147,
      "grad_norm": 0.9932058453559875,
      "learning_rate": 4.8512385095782756e-05,
      "loss": 0.9244,
      "step": 2008
    },
    {
      "epoch": 0.6343542784970003,
      "grad_norm": 0.9059169292449951,
      "learning_rate": 4.850887657006526e-05,
      "loss": 0.9784,
      "step": 2009
    },
    {
      "epoch": 0.634670034733186,
      "grad_norm": 1.0070979595184326,
      "learning_rate": 4.850536804434776e-05,
      "loss": 0.9523,
      "step": 2010
    },
    {
      "epoch": 0.6349857909693717,
      "grad_norm": 0.7713945508003235,
      "learning_rate": 4.850185951863027e-05,
      "loss": 0.8828,
      "step": 2011
    },
    {
      "epoch": 0.6353015472055573,
      "grad_norm": 0.9798288941383362,
      "learning_rate": 4.8498350992912776e-05,
      "loss": 0.9476,
      "step": 2012
    },
    {
      "epoch": 0.6356173034417429,
      "grad_norm": 0.9157301783561707,
      "learning_rate": 4.849484246719529e-05,
      "loss": 0.9465,
      "step": 2013
    },
    {
      "epoch": 0.6359330596779287,
      "grad_norm": 1.0438892841339111,
      "learning_rate": 4.8491333941477796e-05,
      "loss": 0.9075,
      "step": 2014
    },
    {
      "epoch": 0.6362488159141143,
      "grad_norm": 0.9387108683586121,
      "learning_rate": 4.84878254157603e-05,
      "loss": 0.9114,
      "step": 2015
    },
    {
      "epoch": 0.6365645721502999,
      "grad_norm": 0.9482589960098267,
      "learning_rate": 4.848431689004281e-05,
      "loss": 0.9565,
      "step": 2016
    },
    {
      "epoch": 0.6368803283864857,
      "grad_norm": 0.96918785572052,
      "learning_rate": 4.848080836432532e-05,
      "loss": 0.918,
      "step": 2017
    },
    {
      "epoch": 0.6371960846226713,
      "grad_norm": 0.9130452275276184,
      "learning_rate": 4.847729983860782e-05,
      "loss": 0.9684,
      "step": 2018
    },
    {
      "epoch": 0.6375118408588569,
      "grad_norm": 0.804583728313446,
      "learning_rate": 4.8473791312890324e-05,
      "loss": 0.9399,
      "step": 2019
    },
    {
      "epoch": 0.6378275970950427,
      "grad_norm": 0.7872358560562134,
      "learning_rate": 4.847028278717283e-05,
      "loss": 0.9331,
      "step": 2020
    },
    {
      "epoch": 0.6381433533312283,
      "grad_norm": 0.8944685459136963,
      "learning_rate": 4.846677426145534e-05,
      "loss": 0.9168,
      "step": 2021
    },
    {
      "epoch": 0.6384591095674139,
      "grad_norm": 0.8200992345809937,
      "learning_rate": 4.8463265735737844e-05,
      "loss": 0.9069,
      "step": 2022
    },
    {
      "epoch": 0.6387748658035997,
      "grad_norm": 0.8044791221618652,
      "learning_rate": 4.845975721002035e-05,
      "loss": 0.9598,
      "step": 2023
    },
    {
      "epoch": 0.6390906220397853,
      "grad_norm": 0.7977368235588074,
      "learning_rate": 4.845624868430286e-05,
      "loss": 0.9855,
      "step": 2024
    },
    {
      "epoch": 0.6394063782759709,
      "grad_norm": 0.8212038278579712,
      "learning_rate": 4.8452740158585365e-05,
      "loss": 0.9469,
      "step": 2025
    },
    {
      "epoch": 0.6397221345121566,
      "grad_norm": 1.003915786743164,
      "learning_rate": 4.844923163286787e-05,
      "loss": 0.9546,
      "step": 2026
    },
    {
      "epoch": 0.6400378907483423,
      "grad_norm": 0.814720630645752,
      "learning_rate": 4.844572310715038e-05,
      "loss": 0.9003,
      "step": 2027
    },
    {
      "epoch": 0.6403536469845279,
      "grad_norm": 0.8986808657646179,
      "learning_rate": 4.8442214581432885e-05,
      "loss": 0.9073,
      "step": 2028
    },
    {
      "epoch": 0.6406694032207136,
      "grad_norm": 1.1384204626083374,
      "learning_rate": 4.8438706055715385e-05,
      "loss": 0.9236,
      "step": 2029
    },
    {
      "epoch": 0.6409851594568993,
      "grad_norm": 0.8720213174819946,
      "learning_rate": 4.84351975299979e-05,
      "loss": 0.8753,
      "step": 2030
    },
    {
      "epoch": 0.6413009156930849,
      "grad_norm": 0.9339686036109924,
      "learning_rate": 4.8431689004280406e-05,
      "loss": 0.912,
      "step": 2031
    },
    {
      "epoch": 0.6416166719292706,
      "grad_norm": 0.9838172197341919,
      "learning_rate": 4.842818047856291e-05,
      "loss": 0.9735,
      "step": 2032
    },
    {
      "epoch": 0.6419324281654563,
      "grad_norm": 0.8113734722137451,
      "learning_rate": 4.842467195284542e-05,
      "loss": 0.9051,
      "step": 2033
    },
    {
      "epoch": 0.6422481844016419,
      "grad_norm": 1.0606969594955444,
      "learning_rate": 4.8421163427127926e-05,
      "loss": 0.8731,
      "step": 2034
    },
    {
      "epoch": 0.6425639406378276,
      "grad_norm": 1.1281532049179077,
      "learning_rate": 4.841765490141043e-05,
      "loss": 0.9743,
      "step": 2035
    },
    {
      "epoch": 0.6428796968740133,
      "grad_norm": 0.8947644233703613,
      "learning_rate": 4.841414637569294e-05,
      "loss": 0.9772,
      "step": 2036
    },
    {
      "epoch": 0.6431954531101989,
      "grad_norm": 1.0771374702453613,
      "learning_rate": 4.8410637849975446e-05,
      "loss": 0.9409,
      "step": 2037
    },
    {
      "epoch": 0.6435112093463846,
      "grad_norm": 0.9653573632240295,
      "learning_rate": 4.8407129324257946e-05,
      "loss": 0.9515,
      "step": 2038
    },
    {
      "epoch": 0.6438269655825702,
      "grad_norm": 1.016235589981079,
      "learning_rate": 4.840362079854045e-05,
      "loss": 0.9056,
      "step": 2039
    },
    {
      "epoch": 0.6441427218187559,
      "grad_norm": 0.9881219267845154,
      "learning_rate": 4.840011227282296e-05,
      "loss": 0.9309,
      "step": 2040
    },
    {
      "epoch": 0.6444584780549416,
      "grad_norm": 0.856764018535614,
      "learning_rate": 4.839660374710547e-05,
      "loss": 0.9589,
      "step": 2041
    },
    {
      "epoch": 0.6447742342911272,
      "grad_norm": 0.992955207824707,
      "learning_rate": 4.8393095221387974e-05,
      "loss": 0.934,
      "step": 2042
    },
    {
      "epoch": 0.6450899905273129,
      "grad_norm": 0.9156681299209595,
      "learning_rate": 4.838958669567048e-05,
      "loss": 0.9069,
      "step": 2043
    },
    {
      "epoch": 0.6454057467634986,
      "grad_norm": 0.9109468460083008,
      "learning_rate": 4.838607816995299e-05,
      "loss": 0.9522,
      "step": 2044
    },
    {
      "epoch": 0.6457215029996842,
      "grad_norm": 1.2245497703552246,
      "learning_rate": 4.8382569644235494e-05,
      "loss": 0.9848,
      "step": 2045
    },
    {
      "epoch": 0.6460372592358699,
      "grad_norm": 0.954457700252533,
      "learning_rate": 4.8379061118518e-05,
      "loss": 0.9388,
      "step": 2046
    },
    {
      "epoch": 0.6463530154720556,
      "grad_norm": 0.9210532307624817,
      "learning_rate": 4.837555259280051e-05,
      "loss": 0.943,
      "step": 2047
    },
    {
      "epoch": 0.6466687717082412,
      "grad_norm": 0.7135434150695801,
      "learning_rate": 4.8372044067083015e-05,
      "loss": 0.9761,
      "step": 2048
    },
    {
      "epoch": 0.646984527944427,
      "grad_norm": 0.830604612827301,
      "learning_rate": 4.836853554136552e-05,
      "loss": 0.8817,
      "step": 2049
    },
    {
      "epoch": 0.6473002841806126,
      "grad_norm": 0.9199753999710083,
      "learning_rate": 4.836502701564803e-05,
      "loss": 0.9535,
      "step": 2050
    },
    {
      "epoch": 0.6476160404167982,
      "grad_norm": 0.8631553649902344,
      "learning_rate": 4.8361518489930535e-05,
      "loss": 0.9168,
      "step": 2051
    },
    {
      "epoch": 0.647931796652984,
      "grad_norm": 0.8776834607124329,
      "learning_rate": 4.835800996421304e-05,
      "loss": 0.9132,
      "step": 2052
    },
    {
      "epoch": 0.6482475528891696,
      "grad_norm": 0.972796618938446,
      "learning_rate": 4.835450143849555e-05,
      "loss": 0.9619,
      "step": 2053
    },
    {
      "epoch": 0.6485633091253552,
      "grad_norm": 0.9555532336235046,
      "learning_rate": 4.8350992912778055e-05,
      "loss": 0.9782,
      "step": 2054
    },
    {
      "epoch": 0.6488790653615409,
      "grad_norm": 0.909334659576416,
      "learning_rate": 4.834748438706056e-05,
      "loss": 0.927,
      "step": 2055
    },
    {
      "epoch": 0.6491948215977266,
      "grad_norm": 0.9177472591400146,
      "learning_rate": 4.834397586134307e-05,
      "loss": 0.9387,
      "step": 2056
    },
    {
      "epoch": 0.6495105778339122,
      "grad_norm": 1.0882467031478882,
      "learning_rate": 4.834046733562557e-05,
      "loss": 0.9687,
      "step": 2057
    },
    {
      "epoch": 0.6498263340700979,
      "grad_norm": 0.9440739154815674,
      "learning_rate": 4.8336958809908076e-05,
      "loss": 0.8949,
      "step": 2058
    },
    {
      "epoch": 0.6501420903062836,
      "grad_norm": 0.8621885776519775,
      "learning_rate": 4.833345028419058e-05,
      "loss": 0.9807,
      "step": 2059
    },
    {
      "epoch": 0.6504578465424692,
      "grad_norm": 0.8570411801338196,
      "learning_rate": 4.832994175847309e-05,
      "loss": 0.9149,
      "step": 2060
    },
    {
      "epoch": 0.6507736027786549,
      "grad_norm": 0.9416449069976807,
      "learning_rate": 4.8326433232755596e-05,
      "loss": 0.962,
      "step": 2061
    },
    {
      "epoch": 0.6510893590148406,
      "grad_norm": 0.9035398364067078,
      "learning_rate": 4.83229247070381e-05,
      "loss": 0.9467,
      "step": 2062
    },
    {
      "epoch": 0.6514051152510262,
      "grad_norm": 0.8432939648628235,
      "learning_rate": 4.831941618132061e-05,
      "loss": 0.936,
      "step": 2063
    },
    {
      "epoch": 0.6517208714872119,
      "grad_norm": 0.9859336018562317,
      "learning_rate": 4.831590765560312e-05,
      "loss": 0.9439,
      "step": 2064
    },
    {
      "epoch": 0.6520366277233975,
      "grad_norm": 0.8783730268478394,
      "learning_rate": 4.831239912988563e-05,
      "loss": 0.8973,
      "step": 2065
    },
    {
      "epoch": 0.6523523839595832,
      "grad_norm": 0.9620686173439026,
      "learning_rate": 4.830889060416813e-05,
      "loss": 0.9517,
      "step": 2066
    },
    {
      "epoch": 0.6526681401957689,
      "grad_norm": 0.9241499900817871,
      "learning_rate": 4.830538207845064e-05,
      "loss": 0.9293,
      "step": 2067
    },
    {
      "epoch": 0.6529838964319545,
      "grad_norm": 0.8855466246604919,
      "learning_rate": 4.8301873552733144e-05,
      "loss": 0.8891,
      "step": 2068
    },
    {
      "epoch": 0.6532996526681402,
      "grad_norm": 0.8617545962333679,
      "learning_rate": 4.829836502701565e-05,
      "loss": 0.9012,
      "step": 2069
    },
    {
      "epoch": 0.6536154089043259,
      "grad_norm": 1.002083420753479,
      "learning_rate": 4.829485650129816e-05,
      "loss": 0.9735,
      "step": 2070
    },
    {
      "epoch": 0.6539311651405115,
      "grad_norm": 0.9080199599266052,
      "learning_rate": 4.8291347975580664e-05,
      "loss": 0.9295,
      "step": 2071
    },
    {
      "epoch": 0.6542469213766972,
      "grad_norm": 0.9083384275436401,
      "learning_rate": 4.828783944986317e-05,
      "loss": 0.9504,
      "step": 2072
    },
    {
      "epoch": 0.6545626776128829,
      "grad_norm": 0.8225836157798767,
      "learning_rate": 4.828433092414568e-05,
      "loss": 0.8865,
      "step": 2073
    },
    {
      "epoch": 0.6548784338490685,
      "grad_norm": 0.8008926510810852,
      "learning_rate": 4.8280822398428185e-05,
      "loss": 0.9291,
      "step": 2074
    },
    {
      "epoch": 0.6551941900852541,
      "grad_norm": 0.9519299864768982,
      "learning_rate": 4.827731387271069e-05,
      "loss": 0.9494,
      "step": 2075
    },
    {
      "epoch": 0.6555099463214399,
      "grad_norm": 0.8305957317352295,
      "learning_rate": 4.827380534699319e-05,
      "loss": 0.8753,
      "step": 2076
    },
    {
      "epoch": 0.6558257025576255,
      "grad_norm": 0.8491001725196838,
      "learning_rate": 4.82702968212757e-05,
      "loss": 0.946,
      "step": 2077
    },
    {
      "epoch": 0.6561414587938111,
      "grad_norm": 0.8690711855888367,
      "learning_rate": 4.8266788295558205e-05,
      "loss": 0.9186,
      "step": 2078
    },
    {
      "epoch": 0.6564572150299969,
      "grad_norm": 1.0577651262283325,
      "learning_rate": 4.826327976984071e-05,
      "loss": 0.979,
      "step": 2079
    },
    {
      "epoch": 0.6567729712661825,
      "grad_norm": 1.0983853340148926,
      "learning_rate": 4.825977124412322e-05,
      "loss": 0.9453,
      "step": 2080
    },
    {
      "epoch": 0.6570887275023681,
      "grad_norm": 0.8016387820243835,
      "learning_rate": 4.8256262718405726e-05,
      "loss": 0.8997,
      "step": 2081
    },
    {
      "epoch": 0.6574044837385539,
      "grad_norm": 0.8716789484024048,
      "learning_rate": 4.825275419268823e-05,
      "loss": 0.9375,
      "step": 2082
    },
    {
      "epoch": 0.6577202399747395,
      "grad_norm": 0.8880289196968079,
      "learning_rate": 4.8249245666970746e-05,
      "loss": 0.8788,
      "step": 2083
    },
    {
      "epoch": 0.6580359962109251,
      "grad_norm": 1.1746392250061035,
      "learning_rate": 4.824573714125325e-05,
      "loss": 0.8865,
      "step": 2084
    },
    {
      "epoch": 0.6583517524471109,
      "grad_norm": 1.6228083372116089,
      "learning_rate": 4.824222861553575e-05,
      "loss": 0.9222,
      "step": 2085
    },
    {
      "epoch": 0.6586675086832965,
      "grad_norm": 0.9004389047622681,
      "learning_rate": 4.823872008981826e-05,
      "loss": 0.9362,
      "step": 2086
    },
    {
      "epoch": 0.6589832649194821,
      "grad_norm": 0.8536555767059326,
      "learning_rate": 4.823521156410077e-05,
      "loss": 0.9402,
      "step": 2087
    },
    {
      "epoch": 0.6592990211556679,
      "grad_norm": 1.028031349182129,
      "learning_rate": 4.8231703038383274e-05,
      "loss": 0.9315,
      "step": 2088
    },
    {
      "epoch": 0.6596147773918535,
      "grad_norm": 0.9625617265701294,
      "learning_rate": 4.822819451266578e-05,
      "loss": 0.9837,
      "step": 2089
    },
    {
      "epoch": 0.6599305336280391,
      "grad_norm": 1.0757441520690918,
      "learning_rate": 4.822468598694829e-05,
      "loss": 0.9163,
      "step": 2090
    },
    {
      "epoch": 0.6602462898642248,
      "grad_norm": 0.9716647267341614,
      "learning_rate": 4.8221177461230794e-05,
      "loss": 0.9201,
      "step": 2091
    },
    {
      "epoch": 0.6605620461004105,
      "grad_norm": 0.9223290085792542,
      "learning_rate": 4.82176689355133e-05,
      "loss": 0.9754,
      "step": 2092
    },
    {
      "epoch": 0.6608778023365961,
      "grad_norm": 0.8265789747238159,
      "learning_rate": 4.821416040979581e-05,
      "loss": 0.9705,
      "step": 2093
    },
    {
      "epoch": 0.6611935585727818,
      "grad_norm": 1.0431139469146729,
      "learning_rate": 4.8210651884078314e-05,
      "loss": 0.9601,
      "step": 2094
    },
    {
      "epoch": 0.6615093148089675,
      "grad_norm": 1.1407561302185059,
      "learning_rate": 4.820714335836082e-05,
      "loss": 0.9205,
      "step": 2095
    },
    {
      "epoch": 0.6618250710451531,
      "grad_norm": 1.058385968208313,
      "learning_rate": 4.820363483264332e-05,
      "loss": 0.9556,
      "step": 2096
    },
    {
      "epoch": 0.6621408272813388,
      "grad_norm": 1.1743234395980835,
      "learning_rate": 4.820012630692583e-05,
      "loss": 0.9746,
      "step": 2097
    },
    {
      "epoch": 0.6624565835175245,
      "grad_norm": 0.9253515005111694,
      "learning_rate": 4.8196617781208335e-05,
      "loss": 0.95,
      "step": 2098
    },
    {
      "epoch": 0.6627723397537101,
      "grad_norm": 1.1618530750274658,
      "learning_rate": 4.819310925549084e-05,
      "loss": 0.9323,
      "step": 2099
    },
    {
      "epoch": 0.6630880959898958,
      "grad_norm": 1.1319518089294434,
      "learning_rate": 4.8189600729773355e-05,
      "loss": 0.9252,
      "step": 2100
    },
    {
      "epoch": 0.6634038522260814,
      "grad_norm": 1.0473134517669678,
      "learning_rate": 4.818609220405586e-05,
      "loss": 0.9443,
      "step": 2101
    },
    {
      "epoch": 0.6637196084622671,
      "grad_norm": 0.9043784737586975,
      "learning_rate": 4.818258367833837e-05,
      "loss": 0.8639,
      "step": 2102
    },
    {
      "epoch": 0.6640353646984528,
      "grad_norm": 0.9491549730300903,
      "learning_rate": 4.8179075152620876e-05,
      "loss": 0.8967,
      "step": 2103
    },
    {
      "epoch": 0.6643511209346384,
      "grad_norm": 0.8547908067703247,
      "learning_rate": 4.8175566626903376e-05,
      "loss": 0.906,
      "step": 2104
    },
    {
      "epoch": 0.6646668771708242,
      "grad_norm": 1.0865470170974731,
      "learning_rate": 4.817205810118588e-05,
      "loss": 0.9023,
      "step": 2105
    },
    {
      "epoch": 0.6649826334070098,
      "grad_norm": 0.9374250173568726,
      "learning_rate": 4.816854957546839e-05,
      "loss": 0.9195,
      "step": 2106
    },
    {
      "epoch": 0.6652983896431954,
      "grad_norm": 0.9054169058799744,
      "learning_rate": 4.8165041049750896e-05,
      "loss": 0.8924,
      "step": 2107
    },
    {
      "epoch": 0.6656141458793812,
      "grad_norm": 1.173295497894287,
      "learning_rate": 4.81615325240334e-05,
      "loss": 0.9402,
      "step": 2108
    },
    {
      "epoch": 0.6659299021155668,
      "grad_norm": 1.2512705326080322,
      "learning_rate": 4.815802399831591e-05,
      "loss": 0.957,
      "step": 2109
    },
    {
      "epoch": 0.6662456583517524,
      "grad_norm": 0.8751164674758911,
      "learning_rate": 4.815451547259842e-05,
      "loss": 0.9398,
      "step": 2110
    },
    {
      "epoch": 0.6665614145879382,
      "grad_norm": 1.0756975412368774,
      "learning_rate": 4.8151006946880923e-05,
      "loss": 0.9449,
      "step": 2111
    },
    {
      "epoch": 0.6668771708241238,
      "grad_norm": 1.0081050395965576,
      "learning_rate": 4.814749842116343e-05,
      "loss": 0.9462,
      "step": 2112
    },
    {
      "epoch": 0.6671929270603094,
      "grad_norm": 1.1541353464126587,
      "learning_rate": 4.814398989544594e-05,
      "loss": 0.9289,
      "step": 2113
    },
    {
      "epoch": 0.6675086832964952,
      "grad_norm": 1.767268180847168,
      "learning_rate": 4.8140481369728444e-05,
      "loss": 0.9462,
      "step": 2114
    },
    {
      "epoch": 0.6678244395326808,
      "grad_norm": 0.9387238621711731,
      "learning_rate": 4.8136972844010944e-05,
      "loss": 0.9874,
      "step": 2115
    },
    {
      "epoch": 0.6681401957688664,
      "grad_norm": 1.111083984375,
      "learning_rate": 4.813346431829345e-05,
      "loss": 0.9345,
      "step": 2116
    },
    {
      "epoch": 0.6684559520050521,
      "grad_norm": 0.9203670024871826,
      "learning_rate": 4.812995579257596e-05,
      "loss": 0.9075,
      "step": 2117
    },
    {
      "epoch": 0.6687717082412378,
      "grad_norm": 1.2462701797485352,
      "learning_rate": 4.812644726685847e-05,
      "loss": 0.9162,
      "step": 2118
    },
    {
      "epoch": 0.6690874644774234,
      "grad_norm": 1.0394833087921143,
      "learning_rate": 4.812293874114098e-05,
      "loss": 0.9583,
      "step": 2119
    },
    {
      "epoch": 0.6694032207136091,
      "grad_norm": 1.0731309652328491,
      "learning_rate": 4.8119430215423485e-05,
      "loss": 0.9056,
      "step": 2120
    },
    {
      "epoch": 0.6697189769497948,
      "grad_norm": 0.9112296104431152,
      "learning_rate": 4.811592168970599e-05,
      "loss": 0.9288,
      "step": 2121
    },
    {
      "epoch": 0.6700347331859804,
      "grad_norm": 0.9546449184417725,
      "learning_rate": 4.81124131639885e-05,
      "loss": 0.9221,
      "step": 2122
    },
    {
      "epoch": 0.6703504894221661,
      "grad_norm": 1.1501476764678955,
      "learning_rate": 4.8108904638271005e-05,
      "loss": 0.9087,
      "step": 2123
    },
    {
      "epoch": 0.6706662456583518,
      "grad_norm": 1.0434279441833496,
      "learning_rate": 4.8105396112553505e-05,
      "loss": 0.9007,
      "step": 2124
    },
    {
      "epoch": 0.6709820018945374,
      "grad_norm": 0.7647051811218262,
      "learning_rate": 4.810188758683601e-05,
      "loss": 0.8963,
      "step": 2125
    },
    {
      "epoch": 0.6712977581307231,
      "grad_norm": 1.1238497495651245,
      "learning_rate": 4.809837906111852e-05,
      "loss": 0.9525,
      "step": 2126
    },
    {
      "epoch": 0.6716135143669087,
      "grad_norm": 1.0279412269592285,
      "learning_rate": 4.8094870535401026e-05,
      "loss": 0.8947,
      "step": 2127
    },
    {
      "epoch": 0.6719292706030944,
      "grad_norm": 1.6879898309707642,
      "learning_rate": 4.809136200968353e-05,
      "loss": 0.931,
      "step": 2128
    },
    {
      "epoch": 0.6722450268392801,
      "grad_norm": 1.1181367635726929,
      "learning_rate": 4.808785348396604e-05,
      "loss": 0.8731,
      "step": 2129
    },
    {
      "epoch": 0.6725607830754657,
      "grad_norm": 1.0225683450698853,
      "learning_rate": 4.8084344958248546e-05,
      "loss": 0.8758,
      "step": 2130
    },
    {
      "epoch": 0.6728765393116514,
      "grad_norm": 0.8678219318389893,
      "learning_rate": 4.808083643253105e-05,
      "loss": 0.9434,
      "step": 2131
    },
    {
      "epoch": 0.6731922955478371,
      "grad_norm": 1.2551231384277344,
      "learning_rate": 4.807732790681356e-05,
      "loss": 0.9997,
      "step": 2132
    },
    {
      "epoch": 0.6735080517840227,
      "grad_norm": 0.9308462738990784,
      "learning_rate": 4.8073819381096067e-05,
      "loss": 0.9044,
      "step": 2133
    },
    {
      "epoch": 0.6738238080202084,
      "grad_norm": 1.0971746444702148,
      "learning_rate": 4.8070310855378567e-05,
      "loss": 0.8921,
      "step": 2134
    },
    {
      "epoch": 0.6741395642563941,
      "grad_norm": 0.9452659487724304,
      "learning_rate": 4.806680232966108e-05,
      "loss": 1.0102,
      "step": 2135
    },
    {
      "epoch": 0.6744553204925797,
      "grad_norm": 0.8626505732536316,
      "learning_rate": 4.806329380394359e-05,
      "loss": 0.9253,
      "step": 2136
    },
    {
      "epoch": 0.6747710767287654,
      "grad_norm": 1.0569825172424316,
      "learning_rate": 4.8059785278226094e-05,
      "loss": 0.9753,
      "step": 2137
    },
    {
      "epoch": 0.6750868329649511,
      "grad_norm": 0.9538291692733765,
      "learning_rate": 4.80562767525086e-05,
      "loss": 0.9554,
      "step": 2138
    },
    {
      "epoch": 0.6754025892011367,
      "grad_norm": 1.1178488731384277,
      "learning_rate": 4.805276822679111e-05,
      "loss": 0.9868,
      "step": 2139
    },
    {
      "epoch": 0.6757183454373223,
      "grad_norm": 1.0342998504638672,
      "learning_rate": 4.8049259701073614e-05,
      "loss": 0.95,
      "step": 2140
    },
    {
      "epoch": 0.6760341016735081,
      "grad_norm": 0.8696918487548828,
      "learning_rate": 4.804575117535612e-05,
      "loss": 0.9857,
      "step": 2141
    },
    {
      "epoch": 0.6763498579096937,
      "grad_norm": 0.922849178314209,
      "learning_rate": 4.804224264963863e-05,
      "loss": 0.8933,
      "step": 2142
    },
    {
      "epoch": 0.6766656141458793,
      "grad_norm": 0.9558029174804688,
      "learning_rate": 4.803873412392113e-05,
      "loss": 0.9532,
      "step": 2143
    },
    {
      "epoch": 0.6769813703820651,
      "grad_norm": 0.8252430558204651,
      "learning_rate": 4.8035225598203635e-05,
      "loss": 0.9402,
      "step": 2144
    },
    {
      "epoch": 0.6772971266182507,
      "grad_norm": 0.7060983180999756,
      "learning_rate": 4.803171707248614e-05,
      "loss": 0.9178,
      "step": 2145
    },
    {
      "epoch": 0.6776128828544363,
      "grad_norm": 0.7782047390937805,
      "learning_rate": 4.802820854676865e-05,
      "loss": 0.9295,
      "step": 2146
    },
    {
      "epoch": 0.6779286390906221,
      "grad_norm": 0.950173556804657,
      "learning_rate": 4.8024700021051155e-05,
      "loss": 0.936,
      "step": 2147
    },
    {
      "epoch": 0.6782443953268077,
      "grad_norm": 0.8984846472740173,
      "learning_rate": 4.802119149533366e-05,
      "loss": 0.919,
      "step": 2148
    },
    {
      "epoch": 0.6785601515629933,
      "grad_norm": 1.0718544721603394,
      "learning_rate": 4.801768296961617e-05,
      "loss": 0.9553,
      "step": 2149
    },
    {
      "epoch": 0.6788759077991791,
      "grad_norm": 0.8679318428039551,
      "learning_rate": 4.8014174443898676e-05,
      "loss": 0.9236,
      "step": 2150
    },
    {
      "epoch": 0.6791916640353647,
      "grad_norm": 0.8235222697257996,
      "learning_rate": 4.801066591818118e-05,
      "loss": 0.8835,
      "step": 2151
    },
    {
      "epoch": 0.6795074202715503,
      "grad_norm": 0.7066424489021301,
      "learning_rate": 4.800715739246369e-05,
      "loss": 0.9069,
      "step": 2152
    },
    {
      "epoch": 0.679823176507736,
      "grad_norm": 0.8098611235618591,
      "learning_rate": 4.8003648866746196e-05,
      "loss": 0.9117,
      "step": 2153
    },
    {
      "epoch": 0.6801389327439217,
      "grad_norm": 0.9185462594032288,
      "learning_rate": 4.80001403410287e-05,
      "loss": 0.9931,
      "step": 2154
    },
    {
      "epoch": 0.6804546889801073,
      "grad_norm": 0.9494038820266724,
      "learning_rate": 4.799663181531121e-05,
      "loss": 0.9174,
      "step": 2155
    },
    {
      "epoch": 0.680770445216293,
      "grad_norm": 1.077917456626892,
      "learning_rate": 4.7993123289593716e-05,
      "loss": 0.9402,
      "step": 2156
    },
    {
      "epoch": 0.6810862014524787,
      "grad_norm": 1.5739734172821045,
      "learning_rate": 4.798961476387622e-05,
      "loss": 0.9196,
      "step": 2157
    },
    {
      "epoch": 0.6814019576886644,
      "grad_norm": 0.8043046593666077,
      "learning_rate": 4.798610623815873e-05,
      "loss": 0.9612,
      "step": 2158
    },
    {
      "epoch": 0.68171771392485,
      "grad_norm": 0.8549139499664307,
      "learning_rate": 4.798259771244124e-05,
      "loss": 0.9013,
      "step": 2159
    },
    {
      "epoch": 0.6820334701610357,
      "grad_norm": 1.001797080039978,
      "learning_rate": 4.7979089186723744e-05,
      "loss": 0.915,
      "step": 2160
    },
    {
      "epoch": 0.6823492263972214,
      "grad_norm": 0.9424559473991394,
      "learning_rate": 4.797558066100625e-05,
      "loss": 0.8626,
      "step": 2161
    },
    {
      "epoch": 0.682664982633407,
      "grad_norm": 0.9882215261459351,
      "learning_rate": 4.797207213528875e-05,
      "loss": 0.9215,
      "step": 2162
    },
    {
      "epoch": 0.6829807388695927,
      "grad_norm": 0.9761519432067871,
      "learning_rate": 4.796856360957126e-05,
      "loss": 0.9438,
      "step": 2163
    },
    {
      "epoch": 0.6832964951057784,
      "grad_norm": 0.9722285866737366,
      "learning_rate": 4.7965055083853764e-05,
      "loss": 0.9535,
      "step": 2164
    },
    {
      "epoch": 0.683612251341964,
      "grad_norm": 1.0215390920639038,
      "learning_rate": 4.796154655813627e-05,
      "loss": 0.9803,
      "step": 2165
    },
    {
      "epoch": 0.6839280075781496,
      "grad_norm": 1.0236492156982422,
      "learning_rate": 4.795803803241878e-05,
      "loss": 0.8841,
      "step": 2166
    },
    {
      "epoch": 0.6842437638143354,
      "grad_norm": 1.033446192741394,
      "learning_rate": 4.7954529506701285e-05,
      "loss": 0.9348,
      "step": 2167
    },
    {
      "epoch": 0.684559520050521,
      "grad_norm": 1.0206842422485352,
      "learning_rate": 4.795102098098379e-05,
      "loss": 0.8761,
      "step": 2168
    },
    {
      "epoch": 0.6848752762867066,
      "grad_norm": 0.9929082989692688,
      "learning_rate": 4.79475124552663e-05,
      "loss": 0.9261,
      "step": 2169
    },
    {
      "epoch": 0.6851910325228924,
      "grad_norm": 0.8079281449317932,
      "learning_rate": 4.794400392954881e-05,
      "loss": 0.9188,
      "step": 2170
    },
    {
      "epoch": 0.685506788759078,
      "grad_norm": 1.0712321996688843,
      "learning_rate": 4.794049540383131e-05,
      "loss": 0.8793,
      "step": 2171
    },
    {
      "epoch": 0.6858225449952636,
      "grad_norm": 0.9059593677520752,
      "learning_rate": 4.793698687811382e-05,
      "loss": 0.9788,
      "step": 2172
    },
    {
      "epoch": 0.6861383012314494,
      "grad_norm": 1.0053893327713013,
      "learning_rate": 4.7933478352396326e-05,
      "loss": 0.9443,
      "step": 2173
    },
    {
      "epoch": 0.686454057467635,
      "grad_norm": 0.8830248713493347,
      "learning_rate": 4.792996982667883e-05,
      "loss": 0.9455,
      "step": 2174
    },
    {
      "epoch": 0.6867698137038206,
      "grad_norm": 1.052634596824646,
      "learning_rate": 4.792646130096134e-05,
      "loss": 0.9424,
      "step": 2175
    },
    {
      "epoch": 0.6870855699400064,
      "grad_norm": 0.9581133723258972,
      "learning_rate": 4.7922952775243846e-05,
      "loss": 0.9509,
      "step": 2176
    },
    {
      "epoch": 0.687401326176192,
      "grad_norm": 1.0139973163604736,
      "learning_rate": 4.791944424952635e-05,
      "loss": 0.9004,
      "step": 2177
    },
    {
      "epoch": 0.6877170824123776,
      "grad_norm": 0.8967349529266357,
      "learning_rate": 4.791593572380886e-05,
      "loss": 0.8668,
      "step": 2178
    },
    {
      "epoch": 0.6880328386485633,
      "grad_norm": 0.8164037466049194,
      "learning_rate": 4.7912427198091366e-05,
      "loss": 0.9926,
      "step": 2179
    },
    {
      "epoch": 0.688348594884749,
      "grad_norm": 0.852288007736206,
      "learning_rate": 4.790891867237387e-05,
      "loss": 0.9243,
      "step": 2180
    },
    {
      "epoch": 0.6886643511209346,
      "grad_norm": 1.0159907341003418,
      "learning_rate": 4.790541014665637e-05,
      "loss": 0.9113,
      "step": 2181
    },
    {
      "epoch": 0.6889801073571203,
      "grad_norm": 1.0126631259918213,
      "learning_rate": 4.790190162093888e-05,
      "loss": 0.9658,
      "step": 2182
    },
    {
      "epoch": 0.689295863593306,
      "grad_norm": 1.0393211841583252,
      "learning_rate": 4.789839309522139e-05,
      "loss": 0.9069,
      "step": 2183
    },
    {
      "epoch": 0.6896116198294916,
      "grad_norm": 1.1215628385543823,
      "learning_rate": 4.7894884569503894e-05,
      "loss": 0.9109,
      "step": 2184
    },
    {
      "epoch": 0.6899273760656773,
      "grad_norm": 1.273755431175232,
      "learning_rate": 4.78913760437864e-05,
      "loss": 0.918,
      "step": 2185
    },
    {
      "epoch": 0.690243132301863,
      "grad_norm": 1.0634371042251587,
      "learning_rate": 4.788786751806891e-05,
      "loss": 0.8589,
      "step": 2186
    },
    {
      "epoch": 0.6905588885380486,
      "grad_norm": 1.0034253597259521,
      "learning_rate": 4.7884358992351414e-05,
      "loss": 0.8912,
      "step": 2187
    },
    {
      "epoch": 0.6908746447742343,
      "grad_norm": 1.1176636219024658,
      "learning_rate": 4.788085046663393e-05,
      "loss": 0.9343,
      "step": 2188
    },
    {
      "epoch": 0.69119040101042,
      "grad_norm": 0.88018798828125,
      "learning_rate": 4.7877341940916435e-05,
      "loss": 0.9288,
      "step": 2189
    },
    {
      "epoch": 0.6915061572466056,
      "grad_norm": 0.9545112252235413,
      "learning_rate": 4.7873833415198935e-05,
      "loss": 0.935,
      "step": 2190
    },
    {
      "epoch": 0.6918219134827913,
      "grad_norm": 1.1200776100158691,
      "learning_rate": 4.787032488948144e-05,
      "loss": 0.9572,
      "step": 2191
    },
    {
      "epoch": 0.6921376697189769,
      "grad_norm": 0.9607634544372559,
      "learning_rate": 4.786681636376395e-05,
      "loss": 0.9178,
      "step": 2192
    },
    {
      "epoch": 0.6924534259551626,
      "grad_norm": 0.9006031155586243,
      "learning_rate": 4.7863307838046455e-05,
      "loss": 0.9134,
      "step": 2193
    },
    {
      "epoch": 0.6927691821913483,
      "grad_norm": 1.1597894430160522,
      "learning_rate": 4.785979931232896e-05,
      "loss": 0.9493,
      "step": 2194
    },
    {
      "epoch": 0.6930849384275339,
      "grad_norm": 0.9509448409080505,
      "learning_rate": 4.785629078661147e-05,
      "loss": 0.8785,
      "step": 2195
    },
    {
      "epoch": 0.6934006946637196,
      "grad_norm": 1.1580432653427124,
      "learning_rate": 4.7852782260893975e-05,
      "loss": 0.9393,
      "step": 2196
    },
    {
      "epoch": 0.6937164508999053,
      "grad_norm": 0.8308850526809692,
      "learning_rate": 4.784927373517648e-05,
      "loss": 0.8819,
      "step": 2197
    },
    {
      "epoch": 0.6940322071360909,
      "grad_norm": 1.0193897485733032,
      "learning_rate": 4.784576520945899e-05,
      "loss": 0.8991,
      "step": 2198
    },
    {
      "epoch": 0.6943479633722766,
      "grad_norm": 0.9447402358055115,
      "learning_rate": 4.7842256683741496e-05,
      "loss": 0.8901,
      "step": 2199
    },
    {
      "epoch": 0.6946637196084623,
      "grad_norm": 1.1092089414596558,
      "learning_rate": 4.7838748158024e-05,
      "loss": 0.9394,
      "step": 2200
    },
    {
      "epoch": 0.6946637196084623,
      "eval_loss": 0.8896675705909729,
      "eval_runtime": 4.5982,
      "eval_samples_per_second": 541.518,
      "eval_steps_per_second": 4.567,
      "step": 2200
    },
    {
      "epoch": 0.6949794758446479,
      "grad_norm": 1.14857816696167,
      "learning_rate": 4.78352396323065e-05,
      "loss": 0.971,
      "step": 2201
    },
    {
      "epoch": 0.6952952320808335,
      "grad_norm": 1.160018801689148,
      "learning_rate": 4.783173110658901e-05,
      "loss": 0.9159,
      "step": 2202
    },
    {
      "epoch": 0.6956109883170193,
      "grad_norm": 0.9482296109199524,
      "learning_rate": 4.7828222580871516e-05,
      "loss": 0.9007,
      "step": 2203
    },
    {
      "epoch": 0.6959267445532049,
      "grad_norm": 0.9739649295806885,
      "learning_rate": 4.782471405515402e-05,
      "loss": 0.9147,
      "step": 2204
    },
    {
      "epoch": 0.6962425007893905,
      "grad_norm": 0.9409672021865845,
      "learning_rate": 4.782120552943654e-05,
      "loss": 0.9239,
      "step": 2205
    },
    {
      "epoch": 0.6965582570255763,
      "grad_norm": 1.0474107265472412,
      "learning_rate": 4.7817697003719044e-05,
      "loss": 0.9077,
      "step": 2206
    },
    {
      "epoch": 0.6968740132617619,
      "grad_norm": 1.0161563158035278,
      "learning_rate": 4.781418847800155e-05,
      "loss": 0.8982,
      "step": 2207
    },
    {
      "epoch": 0.6971897694979475,
      "grad_norm": 0.9729804992675781,
      "learning_rate": 4.781067995228406e-05,
      "loss": 0.9348,
      "step": 2208
    },
    {
      "epoch": 0.6975055257341333,
      "grad_norm": 0.990343451499939,
      "learning_rate": 4.780717142656656e-05,
      "loss": 0.9019,
      "step": 2209
    },
    {
      "epoch": 0.6978212819703189,
      "grad_norm": 1.0252972841262817,
      "learning_rate": 4.7803662900849064e-05,
      "loss": 0.9464,
      "step": 2210
    },
    {
      "epoch": 0.6981370382065045,
      "grad_norm": 1.0026285648345947,
      "learning_rate": 4.780015437513157e-05,
      "loss": 0.94,
      "step": 2211
    },
    {
      "epoch": 0.6984527944426903,
      "grad_norm": 0.9541838765144348,
      "learning_rate": 4.779664584941408e-05,
      "loss": 0.9336,
      "step": 2212
    },
    {
      "epoch": 0.6987685506788759,
      "grad_norm": 1.1420776844024658,
      "learning_rate": 4.7793137323696584e-05,
      "loss": 0.9514,
      "step": 2213
    },
    {
      "epoch": 0.6990843069150616,
      "grad_norm": 1.143541932106018,
      "learning_rate": 4.778962879797909e-05,
      "loss": 0.9338,
      "step": 2214
    },
    {
      "epoch": 0.6994000631512473,
      "grad_norm": 1.0648571252822876,
      "learning_rate": 4.77861202722616e-05,
      "loss": 0.9262,
      "step": 2215
    },
    {
      "epoch": 0.6997158193874329,
      "grad_norm": 1.278847098350525,
      "learning_rate": 4.7782611746544105e-05,
      "loss": 0.9773,
      "step": 2216
    },
    {
      "epoch": 0.7000315756236186,
      "grad_norm": 0.9674970507621765,
      "learning_rate": 4.777910322082661e-05,
      "loss": 0.9471,
      "step": 2217
    },
    {
      "epoch": 0.7003473318598042,
      "grad_norm": 0.8460709452629089,
      "learning_rate": 4.777559469510912e-05,
      "loss": 0.9141,
      "step": 2218
    },
    {
      "epoch": 0.7006630880959899,
      "grad_norm": 1.1306320428848267,
      "learning_rate": 4.7772086169391625e-05,
      "loss": 0.9471,
      "step": 2219
    },
    {
      "epoch": 0.7009788443321756,
      "grad_norm": 0.9396591186523438,
      "learning_rate": 4.7768577643674125e-05,
      "loss": 0.9006,
      "step": 2220
    },
    {
      "epoch": 0.7012946005683612,
      "grad_norm": 0.9973105788230896,
      "learning_rate": 4.776506911795663e-05,
      "loss": 0.9122,
      "step": 2221
    },
    {
      "epoch": 0.7016103568045469,
      "grad_norm": 0.8573862910270691,
      "learning_rate": 4.776156059223914e-05,
      "loss": 0.9177,
      "step": 2222
    },
    {
      "epoch": 0.7019261130407326,
      "grad_norm": 0.9438825249671936,
      "learning_rate": 4.775805206652165e-05,
      "loss": 0.9029,
      "step": 2223
    },
    {
      "epoch": 0.7022418692769182,
      "grad_norm": 0.9946194887161255,
      "learning_rate": 4.775454354080416e-05,
      "loss": 0.9518,
      "step": 2224
    },
    {
      "epoch": 0.7025576255131039,
      "grad_norm": 0.7770864367485046,
      "learning_rate": 4.7751035015086666e-05,
      "loss": 0.902,
      "step": 2225
    },
    {
      "epoch": 0.7028733817492896,
      "grad_norm": 0.8693189024925232,
      "learning_rate": 4.774752648936917e-05,
      "loss": 0.9616,
      "step": 2226
    },
    {
      "epoch": 0.7031891379854752,
      "grad_norm": 0.8609262704849243,
      "learning_rate": 4.774401796365168e-05,
      "loss": 0.8632,
      "step": 2227
    },
    {
      "epoch": 0.7035048942216608,
      "grad_norm": 0.8197088837623596,
      "learning_rate": 4.774050943793419e-05,
      "loss": 0.9377,
      "step": 2228
    },
    {
      "epoch": 0.7038206504578466,
      "grad_norm": 0.7312625646591187,
      "learning_rate": 4.773700091221669e-05,
      "loss": 0.9858,
      "step": 2229
    },
    {
      "epoch": 0.7041364066940322,
      "grad_norm": 1.1883238554000854,
      "learning_rate": 4.7733492386499194e-05,
      "loss": 0.9163,
      "step": 2230
    },
    {
      "epoch": 0.7044521629302178,
      "grad_norm": 0.8565148711204529,
      "learning_rate": 4.77299838607817e-05,
      "loss": 0.9311,
      "step": 2231
    },
    {
      "epoch": 0.7047679191664036,
      "grad_norm": 0.9193990230560303,
      "learning_rate": 4.772647533506421e-05,
      "loss": 0.8944,
      "step": 2232
    },
    {
      "epoch": 0.7050836754025892,
      "grad_norm": 0.9469957947731018,
      "learning_rate": 4.7722966809346714e-05,
      "loss": 0.937,
      "step": 2233
    },
    {
      "epoch": 0.7053994316387748,
      "grad_norm": 0.838573157787323,
      "learning_rate": 4.771945828362922e-05,
      "loss": 0.8696,
      "step": 2234
    },
    {
      "epoch": 0.7057151878749606,
      "grad_norm": 0.8632515072822571,
      "learning_rate": 4.771594975791173e-05,
      "loss": 0.9047,
      "step": 2235
    },
    {
      "epoch": 0.7060309441111462,
      "grad_norm": 0.8928689360618591,
      "learning_rate": 4.7712441232194234e-05,
      "loss": 0.9443,
      "step": 2236
    },
    {
      "epoch": 0.7063467003473318,
      "grad_norm": 0.8967702984809875,
      "learning_rate": 4.770893270647674e-05,
      "loss": 0.8731,
      "step": 2237
    },
    {
      "epoch": 0.7066624565835176,
      "grad_norm": 0.9720224738121033,
      "learning_rate": 4.770542418075925e-05,
      "loss": 0.9072,
      "step": 2238
    },
    {
      "epoch": 0.7069782128197032,
      "grad_norm": 1.0717461109161377,
      "learning_rate": 4.770191565504175e-05,
      "loss": 0.9032,
      "step": 2239
    },
    {
      "epoch": 0.7072939690558888,
      "grad_norm": 0.8807306885719299,
      "learning_rate": 4.769840712932426e-05,
      "loss": 0.9449,
      "step": 2240
    },
    {
      "epoch": 0.7076097252920746,
      "grad_norm": 0.9873418211936951,
      "learning_rate": 4.769489860360677e-05,
      "loss": 0.9292,
      "step": 2241
    },
    {
      "epoch": 0.7079254815282602,
      "grad_norm": 0.9691063165664673,
      "learning_rate": 4.7691390077889275e-05,
      "loss": 0.9207,
      "step": 2242
    },
    {
      "epoch": 0.7082412377644458,
      "grad_norm": 1.175956130027771,
      "learning_rate": 4.768788155217178e-05,
      "loss": 0.9437,
      "step": 2243
    },
    {
      "epoch": 0.7085569940006315,
      "grad_norm": 1.1338388919830322,
      "learning_rate": 4.768437302645429e-05,
      "loss": 0.963,
      "step": 2244
    },
    {
      "epoch": 0.7088727502368172,
      "grad_norm": 1.0861842632293701,
      "learning_rate": 4.7680864500736796e-05,
      "loss": 0.9032,
      "step": 2245
    },
    {
      "epoch": 0.7091885064730028,
      "grad_norm": 0.9738459587097168,
      "learning_rate": 4.76773559750193e-05,
      "loss": 0.8738,
      "step": 2246
    },
    {
      "epoch": 0.7095042627091885,
      "grad_norm": 1.324927568435669,
      "learning_rate": 4.767384744930181e-05,
      "loss": 0.9269,
      "step": 2247
    },
    {
      "epoch": 0.7098200189453742,
      "grad_norm": 0.8876687288284302,
      "learning_rate": 4.767033892358431e-05,
      "loss": 0.8953,
      "step": 2248
    },
    {
      "epoch": 0.7101357751815598,
      "grad_norm": 1.0507029294967651,
      "learning_rate": 4.7666830397866816e-05,
      "loss": 0.9288,
      "step": 2249
    },
    {
      "epoch": 0.7104515314177455,
      "grad_norm": 0.877196192741394,
      "learning_rate": 4.766332187214932e-05,
      "loss": 0.9357,
      "step": 2250
    },
    {
      "epoch": 0.7107672876539312,
      "grad_norm": 0.8967514634132385,
      "learning_rate": 4.765981334643183e-05,
      "loss": 0.9568,
      "step": 2251
    },
    {
      "epoch": 0.7110830438901168,
      "grad_norm": 0.8865492343902588,
      "learning_rate": 4.765630482071434e-05,
      "loss": 0.9111,
      "step": 2252
    },
    {
      "epoch": 0.7113988001263025,
      "grad_norm": 1.0009853839874268,
      "learning_rate": 4.7652796294996843e-05,
      "loss": 0.9445,
      "step": 2253
    },
    {
      "epoch": 0.7117145563624881,
      "grad_norm": 0.8606018424034119,
      "learning_rate": 4.764928776927935e-05,
      "loss": 0.9336,
      "step": 2254
    },
    {
      "epoch": 0.7120303125986738,
      "grad_norm": 0.9121329188346863,
      "learning_rate": 4.764577924356186e-05,
      "loss": 0.9479,
      "step": 2255
    },
    {
      "epoch": 0.7123460688348595,
      "grad_norm": 0.873526930809021,
      "learning_rate": 4.7642270717844364e-05,
      "loss": 0.8632,
      "step": 2256
    },
    {
      "epoch": 0.7126618250710451,
      "grad_norm": 1.053970217704773,
      "learning_rate": 4.763876219212687e-05,
      "loss": 0.9725,
      "step": 2257
    },
    {
      "epoch": 0.7129775813072308,
      "grad_norm": 0.8546570539474487,
      "learning_rate": 4.763525366640938e-05,
      "loss": 0.9133,
      "step": 2258
    },
    {
      "epoch": 0.7132933375434165,
      "grad_norm": 1.08699369430542,
      "learning_rate": 4.7631745140691884e-05,
      "loss": 0.9819,
      "step": 2259
    },
    {
      "epoch": 0.7136090937796021,
      "grad_norm": 1.0052090883255005,
      "learning_rate": 4.762823661497439e-05,
      "loss": 0.8794,
      "step": 2260
    },
    {
      "epoch": 0.7139248500157878,
      "grad_norm": 0.9568954110145569,
      "learning_rate": 4.76247280892569e-05,
      "loss": 0.9482,
      "step": 2261
    },
    {
      "epoch": 0.7142406062519735,
      "grad_norm": 0.9804130792617798,
      "learning_rate": 4.7621219563539405e-05,
      "loss": 0.9188,
      "step": 2262
    },
    {
      "epoch": 0.7145563624881591,
      "grad_norm": 0.9744316339492798,
      "learning_rate": 4.761771103782191e-05,
      "loss": 0.9274,
      "step": 2263
    },
    {
      "epoch": 0.7148721187243448,
      "grad_norm": 0.8781080842018127,
      "learning_rate": 4.761420251210442e-05,
      "loss": 0.921,
      "step": 2264
    },
    {
      "epoch": 0.7151878749605305,
      "grad_norm": 1.0413039922714233,
      "learning_rate": 4.7610693986386925e-05,
      "loss": 0.8599,
      "step": 2265
    },
    {
      "epoch": 0.7155036311967161,
      "grad_norm": 0.9382190108299255,
      "learning_rate": 4.760718546066943e-05,
      "loss": 0.9384,
      "step": 2266
    },
    {
      "epoch": 0.7158193874329019,
      "grad_norm": 1.1032063961029053,
      "learning_rate": 4.760367693495193e-05,
      "loss": 0.9548,
      "step": 2267
    },
    {
      "epoch": 0.7161351436690875,
      "grad_norm": 1.0266262292861938,
      "learning_rate": 4.760016840923444e-05,
      "loss": 0.9417,
      "step": 2268
    },
    {
      "epoch": 0.7164508999052731,
      "grad_norm": 0.9846084117889404,
      "learning_rate": 4.7596659883516946e-05,
      "loss": 0.9049,
      "step": 2269
    },
    {
      "epoch": 0.7167666561414588,
      "grad_norm": 1.0525825023651123,
      "learning_rate": 4.759315135779945e-05,
      "loss": 0.9197,
      "step": 2270
    },
    {
      "epoch": 0.7170824123776445,
      "grad_norm": 1.0024794340133667,
      "learning_rate": 4.758964283208196e-05,
      "loss": 0.9147,
      "step": 2271
    },
    {
      "epoch": 0.7173981686138301,
      "grad_norm": 0.8826039433479309,
      "learning_rate": 4.7586134306364466e-05,
      "loss": 0.9218,
      "step": 2272
    },
    {
      "epoch": 0.7177139248500158,
      "grad_norm": 0.9865855574607849,
      "learning_rate": 4.758262578064697e-05,
      "loss": 0.9124,
      "step": 2273
    },
    {
      "epoch": 0.7180296810862015,
      "grad_norm": 1.0676836967468262,
      "learning_rate": 4.757911725492948e-05,
      "loss": 0.942,
      "step": 2274
    },
    {
      "epoch": 0.7183454373223871,
      "grad_norm": 1.1050100326538086,
      "learning_rate": 4.757560872921199e-05,
      "loss": 0.8994,
      "step": 2275
    },
    {
      "epoch": 0.7186611935585728,
      "grad_norm": 0.9023856520652771,
      "learning_rate": 4.757210020349449e-05,
      "loss": 0.9314,
      "step": 2276
    },
    {
      "epoch": 0.7189769497947585,
      "grad_norm": 0.8793060183525085,
      "learning_rate": 4.7568591677777e-05,
      "loss": 0.888,
      "step": 2277
    },
    {
      "epoch": 0.7192927060309441,
      "grad_norm": 1.1940664052963257,
      "learning_rate": 4.756508315205951e-05,
      "loss": 0.9183,
      "step": 2278
    },
    {
      "epoch": 0.7196084622671298,
      "grad_norm": 0.9811655879020691,
      "learning_rate": 4.7561574626342014e-05,
      "loss": 0.9528,
      "step": 2279
    },
    {
      "epoch": 0.7199242185033154,
      "grad_norm": 0.9118564128875732,
      "learning_rate": 4.755806610062452e-05,
      "loss": 0.9624,
      "step": 2280
    },
    {
      "epoch": 0.7202399747395011,
      "grad_norm": 0.8215085864067078,
      "learning_rate": 4.755455757490703e-05,
      "loss": 0.8561,
      "step": 2281
    },
    {
      "epoch": 0.7205557309756868,
      "grad_norm": 0.835200846195221,
      "learning_rate": 4.7551049049189534e-05,
      "loss": 0.9078,
      "step": 2282
    },
    {
      "epoch": 0.7208714872118724,
      "grad_norm": 1.0376169681549072,
      "learning_rate": 4.754754052347204e-05,
      "loss": 0.9414,
      "step": 2283
    },
    {
      "epoch": 0.7211872434480581,
      "grad_norm": 0.840149998664856,
      "learning_rate": 4.754403199775455e-05,
      "loss": 0.9093,
      "step": 2284
    },
    {
      "epoch": 0.7215029996842438,
      "grad_norm": 1.206997275352478,
      "learning_rate": 4.7540523472037055e-05,
      "loss": 0.9205,
      "step": 2285
    },
    {
      "epoch": 0.7218187559204294,
      "grad_norm": 1.1440268754959106,
      "learning_rate": 4.7537014946319555e-05,
      "loss": 1.0071,
      "step": 2286
    },
    {
      "epoch": 0.7221345121566151,
      "grad_norm": 0.7800721526145935,
      "learning_rate": 4.753350642060206e-05,
      "loss": 0.9304,
      "step": 2287
    },
    {
      "epoch": 0.7224502683928008,
      "grad_norm": 1.1156502962112427,
      "learning_rate": 4.752999789488457e-05,
      "loss": 0.8947,
      "step": 2288
    },
    {
      "epoch": 0.7227660246289864,
      "grad_norm": 0.9227898120880127,
      "learning_rate": 4.7526489369167075e-05,
      "loss": 0.9401,
      "step": 2289
    },
    {
      "epoch": 0.723081780865172,
      "grad_norm": 0.9131006002426147,
      "learning_rate": 4.752298084344958e-05,
      "loss": 0.8883,
      "step": 2290
    },
    {
      "epoch": 0.7233975371013578,
      "grad_norm": 0.9739009737968445,
      "learning_rate": 4.751947231773209e-05,
      "loss": 0.8906,
      "step": 2291
    },
    {
      "epoch": 0.7237132933375434,
      "grad_norm": 0.9689772129058838,
      "learning_rate": 4.7515963792014596e-05,
      "loss": 0.9581,
      "step": 2292
    },
    {
      "epoch": 0.724029049573729,
      "grad_norm": 1.0806829929351807,
      "learning_rate": 4.751245526629711e-05,
      "loss": 0.9112,
      "step": 2293
    },
    {
      "epoch": 0.7243448058099148,
      "grad_norm": 1.2985413074493408,
      "learning_rate": 4.7508946740579616e-05,
      "loss": 0.9519,
      "step": 2294
    },
    {
      "epoch": 0.7246605620461004,
      "grad_norm": 0.9401666522026062,
      "learning_rate": 4.7505438214862116e-05,
      "loss": 0.9425,
      "step": 2295
    },
    {
      "epoch": 0.724976318282286,
      "grad_norm": 1.0736732482910156,
      "learning_rate": 4.750192968914462e-05,
      "loss": 0.8858,
      "step": 2296
    },
    {
      "epoch": 0.7252920745184718,
      "grad_norm": 1.344950795173645,
      "learning_rate": 4.749842116342713e-05,
      "loss": 0.9268,
      "step": 2297
    },
    {
      "epoch": 0.7256078307546574,
      "grad_norm": 0.9958131909370422,
      "learning_rate": 4.7494912637709636e-05,
      "loss": 0.9318,
      "step": 2298
    },
    {
      "epoch": 0.725923586990843,
      "grad_norm": 1.1805230379104614,
      "learning_rate": 4.749140411199214e-05,
      "loss": 0.942,
      "step": 2299
    },
    {
      "epoch": 0.7262393432270288,
      "grad_norm": 1.1495006084442139,
      "learning_rate": 4.748789558627465e-05,
      "loss": 0.8881,
      "step": 2300
    },
    {
      "epoch": 0.7265550994632144,
      "grad_norm": 1.2691855430603027,
      "learning_rate": 4.748438706055716e-05,
      "loss": 0.9415,
      "step": 2301
    },
    {
      "epoch": 0.7268708556994,
      "grad_norm": 0.9810152649879456,
      "learning_rate": 4.7480878534839664e-05,
      "loss": 0.8959,
      "step": 2302
    },
    {
      "epoch": 0.7271866119355858,
      "grad_norm": 0.9843587875366211,
      "learning_rate": 4.747737000912217e-05,
      "loss": 0.9793,
      "step": 2303
    },
    {
      "epoch": 0.7275023681717714,
      "grad_norm": 0.8970917463302612,
      "learning_rate": 4.747386148340468e-05,
      "loss": 0.9084,
      "step": 2304
    },
    {
      "epoch": 0.727818124407957,
      "grad_norm": 1.0479366779327393,
      "learning_rate": 4.7470352957687184e-05,
      "loss": 0.9552,
      "step": 2305
    },
    {
      "epoch": 0.7281338806441428,
      "grad_norm": 2.6102712154388428,
      "learning_rate": 4.7466844431969684e-05,
      "loss": 0.9898,
      "step": 2306
    },
    {
      "epoch": 0.7284496368803284,
      "grad_norm": 0.936907947063446,
      "learning_rate": 4.746333590625219e-05,
      "loss": 0.952,
      "step": 2307
    },
    {
      "epoch": 0.728765393116514,
      "grad_norm": 1.0764626264572144,
      "learning_rate": 4.74598273805347e-05,
      "loss": 0.9584,
      "step": 2308
    },
    {
      "epoch": 0.7290811493526997,
      "grad_norm": 0.9630387425422668,
      "learning_rate": 4.7456318854817205e-05,
      "loss": 0.9432,
      "step": 2309
    },
    {
      "epoch": 0.7293969055888854,
      "grad_norm": 1.0322999954223633,
      "learning_rate": 4.745281032909972e-05,
      "loss": 0.8978,
      "step": 2310
    },
    {
      "epoch": 0.729712661825071,
      "grad_norm": 0.956575334072113,
      "learning_rate": 4.7449301803382225e-05,
      "loss": 0.9048,
      "step": 2311
    },
    {
      "epoch": 0.7300284180612567,
      "grad_norm": 0.8905308842658997,
      "learning_rate": 4.744579327766473e-05,
      "loss": 0.8644,
      "step": 2312
    },
    {
      "epoch": 0.7303441742974424,
      "grad_norm": 0.9066529273986816,
      "learning_rate": 4.744228475194724e-05,
      "loss": 0.9187,
      "step": 2313
    },
    {
      "epoch": 0.730659930533628,
      "grad_norm": 0.9589291214942932,
      "learning_rate": 4.743877622622974e-05,
      "loss": 0.8945,
      "step": 2314
    },
    {
      "epoch": 0.7309756867698137,
      "grad_norm": 0.9422157406806946,
      "learning_rate": 4.7435267700512246e-05,
      "loss": 0.8876,
      "step": 2315
    },
    {
      "epoch": 0.7312914430059994,
      "grad_norm": 0.9243563413619995,
      "learning_rate": 4.743175917479475e-05,
      "loss": 0.9216,
      "step": 2316
    },
    {
      "epoch": 0.731607199242185,
      "grad_norm": 1.502712368965149,
      "learning_rate": 4.742825064907726e-05,
      "loss": 0.9788,
      "step": 2317
    },
    {
      "epoch": 0.7319229554783707,
      "grad_norm": 1.0867252349853516,
      "learning_rate": 4.7424742123359766e-05,
      "loss": 0.9058,
      "step": 2318
    },
    {
      "epoch": 0.7322387117145563,
      "grad_norm": 0.9679430723190308,
      "learning_rate": 4.742123359764227e-05,
      "loss": 0.8734,
      "step": 2319
    },
    {
      "epoch": 0.732554467950742,
      "grad_norm": 1.1443034410476685,
      "learning_rate": 4.741772507192478e-05,
      "loss": 0.8911,
      "step": 2320
    },
    {
      "epoch": 0.7328702241869277,
      "grad_norm": 1.2351187467575073,
      "learning_rate": 4.7414216546207286e-05,
      "loss": 0.9188,
      "step": 2321
    },
    {
      "epoch": 0.7331859804231133,
      "grad_norm": 1.1379988193511963,
      "learning_rate": 4.741070802048979e-05,
      "loss": 0.955,
      "step": 2322
    },
    {
      "epoch": 0.7335017366592991,
      "grad_norm": 0.9476922154426575,
      "learning_rate": 4.74071994947723e-05,
      "loss": 0.8916,
      "step": 2323
    },
    {
      "epoch": 0.7338174928954847,
      "grad_norm": 1.0605127811431885,
      "learning_rate": 4.740369096905481e-05,
      "loss": 0.9624,
      "step": 2324
    },
    {
      "epoch": 0.7341332491316703,
      "grad_norm": 1.5128973722457886,
      "learning_rate": 4.740018244333731e-05,
      "loss": 0.9651,
      "step": 2325
    },
    {
      "epoch": 0.7344490053678561,
      "grad_norm": 0.9260780811309814,
      "learning_rate": 4.7396673917619814e-05,
      "loss": 0.9105,
      "step": 2326
    },
    {
      "epoch": 0.7347647616040417,
      "grad_norm": 1.085620641708374,
      "learning_rate": 4.739316539190232e-05,
      "loss": 0.8902,
      "step": 2327
    },
    {
      "epoch": 0.7350805178402273,
      "grad_norm": 1.25839364528656,
      "learning_rate": 4.7389656866184834e-05,
      "loss": 0.9203,
      "step": 2328
    },
    {
      "epoch": 0.7353962740764131,
      "grad_norm": 1.200541377067566,
      "learning_rate": 4.738614834046734e-05,
      "loss": 0.9461,
      "step": 2329
    },
    {
      "epoch": 0.7357120303125987,
      "grad_norm": 1.1039355993270874,
      "learning_rate": 4.738263981474985e-05,
      "loss": 0.9485,
      "step": 2330
    },
    {
      "epoch": 0.7360277865487843,
      "grad_norm": 0.8441333174705505,
      "learning_rate": 4.7379131289032355e-05,
      "loss": 0.9371,
      "step": 2331
    },
    {
      "epoch": 0.73634354278497,
      "grad_norm": 1.1380435228347778,
      "learning_rate": 4.737562276331486e-05,
      "loss": 0.8628,
      "step": 2332
    },
    {
      "epoch": 0.7366592990211557,
      "grad_norm": 1.0509103536605835,
      "learning_rate": 4.737211423759737e-05,
      "loss": 0.9102,
      "step": 2333
    },
    {
      "epoch": 0.7369750552573413,
      "grad_norm": 1.3237251043319702,
      "learning_rate": 4.736860571187987e-05,
      "loss": 0.9468,
      "step": 2334
    },
    {
      "epoch": 0.737290811493527,
      "grad_norm": 1.0802918672561646,
      "learning_rate": 4.7365097186162375e-05,
      "loss": 0.9089,
      "step": 2335
    },
    {
      "epoch": 0.7376065677297127,
      "grad_norm": 1.4385875463485718,
      "learning_rate": 4.736158866044488e-05,
      "loss": 0.8481,
      "step": 2336
    },
    {
      "epoch": 0.7379223239658983,
      "grad_norm": 1.0113210678100586,
      "learning_rate": 4.735808013472739e-05,
      "loss": 0.9106,
      "step": 2337
    },
    {
      "epoch": 0.738238080202084,
      "grad_norm": 1.0233670473098755,
      "learning_rate": 4.7354571609009895e-05,
      "loss": 0.8862,
      "step": 2338
    },
    {
      "epoch": 0.7385538364382697,
      "grad_norm": 1.2118414640426636,
      "learning_rate": 4.73510630832924e-05,
      "loss": 0.8976,
      "step": 2339
    },
    {
      "epoch": 0.7388695926744553,
      "grad_norm": 0.9537255764007568,
      "learning_rate": 4.734755455757491e-05,
      "loss": 0.9514,
      "step": 2340
    },
    {
      "epoch": 0.739185348910641,
      "grad_norm": 1.0248031616210938,
      "learning_rate": 4.7344046031857416e-05,
      "loss": 0.9429,
      "step": 2341
    },
    {
      "epoch": 0.7395011051468267,
      "grad_norm": 0.9835237264633179,
      "learning_rate": 4.734053750613992e-05,
      "loss": 0.8586,
      "step": 2342
    },
    {
      "epoch": 0.7398168613830123,
      "grad_norm": 1.1323457956314087,
      "learning_rate": 4.733702898042243e-05,
      "loss": 0.9258,
      "step": 2343
    },
    {
      "epoch": 0.740132617619198,
      "grad_norm": 1.066902756690979,
      "learning_rate": 4.733352045470493e-05,
      "loss": 0.9224,
      "step": 2344
    },
    {
      "epoch": 0.7404483738553836,
      "grad_norm": 1.4770880937576294,
      "learning_rate": 4.733001192898744e-05,
      "loss": 0.9452,
      "step": 2345
    },
    {
      "epoch": 0.7407641300915693,
      "grad_norm": 2.372875690460205,
      "learning_rate": 4.732650340326995e-05,
      "loss": 0.9375,
      "step": 2346
    },
    {
      "epoch": 0.741079886327755,
      "grad_norm": 1.5194586515426636,
      "learning_rate": 4.732299487755246e-05,
      "loss": 0.8828,
      "step": 2347
    },
    {
      "epoch": 0.7413956425639406,
      "grad_norm": 1.3629095554351807,
      "learning_rate": 4.7319486351834964e-05,
      "loss": 0.885,
      "step": 2348
    },
    {
      "epoch": 0.7417113988001263,
      "grad_norm": 1.235119342803955,
      "learning_rate": 4.731597782611747e-05,
      "loss": 0.9053,
      "step": 2349
    },
    {
      "epoch": 0.742027155036312,
      "grad_norm": 0.9796966910362244,
      "learning_rate": 4.731246930039998e-05,
      "loss": 0.9181,
      "step": 2350
    },
    {
      "epoch": 0.7423429112724976,
      "grad_norm": 1.306685447692871,
      "learning_rate": 4.7308960774682484e-05,
      "loss": 0.9101,
      "step": 2351
    },
    {
      "epoch": 0.7426586675086833,
      "grad_norm": 1.0761622190475464,
      "learning_rate": 4.730545224896499e-05,
      "loss": 0.9375,
      "step": 2352
    },
    {
      "epoch": 0.742974423744869,
      "grad_norm": 1.525148630142212,
      "learning_rate": 4.730194372324749e-05,
      "loss": 0.9085,
      "step": 2353
    },
    {
      "epoch": 0.7432901799810546,
      "grad_norm": 1.3216383457183838,
      "learning_rate": 4.729843519753e-05,
      "loss": 0.9336,
      "step": 2354
    },
    {
      "epoch": 0.7436059362172402,
      "grad_norm": 1.3027753829956055,
      "learning_rate": 4.7294926671812504e-05,
      "loss": 0.8763,
      "step": 2355
    },
    {
      "epoch": 0.743921692453426,
      "grad_norm": 0.9309592247009277,
      "learning_rate": 4.729141814609501e-05,
      "loss": 0.8813,
      "step": 2356
    },
    {
      "epoch": 0.7442374486896116,
      "grad_norm": 1.056947112083435,
      "learning_rate": 4.728790962037752e-05,
      "loss": 0.8596,
      "step": 2357
    },
    {
      "epoch": 0.7445532049257972,
      "grad_norm": 1.0086640119552612,
      "learning_rate": 4.7284401094660025e-05,
      "loss": 0.8957,
      "step": 2358
    },
    {
      "epoch": 0.744868961161983,
      "grad_norm": 1.1591912508010864,
      "learning_rate": 4.728089256894253e-05,
      "loss": 0.9264,
      "step": 2359
    },
    {
      "epoch": 0.7451847173981686,
      "grad_norm": 1.103414535522461,
      "learning_rate": 4.727738404322504e-05,
      "loss": 0.9364,
      "step": 2360
    },
    {
      "epoch": 0.7455004736343542,
      "grad_norm": 1.220001220703125,
      "learning_rate": 4.7273875517507545e-05,
      "loss": 0.8759,
      "step": 2361
    },
    {
      "epoch": 0.74581622987054,
      "grad_norm": 0.99479079246521,
      "learning_rate": 4.727036699179005e-05,
      "loss": 0.9258,
      "step": 2362
    },
    {
      "epoch": 0.7461319861067256,
      "grad_norm": 1.044878602027893,
      "learning_rate": 4.726685846607256e-05,
      "loss": 0.914,
      "step": 2363
    },
    {
      "epoch": 0.7464477423429112,
      "grad_norm": 1.1415135860443115,
      "learning_rate": 4.7263349940355066e-05,
      "loss": 0.9343,
      "step": 2364
    },
    {
      "epoch": 0.746763498579097,
      "grad_norm": 1.0046498775482178,
      "learning_rate": 4.725984141463757e-05,
      "loss": 0.8716,
      "step": 2365
    },
    {
      "epoch": 0.7470792548152826,
      "grad_norm": 0.9811848998069763,
      "learning_rate": 4.725633288892008e-05,
      "loss": 0.8996,
      "step": 2366
    },
    {
      "epoch": 0.7473950110514682,
      "grad_norm": 1.002556324005127,
      "learning_rate": 4.7252824363202586e-05,
      "loss": 0.9333,
      "step": 2367
    },
    {
      "epoch": 0.747710767287654,
      "grad_norm": 0.9278732538223267,
      "learning_rate": 4.724931583748509e-05,
      "loss": 0.985,
      "step": 2368
    },
    {
      "epoch": 0.7480265235238396,
      "grad_norm": 0.8661118745803833,
      "learning_rate": 4.72458073117676e-05,
      "loss": 0.9175,
      "step": 2369
    },
    {
      "epoch": 0.7483422797600252,
      "grad_norm": 1.4631303548812866,
      "learning_rate": 4.724229878605011e-05,
      "loss": 1.0072,
      "step": 2370
    },
    {
      "epoch": 0.748658035996211,
      "grad_norm": 1.1179267168045044,
      "learning_rate": 4.7238790260332613e-05,
      "loss": 0.9466,
      "step": 2371
    },
    {
      "epoch": 0.7489737922323966,
      "grad_norm": 1.0272197723388672,
      "learning_rate": 4.7235281734615114e-05,
      "loss": 0.9179,
      "step": 2372
    },
    {
      "epoch": 0.7492895484685822,
      "grad_norm": 0.8711735606193542,
      "learning_rate": 4.723177320889762e-05,
      "loss": 0.8681,
      "step": 2373
    },
    {
      "epoch": 0.7496053047047679,
      "grad_norm": 1.0161163806915283,
      "learning_rate": 4.722826468318013e-05,
      "loss": 0.9262,
      "step": 2374
    },
    {
      "epoch": 0.7499210609409536,
      "grad_norm": 1.0041252374649048,
      "learning_rate": 4.7224756157462634e-05,
      "loss": 0.9033,
      "step": 2375
    },
    {
      "epoch": 0.7502368171771393,
      "grad_norm": 0.974001944065094,
      "learning_rate": 4.722124763174514e-05,
      "loss": 0.9776,
      "step": 2376
    },
    {
      "epoch": 0.7505525734133249,
      "grad_norm": 1.188202977180481,
      "learning_rate": 4.721773910602765e-05,
      "loss": 0.8996,
      "step": 2377
    },
    {
      "epoch": 0.7508683296495106,
      "grad_norm": 0.9414958357810974,
      "learning_rate": 4.7214230580310154e-05,
      "loss": 0.892,
      "step": 2378
    },
    {
      "epoch": 0.7511840858856963,
      "grad_norm": 0.9819070100784302,
      "learning_rate": 4.721072205459266e-05,
      "loss": 0.9264,
      "step": 2379
    },
    {
      "epoch": 0.7514998421218819,
      "grad_norm": 1.3886284828186035,
      "learning_rate": 4.7207213528875175e-05,
      "loss": 0.9039,
      "step": 2380
    },
    {
      "epoch": 0.7518155983580675,
      "grad_norm": 1.053699254989624,
      "learning_rate": 4.7203705003157675e-05,
      "loss": 0.9026,
      "step": 2381
    },
    {
      "epoch": 0.7521313545942533,
      "grad_norm": 0.9277195930480957,
      "learning_rate": 4.720019647744018e-05,
      "loss": 0.9093,
      "step": 2382
    },
    {
      "epoch": 0.7524471108304389,
      "grad_norm": 0.9577584862709045,
      "learning_rate": 4.719668795172269e-05,
      "loss": 0.9475,
      "step": 2383
    },
    {
      "epoch": 0.7527628670666245,
      "grad_norm": 0.9396375417709351,
      "learning_rate": 4.7193179426005195e-05,
      "loss": 0.9221,
      "step": 2384
    },
    {
      "epoch": 0.7530786233028103,
      "grad_norm": 1.2239817380905151,
      "learning_rate": 4.71896709002877e-05,
      "loss": 0.8862,
      "step": 2385
    },
    {
      "epoch": 0.7533943795389959,
      "grad_norm": 1.0646113157272339,
      "learning_rate": 4.718616237457021e-05,
      "loss": 0.9218,
      "step": 2386
    },
    {
      "epoch": 0.7537101357751815,
      "grad_norm": 0.9182193875312805,
      "learning_rate": 4.7182653848852716e-05,
      "loss": 0.9407,
      "step": 2387
    },
    {
      "epoch": 0.7540258920113673,
      "grad_norm": 1.003642201423645,
      "learning_rate": 4.717914532313522e-05,
      "loss": 0.9259,
      "step": 2388
    },
    {
      "epoch": 0.7543416482475529,
      "grad_norm": 1.0132207870483398,
      "learning_rate": 4.717563679741773e-05,
      "loss": 0.9331,
      "step": 2389
    },
    {
      "epoch": 0.7546574044837385,
      "grad_norm": 0.9536680579185486,
      "learning_rate": 4.7172128271700236e-05,
      "loss": 0.9,
      "step": 2390
    },
    {
      "epoch": 0.7549731607199243,
      "grad_norm": 1.1672961711883545,
      "learning_rate": 4.7168619745982736e-05,
      "loss": 0.9199,
      "step": 2391
    },
    {
      "epoch": 0.7552889169561099,
      "grad_norm": 0.9714042544364929,
      "learning_rate": 4.716511122026524e-05,
      "loss": 0.9269,
      "step": 2392
    },
    {
      "epoch": 0.7556046731922955,
      "grad_norm": 0.9416836500167847,
      "learning_rate": 4.716160269454775e-05,
      "loss": 0.9817,
      "step": 2393
    },
    {
      "epoch": 0.7559204294284813,
      "grad_norm": 0.8809195160865784,
      "learning_rate": 4.715809416883026e-05,
      "loss": 0.9471,
      "step": 2394
    },
    {
      "epoch": 0.7562361856646669,
      "grad_norm": 0.825110673904419,
      "learning_rate": 4.7154585643112763e-05,
      "loss": 0.888,
      "step": 2395
    },
    {
      "epoch": 0.7565519419008525,
      "grad_norm": 0.8345769047737122,
      "learning_rate": 4.715107711739527e-05,
      "loss": 0.9478,
      "step": 2396
    },
    {
      "epoch": 0.7568676981370382,
      "grad_norm": 0.9893448352813721,
      "learning_rate": 4.714756859167778e-05,
      "loss": 0.9093,
      "step": 2397
    },
    {
      "epoch": 0.7571834543732239,
      "grad_norm": 1.1800404787063599,
      "learning_rate": 4.714406006596029e-05,
      "loss": 0.9039,
      "step": 2398
    },
    {
      "epoch": 0.7574992106094095,
      "grad_norm": 1.2006659507751465,
      "learning_rate": 4.71405515402428e-05,
      "loss": 0.9836,
      "step": 2399
    },
    {
      "epoch": 0.7578149668455952,
      "grad_norm": 0.9321804046630859,
      "learning_rate": 4.71370430145253e-05,
      "loss": 0.8855,
      "step": 2400
    },
    {
      "epoch": 0.7578149668455952,
      "eval_loss": 0.8853708505630493,
      "eval_runtime": 4.7152,
      "eval_samples_per_second": 528.078,
      "eval_steps_per_second": 4.454,
      "step": 2400
    },
    {
      "epoch": 0.7581307230817809,
      "grad_norm": 1.057551383972168,
      "learning_rate": 4.7133534488807804e-05,
      "loss": 0.9391,
      "step": 2401
    },
    {
      "epoch": 0.7584464793179665,
      "grad_norm": 0.9835776090621948,
      "learning_rate": 4.713002596309031e-05,
      "loss": 0.883,
      "step": 2402
    },
    {
      "epoch": 0.7587622355541522,
      "grad_norm": 1.0707720518112183,
      "learning_rate": 4.712651743737282e-05,
      "loss": 0.9475,
      "step": 2403
    },
    {
      "epoch": 0.7590779917903379,
      "grad_norm": 1.1359708309173584,
      "learning_rate": 4.7123008911655325e-05,
      "loss": 0.9563,
      "step": 2404
    },
    {
      "epoch": 0.7593937480265235,
      "grad_norm": 0.9647411108016968,
      "learning_rate": 4.711950038593783e-05,
      "loss": 0.9453,
      "step": 2405
    },
    {
      "epoch": 0.7597095042627092,
      "grad_norm": 0.8763328194618225,
      "learning_rate": 4.711599186022034e-05,
      "loss": 0.9053,
      "step": 2406
    },
    {
      "epoch": 0.7600252604988949,
      "grad_norm": 0.943067193031311,
      "learning_rate": 4.7112483334502845e-05,
      "loss": 0.9061,
      "step": 2407
    },
    {
      "epoch": 0.7603410167350805,
      "grad_norm": 1.0446947813034058,
      "learning_rate": 4.710897480878535e-05,
      "loss": 0.9214,
      "step": 2408
    },
    {
      "epoch": 0.7606567729712662,
      "grad_norm": 0.9883646368980408,
      "learning_rate": 4.710546628306786e-05,
      "loss": 0.9768,
      "step": 2409
    },
    {
      "epoch": 0.7609725292074518,
      "grad_norm": 1.1656285524368286,
      "learning_rate": 4.7101957757350366e-05,
      "loss": 0.9056,
      "step": 2410
    },
    {
      "epoch": 0.7612882854436375,
      "grad_norm": 1.1079875230789185,
      "learning_rate": 4.7098449231632866e-05,
      "loss": 0.8853,
      "step": 2411
    },
    {
      "epoch": 0.7616040416798232,
      "grad_norm": 0.9276177287101746,
      "learning_rate": 4.709494070591537e-05,
      "loss": 0.898,
      "step": 2412
    },
    {
      "epoch": 0.7619197979160088,
      "grad_norm": 0.8929612636566162,
      "learning_rate": 4.709143218019788e-05,
      "loss": 0.9733,
      "step": 2413
    },
    {
      "epoch": 0.7622355541521945,
      "grad_norm": 0.9183929562568665,
      "learning_rate": 4.7087923654480386e-05,
      "loss": 0.9451,
      "step": 2414
    },
    {
      "epoch": 0.7625513103883802,
      "grad_norm": 1.1048945188522339,
      "learning_rate": 4.70844151287629e-05,
      "loss": 0.887,
      "step": 2415
    },
    {
      "epoch": 0.7628670666245658,
      "grad_norm": 0.8379672169685364,
      "learning_rate": 4.7080906603045407e-05,
      "loss": 0.9312,
      "step": 2416
    },
    {
      "epoch": 0.7631828228607515,
      "grad_norm": 1.0458478927612305,
      "learning_rate": 4.707739807732791e-05,
      "loss": 0.9604,
      "step": 2417
    },
    {
      "epoch": 0.7634985790969372,
      "grad_norm": 1.2742005586624146,
      "learning_rate": 4.707388955161042e-05,
      "loss": 0.9854,
      "step": 2418
    },
    {
      "epoch": 0.7638143353331228,
      "grad_norm": 0.9126197695732117,
      "learning_rate": 4.707038102589292e-05,
      "loss": 0.933,
      "step": 2419
    },
    {
      "epoch": 0.7641300915693084,
      "grad_norm": 1.1386970281600952,
      "learning_rate": 4.706687250017543e-05,
      "loss": 0.939,
      "step": 2420
    },
    {
      "epoch": 0.7644458478054942,
      "grad_norm": 0.9207906723022461,
      "learning_rate": 4.7063363974457934e-05,
      "loss": 0.9232,
      "step": 2421
    },
    {
      "epoch": 0.7647616040416798,
      "grad_norm": 1.0700156688690186,
      "learning_rate": 4.705985544874044e-05,
      "loss": 0.9672,
      "step": 2422
    },
    {
      "epoch": 0.7650773602778654,
      "grad_norm": 1.0199568271636963,
      "learning_rate": 4.705634692302295e-05,
      "loss": 0.8846,
      "step": 2423
    },
    {
      "epoch": 0.7653931165140512,
      "grad_norm": 1.050131916999817,
      "learning_rate": 4.7052838397305454e-05,
      "loss": 0.8811,
      "step": 2424
    },
    {
      "epoch": 0.7657088727502368,
      "grad_norm": 0.9537582993507385,
      "learning_rate": 4.704932987158796e-05,
      "loss": 0.8997,
      "step": 2425
    },
    {
      "epoch": 0.7660246289864224,
      "grad_norm": 0.8640641570091248,
      "learning_rate": 4.704582134587047e-05,
      "loss": 0.9443,
      "step": 2426
    },
    {
      "epoch": 0.7663403852226082,
      "grad_norm": 0.9888316988945007,
      "learning_rate": 4.7042312820152975e-05,
      "loss": 0.9186,
      "step": 2427
    },
    {
      "epoch": 0.7666561414587938,
      "grad_norm": 1.1052166223526,
      "learning_rate": 4.703880429443548e-05,
      "loss": 0.9385,
      "step": 2428
    },
    {
      "epoch": 0.7669718976949794,
      "grad_norm": 1.0135493278503418,
      "learning_rate": 4.703529576871799e-05,
      "loss": 0.8926,
      "step": 2429
    },
    {
      "epoch": 0.7672876539311652,
      "grad_norm": 1.2000917196273804,
      "learning_rate": 4.703178724300049e-05,
      "loss": 0.8951,
      "step": 2430
    },
    {
      "epoch": 0.7676034101673508,
      "grad_norm": 1.2580671310424805,
      "learning_rate": 4.7028278717282995e-05,
      "loss": 0.8677,
      "step": 2431
    },
    {
      "epoch": 0.7679191664035365,
      "grad_norm": 0.9188705086708069,
      "learning_rate": 4.70247701915655e-05,
      "loss": 0.878,
      "step": 2432
    },
    {
      "epoch": 0.7682349226397222,
      "grad_norm": 0.9529401659965515,
      "learning_rate": 4.7021261665848016e-05,
      "loss": 0.9298,
      "step": 2433
    },
    {
      "epoch": 0.7685506788759078,
      "grad_norm": 0.9539476633071899,
      "learning_rate": 4.701775314013052e-05,
      "loss": 0.9763,
      "step": 2434
    },
    {
      "epoch": 0.7688664351120935,
      "grad_norm": 0.9078298211097717,
      "learning_rate": 4.701424461441303e-05,
      "loss": 0.9067,
      "step": 2435
    },
    {
      "epoch": 0.7691821913482791,
      "grad_norm": 0.9847672581672668,
      "learning_rate": 4.7010736088695536e-05,
      "loss": 0.9091,
      "step": 2436
    },
    {
      "epoch": 0.7694979475844648,
      "grad_norm": 1.0050297975540161,
      "learning_rate": 4.700722756297804e-05,
      "loss": 0.9491,
      "step": 2437
    },
    {
      "epoch": 0.7698137038206505,
      "grad_norm": 1.077934741973877,
      "learning_rate": 4.700371903726055e-05,
      "loss": 0.9537,
      "step": 2438
    },
    {
      "epoch": 0.7701294600568361,
      "grad_norm": 0.865628182888031,
      "learning_rate": 4.700021051154305e-05,
      "loss": 0.9285,
      "step": 2439
    },
    {
      "epoch": 0.7704452162930218,
      "grad_norm": 1.0372687578201294,
      "learning_rate": 4.6996701985825556e-05,
      "loss": 0.9146,
      "step": 2440
    },
    {
      "epoch": 0.7707609725292075,
      "grad_norm": 1.059919834136963,
      "learning_rate": 4.699319346010806e-05,
      "loss": 0.9502,
      "step": 2441
    },
    {
      "epoch": 0.7710767287653931,
      "grad_norm": 1.1547670364379883,
      "learning_rate": 4.698968493439057e-05,
      "loss": 0.9264,
      "step": 2442
    },
    {
      "epoch": 0.7713924850015788,
      "grad_norm": 1.0374693870544434,
      "learning_rate": 4.698617640867308e-05,
      "loss": 0.9236,
      "step": 2443
    },
    {
      "epoch": 0.7717082412377645,
      "grad_norm": 1.138913631439209,
      "learning_rate": 4.6982667882955584e-05,
      "loss": 0.8785,
      "step": 2444
    },
    {
      "epoch": 0.7720239974739501,
      "grad_norm": 0.8783249258995056,
      "learning_rate": 4.697915935723809e-05,
      "loss": 0.9373,
      "step": 2445
    },
    {
      "epoch": 0.7723397537101357,
      "grad_norm": 1.7711900472640991,
      "learning_rate": 4.69756508315206e-05,
      "loss": 0.9412,
      "step": 2446
    },
    {
      "epoch": 0.7726555099463215,
      "grad_norm": 0.9840225577354431,
      "learning_rate": 4.6972142305803104e-05,
      "loss": 0.9311,
      "step": 2447
    },
    {
      "epoch": 0.7729712661825071,
      "grad_norm": 0.9134658575057983,
      "learning_rate": 4.696863378008561e-05,
      "loss": 0.9058,
      "step": 2448
    },
    {
      "epoch": 0.7732870224186927,
      "grad_norm": 0.997555136680603,
      "learning_rate": 4.696512525436811e-05,
      "loss": 0.9083,
      "step": 2449
    },
    {
      "epoch": 0.7736027786548785,
      "grad_norm": 1.1154942512512207,
      "learning_rate": 4.6961616728650625e-05,
      "loss": 0.9405,
      "step": 2450
    },
    {
      "epoch": 0.7739185348910641,
      "grad_norm": 1.1193689107894897,
      "learning_rate": 4.695810820293313e-05,
      "loss": 0.828,
      "step": 2451
    },
    {
      "epoch": 0.7742342911272497,
      "grad_norm": 0.7565919756889343,
      "learning_rate": 4.695459967721564e-05,
      "loss": 0.8811,
      "step": 2452
    },
    {
      "epoch": 0.7745500473634355,
      "grad_norm": 1.7679499387741089,
      "learning_rate": 4.6951091151498145e-05,
      "loss": 0.9155,
      "step": 2453
    },
    {
      "epoch": 0.7748658035996211,
      "grad_norm": 0.9557313919067383,
      "learning_rate": 4.694758262578065e-05,
      "loss": 0.9021,
      "step": 2454
    },
    {
      "epoch": 0.7751815598358067,
      "grad_norm": 1.0108293294906616,
      "learning_rate": 4.694407410006316e-05,
      "loss": 0.9332,
      "step": 2455
    },
    {
      "epoch": 0.7754973160719925,
      "grad_norm": 0.9184893369674683,
      "learning_rate": 4.6940565574345665e-05,
      "loss": 0.9261,
      "step": 2456
    },
    {
      "epoch": 0.7758130723081781,
      "grad_norm": 0.9110531806945801,
      "learning_rate": 4.693705704862817e-05,
      "loss": 0.8746,
      "step": 2457
    },
    {
      "epoch": 0.7761288285443637,
      "grad_norm": 1.0241234302520752,
      "learning_rate": 4.693354852291067e-05,
      "loss": 0.9645,
      "step": 2458
    },
    {
      "epoch": 0.7764445847805495,
      "grad_norm": 0.9003347754478455,
      "learning_rate": 4.693003999719318e-05,
      "loss": 0.8606,
      "step": 2459
    },
    {
      "epoch": 0.7767603410167351,
      "grad_norm": 0.9901620745658875,
      "learning_rate": 4.6926531471475686e-05,
      "loss": 0.8978,
      "step": 2460
    },
    {
      "epoch": 0.7770760972529207,
      "grad_norm": 1.0883458852767944,
      "learning_rate": 4.692302294575819e-05,
      "loss": 0.9303,
      "step": 2461
    },
    {
      "epoch": 0.7773918534891064,
      "grad_norm": 1.3450865745544434,
      "learning_rate": 4.69195144200407e-05,
      "loss": 0.8642,
      "step": 2462
    },
    {
      "epoch": 0.7777076097252921,
      "grad_norm": 1.2770307064056396,
      "learning_rate": 4.6916005894323206e-05,
      "loss": 0.9795,
      "step": 2463
    },
    {
      "epoch": 0.7780233659614777,
      "grad_norm": 1.1081246137619019,
      "learning_rate": 4.691249736860571e-05,
      "loss": 0.8731,
      "step": 2464
    },
    {
      "epoch": 0.7783391221976634,
      "grad_norm": 1.1094814538955688,
      "learning_rate": 4.690898884288822e-05,
      "loss": 0.9324,
      "step": 2465
    },
    {
      "epoch": 0.7786548784338491,
      "grad_norm": 1.1313260793685913,
      "learning_rate": 4.690548031717073e-05,
      "loss": 0.9024,
      "step": 2466
    },
    {
      "epoch": 0.7789706346700347,
      "grad_norm": 1.0317285060882568,
      "learning_rate": 4.6901971791453234e-05,
      "loss": 0.899,
      "step": 2467
    },
    {
      "epoch": 0.7792863909062204,
      "grad_norm": 1.2694274187088013,
      "learning_rate": 4.689846326573574e-05,
      "loss": 0.9352,
      "step": 2468
    },
    {
      "epoch": 0.779602147142406,
      "grad_norm": 1.1912763118743896,
      "learning_rate": 4.689495474001825e-05,
      "loss": 0.8958,
      "step": 2469
    },
    {
      "epoch": 0.7799179033785917,
      "grad_norm": 0.8991129398345947,
      "learning_rate": 4.6891446214300754e-05,
      "loss": 0.9113,
      "step": 2470
    },
    {
      "epoch": 0.7802336596147774,
      "grad_norm": 1.0854359865188599,
      "learning_rate": 4.688793768858326e-05,
      "loss": 0.971,
      "step": 2471
    },
    {
      "epoch": 0.780549415850963,
      "grad_norm": 1.126630425453186,
      "learning_rate": 4.688442916286577e-05,
      "loss": 0.8924,
      "step": 2472
    },
    {
      "epoch": 0.7808651720871487,
      "grad_norm": 1.0490480661392212,
      "learning_rate": 4.6880920637148275e-05,
      "loss": 0.9275,
      "step": 2473
    },
    {
      "epoch": 0.7811809283233344,
      "grad_norm": 1.0084460973739624,
      "learning_rate": 4.687741211143078e-05,
      "loss": 0.8347,
      "step": 2474
    },
    {
      "epoch": 0.78149668455952,
      "grad_norm": 1.0831611156463623,
      "learning_rate": 4.687390358571329e-05,
      "loss": 0.9342,
      "step": 2475
    },
    {
      "epoch": 0.7818124407957057,
      "grad_norm": 1.212357997894287,
      "learning_rate": 4.6870395059995795e-05,
      "loss": 0.8888,
      "step": 2476
    },
    {
      "epoch": 0.7821281970318914,
      "grad_norm": 1.1019783020019531,
      "learning_rate": 4.6866886534278295e-05,
      "loss": 0.888,
      "step": 2477
    },
    {
      "epoch": 0.782443953268077,
      "grad_norm": 1.079509973526001,
      "learning_rate": 4.68633780085608e-05,
      "loss": 0.9085,
      "step": 2478
    },
    {
      "epoch": 0.7827597095042627,
      "grad_norm": 1.4191691875457764,
      "learning_rate": 4.685986948284331e-05,
      "loss": 0.9236,
      "step": 2479
    },
    {
      "epoch": 0.7830754657404484,
      "grad_norm": 1.064729928970337,
      "learning_rate": 4.6856360957125815e-05,
      "loss": 0.953,
      "step": 2480
    },
    {
      "epoch": 0.783391221976634,
      "grad_norm": 1.1164205074310303,
      "learning_rate": 4.685285243140832e-05,
      "loss": 0.8678,
      "step": 2481
    },
    {
      "epoch": 0.7837069782128196,
      "grad_norm": 1.1589186191558838,
      "learning_rate": 4.684934390569083e-05,
      "loss": 0.8913,
      "step": 2482
    },
    {
      "epoch": 0.7840227344490054,
      "grad_norm": 1.217752456665039,
      "learning_rate": 4.6845835379973336e-05,
      "loss": 0.8675,
      "step": 2483
    },
    {
      "epoch": 0.784338490685191,
      "grad_norm": 1.4120724201202393,
      "learning_rate": 4.684232685425584e-05,
      "loss": 0.9774,
      "step": 2484
    },
    {
      "epoch": 0.7846542469213768,
      "grad_norm": 1.3064451217651367,
      "learning_rate": 4.6838818328538356e-05,
      "loss": 0.932,
      "step": 2485
    },
    {
      "epoch": 0.7849700031575624,
      "grad_norm": 1.483642339706421,
      "learning_rate": 4.6835309802820856e-05,
      "loss": 0.8883,
      "step": 2486
    },
    {
      "epoch": 0.785285759393748,
      "grad_norm": 1.1087448596954346,
      "learning_rate": 4.683180127710336e-05,
      "loss": 0.8669,
      "step": 2487
    },
    {
      "epoch": 0.7856015156299337,
      "grad_norm": 1.0396878719329834,
      "learning_rate": 4.682829275138587e-05,
      "loss": 0.9925,
      "step": 2488
    },
    {
      "epoch": 0.7859172718661194,
      "grad_norm": 1.1418548822402954,
      "learning_rate": 4.682478422566838e-05,
      "loss": 0.8494,
      "step": 2489
    },
    {
      "epoch": 0.786233028102305,
      "grad_norm": 1.175916075706482,
      "learning_rate": 4.6821275699950884e-05,
      "loss": 0.8903,
      "step": 2490
    },
    {
      "epoch": 0.7865487843384907,
      "grad_norm": 1.262366533279419,
      "learning_rate": 4.681776717423339e-05,
      "loss": 0.9041,
      "step": 2491
    },
    {
      "epoch": 0.7868645405746764,
      "grad_norm": 1.050649642944336,
      "learning_rate": 4.68142586485159e-05,
      "loss": 0.9067,
      "step": 2492
    },
    {
      "epoch": 0.787180296810862,
      "grad_norm": 1.5421031713485718,
      "learning_rate": 4.6810750122798404e-05,
      "loss": 0.9157,
      "step": 2493
    },
    {
      "epoch": 0.7874960530470477,
      "grad_norm": 1.1166247129440308,
      "learning_rate": 4.680724159708091e-05,
      "loss": 0.8581,
      "step": 2494
    },
    {
      "epoch": 0.7878118092832334,
      "grad_norm": 0.992364764213562,
      "learning_rate": 4.680373307136342e-05,
      "loss": 0.8741,
      "step": 2495
    },
    {
      "epoch": 0.788127565519419,
      "grad_norm": 1.125847339630127,
      "learning_rate": 4.680022454564592e-05,
      "loss": 0.9184,
      "step": 2496
    },
    {
      "epoch": 0.7884433217556047,
      "grad_norm": 0.9742746353149414,
      "learning_rate": 4.6796716019928424e-05,
      "loss": 0.869,
      "step": 2497
    },
    {
      "epoch": 0.7887590779917903,
      "grad_norm": 1.0400216579437256,
      "learning_rate": 4.679320749421093e-05,
      "loss": 0.8896,
      "step": 2498
    },
    {
      "epoch": 0.789074834227976,
      "grad_norm": 1.0562700033187866,
      "learning_rate": 4.678969896849344e-05,
      "loss": 0.9585,
      "step": 2499
    },
    {
      "epoch": 0.7893905904641617,
      "grad_norm": 1.2536072731018066,
      "learning_rate": 4.6786190442775945e-05,
      "loss": 0.9055,
      "step": 2500
    },
    {
      "epoch": 0.7897063467003473,
      "grad_norm": 1.0137526988983154,
      "learning_rate": 4.678268191705845e-05,
      "loss": 0.9524,
      "step": 2501
    },
    {
      "epoch": 0.790022102936533,
      "grad_norm": 0.9453123211860657,
      "learning_rate": 4.677917339134096e-05,
      "loss": 0.9947,
      "step": 2502
    },
    {
      "epoch": 0.7903378591727187,
      "grad_norm": 1.4619039297103882,
      "learning_rate": 4.677566486562347e-05,
      "loss": 0.89,
      "step": 2503
    },
    {
      "epoch": 0.7906536154089043,
      "grad_norm": 0.9926608204841614,
      "learning_rate": 4.677215633990598e-05,
      "loss": 0.9201,
      "step": 2504
    },
    {
      "epoch": 0.79096937164509,
      "grad_norm": 0.8761029243469238,
      "learning_rate": 4.676864781418848e-05,
      "loss": 0.8644,
      "step": 2505
    },
    {
      "epoch": 0.7912851278812757,
      "grad_norm": 1.0156453847885132,
      "learning_rate": 4.6765139288470986e-05,
      "loss": 0.9022,
      "step": 2506
    },
    {
      "epoch": 0.7916008841174613,
      "grad_norm": 1.1257244348526,
      "learning_rate": 4.676163076275349e-05,
      "loss": 0.9708,
      "step": 2507
    },
    {
      "epoch": 0.791916640353647,
      "grad_norm": 1.0327608585357666,
      "learning_rate": 4.6758122237036e-05,
      "loss": 0.93,
      "step": 2508
    },
    {
      "epoch": 0.7922323965898327,
      "grad_norm": 1.0810730457305908,
      "learning_rate": 4.6754613711318506e-05,
      "loss": 0.8971,
      "step": 2509
    },
    {
      "epoch": 0.7925481528260183,
      "grad_norm": 1.0213775634765625,
      "learning_rate": 4.675110518560101e-05,
      "loss": 0.9008,
      "step": 2510
    },
    {
      "epoch": 0.7928639090622039,
      "grad_norm": 1.0643651485443115,
      "learning_rate": 4.674759665988352e-05,
      "loss": 0.9462,
      "step": 2511
    },
    {
      "epoch": 0.7931796652983897,
      "grad_norm": 0.9709125757217407,
      "learning_rate": 4.674408813416603e-05,
      "loss": 0.9264,
      "step": 2512
    },
    {
      "epoch": 0.7934954215345753,
      "grad_norm": 0.9423490166664124,
      "learning_rate": 4.6740579608448534e-05,
      "loss": 0.9478,
      "step": 2513
    },
    {
      "epoch": 0.7938111777707609,
      "grad_norm": 1.02461576461792,
      "learning_rate": 4.673707108273104e-05,
      "loss": 0.8977,
      "step": 2514
    },
    {
      "epoch": 0.7941269340069467,
      "grad_norm": 0.9666416645050049,
      "learning_rate": 4.673356255701355e-05,
      "loss": 0.9291,
      "step": 2515
    },
    {
      "epoch": 0.7944426902431323,
      "grad_norm": 0.7893954515457153,
      "learning_rate": 4.673005403129605e-05,
      "loss": 0.9317,
      "step": 2516
    },
    {
      "epoch": 0.7947584464793179,
      "grad_norm": 0.8906834125518799,
      "learning_rate": 4.6726545505578554e-05,
      "loss": 0.8567,
      "step": 2517
    },
    {
      "epoch": 0.7950742027155037,
      "grad_norm": 0.9168080687522888,
      "learning_rate": 4.672303697986106e-05,
      "loss": 0.9641,
      "step": 2518
    },
    {
      "epoch": 0.7953899589516893,
      "grad_norm": 1.5364433526992798,
      "learning_rate": 4.671952845414357e-05,
      "loss": 0.9168,
      "step": 2519
    },
    {
      "epoch": 0.7957057151878749,
      "grad_norm": 1.0429829359054565,
      "learning_rate": 4.671601992842608e-05,
      "loss": 0.9212,
      "step": 2520
    },
    {
      "epoch": 0.7960214714240607,
      "grad_norm": 1.233667016029358,
      "learning_rate": 4.671251140270859e-05,
      "loss": 0.9548,
      "step": 2521
    },
    {
      "epoch": 0.7963372276602463,
      "grad_norm": 1.2007853984832764,
      "learning_rate": 4.6709002876991095e-05,
      "loss": 0.9089,
      "step": 2522
    },
    {
      "epoch": 0.7966529838964319,
      "grad_norm": 0.9994527101516724,
      "learning_rate": 4.67054943512736e-05,
      "loss": 0.876,
      "step": 2523
    },
    {
      "epoch": 0.7969687401326176,
      "grad_norm": 1.155409574508667,
      "learning_rate": 4.67019858255561e-05,
      "loss": 0.8759,
      "step": 2524
    },
    {
      "epoch": 0.7972844963688033,
      "grad_norm": 1.1419814825057983,
      "learning_rate": 4.669847729983861e-05,
      "loss": 0.9217,
      "step": 2525
    },
    {
      "epoch": 0.7976002526049889,
      "grad_norm": 0.9090442657470703,
      "learning_rate": 4.6694968774121115e-05,
      "loss": 0.8922,
      "step": 2526
    },
    {
      "epoch": 0.7979160088411746,
      "grad_norm": 1.1741853952407837,
      "learning_rate": 4.669146024840362e-05,
      "loss": 0.9538,
      "step": 2527
    },
    {
      "epoch": 0.7982317650773603,
      "grad_norm": 1.4247479438781738,
      "learning_rate": 4.668795172268613e-05,
      "loss": 0.9072,
      "step": 2528
    },
    {
      "epoch": 0.7985475213135459,
      "grad_norm": 1.0867724418640137,
      "learning_rate": 4.6684443196968636e-05,
      "loss": 0.8982,
      "step": 2529
    },
    {
      "epoch": 0.7988632775497316,
      "grad_norm": 1.137753963470459,
      "learning_rate": 4.668093467125114e-05,
      "loss": 0.9792,
      "step": 2530
    },
    {
      "epoch": 0.7991790337859173,
      "grad_norm": 1.0444602966308594,
      "learning_rate": 4.667742614553365e-05,
      "loss": 0.8996,
      "step": 2531
    },
    {
      "epoch": 0.7994947900221029,
      "grad_norm": 1.091591238975525,
      "learning_rate": 4.6673917619816156e-05,
      "loss": 0.9375,
      "step": 2532
    },
    {
      "epoch": 0.7998105462582886,
      "grad_norm": 1.0095692873001099,
      "learning_rate": 4.667040909409866e-05,
      "loss": 0.8985,
      "step": 2533
    },
    {
      "epoch": 0.8001263024944743,
      "grad_norm": 1.1311588287353516,
      "learning_rate": 4.666690056838117e-05,
      "loss": 0.8922,
      "step": 2534
    },
    {
      "epoch": 0.8004420587306599,
      "grad_norm": 0.8456923961639404,
      "learning_rate": 4.666339204266367e-05,
      "loss": 0.9292,
      "step": 2535
    },
    {
      "epoch": 0.8007578149668456,
      "grad_norm": 1.1714831590652466,
      "learning_rate": 4.665988351694618e-05,
      "loss": 0.8656,
      "step": 2536
    },
    {
      "epoch": 0.8010735712030312,
      "grad_norm": 1.10301673412323,
      "learning_rate": 4.665637499122869e-05,
      "loss": 0.94,
      "step": 2537
    },
    {
      "epoch": 0.8013893274392169,
      "grad_norm": 1.1577216386795044,
      "learning_rate": 4.66528664655112e-05,
      "loss": 0.9176,
      "step": 2538
    },
    {
      "epoch": 0.8017050836754026,
      "grad_norm": 1.027536392211914,
      "learning_rate": 4.6649357939793704e-05,
      "loss": 0.9365,
      "step": 2539
    },
    {
      "epoch": 0.8020208399115882,
      "grad_norm": 1.0582525730133057,
      "learning_rate": 4.664584941407621e-05,
      "loss": 0.8527,
      "step": 2540
    },
    {
      "epoch": 0.802336596147774,
      "grad_norm": 1.1682722568511963,
      "learning_rate": 4.664234088835872e-05,
      "loss": 0.9155,
      "step": 2541
    },
    {
      "epoch": 0.8026523523839596,
      "grad_norm": 1.307728886604309,
      "learning_rate": 4.6638832362641224e-05,
      "loss": 0.971,
      "step": 2542
    },
    {
      "epoch": 0.8029681086201452,
      "grad_norm": 1.003188133239746,
      "learning_rate": 4.663532383692373e-05,
      "loss": 0.8726,
      "step": 2543
    },
    {
      "epoch": 0.803283864856331,
      "grad_norm": 1.08687424659729,
      "learning_rate": 4.663181531120623e-05,
      "loss": 0.9178,
      "step": 2544
    },
    {
      "epoch": 0.8035996210925166,
      "grad_norm": 0.9033157229423523,
      "learning_rate": 4.662830678548874e-05,
      "loss": 0.8764,
      "step": 2545
    },
    {
      "epoch": 0.8039153773287022,
      "grad_norm": 1.3611619472503662,
      "learning_rate": 4.6624798259771245e-05,
      "loss": 0.8942,
      "step": 2546
    },
    {
      "epoch": 0.804231133564888,
      "grad_norm": 1.257498860359192,
      "learning_rate": 4.662128973405375e-05,
      "loss": 0.8942,
      "step": 2547
    },
    {
      "epoch": 0.8045468898010736,
      "grad_norm": 1.067936897277832,
      "learning_rate": 4.661778120833626e-05,
      "loss": 0.9041,
      "step": 2548
    },
    {
      "epoch": 0.8048626460372592,
      "grad_norm": 1.1998472213745117,
      "learning_rate": 4.6614272682618765e-05,
      "loss": 0.8993,
      "step": 2549
    },
    {
      "epoch": 0.805178402273445,
      "grad_norm": 0.9164643287658691,
      "learning_rate": 4.661076415690127e-05,
      "loss": 0.9517,
      "step": 2550
    },
    {
      "epoch": 0.8054941585096306,
      "grad_norm": 0.9959022998809814,
      "learning_rate": 4.660725563118378e-05,
      "loss": 0.8948,
      "step": 2551
    },
    {
      "epoch": 0.8058099147458162,
      "grad_norm": 1.0842750072479248,
      "learning_rate": 4.6603747105466286e-05,
      "loss": 0.928,
      "step": 2552
    },
    {
      "epoch": 0.8061256709820019,
      "grad_norm": 0.9833402037620544,
      "learning_rate": 4.660023857974879e-05,
      "loss": 0.808,
      "step": 2553
    },
    {
      "epoch": 0.8064414272181876,
      "grad_norm": 1.0647450685501099,
      "learning_rate": 4.659673005403129e-05,
      "loss": 0.9271,
      "step": 2554
    },
    {
      "epoch": 0.8067571834543732,
      "grad_norm": 1.0537164211273193,
      "learning_rate": 4.6593221528313806e-05,
      "loss": 0.9029,
      "step": 2555
    },
    {
      "epoch": 0.8070729396905589,
      "grad_norm": 1.171040654182434,
      "learning_rate": 4.658971300259631e-05,
      "loss": 0.8959,
      "step": 2556
    },
    {
      "epoch": 0.8073886959267446,
      "grad_norm": 1.1169190406799316,
      "learning_rate": 4.658620447687882e-05,
      "loss": 0.9193,
      "step": 2557
    },
    {
      "epoch": 0.8077044521629302,
      "grad_norm": 1.0218583345413208,
      "learning_rate": 4.6582695951161327e-05,
      "loss": 0.8864,
      "step": 2558
    },
    {
      "epoch": 0.8080202083991159,
      "grad_norm": 0.9737411737442017,
      "learning_rate": 4.657918742544383e-05,
      "loss": 0.9303,
      "step": 2559
    },
    {
      "epoch": 0.8083359646353016,
      "grad_norm": 1.715042233467102,
      "learning_rate": 4.657567889972634e-05,
      "loss": 0.8968,
      "step": 2560
    },
    {
      "epoch": 0.8086517208714872,
      "grad_norm": 0.9986684322357178,
      "learning_rate": 4.657217037400885e-05,
      "loss": 0.8855,
      "step": 2561
    },
    {
      "epoch": 0.8089674771076729,
      "grad_norm": 1.0194976329803467,
      "learning_rate": 4.6568661848291354e-05,
      "loss": 1.0186,
      "step": 2562
    },
    {
      "epoch": 0.8092832333438585,
      "grad_norm": 0.9488517642021179,
      "learning_rate": 4.6565153322573854e-05,
      "loss": 0.8758,
      "step": 2563
    },
    {
      "epoch": 0.8095989895800442,
      "grad_norm": 1.0746160745620728,
      "learning_rate": 4.656164479685636e-05,
      "loss": 0.9124,
      "step": 2564
    },
    {
      "epoch": 0.8099147458162299,
      "grad_norm": 0.9158166646957397,
      "learning_rate": 4.655813627113887e-05,
      "loss": 0.8776,
      "step": 2565
    },
    {
      "epoch": 0.8102305020524155,
      "grad_norm": 1.0413742065429688,
      "learning_rate": 4.6554627745421374e-05,
      "loss": 0.8726,
      "step": 2566
    },
    {
      "epoch": 0.8105462582886012,
      "grad_norm": 1.1736472845077515,
      "learning_rate": 4.655111921970388e-05,
      "loss": 0.8926,
      "step": 2567
    },
    {
      "epoch": 0.8108620145247869,
      "grad_norm": 0.958565354347229,
      "learning_rate": 4.654761069398639e-05,
      "loss": 0.8619,
      "step": 2568
    },
    {
      "epoch": 0.8111777707609725,
      "grad_norm": 1.0312566757202148,
      "learning_rate": 4.6544102168268895e-05,
      "loss": 0.9112,
      "step": 2569
    },
    {
      "epoch": 0.8114935269971582,
      "grad_norm": 1.1037001609802246,
      "learning_rate": 4.65405936425514e-05,
      "loss": 0.9353,
      "step": 2570
    },
    {
      "epoch": 0.8118092832333439,
      "grad_norm": 0.8909184336662292,
      "learning_rate": 4.653708511683391e-05,
      "loss": 0.9108,
      "step": 2571
    },
    {
      "epoch": 0.8121250394695295,
      "grad_norm": 1.0792757272720337,
      "learning_rate": 4.6533576591116415e-05,
      "loss": 0.9536,
      "step": 2572
    },
    {
      "epoch": 0.8124407957057151,
      "grad_norm": 1.0808464288711548,
      "learning_rate": 4.653006806539892e-05,
      "loss": 0.9273,
      "step": 2573
    },
    {
      "epoch": 0.8127565519419009,
      "grad_norm": 0.9546955227851868,
      "learning_rate": 4.652655953968143e-05,
      "loss": 0.9466,
      "step": 2574
    },
    {
      "epoch": 0.8130723081780865,
      "grad_norm": 0.9062329530715942,
      "learning_rate": 4.6523051013963936e-05,
      "loss": 0.9287,
      "step": 2575
    },
    {
      "epoch": 0.8133880644142721,
      "grad_norm": 1.2340987920761108,
      "learning_rate": 4.651954248824644e-05,
      "loss": 0.8471,
      "step": 2576
    },
    {
      "epoch": 0.8137038206504579,
      "grad_norm": 0.9625616669654846,
      "learning_rate": 4.651603396252895e-05,
      "loss": 0.88,
      "step": 2577
    },
    {
      "epoch": 0.8140195768866435,
      "grad_norm": 1.1157644987106323,
      "learning_rate": 4.6512525436811456e-05,
      "loss": 0.9281,
      "step": 2578
    },
    {
      "epoch": 0.8143353331228291,
      "grad_norm": 0.8925914168357849,
      "learning_rate": 4.650901691109396e-05,
      "loss": 0.8459,
      "step": 2579
    },
    {
      "epoch": 0.8146510893590149,
      "grad_norm": 1.1432541608810425,
      "learning_rate": 4.650550838537647e-05,
      "loss": 0.9113,
      "step": 2580
    },
    {
      "epoch": 0.8149668455952005,
      "grad_norm": 5.582278728485107,
      "learning_rate": 4.6501999859658976e-05,
      "loss": 0.8596,
      "step": 2581
    },
    {
      "epoch": 0.8152826018313861,
      "grad_norm": 1.0581542253494263,
      "learning_rate": 4.6498491333941476e-05,
      "loss": 0.9357,
      "step": 2582
    },
    {
      "epoch": 0.8155983580675719,
      "grad_norm": 1.0528415441513062,
      "learning_rate": 4.649498280822398e-05,
      "loss": 0.8908,
      "step": 2583
    },
    {
      "epoch": 0.8159141143037575,
      "grad_norm": 0.9982864260673523,
      "learning_rate": 4.649147428250649e-05,
      "loss": 0.9289,
      "step": 2584
    },
    {
      "epoch": 0.8162298705399431,
      "grad_norm": 0.9403671622276306,
      "learning_rate": 4.6487965756789e-05,
      "loss": 0.8885,
      "step": 2585
    },
    {
      "epoch": 0.8165456267761289,
      "grad_norm": 0.9053142070770264,
      "learning_rate": 4.6484457231071504e-05,
      "loss": 0.9552,
      "step": 2586
    },
    {
      "epoch": 0.8168613830123145,
      "grad_norm": 1.4171735048294067,
      "learning_rate": 4.648094870535401e-05,
      "loss": 0.8865,
      "step": 2587
    },
    {
      "epoch": 0.8171771392485001,
      "grad_norm": 1.0008620023727417,
      "learning_rate": 4.647744017963652e-05,
      "loss": 0.9322,
      "step": 2588
    },
    {
      "epoch": 0.8174928954846858,
      "grad_norm": 0.9979502558708191,
      "learning_rate": 4.6473931653919024e-05,
      "loss": 0.8811,
      "step": 2589
    },
    {
      "epoch": 0.8178086517208715,
      "grad_norm": 1.1298319101333618,
      "learning_rate": 4.647042312820154e-05,
      "loss": 0.8868,
      "step": 2590
    },
    {
      "epoch": 0.8181244079570571,
      "grad_norm": 1.6279778480529785,
      "learning_rate": 4.646691460248404e-05,
      "loss": 0.9803,
      "step": 2591
    },
    {
      "epoch": 0.8184401641932428,
      "grad_norm": 1.140729308128357,
      "learning_rate": 4.6463406076766545e-05,
      "loss": 0.9141,
      "step": 2592
    },
    {
      "epoch": 0.8187559204294285,
      "grad_norm": 0.8756786584854126,
      "learning_rate": 4.645989755104905e-05,
      "loss": 0.9657,
      "step": 2593
    },
    {
      "epoch": 0.8190716766656142,
      "grad_norm": 1.2670650482177734,
      "learning_rate": 4.645638902533156e-05,
      "loss": 0.9739,
      "step": 2594
    },
    {
      "epoch": 0.8193874329017998,
      "grad_norm": 0.9150395393371582,
      "learning_rate": 4.6452880499614065e-05,
      "loss": 0.9094,
      "step": 2595
    },
    {
      "epoch": 0.8197031891379855,
      "grad_norm": 1.1084283590316772,
      "learning_rate": 4.644937197389657e-05,
      "loss": 0.8959,
      "step": 2596
    },
    {
      "epoch": 0.8200189453741712,
      "grad_norm": 1.084456443786621,
      "learning_rate": 4.644586344817908e-05,
      "loss": 0.9329,
      "step": 2597
    },
    {
      "epoch": 0.8203347016103568,
      "grad_norm": 0.942467987537384,
      "learning_rate": 4.6442354922461586e-05,
      "loss": 0.8942,
      "step": 2598
    },
    {
      "epoch": 0.8206504578465424,
      "grad_norm": 0.9653447866439819,
      "learning_rate": 4.643884639674409e-05,
      "loss": 0.9221,
      "step": 2599
    },
    {
      "epoch": 0.8209662140827282,
      "grad_norm": 0.9654855132102966,
      "learning_rate": 4.64353378710266e-05,
      "loss": 0.8994,
      "step": 2600
    },
    {
      "epoch": 0.8209662140827282,
      "eval_loss": 0.8691033720970154,
      "eval_runtime": 4.641,
      "eval_samples_per_second": 536.523,
      "eval_steps_per_second": 4.525,
      "step": 2600
    },
    {
      "epoch": 0.8212819703189138,
      "grad_norm": 1.148393988609314,
      "learning_rate": 4.64318293453091e-05,
      "loss": 0.9439,
      "step": 2601
    },
    {
      "epoch": 0.8215977265550994,
      "grad_norm": 1.2259352207183838,
      "learning_rate": 4.6428320819591606e-05,
      "loss": 0.9582,
      "step": 2602
    },
    {
      "epoch": 0.8219134827912852,
      "grad_norm": 1.1526163816452026,
      "learning_rate": 4.642481229387411e-05,
      "loss": 0.9439,
      "step": 2603
    },
    {
      "epoch": 0.8222292390274708,
      "grad_norm": 1.0290998220443726,
      "learning_rate": 4.642130376815662e-05,
      "loss": 0.8823,
      "step": 2604
    },
    {
      "epoch": 0.8225449952636564,
      "grad_norm": 1.434584617614746,
      "learning_rate": 4.6417795242439126e-05,
      "loss": 0.856,
      "step": 2605
    },
    {
      "epoch": 0.8228607514998422,
      "grad_norm": 1.0399680137634277,
      "learning_rate": 4.641428671672163e-05,
      "loss": 0.9783,
      "step": 2606
    },
    {
      "epoch": 0.8231765077360278,
      "grad_norm": 1.1417077779769897,
      "learning_rate": 4.641077819100415e-05,
      "loss": 0.9584,
      "step": 2607
    },
    {
      "epoch": 0.8234922639722134,
      "grad_norm": 1.3663848638534546,
      "learning_rate": 4.6407269665286654e-05,
      "loss": 0.8657,
      "step": 2608
    },
    {
      "epoch": 0.8238080202083992,
      "grad_norm": 1.1044368743896484,
      "learning_rate": 4.640376113956916e-05,
      "loss": 0.9043,
      "step": 2609
    },
    {
      "epoch": 0.8241237764445848,
      "grad_norm": 0.9005938768386841,
      "learning_rate": 4.640025261385166e-05,
      "loss": 0.8347,
      "step": 2610
    },
    {
      "epoch": 0.8244395326807704,
      "grad_norm": 1.214228630065918,
      "learning_rate": 4.639674408813417e-05,
      "loss": 0.9978,
      "step": 2611
    },
    {
      "epoch": 0.8247552889169562,
      "grad_norm": 1.2312196493148804,
      "learning_rate": 4.6393235562416674e-05,
      "loss": 0.9383,
      "step": 2612
    },
    {
      "epoch": 0.8250710451531418,
      "grad_norm": 1.0107920169830322,
      "learning_rate": 4.638972703669918e-05,
      "loss": 0.8734,
      "step": 2613
    },
    {
      "epoch": 0.8253868013893274,
      "grad_norm": 1.0658220052719116,
      "learning_rate": 4.638621851098169e-05,
      "loss": 0.8853,
      "step": 2614
    },
    {
      "epoch": 0.8257025576255131,
      "grad_norm": 1.272376298904419,
      "learning_rate": 4.6382709985264195e-05,
      "loss": 0.8995,
      "step": 2615
    },
    {
      "epoch": 0.8260183138616988,
      "grad_norm": 0.8516282439231873,
      "learning_rate": 4.63792014595467e-05,
      "loss": 0.8671,
      "step": 2616
    },
    {
      "epoch": 0.8263340700978844,
      "grad_norm": 1.039568543434143,
      "learning_rate": 4.637569293382921e-05,
      "loss": 0.9115,
      "step": 2617
    },
    {
      "epoch": 0.8266498263340701,
      "grad_norm": 1.122741460800171,
      "learning_rate": 4.6372184408111715e-05,
      "loss": 0.9018,
      "step": 2618
    },
    {
      "epoch": 0.8269655825702558,
      "grad_norm": 0.9005895256996155,
      "learning_rate": 4.636867588239422e-05,
      "loss": 0.9526,
      "step": 2619
    },
    {
      "epoch": 0.8272813388064414,
      "grad_norm": 1.053305983543396,
      "learning_rate": 4.636516735667673e-05,
      "loss": 0.9891,
      "step": 2620
    },
    {
      "epoch": 0.8275970950426271,
      "grad_norm": 0.8518990278244019,
      "learning_rate": 4.636165883095923e-05,
      "loss": 0.8884,
      "step": 2621
    },
    {
      "epoch": 0.8279128512788128,
      "grad_norm": 0.9512357711791992,
      "learning_rate": 4.6358150305241735e-05,
      "loss": 0.8996,
      "step": 2622
    },
    {
      "epoch": 0.8282286075149984,
      "grad_norm": 1.0468392372131348,
      "learning_rate": 4.635464177952424e-05,
      "loss": 0.9663,
      "step": 2623
    },
    {
      "epoch": 0.8285443637511841,
      "grad_norm": 0.9997020363807678,
      "learning_rate": 4.635113325380675e-05,
      "loss": 0.8608,
      "step": 2624
    },
    {
      "epoch": 0.8288601199873697,
      "grad_norm": 0.9637042880058289,
      "learning_rate": 4.634762472808926e-05,
      "loss": 0.8925,
      "step": 2625
    },
    {
      "epoch": 0.8291758762235554,
      "grad_norm": 1.0603419542312622,
      "learning_rate": 4.634411620237177e-05,
      "loss": 0.8862,
      "step": 2626
    },
    {
      "epoch": 0.8294916324597411,
      "grad_norm": 1.2362290620803833,
      "learning_rate": 4.6340607676654276e-05,
      "loss": 1.0152,
      "step": 2627
    },
    {
      "epoch": 0.8298073886959267,
      "grad_norm": 0.8998485803604126,
      "learning_rate": 4.633709915093678e-05,
      "loss": 0.8884,
      "step": 2628
    },
    {
      "epoch": 0.8301231449321124,
      "grad_norm": 1.1198145151138306,
      "learning_rate": 4.633359062521928e-05,
      "loss": 0.9115,
      "step": 2629
    },
    {
      "epoch": 0.8304389011682981,
      "grad_norm": 0.9187816381454468,
      "learning_rate": 4.633008209950179e-05,
      "loss": 0.8945,
      "step": 2630
    },
    {
      "epoch": 0.8307546574044837,
      "grad_norm": 1.0395992994308472,
      "learning_rate": 4.63265735737843e-05,
      "loss": 0.8983,
      "step": 2631
    },
    {
      "epoch": 0.8310704136406694,
      "grad_norm": 0.959064781665802,
      "learning_rate": 4.6323065048066804e-05,
      "loss": 0.8786,
      "step": 2632
    },
    {
      "epoch": 0.8313861698768551,
      "grad_norm": 0.9190137982368469,
      "learning_rate": 4.631955652234931e-05,
      "loss": 0.9228,
      "step": 2633
    },
    {
      "epoch": 0.8317019261130407,
      "grad_norm": 1.0952447652816772,
      "learning_rate": 4.631604799663182e-05,
      "loss": 0.9427,
      "step": 2634
    },
    {
      "epoch": 0.8320176823492264,
      "grad_norm": 1.0434576272964478,
      "learning_rate": 4.6312539470914324e-05,
      "loss": 0.8874,
      "step": 2635
    },
    {
      "epoch": 0.8323334385854121,
      "grad_norm": 1.0396827459335327,
      "learning_rate": 4.630903094519683e-05,
      "loss": 0.9191,
      "step": 2636
    },
    {
      "epoch": 0.8326491948215977,
      "grad_norm": 1.1930456161499023,
      "learning_rate": 4.630552241947934e-05,
      "loss": 0.9123,
      "step": 2637
    },
    {
      "epoch": 0.8329649510577833,
      "grad_norm": 1.0544923543930054,
      "learning_rate": 4.6302013893761844e-05,
      "loss": 0.9055,
      "step": 2638
    },
    {
      "epoch": 0.8332807072939691,
      "grad_norm": 1.0902093648910522,
      "learning_rate": 4.629850536804435e-05,
      "loss": 0.9816,
      "step": 2639
    },
    {
      "epoch": 0.8335964635301547,
      "grad_norm": 1.0289242267608643,
      "learning_rate": 4.629499684232685e-05,
      "loss": 0.9422,
      "step": 2640
    },
    {
      "epoch": 0.8339122197663403,
      "grad_norm": 1.0434802770614624,
      "learning_rate": 4.629148831660936e-05,
      "loss": 0.8999,
      "step": 2641
    },
    {
      "epoch": 0.8342279760025261,
      "grad_norm": 1.0047171115875244,
      "learning_rate": 4.628797979089187e-05,
      "loss": 0.8655,
      "step": 2642
    },
    {
      "epoch": 0.8345437322387117,
      "grad_norm": 1.0515347719192505,
      "learning_rate": 4.628447126517438e-05,
      "loss": 0.8731,
      "step": 2643
    },
    {
      "epoch": 0.8348594884748973,
      "grad_norm": 0.9243134260177612,
      "learning_rate": 4.6280962739456885e-05,
      "loss": 0.9234,
      "step": 2644
    },
    {
      "epoch": 0.8351752447110831,
      "grad_norm": 1.0212010145187378,
      "learning_rate": 4.627745421373939e-05,
      "loss": 0.8948,
      "step": 2645
    },
    {
      "epoch": 0.8354910009472687,
      "grad_norm": 1.5033270120620728,
      "learning_rate": 4.62739456880219e-05,
      "loss": 0.9117,
      "step": 2646
    },
    {
      "epoch": 0.8358067571834543,
      "grad_norm": 1.0221437215805054,
      "learning_rate": 4.6270437162304406e-05,
      "loss": 0.8884,
      "step": 2647
    },
    {
      "epoch": 0.8361225134196401,
      "grad_norm": 1.1300909519195557,
      "learning_rate": 4.626692863658691e-05,
      "loss": 0.8874,
      "step": 2648
    },
    {
      "epoch": 0.8364382696558257,
      "grad_norm": 1.2555543184280396,
      "learning_rate": 4.626342011086941e-05,
      "loss": 0.9374,
      "step": 2649
    },
    {
      "epoch": 0.8367540258920114,
      "grad_norm": 1.1265660524368286,
      "learning_rate": 4.625991158515192e-05,
      "loss": 0.9408,
      "step": 2650
    },
    {
      "epoch": 0.837069782128197,
      "grad_norm": 0.9728745818138123,
      "learning_rate": 4.6256403059434426e-05,
      "loss": 0.8987,
      "step": 2651
    },
    {
      "epoch": 0.8373855383643827,
      "grad_norm": 1.1680454015731812,
      "learning_rate": 4.625289453371693e-05,
      "loss": 0.9775,
      "step": 2652
    },
    {
      "epoch": 0.8377012946005684,
      "grad_norm": 0.9779984354972839,
      "learning_rate": 4.624938600799944e-05,
      "loss": 0.8579,
      "step": 2653
    },
    {
      "epoch": 0.838017050836754,
      "grad_norm": 0.9406771063804626,
      "learning_rate": 4.624587748228195e-05,
      "loss": 0.863,
      "step": 2654
    },
    {
      "epoch": 0.8383328070729397,
      "grad_norm": 0.978956401348114,
      "learning_rate": 4.6242368956564454e-05,
      "loss": 0.8359,
      "step": 2655
    },
    {
      "epoch": 0.8386485633091254,
      "grad_norm": 0.9994984865188599,
      "learning_rate": 4.623886043084696e-05,
      "loss": 0.9018,
      "step": 2656
    },
    {
      "epoch": 0.838964319545311,
      "grad_norm": 1.164543867111206,
      "learning_rate": 4.623535190512947e-05,
      "loss": 0.9085,
      "step": 2657
    },
    {
      "epoch": 0.8392800757814967,
      "grad_norm": 1.0497210025787354,
      "learning_rate": 4.6231843379411974e-05,
      "loss": 0.8893,
      "step": 2658
    },
    {
      "epoch": 0.8395958320176824,
      "grad_norm": 1.033876895904541,
      "learning_rate": 4.6228334853694474e-05,
      "loss": 0.8921,
      "step": 2659
    },
    {
      "epoch": 0.839911588253868,
      "grad_norm": 1.1162151098251343,
      "learning_rate": 4.622482632797699e-05,
      "loss": 0.8833,
      "step": 2660
    },
    {
      "epoch": 0.8402273444900537,
      "grad_norm": 0.8616299033164978,
      "learning_rate": 4.6221317802259494e-05,
      "loss": 0.9096,
      "step": 2661
    },
    {
      "epoch": 0.8405431007262394,
      "grad_norm": 1.0763897895812988,
      "learning_rate": 4.6217809276542e-05,
      "loss": 0.9331,
      "step": 2662
    },
    {
      "epoch": 0.840858856962425,
      "grad_norm": 1.0147262811660767,
      "learning_rate": 4.621430075082451e-05,
      "loss": 0.8451,
      "step": 2663
    },
    {
      "epoch": 0.8411746131986106,
      "grad_norm": 1.1386847496032715,
      "learning_rate": 4.6210792225107015e-05,
      "loss": 0.8636,
      "step": 2664
    },
    {
      "epoch": 0.8414903694347964,
      "grad_norm": 1.0315696001052856,
      "learning_rate": 4.620728369938952e-05,
      "loss": 0.8456,
      "step": 2665
    },
    {
      "epoch": 0.841806125670982,
      "grad_norm": 1.1792515516281128,
      "learning_rate": 4.620377517367203e-05,
      "loss": 0.8805,
      "step": 2666
    },
    {
      "epoch": 0.8421218819071676,
      "grad_norm": 1.3692259788513184,
      "learning_rate": 4.6200266647954535e-05,
      "loss": 0.9138,
      "step": 2667
    },
    {
      "epoch": 0.8424376381433534,
      "grad_norm": 1.1629782915115356,
      "learning_rate": 4.6196758122237035e-05,
      "loss": 0.8864,
      "step": 2668
    },
    {
      "epoch": 0.842753394379539,
      "grad_norm": 1.1382927894592285,
      "learning_rate": 4.619324959651954e-05,
      "loss": 0.9288,
      "step": 2669
    },
    {
      "epoch": 0.8430691506157246,
      "grad_norm": 1.064125895500183,
      "learning_rate": 4.618974107080205e-05,
      "loss": 0.8246,
      "step": 2670
    },
    {
      "epoch": 0.8433849068519104,
      "grad_norm": 1.0386189222335815,
      "learning_rate": 4.6186232545084556e-05,
      "loss": 0.9463,
      "step": 2671
    },
    {
      "epoch": 0.843700663088096,
      "grad_norm": 1.2270169258117676,
      "learning_rate": 4.618272401936706e-05,
      "loss": 0.9145,
      "step": 2672
    },
    {
      "epoch": 0.8440164193242816,
      "grad_norm": 1.2211166620254517,
      "learning_rate": 4.617921549364957e-05,
      "loss": 0.915,
      "step": 2673
    },
    {
      "epoch": 0.8443321755604674,
      "grad_norm": 1.41981840133667,
      "learning_rate": 4.6175706967932076e-05,
      "loss": 0.9686,
      "step": 2674
    },
    {
      "epoch": 0.844647931796653,
      "grad_norm": 1.2388707399368286,
      "learning_rate": 4.617219844221458e-05,
      "loss": 0.8826,
      "step": 2675
    },
    {
      "epoch": 0.8449636880328386,
      "grad_norm": 1.1144760847091675,
      "learning_rate": 4.616868991649709e-05,
      "loss": 0.8746,
      "step": 2676
    },
    {
      "epoch": 0.8452794442690243,
      "grad_norm": 1.4339210987091064,
      "learning_rate": 4.61651813907796e-05,
      "loss": 0.8772,
      "step": 2677
    },
    {
      "epoch": 0.84559520050521,
      "grad_norm": 1.2534421682357788,
      "learning_rate": 4.6161672865062103e-05,
      "loss": 0.8863,
      "step": 2678
    },
    {
      "epoch": 0.8459109567413956,
      "grad_norm": 1.0200146436691284,
      "learning_rate": 4.615816433934461e-05,
      "loss": 0.8927,
      "step": 2679
    },
    {
      "epoch": 0.8462267129775813,
      "grad_norm": 1.0400927066802979,
      "learning_rate": 4.615465581362712e-05,
      "loss": 0.8222,
      "step": 2680
    },
    {
      "epoch": 0.846542469213767,
      "grad_norm": 0.9867685437202454,
      "learning_rate": 4.6151147287909624e-05,
      "loss": 0.9174,
      "step": 2681
    },
    {
      "epoch": 0.8468582254499526,
      "grad_norm": 1.1033228635787964,
      "learning_rate": 4.614763876219213e-05,
      "loss": 0.8746,
      "step": 2682
    },
    {
      "epoch": 0.8471739816861383,
      "grad_norm": 1.1520752906799316,
      "learning_rate": 4.614413023647464e-05,
      "loss": 0.8791,
      "step": 2683
    },
    {
      "epoch": 0.847489737922324,
      "grad_norm": 1.0276257991790771,
      "learning_rate": 4.6140621710757144e-05,
      "loss": 0.9441,
      "step": 2684
    },
    {
      "epoch": 0.8478054941585096,
      "grad_norm": 1.1899372339248657,
      "learning_rate": 4.613711318503965e-05,
      "loss": 0.9434,
      "step": 2685
    },
    {
      "epoch": 0.8481212503946953,
      "grad_norm": 1.3945205211639404,
      "learning_rate": 4.613360465932216e-05,
      "loss": 0.8811,
      "step": 2686
    },
    {
      "epoch": 0.848437006630881,
      "grad_norm": 1.0523178577423096,
      "learning_rate": 4.613009613360466e-05,
      "loss": 0.8215,
      "step": 2687
    },
    {
      "epoch": 0.8487527628670666,
      "grad_norm": 1.234029769897461,
      "learning_rate": 4.6126587607887165e-05,
      "loss": 0.9398,
      "step": 2688
    },
    {
      "epoch": 0.8490685191032523,
      "grad_norm": 1.0734738111495972,
      "learning_rate": 4.612307908216967e-05,
      "loss": 0.9149,
      "step": 2689
    },
    {
      "epoch": 0.8493842753394379,
      "grad_norm": 0.9638886451721191,
      "learning_rate": 4.611957055645218e-05,
      "loss": 0.9255,
      "step": 2690
    },
    {
      "epoch": 0.8497000315756236,
      "grad_norm": 1.2488864660263062,
      "learning_rate": 4.6116062030734685e-05,
      "loss": 0.8927,
      "step": 2691
    },
    {
      "epoch": 0.8500157878118093,
      "grad_norm": 1.197392463684082,
      "learning_rate": 4.611255350501719e-05,
      "loss": 0.8829,
      "step": 2692
    },
    {
      "epoch": 0.8503315440479949,
      "grad_norm": 1.2193418741226196,
      "learning_rate": 4.61090449792997e-05,
      "loss": 0.8947,
      "step": 2693
    },
    {
      "epoch": 0.8506473002841806,
      "grad_norm": 1.1639360189437866,
      "learning_rate": 4.6105536453582206e-05,
      "loss": 0.9251,
      "step": 2694
    },
    {
      "epoch": 0.8509630565203663,
      "grad_norm": 1.5056523084640503,
      "learning_rate": 4.610202792786472e-05,
      "loss": 0.9312,
      "step": 2695
    },
    {
      "epoch": 0.8512788127565519,
      "grad_norm": 1.1038801670074463,
      "learning_rate": 4.609851940214722e-05,
      "loss": 0.9218,
      "step": 2696
    },
    {
      "epoch": 0.8515945689927376,
      "grad_norm": 1.339214563369751,
      "learning_rate": 4.6095010876429726e-05,
      "loss": 0.9577,
      "step": 2697
    },
    {
      "epoch": 0.8519103252289233,
      "grad_norm": 1.0129718780517578,
      "learning_rate": 4.609150235071223e-05,
      "loss": 0.8734,
      "step": 2698
    },
    {
      "epoch": 0.8522260814651089,
      "grad_norm": 1.1429089307785034,
      "learning_rate": 4.608799382499474e-05,
      "loss": 0.9261,
      "step": 2699
    },
    {
      "epoch": 0.8525418377012945,
      "grad_norm": 0.984790563583374,
      "learning_rate": 4.6084485299277247e-05,
      "loss": 0.9163,
      "step": 2700
    },
    {
      "epoch": 0.8528575939374803,
      "grad_norm": 0.9304472804069519,
      "learning_rate": 4.608097677355975e-05,
      "loss": 0.8659,
      "step": 2701
    },
    {
      "epoch": 0.8531733501736659,
      "grad_norm": 0.9183005094528198,
      "learning_rate": 4.607746824784226e-05,
      "loss": 0.9349,
      "step": 2702
    },
    {
      "epoch": 0.8534891064098516,
      "grad_norm": 1.0505826473236084,
      "learning_rate": 4.607395972212477e-05,
      "loss": 0.902,
      "step": 2703
    },
    {
      "epoch": 0.8538048626460373,
      "grad_norm": 1.157343864440918,
      "learning_rate": 4.6070451196407274e-05,
      "loss": 0.9514,
      "step": 2704
    },
    {
      "epoch": 0.8541206188822229,
      "grad_norm": 1.1378198862075806,
      "learning_rate": 4.606694267068978e-05,
      "loss": 0.8815,
      "step": 2705
    },
    {
      "epoch": 0.8544363751184086,
      "grad_norm": 1.0765589475631714,
      "learning_rate": 4.606343414497228e-05,
      "loss": 0.9184,
      "step": 2706
    },
    {
      "epoch": 0.8547521313545943,
      "grad_norm": 1.2196310758590698,
      "learning_rate": 4.605992561925479e-05,
      "loss": 0.9012,
      "step": 2707
    },
    {
      "epoch": 0.8550678875907799,
      "grad_norm": 1.2838129997253418,
      "learning_rate": 4.6056417093537294e-05,
      "loss": 0.9348,
      "step": 2708
    },
    {
      "epoch": 0.8553836438269656,
      "grad_norm": 1.0261914730072021,
      "learning_rate": 4.60529085678198e-05,
      "loss": 0.8934,
      "step": 2709
    },
    {
      "epoch": 0.8556994000631513,
      "grad_norm": 0.9993560314178467,
      "learning_rate": 4.604940004210231e-05,
      "loss": 0.9618,
      "step": 2710
    },
    {
      "epoch": 0.8560151562993369,
      "grad_norm": 1.1377321481704712,
      "learning_rate": 4.6045891516384815e-05,
      "loss": 0.8944,
      "step": 2711
    },
    {
      "epoch": 0.8563309125355226,
      "grad_norm": 1.01433527469635,
      "learning_rate": 4.604238299066733e-05,
      "loss": 0.9092,
      "step": 2712
    },
    {
      "epoch": 0.8566466687717083,
      "grad_norm": 1.08974289894104,
      "learning_rate": 4.6038874464949835e-05,
      "loss": 0.9156,
      "step": 2713
    },
    {
      "epoch": 0.8569624250078939,
      "grad_norm": 1.1416343450546265,
      "learning_rate": 4.603536593923234e-05,
      "loss": 0.887,
      "step": 2714
    },
    {
      "epoch": 0.8572781812440796,
      "grad_norm": 1.1243611574172974,
      "learning_rate": 4.603185741351484e-05,
      "loss": 0.8792,
      "step": 2715
    },
    {
      "epoch": 0.8575939374802652,
      "grad_norm": 1.3561675548553467,
      "learning_rate": 4.602834888779735e-05,
      "loss": 0.9184,
      "step": 2716
    },
    {
      "epoch": 0.8579096937164509,
      "grad_norm": 0.9882817268371582,
      "learning_rate": 4.6024840362079856e-05,
      "loss": 0.9094,
      "step": 2717
    },
    {
      "epoch": 0.8582254499526366,
      "grad_norm": 1.0638346672058105,
      "learning_rate": 4.602133183636236e-05,
      "loss": 0.9241,
      "step": 2718
    },
    {
      "epoch": 0.8585412061888222,
      "grad_norm": 0.9335153698921204,
      "learning_rate": 4.601782331064487e-05,
      "loss": 0.8663,
      "step": 2719
    },
    {
      "epoch": 0.8588569624250079,
      "grad_norm": 0.9449256062507629,
      "learning_rate": 4.6014314784927376e-05,
      "loss": 0.9483,
      "step": 2720
    },
    {
      "epoch": 0.8591727186611936,
      "grad_norm": 0.9933111667633057,
      "learning_rate": 4.601080625920988e-05,
      "loss": 0.9106,
      "step": 2721
    },
    {
      "epoch": 0.8594884748973792,
      "grad_norm": 1.0517160892486572,
      "learning_rate": 4.600729773349239e-05,
      "loss": 0.9005,
      "step": 2722
    },
    {
      "epoch": 0.8598042311335649,
      "grad_norm": 1.591367244720459,
      "learning_rate": 4.6003789207774896e-05,
      "loss": 0.9332,
      "step": 2723
    },
    {
      "epoch": 0.8601199873697506,
      "grad_norm": 1.2593306303024292,
      "learning_rate": 4.60002806820574e-05,
      "loss": 0.9648,
      "step": 2724
    },
    {
      "epoch": 0.8604357436059362,
      "grad_norm": 1.0459985733032227,
      "learning_rate": 4.599677215633991e-05,
      "loss": 0.858,
      "step": 2725
    },
    {
      "epoch": 0.8607514998421218,
      "grad_norm": 0.9544197916984558,
      "learning_rate": 4.599326363062241e-05,
      "loss": 0.9418,
      "step": 2726
    },
    {
      "epoch": 0.8610672560783076,
      "grad_norm": 0.9784119725227356,
      "learning_rate": 4.598975510490492e-05,
      "loss": 0.9665,
      "step": 2727
    },
    {
      "epoch": 0.8613830123144932,
      "grad_norm": 0.9791545867919922,
      "learning_rate": 4.5986246579187424e-05,
      "loss": 0.9124,
      "step": 2728
    },
    {
      "epoch": 0.8616987685506788,
      "grad_norm": 1.0486890077590942,
      "learning_rate": 4.598273805346993e-05,
      "loss": 0.9635,
      "step": 2729
    },
    {
      "epoch": 0.8620145247868646,
      "grad_norm": 0.964497983455658,
      "learning_rate": 4.5979229527752444e-05,
      "loss": 0.9037,
      "step": 2730
    },
    {
      "epoch": 0.8623302810230502,
      "grad_norm": 1.1791353225708008,
      "learning_rate": 4.597572100203495e-05,
      "loss": 0.9282,
      "step": 2731
    },
    {
      "epoch": 0.8626460372592358,
      "grad_norm": 0.9775570631027222,
      "learning_rate": 4.597221247631746e-05,
      "loss": 0.8945,
      "step": 2732
    },
    {
      "epoch": 0.8629617934954216,
      "grad_norm": 0.9031182527542114,
      "learning_rate": 4.5968703950599965e-05,
      "loss": 0.838,
      "step": 2733
    },
    {
      "epoch": 0.8632775497316072,
      "grad_norm": 0.92740797996521,
      "learning_rate": 4.5965195424882465e-05,
      "loss": 0.9239,
      "step": 2734
    },
    {
      "epoch": 0.8635933059677928,
      "grad_norm": 1.0481830835342407,
      "learning_rate": 4.596168689916497e-05,
      "loss": 0.9145,
      "step": 2735
    },
    {
      "epoch": 0.8639090622039786,
      "grad_norm": 1.0042822360992432,
      "learning_rate": 4.595817837344748e-05,
      "loss": 0.8516,
      "step": 2736
    },
    {
      "epoch": 0.8642248184401642,
      "grad_norm": 0.850605845451355,
      "learning_rate": 4.5954669847729985e-05,
      "loss": 0.86,
      "step": 2737
    },
    {
      "epoch": 0.8645405746763498,
      "grad_norm": 0.9891607165336609,
      "learning_rate": 4.595116132201249e-05,
      "loss": 0.9032,
      "step": 2738
    },
    {
      "epoch": 0.8648563309125356,
      "grad_norm": 0.8862621188163757,
      "learning_rate": 4.5947652796295e-05,
      "loss": 0.8521,
      "step": 2739
    },
    {
      "epoch": 0.8651720871487212,
      "grad_norm": 0.9723474979400635,
      "learning_rate": 4.5944144270577506e-05,
      "loss": 0.8959,
      "step": 2740
    },
    {
      "epoch": 0.8654878433849068,
      "grad_norm": 1.7832742929458618,
      "learning_rate": 4.594063574486001e-05,
      "loss": 0.901,
      "step": 2741
    },
    {
      "epoch": 0.8658035996210925,
      "grad_norm": 1.1447393894195557,
      "learning_rate": 4.593712721914252e-05,
      "loss": 0.8971,
      "step": 2742
    },
    {
      "epoch": 0.8661193558572782,
      "grad_norm": 1.3289326429367065,
      "learning_rate": 4.5933618693425026e-05,
      "loss": 0.9372,
      "step": 2743
    },
    {
      "epoch": 0.8664351120934638,
      "grad_norm": 1.2291117906570435,
      "learning_rate": 4.593011016770753e-05,
      "loss": 0.8676,
      "step": 2744
    },
    {
      "epoch": 0.8667508683296495,
      "grad_norm": 1.1375432014465332,
      "learning_rate": 4.592660164199003e-05,
      "loss": 0.8296,
      "step": 2745
    },
    {
      "epoch": 0.8670666245658352,
      "grad_norm": 0.9583473801612854,
      "learning_rate": 4.592309311627254e-05,
      "loss": 0.9346,
      "step": 2746
    },
    {
      "epoch": 0.8673823808020208,
      "grad_norm": 1.2570602893829346,
      "learning_rate": 4.591958459055505e-05,
      "loss": 0.8403,
      "step": 2747
    },
    {
      "epoch": 0.8676981370382065,
      "grad_norm": 1.1703839302062988,
      "learning_rate": 4.591607606483756e-05,
      "loss": 0.8838,
      "step": 2748
    },
    {
      "epoch": 0.8680138932743922,
      "grad_norm": 1.1967567205429077,
      "learning_rate": 4.591256753912007e-05,
      "loss": 0.8983,
      "step": 2749
    },
    {
      "epoch": 0.8683296495105778,
      "grad_norm": 1.0260295867919922,
      "learning_rate": 4.5909059013402574e-05,
      "loss": 0.8959,
      "step": 2750
    },
    {
      "epoch": 0.8686454057467635,
      "grad_norm": 0.9351486563682556,
      "learning_rate": 4.590555048768508e-05,
      "loss": 0.9106,
      "step": 2751
    },
    {
      "epoch": 0.8689611619829491,
      "grad_norm": 0.9513700008392334,
      "learning_rate": 4.590204196196759e-05,
      "loss": 0.8546,
      "step": 2752
    },
    {
      "epoch": 0.8692769182191348,
      "grad_norm": 0.9411612153053284,
      "learning_rate": 4.5898533436250094e-05,
      "loss": 0.8724,
      "step": 2753
    },
    {
      "epoch": 0.8695926744553205,
      "grad_norm": 0.9388648271560669,
      "learning_rate": 4.5895024910532594e-05,
      "loss": 0.8588,
      "step": 2754
    },
    {
      "epoch": 0.8699084306915061,
      "grad_norm": 2.4780027866363525,
      "learning_rate": 4.58915163848151e-05,
      "loss": 0.9914,
      "step": 2755
    },
    {
      "epoch": 0.8702241869276918,
      "grad_norm": 1.13666570186615,
      "learning_rate": 4.588800785909761e-05,
      "loss": 0.9428,
      "step": 2756
    },
    {
      "epoch": 0.8705399431638775,
      "grad_norm": 1.1147249937057495,
      "learning_rate": 4.5884499333380115e-05,
      "loss": 0.9689,
      "step": 2757
    },
    {
      "epoch": 0.8708556994000631,
      "grad_norm": 1.4276193380355835,
      "learning_rate": 4.588099080766262e-05,
      "loss": 0.8915,
      "step": 2758
    },
    {
      "epoch": 0.8711714556362489,
      "grad_norm": 1.0388157367706299,
      "learning_rate": 4.587748228194513e-05,
      "loss": 0.89,
      "step": 2759
    },
    {
      "epoch": 0.8714872118724345,
      "grad_norm": 0.982055127620697,
      "learning_rate": 4.5873973756227635e-05,
      "loss": 0.9786,
      "step": 2760
    },
    {
      "epoch": 0.8718029681086201,
      "grad_norm": 1.5654001235961914,
      "learning_rate": 4.587046523051014e-05,
      "loss": 0.9336,
      "step": 2761
    },
    {
      "epoch": 0.8721187243448059,
      "grad_norm": 0.9803472757339478,
      "learning_rate": 4.586695670479265e-05,
      "loss": 0.9613,
      "step": 2762
    },
    {
      "epoch": 0.8724344805809915,
      "grad_norm": 1.1245726346969604,
      "learning_rate": 4.5863448179075155e-05,
      "loss": 0.9215,
      "step": 2763
    },
    {
      "epoch": 0.8727502368171771,
      "grad_norm": 1.4704391956329346,
      "learning_rate": 4.5859939653357655e-05,
      "loss": 0.8511,
      "step": 2764
    },
    {
      "epoch": 0.8730659930533629,
      "grad_norm": 3.2456471920013428,
      "learning_rate": 4.585643112764017e-05,
      "loss": 0.9492,
      "step": 2765
    },
    {
      "epoch": 0.8733817492895485,
      "grad_norm": 1.0487864017486572,
      "learning_rate": 4.5852922601922676e-05,
      "loss": 0.8593,
      "step": 2766
    },
    {
      "epoch": 0.8736975055257341,
      "grad_norm": 1.0133205652236938,
      "learning_rate": 4.584941407620518e-05,
      "loss": 0.9373,
      "step": 2767
    },
    {
      "epoch": 0.8740132617619198,
      "grad_norm": 1.0536540746688843,
      "learning_rate": 4.584590555048769e-05,
      "loss": 0.9349,
      "step": 2768
    },
    {
      "epoch": 0.8743290179981055,
      "grad_norm": 1.3426727056503296,
      "learning_rate": 4.5842397024770196e-05,
      "loss": 0.9312,
      "step": 2769
    },
    {
      "epoch": 0.8746447742342911,
      "grad_norm": 1.088197946548462,
      "learning_rate": 4.58388884990527e-05,
      "loss": 0.9169,
      "step": 2770
    },
    {
      "epoch": 0.8749605304704768,
      "grad_norm": 1.0485631227493286,
      "learning_rate": 4.583537997333521e-05,
      "loss": 0.9052,
      "step": 2771
    },
    {
      "epoch": 0.8752762867066625,
      "grad_norm": 1.2327253818511963,
      "learning_rate": 4.583187144761772e-05,
      "loss": 0.8832,
      "step": 2772
    },
    {
      "epoch": 0.8755920429428481,
      "grad_norm": 1.0798962116241455,
      "learning_rate": 4.582836292190022e-05,
      "loss": 0.8151,
      "step": 2773
    },
    {
      "epoch": 0.8759077991790338,
      "grad_norm": 0.938936710357666,
      "learning_rate": 4.5824854396182724e-05,
      "loss": 0.8914,
      "step": 2774
    },
    {
      "epoch": 0.8762235554152195,
      "grad_norm": 1.1588704586029053,
      "learning_rate": 4.582134587046523e-05,
      "loss": 0.895,
      "step": 2775
    },
    {
      "epoch": 0.8765393116514051,
      "grad_norm": 1.025780439376831,
      "learning_rate": 4.581783734474774e-05,
      "loss": 0.942,
      "step": 2776
    },
    {
      "epoch": 0.8768550678875908,
      "grad_norm": 0.874127209186554,
      "learning_rate": 4.5814328819030244e-05,
      "loss": 0.9138,
      "step": 2777
    },
    {
      "epoch": 0.8771708241237764,
      "grad_norm": 0.9808430671691895,
      "learning_rate": 4.581082029331275e-05,
      "loss": 0.8771,
      "step": 2778
    },
    {
      "epoch": 0.8774865803599621,
      "grad_norm": 0.8027412295341492,
      "learning_rate": 4.580731176759526e-05,
      "loss": 0.8488,
      "step": 2779
    },
    {
      "epoch": 0.8778023365961478,
      "grad_norm": 1.0361876487731934,
      "learning_rate": 4.5803803241877764e-05,
      "loss": 0.9763,
      "step": 2780
    },
    {
      "epoch": 0.8781180928323334,
      "grad_norm": 0.7975608706474304,
      "learning_rate": 4.580029471616027e-05,
      "loss": 0.9069,
      "step": 2781
    },
    {
      "epoch": 0.8784338490685191,
      "grad_norm": 1.168525218963623,
      "learning_rate": 4.579678619044278e-05,
      "loss": 1.0029,
      "step": 2782
    },
    {
      "epoch": 0.8787496053047048,
      "grad_norm": 1.0104286670684814,
      "learning_rate": 4.5793277664725285e-05,
      "loss": 0.8674,
      "step": 2783
    },
    {
      "epoch": 0.8790653615408904,
      "grad_norm": 0.824644148349762,
      "learning_rate": 4.578976913900779e-05,
      "loss": 0.8871,
      "step": 2784
    },
    {
      "epoch": 0.8793811177770761,
      "grad_norm": 1.2129366397857666,
      "learning_rate": 4.57862606132903e-05,
      "loss": 0.9374,
      "step": 2785
    },
    {
      "epoch": 0.8796968740132618,
      "grad_norm": 1.065083622932434,
      "learning_rate": 4.5782752087572805e-05,
      "loss": 0.8563,
      "step": 2786
    },
    {
      "epoch": 0.8800126302494474,
      "grad_norm": 0.921405017375946,
      "learning_rate": 4.577924356185531e-05,
      "loss": 0.9006,
      "step": 2787
    },
    {
      "epoch": 0.880328386485633,
      "grad_norm": 1.0205297470092773,
      "learning_rate": 4.577573503613782e-05,
      "loss": 0.9266,
      "step": 2788
    },
    {
      "epoch": 0.8806441427218188,
      "grad_norm": 1.1873255968093872,
      "learning_rate": 4.5772226510420326e-05,
      "loss": 0.9181,
      "step": 2789
    },
    {
      "epoch": 0.8809598989580044,
      "grad_norm": 1.10094153881073,
      "learning_rate": 4.576871798470283e-05,
      "loss": 0.9035,
      "step": 2790
    },
    {
      "epoch": 0.88127565519419,
      "grad_norm": 1.0131163597106934,
      "learning_rate": 4.576520945898534e-05,
      "loss": 0.868,
      "step": 2791
    },
    {
      "epoch": 0.8815914114303758,
      "grad_norm": 1.0100510120391846,
      "learning_rate": 4.576170093326784e-05,
      "loss": 0.9694,
      "step": 2792
    },
    {
      "epoch": 0.8819071676665614,
      "grad_norm": 1.9127845764160156,
      "learning_rate": 4.5758192407550346e-05,
      "loss": 0.9354,
      "step": 2793
    },
    {
      "epoch": 0.882222923902747,
      "grad_norm": 0.929604172706604,
      "learning_rate": 4.575468388183285e-05,
      "loss": 0.8055,
      "step": 2794
    },
    {
      "epoch": 0.8825386801389328,
      "grad_norm": 1.1656545400619507,
      "learning_rate": 4.575117535611536e-05,
      "loss": 0.9307,
      "step": 2795
    },
    {
      "epoch": 0.8828544363751184,
      "grad_norm": 1.099705696105957,
      "learning_rate": 4.574766683039787e-05,
      "loss": 0.9084,
      "step": 2796
    },
    {
      "epoch": 0.883170192611304,
      "grad_norm": 0.997499406337738,
      "learning_rate": 4.5744158304680374e-05,
      "loss": 0.8667,
      "step": 2797
    },
    {
      "epoch": 0.8834859488474898,
      "grad_norm": 0.9263787865638733,
      "learning_rate": 4.574064977896288e-05,
      "loss": 0.9054,
      "step": 2798
    },
    {
      "epoch": 0.8838017050836754,
      "grad_norm": 1.0682822465896606,
      "learning_rate": 4.573714125324539e-05,
      "loss": 0.8584,
      "step": 2799
    },
    {
      "epoch": 0.884117461319861,
      "grad_norm": 0.9610905647277832,
      "learning_rate": 4.57336327275279e-05,
      "loss": 0.9281,
      "step": 2800
    },
    {
      "epoch": 0.884117461319861,
      "eval_loss": 0.8553282618522644,
      "eval_runtime": 4.6777,
      "eval_samples_per_second": 532.315,
      "eval_steps_per_second": 4.489,
      "step": 2800
    },
    {
      "epoch": 0.8844332175560468,
      "grad_norm": 1.0474849939346313,
      "learning_rate": 4.57301242018104e-05,
      "loss": 0.8801,
      "step": 2801
    },
    {
      "epoch": 0.8847489737922324,
      "grad_norm": 1.0202239751815796,
      "learning_rate": 4.572661567609291e-05,
      "loss": 0.8829,
      "step": 2802
    },
    {
      "epoch": 0.885064730028418,
      "grad_norm": 1.3996473550796509,
      "learning_rate": 4.5723107150375414e-05,
      "loss": 0.9131,
      "step": 2803
    },
    {
      "epoch": 0.8853804862646037,
      "grad_norm": 1.2365909814834595,
      "learning_rate": 4.571959862465792e-05,
      "loss": 0.9306,
      "step": 2804
    },
    {
      "epoch": 0.8856962425007894,
      "grad_norm": 1.0441761016845703,
      "learning_rate": 4.571609009894043e-05,
      "loss": 0.9026,
      "step": 2805
    },
    {
      "epoch": 0.886011998736975,
      "grad_norm": 1.050602674484253,
      "learning_rate": 4.5712581573222935e-05,
      "loss": 0.9048,
      "step": 2806
    },
    {
      "epoch": 0.8863277549731607,
      "grad_norm": 1.3812052011489868,
      "learning_rate": 4.570907304750544e-05,
      "loss": 0.899,
      "step": 2807
    },
    {
      "epoch": 0.8866435112093464,
      "grad_norm": 0.9874771237373352,
      "learning_rate": 4.570556452178795e-05,
      "loss": 0.9507,
      "step": 2808
    },
    {
      "epoch": 0.886959267445532,
      "grad_norm": 0.9383509755134583,
      "learning_rate": 4.5702055996070455e-05,
      "loss": 0.9066,
      "step": 2809
    },
    {
      "epoch": 0.8872750236817177,
      "grad_norm": 1.0779870748519897,
      "learning_rate": 4.569854747035296e-05,
      "loss": 0.8868,
      "step": 2810
    },
    {
      "epoch": 0.8875907799179034,
      "grad_norm": 1.0327861309051514,
      "learning_rate": 4.569503894463546e-05,
      "loss": 0.8568,
      "step": 2811
    },
    {
      "epoch": 0.887906536154089,
      "grad_norm": 1.2771103382110596,
      "learning_rate": 4.569153041891797e-05,
      "loss": 0.9681,
      "step": 2812
    },
    {
      "epoch": 0.8882222923902747,
      "grad_norm": 1.0694422721862793,
      "learning_rate": 4.5688021893200476e-05,
      "loss": 0.9173,
      "step": 2813
    },
    {
      "epoch": 0.8885380486264604,
      "grad_norm": 1.08661949634552,
      "learning_rate": 4.568451336748298e-05,
      "loss": 0.9139,
      "step": 2814
    },
    {
      "epoch": 0.8888538048626461,
      "grad_norm": 1.1186068058013916,
      "learning_rate": 4.568100484176549e-05,
      "loss": 0.9595,
      "step": 2815
    },
    {
      "epoch": 0.8891695610988317,
      "grad_norm": 1.415155291557312,
      "learning_rate": 4.5677496316047996e-05,
      "loss": 0.908,
      "step": 2816
    },
    {
      "epoch": 0.8894853173350173,
      "grad_norm": 1.2370996475219727,
      "learning_rate": 4.567398779033051e-05,
      "loss": 0.8652,
      "step": 2817
    },
    {
      "epoch": 0.8898010735712031,
      "grad_norm": 0.9961395263671875,
      "learning_rate": 4.5670479264613017e-05,
      "loss": 0.8832,
      "step": 2818
    },
    {
      "epoch": 0.8901168298073887,
      "grad_norm": 1.348283052444458,
      "learning_rate": 4.5666970738895523e-05,
      "loss": 0.9201,
      "step": 2819
    },
    {
      "epoch": 0.8904325860435743,
      "grad_norm": 1.6682292222976685,
      "learning_rate": 4.5663462213178023e-05,
      "loss": 0.9127,
      "step": 2820
    },
    {
      "epoch": 0.8907483422797601,
      "grad_norm": 1.0839049816131592,
      "learning_rate": 4.565995368746053e-05,
      "loss": 0.9671,
      "step": 2821
    },
    {
      "epoch": 0.8910640985159457,
      "grad_norm": 1.961344838142395,
      "learning_rate": 4.565644516174304e-05,
      "loss": 0.8512,
      "step": 2822
    },
    {
      "epoch": 0.8913798547521313,
      "grad_norm": 1.1335396766662598,
      "learning_rate": 4.5652936636025544e-05,
      "loss": 0.8957,
      "step": 2823
    },
    {
      "epoch": 0.8916956109883171,
      "grad_norm": 1.0924997329711914,
      "learning_rate": 4.564942811030805e-05,
      "loss": 0.9474,
      "step": 2824
    },
    {
      "epoch": 0.8920113672245027,
      "grad_norm": 0.9786571264266968,
      "learning_rate": 4.564591958459056e-05,
      "loss": 0.9543,
      "step": 2825
    },
    {
      "epoch": 0.8923271234606883,
      "grad_norm": 1.1424216032028198,
      "learning_rate": 4.5642411058873064e-05,
      "loss": 0.9619,
      "step": 2826
    },
    {
      "epoch": 0.8926428796968741,
      "grad_norm": 1.1934258937835693,
      "learning_rate": 4.563890253315557e-05,
      "loss": 0.9768,
      "step": 2827
    },
    {
      "epoch": 0.8929586359330597,
      "grad_norm": 1.0498046875,
      "learning_rate": 4.563539400743808e-05,
      "loss": 0.939,
      "step": 2828
    },
    {
      "epoch": 0.8932743921692453,
      "grad_norm": 0.974007785320282,
      "learning_rate": 4.5631885481720585e-05,
      "loss": 0.8879,
      "step": 2829
    },
    {
      "epoch": 0.893590148405431,
      "grad_norm": 0.8048463463783264,
      "learning_rate": 4.562837695600309e-05,
      "loss": 0.8799,
      "step": 2830
    },
    {
      "epoch": 0.8939059046416167,
      "grad_norm": 1.0105617046356201,
      "learning_rate": 4.562486843028559e-05,
      "loss": 0.8759,
      "step": 2831
    },
    {
      "epoch": 0.8942216608778023,
      "grad_norm": 1.1063039302825928,
      "learning_rate": 4.56213599045681e-05,
      "loss": 0.9196,
      "step": 2832
    },
    {
      "epoch": 0.894537417113988,
      "grad_norm": 1.06602144241333,
      "learning_rate": 4.5617851378850605e-05,
      "loss": 0.9367,
      "step": 2833
    },
    {
      "epoch": 0.8948531733501737,
      "grad_norm": 0.9628061056137085,
      "learning_rate": 4.561434285313311e-05,
      "loss": 0.9327,
      "step": 2834
    },
    {
      "epoch": 0.8951689295863593,
      "grad_norm": 0.9874778985977173,
      "learning_rate": 4.5610834327415626e-05,
      "loss": 0.9333,
      "step": 2835
    },
    {
      "epoch": 0.895484685822545,
      "grad_norm": 0.9386575818061829,
      "learning_rate": 4.560732580169813e-05,
      "loss": 0.9621,
      "step": 2836
    },
    {
      "epoch": 0.8958004420587307,
      "grad_norm": 0.9534977674484253,
      "learning_rate": 4.560381727598064e-05,
      "loss": 0.8658,
      "step": 2837
    },
    {
      "epoch": 0.8961161982949163,
      "grad_norm": 1.180923581123352,
      "learning_rate": 4.5600308750263146e-05,
      "loss": 0.8409,
      "step": 2838
    },
    {
      "epoch": 0.896431954531102,
      "grad_norm": 0.8690611124038696,
      "learning_rate": 4.5596800224545646e-05,
      "loss": 0.8259,
      "step": 2839
    },
    {
      "epoch": 0.8967477107672877,
      "grad_norm": 0.9519550800323486,
      "learning_rate": 4.559329169882815e-05,
      "loss": 0.8884,
      "step": 2840
    },
    {
      "epoch": 0.8970634670034733,
      "grad_norm": 1.0141440629959106,
      "learning_rate": 4.558978317311066e-05,
      "loss": 0.8401,
      "step": 2841
    },
    {
      "epoch": 0.897379223239659,
      "grad_norm": 1.0405882596969604,
      "learning_rate": 4.5586274647393167e-05,
      "loss": 0.9282,
      "step": 2842
    },
    {
      "epoch": 0.8976949794758446,
      "grad_norm": 1.382893681526184,
      "learning_rate": 4.558276612167567e-05,
      "loss": 0.9051,
      "step": 2843
    },
    {
      "epoch": 0.8980107357120303,
      "grad_norm": 1.126043677330017,
      "learning_rate": 4.557925759595818e-05,
      "loss": 0.9075,
      "step": 2844
    },
    {
      "epoch": 0.898326491948216,
      "grad_norm": 1.069599986076355,
      "learning_rate": 4.557574907024069e-05,
      "loss": 0.8724,
      "step": 2845
    },
    {
      "epoch": 0.8986422481844016,
      "grad_norm": 0.9448163509368896,
      "learning_rate": 4.5572240544523194e-05,
      "loss": 0.8877,
      "step": 2846
    },
    {
      "epoch": 0.8989580044205873,
      "grad_norm": 0.9394228458404541,
      "learning_rate": 4.55687320188057e-05,
      "loss": 0.8141,
      "step": 2847
    },
    {
      "epoch": 0.899273760656773,
      "grad_norm": 1.119927167892456,
      "learning_rate": 4.556522349308821e-05,
      "loss": 0.9523,
      "step": 2848
    },
    {
      "epoch": 0.8995895168929586,
      "grad_norm": 0.9912116527557373,
      "learning_rate": 4.5561714967370714e-05,
      "loss": 0.935,
      "step": 2849
    },
    {
      "epoch": 0.8999052731291443,
      "grad_norm": 1.3057783842086792,
      "learning_rate": 4.5558206441653214e-05,
      "loss": 0.944,
      "step": 2850
    },
    {
      "epoch": 0.90022102936533,
      "grad_norm": 1.0897692441940308,
      "learning_rate": 4.555469791593572e-05,
      "loss": 0.8929,
      "step": 2851
    },
    {
      "epoch": 0.9005367856015156,
      "grad_norm": 1.0334997177124023,
      "learning_rate": 4.5551189390218235e-05,
      "loss": 0.8754,
      "step": 2852
    },
    {
      "epoch": 0.9008525418377012,
      "grad_norm": 1.0285969972610474,
      "learning_rate": 4.554768086450074e-05,
      "loss": 0.9168,
      "step": 2853
    },
    {
      "epoch": 0.901168298073887,
      "grad_norm": 1.1738855838775635,
      "learning_rate": 4.554417233878325e-05,
      "loss": 0.8563,
      "step": 2854
    },
    {
      "epoch": 0.9014840543100726,
      "grad_norm": 1.183708906173706,
      "learning_rate": 4.5540663813065755e-05,
      "loss": 0.9412,
      "step": 2855
    },
    {
      "epoch": 0.9017998105462582,
      "grad_norm": 1.0404397249221802,
      "learning_rate": 4.553715528734826e-05,
      "loss": 0.8811,
      "step": 2856
    },
    {
      "epoch": 0.902115566782444,
      "grad_norm": 1.079374074935913,
      "learning_rate": 4.553364676163077e-05,
      "loss": 0.8777,
      "step": 2857
    },
    {
      "epoch": 0.9024313230186296,
      "grad_norm": 0.9689428210258484,
      "learning_rate": 4.5530138235913276e-05,
      "loss": 0.9225,
      "step": 2858
    },
    {
      "epoch": 0.9027470792548152,
      "grad_norm": 1.041748285293579,
      "learning_rate": 4.5526629710195776e-05,
      "loss": 0.9368,
      "step": 2859
    },
    {
      "epoch": 0.903062835491001,
      "grad_norm": 0.9246417284011841,
      "learning_rate": 4.552312118447828e-05,
      "loss": 0.8845,
      "step": 2860
    },
    {
      "epoch": 0.9033785917271866,
      "grad_norm": 1.022026538848877,
      "learning_rate": 4.551961265876079e-05,
      "loss": 0.9194,
      "step": 2861
    },
    {
      "epoch": 0.9036943479633722,
      "grad_norm": 0.8084893822669983,
      "learning_rate": 4.5516104133043296e-05,
      "loss": 0.8867,
      "step": 2862
    },
    {
      "epoch": 0.904010104199558,
      "grad_norm": 1.012305498123169,
      "learning_rate": 4.55125956073258e-05,
      "loss": 0.846,
      "step": 2863
    },
    {
      "epoch": 0.9043258604357436,
      "grad_norm": 1.1148563623428345,
      "learning_rate": 4.550908708160831e-05,
      "loss": 0.8931,
      "step": 2864
    },
    {
      "epoch": 0.9046416166719292,
      "grad_norm": 0.916312575340271,
      "learning_rate": 4.5505578555890816e-05,
      "loss": 0.9351,
      "step": 2865
    },
    {
      "epoch": 0.904957372908115,
      "grad_norm": 1.0679224729537964,
      "learning_rate": 4.550207003017332e-05,
      "loss": 0.9096,
      "step": 2866
    },
    {
      "epoch": 0.9052731291443006,
      "grad_norm": 0.9222119450569153,
      "learning_rate": 4.549856150445583e-05,
      "loss": 0.9367,
      "step": 2867
    },
    {
      "epoch": 0.9055888853804863,
      "grad_norm": 0.9513258934020996,
      "learning_rate": 4.549505297873834e-05,
      "loss": 0.9439,
      "step": 2868
    },
    {
      "epoch": 0.905904641616672,
      "grad_norm": 0.9700416326522827,
      "learning_rate": 4.549154445302084e-05,
      "loss": 0.9992,
      "step": 2869
    },
    {
      "epoch": 0.9062203978528576,
      "grad_norm": 0.9239291548728943,
      "learning_rate": 4.548803592730335e-05,
      "loss": 0.9492,
      "step": 2870
    },
    {
      "epoch": 0.9065361540890433,
      "grad_norm": 3.421285629272461,
      "learning_rate": 4.548452740158586e-05,
      "loss": 0.8675,
      "step": 2871
    },
    {
      "epoch": 0.9068519103252289,
      "grad_norm": 0.9730812907218933,
      "learning_rate": 4.5481018875868364e-05,
      "loss": 0.8978,
      "step": 2872
    },
    {
      "epoch": 0.9071676665614146,
      "grad_norm": 0.8398175239562988,
      "learning_rate": 4.547751035015087e-05,
      "loss": 0.9456,
      "step": 2873
    },
    {
      "epoch": 0.9074834227976003,
      "grad_norm": 0.990656316280365,
      "learning_rate": 4.547400182443338e-05,
      "loss": 0.9211,
      "step": 2874
    },
    {
      "epoch": 0.9077991790337859,
      "grad_norm": 1.0789458751678467,
      "learning_rate": 4.5470493298715885e-05,
      "loss": 0.9432,
      "step": 2875
    },
    {
      "epoch": 0.9081149352699716,
      "grad_norm": 0.9680424332618713,
      "learning_rate": 4.546698477299839e-05,
      "loss": 0.9078,
      "step": 2876
    },
    {
      "epoch": 0.9084306915061573,
      "grad_norm": 0.9696246981620789,
      "learning_rate": 4.54634762472809e-05,
      "loss": 0.921,
      "step": 2877
    },
    {
      "epoch": 0.9087464477423429,
      "grad_norm": 1.43312668800354,
      "learning_rate": 4.54599677215634e-05,
      "loss": 0.916,
      "step": 2878
    },
    {
      "epoch": 0.9090622039785285,
      "grad_norm": 0.9391671419143677,
      "learning_rate": 4.5456459195845905e-05,
      "loss": 0.8693,
      "step": 2879
    },
    {
      "epoch": 0.9093779602147143,
      "grad_norm": 1.069283366203308,
      "learning_rate": 4.545295067012841e-05,
      "loss": 0.9148,
      "step": 2880
    },
    {
      "epoch": 0.9096937164508999,
      "grad_norm": 0.9354763031005859,
      "learning_rate": 4.544944214441092e-05,
      "loss": 0.9045,
      "step": 2881
    },
    {
      "epoch": 0.9100094726870855,
      "grad_norm": 1.0743329524993896,
      "learning_rate": 4.5445933618693426e-05,
      "loss": 0.9243,
      "step": 2882
    },
    {
      "epoch": 0.9103252289232713,
      "grad_norm": 1.0252357721328735,
      "learning_rate": 4.544242509297593e-05,
      "loss": 0.9445,
      "step": 2883
    },
    {
      "epoch": 0.9106409851594569,
      "grad_norm": 0.9274823069572449,
      "learning_rate": 4.543891656725844e-05,
      "loss": 0.8677,
      "step": 2884
    },
    {
      "epoch": 0.9109567413956425,
      "grad_norm": 1.0803545713424683,
      "learning_rate": 4.5435408041540946e-05,
      "loss": 0.8813,
      "step": 2885
    },
    {
      "epoch": 0.9112724976318283,
      "grad_norm": 1.0759519338607788,
      "learning_rate": 4.543189951582345e-05,
      "loss": 0.9193,
      "step": 2886
    },
    {
      "epoch": 0.9115882538680139,
      "grad_norm": 1.373956322669983,
      "learning_rate": 4.542839099010596e-05,
      "loss": 0.8614,
      "step": 2887
    },
    {
      "epoch": 0.9119040101041995,
      "grad_norm": 1.2221896648406982,
      "learning_rate": 4.5424882464388466e-05,
      "loss": 0.9456,
      "step": 2888
    },
    {
      "epoch": 0.9122197663403853,
      "grad_norm": 1.1441844701766968,
      "learning_rate": 4.542137393867097e-05,
      "loss": 0.9195,
      "step": 2889
    },
    {
      "epoch": 0.9125355225765709,
      "grad_norm": 0.9834370613098145,
      "learning_rate": 4.541786541295348e-05,
      "loss": 0.9088,
      "step": 2890
    },
    {
      "epoch": 0.9128512788127565,
      "grad_norm": 1.1308224201202393,
      "learning_rate": 4.541435688723599e-05,
      "loss": 0.8632,
      "step": 2891
    },
    {
      "epoch": 0.9131670350489423,
      "grad_norm": 1.3427680730819702,
      "learning_rate": 4.5410848361518494e-05,
      "loss": 0.8975,
      "step": 2892
    },
    {
      "epoch": 0.9134827912851279,
      "grad_norm": 1.2644745111465454,
      "learning_rate": 4.5407339835801e-05,
      "loss": 0.9362,
      "step": 2893
    },
    {
      "epoch": 0.9137985475213135,
      "grad_norm": 1.0677839517593384,
      "learning_rate": 4.540383131008351e-05,
      "loss": 0.9336,
      "step": 2894
    },
    {
      "epoch": 0.9141143037574992,
      "grad_norm": 1.0674268007278442,
      "learning_rate": 4.5400322784366014e-05,
      "loss": 0.9034,
      "step": 2895
    },
    {
      "epoch": 0.9144300599936849,
      "grad_norm": 1.1463627815246582,
      "learning_rate": 4.539681425864852e-05,
      "loss": 0.8919,
      "step": 2896
    },
    {
      "epoch": 0.9147458162298705,
      "grad_norm": 1.0771000385284424,
      "learning_rate": 4.539330573293102e-05,
      "loss": 0.862,
      "step": 2897
    },
    {
      "epoch": 0.9150615724660562,
      "grad_norm": 0.9251135587692261,
      "learning_rate": 4.538979720721353e-05,
      "loss": 0.8995,
      "step": 2898
    },
    {
      "epoch": 0.9153773287022419,
      "grad_norm": 1.0909013748168945,
      "learning_rate": 4.5386288681496035e-05,
      "loss": 0.908,
      "step": 2899
    },
    {
      "epoch": 0.9156930849384275,
      "grad_norm": 1.0060232877731323,
      "learning_rate": 4.538278015577854e-05,
      "loss": 0.9338,
      "step": 2900
    },
    {
      "epoch": 0.9160088411746132,
      "grad_norm": 1.1674187183380127,
      "learning_rate": 4.537927163006105e-05,
      "loss": 0.917,
      "step": 2901
    },
    {
      "epoch": 0.9163245974107989,
      "grad_norm": 1.0613524913787842,
      "learning_rate": 4.5375763104343555e-05,
      "loss": 0.9227,
      "step": 2902
    },
    {
      "epoch": 0.9166403536469845,
      "grad_norm": 1.240087628364563,
      "learning_rate": 4.537225457862606e-05,
      "loss": 0.9004,
      "step": 2903
    },
    {
      "epoch": 0.9169561098831702,
      "grad_norm": 1.1152108907699585,
      "learning_rate": 4.536874605290857e-05,
      "loss": 0.8719,
      "step": 2904
    },
    {
      "epoch": 0.9172718661193558,
      "grad_norm": 1.064544439315796,
      "learning_rate": 4.536523752719108e-05,
      "loss": 0.8969,
      "step": 2905
    },
    {
      "epoch": 0.9175876223555415,
      "grad_norm": 1.125544786453247,
      "learning_rate": 4.536172900147358e-05,
      "loss": 0.9316,
      "step": 2906
    },
    {
      "epoch": 0.9179033785917272,
      "grad_norm": 0.9890003800392151,
      "learning_rate": 4.535822047575609e-05,
      "loss": 0.8735,
      "step": 2907
    },
    {
      "epoch": 0.9182191348279128,
      "grad_norm": 1.457308053970337,
      "learning_rate": 4.5354711950038596e-05,
      "loss": 0.934,
      "step": 2908
    },
    {
      "epoch": 0.9185348910640985,
      "grad_norm": 1.0316928625106812,
      "learning_rate": 4.53512034243211e-05,
      "loss": 0.9063,
      "step": 2909
    },
    {
      "epoch": 0.9188506473002842,
      "grad_norm": 1.0047565698623657,
      "learning_rate": 4.534769489860361e-05,
      "loss": 0.8873,
      "step": 2910
    },
    {
      "epoch": 0.9191664035364698,
      "grad_norm": 1.2738251686096191,
      "learning_rate": 4.5344186372886116e-05,
      "loss": 0.8843,
      "step": 2911
    },
    {
      "epoch": 0.9194821597726555,
      "grad_norm": 1.1915535926818848,
      "learning_rate": 4.534067784716862e-05,
      "loss": 0.8757,
      "step": 2912
    },
    {
      "epoch": 0.9197979160088412,
      "grad_norm": 1.0581080913543701,
      "learning_rate": 4.533716932145113e-05,
      "loss": 0.9874,
      "step": 2913
    },
    {
      "epoch": 0.9201136722450268,
      "grad_norm": 1.027868390083313,
      "learning_rate": 4.533366079573364e-05,
      "loss": 0.9129,
      "step": 2914
    },
    {
      "epoch": 0.9204294284812125,
      "grad_norm": 1.0599786043167114,
      "learning_rate": 4.5330152270016144e-05,
      "loss": 0.8728,
      "step": 2915
    },
    {
      "epoch": 0.9207451847173982,
      "grad_norm": 1.1101906299591064,
      "learning_rate": 4.5326643744298644e-05,
      "loss": 0.8714,
      "step": 2916
    },
    {
      "epoch": 0.9210609409535838,
      "grad_norm": 1.2183983325958252,
      "learning_rate": 4.532313521858115e-05,
      "loss": 0.9337,
      "step": 2917
    },
    {
      "epoch": 0.9213766971897694,
      "grad_norm": 1.2733267545700073,
      "learning_rate": 4.531962669286366e-05,
      "loss": 0.917,
      "step": 2918
    },
    {
      "epoch": 0.9216924534259552,
      "grad_norm": 1.2762457132339478,
      "learning_rate": 4.5316118167146164e-05,
      "loss": 0.8927,
      "step": 2919
    },
    {
      "epoch": 0.9220082096621408,
      "grad_norm": 1.063663125038147,
      "learning_rate": 4.531260964142867e-05,
      "loss": 0.8837,
      "step": 2920
    },
    {
      "epoch": 0.9223239658983264,
      "grad_norm": 1.1678236722946167,
      "learning_rate": 4.530910111571118e-05,
      "loss": 0.9025,
      "step": 2921
    },
    {
      "epoch": 0.9226397221345122,
      "grad_norm": 1.268550992012024,
      "learning_rate": 4.530559258999369e-05,
      "loss": 0.8971,
      "step": 2922
    },
    {
      "epoch": 0.9229554783706978,
      "grad_norm": 1.249167799949646,
      "learning_rate": 4.53020840642762e-05,
      "loss": 0.9029,
      "step": 2923
    },
    {
      "epoch": 0.9232712346068835,
      "grad_norm": 1.0813201665878296,
      "learning_rate": 4.5298575538558705e-05,
      "loss": 0.8515,
      "step": 2924
    },
    {
      "epoch": 0.9235869908430692,
      "grad_norm": 1.0783543586730957,
      "learning_rate": 4.5295067012841205e-05,
      "loss": 0.9227,
      "step": 2925
    },
    {
      "epoch": 0.9239027470792548,
      "grad_norm": 0.9856969118118286,
      "learning_rate": 4.529155848712371e-05,
      "loss": 0.8981,
      "step": 2926
    },
    {
      "epoch": 0.9242185033154405,
      "grad_norm": 1.0245016813278198,
      "learning_rate": 4.528804996140622e-05,
      "loss": 0.9144,
      "step": 2927
    },
    {
      "epoch": 0.9245342595516262,
      "grad_norm": 0.988253116607666,
      "learning_rate": 4.5284541435688725e-05,
      "loss": 0.8899,
      "step": 2928
    },
    {
      "epoch": 0.9248500157878118,
      "grad_norm": 0.95001220703125,
      "learning_rate": 4.528103290997123e-05,
      "loss": 0.9015,
      "step": 2929
    },
    {
      "epoch": 0.9251657720239975,
      "grad_norm": 0.9206025004386902,
      "learning_rate": 4.527752438425374e-05,
      "loss": 0.903,
      "step": 2930
    },
    {
      "epoch": 0.9254815282601832,
      "grad_norm": 1.0454187393188477,
      "learning_rate": 4.5274015858536246e-05,
      "loss": 0.9002,
      "step": 2931
    },
    {
      "epoch": 0.9257972844963688,
      "grad_norm": 1.0832078456878662,
      "learning_rate": 4.527050733281875e-05,
      "loss": 0.9021,
      "step": 2932
    },
    {
      "epoch": 0.9261130407325545,
      "grad_norm": 0.9830026626586914,
      "learning_rate": 4.526699880710126e-05,
      "loss": 0.8716,
      "step": 2933
    },
    {
      "epoch": 0.9264287969687401,
      "grad_norm": 1.097682237625122,
      "learning_rate": 4.5263490281383766e-05,
      "loss": 0.873,
      "step": 2934
    },
    {
      "epoch": 0.9267445532049258,
      "grad_norm": 1.7189642190933228,
      "learning_rate": 4.525998175566627e-05,
      "loss": 0.8968,
      "step": 2935
    },
    {
      "epoch": 0.9270603094411115,
      "grad_norm": 0.9436437487602234,
      "learning_rate": 4.525647322994877e-05,
      "loss": 0.8257,
      "step": 2936
    },
    {
      "epoch": 0.9273760656772971,
      "grad_norm": 1.0276210308074951,
      "learning_rate": 4.525296470423128e-05,
      "loss": 0.9474,
      "step": 2937
    },
    {
      "epoch": 0.9276918219134828,
      "grad_norm": 1.0971757173538208,
      "learning_rate": 4.524945617851379e-05,
      "loss": 0.9268,
      "step": 2938
    },
    {
      "epoch": 0.9280075781496685,
      "grad_norm": 1.7977612018585205,
      "learning_rate": 4.5245947652796294e-05,
      "loss": 0.8558,
      "step": 2939
    },
    {
      "epoch": 0.9283233343858541,
      "grad_norm": 1.1872583627700806,
      "learning_rate": 4.524243912707881e-05,
      "loss": 0.9357,
      "step": 2940
    },
    {
      "epoch": 0.9286390906220398,
      "grad_norm": 0.9923650026321411,
      "learning_rate": 4.5238930601361314e-05,
      "loss": 0.8675,
      "step": 2941
    },
    {
      "epoch": 0.9289548468582255,
      "grad_norm": 1.2634459733963013,
      "learning_rate": 4.523542207564382e-05,
      "loss": 0.9213,
      "step": 2942
    },
    {
      "epoch": 0.9292706030944111,
      "grad_norm": 1.2108501195907593,
      "learning_rate": 4.523191354992633e-05,
      "loss": 0.8914,
      "step": 2943
    },
    {
      "epoch": 0.9295863593305967,
      "grad_norm": 1.0534813404083252,
      "learning_rate": 4.522840502420883e-05,
      "loss": 0.8956,
      "step": 2944
    },
    {
      "epoch": 0.9299021155667825,
      "grad_norm": 0.9897943735122681,
      "learning_rate": 4.5224896498491334e-05,
      "loss": 0.8605,
      "step": 2945
    },
    {
      "epoch": 0.9302178718029681,
      "grad_norm": 0.956579864025116,
      "learning_rate": 4.522138797277384e-05,
      "loss": 0.8782,
      "step": 2946
    },
    {
      "epoch": 0.9305336280391537,
      "grad_norm": 1.0236585140228271,
      "learning_rate": 4.521787944705635e-05,
      "loss": 0.8875,
      "step": 2947
    },
    {
      "epoch": 0.9308493842753395,
      "grad_norm": 1.1353472471237183,
      "learning_rate": 4.5214370921338855e-05,
      "loss": 0.939,
      "step": 2948
    },
    {
      "epoch": 0.9311651405115251,
      "grad_norm": 1.0031490325927734,
      "learning_rate": 4.521086239562136e-05,
      "loss": 0.8233,
      "step": 2949
    },
    {
      "epoch": 0.9314808967477107,
      "grad_norm": 1.120057225227356,
      "learning_rate": 4.520735386990387e-05,
      "loss": 0.9408,
      "step": 2950
    },
    {
      "epoch": 0.9317966529838965,
      "grad_norm": 0.9295412302017212,
      "learning_rate": 4.5203845344186375e-05,
      "loss": 0.9307,
      "step": 2951
    },
    {
      "epoch": 0.9321124092200821,
      "grad_norm": 1.1131649017333984,
      "learning_rate": 4.520033681846888e-05,
      "loss": 0.877,
      "step": 2952
    },
    {
      "epoch": 0.9324281654562677,
      "grad_norm": 1.1381089687347412,
      "learning_rate": 4.519682829275139e-05,
      "loss": 0.929,
      "step": 2953
    },
    {
      "epoch": 0.9327439216924535,
      "grad_norm": 1.000033974647522,
      "learning_rate": 4.5193319767033896e-05,
      "loss": 0.9078,
      "step": 2954
    },
    {
      "epoch": 0.9330596779286391,
      "grad_norm": 1.1874357461929321,
      "learning_rate": 4.5189811241316396e-05,
      "loss": 0.9092,
      "step": 2955
    },
    {
      "epoch": 0.9333754341648247,
      "grad_norm": 1.1662068367004395,
      "learning_rate": 4.51863027155989e-05,
      "loss": 0.9528,
      "step": 2956
    },
    {
      "epoch": 0.9336911904010105,
      "grad_norm": 1.0516247749328613,
      "learning_rate": 4.5182794189881416e-05,
      "loss": 0.9006,
      "step": 2957
    },
    {
      "epoch": 0.9340069466371961,
      "grad_norm": 1.4025721549987793,
      "learning_rate": 4.517928566416392e-05,
      "loss": 0.9548,
      "step": 2958
    },
    {
      "epoch": 0.9343227028733817,
      "grad_norm": 1.8851778507232666,
      "learning_rate": 4.517577713844643e-05,
      "loss": 0.9474,
      "step": 2959
    },
    {
      "epoch": 0.9346384591095674,
      "grad_norm": 1.0349570512771606,
      "learning_rate": 4.517226861272894e-05,
      "loss": 0.904,
      "step": 2960
    },
    {
      "epoch": 0.9349542153457531,
      "grad_norm": 0.988892138004303,
      "learning_rate": 4.5168760087011443e-05,
      "loss": 0.8742,
      "step": 2961
    },
    {
      "epoch": 0.9352699715819387,
      "grad_norm": 1.1185294389724731,
      "learning_rate": 4.516525156129395e-05,
      "loss": 0.8623,
      "step": 2962
    },
    {
      "epoch": 0.9355857278181244,
      "grad_norm": 1.0694365501403809,
      "learning_rate": 4.516174303557646e-05,
      "loss": 0.851,
      "step": 2963
    },
    {
      "epoch": 0.9359014840543101,
      "grad_norm": 1.1154131889343262,
      "learning_rate": 4.515823450985896e-05,
      "loss": 0.8982,
      "step": 2964
    },
    {
      "epoch": 0.9362172402904957,
      "grad_norm": 1.1517608165740967,
      "learning_rate": 4.5154725984141464e-05,
      "loss": 0.8852,
      "step": 2965
    },
    {
      "epoch": 0.9365329965266814,
      "grad_norm": 1.1106007099151611,
      "learning_rate": 4.515121745842397e-05,
      "loss": 0.8869,
      "step": 2966
    },
    {
      "epoch": 0.936848752762867,
      "grad_norm": 1.1184413433074951,
      "learning_rate": 4.514770893270648e-05,
      "loss": 0.8749,
      "step": 2967
    },
    {
      "epoch": 0.9371645089990527,
      "grad_norm": 1.0219825506210327,
      "learning_rate": 4.5144200406988984e-05,
      "loss": 0.8288,
      "step": 2968
    },
    {
      "epoch": 0.9374802652352384,
      "grad_norm": 1.0798217058181763,
      "learning_rate": 4.514069188127149e-05,
      "loss": 0.9506,
      "step": 2969
    },
    {
      "epoch": 0.937796021471424,
      "grad_norm": 1.2466092109680176,
      "learning_rate": 4.5137183355554e-05,
      "loss": 0.9298,
      "step": 2970
    },
    {
      "epoch": 0.9381117777076097,
      "grad_norm": 1.0016347169876099,
      "learning_rate": 4.5133674829836505e-05,
      "loss": 0.846,
      "step": 2971
    },
    {
      "epoch": 0.9384275339437954,
      "grad_norm": 1.4136356115341187,
      "learning_rate": 4.513016630411901e-05,
      "loss": 0.8627,
      "step": 2972
    },
    {
      "epoch": 0.938743290179981,
      "grad_norm": 1.039479374885559,
      "learning_rate": 4.512665777840152e-05,
      "loss": 0.8364,
      "step": 2973
    },
    {
      "epoch": 0.9390590464161667,
      "grad_norm": 1.1516257524490356,
      "learning_rate": 4.512314925268402e-05,
      "loss": 0.9055,
      "step": 2974
    },
    {
      "epoch": 0.9393748026523524,
      "grad_norm": 1.223896861076355,
      "learning_rate": 4.511964072696653e-05,
      "loss": 0.8498,
      "step": 2975
    },
    {
      "epoch": 0.939690558888538,
      "grad_norm": 1.2993367910385132,
      "learning_rate": 4.511613220124904e-05,
      "loss": 0.8886,
      "step": 2976
    },
    {
      "epoch": 0.9400063151247238,
      "grad_norm": 1.2625572681427002,
      "learning_rate": 4.5112623675531546e-05,
      "loss": 0.9474,
      "step": 2977
    },
    {
      "epoch": 0.9403220713609094,
      "grad_norm": 1.2441917657852173,
      "learning_rate": 4.510911514981405e-05,
      "loss": 0.8634,
      "step": 2978
    },
    {
      "epoch": 0.940637827597095,
      "grad_norm": 1.2956151962280273,
      "learning_rate": 4.510560662409656e-05,
      "loss": 0.9178,
      "step": 2979
    },
    {
      "epoch": 0.9409535838332808,
      "grad_norm": 0.9785074591636658,
      "learning_rate": 4.5102098098379066e-05,
      "loss": 0.8979,
      "step": 2980
    },
    {
      "epoch": 0.9412693400694664,
      "grad_norm": 1.117148756980896,
      "learning_rate": 4.509858957266157e-05,
      "loss": 0.995,
      "step": 2981
    },
    {
      "epoch": 0.941585096305652,
      "grad_norm": 0.9936628341674805,
      "learning_rate": 4.509508104694408e-05,
      "loss": 0.844,
      "step": 2982
    },
    {
      "epoch": 0.9419008525418378,
      "grad_norm": 1.2825982570648193,
      "learning_rate": 4.509157252122658e-05,
      "loss": 0.9358,
      "step": 2983
    },
    {
      "epoch": 0.9422166087780234,
      "grad_norm": 1.1412286758422852,
      "learning_rate": 4.5088063995509087e-05,
      "loss": 0.9583,
      "step": 2984
    },
    {
      "epoch": 0.942532365014209,
      "grad_norm": 0.9620031714439392,
      "learning_rate": 4.508455546979159e-05,
      "loss": 0.8897,
      "step": 2985
    },
    {
      "epoch": 0.9428481212503947,
      "grad_norm": 0.9889291524887085,
      "learning_rate": 4.50810469440741e-05,
      "loss": 0.8981,
      "step": 2986
    },
    {
      "epoch": 0.9431638774865804,
      "grad_norm": 1.0185970067977905,
      "learning_rate": 4.507753841835661e-05,
      "loss": 0.9048,
      "step": 2987
    },
    {
      "epoch": 0.943479633722766,
      "grad_norm": 1.0200780630111694,
      "learning_rate": 4.5074029892639114e-05,
      "loss": 0.8474,
      "step": 2988
    },
    {
      "epoch": 0.9437953899589517,
      "grad_norm": 1.2466601133346558,
      "learning_rate": 4.507052136692162e-05,
      "loss": 0.8813,
      "step": 2989
    },
    {
      "epoch": 0.9441111461951374,
      "grad_norm": 1.1296182870864868,
      "learning_rate": 4.506701284120413e-05,
      "loss": 0.9068,
      "step": 2990
    },
    {
      "epoch": 0.944426902431323,
      "grad_norm": 1.2353909015655518,
      "learning_rate": 4.5063504315486634e-05,
      "loss": 0.9269,
      "step": 2991
    },
    {
      "epoch": 0.9447426586675087,
      "grad_norm": 1.0816261768341064,
      "learning_rate": 4.505999578976914e-05,
      "loss": 0.8918,
      "step": 2992
    },
    {
      "epoch": 0.9450584149036944,
      "grad_norm": 1.1084672212600708,
      "learning_rate": 4.505648726405165e-05,
      "loss": 0.9349,
      "step": 2993
    },
    {
      "epoch": 0.94537417113988,
      "grad_norm": 1.14238440990448,
      "learning_rate": 4.5052978738334155e-05,
      "loss": 0.91,
      "step": 2994
    },
    {
      "epoch": 0.9456899273760657,
      "grad_norm": 1.0209029912948608,
      "learning_rate": 4.504947021261666e-05,
      "loss": 0.8891,
      "step": 2995
    },
    {
      "epoch": 0.9460056836122513,
      "grad_norm": 1.0658767223358154,
      "learning_rate": 4.504596168689917e-05,
      "loss": 0.8913,
      "step": 2996
    },
    {
      "epoch": 0.946321439848437,
      "grad_norm": 1.1011661291122437,
      "learning_rate": 4.5042453161181675e-05,
      "loss": 0.8117,
      "step": 2997
    },
    {
      "epoch": 0.9466371960846227,
      "grad_norm": 0.9932137727737427,
      "learning_rate": 4.503894463546418e-05,
      "loss": 0.9405,
      "step": 2998
    },
    {
      "epoch": 0.9469529523208083,
      "grad_norm": 1.083028793334961,
      "learning_rate": 4.503543610974669e-05,
      "loss": 0.9282,
      "step": 2999
    },
    {
      "epoch": 0.947268708556994,
      "grad_norm": 1.0829424858093262,
      "learning_rate": 4.5031927584029196e-05,
      "loss": 0.8793,
      "step": 3000
    },
    {
      "epoch": 0.947268708556994,
      "eval_loss": 0.85349041223526,
      "eval_runtime": 4.6498,
      "eval_samples_per_second": 535.508,
      "eval_steps_per_second": 4.516,
      "step": 3000
    },
    {
      "epoch": 0.9475844647931797,
      "grad_norm": 1.1133885383605957,
      "learning_rate": 4.50284190583117e-05,
      "loss": 0.8968,
      "step": 3001
    },
    {
      "epoch": 0.9479002210293653,
      "grad_norm": 1.0415027141571045,
      "learning_rate": 4.50249105325942e-05,
      "loss": 0.949,
      "step": 3002
    },
    {
      "epoch": 0.948215977265551,
      "grad_norm": 1.0418869256973267,
      "learning_rate": 4.502140200687671e-05,
      "loss": 0.9167,
      "step": 3003
    },
    {
      "epoch": 0.9485317335017367,
      "grad_norm": 1.114972472190857,
      "learning_rate": 4.5017893481159216e-05,
      "loss": 0.9112,
      "step": 3004
    },
    {
      "epoch": 0.9488474897379223,
      "grad_norm": 0.9584295749664307,
      "learning_rate": 4.501438495544172e-05,
      "loss": 0.8306,
      "step": 3005
    },
    {
      "epoch": 0.949163245974108,
      "grad_norm": 0.9401095509529114,
      "learning_rate": 4.501087642972423e-05,
      "loss": 0.8986,
      "step": 3006
    },
    {
      "epoch": 0.9494790022102937,
      "grad_norm": 0.9166591763496399,
      "learning_rate": 4.5007367904006736e-05,
      "loss": 0.8323,
      "step": 3007
    },
    {
      "epoch": 0.9497947584464793,
      "grad_norm": 1.0116363763809204,
      "learning_rate": 4.500385937828924e-05,
      "loss": 0.915,
      "step": 3008
    },
    {
      "epoch": 0.9501105146826649,
      "grad_norm": 1.2658600807189941,
      "learning_rate": 4.500035085257175e-05,
      "loss": 0.8871,
      "step": 3009
    },
    {
      "epoch": 0.9504262709188507,
      "grad_norm": 1.0832006931304932,
      "learning_rate": 4.4996842326854264e-05,
      "loss": 0.8432,
      "step": 3010
    },
    {
      "epoch": 0.9507420271550363,
      "grad_norm": 0.9858319163322449,
      "learning_rate": 4.4993333801136764e-05,
      "loss": 0.8582,
      "step": 3011
    },
    {
      "epoch": 0.9510577833912219,
      "grad_norm": 1.041562795639038,
      "learning_rate": 4.498982527541927e-05,
      "loss": 0.9243,
      "step": 3012
    },
    {
      "epoch": 0.9513735396274077,
      "grad_norm": 1.178750991821289,
      "learning_rate": 4.498631674970178e-05,
      "loss": 0.9378,
      "step": 3013
    },
    {
      "epoch": 0.9516892958635933,
      "grad_norm": 0.9722561240196228,
      "learning_rate": 4.4982808223984284e-05,
      "loss": 0.947,
      "step": 3014
    },
    {
      "epoch": 0.9520050520997789,
      "grad_norm": 1.1689636707305908,
      "learning_rate": 4.497929969826679e-05,
      "loss": 0.9101,
      "step": 3015
    },
    {
      "epoch": 0.9523208083359647,
      "grad_norm": 1.0368140935897827,
      "learning_rate": 4.49757911725493e-05,
      "loss": 0.8917,
      "step": 3016
    },
    {
      "epoch": 0.9526365645721503,
      "grad_norm": 1.126754879951477,
      "learning_rate": 4.4972282646831805e-05,
      "loss": 0.8607,
      "step": 3017
    },
    {
      "epoch": 0.9529523208083359,
      "grad_norm": 1.0815433263778687,
      "learning_rate": 4.496877412111431e-05,
      "loss": 0.9278,
      "step": 3018
    },
    {
      "epoch": 0.9532680770445217,
      "grad_norm": 1.1035547256469727,
      "learning_rate": 4.496526559539682e-05,
      "loss": 0.8995,
      "step": 3019
    },
    {
      "epoch": 0.9535838332807073,
      "grad_norm": 1.245469331741333,
      "learning_rate": 4.4961757069679325e-05,
      "loss": 0.9051,
      "step": 3020
    },
    {
      "epoch": 0.9538995895168929,
      "grad_norm": 1.3107539415359497,
      "learning_rate": 4.4958248543961825e-05,
      "loss": 0.9541,
      "step": 3021
    },
    {
      "epoch": 0.9542153457530786,
      "grad_norm": 1.2437812089920044,
      "learning_rate": 4.495474001824433e-05,
      "loss": 0.9004,
      "step": 3022
    },
    {
      "epoch": 0.9545311019892643,
      "grad_norm": 0.9474031925201416,
      "learning_rate": 4.495123149252684e-05,
      "loss": 0.827,
      "step": 3023
    },
    {
      "epoch": 0.9548468582254499,
      "grad_norm": 1.2521545886993408,
      "learning_rate": 4.4947722966809346e-05,
      "loss": 0.8641,
      "step": 3024
    },
    {
      "epoch": 0.9551626144616356,
      "grad_norm": 1.118855595588684,
      "learning_rate": 4.494421444109185e-05,
      "loss": 0.8961,
      "step": 3025
    },
    {
      "epoch": 0.9554783706978213,
      "grad_norm": 1.1479167938232422,
      "learning_rate": 4.494070591537436e-05,
      "loss": 0.9104,
      "step": 3026
    },
    {
      "epoch": 0.9557941269340069,
      "grad_norm": 1.1611300706863403,
      "learning_rate": 4.493719738965687e-05,
      "loss": 0.9206,
      "step": 3027
    },
    {
      "epoch": 0.9561098831701926,
      "grad_norm": 0.9785721898078918,
      "learning_rate": 4.493368886393938e-05,
      "loss": 0.8086,
      "step": 3028
    },
    {
      "epoch": 0.9564256394063783,
      "grad_norm": 1.1774497032165527,
      "learning_rate": 4.4930180338221886e-05,
      "loss": 0.8865,
      "step": 3029
    },
    {
      "epoch": 0.9567413956425639,
      "grad_norm": 1.1353667974472046,
      "learning_rate": 4.4926671812504386e-05,
      "loss": 0.9476,
      "step": 3030
    },
    {
      "epoch": 0.9570571518787496,
      "grad_norm": 1.152693510055542,
      "learning_rate": 4.492316328678689e-05,
      "loss": 0.855,
      "step": 3031
    },
    {
      "epoch": 0.9573729081149353,
      "grad_norm": 1.1054630279541016,
      "learning_rate": 4.49196547610694e-05,
      "loss": 0.8783,
      "step": 3032
    },
    {
      "epoch": 0.957688664351121,
      "grad_norm": 1.165578842163086,
      "learning_rate": 4.491614623535191e-05,
      "loss": 0.8589,
      "step": 3033
    },
    {
      "epoch": 0.9580044205873066,
      "grad_norm": 1.3960182666778564,
      "learning_rate": 4.4912637709634414e-05,
      "loss": 0.9287,
      "step": 3034
    },
    {
      "epoch": 0.9583201768234922,
      "grad_norm": 0.9414672255516052,
      "learning_rate": 4.490912918391692e-05,
      "loss": 0.8159,
      "step": 3035
    },
    {
      "epoch": 0.958635933059678,
      "grad_norm": 1.0324352979660034,
      "learning_rate": 4.490562065819943e-05,
      "loss": 0.9082,
      "step": 3036
    },
    {
      "epoch": 0.9589516892958636,
      "grad_norm": 0.9721778631210327,
      "learning_rate": 4.4902112132481934e-05,
      "loss": 0.867,
      "step": 3037
    },
    {
      "epoch": 0.9592674455320492,
      "grad_norm": 1.097660779953003,
      "learning_rate": 4.489860360676444e-05,
      "loss": 0.8706,
      "step": 3038
    },
    {
      "epoch": 0.959583201768235,
      "grad_norm": 1.1163232326507568,
      "learning_rate": 4.489509508104695e-05,
      "loss": 0.8419,
      "step": 3039
    },
    {
      "epoch": 0.9598989580044206,
      "grad_norm": 1.1376755237579346,
      "learning_rate": 4.4891586555329455e-05,
      "loss": 0.8837,
      "step": 3040
    },
    {
      "epoch": 0.9602147142406062,
      "grad_norm": 1.087418556213379,
      "learning_rate": 4.4888078029611955e-05,
      "loss": 0.9157,
      "step": 3041
    },
    {
      "epoch": 0.960530470476792,
      "grad_norm": 1.0643036365509033,
      "learning_rate": 4.488456950389446e-05,
      "loss": 0.8711,
      "step": 3042
    },
    {
      "epoch": 0.9608462267129776,
      "grad_norm": 1.12212336063385,
      "learning_rate": 4.488106097817697e-05,
      "loss": 0.8969,
      "step": 3043
    },
    {
      "epoch": 0.9611619829491632,
      "grad_norm": 1.2488645315170288,
      "learning_rate": 4.4877552452459475e-05,
      "loss": 0.8622,
      "step": 3044
    },
    {
      "epoch": 0.961477739185349,
      "grad_norm": 1.0768134593963623,
      "learning_rate": 4.487404392674199e-05,
      "loss": 0.8681,
      "step": 3045
    },
    {
      "epoch": 0.9617934954215346,
      "grad_norm": 1.2107365131378174,
      "learning_rate": 4.4870535401024495e-05,
      "loss": 0.8989,
      "step": 3046
    },
    {
      "epoch": 0.9621092516577202,
      "grad_norm": 1.018805980682373,
      "learning_rate": 4.4867026875307e-05,
      "loss": 0.9024,
      "step": 3047
    },
    {
      "epoch": 0.962425007893906,
      "grad_norm": 1.0466809272766113,
      "learning_rate": 4.486351834958951e-05,
      "loss": 0.8577,
      "step": 3048
    },
    {
      "epoch": 0.9627407641300916,
      "grad_norm": 0.9803243279457092,
      "learning_rate": 4.486000982387201e-05,
      "loss": 0.8949,
      "step": 3049
    },
    {
      "epoch": 0.9630565203662772,
      "grad_norm": 1.0960906744003296,
      "learning_rate": 4.4856501298154516e-05,
      "loss": 0.8428,
      "step": 3050
    },
    {
      "epoch": 0.9633722766024629,
      "grad_norm": 1.1146119832992554,
      "learning_rate": 4.485299277243702e-05,
      "loss": 0.853,
      "step": 3051
    },
    {
      "epoch": 0.9636880328386486,
      "grad_norm": 1.102012276649475,
      "learning_rate": 4.484948424671953e-05,
      "loss": 0.9721,
      "step": 3052
    },
    {
      "epoch": 0.9640037890748342,
      "grad_norm": 1.0478065013885498,
      "learning_rate": 4.4845975721002036e-05,
      "loss": 0.8907,
      "step": 3053
    },
    {
      "epoch": 0.9643195453110199,
      "grad_norm": 1.1385153532028198,
      "learning_rate": 4.484246719528454e-05,
      "loss": 0.8552,
      "step": 3054
    },
    {
      "epoch": 0.9646353015472056,
      "grad_norm": 0.9938973188400269,
      "learning_rate": 4.483895866956705e-05,
      "loss": 0.8187,
      "step": 3055
    },
    {
      "epoch": 0.9649510577833912,
      "grad_norm": 1.0327372550964355,
      "learning_rate": 4.483545014384956e-05,
      "loss": 0.9592,
      "step": 3056
    },
    {
      "epoch": 0.9652668140195769,
      "grad_norm": 1.2061468362808228,
      "learning_rate": 4.4831941618132064e-05,
      "loss": 0.8613,
      "step": 3057
    },
    {
      "epoch": 0.9655825702557626,
      "grad_norm": 1.2126951217651367,
      "learning_rate": 4.482843309241457e-05,
      "loss": 0.875,
      "step": 3058
    },
    {
      "epoch": 0.9658983264919482,
      "grad_norm": 1.0722510814666748,
      "learning_rate": 4.482492456669708e-05,
      "loss": 0.9079,
      "step": 3059
    },
    {
      "epoch": 0.9662140827281339,
      "grad_norm": 1.0523604154586792,
      "learning_rate": 4.482141604097958e-05,
      "loss": 0.8551,
      "step": 3060
    },
    {
      "epoch": 0.9665298389643195,
      "grad_norm": 1.3895678520202637,
      "learning_rate": 4.4817907515262084e-05,
      "loss": 0.8926,
      "step": 3061
    },
    {
      "epoch": 0.9668455952005052,
      "grad_norm": 1.3787956237792969,
      "learning_rate": 4.48143989895446e-05,
      "loss": 0.9868,
      "step": 3062
    },
    {
      "epoch": 0.9671613514366909,
      "grad_norm": 1.341995358467102,
      "learning_rate": 4.4810890463827104e-05,
      "loss": 0.8526,
      "step": 3063
    },
    {
      "epoch": 0.9674771076728765,
      "grad_norm": 1.2154594659805298,
      "learning_rate": 4.480738193810961e-05,
      "loss": 0.8996,
      "step": 3064
    },
    {
      "epoch": 0.9677928639090622,
      "grad_norm": 1.1119743585586548,
      "learning_rate": 4.480387341239212e-05,
      "loss": 0.8859,
      "step": 3065
    },
    {
      "epoch": 0.9681086201452479,
      "grad_norm": 1.16015625,
      "learning_rate": 4.4800364886674625e-05,
      "loss": 0.9303,
      "step": 3066
    },
    {
      "epoch": 0.9684243763814335,
      "grad_norm": 0.9931123852729797,
      "learning_rate": 4.479685636095713e-05,
      "loss": 0.8847,
      "step": 3067
    },
    {
      "epoch": 0.9687401326176192,
      "grad_norm": 1.1112149953842163,
      "learning_rate": 4.479334783523964e-05,
      "loss": 0.915,
      "step": 3068
    },
    {
      "epoch": 0.9690558888538049,
      "grad_norm": 1.002989649772644,
      "learning_rate": 4.478983930952214e-05,
      "loss": 0.9175,
      "step": 3069
    },
    {
      "epoch": 0.9693716450899905,
      "grad_norm": 1.0650256872177124,
      "learning_rate": 4.4786330783804645e-05,
      "loss": 0.929,
      "step": 3070
    },
    {
      "epoch": 0.9696874013261761,
      "grad_norm": 1.162246823310852,
      "learning_rate": 4.478282225808715e-05,
      "loss": 0.9714,
      "step": 3071
    },
    {
      "epoch": 0.9700031575623619,
      "grad_norm": 1.1962890625,
      "learning_rate": 4.477931373236966e-05,
      "loss": 0.9118,
      "step": 3072
    },
    {
      "epoch": 0.9703189137985475,
      "grad_norm": 1.1458452939987183,
      "learning_rate": 4.4775805206652166e-05,
      "loss": 0.968,
      "step": 3073
    },
    {
      "epoch": 0.9706346700347331,
      "grad_norm": 1.001712679862976,
      "learning_rate": 4.477229668093467e-05,
      "loss": 0.8898,
      "step": 3074
    },
    {
      "epoch": 0.9709504262709189,
      "grad_norm": 1.1122463941574097,
      "learning_rate": 4.476878815521718e-05,
      "loss": 0.8838,
      "step": 3075
    },
    {
      "epoch": 0.9712661825071045,
      "grad_norm": 1.0601799488067627,
      "learning_rate": 4.4765279629499686e-05,
      "loss": 0.9165,
      "step": 3076
    },
    {
      "epoch": 0.9715819387432901,
      "grad_norm": 0.9878460764884949,
      "learning_rate": 4.476177110378219e-05,
      "loss": 0.858,
      "step": 3077
    },
    {
      "epoch": 0.9718976949794759,
      "grad_norm": 1.1670392751693726,
      "learning_rate": 4.47582625780647e-05,
      "loss": 0.9213,
      "step": 3078
    },
    {
      "epoch": 0.9722134512156615,
      "grad_norm": 1.0116336345672607,
      "learning_rate": 4.47547540523472e-05,
      "loss": 0.8748,
      "step": 3079
    },
    {
      "epoch": 0.9725292074518471,
      "grad_norm": 0.9041163921356201,
      "learning_rate": 4.4751245526629714e-05,
      "loss": 0.8294,
      "step": 3080
    },
    {
      "epoch": 0.9728449636880329,
      "grad_norm": 1.1185548305511475,
      "learning_rate": 4.474773700091222e-05,
      "loss": 0.9153,
      "step": 3081
    },
    {
      "epoch": 0.9731607199242185,
      "grad_norm": 0.8885275721549988,
      "learning_rate": 4.474422847519473e-05,
      "loss": 0.8701,
      "step": 3082
    },
    {
      "epoch": 0.9734764761604041,
      "grad_norm": 1.1393438577651978,
      "learning_rate": 4.4740719949477234e-05,
      "loss": 0.9554,
      "step": 3083
    },
    {
      "epoch": 0.9737922323965899,
      "grad_norm": 0.9372889399528503,
      "learning_rate": 4.473721142375974e-05,
      "loss": 0.9015,
      "step": 3084
    },
    {
      "epoch": 0.9741079886327755,
      "grad_norm": 0.9301940202713013,
      "learning_rate": 4.473370289804225e-05,
      "loss": 0.8954,
      "step": 3085
    },
    {
      "epoch": 0.9744237448689612,
      "grad_norm": 0.9548868536949158,
      "learning_rate": 4.4730194372324754e-05,
      "loss": 0.9124,
      "step": 3086
    },
    {
      "epoch": 0.9747395011051468,
      "grad_norm": 0.9264155626296997,
      "learning_rate": 4.472668584660726e-05,
      "loss": 0.8554,
      "step": 3087
    },
    {
      "epoch": 0.9750552573413325,
      "grad_norm": 1.31179678440094,
      "learning_rate": 4.472317732088976e-05,
      "loss": 0.9155,
      "step": 3088
    },
    {
      "epoch": 0.9753710135775182,
      "grad_norm": 0.9943866729736328,
      "learning_rate": 4.471966879517227e-05,
      "loss": 0.9273,
      "step": 3089
    },
    {
      "epoch": 0.9756867698137038,
      "grad_norm": 0.874504029750824,
      "learning_rate": 4.4716160269454775e-05,
      "loss": 0.9149,
      "step": 3090
    },
    {
      "epoch": 0.9760025260498895,
      "grad_norm": 1.1772607564926147,
      "learning_rate": 4.471265174373728e-05,
      "loss": 0.885,
      "step": 3091
    },
    {
      "epoch": 0.9763182822860752,
      "grad_norm": 0.9634997844696045,
      "learning_rate": 4.470914321801979e-05,
      "loss": 0.895,
      "step": 3092
    },
    {
      "epoch": 0.9766340385222608,
      "grad_norm": 0.9933551549911499,
      "learning_rate": 4.4705634692302295e-05,
      "loss": 0.8804,
      "step": 3093
    },
    {
      "epoch": 0.9769497947584465,
      "grad_norm": 0.96160489320755,
      "learning_rate": 4.47021261665848e-05,
      "loss": 0.8108,
      "step": 3094
    },
    {
      "epoch": 0.9772655509946322,
      "grad_norm": 0.9287020564079285,
      "learning_rate": 4.469861764086731e-05,
      "loss": 0.8966,
      "step": 3095
    },
    {
      "epoch": 0.9775813072308178,
      "grad_norm": 1.1548281908035278,
      "learning_rate": 4.4695109115149816e-05,
      "loss": 0.921,
      "step": 3096
    },
    {
      "epoch": 0.9778970634670034,
      "grad_norm": 1.08368980884552,
      "learning_rate": 4.469160058943232e-05,
      "loss": 0.8815,
      "step": 3097
    },
    {
      "epoch": 0.9782128197031892,
      "grad_norm": 1.2442452907562256,
      "learning_rate": 4.468809206371483e-05,
      "loss": 0.9746,
      "step": 3098
    },
    {
      "epoch": 0.9785285759393748,
      "grad_norm": 1.1310523748397827,
      "learning_rate": 4.4684583537997336e-05,
      "loss": 0.9383,
      "step": 3099
    },
    {
      "epoch": 0.9788443321755604,
      "grad_norm": 1.041748046875,
      "learning_rate": 4.468107501227984e-05,
      "loss": 0.9186,
      "step": 3100
    },
    {
      "epoch": 0.9791600884117462,
      "grad_norm": 0.9176311492919922,
      "learning_rate": 4.467756648656235e-05,
      "loss": 0.849,
      "step": 3101
    },
    {
      "epoch": 0.9794758446479318,
      "grad_norm": 1.1813466548919678,
      "learning_rate": 4.467405796084486e-05,
      "loss": 0.9041,
      "step": 3102
    },
    {
      "epoch": 0.9797916008841174,
      "grad_norm": 1.0924654006958008,
      "learning_rate": 4.4670549435127363e-05,
      "loss": 0.8914,
      "step": 3103
    },
    {
      "epoch": 0.9801073571203032,
      "grad_norm": 1.0765608549118042,
      "learning_rate": 4.466704090940987e-05,
      "loss": 0.8818,
      "step": 3104
    },
    {
      "epoch": 0.9804231133564888,
      "grad_norm": 1.1642996072769165,
      "learning_rate": 4.466353238369238e-05,
      "loss": 0.9222,
      "step": 3105
    },
    {
      "epoch": 0.9807388695926744,
      "grad_norm": 1.215882420539856,
      "learning_rate": 4.4660023857974884e-05,
      "loss": 0.8739,
      "step": 3106
    },
    {
      "epoch": 0.9810546258288602,
      "grad_norm": 1.0945318937301636,
      "learning_rate": 4.4656515332257384e-05,
      "loss": 0.8826,
      "step": 3107
    },
    {
      "epoch": 0.9813703820650458,
      "grad_norm": 1.073108196258545,
      "learning_rate": 4.465300680653989e-05,
      "loss": 0.871,
      "step": 3108
    },
    {
      "epoch": 0.9816861383012314,
      "grad_norm": 1.278303861618042,
      "learning_rate": 4.46494982808224e-05,
      "loss": 0.9331,
      "step": 3109
    },
    {
      "epoch": 0.9820018945374172,
      "grad_norm": 1.2647119760513306,
      "learning_rate": 4.4645989755104904e-05,
      "loss": 0.9044,
      "step": 3110
    },
    {
      "epoch": 0.9823176507736028,
      "grad_norm": 1.133901596069336,
      "learning_rate": 4.464248122938741e-05,
      "loss": 0.8648,
      "step": 3111
    },
    {
      "epoch": 0.9826334070097884,
      "grad_norm": 1.2925716638565063,
      "learning_rate": 4.463897270366992e-05,
      "loss": 0.8571,
      "step": 3112
    },
    {
      "epoch": 0.9829491632459741,
      "grad_norm": 1.0702893733978271,
      "learning_rate": 4.4635464177952425e-05,
      "loss": 0.8543,
      "step": 3113
    },
    {
      "epoch": 0.9832649194821598,
      "grad_norm": 1.1996163129806519,
      "learning_rate": 4.463195565223493e-05,
      "loss": 0.8872,
      "step": 3114
    },
    {
      "epoch": 0.9835806757183454,
      "grad_norm": 1.0276618003845215,
      "learning_rate": 4.4628447126517445e-05,
      "loss": 0.8419,
      "step": 3115
    },
    {
      "epoch": 0.9838964319545311,
      "grad_norm": 1.0363563299179077,
      "learning_rate": 4.4624938600799945e-05,
      "loss": 0.8802,
      "step": 3116
    },
    {
      "epoch": 0.9842121881907168,
      "grad_norm": 0.9711204171180725,
      "learning_rate": 4.462143007508245e-05,
      "loss": 0.8921,
      "step": 3117
    },
    {
      "epoch": 0.9845279444269024,
      "grad_norm": 1.0328432321548462,
      "learning_rate": 4.461792154936496e-05,
      "loss": 0.9167,
      "step": 3118
    },
    {
      "epoch": 0.9848437006630881,
      "grad_norm": 1.3616207838058472,
      "learning_rate": 4.4614413023647466e-05,
      "loss": 0.8659,
      "step": 3119
    },
    {
      "epoch": 0.9851594568992738,
      "grad_norm": 0.9992517828941345,
      "learning_rate": 4.461090449792997e-05,
      "loss": 0.8816,
      "step": 3120
    },
    {
      "epoch": 0.9854752131354594,
      "grad_norm": 0.9889560341835022,
      "learning_rate": 4.460739597221248e-05,
      "loss": 0.9135,
      "step": 3121
    },
    {
      "epoch": 0.9857909693716451,
      "grad_norm": 0.9427902698516846,
      "learning_rate": 4.4603887446494986e-05,
      "loss": 0.8265,
      "step": 3122
    },
    {
      "epoch": 0.9861067256078307,
      "grad_norm": 1.08927321434021,
      "learning_rate": 4.460037892077749e-05,
      "loss": 0.8775,
      "step": 3123
    },
    {
      "epoch": 0.9864224818440164,
      "grad_norm": 1.0443331003189087,
      "learning_rate": 4.459687039506e-05,
      "loss": 0.8777,
      "step": 3124
    },
    {
      "epoch": 0.9867382380802021,
      "grad_norm": 0.9435437321662903,
      "learning_rate": 4.4593361869342507e-05,
      "loss": 0.8597,
      "step": 3125
    },
    {
      "epoch": 0.9870539943163877,
      "grad_norm": 1.147938847541809,
      "learning_rate": 4.4589853343625007e-05,
      "loss": 0.8605,
      "step": 3126
    },
    {
      "epoch": 0.9873697505525734,
      "grad_norm": 1.281784176826477,
      "learning_rate": 4.4586344817907513e-05,
      "loss": 0.8756,
      "step": 3127
    },
    {
      "epoch": 0.9876855067887591,
      "grad_norm": 1.0001288652420044,
      "learning_rate": 4.458283629219002e-05,
      "loss": 0.9082,
      "step": 3128
    },
    {
      "epoch": 0.9880012630249447,
      "grad_norm": 1.252631425857544,
      "learning_rate": 4.457932776647253e-05,
      "loss": 0.8795,
      "step": 3129
    },
    {
      "epoch": 0.9883170192611304,
      "grad_norm": 1.1504164934158325,
      "learning_rate": 4.4575819240755034e-05,
      "loss": 0.9576,
      "step": 3130
    },
    {
      "epoch": 0.9886327754973161,
      "grad_norm": 1.1728312969207764,
      "learning_rate": 4.457231071503754e-05,
      "loss": 0.9118,
      "step": 3131
    },
    {
      "epoch": 0.9889485317335017,
      "grad_norm": 1.2717264890670776,
      "learning_rate": 4.4568802189320054e-05,
      "loss": 0.873,
      "step": 3132
    },
    {
      "epoch": 0.9892642879696874,
      "grad_norm": 1.2330684661865234,
      "learning_rate": 4.456529366360256e-05,
      "loss": 0.8637,
      "step": 3133
    },
    {
      "epoch": 0.9895800442058731,
      "grad_norm": 1.0636215209960938,
      "learning_rate": 4.456178513788507e-05,
      "loss": 0.8663,
      "step": 3134
    },
    {
      "epoch": 0.9898958004420587,
      "grad_norm": 0.9398119449615479,
      "learning_rate": 4.455827661216757e-05,
      "loss": 0.8413,
      "step": 3135
    },
    {
      "epoch": 0.9902115566782443,
      "grad_norm": 1.0919666290283203,
      "learning_rate": 4.4554768086450075e-05,
      "loss": 0.9045,
      "step": 3136
    },
    {
      "epoch": 0.9905273129144301,
      "grad_norm": 1.0781211853027344,
      "learning_rate": 4.455125956073258e-05,
      "loss": 0.9192,
      "step": 3137
    },
    {
      "epoch": 0.9908430691506157,
      "grad_norm": 1.151046633720398,
      "learning_rate": 4.454775103501509e-05,
      "loss": 0.8943,
      "step": 3138
    },
    {
      "epoch": 0.9911588253868013,
      "grad_norm": 1.112903356552124,
      "learning_rate": 4.4544242509297595e-05,
      "loss": 0.918,
      "step": 3139
    },
    {
      "epoch": 0.9914745816229871,
      "grad_norm": 0.9539350271224976,
      "learning_rate": 4.45407339835801e-05,
      "loss": 0.9308,
      "step": 3140
    },
    {
      "epoch": 0.9917903378591727,
      "grad_norm": 1.0617414712905884,
      "learning_rate": 4.453722545786261e-05,
      "loss": 0.8649,
      "step": 3141
    },
    {
      "epoch": 0.9921060940953584,
      "grad_norm": 1.205958366394043,
      "learning_rate": 4.4533716932145116e-05,
      "loss": 0.8594,
      "step": 3142
    },
    {
      "epoch": 0.9924218503315441,
      "grad_norm": 1.0860803127288818,
      "learning_rate": 4.453020840642762e-05,
      "loss": 0.8532,
      "step": 3143
    },
    {
      "epoch": 0.9927376065677297,
      "grad_norm": 1.2592012882232666,
      "learning_rate": 4.452669988071013e-05,
      "loss": 0.8826,
      "step": 3144
    },
    {
      "epoch": 0.9930533628039154,
      "grad_norm": 1.097312569618225,
      "learning_rate": 4.4523191354992636e-05,
      "loss": 0.8775,
      "step": 3145
    },
    {
      "epoch": 0.9933691190401011,
      "grad_norm": 1.0909771919250488,
      "learning_rate": 4.4519682829275136e-05,
      "loss": 0.9318,
      "step": 3146
    },
    {
      "epoch": 0.9936848752762867,
      "grad_norm": 1.078808069229126,
      "learning_rate": 4.451617430355764e-05,
      "loss": 0.9164,
      "step": 3147
    },
    {
      "epoch": 0.9940006315124724,
      "grad_norm": 1.063821792602539,
      "learning_rate": 4.451266577784015e-05,
      "loss": 0.8572,
      "step": 3148
    },
    {
      "epoch": 0.994316387748658,
      "grad_norm": 1.1115081310272217,
      "learning_rate": 4.450915725212266e-05,
      "loss": 0.9147,
      "step": 3149
    },
    {
      "epoch": 0.9946321439848437,
      "grad_norm": 1.2524313926696777,
      "learning_rate": 4.450564872640517e-05,
      "loss": 0.8936,
      "step": 3150
    },
    {
      "epoch": 0.9949479002210294,
      "grad_norm": 1.0723297595977783,
      "learning_rate": 4.450214020068768e-05,
      "loss": 0.9421,
      "step": 3151
    },
    {
      "epoch": 0.995263656457215,
      "grad_norm": 1.1450793743133545,
      "learning_rate": 4.4498631674970184e-05,
      "loss": 0.8138,
      "step": 3152
    },
    {
      "epoch": 0.9955794126934007,
      "grad_norm": 1.2208491563796997,
      "learning_rate": 4.449512314925269e-05,
      "loss": 0.8818,
      "step": 3153
    },
    {
      "epoch": 0.9958951689295864,
      "grad_norm": 0.9384205937385559,
      "learning_rate": 4.449161462353519e-05,
      "loss": 0.8738,
      "step": 3154
    },
    {
      "epoch": 0.996210925165772,
      "grad_norm": 1.1036208868026733,
      "learning_rate": 4.44881060978177e-05,
      "loss": 0.8403,
      "step": 3155
    },
    {
      "epoch": 0.9965266814019577,
      "grad_norm": 1.0059205293655396,
      "learning_rate": 4.4484597572100204e-05,
      "loss": 0.9184,
      "step": 3156
    },
    {
      "epoch": 0.9968424376381434,
      "grad_norm": 1.2179286479949951,
      "learning_rate": 4.448108904638271e-05,
      "loss": 0.9266,
      "step": 3157
    },
    {
      "epoch": 0.997158193874329,
      "grad_norm": 0.924884557723999,
      "learning_rate": 4.447758052066522e-05,
      "loss": 0.9336,
      "step": 3158
    },
    {
      "epoch": 0.9974739501105147,
      "grad_norm": 1.2534922361373901,
      "learning_rate": 4.4474071994947725e-05,
      "loss": 0.9042,
      "step": 3159
    },
    {
      "epoch": 0.9977897063467004,
      "grad_norm": 1.4259549379348755,
      "learning_rate": 4.447056346923023e-05,
      "loss": 0.8966,
      "step": 3160
    },
    {
      "epoch": 0.998105462582886,
      "grad_norm": 1.2186243534088135,
      "learning_rate": 4.446705494351274e-05,
      "loss": 0.897,
      "step": 3161
    },
    {
      "epoch": 0.9984212188190716,
      "grad_norm": 1.2698390483856201,
      "learning_rate": 4.4463546417795245e-05,
      "loss": 0.889,
      "step": 3162
    },
    {
      "epoch": 0.9987369750552574,
      "grad_norm": 1.0602991580963135,
      "learning_rate": 4.446003789207775e-05,
      "loss": 0.8951,
      "step": 3163
    },
    {
      "epoch": 0.999052731291443,
      "grad_norm": 1.7816029787063599,
      "learning_rate": 4.445652936636026e-05,
      "loss": 0.9149,
      "step": 3164
    },
    {
      "epoch": 0.9993684875276286,
      "grad_norm": 1.0922999382019043,
      "learning_rate": 4.445302084064276e-05,
      "loss": 0.8784,
      "step": 3165
    },
    {
      "epoch": 0.9996842437638144,
      "grad_norm": 1.0448654890060425,
      "learning_rate": 4.4449512314925266e-05,
      "loss": 0.8933,
      "step": 3166
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1639585494995117,
      "learning_rate": 4.444600378920778e-05,
      "loss": 0.8705,
      "step": 3167
    },
    {
      "epoch": 1.0003157562361857,
      "grad_norm": 1.106607437133789,
      "learning_rate": 4.4442495263490286e-05,
      "loss": 0.8889,
      "step": 3168
    },
    {
      "epoch": 1.0006315124723713,
      "grad_norm": 1.1627647876739502,
      "learning_rate": 4.443898673777279e-05,
      "loss": 0.8879,
      "step": 3169
    },
    {
      "epoch": 1.000947268708557,
      "grad_norm": 1.2234270572662354,
      "learning_rate": 4.44354782120553e-05,
      "loss": 0.8359,
      "step": 3170
    },
    {
      "epoch": 1.0012630249447427,
      "grad_norm": 1.1808621883392334,
      "learning_rate": 4.4431969686337806e-05,
      "loss": 0.8842,
      "step": 3171
    },
    {
      "epoch": 1.0015787811809282,
      "grad_norm": 1.2540806531906128,
      "learning_rate": 4.442846116062031e-05,
      "loss": 0.825,
      "step": 3172
    },
    {
      "epoch": 1.001894537417114,
      "grad_norm": 1.325872778892517,
      "learning_rate": 4.442495263490282e-05,
      "loss": 0.8334,
      "step": 3173
    },
    {
      "epoch": 1.0022102936532997,
      "grad_norm": 1.230893850326538,
      "learning_rate": 4.442144410918532e-05,
      "loss": 0.8434,
      "step": 3174
    },
    {
      "epoch": 1.0025260498894852,
      "grad_norm": 1.039170742034912,
      "learning_rate": 4.441793558346783e-05,
      "loss": 0.8409,
      "step": 3175
    },
    {
      "epoch": 1.002841806125671,
      "grad_norm": 1.3620980978012085,
      "learning_rate": 4.4414427057750334e-05,
      "loss": 0.8731,
      "step": 3176
    },
    {
      "epoch": 1.0031575623618567,
      "grad_norm": 0.9641645550727844,
      "learning_rate": 4.441091853203284e-05,
      "loss": 0.8708,
      "step": 3177
    },
    {
      "epoch": 1.0034733185980422,
      "grad_norm": 1.121776819229126,
      "learning_rate": 4.440741000631535e-05,
      "loss": 0.9396,
      "step": 3178
    },
    {
      "epoch": 1.003789074834228,
      "grad_norm": 1.0840133428573608,
      "learning_rate": 4.4403901480597854e-05,
      "loss": 0.8089,
      "step": 3179
    },
    {
      "epoch": 1.0041048310704137,
      "grad_norm": 1.0955281257629395,
      "learning_rate": 4.440039295488036e-05,
      "loss": 0.903,
      "step": 3180
    },
    {
      "epoch": 1.0044205873065992,
      "grad_norm": 0.9368880391120911,
      "learning_rate": 4.439688442916287e-05,
      "loss": 0.8906,
      "step": 3181
    },
    {
      "epoch": 1.004736343542785,
      "grad_norm": 1.0063785314559937,
      "learning_rate": 4.4393375903445375e-05,
      "loss": 0.8429,
      "step": 3182
    },
    {
      "epoch": 1.0050520997789707,
      "grad_norm": 0.9729346632957458,
      "learning_rate": 4.438986737772788e-05,
      "loss": 0.9437,
      "step": 3183
    },
    {
      "epoch": 1.0053678560151562,
      "grad_norm": 1.236242651939392,
      "learning_rate": 4.438635885201039e-05,
      "loss": 0.823,
      "step": 3184
    },
    {
      "epoch": 1.005683612251342,
      "grad_norm": 1.2658931016921997,
      "learning_rate": 4.4382850326292895e-05,
      "loss": 0.9061,
      "step": 3185
    },
    {
      "epoch": 1.0059993684875277,
      "grad_norm": 1.1356275081634521,
      "learning_rate": 4.43793418005754e-05,
      "loss": 0.9373,
      "step": 3186
    },
    {
      "epoch": 1.0063151247237132,
      "grad_norm": 1.1183698177337646,
      "learning_rate": 4.437583327485791e-05,
      "loss": 0.8169,
      "step": 3187
    },
    {
      "epoch": 1.006630880959899,
      "grad_norm": 1.1950238943099976,
      "learning_rate": 4.4372324749140415e-05,
      "loss": 0.802,
      "step": 3188
    },
    {
      "epoch": 1.0069466371960847,
      "grad_norm": 1.0156461000442505,
      "learning_rate": 4.436881622342292e-05,
      "loss": 0.8608,
      "step": 3189
    },
    {
      "epoch": 1.0072623934322702,
      "grad_norm": 1.030962586402893,
      "learning_rate": 4.436530769770543e-05,
      "loss": 0.8545,
      "step": 3190
    },
    {
      "epoch": 1.007578149668456,
      "grad_norm": 1.0515811443328857,
      "learning_rate": 4.4361799171987936e-05,
      "loss": 0.8962,
      "step": 3191
    },
    {
      "epoch": 1.0078939059046417,
      "grad_norm": 1.4673593044281006,
      "learning_rate": 4.435829064627044e-05,
      "loss": 0.7909,
      "step": 3192
    },
    {
      "epoch": 1.0082096621408272,
      "grad_norm": 1.0957027673721313,
      "learning_rate": 4.435478212055294e-05,
      "loss": 0.8516,
      "step": 3193
    },
    {
      "epoch": 1.008525418377013,
      "grad_norm": 1.2799077033996582,
      "learning_rate": 4.435127359483545e-05,
      "loss": 0.8617,
      "step": 3194
    },
    {
      "epoch": 1.0088411746131987,
      "grad_norm": 1.069575309753418,
      "learning_rate": 4.4347765069117956e-05,
      "loss": 0.8383,
      "step": 3195
    },
    {
      "epoch": 1.0091569308493842,
      "grad_norm": 1.164042353630066,
      "learning_rate": 4.434425654340046e-05,
      "loss": 0.7992,
      "step": 3196
    },
    {
      "epoch": 1.00947268708557,
      "grad_norm": 1.1451523303985596,
      "learning_rate": 4.434074801768297e-05,
      "loss": 0.8701,
      "step": 3197
    },
    {
      "epoch": 1.0097884433217557,
      "grad_norm": 1.1476101875305176,
      "learning_rate": 4.433723949196548e-05,
      "loss": 0.8131,
      "step": 3198
    },
    {
      "epoch": 1.0101041995579412,
      "grad_norm": 1.1933821439743042,
      "learning_rate": 4.4333730966247984e-05,
      "loss": 0.7947,
      "step": 3199
    },
    {
      "epoch": 1.010419955794127,
      "grad_norm": 1.2232110500335693,
      "learning_rate": 4.433022244053049e-05,
      "loss": 0.8715,
      "step": 3200
    },
    {
      "epoch": 1.010419955794127,
      "eval_loss": 0.8396367430686951,
      "eval_runtime": 4.6603,
      "eval_samples_per_second": 534.303,
      "eval_steps_per_second": 4.506,
      "step": 3200
    },
    {
      "epoch": 1.0107357120303126,
      "grad_norm": 1.2972713708877563,
      "learning_rate": 4.4326713914813e-05,
      "loss": 0.8785,
      "step": 3201
    },
    {
      "epoch": 1.0110514682664982,
      "grad_norm": 1.3039240837097168,
      "learning_rate": 4.4323205389095504e-05,
      "loss": 0.8956,
      "step": 3202
    },
    {
      "epoch": 1.011367224502684,
      "grad_norm": 1.500199317932129,
      "learning_rate": 4.431969686337801e-05,
      "loss": 0.8917,
      "step": 3203
    },
    {
      "epoch": 1.0116829807388696,
      "grad_norm": 1.2026387453079224,
      "learning_rate": 4.431618833766052e-05,
      "loss": 0.7877,
      "step": 3204
    },
    {
      "epoch": 1.0119987369750552,
      "grad_norm": 1.2348090410232544,
      "learning_rate": 4.4312679811943024e-05,
      "loss": 0.8704,
      "step": 3205
    },
    {
      "epoch": 1.012314493211241,
      "grad_norm": 1.3527272939682007,
      "learning_rate": 4.430917128622553e-05,
      "loss": 0.8708,
      "step": 3206
    },
    {
      "epoch": 1.0126302494474266,
      "grad_norm": 1.01642644405365,
      "learning_rate": 4.430566276050804e-05,
      "loss": 0.8729,
      "step": 3207
    },
    {
      "epoch": 1.0129460056836121,
      "grad_norm": 1.1509177684783936,
      "learning_rate": 4.4302154234790545e-05,
      "loss": 0.8079,
      "step": 3208
    },
    {
      "epoch": 1.0132617619197979,
      "grad_norm": 1.5998523235321045,
      "learning_rate": 4.429864570907305e-05,
      "loss": 0.8139,
      "step": 3209
    },
    {
      "epoch": 1.0135775181559836,
      "grad_norm": 1.1272422075271606,
      "learning_rate": 4.429513718335556e-05,
      "loss": 0.8522,
      "step": 3210
    },
    {
      "epoch": 1.0138932743921691,
      "grad_norm": 1.3864593505859375,
      "learning_rate": 4.4291628657638065e-05,
      "loss": 0.8801,
      "step": 3211
    },
    {
      "epoch": 1.0142090306283549,
      "grad_norm": 1.3164728879928589,
      "learning_rate": 4.4288120131920565e-05,
      "loss": 0.8373,
      "step": 3212
    },
    {
      "epoch": 1.0145247868645406,
      "grad_norm": 1.2056610584259033,
      "learning_rate": 4.428461160620307e-05,
      "loss": 0.8224,
      "step": 3213
    },
    {
      "epoch": 1.0148405431007261,
      "grad_norm": 1.3331960439682007,
      "learning_rate": 4.428110308048558e-05,
      "loss": 0.8707,
      "step": 3214
    },
    {
      "epoch": 1.0151562993369119,
      "grad_norm": 1.2001482248306274,
      "learning_rate": 4.4277594554768086e-05,
      "loss": 0.8547,
      "step": 3215
    },
    {
      "epoch": 1.0154720555730976,
      "grad_norm": 1.3217540979385376,
      "learning_rate": 4.427408602905059e-05,
      "loss": 0.8828,
      "step": 3216
    },
    {
      "epoch": 1.0157878118092831,
      "grad_norm": 1.38961660861969,
      "learning_rate": 4.42705775033331e-05,
      "loss": 0.8994,
      "step": 3217
    },
    {
      "epoch": 1.0161035680454689,
      "grad_norm": 1.268365740776062,
      "learning_rate": 4.4267068977615606e-05,
      "loss": 0.8184,
      "step": 3218
    },
    {
      "epoch": 1.0164193242816546,
      "grad_norm": 1.3952440023422241,
      "learning_rate": 4.426356045189812e-05,
      "loss": 0.8781,
      "step": 3219
    },
    {
      "epoch": 1.0167350805178401,
      "grad_norm": 1.2727631330490112,
      "learning_rate": 4.426005192618063e-05,
      "loss": 0.8625,
      "step": 3220
    },
    {
      "epoch": 1.0170508367540259,
      "grad_norm": 1.1898590326309204,
      "learning_rate": 4.425654340046313e-05,
      "loss": 0.8389,
      "step": 3221
    },
    {
      "epoch": 1.0173665929902116,
      "grad_norm": 1.359311819076538,
      "learning_rate": 4.4253034874745634e-05,
      "loss": 0.8384,
      "step": 3222
    },
    {
      "epoch": 1.0176823492263973,
      "grad_norm": 1.4043477773666382,
      "learning_rate": 4.424952634902814e-05,
      "loss": 0.7932,
      "step": 3223
    },
    {
      "epoch": 1.0179981054625828,
      "grad_norm": 1.1020665168762207,
      "learning_rate": 4.424601782331065e-05,
      "loss": 0.8534,
      "step": 3224
    },
    {
      "epoch": 1.0183138616987686,
      "grad_norm": 1.106798768043518,
      "learning_rate": 4.4242509297593154e-05,
      "loss": 0.8669,
      "step": 3225
    },
    {
      "epoch": 1.0186296179349543,
      "grad_norm": 1.4019180536270142,
      "learning_rate": 4.423900077187566e-05,
      "loss": 0.8856,
      "step": 3226
    },
    {
      "epoch": 1.0189453741711398,
      "grad_norm": 1.373473882675171,
      "learning_rate": 4.423549224615817e-05,
      "loss": 0.8229,
      "step": 3227
    },
    {
      "epoch": 1.0192611304073256,
      "grad_norm": 1.163430094718933,
      "learning_rate": 4.4231983720440674e-05,
      "loss": 0.8692,
      "step": 3228
    },
    {
      "epoch": 1.0195768866435113,
      "grad_norm": 1.156232237815857,
      "learning_rate": 4.422847519472318e-05,
      "loss": 0.8534,
      "step": 3229
    },
    {
      "epoch": 1.0198926428796968,
      "grad_norm": 1.3859307765960693,
      "learning_rate": 4.422496666900569e-05,
      "loss": 0.8811,
      "step": 3230
    },
    {
      "epoch": 1.0202083991158826,
      "grad_norm": 1.5391151905059814,
      "learning_rate": 4.422145814328819e-05,
      "loss": 0.8684,
      "step": 3231
    },
    {
      "epoch": 1.0205241553520683,
      "grad_norm": 1.2807817459106445,
      "learning_rate": 4.4217949617570695e-05,
      "loss": 0.8317,
      "step": 3232
    },
    {
      "epoch": 1.0208399115882538,
      "grad_norm": 1.3482948541641235,
      "learning_rate": 4.42144410918532e-05,
      "loss": 0.8104,
      "step": 3233
    },
    {
      "epoch": 1.0211556678244396,
      "grad_norm": 1.3984806537628174,
      "learning_rate": 4.421093256613571e-05,
      "loss": 0.8299,
      "step": 3234
    },
    {
      "epoch": 1.0214714240606253,
      "grad_norm": 1.2981865406036377,
      "learning_rate": 4.4207424040418215e-05,
      "loss": 0.8721,
      "step": 3235
    },
    {
      "epoch": 1.0217871802968108,
      "grad_norm": 1.4369895458221436,
      "learning_rate": 4.420391551470072e-05,
      "loss": 0.7653,
      "step": 3236
    },
    {
      "epoch": 1.0221029365329966,
      "grad_norm": 1.4937692880630493,
      "learning_rate": 4.4200406988983236e-05,
      "loss": 0.9208,
      "step": 3237
    },
    {
      "epoch": 1.0224186927691823,
      "grad_norm": 1.4507741928100586,
      "learning_rate": 4.419689846326574e-05,
      "loss": 0.8706,
      "step": 3238
    },
    {
      "epoch": 1.0227344490053678,
      "grad_norm": 1.290263295173645,
      "learning_rate": 4.419338993754825e-05,
      "loss": 0.8621,
      "step": 3239
    },
    {
      "epoch": 1.0230502052415535,
      "grad_norm": 1.428330898284912,
      "learning_rate": 4.418988141183075e-05,
      "loss": 0.8164,
      "step": 3240
    },
    {
      "epoch": 1.0233659614777393,
      "grad_norm": 1.4229470491409302,
      "learning_rate": 4.4186372886113256e-05,
      "loss": 0.9541,
      "step": 3241
    },
    {
      "epoch": 1.0236817177139248,
      "grad_norm": 1.4124921560287476,
      "learning_rate": 4.418286436039576e-05,
      "loss": 0.8621,
      "step": 3242
    },
    {
      "epoch": 1.0239974739501105,
      "grad_norm": 1.2276829481124878,
      "learning_rate": 4.417935583467827e-05,
      "loss": 0.8246,
      "step": 3243
    },
    {
      "epoch": 1.0243132301862963,
      "grad_norm": 1.2962654829025269,
      "learning_rate": 4.417584730896078e-05,
      "loss": 0.8164,
      "step": 3244
    },
    {
      "epoch": 1.0246289864224818,
      "grad_norm": 1.2847826480865479,
      "learning_rate": 4.4172338783243283e-05,
      "loss": 0.8887,
      "step": 3245
    },
    {
      "epoch": 1.0249447426586675,
      "grad_norm": 1.9985677003860474,
      "learning_rate": 4.416883025752579e-05,
      "loss": 0.8397,
      "step": 3246
    },
    {
      "epoch": 1.0252604988948533,
      "grad_norm": 1.6124573945999146,
      "learning_rate": 4.41653217318083e-05,
      "loss": 0.8591,
      "step": 3247
    },
    {
      "epoch": 1.0255762551310388,
      "grad_norm": 1.2783136367797852,
      "learning_rate": 4.4161813206090804e-05,
      "loss": 0.8953,
      "step": 3248
    },
    {
      "epoch": 1.0258920113672245,
      "grad_norm": 1.29800546169281,
      "learning_rate": 4.415830468037331e-05,
      "loss": 0.8929,
      "step": 3249
    },
    {
      "epoch": 1.0262077676034103,
      "grad_norm": 1.3556519746780396,
      "learning_rate": 4.415479615465582e-05,
      "loss": 0.8085,
      "step": 3250
    },
    {
      "epoch": 1.0265235238395958,
      "grad_norm": 1.5163060426712036,
      "learning_rate": 4.415128762893832e-05,
      "loss": 0.9056,
      "step": 3251
    },
    {
      "epoch": 1.0268392800757815,
      "grad_norm": 1.5000189542770386,
      "learning_rate": 4.4147779103220824e-05,
      "loss": 0.8494,
      "step": 3252
    },
    {
      "epoch": 1.0271550363119673,
      "grad_norm": 1.268557071685791,
      "learning_rate": 4.414427057750333e-05,
      "loss": 0.8372,
      "step": 3253
    },
    {
      "epoch": 1.0274707925481528,
      "grad_norm": 1.246187448501587,
      "learning_rate": 4.4140762051785845e-05,
      "loss": 0.9085,
      "step": 3254
    },
    {
      "epoch": 1.0277865487843385,
      "grad_norm": 1.255090594291687,
      "learning_rate": 4.413725352606835e-05,
      "loss": 0.8853,
      "step": 3255
    },
    {
      "epoch": 1.0281023050205242,
      "grad_norm": 1.2252596616744995,
      "learning_rate": 4.413374500035086e-05,
      "loss": 0.8333,
      "step": 3256
    },
    {
      "epoch": 1.0284180612567098,
      "grad_norm": 1.4739065170288086,
      "learning_rate": 4.4130236474633365e-05,
      "loss": 0.883,
      "step": 3257
    },
    {
      "epoch": 1.0287338174928955,
      "grad_norm": 1.2338789701461792,
      "learning_rate": 4.412672794891587e-05,
      "loss": 0.8682,
      "step": 3258
    },
    {
      "epoch": 1.0290495737290812,
      "grad_norm": 1.309783935546875,
      "learning_rate": 4.412321942319837e-05,
      "loss": 0.866,
      "step": 3259
    },
    {
      "epoch": 1.0293653299652668,
      "grad_norm": 1.1271470785140991,
      "learning_rate": 4.411971089748088e-05,
      "loss": 0.7779,
      "step": 3260
    },
    {
      "epoch": 1.0296810862014525,
      "grad_norm": 1.2083791494369507,
      "learning_rate": 4.4116202371763386e-05,
      "loss": 0.8878,
      "step": 3261
    },
    {
      "epoch": 1.0299968424376382,
      "grad_norm": 1.191327452659607,
      "learning_rate": 4.411269384604589e-05,
      "loss": 0.8851,
      "step": 3262
    },
    {
      "epoch": 1.0303125986738237,
      "grad_norm": 1.263386607170105,
      "learning_rate": 4.41091853203284e-05,
      "loss": 0.9452,
      "step": 3263
    },
    {
      "epoch": 1.0306283549100095,
      "grad_norm": 1.3068941831588745,
      "learning_rate": 4.4105676794610906e-05,
      "loss": 0.868,
      "step": 3264
    },
    {
      "epoch": 1.0309441111461952,
      "grad_norm": 1.0483072996139526,
      "learning_rate": 4.410216826889341e-05,
      "loss": 0.819,
      "step": 3265
    },
    {
      "epoch": 1.0312598673823807,
      "grad_norm": 1.2853360176086426,
      "learning_rate": 4.409865974317592e-05,
      "loss": 0.8359,
      "step": 3266
    },
    {
      "epoch": 1.0315756236185665,
      "grad_norm": 1.514740228652954,
      "learning_rate": 4.4095151217458427e-05,
      "loss": 0.8272,
      "step": 3267
    },
    {
      "epoch": 1.0318913798547522,
      "grad_norm": 1.3474760055541992,
      "learning_rate": 4.409164269174093e-05,
      "loss": 0.9373,
      "step": 3268
    },
    {
      "epoch": 1.0322071360909377,
      "grad_norm": 1.1786006689071655,
      "learning_rate": 4.408813416602344e-05,
      "loss": 0.821,
      "step": 3269
    },
    {
      "epoch": 1.0325228923271235,
      "grad_norm": 1.4078422784805298,
      "learning_rate": 4.408462564030594e-05,
      "loss": 0.896,
      "step": 3270
    },
    {
      "epoch": 1.0328386485633092,
      "grad_norm": 1.2616363763809204,
      "learning_rate": 4.408111711458845e-05,
      "loss": 0.8497,
      "step": 3271
    },
    {
      "epoch": 1.0331544047994947,
      "grad_norm": 1.3236106634140015,
      "learning_rate": 4.407760858887096e-05,
      "loss": 0.8221,
      "step": 3272
    },
    {
      "epoch": 1.0334701610356805,
      "grad_norm": 1.2116938829421997,
      "learning_rate": 4.407410006315347e-05,
      "loss": 0.8542,
      "step": 3273
    },
    {
      "epoch": 1.0337859172718662,
      "grad_norm": 1.3427273035049438,
      "learning_rate": 4.4070591537435974e-05,
      "loss": 0.8377,
      "step": 3274
    },
    {
      "epoch": 1.0341016735080517,
      "grad_norm": 1.0970596075057983,
      "learning_rate": 4.406708301171848e-05,
      "loss": 0.8328,
      "step": 3275
    },
    {
      "epoch": 1.0344174297442374,
      "grad_norm": 1.1833012104034424,
      "learning_rate": 4.406357448600099e-05,
      "loss": 0.8455,
      "step": 3276
    },
    {
      "epoch": 1.0347331859804232,
      "grad_norm": 1.179491639137268,
      "learning_rate": 4.4060065960283495e-05,
      "loss": 0.8034,
      "step": 3277
    },
    {
      "epoch": 1.0350489422166087,
      "grad_norm": 1.3461120128631592,
      "learning_rate": 4.4056557434566e-05,
      "loss": 0.7937,
      "step": 3278
    },
    {
      "epoch": 1.0353646984527944,
      "grad_norm": 1.244369387626648,
      "learning_rate": 4.40530489088485e-05,
      "loss": 0.8625,
      "step": 3279
    },
    {
      "epoch": 1.0356804546889802,
      "grad_norm": 1.2200156450271606,
      "learning_rate": 4.404954038313101e-05,
      "loss": 0.7922,
      "step": 3280
    },
    {
      "epoch": 1.0359962109251657,
      "grad_norm": 1.3859699964523315,
      "learning_rate": 4.4046031857413515e-05,
      "loss": 0.9372,
      "step": 3281
    },
    {
      "epoch": 1.0363119671613514,
      "grad_norm": 1.2536747455596924,
      "learning_rate": 4.404252333169602e-05,
      "loss": 0.8621,
      "step": 3282
    },
    {
      "epoch": 1.0366277233975372,
      "grad_norm": 1.2468756437301636,
      "learning_rate": 4.403901480597853e-05,
      "loss": 0.8688,
      "step": 3283
    },
    {
      "epoch": 1.0369434796337227,
      "grad_norm": 1.3797259330749512,
      "learning_rate": 4.4035506280261036e-05,
      "loss": 0.9488,
      "step": 3284
    },
    {
      "epoch": 1.0372592358699084,
      "grad_norm": 1.393237590789795,
      "learning_rate": 4.403199775454354e-05,
      "loss": 0.8668,
      "step": 3285
    },
    {
      "epoch": 1.0375749921060942,
      "grad_norm": 1.4611796140670776,
      "learning_rate": 4.402848922882605e-05,
      "loss": 0.8835,
      "step": 3286
    },
    {
      "epoch": 1.0378907483422797,
      "grad_norm": 1.51382577419281,
      "learning_rate": 4.4024980703108556e-05,
      "loss": 0.8849,
      "step": 3287
    },
    {
      "epoch": 1.0382065045784654,
      "grad_norm": 1.4296114444732666,
      "learning_rate": 4.402147217739106e-05,
      "loss": 0.8811,
      "step": 3288
    },
    {
      "epoch": 1.0385222608146512,
      "grad_norm": 2.220496416091919,
      "learning_rate": 4.401796365167357e-05,
      "loss": 0.8433,
      "step": 3289
    },
    {
      "epoch": 1.0388380170508367,
      "grad_norm": 1.3187059164047241,
      "learning_rate": 4.4014455125956076e-05,
      "loss": 0.8768,
      "step": 3290
    },
    {
      "epoch": 1.0391537732870224,
      "grad_norm": 1.3587061166763306,
      "learning_rate": 4.401094660023858e-05,
      "loss": 0.8554,
      "step": 3291
    },
    {
      "epoch": 1.0394695295232081,
      "grad_norm": 1.479017972946167,
      "learning_rate": 4.400743807452109e-05,
      "loss": 0.943,
      "step": 3292
    },
    {
      "epoch": 1.0397852857593937,
      "grad_norm": 1.5564839839935303,
      "learning_rate": 4.40039295488036e-05,
      "loss": 0.8778,
      "step": 3293
    },
    {
      "epoch": 1.0401010419955794,
      "grad_norm": 1.40032160282135,
      "learning_rate": 4.4000421023086104e-05,
      "loss": 0.8404,
      "step": 3294
    },
    {
      "epoch": 1.0404167982317651,
      "grad_norm": 1.1947484016418457,
      "learning_rate": 4.399691249736861e-05,
      "loss": 0.9276,
      "step": 3295
    },
    {
      "epoch": 1.0407325544679507,
      "grad_norm": 1.100346565246582,
      "learning_rate": 4.399340397165112e-05,
      "loss": 0.8725,
      "step": 3296
    },
    {
      "epoch": 1.0410483107041364,
      "grad_norm": 1.5469601154327393,
      "learning_rate": 4.3989895445933624e-05,
      "loss": 0.8667,
      "step": 3297
    },
    {
      "epoch": 1.0413640669403221,
      "grad_norm": 1.291925072669983,
      "learning_rate": 4.3986386920216124e-05,
      "loss": 0.7963,
      "step": 3298
    },
    {
      "epoch": 1.0416798231765076,
      "grad_norm": 1.4034690856933594,
      "learning_rate": 4.398287839449863e-05,
      "loss": 0.9709,
      "step": 3299
    },
    {
      "epoch": 1.0419955794126934,
      "grad_norm": 1.3800048828125,
      "learning_rate": 4.397936986878114e-05,
      "loss": 0.923,
      "step": 3300
    },
    {
      "epoch": 1.0423113356488791,
      "grad_norm": 1.1837154626846313,
      "learning_rate": 4.3975861343063645e-05,
      "loss": 0.8606,
      "step": 3301
    },
    {
      "epoch": 1.0426270918850646,
      "grad_norm": 1.3955007791519165,
      "learning_rate": 4.397235281734615e-05,
      "loss": 0.8199,
      "step": 3302
    },
    {
      "epoch": 1.0429428481212504,
      "grad_norm": 1.3981908559799194,
      "learning_rate": 4.396884429162866e-05,
      "loss": 0.8708,
      "step": 3303
    },
    {
      "epoch": 1.0432586043574361,
      "grad_norm": 1.1202657222747803,
      "learning_rate": 4.3965335765911165e-05,
      "loss": 0.8313,
      "step": 3304
    },
    {
      "epoch": 1.0435743605936216,
      "grad_norm": 1.3202056884765625,
      "learning_rate": 4.396182724019367e-05,
      "loss": 0.8544,
      "step": 3305
    },
    {
      "epoch": 1.0438901168298074,
      "grad_norm": 1.3341093063354492,
      "learning_rate": 4.395831871447618e-05,
      "loss": 0.8726,
      "step": 3306
    },
    {
      "epoch": 1.044205873065993,
      "grad_norm": 1.3478128910064697,
      "learning_rate": 4.3954810188758686e-05,
      "loss": 0.8777,
      "step": 3307
    },
    {
      "epoch": 1.0445216293021786,
      "grad_norm": 1.4499539136886597,
      "learning_rate": 4.395130166304119e-05,
      "loss": 0.8124,
      "step": 3308
    },
    {
      "epoch": 1.0448373855383644,
      "grad_norm": 1.3064926862716675,
      "learning_rate": 4.39477931373237e-05,
      "loss": 0.8058,
      "step": 3309
    },
    {
      "epoch": 1.04515314177455,
      "grad_norm": 1.4711309671401978,
      "learning_rate": 4.3944284611606206e-05,
      "loss": 0.8601,
      "step": 3310
    },
    {
      "epoch": 1.0454688980107356,
      "grad_norm": 1.4210822582244873,
      "learning_rate": 4.394077608588871e-05,
      "loss": 0.8111,
      "step": 3311
    },
    {
      "epoch": 1.0457846542469214,
      "grad_norm": 1.3832855224609375,
      "learning_rate": 4.393726756017122e-05,
      "loss": 0.8311,
      "step": 3312
    },
    {
      "epoch": 1.046100410483107,
      "grad_norm": 1.1218352317810059,
      "learning_rate": 4.3933759034453726e-05,
      "loss": 0.8746,
      "step": 3313
    },
    {
      "epoch": 1.0464161667192926,
      "grad_norm": 1.3539345264434814,
      "learning_rate": 4.393025050873623e-05,
      "loss": 0.8357,
      "step": 3314
    },
    {
      "epoch": 1.0467319229554783,
      "grad_norm": 1.2961746454238892,
      "learning_rate": 4.392674198301874e-05,
      "loss": 0.9109,
      "step": 3315
    },
    {
      "epoch": 1.047047679191664,
      "grad_norm": 1.2215759754180908,
      "learning_rate": 4.392323345730125e-05,
      "loss": 0.8184,
      "step": 3316
    },
    {
      "epoch": 1.0473634354278496,
      "grad_norm": 1.5721526145935059,
      "learning_rate": 4.391972493158375e-05,
      "loss": 0.7561,
      "step": 3317
    },
    {
      "epoch": 1.0476791916640353,
      "grad_norm": 1.4535729885101318,
      "learning_rate": 4.3916216405866254e-05,
      "loss": 0.8846,
      "step": 3318
    },
    {
      "epoch": 1.047994947900221,
      "grad_norm": 1.2666404247283936,
      "learning_rate": 4.391270788014876e-05,
      "loss": 0.7964,
      "step": 3319
    },
    {
      "epoch": 1.0483107041364066,
      "grad_norm": 1.20407235622406,
      "learning_rate": 4.390919935443127e-05,
      "loss": 0.8409,
      "step": 3320
    },
    {
      "epoch": 1.0486264603725923,
      "grad_norm": 1.314408540725708,
      "learning_rate": 4.3905690828713774e-05,
      "loss": 0.8147,
      "step": 3321
    },
    {
      "epoch": 1.048942216608778,
      "grad_norm": 1.3043941259384155,
      "learning_rate": 4.390218230299628e-05,
      "loss": 0.9241,
      "step": 3322
    },
    {
      "epoch": 1.0492579728449636,
      "grad_norm": 1.4820588827133179,
      "learning_rate": 4.389867377727879e-05,
      "loss": 0.8413,
      "step": 3323
    },
    {
      "epoch": 1.0495737290811493,
      "grad_norm": 1.3611506223678589,
      "learning_rate": 4.38951652515613e-05,
      "loss": 0.8188,
      "step": 3324
    },
    {
      "epoch": 1.049889485317335,
      "grad_norm": 1.4361180067062378,
      "learning_rate": 4.389165672584381e-05,
      "loss": 0.8767,
      "step": 3325
    },
    {
      "epoch": 1.0502052415535208,
      "grad_norm": 1.3010070323944092,
      "learning_rate": 4.388814820012631e-05,
      "loss": 0.7982,
      "step": 3326
    },
    {
      "epoch": 1.0505209977897063,
      "grad_norm": 1.6405011415481567,
      "learning_rate": 4.3884639674408815e-05,
      "loss": 0.838,
      "step": 3327
    },
    {
      "epoch": 1.050836754025892,
      "grad_norm": 1.5822410583496094,
      "learning_rate": 4.388113114869132e-05,
      "loss": 0.8894,
      "step": 3328
    },
    {
      "epoch": 1.0511525102620776,
      "grad_norm": 1.4182753562927246,
      "learning_rate": 4.387762262297383e-05,
      "loss": 0.9232,
      "step": 3329
    },
    {
      "epoch": 1.0514682664982633,
      "grad_norm": 1.6888262033462524,
      "learning_rate": 4.3874114097256335e-05,
      "loss": 0.9018,
      "step": 3330
    },
    {
      "epoch": 1.051784022734449,
      "grad_norm": 1.3911628723144531,
      "learning_rate": 4.387060557153884e-05,
      "loss": 0.8564,
      "step": 3331
    },
    {
      "epoch": 1.0520997789706348,
      "grad_norm": 1.493026852607727,
      "learning_rate": 4.386709704582135e-05,
      "loss": 0.8478,
      "step": 3332
    },
    {
      "epoch": 1.0524155352068203,
      "grad_norm": 1.3774484395980835,
      "learning_rate": 4.3863588520103856e-05,
      "loss": 0.8876,
      "step": 3333
    },
    {
      "epoch": 1.052731291443006,
      "grad_norm": 1.4033831357955933,
      "learning_rate": 4.386007999438636e-05,
      "loss": 0.8229,
      "step": 3334
    },
    {
      "epoch": 1.0530470476791916,
      "grad_norm": 1.5387359857559204,
      "learning_rate": 4.385657146866887e-05,
      "loss": 0.8148,
      "step": 3335
    },
    {
      "epoch": 1.0533628039153773,
      "grad_norm": 1.387378215789795,
      "learning_rate": 4.385306294295137e-05,
      "loss": 0.8491,
      "step": 3336
    },
    {
      "epoch": 1.053678560151563,
      "grad_norm": 1.3238996267318726,
      "learning_rate": 4.3849554417233876e-05,
      "loss": 0.8295,
      "step": 3337
    },
    {
      "epoch": 1.0539943163877488,
      "grad_norm": 1.4051023721694946,
      "learning_rate": 4.384604589151638e-05,
      "loss": 0.9198,
      "step": 3338
    },
    {
      "epoch": 1.0543100726239343,
      "grad_norm": 1.420628309249878,
      "learning_rate": 4.384253736579889e-05,
      "loss": 0.9182,
      "step": 3339
    },
    {
      "epoch": 1.05462582886012,
      "grad_norm": 1.251360535621643,
      "learning_rate": 4.38390288400814e-05,
      "loss": 0.8383,
      "step": 3340
    },
    {
      "epoch": 1.0549415850963058,
      "grad_norm": 1.590808629989624,
      "learning_rate": 4.3835520314363904e-05,
      "loss": 0.9334,
      "step": 3341
    },
    {
      "epoch": 1.0552573413324913,
      "grad_norm": 1.2370381355285645,
      "learning_rate": 4.383201178864642e-05,
      "loss": 0.8745,
      "step": 3342
    },
    {
      "epoch": 1.055573097568677,
      "grad_norm": 1.045510172843933,
      "learning_rate": 4.3828503262928924e-05,
      "loss": 0.8337,
      "step": 3343
    },
    {
      "epoch": 1.0558888538048627,
      "grad_norm": 1.4270246028900146,
      "learning_rate": 4.382499473721143e-05,
      "loss": 0.8528,
      "step": 3344
    },
    {
      "epoch": 1.0562046100410483,
      "grad_norm": 1.3455708026885986,
      "learning_rate": 4.382148621149393e-05,
      "loss": 0.8228,
      "step": 3345
    },
    {
      "epoch": 1.056520366277234,
      "grad_norm": 1.4560869932174683,
      "learning_rate": 4.381797768577644e-05,
      "loss": 0.8249,
      "step": 3346
    },
    {
      "epoch": 1.0568361225134197,
      "grad_norm": 1.3973313570022583,
      "learning_rate": 4.3814469160058944e-05,
      "loss": 0.8908,
      "step": 3347
    },
    {
      "epoch": 1.0571518787496053,
      "grad_norm": 1.3589073419570923,
      "learning_rate": 4.381096063434145e-05,
      "loss": 0.8977,
      "step": 3348
    },
    {
      "epoch": 1.057467634985791,
      "grad_norm": 1.334989070892334,
      "learning_rate": 4.380745210862396e-05,
      "loss": 0.9203,
      "step": 3349
    },
    {
      "epoch": 1.0577833912219767,
      "grad_norm": 1.2191540002822876,
      "learning_rate": 4.3803943582906465e-05,
      "loss": 0.8528,
      "step": 3350
    },
    {
      "epoch": 1.0580991474581622,
      "grad_norm": 1.3586493730545044,
      "learning_rate": 4.380043505718897e-05,
      "loss": 0.8658,
      "step": 3351
    },
    {
      "epoch": 1.058414903694348,
      "grad_norm": 1.2337809801101685,
      "learning_rate": 4.379692653147148e-05,
      "loss": 0.8624,
      "step": 3352
    },
    {
      "epoch": 1.0587306599305337,
      "grad_norm": 1.6011161804199219,
      "learning_rate": 4.3793418005753985e-05,
      "loss": 0.8887,
      "step": 3353
    },
    {
      "epoch": 1.0590464161667192,
      "grad_norm": 1.3683332204818726,
      "learning_rate": 4.378990948003649e-05,
      "loss": 0.8659,
      "step": 3354
    },
    {
      "epoch": 1.059362172402905,
      "grad_norm": 1.3744690418243408,
      "learning_rate": 4.3786400954319e-05,
      "loss": 0.863,
      "step": 3355
    },
    {
      "epoch": 1.0596779286390907,
      "grad_norm": 1.5093733072280884,
      "learning_rate": 4.37828924286015e-05,
      "loss": 0.8698,
      "step": 3356
    },
    {
      "epoch": 1.0599936848752762,
      "grad_norm": 1.3363398313522339,
      "learning_rate": 4.3779383902884006e-05,
      "loss": 0.8613,
      "step": 3357
    },
    {
      "epoch": 1.060309441111462,
      "grad_norm": 1.3476241827011108,
      "learning_rate": 4.377587537716651e-05,
      "loss": 0.8248,
      "step": 3358
    },
    {
      "epoch": 1.0606251973476477,
      "grad_norm": 1.3230531215667725,
      "learning_rate": 4.3772366851449026e-05,
      "loss": 0.8026,
      "step": 3359
    },
    {
      "epoch": 1.0609409535838332,
      "grad_norm": 1.282724142074585,
      "learning_rate": 4.376885832573153e-05,
      "loss": 0.8134,
      "step": 3360
    },
    {
      "epoch": 1.061256709820019,
      "grad_norm": 1.4643778800964355,
      "learning_rate": 4.376534980001404e-05,
      "loss": 0.8718,
      "step": 3361
    },
    {
      "epoch": 1.0615724660562047,
      "grad_norm": 1.4662151336669922,
      "learning_rate": 4.376184127429655e-05,
      "loss": 0.9045,
      "step": 3362
    },
    {
      "epoch": 1.0618882222923902,
      "grad_norm": 1.2006057500839233,
      "learning_rate": 4.3758332748579053e-05,
      "loss": 0.8581,
      "step": 3363
    },
    {
      "epoch": 1.062203978528576,
      "grad_norm": 1.4852558374404907,
      "learning_rate": 4.3754824222861554e-05,
      "loss": 0.8768,
      "step": 3364
    },
    {
      "epoch": 1.0625197347647617,
      "grad_norm": 1.2792079448699951,
      "learning_rate": 4.375131569714406e-05,
      "loss": 0.7847,
      "step": 3365
    },
    {
      "epoch": 1.0628354910009472,
      "grad_norm": 1.5508906841278076,
      "learning_rate": 4.374780717142657e-05,
      "loss": 0.9128,
      "step": 3366
    },
    {
      "epoch": 1.063151247237133,
      "grad_norm": 1.2632700204849243,
      "learning_rate": 4.3744298645709074e-05,
      "loss": 0.8922,
      "step": 3367
    },
    {
      "epoch": 1.0634670034733187,
      "grad_norm": 1.3785881996154785,
      "learning_rate": 4.374079011999158e-05,
      "loss": 0.8297,
      "step": 3368
    },
    {
      "epoch": 1.0637827597095042,
      "grad_norm": 1.3067854642868042,
      "learning_rate": 4.373728159427409e-05,
      "loss": 0.8735,
      "step": 3369
    },
    {
      "epoch": 1.06409851594569,
      "grad_norm": 1.6235016584396362,
      "learning_rate": 4.3733773068556594e-05,
      "loss": 0.8285,
      "step": 3370
    },
    {
      "epoch": 1.0644142721818757,
      "grad_norm": 1.3836663961410522,
      "learning_rate": 4.37302645428391e-05,
      "loss": 0.8526,
      "step": 3371
    },
    {
      "epoch": 1.0647300284180612,
      "grad_norm": 1.431514024734497,
      "learning_rate": 4.372675601712161e-05,
      "loss": 0.8773,
      "step": 3372
    },
    {
      "epoch": 1.065045784654247,
      "grad_norm": 1.4515608549118042,
      "learning_rate": 4.3723247491404115e-05,
      "loss": 0.8693,
      "step": 3373
    },
    {
      "epoch": 1.0653615408904327,
      "grad_norm": 1.2145882844924927,
      "learning_rate": 4.371973896568662e-05,
      "loss": 0.8834,
      "step": 3374
    },
    {
      "epoch": 1.0656772971266182,
      "grad_norm": 1.1207077503204346,
      "learning_rate": 4.371623043996912e-05,
      "loss": 0.8167,
      "step": 3375
    },
    {
      "epoch": 1.065993053362804,
      "grad_norm": 1.2099237442016602,
      "learning_rate": 4.371272191425163e-05,
      "loss": 0.9648,
      "step": 3376
    },
    {
      "epoch": 1.0663088095989897,
      "grad_norm": 1.3103021383285522,
      "learning_rate": 4.370921338853414e-05,
      "loss": 0.8665,
      "step": 3377
    },
    {
      "epoch": 1.0666245658351752,
      "grad_norm": 1.5307999849319458,
      "learning_rate": 4.370570486281665e-05,
      "loss": 0.8786,
      "step": 3378
    },
    {
      "epoch": 1.066940322071361,
      "grad_norm": 1.4044684171676636,
      "learning_rate": 4.3702196337099156e-05,
      "loss": 0.8641,
      "step": 3379
    },
    {
      "epoch": 1.0672560783075467,
      "grad_norm": 1.0372867584228516,
      "learning_rate": 4.369868781138166e-05,
      "loss": 0.8313,
      "step": 3380
    },
    {
      "epoch": 1.0675718345437322,
      "grad_norm": 1.3623121976852417,
      "learning_rate": 4.369517928566417e-05,
      "loss": 0.9026,
      "step": 3381
    },
    {
      "epoch": 1.067887590779918,
      "grad_norm": 1.5271263122558594,
      "learning_rate": 4.3691670759946676e-05,
      "loss": 0.8396,
      "step": 3382
    },
    {
      "epoch": 1.0682033470161036,
      "grad_norm": 1.694191575050354,
      "learning_rate": 4.368816223422918e-05,
      "loss": 0.8164,
      "step": 3383
    },
    {
      "epoch": 1.0685191032522892,
      "grad_norm": 1.337134838104248,
      "learning_rate": 4.368465370851168e-05,
      "loss": 0.9285,
      "step": 3384
    },
    {
      "epoch": 1.068834859488475,
      "grad_norm": 1.4319822788238525,
      "learning_rate": 4.368114518279419e-05,
      "loss": 0.8458,
      "step": 3385
    },
    {
      "epoch": 1.0691506157246606,
      "grad_norm": 1.4134976863861084,
      "learning_rate": 4.36776366570767e-05,
      "loss": 0.8727,
      "step": 3386
    },
    {
      "epoch": 1.0694663719608462,
      "grad_norm": 2.2090563774108887,
      "learning_rate": 4.3674128131359203e-05,
      "loss": 0.9051,
      "step": 3387
    },
    {
      "epoch": 1.069782128197032,
      "grad_norm": 1.657466173171997,
      "learning_rate": 4.367061960564171e-05,
      "loss": 0.9665,
      "step": 3388
    },
    {
      "epoch": 1.0700978844332176,
      "grad_norm": 1.1578459739685059,
      "learning_rate": 4.366711107992422e-05,
      "loss": 0.768,
      "step": 3389
    },
    {
      "epoch": 1.0704136406694031,
      "grad_norm": 1.111130714416504,
      "learning_rate": 4.3663602554206724e-05,
      "loss": 0.8312,
      "step": 3390
    },
    {
      "epoch": 1.0707293969055889,
      "grad_norm": 1.3523480892181396,
      "learning_rate": 4.366009402848923e-05,
      "loss": 0.8781,
      "step": 3391
    },
    {
      "epoch": 1.0710451531417746,
      "grad_norm": 1.572722315788269,
      "learning_rate": 4.365658550277174e-05,
      "loss": 0.8716,
      "step": 3392
    },
    {
      "epoch": 1.0713609093779601,
      "grad_norm": 1.2762960195541382,
      "learning_rate": 4.3653076977054244e-05,
      "loss": 0.8508,
      "step": 3393
    },
    {
      "epoch": 1.0716766656141459,
      "grad_norm": 1.4170249700546265,
      "learning_rate": 4.364956845133675e-05,
      "loss": 0.8537,
      "step": 3394
    },
    {
      "epoch": 1.0719924218503316,
      "grad_norm": 1.551807165145874,
      "learning_rate": 4.364605992561926e-05,
      "loss": 0.9381,
      "step": 3395
    },
    {
      "epoch": 1.0723081780865171,
      "grad_norm": 1.481311559677124,
      "learning_rate": 4.3642551399901765e-05,
      "loss": 0.8507,
      "step": 3396
    },
    {
      "epoch": 1.0726239343227029,
      "grad_norm": 1.9883519411087036,
      "learning_rate": 4.363904287418427e-05,
      "loss": 0.8979,
      "step": 3397
    },
    {
      "epoch": 1.0729396905588886,
      "grad_norm": 1.3747484683990479,
      "learning_rate": 4.363553434846678e-05,
      "loss": 0.8208,
      "step": 3398
    },
    {
      "epoch": 1.0732554467950741,
      "grad_norm": 1.72152578830719,
      "learning_rate": 4.3632025822749285e-05,
      "loss": 0.9232,
      "step": 3399
    },
    {
      "epoch": 1.0735712030312599,
      "grad_norm": 1.5364238023757935,
      "learning_rate": 4.362851729703179e-05,
      "loss": 0.8719,
      "step": 3400
    },
    {
      "epoch": 1.0735712030312599,
      "eval_loss": 0.833091139793396,
      "eval_runtime": 4.6616,
      "eval_samples_per_second": 534.155,
      "eval_steps_per_second": 4.505,
      "step": 3400
    },
    {
      "epoch": 1.0738869592674456,
      "grad_norm": 1.192223072052002,
      "learning_rate": 4.36250087713143e-05,
      "loss": 0.8349,
      "step": 3401
    },
    {
      "epoch": 1.0742027155036311,
      "grad_norm": 1.4514942169189453,
      "learning_rate": 4.3621500245596806e-05,
      "loss": 0.8193,
      "step": 3402
    },
    {
      "epoch": 1.0745184717398168,
      "grad_norm": 1.4760054349899292,
      "learning_rate": 4.3617991719879306e-05,
      "loss": 0.9158,
      "step": 3403
    },
    {
      "epoch": 1.0748342279760026,
      "grad_norm": 1.3327314853668213,
      "learning_rate": 4.361448319416181e-05,
      "loss": 0.7927,
      "step": 3404
    },
    {
      "epoch": 1.075149984212188,
      "grad_norm": 1.8883261680603027,
      "learning_rate": 4.361097466844432e-05,
      "loss": 0.8545,
      "step": 3405
    },
    {
      "epoch": 1.0754657404483738,
      "grad_norm": 1.4154291152954102,
      "learning_rate": 4.3607466142726826e-05,
      "loss": 0.931,
      "step": 3406
    },
    {
      "epoch": 1.0757814966845596,
      "grad_norm": 1.4894837141036987,
      "learning_rate": 4.360395761700933e-05,
      "loss": 0.9088,
      "step": 3407
    },
    {
      "epoch": 1.076097252920745,
      "grad_norm": 1.3674039840698242,
      "learning_rate": 4.360044909129184e-05,
      "loss": 0.7968,
      "step": 3408
    },
    {
      "epoch": 1.0764130091569308,
      "grad_norm": 1.1750346422195435,
      "learning_rate": 4.3596940565574347e-05,
      "loss": 0.8817,
      "step": 3409
    },
    {
      "epoch": 1.0767287653931166,
      "grad_norm": 1.4799145460128784,
      "learning_rate": 4.359343203985685e-05,
      "loss": 0.8446,
      "step": 3410
    },
    {
      "epoch": 1.077044521629302,
      "grad_norm": 1.2764378786087036,
      "learning_rate": 4.358992351413936e-05,
      "loss": 0.897,
      "step": 3411
    },
    {
      "epoch": 1.0773602778654878,
      "grad_norm": 1.1890863180160522,
      "learning_rate": 4.358641498842187e-05,
      "loss": 0.8101,
      "step": 3412
    },
    {
      "epoch": 1.0776760341016736,
      "grad_norm": 1.4572267532348633,
      "learning_rate": 4.3582906462704374e-05,
      "loss": 0.8773,
      "step": 3413
    },
    {
      "epoch": 1.077991790337859,
      "grad_norm": 1.3050251007080078,
      "learning_rate": 4.357939793698688e-05,
      "loss": 0.8742,
      "step": 3414
    },
    {
      "epoch": 1.0783075465740448,
      "grad_norm": 1.2649405002593994,
      "learning_rate": 4.357588941126939e-05,
      "loss": 0.7933,
      "step": 3415
    },
    {
      "epoch": 1.0786233028102306,
      "grad_norm": 1.4446775913238525,
      "learning_rate": 4.3572380885551894e-05,
      "loss": 0.9147,
      "step": 3416
    },
    {
      "epoch": 1.078939059046416,
      "grad_norm": 1.3319528102874756,
      "learning_rate": 4.35688723598344e-05,
      "loss": 0.8426,
      "step": 3417
    },
    {
      "epoch": 1.0792548152826018,
      "grad_norm": 1.3228209018707275,
      "learning_rate": 4.356536383411691e-05,
      "loss": 0.882,
      "step": 3418
    },
    {
      "epoch": 1.0795705715187875,
      "grad_norm": 1.306976556777954,
      "learning_rate": 4.3561855308399415e-05,
      "loss": 0.8798,
      "step": 3419
    },
    {
      "epoch": 1.079886327754973,
      "grad_norm": 1.4541617631912231,
      "learning_rate": 4.355834678268192e-05,
      "loss": 0.8567,
      "step": 3420
    },
    {
      "epoch": 1.0802020839911588,
      "grad_norm": 1.213168740272522,
      "learning_rate": 4.355483825696443e-05,
      "loss": 0.8594,
      "step": 3421
    },
    {
      "epoch": 1.0805178402273445,
      "grad_norm": 1.4533123970031738,
      "learning_rate": 4.355132973124693e-05,
      "loss": 0.8314,
      "step": 3422
    },
    {
      "epoch": 1.08083359646353,
      "grad_norm": 1.3639092445373535,
      "learning_rate": 4.3547821205529435e-05,
      "loss": 0.8094,
      "step": 3423
    },
    {
      "epoch": 1.0811493526997158,
      "grad_norm": 1.2042206525802612,
      "learning_rate": 4.354431267981194e-05,
      "loss": 0.8106,
      "step": 3424
    },
    {
      "epoch": 1.0814651089359015,
      "grad_norm": 1.3422497510910034,
      "learning_rate": 4.354080415409445e-05,
      "loss": 0.8744,
      "step": 3425
    },
    {
      "epoch": 1.081780865172087,
      "grad_norm": 1.1426187753677368,
      "learning_rate": 4.3537295628376956e-05,
      "loss": 0.7895,
      "step": 3426
    },
    {
      "epoch": 1.0820966214082728,
      "grad_norm": 1.280764102935791,
      "learning_rate": 4.353378710265946e-05,
      "loss": 0.894,
      "step": 3427
    },
    {
      "epoch": 1.0824123776444585,
      "grad_norm": 1.2460989952087402,
      "learning_rate": 4.353027857694197e-05,
      "loss": 0.8392,
      "step": 3428
    },
    {
      "epoch": 1.082728133880644,
      "grad_norm": 1.2994935512542725,
      "learning_rate": 4.352677005122448e-05,
      "loss": 0.8445,
      "step": 3429
    },
    {
      "epoch": 1.0830438901168298,
      "grad_norm": 1.3414089679718018,
      "learning_rate": 4.352326152550699e-05,
      "loss": 0.8739,
      "step": 3430
    },
    {
      "epoch": 1.0833596463530155,
      "grad_norm": 1.373284935951233,
      "learning_rate": 4.351975299978949e-05,
      "loss": 0.8003,
      "step": 3431
    },
    {
      "epoch": 1.083675402589201,
      "grad_norm": 1.2849303483963013,
      "learning_rate": 4.3516244474071996e-05,
      "loss": 0.8953,
      "step": 3432
    },
    {
      "epoch": 1.0839911588253868,
      "grad_norm": 1.1858386993408203,
      "learning_rate": 4.35127359483545e-05,
      "loss": 0.9254,
      "step": 3433
    },
    {
      "epoch": 1.0843069150615725,
      "grad_norm": 1.5892621278762817,
      "learning_rate": 4.350922742263701e-05,
      "loss": 0.8722,
      "step": 3434
    },
    {
      "epoch": 1.0846226712977582,
      "grad_norm": 1.158524990081787,
      "learning_rate": 4.350571889691952e-05,
      "loss": 0.8331,
      "step": 3435
    },
    {
      "epoch": 1.0849384275339438,
      "grad_norm": 1.449681282043457,
      "learning_rate": 4.3502210371202024e-05,
      "loss": 0.8816,
      "step": 3436
    },
    {
      "epoch": 1.0852541837701295,
      "grad_norm": 1.336367130279541,
      "learning_rate": 4.349870184548453e-05,
      "loss": 0.876,
      "step": 3437
    },
    {
      "epoch": 1.085569940006315,
      "grad_norm": 1.476487398147583,
      "learning_rate": 4.349519331976704e-05,
      "loss": 0.895,
      "step": 3438
    },
    {
      "epoch": 1.0858856962425008,
      "grad_norm": 1.1578559875488281,
      "learning_rate": 4.3491684794049544e-05,
      "loss": 0.849,
      "step": 3439
    },
    {
      "epoch": 1.0862014524786865,
      "grad_norm": 1.200305700302124,
      "learning_rate": 4.348817626833205e-05,
      "loss": 0.8251,
      "step": 3440
    },
    {
      "epoch": 1.0865172087148722,
      "grad_norm": 1.0209311246871948,
      "learning_rate": 4.348466774261455e-05,
      "loss": 0.782,
      "step": 3441
    },
    {
      "epoch": 1.0868329649510577,
      "grad_norm": 1.54669988155365,
      "learning_rate": 4.348115921689706e-05,
      "loss": 0.8399,
      "step": 3442
    },
    {
      "epoch": 1.0871487211872435,
      "grad_norm": 1.2414028644561768,
      "learning_rate": 4.3477650691179565e-05,
      "loss": 0.8939,
      "step": 3443
    },
    {
      "epoch": 1.087464477423429,
      "grad_norm": 1.451322317123413,
      "learning_rate": 4.347414216546207e-05,
      "loss": 0.8559,
      "step": 3444
    },
    {
      "epoch": 1.0877802336596147,
      "grad_norm": 1.3343098163604736,
      "learning_rate": 4.347063363974458e-05,
      "loss": 0.884,
      "step": 3445
    },
    {
      "epoch": 1.0880959898958005,
      "grad_norm": 1.2749762535095215,
      "learning_rate": 4.3467125114027085e-05,
      "loss": 0.7851,
      "step": 3446
    },
    {
      "epoch": 1.0884117461319862,
      "grad_norm": 1.282204031944275,
      "learning_rate": 4.34636165883096e-05,
      "loss": 0.8265,
      "step": 3447
    },
    {
      "epoch": 1.0887275023681717,
      "grad_norm": 1.2104681730270386,
      "learning_rate": 4.3460108062592105e-05,
      "loss": 0.8771,
      "step": 3448
    },
    {
      "epoch": 1.0890432586043575,
      "grad_norm": 1.5833523273468018,
      "learning_rate": 4.345659953687461e-05,
      "loss": 0.821,
      "step": 3449
    },
    {
      "epoch": 1.089359014840543,
      "grad_norm": 1.362027645111084,
      "learning_rate": 4.345309101115711e-05,
      "loss": 0.8673,
      "step": 3450
    },
    {
      "epoch": 1.0896747710767287,
      "grad_norm": 1.5032286643981934,
      "learning_rate": 4.344958248543962e-05,
      "loss": 0.8425,
      "step": 3451
    },
    {
      "epoch": 1.0899905273129145,
      "grad_norm": 1.334510087966919,
      "learning_rate": 4.3446073959722126e-05,
      "loss": 0.7747,
      "step": 3452
    },
    {
      "epoch": 1.0903062835491002,
      "grad_norm": 1.4706002473831177,
      "learning_rate": 4.344256543400463e-05,
      "loss": 0.8844,
      "step": 3453
    },
    {
      "epoch": 1.0906220397852857,
      "grad_norm": 1.6068137884140015,
      "learning_rate": 4.343905690828714e-05,
      "loss": 0.9133,
      "step": 3454
    },
    {
      "epoch": 1.0909377960214715,
      "grad_norm": 1.1799057722091675,
      "learning_rate": 4.3435548382569646e-05,
      "loss": 0.8204,
      "step": 3455
    },
    {
      "epoch": 1.0912535522576572,
      "grad_norm": 1.261662483215332,
      "learning_rate": 4.343203985685215e-05,
      "loss": 0.7826,
      "step": 3456
    },
    {
      "epoch": 1.0915693084938427,
      "grad_norm": 1.394728660583496,
      "learning_rate": 4.342853133113466e-05,
      "loss": 0.8526,
      "step": 3457
    },
    {
      "epoch": 1.0918850647300284,
      "grad_norm": 1.522416114807129,
      "learning_rate": 4.342502280541717e-05,
      "loss": 0.8194,
      "step": 3458
    },
    {
      "epoch": 1.0922008209662142,
      "grad_norm": 1.3711187839508057,
      "learning_rate": 4.3421514279699674e-05,
      "loss": 0.8733,
      "step": 3459
    },
    {
      "epoch": 1.0925165772023997,
      "grad_norm": 1.2825745344161987,
      "learning_rate": 4.341800575398218e-05,
      "loss": 0.8552,
      "step": 3460
    },
    {
      "epoch": 1.0928323334385854,
      "grad_norm": 1.3332918882369995,
      "learning_rate": 4.341449722826468e-05,
      "loss": 0.9185,
      "step": 3461
    },
    {
      "epoch": 1.0931480896747712,
      "grad_norm": 1.3749313354492188,
      "learning_rate": 4.341098870254719e-05,
      "loss": 0.8187,
      "step": 3462
    },
    {
      "epoch": 1.0934638459109567,
      "grad_norm": 1.356250524520874,
      "learning_rate": 4.3407480176829694e-05,
      "loss": 0.806,
      "step": 3463
    },
    {
      "epoch": 1.0937796021471424,
      "grad_norm": 1.162031650543213,
      "learning_rate": 4.340397165111221e-05,
      "loss": 0.8714,
      "step": 3464
    },
    {
      "epoch": 1.0940953583833282,
      "grad_norm": 1.2985845804214478,
      "learning_rate": 4.3400463125394715e-05,
      "loss": 0.9021,
      "step": 3465
    },
    {
      "epoch": 1.0944111146195137,
      "grad_norm": 1.7060338258743286,
      "learning_rate": 4.339695459967722e-05,
      "loss": 0.7833,
      "step": 3466
    },
    {
      "epoch": 1.0947268708556994,
      "grad_norm": 1.548606038093567,
      "learning_rate": 4.339344607395973e-05,
      "loss": 0.9256,
      "step": 3467
    },
    {
      "epoch": 1.0950426270918852,
      "grad_norm": 1.5420989990234375,
      "learning_rate": 4.3389937548242235e-05,
      "loss": 0.8517,
      "step": 3468
    },
    {
      "epoch": 1.0953583833280707,
      "grad_norm": 1.394096851348877,
      "learning_rate": 4.3386429022524735e-05,
      "loss": 0.872,
      "step": 3469
    },
    {
      "epoch": 1.0956741395642564,
      "grad_norm": 2.043003559112549,
      "learning_rate": 4.338292049680724e-05,
      "loss": 0.8941,
      "step": 3470
    },
    {
      "epoch": 1.0959898958004421,
      "grad_norm": 1.4689583778381348,
      "learning_rate": 4.337941197108975e-05,
      "loss": 0.9204,
      "step": 3471
    },
    {
      "epoch": 1.0963056520366277,
      "grad_norm": 1.3515405654907227,
      "learning_rate": 4.3375903445372255e-05,
      "loss": 0.7905,
      "step": 3472
    },
    {
      "epoch": 1.0966214082728134,
      "grad_norm": 1.4842826128005981,
      "learning_rate": 4.337239491965476e-05,
      "loss": 0.9153,
      "step": 3473
    },
    {
      "epoch": 1.0969371645089991,
      "grad_norm": 1.3981314897537231,
      "learning_rate": 4.336888639393727e-05,
      "loss": 0.8198,
      "step": 3474
    },
    {
      "epoch": 1.0972529207451847,
      "grad_norm": 1.4868078231811523,
      "learning_rate": 4.3365377868219776e-05,
      "loss": 0.8958,
      "step": 3475
    },
    {
      "epoch": 1.0975686769813704,
      "grad_norm": 1.6116594076156616,
      "learning_rate": 4.336186934250228e-05,
      "loss": 0.9045,
      "step": 3476
    },
    {
      "epoch": 1.0978844332175561,
      "grad_norm": 1.2400099039077759,
      "learning_rate": 4.335836081678479e-05,
      "loss": 0.8024,
      "step": 3477
    },
    {
      "epoch": 1.0982001894537416,
      "grad_norm": 1.3250614404678345,
      "learning_rate": 4.3354852291067296e-05,
      "loss": 0.8709,
      "step": 3478
    },
    {
      "epoch": 1.0985159456899274,
      "grad_norm": 1.3731950521469116,
      "learning_rate": 4.33513437653498e-05,
      "loss": 0.8501,
      "step": 3479
    },
    {
      "epoch": 1.0988317019261131,
      "grad_norm": 1.242063283920288,
      "learning_rate": 4.33478352396323e-05,
      "loss": 0.8179,
      "step": 3480
    },
    {
      "epoch": 1.0991474581622986,
      "grad_norm": 1.1863209009170532,
      "learning_rate": 4.334432671391481e-05,
      "loss": 0.861,
      "step": 3481
    },
    {
      "epoch": 1.0994632143984844,
      "grad_norm": 1.2087740898132324,
      "learning_rate": 4.3340818188197324e-05,
      "loss": 0.8695,
      "step": 3482
    },
    {
      "epoch": 1.0997789706346701,
      "grad_norm": 1.460970401763916,
      "learning_rate": 4.333730966247983e-05,
      "loss": 0.8529,
      "step": 3483
    },
    {
      "epoch": 1.1000947268708556,
      "grad_norm": 1.2758758068084717,
      "learning_rate": 4.333380113676234e-05,
      "loss": 0.8861,
      "step": 3484
    },
    {
      "epoch": 1.1004104831070414,
      "grad_norm": 1.3460737466812134,
      "learning_rate": 4.3330292611044844e-05,
      "loss": 0.9115,
      "step": 3485
    },
    {
      "epoch": 1.100726239343227,
      "grad_norm": 1.222632646560669,
      "learning_rate": 4.332678408532735e-05,
      "loss": 0.8037,
      "step": 3486
    },
    {
      "epoch": 1.1010419955794126,
      "grad_norm": 1.4024113416671753,
      "learning_rate": 4.332327555960986e-05,
      "loss": 0.8783,
      "step": 3487
    },
    {
      "epoch": 1.1013577518155984,
      "grad_norm": 1.2598655223846436,
      "learning_rate": 4.3319767033892364e-05,
      "loss": 0.8592,
      "step": 3488
    },
    {
      "epoch": 1.101673508051784,
      "grad_norm": 1.3137998580932617,
      "learning_rate": 4.3316258508174865e-05,
      "loss": 0.8804,
      "step": 3489
    },
    {
      "epoch": 1.1019892642879696,
      "grad_norm": 1.5182158946990967,
      "learning_rate": 4.331274998245737e-05,
      "loss": 0.8722,
      "step": 3490
    },
    {
      "epoch": 1.1023050205241554,
      "grad_norm": 1.3660591840744019,
      "learning_rate": 4.330924145673988e-05,
      "loss": 0.8215,
      "step": 3491
    },
    {
      "epoch": 1.102620776760341,
      "grad_norm": 1.21888267993927,
      "learning_rate": 4.3305732931022385e-05,
      "loss": 0.7937,
      "step": 3492
    },
    {
      "epoch": 1.1029365329965266,
      "grad_norm": 1.224900722503662,
      "learning_rate": 4.330222440530489e-05,
      "loss": 0.835,
      "step": 3493
    },
    {
      "epoch": 1.1032522892327123,
      "grad_norm": 1.452567219734192,
      "learning_rate": 4.32987158795874e-05,
      "loss": 0.8257,
      "step": 3494
    },
    {
      "epoch": 1.103568045468898,
      "grad_norm": 1.4291502237319946,
      "learning_rate": 4.3295207353869905e-05,
      "loss": 0.8575,
      "step": 3495
    },
    {
      "epoch": 1.1038838017050836,
      "grad_norm": 1.3047568798065186,
      "learning_rate": 4.329169882815241e-05,
      "loss": 0.8291,
      "step": 3496
    },
    {
      "epoch": 1.1041995579412693,
      "grad_norm": 1.4571539163589478,
      "learning_rate": 4.328819030243492e-05,
      "loss": 0.8279,
      "step": 3497
    },
    {
      "epoch": 1.104515314177455,
      "grad_norm": 1.3924729824066162,
      "learning_rate": 4.3284681776717426e-05,
      "loss": 0.8608,
      "step": 3498
    },
    {
      "epoch": 1.1048310704136406,
      "grad_norm": 1.3535292148590088,
      "learning_rate": 4.328117325099993e-05,
      "loss": 0.8268,
      "step": 3499
    },
    {
      "epoch": 1.1051468266498263,
      "grad_norm": 1.2057039737701416,
      "learning_rate": 4.327766472528244e-05,
      "loss": 0.7979,
      "step": 3500
    },
    {
      "epoch": 1.105462582886012,
      "grad_norm": 1.3945034742355347,
      "learning_rate": 4.3274156199564946e-05,
      "loss": 0.9021,
      "step": 3501
    },
    {
      "epoch": 1.1057783391221976,
      "grad_norm": 1.3474268913269043,
      "learning_rate": 4.327064767384745e-05,
      "loss": 0.8451,
      "step": 3502
    },
    {
      "epoch": 1.1060940953583833,
      "grad_norm": 1.4807653427124023,
      "learning_rate": 4.326713914812996e-05,
      "loss": 0.9006,
      "step": 3503
    },
    {
      "epoch": 1.106409851594569,
      "grad_norm": 1.7543476819992065,
      "learning_rate": 4.326363062241247e-05,
      "loss": 0.9106,
      "step": 3504
    },
    {
      "epoch": 1.1067256078307546,
      "grad_norm": 1.2802422046661377,
      "learning_rate": 4.3260122096694974e-05,
      "loss": 0.8831,
      "step": 3505
    },
    {
      "epoch": 1.1070413640669403,
      "grad_norm": 1.3958791494369507,
      "learning_rate": 4.325661357097748e-05,
      "loss": 0.7669,
      "step": 3506
    },
    {
      "epoch": 1.107357120303126,
      "grad_norm": 1.3496863842010498,
      "learning_rate": 4.325310504525999e-05,
      "loss": 0.8404,
      "step": 3507
    },
    {
      "epoch": 1.1076728765393116,
      "grad_norm": 1.53837251663208,
      "learning_rate": 4.324959651954249e-05,
      "loss": 0.8637,
      "step": 3508
    },
    {
      "epoch": 1.1079886327754973,
      "grad_norm": 1.0757344961166382,
      "learning_rate": 4.3246087993824994e-05,
      "loss": 0.7892,
      "step": 3509
    },
    {
      "epoch": 1.108304389011683,
      "grad_norm": 1.1449978351593018,
      "learning_rate": 4.32425794681075e-05,
      "loss": 0.798,
      "step": 3510
    },
    {
      "epoch": 1.1086201452478686,
      "grad_norm": 1.1055880784988403,
      "learning_rate": 4.323907094239001e-05,
      "loss": 0.8166,
      "step": 3511
    },
    {
      "epoch": 1.1089359014840543,
      "grad_norm": 1.5957673788070679,
      "learning_rate": 4.3235562416672514e-05,
      "loss": 0.8654,
      "step": 3512
    },
    {
      "epoch": 1.10925165772024,
      "grad_norm": 1.6659975051879883,
      "learning_rate": 4.323205389095502e-05,
      "loss": 0.8759,
      "step": 3513
    },
    {
      "epoch": 1.1095674139564256,
      "grad_norm": 1.4316682815551758,
      "learning_rate": 4.322854536523753e-05,
      "loss": 0.8159,
      "step": 3514
    },
    {
      "epoch": 1.1098831701926113,
      "grad_norm": 1.2557705640792847,
      "learning_rate": 4.3225036839520035e-05,
      "loss": 0.8494,
      "step": 3515
    },
    {
      "epoch": 1.110198926428797,
      "grad_norm": 1.2475039958953857,
      "learning_rate": 4.322152831380254e-05,
      "loss": 0.9256,
      "step": 3516
    },
    {
      "epoch": 1.1105146826649825,
      "grad_norm": 1.5248082876205444,
      "learning_rate": 4.321801978808505e-05,
      "loss": 0.8404,
      "step": 3517
    },
    {
      "epoch": 1.1108304389011683,
      "grad_norm": 1.3328133821487427,
      "learning_rate": 4.3214511262367555e-05,
      "loss": 0.7694,
      "step": 3518
    },
    {
      "epoch": 1.111146195137354,
      "grad_norm": 1.2708640098571777,
      "learning_rate": 4.321100273665006e-05,
      "loss": 0.8104,
      "step": 3519
    },
    {
      "epoch": 1.1114619513735395,
      "grad_norm": 1.517042636871338,
      "learning_rate": 4.320749421093257e-05,
      "loss": 0.8245,
      "step": 3520
    },
    {
      "epoch": 1.1117777076097253,
      "grad_norm": 1.372994065284729,
      "learning_rate": 4.3203985685215076e-05,
      "loss": 0.8241,
      "step": 3521
    },
    {
      "epoch": 1.112093463845911,
      "grad_norm": 1.6686723232269287,
      "learning_rate": 4.320047715949758e-05,
      "loss": 0.8534,
      "step": 3522
    },
    {
      "epoch": 1.1124092200820965,
      "grad_norm": 1.3630249500274658,
      "learning_rate": 4.319696863378009e-05,
      "loss": 0.8754,
      "step": 3523
    },
    {
      "epoch": 1.1127249763182823,
      "grad_norm": 1.5104897022247314,
      "learning_rate": 4.3193460108062596e-05,
      "loss": 0.8387,
      "step": 3524
    },
    {
      "epoch": 1.113040732554468,
      "grad_norm": 1.5596264600753784,
      "learning_rate": 4.31899515823451e-05,
      "loss": 0.9532,
      "step": 3525
    },
    {
      "epoch": 1.1133564887906535,
      "grad_norm": 1.9305760860443115,
      "learning_rate": 4.318644305662761e-05,
      "loss": 0.9109,
      "step": 3526
    },
    {
      "epoch": 1.1136722450268393,
      "grad_norm": 1.4699592590332031,
      "learning_rate": 4.318293453091011e-05,
      "loss": 0.8524,
      "step": 3527
    },
    {
      "epoch": 1.113988001263025,
      "grad_norm": 1.3031964302062988,
      "learning_rate": 4.317942600519262e-05,
      "loss": 0.7847,
      "step": 3528
    },
    {
      "epoch": 1.1143037574992105,
      "grad_norm": 2.074492931365967,
      "learning_rate": 4.3175917479475123e-05,
      "loss": 0.8733,
      "step": 3529
    },
    {
      "epoch": 1.1146195137353963,
      "grad_norm": 1.150720477104187,
      "learning_rate": 4.317240895375763e-05,
      "loss": 0.8392,
      "step": 3530
    },
    {
      "epoch": 1.114935269971582,
      "grad_norm": 1.361722469329834,
      "learning_rate": 4.316890042804014e-05,
      "loss": 0.8664,
      "step": 3531
    },
    {
      "epoch": 1.1152510262077675,
      "grad_norm": 1.209101676940918,
      "learning_rate": 4.3165391902322644e-05,
      "loss": 0.8567,
      "step": 3532
    },
    {
      "epoch": 1.1155667824439532,
      "grad_norm": 1.6466710567474365,
      "learning_rate": 4.316188337660515e-05,
      "loss": 0.8838,
      "step": 3533
    },
    {
      "epoch": 1.115882538680139,
      "grad_norm": 1.6506181955337524,
      "learning_rate": 4.3158374850887664e-05,
      "loss": 0.8069,
      "step": 3534
    },
    {
      "epoch": 1.1161982949163245,
      "grad_norm": 1.4016743898391724,
      "learning_rate": 4.315486632517017e-05,
      "loss": 0.7938,
      "step": 3535
    },
    {
      "epoch": 1.1165140511525102,
      "grad_norm": 1.482635259628296,
      "learning_rate": 4.315135779945267e-05,
      "loss": 0.8519,
      "step": 3536
    },
    {
      "epoch": 1.116829807388696,
      "grad_norm": 1.531810998916626,
      "learning_rate": 4.314784927373518e-05,
      "loss": 0.7925,
      "step": 3537
    },
    {
      "epoch": 1.1171455636248815,
      "grad_norm": 1.709409236907959,
      "learning_rate": 4.3144340748017685e-05,
      "loss": 0.9005,
      "step": 3538
    },
    {
      "epoch": 1.1174613198610672,
      "grad_norm": 1.5054655075073242,
      "learning_rate": 4.314083222230019e-05,
      "loss": 0.8222,
      "step": 3539
    },
    {
      "epoch": 1.117777076097253,
      "grad_norm": 1.6588751077651978,
      "learning_rate": 4.31373236965827e-05,
      "loss": 0.8933,
      "step": 3540
    },
    {
      "epoch": 1.1180928323334385,
      "grad_norm": 1.2618117332458496,
      "learning_rate": 4.3133815170865205e-05,
      "loss": 0.8683,
      "step": 3541
    },
    {
      "epoch": 1.1184085885696242,
      "grad_norm": 1.4483941793441772,
      "learning_rate": 4.313030664514771e-05,
      "loss": 0.8267,
      "step": 3542
    },
    {
      "epoch": 1.11872434480581,
      "grad_norm": 1.449885368347168,
      "learning_rate": 4.312679811943022e-05,
      "loss": 0.8338,
      "step": 3543
    },
    {
      "epoch": 1.1190401010419957,
      "grad_norm": 1.4898221492767334,
      "learning_rate": 4.3123289593712726e-05,
      "loss": 0.8151,
      "step": 3544
    },
    {
      "epoch": 1.1193558572781812,
      "grad_norm": 1.4152177572250366,
      "learning_rate": 4.311978106799523e-05,
      "loss": 0.7955,
      "step": 3545
    },
    {
      "epoch": 1.119671613514367,
      "grad_norm": 1.5091129541397095,
      "learning_rate": 4.311627254227773e-05,
      "loss": 0.8162,
      "step": 3546
    },
    {
      "epoch": 1.1199873697505525,
      "grad_norm": 1.3934051990509033,
      "learning_rate": 4.311276401656024e-05,
      "loss": 0.8726,
      "step": 3547
    },
    {
      "epoch": 1.1203031259867382,
      "grad_norm": 1.382577657699585,
      "learning_rate": 4.3109255490842746e-05,
      "loss": 0.8772,
      "step": 3548
    },
    {
      "epoch": 1.120618882222924,
      "grad_norm": 1.8560032844543457,
      "learning_rate": 4.310574696512525e-05,
      "loss": 0.897,
      "step": 3549
    },
    {
      "epoch": 1.1209346384591097,
      "grad_norm": 1.8002102375030518,
      "learning_rate": 4.310223843940776e-05,
      "loss": 0.8587,
      "step": 3550
    },
    {
      "epoch": 1.1212503946952952,
      "grad_norm": 1.4836196899414062,
      "learning_rate": 4.3098729913690267e-05,
      "loss": 0.8657,
      "step": 3551
    },
    {
      "epoch": 1.121566150931481,
      "grad_norm": 1.500075340270996,
      "learning_rate": 4.309522138797278e-05,
      "loss": 0.9281,
      "step": 3552
    },
    {
      "epoch": 1.1218819071676664,
      "grad_norm": 1.6917767524719238,
      "learning_rate": 4.309171286225529e-05,
      "loss": 0.8092,
      "step": 3553
    },
    {
      "epoch": 1.1221976634038522,
      "grad_norm": 1.438799262046814,
      "learning_rate": 4.3088204336537794e-05,
      "loss": 0.8404,
      "step": 3554
    },
    {
      "epoch": 1.122513419640038,
      "grad_norm": 1.346506953239441,
      "learning_rate": 4.3084695810820294e-05,
      "loss": 0.8735,
      "step": 3555
    },
    {
      "epoch": 1.1228291758762237,
      "grad_norm": 1.6324658393859863,
      "learning_rate": 4.30811872851028e-05,
      "loss": 0.8285,
      "step": 3556
    },
    {
      "epoch": 1.1231449321124092,
      "grad_norm": 1.694936990737915,
      "learning_rate": 4.307767875938531e-05,
      "loss": 0.9266,
      "step": 3557
    },
    {
      "epoch": 1.123460688348595,
      "grad_norm": 1.5349112749099731,
      "learning_rate": 4.3074170233667814e-05,
      "loss": 0.8506,
      "step": 3558
    },
    {
      "epoch": 1.1237764445847804,
      "grad_norm": 1.5941627025604248,
      "learning_rate": 4.307066170795032e-05,
      "loss": 0.8602,
      "step": 3559
    },
    {
      "epoch": 1.1240922008209662,
      "grad_norm": 1.502617597579956,
      "learning_rate": 4.306715318223283e-05,
      "loss": 0.8558,
      "step": 3560
    },
    {
      "epoch": 1.124407957057152,
      "grad_norm": 1.8246062994003296,
      "learning_rate": 4.3063644656515335e-05,
      "loss": 0.8973,
      "step": 3561
    },
    {
      "epoch": 1.1247237132933376,
      "grad_norm": 1.7499685287475586,
      "learning_rate": 4.306013613079784e-05,
      "loss": 0.8512,
      "step": 3562
    },
    {
      "epoch": 1.1250394695295232,
      "grad_norm": 1.3705840110778809,
      "learning_rate": 4.305662760508035e-05,
      "loss": 0.8095,
      "step": 3563
    },
    {
      "epoch": 1.125355225765709,
      "grad_norm": 1.5011762380599976,
      "learning_rate": 4.3053119079362855e-05,
      "loss": 0.8414,
      "step": 3564
    },
    {
      "epoch": 1.1256709820018944,
      "grad_norm": 1.3849495649337769,
      "learning_rate": 4.304961055364536e-05,
      "loss": 0.8874,
      "step": 3565
    },
    {
      "epoch": 1.1259867382380802,
      "grad_norm": 1.3658725023269653,
      "learning_rate": 4.304610202792786e-05,
      "loss": 0.8542,
      "step": 3566
    },
    {
      "epoch": 1.126302494474266,
      "grad_norm": 1.3573750257492065,
      "learning_rate": 4.304259350221037e-05,
      "loss": 0.7843,
      "step": 3567
    },
    {
      "epoch": 1.1266182507104516,
      "grad_norm": 1.4323481321334839,
      "learning_rate": 4.3039084976492876e-05,
      "loss": 0.8588,
      "step": 3568
    },
    {
      "epoch": 1.1269340069466371,
      "grad_norm": 1.3811366558074951,
      "learning_rate": 4.303557645077539e-05,
      "loss": 0.8577,
      "step": 3569
    },
    {
      "epoch": 1.1272497631828229,
      "grad_norm": 1.40803861618042,
      "learning_rate": 4.3032067925057896e-05,
      "loss": 0.8612,
      "step": 3570
    },
    {
      "epoch": 1.1275655194190086,
      "grad_norm": 1.5966479778289795,
      "learning_rate": 4.30285593993404e-05,
      "loss": 0.8931,
      "step": 3571
    },
    {
      "epoch": 1.1278812756551941,
      "grad_norm": 1.6083179712295532,
      "learning_rate": 4.302505087362291e-05,
      "loss": 0.9285,
      "step": 3572
    },
    {
      "epoch": 1.1281970318913799,
      "grad_norm": 1.483344316482544,
      "learning_rate": 4.3021542347905416e-05,
      "loss": 0.8598,
      "step": 3573
    },
    {
      "epoch": 1.1285127881275656,
      "grad_norm": 1.7732867002487183,
      "learning_rate": 4.3018033822187916e-05,
      "loss": 0.8512,
      "step": 3574
    },
    {
      "epoch": 1.1288285443637511,
      "grad_norm": 2.4425394535064697,
      "learning_rate": 4.301452529647042e-05,
      "loss": 0.8647,
      "step": 3575
    },
    {
      "epoch": 1.1291443005999369,
      "grad_norm": 1.4102956056594849,
      "learning_rate": 4.301101677075293e-05,
      "loss": 0.8274,
      "step": 3576
    },
    {
      "epoch": 1.1294600568361226,
      "grad_norm": 1.4167829751968384,
      "learning_rate": 4.300750824503544e-05,
      "loss": 0.8793,
      "step": 3577
    },
    {
      "epoch": 1.1297758130723081,
      "grad_norm": 1.6204673051834106,
      "learning_rate": 4.3003999719317944e-05,
      "loss": 0.8945,
      "step": 3578
    },
    {
      "epoch": 1.1300915693084939,
      "grad_norm": 1.47287118434906,
      "learning_rate": 4.300049119360045e-05,
      "loss": 0.894,
      "step": 3579
    },
    {
      "epoch": 1.1304073255446796,
      "grad_norm": 1.2969515323638916,
      "learning_rate": 4.299698266788296e-05,
      "loss": 0.9139,
      "step": 3580
    },
    {
      "epoch": 1.1307230817808651,
      "grad_norm": 1.423932433128357,
      "learning_rate": 4.2993474142165464e-05,
      "loss": 0.8481,
      "step": 3581
    },
    {
      "epoch": 1.1310388380170509,
      "grad_norm": 1.396943211555481,
      "learning_rate": 4.298996561644797e-05,
      "loss": 0.8115,
      "step": 3582
    },
    {
      "epoch": 1.1313545942532366,
      "grad_norm": 1.4635008573532104,
      "learning_rate": 4.298645709073048e-05,
      "loss": 0.868,
      "step": 3583
    },
    {
      "epoch": 1.131670350489422,
      "grad_norm": 1.4096519947052002,
      "learning_rate": 4.2982948565012985e-05,
      "loss": 0.8333,
      "step": 3584
    },
    {
      "epoch": 1.1319861067256078,
      "grad_norm": 1.4122450351715088,
      "learning_rate": 4.2979440039295485e-05,
      "loss": 0.778,
      "step": 3585
    },
    {
      "epoch": 1.1323018629617936,
      "grad_norm": 1.509149193763733,
      "learning_rate": 4.297593151357799e-05,
      "loss": 0.9397,
      "step": 3586
    },
    {
      "epoch": 1.132617619197979,
      "grad_norm": 1.6674365997314453,
      "learning_rate": 4.2972422987860505e-05,
      "loss": 0.864,
      "step": 3587
    },
    {
      "epoch": 1.1329333754341648,
      "grad_norm": 1.473973274230957,
      "learning_rate": 4.296891446214301e-05,
      "loss": 0.7752,
      "step": 3588
    },
    {
      "epoch": 1.1332491316703506,
      "grad_norm": 1.4615284204483032,
      "learning_rate": 4.296540593642552e-05,
      "loss": 0.7991,
      "step": 3589
    },
    {
      "epoch": 1.133564887906536,
      "grad_norm": 1.2806404829025269,
      "learning_rate": 4.2961897410708026e-05,
      "loss": 0.8536,
      "step": 3590
    },
    {
      "epoch": 1.1338806441427218,
      "grad_norm": 1.451107144355774,
      "learning_rate": 4.295838888499053e-05,
      "loss": 0.9157,
      "step": 3591
    },
    {
      "epoch": 1.1341964003789076,
      "grad_norm": 1.4020249843597412,
      "learning_rate": 4.295488035927304e-05,
      "loss": 0.8335,
      "step": 3592
    },
    {
      "epoch": 1.134512156615093,
      "grad_norm": 1.5349435806274414,
      "learning_rate": 4.2951371833555546e-05,
      "loss": 0.8431,
      "step": 3593
    },
    {
      "epoch": 1.1348279128512788,
      "grad_norm": 1.3788851499557495,
      "learning_rate": 4.2947863307838046e-05,
      "loss": 0.8441,
      "step": 3594
    },
    {
      "epoch": 1.1351436690874646,
      "grad_norm": 1.4129714965820312,
      "learning_rate": 4.294435478212055e-05,
      "loss": 0.8999,
      "step": 3595
    },
    {
      "epoch": 1.13545942532365,
      "grad_norm": 1.6429580450057983,
      "learning_rate": 4.294084625640306e-05,
      "loss": 0.9309,
      "step": 3596
    },
    {
      "epoch": 1.1357751815598358,
      "grad_norm": 1.3543789386749268,
      "learning_rate": 4.2937337730685566e-05,
      "loss": 0.8387,
      "step": 3597
    },
    {
      "epoch": 1.1360909377960215,
      "grad_norm": 1.329883098602295,
      "learning_rate": 4.293382920496807e-05,
      "loss": 0.8382,
      "step": 3598
    },
    {
      "epoch": 1.136406694032207,
      "grad_norm": 1.4480037689208984,
      "learning_rate": 4.293032067925058e-05,
      "loss": 0.8341,
      "step": 3599
    },
    {
      "epoch": 1.1367224502683928,
      "grad_norm": 1.5628005266189575,
      "learning_rate": 4.292681215353309e-05,
      "loss": 0.7934,
      "step": 3600
    },
    {
      "epoch": 1.1367224502683928,
      "eval_loss": 0.82826167345047,
      "eval_runtime": 4.6618,
      "eval_samples_per_second": 534.134,
      "eval_steps_per_second": 4.505,
      "step": 3600
    },
    {
      "epoch": 1.1370382065045785,
      "grad_norm": 1.3981295824050903,
      "learning_rate": 4.2923303627815594e-05,
      "loss": 0.808,
      "step": 3601
    },
    {
      "epoch": 1.137353962740764,
      "grad_norm": 1.2758870124816895,
      "learning_rate": 4.29197951020981e-05,
      "loss": 0.8685,
      "step": 3602
    },
    {
      "epoch": 1.1376697189769498,
      "grad_norm": 1.3419984579086304,
      "learning_rate": 4.291628657638061e-05,
      "loss": 0.846,
      "step": 3603
    },
    {
      "epoch": 1.1379854752131355,
      "grad_norm": 1.5731934309005737,
      "learning_rate": 4.2912778050663114e-05,
      "loss": 0.8464,
      "step": 3604
    },
    {
      "epoch": 1.138301231449321,
      "grad_norm": 1.5997461080551147,
      "learning_rate": 4.290926952494562e-05,
      "loss": 0.8665,
      "step": 3605
    },
    {
      "epoch": 1.1386169876855068,
      "grad_norm": 1.3569799661636353,
      "learning_rate": 4.290576099922813e-05,
      "loss": 0.886,
      "step": 3606
    },
    {
      "epoch": 1.1389327439216925,
      "grad_norm": 1.3471125364303589,
      "learning_rate": 4.2902252473510635e-05,
      "loss": 0.8024,
      "step": 3607
    },
    {
      "epoch": 1.139248500157878,
      "grad_norm": 1.528493046760559,
      "learning_rate": 4.289874394779314e-05,
      "loss": 0.7981,
      "step": 3608
    },
    {
      "epoch": 1.1395642563940638,
      "grad_norm": 1.5385621786117554,
      "learning_rate": 4.289523542207565e-05,
      "loss": 0.8847,
      "step": 3609
    },
    {
      "epoch": 1.1398800126302495,
      "grad_norm": 1.4862748384475708,
      "learning_rate": 4.2891726896358155e-05,
      "loss": 0.8309,
      "step": 3610
    },
    {
      "epoch": 1.140195768866435,
      "grad_norm": 1.5725194215774536,
      "learning_rate": 4.288821837064066e-05,
      "loss": 0.8354,
      "step": 3611
    },
    {
      "epoch": 1.1405115251026208,
      "grad_norm": 1.4074792861938477,
      "learning_rate": 4.288470984492317e-05,
      "loss": 0.7969,
      "step": 3612
    },
    {
      "epoch": 1.1408272813388065,
      "grad_norm": 1.4561985731124878,
      "learning_rate": 4.288120131920567e-05,
      "loss": 0.8406,
      "step": 3613
    },
    {
      "epoch": 1.141143037574992,
      "grad_norm": 1.7960768938064575,
      "learning_rate": 4.2877692793488175e-05,
      "loss": 0.8461,
      "step": 3614
    },
    {
      "epoch": 1.1414587938111778,
      "grad_norm": 1.7152395248413086,
      "learning_rate": 4.287418426777068e-05,
      "loss": 0.807,
      "step": 3615
    },
    {
      "epoch": 1.1417745500473635,
      "grad_norm": 1.4921870231628418,
      "learning_rate": 4.287067574205319e-05,
      "loss": 0.8343,
      "step": 3616
    },
    {
      "epoch": 1.142090306283549,
      "grad_norm": 1.675703525543213,
      "learning_rate": 4.2867167216335696e-05,
      "loss": 0.85,
      "step": 3617
    },
    {
      "epoch": 1.1424060625197348,
      "grad_norm": 1.5385804176330566,
      "learning_rate": 4.28636586906182e-05,
      "loss": 0.7884,
      "step": 3618
    },
    {
      "epoch": 1.1427218187559205,
      "grad_norm": 1.7622488737106323,
      "learning_rate": 4.286015016490071e-05,
      "loss": 0.8726,
      "step": 3619
    },
    {
      "epoch": 1.143037574992106,
      "grad_norm": 1.4280339479446411,
      "learning_rate": 4.2856641639183216e-05,
      "loss": 0.8403,
      "step": 3620
    },
    {
      "epoch": 1.1433533312282917,
      "grad_norm": 1.473936676979065,
      "learning_rate": 4.285313311346572e-05,
      "loss": 0.9023,
      "step": 3621
    },
    {
      "epoch": 1.1436690874644775,
      "grad_norm": 1.6198811531066895,
      "learning_rate": 4.284962458774823e-05,
      "loss": 0.8486,
      "step": 3622
    },
    {
      "epoch": 1.143984843700663,
      "grad_norm": 1.5929440259933472,
      "learning_rate": 4.284611606203074e-05,
      "loss": 0.832,
      "step": 3623
    },
    {
      "epoch": 1.1443005999368487,
      "grad_norm": 1.4125012159347534,
      "learning_rate": 4.2842607536313244e-05,
      "loss": 0.8135,
      "step": 3624
    },
    {
      "epoch": 1.1446163561730345,
      "grad_norm": 1.4646763801574707,
      "learning_rate": 4.283909901059575e-05,
      "loss": 0.8392,
      "step": 3625
    },
    {
      "epoch": 1.14493211240922,
      "grad_norm": 1.6534842252731323,
      "learning_rate": 4.283559048487826e-05,
      "loss": 0.8204,
      "step": 3626
    },
    {
      "epoch": 1.1452478686454057,
      "grad_norm": 1.8237513303756714,
      "learning_rate": 4.2832081959160764e-05,
      "loss": 0.8254,
      "step": 3627
    },
    {
      "epoch": 1.1455636248815915,
      "grad_norm": 2.0244812965393066,
      "learning_rate": 4.282857343344327e-05,
      "loss": 0.7824,
      "step": 3628
    },
    {
      "epoch": 1.145879381117777,
      "grad_norm": 1.5637239217758179,
      "learning_rate": 4.282506490772578e-05,
      "loss": 0.9197,
      "step": 3629
    },
    {
      "epoch": 1.1461951373539627,
      "grad_norm": 1.5449140071868896,
      "learning_rate": 4.2821556382008284e-05,
      "loss": 0.9043,
      "step": 3630
    },
    {
      "epoch": 1.1465108935901485,
      "grad_norm": 1.3359190225601196,
      "learning_rate": 4.281804785629079e-05,
      "loss": 0.7866,
      "step": 3631
    },
    {
      "epoch": 1.146826649826334,
      "grad_norm": 1.3544241189956665,
      "learning_rate": 4.281453933057329e-05,
      "loss": 0.8762,
      "step": 3632
    },
    {
      "epoch": 1.1471424060625197,
      "grad_norm": 1.5791523456573486,
      "learning_rate": 4.28110308048558e-05,
      "loss": 0.8723,
      "step": 3633
    },
    {
      "epoch": 1.1474581622987055,
      "grad_norm": 1.3657084703445435,
      "learning_rate": 4.2807522279138305e-05,
      "loss": 0.7415,
      "step": 3634
    },
    {
      "epoch": 1.147773918534891,
      "grad_norm": 1.4277592897415161,
      "learning_rate": 4.280401375342081e-05,
      "loss": 0.8392,
      "step": 3635
    },
    {
      "epoch": 1.1480896747710767,
      "grad_norm": 1.5580801963806152,
      "learning_rate": 4.280050522770332e-05,
      "loss": 0.8039,
      "step": 3636
    },
    {
      "epoch": 1.1484054310072624,
      "grad_norm": 1.4890990257263184,
      "learning_rate": 4.2796996701985825e-05,
      "loss": 0.8676,
      "step": 3637
    },
    {
      "epoch": 1.148721187243448,
      "grad_norm": 1.6304434537887573,
      "learning_rate": 4.279348817626833e-05,
      "loss": 0.8592,
      "step": 3638
    },
    {
      "epoch": 1.1490369434796337,
      "grad_norm": 1.8879693746566772,
      "learning_rate": 4.2789979650550846e-05,
      "loss": 0.9237,
      "step": 3639
    },
    {
      "epoch": 1.1493526997158194,
      "grad_norm": 1.607777714729309,
      "learning_rate": 4.278647112483335e-05,
      "loss": 0.8134,
      "step": 3640
    },
    {
      "epoch": 1.1496684559520052,
      "grad_norm": 1.6893434524536133,
      "learning_rate": 4.278296259911585e-05,
      "loss": 0.853,
      "step": 3641
    },
    {
      "epoch": 1.1499842121881907,
      "grad_norm": 1.648634910583496,
      "learning_rate": 4.277945407339836e-05,
      "loss": 0.8311,
      "step": 3642
    },
    {
      "epoch": 1.1502999684243764,
      "grad_norm": 1.4123278856277466,
      "learning_rate": 4.2775945547680866e-05,
      "loss": 0.8533,
      "step": 3643
    },
    {
      "epoch": 1.150615724660562,
      "grad_norm": 1.6934242248535156,
      "learning_rate": 4.277243702196337e-05,
      "loss": 0.8324,
      "step": 3644
    },
    {
      "epoch": 1.1509314808967477,
      "grad_norm": 1.3733150959014893,
      "learning_rate": 4.276892849624588e-05,
      "loss": 0.8996,
      "step": 3645
    },
    {
      "epoch": 1.1512472371329334,
      "grad_norm": 1.8070018291473389,
      "learning_rate": 4.276541997052839e-05,
      "loss": 0.8002,
      "step": 3646
    },
    {
      "epoch": 1.1515629933691192,
      "grad_norm": 1.573795199394226,
      "learning_rate": 4.2761911444810894e-05,
      "loss": 0.8673,
      "step": 3647
    },
    {
      "epoch": 1.1518787496053047,
      "grad_norm": 1.8072658777236938,
      "learning_rate": 4.27584029190934e-05,
      "loss": 0.7776,
      "step": 3648
    },
    {
      "epoch": 1.1521945058414904,
      "grad_norm": 1.510526418685913,
      "learning_rate": 4.275489439337591e-05,
      "loss": 0.9222,
      "step": 3649
    },
    {
      "epoch": 1.152510262077676,
      "grad_norm": 1.8621726036071777,
      "learning_rate": 4.2751385867658414e-05,
      "loss": 0.8869,
      "step": 3650
    },
    {
      "epoch": 1.1528260183138617,
      "grad_norm": 1.533288836479187,
      "learning_rate": 4.2747877341940914e-05,
      "loss": 0.8159,
      "step": 3651
    },
    {
      "epoch": 1.1531417745500474,
      "grad_norm": 2.3282816410064697,
      "learning_rate": 4.274436881622342e-05,
      "loss": 0.7861,
      "step": 3652
    },
    {
      "epoch": 1.1534575307862331,
      "grad_norm": 1.3775742053985596,
      "learning_rate": 4.274086029050593e-05,
      "loss": 0.9181,
      "step": 3653
    },
    {
      "epoch": 1.1537732870224187,
      "grad_norm": 1.6020113229751587,
      "learning_rate": 4.2737351764788434e-05,
      "loss": 0.8448,
      "step": 3654
    },
    {
      "epoch": 1.1540890432586044,
      "grad_norm": 1.610823392868042,
      "learning_rate": 4.273384323907094e-05,
      "loss": 0.8372,
      "step": 3655
    },
    {
      "epoch": 1.15440479949479,
      "grad_norm": 1.4308178424835205,
      "learning_rate": 4.273033471335345e-05,
      "loss": 0.8358,
      "step": 3656
    },
    {
      "epoch": 1.1547205557309757,
      "grad_norm": 1.5681132078170776,
      "learning_rate": 4.272682618763596e-05,
      "loss": 0.8817,
      "step": 3657
    },
    {
      "epoch": 1.1550363119671614,
      "grad_norm": 1.4211950302124023,
      "learning_rate": 4.272331766191847e-05,
      "loss": 0.8019,
      "step": 3658
    },
    {
      "epoch": 1.1553520682033471,
      "grad_norm": 1.6808040142059326,
      "learning_rate": 4.2719809136200975e-05,
      "loss": 0.7883,
      "step": 3659
    },
    {
      "epoch": 1.1556678244395326,
      "grad_norm": 1.449851393699646,
      "learning_rate": 4.2716300610483475e-05,
      "loss": 0.8036,
      "step": 3660
    },
    {
      "epoch": 1.1559835806757184,
      "grad_norm": 1.3587117195129395,
      "learning_rate": 4.271279208476598e-05,
      "loss": 0.8932,
      "step": 3661
    },
    {
      "epoch": 1.156299336911904,
      "grad_norm": 1.7088351249694824,
      "learning_rate": 4.270928355904849e-05,
      "loss": 0.9102,
      "step": 3662
    },
    {
      "epoch": 1.1566150931480896,
      "grad_norm": 1.3760402202606201,
      "learning_rate": 4.2705775033330996e-05,
      "loss": 0.8781,
      "step": 3663
    },
    {
      "epoch": 1.1569308493842754,
      "grad_norm": 1.5937891006469727,
      "learning_rate": 4.27022665076135e-05,
      "loss": 0.8371,
      "step": 3664
    },
    {
      "epoch": 1.157246605620461,
      "grad_norm": 1.5436561107635498,
      "learning_rate": 4.269875798189601e-05,
      "loss": 0.8031,
      "step": 3665
    },
    {
      "epoch": 1.1575623618566466,
      "grad_norm": 1.4234154224395752,
      "learning_rate": 4.2695249456178516e-05,
      "loss": 0.7794,
      "step": 3666
    },
    {
      "epoch": 1.1578781180928324,
      "grad_norm": 1.4942626953125,
      "learning_rate": 4.269174093046102e-05,
      "loss": 0.844,
      "step": 3667
    },
    {
      "epoch": 1.1581938743290179,
      "grad_norm": 1.311644196510315,
      "learning_rate": 4.268823240474353e-05,
      "loss": 0.8382,
      "step": 3668
    },
    {
      "epoch": 1.1585096305652036,
      "grad_norm": 1.7135343551635742,
      "learning_rate": 4.268472387902604e-05,
      "loss": 0.8449,
      "step": 3669
    },
    {
      "epoch": 1.1588253868013894,
      "grad_norm": 1.5958439111709595,
      "learning_rate": 4.2681215353308543e-05,
      "loss": 0.8829,
      "step": 3670
    },
    {
      "epoch": 1.159141143037575,
      "grad_norm": 1.5884641408920288,
      "learning_rate": 4.2677706827591043e-05,
      "loss": 0.9062,
      "step": 3671
    },
    {
      "epoch": 1.1594568992737606,
      "grad_norm": 1.3574752807617188,
      "learning_rate": 4.267419830187355e-05,
      "loss": 0.8318,
      "step": 3672
    },
    {
      "epoch": 1.1597726555099463,
      "grad_norm": 1.6749953031539917,
      "learning_rate": 4.267068977615606e-05,
      "loss": 0.9046,
      "step": 3673
    },
    {
      "epoch": 1.1600884117461319,
      "grad_norm": 1.413309931755066,
      "learning_rate": 4.266718125043857e-05,
      "loss": 0.851,
      "step": 3674
    },
    {
      "epoch": 1.1604041679823176,
      "grad_norm": 1.2297406196594238,
      "learning_rate": 4.266367272472108e-05,
      "loss": 0.8121,
      "step": 3675
    },
    {
      "epoch": 1.1607199242185033,
      "grad_norm": 1.446869969367981,
      "learning_rate": 4.2660164199003584e-05,
      "loss": 0.8227,
      "step": 3676
    },
    {
      "epoch": 1.161035680454689,
      "grad_norm": 1.4810465574264526,
      "learning_rate": 4.265665567328609e-05,
      "loss": 0.8256,
      "step": 3677
    },
    {
      "epoch": 1.1613514366908746,
      "grad_norm": 1.5193307399749756,
      "learning_rate": 4.26531471475686e-05,
      "loss": 0.8698,
      "step": 3678
    },
    {
      "epoch": 1.1616671929270603,
      "grad_norm": 1.5570194721221924,
      "learning_rate": 4.26496386218511e-05,
      "loss": 0.8776,
      "step": 3679
    },
    {
      "epoch": 1.161982949163246,
      "grad_norm": 1.1957755088806152,
      "learning_rate": 4.2646130096133605e-05,
      "loss": 0.8053,
      "step": 3680
    },
    {
      "epoch": 1.1622987053994316,
      "grad_norm": 1.257508397102356,
      "learning_rate": 4.264262157041611e-05,
      "loss": 0.8038,
      "step": 3681
    },
    {
      "epoch": 1.1626144616356173,
      "grad_norm": 1.318721890449524,
      "learning_rate": 4.263911304469862e-05,
      "loss": 0.7752,
      "step": 3682
    },
    {
      "epoch": 1.162930217871803,
      "grad_norm": 1.670576810836792,
      "learning_rate": 4.2635604518981125e-05,
      "loss": 0.8678,
      "step": 3683
    },
    {
      "epoch": 1.1632459741079886,
      "grad_norm": 1.3444547653198242,
      "learning_rate": 4.263209599326363e-05,
      "loss": 0.846,
      "step": 3684
    },
    {
      "epoch": 1.1635617303441743,
      "grad_norm": 1.3820645809173584,
      "learning_rate": 4.262858746754614e-05,
      "loss": 0.8398,
      "step": 3685
    },
    {
      "epoch": 1.16387748658036,
      "grad_norm": 1.434889316558838,
      "learning_rate": 4.2625078941828646e-05,
      "loss": 0.8717,
      "step": 3686
    },
    {
      "epoch": 1.1641932428165456,
      "grad_norm": 1.4324404001235962,
      "learning_rate": 4.262157041611115e-05,
      "loss": 0.7868,
      "step": 3687
    },
    {
      "epoch": 1.1645089990527313,
      "grad_norm": 1.145082950592041,
      "learning_rate": 4.261806189039366e-05,
      "loss": 0.8334,
      "step": 3688
    },
    {
      "epoch": 1.164824755288917,
      "grad_norm": 1.445884346961975,
      "learning_rate": 4.2614553364676166e-05,
      "loss": 0.7949,
      "step": 3689
    },
    {
      "epoch": 1.1651405115251026,
      "grad_norm": 1.3468658924102783,
      "learning_rate": 4.2611044838958666e-05,
      "loss": 0.7855,
      "step": 3690
    },
    {
      "epoch": 1.1654562677612883,
      "grad_norm": 1.5069907903671265,
      "learning_rate": 4.260753631324117e-05,
      "loss": 0.7697,
      "step": 3691
    },
    {
      "epoch": 1.165772023997474,
      "grad_norm": 2.1671528816223145,
      "learning_rate": 4.2604027787523687e-05,
      "loss": 0.8103,
      "step": 3692
    },
    {
      "epoch": 1.1660877802336596,
      "grad_norm": 1.5896146297454834,
      "learning_rate": 4.260051926180619e-05,
      "loss": 0.9063,
      "step": 3693
    },
    {
      "epoch": 1.1664035364698453,
      "grad_norm": 2.543896436691284,
      "learning_rate": 4.25970107360887e-05,
      "loss": 0.858,
      "step": 3694
    },
    {
      "epoch": 1.166719292706031,
      "grad_norm": 1.9040255546569824,
      "learning_rate": 4.259350221037121e-05,
      "loss": 0.8129,
      "step": 3695
    },
    {
      "epoch": 1.1670350489422165,
      "grad_norm": 1.679237961769104,
      "learning_rate": 4.2589993684653714e-05,
      "loss": 0.8267,
      "step": 3696
    },
    {
      "epoch": 1.1673508051784023,
      "grad_norm": 1.597759485244751,
      "learning_rate": 4.258648515893622e-05,
      "loss": 0.7981,
      "step": 3697
    },
    {
      "epoch": 1.167666561414588,
      "grad_norm": 1.3512784242630005,
      "learning_rate": 4.258297663321873e-05,
      "loss": 0.8378,
      "step": 3698
    },
    {
      "epoch": 1.1679823176507735,
      "grad_norm": 1.380062460899353,
      "learning_rate": 4.257946810750123e-05,
      "loss": 0.8406,
      "step": 3699
    },
    {
      "epoch": 1.1682980738869593,
      "grad_norm": 1.4972820281982422,
      "learning_rate": 4.2575959581783734e-05,
      "loss": 0.8529,
      "step": 3700
    },
    {
      "epoch": 1.168613830123145,
      "grad_norm": 1.5528321266174316,
      "learning_rate": 4.257245105606624e-05,
      "loss": 0.875,
      "step": 3701
    },
    {
      "epoch": 1.1689295863593305,
      "grad_norm": 1.5872782468795776,
      "learning_rate": 4.256894253034875e-05,
      "loss": 0.8261,
      "step": 3702
    },
    {
      "epoch": 1.1692453425955163,
      "grad_norm": 1.3255056142807007,
      "learning_rate": 4.2565434004631255e-05,
      "loss": 0.9101,
      "step": 3703
    },
    {
      "epoch": 1.169561098831702,
      "grad_norm": 1.6173511743545532,
      "learning_rate": 4.256192547891376e-05,
      "loss": 0.9369,
      "step": 3704
    },
    {
      "epoch": 1.1698768550678875,
      "grad_norm": 1.4417248964309692,
      "learning_rate": 4.255841695319627e-05,
      "loss": 0.7713,
      "step": 3705
    },
    {
      "epoch": 1.1701926113040733,
      "grad_norm": 1.2534422874450684,
      "learning_rate": 4.2554908427478775e-05,
      "loss": 0.8329,
      "step": 3706
    },
    {
      "epoch": 1.170508367540259,
      "grad_norm": 1.281675934791565,
      "learning_rate": 4.255139990176128e-05,
      "loss": 0.7613,
      "step": 3707
    },
    {
      "epoch": 1.1708241237764445,
      "grad_norm": 1.2671864032745361,
      "learning_rate": 4.254789137604379e-05,
      "loss": 0.8472,
      "step": 3708
    },
    {
      "epoch": 1.1711398800126303,
      "grad_norm": 1.2809953689575195,
      "learning_rate": 4.2544382850326296e-05,
      "loss": 0.7715,
      "step": 3709
    },
    {
      "epoch": 1.171455636248816,
      "grad_norm": 1.283815860748291,
      "learning_rate": 4.25408743246088e-05,
      "loss": 0.9098,
      "step": 3710
    },
    {
      "epoch": 1.1717713924850015,
      "grad_norm": 1.5737652778625488,
      "learning_rate": 4.253736579889131e-05,
      "loss": 0.8559,
      "step": 3711
    },
    {
      "epoch": 1.1720871487211872,
      "grad_norm": 1.4368433952331543,
      "learning_rate": 4.2533857273173816e-05,
      "loss": 0.8849,
      "step": 3712
    },
    {
      "epoch": 1.172402904957373,
      "grad_norm": 1.572675108909607,
      "learning_rate": 4.253034874745632e-05,
      "loss": 0.8263,
      "step": 3713
    },
    {
      "epoch": 1.1727186611935585,
      "grad_norm": 2.1006226539611816,
      "learning_rate": 4.252684022173883e-05,
      "loss": 0.9664,
      "step": 3714
    },
    {
      "epoch": 1.1730344174297442,
      "grad_norm": 1.402104139328003,
      "learning_rate": 4.2523331696021336e-05,
      "loss": 0.8515,
      "step": 3715
    },
    {
      "epoch": 1.17335017366593,
      "grad_norm": 1.537558674812317,
      "learning_rate": 4.251982317030384e-05,
      "loss": 0.7893,
      "step": 3716
    },
    {
      "epoch": 1.1736659299021155,
      "grad_norm": 1.343255877494812,
      "learning_rate": 4.251631464458635e-05,
      "loss": 0.799,
      "step": 3717
    },
    {
      "epoch": 1.1739816861383012,
      "grad_norm": 1.4543713331222534,
      "learning_rate": 4.251280611886885e-05,
      "loss": 0.8885,
      "step": 3718
    },
    {
      "epoch": 1.174297442374487,
      "grad_norm": 1.5056151151657104,
      "learning_rate": 4.250929759315136e-05,
      "loss": 0.8569,
      "step": 3719
    },
    {
      "epoch": 1.1746131986106725,
      "grad_norm": 1.3976993560791016,
      "learning_rate": 4.2505789067433864e-05,
      "loss": 0.8417,
      "step": 3720
    },
    {
      "epoch": 1.1749289548468582,
      "grad_norm": 1.2969530820846558,
      "learning_rate": 4.250228054171637e-05,
      "loss": 0.7461,
      "step": 3721
    },
    {
      "epoch": 1.175244711083044,
      "grad_norm": 1.3970513343811035,
      "learning_rate": 4.249877201599888e-05,
      "loss": 0.9353,
      "step": 3722
    },
    {
      "epoch": 1.1755604673192295,
      "grad_norm": 1.516650676727295,
      "learning_rate": 4.2495263490281384e-05,
      "loss": 0.8661,
      "step": 3723
    },
    {
      "epoch": 1.1758762235554152,
      "grad_norm": 1.6106929779052734,
      "learning_rate": 4.249175496456389e-05,
      "loss": 0.8483,
      "step": 3724
    },
    {
      "epoch": 1.176191979791601,
      "grad_norm": 1.5135729312896729,
      "learning_rate": 4.24882464388464e-05,
      "loss": 0.8101,
      "step": 3725
    },
    {
      "epoch": 1.1765077360277865,
      "grad_norm": 1.416178584098816,
      "learning_rate": 4.2484737913128905e-05,
      "loss": 0.8432,
      "step": 3726
    },
    {
      "epoch": 1.1768234922639722,
      "grad_norm": 1.4878500699996948,
      "learning_rate": 4.248122938741141e-05,
      "loss": 0.7991,
      "step": 3727
    },
    {
      "epoch": 1.177139248500158,
      "grad_norm": 1.4562584161758423,
      "learning_rate": 4.247772086169392e-05,
      "loss": 0.8824,
      "step": 3728
    },
    {
      "epoch": 1.1774550047363435,
      "grad_norm": 1.3369364738464355,
      "learning_rate": 4.2474212335976425e-05,
      "loss": 0.7953,
      "step": 3729
    },
    {
      "epoch": 1.1777707609725292,
      "grad_norm": 1.5547410249710083,
      "learning_rate": 4.247070381025893e-05,
      "loss": 0.8701,
      "step": 3730
    },
    {
      "epoch": 1.178086517208715,
      "grad_norm": 1.5600937604904175,
      "learning_rate": 4.246719528454144e-05,
      "loss": 0.816,
      "step": 3731
    },
    {
      "epoch": 1.1784022734449004,
      "grad_norm": 1.5440497398376465,
      "learning_rate": 4.2463686758823946e-05,
      "loss": 0.9418,
      "step": 3732
    },
    {
      "epoch": 1.1787180296810862,
      "grad_norm": 1.4081655740737915,
      "learning_rate": 4.246017823310645e-05,
      "loss": 0.8349,
      "step": 3733
    },
    {
      "epoch": 1.179033785917272,
      "grad_norm": 1.6605298519134521,
      "learning_rate": 4.245666970738896e-05,
      "loss": 0.8104,
      "step": 3734
    },
    {
      "epoch": 1.1793495421534574,
      "grad_norm": 1.5328059196472168,
      "learning_rate": 4.2453161181671466e-05,
      "loss": 0.8016,
      "step": 3735
    },
    {
      "epoch": 1.1796652983896432,
      "grad_norm": 1.5660732984542847,
      "learning_rate": 4.244965265595397e-05,
      "loss": 0.8487,
      "step": 3736
    },
    {
      "epoch": 1.179981054625829,
      "grad_norm": 1.6726597547531128,
      "learning_rate": 4.244614413023647e-05,
      "loss": 0.8855,
      "step": 3737
    },
    {
      "epoch": 1.1802968108620144,
      "grad_norm": 1.528181791305542,
      "learning_rate": 4.244263560451898e-05,
      "loss": 0.8633,
      "step": 3738
    },
    {
      "epoch": 1.1806125670982002,
      "grad_norm": 1.6293046474456787,
      "learning_rate": 4.2439127078801486e-05,
      "loss": 0.8806,
      "step": 3739
    },
    {
      "epoch": 1.180928323334386,
      "grad_norm": 1.3301984071731567,
      "learning_rate": 4.243561855308399e-05,
      "loss": 0.8076,
      "step": 3740
    },
    {
      "epoch": 1.1812440795705714,
      "grad_norm": 1.4870843887329102,
      "learning_rate": 4.24321100273665e-05,
      "loss": 0.8056,
      "step": 3741
    },
    {
      "epoch": 1.1815598358067572,
      "grad_norm": 1.5444482564926147,
      "learning_rate": 4.242860150164901e-05,
      "loss": 0.7809,
      "step": 3742
    },
    {
      "epoch": 1.181875592042943,
      "grad_norm": 1.5818939208984375,
      "learning_rate": 4.2425092975931514e-05,
      "loss": 0.7211,
      "step": 3743
    },
    {
      "epoch": 1.1821913482791284,
      "grad_norm": 1.827374815940857,
      "learning_rate": 4.242158445021403e-05,
      "loss": 0.8772,
      "step": 3744
    },
    {
      "epoch": 1.1825071045153142,
      "grad_norm": 1.6781004667282104,
      "learning_rate": 4.2418075924496534e-05,
      "loss": 0.8576,
      "step": 3745
    },
    {
      "epoch": 1.1828228607515,
      "grad_norm": 1.375720739364624,
      "learning_rate": 4.2414567398779034e-05,
      "loss": 0.8595,
      "step": 3746
    },
    {
      "epoch": 1.1831386169876854,
      "grad_norm": 1.5200674533843994,
      "learning_rate": 4.241105887306154e-05,
      "loss": 0.851,
      "step": 3747
    },
    {
      "epoch": 1.1834543732238711,
      "grad_norm": 1.2373559474945068,
      "learning_rate": 4.240755034734405e-05,
      "loss": 0.8466,
      "step": 3748
    },
    {
      "epoch": 1.1837701294600569,
      "grad_norm": 1.6932904720306396,
      "learning_rate": 4.2404041821626555e-05,
      "loss": 0.9279,
      "step": 3749
    },
    {
      "epoch": 1.1840858856962426,
      "grad_norm": 1.6286180019378662,
      "learning_rate": 4.240053329590906e-05,
      "loss": 0.8611,
      "step": 3750
    },
    {
      "epoch": 1.1844016419324281,
      "grad_norm": 1.4975529909133911,
      "learning_rate": 4.239702477019157e-05,
      "loss": 0.8089,
      "step": 3751
    },
    {
      "epoch": 1.1847173981686139,
      "grad_norm": 1.5627734661102295,
      "learning_rate": 4.2393516244474075e-05,
      "loss": 0.7898,
      "step": 3752
    },
    {
      "epoch": 1.1850331544047994,
      "grad_norm": 1.6324865818023682,
      "learning_rate": 4.239000771875658e-05,
      "loss": 0.8376,
      "step": 3753
    },
    {
      "epoch": 1.1853489106409851,
      "grad_norm": 1.2707597017288208,
      "learning_rate": 4.238649919303909e-05,
      "loss": 0.8645,
      "step": 3754
    },
    {
      "epoch": 1.1856646668771709,
      "grad_norm": 1.448202133178711,
      "learning_rate": 4.2382990667321595e-05,
      "loss": 0.8063,
      "step": 3755
    },
    {
      "epoch": 1.1859804231133566,
      "grad_norm": 1.7594614028930664,
      "learning_rate": 4.2379482141604095e-05,
      "loss": 0.8974,
      "step": 3756
    },
    {
      "epoch": 1.1862961793495421,
      "grad_norm": 1.45151686668396,
      "learning_rate": 4.23759736158866e-05,
      "loss": 0.8411,
      "step": 3757
    },
    {
      "epoch": 1.1866119355857279,
      "grad_norm": 1.7614847421646118,
      "learning_rate": 4.237246509016911e-05,
      "loss": 0.8471,
      "step": 3758
    },
    {
      "epoch": 1.1869276918219134,
      "grad_norm": 1.6586607694625854,
      "learning_rate": 4.2368956564451616e-05,
      "loss": 0.8646,
      "step": 3759
    },
    {
      "epoch": 1.1872434480580991,
      "grad_norm": 1.3903013467788696,
      "learning_rate": 4.236544803873412e-05,
      "loss": 0.8796,
      "step": 3760
    },
    {
      "epoch": 1.1875592042942849,
      "grad_norm": 1.747534990310669,
      "learning_rate": 4.2361939513016636e-05,
      "loss": 0.9024,
      "step": 3761
    },
    {
      "epoch": 1.1878749605304706,
      "grad_norm": 1.3030835390090942,
      "learning_rate": 4.235843098729914e-05,
      "loss": 0.7876,
      "step": 3762
    },
    {
      "epoch": 1.188190716766656,
      "grad_norm": 1.2437330484390259,
      "learning_rate": 4.235492246158165e-05,
      "loss": 0.8438,
      "step": 3763
    },
    {
      "epoch": 1.1885064730028418,
      "grad_norm": 1.8603801727294922,
      "learning_rate": 4.235141393586416e-05,
      "loss": 0.8269,
      "step": 3764
    },
    {
      "epoch": 1.1888222292390274,
      "grad_norm": 1.2190220355987549,
      "learning_rate": 4.234790541014666e-05,
      "loss": 0.8549,
      "step": 3765
    },
    {
      "epoch": 1.189137985475213,
      "grad_norm": 1.3303803205490112,
      "learning_rate": 4.2344396884429164e-05,
      "loss": 0.823,
      "step": 3766
    },
    {
      "epoch": 1.1894537417113988,
      "grad_norm": 1.522394061088562,
      "learning_rate": 4.234088835871167e-05,
      "loss": 0.866,
      "step": 3767
    },
    {
      "epoch": 1.1897694979475846,
      "grad_norm": 1.829817771911621,
      "learning_rate": 4.233737983299418e-05,
      "loss": 0.8718,
      "step": 3768
    },
    {
      "epoch": 1.19008525418377,
      "grad_norm": 1.2208274602890015,
      "learning_rate": 4.2333871307276684e-05,
      "loss": 0.8586,
      "step": 3769
    },
    {
      "epoch": 1.1904010104199558,
      "grad_norm": 1.4626319408416748,
      "learning_rate": 4.233036278155919e-05,
      "loss": 0.8516,
      "step": 3770
    },
    {
      "epoch": 1.1907167666561413,
      "grad_norm": 1.4421865940093994,
      "learning_rate": 4.23268542558417e-05,
      "loss": 0.8936,
      "step": 3771
    },
    {
      "epoch": 1.191032522892327,
      "grad_norm": 1.524406909942627,
      "learning_rate": 4.2323345730124204e-05,
      "loss": 0.9041,
      "step": 3772
    },
    {
      "epoch": 1.1913482791285128,
      "grad_norm": 1.5267866849899292,
      "learning_rate": 4.231983720440671e-05,
      "loss": 0.7744,
      "step": 3773
    },
    {
      "epoch": 1.1916640353646986,
      "grad_norm": 1.3156425952911377,
      "learning_rate": 4.231632867868922e-05,
      "loss": 0.8806,
      "step": 3774
    },
    {
      "epoch": 1.191979791600884,
      "grad_norm": 1.3213658332824707,
      "learning_rate": 4.2312820152971725e-05,
      "loss": 0.827,
      "step": 3775
    },
    {
      "epoch": 1.1922955478370698,
      "grad_norm": 1.4986577033996582,
      "learning_rate": 4.2309311627254225e-05,
      "loss": 0.7513,
      "step": 3776
    },
    {
      "epoch": 1.1926113040732553,
      "grad_norm": 1.3473587036132812,
      "learning_rate": 4.230580310153673e-05,
      "loss": 0.8668,
      "step": 3777
    },
    {
      "epoch": 1.192927060309441,
      "grad_norm": 1.3186733722686768,
      "learning_rate": 4.230229457581924e-05,
      "loss": 0.8885,
      "step": 3778
    },
    {
      "epoch": 1.1932428165456268,
      "grad_norm": 1.4066764116287231,
      "learning_rate": 4.229878605010175e-05,
      "loss": 0.8531,
      "step": 3779
    },
    {
      "epoch": 1.1935585727818125,
      "grad_norm": 1.2060154676437378,
      "learning_rate": 4.229527752438426e-05,
      "loss": 0.8718,
      "step": 3780
    },
    {
      "epoch": 1.193874329017998,
      "grad_norm": 1.6499701738357544,
      "learning_rate": 4.2291768998666766e-05,
      "loss": 0.8587,
      "step": 3781
    },
    {
      "epoch": 1.1941900852541838,
      "grad_norm": 1.3461400270462036,
      "learning_rate": 4.228826047294927e-05,
      "loss": 0.8382,
      "step": 3782
    },
    {
      "epoch": 1.1945058414903693,
      "grad_norm": 1.608927845954895,
      "learning_rate": 4.228475194723178e-05,
      "loss": 0.8517,
      "step": 3783
    },
    {
      "epoch": 1.194821597726555,
      "grad_norm": 1.855676293373108,
      "learning_rate": 4.228124342151428e-05,
      "loss": 0.8915,
      "step": 3784
    },
    {
      "epoch": 1.1951373539627408,
      "grad_norm": 1.2431226968765259,
      "learning_rate": 4.2277734895796786e-05,
      "loss": 0.8244,
      "step": 3785
    },
    {
      "epoch": 1.1954531101989265,
      "grad_norm": 1.523222804069519,
      "learning_rate": 4.227422637007929e-05,
      "loss": 0.9424,
      "step": 3786
    },
    {
      "epoch": 1.195768866435112,
      "grad_norm": 1.6323823928833008,
      "learning_rate": 4.22707178443618e-05,
      "loss": 0.8733,
      "step": 3787
    },
    {
      "epoch": 1.1960846226712978,
      "grad_norm": 1.3747648000717163,
      "learning_rate": 4.226720931864431e-05,
      "loss": 0.8405,
      "step": 3788
    },
    {
      "epoch": 1.1964003789074835,
      "grad_norm": 1.562716007232666,
      "learning_rate": 4.2263700792926814e-05,
      "loss": 0.9506,
      "step": 3789
    },
    {
      "epoch": 1.196716135143669,
      "grad_norm": 1.4353854656219482,
      "learning_rate": 4.226019226720932e-05,
      "loss": 0.8001,
      "step": 3790
    },
    {
      "epoch": 1.1970318913798548,
      "grad_norm": 1.5799071788787842,
      "learning_rate": 4.225668374149183e-05,
      "loss": 0.8318,
      "step": 3791
    },
    {
      "epoch": 1.1973476476160405,
      "grad_norm": 1.3843834400177002,
      "learning_rate": 4.2253175215774334e-05,
      "loss": 0.881,
      "step": 3792
    },
    {
      "epoch": 1.197663403852226,
      "grad_norm": 1.535508155822754,
      "learning_rate": 4.224966669005684e-05,
      "loss": 0.8689,
      "step": 3793
    },
    {
      "epoch": 1.1979791600884118,
      "grad_norm": 1.4619882106781006,
      "learning_rate": 4.224615816433935e-05,
      "loss": 0.8592,
      "step": 3794
    },
    {
      "epoch": 1.1982949163245975,
      "grad_norm": 1.3768644332885742,
      "learning_rate": 4.224264963862185e-05,
      "loss": 0.834,
      "step": 3795
    },
    {
      "epoch": 1.198610672560783,
      "grad_norm": 1.3374943733215332,
      "learning_rate": 4.223914111290436e-05,
      "loss": 0.8339,
      "step": 3796
    },
    {
      "epoch": 1.1989264287969688,
      "grad_norm": 1.6215804815292358,
      "learning_rate": 4.223563258718687e-05,
      "loss": 0.8135,
      "step": 3797
    },
    {
      "epoch": 1.1992421850331545,
      "grad_norm": 1.3644299507141113,
      "learning_rate": 4.2232124061469375e-05,
      "loss": 0.8635,
      "step": 3798
    },
    {
      "epoch": 1.19955794126934,
      "grad_norm": 1.4561705589294434,
      "learning_rate": 4.222861553575188e-05,
      "loss": 0.9043,
      "step": 3799
    },
    {
      "epoch": 1.1998736975055257,
      "grad_norm": 1.272659420967102,
      "learning_rate": 4.222510701003439e-05,
      "loss": 0.8063,
      "step": 3800
    },
    {
      "epoch": 1.1998736975055257,
      "eval_loss": 0.8341767191886902,
      "eval_runtime": 4.6503,
      "eval_samples_per_second": 535.453,
      "eval_steps_per_second": 4.516,
      "step": 3800
    },
    {
      "epoch": 1.2001894537417115,
      "grad_norm": 1.325452208518982,
      "learning_rate": 4.2221598484316895e-05,
      "loss": 0.8301,
      "step": 3801
    },
    {
      "epoch": 1.200505209977897,
      "grad_norm": 1.4742480516433716,
      "learning_rate": 4.22180899585994e-05,
      "loss": 0.8331,
      "step": 3802
    },
    {
      "epoch": 1.2008209662140827,
      "grad_norm": 1.4864006042480469,
      "learning_rate": 4.221458143288191e-05,
      "loss": 0.8251,
      "step": 3803
    },
    {
      "epoch": 1.2011367224502685,
      "grad_norm": 1.9110397100448608,
      "learning_rate": 4.221107290716441e-05,
      "loss": 0.9296,
      "step": 3804
    },
    {
      "epoch": 1.201452478686454,
      "grad_norm": 1.618923544883728,
      "learning_rate": 4.2207564381446916e-05,
      "loss": 0.8945,
      "step": 3805
    },
    {
      "epoch": 1.2017682349226397,
      "grad_norm": 1.7346782684326172,
      "learning_rate": 4.220405585572942e-05,
      "loss": 0.8864,
      "step": 3806
    },
    {
      "epoch": 1.2020839911588255,
      "grad_norm": 1.4780226945877075,
      "learning_rate": 4.220054733001193e-05,
      "loss": 0.859,
      "step": 3807
    },
    {
      "epoch": 1.202399747395011,
      "grad_norm": 1.546600341796875,
      "learning_rate": 4.2197038804294436e-05,
      "loss": 0.8715,
      "step": 3808
    },
    {
      "epoch": 1.2027155036311967,
      "grad_norm": 1.329512119293213,
      "learning_rate": 4.219353027857694e-05,
      "loss": 0.8447,
      "step": 3809
    },
    {
      "epoch": 1.2030312598673825,
      "grad_norm": 1.798789381980896,
      "learning_rate": 4.219002175285945e-05,
      "loss": 0.8963,
      "step": 3810
    },
    {
      "epoch": 1.203347016103568,
      "grad_norm": 1.603308916091919,
      "learning_rate": 4.218651322714196e-05,
      "loss": 0.8962,
      "step": 3811
    },
    {
      "epoch": 1.2036627723397537,
      "grad_norm": 1.2910624742507935,
      "learning_rate": 4.2183004701424463e-05,
      "loss": 0.7638,
      "step": 3812
    },
    {
      "epoch": 1.2039785285759395,
      "grad_norm": 1.4012442827224731,
      "learning_rate": 4.217949617570697e-05,
      "loss": 0.8009,
      "step": 3813
    },
    {
      "epoch": 1.204294284812125,
      "grad_norm": 1.4637051820755005,
      "learning_rate": 4.217598764998948e-05,
      "loss": 0.8413,
      "step": 3814
    },
    {
      "epoch": 1.2046100410483107,
      "grad_norm": 1.658493161201477,
      "learning_rate": 4.2172479124271984e-05,
      "loss": 0.8238,
      "step": 3815
    },
    {
      "epoch": 1.2049257972844964,
      "grad_norm": 1.581385850906372,
      "learning_rate": 4.216897059855449e-05,
      "loss": 0.8642,
      "step": 3816
    },
    {
      "epoch": 1.205241553520682,
      "grad_norm": 1.6865057945251465,
      "learning_rate": 4.2165462072837e-05,
      "loss": 0.8407,
      "step": 3817
    },
    {
      "epoch": 1.2055573097568677,
      "grad_norm": 1.442489743232727,
      "learning_rate": 4.2161953547119504e-05,
      "loss": 0.8635,
      "step": 3818
    },
    {
      "epoch": 1.2058730659930534,
      "grad_norm": 1.4933756589889526,
      "learning_rate": 4.215844502140201e-05,
      "loss": 0.8789,
      "step": 3819
    },
    {
      "epoch": 1.206188822229239,
      "grad_norm": 1.23209547996521,
      "learning_rate": 4.215493649568452e-05,
      "loss": 0.8447,
      "step": 3820
    },
    {
      "epoch": 1.2065045784654247,
      "grad_norm": 1.46611750125885,
      "learning_rate": 4.2151427969967025e-05,
      "loss": 0.8674,
      "step": 3821
    },
    {
      "epoch": 1.2068203347016104,
      "grad_norm": 1.5419644117355347,
      "learning_rate": 4.214791944424953e-05,
      "loss": 0.8324,
      "step": 3822
    },
    {
      "epoch": 1.207136090937796,
      "grad_norm": 1.4790115356445312,
      "learning_rate": 4.214441091853203e-05,
      "loss": 0.8923,
      "step": 3823
    },
    {
      "epoch": 1.2074518471739817,
      "grad_norm": 1.3646835088729858,
      "learning_rate": 4.214090239281454e-05,
      "loss": 0.8653,
      "step": 3824
    },
    {
      "epoch": 1.2077676034101674,
      "grad_norm": 1.5480709075927734,
      "learning_rate": 4.2137393867097045e-05,
      "loss": 0.862,
      "step": 3825
    },
    {
      "epoch": 1.208083359646353,
      "grad_norm": 1.4514485597610474,
      "learning_rate": 4.213388534137955e-05,
      "loss": 0.794,
      "step": 3826
    },
    {
      "epoch": 1.2083991158825387,
      "grad_norm": 1.4364476203918457,
      "learning_rate": 4.213037681566206e-05,
      "loss": 0.8654,
      "step": 3827
    },
    {
      "epoch": 1.2087148721187244,
      "grad_norm": 1.758071780204773,
      "learning_rate": 4.2126868289944566e-05,
      "loss": 0.9396,
      "step": 3828
    },
    {
      "epoch": 1.20903062835491,
      "grad_norm": 1.3764573335647583,
      "learning_rate": 4.212335976422707e-05,
      "loss": 0.8794,
      "step": 3829
    },
    {
      "epoch": 1.2093463845910957,
      "grad_norm": 1.3412882089614868,
      "learning_rate": 4.211985123850958e-05,
      "loss": 0.851,
      "step": 3830
    },
    {
      "epoch": 1.2096621408272814,
      "grad_norm": 1.3702799081802368,
      "learning_rate": 4.211634271279209e-05,
      "loss": 0.8596,
      "step": 3831
    },
    {
      "epoch": 1.209977897063467,
      "grad_norm": 1.5423040390014648,
      "learning_rate": 4.211283418707459e-05,
      "loss": 0.8637,
      "step": 3832
    },
    {
      "epoch": 1.2102936532996527,
      "grad_norm": 1.720055103302002,
      "learning_rate": 4.21093256613571e-05,
      "loss": 0.9107,
      "step": 3833
    },
    {
      "epoch": 1.2106094095358384,
      "grad_norm": 1.44645357131958,
      "learning_rate": 4.2105817135639607e-05,
      "loss": 0.7984,
      "step": 3834
    },
    {
      "epoch": 1.210925165772024,
      "grad_norm": 1.7347885370254517,
      "learning_rate": 4.210230860992211e-05,
      "loss": 0.8381,
      "step": 3835
    },
    {
      "epoch": 1.2112409220082097,
      "grad_norm": 1.5060383081436157,
      "learning_rate": 4.209880008420462e-05,
      "loss": 0.8179,
      "step": 3836
    },
    {
      "epoch": 1.2115566782443954,
      "grad_norm": 1.281046748161316,
      "learning_rate": 4.209529155848713e-05,
      "loss": 0.8015,
      "step": 3837
    },
    {
      "epoch": 1.211872434480581,
      "grad_norm": 1.426518440246582,
      "learning_rate": 4.2091783032769634e-05,
      "loss": 0.8173,
      "step": 3838
    },
    {
      "epoch": 1.2121881907167666,
      "grad_norm": 1.2682271003723145,
      "learning_rate": 4.208827450705214e-05,
      "loss": 0.9129,
      "step": 3839
    },
    {
      "epoch": 1.2125039469529524,
      "grad_norm": 1.3890272378921509,
      "learning_rate": 4.208476598133465e-05,
      "loss": 0.8974,
      "step": 3840
    },
    {
      "epoch": 1.212819703189138,
      "grad_norm": 1.3431968688964844,
      "learning_rate": 4.2081257455617154e-05,
      "loss": 0.8404,
      "step": 3841
    },
    {
      "epoch": 1.2131354594253236,
      "grad_norm": 1.6054831743240356,
      "learning_rate": 4.2077748929899654e-05,
      "loss": 0.8542,
      "step": 3842
    },
    {
      "epoch": 1.2134512156615094,
      "grad_norm": 1.4830769300460815,
      "learning_rate": 4.207424040418216e-05,
      "loss": 0.9411,
      "step": 3843
    },
    {
      "epoch": 1.213766971897695,
      "grad_norm": 1.7298551797866821,
      "learning_rate": 4.207073187846467e-05,
      "loss": 0.9736,
      "step": 3844
    },
    {
      "epoch": 1.2140827281338806,
      "grad_norm": 1.3191176652908325,
      "learning_rate": 4.2067223352747175e-05,
      "loss": 0.8619,
      "step": 3845
    },
    {
      "epoch": 1.2143984843700664,
      "grad_norm": 1.0526477098464966,
      "learning_rate": 4.206371482702968e-05,
      "loss": 0.8179,
      "step": 3846
    },
    {
      "epoch": 1.2147142406062519,
      "grad_norm": 1.5243011713027954,
      "learning_rate": 4.206020630131219e-05,
      "loss": 0.8657,
      "step": 3847
    },
    {
      "epoch": 1.2150299968424376,
      "grad_norm": 1.5896483659744263,
      "learning_rate": 4.2056697775594695e-05,
      "loss": 0.8756,
      "step": 3848
    },
    {
      "epoch": 1.2153457530786234,
      "grad_norm": 1.517396330833435,
      "learning_rate": 4.205318924987721e-05,
      "loss": 0.8397,
      "step": 3849
    },
    {
      "epoch": 1.2156615093148089,
      "grad_norm": 1.467668056488037,
      "learning_rate": 4.2049680724159716e-05,
      "loss": 0.8639,
      "step": 3850
    },
    {
      "epoch": 1.2159772655509946,
      "grad_norm": 1.443577766418457,
      "learning_rate": 4.2046172198442216e-05,
      "loss": 0.8883,
      "step": 3851
    },
    {
      "epoch": 1.2162930217871804,
      "grad_norm": 1.5345938205718994,
      "learning_rate": 4.204266367272472e-05,
      "loss": 0.8979,
      "step": 3852
    },
    {
      "epoch": 1.2166087780233659,
      "grad_norm": 1.3331851959228516,
      "learning_rate": 4.203915514700723e-05,
      "loss": 0.7976,
      "step": 3853
    },
    {
      "epoch": 1.2169245342595516,
      "grad_norm": 1.4128836393356323,
      "learning_rate": 4.2035646621289736e-05,
      "loss": 0.8231,
      "step": 3854
    },
    {
      "epoch": 1.2172402904957373,
      "grad_norm": 1.48471999168396,
      "learning_rate": 4.203213809557224e-05,
      "loss": 0.8891,
      "step": 3855
    },
    {
      "epoch": 1.2175560467319229,
      "grad_norm": 1.4415122270584106,
      "learning_rate": 4.202862956985475e-05,
      "loss": 0.8627,
      "step": 3856
    },
    {
      "epoch": 1.2178718029681086,
      "grad_norm": 1.2999764680862427,
      "learning_rate": 4.2025121044137256e-05,
      "loss": 0.8632,
      "step": 3857
    },
    {
      "epoch": 1.2181875592042943,
      "grad_norm": 1.6248998641967773,
      "learning_rate": 4.202161251841976e-05,
      "loss": 0.9086,
      "step": 3858
    },
    {
      "epoch": 1.21850331544048,
      "grad_norm": 1.399597406387329,
      "learning_rate": 4.201810399270227e-05,
      "loss": 0.8025,
      "step": 3859
    },
    {
      "epoch": 1.2188190716766656,
      "grad_norm": 1.6547726392745972,
      "learning_rate": 4.201459546698478e-05,
      "loss": 0.8417,
      "step": 3860
    },
    {
      "epoch": 1.2191348279128513,
      "grad_norm": 1.4070947170257568,
      "learning_rate": 4.201108694126728e-05,
      "loss": 0.9155,
      "step": 3861
    },
    {
      "epoch": 1.2194505841490368,
      "grad_norm": 1.5363483428955078,
      "learning_rate": 4.2007578415549784e-05,
      "loss": 0.8021,
      "step": 3862
    },
    {
      "epoch": 1.2197663403852226,
      "grad_norm": 1.3492803573608398,
      "learning_rate": 4.200406988983229e-05,
      "loss": 0.8811,
      "step": 3863
    },
    {
      "epoch": 1.2200820966214083,
      "grad_norm": 1.6441287994384766,
      "learning_rate": 4.20005613641148e-05,
      "loss": 0.8956,
      "step": 3864
    },
    {
      "epoch": 1.220397852857594,
      "grad_norm": 1.449493408203125,
      "learning_rate": 4.1997052838397304e-05,
      "loss": 0.8643,
      "step": 3865
    },
    {
      "epoch": 1.2207136090937796,
      "grad_norm": 1.5367226600646973,
      "learning_rate": 4.199354431267982e-05,
      "loss": 0.8157,
      "step": 3866
    },
    {
      "epoch": 1.2210293653299653,
      "grad_norm": 1.4341907501220703,
      "learning_rate": 4.1990035786962325e-05,
      "loss": 0.8619,
      "step": 3867
    },
    {
      "epoch": 1.2213451215661508,
      "grad_norm": 1.2538748979568481,
      "learning_rate": 4.198652726124483e-05,
      "loss": 0.8223,
      "step": 3868
    },
    {
      "epoch": 1.2216608778023366,
      "grad_norm": 1.4275341033935547,
      "learning_rate": 4.198301873552734e-05,
      "loss": 0.8146,
      "step": 3869
    },
    {
      "epoch": 1.2219766340385223,
      "grad_norm": 1.564934492111206,
      "learning_rate": 4.197951020980984e-05,
      "loss": 0.8292,
      "step": 3870
    },
    {
      "epoch": 1.222292390274708,
      "grad_norm": 1.3230438232421875,
      "learning_rate": 4.1976001684092345e-05,
      "loss": 0.8836,
      "step": 3871
    },
    {
      "epoch": 1.2226081465108936,
      "grad_norm": 1.3014353513717651,
      "learning_rate": 4.197249315837485e-05,
      "loss": 0.9057,
      "step": 3872
    },
    {
      "epoch": 1.2229239027470793,
      "grad_norm": 1.3074311017990112,
      "learning_rate": 4.196898463265736e-05,
      "loss": 0.7815,
      "step": 3873
    },
    {
      "epoch": 1.2232396589832648,
      "grad_norm": 1.4844794273376465,
      "learning_rate": 4.1965476106939866e-05,
      "loss": 0.8187,
      "step": 3874
    },
    {
      "epoch": 1.2235554152194505,
      "grad_norm": 1.2235418558120728,
      "learning_rate": 4.196196758122237e-05,
      "loss": 0.8898,
      "step": 3875
    },
    {
      "epoch": 1.2238711714556363,
      "grad_norm": 1.4762402772903442,
      "learning_rate": 4.195845905550488e-05,
      "loss": 0.8358,
      "step": 3876
    },
    {
      "epoch": 1.224186927691822,
      "grad_norm": 1.5340626239776611,
      "learning_rate": 4.1954950529787386e-05,
      "loss": 0.8543,
      "step": 3877
    },
    {
      "epoch": 1.2245026839280075,
      "grad_norm": 1.3282874822616577,
      "learning_rate": 4.195144200406989e-05,
      "loss": 0.7867,
      "step": 3878
    },
    {
      "epoch": 1.2248184401641933,
      "grad_norm": 1.6872862577438354,
      "learning_rate": 4.19479334783524e-05,
      "loss": 0.8322,
      "step": 3879
    },
    {
      "epoch": 1.2251341964003788,
      "grad_norm": 1.5591706037521362,
      "learning_rate": 4.1944424952634906e-05,
      "loss": 0.939,
      "step": 3880
    },
    {
      "epoch": 1.2254499526365645,
      "grad_norm": 1.4701764583587646,
      "learning_rate": 4.1940916426917406e-05,
      "loss": 0.8192,
      "step": 3881
    },
    {
      "epoch": 1.2257657088727503,
      "grad_norm": 1.531922459602356,
      "learning_rate": 4.193740790119991e-05,
      "loss": 0.8119,
      "step": 3882
    },
    {
      "epoch": 1.226081465108936,
      "grad_norm": 1.54513418674469,
      "learning_rate": 4.193389937548242e-05,
      "loss": 0.8263,
      "step": 3883
    },
    {
      "epoch": 1.2263972213451215,
      "grad_norm": 1.5531264543533325,
      "learning_rate": 4.1930390849764934e-05,
      "loss": 0.8951,
      "step": 3884
    },
    {
      "epoch": 1.2267129775813073,
      "grad_norm": 1.4373369216918945,
      "learning_rate": 4.192688232404744e-05,
      "loss": 0.8212,
      "step": 3885
    },
    {
      "epoch": 1.2270287338174928,
      "grad_norm": 1.7146066427230835,
      "learning_rate": 4.192337379832995e-05,
      "loss": 0.8415,
      "step": 3886
    },
    {
      "epoch": 1.2273444900536785,
      "grad_norm": 1.5877792835235596,
      "learning_rate": 4.1919865272612454e-05,
      "loss": 0.8617,
      "step": 3887
    },
    {
      "epoch": 1.2276602462898643,
      "grad_norm": 1.5778800249099731,
      "learning_rate": 4.191635674689496e-05,
      "loss": 0.8604,
      "step": 3888
    },
    {
      "epoch": 1.22797600252605,
      "grad_norm": 1.7164825201034546,
      "learning_rate": 4.191284822117746e-05,
      "loss": 0.8629,
      "step": 3889
    },
    {
      "epoch": 1.2282917587622355,
      "grad_norm": 1.470838189125061,
      "learning_rate": 4.190933969545997e-05,
      "loss": 0.8902,
      "step": 3890
    },
    {
      "epoch": 1.2286075149984212,
      "grad_norm": 1.5728129148483276,
      "learning_rate": 4.1905831169742475e-05,
      "loss": 0.9189,
      "step": 3891
    },
    {
      "epoch": 1.2289232712346068,
      "grad_norm": 1.35187828540802,
      "learning_rate": 4.190232264402498e-05,
      "loss": 0.8182,
      "step": 3892
    },
    {
      "epoch": 1.2292390274707925,
      "grad_norm": 1.5535212755203247,
      "learning_rate": 4.189881411830749e-05,
      "loss": 0.8023,
      "step": 3893
    },
    {
      "epoch": 1.2295547837069782,
      "grad_norm": 1.5296915769577026,
      "learning_rate": 4.1895305592589995e-05,
      "loss": 0.8757,
      "step": 3894
    },
    {
      "epoch": 1.229870539943164,
      "grad_norm": 1.6527413129806519,
      "learning_rate": 4.18917970668725e-05,
      "loss": 0.8198,
      "step": 3895
    },
    {
      "epoch": 1.2301862961793495,
      "grad_norm": 1.2901008129119873,
      "learning_rate": 4.188828854115501e-05,
      "loss": 0.7978,
      "step": 3896
    },
    {
      "epoch": 1.2305020524155352,
      "grad_norm": 2.045659303665161,
      "learning_rate": 4.1884780015437515e-05,
      "loss": 0.8703,
      "step": 3897
    },
    {
      "epoch": 1.230817808651721,
      "grad_norm": 1.620684266090393,
      "learning_rate": 4.188127148972002e-05,
      "loss": 0.7954,
      "step": 3898
    },
    {
      "epoch": 1.2311335648879065,
      "grad_norm": 1.5619772672653198,
      "learning_rate": 4.187776296400253e-05,
      "loss": 0.7863,
      "step": 3899
    },
    {
      "epoch": 1.2314493211240922,
      "grad_norm": 1.7467846870422363,
      "learning_rate": 4.187425443828503e-05,
      "loss": 0.8725,
      "step": 3900
    },
    {
      "epoch": 1.231765077360278,
      "grad_norm": 1.4695998430252075,
      "learning_rate": 4.187074591256754e-05,
      "loss": 0.839,
      "step": 3901
    },
    {
      "epoch": 1.2320808335964635,
      "grad_norm": 1.7443143129348755,
      "learning_rate": 4.186723738685005e-05,
      "loss": 0.8005,
      "step": 3902
    },
    {
      "epoch": 1.2323965898326492,
      "grad_norm": 1.3413150310516357,
      "learning_rate": 4.1863728861132556e-05,
      "loss": 0.7327,
      "step": 3903
    },
    {
      "epoch": 1.232712346068835,
      "grad_norm": 1.3886475563049316,
      "learning_rate": 4.186022033541506e-05,
      "loss": 0.8785,
      "step": 3904
    },
    {
      "epoch": 1.2330281023050205,
      "grad_norm": 1.6796720027923584,
      "learning_rate": 4.185671180969757e-05,
      "loss": 0.8569,
      "step": 3905
    },
    {
      "epoch": 1.2333438585412062,
      "grad_norm": 1.5373895168304443,
      "learning_rate": 4.185320328398008e-05,
      "loss": 0.8138,
      "step": 3906
    },
    {
      "epoch": 1.233659614777392,
      "grad_norm": 1.4316993951797485,
      "learning_rate": 4.1849694758262584e-05,
      "loss": 0.8274,
      "step": 3907
    },
    {
      "epoch": 1.2339753710135775,
      "grad_norm": 1.6877665519714355,
      "learning_rate": 4.184618623254509e-05,
      "loss": 0.8226,
      "step": 3908
    },
    {
      "epoch": 1.2342911272497632,
      "grad_norm": 1.6799288988113403,
      "learning_rate": 4.184267770682759e-05,
      "loss": 0.8154,
      "step": 3909
    },
    {
      "epoch": 1.234606883485949,
      "grad_norm": 1.4573338031768799,
      "learning_rate": 4.18391691811101e-05,
      "loss": 0.8775,
      "step": 3910
    },
    {
      "epoch": 1.2349226397221345,
      "grad_norm": 1.4933159351348877,
      "learning_rate": 4.1835660655392604e-05,
      "loss": 0.8618,
      "step": 3911
    },
    {
      "epoch": 1.2352383959583202,
      "grad_norm": 1.4587494134902954,
      "learning_rate": 4.183215212967511e-05,
      "loss": 0.8611,
      "step": 3912
    },
    {
      "epoch": 1.235554152194506,
      "grad_norm": 1.379380226135254,
      "learning_rate": 4.182864360395762e-05,
      "loss": 0.8323,
      "step": 3913
    },
    {
      "epoch": 1.2358699084306914,
      "grad_norm": 1.583358883857727,
      "learning_rate": 4.1825135078240124e-05,
      "loss": 0.8276,
      "step": 3914
    },
    {
      "epoch": 1.2361856646668772,
      "grad_norm": 1.5200313329696655,
      "learning_rate": 4.182162655252263e-05,
      "loss": 0.8998,
      "step": 3915
    },
    {
      "epoch": 1.236501420903063,
      "grad_norm": 1.5336095094680786,
      "learning_rate": 4.181811802680514e-05,
      "loss": 0.9153,
      "step": 3916
    },
    {
      "epoch": 1.2368171771392484,
      "grad_norm": 1.580023169517517,
      "learning_rate": 4.1814609501087645e-05,
      "loss": 0.8626,
      "step": 3917
    },
    {
      "epoch": 1.2371329333754342,
      "grad_norm": 1.5615235567092896,
      "learning_rate": 4.181110097537015e-05,
      "loss": 0.9111,
      "step": 3918
    },
    {
      "epoch": 1.23744868961162,
      "grad_norm": 1.520442247390747,
      "learning_rate": 4.180759244965266e-05,
      "loss": 0.8215,
      "step": 3919
    },
    {
      "epoch": 1.2377644458478054,
      "grad_norm": 1.464592695236206,
      "learning_rate": 4.1804083923935165e-05,
      "loss": 0.8361,
      "step": 3920
    },
    {
      "epoch": 1.2380802020839912,
      "grad_norm": 1.356971025466919,
      "learning_rate": 4.180057539821767e-05,
      "loss": 0.8923,
      "step": 3921
    },
    {
      "epoch": 1.238395958320177,
      "grad_norm": 1.2866865396499634,
      "learning_rate": 4.179706687250018e-05,
      "loss": 0.7896,
      "step": 3922
    },
    {
      "epoch": 1.2387117145563624,
      "grad_norm": 1.4942744970321655,
      "learning_rate": 4.1793558346782686e-05,
      "loss": 0.8414,
      "step": 3923
    },
    {
      "epoch": 1.2390274707925482,
      "grad_norm": 1.3535505533218384,
      "learning_rate": 4.179004982106519e-05,
      "loss": 0.7503,
      "step": 3924
    },
    {
      "epoch": 1.239343227028734,
      "grad_norm": 1.5723042488098145,
      "learning_rate": 4.17865412953477e-05,
      "loss": 0.8393,
      "step": 3925
    },
    {
      "epoch": 1.2396589832649194,
      "grad_norm": 1.45696222782135,
      "learning_rate": 4.1783032769630206e-05,
      "loss": 0.827,
      "step": 3926
    },
    {
      "epoch": 1.2399747395011051,
      "grad_norm": 1.6274158954620361,
      "learning_rate": 4.177952424391271e-05,
      "loss": 0.8733,
      "step": 3927
    },
    {
      "epoch": 1.2402904957372909,
      "grad_norm": 1.4010738134384155,
      "learning_rate": 4.177601571819521e-05,
      "loss": 0.7995,
      "step": 3928
    },
    {
      "epoch": 1.2406062519734764,
      "grad_norm": 1.695074439048767,
      "learning_rate": 4.177250719247772e-05,
      "loss": 0.839,
      "step": 3929
    },
    {
      "epoch": 1.2409220082096621,
      "grad_norm": 1.4634625911712646,
      "learning_rate": 4.176899866676023e-05,
      "loss": 0.8667,
      "step": 3930
    },
    {
      "epoch": 1.2412377644458479,
      "grad_norm": 1.7929047346115112,
      "learning_rate": 4.1765490141042734e-05,
      "loss": 0.8414,
      "step": 3931
    },
    {
      "epoch": 1.2415535206820334,
      "grad_norm": 1.4586666822433472,
      "learning_rate": 4.176198161532524e-05,
      "loss": 0.8417,
      "step": 3932
    },
    {
      "epoch": 1.2418692769182191,
      "grad_norm": 1.8175556659698486,
      "learning_rate": 4.175847308960775e-05,
      "loss": 0.9709,
      "step": 3933
    },
    {
      "epoch": 1.2421850331544049,
      "grad_norm": 1.4801464080810547,
      "learning_rate": 4.1754964563890254e-05,
      "loss": 0.8405,
      "step": 3934
    },
    {
      "epoch": 1.2425007893905904,
      "grad_norm": 1.4470648765563965,
      "learning_rate": 4.175145603817276e-05,
      "loss": 0.8124,
      "step": 3935
    },
    {
      "epoch": 1.2428165456267761,
      "grad_norm": 1.5876513719558716,
      "learning_rate": 4.1747947512455274e-05,
      "loss": 0.8546,
      "step": 3936
    },
    {
      "epoch": 1.2431323018629619,
      "grad_norm": 1.4106171131134033,
      "learning_rate": 4.1744438986737774e-05,
      "loss": 0.8004,
      "step": 3937
    },
    {
      "epoch": 1.2434480580991474,
      "grad_norm": 1.9288426637649536,
      "learning_rate": 4.174093046102028e-05,
      "loss": 0.8397,
      "step": 3938
    },
    {
      "epoch": 1.2437638143353331,
      "grad_norm": 1.867264747619629,
      "learning_rate": 4.173742193530279e-05,
      "loss": 0.8626,
      "step": 3939
    },
    {
      "epoch": 1.2440795705715189,
      "grad_norm": 1.822218656539917,
      "learning_rate": 4.1733913409585295e-05,
      "loss": 0.7756,
      "step": 3940
    },
    {
      "epoch": 1.2443953268077044,
      "grad_norm": 1.6663250923156738,
      "learning_rate": 4.17304048838678e-05,
      "loss": 0.879,
      "step": 3941
    },
    {
      "epoch": 1.24471108304389,
      "grad_norm": 1.4544328451156616,
      "learning_rate": 4.172689635815031e-05,
      "loss": 0.8555,
      "step": 3942
    },
    {
      "epoch": 1.2450268392800758,
      "grad_norm": 1.5109134912490845,
      "learning_rate": 4.1723387832432815e-05,
      "loss": 0.8252,
      "step": 3943
    },
    {
      "epoch": 1.2453425955162614,
      "grad_norm": 1.7843502759933472,
      "learning_rate": 4.171987930671532e-05,
      "loss": 0.8343,
      "step": 3944
    },
    {
      "epoch": 1.245658351752447,
      "grad_norm": 1.7015169858932495,
      "learning_rate": 4.171637078099783e-05,
      "loss": 0.8564,
      "step": 3945
    },
    {
      "epoch": 1.2459741079886328,
      "grad_norm": 1.8349554538726807,
      "learning_rate": 4.1712862255280336e-05,
      "loss": 0.8924,
      "step": 3946
    },
    {
      "epoch": 1.2462898642248184,
      "grad_norm": 1.9611668586730957,
      "learning_rate": 4.1709353729562836e-05,
      "loss": 0.7908,
      "step": 3947
    },
    {
      "epoch": 1.246605620461004,
      "grad_norm": 2.0568509101867676,
      "learning_rate": 4.170584520384534e-05,
      "loss": 0.8535,
      "step": 3948
    },
    {
      "epoch": 1.2469213766971898,
      "grad_norm": 1.5097066164016724,
      "learning_rate": 4.170233667812785e-05,
      "loss": 0.9057,
      "step": 3949
    },
    {
      "epoch": 1.2472371329333753,
      "grad_norm": 1.4783639907836914,
      "learning_rate": 4.1698828152410356e-05,
      "loss": 0.7888,
      "step": 3950
    },
    {
      "epoch": 1.247552889169561,
      "grad_norm": 1.7196366786956787,
      "learning_rate": 4.169531962669286e-05,
      "loss": 0.8692,
      "step": 3951
    },
    {
      "epoch": 1.2478686454057468,
      "grad_norm": 1.5030310153961182,
      "learning_rate": 4.169181110097537e-05,
      "loss": 0.7883,
      "step": 3952
    },
    {
      "epoch": 1.2481844016419323,
      "grad_norm": 1.4365334510803223,
      "learning_rate": 4.168830257525788e-05,
      "loss": 0.8921,
      "step": 3953
    },
    {
      "epoch": 1.248500157878118,
      "grad_norm": 1.655436635017395,
      "learning_rate": 4.168479404954039e-05,
      "loss": 0.8084,
      "step": 3954
    },
    {
      "epoch": 1.2488159141143038,
      "grad_norm": 1.6098144054412842,
      "learning_rate": 4.16812855238229e-05,
      "loss": 0.8569,
      "step": 3955
    },
    {
      "epoch": 1.2491316703504893,
      "grad_norm": 1.4326694011688232,
      "learning_rate": 4.16777769981054e-05,
      "loss": 0.8576,
      "step": 3956
    },
    {
      "epoch": 1.249447426586675,
      "grad_norm": 1.4316906929016113,
      "learning_rate": 4.1674268472387904e-05,
      "loss": 0.8875,
      "step": 3957
    },
    {
      "epoch": 1.2497631828228608,
      "grad_norm": 1.508374571800232,
      "learning_rate": 4.167075994667041e-05,
      "loss": 0.8779,
      "step": 3958
    },
    {
      "epoch": 1.2500789390590463,
      "grad_norm": 1.750179409980774,
      "learning_rate": 4.166725142095292e-05,
      "loss": 0.7935,
      "step": 3959
    },
    {
      "epoch": 1.250394695295232,
      "grad_norm": 1.6434983015060425,
      "learning_rate": 4.1663742895235424e-05,
      "loss": 0.8277,
      "step": 3960
    },
    {
      "epoch": 1.2507104515314178,
      "grad_norm": 1.4686481952667236,
      "learning_rate": 4.166023436951793e-05,
      "loss": 0.843,
      "step": 3961
    },
    {
      "epoch": 1.2510262077676035,
      "grad_norm": 1.5459500551223755,
      "learning_rate": 4.165672584380044e-05,
      "loss": 0.8272,
      "step": 3962
    },
    {
      "epoch": 1.251341964003789,
      "grad_norm": 1.5343471765518188,
      "learning_rate": 4.1653217318082945e-05,
      "loss": 0.8308,
      "step": 3963
    },
    {
      "epoch": 1.2516577202399748,
      "grad_norm": 1.3424772024154663,
      "learning_rate": 4.164970879236545e-05,
      "loss": 0.7776,
      "step": 3964
    },
    {
      "epoch": 1.2519734764761603,
      "grad_norm": 1.4925307035446167,
      "learning_rate": 4.164620026664796e-05,
      "loss": 0.8551,
      "step": 3965
    },
    {
      "epoch": 1.252289232712346,
      "grad_norm": 1.4465644359588623,
      "learning_rate": 4.164269174093046e-05,
      "loss": 0.8811,
      "step": 3966
    },
    {
      "epoch": 1.2526049889485318,
      "grad_norm": 1.4816621541976929,
      "learning_rate": 4.1639183215212965e-05,
      "loss": 0.8704,
      "step": 3967
    },
    {
      "epoch": 1.2529207451847175,
      "grad_norm": 1.8476746082305908,
      "learning_rate": 4.163567468949547e-05,
      "loss": 0.9335,
      "step": 3968
    },
    {
      "epoch": 1.253236501420903,
      "grad_norm": 1.3277174234390259,
      "learning_rate": 4.163216616377798e-05,
      "loss": 0.8358,
      "step": 3969
    },
    {
      "epoch": 1.2535522576570888,
      "grad_norm": 1.796445369720459,
      "learning_rate": 4.1628657638060486e-05,
      "loss": 0.8929,
      "step": 3970
    },
    {
      "epoch": 1.2538680138932743,
      "grad_norm": 1.5302801132202148,
      "learning_rate": 4.1625149112343e-05,
      "loss": 0.7907,
      "step": 3971
    },
    {
      "epoch": 1.25418377012946,
      "grad_norm": 2.180164337158203,
      "learning_rate": 4.1621640586625506e-05,
      "loss": 0.8379,
      "step": 3972
    },
    {
      "epoch": 1.2544995263656458,
      "grad_norm": 2.0235002040863037,
      "learning_rate": 4.161813206090801e-05,
      "loss": 0.9253,
      "step": 3973
    },
    {
      "epoch": 1.2548152826018315,
      "grad_norm": 1.7548242807388306,
      "learning_rate": 4.161462353519052e-05,
      "loss": 0.8716,
      "step": 3974
    },
    {
      "epoch": 1.255131038838017,
      "grad_norm": 1.7052687406539917,
      "learning_rate": 4.161111500947302e-05,
      "loss": 0.8718,
      "step": 3975
    },
    {
      "epoch": 1.2554467950742028,
      "grad_norm": 1.1791009902954102,
      "learning_rate": 4.1607606483755527e-05,
      "loss": 0.805,
      "step": 3976
    },
    {
      "epoch": 1.2557625513103883,
      "grad_norm": 1.6199541091918945,
      "learning_rate": 4.160409795803803e-05,
      "loss": 0.8779,
      "step": 3977
    },
    {
      "epoch": 1.256078307546574,
      "grad_norm": 1.6714086532592773,
      "learning_rate": 4.160058943232054e-05,
      "loss": 0.7765,
      "step": 3978
    },
    {
      "epoch": 1.2563940637827598,
      "grad_norm": 1.4039887189865112,
      "learning_rate": 4.159708090660305e-05,
      "loss": 0.8868,
      "step": 3979
    },
    {
      "epoch": 1.2567098200189455,
      "grad_norm": 1.6768770217895508,
      "learning_rate": 4.1593572380885554e-05,
      "loss": 0.8998,
      "step": 3980
    },
    {
      "epoch": 1.257025576255131,
      "grad_norm": 1.4898263216018677,
      "learning_rate": 4.159006385516806e-05,
      "loss": 0.8356,
      "step": 3981
    },
    {
      "epoch": 1.2573413324913167,
      "grad_norm": 1.640607237815857,
      "learning_rate": 4.158655532945057e-05,
      "loss": 0.8686,
      "step": 3982
    },
    {
      "epoch": 1.2576570887275023,
      "grad_norm": 1.7564013004302979,
      "learning_rate": 4.1583046803733074e-05,
      "loss": 0.9172,
      "step": 3983
    },
    {
      "epoch": 1.257972844963688,
      "grad_norm": 1.480151891708374,
      "learning_rate": 4.157953827801558e-05,
      "loss": 0.9138,
      "step": 3984
    },
    {
      "epoch": 1.2582886011998737,
      "grad_norm": 1.586480736732483,
      "learning_rate": 4.157602975229809e-05,
      "loss": 0.8234,
      "step": 3985
    },
    {
      "epoch": 1.2586043574360595,
      "grad_norm": 1.5808840990066528,
      "learning_rate": 4.157252122658059e-05,
      "loss": 0.8441,
      "step": 3986
    },
    {
      "epoch": 1.258920113672245,
      "grad_norm": 1.4451979398727417,
      "learning_rate": 4.1569012700863095e-05,
      "loss": 0.7867,
      "step": 3987
    },
    {
      "epoch": 1.2592358699084307,
      "grad_norm": 1.9020767211914062,
      "learning_rate": 4.15655041751456e-05,
      "loss": 0.8328,
      "step": 3988
    },
    {
      "epoch": 1.2595516261446162,
      "grad_norm": 1.4088338613510132,
      "learning_rate": 4.1561995649428115e-05,
      "loss": 0.8147,
      "step": 3989
    },
    {
      "epoch": 1.259867382380802,
      "grad_norm": 1.6336491107940674,
      "learning_rate": 4.155848712371062e-05,
      "loss": 0.8701,
      "step": 3990
    },
    {
      "epoch": 1.2601831386169877,
      "grad_norm": 1.480569839477539,
      "learning_rate": 4.155497859799313e-05,
      "loss": 0.7871,
      "step": 3991
    },
    {
      "epoch": 1.2604988948531735,
      "grad_norm": 1.5152236223220825,
      "learning_rate": 4.1551470072275636e-05,
      "loss": 0.878,
      "step": 3992
    },
    {
      "epoch": 1.260814651089359,
      "grad_norm": 1.5319254398345947,
      "learning_rate": 4.154796154655814e-05,
      "loss": 0.8673,
      "step": 3993
    },
    {
      "epoch": 1.2611304073255447,
      "grad_norm": 1.6211885213851929,
      "learning_rate": 4.154445302084064e-05,
      "loss": 0.8824,
      "step": 3994
    },
    {
      "epoch": 1.2614461635617302,
      "grad_norm": 1.468929409980774,
      "learning_rate": 4.154094449512315e-05,
      "loss": 0.8519,
      "step": 3995
    },
    {
      "epoch": 1.261761919797916,
      "grad_norm": 1.6606769561767578,
      "learning_rate": 4.1537435969405656e-05,
      "loss": 0.9071,
      "step": 3996
    },
    {
      "epoch": 1.2620776760341017,
      "grad_norm": 1.385574460029602,
      "learning_rate": 4.153392744368816e-05,
      "loss": 0.7969,
      "step": 3997
    },
    {
      "epoch": 1.2623934322702874,
      "grad_norm": 1.321311593055725,
      "learning_rate": 4.153041891797067e-05,
      "loss": 0.8181,
      "step": 3998
    },
    {
      "epoch": 1.262709188506473,
      "grad_norm": 1.3997597694396973,
      "learning_rate": 4.1526910392253176e-05,
      "loss": 0.7952,
      "step": 3999
    },
    {
      "epoch": 1.2630249447426587,
      "grad_norm": 1.6618342399597168,
      "learning_rate": 4.152340186653568e-05,
      "loss": 0.78,
      "step": 4000
    },
    {
      "epoch": 1.2630249447426587,
      "eval_loss": 0.8333959579467773,
      "eval_runtime": 4.6403,
      "eval_samples_per_second": 536.607,
      "eval_steps_per_second": 4.526,
      "step": 4000
    },
    {
      "epoch": 1.2633407009788442,
      "grad_norm": 1.4699405431747437,
      "learning_rate": 4.151989334081819e-05,
      "loss": 0.7394,
      "step": 4001
    },
    {
      "epoch": 1.26365645721503,
      "grad_norm": 1.4466103315353394,
      "learning_rate": 4.15163848151007e-05,
      "loss": 0.8695,
      "step": 4002
    },
    {
      "epoch": 1.2639722134512157,
      "grad_norm": 1.5211594104766846,
      "learning_rate": 4.1512876289383204e-05,
      "loss": 0.841,
      "step": 4003
    },
    {
      "epoch": 1.2642879696874014,
      "grad_norm": 1.7041658163070679,
      "learning_rate": 4.150936776366571e-05,
      "loss": 0.8843,
      "step": 4004
    },
    {
      "epoch": 1.264603725923587,
      "grad_norm": 1.4558308124542236,
      "learning_rate": 4.150585923794821e-05,
      "loss": 0.8508,
      "step": 4005
    },
    {
      "epoch": 1.2649194821597727,
      "grad_norm": 1.6265546083450317,
      "learning_rate": 4.1502350712230724e-05,
      "loss": 0.7659,
      "step": 4006
    },
    {
      "epoch": 1.2652352383959582,
      "grad_norm": 1.5147342681884766,
      "learning_rate": 4.149884218651323e-05,
      "loss": 0.8748,
      "step": 4007
    },
    {
      "epoch": 1.265550994632144,
      "grad_norm": 1.420322299003601,
      "learning_rate": 4.149533366079574e-05,
      "loss": 0.8132,
      "step": 4008
    },
    {
      "epoch": 1.2658667508683297,
      "grad_norm": 1.8604329824447632,
      "learning_rate": 4.1491825135078245e-05,
      "loss": 0.8037,
      "step": 4009
    },
    {
      "epoch": 1.2661825071045154,
      "grad_norm": 1.584606647491455,
      "learning_rate": 4.148831660936075e-05,
      "loss": 0.886,
      "step": 4010
    },
    {
      "epoch": 1.266498263340701,
      "grad_norm": 1.5213383436203003,
      "learning_rate": 4.148480808364326e-05,
      "loss": 0.8482,
      "step": 4011
    },
    {
      "epoch": 1.2668140195768867,
      "grad_norm": 1.5686222314834595,
      "learning_rate": 4.1481299557925765e-05,
      "loss": 0.7723,
      "step": 4012
    },
    {
      "epoch": 1.2671297758130722,
      "grad_norm": 1.6569563150405884,
      "learning_rate": 4.147779103220827e-05,
      "loss": 0.9005,
      "step": 4013
    },
    {
      "epoch": 1.267445532049258,
      "grad_norm": 1.590052604675293,
      "learning_rate": 4.147428250649077e-05,
      "loss": 0.853,
      "step": 4014
    },
    {
      "epoch": 1.2677612882854437,
      "grad_norm": 1.6000232696533203,
      "learning_rate": 4.147077398077328e-05,
      "loss": 0.8919,
      "step": 4015
    },
    {
      "epoch": 1.2680770445216294,
      "grad_norm": 1.4781519174575806,
      "learning_rate": 4.1467265455055786e-05,
      "loss": 0.8599,
      "step": 4016
    },
    {
      "epoch": 1.268392800757815,
      "grad_norm": 1.5336225032806396,
      "learning_rate": 4.146375692933829e-05,
      "loss": 0.8319,
      "step": 4017
    },
    {
      "epoch": 1.2687085569940006,
      "grad_norm": 1.710074782371521,
      "learning_rate": 4.14602484036208e-05,
      "loss": 0.8383,
      "step": 4018
    },
    {
      "epoch": 1.2690243132301862,
      "grad_norm": 1.62420654296875,
      "learning_rate": 4.1456739877903306e-05,
      "loss": 0.827,
      "step": 4019
    },
    {
      "epoch": 1.269340069466372,
      "grad_norm": 1.4683064222335815,
      "learning_rate": 4.145323135218581e-05,
      "loss": 0.9012,
      "step": 4020
    },
    {
      "epoch": 1.2696558257025576,
      "grad_norm": 1.7273567914962769,
      "learning_rate": 4.144972282646832e-05,
      "loss": 0.9062,
      "step": 4021
    },
    {
      "epoch": 1.2699715819387434,
      "grad_norm": 1.7483359575271606,
      "learning_rate": 4.1446214300750826e-05,
      "loss": 0.8749,
      "step": 4022
    },
    {
      "epoch": 1.270287338174929,
      "grad_norm": 1.380494236946106,
      "learning_rate": 4.144270577503333e-05,
      "loss": 0.8052,
      "step": 4023
    },
    {
      "epoch": 1.2706030944111146,
      "grad_norm": 1.623705267906189,
      "learning_rate": 4.143919724931584e-05,
      "loss": 0.8657,
      "step": 4024
    },
    {
      "epoch": 1.2709188506473001,
      "grad_norm": 1.4989138841629028,
      "learning_rate": 4.143568872359835e-05,
      "loss": 0.8857,
      "step": 4025
    },
    {
      "epoch": 1.2712346068834859,
      "grad_norm": 1.474645733833313,
      "learning_rate": 4.1432180197880854e-05,
      "loss": 0.7955,
      "step": 4026
    },
    {
      "epoch": 1.2715503631196716,
      "grad_norm": 1.43124258518219,
      "learning_rate": 4.142867167216336e-05,
      "loss": 0.8631,
      "step": 4027
    },
    {
      "epoch": 1.2718661193558574,
      "grad_norm": 1.626041054725647,
      "learning_rate": 4.142516314644587e-05,
      "loss": 0.8437,
      "step": 4028
    },
    {
      "epoch": 1.2721818755920429,
      "grad_norm": 1.4264518022537231,
      "learning_rate": 4.1421654620728374e-05,
      "loss": 0.8303,
      "step": 4029
    },
    {
      "epoch": 1.2724976318282286,
      "grad_norm": 1.4568678140640259,
      "learning_rate": 4.141814609501088e-05,
      "loss": 0.7766,
      "step": 4030
    },
    {
      "epoch": 1.2728133880644144,
      "grad_norm": 1.4676798582077026,
      "learning_rate": 4.141463756929339e-05,
      "loss": 0.8508,
      "step": 4031
    },
    {
      "epoch": 1.2731291443005999,
      "grad_norm": 1.5388050079345703,
      "learning_rate": 4.1411129043575895e-05,
      "loss": 0.791,
      "step": 4032
    },
    {
      "epoch": 1.2734449005367856,
      "grad_norm": 1.4092440605163574,
      "learning_rate": 4.1407620517858395e-05,
      "loss": 0.8608,
      "step": 4033
    },
    {
      "epoch": 1.2737606567729713,
      "grad_norm": 1.6175249814987183,
      "learning_rate": 4.14041119921409e-05,
      "loss": 0.7907,
      "step": 4034
    },
    {
      "epoch": 1.2740764130091569,
      "grad_norm": 1.446447730064392,
      "learning_rate": 4.140060346642341e-05,
      "loss": 0.8362,
      "step": 4035
    },
    {
      "epoch": 1.2743921692453426,
      "grad_norm": 2.9028213024139404,
      "learning_rate": 4.1397094940705915e-05,
      "loss": 0.8512,
      "step": 4036
    },
    {
      "epoch": 1.2747079254815283,
      "grad_norm": 1.3936593532562256,
      "learning_rate": 4.139358641498842e-05,
      "loss": 0.8122,
      "step": 4037
    },
    {
      "epoch": 1.2750236817177139,
      "grad_norm": 1.4408107995986938,
      "learning_rate": 4.139007788927093e-05,
      "loss": 0.9244,
      "step": 4038
    },
    {
      "epoch": 1.2753394379538996,
      "grad_norm": 1.563306212425232,
      "learning_rate": 4.1386569363553435e-05,
      "loss": 0.8924,
      "step": 4039
    },
    {
      "epoch": 1.2756551941900853,
      "grad_norm": 1.4983826875686646,
      "learning_rate": 4.138306083783594e-05,
      "loss": 0.8693,
      "step": 4040
    },
    {
      "epoch": 1.2759709504262708,
      "grad_norm": 1.4815597534179688,
      "learning_rate": 4.1379552312118456e-05,
      "loss": 0.8424,
      "step": 4041
    },
    {
      "epoch": 1.2762867066624566,
      "grad_norm": 1.3197910785675049,
      "learning_rate": 4.1376043786400956e-05,
      "loss": 0.8233,
      "step": 4042
    },
    {
      "epoch": 1.2766024628986423,
      "grad_norm": 1.440425992012024,
      "learning_rate": 4.137253526068346e-05,
      "loss": 0.7874,
      "step": 4043
    },
    {
      "epoch": 1.2769182191348278,
      "grad_norm": 1.6186802387237549,
      "learning_rate": 4.136902673496597e-05,
      "loss": 0.8671,
      "step": 4044
    },
    {
      "epoch": 1.2772339753710136,
      "grad_norm": 1.6998566389083862,
      "learning_rate": 4.1365518209248476e-05,
      "loss": 0.9204,
      "step": 4045
    },
    {
      "epoch": 1.2775497316071993,
      "grad_norm": 1.5972449779510498,
      "learning_rate": 4.136200968353098e-05,
      "loss": 0.8591,
      "step": 4046
    },
    {
      "epoch": 1.2778654878433848,
      "grad_norm": 1.5569568872451782,
      "learning_rate": 4.135850115781349e-05,
      "loss": 0.7908,
      "step": 4047
    },
    {
      "epoch": 1.2781812440795706,
      "grad_norm": 1.2589541673660278,
      "learning_rate": 4.1354992632096e-05,
      "loss": 0.8168,
      "step": 4048
    },
    {
      "epoch": 1.2784970003157563,
      "grad_norm": 1.4311097860336304,
      "learning_rate": 4.1351484106378504e-05,
      "loss": 0.798,
      "step": 4049
    },
    {
      "epoch": 1.2788127565519418,
      "grad_norm": 1.4893255233764648,
      "learning_rate": 4.134797558066101e-05,
      "loss": 0.8437,
      "step": 4050
    },
    {
      "epoch": 1.2791285127881276,
      "grad_norm": 1.7375470399856567,
      "learning_rate": 4.134446705494352e-05,
      "loss": 0.8069,
      "step": 4051
    },
    {
      "epoch": 1.2794442690243133,
      "grad_norm": 1.563711404800415,
      "learning_rate": 4.134095852922602e-05,
      "loss": 0.7993,
      "step": 4052
    },
    {
      "epoch": 1.2797600252604988,
      "grad_norm": 1.529415249824524,
      "learning_rate": 4.1337450003508524e-05,
      "loss": 0.8182,
      "step": 4053
    },
    {
      "epoch": 1.2800757814966846,
      "grad_norm": 1.3801829814910889,
      "learning_rate": 4.133394147779103e-05,
      "loss": 0.7926,
      "step": 4054
    },
    {
      "epoch": 1.2803915377328703,
      "grad_norm": 1.6253492832183838,
      "learning_rate": 4.133043295207354e-05,
      "loss": 0.8958,
      "step": 4055
    },
    {
      "epoch": 1.2807072939690558,
      "grad_norm": 1.4411224126815796,
      "learning_rate": 4.1326924426356045e-05,
      "loss": 0.8968,
      "step": 4056
    },
    {
      "epoch": 1.2810230502052415,
      "grad_norm": 1.5180435180664062,
      "learning_rate": 4.132341590063855e-05,
      "loss": 0.8259,
      "step": 4057
    },
    {
      "epoch": 1.2813388064414273,
      "grad_norm": 1.8099654912948608,
      "learning_rate": 4.131990737492106e-05,
      "loss": 0.8333,
      "step": 4058
    },
    {
      "epoch": 1.281654562677613,
      "grad_norm": 1.5632810592651367,
      "learning_rate": 4.131639884920357e-05,
      "loss": 0.8934,
      "step": 4059
    },
    {
      "epoch": 1.2819703189137985,
      "grad_norm": 1.7994818687438965,
      "learning_rate": 4.131289032348608e-05,
      "loss": 0.9138,
      "step": 4060
    },
    {
      "epoch": 1.2822860751499843,
      "grad_norm": 1.4595025777816772,
      "learning_rate": 4.130938179776858e-05,
      "loss": 0.847,
      "step": 4061
    },
    {
      "epoch": 1.2826018313861698,
      "grad_norm": 1.5881859064102173,
      "learning_rate": 4.1305873272051085e-05,
      "loss": 0.829,
      "step": 4062
    },
    {
      "epoch": 1.2829175876223555,
      "grad_norm": 1.6102286577224731,
      "learning_rate": 4.130236474633359e-05,
      "loss": 0.8565,
      "step": 4063
    },
    {
      "epoch": 1.2832333438585413,
      "grad_norm": 1.4733242988586426,
      "learning_rate": 4.12988562206161e-05,
      "loss": 0.8543,
      "step": 4064
    },
    {
      "epoch": 1.283549100094727,
      "grad_norm": 1.8382903337478638,
      "learning_rate": 4.1295347694898606e-05,
      "loss": 0.9468,
      "step": 4065
    },
    {
      "epoch": 1.2838648563309125,
      "grad_norm": 1.7715463638305664,
      "learning_rate": 4.129183916918111e-05,
      "loss": 0.8644,
      "step": 4066
    },
    {
      "epoch": 1.2841806125670983,
      "grad_norm": 1.4031401872634888,
      "learning_rate": 4.128833064346362e-05,
      "loss": 0.8078,
      "step": 4067
    },
    {
      "epoch": 1.2844963688032838,
      "grad_norm": 1.5900565385818481,
      "learning_rate": 4.1284822117746126e-05,
      "loss": 0.7906,
      "step": 4068
    },
    {
      "epoch": 1.2848121250394695,
      "grad_norm": 1.5115939378738403,
      "learning_rate": 4.128131359202863e-05,
      "loss": 0.8831,
      "step": 4069
    },
    {
      "epoch": 1.2851278812756552,
      "grad_norm": 1.336388349533081,
      "learning_rate": 4.127780506631114e-05,
      "loss": 0.794,
      "step": 4070
    },
    {
      "epoch": 1.285443637511841,
      "grad_norm": 1.3074452877044678,
      "learning_rate": 4.127429654059364e-05,
      "loss": 0.8896,
      "step": 4071
    },
    {
      "epoch": 1.2857593937480265,
      "grad_norm": 1.2016587257385254,
      "learning_rate": 4.127078801487615e-05,
      "loss": 0.8503,
      "step": 4072
    },
    {
      "epoch": 1.2860751499842122,
      "grad_norm": 1.5935674905776978,
      "learning_rate": 4.1267279489158654e-05,
      "loss": 0.8735,
      "step": 4073
    },
    {
      "epoch": 1.2863909062203978,
      "grad_norm": 1.2957243919372559,
      "learning_rate": 4.126377096344116e-05,
      "loss": 0.8455,
      "step": 4074
    },
    {
      "epoch": 1.2867066624565835,
      "grad_norm": 1.302349328994751,
      "learning_rate": 4.126026243772367e-05,
      "loss": 0.8082,
      "step": 4075
    },
    {
      "epoch": 1.2870224186927692,
      "grad_norm": 1.5224480628967285,
      "learning_rate": 4.125675391200618e-05,
      "loss": 0.9333,
      "step": 4076
    },
    {
      "epoch": 1.287338174928955,
      "grad_norm": 1.4741299152374268,
      "learning_rate": 4.125324538628869e-05,
      "loss": 0.8897,
      "step": 4077
    },
    {
      "epoch": 1.2876539311651405,
      "grad_norm": 1.5227241516113281,
      "learning_rate": 4.1249736860571194e-05,
      "loss": 0.8253,
      "step": 4078
    },
    {
      "epoch": 1.2879696874013262,
      "grad_norm": 1.4831061363220215,
      "learning_rate": 4.12462283348537e-05,
      "loss": 0.8148,
      "step": 4079
    },
    {
      "epoch": 1.2882854436375117,
      "grad_norm": 1.5665936470031738,
      "learning_rate": 4.12427198091362e-05,
      "loss": 0.8573,
      "step": 4080
    },
    {
      "epoch": 1.2886011998736975,
      "grad_norm": 2.081831693649292,
      "learning_rate": 4.123921128341871e-05,
      "loss": 0.8575,
      "step": 4081
    },
    {
      "epoch": 1.2889169561098832,
      "grad_norm": 1.8105084896087646,
      "learning_rate": 4.1235702757701215e-05,
      "loss": 0.9369,
      "step": 4082
    },
    {
      "epoch": 1.289232712346069,
      "grad_norm": 1.7152694463729858,
      "learning_rate": 4.123219423198372e-05,
      "loss": 0.8393,
      "step": 4083
    },
    {
      "epoch": 1.2895484685822545,
      "grad_norm": 1.7008373737335205,
      "learning_rate": 4.122868570626623e-05,
      "loss": 0.8687,
      "step": 4084
    },
    {
      "epoch": 1.2898642248184402,
      "grad_norm": 1.5407357215881348,
      "learning_rate": 4.1225177180548735e-05,
      "loss": 0.887,
      "step": 4085
    },
    {
      "epoch": 1.2901799810546257,
      "grad_norm": 1.7591028213500977,
      "learning_rate": 4.122166865483124e-05,
      "loss": 0.8216,
      "step": 4086
    },
    {
      "epoch": 1.2904957372908115,
      "grad_norm": 1.5425440073013306,
      "learning_rate": 4.121816012911375e-05,
      "loss": 0.8752,
      "step": 4087
    },
    {
      "epoch": 1.2908114935269972,
      "grad_norm": 1.5508562326431274,
      "learning_rate": 4.1214651603396256e-05,
      "loss": 0.8906,
      "step": 4088
    },
    {
      "epoch": 1.291127249763183,
      "grad_norm": 1.7251255512237549,
      "learning_rate": 4.121114307767876e-05,
      "loss": 0.8326,
      "step": 4089
    },
    {
      "epoch": 1.2914430059993685,
      "grad_norm": 2.299179792404175,
      "learning_rate": 4.120763455196127e-05,
      "loss": 0.889,
      "step": 4090
    },
    {
      "epoch": 1.2917587622355542,
      "grad_norm": 1.4324544668197632,
      "learning_rate": 4.120412602624377e-05,
      "loss": 0.7992,
      "step": 4091
    },
    {
      "epoch": 1.2920745184717397,
      "grad_norm": 1.4392399787902832,
      "learning_rate": 4.1200617500526276e-05,
      "loss": 0.8613,
      "step": 4092
    },
    {
      "epoch": 1.2923902747079254,
      "grad_norm": 1.7793991565704346,
      "learning_rate": 4.119710897480878e-05,
      "loss": 0.7784,
      "step": 4093
    },
    {
      "epoch": 1.2927060309441112,
      "grad_norm": 1.4173027276992798,
      "learning_rate": 4.11936004490913e-05,
      "loss": 0.8337,
      "step": 4094
    },
    {
      "epoch": 1.293021787180297,
      "grad_norm": 1.451509952545166,
      "learning_rate": 4.1190091923373803e-05,
      "loss": 0.8846,
      "step": 4095
    },
    {
      "epoch": 1.2933375434164824,
      "grad_norm": 1.3861114978790283,
      "learning_rate": 4.118658339765631e-05,
      "loss": 0.8708,
      "step": 4096
    },
    {
      "epoch": 1.2936532996526682,
      "grad_norm": 1.6360199451446533,
      "learning_rate": 4.118307487193882e-05,
      "loss": 0.835,
      "step": 4097
    },
    {
      "epoch": 1.2939690558888537,
      "grad_norm": 1.9233113527297974,
      "learning_rate": 4.1179566346221324e-05,
      "loss": 0.8954,
      "step": 4098
    },
    {
      "epoch": 1.2942848121250394,
      "grad_norm": 1.4117727279663086,
      "learning_rate": 4.1176057820503824e-05,
      "loss": 0.9081,
      "step": 4099
    },
    {
      "epoch": 1.2946005683612252,
      "grad_norm": 1.7493565082550049,
      "learning_rate": 4.117254929478633e-05,
      "loss": 0.8589,
      "step": 4100
    },
    {
      "epoch": 1.294916324597411,
      "grad_norm": 1.526517629623413,
      "learning_rate": 4.116904076906884e-05,
      "loss": 0.8002,
      "step": 4101
    },
    {
      "epoch": 1.2952320808335964,
      "grad_norm": 1.2674895524978638,
      "learning_rate": 4.1165532243351344e-05,
      "loss": 0.8756,
      "step": 4102
    },
    {
      "epoch": 1.2955478370697822,
      "grad_norm": 1.394714117050171,
      "learning_rate": 4.116202371763385e-05,
      "loss": 0.8572,
      "step": 4103
    },
    {
      "epoch": 1.2958635933059677,
      "grad_norm": 1.5540506839752197,
      "learning_rate": 4.115851519191636e-05,
      "loss": 0.8781,
      "step": 4104
    },
    {
      "epoch": 1.2961793495421534,
      "grad_norm": 1.4024643898010254,
      "learning_rate": 4.1155006666198865e-05,
      "loss": 0.8494,
      "step": 4105
    },
    {
      "epoch": 1.2964951057783392,
      "grad_norm": 1.46404230594635,
      "learning_rate": 4.115149814048137e-05,
      "loss": 0.7978,
      "step": 4106
    },
    {
      "epoch": 1.296810862014525,
      "grad_norm": 1.5240685939788818,
      "learning_rate": 4.114798961476388e-05,
      "loss": 0.7927,
      "step": 4107
    },
    {
      "epoch": 1.2971266182507104,
      "grad_norm": 1.4599984884262085,
      "learning_rate": 4.1144481089046385e-05,
      "loss": 0.8092,
      "step": 4108
    },
    {
      "epoch": 1.2974423744868961,
      "grad_norm": 1.4604295492172241,
      "learning_rate": 4.114097256332889e-05,
      "loss": 0.8323,
      "step": 4109
    },
    {
      "epoch": 1.2977581307230817,
      "grad_norm": 1.4150848388671875,
      "learning_rate": 4.113746403761139e-05,
      "loss": 0.8376,
      "step": 4110
    },
    {
      "epoch": 1.2980738869592674,
      "grad_norm": 1.6252601146697998,
      "learning_rate": 4.1133955511893906e-05,
      "loss": 0.8605,
      "step": 4111
    },
    {
      "epoch": 1.2983896431954531,
      "grad_norm": 1.5255917310714722,
      "learning_rate": 4.113044698617641e-05,
      "loss": 0.8497,
      "step": 4112
    },
    {
      "epoch": 1.2987053994316389,
      "grad_norm": 1.3720515966415405,
      "learning_rate": 4.112693846045892e-05,
      "loss": 0.8122,
      "step": 4113
    },
    {
      "epoch": 1.2990211556678244,
      "grad_norm": 1.554726243019104,
      "learning_rate": 4.1123429934741426e-05,
      "loss": 0.8129,
      "step": 4114
    },
    {
      "epoch": 1.2993369119040101,
      "grad_norm": 1.297245740890503,
      "learning_rate": 4.111992140902393e-05,
      "loss": 0.7318,
      "step": 4115
    },
    {
      "epoch": 1.2996526681401956,
      "grad_norm": 1.4918524026870728,
      "learning_rate": 4.111641288330644e-05,
      "loss": 0.9208,
      "step": 4116
    },
    {
      "epoch": 1.2999684243763814,
      "grad_norm": 1.5354169607162476,
      "learning_rate": 4.1112904357588947e-05,
      "loss": 0.8362,
      "step": 4117
    },
    {
      "epoch": 1.3002841806125671,
      "grad_norm": 1.772107481956482,
      "learning_rate": 4.110939583187145e-05,
      "loss": 0.9012,
      "step": 4118
    },
    {
      "epoch": 1.3005999368487529,
      "grad_norm": 1.4610967636108398,
      "learning_rate": 4.1105887306153953e-05,
      "loss": 0.838,
      "step": 4119
    },
    {
      "epoch": 1.3009156930849384,
      "grad_norm": 1.7545949220657349,
      "learning_rate": 4.110237878043646e-05,
      "loss": 0.8757,
      "step": 4120
    },
    {
      "epoch": 1.3012314493211241,
      "grad_norm": 1.4918396472930908,
      "learning_rate": 4.109887025471897e-05,
      "loss": 0.77,
      "step": 4121
    },
    {
      "epoch": 1.3015472055573096,
      "grad_norm": 1.485168695449829,
      "learning_rate": 4.1095361729001474e-05,
      "loss": 0.8695,
      "step": 4122
    },
    {
      "epoch": 1.3018629617934954,
      "grad_norm": 1.6096550226211548,
      "learning_rate": 4.109185320328398e-05,
      "loss": 0.8759,
      "step": 4123
    },
    {
      "epoch": 1.302178718029681,
      "grad_norm": 1.5670280456542969,
      "learning_rate": 4.108834467756649e-05,
      "loss": 0.8762,
      "step": 4124
    },
    {
      "epoch": 1.3024944742658668,
      "grad_norm": 1.6768282651901245,
      "learning_rate": 4.1084836151848994e-05,
      "loss": 0.829,
      "step": 4125
    },
    {
      "epoch": 1.3028102305020524,
      "grad_norm": 1.6085538864135742,
      "learning_rate": 4.10813276261315e-05,
      "loss": 0.8518,
      "step": 4126
    },
    {
      "epoch": 1.303125986738238,
      "grad_norm": 1.4793001413345337,
      "learning_rate": 4.107781910041401e-05,
      "loss": 0.8486,
      "step": 4127
    },
    {
      "epoch": 1.3034417429744236,
      "grad_norm": 1.4142429828643799,
      "learning_rate": 4.1074310574696515e-05,
      "loss": 0.8309,
      "step": 4128
    },
    {
      "epoch": 1.3037574992106093,
      "grad_norm": 1.5241317749023438,
      "learning_rate": 4.107080204897902e-05,
      "loss": 0.826,
      "step": 4129
    },
    {
      "epoch": 1.304073255446795,
      "grad_norm": 1.3584460020065308,
      "learning_rate": 4.106729352326153e-05,
      "loss": 0.8435,
      "step": 4130
    },
    {
      "epoch": 1.3043890116829808,
      "grad_norm": 1.627820372581482,
      "learning_rate": 4.1063784997544035e-05,
      "loss": 0.9063,
      "step": 4131
    },
    {
      "epoch": 1.3047047679191663,
      "grad_norm": 1.5434234142303467,
      "learning_rate": 4.106027647182654e-05,
      "loss": 0.8773,
      "step": 4132
    },
    {
      "epoch": 1.305020524155352,
      "grad_norm": 1.6836438179016113,
      "learning_rate": 4.105676794610905e-05,
      "loss": 0.899,
      "step": 4133
    },
    {
      "epoch": 1.3053362803915376,
      "grad_norm": 1.5206223726272583,
      "learning_rate": 4.1053259420391556e-05,
      "loss": 0.8313,
      "step": 4134
    },
    {
      "epoch": 1.3056520366277233,
      "grad_norm": 1.2415276765823364,
      "learning_rate": 4.104975089467406e-05,
      "loss": 0.7883,
      "step": 4135
    },
    {
      "epoch": 1.305967792863909,
      "grad_norm": 1.4448535442352295,
      "learning_rate": 4.104624236895657e-05,
      "loss": 0.8589,
      "step": 4136
    },
    {
      "epoch": 1.3062835491000948,
      "grad_norm": 1.4567493200302124,
      "learning_rate": 4.1042733843239076e-05,
      "loss": 0.9201,
      "step": 4137
    },
    {
      "epoch": 1.3065993053362803,
      "grad_norm": 1.4603203535079956,
      "learning_rate": 4.1039225317521576e-05,
      "loss": 0.7638,
      "step": 4138
    },
    {
      "epoch": 1.306915061572466,
      "grad_norm": 1.493293046951294,
      "learning_rate": 4.103571679180408e-05,
      "loss": 0.8661,
      "step": 4139
    },
    {
      "epoch": 1.3072308178086518,
      "grad_norm": 1.3248999118804932,
      "learning_rate": 4.103220826608659e-05,
      "loss": 0.8299,
      "step": 4140
    },
    {
      "epoch": 1.3075465740448373,
      "grad_norm": 1.5522819757461548,
      "learning_rate": 4.1028699740369097e-05,
      "loss": 0.89,
      "step": 4141
    },
    {
      "epoch": 1.307862330281023,
      "grad_norm": 1.2779945135116577,
      "learning_rate": 4.10251912146516e-05,
      "loss": 0.8065,
      "step": 4142
    },
    {
      "epoch": 1.3081780865172088,
      "grad_norm": 1.5921071767807007,
      "learning_rate": 4.102168268893411e-05,
      "loss": 0.8178,
      "step": 4143
    },
    {
      "epoch": 1.3084938427533943,
      "grad_norm": 1.3126806020736694,
      "learning_rate": 4.101817416321662e-05,
      "loss": 0.7924,
      "step": 4144
    },
    {
      "epoch": 1.30880959898958,
      "grad_norm": 1.3859214782714844,
      "learning_rate": 4.1014665637499124e-05,
      "loss": 0.8892,
      "step": 4145
    },
    {
      "epoch": 1.3091253552257658,
      "grad_norm": 1.770031213760376,
      "learning_rate": 4.101115711178164e-05,
      "loss": 0.9576,
      "step": 4146
    },
    {
      "epoch": 1.3094411114619513,
      "grad_norm": 1.5196037292480469,
      "learning_rate": 4.100764858606414e-05,
      "loss": 0.8747,
      "step": 4147
    },
    {
      "epoch": 1.309756867698137,
      "grad_norm": 1.3362215757369995,
      "learning_rate": 4.1004140060346644e-05,
      "loss": 0.9028,
      "step": 4148
    },
    {
      "epoch": 1.3100726239343228,
      "grad_norm": 1.6856639385223389,
      "learning_rate": 4.100063153462915e-05,
      "loss": 0.7504,
      "step": 4149
    },
    {
      "epoch": 1.3103883801705083,
      "grad_norm": 1.7020567655563354,
      "learning_rate": 4.099712300891166e-05,
      "loss": 0.8087,
      "step": 4150
    },
    {
      "epoch": 1.310704136406694,
      "grad_norm": 1.7378498315811157,
      "learning_rate": 4.0993614483194165e-05,
      "loss": 0.8586,
      "step": 4151
    },
    {
      "epoch": 1.3110198926428798,
      "grad_norm": 1.4788835048675537,
      "learning_rate": 4.099010595747667e-05,
      "loss": 0.8341,
      "step": 4152
    },
    {
      "epoch": 1.3113356488790653,
      "grad_norm": 1.5165610313415527,
      "learning_rate": 4.098659743175918e-05,
      "loss": 0.8432,
      "step": 4153
    },
    {
      "epoch": 1.311651405115251,
      "grad_norm": 1.5807034969329834,
      "learning_rate": 4.0983088906041685e-05,
      "loss": 0.7819,
      "step": 4154
    },
    {
      "epoch": 1.3119671613514368,
      "grad_norm": 2.020185947418213,
      "learning_rate": 4.097958038032419e-05,
      "loss": 0.9109,
      "step": 4155
    },
    {
      "epoch": 1.3122829175876223,
      "grad_norm": 1.6231521368026733,
      "learning_rate": 4.09760718546067e-05,
      "loss": 0.8455,
      "step": 4156
    },
    {
      "epoch": 1.312598673823808,
      "grad_norm": 1.8280638456344604,
      "learning_rate": 4.09725633288892e-05,
      "loss": 0.8802,
      "step": 4157
    },
    {
      "epoch": 1.3129144300599938,
      "grad_norm": 1.7295873165130615,
      "learning_rate": 4.0969054803171706e-05,
      "loss": 0.8419,
      "step": 4158
    },
    {
      "epoch": 1.3132301862961793,
      "grad_norm": 1.2766257524490356,
      "learning_rate": 4.096554627745421e-05,
      "loss": 0.7596,
      "step": 4159
    },
    {
      "epoch": 1.313545942532365,
      "grad_norm": 1.2651698589324951,
      "learning_rate": 4.096203775173672e-05,
      "loss": 0.8876,
      "step": 4160
    },
    {
      "epoch": 1.3138616987685507,
      "grad_norm": 2.9703335762023926,
      "learning_rate": 4.0958529226019226e-05,
      "loss": 0.7679,
      "step": 4161
    },
    {
      "epoch": 1.3141774550047363,
      "grad_norm": 1.6787188053131104,
      "learning_rate": 4.095502070030173e-05,
      "loss": 0.8873,
      "step": 4162
    },
    {
      "epoch": 1.314493211240922,
      "grad_norm": 1.5189855098724365,
      "learning_rate": 4.095151217458424e-05,
      "loss": 0.9215,
      "step": 4163
    },
    {
      "epoch": 1.3148089674771077,
      "grad_norm": 1.9502607583999634,
      "learning_rate": 4.094800364886675e-05,
      "loss": 0.7843,
      "step": 4164
    },
    {
      "epoch": 1.3151247237132933,
      "grad_norm": 1.5806443691253662,
      "learning_rate": 4.094449512314926e-05,
      "loss": 0.905,
      "step": 4165
    },
    {
      "epoch": 1.315440479949479,
      "grad_norm": 1.353838562965393,
      "learning_rate": 4.094098659743176e-05,
      "loss": 0.8741,
      "step": 4166
    },
    {
      "epoch": 1.3157562361856647,
      "grad_norm": 1.7327853441238403,
      "learning_rate": 4.093747807171427e-05,
      "loss": 0.9522,
      "step": 4167
    },
    {
      "epoch": 1.3160719924218505,
      "grad_norm": 1.461549162864685,
      "learning_rate": 4.0933969545996774e-05,
      "loss": 0.8335,
      "step": 4168
    },
    {
      "epoch": 1.316387748658036,
      "grad_norm": 1.3628993034362793,
      "learning_rate": 4.093046102027928e-05,
      "loss": 0.8329,
      "step": 4169
    },
    {
      "epoch": 1.3167035048942217,
      "grad_norm": 3.1600191593170166,
      "learning_rate": 4.092695249456179e-05,
      "loss": 0.8366,
      "step": 4170
    },
    {
      "epoch": 1.3170192611304072,
      "grad_norm": 1.7100918292999268,
      "learning_rate": 4.0923443968844294e-05,
      "loss": 0.8586,
      "step": 4171
    },
    {
      "epoch": 1.317335017366593,
      "grad_norm": 1.4550524950027466,
      "learning_rate": 4.09199354431268e-05,
      "loss": 0.816,
      "step": 4172
    },
    {
      "epoch": 1.3176507736027787,
      "grad_norm": 2.182666540145874,
      "learning_rate": 4.091642691740931e-05,
      "loss": 0.8208,
      "step": 4173
    },
    {
      "epoch": 1.3179665298389645,
      "grad_norm": 1.3032244443893433,
      "learning_rate": 4.0912918391691815e-05,
      "loss": 0.8281,
      "step": 4174
    },
    {
      "epoch": 1.31828228607515,
      "grad_norm": 1.5784621238708496,
      "learning_rate": 4.090940986597432e-05,
      "loss": 0.8229,
      "step": 4175
    },
    {
      "epoch": 1.3185980423113357,
      "grad_norm": 1.5329428911209106,
      "learning_rate": 4.090590134025682e-05,
      "loss": 0.8197,
      "step": 4176
    },
    {
      "epoch": 1.3189137985475212,
      "grad_norm": 1.3374160528182983,
      "learning_rate": 4.090239281453933e-05,
      "loss": 0.8021,
      "step": 4177
    },
    {
      "epoch": 1.319229554783707,
      "grad_norm": 1.5519901514053345,
      "learning_rate": 4.0898884288821835e-05,
      "loss": 0.7837,
      "step": 4178
    },
    {
      "epoch": 1.3195453110198927,
      "grad_norm": 1.3509488105773926,
      "learning_rate": 4.089537576310434e-05,
      "loss": 0.8491,
      "step": 4179
    },
    {
      "epoch": 1.3198610672560784,
      "grad_norm": 1.359291672706604,
      "learning_rate": 4.089186723738685e-05,
      "loss": 0.7971,
      "step": 4180
    },
    {
      "epoch": 1.320176823492264,
      "grad_norm": 1.3950834274291992,
      "learning_rate": 4.088835871166936e-05,
      "loss": 0.836,
      "step": 4181
    },
    {
      "epoch": 1.3204925797284497,
      "grad_norm": 1.6823608875274658,
      "learning_rate": 4.088485018595187e-05,
      "loss": 0.8717,
      "step": 4182
    },
    {
      "epoch": 1.3208083359646352,
      "grad_norm": 1.6587828397750854,
      "learning_rate": 4.0881341660234376e-05,
      "loss": 0.8382,
      "step": 4183
    },
    {
      "epoch": 1.321124092200821,
      "grad_norm": 1.657652735710144,
      "learning_rate": 4.087783313451688e-05,
      "loss": 0.8283,
      "step": 4184
    },
    {
      "epoch": 1.3214398484370067,
      "grad_norm": 1.5184135437011719,
      "learning_rate": 4.087432460879938e-05,
      "loss": 0.8348,
      "step": 4185
    },
    {
      "epoch": 1.3217556046731924,
      "grad_norm": 1.3703439235687256,
      "learning_rate": 4.087081608308189e-05,
      "loss": 0.7886,
      "step": 4186
    },
    {
      "epoch": 1.322071360909378,
      "grad_norm": 1.4681191444396973,
      "learning_rate": 4.0867307557364396e-05,
      "loss": 0.7888,
      "step": 4187
    },
    {
      "epoch": 1.3223871171455637,
      "grad_norm": 1.451773762702942,
      "learning_rate": 4.08637990316469e-05,
      "loss": 0.92,
      "step": 4188
    },
    {
      "epoch": 1.3227028733817492,
      "grad_norm": 1.7086633443832397,
      "learning_rate": 4.086029050592941e-05,
      "loss": 0.8598,
      "step": 4189
    },
    {
      "epoch": 1.323018629617935,
      "grad_norm": 1.6083409786224365,
      "learning_rate": 4.085678198021192e-05,
      "loss": 0.8599,
      "step": 4190
    },
    {
      "epoch": 1.3233343858541207,
      "grad_norm": 1.8837437629699707,
      "learning_rate": 4.0853273454494424e-05,
      "loss": 0.7658,
      "step": 4191
    },
    {
      "epoch": 1.3236501420903064,
      "grad_norm": 1.3768987655639648,
      "learning_rate": 4.084976492877693e-05,
      "loss": 0.8167,
      "step": 4192
    },
    {
      "epoch": 1.323965898326492,
      "grad_norm": 1.739601731300354,
      "learning_rate": 4.084625640305944e-05,
      "loss": 0.8455,
      "step": 4193
    },
    {
      "epoch": 1.3242816545626777,
      "grad_norm": 1.8997716903686523,
      "learning_rate": 4.0842747877341944e-05,
      "loss": 0.9102,
      "step": 4194
    },
    {
      "epoch": 1.3245974107988632,
      "grad_norm": 1.556579828262329,
      "learning_rate": 4.083923935162445e-05,
      "loss": 0.9184,
      "step": 4195
    },
    {
      "epoch": 1.324913167035049,
      "grad_norm": 1.591252088546753,
      "learning_rate": 4.083573082590695e-05,
      "loss": 0.8739,
      "step": 4196
    },
    {
      "epoch": 1.3252289232712346,
      "grad_norm": 1.4133970737457275,
      "learning_rate": 4.083222230018946e-05,
      "loss": 0.7816,
      "step": 4197
    },
    {
      "epoch": 1.3255446795074204,
      "grad_norm": 1.7922403812408447,
      "learning_rate": 4.0828713774471965e-05,
      "loss": 0.8518,
      "step": 4198
    },
    {
      "epoch": 1.325860435743606,
      "grad_norm": 1.762742280960083,
      "learning_rate": 4.082520524875448e-05,
      "loss": 0.7802,
      "step": 4199
    },
    {
      "epoch": 1.3261761919797916,
      "grad_norm": 2.30027437210083,
      "learning_rate": 4.0821696723036985e-05,
      "loss": 0.8917,
      "step": 4200
    },
    {
      "epoch": 1.3261761919797916,
      "eval_loss": 0.8299435973167419,
      "eval_runtime": 4.635,
      "eval_samples_per_second": 537.215,
      "eval_steps_per_second": 4.531,
      "step": 4200
    },
    {
      "epoch": 1.3264919482159772,
      "grad_norm": 1.4981831312179565,
      "learning_rate": 4.081818819731949e-05,
      "loss": 0.9288,
      "step": 4201
    },
    {
      "epoch": 1.326807704452163,
      "grad_norm": 1.748822808265686,
      "learning_rate": 4.0814679671602e-05,
      "loss": 0.8607,
      "step": 4202
    },
    {
      "epoch": 1.3271234606883486,
      "grad_norm": 1.6646610498428345,
      "learning_rate": 4.0811171145884505e-05,
      "loss": 0.851,
      "step": 4203
    },
    {
      "epoch": 1.3274392169245344,
      "grad_norm": 1.5848709344863892,
      "learning_rate": 4.0807662620167005e-05,
      "loss": 0.8855,
      "step": 4204
    },
    {
      "epoch": 1.3277549731607199,
      "grad_norm": 2.4155590534210205,
      "learning_rate": 4.080415409444951e-05,
      "loss": 0.8707,
      "step": 4205
    },
    {
      "epoch": 1.3280707293969056,
      "grad_norm": 1.3104689121246338,
      "learning_rate": 4.080064556873202e-05,
      "loss": 0.8845,
      "step": 4206
    },
    {
      "epoch": 1.3283864856330911,
      "grad_norm": 1.743078351020813,
      "learning_rate": 4.0797137043014526e-05,
      "loss": 0.8008,
      "step": 4207
    },
    {
      "epoch": 1.3287022418692769,
      "grad_norm": 1.528314232826233,
      "learning_rate": 4.079362851729703e-05,
      "loss": 0.8618,
      "step": 4208
    },
    {
      "epoch": 1.3290179981054626,
      "grad_norm": 1.3603640794754028,
      "learning_rate": 4.079011999157954e-05,
      "loss": 0.8429,
      "step": 4209
    },
    {
      "epoch": 1.3293337543416484,
      "grad_norm": 1.558775782585144,
      "learning_rate": 4.0786611465862046e-05,
      "loss": 0.8212,
      "step": 4210
    },
    {
      "epoch": 1.3296495105778339,
      "grad_norm": 1.209812879562378,
      "learning_rate": 4.078310294014455e-05,
      "loss": 0.7979,
      "step": 4211
    },
    {
      "epoch": 1.3299652668140196,
      "grad_norm": 1.2599406242370605,
      "learning_rate": 4.077959441442706e-05,
      "loss": 0.8368,
      "step": 4212
    },
    {
      "epoch": 1.3302810230502051,
      "grad_norm": 1.328417181968689,
      "learning_rate": 4.077608588870957e-05,
      "loss": 0.9024,
      "step": 4213
    },
    {
      "epoch": 1.3305967792863909,
      "grad_norm": 1.2994463443756104,
      "learning_rate": 4.0772577362992074e-05,
      "loss": 0.7881,
      "step": 4214
    },
    {
      "epoch": 1.3309125355225766,
      "grad_norm": 1.4884603023529053,
      "learning_rate": 4.0769068837274574e-05,
      "loss": 0.8692,
      "step": 4215
    },
    {
      "epoch": 1.3312282917587623,
      "grad_norm": 1.500399112701416,
      "learning_rate": 4.076556031155709e-05,
      "loss": 0.8496,
      "step": 4216
    },
    {
      "epoch": 1.3315440479949479,
      "grad_norm": 1.431530475616455,
      "learning_rate": 4.0762051785839594e-05,
      "loss": 0.8174,
      "step": 4217
    },
    {
      "epoch": 1.3318598042311336,
      "grad_norm": 1.371669888496399,
      "learning_rate": 4.07585432601221e-05,
      "loss": 0.9499,
      "step": 4218
    },
    {
      "epoch": 1.332175560467319,
      "grad_norm": 1.3866008520126343,
      "learning_rate": 4.075503473440461e-05,
      "loss": 0.8283,
      "step": 4219
    },
    {
      "epoch": 1.3324913167035048,
      "grad_norm": 1.3367058038711548,
      "learning_rate": 4.0751526208687114e-05,
      "loss": 0.8215,
      "step": 4220
    },
    {
      "epoch": 1.3328070729396906,
      "grad_norm": 1.2517547607421875,
      "learning_rate": 4.074801768296962e-05,
      "loss": 0.8679,
      "step": 4221
    },
    {
      "epoch": 1.3331228291758763,
      "grad_norm": 1.4412049055099487,
      "learning_rate": 4.074450915725213e-05,
      "loss": 0.7603,
      "step": 4222
    },
    {
      "epoch": 1.3334385854120618,
      "grad_norm": 1.4333962202072144,
      "learning_rate": 4.0741000631534635e-05,
      "loss": 0.8289,
      "step": 4223
    },
    {
      "epoch": 1.3337543416482476,
      "grad_norm": 1.2082401514053345,
      "learning_rate": 4.0737492105817135e-05,
      "loss": 0.8093,
      "step": 4224
    },
    {
      "epoch": 1.334070097884433,
      "grad_norm": 1.596693754196167,
      "learning_rate": 4.073398358009964e-05,
      "loss": 0.8793,
      "step": 4225
    },
    {
      "epoch": 1.3343858541206188,
      "grad_norm": 1.3153425455093384,
      "learning_rate": 4.073047505438215e-05,
      "loss": 0.791,
      "step": 4226
    },
    {
      "epoch": 1.3347016103568046,
      "grad_norm": 1.3912310600280762,
      "learning_rate": 4.0726966528664655e-05,
      "loss": 0.8605,
      "step": 4227
    },
    {
      "epoch": 1.3350173665929903,
      "grad_norm": 1.443097472190857,
      "learning_rate": 4.072345800294716e-05,
      "loss": 0.7786,
      "step": 4228
    },
    {
      "epoch": 1.3353331228291758,
      "grad_norm": 1.4670567512512207,
      "learning_rate": 4.071994947722967e-05,
      "loss": 0.8465,
      "step": 4229
    },
    {
      "epoch": 1.3356488790653616,
      "grad_norm": 1.3289449214935303,
      "learning_rate": 4.0716440951512176e-05,
      "loss": 0.862,
      "step": 4230
    },
    {
      "epoch": 1.335964635301547,
      "grad_norm": 1.409098505973816,
      "learning_rate": 4.071293242579468e-05,
      "loss": 0.881,
      "step": 4231
    },
    {
      "epoch": 1.3362803915377328,
      "grad_norm": 1.549578309059143,
      "learning_rate": 4.070942390007719e-05,
      "loss": 0.9069,
      "step": 4232
    },
    {
      "epoch": 1.3365961477739186,
      "grad_norm": 1.4080777168273926,
      "learning_rate": 4.0705915374359696e-05,
      "loss": 0.8789,
      "step": 4233
    },
    {
      "epoch": 1.3369119040101043,
      "grad_norm": 1.334276556968689,
      "learning_rate": 4.07024068486422e-05,
      "loss": 0.8549,
      "step": 4234
    },
    {
      "epoch": 1.3372276602462898,
      "grad_norm": 1.640486240386963,
      "learning_rate": 4.069889832292471e-05,
      "loss": 0.8772,
      "step": 4235
    },
    {
      "epoch": 1.3375434164824755,
      "grad_norm": 1.3477896451950073,
      "learning_rate": 4.069538979720722e-05,
      "loss": 0.7786,
      "step": 4236
    },
    {
      "epoch": 1.337859172718661,
      "grad_norm": 1.380759835243225,
      "learning_rate": 4.0691881271489723e-05,
      "loss": 0.8266,
      "step": 4237
    },
    {
      "epoch": 1.3381749289548468,
      "grad_norm": 1.4605956077575684,
      "learning_rate": 4.068837274577223e-05,
      "loss": 0.8436,
      "step": 4238
    },
    {
      "epoch": 1.3384906851910325,
      "grad_norm": 1.5546923875808716,
      "learning_rate": 4.068486422005474e-05,
      "loss": 0.7707,
      "step": 4239
    },
    {
      "epoch": 1.3388064414272183,
      "grad_norm": 1.40436589717865,
      "learning_rate": 4.0681355694337244e-05,
      "loss": 0.7581,
      "step": 4240
    },
    {
      "epoch": 1.3391221976634038,
      "grad_norm": 1.630941390991211,
      "learning_rate": 4.067784716861975e-05,
      "loss": 0.8511,
      "step": 4241
    },
    {
      "epoch": 1.3394379538995895,
      "grad_norm": 1.8147104978561401,
      "learning_rate": 4.067433864290226e-05,
      "loss": 0.8858,
      "step": 4242
    },
    {
      "epoch": 1.339753710135775,
      "grad_norm": 1.6649870872497559,
      "learning_rate": 4.067083011718476e-05,
      "loss": 0.8081,
      "step": 4243
    },
    {
      "epoch": 1.3400694663719608,
      "grad_norm": 1.6920303106307983,
      "learning_rate": 4.0667321591467264e-05,
      "loss": 0.8079,
      "step": 4244
    },
    {
      "epoch": 1.3403852226081465,
      "grad_norm": 1.582308292388916,
      "learning_rate": 4.066381306574977e-05,
      "loss": 0.8104,
      "step": 4245
    },
    {
      "epoch": 1.3407009788443323,
      "grad_norm": 1.971217393875122,
      "learning_rate": 4.066030454003228e-05,
      "loss": 0.8883,
      "step": 4246
    },
    {
      "epoch": 1.3410167350805178,
      "grad_norm": 1.5682759284973145,
      "learning_rate": 4.0656796014314785e-05,
      "loss": 0.8255,
      "step": 4247
    },
    {
      "epoch": 1.3413324913167035,
      "grad_norm": 1.5358647108078003,
      "learning_rate": 4.065328748859729e-05,
      "loss": 0.847,
      "step": 4248
    },
    {
      "epoch": 1.3416482475528893,
      "grad_norm": 1.87563157081604,
      "learning_rate": 4.06497789628798e-05,
      "loss": 0.8344,
      "step": 4249
    },
    {
      "epoch": 1.3419640037890748,
      "grad_norm": 1.8260561227798462,
      "learning_rate": 4.0646270437162305e-05,
      "loss": 0.9115,
      "step": 4250
    },
    {
      "epoch": 1.3422797600252605,
      "grad_norm": 1.8222332000732422,
      "learning_rate": 4.064276191144482e-05,
      "loss": 0.9078,
      "step": 4251
    },
    {
      "epoch": 1.3425955162614462,
      "grad_norm": 1.565817952156067,
      "learning_rate": 4.063925338572732e-05,
      "loss": 0.9317,
      "step": 4252
    },
    {
      "epoch": 1.3429112724976318,
      "grad_norm": 1.4876292943954468,
      "learning_rate": 4.0635744860009826e-05,
      "loss": 0.8581,
      "step": 4253
    },
    {
      "epoch": 1.3432270287338175,
      "grad_norm": 1.7025344371795654,
      "learning_rate": 4.063223633429233e-05,
      "loss": 0.814,
      "step": 4254
    },
    {
      "epoch": 1.3435427849700032,
      "grad_norm": 1.6009862422943115,
      "learning_rate": 4.062872780857484e-05,
      "loss": 0.8486,
      "step": 4255
    },
    {
      "epoch": 1.3438585412061888,
      "grad_norm": 1.5270143747329712,
      "learning_rate": 4.0625219282857346e-05,
      "loss": 0.8889,
      "step": 4256
    },
    {
      "epoch": 1.3441742974423745,
      "grad_norm": 1.7627493143081665,
      "learning_rate": 4.062171075713985e-05,
      "loss": 0.8205,
      "step": 4257
    },
    {
      "epoch": 1.3444900536785602,
      "grad_norm": 1.4805049896240234,
      "learning_rate": 4.061820223142236e-05,
      "loss": 0.945,
      "step": 4258
    },
    {
      "epoch": 1.3448058099147457,
      "grad_norm": 1.4059739112854004,
      "learning_rate": 4.0614693705704867e-05,
      "loss": 0.8689,
      "step": 4259
    },
    {
      "epoch": 1.3451215661509315,
      "grad_norm": 1.8653526306152344,
      "learning_rate": 4.061118517998737e-05,
      "loss": 0.867,
      "step": 4260
    },
    {
      "epoch": 1.3454373223871172,
      "grad_norm": 1.5978461503982544,
      "learning_rate": 4.060767665426988e-05,
      "loss": 0.9291,
      "step": 4261
    },
    {
      "epoch": 1.3457530786233027,
      "grad_norm": 1.5291635990142822,
      "learning_rate": 4.060416812855238e-05,
      "loss": 0.8843,
      "step": 4262
    },
    {
      "epoch": 1.3460688348594885,
      "grad_norm": 1.4896084070205688,
      "learning_rate": 4.060065960283489e-05,
      "loss": 0.8825,
      "step": 4263
    },
    {
      "epoch": 1.3463845910956742,
      "grad_norm": 1.41112220287323,
      "learning_rate": 4.0597151077117394e-05,
      "loss": 0.8312,
      "step": 4264
    },
    {
      "epoch": 1.3467003473318597,
      "grad_norm": 1.6176544427871704,
      "learning_rate": 4.05936425513999e-05,
      "loss": 0.8748,
      "step": 4265
    },
    {
      "epoch": 1.3470161035680455,
      "grad_norm": 1.2759147882461548,
      "learning_rate": 4.059013402568241e-05,
      "loss": 0.8039,
      "step": 4266
    },
    {
      "epoch": 1.3473318598042312,
      "grad_norm": 1.7099531888961792,
      "learning_rate": 4.0586625499964914e-05,
      "loss": 0.8483,
      "step": 4267
    },
    {
      "epoch": 1.3476476160404167,
      "grad_norm": 1.5534803867340088,
      "learning_rate": 4.058311697424742e-05,
      "loss": 0.7985,
      "step": 4268
    },
    {
      "epoch": 1.3479633722766025,
      "grad_norm": 1.3923829793930054,
      "learning_rate": 4.0579608448529935e-05,
      "loss": 0.8383,
      "step": 4269
    },
    {
      "epoch": 1.3482791285127882,
      "grad_norm": 1.3401122093200684,
      "learning_rate": 4.057609992281244e-05,
      "loss": 0.8101,
      "step": 4270
    },
    {
      "epoch": 1.3485948847489737,
      "grad_norm": 1.6832118034362793,
      "learning_rate": 4.057259139709494e-05,
      "loss": 0.8523,
      "step": 4271
    },
    {
      "epoch": 1.3489106409851594,
      "grad_norm": 1.577630639076233,
      "learning_rate": 4.056908287137745e-05,
      "loss": 0.838,
      "step": 4272
    },
    {
      "epoch": 1.3492263972213452,
      "grad_norm": 1.298061728477478,
      "learning_rate": 4.0565574345659955e-05,
      "loss": 0.7767,
      "step": 4273
    },
    {
      "epoch": 1.3495421534575307,
      "grad_norm": 1.341559648513794,
      "learning_rate": 4.056206581994246e-05,
      "loss": 0.8326,
      "step": 4274
    },
    {
      "epoch": 1.3498579096937164,
      "grad_norm": 1.532920241355896,
      "learning_rate": 4.055855729422497e-05,
      "loss": 0.873,
      "step": 4275
    },
    {
      "epoch": 1.3501736659299022,
      "grad_norm": 1.3661079406738281,
      "learning_rate": 4.0555048768507476e-05,
      "loss": 0.8436,
      "step": 4276
    },
    {
      "epoch": 1.350489422166088,
      "grad_norm": 1.4375253915786743,
      "learning_rate": 4.055154024278998e-05,
      "loss": 0.8764,
      "step": 4277
    },
    {
      "epoch": 1.3508051784022734,
      "grad_norm": 1.6836926937103271,
      "learning_rate": 4.054803171707249e-05,
      "loss": 0.8086,
      "step": 4278
    },
    {
      "epoch": 1.3511209346384592,
      "grad_norm": 1.498545527458191,
      "learning_rate": 4.0544523191354996e-05,
      "loss": 0.8098,
      "step": 4279
    },
    {
      "epoch": 1.3514366908746447,
      "grad_norm": 1.4759293794631958,
      "learning_rate": 4.05410146656375e-05,
      "loss": 0.7581,
      "step": 4280
    },
    {
      "epoch": 1.3517524471108304,
      "grad_norm": 1.5008151531219482,
      "learning_rate": 4.053750613992e-05,
      "loss": 0.7597,
      "step": 4281
    },
    {
      "epoch": 1.3520682033470162,
      "grad_norm": 1.4683606624603271,
      "learning_rate": 4.053399761420251e-05,
      "loss": 0.8309,
      "step": 4282
    },
    {
      "epoch": 1.352383959583202,
      "grad_norm": 1.5259125232696533,
      "learning_rate": 4.0530489088485017e-05,
      "loss": 0.8453,
      "step": 4283
    },
    {
      "epoch": 1.3526997158193874,
      "grad_norm": 1.5164010524749756,
      "learning_rate": 4.052698056276752e-05,
      "loss": 0.7607,
      "step": 4284
    },
    {
      "epoch": 1.3530154720555732,
      "grad_norm": 1.4902886152267456,
      "learning_rate": 4.052347203705003e-05,
      "loss": 0.8302,
      "step": 4285
    },
    {
      "epoch": 1.3533312282917587,
      "grad_norm": 1.6131035089492798,
      "learning_rate": 4.0519963511332544e-05,
      "loss": 0.8111,
      "step": 4286
    },
    {
      "epoch": 1.3536469845279444,
      "grad_norm": 1.5102654695510864,
      "learning_rate": 4.051645498561505e-05,
      "loss": 0.87,
      "step": 4287
    },
    {
      "epoch": 1.3539627407641301,
      "grad_norm": 1.7544292211532593,
      "learning_rate": 4.051294645989756e-05,
      "loss": 0.9156,
      "step": 4288
    },
    {
      "epoch": 1.3542784970003159,
      "grad_norm": 1.5801669359207153,
      "learning_rate": 4.0509437934180064e-05,
      "loss": 0.8176,
      "step": 4289
    },
    {
      "epoch": 1.3545942532365014,
      "grad_norm": 1.4845494031906128,
      "learning_rate": 4.0505929408462564e-05,
      "loss": 0.8553,
      "step": 4290
    },
    {
      "epoch": 1.3549100094726871,
      "grad_norm": 1.5227947235107422,
      "learning_rate": 4.050242088274507e-05,
      "loss": 0.7767,
      "step": 4291
    },
    {
      "epoch": 1.3552257657088727,
      "grad_norm": 1.3975945711135864,
      "learning_rate": 4.049891235702758e-05,
      "loss": 0.8806,
      "step": 4292
    },
    {
      "epoch": 1.3555415219450584,
      "grad_norm": 1.4620012044906616,
      "learning_rate": 4.0495403831310085e-05,
      "loss": 0.8265,
      "step": 4293
    },
    {
      "epoch": 1.3558572781812441,
      "grad_norm": 1.8225988149642944,
      "learning_rate": 4.049189530559259e-05,
      "loss": 0.8588,
      "step": 4294
    },
    {
      "epoch": 1.3561730344174299,
      "grad_norm": 1.350689172744751,
      "learning_rate": 4.04883867798751e-05,
      "loss": 0.8176,
      "step": 4295
    },
    {
      "epoch": 1.3564887906536154,
      "grad_norm": 1.5759472846984863,
      "learning_rate": 4.0484878254157605e-05,
      "loss": 0.7617,
      "step": 4296
    },
    {
      "epoch": 1.3568045468898011,
      "grad_norm": 1.6541259288787842,
      "learning_rate": 4.048136972844011e-05,
      "loss": 0.8523,
      "step": 4297
    },
    {
      "epoch": 1.3571203031259866,
      "grad_norm": 1.4184266328811646,
      "learning_rate": 4.047786120272262e-05,
      "loss": 0.7188,
      "step": 4298
    },
    {
      "epoch": 1.3574360593621724,
      "grad_norm": 1.7785192728042603,
      "learning_rate": 4.0474352677005126e-05,
      "loss": 0.8068,
      "step": 4299
    },
    {
      "epoch": 1.3577518155983581,
      "grad_norm": 1.6134536266326904,
      "learning_rate": 4.047084415128763e-05,
      "loss": 0.9302,
      "step": 4300
    },
    {
      "epoch": 1.3580675718345439,
      "grad_norm": 1.524966835975647,
      "learning_rate": 4.046733562557013e-05,
      "loss": 0.8647,
      "step": 4301
    },
    {
      "epoch": 1.3583833280707294,
      "grad_norm": 1.5535908937454224,
      "learning_rate": 4.046382709985264e-05,
      "loss": 0.817,
      "step": 4302
    },
    {
      "epoch": 1.358699084306915,
      "grad_norm": 1.6743041276931763,
      "learning_rate": 4.0460318574135146e-05,
      "loss": 0.8125,
      "step": 4303
    },
    {
      "epoch": 1.3590148405431006,
      "grad_norm": 1.663089394569397,
      "learning_rate": 4.045681004841766e-05,
      "loss": 0.8489,
      "step": 4304
    },
    {
      "epoch": 1.3593305967792864,
      "grad_norm": 2.000784158706665,
      "learning_rate": 4.0453301522700166e-05,
      "loss": 0.8355,
      "step": 4305
    },
    {
      "epoch": 1.359646353015472,
      "grad_norm": 1.6361922025680542,
      "learning_rate": 4.044979299698267e-05,
      "loss": 0.8758,
      "step": 4306
    },
    {
      "epoch": 1.3599621092516578,
      "grad_norm": 2.1130285263061523,
      "learning_rate": 4.044628447126518e-05,
      "loss": 0.8181,
      "step": 4307
    },
    {
      "epoch": 1.3602778654878434,
      "grad_norm": 1.3804148435592651,
      "learning_rate": 4.044277594554769e-05,
      "loss": 0.7893,
      "step": 4308
    },
    {
      "epoch": 1.360593621724029,
      "grad_norm": 1.4941576719284058,
      "learning_rate": 4.043926741983019e-05,
      "loss": 0.8061,
      "step": 4309
    },
    {
      "epoch": 1.3609093779602146,
      "grad_norm": 1.3997608423233032,
      "learning_rate": 4.0435758894112694e-05,
      "loss": 0.7764,
      "step": 4310
    },
    {
      "epoch": 1.3612251341964003,
      "grad_norm": 1.6916966438293457,
      "learning_rate": 4.04322503683952e-05,
      "loss": 0.8877,
      "step": 4311
    },
    {
      "epoch": 1.361540890432586,
      "grad_norm": 1.8327308893203735,
      "learning_rate": 4.042874184267771e-05,
      "loss": 0.8179,
      "step": 4312
    },
    {
      "epoch": 1.3618566466687718,
      "grad_norm": 1.6867951154708862,
      "learning_rate": 4.0425233316960214e-05,
      "loss": 0.8731,
      "step": 4313
    },
    {
      "epoch": 1.3621724029049573,
      "grad_norm": 1.986506700515747,
      "learning_rate": 4.042172479124272e-05,
      "loss": 0.8409,
      "step": 4314
    },
    {
      "epoch": 1.362488159141143,
      "grad_norm": 1.6442300081253052,
      "learning_rate": 4.041821626552523e-05,
      "loss": 0.7648,
      "step": 4315
    },
    {
      "epoch": 1.3628039153773286,
      "grad_norm": 1.533205270767212,
      "learning_rate": 4.0414707739807735e-05,
      "loss": 0.8411,
      "step": 4316
    },
    {
      "epoch": 1.3631196716135143,
      "grad_norm": 2.44989013671875,
      "learning_rate": 4.041119921409024e-05,
      "loss": 0.8462,
      "step": 4317
    },
    {
      "epoch": 1.3634354278497,
      "grad_norm": 1.6178470849990845,
      "learning_rate": 4.040769068837275e-05,
      "loss": 0.7945,
      "step": 4318
    },
    {
      "epoch": 1.3637511840858858,
      "grad_norm": 1.993066430091858,
      "learning_rate": 4.0404182162655255e-05,
      "loss": 0.8712,
      "step": 4319
    },
    {
      "epoch": 1.3640669403220713,
      "grad_norm": 1.3755475282669067,
      "learning_rate": 4.0400673636937755e-05,
      "loss": 0.7969,
      "step": 4320
    },
    {
      "epoch": 1.364382696558257,
      "grad_norm": 1.5837256908416748,
      "learning_rate": 4.039716511122027e-05,
      "loss": 0.7989,
      "step": 4321
    },
    {
      "epoch": 1.3646984527944426,
      "grad_norm": 1.7121816873550415,
      "learning_rate": 4.0393656585502775e-05,
      "loss": 0.8221,
      "step": 4322
    },
    {
      "epoch": 1.3650142090306283,
      "grad_norm": 1.5173497200012207,
      "learning_rate": 4.039014805978528e-05,
      "loss": 0.8503,
      "step": 4323
    },
    {
      "epoch": 1.365329965266814,
      "grad_norm": 1.81448233127594,
      "learning_rate": 4.038663953406779e-05,
      "loss": 0.8811,
      "step": 4324
    },
    {
      "epoch": 1.3656457215029998,
      "grad_norm": 1.5625007152557373,
      "learning_rate": 4.0383131008350296e-05,
      "loss": 0.8278,
      "step": 4325
    },
    {
      "epoch": 1.3659614777391853,
      "grad_norm": 1.5580092668533325,
      "learning_rate": 4.03796224826328e-05,
      "loss": 0.816,
      "step": 4326
    },
    {
      "epoch": 1.366277233975371,
      "grad_norm": 1.4841119050979614,
      "learning_rate": 4.037611395691531e-05,
      "loss": 0.8849,
      "step": 4327
    },
    {
      "epoch": 1.3665929902115566,
      "grad_norm": 1.52042818069458,
      "learning_rate": 4.0372605431197816e-05,
      "loss": 0.8107,
      "step": 4328
    },
    {
      "epoch": 1.3669087464477423,
      "grad_norm": 1.4520478248596191,
      "learning_rate": 4.0369096905480316e-05,
      "loss": 0.8368,
      "step": 4329
    },
    {
      "epoch": 1.367224502683928,
      "grad_norm": 1.4769699573516846,
      "learning_rate": 4.036558837976282e-05,
      "loss": 0.7965,
      "step": 4330
    },
    {
      "epoch": 1.3675402589201138,
      "grad_norm": 1.756090760231018,
      "learning_rate": 4.036207985404533e-05,
      "loss": 0.8193,
      "step": 4331
    },
    {
      "epoch": 1.3678560151562993,
      "grad_norm": 1.5212411880493164,
      "learning_rate": 4.035857132832784e-05,
      "loss": 0.8052,
      "step": 4332
    },
    {
      "epoch": 1.368171771392485,
      "grad_norm": 1.5918939113616943,
      "learning_rate": 4.0355062802610344e-05,
      "loss": 0.8194,
      "step": 4333
    },
    {
      "epoch": 1.3684875276286705,
      "grad_norm": 1.5922393798828125,
      "learning_rate": 4.035155427689285e-05,
      "loss": 0.8483,
      "step": 4334
    },
    {
      "epoch": 1.3688032838648563,
      "grad_norm": 1.4117118120193481,
      "learning_rate": 4.034804575117536e-05,
      "loss": 0.8441,
      "step": 4335
    },
    {
      "epoch": 1.369119040101042,
      "grad_norm": 1.3126492500305176,
      "learning_rate": 4.0344537225457864e-05,
      "loss": 0.8371,
      "step": 4336
    },
    {
      "epoch": 1.3694347963372278,
      "grad_norm": 1.6128658056259155,
      "learning_rate": 4.034102869974037e-05,
      "loss": 0.8057,
      "step": 4337
    },
    {
      "epoch": 1.3697505525734133,
      "grad_norm": 1.5638086795806885,
      "learning_rate": 4.033752017402288e-05,
      "loss": 0.8874,
      "step": 4338
    },
    {
      "epoch": 1.370066308809599,
      "grad_norm": 1.359797477722168,
      "learning_rate": 4.0334011648305384e-05,
      "loss": 0.8352,
      "step": 4339
    },
    {
      "epoch": 1.3703820650457845,
      "grad_norm": 1.381535530090332,
      "learning_rate": 4.033050312258789e-05,
      "loss": 0.819,
      "step": 4340
    },
    {
      "epoch": 1.3706978212819703,
      "grad_norm": 1.3254715204238892,
      "learning_rate": 4.03269945968704e-05,
      "loss": 0.8961,
      "step": 4341
    },
    {
      "epoch": 1.371013577518156,
      "grad_norm": 1.3220685720443726,
      "learning_rate": 4.0323486071152905e-05,
      "loss": 0.8405,
      "step": 4342
    },
    {
      "epoch": 1.3713293337543417,
      "grad_norm": 1.5892630815505981,
      "learning_rate": 4.031997754543541e-05,
      "loss": 0.9414,
      "step": 4343
    },
    {
      "epoch": 1.3716450899905273,
      "grad_norm": 1.90834379196167,
      "learning_rate": 4.031646901971792e-05,
      "loss": 0.8145,
      "step": 4344
    },
    {
      "epoch": 1.371960846226713,
      "grad_norm": 1.6020350456237793,
      "learning_rate": 4.0312960494000425e-05,
      "loss": 0.7716,
      "step": 4345
    },
    {
      "epoch": 1.3722766024628985,
      "grad_norm": 1.5287340879440308,
      "learning_rate": 4.030945196828293e-05,
      "loss": 0.8504,
      "step": 4346
    },
    {
      "epoch": 1.3725923586990842,
      "grad_norm": 1.589765191078186,
      "learning_rate": 4.030594344256544e-05,
      "loss": 0.8859,
      "step": 4347
    },
    {
      "epoch": 1.37290811493527,
      "grad_norm": 1.386569619178772,
      "learning_rate": 4.030243491684794e-05,
      "loss": 0.8317,
      "step": 4348
    },
    {
      "epoch": 1.3732238711714557,
      "grad_norm": 1.7982304096221924,
      "learning_rate": 4.0298926391130446e-05,
      "loss": 0.8978,
      "step": 4349
    },
    {
      "epoch": 1.3735396274076412,
      "grad_norm": 1.6590758562088013,
      "learning_rate": 4.029541786541295e-05,
      "loss": 0.825,
      "step": 4350
    },
    {
      "epoch": 1.373855383643827,
      "grad_norm": 1.5668187141418457,
      "learning_rate": 4.029190933969546e-05,
      "loss": 0.8534,
      "step": 4351
    },
    {
      "epoch": 1.3741711398800125,
      "grad_norm": 1.2286620140075684,
      "learning_rate": 4.0288400813977966e-05,
      "loss": 0.7853,
      "step": 4352
    },
    {
      "epoch": 1.3744868961161982,
      "grad_norm": 1.7132081985473633,
      "learning_rate": 4.028489228826047e-05,
      "loss": 0.8227,
      "step": 4353
    },
    {
      "epoch": 1.374802652352384,
      "grad_norm": 1.5316709280014038,
      "learning_rate": 4.028138376254298e-05,
      "loss": 0.882,
      "step": 4354
    },
    {
      "epoch": 1.3751184085885697,
      "grad_norm": 1.3627955913543701,
      "learning_rate": 4.027787523682549e-05,
      "loss": 0.856,
      "step": 4355
    },
    {
      "epoch": 1.3754341648247552,
      "grad_norm": 1.5007730722427368,
      "learning_rate": 4.0274366711108e-05,
      "loss": 0.8202,
      "step": 4356
    },
    {
      "epoch": 1.375749921060941,
      "grad_norm": 1.736571192741394,
      "learning_rate": 4.02708581853905e-05,
      "loss": 0.9436,
      "step": 4357
    },
    {
      "epoch": 1.3760656772971267,
      "grad_norm": 1.8944437503814697,
      "learning_rate": 4.026734965967301e-05,
      "loss": 0.8367,
      "step": 4358
    },
    {
      "epoch": 1.3763814335333122,
      "grad_norm": 1.3194036483764648,
      "learning_rate": 4.0263841133955514e-05,
      "loss": 0.8585,
      "step": 4359
    },
    {
      "epoch": 1.376697189769498,
      "grad_norm": 1.5142103433609009,
      "learning_rate": 4.026033260823802e-05,
      "loss": 0.7854,
      "step": 4360
    },
    {
      "epoch": 1.3770129460056837,
      "grad_norm": 1.6060140132904053,
      "learning_rate": 4.025682408252053e-05,
      "loss": 0.7862,
      "step": 4361
    },
    {
      "epoch": 1.3773287022418692,
      "grad_norm": 1.620501160621643,
      "learning_rate": 4.0253315556803034e-05,
      "loss": 0.8552,
      "step": 4362
    },
    {
      "epoch": 1.377644458478055,
      "grad_norm": 1.7089436054229736,
      "learning_rate": 4.024980703108554e-05,
      "loss": 0.8189,
      "step": 4363
    },
    {
      "epoch": 1.3779602147142407,
      "grad_norm": 1.4465779066085815,
      "learning_rate": 4.024629850536805e-05,
      "loss": 0.8834,
      "step": 4364
    },
    {
      "epoch": 1.3782759709504262,
      "grad_norm": 1.299920916557312,
      "learning_rate": 4.0242789979650555e-05,
      "loss": 0.8054,
      "step": 4365
    },
    {
      "epoch": 1.378591727186612,
      "grad_norm": 1.814666986465454,
      "learning_rate": 4.023928145393306e-05,
      "loss": 0.8649,
      "step": 4366
    },
    {
      "epoch": 1.3789074834227977,
      "grad_norm": 1.4750208854675293,
      "learning_rate": 4.023577292821556e-05,
      "loss": 0.8218,
      "step": 4367
    },
    {
      "epoch": 1.3792232396589832,
      "grad_norm": 1.3791730403900146,
      "learning_rate": 4.023226440249807e-05,
      "loss": 0.8687,
      "step": 4368
    },
    {
      "epoch": 1.379538995895169,
      "grad_norm": 1.4300698041915894,
      "learning_rate": 4.0228755876780575e-05,
      "loss": 0.8414,
      "step": 4369
    },
    {
      "epoch": 1.3798547521313547,
      "grad_norm": 1.6019965410232544,
      "learning_rate": 4.022524735106308e-05,
      "loss": 0.9266,
      "step": 4370
    },
    {
      "epoch": 1.3801705083675402,
      "grad_norm": 1.6146955490112305,
      "learning_rate": 4.022173882534559e-05,
      "loss": 0.8819,
      "step": 4371
    },
    {
      "epoch": 1.380486264603726,
      "grad_norm": 1.322348713874817,
      "learning_rate": 4.0218230299628096e-05,
      "loss": 0.9037,
      "step": 4372
    },
    {
      "epoch": 1.3808020208399117,
      "grad_norm": 1.4307148456573486,
      "learning_rate": 4.021472177391061e-05,
      "loss": 0.8023,
      "step": 4373
    },
    {
      "epoch": 1.3811177770760972,
      "grad_norm": 1.4009766578674316,
      "learning_rate": 4.0211213248193116e-05,
      "loss": 0.8532,
      "step": 4374
    },
    {
      "epoch": 1.381433533312283,
      "grad_norm": 1.4506936073303223,
      "learning_rate": 4.020770472247562e-05,
      "loss": 0.8085,
      "step": 4375
    },
    {
      "epoch": 1.3817492895484687,
      "grad_norm": 1.5847225189208984,
      "learning_rate": 4.020419619675812e-05,
      "loss": 0.8863,
      "step": 4376
    },
    {
      "epoch": 1.3820650457846542,
      "grad_norm": 1.4637672901153564,
      "learning_rate": 4.020068767104063e-05,
      "loss": 0.8424,
      "step": 4377
    },
    {
      "epoch": 1.38238080202084,
      "grad_norm": 1.5258926153182983,
      "learning_rate": 4.019717914532314e-05,
      "loss": 0.9135,
      "step": 4378
    },
    {
      "epoch": 1.3826965582570256,
      "grad_norm": 1.3563921451568604,
      "learning_rate": 4.0193670619605643e-05,
      "loss": 0.8223,
      "step": 4379
    },
    {
      "epoch": 1.3830123144932112,
      "grad_norm": 1.4117552042007446,
      "learning_rate": 4.019016209388815e-05,
      "loss": 0.7727,
      "step": 4380
    },
    {
      "epoch": 1.383328070729397,
      "grad_norm": 1.3766813278198242,
      "learning_rate": 4.018665356817066e-05,
      "loss": 0.8146,
      "step": 4381
    },
    {
      "epoch": 1.3836438269655826,
      "grad_norm": 1.5329054594039917,
      "learning_rate": 4.0183145042453164e-05,
      "loss": 0.9157,
      "step": 4382
    },
    {
      "epoch": 1.3839595832017682,
      "grad_norm": 1.7424246072769165,
      "learning_rate": 4.017963651673567e-05,
      "loss": 0.8774,
      "step": 4383
    },
    {
      "epoch": 1.3842753394379539,
      "grad_norm": 1.6022025346755981,
      "learning_rate": 4.017612799101818e-05,
      "loss": 0.8083,
      "step": 4384
    },
    {
      "epoch": 1.3845910956741396,
      "grad_norm": 1.473455786705017,
      "learning_rate": 4.0172619465300684e-05,
      "loss": 0.7552,
      "step": 4385
    },
    {
      "epoch": 1.3849068519103254,
      "grad_norm": 1.5410815477371216,
      "learning_rate": 4.0169110939583184e-05,
      "loss": 0.8486,
      "step": 4386
    },
    {
      "epoch": 1.3852226081465109,
      "grad_norm": 1.5200222730636597,
      "learning_rate": 4.016560241386569e-05,
      "loss": 0.8329,
      "step": 4387
    },
    {
      "epoch": 1.3855383643826966,
      "grad_norm": 1.4593535661697388,
      "learning_rate": 4.01620938881482e-05,
      "loss": 0.8408,
      "step": 4388
    },
    {
      "epoch": 1.3858541206188821,
      "grad_norm": 1.6265389919281006,
      "learning_rate": 4.0158585362430705e-05,
      "loss": 0.8483,
      "step": 4389
    },
    {
      "epoch": 1.3861698768550679,
      "grad_norm": 1.4187952280044556,
      "learning_rate": 4.015507683671321e-05,
      "loss": 0.9,
      "step": 4390
    },
    {
      "epoch": 1.3864856330912536,
      "grad_norm": 1.5724380016326904,
      "learning_rate": 4.0151568310995725e-05,
      "loss": 0.8574,
      "step": 4391
    },
    {
      "epoch": 1.3868013893274393,
      "grad_norm": 1.570624589920044,
      "learning_rate": 4.014805978527823e-05,
      "loss": 0.8814,
      "step": 4392
    },
    {
      "epoch": 1.3871171455636249,
      "grad_norm": 1.71418035030365,
      "learning_rate": 4.014455125956074e-05,
      "loss": 0.9405,
      "step": 4393
    },
    {
      "epoch": 1.3874329017998106,
      "grad_norm": 1.3368144035339355,
      "learning_rate": 4.0141042733843246e-05,
      "loss": 0.8532,
      "step": 4394
    },
    {
      "epoch": 1.3877486580359961,
      "grad_norm": 1.279923439025879,
      "learning_rate": 4.0137534208125746e-05,
      "loss": 0.7944,
      "step": 4395
    },
    {
      "epoch": 1.3880644142721819,
      "grad_norm": 1.610897183418274,
      "learning_rate": 4.013402568240825e-05,
      "loss": 0.8484,
      "step": 4396
    },
    {
      "epoch": 1.3883801705083676,
      "grad_norm": 1.5528974533081055,
      "learning_rate": 4.013051715669076e-05,
      "loss": 0.7954,
      "step": 4397
    },
    {
      "epoch": 1.3886959267445533,
      "grad_norm": 1.6842387914657593,
      "learning_rate": 4.0127008630973266e-05,
      "loss": 0.8107,
      "step": 4398
    },
    {
      "epoch": 1.3890116829807388,
      "grad_norm": 1.4389299154281616,
      "learning_rate": 4.012350010525577e-05,
      "loss": 0.7803,
      "step": 4399
    },
    {
      "epoch": 1.3893274392169246,
      "grad_norm": 1.6746896505355835,
      "learning_rate": 4.011999157953828e-05,
      "loss": 0.823,
      "step": 4400
    },
    {
      "epoch": 1.3893274392169246,
      "eval_loss": 0.8218422532081604,
      "eval_runtime": 4.6395,
      "eval_samples_per_second": 536.698,
      "eval_steps_per_second": 4.526,
      "step": 4400
    },
    {
      "epoch": 1.38964319545311,
      "grad_norm": 1.5333735942840576,
      "learning_rate": 4.0116483053820787e-05,
      "loss": 0.8158,
      "step": 4401
    },
    {
      "epoch": 1.3899589516892958,
      "grad_norm": 2.0612874031066895,
      "learning_rate": 4.011297452810329e-05,
      "loss": 0.7946,
      "step": 4402
    },
    {
      "epoch": 1.3902747079254816,
      "grad_norm": 1.7265790700912476,
      "learning_rate": 4.01094660023858e-05,
      "loss": 0.8894,
      "step": 4403
    },
    {
      "epoch": 1.3905904641616673,
      "grad_norm": 1.5670524835586548,
      "learning_rate": 4.010595747666831e-05,
      "loss": 0.8429,
      "step": 4404
    },
    {
      "epoch": 1.3909062203978528,
      "grad_norm": 1.537812352180481,
      "learning_rate": 4.0102448950950814e-05,
      "loss": 0.8994,
      "step": 4405
    },
    {
      "epoch": 1.3912219766340386,
      "grad_norm": 1.6331697702407837,
      "learning_rate": 4.0098940425233314e-05,
      "loss": 0.8515,
      "step": 4406
    },
    {
      "epoch": 1.391537732870224,
      "grad_norm": 1.5537397861480713,
      "learning_rate": 4.009543189951582e-05,
      "loss": 0.8441,
      "step": 4407
    },
    {
      "epoch": 1.3918534891064098,
      "grad_norm": 1.9069193601608276,
      "learning_rate": 4.0091923373798334e-05,
      "loss": 0.7968,
      "step": 4408
    },
    {
      "epoch": 1.3921692453425956,
      "grad_norm": 1.6036912202835083,
      "learning_rate": 4.008841484808084e-05,
      "loss": 0.8045,
      "step": 4409
    },
    {
      "epoch": 1.3924850015787813,
      "grad_norm": 1.4685795307159424,
      "learning_rate": 4.008490632236335e-05,
      "loss": 0.8176,
      "step": 4410
    },
    {
      "epoch": 1.3928007578149668,
      "grad_norm": 1.6028308868408203,
      "learning_rate": 4.0081397796645855e-05,
      "loss": 0.8406,
      "step": 4411
    },
    {
      "epoch": 1.3931165140511526,
      "grad_norm": 1.7942606210708618,
      "learning_rate": 4.007788927092836e-05,
      "loss": 0.9182,
      "step": 4412
    },
    {
      "epoch": 1.393432270287338,
      "grad_norm": 1.6014580726623535,
      "learning_rate": 4.007438074521087e-05,
      "loss": 0.8258,
      "step": 4413
    },
    {
      "epoch": 1.3937480265235238,
      "grad_norm": 1.5398492813110352,
      "learning_rate": 4.007087221949337e-05,
      "loss": 0.7734,
      "step": 4414
    },
    {
      "epoch": 1.3940637827597095,
      "grad_norm": 1.948588490486145,
      "learning_rate": 4.0067363693775875e-05,
      "loss": 0.7664,
      "step": 4415
    },
    {
      "epoch": 1.3943795389958953,
      "grad_norm": 1.8107151985168457,
      "learning_rate": 4.006385516805838e-05,
      "loss": 0.8275,
      "step": 4416
    },
    {
      "epoch": 1.3946952952320808,
      "grad_norm": 1.4763538837432861,
      "learning_rate": 4.006034664234089e-05,
      "loss": 0.7943,
      "step": 4417
    },
    {
      "epoch": 1.3950110514682665,
      "grad_norm": 1.512760043144226,
      "learning_rate": 4.0056838116623396e-05,
      "loss": 0.8466,
      "step": 4418
    },
    {
      "epoch": 1.395326807704452,
      "grad_norm": 1.648298978805542,
      "learning_rate": 4.00533295909059e-05,
      "loss": 0.8649,
      "step": 4419
    },
    {
      "epoch": 1.3956425639406378,
      "grad_norm": 1.533797264099121,
      "learning_rate": 4.004982106518841e-05,
      "loss": 0.8747,
      "step": 4420
    },
    {
      "epoch": 1.3959583201768235,
      "grad_norm": 1.6989260911941528,
      "learning_rate": 4.0046312539470916e-05,
      "loss": 0.9006,
      "step": 4421
    },
    {
      "epoch": 1.3962740764130093,
      "grad_norm": 1.6647560596466064,
      "learning_rate": 4.004280401375342e-05,
      "loss": 0.8239,
      "step": 4422
    },
    {
      "epoch": 1.3965898326491948,
      "grad_norm": 1.734420657157898,
      "learning_rate": 4.003929548803593e-05,
      "loss": 0.8266,
      "step": 4423
    },
    {
      "epoch": 1.3969055888853805,
      "grad_norm": 1.8874523639678955,
      "learning_rate": 4.0035786962318436e-05,
      "loss": 0.8727,
      "step": 4424
    },
    {
      "epoch": 1.397221345121566,
      "grad_norm": 1.7624012231826782,
      "learning_rate": 4.0032278436600937e-05,
      "loss": 0.877,
      "step": 4425
    },
    {
      "epoch": 1.3975371013577518,
      "grad_norm": 1.5006569623947144,
      "learning_rate": 4.002876991088345e-05,
      "loss": 0.7901,
      "step": 4426
    },
    {
      "epoch": 1.3978528575939375,
      "grad_norm": 1.6261563301086426,
      "learning_rate": 4.002526138516596e-05,
      "loss": 0.9325,
      "step": 4427
    },
    {
      "epoch": 1.3981686138301233,
      "grad_norm": 1.6563310623168945,
      "learning_rate": 4.0021752859448464e-05,
      "loss": 0.7895,
      "step": 4428
    },
    {
      "epoch": 1.3984843700663088,
      "grad_norm": 1.7374765872955322,
      "learning_rate": 4.001824433373097e-05,
      "loss": 0.8241,
      "step": 4429
    },
    {
      "epoch": 1.3988001263024945,
      "grad_norm": 1.7274373769760132,
      "learning_rate": 4.001473580801348e-05,
      "loss": 0.8828,
      "step": 4430
    },
    {
      "epoch": 1.39911588253868,
      "grad_norm": 1.5714603662490845,
      "learning_rate": 4.0011227282295984e-05,
      "loss": 0.8992,
      "step": 4431
    },
    {
      "epoch": 1.3994316387748658,
      "grad_norm": 1.6167879104614258,
      "learning_rate": 4.000771875657849e-05,
      "loss": 0.8015,
      "step": 4432
    },
    {
      "epoch": 1.3997473950110515,
      "grad_norm": 1.7456609010696411,
      "learning_rate": 4.0004210230861e-05,
      "loss": 0.8482,
      "step": 4433
    },
    {
      "epoch": 1.4000631512472372,
      "grad_norm": 1.9982733726501465,
      "learning_rate": 4.00007017051435e-05,
      "loss": 0.8447,
      "step": 4434
    },
    {
      "epoch": 1.4003789074834228,
      "grad_norm": 1.8143211603164673,
      "learning_rate": 3.9997193179426005e-05,
      "loss": 0.8162,
      "step": 4435
    },
    {
      "epoch": 1.4006946637196085,
      "grad_norm": 1.560353398323059,
      "learning_rate": 3.999368465370851e-05,
      "loss": 0.8829,
      "step": 4436
    },
    {
      "epoch": 1.401010419955794,
      "grad_norm": 1.6265509128570557,
      "learning_rate": 3.999017612799102e-05,
      "loss": 0.8302,
      "step": 4437
    },
    {
      "epoch": 1.4013261761919797,
      "grad_norm": 1.4746273756027222,
      "learning_rate": 3.9986667602273525e-05,
      "loss": 0.8603,
      "step": 4438
    },
    {
      "epoch": 1.4016419324281655,
      "grad_norm": 1.5032069683074951,
      "learning_rate": 3.998315907655603e-05,
      "loss": 0.9328,
      "step": 4439
    },
    {
      "epoch": 1.4019576886643512,
      "grad_norm": 2.424142360687256,
      "learning_rate": 3.997965055083854e-05,
      "loss": 0.7358,
      "step": 4440
    },
    {
      "epoch": 1.4022734449005367,
      "grad_norm": 1.674362301826477,
      "learning_rate": 3.9976142025121046e-05,
      "loss": 0.8085,
      "step": 4441
    },
    {
      "epoch": 1.4025892011367225,
      "grad_norm": 1.5993857383728027,
      "learning_rate": 3.997263349940355e-05,
      "loss": 0.7898,
      "step": 4442
    },
    {
      "epoch": 1.402904957372908,
      "grad_norm": 1.6421235799789429,
      "learning_rate": 3.996912497368606e-05,
      "loss": 0.8402,
      "step": 4443
    },
    {
      "epoch": 1.4032207136090937,
      "grad_norm": 1.6601885557174683,
      "learning_rate": 3.9965616447968566e-05,
      "loss": 0.8562,
      "step": 4444
    },
    {
      "epoch": 1.4035364698452795,
      "grad_norm": 1.443986415863037,
      "learning_rate": 3.996210792225107e-05,
      "loss": 0.8272,
      "step": 4445
    },
    {
      "epoch": 1.4038522260814652,
      "grad_norm": 1.6107374429702759,
      "learning_rate": 3.995859939653358e-05,
      "loss": 0.8197,
      "step": 4446
    },
    {
      "epoch": 1.4041679823176507,
      "grad_norm": 1.6746281385421753,
      "learning_rate": 3.9955090870816086e-05,
      "loss": 0.8189,
      "step": 4447
    },
    {
      "epoch": 1.4044837385538365,
      "grad_norm": 1.5008562803268433,
      "learning_rate": 3.995158234509859e-05,
      "loss": 0.8194,
      "step": 4448
    },
    {
      "epoch": 1.404799494790022,
      "grad_norm": 1.4710688591003418,
      "learning_rate": 3.99480738193811e-05,
      "loss": 0.7957,
      "step": 4449
    },
    {
      "epoch": 1.4051152510262077,
      "grad_norm": 1.5606199502944946,
      "learning_rate": 3.994456529366361e-05,
      "loss": 0.9085,
      "step": 4450
    },
    {
      "epoch": 1.4054310072623935,
      "grad_norm": 1.5862246751785278,
      "learning_rate": 3.9941056767946114e-05,
      "loss": 0.8825,
      "step": 4451
    },
    {
      "epoch": 1.4057467634985792,
      "grad_norm": 1.6644660234451294,
      "learning_rate": 3.993754824222862e-05,
      "loss": 0.8086,
      "step": 4452
    },
    {
      "epoch": 1.4060625197347647,
      "grad_norm": 1.581642508506775,
      "learning_rate": 3.993403971651112e-05,
      "loss": 0.8067,
      "step": 4453
    },
    {
      "epoch": 1.4063782759709504,
      "grad_norm": 1.5006800889968872,
      "learning_rate": 3.993053119079363e-05,
      "loss": 0.9636,
      "step": 4454
    },
    {
      "epoch": 1.406694032207136,
      "grad_norm": 1.4333081245422363,
      "learning_rate": 3.9927022665076134e-05,
      "loss": 0.8505,
      "step": 4455
    },
    {
      "epoch": 1.4070097884433217,
      "grad_norm": 1.5359361171722412,
      "learning_rate": 3.992351413935864e-05,
      "loss": 0.8586,
      "step": 4456
    },
    {
      "epoch": 1.4073255446795074,
      "grad_norm": 1.4676713943481445,
      "learning_rate": 3.992000561364115e-05,
      "loss": 0.8362,
      "step": 4457
    },
    {
      "epoch": 1.4076413009156932,
      "grad_norm": 1.349523663520813,
      "learning_rate": 3.9916497087923655e-05,
      "loss": 0.8222,
      "step": 4458
    },
    {
      "epoch": 1.4079570571518787,
      "grad_norm": 1.438546061515808,
      "learning_rate": 3.991298856220616e-05,
      "loss": 0.8534,
      "step": 4459
    },
    {
      "epoch": 1.4082728133880644,
      "grad_norm": 1.4059104919433594,
      "learning_rate": 3.990948003648867e-05,
      "loss": 0.8756,
      "step": 4460
    },
    {
      "epoch": 1.40858856962425,
      "grad_norm": 1.5857878923416138,
      "learning_rate": 3.990597151077118e-05,
      "loss": 0.86,
      "step": 4461
    },
    {
      "epoch": 1.4089043258604357,
      "grad_norm": 1.5693187713623047,
      "learning_rate": 3.990246298505368e-05,
      "loss": 0.7478,
      "step": 4462
    },
    {
      "epoch": 1.4092200820966214,
      "grad_norm": 1.553001880645752,
      "learning_rate": 3.989895445933619e-05,
      "loss": 0.7802,
      "step": 4463
    },
    {
      "epoch": 1.4095358383328072,
      "grad_norm": 1.5786094665527344,
      "learning_rate": 3.9895445933618695e-05,
      "loss": 0.8555,
      "step": 4464
    },
    {
      "epoch": 1.4098515945689927,
      "grad_norm": 1.5482910871505737,
      "learning_rate": 3.98919374079012e-05,
      "loss": 0.7573,
      "step": 4465
    },
    {
      "epoch": 1.4101673508051784,
      "grad_norm": 1.774703025817871,
      "learning_rate": 3.988842888218371e-05,
      "loss": 0.8605,
      "step": 4466
    },
    {
      "epoch": 1.4104831070413641,
      "grad_norm": 1.5766854286193848,
      "learning_rate": 3.9884920356466216e-05,
      "loss": 0.8593,
      "step": 4467
    },
    {
      "epoch": 1.4107988632775497,
      "grad_norm": 1.4903181791305542,
      "learning_rate": 3.988141183074872e-05,
      "loss": 0.8106,
      "step": 4468
    },
    {
      "epoch": 1.4111146195137354,
      "grad_norm": 1.5843738317489624,
      "learning_rate": 3.987790330503123e-05,
      "loss": 0.8278,
      "step": 4469
    },
    {
      "epoch": 1.4114303757499211,
      "grad_norm": 2.326145887374878,
      "learning_rate": 3.9874394779313736e-05,
      "loss": 0.9411,
      "step": 4470
    },
    {
      "epoch": 1.4117461319861067,
      "grad_norm": 1.6174696683883667,
      "learning_rate": 3.987088625359624e-05,
      "loss": 0.8158,
      "step": 4471
    },
    {
      "epoch": 1.4120618882222924,
      "grad_norm": 1.6002799272537231,
      "learning_rate": 3.986737772787874e-05,
      "loss": 0.8873,
      "step": 4472
    },
    {
      "epoch": 1.4123776444584781,
      "grad_norm": 1.5723081827163696,
      "learning_rate": 3.986386920216125e-05,
      "loss": 0.7745,
      "step": 4473
    },
    {
      "epoch": 1.4126934006946636,
      "grad_norm": 1.3516496419906616,
      "learning_rate": 3.986036067644376e-05,
      "loss": 0.755,
      "step": 4474
    },
    {
      "epoch": 1.4130091569308494,
      "grad_norm": 1.5879884958267212,
      "learning_rate": 3.9856852150726264e-05,
      "loss": 0.818,
      "step": 4475
    },
    {
      "epoch": 1.4133249131670351,
      "grad_norm": 1.5955134630203247,
      "learning_rate": 3.985334362500877e-05,
      "loss": 0.8627,
      "step": 4476
    },
    {
      "epoch": 1.4136406694032206,
      "grad_norm": 1.4644721746444702,
      "learning_rate": 3.984983509929128e-05,
      "loss": 0.8552,
      "step": 4477
    },
    {
      "epoch": 1.4139564256394064,
      "grad_norm": 1.6134867668151855,
      "learning_rate": 3.984632657357379e-05,
      "loss": 0.7667,
      "step": 4478
    },
    {
      "epoch": 1.4142721818755921,
      "grad_norm": 1.604172945022583,
      "learning_rate": 3.98428180478563e-05,
      "loss": 0.7605,
      "step": 4479
    },
    {
      "epoch": 1.4145879381117776,
      "grad_norm": 1.8545411825180054,
      "learning_rate": 3.9839309522138804e-05,
      "loss": 0.861,
      "step": 4480
    },
    {
      "epoch": 1.4149036943479634,
      "grad_norm": 1.6852694749832153,
      "learning_rate": 3.9835800996421305e-05,
      "loss": 0.8217,
      "step": 4481
    },
    {
      "epoch": 1.415219450584149,
      "grad_norm": 1.7566553354263306,
      "learning_rate": 3.983229247070381e-05,
      "loss": 0.8372,
      "step": 4482
    },
    {
      "epoch": 1.4155352068203346,
      "grad_norm": 1.6489884853363037,
      "learning_rate": 3.982878394498632e-05,
      "loss": 0.775,
      "step": 4483
    },
    {
      "epoch": 1.4158509630565204,
      "grad_norm": 1.326647162437439,
      "learning_rate": 3.9825275419268825e-05,
      "loss": 0.7535,
      "step": 4484
    },
    {
      "epoch": 1.416166719292706,
      "grad_norm": 1.5317516326904297,
      "learning_rate": 3.982176689355133e-05,
      "loss": 0.8233,
      "step": 4485
    },
    {
      "epoch": 1.4164824755288916,
      "grad_norm": 1.5377919673919678,
      "learning_rate": 3.981825836783384e-05,
      "loss": 0.9125,
      "step": 4486
    },
    {
      "epoch": 1.4167982317650774,
      "grad_norm": 1.5106792449951172,
      "learning_rate": 3.9814749842116345e-05,
      "loss": 0.8733,
      "step": 4487
    },
    {
      "epoch": 1.417113988001263,
      "grad_norm": 1.6283109188079834,
      "learning_rate": 3.981124131639885e-05,
      "loss": 0.7892,
      "step": 4488
    },
    {
      "epoch": 1.4174297442374486,
      "grad_norm": 1.6357195377349854,
      "learning_rate": 3.980773279068136e-05,
      "loss": 0.8636,
      "step": 4489
    },
    {
      "epoch": 1.4177455004736343,
      "grad_norm": 1.6000878810882568,
      "learning_rate": 3.9804224264963866e-05,
      "loss": 0.8313,
      "step": 4490
    },
    {
      "epoch": 1.41806125670982,
      "grad_norm": 1.5470213890075684,
      "learning_rate": 3.9800715739246366e-05,
      "loss": 0.8221,
      "step": 4491
    },
    {
      "epoch": 1.4183770129460056,
      "grad_norm": 1.5627228021621704,
      "learning_rate": 3.979720721352887e-05,
      "loss": 0.9271,
      "step": 4492
    },
    {
      "epoch": 1.4186927691821913,
      "grad_norm": 1.483372688293457,
      "learning_rate": 3.979369868781138e-05,
      "loss": 0.7891,
      "step": 4493
    },
    {
      "epoch": 1.419008525418377,
      "grad_norm": 1.7114973068237305,
      "learning_rate": 3.9790190162093886e-05,
      "loss": 0.8993,
      "step": 4494
    },
    {
      "epoch": 1.4193242816545628,
      "grad_norm": 1.720152735710144,
      "learning_rate": 3.978668163637639e-05,
      "loss": 0.8415,
      "step": 4495
    },
    {
      "epoch": 1.4196400378907483,
      "grad_norm": 1.570217490196228,
      "learning_rate": 3.978317311065891e-05,
      "loss": 0.8481,
      "step": 4496
    },
    {
      "epoch": 1.419955794126934,
      "grad_norm": 1.567944049835205,
      "learning_rate": 3.9779664584941414e-05,
      "loss": 0.8292,
      "step": 4497
    },
    {
      "epoch": 1.4202715503631196,
      "grad_norm": 1.3473296165466309,
      "learning_rate": 3.977615605922392e-05,
      "loss": 0.8214,
      "step": 4498
    },
    {
      "epoch": 1.4205873065993053,
      "grad_norm": 1.5634629726409912,
      "learning_rate": 3.977264753350643e-05,
      "loss": 0.7551,
      "step": 4499
    },
    {
      "epoch": 1.420903062835491,
      "grad_norm": 1.7381242513656616,
      "learning_rate": 3.976913900778893e-05,
      "loss": 0.8914,
      "step": 4500
    },
    {
      "epoch": 1.4212188190716768,
      "grad_norm": 1.6951957941055298,
      "learning_rate": 3.9765630482071434e-05,
      "loss": 0.8664,
      "step": 4501
    },
    {
      "epoch": 1.4215345753078623,
      "grad_norm": 1.5054655075073242,
      "learning_rate": 3.976212195635394e-05,
      "loss": 0.8106,
      "step": 4502
    },
    {
      "epoch": 1.421850331544048,
      "grad_norm": 1.5153274536132812,
      "learning_rate": 3.975861343063645e-05,
      "loss": 0.7375,
      "step": 4503
    },
    {
      "epoch": 1.4221660877802336,
      "grad_norm": 1.7567576169967651,
      "learning_rate": 3.9755104904918954e-05,
      "loss": 0.9642,
      "step": 4504
    },
    {
      "epoch": 1.4224818440164193,
      "grad_norm": 1.5679305791854858,
      "learning_rate": 3.975159637920146e-05,
      "loss": 0.899,
      "step": 4505
    },
    {
      "epoch": 1.422797600252605,
      "grad_norm": 1.5323504209518433,
      "learning_rate": 3.974808785348397e-05,
      "loss": 0.8153,
      "step": 4506
    },
    {
      "epoch": 1.4231133564887908,
      "grad_norm": 1.7465966939926147,
      "learning_rate": 3.9744579327766475e-05,
      "loss": 0.797,
      "step": 4507
    },
    {
      "epoch": 1.4234291127249763,
      "grad_norm": 1.5208324193954468,
      "learning_rate": 3.974107080204898e-05,
      "loss": 0.9096,
      "step": 4508
    },
    {
      "epoch": 1.423744868961162,
      "grad_norm": 1.2950056791305542,
      "learning_rate": 3.973756227633149e-05,
      "loss": 0.8109,
      "step": 4509
    },
    {
      "epoch": 1.4240606251973476,
      "grad_norm": 1.3097139596939087,
      "learning_rate": 3.9734053750613995e-05,
      "loss": 0.8193,
      "step": 4510
    },
    {
      "epoch": 1.4243763814335333,
      "grad_norm": 1.5211713314056396,
      "learning_rate": 3.9730545224896495e-05,
      "loss": 0.8489,
      "step": 4511
    },
    {
      "epoch": 1.424692137669719,
      "grad_norm": 1.4125317335128784,
      "learning_rate": 3.9727036699179e-05,
      "loss": 0.789,
      "step": 4512
    },
    {
      "epoch": 1.4250078939059048,
      "grad_norm": 1.7332508563995361,
      "learning_rate": 3.9723528173461516e-05,
      "loss": 0.8264,
      "step": 4513
    },
    {
      "epoch": 1.4253236501420903,
      "grad_norm": 1.4779351949691772,
      "learning_rate": 3.972001964774402e-05,
      "loss": 0.8124,
      "step": 4514
    },
    {
      "epoch": 1.425639406378276,
      "grad_norm": 1.5072672367095947,
      "learning_rate": 3.971651112202653e-05,
      "loss": 0.8529,
      "step": 4515
    },
    {
      "epoch": 1.4259551626144615,
      "grad_norm": 1.4805810451507568,
      "learning_rate": 3.9713002596309036e-05,
      "loss": 0.8047,
      "step": 4516
    },
    {
      "epoch": 1.4262709188506473,
      "grad_norm": 1.4890683889389038,
      "learning_rate": 3.970949407059154e-05,
      "loss": 0.7631,
      "step": 4517
    },
    {
      "epoch": 1.426586675086833,
      "grad_norm": 1.5550403594970703,
      "learning_rate": 3.970598554487405e-05,
      "loss": 0.8199,
      "step": 4518
    },
    {
      "epoch": 1.4269024313230187,
      "grad_norm": 1.6127912998199463,
      "learning_rate": 3.970247701915655e-05,
      "loss": 0.8228,
      "step": 4519
    },
    {
      "epoch": 1.4272181875592043,
      "grad_norm": 1.6355823278427124,
      "learning_rate": 3.969896849343906e-05,
      "loss": 0.8186,
      "step": 4520
    },
    {
      "epoch": 1.42753394379539,
      "grad_norm": 1.6705576181411743,
      "learning_rate": 3.9695459967721563e-05,
      "loss": 0.9266,
      "step": 4521
    },
    {
      "epoch": 1.4278497000315755,
      "grad_norm": 1.4630117416381836,
      "learning_rate": 3.969195144200407e-05,
      "loss": 0.8612,
      "step": 4522
    },
    {
      "epoch": 1.4281654562677613,
      "grad_norm": 1.5942927598953247,
      "learning_rate": 3.968844291628658e-05,
      "loss": 0.8102,
      "step": 4523
    },
    {
      "epoch": 1.428481212503947,
      "grad_norm": 1.64378821849823,
      "learning_rate": 3.9684934390569084e-05,
      "loss": 0.8642,
      "step": 4524
    },
    {
      "epoch": 1.4287969687401327,
      "grad_norm": 1.4307700395584106,
      "learning_rate": 3.968142586485159e-05,
      "loss": 0.8175,
      "step": 4525
    },
    {
      "epoch": 1.4291127249763182,
      "grad_norm": 1.6677446365356445,
      "learning_rate": 3.96779173391341e-05,
      "loss": 0.8062,
      "step": 4526
    },
    {
      "epoch": 1.429428481212504,
      "grad_norm": 1.5548348426818848,
      "learning_rate": 3.9674408813416604e-05,
      "loss": 0.8604,
      "step": 4527
    },
    {
      "epoch": 1.4297442374486895,
      "grad_norm": 1.43466317653656,
      "learning_rate": 3.967090028769911e-05,
      "loss": 0.8276,
      "step": 4528
    },
    {
      "epoch": 1.4300599936848752,
      "grad_norm": 1.574773907661438,
      "learning_rate": 3.966739176198162e-05,
      "loss": 0.9051,
      "step": 4529
    },
    {
      "epoch": 1.430375749921061,
      "grad_norm": 1.6242785453796387,
      "learning_rate": 3.966388323626412e-05,
      "loss": 0.8459,
      "step": 4530
    },
    {
      "epoch": 1.4306915061572467,
      "grad_norm": 1.4519020318984985,
      "learning_rate": 3.966037471054663e-05,
      "loss": 0.8254,
      "step": 4531
    },
    {
      "epoch": 1.4310072623934322,
      "grad_norm": 1.5629212856292725,
      "learning_rate": 3.965686618482914e-05,
      "loss": 0.8485,
      "step": 4532
    },
    {
      "epoch": 1.431323018629618,
      "grad_norm": 1.5345183610916138,
      "learning_rate": 3.9653357659111645e-05,
      "loss": 0.8898,
      "step": 4533
    },
    {
      "epoch": 1.4316387748658035,
      "grad_norm": 1.5418620109558105,
      "learning_rate": 3.964984913339415e-05,
      "loss": 0.8778,
      "step": 4534
    },
    {
      "epoch": 1.4319545311019892,
      "grad_norm": 1.545526146888733,
      "learning_rate": 3.964634060767666e-05,
      "loss": 0.8416,
      "step": 4535
    },
    {
      "epoch": 1.432270287338175,
      "grad_norm": 1.419021725654602,
      "learning_rate": 3.9642832081959166e-05,
      "loss": 0.8429,
      "step": 4536
    },
    {
      "epoch": 1.4325860435743607,
      "grad_norm": 1.7146384716033936,
      "learning_rate": 3.963932355624167e-05,
      "loss": 0.8332,
      "step": 4537
    },
    {
      "epoch": 1.4329017998105462,
      "grad_norm": 1.4677722454071045,
      "learning_rate": 3.963581503052418e-05,
      "loss": 0.8056,
      "step": 4538
    },
    {
      "epoch": 1.433217556046732,
      "grad_norm": 1.5269403457641602,
      "learning_rate": 3.963230650480668e-05,
      "loss": 0.7259,
      "step": 4539
    },
    {
      "epoch": 1.4335333122829175,
      "grad_norm": 1.7425200939178467,
      "learning_rate": 3.9628797979089186e-05,
      "loss": 0.887,
      "step": 4540
    },
    {
      "epoch": 1.4338490685191032,
      "grad_norm": 1.5912493467330933,
      "learning_rate": 3.962528945337169e-05,
      "loss": 0.7756,
      "step": 4541
    },
    {
      "epoch": 1.434164824755289,
      "grad_norm": 1.7142442464828491,
      "learning_rate": 3.96217809276542e-05,
      "loss": 0.8366,
      "step": 4542
    },
    {
      "epoch": 1.4344805809914747,
      "grad_norm": 1.4664912223815918,
      "learning_rate": 3.9618272401936707e-05,
      "loss": 0.7832,
      "step": 4543
    },
    {
      "epoch": 1.4347963372276602,
      "grad_norm": 1.6268521547317505,
      "learning_rate": 3.9614763876219213e-05,
      "loss": 0.8955,
      "step": 4544
    },
    {
      "epoch": 1.435112093463846,
      "grad_norm": 1.7580329179763794,
      "learning_rate": 3.961125535050172e-05,
      "loss": 0.8259,
      "step": 4545
    },
    {
      "epoch": 1.4354278497000315,
      "grad_norm": 1.864894986152649,
      "learning_rate": 3.960774682478423e-05,
      "loss": 0.7982,
      "step": 4546
    },
    {
      "epoch": 1.4357436059362172,
      "grad_norm": 1.57701575756073,
      "learning_rate": 3.9604238299066734e-05,
      "loss": 0.8643,
      "step": 4547
    },
    {
      "epoch": 1.436059362172403,
      "grad_norm": 1.883619785308838,
      "learning_rate": 3.960072977334924e-05,
      "loss": 0.7765,
      "step": 4548
    },
    {
      "epoch": 1.4363751184085887,
      "grad_norm": 1.6192280054092407,
      "learning_rate": 3.959722124763175e-05,
      "loss": 0.8058,
      "step": 4549
    },
    {
      "epoch": 1.4366908746447742,
      "grad_norm": 1.6511085033416748,
      "learning_rate": 3.9593712721914254e-05,
      "loss": 0.7877,
      "step": 4550
    },
    {
      "epoch": 1.43700663088096,
      "grad_norm": 1.867389440536499,
      "learning_rate": 3.959020419619676e-05,
      "loss": 0.862,
      "step": 4551
    },
    {
      "epoch": 1.4373223871171454,
      "grad_norm": 1.5066741704940796,
      "learning_rate": 3.958669567047927e-05,
      "loss": 0.8163,
      "step": 4552
    },
    {
      "epoch": 1.4376381433533312,
      "grad_norm": 1.9731134176254272,
      "learning_rate": 3.9583187144761775e-05,
      "loss": 0.8201,
      "step": 4553
    },
    {
      "epoch": 1.437953899589517,
      "grad_norm": 1.4177762269973755,
      "learning_rate": 3.957967861904428e-05,
      "loss": 0.8289,
      "step": 4554
    },
    {
      "epoch": 1.4382696558257027,
      "grad_norm": 1.839299201965332,
      "learning_rate": 3.957617009332679e-05,
      "loss": 0.8288,
      "step": 4555
    },
    {
      "epoch": 1.4385854120618882,
      "grad_norm": 1.669199824333191,
      "learning_rate": 3.9572661567609295e-05,
      "loss": 0.799,
      "step": 4556
    },
    {
      "epoch": 1.438901168298074,
      "grad_norm": 1.4671573638916016,
      "learning_rate": 3.95691530418918e-05,
      "loss": 0.8016,
      "step": 4557
    },
    {
      "epoch": 1.4392169245342594,
      "grad_norm": 2.4713852405548096,
      "learning_rate": 3.95656445161743e-05,
      "loss": 0.8443,
      "step": 4558
    },
    {
      "epoch": 1.4395326807704452,
      "grad_norm": 1.80170476436615,
      "learning_rate": 3.956213599045681e-05,
      "loss": 0.858,
      "step": 4559
    },
    {
      "epoch": 1.439848437006631,
      "grad_norm": 1.798817753791809,
      "learning_rate": 3.9558627464739316e-05,
      "loss": 0.8655,
      "step": 4560
    },
    {
      "epoch": 1.4401641932428166,
      "grad_norm": 1.5788763761520386,
      "learning_rate": 3.955511893902182e-05,
      "loss": 0.8584,
      "step": 4561
    },
    {
      "epoch": 1.4404799494790022,
      "grad_norm": 1.4755996465682983,
      "learning_rate": 3.955161041330433e-05,
      "loss": 0.8957,
      "step": 4562
    },
    {
      "epoch": 1.440795705715188,
      "grad_norm": 1.538393259048462,
      "learning_rate": 3.9548101887586836e-05,
      "loss": 0.7092,
      "step": 4563
    },
    {
      "epoch": 1.4411114619513734,
      "grad_norm": 1.6732295751571655,
      "learning_rate": 3.954459336186934e-05,
      "loss": 0.7402,
      "step": 4564
    },
    {
      "epoch": 1.4414272181875591,
      "grad_norm": 1.3982229232788086,
      "learning_rate": 3.954108483615185e-05,
      "loss": 0.833,
      "step": 4565
    },
    {
      "epoch": 1.4417429744237449,
      "grad_norm": 1.8004560470581055,
      "learning_rate": 3.953757631043436e-05,
      "loss": 0.8546,
      "step": 4566
    },
    {
      "epoch": 1.4420587306599306,
      "grad_norm": 1.9105366468429565,
      "learning_rate": 3.953406778471686e-05,
      "loss": 0.8321,
      "step": 4567
    },
    {
      "epoch": 1.4423744868961161,
      "grad_norm": 1.5002622604370117,
      "learning_rate": 3.953055925899937e-05,
      "loss": 0.8562,
      "step": 4568
    },
    {
      "epoch": 1.4426902431323019,
      "grad_norm": 1.482658863067627,
      "learning_rate": 3.952705073328188e-05,
      "loss": 0.8901,
      "step": 4569
    },
    {
      "epoch": 1.4430059993684874,
      "grad_norm": 1.6804170608520508,
      "learning_rate": 3.9523542207564384e-05,
      "loss": 0.864,
      "step": 4570
    },
    {
      "epoch": 1.4433217556046731,
      "grad_norm": 1.639563798904419,
      "learning_rate": 3.952003368184689e-05,
      "loss": 0.8319,
      "step": 4571
    },
    {
      "epoch": 1.4436375118408589,
      "grad_norm": 1.9759910106658936,
      "learning_rate": 3.95165251561294e-05,
      "loss": 0.8656,
      "step": 4572
    },
    {
      "epoch": 1.4439532680770446,
      "grad_norm": 1.5124400854110718,
      "learning_rate": 3.9513016630411904e-05,
      "loss": 0.8058,
      "step": 4573
    },
    {
      "epoch": 1.4442690243132301,
      "grad_norm": 2.1244072914123535,
      "learning_rate": 3.950950810469441e-05,
      "loss": 0.8393,
      "step": 4574
    },
    {
      "epoch": 1.4445847805494159,
      "grad_norm": 1.5381557941436768,
      "learning_rate": 3.950599957897692e-05,
      "loss": 0.8657,
      "step": 4575
    },
    {
      "epoch": 1.4449005367856016,
      "grad_norm": 1.711056113243103,
      "learning_rate": 3.9502491053259425e-05,
      "loss": 0.8591,
      "step": 4576
    },
    {
      "epoch": 1.4452162930217871,
      "grad_norm": 1.8117634057998657,
      "learning_rate": 3.9498982527541925e-05,
      "loss": 0.8578,
      "step": 4577
    },
    {
      "epoch": 1.4455320492579729,
      "grad_norm": 1.576459288597107,
      "learning_rate": 3.949547400182443e-05,
      "loss": 0.8191,
      "step": 4578
    },
    {
      "epoch": 1.4458478054941586,
      "grad_norm": 1.7516590356826782,
      "learning_rate": 3.949196547610694e-05,
      "loss": 0.9205,
      "step": 4579
    },
    {
      "epoch": 1.446163561730344,
      "grad_norm": 1.4096653461456299,
      "learning_rate": 3.9488456950389445e-05,
      "loss": 0.8607,
      "step": 4580
    },
    {
      "epoch": 1.4464793179665298,
      "grad_norm": 1.6379708051681519,
      "learning_rate": 3.948494842467195e-05,
      "loss": 0.8862,
      "step": 4581
    },
    {
      "epoch": 1.4467950742027156,
      "grad_norm": 1.2705265283584595,
      "learning_rate": 3.948143989895446e-05,
      "loss": 0.7893,
      "step": 4582
    },
    {
      "epoch": 1.447110830438901,
      "grad_norm": 1.6467739343643188,
      "learning_rate": 3.947793137323697e-05,
      "loss": 0.8615,
      "step": 4583
    },
    {
      "epoch": 1.4474265866750868,
      "grad_norm": 1.5900639295578003,
      "learning_rate": 3.947442284751948e-05,
      "loss": 0.8415,
      "step": 4584
    },
    {
      "epoch": 1.4477423429112726,
      "grad_norm": 1.7197202444076538,
      "learning_rate": 3.9470914321801986e-05,
      "loss": 0.8811,
      "step": 4585
    },
    {
      "epoch": 1.448058099147458,
      "grad_norm": 1.7301826477050781,
      "learning_rate": 3.9467405796084486e-05,
      "loss": 0.9165,
      "step": 4586
    },
    {
      "epoch": 1.4483738553836438,
      "grad_norm": 1.6005326509475708,
      "learning_rate": 3.946389727036699e-05,
      "loss": 0.8668,
      "step": 4587
    },
    {
      "epoch": 1.4486896116198296,
      "grad_norm": 1.6782310009002686,
      "learning_rate": 3.94603887446495e-05,
      "loss": 0.8593,
      "step": 4588
    },
    {
      "epoch": 1.449005367856015,
      "grad_norm": 1.460198163986206,
      "learning_rate": 3.9456880218932006e-05,
      "loss": 0.8815,
      "step": 4589
    },
    {
      "epoch": 1.4493211240922008,
      "grad_norm": 1.4027708768844604,
      "learning_rate": 3.945337169321451e-05,
      "loss": 0.8677,
      "step": 4590
    },
    {
      "epoch": 1.4496368803283866,
      "grad_norm": 1.679915428161621,
      "learning_rate": 3.944986316749702e-05,
      "loss": 0.7436,
      "step": 4591
    },
    {
      "epoch": 1.449952636564572,
      "grad_norm": 1.6182562112808228,
      "learning_rate": 3.944635464177953e-05,
      "loss": 0.8565,
      "step": 4592
    },
    {
      "epoch": 1.4502683928007578,
      "grad_norm": 1.936491847038269,
      "learning_rate": 3.9442846116062034e-05,
      "loss": 0.8791,
      "step": 4593
    },
    {
      "epoch": 1.4505841490369435,
      "grad_norm": 1.3586446046829224,
      "learning_rate": 3.943933759034454e-05,
      "loss": 0.8419,
      "step": 4594
    },
    {
      "epoch": 1.450899905273129,
      "grad_norm": 1.645775318145752,
      "learning_rate": 3.943582906462705e-05,
      "loss": 0.89,
      "step": 4595
    },
    {
      "epoch": 1.4512156615093148,
      "grad_norm": 1.5189992189407349,
      "learning_rate": 3.943232053890955e-05,
      "loss": 0.8541,
      "step": 4596
    },
    {
      "epoch": 1.4515314177455005,
      "grad_norm": 1.4682941436767578,
      "learning_rate": 3.9428812013192054e-05,
      "loss": 0.7673,
      "step": 4597
    },
    {
      "epoch": 1.451847173981686,
      "grad_norm": 1.5653598308563232,
      "learning_rate": 3.942530348747456e-05,
      "loss": 0.851,
      "step": 4598
    },
    {
      "epoch": 1.4521629302178718,
      "grad_norm": 2.1783013343811035,
      "learning_rate": 3.942179496175707e-05,
      "loss": 0.9017,
      "step": 4599
    },
    {
      "epoch": 1.4524786864540575,
      "grad_norm": 1.6616880893707275,
      "learning_rate": 3.9418286436039575e-05,
      "loss": 0.8166,
      "step": 4600
    },
    {
      "epoch": 1.4524786864540575,
      "eval_loss": 0.8201206922531128,
      "eval_runtime": 4.6414,
      "eval_samples_per_second": 536.472,
      "eval_steps_per_second": 4.524,
      "step": 4600
    },
    {
      "epoch": 1.452794442690243,
      "grad_norm": 1.5624467134475708,
      "learning_rate": 3.941477791032209e-05,
      "loss": 0.8428,
      "step": 4601
    },
    {
      "epoch": 1.4531101989264288,
      "grad_norm": 1.6602729558944702,
      "learning_rate": 3.9411269384604595e-05,
      "loss": 0.8474,
      "step": 4602
    },
    {
      "epoch": 1.4534259551626145,
      "grad_norm": 1.768412709236145,
      "learning_rate": 3.94077608588871e-05,
      "loss": 0.8333,
      "step": 4603
    },
    {
      "epoch": 1.4537417113988003,
      "grad_norm": 1.4076181650161743,
      "learning_rate": 3.940425233316961e-05,
      "loss": 0.8256,
      "step": 4604
    },
    {
      "epoch": 1.4540574676349858,
      "grad_norm": 1.4410710334777832,
      "learning_rate": 3.940074380745211e-05,
      "loss": 0.7972,
      "step": 4605
    },
    {
      "epoch": 1.4543732238711715,
      "grad_norm": 1.5973460674285889,
      "learning_rate": 3.9397235281734615e-05,
      "loss": 0.8853,
      "step": 4606
    },
    {
      "epoch": 1.454688980107357,
      "grad_norm": 1.7262035608291626,
      "learning_rate": 3.939372675601712e-05,
      "loss": 0.8433,
      "step": 4607
    },
    {
      "epoch": 1.4550047363435428,
      "grad_norm": 1.690081238746643,
      "learning_rate": 3.939021823029963e-05,
      "loss": 0.7337,
      "step": 4608
    },
    {
      "epoch": 1.4553204925797285,
      "grad_norm": 1.8056745529174805,
      "learning_rate": 3.9386709704582136e-05,
      "loss": 0.878,
      "step": 4609
    },
    {
      "epoch": 1.4556362488159142,
      "grad_norm": 1.5484894514083862,
      "learning_rate": 3.938320117886464e-05,
      "loss": 0.8209,
      "step": 4610
    },
    {
      "epoch": 1.4559520050520998,
      "grad_norm": 1.7405664920806885,
      "learning_rate": 3.937969265314715e-05,
      "loss": 0.8786,
      "step": 4611
    },
    {
      "epoch": 1.4562677612882855,
      "grad_norm": 1.652255892753601,
      "learning_rate": 3.9376184127429656e-05,
      "loss": 0.8963,
      "step": 4612
    },
    {
      "epoch": 1.456583517524471,
      "grad_norm": 1.5860185623168945,
      "learning_rate": 3.937267560171216e-05,
      "loss": 0.8919,
      "step": 4613
    },
    {
      "epoch": 1.4568992737606568,
      "grad_norm": 1.6306465864181519,
      "learning_rate": 3.936916707599467e-05,
      "loss": 0.8985,
      "step": 4614
    },
    {
      "epoch": 1.4572150299968425,
      "grad_norm": 1.7915350198745728,
      "learning_rate": 3.936565855027718e-05,
      "loss": 0.9105,
      "step": 4615
    },
    {
      "epoch": 1.4575307862330282,
      "grad_norm": 1.4912025928497314,
      "learning_rate": 3.936215002455968e-05,
      "loss": 0.8332,
      "step": 4616
    },
    {
      "epoch": 1.4578465424692137,
      "grad_norm": 1.4856559038162231,
      "learning_rate": 3.9358641498842184e-05,
      "loss": 0.8627,
      "step": 4617
    },
    {
      "epoch": 1.4581622987053995,
      "grad_norm": 1.6728370189666748,
      "learning_rate": 3.93551329731247e-05,
      "loss": 0.8509,
      "step": 4618
    },
    {
      "epoch": 1.458478054941585,
      "grad_norm": 1.519215703010559,
      "learning_rate": 3.9351624447407204e-05,
      "loss": 0.8165,
      "step": 4619
    },
    {
      "epoch": 1.4587938111777707,
      "grad_norm": 1.5332660675048828,
      "learning_rate": 3.934811592168971e-05,
      "loss": 0.8773,
      "step": 4620
    },
    {
      "epoch": 1.4591095674139565,
      "grad_norm": 1.5925939083099365,
      "learning_rate": 3.934460739597222e-05,
      "loss": 0.9607,
      "step": 4621
    },
    {
      "epoch": 1.4594253236501422,
      "grad_norm": 1.6545419692993164,
      "learning_rate": 3.9341098870254724e-05,
      "loss": 0.8551,
      "step": 4622
    },
    {
      "epoch": 1.4597410798863277,
      "grad_norm": 1.425685167312622,
      "learning_rate": 3.933759034453723e-05,
      "loss": 0.8204,
      "step": 4623
    },
    {
      "epoch": 1.4600568361225135,
      "grad_norm": 1.4133869409561157,
      "learning_rate": 3.933408181881973e-05,
      "loss": 0.7693,
      "step": 4624
    },
    {
      "epoch": 1.460372592358699,
      "grad_norm": 1.4362865686416626,
      "learning_rate": 3.933057329310224e-05,
      "loss": 0.7862,
      "step": 4625
    },
    {
      "epoch": 1.4606883485948847,
      "grad_norm": 1.4716010093688965,
      "learning_rate": 3.9327064767384745e-05,
      "loss": 0.8735,
      "step": 4626
    },
    {
      "epoch": 1.4610041048310705,
      "grad_norm": 1.4303773641586304,
      "learning_rate": 3.932355624166725e-05,
      "loss": 0.7791,
      "step": 4627
    },
    {
      "epoch": 1.4613198610672562,
      "grad_norm": 1.4327508211135864,
      "learning_rate": 3.932004771594976e-05,
      "loss": 0.7917,
      "step": 4628
    },
    {
      "epoch": 1.4616356173034417,
      "grad_norm": 1.6997747421264648,
      "learning_rate": 3.9316539190232265e-05,
      "loss": 0.8483,
      "step": 4629
    },
    {
      "epoch": 1.4619513735396275,
      "grad_norm": 1.7507926225662231,
      "learning_rate": 3.931303066451477e-05,
      "loss": 0.8943,
      "step": 4630
    },
    {
      "epoch": 1.462267129775813,
      "grad_norm": 1.4732879400253296,
      "learning_rate": 3.930952213879728e-05,
      "loss": 0.7992,
      "step": 4631
    },
    {
      "epoch": 1.4625828860119987,
      "grad_norm": 1.5090649127960205,
      "learning_rate": 3.9306013613079786e-05,
      "loss": 0.7921,
      "step": 4632
    },
    {
      "epoch": 1.4628986422481844,
      "grad_norm": 1.5671354532241821,
      "learning_rate": 3.930250508736229e-05,
      "loss": 0.838,
      "step": 4633
    },
    {
      "epoch": 1.4632143984843702,
      "grad_norm": 1.6738996505737305,
      "learning_rate": 3.92989965616448e-05,
      "loss": 0.8556,
      "step": 4634
    },
    {
      "epoch": 1.4635301547205557,
      "grad_norm": 1.6274397373199463,
      "learning_rate": 3.92954880359273e-05,
      "loss": 0.9086,
      "step": 4635
    },
    {
      "epoch": 1.4638459109567414,
      "grad_norm": 1.6783688068389893,
      "learning_rate": 3.929197951020981e-05,
      "loss": 0.8047,
      "step": 4636
    },
    {
      "epoch": 1.464161667192927,
      "grad_norm": 1.55343496799469,
      "learning_rate": 3.928847098449232e-05,
      "loss": 0.7998,
      "step": 4637
    },
    {
      "epoch": 1.4644774234291127,
      "grad_norm": 1.7193371057510376,
      "learning_rate": 3.928496245877483e-05,
      "loss": 0.8508,
      "step": 4638
    },
    {
      "epoch": 1.4647931796652984,
      "grad_norm": 1.5816071033477783,
      "learning_rate": 3.9281453933057334e-05,
      "loss": 0.8205,
      "step": 4639
    },
    {
      "epoch": 1.4651089359014842,
      "grad_norm": 1.560926079750061,
      "learning_rate": 3.927794540733984e-05,
      "loss": 0.7897,
      "step": 4640
    },
    {
      "epoch": 1.4654246921376697,
      "grad_norm": 1.547070860862732,
      "learning_rate": 3.927443688162235e-05,
      "loss": 0.7497,
      "step": 4641
    },
    {
      "epoch": 1.4657404483738554,
      "grad_norm": 1.8873211145401,
      "learning_rate": 3.9270928355904854e-05,
      "loss": 0.9028,
      "step": 4642
    },
    {
      "epoch": 1.466056204610041,
      "grad_norm": 1.7548028230667114,
      "learning_rate": 3.926741983018736e-05,
      "loss": 0.8152,
      "step": 4643
    },
    {
      "epoch": 1.4663719608462267,
      "grad_norm": 1.5016502141952515,
      "learning_rate": 3.926391130446986e-05,
      "loss": 0.8449,
      "step": 4644
    },
    {
      "epoch": 1.4666877170824124,
      "grad_norm": 1.4958807229995728,
      "learning_rate": 3.926040277875237e-05,
      "loss": 0.8914,
      "step": 4645
    },
    {
      "epoch": 1.4670034733185982,
      "grad_norm": 1.2564594745635986,
      "learning_rate": 3.9256894253034874e-05,
      "loss": 0.7689,
      "step": 4646
    },
    {
      "epoch": 1.4673192295547837,
      "grad_norm": 1.7672631740570068,
      "learning_rate": 3.925338572731738e-05,
      "loss": 0.9092,
      "step": 4647
    },
    {
      "epoch": 1.4676349857909694,
      "grad_norm": 1.404791235923767,
      "learning_rate": 3.924987720159989e-05,
      "loss": 0.7468,
      "step": 4648
    },
    {
      "epoch": 1.467950742027155,
      "grad_norm": 1.4610538482666016,
      "learning_rate": 3.9246368675882395e-05,
      "loss": 0.8546,
      "step": 4649
    },
    {
      "epoch": 1.4682664982633407,
      "grad_norm": 1.730507493019104,
      "learning_rate": 3.92428601501649e-05,
      "loss": 0.8245,
      "step": 4650
    },
    {
      "epoch": 1.4685822544995264,
      "grad_norm": 1.5226496458053589,
      "learning_rate": 3.923935162444741e-05,
      "loss": 0.8245,
      "step": 4651
    },
    {
      "epoch": 1.4688980107357121,
      "grad_norm": 1.5795046091079712,
      "learning_rate": 3.9235843098729915e-05,
      "loss": 0.8372,
      "step": 4652
    },
    {
      "epoch": 1.4692137669718977,
      "grad_norm": 1.474113941192627,
      "learning_rate": 3.923233457301242e-05,
      "loss": 0.79,
      "step": 4653
    },
    {
      "epoch": 1.4695295232080834,
      "grad_norm": 1.4349708557128906,
      "learning_rate": 3.922882604729493e-05,
      "loss": 0.8141,
      "step": 4654
    },
    {
      "epoch": 1.469845279444269,
      "grad_norm": 1.3993587493896484,
      "learning_rate": 3.9225317521577436e-05,
      "loss": 0.8814,
      "step": 4655
    },
    {
      "epoch": 1.4701610356804546,
      "grad_norm": 1.438961386680603,
      "learning_rate": 3.922180899585994e-05,
      "loss": 0.8459,
      "step": 4656
    },
    {
      "epoch": 1.4704767919166404,
      "grad_norm": 1.4333739280700684,
      "learning_rate": 3.921830047014245e-05,
      "loss": 0.8586,
      "step": 4657
    },
    {
      "epoch": 1.4707925481528261,
      "grad_norm": 1.5605816841125488,
      "learning_rate": 3.9214791944424956e-05,
      "loss": 0.8784,
      "step": 4658
    },
    {
      "epoch": 1.4711083043890116,
      "grad_norm": 1.228149652481079,
      "learning_rate": 3.921128341870746e-05,
      "loss": 0.7718,
      "step": 4659
    },
    {
      "epoch": 1.4714240606251974,
      "grad_norm": 1.5439503192901611,
      "learning_rate": 3.920777489298997e-05,
      "loss": 0.7921,
      "step": 4660
    },
    {
      "epoch": 1.4717398168613829,
      "grad_norm": 1.6866742372512817,
      "learning_rate": 3.920426636727248e-05,
      "loss": 0.838,
      "step": 4661
    },
    {
      "epoch": 1.4720555730975686,
      "grad_norm": 1.3130435943603516,
      "learning_rate": 3.9200757841554983e-05,
      "loss": 0.7471,
      "step": 4662
    },
    {
      "epoch": 1.4723713293337544,
      "grad_norm": 1.3303450345993042,
      "learning_rate": 3.9197249315837483e-05,
      "loss": 0.839,
      "step": 4663
    },
    {
      "epoch": 1.47268708556994,
      "grad_norm": 1.4809139966964722,
      "learning_rate": 3.919374079011999e-05,
      "loss": 0.814,
      "step": 4664
    },
    {
      "epoch": 1.4730028418061256,
      "grad_norm": 1.274507761001587,
      "learning_rate": 3.91902322644025e-05,
      "loss": 0.7201,
      "step": 4665
    },
    {
      "epoch": 1.4733185980423114,
      "grad_norm": 1.4779845476150513,
      "learning_rate": 3.9186723738685004e-05,
      "loss": 0.8823,
      "step": 4666
    },
    {
      "epoch": 1.4736343542784969,
      "grad_norm": 1.7506837844848633,
      "learning_rate": 3.918321521296751e-05,
      "loss": 0.9447,
      "step": 4667
    },
    {
      "epoch": 1.4739501105146826,
      "grad_norm": 1.491930603981018,
      "learning_rate": 3.917970668725002e-05,
      "loss": 0.8158,
      "step": 4668
    },
    {
      "epoch": 1.4742658667508683,
      "grad_norm": 1.3493465185165405,
      "learning_rate": 3.9176198161532524e-05,
      "loss": 0.8336,
      "step": 4669
    },
    {
      "epoch": 1.474581622987054,
      "grad_norm": 1.6757386922836304,
      "learning_rate": 3.917268963581503e-05,
      "loss": 0.8657,
      "step": 4670
    },
    {
      "epoch": 1.4748973792232396,
      "grad_norm": 1.5359625816345215,
      "learning_rate": 3.9169181110097545e-05,
      "loss": 0.8521,
      "step": 4671
    },
    {
      "epoch": 1.4752131354594253,
      "grad_norm": 1.4655091762542725,
      "learning_rate": 3.9165672584380045e-05,
      "loss": 0.8503,
      "step": 4672
    },
    {
      "epoch": 1.4755288916956109,
      "grad_norm": 1.6890323162078857,
      "learning_rate": 3.916216405866255e-05,
      "loss": 0.8495,
      "step": 4673
    },
    {
      "epoch": 1.4758446479317966,
      "grad_norm": 1.297249674797058,
      "learning_rate": 3.915865553294506e-05,
      "loss": 0.8552,
      "step": 4674
    },
    {
      "epoch": 1.4761604041679823,
      "grad_norm": 1.4702285528182983,
      "learning_rate": 3.9155147007227565e-05,
      "loss": 0.8241,
      "step": 4675
    },
    {
      "epoch": 1.476476160404168,
      "grad_norm": 1.5535651445388794,
      "learning_rate": 3.915163848151007e-05,
      "loss": 0.8584,
      "step": 4676
    },
    {
      "epoch": 1.4767919166403536,
      "grad_norm": 1.566666841506958,
      "learning_rate": 3.914812995579258e-05,
      "loss": 0.8535,
      "step": 4677
    },
    {
      "epoch": 1.4771076728765393,
      "grad_norm": 1.7296886444091797,
      "learning_rate": 3.9144621430075086e-05,
      "loss": 0.8549,
      "step": 4678
    },
    {
      "epoch": 1.4774234291127248,
      "grad_norm": 1.615769624710083,
      "learning_rate": 3.914111290435759e-05,
      "loss": 0.8191,
      "step": 4679
    },
    {
      "epoch": 1.4777391853489106,
      "grad_norm": 1.5854003429412842,
      "learning_rate": 3.91376043786401e-05,
      "loss": 0.8708,
      "step": 4680
    },
    {
      "epoch": 1.4780549415850963,
      "grad_norm": 1.6196036338806152,
      "learning_rate": 3.9134095852922606e-05,
      "loss": 0.9275,
      "step": 4681
    },
    {
      "epoch": 1.478370697821282,
      "grad_norm": 1.6865230798721313,
      "learning_rate": 3.9130587327205106e-05,
      "loss": 0.9055,
      "step": 4682
    },
    {
      "epoch": 1.4786864540574676,
      "grad_norm": 1.4032491445541382,
      "learning_rate": 3.912707880148761e-05,
      "loss": 0.8442,
      "step": 4683
    },
    {
      "epoch": 1.4790022102936533,
      "grad_norm": 1.3028146028518677,
      "learning_rate": 3.912357027577012e-05,
      "loss": 0.8092,
      "step": 4684
    },
    {
      "epoch": 1.479317966529839,
      "grad_norm": 1.620944857597351,
      "learning_rate": 3.9120061750052627e-05,
      "loss": 0.9665,
      "step": 4685
    },
    {
      "epoch": 1.4796337227660246,
      "grad_norm": 1.5517371892929077,
      "learning_rate": 3.9116553224335133e-05,
      "loss": 0.8583,
      "step": 4686
    },
    {
      "epoch": 1.4799494790022103,
      "grad_norm": 1.4960241317749023,
      "learning_rate": 3.911304469861764e-05,
      "loss": 0.8194,
      "step": 4687
    },
    {
      "epoch": 1.480265235238396,
      "grad_norm": 1.3401703834533691,
      "learning_rate": 3.9109536172900154e-05,
      "loss": 0.7739,
      "step": 4688
    },
    {
      "epoch": 1.4805809914745816,
      "grad_norm": 1.4302645921707153,
      "learning_rate": 3.910602764718266e-05,
      "loss": 0.7876,
      "step": 4689
    },
    {
      "epoch": 1.4808967477107673,
      "grad_norm": 1.3229533433914185,
      "learning_rate": 3.910251912146517e-05,
      "loss": 0.8232,
      "step": 4690
    },
    {
      "epoch": 1.481212503946953,
      "grad_norm": 1.4940483570098877,
      "learning_rate": 3.909901059574767e-05,
      "loss": 0.8871,
      "step": 4691
    },
    {
      "epoch": 1.4815282601831385,
      "grad_norm": 1.3620089292526245,
      "learning_rate": 3.9095502070030174e-05,
      "loss": 0.9126,
      "step": 4692
    },
    {
      "epoch": 1.4818440164193243,
      "grad_norm": 1.6312066316604614,
      "learning_rate": 3.909199354431268e-05,
      "loss": 0.8596,
      "step": 4693
    },
    {
      "epoch": 1.48215977265551,
      "grad_norm": 1.3037011623382568,
      "learning_rate": 3.908848501859519e-05,
      "loss": 0.8808,
      "step": 4694
    },
    {
      "epoch": 1.4824755288916955,
      "grad_norm": 1.4708293676376343,
      "learning_rate": 3.9084976492877695e-05,
      "loss": 0.7811,
      "step": 4695
    },
    {
      "epoch": 1.4827912851278813,
      "grad_norm": 1.6734509468078613,
      "learning_rate": 3.90814679671602e-05,
      "loss": 0.8295,
      "step": 4696
    },
    {
      "epoch": 1.483107041364067,
      "grad_norm": 1.7514636516571045,
      "learning_rate": 3.907795944144271e-05,
      "loss": 0.8305,
      "step": 4697
    },
    {
      "epoch": 1.4834227976002525,
      "grad_norm": 1.26260244846344,
      "learning_rate": 3.9074450915725215e-05,
      "loss": 0.8163,
      "step": 4698
    },
    {
      "epoch": 1.4837385538364383,
      "grad_norm": 1.2365955114364624,
      "learning_rate": 3.907094239000772e-05,
      "loss": 0.8539,
      "step": 4699
    },
    {
      "epoch": 1.484054310072624,
      "grad_norm": 1.3348604440689087,
      "learning_rate": 3.906743386429023e-05,
      "loss": 0.8127,
      "step": 4700
    },
    {
      "epoch": 1.4843700663088095,
      "grad_norm": 1.5698764324188232,
      "learning_rate": 3.906392533857273e-05,
      "loss": 0.8605,
      "step": 4701
    },
    {
      "epoch": 1.4846858225449953,
      "grad_norm": 1.4569913148880005,
      "learning_rate": 3.9060416812855236e-05,
      "loss": 0.8083,
      "step": 4702
    },
    {
      "epoch": 1.485001578781181,
      "grad_norm": 1.7658709287643433,
      "learning_rate": 3.905690828713774e-05,
      "loss": 0.8135,
      "step": 4703
    },
    {
      "epoch": 1.4853173350173665,
      "grad_norm": 1.4752535820007324,
      "learning_rate": 3.905339976142025e-05,
      "loss": 0.881,
      "step": 4704
    },
    {
      "epoch": 1.4856330912535523,
      "grad_norm": 1.6538231372833252,
      "learning_rate": 3.9049891235702756e-05,
      "loss": 0.8095,
      "step": 4705
    },
    {
      "epoch": 1.485948847489738,
      "grad_norm": 1.5558148622512817,
      "learning_rate": 3.904638270998527e-05,
      "loss": 0.9068,
      "step": 4706
    },
    {
      "epoch": 1.4862646037259235,
      "grad_norm": 1.4773924350738525,
      "learning_rate": 3.9042874184267776e-05,
      "loss": 0.8457,
      "step": 4707
    },
    {
      "epoch": 1.4865803599621092,
      "grad_norm": 1.5849956274032593,
      "learning_rate": 3.903936565855028e-05,
      "loss": 0.7966,
      "step": 4708
    },
    {
      "epoch": 1.486896116198295,
      "grad_norm": 1.8696885108947754,
      "learning_rate": 3.903585713283279e-05,
      "loss": 0.9577,
      "step": 4709
    },
    {
      "epoch": 1.4872118724344805,
      "grad_norm": 1.4024370908737183,
      "learning_rate": 3.903234860711529e-05,
      "loss": 0.7755,
      "step": 4710
    },
    {
      "epoch": 1.4875276286706662,
      "grad_norm": 1.3735047578811646,
      "learning_rate": 3.90288400813978e-05,
      "loss": 0.7994,
      "step": 4711
    },
    {
      "epoch": 1.487843384906852,
      "grad_norm": 1.5606880187988281,
      "learning_rate": 3.9025331555680304e-05,
      "loss": 0.8432,
      "step": 4712
    },
    {
      "epoch": 1.4881591411430377,
      "grad_norm": 1.8677892684936523,
      "learning_rate": 3.902182302996281e-05,
      "loss": 0.868,
      "step": 4713
    },
    {
      "epoch": 1.4884748973792232,
      "grad_norm": 1.2619835138320923,
      "learning_rate": 3.901831450424532e-05,
      "loss": 0.8237,
      "step": 4714
    },
    {
      "epoch": 1.488790653615409,
      "grad_norm": 1.428032398223877,
      "learning_rate": 3.9014805978527824e-05,
      "loss": 0.885,
      "step": 4715
    },
    {
      "epoch": 1.4891064098515945,
      "grad_norm": 1.383660078048706,
      "learning_rate": 3.901129745281033e-05,
      "loss": 0.8397,
      "step": 4716
    },
    {
      "epoch": 1.4894221660877802,
      "grad_norm": 1.9475754499435425,
      "learning_rate": 3.900778892709284e-05,
      "loss": 0.8032,
      "step": 4717
    },
    {
      "epoch": 1.489737922323966,
      "grad_norm": 1.453322410583496,
      "learning_rate": 3.9004280401375345e-05,
      "loss": 0.8014,
      "step": 4718
    },
    {
      "epoch": 1.4900536785601517,
      "grad_norm": 1.5051252841949463,
      "learning_rate": 3.900077187565785e-05,
      "loss": 0.8641,
      "step": 4719
    },
    {
      "epoch": 1.4903694347963372,
      "grad_norm": 1.5723737478256226,
      "learning_rate": 3.899726334994036e-05,
      "loss": 0.8502,
      "step": 4720
    },
    {
      "epoch": 1.490685191032523,
      "grad_norm": 1.45952570438385,
      "learning_rate": 3.899375482422286e-05,
      "loss": 0.8661,
      "step": 4721
    },
    {
      "epoch": 1.4910009472687085,
      "grad_norm": 1.574723482131958,
      "learning_rate": 3.8990246298505365e-05,
      "loss": 0.9059,
      "step": 4722
    },
    {
      "epoch": 1.4913167035048942,
      "grad_norm": 1.454137921333313,
      "learning_rate": 3.898673777278788e-05,
      "loss": 0.8238,
      "step": 4723
    },
    {
      "epoch": 1.49163245974108,
      "grad_norm": 1.308060884475708,
      "learning_rate": 3.8983229247070386e-05,
      "loss": 0.7958,
      "step": 4724
    },
    {
      "epoch": 1.4919482159772657,
      "grad_norm": 1.6803967952728271,
      "learning_rate": 3.897972072135289e-05,
      "loss": 0.8497,
      "step": 4725
    },
    {
      "epoch": 1.4922639722134512,
      "grad_norm": 1.5085259675979614,
      "learning_rate": 3.89762121956354e-05,
      "loss": 0.7931,
      "step": 4726
    },
    {
      "epoch": 1.492579728449637,
      "grad_norm": 1.6440339088439941,
      "learning_rate": 3.8972703669917906e-05,
      "loss": 0.8374,
      "step": 4727
    },
    {
      "epoch": 1.4928954846858224,
      "grad_norm": 1.7148555517196655,
      "learning_rate": 3.896919514420041e-05,
      "loss": 0.9089,
      "step": 4728
    },
    {
      "epoch": 1.4932112409220082,
      "grad_norm": 1.3897795677185059,
      "learning_rate": 3.896568661848291e-05,
      "loss": 0.9053,
      "step": 4729
    },
    {
      "epoch": 1.493526997158194,
      "grad_norm": 1.4018199443817139,
      "learning_rate": 3.896217809276542e-05,
      "loss": 0.7662,
      "step": 4730
    },
    {
      "epoch": 1.4938427533943797,
      "grad_norm": 1.5593526363372803,
      "learning_rate": 3.8958669567047926e-05,
      "loss": 0.8299,
      "step": 4731
    },
    {
      "epoch": 1.4941585096305652,
      "grad_norm": 1.3806626796722412,
      "learning_rate": 3.895516104133043e-05,
      "loss": 0.7897,
      "step": 4732
    },
    {
      "epoch": 1.494474265866751,
      "grad_norm": 1.7311768531799316,
      "learning_rate": 3.895165251561294e-05,
      "loss": 0.9005,
      "step": 4733
    },
    {
      "epoch": 1.4947900221029364,
      "grad_norm": 1.3958072662353516,
      "learning_rate": 3.894814398989545e-05,
      "loss": 0.7724,
      "step": 4734
    },
    {
      "epoch": 1.4951057783391222,
      "grad_norm": 1.6490064859390259,
      "learning_rate": 3.8944635464177954e-05,
      "loss": 0.8001,
      "step": 4735
    },
    {
      "epoch": 1.495421534575308,
      "grad_norm": 1.6494758129119873,
      "learning_rate": 3.894112693846046e-05,
      "loss": 0.9076,
      "step": 4736
    },
    {
      "epoch": 1.4957372908114936,
      "grad_norm": 1.7116785049438477,
      "learning_rate": 3.893761841274297e-05,
      "loss": 0.7663,
      "step": 4737
    },
    {
      "epoch": 1.4960530470476792,
      "grad_norm": 1.5092246532440186,
      "learning_rate": 3.8934109887025474e-05,
      "loss": 0.8458,
      "step": 4738
    },
    {
      "epoch": 1.496368803283865,
      "grad_norm": 1.6534548997879028,
      "learning_rate": 3.893060136130798e-05,
      "loss": 0.8846,
      "step": 4739
    },
    {
      "epoch": 1.4966845595200504,
      "grad_norm": 1.7784606218338013,
      "learning_rate": 3.892709283559048e-05,
      "loss": 0.8765,
      "step": 4740
    },
    {
      "epoch": 1.4970003157562362,
      "grad_norm": 1.5050069093704224,
      "learning_rate": 3.8923584309872995e-05,
      "loss": 0.8184,
      "step": 4741
    },
    {
      "epoch": 1.497316071992422,
      "grad_norm": 1.3976389169692993,
      "learning_rate": 3.89200757841555e-05,
      "loss": 0.7905,
      "step": 4742
    },
    {
      "epoch": 1.4976318282286076,
      "grad_norm": 1.569875717163086,
      "learning_rate": 3.891656725843801e-05,
      "loss": 0.8448,
      "step": 4743
    },
    {
      "epoch": 1.4979475844647931,
      "grad_norm": 1.4596391916275024,
      "learning_rate": 3.8913058732720515e-05,
      "loss": 0.8896,
      "step": 4744
    },
    {
      "epoch": 1.4982633407009789,
      "grad_norm": 1.5294045209884644,
      "learning_rate": 3.890955020700302e-05,
      "loss": 0.8516,
      "step": 4745
    },
    {
      "epoch": 1.4985790969371644,
      "grad_norm": 1.6820499897003174,
      "learning_rate": 3.890604168128553e-05,
      "loss": 0.8556,
      "step": 4746
    },
    {
      "epoch": 1.4988948531733501,
      "grad_norm": 1.6318995952606201,
      "learning_rate": 3.8902533155568035e-05,
      "loss": 0.8238,
      "step": 4747
    },
    {
      "epoch": 1.4992106094095359,
      "grad_norm": 1.3485645055770874,
      "learning_rate": 3.889902462985054e-05,
      "loss": 0.7921,
      "step": 4748
    },
    {
      "epoch": 1.4995263656457216,
      "grad_norm": 1.4214975833892822,
      "learning_rate": 3.889551610413304e-05,
      "loss": 0.8261,
      "step": 4749
    },
    {
      "epoch": 1.4998421218819071,
      "grad_norm": 1.4903459548950195,
      "learning_rate": 3.889200757841555e-05,
      "loss": 0.7487,
      "step": 4750
    },
    {
      "epoch": 1.5001578781180929,
      "grad_norm": 1.5824497938156128,
      "learning_rate": 3.8888499052698056e-05,
      "loss": 0.8089,
      "step": 4751
    },
    {
      "epoch": 1.5004736343542784,
      "grad_norm": 1.381673812866211,
      "learning_rate": 3.888499052698056e-05,
      "loss": 0.7788,
      "step": 4752
    },
    {
      "epoch": 1.5007893905904641,
      "grad_norm": 1.5604487657546997,
      "learning_rate": 3.888148200126307e-05,
      "loss": 0.7858,
      "step": 4753
    },
    {
      "epoch": 1.5011051468266499,
      "grad_norm": 1.3632744550704956,
      "learning_rate": 3.8877973475545576e-05,
      "loss": 0.7832,
      "step": 4754
    },
    {
      "epoch": 1.5014209030628356,
      "grad_norm": 1.4881318807601929,
      "learning_rate": 3.887446494982808e-05,
      "loss": 0.8242,
      "step": 4755
    },
    {
      "epoch": 1.5017366592990211,
      "grad_norm": 1.7879523038864136,
      "learning_rate": 3.887095642411059e-05,
      "loss": 0.8903,
      "step": 4756
    },
    {
      "epoch": 1.5020524155352069,
      "grad_norm": 1.3966681957244873,
      "learning_rate": 3.88674478983931e-05,
      "loss": 0.8226,
      "step": 4757
    },
    {
      "epoch": 1.5023681717713924,
      "grad_norm": 1.4724209308624268,
      "learning_rate": 3.8863939372675604e-05,
      "loss": 0.7783,
      "step": 4758
    },
    {
      "epoch": 1.502683928007578,
      "grad_norm": 1.766901969909668,
      "learning_rate": 3.886043084695811e-05,
      "loss": 0.9176,
      "step": 4759
    },
    {
      "epoch": 1.5029996842437638,
      "grad_norm": 1.5829044580459595,
      "learning_rate": 3.885692232124062e-05,
      "loss": 0.8843,
      "step": 4760
    },
    {
      "epoch": 1.5033154404799496,
      "grad_norm": 1.7893955707550049,
      "learning_rate": 3.8853413795523124e-05,
      "loss": 0.8911,
      "step": 4761
    },
    {
      "epoch": 1.503631196716135,
      "grad_norm": 1.6440526247024536,
      "learning_rate": 3.884990526980563e-05,
      "loss": 0.8209,
      "step": 4762
    },
    {
      "epoch": 1.5039469529523208,
      "grad_norm": 1.6354047060012817,
      "learning_rate": 3.884639674408814e-05,
      "loss": 0.8257,
      "step": 4763
    },
    {
      "epoch": 1.5042627091885064,
      "grad_norm": 1.6097866296768188,
      "learning_rate": 3.8842888218370644e-05,
      "loss": 0.8227,
      "step": 4764
    },
    {
      "epoch": 1.504578465424692,
      "grad_norm": 1.6272201538085938,
      "learning_rate": 3.883937969265315e-05,
      "loss": 0.7891,
      "step": 4765
    },
    {
      "epoch": 1.5048942216608778,
      "grad_norm": 1.2800514698028564,
      "learning_rate": 3.883587116693566e-05,
      "loss": 0.8132,
      "step": 4766
    },
    {
      "epoch": 1.5052099778970636,
      "grad_norm": 1.4992262125015259,
      "learning_rate": 3.8832362641218165e-05,
      "loss": 0.8507,
      "step": 4767
    },
    {
      "epoch": 1.505525734133249,
      "grad_norm": 1.3803297281265259,
      "learning_rate": 3.8828854115500665e-05,
      "loss": 0.9081,
      "step": 4768
    },
    {
      "epoch": 1.5058414903694348,
      "grad_norm": 1.611128807067871,
      "learning_rate": 3.882534558978317e-05,
      "loss": 0.8928,
      "step": 4769
    },
    {
      "epoch": 1.5061572466056203,
      "grad_norm": 1.562736988067627,
      "learning_rate": 3.882183706406568e-05,
      "loss": 0.8103,
      "step": 4770
    },
    {
      "epoch": 1.506473002841806,
      "grad_norm": 1.4529516696929932,
      "learning_rate": 3.8818328538348185e-05,
      "loss": 0.833,
      "step": 4771
    },
    {
      "epoch": 1.5067887590779918,
      "grad_norm": 1.4256623983383179,
      "learning_rate": 3.881482001263069e-05,
      "loss": 0.8052,
      "step": 4772
    },
    {
      "epoch": 1.5071045153141776,
      "grad_norm": 1.4643378257751465,
      "learning_rate": 3.88113114869132e-05,
      "loss": 0.8311,
      "step": 4773
    },
    {
      "epoch": 1.507420271550363,
      "grad_norm": 1.4209450483322144,
      "learning_rate": 3.8807802961195706e-05,
      "loss": 0.8326,
      "step": 4774
    },
    {
      "epoch": 1.5077360277865488,
      "grad_norm": 1.421772837638855,
      "learning_rate": 3.880429443547821e-05,
      "loss": 0.8242,
      "step": 4775
    },
    {
      "epoch": 1.5080517840227343,
      "grad_norm": 1.3281100988388062,
      "learning_rate": 3.8800785909760726e-05,
      "loss": 0.759,
      "step": 4776
    },
    {
      "epoch": 1.50836754025892,
      "grad_norm": 1.5192646980285645,
      "learning_rate": 3.8797277384043226e-05,
      "loss": 0.8592,
      "step": 4777
    },
    {
      "epoch": 1.5086832964951058,
      "grad_norm": 1.4933596849441528,
      "learning_rate": 3.879376885832573e-05,
      "loss": 0.7721,
      "step": 4778
    },
    {
      "epoch": 1.5089990527312915,
      "grad_norm": 1.5384392738342285,
      "learning_rate": 3.879026033260824e-05,
      "loss": 0.922,
      "step": 4779
    },
    {
      "epoch": 1.5093148089674773,
      "grad_norm": 1.603692889213562,
      "learning_rate": 3.878675180689075e-05,
      "loss": 0.8515,
      "step": 4780
    },
    {
      "epoch": 1.5096305652036628,
      "grad_norm": 1.7550874948501587,
      "learning_rate": 3.8783243281173254e-05,
      "loss": 0.8271,
      "step": 4781
    },
    {
      "epoch": 1.5099463214398483,
      "grad_norm": 1.6288964748382568,
      "learning_rate": 3.877973475545576e-05,
      "loss": 0.7515,
      "step": 4782
    },
    {
      "epoch": 1.510262077676034,
      "grad_norm": 1.3227311372756958,
      "learning_rate": 3.877622622973827e-05,
      "loss": 0.8188,
      "step": 4783
    },
    {
      "epoch": 1.5105778339122198,
      "grad_norm": 1.4396750926971436,
      "learning_rate": 3.8772717704020774e-05,
      "loss": 0.797,
      "step": 4784
    },
    {
      "epoch": 1.5108935901484055,
      "grad_norm": 1.6676241159439087,
      "learning_rate": 3.876920917830328e-05,
      "loss": 0.786,
      "step": 4785
    },
    {
      "epoch": 1.5112093463845913,
      "grad_norm": 1.844314694404602,
      "learning_rate": 3.876570065258579e-05,
      "loss": 0.8863,
      "step": 4786
    },
    {
      "epoch": 1.5115251026207768,
      "grad_norm": 1.7386523485183716,
      "learning_rate": 3.876219212686829e-05,
      "loss": 0.8248,
      "step": 4787
    },
    {
      "epoch": 1.5118408588569623,
      "grad_norm": 1.574926495552063,
      "learning_rate": 3.8758683601150794e-05,
      "loss": 0.8706,
      "step": 4788
    },
    {
      "epoch": 1.512156615093148,
      "grad_norm": 1.6961305141448975,
      "learning_rate": 3.87551750754333e-05,
      "loss": 0.8534,
      "step": 4789
    },
    {
      "epoch": 1.5124723713293338,
      "grad_norm": 1.4519894123077393,
      "learning_rate": 3.875166654971581e-05,
      "loss": 0.776,
      "step": 4790
    },
    {
      "epoch": 1.5127881275655195,
      "grad_norm": 1.7289983034133911,
      "learning_rate": 3.8748158023998315e-05,
      "loss": 0.8699,
      "step": 4791
    },
    {
      "epoch": 1.5131038838017052,
      "grad_norm": 1.4284460544586182,
      "learning_rate": 3.874464949828082e-05,
      "loss": 0.7731,
      "step": 4792
    },
    {
      "epoch": 1.5134196400378908,
      "grad_norm": 1.6451168060302734,
      "learning_rate": 3.8741140972563335e-05,
      "loss": 0.841,
      "step": 4793
    },
    {
      "epoch": 1.5137353962740763,
      "grad_norm": 1.51224946975708,
      "learning_rate": 3.873763244684584e-05,
      "loss": 0.8208,
      "step": 4794
    },
    {
      "epoch": 1.514051152510262,
      "grad_norm": 1.923775553703308,
      "learning_rate": 3.873412392112835e-05,
      "loss": 0.8601,
      "step": 4795
    },
    {
      "epoch": 1.5143669087464477,
      "grad_norm": 1.904848575592041,
      "learning_rate": 3.873061539541085e-05,
      "loss": 0.8507,
      "step": 4796
    },
    {
      "epoch": 1.5146826649826335,
      "grad_norm": 1.721415400505066,
      "learning_rate": 3.8727106869693356e-05,
      "loss": 0.8946,
      "step": 4797
    },
    {
      "epoch": 1.5149984212188192,
      "grad_norm": 1.6277210712432861,
      "learning_rate": 3.872359834397586e-05,
      "loss": 0.8471,
      "step": 4798
    },
    {
      "epoch": 1.5153141774550047,
      "grad_norm": 1.4980125427246094,
      "learning_rate": 3.872008981825837e-05,
      "loss": 0.8054,
      "step": 4799
    },
    {
      "epoch": 1.5156299336911903,
      "grad_norm": 1.8494031429290771,
      "learning_rate": 3.8716581292540876e-05,
      "loss": 0.7925,
      "step": 4800
    },
    {
      "epoch": 1.5156299336911903,
      "eval_loss": 0.8110987544059753,
      "eval_runtime": 4.6593,
      "eval_samples_per_second": 534.413,
      "eval_steps_per_second": 4.507,
      "step": 4800
    },
    {
      "epoch": 1.515945689927376,
      "grad_norm": 1.557928442955017,
      "learning_rate": 3.871307276682338e-05,
      "loss": 0.8681,
      "step": 4801
    },
    {
      "epoch": 1.5162614461635617,
      "grad_norm": 1.622336745262146,
      "learning_rate": 3.870956424110589e-05,
      "loss": 0.8208,
      "step": 4802
    },
    {
      "epoch": 1.5165772023997475,
      "grad_norm": 1.659687876701355,
      "learning_rate": 3.87060557153884e-05,
      "loss": 0.8904,
      "step": 4803
    },
    {
      "epoch": 1.5168929586359332,
      "grad_norm": 1.3273651599884033,
      "learning_rate": 3.8702547189670903e-05,
      "loss": 0.8609,
      "step": 4804
    },
    {
      "epoch": 1.5172087148721187,
      "grad_norm": 1.5496708154678345,
      "learning_rate": 3.869903866395341e-05,
      "loss": 0.8863,
      "step": 4805
    },
    {
      "epoch": 1.5175244711083042,
      "grad_norm": 1.7568033933639526,
      "learning_rate": 3.869553013823591e-05,
      "loss": 0.8412,
      "step": 4806
    },
    {
      "epoch": 1.51784022734449,
      "grad_norm": 1.6295356750488281,
      "learning_rate": 3.869202161251842e-05,
      "loss": 0.8218,
      "step": 4807
    },
    {
      "epoch": 1.5181559835806757,
      "grad_norm": 1.701931357383728,
      "learning_rate": 3.8688513086800924e-05,
      "loss": 0.8661,
      "step": 4808
    },
    {
      "epoch": 1.5184717398168615,
      "grad_norm": 1.6576268672943115,
      "learning_rate": 3.868500456108343e-05,
      "loss": 0.7816,
      "step": 4809
    },
    {
      "epoch": 1.5187874960530472,
      "grad_norm": 1.6707134246826172,
      "learning_rate": 3.868149603536594e-05,
      "loss": 0.8597,
      "step": 4810
    },
    {
      "epoch": 1.5191032522892327,
      "grad_norm": 1.4048521518707275,
      "learning_rate": 3.867798750964845e-05,
      "loss": 0.8469,
      "step": 4811
    },
    {
      "epoch": 1.5194190085254182,
      "grad_norm": 1.7645864486694336,
      "learning_rate": 3.867447898393096e-05,
      "loss": 0.876,
      "step": 4812
    },
    {
      "epoch": 1.519734764761604,
      "grad_norm": 1.5582526922225952,
      "learning_rate": 3.8670970458213465e-05,
      "loss": 0.7997,
      "step": 4813
    },
    {
      "epoch": 1.5200505209977897,
      "grad_norm": 1.6120580434799194,
      "learning_rate": 3.866746193249597e-05,
      "loss": 0.8741,
      "step": 4814
    },
    {
      "epoch": 1.5203662772339754,
      "grad_norm": 1.383312702178955,
      "learning_rate": 3.866395340677847e-05,
      "loss": 0.8502,
      "step": 4815
    },
    {
      "epoch": 1.5206820334701612,
      "grad_norm": 1.6687371730804443,
      "learning_rate": 3.866044488106098e-05,
      "loss": 0.9071,
      "step": 4816
    },
    {
      "epoch": 1.5209977897063467,
      "grad_norm": 1.5321872234344482,
      "learning_rate": 3.8656936355343485e-05,
      "loss": 0.8424,
      "step": 4817
    },
    {
      "epoch": 1.5213135459425322,
      "grad_norm": 1.6236385107040405,
      "learning_rate": 3.865342782962599e-05,
      "loss": 0.834,
      "step": 4818
    },
    {
      "epoch": 1.521629302178718,
      "grad_norm": 1.3369722366333008,
      "learning_rate": 3.86499193039085e-05,
      "loss": 0.8235,
      "step": 4819
    },
    {
      "epoch": 1.5219450584149037,
      "grad_norm": 1.6542962789535522,
      "learning_rate": 3.8646410778191006e-05,
      "loss": 0.8836,
      "step": 4820
    },
    {
      "epoch": 1.5222608146510894,
      "grad_norm": 1.407938838005066,
      "learning_rate": 3.864290225247351e-05,
      "loss": 0.8271,
      "step": 4821
    },
    {
      "epoch": 1.5225765708872752,
      "grad_norm": 1.5832149982452393,
      "learning_rate": 3.863939372675602e-05,
      "loss": 0.8749,
      "step": 4822
    },
    {
      "epoch": 1.5228923271234607,
      "grad_norm": 1.691169023513794,
      "learning_rate": 3.8635885201038526e-05,
      "loss": 0.8767,
      "step": 4823
    },
    {
      "epoch": 1.5232080833596462,
      "grad_norm": 1.5094856023788452,
      "learning_rate": 3.863237667532103e-05,
      "loss": 0.7696,
      "step": 4824
    },
    {
      "epoch": 1.523523839595832,
      "grad_norm": 1.5296095609664917,
      "learning_rate": 3.862886814960354e-05,
      "loss": 0.84,
      "step": 4825
    },
    {
      "epoch": 1.5238395958320177,
      "grad_norm": 1.3044872283935547,
      "learning_rate": 3.862535962388604e-05,
      "loss": 0.7814,
      "step": 4826
    },
    {
      "epoch": 1.5241553520682034,
      "grad_norm": 1.4256916046142578,
      "learning_rate": 3.8621851098168547e-05,
      "loss": 0.7659,
      "step": 4827
    },
    {
      "epoch": 1.5244711083043891,
      "grad_norm": 1.3802675008773804,
      "learning_rate": 3.861834257245106e-05,
      "loss": 0.8175,
      "step": 4828
    },
    {
      "epoch": 1.5247868645405747,
      "grad_norm": 1.5873079299926758,
      "learning_rate": 3.861483404673357e-05,
      "loss": 0.837,
      "step": 4829
    },
    {
      "epoch": 1.5251026207767602,
      "grad_norm": 1.4891316890716553,
      "learning_rate": 3.8611325521016074e-05,
      "loss": 0.9449,
      "step": 4830
    },
    {
      "epoch": 1.525418377012946,
      "grad_norm": 1.6417300701141357,
      "learning_rate": 3.860781699529858e-05,
      "loss": 0.8701,
      "step": 4831
    },
    {
      "epoch": 1.5257341332491317,
      "grad_norm": 1.6516119241714478,
      "learning_rate": 3.860430846958109e-05,
      "loss": 0.8612,
      "step": 4832
    },
    {
      "epoch": 1.5260498894853174,
      "grad_norm": 1.851173758506775,
      "learning_rate": 3.8600799943863594e-05,
      "loss": 0.8642,
      "step": 4833
    },
    {
      "epoch": 1.5263656457215031,
      "grad_norm": 1.4026970863342285,
      "learning_rate": 3.8597291418146094e-05,
      "loss": 0.8839,
      "step": 4834
    },
    {
      "epoch": 1.5266814019576886,
      "grad_norm": 1.497975468635559,
      "learning_rate": 3.85937828924286e-05,
      "loss": 0.9092,
      "step": 4835
    },
    {
      "epoch": 1.5269971581938744,
      "grad_norm": 1.5079158544540405,
      "learning_rate": 3.859027436671111e-05,
      "loss": 0.7999,
      "step": 4836
    },
    {
      "epoch": 1.52731291443006,
      "grad_norm": 1.3726541996002197,
      "learning_rate": 3.8586765840993615e-05,
      "loss": 0.7582,
      "step": 4837
    },
    {
      "epoch": 1.5276286706662456,
      "grad_norm": 1.7559317350387573,
      "learning_rate": 3.858325731527612e-05,
      "loss": 0.9084,
      "step": 4838
    },
    {
      "epoch": 1.5279444269024314,
      "grad_norm": 1.439076542854309,
      "learning_rate": 3.857974878955863e-05,
      "loss": 0.7612,
      "step": 4839
    },
    {
      "epoch": 1.5282601831386171,
      "grad_norm": 1.7206900119781494,
      "learning_rate": 3.8576240263841135e-05,
      "loss": 0.8149,
      "step": 4840
    },
    {
      "epoch": 1.5285759393748026,
      "grad_norm": 1.3798173666000366,
      "learning_rate": 3.857273173812364e-05,
      "loss": 0.8686,
      "step": 4841
    },
    {
      "epoch": 1.5288916956109884,
      "grad_norm": 1.5692943334579468,
      "learning_rate": 3.856922321240615e-05,
      "loss": 0.7756,
      "step": 4842
    },
    {
      "epoch": 1.5292074518471739,
      "grad_norm": 1.6462910175323486,
      "learning_rate": 3.8565714686688656e-05,
      "loss": 0.7872,
      "step": 4843
    },
    {
      "epoch": 1.5295232080833596,
      "grad_norm": 1.565780520439148,
      "learning_rate": 3.856220616097116e-05,
      "loss": 0.862,
      "step": 4844
    },
    {
      "epoch": 1.5298389643195454,
      "grad_norm": 1.4828801155090332,
      "learning_rate": 3.855869763525366e-05,
      "loss": 0.8067,
      "step": 4845
    },
    {
      "epoch": 1.530154720555731,
      "grad_norm": 1.6439982652664185,
      "learning_rate": 3.8555189109536176e-05,
      "loss": 0.819,
      "step": 4846
    },
    {
      "epoch": 1.5304704767919166,
      "grad_norm": 1.600926399230957,
      "learning_rate": 3.855168058381868e-05,
      "loss": 0.8306,
      "step": 4847
    },
    {
      "epoch": 1.5307862330281023,
      "grad_norm": 1.5945336818695068,
      "learning_rate": 3.854817205810119e-05,
      "loss": 0.8227,
      "step": 4848
    },
    {
      "epoch": 1.5311019892642879,
      "grad_norm": 1.3774665594100952,
      "learning_rate": 3.8544663532383696e-05,
      "loss": 0.7651,
      "step": 4849
    },
    {
      "epoch": 1.5314177455004736,
      "grad_norm": 1.5834307670593262,
      "learning_rate": 3.85411550066662e-05,
      "loss": 0.8364,
      "step": 4850
    },
    {
      "epoch": 1.5317335017366593,
      "grad_norm": 1.3275132179260254,
      "learning_rate": 3.853764648094871e-05,
      "loss": 0.8148,
      "step": 4851
    },
    {
      "epoch": 1.532049257972845,
      "grad_norm": 1.2578389644622803,
      "learning_rate": 3.853413795523122e-05,
      "loss": 0.8013,
      "step": 4852
    },
    {
      "epoch": 1.5323650142090306,
      "grad_norm": 1.513444423675537,
      "learning_rate": 3.8530629429513724e-05,
      "loss": 0.7608,
      "step": 4853
    },
    {
      "epoch": 1.5326807704452163,
      "grad_norm": 1.6977546215057373,
      "learning_rate": 3.8527120903796224e-05,
      "loss": 0.8002,
      "step": 4854
    },
    {
      "epoch": 1.5329965266814018,
      "grad_norm": 1.3984348773956299,
      "learning_rate": 3.852361237807873e-05,
      "loss": 0.8168,
      "step": 4855
    },
    {
      "epoch": 1.5333122829175876,
      "grad_norm": 1.4406898021697998,
      "learning_rate": 3.852010385236124e-05,
      "loss": 0.8481,
      "step": 4856
    },
    {
      "epoch": 1.5336280391537733,
      "grad_norm": 1.763863205909729,
      "learning_rate": 3.8516595326643744e-05,
      "loss": 0.8513,
      "step": 4857
    },
    {
      "epoch": 1.533943795389959,
      "grad_norm": 1.5561888217926025,
      "learning_rate": 3.851308680092625e-05,
      "loss": 0.8916,
      "step": 4858
    },
    {
      "epoch": 1.5342595516261446,
      "grad_norm": 1.408018946647644,
      "learning_rate": 3.850957827520876e-05,
      "loss": 0.7696,
      "step": 4859
    },
    {
      "epoch": 1.5345753078623303,
      "grad_norm": 1.5636624097824097,
      "learning_rate": 3.8506069749491265e-05,
      "loss": 0.8089,
      "step": 4860
    },
    {
      "epoch": 1.5348910640985158,
      "grad_norm": 1.4446289539337158,
      "learning_rate": 3.850256122377377e-05,
      "loss": 0.8622,
      "step": 4861
    },
    {
      "epoch": 1.5352068203347016,
      "grad_norm": 1.4009462594985962,
      "learning_rate": 3.849905269805628e-05,
      "loss": 0.8314,
      "step": 4862
    },
    {
      "epoch": 1.5355225765708873,
      "grad_norm": 1.4945865869522095,
      "learning_rate": 3.8495544172338785e-05,
      "loss": 0.8383,
      "step": 4863
    },
    {
      "epoch": 1.535838332807073,
      "grad_norm": 1.83964204788208,
      "learning_rate": 3.849203564662129e-05,
      "loss": 0.9091,
      "step": 4864
    },
    {
      "epoch": 1.5361540890432586,
      "grad_norm": 1.7387256622314453,
      "learning_rate": 3.84885271209038e-05,
      "loss": 0.8322,
      "step": 4865
    },
    {
      "epoch": 1.5364698452794443,
      "grad_norm": 1.4760903120040894,
      "learning_rate": 3.8485018595186306e-05,
      "loss": 0.8031,
      "step": 4866
    },
    {
      "epoch": 1.5367856015156298,
      "grad_norm": 1.741107702255249,
      "learning_rate": 3.848151006946881e-05,
      "loss": 0.9569,
      "step": 4867
    },
    {
      "epoch": 1.5371013577518156,
      "grad_norm": 1.2210583686828613,
      "learning_rate": 3.847800154375132e-05,
      "loss": 0.7559,
      "step": 4868
    },
    {
      "epoch": 1.5374171139880013,
      "grad_norm": 1.118079662322998,
      "learning_rate": 3.8474493018033826e-05,
      "loss": 0.8018,
      "step": 4869
    },
    {
      "epoch": 1.537732870224187,
      "grad_norm": 1.625740647315979,
      "learning_rate": 3.847098449231633e-05,
      "loss": 0.8047,
      "step": 4870
    },
    {
      "epoch": 1.5380486264603725,
      "grad_norm": 1.7008445262908936,
      "learning_rate": 3.846747596659884e-05,
      "loss": 0.8583,
      "step": 4871
    },
    {
      "epoch": 1.5383643826965583,
      "grad_norm": 1.8076567649841309,
      "learning_rate": 3.8463967440881346e-05,
      "loss": 0.79,
      "step": 4872
    },
    {
      "epoch": 1.5386801389327438,
      "grad_norm": 1.4689642190933228,
      "learning_rate": 3.8460458915163846e-05,
      "loss": 0.8419,
      "step": 4873
    },
    {
      "epoch": 1.5389958951689295,
      "grad_norm": 1.6101197004318237,
      "learning_rate": 3.845695038944635e-05,
      "loss": 0.8728,
      "step": 4874
    },
    {
      "epoch": 1.5393116514051153,
      "grad_norm": 1.5637474060058594,
      "learning_rate": 3.845344186372886e-05,
      "loss": 0.7842,
      "step": 4875
    },
    {
      "epoch": 1.539627407641301,
      "grad_norm": 1.3783553838729858,
      "learning_rate": 3.844993333801137e-05,
      "loss": 0.7856,
      "step": 4876
    },
    {
      "epoch": 1.5399431638774865,
      "grad_norm": 1.9041773080825806,
      "learning_rate": 3.8446424812293874e-05,
      "loss": 0.8969,
      "step": 4877
    },
    {
      "epoch": 1.5402589201136723,
      "grad_norm": 1.4730292558670044,
      "learning_rate": 3.844291628657638e-05,
      "loss": 0.7389,
      "step": 4878
    },
    {
      "epoch": 1.5405746763498578,
      "grad_norm": 2.541990041732788,
      "learning_rate": 3.843940776085889e-05,
      "loss": 0.789,
      "step": 4879
    },
    {
      "epoch": 1.5408904325860435,
      "grad_norm": 1.430769443511963,
      "learning_rate": 3.8435899235141394e-05,
      "loss": 0.8196,
      "step": 4880
    },
    {
      "epoch": 1.5412061888222293,
      "grad_norm": 1.8230094909667969,
      "learning_rate": 3.843239070942391e-05,
      "loss": 0.9184,
      "step": 4881
    },
    {
      "epoch": 1.541521945058415,
      "grad_norm": 1.8350028991699219,
      "learning_rate": 3.842888218370641e-05,
      "loss": 0.8045,
      "step": 4882
    },
    {
      "epoch": 1.5418377012946005,
      "grad_norm": 1.7163348197937012,
      "learning_rate": 3.8425373657988915e-05,
      "loss": 0.8752,
      "step": 4883
    },
    {
      "epoch": 1.5421534575307863,
      "grad_norm": 1.723140835762024,
      "learning_rate": 3.842186513227142e-05,
      "loss": 0.7638,
      "step": 4884
    },
    {
      "epoch": 1.5424692137669718,
      "grad_norm": 1.6653889417648315,
      "learning_rate": 3.841835660655393e-05,
      "loss": 0.8867,
      "step": 4885
    },
    {
      "epoch": 1.5427849700031575,
      "grad_norm": 1.462311029434204,
      "learning_rate": 3.8414848080836435e-05,
      "loss": 0.7987,
      "step": 4886
    },
    {
      "epoch": 1.5431007262393432,
      "grad_norm": 1.8944953680038452,
      "learning_rate": 3.841133955511894e-05,
      "loss": 0.8993,
      "step": 4887
    },
    {
      "epoch": 1.543416482475529,
      "grad_norm": 1.9390504360198975,
      "learning_rate": 3.840783102940145e-05,
      "loss": 0.8322,
      "step": 4888
    },
    {
      "epoch": 1.5437322387117147,
      "grad_norm": 2.089503288269043,
      "learning_rate": 3.8404322503683955e-05,
      "loss": 0.8294,
      "step": 4889
    },
    {
      "epoch": 1.5440479949479002,
      "grad_norm": 1.7328399419784546,
      "learning_rate": 3.840081397796646e-05,
      "loss": 0.9147,
      "step": 4890
    },
    {
      "epoch": 1.5443637511840858,
      "grad_norm": 1.5798628330230713,
      "learning_rate": 3.839730545224897e-05,
      "loss": 0.8407,
      "step": 4891
    },
    {
      "epoch": 1.5446795074202715,
      "grad_norm": 1.6291977167129517,
      "learning_rate": 3.839379692653147e-05,
      "loss": 0.8323,
      "step": 4892
    },
    {
      "epoch": 1.5449952636564572,
      "grad_norm": 2.065135955810547,
      "learning_rate": 3.8390288400813976e-05,
      "loss": 0.9002,
      "step": 4893
    },
    {
      "epoch": 1.545311019892643,
      "grad_norm": 1.759584903717041,
      "learning_rate": 3.838677987509648e-05,
      "loss": 0.8735,
      "step": 4894
    },
    {
      "epoch": 1.5456267761288287,
      "grad_norm": 1.3735668659210205,
      "learning_rate": 3.838327134937899e-05,
      "loss": 0.7847,
      "step": 4895
    },
    {
      "epoch": 1.5459425323650142,
      "grad_norm": 1.7690284252166748,
      "learning_rate": 3.8379762823661496e-05,
      "loss": 0.8311,
      "step": 4896
    },
    {
      "epoch": 1.5462582886011997,
      "grad_norm": 1.5446250438690186,
      "learning_rate": 3.8376254297944e-05,
      "loss": 0.8153,
      "step": 4897
    },
    {
      "epoch": 1.5465740448373855,
      "grad_norm": 1.427620530128479,
      "learning_rate": 3.837274577222652e-05,
      "loss": 0.804,
      "step": 4898
    },
    {
      "epoch": 1.5468898010735712,
      "grad_norm": 1.6636675596237183,
      "learning_rate": 3.8369237246509024e-05,
      "loss": 0.9249,
      "step": 4899
    },
    {
      "epoch": 1.547205557309757,
      "grad_norm": 1.5591552257537842,
      "learning_rate": 3.836572872079153e-05,
      "loss": 0.8704,
      "step": 4900
    },
    {
      "epoch": 1.5475213135459427,
      "grad_norm": 1.444417119026184,
      "learning_rate": 3.836222019507403e-05,
      "loss": 0.7824,
      "step": 4901
    },
    {
      "epoch": 1.5478370697821282,
      "grad_norm": 1.5822445154190063,
      "learning_rate": 3.835871166935654e-05,
      "loss": 0.8326,
      "step": 4902
    },
    {
      "epoch": 1.5481528260183137,
      "grad_norm": 1.4268358945846558,
      "learning_rate": 3.8355203143639044e-05,
      "loss": 0.8668,
      "step": 4903
    },
    {
      "epoch": 1.5484685822544995,
      "grad_norm": 1.5124306678771973,
      "learning_rate": 3.835169461792155e-05,
      "loss": 0.8026,
      "step": 4904
    },
    {
      "epoch": 1.5487843384906852,
      "grad_norm": 1.392275333404541,
      "learning_rate": 3.834818609220406e-05,
      "loss": 0.8511,
      "step": 4905
    },
    {
      "epoch": 1.549100094726871,
      "grad_norm": 1.4028733968734741,
      "learning_rate": 3.8344677566486564e-05,
      "loss": 0.7895,
      "step": 4906
    },
    {
      "epoch": 1.5494158509630567,
      "grad_norm": 1.4605664014816284,
      "learning_rate": 3.834116904076907e-05,
      "loss": 0.7438,
      "step": 4907
    },
    {
      "epoch": 1.5497316071992422,
      "grad_norm": 1.6726489067077637,
      "learning_rate": 3.833766051505158e-05,
      "loss": 0.8396,
      "step": 4908
    },
    {
      "epoch": 1.5500473634354277,
      "grad_norm": 1.4382274150848389,
      "learning_rate": 3.8334151989334085e-05,
      "loss": 0.8107,
      "step": 4909
    },
    {
      "epoch": 1.5503631196716134,
      "grad_norm": 1.336544394493103,
      "learning_rate": 3.833064346361659e-05,
      "loss": 0.838,
      "step": 4910
    },
    {
      "epoch": 1.5506788759077992,
      "grad_norm": 1.5152784585952759,
      "learning_rate": 3.832713493789909e-05,
      "loss": 0.8864,
      "step": 4911
    },
    {
      "epoch": 1.550994632143985,
      "grad_norm": 1.4554141759872437,
      "learning_rate": 3.83236264121816e-05,
      "loss": 0.7246,
      "step": 4912
    },
    {
      "epoch": 1.5513103883801707,
      "grad_norm": 1.7215549945831299,
      "learning_rate": 3.8320117886464105e-05,
      "loss": 0.8249,
      "step": 4913
    },
    {
      "epoch": 1.5516261446163562,
      "grad_norm": 1.6354501247406006,
      "learning_rate": 3.831660936074661e-05,
      "loss": 0.7833,
      "step": 4914
    },
    {
      "epoch": 1.5519419008525417,
      "grad_norm": 1.338503122329712,
      "learning_rate": 3.831310083502912e-05,
      "loss": 0.8166,
      "step": 4915
    },
    {
      "epoch": 1.5522576570887274,
      "grad_norm": 1.67670738697052,
      "learning_rate": 3.830959230931163e-05,
      "loss": 0.8622,
      "step": 4916
    },
    {
      "epoch": 1.5525734133249132,
      "grad_norm": 1.6159229278564453,
      "learning_rate": 3.830608378359414e-05,
      "loss": 0.7424,
      "step": 4917
    },
    {
      "epoch": 1.552889169561099,
      "grad_norm": 1.4242947101593018,
      "learning_rate": 3.8302575257876646e-05,
      "loss": 0.8504,
      "step": 4918
    },
    {
      "epoch": 1.5532049257972846,
      "grad_norm": 1.7640541791915894,
      "learning_rate": 3.829906673215915e-05,
      "loss": 0.8038,
      "step": 4919
    },
    {
      "epoch": 1.5535206820334702,
      "grad_norm": 1.6106517314910889,
      "learning_rate": 3.829555820644165e-05,
      "loss": 0.8231,
      "step": 4920
    },
    {
      "epoch": 1.5538364382696557,
      "grad_norm": 2.3097479343414307,
      "learning_rate": 3.829204968072416e-05,
      "loss": 0.8503,
      "step": 4921
    },
    {
      "epoch": 1.5541521945058414,
      "grad_norm": 1.8829755783081055,
      "learning_rate": 3.828854115500667e-05,
      "loss": 0.836,
      "step": 4922
    },
    {
      "epoch": 1.5544679507420271,
      "grad_norm": 1.7814255952835083,
      "learning_rate": 3.8285032629289174e-05,
      "loss": 0.8175,
      "step": 4923
    },
    {
      "epoch": 1.5547837069782129,
      "grad_norm": 1.586068868637085,
      "learning_rate": 3.828152410357168e-05,
      "loss": 0.7511,
      "step": 4924
    },
    {
      "epoch": 1.5550994632143986,
      "grad_norm": 1.8101311922073364,
      "learning_rate": 3.827801557785419e-05,
      "loss": 0.8282,
      "step": 4925
    },
    {
      "epoch": 1.5554152194505841,
      "grad_norm": 1.758104920387268,
      "learning_rate": 3.8274507052136694e-05,
      "loss": 0.8581,
      "step": 4926
    },
    {
      "epoch": 1.5557309756867697,
      "grad_norm": 1.6319857835769653,
      "learning_rate": 3.82709985264192e-05,
      "loss": 0.7701,
      "step": 4927
    },
    {
      "epoch": 1.5560467319229554,
      "grad_norm": 1.7719489336013794,
      "learning_rate": 3.826749000070171e-05,
      "loss": 0.8708,
      "step": 4928
    },
    {
      "epoch": 1.5563624881591411,
      "grad_norm": 1.6629509925842285,
      "learning_rate": 3.8263981474984214e-05,
      "loss": 0.8087,
      "step": 4929
    },
    {
      "epoch": 1.5566782443953269,
      "grad_norm": 1.5563396215438843,
      "learning_rate": 3.826047294926672e-05,
      "loss": 0.8158,
      "step": 4930
    },
    {
      "epoch": 1.5569940006315126,
      "grad_norm": 1.9165459871292114,
      "learning_rate": 3.825696442354922e-05,
      "loss": 0.8418,
      "step": 4931
    },
    {
      "epoch": 1.5573097568676981,
      "grad_norm": 1.5768400430679321,
      "learning_rate": 3.825345589783173e-05,
      "loss": 0.8196,
      "step": 4932
    },
    {
      "epoch": 1.5576255131038836,
      "grad_norm": 1.7514007091522217,
      "learning_rate": 3.824994737211424e-05,
      "loss": 0.8457,
      "step": 4933
    },
    {
      "epoch": 1.5579412693400694,
      "grad_norm": 1.9323853254318237,
      "learning_rate": 3.824643884639675e-05,
      "loss": 0.8129,
      "step": 4934
    },
    {
      "epoch": 1.5582570255762551,
      "grad_norm": 1.3419442176818848,
      "learning_rate": 3.8242930320679255e-05,
      "loss": 0.8153,
      "step": 4935
    },
    {
      "epoch": 1.5585727818124409,
      "grad_norm": 1.6170603036880493,
      "learning_rate": 3.823942179496176e-05,
      "loss": 0.8996,
      "step": 4936
    },
    {
      "epoch": 1.5588885380486266,
      "grad_norm": 1.7915517091751099,
      "learning_rate": 3.823591326924427e-05,
      "loss": 0.7959,
      "step": 4937
    },
    {
      "epoch": 1.559204294284812,
      "grad_norm": 1.457487940788269,
      "learning_rate": 3.8232404743526776e-05,
      "loss": 0.8206,
      "step": 4938
    },
    {
      "epoch": 1.5595200505209976,
      "grad_norm": 1.595802903175354,
      "learning_rate": 3.8228896217809276e-05,
      "loss": 0.8554,
      "step": 4939
    },
    {
      "epoch": 1.5598358067571834,
      "grad_norm": 1.6154826879501343,
      "learning_rate": 3.822538769209178e-05,
      "loss": 0.797,
      "step": 4940
    },
    {
      "epoch": 1.560151562993369,
      "grad_norm": 1.750280499458313,
      "learning_rate": 3.822187916637429e-05,
      "loss": 0.8466,
      "step": 4941
    },
    {
      "epoch": 1.5604673192295548,
      "grad_norm": 1.5777792930603027,
      "learning_rate": 3.8218370640656796e-05,
      "loss": 0.829,
      "step": 4942
    },
    {
      "epoch": 1.5607830754657406,
      "grad_norm": 1.7009022235870361,
      "learning_rate": 3.82148621149393e-05,
      "loss": 0.8383,
      "step": 4943
    },
    {
      "epoch": 1.561098831701926,
      "grad_norm": 1.4791483879089355,
      "learning_rate": 3.821135358922181e-05,
      "loss": 0.8208,
      "step": 4944
    },
    {
      "epoch": 1.5614145879381118,
      "grad_norm": 1.7833298444747925,
      "learning_rate": 3.820784506350432e-05,
      "loss": 0.9008,
      "step": 4945
    },
    {
      "epoch": 1.5617303441742973,
      "grad_norm": 1.6473500728607178,
      "learning_rate": 3.8204336537786823e-05,
      "loss": 0.7609,
      "step": 4946
    },
    {
      "epoch": 1.562046100410483,
      "grad_norm": 1.7140107154846191,
      "learning_rate": 3.820082801206933e-05,
      "loss": 0.8776,
      "step": 4947
    },
    {
      "epoch": 1.5623618566466688,
      "grad_norm": 1.923792839050293,
      "learning_rate": 3.819731948635184e-05,
      "loss": 0.8209,
      "step": 4948
    },
    {
      "epoch": 1.5626776128828546,
      "grad_norm": 1.8174283504486084,
      "learning_rate": 3.8193810960634344e-05,
      "loss": 0.8502,
      "step": 4949
    },
    {
      "epoch": 1.56299336911904,
      "grad_norm": 1.7268078327178955,
      "learning_rate": 3.8190302434916844e-05,
      "loss": 0.8223,
      "step": 4950
    },
    {
      "epoch": 1.5633091253552258,
      "grad_norm": 1.6399173736572266,
      "learning_rate": 3.818679390919936e-05,
      "loss": 0.7841,
      "step": 4951
    },
    {
      "epoch": 1.5636248815914113,
      "grad_norm": 1.6771737337112427,
      "learning_rate": 3.8183285383481864e-05,
      "loss": 0.8874,
      "step": 4952
    },
    {
      "epoch": 1.563940637827597,
      "grad_norm": 1.7858998775482178,
      "learning_rate": 3.817977685776437e-05,
      "loss": 0.8827,
      "step": 4953
    },
    {
      "epoch": 1.5642563940637828,
      "grad_norm": 1.5930967330932617,
      "learning_rate": 3.817626833204688e-05,
      "loss": 0.8519,
      "step": 4954
    },
    {
      "epoch": 1.5645721502999685,
      "grad_norm": 1.6255393028259277,
      "learning_rate": 3.8172759806329385e-05,
      "loss": 0.8997,
      "step": 4955
    },
    {
      "epoch": 1.564887906536154,
      "grad_norm": 1.5229499340057373,
      "learning_rate": 3.816925128061189e-05,
      "loss": 0.8063,
      "step": 4956
    },
    {
      "epoch": 1.5652036627723398,
      "grad_norm": 1.7632403373718262,
      "learning_rate": 3.81657427548944e-05,
      "loss": 0.8765,
      "step": 4957
    },
    {
      "epoch": 1.5655194190085253,
      "grad_norm": 1.5409215688705444,
      "learning_rate": 3.8162234229176905e-05,
      "loss": 0.8932,
      "step": 4958
    },
    {
      "epoch": 1.565835175244711,
      "grad_norm": 1.9288300275802612,
      "learning_rate": 3.8158725703459405e-05,
      "loss": 0.8017,
      "step": 4959
    },
    {
      "epoch": 1.5661509314808968,
      "grad_norm": 1.5460902452468872,
      "learning_rate": 3.815521717774191e-05,
      "loss": 0.7568,
      "step": 4960
    },
    {
      "epoch": 1.5664666877170825,
      "grad_norm": 1.5431033372879028,
      "learning_rate": 3.815170865202442e-05,
      "loss": 0.8018,
      "step": 4961
    },
    {
      "epoch": 1.566782443953268,
      "grad_norm": 1.6795482635498047,
      "learning_rate": 3.8148200126306926e-05,
      "loss": 0.8566,
      "step": 4962
    },
    {
      "epoch": 1.5670982001894538,
      "grad_norm": 1.6182690858840942,
      "learning_rate": 3.814469160058943e-05,
      "loss": 0.8196,
      "step": 4963
    },
    {
      "epoch": 1.5674139564256393,
      "grad_norm": 1.7207424640655518,
      "learning_rate": 3.814118307487194e-05,
      "loss": 0.8077,
      "step": 4964
    },
    {
      "epoch": 1.567729712661825,
      "grad_norm": 1.5020034313201904,
      "learning_rate": 3.8137674549154446e-05,
      "loss": 0.8165,
      "step": 4965
    },
    {
      "epoch": 1.5680454688980108,
      "grad_norm": 1.807464361190796,
      "learning_rate": 3.813416602343695e-05,
      "loss": 0.8593,
      "step": 4966
    },
    {
      "epoch": 1.5683612251341965,
      "grad_norm": 1.4208084344863892,
      "learning_rate": 3.813065749771946e-05,
      "loss": 0.8063,
      "step": 4967
    },
    {
      "epoch": 1.568676981370382,
      "grad_norm": 1.4886223077774048,
      "learning_rate": 3.8127148972001967e-05,
      "loss": 0.8273,
      "step": 4968
    },
    {
      "epoch": 1.5689927376065678,
      "grad_norm": 1.5626285076141357,
      "learning_rate": 3.812364044628447e-05,
      "loss": 0.7926,
      "step": 4969
    },
    {
      "epoch": 1.5693084938427533,
      "grad_norm": 1.5983699560165405,
      "learning_rate": 3.812013192056698e-05,
      "loss": 0.7899,
      "step": 4970
    },
    {
      "epoch": 1.569624250078939,
      "grad_norm": 1.7395871877670288,
      "learning_rate": 3.811662339484949e-05,
      "loss": 0.875,
      "step": 4971
    },
    {
      "epoch": 1.5699400063151248,
      "grad_norm": 1.6556960344314575,
      "learning_rate": 3.8113114869131994e-05,
      "loss": 0.8242,
      "step": 4972
    },
    {
      "epoch": 1.5702557625513105,
      "grad_norm": 1.4643417596817017,
      "learning_rate": 3.81096063434145e-05,
      "loss": 0.8208,
      "step": 4973
    },
    {
      "epoch": 1.570571518787496,
      "grad_norm": 1.9650533199310303,
      "learning_rate": 3.810609781769701e-05,
      "loss": 0.8254,
      "step": 4974
    },
    {
      "epoch": 1.5708872750236818,
      "grad_norm": 1.830865740776062,
      "learning_rate": 3.8102589291979514e-05,
      "loss": 0.8734,
      "step": 4975
    },
    {
      "epoch": 1.5712030312598673,
      "grad_norm": 1.3290743827819824,
      "learning_rate": 3.809908076626202e-05,
      "loss": 0.8417,
      "step": 4976
    },
    {
      "epoch": 1.571518787496053,
      "grad_norm": 1.42648446559906,
      "learning_rate": 3.809557224054453e-05,
      "loss": 0.8071,
      "step": 4977
    },
    {
      "epoch": 1.5718345437322387,
      "grad_norm": 1.6476466655731201,
      "learning_rate": 3.809206371482703e-05,
      "loss": 0.7837,
      "step": 4978
    },
    {
      "epoch": 1.5721502999684245,
      "grad_norm": 1.7402434349060059,
      "learning_rate": 3.8088555189109535e-05,
      "loss": 0.8415,
      "step": 4979
    },
    {
      "epoch": 1.57246605620461,
      "grad_norm": 1.7241671085357666,
      "learning_rate": 3.808504666339204e-05,
      "loss": 0.7678,
      "step": 4980
    },
    {
      "epoch": 1.5727818124407957,
      "grad_norm": 1.4532774686813354,
      "learning_rate": 3.808153813767455e-05,
      "loss": 0.8223,
      "step": 4981
    },
    {
      "epoch": 1.5730975686769813,
      "grad_norm": 1.4341390132904053,
      "learning_rate": 3.8078029611957055e-05,
      "loss": 0.8575,
      "step": 4982
    },
    {
      "epoch": 1.573413324913167,
      "grad_norm": 1.5381768941879272,
      "learning_rate": 3.807452108623956e-05,
      "loss": 0.8303,
      "step": 4983
    },
    {
      "epoch": 1.5737290811493527,
      "grad_norm": 1.4578547477722168,
      "learning_rate": 3.807101256052207e-05,
      "loss": 0.9271,
      "step": 4984
    },
    {
      "epoch": 1.5740448373855385,
      "grad_norm": 1.2874937057495117,
      "learning_rate": 3.8067504034804576e-05,
      "loss": 0.788,
      "step": 4985
    },
    {
      "epoch": 1.574360593621724,
      "grad_norm": 2.298954963684082,
      "learning_rate": 3.806399550908709e-05,
      "loss": 0.8651,
      "step": 4986
    },
    {
      "epoch": 1.5746763498579097,
      "grad_norm": 1.568679690361023,
      "learning_rate": 3.806048698336959e-05,
      "loss": 0.8113,
      "step": 4987
    },
    {
      "epoch": 1.5749921060940952,
      "grad_norm": 1.7662888765335083,
      "learning_rate": 3.8056978457652096e-05,
      "loss": 0.8635,
      "step": 4988
    },
    {
      "epoch": 1.575307862330281,
      "grad_norm": 1.678344488143921,
      "learning_rate": 3.80534699319346e-05,
      "loss": 0.8241,
      "step": 4989
    },
    {
      "epoch": 1.5756236185664667,
      "grad_norm": 1.4485492706298828,
      "learning_rate": 3.804996140621711e-05,
      "loss": 0.8541,
      "step": 4990
    },
    {
      "epoch": 1.5759393748026524,
      "grad_norm": 1.4470220804214478,
      "learning_rate": 3.8046452880499616e-05,
      "loss": 0.7878,
      "step": 4991
    },
    {
      "epoch": 1.576255131038838,
      "grad_norm": 1.4324150085449219,
      "learning_rate": 3.804294435478212e-05,
      "loss": 0.8558,
      "step": 4992
    },
    {
      "epoch": 1.5765708872750237,
      "grad_norm": 1.5005615949630737,
      "learning_rate": 3.803943582906463e-05,
      "loss": 0.841,
      "step": 4993
    },
    {
      "epoch": 1.5768866435112092,
      "grad_norm": 1.5563926696777344,
      "learning_rate": 3.803592730334714e-05,
      "loss": 0.8386,
      "step": 4994
    },
    {
      "epoch": 1.577202399747395,
      "grad_norm": 1.789993166923523,
      "learning_rate": 3.8032418777629644e-05,
      "loss": 0.8893,
      "step": 4995
    },
    {
      "epoch": 1.5775181559835807,
      "grad_norm": 1.5692516565322876,
      "learning_rate": 3.802891025191215e-05,
      "loss": 0.756,
      "step": 4996
    },
    {
      "epoch": 1.5778339122197664,
      "grad_norm": 1.548466444015503,
      "learning_rate": 3.802540172619465e-05,
      "loss": 0.7611,
      "step": 4997
    },
    {
      "epoch": 1.5781496684559522,
      "grad_norm": 1.4604586362838745,
      "learning_rate": 3.802189320047716e-05,
      "loss": 0.8184,
      "step": 4998
    },
    {
      "epoch": 1.5784654246921377,
      "grad_norm": 1.3909075260162354,
      "learning_rate": 3.8018384674759664e-05,
      "loss": 0.8218,
      "step": 4999
    },
    {
      "epoch": 1.5787811809283232,
      "grad_norm": 1.4840290546417236,
      "learning_rate": 3.801487614904217e-05,
      "loss": 0.8134,
      "step": 5000
    },
    {
      "epoch": 1.5787811809283232,
      "eval_loss": 0.8073826432228088,
      "eval_runtime": 4.6201,
      "eval_samples_per_second": 538.951,
      "eval_steps_per_second": 4.545,
      "step": 5000
    },
    {
      "epoch": 1.579096937164509,
      "grad_norm": 1.484700322151184,
      "learning_rate": 3.801136762332468e-05,
      "loss": 0.8139,
      "step": 5001
    },
    {
      "epoch": 1.5794126934006947,
      "grad_norm": 1.5958552360534668,
      "learning_rate": 3.8007859097607185e-05,
      "loss": 0.8425,
      "step": 5002
    },
    {
      "epoch": 1.5797284496368804,
      "grad_norm": 1.5217584371566772,
      "learning_rate": 3.80043505718897e-05,
      "loss": 0.8005,
      "step": 5003
    },
    {
      "epoch": 1.5800442058730662,
      "grad_norm": 1.4056402444839478,
      "learning_rate": 3.8000842046172205e-05,
      "loss": 0.7269,
      "step": 5004
    },
    {
      "epoch": 1.5803599621092517,
      "grad_norm": 1.6808857917785645,
      "learning_rate": 3.799733352045471e-05,
      "loss": 0.8195,
      "step": 5005
    },
    {
      "epoch": 1.5806757183454372,
      "grad_norm": 1.5727094411849976,
      "learning_rate": 3.799382499473721e-05,
      "loss": 0.8029,
      "step": 5006
    },
    {
      "epoch": 1.580991474581623,
      "grad_norm": 1.8352006673812866,
      "learning_rate": 3.799031646901972e-05,
      "loss": 0.8656,
      "step": 5007
    },
    {
      "epoch": 1.5813072308178087,
      "grad_norm": 1.7535638809204102,
      "learning_rate": 3.7986807943302226e-05,
      "loss": 0.8522,
      "step": 5008
    },
    {
      "epoch": 1.5816229870539944,
      "grad_norm": 1.423492193222046,
      "learning_rate": 3.798329941758473e-05,
      "loss": 0.7918,
      "step": 5009
    },
    {
      "epoch": 1.5819387432901801,
      "grad_norm": 1.5691829919815063,
      "learning_rate": 3.797979089186724e-05,
      "loss": 0.7807,
      "step": 5010
    },
    {
      "epoch": 1.5822544995263657,
      "grad_norm": 1.5819690227508545,
      "learning_rate": 3.7976282366149746e-05,
      "loss": 0.7943,
      "step": 5011
    },
    {
      "epoch": 1.5825702557625512,
      "grad_norm": 1.5769157409667969,
      "learning_rate": 3.797277384043225e-05,
      "loss": 0.8363,
      "step": 5012
    },
    {
      "epoch": 1.582886011998737,
      "grad_norm": 1.8106718063354492,
      "learning_rate": 3.796926531471476e-05,
      "loss": 0.8013,
      "step": 5013
    },
    {
      "epoch": 1.5832017682349226,
      "grad_norm": 1.6587705612182617,
      "learning_rate": 3.7965756788997266e-05,
      "loss": 0.8678,
      "step": 5014
    },
    {
      "epoch": 1.5835175244711084,
      "grad_norm": 1.5615488290786743,
      "learning_rate": 3.796224826327977e-05,
      "loss": 0.8383,
      "step": 5015
    },
    {
      "epoch": 1.5838332807072941,
      "grad_norm": 1.7106475830078125,
      "learning_rate": 3.795873973756227e-05,
      "loss": 0.7977,
      "step": 5016
    },
    {
      "epoch": 1.5841490369434796,
      "grad_norm": 1.8201978206634521,
      "learning_rate": 3.795523121184478e-05,
      "loss": 0.8553,
      "step": 5017
    },
    {
      "epoch": 1.5844647931796652,
      "grad_norm": 1.7020034790039062,
      "learning_rate": 3.795172268612729e-05,
      "loss": 0.8177,
      "step": 5018
    },
    {
      "epoch": 1.584780549415851,
      "grad_norm": 1.7807979583740234,
      "learning_rate": 3.7948214160409794e-05,
      "loss": 0.8202,
      "step": 5019
    },
    {
      "epoch": 1.5850963056520366,
      "grad_norm": 1.721632480621338,
      "learning_rate": 3.794470563469231e-05,
      "loss": 0.8756,
      "step": 5020
    },
    {
      "epoch": 1.5854120618882224,
      "grad_norm": 1.5258888006210327,
      "learning_rate": 3.7941197108974814e-05,
      "loss": 0.7723,
      "step": 5021
    },
    {
      "epoch": 1.585727818124408,
      "grad_norm": 1.6635676622390747,
      "learning_rate": 3.793768858325732e-05,
      "loss": 0.8513,
      "step": 5022
    },
    {
      "epoch": 1.5860435743605936,
      "grad_norm": 1.4835783243179321,
      "learning_rate": 3.793418005753983e-05,
      "loss": 0.8187,
      "step": 5023
    },
    {
      "epoch": 1.5863593305967791,
      "grad_norm": 1.6236728429794312,
      "learning_rate": 3.7930671531822335e-05,
      "loss": 0.779,
      "step": 5024
    },
    {
      "epoch": 1.5866750868329649,
      "grad_norm": 1.3114032745361328,
      "learning_rate": 3.7927163006104835e-05,
      "loss": 0.8061,
      "step": 5025
    },
    {
      "epoch": 1.5869908430691506,
      "grad_norm": 1.8985062837600708,
      "learning_rate": 3.792365448038734e-05,
      "loss": 0.8698,
      "step": 5026
    },
    {
      "epoch": 1.5873065993053364,
      "grad_norm": 1.5194205045700073,
      "learning_rate": 3.792014595466985e-05,
      "loss": 0.8164,
      "step": 5027
    },
    {
      "epoch": 1.587622355541522,
      "grad_norm": 1.5948476791381836,
      "learning_rate": 3.7916637428952355e-05,
      "loss": 0.8002,
      "step": 5028
    },
    {
      "epoch": 1.5879381117777076,
      "grad_norm": 1.44679856300354,
      "learning_rate": 3.791312890323486e-05,
      "loss": 0.7837,
      "step": 5029
    },
    {
      "epoch": 1.5882538680138931,
      "grad_norm": 1.758924126625061,
      "learning_rate": 3.790962037751737e-05,
      "loss": 0.7766,
      "step": 5030
    },
    {
      "epoch": 1.5885696242500789,
      "grad_norm": 1.6094425916671753,
      "learning_rate": 3.7906111851799875e-05,
      "loss": 0.9034,
      "step": 5031
    },
    {
      "epoch": 1.5888853804862646,
      "grad_norm": 1.5100337266921997,
      "learning_rate": 3.790260332608238e-05,
      "loss": 0.8815,
      "step": 5032
    },
    {
      "epoch": 1.5892011367224503,
      "grad_norm": 1.6240681409835815,
      "learning_rate": 3.789909480036489e-05,
      "loss": 0.8815,
      "step": 5033
    },
    {
      "epoch": 1.589516892958636,
      "grad_norm": 1.452844500541687,
      "learning_rate": 3.7895586274647396e-05,
      "loss": 0.7844,
      "step": 5034
    },
    {
      "epoch": 1.5898326491948216,
      "grad_norm": 1.41586172580719,
      "learning_rate": 3.78920777489299e-05,
      "loss": 0.7979,
      "step": 5035
    },
    {
      "epoch": 1.590148405431007,
      "grad_norm": 1.5080628395080566,
      "learning_rate": 3.78885692232124e-05,
      "loss": 0.8231,
      "step": 5036
    },
    {
      "epoch": 1.5904641616671928,
      "grad_norm": 2.23954176902771,
      "learning_rate": 3.788506069749491e-05,
      "loss": 0.8391,
      "step": 5037
    },
    {
      "epoch": 1.5907799179033786,
      "grad_norm": 1.5186479091644287,
      "learning_rate": 3.788155217177742e-05,
      "loss": 0.771,
      "step": 5038
    },
    {
      "epoch": 1.5910956741395643,
      "grad_norm": 1.6261610984802246,
      "learning_rate": 3.787804364605993e-05,
      "loss": 0.8248,
      "step": 5039
    },
    {
      "epoch": 1.59141143037575,
      "grad_norm": 1.482157826423645,
      "learning_rate": 3.787453512034244e-05,
      "loss": 0.8222,
      "step": 5040
    },
    {
      "epoch": 1.5917271866119356,
      "grad_norm": 1.705585241317749,
      "learning_rate": 3.7871026594624944e-05,
      "loss": 0.8451,
      "step": 5041
    },
    {
      "epoch": 1.592042942848121,
      "grad_norm": 1.7123255729675293,
      "learning_rate": 3.786751806890745e-05,
      "loss": 0.8683,
      "step": 5042
    },
    {
      "epoch": 1.5923586990843068,
      "grad_norm": 1.2824320793151855,
      "learning_rate": 3.786400954318996e-05,
      "loss": 0.7718,
      "step": 5043
    },
    {
      "epoch": 1.5926744553204926,
      "grad_norm": 1.7812381982803345,
      "learning_rate": 3.786050101747246e-05,
      "loss": 0.8025,
      "step": 5044
    },
    {
      "epoch": 1.5929902115566783,
      "grad_norm": 1.607507348060608,
      "learning_rate": 3.7856992491754964e-05,
      "loss": 0.7501,
      "step": 5045
    },
    {
      "epoch": 1.593305967792864,
      "grad_norm": 1.677405834197998,
      "learning_rate": 3.785348396603747e-05,
      "loss": 0.8025,
      "step": 5046
    },
    {
      "epoch": 1.5936217240290496,
      "grad_norm": 1.595523476600647,
      "learning_rate": 3.784997544031998e-05,
      "loss": 0.8043,
      "step": 5047
    },
    {
      "epoch": 1.593937480265235,
      "grad_norm": 1.6912221908569336,
      "learning_rate": 3.7846466914602485e-05,
      "loss": 0.8321,
      "step": 5048
    },
    {
      "epoch": 1.5942532365014208,
      "grad_norm": 1.647857427597046,
      "learning_rate": 3.784295838888499e-05,
      "loss": 0.8021,
      "step": 5049
    },
    {
      "epoch": 1.5945689927376065,
      "grad_norm": 1.6624709367752075,
      "learning_rate": 3.78394498631675e-05,
      "loss": 0.8422,
      "step": 5050
    },
    {
      "epoch": 1.5948847489737923,
      "grad_norm": 1.5973066091537476,
      "learning_rate": 3.7835941337450005e-05,
      "loss": 0.8458,
      "step": 5051
    },
    {
      "epoch": 1.595200505209978,
      "grad_norm": 1.7343462705612183,
      "learning_rate": 3.783243281173251e-05,
      "loss": 0.7887,
      "step": 5052
    },
    {
      "epoch": 1.5955162614461635,
      "grad_norm": 1.7077133655548096,
      "learning_rate": 3.782892428601502e-05,
      "loss": 0.8702,
      "step": 5053
    },
    {
      "epoch": 1.5958320176823493,
      "grad_norm": 1.6978346109390259,
      "learning_rate": 3.7825415760297525e-05,
      "loss": 0.8745,
      "step": 5054
    },
    {
      "epoch": 1.5961477739185348,
      "grad_norm": 1.585986852645874,
      "learning_rate": 3.782190723458003e-05,
      "loss": 0.8287,
      "step": 5055
    },
    {
      "epoch": 1.5964635301547205,
      "grad_norm": 1.7478224039077759,
      "learning_rate": 3.781839870886254e-05,
      "loss": 0.7967,
      "step": 5056
    },
    {
      "epoch": 1.5967792863909063,
      "grad_norm": 1.5360087156295776,
      "learning_rate": 3.7814890183145046e-05,
      "loss": 0.8966,
      "step": 5057
    },
    {
      "epoch": 1.597095042627092,
      "grad_norm": 1.7335553169250488,
      "learning_rate": 3.781138165742755e-05,
      "loss": 0.861,
      "step": 5058
    },
    {
      "epoch": 1.5974107988632775,
      "grad_norm": 1.5264983177185059,
      "learning_rate": 3.780787313171006e-05,
      "loss": 0.8052,
      "step": 5059
    },
    {
      "epoch": 1.5977265550994633,
      "grad_norm": 1.7247560024261475,
      "learning_rate": 3.7804364605992566e-05,
      "loss": 0.8736,
      "step": 5060
    },
    {
      "epoch": 1.5980423113356488,
      "grad_norm": 1.6035040616989136,
      "learning_rate": 3.780085608027507e-05,
      "loss": 0.8535,
      "step": 5061
    },
    {
      "epoch": 1.5983580675718345,
      "grad_norm": 1.6126066446304321,
      "learning_rate": 3.779734755455758e-05,
      "loss": 0.7825,
      "step": 5062
    },
    {
      "epoch": 1.5986738238080203,
      "grad_norm": 1.6660112142562866,
      "learning_rate": 3.779383902884009e-05,
      "loss": 0.8188,
      "step": 5063
    },
    {
      "epoch": 1.598989580044206,
      "grad_norm": 1.3551644086837769,
      "learning_rate": 3.779033050312259e-05,
      "loss": 0.8222,
      "step": 5064
    },
    {
      "epoch": 1.5993053362803915,
      "grad_norm": 1.921560525894165,
      "learning_rate": 3.7786821977405094e-05,
      "loss": 0.8552,
      "step": 5065
    },
    {
      "epoch": 1.5996210925165772,
      "grad_norm": 3.088942766189575,
      "learning_rate": 3.77833134516876e-05,
      "loss": 0.8619,
      "step": 5066
    },
    {
      "epoch": 1.5999368487527628,
      "grad_norm": 1.7861450910568237,
      "learning_rate": 3.777980492597011e-05,
      "loss": 0.8336,
      "step": 5067
    },
    {
      "epoch": 1.6002526049889485,
      "grad_norm": 1.5119320154190063,
      "learning_rate": 3.7776296400252614e-05,
      "loss": 0.823,
      "step": 5068
    },
    {
      "epoch": 1.6005683612251342,
      "grad_norm": 1.5905479192733765,
      "learning_rate": 3.777278787453512e-05,
      "loss": 0.7956,
      "step": 5069
    },
    {
      "epoch": 1.60088411746132,
      "grad_norm": 1.7036293745040894,
      "learning_rate": 3.776927934881763e-05,
      "loss": 0.8779,
      "step": 5070
    },
    {
      "epoch": 1.6011998736975055,
      "grad_norm": 1.586238980293274,
      "learning_rate": 3.7765770823100134e-05,
      "loss": 0.8725,
      "step": 5071
    },
    {
      "epoch": 1.6015156299336912,
      "grad_norm": 1.5870521068572998,
      "learning_rate": 3.776226229738264e-05,
      "loss": 0.8248,
      "step": 5072
    },
    {
      "epoch": 1.6018313861698767,
      "grad_norm": 1.5312579870224,
      "learning_rate": 3.775875377166515e-05,
      "loss": 0.8314,
      "step": 5073
    },
    {
      "epoch": 1.6021471424060625,
      "grad_norm": 1.5732861757278442,
      "learning_rate": 3.7755245245947655e-05,
      "loss": 0.8402,
      "step": 5074
    },
    {
      "epoch": 1.6024628986422482,
      "grad_norm": 1.6943022012710571,
      "learning_rate": 3.775173672023016e-05,
      "loss": 0.8176,
      "step": 5075
    },
    {
      "epoch": 1.602778654878434,
      "grad_norm": 1.3225643634796143,
      "learning_rate": 3.774822819451267e-05,
      "loss": 0.8008,
      "step": 5076
    },
    {
      "epoch": 1.6030944111146195,
      "grad_norm": 1.5207359790802002,
      "learning_rate": 3.7744719668795175e-05,
      "loss": 0.8292,
      "step": 5077
    },
    {
      "epoch": 1.6034101673508052,
      "grad_norm": 1.5774452686309814,
      "learning_rate": 3.774121114307768e-05,
      "loss": 0.8515,
      "step": 5078
    },
    {
      "epoch": 1.6037259235869907,
      "grad_norm": 1.4081134796142578,
      "learning_rate": 3.773770261736019e-05,
      "loss": 0.735,
      "step": 5079
    },
    {
      "epoch": 1.6040416798231765,
      "grad_norm": 1.5351479053497314,
      "learning_rate": 3.7734194091642696e-05,
      "loss": 0.8641,
      "step": 5080
    },
    {
      "epoch": 1.6043574360593622,
      "grad_norm": 1.615328311920166,
      "learning_rate": 3.77306855659252e-05,
      "loss": 0.8554,
      "step": 5081
    },
    {
      "epoch": 1.604673192295548,
      "grad_norm": 1.4234740734100342,
      "learning_rate": 3.772717704020771e-05,
      "loss": 0.8681,
      "step": 5082
    },
    {
      "epoch": 1.6049889485317335,
      "grad_norm": 1.3264122009277344,
      "learning_rate": 3.772366851449021e-05,
      "loss": 0.7804,
      "step": 5083
    },
    {
      "epoch": 1.6053047047679192,
      "grad_norm": 1.6425964832305908,
      "learning_rate": 3.7720159988772716e-05,
      "loss": 0.8428,
      "step": 5084
    },
    {
      "epoch": 1.6056204610041047,
      "grad_norm": 1.6083643436431885,
      "learning_rate": 3.771665146305522e-05,
      "loss": 0.7817,
      "step": 5085
    },
    {
      "epoch": 1.6059362172402905,
      "grad_norm": 1.6263821125030518,
      "learning_rate": 3.771314293733773e-05,
      "loss": 0.7865,
      "step": 5086
    },
    {
      "epoch": 1.6062519734764762,
      "grad_norm": 1.8989205360412598,
      "learning_rate": 3.770963441162024e-05,
      "loss": 0.8612,
      "step": 5087
    },
    {
      "epoch": 1.606567729712662,
      "grad_norm": 1.6726651191711426,
      "learning_rate": 3.7706125885902743e-05,
      "loss": 0.9054,
      "step": 5088
    },
    {
      "epoch": 1.6068834859488474,
      "grad_norm": 1.8127450942993164,
      "learning_rate": 3.770261736018525e-05,
      "loss": 0.8246,
      "step": 5089
    },
    {
      "epoch": 1.6071992421850332,
      "grad_norm": 1.472322940826416,
      "learning_rate": 3.7699108834467764e-05,
      "loss": 0.764,
      "step": 5090
    },
    {
      "epoch": 1.6075149984212187,
      "grad_norm": 1.573603868484497,
      "learning_rate": 3.769560030875027e-05,
      "loss": 0.7938,
      "step": 5091
    },
    {
      "epoch": 1.6078307546574044,
      "grad_norm": 1.5486643314361572,
      "learning_rate": 3.769209178303277e-05,
      "loss": 0.8347,
      "step": 5092
    },
    {
      "epoch": 1.6081465108935902,
      "grad_norm": 3.822458028793335,
      "learning_rate": 3.768858325731528e-05,
      "loss": 0.8362,
      "step": 5093
    },
    {
      "epoch": 1.608462267129776,
      "grad_norm": 1.6220942735671997,
      "learning_rate": 3.7685074731597784e-05,
      "loss": 0.8332,
      "step": 5094
    },
    {
      "epoch": 1.6087780233659614,
      "grad_norm": 1.6034934520721436,
      "learning_rate": 3.768156620588029e-05,
      "loss": 0.8406,
      "step": 5095
    },
    {
      "epoch": 1.6090937796021472,
      "grad_norm": 1.5629174709320068,
      "learning_rate": 3.76780576801628e-05,
      "loss": 0.7782,
      "step": 5096
    },
    {
      "epoch": 1.6094095358383327,
      "grad_norm": 1.8423640727996826,
      "learning_rate": 3.7674549154445305e-05,
      "loss": 0.8295,
      "step": 5097
    },
    {
      "epoch": 1.6097252920745184,
      "grad_norm": 1.6807990074157715,
      "learning_rate": 3.767104062872781e-05,
      "loss": 0.869,
      "step": 5098
    },
    {
      "epoch": 1.6100410483107042,
      "grad_norm": 1.5227093696594238,
      "learning_rate": 3.766753210301032e-05,
      "loss": 0.7992,
      "step": 5099
    },
    {
      "epoch": 1.61035680454689,
      "grad_norm": 1.4859997034072876,
      "learning_rate": 3.7664023577292825e-05,
      "loss": 0.8203,
      "step": 5100
    },
    {
      "epoch": 1.6106725607830754,
      "grad_norm": 1.606156349182129,
      "learning_rate": 3.766051505157533e-05,
      "loss": 0.8017,
      "step": 5101
    },
    {
      "epoch": 1.6109883170192612,
      "grad_norm": 2.414417028427124,
      "learning_rate": 3.765700652585783e-05,
      "loss": 0.9397,
      "step": 5102
    },
    {
      "epoch": 1.6113040732554467,
      "grad_norm": 1.6797153949737549,
      "learning_rate": 3.765349800014034e-05,
      "loss": 0.8186,
      "step": 5103
    },
    {
      "epoch": 1.6116198294916324,
      "grad_norm": 1.4573915004730225,
      "learning_rate": 3.7649989474422846e-05,
      "loss": 0.8291,
      "step": 5104
    },
    {
      "epoch": 1.6119355857278181,
      "grad_norm": 1.8462940454483032,
      "learning_rate": 3.764648094870535e-05,
      "loss": 0.8984,
      "step": 5105
    },
    {
      "epoch": 1.6122513419640039,
      "grad_norm": 1.6135650873184204,
      "learning_rate": 3.764297242298786e-05,
      "loss": 0.8964,
      "step": 5106
    },
    {
      "epoch": 1.6125670982001896,
      "grad_norm": 1.782224178314209,
      "learning_rate": 3.7639463897270366e-05,
      "loss": 0.8657,
      "step": 5107
    },
    {
      "epoch": 1.6128828544363751,
      "grad_norm": 1.9702509641647339,
      "learning_rate": 3.763595537155288e-05,
      "loss": 0.8361,
      "step": 5108
    },
    {
      "epoch": 1.6131986106725607,
      "grad_norm": 1.5906575918197632,
      "learning_rate": 3.7632446845835387e-05,
      "loss": 0.886,
      "step": 5109
    },
    {
      "epoch": 1.6135143669087464,
      "grad_norm": 1.8122599124908447,
      "learning_rate": 3.762893832011789e-05,
      "loss": 0.8272,
      "step": 5110
    },
    {
      "epoch": 1.6138301231449321,
      "grad_norm": 1.5467407703399658,
      "learning_rate": 3.7625429794400393e-05,
      "loss": 0.7987,
      "step": 5111
    },
    {
      "epoch": 1.6141458793811179,
      "grad_norm": 2.1135921478271484,
      "learning_rate": 3.76219212686829e-05,
      "loss": 0.8368,
      "step": 5112
    },
    {
      "epoch": 1.6144616356173036,
      "grad_norm": 1.7232626676559448,
      "learning_rate": 3.761841274296541e-05,
      "loss": 0.7995,
      "step": 5113
    },
    {
      "epoch": 1.6147773918534891,
      "grad_norm": 1.708240270614624,
      "learning_rate": 3.7614904217247914e-05,
      "loss": 0.7819,
      "step": 5114
    },
    {
      "epoch": 1.6150931480896746,
      "grad_norm": 1.4465301036834717,
      "learning_rate": 3.761139569153042e-05,
      "loss": 0.7904,
      "step": 5115
    },
    {
      "epoch": 1.6154089043258604,
      "grad_norm": 1.7587947845458984,
      "learning_rate": 3.760788716581293e-05,
      "loss": 0.8798,
      "step": 5116
    },
    {
      "epoch": 1.6157246605620461,
      "grad_norm": 1.4866936206817627,
      "learning_rate": 3.7604378640095434e-05,
      "loss": 0.7661,
      "step": 5117
    },
    {
      "epoch": 1.6160404167982318,
      "grad_norm": 1.7399965524673462,
      "learning_rate": 3.760087011437794e-05,
      "loss": 0.8814,
      "step": 5118
    },
    {
      "epoch": 1.6163561730344176,
      "grad_norm": 1.4643458127975464,
      "learning_rate": 3.759736158866045e-05,
      "loss": 0.8509,
      "step": 5119
    },
    {
      "epoch": 1.616671929270603,
      "grad_norm": 1.393607258796692,
      "learning_rate": 3.7593853062942955e-05,
      "loss": 0.8821,
      "step": 5120
    },
    {
      "epoch": 1.6169876855067886,
      "grad_norm": 1.6108826398849487,
      "learning_rate": 3.7590344537225455e-05,
      "loss": 0.7595,
      "step": 5121
    },
    {
      "epoch": 1.6173034417429744,
      "grad_norm": 1.8430935144424438,
      "learning_rate": 3.758683601150796e-05,
      "loss": 0.8347,
      "step": 5122
    },
    {
      "epoch": 1.61761919797916,
      "grad_norm": 1.5073673725128174,
      "learning_rate": 3.758332748579047e-05,
      "loss": 0.821,
      "step": 5123
    },
    {
      "epoch": 1.6179349542153458,
      "grad_norm": 1.5474271774291992,
      "learning_rate": 3.7579818960072975e-05,
      "loss": 0.7701,
      "step": 5124
    },
    {
      "epoch": 1.6182507104515316,
      "grad_norm": 1.5504217147827148,
      "learning_rate": 3.757631043435549e-05,
      "loss": 0.8211,
      "step": 5125
    },
    {
      "epoch": 1.618566466687717,
      "grad_norm": 1.6862438917160034,
      "learning_rate": 3.7572801908637996e-05,
      "loss": 0.7868,
      "step": 5126
    },
    {
      "epoch": 1.6188822229239026,
      "grad_norm": 1.4331867694854736,
      "learning_rate": 3.75692933829205e-05,
      "loss": 0.7731,
      "step": 5127
    },
    {
      "epoch": 1.6191979791600883,
      "grad_norm": 1.3537402153015137,
      "learning_rate": 3.756578485720301e-05,
      "loss": 0.856,
      "step": 5128
    },
    {
      "epoch": 1.619513735396274,
      "grad_norm": 1.5308903455734253,
      "learning_rate": 3.7562276331485516e-05,
      "loss": 0.8614,
      "step": 5129
    },
    {
      "epoch": 1.6198294916324598,
      "grad_norm": 1.3987387418746948,
      "learning_rate": 3.7558767805768016e-05,
      "loss": 0.784,
      "step": 5130
    },
    {
      "epoch": 1.6201452478686456,
      "grad_norm": 1.607672095298767,
      "learning_rate": 3.755525928005052e-05,
      "loss": 0.9373,
      "step": 5131
    },
    {
      "epoch": 1.620461004104831,
      "grad_norm": 1.4741718769073486,
      "learning_rate": 3.755175075433303e-05,
      "loss": 0.751,
      "step": 5132
    },
    {
      "epoch": 1.6207767603410166,
      "grad_norm": 1.7237701416015625,
      "learning_rate": 3.7548242228615537e-05,
      "loss": 0.8811,
      "step": 5133
    },
    {
      "epoch": 1.6210925165772023,
      "grad_norm": 1.6561201810836792,
      "learning_rate": 3.754473370289804e-05,
      "loss": 0.8417,
      "step": 5134
    },
    {
      "epoch": 1.621408272813388,
      "grad_norm": 1.4816585779190063,
      "learning_rate": 3.754122517718055e-05,
      "loss": 0.7754,
      "step": 5135
    },
    {
      "epoch": 1.6217240290495738,
      "grad_norm": 1.710735559463501,
      "learning_rate": 3.753771665146306e-05,
      "loss": 0.8106,
      "step": 5136
    },
    {
      "epoch": 1.6220397852857595,
      "grad_norm": 1.652948021888733,
      "learning_rate": 3.7534208125745564e-05,
      "loss": 0.8595,
      "step": 5137
    },
    {
      "epoch": 1.622355541521945,
      "grad_norm": 1.4414119720458984,
      "learning_rate": 3.753069960002807e-05,
      "loss": 0.7347,
      "step": 5138
    },
    {
      "epoch": 1.6226712977581306,
      "grad_norm": 1.3766125440597534,
      "learning_rate": 3.752719107431058e-05,
      "loss": 0.8683,
      "step": 5139
    },
    {
      "epoch": 1.6229870539943163,
      "grad_norm": 1.6057956218719482,
      "learning_rate": 3.7523682548593084e-05,
      "loss": 0.7446,
      "step": 5140
    },
    {
      "epoch": 1.623302810230502,
      "grad_norm": 1.5793780088424683,
      "learning_rate": 3.7520174022875584e-05,
      "loss": 0.983,
      "step": 5141
    },
    {
      "epoch": 1.6236185664666878,
      "grad_norm": 1.7132762670516968,
      "learning_rate": 3.751666549715809e-05,
      "loss": 0.8013,
      "step": 5142
    },
    {
      "epoch": 1.6239343227028735,
      "grad_norm": 1.4810545444488525,
      "learning_rate": 3.7513156971440605e-05,
      "loss": 0.8408,
      "step": 5143
    },
    {
      "epoch": 1.624250078939059,
      "grad_norm": 1.5929893255233765,
      "learning_rate": 3.750964844572311e-05,
      "loss": 0.7891,
      "step": 5144
    },
    {
      "epoch": 1.6245658351752446,
      "grad_norm": 1.5730725526809692,
      "learning_rate": 3.750613992000562e-05,
      "loss": 0.7703,
      "step": 5145
    },
    {
      "epoch": 1.6248815914114303,
      "grad_norm": 1.5051493644714355,
      "learning_rate": 3.7502631394288125e-05,
      "loss": 0.8459,
      "step": 5146
    },
    {
      "epoch": 1.625197347647616,
      "grad_norm": 1.9080697298049927,
      "learning_rate": 3.749912286857063e-05,
      "loss": 0.8175,
      "step": 5147
    },
    {
      "epoch": 1.6255131038838018,
      "grad_norm": 1.573887586593628,
      "learning_rate": 3.749561434285314e-05,
      "loss": 0.8646,
      "step": 5148
    },
    {
      "epoch": 1.6258288601199875,
      "grad_norm": 2.545718193054199,
      "learning_rate": 3.749210581713564e-05,
      "loss": 0.8449,
      "step": 5149
    },
    {
      "epoch": 1.626144616356173,
      "grad_norm": 1.5267207622528076,
      "learning_rate": 3.7488597291418146e-05,
      "loss": 0.9189,
      "step": 5150
    },
    {
      "epoch": 1.6264603725923585,
      "grad_norm": 1.692452073097229,
      "learning_rate": 3.748508876570065e-05,
      "loss": 0.8025,
      "step": 5151
    },
    {
      "epoch": 1.6267761288285443,
      "grad_norm": 1.631395697593689,
      "learning_rate": 3.748158023998316e-05,
      "loss": 0.8275,
      "step": 5152
    },
    {
      "epoch": 1.62709188506473,
      "grad_norm": 2.397040367126465,
      "learning_rate": 3.7478071714265666e-05,
      "loss": 0.8116,
      "step": 5153
    },
    {
      "epoch": 1.6274076413009158,
      "grad_norm": 1.8352304697036743,
      "learning_rate": 3.747456318854817e-05,
      "loss": 0.7831,
      "step": 5154
    },
    {
      "epoch": 1.6277233975371015,
      "grad_norm": 1.5617971420288086,
      "learning_rate": 3.747105466283068e-05,
      "loss": 0.7758,
      "step": 5155
    },
    {
      "epoch": 1.628039153773287,
      "grad_norm": 1.707202672958374,
      "learning_rate": 3.7467546137113186e-05,
      "loss": 0.8342,
      "step": 5156
    },
    {
      "epoch": 1.6283549100094725,
      "grad_norm": 1.8691133260726929,
      "learning_rate": 3.746403761139569e-05,
      "loss": 0.8846,
      "step": 5157
    },
    {
      "epoch": 1.6286706662456583,
      "grad_norm": 1.3866554498672485,
      "learning_rate": 3.74605290856782e-05,
      "loss": 0.8005,
      "step": 5158
    },
    {
      "epoch": 1.628986422481844,
      "grad_norm": 1.5791021585464478,
      "learning_rate": 3.745702055996071e-05,
      "loss": 0.9237,
      "step": 5159
    },
    {
      "epoch": 1.6293021787180297,
      "grad_norm": 1.6687885522842407,
      "learning_rate": 3.7453512034243214e-05,
      "loss": 0.8469,
      "step": 5160
    },
    {
      "epoch": 1.6296179349542155,
      "grad_norm": 1.6584012508392334,
      "learning_rate": 3.745000350852572e-05,
      "loss": 0.8419,
      "step": 5161
    },
    {
      "epoch": 1.629933691190401,
      "grad_norm": 1.5284264087677002,
      "learning_rate": 3.744649498280823e-05,
      "loss": 0.7413,
      "step": 5162
    },
    {
      "epoch": 1.6302494474265867,
      "grad_norm": 1.513537883758545,
      "learning_rate": 3.7442986457090734e-05,
      "loss": 0.8504,
      "step": 5163
    },
    {
      "epoch": 1.6305652036627722,
      "grad_norm": 1.6262078285217285,
      "learning_rate": 3.743947793137324e-05,
      "loss": 0.8506,
      "step": 5164
    },
    {
      "epoch": 1.630880959898958,
      "grad_norm": 1.7050657272338867,
      "learning_rate": 3.743596940565575e-05,
      "loss": 0.8437,
      "step": 5165
    },
    {
      "epoch": 1.6311967161351437,
      "grad_norm": 1.630054235458374,
      "learning_rate": 3.7432460879938255e-05,
      "loss": 0.8187,
      "step": 5166
    },
    {
      "epoch": 1.6315124723713295,
      "grad_norm": 1.7192258834838867,
      "learning_rate": 3.742895235422076e-05,
      "loss": 0.7933,
      "step": 5167
    },
    {
      "epoch": 1.631828228607515,
      "grad_norm": 1.2983051538467407,
      "learning_rate": 3.742544382850327e-05,
      "loss": 0.8104,
      "step": 5168
    },
    {
      "epoch": 1.6321439848437007,
      "grad_norm": 1.783172369003296,
      "learning_rate": 3.742193530278577e-05,
      "loss": 0.8586,
      "step": 5169
    },
    {
      "epoch": 1.6324597410798862,
      "grad_norm": 1.6884504556655884,
      "learning_rate": 3.7418426777068275e-05,
      "loss": 0.8902,
      "step": 5170
    },
    {
      "epoch": 1.632775497316072,
      "grad_norm": 1.4973958730697632,
      "learning_rate": 3.741491825135078e-05,
      "loss": 0.7527,
      "step": 5171
    },
    {
      "epoch": 1.6330912535522577,
      "grad_norm": 1.6182132959365845,
      "learning_rate": 3.741140972563329e-05,
      "loss": 0.8348,
      "step": 5172
    },
    {
      "epoch": 1.6334070097884434,
      "grad_norm": 1.4643468856811523,
      "learning_rate": 3.7407901199915795e-05,
      "loss": 0.8692,
      "step": 5173
    },
    {
      "epoch": 1.633722766024629,
      "grad_norm": 1.6358442306518555,
      "learning_rate": 3.74043926741983e-05,
      "loss": 0.8877,
      "step": 5174
    },
    {
      "epoch": 1.6340385222608147,
      "grad_norm": 1.3535360097885132,
      "learning_rate": 3.740088414848081e-05,
      "loss": 0.813,
      "step": 5175
    },
    {
      "epoch": 1.6343542784970002,
      "grad_norm": 1.3456146717071533,
      "learning_rate": 3.7397375622763316e-05,
      "loss": 0.8456,
      "step": 5176
    },
    {
      "epoch": 1.634670034733186,
      "grad_norm": 1.4897866249084473,
      "learning_rate": 3.739386709704582e-05,
      "loss": 0.8256,
      "step": 5177
    },
    {
      "epoch": 1.6349857909693717,
      "grad_norm": 1.4344450235366821,
      "learning_rate": 3.739035857132833e-05,
      "loss": 0.8405,
      "step": 5178
    },
    {
      "epoch": 1.6353015472055574,
      "grad_norm": 1.5324552059173584,
      "learning_rate": 3.7386850045610836e-05,
      "loss": 0.8345,
      "step": 5179
    },
    {
      "epoch": 1.635617303441743,
      "grad_norm": 1.4531201124191284,
      "learning_rate": 3.738334151989334e-05,
      "loss": 0.8022,
      "step": 5180
    },
    {
      "epoch": 1.6359330596779287,
      "grad_norm": 1.6429578065872192,
      "learning_rate": 3.737983299417585e-05,
      "loss": 0.8361,
      "step": 5181
    },
    {
      "epoch": 1.6362488159141142,
      "grad_norm": 1.4572690725326538,
      "learning_rate": 3.737632446845836e-05,
      "loss": 0.7944,
      "step": 5182
    },
    {
      "epoch": 1.6365645721503,
      "grad_norm": 1.5852808952331543,
      "learning_rate": 3.7372815942740864e-05,
      "loss": 0.8906,
      "step": 5183
    },
    {
      "epoch": 1.6368803283864857,
      "grad_norm": 1.5711814165115356,
      "learning_rate": 3.736930741702337e-05,
      "loss": 0.8981,
      "step": 5184
    },
    {
      "epoch": 1.6371960846226714,
      "grad_norm": 1.3497421741485596,
      "learning_rate": 3.736579889130588e-05,
      "loss": 0.7984,
      "step": 5185
    },
    {
      "epoch": 1.637511840858857,
      "grad_norm": 2.408278703689575,
      "learning_rate": 3.7362290365588384e-05,
      "loss": 0.889,
      "step": 5186
    },
    {
      "epoch": 1.6378275970950427,
      "grad_norm": 1.5867490768432617,
      "learning_rate": 3.735878183987089e-05,
      "loss": 0.9011,
      "step": 5187
    },
    {
      "epoch": 1.6381433533312282,
      "grad_norm": 1.7212071418762207,
      "learning_rate": 3.735527331415339e-05,
      "loss": 0.8759,
      "step": 5188
    },
    {
      "epoch": 1.638459109567414,
      "grad_norm": 1.363734483718872,
      "learning_rate": 3.73517647884359e-05,
      "loss": 0.88,
      "step": 5189
    },
    {
      "epoch": 1.6387748658035997,
      "grad_norm": 1.5886611938476562,
      "learning_rate": 3.7348256262718405e-05,
      "loss": 0.8198,
      "step": 5190
    },
    {
      "epoch": 1.6390906220397854,
      "grad_norm": 1.6794050931930542,
      "learning_rate": 3.734474773700091e-05,
      "loss": 0.9161,
      "step": 5191
    },
    {
      "epoch": 1.639406378275971,
      "grad_norm": 1.8658729791641235,
      "learning_rate": 3.734123921128342e-05,
      "loss": 0.8464,
      "step": 5192
    },
    {
      "epoch": 1.6397221345121566,
      "grad_norm": 1.3696396350860596,
      "learning_rate": 3.7337730685565925e-05,
      "loss": 0.805,
      "step": 5193
    },
    {
      "epoch": 1.6400378907483422,
      "grad_norm": 1.3785524368286133,
      "learning_rate": 3.733422215984843e-05,
      "loss": 0.7488,
      "step": 5194
    },
    {
      "epoch": 1.640353646984528,
      "grad_norm": 1.7262612581253052,
      "learning_rate": 3.7330713634130945e-05,
      "loss": 0.8516,
      "step": 5195
    },
    {
      "epoch": 1.6406694032207136,
      "grad_norm": 1.5183308124542236,
      "learning_rate": 3.732720510841345e-05,
      "loss": 0.8004,
      "step": 5196
    },
    {
      "epoch": 1.6409851594568994,
      "grad_norm": 1.4178286790847778,
      "learning_rate": 3.732369658269595e-05,
      "loss": 0.7629,
      "step": 5197
    },
    {
      "epoch": 1.641300915693085,
      "grad_norm": 1.4066914319992065,
      "learning_rate": 3.732018805697846e-05,
      "loss": 0.7986,
      "step": 5198
    },
    {
      "epoch": 1.6416166719292706,
      "grad_norm": 1.5659650564193726,
      "learning_rate": 3.7316679531260966e-05,
      "loss": 0.8447,
      "step": 5199
    },
    {
      "epoch": 1.6419324281654561,
      "grad_norm": 1.5208096504211426,
      "learning_rate": 3.731317100554347e-05,
      "loss": 0.8069,
      "step": 5200
    },
    {
      "epoch": 1.6419324281654561,
      "eval_loss": 0.8032549619674683,
      "eval_runtime": 4.7448,
      "eval_samples_per_second": 524.786,
      "eval_steps_per_second": 4.426,
      "step": 5200
    },
    {
      "epoch": 1.6422481844016419,
      "grad_norm": 1.5353869199752808,
      "learning_rate": 3.730966247982598e-05,
      "loss": 0.7892,
      "step": 5201
    },
    {
      "epoch": 1.6425639406378276,
      "grad_norm": 2.5691771507263184,
      "learning_rate": 3.7306153954108486e-05,
      "loss": 0.9046,
      "step": 5202
    },
    {
      "epoch": 1.6428796968740134,
      "grad_norm": 1.4076288938522339,
      "learning_rate": 3.730264542839099e-05,
      "loss": 0.7971,
      "step": 5203
    },
    {
      "epoch": 1.6431954531101989,
      "grad_norm": 1.747169852256775,
      "learning_rate": 3.72991369026735e-05,
      "loss": 0.8266,
      "step": 5204
    },
    {
      "epoch": 1.6435112093463846,
      "grad_norm": 1.609574556350708,
      "learning_rate": 3.729562837695601e-05,
      "loss": 0.8171,
      "step": 5205
    },
    {
      "epoch": 1.6438269655825701,
      "grad_norm": 1.379012107849121,
      "learning_rate": 3.7292119851238514e-05,
      "loss": 0.7057,
      "step": 5206
    },
    {
      "epoch": 1.6441427218187559,
      "grad_norm": 1.6667073965072632,
      "learning_rate": 3.7288611325521014e-05,
      "loss": 0.8104,
      "step": 5207
    },
    {
      "epoch": 1.6444584780549416,
      "grad_norm": 1.397088646888733,
      "learning_rate": 3.728510279980352e-05,
      "loss": 0.8602,
      "step": 5208
    },
    {
      "epoch": 1.6447742342911273,
      "grad_norm": 1.6937955617904663,
      "learning_rate": 3.728159427408603e-05,
      "loss": 0.8166,
      "step": 5209
    },
    {
      "epoch": 1.6450899905273129,
      "grad_norm": 1.731307029724121,
      "learning_rate": 3.7278085748368534e-05,
      "loss": 0.9159,
      "step": 5210
    },
    {
      "epoch": 1.6454057467634986,
      "grad_norm": 1.6946280002593994,
      "learning_rate": 3.727457722265104e-05,
      "loss": 0.8342,
      "step": 5211
    },
    {
      "epoch": 1.6457215029996841,
      "grad_norm": 1.5929213762283325,
      "learning_rate": 3.727106869693355e-05,
      "loss": 0.8487,
      "step": 5212
    },
    {
      "epoch": 1.6460372592358699,
      "grad_norm": 1.7374563217163086,
      "learning_rate": 3.726756017121606e-05,
      "loss": 0.9084,
      "step": 5213
    },
    {
      "epoch": 1.6463530154720556,
      "grad_norm": 1.461136817932129,
      "learning_rate": 3.726405164549857e-05,
      "loss": 0.8424,
      "step": 5214
    },
    {
      "epoch": 1.6466687717082413,
      "grad_norm": 1.2923390865325928,
      "learning_rate": 3.7260543119781075e-05,
      "loss": 0.744,
      "step": 5215
    },
    {
      "epoch": 1.646984527944427,
      "grad_norm": 1.6055855751037598,
      "learning_rate": 3.7257034594063575e-05,
      "loss": 0.8053,
      "step": 5216
    },
    {
      "epoch": 1.6473002841806126,
      "grad_norm": 1.560577392578125,
      "learning_rate": 3.725352606834608e-05,
      "loss": 0.8191,
      "step": 5217
    },
    {
      "epoch": 1.647616040416798,
      "grad_norm": 1.7458409070968628,
      "learning_rate": 3.725001754262859e-05,
      "loss": 0.8283,
      "step": 5218
    },
    {
      "epoch": 1.6479317966529838,
      "grad_norm": 3.849846363067627,
      "learning_rate": 3.7246509016911095e-05,
      "loss": 0.8007,
      "step": 5219
    },
    {
      "epoch": 1.6482475528891696,
      "grad_norm": 1.472739338874817,
      "learning_rate": 3.72430004911936e-05,
      "loss": 0.7333,
      "step": 5220
    },
    {
      "epoch": 1.6485633091253553,
      "grad_norm": 1.5081013441085815,
      "learning_rate": 3.723949196547611e-05,
      "loss": 0.7998,
      "step": 5221
    },
    {
      "epoch": 1.648879065361541,
      "grad_norm": 1.5693519115447998,
      "learning_rate": 3.7235983439758616e-05,
      "loss": 0.812,
      "step": 5222
    },
    {
      "epoch": 1.6491948215977266,
      "grad_norm": 1.805011510848999,
      "learning_rate": 3.723247491404112e-05,
      "loss": 0.8279,
      "step": 5223
    },
    {
      "epoch": 1.649510577833912,
      "grad_norm": 1.8493775129318237,
      "learning_rate": 3.722896638832363e-05,
      "loss": 0.8454,
      "step": 5224
    },
    {
      "epoch": 1.6498263340700978,
      "grad_norm": 1.3976987600326538,
      "learning_rate": 3.7225457862606136e-05,
      "loss": 0.8342,
      "step": 5225
    },
    {
      "epoch": 1.6501420903062836,
      "grad_norm": 1.3794821500778198,
      "learning_rate": 3.7221949336888636e-05,
      "loss": 0.8616,
      "step": 5226
    },
    {
      "epoch": 1.6504578465424693,
      "grad_norm": 1.4706273078918457,
      "learning_rate": 3.721844081117114e-05,
      "loss": 0.8098,
      "step": 5227
    },
    {
      "epoch": 1.650773602778655,
      "grad_norm": 1.5227131843566895,
      "learning_rate": 3.721493228545365e-05,
      "loss": 0.8385,
      "step": 5228
    },
    {
      "epoch": 1.6510893590148406,
      "grad_norm": 1.3529319763183594,
      "learning_rate": 3.721142375973616e-05,
      "loss": 0.8665,
      "step": 5229
    },
    {
      "epoch": 1.651405115251026,
      "grad_norm": 1.6496703624725342,
      "learning_rate": 3.720791523401867e-05,
      "loss": 0.8195,
      "step": 5230
    },
    {
      "epoch": 1.6517208714872118,
      "grad_norm": 1.55513334274292,
      "learning_rate": 3.720440670830118e-05,
      "loss": 0.8166,
      "step": 5231
    },
    {
      "epoch": 1.6520366277233975,
      "grad_norm": 1.51381516456604,
      "learning_rate": 3.7200898182583684e-05,
      "loss": 0.8063,
      "step": 5232
    },
    {
      "epoch": 1.6523523839595833,
      "grad_norm": 1.487753987312317,
      "learning_rate": 3.719738965686619e-05,
      "loss": 0.8731,
      "step": 5233
    },
    {
      "epoch": 1.652668140195769,
      "grad_norm": 1.4157350063323975,
      "learning_rate": 3.71938811311487e-05,
      "loss": 0.8116,
      "step": 5234
    },
    {
      "epoch": 1.6529838964319545,
      "grad_norm": 1.6238561868667603,
      "learning_rate": 3.71903726054312e-05,
      "loss": 0.8155,
      "step": 5235
    },
    {
      "epoch": 1.65329965266814,
      "grad_norm": 1.5540248155593872,
      "learning_rate": 3.7186864079713704e-05,
      "loss": 0.8772,
      "step": 5236
    },
    {
      "epoch": 1.6536154089043258,
      "grad_norm": 1.8155204057693481,
      "learning_rate": 3.718335555399621e-05,
      "loss": 0.7723,
      "step": 5237
    },
    {
      "epoch": 1.6539311651405115,
      "grad_norm": 1.4586080312728882,
      "learning_rate": 3.717984702827872e-05,
      "loss": 0.7924,
      "step": 5238
    },
    {
      "epoch": 1.6542469213766973,
      "grad_norm": 1.7099891901016235,
      "learning_rate": 3.7176338502561225e-05,
      "loss": 0.8037,
      "step": 5239
    },
    {
      "epoch": 1.654562677612883,
      "grad_norm": 1.7398186922073364,
      "learning_rate": 3.717282997684373e-05,
      "loss": 0.8932,
      "step": 5240
    },
    {
      "epoch": 1.6548784338490685,
      "grad_norm": 1.7840981483459473,
      "learning_rate": 3.716932145112624e-05,
      "loss": 0.768,
      "step": 5241
    },
    {
      "epoch": 1.655194190085254,
      "grad_norm": 2.01460862159729,
      "learning_rate": 3.7165812925408745e-05,
      "loss": 0.8058,
      "step": 5242
    },
    {
      "epoch": 1.6555099463214398,
      "grad_norm": 2.069404363632202,
      "learning_rate": 3.716230439969125e-05,
      "loss": 0.9045,
      "step": 5243
    },
    {
      "epoch": 1.6558257025576255,
      "grad_norm": 1.7984302043914795,
      "learning_rate": 3.715879587397376e-05,
      "loss": 0.8521,
      "step": 5244
    },
    {
      "epoch": 1.6561414587938112,
      "grad_norm": 1.6282991170883179,
      "learning_rate": 3.7155287348256266e-05,
      "loss": 0.8589,
      "step": 5245
    },
    {
      "epoch": 1.656457215029997,
      "grad_norm": 1.5746698379516602,
      "learning_rate": 3.7151778822538766e-05,
      "loss": 0.7673,
      "step": 5246
    },
    {
      "epoch": 1.6567729712661825,
      "grad_norm": 1.5195739269256592,
      "learning_rate": 3.714827029682127e-05,
      "loss": 0.7982,
      "step": 5247
    },
    {
      "epoch": 1.657088727502368,
      "grad_norm": 1.4736098051071167,
      "learning_rate": 3.7144761771103786e-05,
      "loss": 0.8397,
      "step": 5248
    },
    {
      "epoch": 1.6574044837385538,
      "grad_norm": 1.7340282201766968,
      "learning_rate": 3.714125324538629e-05,
      "loss": 0.8658,
      "step": 5249
    },
    {
      "epoch": 1.6577202399747395,
      "grad_norm": 1.6198762655258179,
      "learning_rate": 3.71377447196688e-05,
      "loss": 0.8054,
      "step": 5250
    },
    {
      "epoch": 1.6580359962109252,
      "grad_norm": 1.8169491291046143,
      "learning_rate": 3.7134236193951307e-05,
      "loss": 0.8918,
      "step": 5251
    },
    {
      "epoch": 1.658351752447111,
      "grad_norm": 1.4758460521697998,
      "learning_rate": 3.713072766823381e-05,
      "loss": 0.7663,
      "step": 5252
    },
    {
      "epoch": 1.6586675086832965,
      "grad_norm": 1.6086297035217285,
      "learning_rate": 3.712721914251632e-05,
      "loss": 0.8814,
      "step": 5253
    },
    {
      "epoch": 1.658983264919482,
      "grad_norm": 1.6815297603607178,
      "learning_rate": 3.712371061679882e-05,
      "loss": 0.8129,
      "step": 5254
    },
    {
      "epoch": 1.6592990211556677,
      "grad_norm": 1.501027226448059,
      "learning_rate": 3.712020209108133e-05,
      "loss": 0.7651,
      "step": 5255
    },
    {
      "epoch": 1.6596147773918535,
      "grad_norm": 1.4576534032821655,
      "learning_rate": 3.7116693565363834e-05,
      "loss": 0.8465,
      "step": 5256
    },
    {
      "epoch": 1.6599305336280392,
      "grad_norm": 1.9207993745803833,
      "learning_rate": 3.711318503964634e-05,
      "loss": 0.9208,
      "step": 5257
    },
    {
      "epoch": 1.660246289864225,
      "grad_norm": 1.6832221746444702,
      "learning_rate": 3.710967651392885e-05,
      "loss": 0.8005,
      "step": 5258
    },
    {
      "epoch": 1.6605620461004105,
      "grad_norm": 1.5594749450683594,
      "learning_rate": 3.7106167988211354e-05,
      "loss": 0.7457,
      "step": 5259
    },
    {
      "epoch": 1.660877802336596,
      "grad_norm": 1.6744896173477173,
      "learning_rate": 3.710265946249386e-05,
      "loss": 0.8835,
      "step": 5260
    },
    {
      "epoch": 1.6611935585727817,
      "grad_norm": 10.113781929016113,
      "learning_rate": 3.709915093677637e-05,
      "loss": 0.7704,
      "step": 5261
    },
    {
      "epoch": 1.6615093148089675,
      "grad_norm": 1.560686469078064,
      "learning_rate": 3.7095642411058875e-05,
      "loss": 0.8697,
      "step": 5262
    },
    {
      "epoch": 1.6618250710451532,
      "grad_norm": 1.7674793004989624,
      "learning_rate": 3.709213388534138e-05,
      "loss": 0.8314,
      "step": 5263
    },
    {
      "epoch": 1.662140827281339,
      "grad_norm": 1.8246595859527588,
      "learning_rate": 3.708862535962389e-05,
      "loss": 0.8377,
      "step": 5264
    },
    {
      "epoch": 1.6624565835175245,
      "grad_norm": 1.76127028465271,
      "learning_rate": 3.7085116833906395e-05,
      "loss": 0.8529,
      "step": 5265
    },
    {
      "epoch": 1.66277233975371,
      "grad_norm": 1.391167402267456,
      "learning_rate": 3.70816083081889e-05,
      "loss": 0.774,
      "step": 5266
    },
    {
      "epoch": 1.6630880959898957,
      "grad_norm": 1.5596741437911987,
      "learning_rate": 3.707809978247141e-05,
      "loss": 0.8082,
      "step": 5267
    },
    {
      "epoch": 1.6634038522260814,
      "grad_norm": 1.6280179023742676,
      "learning_rate": 3.7074591256753916e-05,
      "loss": 0.8444,
      "step": 5268
    },
    {
      "epoch": 1.6637196084622672,
      "grad_norm": 1.9853307008743286,
      "learning_rate": 3.707108273103642e-05,
      "loss": 0.8577,
      "step": 5269
    },
    {
      "epoch": 1.664035364698453,
      "grad_norm": 1.5723598003387451,
      "learning_rate": 3.706757420531893e-05,
      "loss": 0.8147,
      "step": 5270
    },
    {
      "epoch": 1.6643511209346384,
      "grad_norm": 1.5264301300048828,
      "learning_rate": 3.7064065679601436e-05,
      "loss": 0.7822,
      "step": 5271
    },
    {
      "epoch": 1.6646668771708242,
      "grad_norm": 1.705899953842163,
      "learning_rate": 3.706055715388394e-05,
      "loss": 0.868,
      "step": 5272
    },
    {
      "epoch": 1.6649826334070097,
      "grad_norm": 1.5429391860961914,
      "learning_rate": 3.705704862816645e-05,
      "loss": 0.7859,
      "step": 5273
    },
    {
      "epoch": 1.6652983896431954,
      "grad_norm": 1.5773643255233765,
      "learning_rate": 3.705354010244895e-05,
      "loss": 0.8325,
      "step": 5274
    },
    {
      "epoch": 1.6656141458793812,
      "grad_norm": 1.3403648138046265,
      "learning_rate": 3.7050031576731457e-05,
      "loss": 0.8,
      "step": 5275
    },
    {
      "epoch": 1.665929902115567,
      "grad_norm": 1.5176193714141846,
      "learning_rate": 3.704652305101396e-05,
      "loss": 0.7831,
      "step": 5276
    },
    {
      "epoch": 1.6662456583517524,
      "grad_norm": 1.3966823816299438,
      "learning_rate": 3.704301452529647e-05,
      "loss": 0.7677,
      "step": 5277
    },
    {
      "epoch": 1.6665614145879382,
      "grad_norm": 1.9365239143371582,
      "learning_rate": 3.703950599957898e-05,
      "loss": 0.9346,
      "step": 5278
    },
    {
      "epoch": 1.6668771708241237,
      "grad_norm": 1.5025556087493896,
      "learning_rate": 3.7035997473861484e-05,
      "loss": 0.7158,
      "step": 5279
    },
    {
      "epoch": 1.6671929270603094,
      "grad_norm": 1.7814093828201294,
      "learning_rate": 3.703248894814399e-05,
      "loss": 0.8976,
      "step": 5280
    },
    {
      "epoch": 1.6675086832964952,
      "grad_norm": 1.481447696685791,
      "learning_rate": 3.70289804224265e-05,
      "loss": 0.8237,
      "step": 5281
    },
    {
      "epoch": 1.667824439532681,
      "grad_norm": 1.5727640390396118,
      "learning_rate": 3.7025471896709004e-05,
      "loss": 0.8577,
      "step": 5282
    },
    {
      "epoch": 1.6681401957688664,
      "grad_norm": 7.5696282386779785,
      "learning_rate": 3.702196337099151e-05,
      "loss": 0.8311,
      "step": 5283
    },
    {
      "epoch": 1.6684559520050521,
      "grad_norm": 1.5640032291412354,
      "learning_rate": 3.701845484527402e-05,
      "loss": 0.8023,
      "step": 5284
    },
    {
      "epoch": 1.6687717082412377,
      "grad_norm": 1.4371219873428345,
      "learning_rate": 3.7014946319556525e-05,
      "loss": 0.8132,
      "step": 5285
    },
    {
      "epoch": 1.6690874644774234,
      "grad_norm": 1.751827597618103,
      "learning_rate": 3.701143779383903e-05,
      "loss": 0.8687,
      "step": 5286
    },
    {
      "epoch": 1.6694032207136091,
      "grad_norm": 1.5770227909088135,
      "learning_rate": 3.700792926812154e-05,
      "loss": 0.8481,
      "step": 5287
    },
    {
      "epoch": 1.6697189769497949,
      "grad_norm": 1.335878849029541,
      "learning_rate": 3.7004420742404045e-05,
      "loss": 0.7697,
      "step": 5288
    },
    {
      "epoch": 1.6700347331859804,
      "grad_norm": 1.4449381828308105,
      "learning_rate": 3.700091221668655e-05,
      "loss": 0.7977,
      "step": 5289
    },
    {
      "epoch": 1.6703504894221661,
      "grad_norm": 1.9156550168991089,
      "learning_rate": 3.699740369096906e-05,
      "loss": 0.792,
      "step": 5290
    },
    {
      "epoch": 1.6706662456583516,
      "grad_norm": 1.5881145000457764,
      "learning_rate": 3.6993895165251566e-05,
      "loss": 0.7728,
      "step": 5291
    },
    {
      "epoch": 1.6709820018945374,
      "grad_norm": 1.6013909578323364,
      "learning_rate": 3.699038663953407e-05,
      "loss": 0.8305,
      "step": 5292
    },
    {
      "epoch": 1.6712977581307231,
      "grad_norm": 1.5802183151245117,
      "learning_rate": 3.698687811381657e-05,
      "loss": 0.8469,
      "step": 5293
    },
    {
      "epoch": 1.6716135143669089,
      "grad_norm": 1.5333560705184937,
      "learning_rate": 3.698336958809908e-05,
      "loss": 0.8056,
      "step": 5294
    },
    {
      "epoch": 1.6719292706030944,
      "grad_norm": 1.5872652530670166,
      "learning_rate": 3.6979861062381586e-05,
      "loss": 0.8086,
      "step": 5295
    },
    {
      "epoch": 1.6722450268392801,
      "grad_norm": 1.9169071912765503,
      "learning_rate": 3.697635253666409e-05,
      "loss": 0.7731,
      "step": 5296
    },
    {
      "epoch": 1.6725607830754656,
      "grad_norm": 1.6068248748779297,
      "learning_rate": 3.69728440109466e-05,
      "loss": 0.7968,
      "step": 5297
    },
    {
      "epoch": 1.6728765393116514,
      "grad_norm": 1.4674222469329834,
      "learning_rate": 3.6969335485229106e-05,
      "loss": 0.7483,
      "step": 5298
    },
    {
      "epoch": 1.673192295547837,
      "grad_norm": 1.6474252939224243,
      "learning_rate": 3.696582695951161e-05,
      "loss": 0.8033,
      "step": 5299
    },
    {
      "epoch": 1.6735080517840228,
      "grad_norm": 1.4793139696121216,
      "learning_rate": 3.696231843379413e-05,
      "loss": 0.7571,
      "step": 5300
    },
    {
      "epoch": 1.6738238080202084,
      "grad_norm": 1.6707531213760376,
      "learning_rate": 3.6958809908076634e-05,
      "loss": 0.813,
      "step": 5301
    },
    {
      "epoch": 1.674139564256394,
      "grad_norm": 1.5448542833328247,
      "learning_rate": 3.6955301382359134e-05,
      "loss": 0.8489,
      "step": 5302
    },
    {
      "epoch": 1.6744553204925796,
      "grad_norm": 1.5530883073806763,
      "learning_rate": 3.695179285664164e-05,
      "loss": 0.9313,
      "step": 5303
    },
    {
      "epoch": 1.6747710767287654,
      "grad_norm": 1.7927192449569702,
      "learning_rate": 3.694828433092415e-05,
      "loss": 0.7749,
      "step": 5304
    },
    {
      "epoch": 1.675086832964951,
      "grad_norm": 1.9379810094833374,
      "learning_rate": 3.6944775805206654e-05,
      "loss": 0.7392,
      "step": 5305
    },
    {
      "epoch": 1.6754025892011368,
      "grad_norm": 1.6348607540130615,
      "learning_rate": 3.694126727948916e-05,
      "loss": 0.8782,
      "step": 5306
    },
    {
      "epoch": 1.6757183454373223,
      "grad_norm": 2.040285348892212,
      "learning_rate": 3.693775875377167e-05,
      "loss": 0.8116,
      "step": 5307
    },
    {
      "epoch": 1.676034101673508,
      "grad_norm": 2.0016911029815674,
      "learning_rate": 3.6934250228054175e-05,
      "loss": 0.8672,
      "step": 5308
    },
    {
      "epoch": 1.6763498579096936,
      "grad_norm": 2.1011314392089844,
      "learning_rate": 3.693074170233668e-05,
      "loss": 0.8857,
      "step": 5309
    },
    {
      "epoch": 1.6766656141458793,
      "grad_norm": 1.4649302959442139,
      "learning_rate": 3.692723317661919e-05,
      "loss": 0.7905,
      "step": 5310
    },
    {
      "epoch": 1.676981370382065,
      "grad_norm": 1.6796393394470215,
      "learning_rate": 3.6923724650901695e-05,
      "loss": 0.8504,
      "step": 5311
    },
    {
      "epoch": 1.6772971266182508,
      "grad_norm": 1.565549373626709,
      "learning_rate": 3.6920216125184195e-05,
      "loss": 0.8633,
      "step": 5312
    },
    {
      "epoch": 1.6776128828544363,
      "grad_norm": 1.6505177021026611,
      "learning_rate": 3.69167075994667e-05,
      "loss": 0.7332,
      "step": 5313
    },
    {
      "epoch": 1.677928639090622,
      "grad_norm": 1.4225116968154907,
      "learning_rate": 3.691319907374921e-05,
      "loss": 0.8254,
      "step": 5314
    },
    {
      "epoch": 1.6782443953268076,
      "grad_norm": 1.7235530614852905,
      "learning_rate": 3.6909690548031715e-05,
      "loss": 0.8881,
      "step": 5315
    },
    {
      "epoch": 1.6785601515629933,
      "grad_norm": 1.3982694149017334,
      "learning_rate": 3.690618202231422e-05,
      "loss": 0.8106,
      "step": 5316
    },
    {
      "epoch": 1.678875907799179,
      "grad_norm": 1.535114049911499,
      "learning_rate": 3.690267349659673e-05,
      "loss": 0.8213,
      "step": 5317
    },
    {
      "epoch": 1.6791916640353648,
      "grad_norm": 1.7203683853149414,
      "learning_rate": 3.689916497087924e-05,
      "loss": 0.8048,
      "step": 5318
    },
    {
      "epoch": 1.6795074202715503,
      "grad_norm": 1.5576292276382446,
      "learning_rate": 3.689565644516175e-05,
      "loss": 0.8353,
      "step": 5319
    },
    {
      "epoch": 1.679823176507736,
      "grad_norm": 1.5687229633331299,
      "learning_rate": 3.6892147919444256e-05,
      "loss": 0.8517,
      "step": 5320
    },
    {
      "epoch": 1.6801389327439216,
      "grad_norm": 1.4814887046813965,
      "learning_rate": 3.6888639393726756e-05,
      "loss": 0.8274,
      "step": 5321
    },
    {
      "epoch": 1.6804546889801073,
      "grad_norm": 1.7158751487731934,
      "learning_rate": 3.688513086800926e-05,
      "loss": 0.7819,
      "step": 5322
    },
    {
      "epoch": 1.680770445216293,
      "grad_norm": 1.442016363143921,
      "learning_rate": 3.688162234229177e-05,
      "loss": 0.834,
      "step": 5323
    },
    {
      "epoch": 1.6810862014524788,
      "grad_norm": 1.8938161134719849,
      "learning_rate": 3.687811381657428e-05,
      "loss": 0.8848,
      "step": 5324
    },
    {
      "epoch": 1.6814019576886645,
      "grad_norm": 1.7090107202529907,
      "learning_rate": 3.6874605290856784e-05,
      "loss": 0.8473,
      "step": 5325
    },
    {
      "epoch": 1.68171771392485,
      "grad_norm": 1.5245699882507324,
      "learning_rate": 3.687109676513929e-05,
      "loss": 0.8295,
      "step": 5326
    },
    {
      "epoch": 1.6820334701610355,
      "grad_norm": 1.683974027633667,
      "learning_rate": 3.68675882394218e-05,
      "loss": 0.8466,
      "step": 5327
    },
    {
      "epoch": 1.6823492263972213,
      "grad_norm": 1.4824546575546265,
      "learning_rate": 3.6864079713704304e-05,
      "loss": 0.778,
      "step": 5328
    },
    {
      "epoch": 1.682664982633407,
      "grad_norm": 1.4967994689941406,
      "learning_rate": 3.686057118798681e-05,
      "loss": 0.7572,
      "step": 5329
    },
    {
      "epoch": 1.6829807388695928,
      "grad_norm": 1.6956239938735962,
      "learning_rate": 3.685706266226932e-05,
      "loss": 0.8227,
      "step": 5330
    },
    {
      "epoch": 1.6832964951057785,
      "grad_norm": 1.7145665884017944,
      "learning_rate": 3.685355413655182e-05,
      "loss": 0.9125,
      "step": 5331
    },
    {
      "epoch": 1.683612251341964,
      "grad_norm": 1.6463778018951416,
      "learning_rate": 3.6850045610834325e-05,
      "loss": 0.7953,
      "step": 5332
    },
    {
      "epoch": 1.6839280075781495,
      "grad_norm": 1.516180396080017,
      "learning_rate": 3.684653708511683e-05,
      "loss": 0.7917,
      "step": 5333
    },
    {
      "epoch": 1.6842437638143353,
      "grad_norm": 1.6173187494277954,
      "learning_rate": 3.684302855939934e-05,
      "loss": 0.788,
      "step": 5334
    },
    {
      "epoch": 1.684559520050521,
      "grad_norm": 1.5720996856689453,
      "learning_rate": 3.683952003368185e-05,
      "loss": 0.8682,
      "step": 5335
    },
    {
      "epoch": 1.6848752762867067,
      "grad_norm": 1.7636570930480957,
      "learning_rate": 3.683601150796436e-05,
      "loss": 0.7052,
      "step": 5336
    },
    {
      "epoch": 1.6851910325228925,
      "grad_norm": 1.5413808822631836,
      "learning_rate": 3.6832502982246865e-05,
      "loss": 0.8232,
      "step": 5337
    },
    {
      "epoch": 1.685506788759078,
      "grad_norm": 1.6175262928009033,
      "learning_rate": 3.682899445652937e-05,
      "loss": 0.7691,
      "step": 5338
    },
    {
      "epoch": 1.6858225449952635,
      "grad_norm": 1.8955190181732178,
      "learning_rate": 3.682548593081188e-05,
      "loss": 0.8033,
      "step": 5339
    },
    {
      "epoch": 1.6861383012314493,
      "grad_norm": 1.503516435623169,
      "learning_rate": 3.682197740509438e-05,
      "loss": 0.8548,
      "step": 5340
    },
    {
      "epoch": 1.686454057467635,
      "grad_norm": 1.6797317266464233,
      "learning_rate": 3.6818468879376886e-05,
      "loss": 0.8981,
      "step": 5341
    },
    {
      "epoch": 1.6867698137038207,
      "grad_norm": 1.4207340478897095,
      "learning_rate": 3.681496035365939e-05,
      "loss": 0.8285,
      "step": 5342
    },
    {
      "epoch": 1.6870855699400065,
      "grad_norm": 1.467647910118103,
      "learning_rate": 3.68114518279419e-05,
      "loss": 0.7828,
      "step": 5343
    },
    {
      "epoch": 1.687401326176192,
      "grad_norm": 1.8754521608352661,
      "learning_rate": 3.6807943302224406e-05,
      "loss": 0.7803,
      "step": 5344
    },
    {
      "epoch": 1.6877170824123775,
      "grad_norm": 1.5523757934570312,
      "learning_rate": 3.680443477650691e-05,
      "loss": 0.8502,
      "step": 5345
    },
    {
      "epoch": 1.6880328386485632,
      "grad_norm": 1.7044315338134766,
      "learning_rate": 3.680092625078942e-05,
      "loss": 0.7931,
      "step": 5346
    },
    {
      "epoch": 1.688348594884749,
      "grad_norm": 1.7536476850509644,
      "learning_rate": 3.679741772507193e-05,
      "loss": 0.8914,
      "step": 5347
    },
    {
      "epoch": 1.6886643511209347,
      "grad_norm": 1.5466448068618774,
      "learning_rate": 3.6793909199354434e-05,
      "loss": 0.7603,
      "step": 5348
    },
    {
      "epoch": 1.6889801073571205,
      "grad_norm": 1.7107092142105103,
      "learning_rate": 3.679040067363694e-05,
      "loss": 0.815,
      "step": 5349
    },
    {
      "epoch": 1.689295863593306,
      "grad_norm": 1.6244200468063354,
      "learning_rate": 3.678689214791945e-05,
      "loss": 0.8037,
      "step": 5350
    },
    {
      "epoch": 1.6896116198294915,
      "grad_norm": 1.4440016746520996,
      "learning_rate": 3.678338362220195e-05,
      "loss": 0.8196,
      "step": 5351
    },
    {
      "epoch": 1.6899273760656772,
      "grad_norm": 1.578189730644226,
      "learning_rate": 3.6779875096484454e-05,
      "loss": 0.8526,
      "step": 5352
    },
    {
      "epoch": 1.690243132301863,
      "grad_norm": 1.5124180316925049,
      "learning_rate": 3.677636657076697e-05,
      "loss": 0.8043,
      "step": 5353
    },
    {
      "epoch": 1.6905588885380487,
      "grad_norm": 1.521615743637085,
      "learning_rate": 3.6772858045049474e-05,
      "loss": 0.8411,
      "step": 5354
    },
    {
      "epoch": 1.6908746447742344,
      "grad_norm": 1.6574499607086182,
      "learning_rate": 3.676934951933198e-05,
      "loss": 0.9287,
      "step": 5355
    },
    {
      "epoch": 1.69119040101042,
      "grad_norm": 1.5820316076278687,
      "learning_rate": 3.676584099361449e-05,
      "loss": 0.8157,
      "step": 5356
    },
    {
      "epoch": 1.6915061572466055,
      "grad_norm": 1.6218682527542114,
      "learning_rate": 3.6762332467896995e-05,
      "loss": 0.8287,
      "step": 5357
    },
    {
      "epoch": 1.6918219134827912,
      "grad_norm": 1.6607186794281006,
      "learning_rate": 3.67588239421795e-05,
      "loss": 0.7311,
      "step": 5358
    },
    {
      "epoch": 1.692137669718977,
      "grad_norm": 1.4662470817565918,
      "learning_rate": 3.6755315416462e-05,
      "loss": 0.813,
      "step": 5359
    },
    {
      "epoch": 1.6924534259551627,
      "grad_norm": 1.8940560817718506,
      "learning_rate": 3.675180689074451e-05,
      "loss": 0.8181,
      "step": 5360
    },
    {
      "epoch": 1.6927691821913484,
      "grad_norm": 1.6828205585479736,
      "learning_rate": 3.6748298365027015e-05,
      "loss": 0.9507,
      "step": 5361
    },
    {
      "epoch": 1.693084938427534,
      "grad_norm": 1.4755808115005493,
      "learning_rate": 3.674478983930952e-05,
      "loss": 0.82,
      "step": 5362
    },
    {
      "epoch": 1.6934006946637195,
      "grad_norm": 1.547800898551941,
      "learning_rate": 3.674128131359203e-05,
      "loss": 0.8399,
      "step": 5363
    },
    {
      "epoch": 1.6937164508999052,
      "grad_norm": 1.6143141984939575,
      "learning_rate": 3.6737772787874536e-05,
      "loss": 0.8844,
      "step": 5364
    },
    {
      "epoch": 1.694032207136091,
      "grad_norm": 1.523942232131958,
      "learning_rate": 3.673426426215704e-05,
      "loss": 0.8133,
      "step": 5365
    },
    {
      "epoch": 1.6943479633722767,
      "grad_norm": 1.593870997428894,
      "learning_rate": 3.673075573643955e-05,
      "loss": 0.7958,
      "step": 5366
    },
    {
      "epoch": 1.6946637196084624,
      "grad_norm": 1.6371946334838867,
      "learning_rate": 3.6727247210722056e-05,
      "loss": 0.8113,
      "step": 5367
    },
    {
      "epoch": 1.694979475844648,
      "grad_norm": 1.2689663171768188,
      "learning_rate": 3.672373868500456e-05,
      "loss": 0.8137,
      "step": 5368
    },
    {
      "epoch": 1.6952952320808334,
      "grad_norm": 1.6768923997879028,
      "learning_rate": 3.672023015928707e-05,
      "loss": 0.8211,
      "step": 5369
    },
    {
      "epoch": 1.6956109883170192,
      "grad_norm": 2.0047414302825928,
      "learning_rate": 3.671672163356958e-05,
      "loss": 0.9108,
      "step": 5370
    },
    {
      "epoch": 1.695926744553205,
      "grad_norm": 1.4266271591186523,
      "learning_rate": 3.6713213107852083e-05,
      "loss": 0.8073,
      "step": 5371
    },
    {
      "epoch": 1.6962425007893907,
      "grad_norm": 1.5985187292099,
      "learning_rate": 3.670970458213459e-05,
      "loss": 0.7598,
      "step": 5372
    },
    {
      "epoch": 1.6965582570255764,
      "grad_norm": 1.5697962045669556,
      "learning_rate": 3.67061960564171e-05,
      "loss": 0.8016,
      "step": 5373
    },
    {
      "epoch": 1.696874013261762,
      "grad_norm": 1.635013222694397,
      "learning_rate": 3.6702687530699604e-05,
      "loss": 0.8599,
      "step": 5374
    },
    {
      "epoch": 1.6971897694979474,
      "grad_norm": 1.995598554611206,
      "learning_rate": 3.669917900498211e-05,
      "loss": 0.8023,
      "step": 5375
    },
    {
      "epoch": 1.6975055257341332,
      "grad_norm": 1.5498782396316528,
      "learning_rate": 3.669567047926462e-05,
      "loss": 0.7854,
      "step": 5376
    },
    {
      "epoch": 1.697821281970319,
      "grad_norm": 1.9129539728164673,
      "learning_rate": 3.6692161953547124e-05,
      "loss": 0.8481,
      "step": 5377
    },
    {
      "epoch": 1.6981370382065046,
      "grad_norm": 1.6221389770507812,
      "learning_rate": 3.668865342782963e-05,
      "loss": 0.8679,
      "step": 5378
    },
    {
      "epoch": 1.6984527944426904,
      "grad_norm": 1.7163769006729126,
      "learning_rate": 3.668514490211213e-05,
      "loss": 0.8783,
      "step": 5379
    },
    {
      "epoch": 1.6987685506788759,
      "grad_norm": 1.5488862991333008,
      "learning_rate": 3.668163637639464e-05,
      "loss": 0.9409,
      "step": 5380
    },
    {
      "epoch": 1.6990843069150616,
      "grad_norm": 1.4185004234313965,
      "learning_rate": 3.6678127850677145e-05,
      "loss": 0.797,
      "step": 5381
    },
    {
      "epoch": 1.6994000631512471,
      "grad_norm": 1.5311391353607178,
      "learning_rate": 3.667461932495965e-05,
      "loss": 0.8488,
      "step": 5382
    },
    {
      "epoch": 1.6997158193874329,
      "grad_norm": 1.4914088249206543,
      "learning_rate": 3.667111079924216e-05,
      "loss": 0.8203,
      "step": 5383
    },
    {
      "epoch": 1.7000315756236186,
      "grad_norm": 1.6261134147644043,
      "learning_rate": 3.6667602273524665e-05,
      "loss": 0.9146,
      "step": 5384
    },
    {
      "epoch": 1.7003473318598044,
      "grad_norm": 1.5764397382736206,
      "learning_rate": 3.666409374780717e-05,
      "loss": 0.8374,
      "step": 5385
    },
    {
      "epoch": 1.7006630880959899,
      "grad_norm": 1.735280990600586,
      "learning_rate": 3.666058522208968e-05,
      "loss": 0.8336,
      "step": 5386
    },
    {
      "epoch": 1.7009788443321756,
      "grad_norm": 1.802237868309021,
      "learning_rate": 3.6657076696372186e-05,
      "loss": 0.8267,
      "step": 5387
    },
    {
      "epoch": 1.7012946005683611,
      "grad_norm": 1.7732261419296265,
      "learning_rate": 3.665356817065469e-05,
      "loss": 0.8066,
      "step": 5388
    },
    {
      "epoch": 1.7016103568045469,
      "grad_norm": 1.6837910413742065,
      "learning_rate": 3.66500596449372e-05,
      "loss": 0.8137,
      "step": 5389
    },
    {
      "epoch": 1.7019261130407326,
      "grad_norm": 1.5621986389160156,
      "learning_rate": 3.6646551119219706e-05,
      "loss": 0.8893,
      "step": 5390
    },
    {
      "epoch": 1.7022418692769183,
      "grad_norm": 1.5226351022720337,
      "learning_rate": 3.664304259350221e-05,
      "loss": 0.7827,
      "step": 5391
    },
    {
      "epoch": 1.7025576255131039,
      "grad_norm": 1.360098123550415,
      "learning_rate": 3.663953406778472e-05,
      "loss": 0.7639,
      "step": 5392
    },
    {
      "epoch": 1.7028733817492896,
      "grad_norm": 1.5372247695922852,
      "learning_rate": 3.6636025542067227e-05,
      "loss": 0.8428,
      "step": 5393
    },
    {
      "epoch": 1.703189137985475,
      "grad_norm": 1.3933289051055908,
      "learning_rate": 3.663251701634973e-05,
      "loss": 0.7155,
      "step": 5394
    },
    {
      "epoch": 1.7035048942216608,
      "grad_norm": 1.6725527048110962,
      "learning_rate": 3.662900849063224e-05,
      "loss": 0.8223,
      "step": 5395
    },
    {
      "epoch": 1.7038206504578466,
      "grad_norm": 1.5792361497879028,
      "learning_rate": 3.662549996491475e-05,
      "loss": 0.8816,
      "step": 5396
    },
    {
      "epoch": 1.7041364066940323,
      "grad_norm": 1.5207762718200684,
      "learning_rate": 3.6621991439197254e-05,
      "loss": 0.7673,
      "step": 5397
    },
    {
      "epoch": 1.7044521629302178,
      "grad_norm": 1.8278863430023193,
      "learning_rate": 3.6618482913479754e-05,
      "loss": 0.8558,
      "step": 5398
    },
    {
      "epoch": 1.7047679191664036,
      "grad_norm": 1.545057773590088,
      "learning_rate": 3.661497438776226e-05,
      "loss": 0.7816,
      "step": 5399
    },
    {
      "epoch": 1.705083675402589,
      "grad_norm": 1.7084436416625977,
      "learning_rate": 3.661146586204477e-05,
      "loss": 0.8003,
      "step": 5400
    },
    {
      "epoch": 1.705083675402589,
      "eval_loss": 0.8066611289978027,
      "eval_runtime": 4.6441,
      "eval_samples_per_second": 536.17,
      "eval_steps_per_second": 4.522,
      "step": 5400
    },
    {
      "epoch": 1.7053994316387748,
      "grad_norm": 1.5998367071151733,
      "learning_rate": 3.6607957336327274e-05,
      "loss": 0.9776,
      "step": 5401
    },
    {
      "epoch": 1.7057151878749606,
      "grad_norm": 1.6852201223373413,
      "learning_rate": 3.660444881060978e-05,
      "loss": 0.9129,
      "step": 5402
    },
    {
      "epoch": 1.7060309441111463,
      "grad_norm": 1.6975455284118652,
      "learning_rate": 3.660094028489229e-05,
      "loss": 0.8612,
      "step": 5403
    },
    {
      "epoch": 1.7063467003473318,
      "grad_norm": 1.571816086769104,
      "learning_rate": 3.6597431759174795e-05,
      "loss": 0.8384,
      "step": 5404
    },
    {
      "epoch": 1.7066624565835176,
      "grad_norm": 1.6761680841445923,
      "learning_rate": 3.659392323345731e-05,
      "loss": 0.8241,
      "step": 5405
    },
    {
      "epoch": 1.706978212819703,
      "grad_norm": 1.606250286102295,
      "learning_rate": 3.6590414707739815e-05,
      "loss": 0.7533,
      "step": 5406
    },
    {
      "epoch": 1.7072939690558888,
      "grad_norm": 1.435547113418579,
      "learning_rate": 3.6586906182022315e-05,
      "loss": 0.7925,
      "step": 5407
    },
    {
      "epoch": 1.7076097252920746,
      "grad_norm": 1.520980954170227,
      "learning_rate": 3.658339765630482e-05,
      "loss": 0.7215,
      "step": 5408
    },
    {
      "epoch": 1.7079254815282603,
      "grad_norm": 1.4826123714447021,
      "learning_rate": 3.657988913058733e-05,
      "loss": 0.8533,
      "step": 5409
    },
    {
      "epoch": 1.7082412377644458,
      "grad_norm": 1.454796314239502,
      "learning_rate": 3.6576380604869836e-05,
      "loss": 0.8113,
      "step": 5410
    },
    {
      "epoch": 1.7085569940006315,
      "grad_norm": 1.6367172002792358,
      "learning_rate": 3.657287207915234e-05,
      "loss": 0.7758,
      "step": 5411
    },
    {
      "epoch": 1.708872750236817,
      "grad_norm": 1.5822339057922363,
      "learning_rate": 3.656936355343485e-05,
      "loss": 0.7851,
      "step": 5412
    },
    {
      "epoch": 1.7091885064730028,
      "grad_norm": 1.5443233251571655,
      "learning_rate": 3.6565855027717356e-05,
      "loss": 0.6597,
      "step": 5413
    },
    {
      "epoch": 1.7095042627091885,
      "grad_norm": 1.4669326543807983,
      "learning_rate": 3.656234650199986e-05,
      "loss": 0.8106,
      "step": 5414
    },
    {
      "epoch": 1.7098200189453743,
      "grad_norm": 1.5438441038131714,
      "learning_rate": 3.655883797628237e-05,
      "loss": 0.8327,
      "step": 5415
    },
    {
      "epoch": 1.7101357751815598,
      "grad_norm": 1.5410534143447876,
      "learning_rate": 3.6555329450564876e-05,
      "loss": 0.7846,
      "step": 5416
    },
    {
      "epoch": 1.7104515314177455,
      "grad_norm": 1.7028733491897583,
      "learning_rate": 3.6551820924847377e-05,
      "loss": 0.8289,
      "step": 5417
    },
    {
      "epoch": 1.710767287653931,
      "grad_norm": 1.601044774055481,
      "learning_rate": 3.654831239912988e-05,
      "loss": 0.7382,
      "step": 5418
    },
    {
      "epoch": 1.7110830438901168,
      "grad_norm": 1.999097228050232,
      "learning_rate": 3.654480387341239e-05,
      "loss": 0.8339,
      "step": 5419
    },
    {
      "epoch": 1.7113988001263025,
      "grad_norm": 1.6137597560882568,
      "learning_rate": 3.65412953476949e-05,
      "loss": 0.7907,
      "step": 5420
    },
    {
      "epoch": 1.7117145563624883,
      "grad_norm": 1.7758296728134155,
      "learning_rate": 3.6537786821977404e-05,
      "loss": 0.8247,
      "step": 5421
    },
    {
      "epoch": 1.7120303125986738,
      "grad_norm": 1.618868112564087,
      "learning_rate": 3.653427829625991e-05,
      "loss": 0.8392,
      "step": 5422
    },
    {
      "epoch": 1.7123460688348595,
      "grad_norm": 1.6522029638290405,
      "learning_rate": 3.6530769770542424e-05,
      "loss": 0.8142,
      "step": 5423
    },
    {
      "epoch": 1.712661825071045,
      "grad_norm": 1.4241007566452026,
      "learning_rate": 3.652726124482493e-05,
      "loss": 0.806,
      "step": 5424
    },
    {
      "epoch": 1.7129775813072308,
      "grad_norm": 1.5827754735946655,
      "learning_rate": 3.652375271910744e-05,
      "loss": 0.8195,
      "step": 5425
    },
    {
      "epoch": 1.7132933375434165,
      "grad_norm": 1.3041330575942993,
      "learning_rate": 3.652024419338994e-05,
      "loss": 0.7334,
      "step": 5426
    },
    {
      "epoch": 1.7136090937796022,
      "grad_norm": 1.6899603605270386,
      "learning_rate": 3.6516735667672445e-05,
      "loss": 0.7656,
      "step": 5427
    },
    {
      "epoch": 1.7139248500157878,
      "grad_norm": 1.5089499950408936,
      "learning_rate": 3.651322714195495e-05,
      "loss": 0.8739,
      "step": 5428
    },
    {
      "epoch": 1.7142406062519735,
      "grad_norm": 2.0166261196136475,
      "learning_rate": 3.650971861623746e-05,
      "loss": 0.7836,
      "step": 5429
    },
    {
      "epoch": 1.714556362488159,
      "grad_norm": 1.6232038736343384,
      "learning_rate": 3.6506210090519965e-05,
      "loss": 0.7895,
      "step": 5430
    },
    {
      "epoch": 1.7148721187243448,
      "grad_norm": 1.6756688356399536,
      "learning_rate": 3.650270156480247e-05,
      "loss": 0.8159,
      "step": 5431
    },
    {
      "epoch": 1.7151878749605305,
      "grad_norm": 1.6273654699325562,
      "learning_rate": 3.649919303908498e-05,
      "loss": 0.8007,
      "step": 5432
    },
    {
      "epoch": 1.7155036311967162,
      "grad_norm": 1.743805170059204,
      "learning_rate": 3.6495684513367486e-05,
      "loss": 0.8152,
      "step": 5433
    },
    {
      "epoch": 1.715819387432902,
      "grad_norm": 1.544948697090149,
      "learning_rate": 3.649217598764999e-05,
      "loss": 0.8514,
      "step": 5434
    },
    {
      "epoch": 1.7161351436690875,
      "grad_norm": 1.5747015476226807,
      "learning_rate": 3.64886674619325e-05,
      "loss": 0.8302,
      "step": 5435
    },
    {
      "epoch": 1.716450899905273,
      "grad_norm": 1.7342487573623657,
      "learning_rate": 3.6485158936215e-05,
      "loss": 0.8805,
      "step": 5436
    },
    {
      "epoch": 1.7167666561414587,
      "grad_norm": 1.7901781797409058,
      "learning_rate": 3.6481650410497506e-05,
      "loss": 0.8171,
      "step": 5437
    },
    {
      "epoch": 1.7170824123776445,
      "grad_norm": 1.677917718887329,
      "learning_rate": 3.647814188478001e-05,
      "loss": 0.8797,
      "step": 5438
    },
    {
      "epoch": 1.7173981686138302,
      "grad_norm": 1.866264820098877,
      "learning_rate": 3.647463335906252e-05,
      "loss": 0.8736,
      "step": 5439
    },
    {
      "epoch": 1.717713924850016,
      "grad_norm": 1.5915265083312988,
      "learning_rate": 3.647112483334503e-05,
      "loss": 0.829,
      "step": 5440
    },
    {
      "epoch": 1.7180296810862015,
      "grad_norm": 1.7068495750427246,
      "learning_rate": 3.646761630762754e-05,
      "loss": 0.8655,
      "step": 5441
    },
    {
      "epoch": 1.718345437322387,
      "grad_norm": 2.061764717102051,
      "learning_rate": 3.646410778191005e-05,
      "loss": 0.767,
      "step": 5442
    },
    {
      "epoch": 1.7186611935585727,
      "grad_norm": 1.7669097185134888,
      "learning_rate": 3.6460599256192554e-05,
      "loss": 0.7942,
      "step": 5443
    },
    {
      "epoch": 1.7189769497947585,
      "grad_norm": 1.6115165948867798,
      "learning_rate": 3.645709073047506e-05,
      "loss": 0.8786,
      "step": 5444
    },
    {
      "epoch": 1.7192927060309442,
      "grad_norm": 1.625283122062683,
      "learning_rate": 3.645358220475756e-05,
      "loss": 0.8747,
      "step": 5445
    },
    {
      "epoch": 1.71960846226713,
      "grad_norm": 1.6089359521865845,
      "learning_rate": 3.645007367904007e-05,
      "loss": 0.8389,
      "step": 5446
    },
    {
      "epoch": 1.7199242185033154,
      "grad_norm": 1.4785311222076416,
      "learning_rate": 3.6446565153322574e-05,
      "loss": 0.8374,
      "step": 5447
    },
    {
      "epoch": 1.720239974739501,
      "grad_norm": 1.440779685974121,
      "learning_rate": 3.644305662760508e-05,
      "loss": 0.8155,
      "step": 5448
    },
    {
      "epoch": 1.7205557309756867,
      "grad_norm": 1.5773162841796875,
      "learning_rate": 3.643954810188759e-05,
      "loss": 0.7939,
      "step": 5449
    },
    {
      "epoch": 1.7208714872118724,
      "grad_norm": 1.979898452758789,
      "learning_rate": 3.6436039576170095e-05,
      "loss": 0.8191,
      "step": 5450
    },
    {
      "epoch": 1.7211872434480582,
      "grad_norm": 1.6836721897125244,
      "learning_rate": 3.64325310504526e-05,
      "loss": 0.9252,
      "step": 5451
    },
    {
      "epoch": 1.721502999684244,
      "grad_norm": 1.6722766160964966,
      "learning_rate": 3.642902252473511e-05,
      "loss": 0.8877,
      "step": 5452
    },
    {
      "epoch": 1.7218187559204294,
      "grad_norm": 1.5422343015670776,
      "learning_rate": 3.6425513999017615e-05,
      "loss": 0.8143,
      "step": 5453
    },
    {
      "epoch": 1.722134512156615,
      "grad_norm": 1.8262113332748413,
      "learning_rate": 3.642200547330012e-05,
      "loss": 0.849,
      "step": 5454
    },
    {
      "epoch": 1.7224502683928007,
      "grad_norm": 1.4413524866104126,
      "learning_rate": 3.641849694758263e-05,
      "loss": 0.8017,
      "step": 5455
    },
    {
      "epoch": 1.7227660246289864,
      "grad_norm": 1.6190541982650757,
      "learning_rate": 3.641498842186513e-05,
      "loss": 0.8492,
      "step": 5456
    },
    {
      "epoch": 1.7230817808651722,
      "grad_norm": 1.5228039026260376,
      "learning_rate": 3.6411479896147635e-05,
      "loss": 0.843,
      "step": 5457
    },
    {
      "epoch": 1.723397537101358,
      "grad_norm": 1.4095529317855835,
      "learning_rate": 3.640797137043015e-05,
      "loss": 0.7507,
      "step": 5458
    },
    {
      "epoch": 1.7237132933375434,
      "grad_norm": 1.422058343887329,
      "learning_rate": 3.6404462844712656e-05,
      "loss": 0.7987,
      "step": 5459
    },
    {
      "epoch": 1.724029049573729,
      "grad_norm": 1.7065179347991943,
      "learning_rate": 3.640095431899516e-05,
      "loss": 0.8332,
      "step": 5460
    },
    {
      "epoch": 1.7243448058099147,
      "grad_norm": 1.485853672027588,
      "learning_rate": 3.639744579327767e-05,
      "loss": 0.7682,
      "step": 5461
    },
    {
      "epoch": 1.7246605620461004,
      "grad_norm": 1.5064595937728882,
      "learning_rate": 3.6393937267560176e-05,
      "loss": 0.7923,
      "step": 5462
    },
    {
      "epoch": 1.7249763182822861,
      "grad_norm": 1.6015124320983887,
      "learning_rate": 3.639042874184268e-05,
      "loss": 0.837,
      "step": 5463
    },
    {
      "epoch": 1.7252920745184719,
      "grad_norm": 1.3775262832641602,
      "learning_rate": 3.638692021612518e-05,
      "loss": 0.8211,
      "step": 5464
    },
    {
      "epoch": 1.7256078307546574,
      "grad_norm": 1.4006123542785645,
      "learning_rate": 3.638341169040769e-05,
      "loss": 0.7636,
      "step": 5465
    },
    {
      "epoch": 1.725923586990843,
      "grad_norm": 1.525894284248352,
      "learning_rate": 3.63799031646902e-05,
      "loss": 0.8275,
      "step": 5466
    },
    {
      "epoch": 1.7262393432270287,
      "grad_norm": 1.5481616258621216,
      "learning_rate": 3.6376394638972704e-05,
      "loss": 0.8059,
      "step": 5467
    },
    {
      "epoch": 1.7265550994632144,
      "grad_norm": 1.45096755027771,
      "learning_rate": 3.637288611325521e-05,
      "loss": 0.7727,
      "step": 5468
    },
    {
      "epoch": 1.7268708556994001,
      "grad_norm": 1.5742859840393066,
      "learning_rate": 3.636937758753772e-05,
      "loss": 0.8445,
      "step": 5469
    },
    {
      "epoch": 1.7271866119355859,
      "grad_norm": 1.4288631677627563,
      "learning_rate": 3.6365869061820224e-05,
      "loss": 0.8125,
      "step": 5470
    },
    {
      "epoch": 1.7275023681717714,
      "grad_norm": 1.8203970193862915,
      "learning_rate": 3.636236053610273e-05,
      "loss": 0.8226,
      "step": 5471
    },
    {
      "epoch": 1.727818124407957,
      "grad_norm": 1.6961601972579956,
      "learning_rate": 3.635885201038524e-05,
      "loss": 0.8898,
      "step": 5472
    },
    {
      "epoch": 1.7281338806441426,
      "grad_norm": 2.016982078552246,
      "learning_rate": 3.6355343484667745e-05,
      "loss": 0.9445,
      "step": 5473
    },
    {
      "epoch": 1.7284496368803284,
      "grad_norm": 2.8959052562713623,
      "learning_rate": 3.635183495895025e-05,
      "loss": 0.7595,
      "step": 5474
    },
    {
      "epoch": 1.7287653931165141,
      "grad_norm": 1.3755466938018799,
      "learning_rate": 3.634832643323276e-05,
      "loss": 0.8124,
      "step": 5475
    },
    {
      "epoch": 1.7290811493526999,
      "grad_norm": 1.6468219757080078,
      "learning_rate": 3.6344817907515265e-05,
      "loss": 0.8798,
      "step": 5476
    },
    {
      "epoch": 1.7293969055888854,
      "grad_norm": 1.5688143968582153,
      "learning_rate": 3.634130938179777e-05,
      "loss": 0.8281,
      "step": 5477
    },
    {
      "epoch": 1.7297126618250709,
      "grad_norm": 1.6109962463378906,
      "learning_rate": 3.633780085608028e-05,
      "loss": 0.8618,
      "step": 5478
    },
    {
      "epoch": 1.7300284180612566,
      "grad_norm": 1.5773261785507202,
      "learning_rate": 3.6334292330362785e-05,
      "loss": 0.798,
      "step": 5479
    },
    {
      "epoch": 1.7303441742974424,
      "grad_norm": 1.577972412109375,
      "learning_rate": 3.633078380464529e-05,
      "loss": 0.8812,
      "step": 5480
    },
    {
      "epoch": 1.730659930533628,
      "grad_norm": 1.5452653169631958,
      "learning_rate": 3.63272752789278e-05,
      "loss": 0.8343,
      "step": 5481
    },
    {
      "epoch": 1.7309756867698138,
      "grad_norm": 1.6169922351837158,
      "learning_rate": 3.6323766753210306e-05,
      "loss": 0.8314,
      "step": 5482
    },
    {
      "epoch": 1.7312914430059994,
      "grad_norm": 1.6311897039413452,
      "learning_rate": 3.632025822749281e-05,
      "loss": 0.827,
      "step": 5483
    },
    {
      "epoch": 1.7316071992421849,
      "grad_norm": 1.755847454071045,
      "learning_rate": 3.631674970177531e-05,
      "loss": 0.8727,
      "step": 5484
    },
    {
      "epoch": 1.7319229554783706,
      "grad_norm": 1.5499932765960693,
      "learning_rate": 3.631324117605782e-05,
      "loss": 0.8099,
      "step": 5485
    },
    {
      "epoch": 1.7322387117145563,
      "grad_norm": 1.9953112602233887,
      "learning_rate": 3.6309732650340326e-05,
      "loss": 0.8158,
      "step": 5486
    },
    {
      "epoch": 1.732554467950742,
      "grad_norm": 1.3896106481552124,
      "learning_rate": 3.630622412462283e-05,
      "loss": 0.8331,
      "step": 5487
    },
    {
      "epoch": 1.7328702241869278,
      "grad_norm": 1.7111977338790894,
      "learning_rate": 3.630271559890534e-05,
      "loss": 0.7958,
      "step": 5488
    },
    {
      "epoch": 1.7331859804231133,
      "grad_norm": 1.4546087980270386,
      "learning_rate": 3.629920707318785e-05,
      "loss": 0.8364,
      "step": 5489
    },
    {
      "epoch": 1.733501736659299,
      "grad_norm": 1.6529803276062012,
      "learning_rate": 3.6295698547470354e-05,
      "loss": 0.7831,
      "step": 5490
    },
    {
      "epoch": 1.7338174928954846,
      "grad_norm": 1.4933691024780273,
      "learning_rate": 3.629219002175286e-05,
      "loss": 0.9758,
      "step": 5491
    },
    {
      "epoch": 1.7341332491316703,
      "grad_norm": 1.4601681232452393,
      "learning_rate": 3.628868149603537e-05,
      "loss": 0.8652,
      "step": 5492
    },
    {
      "epoch": 1.734449005367856,
      "grad_norm": 1.9701600074768066,
      "learning_rate": 3.6285172970317874e-05,
      "loss": 0.8308,
      "step": 5493
    },
    {
      "epoch": 1.7347647616040418,
      "grad_norm": 1.4474767446517944,
      "learning_rate": 3.628166444460038e-05,
      "loss": 0.7962,
      "step": 5494
    },
    {
      "epoch": 1.7350805178402273,
      "grad_norm": 1.443659782409668,
      "learning_rate": 3.627815591888289e-05,
      "loss": 0.8167,
      "step": 5495
    },
    {
      "epoch": 1.735396274076413,
      "grad_norm": 1.6698609590530396,
      "learning_rate": 3.6274647393165394e-05,
      "loss": 0.8071,
      "step": 5496
    },
    {
      "epoch": 1.7357120303125986,
      "grad_norm": 1.4016203880310059,
      "learning_rate": 3.62711388674479e-05,
      "loss": 0.7798,
      "step": 5497
    },
    {
      "epoch": 1.7360277865487843,
      "grad_norm": 1.4374622106552124,
      "learning_rate": 3.626763034173041e-05,
      "loss": 0.77,
      "step": 5498
    },
    {
      "epoch": 1.73634354278497,
      "grad_norm": 1.4344229698181152,
      "learning_rate": 3.6264121816012915e-05,
      "loss": 0.833,
      "step": 5499
    },
    {
      "epoch": 1.7366592990211558,
      "grad_norm": 1.3642518520355225,
      "learning_rate": 3.626061329029542e-05,
      "loss": 0.8197,
      "step": 5500
    },
    {
      "epoch": 1.7369750552573413,
      "grad_norm": 1.505143404006958,
      "learning_rate": 3.625710476457793e-05,
      "loss": 0.8345,
      "step": 5501
    },
    {
      "epoch": 1.737290811493527,
      "grad_norm": 1.3257687091827393,
      "learning_rate": 3.6253596238860435e-05,
      "loss": 0.816,
      "step": 5502
    },
    {
      "epoch": 1.7376065677297126,
      "grad_norm": 1.4487271308898926,
      "learning_rate": 3.6250087713142935e-05,
      "loss": 0.8181,
      "step": 5503
    },
    {
      "epoch": 1.7379223239658983,
      "grad_norm": 1.5473973751068115,
      "learning_rate": 3.624657918742544e-05,
      "loss": 0.8703,
      "step": 5504
    },
    {
      "epoch": 1.738238080202084,
      "grad_norm": 1.4732823371887207,
      "learning_rate": 3.624307066170795e-05,
      "loss": 0.8347,
      "step": 5505
    },
    {
      "epoch": 1.7385538364382698,
      "grad_norm": 1.7247536182403564,
      "learning_rate": 3.6239562135990456e-05,
      "loss": 0.8183,
      "step": 5506
    },
    {
      "epoch": 1.7388695926744553,
      "grad_norm": 1.490932583808899,
      "learning_rate": 3.623605361027296e-05,
      "loss": 0.8322,
      "step": 5507
    },
    {
      "epoch": 1.739185348910641,
      "grad_norm": 1.3252382278442383,
      "learning_rate": 3.623254508455547e-05,
      "loss": 0.8656,
      "step": 5508
    },
    {
      "epoch": 1.7395011051468265,
      "grad_norm": 1.6951146125793457,
      "learning_rate": 3.6229036558837976e-05,
      "loss": 0.8749,
      "step": 5509
    },
    {
      "epoch": 1.7398168613830123,
      "grad_norm": 1.7702513933181763,
      "learning_rate": 3.622552803312049e-05,
      "loss": 0.8906,
      "step": 5510
    },
    {
      "epoch": 1.740132617619198,
      "grad_norm": 1.7215651273727417,
      "learning_rate": 3.6222019507403e-05,
      "loss": 0.8338,
      "step": 5511
    },
    {
      "epoch": 1.7404483738553838,
      "grad_norm": 1.6573505401611328,
      "learning_rate": 3.62185109816855e-05,
      "loss": 0.8391,
      "step": 5512
    },
    {
      "epoch": 1.7407641300915693,
      "grad_norm": 1.384750485420227,
      "learning_rate": 3.6215002455968003e-05,
      "loss": 0.8666,
      "step": 5513
    },
    {
      "epoch": 1.741079886327755,
      "grad_norm": 1.572790503501892,
      "learning_rate": 3.621149393025051e-05,
      "loss": 0.8269,
      "step": 5514
    },
    {
      "epoch": 1.7413956425639405,
      "grad_norm": 1.4838720560073853,
      "learning_rate": 3.620798540453302e-05,
      "loss": 0.8262,
      "step": 5515
    },
    {
      "epoch": 1.7417113988001263,
      "grad_norm": 1.5948370695114136,
      "learning_rate": 3.6204476878815524e-05,
      "loss": 0.883,
      "step": 5516
    },
    {
      "epoch": 1.742027155036312,
      "grad_norm": 1.6506969928741455,
      "learning_rate": 3.620096835309803e-05,
      "loss": 0.7918,
      "step": 5517
    },
    {
      "epoch": 1.7423429112724977,
      "grad_norm": 1.5805619955062866,
      "learning_rate": 3.619745982738054e-05,
      "loss": 0.7616,
      "step": 5518
    },
    {
      "epoch": 1.7426586675086833,
      "grad_norm": 1.5988215208053589,
      "learning_rate": 3.6193951301663044e-05,
      "loss": 0.7918,
      "step": 5519
    },
    {
      "epoch": 1.742974423744869,
      "grad_norm": 1.4710763692855835,
      "learning_rate": 3.619044277594555e-05,
      "loss": 0.8166,
      "step": 5520
    },
    {
      "epoch": 1.7432901799810545,
      "grad_norm": 1.7234500646591187,
      "learning_rate": 3.618693425022806e-05,
      "loss": 0.8136,
      "step": 5521
    },
    {
      "epoch": 1.7436059362172402,
      "grad_norm": 1.5908420085906982,
      "learning_rate": 3.618342572451056e-05,
      "loss": 0.8271,
      "step": 5522
    },
    {
      "epoch": 1.743921692453426,
      "grad_norm": 1.610688328742981,
      "learning_rate": 3.6179917198793065e-05,
      "loss": 0.7146,
      "step": 5523
    },
    {
      "epoch": 1.7442374486896117,
      "grad_norm": 1.9606201648712158,
      "learning_rate": 3.617640867307557e-05,
      "loss": 0.869,
      "step": 5524
    },
    {
      "epoch": 1.7445532049257972,
      "grad_norm": 1.6325223445892334,
      "learning_rate": 3.617290014735808e-05,
      "loss": 0.8743,
      "step": 5525
    },
    {
      "epoch": 1.744868961161983,
      "grad_norm": 1.7897604703903198,
      "learning_rate": 3.6169391621640585e-05,
      "loss": 0.8518,
      "step": 5526
    },
    {
      "epoch": 1.7451847173981685,
      "grad_norm": 6.693898677825928,
      "learning_rate": 3.616588309592309e-05,
      "loss": 0.8313,
      "step": 5527
    },
    {
      "epoch": 1.7455004736343542,
      "grad_norm": 1.8072121143341064,
      "learning_rate": 3.6162374570205606e-05,
      "loss": 0.8483,
      "step": 5528
    },
    {
      "epoch": 1.74581622987054,
      "grad_norm": 1.480787992477417,
      "learning_rate": 3.615886604448811e-05,
      "loss": 0.7626,
      "step": 5529
    },
    {
      "epoch": 1.7461319861067257,
      "grad_norm": 1.6044607162475586,
      "learning_rate": 3.615535751877062e-05,
      "loss": 0.7671,
      "step": 5530
    },
    {
      "epoch": 1.7464477423429112,
      "grad_norm": 2.054748058319092,
      "learning_rate": 3.615184899305312e-05,
      "loss": 0.8779,
      "step": 5531
    },
    {
      "epoch": 1.746763498579097,
      "grad_norm": 1.8606470823287964,
      "learning_rate": 3.6148340467335626e-05,
      "loss": 0.8152,
      "step": 5532
    },
    {
      "epoch": 1.7470792548152825,
      "grad_norm": 1.4883506298065186,
      "learning_rate": 3.614483194161813e-05,
      "loss": 0.8045,
      "step": 5533
    },
    {
      "epoch": 1.7473950110514682,
      "grad_norm": 1.3926432132720947,
      "learning_rate": 3.614132341590064e-05,
      "loss": 0.7478,
      "step": 5534
    },
    {
      "epoch": 1.747710767287654,
      "grad_norm": 1.6653136014938354,
      "learning_rate": 3.6137814890183147e-05,
      "loss": 0.8184,
      "step": 5535
    },
    {
      "epoch": 1.7480265235238397,
      "grad_norm": 1.5248104333877563,
      "learning_rate": 3.6134306364465653e-05,
      "loss": 0.8559,
      "step": 5536
    },
    {
      "epoch": 1.7483422797600252,
      "grad_norm": 2.0515177249908447,
      "learning_rate": 3.613079783874816e-05,
      "loss": 0.8138,
      "step": 5537
    },
    {
      "epoch": 1.748658035996211,
      "grad_norm": 1.7027186155319214,
      "learning_rate": 3.612728931303067e-05,
      "loss": 0.8354,
      "step": 5538
    },
    {
      "epoch": 1.7489737922323965,
      "grad_norm": 1.744339108467102,
      "learning_rate": 3.6123780787313174e-05,
      "loss": 0.8767,
      "step": 5539
    },
    {
      "epoch": 1.7492895484685822,
      "grad_norm": 1.5631431341171265,
      "learning_rate": 3.612027226159568e-05,
      "loss": 0.805,
      "step": 5540
    },
    {
      "epoch": 1.749605304704768,
      "grad_norm": 1.6552493572235107,
      "learning_rate": 3.611676373587819e-05,
      "loss": 0.8093,
      "step": 5541
    },
    {
      "epoch": 1.7499210609409537,
      "grad_norm": 1.699126124382019,
      "learning_rate": 3.611325521016069e-05,
      "loss": 0.7502,
      "step": 5542
    },
    {
      "epoch": 1.7502368171771394,
      "grad_norm": 1.6203258037567139,
      "learning_rate": 3.6109746684443194e-05,
      "loss": 0.8239,
      "step": 5543
    },
    {
      "epoch": 1.750552573413325,
      "grad_norm": 1.623042106628418,
      "learning_rate": 3.61062381587257e-05,
      "loss": 0.7534,
      "step": 5544
    },
    {
      "epoch": 1.7508683296495104,
      "grad_norm": 1.8528577089309692,
      "learning_rate": 3.6102729633008215e-05,
      "loss": 0.8422,
      "step": 5545
    },
    {
      "epoch": 1.7511840858856962,
      "grad_norm": 1.8386491537094116,
      "learning_rate": 3.609922110729072e-05,
      "loss": 0.8937,
      "step": 5546
    },
    {
      "epoch": 1.751499842121882,
      "grad_norm": 1.8359670639038086,
      "learning_rate": 3.609571258157323e-05,
      "loss": 0.9166,
      "step": 5547
    },
    {
      "epoch": 1.7518155983580677,
      "grad_norm": 1.8004486560821533,
      "learning_rate": 3.6092204055855735e-05,
      "loss": 0.8563,
      "step": 5548
    },
    {
      "epoch": 1.7521313545942534,
      "grad_norm": 2.2515616416931152,
      "learning_rate": 3.608869553013824e-05,
      "loss": 0.9053,
      "step": 5549
    },
    {
      "epoch": 1.752447110830439,
      "grad_norm": 1.643393874168396,
      "learning_rate": 3.608518700442074e-05,
      "loss": 0.8408,
      "step": 5550
    },
    {
      "epoch": 1.7527628670666244,
      "grad_norm": 1.5445871353149414,
      "learning_rate": 3.608167847870325e-05,
      "loss": 0.782,
      "step": 5551
    },
    {
      "epoch": 1.7530786233028102,
      "grad_norm": 1.8021868467330933,
      "learning_rate": 3.6078169952985756e-05,
      "loss": 0.8545,
      "step": 5552
    },
    {
      "epoch": 1.753394379538996,
      "grad_norm": 1.5857800245285034,
      "learning_rate": 3.607466142726826e-05,
      "loss": 0.8276,
      "step": 5553
    },
    {
      "epoch": 1.7537101357751816,
      "grad_norm": 1.535327434539795,
      "learning_rate": 3.607115290155077e-05,
      "loss": 0.8434,
      "step": 5554
    },
    {
      "epoch": 1.7540258920113674,
      "grad_norm": 1.453522801399231,
      "learning_rate": 3.6067644375833276e-05,
      "loss": 0.7758,
      "step": 5555
    },
    {
      "epoch": 1.754341648247553,
      "grad_norm": 1.793479084968567,
      "learning_rate": 3.606413585011578e-05,
      "loss": 0.7962,
      "step": 5556
    },
    {
      "epoch": 1.7546574044837384,
      "grad_norm": 1.5809346437454224,
      "learning_rate": 3.606062732439829e-05,
      "loss": 0.8035,
      "step": 5557
    },
    {
      "epoch": 1.7549731607199242,
      "grad_norm": 1.529865026473999,
      "learning_rate": 3.6057118798680797e-05,
      "loss": 0.7854,
      "step": 5558
    },
    {
      "epoch": 1.75528891695611,
      "grad_norm": 1.5832805633544922,
      "learning_rate": 3.60536102729633e-05,
      "loss": 0.7991,
      "step": 5559
    },
    {
      "epoch": 1.7556046731922956,
      "grad_norm": 1.5097464323043823,
      "learning_rate": 3.605010174724581e-05,
      "loss": 0.747,
      "step": 5560
    },
    {
      "epoch": 1.7559204294284814,
      "grad_norm": 1.8153765201568604,
      "learning_rate": 3.604659322152831e-05,
      "loss": 0.812,
      "step": 5561
    },
    {
      "epoch": 1.7562361856646669,
      "grad_norm": 1.6798441410064697,
      "learning_rate": 3.604308469581082e-05,
      "loss": 0.8954,
      "step": 5562
    },
    {
      "epoch": 1.7565519419008524,
      "grad_norm": 1.6463536024093628,
      "learning_rate": 3.603957617009333e-05,
      "loss": 0.8196,
      "step": 5563
    },
    {
      "epoch": 1.7568676981370381,
      "grad_norm": 1.6974390745162964,
      "learning_rate": 3.603606764437584e-05,
      "loss": 0.8208,
      "step": 5564
    },
    {
      "epoch": 1.7571834543732239,
      "grad_norm": 1.4082725048065186,
      "learning_rate": 3.6032559118658344e-05,
      "loss": 0.8665,
      "step": 5565
    },
    {
      "epoch": 1.7574992106094096,
      "grad_norm": 1.574893832206726,
      "learning_rate": 3.602905059294085e-05,
      "loss": 0.8875,
      "step": 5566
    },
    {
      "epoch": 1.7578149668455954,
      "grad_norm": 1.394145131111145,
      "learning_rate": 3.602554206722336e-05,
      "loss": 0.8051,
      "step": 5567
    },
    {
      "epoch": 1.7581307230817809,
      "grad_norm": 1.7446305751800537,
      "learning_rate": 3.6022033541505865e-05,
      "loss": 0.8114,
      "step": 5568
    },
    {
      "epoch": 1.7584464793179664,
      "grad_norm": 1.862027645111084,
      "learning_rate": 3.6018525015788365e-05,
      "loss": 0.8276,
      "step": 5569
    },
    {
      "epoch": 1.7587622355541521,
      "grad_norm": 1.2824172973632812,
      "learning_rate": 3.601501649007087e-05,
      "loss": 0.782,
      "step": 5570
    },
    {
      "epoch": 1.7590779917903379,
      "grad_norm": 1.4506739377975464,
      "learning_rate": 3.601150796435338e-05,
      "loss": 0.8185,
      "step": 5571
    },
    {
      "epoch": 1.7593937480265236,
      "grad_norm": 1.4948301315307617,
      "learning_rate": 3.6007999438635885e-05,
      "loss": 0.8479,
      "step": 5572
    },
    {
      "epoch": 1.7597095042627093,
      "grad_norm": 1.4741679430007935,
      "learning_rate": 3.600449091291839e-05,
      "loss": 0.7746,
      "step": 5573
    },
    {
      "epoch": 1.7600252604988949,
      "grad_norm": 1.3628517389297485,
      "learning_rate": 3.60009823872009e-05,
      "loss": 0.7887,
      "step": 5574
    },
    {
      "epoch": 1.7603410167350804,
      "grad_norm": 1.6605803966522217,
      "learning_rate": 3.5997473861483406e-05,
      "loss": 0.7646,
      "step": 5575
    },
    {
      "epoch": 1.760656772971266,
      "grad_norm": 1.6739106178283691,
      "learning_rate": 3.599396533576591e-05,
      "loss": 0.8425,
      "step": 5576
    },
    {
      "epoch": 1.7609725292074518,
      "grad_norm": 1.585147500038147,
      "learning_rate": 3.599045681004842e-05,
      "loss": 0.8039,
      "step": 5577
    },
    {
      "epoch": 1.7612882854436376,
      "grad_norm": 1.547345757484436,
      "learning_rate": 3.5986948284330926e-05,
      "loss": 0.8277,
      "step": 5578
    },
    {
      "epoch": 1.7616040416798233,
      "grad_norm": 1.6580921411514282,
      "learning_rate": 3.598343975861343e-05,
      "loss": 0.8245,
      "step": 5579
    },
    {
      "epoch": 1.7619197979160088,
      "grad_norm": 1.8122353553771973,
      "learning_rate": 3.597993123289594e-05,
      "loss": 0.8646,
      "step": 5580
    },
    {
      "epoch": 1.7622355541521944,
      "grad_norm": 1.7994346618652344,
      "learning_rate": 3.5976422707178446e-05,
      "loss": 0.8574,
      "step": 5581
    },
    {
      "epoch": 1.76255131038838,
      "grad_norm": 1.7144434452056885,
      "learning_rate": 3.597291418146095e-05,
      "loss": 0.8056,
      "step": 5582
    },
    {
      "epoch": 1.7628670666245658,
      "grad_norm": 1.5355205535888672,
      "learning_rate": 3.596940565574346e-05,
      "loss": 0.8327,
      "step": 5583
    },
    {
      "epoch": 1.7631828228607516,
      "grad_norm": 1.6545188426971436,
      "learning_rate": 3.596589713002597e-05,
      "loss": 0.8599,
      "step": 5584
    },
    {
      "epoch": 1.7634985790969373,
      "grad_norm": 1.6246957778930664,
      "learning_rate": 3.5962388604308474e-05,
      "loss": 0.8002,
      "step": 5585
    },
    {
      "epoch": 1.7638143353331228,
      "grad_norm": 1.5842392444610596,
      "learning_rate": 3.595888007859098e-05,
      "loss": 0.8191,
      "step": 5586
    },
    {
      "epoch": 1.7641300915693083,
      "grad_norm": 1.985959529876709,
      "learning_rate": 3.595537155287349e-05,
      "loss": 0.8434,
      "step": 5587
    },
    {
      "epoch": 1.764445847805494,
      "grad_norm": 1.5069947242736816,
      "learning_rate": 3.5951863027155994e-05,
      "loss": 0.8215,
      "step": 5588
    },
    {
      "epoch": 1.7647616040416798,
      "grad_norm": 1.530216932296753,
      "learning_rate": 3.5948354501438494e-05,
      "loss": 0.7721,
      "step": 5589
    },
    {
      "epoch": 1.7650773602778655,
      "grad_norm": 1.7700178623199463,
      "learning_rate": 3.5944845975721e-05,
      "loss": 0.8451,
      "step": 5590
    },
    {
      "epoch": 1.7653931165140513,
      "grad_norm": 1.5672129392623901,
      "learning_rate": 3.594133745000351e-05,
      "loss": 0.8005,
      "step": 5591
    },
    {
      "epoch": 1.7657088727502368,
      "grad_norm": 1.5044097900390625,
      "learning_rate": 3.5937828924286015e-05,
      "loss": 0.767,
      "step": 5592
    },
    {
      "epoch": 1.7660246289864223,
      "grad_norm": 1.688894510269165,
      "learning_rate": 3.593432039856852e-05,
      "loss": 0.8796,
      "step": 5593
    },
    {
      "epoch": 1.766340385222608,
      "grad_norm": 1.5235748291015625,
      "learning_rate": 3.593081187285103e-05,
      "loss": 0.7694,
      "step": 5594
    },
    {
      "epoch": 1.7666561414587938,
      "grad_norm": 1.6775933504104614,
      "learning_rate": 3.5927303347133535e-05,
      "loss": 0.8588,
      "step": 5595
    },
    {
      "epoch": 1.7669718976949795,
      "grad_norm": 1.5897183418273926,
      "learning_rate": 3.592379482141604e-05,
      "loss": 0.7929,
      "step": 5596
    },
    {
      "epoch": 1.7672876539311653,
      "grad_norm": 1.7714300155639648,
      "learning_rate": 3.592028629569855e-05,
      "loss": 0.8443,
      "step": 5597
    },
    {
      "epoch": 1.7676034101673508,
      "grad_norm": 1.6358953714370728,
      "learning_rate": 3.5916777769981055e-05,
      "loss": 0.8333,
      "step": 5598
    },
    {
      "epoch": 1.7679191664035365,
      "grad_norm": 2.3594632148742676,
      "learning_rate": 3.591326924426356e-05,
      "loss": 0.7555,
      "step": 5599
    },
    {
      "epoch": 1.768234922639722,
      "grad_norm": 1.860563039779663,
      "learning_rate": 3.590976071854607e-05,
      "loss": 0.7766,
      "step": 5600
    },
    {
      "epoch": 1.768234922639722,
      "eval_loss": 0.7997680902481079,
      "eval_runtime": 4.6427,
      "eval_samples_per_second": 536.325,
      "eval_steps_per_second": 4.523,
      "step": 5600
    },
    {
      "epoch": 1.7685506788759078,
      "grad_norm": 1.5731174945831299,
      "learning_rate": 3.5906252192828576e-05,
      "loss": 0.7222,
      "step": 5601
    },
    {
      "epoch": 1.7688664351120935,
      "grad_norm": 1.8291183710098267,
      "learning_rate": 3.590274366711108e-05,
      "loss": 0.8183,
      "step": 5602
    },
    {
      "epoch": 1.7691821913482793,
      "grad_norm": 1.6165671348571777,
      "learning_rate": 3.589923514139359e-05,
      "loss": 0.8274,
      "step": 5603
    },
    {
      "epoch": 1.7694979475844648,
      "grad_norm": 1.584455966949463,
      "learning_rate": 3.5895726615676096e-05,
      "loss": 0.8412,
      "step": 5604
    },
    {
      "epoch": 1.7698137038206505,
      "grad_norm": 1.7998765707015991,
      "learning_rate": 3.58922180899586e-05,
      "loss": 0.7953,
      "step": 5605
    },
    {
      "epoch": 1.770129460056836,
      "grad_norm": 1.415165901184082,
      "learning_rate": 3.588870956424111e-05,
      "loss": 0.705,
      "step": 5606
    },
    {
      "epoch": 1.7704452162930218,
      "grad_norm": 1.5970851182937622,
      "learning_rate": 3.588520103852362e-05,
      "loss": 0.8308,
      "step": 5607
    },
    {
      "epoch": 1.7707609725292075,
      "grad_norm": 1.6645166873931885,
      "learning_rate": 3.588169251280612e-05,
      "loss": 0.8993,
      "step": 5608
    },
    {
      "epoch": 1.7710767287653932,
      "grad_norm": 1.628039836883545,
      "learning_rate": 3.5878183987088624e-05,
      "loss": 0.7829,
      "step": 5609
    },
    {
      "epoch": 1.7713924850015788,
      "grad_norm": 1.7771400213241577,
      "learning_rate": 3.587467546137113e-05,
      "loss": 0.8222,
      "step": 5610
    },
    {
      "epoch": 1.7717082412377645,
      "grad_norm": 1.5714292526245117,
      "learning_rate": 3.587116693565364e-05,
      "loss": 0.8615,
      "step": 5611
    },
    {
      "epoch": 1.77202399747395,
      "grad_norm": 1.8475581407546997,
      "learning_rate": 3.5867658409936144e-05,
      "loss": 0.8647,
      "step": 5612
    },
    {
      "epoch": 1.7723397537101357,
      "grad_norm": 1.8199416399002075,
      "learning_rate": 3.586414988421865e-05,
      "loss": 0.8708,
      "step": 5613
    },
    {
      "epoch": 1.7726555099463215,
      "grad_norm": 1.474994421005249,
      "learning_rate": 3.586064135850116e-05,
      "loss": 0.8663,
      "step": 5614
    },
    {
      "epoch": 1.7729712661825072,
      "grad_norm": 1.802014946937561,
      "learning_rate": 3.585713283278367e-05,
      "loss": 0.8878,
      "step": 5615
    },
    {
      "epoch": 1.7732870224186927,
      "grad_norm": 1.5896998643875122,
      "learning_rate": 3.585362430706618e-05,
      "loss": 0.8214,
      "step": 5616
    },
    {
      "epoch": 1.7736027786548785,
      "grad_norm": 1.5054792165756226,
      "learning_rate": 3.585011578134868e-05,
      "loss": 0.8535,
      "step": 5617
    },
    {
      "epoch": 1.773918534891064,
      "grad_norm": 1.8819308280944824,
      "learning_rate": 3.5846607255631185e-05,
      "loss": 0.8303,
      "step": 5618
    },
    {
      "epoch": 1.7742342911272497,
      "grad_norm": 1.5306286811828613,
      "learning_rate": 3.584309872991369e-05,
      "loss": 0.7539,
      "step": 5619
    },
    {
      "epoch": 1.7745500473634355,
      "grad_norm": 1.4349080324172974,
      "learning_rate": 3.58395902041962e-05,
      "loss": 0.8192,
      "step": 5620
    },
    {
      "epoch": 1.7748658035996212,
      "grad_norm": 1.7494382858276367,
      "learning_rate": 3.5836081678478705e-05,
      "loss": 0.8393,
      "step": 5621
    },
    {
      "epoch": 1.7751815598358067,
      "grad_norm": 1.8895246982574463,
      "learning_rate": 3.583257315276121e-05,
      "loss": 0.9106,
      "step": 5622
    },
    {
      "epoch": 1.7754973160719925,
      "grad_norm": 1.6167203187942505,
      "learning_rate": 3.582906462704372e-05,
      "loss": 0.8874,
      "step": 5623
    },
    {
      "epoch": 1.775813072308178,
      "grad_norm": 1.8301289081573486,
      "learning_rate": 3.5825556101326226e-05,
      "loss": 0.8479,
      "step": 5624
    },
    {
      "epoch": 1.7761288285443637,
      "grad_norm": 1.4793058633804321,
      "learning_rate": 3.582204757560873e-05,
      "loss": 0.7956,
      "step": 5625
    },
    {
      "epoch": 1.7764445847805495,
      "grad_norm": 1.8276954889297485,
      "learning_rate": 3.581853904989124e-05,
      "loss": 0.8817,
      "step": 5626
    },
    {
      "epoch": 1.7767603410167352,
      "grad_norm": 1.437646746635437,
      "learning_rate": 3.581503052417374e-05,
      "loss": 0.8245,
      "step": 5627
    },
    {
      "epoch": 1.7770760972529207,
      "grad_norm": 1.904840350151062,
      "learning_rate": 3.5811521998456246e-05,
      "loss": 0.8819,
      "step": 5628
    },
    {
      "epoch": 1.7773918534891064,
      "grad_norm": 2.084947109222412,
      "learning_rate": 3.580801347273875e-05,
      "loss": 0.8376,
      "step": 5629
    },
    {
      "epoch": 1.777707609725292,
      "grad_norm": 1.6400953531265259,
      "learning_rate": 3.580450494702126e-05,
      "loss": 0.8681,
      "step": 5630
    },
    {
      "epoch": 1.7780233659614777,
      "grad_norm": 1.655770182609558,
      "learning_rate": 3.580099642130377e-05,
      "loss": 0.7861,
      "step": 5631
    },
    {
      "epoch": 1.7783391221976634,
      "grad_norm": 1.5104814767837524,
      "learning_rate": 3.579748789558628e-05,
      "loss": 0.8407,
      "step": 5632
    },
    {
      "epoch": 1.7786548784338492,
      "grad_norm": 1.7937029600143433,
      "learning_rate": 3.579397936986879e-05,
      "loss": 0.759,
      "step": 5633
    },
    {
      "epoch": 1.7789706346700347,
      "grad_norm": 1.6593916416168213,
      "learning_rate": 3.5790470844151294e-05,
      "loss": 0.9252,
      "step": 5634
    },
    {
      "epoch": 1.7792863909062204,
      "grad_norm": 1.613574743270874,
      "learning_rate": 3.57869623184338e-05,
      "loss": 0.8511,
      "step": 5635
    },
    {
      "epoch": 1.779602147142406,
      "grad_norm": 1.4485881328582764,
      "learning_rate": 3.57834537927163e-05,
      "loss": 0.7852,
      "step": 5636
    },
    {
      "epoch": 1.7799179033785917,
      "grad_norm": 1.4545005559921265,
      "learning_rate": 3.577994526699881e-05,
      "loss": 0.8507,
      "step": 5637
    },
    {
      "epoch": 1.7802336596147774,
      "grad_norm": 1.3921209573745728,
      "learning_rate": 3.5776436741281314e-05,
      "loss": 0.8248,
      "step": 5638
    },
    {
      "epoch": 1.7805494158509632,
      "grad_norm": 1.5735197067260742,
      "learning_rate": 3.577292821556382e-05,
      "loss": 0.8107,
      "step": 5639
    },
    {
      "epoch": 1.7808651720871487,
      "grad_norm": 1.3740960359573364,
      "learning_rate": 3.576941968984633e-05,
      "loss": 0.8237,
      "step": 5640
    },
    {
      "epoch": 1.7811809283233344,
      "grad_norm": 1.5857830047607422,
      "learning_rate": 3.5765911164128835e-05,
      "loss": 0.8503,
      "step": 5641
    },
    {
      "epoch": 1.78149668455952,
      "grad_norm": 1.269473671913147,
      "learning_rate": 3.576240263841134e-05,
      "loss": 0.8114,
      "step": 5642
    },
    {
      "epoch": 1.7818124407957057,
      "grad_norm": 1.63279128074646,
      "learning_rate": 3.575889411269385e-05,
      "loss": 0.8128,
      "step": 5643
    },
    {
      "epoch": 1.7821281970318914,
      "grad_norm": 1.7472383975982666,
      "learning_rate": 3.5755385586976355e-05,
      "loss": 0.8268,
      "step": 5644
    },
    {
      "epoch": 1.7824439532680771,
      "grad_norm": 1.5889062881469727,
      "learning_rate": 3.575187706125886e-05,
      "loss": 0.8397,
      "step": 5645
    },
    {
      "epoch": 1.7827597095042627,
      "grad_norm": 1.4051973819732666,
      "learning_rate": 3.574836853554137e-05,
      "loss": 0.8915,
      "step": 5646
    },
    {
      "epoch": 1.7830754657404484,
      "grad_norm": 1.468736171722412,
      "learning_rate": 3.574486000982387e-05,
      "loss": 0.8292,
      "step": 5647
    },
    {
      "epoch": 1.783391221976634,
      "grad_norm": 1.5180410146713257,
      "learning_rate": 3.5741351484106376e-05,
      "loss": 0.8154,
      "step": 5648
    },
    {
      "epoch": 1.7837069782128196,
      "grad_norm": 1.402213454246521,
      "learning_rate": 3.573784295838888e-05,
      "loss": 0.7678,
      "step": 5649
    },
    {
      "epoch": 1.7840227344490054,
      "grad_norm": 1.4519771337509155,
      "learning_rate": 3.5734334432671396e-05,
      "loss": 0.8198,
      "step": 5650
    },
    {
      "epoch": 1.7843384906851911,
      "grad_norm": 1.6240631341934204,
      "learning_rate": 3.57308259069539e-05,
      "loss": 0.8351,
      "step": 5651
    },
    {
      "epoch": 1.7846542469213769,
      "grad_norm": 1.5876895189285278,
      "learning_rate": 3.572731738123641e-05,
      "loss": 0.8474,
      "step": 5652
    },
    {
      "epoch": 1.7849700031575624,
      "grad_norm": 1.50093412399292,
      "learning_rate": 3.572380885551892e-05,
      "loss": 0.8591,
      "step": 5653
    },
    {
      "epoch": 1.785285759393748,
      "grad_norm": 1.4289923906326294,
      "learning_rate": 3.5720300329801423e-05,
      "loss": 0.7765,
      "step": 5654
    },
    {
      "epoch": 1.7856015156299336,
      "grad_norm": 1.5946141481399536,
      "learning_rate": 3.5716791804083923e-05,
      "loss": 0.8517,
      "step": 5655
    },
    {
      "epoch": 1.7859172718661194,
      "grad_norm": 1.530097484588623,
      "learning_rate": 3.571328327836643e-05,
      "loss": 0.862,
      "step": 5656
    },
    {
      "epoch": 1.786233028102305,
      "grad_norm": 1.7180123329162598,
      "learning_rate": 3.570977475264894e-05,
      "loss": 0.8888,
      "step": 5657
    },
    {
      "epoch": 1.7865487843384908,
      "grad_norm": 1.9452617168426514,
      "learning_rate": 3.5706266226931444e-05,
      "loss": 0.82,
      "step": 5658
    },
    {
      "epoch": 1.7868645405746764,
      "grad_norm": 1.5702576637268066,
      "learning_rate": 3.570275770121395e-05,
      "loss": 0.7713,
      "step": 5659
    },
    {
      "epoch": 1.7871802968108619,
      "grad_norm": 1.6225104331970215,
      "learning_rate": 3.569924917549646e-05,
      "loss": 0.8125,
      "step": 5660
    },
    {
      "epoch": 1.7874960530470476,
      "grad_norm": 1.3766810894012451,
      "learning_rate": 3.5695740649778964e-05,
      "loss": 0.836,
      "step": 5661
    },
    {
      "epoch": 1.7878118092832334,
      "grad_norm": 1.5156443119049072,
      "learning_rate": 3.569223212406147e-05,
      "loss": 0.8295,
      "step": 5662
    },
    {
      "epoch": 1.788127565519419,
      "grad_norm": 1.5481019020080566,
      "learning_rate": 3.568872359834398e-05,
      "loss": 0.8217,
      "step": 5663
    },
    {
      "epoch": 1.7884433217556048,
      "grad_norm": 1.8259557485580444,
      "learning_rate": 3.5685215072626485e-05,
      "loss": 0.7876,
      "step": 5664
    },
    {
      "epoch": 1.7887590779917903,
      "grad_norm": 2.083125591278076,
      "learning_rate": 3.568170654690899e-05,
      "loss": 0.9552,
      "step": 5665
    },
    {
      "epoch": 1.7890748342279759,
      "grad_norm": 1.697433352470398,
      "learning_rate": 3.567819802119149e-05,
      "loss": 0.8574,
      "step": 5666
    },
    {
      "epoch": 1.7893905904641616,
      "grad_norm": 1.6174657344818115,
      "learning_rate": 3.5674689495474005e-05,
      "loss": 0.8143,
      "step": 5667
    },
    {
      "epoch": 1.7897063467003473,
      "grad_norm": 1.479853868484497,
      "learning_rate": 3.567118096975651e-05,
      "loss": 0.7933,
      "step": 5668
    },
    {
      "epoch": 1.790022102936533,
      "grad_norm": 1.4573314189910889,
      "learning_rate": 3.566767244403902e-05,
      "loss": 0.8349,
      "step": 5669
    },
    {
      "epoch": 1.7903378591727188,
      "grad_norm": 1.545058250427246,
      "learning_rate": 3.5664163918321526e-05,
      "loss": 0.8186,
      "step": 5670
    },
    {
      "epoch": 1.7906536154089043,
      "grad_norm": 1.5222078561782837,
      "learning_rate": 3.566065539260403e-05,
      "loss": 0.8288,
      "step": 5671
    },
    {
      "epoch": 1.7909693716450898,
      "grad_norm": 1.8969649076461792,
      "learning_rate": 3.565714686688654e-05,
      "loss": 0.8037,
      "step": 5672
    },
    {
      "epoch": 1.7912851278812756,
      "grad_norm": 1.4693795442581177,
      "learning_rate": 3.5653638341169046e-05,
      "loss": 0.7495,
      "step": 5673
    },
    {
      "epoch": 1.7916008841174613,
      "grad_norm": 1.6741588115692139,
      "learning_rate": 3.5650129815451546e-05,
      "loss": 0.8864,
      "step": 5674
    },
    {
      "epoch": 1.791916640353647,
      "grad_norm": 1.6865875720977783,
      "learning_rate": 3.564662128973405e-05,
      "loss": 0.8746,
      "step": 5675
    },
    {
      "epoch": 1.7922323965898328,
      "grad_norm": 1.779848575592041,
      "learning_rate": 3.564311276401656e-05,
      "loss": 0.8443,
      "step": 5676
    },
    {
      "epoch": 1.7925481528260183,
      "grad_norm": 1.3887736797332764,
      "learning_rate": 3.5639604238299067e-05,
      "loss": 0.8195,
      "step": 5677
    },
    {
      "epoch": 1.7928639090622038,
      "grad_norm": 1.768330454826355,
      "learning_rate": 3.5636095712581573e-05,
      "loss": 0.7993,
      "step": 5678
    },
    {
      "epoch": 1.7931796652983896,
      "grad_norm": 1.4553890228271484,
      "learning_rate": 3.563258718686408e-05,
      "loss": 0.7795,
      "step": 5679
    },
    {
      "epoch": 1.7934954215345753,
      "grad_norm": 1.4726961851119995,
      "learning_rate": 3.562907866114659e-05,
      "loss": 0.7838,
      "step": 5680
    },
    {
      "epoch": 1.793811177770761,
      "grad_norm": 1.5642398595809937,
      "learning_rate": 3.5625570135429094e-05,
      "loss": 0.8574,
      "step": 5681
    },
    {
      "epoch": 1.7941269340069468,
      "grad_norm": 1.4276130199432373,
      "learning_rate": 3.56220616097116e-05,
      "loss": 0.7852,
      "step": 5682
    },
    {
      "epoch": 1.7944426902431323,
      "grad_norm": 1.6577917337417603,
      "learning_rate": 3.561855308399411e-05,
      "loss": 0.8089,
      "step": 5683
    },
    {
      "epoch": 1.7947584464793178,
      "grad_norm": 1.413488745689392,
      "learning_rate": 3.5615044558276614e-05,
      "loss": 0.8533,
      "step": 5684
    },
    {
      "epoch": 1.7950742027155036,
      "grad_norm": 1.5353004932403564,
      "learning_rate": 3.561153603255912e-05,
      "loss": 0.8937,
      "step": 5685
    },
    {
      "epoch": 1.7953899589516893,
      "grad_norm": 1.6917661428451538,
      "learning_rate": 3.560802750684163e-05,
      "loss": 0.7723,
      "step": 5686
    },
    {
      "epoch": 1.795705715187875,
      "grad_norm": 1.4890729188919067,
      "learning_rate": 3.5604518981124135e-05,
      "loss": 0.8107,
      "step": 5687
    },
    {
      "epoch": 1.7960214714240608,
      "grad_norm": 1.5204569101333618,
      "learning_rate": 3.560101045540664e-05,
      "loss": 0.7539,
      "step": 5688
    },
    {
      "epoch": 1.7963372276602463,
      "grad_norm": 1.7445851564407349,
      "learning_rate": 3.559750192968915e-05,
      "loss": 0.8488,
      "step": 5689
    },
    {
      "epoch": 1.7966529838964318,
      "grad_norm": 1.6797038316726685,
      "learning_rate": 3.5593993403971655e-05,
      "loss": 0.7115,
      "step": 5690
    },
    {
      "epoch": 1.7969687401326175,
      "grad_norm": 1.850656270980835,
      "learning_rate": 3.559048487825416e-05,
      "loss": 0.7931,
      "step": 5691
    },
    {
      "epoch": 1.7972844963688033,
      "grad_norm": 1.6683827638626099,
      "learning_rate": 3.558697635253667e-05,
      "loss": 0.8421,
      "step": 5692
    },
    {
      "epoch": 1.797600252604989,
      "grad_norm": 1.4796149730682373,
      "learning_rate": 3.5583467826819176e-05,
      "loss": 0.8303,
      "step": 5693
    },
    {
      "epoch": 1.7979160088411748,
      "grad_norm": 1.569527506828308,
      "learning_rate": 3.5579959301101676e-05,
      "loss": 0.8173,
      "step": 5694
    },
    {
      "epoch": 1.7982317650773603,
      "grad_norm": 1.3918882608413696,
      "learning_rate": 3.557645077538418e-05,
      "loss": 0.7398,
      "step": 5695
    },
    {
      "epoch": 1.7985475213135458,
      "grad_norm": 1.7282501459121704,
      "learning_rate": 3.557294224966669e-05,
      "loss": 0.8687,
      "step": 5696
    },
    {
      "epoch": 1.7988632775497315,
      "grad_norm": 1.7265143394470215,
      "learning_rate": 3.5569433723949196e-05,
      "loss": 0.8278,
      "step": 5697
    },
    {
      "epoch": 1.7991790337859173,
      "grad_norm": 1.3740073442459106,
      "learning_rate": 3.55659251982317e-05,
      "loss": 0.7864,
      "step": 5698
    },
    {
      "epoch": 1.799494790022103,
      "grad_norm": 1.4146409034729004,
      "learning_rate": 3.556241667251421e-05,
      "loss": 0.7898,
      "step": 5699
    },
    {
      "epoch": 1.7998105462582887,
      "grad_norm": 1.5840274095535278,
      "learning_rate": 3.5558908146796717e-05,
      "loss": 0.8388,
      "step": 5700
    },
    {
      "epoch": 1.8001263024944743,
      "grad_norm": 1.7844014167785645,
      "learning_rate": 3.555539962107922e-05,
      "loss": 0.885,
      "step": 5701
    },
    {
      "epoch": 1.8004420587306598,
      "grad_norm": 1.7451914548873901,
      "learning_rate": 3.555189109536173e-05,
      "loss": 0.7785,
      "step": 5702
    },
    {
      "epoch": 1.8007578149668455,
      "grad_norm": 1.4865918159484863,
      "learning_rate": 3.554838256964424e-05,
      "loss": 0.8013,
      "step": 5703
    },
    {
      "epoch": 1.8010735712030312,
      "grad_norm": 1.5442225933074951,
      "learning_rate": 3.5544874043926744e-05,
      "loss": 0.7701,
      "step": 5704
    },
    {
      "epoch": 1.801389327439217,
      "grad_norm": 1.672167420387268,
      "learning_rate": 3.554136551820925e-05,
      "loss": 0.8168,
      "step": 5705
    },
    {
      "epoch": 1.8017050836754027,
      "grad_norm": 1.6197134256362915,
      "learning_rate": 3.553785699249176e-05,
      "loss": 0.8154,
      "step": 5706
    },
    {
      "epoch": 1.8020208399115882,
      "grad_norm": 1.5226248502731323,
      "learning_rate": 3.5534348466774264e-05,
      "loss": 0.7189,
      "step": 5707
    },
    {
      "epoch": 1.802336596147774,
      "grad_norm": 1.7332113981246948,
      "learning_rate": 3.553083994105677e-05,
      "loss": 0.8019,
      "step": 5708
    },
    {
      "epoch": 1.8026523523839595,
      "grad_norm": 1.799607515335083,
      "learning_rate": 3.552733141533928e-05,
      "loss": 0.8522,
      "step": 5709
    },
    {
      "epoch": 1.8029681086201452,
      "grad_norm": 1.3847594261169434,
      "learning_rate": 3.5523822889621785e-05,
      "loss": 0.7665,
      "step": 5710
    },
    {
      "epoch": 1.803283864856331,
      "grad_norm": 1.8985986709594727,
      "learning_rate": 3.552031436390429e-05,
      "loss": 0.7935,
      "step": 5711
    },
    {
      "epoch": 1.8035996210925167,
      "grad_norm": 1.698777437210083,
      "learning_rate": 3.55168058381868e-05,
      "loss": 0.8713,
      "step": 5712
    },
    {
      "epoch": 1.8039153773287022,
      "grad_norm": 1.5903116464614868,
      "learning_rate": 3.55132973124693e-05,
      "loss": 0.8419,
      "step": 5713
    },
    {
      "epoch": 1.804231133564888,
      "grad_norm": 1.7107106447219849,
      "learning_rate": 3.5509788786751805e-05,
      "loss": 0.8688,
      "step": 5714
    },
    {
      "epoch": 1.8045468898010735,
      "grad_norm": 1.5210397243499756,
      "learning_rate": 3.550628026103431e-05,
      "loss": 0.7799,
      "step": 5715
    },
    {
      "epoch": 1.8048626460372592,
      "grad_norm": 1.5112204551696777,
      "learning_rate": 3.550277173531682e-05,
      "loss": 0.8404,
      "step": 5716
    },
    {
      "epoch": 1.805178402273445,
      "grad_norm": 1.6892962455749512,
      "learning_rate": 3.5499263209599326e-05,
      "loss": 0.7883,
      "step": 5717
    },
    {
      "epoch": 1.8054941585096307,
      "grad_norm": 1.4952746629714966,
      "learning_rate": 3.549575468388183e-05,
      "loss": 0.8345,
      "step": 5718
    },
    {
      "epoch": 1.8058099147458162,
      "grad_norm": 1.4647941589355469,
      "learning_rate": 3.549224615816434e-05,
      "loss": 0.8048,
      "step": 5719
    },
    {
      "epoch": 1.806125670982002,
      "grad_norm": 1.6424485445022583,
      "learning_rate": 3.548873763244685e-05,
      "loss": 0.8002,
      "step": 5720
    },
    {
      "epoch": 1.8064414272181875,
      "grad_norm": 1.404929757118225,
      "learning_rate": 3.548522910672936e-05,
      "loss": 0.7943,
      "step": 5721
    },
    {
      "epoch": 1.8067571834543732,
      "grad_norm": 1.5542352199554443,
      "learning_rate": 3.548172058101186e-05,
      "loss": 0.8957,
      "step": 5722
    },
    {
      "epoch": 1.807072939690559,
      "grad_norm": 1.8336302042007446,
      "learning_rate": 3.5478212055294366e-05,
      "loss": 0.8447,
      "step": 5723
    },
    {
      "epoch": 1.8073886959267447,
      "grad_norm": 1.6597037315368652,
      "learning_rate": 3.547470352957687e-05,
      "loss": 0.8385,
      "step": 5724
    },
    {
      "epoch": 1.8077044521629302,
      "grad_norm": 1.6612048149108887,
      "learning_rate": 3.547119500385938e-05,
      "loss": 0.7487,
      "step": 5725
    },
    {
      "epoch": 1.808020208399116,
      "grad_norm": 1.622497797012329,
      "learning_rate": 3.546768647814189e-05,
      "loss": 0.8114,
      "step": 5726
    },
    {
      "epoch": 1.8083359646353014,
      "grad_norm": 1.7218443155288696,
      "learning_rate": 3.5464177952424394e-05,
      "loss": 0.8024,
      "step": 5727
    },
    {
      "epoch": 1.8086517208714872,
      "grad_norm": 1.4281755685806274,
      "learning_rate": 3.54606694267069e-05,
      "loss": 0.7804,
      "step": 5728
    },
    {
      "epoch": 1.808967477107673,
      "grad_norm": 1.4043952226638794,
      "learning_rate": 3.545716090098941e-05,
      "loss": 0.8239,
      "step": 5729
    },
    {
      "epoch": 1.8092832333438587,
      "grad_norm": 1.5362807512283325,
      "learning_rate": 3.5453652375271914e-05,
      "loss": 0.8183,
      "step": 5730
    },
    {
      "epoch": 1.8095989895800442,
      "grad_norm": 1.494787335395813,
      "learning_rate": 3.545014384955442e-05,
      "loss": 0.7428,
      "step": 5731
    },
    {
      "epoch": 1.80991474581623,
      "grad_norm": 1.2170345783233643,
      "learning_rate": 3.544663532383692e-05,
      "loss": 0.7937,
      "step": 5732
    },
    {
      "epoch": 1.8102305020524154,
      "grad_norm": 1.7301284074783325,
      "learning_rate": 3.544312679811943e-05,
      "loss": 0.8024,
      "step": 5733
    },
    {
      "epoch": 1.8105462582886012,
      "grad_norm": 1.5486667156219482,
      "learning_rate": 3.5439618272401935e-05,
      "loss": 0.7648,
      "step": 5734
    },
    {
      "epoch": 1.810862014524787,
      "grad_norm": 1.5176335573196411,
      "learning_rate": 3.543610974668444e-05,
      "loss": 0.8001,
      "step": 5735
    },
    {
      "epoch": 1.8111777707609726,
      "grad_norm": 1.627005934715271,
      "learning_rate": 3.543260122096695e-05,
      "loss": 0.8347,
      "step": 5736
    },
    {
      "epoch": 1.8114935269971582,
      "grad_norm": 1.8342612981796265,
      "learning_rate": 3.542909269524946e-05,
      "loss": 0.7714,
      "step": 5737
    },
    {
      "epoch": 1.811809283233344,
      "grad_norm": 1.6290276050567627,
      "learning_rate": 3.542558416953197e-05,
      "loss": 0.8916,
      "step": 5738
    },
    {
      "epoch": 1.8121250394695294,
      "grad_norm": 2.0638303756713867,
      "learning_rate": 3.5422075643814475e-05,
      "loss": 0.8744,
      "step": 5739
    },
    {
      "epoch": 1.8124407957057151,
      "grad_norm": 1.690139651298523,
      "learning_rate": 3.541856711809698e-05,
      "loss": 0.7588,
      "step": 5740
    },
    {
      "epoch": 1.8127565519419009,
      "grad_norm": 1.738269567489624,
      "learning_rate": 3.541505859237948e-05,
      "loss": 0.8079,
      "step": 5741
    },
    {
      "epoch": 1.8130723081780866,
      "grad_norm": 1.8335567712783813,
      "learning_rate": 3.541155006666199e-05,
      "loss": 0.8527,
      "step": 5742
    },
    {
      "epoch": 1.8133880644142721,
      "grad_norm": 1.7160004377365112,
      "learning_rate": 3.5408041540944496e-05,
      "loss": 0.839,
      "step": 5743
    },
    {
      "epoch": 1.8137038206504579,
      "grad_norm": 1.7700188159942627,
      "learning_rate": 3.5404533015227e-05,
      "loss": 0.9251,
      "step": 5744
    },
    {
      "epoch": 1.8140195768866434,
      "grad_norm": 1.7094638347625732,
      "learning_rate": 3.540102448950951e-05,
      "loss": 0.9065,
      "step": 5745
    },
    {
      "epoch": 1.8143353331228291,
      "grad_norm": 1.538648009300232,
      "learning_rate": 3.5397515963792016e-05,
      "loss": 0.8449,
      "step": 5746
    },
    {
      "epoch": 1.8146510893590149,
      "grad_norm": 1.4230469465255737,
      "learning_rate": 3.539400743807452e-05,
      "loss": 0.8006,
      "step": 5747
    },
    {
      "epoch": 1.8149668455952006,
      "grad_norm": 1.3585429191589355,
      "learning_rate": 3.539049891235703e-05,
      "loss": 0.7891,
      "step": 5748
    },
    {
      "epoch": 1.8152826018313861,
      "grad_norm": 1.5526738166809082,
      "learning_rate": 3.538699038663954e-05,
      "loss": 0.8572,
      "step": 5749
    },
    {
      "epoch": 1.8155983580675719,
      "grad_norm": 1.5878121852874756,
      "learning_rate": 3.5383481860922044e-05,
      "loss": 0.7893,
      "step": 5750
    },
    {
      "epoch": 1.8159141143037574,
      "grad_norm": 1.7220971584320068,
      "learning_rate": 3.537997333520455e-05,
      "loss": 0.8577,
      "step": 5751
    },
    {
      "epoch": 1.8162298705399431,
      "grad_norm": 1.4755959510803223,
      "learning_rate": 3.537646480948705e-05,
      "loss": 0.8305,
      "step": 5752
    },
    {
      "epoch": 1.8165456267761289,
      "grad_norm": 1.5470174551010132,
      "learning_rate": 3.537295628376956e-05,
      "loss": 0.7809,
      "step": 5753
    },
    {
      "epoch": 1.8168613830123146,
      "grad_norm": 2.277376174926758,
      "learning_rate": 3.5369447758052064e-05,
      "loss": 0.8502,
      "step": 5754
    },
    {
      "epoch": 1.8171771392485,
      "grad_norm": 1.4327343702316284,
      "learning_rate": 3.536593923233458e-05,
      "loss": 0.7765,
      "step": 5755
    },
    {
      "epoch": 1.8174928954846858,
      "grad_norm": 1.77643883228302,
      "learning_rate": 3.5362430706617084e-05,
      "loss": 0.8028,
      "step": 5756
    },
    {
      "epoch": 1.8178086517208714,
      "grad_norm": 1.5304789543151855,
      "learning_rate": 3.535892218089959e-05,
      "loss": 0.8225,
      "step": 5757
    },
    {
      "epoch": 1.818124407957057,
      "grad_norm": 1.618960976600647,
      "learning_rate": 3.53554136551821e-05,
      "loss": 0.8752,
      "step": 5758
    },
    {
      "epoch": 1.8184401641932428,
      "grad_norm": 1.8889859914779663,
      "learning_rate": 3.5351905129464605e-05,
      "loss": 0.6812,
      "step": 5759
    },
    {
      "epoch": 1.8187559204294286,
      "grad_norm": 1.5909082889556885,
      "learning_rate": 3.5348396603747105e-05,
      "loss": 0.7612,
      "step": 5760
    },
    {
      "epoch": 1.8190716766656143,
      "grad_norm": 1.5842199325561523,
      "learning_rate": 3.534488807802961e-05,
      "loss": 0.8592,
      "step": 5761
    },
    {
      "epoch": 1.8193874329017998,
      "grad_norm": 1.7698986530303955,
      "learning_rate": 3.534137955231212e-05,
      "loss": 0.8144,
      "step": 5762
    },
    {
      "epoch": 1.8197031891379853,
      "grad_norm": 1.5370988845825195,
      "learning_rate": 3.5337871026594625e-05,
      "loss": 0.7583,
      "step": 5763
    },
    {
      "epoch": 1.820018945374171,
      "grad_norm": 1.5239393711090088,
      "learning_rate": 3.533436250087713e-05,
      "loss": 0.8363,
      "step": 5764
    },
    {
      "epoch": 1.8203347016103568,
      "grad_norm": 1.8667972087860107,
      "learning_rate": 3.533085397515964e-05,
      "loss": 0.7662,
      "step": 5765
    },
    {
      "epoch": 1.8206504578465426,
      "grad_norm": 2.0647966861724854,
      "learning_rate": 3.5327345449442146e-05,
      "loss": 0.8568,
      "step": 5766
    },
    {
      "epoch": 1.8209662140827283,
      "grad_norm": 1.5912829637527466,
      "learning_rate": 3.532383692372465e-05,
      "loss": 0.8275,
      "step": 5767
    },
    {
      "epoch": 1.8212819703189138,
      "grad_norm": 1.9510905742645264,
      "learning_rate": 3.532032839800716e-05,
      "loss": 0.841,
      "step": 5768
    },
    {
      "epoch": 1.8215977265550993,
      "grad_norm": 1.74018132686615,
      "learning_rate": 3.5316819872289666e-05,
      "loss": 0.8258,
      "step": 5769
    },
    {
      "epoch": 1.821913482791285,
      "grad_norm": 1.7118306159973145,
      "learning_rate": 3.531331134657217e-05,
      "loss": 0.8093,
      "step": 5770
    },
    {
      "epoch": 1.8222292390274708,
      "grad_norm": 1.6666886806488037,
      "learning_rate": 3.530980282085467e-05,
      "loss": 0.8404,
      "step": 5771
    },
    {
      "epoch": 1.8225449952636565,
      "grad_norm": 1.816405177116394,
      "learning_rate": 3.530629429513719e-05,
      "loss": 0.7583,
      "step": 5772
    },
    {
      "epoch": 1.8228607514998423,
      "grad_norm": 1.9241434335708618,
      "learning_rate": 3.5302785769419694e-05,
      "loss": 0.8267,
      "step": 5773
    },
    {
      "epoch": 1.8231765077360278,
      "grad_norm": 1.7839720249176025,
      "learning_rate": 3.52992772437022e-05,
      "loss": 0.8206,
      "step": 5774
    },
    {
      "epoch": 1.8234922639722133,
      "grad_norm": 1.6137515306472778,
      "learning_rate": 3.529576871798471e-05,
      "loss": 0.7744,
      "step": 5775
    },
    {
      "epoch": 1.823808020208399,
      "grad_norm": 1.7122888565063477,
      "learning_rate": 3.5292260192267214e-05,
      "loss": 0.8147,
      "step": 5776
    },
    {
      "epoch": 1.8241237764445848,
      "grad_norm": 1.6840680837631226,
      "learning_rate": 3.528875166654972e-05,
      "loss": 0.8644,
      "step": 5777
    },
    {
      "epoch": 1.8244395326807705,
      "grad_norm": 1.8183412551879883,
      "learning_rate": 3.528524314083223e-05,
      "loss": 0.8912,
      "step": 5778
    },
    {
      "epoch": 1.8247552889169563,
      "grad_norm": 1.6460973024368286,
      "learning_rate": 3.528173461511473e-05,
      "loss": 0.7904,
      "step": 5779
    },
    {
      "epoch": 1.8250710451531418,
      "grad_norm": 1.8295189142227173,
      "learning_rate": 3.5278226089397234e-05,
      "loss": 0.8128,
      "step": 5780
    },
    {
      "epoch": 1.8253868013893273,
      "grad_norm": 2.1648080348968506,
      "learning_rate": 3.527471756367974e-05,
      "loss": 0.8223,
      "step": 5781
    },
    {
      "epoch": 1.825702557625513,
      "grad_norm": 1.3717033863067627,
      "learning_rate": 3.527120903796225e-05,
      "loss": 0.7793,
      "step": 5782
    },
    {
      "epoch": 1.8260183138616988,
      "grad_norm": 1.6923046112060547,
      "learning_rate": 3.5267700512244755e-05,
      "loss": 0.814,
      "step": 5783
    },
    {
      "epoch": 1.8263340700978845,
      "grad_norm": 1.6101430654525757,
      "learning_rate": 3.526419198652726e-05,
      "loss": 0.8136,
      "step": 5784
    },
    {
      "epoch": 1.8266498263340702,
      "grad_norm": 1.528930902481079,
      "learning_rate": 3.526068346080977e-05,
      "loss": 0.832,
      "step": 5785
    },
    {
      "epoch": 1.8269655825702558,
      "grad_norm": 1.669541597366333,
      "learning_rate": 3.5257174935092275e-05,
      "loss": 0.873,
      "step": 5786
    },
    {
      "epoch": 1.8272813388064413,
      "grad_norm": 1.721664547920227,
      "learning_rate": 3.525366640937478e-05,
      "loss": 0.8051,
      "step": 5787
    },
    {
      "epoch": 1.827597095042627,
      "grad_norm": 1.6607335805892944,
      "learning_rate": 3.525015788365729e-05,
      "loss": 0.8234,
      "step": 5788
    },
    {
      "epoch": 1.8279128512788128,
      "grad_norm": 1.5749949216842651,
      "learning_rate": 3.5246649357939796e-05,
      "loss": 0.8521,
      "step": 5789
    },
    {
      "epoch": 1.8282286075149985,
      "grad_norm": 1.6274627447128296,
      "learning_rate": 3.52431408322223e-05,
      "loss": 0.7085,
      "step": 5790
    },
    {
      "epoch": 1.8285443637511842,
      "grad_norm": 1.5711743831634521,
      "learning_rate": 3.523963230650481e-05,
      "loss": 0.8287,
      "step": 5791
    },
    {
      "epoch": 1.8288601199873697,
      "grad_norm": 1.8797237873077393,
      "learning_rate": 3.5236123780787316e-05,
      "loss": 0.8043,
      "step": 5792
    },
    {
      "epoch": 1.8291758762235553,
      "grad_norm": 1.55975341796875,
      "learning_rate": 3.523261525506982e-05,
      "loss": 0.8539,
      "step": 5793
    },
    {
      "epoch": 1.829491632459741,
      "grad_norm": 1.408881425857544,
      "learning_rate": 3.522910672935233e-05,
      "loss": 0.7465,
      "step": 5794
    },
    {
      "epoch": 1.8298073886959267,
      "grad_norm": 1.77203369140625,
      "learning_rate": 3.522559820363484e-05,
      "loss": 0.8778,
      "step": 5795
    },
    {
      "epoch": 1.8301231449321125,
      "grad_norm": 1.741595983505249,
      "learning_rate": 3.5222089677917343e-05,
      "loss": 0.7018,
      "step": 5796
    },
    {
      "epoch": 1.8304389011682982,
      "grad_norm": 1.7076911926269531,
      "learning_rate": 3.521858115219985e-05,
      "loss": 0.8328,
      "step": 5797
    },
    {
      "epoch": 1.8307546574044837,
      "grad_norm": 1.5112513303756714,
      "learning_rate": 3.521507262648236e-05,
      "loss": 0.8394,
      "step": 5798
    },
    {
      "epoch": 1.8310704136406692,
      "grad_norm": 1.8096253871917725,
      "learning_rate": 3.521156410076486e-05,
      "loss": 0.8501,
      "step": 5799
    },
    {
      "epoch": 1.831386169876855,
      "grad_norm": 1.556073546409607,
      "learning_rate": 3.5208055575047364e-05,
      "loss": 0.7603,
      "step": 5800
    },
    {
      "epoch": 1.831386169876855,
      "eval_loss": 0.8020087480545044,
      "eval_runtime": 4.6637,
      "eval_samples_per_second": 533.914,
      "eval_steps_per_second": 4.503,
      "step": 5800
    },
    {
      "epoch": 1.8317019261130407,
      "grad_norm": 1.6009166240692139,
      "learning_rate": 3.520454704932987e-05,
      "loss": 0.8295,
      "step": 5801
    },
    {
      "epoch": 1.8320176823492265,
      "grad_norm": 1.7514853477478027,
      "learning_rate": 3.520103852361238e-05,
      "loss": 0.7682,
      "step": 5802
    },
    {
      "epoch": 1.8323334385854122,
      "grad_norm": 1.6241828203201294,
      "learning_rate": 3.5197529997894884e-05,
      "loss": 0.9013,
      "step": 5803
    },
    {
      "epoch": 1.8326491948215977,
      "grad_norm": 1.421126127243042,
      "learning_rate": 3.519402147217739e-05,
      "loss": 0.8023,
      "step": 5804
    },
    {
      "epoch": 1.8329649510577832,
      "grad_norm": 1.56403386592865,
      "learning_rate": 3.51905129464599e-05,
      "loss": 0.7696,
      "step": 5805
    },
    {
      "epoch": 1.833280707293969,
      "grad_norm": 1.7297053337097168,
      "learning_rate": 3.5187004420742405e-05,
      "loss": 0.8652,
      "step": 5806
    },
    {
      "epoch": 1.8335964635301547,
      "grad_norm": 1.7932279109954834,
      "learning_rate": 3.518349589502491e-05,
      "loss": 0.9303,
      "step": 5807
    },
    {
      "epoch": 1.8339122197663404,
      "grad_norm": 1.542431354522705,
      "learning_rate": 3.517998736930742e-05,
      "loss": 0.8159,
      "step": 5808
    },
    {
      "epoch": 1.8342279760025262,
      "grad_norm": 1.6829142570495605,
      "learning_rate": 3.5176478843589925e-05,
      "loss": 0.8081,
      "step": 5809
    },
    {
      "epoch": 1.8345437322387117,
      "grad_norm": 1.5777701139450073,
      "learning_rate": 3.517297031787243e-05,
      "loss": 0.8266,
      "step": 5810
    },
    {
      "epoch": 1.8348594884748972,
      "grad_norm": 1.5216739177703857,
      "learning_rate": 3.516946179215494e-05,
      "loss": 0.8339,
      "step": 5811
    },
    {
      "epoch": 1.835175244711083,
      "grad_norm": 1.6322482824325562,
      "learning_rate": 3.5165953266437446e-05,
      "loss": 0.833,
      "step": 5812
    },
    {
      "epoch": 1.8354910009472687,
      "grad_norm": 1.7127716541290283,
      "learning_rate": 3.516244474071995e-05,
      "loss": 0.8399,
      "step": 5813
    },
    {
      "epoch": 1.8358067571834544,
      "grad_norm": 1.6222598552703857,
      "learning_rate": 3.515893621500246e-05,
      "loss": 0.8215,
      "step": 5814
    },
    {
      "epoch": 1.8361225134196402,
      "grad_norm": 1.791230320930481,
      "learning_rate": 3.5155427689284966e-05,
      "loss": 0.8126,
      "step": 5815
    },
    {
      "epoch": 1.8364382696558257,
      "grad_norm": 1.5561192035675049,
      "learning_rate": 3.515191916356747e-05,
      "loss": 0.7585,
      "step": 5816
    },
    {
      "epoch": 1.8367540258920114,
      "grad_norm": 1.5561288595199585,
      "learning_rate": 3.514841063784998e-05,
      "loss": 0.7882,
      "step": 5817
    },
    {
      "epoch": 1.837069782128197,
      "grad_norm": 1.499991536140442,
      "learning_rate": 3.514490211213248e-05,
      "loss": 0.8146,
      "step": 5818
    },
    {
      "epoch": 1.8373855383643827,
      "grad_norm": 1.54446280002594,
      "learning_rate": 3.5141393586414987e-05,
      "loss": 0.8268,
      "step": 5819
    },
    {
      "epoch": 1.8377012946005684,
      "grad_norm": 1.60240638256073,
      "learning_rate": 3.5137885060697493e-05,
      "loss": 0.6937,
      "step": 5820
    },
    {
      "epoch": 1.8380170508367542,
      "grad_norm": 1.474940299987793,
      "learning_rate": 3.513437653498e-05,
      "loss": 0.8134,
      "step": 5821
    },
    {
      "epoch": 1.8383328070729397,
      "grad_norm": 1.9370241165161133,
      "learning_rate": 3.513086800926251e-05,
      "loss": 0.8478,
      "step": 5822
    },
    {
      "epoch": 1.8386485633091254,
      "grad_norm": 1.7302250862121582,
      "learning_rate": 3.5127359483545014e-05,
      "loss": 0.869,
      "step": 5823
    },
    {
      "epoch": 1.838964319545311,
      "grad_norm": 1.8004552125930786,
      "learning_rate": 3.512385095782752e-05,
      "loss": 0.8583,
      "step": 5824
    },
    {
      "epoch": 1.8392800757814967,
      "grad_norm": 1.4889637231826782,
      "learning_rate": 3.5120342432110034e-05,
      "loss": 0.8316,
      "step": 5825
    },
    {
      "epoch": 1.8395958320176824,
      "grad_norm": 1.759369969367981,
      "learning_rate": 3.511683390639254e-05,
      "loss": 0.8301,
      "step": 5826
    },
    {
      "epoch": 1.8399115882538681,
      "grad_norm": 1.6872749328613281,
      "learning_rate": 3.511332538067504e-05,
      "loss": 0.8927,
      "step": 5827
    },
    {
      "epoch": 1.8402273444900537,
      "grad_norm": 1.9752485752105713,
      "learning_rate": 3.510981685495755e-05,
      "loss": 0.7611,
      "step": 5828
    },
    {
      "epoch": 1.8405431007262394,
      "grad_norm": 1.7503858804702759,
      "learning_rate": 3.5106308329240055e-05,
      "loss": 0.8079,
      "step": 5829
    },
    {
      "epoch": 1.840858856962425,
      "grad_norm": 1.7351558208465576,
      "learning_rate": 3.510279980352256e-05,
      "loss": 0.8484,
      "step": 5830
    },
    {
      "epoch": 1.8411746131986106,
      "grad_norm": 1.6016924381256104,
      "learning_rate": 3.509929127780507e-05,
      "loss": 0.817,
      "step": 5831
    },
    {
      "epoch": 1.8414903694347964,
      "grad_norm": 1.9724829196929932,
      "learning_rate": 3.5095782752087575e-05,
      "loss": 0.8272,
      "step": 5832
    },
    {
      "epoch": 1.8418061256709821,
      "grad_norm": 1.4887669086456299,
      "learning_rate": 3.509227422637008e-05,
      "loss": 0.821,
      "step": 5833
    },
    {
      "epoch": 1.8421218819071676,
      "grad_norm": 1.7182756662368774,
      "learning_rate": 3.508876570065259e-05,
      "loss": 0.8419,
      "step": 5834
    },
    {
      "epoch": 1.8424376381433534,
      "grad_norm": 1.6058084964752197,
      "learning_rate": 3.5085257174935096e-05,
      "loss": 0.7939,
      "step": 5835
    },
    {
      "epoch": 1.842753394379539,
      "grad_norm": 1.722802758216858,
      "learning_rate": 3.50817486492176e-05,
      "loss": 0.8965,
      "step": 5836
    },
    {
      "epoch": 1.8430691506157246,
      "grad_norm": 2.3667733669281006,
      "learning_rate": 3.50782401235001e-05,
      "loss": 0.8794,
      "step": 5837
    },
    {
      "epoch": 1.8433849068519104,
      "grad_norm": 1.7319821119308472,
      "learning_rate": 3.507473159778261e-05,
      "loss": 0.7401,
      "step": 5838
    },
    {
      "epoch": 1.843700663088096,
      "grad_norm": 1.4555222988128662,
      "learning_rate": 3.5071223072065116e-05,
      "loss": 0.7807,
      "step": 5839
    },
    {
      "epoch": 1.8440164193242816,
      "grad_norm": 1.6396267414093018,
      "learning_rate": 3.506771454634762e-05,
      "loss": 0.8705,
      "step": 5840
    },
    {
      "epoch": 1.8443321755604674,
      "grad_norm": 1.5137939453125,
      "learning_rate": 3.506420602063013e-05,
      "loss": 0.8339,
      "step": 5841
    },
    {
      "epoch": 1.8446479317966529,
      "grad_norm": 1.4370001554489136,
      "learning_rate": 3.506069749491264e-05,
      "loss": 0.7478,
      "step": 5842
    },
    {
      "epoch": 1.8449636880328386,
      "grad_norm": 1.6320725679397583,
      "learning_rate": 3.505718896919515e-05,
      "loss": 0.8173,
      "step": 5843
    },
    {
      "epoch": 1.8452794442690243,
      "grad_norm": 1.7061631679534912,
      "learning_rate": 3.505368044347766e-05,
      "loss": 0.8001,
      "step": 5844
    },
    {
      "epoch": 1.84559520050521,
      "grad_norm": 1.495880365371704,
      "learning_rate": 3.5050171917760164e-05,
      "loss": 0.8166,
      "step": 5845
    },
    {
      "epoch": 1.8459109567413956,
      "grad_norm": 1.7644630670547485,
      "learning_rate": 3.5046663392042664e-05,
      "loss": 0.8322,
      "step": 5846
    },
    {
      "epoch": 1.8462267129775813,
      "grad_norm": 1.3992202281951904,
      "learning_rate": 3.504315486632517e-05,
      "loss": 0.7886,
      "step": 5847
    },
    {
      "epoch": 1.8465424692137669,
      "grad_norm": 1.4376153945922852,
      "learning_rate": 3.503964634060768e-05,
      "loss": 0.7736,
      "step": 5848
    },
    {
      "epoch": 1.8468582254499526,
      "grad_norm": 1.7922730445861816,
      "learning_rate": 3.5036137814890184e-05,
      "loss": 0.7931,
      "step": 5849
    },
    {
      "epoch": 1.8471739816861383,
      "grad_norm": 1.4094620943069458,
      "learning_rate": 3.503262928917269e-05,
      "loss": 0.7626,
      "step": 5850
    },
    {
      "epoch": 1.847489737922324,
      "grad_norm": 1.4979889392852783,
      "learning_rate": 3.50291207634552e-05,
      "loss": 0.8565,
      "step": 5851
    },
    {
      "epoch": 1.8478054941585096,
      "grad_norm": 1.5990864038467407,
      "learning_rate": 3.5025612237737705e-05,
      "loss": 0.8482,
      "step": 5852
    },
    {
      "epoch": 1.8481212503946953,
      "grad_norm": 1.6632698774337769,
      "learning_rate": 3.502210371202021e-05,
      "loss": 0.8328,
      "step": 5853
    },
    {
      "epoch": 1.8484370066308808,
      "grad_norm": 1.3993018865585327,
      "learning_rate": 3.501859518630272e-05,
      "loss": 0.7741,
      "step": 5854
    },
    {
      "epoch": 1.8487527628670666,
      "grad_norm": 1.51312255859375,
      "learning_rate": 3.5015086660585225e-05,
      "loss": 0.8195,
      "step": 5855
    },
    {
      "epoch": 1.8490685191032523,
      "grad_norm": 1.5059975385665894,
      "learning_rate": 3.501157813486773e-05,
      "loss": 0.8515,
      "step": 5856
    },
    {
      "epoch": 1.849384275339438,
      "grad_norm": 1.3958325386047363,
      "learning_rate": 3.500806960915023e-05,
      "loss": 0.7797,
      "step": 5857
    },
    {
      "epoch": 1.8497000315756236,
      "grad_norm": 1.517745018005371,
      "learning_rate": 3.500456108343274e-05,
      "loss": 0.8267,
      "step": 5858
    },
    {
      "epoch": 1.8500157878118093,
      "grad_norm": 1.5102554559707642,
      "learning_rate": 3.5001052557715246e-05,
      "loss": 0.8157,
      "step": 5859
    },
    {
      "epoch": 1.8503315440479948,
      "grad_norm": 1.4988510608673096,
      "learning_rate": 3.499754403199776e-05,
      "loss": 0.8132,
      "step": 5860
    },
    {
      "epoch": 1.8506473002841806,
      "grad_norm": 1.3416732549667358,
      "learning_rate": 3.4994035506280266e-05,
      "loss": 0.8036,
      "step": 5861
    },
    {
      "epoch": 1.8509630565203663,
      "grad_norm": 1.5910085439682007,
      "learning_rate": 3.499052698056277e-05,
      "loss": 0.8391,
      "step": 5862
    },
    {
      "epoch": 1.851278812756552,
      "grad_norm": 1.5653626918792725,
      "learning_rate": 3.498701845484528e-05,
      "loss": 0.8172,
      "step": 5863
    },
    {
      "epoch": 1.8515945689927376,
      "grad_norm": 1.9395679235458374,
      "learning_rate": 3.4983509929127786e-05,
      "loss": 0.8152,
      "step": 5864
    },
    {
      "epoch": 1.8519103252289233,
      "grad_norm": 1.457707166671753,
      "learning_rate": 3.4980001403410286e-05,
      "loss": 0.7843,
      "step": 5865
    },
    {
      "epoch": 1.8522260814651088,
      "grad_norm": 1.5153781175613403,
      "learning_rate": 3.497649287769279e-05,
      "loss": 0.8278,
      "step": 5866
    },
    {
      "epoch": 1.8525418377012945,
      "grad_norm": 1.7235302925109863,
      "learning_rate": 3.49729843519753e-05,
      "loss": 0.7933,
      "step": 5867
    },
    {
      "epoch": 1.8528575939374803,
      "grad_norm": 1.4321563243865967,
      "learning_rate": 3.496947582625781e-05,
      "loss": 0.8003,
      "step": 5868
    },
    {
      "epoch": 1.853173350173666,
      "grad_norm": 1.5199068784713745,
      "learning_rate": 3.4965967300540314e-05,
      "loss": 0.8185,
      "step": 5869
    },
    {
      "epoch": 1.8534891064098518,
      "grad_norm": 1.6151487827301025,
      "learning_rate": 3.496245877482282e-05,
      "loss": 0.807,
      "step": 5870
    },
    {
      "epoch": 1.8538048626460373,
      "grad_norm": 1.6770923137664795,
      "learning_rate": 3.495895024910533e-05,
      "loss": 0.867,
      "step": 5871
    },
    {
      "epoch": 1.8541206188822228,
      "grad_norm": 1.5735334157943726,
      "learning_rate": 3.4955441723387834e-05,
      "loss": 0.7319,
      "step": 5872
    },
    {
      "epoch": 1.8544363751184085,
      "grad_norm": 1.4910534620285034,
      "learning_rate": 3.495193319767034e-05,
      "loss": 0.8433,
      "step": 5873
    },
    {
      "epoch": 1.8547521313545943,
      "grad_norm": 1.924100637435913,
      "learning_rate": 3.494842467195285e-05,
      "loss": 0.8579,
      "step": 5874
    },
    {
      "epoch": 1.85506788759078,
      "grad_norm": 1.5957162380218506,
      "learning_rate": 3.4944916146235355e-05,
      "loss": 0.8462,
      "step": 5875
    },
    {
      "epoch": 1.8553836438269657,
      "grad_norm": 1.9642044305801392,
      "learning_rate": 3.4941407620517855e-05,
      "loss": 0.7973,
      "step": 5876
    },
    {
      "epoch": 1.8556994000631513,
      "grad_norm": 1.4280966520309448,
      "learning_rate": 3.493789909480037e-05,
      "loss": 0.8133,
      "step": 5877
    },
    {
      "epoch": 1.8560151562993368,
      "grad_norm": 1.5283044576644897,
      "learning_rate": 3.4934390569082875e-05,
      "loss": 0.8113,
      "step": 5878
    },
    {
      "epoch": 1.8563309125355225,
      "grad_norm": 1.5749281644821167,
      "learning_rate": 3.493088204336538e-05,
      "loss": 0.8179,
      "step": 5879
    },
    {
      "epoch": 1.8566466687717083,
      "grad_norm": 1.6730461120605469,
      "learning_rate": 3.492737351764789e-05,
      "loss": 0.7972,
      "step": 5880
    },
    {
      "epoch": 1.856962425007894,
      "grad_norm": 1.576575756072998,
      "learning_rate": 3.4923864991930395e-05,
      "loss": 0.7354,
      "step": 5881
    },
    {
      "epoch": 1.8572781812440797,
      "grad_norm": 1.607146143913269,
      "learning_rate": 3.49203564662129e-05,
      "loss": 0.7686,
      "step": 5882
    },
    {
      "epoch": 1.8575939374802652,
      "grad_norm": 1.450456142425537,
      "learning_rate": 3.491684794049541e-05,
      "loss": 0.8279,
      "step": 5883
    },
    {
      "epoch": 1.8579096937164508,
      "grad_norm": 1.8746423721313477,
      "learning_rate": 3.491333941477791e-05,
      "loss": 0.913,
      "step": 5884
    },
    {
      "epoch": 1.8582254499526365,
      "grad_norm": 1.4949873685836792,
      "learning_rate": 3.4909830889060416e-05,
      "loss": 0.7902,
      "step": 5885
    },
    {
      "epoch": 1.8585412061888222,
      "grad_norm": 1.5893785953521729,
      "learning_rate": 3.490632236334292e-05,
      "loss": 0.8889,
      "step": 5886
    },
    {
      "epoch": 1.858856962425008,
      "grad_norm": 1.405253291130066,
      "learning_rate": 3.490281383762543e-05,
      "loss": 0.7951,
      "step": 5887
    },
    {
      "epoch": 1.8591727186611937,
      "grad_norm": 1.6688015460968018,
      "learning_rate": 3.4899305311907936e-05,
      "loss": 0.8946,
      "step": 5888
    },
    {
      "epoch": 1.8594884748973792,
      "grad_norm": 1.4087270498275757,
      "learning_rate": 3.489579678619044e-05,
      "loss": 0.7749,
      "step": 5889
    },
    {
      "epoch": 1.8598042311335647,
      "grad_norm": 1.5927802324295044,
      "learning_rate": 3.489228826047295e-05,
      "loss": 0.7531,
      "step": 5890
    },
    {
      "epoch": 1.8601199873697505,
      "grad_norm": 1.6765002012252808,
      "learning_rate": 3.488877973475546e-05,
      "loss": 0.8436,
      "step": 5891
    },
    {
      "epoch": 1.8604357436059362,
      "grad_norm": 1.5520349740982056,
      "learning_rate": 3.4885271209037964e-05,
      "loss": 0.8715,
      "step": 5892
    },
    {
      "epoch": 1.860751499842122,
      "grad_norm": 1.3830060958862305,
      "learning_rate": 3.488176268332047e-05,
      "loss": 0.8171,
      "step": 5893
    },
    {
      "epoch": 1.8610672560783077,
      "grad_norm": 1.7463703155517578,
      "learning_rate": 3.487825415760298e-05,
      "loss": 0.8347,
      "step": 5894
    },
    {
      "epoch": 1.8613830123144932,
      "grad_norm": 1.8890496492385864,
      "learning_rate": 3.4874745631885484e-05,
      "loss": 0.8087,
      "step": 5895
    },
    {
      "epoch": 1.8616987685506787,
      "grad_norm": 1.514704704284668,
      "learning_rate": 3.487123710616799e-05,
      "loss": 0.741,
      "step": 5896
    },
    {
      "epoch": 1.8620145247868645,
      "grad_norm": 1.4326145648956299,
      "learning_rate": 3.48677285804505e-05,
      "loss": 0.8139,
      "step": 5897
    },
    {
      "epoch": 1.8623302810230502,
      "grad_norm": 1.4706913232803345,
      "learning_rate": 3.4864220054733004e-05,
      "loss": 0.8033,
      "step": 5898
    },
    {
      "epoch": 1.862646037259236,
      "grad_norm": 1.7480053901672363,
      "learning_rate": 3.486071152901551e-05,
      "loss": 0.8145,
      "step": 5899
    },
    {
      "epoch": 1.8629617934954217,
      "grad_norm": 1.6722049713134766,
      "learning_rate": 3.485720300329802e-05,
      "loss": 0.7604,
      "step": 5900
    },
    {
      "epoch": 1.8632775497316072,
      "grad_norm": 1.23417067527771,
      "learning_rate": 3.4853694477580525e-05,
      "loss": 0.7157,
      "step": 5901
    },
    {
      "epoch": 1.8635933059677927,
      "grad_norm": 1.6794359683990479,
      "learning_rate": 3.485018595186303e-05,
      "loss": 0.8756,
      "step": 5902
    },
    {
      "epoch": 1.8639090622039785,
      "grad_norm": 1.439768671989441,
      "learning_rate": 3.484667742614554e-05,
      "loss": 0.8196,
      "step": 5903
    },
    {
      "epoch": 1.8642248184401642,
      "grad_norm": 1.3198407888412476,
      "learning_rate": 3.484316890042804e-05,
      "loss": 0.7804,
      "step": 5904
    },
    {
      "epoch": 1.86454057467635,
      "grad_norm": 1.7481340169906616,
      "learning_rate": 3.4839660374710545e-05,
      "loss": 0.8375,
      "step": 5905
    },
    {
      "epoch": 1.8648563309125357,
      "grad_norm": 1.6905640363693237,
      "learning_rate": 3.483615184899305e-05,
      "loss": 0.7771,
      "step": 5906
    },
    {
      "epoch": 1.8651720871487212,
      "grad_norm": 1.6961987018585205,
      "learning_rate": 3.483264332327556e-05,
      "loss": 0.8391,
      "step": 5907
    },
    {
      "epoch": 1.8654878433849067,
      "grad_norm": 2.0850746631622314,
      "learning_rate": 3.4829134797558066e-05,
      "loss": 0.8533,
      "step": 5908
    },
    {
      "epoch": 1.8658035996210924,
      "grad_norm": 2.0472161769866943,
      "learning_rate": 3.482562627184057e-05,
      "loss": 0.868,
      "step": 5909
    },
    {
      "epoch": 1.8661193558572782,
      "grad_norm": 1.4708845615386963,
      "learning_rate": 3.482211774612308e-05,
      "loss": 0.8581,
      "step": 5910
    },
    {
      "epoch": 1.866435112093464,
      "grad_norm": 1.594724178314209,
      "learning_rate": 3.4818609220405586e-05,
      "loss": 0.8285,
      "step": 5911
    },
    {
      "epoch": 1.8667508683296496,
      "grad_norm": 1.5660326480865479,
      "learning_rate": 3.481510069468809e-05,
      "loss": 0.7661,
      "step": 5912
    },
    {
      "epoch": 1.8670666245658352,
      "grad_norm": 1.6207331418991089,
      "learning_rate": 3.48115921689706e-05,
      "loss": 0.8213,
      "step": 5913
    },
    {
      "epoch": 1.8673823808020207,
      "grad_norm": 1.4295165538787842,
      "learning_rate": 3.480808364325311e-05,
      "loss": 0.7619,
      "step": 5914
    },
    {
      "epoch": 1.8676981370382064,
      "grad_norm": 1.683862566947937,
      "learning_rate": 3.4804575117535614e-05,
      "loss": 0.7924,
      "step": 5915
    },
    {
      "epoch": 1.8680138932743922,
      "grad_norm": 1.6118426322937012,
      "learning_rate": 3.480106659181812e-05,
      "loss": 0.7543,
      "step": 5916
    },
    {
      "epoch": 1.868329649510578,
      "grad_norm": 2.3119704723358154,
      "learning_rate": 3.479755806610063e-05,
      "loss": 0.8794,
      "step": 5917
    },
    {
      "epoch": 1.8686454057467636,
      "grad_norm": 1.5308163166046143,
      "learning_rate": 3.4794049540383134e-05,
      "loss": 0.8324,
      "step": 5918
    },
    {
      "epoch": 1.8689611619829491,
      "grad_norm": 2.139662742614746,
      "learning_rate": 3.479054101466564e-05,
      "loss": 0.8277,
      "step": 5919
    },
    {
      "epoch": 1.8692769182191347,
      "grad_norm": 1.5550779104232788,
      "learning_rate": 3.478703248894815e-05,
      "loss": 0.8521,
      "step": 5920
    },
    {
      "epoch": 1.8695926744553204,
      "grad_norm": 1.7536754608154297,
      "learning_rate": 3.4783523963230654e-05,
      "loss": 0.7564,
      "step": 5921
    },
    {
      "epoch": 1.8699084306915061,
      "grad_norm": 1.5094937086105347,
      "learning_rate": 3.478001543751316e-05,
      "loss": 0.7368,
      "step": 5922
    },
    {
      "epoch": 1.8702241869276919,
      "grad_norm": 1.578058123588562,
      "learning_rate": 3.477650691179566e-05,
      "loss": 0.8288,
      "step": 5923
    },
    {
      "epoch": 1.8705399431638776,
      "grad_norm": 1.5651580095291138,
      "learning_rate": 3.477299838607817e-05,
      "loss": 0.7541,
      "step": 5924
    },
    {
      "epoch": 1.8708556994000631,
      "grad_norm": 1.4915457963943481,
      "learning_rate": 3.4769489860360675e-05,
      "loss": 0.8234,
      "step": 5925
    },
    {
      "epoch": 1.8711714556362489,
      "grad_norm": 1.578057050704956,
      "learning_rate": 3.476598133464318e-05,
      "loss": 0.8353,
      "step": 5926
    },
    {
      "epoch": 1.8714872118724344,
      "grad_norm": 1.5771353244781494,
      "learning_rate": 3.476247280892569e-05,
      "loss": 0.7612,
      "step": 5927
    },
    {
      "epoch": 1.8718029681086201,
      "grad_norm": 1.7461971044540405,
      "learning_rate": 3.4758964283208195e-05,
      "loss": 0.7674,
      "step": 5928
    },
    {
      "epoch": 1.8721187243448059,
      "grad_norm": 1.5831201076507568,
      "learning_rate": 3.47554557574907e-05,
      "loss": 0.794,
      "step": 5929
    },
    {
      "epoch": 1.8724344805809916,
      "grad_norm": 1.5036020278930664,
      "learning_rate": 3.4751947231773216e-05,
      "loss": 0.8395,
      "step": 5930
    },
    {
      "epoch": 1.8727502368171771,
      "grad_norm": 1.6445670127868652,
      "learning_rate": 3.474843870605572e-05,
      "loss": 0.8141,
      "step": 5931
    },
    {
      "epoch": 1.8730659930533629,
      "grad_norm": 1.6093717813491821,
      "learning_rate": 3.474493018033822e-05,
      "loss": 0.8399,
      "step": 5932
    },
    {
      "epoch": 1.8733817492895484,
      "grad_norm": 1.7797726392745972,
      "learning_rate": 3.474142165462073e-05,
      "loss": 0.8473,
      "step": 5933
    },
    {
      "epoch": 1.873697505525734,
      "grad_norm": 1.551011085510254,
      "learning_rate": 3.4737913128903236e-05,
      "loss": 0.8586,
      "step": 5934
    },
    {
      "epoch": 1.8740132617619198,
      "grad_norm": 1.4278476238250732,
      "learning_rate": 3.473440460318574e-05,
      "loss": 0.8198,
      "step": 5935
    },
    {
      "epoch": 1.8743290179981056,
      "grad_norm": 1.4769338369369507,
      "learning_rate": 3.473089607746825e-05,
      "loss": 0.7877,
      "step": 5936
    },
    {
      "epoch": 1.874644774234291,
      "grad_norm": 1.7522413730621338,
      "learning_rate": 3.472738755175076e-05,
      "loss": 0.8007,
      "step": 5937
    },
    {
      "epoch": 1.8749605304704768,
      "grad_norm": 1.6246000528335571,
      "learning_rate": 3.4723879026033263e-05,
      "loss": 0.7347,
      "step": 5938
    },
    {
      "epoch": 1.8752762867066624,
      "grad_norm": 1.5087512731552124,
      "learning_rate": 3.472037050031577e-05,
      "loss": 0.7139,
      "step": 5939
    },
    {
      "epoch": 1.875592042942848,
      "grad_norm": 1.4432265758514404,
      "learning_rate": 3.471686197459828e-05,
      "loss": 0.8047,
      "step": 5940
    },
    {
      "epoch": 1.8759077991790338,
      "grad_norm": 1.5028444528579712,
      "learning_rate": 3.4713353448880784e-05,
      "loss": 0.7556,
      "step": 5941
    },
    {
      "epoch": 1.8762235554152196,
      "grad_norm": 1.756437063217163,
      "learning_rate": 3.4709844923163284e-05,
      "loss": 0.8586,
      "step": 5942
    },
    {
      "epoch": 1.876539311651405,
      "grad_norm": 1.744747281074524,
      "learning_rate": 3.470633639744579e-05,
      "loss": 0.8809,
      "step": 5943
    },
    {
      "epoch": 1.8768550678875908,
      "grad_norm": 1.5890856981277466,
      "learning_rate": 3.47028278717283e-05,
      "loss": 0.791,
      "step": 5944
    },
    {
      "epoch": 1.8771708241237763,
      "grad_norm": 1.7070366144180298,
      "learning_rate": 3.4699319346010804e-05,
      "loss": 0.9192,
      "step": 5945
    },
    {
      "epoch": 1.877486580359962,
      "grad_norm": 1.7561713457107544,
      "learning_rate": 3.469581082029331e-05,
      "loss": 0.803,
      "step": 5946
    },
    {
      "epoch": 1.8778023365961478,
      "grad_norm": 1.763536810874939,
      "learning_rate": 3.4692302294575825e-05,
      "loss": 0.793,
      "step": 5947
    },
    {
      "epoch": 1.8781180928323336,
      "grad_norm": 1.6366345882415771,
      "learning_rate": 3.468879376885833e-05,
      "loss": 0.7969,
      "step": 5948
    },
    {
      "epoch": 1.878433849068519,
      "grad_norm": 1.7211172580718994,
      "learning_rate": 3.468528524314084e-05,
      "loss": 0.8639,
      "step": 5949
    },
    {
      "epoch": 1.8787496053047048,
      "grad_norm": 1.78987455368042,
      "learning_rate": 3.4681776717423345e-05,
      "loss": 0.8269,
      "step": 5950
    },
    {
      "epoch": 1.8790653615408903,
      "grad_norm": 1.7112611532211304,
      "learning_rate": 3.4678268191705845e-05,
      "loss": 0.799,
      "step": 5951
    },
    {
      "epoch": 1.879381117777076,
      "grad_norm": 1.6511179208755493,
      "learning_rate": 3.467475966598835e-05,
      "loss": 0.8215,
      "step": 5952
    },
    {
      "epoch": 1.8796968740132618,
      "grad_norm": 1.539608120918274,
      "learning_rate": 3.467125114027086e-05,
      "loss": 0.8669,
      "step": 5953
    },
    {
      "epoch": 1.8800126302494475,
      "grad_norm": 1.7341794967651367,
      "learning_rate": 3.4667742614553366e-05,
      "loss": 0.8715,
      "step": 5954
    },
    {
      "epoch": 1.880328386485633,
      "grad_norm": 1.9370721578598022,
      "learning_rate": 3.466423408883587e-05,
      "loss": 0.8548,
      "step": 5955
    },
    {
      "epoch": 1.8806441427218188,
      "grad_norm": 1.4808419942855835,
      "learning_rate": 3.466072556311838e-05,
      "loss": 0.8549,
      "step": 5956
    },
    {
      "epoch": 1.8809598989580043,
      "grad_norm": 1.6106830835342407,
      "learning_rate": 3.4657217037400886e-05,
      "loss": 0.7882,
      "step": 5957
    },
    {
      "epoch": 1.88127565519419,
      "grad_norm": 1.866086721420288,
      "learning_rate": 3.465370851168339e-05,
      "loss": 0.8327,
      "step": 5958
    },
    {
      "epoch": 1.8815914114303758,
      "grad_norm": 1.8336613178253174,
      "learning_rate": 3.46501999859659e-05,
      "loss": 0.8586,
      "step": 5959
    },
    {
      "epoch": 1.8819071676665615,
      "grad_norm": 1.550226092338562,
      "learning_rate": 3.4646691460248407e-05,
      "loss": 0.7525,
      "step": 5960
    },
    {
      "epoch": 1.882222923902747,
      "grad_norm": 1.7071210145950317,
      "learning_rate": 3.464318293453091e-05,
      "loss": 0.9349,
      "step": 5961
    },
    {
      "epoch": 1.8825386801389328,
      "grad_norm": 1.658603310585022,
      "learning_rate": 3.4639674408813413e-05,
      "loss": 0.8619,
      "step": 5962
    },
    {
      "epoch": 1.8828544363751183,
      "grad_norm": 1.4947222471237183,
      "learning_rate": 3.463616588309592e-05,
      "loss": 0.8978,
      "step": 5963
    },
    {
      "epoch": 1.883170192611304,
      "grad_norm": 1.7277193069458008,
      "learning_rate": 3.463265735737843e-05,
      "loss": 0.8265,
      "step": 5964
    },
    {
      "epoch": 1.8834859488474898,
      "grad_norm": 1.7787446975708008,
      "learning_rate": 3.462914883166094e-05,
      "loss": 0.7538,
      "step": 5965
    },
    {
      "epoch": 1.8838017050836755,
      "grad_norm": 1.556546926498413,
      "learning_rate": 3.462564030594345e-05,
      "loss": 0.8566,
      "step": 5966
    },
    {
      "epoch": 1.884117461319861,
      "grad_norm": 1.5876212120056152,
      "learning_rate": 3.4622131780225954e-05,
      "loss": 0.7752,
      "step": 5967
    },
    {
      "epoch": 1.8844332175560468,
      "grad_norm": 1.669906735420227,
      "learning_rate": 3.461862325450846e-05,
      "loss": 0.8094,
      "step": 5968
    },
    {
      "epoch": 1.8847489737922323,
      "grad_norm": 1.6791768074035645,
      "learning_rate": 3.461511472879097e-05,
      "loss": 0.7673,
      "step": 5969
    },
    {
      "epoch": 1.885064730028418,
      "grad_norm": 1.6261039972305298,
      "learning_rate": 3.461160620307347e-05,
      "loss": 0.8421,
      "step": 5970
    },
    {
      "epoch": 1.8853804862646037,
      "grad_norm": 1.6963905096054077,
      "learning_rate": 3.4608097677355975e-05,
      "loss": 0.8696,
      "step": 5971
    },
    {
      "epoch": 1.8856962425007895,
      "grad_norm": 1.4128772020339966,
      "learning_rate": 3.460458915163848e-05,
      "loss": 0.7728,
      "step": 5972
    },
    {
      "epoch": 1.886011998736975,
      "grad_norm": 1.4653892517089844,
      "learning_rate": 3.460108062592099e-05,
      "loss": 0.7742,
      "step": 5973
    },
    {
      "epoch": 1.8863277549731607,
      "grad_norm": 1.8178144693374634,
      "learning_rate": 3.4597572100203495e-05,
      "loss": 0.8171,
      "step": 5974
    },
    {
      "epoch": 1.8866435112093463,
      "grad_norm": 1.571824312210083,
      "learning_rate": 3.4594063574486e-05,
      "loss": 0.7529,
      "step": 5975
    },
    {
      "epoch": 1.886959267445532,
      "grad_norm": 1.714598536491394,
      "learning_rate": 3.459055504876851e-05,
      "loss": 0.7718,
      "step": 5976
    },
    {
      "epoch": 1.8872750236817177,
      "grad_norm": 1.6628999710083008,
      "learning_rate": 3.4587046523051016e-05,
      "loss": 0.8462,
      "step": 5977
    },
    {
      "epoch": 1.8875907799179035,
      "grad_norm": 1.7542319297790527,
      "learning_rate": 3.458353799733352e-05,
      "loss": 0.7663,
      "step": 5978
    },
    {
      "epoch": 1.887906536154089,
      "grad_norm": 1.6935679912567139,
      "learning_rate": 3.458002947161603e-05,
      "loss": 0.8083,
      "step": 5979
    },
    {
      "epoch": 1.8882222923902747,
      "grad_norm": 1.6027164459228516,
      "learning_rate": 3.4576520945898536e-05,
      "loss": 0.8854,
      "step": 5980
    },
    {
      "epoch": 1.8885380486264602,
      "grad_norm": 1.685696005821228,
      "learning_rate": 3.4573012420181036e-05,
      "loss": 0.793,
      "step": 5981
    },
    {
      "epoch": 1.888853804862646,
      "grad_norm": 1.558816909790039,
      "learning_rate": 3.456950389446355e-05,
      "loss": 0.743,
      "step": 5982
    },
    {
      "epoch": 1.8891695610988317,
      "grad_norm": 1.7199510335922241,
      "learning_rate": 3.4565995368746056e-05,
      "loss": 0.8291,
      "step": 5983
    },
    {
      "epoch": 1.8894853173350175,
      "grad_norm": 1.7398945093154907,
      "learning_rate": 3.456248684302856e-05,
      "loss": 0.9085,
      "step": 5984
    },
    {
      "epoch": 1.8898010735712032,
      "grad_norm": 2.092740774154663,
      "learning_rate": 3.455897831731107e-05,
      "loss": 0.7826,
      "step": 5985
    },
    {
      "epoch": 1.8901168298073887,
      "grad_norm": 1.7366584539413452,
      "learning_rate": 3.455546979159358e-05,
      "loss": 0.8077,
      "step": 5986
    },
    {
      "epoch": 1.8904325860435742,
      "grad_norm": 1.8012280464172363,
      "learning_rate": 3.4551961265876084e-05,
      "loss": 0.8245,
      "step": 5987
    },
    {
      "epoch": 1.89074834227976,
      "grad_norm": 1.5165512561798096,
      "learning_rate": 3.454845274015859e-05,
      "loss": 0.8435,
      "step": 5988
    },
    {
      "epoch": 1.8910640985159457,
      "grad_norm": 1.8824572563171387,
      "learning_rate": 3.454494421444109e-05,
      "loss": 0.8534,
      "step": 5989
    },
    {
      "epoch": 1.8913798547521314,
      "grad_norm": 1.6762734651565552,
      "learning_rate": 3.45414356887236e-05,
      "loss": 0.8463,
      "step": 5990
    },
    {
      "epoch": 1.8916956109883172,
      "grad_norm": 1.6500449180603027,
      "learning_rate": 3.4537927163006104e-05,
      "loss": 0.7584,
      "step": 5991
    },
    {
      "epoch": 1.8920113672245027,
      "grad_norm": 2.2236437797546387,
      "learning_rate": 3.453441863728861e-05,
      "loss": 0.7324,
      "step": 5992
    },
    {
      "epoch": 1.8923271234606882,
      "grad_norm": 1.597829818725586,
      "learning_rate": 3.453091011157112e-05,
      "loss": 0.7597,
      "step": 5993
    },
    {
      "epoch": 1.892642879696874,
      "grad_norm": 1.779622197151184,
      "learning_rate": 3.4527401585853625e-05,
      "loss": 0.8354,
      "step": 5994
    },
    {
      "epoch": 1.8929586359330597,
      "grad_norm": 1.6349881887435913,
      "learning_rate": 3.452389306013613e-05,
      "loss": 0.7506,
      "step": 5995
    },
    {
      "epoch": 1.8932743921692454,
      "grad_norm": 1.7420233488082886,
      "learning_rate": 3.452038453441864e-05,
      "loss": 0.8236,
      "step": 5996
    },
    {
      "epoch": 1.8935901484054312,
      "grad_norm": 1.582596778869629,
      "learning_rate": 3.4516876008701145e-05,
      "loss": 0.8398,
      "step": 5997
    },
    {
      "epoch": 1.8939059046416167,
      "grad_norm": 1.6589572429656982,
      "learning_rate": 3.451336748298365e-05,
      "loss": 0.8152,
      "step": 5998
    },
    {
      "epoch": 1.8942216608778022,
      "grad_norm": 1.8514900207519531,
      "learning_rate": 3.450985895726616e-05,
      "loss": 0.8496,
      "step": 5999
    },
    {
      "epoch": 1.894537417113988,
      "grad_norm": 1.9089871644973755,
      "learning_rate": 3.4506350431548666e-05,
      "loss": 0.8276,
      "step": 6000
    },
    {
      "epoch": 1.894537417113988,
      "eval_loss": 0.8018169403076172,
      "eval_runtime": 4.6532,
      "eval_samples_per_second": 535.116,
      "eval_steps_per_second": 4.513,
      "step": 6000
    },
    {
      "epoch": 1.8948531733501737,
      "grad_norm": 1.4901862144470215,
      "learning_rate": 3.450284190583117e-05,
      "loss": 0.8156,
      "step": 6001
    },
    {
      "epoch": 1.8951689295863594,
      "grad_norm": 1.3226667642593384,
      "learning_rate": 3.449933338011368e-05,
      "loss": 0.7755,
      "step": 6002
    },
    {
      "epoch": 1.8954846858225451,
      "grad_norm": 1.7962757349014282,
      "learning_rate": 3.4495824854396186e-05,
      "loss": 0.7488,
      "step": 6003
    },
    {
      "epoch": 1.8958004420587307,
      "grad_norm": 1.5010707378387451,
      "learning_rate": 3.449231632867869e-05,
      "loss": 0.7807,
      "step": 6004
    },
    {
      "epoch": 1.8961161982949162,
      "grad_norm": 1.6418774127960205,
      "learning_rate": 3.44888078029612e-05,
      "loss": 0.8825,
      "step": 6005
    },
    {
      "epoch": 1.896431954531102,
      "grad_norm": 1.5780160427093506,
      "learning_rate": 3.4485299277243706e-05,
      "loss": 0.7634,
      "step": 6006
    },
    {
      "epoch": 1.8967477107672877,
      "grad_norm": 1.7857470512390137,
      "learning_rate": 3.448179075152621e-05,
      "loss": 0.7169,
      "step": 6007
    },
    {
      "epoch": 1.8970634670034734,
      "grad_norm": 1.6138498783111572,
      "learning_rate": 3.447828222580872e-05,
      "loss": 0.7646,
      "step": 6008
    },
    {
      "epoch": 1.8973792232396591,
      "grad_norm": 1.6302531957626343,
      "learning_rate": 3.447477370009122e-05,
      "loss": 0.7698,
      "step": 6009
    },
    {
      "epoch": 1.8976949794758446,
      "grad_norm": 1.4761991500854492,
      "learning_rate": 3.447126517437373e-05,
      "loss": 0.809,
      "step": 6010
    },
    {
      "epoch": 1.8980107357120302,
      "grad_norm": 1.4772412776947021,
      "learning_rate": 3.4467756648656234e-05,
      "loss": 0.7474,
      "step": 6011
    },
    {
      "epoch": 1.898326491948216,
      "grad_norm": 1.593679428100586,
      "learning_rate": 3.446424812293874e-05,
      "loss": 0.7592,
      "step": 6012
    },
    {
      "epoch": 1.8986422481844016,
      "grad_norm": 1.6071380376815796,
      "learning_rate": 3.446073959722125e-05,
      "loss": 0.7807,
      "step": 6013
    },
    {
      "epoch": 1.8989580044205874,
      "grad_norm": 1.6962323188781738,
      "learning_rate": 3.4457231071503754e-05,
      "loss": 0.8339,
      "step": 6014
    },
    {
      "epoch": 1.8992737606567731,
      "grad_norm": 1.4001479148864746,
      "learning_rate": 3.445372254578626e-05,
      "loss": 0.763,
      "step": 6015
    },
    {
      "epoch": 1.8995895168929586,
      "grad_norm": 1.6615264415740967,
      "learning_rate": 3.445021402006877e-05,
      "loss": 0.8688,
      "step": 6016
    },
    {
      "epoch": 1.8999052731291441,
      "grad_norm": 1.754500150680542,
      "learning_rate": 3.4446705494351275e-05,
      "loss": 0.892,
      "step": 6017
    },
    {
      "epoch": 1.9002210293653299,
      "grad_norm": 2.000978708267212,
      "learning_rate": 3.444319696863378e-05,
      "loss": 0.8109,
      "step": 6018
    },
    {
      "epoch": 1.9005367856015156,
      "grad_norm": 1.8442118167877197,
      "learning_rate": 3.443968844291629e-05,
      "loss": 0.9181,
      "step": 6019
    },
    {
      "epoch": 1.9008525418377014,
      "grad_norm": 1.732500672340393,
      "learning_rate": 3.4436179917198795e-05,
      "loss": 0.7343,
      "step": 6020
    },
    {
      "epoch": 1.901168298073887,
      "grad_norm": 2.0572307109832764,
      "learning_rate": 3.44326713914813e-05,
      "loss": 0.8349,
      "step": 6021
    },
    {
      "epoch": 1.9014840543100726,
      "grad_norm": 1.6267365217208862,
      "learning_rate": 3.442916286576381e-05,
      "loss": 0.7623,
      "step": 6022
    },
    {
      "epoch": 1.9017998105462581,
      "grad_norm": 1.5624520778656006,
      "learning_rate": 3.4425654340046315e-05,
      "loss": 0.8032,
      "step": 6023
    },
    {
      "epoch": 1.9021155667824439,
      "grad_norm": 1.984693169593811,
      "learning_rate": 3.442214581432882e-05,
      "loss": 0.8361,
      "step": 6024
    },
    {
      "epoch": 1.9024313230186296,
      "grad_norm": 1.7114468812942505,
      "learning_rate": 3.441863728861133e-05,
      "loss": 0.763,
      "step": 6025
    },
    {
      "epoch": 1.9027470792548153,
      "grad_norm": 1.9881138801574707,
      "learning_rate": 3.4415128762893836e-05,
      "loss": 0.8455,
      "step": 6026
    },
    {
      "epoch": 1.903062835491001,
      "grad_norm": 1.564346194267273,
      "learning_rate": 3.441162023717634e-05,
      "loss": 0.8136,
      "step": 6027
    },
    {
      "epoch": 1.9033785917271866,
      "grad_norm": 1.552574872970581,
      "learning_rate": 3.440811171145884e-05,
      "loss": 0.8145,
      "step": 6028
    },
    {
      "epoch": 1.9036943479633721,
      "grad_norm": 1.5122597217559814,
      "learning_rate": 3.440460318574135e-05,
      "loss": 0.7603,
      "step": 6029
    },
    {
      "epoch": 1.9040101041995579,
      "grad_norm": 1.5370080471038818,
      "learning_rate": 3.4401094660023856e-05,
      "loss": 0.8529,
      "step": 6030
    },
    {
      "epoch": 1.9043258604357436,
      "grad_norm": 1.592112421989441,
      "learning_rate": 3.439758613430636e-05,
      "loss": 0.7861,
      "step": 6031
    },
    {
      "epoch": 1.9046416166719293,
      "grad_norm": 1.7260569334030151,
      "learning_rate": 3.439407760858887e-05,
      "loss": 0.8169,
      "step": 6032
    },
    {
      "epoch": 1.904957372908115,
      "grad_norm": 2.0342743396759033,
      "learning_rate": 3.439056908287138e-05,
      "loss": 0.8477,
      "step": 6033
    },
    {
      "epoch": 1.9052731291443006,
      "grad_norm": 1.9562392234802246,
      "learning_rate": 3.4387060557153884e-05,
      "loss": 0.9415,
      "step": 6034
    },
    {
      "epoch": 1.9055888853804863,
      "grad_norm": 1.9631927013397217,
      "learning_rate": 3.43835520314364e-05,
      "loss": 0.8139,
      "step": 6035
    },
    {
      "epoch": 1.9059046416166718,
      "grad_norm": 1.5464006662368774,
      "learning_rate": 3.4380043505718904e-05,
      "loss": 0.8301,
      "step": 6036
    },
    {
      "epoch": 1.9062203978528576,
      "grad_norm": 1.676655888557434,
      "learning_rate": 3.4376534980001404e-05,
      "loss": 0.8479,
      "step": 6037
    },
    {
      "epoch": 1.9065361540890433,
      "grad_norm": 1.792374849319458,
      "learning_rate": 3.437302645428391e-05,
      "loss": 0.8448,
      "step": 6038
    },
    {
      "epoch": 1.906851910325229,
      "grad_norm": 1.9529852867126465,
      "learning_rate": 3.436951792856642e-05,
      "loss": 0.7948,
      "step": 6039
    },
    {
      "epoch": 1.9071676665614146,
      "grad_norm": 1.7871599197387695,
      "learning_rate": 3.4366009402848925e-05,
      "loss": 0.8127,
      "step": 6040
    },
    {
      "epoch": 1.9074834227976003,
      "grad_norm": 1.7279584407806396,
      "learning_rate": 3.436250087713143e-05,
      "loss": 0.8354,
      "step": 6041
    },
    {
      "epoch": 1.9077991790337858,
      "grad_norm": 1.5021716356277466,
      "learning_rate": 3.435899235141394e-05,
      "loss": 0.7834,
      "step": 6042
    },
    {
      "epoch": 1.9081149352699716,
      "grad_norm": 1.5199625492095947,
      "learning_rate": 3.4355483825696445e-05,
      "loss": 0.8016,
      "step": 6043
    },
    {
      "epoch": 1.9084306915061573,
      "grad_norm": 1.5900851488113403,
      "learning_rate": 3.435197529997895e-05,
      "loss": 0.81,
      "step": 6044
    },
    {
      "epoch": 1.908746447742343,
      "grad_norm": 1.447289228439331,
      "learning_rate": 3.434846677426146e-05,
      "loss": 0.7902,
      "step": 6045
    },
    {
      "epoch": 1.9090622039785285,
      "grad_norm": 1.6488759517669678,
      "learning_rate": 3.4344958248543965e-05,
      "loss": 0.7682,
      "step": 6046
    },
    {
      "epoch": 1.9093779602147143,
      "grad_norm": 1.712686538696289,
      "learning_rate": 3.4341449722826465e-05,
      "loss": 0.9344,
      "step": 6047
    },
    {
      "epoch": 1.9096937164508998,
      "grad_norm": 1.6084325313568115,
      "learning_rate": 3.433794119710897e-05,
      "loss": 0.8063,
      "step": 6048
    },
    {
      "epoch": 1.9100094726870855,
      "grad_norm": 1.8277891874313354,
      "learning_rate": 3.433443267139148e-05,
      "loss": 0.767,
      "step": 6049
    },
    {
      "epoch": 1.9103252289232713,
      "grad_norm": 1.7767585515975952,
      "learning_rate": 3.4330924145673986e-05,
      "loss": 0.8349,
      "step": 6050
    },
    {
      "epoch": 1.910640985159457,
      "grad_norm": 1.666571021080017,
      "learning_rate": 3.432741561995649e-05,
      "loss": 0.8244,
      "step": 6051
    },
    {
      "epoch": 1.9109567413956425,
      "grad_norm": 1.401606559753418,
      "learning_rate": 3.4323907094239006e-05,
      "loss": 0.7772,
      "step": 6052
    },
    {
      "epoch": 1.9112724976318283,
      "grad_norm": 1.6714173555374146,
      "learning_rate": 3.432039856852151e-05,
      "loss": 0.8275,
      "step": 6053
    },
    {
      "epoch": 1.9115882538680138,
      "grad_norm": 1.7354278564453125,
      "learning_rate": 3.431689004280402e-05,
      "loss": 0.8231,
      "step": 6054
    },
    {
      "epoch": 1.9119040101041995,
      "grad_norm": 1.8952463865280151,
      "learning_rate": 3.431338151708653e-05,
      "loss": 0.8384,
      "step": 6055
    },
    {
      "epoch": 1.9122197663403853,
      "grad_norm": 1.6706159114837646,
      "learning_rate": 3.430987299136903e-05,
      "loss": 0.806,
      "step": 6056
    },
    {
      "epoch": 1.912535522576571,
      "grad_norm": 1.4987882375717163,
      "learning_rate": 3.4306364465651534e-05,
      "loss": 0.7842,
      "step": 6057
    },
    {
      "epoch": 1.9128512788127565,
      "grad_norm": 1.5363301038742065,
      "learning_rate": 3.430285593993404e-05,
      "loss": 0.8047,
      "step": 6058
    },
    {
      "epoch": 1.9131670350489423,
      "grad_norm": 1.6370660066604614,
      "learning_rate": 3.429934741421655e-05,
      "loss": 0.8881,
      "step": 6059
    },
    {
      "epoch": 1.9134827912851278,
      "grad_norm": 1.8531221151351929,
      "learning_rate": 3.4295838888499054e-05,
      "loss": 0.8629,
      "step": 6060
    },
    {
      "epoch": 1.9137985475213135,
      "grad_norm": 1.416321039199829,
      "learning_rate": 3.429233036278156e-05,
      "loss": 0.7563,
      "step": 6061
    },
    {
      "epoch": 1.9141143037574992,
      "grad_norm": 1.6741551160812378,
      "learning_rate": 3.428882183706407e-05,
      "loss": 0.8765,
      "step": 6062
    },
    {
      "epoch": 1.914430059993685,
      "grad_norm": 1.6431457996368408,
      "learning_rate": 3.4285313311346574e-05,
      "loss": 0.7953,
      "step": 6063
    },
    {
      "epoch": 1.9147458162298705,
      "grad_norm": 1.6028872728347778,
      "learning_rate": 3.428180478562908e-05,
      "loss": 0.8249,
      "step": 6064
    },
    {
      "epoch": 1.9150615724660562,
      "grad_norm": 1.7014741897583008,
      "learning_rate": 3.427829625991159e-05,
      "loss": 0.839,
      "step": 6065
    },
    {
      "epoch": 1.9153773287022418,
      "grad_norm": 1.6454497575759888,
      "learning_rate": 3.4274787734194095e-05,
      "loss": 0.8637,
      "step": 6066
    },
    {
      "epoch": 1.9156930849384275,
      "grad_norm": 1.5127921104431152,
      "learning_rate": 3.4271279208476595e-05,
      "loss": 0.8294,
      "step": 6067
    },
    {
      "epoch": 1.9160088411746132,
      "grad_norm": 1.5776758193969727,
      "learning_rate": 3.42677706827591e-05,
      "loss": 0.8378,
      "step": 6068
    },
    {
      "epoch": 1.916324597410799,
      "grad_norm": 1.6622676849365234,
      "learning_rate": 3.426426215704161e-05,
      "loss": 0.7532,
      "step": 6069
    },
    {
      "epoch": 1.9166403536469845,
      "grad_norm": 1.4396865367889404,
      "learning_rate": 3.426075363132412e-05,
      "loss": 0.7522,
      "step": 6070
    },
    {
      "epoch": 1.9169561098831702,
      "grad_norm": 1.6908659934997559,
      "learning_rate": 3.425724510560663e-05,
      "loss": 0.7878,
      "step": 6071
    },
    {
      "epoch": 1.9172718661193557,
      "grad_norm": 1.335086464881897,
      "learning_rate": 3.4253736579889136e-05,
      "loss": 0.8035,
      "step": 6072
    },
    {
      "epoch": 1.9175876223555415,
      "grad_norm": 1.6378828287124634,
      "learning_rate": 3.425022805417164e-05,
      "loss": 0.8668,
      "step": 6073
    },
    {
      "epoch": 1.9179033785917272,
      "grad_norm": 1.4509700536727905,
      "learning_rate": 3.424671952845415e-05,
      "loss": 0.8343,
      "step": 6074
    },
    {
      "epoch": 1.918219134827913,
      "grad_norm": 1.6996731758117676,
      "learning_rate": 3.424321100273665e-05,
      "loss": 0.7986,
      "step": 6075
    },
    {
      "epoch": 1.9185348910640985,
      "grad_norm": 1.4782954454421997,
      "learning_rate": 3.4239702477019156e-05,
      "loss": 0.8165,
      "step": 6076
    },
    {
      "epoch": 1.9188506473002842,
      "grad_norm": 1.478989601135254,
      "learning_rate": 3.423619395130166e-05,
      "loss": 0.7802,
      "step": 6077
    },
    {
      "epoch": 1.9191664035364697,
      "grad_norm": 1.6461255550384521,
      "learning_rate": 3.423268542558417e-05,
      "loss": 0.72,
      "step": 6078
    },
    {
      "epoch": 1.9194821597726555,
      "grad_norm": 1.868127465248108,
      "learning_rate": 3.422917689986668e-05,
      "loss": 0.9475,
      "step": 6079
    },
    {
      "epoch": 1.9197979160088412,
      "grad_norm": 1.73765230178833,
      "learning_rate": 3.4225668374149183e-05,
      "loss": 0.8678,
      "step": 6080
    },
    {
      "epoch": 1.920113672245027,
      "grad_norm": 1.438746690750122,
      "learning_rate": 3.422215984843169e-05,
      "loss": 0.7746,
      "step": 6081
    },
    {
      "epoch": 1.9204294284812125,
      "grad_norm": 1.9830878973007202,
      "learning_rate": 3.42186513227142e-05,
      "loss": 0.862,
      "step": 6082
    },
    {
      "epoch": 1.9207451847173982,
      "grad_norm": 1.7625255584716797,
      "learning_rate": 3.4215142796996704e-05,
      "loss": 0.8338,
      "step": 6083
    },
    {
      "epoch": 1.9210609409535837,
      "grad_norm": 1.6853363513946533,
      "learning_rate": 3.421163427127921e-05,
      "loss": 0.8193,
      "step": 6084
    },
    {
      "epoch": 1.9213766971897694,
      "grad_norm": 1.5520851612091064,
      "learning_rate": 3.420812574556172e-05,
      "loss": 0.7799,
      "step": 6085
    },
    {
      "epoch": 1.9216924534259552,
      "grad_norm": 1.7553693056106567,
      "learning_rate": 3.420461721984422e-05,
      "loss": 0.8375,
      "step": 6086
    },
    {
      "epoch": 1.922008209662141,
      "grad_norm": 1.6750457286834717,
      "learning_rate": 3.420110869412673e-05,
      "loss": 0.8137,
      "step": 6087
    },
    {
      "epoch": 1.9223239658983264,
      "grad_norm": 1.5726293325424194,
      "learning_rate": 3.419760016840924e-05,
      "loss": 0.8154,
      "step": 6088
    },
    {
      "epoch": 1.9226397221345122,
      "grad_norm": 1.6752351522445679,
      "learning_rate": 3.4194091642691745e-05,
      "loss": 0.7762,
      "step": 6089
    },
    {
      "epoch": 1.9229554783706977,
      "grad_norm": 1.5520119667053223,
      "learning_rate": 3.419058311697425e-05,
      "loss": 0.7944,
      "step": 6090
    },
    {
      "epoch": 1.9232712346068834,
      "grad_norm": 1.549586534500122,
      "learning_rate": 3.418707459125676e-05,
      "loss": 0.8222,
      "step": 6091
    },
    {
      "epoch": 1.9235869908430692,
      "grad_norm": 1.6558783054351807,
      "learning_rate": 3.4183566065539265e-05,
      "loss": 0.8292,
      "step": 6092
    },
    {
      "epoch": 1.923902747079255,
      "grad_norm": 1.7367820739746094,
      "learning_rate": 3.418005753982177e-05,
      "loss": 0.8134,
      "step": 6093
    },
    {
      "epoch": 1.9242185033154406,
      "grad_norm": 1.995306134223938,
      "learning_rate": 3.417654901410427e-05,
      "loss": 0.7887,
      "step": 6094
    },
    {
      "epoch": 1.9245342595516262,
      "grad_norm": 2.0611982345581055,
      "learning_rate": 3.417304048838678e-05,
      "loss": 0.8391,
      "step": 6095
    },
    {
      "epoch": 1.9248500157878117,
      "grad_norm": 1.9915931224822998,
      "learning_rate": 3.4169531962669286e-05,
      "loss": 0.8352,
      "step": 6096
    },
    {
      "epoch": 1.9251657720239974,
      "grad_norm": 1.7397074699401855,
      "learning_rate": 3.416602343695179e-05,
      "loss": 0.74,
      "step": 6097
    },
    {
      "epoch": 1.9254815282601832,
      "grad_norm": 1.7097278833389282,
      "learning_rate": 3.41625149112343e-05,
      "loss": 0.8511,
      "step": 6098
    },
    {
      "epoch": 1.9257972844963689,
      "grad_norm": 1.648420810699463,
      "learning_rate": 3.4159006385516806e-05,
      "loss": 0.8389,
      "step": 6099
    },
    {
      "epoch": 1.9261130407325546,
      "grad_norm": 1.9523298740386963,
      "learning_rate": 3.415549785979931e-05,
      "loss": 0.8742,
      "step": 6100
    },
    {
      "epoch": 1.9264287969687401,
      "grad_norm": 1.4752002954483032,
      "learning_rate": 3.415198933408182e-05,
      "loss": 0.7163,
      "step": 6101
    },
    {
      "epoch": 1.9267445532049257,
      "grad_norm": 1.509879469871521,
      "learning_rate": 3.4148480808364327e-05,
      "loss": 0.8173,
      "step": 6102
    },
    {
      "epoch": 1.9270603094411114,
      "grad_norm": 1.8682430982589722,
      "learning_rate": 3.4144972282646833e-05,
      "loss": 0.9153,
      "step": 6103
    },
    {
      "epoch": 1.9273760656772971,
      "grad_norm": 1.5972297191619873,
      "learning_rate": 3.414146375692934e-05,
      "loss": 0.807,
      "step": 6104
    },
    {
      "epoch": 1.9276918219134829,
      "grad_norm": 1.8454513549804688,
      "learning_rate": 3.413795523121185e-05,
      "loss": 0.8375,
      "step": 6105
    },
    {
      "epoch": 1.9280075781496686,
      "grad_norm": 1.5727429389953613,
      "learning_rate": 3.4134446705494354e-05,
      "loss": 0.7224,
      "step": 6106
    },
    {
      "epoch": 1.9283233343858541,
      "grad_norm": 1.6331605911254883,
      "learning_rate": 3.413093817977686e-05,
      "loss": 0.8235,
      "step": 6107
    },
    {
      "epoch": 1.9286390906220396,
      "grad_norm": 1.6751604080200195,
      "learning_rate": 3.412742965405937e-05,
      "loss": 0.7359,
      "step": 6108
    },
    {
      "epoch": 1.9289548468582254,
      "grad_norm": 1.7058850526809692,
      "learning_rate": 3.4123921128341874e-05,
      "loss": 0.8255,
      "step": 6109
    },
    {
      "epoch": 1.9292706030944111,
      "grad_norm": 1.7451233863830566,
      "learning_rate": 3.412041260262438e-05,
      "loss": 0.8878,
      "step": 6110
    },
    {
      "epoch": 1.9295863593305969,
      "grad_norm": 1.7267180681228638,
      "learning_rate": 3.411690407690689e-05,
      "loss": 0.8038,
      "step": 6111
    },
    {
      "epoch": 1.9299021155667826,
      "grad_norm": 1.766503930091858,
      "learning_rate": 3.4113395551189395e-05,
      "loss": 0.8127,
      "step": 6112
    },
    {
      "epoch": 1.930217871802968,
      "grad_norm": 1.5206016302108765,
      "learning_rate": 3.41098870254719e-05,
      "loss": 0.8219,
      "step": 6113
    },
    {
      "epoch": 1.9305336280391536,
      "grad_norm": 1.989485263824463,
      "learning_rate": 3.41063784997544e-05,
      "loss": 0.8061,
      "step": 6114
    },
    {
      "epoch": 1.9308493842753394,
      "grad_norm": 2.2124698162078857,
      "learning_rate": 3.410286997403691e-05,
      "loss": 0.7327,
      "step": 6115
    },
    {
      "epoch": 1.931165140511525,
      "grad_norm": 2.0208988189697266,
      "learning_rate": 3.4099361448319415e-05,
      "loss": 0.8425,
      "step": 6116
    },
    {
      "epoch": 1.9314808967477108,
      "grad_norm": 1.7943804264068604,
      "learning_rate": 3.409585292260192e-05,
      "loss": 0.8079,
      "step": 6117
    },
    {
      "epoch": 1.9317966529838966,
      "grad_norm": 1.633555293083191,
      "learning_rate": 3.409234439688443e-05,
      "loss": 0.8503,
      "step": 6118
    },
    {
      "epoch": 1.932112409220082,
      "grad_norm": 1.4809001684188843,
      "learning_rate": 3.4088835871166936e-05,
      "loss": 0.7612,
      "step": 6119
    },
    {
      "epoch": 1.9324281654562676,
      "grad_norm": 1.7119665145874023,
      "learning_rate": 3.408532734544944e-05,
      "loss": 0.7883,
      "step": 6120
    },
    {
      "epoch": 1.9327439216924533,
      "grad_norm": 2.336754083633423,
      "learning_rate": 3.408181881973195e-05,
      "loss": 0.8214,
      "step": 6121
    },
    {
      "epoch": 1.933059677928639,
      "grad_norm": 1.5611931085586548,
      "learning_rate": 3.4078310294014456e-05,
      "loss": 0.7969,
      "step": 6122
    },
    {
      "epoch": 1.9333754341648248,
      "grad_norm": 1.4486987590789795,
      "learning_rate": 3.407480176829696e-05,
      "loss": 0.7466,
      "step": 6123
    },
    {
      "epoch": 1.9336911904010106,
      "grad_norm": 1.698880672454834,
      "learning_rate": 3.407129324257947e-05,
      "loss": 0.7814,
      "step": 6124
    },
    {
      "epoch": 1.934006946637196,
      "grad_norm": 1.8360825777053833,
      "learning_rate": 3.4067784716861977e-05,
      "loss": 0.7711,
      "step": 6125
    },
    {
      "epoch": 1.9343227028733816,
      "grad_norm": 1.813863754272461,
      "learning_rate": 3.406427619114448e-05,
      "loss": 0.807,
      "step": 6126
    },
    {
      "epoch": 1.9346384591095673,
      "grad_norm": 1.7066292762756348,
      "learning_rate": 3.406076766542699e-05,
      "loss": 0.7412,
      "step": 6127
    },
    {
      "epoch": 1.934954215345753,
      "grad_norm": 1.4609919786453247,
      "learning_rate": 3.40572591397095e-05,
      "loss": 0.7805,
      "step": 6128
    },
    {
      "epoch": 1.9352699715819388,
      "grad_norm": 1.5187745094299316,
      "learning_rate": 3.4053750613992004e-05,
      "loss": 0.7936,
      "step": 6129
    },
    {
      "epoch": 1.9355857278181245,
      "grad_norm": 1.4678035974502563,
      "learning_rate": 3.405024208827451e-05,
      "loss": 0.7766,
      "step": 6130
    },
    {
      "epoch": 1.93590148405431,
      "grad_norm": 1.5527604818344116,
      "learning_rate": 3.404673356255702e-05,
      "loss": 0.7595,
      "step": 6131
    },
    {
      "epoch": 1.9362172402904956,
      "grad_norm": 1.7235363721847534,
      "learning_rate": 3.4043225036839524e-05,
      "loss": 0.8457,
      "step": 6132
    },
    {
      "epoch": 1.9365329965266813,
      "grad_norm": 1.8999508619308472,
      "learning_rate": 3.4039716511122024e-05,
      "loss": 0.886,
      "step": 6133
    },
    {
      "epoch": 1.936848752762867,
      "grad_norm": 1.4137821197509766,
      "learning_rate": 3.403620798540453e-05,
      "loss": 0.7187,
      "step": 6134
    },
    {
      "epoch": 1.9371645089990528,
      "grad_norm": 1.642025113105774,
      "learning_rate": 3.403269945968704e-05,
      "loss": 0.7893,
      "step": 6135
    },
    {
      "epoch": 1.9374802652352385,
      "grad_norm": 1.6034938097000122,
      "learning_rate": 3.4029190933969545e-05,
      "loss": 0.7812,
      "step": 6136
    },
    {
      "epoch": 1.937796021471424,
      "grad_norm": 1.7245558500289917,
      "learning_rate": 3.402568240825205e-05,
      "loss": 0.854,
      "step": 6137
    },
    {
      "epoch": 1.9381117777076096,
      "grad_norm": 1.6643686294555664,
      "learning_rate": 3.402217388253456e-05,
      "loss": 0.7507,
      "step": 6138
    },
    {
      "epoch": 1.9384275339437953,
      "grad_norm": 4.51622200012207,
      "learning_rate": 3.4018665356817065e-05,
      "loss": 0.7434,
      "step": 6139
    },
    {
      "epoch": 1.938743290179981,
      "grad_norm": 1.6777355670928955,
      "learning_rate": 3.401515683109958e-05,
      "loss": 0.8285,
      "step": 6140
    },
    {
      "epoch": 1.9390590464161668,
      "grad_norm": 1.7170981168746948,
      "learning_rate": 3.4011648305382086e-05,
      "loss": 0.8294,
      "step": 6141
    },
    {
      "epoch": 1.9393748026523525,
      "grad_norm": 1.848675012588501,
      "learning_rate": 3.4008139779664586e-05,
      "loss": 0.7709,
      "step": 6142
    },
    {
      "epoch": 1.939690558888538,
      "grad_norm": 1.8355568647384644,
      "learning_rate": 3.400463125394709e-05,
      "loss": 0.8421,
      "step": 6143
    },
    {
      "epoch": 1.9400063151247238,
      "grad_norm": 1.8701810836791992,
      "learning_rate": 3.40011227282296e-05,
      "loss": 0.8209,
      "step": 6144
    },
    {
      "epoch": 1.9403220713609093,
      "grad_norm": 1.8664400577545166,
      "learning_rate": 3.3997614202512106e-05,
      "loss": 0.7833,
      "step": 6145
    },
    {
      "epoch": 1.940637827597095,
      "grad_norm": 1.8227351903915405,
      "learning_rate": 3.399410567679461e-05,
      "loss": 0.8175,
      "step": 6146
    },
    {
      "epoch": 1.9409535838332808,
      "grad_norm": 1.5938326120376587,
      "learning_rate": 3.399059715107712e-05,
      "loss": 0.8,
      "step": 6147
    },
    {
      "epoch": 1.9412693400694665,
      "grad_norm": 1.6872050762176514,
      "learning_rate": 3.3987088625359626e-05,
      "loss": 0.7862,
      "step": 6148
    },
    {
      "epoch": 1.941585096305652,
      "grad_norm": 1.6337729692459106,
      "learning_rate": 3.398358009964213e-05,
      "loss": 0.8415,
      "step": 6149
    },
    {
      "epoch": 1.9419008525418378,
      "grad_norm": 1.556384563446045,
      "learning_rate": 3.398007157392464e-05,
      "loss": 0.7697,
      "step": 6150
    },
    {
      "epoch": 1.9422166087780233,
      "grad_norm": 1.5426944494247437,
      "learning_rate": 3.397656304820715e-05,
      "loss": 0.7549,
      "step": 6151
    },
    {
      "epoch": 1.942532365014209,
      "grad_norm": 1.5192077159881592,
      "learning_rate": 3.397305452248965e-05,
      "loss": 0.8588,
      "step": 6152
    },
    {
      "epoch": 1.9428481212503947,
      "grad_norm": 1.6471771001815796,
      "learning_rate": 3.3969545996772154e-05,
      "loss": 0.8293,
      "step": 6153
    },
    {
      "epoch": 1.9431638774865805,
      "grad_norm": 1.514683723449707,
      "learning_rate": 3.396603747105466e-05,
      "loss": 0.7773,
      "step": 6154
    },
    {
      "epoch": 1.943479633722766,
      "grad_norm": 1.651056170463562,
      "learning_rate": 3.396252894533717e-05,
      "loss": 0.8381,
      "step": 6155
    },
    {
      "epoch": 1.9437953899589517,
      "grad_norm": 2.173837900161743,
      "learning_rate": 3.3959020419619674e-05,
      "loss": 0.8587,
      "step": 6156
    },
    {
      "epoch": 1.9441111461951373,
      "grad_norm": 1.8896266222000122,
      "learning_rate": 3.395551189390219e-05,
      "loss": 0.8545,
      "step": 6157
    },
    {
      "epoch": 1.944426902431323,
      "grad_norm": 1.9384928941726685,
      "learning_rate": 3.3952003368184695e-05,
      "loss": 0.8153,
      "step": 6158
    },
    {
      "epoch": 1.9447426586675087,
      "grad_norm": 1.6518454551696777,
      "learning_rate": 3.39484948424672e-05,
      "loss": 0.9155,
      "step": 6159
    },
    {
      "epoch": 1.9450584149036945,
      "grad_norm": 1.41838800907135,
      "learning_rate": 3.394498631674971e-05,
      "loss": 0.7615,
      "step": 6160
    },
    {
      "epoch": 1.94537417113988,
      "grad_norm": 1.6718074083328247,
      "learning_rate": 3.394147779103221e-05,
      "loss": 0.8312,
      "step": 6161
    },
    {
      "epoch": 1.9456899273760657,
      "grad_norm": 1.5072894096374512,
      "learning_rate": 3.3937969265314715e-05,
      "loss": 0.7973,
      "step": 6162
    },
    {
      "epoch": 1.9460056836122512,
      "grad_norm": 1.744485855102539,
      "learning_rate": 3.393446073959722e-05,
      "loss": 0.7684,
      "step": 6163
    },
    {
      "epoch": 1.946321439848437,
      "grad_norm": 1.475874662399292,
      "learning_rate": 3.393095221387973e-05,
      "loss": 0.7872,
      "step": 6164
    },
    {
      "epoch": 1.9466371960846227,
      "grad_norm": 1.8901283740997314,
      "learning_rate": 3.3927443688162235e-05,
      "loss": 0.8497,
      "step": 6165
    },
    {
      "epoch": 1.9469529523208084,
      "grad_norm": 1.4627752304077148,
      "learning_rate": 3.392393516244474e-05,
      "loss": 0.8226,
      "step": 6166
    },
    {
      "epoch": 1.947268708556994,
      "grad_norm": 1.5732786655426025,
      "learning_rate": 3.392042663672725e-05,
      "loss": 0.7796,
      "step": 6167
    },
    {
      "epoch": 1.9475844647931797,
      "grad_norm": 1.6568505764007568,
      "learning_rate": 3.3916918111009756e-05,
      "loss": 0.784,
      "step": 6168
    },
    {
      "epoch": 1.9479002210293652,
      "grad_norm": 1.6477205753326416,
      "learning_rate": 3.391340958529226e-05,
      "loss": 0.8656,
      "step": 6169
    },
    {
      "epoch": 1.948215977265551,
      "grad_norm": 1.678008794784546,
      "learning_rate": 3.390990105957477e-05,
      "loss": 0.8287,
      "step": 6170
    },
    {
      "epoch": 1.9485317335017367,
      "grad_norm": 1.5315500497817993,
      "learning_rate": 3.3906392533857276e-05,
      "loss": 0.805,
      "step": 6171
    },
    {
      "epoch": 1.9488474897379224,
      "grad_norm": 1.654765248298645,
      "learning_rate": 3.3902884008139776e-05,
      "loss": 0.8739,
      "step": 6172
    },
    {
      "epoch": 1.949163245974108,
      "grad_norm": 1.6595596075057983,
      "learning_rate": 3.389937548242228e-05,
      "loss": 0.7405,
      "step": 6173
    },
    {
      "epoch": 1.9494790022102937,
      "grad_norm": 1.5826159715652466,
      "learning_rate": 3.389586695670479e-05,
      "loss": 0.9589,
      "step": 6174
    },
    {
      "epoch": 1.9497947584464792,
      "grad_norm": 1.8422529697418213,
      "learning_rate": 3.3892358430987304e-05,
      "loss": 0.8476,
      "step": 6175
    },
    {
      "epoch": 1.950110514682665,
      "grad_norm": 1.9669815301895142,
      "learning_rate": 3.388884990526981e-05,
      "loss": 0.9064,
      "step": 6176
    },
    {
      "epoch": 1.9504262709188507,
      "grad_norm": 1.603087067604065,
      "learning_rate": 3.388534137955232e-05,
      "loss": 0.8227,
      "step": 6177
    },
    {
      "epoch": 1.9507420271550364,
      "grad_norm": 1.8123682737350464,
      "learning_rate": 3.3881832853834824e-05,
      "loss": 0.7852,
      "step": 6178
    },
    {
      "epoch": 1.951057783391222,
      "grad_norm": 1.4978781938552856,
      "learning_rate": 3.387832432811733e-05,
      "loss": 0.8423,
      "step": 6179
    },
    {
      "epoch": 1.9513735396274077,
      "grad_norm": 1.5861334800720215,
      "learning_rate": 3.387481580239983e-05,
      "loss": 0.7857,
      "step": 6180
    },
    {
      "epoch": 1.9516892958635932,
      "grad_norm": 1.9406757354736328,
      "learning_rate": 3.387130727668234e-05,
      "loss": 0.8475,
      "step": 6181
    },
    {
      "epoch": 1.952005052099779,
      "grad_norm": 1.4665088653564453,
      "learning_rate": 3.3867798750964845e-05,
      "loss": 0.9069,
      "step": 6182
    },
    {
      "epoch": 1.9523208083359647,
      "grad_norm": 1.3597332239151,
      "learning_rate": 3.386429022524735e-05,
      "loss": 0.7846,
      "step": 6183
    },
    {
      "epoch": 1.9526365645721504,
      "grad_norm": 1.521143913269043,
      "learning_rate": 3.386078169952986e-05,
      "loss": 0.804,
      "step": 6184
    },
    {
      "epoch": 1.952952320808336,
      "grad_norm": 2.040517807006836,
      "learning_rate": 3.3857273173812365e-05,
      "loss": 0.764,
      "step": 6185
    },
    {
      "epoch": 1.9532680770445217,
      "grad_norm": 1.7236990928649902,
      "learning_rate": 3.385376464809487e-05,
      "loss": 0.7861,
      "step": 6186
    },
    {
      "epoch": 1.9535838332807072,
      "grad_norm": 1.6765058040618896,
      "learning_rate": 3.385025612237738e-05,
      "loss": 0.7713,
      "step": 6187
    },
    {
      "epoch": 1.953899589516893,
      "grad_norm": 1.3860074281692505,
      "learning_rate": 3.3846747596659885e-05,
      "loss": 0.8507,
      "step": 6188
    },
    {
      "epoch": 1.9542153457530786,
      "grad_norm": 1.723057746887207,
      "learning_rate": 3.384323907094239e-05,
      "loss": 0.836,
      "step": 6189
    },
    {
      "epoch": 1.9545311019892644,
      "grad_norm": 1.4935861825942993,
      "learning_rate": 3.38397305452249e-05,
      "loss": 0.784,
      "step": 6190
    },
    {
      "epoch": 1.95484685822545,
      "grad_norm": 1.4397743940353394,
      "learning_rate": 3.38362220195074e-05,
      "loss": 0.7885,
      "step": 6191
    },
    {
      "epoch": 1.9551626144616356,
      "grad_norm": 1.5517668724060059,
      "learning_rate": 3.383271349378991e-05,
      "loss": 0.8655,
      "step": 6192
    },
    {
      "epoch": 1.9554783706978212,
      "grad_norm": 1.2996960878372192,
      "learning_rate": 3.382920496807242e-05,
      "loss": 0.7884,
      "step": 6193
    },
    {
      "epoch": 1.955794126934007,
      "grad_norm": 1.4839069843292236,
      "learning_rate": 3.3825696442354926e-05,
      "loss": 0.8124,
      "step": 6194
    },
    {
      "epoch": 1.9561098831701926,
      "grad_norm": 1.5454908609390259,
      "learning_rate": 3.382218791663743e-05,
      "loss": 0.7836,
      "step": 6195
    },
    {
      "epoch": 1.9564256394063784,
      "grad_norm": 1.6608678102493286,
      "learning_rate": 3.381867939091994e-05,
      "loss": 0.8328,
      "step": 6196
    },
    {
      "epoch": 1.9567413956425639,
      "grad_norm": 1.499741792678833,
      "learning_rate": 3.381517086520245e-05,
      "loss": 0.8202,
      "step": 6197
    },
    {
      "epoch": 1.9570571518787496,
      "grad_norm": 1.5932340621948242,
      "learning_rate": 3.3811662339484954e-05,
      "loss": 0.8295,
      "step": 6198
    },
    {
      "epoch": 1.9573729081149351,
      "grad_norm": 1.4120149612426758,
      "learning_rate": 3.3808153813767454e-05,
      "loss": 0.702,
      "step": 6199
    },
    {
      "epoch": 1.9576886643511209,
      "grad_norm": 1.3958516120910645,
      "learning_rate": 3.380464528804996e-05,
      "loss": 0.859,
      "step": 6200
    },
    {
      "epoch": 1.9576886643511209,
      "eval_loss": 0.7943219542503357,
      "eval_runtime": 4.6372,
      "eval_samples_per_second": 536.968,
      "eval_steps_per_second": 4.529,
      "step": 6200
    },
    {
      "epoch": 1.9580044205873066,
      "grad_norm": 1.6435725688934326,
      "learning_rate": 3.380113676233247e-05,
      "loss": 0.7711,
      "step": 6201
    },
    {
      "epoch": 1.9583201768234924,
      "grad_norm": 1.533800482749939,
      "learning_rate": 3.3797628236614974e-05,
      "loss": 0.8539,
      "step": 6202
    },
    {
      "epoch": 1.958635933059678,
      "grad_norm": 1.4997267723083496,
      "learning_rate": 3.379411971089748e-05,
      "loss": 0.8882,
      "step": 6203
    },
    {
      "epoch": 1.9589516892958636,
      "grad_norm": 1.5947785377502441,
      "learning_rate": 3.379061118517999e-05,
      "loss": 0.8503,
      "step": 6204
    },
    {
      "epoch": 1.9592674455320491,
      "grad_norm": 1.8210787773132324,
      "learning_rate": 3.3787102659462494e-05,
      "loss": 0.8085,
      "step": 6205
    },
    {
      "epoch": 1.9595832017682349,
      "grad_norm": 1.481123685836792,
      "learning_rate": 3.3783594133745e-05,
      "loss": 0.7225,
      "step": 6206
    },
    {
      "epoch": 1.9598989580044206,
      "grad_norm": 1.3731184005737305,
      "learning_rate": 3.378008560802751e-05,
      "loss": 0.8451,
      "step": 6207
    },
    {
      "epoch": 1.9602147142406063,
      "grad_norm": 1.9007949829101562,
      "learning_rate": 3.3776577082310015e-05,
      "loss": 0.8689,
      "step": 6208
    },
    {
      "epoch": 1.960530470476792,
      "grad_norm": 1.569210171699524,
      "learning_rate": 3.377306855659252e-05,
      "loss": 0.8268,
      "step": 6209
    },
    {
      "epoch": 1.9608462267129776,
      "grad_norm": 1.5506314039230347,
      "learning_rate": 3.376956003087503e-05,
      "loss": 0.7786,
      "step": 6210
    },
    {
      "epoch": 1.961161982949163,
      "grad_norm": 1.5429105758666992,
      "learning_rate": 3.3766051505157535e-05,
      "loss": 0.7576,
      "step": 6211
    },
    {
      "epoch": 1.9614777391853488,
      "grad_norm": 2.2994437217712402,
      "learning_rate": 3.376254297944004e-05,
      "loss": 0.8516,
      "step": 6212
    },
    {
      "epoch": 1.9617934954215346,
      "grad_norm": 1.7567378282546997,
      "learning_rate": 3.375903445372255e-05,
      "loss": 0.8234,
      "step": 6213
    },
    {
      "epoch": 1.9621092516577203,
      "grad_norm": 1.7361055612564087,
      "learning_rate": 3.3755525928005056e-05,
      "loss": 0.9248,
      "step": 6214
    },
    {
      "epoch": 1.962425007893906,
      "grad_norm": 1.6721558570861816,
      "learning_rate": 3.375201740228756e-05,
      "loss": 0.8278,
      "step": 6215
    },
    {
      "epoch": 1.9627407641300916,
      "grad_norm": 1.4202831983566284,
      "learning_rate": 3.374850887657007e-05,
      "loss": 0.8006,
      "step": 6216
    },
    {
      "epoch": 1.963056520366277,
      "grad_norm": 1.5935773849487305,
      "learning_rate": 3.3745000350852576e-05,
      "loss": 0.7688,
      "step": 6217
    },
    {
      "epoch": 1.9633722766024628,
      "grad_norm": 1.6953234672546387,
      "learning_rate": 3.374149182513508e-05,
      "loss": 0.8609,
      "step": 6218
    },
    {
      "epoch": 1.9636880328386486,
      "grad_norm": 1.7165930271148682,
      "learning_rate": 3.373798329941758e-05,
      "loss": 0.894,
      "step": 6219
    },
    {
      "epoch": 1.9640037890748343,
      "grad_norm": 1.5531716346740723,
      "learning_rate": 3.373447477370009e-05,
      "loss": 0.8159,
      "step": 6220
    },
    {
      "epoch": 1.96431954531102,
      "grad_norm": 1.7121466398239136,
      "learning_rate": 3.37309662479826e-05,
      "loss": 0.8731,
      "step": 6221
    },
    {
      "epoch": 1.9646353015472056,
      "grad_norm": 1.465232491493225,
      "learning_rate": 3.3727457722265103e-05,
      "loss": 0.7688,
      "step": 6222
    },
    {
      "epoch": 1.964951057783391,
      "grad_norm": 1.4799988269805908,
      "learning_rate": 3.372394919654761e-05,
      "loss": 0.8137,
      "step": 6223
    },
    {
      "epoch": 1.9652668140195768,
      "grad_norm": 1.6142183542251587,
      "learning_rate": 3.372044067083012e-05,
      "loss": 0.8295,
      "step": 6224
    },
    {
      "epoch": 1.9655825702557626,
      "grad_norm": 1.4556670188903809,
      "learning_rate": 3.3716932145112624e-05,
      "loss": 0.8083,
      "step": 6225
    },
    {
      "epoch": 1.9658983264919483,
      "grad_norm": 1.7564135789871216,
      "learning_rate": 3.371342361939513e-05,
      "loss": 0.8095,
      "step": 6226
    },
    {
      "epoch": 1.966214082728134,
      "grad_norm": 1.8511929512023926,
      "learning_rate": 3.370991509367764e-05,
      "loss": 0.818,
      "step": 6227
    },
    {
      "epoch": 1.9665298389643195,
      "grad_norm": 2.1073665618896484,
      "learning_rate": 3.3706406567960144e-05,
      "loss": 0.7953,
      "step": 6228
    },
    {
      "epoch": 1.966845595200505,
      "grad_norm": 1.6273795366287231,
      "learning_rate": 3.370289804224265e-05,
      "loss": 0.8129,
      "step": 6229
    },
    {
      "epoch": 1.9671613514366908,
      "grad_norm": 1.8610087633132935,
      "learning_rate": 3.369938951652516e-05,
      "loss": 0.8889,
      "step": 6230
    },
    {
      "epoch": 1.9674771076728765,
      "grad_norm": 1.3990851640701294,
      "learning_rate": 3.3695880990807665e-05,
      "loss": 0.8366,
      "step": 6231
    },
    {
      "epoch": 1.9677928639090623,
      "grad_norm": 1.528499722480774,
      "learning_rate": 3.369237246509017e-05,
      "loss": 0.7793,
      "step": 6232
    },
    {
      "epoch": 1.968108620145248,
      "grad_norm": 1.6284334659576416,
      "learning_rate": 3.368886393937268e-05,
      "loss": 0.8662,
      "step": 6233
    },
    {
      "epoch": 1.9684243763814335,
      "grad_norm": 1.8746371269226074,
      "learning_rate": 3.3685355413655185e-05,
      "loss": 0.911,
      "step": 6234
    },
    {
      "epoch": 1.968740132617619,
      "grad_norm": 1.53440260887146,
      "learning_rate": 3.368184688793769e-05,
      "loss": 0.8139,
      "step": 6235
    },
    {
      "epoch": 1.9690558888538048,
      "grad_norm": 1.4080958366394043,
      "learning_rate": 3.36783383622202e-05,
      "loss": 0.74,
      "step": 6236
    },
    {
      "epoch": 1.9693716450899905,
      "grad_norm": 1.6428523063659668,
      "learning_rate": 3.3674829836502706e-05,
      "loss": 0.8381,
      "step": 6237
    },
    {
      "epoch": 1.9696874013261763,
      "grad_norm": 1.4250004291534424,
      "learning_rate": 3.3671321310785206e-05,
      "loss": 0.7711,
      "step": 6238
    },
    {
      "epoch": 1.970003157562362,
      "grad_norm": 1.7756762504577637,
      "learning_rate": 3.366781278506771e-05,
      "loss": 0.8887,
      "step": 6239
    },
    {
      "epoch": 1.9703189137985475,
      "grad_norm": 1.6838575601577759,
      "learning_rate": 3.366430425935022e-05,
      "loss": 0.7815,
      "step": 6240
    },
    {
      "epoch": 1.970634670034733,
      "grad_norm": 1.6821503639221191,
      "learning_rate": 3.3660795733632726e-05,
      "loss": 0.8808,
      "step": 6241
    },
    {
      "epoch": 1.9709504262709188,
      "grad_norm": 2.0150651931762695,
      "learning_rate": 3.365728720791523e-05,
      "loss": 0.8295,
      "step": 6242
    },
    {
      "epoch": 1.9712661825071045,
      "grad_norm": 1.4235252141952515,
      "learning_rate": 3.365377868219774e-05,
      "loss": 0.7678,
      "step": 6243
    },
    {
      "epoch": 1.9715819387432902,
      "grad_norm": 1.5674543380737305,
      "learning_rate": 3.365027015648025e-05,
      "loss": 0.872,
      "step": 6244
    },
    {
      "epoch": 1.971897694979476,
      "grad_norm": 1.3686305284500122,
      "learning_rate": 3.364676163076276e-05,
      "loss": 0.7109,
      "step": 6245
    },
    {
      "epoch": 1.9722134512156615,
      "grad_norm": 1.65183424949646,
      "learning_rate": 3.364325310504527e-05,
      "loss": 0.8932,
      "step": 6246
    },
    {
      "epoch": 1.972529207451847,
      "grad_norm": 1.7164069414138794,
      "learning_rate": 3.363974457932777e-05,
      "loss": 0.7763,
      "step": 6247
    },
    {
      "epoch": 1.9728449636880327,
      "grad_norm": 1.5926487445831299,
      "learning_rate": 3.3636236053610274e-05,
      "loss": 0.825,
      "step": 6248
    },
    {
      "epoch": 1.9731607199242185,
      "grad_norm": 1.6028212308883667,
      "learning_rate": 3.363272752789278e-05,
      "loss": 0.8518,
      "step": 6249
    },
    {
      "epoch": 1.9734764761604042,
      "grad_norm": 1.74600088596344,
      "learning_rate": 3.362921900217529e-05,
      "loss": 0.8966,
      "step": 6250
    },
    {
      "epoch": 1.97379223239659,
      "grad_norm": 1.3179709911346436,
      "learning_rate": 3.3625710476457794e-05,
      "loss": 0.749,
      "step": 6251
    },
    {
      "epoch": 1.9741079886327755,
      "grad_norm": 1.4125717878341675,
      "learning_rate": 3.36222019507403e-05,
      "loss": 0.8169,
      "step": 6252
    },
    {
      "epoch": 1.9744237448689612,
      "grad_norm": 1.871294379234314,
      "learning_rate": 3.361869342502281e-05,
      "loss": 0.8595,
      "step": 6253
    },
    {
      "epoch": 1.9747395011051467,
      "grad_norm": 1.6700377464294434,
      "learning_rate": 3.3615184899305315e-05,
      "loss": 0.7998,
      "step": 6254
    },
    {
      "epoch": 1.9750552573413325,
      "grad_norm": 1.551812767982483,
      "learning_rate": 3.361167637358782e-05,
      "loss": 0.7847,
      "step": 6255
    },
    {
      "epoch": 1.9753710135775182,
      "grad_norm": 1.5368926525115967,
      "learning_rate": 3.360816784787033e-05,
      "loss": 0.7881,
      "step": 6256
    },
    {
      "epoch": 1.975686769813704,
      "grad_norm": 1.7791368961334229,
      "learning_rate": 3.360465932215283e-05,
      "loss": 0.8061,
      "step": 6257
    },
    {
      "epoch": 1.9760025260498895,
      "grad_norm": 1.6058894395828247,
      "learning_rate": 3.3601150796435335e-05,
      "loss": 0.7802,
      "step": 6258
    },
    {
      "epoch": 1.9763182822860752,
      "grad_norm": 1.7255967855453491,
      "learning_rate": 3.359764227071784e-05,
      "loss": 0.9258,
      "step": 6259
    },
    {
      "epoch": 1.9766340385222607,
      "grad_norm": 1.670345425605774,
      "learning_rate": 3.359413374500035e-05,
      "loss": 0.8222,
      "step": 6260
    },
    {
      "epoch": 1.9769497947584465,
      "grad_norm": 1.6544480323791504,
      "learning_rate": 3.3590625219282856e-05,
      "loss": 0.8253,
      "step": 6261
    },
    {
      "epoch": 1.9772655509946322,
      "grad_norm": 1.6839596033096313,
      "learning_rate": 3.358711669356537e-05,
      "loss": 0.8435,
      "step": 6262
    },
    {
      "epoch": 1.977581307230818,
      "grad_norm": 1.8022363185882568,
      "learning_rate": 3.3583608167847876e-05,
      "loss": 0.8179,
      "step": 6263
    },
    {
      "epoch": 1.9778970634670034,
      "grad_norm": 1.8077480792999268,
      "learning_rate": 3.358009964213038e-05,
      "loss": 0.8368,
      "step": 6264
    },
    {
      "epoch": 1.9782128197031892,
      "grad_norm": 1.5543925762176514,
      "learning_rate": 3.357659111641289e-05,
      "loss": 0.6701,
      "step": 6265
    },
    {
      "epoch": 1.9785285759393747,
      "grad_norm": 1.4966018199920654,
      "learning_rate": 3.357308259069539e-05,
      "loss": 0.8197,
      "step": 6266
    },
    {
      "epoch": 1.9788443321755604,
      "grad_norm": 1.8260619640350342,
      "learning_rate": 3.3569574064977897e-05,
      "loss": 0.8449,
      "step": 6267
    },
    {
      "epoch": 1.9791600884117462,
      "grad_norm": 1.623049020767212,
      "learning_rate": 3.35660655392604e-05,
      "loss": 0.8497,
      "step": 6268
    },
    {
      "epoch": 1.979475844647932,
      "grad_norm": 1.6246464252471924,
      "learning_rate": 3.356255701354291e-05,
      "loss": 0.8619,
      "step": 6269
    },
    {
      "epoch": 1.9797916008841174,
      "grad_norm": 1.683592677116394,
      "learning_rate": 3.355904848782542e-05,
      "loss": 0.8289,
      "step": 6270
    },
    {
      "epoch": 1.9801073571203032,
      "grad_norm": 1.797419786453247,
      "learning_rate": 3.3555539962107924e-05,
      "loss": 0.781,
      "step": 6271
    },
    {
      "epoch": 1.9804231133564887,
      "grad_norm": 1.5305997133255005,
      "learning_rate": 3.355203143639043e-05,
      "loss": 0.7579,
      "step": 6272
    },
    {
      "epoch": 1.9807388695926744,
      "grad_norm": 1.5542017221450806,
      "learning_rate": 3.354852291067294e-05,
      "loss": 0.7015,
      "step": 6273
    },
    {
      "epoch": 1.9810546258288602,
      "grad_norm": 1.8380526304244995,
      "learning_rate": 3.3545014384955444e-05,
      "loss": 0.7601,
      "step": 6274
    },
    {
      "epoch": 1.981370382065046,
      "grad_norm": 2.2433202266693115,
      "learning_rate": 3.354150585923795e-05,
      "loss": 0.8352,
      "step": 6275
    },
    {
      "epoch": 1.9816861383012314,
      "grad_norm": 1.6627756357192993,
      "learning_rate": 3.353799733352046e-05,
      "loss": 0.7511,
      "step": 6276
    },
    {
      "epoch": 1.9820018945374172,
      "grad_norm": 1.5738400220870972,
      "learning_rate": 3.353448880780296e-05,
      "loss": 0.8048,
      "step": 6277
    },
    {
      "epoch": 1.9823176507736027,
      "grad_norm": 2.1512889862060547,
      "learning_rate": 3.3530980282085465e-05,
      "loss": 0.9225,
      "step": 6278
    },
    {
      "epoch": 1.9826334070097884,
      "grad_norm": 1.7776243686676025,
      "learning_rate": 3.352747175636798e-05,
      "loss": 0.8628,
      "step": 6279
    },
    {
      "epoch": 1.9829491632459741,
      "grad_norm": 1.7143806219100952,
      "learning_rate": 3.3523963230650485e-05,
      "loss": 0.7762,
      "step": 6280
    },
    {
      "epoch": 1.9832649194821599,
      "grad_norm": 1.664047360420227,
      "learning_rate": 3.352045470493299e-05,
      "loss": 0.8069,
      "step": 6281
    },
    {
      "epoch": 1.9835806757183454,
      "grad_norm": 1.6461913585662842,
      "learning_rate": 3.35169461792155e-05,
      "loss": 0.8559,
      "step": 6282
    },
    {
      "epoch": 1.9838964319545311,
      "grad_norm": 1.8188468217849731,
      "learning_rate": 3.3513437653498006e-05,
      "loss": 0.8515,
      "step": 6283
    },
    {
      "epoch": 1.9842121881907167,
      "grad_norm": 1.4870647192001343,
      "learning_rate": 3.350992912778051e-05,
      "loss": 0.7674,
      "step": 6284
    },
    {
      "epoch": 1.9845279444269024,
      "grad_norm": 1.719865322113037,
      "learning_rate": 3.350642060206301e-05,
      "loss": 0.8445,
      "step": 6285
    },
    {
      "epoch": 1.9848437006630881,
      "grad_norm": 1.7451304197311401,
      "learning_rate": 3.350291207634552e-05,
      "loss": 0.7852,
      "step": 6286
    },
    {
      "epoch": 1.9851594568992739,
      "grad_norm": 1.8049710988998413,
      "learning_rate": 3.3499403550628026e-05,
      "loss": 0.7955,
      "step": 6287
    },
    {
      "epoch": 1.9854752131354594,
      "grad_norm": 1.6886123418807983,
      "learning_rate": 3.349589502491053e-05,
      "loss": 0.7983,
      "step": 6288
    },
    {
      "epoch": 1.9857909693716451,
      "grad_norm": 1.6011416912078857,
      "learning_rate": 3.349238649919304e-05,
      "loss": 0.7226,
      "step": 6289
    },
    {
      "epoch": 1.9861067256078306,
      "grad_norm": 1.6888787746429443,
      "learning_rate": 3.3488877973475546e-05,
      "loss": 0.8146,
      "step": 6290
    },
    {
      "epoch": 1.9864224818440164,
      "grad_norm": 1.834202766418457,
      "learning_rate": 3.348536944775805e-05,
      "loss": 0.8326,
      "step": 6291
    },
    {
      "epoch": 1.9867382380802021,
      "grad_norm": 1.62791907787323,
      "learning_rate": 3.348186092204056e-05,
      "loss": 0.7055,
      "step": 6292
    },
    {
      "epoch": 1.9870539943163879,
      "grad_norm": 1.7137739658355713,
      "learning_rate": 3.347835239632307e-05,
      "loss": 0.8275,
      "step": 6293
    },
    {
      "epoch": 1.9873697505525734,
      "grad_norm": 1.8038707971572876,
      "learning_rate": 3.3474843870605574e-05,
      "loss": 0.7917,
      "step": 6294
    },
    {
      "epoch": 1.987685506788759,
      "grad_norm": 1.809990644454956,
      "learning_rate": 3.347133534488808e-05,
      "loss": 0.7832,
      "step": 6295
    },
    {
      "epoch": 1.9880012630249446,
      "grad_norm": 1.633845567703247,
      "learning_rate": 3.346782681917058e-05,
      "loss": 0.9004,
      "step": 6296
    },
    {
      "epoch": 1.9883170192611304,
      "grad_norm": 1.6250258684158325,
      "learning_rate": 3.3464318293453094e-05,
      "loss": 0.8747,
      "step": 6297
    },
    {
      "epoch": 1.988632775497316,
      "grad_norm": 1.8823883533477783,
      "learning_rate": 3.34608097677356e-05,
      "loss": 0.7931,
      "step": 6298
    },
    {
      "epoch": 1.9889485317335018,
      "grad_norm": 1.623695731163025,
      "learning_rate": 3.345730124201811e-05,
      "loss": 0.8351,
      "step": 6299
    },
    {
      "epoch": 1.9892642879696874,
      "grad_norm": 1.6660754680633545,
      "learning_rate": 3.3453792716300615e-05,
      "loss": 0.7715,
      "step": 6300
    },
    {
      "epoch": 1.989580044205873,
      "grad_norm": 1.5366994142532349,
      "learning_rate": 3.345028419058312e-05,
      "loss": 0.8153,
      "step": 6301
    },
    {
      "epoch": 1.9898958004420586,
      "grad_norm": 2.043386220932007,
      "learning_rate": 3.344677566486563e-05,
      "loss": 0.8674,
      "step": 6302
    },
    {
      "epoch": 1.9902115566782443,
      "grad_norm": 1.5437418222427368,
      "learning_rate": 3.3443267139148135e-05,
      "loss": 0.7959,
      "step": 6303
    },
    {
      "epoch": 1.99052731291443,
      "grad_norm": 1.9451091289520264,
      "learning_rate": 3.3439758613430635e-05,
      "loss": 0.8777,
      "step": 6304
    },
    {
      "epoch": 1.9908430691506158,
      "grad_norm": 1.5307797193527222,
      "learning_rate": 3.343625008771314e-05,
      "loss": 0.778,
      "step": 6305
    },
    {
      "epoch": 1.9911588253868013,
      "grad_norm": 1.6304341554641724,
      "learning_rate": 3.343274156199565e-05,
      "loss": 0.8142,
      "step": 6306
    },
    {
      "epoch": 1.991474581622987,
      "grad_norm": 1.4863299131393433,
      "learning_rate": 3.3429233036278155e-05,
      "loss": 0.8975,
      "step": 6307
    },
    {
      "epoch": 1.9917903378591726,
      "grad_norm": 1.6054565906524658,
      "learning_rate": 3.342572451056066e-05,
      "loss": 0.7905,
      "step": 6308
    },
    {
      "epoch": 1.9921060940953583,
      "grad_norm": 1.5341846942901611,
      "learning_rate": 3.342221598484317e-05,
      "loss": 0.8578,
      "step": 6309
    },
    {
      "epoch": 1.992421850331544,
      "grad_norm": 1.7488003969192505,
      "learning_rate": 3.3418707459125676e-05,
      "loss": 0.8495,
      "step": 6310
    },
    {
      "epoch": 1.9927376065677298,
      "grad_norm": 1.475197672843933,
      "learning_rate": 3.341519893340818e-05,
      "loss": 0.8126,
      "step": 6311
    },
    {
      "epoch": 1.9930533628039155,
      "grad_norm": 1.6037076711654663,
      "learning_rate": 3.341169040769069e-05,
      "loss": 0.9195,
      "step": 6312
    },
    {
      "epoch": 1.993369119040101,
      "grad_norm": 1.5877832174301147,
      "learning_rate": 3.3408181881973196e-05,
      "loss": 0.7762,
      "step": 6313
    },
    {
      "epoch": 1.9936848752762866,
      "grad_norm": 1.7128181457519531,
      "learning_rate": 3.34046733562557e-05,
      "loss": 0.8411,
      "step": 6314
    },
    {
      "epoch": 1.9940006315124723,
      "grad_norm": 1.4810059070587158,
      "learning_rate": 3.340116483053821e-05,
      "loss": 0.7668,
      "step": 6315
    },
    {
      "epoch": 1.994316387748658,
      "grad_norm": 1.596949815750122,
      "learning_rate": 3.339765630482072e-05,
      "loss": 0.7907,
      "step": 6316
    },
    {
      "epoch": 1.9946321439848438,
      "grad_norm": 1.6096034049987793,
      "learning_rate": 3.3394147779103224e-05,
      "loss": 0.8888,
      "step": 6317
    },
    {
      "epoch": 1.9949479002210295,
      "grad_norm": 1.70423424243927,
      "learning_rate": 3.339063925338573e-05,
      "loss": 0.8535,
      "step": 6318
    },
    {
      "epoch": 1.995263656457215,
      "grad_norm": 1.839213252067566,
      "learning_rate": 3.338713072766824e-05,
      "loss": 0.8218,
      "step": 6319
    },
    {
      "epoch": 1.9955794126934006,
      "grad_norm": 1.6469316482543945,
      "learning_rate": 3.3383622201950744e-05,
      "loss": 0.7754,
      "step": 6320
    },
    {
      "epoch": 1.9958951689295863,
      "grad_norm": 1.73347806930542,
      "learning_rate": 3.338011367623325e-05,
      "loss": 0.8962,
      "step": 6321
    },
    {
      "epoch": 1.996210925165772,
      "grad_norm": 1.5520660877227783,
      "learning_rate": 3.337660515051576e-05,
      "loss": 0.8361,
      "step": 6322
    },
    {
      "epoch": 1.9965266814019578,
      "grad_norm": 1.9288694858551025,
      "learning_rate": 3.3373096624798264e-05,
      "loss": 0.8124,
      "step": 6323
    },
    {
      "epoch": 1.9968424376381435,
      "grad_norm": 1.7653350830078125,
      "learning_rate": 3.3369588099080765e-05,
      "loss": 0.8251,
      "step": 6324
    },
    {
      "epoch": 1.997158193874329,
      "grad_norm": 1.6585365533828735,
      "learning_rate": 3.336607957336327e-05,
      "loss": 0.8067,
      "step": 6325
    },
    {
      "epoch": 1.9974739501105145,
      "grad_norm": 1.6485157012939453,
      "learning_rate": 3.336257104764578e-05,
      "loss": 0.8547,
      "step": 6326
    },
    {
      "epoch": 1.9977897063467003,
      "grad_norm": 1.7723032236099243,
      "learning_rate": 3.3359062521928285e-05,
      "loss": 0.8254,
      "step": 6327
    },
    {
      "epoch": 1.998105462582886,
      "grad_norm": 1.609591007232666,
      "learning_rate": 3.335555399621079e-05,
      "loss": 0.8263,
      "step": 6328
    },
    {
      "epoch": 1.9984212188190718,
      "grad_norm": 1.455913782119751,
      "learning_rate": 3.33520454704933e-05,
      "loss": 0.7977,
      "step": 6329
    },
    {
      "epoch": 1.9987369750552575,
      "grad_norm": 1.7227437496185303,
      "learning_rate": 3.3348536944775805e-05,
      "loss": 0.7908,
      "step": 6330
    },
    {
      "epoch": 1.999052731291443,
      "grad_norm": 1.6630346775054932,
      "learning_rate": 3.334502841905831e-05,
      "loss": 0.8698,
      "step": 6331
    },
    {
      "epoch": 1.9993684875276285,
      "grad_norm": 1.3162474632263184,
      "learning_rate": 3.334151989334082e-05,
      "loss": 0.8077,
      "step": 6332
    },
    {
      "epoch": 1.9996842437638143,
      "grad_norm": 1.4574191570281982,
      "learning_rate": 3.3338011367623326e-05,
      "loss": 0.846,
      "step": 6333
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.6404056549072266,
      "learning_rate": 3.333450284190583e-05,
      "loss": 0.7919,
      "step": 6334
    },
    {
      "epoch": 2.0003157562361857,
      "grad_norm": 1.492295503616333,
      "learning_rate": 3.333099431618834e-05,
      "loss": 0.7683,
      "step": 6335
    },
    {
      "epoch": 2.0006315124723715,
      "grad_norm": 1.396347165107727,
      "learning_rate": 3.3327485790470846e-05,
      "loss": 0.8379,
      "step": 6336
    },
    {
      "epoch": 2.000947268708557,
      "grad_norm": 1.8005321025848389,
      "learning_rate": 3.332397726475335e-05,
      "loss": 0.7186,
      "step": 6337
    },
    {
      "epoch": 2.0012630249447425,
      "grad_norm": 1.5641192197799683,
      "learning_rate": 3.332046873903586e-05,
      "loss": 0.7828,
      "step": 6338
    },
    {
      "epoch": 2.0015787811809282,
      "grad_norm": 1.5769257545471191,
      "learning_rate": 3.331696021331837e-05,
      "loss": 0.7568,
      "step": 6339
    },
    {
      "epoch": 2.001894537417114,
      "grad_norm": 1.5210051536560059,
      "learning_rate": 3.3313451687600874e-05,
      "loss": 0.7987,
      "step": 6340
    },
    {
      "epoch": 2.0022102936532997,
      "grad_norm": 1.4692426919937134,
      "learning_rate": 3.330994316188338e-05,
      "loss": 0.735,
      "step": 6341
    },
    {
      "epoch": 2.0025260498894855,
      "grad_norm": 1.5477075576782227,
      "learning_rate": 3.330643463616589e-05,
      "loss": 0.7894,
      "step": 6342
    },
    {
      "epoch": 2.002841806125671,
      "grad_norm": 1.454825520515442,
      "learning_rate": 3.330292611044839e-05,
      "loss": 0.8186,
      "step": 6343
    },
    {
      "epoch": 2.0031575623618565,
      "grad_norm": 1.6488242149353027,
      "learning_rate": 3.3299417584730894e-05,
      "loss": 0.8304,
      "step": 6344
    },
    {
      "epoch": 2.0034733185980422,
      "grad_norm": 1.9685813188552856,
      "learning_rate": 3.32959090590134e-05,
      "loss": 0.7402,
      "step": 6345
    },
    {
      "epoch": 2.003789074834228,
      "grad_norm": 1.6732372045516968,
      "learning_rate": 3.329240053329591e-05,
      "loss": 0.7621,
      "step": 6346
    },
    {
      "epoch": 2.0041048310704137,
      "grad_norm": 1.7349416017532349,
      "learning_rate": 3.3288892007578414e-05,
      "loss": 0.7313,
      "step": 6347
    },
    {
      "epoch": 2.0044205873065994,
      "grad_norm": 1.392106533050537,
      "learning_rate": 3.328538348186092e-05,
      "loss": 0.686,
      "step": 6348
    },
    {
      "epoch": 2.004736343542785,
      "grad_norm": 1.7619165182113647,
      "learning_rate": 3.3281874956143435e-05,
      "loss": 0.762,
      "step": 6349
    },
    {
      "epoch": 2.0050520997789705,
      "grad_norm": 1.768504023551941,
      "learning_rate": 3.327836643042594e-05,
      "loss": 0.8294,
      "step": 6350
    },
    {
      "epoch": 2.005367856015156,
      "grad_norm": 1.8834800720214844,
      "learning_rate": 3.327485790470845e-05,
      "loss": 0.7581,
      "step": 6351
    },
    {
      "epoch": 2.005683612251342,
      "grad_norm": 1.2648040056228638,
      "learning_rate": 3.327134937899095e-05,
      "loss": 0.6829,
      "step": 6352
    },
    {
      "epoch": 2.0059993684875277,
      "grad_norm": 1.5889233350753784,
      "learning_rate": 3.3267840853273455e-05,
      "loss": 0.7533,
      "step": 6353
    },
    {
      "epoch": 2.0063151247237134,
      "grad_norm": 1.4998031854629517,
      "learning_rate": 3.326433232755596e-05,
      "loss": 0.7409,
      "step": 6354
    },
    {
      "epoch": 2.006630880959899,
      "grad_norm": 1.5897468328475952,
      "learning_rate": 3.326082380183847e-05,
      "loss": 0.7026,
      "step": 6355
    },
    {
      "epoch": 2.0069466371960845,
      "grad_norm": 1.7565042972564697,
      "learning_rate": 3.3257315276120976e-05,
      "loss": 0.7351,
      "step": 6356
    },
    {
      "epoch": 2.00726239343227,
      "grad_norm": 1.5232347249984741,
      "learning_rate": 3.325380675040348e-05,
      "loss": 0.7533,
      "step": 6357
    },
    {
      "epoch": 2.007578149668456,
      "grad_norm": 1.6367688179016113,
      "learning_rate": 3.325029822468599e-05,
      "loss": 0.665,
      "step": 6358
    },
    {
      "epoch": 2.0078939059046417,
      "grad_norm": 1.6221675872802734,
      "learning_rate": 3.3246789698968496e-05,
      "loss": 0.7465,
      "step": 6359
    },
    {
      "epoch": 2.0082096621408274,
      "grad_norm": 1.7685256004333496,
      "learning_rate": 3.3243281173251e-05,
      "loss": 0.7124,
      "step": 6360
    },
    {
      "epoch": 2.008525418377013,
      "grad_norm": 1.5496512651443481,
      "learning_rate": 3.323977264753351e-05,
      "loss": 0.7204,
      "step": 6361
    },
    {
      "epoch": 2.0088411746131984,
      "grad_norm": 1.6250983476638794,
      "learning_rate": 3.323626412181601e-05,
      "loss": 0.6912,
      "step": 6362
    },
    {
      "epoch": 2.009156930849384,
      "grad_norm": 1.757332682609558,
      "learning_rate": 3.323275559609852e-05,
      "loss": 0.7671,
      "step": 6363
    },
    {
      "epoch": 2.00947268708557,
      "grad_norm": 1.54448664188385,
      "learning_rate": 3.3229247070381023e-05,
      "loss": 0.6614,
      "step": 6364
    },
    {
      "epoch": 2.0097884433217557,
      "grad_norm": 1.8627994060516357,
      "learning_rate": 3.322573854466353e-05,
      "loss": 0.757,
      "step": 6365
    },
    {
      "epoch": 2.0101041995579414,
      "grad_norm": 1.809627890586853,
      "learning_rate": 3.322223001894604e-05,
      "loss": 0.6318,
      "step": 6366
    },
    {
      "epoch": 2.010419955794127,
      "grad_norm": 2.020839214324951,
      "learning_rate": 3.321872149322855e-05,
      "loss": 0.7425,
      "step": 6367
    },
    {
      "epoch": 2.0107357120303124,
      "grad_norm": 1.6270614862442017,
      "learning_rate": 3.321521296751106e-05,
      "loss": 0.6721,
      "step": 6368
    },
    {
      "epoch": 2.011051468266498,
      "grad_norm": 1.9273289442062378,
      "learning_rate": 3.3211704441793564e-05,
      "loss": 0.6945,
      "step": 6369
    },
    {
      "epoch": 2.011367224502684,
      "grad_norm": 1.9685349464416504,
      "learning_rate": 3.320819591607607e-05,
      "loss": 0.8003,
      "step": 6370
    },
    {
      "epoch": 2.0116829807388696,
      "grad_norm": 2.0947742462158203,
      "learning_rate": 3.320468739035857e-05,
      "loss": 0.7876,
      "step": 6371
    },
    {
      "epoch": 2.0119987369750554,
      "grad_norm": 2.198631763458252,
      "learning_rate": 3.320117886464108e-05,
      "loss": 0.831,
      "step": 6372
    },
    {
      "epoch": 2.012314493211241,
      "grad_norm": 1.889513611793518,
      "learning_rate": 3.3197670338923585e-05,
      "loss": 0.7715,
      "step": 6373
    },
    {
      "epoch": 2.0126302494474264,
      "grad_norm": 1.9071953296661377,
      "learning_rate": 3.319416181320609e-05,
      "loss": 0.8136,
      "step": 6374
    },
    {
      "epoch": 2.012946005683612,
      "grad_norm": 1.99886155128479,
      "learning_rate": 3.31906532874886e-05,
      "loss": 0.7642,
      "step": 6375
    },
    {
      "epoch": 2.013261761919798,
      "grad_norm": 1.9066705703735352,
      "learning_rate": 3.3187144761771105e-05,
      "loss": 0.8276,
      "step": 6376
    },
    {
      "epoch": 2.0135775181559836,
      "grad_norm": 1.8930857181549072,
      "learning_rate": 3.318363623605361e-05,
      "loss": 0.8309,
      "step": 6377
    },
    {
      "epoch": 2.0138932743921694,
      "grad_norm": 1.792205810546875,
      "learning_rate": 3.318012771033612e-05,
      "loss": 0.7596,
      "step": 6378
    },
    {
      "epoch": 2.014209030628355,
      "grad_norm": 1.7352725267410278,
      "learning_rate": 3.3176619184618626e-05,
      "loss": 0.7807,
      "step": 6379
    },
    {
      "epoch": 2.0145247868645404,
      "grad_norm": 1.9491454362869263,
      "learning_rate": 3.317311065890113e-05,
      "loss": 0.7736,
      "step": 6380
    },
    {
      "epoch": 2.014840543100726,
      "grad_norm": 2.0065839290618896,
      "learning_rate": 3.316960213318364e-05,
      "loss": 0.7107,
      "step": 6381
    },
    {
      "epoch": 2.015156299336912,
      "grad_norm": 1.945356011390686,
      "learning_rate": 3.316609360746614e-05,
      "loss": 0.7756,
      "step": 6382
    },
    {
      "epoch": 2.0154720555730976,
      "grad_norm": 1.7113606929779053,
      "learning_rate": 3.3162585081748646e-05,
      "loss": 0.6965,
      "step": 6383
    },
    {
      "epoch": 2.0157878118092833,
      "grad_norm": 1.9262731075286865,
      "learning_rate": 3.315907655603116e-05,
      "loss": 0.6711,
      "step": 6384
    },
    {
      "epoch": 2.016103568045469,
      "grad_norm": 1.920875072479248,
      "learning_rate": 3.3155568030313667e-05,
      "loss": 0.725,
      "step": 6385
    },
    {
      "epoch": 2.0164193242816544,
      "grad_norm": 2.1092772483825684,
      "learning_rate": 3.315205950459617e-05,
      "loss": 0.7567,
      "step": 6386
    },
    {
      "epoch": 2.01673508051784,
      "grad_norm": 2.0090203285217285,
      "learning_rate": 3.314855097887868e-05,
      "loss": 0.7957,
      "step": 6387
    },
    {
      "epoch": 2.017050836754026,
      "grad_norm": 1.894662857055664,
      "learning_rate": 3.314504245316119e-05,
      "loss": 0.7428,
      "step": 6388
    },
    {
      "epoch": 2.0173665929902116,
      "grad_norm": 2.207106828689575,
      "learning_rate": 3.3141533927443694e-05,
      "loss": 0.7933,
      "step": 6389
    },
    {
      "epoch": 2.0176823492263973,
      "grad_norm": 1.8250393867492676,
      "learning_rate": 3.3138025401726194e-05,
      "loss": 0.6856,
      "step": 6390
    },
    {
      "epoch": 2.017998105462583,
      "grad_norm": 2.1612703800201416,
      "learning_rate": 3.31345168760087e-05,
      "loss": 0.6796,
      "step": 6391
    },
    {
      "epoch": 2.0183138616987684,
      "grad_norm": 1.8539985418319702,
      "learning_rate": 3.313100835029121e-05,
      "loss": 0.7949,
      "step": 6392
    },
    {
      "epoch": 2.018629617934954,
      "grad_norm": 2.0155324935913086,
      "learning_rate": 3.3127499824573714e-05,
      "loss": 0.7694,
      "step": 6393
    },
    {
      "epoch": 2.01894537417114,
      "grad_norm": 1.5613377094268799,
      "learning_rate": 3.312399129885622e-05,
      "loss": 0.6675,
      "step": 6394
    },
    {
      "epoch": 2.0192611304073256,
      "grad_norm": 3.192206621170044,
      "learning_rate": 3.312048277313873e-05,
      "loss": 0.7304,
      "step": 6395
    },
    {
      "epoch": 2.0195768866435113,
      "grad_norm": 1.8701496124267578,
      "learning_rate": 3.3116974247421235e-05,
      "loss": 0.674,
      "step": 6396
    },
    {
      "epoch": 2.019892642879697,
      "grad_norm": 1.8361037969589233,
      "learning_rate": 3.311346572170374e-05,
      "loss": 0.7497,
      "step": 6397
    },
    {
      "epoch": 2.0202083991158823,
      "grad_norm": 1.8061306476593018,
      "learning_rate": 3.310995719598625e-05,
      "loss": 0.7669,
      "step": 6398
    },
    {
      "epoch": 2.020524155352068,
      "grad_norm": 1.7941968441009521,
      "learning_rate": 3.3106448670268755e-05,
      "loss": 0.7057,
      "step": 6399
    },
    {
      "epoch": 2.020839911588254,
      "grad_norm": 2.049414873123169,
      "learning_rate": 3.310294014455126e-05,
      "loss": 0.7396,
      "step": 6400
    },
    {
      "epoch": 2.020839911588254,
      "eval_loss": 0.8002108335494995,
      "eval_runtime": 4.6508,
      "eval_samples_per_second": 535.396,
      "eval_steps_per_second": 4.515,
      "step": 6400
    },
    {
      "epoch": 2.0211556678244396,
      "grad_norm": 1.908068060874939,
      "learning_rate": 3.309943161883376e-05,
      "loss": 0.7355,
      "step": 6401
    },
    {
      "epoch": 2.0214714240606253,
      "grad_norm": 2.128443956375122,
      "learning_rate": 3.3095923093116276e-05,
      "loss": 0.8345,
      "step": 6402
    },
    {
      "epoch": 2.021787180296811,
      "grad_norm": 1.9959487915039062,
      "learning_rate": 3.309241456739878e-05,
      "loss": 0.7723,
      "step": 6403
    },
    {
      "epoch": 2.0221029365329963,
      "grad_norm": 1.814018726348877,
      "learning_rate": 3.308890604168129e-05,
      "loss": 0.6625,
      "step": 6404
    },
    {
      "epoch": 2.022418692769182,
      "grad_norm": 2.0185599327087402,
      "learning_rate": 3.3085397515963796e-05,
      "loss": 0.8061,
      "step": 6405
    },
    {
      "epoch": 2.022734449005368,
      "grad_norm": 2.0665056705474854,
      "learning_rate": 3.30818889902463e-05,
      "loss": 0.7487,
      "step": 6406
    },
    {
      "epoch": 2.0230502052415535,
      "grad_norm": 2.1538619995117188,
      "learning_rate": 3.307838046452881e-05,
      "loss": 0.7589,
      "step": 6407
    },
    {
      "epoch": 2.0233659614777393,
      "grad_norm": 2.033512592315674,
      "learning_rate": 3.3074871938811316e-05,
      "loss": 0.7257,
      "step": 6408
    },
    {
      "epoch": 2.023681717713925,
      "grad_norm": 1.7454860210418701,
      "learning_rate": 3.3071363413093817e-05,
      "loss": 0.6355,
      "step": 6409
    },
    {
      "epoch": 2.0239974739501103,
      "grad_norm": 2.458042621612549,
      "learning_rate": 3.306785488737632e-05,
      "loss": 0.7238,
      "step": 6410
    },
    {
      "epoch": 2.024313230186296,
      "grad_norm": 1.859931468963623,
      "learning_rate": 3.306434636165883e-05,
      "loss": 0.6977,
      "step": 6411
    },
    {
      "epoch": 2.024628986422482,
      "grad_norm": 2.038628339767456,
      "learning_rate": 3.306083783594134e-05,
      "loss": 0.6974,
      "step": 6412
    },
    {
      "epoch": 2.0249447426586675,
      "grad_norm": 2.1263861656188965,
      "learning_rate": 3.3057329310223844e-05,
      "loss": 0.7063,
      "step": 6413
    },
    {
      "epoch": 2.0252604988948533,
      "grad_norm": 2.1232004165649414,
      "learning_rate": 3.305382078450635e-05,
      "loss": 0.7738,
      "step": 6414
    },
    {
      "epoch": 2.025576255131039,
      "grad_norm": 1.9384418725967407,
      "learning_rate": 3.305031225878886e-05,
      "loss": 0.686,
      "step": 6415
    },
    {
      "epoch": 2.0258920113672243,
      "grad_norm": 1.8478412628173828,
      "learning_rate": 3.3046803733071364e-05,
      "loss": 0.7242,
      "step": 6416
    },
    {
      "epoch": 2.02620776760341,
      "grad_norm": 2.1112287044525146,
      "learning_rate": 3.304329520735387e-05,
      "loss": 0.6917,
      "step": 6417
    },
    {
      "epoch": 2.0265235238395958,
      "grad_norm": 2.379385471343994,
      "learning_rate": 3.303978668163638e-05,
      "loss": 0.7607,
      "step": 6418
    },
    {
      "epoch": 2.0268392800757815,
      "grad_norm": 2.2772223949432373,
      "learning_rate": 3.3036278155918885e-05,
      "loss": 0.7273,
      "step": 6419
    },
    {
      "epoch": 2.0271550363119673,
      "grad_norm": 1.9951714277267456,
      "learning_rate": 3.303276963020139e-05,
      "loss": 0.7387,
      "step": 6420
    },
    {
      "epoch": 2.027470792548153,
      "grad_norm": 2.213594675064087,
      "learning_rate": 3.30292611044839e-05,
      "loss": 0.8082,
      "step": 6421
    },
    {
      "epoch": 2.0277865487843383,
      "grad_norm": 2.148743152618408,
      "learning_rate": 3.3025752578766405e-05,
      "loss": 0.7703,
      "step": 6422
    },
    {
      "epoch": 2.028102305020524,
      "grad_norm": 2.0018413066864014,
      "learning_rate": 3.302224405304891e-05,
      "loss": 0.7694,
      "step": 6423
    },
    {
      "epoch": 2.0284180612567098,
      "grad_norm": 2.0477826595306396,
      "learning_rate": 3.301873552733142e-05,
      "loss": 0.678,
      "step": 6424
    },
    {
      "epoch": 2.0287338174928955,
      "grad_norm": 2.3799986839294434,
      "learning_rate": 3.3015227001613926e-05,
      "loss": 0.7604,
      "step": 6425
    },
    {
      "epoch": 2.0290495737290812,
      "grad_norm": 2.042985439300537,
      "learning_rate": 3.301171847589643e-05,
      "loss": 0.6621,
      "step": 6426
    },
    {
      "epoch": 2.029365329965267,
      "grad_norm": 2.121739149093628,
      "learning_rate": 3.300820995017894e-05,
      "loss": 0.8757,
      "step": 6427
    },
    {
      "epoch": 2.0296810862014523,
      "grad_norm": 2.3903725147247314,
      "learning_rate": 3.3004701424461446e-05,
      "loss": 0.7577,
      "step": 6428
    },
    {
      "epoch": 2.029996842437638,
      "grad_norm": 2.3791189193725586,
      "learning_rate": 3.3001192898743946e-05,
      "loss": 0.7658,
      "step": 6429
    },
    {
      "epoch": 2.0303125986738237,
      "grad_norm": 1.9489237070083618,
      "learning_rate": 3.299768437302645e-05,
      "loss": 0.8227,
      "step": 6430
    },
    {
      "epoch": 2.0306283549100095,
      "grad_norm": 2.0425682067871094,
      "learning_rate": 3.299417584730896e-05,
      "loss": 0.7174,
      "step": 6431
    },
    {
      "epoch": 2.030944111146195,
      "grad_norm": 2.042924642562866,
      "learning_rate": 3.2990667321591466e-05,
      "loss": 0.7808,
      "step": 6432
    },
    {
      "epoch": 2.031259867382381,
      "grad_norm": 2.0211596488952637,
      "learning_rate": 3.298715879587397e-05,
      "loss": 0.7898,
      "step": 6433
    },
    {
      "epoch": 2.0315756236185663,
      "grad_norm": 2.264843702316284,
      "learning_rate": 3.298365027015648e-05,
      "loss": 0.8183,
      "step": 6434
    },
    {
      "epoch": 2.031891379854752,
      "grad_norm": 1.9764312505722046,
      "learning_rate": 3.298014174443899e-05,
      "loss": 0.7705,
      "step": 6435
    },
    {
      "epoch": 2.0322071360909377,
      "grad_norm": 2.0788519382476807,
      "learning_rate": 3.2976633218721494e-05,
      "loss": 0.7246,
      "step": 6436
    },
    {
      "epoch": 2.0325228923271235,
      "grad_norm": 2.3481295108795166,
      "learning_rate": 3.2973124693004e-05,
      "loss": 0.8018,
      "step": 6437
    },
    {
      "epoch": 2.032838648563309,
      "grad_norm": 2.3540287017822266,
      "learning_rate": 3.296961616728651e-05,
      "loss": 0.6858,
      "step": 6438
    },
    {
      "epoch": 2.033154404799495,
      "grad_norm": 1.8672012090682983,
      "learning_rate": 3.2966107641569014e-05,
      "loss": 0.731,
      "step": 6439
    },
    {
      "epoch": 2.0334701610356802,
      "grad_norm": 2.4494681358337402,
      "learning_rate": 3.296259911585152e-05,
      "loss": 0.7101,
      "step": 6440
    },
    {
      "epoch": 2.033785917271866,
      "grad_norm": 1.9512306451797485,
      "learning_rate": 3.295909059013403e-05,
      "loss": 0.7836,
      "step": 6441
    },
    {
      "epoch": 2.0341016735080517,
      "grad_norm": 2.17856764793396,
      "learning_rate": 3.2955582064416535e-05,
      "loss": 0.8247,
      "step": 6442
    },
    {
      "epoch": 2.0344174297442374,
      "grad_norm": 1.94393789768219,
      "learning_rate": 3.295207353869904e-05,
      "loss": 0.7773,
      "step": 6443
    },
    {
      "epoch": 2.034733185980423,
      "grad_norm": 1.6540597677230835,
      "learning_rate": 3.294856501298155e-05,
      "loss": 0.6775,
      "step": 6444
    },
    {
      "epoch": 2.035048942216609,
      "grad_norm": 1.9585721492767334,
      "learning_rate": 3.2945056487264055e-05,
      "loss": 0.6189,
      "step": 6445
    },
    {
      "epoch": 2.0353646984527947,
      "grad_norm": 1.9853477478027344,
      "learning_rate": 3.294154796154656e-05,
      "loss": 0.8075,
      "step": 6446
    },
    {
      "epoch": 2.03568045468898,
      "grad_norm": 2.197573661804199,
      "learning_rate": 3.293803943582907e-05,
      "loss": 0.7279,
      "step": 6447
    },
    {
      "epoch": 2.0359962109251657,
      "grad_norm": 1.8134406805038452,
      "learning_rate": 3.293453091011157e-05,
      "loss": 0.7414,
      "step": 6448
    },
    {
      "epoch": 2.0363119671613514,
      "grad_norm": 2.1368350982666016,
      "learning_rate": 3.2931022384394075e-05,
      "loss": 0.7785,
      "step": 6449
    },
    {
      "epoch": 2.036627723397537,
      "grad_norm": 2.0109541416168213,
      "learning_rate": 3.292751385867658e-05,
      "loss": 0.7745,
      "step": 6450
    },
    {
      "epoch": 2.036943479633723,
      "grad_norm": 2.2689733505249023,
      "learning_rate": 3.292400533295909e-05,
      "loss": 0.7852,
      "step": 6451
    },
    {
      "epoch": 2.0372592358699086,
      "grad_norm": 2.136427402496338,
      "learning_rate": 3.2920496807241596e-05,
      "loss": 0.7403,
      "step": 6452
    },
    {
      "epoch": 2.037574992106094,
      "grad_norm": 1.8920031785964966,
      "learning_rate": 3.29169882815241e-05,
      "loss": 0.7151,
      "step": 6453
    },
    {
      "epoch": 2.0378907483422797,
      "grad_norm": 2.134596586227417,
      "learning_rate": 3.2913479755806616e-05,
      "loss": 0.8522,
      "step": 6454
    },
    {
      "epoch": 2.0382065045784654,
      "grad_norm": 1.983984351158142,
      "learning_rate": 3.290997123008912e-05,
      "loss": 0.6664,
      "step": 6455
    },
    {
      "epoch": 2.038522260814651,
      "grad_norm": 1.7447128295898438,
      "learning_rate": 3.290646270437163e-05,
      "loss": 0.6868,
      "step": 6456
    },
    {
      "epoch": 2.038838017050837,
      "grad_norm": 3.0061018466949463,
      "learning_rate": 3.290295417865413e-05,
      "loss": 0.7204,
      "step": 6457
    },
    {
      "epoch": 2.0391537732870226,
      "grad_norm": 2.0019466876983643,
      "learning_rate": 3.289944565293664e-05,
      "loss": 0.7463,
      "step": 6458
    },
    {
      "epoch": 2.039469529523208,
      "grad_norm": 2.247622013092041,
      "learning_rate": 3.2895937127219144e-05,
      "loss": 0.7882,
      "step": 6459
    },
    {
      "epoch": 2.0397852857593937,
      "grad_norm": 1.9882423877716064,
      "learning_rate": 3.289242860150165e-05,
      "loss": 0.8703,
      "step": 6460
    },
    {
      "epoch": 2.0401010419955794,
      "grad_norm": 2.266451597213745,
      "learning_rate": 3.288892007578416e-05,
      "loss": 0.7483,
      "step": 6461
    },
    {
      "epoch": 2.040416798231765,
      "grad_norm": 2.0641090869903564,
      "learning_rate": 3.2885411550066664e-05,
      "loss": 0.7365,
      "step": 6462
    },
    {
      "epoch": 2.040732554467951,
      "grad_norm": 2.055565357208252,
      "learning_rate": 3.288190302434917e-05,
      "loss": 0.7126,
      "step": 6463
    },
    {
      "epoch": 2.0410483107041366,
      "grad_norm": 2.102862596511841,
      "learning_rate": 3.287839449863168e-05,
      "loss": 0.7158,
      "step": 6464
    },
    {
      "epoch": 2.041364066940322,
      "grad_norm": 2.1754610538482666,
      "learning_rate": 3.2874885972914185e-05,
      "loss": 0.7603,
      "step": 6465
    },
    {
      "epoch": 2.0416798231765076,
      "grad_norm": 1.9025212526321411,
      "learning_rate": 3.287137744719669e-05,
      "loss": 0.6917,
      "step": 6466
    },
    {
      "epoch": 2.0419955794126934,
      "grad_norm": 1.8822622299194336,
      "learning_rate": 3.286786892147919e-05,
      "loss": 0.6254,
      "step": 6467
    },
    {
      "epoch": 2.042311335648879,
      "grad_norm": 2.178542137145996,
      "learning_rate": 3.28643603957617e-05,
      "loss": 0.8083,
      "step": 6468
    },
    {
      "epoch": 2.042627091885065,
      "grad_norm": 2.089355707168579,
      "learning_rate": 3.2860851870044205e-05,
      "loss": 0.7585,
      "step": 6469
    },
    {
      "epoch": 2.0429428481212506,
      "grad_norm": 2.0004284381866455,
      "learning_rate": 3.285734334432671e-05,
      "loss": 0.6913,
      "step": 6470
    },
    {
      "epoch": 2.043258604357436,
      "grad_norm": 2.2849936485290527,
      "learning_rate": 3.285383481860922e-05,
      "loss": 0.8095,
      "step": 6471
    },
    {
      "epoch": 2.0435743605936216,
      "grad_norm": 2.0796287059783936,
      "learning_rate": 3.285032629289173e-05,
      "loss": 0.7719,
      "step": 6472
    },
    {
      "epoch": 2.0438901168298074,
      "grad_norm": 2.227362632751465,
      "learning_rate": 3.284681776717424e-05,
      "loss": 0.7674,
      "step": 6473
    },
    {
      "epoch": 2.044205873065993,
      "grad_norm": 1.8189998865127563,
      "learning_rate": 3.2843309241456746e-05,
      "loss": 0.686,
      "step": 6474
    },
    {
      "epoch": 2.044521629302179,
      "grad_norm": 2.191072702407837,
      "learning_rate": 3.283980071573925e-05,
      "loss": 0.6738,
      "step": 6475
    },
    {
      "epoch": 2.0448373855383646,
      "grad_norm": 2.047302484512329,
      "learning_rate": 3.283629219002175e-05,
      "loss": 0.7357,
      "step": 6476
    },
    {
      "epoch": 2.04515314177455,
      "grad_norm": 1.833237648010254,
      "learning_rate": 3.283278366430426e-05,
      "loss": 0.7328,
      "step": 6477
    },
    {
      "epoch": 2.0454688980107356,
      "grad_norm": 1.9855940341949463,
      "learning_rate": 3.2829275138586766e-05,
      "loss": 0.8008,
      "step": 6478
    },
    {
      "epoch": 2.0457846542469214,
      "grad_norm": 2.275557041168213,
      "learning_rate": 3.282576661286927e-05,
      "loss": 0.757,
      "step": 6479
    },
    {
      "epoch": 2.046100410483107,
      "grad_norm": 2.1609694957733154,
      "learning_rate": 3.282225808715178e-05,
      "loss": 0.7131,
      "step": 6480
    },
    {
      "epoch": 2.046416166719293,
      "grad_norm": 1.9149377346038818,
      "learning_rate": 3.281874956143429e-05,
      "loss": 0.6425,
      "step": 6481
    },
    {
      "epoch": 2.0467319229554786,
      "grad_norm": 1.9828675985336304,
      "learning_rate": 3.2815241035716794e-05,
      "loss": 0.8167,
      "step": 6482
    },
    {
      "epoch": 2.047047679191664,
      "grad_norm": 1.9869507551193237,
      "learning_rate": 3.28117325099993e-05,
      "loss": 0.7314,
      "step": 6483
    },
    {
      "epoch": 2.0473634354278496,
      "grad_norm": 2.3055427074432373,
      "learning_rate": 3.280822398428181e-05,
      "loss": 0.6765,
      "step": 6484
    },
    {
      "epoch": 2.0476791916640353,
      "grad_norm": 2.312377691268921,
      "learning_rate": 3.2804715458564314e-05,
      "loss": 0.8236,
      "step": 6485
    },
    {
      "epoch": 2.047994947900221,
      "grad_norm": 1.9812113046646118,
      "learning_rate": 3.280120693284682e-05,
      "loss": 0.718,
      "step": 6486
    },
    {
      "epoch": 2.048310704136407,
      "grad_norm": 2.4211349487304688,
      "learning_rate": 3.279769840712932e-05,
      "loss": 0.7324,
      "step": 6487
    },
    {
      "epoch": 2.0486264603725926,
      "grad_norm": 2.343109607696533,
      "learning_rate": 3.279418988141183e-05,
      "loss": 0.8108,
      "step": 6488
    },
    {
      "epoch": 2.048942216608778,
      "grad_norm": 2.057954788208008,
      "learning_rate": 3.279068135569434e-05,
      "loss": 0.7534,
      "step": 6489
    },
    {
      "epoch": 2.0492579728449636,
      "grad_norm": 2.2559096813201904,
      "learning_rate": 3.278717282997685e-05,
      "loss": 0.8236,
      "step": 6490
    },
    {
      "epoch": 2.0495737290811493,
      "grad_norm": 2.177590847015381,
      "learning_rate": 3.2783664304259355e-05,
      "loss": 0.7515,
      "step": 6491
    },
    {
      "epoch": 2.049889485317335,
      "grad_norm": 1.9091200828552246,
      "learning_rate": 3.278015577854186e-05,
      "loss": 0.7216,
      "step": 6492
    },
    {
      "epoch": 2.050205241553521,
      "grad_norm": 2.1055219173431396,
      "learning_rate": 3.277664725282437e-05,
      "loss": 0.7581,
      "step": 6493
    },
    {
      "epoch": 2.0505209977897065,
      "grad_norm": 2.1054258346557617,
      "learning_rate": 3.2773138727106875e-05,
      "loss": 0.7825,
      "step": 6494
    },
    {
      "epoch": 2.050836754025892,
      "grad_norm": 1.740094780921936,
      "learning_rate": 3.2769630201389375e-05,
      "loss": 0.6495,
      "step": 6495
    },
    {
      "epoch": 2.0511525102620776,
      "grad_norm": 2.0872693061828613,
      "learning_rate": 3.276612167567188e-05,
      "loss": 0.7859,
      "step": 6496
    },
    {
      "epoch": 2.0514682664982633,
      "grad_norm": 2.2490272521972656,
      "learning_rate": 3.276261314995439e-05,
      "loss": 0.7421,
      "step": 6497
    },
    {
      "epoch": 2.051784022734449,
      "grad_norm": 2.133593797683716,
      "learning_rate": 3.2759104624236896e-05,
      "loss": 0.7421,
      "step": 6498
    },
    {
      "epoch": 2.052099778970635,
      "grad_norm": 2.2421038150787354,
      "learning_rate": 3.27555960985194e-05,
      "loss": 0.7567,
      "step": 6499
    },
    {
      "epoch": 2.0524155352068205,
      "grad_norm": 2.1320831775665283,
      "learning_rate": 3.275208757280191e-05,
      "loss": 0.7974,
      "step": 6500
    },
    {
      "epoch": 2.052731291443006,
      "grad_norm": 2.296044111251831,
      "learning_rate": 3.2748579047084416e-05,
      "loss": 0.7093,
      "step": 6501
    },
    {
      "epoch": 2.0530470476791916,
      "grad_norm": 2.032609462738037,
      "learning_rate": 3.274507052136692e-05,
      "loss": 0.7786,
      "step": 6502
    },
    {
      "epoch": 2.0533628039153773,
      "grad_norm": 2.083488941192627,
      "learning_rate": 3.274156199564943e-05,
      "loss": 0.6981,
      "step": 6503
    },
    {
      "epoch": 2.053678560151563,
      "grad_norm": 2.1647591590881348,
      "learning_rate": 3.273805346993194e-05,
      "loss": 0.7373,
      "step": 6504
    },
    {
      "epoch": 2.0539943163877488,
      "grad_norm": 1.9790384769439697,
      "learning_rate": 3.2734544944214443e-05,
      "loss": 0.7794,
      "step": 6505
    },
    {
      "epoch": 2.0543100726239345,
      "grad_norm": 1.8212674856185913,
      "learning_rate": 3.2731036418496944e-05,
      "loss": 0.751,
      "step": 6506
    },
    {
      "epoch": 2.05462582886012,
      "grad_norm": 2.367886543273926,
      "learning_rate": 3.272752789277946e-05,
      "loss": 0.7868,
      "step": 6507
    },
    {
      "epoch": 2.0549415850963055,
      "grad_norm": 2.009960651397705,
      "learning_rate": 3.2724019367061964e-05,
      "loss": 0.8188,
      "step": 6508
    },
    {
      "epoch": 2.0552573413324913,
      "grad_norm": 2.0048441886901855,
      "learning_rate": 3.272051084134447e-05,
      "loss": 0.6598,
      "step": 6509
    },
    {
      "epoch": 2.055573097568677,
      "grad_norm": 2.0312254428863525,
      "learning_rate": 3.271700231562698e-05,
      "loss": 0.7582,
      "step": 6510
    },
    {
      "epoch": 2.0558888538048627,
      "grad_norm": 1.72425377368927,
      "learning_rate": 3.2713493789909484e-05,
      "loss": 0.6026,
      "step": 6511
    },
    {
      "epoch": 2.0562046100410485,
      "grad_norm": 2.447788715362549,
      "learning_rate": 3.270998526419199e-05,
      "loss": 0.7178,
      "step": 6512
    },
    {
      "epoch": 2.056520366277234,
      "grad_norm": 1.968916654586792,
      "learning_rate": 3.27064767384745e-05,
      "loss": 0.8042,
      "step": 6513
    },
    {
      "epoch": 2.0568361225134195,
      "grad_norm": 1.8967713117599487,
      "learning_rate": 3.2702968212757005e-05,
      "loss": 0.7247,
      "step": 6514
    },
    {
      "epoch": 2.0571518787496053,
      "grad_norm": 1.9737569093704224,
      "learning_rate": 3.2699459687039505e-05,
      "loss": 0.696,
      "step": 6515
    },
    {
      "epoch": 2.057467634985791,
      "grad_norm": 2.1453864574432373,
      "learning_rate": 3.269595116132201e-05,
      "loss": 0.8653,
      "step": 6516
    },
    {
      "epoch": 2.0577833912219767,
      "grad_norm": 2.123584747314453,
      "learning_rate": 3.269244263560452e-05,
      "loss": 0.7152,
      "step": 6517
    },
    {
      "epoch": 2.0580991474581625,
      "grad_norm": 2.034045696258545,
      "learning_rate": 3.2688934109887025e-05,
      "loss": 0.722,
      "step": 6518
    },
    {
      "epoch": 2.0584149036943478,
      "grad_norm": 1.890863060951233,
      "learning_rate": 3.268542558416953e-05,
      "loss": 0.6969,
      "step": 6519
    },
    {
      "epoch": 2.0587306599305335,
      "grad_norm": 2.4930038452148438,
      "learning_rate": 3.268191705845204e-05,
      "loss": 0.7955,
      "step": 6520
    },
    {
      "epoch": 2.0590464161667192,
      "grad_norm": 1.8160390853881836,
      "learning_rate": 3.2678408532734546e-05,
      "loss": 0.7251,
      "step": 6521
    },
    {
      "epoch": 2.059362172402905,
      "grad_norm": 1.93675696849823,
      "learning_rate": 3.267490000701705e-05,
      "loss": 0.7677,
      "step": 6522
    },
    {
      "epoch": 2.0596779286390907,
      "grad_norm": 2.20833420753479,
      "learning_rate": 3.267139148129956e-05,
      "loss": 0.7818,
      "step": 6523
    },
    {
      "epoch": 2.0599936848752765,
      "grad_norm": 1.7513138055801392,
      "learning_rate": 3.2667882955582066e-05,
      "loss": 0.6659,
      "step": 6524
    },
    {
      "epoch": 2.0603094411114617,
      "grad_norm": 2.018664836883545,
      "learning_rate": 3.266437442986457e-05,
      "loss": 0.8012,
      "step": 6525
    },
    {
      "epoch": 2.0606251973476475,
      "grad_norm": 1.7381110191345215,
      "learning_rate": 3.266086590414708e-05,
      "loss": 0.7054,
      "step": 6526
    },
    {
      "epoch": 2.0609409535838332,
      "grad_norm": 2.135803699493408,
      "learning_rate": 3.2657357378429587e-05,
      "loss": 0.7775,
      "step": 6527
    },
    {
      "epoch": 2.061256709820019,
      "grad_norm": 2.171966075897217,
      "learning_rate": 3.2653848852712093e-05,
      "loss": 0.8147,
      "step": 6528
    },
    {
      "epoch": 2.0615724660562047,
      "grad_norm": 2.34478497505188,
      "learning_rate": 3.26503403269946e-05,
      "loss": 0.7621,
      "step": 6529
    },
    {
      "epoch": 2.0618882222923904,
      "grad_norm": 2.5370421409606934,
      "learning_rate": 3.264683180127711e-05,
      "loss": 0.7672,
      "step": 6530
    },
    {
      "epoch": 2.0622039785285757,
      "grad_norm": 2.373250722885132,
      "learning_rate": 3.2643323275559614e-05,
      "loss": 0.82,
      "step": 6531
    },
    {
      "epoch": 2.0625197347647615,
      "grad_norm": 2.3186488151550293,
      "learning_rate": 3.263981474984212e-05,
      "loss": 0.7109,
      "step": 6532
    },
    {
      "epoch": 2.062835491000947,
      "grad_norm": 2.0560848712921143,
      "learning_rate": 3.263630622412463e-05,
      "loss": 0.7422,
      "step": 6533
    },
    {
      "epoch": 2.063151247237133,
      "grad_norm": 1.8964918851852417,
      "learning_rate": 3.263279769840713e-05,
      "loss": 0.7869,
      "step": 6534
    },
    {
      "epoch": 2.0634670034733187,
      "grad_norm": 2.066243886947632,
      "learning_rate": 3.2629289172689634e-05,
      "loss": 0.7361,
      "step": 6535
    },
    {
      "epoch": 2.0637827597095044,
      "grad_norm": 1.9111579656600952,
      "learning_rate": 3.262578064697214e-05,
      "loss": 0.6331,
      "step": 6536
    },
    {
      "epoch": 2.0640985159456897,
      "grad_norm": 2.073075771331787,
      "learning_rate": 3.262227212125465e-05,
      "loss": 0.7403,
      "step": 6537
    },
    {
      "epoch": 2.0644142721818755,
      "grad_norm": 1.8149272203445435,
      "learning_rate": 3.2618763595537155e-05,
      "loss": 0.7175,
      "step": 6538
    },
    {
      "epoch": 2.064730028418061,
      "grad_norm": 2.0119590759277344,
      "learning_rate": 3.261525506981966e-05,
      "loss": 0.7928,
      "step": 6539
    },
    {
      "epoch": 2.065045784654247,
      "grad_norm": 2.1135449409484863,
      "learning_rate": 3.261174654410217e-05,
      "loss": 0.8316,
      "step": 6540
    },
    {
      "epoch": 2.0653615408904327,
      "grad_norm": 2.1925501823425293,
      "learning_rate": 3.2608238018384675e-05,
      "loss": 0.7847,
      "step": 6541
    },
    {
      "epoch": 2.0656772971266184,
      "grad_norm": 2.2102808952331543,
      "learning_rate": 3.260472949266718e-05,
      "loss": 0.7857,
      "step": 6542
    },
    {
      "epoch": 2.0659930533628037,
      "grad_norm": 2.32311749458313,
      "learning_rate": 3.260122096694969e-05,
      "loss": 0.7892,
      "step": 6543
    },
    {
      "epoch": 2.0663088095989894,
      "grad_norm": 2.0817503929138184,
      "learning_rate": 3.2597712441232196e-05,
      "loss": 0.7402,
      "step": 6544
    },
    {
      "epoch": 2.066624565835175,
      "grad_norm": 2.434041738510132,
      "learning_rate": 3.25942039155147e-05,
      "loss": 0.7963,
      "step": 6545
    },
    {
      "epoch": 2.066940322071361,
      "grad_norm": 2.0663223266601562,
      "learning_rate": 3.259069538979721e-05,
      "loss": 0.7502,
      "step": 6546
    },
    {
      "epoch": 2.0672560783075467,
      "grad_norm": 2.214555025100708,
      "learning_rate": 3.2587186864079716e-05,
      "loss": 0.7588,
      "step": 6547
    },
    {
      "epoch": 2.0675718345437324,
      "grad_norm": 2.1099555492401123,
      "learning_rate": 3.258367833836222e-05,
      "loss": 0.7724,
      "step": 6548
    },
    {
      "epoch": 2.067887590779918,
      "grad_norm": 2.0489418506622314,
      "learning_rate": 3.258016981264473e-05,
      "loss": 0.7301,
      "step": 6549
    },
    {
      "epoch": 2.0682033470161034,
      "grad_norm": 2.1683804988861084,
      "learning_rate": 3.2576661286927237e-05,
      "loss": 0.7623,
      "step": 6550
    },
    {
      "epoch": 2.068519103252289,
      "grad_norm": 2.1479246616363525,
      "learning_rate": 3.257315276120974e-05,
      "loss": 0.7783,
      "step": 6551
    },
    {
      "epoch": 2.068834859488475,
      "grad_norm": 2.182572841644287,
      "learning_rate": 3.256964423549225e-05,
      "loss": 0.8296,
      "step": 6552
    },
    {
      "epoch": 2.0691506157246606,
      "grad_norm": 2.3735098838806152,
      "learning_rate": 3.256613570977475e-05,
      "loss": 0.7578,
      "step": 6553
    },
    {
      "epoch": 2.0694663719608464,
      "grad_norm": 2.52626633644104,
      "learning_rate": 3.256262718405726e-05,
      "loss": 0.7615,
      "step": 6554
    },
    {
      "epoch": 2.0697821281970317,
      "grad_norm": 2.068345546722412,
      "learning_rate": 3.2559118658339764e-05,
      "loss": 0.8015,
      "step": 6555
    },
    {
      "epoch": 2.0700978844332174,
      "grad_norm": 2.2631423473358154,
      "learning_rate": 3.255561013262227e-05,
      "loss": 0.8047,
      "step": 6556
    },
    {
      "epoch": 2.070413640669403,
      "grad_norm": 1.9937427043914795,
      "learning_rate": 3.255210160690478e-05,
      "loss": 0.7305,
      "step": 6557
    },
    {
      "epoch": 2.070729396905589,
      "grad_norm": 2.4205543994903564,
      "learning_rate": 3.2548593081187284e-05,
      "loss": 0.8025,
      "step": 6558
    },
    {
      "epoch": 2.0710451531417746,
      "grad_norm": 1.711172342300415,
      "learning_rate": 3.25450845554698e-05,
      "loss": 0.703,
      "step": 6559
    },
    {
      "epoch": 2.0713609093779604,
      "grad_norm": 2.0886597633361816,
      "learning_rate": 3.2541576029752305e-05,
      "loss": 0.766,
      "step": 6560
    },
    {
      "epoch": 2.071676665614146,
      "grad_norm": 2.3911187648773193,
      "learning_rate": 3.253806750403481e-05,
      "loss": 0.7389,
      "step": 6561
    },
    {
      "epoch": 2.0719924218503314,
      "grad_norm": 1.953726053237915,
      "learning_rate": 3.253455897831731e-05,
      "loss": 0.6777,
      "step": 6562
    },
    {
      "epoch": 2.072308178086517,
      "grad_norm": 2.235081911087036,
      "learning_rate": 3.253105045259982e-05,
      "loss": 0.7517,
      "step": 6563
    },
    {
      "epoch": 2.072623934322703,
      "grad_norm": 2.3326988220214844,
      "learning_rate": 3.2527541926882325e-05,
      "loss": 0.8918,
      "step": 6564
    },
    {
      "epoch": 2.0729396905588886,
      "grad_norm": 1.8050447702407837,
      "learning_rate": 3.252403340116483e-05,
      "loss": 0.7157,
      "step": 6565
    },
    {
      "epoch": 2.0732554467950743,
      "grad_norm": 2.251981496810913,
      "learning_rate": 3.252052487544734e-05,
      "loss": 0.8361,
      "step": 6566
    },
    {
      "epoch": 2.07357120303126,
      "grad_norm": 2.100240468978882,
      "learning_rate": 3.2517016349729846e-05,
      "loss": 0.8034,
      "step": 6567
    },
    {
      "epoch": 2.0738869592674454,
      "grad_norm": 2.5308914184570312,
      "learning_rate": 3.251350782401235e-05,
      "loss": 0.8158,
      "step": 6568
    },
    {
      "epoch": 2.074202715503631,
      "grad_norm": 1.9172786474227905,
      "learning_rate": 3.250999929829486e-05,
      "loss": 0.7116,
      "step": 6569
    },
    {
      "epoch": 2.074518471739817,
      "grad_norm": 1.7444581985473633,
      "learning_rate": 3.2506490772577366e-05,
      "loss": 0.7602,
      "step": 6570
    },
    {
      "epoch": 2.0748342279760026,
      "grad_norm": 2.0086705684661865,
      "learning_rate": 3.250298224685987e-05,
      "loss": 0.8353,
      "step": 6571
    },
    {
      "epoch": 2.0751499842121883,
      "grad_norm": 1.921678066253662,
      "learning_rate": 3.249947372114237e-05,
      "loss": 0.6899,
      "step": 6572
    },
    {
      "epoch": 2.075465740448374,
      "grad_norm": 1.9560898542404175,
      "learning_rate": 3.249596519542488e-05,
      "loss": 0.6534,
      "step": 6573
    },
    {
      "epoch": 2.0757814966845594,
      "grad_norm": 1.8242032527923584,
      "learning_rate": 3.2492456669707386e-05,
      "loss": 0.7565,
      "step": 6574
    },
    {
      "epoch": 2.076097252920745,
      "grad_norm": 1.7818056344985962,
      "learning_rate": 3.248894814398989e-05,
      "loss": 0.7443,
      "step": 6575
    },
    {
      "epoch": 2.076413009156931,
      "grad_norm": 2.162964105606079,
      "learning_rate": 3.24854396182724e-05,
      "loss": 0.8121,
      "step": 6576
    },
    {
      "epoch": 2.0767287653931166,
      "grad_norm": 1.8250137567520142,
      "learning_rate": 3.2481931092554914e-05,
      "loss": 0.7717,
      "step": 6577
    },
    {
      "epoch": 2.0770445216293023,
      "grad_norm": 2.0499258041381836,
      "learning_rate": 3.247842256683742e-05,
      "loss": 0.7264,
      "step": 6578
    },
    {
      "epoch": 2.077360277865488,
      "grad_norm": 2.3792617321014404,
      "learning_rate": 3.247491404111993e-05,
      "loss": 0.7157,
      "step": 6579
    },
    {
      "epoch": 2.0776760341016733,
      "grad_norm": 1.8136591911315918,
      "learning_rate": 3.2471405515402434e-05,
      "loss": 0.7353,
      "step": 6580
    },
    {
      "epoch": 2.077991790337859,
      "grad_norm": 1.8944909572601318,
      "learning_rate": 3.2467896989684934e-05,
      "loss": 0.6859,
      "step": 6581
    },
    {
      "epoch": 2.078307546574045,
      "grad_norm": 1.9601718187332153,
      "learning_rate": 3.246438846396744e-05,
      "loss": 0.7371,
      "step": 6582
    },
    {
      "epoch": 2.0786233028102306,
      "grad_norm": 1.785563588142395,
      "learning_rate": 3.246087993824995e-05,
      "loss": 0.6519,
      "step": 6583
    },
    {
      "epoch": 2.0789390590464163,
      "grad_norm": 1.8035155534744263,
      "learning_rate": 3.2457371412532455e-05,
      "loss": 0.7624,
      "step": 6584
    },
    {
      "epoch": 2.079254815282602,
      "grad_norm": 2.622244119644165,
      "learning_rate": 3.245386288681496e-05,
      "loss": 0.7161,
      "step": 6585
    },
    {
      "epoch": 2.0795705715187873,
      "grad_norm": 2.0769288539886475,
      "learning_rate": 3.245035436109747e-05,
      "loss": 0.7644,
      "step": 6586
    },
    {
      "epoch": 2.079886327754973,
      "grad_norm": 2.1280462741851807,
      "learning_rate": 3.2446845835379975e-05,
      "loss": 0.7723,
      "step": 6587
    },
    {
      "epoch": 2.080202083991159,
      "grad_norm": 2.115908145904541,
      "learning_rate": 3.244333730966248e-05,
      "loss": 0.7135,
      "step": 6588
    },
    {
      "epoch": 2.0805178402273445,
      "grad_norm": 2.0026094913482666,
      "learning_rate": 3.243982878394499e-05,
      "loss": 0.753,
      "step": 6589
    },
    {
      "epoch": 2.0808335964635303,
      "grad_norm": 2.3101110458374023,
      "learning_rate": 3.2436320258227495e-05,
      "loss": 0.7933,
      "step": 6590
    },
    {
      "epoch": 2.081149352699716,
      "grad_norm": 2.1705563068389893,
      "learning_rate": 3.243281173251e-05,
      "loss": 0.7016,
      "step": 6591
    },
    {
      "epoch": 2.0814651089359013,
      "grad_norm": 1.9329017400741577,
      "learning_rate": 3.24293032067925e-05,
      "loss": 0.706,
      "step": 6592
    },
    {
      "epoch": 2.081780865172087,
      "grad_norm": 2.1450281143188477,
      "learning_rate": 3.242579468107501e-05,
      "loss": 0.7486,
      "step": 6593
    },
    {
      "epoch": 2.082096621408273,
      "grad_norm": 1.8609877824783325,
      "learning_rate": 3.242228615535752e-05,
      "loss": 0.7174,
      "step": 6594
    },
    {
      "epoch": 2.0824123776444585,
      "grad_norm": 1.830841302871704,
      "learning_rate": 3.241877762964003e-05,
      "loss": 0.6525,
      "step": 6595
    },
    {
      "epoch": 2.0827281338806443,
      "grad_norm": 2.0177626609802246,
      "learning_rate": 3.2415269103922536e-05,
      "loss": 0.7681,
      "step": 6596
    },
    {
      "epoch": 2.08304389011683,
      "grad_norm": 2.0386557579040527,
      "learning_rate": 3.241176057820504e-05,
      "loss": 0.6726,
      "step": 6597
    },
    {
      "epoch": 2.0833596463530153,
      "grad_norm": 2.0549044609069824,
      "learning_rate": 3.240825205248755e-05,
      "loss": 0.7659,
      "step": 6598
    },
    {
      "epoch": 2.083675402589201,
      "grad_norm": 2.3763279914855957,
      "learning_rate": 3.240474352677006e-05,
      "loss": 0.7315,
      "step": 6599
    },
    {
      "epoch": 2.0839911588253868,
      "grad_norm": 2.0496819019317627,
      "learning_rate": 3.240123500105256e-05,
      "loss": 0.6598,
      "step": 6600
    },
    {
      "epoch": 2.0839911588253868,
      "eval_loss": 0.8058702349662781,
      "eval_runtime": 4.6484,
      "eval_samples_per_second": 535.666,
      "eval_steps_per_second": 4.518,
      "step": 6600
    },
    {
      "epoch": 2.0843069150615725,
      "grad_norm": 2.6891376972198486,
      "learning_rate": 3.2397726475335064e-05,
      "loss": 0.823,
      "step": 6601
    },
    {
      "epoch": 2.0846226712977582,
      "grad_norm": 1.8229730129241943,
      "learning_rate": 3.239421794961757e-05,
      "loss": 0.6651,
      "step": 6602
    },
    {
      "epoch": 2.084938427533944,
      "grad_norm": 2.2517082691192627,
      "learning_rate": 3.239070942390008e-05,
      "loss": 0.7831,
      "step": 6603
    },
    {
      "epoch": 2.0852541837701293,
      "grad_norm": 1.894599199295044,
      "learning_rate": 3.2387200898182584e-05,
      "loss": 0.7288,
      "step": 6604
    },
    {
      "epoch": 2.085569940006315,
      "grad_norm": 2.013705015182495,
      "learning_rate": 3.238369237246509e-05,
      "loss": 0.7775,
      "step": 6605
    },
    {
      "epoch": 2.0858856962425008,
      "grad_norm": 2.4021410942077637,
      "learning_rate": 3.23801838467476e-05,
      "loss": 0.769,
      "step": 6606
    },
    {
      "epoch": 2.0862014524786865,
      "grad_norm": 2.3106143474578857,
      "learning_rate": 3.2376675321030105e-05,
      "loss": 0.7629,
      "step": 6607
    },
    {
      "epoch": 2.0865172087148722,
      "grad_norm": 2.153887987136841,
      "learning_rate": 3.237316679531261e-05,
      "loss": 0.7489,
      "step": 6608
    },
    {
      "epoch": 2.086832964951058,
      "grad_norm": 2.3271396160125732,
      "learning_rate": 3.236965826959512e-05,
      "loss": 0.7303,
      "step": 6609
    },
    {
      "epoch": 2.0871487211872433,
      "grad_norm": 2.3150453567504883,
      "learning_rate": 3.2366149743877625e-05,
      "loss": 0.7499,
      "step": 6610
    },
    {
      "epoch": 2.087464477423429,
      "grad_norm": 2.35286021232605,
      "learning_rate": 3.2362641218160125e-05,
      "loss": 0.7701,
      "step": 6611
    },
    {
      "epoch": 2.0877802336596147,
      "grad_norm": 2.4319615364074707,
      "learning_rate": 3.235913269244264e-05,
      "loss": 0.7796,
      "step": 6612
    },
    {
      "epoch": 2.0880959898958005,
      "grad_norm": 2.418668746948242,
      "learning_rate": 3.2355624166725145e-05,
      "loss": 0.8046,
      "step": 6613
    },
    {
      "epoch": 2.088411746131986,
      "grad_norm": 2.4616994857788086,
      "learning_rate": 3.235211564100765e-05,
      "loss": 0.7091,
      "step": 6614
    },
    {
      "epoch": 2.088727502368172,
      "grad_norm": 1.9450907707214355,
      "learning_rate": 3.234860711529016e-05,
      "loss": 0.7026,
      "step": 6615
    },
    {
      "epoch": 2.0890432586043572,
      "grad_norm": 2.109023332595825,
      "learning_rate": 3.2345098589572666e-05,
      "loss": 0.7776,
      "step": 6616
    },
    {
      "epoch": 2.089359014840543,
      "grad_norm": 2.3477461338043213,
      "learning_rate": 3.234159006385517e-05,
      "loss": 0.7879,
      "step": 6617
    },
    {
      "epoch": 2.0896747710767287,
      "grad_norm": 2.1826770305633545,
      "learning_rate": 3.233808153813768e-05,
      "loss": 0.7242,
      "step": 6618
    },
    {
      "epoch": 2.0899905273129145,
      "grad_norm": 2.3482561111450195,
      "learning_rate": 3.2334573012420186e-05,
      "loss": 0.8173,
      "step": 6619
    },
    {
      "epoch": 2.0903062835491,
      "grad_norm": 2.559236526489258,
      "learning_rate": 3.2331064486702686e-05,
      "loss": 0.7848,
      "step": 6620
    },
    {
      "epoch": 2.090622039785286,
      "grad_norm": 2.4226417541503906,
      "learning_rate": 3.232755596098519e-05,
      "loss": 0.7063,
      "step": 6621
    },
    {
      "epoch": 2.0909377960214712,
      "grad_norm": 2.2333264350891113,
      "learning_rate": 3.23240474352677e-05,
      "loss": 0.6761,
      "step": 6622
    },
    {
      "epoch": 2.091253552257657,
      "grad_norm": 3.2846360206604004,
      "learning_rate": 3.232053890955021e-05,
      "loss": 0.7774,
      "step": 6623
    },
    {
      "epoch": 2.0915693084938427,
      "grad_norm": 4.119761943817139,
      "learning_rate": 3.2317030383832714e-05,
      "loss": 0.7788,
      "step": 6624
    },
    {
      "epoch": 2.0918850647300284,
      "grad_norm": 2.540391206741333,
      "learning_rate": 3.231352185811522e-05,
      "loss": 0.8278,
      "step": 6625
    },
    {
      "epoch": 2.092200820966214,
      "grad_norm": 1.9324605464935303,
      "learning_rate": 3.231001333239773e-05,
      "loss": 0.6958,
      "step": 6626
    },
    {
      "epoch": 2.0925165772024,
      "grad_norm": 2.2033236026763916,
      "learning_rate": 3.2306504806680234e-05,
      "loss": 0.8678,
      "step": 6627
    },
    {
      "epoch": 2.092832333438585,
      "grad_norm": 2.2796459197998047,
      "learning_rate": 3.230299628096274e-05,
      "loss": 0.7226,
      "step": 6628
    },
    {
      "epoch": 2.093148089674771,
      "grad_norm": 1.8355286121368408,
      "learning_rate": 3.229948775524525e-05,
      "loss": 0.7447,
      "step": 6629
    },
    {
      "epoch": 2.0934638459109567,
      "grad_norm": 1.8519748449325562,
      "learning_rate": 3.2295979229527754e-05,
      "loss": 0.6507,
      "step": 6630
    },
    {
      "epoch": 2.0937796021471424,
      "grad_norm": 2.008453369140625,
      "learning_rate": 3.229247070381026e-05,
      "loss": 0.7081,
      "step": 6631
    },
    {
      "epoch": 2.094095358383328,
      "grad_norm": 2.132897138595581,
      "learning_rate": 3.228896217809277e-05,
      "loss": 0.7647,
      "step": 6632
    },
    {
      "epoch": 2.094411114619514,
      "grad_norm": 1.9708654880523682,
      "learning_rate": 3.2285453652375275e-05,
      "loss": 0.8568,
      "step": 6633
    },
    {
      "epoch": 2.094726870855699,
      "grad_norm": 2.253495454788208,
      "learning_rate": 3.228194512665778e-05,
      "loss": 0.7502,
      "step": 6634
    },
    {
      "epoch": 2.095042627091885,
      "grad_norm": 1.9862525463104248,
      "learning_rate": 3.227843660094029e-05,
      "loss": 0.7646,
      "step": 6635
    },
    {
      "epoch": 2.0953583833280707,
      "grad_norm": 2.471459150314331,
      "learning_rate": 3.2274928075222795e-05,
      "loss": 0.7596,
      "step": 6636
    },
    {
      "epoch": 2.0956741395642564,
      "grad_norm": 2.0238120555877686,
      "learning_rate": 3.22714195495053e-05,
      "loss": 0.7571,
      "step": 6637
    },
    {
      "epoch": 2.095989895800442,
      "grad_norm": 2.119464874267578,
      "learning_rate": 3.226791102378781e-05,
      "loss": 0.7678,
      "step": 6638
    },
    {
      "epoch": 2.096305652036628,
      "grad_norm": 2.2442867755889893,
      "learning_rate": 3.226440249807031e-05,
      "loss": 0.7325,
      "step": 6639
    },
    {
      "epoch": 2.096621408272813,
      "grad_norm": 2.3837075233459473,
      "learning_rate": 3.2260893972352816e-05,
      "loss": 0.764,
      "step": 6640
    },
    {
      "epoch": 2.096937164508999,
      "grad_norm": 2.0504939556121826,
      "learning_rate": 3.225738544663532e-05,
      "loss": 0.7538,
      "step": 6641
    },
    {
      "epoch": 2.0972529207451847,
      "grad_norm": 2.1267056465148926,
      "learning_rate": 3.225387692091783e-05,
      "loss": 0.8048,
      "step": 6642
    },
    {
      "epoch": 2.0975686769813704,
      "grad_norm": 2.260951280593872,
      "learning_rate": 3.2250368395200336e-05,
      "loss": 0.7235,
      "step": 6643
    },
    {
      "epoch": 2.097884433217556,
      "grad_norm": 2.212188243865967,
      "learning_rate": 3.224685986948284e-05,
      "loss": 0.704,
      "step": 6644
    },
    {
      "epoch": 2.098200189453742,
      "grad_norm": 1.8566049337387085,
      "learning_rate": 3.224335134376535e-05,
      "loss": 0.7806,
      "step": 6645
    },
    {
      "epoch": 2.098515945689927,
      "grad_norm": 2.399444341659546,
      "learning_rate": 3.223984281804786e-05,
      "loss": 0.7613,
      "step": 6646
    },
    {
      "epoch": 2.098831701926113,
      "grad_norm": 2.0126001834869385,
      "learning_rate": 3.2236334292330363e-05,
      "loss": 0.7676,
      "step": 6647
    },
    {
      "epoch": 2.0991474581622986,
      "grad_norm": 2.2911036014556885,
      "learning_rate": 3.223282576661287e-05,
      "loss": 0.7504,
      "step": 6648
    },
    {
      "epoch": 2.0994632143984844,
      "grad_norm": 2.2976768016815186,
      "learning_rate": 3.222931724089538e-05,
      "loss": 0.6736,
      "step": 6649
    },
    {
      "epoch": 2.09977897063467,
      "grad_norm": 2.1218976974487305,
      "learning_rate": 3.2225808715177884e-05,
      "loss": 0.78,
      "step": 6650
    },
    {
      "epoch": 2.100094726870856,
      "grad_norm": 2.0603344440460205,
      "learning_rate": 3.222230018946039e-05,
      "loss": 0.7309,
      "step": 6651
    },
    {
      "epoch": 2.1004104831070416,
      "grad_norm": 2.337451219558716,
      "learning_rate": 3.22187916637429e-05,
      "loss": 0.7605,
      "step": 6652
    },
    {
      "epoch": 2.100726239343227,
      "grad_norm": 2.122586250305176,
      "learning_rate": 3.2215283138025404e-05,
      "loss": 0.7882,
      "step": 6653
    },
    {
      "epoch": 2.1010419955794126,
      "grad_norm": 2.134432077407837,
      "learning_rate": 3.221177461230791e-05,
      "loss": 0.7355,
      "step": 6654
    },
    {
      "epoch": 2.1013577518155984,
      "grad_norm": 2.206804037094116,
      "learning_rate": 3.220826608659042e-05,
      "loss": 0.7478,
      "step": 6655
    },
    {
      "epoch": 2.101673508051784,
      "grad_norm": 2.5055766105651855,
      "learning_rate": 3.2204757560872925e-05,
      "loss": 0.762,
      "step": 6656
    },
    {
      "epoch": 2.10198926428797,
      "grad_norm": 2.4882583618164062,
      "learning_rate": 3.220124903515543e-05,
      "loss": 0.8167,
      "step": 6657
    },
    {
      "epoch": 2.102305020524155,
      "grad_norm": 2.1706702709198,
      "learning_rate": 3.219774050943793e-05,
      "loss": 0.7274,
      "step": 6658
    },
    {
      "epoch": 2.102620776760341,
      "grad_norm": 2.2044060230255127,
      "learning_rate": 3.219423198372044e-05,
      "loss": 0.7698,
      "step": 6659
    },
    {
      "epoch": 2.1029365329965266,
      "grad_norm": 2.76092267036438,
      "learning_rate": 3.2190723458002945e-05,
      "loss": 0.7307,
      "step": 6660
    },
    {
      "epoch": 2.1032522892327123,
      "grad_norm": 2.174201011657715,
      "learning_rate": 3.218721493228545e-05,
      "loss": 0.6941,
      "step": 6661
    },
    {
      "epoch": 2.103568045468898,
      "grad_norm": 2.209679126739502,
      "learning_rate": 3.218370640656796e-05,
      "loss": 0.6774,
      "step": 6662
    },
    {
      "epoch": 2.103883801705084,
      "grad_norm": 1.9446032047271729,
      "learning_rate": 3.2180197880850466e-05,
      "loss": 0.7053,
      "step": 6663
    },
    {
      "epoch": 2.1041995579412696,
      "grad_norm": 2.1434407234191895,
      "learning_rate": 3.217668935513298e-05,
      "loss": 0.7349,
      "step": 6664
    },
    {
      "epoch": 2.104515314177455,
      "grad_norm": 2.3941092491149902,
      "learning_rate": 3.2173180829415486e-05,
      "loss": 0.7164,
      "step": 6665
    },
    {
      "epoch": 2.1048310704136406,
      "grad_norm": 2.3618533611297607,
      "learning_rate": 3.216967230369799e-05,
      "loss": 0.7517,
      "step": 6666
    },
    {
      "epoch": 2.1051468266498263,
      "grad_norm": 2.2143232822418213,
      "learning_rate": 3.216616377798049e-05,
      "loss": 0.7478,
      "step": 6667
    },
    {
      "epoch": 2.105462582886012,
      "grad_norm": 2.3214919567108154,
      "learning_rate": 3.2162655252263e-05,
      "loss": 0.765,
      "step": 6668
    },
    {
      "epoch": 2.105778339122198,
      "grad_norm": 2.1871297359466553,
      "learning_rate": 3.2159146726545507e-05,
      "loss": 0.6756,
      "step": 6669
    },
    {
      "epoch": 2.106094095358383,
      "grad_norm": 2.106444835662842,
      "learning_rate": 3.2155638200828013e-05,
      "loss": 0.7266,
      "step": 6670
    },
    {
      "epoch": 2.106409851594569,
      "grad_norm": 2.186549425125122,
      "learning_rate": 3.215212967511052e-05,
      "loss": 0.7553,
      "step": 6671
    },
    {
      "epoch": 2.1067256078307546,
      "grad_norm": 2.1746933460235596,
      "learning_rate": 3.214862114939303e-05,
      "loss": 0.7961,
      "step": 6672
    },
    {
      "epoch": 2.1070413640669403,
      "grad_norm": 2.2364883422851562,
      "learning_rate": 3.2145112623675534e-05,
      "loss": 0.7773,
      "step": 6673
    },
    {
      "epoch": 2.107357120303126,
      "grad_norm": 2.2244014739990234,
      "learning_rate": 3.214160409795804e-05,
      "loss": 0.7429,
      "step": 6674
    },
    {
      "epoch": 2.107672876539312,
      "grad_norm": 2.1671805381774902,
      "learning_rate": 3.213809557224055e-05,
      "loss": 0.6736,
      "step": 6675
    },
    {
      "epoch": 2.1079886327754975,
      "grad_norm": 2.068904161453247,
      "learning_rate": 3.2134587046523054e-05,
      "loss": 0.7393,
      "step": 6676
    },
    {
      "epoch": 2.108304389011683,
      "grad_norm": 2.334487199783325,
      "learning_rate": 3.2131078520805554e-05,
      "loss": 0.8331,
      "step": 6677
    },
    {
      "epoch": 2.1086201452478686,
      "grad_norm": 2.003533124923706,
      "learning_rate": 3.212756999508806e-05,
      "loss": 0.6909,
      "step": 6678
    },
    {
      "epoch": 2.1089359014840543,
      "grad_norm": 2.2931978702545166,
      "learning_rate": 3.212406146937057e-05,
      "loss": 0.842,
      "step": 6679
    },
    {
      "epoch": 2.10925165772024,
      "grad_norm": 2.3892064094543457,
      "learning_rate": 3.2120552943653075e-05,
      "loss": 0.8252,
      "step": 6680
    },
    {
      "epoch": 2.1095674139564258,
      "grad_norm": 2.152752161026001,
      "learning_rate": 3.211704441793558e-05,
      "loss": 0.6458,
      "step": 6681
    },
    {
      "epoch": 2.1098831701926115,
      "grad_norm": 2.5352835655212402,
      "learning_rate": 3.2113535892218095e-05,
      "loss": 0.7863,
      "step": 6682
    },
    {
      "epoch": 2.110198926428797,
      "grad_norm": 2.2055249214172363,
      "learning_rate": 3.21100273665006e-05,
      "loss": 0.7808,
      "step": 6683
    },
    {
      "epoch": 2.1105146826649825,
      "grad_norm": 2.4185030460357666,
      "learning_rate": 3.210651884078311e-05,
      "loss": 0.6962,
      "step": 6684
    },
    {
      "epoch": 2.1108304389011683,
      "grad_norm": 1.9438625574111938,
      "learning_rate": 3.2103010315065616e-05,
      "loss": 0.7591,
      "step": 6685
    },
    {
      "epoch": 2.111146195137354,
      "grad_norm": 2.2147018909454346,
      "learning_rate": 3.2099501789348116e-05,
      "loss": 0.8066,
      "step": 6686
    },
    {
      "epoch": 2.1114619513735398,
      "grad_norm": 2.644827365875244,
      "learning_rate": 3.209599326363062e-05,
      "loss": 0.817,
      "step": 6687
    },
    {
      "epoch": 2.1117777076097255,
      "grad_norm": 2.951103448867798,
      "learning_rate": 3.209248473791313e-05,
      "loss": 0.6517,
      "step": 6688
    },
    {
      "epoch": 2.112093463845911,
      "grad_norm": 2.31569766998291,
      "learning_rate": 3.2088976212195636e-05,
      "loss": 0.7723,
      "step": 6689
    },
    {
      "epoch": 2.1124092200820965,
      "grad_norm": 2.011406660079956,
      "learning_rate": 3.208546768647814e-05,
      "loss": 0.7392,
      "step": 6690
    },
    {
      "epoch": 2.1127249763182823,
      "grad_norm": 1.8817501068115234,
      "learning_rate": 3.208195916076065e-05,
      "loss": 0.7277,
      "step": 6691
    },
    {
      "epoch": 2.113040732554468,
      "grad_norm": 2.427227735519409,
      "learning_rate": 3.2078450635043157e-05,
      "loss": 0.7541,
      "step": 6692
    },
    {
      "epoch": 2.1133564887906537,
      "grad_norm": 2.2374625205993652,
      "learning_rate": 3.207494210932566e-05,
      "loss": 0.635,
      "step": 6693
    },
    {
      "epoch": 2.1136722450268395,
      "grad_norm": 2.3208086490631104,
      "learning_rate": 3.207143358360817e-05,
      "loss": 0.7036,
      "step": 6694
    },
    {
      "epoch": 2.1139880012630248,
      "grad_norm": 2.465662717819214,
      "learning_rate": 3.206792505789068e-05,
      "loss": 0.7734,
      "step": 6695
    },
    {
      "epoch": 2.1143037574992105,
      "grad_norm": 2.05859375,
      "learning_rate": 3.2064416532173184e-05,
      "loss": 0.7248,
      "step": 6696
    },
    {
      "epoch": 2.1146195137353963,
      "grad_norm": 2.855233907699585,
      "learning_rate": 3.2060908006455684e-05,
      "loss": 0.7911,
      "step": 6697
    },
    {
      "epoch": 2.114935269971582,
      "grad_norm": 1.9937186241149902,
      "learning_rate": 3.205739948073819e-05,
      "loss": 0.6522,
      "step": 6698
    },
    {
      "epoch": 2.1152510262077677,
      "grad_norm": 2.1785531044006348,
      "learning_rate": 3.2053890955020704e-05,
      "loss": 0.7531,
      "step": 6699
    },
    {
      "epoch": 2.1155667824439535,
      "grad_norm": 2.326443910598755,
      "learning_rate": 3.205038242930321e-05,
      "loss": 0.7283,
      "step": 6700
    },
    {
      "epoch": 2.1158825386801388,
      "grad_norm": 1.9623833894729614,
      "learning_rate": 3.204687390358572e-05,
      "loss": 0.7657,
      "step": 6701
    },
    {
      "epoch": 2.1161982949163245,
      "grad_norm": 2.097893238067627,
      "learning_rate": 3.2043365377868225e-05,
      "loss": 0.7241,
      "step": 6702
    },
    {
      "epoch": 2.1165140511525102,
      "grad_norm": 2.040245532989502,
      "learning_rate": 3.203985685215073e-05,
      "loss": 0.6982,
      "step": 6703
    },
    {
      "epoch": 2.116829807388696,
      "grad_norm": 1.9357296228408813,
      "learning_rate": 3.203634832643324e-05,
      "loss": 0.7139,
      "step": 6704
    },
    {
      "epoch": 2.1171455636248817,
      "grad_norm": 1.9025906324386597,
      "learning_rate": 3.203283980071574e-05,
      "loss": 0.7292,
      "step": 6705
    },
    {
      "epoch": 2.1174613198610674,
      "grad_norm": 2.295788526535034,
      "learning_rate": 3.2029331274998245e-05,
      "loss": 0.6867,
      "step": 6706
    },
    {
      "epoch": 2.1177770760972527,
      "grad_norm": 2.060145616531372,
      "learning_rate": 3.202582274928075e-05,
      "loss": 0.7352,
      "step": 6707
    },
    {
      "epoch": 2.1180928323334385,
      "grad_norm": 2.1890928745269775,
      "learning_rate": 3.202231422356326e-05,
      "loss": 0.7294,
      "step": 6708
    },
    {
      "epoch": 2.118408588569624,
      "grad_norm": 2.1516523361206055,
      "learning_rate": 3.2018805697845766e-05,
      "loss": 0.7491,
      "step": 6709
    },
    {
      "epoch": 2.11872434480581,
      "grad_norm": 1.8783628940582275,
      "learning_rate": 3.201529717212827e-05,
      "loss": 0.6799,
      "step": 6710
    },
    {
      "epoch": 2.1190401010419957,
      "grad_norm": 2.2161641120910645,
      "learning_rate": 3.201178864641078e-05,
      "loss": 0.7159,
      "step": 6711
    },
    {
      "epoch": 2.1193558572781814,
      "grad_norm": 2.025261402130127,
      "learning_rate": 3.2008280120693286e-05,
      "loss": 0.7446,
      "step": 6712
    },
    {
      "epoch": 2.1196716135143667,
      "grad_norm": 2.1067612171173096,
      "learning_rate": 3.200477159497579e-05,
      "loss": 0.7821,
      "step": 6713
    },
    {
      "epoch": 2.1199873697505525,
      "grad_norm": 2.2117624282836914,
      "learning_rate": 3.20012630692583e-05,
      "loss": 0.7289,
      "step": 6714
    },
    {
      "epoch": 2.120303125986738,
      "grad_norm": 1.9223852157592773,
      "learning_rate": 3.1997754543540806e-05,
      "loss": 0.7098,
      "step": 6715
    },
    {
      "epoch": 2.120618882222924,
      "grad_norm": 2.2396163940429688,
      "learning_rate": 3.1994246017823306e-05,
      "loss": 0.7442,
      "step": 6716
    },
    {
      "epoch": 2.1209346384591097,
      "grad_norm": 2.4525556564331055,
      "learning_rate": 3.199073749210582e-05,
      "loss": 0.7508,
      "step": 6717
    },
    {
      "epoch": 2.1212503946952954,
      "grad_norm": 2.2112276554107666,
      "learning_rate": 3.198722896638833e-05,
      "loss": 0.704,
      "step": 6718
    },
    {
      "epoch": 2.1215661509314807,
      "grad_norm": 2.0044429302215576,
      "learning_rate": 3.1983720440670834e-05,
      "loss": 0.6681,
      "step": 6719
    },
    {
      "epoch": 2.1218819071676664,
      "grad_norm": 2.0736711025238037,
      "learning_rate": 3.198021191495334e-05,
      "loss": 0.7487,
      "step": 6720
    },
    {
      "epoch": 2.122197663403852,
      "grad_norm": 2.2804269790649414,
      "learning_rate": 3.197670338923585e-05,
      "loss": 0.8178,
      "step": 6721
    },
    {
      "epoch": 2.122513419640038,
      "grad_norm": 2.278089761734009,
      "learning_rate": 3.1973194863518354e-05,
      "loss": 0.7696,
      "step": 6722
    },
    {
      "epoch": 2.1228291758762237,
      "grad_norm": 2.9240622520446777,
      "learning_rate": 3.196968633780086e-05,
      "loss": 0.6674,
      "step": 6723
    },
    {
      "epoch": 2.1231449321124094,
      "grad_norm": 2.296034574508667,
      "learning_rate": 3.196617781208337e-05,
      "loss": 0.7114,
      "step": 6724
    },
    {
      "epoch": 2.1234606883485947,
      "grad_norm": 2.5518460273742676,
      "learning_rate": 3.196266928636587e-05,
      "loss": 0.7637,
      "step": 6725
    },
    {
      "epoch": 2.1237764445847804,
      "grad_norm": 2.601506233215332,
      "learning_rate": 3.1959160760648375e-05,
      "loss": 0.8092,
      "step": 6726
    },
    {
      "epoch": 2.124092200820966,
      "grad_norm": 2.4227237701416016,
      "learning_rate": 3.195565223493088e-05,
      "loss": 0.7178,
      "step": 6727
    },
    {
      "epoch": 2.124407957057152,
      "grad_norm": 2.310181140899658,
      "learning_rate": 3.195214370921339e-05,
      "loss": 0.7489,
      "step": 6728
    },
    {
      "epoch": 2.1247237132933376,
      "grad_norm": 2.4962263107299805,
      "learning_rate": 3.1948635183495895e-05,
      "loss": 0.763,
      "step": 6729
    },
    {
      "epoch": 2.1250394695295234,
      "grad_norm": 2.6735711097717285,
      "learning_rate": 3.19451266577784e-05,
      "loss": 0.799,
      "step": 6730
    },
    {
      "epoch": 2.1253552257657087,
      "grad_norm": 2.4510834217071533,
      "learning_rate": 3.194161813206091e-05,
      "loss": 0.82,
      "step": 6731
    },
    {
      "epoch": 2.1256709820018944,
      "grad_norm": 2.046588659286499,
      "learning_rate": 3.1938109606343415e-05,
      "loss": 0.7395,
      "step": 6732
    },
    {
      "epoch": 2.12598673823808,
      "grad_norm": 8.734180450439453,
      "learning_rate": 3.193460108062592e-05,
      "loss": 0.751,
      "step": 6733
    },
    {
      "epoch": 2.126302494474266,
      "grad_norm": 2.026726722717285,
      "learning_rate": 3.193109255490843e-05,
      "loss": 0.7027,
      "step": 6734
    },
    {
      "epoch": 2.1266182507104516,
      "grad_norm": 2.338193416595459,
      "learning_rate": 3.1927584029190936e-05,
      "loss": 0.7002,
      "step": 6735
    },
    {
      "epoch": 2.1269340069466374,
      "grad_norm": 2.2358694076538086,
      "learning_rate": 3.192407550347344e-05,
      "loss": 0.7986,
      "step": 6736
    },
    {
      "epoch": 2.1272497631828227,
      "grad_norm": 2.3613362312316895,
      "learning_rate": 3.192056697775595e-05,
      "loss": 0.7586,
      "step": 6737
    },
    {
      "epoch": 2.1275655194190084,
      "grad_norm": 2.2941699028015137,
      "learning_rate": 3.1917058452038456e-05,
      "loss": 0.7031,
      "step": 6738
    },
    {
      "epoch": 2.127881275655194,
      "grad_norm": 2.288527727127075,
      "learning_rate": 3.191354992632096e-05,
      "loss": 0.7278,
      "step": 6739
    },
    {
      "epoch": 2.12819703189138,
      "grad_norm": 2.177370548248291,
      "learning_rate": 3.191004140060347e-05,
      "loss": 0.7113,
      "step": 6740
    },
    {
      "epoch": 2.1285127881275656,
      "grad_norm": 2.205658435821533,
      "learning_rate": 3.190653287488598e-05,
      "loss": 0.7555,
      "step": 6741
    },
    {
      "epoch": 2.1288285443637514,
      "grad_norm": 2.2705488204956055,
      "learning_rate": 3.1903024349168484e-05,
      "loss": 0.839,
      "step": 6742
    },
    {
      "epoch": 2.1291443005999366,
      "grad_norm": 2.384528160095215,
      "learning_rate": 3.189951582345099e-05,
      "loss": 0.7025,
      "step": 6743
    },
    {
      "epoch": 2.1294600568361224,
      "grad_norm": 2.3578848838806152,
      "learning_rate": 3.189600729773349e-05,
      "loss": 0.6854,
      "step": 6744
    },
    {
      "epoch": 2.129775813072308,
      "grad_norm": 2.239368438720703,
      "learning_rate": 3.1892498772016e-05,
      "loss": 0.734,
      "step": 6745
    },
    {
      "epoch": 2.130091569308494,
      "grad_norm": 2.1885311603546143,
      "learning_rate": 3.1888990246298504e-05,
      "loss": 0.7967,
      "step": 6746
    },
    {
      "epoch": 2.1304073255446796,
      "grad_norm": 2.2505903244018555,
      "learning_rate": 3.188548172058101e-05,
      "loss": 0.6948,
      "step": 6747
    },
    {
      "epoch": 2.1307230817808653,
      "grad_norm": 2.0296664237976074,
      "learning_rate": 3.188197319486352e-05,
      "loss": 0.7331,
      "step": 6748
    },
    {
      "epoch": 2.1310388380170506,
      "grad_norm": 2.6570169925689697,
      "learning_rate": 3.1878464669146025e-05,
      "loss": 0.757,
      "step": 6749
    },
    {
      "epoch": 2.1313545942532364,
      "grad_norm": 2.4091176986694336,
      "learning_rate": 3.187495614342853e-05,
      "loss": 0.7774,
      "step": 6750
    },
    {
      "epoch": 2.131670350489422,
      "grad_norm": 2.3863558769226074,
      "learning_rate": 3.187144761771104e-05,
      "loss": 0.7403,
      "step": 6751
    },
    {
      "epoch": 2.131986106725608,
      "grad_norm": 2.34324049949646,
      "learning_rate": 3.1867939091993545e-05,
      "loss": 0.7109,
      "step": 6752
    },
    {
      "epoch": 2.1323018629617936,
      "grad_norm": 2.0983612537384033,
      "learning_rate": 3.186443056627605e-05,
      "loss": 0.7797,
      "step": 6753
    },
    {
      "epoch": 2.1326176191979793,
      "grad_norm": 2.2774460315704346,
      "learning_rate": 3.186092204055856e-05,
      "loss": 0.7583,
      "step": 6754
    },
    {
      "epoch": 2.132933375434165,
      "grad_norm": 2.1234469413757324,
      "learning_rate": 3.1857413514841065e-05,
      "loss": 0.7175,
      "step": 6755
    },
    {
      "epoch": 2.1332491316703504,
      "grad_norm": 2.168590545654297,
      "learning_rate": 3.185390498912357e-05,
      "loss": 0.7617,
      "step": 6756
    },
    {
      "epoch": 2.133564887906536,
      "grad_norm": 2.0537455081939697,
      "learning_rate": 3.185039646340608e-05,
      "loss": 0.6937,
      "step": 6757
    },
    {
      "epoch": 2.133880644142722,
      "grad_norm": 2.096879243850708,
      "learning_rate": 3.1846887937688586e-05,
      "loss": 0.6424,
      "step": 6758
    },
    {
      "epoch": 2.1341964003789076,
      "grad_norm": 2.252772331237793,
      "learning_rate": 3.184337941197109e-05,
      "loss": 0.756,
      "step": 6759
    },
    {
      "epoch": 2.1345121566150933,
      "grad_norm": 2.280225992202759,
      "learning_rate": 3.18398708862536e-05,
      "loss": 0.6788,
      "step": 6760
    },
    {
      "epoch": 2.1348279128512786,
      "grad_norm": 2.3396100997924805,
      "learning_rate": 3.1836362360536106e-05,
      "loss": 0.8258,
      "step": 6761
    },
    {
      "epoch": 2.1351436690874643,
      "grad_norm": 2.2950425148010254,
      "learning_rate": 3.183285383481861e-05,
      "loss": 0.8897,
      "step": 6762
    },
    {
      "epoch": 2.13545942532365,
      "grad_norm": 2.223811388015747,
      "learning_rate": 3.182934530910111e-05,
      "loss": 0.7359,
      "step": 6763
    },
    {
      "epoch": 2.135775181559836,
      "grad_norm": 2.0260560512542725,
      "learning_rate": 3.182583678338362e-05,
      "loss": 0.8081,
      "step": 6764
    },
    {
      "epoch": 2.1360909377960215,
      "grad_norm": 2.3075947761535645,
      "learning_rate": 3.182232825766613e-05,
      "loss": 0.7282,
      "step": 6765
    },
    {
      "epoch": 2.1364066940322073,
      "grad_norm": 2.251823902130127,
      "learning_rate": 3.1818819731948634e-05,
      "loss": 0.7253,
      "step": 6766
    },
    {
      "epoch": 2.136722450268393,
      "grad_norm": 2.0623443126678467,
      "learning_rate": 3.181531120623114e-05,
      "loss": 0.7294,
      "step": 6767
    },
    {
      "epoch": 2.1370382065045783,
      "grad_norm": 2.113076686859131,
      "learning_rate": 3.181180268051365e-05,
      "loss": 0.7374,
      "step": 6768
    },
    {
      "epoch": 2.137353962740764,
      "grad_norm": 2.1410305500030518,
      "learning_rate": 3.180829415479616e-05,
      "loss": 0.8654,
      "step": 6769
    },
    {
      "epoch": 2.13766971897695,
      "grad_norm": 2.1354429721832275,
      "learning_rate": 3.180478562907867e-05,
      "loss": 0.7116,
      "step": 6770
    },
    {
      "epoch": 2.1379854752131355,
      "grad_norm": 2.1806554794311523,
      "learning_rate": 3.1801277103361174e-05,
      "loss": 0.7362,
      "step": 6771
    },
    {
      "epoch": 2.1383012314493213,
      "grad_norm": 2.3492307662963867,
      "learning_rate": 3.1797768577643674e-05,
      "loss": 0.6724,
      "step": 6772
    },
    {
      "epoch": 2.1386169876855066,
      "grad_norm": 2.096997022628784,
      "learning_rate": 3.179426005192618e-05,
      "loss": 0.7573,
      "step": 6773
    },
    {
      "epoch": 2.1389327439216923,
      "grad_norm": 2.232541799545288,
      "learning_rate": 3.179075152620869e-05,
      "loss": 0.7568,
      "step": 6774
    },
    {
      "epoch": 2.139248500157878,
      "grad_norm": 2.0776607990264893,
      "learning_rate": 3.1787243000491195e-05,
      "loss": 0.7779,
      "step": 6775
    },
    {
      "epoch": 2.139564256394064,
      "grad_norm": 2.64874529838562,
      "learning_rate": 3.17837344747737e-05,
      "loss": 0.8104,
      "step": 6776
    },
    {
      "epoch": 2.1398800126302495,
      "grad_norm": 2.4701123237609863,
      "learning_rate": 3.178022594905621e-05,
      "loss": 0.8485,
      "step": 6777
    },
    {
      "epoch": 2.1401957688664353,
      "grad_norm": 2.3568387031555176,
      "learning_rate": 3.1776717423338715e-05,
      "loss": 0.7434,
      "step": 6778
    },
    {
      "epoch": 2.140511525102621,
      "grad_norm": 2.219266176223755,
      "learning_rate": 3.177320889762122e-05,
      "loss": 0.766,
      "step": 6779
    },
    {
      "epoch": 2.1408272813388063,
      "grad_norm": 2.5131964683532715,
      "learning_rate": 3.176970037190373e-05,
      "loss": 0.695,
      "step": 6780
    },
    {
      "epoch": 2.141143037574992,
      "grad_norm": 2.374833822250366,
      "learning_rate": 3.1766191846186236e-05,
      "loss": 0.7359,
      "step": 6781
    },
    {
      "epoch": 2.1414587938111778,
      "grad_norm": 2.48514461517334,
      "learning_rate": 3.1762683320468736e-05,
      "loss": 0.7355,
      "step": 6782
    },
    {
      "epoch": 2.1417745500473635,
      "grad_norm": 1.9903830289840698,
      "learning_rate": 3.175917479475124e-05,
      "loss": 0.7253,
      "step": 6783
    },
    {
      "epoch": 2.1420903062835492,
      "grad_norm": 2.3608052730560303,
      "learning_rate": 3.175566626903375e-05,
      "loss": 0.7734,
      "step": 6784
    },
    {
      "epoch": 2.1424060625197345,
      "grad_norm": 2.1928398609161377,
      "learning_rate": 3.1752157743316256e-05,
      "loss": 0.6961,
      "step": 6785
    },
    {
      "epoch": 2.1427218187559203,
      "grad_norm": 2.636944532394409,
      "learning_rate": 3.174864921759876e-05,
      "loss": 0.7646,
      "step": 6786
    },
    {
      "epoch": 2.143037574992106,
      "grad_norm": 2.1086061000823975,
      "learning_rate": 3.174514069188128e-05,
      "loss": 0.783,
      "step": 6787
    },
    {
      "epoch": 2.1433533312282917,
      "grad_norm": 2.2189481258392334,
      "learning_rate": 3.1741632166163783e-05,
      "loss": 0.794,
      "step": 6788
    },
    {
      "epoch": 2.1436690874644775,
      "grad_norm": 1.875689148902893,
      "learning_rate": 3.173812364044629e-05,
      "loss": 0.6846,
      "step": 6789
    },
    {
      "epoch": 2.1439848437006632,
      "grad_norm": 2.086249589920044,
      "learning_rate": 3.17346151147288e-05,
      "loss": 0.6852,
      "step": 6790
    },
    {
      "epoch": 2.144300599936849,
      "grad_norm": 2.0313150882720947,
      "learning_rate": 3.17311065890113e-05,
      "loss": 0.7336,
      "step": 6791
    },
    {
      "epoch": 2.1446163561730343,
      "grad_norm": 2.315901279449463,
      "learning_rate": 3.1727598063293804e-05,
      "loss": 0.8142,
      "step": 6792
    },
    {
      "epoch": 2.14493211240922,
      "grad_norm": 2.0185859203338623,
      "learning_rate": 3.172408953757631e-05,
      "loss": 0.7826,
      "step": 6793
    },
    {
      "epoch": 2.1452478686454057,
      "grad_norm": 2.1442131996154785,
      "learning_rate": 3.172058101185882e-05,
      "loss": 0.8047,
      "step": 6794
    },
    {
      "epoch": 2.1455636248815915,
      "grad_norm": 2.0098421573638916,
      "learning_rate": 3.1717072486141324e-05,
      "loss": 0.693,
      "step": 6795
    },
    {
      "epoch": 2.145879381117777,
      "grad_norm": 2.7889204025268555,
      "learning_rate": 3.171356396042383e-05,
      "loss": 0.7925,
      "step": 6796
    },
    {
      "epoch": 2.146195137353963,
      "grad_norm": 1.9726636409759521,
      "learning_rate": 3.171005543470634e-05,
      "loss": 0.6881,
      "step": 6797
    },
    {
      "epoch": 2.1465108935901482,
      "grad_norm": 2.1095829010009766,
      "learning_rate": 3.1706546908988845e-05,
      "loss": 0.8014,
      "step": 6798
    },
    {
      "epoch": 2.146826649826334,
      "grad_norm": 2.6707170009613037,
      "learning_rate": 3.170303838327135e-05,
      "loss": 0.757,
      "step": 6799
    },
    {
      "epoch": 2.1471424060625197,
      "grad_norm": 2.255384683609009,
      "learning_rate": 3.169952985755386e-05,
      "loss": 0.753,
      "step": 6800
    },
    {
      "epoch": 2.1471424060625197,
      "eval_loss": 0.7968311905860901,
      "eval_runtime": 4.6581,
      "eval_samples_per_second": 534.55,
      "eval_steps_per_second": 4.508,
      "step": 6800
    },
    {
      "epoch": 2.1474581622987055,
      "grad_norm": 2.957594156265259,
      "learning_rate": 3.1696021331836365e-05,
      "loss": 0.8007,
      "step": 6801
    },
    {
      "epoch": 2.147773918534891,
      "grad_norm": 2.2774479389190674,
      "learning_rate": 3.1692512806118865e-05,
      "loss": 0.8126,
      "step": 6802
    },
    {
      "epoch": 2.148089674771077,
      "grad_norm": 2.3047308921813965,
      "learning_rate": 3.168900428040137e-05,
      "loss": 0.785,
      "step": 6803
    },
    {
      "epoch": 2.1484054310072622,
      "grad_norm": 2.1418721675872803,
      "learning_rate": 3.1685495754683886e-05,
      "loss": 0.768,
      "step": 6804
    },
    {
      "epoch": 2.148721187243448,
      "grad_norm": 2.1386265754699707,
      "learning_rate": 3.168198722896639e-05,
      "loss": 0.7355,
      "step": 6805
    },
    {
      "epoch": 2.1490369434796337,
      "grad_norm": 2.3132195472717285,
      "learning_rate": 3.16784787032489e-05,
      "loss": 0.8113,
      "step": 6806
    },
    {
      "epoch": 2.1493526997158194,
      "grad_norm": 2.304617404937744,
      "learning_rate": 3.1674970177531406e-05,
      "loss": 0.6522,
      "step": 6807
    },
    {
      "epoch": 2.149668455952005,
      "grad_norm": 2.4355862140655518,
      "learning_rate": 3.167146165181391e-05,
      "loss": 0.7601,
      "step": 6808
    },
    {
      "epoch": 2.149984212188191,
      "grad_norm": 2.5703375339508057,
      "learning_rate": 3.166795312609642e-05,
      "loss": 0.7039,
      "step": 6809
    },
    {
      "epoch": 2.150299968424376,
      "grad_norm": 1.9978405237197876,
      "learning_rate": 3.166444460037892e-05,
      "loss": 0.8009,
      "step": 6810
    },
    {
      "epoch": 2.150615724660562,
      "grad_norm": 2.2062342166900635,
      "learning_rate": 3.1660936074661427e-05,
      "loss": 0.8071,
      "step": 6811
    },
    {
      "epoch": 2.1509314808967477,
      "grad_norm": 2.3229968547821045,
      "learning_rate": 3.1657427548943933e-05,
      "loss": 0.7451,
      "step": 6812
    },
    {
      "epoch": 2.1512472371329334,
      "grad_norm": 2.581404685974121,
      "learning_rate": 3.165391902322644e-05,
      "loss": 0.7512,
      "step": 6813
    },
    {
      "epoch": 2.151562993369119,
      "grad_norm": 2.3853759765625,
      "learning_rate": 3.165041049750895e-05,
      "loss": 0.7198,
      "step": 6814
    },
    {
      "epoch": 2.151878749605305,
      "grad_norm": 2.384294271469116,
      "learning_rate": 3.1646901971791454e-05,
      "loss": 0.77,
      "step": 6815
    },
    {
      "epoch": 2.15219450584149,
      "grad_norm": 2.107637405395508,
      "learning_rate": 3.164339344607396e-05,
      "loss": 0.7165,
      "step": 6816
    },
    {
      "epoch": 2.152510262077676,
      "grad_norm": 2.6272058486938477,
      "learning_rate": 3.163988492035647e-05,
      "loss": 0.7719,
      "step": 6817
    },
    {
      "epoch": 2.1528260183138617,
      "grad_norm": 2.0652401447296143,
      "learning_rate": 3.1636376394638974e-05,
      "loss": 0.8165,
      "step": 6818
    },
    {
      "epoch": 2.1531417745500474,
      "grad_norm": 2.297610282897949,
      "learning_rate": 3.163286786892148e-05,
      "loss": 0.7347,
      "step": 6819
    },
    {
      "epoch": 2.153457530786233,
      "grad_norm": 2.146357297897339,
      "learning_rate": 3.162935934320399e-05,
      "loss": 0.7619,
      "step": 6820
    },
    {
      "epoch": 2.153773287022419,
      "grad_norm": 1.901348352432251,
      "learning_rate": 3.162585081748649e-05,
      "loss": 0.6986,
      "step": 6821
    },
    {
      "epoch": 2.154089043258604,
      "grad_norm": 2.0477712154388428,
      "learning_rate": 3.1622342291769e-05,
      "loss": 0.6769,
      "step": 6822
    },
    {
      "epoch": 2.15440479949479,
      "grad_norm": 2.2920305728912354,
      "learning_rate": 3.161883376605151e-05,
      "loss": 0.7142,
      "step": 6823
    },
    {
      "epoch": 2.1547205557309757,
      "grad_norm": 2.2461605072021484,
      "learning_rate": 3.1615325240334015e-05,
      "loss": 0.8531,
      "step": 6824
    },
    {
      "epoch": 2.1550363119671614,
      "grad_norm": 2.3132283687591553,
      "learning_rate": 3.161181671461652e-05,
      "loss": 0.8353,
      "step": 6825
    },
    {
      "epoch": 2.155352068203347,
      "grad_norm": 2.0284183025360107,
      "learning_rate": 3.160830818889903e-05,
      "loss": 0.7212,
      "step": 6826
    },
    {
      "epoch": 2.155667824439533,
      "grad_norm": 2.4074413776397705,
      "learning_rate": 3.1604799663181536e-05,
      "loss": 0.6749,
      "step": 6827
    },
    {
      "epoch": 2.155983580675718,
      "grad_norm": 2.2629809379577637,
      "learning_rate": 3.160129113746404e-05,
      "loss": 0.7881,
      "step": 6828
    },
    {
      "epoch": 2.156299336911904,
      "grad_norm": 2.0158896446228027,
      "learning_rate": 3.159778261174655e-05,
      "loss": 0.7734,
      "step": 6829
    },
    {
      "epoch": 2.1566150931480896,
      "grad_norm": 1.935248613357544,
      "learning_rate": 3.159427408602905e-05,
      "loss": 0.644,
      "step": 6830
    },
    {
      "epoch": 2.1569308493842754,
      "grad_norm": 1.8602139949798584,
      "learning_rate": 3.1590765560311556e-05,
      "loss": 0.7027,
      "step": 6831
    },
    {
      "epoch": 2.157246605620461,
      "grad_norm": 2.281571626663208,
      "learning_rate": 3.158725703459406e-05,
      "loss": 0.7344,
      "step": 6832
    },
    {
      "epoch": 2.157562361856647,
      "grad_norm": 2.3627426624298096,
      "learning_rate": 3.158374850887657e-05,
      "loss": 0.695,
      "step": 6833
    },
    {
      "epoch": 2.157878118092832,
      "grad_norm": 2.24454927444458,
      "learning_rate": 3.1580239983159077e-05,
      "loss": 0.7971,
      "step": 6834
    },
    {
      "epoch": 2.158193874329018,
      "grad_norm": 2.3292007446289062,
      "learning_rate": 3.157673145744158e-05,
      "loss": 0.7772,
      "step": 6835
    },
    {
      "epoch": 2.1585096305652036,
      "grad_norm": 2.6023976802825928,
      "learning_rate": 3.157322293172409e-05,
      "loss": 0.8707,
      "step": 6836
    },
    {
      "epoch": 2.1588253868013894,
      "grad_norm": 2.3543670177459717,
      "learning_rate": 3.15697144060066e-05,
      "loss": 0.8274,
      "step": 6837
    },
    {
      "epoch": 2.159141143037575,
      "grad_norm": 2.2895290851593018,
      "learning_rate": 3.1566205880289104e-05,
      "loss": 0.7837,
      "step": 6838
    },
    {
      "epoch": 2.159456899273761,
      "grad_norm": 2.161741018295288,
      "learning_rate": 3.156269735457161e-05,
      "loss": 0.7199,
      "step": 6839
    },
    {
      "epoch": 2.159772655509946,
      "grad_norm": 2.450916290283203,
      "learning_rate": 3.155918882885412e-05,
      "loss": 0.7461,
      "step": 6840
    },
    {
      "epoch": 2.160088411746132,
      "grad_norm": 2.0598714351654053,
      "learning_rate": 3.1555680303136624e-05,
      "loss": 0.6779,
      "step": 6841
    },
    {
      "epoch": 2.1604041679823176,
      "grad_norm": 2.376969575881958,
      "learning_rate": 3.155217177741913e-05,
      "loss": 0.8265,
      "step": 6842
    },
    {
      "epoch": 2.1607199242185033,
      "grad_norm": 2.177891731262207,
      "learning_rate": 3.154866325170164e-05,
      "loss": 0.7833,
      "step": 6843
    },
    {
      "epoch": 2.161035680454689,
      "grad_norm": 2.635547399520874,
      "learning_rate": 3.1545154725984145e-05,
      "loss": 0.722,
      "step": 6844
    },
    {
      "epoch": 2.161351436690875,
      "grad_norm": 2.4670073986053467,
      "learning_rate": 3.154164620026665e-05,
      "loss": 0.8138,
      "step": 6845
    },
    {
      "epoch": 2.16166719292706,
      "grad_norm": 1.8363803625106812,
      "learning_rate": 3.153813767454916e-05,
      "loss": 0.6743,
      "step": 6846
    },
    {
      "epoch": 2.161982949163246,
      "grad_norm": 2.29772686958313,
      "learning_rate": 3.1534629148831665e-05,
      "loss": 0.6785,
      "step": 6847
    },
    {
      "epoch": 2.1622987053994316,
      "grad_norm": 2.3353123664855957,
      "learning_rate": 3.153112062311417e-05,
      "loss": 0.797,
      "step": 6848
    },
    {
      "epoch": 2.1626144616356173,
      "grad_norm": 2.1983542442321777,
      "learning_rate": 3.152761209739667e-05,
      "loss": 0.7534,
      "step": 6849
    },
    {
      "epoch": 2.162930217871803,
      "grad_norm": 1.8975646495819092,
      "learning_rate": 3.152410357167918e-05,
      "loss": 0.6698,
      "step": 6850
    },
    {
      "epoch": 2.163245974107989,
      "grad_norm": 2.068563938140869,
      "learning_rate": 3.1520595045961686e-05,
      "loss": 0.7305,
      "step": 6851
    },
    {
      "epoch": 2.163561730344174,
      "grad_norm": 2.348878860473633,
      "learning_rate": 3.151708652024419e-05,
      "loss": 0.8405,
      "step": 6852
    },
    {
      "epoch": 2.16387748658036,
      "grad_norm": 2.199399471282959,
      "learning_rate": 3.15135779945267e-05,
      "loss": 0.7456,
      "step": 6853
    },
    {
      "epoch": 2.1641932428165456,
      "grad_norm": 2.079941749572754,
      "learning_rate": 3.1510069468809206e-05,
      "loss": 0.7065,
      "step": 6854
    },
    {
      "epoch": 2.1645089990527313,
      "grad_norm": 2.1534173488616943,
      "learning_rate": 3.150656094309171e-05,
      "loss": 0.777,
      "step": 6855
    },
    {
      "epoch": 2.164824755288917,
      "grad_norm": 2.0500667095184326,
      "learning_rate": 3.1503052417374226e-05,
      "loss": 0.746,
      "step": 6856
    },
    {
      "epoch": 2.165140511525103,
      "grad_norm": 2.292743444442749,
      "learning_rate": 3.1499543891656726e-05,
      "loss": 0.7078,
      "step": 6857
    },
    {
      "epoch": 2.165456267761288,
      "grad_norm": 2.0364232063293457,
      "learning_rate": 3.149603536593923e-05,
      "loss": 0.7237,
      "step": 6858
    },
    {
      "epoch": 2.165772023997474,
      "grad_norm": 2.129047155380249,
      "learning_rate": 3.149252684022174e-05,
      "loss": 0.657,
      "step": 6859
    },
    {
      "epoch": 2.1660877802336596,
      "grad_norm": 2.0122857093811035,
      "learning_rate": 3.148901831450425e-05,
      "loss": 0.6892,
      "step": 6860
    },
    {
      "epoch": 2.1664035364698453,
      "grad_norm": 1.9959759712219238,
      "learning_rate": 3.1485509788786754e-05,
      "loss": 0.7109,
      "step": 6861
    },
    {
      "epoch": 2.166719292706031,
      "grad_norm": 2.2197518348693848,
      "learning_rate": 3.148200126306926e-05,
      "loss": 0.7006,
      "step": 6862
    },
    {
      "epoch": 2.1670350489422168,
      "grad_norm": 2.091059923171997,
      "learning_rate": 3.147849273735177e-05,
      "loss": 0.7517,
      "step": 6863
    },
    {
      "epoch": 2.167350805178402,
      "grad_norm": 2.285926342010498,
      "learning_rate": 3.1474984211634274e-05,
      "loss": 0.7692,
      "step": 6864
    },
    {
      "epoch": 2.167666561414588,
      "grad_norm": 2.0707528591156006,
      "learning_rate": 3.147147568591678e-05,
      "loss": 0.7256,
      "step": 6865
    },
    {
      "epoch": 2.1679823176507735,
      "grad_norm": 2.3035848140716553,
      "learning_rate": 3.146796716019929e-05,
      "loss": 0.6866,
      "step": 6866
    },
    {
      "epoch": 2.1682980738869593,
      "grad_norm": 2.065610885620117,
      "learning_rate": 3.1464458634481795e-05,
      "loss": 0.8014,
      "step": 6867
    },
    {
      "epoch": 2.168613830123145,
      "grad_norm": 2.079629898071289,
      "learning_rate": 3.1460950108764295e-05,
      "loss": 0.7146,
      "step": 6868
    },
    {
      "epoch": 2.1689295863593308,
      "grad_norm": 2.543529510498047,
      "learning_rate": 3.14574415830468e-05,
      "loss": 0.8081,
      "step": 6869
    },
    {
      "epoch": 2.1692453425955165,
      "grad_norm": 2.0147557258605957,
      "learning_rate": 3.145393305732931e-05,
      "loss": 0.8092,
      "step": 6870
    },
    {
      "epoch": 2.169561098831702,
      "grad_norm": 2.269550085067749,
      "learning_rate": 3.1450424531611815e-05,
      "loss": 0.7312,
      "step": 6871
    },
    {
      "epoch": 2.1698768550678875,
      "grad_norm": 2.495201349258423,
      "learning_rate": 3.144691600589432e-05,
      "loss": 0.7507,
      "step": 6872
    },
    {
      "epoch": 2.1701926113040733,
      "grad_norm": 2.204404354095459,
      "learning_rate": 3.144340748017683e-05,
      "loss": 0.7824,
      "step": 6873
    },
    {
      "epoch": 2.170508367540259,
      "grad_norm": 2.177433729171753,
      "learning_rate": 3.143989895445934e-05,
      "loss": 0.6996,
      "step": 6874
    },
    {
      "epoch": 2.1708241237764447,
      "grad_norm": 2.1818623542785645,
      "learning_rate": 3.143639042874185e-05,
      "loss": 0.7598,
      "step": 6875
    },
    {
      "epoch": 2.17113988001263,
      "grad_norm": 2.4439516067504883,
      "learning_rate": 3.1432881903024356e-05,
      "loss": 0.8095,
      "step": 6876
    },
    {
      "epoch": 2.1714556362488158,
      "grad_norm": 2.12886381149292,
      "learning_rate": 3.1429373377306856e-05,
      "loss": 0.6936,
      "step": 6877
    },
    {
      "epoch": 2.1717713924850015,
      "grad_norm": 2.0463461875915527,
      "learning_rate": 3.142586485158936e-05,
      "loss": 0.6873,
      "step": 6878
    },
    {
      "epoch": 2.1720871487211872,
      "grad_norm": 2.4373247623443604,
      "learning_rate": 3.142235632587187e-05,
      "loss": 0.8351,
      "step": 6879
    },
    {
      "epoch": 2.172402904957373,
      "grad_norm": 2.3773300647735596,
      "learning_rate": 3.1418847800154376e-05,
      "loss": 0.7157,
      "step": 6880
    },
    {
      "epoch": 2.1727186611935587,
      "grad_norm": 2.4159703254699707,
      "learning_rate": 3.141533927443688e-05,
      "loss": 0.7499,
      "step": 6881
    },
    {
      "epoch": 2.1730344174297445,
      "grad_norm": 2.21565842628479,
      "learning_rate": 3.141183074871939e-05,
      "loss": 0.7113,
      "step": 6882
    },
    {
      "epoch": 2.1733501736659298,
      "grad_norm": 2.5401837825775146,
      "learning_rate": 3.14083222230019e-05,
      "loss": 0.7155,
      "step": 6883
    },
    {
      "epoch": 2.1736659299021155,
      "grad_norm": 2.0962510108947754,
      "learning_rate": 3.1404813697284404e-05,
      "loss": 0.7806,
      "step": 6884
    },
    {
      "epoch": 2.1739816861383012,
      "grad_norm": 2.482536792755127,
      "learning_rate": 3.140130517156691e-05,
      "loss": 0.7719,
      "step": 6885
    },
    {
      "epoch": 2.174297442374487,
      "grad_norm": 2.0880324840545654,
      "learning_rate": 3.139779664584942e-05,
      "loss": 0.7213,
      "step": 6886
    },
    {
      "epoch": 2.1746131986106727,
      "grad_norm": 2.402268409729004,
      "learning_rate": 3.139428812013192e-05,
      "loss": 0.7949,
      "step": 6887
    },
    {
      "epoch": 2.174928954846858,
      "grad_norm": 2.2460930347442627,
      "learning_rate": 3.1390779594414424e-05,
      "loss": 0.6827,
      "step": 6888
    },
    {
      "epoch": 2.1752447110830437,
      "grad_norm": 2.388453483581543,
      "learning_rate": 3.138727106869693e-05,
      "loss": 0.7315,
      "step": 6889
    },
    {
      "epoch": 2.1755604673192295,
      "grad_norm": 2.065554618835449,
      "learning_rate": 3.138376254297944e-05,
      "loss": 0.7709,
      "step": 6890
    },
    {
      "epoch": 2.175876223555415,
      "grad_norm": 2.408198118209839,
      "learning_rate": 3.138025401726195e-05,
      "loss": 0.7716,
      "step": 6891
    },
    {
      "epoch": 2.176191979791601,
      "grad_norm": 1.8749502897262573,
      "learning_rate": 3.137674549154446e-05,
      "loss": 0.7178,
      "step": 6892
    },
    {
      "epoch": 2.1765077360277867,
      "grad_norm": 2.294121026992798,
      "learning_rate": 3.1373236965826965e-05,
      "loss": 0.792,
      "step": 6893
    },
    {
      "epoch": 2.1768234922639724,
      "grad_norm": 2.3178462982177734,
      "learning_rate": 3.136972844010947e-05,
      "loss": 0.7677,
      "step": 6894
    },
    {
      "epoch": 2.1771392485001577,
      "grad_norm": 2.251103401184082,
      "learning_rate": 3.136621991439198e-05,
      "loss": 0.769,
      "step": 6895
    },
    {
      "epoch": 2.1774550047363435,
      "grad_norm": 2.4031431674957275,
      "learning_rate": 3.136271138867448e-05,
      "loss": 0.7173,
      "step": 6896
    },
    {
      "epoch": 2.177770760972529,
      "grad_norm": 2.0890727043151855,
      "learning_rate": 3.1359202862956985e-05,
      "loss": 0.7539,
      "step": 6897
    },
    {
      "epoch": 2.178086517208715,
      "grad_norm": 2.3998680114746094,
      "learning_rate": 3.135569433723949e-05,
      "loss": 0.6937,
      "step": 6898
    },
    {
      "epoch": 2.1784022734449007,
      "grad_norm": 2.400709867477417,
      "learning_rate": 3.1352185811522e-05,
      "loss": 0.7921,
      "step": 6899
    },
    {
      "epoch": 2.178718029681086,
      "grad_norm": 1.9990376234054565,
      "learning_rate": 3.1348677285804506e-05,
      "loss": 0.7561,
      "step": 6900
    },
    {
      "epoch": 2.1790337859172717,
      "grad_norm": 1.9712812900543213,
      "learning_rate": 3.134516876008701e-05,
      "loss": 0.8284,
      "step": 6901
    },
    {
      "epoch": 2.1793495421534574,
      "grad_norm": 1.8277530670166016,
      "learning_rate": 3.134166023436952e-05,
      "loss": 0.7423,
      "step": 6902
    },
    {
      "epoch": 2.179665298389643,
      "grad_norm": 2.3931634426116943,
      "learning_rate": 3.1338151708652026e-05,
      "loss": 0.7046,
      "step": 6903
    },
    {
      "epoch": 2.179981054625829,
      "grad_norm": 2.468158006668091,
      "learning_rate": 3.133464318293453e-05,
      "loss": 0.7552,
      "step": 6904
    },
    {
      "epoch": 2.1802968108620147,
      "grad_norm": 2.123133420944214,
      "learning_rate": 3.133113465721704e-05,
      "loss": 0.7362,
      "step": 6905
    },
    {
      "epoch": 2.1806125670982004,
      "grad_norm": 2.6154234409332275,
      "learning_rate": 3.132762613149955e-05,
      "loss": 0.7735,
      "step": 6906
    },
    {
      "epoch": 2.1809283233343857,
      "grad_norm": 2.0382280349731445,
      "learning_rate": 3.132411760578205e-05,
      "loss": 0.6917,
      "step": 6907
    },
    {
      "epoch": 2.1812440795705714,
      "grad_norm": 2.030731201171875,
      "learning_rate": 3.1320609080064554e-05,
      "loss": 0.7882,
      "step": 6908
    },
    {
      "epoch": 2.181559835806757,
      "grad_norm": 2.3782784938812256,
      "learning_rate": 3.131710055434707e-05,
      "loss": 0.7875,
      "step": 6909
    },
    {
      "epoch": 2.181875592042943,
      "grad_norm": 3.6744272708892822,
      "learning_rate": 3.1313592028629574e-05,
      "loss": 0.7966,
      "step": 6910
    },
    {
      "epoch": 2.1821913482791286,
      "grad_norm": 2.7462878227233887,
      "learning_rate": 3.131008350291208e-05,
      "loss": 0.6878,
      "step": 6911
    },
    {
      "epoch": 2.1825071045153144,
      "grad_norm": 2.753389358520508,
      "learning_rate": 3.130657497719459e-05,
      "loss": 0.7048,
      "step": 6912
    },
    {
      "epoch": 2.1828228607514997,
      "grad_norm": 2.341193199157715,
      "learning_rate": 3.1303066451477094e-05,
      "loss": 0.8376,
      "step": 6913
    },
    {
      "epoch": 2.1831386169876854,
      "grad_norm": 2.6018967628479004,
      "learning_rate": 3.12995579257596e-05,
      "loss": 0.8154,
      "step": 6914
    },
    {
      "epoch": 2.183454373223871,
      "grad_norm": 2.467395544052124,
      "learning_rate": 3.12960494000421e-05,
      "loss": 0.6676,
      "step": 6915
    },
    {
      "epoch": 2.183770129460057,
      "grad_norm": 2.599199056625366,
      "learning_rate": 3.129254087432461e-05,
      "loss": 0.7164,
      "step": 6916
    },
    {
      "epoch": 2.1840858856962426,
      "grad_norm": 2.606008768081665,
      "learning_rate": 3.1289032348607115e-05,
      "loss": 0.8072,
      "step": 6917
    },
    {
      "epoch": 2.1844016419324284,
      "grad_norm": 3.078294038772583,
      "learning_rate": 3.128552382288962e-05,
      "loss": 0.8161,
      "step": 6918
    },
    {
      "epoch": 2.1847173981686137,
      "grad_norm": 2.0898895263671875,
      "learning_rate": 3.128201529717213e-05,
      "loss": 0.6562,
      "step": 6919
    },
    {
      "epoch": 2.1850331544047994,
      "grad_norm": 2.338548183441162,
      "learning_rate": 3.1278506771454635e-05,
      "loss": 0.745,
      "step": 6920
    },
    {
      "epoch": 2.185348910640985,
      "grad_norm": 1.945318579673767,
      "learning_rate": 3.127499824573714e-05,
      "loss": 0.7081,
      "step": 6921
    },
    {
      "epoch": 2.185664666877171,
      "grad_norm": 2.260446786880493,
      "learning_rate": 3.127148972001965e-05,
      "loss": 0.7683,
      "step": 6922
    },
    {
      "epoch": 2.1859804231133566,
      "grad_norm": 1.961209774017334,
      "learning_rate": 3.1267981194302156e-05,
      "loss": 0.7405,
      "step": 6923
    },
    {
      "epoch": 2.1862961793495423,
      "grad_norm": 2.590855598449707,
      "learning_rate": 3.126447266858466e-05,
      "loss": 0.7257,
      "step": 6924
    },
    {
      "epoch": 2.1866119355857276,
      "grad_norm": 2.522657632827759,
      "learning_rate": 3.126096414286717e-05,
      "loss": 0.7211,
      "step": 6925
    },
    {
      "epoch": 2.1869276918219134,
      "grad_norm": 2.5991055965423584,
      "learning_rate": 3.1257455617149676e-05,
      "loss": 0.7965,
      "step": 6926
    },
    {
      "epoch": 2.187243448058099,
      "grad_norm": 2.4663400650024414,
      "learning_rate": 3.125394709143218e-05,
      "loss": 0.7883,
      "step": 6927
    },
    {
      "epoch": 2.187559204294285,
      "grad_norm": 1.954297423362732,
      "learning_rate": 3.125043856571469e-05,
      "loss": 0.6561,
      "step": 6928
    },
    {
      "epoch": 2.1878749605304706,
      "grad_norm": 2.1081156730651855,
      "learning_rate": 3.12469300399972e-05,
      "loss": 0.7788,
      "step": 6929
    },
    {
      "epoch": 2.1881907167666563,
      "grad_norm": 2.9626495838165283,
      "learning_rate": 3.1243421514279703e-05,
      "loss": 0.7672,
      "step": 6930
    },
    {
      "epoch": 2.1885064730028416,
      "grad_norm": 2.310108184814453,
      "learning_rate": 3.123991298856221e-05,
      "loss": 0.7901,
      "step": 6931
    },
    {
      "epoch": 2.1888222292390274,
      "grad_norm": 2.459296464920044,
      "learning_rate": 3.123640446284472e-05,
      "loss": 0.7412,
      "step": 6932
    },
    {
      "epoch": 2.189137985475213,
      "grad_norm": 2.314457654953003,
      "learning_rate": 3.1232895937127224e-05,
      "loss": 0.765,
      "step": 6933
    },
    {
      "epoch": 2.189453741711399,
      "grad_norm": 2.3434197902679443,
      "learning_rate": 3.122938741140973e-05,
      "loss": 0.8,
      "step": 6934
    },
    {
      "epoch": 2.1897694979475846,
      "grad_norm": 2.2932345867156982,
      "learning_rate": 3.122587888569223e-05,
      "loss": 0.7724,
      "step": 6935
    },
    {
      "epoch": 2.1900852541837703,
      "grad_norm": 2.5003488063812256,
      "learning_rate": 3.122237035997474e-05,
      "loss": 0.7551,
      "step": 6936
    },
    {
      "epoch": 2.1904010104199556,
      "grad_norm": 2.3951499462127686,
      "learning_rate": 3.1218861834257244e-05,
      "loss": 0.6998,
      "step": 6937
    },
    {
      "epoch": 2.1907167666561413,
      "grad_norm": 2.454188346862793,
      "learning_rate": 3.121535330853975e-05,
      "loss": 0.7244,
      "step": 6938
    },
    {
      "epoch": 2.191032522892327,
      "grad_norm": 2.182811975479126,
      "learning_rate": 3.121184478282226e-05,
      "loss": 0.7166,
      "step": 6939
    },
    {
      "epoch": 2.191348279128513,
      "grad_norm": 2.2786381244659424,
      "learning_rate": 3.1208336257104765e-05,
      "loss": 0.7575,
      "step": 6940
    },
    {
      "epoch": 2.1916640353646986,
      "grad_norm": 2.3486509323120117,
      "learning_rate": 3.120482773138727e-05,
      "loss": 0.7191,
      "step": 6941
    },
    {
      "epoch": 2.1919797916008843,
      "grad_norm": 2.508148193359375,
      "learning_rate": 3.120131920566978e-05,
      "loss": 0.7947,
      "step": 6942
    },
    {
      "epoch": 2.1922955478370696,
      "grad_norm": 2.0271053314208984,
      "learning_rate": 3.1197810679952285e-05,
      "loss": 0.7061,
      "step": 6943
    },
    {
      "epoch": 2.1926113040732553,
      "grad_norm": 2.330192804336548,
      "learning_rate": 3.119430215423479e-05,
      "loss": 0.7905,
      "step": 6944
    },
    {
      "epoch": 2.192927060309441,
      "grad_norm": 2.110102891921997,
      "learning_rate": 3.11907936285173e-05,
      "loss": 0.7515,
      "step": 6945
    },
    {
      "epoch": 2.193242816545627,
      "grad_norm": 2.139892339706421,
      "learning_rate": 3.1187285102799806e-05,
      "loss": 0.7434,
      "step": 6946
    },
    {
      "epoch": 2.1935585727818125,
      "grad_norm": 2.089160203933716,
      "learning_rate": 3.118377657708231e-05,
      "loss": 0.786,
      "step": 6947
    },
    {
      "epoch": 2.1938743290179983,
      "grad_norm": 2.0158069133758545,
      "learning_rate": 3.118026805136482e-05,
      "loss": 0.7676,
      "step": 6948
    },
    {
      "epoch": 2.1941900852541836,
      "grad_norm": 2.314312696456909,
      "learning_rate": 3.1176759525647326e-05,
      "loss": 0.6919,
      "step": 6949
    },
    {
      "epoch": 2.1945058414903693,
      "grad_norm": 2.3269805908203125,
      "learning_rate": 3.117325099992983e-05,
      "loss": 0.6952,
      "step": 6950
    },
    {
      "epoch": 2.194821597726555,
      "grad_norm": 2.7554454803466797,
      "learning_rate": 3.116974247421234e-05,
      "loss": 0.7627,
      "step": 6951
    },
    {
      "epoch": 2.195137353962741,
      "grad_norm": 2.53166127204895,
      "learning_rate": 3.1166233948494847e-05,
      "loss": 0.707,
      "step": 6952
    },
    {
      "epoch": 2.1954531101989265,
      "grad_norm": 2.249422073364258,
      "learning_rate": 3.1162725422777353e-05,
      "loss": 0.6835,
      "step": 6953
    },
    {
      "epoch": 2.1957688664351123,
      "grad_norm": 2.167524576187134,
      "learning_rate": 3.1159216897059853e-05,
      "loss": 0.7065,
      "step": 6954
    },
    {
      "epoch": 2.1960846226712976,
      "grad_norm": 1.9959344863891602,
      "learning_rate": 3.115570837134236e-05,
      "loss": 0.7488,
      "step": 6955
    },
    {
      "epoch": 2.1964003789074833,
      "grad_norm": 2.4210193157196045,
      "learning_rate": 3.115219984562487e-05,
      "loss": 0.7275,
      "step": 6956
    },
    {
      "epoch": 2.196716135143669,
      "grad_norm": 2.418027400970459,
      "learning_rate": 3.1148691319907374e-05,
      "loss": 0.7805,
      "step": 6957
    },
    {
      "epoch": 2.1970318913798548,
      "grad_norm": 1.9333016872406006,
      "learning_rate": 3.114518279418988e-05,
      "loss": 0.6913,
      "step": 6958
    },
    {
      "epoch": 2.1973476476160405,
      "grad_norm": 2.4277188777923584,
      "learning_rate": 3.114167426847239e-05,
      "loss": 0.7705,
      "step": 6959
    },
    {
      "epoch": 2.1976634038522262,
      "grad_norm": 2.395953893661499,
      "learning_rate": 3.1138165742754894e-05,
      "loss": 0.7648,
      "step": 6960
    },
    {
      "epoch": 2.1979791600884115,
      "grad_norm": 2.9151248931884766,
      "learning_rate": 3.113465721703741e-05,
      "loss": 0.7191,
      "step": 6961
    },
    {
      "epoch": 2.1982949163245973,
      "grad_norm": 2.548158884048462,
      "learning_rate": 3.113114869131991e-05,
      "loss": 0.835,
      "step": 6962
    },
    {
      "epoch": 2.198610672560783,
      "grad_norm": 2.597038745880127,
      "learning_rate": 3.1127640165602415e-05,
      "loss": 0.7993,
      "step": 6963
    },
    {
      "epoch": 2.1989264287969688,
      "grad_norm": 1.9257099628448486,
      "learning_rate": 3.112413163988492e-05,
      "loss": 0.6841,
      "step": 6964
    },
    {
      "epoch": 2.1992421850331545,
      "grad_norm": 1.8968545198440552,
      "learning_rate": 3.112062311416743e-05,
      "loss": 0.7152,
      "step": 6965
    },
    {
      "epoch": 2.1995579412693402,
      "grad_norm": 2.001333475112915,
      "learning_rate": 3.1117114588449935e-05,
      "loss": 0.7109,
      "step": 6966
    },
    {
      "epoch": 2.1998736975055255,
      "grad_norm": 2.2053918838500977,
      "learning_rate": 3.111360606273244e-05,
      "loss": 0.7192,
      "step": 6967
    },
    {
      "epoch": 2.2001894537417113,
      "grad_norm": 2.525728225708008,
      "learning_rate": 3.111009753701495e-05,
      "loss": 0.8896,
      "step": 6968
    },
    {
      "epoch": 2.200505209977897,
      "grad_norm": 2.246081590652466,
      "learning_rate": 3.1106589011297456e-05,
      "loss": 0.7927,
      "step": 6969
    },
    {
      "epoch": 2.2008209662140827,
      "grad_norm": 2.549044132232666,
      "learning_rate": 3.110308048557996e-05,
      "loss": 0.7013,
      "step": 6970
    },
    {
      "epoch": 2.2011367224502685,
      "grad_norm": 2.147968053817749,
      "learning_rate": 3.109957195986247e-05,
      "loss": 0.6567,
      "step": 6971
    },
    {
      "epoch": 2.201452478686454,
      "grad_norm": 2.5668222904205322,
      "learning_rate": 3.1096063434144976e-05,
      "loss": 0.7088,
      "step": 6972
    },
    {
      "epoch": 2.20176823492264,
      "grad_norm": 2.7289676666259766,
      "learning_rate": 3.1092554908427476e-05,
      "loss": 0.8237,
      "step": 6973
    },
    {
      "epoch": 2.2020839911588252,
      "grad_norm": 2.1951205730438232,
      "learning_rate": 3.108904638270998e-05,
      "loss": 0.7698,
      "step": 6974
    },
    {
      "epoch": 2.202399747395011,
      "grad_norm": 2.619642734527588,
      "learning_rate": 3.108553785699249e-05,
      "loss": 0.7881,
      "step": 6975
    },
    {
      "epoch": 2.2027155036311967,
      "grad_norm": 2.192633867263794,
      "learning_rate": 3.1082029331274997e-05,
      "loss": 0.6781,
      "step": 6976
    },
    {
      "epoch": 2.2030312598673825,
      "grad_norm": 2.501967668533325,
      "learning_rate": 3.10785208055575e-05,
      "loss": 0.7398,
      "step": 6977
    },
    {
      "epoch": 2.203347016103568,
      "grad_norm": 2.253826856613159,
      "learning_rate": 3.107501227984001e-05,
      "loss": 0.7756,
      "step": 6978
    },
    {
      "epoch": 2.2036627723397535,
      "grad_norm": 2.149857997894287,
      "learning_rate": 3.1071503754122524e-05,
      "loss": 0.7253,
      "step": 6979
    },
    {
      "epoch": 2.2039785285759392,
      "grad_norm": 2.222339153289795,
      "learning_rate": 3.106799522840503e-05,
      "loss": 0.8098,
      "step": 6980
    },
    {
      "epoch": 2.204294284812125,
      "grad_norm": 2.35911226272583,
      "learning_rate": 3.106448670268754e-05,
      "loss": 0.7558,
      "step": 6981
    },
    {
      "epoch": 2.2046100410483107,
      "grad_norm": 2.548246383666992,
      "learning_rate": 3.106097817697004e-05,
      "loss": 0.723,
      "step": 6982
    },
    {
      "epoch": 2.2049257972844964,
      "grad_norm": 2.4246907234191895,
      "learning_rate": 3.1057469651252544e-05,
      "loss": 0.7119,
      "step": 6983
    },
    {
      "epoch": 2.205241553520682,
      "grad_norm": 2.5319912433624268,
      "learning_rate": 3.105396112553505e-05,
      "loss": 0.7019,
      "step": 6984
    },
    {
      "epoch": 2.205557309756868,
      "grad_norm": 2.17582368850708,
      "learning_rate": 3.105045259981756e-05,
      "loss": 0.7354,
      "step": 6985
    },
    {
      "epoch": 2.205873065993053,
      "grad_norm": 2.6091132164001465,
      "learning_rate": 3.1046944074100065e-05,
      "loss": 0.8014,
      "step": 6986
    },
    {
      "epoch": 2.206188822229239,
      "grad_norm": 2.5285701751708984,
      "learning_rate": 3.104343554838257e-05,
      "loss": 0.7776,
      "step": 6987
    },
    {
      "epoch": 2.2065045784654247,
      "grad_norm": 2.444873809814453,
      "learning_rate": 3.103992702266508e-05,
      "loss": 0.8246,
      "step": 6988
    },
    {
      "epoch": 2.2068203347016104,
      "grad_norm": 2.101264715194702,
      "learning_rate": 3.1036418496947585e-05,
      "loss": 0.7108,
      "step": 6989
    },
    {
      "epoch": 2.207136090937796,
      "grad_norm": 2.6864399909973145,
      "learning_rate": 3.103290997123009e-05,
      "loss": 0.7001,
      "step": 6990
    },
    {
      "epoch": 2.2074518471739815,
      "grad_norm": 2.397538423538208,
      "learning_rate": 3.10294014455126e-05,
      "loss": 0.6601,
      "step": 6991
    },
    {
      "epoch": 2.207767603410167,
      "grad_norm": 2.7886643409729004,
      "learning_rate": 3.10258929197951e-05,
      "loss": 0.7939,
      "step": 6992
    },
    {
      "epoch": 2.208083359646353,
      "grad_norm": 2.1274120807647705,
      "learning_rate": 3.1022384394077606e-05,
      "loss": 0.7578,
      "step": 6993
    },
    {
      "epoch": 2.2083991158825387,
      "grad_norm": 2.257737636566162,
      "learning_rate": 3.101887586836011e-05,
      "loss": 0.7495,
      "step": 6994
    },
    {
      "epoch": 2.2087148721187244,
      "grad_norm": 2.2747137546539307,
      "learning_rate": 3.101536734264262e-05,
      "loss": 0.7763,
      "step": 6995
    },
    {
      "epoch": 2.20903062835491,
      "grad_norm": 2.5156853199005127,
      "learning_rate": 3.101185881692513e-05,
      "loss": 0.7994,
      "step": 6996
    },
    {
      "epoch": 2.209346384591096,
      "grad_norm": 2.433133363723755,
      "learning_rate": 3.100835029120764e-05,
      "loss": 0.7133,
      "step": 6997
    },
    {
      "epoch": 2.209662140827281,
      "grad_norm": 2.281233549118042,
      "learning_rate": 3.1004841765490146e-05,
      "loss": 0.7651,
      "step": 6998
    },
    {
      "epoch": 2.209977897063467,
      "grad_norm": 2.5666158199310303,
      "learning_rate": 3.100133323977265e-05,
      "loss": 0.7301,
      "step": 6999
    },
    {
      "epoch": 2.2102936532996527,
      "grad_norm": 2.3010194301605225,
      "learning_rate": 3.099782471405516e-05,
      "loss": 0.827,
      "step": 7000
    },
    {
      "epoch": 2.2102936532996527,
      "eval_loss": 0.794898509979248,
      "eval_runtime": 4.6519,
      "eval_samples_per_second": 535.268,
      "eval_steps_per_second": 4.514,
      "step": 7000
    },
    {
      "epoch": 2.2106094095358384,
      "grad_norm": 2.982457399368286,
      "learning_rate": 3.099431618833766e-05,
      "loss": 0.877,
      "step": 7001
    },
    {
      "epoch": 2.210925165772024,
      "grad_norm": 2.829554557800293,
      "learning_rate": 3.099080766262017e-05,
      "loss": 0.7844,
      "step": 7002
    },
    {
      "epoch": 2.2112409220082094,
      "grad_norm": 2.21689772605896,
      "learning_rate": 3.0987299136902674e-05,
      "loss": 0.7476,
      "step": 7003
    },
    {
      "epoch": 2.211556678244395,
      "grad_norm": 2.553528308868408,
      "learning_rate": 3.098379061118518e-05,
      "loss": 0.7413,
      "step": 7004
    },
    {
      "epoch": 2.211872434480581,
      "grad_norm": 2.625643253326416,
      "learning_rate": 3.098028208546769e-05,
      "loss": 0.7475,
      "step": 7005
    },
    {
      "epoch": 2.2121881907167666,
      "grad_norm": 2.0798192024230957,
      "learning_rate": 3.0976773559750194e-05,
      "loss": 0.7425,
      "step": 7006
    },
    {
      "epoch": 2.2125039469529524,
      "grad_norm": 2.423009157180786,
      "learning_rate": 3.09732650340327e-05,
      "loss": 0.7629,
      "step": 7007
    },
    {
      "epoch": 2.212819703189138,
      "grad_norm": 2.3721702098846436,
      "learning_rate": 3.096975650831521e-05,
      "loss": 0.8515,
      "step": 7008
    },
    {
      "epoch": 2.213135459425324,
      "grad_norm": 2.2181951999664307,
      "learning_rate": 3.0966247982597715e-05,
      "loss": 0.7356,
      "step": 7009
    },
    {
      "epoch": 2.213451215661509,
      "grad_norm": 2.475280284881592,
      "learning_rate": 3.096273945688022e-05,
      "loss": 0.8286,
      "step": 7010
    },
    {
      "epoch": 2.213766971897695,
      "grad_norm": 3.2498111724853516,
      "learning_rate": 3.095923093116273e-05,
      "loss": 0.7756,
      "step": 7011
    },
    {
      "epoch": 2.2140827281338806,
      "grad_norm": 2.55499005317688,
      "learning_rate": 3.095572240544523e-05,
      "loss": 0.8056,
      "step": 7012
    },
    {
      "epoch": 2.2143984843700664,
      "grad_norm": 2.441451072692871,
      "learning_rate": 3.0952213879727735e-05,
      "loss": 0.6911,
      "step": 7013
    },
    {
      "epoch": 2.214714240606252,
      "grad_norm": 2.503180503845215,
      "learning_rate": 3.094870535401025e-05,
      "loss": 0.7289,
      "step": 7014
    },
    {
      "epoch": 2.215029996842438,
      "grad_norm": 2.4630000591278076,
      "learning_rate": 3.0945196828292755e-05,
      "loss": 0.7304,
      "step": 7015
    },
    {
      "epoch": 2.215345753078623,
      "grad_norm": 2.4726457595825195,
      "learning_rate": 3.094168830257526e-05,
      "loss": 0.7961,
      "step": 7016
    },
    {
      "epoch": 2.215661509314809,
      "grad_norm": 2.2462337017059326,
      "learning_rate": 3.093817977685777e-05,
      "loss": 0.7297,
      "step": 7017
    },
    {
      "epoch": 2.2159772655509946,
      "grad_norm": 2.4599127769470215,
      "learning_rate": 3.0934671251140276e-05,
      "loss": 0.7008,
      "step": 7018
    },
    {
      "epoch": 2.2162930217871804,
      "grad_norm": 2.269935131072998,
      "learning_rate": 3.093116272542278e-05,
      "loss": 0.7697,
      "step": 7019
    },
    {
      "epoch": 2.216608778023366,
      "grad_norm": 2.2348005771636963,
      "learning_rate": 3.092765419970528e-05,
      "loss": 0.7079,
      "step": 7020
    },
    {
      "epoch": 2.216924534259552,
      "grad_norm": 2.442119598388672,
      "learning_rate": 3.092414567398779e-05,
      "loss": 0.7483,
      "step": 7021
    },
    {
      "epoch": 2.217240290495737,
      "grad_norm": 3.032539129257202,
      "learning_rate": 3.0920637148270296e-05,
      "loss": 0.7892,
      "step": 7022
    },
    {
      "epoch": 2.217556046731923,
      "grad_norm": 2.682512044906616,
      "learning_rate": 3.09171286225528e-05,
      "loss": 0.7158,
      "step": 7023
    },
    {
      "epoch": 2.2178718029681086,
      "grad_norm": 2.399273157119751,
      "learning_rate": 3.091362009683531e-05,
      "loss": 0.6552,
      "step": 7024
    },
    {
      "epoch": 2.2181875592042943,
      "grad_norm": 1.8452287912368774,
      "learning_rate": 3.091011157111782e-05,
      "loss": 0.7949,
      "step": 7025
    },
    {
      "epoch": 2.21850331544048,
      "grad_norm": 2.2427725791931152,
      "learning_rate": 3.0906603045400324e-05,
      "loss": 0.7843,
      "step": 7026
    },
    {
      "epoch": 2.218819071676666,
      "grad_norm": 2.1445887088775635,
      "learning_rate": 3.090309451968283e-05,
      "loss": 0.7129,
      "step": 7027
    },
    {
      "epoch": 2.219134827912851,
      "grad_norm": 2.140022039413452,
      "learning_rate": 3.089958599396534e-05,
      "loss": 0.7261,
      "step": 7028
    },
    {
      "epoch": 2.219450584149037,
      "grad_norm": 2.14717435836792,
      "learning_rate": 3.0896077468247844e-05,
      "loss": 0.7526,
      "step": 7029
    },
    {
      "epoch": 2.2197663403852226,
      "grad_norm": 2.1151702404022217,
      "learning_rate": 3.089256894253035e-05,
      "loss": 0.7137,
      "step": 7030
    },
    {
      "epoch": 2.2200820966214083,
      "grad_norm": 2.041774034500122,
      "learning_rate": 3.088906041681286e-05,
      "loss": 0.6801,
      "step": 7031
    },
    {
      "epoch": 2.220397852857594,
      "grad_norm": 2.6766180992126465,
      "learning_rate": 3.0885551891095365e-05,
      "loss": 0.7181,
      "step": 7032
    },
    {
      "epoch": 2.22071360909378,
      "grad_norm": 1.9282101392745972,
      "learning_rate": 3.088204336537787e-05,
      "loss": 0.6709,
      "step": 7033
    },
    {
      "epoch": 2.221029365329965,
      "grad_norm": 1.8952668905258179,
      "learning_rate": 3.087853483966038e-05,
      "loss": 0.6508,
      "step": 7034
    },
    {
      "epoch": 2.221345121566151,
      "grad_norm": 2.521449089050293,
      "learning_rate": 3.0875026313942885e-05,
      "loss": 0.7762,
      "step": 7035
    },
    {
      "epoch": 2.2216608778023366,
      "grad_norm": 2.5264554023742676,
      "learning_rate": 3.087151778822539e-05,
      "loss": 0.6731,
      "step": 7036
    },
    {
      "epoch": 2.2219766340385223,
      "grad_norm": 1.9786850214004517,
      "learning_rate": 3.08680092625079e-05,
      "loss": 0.6901,
      "step": 7037
    },
    {
      "epoch": 2.222292390274708,
      "grad_norm": 2.0600435733795166,
      "learning_rate": 3.0864500736790405e-05,
      "loss": 0.6257,
      "step": 7038
    },
    {
      "epoch": 2.222608146510894,
      "grad_norm": 2.0374197959899902,
      "learning_rate": 3.086099221107291e-05,
      "loss": 0.6923,
      "step": 7039
    },
    {
      "epoch": 2.222923902747079,
      "grad_norm": 2.310828447341919,
      "learning_rate": 3.085748368535541e-05,
      "loss": 0.7064,
      "step": 7040
    },
    {
      "epoch": 2.223239658983265,
      "grad_norm": 3.012671947479248,
      "learning_rate": 3.085397515963792e-05,
      "loss": 0.7992,
      "step": 7041
    },
    {
      "epoch": 2.2235554152194505,
      "grad_norm": 2.1257553100585938,
      "learning_rate": 3.0850466633920426e-05,
      "loss": 0.7503,
      "step": 7042
    },
    {
      "epoch": 2.2238711714556363,
      "grad_norm": 2.481330156326294,
      "learning_rate": 3.084695810820293e-05,
      "loss": 0.792,
      "step": 7043
    },
    {
      "epoch": 2.224186927691822,
      "grad_norm": 2.17275071144104,
      "learning_rate": 3.084344958248544e-05,
      "loss": 0.7595,
      "step": 7044
    },
    {
      "epoch": 2.2245026839280078,
      "grad_norm": 2.651052474975586,
      "learning_rate": 3.0839941056767946e-05,
      "loss": 0.7085,
      "step": 7045
    },
    {
      "epoch": 2.224818440164193,
      "grad_norm": 2.5374433994293213,
      "learning_rate": 3.083643253105045e-05,
      "loss": 0.7344,
      "step": 7046
    },
    {
      "epoch": 2.225134196400379,
      "grad_norm": 2.7245404720306396,
      "learning_rate": 3.083292400533296e-05,
      "loss": 0.6964,
      "step": 7047
    },
    {
      "epoch": 2.2254499526365645,
      "grad_norm": 2.45613694190979,
      "learning_rate": 3.082941547961547e-05,
      "loss": 0.7657,
      "step": 7048
    },
    {
      "epoch": 2.2257657088727503,
      "grad_norm": 2.5468358993530273,
      "learning_rate": 3.0825906953897974e-05,
      "loss": 0.7862,
      "step": 7049
    },
    {
      "epoch": 2.226081465108936,
      "grad_norm": 2.3698441982269287,
      "learning_rate": 3.082239842818048e-05,
      "loss": 0.7095,
      "step": 7050
    },
    {
      "epoch": 2.2263972213451217,
      "grad_norm": 2.230936050415039,
      "learning_rate": 3.081888990246299e-05,
      "loss": 0.7093,
      "step": 7051
    },
    {
      "epoch": 2.226712977581307,
      "grad_norm": 2.2393150329589844,
      "learning_rate": 3.0815381376745494e-05,
      "loss": 0.7115,
      "step": 7052
    },
    {
      "epoch": 2.227028733817493,
      "grad_norm": 2.3430168628692627,
      "learning_rate": 3.0811872851028e-05,
      "loss": 0.7202,
      "step": 7053
    },
    {
      "epoch": 2.2273444900536785,
      "grad_norm": 2.245939254760742,
      "learning_rate": 3.080836432531051e-05,
      "loss": 0.7391,
      "step": 7054
    },
    {
      "epoch": 2.2276602462898643,
      "grad_norm": 2.451160430908203,
      "learning_rate": 3.0804855799593014e-05,
      "loss": 0.7409,
      "step": 7055
    },
    {
      "epoch": 2.22797600252605,
      "grad_norm": 2.522615432739258,
      "learning_rate": 3.080134727387552e-05,
      "loss": 0.7929,
      "step": 7056
    },
    {
      "epoch": 2.2282917587622357,
      "grad_norm": 2.3565144538879395,
      "learning_rate": 3.079783874815803e-05,
      "loss": 0.8147,
      "step": 7057
    },
    {
      "epoch": 2.228607514998421,
      "grad_norm": 2.1623427867889404,
      "learning_rate": 3.0794330222440535e-05,
      "loss": 0.7663,
      "step": 7058
    },
    {
      "epoch": 2.2289232712346068,
      "grad_norm": 2.1652281284332275,
      "learning_rate": 3.0790821696723035e-05,
      "loss": 0.7231,
      "step": 7059
    },
    {
      "epoch": 2.2292390274707925,
      "grad_norm": 2.193481206893921,
      "learning_rate": 3.078731317100554e-05,
      "loss": 0.7416,
      "step": 7060
    },
    {
      "epoch": 2.2295547837069782,
      "grad_norm": 2.2212977409362793,
      "learning_rate": 3.078380464528805e-05,
      "loss": 0.68,
      "step": 7061
    },
    {
      "epoch": 2.229870539943164,
      "grad_norm": 2.3086159229278564,
      "learning_rate": 3.0780296119570555e-05,
      "loss": 0.7193,
      "step": 7062
    },
    {
      "epoch": 2.2301862961793497,
      "grad_norm": 2.453998565673828,
      "learning_rate": 3.077678759385306e-05,
      "loss": 0.7522,
      "step": 7063
    },
    {
      "epoch": 2.230502052415535,
      "grad_norm": 2.469996929168701,
      "learning_rate": 3.077327906813557e-05,
      "loss": 0.7179,
      "step": 7064
    },
    {
      "epoch": 2.2308178086517207,
      "grad_norm": 2.3959169387817383,
      "learning_rate": 3.0769770542418076e-05,
      "loss": 0.7178,
      "step": 7065
    },
    {
      "epoch": 2.2311335648879065,
      "grad_norm": 2.1577577590942383,
      "learning_rate": 3.076626201670059e-05,
      "loss": 0.7618,
      "step": 7066
    },
    {
      "epoch": 2.2314493211240922,
      "grad_norm": 2.1654553413391113,
      "learning_rate": 3.076275349098309e-05,
      "loss": 0.7222,
      "step": 7067
    },
    {
      "epoch": 2.231765077360278,
      "grad_norm": 2.2604920864105225,
      "learning_rate": 3.0759244965265596e-05,
      "loss": 0.7142,
      "step": 7068
    },
    {
      "epoch": 2.2320808335964637,
      "grad_norm": 2.404649257659912,
      "learning_rate": 3.07557364395481e-05,
      "loss": 0.7116,
      "step": 7069
    },
    {
      "epoch": 2.232396589832649,
      "grad_norm": 2.368260622024536,
      "learning_rate": 3.075222791383061e-05,
      "loss": 0.7887,
      "step": 7070
    },
    {
      "epoch": 2.2327123460688347,
      "grad_norm": 2.508551597595215,
      "learning_rate": 3.074871938811312e-05,
      "loss": 0.8059,
      "step": 7071
    },
    {
      "epoch": 2.2330281023050205,
      "grad_norm": 2.006636142730713,
      "learning_rate": 3.0745210862395623e-05,
      "loss": 0.7135,
      "step": 7072
    },
    {
      "epoch": 2.233343858541206,
      "grad_norm": 2.3622262477874756,
      "learning_rate": 3.074170233667813e-05,
      "loss": 0.7596,
      "step": 7073
    },
    {
      "epoch": 2.233659614777392,
      "grad_norm": 2.3288979530334473,
      "learning_rate": 3.073819381096064e-05,
      "loss": 0.7507,
      "step": 7074
    },
    {
      "epoch": 2.2339753710135777,
      "grad_norm": 2.292705535888672,
      "learning_rate": 3.0734685285243144e-05,
      "loss": 0.7319,
      "step": 7075
    },
    {
      "epoch": 2.234291127249763,
      "grad_norm": 2.7337534427642822,
      "learning_rate": 3.073117675952565e-05,
      "loss": 0.7346,
      "step": 7076
    },
    {
      "epoch": 2.2346068834859487,
      "grad_norm": 2.415074348449707,
      "learning_rate": 3.072766823380816e-05,
      "loss": 0.6979,
      "step": 7077
    },
    {
      "epoch": 2.2349226397221345,
      "grad_norm": 2.207068681716919,
      "learning_rate": 3.072415970809066e-05,
      "loss": 0.709,
      "step": 7078
    },
    {
      "epoch": 2.23523839595832,
      "grad_norm": 2.4903416633605957,
      "learning_rate": 3.0720651182373164e-05,
      "loss": 0.7708,
      "step": 7079
    },
    {
      "epoch": 2.235554152194506,
      "grad_norm": 2.082897186279297,
      "learning_rate": 3.071714265665567e-05,
      "loss": 0.7634,
      "step": 7080
    },
    {
      "epoch": 2.2358699084306917,
      "grad_norm": 2.4188179969787598,
      "learning_rate": 3.071363413093818e-05,
      "loss": 0.8194,
      "step": 7081
    },
    {
      "epoch": 2.236185664666877,
      "grad_norm": 1.9610064029693604,
      "learning_rate": 3.0710125605220685e-05,
      "loss": 0.7268,
      "step": 7082
    },
    {
      "epoch": 2.2365014209030627,
      "grad_norm": 2.0312602519989014,
      "learning_rate": 3.070661707950319e-05,
      "loss": 0.6942,
      "step": 7083
    },
    {
      "epoch": 2.2368171771392484,
      "grad_norm": 2.4487180709838867,
      "learning_rate": 3.0703108553785705e-05,
      "loss": 0.6592,
      "step": 7084
    },
    {
      "epoch": 2.237132933375434,
      "grad_norm": 2.225050926208496,
      "learning_rate": 3.069960002806821e-05,
      "loss": 0.7035,
      "step": 7085
    },
    {
      "epoch": 2.23744868961162,
      "grad_norm": 2.2011630535125732,
      "learning_rate": 3.069609150235072e-05,
      "loss": 0.7341,
      "step": 7086
    },
    {
      "epoch": 2.2377644458478056,
      "grad_norm": 2.0928163528442383,
      "learning_rate": 3.069258297663322e-05,
      "loss": 0.7169,
      "step": 7087
    },
    {
      "epoch": 2.2380802020839914,
      "grad_norm": 2.532163619995117,
      "learning_rate": 3.0689074450915726e-05,
      "loss": 0.7003,
      "step": 7088
    },
    {
      "epoch": 2.2383959583201767,
      "grad_norm": 2.357783555984497,
      "learning_rate": 3.068556592519823e-05,
      "loss": 0.7247,
      "step": 7089
    },
    {
      "epoch": 2.2387117145563624,
      "grad_norm": 2.304234504699707,
      "learning_rate": 3.068205739948074e-05,
      "loss": 0.6766,
      "step": 7090
    },
    {
      "epoch": 2.239027470792548,
      "grad_norm": 2.233611822128296,
      "learning_rate": 3.0678548873763246e-05,
      "loss": 0.6802,
      "step": 7091
    },
    {
      "epoch": 2.239343227028734,
      "grad_norm": 2.183241128921509,
      "learning_rate": 3.067504034804575e-05,
      "loss": 0.6847,
      "step": 7092
    },
    {
      "epoch": 2.2396589832649196,
      "grad_norm": 2.567885637283325,
      "learning_rate": 3.067153182232826e-05,
      "loss": 0.7915,
      "step": 7093
    },
    {
      "epoch": 2.239974739501105,
      "grad_norm": 2.160076141357422,
      "learning_rate": 3.0668023296610767e-05,
      "loss": 0.8296,
      "step": 7094
    },
    {
      "epoch": 2.2402904957372907,
      "grad_norm": 2.593989610671997,
      "learning_rate": 3.0664514770893273e-05,
      "loss": 0.7153,
      "step": 7095
    },
    {
      "epoch": 2.2406062519734764,
      "grad_norm": 2.6350581645965576,
      "learning_rate": 3.066100624517578e-05,
      "loss": 0.7838,
      "step": 7096
    },
    {
      "epoch": 2.240922008209662,
      "grad_norm": 2.2070255279541016,
      "learning_rate": 3.065749771945828e-05,
      "loss": 0.6684,
      "step": 7097
    },
    {
      "epoch": 2.241237764445848,
      "grad_norm": 2.4243834018707275,
      "learning_rate": 3.065398919374079e-05,
      "loss": 0.7059,
      "step": 7098
    },
    {
      "epoch": 2.2415535206820336,
      "grad_norm": 2.2357306480407715,
      "learning_rate": 3.0650480668023294e-05,
      "loss": 0.6213,
      "step": 7099
    },
    {
      "epoch": 2.2418692769182194,
      "grad_norm": 2.0454835891723633,
      "learning_rate": 3.06469721423058e-05,
      "loss": 0.6776,
      "step": 7100
    },
    {
      "epoch": 2.2421850331544046,
      "grad_norm": 2.7366714477539062,
      "learning_rate": 3.0643463616588314e-05,
      "loss": 0.7948,
      "step": 7101
    },
    {
      "epoch": 2.2425007893905904,
      "grad_norm": 2.7008073329925537,
      "learning_rate": 3.063995509087082e-05,
      "loss": 0.7859,
      "step": 7102
    },
    {
      "epoch": 2.242816545626776,
      "grad_norm": 2.5692248344421387,
      "learning_rate": 3.063644656515333e-05,
      "loss": 0.7382,
      "step": 7103
    },
    {
      "epoch": 2.243132301862962,
      "grad_norm": 1.87924325466156,
      "learning_rate": 3.0632938039435835e-05,
      "loss": 0.7191,
      "step": 7104
    },
    {
      "epoch": 2.2434480580991476,
      "grad_norm": 2.757094621658325,
      "learning_rate": 3.062942951371834e-05,
      "loss": 0.7369,
      "step": 7105
    },
    {
      "epoch": 2.243763814335333,
      "grad_norm": 2.312746047973633,
      "learning_rate": 3.062592098800084e-05,
      "loss": 0.7614,
      "step": 7106
    },
    {
      "epoch": 2.2440795705715186,
      "grad_norm": 2.341064691543579,
      "learning_rate": 3.062241246228335e-05,
      "loss": 0.698,
      "step": 7107
    },
    {
      "epoch": 2.2443953268077044,
      "grad_norm": 2.332651376724243,
      "learning_rate": 3.0618903936565855e-05,
      "loss": 0.7466,
      "step": 7108
    },
    {
      "epoch": 2.24471108304389,
      "grad_norm": 2.0048224925994873,
      "learning_rate": 3.061539541084836e-05,
      "loss": 0.7024,
      "step": 7109
    },
    {
      "epoch": 2.245026839280076,
      "grad_norm": 2.422097682952881,
      "learning_rate": 3.061188688513087e-05,
      "loss": 0.7839,
      "step": 7110
    },
    {
      "epoch": 2.2453425955162616,
      "grad_norm": 2.3646738529205322,
      "learning_rate": 3.0608378359413376e-05,
      "loss": 0.7203,
      "step": 7111
    },
    {
      "epoch": 2.2456583517524473,
      "grad_norm": 2.891148567199707,
      "learning_rate": 3.060486983369588e-05,
      "loss": 0.7988,
      "step": 7112
    },
    {
      "epoch": 2.2459741079886326,
      "grad_norm": 2.5557985305786133,
      "learning_rate": 3.060136130797839e-05,
      "loss": 0.7607,
      "step": 7113
    },
    {
      "epoch": 2.2462898642248184,
      "grad_norm": 2.367520809173584,
      "learning_rate": 3.0597852782260896e-05,
      "loss": 0.7053,
      "step": 7114
    },
    {
      "epoch": 2.246605620461004,
      "grad_norm": 2.635014295578003,
      "learning_rate": 3.05943442565434e-05,
      "loss": 0.7035,
      "step": 7115
    },
    {
      "epoch": 2.24692137669719,
      "grad_norm": 2.514634132385254,
      "learning_rate": 3.059083573082591e-05,
      "loss": 0.7469,
      "step": 7116
    },
    {
      "epoch": 2.2472371329333756,
      "grad_norm": 2.337655782699585,
      "learning_rate": 3.058732720510841e-05,
      "loss": 0.7578,
      "step": 7117
    },
    {
      "epoch": 2.247552889169561,
      "grad_norm": 2.3231916427612305,
      "learning_rate": 3.0583818679390917e-05,
      "loss": 0.726,
      "step": 7118
    },
    {
      "epoch": 2.2478686454057466,
      "grad_norm": 2.779519557952881,
      "learning_rate": 3.058031015367343e-05,
      "loss": 0.7697,
      "step": 7119
    },
    {
      "epoch": 2.2481844016419323,
      "grad_norm": 2.124307155609131,
      "learning_rate": 3.057680162795594e-05,
      "loss": 0.6881,
      "step": 7120
    },
    {
      "epoch": 2.248500157878118,
      "grad_norm": 2.7664802074432373,
      "learning_rate": 3.0573293102238444e-05,
      "loss": 0.7377,
      "step": 7121
    },
    {
      "epoch": 2.248815914114304,
      "grad_norm": 2.3181910514831543,
      "learning_rate": 3.056978457652095e-05,
      "loss": 0.7526,
      "step": 7122
    },
    {
      "epoch": 2.2491316703504896,
      "grad_norm": 2.2994322776794434,
      "learning_rate": 3.056627605080346e-05,
      "loss": 0.7209,
      "step": 7123
    },
    {
      "epoch": 2.2494474265866753,
      "grad_norm": 2.2735352516174316,
      "learning_rate": 3.0562767525085964e-05,
      "loss": 0.6598,
      "step": 7124
    },
    {
      "epoch": 2.2497631828228606,
      "grad_norm": 2.5197255611419678,
      "learning_rate": 3.0559258999368464e-05,
      "loss": 0.8179,
      "step": 7125
    },
    {
      "epoch": 2.2500789390590463,
      "grad_norm": 2.497270107269287,
      "learning_rate": 3.055575047365097e-05,
      "loss": 0.7014,
      "step": 7126
    },
    {
      "epoch": 2.250394695295232,
      "grad_norm": 2.405535936355591,
      "learning_rate": 3.055224194793348e-05,
      "loss": 0.6831,
      "step": 7127
    },
    {
      "epoch": 2.250710451531418,
      "grad_norm": 2.3084027767181396,
      "learning_rate": 3.0548733422215985e-05,
      "loss": 0.7223,
      "step": 7128
    },
    {
      "epoch": 2.2510262077676035,
      "grad_norm": 2.435748815536499,
      "learning_rate": 3.054522489649849e-05,
      "loss": 0.6787,
      "step": 7129
    },
    {
      "epoch": 2.251341964003789,
      "grad_norm": 2.5228536128997803,
      "learning_rate": 3.0541716370781e-05,
      "loss": 0.7126,
      "step": 7130
    },
    {
      "epoch": 2.2516577202399746,
      "grad_norm": 2.464480400085449,
      "learning_rate": 3.0538207845063505e-05,
      "loss": 0.7688,
      "step": 7131
    },
    {
      "epoch": 2.2519734764761603,
      "grad_norm": 2.805652379989624,
      "learning_rate": 3.053469931934601e-05,
      "loss": 0.7503,
      "step": 7132
    },
    {
      "epoch": 2.252289232712346,
      "grad_norm": 2.765622854232788,
      "learning_rate": 3.053119079362852e-05,
      "loss": 0.7375,
      "step": 7133
    },
    {
      "epoch": 2.252604988948532,
      "grad_norm": 2.148521900177002,
      "learning_rate": 3.0527682267911026e-05,
      "loss": 0.687,
      "step": 7134
    },
    {
      "epoch": 2.2529207451847175,
      "grad_norm": 7.946035861968994,
      "learning_rate": 3.052417374219353e-05,
      "loss": 0.6901,
      "step": 7135
    },
    {
      "epoch": 2.2532365014209033,
      "grad_norm": 2.1461427211761475,
      "learning_rate": 3.052066521647604e-05,
      "loss": 0.625,
      "step": 7136
    },
    {
      "epoch": 2.2535522576570886,
      "grad_norm": 2.4350600242614746,
      "learning_rate": 3.0517156690758546e-05,
      "loss": 0.7433,
      "step": 7137
    },
    {
      "epoch": 2.2538680138932743,
      "grad_norm": 2.779191255569458,
      "learning_rate": 3.0513648165041053e-05,
      "loss": 0.9322,
      "step": 7138
    },
    {
      "epoch": 2.25418377012946,
      "grad_norm": 2.7176315784454346,
      "learning_rate": 3.051013963932356e-05,
      "loss": 0.7588,
      "step": 7139
    },
    {
      "epoch": 2.2544995263656458,
      "grad_norm": 2.8286473751068115,
      "learning_rate": 3.0506631113606066e-05,
      "loss": 0.7454,
      "step": 7140
    },
    {
      "epoch": 2.2548152826018315,
      "grad_norm": 2.5955464839935303,
      "learning_rate": 3.0503122587888573e-05,
      "loss": 0.7025,
      "step": 7141
    },
    {
      "epoch": 2.2551310388380172,
      "grad_norm": 2.48018741607666,
      "learning_rate": 3.0499614062171077e-05,
      "loss": 0.6958,
      "step": 7142
    },
    {
      "epoch": 2.2554467950742025,
      "grad_norm": 2.6654248237609863,
      "learning_rate": 3.0496105536453583e-05,
      "loss": 0.7862,
      "step": 7143
    },
    {
      "epoch": 2.2557625513103883,
      "grad_norm": 2.5904946327209473,
      "learning_rate": 3.049259701073609e-05,
      "loss": 0.801,
      "step": 7144
    },
    {
      "epoch": 2.256078307546574,
      "grad_norm": 2.0754454135894775,
      "learning_rate": 3.0489088485018597e-05,
      "loss": 0.7598,
      "step": 7145
    },
    {
      "epoch": 2.2563940637827598,
      "grad_norm": 2.525352716445923,
      "learning_rate": 3.0485579959301104e-05,
      "loss": 0.7976,
      "step": 7146
    },
    {
      "epoch": 2.2567098200189455,
      "grad_norm": 2.344233751296997,
      "learning_rate": 3.0482071433583607e-05,
      "loss": 0.7225,
      "step": 7147
    },
    {
      "epoch": 2.2570255762551312,
      "grad_norm": 2.2833056449890137,
      "learning_rate": 3.0478562907866114e-05,
      "loss": 0.7292,
      "step": 7148
    },
    {
      "epoch": 2.2573413324913165,
      "grad_norm": 2.550870418548584,
      "learning_rate": 3.047505438214862e-05,
      "loss": 0.7341,
      "step": 7149
    },
    {
      "epoch": 2.2576570887275023,
      "grad_norm": 2.5248491764068604,
      "learning_rate": 3.0471545856431128e-05,
      "loss": 0.7248,
      "step": 7150
    },
    {
      "epoch": 2.257972844963688,
      "grad_norm": 2.415940999984741,
      "learning_rate": 3.0468037330713635e-05,
      "loss": 0.7307,
      "step": 7151
    },
    {
      "epoch": 2.2582886011998737,
      "grad_norm": 2.240694761276245,
      "learning_rate": 3.0464528804996138e-05,
      "loss": 0.8362,
      "step": 7152
    },
    {
      "epoch": 2.2586043574360595,
      "grad_norm": 2.5536751747131348,
      "learning_rate": 3.0461020279278645e-05,
      "loss": 0.7504,
      "step": 7153
    },
    {
      "epoch": 2.258920113672245,
      "grad_norm": 2.2589380741119385,
      "learning_rate": 3.045751175356116e-05,
      "loss": 0.7076,
      "step": 7154
    },
    {
      "epoch": 2.2592358699084305,
      "grad_norm": 2.3906195163726807,
      "learning_rate": 3.0454003227843665e-05,
      "loss": 0.7097,
      "step": 7155
    },
    {
      "epoch": 2.2595516261446162,
      "grad_norm": 2.3084654808044434,
      "learning_rate": 3.045049470212617e-05,
      "loss": 0.7118,
      "step": 7156
    },
    {
      "epoch": 2.259867382380802,
      "grad_norm": 2.0279760360717773,
      "learning_rate": 3.0446986176408675e-05,
      "loss": 0.7853,
      "step": 7157
    },
    {
      "epoch": 2.2601831386169877,
      "grad_norm": 2.045332431793213,
      "learning_rate": 3.0443477650691182e-05,
      "loss": 0.6747,
      "step": 7158
    },
    {
      "epoch": 2.2604988948531735,
      "grad_norm": 2.1987600326538086,
      "learning_rate": 3.043996912497369e-05,
      "loss": 0.6962,
      "step": 7159
    },
    {
      "epoch": 2.260814651089359,
      "grad_norm": 2.0998942852020264,
      "learning_rate": 3.0436460599256196e-05,
      "loss": 0.7408,
      "step": 7160
    },
    {
      "epoch": 2.2611304073255445,
      "grad_norm": 1.9341390132904053,
      "learning_rate": 3.04329520735387e-05,
      "loss": 0.6853,
      "step": 7161
    },
    {
      "epoch": 2.2614461635617302,
      "grad_norm": 2.2693674564361572,
      "learning_rate": 3.0429443547821206e-05,
      "loss": 0.7549,
      "step": 7162
    },
    {
      "epoch": 2.261761919797916,
      "grad_norm": 2.476466655731201,
      "learning_rate": 3.0425935022103713e-05,
      "loss": 0.7856,
      "step": 7163
    },
    {
      "epoch": 2.2620776760341017,
      "grad_norm": 2.2677924633026123,
      "learning_rate": 3.042242649638622e-05,
      "loss": 0.8106,
      "step": 7164
    },
    {
      "epoch": 2.2623934322702874,
      "grad_norm": 2.1334965229034424,
      "learning_rate": 3.0418917970668727e-05,
      "loss": 0.8069,
      "step": 7165
    },
    {
      "epoch": 2.262709188506473,
      "grad_norm": 2.022559881210327,
      "learning_rate": 3.041540944495123e-05,
      "loss": 0.6823,
      "step": 7166
    },
    {
      "epoch": 2.263024944742659,
      "grad_norm": 2.3671505451202393,
      "learning_rate": 3.0411900919233737e-05,
      "loss": 0.7607,
      "step": 7167
    },
    {
      "epoch": 2.263340700978844,
      "grad_norm": 2.2883100509643555,
      "learning_rate": 3.0408392393516244e-05,
      "loss": 0.819,
      "step": 7168
    },
    {
      "epoch": 2.26365645721503,
      "grad_norm": 2.3608415126800537,
      "learning_rate": 3.040488386779875e-05,
      "loss": 0.7373,
      "step": 7169
    },
    {
      "epoch": 2.2639722134512157,
      "grad_norm": 2.4908761978149414,
      "learning_rate": 3.0401375342081257e-05,
      "loss": 0.7519,
      "step": 7170
    },
    {
      "epoch": 2.2642879696874014,
      "grad_norm": 2.4139370918273926,
      "learning_rate": 3.0397866816363767e-05,
      "loss": 0.7694,
      "step": 7171
    },
    {
      "epoch": 2.264603725923587,
      "grad_norm": 2.162668466567993,
      "learning_rate": 3.0394358290646274e-05,
      "loss": 0.7656,
      "step": 7172
    },
    {
      "epoch": 2.2649194821597725,
      "grad_norm": 2.400251626968384,
      "learning_rate": 3.039084976492878e-05,
      "loss": 0.8015,
      "step": 7173
    },
    {
      "epoch": 2.265235238395958,
      "grad_norm": 2.2222023010253906,
      "learning_rate": 3.0387341239211288e-05,
      "loss": 0.7896,
      "step": 7174
    },
    {
      "epoch": 2.265550994632144,
      "grad_norm": 4.227694034576416,
      "learning_rate": 3.038383271349379e-05,
      "loss": 0.8221,
      "step": 7175
    },
    {
      "epoch": 2.2658667508683297,
      "grad_norm": 2.6431877613067627,
      "learning_rate": 3.0380324187776298e-05,
      "loss": 0.7798,
      "step": 7176
    },
    {
      "epoch": 2.2661825071045154,
      "grad_norm": 2.7466747760772705,
      "learning_rate": 3.0376815662058805e-05,
      "loss": 0.7712,
      "step": 7177
    },
    {
      "epoch": 2.266498263340701,
      "grad_norm": 2.1920578479766846,
      "learning_rate": 3.0373307136341312e-05,
      "loss": 0.7062,
      "step": 7178
    },
    {
      "epoch": 2.266814019576887,
      "grad_norm": 2.567270278930664,
      "learning_rate": 3.036979861062382e-05,
      "loss": 0.8018,
      "step": 7179
    },
    {
      "epoch": 2.267129775813072,
      "grad_norm": 2.1278891563415527,
      "learning_rate": 3.0366290084906322e-05,
      "loss": 0.7606,
      "step": 7180
    },
    {
      "epoch": 2.267445532049258,
      "grad_norm": 2.7066848278045654,
      "learning_rate": 3.036278155918883e-05,
      "loss": 0.7306,
      "step": 7181
    },
    {
      "epoch": 2.2677612882854437,
      "grad_norm": 2.763791561126709,
      "learning_rate": 3.0359273033471336e-05,
      "loss": 0.7851,
      "step": 7182
    },
    {
      "epoch": 2.2680770445216294,
      "grad_norm": 2.4486947059631348,
      "learning_rate": 3.0355764507753842e-05,
      "loss": 0.7221,
      "step": 7183
    },
    {
      "epoch": 2.268392800757815,
      "grad_norm": 2.204829454421997,
      "learning_rate": 3.035225598203635e-05,
      "loss": 0.6519,
      "step": 7184
    },
    {
      "epoch": 2.2687085569940004,
      "grad_norm": 2.8860323429107666,
      "learning_rate": 3.0348747456318856e-05,
      "loss": 0.7514,
      "step": 7185
    },
    {
      "epoch": 2.269024313230186,
      "grad_norm": 2.211028575897217,
      "learning_rate": 3.034523893060136e-05,
      "loss": 0.6903,
      "step": 7186
    },
    {
      "epoch": 2.269340069466372,
      "grad_norm": 3.034632682800293,
      "learning_rate": 3.0341730404883866e-05,
      "loss": 0.7039,
      "step": 7187
    },
    {
      "epoch": 2.2696558257025576,
      "grad_norm": 2.561779499053955,
      "learning_rate": 3.0338221879166373e-05,
      "loss": 0.637,
      "step": 7188
    },
    {
      "epoch": 2.2699715819387434,
      "grad_norm": 2.656236171722412,
      "learning_rate": 3.0334713353448883e-05,
      "loss": 0.8267,
      "step": 7189
    },
    {
      "epoch": 2.270287338174929,
      "grad_norm": 2.5739450454711914,
      "learning_rate": 3.033120482773139e-05,
      "loss": 0.831,
      "step": 7190
    },
    {
      "epoch": 2.270603094411115,
      "grad_norm": 2.533402442932129,
      "learning_rate": 3.0327696302013897e-05,
      "loss": 0.7025,
      "step": 7191
    },
    {
      "epoch": 2.2709188506473,
      "grad_norm": 2.2591474056243896,
      "learning_rate": 3.0324187776296404e-05,
      "loss": 0.7571,
      "step": 7192
    },
    {
      "epoch": 2.271234606883486,
      "grad_norm": 3.238717555999756,
      "learning_rate": 3.032067925057891e-05,
      "loss": 0.8664,
      "step": 7193
    },
    {
      "epoch": 2.2715503631196716,
      "grad_norm": 2.675039529800415,
      "learning_rate": 3.0317170724861414e-05,
      "loss": 0.776,
      "step": 7194
    },
    {
      "epoch": 2.2718661193558574,
      "grad_norm": 2.761537551879883,
      "learning_rate": 3.031366219914392e-05,
      "loss": 0.8753,
      "step": 7195
    },
    {
      "epoch": 2.272181875592043,
      "grad_norm": 2.130873441696167,
      "learning_rate": 3.0310153673426428e-05,
      "loss": 0.8223,
      "step": 7196
    },
    {
      "epoch": 2.2724976318282284,
      "grad_norm": 2.15651273727417,
      "learning_rate": 3.0306645147708934e-05,
      "loss": 0.8242,
      "step": 7197
    },
    {
      "epoch": 2.272813388064414,
      "grad_norm": 2.460064649581909,
      "learning_rate": 3.030313662199144e-05,
      "loss": 0.7727,
      "step": 7198
    },
    {
      "epoch": 2.2731291443006,
      "grad_norm": 2.3717041015625,
      "learning_rate": 3.0299628096273948e-05,
      "loss": 0.6641,
      "step": 7199
    },
    {
      "epoch": 2.2734449005367856,
      "grad_norm": 2.1239569187164307,
      "learning_rate": 3.029611957055645e-05,
      "loss": 0.6215,
      "step": 7200
    },
    {
      "epoch": 2.2734449005367856,
      "eval_loss": 0.7997857928276062,
      "eval_runtime": 4.7621,
      "eval_samples_per_second": 522.88,
      "eval_steps_per_second": 4.41,
      "step": 7200
    },
    {
      "epoch": 2.2737606567729713,
      "grad_norm": 2.479995012283325,
      "learning_rate": 3.029261104483896e-05,
      "loss": 0.7393,
      "step": 7201
    },
    {
      "epoch": 2.274076413009157,
      "grad_norm": 2.1790311336517334,
      "learning_rate": 3.0289102519121465e-05,
      "loss": 0.7556,
      "step": 7202
    },
    {
      "epoch": 2.274392169245343,
      "grad_norm": 2.037801742553711,
      "learning_rate": 3.0285593993403972e-05,
      "loss": 0.7108,
      "step": 7203
    },
    {
      "epoch": 2.274707925481528,
      "grad_norm": 2.3947386741638184,
      "learning_rate": 3.028208546768648e-05,
      "loss": 0.7981,
      "step": 7204
    },
    {
      "epoch": 2.275023681717714,
      "grad_norm": 2.578749895095825,
      "learning_rate": 3.0278576941968982e-05,
      "loss": 0.7852,
      "step": 7205
    },
    {
      "epoch": 2.2753394379538996,
      "grad_norm": 2.826115131378174,
      "learning_rate": 3.0275068416251496e-05,
      "loss": 0.7647,
      "step": 7206
    },
    {
      "epoch": 2.2756551941900853,
      "grad_norm": 2.2345645427703857,
      "learning_rate": 3.0271559890534003e-05,
      "loss": 0.7635,
      "step": 7207
    },
    {
      "epoch": 2.275970950426271,
      "grad_norm": 2.5881919860839844,
      "learning_rate": 3.0268051364816506e-05,
      "loss": 0.7782,
      "step": 7208
    },
    {
      "epoch": 2.2762867066624564,
      "grad_norm": 2.515204429626465,
      "learning_rate": 3.0264542839099013e-05,
      "loss": 0.7574,
      "step": 7209
    },
    {
      "epoch": 2.276602462898642,
      "grad_norm": 2.371365785598755,
      "learning_rate": 3.026103431338152e-05,
      "loss": 0.7005,
      "step": 7210
    },
    {
      "epoch": 2.276918219134828,
      "grad_norm": 2.492860794067383,
      "learning_rate": 3.0257525787664026e-05,
      "loss": 0.7557,
      "step": 7211
    },
    {
      "epoch": 2.2772339753710136,
      "grad_norm": 2.5127508640289307,
      "learning_rate": 3.0254017261946533e-05,
      "loss": 0.8084,
      "step": 7212
    },
    {
      "epoch": 2.2775497316071993,
      "grad_norm": 2.3971304893493652,
      "learning_rate": 3.025050873622904e-05,
      "loss": 0.728,
      "step": 7213
    },
    {
      "epoch": 2.277865487843385,
      "grad_norm": 2.0799014568328857,
      "learning_rate": 3.0247000210511543e-05,
      "loss": 0.6832,
      "step": 7214
    },
    {
      "epoch": 2.278181244079571,
      "grad_norm": 2.623058319091797,
      "learning_rate": 3.024349168479405e-05,
      "loss": 0.7656,
      "step": 7215
    },
    {
      "epoch": 2.278497000315756,
      "grad_norm": 2.8353917598724365,
      "learning_rate": 3.0239983159076557e-05,
      "loss": 0.7663,
      "step": 7216
    },
    {
      "epoch": 2.278812756551942,
      "grad_norm": 2.4248063564300537,
      "learning_rate": 3.0236474633359064e-05,
      "loss": 0.7898,
      "step": 7217
    },
    {
      "epoch": 2.2791285127881276,
      "grad_norm": 3.1118452548980713,
      "learning_rate": 3.023296610764157e-05,
      "loss": 0.8454,
      "step": 7218
    },
    {
      "epoch": 2.2794442690243133,
      "grad_norm": 2.870692729949951,
      "learning_rate": 3.0229457581924074e-05,
      "loss": 0.7768,
      "step": 7219
    },
    {
      "epoch": 2.279760025260499,
      "grad_norm": 2.666354179382324,
      "learning_rate": 3.022594905620658e-05,
      "loss": 0.7329,
      "step": 7220
    },
    {
      "epoch": 2.2800757814966843,
      "grad_norm": 2.4354429244995117,
      "learning_rate": 3.0222440530489088e-05,
      "loss": 0.8391,
      "step": 7221
    },
    {
      "epoch": 2.28039153773287,
      "grad_norm": 2.533961296081543,
      "learning_rate": 3.0218932004771595e-05,
      "loss": 0.9197,
      "step": 7222
    },
    {
      "epoch": 2.280707293969056,
      "grad_norm": 2.1453311443328857,
      "learning_rate": 3.02154234790541e-05,
      "loss": 0.6927,
      "step": 7223
    },
    {
      "epoch": 2.2810230502052415,
      "grad_norm": 2.3703842163085938,
      "learning_rate": 3.021191495333661e-05,
      "loss": 0.6827,
      "step": 7224
    },
    {
      "epoch": 2.2813388064414273,
      "grad_norm": 2.7555010318756104,
      "learning_rate": 3.020840642761912e-05,
      "loss": 0.8009,
      "step": 7225
    },
    {
      "epoch": 2.281654562677613,
      "grad_norm": 2.5225627422332764,
      "learning_rate": 3.0204897901901625e-05,
      "loss": 0.6893,
      "step": 7226
    },
    {
      "epoch": 2.2819703189137988,
      "grad_norm": 2.4972474575042725,
      "learning_rate": 3.0201389376184132e-05,
      "loss": 0.6993,
      "step": 7227
    },
    {
      "epoch": 2.282286075149984,
      "grad_norm": 2.2924482822418213,
      "learning_rate": 3.0197880850466635e-05,
      "loss": 0.7017,
      "step": 7228
    },
    {
      "epoch": 2.28260183138617,
      "grad_norm": 2.373781681060791,
      "learning_rate": 3.0194372324749142e-05,
      "loss": 0.7862,
      "step": 7229
    },
    {
      "epoch": 2.2829175876223555,
      "grad_norm": 3.0161471366882324,
      "learning_rate": 3.019086379903165e-05,
      "loss": 0.6698,
      "step": 7230
    },
    {
      "epoch": 2.2832333438585413,
      "grad_norm": 2.474897861480713,
      "learning_rate": 3.0187355273314156e-05,
      "loss": 0.8156,
      "step": 7231
    },
    {
      "epoch": 2.283549100094727,
      "grad_norm": 2.3785996437072754,
      "learning_rate": 3.0183846747596663e-05,
      "loss": 0.6828,
      "step": 7232
    },
    {
      "epoch": 2.2838648563309123,
      "grad_norm": 2.9090404510498047,
      "learning_rate": 3.0180338221879166e-05,
      "loss": 0.855,
      "step": 7233
    },
    {
      "epoch": 2.284180612567098,
      "grad_norm": 2.4153780937194824,
      "learning_rate": 3.0176829696161673e-05,
      "loss": 0.7434,
      "step": 7234
    },
    {
      "epoch": 2.2844963688032838,
      "grad_norm": 2.5012056827545166,
      "learning_rate": 3.017332117044418e-05,
      "loss": 0.7845,
      "step": 7235
    },
    {
      "epoch": 2.2848121250394695,
      "grad_norm": 2.2408108711242676,
      "learning_rate": 3.0169812644726687e-05,
      "loss": 0.7497,
      "step": 7236
    },
    {
      "epoch": 2.2851278812756552,
      "grad_norm": 2.394666910171509,
      "learning_rate": 3.0166304119009193e-05,
      "loss": 0.7305,
      "step": 7237
    },
    {
      "epoch": 2.285443637511841,
      "grad_norm": 2.350139617919922,
      "learning_rate": 3.0162795593291697e-05,
      "loss": 0.7338,
      "step": 7238
    },
    {
      "epoch": 2.2857593937480267,
      "grad_norm": 2.5713658332824707,
      "learning_rate": 3.0159287067574204e-05,
      "loss": 0.7926,
      "step": 7239
    },
    {
      "epoch": 2.286075149984212,
      "grad_norm": 1.8615607023239136,
      "learning_rate": 3.015577854185671e-05,
      "loss": 0.6295,
      "step": 7240
    },
    {
      "epoch": 2.2863909062203978,
      "grad_norm": 2.1719486713409424,
      "learning_rate": 3.0152270016139224e-05,
      "loss": 0.7474,
      "step": 7241
    },
    {
      "epoch": 2.2867066624565835,
      "grad_norm": 2.6736607551574707,
      "learning_rate": 3.0148761490421727e-05,
      "loss": 0.749,
      "step": 7242
    },
    {
      "epoch": 2.2870224186927692,
      "grad_norm": 2.4535696506500244,
      "learning_rate": 3.0145252964704234e-05,
      "loss": 0.695,
      "step": 7243
    },
    {
      "epoch": 2.287338174928955,
      "grad_norm": 2.524822473526001,
      "learning_rate": 3.014174443898674e-05,
      "loss": 0.7213,
      "step": 7244
    },
    {
      "epoch": 2.2876539311651403,
      "grad_norm": 2.5155208110809326,
      "learning_rate": 3.0138235913269248e-05,
      "loss": 0.7648,
      "step": 7245
    },
    {
      "epoch": 2.287969687401326,
      "grad_norm": 2.2873315811157227,
      "learning_rate": 3.0134727387551755e-05,
      "loss": 0.7329,
      "step": 7246
    },
    {
      "epoch": 2.2882854436375117,
      "grad_norm": 1.9183903932571411,
      "learning_rate": 3.0131218861834258e-05,
      "loss": 0.7335,
      "step": 7247
    },
    {
      "epoch": 2.2886011998736975,
      "grad_norm": 2.5903446674346924,
      "learning_rate": 3.0127710336116765e-05,
      "loss": 0.8084,
      "step": 7248
    },
    {
      "epoch": 2.288916956109883,
      "grad_norm": 2.4433419704437256,
      "learning_rate": 3.0124201810399272e-05,
      "loss": 0.764,
      "step": 7249
    },
    {
      "epoch": 2.289232712346069,
      "grad_norm": 2.4299283027648926,
      "learning_rate": 3.012069328468178e-05,
      "loss": 0.7521,
      "step": 7250
    },
    {
      "epoch": 2.2895484685822547,
      "grad_norm": 2.851719617843628,
      "learning_rate": 3.0117184758964285e-05,
      "loss": 0.8123,
      "step": 7251
    },
    {
      "epoch": 2.28986422481844,
      "grad_norm": 7.80819845199585,
      "learning_rate": 3.011367623324679e-05,
      "loss": 0.7482,
      "step": 7252
    },
    {
      "epoch": 2.2901799810546257,
      "grad_norm": 2.2653703689575195,
      "learning_rate": 3.0110167707529296e-05,
      "loss": 0.699,
      "step": 7253
    },
    {
      "epoch": 2.2904957372908115,
      "grad_norm": 2.387716293334961,
      "learning_rate": 3.0106659181811802e-05,
      "loss": 0.7831,
      "step": 7254
    },
    {
      "epoch": 2.290811493526997,
      "grad_norm": 2.279770612716675,
      "learning_rate": 3.010315065609431e-05,
      "loss": 0.8056,
      "step": 7255
    },
    {
      "epoch": 2.291127249763183,
      "grad_norm": 2.272672414779663,
      "learning_rate": 3.0099642130376816e-05,
      "loss": 0.8212,
      "step": 7256
    },
    {
      "epoch": 2.2914430059993687,
      "grad_norm": 2.257164716720581,
      "learning_rate": 3.009613360465932e-05,
      "loss": 0.7527,
      "step": 7257
    },
    {
      "epoch": 2.291758762235554,
      "grad_norm": 2.6371030807495117,
      "learning_rate": 3.0092625078941826e-05,
      "loss": 0.8153,
      "step": 7258
    },
    {
      "epoch": 2.2920745184717397,
      "grad_norm": 2.1495399475097656,
      "learning_rate": 3.008911655322434e-05,
      "loss": 0.768,
      "step": 7259
    },
    {
      "epoch": 2.2923902747079254,
      "grad_norm": 2.133964776992798,
      "learning_rate": 3.0085608027506847e-05,
      "loss": 0.6462,
      "step": 7260
    },
    {
      "epoch": 2.292706030944111,
      "grad_norm": 2.32861065864563,
      "learning_rate": 3.008209950178935e-05,
      "loss": 0.7796,
      "step": 7261
    },
    {
      "epoch": 2.293021787180297,
      "grad_norm": 2.3527207374572754,
      "learning_rate": 3.0078590976071857e-05,
      "loss": 0.7367,
      "step": 7262
    },
    {
      "epoch": 2.2933375434164827,
      "grad_norm": 2.2888424396514893,
      "learning_rate": 3.0075082450354364e-05,
      "loss": 0.7619,
      "step": 7263
    },
    {
      "epoch": 2.293653299652668,
      "grad_norm": 2.232847213745117,
      "learning_rate": 3.007157392463687e-05,
      "loss": 0.8068,
      "step": 7264
    },
    {
      "epoch": 2.2939690558888537,
      "grad_norm": 2.19287109375,
      "learning_rate": 3.0068065398919377e-05,
      "loss": 0.7492,
      "step": 7265
    },
    {
      "epoch": 2.2942848121250394,
      "grad_norm": 2.5348098278045654,
      "learning_rate": 3.006455687320188e-05,
      "loss": 0.7454,
      "step": 7266
    },
    {
      "epoch": 2.294600568361225,
      "grad_norm": 1.8643932342529297,
      "learning_rate": 3.0061048347484388e-05,
      "loss": 0.69,
      "step": 7267
    },
    {
      "epoch": 2.294916324597411,
      "grad_norm": 2.3942253589630127,
      "learning_rate": 3.0057539821766894e-05,
      "loss": 0.804,
      "step": 7268
    },
    {
      "epoch": 2.2952320808335966,
      "grad_norm": 2.404303550720215,
      "learning_rate": 3.00540312960494e-05,
      "loss": 0.8237,
      "step": 7269
    },
    {
      "epoch": 2.295547837069782,
      "grad_norm": 1.9796595573425293,
      "learning_rate": 3.0050522770331908e-05,
      "loss": 0.7041,
      "step": 7270
    },
    {
      "epoch": 2.2958635933059677,
      "grad_norm": 1.940141201019287,
      "learning_rate": 3.004701424461441e-05,
      "loss": 0.7747,
      "step": 7271
    },
    {
      "epoch": 2.2961793495421534,
      "grad_norm": 2.5319035053253174,
      "learning_rate": 3.004350571889692e-05,
      "loss": 0.721,
      "step": 7272
    },
    {
      "epoch": 2.296495105778339,
      "grad_norm": 2.0930335521698,
      "learning_rate": 3.0039997193179425e-05,
      "loss": 0.737,
      "step": 7273
    },
    {
      "epoch": 2.296810862014525,
      "grad_norm": 2.406712770462036,
      "learning_rate": 3.0036488667461932e-05,
      "loss": 0.7987,
      "step": 7274
    },
    {
      "epoch": 2.2971266182507106,
      "grad_norm": 2.8362886905670166,
      "learning_rate": 3.003298014174444e-05,
      "loss": 0.74,
      "step": 7275
    },
    {
      "epoch": 2.297442374486896,
      "grad_norm": 2.115239381790161,
      "learning_rate": 3.002947161602695e-05,
      "loss": 0.7273,
      "step": 7276
    },
    {
      "epoch": 2.2977581307230817,
      "grad_norm": 1.8993161916732788,
      "learning_rate": 3.0025963090309456e-05,
      "loss": 0.7241,
      "step": 7277
    },
    {
      "epoch": 2.2980738869592674,
      "grad_norm": 2.135775566101074,
      "learning_rate": 3.0022454564591963e-05,
      "loss": 0.6881,
      "step": 7278
    },
    {
      "epoch": 2.298389643195453,
      "grad_norm": 2.130557060241699,
      "learning_rate": 3.001894603887447e-05,
      "loss": 0.695,
      "step": 7279
    },
    {
      "epoch": 2.298705399431639,
      "grad_norm": 2.2168092727661133,
      "learning_rate": 3.0015437513156973e-05,
      "loss": 0.7051,
      "step": 7280
    },
    {
      "epoch": 2.2990211556678246,
      "grad_norm": 2.1352195739746094,
      "learning_rate": 3.001192898743948e-05,
      "loss": 0.7272,
      "step": 7281
    },
    {
      "epoch": 2.2993369119040103,
      "grad_norm": 2.1447126865386963,
      "learning_rate": 3.0008420461721986e-05,
      "loss": 0.7194,
      "step": 7282
    },
    {
      "epoch": 2.2996526681401956,
      "grad_norm": 2.1734869480133057,
      "learning_rate": 3.0004911936004493e-05,
      "loss": 0.7656,
      "step": 7283
    },
    {
      "epoch": 2.2999684243763814,
      "grad_norm": 2.460042953491211,
      "learning_rate": 3.0001403410287e-05,
      "loss": 0.713,
      "step": 7284
    },
    {
      "epoch": 2.300284180612567,
      "grad_norm": 1.99477219581604,
      "learning_rate": 2.9997894884569503e-05,
      "loss": 0.7571,
      "step": 7285
    },
    {
      "epoch": 2.300599936848753,
      "grad_norm": 2.1291539669036865,
      "learning_rate": 2.999438635885201e-05,
      "loss": 0.8196,
      "step": 7286
    },
    {
      "epoch": 2.3009156930849386,
      "grad_norm": 2.1616289615631104,
      "learning_rate": 2.9990877833134517e-05,
      "loss": 0.7062,
      "step": 7287
    },
    {
      "epoch": 2.301231449321124,
      "grad_norm": 2.254028081893921,
      "learning_rate": 2.9987369307417024e-05,
      "loss": 0.8098,
      "step": 7288
    },
    {
      "epoch": 2.3015472055573096,
      "grad_norm": 2.466989040374756,
      "learning_rate": 2.998386078169953e-05,
      "loss": 0.7321,
      "step": 7289
    },
    {
      "epoch": 2.3018629617934954,
      "grad_norm": 2.60029673576355,
      "learning_rate": 2.9980352255982038e-05,
      "loss": 0.757,
      "step": 7290
    },
    {
      "epoch": 2.302178718029681,
      "grad_norm": 2.155792713165283,
      "learning_rate": 2.997684373026454e-05,
      "loss": 0.6383,
      "step": 7291
    },
    {
      "epoch": 2.302494474265867,
      "grad_norm": 2.219571828842163,
      "learning_rate": 2.9973335204547048e-05,
      "loss": 0.6792,
      "step": 7292
    },
    {
      "epoch": 2.3028102305020526,
      "grad_norm": 2.595917224884033,
      "learning_rate": 2.9969826678829555e-05,
      "loss": 0.7857,
      "step": 7293
    },
    {
      "epoch": 2.3031259867382383,
      "grad_norm": 2.58137845993042,
      "learning_rate": 2.9966318153112065e-05,
      "loss": 0.7642,
      "step": 7294
    },
    {
      "epoch": 2.3034417429744236,
      "grad_norm": 3.1345338821411133,
      "learning_rate": 2.996280962739457e-05,
      "loss": 0.7063,
      "step": 7295
    },
    {
      "epoch": 2.3037574992106093,
      "grad_norm": 2.876901388168335,
      "learning_rate": 2.995930110167708e-05,
      "loss": 0.8255,
      "step": 7296
    },
    {
      "epoch": 2.304073255446795,
      "grad_norm": 2.5897183418273926,
      "learning_rate": 2.9955792575959585e-05,
      "loss": 0.7486,
      "step": 7297
    },
    {
      "epoch": 2.304389011682981,
      "grad_norm": 2.626077890396118,
      "learning_rate": 2.9952284050242092e-05,
      "loss": 0.7538,
      "step": 7298
    },
    {
      "epoch": 2.3047047679191666,
      "grad_norm": 2.694746494293213,
      "learning_rate": 2.9948775524524595e-05,
      "loss": 0.7358,
      "step": 7299
    },
    {
      "epoch": 2.305020524155352,
      "grad_norm": 2.6256754398345947,
      "learning_rate": 2.9945266998807102e-05,
      "loss": 0.7726,
      "step": 7300
    },
    {
      "epoch": 2.3053362803915376,
      "grad_norm": 2.3105099201202393,
      "learning_rate": 2.994175847308961e-05,
      "loss": 0.8097,
      "step": 7301
    },
    {
      "epoch": 2.3056520366277233,
      "grad_norm": 2.0281898975372314,
      "learning_rate": 2.9938249947372116e-05,
      "loss": 0.721,
      "step": 7302
    },
    {
      "epoch": 2.305967792863909,
      "grad_norm": 2.2041540145874023,
      "learning_rate": 2.9934741421654623e-05,
      "loss": 0.7666,
      "step": 7303
    },
    {
      "epoch": 2.306283549100095,
      "grad_norm": 2.409923553466797,
      "learning_rate": 2.993123289593713e-05,
      "loss": 0.7334,
      "step": 7304
    },
    {
      "epoch": 2.3065993053362805,
      "grad_norm": 2.463165521621704,
      "learning_rate": 2.9927724370219633e-05,
      "loss": 0.8002,
      "step": 7305
    },
    {
      "epoch": 2.3069150615724663,
      "grad_norm": 2.4527652263641357,
      "learning_rate": 2.992421584450214e-05,
      "loss": 0.6791,
      "step": 7306
    },
    {
      "epoch": 2.3072308178086516,
      "grad_norm": 2.4457430839538574,
      "learning_rate": 2.9920707318784647e-05,
      "loss": 0.8123,
      "step": 7307
    },
    {
      "epoch": 2.3075465740448373,
      "grad_norm": 2.505138635635376,
      "learning_rate": 2.9917198793067153e-05,
      "loss": 0.7776,
      "step": 7308
    },
    {
      "epoch": 2.307862330281023,
      "grad_norm": 2.2123680114746094,
      "learning_rate": 2.991369026734966e-05,
      "loss": 0.6773,
      "step": 7309
    },
    {
      "epoch": 2.308178086517209,
      "grad_norm": 2.52773380279541,
      "learning_rate": 2.9910181741632164e-05,
      "loss": 0.7525,
      "step": 7310
    },
    {
      "epoch": 2.3084938427533945,
      "grad_norm": 2.443413496017456,
      "learning_rate": 2.9906673215914677e-05,
      "loss": 0.8293,
      "step": 7311
    },
    {
      "epoch": 2.30880959898958,
      "grad_norm": 2.667217969894409,
      "learning_rate": 2.9903164690197184e-05,
      "loss": 0.7614,
      "step": 7312
    },
    {
      "epoch": 2.3091253552257656,
      "grad_norm": 2.1208202838897705,
      "learning_rate": 2.9899656164479687e-05,
      "loss": 0.6814,
      "step": 7313
    },
    {
      "epoch": 2.3094411114619513,
      "grad_norm": 2.703173875808716,
      "learning_rate": 2.9896147638762194e-05,
      "loss": 0.818,
      "step": 7314
    },
    {
      "epoch": 2.309756867698137,
      "grad_norm": 2.4993746280670166,
      "learning_rate": 2.98926391130447e-05,
      "loss": 0.8061,
      "step": 7315
    },
    {
      "epoch": 2.3100726239343228,
      "grad_norm": 2.5123698711395264,
      "learning_rate": 2.9889130587327208e-05,
      "loss": 0.6822,
      "step": 7316
    },
    {
      "epoch": 2.3103883801705085,
      "grad_norm": 2.139251708984375,
      "learning_rate": 2.9885622061609715e-05,
      "loss": 0.7371,
      "step": 7317
    },
    {
      "epoch": 2.3107041364066943,
      "grad_norm": 2.6280529499053955,
      "learning_rate": 2.988211353589222e-05,
      "loss": 0.6716,
      "step": 7318
    },
    {
      "epoch": 2.3110198926428795,
      "grad_norm": 2.7334890365600586,
      "learning_rate": 2.9878605010174725e-05,
      "loss": 0.8302,
      "step": 7319
    },
    {
      "epoch": 2.3113356488790653,
      "grad_norm": 2.2396504878997803,
      "learning_rate": 2.9875096484457232e-05,
      "loss": 0.6377,
      "step": 7320
    },
    {
      "epoch": 2.311651405115251,
      "grad_norm": 2.457080841064453,
      "learning_rate": 2.987158795873974e-05,
      "loss": 0.743,
      "step": 7321
    },
    {
      "epoch": 2.3119671613514368,
      "grad_norm": 2.3163976669311523,
      "learning_rate": 2.9868079433022245e-05,
      "loss": 0.6871,
      "step": 7322
    },
    {
      "epoch": 2.3122829175876225,
      "grad_norm": 2.6265053749084473,
      "learning_rate": 2.9864570907304752e-05,
      "loss": 0.7871,
      "step": 7323
    },
    {
      "epoch": 2.312598673823808,
      "grad_norm": 2.4547927379608154,
      "learning_rate": 2.9861062381587256e-05,
      "loss": 0.7826,
      "step": 7324
    },
    {
      "epoch": 2.3129144300599935,
      "grad_norm": 2.107273578643799,
      "learning_rate": 2.9857553855869762e-05,
      "loss": 0.669,
      "step": 7325
    },
    {
      "epoch": 2.3132301862961793,
      "grad_norm": 2.1450693607330322,
      "learning_rate": 2.985404533015227e-05,
      "loss": 0.7044,
      "step": 7326
    },
    {
      "epoch": 2.313545942532365,
      "grad_norm": 2.6026840209960938,
      "learning_rate": 2.9850536804434776e-05,
      "loss": 0.7817,
      "step": 7327
    },
    {
      "epoch": 2.3138616987685507,
      "grad_norm": 2.221754550933838,
      "learning_rate": 2.9847028278717283e-05,
      "loss": 0.7248,
      "step": 7328
    },
    {
      "epoch": 2.3141774550047365,
      "grad_norm": 2.557701587677002,
      "learning_rate": 2.9843519752999793e-05,
      "loss": 0.7729,
      "step": 7329
    },
    {
      "epoch": 2.314493211240922,
      "grad_norm": 2.5737781524658203,
      "learning_rate": 2.98400112272823e-05,
      "loss": 0.7886,
      "step": 7330
    },
    {
      "epoch": 2.3148089674771075,
      "grad_norm": 2.699639081954956,
      "learning_rate": 2.9836502701564807e-05,
      "loss": 0.775,
      "step": 7331
    },
    {
      "epoch": 2.3151247237132933,
      "grad_norm": 3.0864360332489014,
      "learning_rate": 2.9832994175847314e-05,
      "loss": 0.7101,
      "step": 7332
    },
    {
      "epoch": 2.315440479949479,
      "grad_norm": 2.416002035140991,
      "learning_rate": 2.9829485650129817e-05,
      "loss": 0.7057,
      "step": 7333
    },
    {
      "epoch": 2.3157562361856647,
      "grad_norm": 3.2383041381835938,
      "learning_rate": 2.9825977124412324e-05,
      "loss": 0.8272,
      "step": 7334
    },
    {
      "epoch": 2.3160719924218505,
      "grad_norm": 2.371309757232666,
      "learning_rate": 2.982246859869483e-05,
      "loss": 0.6538,
      "step": 7335
    },
    {
      "epoch": 2.3163877486580358,
      "grad_norm": 2.7523980140686035,
      "learning_rate": 2.9818960072977337e-05,
      "loss": 0.7539,
      "step": 7336
    },
    {
      "epoch": 2.3167035048942215,
      "grad_norm": 2.741077423095703,
      "learning_rate": 2.9815451547259844e-05,
      "loss": 0.7687,
      "step": 7337
    },
    {
      "epoch": 2.3170192611304072,
      "grad_norm": 2.6032168865203857,
      "learning_rate": 2.9811943021542348e-05,
      "loss": 0.7213,
      "step": 7338
    },
    {
      "epoch": 2.317335017366593,
      "grad_norm": 2.4549508094787598,
      "learning_rate": 2.9808434495824854e-05,
      "loss": 0.7339,
      "step": 7339
    },
    {
      "epoch": 2.3176507736027787,
      "grad_norm": 2.5298562049865723,
      "learning_rate": 2.980492597010736e-05,
      "loss": 0.7352,
      "step": 7340
    },
    {
      "epoch": 2.3179665298389645,
      "grad_norm": 2.5519349575042725,
      "learning_rate": 2.9801417444389868e-05,
      "loss": 0.7026,
      "step": 7341
    },
    {
      "epoch": 2.31828228607515,
      "grad_norm": 2.7343759536743164,
      "learning_rate": 2.9797908918672375e-05,
      "loss": 0.8394,
      "step": 7342
    },
    {
      "epoch": 2.3185980423113355,
      "grad_norm": 2.7485849857330322,
      "learning_rate": 2.979440039295488e-05,
      "loss": 0.7291,
      "step": 7343
    },
    {
      "epoch": 2.318913798547521,
      "grad_norm": 2.4107861518859863,
      "learning_rate": 2.9790891867237385e-05,
      "loss": 0.7831,
      "step": 7344
    },
    {
      "epoch": 2.319229554783707,
      "grad_norm": 2.2472264766693115,
      "learning_rate": 2.9787383341519892e-05,
      "loss": 0.6826,
      "step": 7345
    },
    {
      "epoch": 2.3195453110198927,
      "grad_norm": 2.4410629272460938,
      "learning_rate": 2.9783874815802406e-05,
      "loss": 0.78,
      "step": 7346
    },
    {
      "epoch": 2.3198610672560784,
      "grad_norm": 2.34069561958313,
      "learning_rate": 2.978036629008491e-05,
      "loss": 0.7601,
      "step": 7347
    },
    {
      "epoch": 2.3201768234922637,
      "grad_norm": 2.2860453128814697,
      "learning_rate": 2.9776857764367416e-05,
      "loss": 0.7638,
      "step": 7348
    },
    {
      "epoch": 2.3204925797284495,
      "grad_norm": 2.4961392879486084,
      "learning_rate": 2.9773349238649923e-05,
      "loss": 0.7022,
      "step": 7349
    },
    {
      "epoch": 2.320808335964635,
      "grad_norm": 2.299290180206299,
      "learning_rate": 2.976984071293243e-05,
      "loss": 0.8174,
      "step": 7350
    },
    {
      "epoch": 2.321124092200821,
      "grad_norm": 2.8015012741088867,
      "learning_rate": 2.9766332187214936e-05,
      "loss": 0.7625,
      "step": 7351
    },
    {
      "epoch": 2.3214398484370067,
      "grad_norm": 2.360774040222168,
      "learning_rate": 2.976282366149744e-05,
      "loss": 0.7876,
      "step": 7352
    },
    {
      "epoch": 2.3217556046731924,
      "grad_norm": 2.2719409465789795,
      "learning_rate": 2.9759315135779946e-05,
      "loss": 0.7547,
      "step": 7353
    },
    {
      "epoch": 2.322071360909378,
      "grad_norm": 2.197556257247925,
      "learning_rate": 2.9755806610062453e-05,
      "loss": 0.7484,
      "step": 7354
    },
    {
      "epoch": 2.3223871171455635,
      "grad_norm": 1.951106071472168,
      "learning_rate": 2.975229808434496e-05,
      "loss": 0.7407,
      "step": 7355
    },
    {
      "epoch": 2.322702873381749,
      "grad_norm": 2.1883058547973633,
      "learning_rate": 2.9748789558627467e-05,
      "loss": 0.6827,
      "step": 7356
    },
    {
      "epoch": 2.323018629617935,
      "grad_norm": 2.715224027633667,
      "learning_rate": 2.974528103290997e-05,
      "loss": 0.7179,
      "step": 7357
    },
    {
      "epoch": 2.3233343858541207,
      "grad_norm": 2.380234956741333,
      "learning_rate": 2.9741772507192477e-05,
      "loss": 0.7292,
      "step": 7358
    },
    {
      "epoch": 2.3236501420903064,
      "grad_norm": 2.3175301551818848,
      "learning_rate": 2.9738263981474984e-05,
      "loss": 0.7538,
      "step": 7359
    },
    {
      "epoch": 2.323965898326492,
      "grad_norm": 1.99830162525177,
      "learning_rate": 2.973475545575749e-05,
      "loss": 0.6562,
      "step": 7360
    },
    {
      "epoch": 2.3242816545626774,
      "grad_norm": 2.1011064052581787,
      "learning_rate": 2.9731246930039998e-05,
      "loss": 0.6784,
      "step": 7361
    },
    {
      "epoch": 2.324597410798863,
      "grad_norm": 2.609375476837158,
      "learning_rate": 2.97277384043225e-05,
      "loss": 0.7836,
      "step": 7362
    },
    {
      "epoch": 2.324913167035049,
      "grad_norm": 2.055600881576538,
      "learning_rate": 2.9724229878605008e-05,
      "loss": 0.822,
      "step": 7363
    },
    {
      "epoch": 2.3252289232712346,
      "grad_norm": 2.090005874633789,
      "learning_rate": 2.972072135288752e-05,
      "loss": 0.67,
      "step": 7364
    },
    {
      "epoch": 2.3255446795074204,
      "grad_norm": 2.3764986991882324,
      "learning_rate": 2.9717212827170028e-05,
      "loss": 0.7123,
      "step": 7365
    },
    {
      "epoch": 2.325860435743606,
      "grad_norm": 2.8861587047576904,
      "learning_rate": 2.971370430145253e-05,
      "loss": 0.7731,
      "step": 7366
    },
    {
      "epoch": 2.3261761919797914,
      "grad_norm": 2.256805181503296,
      "learning_rate": 2.971019577573504e-05,
      "loss": 0.765,
      "step": 7367
    },
    {
      "epoch": 2.326491948215977,
      "grad_norm": 2.054564952850342,
      "learning_rate": 2.9706687250017545e-05,
      "loss": 0.6604,
      "step": 7368
    },
    {
      "epoch": 2.326807704452163,
      "grad_norm": 2.139801025390625,
      "learning_rate": 2.9703178724300052e-05,
      "loss": 0.6979,
      "step": 7369
    },
    {
      "epoch": 2.3271234606883486,
      "grad_norm": 2.3885324001312256,
      "learning_rate": 2.969967019858256e-05,
      "loss": 0.8179,
      "step": 7370
    },
    {
      "epoch": 2.3274392169245344,
      "grad_norm": 2.275419235229492,
      "learning_rate": 2.9696161672865062e-05,
      "loss": 0.7228,
      "step": 7371
    },
    {
      "epoch": 2.32775497316072,
      "grad_norm": 2.6995534896850586,
      "learning_rate": 2.969265314714757e-05,
      "loss": 0.8168,
      "step": 7372
    },
    {
      "epoch": 2.3280707293969054,
      "grad_norm": 2.305575132369995,
      "learning_rate": 2.9689144621430076e-05,
      "loss": 0.7038,
      "step": 7373
    },
    {
      "epoch": 2.328386485633091,
      "grad_norm": 2.277045965194702,
      "learning_rate": 2.9685636095712583e-05,
      "loss": 0.7411,
      "step": 7374
    },
    {
      "epoch": 2.328702241869277,
      "grad_norm": 2.330233573913574,
      "learning_rate": 2.968212756999509e-05,
      "loss": 0.6619,
      "step": 7375
    },
    {
      "epoch": 2.3290179981054626,
      "grad_norm": 2.6195642948150635,
      "learning_rate": 2.9678619044277593e-05,
      "loss": 0.6614,
      "step": 7376
    },
    {
      "epoch": 2.3293337543416484,
      "grad_norm": 1.9686079025268555,
      "learning_rate": 2.96751105185601e-05,
      "loss": 0.7511,
      "step": 7377
    },
    {
      "epoch": 2.329649510577834,
      "grad_norm": 1.986945390701294,
      "learning_rate": 2.9671601992842607e-05,
      "loss": 0.6227,
      "step": 7378
    },
    {
      "epoch": 2.3299652668140194,
      "grad_norm": 2.0389959812164307,
      "learning_rate": 2.9668093467125113e-05,
      "loss": 0.768,
      "step": 7379
    },
    {
      "epoch": 2.330281023050205,
      "grad_norm": 2.2983319759368896,
      "learning_rate": 2.966458494140762e-05,
      "loss": 0.8504,
      "step": 7380
    },
    {
      "epoch": 2.330596779286391,
      "grad_norm": 1.8706932067871094,
      "learning_rate": 2.966107641569013e-05,
      "loss": 0.6849,
      "step": 7381
    },
    {
      "epoch": 2.3309125355225766,
      "grad_norm": 2.4476423263549805,
      "learning_rate": 2.9657567889972637e-05,
      "loss": 0.7517,
      "step": 7382
    },
    {
      "epoch": 2.3312282917587623,
      "grad_norm": 2.3378243446350098,
      "learning_rate": 2.9654059364255144e-05,
      "loss": 0.6671,
      "step": 7383
    },
    {
      "epoch": 2.331544047994948,
      "grad_norm": 2.525559186935425,
      "learning_rate": 2.965055083853765e-05,
      "loss": 0.7148,
      "step": 7384
    },
    {
      "epoch": 2.3318598042311334,
      "grad_norm": 2.4138453006744385,
      "learning_rate": 2.9647042312820154e-05,
      "loss": 0.7901,
      "step": 7385
    },
    {
      "epoch": 2.332175560467319,
      "grad_norm": 2.5120716094970703,
      "learning_rate": 2.964353378710266e-05,
      "loss": 0.7479,
      "step": 7386
    },
    {
      "epoch": 2.332491316703505,
      "grad_norm": 2.065519332885742,
      "learning_rate": 2.9640025261385168e-05,
      "loss": 0.792,
      "step": 7387
    },
    {
      "epoch": 2.3328070729396906,
      "grad_norm": 2.5150938034057617,
      "learning_rate": 2.9636516735667675e-05,
      "loss": 0.7202,
      "step": 7388
    },
    {
      "epoch": 2.3331228291758763,
      "grad_norm": 2.034487724304199,
      "learning_rate": 2.963300820995018e-05,
      "loss": 0.7883,
      "step": 7389
    },
    {
      "epoch": 2.333438585412062,
      "grad_norm": 2.4592161178588867,
      "learning_rate": 2.9629499684232685e-05,
      "loss": 0.7753,
      "step": 7390
    },
    {
      "epoch": 2.3337543416482474,
      "grad_norm": 2.4378509521484375,
      "learning_rate": 2.9625991158515192e-05,
      "loss": 0.6757,
      "step": 7391
    },
    {
      "epoch": 2.334070097884433,
      "grad_norm": 2.4612927436828613,
      "learning_rate": 2.96224826327977e-05,
      "loss": 0.8291,
      "step": 7392
    },
    {
      "epoch": 2.334385854120619,
      "grad_norm": 2.348860502243042,
      "learning_rate": 2.9618974107080205e-05,
      "loss": 0.6545,
      "step": 7393
    },
    {
      "epoch": 2.3347016103568046,
      "grad_norm": 2.478714942932129,
      "learning_rate": 2.9615465581362712e-05,
      "loss": 0.7395,
      "step": 7394
    },
    {
      "epoch": 2.3350173665929903,
      "grad_norm": 2.8029351234436035,
      "learning_rate": 2.961195705564522e-05,
      "loss": 0.8049,
      "step": 7395
    },
    {
      "epoch": 2.335333122829176,
      "grad_norm": 2.272515296936035,
      "learning_rate": 2.9608448529927722e-05,
      "loss": 0.7218,
      "step": 7396
    },
    {
      "epoch": 2.335648879065362,
      "grad_norm": 2.2863895893096924,
      "learning_rate": 2.960494000421023e-05,
      "loss": 0.7828,
      "step": 7397
    },
    {
      "epoch": 2.335964635301547,
      "grad_norm": 2.1787779331207275,
      "learning_rate": 2.9601431478492736e-05,
      "loss": 0.718,
      "step": 7398
    },
    {
      "epoch": 2.336280391537733,
      "grad_norm": 2.4618868827819824,
      "learning_rate": 2.9597922952775246e-05,
      "loss": 0.6957,
      "step": 7399
    },
    {
      "epoch": 2.3365961477739186,
      "grad_norm": 2.4788291454315186,
      "learning_rate": 2.9594414427057753e-05,
      "loss": 0.7167,
      "step": 7400
    },
    {
      "epoch": 2.3365961477739186,
      "eval_loss": 0.8035600185394287,
      "eval_runtime": 4.6422,
      "eval_samples_per_second": 536.386,
      "eval_steps_per_second": 4.524,
      "step": 7400
    },
    {
      "epoch": 2.3369119040101043,
      "grad_norm": 2.235018253326416,
      "learning_rate": 2.959090590134026e-05,
      "loss": 0.6969,
      "step": 7401
    },
    {
      "epoch": 2.33722766024629,
      "grad_norm": 2.6314194202423096,
      "learning_rate": 2.9587397375622767e-05,
      "loss": 0.7422,
      "step": 7402
    },
    {
      "epoch": 2.3375434164824753,
      "grad_norm": 2.4407620429992676,
      "learning_rate": 2.9583888849905274e-05,
      "loss": 0.7474,
      "step": 7403
    },
    {
      "epoch": 2.337859172718661,
      "grad_norm": 2.1666862964630127,
      "learning_rate": 2.9580380324187777e-05,
      "loss": 0.7064,
      "step": 7404
    },
    {
      "epoch": 2.338174928954847,
      "grad_norm": 2.0385549068450928,
      "learning_rate": 2.9576871798470284e-05,
      "loss": 0.6772,
      "step": 7405
    },
    {
      "epoch": 2.3384906851910325,
      "grad_norm": 2.0688464641571045,
      "learning_rate": 2.957336327275279e-05,
      "loss": 0.7269,
      "step": 7406
    },
    {
      "epoch": 2.3388064414272183,
      "grad_norm": 2.455272912979126,
      "learning_rate": 2.9569854747035297e-05,
      "loss": 0.8321,
      "step": 7407
    },
    {
      "epoch": 2.339122197663404,
      "grad_norm": 2.3672099113464355,
      "learning_rate": 2.9566346221317804e-05,
      "loss": 0.7883,
      "step": 7408
    },
    {
      "epoch": 2.3394379538995898,
      "grad_norm": 2.3462321758270264,
      "learning_rate": 2.956283769560031e-05,
      "loss": 0.7947,
      "step": 7409
    },
    {
      "epoch": 2.339753710135775,
      "grad_norm": 2.6598012447357178,
      "learning_rate": 2.9559329169882814e-05,
      "loss": 0.7301,
      "step": 7410
    },
    {
      "epoch": 2.340069466371961,
      "grad_norm": 2.564700126647949,
      "learning_rate": 2.955582064416532e-05,
      "loss": 0.7953,
      "step": 7411
    },
    {
      "epoch": 2.3403852226081465,
      "grad_norm": 2.294672966003418,
      "learning_rate": 2.9552312118447828e-05,
      "loss": 0.7568,
      "step": 7412
    },
    {
      "epoch": 2.3407009788443323,
      "grad_norm": 2.162868022918701,
      "learning_rate": 2.9548803592730335e-05,
      "loss": 0.7104,
      "step": 7413
    },
    {
      "epoch": 2.341016735080518,
      "grad_norm": 2.26338791847229,
      "learning_rate": 2.9545295067012842e-05,
      "loss": 0.7815,
      "step": 7414
    },
    {
      "epoch": 2.3413324913167033,
      "grad_norm": 2.349160671234131,
      "learning_rate": 2.9541786541295345e-05,
      "loss": 0.6821,
      "step": 7415
    },
    {
      "epoch": 2.341648247552889,
      "grad_norm": 2.385068416595459,
      "learning_rate": 2.953827801557786e-05,
      "loss": 0.7421,
      "step": 7416
    },
    {
      "epoch": 2.3419640037890748,
      "grad_norm": 2.288083553314209,
      "learning_rate": 2.9534769489860366e-05,
      "loss": 0.7226,
      "step": 7417
    },
    {
      "epoch": 2.3422797600252605,
      "grad_norm": 2.219041347503662,
      "learning_rate": 2.953126096414287e-05,
      "loss": 0.7485,
      "step": 7418
    },
    {
      "epoch": 2.3425955162614462,
      "grad_norm": 2.3648176193237305,
      "learning_rate": 2.9527752438425376e-05,
      "loss": 0.8088,
      "step": 7419
    },
    {
      "epoch": 2.342911272497632,
      "grad_norm": 2.5157456398010254,
      "learning_rate": 2.9524243912707883e-05,
      "loss": 0.7915,
      "step": 7420
    },
    {
      "epoch": 2.3432270287338177,
      "grad_norm": 2.265714645385742,
      "learning_rate": 2.952073538699039e-05,
      "loss": 0.7612,
      "step": 7421
    },
    {
      "epoch": 2.343542784970003,
      "grad_norm": 2.120945930480957,
      "learning_rate": 2.9517226861272896e-05,
      "loss": 0.7576,
      "step": 7422
    },
    {
      "epoch": 2.3438585412061888,
      "grad_norm": 2.0959811210632324,
      "learning_rate": 2.9513718335555403e-05,
      "loss": 0.8624,
      "step": 7423
    },
    {
      "epoch": 2.3441742974423745,
      "grad_norm": 2.2487218379974365,
      "learning_rate": 2.9510209809837906e-05,
      "loss": 0.7364,
      "step": 7424
    },
    {
      "epoch": 2.3444900536785602,
      "grad_norm": 2.052297592163086,
      "learning_rate": 2.9506701284120413e-05,
      "loss": 0.7291,
      "step": 7425
    },
    {
      "epoch": 2.344805809914746,
      "grad_norm": 2.551802158355713,
      "learning_rate": 2.950319275840292e-05,
      "loss": 0.6983,
      "step": 7426
    },
    {
      "epoch": 2.3451215661509313,
      "grad_norm": 2.852414131164551,
      "learning_rate": 2.9499684232685427e-05,
      "loss": 0.6973,
      "step": 7427
    },
    {
      "epoch": 2.345437322387117,
      "grad_norm": 2.131692886352539,
      "learning_rate": 2.9496175706967934e-05,
      "loss": 0.6837,
      "step": 7428
    },
    {
      "epoch": 2.3457530786233027,
      "grad_norm": 2.288992166519165,
      "learning_rate": 2.9492667181250437e-05,
      "loss": 0.6816,
      "step": 7429
    },
    {
      "epoch": 2.3460688348594885,
      "grad_norm": 2.398301601409912,
      "learning_rate": 2.9489158655532944e-05,
      "loss": 0.7529,
      "step": 7430
    },
    {
      "epoch": 2.346384591095674,
      "grad_norm": 2.374530076980591,
      "learning_rate": 2.948565012981545e-05,
      "loss": 0.6959,
      "step": 7431
    },
    {
      "epoch": 2.34670034733186,
      "grad_norm": 2.1648707389831543,
      "learning_rate": 2.9482141604097958e-05,
      "loss": 0.7428,
      "step": 7432
    },
    {
      "epoch": 2.3470161035680457,
      "grad_norm": 2.0907163619995117,
      "learning_rate": 2.9478633078380464e-05,
      "loss": 0.7873,
      "step": 7433
    },
    {
      "epoch": 2.347331859804231,
      "grad_norm": 2.2659311294555664,
      "learning_rate": 2.9475124552662975e-05,
      "loss": 0.7489,
      "step": 7434
    },
    {
      "epoch": 2.3476476160404167,
      "grad_norm": 2.2016830444335938,
      "learning_rate": 2.947161602694548e-05,
      "loss": 0.6893,
      "step": 7435
    },
    {
      "epoch": 2.3479633722766025,
      "grad_norm": 2.2501957416534424,
      "learning_rate": 2.9468107501227988e-05,
      "loss": 0.7223,
      "step": 7436
    },
    {
      "epoch": 2.348279128512788,
      "grad_norm": 1.9694714546203613,
      "learning_rate": 2.9464598975510495e-05,
      "loss": 0.7145,
      "step": 7437
    },
    {
      "epoch": 2.348594884748974,
      "grad_norm": 1.9544230699539185,
      "learning_rate": 2.9461090449793e-05,
      "loss": 0.7525,
      "step": 7438
    },
    {
      "epoch": 2.3489106409851592,
      "grad_norm": 2.2328619956970215,
      "learning_rate": 2.9457581924075505e-05,
      "loss": 0.6792,
      "step": 7439
    },
    {
      "epoch": 2.349226397221345,
      "grad_norm": 2.0488662719726562,
      "learning_rate": 2.9454073398358012e-05,
      "loss": 0.6626,
      "step": 7440
    },
    {
      "epoch": 2.3495421534575307,
      "grad_norm": 2.0848724842071533,
      "learning_rate": 2.945056487264052e-05,
      "loss": 0.7371,
      "step": 7441
    },
    {
      "epoch": 2.3498579096937164,
      "grad_norm": 2.4979360103607178,
      "learning_rate": 2.9447056346923026e-05,
      "loss": 0.828,
      "step": 7442
    },
    {
      "epoch": 2.350173665929902,
      "grad_norm": 2.2328217029571533,
      "learning_rate": 2.944354782120553e-05,
      "loss": 0.7928,
      "step": 7443
    },
    {
      "epoch": 2.350489422166088,
      "grad_norm": 2.795334577560425,
      "learning_rate": 2.9440039295488036e-05,
      "loss": 0.6807,
      "step": 7444
    },
    {
      "epoch": 2.3508051784022737,
      "grad_norm": 2.096012830734253,
      "learning_rate": 2.9436530769770543e-05,
      "loss": 0.6654,
      "step": 7445
    },
    {
      "epoch": 2.351120934638459,
      "grad_norm": 2.13388729095459,
      "learning_rate": 2.943302224405305e-05,
      "loss": 0.7453,
      "step": 7446
    },
    {
      "epoch": 2.3514366908746447,
      "grad_norm": 2.284966230392456,
      "learning_rate": 2.9429513718335556e-05,
      "loss": 0.7378,
      "step": 7447
    },
    {
      "epoch": 2.3517524471108304,
      "grad_norm": 2.3433926105499268,
      "learning_rate": 2.942600519261806e-05,
      "loss": 0.7906,
      "step": 7448
    },
    {
      "epoch": 2.352068203347016,
      "grad_norm": 2.2119076251983643,
      "learning_rate": 2.9422496666900567e-05,
      "loss": 0.7817,
      "step": 7449
    },
    {
      "epoch": 2.352383959583202,
      "grad_norm": 2.2864718437194824,
      "learning_rate": 2.9418988141183073e-05,
      "loss": 0.7157,
      "step": 7450
    },
    {
      "epoch": 2.352699715819387,
      "grad_norm": 2.7828001976013184,
      "learning_rate": 2.9415479615465587e-05,
      "loss": 0.7452,
      "step": 7451
    },
    {
      "epoch": 2.353015472055573,
      "grad_norm": 2.7065234184265137,
      "learning_rate": 2.941197108974809e-05,
      "loss": 0.7888,
      "step": 7452
    },
    {
      "epoch": 2.3533312282917587,
      "grad_norm": 1.988437533378601,
      "learning_rate": 2.9408462564030597e-05,
      "loss": 0.7645,
      "step": 7453
    },
    {
      "epoch": 2.3536469845279444,
      "grad_norm": 2.4566433429718018,
      "learning_rate": 2.9404954038313104e-05,
      "loss": 0.7925,
      "step": 7454
    },
    {
      "epoch": 2.35396274076413,
      "grad_norm": 2.5806195735931396,
      "learning_rate": 2.940144551259561e-05,
      "loss": 0.7361,
      "step": 7455
    },
    {
      "epoch": 2.354278497000316,
      "grad_norm": 2.1439549922943115,
      "learning_rate": 2.9397936986878118e-05,
      "loss": 0.6998,
      "step": 7456
    },
    {
      "epoch": 2.3545942532365016,
      "grad_norm": 2.5474462509155273,
      "learning_rate": 2.939442846116062e-05,
      "loss": 0.8584,
      "step": 7457
    },
    {
      "epoch": 2.354910009472687,
      "grad_norm": 2.1750094890594482,
      "learning_rate": 2.9390919935443128e-05,
      "loss": 0.7359,
      "step": 7458
    },
    {
      "epoch": 2.3552257657088727,
      "grad_norm": 2.408705711364746,
      "learning_rate": 2.9387411409725635e-05,
      "loss": 0.728,
      "step": 7459
    },
    {
      "epoch": 2.3555415219450584,
      "grad_norm": 2.0384206771850586,
      "learning_rate": 2.938390288400814e-05,
      "loss": 0.7313,
      "step": 7460
    },
    {
      "epoch": 2.355857278181244,
      "grad_norm": 2.6000070571899414,
      "learning_rate": 2.938039435829065e-05,
      "loss": 0.7641,
      "step": 7461
    },
    {
      "epoch": 2.35617303441743,
      "grad_norm": 2.432255268096924,
      "learning_rate": 2.9376885832573152e-05,
      "loss": 0.6793,
      "step": 7462
    },
    {
      "epoch": 2.356488790653615,
      "grad_norm": 2.3056976795196533,
      "learning_rate": 2.937337730685566e-05,
      "loss": 0.8077,
      "step": 7463
    },
    {
      "epoch": 2.356804546889801,
      "grad_norm": 2.5025837421417236,
      "learning_rate": 2.9369868781138165e-05,
      "loss": 0.7248,
      "step": 7464
    },
    {
      "epoch": 2.3571203031259866,
      "grad_norm": 2.3907036781311035,
      "learning_rate": 2.9366360255420672e-05,
      "loss": 0.7894,
      "step": 7465
    },
    {
      "epoch": 2.3574360593621724,
      "grad_norm": 2.8402698040008545,
      "learning_rate": 2.936285172970318e-05,
      "loss": 0.7465,
      "step": 7466
    },
    {
      "epoch": 2.357751815598358,
      "grad_norm": 2.3583929538726807,
      "learning_rate": 2.9359343203985686e-05,
      "loss": 0.7461,
      "step": 7467
    },
    {
      "epoch": 2.358067571834544,
      "grad_norm": 2.286571741104126,
      "learning_rate": 2.935583467826819e-05,
      "loss": 0.815,
      "step": 7468
    },
    {
      "epoch": 2.3583833280707296,
      "grad_norm": 2.4988346099853516,
      "learning_rate": 2.9352326152550703e-05,
      "loss": 0.7647,
      "step": 7469
    },
    {
      "epoch": 2.358699084306915,
      "grad_norm": 2.139862060546875,
      "learning_rate": 2.934881762683321e-05,
      "loss": 0.6481,
      "step": 7470
    },
    {
      "epoch": 2.3590148405431006,
      "grad_norm": 2.6962850093841553,
      "learning_rate": 2.9345309101115713e-05,
      "loss": 0.7733,
      "step": 7471
    },
    {
      "epoch": 2.3593305967792864,
      "grad_norm": 2.435642957687378,
      "learning_rate": 2.934180057539822e-05,
      "loss": 0.702,
      "step": 7472
    },
    {
      "epoch": 2.359646353015472,
      "grad_norm": 2.5130226612091064,
      "learning_rate": 2.9338292049680727e-05,
      "loss": 0.7331,
      "step": 7473
    },
    {
      "epoch": 2.359962109251658,
      "grad_norm": 2.325961112976074,
      "learning_rate": 2.9334783523963234e-05,
      "loss": 0.7705,
      "step": 7474
    },
    {
      "epoch": 2.3602778654878436,
      "grad_norm": 1.9589663743972778,
      "learning_rate": 2.933127499824574e-05,
      "loss": 0.6277,
      "step": 7475
    },
    {
      "epoch": 2.360593621724029,
      "grad_norm": 2.359126329421997,
      "learning_rate": 2.9327766472528244e-05,
      "loss": 0.7504,
      "step": 7476
    },
    {
      "epoch": 2.3609093779602146,
      "grad_norm": 2.676604986190796,
      "learning_rate": 2.932425794681075e-05,
      "loss": 0.6986,
      "step": 7477
    },
    {
      "epoch": 2.3612251341964003,
      "grad_norm": 2.2030200958251953,
      "learning_rate": 2.9320749421093257e-05,
      "loss": 0.7662,
      "step": 7478
    },
    {
      "epoch": 2.361540890432586,
      "grad_norm": 2.2601187229156494,
      "learning_rate": 2.9317240895375764e-05,
      "loss": 0.7461,
      "step": 7479
    },
    {
      "epoch": 2.361856646668772,
      "grad_norm": 2.6178057193756104,
      "learning_rate": 2.931373236965827e-05,
      "loss": 0.6954,
      "step": 7480
    },
    {
      "epoch": 2.3621724029049576,
      "grad_norm": 2.9918744564056396,
      "learning_rate": 2.9310223843940774e-05,
      "loss": 0.7465,
      "step": 7481
    },
    {
      "epoch": 2.362488159141143,
      "grad_norm": 2.6947402954101562,
      "learning_rate": 2.930671531822328e-05,
      "loss": 0.7445,
      "step": 7482
    },
    {
      "epoch": 2.3628039153773286,
      "grad_norm": 2.348411798477173,
      "learning_rate": 2.9303206792505788e-05,
      "loss": 0.7517,
      "step": 7483
    },
    {
      "epoch": 2.3631196716135143,
      "grad_norm": 2.3570075035095215,
      "learning_rate": 2.9299698266788295e-05,
      "loss": 0.7515,
      "step": 7484
    },
    {
      "epoch": 2.3634354278497,
      "grad_norm": 2.0469462871551514,
      "learning_rate": 2.9296189741070802e-05,
      "loss": 0.7813,
      "step": 7485
    },
    {
      "epoch": 2.363751184085886,
      "grad_norm": 2.372854709625244,
      "learning_rate": 2.9292681215353312e-05,
      "loss": 0.678,
      "step": 7486
    },
    {
      "epoch": 2.3640669403220715,
      "grad_norm": 2.7495646476745605,
      "learning_rate": 2.928917268963582e-05,
      "loss": 0.7526,
      "step": 7487
    },
    {
      "epoch": 2.364382696558257,
      "grad_norm": 2.6166694164276123,
      "learning_rate": 2.9285664163918326e-05,
      "loss": 0.6486,
      "step": 7488
    },
    {
      "epoch": 2.3646984527944426,
      "grad_norm": 2.480560541152954,
      "learning_rate": 2.9282155638200832e-05,
      "loss": 0.7752,
      "step": 7489
    },
    {
      "epoch": 2.3650142090306283,
      "grad_norm": 2.2966151237487793,
      "learning_rate": 2.9278647112483336e-05,
      "loss": 0.6767,
      "step": 7490
    },
    {
      "epoch": 2.365329965266814,
      "grad_norm": 2.223629951477051,
      "learning_rate": 2.9275138586765843e-05,
      "loss": 0.7743,
      "step": 7491
    },
    {
      "epoch": 2.365645721503,
      "grad_norm": 2.326798439025879,
      "learning_rate": 2.927163006104835e-05,
      "loss": 0.6826,
      "step": 7492
    },
    {
      "epoch": 2.3659614777391855,
      "grad_norm": 2.1681854724884033,
      "learning_rate": 2.9268121535330856e-05,
      "loss": 0.6697,
      "step": 7493
    },
    {
      "epoch": 2.366277233975371,
      "grad_norm": 2.963494300842285,
      "learning_rate": 2.9264613009613363e-05,
      "loss": 0.7963,
      "step": 7494
    },
    {
      "epoch": 2.3665929902115566,
      "grad_norm": 2.5618464946746826,
      "learning_rate": 2.9261104483895866e-05,
      "loss": 0.8396,
      "step": 7495
    },
    {
      "epoch": 2.3669087464477423,
      "grad_norm": 2.5814571380615234,
      "learning_rate": 2.9257595958178373e-05,
      "loss": 0.7723,
      "step": 7496
    },
    {
      "epoch": 2.367224502683928,
      "grad_norm": 2.307142972946167,
      "learning_rate": 2.925408743246088e-05,
      "loss": 0.7234,
      "step": 7497
    },
    {
      "epoch": 2.3675402589201138,
      "grad_norm": 2.140676975250244,
      "learning_rate": 2.9250578906743387e-05,
      "loss": 0.705,
      "step": 7498
    },
    {
      "epoch": 2.3678560151562995,
      "grad_norm": 2.3716683387756348,
      "learning_rate": 2.9247070381025894e-05,
      "loss": 0.844,
      "step": 7499
    },
    {
      "epoch": 2.3681717713924852,
      "grad_norm": 2.1812894344329834,
      "learning_rate": 2.92435618553084e-05,
      "loss": 0.7592,
      "step": 7500
    },
    {
      "epoch": 2.3684875276286705,
      "grad_norm": 2.51287579536438,
      "learning_rate": 2.9240053329590904e-05,
      "loss": 0.7932,
      "step": 7501
    },
    {
      "epoch": 2.3688032838648563,
      "grad_norm": 2.2117486000061035,
      "learning_rate": 2.923654480387341e-05,
      "loss": 0.6569,
      "step": 7502
    },
    {
      "epoch": 2.369119040101042,
      "grad_norm": 2.69036865234375,
      "learning_rate": 2.9233036278155924e-05,
      "loss": 0.68,
      "step": 7503
    },
    {
      "epoch": 2.3694347963372278,
      "grad_norm": 2.278089761734009,
      "learning_rate": 2.9229527752438428e-05,
      "loss": 0.7347,
      "step": 7504
    },
    {
      "epoch": 2.3697505525734135,
      "grad_norm": 2.1983838081359863,
      "learning_rate": 2.9226019226720935e-05,
      "loss": 0.6546,
      "step": 7505
    },
    {
      "epoch": 2.370066308809599,
      "grad_norm": 2.5673859119415283,
      "learning_rate": 2.922251070100344e-05,
      "loss": 0.7457,
      "step": 7506
    },
    {
      "epoch": 2.3703820650457845,
      "grad_norm": 2.5021610260009766,
      "learning_rate": 2.9219002175285948e-05,
      "loss": 0.7176,
      "step": 7507
    },
    {
      "epoch": 2.3706978212819703,
      "grad_norm": 2.8065731525421143,
      "learning_rate": 2.9215493649568455e-05,
      "loss": 0.7611,
      "step": 7508
    },
    {
      "epoch": 2.371013577518156,
      "grad_norm": 2.1321070194244385,
      "learning_rate": 2.921198512385096e-05,
      "loss": 0.7298,
      "step": 7509
    },
    {
      "epoch": 2.3713293337543417,
      "grad_norm": 2.7789270877838135,
      "learning_rate": 2.9208476598133465e-05,
      "loss": 0.7186,
      "step": 7510
    },
    {
      "epoch": 2.3716450899905275,
      "grad_norm": 2.430326223373413,
      "learning_rate": 2.9204968072415972e-05,
      "loss": 0.739,
      "step": 7511
    },
    {
      "epoch": 2.371960846226713,
      "grad_norm": 2.0115556716918945,
      "learning_rate": 2.920145954669848e-05,
      "loss": 0.7187,
      "step": 7512
    },
    {
      "epoch": 2.3722766024628985,
      "grad_norm": 2.2092037200927734,
      "learning_rate": 2.9197951020980986e-05,
      "loss": 0.7049,
      "step": 7513
    },
    {
      "epoch": 2.3725923586990842,
      "grad_norm": 2.0526394844055176,
      "learning_rate": 2.9194442495263493e-05,
      "loss": 0.6785,
      "step": 7514
    },
    {
      "epoch": 2.37290811493527,
      "grad_norm": 2.3718132972717285,
      "learning_rate": 2.9190933969545996e-05,
      "loss": 0.7652,
      "step": 7515
    },
    {
      "epoch": 2.3732238711714557,
      "grad_norm": 2.4179224967956543,
      "learning_rate": 2.9187425443828503e-05,
      "loss": 0.8141,
      "step": 7516
    },
    {
      "epoch": 2.3735396274076415,
      "grad_norm": 2.490781307220459,
      "learning_rate": 2.918391691811101e-05,
      "loss": 0.6813,
      "step": 7517
    },
    {
      "epoch": 2.3738553836438268,
      "grad_norm": 2.419264793395996,
      "learning_rate": 2.9180408392393516e-05,
      "loss": 0.737,
      "step": 7518
    },
    {
      "epoch": 2.3741711398800125,
      "grad_norm": 3.0026051998138428,
      "learning_rate": 2.9176899866676023e-05,
      "loss": 0.8376,
      "step": 7519
    },
    {
      "epoch": 2.3744868961161982,
      "grad_norm": 2.1775519847869873,
      "learning_rate": 2.9173391340958527e-05,
      "loss": 0.821,
      "step": 7520
    },
    {
      "epoch": 2.374802652352384,
      "grad_norm": 2.1790108680725098,
      "learning_rate": 2.916988281524104e-05,
      "loss": 0.6639,
      "step": 7521
    },
    {
      "epoch": 2.3751184085885697,
      "grad_norm": 2.614532709121704,
      "learning_rate": 2.9166374289523547e-05,
      "loss": 0.7971,
      "step": 7522
    },
    {
      "epoch": 2.3754341648247554,
      "grad_norm": 2.4363787174224854,
      "learning_rate": 2.916286576380605e-05,
      "loss": 0.7983,
      "step": 7523
    },
    {
      "epoch": 2.375749921060941,
      "grad_norm": 2.5050036907196045,
      "learning_rate": 2.9159357238088557e-05,
      "loss": 0.775,
      "step": 7524
    },
    {
      "epoch": 2.3760656772971265,
      "grad_norm": 2.499716281890869,
      "learning_rate": 2.9155848712371064e-05,
      "loss": 0.8174,
      "step": 7525
    },
    {
      "epoch": 2.376381433533312,
      "grad_norm": 2.59696626663208,
      "learning_rate": 2.915234018665357e-05,
      "loss": 0.8141,
      "step": 7526
    },
    {
      "epoch": 2.376697189769498,
      "grad_norm": 2.124676465988159,
      "learning_rate": 2.9148831660936078e-05,
      "loss": 0.7018,
      "step": 7527
    },
    {
      "epoch": 2.3770129460056837,
      "grad_norm": 2.5926177501678467,
      "learning_rate": 2.9145323135218585e-05,
      "loss": 0.7984,
      "step": 7528
    },
    {
      "epoch": 2.3773287022418694,
      "grad_norm": 2.1478431224823,
      "learning_rate": 2.9141814609501088e-05,
      "loss": 0.7325,
      "step": 7529
    },
    {
      "epoch": 2.3776444584780547,
      "grad_norm": 1.9406179189682007,
      "learning_rate": 2.9138306083783595e-05,
      "loss": 0.6962,
      "step": 7530
    },
    {
      "epoch": 2.3779602147142405,
      "grad_norm": 2.7800700664520264,
      "learning_rate": 2.91347975580661e-05,
      "loss": 0.7764,
      "step": 7531
    },
    {
      "epoch": 2.378275970950426,
      "grad_norm": 2.5579323768615723,
      "learning_rate": 2.913128903234861e-05,
      "loss": 0.7401,
      "step": 7532
    },
    {
      "epoch": 2.378591727186612,
      "grad_norm": 2.0257279872894287,
      "learning_rate": 2.9127780506631115e-05,
      "loss": 0.7821,
      "step": 7533
    },
    {
      "epoch": 2.3789074834227977,
      "grad_norm": 2.431100845336914,
      "learning_rate": 2.912427198091362e-05,
      "loss": 0.7321,
      "step": 7534
    },
    {
      "epoch": 2.3792232396589834,
      "grad_norm": 2.635514497756958,
      "learning_rate": 2.9120763455196125e-05,
      "loss": 0.7789,
      "step": 7535
    },
    {
      "epoch": 2.379538995895169,
      "grad_norm": 2.185744524002075,
      "learning_rate": 2.9117254929478632e-05,
      "loss": 0.7439,
      "step": 7536
    },
    {
      "epoch": 2.3798547521313544,
      "grad_norm": 4.003466606140137,
      "learning_rate": 2.911374640376114e-05,
      "loss": 0.7456,
      "step": 7537
    },
    {
      "epoch": 2.38017050836754,
      "grad_norm": 2.179377794265747,
      "learning_rate": 2.911023787804365e-05,
      "loss": 0.7873,
      "step": 7538
    },
    {
      "epoch": 2.380486264603726,
      "grad_norm": 2.1211819648742676,
      "learning_rate": 2.9106729352326156e-05,
      "loss": 0.8385,
      "step": 7539
    },
    {
      "epoch": 2.3808020208399117,
      "grad_norm": 2.3336894512176514,
      "learning_rate": 2.9103220826608663e-05,
      "loss": 0.7082,
      "step": 7540
    },
    {
      "epoch": 2.3811177770760974,
      "grad_norm": 2.6049082279205322,
      "learning_rate": 2.909971230089117e-05,
      "loss": 0.8439,
      "step": 7541
    },
    {
      "epoch": 2.3814335333122827,
      "grad_norm": 2.2854316234588623,
      "learning_rate": 2.9096203775173677e-05,
      "loss": 0.7314,
      "step": 7542
    },
    {
      "epoch": 2.3817492895484684,
      "grad_norm": 2.792344331741333,
      "learning_rate": 2.909269524945618e-05,
      "loss": 0.7574,
      "step": 7543
    },
    {
      "epoch": 2.382065045784654,
      "grad_norm": 2.131192922592163,
      "learning_rate": 2.9089186723738687e-05,
      "loss": 0.7757,
      "step": 7544
    },
    {
      "epoch": 2.38238080202084,
      "grad_norm": 2.4767322540283203,
      "learning_rate": 2.9085678198021194e-05,
      "loss": 0.7344,
      "step": 7545
    },
    {
      "epoch": 2.3826965582570256,
      "grad_norm": 2.4084768295288086,
      "learning_rate": 2.90821696723037e-05,
      "loss": 0.7654,
      "step": 7546
    },
    {
      "epoch": 2.3830123144932114,
      "grad_norm": 2.2301435470581055,
      "learning_rate": 2.9078661146586207e-05,
      "loss": 0.7176,
      "step": 7547
    },
    {
      "epoch": 2.383328070729397,
      "grad_norm": 2.220057249069214,
      "learning_rate": 2.907515262086871e-05,
      "loss": 0.7264,
      "step": 7548
    },
    {
      "epoch": 2.3836438269655824,
      "grad_norm": 2.209995985031128,
      "learning_rate": 2.9071644095151217e-05,
      "loss": 0.7376,
      "step": 7549
    },
    {
      "epoch": 2.383959583201768,
      "grad_norm": 2.475656509399414,
      "learning_rate": 2.9068135569433724e-05,
      "loss": 0.7562,
      "step": 7550
    },
    {
      "epoch": 2.384275339437954,
      "grad_norm": 2.1142029762268066,
      "learning_rate": 2.906462704371623e-05,
      "loss": 0.8,
      "step": 7551
    },
    {
      "epoch": 2.3845910956741396,
      "grad_norm": 2.162642478942871,
      "learning_rate": 2.9061118517998738e-05,
      "loss": 0.7264,
      "step": 7552
    },
    {
      "epoch": 2.3849068519103254,
      "grad_norm": 1.972994327545166,
      "learning_rate": 2.905760999228124e-05,
      "loss": 0.7496,
      "step": 7553
    },
    {
      "epoch": 2.3852226081465107,
      "grad_norm": 2.393960475921631,
      "learning_rate": 2.9054101466563748e-05,
      "loss": 0.8221,
      "step": 7554
    },
    {
      "epoch": 2.3855383643826964,
      "grad_norm": 2.333997964859009,
      "learning_rate": 2.9050592940846255e-05,
      "loss": 0.7991,
      "step": 7555
    },
    {
      "epoch": 2.385854120618882,
      "grad_norm": 2.1647629737854004,
      "learning_rate": 2.904708441512877e-05,
      "loss": 0.6889,
      "step": 7556
    },
    {
      "epoch": 2.386169876855068,
      "grad_norm": 2.1720919609069824,
      "learning_rate": 2.9043575889411272e-05,
      "loss": 0.7411,
      "step": 7557
    },
    {
      "epoch": 2.3864856330912536,
      "grad_norm": 2.854523181915283,
      "learning_rate": 2.904006736369378e-05,
      "loss": 0.784,
      "step": 7558
    },
    {
      "epoch": 2.3868013893274393,
      "grad_norm": 2.472994089126587,
      "learning_rate": 2.9036558837976286e-05,
      "loss": 0.7975,
      "step": 7559
    },
    {
      "epoch": 2.387117145563625,
      "grad_norm": 2.4374210834503174,
      "learning_rate": 2.9033050312258792e-05,
      "loss": 0.6512,
      "step": 7560
    },
    {
      "epoch": 2.3874329017998104,
      "grad_norm": 2.0756728649139404,
      "learning_rate": 2.90295417865413e-05,
      "loss": 0.7141,
      "step": 7561
    },
    {
      "epoch": 2.387748658035996,
      "grad_norm": 2.3933730125427246,
      "learning_rate": 2.9026033260823803e-05,
      "loss": 0.7151,
      "step": 7562
    },
    {
      "epoch": 2.388064414272182,
      "grad_norm": 20.099300384521484,
      "learning_rate": 2.902252473510631e-05,
      "loss": 0.8046,
      "step": 7563
    },
    {
      "epoch": 2.3883801705083676,
      "grad_norm": 2.5297629833221436,
      "learning_rate": 2.9019016209388816e-05,
      "loss": 0.7871,
      "step": 7564
    },
    {
      "epoch": 2.3886959267445533,
      "grad_norm": 2.325427532196045,
      "learning_rate": 2.9015507683671323e-05,
      "loss": 0.6952,
      "step": 7565
    },
    {
      "epoch": 2.3890116829807386,
      "grad_norm": 2.469228506088257,
      "learning_rate": 2.901199915795383e-05,
      "loss": 0.753,
      "step": 7566
    },
    {
      "epoch": 2.3893274392169244,
      "grad_norm": 2.5691096782684326,
      "learning_rate": 2.9008490632236333e-05,
      "loss": 0.7572,
      "step": 7567
    },
    {
      "epoch": 2.38964319545311,
      "grad_norm": 2.460206985473633,
      "learning_rate": 2.900498210651884e-05,
      "loss": 0.6917,
      "step": 7568
    },
    {
      "epoch": 2.389958951689296,
      "grad_norm": 2.719127655029297,
      "learning_rate": 2.9001473580801347e-05,
      "loss": 0.7953,
      "step": 7569
    },
    {
      "epoch": 2.3902747079254816,
      "grad_norm": 2.1570441722869873,
      "learning_rate": 2.8997965055083854e-05,
      "loss": 0.7487,
      "step": 7570
    },
    {
      "epoch": 2.3905904641616673,
      "grad_norm": 2.550302743911743,
      "learning_rate": 2.899445652936636e-05,
      "loss": 0.789,
      "step": 7571
    },
    {
      "epoch": 2.390906220397853,
      "grad_norm": 2.2833898067474365,
      "learning_rate": 2.8990948003648867e-05,
      "loss": 0.723,
      "step": 7572
    },
    {
      "epoch": 2.3912219766340383,
      "grad_norm": 2.6677935123443604,
      "learning_rate": 2.8987439477931378e-05,
      "loss": 0.7599,
      "step": 7573
    },
    {
      "epoch": 2.391537732870224,
      "grad_norm": 1.9999110698699951,
      "learning_rate": 2.8983930952213884e-05,
      "loss": 0.667,
      "step": 7574
    },
    {
      "epoch": 2.39185348910641,
      "grad_norm": 2.359208345413208,
      "learning_rate": 2.898042242649639e-05,
      "loss": 0.7034,
      "step": 7575
    },
    {
      "epoch": 2.3921692453425956,
      "grad_norm": 2.0094001293182373,
      "learning_rate": 2.8976913900778895e-05,
      "loss": 0.7178,
      "step": 7576
    },
    {
      "epoch": 2.3924850015787813,
      "grad_norm": 2.398533582687378,
      "learning_rate": 2.89734053750614e-05,
      "loss": 0.7207,
      "step": 7577
    },
    {
      "epoch": 2.392800757814967,
      "grad_norm": 2.0624513626098633,
      "learning_rate": 2.8969896849343908e-05,
      "loss": 0.6441,
      "step": 7578
    },
    {
      "epoch": 2.3931165140511523,
      "grad_norm": 2.4364500045776367,
      "learning_rate": 2.8966388323626415e-05,
      "loss": 0.734,
      "step": 7579
    },
    {
      "epoch": 2.393432270287338,
      "grad_norm": 2.1480374336242676,
      "learning_rate": 2.8962879797908922e-05,
      "loss": 0.6913,
      "step": 7580
    },
    {
      "epoch": 2.393748026523524,
      "grad_norm": 2.3899426460266113,
      "learning_rate": 2.8959371272191425e-05,
      "loss": 0.6846,
      "step": 7581
    },
    {
      "epoch": 2.3940637827597095,
      "grad_norm": 2.295400619506836,
      "learning_rate": 2.8955862746473932e-05,
      "loss": 0.782,
      "step": 7582
    },
    {
      "epoch": 2.3943795389958953,
      "grad_norm": 2.3174333572387695,
      "learning_rate": 2.895235422075644e-05,
      "loss": 0.7246,
      "step": 7583
    },
    {
      "epoch": 2.394695295232081,
      "grad_norm": 2.151867151260376,
      "learning_rate": 2.8948845695038946e-05,
      "loss": 0.7496,
      "step": 7584
    },
    {
      "epoch": 2.3950110514682663,
      "grad_norm": 2.4814796447753906,
      "learning_rate": 2.8945337169321453e-05,
      "loss": 0.7356,
      "step": 7585
    },
    {
      "epoch": 2.395326807704452,
      "grad_norm": 2.0832927227020264,
      "learning_rate": 2.8941828643603956e-05,
      "loss": 0.7561,
      "step": 7586
    },
    {
      "epoch": 2.395642563940638,
      "grad_norm": 2.295858144760132,
      "learning_rate": 2.8938320117886463e-05,
      "loss": 0.6491,
      "step": 7587
    },
    {
      "epoch": 2.3959583201768235,
      "grad_norm": 2.5979483127593994,
      "learning_rate": 2.893481159216897e-05,
      "loss": 0.6936,
      "step": 7588
    },
    {
      "epoch": 2.3962740764130093,
      "grad_norm": 2.5999701023101807,
      "learning_rate": 2.8931303066451476e-05,
      "loss": 0.7178,
      "step": 7589
    },
    {
      "epoch": 2.396589832649195,
      "grad_norm": 2.243502378463745,
      "learning_rate": 2.8927794540733983e-05,
      "loss": 0.6786,
      "step": 7590
    },
    {
      "epoch": 2.3969055888853803,
      "grad_norm": 2.5483105182647705,
      "learning_rate": 2.8924286015016493e-05,
      "loss": 0.7304,
      "step": 7591
    },
    {
      "epoch": 2.397221345121566,
      "grad_norm": 2.2263283729553223,
      "learning_rate": 2.8920777489299e-05,
      "loss": 0.713,
      "step": 7592
    },
    {
      "epoch": 2.3975371013577518,
      "grad_norm": 2.881471872329712,
      "learning_rate": 2.8917268963581507e-05,
      "loss": 0.7887,
      "step": 7593
    },
    {
      "epoch": 2.3978528575939375,
      "grad_norm": 2.6122899055480957,
      "learning_rate": 2.8913760437864014e-05,
      "loss": 0.8816,
      "step": 7594
    },
    {
      "epoch": 2.3981686138301233,
      "grad_norm": 2.3309762477874756,
      "learning_rate": 2.8910251912146517e-05,
      "loss": 0.7693,
      "step": 7595
    },
    {
      "epoch": 2.398484370066309,
      "grad_norm": 2.586186408996582,
      "learning_rate": 2.8906743386429024e-05,
      "loss": 0.7424,
      "step": 7596
    },
    {
      "epoch": 2.3988001263024943,
      "grad_norm": 2.7314932346343994,
      "learning_rate": 2.890323486071153e-05,
      "loss": 0.7981,
      "step": 7597
    },
    {
      "epoch": 2.39911588253868,
      "grad_norm": 2.356030225753784,
      "learning_rate": 2.8899726334994038e-05,
      "loss": 0.6929,
      "step": 7598
    },
    {
      "epoch": 2.3994316387748658,
      "grad_norm": 2.4049620628356934,
      "learning_rate": 2.8896217809276545e-05,
      "loss": 0.7219,
      "step": 7599
    },
    {
      "epoch": 2.3997473950110515,
      "grad_norm": 2.4922211170196533,
      "learning_rate": 2.8892709283559048e-05,
      "loss": 0.7983,
      "step": 7600
    },
    {
      "epoch": 2.3997473950110515,
      "eval_loss": 0.7912317514419556,
      "eval_runtime": 4.646,
      "eval_samples_per_second": 535.943,
      "eval_steps_per_second": 4.52,
      "step": 7600
    },
    {
      "epoch": 2.4000631512472372,
      "grad_norm": 2.1001534461975098,
      "learning_rate": 2.8889200757841555e-05,
      "loss": 0.7524,
      "step": 7601
    },
    {
      "epoch": 2.400378907483423,
      "grad_norm": 2.4967002868652344,
      "learning_rate": 2.888569223212406e-05,
      "loss": 0.8288,
      "step": 7602
    },
    {
      "epoch": 2.4006946637196083,
      "grad_norm": 2.227509021759033,
      "learning_rate": 2.888218370640657e-05,
      "loss": 0.7547,
      "step": 7603
    },
    {
      "epoch": 2.401010419955794,
      "grad_norm": 2.633049726486206,
      "learning_rate": 2.8878675180689075e-05,
      "loss": 0.7978,
      "step": 7604
    },
    {
      "epoch": 2.4013261761919797,
      "grad_norm": 2.5912160873413086,
      "learning_rate": 2.8875166654971582e-05,
      "loss": 0.7998,
      "step": 7605
    },
    {
      "epoch": 2.4016419324281655,
      "grad_norm": 2.3937876224517822,
      "learning_rate": 2.8871658129254085e-05,
      "loss": 0.7535,
      "step": 7606
    },
    {
      "epoch": 2.401957688664351,
      "grad_norm": 2.4839625358581543,
      "learning_rate": 2.8868149603536592e-05,
      "loss": 0.6771,
      "step": 7607
    },
    {
      "epoch": 2.402273444900537,
      "grad_norm": 2.337853193283081,
      "learning_rate": 2.8864641077819106e-05,
      "loss": 0.7992,
      "step": 7608
    },
    {
      "epoch": 2.4025892011367223,
      "grad_norm": 2.1940224170684814,
      "learning_rate": 2.886113255210161e-05,
      "loss": 0.7916,
      "step": 7609
    },
    {
      "epoch": 2.402904957372908,
      "grad_norm": 2.390580415725708,
      "learning_rate": 2.8857624026384116e-05,
      "loss": 0.6889,
      "step": 7610
    },
    {
      "epoch": 2.4032207136090937,
      "grad_norm": 1.9013516902923584,
      "learning_rate": 2.8854115500666623e-05,
      "loss": 0.6694,
      "step": 7611
    },
    {
      "epoch": 2.4035364698452795,
      "grad_norm": 2.409512996673584,
      "learning_rate": 2.885060697494913e-05,
      "loss": 0.6872,
      "step": 7612
    },
    {
      "epoch": 2.403852226081465,
      "grad_norm": 2.3823490142822266,
      "learning_rate": 2.8847098449231637e-05,
      "loss": 0.7591,
      "step": 7613
    },
    {
      "epoch": 2.404167982317651,
      "grad_norm": 2.1817009449005127,
      "learning_rate": 2.884358992351414e-05,
      "loss": 0.7125,
      "step": 7614
    },
    {
      "epoch": 2.4044837385538367,
      "grad_norm": 2.282644510269165,
      "learning_rate": 2.8840081397796647e-05,
      "loss": 0.7225,
      "step": 7615
    },
    {
      "epoch": 2.404799494790022,
      "grad_norm": 2.236366033554077,
      "learning_rate": 2.8836572872079154e-05,
      "loss": 0.6814,
      "step": 7616
    },
    {
      "epoch": 2.4051152510262077,
      "grad_norm": 2.43698787689209,
      "learning_rate": 2.883306434636166e-05,
      "loss": 0.8306,
      "step": 7617
    },
    {
      "epoch": 2.4054310072623935,
      "grad_norm": 2.5409533977508545,
      "learning_rate": 2.8829555820644167e-05,
      "loss": 0.8089,
      "step": 7618
    },
    {
      "epoch": 2.405746763498579,
      "grad_norm": 2.2712771892547607,
      "learning_rate": 2.8826047294926674e-05,
      "loss": 0.7397,
      "step": 7619
    },
    {
      "epoch": 2.406062519734765,
      "grad_norm": 2.4282281398773193,
      "learning_rate": 2.8822538769209177e-05,
      "loss": 0.7763,
      "step": 7620
    },
    {
      "epoch": 2.40637827597095,
      "grad_norm": 2.0837953090667725,
      "learning_rate": 2.8819030243491684e-05,
      "loss": 0.7871,
      "step": 7621
    },
    {
      "epoch": 2.406694032207136,
      "grad_norm": 2.827772378921509,
      "learning_rate": 2.881552171777419e-05,
      "loss": 0.8202,
      "step": 7622
    },
    {
      "epoch": 2.4070097884433217,
      "grad_norm": 2.813912868499756,
      "learning_rate": 2.8812013192056698e-05,
      "loss": 0.7353,
      "step": 7623
    },
    {
      "epoch": 2.4073255446795074,
      "grad_norm": 2.4461209774017334,
      "learning_rate": 2.8808504666339205e-05,
      "loss": 0.7895,
      "step": 7624
    },
    {
      "epoch": 2.407641300915693,
      "grad_norm": 2.5072338581085205,
      "learning_rate": 2.8804996140621708e-05,
      "loss": 0.7723,
      "step": 7625
    },
    {
      "epoch": 2.407957057151879,
      "grad_norm": 3.0217785835266113,
      "learning_rate": 2.880148761490422e-05,
      "loss": 0.7526,
      "step": 7626
    },
    {
      "epoch": 2.4082728133880646,
      "grad_norm": 2.741884708404541,
      "learning_rate": 2.879797908918673e-05,
      "loss": 0.723,
      "step": 7627
    },
    {
      "epoch": 2.40858856962425,
      "grad_norm": 2.1257529258728027,
      "learning_rate": 2.8794470563469232e-05,
      "loss": 0.6308,
      "step": 7628
    },
    {
      "epoch": 2.4089043258604357,
      "grad_norm": 2.425036907196045,
      "learning_rate": 2.879096203775174e-05,
      "loss": 0.7481,
      "step": 7629
    },
    {
      "epoch": 2.4092200820966214,
      "grad_norm": 2.3721234798431396,
      "learning_rate": 2.8787453512034246e-05,
      "loss": 0.6585,
      "step": 7630
    },
    {
      "epoch": 2.409535838332807,
      "grad_norm": 2.6493728160858154,
      "learning_rate": 2.8783944986316752e-05,
      "loss": 0.7436,
      "step": 7631
    },
    {
      "epoch": 2.409851594568993,
      "grad_norm": 2.327390670776367,
      "learning_rate": 2.878043646059926e-05,
      "loss": 0.7648,
      "step": 7632
    },
    {
      "epoch": 2.410167350805178,
      "grad_norm": 2.1246867179870605,
      "learning_rate": 2.8776927934881766e-05,
      "loss": 0.6497,
      "step": 7633
    },
    {
      "epoch": 2.410483107041364,
      "grad_norm": 2.6984927654266357,
      "learning_rate": 2.877341940916427e-05,
      "loss": 0.7724,
      "step": 7634
    },
    {
      "epoch": 2.4107988632775497,
      "grad_norm": 2.206223487854004,
      "learning_rate": 2.8769910883446776e-05,
      "loss": 0.797,
      "step": 7635
    },
    {
      "epoch": 2.4111146195137354,
      "grad_norm": 2.3960185050964355,
      "learning_rate": 2.8766402357729283e-05,
      "loss": 0.7855,
      "step": 7636
    },
    {
      "epoch": 2.411430375749921,
      "grad_norm": 2.43208646774292,
      "learning_rate": 2.876289383201179e-05,
      "loss": 0.7373,
      "step": 7637
    },
    {
      "epoch": 2.411746131986107,
      "grad_norm": 2.1386830806732178,
      "learning_rate": 2.8759385306294297e-05,
      "loss": 0.7559,
      "step": 7638
    },
    {
      "epoch": 2.4120618882222926,
      "grad_norm": 2.4009485244750977,
      "learning_rate": 2.87558767805768e-05,
      "loss": 0.7385,
      "step": 7639
    },
    {
      "epoch": 2.412377644458478,
      "grad_norm": 2.5974667072296143,
      "learning_rate": 2.8752368254859307e-05,
      "loss": 0.7774,
      "step": 7640
    },
    {
      "epoch": 2.4126934006946636,
      "grad_norm": 2.6378698348999023,
      "learning_rate": 2.8748859729141814e-05,
      "loss": 0.6742,
      "step": 7641
    },
    {
      "epoch": 2.4130091569308494,
      "grad_norm": 2.4827983379364014,
      "learning_rate": 2.874535120342432e-05,
      "loss": 0.8265,
      "step": 7642
    },
    {
      "epoch": 2.413324913167035,
      "grad_norm": 2.5685155391693115,
      "learning_rate": 2.874184267770683e-05,
      "loss": 0.7155,
      "step": 7643
    },
    {
      "epoch": 2.413640669403221,
      "grad_norm": 2.535443067550659,
      "learning_rate": 2.8738334151989338e-05,
      "loss": 0.8901,
      "step": 7644
    },
    {
      "epoch": 2.413956425639406,
      "grad_norm": 2.555335283279419,
      "learning_rate": 2.8734825626271844e-05,
      "loss": 0.7702,
      "step": 7645
    },
    {
      "epoch": 2.414272181875592,
      "grad_norm": 2.3751912117004395,
      "learning_rate": 2.873131710055435e-05,
      "loss": 0.7649,
      "step": 7646
    },
    {
      "epoch": 2.4145879381117776,
      "grad_norm": 2.1492176055908203,
      "learning_rate": 2.8727808574836858e-05,
      "loss": 0.7072,
      "step": 7647
    },
    {
      "epoch": 2.4149036943479634,
      "grad_norm": 2.210137128829956,
      "learning_rate": 2.872430004911936e-05,
      "loss": 0.6559,
      "step": 7648
    },
    {
      "epoch": 2.415219450584149,
      "grad_norm": 2.0029709339141846,
      "learning_rate": 2.8720791523401868e-05,
      "loss": 0.753,
      "step": 7649
    },
    {
      "epoch": 2.415535206820335,
      "grad_norm": 2.377749443054199,
      "learning_rate": 2.8717282997684375e-05,
      "loss": 0.676,
      "step": 7650
    },
    {
      "epoch": 2.4158509630565206,
      "grad_norm": 2.464383363723755,
      "learning_rate": 2.8713774471966882e-05,
      "loss": 0.7386,
      "step": 7651
    },
    {
      "epoch": 2.416166719292706,
      "grad_norm": 2.2709133625030518,
      "learning_rate": 2.871026594624939e-05,
      "loss": 0.7615,
      "step": 7652
    },
    {
      "epoch": 2.4164824755288916,
      "grad_norm": 2.3370187282562256,
      "learning_rate": 2.8706757420531892e-05,
      "loss": 0.7624,
      "step": 7653
    },
    {
      "epoch": 2.4167982317650774,
      "grad_norm": 2.3838319778442383,
      "learning_rate": 2.87032488948144e-05,
      "loss": 0.8077,
      "step": 7654
    },
    {
      "epoch": 2.417113988001263,
      "grad_norm": 2.518190860748291,
      "learning_rate": 2.8699740369096906e-05,
      "loss": 0.7685,
      "step": 7655
    },
    {
      "epoch": 2.417429744237449,
      "grad_norm": 2.2337231636047363,
      "learning_rate": 2.8696231843379413e-05,
      "loss": 0.7024,
      "step": 7656
    },
    {
      "epoch": 2.417745500473634,
      "grad_norm": 2.340545654296875,
      "learning_rate": 2.869272331766192e-05,
      "loss": 0.6519,
      "step": 7657
    },
    {
      "epoch": 2.41806125670982,
      "grad_norm": 2.1071722507476807,
      "learning_rate": 2.8689214791944423e-05,
      "loss": 0.7024,
      "step": 7658
    },
    {
      "epoch": 2.4183770129460056,
      "grad_norm": 2.4532523155212402,
      "learning_rate": 2.868570626622693e-05,
      "loss": 0.8547,
      "step": 7659
    },
    {
      "epoch": 2.4186927691821913,
      "grad_norm": 2.2174508571624756,
      "learning_rate": 2.8682197740509436e-05,
      "loss": 0.7109,
      "step": 7660
    },
    {
      "epoch": 2.419008525418377,
      "grad_norm": 2.086230516433716,
      "learning_rate": 2.867868921479195e-05,
      "loss": 0.7439,
      "step": 7661
    },
    {
      "epoch": 2.419324281654563,
      "grad_norm": 2.261385917663574,
      "learning_rate": 2.8675180689074453e-05,
      "loss": 0.7191,
      "step": 7662
    },
    {
      "epoch": 2.4196400378907486,
      "grad_norm": 2.4019076824188232,
      "learning_rate": 2.867167216335696e-05,
      "loss": 0.7586,
      "step": 7663
    },
    {
      "epoch": 2.419955794126934,
      "grad_norm": 2.328317642211914,
      "learning_rate": 2.8668163637639467e-05,
      "loss": 0.7727,
      "step": 7664
    },
    {
      "epoch": 2.4202715503631196,
      "grad_norm": 2.281101703643799,
      "learning_rate": 2.8664655111921974e-05,
      "loss": 0.7277,
      "step": 7665
    },
    {
      "epoch": 2.4205873065993053,
      "grad_norm": 2.229642629623413,
      "learning_rate": 2.866114658620448e-05,
      "loss": 0.7668,
      "step": 7666
    },
    {
      "epoch": 2.420903062835491,
      "grad_norm": 3.0005760192871094,
      "learning_rate": 2.8657638060486984e-05,
      "loss": 0.8534,
      "step": 7667
    },
    {
      "epoch": 2.421218819071677,
      "grad_norm": 2.379782199859619,
      "learning_rate": 2.865412953476949e-05,
      "loss": 0.8101,
      "step": 7668
    },
    {
      "epoch": 2.421534575307862,
      "grad_norm": 2.1543023586273193,
      "learning_rate": 2.8650621009051998e-05,
      "loss": 0.7603,
      "step": 7669
    },
    {
      "epoch": 2.421850331544048,
      "grad_norm": 2.881096124649048,
      "learning_rate": 2.8647112483334505e-05,
      "loss": 0.7606,
      "step": 7670
    },
    {
      "epoch": 2.4221660877802336,
      "grad_norm": 2.03995943069458,
      "learning_rate": 2.864360395761701e-05,
      "loss": 0.6719,
      "step": 7671
    },
    {
      "epoch": 2.4224818440164193,
      "grad_norm": 2.380509614944458,
      "learning_rate": 2.8640095431899515e-05,
      "loss": 0.736,
      "step": 7672
    },
    {
      "epoch": 2.422797600252605,
      "grad_norm": 2.2206733226776123,
      "learning_rate": 2.863658690618202e-05,
      "loss": 0.7034,
      "step": 7673
    },
    {
      "epoch": 2.423113356488791,
      "grad_norm": 1.9876405000686646,
      "learning_rate": 2.863307838046453e-05,
      "loss": 0.7119,
      "step": 7674
    },
    {
      "epoch": 2.4234291127249765,
      "grad_norm": 2.1760857105255127,
      "learning_rate": 2.8629569854747035e-05,
      "loss": 0.6858,
      "step": 7675
    },
    {
      "epoch": 2.423744868961162,
      "grad_norm": 2.5893588066101074,
      "learning_rate": 2.8626061329029542e-05,
      "loss": 0.7333,
      "step": 7676
    },
    {
      "epoch": 2.4240606251973476,
      "grad_norm": 2.578601598739624,
      "learning_rate": 2.862255280331205e-05,
      "loss": 0.7984,
      "step": 7677
    },
    {
      "epoch": 2.4243763814335333,
      "grad_norm": 2.279707431793213,
      "learning_rate": 2.861904427759456e-05,
      "loss": 0.7327,
      "step": 7678
    },
    {
      "epoch": 2.424692137669719,
      "grad_norm": 2.5000035762786865,
      "learning_rate": 2.8615535751877066e-05,
      "loss": 0.8039,
      "step": 7679
    },
    {
      "epoch": 2.4250078939059048,
      "grad_norm": 2.192425012588501,
      "learning_rate": 2.8612027226159573e-05,
      "loss": 0.7376,
      "step": 7680
    },
    {
      "epoch": 2.42532365014209,
      "grad_norm": 2.2738234996795654,
      "learning_rate": 2.8608518700442076e-05,
      "loss": 0.7908,
      "step": 7681
    },
    {
      "epoch": 2.425639406378276,
      "grad_norm": 2.3572921752929688,
      "learning_rate": 2.8605010174724583e-05,
      "loss": 0.7589,
      "step": 7682
    },
    {
      "epoch": 2.4259551626144615,
      "grad_norm": 2.270810127258301,
      "learning_rate": 2.860150164900709e-05,
      "loss": 0.6401,
      "step": 7683
    },
    {
      "epoch": 2.4262709188506473,
      "grad_norm": 2.2342569828033447,
      "learning_rate": 2.8597993123289597e-05,
      "loss": 0.7999,
      "step": 7684
    },
    {
      "epoch": 2.426586675086833,
      "grad_norm": 2.6972498893737793,
      "learning_rate": 2.8594484597572103e-05,
      "loss": 0.7724,
      "step": 7685
    },
    {
      "epoch": 2.4269024313230187,
      "grad_norm": 2.1097421646118164,
      "learning_rate": 2.8590976071854607e-05,
      "loss": 0.743,
      "step": 7686
    },
    {
      "epoch": 2.4272181875592045,
      "grad_norm": 2.2763214111328125,
      "learning_rate": 2.8587467546137114e-05,
      "loss": 0.701,
      "step": 7687
    },
    {
      "epoch": 2.42753394379539,
      "grad_norm": 2.272642135620117,
      "learning_rate": 2.858395902041962e-05,
      "loss": 0.7209,
      "step": 7688
    },
    {
      "epoch": 2.4278497000315755,
      "grad_norm": 2.625168561935425,
      "learning_rate": 2.8580450494702127e-05,
      "loss": 0.7508,
      "step": 7689
    },
    {
      "epoch": 2.4281654562677613,
      "grad_norm": 2.139834403991699,
      "learning_rate": 2.8576941968984634e-05,
      "loss": 0.6809,
      "step": 7690
    },
    {
      "epoch": 2.428481212503947,
      "grad_norm": 2.4825310707092285,
      "learning_rate": 2.8573433443267137e-05,
      "loss": 0.7051,
      "step": 7691
    },
    {
      "epoch": 2.4287969687401327,
      "grad_norm": 2.4181926250457764,
      "learning_rate": 2.8569924917549644e-05,
      "loss": 0.6352,
      "step": 7692
    },
    {
      "epoch": 2.4291127249763185,
      "grad_norm": 2.1215713024139404,
      "learning_rate": 2.856641639183215e-05,
      "loss": 0.7784,
      "step": 7693
    },
    {
      "epoch": 2.4294284812125038,
      "grad_norm": 2.5258188247680664,
      "learning_rate": 2.8562907866114658e-05,
      "loss": 0.7724,
      "step": 7694
    },
    {
      "epoch": 2.4297442374486895,
      "grad_norm": 2.696810245513916,
      "learning_rate": 2.8559399340397165e-05,
      "loss": 0.8032,
      "step": 7695
    },
    {
      "epoch": 2.4300599936848752,
      "grad_norm": 2.411675214767456,
      "learning_rate": 2.8555890814679675e-05,
      "loss": 0.8035,
      "step": 7696
    },
    {
      "epoch": 2.430375749921061,
      "grad_norm": 1.9283158779144287,
      "learning_rate": 2.8552382288962182e-05,
      "loss": 0.6631,
      "step": 7697
    },
    {
      "epoch": 2.4306915061572467,
      "grad_norm": 2.5783939361572266,
      "learning_rate": 2.854887376324469e-05,
      "loss": 0.7479,
      "step": 7698
    },
    {
      "epoch": 2.4310072623934325,
      "grad_norm": 2.440401554107666,
      "learning_rate": 2.8545365237527195e-05,
      "loss": 0.7737,
      "step": 7699
    },
    {
      "epoch": 2.4313230186296177,
      "grad_norm": 2.6944591999053955,
      "learning_rate": 2.85418567118097e-05,
      "loss": 0.8004,
      "step": 7700
    },
    {
      "epoch": 2.4316387748658035,
      "grad_norm": 2.167044162750244,
      "learning_rate": 2.8538348186092206e-05,
      "loss": 0.7003,
      "step": 7701
    },
    {
      "epoch": 2.4319545311019892,
      "grad_norm": 2.17741060256958,
      "learning_rate": 2.8534839660374712e-05,
      "loss": 0.779,
      "step": 7702
    },
    {
      "epoch": 2.432270287338175,
      "grad_norm": 2.2927565574645996,
      "learning_rate": 2.853133113465722e-05,
      "loss": 0.7912,
      "step": 7703
    },
    {
      "epoch": 2.4325860435743607,
      "grad_norm": 2.265805959701538,
      "learning_rate": 2.8527822608939726e-05,
      "loss": 0.7282,
      "step": 7704
    },
    {
      "epoch": 2.4329017998105464,
      "grad_norm": 2.60117244720459,
      "learning_rate": 2.852431408322223e-05,
      "loss": 0.7411,
      "step": 7705
    },
    {
      "epoch": 2.4332175560467317,
      "grad_norm": 2.0044736862182617,
      "learning_rate": 2.8520805557504736e-05,
      "loss": 0.6959,
      "step": 7706
    },
    {
      "epoch": 2.4335333122829175,
      "grad_norm": 2.282301902770996,
      "learning_rate": 2.8517297031787243e-05,
      "loss": 0.8102,
      "step": 7707
    },
    {
      "epoch": 2.433849068519103,
      "grad_norm": 2.1658644676208496,
      "learning_rate": 2.851378850606975e-05,
      "loss": 0.7005,
      "step": 7708
    },
    {
      "epoch": 2.434164824755289,
      "grad_norm": 2.281977891921997,
      "learning_rate": 2.8510279980352257e-05,
      "loss": 0.7277,
      "step": 7709
    },
    {
      "epoch": 2.4344805809914747,
      "grad_norm": 2.2611544132232666,
      "learning_rate": 2.8506771454634763e-05,
      "loss": 0.7332,
      "step": 7710
    },
    {
      "epoch": 2.4347963372276604,
      "grad_norm": 2.1976494789123535,
      "learning_rate": 2.8503262928917267e-05,
      "loss": 0.7279,
      "step": 7711
    },
    {
      "epoch": 2.4351120934638457,
      "grad_norm": 2.644869804382324,
      "learning_rate": 2.8499754403199774e-05,
      "loss": 0.7926,
      "step": 7712
    },
    {
      "epoch": 2.4354278497000315,
      "grad_norm": 2.724968194961548,
      "learning_rate": 2.8496245877482287e-05,
      "loss": 0.7081,
      "step": 7713
    },
    {
      "epoch": 2.435743605936217,
      "grad_norm": 2.391479969024658,
      "learning_rate": 2.849273735176479e-05,
      "loss": 0.8209,
      "step": 7714
    },
    {
      "epoch": 2.436059362172403,
      "grad_norm": 2.5117380619049072,
      "learning_rate": 2.8489228826047298e-05,
      "loss": 0.7302,
      "step": 7715
    },
    {
      "epoch": 2.4363751184085887,
      "grad_norm": 2.252516269683838,
      "learning_rate": 2.8485720300329804e-05,
      "loss": 0.7585,
      "step": 7716
    },
    {
      "epoch": 2.4366908746447744,
      "grad_norm": 2.6872568130493164,
      "learning_rate": 2.848221177461231e-05,
      "loss": 0.7802,
      "step": 7717
    },
    {
      "epoch": 2.43700663088096,
      "grad_norm": 2.281118631362915,
      "learning_rate": 2.8478703248894818e-05,
      "loss": 0.7217,
      "step": 7718
    },
    {
      "epoch": 2.4373223871171454,
      "grad_norm": 2.6029117107391357,
      "learning_rate": 2.847519472317732e-05,
      "loss": 0.8431,
      "step": 7719
    },
    {
      "epoch": 2.437638143353331,
      "grad_norm": 2.493500232696533,
      "learning_rate": 2.8471686197459828e-05,
      "loss": 0.7773,
      "step": 7720
    },
    {
      "epoch": 2.437953899589517,
      "grad_norm": 2.405489683151245,
      "learning_rate": 2.8468177671742335e-05,
      "loss": 0.7538,
      "step": 7721
    },
    {
      "epoch": 2.4382696558257027,
      "grad_norm": 2.4852843284606934,
      "learning_rate": 2.8464669146024842e-05,
      "loss": 0.7382,
      "step": 7722
    },
    {
      "epoch": 2.4385854120618884,
      "grad_norm": 2.0879263877868652,
      "learning_rate": 2.846116062030735e-05,
      "loss": 0.7666,
      "step": 7723
    },
    {
      "epoch": 2.4389011682980737,
      "grad_norm": 2.477461576461792,
      "learning_rate": 2.8457652094589855e-05,
      "loss": 0.7452,
      "step": 7724
    },
    {
      "epoch": 2.4392169245342594,
      "grad_norm": 2.768342971801758,
      "learning_rate": 2.845414356887236e-05,
      "loss": 0.765,
      "step": 7725
    },
    {
      "epoch": 2.439532680770445,
      "grad_norm": 2.504223108291626,
      "learning_rate": 2.8450635043154866e-05,
      "loss": 0.7484,
      "step": 7726
    },
    {
      "epoch": 2.439848437006631,
      "grad_norm": 2.573714256286621,
      "learning_rate": 2.8447126517437373e-05,
      "loss": 0.7275,
      "step": 7727
    },
    {
      "epoch": 2.4401641932428166,
      "grad_norm": 2.770782470703125,
      "learning_rate": 2.844361799171988e-05,
      "loss": 0.7071,
      "step": 7728
    },
    {
      "epoch": 2.4404799494790024,
      "grad_norm": 2.4680685997009277,
      "learning_rate": 2.8440109466002386e-05,
      "loss": 0.7515,
      "step": 7729
    },
    {
      "epoch": 2.440795705715188,
      "grad_norm": 2.449904680252075,
      "learning_rate": 2.843660094028489e-05,
      "loss": 0.7552,
      "step": 7730
    },
    {
      "epoch": 2.4411114619513734,
      "grad_norm": 2.459946870803833,
      "learning_rate": 2.8433092414567403e-05,
      "loss": 0.7863,
      "step": 7731
    },
    {
      "epoch": 2.441427218187559,
      "grad_norm": 2.3294801712036133,
      "learning_rate": 2.842958388884991e-05,
      "loss": 0.6762,
      "step": 7732
    },
    {
      "epoch": 2.441742974423745,
      "grad_norm": 2.1153998374938965,
      "learning_rate": 2.8426075363132413e-05,
      "loss": 0.6692,
      "step": 7733
    },
    {
      "epoch": 2.4420587306599306,
      "grad_norm": 1.9492788314819336,
      "learning_rate": 2.842256683741492e-05,
      "loss": 0.762,
      "step": 7734
    },
    {
      "epoch": 2.4423744868961164,
      "grad_norm": 2.1991779804229736,
      "learning_rate": 2.8419058311697427e-05,
      "loss": 0.7217,
      "step": 7735
    },
    {
      "epoch": 2.4426902431323017,
      "grad_norm": 2.195894956588745,
      "learning_rate": 2.8415549785979934e-05,
      "loss": 0.7017,
      "step": 7736
    },
    {
      "epoch": 2.4430059993684874,
      "grad_norm": 2.218366861343384,
      "learning_rate": 2.841204126026244e-05,
      "loss": 0.7606,
      "step": 7737
    },
    {
      "epoch": 2.443321755604673,
      "grad_norm": 2.060709238052368,
      "learning_rate": 2.8408532734544947e-05,
      "loss": 0.7379,
      "step": 7738
    },
    {
      "epoch": 2.443637511840859,
      "grad_norm": 2.4329493045806885,
      "learning_rate": 2.840502420882745e-05,
      "loss": 0.7094,
      "step": 7739
    },
    {
      "epoch": 2.4439532680770446,
      "grad_norm": 2.7848832607269287,
      "learning_rate": 2.8401515683109958e-05,
      "loss": 0.7574,
      "step": 7740
    },
    {
      "epoch": 2.4442690243132303,
      "grad_norm": 2.322451114654541,
      "learning_rate": 2.8398007157392465e-05,
      "loss": 0.7309,
      "step": 7741
    },
    {
      "epoch": 2.444584780549416,
      "grad_norm": 2.3492846488952637,
      "learning_rate": 2.839449863167497e-05,
      "loss": 0.735,
      "step": 7742
    },
    {
      "epoch": 2.4449005367856014,
      "grad_norm": 2.287339448928833,
      "learning_rate": 2.8390990105957478e-05,
      "loss": 0.703,
      "step": 7743
    },
    {
      "epoch": 2.445216293021787,
      "grad_norm": 2.2510428428649902,
      "learning_rate": 2.838748158023998e-05,
      "loss": 0.6897,
      "step": 7744
    },
    {
      "epoch": 2.445532049257973,
      "grad_norm": 2.506152391433716,
      "learning_rate": 2.838397305452249e-05,
      "loss": 0.8254,
      "step": 7745
    },
    {
      "epoch": 2.4458478054941586,
      "grad_norm": 2.4801883697509766,
      "learning_rate": 2.8380464528804995e-05,
      "loss": 0.7223,
      "step": 7746
    },
    {
      "epoch": 2.4461635617303443,
      "grad_norm": 2.622631072998047,
      "learning_rate": 2.8376956003087502e-05,
      "loss": 0.717,
      "step": 7747
    },
    {
      "epoch": 2.4464793179665296,
      "grad_norm": 2.02839732170105,
      "learning_rate": 2.8373447477370012e-05,
      "loss": 0.6837,
      "step": 7748
    },
    {
      "epoch": 2.4467950742027154,
      "grad_norm": 2.424133539199829,
      "learning_rate": 2.836993895165252e-05,
      "loss": 0.7191,
      "step": 7749
    },
    {
      "epoch": 2.447110830438901,
      "grad_norm": 2.6341497898101807,
      "learning_rate": 2.8366430425935026e-05,
      "loss": 0.7389,
      "step": 7750
    },
    {
      "epoch": 2.447426586675087,
      "grad_norm": 2.6019773483276367,
      "learning_rate": 2.8362921900217533e-05,
      "loss": 0.8041,
      "step": 7751
    },
    {
      "epoch": 2.4477423429112726,
      "grad_norm": 2.512977361679077,
      "learning_rate": 2.835941337450004e-05,
      "loss": 0.7424,
      "step": 7752
    },
    {
      "epoch": 2.4480580991474583,
      "grad_norm": 2.118992567062378,
      "learning_rate": 2.8355904848782543e-05,
      "loss": 0.7023,
      "step": 7753
    },
    {
      "epoch": 2.448373855383644,
      "grad_norm": 1.9699645042419434,
      "learning_rate": 2.835239632306505e-05,
      "loss": 0.6905,
      "step": 7754
    },
    {
      "epoch": 2.4486896116198293,
      "grad_norm": 2.891951560974121,
      "learning_rate": 2.8348887797347557e-05,
      "loss": 0.7116,
      "step": 7755
    },
    {
      "epoch": 2.449005367856015,
      "grad_norm": 2.133965253829956,
      "learning_rate": 2.8345379271630063e-05,
      "loss": 0.7275,
      "step": 7756
    },
    {
      "epoch": 2.449321124092201,
      "grad_norm": 2.3971898555755615,
      "learning_rate": 2.834187074591257e-05,
      "loss": 0.7506,
      "step": 7757
    },
    {
      "epoch": 2.4496368803283866,
      "grad_norm": 2.6870806217193604,
      "learning_rate": 2.8338362220195074e-05,
      "loss": 0.7064,
      "step": 7758
    },
    {
      "epoch": 2.4499526365645723,
      "grad_norm": 2.7624027729034424,
      "learning_rate": 2.833485369447758e-05,
      "loss": 0.6863,
      "step": 7759
    },
    {
      "epoch": 2.4502683928007576,
      "grad_norm": 2.484246015548706,
      "learning_rate": 2.8331345168760087e-05,
      "loss": 0.7085,
      "step": 7760
    },
    {
      "epoch": 2.4505841490369433,
      "grad_norm": 16.37071418762207,
      "learning_rate": 2.8327836643042594e-05,
      "loss": 0.7955,
      "step": 7761
    },
    {
      "epoch": 2.450899905273129,
      "grad_norm": 2.3201937675476074,
      "learning_rate": 2.83243281173251e-05,
      "loss": 0.6986,
      "step": 7762
    },
    {
      "epoch": 2.451215661509315,
      "grad_norm": 2.4304678440093994,
      "learning_rate": 2.8320819591607604e-05,
      "loss": 0.7907,
      "step": 7763
    },
    {
      "epoch": 2.4515314177455005,
      "grad_norm": 2.3980095386505127,
      "learning_rate": 2.831731106589011e-05,
      "loss": 0.8093,
      "step": 7764
    },
    {
      "epoch": 2.4518471739816863,
      "grad_norm": 2.7226264476776123,
      "learning_rate": 2.8313802540172618e-05,
      "loss": 0.6937,
      "step": 7765
    },
    {
      "epoch": 2.452162930217872,
      "grad_norm": 2.5921566486358643,
      "learning_rate": 2.831029401445513e-05,
      "loss": 0.8062,
      "step": 7766
    },
    {
      "epoch": 2.4524786864540573,
      "grad_norm": 2.329010248184204,
      "learning_rate": 2.8306785488737635e-05,
      "loss": 0.7596,
      "step": 7767
    },
    {
      "epoch": 2.452794442690243,
      "grad_norm": 2.38631534576416,
      "learning_rate": 2.8303276963020142e-05,
      "loss": 0.8311,
      "step": 7768
    },
    {
      "epoch": 2.453110198926429,
      "grad_norm": 2.2914071083068848,
      "learning_rate": 2.829976843730265e-05,
      "loss": 0.8805,
      "step": 7769
    },
    {
      "epoch": 2.4534259551626145,
      "grad_norm": 2.307739734649658,
      "learning_rate": 2.8296259911585155e-05,
      "loss": 0.7861,
      "step": 7770
    },
    {
      "epoch": 2.4537417113988003,
      "grad_norm": 2.351430654525757,
      "learning_rate": 2.8292751385867662e-05,
      "loss": 0.7481,
      "step": 7771
    },
    {
      "epoch": 2.4540574676349856,
      "grad_norm": 2.2970633506774902,
      "learning_rate": 2.8289242860150166e-05,
      "loss": 0.6775,
      "step": 7772
    },
    {
      "epoch": 2.4543732238711713,
      "grad_norm": 2.3701975345611572,
      "learning_rate": 2.8285734334432672e-05,
      "loss": 0.7621,
      "step": 7773
    },
    {
      "epoch": 2.454688980107357,
      "grad_norm": 1.9376730918884277,
      "learning_rate": 2.828222580871518e-05,
      "loss": 0.7205,
      "step": 7774
    },
    {
      "epoch": 2.4550047363435428,
      "grad_norm": 2.019841432571411,
      "learning_rate": 2.8278717282997686e-05,
      "loss": 0.7269,
      "step": 7775
    },
    {
      "epoch": 2.4553204925797285,
      "grad_norm": 2.4113385677337646,
      "learning_rate": 2.8275208757280193e-05,
      "loss": 0.8193,
      "step": 7776
    },
    {
      "epoch": 2.4556362488159142,
      "grad_norm": 2.0942764282226562,
      "learning_rate": 2.8271700231562696e-05,
      "loss": 0.6823,
      "step": 7777
    },
    {
      "epoch": 2.4559520050521,
      "grad_norm": 2.127964973449707,
      "learning_rate": 2.8268191705845203e-05,
      "loss": 0.7047,
      "step": 7778
    },
    {
      "epoch": 2.4562677612882853,
      "grad_norm": 2.3988142013549805,
      "learning_rate": 2.826468318012771e-05,
      "loss": 0.7083,
      "step": 7779
    },
    {
      "epoch": 2.456583517524471,
      "grad_norm": 2.3403193950653076,
      "learning_rate": 2.8261174654410217e-05,
      "loss": 0.8466,
      "step": 7780
    },
    {
      "epoch": 2.4568992737606568,
      "grad_norm": 2.0409491062164307,
      "learning_rate": 2.8257666128692723e-05,
      "loss": 0.7159,
      "step": 7781
    },
    {
      "epoch": 2.4572150299968425,
      "grad_norm": 2.2697956562042236,
      "learning_rate": 2.825415760297523e-05,
      "loss": 0.6988,
      "step": 7782
    },
    {
      "epoch": 2.4575307862330282,
      "grad_norm": 2.364821672439575,
      "learning_rate": 2.825064907725774e-05,
      "loss": 0.754,
      "step": 7783
    },
    {
      "epoch": 2.4578465424692135,
      "grad_norm": 2.4860057830810547,
      "learning_rate": 2.8247140551540247e-05,
      "loss": 0.7787,
      "step": 7784
    },
    {
      "epoch": 2.4581622987053993,
      "grad_norm": 2.640302896499634,
      "learning_rate": 2.8243632025822754e-05,
      "loss": 0.7305,
      "step": 7785
    },
    {
      "epoch": 2.458478054941585,
      "grad_norm": 2.5148797035217285,
      "learning_rate": 2.8240123500105258e-05,
      "loss": 0.7608,
      "step": 7786
    },
    {
      "epoch": 2.4587938111777707,
      "grad_norm": 2.61014461517334,
      "learning_rate": 2.8236614974387764e-05,
      "loss": 0.7155,
      "step": 7787
    },
    {
      "epoch": 2.4591095674139565,
      "grad_norm": 2.0519485473632812,
      "learning_rate": 2.823310644867027e-05,
      "loss": 0.6905,
      "step": 7788
    },
    {
      "epoch": 2.459425323650142,
      "grad_norm": 2.2300052642822266,
      "learning_rate": 2.8229597922952778e-05,
      "loss": 0.8122,
      "step": 7789
    },
    {
      "epoch": 2.459741079886328,
      "grad_norm": 2.6219449043273926,
      "learning_rate": 2.8226089397235285e-05,
      "loss": 0.7365,
      "step": 7790
    },
    {
      "epoch": 2.4600568361225132,
      "grad_norm": 2.2139534950256348,
      "learning_rate": 2.8222580871517788e-05,
      "loss": 0.6989,
      "step": 7791
    },
    {
      "epoch": 2.460372592358699,
      "grad_norm": 2.7839269638061523,
      "learning_rate": 2.8219072345800295e-05,
      "loss": 0.8257,
      "step": 7792
    },
    {
      "epoch": 2.4606883485948847,
      "grad_norm": 2.33667254447937,
      "learning_rate": 2.8215563820082802e-05,
      "loss": 0.7503,
      "step": 7793
    },
    {
      "epoch": 2.4610041048310705,
      "grad_norm": 2.23229718208313,
      "learning_rate": 2.821205529436531e-05,
      "loss": 0.7881,
      "step": 7794
    },
    {
      "epoch": 2.461319861067256,
      "grad_norm": 9.177754402160645,
      "learning_rate": 2.8208546768647815e-05,
      "loss": 0.6363,
      "step": 7795
    },
    {
      "epoch": 2.461635617303442,
      "grad_norm": 2.544511318206787,
      "learning_rate": 2.820503824293032e-05,
      "loss": 0.6878,
      "step": 7796
    },
    {
      "epoch": 2.4619513735396272,
      "grad_norm": 2.6057910919189453,
      "learning_rate": 2.8201529717212826e-05,
      "loss": 0.8149,
      "step": 7797
    },
    {
      "epoch": 2.462267129775813,
      "grad_norm": 2.2863526344299316,
      "learning_rate": 2.8198021191495333e-05,
      "loss": 0.7807,
      "step": 7798
    },
    {
      "epoch": 2.4625828860119987,
      "grad_norm": 6.27046537399292,
      "learning_rate": 2.819451266577784e-05,
      "loss": 0.7465,
      "step": 7799
    },
    {
      "epoch": 2.4628986422481844,
      "grad_norm": 2.3829357624053955,
      "learning_rate": 2.8191004140060346e-05,
      "loss": 0.8024,
      "step": 7800
    },
    {
      "epoch": 2.4628986422481844,
      "eval_loss": 0.7955511212348938,
      "eval_runtime": 4.6267,
      "eval_samples_per_second": 538.182,
      "eval_steps_per_second": 4.539,
      "step": 7800
    },
    {
      "epoch": 2.46321439848437,
      "grad_norm": 2.1911189556121826,
      "learning_rate": 2.8187495614342856e-05,
      "loss": 0.6773,
      "step": 7801
    },
    {
      "epoch": 2.463530154720556,
      "grad_norm": 2.5137064456939697,
      "learning_rate": 2.8183987088625363e-05,
      "loss": 0.7841,
      "step": 7802
    },
    {
      "epoch": 2.463845910956741,
      "grad_norm": 2.4502947330474854,
      "learning_rate": 2.818047856290787e-05,
      "loss": 0.7359,
      "step": 7803
    },
    {
      "epoch": 2.464161667192927,
      "grad_norm": 1.9746136665344238,
      "learning_rate": 2.8176970037190377e-05,
      "loss": 0.6739,
      "step": 7804
    },
    {
      "epoch": 2.4644774234291127,
      "grad_norm": 2.0744900703430176,
      "learning_rate": 2.817346151147288e-05,
      "loss": 0.7396,
      "step": 7805
    },
    {
      "epoch": 2.4647931796652984,
      "grad_norm": 2.4486775398254395,
      "learning_rate": 2.8169952985755387e-05,
      "loss": 0.7823,
      "step": 7806
    },
    {
      "epoch": 2.465108935901484,
      "grad_norm": 2.0709025859832764,
      "learning_rate": 2.8166444460037894e-05,
      "loss": 0.751,
      "step": 7807
    },
    {
      "epoch": 2.46542469213767,
      "grad_norm": 2.633234977722168,
      "learning_rate": 2.81629359343204e-05,
      "loss": 0.7602,
      "step": 7808
    },
    {
      "epoch": 2.465740448373855,
      "grad_norm": 2.468092441558838,
      "learning_rate": 2.8159427408602907e-05,
      "loss": 0.7761,
      "step": 7809
    },
    {
      "epoch": 2.466056204610041,
      "grad_norm": 2.4878504276275635,
      "learning_rate": 2.815591888288541e-05,
      "loss": 0.7738,
      "step": 7810
    },
    {
      "epoch": 2.4663719608462267,
      "grad_norm": 2.1919963359832764,
      "learning_rate": 2.8152410357167918e-05,
      "loss": 0.7523,
      "step": 7811
    },
    {
      "epoch": 2.4666877170824124,
      "grad_norm": 2.332583427429199,
      "learning_rate": 2.8148901831450425e-05,
      "loss": 0.7805,
      "step": 7812
    },
    {
      "epoch": 2.467003473318598,
      "grad_norm": 2.4489905834198,
      "learning_rate": 2.814539330573293e-05,
      "loss": 0.7419,
      "step": 7813
    },
    {
      "epoch": 2.467319229554784,
      "grad_norm": 2.183854579925537,
      "learning_rate": 2.8141884780015438e-05,
      "loss": 0.6888,
      "step": 7814
    },
    {
      "epoch": 2.467634985790969,
      "grad_norm": 2.2216954231262207,
      "learning_rate": 2.8138376254297945e-05,
      "loss": 0.7753,
      "step": 7815
    },
    {
      "epoch": 2.467950742027155,
      "grad_norm": 2.664097547531128,
      "learning_rate": 2.813486772858045e-05,
      "loss": 0.7614,
      "step": 7816
    },
    {
      "epoch": 2.4682664982633407,
      "grad_norm": 2.284264326095581,
      "learning_rate": 2.8131359202862955e-05,
      "loss": 0.7527,
      "step": 7817
    },
    {
      "epoch": 2.4685822544995264,
      "grad_norm": 2.470189094543457,
      "learning_rate": 2.812785067714547e-05,
      "loss": 0.7217,
      "step": 7818
    },
    {
      "epoch": 2.468898010735712,
      "grad_norm": 2.133427143096924,
      "learning_rate": 2.8124342151427972e-05,
      "loss": 0.7914,
      "step": 7819
    },
    {
      "epoch": 2.469213766971898,
      "grad_norm": 2.1593017578125,
      "learning_rate": 2.812083362571048e-05,
      "loss": 0.7195,
      "step": 7820
    },
    {
      "epoch": 2.469529523208083,
      "grad_norm": 2.3472225666046143,
      "learning_rate": 2.8117325099992986e-05,
      "loss": 0.6206,
      "step": 7821
    },
    {
      "epoch": 2.469845279444269,
      "grad_norm": 2.3892202377319336,
      "learning_rate": 2.8113816574275493e-05,
      "loss": 0.7927,
      "step": 7822
    },
    {
      "epoch": 2.4701610356804546,
      "grad_norm": 2.3585803508758545,
      "learning_rate": 2.8110308048558e-05,
      "loss": 0.7147,
      "step": 7823
    },
    {
      "epoch": 2.4704767919166404,
      "grad_norm": 2.3597376346588135,
      "learning_rate": 2.8106799522840503e-05,
      "loss": 0.7292,
      "step": 7824
    },
    {
      "epoch": 2.470792548152826,
      "grad_norm": 2.255868911743164,
      "learning_rate": 2.810329099712301e-05,
      "loss": 0.7679,
      "step": 7825
    },
    {
      "epoch": 2.471108304389012,
      "grad_norm": 2.415968656539917,
      "learning_rate": 2.8099782471405517e-05,
      "loss": 0.7566,
      "step": 7826
    },
    {
      "epoch": 2.471424060625197,
      "grad_norm": 2.3935115337371826,
      "learning_rate": 2.8096273945688023e-05,
      "loss": 0.631,
      "step": 7827
    },
    {
      "epoch": 2.471739816861383,
      "grad_norm": 2.3162457942962646,
      "learning_rate": 2.809276541997053e-05,
      "loss": 0.7716,
      "step": 7828
    },
    {
      "epoch": 2.4720555730975686,
      "grad_norm": 2.250883102416992,
      "learning_rate": 2.8089256894253037e-05,
      "loss": 0.732,
      "step": 7829
    },
    {
      "epoch": 2.4723713293337544,
      "grad_norm": 2.277256965637207,
      "learning_rate": 2.808574836853554e-05,
      "loss": 0.7008,
      "step": 7830
    },
    {
      "epoch": 2.47268708556994,
      "grad_norm": 2.610682964324951,
      "learning_rate": 2.8082239842818047e-05,
      "loss": 0.7287,
      "step": 7831
    },
    {
      "epoch": 2.473002841806126,
      "grad_norm": 2.1405746936798096,
      "learning_rate": 2.8078731317100554e-05,
      "loss": 0.6856,
      "step": 7832
    },
    {
      "epoch": 2.4733185980423116,
      "grad_norm": 2.183805227279663,
      "learning_rate": 2.807522279138306e-05,
      "loss": 0.6793,
      "step": 7833
    },
    {
      "epoch": 2.473634354278497,
      "grad_norm": 2.087543487548828,
      "learning_rate": 2.8071714265665568e-05,
      "loss": 0.6904,
      "step": 7834
    },
    {
      "epoch": 2.4739501105146826,
      "grad_norm": 2.4307799339294434,
      "learning_rate": 2.806820573994807e-05,
      "loss": 0.7229,
      "step": 7835
    },
    {
      "epoch": 2.4742658667508683,
      "grad_norm": 2.844841480255127,
      "learning_rate": 2.8064697214230585e-05,
      "loss": 0.7487,
      "step": 7836
    },
    {
      "epoch": 2.474581622987054,
      "grad_norm": 2.7362899780273438,
      "learning_rate": 2.806118868851309e-05,
      "loss": 0.7725,
      "step": 7837
    },
    {
      "epoch": 2.47489737922324,
      "grad_norm": 2.343569040298462,
      "learning_rate": 2.8057680162795595e-05,
      "loss": 0.7824,
      "step": 7838
    },
    {
      "epoch": 2.475213135459425,
      "grad_norm": 2.5190138816833496,
      "learning_rate": 2.8054171637078102e-05,
      "loss": 0.7181,
      "step": 7839
    },
    {
      "epoch": 2.475528891695611,
      "grad_norm": 2.467163562774658,
      "learning_rate": 2.805066311136061e-05,
      "loss": 0.7545,
      "step": 7840
    },
    {
      "epoch": 2.4758446479317966,
      "grad_norm": 2.58988094329834,
      "learning_rate": 2.8047154585643115e-05,
      "loss": 0.7764,
      "step": 7841
    },
    {
      "epoch": 2.4761604041679823,
      "grad_norm": 2.3242063522338867,
      "learning_rate": 2.8043646059925622e-05,
      "loss": 0.7057,
      "step": 7842
    },
    {
      "epoch": 2.476476160404168,
      "grad_norm": 2.401400327682495,
      "learning_rate": 2.804013753420813e-05,
      "loss": 0.6801,
      "step": 7843
    },
    {
      "epoch": 2.476791916640354,
      "grad_norm": 2.4459807872772217,
      "learning_rate": 2.8036629008490632e-05,
      "loss": 0.7366,
      "step": 7844
    },
    {
      "epoch": 2.4771076728765395,
      "grad_norm": 2.429370403289795,
      "learning_rate": 2.803312048277314e-05,
      "loss": 0.6734,
      "step": 7845
    },
    {
      "epoch": 2.477423429112725,
      "grad_norm": 2.477447986602783,
      "learning_rate": 2.8029611957055646e-05,
      "loss": 0.6654,
      "step": 7846
    },
    {
      "epoch": 2.4777391853489106,
      "grad_norm": 2.8639168739318848,
      "learning_rate": 2.8026103431338153e-05,
      "loss": 0.6818,
      "step": 7847
    },
    {
      "epoch": 2.4780549415850963,
      "grad_norm": 2.277315616607666,
      "learning_rate": 2.802259490562066e-05,
      "loss": 0.714,
      "step": 7848
    },
    {
      "epoch": 2.478370697821282,
      "grad_norm": 2.210090398788452,
      "learning_rate": 2.8019086379903163e-05,
      "loss": 0.7368,
      "step": 7849
    },
    {
      "epoch": 2.478686454057468,
      "grad_norm": 2.2959659099578857,
      "learning_rate": 2.801557785418567e-05,
      "loss": 0.7667,
      "step": 7850
    },
    {
      "epoch": 2.479002210293653,
      "grad_norm": 2.6856701374053955,
      "learning_rate": 2.8012069328468177e-05,
      "loss": 0.7243,
      "step": 7851
    },
    {
      "epoch": 2.479317966529839,
      "grad_norm": 2.6359176635742188,
      "learning_rate": 2.8008560802750683e-05,
      "loss": 0.8474,
      "step": 7852
    },
    {
      "epoch": 2.4796337227660246,
      "grad_norm": 2.3390634059906006,
      "learning_rate": 2.8005052277033194e-05,
      "loss": 0.7775,
      "step": 7853
    },
    {
      "epoch": 2.4799494790022103,
      "grad_norm": 2.628613233566284,
      "learning_rate": 2.80015437513157e-05,
      "loss": 0.8278,
      "step": 7854
    },
    {
      "epoch": 2.480265235238396,
      "grad_norm": 2.5541603565216064,
      "learning_rate": 2.7998035225598207e-05,
      "loss": 0.6995,
      "step": 7855
    },
    {
      "epoch": 2.4805809914745818,
      "grad_norm": 2.4217591285705566,
      "learning_rate": 2.7994526699880714e-05,
      "loss": 0.7005,
      "step": 7856
    },
    {
      "epoch": 2.4808967477107675,
      "grad_norm": 2.6519956588745117,
      "learning_rate": 2.799101817416322e-05,
      "loss": 0.8001,
      "step": 7857
    },
    {
      "epoch": 2.481212503946953,
      "grad_norm": 2.37377667427063,
      "learning_rate": 2.7987509648445724e-05,
      "loss": 0.6323,
      "step": 7858
    },
    {
      "epoch": 2.4815282601831385,
      "grad_norm": 2.624235153198242,
      "learning_rate": 2.798400112272823e-05,
      "loss": 0.7496,
      "step": 7859
    },
    {
      "epoch": 2.4818440164193243,
      "grad_norm": 2.4483530521392822,
      "learning_rate": 2.7980492597010738e-05,
      "loss": 0.7747,
      "step": 7860
    },
    {
      "epoch": 2.48215977265551,
      "grad_norm": 2.9636473655700684,
      "learning_rate": 2.7976984071293245e-05,
      "loss": 0.6588,
      "step": 7861
    },
    {
      "epoch": 2.4824755288916958,
      "grad_norm": 2.4214236736297607,
      "learning_rate": 2.797347554557575e-05,
      "loss": 0.7003,
      "step": 7862
    },
    {
      "epoch": 2.482791285127881,
      "grad_norm": 2.476369857788086,
      "learning_rate": 2.7969967019858255e-05,
      "loss": 0.8143,
      "step": 7863
    },
    {
      "epoch": 2.483107041364067,
      "grad_norm": 2.969175100326538,
      "learning_rate": 2.7966458494140762e-05,
      "loss": 0.7454,
      "step": 7864
    },
    {
      "epoch": 2.4834227976002525,
      "grad_norm": 2.572073459625244,
      "learning_rate": 2.796294996842327e-05,
      "loss": 0.8051,
      "step": 7865
    },
    {
      "epoch": 2.4837385538364383,
      "grad_norm": 2.400617837905884,
      "learning_rate": 2.7959441442705775e-05,
      "loss": 0.7007,
      "step": 7866
    },
    {
      "epoch": 2.484054310072624,
      "grad_norm": 2.6322731971740723,
      "learning_rate": 2.7955932916988282e-05,
      "loss": 0.6602,
      "step": 7867
    },
    {
      "epoch": 2.4843700663088097,
      "grad_norm": 2.187419891357422,
      "learning_rate": 2.7952424391270786e-05,
      "loss": 0.6799,
      "step": 7868
    },
    {
      "epoch": 2.4846858225449955,
      "grad_norm": 2.4301605224609375,
      "learning_rate": 2.7948915865553293e-05,
      "loss": 0.7434,
      "step": 7869
    },
    {
      "epoch": 2.4850015787811808,
      "grad_norm": 2.4018242359161377,
      "learning_rate": 2.79454073398358e-05,
      "loss": 0.8372,
      "step": 7870
    },
    {
      "epoch": 2.4853173350173665,
      "grad_norm": 2.419034481048584,
      "learning_rate": 2.7941898814118313e-05,
      "loss": 0.759,
      "step": 7871
    },
    {
      "epoch": 2.4856330912535523,
      "grad_norm": 2.394876480102539,
      "learning_rate": 2.7938390288400816e-05,
      "loss": 0.6638,
      "step": 7872
    },
    {
      "epoch": 2.485948847489738,
      "grad_norm": 2.5141539573669434,
      "learning_rate": 2.7934881762683323e-05,
      "loss": 0.7068,
      "step": 7873
    },
    {
      "epoch": 2.4862646037259237,
      "grad_norm": 3.3256542682647705,
      "learning_rate": 2.793137323696583e-05,
      "loss": 0.7367,
      "step": 7874
    },
    {
      "epoch": 2.486580359962109,
      "grad_norm": 2.73638916015625,
      "learning_rate": 2.7927864711248337e-05,
      "loss": 0.7477,
      "step": 7875
    },
    {
      "epoch": 2.4868961161982948,
      "grad_norm": 2.1993014812469482,
      "learning_rate": 2.7924356185530844e-05,
      "loss": 0.6758,
      "step": 7876
    },
    {
      "epoch": 2.4872118724344805,
      "grad_norm": 2.4741551876068115,
      "learning_rate": 2.7920847659813347e-05,
      "loss": 0.8704,
      "step": 7877
    },
    {
      "epoch": 2.4875276286706662,
      "grad_norm": 2.4817609786987305,
      "learning_rate": 2.7917339134095854e-05,
      "loss": 0.7593,
      "step": 7878
    },
    {
      "epoch": 2.487843384906852,
      "grad_norm": 2.3690831661224365,
      "learning_rate": 2.791383060837836e-05,
      "loss": 0.716,
      "step": 7879
    },
    {
      "epoch": 2.4881591411430377,
      "grad_norm": 2.0725345611572266,
      "learning_rate": 2.7910322082660867e-05,
      "loss": 0.7442,
      "step": 7880
    },
    {
      "epoch": 2.4884748973792234,
      "grad_norm": 2.355931520462036,
      "learning_rate": 2.7906813556943374e-05,
      "loss": 0.8229,
      "step": 7881
    },
    {
      "epoch": 2.4887906536154087,
      "grad_norm": 2.3445324897766113,
      "learning_rate": 2.7903305031225878e-05,
      "loss": 0.778,
      "step": 7882
    },
    {
      "epoch": 2.4891064098515945,
      "grad_norm": 2.544325113296509,
      "learning_rate": 2.7899796505508385e-05,
      "loss": 0.7133,
      "step": 7883
    },
    {
      "epoch": 2.48942216608778,
      "grad_norm": 2.2660303115844727,
      "learning_rate": 2.789628797979089e-05,
      "loss": 0.7189,
      "step": 7884
    },
    {
      "epoch": 2.489737922323966,
      "grad_norm": 2.211211919784546,
      "learning_rate": 2.7892779454073398e-05,
      "loss": 0.6776,
      "step": 7885
    },
    {
      "epoch": 2.4900536785601517,
      "grad_norm": 2.2115790843963623,
      "learning_rate": 2.7889270928355905e-05,
      "loss": 0.7173,
      "step": 7886
    },
    {
      "epoch": 2.490369434796337,
      "grad_norm": 2.6991655826568604,
      "learning_rate": 2.7885762402638412e-05,
      "loss": 0.7654,
      "step": 7887
    },
    {
      "epoch": 2.4906851910325227,
      "grad_norm": 2.7313175201416016,
      "learning_rate": 2.7882253876920922e-05,
      "loss": 0.7635,
      "step": 7888
    },
    {
      "epoch": 2.4910009472687085,
      "grad_norm": 3.2592592239379883,
      "learning_rate": 2.787874535120343e-05,
      "loss": 0.7682,
      "step": 7889
    },
    {
      "epoch": 2.491316703504894,
      "grad_norm": 3.1037557125091553,
      "learning_rate": 2.7875236825485936e-05,
      "loss": 0.82,
      "step": 7890
    },
    {
      "epoch": 2.49163245974108,
      "grad_norm": 2.3602840900421143,
      "learning_rate": 2.787172829976844e-05,
      "loss": 0.6315,
      "step": 7891
    },
    {
      "epoch": 2.4919482159772657,
      "grad_norm": 2.5900259017944336,
      "learning_rate": 2.7868219774050946e-05,
      "loss": 0.8301,
      "step": 7892
    },
    {
      "epoch": 2.4922639722134514,
      "grad_norm": 2.418797254562378,
      "learning_rate": 2.7864711248333453e-05,
      "loss": 0.7439,
      "step": 7893
    },
    {
      "epoch": 2.4925797284496367,
      "grad_norm": 2.944286584854126,
      "learning_rate": 2.786120272261596e-05,
      "loss": 0.7913,
      "step": 7894
    },
    {
      "epoch": 2.4928954846858224,
      "grad_norm": 2.264598846435547,
      "learning_rate": 2.7857694196898466e-05,
      "loss": 0.7245,
      "step": 7895
    },
    {
      "epoch": 2.493211240922008,
      "grad_norm": 2.308166265487671,
      "learning_rate": 2.785418567118097e-05,
      "loss": 0.6907,
      "step": 7896
    },
    {
      "epoch": 2.493526997158194,
      "grad_norm": 2.416511058807373,
      "learning_rate": 2.7850677145463477e-05,
      "loss": 0.7256,
      "step": 7897
    },
    {
      "epoch": 2.4938427533943797,
      "grad_norm": 2.35420298576355,
      "learning_rate": 2.7847168619745983e-05,
      "loss": 0.726,
      "step": 7898
    },
    {
      "epoch": 2.494158509630565,
      "grad_norm": 2.144544839859009,
      "learning_rate": 2.784366009402849e-05,
      "loss": 0.6385,
      "step": 7899
    },
    {
      "epoch": 2.4944742658667507,
      "grad_norm": 3.389977216720581,
      "learning_rate": 2.7840151568310997e-05,
      "loss": 0.7849,
      "step": 7900
    },
    {
      "epoch": 2.4947900221029364,
      "grad_norm": 2.643062114715576,
      "learning_rate": 2.78366430425935e-05,
      "loss": 0.7384,
      "step": 7901
    },
    {
      "epoch": 2.495105778339122,
      "grad_norm": 2.3465049266815186,
      "learning_rate": 2.7833134516876007e-05,
      "loss": 0.7429,
      "step": 7902
    },
    {
      "epoch": 2.495421534575308,
      "grad_norm": 2.2996137142181396,
      "learning_rate": 2.7829625991158514e-05,
      "loss": 0.7754,
      "step": 7903
    },
    {
      "epoch": 2.4957372908114936,
      "grad_norm": 2.136007070541382,
      "learning_rate": 2.782611746544102e-05,
      "loss": 0.6622,
      "step": 7904
    },
    {
      "epoch": 2.4960530470476794,
      "grad_norm": 2.3708810806274414,
      "learning_rate": 2.7822608939723528e-05,
      "loss": 0.7159,
      "step": 7905
    },
    {
      "epoch": 2.4963688032838647,
      "grad_norm": 2.3840689659118652,
      "learning_rate": 2.7819100414006038e-05,
      "loss": 0.7169,
      "step": 7906
    },
    {
      "epoch": 2.4966845595200504,
      "grad_norm": 2.689486503601074,
      "learning_rate": 2.7815591888288545e-05,
      "loss": 0.6968,
      "step": 7907
    },
    {
      "epoch": 2.497000315756236,
      "grad_norm": 2.379429340362549,
      "learning_rate": 2.781208336257105e-05,
      "loss": 0.7513,
      "step": 7908
    },
    {
      "epoch": 2.497316071992422,
      "grad_norm": 2.71352219581604,
      "learning_rate": 2.7808574836853558e-05,
      "loss": 0.7944,
      "step": 7909
    },
    {
      "epoch": 2.4976318282286076,
      "grad_norm": 2.3644139766693115,
      "learning_rate": 2.7805066311136062e-05,
      "loss": 0.8672,
      "step": 7910
    },
    {
      "epoch": 2.4979475844647934,
      "grad_norm": 2.6670258045196533,
      "learning_rate": 2.780155778541857e-05,
      "loss": 0.6849,
      "step": 7911
    },
    {
      "epoch": 2.4982633407009787,
      "grad_norm": 2.5418779850006104,
      "learning_rate": 2.7798049259701075e-05,
      "loss": 0.8249,
      "step": 7912
    },
    {
      "epoch": 2.4985790969371644,
      "grad_norm": 1.9803799390792847,
      "learning_rate": 2.7794540733983582e-05,
      "loss": 0.6896,
      "step": 7913
    },
    {
      "epoch": 2.49889485317335,
      "grad_norm": 2.265249490737915,
      "learning_rate": 2.779103220826609e-05,
      "loss": 0.6759,
      "step": 7914
    },
    {
      "epoch": 2.499210609409536,
      "grad_norm": 2.415834426879883,
      "learning_rate": 2.7787523682548592e-05,
      "loss": 0.6709,
      "step": 7915
    },
    {
      "epoch": 2.4995263656457216,
      "grad_norm": 3.0948057174682617,
      "learning_rate": 2.77840151568311e-05,
      "loss": 0.7301,
      "step": 7916
    },
    {
      "epoch": 2.4998421218819074,
      "grad_norm": 2.397331476211548,
      "learning_rate": 2.7780506631113606e-05,
      "loss": 0.7702,
      "step": 7917
    },
    {
      "epoch": 2.5001578781180926,
      "grad_norm": 2.410745620727539,
      "learning_rate": 2.7776998105396113e-05,
      "loss": 0.7555,
      "step": 7918
    },
    {
      "epoch": 2.5004736343542784,
      "grad_norm": 2.4107954502105713,
      "learning_rate": 2.777348957967862e-05,
      "loss": 0.6749,
      "step": 7919
    },
    {
      "epoch": 2.500789390590464,
      "grad_norm": 2.3475112915039062,
      "learning_rate": 2.7769981053961126e-05,
      "loss": 0.7336,
      "step": 7920
    },
    {
      "epoch": 2.50110514682665,
      "grad_norm": 2.8081724643707275,
      "learning_rate": 2.776647252824363e-05,
      "loss": 0.7622,
      "step": 7921
    },
    {
      "epoch": 2.5014209030628356,
      "grad_norm": 2.4061834812164307,
      "learning_rate": 2.7762964002526137e-05,
      "loss": 0.6696,
      "step": 7922
    },
    {
      "epoch": 2.501736659299021,
      "grad_norm": 2.065983772277832,
      "learning_rate": 2.775945547680865e-05,
      "loss": 0.7613,
      "step": 7923
    },
    {
      "epoch": 2.502052415535207,
      "grad_norm": 2.6889243125915527,
      "learning_rate": 2.7755946951091154e-05,
      "loss": 0.7024,
      "step": 7924
    },
    {
      "epoch": 2.5023681717713924,
      "grad_norm": 2.6158463954925537,
      "learning_rate": 2.775243842537366e-05,
      "loss": 0.6718,
      "step": 7925
    },
    {
      "epoch": 2.502683928007578,
      "grad_norm": 2.386751413345337,
      "learning_rate": 2.7748929899656167e-05,
      "loss": 0.7136,
      "step": 7926
    },
    {
      "epoch": 2.502999684243764,
      "grad_norm": 2.3853461742401123,
      "learning_rate": 2.7745421373938674e-05,
      "loss": 0.6937,
      "step": 7927
    },
    {
      "epoch": 2.5033154404799496,
      "grad_norm": 2.3774282932281494,
      "learning_rate": 2.774191284822118e-05,
      "loss": 0.723,
      "step": 7928
    },
    {
      "epoch": 2.5036311967161353,
      "grad_norm": 2.5409486293792725,
      "learning_rate": 2.7738404322503684e-05,
      "loss": 0.7361,
      "step": 7929
    },
    {
      "epoch": 2.5039469529523206,
      "grad_norm": 2.7457022666931152,
      "learning_rate": 2.773489579678619e-05,
      "loss": 0.7066,
      "step": 7930
    },
    {
      "epoch": 2.5042627091885064,
      "grad_norm": 2.303001880645752,
      "learning_rate": 2.7731387271068698e-05,
      "loss": 0.7209,
      "step": 7931
    },
    {
      "epoch": 2.504578465424692,
      "grad_norm": 2.7409348487854004,
      "learning_rate": 2.7727878745351205e-05,
      "loss": 0.7912,
      "step": 7932
    },
    {
      "epoch": 2.504894221660878,
      "grad_norm": 1.934404969215393,
      "learning_rate": 2.772437021963371e-05,
      "loss": 0.7261,
      "step": 7933
    },
    {
      "epoch": 2.5052099778970636,
      "grad_norm": 2.526308059692383,
      "learning_rate": 2.772086169391622e-05,
      "loss": 0.8449,
      "step": 7934
    },
    {
      "epoch": 2.505525734133249,
      "grad_norm": 2.3379437923431396,
      "learning_rate": 2.7717353168198722e-05,
      "loss": 0.729,
      "step": 7935
    },
    {
      "epoch": 2.505841490369435,
      "grad_norm": 2.321218252182007,
      "learning_rate": 2.771384464248123e-05,
      "loss": 0.7237,
      "step": 7936
    },
    {
      "epoch": 2.5061572466056203,
      "grad_norm": 2.5295259952545166,
      "learning_rate": 2.7710336116763735e-05,
      "loss": 0.8507,
      "step": 7937
    },
    {
      "epoch": 2.506473002841806,
      "grad_norm": 2.4350407123565674,
      "learning_rate": 2.7706827591046242e-05,
      "loss": 0.7069,
      "step": 7938
    },
    {
      "epoch": 2.506788759077992,
      "grad_norm": 2.1288633346557617,
      "learning_rate": 2.770331906532875e-05,
      "loss": 0.6553,
      "step": 7939
    },
    {
      "epoch": 2.5071045153141776,
      "grad_norm": 2.712320566177368,
      "learning_rate": 2.7699810539611253e-05,
      "loss": 0.6779,
      "step": 7940
    },
    {
      "epoch": 2.5074202715503633,
      "grad_norm": 2.5447661876678467,
      "learning_rate": 2.7696302013893766e-05,
      "loss": 0.7492,
      "step": 7941
    },
    {
      "epoch": 2.5077360277865486,
      "grad_norm": 2.4537196159362793,
      "learning_rate": 2.7692793488176273e-05,
      "loss": 0.6752,
      "step": 7942
    },
    {
      "epoch": 2.5080517840227343,
      "grad_norm": 2.3235933780670166,
      "learning_rate": 2.7689284962458776e-05,
      "loss": 0.7745,
      "step": 7943
    },
    {
      "epoch": 2.50836754025892,
      "grad_norm": 2.8655431270599365,
      "learning_rate": 2.7685776436741283e-05,
      "loss": 0.7338,
      "step": 7944
    },
    {
      "epoch": 2.508683296495106,
      "grad_norm": 2.045456647872925,
      "learning_rate": 2.768226791102379e-05,
      "loss": 0.5939,
      "step": 7945
    },
    {
      "epoch": 2.5089990527312915,
      "grad_norm": 2.088391065597534,
      "learning_rate": 2.7678759385306297e-05,
      "loss": 0.6948,
      "step": 7946
    },
    {
      "epoch": 2.5093148089674773,
      "grad_norm": 2.8709726333618164,
      "learning_rate": 2.7675250859588804e-05,
      "loss": 0.7281,
      "step": 7947
    },
    {
      "epoch": 2.509630565203663,
      "grad_norm": 2.3231163024902344,
      "learning_rate": 2.767174233387131e-05,
      "loss": 0.704,
      "step": 7948
    },
    {
      "epoch": 2.5099463214398483,
      "grad_norm": 2.7960054874420166,
      "learning_rate": 2.7668233808153814e-05,
      "loss": 0.7618,
      "step": 7949
    },
    {
      "epoch": 2.510262077676034,
      "grad_norm": 2.6462509632110596,
      "learning_rate": 2.766472528243632e-05,
      "loss": 0.7265,
      "step": 7950
    },
    {
      "epoch": 2.51057783391222,
      "grad_norm": 2.7217326164245605,
      "learning_rate": 2.7661216756718827e-05,
      "loss": 0.7651,
      "step": 7951
    },
    {
      "epoch": 2.5108935901484055,
      "grad_norm": 2.327256679534912,
      "learning_rate": 2.7657708231001334e-05,
      "loss": 0.7313,
      "step": 7952
    },
    {
      "epoch": 2.5112093463845913,
      "grad_norm": 2.3031861782073975,
      "learning_rate": 2.765419970528384e-05,
      "loss": 0.7871,
      "step": 7953
    },
    {
      "epoch": 2.5115251026207766,
      "grad_norm": 2.7553248405456543,
      "learning_rate": 2.7650691179566345e-05,
      "loss": 0.7851,
      "step": 7954
    },
    {
      "epoch": 2.5118408588569623,
      "grad_norm": 2.29787015914917,
      "learning_rate": 2.764718265384885e-05,
      "loss": 0.6809,
      "step": 7955
    },
    {
      "epoch": 2.512156615093148,
      "grad_norm": 2.7689268589019775,
      "learning_rate": 2.7643674128131358e-05,
      "loss": 0.7583,
      "step": 7956
    },
    {
      "epoch": 2.5124723713293338,
      "grad_norm": 2.67826771736145,
      "learning_rate": 2.7640165602413865e-05,
      "loss": 0.7323,
      "step": 7957
    },
    {
      "epoch": 2.5127881275655195,
      "grad_norm": 2.370474100112915,
      "learning_rate": 2.7636657076696375e-05,
      "loss": 0.73,
      "step": 7958
    },
    {
      "epoch": 2.5131038838017052,
      "grad_norm": 2.4333252906799316,
      "learning_rate": 2.7633148550978882e-05,
      "loss": 0.7276,
      "step": 7959
    },
    {
      "epoch": 2.513419640037891,
      "grad_norm": 2.2231945991516113,
      "learning_rate": 2.762964002526139e-05,
      "loss": 0.7761,
      "step": 7960
    },
    {
      "epoch": 2.5137353962740763,
      "grad_norm": 2.8575077056884766,
      "learning_rate": 2.7626131499543896e-05,
      "loss": 0.7103,
      "step": 7961
    },
    {
      "epoch": 2.514051152510262,
      "grad_norm": 2.5254480838775635,
      "learning_rate": 2.7622622973826402e-05,
      "loss": 0.7069,
      "step": 7962
    },
    {
      "epoch": 2.5143669087464477,
      "grad_norm": 2.623673677444458,
      "learning_rate": 2.7619114448108906e-05,
      "loss": 0.7384,
      "step": 7963
    },
    {
      "epoch": 2.5146826649826335,
      "grad_norm": 3.204202890396118,
      "learning_rate": 2.7615605922391413e-05,
      "loss": 0.8515,
      "step": 7964
    },
    {
      "epoch": 2.5149984212188192,
      "grad_norm": 2.281148672103882,
      "learning_rate": 2.761209739667392e-05,
      "loss": 0.6984,
      "step": 7965
    },
    {
      "epoch": 2.5153141774550045,
      "grad_norm": 2.939025402069092,
      "learning_rate": 2.7608588870956426e-05,
      "loss": 0.7118,
      "step": 7966
    },
    {
      "epoch": 2.5156299336911903,
      "grad_norm": 2.5626325607299805,
      "learning_rate": 2.7605080345238933e-05,
      "loss": 0.7334,
      "step": 7967
    },
    {
      "epoch": 2.515945689927376,
      "grad_norm": 2.4959731101989746,
      "learning_rate": 2.7601571819521437e-05,
      "loss": 0.7703,
      "step": 7968
    },
    {
      "epoch": 2.5162614461635617,
      "grad_norm": 2.405776262283325,
      "learning_rate": 2.7598063293803943e-05,
      "loss": 0.791,
      "step": 7969
    },
    {
      "epoch": 2.5165772023997475,
      "grad_norm": 2.1625852584838867,
      "learning_rate": 2.759455476808645e-05,
      "loss": 0.7208,
      "step": 7970
    },
    {
      "epoch": 2.516892958635933,
      "grad_norm": 2.4721813201904297,
      "learning_rate": 2.7591046242368957e-05,
      "loss": 0.7348,
      "step": 7971
    },
    {
      "epoch": 2.517208714872119,
      "grad_norm": 2.2660958766937256,
      "learning_rate": 2.7587537716651464e-05,
      "loss": 0.67,
      "step": 7972
    },
    {
      "epoch": 2.5175244711083042,
      "grad_norm": 2.7491722106933594,
      "learning_rate": 2.7584029190933967e-05,
      "loss": 0.7033,
      "step": 7973
    },
    {
      "epoch": 2.51784022734449,
      "grad_norm": 2.3038158416748047,
      "learning_rate": 2.7580520665216474e-05,
      "loss": 0.6774,
      "step": 7974
    },
    {
      "epoch": 2.5181559835806757,
      "grad_norm": 2.319641351699829,
      "learning_rate": 2.757701213949898e-05,
      "loss": 0.7273,
      "step": 7975
    },
    {
      "epoch": 2.5184717398168615,
      "grad_norm": 2.575176954269409,
      "learning_rate": 2.7573503613781494e-05,
      "loss": 0.6405,
      "step": 7976
    },
    {
      "epoch": 2.518787496053047,
      "grad_norm": 2.2330894470214844,
      "learning_rate": 2.7569995088063998e-05,
      "loss": 0.6687,
      "step": 7977
    },
    {
      "epoch": 2.5191032522892325,
      "grad_norm": 2.8626718521118164,
      "learning_rate": 2.7566486562346505e-05,
      "loss": 0.8114,
      "step": 7978
    },
    {
      "epoch": 2.5194190085254182,
      "grad_norm": 2.9950969219207764,
      "learning_rate": 2.756297803662901e-05,
      "loss": 0.8135,
      "step": 7979
    },
    {
      "epoch": 2.519734764761604,
      "grad_norm": 2.15744948387146,
      "learning_rate": 2.7559469510911518e-05,
      "loss": 0.6706,
      "step": 7980
    },
    {
      "epoch": 2.5200505209977897,
      "grad_norm": 3.2104902267456055,
      "learning_rate": 2.7555960985194025e-05,
      "loss": 0.8694,
      "step": 7981
    },
    {
      "epoch": 2.5203662772339754,
      "grad_norm": 2.450117349624634,
      "learning_rate": 2.755245245947653e-05,
      "loss": 0.7274,
      "step": 7982
    },
    {
      "epoch": 2.520682033470161,
      "grad_norm": 2.234478235244751,
      "learning_rate": 2.7548943933759035e-05,
      "loss": 0.7066,
      "step": 7983
    },
    {
      "epoch": 2.520997789706347,
      "grad_norm": 2.559617757797241,
      "learning_rate": 2.7545435408041542e-05,
      "loss": 0.8363,
      "step": 7984
    },
    {
      "epoch": 2.521313545942532,
      "grad_norm": 2.554666757583618,
      "learning_rate": 2.754192688232405e-05,
      "loss": 0.7184,
      "step": 7985
    },
    {
      "epoch": 2.521629302178718,
      "grad_norm": 2.3382070064544678,
      "learning_rate": 2.7538418356606556e-05,
      "loss": 0.7293,
      "step": 7986
    },
    {
      "epoch": 2.5219450584149037,
      "grad_norm": 2.771143674850464,
      "learning_rate": 2.753490983088906e-05,
      "loss": 0.72,
      "step": 7987
    },
    {
      "epoch": 2.5222608146510894,
      "grad_norm": 2.5095536708831787,
      "learning_rate": 2.7531401305171566e-05,
      "loss": 0.7956,
      "step": 7988
    },
    {
      "epoch": 2.522576570887275,
      "grad_norm": 2.2074074745178223,
      "learning_rate": 2.7527892779454073e-05,
      "loss": 0.6876,
      "step": 7989
    },
    {
      "epoch": 2.5228923271234605,
      "grad_norm": 2.619969129562378,
      "learning_rate": 2.752438425373658e-05,
      "loss": 0.8544,
      "step": 7990
    },
    {
      "epoch": 2.523208083359646,
      "grad_norm": 2.3631112575531006,
      "learning_rate": 2.7520875728019086e-05,
      "loss": 0.6662,
      "step": 7991
    },
    {
      "epoch": 2.523523839595832,
      "grad_norm": 2.494305372238159,
      "learning_rate": 2.7517367202301593e-05,
      "loss": 0.7141,
      "step": 7992
    },
    {
      "epoch": 2.5238395958320177,
      "grad_norm": 3.039639472961426,
      "learning_rate": 2.7513858676584103e-05,
      "loss": 0.7918,
      "step": 7993
    },
    {
      "epoch": 2.5241553520682034,
      "grad_norm": 2.5698022842407227,
      "learning_rate": 2.751035015086661e-05,
      "loss": 0.7693,
      "step": 7994
    },
    {
      "epoch": 2.524471108304389,
      "grad_norm": 2.485750198364258,
      "learning_rate": 2.7506841625149117e-05,
      "loss": 0.7818,
      "step": 7995
    },
    {
      "epoch": 2.524786864540575,
      "grad_norm": 2.1735055446624756,
      "learning_rate": 2.750333309943162e-05,
      "loss": 0.6565,
      "step": 7996
    },
    {
      "epoch": 2.52510262077676,
      "grad_norm": 3.031449556350708,
      "learning_rate": 2.7499824573714127e-05,
      "loss": 0.7045,
      "step": 7997
    },
    {
      "epoch": 2.525418377012946,
      "grad_norm": 2.1990320682525635,
      "learning_rate": 2.7496316047996634e-05,
      "loss": 0.699,
      "step": 7998
    },
    {
      "epoch": 2.5257341332491317,
      "grad_norm": 2.451413154602051,
      "learning_rate": 2.749280752227914e-05,
      "loss": 0.7175,
      "step": 7999
    },
    {
      "epoch": 2.5260498894853174,
      "grad_norm": 2.304077386856079,
      "learning_rate": 2.7489298996561648e-05,
      "loss": 0.6471,
      "step": 8000
    },
    {
      "epoch": 2.5260498894853174,
      "eval_loss": 0.7992247939109802,
      "eval_runtime": 4.6268,
      "eval_samples_per_second": 538.166,
      "eval_steps_per_second": 4.539,
      "step": 8000
    },
    {
      "epoch": 2.526365645721503,
      "grad_norm": 2.1942520141601562,
      "learning_rate": 2.748579047084415e-05,
      "loss": 0.7037,
      "step": 8001
    },
    {
      "epoch": 2.5266814019576884,
      "grad_norm": 2.6961333751678467,
      "learning_rate": 2.7482281945126658e-05,
      "loss": 0.7277,
      "step": 8002
    },
    {
      "epoch": 2.5269971581938746,
      "grad_norm": 2.216153860092163,
      "learning_rate": 2.7478773419409165e-05,
      "loss": 0.7524,
      "step": 8003
    },
    {
      "epoch": 2.52731291443006,
      "grad_norm": 2.2780680656433105,
      "learning_rate": 2.747526489369167e-05,
      "loss": 0.7426,
      "step": 8004
    },
    {
      "epoch": 2.5276286706662456,
      "grad_norm": 2.2658138275146484,
      "learning_rate": 2.747175636797418e-05,
      "loss": 0.768,
      "step": 8005
    },
    {
      "epoch": 2.5279444269024314,
      "grad_norm": 2.469113826751709,
      "learning_rate": 2.7468247842256685e-05,
      "loss": 0.7126,
      "step": 8006
    },
    {
      "epoch": 2.528260183138617,
      "grad_norm": 2.3873491287231445,
      "learning_rate": 2.746473931653919e-05,
      "loss": 0.7163,
      "step": 8007
    },
    {
      "epoch": 2.528575939374803,
      "grad_norm": 2.1008806228637695,
      "learning_rate": 2.7461230790821696e-05,
      "loss": 0.717,
      "step": 8008
    },
    {
      "epoch": 2.528891695610988,
      "grad_norm": 2.6623382568359375,
      "learning_rate": 2.7457722265104202e-05,
      "loss": 0.7583,
      "step": 8009
    },
    {
      "epoch": 2.529207451847174,
      "grad_norm": 2.3446550369262695,
      "learning_rate": 2.745421373938671e-05,
      "loss": 0.6838,
      "step": 8010
    },
    {
      "epoch": 2.5295232080833596,
      "grad_norm": 2.428043842315674,
      "learning_rate": 2.745070521366922e-05,
      "loss": 0.6839,
      "step": 8011
    },
    {
      "epoch": 2.5298389643195454,
      "grad_norm": 2.4026782512664795,
      "learning_rate": 2.7447196687951726e-05,
      "loss": 0.8311,
      "step": 8012
    },
    {
      "epoch": 2.530154720555731,
      "grad_norm": 2.3166725635528564,
      "learning_rate": 2.7443688162234233e-05,
      "loss": 0.7685,
      "step": 8013
    },
    {
      "epoch": 2.5304704767919164,
      "grad_norm": 2.561699151992798,
      "learning_rate": 2.744017963651674e-05,
      "loss": 0.8094,
      "step": 8014
    },
    {
      "epoch": 2.5307862330281026,
      "grad_norm": 2.338182210922241,
      "learning_rate": 2.7436671110799243e-05,
      "loss": 0.7918,
      "step": 8015
    },
    {
      "epoch": 2.531101989264288,
      "grad_norm": 2.4417104721069336,
      "learning_rate": 2.743316258508175e-05,
      "loss": 0.8612,
      "step": 8016
    },
    {
      "epoch": 2.5314177455004736,
      "grad_norm": 2.6747395992279053,
      "learning_rate": 2.7429654059364257e-05,
      "loss": 0.7593,
      "step": 8017
    },
    {
      "epoch": 2.5317335017366593,
      "grad_norm": 2.3994271755218506,
      "learning_rate": 2.7426145533646764e-05,
      "loss": 0.7487,
      "step": 8018
    },
    {
      "epoch": 2.532049257972845,
      "grad_norm": 2.755901336669922,
      "learning_rate": 2.742263700792927e-05,
      "loss": 0.7052,
      "step": 8019
    },
    {
      "epoch": 2.532365014209031,
      "grad_norm": 2.5777928829193115,
      "learning_rate": 2.7419128482211774e-05,
      "loss": 0.7,
      "step": 8020
    },
    {
      "epoch": 2.532680770445216,
      "grad_norm": 2.233024835586548,
      "learning_rate": 2.741561995649428e-05,
      "loss": 0.7945,
      "step": 8021
    },
    {
      "epoch": 2.532996526681402,
      "grad_norm": 2.3469080924987793,
      "learning_rate": 2.7412111430776787e-05,
      "loss": 0.7024,
      "step": 8022
    },
    {
      "epoch": 2.5333122829175876,
      "grad_norm": 2.5457849502563477,
      "learning_rate": 2.7408602905059294e-05,
      "loss": 0.6726,
      "step": 8023
    },
    {
      "epoch": 2.5336280391537733,
      "grad_norm": 2.338838815689087,
      "learning_rate": 2.74050943793418e-05,
      "loss": 0.7513,
      "step": 8024
    },
    {
      "epoch": 2.533943795389959,
      "grad_norm": 2.5848019123077393,
      "learning_rate": 2.7401585853624308e-05,
      "loss": 0.7585,
      "step": 8025
    },
    {
      "epoch": 2.5342595516261444,
      "grad_norm": 2.6091928482055664,
      "learning_rate": 2.739807732790681e-05,
      "loss": 0.8361,
      "step": 8026
    },
    {
      "epoch": 2.5345753078623305,
      "grad_norm": 2.3618180751800537,
      "learning_rate": 2.7394568802189318e-05,
      "loss": 0.737,
      "step": 8027
    },
    {
      "epoch": 2.534891064098516,
      "grad_norm": 2.438875913619995,
      "learning_rate": 2.7391060276471832e-05,
      "loss": 0.63,
      "step": 8028
    },
    {
      "epoch": 2.5352068203347016,
      "grad_norm": 2.4659957885742188,
      "learning_rate": 2.7387551750754335e-05,
      "loss": 0.7301,
      "step": 8029
    },
    {
      "epoch": 2.5355225765708873,
      "grad_norm": 2.2290711402893066,
      "learning_rate": 2.7384043225036842e-05,
      "loss": 0.7434,
      "step": 8030
    },
    {
      "epoch": 2.535838332807073,
      "grad_norm": 2.8138296604156494,
      "learning_rate": 2.738053469931935e-05,
      "loss": 0.8661,
      "step": 8031
    },
    {
      "epoch": 2.536154089043259,
      "grad_norm": 2.3989598751068115,
      "learning_rate": 2.7377026173601856e-05,
      "loss": 0.798,
      "step": 8032
    },
    {
      "epoch": 2.536469845279444,
      "grad_norm": 2.108494758605957,
      "learning_rate": 2.7373517647884362e-05,
      "loss": 0.6913,
      "step": 8033
    },
    {
      "epoch": 2.53678560151563,
      "grad_norm": 2.421457052230835,
      "learning_rate": 2.7370009122166866e-05,
      "loss": 0.7528,
      "step": 8034
    },
    {
      "epoch": 2.5371013577518156,
      "grad_norm": 2.2735564708709717,
      "learning_rate": 2.7366500596449373e-05,
      "loss": 0.7191,
      "step": 8035
    },
    {
      "epoch": 2.5374171139880013,
      "grad_norm": 2.2054097652435303,
      "learning_rate": 2.736299207073188e-05,
      "loss": 0.7066,
      "step": 8036
    },
    {
      "epoch": 2.537732870224187,
      "grad_norm": 2.3818764686584473,
      "learning_rate": 2.7359483545014386e-05,
      "loss": 0.768,
      "step": 8037
    },
    {
      "epoch": 2.5380486264603723,
      "grad_norm": 2.377128839492798,
      "learning_rate": 2.7355975019296893e-05,
      "loss": 0.7404,
      "step": 8038
    },
    {
      "epoch": 2.5383643826965585,
      "grad_norm": 2.6771066188812256,
      "learning_rate": 2.73524664935794e-05,
      "loss": 0.7318,
      "step": 8039
    },
    {
      "epoch": 2.538680138932744,
      "grad_norm": 2.513582706451416,
      "learning_rate": 2.7348957967861903e-05,
      "loss": 0.7365,
      "step": 8040
    },
    {
      "epoch": 2.5389958951689295,
      "grad_norm": 2.50449800491333,
      "learning_rate": 2.734544944214441e-05,
      "loss": 0.713,
      "step": 8041
    },
    {
      "epoch": 2.5393116514051153,
      "grad_norm": 1.991033673286438,
      "learning_rate": 2.7341940916426917e-05,
      "loss": 0.6922,
      "step": 8042
    },
    {
      "epoch": 2.539627407641301,
      "grad_norm": 2.1305811405181885,
      "learning_rate": 2.7338432390709424e-05,
      "loss": 0.6573,
      "step": 8043
    },
    {
      "epoch": 2.5399431638774868,
      "grad_norm": 2.634768009185791,
      "learning_rate": 2.733492386499193e-05,
      "loss": 0.6816,
      "step": 8044
    },
    {
      "epoch": 2.540258920113672,
      "grad_norm": 2.6790571212768555,
      "learning_rate": 2.7331415339274434e-05,
      "loss": 0.8605,
      "step": 8045
    },
    {
      "epoch": 2.540574676349858,
      "grad_norm": 2.3116037845611572,
      "learning_rate": 2.7327906813556948e-05,
      "loss": 0.7381,
      "step": 8046
    },
    {
      "epoch": 2.5408904325860435,
      "grad_norm": 2.4327709674835205,
      "learning_rate": 2.7324398287839454e-05,
      "loss": 0.7618,
      "step": 8047
    },
    {
      "epoch": 2.5412061888222293,
      "grad_norm": 2.4486615657806396,
      "learning_rate": 2.7320889762121958e-05,
      "loss": 0.7465,
      "step": 8048
    },
    {
      "epoch": 2.541521945058415,
      "grad_norm": 2.4791343212127686,
      "learning_rate": 2.7317381236404465e-05,
      "loss": 0.7173,
      "step": 8049
    },
    {
      "epoch": 2.5418377012946003,
      "grad_norm": 2.4395029544830322,
      "learning_rate": 2.731387271068697e-05,
      "loss": 0.7388,
      "step": 8050
    },
    {
      "epoch": 2.5421534575307865,
      "grad_norm": 2.248538017272949,
      "learning_rate": 2.7310364184969478e-05,
      "loss": 0.8169,
      "step": 8051
    },
    {
      "epoch": 2.5424692137669718,
      "grad_norm": 2.8700623512268066,
      "learning_rate": 2.7306855659251985e-05,
      "loss": 0.8426,
      "step": 8052
    },
    {
      "epoch": 2.5427849700031575,
      "grad_norm": 2.3379485607147217,
      "learning_rate": 2.7303347133534492e-05,
      "loss": 0.768,
      "step": 8053
    },
    {
      "epoch": 2.5431007262393432,
      "grad_norm": 2.954026222229004,
      "learning_rate": 2.7299838607816995e-05,
      "loss": 0.7969,
      "step": 8054
    },
    {
      "epoch": 2.543416482475529,
      "grad_norm": 2.3549723625183105,
      "learning_rate": 2.7296330082099502e-05,
      "loss": 0.7495,
      "step": 8055
    },
    {
      "epoch": 2.5437322387117147,
      "grad_norm": 2.43685245513916,
      "learning_rate": 2.729282155638201e-05,
      "loss": 0.7064,
      "step": 8056
    },
    {
      "epoch": 2.5440479949479,
      "grad_norm": 2.183501720428467,
      "learning_rate": 2.7289313030664516e-05,
      "loss": 0.6781,
      "step": 8057
    },
    {
      "epoch": 2.5443637511840858,
      "grad_norm": 2.3883416652679443,
      "learning_rate": 2.7285804504947023e-05,
      "loss": 0.7392,
      "step": 8058
    },
    {
      "epoch": 2.5446795074202715,
      "grad_norm": 2.3664956092834473,
      "learning_rate": 2.7282295979229526e-05,
      "loss": 0.6515,
      "step": 8059
    },
    {
      "epoch": 2.5449952636564572,
      "grad_norm": 2.2533278465270996,
      "learning_rate": 2.7278787453512033e-05,
      "loss": 0.7124,
      "step": 8060
    },
    {
      "epoch": 2.545311019892643,
      "grad_norm": 2.2082018852233887,
      "learning_rate": 2.727527892779454e-05,
      "loss": 0.6772,
      "step": 8061
    },
    {
      "epoch": 2.5456267761288287,
      "grad_norm": 2.277820587158203,
      "learning_rate": 2.7271770402077046e-05,
      "loss": 0.7664,
      "step": 8062
    },
    {
      "epoch": 2.5459425323650144,
      "grad_norm": 2.3648135662078857,
      "learning_rate": 2.7268261876359557e-05,
      "loss": 0.7993,
      "step": 8063
    },
    {
      "epoch": 2.5462582886011997,
      "grad_norm": 2.282866954803467,
      "learning_rate": 2.7264753350642063e-05,
      "loss": 0.7014,
      "step": 8064
    },
    {
      "epoch": 2.5465740448373855,
      "grad_norm": 2.2769625186920166,
      "learning_rate": 2.726124482492457e-05,
      "loss": 0.7251,
      "step": 8065
    },
    {
      "epoch": 2.546889801073571,
      "grad_norm": 2.6304590702056885,
      "learning_rate": 2.7257736299207077e-05,
      "loss": 0.7475,
      "step": 8066
    },
    {
      "epoch": 2.547205557309757,
      "grad_norm": 2.311046600341797,
      "learning_rate": 2.7254227773489584e-05,
      "loss": 0.7449,
      "step": 8067
    },
    {
      "epoch": 2.5475213135459427,
      "grad_norm": 2.27382755279541,
      "learning_rate": 2.7250719247772087e-05,
      "loss": 0.7942,
      "step": 8068
    },
    {
      "epoch": 2.547837069782128,
      "grad_norm": 2.755678415298462,
      "learning_rate": 2.7247210722054594e-05,
      "loss": 0.7228,
      "step": 8069
    },
    {
      "epoch": 2.5481528260183137,
      "grad_norm": 2.5086984634399414,
      "learning_rate": 2.72437021963371e-05,
      "loss": 0.6503,
      "step": 8070
    },
    {
      "epoch": 2.5484685822544995,
      "grad_norm": 2.6843223571777344,
      "learning_rate": 2.7240193670619608e-05,
      "loss": 0.8194,
      "step": 8071
    },
    {
      "epoch": 2.548784338490685,
      "grad_norm": 2.315828561782837,
      "learning_rate": 2.7236685144902115e-05,
      "loss": 0.6941,
      "step": 8072
    },
    {
      "epoch": 2.549100094726871,
      "grad_norm": 2.4072844982147217,
      "learning_rate": 2.7233176619184618e-05,
      "loss": 0.7178,
      "step": 8073
    },
    {
      "epoch": 2.5494158509630567,
      "grad_norm": 2.2574756145477295,
      "learning_rate": 2.7229668093467125e-05,
      "loss": 0.7596,
      "step": 8074
    },
    {
      "epoch": 2.5497316071992424,
      "grad_norm": 2.562954902648926,
      "learning_rate": 2.722615956774963e-05,
      "loss": 0.8178,
      "step": 8075
    },
    {
      "epoch": 2.5500473634354277,
      "grad_norm": 2.5442726612091064,
      "learning_rate": 2.722265104203214e-05,
      "loss": 0.7246,
      "step": 8076
    },
    {
      "epoch": 2.5503631196716134,
      "grad_norm": 2.4919166564941406,
      "learning_rate": 2.7219142516314645e-05,
      "loss": 0.7461,
      "step": 8077
    },
    {
      "epoch": 2.550678875907799,
      "grad_norm": 2.3777482509613037,
      "learning_rate": 2.721563399059715e-05,
      "loss": 0.6872,
      "step": 8078
    },
    {
      "epoch": 2.550994632143985,
      "grad_norm": 2.1540489196777344,
      "learning_rate": 2.7212125464879656e-05,
      "loss": 0.6874,
      "step": 8079
    },
    {
      "epoch": 2.5513103883801707,
      "grad_norm": 2.152291774749756,
      "learning_rate": 2.7208616939162162e-05,
      "loss": 0.8421,
      "step": 8080
    },
    {
      "epoch": 2.551626144616356,
      "grad_norm": 2.9231975078582764,
      "learning_rate": 2.7205108413444676e-05,
      "loss": 0.7856,
      "step": 8081
    },
    {
      "epoch": 2.5519419008525417,
      "grad_norm": 2.215616226196289,
      "learning_rate": 2.720159988772718e-05,
      "loss": 0.7859,
      "step": 8082
    },
    {
      "epoch": 2.5522576570887274,
      "grad_norm": 2.259478807449341,
      "learning_rate": 2.7198091362009686e-05,
      "loss": 0.6728,
      "step": 8083
    },
    {
      "epoch": 2.552573413324913,
      "grad_norm": 2.41896653175354,
      "learning_rate": 2.7194582836292193e-05,
      "loss": 0.7585,
      "step": 8084
    },
    {
      "epoch": 2.552889169561099,
      "grad_norm": 2.784658670425415,
      "learning_rate": 2.71910743105747e-05,
      "loss": 0.7367,
      "step": 8085
    },
    {
      "epoch": 2.5532049257972846,
      "grad_norm": 2.3183276653289795,
      "learning_rate": 2.7187565784857207e-05,
      "loss": 0.7991,
      "step": 8086
    },
    {
      "epoch": 2.5535206820334704,
      "grad_norm": 2.575730800628662,
      "learning_rate": 2.718405725913971e-05,
      "loss": 0.7216,
      "step": 8087
    },
    {
      "epoch": 2.5538364382696557,
      "grad_norm": 2.2142953872680664,
      "learning_rate": 2.7180548733422217e-05,
      "loss": 0.7659,
      "step": 8088
    },
    {
      "epoch": 2.5541521945058414,
      "grad_norm": 2.3715288639068604,
      "learning_rate": 2.7177040207704724e-05,
      "loss": 0.7666,
      "step": 8089
    },
    {
      "epoch": 2.554467950742027,
      "grad_norm": 2.1916825771331787,
      "learning_rate": 2.717353168198723e-05,
      "loss": 0.7286,
      "step": 8090
    },
    {
      "epoch": 2.554783706978213,
      "grad_norm": 2.117384433746338,
      "learning_rate": 2.7170023156269737e-05,
      "loss": 0.7556,
      "step": 8091
    },
    {
      "epoch": 2.5550994632143986,
      "grad_norm": 2.4810941219329834,
      "learning_rate": 2.716651463055224e-05,
      "loss": 0.776,
      "step": 8092
    },
    {
      "epoch": 2.555415219450584,
      "grad_norm": 2.612602472305298,
      "learning_rate": 2.7163006104834748e-05,
      "loss": 0.8033,
      "step": 8093
    },
    {
      "epoch": 2.5557309756867697,
      "grad_norm": 2.4239864349365234,
      "learning_rate": 2.7159497579117254e-05,
      "loss": 0.8114,
      "step": 8094
    },
    {
      "epoch": 2.5560467319229554,
      "grad_norm": 2.24149489402771,
      "learning_rate": 2.715598905339976e-05,
      "loss": 0.737,
      "step": 8095
    },
    {
      "epoch": 2.556362488159141,
      "grad_norm": 2.0633201599121094,
      "learning_rate": 2.7152480527682268e-05,
      "loss": 0.8116,
      "step": 8096
    },
    {
      "epoch": 2.556678244395327,
      "grad_norm": 2.361629009246826,
      "learning_rate": 2.7148972001964775e-05,
      "loss": 0.8739,
      "step": 8097
    },
    {
      "epoch": 2.5569940006315126,
      "grad_norm": 2.35791015625,
      "learning_rate": 2.7145463476247285e-05,
      "loss": 0.7005,
      "step": 8098
    },
    {
      "epoch": 2.5573097568676983,
      "grad_norm": 2.4060778617858887,
      "learning_rate": 2.7141954950529792e-05,
      "loss": 0.7222,
      "step": 8099
    },
    {
      "epoch": 2.5576255131038836,
      "grad_norm": 2.621516704559326,
      "learning_rate": 2.71384464248123e-05,
      "loss": 0.7069,
      "step": 8100
    },
    {
      "epoch": 2.5579412693400694,
      "grad_norm": 2.940481424331665,
      "learning_rate": 2.7134937899094802e-05,
      "loss": 0.7863,
      "step": 8101
    },
    {
      "epoch": 2.558257025576255,
      "grad_norm": 2.309562921524048,
      "learning_rate": 2.713142937337731e-05,
      "loss": 0.716,
      "step": 8102
    },
    {
      "epoch": 2.558572781812441,
      "grad_norm": 2.643324613571167,
      "learning_rate": 2.7127920847659816e-05,
      "loss": 0.78,
      "step": 8103
    },
    {
      "epoch": 2.5588885380486266,
      "grad_norm": 2.3618674278259277,
      "learning_rate": 2.7124412321942322e-05,
      "loss": 0.7202,
      "step": 8104
    },
    {
      "epoch": 2.559204294284812,
      "grad_norm": 2.487049102783203,
      "learning_rate": 2.712090379622483e-05,
      "loss": 0.8111,
      "step": 8105
    },
    {
      "epoch": 2.5595200505209976,
      "grad_norm": 2.3368451595306396,
      "learning_rate": 2.7117395270507333e-05,
      "loss": 0.7266,
      "step": 8106
    },
    {
      "epoch": 2.5598358067571834,
      "grad_norm": 2.6494929790496826,
      "learning_rate": 2.711388674478984e-05,
      "loss": 0.7076,
      "step": 8107
    },
    {
      "epoch": 2.560151562993369,
      "grad_norm": 2.091196298599243,
      "learning_rate": 2.7110378219072346e-05,
      "loss": 0.6694,
      "step": 8108
    },
    {
      "epoch": 2.560467319229555,
      "grad_norm": 2.1751017570495605,
      "learning_rate": 2.7106869693354853e-05,
      "loss": 0.6701,
      "step": 8109
    },
    {
      "epoch": 2.5607830754657406,
      "grad_norm": 2.4591405391693115,
      "learning_rate": 2.710336116763736e-05,
      "loss": 0.7729,
      "step": 8110
    },
    {
      "epoch": 2.5610988317019263,
      "grad_norm": 2.3246991634368896,
      "learning_rate": 2.7099852641919867e-05,
      "loss": 0.7404,
      "step": 8111
    },
    {
      "epoch": 2.5614145879381116,
      "grad_norm": 2.4574527740478516,
      "learning_rate": 2.709634411620237e-05,
      "loss": 0.828,
      "step": 8112
    },
    {
      "epoch": 2.5617303441742973,
      "grad_norm": 2.379636287689209,
      "learning_rate": 2.7092835590484877e-05,
      "loss": 0.6884,
      "step": 8113
    },
    {
      "epoch": 2.562046100410483,
      "grad_norm": 2.6298599243164062,
      "learning_rate": 2.7089327064767384e-05,
      "loss": 0.7359,
      "step": 8114
    },
    {
      "epoch": 2.562361856646669,
      "grad_norm": 2.4505839347839355,
      "learning_rate": 2.7085818539049894e-05,
      "loss": 0.6868,
      "step": 8115
    },
    {
      "epoch": 2.5626776128828546,
      "grad_norm": 2.8748483657836914,
      "learning_rate": 2.70823100133324e-05,
      "loss": 0.7029,
      "step": 8116
    },
    {
      "epoch": 2.56299336911904,
      "grad_norm": 2.098757266998291,
      "learning_rate": 2.7078801487614908e-05,
      "loss": 0.7221,
      "step": 8117
    },
    {
      "epoch": 2.563309125355226,
      "grad_norm": 2.418691873550415,
      "learning_rate": 2.7075292961897414e-05,
      "loss": 0.6826,
      "step": 8118
    },
    {
      "epoch": 2.5636248815914113,
      "grad_norm": 2.689342737197876,
      "learning_rate": 2.707178443617992e-05,
      "loss": 0.836,
      "step": 8119
    },
    {
      "epoch": 2.563940637827597,
      "grad_norm": 2.1677663326263428,
      "learning_rate": 2.7068275910462425e-05,
      "loss": 0.6478,
      "step": 8120
    },
    {
      "epoch": 2.564256394063783,
      "grad_norm": 2.685389757156372,
      "learning_rate": 2.706476738474493e-05,
      "loss": 0.6952,
      "step": 8121
    },
    {
      "epoch": 2.5645721502999685,
      "grad_norm": 2.587181806564331,
      "learning_rate": 2.7061258859027438e-05,
      "loss": 0.816,
      "step": 8122
    },
    {
      "epoch": 2.5648879065361543,
      "grad_norm": 2.2215113639831543,
      "learning_rate": 2.7057750333309945e-05,
      "loss": 0.7321,
      "step": 8123
    },
    {
      "epoch": 2.5652036627723396,
      "grad_norm": 2.8568642139434814,
      "learning_rate": 2.7054241807592452e-05,
      "loss": 0.7595,
      "step": 8124
    },
    {
      "epoch": 2.5655194190085253,
      "grad_norm": 2.5096592903137207,
      "learning_rate": 2.7050733281874955e-05,
      "loss": 0.66,
      "step": 8125
    },
    {
      "epoch": 2.565835175244711,
      "grad_norm": 2.3731789588928223,
      "learning_rate": 2.7047224756157462e-05,
      "loss": 0.6986,
      "step": 8126
    },
    {
      "epoch": 2.566150931480897,
      "grad_norm": 2.430569887161255,
      "learning_rate": 2.704371623043997e-05,
      "loss": 0.7575,
      "step": 8127
    },
    {
      "epoch": 2.5664666877170825,
      "grad_norm": 2.630488157272339,
      "learning_rate": 2.7040207704722476e-05,
      "loss": 0.7919,
      "step": 8128
    },
    {
      "epoch": 2.566782443953268,
      "grad_norm": 2.322277545928955,
      "learning_rate": 2.7036699179004983e-05,
      "loss": 0.708,
      "step": 8129
    },
    {
      "epoch": 2.567098200189454,
      "grad_norm": 2.6902620792388916,
      "learning_rate": 2.703319065328749e-05,
      "loss": 0.8132,
      "step": 8130
    },
    {
      "epoch": 2.5674139564256393,
      "grad_norm": 2.8716793060302734,
      "learning_rate": 2.7029682127569993e-05,
      "loss": 0.7095,
      "step": 8131
    },
    {
      "epoch": 2.567729712661825,
      "grad_norm": 2.171515464782715,
      "learning_rate": 2.70261736018525e-05,
      "loss": 0.6776,
      "step": 8132
    },
    {
      "epoch": 2.5680454688980108,
      "grad_norm": 2.7370657920837402,
      "learning_rate": 2.7022665076135013e-05,
      "loss": 0.746,
      "step": 8133
    },
    {
      "epoch": 2.5683612251341965,
      "grad_norm": 2.9791412353515625,
      "learning_rate": 2.7019156550417517e-05,
      "loss": 0.7526,
      "step": 8134
    },
    {
      "epoch": 2.5686769813703823,
      "grad_norm": 2.22898530960083,
      "learning_rate": 2.7015648024700023e-05,
      "loss": 0.7032,
      "step": 8135
    },
    {
      "epoch": 2.5689927376065675,
      "grad_norm": 2.283402442932129,
      "learning_rate": 2.701213949898253e-05,
      "loss": 0.7063,
      "step": 8136
    },
    {
      "epoch": 2.5693084938427533,
      "grad_norm": 2.2978105545043945,
      "learning_rate": 2.7008630973265037e-05,
      "loss": 0.6942,
      "step": 8137
    },
    {
      "epoch": 2.569624250078939,
      "grad_norm": 2.588351249694824,
      "learning_rate": 2.7005122447547544e-05,
      "loss": 0.783,
      "step": 8138
    },
    {
      "epoch": 2.5699400063151248,
      "grad_norm": 2.6444263458251953,
      "learning_rate": 2.7001613921830047e-05,
      "loss": 0.7904,
      "step": 8139
    },
    {
      "epoch": 2.5702557625513105,
      "grad_norm": 2.4859888553619385,
      "learning_rate": 2.6998105396112554e-05,
      "loss": 0.6802,
      "step": 8140
    },
    {
      "epoch": 2.570571518787496,
      "grad_norm": 2.6018075942993164,
      "learning_rate": 2.699459687039506e-05,
      "loss": 0.6668,
      "step": 8141
    },
    {
      "epoch": 2.570887275023682,
      "grad_norm": 2.1325674057006836,
      "learning_rate": 2.6991088344677568e-05,
      "loss": 0.7041,
      "step": 8142
    },
    {
      "epoch": 2.5712030312598673,
      "grad_norm": 2.4690868854522705,
      "learning_rate": 2.6987579818960075e-05,
      "loss": 0.7246,
      "step": 8143
    },
    {
      "epoch": 2.571518787496053,
      "grad_norm": 2.55999755859375,
      "learning_rate": 2.698407129324258e-05,
      "loss": 0.7332,
      "step": 8144
    },
    {
      "epoch": 2.5718345437322387,
      "grad_norm": 2.618525266647339,
      "learning_rate": 2.6980562767525085e-05,
      "loss": 0.8255,
      "step": 8145
    },
    {
      "epoch": 2.5721502999684245,
      "grad_norm": 2.242377519607544,
      "learning_rate": 2.697705424180759e-05,
      "loss": 0.7684,
      "step": 8146
    },
    {
      "epoch": 2.57246605620461,
      "grad_norm": 2.7158102989196777,
      "learning_rate": 2.69735457160901e-05,
      "loss": 0.7301,
      "step": 8147
    },
    {
      "epoch": 2.5727818124407955,
      "grad_norm": 2.944687604904175,
      "learning_rate": 2.6970037190372605e-05,
      "loss": 0.7122,
      "step": 8148
    },
    {
      "epoch": 2.5730975686769813,
      "grad_norm": 2.3071846961975098,
      "learning_rate": 2.6966528664655112e-05,
      "loss": 0.6396,
      "step": 8149
    },
    {
      "epoch": 2.573413324913167,
      "grad_norm": 2.2843475341796875,
      "learning_rate": 2.6963020138937622e-05,
      "loss": 0.7763,
      "step": 8150
    },
    {
      "epoch": 2.5737290811493527,
      "grad_norm": 2.650712013244629,
      "learning_rate": 2.695951161322013e-05,
      "loss": 0.7989,
      "step": 8151
    },
    {
      "epoch": 2.5740448373855385,
      "grad_norm": 2.741014003753662,
      "learning_rate": 2.6956003087502636e-05,
      "loss": 0.8421,
      "step": 8152
    },
    {
      "epoch": 2.5743605936217238,
      "grad_norm": 2.8536412715911865,
      "learning_rate": 2.695249456178514e-05,
      "loss": 0.729,
      "step": 8153
    },
    {
      "epoch": 2.57467634985791,
      "grad_norm": 2.452180862426758,
      "learning_rate": 2.6948986036067646e-05,
      "loss": 0.758,
      "step": 8154
    },
    {
      "epoch": 2.5749921060940952,
      "grad_norm": 2.2709877490997314,
      "learning_rate": 2.6945477510350153e-05,
      "loss": 0.7018,
      "step": 8155
    },
    {
      "epoch": 2.575307862330281,
      "grad_norm": 2.237661600112915,
      "learning_rate": 2.694196898463266e-05,
      "loss": 0.7251,
      "step": 8156
    },
    {
      "epoch": 2.5756236185664667,
      "grad_norm": 2.3241055011749268,
      "learning_rate": 2.6938460458915167e-05,
      "loss": 0.7483,
      "step": 8157
    },
    {
      "epoch": 2.5759393748026524,
      "grad_norm": 2.4286935329437256,
      "learning_rate": 2.6934951933197673e-05,
      "loss": 0.8137,
      "step": 8158
    },
    {
      "epoch": 2.576255131038838,
      "grad_norm": 2.1110410690307617,
      "learning_rate": 2.6931443407480177e-05,
      "loss": 0.6615,
      "step": 8159
    },
    {
      "epoch": 2.5765708872750235,
      "grad_norm": 2.1581385135650635,
      "learning_rate": 2.6927934881762684e-05,
      "loss": 0.7195,
      "step": 8160
    },
    {
      "epoch": 2.576886643511209,
      "grad_norm": 2.2462921142578125,
      "learning_rate": 2.692442635604519e-05,
      "loss": 0.672,
      "step": 8161
    },
    {
      "epoch": 2.577202399747395,
      "grad_norm": 2.0870392322540283,
      "learning_rate": 2.6920917830327697e-05,
      "loss": 0.7728,
      "step": 8162
    },
    {
      "epoch": 2.5775181559835807,
      "grad_norm": 2.2996981143951416,
      "learning_rate": 2.6917409304610204e-05,
      "loss": 0.6393,
      "step": 8163
    },
    {
      "epoch": 2.5778339122197664,
      "grad_norm": 2.5098531246185303,
      "learning_rate": 2.6913900778892708e-05,
      "loss": 0.8471,
      "step": 8164
    },
    {
      "epoch": 2.578149668455952,
      "grad_norm": 2.608076333999634,
      "learning_rate": 2.6910392253175214e-05,
      "loss": 0.7829,
      "step": 8165
    },
    {
      "epoch": 2.578465424692138,
      "grad_norm": 2.406715154647827,
      "learning_rate": 2.690688372745772e-05,
      "loss": 0.741,
      "step": 8166
    },
    {
      "epoch": 2.578781180928323,
      "grad_norm": 2.4719409942626953,
      "learning_rate": 2.6903375201740228e-05,
      "loss": 0.7831,
      "step": 8167
    },
    {
      "epoch": 2.579096937164509,
      "grad_norm": 2.5428037643432617,
      "learning_rate": 2.6899866676022738e-05,
      "loss": 0.7301,
      "step": 8168
    },
    {
      "epoch": 2.5794126934006947,
      "grad_norm": 2.9547014236450195,
      "learning_rate": 2.6896358150305245e-05,
      "loss": 0.7921,
      "step": 8169
    },
    {
      "epoch": 2.5797284496368804,
      "grad_norm": 2.2144856452941895,
      "learning_rate": 2.6892849624587752e-05,
      "loss": 0.7432,
      "step": 8170
    },
    {
      "epoch": 2.580044205873066,
      "grad_norm": 2.6687722206115723,
      "learning_rate": 2.688934109887026e-05,
      "loss": 0.7177,
      "step": 8171
    },
    {
      "epoch": 2.5803599621092514,
      "grad_norm": 2.669297695159912,
      "learning_rate": 2.6885832573152765e-05,
      "loss": 0.7916,
      "step": 8172
    },
    {
      "epoch": 2.580675718345437,
      "grad_norm": 2.1567306518554688,
      "learning_rate": 2.688232404743527e-05,
      "loss": 0.7077,
      "step": 8173
    },
    {
      "epoch": 2.580991474581623,
      "grad_norm": 2.828294515609741,
      "learning_rate": 2.6878815521717776e-05,
      "loss": 0.7432,
      "step": 8174
    },
    {
      "epoch": 2.5813072308178087,
      "grad_norm": 2.467162847518921,
      "learning_rate": 2.6875306996000282e-05,
      "loss": 0.7321,
      "step": 8175
    },
    {
      "epoch": 2.5816229870539944,
      "grad_norm": 2.6355748176574707,
      "learning_rate": 2.687179847028279e-05,
      "loss": 0.8168,
      "step": 8176
    },
    {
      "epoch": 2.58193874329018,
      "grad_norm": 1.9113215208053589,
      "learning_rate": 2.6868289944565296e-05,
      "loss": 0.7033,
      "step": 8177
    },
    {
      "epoch": 2.582254499526366,
      "grad_norm": 2.040447235107422,
      "learning_rate": 2.68647814188478e-05,
      "loss": 0.6952,
      "step": 8178
    },
    {
      "epoch": 2.582570255762551,
      "grad_norm": 2.573671340942383,
      "learning_rate": 2.6861272893130306e-05,
      "loss": 0.7889,
      "step": 8179
    },
    {
      "epoch": 2.582886011998737,
      "grad_norm": 2.3730015754699707,
      "learning_rate": 2.6857764367412813e-05,
      "loss": 0.7843,
      "step": 8180
    },
    {
      "epoch": 2.5832017682349226,
      "grad_norm": 2.5274012088775635,
      "learning_rate": 2.685425584169532e-05,
      "loss": 0.7605,
      "step": 8181
    },
    {
      "epoch": 2.5835175244711084,
      "grad_norm": 2.110501527786255,
      "learning_rate": 2.6850747315977827e-05,
      "loss": 0.7406,
      "step": 8182
    },
    {
      "epoch": 2.583833280707294,
      "grad_norm": 2.537543296813965,
      "learning_rate": 2.684723879026033e-05,
      "loss": 0.7738,
      "step": 8183
    },
    {
      "epoch": 2.5841490369434794,
      "grad_norm": 2.3905465602874756,
      "learning_rate": 2.6843730264542837e-05,
      "loss": 0.6973,
      "step": 8184
    },
    {
      "epoch": 2.584464793179665,
      "grad_norm": 2.047954559326172,
      "learning_rate": 2.684022173882535e-05,
      "loss": 0.7668,
      "step": 8185
    },
    {
      "epoch": 2.584780549415851,
      "grad_norm": 2.1743006706237793,
      "learning_rate": 2.6836713213107857e-05,
      "loss": 0.6435,
      "step": 8186
    },
    {
      "epoch": 2.5850963056520366,
      "grad_norm": 2.3153512477874756,
      "learning_rate": 2.683320468739036e-05,
      "loss": 0.7347,
      "step": 8187
    },
    {
      "epoch": 2.5854120618882224,
      "grad_norm": 2.283766508102417,
      "learning_rate": 2.6829696161672868e-05,
      "loss": 0.7923,
      "step": 8188
    },
    {
      "epoch": 2.585727818124408,
      "grad_norm": 2.429183006286621,
      "learning_rate": 2.6826187635955374e-05,
      "loss": 0.7014,
      "step": 8189
    },
    {
      "epoch": 2.586043574360594,
      "grad_norm": 2.498152256011963,
      "learning_rate": 2.682267911023788e-05,
      "loss": 0.8436,
      "step": 8190
    },
    {
      "epoch": 2.586359330596779,
      "grad_norm": 2.2224090099334717,
      "learning_rate": 2.6819170584520388e-05,
      "loss": 0.7429,
      "step": 8191
    },
    {
      "epoch": 2.586675086832965,
      "grad_norm": 2.1991820335388184,
      "learning_rate": 2.681566205880289e-05,
      "loss": 0.7586,
      "step": 8192
    },
    {
      "epoch": 2.5869908430691506,
      "grad_norm": 2.377720594406128,
      "learning_rate": 2.68121535330854e-05,
      "loss": 0.732,
      "step": 8193
    },
    {
      "epoch": 2.5873065993053364,
      "grad_norm": 2.6544625759124756,
      "learning_rate": 2.6808645007367905e-05,
      "loss": 0.7678,
      "step": 8194
    },
    {
      "epoch": 2.587622355541522,
      "grad_norm": 2.4929847717285156,
      "learning_rate": 2.6805136481650412e-05,
      "loss": 0.7959,
      "step": 8195
    },
    {
      "epoch": 2.5879381117777074,
      "grad_norm": 2.3014907836914062,
      "learning_rate": 2.680162795593292e-05,
      "loss": 0.7607,
      "step": 8196
    },
    {
      "epoch": 2.588253868013893,
      "grad_norm": 2.5811197757720947,
      "learning_rate": 2.6798119430215422e-05,
      "loss": 0.7391,
      "step": 8197
    },
    {
      "epoch": 2.588569624250079,
      "grad_norm": 2.305361032485962,
      "learning_rate": 2.679461090449793e-05,
      "loss": 0.7593,
      "step": 8198
    },
    {
      "epoch": 2.5888853804862646,
      "grad_norm": 2.52787184715271,
      "learning_rate": 2.6791102378780436e-05,
      "loss": 0.8036,
      "step": 8199
    },
    {
      "epoch": 2.5892011367224503,
      "grad_norm": 2.107888698577881,
      "learning_rate": 2.6787593853062943e-05,
      "loss": 0.7198,
      "step": 8200
    },
    {
      "epoch": 2.5892011367224503,
      "eval_loss": 0.7975435853004456,
      "eval_runtime": 4.6731,
      "eval_samples_per_second": 532.834,
      "eval_steps_per_second": 4.494,
      "step": 8200
    },
    {
      "epoch": 2.589516892958636,
      "grad_norm": 2.498654842376709,
      "learning_rate": 2.678408532734545e-05,
      "loss": 0.6922,
      "step": 8201
    },
    {
      "epoch": 2.589832649194822,
      "grad_norm": 2.2758748531341553,
      "learning_rate": 2.6780576801627956e-05,
      "loss": 0.7323,
      "step": 8202
    },
    {
      "epoch": 2.590148405431007,
      "grad_norm": 2.2553553581237793,
      "learning_rate": 2.6777068275910466e-05,
      "loss": 0.6909,
      "step": 8203
    },
    {
      "epoch": 2.590464161667193,
      "grad_norm": 2.5407135486602783,
      "learning_rate": 2.6773559750192973e-05,
      "loss": 0.6318,
      "step": 8204
    },
    {
      "epoch": 2.5907799179033786,
      "grad_norm": 2.4508113861083984,
      "learning_rate": 2.677005122447548e-05,
      "loss": 0.7337,
      "step": 8205
    },
    {
      "epoch": 2.5910956741395643,
      "grad_norm": 2.324509859085083,
      "learning_rate": 2.6766542698757983e-05,
      "loss": 0.671,
      "step": 8206
    },
    {
      "epoch": 2.59141143037575,
      "grad_norm": 2.385906219482422,
      "learning_rate": 2.676303417304049e-05,
      "loss": 0.7882,
      "step": 8207
    },
    {
      "epoch": 2.5917271866119354,
      "grad_norm": 2.3472607135772705,
      "learning_rate": 2.6759525647322997e-05,
      "loss": 0.7522,
      "step": 8208
    },
    {
      "epoch": 2.592042942848121,
      "grad_norm": 2.6224818229675293,
      "learning_rate": 2.6756017121605504e-05,
      "loss": 0.7742,
      "step": 8209
    },
    {
      "epoch": 2.592358699084307,
      "grad_norm": 2.4453258514404297,
      "learning_rate": 2.675250859588801e-05,
      "loss": 0.7764,
      "step": 8210
    },
    {
      "epoch": 2.5926744553204926,
      "grad_norm": 2.6649010181427,
      "learning_rate": 2.6749000070170514e-05,
      "loss": 0.8117,
      "step": 8211
    },
    {
      "epoch": 2.5929902115566783,
      "grad_norm": 2.4305777549743652,
      "learning_rate": 2.674549154445302e-05,
      "loss": 0.7045,
      "step": 8212
    },
    {
      "epoch": 2.593305967792864,
      "grad_norm": 2.6591272354125977,
      "learning_rate": 2.6741983018735528e-05,
      "loss": 0.7426,
      "step": 8213
    },
    {
      "epoch": 2.59362172402905,
      "grad_norm": 2.3387465476989746,
      "learning_rate": 2.6738474493018035e-05,
      "loss": 0.7345,
      "step": 8214
    },
    {
      "epoch": 2.593937480265235,
      "grad_norm": 2.346031904220581,
      "learning_rate": 2.673496596730054e-05,
      "loss": 0.7152,
      "step": 8215
    },
    {
      "epoch": 2.594253236501421,
      "grad_norm": 2.3768062591552734,
      "learning_rate": 2.6731457441583048e-05,
      "loss": 0.7393,
      "step": 8216
    },
    {
      "epoch": 2.5945689927376065,
      "grad_norm": 2.7391059398651123,
      "learning_rate": 2.672794891586555e-05,
      "loss": 0.7053,
      "step": 8217
    },
    {
      "epoch": 2.5948847489737923,
      "grad_norm": 2.3322291374206543,
      "learning_rate": 2.672444039014806e-05,
      "loss": 0.6921,
      "step": 8218
    },
    {
      "epoch": 2.595200505209978,
      "grad_norm": 2.556844711303711,
      "learning_rate": 2.6720931864430565e-05,
      "loss": 0.748,
      "step": 8219
    },
    {
      "epoch": 2.5955162614461633,
      "grad_norm": 2.4072134494781494,
      "learning_rate": 2.6717423338713075e-05,
      "loss": 0.7225,
      "step": 8220
    },
    {
      "epoch": 2.5958320176823495,
      "grad_norm": 2.9948530197143555,
      "learning_rate": 2.6713914812995582e-05,
      "loss": 0.8733,
      "step": 8221
    },
    {
      "epoch": 2.596147773918535,
      "grad_norm": 2.356997489929199,
      "learning_rate": 2.671040628727809e-05,
      "loss": 0.6415,
      "step": 8222
    },
    {
      "epoch": 2.5964635301547205,
      "grad_norm": 2.221707344055176,
      "learning_rate": 2.6706897761560596e-05,
      "loss": 0.6619,
      "step": 8223
    },
    {
      "epoch": 2.5967792863909063,
      "grad_norm": 2.3513686656951904,
      "learning_rate": 2.6703389235843103e-05,
      "loss": 0.6991,
      "step": 8224
    },
    {
      "epoch": 2.597095042627092,
      "grad_norm": 2.6749329566955566,
      "learning_rate": 2.6699880710125606e-05,
      "loss": 0.8266,
      "step": 8225
    },
    {
      "epoch": 2.5974107988632777,
      "grad_norm": 2.5447113513946533,
      "learning_rate": 2.6696372184408113e-05,
      "loss": 0.7682,
      "step": 8226
    },
    {
      "epoch": 2.597726555099463,
      "grad_norm": 2.2701568603515625,
      "learning_rate": 2.669286365869062e-05,
      "loss": 0.7138,
      "step": 8227
    },
    {
      "epoch": 2.598042311335649,
      "grad_norm": 2.1883180141448975,
      "learning_rate": 2.6689355132973127e-05,
      "loss": 0.6304,
      "step": 8228
    },
    {
      "epoch": 2.5983580675718345,
      "grad_norm": 2.641484022140503,
      "learning_rate": 2.6685846607255633e-05,
      "loss": 0.6472,
      "step": 8229
    },
    {
      "epoch": 2.5986738238080203,
      "grad_norm": 2.617219924926758,
      "learning_rate": 2.6682338081538137e-05,
      "loss": 0.7127,
      "step": 8230
    },
    {
      "epoch": 2.598989580044206,
      "grad_norm": 2.5746383666992188,
      "learning_rate": 2.6678829555820644e-05,
      "loss": 0.8158,
      "step": 8231
    },
    {
      "epoch": 2.5993053362803913,
      "grad_norm": 2.697220802307129,
      "learning_rate": 2.667532103010315e-05,
      "loss": 0.7655,
      "step": 8232
    },
    {
      "epoch": 2.5996210925165775,
      "grad_norm": 3.0824365615844727,
      "learning_rate": 2.6671812504385657e-05,
      "loss": 0.8273,
      "step": 8233
    },
    {
      "epoch": 2.5999368487527628,
      "grad_norm": 2.3496737480163574,
      "learning_rate": 2.6668303978668164e-05,
      "loss": 0.6922,
      "step": 8234
    },
    {
      "epoch": 2.6002526049889485,
      "grad_norm": 2.335934638977051,
      "learning_rate": 2.666479545295067e-05,
      "loss": 0.7634,
      "step": 8235
    },
    {
      "epoch": 2.6005683612251342,
      "grad_norm": 2.590859889984131,
      "learning_rate": 2.6661286927233174e-05,
      "loss": 0.7587,
      "step": 8236
    },
    {
      "epoch": 2.60088411746132,
      "grad_norm": 3.3288168907165527,
      "learning_rate": 2.665777840151568e-05,
      "loss": 0.8244,
      "step": 8237
    },
    {
      "epoch": 2.6011998736975057,
      "grad_norm": 2.3495476245880127,
      "learning_rate": 2.6654269875798195e-05,
      "loss": 0.698,
      "step": 8238
    },
    {
      "epoch": 2.601515629933691,
      "grad_norm": 2.6537117958068848,
      "learning_rate": 2.6650761350080698e-05,
      "loss": 0.7364,
      "step": 8239
    },
    {
      "epoch": 2.6018313861698767,
      "grad_norm": 2.2571442127227783,
      "learning_rate": 2.6647252824363205e-05,
      "loss": 0.7815,
      "step": 8240
    },
    {
      "epoch": 2.6021471424060625,
      "grad_norm": 2.462670087814331,
      "learning_rate": 2.6643744298645712e-05,
      "loss": 0.6815,
      "step": 8241
    },
    {
      "epoch": 2.6024628986422482,
      "grad_norm": 2.549516439437866,
      "learning_rate": 2.664023577292822e-05,
      "loss": 0.7849,
      "step": 8242
    },
    {
      "epoch": 2.602778654878434,
      "grad_norm": 2.5031890869140625,
      "learning_rate": 2.6636727247210725e-05,
      "loss": 0.7608,
      "step": 8243
    },
    {
      "epoch": 2.6030944111146193,
      "grad_norm": 2.7474565505981445,
      "learning_rate": 2.663321872149323e-05,
      "loss": 0.6728,
      "step": 8244
    },
    {
      "epoch": 2.6034101673508054,
      "grad_norm": 2.601865291595459,
      "learning_rate": 2.6629710195775736e-05,
      "loss": 0.7005,
      "step": 8245
    },
    {
      "epoch": 2.6037259235869907,
      "grad_norm": 2.2216641902923584,
      "learning_rate": 2.6626201670058242e-05,
      "loss": 0.7764,
      "step": 8246
    },
    {
      "epoch": 2.6040416798231765,
      "grad_norm": 2.4890429973602295,
      "learning_rate": 2.662269314434075e-05,
      "loss": 0.7866,
      "step": 8247
    },
    {
      "epoch": 2.604357436059362,
      "grad_norm": 2.2513744831085205,
      "learning_rate": 2.6619184618623256e-05,
      "loss": 0.7651,
      "step": 8248
    },
    {
      "epoch": 2.604673192295548,
      "grad_norm": 2.8180534839630127,
      "learning_rate": 2.6615676092905763e-05,
      "loss": 0.7746,
      "step": 8249
    },
    {
      "epoch": 2.6049889485317337,
      "grad_norm": 2.4446911811828613,
      "learning_rate": 2.6612167567188266e-05,
      "loss": 0.7955,
      "step": 8250
    },
    {
      "epoch": 2.605304704767919,
      "grad_norm": 2.4537837505340576,
      "learning_rate": 2.6608659041470773e-05,
      "loss": 0.6131,
      "step": 8251
    },
    {
      "epoch": 2.6056204610041047,
      "grad_norm": 2.3210275173187256,
      "learning_rate": 2.660515051575328e-05,
      "loss": 0.7067,
      "step": 8252
    },
    {
      "epoch": 2.6059362172402905,
      "grad_norm": 2.1888844966888428,
      "learning_rate": 2.6601641990035787e-05,
      "loss": 0.6601,
      "step": 8253
    },
    {
      "epoch": 2.606251973476476,
      "grad_norm": 2.544037103652954,
      "learning_rate": 2.6598133464318294e-05,
      "loss": 0.722,
      "step": 8254
    },
    {
      "epoch": 2.606567729712662,
      "grad_norm": 2.5781266689300537,
      "learning_rate": 2.6594624938600804e-05,
      "loss": 0.6477,
      "step": 8255
    },
    {
      "epoch": 2.6068834859488472,
      "grad_norm": 2.2768683433532715,
      "learning_rate": 2.659111641288331e-05,
      "loss": 0.7367,
      "step": 8256
    },
    {
      "epoch": 2.6071992421850334,
      "grad_norm": 2.4042880535125732,
      "learning_rate": 2.6587607887165817e-05,
      "loss": 0.7434,
      "step": 8257
    },
    {
      "epoch": 2.6075149984212187,
      "grad_norm": 2.5086917877197266,
      "learning_rate": 2.658409936144832e-05,
      "loss": 0.7688,
      "step": 8258
    },
    {
      "epoch": 2.6078307546574044,
      "grad_norm": 2.6480512619018555,
      "learning_rate": 2.6580590835730828e-05,
      "loss": 0.8554,
      "step": 8259
    },
    {
      "epoch": 2.60814651089359,
      "grad_norm": 2.085913896560669,
      "learning_rate": 2.6577082310013334e-05,
      "loss": 0.6875,
      "step": 8260
    },
    {
      "epoch": 2.608462267129776,
      "grad_norm": 2.415037155151367,
      "learning_rate": 2.657357378429584e-05,
      "loss": 0.7047,
      "step": 8261
    },
    {
      "epoch": 2.6087780233659617,
      "grad_norm": 3.1793675422668457,
      "learning_rate": 2.6570065258578348e-05,
      "loss": 0.8156,
      "step": 8262
    },
    {
      "epoch": 2.609093779602147,
      "grad_norm": 2.1304075717926025,
      "learning_rate": 2.6566556732860855e-05,
      "loss": 0.6684,
      "step": 8263
    },
    {
      "epoch": 2.6094095358383327,
      "grad_norm": 2.6370720863342285,
      "learning_rate": 2.656304820714336e-05,
      "loss": 0.7802,
      "step": 8264
    },
    {
      "epoch": 2.6097252920745184,
      "grad_norm": 2.6667051315307617,
      "learning_rate": 2.6559539681425865e-05,
      "loss": 0.7051,
      "step": 8265
    },
    {
      "epoch": 2.610041048310704,
      "grad_norm": 2.631880521774292,
      "learning_rate": 2.6556031155708372e-05,
      "loss": 0.7553,
      "step": 8266
    },
    {
      "epoch": 2.61035680454689,
      "grad_norm": 2.6349921226501465,
      "learning_rate": 2.655252262999088e-05,
      "loss": 0.7765,
      "step": 8267
    },
    {
      "epoch": 2.610672560783075,
      "grad_norm": 2.4545443058013916,
      "learning_rate": 2.6549014104273386e-05,
      "loss": 0.6953,
      "step": 8268
    },
    {
      "epoch": 2.6109883170192614,
      "grad_norm": 2.6223623752593994,
      "learning_rate": 2.654550557855589e-05,
      "loss": 0.7428,
      "step": 8269
    },
    {
      "epoch": 2.6113040732554467,
      "grad_norm": 2.3910183906555176,
      "learning_rate": 2.6541997052838396e-05,
      "loss": 0.7819,
      "step": 8270
    },
    {
      "epoch": 2.6116198294916324,
      "grad_norm": 2.7425694465637207,
      "learning_rate": 2.6538488527120903e-05,
      "loss": 0.7815,
      "step": 8271
    },
    {
      "epoch": 2.611935585727818,
      "grad_norm": 2.381471872329712,
      "learning_rate": 2.653498000140341e-05,
      "loss": 0.7204,
      "step": 8272
    },
    {
      "epoch": 2.612251341964004,
      "grad_norm": 2.2604787349700928,
      "learning_rate": 2.653147147568592e-05,
      "loss": 0.7909,
      "step": 8273
    },
    {
      "epoch": 2.6125670982001896,
      "grad_norm": 2.4422695636749268,
      "learning_rate": 2.6527962949968426e-05,
      "loss": 0.7912,
      "step": 8274
    },
    {
      "epoch": 2.612882854436375,
      "grad_norm": 2.3930859565734863,
      "learning_rate": 2.6524454424250933e-05,
      "loss": 0.7095,
      "step": 8275
    },
    {
      "epoch": 2.6131986106725607,
      "grad_norm": 2.3547182083129883,
      "learning_rate": 2.652094589853344e-05,
      "loss": 0.6978,
      "step": 8276
    },
    {
      "epoch": 2.6135143669087464,
      "grad_norm": 2.643425464630127,
      "learning_rate": 2.6517437372815947e-05,
      "loss": 0.7723,
      "step": 8277
    },
    {
      "epoch": 2.613830123144932,
      "grad_norm": 2.364976644515991,
      "learning_rate": 2.651392884709845e-05,
      "loss": 0.6525,
      "step": 8278
    },
    {
      "epoch": 2.614145879381118,
      "grad_norm": 2.812681198120117,
      "learning_rate": 2.6510420321380957e-05,
      "loss": 0.7385,
      "step": 8279
    },
    {
      "epoch": 2.6144616356173036,
      "grad_norm": 2.7351982593536377,
      "learning_rate": 2.6506911795663464e-05,
      "loss": 0.7353,
      "step": 8280
    },
    {
      "epoch": 2.6147773918534893,
      "grad_norm": 3.00227427482605,
      "learning_rate": 2.650340326994597e-05,
      "loss": 0.7257,
      "step": 8281
    },
    {
      "epoch": 2.6150931480896746,
      "grad_norm": 2.4357552528381348,
      "learning_rate": 2.6499894744228478e-05,
      "loss": 0.7928,
      "step": 8282
    },
    {
      "epoch": 2.6154089043258604,
      "grad_norm": 2.186267375946045,
      "learning_rate": 2.649638621851098e-05,
      "loss": 0.7407,
      "step": 8283
    },
    {
      "epoch": 2.615724660562046,
      "grad_norm": 2.7354888916015625,
      "learning_rate": 2.6492877692793488e-05,
      "loss": 0.8288,
      "step": 8284
    },
    {
      "epoch": 2.616040416798232,
      "grad_norm": 2.348182201385498,
      "learning_rate": 2.6489369167075995e-05,
      "loss": 0.7232,
      "step": 8285
    },
    {
      "epoch": 2.6163561730344176,
      "grad_norm": 2.239596128463745,
      "learning_rate": 2.64858606413585e-05,
      "loss": 0.7415,
      "step": 8286
    },
    {
      "epoch": 2.616671929270603,
      "grad_norm": 2.1248116493225098,
      "learning_rate": 2.6482352115641008e-05,
      "loss": 0.6494,
      "step": 8287
    },
    {
      "epoch": 2.6169876855067886,
      "grad_norm": 2.2019712924957275,
      "learning_rate": 2.647884358992351e-05,
      "loss": 0.7126,
      "step": 8288
    },
    {
      "epoch": 2.6173034417429744,
      "grad_norm": 2.490211009979248,
      "learning_rate": 2.647533506420602e-05,
      "loss": 0.7568,
      "step": 8289
    },
    {
      "epoch": 2.61761919797916,
      "grad_norm": 2.488410234451294,
      "learning_rate": 2.6471826538488532e-05,
      "loss": 0.7181,
      "step": 8290
    },
    {
      "epoch": 2.617934954215346,
      "grad_norm": 2.1755330562591553,
      "learning_rate": 2.646831801277104e-05,
      "loss": 0.5978,
      "step": 8291
    },
    {
      "epoch": 2.6182507104515316,
      "grad_norm": 2.547506332397461,
      "learning_rate": 2.6464809487053542e-05,
      "loss": 0.6826,
      "step": 8292
    },
    {
      "epoch": 2.6185664666877173,
      "grad_norm": 2.3873815536499023,
      "learning_rate": 2.646130096133605e-05,
      "loss": 0.7644,
      "step": 8293
    },
    {
      "epoch": 2.6188822229239026,
      "grad_norm": 2.7988219261169434,
      "learning_rate": 2.6457792435618556e-05,
      "loss": 0.7959,
      "step": 8294
    },
    {
      "epoch": 2.6191979791600883,
      "grad_norm": 2.598907232284546,
      "learning_rate": 2.6454283909901063e-05,
      "loss": 0.7493,
      "step": 8295
    },
    {
      "epoch": 2.619513735396274,
      "grad_norm": 2.6280264854431152,
      "learning_rate": 2.645077538418357e-05,
      "loss": 0.8258,
      "step": 8296
    },
    {
      "epoch": 2.61982949163246,
      "grad_norm": 3.196394205093384,
      "learning_rate": 2.6447266858466073e-05,
      "loss": 0.8708,
      "step": 8297
    },
    {
      "epoch": 2.6201452478686456,
      "grad_norm": 2.4897289276123047,
      "learning_rate": 2.644375833274858e-05,
      "loss": 0.6689,
      "step": 8298
    },
    {
      "epoch": 2.620461004104831,
      "grad_norm": 2.751585006713867,
      "learning_rate": 2.6440249807031087e-05,
      "loss": 0.8334,
      "step": 8299
    },
    {
      "epoch": 2.6207767603410166,
      "grad_norm": 2.4268062114715576,
      "learning_rate": 2.6436741281313593e-05,
      "loss": 0.718,
      "step": 8300
    },
    {
      "epoch": 2.6210925165772023,
      "grad_norm": 2.137636423110962,
      "learning_rate": 2.64332327555961e-05,
      "loss": 0.7042,
      "step": 8301
    },
    {
      "epoch": 2.621408272813388,
      "grad_norm": 2.444772958755493,
      "learning_rate": 2.6429724229878604e-05,
      "loss": 0.7451,
      "step": 8302
    },
    {
      "epoch": 2.621724029049574,
      "grad_norm": 2.7853593826293945,
      "learning_rate": 2.642621570416111e-05,
      "loss": 0.739,
      "step": 8303
    },
    {
      "epoch": 2.6220397852857595,
      "grad_norm": 2.7158329486846924,
      "learning_rate": 2.6422707178443617e-05,
      "loss": 0.7929,
      "step": 8304
    },
    {
      "epoch": 2.6223555415219453,
      "grad_norm": 2.7723662853240967,
      "learning_rate": 2.6419198652726124e-05,
      "loss": 0.7329,
      "step": 8305
    },
    {
      "epoch": 2.6226712977581306,
      "grad_norm": 3.1131927967071533,
      "learning_rate": 2.641569012700863e-05,
      "loss": 0.7012,
      "step": 8306
    },
    {
      "epoch": 2.6229870539943163,
      "grad_norm": 2.6569581031799316,
      "learning_rate": 2.6412181601291138e-05,
      "loss": 0.8034,
      "step": 8307
    },
    {
      "epoch": 2.623302810230502,
      "grad_norm": 2.7217390537261963,
      "learning_rate": 2.6408673075573648e-05,
      "loss": 0.7313,
      "step": 8308
    },
    {
      "epoch": 2.623618566466688,
      "grad_norm": 2.5569090843200684,
      "learning_rate": 2.6405164549856155e-05,
      "loss": 0.7574,
      "step": 8309
    },
    {
      "epoch": 2.6239343227028735,
      "grad_norm": 2.8335981369018555,
      "learning_rate": 2.640165602413866e-05,
      "loss": 0.7531,
      "step": 8310
    },
    {
      "epoch": 2.624250078939059,
      "grad_norm": 2.5496299266815186,
      "learning_rate": 2.6398147498421165e-05,
      "loss": 0.7404,
      "step": 8311
    },
    {
      "epoch": 2.6245658351752446,
      "grad_norm": 2.227196216583252,
      "learning_rate": 2.6394638972703672e-05,
      "loss": 0.6818,
      "step": 8312
    },
    {
      "epoch": 2.6248815914114303,
      "grad_norm": 2.909942626953125,
      "learning_rate": 2.639113044698618e-05,
      "loss": 0.6412,
      "step": 8313
    },
    {
      "epoch": 2.625197347647616,
      "grad_norm": 2.3182740211486816,
      "learning_rate": 2.6387621921268685e-05,
      "loss": 0.7012,
      "step": 8314
    },
    {
      "epoch": 2.6255131038838018,
      "grad_norm": 2.9206762313842773,
      "learning_rate": 2.6384113395551192e-05,
      "loss": 0.7545,
      "step": 8315
    },
    {
      "epoch": 2.6258288601199875,
      "grad_norm": 2.8397469520568848,
      "learning_rate": 2.6380604869833696e-05,
      "loss": 0.7685,
      "step": 8316
    },
    {
      "epoch": 2.6261446163561732,
      "grad_norm": 2.1312105655670166,
      "learning_rate": 2.6377096344116202e-05,
      "loss": 0.7326,
      "step": 8317
    },
    {
      "epoch": 2.6264603725923585,
      "grad_norm": 2.9464359283447266,
      "learning_rate": 2.637358781839871e-05,
      "loss": 0.6776,
      "step": 8318
    },
    {
      "epoch": 2.6267761288285443,
      "grad_norm": 2.3246896266937256,
      "learning_rate": 2.6370079292681216e-05,
      "loss": 0.7023,
      "step": 8319
    },
    {
      "epoch": 2.62709188506473,
      "grad_norm": 2.9403493404388428,
      "learning_rate": 2.6366570766963723e-05,
      "loss": 0.8722,
      "step": 8320
    },
    {
      "epoch": 2.6274076413009158,
      "grad_norm": 2.4560766220092773,
      "learning_rate": 2.636306224124623e-05,
      "loss": 0.7539,
      "step": 8321
    },
    {
      "epoch": 2.6277233975371015,
      "grad_norm": 2.568030834197998,
      "learning_rate": 2.6359553715528733e-05,
      "loss": 0.6759,
      "step": 8322
    },
    {
      "epoch": 2.628039153773287,
      "grad_norm": 2.60827898979187,
      "learning_rate": 2.635604518981124e-05,
      "loss": 0.7684,
      "step": 8323
    },
    {
      "epoch": 2.6283549100094725,
      "grad_norm": 2.615285634994507,
      "learning_rate": 2.6352536664093747e-05,
      "loss": 0.7887,
      "step": 8324
    },
    {
      "epoch": 2.6286706662456583,
      "grad_norm": 2.7224783897399902,
      "learning_rate": 2.6349028138376257e-05,
      "loss": 0.8106,
      "step": 8325
    },
    {
      "epoch": 2.628986422481844,
      "grad_norm": 2.3194026947021484,
      "learning_rate": 2.6345519612658764e-05,
      "loss": 0.6498,
      "step": 8326
    },
    {
      "epoch": 2.6293021787180297,
      "grad_norm": 2.739100456237793,
      "learning_rate": 2.634201108694127e-05,
      "loss": 0.6793,
      "step": 8327
    },
    {
      "epoch": 2.6296179349542155,
      "grad_norm": 2.462578058242798,
      "learning_rate": 2.6338502561223777e-05,
      "loss": 0.7784,
      "step": 8328
    },
    {
      "epoch": 2.629933691190401,
      "grad_norm": 2.776928424835205,
      "learning_rate": 2.6334994035506284e-05,
      "loss": 0.7369,
      "step": 8329
    },
    {
      "epoch": 2.6302494474265865,
      "grad_norm": 2.3426976203918457,
      "learning_rate": 2.6331485509788788e-05,
      "loss": 0.747,
      "step": 8330
    },
    {
      "epoch": 2.6305652036627722,
      "grad_norm": 2.6124966144561768,
      "learning_rate": 2.6327976984071294e-05,
      "loss": 0.7995,
      "step": 8331
    },
    {
      "epoch": 2.630880959898958,
      "grad_norm": 2.4600625038146973,
      "learning_rate": 2.63244684583538e-05,
      "loss": 0.686,
      "step": 8332
    },
    {
      "epoch": 2.6311967161351437,
      "grad_norm": 2.3129255771636963,
      "learning_rate": 2.6320959932636308e-05,
      "loss": 0.7,
      "step": 8333
    },
    {
      "epoch": 2.6315124723713295,
      "grad_norm": 2.2975759506225586,
      "learning_rate": 2.6317451406918815e-05,
      "loss": 0.7095,
      "step": 8334
    },
    {
      "epoch": 2.6318282286075148,
      "grad_norm": 2.450422525405884,
      "learning_rate": 2.631394288120132e-05,
      "loss": 0.6678,
      "step": 8335
    },
    {
      "epoch": 2.632143984843701,
      "grad_norm": 2.805755615234375,
      "learning_rate": 2.6310434355483825e-05,
      "loss": 0.7734,
      "step": 8336
    },
    {
      "epoch": 2.6324597410798862,
      "grad_norm": 2.837087631225586,
      "learning_rate": 2.6306925829766332e-05,
      "loss": 0.8042,
      "step": 8337
    },
    {
      "epoch": 2.632775497316072,
      "grad_norm": 2.479093074798584,
      "learning_rate": 2.630341730404884e-05,
      "loss": 0.7937,
      "step": 8338
    },
    {
      "epoch": 2.6330912535522577,
      "grad_norm": 2.553372621536255,
      "learning_rate": 2.6299908778331346e-05,
      "loss": 0.7631,
      "step": 8339
    },
    {
      "epoch": 2.6334070097884434,
      "grad_norm": 2.3548896312713623,
      "learning_rate": 2.6296400252613852e-05,
      "loss": 0.8117,
      "step": 8340
    },
    {
      "epoch": 2.633722766024629,
      "grad_norm": 2.3309485912323,
      "learning_rate": 2.6292891726896356e-05,
      "loss": 0.8127,
      "step": 8341
    },
    {
      "epoch": 2.6340385222608145,
      "grad_norm": 2.3465702533721924,
      "learning_rate": 2.6289383201178863e-05,
      "loss": 0.7605,
      "step": 8342
    },
    {
      "epoch": 2.634354278497,
      "grad_norm": 2.2522268295288086,
      "learning_rate": 2.6285874675461376e-05,
      "loss": 0.6787,
      "step": 8343
    },
    {
      "epoch": 2.634670034733186,
      "grad_norm": 2.404404640197754,
      "learning_rate": 2.628236614974388e-05,
      "loss": 0.7729,
      "step": 8344
    },
    {
      "epoch": 2.6349857909693717,
      "grad_norm": 2.8147504329681396,
      "learning_rate": 2.6278857624026386e-05,
      "loss": 0.8058,
      "step": 8345
    },
    {
      "epoch": 2.6353015472055574,
      "grad_norm": 2.0442709922790527,
      "learning_rate": 2.6275349098308893e-05,
      "loss": 0.7091,
      "step": 8346
    },
    {
      "epoch": 2.6356173034417427,
      "grad_norm": 2.3553738594055176,
      "learning_rate": 2.62718405725914e-05,
      "loss": 0.7408,
      "step": 8347
    },
    {
      "epoch": 2.635933059677929,
      "grad_norm": 2.640146493911743,
      "learning_rate": 2.6268332046873907e-05,
      "loss": 0.7825,
      "step": 8348
    },
    {
      "epoch": 2.636248815914114,
      "grad_norm": 2.650731086730957,
      "learning_rate": 2.626482352115641e-05,
      "loss": 0.7853,
      "step": 8349
    },
    {
      "epoch": 2.6365645721503,
      "grad_norm": 2.4138336181640625,
      "learning_rate": 2.6261314995438917e-05,
      "loss": 0.84,
      "step": 8350
    },
    {
      "epoch": 2.6368803283864857,
      "grad_norm": 2.2141575813293457,
      "learning_rate": 2.6257806469721424e-05,
      "loss": 0.818,
      "step": 8351
    },
    {
      "epoch": 2.6371960846226714,
      "grad_norm": 2.1419949531555176,
      "learning_rate": 2.625429794400393e-05,
      "loss": 0.6533,
      "step": 8352
    },
    {
      "epoch": 2.637511840858857,
      "grad_norm": 2.3818440437316895,
      "learning_rate": 2.6250789418286438e-05,
      "loss": 0.6746,
      "step": 8353
    },
    {
      "epoch": 2.6378275970950424,
      "grad_norm": 2.2820205688476562,
      "learning_rate": 2.6247280892568944e-05,
      "loss": 0.8354,
      "step": 8354
    },
    {
      "epoch": 2.638143353331228,
      "grad_norm": 2.351304054260254,
      "learning_rate": 2.6243772366851448e-05,
      "loss": 0.7781,
      "step": 8355
    },
    {
      "epoch": 2.638459109567414,
      "grad_norm": 2.7107982635498047,
      "learning_rate": 2.6240263841133955e-05,
      "loss": 0.7304,
      "step": 8356
    },
    {
      "epoch": 2.6387748658035997,
      "grad_norm": 2.34439754486084,
      "learning_rate": 2.623675531541646e-05,
      "loss": 0.7235,
      "step": 8357
    },
    {
      "epoch": 2.6390906220397854,
      "grad_norm": 2.1828055381774902,
      "learning_rate": 2.6233246789698968e-05,
      "loss": 0.7006,
      "step": 8358
    },
    {
      "epoch": 2.6394063782759707,
      "grad_norm": 2.2582762241363525,
      "learning_rate": 2.6229738263981475e-05,
      "loss": 0.6985,
      "step": 8359
    },
    {
      "epoch": 2.639722134512157,
      "grad_norm": 2.2207393646240234,
      "learning_rate": 2.6226229738263985e-05,
      "loss": 0.7847,
      "step": 8360
    },
    {
      "epoch": 2.640037890748342,
      "grad_norm": 2.187274694442749,
      "learning_rate": 2.6222721212546492e-05,
      "loss": 0.6641,
      "step": 8361
    },
    {
      "epoch": 2.640353646984528,
      "grad_norm": 2.104440927505493,
      "learning_rate": 2.6219212686829e-05,
      "loss": 0.7032,
      "step": 8362
    },
    {
      "epoch": 2.6406694032207136,
      "grad_norm": 2.552621364593506,
      "learning_rate": 2.6215704161111502e-05,
      "loss": 0.7532,
      "step": 8363
    },
    {
      "epoch": 2.6409851594568994,
      "grad_norm": 2.007997751235962,
      "learning_rate": 2.621219563539401e-05,
      "loss": 0.6821,
      "step": 8364
    },
    {
      "epoch": 2.641300915693085,
      "grad_norm": 2.3894495964050293,
      "learning_rate": 2.6208687109676516e-05,
      "loss": 0.6184,
      "step": 8365
    },
    {
      "epoch": 2.6416166719292704,
      "grad_norm": 2.4126617908477783,
      "learning_rate": 2.6205178583959023e-05,
      "loss": 0.7512,
      "step": 8366
    },
    {
      "epoch": 2.641932428165456,
      "grad_norm": 2.1595256328582764,
      "learning_rate": 2.620167005824153e-05,
      "loss": 0.7454,
      "step": 8367
    },
    {
      "epoch": 2.642248184401642,
      "grad_norm": 2.7071237564086914,
      "learning_rate": 2.6198161532524036e-05,
      "loss": 0.8405,
      "step": 8368
    },
    {
      "epoch": 2.6425639406378276,
      "grad_norm": 2.4277212619781494,
      "learning_rate": 2.619465300680654e-05,
      "loss": 0.6618,
      "step": 8369
    },
    {
      "epoch": 2.6428796968740134,
      "grad_norm": 2.5309371948242188,
      "learning_rate": 2.6191144481089047e-05,
      "loss": 0.6707,
      "step": 8370
    },
    {
      "epoch": 2.6431954531101987,
      "grad_norm": 2.524881601333618,
      "learning_rate": 2.6187635955371553e-05,
      "loss": 0.7685,
      "step": 8371
    },
    {
      "epoch": 2.643511209346385,
      "grad_norm": 2.1235036849975586,
      "learning_rate": 2.618412742965406e-05,
      "loss": 0.7288,
      "step": 8372
    },
    {
      "epoch": 2.64382696558257,
      "grad_norm": 2.5603647232055664,
      "learning_rate": 2.6180618903936567e-05,
      "loss": 0.7311,
      "step": 8373
    },
    {
      "epoch": 2.644142721818756,
      "grad_norm": 2.5626039505004883,
      "learning_rate": 2.617711037821907e-05,
      "loss": 0.789,
      "step": 8374
    },
    {
      "epoch": 2.6444584780549416,
      "grad_norm": 2.3375132083892822,
      "learning_rate": 2.6173601852501577e-05,
      "loss": 0.8167,
      "step": 8375
    },
    {
      "epoch": 2.6447742342911273,
      "grad_norm": 2.6633896827697754,
      "learning_rate": 2.6170093326784084e-05,
      "loss": 0.7749,
      "step": 8376
    },
    {
      "epoch": 2.645089990527313,
      "grad_norm": 2.2674882411956787,
      "learning_rate": 2.616658480106659e-05,
      "loss": 0.7295,
      "step": 8377
    },
    {
      "epoch": 2.6454057467634984,
      "grad_norm": 2.494095802307129,
      "learning_rate": 2.61630762753491e-05,
      "loss": 0.7111,
      "step": 8378
    },
    {
      "epoch": 2.645721502999684,
      "grad_norm": 2.1667428016662598,
      "learning_rate": 2.6159567749631608e-05,
      "loss": 0.7627,
      "step": 8379
    },
    {
      "epoch": 2.64603725923587,
      "grad_norm": 2.293172597885132,
      "learning_rate": 2.6156059223914115e-05,
      "loss": 0.7171,
      "step": 8380
    },
    {
      "epoch": 2.6463530154720556,
      "grad_norm": 9.224043846130371,
      "learning_rate": 2.615255069819662e-05,
      "loss": 0.702,
      "step": 8381
    },
    {
      "epoch": 2.6466687717082413,
      "grad_norm": 2.363206386566162,
      "learning_rate": 2.614904217247913e-05,
      "loss": 0.7622,
      "step": 8382
    },
    {
      "epoch": 2.646984527944427,
      "grad_norm": 2.320366144180298,
      "learning_rate": 2.6145533646761632e-05,
      "loss": 0.8315,
      "step": 8383
    },
    {
      "epoch": 2.647300284180613,
      "grad_norm": 2.7436022758483887,
      "learning_rate": 2.614202512104414e-05,
      "loss": 0.7683,
      "step": 8384
    },
    {
      "epoch": 2.647616040416798,
      "grad_norm": 2.3347585201263428,
      "learning_rate": 2.6138516595326645e-05,
      "loss": 0.7611,
      "step": 8385
    },
    {
      "epoch": 2.647931796652984,
      "grad_norm": 2.1638951301574707,
      "learning_rate": 2.6135008069609152e-05,
      "loss": 0.7106,
      "step": 8386
    },
    {
      "epoch": 2.6482475528891696,
      "grad_norm": 2.3831961154937744,
      "learning_rate": 2.613149954389166e-05,
      "loss": 0.7106,
      "step": 8387
    },
    {
      "epoch": 2.6485633091253553,
      "grad_norm": 2.767533540725708,
      "learning_rate": 2.6127991018174162e-05,
      "loss": 0.7244,
      "step": 8388
    },
    {
      "epoch": 2.648879065361541,
      "grad_norm": 2.1633284091949463,
      "learning_rate": 2.612448249245667e-05,
      "loss": 0.7165,
      "step": 8389
    },
    {
      "epoch": 2.6491948215977263,
      "grad_norm": 2.4278359413146973,
      "learning_rate": 2.6120973966739176e-05,
      "loss": 0.8395,
      "step": 8390
    },
    {
      "epoch": 2.649510577833912,
      "grad_norm": 2.3870835304260254,
      "learning_rate": 2.6117465441021683e-05,
      "loss": 0.7638,
      "step": 8391
    },
    {
      "epoch": 2.649826334070098,
      "grad_norm": 2.3961243629455566,
      "learning_rate": 2.611395691530419e-05,
      "loss": 0.7289,
      "step": 8392
    },
    {
      "epoch": 2.6501420903062836,
      "grad_norm": 2.107114553451538,
      "learning_rate": 2.6110448389586693e-05,
      "loss": 0.631,
      "step": 8393
    },
    {
      "epoch": 2.6504578465424693,
      "grad_norm": 2.031723737716675,
      "learning_rate": 2.61069398638692e-05,
      "loss": 0.6706,
      "step": 8394
    },
    {
      "epoch": 2.650773602778655,
      "grad_norm": 2.5993707180023193,
      "learning_rate": 2.6103431338151714e-05,
      "loss": 0.6683,
      "step": 8395
    },
    {
      "epoch": 2.6510893590148408,
      "grad_norm": 2.049783945083618,
      "learning_rate": 2.609992281243422e-05,
      "loss": 0.7032,
      "step": 8396
    },
    {
      "epoch": 2.651405115251026,
      "grad_norm": 2.426870584487915,
      "learning_rate": 2.6096414286716724e-05,
      "loss": 0.7426,
      "step": 8397
    },
    {
      "epoch": 2.651720871487212,
      "grad_norm": 2.563704013824463,
      "learning_rate": 2.609290576099923e-05,
      "loss": 0.7408,
      "step": 8398
    },
    {
      "epoch": 2.6520366277233975,
      "grad_norm": 2.167780876159668,
      "learning_rate": 2.6089397235281737e-05,
      "loss": 0.7788,
      "step": 8399
    },
    {
      "epoch": 2.6523523839595833,
      "grad_norm": 3.0001113414764404,
      "learning_rate": 2.6085888709564244e-05,
      "loss": 0.7352,
      "step": 8400
    },
    {
      "epoch": 2.6523523839595833,
      "eval_loss": 0.7968266606330872,
      "eval_runtime": 4.6101,
      "eval_samples_per_second": 540.123,
      "eval_steps_per_second": 4.555,
      "step": 8400
    },
    {
      "epoch": 2.652668140195769,
      "grad_norm": 2.8235373497009277,
      "learning_rate": 2.608238018384675e-05,
      "loss": 0.7541,
      "step": 8401
    },
    {
      "epoch": 2.6529838964319543,
      "grad_norm": 2.4974043369293213,
      "learning_rate": 2.6078871658129254e-05,
      "loss": 0.7089,
      "step": 8402
    },
    {
      "epoch": 2.65329965266814,
      "grad_norm": 2.2309467792510986,
      "learning_rate": 2.607536313241176e-05,
      "loss": 0.7204,
      "step": 8403
    },
    {
      "epoch": 2.653615408904326,
      "grad_norm": 2.5267906188964844,
      "learning_rate": 2.6071854606694268e-05,
      "loss": 0.7981,
      "step": 8404
    },
    {
      "epoch": 2.6539311651405115,
      "grad_norm": 2.3236753940582275,
      "learning_rate": 2.6068346080976775e-05,
      "loss": 0.7677,
      "step": 8405
    },
    {
      "epoch": 2.6542469213766973,
      "grad_norm": 2.221526622772217,
      "learning_rate": 2.6064837555259282e-05,
      "loss": 0.7117,
      "step": 8406
    },
    {
      "epoch": 2.654562677612883,
      "grad_norm": 2.194565773010254,
      "learning_rate": 2.6061329029541785e-05,
      "loss": 0.7297,
      "step": 8407
    },
    {
      "epoch": 2.6548784338490687,
      "grad_norm": 2.4610109329223633,
      "learning_rate": 2.6057820503824292e-05,
      "loss": 0.7511,
      "step": 8408
    },
    {
      "epoch": 2.655194190085254,
      "grad_norm": 2.633988857269287,
      "learning_rate": 2.60543119781068e-05,
      "loss": 0.7014,
      "step": 8409
    },
    {
      "epoch": 2.6555099463214398,
      "grad_norm": 2.1860010623931885,
      "learning_rate": 2.6050803452389306e-05,
      "loss": 0.7407,
      "step": 8410
    },
    {
      "epoch": 2.6558257025576255,
      "grad_norm": 2.2085788249969482,
      "learning_rate": 2.6047294926671812e-05,
      "loss": 0.7365,
      "step": 8411
    },
    {
      "epoch": 2.6561414587938112,
      "grad_norm": 2.3676583766937256,
      "learning_rate": 2.604378640095432e-05,
      "loss": 0.7129,
      "step": 8412
    },
    {
      "epoch": 2.656457215029997,
      "grad_norm": 2.0445642471313477,
      "learning_rate": 2.604027787523683e-05,
      "loss": 0.6519,
      "step": 8413
    },
    {
      "epoch": 2.6567729712661823,
      "grad_norm": 2.5225515365600586,
      "learning_rate": 2.6036769349519336e-05,
      "loss": 0.7776,
      "step": 8414
    },
    {
      "epoch": 2.657088727502368,
      "grad_norm": 2.3624539375305176,
      "learning_rate": 2.6033260823801843e-05,
      "loss": 0.6945,
      "step": 8415
    },
    {
      "epoch": 2.6574044837385538,
      "grad_norm": 2.2429039478302,
      "learning_rate": 2.6029752298084346e-05,
      "loss": 0.6599,
      "step": 8416
    },
    {
      "epoch": 2.6577202399747395,
      "grad_norm": 2.302834987640381,
      "learning_rate": 2.6026243772366853e-05,
      "loss": 0.7193,
      "step": 8417
    },
    {
      "epoch": 2.6580359962109252,
      "grad_norm": 2.260072946548462,
      "learning_rate": 2.602273524664936e-05,
      "loss": 0.7011,
      "step": 8418
    },
    {
      "epoch": 2.658351752447111,
      "grad_norm": 2.1160614490509033,
      "learning_rate": 2.6019226720931867e-05,
      "loss": 0.74,
      "step": 8419
    },
    {
      "epoch": 2.6586675086832967,
      "grad_norm": 2.061440944671631,
      "learning_rate": 2.6015718195214374e-05,
      "loss": 0.6539,
      "step": 8420
    },
    {
      "epoch": 2.658983264919482,
      "grad_norm": 2.417125701904297,
      "learning_rate": 2.6012209669496877e-05,
      "loss": 0.7549,
      "step": 8421
    },
    {
      "epoch": 2.6592990211556677,
      "grad_norm": 2.3445324897766113,
      "learning_rate": 2.6008701143779384e-05,
      "loss": 0.7062,
      "step": 8422
    },
    {
      "epoch": 2.6596147773918535,
      "grad_norm": 2.5170319080352783,
      "learning_rate": 2.600519261806189e-05,
      "loss": 0.735,
      "step": 8423
    },
    {
      "epoch": 2.659930533628039,
      "grad_norm": 2.2394402027130127,
      "learning_rate": 2.6001684092344398e-05,
      "loss": 0.6606,
      "step": 8424
    },
    {
      "epoch": 2.660246289864225,
      "grad_norm": 2.483062982559204,
      "learning_rate": 2.5998175566626904e-05,
      "loss": 0.7771,
      "step": 8425
    },
    {
      "epoch": 2.6605620461004102,
      "grad_norm": 2.7876944541931152,
      "learning_rate": 2.599466704090941e-05,
      "loss": 0.7509,
      "step": 8426
    },
    {
      "epoch": 2.660877802336596,
      "grad_norm": 2.660281181335449,
      "learning_rate": 2.5991158515191915e-05,
      "loss": 0.6792,
      "step": 8427
    },
    {
      "epoch": 2.6611935585727817,
      "grad_norm": 2.818680763244629,
      "learning_rate": 2.598764998947442e-05,
      "loss": 0.8398,
      "step": 8428
    },
    {
      "epoch": 2.6615093148089675,
      "grad_norm": 2.803900718688965,
      "learning_rate": 2.5984141463756928e-05,
      "loss": 0.7893,
      "step": 8429
    },
    {
      "epoch": 2.661825071045153,
      "grad_norm": 2.4887654781341553,
      "learning_rate": 2.598063293803944e-05,
      "loss": 0.6941,
      "step": 8430
    },
    {
      "epoch": 2.662140827281339,
      "grad_norm": 2.4038164615631104,
      "learning_rate": 2.5977124412321945e-05,
      "loss": 0.7056,
      "step": 8431
    },
    {
      "epoch": 2.6624565835175247,
      "grad_norm": 2.6170549392700195,
      "learning_rate": 2.5973615886604452e-05,
      "loss": 0.7302,
      "step": 8432
    },
    {
      "epoch": 2.66277233975371,
      "grad_norm": 2.829526424407959,
      "learning_rate": 2.597010736088696e-05,
      "loss": 0.7421,
      "step": 8433
    },
    {
      "epoch": 2.6630880959898957,
      "grad_norm": 2.3505613803863525,
      "learning_rate": 2.5966598835169466e-05,
      "loss": 0.7342,
      "step": 8434
    },
    {
      "epoch": 2.6634038522260814,
      "grad_norm": 2.1942179203033447,
      "learning_rate": 2.596309030945197e-05,
      "loss": 0.7143,
      "step": 8435
    },
    {
      "epoch": 2.663719608462267,
      "grad_norm": 2.4086501598358154,
      "learning_rate": 2.5959581783734476e-05,
      "loss": 0.7275,
      "step": 8436
    },
    {
      "epoch": 2.664035364698453,
      "grad_norm": 2.3911633491516113,
      "learning_rate": 2.5956073258016983e-05,
      "loss": 0.6963,
      "step": 8437
    },
    {
      "epoch": 2.664351120934638,
      "grad_norm": 2.2968788146972656,
      "learning_rate": 2.595256473229949e-05,
      "loss": 0.747,
      "step": 8438
    },
    {
      "epoch": 2.6646668771708244,
      "grad_norm": 2.3520095348358154,
      "learning_rate": 2.5949056206581996e-05,
      "loss": 0.7506,
      "step": 8439
    },
    {
      "epoch": 2.6649826334070097,
      "grad_norm": 2.3529250621795654,
      "learning_rate": 2.59455476808645e-05,
      "loss": 0.7872,
      "step": 8440
    },
    {
      "epoch": 2.6652983896431954,
      "grad_norm": 2.641555070877075,
      "learning_rate": 2.5942039155147007e-05,
      "loss": 0.7072,
      "step": 8441
    },
    {
      "epoch": 2.665614145879381,
      "grad_norm": 2.5218889713287354,
      "learning_rate": 2.5938530629429513e-05,
      "loss": 0.7417,
      "step": 8442
    },
    {
      "epoch": 2.665929902115567,
      "grad_norm": 2.656703472137451,
      "learning_rate": 2.593502210371202e-05,
      "loss": 0.7669,
      "step": 8443
    },
    {
      "epoch": 2.6662456583517526,
      "grad_norm": 2.252285957336426,
      "learning_rate": 2.5931513577994527e-05,
      "loss": 0.7165,
      "step": 8444
    },
    {
      "epoch": 2.666561414587938,
      "grad_norm": 2.6459202766418457,
      "learning_rate": 2.5928005052277034e-05,
      "loss": 0.8064,
      "step": 8445
    },
    {
      "epoch": 2.6668771708241237,
      "grad_norm": 2.0463249683380127,
      "learning_rate": 2.5924496526559537e-05,
      "loss": 0.7716,
      "step": 8446
    },
    {
      "epoch": 2.6671929270603094,
      "grad_norm": 2.4274847507476807,
      "learning_rate": 2.5920988000842044e-05,
      "loss": 0.709,
      "step": 8447
    },
    {
      "epoch": 2.667508683296495,
      "grad_norm": 2.2339375019073486,
      "learning_rate": 2.5917479475124558e-05,
      "loss": 0.7451,
      "step": 8448
    },
    {
      "epoch": 2.667824439532681,
      "grad_norm": 3.140960216522217,
      "learning_rate": 2.591397094940706e-05,
      "loss": 0.7413,
      "step": 8449
    },
    {
      "epoch": 2.668140195768866,
      "grad_norm": 2.865203619003296,
      "learning_rate": 2.5910462423689568e-05,
      "loss": 0.8517,
      "step": 8450
    },
    {
      "epoch": 2.6684559520050524,
      "grad_norm": 2.341926097869873,
      "learning_rate": 2.5906953897972075e-05,
      "loss": 0.7375,
      "step": 8451
    },
    {
      "epoch": 2.6687717082412377,
      "grad_norm": 2.310011148452759,
      "learning_rate": 2.590344537225458e-05,
      "loss": 0.7649,
      "step": 8452
    },
    {
      "epoch": 2.6690874644774234,
      "grad_norm": 2.4710991382598877,
      "learning_rate": 2.589993684653709e-05,
      "loss": 0.7795,
      "step": 8453
    },
    {
      "epoch": 2.669403220713609,
      "grad_norm": 2.4192450046539307,
      "learning_rate": 2.5896428320819592e-05,
      "loss": 0.7264,
      "step": 8454
    },
    {
      "epoch": 2.669718976949795,
      "grad_norm": 2.4949088096618652,
      "learning_rate": 2.58929197951021e-05,
      "loss": 0.7787,
      "step": 8455
    },
    {
      "epoch": 2.6700347331859806,
      "grad_norm": 2.4507617950439453,
      "learning_rate": 2.5889411269384605e-05,
      "loss": 0.7433,
      "step": 8456
    },
    {
      "epoch": 2.670350489422166,
      "grad_norm": 2.2427921295166016,
      "learning_rate": 2.5885902743667112e-05,
      "loss": 0.779,
      "step": 8457
    },
    {
      "epoch": 2.6706662456583516,
      "grad_norm": 2.372410297393799,
      "learning_rate": 2.588239421794962e-05,
      "loss": 0.7392,
      "step": 8458
    },
    {
      "epoch": 2.6709820018945374,
      "grad_norm": 2.680722951889038,
      "learning_rate": 2.5878885692232126e-05,
      "loss": 0.828,
      "step": 8459
    },
    {
      "epoch": 2.671297758130723,
      "grad_norm": 2.5191118717193604,
      "learning_rate": 2.587537716651463e-05,
      "loss": 0.7064,
      "step": 8460
    },
    {
      "epoch": 2.671613514366909,
      "grad_norm": 2.500908613204956,
      "learning_rate": 2.5871868640797136e-05,
      "loss": 0.8285,
      "step": 8461
    },
    {
      "epoch": 2.671929270603094,
      "grad_norm": 2.6274044513702393,
      "learning_rate": 2.5868360115079643e-05,
      "loss": 0.8343,
      "step": 8462
    },
    {
      "epoch": 2.6722450268392803,
      "grad_norm": 2.4403154850006104,
      "learning_rate": 2.586485158936215e-05,
      "loss": 0.7397,
      "step": 8463
    },
    {
      "epoch": 2.6725607830754656,
      "grad_norm": 2.363140821456909,
      "learning_rate": 2.5861343063644657e-05,
      "loss": 0.7692,
      "step": 8464
    },
    {
      "epoch": 2.6728765393116514,
      "grad_norm": 2.525252103805542,
      "learning_rate": 2.5857834537927167e-05,
      "loss": 0.7106,
      "step": 8465
    },
    {
      "epoch": 2.673192295547837,
      "grad_norm": 2.231515645980835,
      "learning_rate": 2.5854326012209674e-05,
      "loss": 0.6948,
      "step": 8466
    },
    {
      "epoch": 2.673508051784023,
      "grad_norm": 2.6129183769226074,
      "learning_rate": 2.585081748649218e-05,
      "loss": 0.7869,
      "step": 8467
    },
    {
      "epoch": 2.6738238080202086,
      "grad_norm": 2.2508139610290527,
      "learning_rate": 2.5847308960774684e-05,
      "loss": 0.8117,
      "step": 8468
    },
    {
      "epoch": 2.674139564256394,
      "grad_norm": 2.699800968170166,
      "learning_rate": 2.584380043505719e-05,
      "loss": 0.8856,
      "step": 8469
    },
    {
      "epoch": 2.6744553204925796,
      "grad_norm": 2.2542290687561035,
      "learning_rate": 2.5840291909339697e-05,
      "loss": 0.7858,
      "step": 8470
    },
    {
      "epoch": 2.6747710767287654,
      "grad_norm": 2.0294015407562256,
      "learning_rate": 2.5836783383622204e-05,
      "loss": 0.7071,
      "step": 8471
    },
    {
      "epoch": 2.675086832964951,
      "grad_norm": 2.432213068008423,
      "learning_rate": 2.583327485790471e-05,
      "loss": 0.7929,
      "step": 8472
    },
    {
      "epoch": 2.675402589201137,
      "grad_norm": 1.9975942373275757,
      "learning_rate": 2.5829766332187218e-05,
      "loss": 0.6614,
      "step": 8473
    },
    {
      "epoch": 2.675718345437322,
      "grad_norm": 2.1554436683654785,
      "learning_rate": 2.582625780646972e-05,
      "loss": 0.6712,
      "step": 8474
    },
    {
      "epoch": 2.6760341016735083,
      "grad_norm": 2.093787670135498,
      "learning_rate": 2.5822749280752228e-05,
      "loss": 0.6668,
      "step": 8475
    },
    {
      "epoch": 2.6763498579096936,
      "grad_norm": 2.326030731201172,
      "learning_rate": 2.5819240755034735e-05,
      "loss": 0.7838,
      "step": 8476
    },
    {
      "epoch": 2.6766656141458793,
      "grad_norm": 2.5169622898101807,
      "learning_rate": 2.5815732229317242e-05,
      "loss": 0.7891,
      "step": 8477
    },
    {
      "epoch": 2.676981370382065,
      "grad_norm": 2.5378623008728027,
      "learning_rate": 2.581222370359975e-05,
      "loss": 0.8178,
      "step": 8478
    },
    {
      "epoch": 2.677297126618251,
      "grad_norm": 2.6857197284698486,
      "learning_rate": 2.5808715177882252e-05,
      "loss": 0.8182,
      "step": 8479
    },
    {
      "epoch": 2.6776128828544365,
      "grad_norm": 2.3159754276275635,
      "learning_rate": 2.580520665216476e-05,
      "loss": 0.663,
      "step": 8480
    },
    {
      "epoch": 2.677928639090622,
      "grad_norm": 2.428288221359253,
      "learning_rate": 2.5801698126447266e-05,
      "loss": 0.7461,
      "step": 8481
    },
    {
      "epoch": 2.6782443953268076,
      "grad_norm": 2.735147714614868,
      "learning_rate": 2.5798189600729772e-05,
      "loss": 0.7455,
      "step": 8482
    },
    {
      "epoch": 2.6785601515629933,
      "grad_norm": 2.276893377304077,
      "learning_rate": 2.5794681075012283e-05,
      "loss": 0.7464,
      "step": 8483
    },
    {
      "epoch": 2.678875907799179,
      "grad_norm": 2.483107089996338,
      "learning_rate": 2.579117254929479e-05,
      "loss": 0.6974,
      "step": 8484
    },
    {
      "epoch": 2.679191664035365,
      "grad_norm": 2.7476325035095215,
      "learning_rate": 2.5787664023577296e-05,
      "loss": 0.7357,
      "step": 8485
    },
    {
      "epoch": 2.67950742027155,
      "grad_norm": 2.9310269355773926,
      "learning_rate": 2.5784155497859803e-05,
      "loss": 0.7894,
      "step": 8486
    },
    {
      "epoch": 2.6798231765077363,
      "grad_norm": 2.468895196914673,
      "learning_rate": 2.578064697214231e-05,
      "loss": 0.7546,
      "step": 8487
    },
    {
      "epoch": 2.6801389327439216,
      "grad_norm": 2.29955792427063,
      "learning_rate": 2.5777138446424813e-05,
      "loss": 0.6408,
      "step": 8488
    },
    {
      "epoch": 2.6804546889801073,
      "grad_norm": 2.4463388919830322,
      "learning_rate": 2.577362992070732e-05,
      "loss": 0.7636,
      "step": 8489
    },
    {
      "epoch": 2.680770445216293,
      "grad_norm": 2.384150266647339,
      "learning_rate": 2.5770121394989827e-05,
      "loss": 0.8132,
      "step": 8490
    },
    {
      "epoch": 2.681086201452479,
      "grad_norm": 2.9806296825408936,
      "learning_rate": 2.5766612869272334e-05,
      "loss": 0.7973,
      "step": 8491
    },
    {
      "epoch": 2.6814019576886645,
      "grad_norm": 2.3597569465637207,
      "learning_rate": 2.576310434355484e-05,
      "loss": 0.6627,
      "step": 8492
    },
    {
      "epoch": 2.68171771392485,
      "grad_norm": 2.489423990249634,
      "learning_rate": 2.5759595817837344e-05,
      "loss": 0.6706,
      "step": 8493
    },
    {
      "epoch": 2.6820334701610355,
      "grad_norm": 2.4898953437805176,
      "learning_rate": 2.575608729211985e-05,
      "loss": 0.7464,
      "step": 8494
    },
    {
      "epoch": 2.6823492263972213,
      "grad_norm": 2.413421869277954,
      "learning_rate": 2.5752578766402358e-05,
      "loss": 0.6715,
      "step": 8495
    },
    {
      "epoch": 2.682664982633407,
      "grad_norm": 7.5998005867004395,
      "learning_rate": 2.5749070240684864e-05,
      "loss": 0.7287,
      "step": 8496
    },
    {
      "epoch": 2.6829807388695928,
      "grad_norm": 2.608133554458618,
      "learning_rate": 2.574556171496737e-05,
      "loss": 0.7184,
      "step": 8497
    },
    {
      "epoch": 2.6832964951057785,
      "grad_norm": 2.5612409114837646,
      "learning_rate": 2.5742053189249875e-05,
      "loss": 0.8706,
      "step": 8498
    },
    {
      "epoch": 2.6836122513419642,
      "grad_norm": 2.6912648677825928,
      "learning_rate": 2.573854466353238e-05,
      "loss": 0.6846,
      "step": 8499
    },
    {
      "epoch": 2.6839280075781495,
      "grad_norm": 1.939754605293274,
      "learning_rate": 2.5735036137814895e-05,
      "loss": 0.6249,
      "step": 8500
    },
    {
      "epoch": 2.6842437638143353,
      "grad_norm": 2.826324701309204,
      "learning_rate": 2.5731527612097402e-05,
      "loss": 0.8416,
      "step": 8501
    },
    {
      "epoch": 2.684559520050521,
      "grad_norm": 2.8148906230926514,
      "learning_rate": 2.5728019086379905e-05,
      "loss": 0.9016,
      "step": 8502
    },
    {
      "epoch": 2.6848752762867067,
      "grad_norm": 2.8792617321014404,
      "learning_rate": 2.5724510560662412e-05,
      "loss": 0.7875,
      "step": 8503
    },
    {
      "epoch": 2.6851910325228925,
      "grad_norm": 2.5082626342773438,
      "learning_rate": 2.572100203494492e-05,
      "loss": 0.7148,
      "step": 8504
    },
    {
      "epoch": 2.685506788759078,
      "grad_norm": 2.6201019287109375,
      "learning_rate": 2.5717493509227426e-05,
      "loss": 0.6223,
      "step": 8505
    },
    {
      "epoch": 2.6858225449952635,
      "grad_norm": 2.479734182357788,
      "learning_rate": 2.5713984983509933e-05,
      "loss": 0.7948,
      "step": 8506
    },
    {
      "epoch": 2.6861383012314493,
      "grad_norm": 2.4883720874786377,
      "learning_rate": 2.5710476457792436e-05,
      "loss": 0.775,
      "step": 8507
    },
    {
      "epoch": 2.686454057467635,
      "grad_norm": 2.5040390491485596,
      "learning_rate": 2.5706967932074943e-05,
      "loss": 0.7732,
      "step": 8508
    },
    {
      "epoch": 2.6867698137038207,
      "grad_norm": 2.645108699798584,
      "learning_rate": 2.570345940635745e-05,
      "loss": 0.7552,
      "step": 8509
    },
    {
      "epoch": 2.6870855699400065,
      "grad_norm": 2.5283992290496826,
      "learning_rate": 2.5699950880639956e-05,
      "loss": 0.7891,
      "step": 8510
    },
    {
      "epoch": 2.687401326176192,
      "grad_norm": 2.297412395477295,
      "learning_rate": 2.5696442354922463e-05,
      "loss": 0.8055,
      "step": 8511
    },
    {
      "epoch": 2.6877170824123775,
      "grad_norm": 2.3031227588653564,
      "learning_rate": 2.5692933829204967e-05,
      "loss": 0.7063,
      "step": 8512
    },
    {
      "epoch": 2.6880328386485632,
      "grad_norm": 2.5275275707244873,
      "learning_rate": 2.5689425303487473e-05,
      "loss": 0.6389,
      "step": 8513
    },
    {
      "epoch": 2.688348594884749,
      "grad_norm": 2.1306684017181396,
      "learning_rate": 2.568591677776998e-05,
      "loss": 0.6747,
      "step": 8514
    },
    {
      "epoch": 2.6886643511209347,
      "grad_norm": 2.5287129878997803,
      "learning_rate": 2.5682408252052487e-05,
      "loss": 0.7596,
      "step": 8515
    },
    {
      "epoch": 2.6889801073571205,
      "grad_norm": 2.1655378341674805,
      "learning_rate": 2.5678899726334994e-05,
      "loss": 0.6342,
      "step": 8516
    },
    {
      "epoch": 2.6892958635933057,
      "grad_norm": 3.5965065956115723,
      "learning_rate": 2.56753912006175e-05,
      "loss": 0.7779,
      "step": 8517
    },
    {
      "epoch": 2.6896116198294915,
      "grad_norm": 2.588719367980957,
      "learning_rate": 2.567188267490001e-05,
      "loss": 0.8033,
      "step": 8518
    },
    {
      "epoch": 2.6899273760656772,
      "grad_norm": 2.3880105018615723,
      "learning_rate": 2.5668374149182518e-05,
      "loss": 0.6702,
      "step": 8519
    },
    {
      "epoch": 2.690243132301863,
      "grad_norm": 2.76770281791687,
      "learning_rate": 2.5664865623465025e-05,
      "loss": 0.7822,
      "step": 8520
    },
    {
      "epoch": 2.6905588885380487,
      "grad_norm": 2.5445058345794678,
      "learning_rate": 2.5661357097747528e-05,
      "loss": 0.7128,
      "step": 8521
    },
    {
      "epoch": 2.6908746447742344,
      "grad_norm": 2.3921186923980713,
      "learning_rate": 2.5657848572030035e-05,
      "loss": 0.7308,
      "step": 8522
    },
    {
      "epoch": 2.69119040101042,
      "grad_norm": 2.497887134552002,
      "learning_rate": 2.565434004631254e-05,
      "loss": 0.6954,
      "step": 8523
    },
    {
      "epoch": 2.6915061572466055,
      "grad_norm": 2.407344341278076,
      "learning_rate": 2.565083152059505e-05,
      "loss": 0.7779,
      "step": 8524
    },
    {
      "epoch": 2.691821913482791,
      "grad_norm": 2.680431604385376,
      "learning_rate": 2.5647322994877555e-05,
      "loss": 0.8058,
      "step": 8525
    },
    {
      "epoch": 2.692137669718977,
      "grad_norm": 2.6604864597320557,
      "learning_rate": 2.564381446916006e-05,
      "loss": 0.7658,
      "step": 8526
    },
    {
      "epoch": 2.6924534259551627,
      "grad_norm": 2.216351270675659,
      "learning_rate": 2.5640305943442565e-05,
      "loss": 0.6609,
      "step": 8527
    },
    {
      "epoch": 2.6927691821913484,
      "grad_norm": 2.7572269439697266,
      "learning_rate": 2.5636797417725072e-05,
      "loss": 0.8579,
      "step": 8528
    },
    {
      "epoch": 2.6930849384275337,
      "grad_norm": 3.9520363807678223,
      "learning_rate": 2.563328889200758e-05,
      "loss": 0.7447,
      "step": 8529
    },
    {
      "epoch": 2.6934006946637195,
      "grad_norm": 2.501776933670044,
      "learning_rate": 2.5629780366290086e-05,
      "loss": 0.6522,
      "step": 8530
    },
    {
      "epoch": 2.693716450899905,
      "grad_norm": 2.2578489780426025,
      "learning_rate": 2.5626271840572593e-05,
      "loss": 0.6709,
      "step": 8531
    },
    {
      "epoch": 2.694032207136091,
      "grad_norm": 2.737102508544922,
      "learning_rate": 2.5622763314855096e-05,
      "loss": 0.8655,
      "step": 8532
    },
    {
      "epoch": 2.6943479633722767,
      "grad_norm": 2.5721981525421143,
      "learning_rate": 2.5619254789137603e-05,
      "loss": 0.8144,
      "step": 8533
    },
    {
      "epoch": 2.6946637196084624,
      "grad_norm": 2.368241786956787,
      "learning_rate": 2.561574626342011e-05,
      "loss": 0.6799,
      "step": 8534
    },
    {
      "epoch": 2.694979475844648,
      "grad_norm": 2.105212926864624,
      "learning_rate": 2.561223773770262e-05,
      "loss": 0.7424,
      "step": 8535
    },
    {
      "epoch": 2.6952952320808334,
      "grad_norm": 2.3432092666625977,
      "learning_rate": 2.5608729211985127e-05,
      "loss": 0.6782,
      "step": 8536
    },
    {
      "epoch": 2.695610988317019,
      "grad_norm": 2.7130136489868164,
      "learning_rate": 2.5605220686267634e-05,
      "loss": 0.8176,
      "step": 8537
    },
    {
      "epoch": 2.695926744553205,
      "grad_norm": 2.47017502784729,
      "learning_rate": 2.560171216055014e-05,
      "loss": 0.7843,
      "step": 8538
    },
    {
      "epoch": 2.6962425007893907,
      "grad_norm": 2.4191250801086426,
      "learning_rate": 2.5598203634832647e-05,
      "loss": 0.761,
      "step": 8539
    },
    {
      "epoch": 2.6965582570255764,
      "grad_norm": 2.8323097229003906,
      "learning_rate": 2.559469510911515e-05,
      "loss": 0.6974,
      "step": 8540
    },
    {
      "epoch": 2.6968740132617617,
      "grad_norm": 2.5637505054473877,
      "learning_rate": 2.5591186583397657e-05,
      "loss": 0.7355,
      "step": 8541
    },
    {
      "epoch": 2.6971897694979474,
      "grad_norm": 1.9644272327423096,
      "learning_rate": 2.5587678057680164e-05,
      "loss": 0.7353,
      "step": 8542
    },
    {
      "epoch": 2.697505525734133,
      "grad_norm": 2.6788384914398193,
      "learning_rate": 2.558416953196267e-05,
      "loss": 0.7279,
      "step": 8543
    },
    {
      "epoch": 2.697821281970319,
      "grad_norm": 2.3976430892944336,
      "learning_rate": 2.5580661006245178e-05,
      "loss": 0.7642,
      "step": 8544
    },
    {
      "epoch": 2.6981370382065046,
      "grad_norm": 2.137380361557007,
      "learning_rate": 2.5577152480527685e-05,
      "loss": 0.6962,
      "step": 8545
    },
    {
      "epoch": 2.6984527944426904,
      "grad_norm": 2.6858417987823486,
      "learning_rate": 2.5573643954810188e-05,
      "loss": 0.7556,
      "step": 8546
    },
    {
      "epoch": 2.698768550678876,
      "grad_norm": 2.214200496673584,
      "learning_rate": 2.5570135429092695e-05,
      "loss": 0.6242,
      "step": 8547
    },
    {
      "epoch": 2.6990843069150614,
      "grad_norm": 2.704279661178589,
      "learning_rate": 2.5566626903375202e-05,
      "loss": 0.8491,
      "step": 8548
    },
    {
      "epoch": 2.699400063151247,
      "grad_norm": 2.1914150714874268,
      "learning_rate": 2.556311837765771e-05,
      "loss": 0.7241,
      "step": 8549
    },
    {
      "epoch": 2.699715819387433,
      "grad_norm": 2.4098730087280273,
      "learning_rate": 2.5559609851940215e-05,
      "loss": 0.7289,
      "step": 8550
    },
    {
      "epoch": 2.7000315756236186,
      "grad_norm": 2.424496650695801,
      "learning_rate": 2.555610132622272e-05,
      "loss": 0.8259,
      "step": 8551
    },
    {
      "epoch": 2.7003473318598044,
      "grad_norm": 2.5399115085601807,
      "learning_rate": 2.5552592800505226e-05,
      "loss": 0.7059,
      "step": 8552
    },
    {
      "epoch": 2.7006630880959897,
      "grad_norm": 2.2579538822174072,
      "learning_rate": 2.554908427478774e-05,
      "loss": 0.7034,
      "step": 8553
    },
    {
      "epoch": 2.700978844332176,
      "grad_norm": 2.6269469261169434,
      "learning_rate": 2.5545575749070243e-05,
      "loss": 0.7544,
      "step": 8554
    },
    {
      "epoch": 2.701294600568361,
      "grad_norm": 2.194556951522827,
      "learning_rate": 2.554206722335275e-05,
      "loss": 0.6699,
      "step": 8555
    },
    {
      "epoch": 2.701610356804547,
      "grad_norm": 2.3527541160583496,
      "learning_rate": 2.5538558697635256e-05,
      "loss": 0.726,
      "step": 8556
    },
    {
      "epoch": 2.7019261130407326,
      "grad_norm": 2.820737838745117,
      "learning_rate": 2.5535050171917763e-05,
      "loss": 0.751,
      "step": 8557
    },
    {
      "epoch": 2.7022418692769183,
      "grad_norm": 2.8298513889312744,
      "learning_rate": 2.553154164620027e-05,
      "loss": 0.751,
      "step": 8558
    },
    {
      "epoch": 2.702557625513104,
      "grad_norm": 2.8962788581848145,
      "learning_rate": 2.5528033120482773e-05,
      "loss": 0.6777,
      "step": 8559
    },
    {
      "epoch": 2.7028733817492894,
      "grad_norm": 2.566971778869629,
      "learning_rate": 2.552452459476528e-05,
      "loss": 0.5685,
      "step": 8560
    },
    {
      "epoch": 2.703189137985475,
      "grad_norm": 2.4354331493377686,
      "learning_rate": 2.5521016069047787e-05,
      "loss": 0.7648,
      "step": 8561
    },
    {
      "epoch": 2.703504894221661,
      "grad_norm": 2.267086982727051,
      "learning_rate": 2.5517507543330294e-05,
      "loss": 0.7031,
      "step": 8562
    },
    {
      "epoch": 2.7038206504578466,
      "grad_norm": 2.9851248264312744,
      "learning_rate": 2.55139990176128e-05,
      "loss": 0.7652,
      "step": 8563
    },
    {
      "epoch": 2.7041364066940323,
      "grad_norm": 2.7841110229492188,
      "learning_rate": 2.5510490491895307e-05,
      "loss": 0.8049,
      "step": 8564
    },
    {
      "epoch": 2.7044521629302176,
      "grad_norm": 2.616584062576294,
      "learning_rate": 2.550698196617781e-05,
      "loss": 0.7879,
      "step": 8565
    },
    {
      "epoch": 2.704767919166404,
      "grad_norm": 2.549264907836914,
      "learning_rate": 2.5503473440460318e-05,
      "loss": 0.6868,
      "step": 8566
    },
    {
      "epoch": 2.705083675402589,
      "grad_norm": 2.493360996246338,
      "learning_rate": 2.5499964914742824e-05,
      "loss": 0.745,
      "step": 8567
    },
    {
      "epoch": 2.705399431638775,
      "grad_norm": 2.4882962703704834,
      "learning_rate": 2.549645638902533e-05,
      "loss": 0.6413,
      "step": 8568
    },
    {
      "epoch": 2.7057151878749606,
      "grad_norm": 2.4755661487579346,
      "learning_rate": 2.5492947863307838e-05,
      "loss": 0.7365,
      "step": 8569
    },
    {
      "epoch": 2.7060309441111463,
      "grad_norm": 2.593780279159546,
      "learning_rate": 2.5489439337590348e-05,
      "loss": 0.699,
      "step": 8570
    },
    {
      "epoch": 2.706346700347332,
      "grad_norm": 2.7914586067199707,
      "learning_rate": 2.5485930811872855e-05,
      "loss": 0.756,
      "step": 8571
    },
    {
      "epoch": 2.7066624565835173,
      "grad_norm": 2.790888786315918,
      "learning_rate": 2.5482422286155362e-05,
      "loss": 0.7629,
      "step": 8572
    },
    {
      "epoch": 2.706978212819703,
      "grad_norm": 2.6854963302612305,
      "learning_rate": 2.5478913760437865e-05,
      "loss": 0.7081,
      "step": 8573
    },
    {
      "epoch": 2.707293969055889,
      "grad_norm": 2.539830446243286,
      "learning_rate": 2.5475405234720372e-05,
      "loss": 0.7555,
      "step": 8574
    },
    {
      "epoch": 2.7076097252920746,
      "grad_norm": 2.7081305980682373,
      "learning_rate": 2.547189670900288e-05,
      "loss": 0.8436,
      "step": 8575
    },
    {
      "epoch": 2.7079254815282603,
      "grad_norm": 2.4429826736450195,
      "learning_rate": 2.5468388183285386e-05,
      "loss": 0.7644,
      "step": 8576
    },
    {
      "epoch": 2.7082412377644456,
      "grad_norm": 2.7234292030334473,
      "learning_rate": 2.5464879657567893e-05,
      "loss": 0.737,
      "step": 8577
    },
    {
      "epoch": 2.7085569940006318,
      "grad_norm": 2.6703391075134277,
      "learning_rate": 2.54613711318504e-05,
      "loss": 0.7239,
      "step": 8578
    },
    {
      "epoch": 2.708872750236817,
      "grad_norm": 2.8058104515075684,
      "learning_rate": 2.5457862606132903e-05,
      "loss": 0.7664,
      "step": 8579
    },
    {
      "epoch": 2.709188506473003,
      "grad_norm": 2.504486083984375,
      "learning_rate": 2.545435408041541e-05,
      "loss": 0.7577,
      "step": 8580
    },
    {
      "epoch": 2.7095042627091885,
      "grad_norm": 3.0798192024230957,
      "learning_rate": 2.5450845554697916e-05,
      "loss": 0.8035,
      "step": 8581
    },
    {
      "epoch": 2.7098200189453743,
      "grad_norm": 2.5639278888702393,
      "learning_rate": 2.5447337028980423e-05,
      "loss": 0.757,
      "step": 8582
    },
    {
      "epoch": 2.71013577518156,
      "grad_norm": 2.711087942123413,
      "learning_rate": 2.544382850326293e-05,
      "loss": 0.6706,
      "step": 8583
    },
    {
      "epoch": 2.7104515314177453,
      "grad_norm": 3.165513753890991,
      "learning_rate": 2.5440319977545433e-05,
      "loss": 0.7673,
      "step": 8584
    },
    {
      "epoch": 2.710767287653931,
      "grad_norm": 2.5625088214874268,
      "learning_rate": 2.543681145182794e-05,
      "loss": 0.766,
      "step": 8585
    },
    {
      "epoch": 2.711083043890117,
      "grad_norm": 2.4975802898406982,
      "learning_rate": 2.5433302926110447e-05,
      "loss": 0.804,
      "step": 8586
    },
    {
      "epoch": 2.7113988001263025,
      "grad_norm": 2.5460798740386963,
      "learning_rate": 2.5429794400392954e-05,
      "loss": 0.6935,
      "step": 8587
    },
    {
      "epoch": 2.7117145563624883,
      "grad_norm": 2.4450571537017822,
      "learning_rate": 2.5426285874675464e-05,
      "loss": 0.6732,
      "step": 8588
    },
    {
      "epoch": 2.7120303125986736,
      "grad_norm": 2.6979546546936035,
      "learning_rate": 2.542277734895797e-05,
      "loss": 0.7762,
      "step": 8589
    },
    {
      "epoch": 2.7123460688348597,
      "grad_norm": 2.4734864234924316,
      "learning_rate": 2.5419268823240478e-05,
      "loss": 0.6883,
      "step": 8590
    },
    {
      "epoch": 2.712661825071045,
      "grad_norm": 2.566009044647217,
      "learning_rate": 2.5415760297522985e-05,
      "loss": 0.7586,
      "step": 8591
    },
    {
      "epoch": 2.7129775813072308,
      "grad_norm": 2.4710426330566406,
      "learning_rate": 2.541225177180549e-05,
      "loss": 0.6933,
      "step": 8592
    },
    {
      "epoch": 2.7132933375434165,
      "grad_norm": 2.455078125,
      "learning_rate": 2.5408743246087995e-05,
      "loss": 0.7837,
      "step": 8593
    },
    {
      "epoch": 2.7136090937796022,
      "grad_norm": 3.0239169597625732,
      "learning_rate": 2.54052347203705e-05,
      "loss": 0.7604,
      "step": 8594
    },
    {
      "epoch": 2.713924850015788,
      "grad_norm": 2.3072690963745117,
      "learning_rate": 2.540172619465301e-05,
      "loss": 0.8249,
      "step": 8595
    },
    {
      "epoch": 2.7142406062519733,
      "grad_norm": 2.8031017780303955,
      "learning_rate": 2.5398217668935515e-05,
      "loss": 0.8043,
      "step": 8596
    },
    {
      "epoch": 2.714556362488159,
      "grad_norm": 2.375964403152466,
      "learning_rate": 2.5394709143218022e-05,
      "loss": 0.6325,
      "step": 8597
    },
    {
      "epoch": 2.7148721187243448,
      "grad_norm": 2.189552068710327,
      "learning_rate": 2.5391200617500525e-05,
      "loss": 0.6952,
      "step": 8598
    },
    {
      "epoch": 2.7151878749605305,
      "grad_norm": 2.439049005508423,
      "learning_rate": 2.5387692091783032e-05,
      "loss": 0.7034,
      "step": 8599
    },
    {
      "epoch": 2.7155036311967162,
      "grad_norm": 2.404221534729004,
      "learning_rate": 2.538418356606554e-05,
      "loss": 0.7776,
      "step": 8600
    },
    {
      "epoch": 2.7155036311967162,
      "eval_loss": 0.7956727147102356,
      "eval_runtime": 4.5847,
      "eval_samples_per_second": 543.108,
      "eval_steps_per_second": 4.58,
      "step": 8600
    },
    {
      "epoch": 2.715819387432902,
      "grad_norm": 2.398512125015259,
      "learning_rate": 2.5380675040348046e-05,
      "loss": 0.7736,
      "step": 8601
    },
    {
      "epoch": 2.7161351436690877,
      "grad_norm": 2.8060193061828613,
      "learning_rate": 2.5377166514630553e-05,
      "loss": 0.7435,
      "step": 8602
    },
    {
      "epoch": 2.716450899905273,
      "grad_norm": 3.03456974029541,
      "learning_rate": 2.5373657988913056e-05,
      "loss": 0.7862,
      "step": 8603
    },
    {
      "epoch": 2.7167666561414587,
      "grad_norm": 2.3733229637145996,
      "learning_rate": 2.5370149463195563e-05,
      "loss": 0.7359,
      "step": 8604
    },
    {
      "epoch": 2.7170824123776445,
      "grad_norm": 2.8188045024871826,
      "learning_rate": 2.5366640937478077e-05,
      "loss": 0.7399,
      "step": 8605
    },
    {
      "epoch": 2.71739816861383,
      "grad_norm": 2.7934887409210205,
      "learning_rate": 2.5363132411760583e-05,
      "loss": 0.7343,
      "step": 8606
    },
    {
      "epoch": 2.717713924850016,
      "grad_norm": 2.1278321743011475,
      "learning_rate": 2.5359623886043087e-05,
      "loss": 0.6987,
      "step": 8607
    },
    {
      "epoch": 2.7180296810862012,
      "grad_norm": 2.4676947593688965,
      "learning_rate": 2.5356115360325594e-05,
      "loss": 0.6535,
      "step": 8608
    },
    {
      "epoch": 2.718345437322387,
      "grad_norm": 2.270416498184204,
      "learning_rate": 2.53526068346081e-05,
      "loss": 0.7843,
      "step": 8609
    },
    {
      "epoch": 2.7186611935585727,
      "grad_norm": 2.8111677169799805,
      "learning_rate": 2.5349098308890607e-05,
      "loss": 0.7978,
      "step": 8610
    },
    {
      "epoch": 2.7189769497947585,
      "grad_norm": 2.17918062210083,
      "learning_rate": 2.5345589783173114e-05,
      "loss": 0.749,
      "step": 8611
    },
    {
      "epoch": 2.719292706030944,
      "grad_norm": 2.80248761177063,
      "learning_rate": 2.5342081257455617e-05,
      "loss": 0.754,
      "step": 8612
    },
    {
      "epoch": 2.71960846226713,
      "grad_norm": 2.205766201019287,
      "learning_rate": 2.5338572731738124e-05,
      "loss": 0.7003,
      "step": 8613
    },
    {
      "epoch": 2.7199242185033157,
      "grad_norm": 2.294405937194824,
      "learning_rate": 2.533506420602063e-05,
      "loss": 0.6615,
      "step": 8614
    },
    {
      "epoch": 2.720239974739501,
      "grad_norm": 2.2776732444763184,
      "learning_rate": 2.5331555680303138e-05,
      "loss": 0.695,
      "step": 8615
    },
    {
      "epoch": 2.7205557309756867,
      "grad_norm": 2.970366954803467,
      "learning_rate": 2.5328047154585645e-05,
      "loss": 0.7577,
      "step": 8616
    },
    {
      "epoch": 2.7208714872118724,
      "grad_norm": 2.5004851818084717,
      "learning_rate": 2.5324538628868148e-05,
      "loss": 0.7309,
      "step": 8617
    },
    {
      "epoch": 2.721187243448058,
      "grad_norm": 2.4846043586730957,
      "learning_rate": 2.5321030103150655e-05,
      "loss": 0.803,
      "step": 8618
    },
    {
      "epoch": 2.721502999684244,
      "grad_norm": 2.1538872718811035,
      "learning_rate": 2.5317521577433162e-05,
      "loss": 0.7022,
      "step": 8619
    },
    {
      "epoch": 2.721818755920429,
      "grad_norm": 2.5873684883117676,
      "learning_rate": 2.531401305171567e-05,
      "loss": 0.7645,
      "step": 8620
    },
    {
      "epoch": 2.722134512156615,
      "grad_norm": 2.3748691082000732,
      "learning_rate": 2.5310504525998175e-05,
      "loss": 0.7711,
      "step": 8621
    },
    {
      "epoch": 2.7224502683928007,
      "grad_norm": 1.9592511653900146,
      "learning_rate": 2.5306996000280682e-05,
      "loss": 0.7114,
      "step": 8622
    },
    {
      "epoch": 2.7227660246289864,
      "grad_norm": 1.9664400815963745,
      "learning_rate": 2.5303487474563192e-05,
      "loss": 0.7235,
      "step": 8623
    },
    {
      "epoch": 2.723081780865172,
      "grad_norm": 2.4833695888519287,
      "learning_rate": 2.52999789488457e-05,
      "loss": 0.7551,
      "step": 8624
    },
    {
      "epoch": 2.723397537101358,
      "grad_norm": 2.777468681335449,
      "learning_rate": 2.5296470423128206e-05,
      "loss": 0.7785,
      "step": 8625
    },
    {
      "epoch": 2.7237132933375436,
      "grad_norm": 2.7684881687164307,
      "learning_rate": 2.529296189741071e-05,
      "loss": 0.7967,
      "step": 8626
    },
    {
      "epoch": 2.724029049573729,
      "grad_norm": 2.335956335067749,
      "learning_rate": 2.5289453371693216e-05,
      "loss": 0.7587,
      "step": 8627
    },
    {
      "epoch": 2.7243448058099147,
      "grad_norm": 2.145386219024658,
      "learning_rate": 2.5285944845975723e-05,
      "loss": 0.671,
      "step": 8628
    },
    {
      "epoch": 2.7246605620461004,
      "grad_norm": 2.3529231548309326,
      "learning_rate": 2.528243632025823e-05,
      "loss": 0.7641,
      "step": 8629
    },
    {
      "epoch": 2.724976318282286,
      "grad_norm": 2.7828328609466553,
      "learning_rate": 2.5278927794540737e-05,
      "loss": 0.7466,
      "step": 8630
    },
    {
      "epoch": 2.725292074518472,
      "grad_norm": 3.034541130065918,
      "learning_rate": 2.527541926882324e-05,
      "loss": 0.7955,
      "step": 8631
    },
    {
      "epoch": 2.725607830754657,
      "grad_norm": 2.3129239082336426,
      "learning_rate": 2.5271910743105747e-05,
      "loss": 0.6458,
      "step": 8632
    },
    {
      "epoch": 2.725923586990843,
      "grad_norm": 2.6252601146698,
      "learning_rate": 2.5268402217388254e-05,
      "loss": 0.8164,
      "step": 8633
    },
    {
      "epoch": 2.7262393432270287,
      "grad_norm": 2.2651212215423584,
      "learning_rate": 2.526489369167076e-05,
      "loss": 0.6986,
      "step": 8634
    },
    {
      "epoch": 2.7265550994632144,
      "grad_norm": 2.2818198204040527,
      "learning_rate": 2.5261385165953267e-05,
      "loss": 0.6896,
      "step": 8635
    },
    {
      "epoch": 2.7268708556994,
      "grad_norm": 2.518920660018921,
      "learning_rate": 2.5257876640235774e-05,
      "loss": 0.8042,
      "step": 8636
    },
    {
      "epoch": 2.727186611935586,
      "grad_norm": 2.4550328254699707,
      "learning_rate": 2.5254368114518278e-05,
      "loss": 0.7528,
      "step": 8637
    },
    {
      "epoch": 2.7275023681717716,
      "grad_norm": 2.728372097015381,
      "learning_rate": 2.5250859588800784e-05,
      "loss": 0.8267,
      "step": 8638
    },
    {
      "epoch": 2.727818124407957,
      "grad_norm": 2.2185325622558594,
      "learning_rate": 2.524735106308329e-05,
      "loss": 0.7532,
      "step": 8639
    },
    {
      "epoch": 2.7281338806441426,
      "grad_norm": 2.1943140029907227,
      "learning_rate": 2.52438425373658e-05,
      "loss": 0.6462,
      "step": 8640
    },
    {
      "epoch": 2.7284496368803284,
      "grad_norm": 2.4769527912139893,
      "learning_rate": 2.5240334011648308e-05,
      "loss": 0.7883,
      "step": 8641
    },
    {
      "epoch": 2.728765393116514,
      "grad_norm": 2.4144506454467773,
      "learning_rate": 2.5236825485930815e-05,
      "loss": 0.7319,
      "step": 8642
    },
    {
      "epoch": 2.7290811493527,
      "grad_norm": 2.5063633918762207,
      "learning_rate": 2.5233316960213322e-05,
      "loss": 0.7447,
      "step": 8643
    },
    {
      "epoch": 2.729396905588885,
      "grad_norm": 2.1326005458831787,
      "learning_rate": 2.522980843449583e-05,
      "loss": 0.6402,
      "step": 8644
    },
    {
      "epoch": 2.729712661825071,
      "grad_norm": 2.2033376693725586,
      "learning_rate": 2.5226299908778332e-05,
      "loss": 0.7783,
      "step": 8645
    },
    {
      "epoch": 2.7300284180612566,
      "grad_norm": 2.2193806171417236,
      "learning_rate": 2.522279138306084e-05,
      "loss": 0.68,
      "step": 8646
    },
    {
      "epoch": 2.7303441742974424,
      "grad_norm": 2.2848684787750244,
      "learning_rate": 2.5219282857343346e-05,
      "loss": 0.7254,
      "step": 8647
    },
    {
      "epoch": 2.730659930533628,
      "grad_norm": 2.29410719871521,
      "learning_rate": 2.5215774331625853e-05,
      "loss": 0.8182,
      "step": 8648
    },
    {
      "epoch": 2.730975686769814,
      "grad_norm": 2.4109294414520264,
      "learning_rate": 2.521226580590836e-05,
      "loss": 0.7932,
      "step": 8649
    },
    {
      "epoch": 2.7312914430059996,
      "grad_norm": 2.573957920074463,
      "learning_rate": 2.5208757280190866e-05,
      "loss": 0.6973,
      "step": 8650
    },
    {
      "epoch": 2.731607199242185,
      "grad_norm": 2.2554540634155273,
      "learning_rate": 2.520524875447337e-05,
      "loss": 0.7359,
      "step": 8651
    },
    {
      "epoch": 2.7319229554783706,
      "grad_norm": 2.3038408756256104,
      "learning_rate": 2.5201740228755876e-05,
      "loss": 0.6265,
      "step": 8652
    },
    {
      "epoch": 2.7322387117145563,
      "grad_norm": 2.7212741374969482,
      "learning_rate": 2.5198231703038383e-05,
      "loss": 0.7609,
      "step": 8653
    },
    {
      "epoch": 2.732554467950742,
      "grad_norm": 2.5690295696258545,
      "learning_rate": 2.519472317732089e-05,
      "loss": 0.7621,
      "step": 8654
    },
    {
      "epoch": 2.732870224186928,
      "grad_norm": 2.413356304168701,
      "learning_rate": 2.5191214651603397e-05,
      "loss": 0.7343,
      "step": 8655
    },
    {
      "epoch": 2.733185980423113,
      "grad_norm": 2.7224414348602295,
      "learning_rate": 2.51877061258859e-05,
      "loss": 0.7285,
      "step": 8656
    },
    {
      "epoch": 2.7335017366592993,
      "grad_norm": 2.4262540340423584,
      "learning_rate": 2.5184197600168407e-05,
      "loss": 0.8107,
      "step": 8657
    },
    {
      "epoch": 2.7338174928954846,
      "grad_norm": 3.0944371223449707,
      "learning_rate": 2.518068907445092e-05,
      "loss": 0.7489,
      "step": 8658
    },
    {
      "epoch": 2.7341332491316703,
      "grad_norm": 2.8042850494384766,
      "learning_rate": 2.5177180548733424e-05,
      "loss": 0.7585,
      "step": 8659
    },
    {
      "epoch": 2.734449005367856,
      "grad_norm": 2.4174530506134033,
      "learning_rate": 2.517367202301593e-05,
      "loss": 0.7852,
      "step": 8660
    },
    {
      "epoch": 2.734764761604042,
      "grad_norm": 2.34775710105896,
      "learning_rate": 2.5170163497298438e-05,
      "loss": 0.7795,
      "step": 8661
    },
    {
      "epoch": 2.7350805178402275,
      "grad_norm": 2.9504575729370117,
      "learning_rate": 2.5166654971580945e-05,
      "loss": 0.675,
      "step": 8662
    },
    {
      "epoch": 2.735396274076413,
      "grad_norm": 2.3844475746154785,
      "learning_rate": 2.516314644586345e-05,
      "loss": 0.7671,
      "step": 8663
    },
    {
      "epoch": 2.7357120303125986,
      "grad_norm": 2.489434003829956,
      "learning_rate": 2.5159637920145955e-05,
      "loss": 0.6727,
      "step": 8664
    },
    {
      "epoch": 2.7360277865487843,
      "grad_norm": 2.5026707649230957,
      "learning_rate": 2.515612939442846e-05,
      "loss": 0.655,
      "step": 8665
    },
    {
      "epoch": 2.73634354278497,
      "grad_norm": 2.3152899742126465,
      "learning_rate": 2.515262086871097e-05,
      "loss": 0.6952,
      "step": 8666
    },
    {
      "epoch": 2.736659299021156,
      "grad_norm": 2.5510880947113037,
      "learning_rate": 2.5149112342993475e-05,
      "loss": 0.7877,
      "step": 8667
    },
    {
      "epoch": 2.736975055257341,
      "grad_norm": 2.8963582515716553,
      "learning_rate": 2.5145603817275982e-05,
      "loss": 0.8195,
      "step": 8668
    },
    {
      "epoch": 2.7372908114935273,
      "grad_norm": 2.544018030166626,
      "learning_rate": 2.514209529155849e-05,
      "loss": 0.7724,
      "step": 8669
    },
    {
      "epoch": 2.7376065677297126,
      "grad_norm": 2.7148170471191406,
      "learning_rate": 2.5138586765840992e-05,
      "loss": 0.7113,
      "step": 8670
    },
    {
      "epoch": 2.7379223239658983,
      "grad_norm": 2.9701101779937744,
      "learning_rate": 2.51350782401235e-05,
      "loss": 0.7007,
      "step": 8671
    },
    {
      "epoch": 2.738238080202084,
      "grad_norm": 2.9118762016296387,
      "learning_rate": 2.5131569714406006e-05,
      "loss": 0.7945,
      "step": 8672
    },
    {
      "epoch": 2.7385538364382698,
      "grad_norm": 2.6592025756835938,
      "learning_rate": 2.5128061188688513e-05,
      "loss": 0.7595,
      "step": 8673
    },
    {
      "epoch": 2.7388695926744555,
      "grad_norm": 2.375293493270874,
      "learning_rate": 2.512455266297102e-05,
      "loss": 0.7108,
      "step": 8674
    },
    {
      "epoch": 2.739185348910641,
      "grad_norm": 2.306117057800293,
      "learning_rate": 2.512104413725353e-05,
      "loss": 0.751,
      "step": 8675
    },
    {
      "epoch": 2.7395011051468265,
      "grad_norm": 2.542980909347534,
      "learning_rate": 2.5117535611536037e-05,
      "loss": 0.7841,
      "step": 8676
    },
    {
      "epoch": 2.7398168613830123,
      "grad_norm": 2.2063374519348145,
      "learning_rate": 2.5114027085818543e-05,
      "loss": 0.6435,
      "step": 8677
    },
    {
      "epoch": 2.740132617619198,
      "grad_norm": 2.750098943710327,
      "learning_rate": 2.5110518560101047e-05,
      "loss": 0.7986,
      "step": 8678
    },
    {
      "epoch": 2.7404483738553838,
      "grad_norm": 2.7093241214752197,
      "learning_rate": 2.5107010034383554e-05,
      "loss": 0.7553,
      "step": 8679
    },
    {
      "epoch": 2.740764130091569,
      "grad_norm": 2.2277870178222656,
      "learning_rate": 2.510350150866606e-05,
      "loss": 0.6761,
      "step": 8680
    },
    {
      "epoch": 2.7410798863277552,
      "grad_norm": 2.104266881942749,
      "learning_rate": 2.5099992982948567e-05,
      "loss": 0.707,
      "step": 8681
    },
    {
      "epoch": 2.7413956425639405,
      "grad_norm": 2.2818784713745117,
      "learning_rate": 2.5096484457231074e-05,
      "loss": 0.733,
      "step": 8682
    },
    {
      "epoch": 2.7417113988001263,
      "grad_norm": 2.4536054134368896,
      "learning_rate": 2.509297593151358e-05,
      "loss": 0.7664,
      "step": 8683
    },
    {
      "epoch": 2.742027155036312,
      "grad_norm": 2.030329704284668,
      "learning_rate": 2.5089467405796084e-05,
      "loss": 0.7254,
      "step": 8684
    },
    {
      "epoch": 2.7423429112724977,
      "grad_norm": 2.2237117290496826,
      "learning_rate": 2.508595888007859e-05,
      "loss": 0.8133,
      "step": 8685
    },
    {
      "epoch": 2.7426586675086835,
      "grad_norm": 2.4592156410217285,
      "learning_rate": 2.5082450354361098e-05,
      "loss": 0.6966,
      "step": 8686
    },
    {
      "epoch": 2.7429744237448688,
      "grad_norm": 2.2240867614746094,
      "learning_rate": 2.5078941828643605e-05,
      "loss": 0.6877,
      "step": 8687
    },
    {
      "epoch": 2.7432901799810545,
      "grad_norm": 2.406949758529663,
      "learning_rate": 2.507543330292611e-05,
      "loss": 0.8104,
      "step": 8688
    },
    {
      "epoch": 2.7436059362172402,
      "grad_norm": 2.188161849975586,
      "learning_rate": 2.5071924777208615e-05,
      "loss": 0.719,
      "step": 8689
    },
    {
      "epoch": 2.743921692453426,
      "grad_norm": 2.8105578422546387,
      "learning_rate": 2.5068416251491122e-05,
      "loss": 0.6916,
      "step": 8690
    },
    {
      "epoch": 2.7442374486896117,
      "grad_norm": 2.193856716156006,
      "learning_rate": 2.506490772577363e-05,
      "loss": 0.7398,
      "step": 8691
    },
    {
      "epoch": 2.744553204925797,
      "grad_norm": 2.624091625213623,
      "learning_rate": 2.5061399200056135e-05,
      "loss": 0.7626,
      "step": 8692
    },
    {
      "epoch": 2.744868961161983,
      "grad_norm": 2.425983428955078,
      "learning_rate": 2.5057890674338646e-05,
      "loss": 0.7682,
      "step": 8693
    },
    {
      "epoch": 2.7451847173981685,
      "grad_norm": 2.2943856716156006,
      "learning_rate": 2.5054382148621152e-05,
      "loss": 0.7117,
      "step": 8694
    },
    {
      "epoch": 2.7455004736343542,
      "grad_norm": 2.6055188179016113,
      "learning_rate": 2.505087362290366e-05,
      "loss": 0.7502,
      "step": 8695
    },
    {
      "epoch": 2.74581622987054,
      "grad_norm": 2.503326416015625,
      "learning_rate": 2.5047365097186166e-05,
      "loss": 0.6865,
      "step": 8696
    },
    {
      "epoch": 2.7461319861067257,
      "grad_norm": 2.3191077709198,
      "learning_rate": 2.5043856571468673e-05,
      "loss": 0.7405,
      "step": 8697
    },
    {
      "epoch": 2.7464477423429114,
      "grad_norm": 2.343081474304199,
      "learning_rate": 2.5040348045751176e-05,
      "loss": 0.6862,
      "step": 8698
    },
    {
      "epoch": 2.7467634985790967,
      "grad_norm": 2.332277297973633,
      "learning_rate": 2.5036839520033683e-05,
      "loss": 0.7278,
      "step": 8699
    },
    {
      "epoch": 2.7470792548152825,
      "grad_norm": 2.1648943424224854,
      "learning_rate": 2.503333099431619e-05,
      "loss": 0.6972,
      "step": 8700
    },
    {
      "epoch": 2.747395011051468,
      "grad_norm": 2.2179505825042725,
      "learning_rate": 2.5029822468598697e-05,
      "loss": 0.7854,
      "step": 8701
    },
    {
      "epoch": 2.747710767287654,
      "grad_norm": 3.432671308517456,
      "learning_rate": 2.5026313942881203e-05,
      "loss": 0.7611,
      "step": 8702
    },
    {
      "epoch": 2.7480265235238397,
      "grad_norm": 2.3657686710357666,
      "learning_rate": 2.5022805417163707e-05,
      "loss": 0.7446,
      "step": 8703
    },
    {
      "epoch": 2.748342279760025,
      "grad_norm": 2.885667085647583,
      "learning_rate": 2.5019296891446214e-05,
      "loss": 0.7011,
      "step": 8704
    },
    {
      "epoch": 2.748658035996211,
      "grad_norm": 2.196674108505249,
      "learning_rate": 2.501578836572872e-05,
      "loss": 0.7386,
      "step": 8705
    },
    {
      "epoch": 2.7489737922323965,
      "grad_norm": 2.5343081951141357,
      "learning_rate": 2.5012279840011227e-05,
      "loss": 0.7756,
      "step": 8706
    },
    {
      "epoch": 2.749289548468582,
      "grad_norm": 2.3830745220184326,
      "learning_rate": 2.5008771314293734e-05,
      "loss": 0.7365,
      "step": 8707
    },
    {
      "epoch": 2.749605304704768,
      "grad_norm": 2.6204917430877686,
      "learning_rate": 2.5005262788576238e-05,
      "loss": 0.743,
      "step": 8708
    },
    {
      "epoch": 2.7499210609409537,
      "grad_norm": 2.0853540897369385,
      "learning_rate": 2.5001754262858744e-05,
      "loss": 0.6501,
      "step": 8709
    },
    {
      "epoch": 2.7502368171771394,
      "grad_norm": 2.3088619709014893,
      "learning_rate": 2.4998245737141255e-05,
      "loss": 0.7281,
      "step": 8710
    },
    {
      "epoch": 2.7505525734133247,
      "grad_norm": 2.312005043029785,
      "learning_rate": 2.499473721142376e-05,
      "loss": 0.744,
      "step": 8711
    },
    {
      "epoch": 2.7508683296495104,
      "grad_norm": 2.520331621170044,
      "learning_rate": 2.4991228685706268e-05,
      "loss": 0.8332,
      "step": 8712
    },
    {
      "epoch": 2.751184085885696,
      "grad_norm": 2.357869863510132,
      "learning_rate": 2.498772015998877e-05,
      "loss": 0.6781,
      "step": 8713
    },
    {
      "epoch": 2.751499842121882,
      "grad_norm": 2.1748087406158447,
      "learning_rate": 2.4984211634271282e-05,
      "loss": 0.6837,
      "step": 8714
    },
    {
      "epoch": 2.7518155983580677,
      "grad_norm": 2.2851173877716064,
      "learning_rate": 2.498070310855379e-05,
      "loss": 0.6844,
      "step": 8715
    },
    {
      "epoch": 2.7521313545942534,
      "grad_norm": 2.5445568561553955,
      "learning_rate": 2.4977194582836295e-05,
      "loss": 0.6673,
      "step": 8716
    },
    {
      "epoch": 2.752447110830439,
      "grad_norm": 2.7293434143066406,
      "learning_rate": 2.49736860571188e-05,
      "loss": 0.7908,
      "step": 8717
    },
    {
      "epoch": 2.7527628670666244,
      "grad_norm": 2.6352765560150146,
      "learning_rate": 2.4970177531401306e-05,
      "loss": 0.7841,
      "step": 8718
    },
    {
      "epoch": 2.75307862330281,
      "grad_norm": 2.5289971828460693,
      "learning_rate": 2.4966669005683813e-05,
      "loss": 0.6587,
      "step": 8719
    },
    {
      "epoch": 2.753394379538996,
      "grad_norm": 2.0427892208099365,
      "learning_rate": 2.496316047996632e-05,
      "loss": 0.7242,
      "step": 8720
    },
    {
      "epoch": 2.7537101357751816,
      "grad_norm": 2.329047441482544,
      "learning_rate": 2.4959651954248826e-05,
      "loss": 0.7925,
      "step": 8721
    },
    {
      "epoch": 2.7540258920113674,
      "grad_norm": 3.0790438652038574,
      "learning_rate": 2.495614342853133e-05,
      "loss": 0.8827,
      "step": 8722
    },
    {
      "epoch": 2.7543416482475527,
      "grad_norm": 2.3797643184661865,
      "learning_rate": 2.495263490281384e-05,
      "loss": 0.71,
      "step": 8723
    },
    {
      "epoch": 2.7546574044837384,
      "grad_norm": 2.5875489711761475,
      "learning_rate": 2.4949126377096347e-05,
      "loss": 0.7125,
      "step": 8724
    },
    {
      "epoch": 2.754973160719924,
      "grad_norm": 2.4357028007507324,
      "learning_rate": 2.4945617851378853e-05,
      "loss": 0.7288,
      "step": 8725
    },
    {
      "epoch": 2.75528891695611,
      "grad_norm": 2.252009391784668,
      "learning_rate": 2.494210932566136e-05,
      "loss": 0.6754,
      "step": 8726
    },
    {
      "epoch": 2.7556046731922956,
      "grad_norm": 2.3603813648223877,
      "learning_rate": 2.4938600799943864e-05,
      "loss": 0.6657,
      "step": 8727
    },
    {
      "epoch": 2.7559204294284814,
      "grad_norm": 2.4854273796081543,
      "learning_rate": 2.493509227422637e-05,
      "loss": 0.7308,
      "step": 8728
    },
    {
      "epoch": 2.756236185664667,
      "grad_norm": 2.8326845169067383,
      "learning_rate": 2.4931583748508877e-05,
      "loss": 0.7711,
      "step": 8729
    },
    {
      "epoch": 2.7565519419008524,
      "grad_norm": 2.250988006591797,
      "learning_rate": 2.4928075222791384e-05,
      "loss": 0.6926,
      "step": 8730
    },
    {
      "epoch": 2.756867698137038,
      "grad_norm": 2.433713674545288,
      "learning_rate": 2.492456669707389e-05,
      "loss": 0.7683,
      "step": 8731
    },
    {
      "epoch": 2.757183454373224,
      "grad_norm": 2.841822624206543,
      "learning_rate": 2.4921058171356398e-05,
      "loss": 0.6988,
      "step": 8732
    },
    {
      "epoch": 2.7574992106094096,
      "grad_norm": 2.7322611808776855,
      "learning_rate": 2.4917549645638905e-05,
      "loss": 0.7279,
      "step": 8733
    },
    {
      "epoch": 2.7578149668455954,
      "grad_norm": 2.1139185428619385,
      "learning_rate": 2.491404111992141e-05,
      "loss": 0.6313,
      "step": 8734
    },
    {
      "epoch": 2.7581307230817806,
      "grad_norm": 2.6493611335754395,
      "learning_rate": 2.4910532594203918e-05,
      "loss": 0.7574,
      "step": 8735
    },
    {
      "epoch": 2.7584464793179664,
      "grad_norm": 2.32263445854187,
      "learning_rate": 2.490702406848642e-05,
      "loss": 0.7826,
      "step": 8736
    },
    {
      "epoch": 2.758762235554152,
      "grad_norm": 2.3841497898101807,
      "learning_rate": 2.490351554276893e-05,
      "loss": 0.6387,
      "step": 8737
    },
    {
      "epoch": 2.759077991790338,
      "grad_norm": 2.5092151165008545,
      "learning_rate": 2.4900007017051435e-05,
      "loss": 0.779,
      "step": 8738
    },
    {
      "epoch": 2.7593937480265236,
      "grad_norm": 2.3928847312927246,
      "learning_rate": 2.4896498491333942e-05,
      "loss": 0.7652,
      "step": 8739
    },
    {
      "epoch": 2.7597095042627093,
      "grad_norm": 2.833920478820801,
      "learning_rate": 2.489298996561645e-05,
      "loss": 0.7783,
      "step": 8740
    },
    {
      "epoch": 2.760025260498895,
      "grad_norm": 2.7485222816467285,
      "learning_rate": 2.4889481439898956e-05,
      "loss": 0.7726,
      "step": 8741
    },
    {
      "epoch": 2.7603410167350804,
      "grad_norm": 2.4399282932281494,
      "learning_rate": 2.4885972914181462e-05,
      "loss": 0.6898,
      "step": 8742
    },
    {
      "epoch": 2.760656772971266,
      "grad_norm": 2.4301130771636963,
      "learning_rate": 2.488246438846397e-05,
      "loss": 0.7108,
      "step": 8743
    },
    {
      "epoch": 2.760972529207452,
      "grad_norm": 2.469423770904541,
      "learning_rate": 2.4878955862746476e-05,
      "loss": 0.7485,
      "step": 8744
    },
    {
      "epoch": 2.7612882854436376,
      "grad_norm": 2.395707607269287,
      "learning_rate": 2.4875447337028983e-05,
      "loss": 0.6189,
      "step": 8745
    },
    {
      "epoch": 2.7616040416798233,
      "grad_norm": 2.6025943756103516,
      "learning_rate": 2.4871938811311486e-05,
      "loss": 0.7824,
      "step": 8746
    },
    {
      "epoch": 2.7619197979160086,
      "grad_norm": 2.347276210784912,
      "learning_rate": 2.4868430285593993e-05,
      "loss": 0.8175,
      "step": 8747
    },
    {
      "epoch": 2.7622355541521944,
      "grad_norm": 2.467728614807129,
      "learning_rate": 2.48649217598765e-05,
      "loss": 0.7849,
      "step": 8748
    },
    {
      "epoch": 2.76255131038838,
      "grad_norm": 2.591505527496338,
      "learning_rate": 2.486141323415901e-05,
      "loss": 0.8079,
      "step": 8749
    },
    {
      "epoch": 2.762867066624566,
      "grad_norm": 2.633042573928833,
      "learning_rate": 2.4857904708441514e-05,
      "loss": 0.8211,
      "step": 8750
    },
    {
      "epoch": 2.7631828228607516,
      "grad_norm": 2.757920503616333,
      "learning_rate": 2.485439618272402e-05,
      "loss": 0.742,
      "step": 8751
    },
    {
      "epoch": 2.7634985790969373,
      "grad_norm": 2.4431166648864746,
      "learning_rate": 2.4850887657006527e-05,
      "loss": 0.8238,
      "step": 8752
    },
    {
      "epoch": 2.763814335333123,
      "grad_norm": 2.630828857421875,
      "learning_rate": 2.4847379131289034e-05,
      "loss": 0.7966,
      "step": 8753
    },
    {
      "epoch": 2.7641300915693083,
      "grad_norm": 2.1994149684906006,
      "learning_rate": 2.484387060557154e-05,
      "loss": 0.6159,
      "step": 8754
    },
    {
      "epoch": 2.764445847805494,
      "grad_norm": 2.837745189666748,
      "learning_rate": 2.4840362079854048e-05,
      "loss": 0.7576,
      "step": 8755
    },
    {
      "epoch": 2.76476160404168,
      "grad_norm": 2.6749799251556396,
      "learning_rate": 2.483685355413655e-05,
      "loss": 0.6815,
      "step": 8756
    },
    {
      "epoch": 2.7650773602778655,
      "grad_norm": 2.9207582473754883,
      "learning_rate": 2.4833345028419058e-05,
      "loss": 0.8557,
      "step": 8757
    },
    {
      "epoch": 2.7653931165140513,
      "grad_norm": 2.5272884368896484,
      "learning_rate": 2.4829836502701568e-05,
      "loss": 0.7201,
      "step": 8758
    },
    {
      "epoch": 2.7657088727502366,
      "grad_norm": 2.5060853958129883,
      "learning_rate": 2.4826327976984075e-05,
      "loss": 0.6733,
      "step": 8759
    },
    {
      "epoch": 2.7660246289864223,
      "grad_norm": 2.3948781490325928,
      "learning_rate": 2.482281945126658e-05,
      "loss": 0.8658,
      "step": 8760
    },
    {
      "epoch": 2.766340385222608,
      "grad_norm": 2.1983227729797363,
      "learning_rate": 2.4819310925549085e-05,
      "loss": 0.7633,
      "step": 8761
    },
    {
      "epoch": 2.766656141458794,
      "grad_norm": 2.743190288543701,
      "learning_rate": 2.4815802399831592e-05,
      "loss": 0.8011,
      "step": 8762
    },
    {
      "epoch": 2.7669718976949795,
      "grad_norm": 2.474454879760742,
      "learning_rate": 2.48122938741141e-05,
      "loss": 0.7061,
      "step": 8763
    },
    {
      "epoch": 2.7672876539311653,
      "grad_norm": 4.203080177307129,
      "learning_rate": 2.4808785348396606e-05,
      "loss": 0.6815,
      "step": 8764
    },
    {
      "epoch": 2.767603410167351,
      "grad_norm": 2.2117502689361572,
      "learning_rate": 2.480527682267911e-05,
      "loss": 0.7311,
      "step": 8765
    },
    {
      "epoch": 2.7679191664035363,
      "grad_norm": 2.0496182441711426,
      "learning_rate": 2.4801768296961616e-05,
      "loss": 0.7133,
      "step": 8766
    },
    {
      "epoch": 2.768234922639722,
      "grad_norm": 2.7592153549194336,
      "learning_rate": 2.4798259771244126e-05,
      "loss": 0.7737,
      "step": 8767
    },
    {
      "epoch": 2.7685506788759078,
      "grad_norm": 2.286083698272705,
      "learning_rate": 2.4794751245526633e-05,
      "loss": 0.6797,
      "step": 8768
    },
    {
      "epoch": 2.7688664351120935,
      "grad_norm": 2.6083996295928955,
      "learning_rate": 2.4791242719809136e-05,
      "loss": 0.7068,
      "step": 8769
    },
    {
      "epoch": 2.7691821913482793,
      "grad_norm": 2.8557333946228027,
      "learning_rate": 2.4787734194091643e-05,
      "loss": 0.7664,
      "step": 8770
    },
    {
      "epoch": 2.7694979475844645,
      "grad_norm": 2.4113056659698486,
      "learning_rate": 2.478422566837415e-05,
      "loss": 0.7185,
      "step": 8771
    },
    {
      "epoch": 2.7698137038206507,
      "grad_norm": 2.67244029045105,
      "learning_rate": 2.4780717142656657e-05,
      "loss": 0.7123,
      "step": 8772
    },
    {
      "epoch": 2.770129460056836,
      "grad_norm": 2.7794880867004395,
      "learning_rate": 2.4777208616939163e-05,
      "loss": 0.7397,
      "step": 8773
    },
    {
      "epoch": 2.7704452162930218,
      "grad_norm": 2.3524138927459717,
      "learning_rate": 2.477370009122167e-05,
      "loss": 0.6616,
      "step": 8774
    },
    {
      "epoch": 2.7707609725292075,
      "grad_norm": 2.633147716522217,
      "learning_rate": 2.4770191565504174e-05,
      "loss": 0.6742,
      "step": 8775
    },
    {
      "epoch": 2.7710767287653932,
      "grad_norm": 3.0608320236206055,
      "learning_rate": 2.4766683039786684e-05,
      "loss": 0.7752,
      "step": 8776
    },
    {
      "epoch": 2.771392485001579,
      "grad_norm": 2.815498113632202,
      "learning_rate": 2.476317451406919e-05,
      "loss": 0.798,
      "step": 8777
    },
    {
      "epoch": 2.7717082412377643,
      "grad_norm": 2.664598226547241,
      "learning_rate": 2.4759665988351698e-05,
      "loss": 0.7841,
      "step": 8778
    },
    {
      "epoch": 2.77202399747395,
      "grad_norm": 2.42587947845459,
      "learning_rate": 2.47561574626342e-05,
      "loss": 0.7373,
      "step": 8779
    },
    {
      "epoch": 2.7723397537101357,
      "grad_norm": 2.8025503158569336,
      "learning_rate": 2.4752648936916708e-05,
      "loss": 0.7418,
      "step": 8780
    },
    {
      "epoch": 2.7726555099463215,
      "grad_norm": 2.3666610717773438,
      "learning_rate": 2.4749140411199215e-05,
      "loss": 0.6436,
      "step": 8781
    },
    {
      "epoch": 2.772971266182507,
      "grad_norm": 2.3482296466827393,
      "learning_rate": 2.474563188548172e-05,
      "loss": 0.8583,
      "step": 8782
    },
    {
      "epoch": 2.7732870224186925,
      "grad_norm": 2.512878656387329,
      "learning_rate": 2.4742123359764228e-05,
      "loss": 0.752,
      "step": 8783
    },
    {
      "epoch": 2.7736027786548787,
      "grad_norm": 2.1747775077819824,
      "learning_rate": 2.4738614834046735e-05,
      "loss": 0.6523,
      "step": 8784
    },
    {
      "epoch": 2.773918534891064,
      "grad_norm": 2.45847749710083,
      "learning_rate": 2.4735106308329242e-05,
      "loss": 0.7602,
      "step": 8785
    },
    {
      "epoch": 2.7742342911272497,
      "grad_norm": 2.6347579956054688,
      "learning_rate": 2.473159778261175e-05,
      "loss": 0.7324,
      "step": 8786
    },
    {
      "epoch": 2.7745500473634355,
      "grad_norm": 2.189237356185913,
      "learning_rate": 2.4728089256894255e-05,
      "loss": 0.7519,
      "step": 8787
    },
    {
      "epoch": 2.774865803599621,
      "grad_norm": 2.577925443649292,
      "learning_rate": 2.4724580731176762e-05,
      "loss": 0.7435,
      "step": 8788
    },
    {
      "epoch": 2.775181559835807,
      "grad_norm": 2.4004664421081543,
      "learning_rate": 2.4721072205459266e-05,
      "loss": 0.7504,
      "step": 8789
    },
    {
      "epoch": 2.7754973160719922,
      "grad_norm": 5.02599573135376,
      "learning_rate": 2.4717563679741773e-05,
      "loss": 0.7196,
      "step": 8790
    },
    {
      "epoch": 2.775813072308178,
      "grad_norm": 2.6962311267852783,
      "learning_rate": 2.471405515402428e-05,
      "loss": 0.8158,
      "step": 8791
    },
    {
      "epoch": 2.7761288285443637,
      "grad_norm": 2.3097681999206543,
      "learning_rate": 2.4710546628306786e-05,
      "loss": 0.6969,
      "step": 8792
    },
    {
      "epoch": 2.7764445847805495,
      "grad_norm": 2.4262685775756836,
      "learning_rate": 2.4707038102589293e-05,
      "loss": 0.6749,
      "step": 8793
    },
    {
      "epoch": 2.776760341016735,
      "grad_norm": 2.094133138656616,
      "learning_rate": 2.47035295768718e-05,
      "loss": 0.7473,
      "step": 8794
    },
    {
      "epoch": 2.7770760972529205,
      "grad_norm": 2.479565382003784,
      "learning_rate": 2.4700021051154307e-05,
      "loss": 0.8165,
      "step": 8795
    },
    {
      "epoch": 2.7773918534891067,
      "grad_norm": 2.071505069732666,
      "learning_rate": 2.4696512525436813e-05,
      "loss": 0.6354,
      "step": 8796
    },
    {
      "epoch": 2.777707609725292,
      "grad_norm": 2.458650827407837,
      "learning_rate": 2.469300399971932e-05,
      "loss": 0.772,
      "step": 8797
    },
    {
      "epoch": 2.7780233659614777,
      "grad_norm": 2.370053291320801,
      "learning_rate": 2.4689495474001824e-05,
      "loss": 0.7249,
      "step": 8798
    },
    {
      "epoch": 2.7783391221976634,
      "grad_norm": 2.449652671813965,
      "learning_rate": 2.468598694828433e-05,
      "loss": 0.6491,
      "step": 8799
    },
    {
      "epoch": 2.778654878433849,
      "grad_norm": 2.1081035137176514,
      "learning_rate": 2.4682478422566837e-05,
      "loss": 0.7048,
      "step": 8800
    },
    {
      "epoch": 2.778654878433849,
      "eval_loss": 0.7912099361419678,
      "eval_runtime": 4.6307,
      "eval_samples_per_second": 537.721,
      "eval_steps_per_second": 4.535,
      "step": 8800
    },
    {
      "epoch": 2.778970634670035,
      "grad_norm": 2.296736478805542,
      "learning_rate": 2.4678969896849344e-05,
      "loss": 0.7207,
      "step": 8801
    },
    {
      "epoch": 2.77928639090622,
      "grad_norm": 2.2039380073547363,
      "learning_rate": 2.4675461371131854e-05,
      "loss": 0.6917,
      "step": 8802
    },
    {
      "epoch": 2.779602147142406,
      "grad_norm": 2.3588624000549316,
      "learning_rate": 2.4671952845414358e-05,
      "loss": 0.7502,
      "step": 8803
    },
    {
      "epoch": 2.7799179033785917,
      "grad_norm": 2.5736563205718994,
      "learning_rate": 2.4668444319696865e-05,
      "loss": 0.8378,
      "step": 8804
    },
    {
      "epoch": 2.7802336596147774,
      "grad_norm": 2.5594122409820557,
      "learning_rate": 2.466493579397937e-05,
      "loss": 0.7767,
      "step": 8805
    },
    {
      "epoch": 2.780549415850963,
      "grad_norm": 2.234895706176758,
      "learning_rate": 2.4661427268261878e-05,
      "loss": 0.6977,
      "step": 8806
    },
    {
      "epoch": 2.7808651720871485,
      "grad_norm": 2.6796302795410156,
      "learning_rate": 2.4657918742544385e-05,
      "loss": 0.6674,
      "step": 8807
    },
    {
      "epoch": 2.7811809283233346,
      "grad_norm": 2.6774864196777344,
      "learning_rate": 2.465441021682689e-05,
      "loss": 0.7445,
      "step": 8808
    },
    {
      "epoch": 2.78149668455952,
      "grad_norm": 2.2757599353790283,
      "learning_rate": 2.4650901691109395e-05,
      "loss": 0.8044,
      "step": 8809
    },
    {
      "epoch": 2.7818124407957057,
      "grad_norm": 2.5107040405273438,
      "learning_rate": 2.4647393165391902e-05,
      "loss": 0.8126,
      "step": 8810
    },
    {
      "epoch": 2.7821281970318914,
      "grad_norm": 2.7912750244140625,
      "learning_rate": 2.4643884639674412e-05,
      "loss": 0.7611,
      "step": 8811
    },
    {
      "epoch": 2.782443953268077,
      "grad_norm": 2.364062547683716,
      "learning_rate": 2.4640376113956916e-05,
      "loss": 0.6742,
      "step": 8812
    },
    {
      "epoch": 2.782759709504263,
      "grad_norm": 2.605483293533325,
      "learning_rate": 2.4636867588239422e-05,
      "loss": 0.7674,
      "step": 8813
    },
    {
      "epoch": 2.783075465740448,
      "grad_norm": 2.3831043243408203,
      "learning_rate": 2.463335906252193e-05,
      "loss": 0.7232,
      "step": 8814
    },
    {
      "epoch": 2.783391221976634,
      "grad_norm": 2.2862775325775146,
      "learning_rate": 2.4629850536804436e-05,
      "loss": 0.7029,
      "step": 8815
    },
    {
      "epoch": 2.7837069782128196,
      "grad_norm": 2.4856202602386475,
      "learning_rate": 2.4626342011086943e-05,
      "loss": 0.652,
      "step": 8816
    },
    {
      "epoch": 2.7840227344490054,
      "grad_norm": 2.57822585105896,
      "learning_rate": 2.462283348536945e-05,
      "loss": 0.7896,
      "step": 8817
    },
    {
      "epoch": 2.784338490685191,
      "grad_norm": 2.599655866622925,
      "learning_rate": 2.4619324959651953e-05,
      "loss": 0.841,
      "step": 8818
    },
    {
      "epoch": 2.784654246921377,
      "grad_norm": 2.313767433166504,
      "learning_rate": 2.4615816433934463e-05,
      "loss": 0.6485,
      "step": 8819
    },
    {
      "epoch": 2.7849700031575626,
      "grad_norm": 2.2114686965942383,
      "learning_rate": 2.461230790821697e-05,
      "loss": 0.7473,
      "step": 8820
    },
    {
      "epoch": 2.785285759393748,
      "grad_norm": 2.7597157955169678,
      "learning_rate": 2.4608799382499477e-05,
      "loss": 0.7677,
      "step": 8821
    },
    {
      "epoch": 2.7856015156299336,
      "grad_norm": 2.4670064449310303,
      "learning_rate": 2.460529085678198e-05,
      "loss": 0.8122,
      "step": 8822
    },
    {
      "epoch": 2.7859172718661194,
      "grad_norm": 2.353789806365967,
      "learning_rate": 2.4601782331064487e-05,
      "loss": 0.791,
      "step": 8823
    },
    {
      "epoch": 2.786233028102305,
      "grad_norm": 2.3865504264831543,
      "learning_rate": 2.4598273805346994e-05,
      "loss": 0.7396,
      "step": 8824
    },
    {
      "epoch": 2.786548784338491,
      "grad_norm": 2.6431260108947754,
      "learning_rate": 2.45947652796295e-05,
      "loss": 0.7665,
      "step": 8825
    },
    {
      "epoch": 2.786864540574676,
      "grad_norm": 2.4049642086029053,
      "learning_rate": 2.4591256753912008e-05,
      "loss": 0.6759,
      "step": 8826
    },
    {
      "epoch": 2.787180296810862,
      "grad_norm": 2.2359349727630615,
      "learning_rate": 2.458774822819451e-05,
      "loss": 0.6539,
      "step": 8827
    },
    {
      "epoch": 2.7874960530470476,
      "grad_norm": 2.6221604347229004,
      "learning_rate": 2.458423970247702e-05,
      "loss": 0.6745,
      "step": 8828
    },
    {
      "epoch": 2.7878118092832334,
      "grad_norm": 2.606412887573242,
      "learning_rate": 2.4580731176759528e-05,
      "loss": 0.7207,
      "step": 8829
    },
    {
      "epoch": 2.788127565519419,
      "grad_norm": 2.4785702228546143,
      "learning_rate": 2.4577222651042035e-05,
      "loss": 0.7333,
      "step": 8830
    },
    {
      "epoch": 2.788443321755605,
      "grad_norm": 2.0348563194274902,
      "learning_rate": 2.4573714125324542e-05,
      "loss": 0.6143,
      "step": 8831
    },
    {
      "epoch": 2.7887590779917906,
      "grad_norm": 2.686310052871704,
      "learning_rate": 2.4570205599607045e-05,
      "loss": 0.8002,
      "step": 8832
    },
    {
      "epoch": 2.789074834227976,
      "grad_norm": 2.5981104373931885,
      "learning_rate": 2.4566697073889552e-05,
      "loss": 0.6843,
      "step": 8833
    },
    {
      "epoch": 2.7893905904641616,
      "grad_norm": 2.3547494411468506,
      "learning_rate": 2.456318854817206e-05,
      "loss": 0.6207,
      "step": 8834
    },
    {
      "epoch": 2.7897063467003473,
      "grad_norm": 3.112903594970703,
      "learning_rate": 2.4559680022454566e-05,
      "loss": 0.7843,
      "step": 8835
    },
    {
      "epoch": 2.790022102936533,
      "grad_norm": 2.5312821865081787,
      "learning_rate": 2.4556171496737072e-05,
      "loss": 0.7498,
      "step": 8836
    },
    {
      "epoch": 2.790337859172719,
      "grad_norm": 3.325485944747925,
      "learning_rate": 2.455266297101958e-05,
      "loss": 0.7709,
      "step": 8837
    },
    {
      "epoch": 2.790653615408904,
      "grad_norm": 2.409001588821411,
      "learning_rate": 2.4549154445302086e-05,
      "loss": 0.7032,
      "step": 8838
    },
    {
      "epoch": 2.79096937164509,
      "grad_norm": 2.39467453956604,
      "learning_rate": 2.4545645919584593e-05,
      "loss": 0.7078,
      "step": 8839
    },
    {
      "epoch": 2.7912851278812756,
      "grad_norm": 2.6620101928710938,
      "learning_rate": 2.45421373938671e-05,
      "loss": 0.713,
      "step": 8840
    },
    {
      "epoch": 2.7916008841174613,
      "grad_norm": 2.4168310165405273,
      "learning_rate": 2.4538628868149603e-05,
      "loss": 0.6899,
      "step": 8841
    },
    {
      "epoch": 2.791916640353647,
      "grad_norm": 2.6692354679107666,
      "learning_rate": 2.453512034243211e-05,
      "loss": 0.8242,
      "step": 8842
    },
    {
      "epoch": 2.792232396589833,
      "grad_norm": 2.709977626800537,
      "learning_rate": 2.4531611816714617e-05,
      "loss": 0.7671,
      "step": 8843
    },
    {
      "epoch": 2.7925481528260185,
      "grad_norm": 2.7314987182617188,
      "learning_rate": 2.4528103290997123e-05,
      "loss": 0.7775,
      "step": 8844
    },
    {
      "epoch": 2.792863909062204,
      "grad_norm": 2.389169454574585,
      "learning_rate": 2.452459476527963e-05,
      "loss": 0.7264,
      "step": 8845
    },
    {
      "epoch": 2.7931796652983896,
      "grad_norm": 2.3564536571502686,
      "learning_rate": 2.4521086239562137e-05,
      "loss": 0.736,
      "step": 8846
    },
    {
      "epoch": 2.7934954215345753,
      "grad_norm": 2.545783042907715,
      "learning_rate": 2.4517577713844644e-05,
      "loss": 0.7732,
      "step": 8847
    },
    {
      "epoch": 2.793811177770761,
      "grad_norm": 2.345135450363159,
      "learning_rate": 2.451406918812715e-05,
      "loss": 0.6733,
      "step": 8848
    },
    {
      "epoch": 2.794126934006947,
      "grad_norm": 2.394545555114746,
      "learning_rate": 2.4510560662409658e-05,
      "loss": 0.7199,
      "step": 8849
    },
    {
      "epoch": 2.794442690243132,
      "grad_norm": 2.3654820919036865,
      "learning_rate": 2.4507052136692164e-05,
      "loss": 0.7636,
      "step": 8850
    },
    {
      "epoch": 2.794758446479318,
      "grad_norm": 3.3083817958831787,
      "learning_rate": 2.4503543610974668e-05,
      "loss": 0.7646,
      "step": 8851
    },
    {
      "epoch": 2.7950742027155036,
      "grad_norm": 2.414288282394409,
      "learning_rate": 2.4500035085257175e-05,
      "loss": 0.801,
      "step": 8852
    },
    {
      "epoch": 2.7953899589516893,
      "grad_norm": 2.6436994075775146,
      "learning_rate": 2.449652655953968e-05,
      "loss": 0.7581,
      "step": 8853
    },
    {
      "epoch": 2.795705715187875,
      "grad_norm": 3.165316343307495,
      "learning_rate": 2.449301803382219e-05,
      "loss": 0.8418,
      "step": 8854
    },
    {
      "epoch": 2.7960214714240608,
      "grad_norm": 2.770082712173462,
      "learning_rate": 2.4489509508104695e-05,
      "loss": 0.7957,
      "step": 8855
    },
    {
      "epoch": 2.7963372276602465,
      "grad_norm": 2.267373561859131,
      "learning_rate": 2.4486000982387202e-05,
      "loss": 0.6986,
      "step": 8856
    },
    {
      "epoch": 2.796652983896432,
      "grad_norm": 2.080335855484009,
      "learning_rate": 2.448249245666971e-05,
      "loss": 0.7084,
      "step": 8857
    },
    {
      "epoch": 2.7969687401326175,
      "grad_norm": 2.2873733043670654,
      "learning_rate": 2.4478983930952215e-05,
      "loss": 0.7211,
      "step": 8858
    },
    {
      "epoch": 2.7972844963688033,
      "grad_norm": 2.3042309284210205,
      "learning_rate": 2.4475475405234722e-05,
      "loss": 0.7922,
      "step": 8859
    },
    {
      "epoch": 2.797600252604989,
      "grad_norm": 2.2039432525634766,
      "learning_rate": 2.447196687951723e-05,
      "loss": 0.6463,
      "step": 8860
    },
    {
      "epoch": 2.7979160088411748,
      "grad_norm": 2.3461532592773438,
      "learning_rate": 2.4468458353799733e-05,
      "loss": 0.7459,
      "step": 8861
    },
    {
      "epoch": 2.79823176507736,
      "grad_norm": 2.1510777473449707,
      "learning_rate": 2.446494982808224e-05,
      "loss": 0.7191,
      "step": 8862
    },
    {
      "epoch": 2.798547521313546,
      "grad_norm": 2.725400924682617,
      "learning_rate": 2.446144130236475e-05,
      "loss": 0.7235,
      "step": 8863
    },
    {
      "epoch": 2.7988632775497315,
      "grad_norm": 2.3533647060394287,
      "learning_rate": 2.4457932776647256e-05,
      "loss": 0.6882,
      "step": 8864
    },
    {
      "epoch": 2.7991790337859173,
      "grad_norm": 2.2957916259765625,
      "learning_rate": 2.445442425092976e-05,
      "loss": 0.7866,
      "step": 8865
    },
    {
      "epoch": 2.799494790022103,
      "grad_norm": 2.243476152420044,
      "learning_rate": 2.4450915725212267e-05,
      "loss": 0.6916,
      "step": 8866
    },
    {
      "epoch": 2.7998105462582887,
      "grad_norm": 2.1120102405548096,
      "learning_rate": 2.4447407199494773e-05,
      "loss": 0.7727,
      "step": 8867
    },
    {
      "epoch": 2.8001263024944745,
      "grad_norm": 2.2677130699157715,
      "learning_rate": 2.444389867377728e-05,
      "loss": 0.7789,
      "step": 8868
    },
    {
      "epoch": 2.8004420587306598,
      "grad_norm": 2.563581943511963,
      "learning_rate": 2.4440390148059787e-05,
      "loss": 0.8402,
      "step": 8869
    },
    {
      "epoch": 2.8007578149668455,
      "grad_norm": 2.43405818939209,
      "learning_rate": 2.443688162234229e-05,
      "loss": 0.6743,
      "step": 8870
    },
    {
      "epoch": 2.8010735712030312,
      "grad_norm": 2.3496007919311523,
      "learning_rate": 2.4433373096624797e-05,
      "loss": 0.71,
      "step": 8871
    },
    {
      "epoch": 2.801389327439217,
      "grad_norm": 2.3088676929473877,
      "learning_rate": 2.4429864570907307e-05,
      "loss": 0.6929,
      "step": 8872
    },
    {
      "epoch": 2.8017050836754027,
      "grad_norm": 2.44026780128479,
      "learning_rate": 2.4426356045189814e-05,
      "loss": 0.7604,
      "step": 8873
    },
    {
      "epoch": 2.802020839911588,
      "grad_norm": 2.652161121368408,
      "learning_rate": 2.4422847519472318e-05,
      "loss": 0.7335,
      "step": 8874
    },
    {
      "epoch": 2.802336596147774,
      "grad_norm": 2.5153133869171143,
      "learning_rate": 2.4419338993754825e-05,
      "loss": 0.6449,
      "step": 8875
    },
    {
      "epoch": 2.8026523523839595,
      "grad_norm": 2.8711774349212646,
      "learning_rate": 2.441583046803733e-05,
      "loss": 0.7684,
      "step": 8876
    },
    {
      "epoch": 2.8029681086201452,
      "grad_norm": 2.7176833152770996,
      "learning_rate": 2.4412321942319838e-05,
      "loss": 0.7736,
      "step": 8877
    },
    {
      "epoch": 2.803283864856331,
      "grad_norm": 2.3761277198791504,
      "learning_rate": 2.4408813416602345e-05,
      "loss": 0.6704,
      "step": 8878
    },
    {
      "epoch": 2.8035996210925167,
      "grad_norm": 2.9295523166656494,
      "learning_rate": 2.4405304890884852e-05,
      "loss": 0.7478,
      "step": 8879
    },
    {
      "epoch": 2.8039153773287024,
      "grad_norm": 2.578779935836792,
      "learning_rate": 2.440179636516736e-05,
      "loss": 0.7875,
      "step": 8880
    },
    {
      "epoch": 2.8042311335648877,
      "grad_norm": 2.347778558731079,
      "learning_rate": 2.4398287839449865e-05,
      "loss": 0.6922,
      "step": 8881
    },
    {
      "epoch": 2.8045468898010735,
      "grad_norm": 2.7005348205566406,
      "learning_rate": 2.4394779313732372e-05,
      "loss": 0.7595,
      "step": 8882
    },
    {
      "epoch": 2.804862646037259,
      "grad_norm": 2.3141396045684814,
      "learning_rate": 2.439127078801488e-05,
      "loss": 0.731,
      "step": 8883
    },
    {
      "epoch": 2.805178402273445,
      "grad_norm": 2.626385450363159,
      "learning_rate": 2.4387762262297382e-05,
      "loss": 0.8746,
      "step": 8884
    },
    {
      "epoch": 2.8054941585096307,
      "grad_norm": 2.4689745903015137,
      "learning_rate": 2.438425373657989e-05,
      "loss": 0.7645,
      "step": 8885
    },
    {
      "epoch": 2.805809914745816,
      "grad_norm": 2.236253499984741,
      "learning_rate": 2.4380745210862396e-05,
      "loss": 0.7146,
      "step": 8886
    },
    {
      "epoch": 2.806125670982002,
      "grad_norm": 2.6227681636810303,
      "learning_rate": 2.4377236685144903e-05,
      "loss": 0.7953,
      "step": 8887
    },
    {
      "epoch": 2.8064414272181875,
      "grad_norm": 2.477125883102417,
      "learning_rate": 2.437372815942741e-05,
      "loss": 0.6917,
      "step": 8888
    },
    {
      "epoch": 2.806757183454373,
      "grad_norm": 2.4142353534698486,
      "learning_rate": 2.4370219633709917e-05,
      "loss": 0.7516,
      "step": 8889
    },
    {
      "epoch": 2.807072939690559,
      "grad_norm": 2.460064649581909,
      "learning_rate": 2.4366711107992423e-05,
      "loss": 0.7168,
      "step": 8890
    },
    {
      "epoch": 2.8073886959267447,
      "grad_norm": 2.492764949798584,
      "learning_rate": 2.436320258227493e-05,
      "loss": 0.6353,
      "step": 8891
    },
    {
      "epoch": 2.8077044521629304,
      "grad_norm": 2.455078601837158,
      "learning_rate": 2.4359694056557437e-05,
      "loss": 0.8272,
      "step": 8892
    },
    {
      "epoch": 2.8080202083991157,
      "grad_norm": 2.4217610359191895,
      "learning_rate": 2.4356185530839944e-05,
      "loss": 0.741,
      "step": 8893
    },
    {
      "epoch": 2.8083359646353014,
      "grad_norm": 2.352480888366699,
      "learning_rate": 2.4352677005122447e-05,
      "loss": 0.7393,
      "step": 8894
    },
    {
      "epoch": 2.808651720871487,
      "grad_norm": 2.1874117851257324,
      "learning_rate": 2.4349168479404954e-05,
      "loss": 0.7097,
      "step": 8895
    },
    {
      "epoch": 2.808967477107673,
      "grad_norm": 2.7552008628845215,
      "learning_rate": 2.434565995368746e-05,
      "loss": 0.6929,
      "step": 8896
    },
    {
      "epoch": 2.8092832333438587,
      "grad_norm": 2.2107269763946533,
      "learning_rate": 2.4342151427969968e-05,
      "loss": 0.7645,
      "step": 8897
    },
    {
      "epoch": 2.809598989580044,
      "grad_norm": 2.5138235092163086,
      "learning_rate": 2.4338642902252474e-05,
      "loss": 0.7733,
      "step": 8898
    },
    {
      "epoch": 2.80991474581623,
      "grad_norm": 2.408143997192383,
      "learning_rate": 2.433513437653498e-05,
      "loss": 0.7212,
      "step": 8899
    },
    {
      "epoch": 2.8102305020524154,
      "grad_norm": 2.937500238418579,
      "learning_rate": 2.4331625850817488e-05,
      "loss": 0.7614,
      "step": 8900
    },
    {
      "epoch": 2.810546258288601,
      "grad_norm": 2.853072166442871,
      "learning_rate": 2.4328117325099995e-05,
      "loss": 0.7105,
      "step": 8901
    },
    {
      "epoch": 2.810862014524787,
      "grad_norm": 2.4819347858428955,
      "learning_rate": 2.4324608799382502e-05,
      "loss": 0.728,
      "step": 8902
    },
    {
      "epoch": 2.8111777707609726,
      "grad_norm": 2.777508497238159,
      "learning_rate": 2.4321100273665005e-05,
      "loss": 0.7537,
      "step": 8903
    },
    {
      "epoch": 2.8114935269971584,
      "grad_norm": 2.332474708557129,
      "learning_rate": 2.4317591747947512e-05,
      "loss": 0.7214,
      "step": 8904
    },
    {
      "epoch": 2.8118092832333437,
      "grad_norm": 2.53784441947937,
      "learning_rate": 2.431408322223002e-05,
      "loss": 0.7604,
      "step": 8905
    },
    {
      "epoch": 2.8121250394695294,
      "grad_norm": 2.027630567550659,
      "learning_rate": 2.4310574696512526e-05,
      "loss": 0.5952,
      "step": 8906
    },
    {
      "epoch": 2.812440795705715,
      "grad_norm": 3.101142644882202,
      "learning_rate": 2.4307066170795036e-05,
      "loss": 0.7162,
      "step": 8907
    },
    {
      "epoch": 2.812756551941901,
      "grad_norm": 2.692864418029785,
      "learning_rate": 2.430355764507754e-05,
      "loss": 0.7038,
      "step": 8908
    },
    {
      "epoch": 2.8130723081780866,
      "grad_norm": 2.3094639778137207,
      "learning_rate": 2.4300049119360046e-05,
      "loss": 0.6485,
      "step": 8909
    },
    {
      "epoch": 2.813388064414272,
      "grad_norm": 2.46431303024292,
      "learning_rate": 2.4296540593642553e-05,
      "loss": 0.7638,
      "step": 8910
    },
    {
      "epoch": 2.813703820650458,
      "grad_norm": 2.694766044616699,
      "learning_rate": 2.429303206792506e-05,
      "loss": 0.6714,
      "step": 8911
    },
    {
      "epoch": 2.8140195768866434,
      "grad_norm": 3.1222031116485596,
      "learning_rate": 2.4289523542207566e-05,
      "loss": 0.823,
      "step": 8912
    },
    {
      "epoch": 2.814335333122829,
      "grad_norm": 2.5324549674987793,
      "learning_rate": 2.428601501649007e-05,
      "loss": 0.7575,
      "step": 8913
    },
    {
      "epoch": 2.814651089359015,
      "grad_norm": 2.173082113265991,
      "learning_rate": 2.4282506490772577e-05,
      "loss": 0.6504,
      "step": 8914
    },
    {
      "epoch": 2.8149668455952006,
      "grad_norm": 3.3686392307281494,
      "learning_rate": 2.4278997965055087e-05,
      "loss": 0.6886,
      "step": 8915
    },
    {
      "epoch": 2.8152826018313863,
      "grad_norm": 2.6902499198913574,
      "learning_rate": 2.4275489439337594e-05,
      "loss": 0.7241,
      "step": 8916
    },
    {
      "epoch": 2.8155983580675716,
      "grad_norm": 2.458083391189575,
      "learning_rate": 2.4271980913620097e-05,
      "loss": 0.7256,
      "step": 8917
    },
    {
      "epoch": 2.8159141143037574,
      "grad_norm": 2.7331430912017822,
      "learning_rate": 2.4268472387902604e-05,
      "loss": 0.7227,
      "step": 8918
    },
    {
      "epoch": 2.816229870539943,
      "grad_norm": 2.503181219100952,
      "learning_rate": 2.426496386218511e-05,
      "loss": 0.6687,
      "step": 8919
    },
    {
      "epoch": 2.816545626776129,
      "grad_norm": 2.4078495502471924,
      "learning_rate": 2.4261455336467618e-05,
      "loss": 0.658,
      "step": 8920
    },
    {
      "epoch": 2.8168613830123146,
      "grad_norm": 2.3793604373931885,
      "learning_rate": 2.4257946810750124e-05,
      "loss": 0.7445,
      "step": 8921
    },
    {
      "epoch": 2.8171771392485,
      "grad_norm": 2.4649345874786377,
      "learning_rate": 2.425443828503263e-05,
      "loss": 0.6729,
      "step": 8922
    },
    {
      "epoch": 2.817492895484686,
      "grad_norm": 2.4503300189971924,
      "learning_rate": 2.4250929759315135e-05,
      "loss": 0.7754,
      "step": 8923
    },
    {
      "epoch": 2.8178086517208714,
      "grad_norm": 2.644202947616577,
      "learning_rate": 2.4247421233597645e-05,
      "loss": 0.6893,
      "step": 8924
    },
    {
      "epoch": 2.818124407957057,
      "grad_norm": 2.381244659423828,
      "learning_rate": 2.424391270788015e-05,
      "loss": 0.6799,
      "step": 8925
    },
    {
      "epoch": 2.818440164193243,
      "grad_norm": 2.6077303886413574,
      "learning_rate": 2.424040418216266e-05,
      "loss": 0.7921,
      "step": 8926
    },
    {
      "epoch": 2.8187559204294286,
      "grad_norm": 2.9424901008605957,
      "learning_rate": 2.4236895656445162e-05,
      "loss": 0.7211,
      "step": 8927
    },
    {
      "epoch": 2.8190716766656143,
      "grad_norm": 2.347673177719116,
      "learning_rate": 2.423338713072767e-05,
      "loss": 0.6764,
      "step": 8928
    },
    {
      "epoch": 2.8193874329017996,
      "grad_norm": 2.4047913551330566,
      "learning_rate": 2.4229878605010175e-05,
      "loss": 0.7307,
      "step": 8929
    },
    {
      "epoch": 2.8197031891379853,
      "grad_norm": 2.4675450325012207,
      "learning_rate": 2.4226370079292682e-05,
      "loss": 0.7378,
      "step": 8930
    },
    {
      "epoch": 2.820018945374171,
      "grad_norm": 2.2592525482177734,
      "learning_rate": 2.422286155357519e-05,
      "loss": 0.7144,
      "step": 8931
    },
    {
      "epoch": 2.820334701610357,
      "grad_norm": 2.5419437885284424,
      "learning_rate": 2.4219353027857693e-05,
      "loss": 0.6967,
      "step": 8932
    },
    {
      "epoch": 2.8206504578465426,
      "grad_norm": 2.330554485321045,
      "learning_rate": 2.4215844502140203e-05,
      "loss": 0.6425,
      "step": 8933
    },
    {
      "epoch": 2.8209662140827283,
      "grad_norm": 2.614145278930664,
      "learning_rate": 2.421233597642271e-05,
      "loss": 0.8405,
      "step": 8934
    },
    {
      "epoch": 2.821281970318914,
      "grad_norm": 2.3799023628234863,
      "learning_rate": 2.4208827450705216e-05,
      "loss": 0.8213,
      "step": 8935
    },
    {
      "epoch": 2.8215977265550993,
      "grad_norm": 2.498802900314331,
      "learning_rate": 2.4205318924987723e-05,
      "loss": 0.7542,
      "step": 8936
    },
    {
      "epoch": 2.821913482791285,
      "grad_norm": 2.5385384559631348,
      "learning_rate": 2.4201810399270227e-05,
      "loss": 0.7875,
      "step": 8937
    },
    {
      "epoch": 2.822229239027471,
      "grad_norm": 2.2742788791656494,
      "learning_rate": 2.4198301873552733e-05,
      "loss": 0.7094,
      "step": 8938
    },
    {
      "epoch": 2.8225449952636565,
      "grad_norm": 2.648491382598877,
      "learning_rate": 2.419479334783524e-05,
      "loss": 0.6849,
      "step": 8939
    },
    {
      "epoch": 2.8228607514998423,
      "grad_norm": 2.523261547088623,
      "learning_rate": 2.4191284822117747e-05,
      "loss": 0.7399,
      "step": 8940
    },
    {
      "epoch": 2.8231765077360276,
      "grad_norm": 2.395794153213501,
      "learning_rate": 2.4187776296400254e-05,
      "loss": 0.8087,
      "step": 8941
    },
    {
      "epoch": 2.8234922639722133,
      "grad_norm": 2.2648534774780273,
      "learning_rate": 2.418426777068276e-05,
      "loss": 0.6321,
      "step": 8942
    },
    {
      "epoch": 2.823808020208399,
      "grad_norm": 2.7663798332214355,
      "learning_rate": 2.4180759244965267e-05,
      "loss": 0.7633,
      "step": 8943
    },
    {
      "epoch": 2.824123776444585,
      "grad_norm": 2.5008766651153564,
      "learning_rate": 2.4177250719247774e-05,
      "loss": 0.7953,
      "step": 8944
    },
    {
      "epoch": 2.8244395326807705,
      "grad_norm": 2.8454267978668213,
      "learning_rate": 2.417374219353028e-05,
      "loss": 0.763,
      "step": 8945
    },
    {
      "epoch": 2.8247552889169563,
      "grad_norm": 2.494990110397339,
      "learning_rate": 2.4170233667812785e-05,
      "loss": 0.8077,
      "step": 8946
    },
    {
      "epoch": 2.825071045153142,
      "grad_norm": 2.3960824012756348,
      "learning_rate": 2.416672514209529e-05,
      "loss": 0.67,
      "step": 8947
    },
    {
      "epoch": 2.8253868013893273,
      "grad_norm": 2.2051570415496826,
      "learning_rate": 2.4163216616377798e-05,
      "loss": 0.6882,
      "step": 8948
    },
    {
      "epoch": 2.825702557625513,
      "grad_norm": 2.642582654953003,
      "learning_rate": 2.4159708090660305e-05,
      "loss": 0.7547,
      "step": 8949
    },
    {
      "epoch": 2.8260183138616988,
      "grad_norm": 2.318605422973633,
      "learning_rate": 2.4156199564942815e-05,
      "loss": 0.6819,
      "step": 8950
    },
    {
      "epoch": 2.8263340700978845,
      "grad_norm": 2.2740347385406494,
      "learning_rate": 2.415269103922532e-05,
      "loss": 0.775,
      "step": 8951
    },
    {
      "epoch": 2.8266498263340702,
      "grad_norm": 2.4152424335479736,
      "learning_rate": 2.4149182513507825e-05,
      "loss": 0.7808,
      "step": 8952
    },
    {
      "epoch": 2.8269655825702555,
      "grad_norm": 2.7690694332122803,
      "learning_rate": 2.4145673987790332e-05,
      "loss": 0.7788,
      "step": 8953
    },
    {
      "epoch": 2.8272813388064413,
      "grad_norm": 2.729459762573242,
      "learning_rate": 2.414216546207284e-05,
      "loss": 0.7685,
      "step": 8954
    },
    {
      "epoch": 2.827597095042627,
      "grad_norm": 2.6397829055786133,
      "learning_rate": 2.4138656936355346e-05,
      "loss": 0.7262,
      "step": 8955
    },
    {
      "epoch": 2.8279128512788128,
      "grad_norm": 2.6219067573547363,
      "learning_rate": 2.413514841063785e-05,
      "loss": 0.7167,
      "step": 8956
    },
    {
      "epoch": 2.8282286075149985,
      "grad_norm": 2.767320394515991,
      "learning_rate": 2.4131639884920356e-05,
      "loss": 0.7031,
      "step": 8957
    },
    {
      "epoch": 2.8285443637511842,
      "grad_norm": 2.328127861022949,
      "learning_rate": 2.4128131359202863e-05,
      "loss": 0.6439,
      "step": 8958
    },
    {
      "epoch": 2.82886011998737,
      "grad_norm": 2.823274850845337,
      "learning_rate": 2.4124622833485373e-05,
      "loss": 0.7601,
      "step": 8959
    },
    {
      "epoch": 2.8291758762235553,
      "grad_norm": 2.9090325832366943,
      "learning_rate": 2.4121114307767877e-05,
      "loss": 0.7692,
      "step": 8960
    },
    {
      "epoch": 2.829491632459741,
      "grad_norm": 2.4672915935516357,
      "learning_rate": 2.4117605782050383e-05,
      "loss": 0.6949,
      "step": 8961
    },
    {
      "epoch": 2.8298073886959267,
      "grad_norm": 2.3325703144073486,
      "learning_rate": 2.411409725633289e-05,
      "loss": 0.7474,
      "step": 8962
    },
    {
      "epoch": 2.8301231449321125,
      "grad_norm": 2.0751712322235107,
      "learning_rate": 2.4110588730615397e-05,
      "loss": 0.7595,
      "step": 8963
    },
    {
      "epoch": 2.830438901168298,
      "grad_norm": 2.558487892150879,
      "learning_rate": 2.4107080204897904e-05,
      "loss": 0.7789,
      "step": 8964
    },
    {
      "epoch": 2.8307546574044835,
      "grad_norm": 2.3760058879852295,
      "learning_rate": 2.410357167918041e-05,
      "loss": 0.7403,
      "step": 8965
    },
    {
      "epoch": 2.8310704136406692,
      "grad_norm": 3.018310070037842,
      "learning_rate": 2.4100063153462914e-05,
      "loss": 0.7998,
      "step": 8966
    },
    {
      "epoch": 2.831386169876855,
      "grad_norm": 2.477400064468384,
      "learning_rate": 2.409655462774542e-05,
      "loss": 0.7506,
      "step": 8967
    },
    {
      "epoch": 2.8317019261130407,
      "grad_norm": 2.4725255966186523,
      "learning_rate": 2.409304610202793e-05,
      "loss": 0.7317,
      "step": 8968
    },
    {
      "epoch": 2.8320176823492265,
      "grad_norm": 2.2943365573883057,
      "learning_rate": 2.4089537576310438e-05,
      "loss": 0.754,
      "step": 8969
    },
    {
      "epoch": 2.832333438585412,
      "grad_norm": 2.5440425872802734,
      "learning_rate": 2.408602905059294e-05,
      "loss": 0.7346,
      "step": 8970
    },
    {
      "epoch": 2.832649194821598,
      "grad_norm": 2.256920099258423,
      "learning_rate": 2.4082520524875448e-05,
      "loss": 0.8281,
      "step": 8971
    },
    {
      "epoch": 2.8329649510577832,
      "grad_norm": 2.212306261062622,
      "learning_rate": 2.4079011999157955e-05,
      "loss": 0.6963,
      "step": 8972
    },
    {
      "epoch": 2.833280707293969,
      "grad_norm": 2.457930088043213,
      "learning_rate": 2.4075503473440462e-05,
      "loss": 0.6873,
      "step": 8973
    },
    {
      "epoch": 2.8335964635301547,
      "grad_norm": 2.347371816635132,
      "learning_rate": 2.407199494772297e-05,
      "loss": 0.7471,
      "step": 8974
    },
    {
      "epoch": 2.8339122197663404,
      "grad_norm": 2.2389607429504395,
      "learning_rate": 2.4068486422005472e-05,
      "loss": 0.7745,
      "step": 8975
    },
    {
      "epoch": 2.834227976002526,
      "grad_norm": 2.4361109733581543,
      "learning_rate": 2.406497789628798e-05,
      "loss": 0.7622,
      "step": 8976
    },
    {
      "epoch": 2.8345437322387115,
      "grad_norm": 2.478696346282959,
      "learning_rate": 2.406146937057049e-05,
      "loss": 0.6939,
      "step": 8977
    },
    {
      "epoch": 2.834859488474897,
      "grad_norm": 3.2518444061279297,
      "learning_rate": 2.4057960844852996e-05,
      "loss": 0.8243,
      "step": 8978
    },
    {
      "epoch": 2.835175244711083,
      "grad_norm": 2.367558479309082,
      "learning_rate": 2.4054452319135503e-05,
      "loss": 0.6733,
      "step": 8979
    },
    {
      "epoch": 2.8354910009472687,
      "grad_norm": 2.584381103515625,
      "learning_rate": 2.4050943793418006e-05,
      "loss": 0.8255,
      "step": 8980
    },
    {
      "epoch": 2.8358067571834544,
      "grad_norm": 2.4341483116149902,
      "learning_rate": 2.4047435267700513e-05,
      "loss": 0.7089,
      "step": 8981
    },
    {
      "epoch": 2.83612251341964,
      "grad_norm": 2.578202486038208,
      "learning_rate": 2.404392674198302e-05,
      "loss": 0.7417,
      "step": 8982
    },
    {
      "epoch": 2.836438269655826,
      "grad_norm": 2.4207763671875,
      "learning_rate": 2.4040418216265526e-05,
      "loss": 0.7375,
      "step": 8983
    },
    {
      "epoch": 2.836754025892011,
      "grad_norm": 2.401876449584961,
      "learning_rate": 2.4036909690548033e-05,
      "loss": 0.7591,
      "step": 8984
    },
    {
      "epoch": 2.837069782128197,
      "grad_norm": 3.1248977184295654,
      "learning_rate": 2.403340116483054e-05,
      "loss": 0.8066,
      "step": 8985
    },
    {
      "epoch": 2.8373855383643827,
      "grad_norm": 2.577383518218994,
      "learning_rate": 2.4029892639113047e-05,
      "loss": 0.7266,
      "step": 8986
    },
    {
      "epoch": 2.8377012946005684,
      "grad_norm": 2.535968780517578,
      "learning_rate": 2.4026384113395554e-05,
      "loss": 0.7887,
      "step": 8987
    },
    {
      "epoch": 2.838017050836754,
      "grad_norm": 2.162317991256714,
      "learning_rate": 2.402287558767806e-05,
      "loss": 0.6805,
      "step": 8988
    },
    {
      "epoch": 2.8383328070729394,
      "grad_norm": 2.605703115463257,
      "learning_rate": 2.4019367061960564e-05,
      "loss": 0.7094,
      "step": 8989
    },
    {
      "epoch": 2.8386485633091256,
      "grad_norm": 2.21221923828125,
      "learning_rate": 2.401585853624307e-05,
      "loss": 0.6473,
      "step": 8990
    },
    {
      "epoch": 2.838964319545311,
      "grad_norm": 2.223137617111206,
      "learning_rate": 2.4012350010525578e-05,
      "loss": 0.6705,
      "step": 8991
    },
    {
      "epoch": 2.8392800757814967,
      "grad_norm": 3.0717239379882812,
      "learning_rate": 2.4008841484808084e-05,
      "loss": 0.784,
      "step": 8992
    },
    {
      "epoch": 2.8395958320176824,
      "grad_norm": 2.4194369316101074,
      "learning_rate": 2.400533295909059e-05,
      "loss": 0.8359,
      "step": 8993
    },
    {
      "epoch": 2.839911588253868,
      "grad_norm": 2.421983003616333,
      "learning_rate": 2.4001824433373098e-05,
      "loss": 0.7049,
      "step": 8994
    },
    {
      "epoch": 2.840227344490054,
      "grad_norm": 2.5765492916107178,
      "learning_rate": 2.3998315907655605e-05,
      "loss": 0.7329,
      "step": 8995
    },
    {
      "epoch": 2.840543100726239,
      "grad_norm": 2.5019514560699463,
      "learning_rate": 2.399480738193811e-05,
      "loss": 0.7795,
      "step": 8996
    },
    {
      "epoch": 2.840858856962425,
      "grad_norm": 2.203559160232544,
      "learning_rate": 2.399129885622062e-05,
      "loss": 0.6959,
      "step": 8997
    },
    {
      "epoch": 2.8411746131986106,
      "grad_norm": 2.4113969802856445,
      "learning_rate": 2.3987790330503125e-05,
      "loss": 0.6095,
      "step": 8998
    },
    {
      "epoch": 2.8414903694347964,
      "grad_norm": 2.355125665664673,
      "learning_rate": 2.398428180478563e-05,
      "loss": 0.6705,
      "step": 8999
    },
    {
      "epoch": 2.841806125670982,
      "grad_norm": 2.4890828132629395,
      "learning_rate": 2.3980773279068136e-05,
      "loss": 0.6209,
      "step": 9000
    },
    {
      "epoch": 2.841806125670982,
      "eval_loss": 0.7833516597747803,
      "eval_runtime": 4.5893,
      "eval_samples_per_second": 542.563,
      "eval_steps_per_second": 4.576,
      "step": 9000
    },
    {
      "epoch": 2.8421218819071674,
      "grad_norm": 2.4989349842071533,
      "learning_rate": 2.3977264753350642e-05,
      "loss": 0.7043,
      "step": 9001
    },
    {
      "epoch": 2.8424376381433536,
      "grad_norm": 2.4280457496643066,
      "learning_rate": 2.397375622763315e-05,
      "loss": 0.7647,
      "step": 9002
    },
    {
      "epoch": 2.842753394379539,
      "grad_norm": 2.5729446411132812,
      "learning_rate": 2.3970247701915656e-05,
      "loss": 0.7829,
      "step": 9003
    },
    {
      "epoch": 2.8430691506157246,
      "grad_norm": 2.474022626876831,
      "learning_rate": 2.3966739176198163e-05,
      "loss": 0.8216,
      "step": 9004
    },
    {
      "epoch": 2.8433849068519104,
      "grad_norm": 2.2606236934661865,
      "learning_rate": 2.396323065048067e-05,
      "loss": 0.7046,
      "step": 9005
    },
    {
      "epoch": 2.843700663088096,
      "grad_norm": 2.3775739669799805,
      "learning_rate": 2.3959722124763176e-05,
      "loss": 0.7554,
      "step": 9006
    },
    {
      "epoch": 2.844016419324282,
      "grad_norm": 2.368159532546997,
      "learning_rate": 2.3956213599045683e-05,
      "loss": 0.6395,
      "step": 9007
    },
    {
      "epoch": 2.844332175560467,
      "grad_norm": 2.396737813949585,
      "learning_rate": 2.3952705073328187e-05,
      "loss": 0.7313,
      "step": 9008
    },
    {
      "epoch": 2.844647931796653,
      "grad_norm": 2.358311414718628,
      "learning_rate": 2.3949196547610693e-05,
      "loss": 0.8286,
      "step": 9009
    },
    {
      "epoch": 2.8449636880328386,
      "grad_norm": 2.9236340522766113,
      "learning_rate": 2.39456880218932e-05,
      "loss": 0.7509,
      "step": 9010
    },
    {
      "epoch": 2.8452794442690243,
      "grad_norm": 2.6499338150024414,
      "learning_rate": 2.3942179496175707e-05,
      "loss": 0.7413,
      "step": 9011
    },
    {
      "epoch": 2.84559520050521,
      "grad_norm": 2.876486301422119,
      "learning_rate": 2.3938670970458217e-05,
      "loss": 0.7794,
      "step": 9012
    },
    {
      "epoch": 2.8459109567413954,
      "grad_norm": 2.617595672607422,
      "learning_rate": 2.393516244474072e-05,
      "loss": 0.7766,
      "step": 9013
    },
    {
      "epoch": 2.8462267129775816,
      "grad_norm": 2.9632253646850586,
      "learning_rate": 2.3931653919023227e-05,
      "loss": 0.7634,
      "step": 9014
    },
    {
      "epoch": 2.846542469213767,
      "grad_norm": 2.6914446353912354,
      "learning_rate": 2.3928145393305734e-05,
      "loss": 0.6899,
      "step": 9015
    },
    {
      "epoch": 2.8468582254499526,
      "grad_norm": 2.57963490486145,
      "learning_rate": 2.392463686758824e-05,
      "loss": 0.7794,
      "step": 9016
    },
    {
      "epoch": 2.8471739816861383,
      "grad_norm": 7.857200622558594,
      "learning_rate": 2.3921128341870748e-05,
      "loss": 0.6928,
      "step": 9017
    },
    {
      "epoch": 2.847489737922324,
      "grad_norm": 2.553574562072754,
      "learning_rate": 2.391761981615325e-05,
      "loss": 0.8373,
      "step": 9018
    },
    {
      "epoch": 2.84780549415851,
      "grad_norm": 2.5766875743865967,
      "learning_rate": 2.3914111290435758e-05,
      "loss": 0.7148,
      "step": 9019
    },
    {
      "epoch": 2.848121250394695,
      "grad_norm": 2.472838878631592,
      "learning_rate": 2.391060276471827e-05,
      "loss": 0.8255,
      "step": 9020
    },
    {
      "epoch": 2.848437006630881,
      "grad_norm": 2.626985549926758,
      "learning_rate": 2.3907094239000775e-05,
      "loss": 0.7698,
      "step": 9021
    },
    {
      "epoch": 2.8487527628670666,
      "grad_norm": 2.293623208999634,
      "learning_rate": 2.390358571328328e-05,
      "loss": 0.7387,
      "step": 9022
    },
    {
      "epoch": 2.8490685191032523,
      "grad_norm": 2.453278064727783,
      "learning_rate": 2.3900077187565785e-05,
      "loss": 0.788,
      "step": 9023
    },
    {
      "epoch": 2.849384275339438,
      "grad_norm": 2.4561777114868164,
      "learning_rate": 2.3896568661848292e-05,
      "loss": 0.7135,
      "step": 9024
    },
    {
      "epoch": 2.8497000315756233,
      "grad_norm": 3.044234037399292,
      "learning_rate": 2.38930601361308e-05,
      "loss": 0.7428,
      "step": 9025
    },
    {
      "epoch": 2.8500157878118095,
      "grad_norm": 2.651215076446533,
      "learning_rate": 2.3889551610413306e-05,
      "loss": 0.8436,
      "step": 9026
    },
    {
      "epoch": 2.850331544047995,
      "grad_norm": 2.34616756439209,
      "learning_rate": 2.3886043084695813e-05,
      "loss": 0.7653,
      "step": 9027
    },
    {
      "epoch": 2.8506473002841806,
      "grad_norm": 2.405510187149048,
      "learning_rate": 2.3882534558978316e-05,
      "loss": 0.7054,
      "step": 9028
    },
    {
      "epoch": 2.8509630565203663,
      "grad_norm": 2.861241102218628,
      "learning_rate": 2.3879026033260826e-05,
      "loss": 0.8119,
      "step": 9029
    },
    {
      "epoch": 2.851278812756552,
      "grad_norm": 2.97275710105896,
      "learning_rate": 2.3875517507543333e-05,
      "loss": 0.7949,
      "step": 9030
    },
    {
      "epoch": 2.8515945689927378,
      "grad_norm": 2.378232002258301,
      "learning_rate": 2.387200898182584e-05,
      "loss": 0.7618,
      "step": 9031
    },
    {
      "epoch": 2.851910325228923,
      "grad_norm": 2.911041021347046,
      "learning_rate": 2.3868500456108343e-05,
      "loss": 0.6554,
      "step": 9032
    },
    {
      "epoch": 2.852226081465109,
      "grad_norm": 2.8265836238861084,
      "learning_rate": 2.386499193039085e-05,
      "loss": 0.7967,
      "step": 9033
    },
    {
      "epoch": 2.8525418377012945,
      "grad_norm": 2.673950672149658,
      "learning_rate": 2.3861483404673357e-05,
      "loss": 0.7297,
      "step": 9034
    },
    {
      "epoch": 2.8528575939374803,
      "grad_norm": 2.6463687419891357,
      "learning_rate": 2.3857974878955864e-05,
      "loss": 0.7837,
      "step": 9035
    },
    {
      "epoch": 2.853173350173666,
      "grad_norm": 2.543220043182373,
      "learning_rate": 2.385446635323837e-05,
      "loss": 0.6987,
      "step": 9036
    },
    {
      "epoch": 2.8534891064098518,
      "grad_norm": 2.3452885150909424,
      "learning_rate": 2.3850957827520874e-05,
      "loss": 0.721,
      "step": 9037
    },
    {
      "epoch": 2.8538048626460375,
      "grad_norm": 2.3449172973632812,
      "learning_rate": 2.3847449301803384e-05,
      "loss": 0.758,
      "step": 9038
    },
    {
      "epoch": 2.854120618882223,
      "grad_norm": 1.89924156665802,
      "learning_rate": 2.384394077608589e-05,
      "loss": 0.667,
      "step": 9039
    },
    {
      "epoch": 2.8544363751184085,
      "grad_norm": 2.5300891399383545,
      "learning_rate": 2.3840432250368398e-05,
      "loss": 0.7355,
      "step": 9040
    },
    {
      "epoch": 2.8547521313545943,
      "grad_norm": 2.2361464500427246,
      "learning_rate": 2.3836923724650905e-05,
      "loss": 0.8082,
      "step": 9041
    },
    {
      "epoch": 2.85506788759078,
      "grad_norm": 2.3654205799102783,
      "learning_rate": 2.3833415198933408e-05,
      "loss": 0.6949,
      "step": 9042
    },
    {
      "epoch": 2.8553836438269657,
      "grad_norm": 2.3401570320129395,
      "learning_rate": 2.3829906673215915e-05,
      "loss": 0.7374,
      "step": 9043
    },
    {
      "epoch": 2.855699400063151,
      "grad_norm": 2.6891989707946777,
      "learning_rate": 2.3826398147498422e-05,
      "loss": 0.7592,
      "step": 9044
    },
    {
      "epoch": 2.8560151562993368,
      "grad_norm": 2.588395357131958,
      "learning_rate": 2.382288962178093e-05,
      "loss": 0.6355,
      "step": 9045
    },
    {
      "epoch": 2.8563309125355225,
      "grad_norm": 2.9519832134246826,
      "learning_rate": 2.3819381096063435e-05,
      "loss": 0.7265,
      "step": 9046
    },
    {
      "epoch": 2.8566466687717083,
      "grad_norm": 2.585479974746704,
      "learning_rate": 2.3815872570345942e-05,
      "loss": 0.7085,
      "step": 9047
    },
    {
      "epoch": 2.856962425007894,
      "grad_norm": 2.6206698417663574,
      "learning_rate": 2.381236404462845e-05,
      "loss": 0.7757,
      "step": 9048
    },
    {
      "epoch": 2.8572781812440797,
      "grad_norm": 2.6471076011657715,
      "learning_rate": 2.3808855518910956e-05,
      "loss": 0.7353,
      "step": 9049
    },
    {
      "epoch": 2.8575939374802655,
      "grad_norm": 2.4812021255493164,
      "learning_rate": 2.3805346993193463e-05,
      "loss": 0.7523,
      "step": 9050
    },
    {
      "epoch": 2.8579096937164508,
      "grad_norm": 2.540060043334961,
      "learning_rate": 2.3801838467475966e-05,
      "loss": 0.779,
      "step": 9051
    },
    {
      "epoch": 2.8582254499526365,
      "grad_norm": 2.7148373126983643,
      "learning_rate": 2.3798329941758473e-05,
      "loss": 0.7406,
      "step": 9052
    },
    {
      "epoch": 2.8585412061888222,
      "grad_norm": 2.5345616340637207,
      "learning_rate": 2.379482141604098e-05,
      "loss": 0.7239,
      "step": 9053
    },
    {
      "epoch": 2.858856962425008,
      "grad_norm": 2.2264046669006348,
      "learning_rate": 2.3791312890323486e-05,
      "loss": 0.6774,
      "step": 9054
    },
    {
      "epoch": 2.8591727186611937,
      "grad_norm": 2.5494492053985596,
      "learning_rate": 2.3787804364605997e-05,
      "loss": 0.6944,
      "step": 9055
    },
    {
      "epoch": 2.859488474897379,
      "grad_norm": 2.7544848918914795,
      "learning_rate": 2.37842958388885e-05,
      "loss": 0.674,
      "step": 9056
    },
    {
      "epoch": 2.8598042311335647,
      "grad_norm": 2.91160249710083,
      "learning_rate": 2.3780787313171007e-05,
      "loss": 0.6667,
      "step": 9057
    },
    {
      "epoch": 2.8601199873697505,
      "grad_norm": 2.74723482131958,
      "learning_rate": 2.3777278787453514e-05,
      "loss": 0.7335,
      "step": 9058
    },
    {
      "epoch": 2.860435743605936,
      "grad_norm": 2.6109490394592285,
      "learning_rate": 2.377377026173602e-05,
      "loss": 0.7331,
      "step": 9059
    },
    {
      "epoch": 2.860751499842122,
      "grad_norm": 2.697392225265503,
      "learning_rate": 2.3770261736018527e-05,
      "loss": 0.7574,
      "step": 9060
    },
    {
      "epoch": 2.8610672560783077,
      "grad_norm": 3.0494914054870605,
      "learning_rate": 2.376675321030103e-05,
      "loss": 0.7804,
      "step": 9061
    },
    {
      "epoch": 2.8613830123144934,
      "grad_norm": 2.207876682281494,
      "learning_rate": 2.3763244684583538e-05,
      "loss": 0.6927,
      "step": 9062
    },
    {
      "epoch": 2.8616987685506787,
      "grad_norm": 2.6762564182281494,
      "learning_rate": 2.3759736158866044e-05,
      "loss": 0.8405,
      "step": 9063
    },
    {
      "epoch": 2.8620145247868645,
      "grad_norm": 2.9363458156585693,
      "learning_rate": 2.3756227633148555e-05,
      "loss": 0.7984,
      "step": 9064
    },
    {
      "epoch": 2.86233028102305,
      "grad_norm": 2.596621036529541,
      "learning_rate": 2.3752719107431058e-05,
      "loss": 0.6975,
      "step": 9065
    },
    {
      "epoch": 2.862646037259236,
      "grad_norm": 2.6599295139312744,
      "learning_rate": 2.3749210581713565e-05,
      "loss": 0.7218,
      "step": 9066
    },
    {
      "epoch": 2.8629617934954217,
      "grad_norm": 2.590156078338623,
      "learning_rate": 2.374570205599607e-05,
      "loss": 0.7386,
      "step": 9067
    },
    {
      "epoch": 2.863277549731607,
      "grad_norm": 3.080420970916748,
      "learning_rate": 2.374219353027858e-05,
      "loss": 0.7504,
      "step": 9068
    },
    {
      "epoch": 2.8635933059677927,
      "grad_norm": 2.4294815063476562,
      "learning_rate": 2.3738685004561085e-05,
      "loss": 0.6769,
      "step": 9069
    },
    {
      "epoch": 2.8639090622039785,
      "grad_norm": 2.6496896743774414,
      "learning_rate": 2.3735176478843592e-05,
      "loss": 0.6816,
      "step": 9070
    },
    {
      "epoch": 2.864224818440164,
      "grad_norm": 2.854438304901123,
      "learning_rate": 2.3731667953126096e-05,
      "loss": 0.693,
      "step": 9071
    },
    {
      "epoch": 2.86454057467635,
      "grad_norm": 2.3079705238342285,
      "learning_rate": 2.3728159427408602e-05,
      "loss": 0.7245,
      "step": 9072
    },
    {
      "epoch": 2.8648563309125357,
      "grad_norm": 2.639106512069702,
      "learning_rate": 2.3724650901691113e-05,
      "loss": 0.7821,
      "step": 9073
    },
    {
      "epoch": 2.8651720871487214,
      "grad_norm": 2.5950684547424316,
      "learning_rate": 2.372114237597362e-05,
      "loss": 0.6854,
      "step": 9074
    },
    {
      "epoch": 2.8654878433849067,
      "grad_norm": 2.776165008544922,
      "learning_rate": 2.3717633850256123e-05,
      "loss": 0.7652,
      "step": 9075
    },
    {
      "epoch": 2.8658035996210924,
      "grad_norm": 3.3803415298461914,
      "learning_rate": 2.371412532453863e-05,
      "loss": 0.8326,
      "step": 9076
    },
    {
      "epoch": 2.866119355857278,
      "grad_norm": 3.002110481262207,
      "learning_rate": 2.3710616798821136e-05,
      "loss": 0.7902,
      "step": 9077
    },
    {
      "epoch": 2.866435112093464,
      "grad_norm": 2.8484630584716797,
      "learning_rate": 2.3707108273103643e-05,
      "loss": 0.8342,
      "step": 9078
    },
    {
      "epoch": 2.8667508683296496,
      "grad_norm": 2.587928533554077,
      "learning_rate": 2.370359974738615e-05,
      "loss": 0.7667,
      "step": 9079
    },
    {
      "epoch": 2.867066624565835,
      "grad_norm": 2.7954626083374023,
      "learning_rate": 2.3700091221668653e-05,
      "loss": 0.7354,
      "step": 9080
    },
    {
      "epoch": 2.8673823808020207,
      "grad_norm": 2.19822096824646,
      "learning_rate": 2.369658269595116e-05,
      "loss": 0.6712,
      "step": 9081
    },
    {
      "epoch": 2.8676981370382064,
      "grad_norm": 2.5663533210754395,
      "learning_rate": 2.369307417023367e-05,
      "loss": 0.7303,
      "step": 9082
    },
    {
      "epoch": 2.868013893274392,
      "grad_norm": 2.4648401737213135,
      "learning_rate": 2.3689565644516177e-05,
      "loss": 0.7113,
      "step": 9083
    },
    {
      "epoch": 2.868329649510578,
      "grad_norm": 2.5340054035186768,
      "learning_rate": 2.3686057118798684e-05,
      "loss": 0.7172,
      "step": 9084
    },
    {
      "epoch": 2.8686454057467636,
      "grad_norm": 3.129338264465332,
      "learning_rate": 2.3682548593081188e-05,
      "loss": 0.7477,
      "step": 9085
    },
    {
      "epoch": 2.8689611619829494,
      "grad_norm": 2.723435878753662,
      "learning_rate": 2.3679040067363694e-05,
      "loss": 0.672,
      "step": 9086
    },
    {
      "epoch": 2.8692769182191347,
      "grad_norm": 2.4318511486053467,
      "learning_rate": 2.36755315416462e-05,
      "loss": 0.698,
      "step": 9087
    },
    {
      "epoch": 2.8695926744553204,
      "grad_norm": 2.725173234939575,
      "learning_rate": 2.3672023015928708e-05,
      "loss": 0.7792,
      "step": 9088
    },
    {
      "epoch": 2.869908430691506,
      "grad_norm": 3.406689405441284,
      "learning_rate": 2.3668514490211215e-05,
      "loss": 0.7252,
      "step": 9089
    },
    {
      "epoch": 2.870224186927692,
      "grad_norm": 2.4025793075561523,
      "learning_rate": 2.366500596449372e-05,
      "loss": 0.7546,
      "step": 9090
    },
    {
      "epoch": 2.8705399431638776,
      "grad_norm": 2.727773427963257,
      "learning_rate": 2.366149743877623e-05,
      "loss": 0.6824,
      "step": 9091
    },
    {
      "epoch": 2.870855699400063,
      "grad_norm": 2.7964718341827393,
      "learning_rate": 2.3657988913058735e-05,
      "loss": 0.7388,
      "step": 9092
    },
    {
      "epoch": 2.871171455636249,
      "grad_norm": 2.9188055992126465,
      "learning_rate": 2.3654480387341242e-05,
      "loss": 0.8156,
      "step": 9093
    },
    {
      "epoch": 2.8714872118724344,
      "grad_norm": 2.79158878326416,
      "learning_rate": 2.3650971861623745e-05,
      "loss": 0.7932,
      "step": 9094
    },
    {
      "epoch": 2.87180296810862,
      "grad_norm": 2.632347583770752,
      "learning_rate": 2.3647463335906252e-05,
      "loss": 0.8103,
      "step": 9095
    },
    {
      "epoch": 2.872118724344806,
      "grad_norm": 2.4270901679992676,
      "learning_rate": 2.364395481018876e-05,
      "loss": 0.7288,
      "step": 9096
    },
    {
      "epoch": 2.8724344805809916,
      "grad_norm": 2.594613552093506,
      "learning_rate": 2.3640446284471266e-05,
      "loss": 0.7873,
      "step": 9097
    },
    {
      "epoch": 2.8727502368171773,
      "grad_norm": 2.7486307621002197,
      "learning_rate": 2.3636937758753773e-05,
      "loss": 0.8304,
      "step": 9098
    },
    {
      "epoch": 2.8730659930533626,
      "grad_norm": 2.4536545276641846,
      "learning_rate": 2.363342923303628e-05,
      "loss": 0.8002,
      "step": 9099
    },
    {
      "epoch": 2.8733817492895484,
      "grad_norm": 2.6922390460968018,
      "learning_rate": 2.3629920707318786e-05,
      "loss": 0.7485,
      "step": 9100
    },
    {
      "epoch": 2.873697505525734,
      "grad_norm": 2.3595759868621826,
      "learning_rate": 2.3626412181601293e-05,
      "loss": 0.7255,
      "step": 9101
    },
    {
      "epoch": 2.87401326176192,
      "grad_norm": 2.7803802490234375,
      "learning_rate": 2.36229036558838e-05,
      "loss": 0.7552,
      "step": 9102
    },
    {
      "epoch": 2.8743290179981056,
      "grad_norm": 3.0408644676208496,
      "learning_rate": 2.3619395130166307e-05,
      "loss": 0.769,
      "step": 9103
    },
    {
      "epoch": 2.874644774234291,
      "grad_norm": 2.5662307739257812,
      "learning_rate": 2.361588660444881e-05,
      "loss": 0.7344,
      "step": 9104
    },
    {
      "epoch": 2.874960530470477,
      "grad_norm": 2.4041924476623535,
      "learning_rate": 2.3612378078731317e-05,
      "loss": 0.7125,
      "step": 9105
    },
    {
      "epoch": 2.8752762867066624,
      "grad_norm": 2.488541841506958,
      "learning_rate": 2.3608869553013824e-05,
      "loss": 0.7547,
      "step": 9106
    },
    {
      "epoch": 2.875592042942848,
      "grad_norm": 2.954252004623413,
      "learning_rate": 2.360536102729633e-05,
      "loss": 0.8089,
      "step": 9107
    },
    {
      "epoch": 2.875907799179034,
      "grad_norm": 2.7843997478485107,
      "learning_rate": 2.3601852501578837e-05,
      "loss": 0.8501,
      "step": 9108
    },
    {
      "epoch": 2.8762235554152196,
      "grad_norm": 2.378739833831787,
      "learning_rate": 2.3598343975861344e-05,
      "loss": 0.7002,
      "step": 9109
    },
    {
      "epoch": 2.8765393116514053,
      "grad_norm": 2.599228620529175,
      "learning_rate": 2.359483545014385e-05,
      "loss": 0.749,
      "step": 9110
    },
    {
      "epoch": 2.8768550678875906,
      "grad_norm": 2.9740164279937744,
      "learning_rate": 2.3591326924426358e-05,
      "loss": 0.7097,
      "step": 9111
    },
    {
      "epoch": 2.8771708241237763,
      "grad_norm": 2.5670740604400635,
      "learning_rate": 2.3587818398708865e-05,
      "loss": 0.7151,
      "step": 9112
    },
    {
      "epoch": 2.877486580359962,
      "grad_norm": 2.5880842208862305,
      "learning_rate": 2.3584309872991368e-05,
      "loss": 0.7656,
      "step": 9113
    },
    {
      "epoch": 2.877802336596148,
      "grad_norm": 2.361635208129883,
      "learning_rate": 2.3580801347273875e-05,
      "loss": 0.7347,
      "step": 9114
    },
    {
      "epoch": 2.8781180928323336,
      "grad_norm": 2.8691399097442627,
      "learning_rate": 2.3577292821556382e-05,
      "loss": 0.6902,
      "step": 9115
    },
    {
      "epoch": 2.878433849068519,
      "grad_norm": 2.683009386062622,
      "learning_rate": 2.357378429583889e-05,
      "loss": 0.7006,
      "step": 9116
    },
    {
      "epoch": 2.878749605304705,
      "grad_norm": 2.414806842803955,
      "learning_rate": 2.35702757701214e-05,
      "loss": 0.6914,
      "step": 9117
    },
    {
      "epoch": 2.8790653615408903,
      "grad_norm": 2.3658559322357178,
      "learning_rate": 2.3566767244403902e-05,
      "loss": 0.6464,
      "step": 9118
    },
    {
      "epoch": 2.879381117777076,
      "grad_norm": 2.353794813156128,
      "learning_rate": 2.356325871868641e-05,
      "loss": 0.7974,
      "step": 9119
    },
    {
      "epoch": 2.879696874013262,
      "grad_norm": 2.2635226249694824,
      "learning_rate": 2.3559750192968916e-05,
      "loss": 0.7855,
      "step": 9120
    },
    {
      "epoch": 2.8800126302494475,
      "grad_norm": 2.3649823665618896,
      "learning_rate": 2.3556241667251423e-05,
      "loss": 0.7244,
      "step": 9121
    },
    {
      "epoch": 2.8803283864856333,
      "grad_norm": 2.5839126110076904,
      "learning_rate": 2.355273314153393e-05,
      "loss": 0.7464,
      "step": 9122
    },
    {
      "epoch": 2.8806441427218186,
      "grad_norm": 2.977485418319702,
      "learning_rate": 2.3549224615816433e-05,
      "loss": 0.7256,
      "step": 9123
    },
    {
      "epoch": 2.8809598989580043,
      "grad_norm": 2.6951887607574463,
      "learning_rate": 2.354571609009894e-05,
      "loss": 0.7827,
      "step": 9124
    },
    {
      "epoch": 2.88127565519419,
      "grad_norm": 2.4694252014160156,
      "learning_rate": 2.354220756438145e-05,
      "loss": 0.8331,
      "step": 9125
    },
    {
      "epoch": 2.881591411430376,
      "grad_norm": 2.3648030757904053,
      "learning_rate": 2.3538699038663957e-05,
      "loss": 0.7156,
      "step": 9126
    },
    {
      "epoch": 2.8819071676665615,
      "grad_norm": 2.4061827659606934,
      "learning_rate": 2.353519051294646e-05,
      "loss": 0.7455,
      "step": 9127
    },
    {
      "epoch": 2.882222923902747,
      "grad_norm": 2.6081135272979736,
      "learning_rate": 2.3531681987228967e-05,
      "loss": 0.7597,
      "step": 9128
    },
    {
      "epoch": 2.882538680138933,
      "grad_norm": 2.5671730041503906,
      "learning_rate": 2.3528173461511474e-05,
      "loss": 0.7858,
      "step": 9129
    },
    {
      "epoch": 2.8828544363751183,
      "grad_norm": 2.6901657581329346,
      "learning_rate": 2.352466493579398e-05,
      "loss": 0.7403,
      "step": 9130
    },
    {
      "epoch": 2.883170192611304,
      "grad_norm": 2.815809965133667,
      "learning_rate": 2.3521156410076487e-05,
      "loss": 0.7179,
      "step": 9131
    },
    {
      "epoch": 2.8834859488474898,
      "grad_norm": 2.1767375469207764,
      "learning_rate": 2.3517647884358994e-05,
      "loss": 0.6676,
      "step": 9132
    },
    {
      "epoch": 2.8838017050836755,
      "grad_norm": 2.4508140087127686,
      "learning_rate": 2.3514139358641498e-05,
      "loss": 0.7138,
      "step": 9133
    },
    {
      "epoch": 2.8841174613198612,
      "grad_norm": 2.31829571723938,
      "learning_rate": 2.3510630832924008e-05,
      "loss": 0.6531,
      "step": 9134
    },
    {
      "epoch": 2.8844332175560465,
      "grad_norm": 2.5408732891082764,
      "learning_rate": 2.3507122307206515e-05,
      "loss": 0.6957,
      "step": 9135
    },
    {
      "epoch": 2.8847489737922323,
      "grad_norm": 2.224029302597046,
      "learning_rate": 2.350361378148902e-05,
      "loss": 0.7121,
      "step": 9136
    },
    {
      "epoch": 2.885064730028418,
      "grad_norm": 2.678548812866211,
      "learning_rate": 2.3500105255771525e-05,
      "loss": 0.7984,
      "step": 9137
    },
    {
      "epoch": 2.8853804862646037,
      "grad_norm": 2.498677968978882,
      "learning_rate": 2.349659673005403e-05,
      "loss": 0.7444,
      "step": 9138
    },
    {
      "epoch": 2.8856962425007895,
      "grad_norm": 2.6704139709472656,
      "learning_rate": 2.349308820433654e-05,
      "loss": 0.6944,
      "step": 9139
    },
    {
      "epoch": 2.886011998736975,
      "grad_norm": 2.5774974822998047,
      "learning_rate": 2.3489579678619045e-05,
      "loss": 0.8659,
      "step": 9140
    },
    {
      "epoch": 2.886327754973161,
      "grad_norm": 2.381798505783081,
      "learning_rate": 2.3486071152901552e-05,
      "loss": 0.6512,
      "step": 9141
    },
    {
      "epoch": 2.8866435112093463,
      "grad_norm": 2.8403561115264893,
      "learning_rate": 2.3482562627184056e-05,
      "loss": 0.7022,
      "step": 9142
    },
    {
      "epoch": 2.886959267445532,
      "grad_norm": 2.661376476287842,
      "learning_rate": 2.3479054101466566e-05,
      "loss": 0.6967,
      "step": 9143
    },
    {
      "epoch": 2.8872750236817177,
      "grad_norm": 2.285708427429199,
      "learning_rate": 2.3475545575749073e-05,
      "loss": 0.7336,
      "step": 9144
    },
    {
      "epoch": 2.8875907799179035,
      "grad_norm": 2.317196846008301,
      "learning_rate": 2.347203705003158e-05,
      "loss": 0.7428,
      "step": 9145
    },
    {
      "epoch": 2.887906536154089,
      "grad_norm": 2.2811105251312256,
      "learning_rate": 2.3468528524314086e-05,
      "loss": 0.7459,
      "step": 9146
    },
    {
      "epoch": 2.8882222923902745,
      "grad_norm": 2.5759265422821045,
      "learning_rate": 2.346501999859659e-05,
      "loss": 0.6901,
      "step": 9147
    },
    {
      "epoch": 2.8885380486264602,
      "grad_norm": 2.399906873703003,
      "learning_rate": 2.3461511472879096e-05,
      "loss": 0.7919,
      "step": 9148
    },
    {
      "epoch": 2.888853804862646,
      "grad_norm": 2.6164801120758057,
      "learning_rate": 2.3458002947161603e-05,
      "loss": 0.7064,
      "step": 9149
    },
    {
      "epoch": 2.8891695610988317,
      "grad_norm": 2.515626907348633,
      "learning_rate": 2.345449442144411e-05,
      "loss": 0.7201,
      "step": 9150
    },
    {
      "epoch": 2.8894853173350175,
      "grad_norm": 2.1886205673217773,
      "learning_rate": 2.3450985895726617e-05,
      "loss": 0.7664,
      "step": 9151
    },
    {
      "epoch": 2.889801073571203,
      "grad_norm": 2.4065191745758057,
      "learning_rate": 2.3447477370009124e-05,
      "loss": 0.6935,
      "step": 9152
    },
    {
      "epoch": 2.890116829807389,
      "grad_norm": 2.928807020187378,
      "learning_rate": 2.344396884429163e-05,
      "loss": 0.7561,
      "step": 9153
    },
    {
      "epoch": 2.8904325860435742,
      "grad_norm": 2.7622382640838623,
      "learning_rate": 2.3440460318574137e-05,
      "loss": 0.7947,
      "step": 9154
    },
    {
      "epoch": 2.89074834227976,
      "grad_norm": 2.3261139392852783,
      "learning_rate": 2.3436951792856644e-05,
      "loss": 0.6941,
      "step": 9155
    },
    {
      "epoch": 2.8910640985159457,
      "grad_norm": 2.677812337875366,
      "learning_rate": 2.3433443267139148e-05,
      "loss": 0.8037,
      "step": 9156
    },
    {
      "epoch": 2.8913798547521314,
      "grad_norm": 2.4560165405273438,
      "learning_rate": 2.3429934741421654e-05,
      "loss": 0.75,
      "step": 9157
    },
    {
      "epoch": 2.891695610988317,
      "grad_norm": 2.927993059158325,
      "learning_rate": 2.342642621570416e-05,
      "loss": 0.7926,
      "step": 9158
    },
    {
      "epoch": 2.8920113672245025,
      "grad_norm": 2.4252138137817383,
      "learning_rate": 2.3422917689986668e-05,
      "loss": 0.7315,
      "step": 9159
    },
    {
      "epoch": 2.892327123460688,
      "grad_norm": 2.331474781036377,
      "learning_rate": 2.3419409164269178e-05,
      "loss": 0.7559,
      "step": 9160
    },
    {
      "epoch": 2.892642879696874,
      "grad_norm": 2.182673931121826,
      "learning_rate": 2.341590063855168e-05,
      "loss": 0.7901,
      "step": 9161
    },
    {
      "epoch": 2.8929586359330597,
      "grad_norm": 3.1507253646850586,
      "learning_rate": 2.341239211283419e-05,
      "loss": 0.8072,
      "step": 9162
    },
    {
      "epoch": 2.8932743921692454,
      "grad_norm": 2.446741819381714,
      "learning_rate": 2.3408883587116695e-05,
      "loss": 0.8637,
      "step": 9163
    },
    {
      "epoch": 2.893590148405431,
      "grad_norm": 2.4556102752685547,
      "learning_rate": 2.3405375061399202e-05,
      "loss": 0.6871,
      "step": 9164
    },
    {
      "epoch": 2.893905904641617,
      "grad_norm": 2.3557090759277344,
      "learning_rate": 2.340186653568171e-05,
      "loss": 0.6747,
      "step": 9165
    },
    {
      "epoch": 2.894221660877802,
      "grad_norm": 3.2370240688323975,
      "learning_rate": 2.3398358009964212e-05,
      "loss": 0.724,
      "step": 9166
    },
    {
      "epoch": 2.894537417113988,
      "grad_norm": 2.1850454807281494,
      "learning_rate": 2.339484948424672e-05,
      "loss": 0.602,
      "step": 9167
    },
    {
      "epoch": 2.8948531733501737,
      "grad_norm": 2.444310426712036,
      "learning_rate": 2.3391340958529226e-05,
      "loss": 0.7781,
      "step": 9168
    },
    {
      "epoch": 2.8951689295863594,
      "grad_norm": 2.1039822101593018,
      "learning_rate": 2.3387832432811736e-05,
      "loss": 0.705,
      "step": 9169
    },
    {
      "epoch": 2.895484685822545,
      "grad_norm": 2.704878330230713,
      "learning_rate": 2.338432390709424e-05,
      "loss": 0.7085,
      "step": 9170
    },
    {
      "epoch": 2.8958004420587304,
      "grad_norm": 2.7028064727783203,
      "learning_rate": 2.3380815381376746e-05,
      "loss": 0.7518,
      "step": 9171
    },
    {
      "epoch": 2.896116198294916,
      "grad_norm": 2.270528793334961,
      "learning_rate": 2.3377306855659253e-05,
      "loss": 0.634,
      "step": 9172
    },
    {
      "epoch": 2.896431954531102,
      "grad_norm": 2.4331862926483154,
      "learning_rate": 2.337379832994176e-05,
      "loss": 0.8079,
      "step": 9173
    },
    {
      "epoch": 2.8967477107672877,
      "grad_norm": 2.6804826259613037,
      "learning_rate": 2.3370289804224267e-05,
      "loss": 0.734,
      "step": 9174
    },
    {
      "epoch": 2.8970634670034734,
      "grad_norm": 2.3753066062927246,
      "learning_rate": 2.3366781278506774e-05,
      "loss": 0.7216,
      "step": 9175
    },
    {
      "epoch": 2.897379223239659,
      "grad_norm": 2.3996944427490234,
      "learning_rate": 2.3363272752789277e-05,
      "loss": 0.7649,
      "step": 9176
    },
    {
      "epoch": 2.897694979475845,
      "grad_norm": 2.5006465911865234,
      "learning_rate": 2.3359764227071784e-05,
      "loss": 0.6394,
      "step": 9177
    },
    {
      "epoch": 2.89801073571203,
      "grad_norm": 2.4220168590545654,
      "learning_rate": 2.3356255701354294e-05,
      "loss": 0.8413,
      "step": 9178
    },
    {
      "epoch": 2.898326491948216,
      "grad_norm": 2.1632838249206543,
      "learning_rate": 2.33527471756368e-05,
      "loss": 0.719,
      "step": 9179
    },
    {
      "epoch": 2.8986422481844016,
      "grad_norm": 2.402865171432495,
      "learning_rate": 2.3349238649919304e-05,
      "loss": 0.8039,
      "step": 9180
    },
    {
      "epoch": 2.8989580044205874,
      "grad_norm": 2.5838332176208496,
      "learning_rate": 2.334573012420181e-05,
      "loss": 0.7666,
      "step": 9181
    },
    {
      "epoch": 2.899273760656773,
      "grad_norm": 2.3073391914367676,
      "learning_rate": 2.3342221598484318e-05,
      "loss": 0.6958,
      "step": 9182
    },
    {
      "epoch": 2.8995895168929584,
      "grad_norm": 2.605236768722534,
      "learning_rate": 2.3338713072766825e-05,
      "loss": 0.7798,
      "step": 9183
    },
    {
      "epoch": 2.899905273129144,
      "grad_norm": 2.7228565216064453,
      "learning_rate": 2.333520454704933e-05,
      "loss": 0.7328,
      "step": 9184
    },
    {
      "epoch": 2.90022102936533,
      "grad_norm": 2.7019946575164795,
      "learning_rate": 2.3331696021331835e-05,
      "loss": 0.7565,
      "step": 9185
    },
    {
      "epoch": 2.9005367856015156,
      "grad_norm": 2.143986463546753,
      "learning_rate": 2.3328187495614345e-05,
      "loss": 0.7755,
      "step": 9186
    },
    {
      "epoch": 2.9008525418377014,
      "grad_norm": 2.5695323944091797,
      "learning_rate": 2.3324678969896852e-05,
      "loss": 0.7567,
      "step": 9187
    },
    {
      "epoch": 2.901168298073887,
      "grad_norm": 2.6235575675964355,
      "learning_rate": 2.332117044417936e-05,
      "loss": 0.7443,
      "step": 9188
    },
    {
      "epoch": 2.901484054310073,
      "grad_norm": 2.4483039379119873,
      "learning_rate": 2.3317661918461866e-05,
      "loss": 0.7063,
      "step": 9189
    },
    {
      "epoch": 2.901799810546258,
      "grad_norm": 2.3892669677734375,
      "learning_rate": 2.331415339274437e-05,
      "loss": 0.8403,
      "step": 9190
    },
    {
      "epoch": 2.902115566782444,
      "grad_norm": 2.4775660037994385,
      "learning_rate": 2.3310644867026876e-05,
      "loss": 0.6424,
      "step": 9191
    },
    {
      "epoch": 2.9024313230186296,
      "grad_norm": 2.2743430137634277,
      "learning_rate": 2.3307136341309383e-05,
      "loss": 0.7345,
      "step": 9192
    },
    {
      "epoch": 2.9027470792548153,
      "grad_norm": 2.886209011077881,
      "learning_rate": 2.330362781559189e-05,
      "loss": 0.6533,
      "step": 9193
    },
    {
      "epoch": 2.903062835491001,
      "grad_norm": 2.052237033843994,
      "learning_rate": 2.3300119289874396e-05,
      "loss": 0.6539,
      "step": 9194
    },
    {
      "epoch": 2.9033785917271864,
      "grad_norm": 2.377558708190918,
      "learning_rate": 2.3296610764156903e-05,
      "loss": 0.7381,
      "step": 9195
    },
    {
      "epoch": 2.903694347963372,
      "grad_norm": 2.314160108566284,
      "learning_rate": 2.329310223843941e-05,
      "loss": 0.7144,
      "step": 9196
    },
    {
      "epoch": 2.904010104199558,
      "grad_norm": 2.4018428325653076,
      "learning_rate": 2.3289593712721917e-05,
      "loss": 0.7166,
      "step": 9197
    },
    {
      "epoch": 2.9043258604357436,
      "grad_norm": 2.439297676086426,
      "learning_rate": 2.3286085187004423e-05,
      "loss": 0.7064,
      "step": 9198
    },
    {
      "epoch": 2.9046416166719293,
      "grad_norm": 2.479656457901001,
      "learning_rate": 2.3282576661286927e-05,
      "loss": 0.7374,
      "step": 9199
    },
    {
      "epoch": 2.904957372908115,
      "grad_norm": 2.514787435531616,
      "learning_rate": 2.3279068135569434e-05,
      "loss": 0.7111,
      "step": 9200
    },
    {
      "epoch": 2.904957372908115,
      "eval_loss": 0.7907506823539734,
      "eval_runtime": 4.5761,
      "eval_samples_per_second": 544.131,
      "eval_steps_per_second": 4.589,
      "step": 9200
    },
    {
      "epoch": 2.905273129144301,
      "grad_norm": 2.3527374267578125,
      "learning_rate": 2.327555960985194e-05,
      "loss": 0.6828,
      "step": 9201
    },
    {
      "epoch": 2.905588885380486,
      "grad_norm": 2.7911617755889893,
      "learning_rate": 2.3272051084134447e-05,
      "loss": 0.7265,
      "step": 9202
    },
    {
      "epoch": 2.905904641616672,
      "grad_norm": 2.478618621826172,
      "learning_rate": 2.3268542558416954e-05,
      "loss": 0.7161,
      "step": 9203
    },
    {
      "epoch": 2.9062203978528576,
      "grad_norm": 2.6273932456970215,
      "learning_rate": 2.326503403269946e-05,
      "loss": 0.6341,
      "step": 9204
    },
    {
      "epoch": 2.9065361540890433,
      "grad_norm": 2.9348721504211426,
      "learning_rate": 2.3261525506981968e-05,
      "loss": 0.8427,
      "step": 9205
    },
    {
      "epoch": 2.906851910325229,
      "grad_norm": 2.5915191173553467,
      "learning_rate": 2.3258016981264475e-05,
      "loss": 0.6971,
      "step": 9206
    },
    {
      "epoch": 2.9071676665614143,
      "grad_norm": 3.8210690021514893,
      "learning_rate": 2.325450845554698e-05,
      "loss": 0.7703,
      "step": 9207
    },
    {
      "epoch": 2.9074834227976005,
      "grad_norm": 2.5246222019195557,
      "learning_rate": 2.3250999929829488e-05,
      "loss": 0.6878,
      "step": 9208
    },
    {
      "epoch": 2.907799179033786,
      "grad_norm": 2.448314666748047,
      "learning_rate": 2.324749140411199e-05,
      "loss": 0.7904,
      "step": 9209
    },
    {
      "epoch": 2.9081149352699716,
      "grad_norm": 2.1110124588012695,
      "learning_rate": 2.32439828783945e-05,
      "loss": 0.7379,
      "step": 9210
    },
    {
      "epoch": 2.9084306915061573,
      "grad_norm": 2.8572466373443604,
      "learning_rate": 2.3240474352677005e-05,
      "loss": 0.6891,
      "step": 9211
    },
    {
      "epoch": 2.908746447742343,
      "grad_norm": 2.516035795211792,
      "learning_rate": 2.3236965826959512e-05,
      "loss": 0.7016,
      "step": 9212
    },
    {
      "epoch": 2.9090622039785288,
      "grad_norm": 2.71418833732605,
      "learning_rate": 2.323345730124202e-05,
      "loss": 0.7713,
      "step": 9213
    },
    {
      "epoch": 2.909377960214714,
      "grad_norm": 3.198737382888794,
      "learning_rate": 2.3229948775524526e-05,
      "loss": 0.7062,
      "step": 9214
    },
    {
      "epoch": 2.9096937164509,
      "grad_norm": 2.3342270851135254,
      "learning_rate": 2.3226440249807033e-05,
      "loss": 0.7106,
      "step": 9215
    },
    {
      "epoch": 2.9100094726870855,
      "grad_norm": 2.897231101989746,
      "learning_rate": 2.322293172408954e-05,
      "loss": 0.7663,
      "step": 9216
    },
    {
      "epoch": 2.9103252289232713,
      "grad_norm": 2.7172648906707764,
      "learning_rate": 2.3219423198372046e-05,
      "loss": 0.6911,
      "step": 9217
    },
    {
      "epoch": 2.910640985159457,
      "grad_norm": 2.9866440296173096,
      "learning_rate": 2.321591467265455e-05,
      "loss": 0.7377,
      "step": 9218
    },
    {
      "epoch": 2.9109567413956423,
      "grad_norm": 2.442868232727051,
      "learning_rate": 2.3212406146937056e-05,
      "loss": 0.765,
      "step": 9219
    },
    {
      "epoch": 2.9112724976318285,
      "grad_norm": 2.6033260822296143,
      "learning_rate": 2.3208897621219563e-05,
      "loss": 0.7324,
      "step": 9220
    },
    {
      "epoch": 2.911588253868014,
      "grad_norm": 2.686974048614502,
      "learning_rate": 2.3205389095502073e-05,
      "loss": 0.7643,
      "step": 9221
    },
    {
      "epoch": 2.9119040101041995,
      "grad_norm": 2.550422191619873,
      "learning_rate": 2.320188056978458e-05,
      "loss": 0.6631,
      "step": 9222
    },
    {
      "epoch": 2.9122197663403853,
      "grad_norm": 2.35408616065979,
      "learning_rate": 2.3198372044067084e-05,
      "loss": 0.6548,
      "step": 9223
    },
    {
      "epoch": 2.912535522576571,
      "grad_norm": 2.439974308013916,
      "learning_rate": 2.319486351834959e-05,
      "loss": 0.669,
      "step": 9224
    },
    {
      "epoch": 2.9128512788127567,
      "grad_norm": 2.5757105350494385,
      "learning_rate": 2.3191354992632097e-05,
      "loss": 0.8301,
      "step": 9225
    },
    {
      "epoch": 2.913167035048942,
      "grad_norm": 3.304941177368164,
      "learning_rate": 2.3187846466914604e-05,
      "loss": 0.7786,
      "step": 9226
    },
    {
      "epoch": 2.9134827912851278,
      "grad_norm": 2.642946481704712,
      "learning_rate": 2.318433794119711e-05,
      "loss": 0.7359,
      "step": 9227
    },
    {
      "epoch": 2.9137985475213135,
      "grad_norm": 2.188997745513916,
      "learning_rate": 2.3180829415479614e-05,
      "loss": 0.6688,
      "step": 9228
    },
    {
      "epoch": 2.9141143037574992,
      "grad_norm": 2.67923641204834,
      "learning_rate": 2.317732088976212e-05,
      "loss": 0.7362,
      "step": 9229
    },
    {
      "epoch": 2.914430059993685,
      "grad_norm": 2.586707592010498,
      "learning_rate": 2.317381236404463e-05,
      "loss": 0.7955,
      "step": 9230
    },
    {
      "epoch": 2.9147458162298703,
      "grad_norm": 2.802046537399292,
      "learning_rate": 2.3170303838327138e-05,
      "loss": 0.7692,
      "step": 9231
    },
    {
      "epoch": 2.9150615724660565,
      "grad_norm": 2.211536407470703,
      "learning_rate": 2.316679531260964e-05,
      "loss": 0.7322,
      "step": 9232
    },
    {
      "epoch": 2.9153773287022418,
      "grad_norm": 2.1226494312286377,
      "learning_rate": 2.316328678689215e-05,
      "loss": 0.6952,
      "step": 9233
    },
    {
      "epoch": 2.9156930849384275,
      "grad_norm": 2.3901278972625732,
      "learning_rate": 2.3159778261174655e-05,
      "loss": 0.7487,
      "step": 9234
    },
    {
      "epoch": 2.9160088411746132,
      "grad_norm": 2.7829437255859375,
      "learning_rate": 2.3156269735457162e-05,
      "loss": 0.8001,
      "step": 9235
    },
    {
      "epoch": 2.916324597410799,
      "grad_norm": 2.5125553607940674,
      "learning_rate": 2.315276120973967e-05,
      "loss": 0.8111,
      "step": 9236
    },
    {
      "epoch": 2.9166403536469847,
      "grad_norm": 2.4822020530700684,
      "learning_rate": 2.3149252684022176e-05,
      "loss": 0.7156,
      "step": 9237
    },
    {
      "epoch": 2.91695610988317,
      "grad_norm": 2.672137975692749,
      "learning_rate": 2.314574415830468e-05,
      "loss": 0.8335,
      "step": 9238
    },
    {
      "epoch": 2.9172718661193557,
      "grad_norm": 2.5146148204803467,
      "learning_rate": 2.314223563258719e-05,
      "loss": 0.8109,
      "step": 9239
    },
    {
      "epoch": 2.9175876223555415,
      "grad_norm": 2.4630775451660156,
      "learning_rate": 2.3138727106869696e-05,
      "loss": 0.6802,
      "step": 9240
    },
    {
      "epoch": 2.917903378591727,
      "grad_norm": 2.975440502166748,
      "learning_rate": 2.3135218581152203e-05,
      "loss": 0.6267,
      "step": 9241
    },
    {
      "epoch": 2.918219134827913,
      "grad_norm": 2.4080777168273926,
      "learning_rate": 2.3131710055434706e-05,
      "loss": 0.6899,
      "step": 9242
    },
    {
      "epoch": 2.9185348910640982,
      "grad_norm": 2.655993700027466,
      "learning_rate": 2.3128201529717213e-05,
      "loss": 0.7164,
      "step": 9243
    },
    {
      "epoch": 2.9188506473002844,
      "grad_norm": 2.519840955734253,
      "learning_rate": 2.312469300399972e-05,
      "loss": 0.8745,
      "step": 9244
    },
    {
      "epoch": 2.9191664035364697,
      "grad_norm": 3.1983628273010254,
      "learning_rate": 2.3121184478282227e-05,
      "loss": 0.7983,
      "step": 9245
    },
    {
      "epoch": 2.9194821597726555,
      "grad_norm": 2.204572916030884,
      "learning_rate": 2.3117675952564734e-05,
      "loss": 0.7993,
      "step": 9246
    },
    {
      "epoch": 2.919797916008841,
      "grad_norm": 2.5970592498779297,
      "learning_rate": 2.3114167426847237e-05,
      "loss": 0.7472,
      "step": 9247
    },
    {
      "epoch": 2.920113672245027,
      "grad_norm": 2.5810210704803467,
      "learning_rate": 2.3110658901129747e-05,
      "loss": 0.7308,
      "step": 9248
    },
    {
      "epoch": 2.9204294284812127,
      "grad_norm": 2.1568260192871094,
      "learning_rate": 2.3107150375412254e-05,
      "loss": 0.6906,
      "step": 9249
    },
    {
      "epoch": 2.920745184717398,
      "grad_norm": 2.3935739994049072,
      "learning_rate": 2.310364184969476e-05,
      "loss": 0.795,
      "step": 9250
    },
    {
      "epoch": 2.9210609409535837,
      "grad_norm": 2.2799885272979736,
      "learning_rate": 2.3100133323977268e-05,
      "loss": 0.6987,
      "step": 9251
    },
    {
      "epoch": 2.9213766971897694,
      "grad_norm": 2.4473540782928467,
      "learning_rate": 2.309662479825977e-05,
      "loss": 0.7544,
      "step": 9252
    },
    {
      "epoch": 2.921692453425955,
      "grad_norm": 2.5927000045776367,
      "learning_rate": 2.3093116272542278e-05,
      "loss": 0.7454,
      "step": 9253
    },
    {
      "epoch": 2.922008209662141,
      "grad_norm": 2.446566343307495,
      "learning_rate": 2.3089607746824785e-05,
      "loss": 0.7957,
      "step": 9254
    },
    {
      "epoch": 2.922323965898326,
      "grad_norm": 2.289015054702759,
      "learning_rate": 2.308609922110729e-05,
      "loss": 0.7334,
      "step": 9255
    },
    {
      "epoch": 2.9226397221345124,
      "grad_norm": 2.4502456188201904,
      "learning_rate": 2.30825906953898e-05,
      "loss": 0.7504,
      "step": 9256
    },
    {
      "epoch": 2.9229554783706977,
      "grad_norm": 2.494638681411743,
      "learning_rate": 2.3079082169672305e-05,
      "loss": 0.7337,
      "step": 9257
    },
    {
      "epoch": 2.9232712346068834,
      "grad_norm": 2.7408828735351562,
      "learning_rate": 2.3075573643954812e-05,
      "loss": 0.7054,
      "step": 9258
    },
    {
      "epoch": 2.923586990843069,
      "grad_norm": 3.705935478210449,
      "learning_rate": 2.307206511823732e-05,
      "loss": 0.7181,
      "step": 9259
    },
    {
      "epoch": 2.923902747079255,
      "grad_norm": 2.176945209503174,
      "learning_rate": 2.3068556592519826e-05,
      "loss": 0.7039,
      "step": 9260
    },
    {
      "epoch": 2.9242185033154406,
      "grad_norm": 2.469611644744873,
      "learning_rate": 2.306504806680233e-05,
      "loss": 0.7777,
      "step": 9261
    },
    {
      "epoch": 2.924534259551626,
      "grad_norm": 2.8257594108581543,
      "learning_rate": 2.3061539541084836e-05,
      "loss": 0.6441,
      "step": 9262
    },
    {
      "epoch": 2.9248500157878117,
      "grad_norm": 3.19709849357605,
      "learning_rate": 2.3058031015367343e-05,
      "loss": 0.7018,
      "step": 9263
    },
    {
      "epoch": 2.9251657720239974,
      "grad_norm": 2.3998069763183594,
      "learning_rate": 2.305452248964985e-05,
      "loss": 0.7012,
      "step": 9264
    },
    {
      "epoch": 2.925481528260183,
      "grad_norm": 2.619664192199707,
      "learning_rate": 2.305101396393236e-05,
      "loss": 0.7853,
      "step": 9265
    },
    {
      "epoch": 2.925797284496369,
      "grad_norm": 2.7568752765655518,
      "learning_rate": 2.3047505438214863e-05,
      "loss": 0.6527,
      "step": 9266
    },
    {
      "epoch": 2.9261130407325546,
      "grad_norm": 2.3987815380096436,
      "learning_rate": 2.304399691249737e-05,
      "loss": 0.6548,
      "step": 9267
    },
    {
      "epoch": 2.9264287969687404,
      "grad_norm": 2.6777021884918213,
      "learning_rate": 2.3040488386779877e-05,
      "loss": 0.8004,
      "step": 9268
    },
    {
      "epoch": 2.9267445532049257,
      "grad_norm": 2.3371469974517822,
      "learning_rate": 2.3036979861062383e-05,
      "loss": 0.7647,
      "step": 9269
    },
    {
      "epoch": 2.9270603094411114,
      "grad_norm": 2.7446107864379883,
      "learning_rate": 2.303347133534489e-05,
      "loss": 0.6506,
      "step": 9270
    },
    {
      "epoch": 2.927376065677297,
      "grad_norm": 2.520533323287964,
      "learning_rate": 2.3029962809627394e-05,
      "loss": 0.7571,
      "step": 9271
    },
    {
      "epoch": 2.927691821913483,
      "grad_norm": 2.875678539276123,
      "learning_rate": 2.30264542839099e-05,
      "loss": 0.7086,
      "step": 9272
    },
    {
      "epoch": 2.9280075781496686,
      "grad_norm": 2.1575591564178467,
      "learning_rate": 2.3022945758192407e-05,
      "loss": 0.6945,
      "step": 9273
    },
    {
      "epoch": 2.928323334385854,
      "grad_norm": 3.0031895637512207,
      "learning_rate": 2.3019437232474918e-05,
      "loss": 0.7541,
      "step": 9274
    },
    {
      "epoch": 2.9286390906220396,
      "grad_norm": 2.690725326538086,
      "learning_rate": 2.301592870675742e-05,
      "loss": 0.7384,
      "step": 9275
    },
    {
      "epoch": 2.9289548468582254,
      "grad_norm": 2.770191192626953,
      "learning_rate": 2.3012420181039928e-05,
      "loss": 0.6898,
      "step": 9276
    },
    {
      "epoch": 2.929270603094411,
      "grad_norm": 2.6355934143066406,
      "learning_rate": 2.3008911655322435e-05,
      "loss": 0.8556,
      "step": 9277
    },
    {
      "epoch": 2.929586359330597,
      "grad_norm": 2.7829644680023193,
      "learning_rate": 2.300540312960494e-05,
      "loss": 0.7039,
      "step": 9278
    },
    {
      "epoch": 2.9299021155667826,
      "grad_norm": 2.635843515396118,
      "learning_rate": 2.3001894603887448e-05,
      "loss": 0.7121,
      "step": 9279
    },
    {
      "epoch": 2.9302178718029683,
      "grad_norm": 2.2910680770874023,
      "learning_rate": 2.2998386078169955e-05,
      "loss": 0.6867,
      "step": 9280
    },
    {
      "epoch": 2.9305336280391536,
      "grad_norm": 2.879244804382324,
      "learning_rate": 2.299487755245246e-05,
      "loss": 0.7215,
      "step": 9281
    },
    {
      "epoch": 2.9308493842753394,
      "grad_norm": 2.5234265327453613,
      "learning_rate": 2.2991369026734965e-05,
      "loss": 0.7396,
      "step": 9282
    },
    {
      "epoch": 2.931165140511525,
      "grad_norm": 3.1845791339874268,
      "learning_rate": 2.2987860501017475e-05,
      "loss": 0.7697,
      "step": 9283
    },
    {
      "epoch": 2.931480896747711,
      "grad_norm": 2.8331546783447266,
      "learning_rate": 2.2984351975299982e-05,
      "loss": 0.6977,
      "step": 9284
    },
    {
      "epoch": 2.9317966529838966,
      "grad_norm": 2.5259342193603516,
      "learning_rate": 2.2980843449582486e-05,
      "loss": 0.7082,
      "step": 9285
    },
    {
      "epoch": 2.932112409220082,
      "grad_norm": 2.6455578804016113,
      "learning_rate": 2.2977334923864993e-05,
      "loss": 0.7093,
      "step": 9286
    },
    {
      "epoch": 2.9324281654562676,
      "grad_norm": 2.8479371070861816,
      "learning_rate": 2.29738263981475e-05,
      "loss": 0.773,
      "step": 9287
    },
    {
      "epoch": 2.9327439216924533,
      "grad_norm": 2.9120547771453857,
      "learning_rate": 2.2970317872430006e-05,
      "loss": 0.7796,
      "step": 9288
    },
    {
      "epoch": 2.933059677928639,
      "grad_norm": 3.5225415229797363,
      "learning_rate": 2.2966809346712513e-05,
      "loss": 0.7012,
      "step": 9289
    },
    {
      "epoch": 2.933375434164825,
      "grad_norm": 2.41093111038208,
      "learning_rate": 2.2963300820995016e-05,
      "loss": 0.776,
      "step": 9290
    },
    {
      "epoch": 2.9336911904010106,
      "grad_norm": 2.4169137477874756,
      "learning_rate": 2.2959792295277527e-05,
      "loss": 0.7209,
      "step": 9291
    },
    {
      "epoch": 2.9340069466371963,
      "grad_norm": 2.7657926082611084,
      "learning_rate": 2.2956283769560033e-05,
      "loss": 0.8253,
      "step": 9292
    },
    {
      "epoch": 2.9343227028733816,
      "grad_norm": 2.7613589763641357,
      "learning_rate": 2.295277524384254e-05,
      "loss": 0.7201,
      "step": 9293
    },
    {
      "epoch": 2.9346384591095673,
      "grad_norm": 2.905285120010376,
      "learning_rate": 2.2949266718125047e-05,
      "loss": 0.7265,
      "step": 9294
    },
    {
      "epoch": 2.934954215345753,
      "grad_norm": 2.707615613937378,
      "learning_rate": 2.294575819240755e-05,
      "loss": 0.6745,
      "step": 9295
    },
    {
      "epoch": 2.935269971581939,
      "grad_norm": 3.3419899940490723,
      "learning_rate": 2.2942249666690057e-05,
      "loss": 0.754,
      "step": 9296
    },
    {
      "epoch": 2.9355857278181245,
      "grad_norm": 2.7203922271728516,
      "learning_rate": 2.2938741140972564e-05,
      "loss": 0.7009,
      "step": 9297
    },
    {
      "epoch": 2.93590148405431,
      "grad_norm": 3.031446933746338,
      "learning_rate": 2.293523261525507e-05,
      "loss": 0.7302,
      "step": 9298
    },
    {
      "epoch": 2.9362172402904956,
      "grad_norm": 2.667414665222168,
      "learning_rate": 2.2931724089537578e-05,
      "loss": 0.7433,
      "step": 9299
    },
    {
      "epoch": 2.9365329965266813,
      "grad_norm": 2.6424765586853027,
      "learning_rate": 2.2928215563820085e-05,
      "loss": 0.7439,
      "step": 9300
    },
    {
      "epoch": 2.936848752762867,
      "grad_norm": 2.593205690383911,
      "learning_rate": 2.292470703810259e-05,
      "loss": 0.7018,
      "step": 9301
    },
    {
      "epoch": 2.937164508999053,
      "grad_norm": 2.422253131866455,
      "learning_rate": 2.2921198512385098e-05,
      "loss": 0.6973,
      "step": 9302
    },
    {
      "epoch": 2.9374802652352385,
      "grad_norm": 2.4933977127075195,
      "learning_rate": 2.2917689986667605e-05,
      "loss": 0.7416,
      "step": 9303
    },
    {
      "epoch": 2.9377960214714243,
      "grad_norm": 2.5614523887634277,
      "learning_rate": 2.291418146095011e-05,
      "loss": 0.7329,
      "step": 9304
    },
    {
      "epoch": 2.9381117777076096,
      "grad_norm": 2.583408832550049,
      "learning_rate": 2.2910672935232615e-05,
      "loss": 0.784,
      "step": 9305
    },
    {
      "epoch": 2.9384275339437953,
      "grad_norm": 2.5303988456726074,
      "learning_rate": 2.2907164409515122e-05,
      "loss": 0.7826,
      "step": 9306
    },
    {
      "epoch": 2.938743290179981,
      "grad_norm": 2.3865888118743896,
      "learning_rate": 2.290365588379763e-05,
      "loss": 0.7003,
      "step": 9307
    },
    {
      "epoch": 2.9390590464161668,
      "grad_norm": 3.024289131164551,
      "learning_rate": 2.2900147358080136e-05,
      "loss": 0.748,
      "step": 9308
    },
    {
      "epoch": 2.9393748026523525,
      "grad_norm": 2.6496422290802,
      "learning_rate": 2.2896638832362642e-05,
      "loss": 0.6829,
      "step": 9309
    },
    {
      "epoch": 2.939690558888538,
      "grad_norm": 2.6678364276885986,
      "learning_rate": 2.289313030664515e-05,
      "loss": 0.8478,
      "step": 9310
    },
    {
      "epoch": 2.940006315124724,
      "grad_norm": 2.372049331665039,
      "learning_rate": 2.2889621780927656e-05,
      "loss": 0.6491,
      "step": 9311
    },
    {
      "epoch": 2.9403220713609093,
      "grad_norm": 2.5289385318756104,
      "learning_rate": 2.2886113255210163e-05,
      "loss": 0.729,
      "step": 9312
    },
    {
      "epoch": 2.940637827597095,
      "grad_norm": 2.2929608821868896,
      "learning_rate": 2.288260472949267e-05,
      "loss": 0.7885,
      "step": 9313
    },
    {
      "epoch": 2.9409535838332808,
      "grad_norm": 2.7941489219665527,
      "learning_rate": 2.2879096203775173e-05,
      "loss": 0.7517,
      "step": 9314
    },
    {
      "epoch": 2.9412693400694665,
      "grad_norm": 2.4989917278289795,
      "learning_rate": 2.287558767805768e-05,
      "loss": 0.7494,
      "step": 9315
    },
    {
      "epoch": 2.9415850963056522,
      "grad_norm": 2.5256569385528564,
      "learning_rate": 2.2872079152340187e-05,
      "loss": 0.6575,
      "step": 9316
    },
    {
      "epoch": 2.9419008525418375,
      "grad_norm": 2.280614137649536,
      "learning_rate": 2.2868570626622694e-05,
      "loss": 0.7199,
      "step": 9317
    },
    {
      "epoch": 2.9422166087780233,
      "grad_norm": 2.3302857875823975,
      "learning_rate": 2.28650621009052e-05,
      "loss": 0.7342,
      "step": 9318
    },
    {
      "epoch": 2.942532365014209,
      "grad_norm": 2.2698307037353516,
      "learning_rate": 2.2861553575187707e-05,
      "loss": 0.7477,
      "step": 9319
    },
    {
      "epoch": 2.9428481212503947,
      "grad_norm": 2.8854482173919678,
      "learning_rate": 2.2858045049470214e-05,
      "loss": 0.7319,
      "step": 9320
    },
    {
      "epoch": 2.9431638774865805,
      "grad_norm": 2.921156167984009,
      "learning_rate": 2.285453652375272e-05,
      "loss": 0.7343,
      "step": 9321
    },
    {
      "epoch": 2.9434796337227658,
      "grad_norm": 2.2317609786987305,
      "learning_rate": 2.2851027998035228e-05,
      "loss": 0.787,
      "step": 9322
    },
    {
      "epoch": 2.943795389958952,
      "grad_norm": 2.591343641281128,
      "learning_rate": 2.284751947231773e-05,
      "loss": 0.7551,
      "step": 9323
    },
    {
      "epoch": 2.9441111461951373,
      "grad_norm": 2.561541795730591,
      "learning_rate": 2.2844010946600238e-05,
      "loss": 0.6452,
      "step": 9324
    },
    {
      "epoch": 2.944426902431323,
      "grad_norm": 2.319946050643921,
      "learning_rate": 2.2840502420882745e-05,
      "loss": 0.7733,
      "step": 9325
    },
    {
      "epoch": 2.9447426586675087,
      "grad_norm": 3.149702548980713,
      "learning_rate": 2.2836993895165255e-05,
      "loss": 0.6917,
      "step": 9326
    },
    {
      "epoch": 2.9450584149036945,
      "grad_norm": 2.4643328189849854,
      "learning_rate": 2.2833485369447762e-05,
      "loss": 0.7192,
      "step": 9327
    },
    {
      "epoch": 2.94537417113988,
      "grad_norm": 2.9664700031280518,
      "learning_rate": 2.2829976843730265e-05,
      "loss": 0.8335,
      "step": 9328
    },
    {
      "epoch": 2.9456899273760655,
      "grad_norm": 2.53157114982605,
      "learning_rate": 2.2826468318012772e-05,
      "loss": 0.7873,
      "step": 9329
    },
    {
      "epoch": 2.9460056836122512,
      "grad_norm": 2.3446097373962402,
      "learning_rate": 2.282295979229528e-05,
      "loss": 0.6687,
      "step": 9330
    },
    {
      "epoch": 2.946321439848437,
      "grad_norm": 2.3535728454589844,
      "learning_rate": 2.2819451266577786e-05,
      "loss": 0.7316,
      "step": 9331
    },
    {
      "epoch": 2.9466371960846227,
      "grad_norm": 2.527705430984497,
      "learning_rate": 2.2815942740860292e-05,
      "loss": 0.7524,
      "step": 9332
    },
    {
      "epoch": 2.9469529523208084,
      "grad_norm": 2.1682631969451904,
      "learning_rate": 2.2812434215142796e-05,
      "loss": 0.7618,
      "step": 9333
    },
    {
      "epoch": 2.9472687085569937,
      "grad_norm": 2.7051892280578613,
      "learning_rate": 2.2808925689425303e-05,
      "loss": 0.7881,
      "step": 9334
    },
    {
      "epoch": 2.94758446479318,
      "grad_norm": 2.185054302215576,
      "learning_rate": 2.2805417163707813e-05,
      "loss": 0.7151,
      "step": 9335
    },
    {
      "epoch": 2.947900221029365,
      "grad_norm": 2.4594883918762207,
      "learning_rate": 2.280190863799032e-05,
      "loss": 0.8077,
      "step": 9336
    },
    {
      "epoch": 2.948215977265551,
      "grad_norm": 2.5111420154571533,
      "learning_rate": 2.2798400112272823e-05,
      "loss": 0.7673,
      "step": 9337
    },
    {
      "epoch": 2.9485317335017367,
      "grad_norm": 2.325746774673462,
      "learning_rate": 2.279489158655533e-05,
      "loss": 0.7433,
      "step": 9338
    },
    {
      "epoch": 2.9488474897379224,
      "grad_norm": 2.3583321571350098,
      "learning_rate": 2.2791383060837837e-05,
      "loss": 0.7412,
      "step": 9339
    },
    {
      "epoch": 2.949163245974108,
      "grad_norm": 2.8629915714263916,
      "learning_rate": 2.2787874535120343e-05,
      "loss": 0.754,
      "step": 9340
    },
    {
      "epoch": 2.9494790022102935,
      "grad_norm": 2.54457688331604,
      "learning_rate": 2.278436600940285e-05,
      "loss": 0.7737,
      "step": 9341
    },
    {
      "epoch": 2.949794758446479,
      "grad_norm": 2.4061965942382812,
      "learning_rate": 2.2780857483685357e-05,
      "loss": 0.8298,
      "step": 9342
    },
    {
      "epoch": 2.950110514682665,
      "grad_norm": 2.4773073196411133,
      "learning_rate": 2.277734895796786e-05,
      "loss": 0.72,
      "step": 9343
    },
    {
      "epoch": 2.9504262709188507,
      "grad_norm": 2.5703821182250977,
      "learning_rate": 2.277384043225037e-05,
      "loss": 0.7123,
      "step": 9344
    },
    {
      "epoch": 2.9507420271550364,
      "grad_norm": 2.2407100200653076,
      "learning_rate": 2.2770331906532878e-05,
      "loss": 0.743,
      "step": 9345
    },
    {
      "epoch": 2.9510577833912217,
      "grad_norm": 2.4526469707489014,
      "learning_rate": 2.2766823380815384e-05,
      "loss": 0.8135,
      "step": 9346
    },
    {
      "epoch": 2.951373539627408,
      "grad_norm": 2.7226855754852295,
      "learning_rate": 2.2763314855097888e-05,
      "loss": 0.7637,
      "step": 9347
    },
    {
      "epoch": 2.951689295863593,
      "grad_norm": 3.0358846187591553,
      "learning_rate": 2.2759806329380395e-05,
      "loss": 0.7455,
      "step": 9348
    },
    {
      "epoch": 2.952005052099779,
      "grad_norm": 2.347381114959717,
      "learning_rate": 2.27562978036629e-05,
      "loss": 0.7089,
      "step": 9349
    },
    {
      "epoch": 2.9523208083359647,
      "grad_norm": 2.87960147857666,
      "learning_rate": 2.2752789277945408e-05,
      "loss": 0.7479,
      "step": 9350
    },
    {
      "epoch": 2.9526365645721504,
      "grad_norm": 2.301072597503662,
      "learning_rate": 2.2749280752227915e-05,
      "loss": 0.7636,
      "step": 9351
    },
    {
      "epoch": 2.952952320808336,
      "grad_norm": 2.3202366828918457,
      "learning_rate": 2.274577222651042e-05,
      "loss": 0.7108,
      "step": 9352
    },
    {
      "epoch": 2.9532680770445214,
      "grad_norm": 2.624488353729248,
      "learning_rate": 2.274226370079293e-05,
      "loss": 0.7619,
      "step": 9353
    },
    {
      "epoch": 2.953583833280707,
      "grad_norm": 2.502105474472046,
      "learning_rate": 2.2738755175075435e-05,
      "loss": 0.7465,
      "step": 9354
    },
    {
      "epoch": 2.953899589516893,
      "grad_norm": 2.403258800506592,
      "learning_rate": 2.2735246649357942e-05,
      "loss": 0.7091,
      "step": 9355
    },
    {
      "epoch": 2.9542153457530786,
      "grad_norm": 2.3709185123443604,
      "learning_rate": 2.273173812364045e-05,
      "loss": 0.7785,
      "step": 9356
    },
    {
      "epoch": 2.9545311019892644,
      "grad_norm": 2.5848066806793213,
      "learning_rate": 2.2728229597922953e-05,
      "loss": 0.7608,
      "step": 9357
    },
    {
      "epoch": 2.9548468582254497,
      "grad_norm": 2.896237850189209,
      "learning_rate": 2.272472107220546e-05,
      "loss": 0.815,
      "step": 9358
    },
    {
      "epoch": 2.955162614461636,
      "grad_norm": 2.388188123703003,
      "learning_rate": 2.2721212546487966e-05,
      "loss": 0.7255,
      "step": 9359
    },
    {
      "epoch": 2.955478370697821,
      "grad_norm": 2.38116192817688,
      "learning_rate": 2.2717704020770473e-05,
      "loss": 0.7265,
      "step": 9360
    },
    {
      "epoch": 2.955794126934007,
      "grad_norm": 3.343810796737671,
      "learning_rate": 2.271419549505298e-05,
      "loss": 0.7238,
      "step": 9361
    },
    {
      "epoch": 2.9561098831701926,
      "grad_norm": 2.435420274734497,
      "learning_rate": 2.2710686969335487e-05,
      "loss": 0.7889,
      "step": 9362
    },
    {
      "epoch": 2.9564256394063784,
      "grad_norm": 2.4869956970214844,
      "learning_rate": 2.2707178443617993e-05,
      "loss": 0.6879,
      "step": 9363
    },
    {
      "epoch": 2.956741395642564,
      "grad_norm": 2.441324472427368,
      "learning_rate": 2.27036699179005e-05,
      "loss": 0.7308,
      "step": 9364
    },
    {
      "epoch": 2.9570571518787494,
      "grad_norm": 2.3021225929260254,
      "learning_rate": 2.2700161392183007e-05,
      "loss": 0.6763,
      "step": 9365
    },
    {
      "epoch": 2.957372908114935,
      "grad_norm": 2.499697685241699,
      "learning_rate": 2.269665286646551e-05,
      "loss": 0.7412,
      "step": 9366
    },
    {
      "epoch": 2.957688664351121,
      "grad_norm": 2.4469594955444336,
      "learning_rate": 2.2693144340748017e-05,
      "loss": 0.753,
      "step": 9367
    },
    {
      "epoch": 2.9580044205873066,
      "grad_norm": 2.358693838119507,
      "learning_rate": 2.2689635815030524e-05,
      "loss": 0.6421,
      "step": 9368
    },
    {
      "epoch": 2.9583201768234924,
      "grad_norm": 2.3484089374542236,
      "learning_rate": 2.268612728931303e-05,
      "loss": 0.6543,
      "step": 9369
    },
    {
      "epoch": 2.958635933059678,
      "grad_norm": 2.087099313735962,
      "learning_rate": 2.268261876359554e-05,
      "loss": 0.7007,
      "step": 9370
    },
    {
      "epoch": 2.958951689295864,
      "grad_norm": 2.6972885131835938,
      "learning_rate": 2.2679110237878045e-05,
      "loss": 0.7177,
      "step": 9371
    },
    {
      "epoch": 2.959267445532049,
      "grad_norm": 2.511280059814453,
      "learning_rate": 2.267560171216055e-05,
      "loss": 0.6804,
      "step": 9372
    },
    {
      "epoch": 2.959583201768235,
      "grad_norm": 2.834104537963867,
      "learning_rate": 2.2672093186443058e-05,
      "loss": 0.801,
      "step": 9373
    },
    {
      "epoch": 2.9598989580044206,
      "grad_norm": 3.133873224258423,
      "learning_rate": 2.2668584660725565e-05,
      "loss": 0.7278,
      "step": 9374
    },
    {
      "epoch": 2.9602147142406063,
      "grad_norm": 2.6146724224090576,
      "learning_rate": 2.2665076135008072e-05,
      "loss": 0.715,
      "step": 9375
    },
    {
      "epoch": 2.960530470476792,
      "grad_norm": 2.7582709789276123,
      "learning_rate": 2.2661567609290575e-05,
      "loss": 0.74,
      "step": 9376
    },
    {
      "epoch": 2.9608462267129774,
      "grad_norm": 2.224372148513794,
      "learning_rate": 2.2658059083573082e-05,
      "loss": 0.7153,
      "step": 9377
    },
    {
      "epoch": 2.961161982949163,
      "grad_norm": 2.6538655757904053,
      "learning_rate": 2.265455055785559e-05,
      "loss": 0.6491,
      "step": 9378
    },
    {
      "epoch": 2.961477739185349,
      "grad_norm": 3.035860776901245,
      "learning_rate": 2.26510420321381e-05,
      "loss": 0.8121,
      "step": 9379
    },
    {
      "epoch": 2.9617934954215346,
      "grad_norm": 2.7879831790924072,
      "learning_rate": 2.2647533506420602e-05,
      "loss": 0.8778,
      "step": 9380
    },
    {
      "epoch": 2.9621092516577203,
      "grad_norm": 2.9267659187316895,
      "learning_rate": 2.264402498070311e-05,
      "loss": 0.7493,
      "step": 9381
    },
    {
      "epoch": 2.962425007893906,
      "grad_norm": 2.3777501583099365,
      "learning_rate": 2.2640516454985616e-05,
      "loss": 0.7491,
      "step": 9382
    },
    {
      "epoch": 2.962740764130092,
      "grad_norm": 2.4229063987731934,
      "learning_rate": 2.2637007929268123e-05,
      "loss": 0.6547,
      "step": 9383
    },
    {
      "epoch": 2.963056520366277,
      "grad_norm": 3.3790550231933594,
      "learning_rate": 2.263349940355063e-05,
      "loss": 0.8572,
      "step": 9384
    },
    {
      "epoch": 2.963372276602463,
      "grad_norm": 2.1625850200653076,
      "learning_rate": 2.2629990877833137e-05,
      "loss": 0.7985,
      "step": 9385
    },
    {
      "epoch": 2.9636880328386486,
      "grad_norm": 2.0603485107421875,
      "learning_rate": 2.262648235211564e-05,
      "loss": 0.6371,
      "step": 9386
    },
    {
      "epoch": 2.9640037890748343,
      "grad_norm": 2.50988507270813,
      "learning_rate": 2.2622973826398147e-05,
      "loss": 0.6701,
      "step": 9387
    },
    {
      "epoch": 2.96431954531102,
      "grad_norm": 2.484091281890869,
      "learning_rate": 2.2619465300680657e-05,
      "loss": 0.7588,
      "step": 9388
    },
    {
      "epoch": 2.9646353015472053,
      "grad_norm": 2.635068655014038,
      "learning_rate": 2.2615956774963164e-05,
      "loss": 0.7331,
      "step": 9389
    },
    {
      "epoch": 2.964951057783391,
      "grad_norm": 2.940302610397339,
      "learning_rate": 2.2612448249245667e-05,
      "loss": 0.7365,
      "step": 9390
    },
    {
      "epoch": 2.965266814019577,
      "grad_norm": 2.9274539947509766,
      "learning_rate": 2.2608939723528174e-05,
      "loss": 0.7907,
      "step": 9391
    },
    {
      "epoch": 2.9655825702557626,
      "grad_norm": 2.4936444759368896,
      "learning_rate": 2.260543119781068e-05,
      "loss": 0.7848,
      "step": 9392
    },
    {
      "epoch": 2.9658983264919483,
      "grad_norm": 2.9496941566467285,
      "learning_rate": 2.2601922672093188e-05,
      "loss": 0.7795,
      "step": 9393
    },
    {
      "epoch": 2.966214082728134,
      "grad_norm": 2.301114797592163,
      "learning_rate": 2.2598414146375694e-05,
      "loss": 0.7129,
      "step": 9394
    },
    {
      "epoch": 2.9665298389643198,
      "grad_norm": 2.5499045848846436,
      "learning_rate": 2.2594905620658198e-05,
      "loss": 0.714,
      "step": 9395
    },
    {
      "epoch": 2.966845595200505,
      "grad_norm": 2.8831191062927246,
      "learning_rate": 2.2591397094940708e-05,
      "loss": 0.743,
      "step": 9396
    },
    {
      "epoch": 2.967161351436691,
      "grad_norm": 2.6918575763702393,
      "learning_rate": 2.2587888569223215e-05,
      "loss": 0.75,
      "step": 9397
    },
    {
      "epoch": 2.9674771076728765,
      "grad_norm": 2.8680601119995117,
      "learning_rate": 2.2584380043505722e-05,
      "loss": 0.6918,
      "step": 9398
    },
    {
      "epoch": 2.9677928639090623,
      "grad_norm": 2.850862503051758,
      "learning_rate": 2.258087151778823e-05,
      "loss": 0.7714,
      "step": 9399
    },
    {
      "epoch": 2.968108620145248,
      "grad_norm": 2.7848963737487793,
      "learning_rate": 2.2577362992070732e-05,
      "loss": 0.7573,
      "step": 9400
    },
    {
      "epoch": 2.968108620145248,
      "eval_loss": 0.7882715463638306,
      "eval_runtime": 4.5826,
      "eval_samples_per_second": 543.358,
      "eval_steps_per_second": 4.583,
      "step": 9400
    },
    {
      "epoch": 2.9684243763814333,
      "grad_norm": 2.1458046436309814,
      "learning_rate": 2.257385446635324e-05,
      "loss": 0.6553,
      "step": 9401
    },
    {
      "epoch": 2.968740132617619,
      "grad_norm": 2.385822296142578,
      "learning_rate": 2.2570345940635746e-05,
      "loss": 0.7871,
      "step": 9402
    },
    {
      "epoch": 2.969055888853805,
      "grad_norm": 2.50099778175354,
      "learning_rate": 2.2566837414918252e-05,
      "loss": 0.7819,
      "step": 9403
    },
    {
      "epoch": 2.9693716450899905,
      "grad_norm": 2.1546623706817627,
      "learning_rate": 2.256332888920076e-05,
      "loss": 0.6866,
      "step": 9404
    },
    {
      "epoch": 2.9696874013261763,
      "grad_norm": 2.6214873790740967,
      "learning_rate": 2.2559820363483266e-05,
      "loss": 0.7154,
      "step": 9405
    },
    {
      "epoch": 2.970003157562362,
      "grad_norm": 2.6981287002563477,
      "learning_rate": 2.2556311837765773e-05,
      "loss": 0.7911,
      "step": 9406
    },
    {
      "epoch": 2.9703189137985477,
      "grad_norm": 2.3443069458007812,
      "learning_rate": 2.255280331204828e-05,
      "loss": 0.8287,
      "step": 9407
    },
    {
      "epoch": 2.970634670034733,
      "grad_norm": 2.6191508769989014,
      "learning_rate": 2.2549294786330786e-05,
      "loss": 0.7553,
      "step": 9408
    },
    {
      "epoch": 2.9709504262709188,
      "grad_norm": 2.465024948120117,
      "learning_rate": 2.254578626061329e-05,
      "loss": 0.6107,
      "step": 9409
    },
    {
      "epoch": 2.9712661825071045,
      "grad_norm": 2.213024616241455,
      "learning_rate": 2.2542277734895797e-05,
      "loss": 0.6384,
      "step": 9410
    },
    {
      "epoch": 2.9715819387432902,
      "grad_norm": 2.932532548904419,
      "learning_rate": 2.2538769209178304e-05,
      "loss": 0.7532,
      "step": 9411
    },
    {
      "epoch": 2.971897694979476,
      "grad_norm": 2.8576862812042236,
      "learning_rate": 2.253526068346081e-05,
      "loss": 0.7556,
      "step": 9412
    },
    {
      "epoch": 2.9722134512156613,
      "grad_norm": 2.711447238922119,
      "learning_rate": 2.2531752157743317e-05,
      "loss": 0.6769,
      "step": 9413
    },
    {
      "epoch": 2.972529207451847,
      "grad_norm": 2.4774630069732666,
      "learning_rate": 2.2528243632025824e-05,
      "loss": 0.7917,
      "step": 9414
    },
    {
      "epoch": 2.9728449636880327,
      "grad_norm": 2.654808759689331,
      "learning_rate": 2.252473510630833e-05,
      "loss": 0.7151,
      "step": 9415
    },
    {
      "epoch": 2.9731607199242185,
      "grad_norm": 2.365800380706787,
      "learning_rate": 2.2521226580590838e-05,
      "loss": 0.7325,
      "step": 9416
    },
    {
      "epoch": 2.9734764761604042,
      "grad_norm": 2.8110358715057373,
      "learning_rate": 2.2517718054873344e-05,
      "loss": 0.7276,
      "step": 9417
    },
    {
      "epoch": 2.97379223239659,
      "grad_norm": 2.3555431365966797,
      "learning_rate": 2.251420952915585e-05,
      "loss": 0.7907,
      "step": 9418
    },
    {
      "epoch": 2.9741079886327757,
      "grad_norm": 2.4302377700805664,
      "learning_rate": 2.2510701003438355e-05,
      "loss": 0.7086,
      "step": 9419
    },
    {
      "epoch": 2.974423744868961,
      "grad_norm": 2.803321123123169,
      "learning_rate": 2.250719247772086e-05,
      "loss": 0.7718,
      "step": 9420
    },
    {
      "epoch": 2.9747395011051467,
      "grad_norm": 2.635627508163452,
      "learning_rate": 2.2503683952003368e-05,
      "loss": 0.6909,
      "step": 9421
    },
    {
      "epoch": 2.9750552573413325,
      "grad_norm": 2.6837337017059326,
      "learning_rate": 2.2500175426285875e-05,
      "loss": 0.7385,
      "step": 9422
    },
    {
      "epoch": 2.975371013577518,
      "grad_norm": 2.672274112701416,
      "learning_rate": 2.2496666900568382e-05,
      "loss": 0.7312,
      "step": 9423
    },
    {
      "epoch": 2.975686769813704,
      "grad_norm": 2.689730167388916,
      "learning_rate": 2.249315837485089e-05,
      "loss": 0.7696,
      "step": 9424
    },
    {
      "epoch": 2.9760025260498892,
      "grad_norm": 2.631971597671509,
      "learning_rate": 2.2489649849133395e-05,
      "loss": 0.7309,
      "step": 9425
    },
    {
      "epoch": 2.9763182822860754,
      "grad_norm": 2.7689850330352783,
      "learning_rate": 2.2486141323415902e-05,
      "loss": 0.7946,
      "step": 9426
    },
    {
      "epoch": 2.9766340385222607,
      "grad_norm": 2.5473520755767822,
      "learning_rate": 2.248263279769841e-05,
      "loss": 0.7244,
      "step": 9427
    },
    {
      "epoch": 2.9769497947584465,
      "grad_norm": 2.4865119457244873,
      "learning_rate": 2.2479124271980913e-05,
      "loss": 0.6867,
      "step": 9428
    },
    {
      "epoch": 2.977265550994632,
      "grad_norm": 2.835289478302002,
      "learning_rate": 2.247561574626342e-05,
      "loss": 0.677,
      "step": 9429
    },
    {
      "epoch": 2.977581307230818,
      "grad_norm": 2.601628303527832,
      "learning_rate": 2.2472107220545926e-05,
      "loss": 0.7858,
      "step": 9430
    },
    {
      "epoch": 2.9778970634670037,
      "grad_norm": 3.019651412963867,
      "learning_rate": 2.2468598694828436e-05,
      "loss": 0.6954,
      "step": 9431
    },
    {
      "epoch": 2.978212819703189,
      "grad_norm": 2.7411277294158936,
      "learning_rate": 2.2465090169110943e-05,
      "loss": 0.7652,
      "step": 9432
    },
    {
      "epoch": 2.9785285759393747,
      "grad_norm": 2.7316291332244873,
      "learning_rate": 2.2461581643393447e-05,
      "loss": 0.7761,
      "step": 9433
    },
    {
      "epoch": 2.9788443321755604,
      "grad_norm": 2.573601007461548,
      "learning_rate": 2.2458073117675953e-05,
      "loss": 0.6549,
      "step": 9434
    },
    {
      "epoch": 2.979160088411746,
      "grad_norm": 3.004300594329834,
      "learning_rate": 2.245456459195846e-05,
      "loss": 0.7113,
      "step": 9435
    },
    {
      "epoch": 2.979475844647932,
      "grad_norm": 2.4461488723754883,
      "learning_rate": 2.2451056066240967e-05,
      "loss": 0.6971,
      "step": 9436
    },
    {
      "epoch": 2.979791600884117,
      "grad_norm": 2.9442760944366455,
      "learning_rate": 2.2447547540523474e-05,
      "loss": 0.7152,
      "step": 9437
    },
    {
      "epoch": 2.9801073571203034,
      "grad_norm": 2.24629807472229,
      "learning_rate": 2.2444039014805977e-05,
      "loss": 0.6167,
      "step": 9438
    },
    {
      "epoch": 2.9804231133564887,
      "grad_norm": 3.034132242202759,
      "learning_rate": 2.2440530489088484e-05,
      "loss": 0.7488,
      "step": 9439
    },
    {
      "epoch": 2.9807388695926744,
      "grad_norm": 2.701209545135498,
      "learning_rate": 2.2437021963370994e-05,
      "loss": 0.7542,
      "step": 9440
    },
    {
      "epoch": 2.98105462582886,
      "grad_norm": 2.578453540802002,
      "learning_rate": 2.24335134376535e-05,
      "loss": 0.7049,
      "step": 9441
    },
    {
      "epoch": 2.981370382065046,
      "grad_norm": 2.332347869873047,
      "learning_rate": 2.2430004911936005e-05,
      "loss": 0.6878,
      "step": 9442
    },
    {
      "epoch": 2.9816861383012316,
      "grad_norm": 2.2839841842651367,
      "learning_rate": 2.242649638621851e-05,
      "loss": 0.6951,
      "step": 9443
    },
    {
      "epoch": 2.982001894537417,
      "grad_norm": 2.4879915714263916,
      "learning_rate": 2.2422987860501018e-05,
      "loss": 0.7907,
      "step": 9444
    },
    {
      "epoch": 2.9823176507736027,
      "grad_norm": 2.481168508529663,
      "learning_rate": 2.2419479334783525e-05,
      "loss": 0.7727,
      "step": 9445
    },
    {
      "epoch": 2.9826334070097884,
      "grad_norm": 2.4531731605529785,
      "learning_rate": 2.2415970809066032e-05,
      "loss": 0.7386,
      "step": 9446
    },
    {
      "epoch": 2.982949163245974,
      "grad_norm": 2.503511905670166,
      "learning_rate": 2.241246228334854e-05,
      "loss": 0.7218,
      "step": 9447
    },
    {
      "epoch": 2.98326491948216,
      "grad_norm": 3.0113608837127686,
      "learning_rate": 2.2408953757631042e-05,
      "loss": 0.6928,
      "step": 9448
    },
    {
      "epoch": 2.983580675718345,
      "grad_norm": 2.7119710445404053,
      "learning_rate": 2.2405445231913552e-05,
      "loss": 0.7208,
      "step": 9449
    },
    {
      "epoch": 2.9838964319545314,
      "grad_norm": 2.4296443462371826,
      "learning_rate": 2.240193670619606e-05,
      "loss": 0.6733,
      "step": 9450
    },
    {
      "epoch": 2.9842121881907167,
      "grad_norm": 2.376213550567627,
      "learning_rate": 2.2398428180478566e-05,
      "loss": 0.8242,
      "step": 9451
    },
    {
      "epoch": 2.9845279444269024,
      "grad_norm": 2.1994616985321045,
      "learning_rate": 2.239491965476107e-05,
      "loss": 0.7699,
      "step": 9452
    },
    {
      "epoch": 2.984843700663088,
      "grad_norm": 2.7046713829040527,
      "learning_rate": 2.2391411129043576e-05,
      "loss": 0.8565,
      "step": 9453
    },
    {
      "epoch": 2.985159456899274,
      "grad_norm": 2.2284021377563477,
      "learning_rate": 2.2387902603326083e-05,
      "loss": 0.6986,
      "step": 9454
    },
    {
      "epoch": 2.9854752131354596,
      "grad_norm": 2.3048794269561768,
      "learning_rate": 2.238439407760859e-05,
      "loss": 0.7016,
      "step": 9455
    },
    {
      "epoch": 2.985790969371645,
      "grad_norm": 2.1571717262268066,
      "learning_rate": 2.2380885551891097e-05,
      "loss": 0.7396,
      "step": 9456
    },
    {
      "epoch": 2.9861067256078306,
      "grad_norm": 2.146087169647217,
      "learning_rate": 2.23773770261736e-05,
      "loss": 0.6829,
      "step": 9457
    },
    {
      "epoch": 2.9864224818440164,
      "grad_norm": 2.639073371887207,
      "learning_rate": 2.237386850045611e-05,
      "loss": 0.7657,
      "step": 9458
    },
    {
      "epoch": 2.986738238080202,
      "grad_norm": 2.520195245742798,
      "learning_rate": 2.2370359974738617e-05,
      "loss": 0.7024,
      "step": 9459
    },
    {
      "epoch": 2.987053994316388,
      "grad_norm": 1.9724764823913574,
      "learning_rate": 2.2366851449021124e-05,
      "loss": 0.6993,
      "step": 9460
    },
    {
      "epoch": 2.987369750552573,
      "grad_norm": 2.180185079574585,
      "learning_rate": 2.236334292330363e-05,
      "loss": 0.7056,
      "step": 9461
    },
    {
      "epoch": 2.9876855067887593,
      "grad_norm": 2.6492536067962646,
      "learning_rate": 2.2359834397586134e-05,
      "loss": 0.7873,
      "step": 9462
    },
    {
      "epoch": 2.9880012630249446,
      "grad_norm": 2.764824867248535,
      "learning_rate": 2.235632587186864e-05,
      "loss": 0.7731,
      "step": 9463
    },
    {
      "epoch": 2.9883170192611304,
      "grad_norm": 3.01835298538208,
      "learning_rate": 2.2352817346151148e-05,
      "loss": 0.717,
      "step": 9464
    },
    {
      "epoch": 2.988632775497316,
      "grad_norm": 2.2452502250671387,
      "learning_rate": 2.2349308820433654e-05,
      "loss": 0.7514,
      "step": 9465
    },
    {
      "epoch": 2.988948531733502,
      "grad_norm": 2.3500478267669678,
      "learning_rate": 2.234580029471616e-05,
      "loss": 0.748,
      "step": 9466
    },
    {
      "epoch": 2.9892642879696876,
      "grad_norm": 2.412283182144165,
      "learning_rate": 2.2342291768998668e-05,
      "loss": 0.7731,
      "step": 9467
    },
    {
      "epoch": 2.989580044205873,
      "grad_norm": 2.2623610496520996,
      "learning_rate": 2.2338783243281175e-05,
      "loss": 0.637,
      "step": 9468
    },
    {
      "epoch": 2.9898958004420586,
      "grad_norm": 2.536012649536133,
      "learning_rate": 2.2335274717563682e-05,
      "loss": 0.7535,
      "step": 9469
    },
    {
      "epoch": 2.9902115566782443,
      "grad_norm": 2.404437780380249,
      "learning_rate": 2.233176619184619e-05,
      "loss": 0.7498,
      "step": 9470
    },
    {
      "epoch": 2.99052731291443,
      "grad_norm": 2.225512981414795,
      "learning_rate": 2.2328257666128692e-05,
      "loss": 0.6741,
      "step": 9471
    },
    {
      "epoch": 2.990843069150616,
      "grad_norm": 2.153644323348999,
      "learning_rate": 2.23247491404112e-05,
      "loss": 0.6506,
      "step": 9472
    },
    {
      "epoch": 2.991158825386801,
      "grad_norm": 2.4269330501556396,
      "learning_rate": 2.2321240614693706e-05,
      "loss": 0.7896,
      "step": 9473
    },
    {
      "epoch": 2.9914745816229873,
      "grad_norm": 2.380992889404297,
      "learning_rate": 2.2317732088976212e-05,
      "loss": 0.7008,
      "step": 9474
    },
    {
      "epoch": 2.9917903378591726,
      "grad_norm": 2.2535619735717773,
      "learning_rate": 2.2314223563258723e-05,
      "loss": 0.6349,
      "step": 9475
    },
    {
      "epoch": 2.9921060940953583,
      "grad_norm": 2.505284547805786,
      "learning_rate": 2.2310715037541226e-05,
      "loss": 0.7275,
      "step": 9476
    },
    {
      "epoch": 2.992421850331544,
      "grad_norm": 2.447540283203125,
      "learning_rate": 2.2307206511823733e-05,
      "loss": 0.6637,
      "step": 9477
    },
    {
      "epoch": 2.99273760656773,
      "grad_norm": 2.664951801300049,
      "learning_rate": 2.230369798610624e-05,
      "loss": 0.8635,
      "step": 9478
    },
    {
      "epoch": 2.9930533628039155,
      "grad_norm": 2.396092653274536,
      "learning_rate": 2.2300189460388746e-05,
      "loss": 0.7546,
      "step": 9479
    },
    {
      "epoch": 2.993369119040101,
      "grad_norm": 2.4796979427337646,
      "learning_rate": 2.2296680934671253e-05,
      "loss": 0.6932,
      "step": 9480
    },
    {
      "epoch": 2.9936848752762866,
      "grad_norm": 2.799229621887207,
      "learning_rate": 2.2293172408953757e-05,
      "loss": 0.7453,
      "step": 9481
    },
    {
      "epoch": 2.9940006315124723,
      "grad_norm": 2.0823090076446533,
      "learning_rate": 2.2289663883236264e-05,
      "loss": 0.7346,
      "step": 9482
    },
    {
      "epoch": 2.994316387748658,
      "grad_norm": 2.40826678276062,
      "learning_rate": 2.228615535751877e-05,
      "loss": 0.7725,
      "step": 9483
    },
    {
      "epoch": 2.994632143984844,
      "grad_norm": 2.752639055252075,
      "learning_rate": 2.228264683180128e-05,
      "loss": 0.6254,
      "step": 9484
    },
    {
      "epoch": 2.9949479002210295,
      "grad_norm": 2.4205355644226074,
      "learning_rate": 2.2279138306083784e-05,
      "loss": 0.7326,
      "step": 9485
    },
    {
      "epoch": 2.9952636564572153,
      "grad_norm": 2.8521881103515625,
      "learning_rate": 2.227562978036629e-05,
      "loss": 0.7036,
      "step": 9486
    },
    {
      "epoch": 2.9955794126934006,
      "grad_norm": 2.5124008655548096,
      "learning_rate": 2.2272121254648798e-05,
      "loss": 0.7394,
      "step": 9487
    },
    {
      "epoch": 2.9958951689295863,
      "grad_norm": 2.416018486022949,
      "learning_rate": 2.2268612728931304e-05,
      "loss": 0.7226,
      "step": 9488
    },
    {
      "epoch": 2.996210925165772,
      "grad_norm": 2.524005889892578,
      "learning_rate": 2.226510420321381e-05,
      "loss": 0.7682,
      "step": 9489
    },
    {
      "epoch": 2.9965266814019578,
      "grad_norm": 3.120187520980835,
      "learning_rate": 2.2261595677496318e-05,
      "loss": 0.7382,
      "step": 9490
    },
    {
      "epoch": 2.9968424376381435,
      "grad_norm": 2.3682308197021484,
      "learning_rate": 2.225808715177882e-05,
      "loss": 0.6342,
      "step": 9491
    },
    {
      "epoch": 2.997158193874329,
      "grad_norm": 2.331914186477661,
      "learning_rate": 2.225457862606133e-05,
      "loss": 0.7643,
      "step": 9492
    },
    {
      "epoch": 2.9974739501105145,
      "grad_norm": 2.4755170345306396,
      "learning_rate": 2.225107010034384e-05,
      "loss": 0.6959,
      "step": 9493
    },
    {
      "epoch": 2.9977897063467003,
      "grad_norm": 2.369723320007324,
      "learning_rate": 2.2247561574626345e-05,
      "loss": 0.7255,
      "step": 9494
    },
    {
      "epoch": 2.998105462582886,
      "grad_norm": 2.277313709259033,
      "learning_rate": 2.224405304890885e-05,
      "loss": 0.6426,
      "step": 9495
    },
    {
      "epoch": 2.9984212188190718,
      "grad_norm": 2.309422254562378,
      "learning_rate": 2.2240544523191356e-05,
      "loss": 0.6765,
      "step": 9496
    },
    {
      "epoch": 2.9987369750552575,
      "grad_norm": 2.042219638824463,
      "learning_rate": 2.2237035997473862e-05,
      "loss": 0.6188,
      "step": 9497
    },
    {
      "epoch": 2.9990527312914432,
      "grad_norm": 2.4655420780181885,
      "learning_rate": 2.223352747175637e-05,
      "loss": 0.7534,
      "step": 9498
    },
    {
      "epoch": 2.9993684875276285,
      "grad_norm": 2.3785157203674316,
      "learning_rate": 2.2230018946038876e-05,
      "loss": 0.7536,
      "step": 9499
    },
    {
      "epoch": 2.9996842437638143,
      "grad_norm": 2.7781314849853516,
      "learning_rate": 2.222651042032138e-05,
      "loss": 0.7231,
      "step": 9500
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.567218780517578,
      "learning_rate": 2.222300189460389e-05,
      "loss": 0.7313,
      "step": 9501
    },
    {
      "epoch": 3.0003157562361857,
      "grad_norm": 2.2923483848571777,
      "learning_rate": 2.2219493368886396e-05,
      "loss": 0.6765,
      "step": 9502
    },
    {
      "epoch": 3.0006315124723715,
      "grad_norm": 2.390303611755371,
      "learning_rate": 2.2215984843168903e-05,
      "loss": 0.6167,
      "step": 9503
    },
    {
      "epoch": 3.000947268708557,
      "grad_norm": 2.73435378074646,
      "learning_rate": 2.221247631745141e-05,
      "loss": 0.691,
      "step": 9504
    },
    {
      "epoch": 3.0012630249447425,
      "grad_norm": 2.642695665359497,
      "learning_rate": 2.2208967791733913e-05,
      "loss": 0.7211,
      "step": 9505
    },
    {
      "epoch": 3.0015787811809282,
      "grad_norm": 2.6673355102539062,
      "learning_rate": 2.220545926601642e-05,
      "loss": 0.6519,
      "step": 9506
    },
    {
      "epoch": 3.001894537417114,
      "grad_norm": 2.468641757965088,
      "learning_rate": 2.2201950740298927e-05,
      "loss": 0.7159,
      "step": 9507
    },
    {
      "epoch": 3.0022102936532997,
      "grad_norm": 2.507878541946411,
      "learning_rate": 2.2198442214581434e-05,
      "loss": 0.7153,
      "step": 9508
    },
    {
      "epoch": 3.0025260498894855,
      "grad_norm": 2.565387725830078,
      "learning_rate": 2.219493368886394e-05,
      "loss": 0.6805,
      "step": 9509
    },
    {
      "epoch": 3.002841806125671,
      "grad_norm": 3.161234140396118,
      "learning_rate": 2.2191425163146447e-05,
      "loss": 0.7093,
      "step": 9510
    },
    {
      "epoch": 3.0031575623618565,
      "grad_norm": 2.5710928440093994,
      "learning_rate": 2.2187916637428954e-05,
      "loss": 0.6476,
      "step": 9511
    },
    {
      "epoch": 3.0034733185980422,
      "grad_norm": 2.224478006362915,
      "learning_rate": 2.218440811171146e-05,
      "loss": 0.6485,
      "step": 9512
    },
    {
      "epoch": 3.003789074834228,
      "grad_norm": 2.351905584335327,
      "learning_rate": 2.2180899585993968e-05,
      "loss": 0.6703,
      "step": 9513
    },
    {
      "epoch": 3.0041048310704137,
      "grad_norm": 2.1915500164031982,
      "learning_rate": 2.217739106027647e-05,
      "loss": 0.7194,
      "step": 9514
    },
    {
      "epoch": 3.0044205873065994,
      "grad_norm": 2.8645036220550537,
      "learning_rate": 2.2173882534558978e-05,
      "loss": 0.6478,
      "step": 9515
    },
    {
      "epoch": 3.004736343542785,
      "grad_norm": 2.3021602630615234,
      "learning_rate": 2.2170374008841485e-05,
      "loss": 0.6253,
      "step": 9516
    },
    {
      "epoch": 3.0050520997789705,
      "grad_norm": 2.6329762935638428,
      "learning_rate": 2.2166865483123992e-05,
      "loss": 0.7142,
      "step": 9517
    },
    {
      "epoch": 3.005367856015156,
      "grad_norm": 2.5033791065216064,
      "learning_rate": 2.21633569574065e-05,
      "loss": 0.6194,
      "step": 9518
    },
    {
      "epoch": 3.005683612251342,
      "grad_norm": 2.6306347846984863,
      "learning_rate": 2.2159848431689005e-05,
      "loss": 0.68,
      "step": 9519
    },
    {
      "epoch": 3.0059993684875277,
      "grad_norm": 2.5388002395629883,
      "learning_rate": 2.2156339905971512e-05,
      "loss": 0.6496,
      "step": 9520
    },
    {
      "epoch": 3.0063151247237134,
      "grad_norm": 2.55238938331604,
      "learning_rate": 2.215283138025402e-05,
      "loss": 0.6893,
      "step": 9521
    },
    {
      "epoch": 3.006630880959899,
      "grad_norm": 2.7668662071228027,
      "learning_rate": 2.2149322854536526e-05,
      "loss": 0.5909,
      "step": 9522
    },
    {
      "epoch": 3.0069466371960845,
      "grad_norm": 2.527935266494751,
      "learning_rate": 2.2145814328819033e-05,
      "loss": 0.7,
      "step": 9523
    },
    {
      "epoch": 3.00726239343227,
      "grad_norm": 2.4348082542419434,
      "learning_rate": 2.2142305803101536e-05,
      "loss": 0.5777,
      "step": 9524
    },
    {
      "epoch": 3.007578149668456,
      "grad_norm": 3.0419819355010986,
      "learning_rate": 2.2138797277384043e-05,
      "loss": 0.626,
      "step": 9525
    },
    {
      "epoch": 3.0078939059046417,
      "grad_norm": 2.7957727909088135,
      "learning_rate": 2.213528875166655e-05,
      "loss": 0.6734,
      "step": 9526
    },
    {
      "epoch": 3.0082096621408274,
      "grad_norm": 2.9344482421875,
      "learning_rate": 2.213178022594906e-05,
      "loss": 0.7144,
      "step": 9527
    },
    {
      "epoch": 3.008525418377013,
      "grad_norm": 2.537337303161621,
      "learning_rate": 2.2128271700231563e-05,
      "loss": 0.6499,
      "step": 9528
    },
    {
      "epoch": 3.0088411746131984,
      "grad_norm": 2.669757604598999,
      "learning_rate": 2.212476317451407e-05,
      "loss": 0.6779,
      "step": 9529
    },
    {
      "epoch": 3.009156930849384,
      "grad_norm": 2.6654419898986816,
      "learning_rate": 2.2121254648796577e-05,
      "loss": 0.6113,
      "step": 9530
    },
    {
      "epoch": 3.00947268708557,
      "grad_norm": 2.639533519744873,
      "learning_rate": 2.2117746123079084e-05,
      "loss": 0.6459,
      "step": 9531
    },
    {
      "epoch": 3.0097884433217557,
      "grad_norm": 2.752302646636963,
      "learning_rate": 2.211423759736159e-05,
      "loss": 0.6978,
      "step": 9532
    },
    {
      "epoch": 3.0101041995579414,
      "grad_norm": 3.016140937805176,
      "learning_rate": 2.2110729071644094e-05,
      "loss": 0.6467,
      "step": 9533
    },
    {
      "epoch": 3.010419955794127,
      "grad_norm": 3.0564889907836914,
      "learning_rate": 2.21072205459266e-05,
      "loss": 0.707,
      "step": 9534
    },
    {
      "epoch": 3.0107357120303124,
      "grad_norm": 3.0336427688598633,
      "learning_rate": 2.2103712020209108e-05,
      "loss": 0.7044,
      "step": 9535
    },
    {
      "epoch": 3.011051468266498,
      "grad_norm": 2.4822819232940674,
      "learning_rate": 2.2100203494491618e-05,
      "loss": 0.6026,
      "step": 9536
    },
    {
      "epoch": 3.011367224502684,
      "grad_norm": 2.7683067321777344,
      "learning_rate": 2.2096694968774125e-05,
      "loss": 0.6326,
      "step": 9537
    },
    {
      "epoch": 3.0116829807388696,
      "grad_norm": 2.457064628601074,
      "learning_rate": 2.2093186443056628e-05,
      "loss": 0.7102,
      "step": 9538
    },
    {
      "epoch": 3.0119987369750554,
      "grad_norm": 3.4380903244018555,
      "learning_rate": 2.2089677917339135e-05,
      "loss": 0.7419,
      "step": 9539
    },
    {
      "epoch": 3.012314493211241,
      "grad_norm": 3.352738857269287,
      "learning_rate": 2.2086169391621642e-05,
      "loss": 0.6553,
      "step": 9540
    },
    {
      "epoch": 3.0126302494474264,
      "grad_norm": 2.90500807762146,
      "learning_rate": 2.208266086590415e-05,
      "loss": 0.6607,
      "step": 9541
    },
    {
      "epoch": 3.012946005683612,
      "grad_norm": 2.1016643047332764,
      "learning_rate": 2.2079152340186655e-05,
      "loss": 0.6085,
      "step": 9542
    },
    {
      "epoch": 3.013261761919798,
      "grad_norm": 2.9785819053649902,
      "learning_rate": 2.207564381446916e-05,
      "loss": 0.7002,
      "step": 9543
    },
    {
      "epoch": 3.0135775181559836,
      "grad_norm": 2.9621825218200684,
      "learning_rate": 2.2072135288751666e-05,
      "loss": 0.7005,
      "step": 9544
    },
    {
      "epoch": 3.0138932743921694,
      "grad_norm": 2.666102170944214,
      "learning_rate": 2.2068626763034176e-05,
      "loss": 0.6715,
      "step": 9545
    },
    {
      "epoch": 3.014209030628355,
      "grad_norm": 2.6532950401306152,
      "learning_rate": 2.2065118237316683e-05,
      "loss": 0.7499,
      "step": 9546
    },
    {
      "epoch": 3.0145247868645404,
      "grad_norm": 2.913890838623047,
      "learning_rate": 2.2061609711599186e-05,
      "loss": 0.6616,
      "step": 9547
    },
    {
      "epoch": 3.014840543100726,
      "grad_norm": 2.8629658222198486,
      "learning_rate": 2.2058101185881693e-05,
      "loss": 0.6652,
      "step": 9548
    },
    {
      "epoch": 3.015156299336912,
      "grad_norm": 2.3988595008850098,
      "learning_rate": 2.20545926601642e-05,
      "loss": 0.6309,
      "step": 9549
    },
    {
      "epoch": 3.0154720555730976,
      "grad_norm": 3.1628241539001465,
      "learning_rate": 2.2051084134446706e-05,
      "loss": 0.6853,
      "step": 9550
    },
    {
      "epoch": 3.0157878118092833,
      "grad_norm": 2.7742152214050293,
      "learning_rate": 2.2047575608729213e-05,
      "loss": 0.6549,
      "step": 9551
    },
    {
      "epoch": 3.016103568045469,
      "grad_norm": 2.937753915786743,
      "learning_rate": 2.204406708301172e-05,
      "loss": 0.6644,
      "step": 9552
    },
    {
      "epoch": 3.0164193242816544,
      "grad_norm": 3.208048105239868,
      "learning_rate": 2.2040558557294224e-05,
      "loss": 0.6096,
      "step": 9553
    },
    {
      "epoch": 3.01673508051784,
      "grad_norm": 2.8475377559661865,
      "learning_rate": 2.2037050031576734e-05,
      "loss": 0.5841,
      "step": 9554
    },
    {
      "epoch": 3.017050836754026,
      "grad_norm": 3.6077914237976074,
      "learning_rate": 2.203354150585924e-05,
      "loss": 0.6522,
      "step": 9555
    },
    {
      "epoch": 3.0173665929902116,
      "grad_norm": 3.0725550651550293,
      "learning_rate": 2.2030032980141747e-05,
      "loss": 0.6854,
      "step": 9556
    },
    {
      "epoch": 3.0176823492263973,
      "grad_norm": 2.6072847843170166,
      "learning_rate": 2.202652445442425e-05,
      "loss": 0.6393,
      "step": 9557
    },
    {
      "epoch": 3.017998105462583,
      "grad_norm": 3.113889455795288,
      "learning_rate": 2.2023015928706758e-05,
      "loss": 0.7181,
      "step": 9558
    },
    {
      "epoch": 3.0183138616987684,
      "grad_norm": 2.67492938041687,
      "learning_rate": 2.2019507402989264e-05,
      "loss": 0.6789,
      "step": 9559
    },
    {
      "epoch": 3.018629617934954,
      "grad_norm": 2.6990251541137695,
      "learning_rate": 2.201599887727177e-05,
      "loss": 0.6171,
      "step": 9560
    },
    {
      "epoch": 3.01894537417114,
      "grad_norm": 2.745776891708374,
      "learning_rate": 2.2012490351554278e-05,
      "loss": 0.6313,
      "step": 9561
    },
    {
      "epoch": 3.0192611304073256,
      "grad_norm": 2.96226167678833,
      "learning_rate": 2.2008981825836785e-05,
      "loss": 0.6547,
      "step": 9562
    },
    {
      "epoch": 3.0195768866435113,
      "grad_norm": 3.9209048748016357,
      "learning_rate": 2.200547330011929e-05,
      "loss": 0.6,
      "step": 9563
    },
    {
      "epoch": 3.019892642879697,
      "grad_norm": 2.7523226737976074,
      "learning_rate": 2.20019647744018e-05,
      "loss": 0.7793,
      "step": 9564
    },
    {
      "epoch": 3.0202083991158823,
      "grad_norm": 3.6282057762145996,
      "learning_rate": 2.1998456248684305e-05,
      "loss": 0.6771,
      "step": 9565
    },
    {
      "epoch": 3.020524155352068,
      "grad_norm": 2.711419105529785,
      "learning_rate": 2.1994947722966812e-05,
      "loss": 0.582,
      "step": 9566
    },
    {
      "epoch": 3.020839911588254,
      "grad_norm": 2.7873950004577637,
      "learning_rate": 2.1991439197249316e-05,
      "loss": 0.7027,
      "step": 9567
    },
    {
      "epoch": 3.0211556678244396,
      "grad_norm": 2.8625128269195557,
      "learning_rate": 2.1987930671531822e-05,
      "loss": 0.5845,
      "step": 9568
    },
    {
      "epoch": 3.0214714240606253,
      "grad_norm": 2.804888963699341,
      "learning_rate": 2.198442214581433e-05,
      "loss": 0.5975,
      "step": 9569
    },
    {
      "epoch": 3.021787180296811,
      "grad_norm": 2.3798463344573975,
      "learning_rate": 2.1980913620096836e-05,
      "loss": 0.6938,
      "step": 9570
    },
    {
      "epoch": 3.0221029365329963,
      "grad_norm": 2.851135492324829,
      "learning_rate": 2.1977405094379343e-05,
      "loss": 0.6549,
      "step": 9571
    },
    {
      "epoch": 3.022418692769182,
      "grad_norm": 3.0329692363739014,
      "learning_rate": 2.197389656866185e-05,
      "loss": 0.7201,
      "step": 9572
    },
    {
      "epoch": 3.022734449005368,
      "grad_norm": 3.13747501373291,
      "learning_rate": 2.1970388042944356e-05,
      "loss": 0.7127,
      "step": 9573
    },
    {
      "epoch": 3.0230502052415535,
      "grad_norm": 2.7635459899902344,
      "learning_rate": 2.1966879517226863e-05,
      "loss": 0.5893,
      "step": 9574
    },
    {
      "epoch": 3.0233659614777393,
      "grad_norm": 2.8861048221588135,
      "learning_rate": 2.196337099150937e-05,
      "loss": 0.6332,
      "step": 9575
    },
    {
      "epoch": 3.023681717713925,
      "grad_norm": 2.5466971397399902,
      "learning_rate": 2.1959862465791873e-05,
      "loss": 0.5869,
      "step": 9576
    },
    {
      "epoch": 3.0239974739501103,
      "grad_norm": 3.138441324234009,
      "learning_rate": 2.195635394007438e-05,
      "loss": 0.7039,
      "step": 9577
    },
    {
      "epoch": 3.024313230186296,
      "grad_norm": 2.8343029022216797,
      "learning_rate": 2.1952845414356887e-05,
      "loss": 0.7371,
      "step": 9578
    },
    {
      "epoch": 3.024628986422482,
      "grad_norm": 2.7146263122558594,
      "learning_rate": 2.1949336888639394e-05,
      "loss": 0.6332,
      "step": 9579
    },
    {
      "epoch": 3.0249447426586675,
      "grad_norm": 3.6055362224578857,
      "learning_rate": 2.1945828362921904e-05,
      "loss": 0.7146,
      "step": 9580
    },
    {
      "epoch": 3.0252604988948533,
      "grad_norm": 2.66831111907959,
      "learning_rate": 2.1942319837204408e-05,
      "loss": 0.6314,
      "step": 9581
    },
    {
      "epoch": 3.025576255131039,
      "grad_norm": 2.90625,
      "learning_rate": 2.1938811311486914e-05,
      "loss": 0.7178,
      "step": 9582
    },
    {
      "epoch": 3.0258920113672243,
      "grad_norm": 2.840942859649658,
      "learning_rate": 2.193530278576942e-05,
      "loss": 0.7508,
      "step": 9583
    },
    {
      "epoch": 3.02620776760341,
      "grad_norm": 2.823155403137207,
      "learning_rate": 2.1931794260051928e-05,
      "loss": 0.7176,
      "step": 9584
    },
    {
      "epoch": 3.0265235238395958,
      "grad_norm": 3.0534889698028564,
      "learning_rate": 2.1928285734334435e-05,
      "loss": 0.6849,
      "step": 9585
    },
    {
      "epoch": 3.0268392800757815,
      "grad_norm": 2.5363831520080566,
      "learning_rate": 2.1924777208616938e-05,
      "loss": 0.6365,
      "step": 9586
    },
    {
      "epoch": 3.0271550363119673,
      "grad_norm": 3.40232253074646,
      "learning_rate": 2.1921268682899445e-05,
      "loss": 0.7435,
      "step": 9587
    },
    {
      "epoch": 3.027470792548153,
      "grad_norm": 2.7407419681549072,
      "learning_rate": 2.1917760157181952e-05,
      "loss": 0.6119,
      "step": 9588
    },
    {
      "epoch": 3.0277865487843383,
      "grad_norm": 2.7686607837677,
      "learning_rate": 2.1914251631464462e-05,
      "loss": 0.6805,
      "step": 9589
    },
    {
      "epoch": 3.028102305020524,
      "grad_norm": 2.8923964500427246,
      "learning_rate": 2.1910743105746965e-05,
      "loss": 0.7032,
      "step": 9590
    },
    {
      "epoch": 3.0284180612567098,
      "grad_norm": 3.213587999343872,
      "learning_rate": 2.1907234580029472e-05,
      "loss": 0.6551,
      "step": 9591
    },
    {
      "epoch": 3.0287338174928955,
      "grad_norm": 2.3602192401885986,
      "learning_rate": 2.190372605431198e-05,
      "loss": 0.5962,
      "step": 9592
    },
    {
      "epoch": 3.0290495737290812,
      "grad_norm": 3.0193827152252197,
      "learning_rate": 2.1900217528594486e-05,
      "loss": 0.7393,
      "step": 9593
    },
    {
      "epoch": 3.029365329965267,
      "grad_norm": 2.8035659790039062,
      "learning_rate": 2.1896709002876993e-05,
      "loss": 0.6098,
      "step": 9594
    },
    {
      "epoch": 3.0296810862014523,
      "grad_norm": 3.1035594940185547,
      "learning_rate": 2.18932004771595e-05,
      "loss": 0.7003,
      "step": 9595
    },
    {
      "epoch": 3.029996842437638,
      "grad_norm": 2.959583044052124,
      "learning_rate": 2.1889691951442003e-05,
      "loss": 0.6627,
      "step": 9596
    },
    {
      "epoch": 3.0303125986738237,
      "grad_norm": 2.778021812438965,
      "learning_rate": 2.1886183425724513e-05,
      "loss": 0.6013,
      "step": 9597
    },
    {
      "epoch": 3.0306283549100095,
      "grad_norm": 2.7092878818511963,
      "learning_rate": 2.188267490000702e-05,
      "loss": 0.6072,
      "step": 9598
    },
    {
      "epoch": 3.030944111146195,
      "grad_norm": 2.387993097305298,
      "learning_rate": 2.1879166374289527e-05,
      "loss": 0.6511,
      "step": 9599
    },
    {
      "epoch": 3.031259867382381,
      "grad_norm": 2.998387336730957,
      "learning_rate": 2.187565784857203e-05,
      "loss": 0.5777,
      "step": 9600
    },
    {
      "epoch": 3.031259867382381,
      "eval_loss": 0.7950844168663025,
      "eval_runtime": 4.5975,
      "eval_samples_per_second": 541.6,
      "eval_steps_per_second": 4.568,
      "step": 9600
    },
    {
      "epoch": 3.0315756236185663,
      "grad_norm": 2.7731103897094727,
      "learning_rate": 2.1872149322854537e-05,
      "loss": 0.6171,
      "step": 9601
    },
    {
      "epoch": 3.031891379854752,
      "grad_norm": 2.8414316177368164,
      "learning_rate": 2.1868640797137044e-05,
      "loss": 0.654,
      "step": 9602
    },
    {
      "epoch": 3.0322071360909377,
      "grad_norm": 3.5313236713409424,
      "learning_rate": 2.186513227141955e-05,
      "loss": 0.7389,
      "step": 9603
    },
    {
      "epoch": 3.0325228923271235,
      "grad_norm": 2.8909530639648438,
      "learning_rate": 2.1861623745702057e-05,
      "loss": 0.603,
      "step": 9604
    },
    {
      "epoch": 3.032838648563309,
      "grad_norm": 3.779252529144287,
      "learning_rate": 2.185811521998456e-05,
      "loss": 0.709,
      "step": 9605
    },
    {
      "epoch": 3.033154404799495,
      "grad_norm": 2.8411290645599365,
      "learning_rate": 2.185460669426707e-05,
      "loss": 0.7377,
      "step": 9606
    },
    {
      "epoch": 3.0334701610356802,
      "grad_norm": 2.7601656913757324,
      "learning_rate": 2.1851098168549578e-05,
      "loss": 0.6622,
      "step": 9607
    },
    {
      "epoch": 3.033785917271866,
      "grad_norm": 2.6784827709198,
      "learning_rate": 2.1847589642832085e-05,
      "loss": 0.6545,
      "step": 9608
    },
    {
      "epoch": 3.0341016735080517,
      "grad_norm": 2.6275341510772705,
      "learning_rate": 2.184408111711459e-05,
      "loss": 0.7475,
      "step": 9609
    },
    {
      "epoch": 3.0344174297442374,
      "grad_norm": 3.4160425662994385,
      "learning_rate": 2.1840572591397095e-05,
      "loss": 0.7136,
      "step": 9610
    },
    {
      "epoch": 3.034733185980423,
      "grad_norm": 2.575544595718384,
      "learning_rate": 2.1837064065679602e-05,
      "loss": 0.6199,
      "step": 9611
    },
    {
      "epoch": 3.035048942216609,
      "grad_norm": 2.482480764389038,
      "learning_rate": 2.183355553996211e-05,
      "loss": 0.5744,
      "step": 9612
    },
    {
      "epoch": 3.0353646984527947,
      "grad_norm": 2.8000869750976562,
      "learning_rate": 2.1830047014244615e-05,
      "loss": 0.6453,
      "step": 9613
    },
    {
      "epoch": 3.03568045468898,
      "grad_norm": 3.7554831504821777,
      "learning_rate": 2.1826538488527122e-05,
      "loss": 0.7098,
      "step": 9614
    },
    {
      "epoch": 3.0359962109251657,
      "grad_norm": 2.762352228164673,
      "learning_rate": 2.182302996280963e-05,
      "loss": 0.7033,
      "step": 9615
    },
    {
      "epoch": 3.0363119671613514,
      "grad_norm": 3.143038511276245,
      "learning_rate": 2.1819521437092136e-05,
      "loss": 0.6337,
      "step": 9616
    },
    {
      "epoch": 3.036627723397537,
      "grad_norm": 2.55827260017395,
      "learning_rate": 2.1816012911374643e-05,
      "loss": 0.5819,
      "step": 9617
    },
    {
      "epoch": 3.036943479633723,
      "grad_norm": 2.622558116912842,
      "learning_rate": 2.181250438565715e-05,
      "loss": 0.6744,
      "step": 9618
    },
    {
      "epoch": 3.0372592358699086,
      "grad_norm": 3.009582281112671,
      "learning_rate": 2.1808995859939653e-05,
      "loss": 0.589,
      "step": 9619
    },
    {
      "epoch": 3.037574992106094,
      "grad_norm": 3.0914580821990967,
      "learning_rate": 2.180548733422216e-05,
      "loss": 0.6323,
      "step": 9620
    },
    {
      "epoch": 3.0378907483422797,
      "grad_norm": 2.740321636199951,
      "learning_rate": 2.1801978808504666e-05,
      "loss": 0.7211,
      "step": 9621
    },
    {
      "epoch": 3.0382065045784654,
      "grad_norm": 2.6455905437469482,
      "learning_rate": 2.1798470282787173e-05,
      "loss": 0.6674,
      "step": 9622
    },
    {
      "epoch": 3.038522260814651,
      "grad_norm": 2.877906084060669,
      "learning_rate": 2.179496175706968e-05,
      "loss": 0.6409,
      "step": 9623
    },
    {
      "epoch": 3.038838017050837,
      "grad_norm": 2.845367193222046,
      "learning_rate": 2.1791453231352187e-05,
      "loss": 0.5746,
      "step": 9624
    },
    {
      "epoch": 3.0391537732870226,
      "grad_norm": 2.896362066268921,
      "learning_rate": 2.1787944705634694e-05,
      "loss": 0.6763,
      "step": 9625
    },
    {
      "epoch": 3.039469529523208,
      "grad_norm": 2.782477378845215,
      "learning_rate": 2.17844361799172e-05,
      "loss": 0.6442,
      "step": 9626
    },
    {
      "epoch": 3.0397852857593937,
      "grad_norm": 4.3072357177734375,
      "learning_rate": 2.1780927654199707e-05,
      "loss": 0.5794,
      "step": 9627
    },
    {
      "epoch": 3.0401010419955794,
      "grad_norm": 3.0062804222106934,
      "learning_rate": 2.1777419128482214e-05,
      "loss": 0.6189,
      "step": 9628
    },
    {
      "epoch": 3.040416798231765,
      "grad_norm": 2.5883848667144775,
      "learning_rate": 2.1773910602764718e-05,
      "loss": 0.6794,
      "step": 9629
    },
    {
      "epoch": 3.040732554467951,
      "grad_norm": 3.0680086612701416,
      "learning_rate": 2.1770402077047224e-05,
      "loss": 0.6753,
      "step": 9630
    },
    {
      "epoch": 3.0410483107041366,
      "grad_norm": 2.9180967807769775,
      "learning_rate": 2.176689355132973e-05,
      "loss": 0.6765,
      "step": 9631
    },
    {
      "epoch": 3.041364066940322,
      "grad_norm": 2.9640653133392334,
      "learning_rate": 2.176338502561224e-05,
      "loss": 0.6272,
      "step": 9632
    },
    {
      "epoch": 3.0416798231765076,
      "grad_norm": 2.921194314956665,
      "learning_rate": 2.1759876499894745e-05,
      "loss": 0.6467,
      "step": 9633
    },
    {
      "epoch": 3.0419955794126934,
      "grad_norm": 2.8514816761016846,
      "learning_rate": 2.175636797417725e-05,
      "loss": 0.6457,
      "step": 9634
    },
    {
      "epoch": 3.042311335648879,
      "grad_norm": 2.8518412113189697,
      "learning_rate": 2.175285944845976e-05,
      "loss": 0.7584,
      "step": 9635
    },
    {
      "epoch": 3.042627091885065,
      "grad_norm": 3.3427557945251465,
      "learning_rate": 2.1749350922742265e-05,
      "loss": 0.6901,
      "step": 9636
    },
    {
      "epoch": 3.0429428481212506,
      "grad_norm": 3.171706438064575,
      "learning_rate": 2.1745842397024772e-05,
      "loss": 0.5519,
      "step": 9637
    },
    {
      "epoch": 3.043258604357436,
      "grad_norm": 3.240757942199707,
      "learning_rate": 2.1742333871307276e-05,
      "loss": 0.5995,
      "step": 9638
    },
    {
      "epoch": 3.0435743605936216,
      "grad_norm": 2.63503360748291,
      "learning_rate": 2.1738825345589782e-05,
      "loss": 0.645,
      "step": 9639
    },
    {
      "epoch": 3.0438901168298074,
      "grad_norm": 2.9493603706359863,
      "learning_rate": 2.173531681987229e-05,
      "loss": 0.6763,
      "step": 9640
    },
    {
      "epoch": 3.044205873065993,
      "grad_norm": 3.1331100463867188,
      "learning_rate": 2.17318082941548e-05,
      "loss": 0.6453,
      "step": 9641
    },
    {
      "epoch": 3.044521629302179,
      "grad_norm": 2.5853538513183594,
      "learning_rate": 2.1728299768437306e-05,
      "loss": 0.6043,
      "step": 9642
    },
    {
      "epoch": 3.0448373855383646,
      "grad_norm": 2.7418253421783447,
      "learning_rate": 2.172479124271981e-05,
      "loss": 0.6722,
      "step": 9643
    },
    {
      "epoch": 3.04515314177455,
      "grad_norm": 2.5591166019439697,
      "learning_rate": 2.1721282717002316e-05,
      "loss": 0.6279,
      "step": 9644
    },
    {
      "epoch": 3.0454688980107356,
      "grad_norm": 2.728153944015503,
      "learning_rate": 2.1717774191284823e-05,
      "loss": 0.6991,
      "step": 9645
    },
    {
      "epoch": 3.0457846542469214,
      "grad_norm": 3.3016343116760254,
      "learning_rate": 2.171426566556733e-05,
      "loss": 0.6046,
      "step": 9646
    },
    {
      "epoch": 3.046100410483107,
      "grad_norm": 2.306528329849243,
      "learning_rate": 2.1710757139849837e-05,
      "loss": 0.5744,
      "step": 9647
    },
    {
      "epoch": 3.046416166719293,
      "grad_norm": 3.0932059288024902,
      "learning_rate": 2.170724861413234e-05,
      "loss": 0.6671,
      "step": 9648
    },
    {
      "epoch": 3.0467319229554786,
      "grad_norm": 2.6485323905944824,
      "learning_rate": 2.1703740088414847e-05,
      "loss": 0.6354,
      "step": 9649
    },
    {
      "epoch": 3.047047679191664,
      "grad_norm": 2.6466550827026367,
      "learning_rate": 2.1700231562697357e-05,
      "loss": 0.616,
      "step": 9650
    },
    {
      "epoch": 3.0473634354278496,
      "grad_norm": 3.103739023208618,
      "learning_rate": 2.1696723036979864e-05,
      "loss": 0.6873,
      "step": 9651
    },
    {
      "epoch": 3.0476791916640353,
      "grad_norm": 3.1080381870269775,
      "learning_rate": 2.1693214511262368e-05,
      "loss": 0.7376,
      "step": 9652
    },
    {
      "epoch": 3.047994947900221,
      "grad_norm": 2.821160316467285,
      "learning_rate": 2.1689705985544874e-05,
      "loss": 0.7733,
      "step": 9653
    },
    {
      "epoch": 3.048310704136407,
      "grad_norm": 2.62251615524292,
      "learning_rate": 2.168619745982738e-05,
      "loss": 0.5932,
      "step": 9654
    },
    {
      "epoch": 3.0486264603725926,
      "grad_norm": 3.7254297733306885,
      "learning_rate": 2.1682688934109888e-05,
      "loss": 0.651,
      "step": 9655
    },
    {
      "epoch": 3.048942216608778,
      "grad_norm": 2.629586935043335,
      "learning_rate": 2.1679180408392395e-05,
      "loss": 0.6077,
      "step": 9656
    },
    {
      "epoch": 3.0492579728449636,
      "grad_norm": 3.5619497299194336,
      "learning_rate": 2.16756718826749e-05,
      "loss": 0.701,
      "step": 9657
    },
    {
      "epoch": 3.0495737290811493,
      "grad_norm": 3.0094730854034424,
      "learning_rate": 2.1672163356957405e-05,
      "loss": 0.6628,
      "step": 9658
    },
    {
      "epoch": 3.049889485317335,
      "grad_norm": 3.1072890758514404,
      "learning_rate": 2.1668654831239915e-05,
      "loss": 0.7073,
      "step": 9659
    },
    {
      "epoch": 3.050205241553521,
      "grad_norm": 3.1592020988464355,
      "learning_rate": 2.1665146305522422e-05,
      "loss": 0.6441,
      "step": 9660
    },
    {
      "epoch": 3.0505209977897065,
      "grad_norm": 3.139833450317383,
      "learning_rate": 2.166163777980493e-05,
      "loss": 0.7924,
      "step": 9661
    },
    {
      "epoch": 3.050836754025892,
      "grad_norm": 2.9128787517547607,
      "learning_rate": 2.1658129254087432e-05,
      "loss": 0.7862,
      "step": 9662
    },
    {
      "epoch": 3.0511525102620776,
      "grad_norm": 2.836887836456299,
      "learning_rate": 2.165462072836994e-05,
      "loss": 0.7435,
      "step": 9663
    },
    {
      "epoch": 3.0514682664982633,
      "grad_norm": 2.876826763153076,
      "learning_rate": 2.1651112202652446e-05,
      "loss": 0.6139,
      "step": 9664
    },
    {
      "epoch": 3.051784022734449,
      "grad_norm": 2.9566571712493896,
      "learning_rate": 2.1647603676934953e-05,
      "loss": 0.7014,
      "step": 9665
    },
    {
      "epoch": 3.052099778970635,
      "grad_norm": 3.2445414066314697,
      "learning_rate": 2.164409515121746e-05,
      "loss": 0.7477,
      "step": 9666
    },
    {
      "epoch": 3.0524155352068205,
      "grad_norm": 2.808030605316162,
      "learning_rate": 2.1640586625499966e-05,
      "loss": 0.626,
      "step": 9667
    },
    {
      "epoch": 3.052731291443006,
      "grad_norm": 2.8455991744995117,
      "learning_rate": 2.1637078099782473e-05,
      "loss": 0.6465,
      "step": 9668
    },
    {
      "epoch": 3.0530470476791916,
      "grad_norm": 2.6729369163513184,
      "learning_rate": 2.163356957406498e-05,
      "loss": 0.603,
      "step": 9669
    },
    {
      "epoch": 3.0533628039153773,
      "grad_norm": 2.9384281635284424,
      "learning_rate": 2.1630061048347487e-05,
      "loss": 0.6931,
      "step": 9670
    },
    {
      "epoch": 3.053678560151563,
      "grad_norm": 2.6920058727264404,
      "learning_rate": 2.1626552522629994e-05,
      "loss": 0.7497,
      "step": 9671
    },
    {
      "epoch": 3.0539943163877488,
      "grad_norm": 3.069693088531494,
      "learning_rate": 2.1623043996912497e-05,
      "loss": 0.5685,
      "step": 9672
    },
    {
      "epoch": 3.0543100726239345,
      "grad_norm": 3.2745280265808105,
      "learning_rate": 2.1619535471195004e-05,
      "loss": 0.6526,
      "step": 9673
    },
    {
      "epoch": 3.05462582886012,
      "grad_norm": 3.0746846199035645,
      "learning_rate": 2.161602694547751e-05,
      "loss": 0.6581,
      "step": 9674
    },
    {
      "epoch": 3.0549415850963055,
      "grad_norm": 2.8823838233947754,
      "learning_rate": 2.1612518419760017e-05,
      "loss": 0.696,
      "step": 9675
    },
    {
      "epoch": 3.0552573413324913,
      "grad_norm": 2.9484610557556152,
      "learning_rate": 2.1609009894042524e-05,
      "loss": 0.7034,
      "step": 9676
    },
    {
      "epoch": 3.055573097568677,
      "grad_norm": 2.9007060527801514,
      "learning_rate": 2.160550136832503e-05,
      "loss": 0.6998,
      "step": 9677
    },
    {
      "epoch": 3.0558888538048627,
      "grad_norm": 3.053295135498047,
      "learning_rate": 2.1601992842607538e-05,
      "loss": 0.7097,
      "step": 9678
    },
    {
      "epoch": 3.0562046100410485,
      "grad_norm": 3.5388336181640625,
      "learning_rate": 2.1598484316890045e-05,
      "loss": 0.7744,
      "step": 9679
    },
    {
      "epoch": 3.056520366277234,
      "grad_norm": 3.608727216720581,
      "learning_rate": 2.159497579117255e-05,
      "loss": 0.6194,
      "step": 9680
    },
    {
      "epoch": 3.0568361225134195,
      "grad_norm": 4.3659257888793945,
      "learning_rate": 2.1591467265455055e-05,
      "loss": 0.6834,
      "step": 9681
    },
    {
      "epoch": 3.0571518787496053,
      "grad_norm": 2.865905523300171,
      "learning_rate": 2.1587958739737562e-05,
      "loss": 0.6679,
      "step": 9682
    },
    {
      "epoch": 3.057467634985791,
      "grad_norm": 2.6739094257354736,
      "learning_rate": 2.158445021402007e-05,
      "loss": 0.6393,
      "step": 9683
    },
    {
      "epoch": 3.0577833912219767,
      "grad_norm": 3.0542075634002686,
      "learning_rate": 2.1580941688302575e-05,
      "loss": 0.6255,
      "step": 9684
    },
    {
      "epoch": 3.0580991474581625,
      "grad_norm": 2.801018714904785,
      "learning_rate": 2.1577433162585086e-05,
      "loss": 0.6672,
      "step": 9685
    },
    {
      "epoch": 3.0584149036943478,
      "grad_norm": 3.0983760356903076,
      "learning_rate": 2.157392463686759e-05,
      "loss": 0.6795,
      "step": 9686
    },
    {
      "epoch": 3.0587306599305335,
      "grad_norm": 2.9365506172180176,
      "learning_rate": 2.1570416111150096e-05,
      "loss": 0.6652,
      "step": 9687
    },
    {
      "epoch": 3.0590464161667192,
      "grad_norm": 3.403839588165283,
      "learning_rate": 2.1566907585432603e-05,
      "loss": 0.6706,
      "step": 9688
    },
    {
      "epoch": 3.059362172402905,
      "grad_norm": 2.9801182746887207,
      "learning_rate": 2.156339905971511e-05,
      "loss": 0.672,
      "step": 9689
    },
    {
      "epoch": 3.0596779286390907,
      "grad_norm": 2.7023870944976807,
      "learning_rate": 2.1559890533997616e-05,
      "loss": 0.7856,
      "step": 9690
    },
    {
      "epoch": 3.0599936848752765,
      "grad_norm": 3.006136417388916,
      "learning_rate": 2.155638200828012e-05,
      "loss": 0.702,
      "step": 9691
    },
    {
      "epoch": 3.0603094411114617,
      "grad_norm": 2.796168327331543,
      "learning_rate": 2.1552873482562626e-05,
      "loss": 0.6653,
      "step": 9692
    },
    {
      "epoch": 3.0606251973476475,
      "grad_norm": 3.3362717628479004,
      "learning_rate": 2.1549364956845133e-05,
      "loss": 0.6885,
      "step": 9693
    },
    {
      "epoch": 3.0609409535838332,
      "grad_norm": 4.052707195281982,
      "learning_rate": 2.1545856431127643e-05,
      "loss": 0.6436,
      "step": 9694
    },
    {
      "epoch": 3.061256709820019,
      "grad_norm": 2.8478708267211914,
      "learning_rate": 2.1542347905410147e-05,
      "loss": 0.6174,
      "step": 9695
    },
    {
      "epoch": 3.0615724660562047,
      "grad_norm": 2.715634346008301,
      "learning_rate": 2.1538839379692654e-05,
      "loss": 0.6552,
      "step": 9696
    },
    {
      "epoch": 3.0618882222923904,
      "grad_norm": 2.7165095806121826,
      "learning_rate": 2.153533085397516e-05,
      "loss": 0.653,
      "step": 9697
    },
    {
      "epoch": 3.0622039785285757,
      "grad_norm": 2.703615188598633,
      "learning_rate": 2.1531822328257667e-05,
      "loss": 0.6741,
      "step": 9698
    },
    {
      "epoch": 3.0625197347647615,
      "grad_norm": 3.089745044708252,
      "learning_rate": 2.1528313802540174e-05,
      "loss": 0.6323,
      "step": 9699
    },
    {
      "epoch": 3.062835491000947,
      "grad_norm": 2.7528433799743652,
      "learning_rate": 2.152480527682268e-05,
      "loss": 0.7079,
      "step": 9700
    },
    {
      "epoch": 3.063151247237133,
      "grad_norm": 2.57442307472229,
      "learning_rate": 2.1521296751105184e-05,
      "loss": 0.5807,
      "step": 9701
    },
    {
      "epoch": 3.0634670034733187,
      "grad_norm": 3.05263352394104,
      "learning_rate": 2.1517788225387695e-05,
      "loss": 0.6579,
      "step": 9702
    },
    {
      "epoch": 3.0637827597095044,
      "grad_norm": 2.98091983795166,
      "learning_rate": 2.15142796996702e-05,
      "loss": 0.6698,
      "step": 9703
    },
    {
      "epoch": 3.0640985159456897,
      "grad_norm": 2.994478940963745,
      "learning_rate": 2.1510771173952708e-05,
      "loss": 0.6832,
      "step": 9704
    },
    {
      "epoch": 3.0644142721818755,
      "grad_norm": 2.9702680110931396,
      "learning_rate": 2.150726264823521e-05,
      "loss": 0.6632,
      "step": 9705
    },
    {
      "epoch": 3.064730028418061,
      "grad_norm": 2.9567222595214844,
      "learning_rate": 2.150375412251772e-05,
      "loss": 0.6611,
      "step": 9706
    },
    {
      "epoch": 3.065045784654247,
      "grad_norm": 3.26708984375,
      "learning_rate": 2.1500245596800225e-05,
      "loss": 0.6885,
      "step": 9707
    },
    {
      "epoch": 3.0653615408904327,
      "grad_norm": 3.226184129714966,
      "learning_rate": 2.1496737071082732e-05,
      "loss": 0.673,
      "step": 9708
    },
    {
      "epoch": 3.0656772971266184,
      "grad_norm": 3.078472852706909,
      "learning_rate": 2.149322854536524e-05,
      "loss": 0.6042,
      "step": 9709
    },
    {
      "epoch": 3.0659930533628037,
      "grad_norm": 2.9430062770843506,
      "learning_rate": 2.1489720019647742e-05,
      "loss": 0.5851,
      "step": 9710
    },
    {
      "epoch": 3.0663088095989894,
      "grad_norm": 3.009984254837036,
      "learning_rate": 2.1486211493930253e-05,
      "loss": 0.6018,
      "step": 9711
    },
    {
      "epoch": 3.066624565835175,
      "grad_norm": 3.037487030029297,
      "learning_rate": 2.148270296821276e-05,
      "loss": 0.6725,
      "step": 9712
    },
    {
      "epoch": 3.066940322071361,
      "grad_norm": 3.2752022743225098,
      "learning_rate": 2.1479194442495266e-05,
      "loss": 0.6765,
      "step": 9713
    },
    {
      "epoch": 3.0672560783075467,
      "grad_norm": 2.6045615673065186,
      "learning_rate": 2.1475685916777773e-05,
      "loss": 0.5661,
      "step": 9714
    },
    {
      "epoch": 3.0675718345437324,
      "grad_norm": 3.2985575199127197,
      "learning_rate": 2.1472177391060276e-05,
      "loss": 0.7849,
      "step": 9715
    },
    {
      "epoch": 3.067887590779918,
      "grad_norm": 4.23221492767334,
      "learning_rate": 2.1468668865342783e-05,
      "loss": 0.6871,
      "step": 9716
    },
    {
      "epoch": 3.0682033470161034,
      "grad_norm": 3.0433847904205322,
      "learning_rate": 2.146516033962529e-05,
      "loss": 0.7372,
      "step": 9717
    },
    {
      "epoch": 3.068519103252289,
      "grad_norm": 2.880638599395752,
      "learning_rate": 2.1461651813907797e-05,
      "loss": 0.6461,
      "step": 9718
    },
    {
      "epoch": 3.068834859488475,
      "grad_norm": 3.729175329208374,
      "learning_rate": 2.1458143288190304e-05,
      "loss": 0.692,
      "step": 9719
    },
    {
      "epoch": 3.0691506157246606,
      "grad_norm": 2.8047611713409424,
      "learning_rate": 2.145463476247281e-05,
      "loss": 0.6179,
      "step": 9720
    },
    {
      "epoch": 3.0694663719608464,
      "grad_norm": 2.867440938949585,
      "learning_rate": 2.1451126236755317e-05,
      "loss": 0.6321,
      "step": 9721
    },
    {
      "epoch": 3.0697821281970317,
      "grad_norm": 3.145787477493286,
      "learning_rate": 2.1447617711037824e-05,
      "loss": 0.618,
      "step": 9722
    },
    {
      "epoch": 3.0700978844332174,
      "grad_norm": 2.8970401287078857,
      "learning_rate": 2.144410918532033e-05,
      "loss": 0.6876,
      "step": 9723
    },
    {
      "epoch": 3.070413640669403,
      "grad_norm": 3.2040016651153564,
      "learning_rate": 2.1440600659602834e-05,
      "loss": 0.6531,
      "step": 9724
    },
    {
      "epoch": 3.070729396905589,
      "grad_norm": 3.305572271347046,
      "learning_rate": 2.143709213388534e-05,
      "loss": 0.6223,
      "step": 9725
    },
    {
      "epoch": 3.0710451531417746,
      "grad_norm": 4.161295413970947,
      "learning_rate": 2.1433583608167848e-05,
      "loss": 0.6616,
      "step": 9726
    },
    {
      "epoch": 3.0713609093779604,
      "grad_norm": 3.023098945617676,
      "learning_rate": 2.1430075082450355e-05,
      "loss": 0.6796,
      "step": 9727
    },
    {
      "epoch": 3.071676665614146,
      "grad_norm": 3.3774795532226562,
      "learning_rate": 2.142656655673286e-05,
      "loss": 0.6615,
      "step": 9728
    },
    {
      "epoch": 3.0719924218503314,
      "grad_norm": 3.091378927230835,
      "learning_rate": 2.142305803101537e-05,
      "loss": 0.7137,
      "step": 9729
    },
    {
      "epoch": 3.072308178086517,
      "grad_norm": 3.0741372108459473,
      "learning_rate": 2.1419549505297875e-05,
      "loss": 0.7472,
      "step": 9730
    },
    {
      "epoch": 3.072623934322703,
      "grad_norm": 3.102987289428711,
      "learning_rate": 2.1416040979580382e-05,
      "loss": 0.6925,
      "step": 9731
    },
    {
      "epoch": 3.0729396905588886,
      "grad_norm": 3.979499578475952,
      "learning_rate": 2.141253245386289e-05,
      "loss": 0.6373,
      "step": 9732
    },
    {
      "epoch": 3.0732554467950743,
      "grad_norm": 2.920081615447998,
      "learning_rate": 2.1409023928145396e-05,
      "loss": 0.6378,
      "step": 9733
    },
    {
      "epoch": 3.07357120303126,
      "grad_norm": 3.177785873413086,
      "learning_rate": 2.14055154024279e-05,
      "loss": 0.6623,
      "step": 9734
    },
    {
      "epoch": 3.0738869592674454,
      "grad_norm": 3.280080556869507,
      "learning_rate": 2.1402006876710406e-05,
      "loss": 0.6724,
      "step": 9735
    },
    {
      "epoch": 3.074202715503631,
      "grad_norm": 4.049752235412598,
      "learning_rate": 2.1398498350992913e-05,
      "loss": 0.612,
      "step": 9736
    },
    {
      "epoch": 3.074518471739817,
      "grad_norm": 3.1503217220306396,
      "learning_rate": 2.1394989825275423e-05,
      "loss": 0.6114,
      "step": 9737
    },
    {
      "epoch": 3.0748342279760026,
      "grad_norm": 3.0363473892211914,
      "learning_rate": 2.1391481299557926e-05,
      "loss": 0.7011,
      "step": 9738
    },
    {
      "epoch": 3.0751499842121883,
      "grad_norm": 2.970588445663452,
      "learning_rate": 2.1387972773840433e-05,
      "loss": 0.6546,
      "step": 9739
    },
    {
      "epoch": 3.075465740448374,
      "grad_norm": 3.0147571563720703,
      "learning_rate": 2.138446424812294e-05,
      "loss": 0.6448,
      "step": 9740
    },
    {
      "epoch": 3.0757814966845594,
      "grad_norm": 3.2441868782043457,
      "learning_rate": 2.1380955722405447e-05,
      "loss": 0.658,
      "step": 9741
    },
    {
      "epoch": 3.076097252920745,
      "grad_norm": 3.0588459968566895,
      "learning_rate": 2.1377447196687954e-05,
      "loss": 0.7239,
      "step": 9742
    },
    {
      "epoch": 3.076413009156931,
      "grad_norm": 3.191218852996826,
      "learning_rate": 2.1373938670970457e-05,
      "loss": 0.6494,
      "step": 9743
    },
    {
      "epoch": 3.0767287653931166,
      "grad_norm": 3.011178493499756,
      "learning_rate": 2.1370430145252964e-05,
      "loss": 0.6585,
      "step": 9744
    },
    {
      "epoch": 3.0770445216293023,
      "grad_norm": 3.063181161880493,
      "learning_rate": 2.136692161953547e-05,
      "loss": 0.6253,
      "step": 9745
    },
    {
      "epoch": 3.077360277865488,
      "grad_norm": 3.348665475845337,
      "learning_rate": 2.136341309381798e-05,
      "loss": 0.6116,
      "step": 9746
    },
    {
      "epoch": 3.0776760341016733,
      "grad_norm": 3.2147879600524902,
      "learning_rate": 2.1359904568100488e-05,
      "loss": 0.6624,
      "step": 9747
    },
    {
      "epoch": 3.077991790337859,
      "grad_norm": 3.5007283687591553,
      "learning_rate": 2.135639604238299e-05,
      "loss": 0.6826,
      "step": 9748
    },
    {
      "epoch": 3.078307546574045,
      "grad_norm": 3.07448410987854,
      "learning_rate": 2.1352887516665498e-05,
      "loss": 0.6649,
      "step": 9749
    },
    {
      "epoch": 3.0786233028102306,
      "grad_norm": 3.2770614624023438,
      "learning_rate": 2.1349378990948005e-05,
      "loss": 0.715,
      "step": 9750
    },
    {
      "epoch": 3.0789390590464163,
      "grad_norm": 2.688957452774048,
      "learning_rate": 2.134587046523051e-05,
      "loss": 0.6486,
      "step": 9751
    },
    {
      "epoch": 3.079254815282602,
      "grad_norm": 3.397622585296631,
      "learning_rate": 2.134236193951302e-05,
      "loss": 0.6477,
      "step": 9752
    },
    {
      "epoch": 3.0795705715187873,
      "grad_norm": 2.716944932937622,
      "learning_rate": 2.1338853413795522e-05,
      "loss": 0.7071,
      "step": 9753
    },
    {
      "epoch": 3.079886327754973,
      "grad_norm": 3.1009976863861084,
      "learning_rate": 2.133534488807803e-05,
      "loss": 0.6708,
      "step": 9754
    },
    {
      "epoch": 3.080202083991159,
      "grad_norm": 2.6938469409942627,
      "learning_rate": 2.133183636236054e-05,
      "loss": 0.5912,
      "step": 9755
    },
    {
      "epoch": 3.0805178402273445,
      "grad_norm": 3.261357545852661,
      "learning_rate": 2.1328327836643046e-05,
      "loss": 0.8043,
      "step": 9756
    },
    {
      "epoch": 3.0808335964635303,
      "grad_norm": 2.8665435314178467,
      "learning_rate": 2.132481931092555e-05,
      "loss": 0.6737,
      "step": 9757
    },
    {
      "epoch": 3.081149352699716,
      "grad_norm": 3.155472755432129,
      "learning_rate": 2.1321310785208056e-05,
      "loss": 0.654,
      "step": 9758
    },
    {
      "epoch": 3.0814651089359013,
      "grad_norm": 3.30456280708313,
      "learning_rate": 2.1317802259490563e-05,
      "loss": 0.631,
      "step": 9759
    },
    {
      "epoch": 3.081780865172087,
      "grad_norm": 3.070479393005371,
      "learning_rate": 2.131429373377307e-05,
      "loss": 0.6406,
      "step": 9760
    },
    {
      "epoch": 3.082096621408273,
      "grad_norm": 3.157705783843994,
      "learning_rate": 2.1310785208055576e-05,
      "loss": 0.6481,
      "step": 9761
    },
    {
      "epoch": 3.0824123776444585,
      "grad_norm": 3.235574960708618,
      "learning_rate": 2.1307276682338083e-05,
      "loss": 0.656,
      "step": 9762
    },
    {
      "epoch": 3.0827281338806443,
      "grad_norm": 2.8556153774261475,
      "learning_rate": 2.1303768156620586e-05,
      "loss": 0.6007,
      "step": 9763
    },
    {
      "epoch": 3.08304389011683,
      "grad_norm": 2.992816686630249,
      "learning_rate": 2.1300259630903097e-05,
      "loss": 0.6593,
      "step": 9764
    },
    {
      "epoch": 3.0833596463530153,
      "grad_norm": 3.0744895935058594,
      "learning_rate": 2.1296751105185603e-05,
      "loss": 0.6565,
      "step": 9765
    },
    {
      "epoch": 3.083675402589201,
      "grad_norm": 2.815242052078247,
      "learning_rate": 2.129324257946811e-05,
      "loss": 0.5985,
      "step": 9766
    },
    {
      "epoch": 3.0839911588253868,
      "grad_norm": 3.010310173034668,
      "learning_rate": 2.1289734053750614e-05,
      "loss": 0.6454,
      "step": 9767
    },
    {
      "epoch": 3.0843069150615725,
      "grad_norm": 3.0138514041900635,
      "learning_rate": 2.128622552803312e-05,
      "loss": 0.6235,
      "step": 9768
    },
    {
      "epoch": 3.0846226712977582,
      "grad_norm": 2.544187545776367,
      "learning_rate": 2.1282717002315627e-05,
      "loss": 0.6258,
      "step": 9769
    },
    {
      "epoch": 3.084938427533944,
      "grad_norm": 2.7156240940093994,
      "learning_rate": 2.1279208476598134e-05,
      "loss": 0.6192,
      "step": 9770
    },
    {
      "epoch": 3.0852541837701293,
      "grad_norm": 3.8164565563201904,
      "learning_rate": 2.127569995088064e-05,
      "loss": 0.7175,
      "step": 9771
    },
    {
      "epoch": 3.085569940006315,
      "grad_norm": 2.806731939315796,
      "learning_rate": 2.1272191425163148e-05,
      "loss": 0.6082,
      "step": 9772
    },
    {
      "epoch": 3.0858856962425008,
      "grad_norm": 3.4350743293762207,
      "learning_rate": 2.1268682899445655e-05,
      "loss": 0.7268,
      "step": 9773
    },
    {
      "epoch": 3.0862014524786865,
      "grad_norm": 3.346261501312256,
      "learning_rate": 2.126517437372816e-05,
      "loss": 0.6914,
      "step": 9774
    },
    {
      "epoch": 3.0865172087148722,
      "grad_norm": 2.5449044704437256,
      "learning_rate": 2.1261665848010668e-05,
      "loss": 0.6608,
      "step": 9775
    },
    {
      "epoch": 3.086832964951058,
      "grad_norm": 2.8137447834014893,
      "learning_rate": 2.1258157322293175e-05,
      "loss": 0.6628,
      "step": 9776
    },
    {
      "epoch": 3.0871487211872433,
      "grad_norm": 2.6923294067382812,
      "learning_rate": 2.125464879657568e-05,
      "loss": 0.6767,
      "step": 9777
    },
    {
      "epoch": 3.087464477423429,
      "grad_norm": 2.991058826446533,
      "learning_rate": 2.1251140270858185e-05,
      "loss": 0.625,
      "step": 9778
    },
    {
      "epoch": 3.0877802336596147,
      "grad_norm": 2.711066961288452,
      "learning_rate": 2.1247631745140692e-05,
      "loss": 0.6895,
      "step": 9779
    },
    {
      "epoch": 3.0880959898958005,
      "grad_norm": 3.414607048034668,
      "learning_rate": 2.12441232194232e-05,
      "loss": 0.5793,
      "step": 9780
    },
    {
      "epoch": 3.088411746131986,
      "grad_norm": 3.0183091163635254,
      "learning_rate": 2.1240614693705706e-05,
      "loss": 0.6559,
      "step": 9781
    },
    {
      "epoch": 3.088727502368172,
      "grad_norm": 3.131182909011841,
      "learning_rate": 2.1237106167988213e-05,
      "loss": 0.6463,
      "step": 9782
    },
    {
      "epoch": 3.0890432586043572,
      "grad_norm": 2.7816975116729736,
      "learning_rate": 2.123359764227072e-05,
      "loss": 0.6027,
      "step": 9783
    },
    {
      "epoch": 3.089359014840543,
      "grad_norm": 3.0036323070526123,
      "learning_rate": 2.1230089116553226e-05,
      "loss": 0.6377,
      "step": 9784
    },
    {
      "epoch": 3.0896747710767287,
      "grad_norm": 3.373335599899292,
      "learning_rate": 2.1226580590835733e-05,
      "loss": 0.6517,
      "step": 9785
    },
    {
      "epoch": 3.0899905273129145,
      "grad_norm": 3.880375862121582,
      "learning_rate": 2.1223072065118236e-05,
      "loss": 0.7004,
      "step": 9786
    },
    {
      "epoch": 3.0903062835491,
      "grad_norm": 4.0190863609313965,
      "learning_rate": 2.1219563539400743e-05,
      "loss": 0.6354,
      "step": 9787
    },
    {
      "epoch": 3.090622039785286,
      "grad_norm": 3.4637904167175293,
      "learning_rate": 2.121605501368325e-05,
      "loss": 0.6605,
      "step": 9788
    },
    {
      "epoch": 3.0909377960214712,
      "grad_norm": 2.8160204887390137,
      "learning_rate": 2.1212546487965757e-05,
      "loss": 0.6286,
      "step": 9789
    },
    {
      "epoch": 3.091253552257657,
      "grad_norm": 4.833189487457275,
      "learning_rate": 2.1209037962248267e-05,
      "loss": 0.6911,
      "step": 9790
    },
    {
      "epoch": 3.0915693084938427,
      "grad_norm": 3.705836296081543,
      "learning_rate": 2.120552943653077e-05,
      "loss": 0.6924,
      "step": 9791
    },
    {
      "epoch": 3.0918850647300284,
      "grad_norm": 3.106459379196167,
      "learning_rate": 2.1202020910813277e-05,
      "loss": 0.6526,
      "step": 9792
    },
    {
      "epoch": 3.092200820966214,
      "grad_norm": 2.6086108684539795,
      "learning_rate": 2.1198512385095784e-05,
      "loss": 0.623,
      "step": 9793
    },
    {
      "epoch": 3.0925165772024,
      "grad_norm": 2.780651569366455,
      "learning_rate": 2.119500385937829e-05,
      "loss": 0.6501,
      "step": 9794
    },
    {
      "epoch": 3.092832333438585,
      "grad_norm": 2.5692574977874756,
      "learning_rate": 2.1191495333660798e-05,
      "loss": 0.6529,
      "step": 9795
    },
    {
      "epoch": 3.093148089674771,
      "grad_norm": 3.2761521339416504,
      "learning_rate": 2.11879868079433e-05,
      "loss": 0.72,
      "step": 9796
    },
    {
      "epoch": 3.0934638459109567,
      "grad_norm": 3.1282811164855957,
      "learning_rate": 2.1184478282225808e-05,
      "loss": 0.6619,
      "step": 9797
    },
    {
      "epoch": 3.0937796021471424,
      "grad_norm": 2.837263822555542,
      "learning_rate": 2.1180969756508318e-05,
      "loss": 0.6237,
      "step": 9798
    },
    {
      "epoch": 3.094095358383328,
      "grad_norm": 3.1135191917419434,
      "learning_rate": 2.1177461230790825e-05,
      "loss": 0.6708,
      "step": 9799
    },
    {
      "epoch": 3.094411114619514,
      "grad_norm": 3.1005945205688477,
      "learning_rate": 2.117395270507333e-05,
      "loss": 0.6553,
      "step": 9800
    },
    {
      "epoch": 3.094411114619514,
      "eval_loss": 0.8121901750564575,
      "eval_runtime": 4.5954,
      "eval_samples_per_second": 541.843,
      "eval_steps_per_second": 4.57,
      "step": 9800
    },
    {
      "epoch": 3.094726870855699,
      "grad_norm": 3.0395193099975586,
      "learning_rate": 2.1170444179355835e-05,
      "loss": 0.6664,
      "step": 9801
    },
    {
      "epoch": 3.095042627091885,
      "grad_norm": 2.9896631240844727,
      "learning_rate": 2.1166935653638342e-05,
      "loss": 0.6805,
      "step": 9802
    },
    {
      "epoch": 3.0953583833280707,
      "grad_norm": 2.7803544998168945,
      "learning_rate": 2.116342712792085e-05,
      "loss": 0.7602,
      "step": 9803
    },
    {
      "epoch": 3.0956741395642564,
      "grad_norm": 3.051438093185425,
      "learning_rate": 2.1159918602203356e-05,
      "loss": 0.6569,
      "step": 9804
    },
    {
      "epoch": 3.095989895800442,
      "grad_norm": 3.1620922088623047,
      "learning_rate": 2.1156410076485862e-05,
      "loss": 0.6763,
      "step": 9805
    },
    {
      "epoch": 3.096305652036628,
      "grad_norm": 3.3443291187286377,
      "learning_rate": 2.1152901550768366e-05,
      "loss": 0.7147,
      "step": 9806
    },
    {
      "epoch": 3.096621408272813,
      "grad_norm": 2.695002317428589,
      "learning_rate": 2.1149393025050876e-05,
      "loss": 0.666,
      "step": 9807
    },
    {
      "epoch": 3.096937164508999,
      "grad_norm": 3.1264610290527344,
      "learning_rate": 2.1145884499333383e-05,
      "loss": 0.6128,
      "step": 9808
    },
    {
      "epoch": 3.0972529207451847,
      "grad_norm": 2.9238779544830322,
      "learning_rate": 2.114237597361589e-05,
      "loss": 0.6322,
      "step": 9809
    },
    {
      "epoch": 3.0975686769813704,
      "grad_norm": 2.939746141433716,
      "learning_rate": 2.1138867447898393e-05,
      "loss": 0.6321,
      "step": 9810
    },
    {
      "epoch": 3.097884433217556,
      "grad_norm": 2.83288836479187,
      "learning_rate": 2.11353589221809e-05,
      "loss": 0.6106,
      "step": 9811
    },
    {
      "epoch": 3.098200189453742,
      "grad_norm": 2.9361207485198975,
      "learning_rate": 2.1131850396463407e-05,
      "loss": 0.6983,
      "step": 9812
    },
    {
      "epoch": 3.098515945689927,
      "grad_norm": 3.5659966468811035,
      "learning_rate": 2.1128341870745914e-05,
      "loss": 0.7132,
      "step": 9813
    },
    {
      "epoch": 3.098831701926113,
      "grad_norm": 3.0681087970733643,
      "learning_rate": 2.112483334502842e-05,
      "loss": 0.672,
      "step": 9814
    },
    {
      "epoch": 3.0991474581622986,
      "grad_norm": 2.9914352893829346,
      "learning_rate": 2.1121324819310924e-05,
      "loss": 0.7118,
      "step": 9815
    },
    {
      "epoch": 3.0994632143984844,
      "grad_norm": 2.7325849533081055,
      "learning_rate": 2.1117816293593434e-05,
      "loss": 0.5885,
      "step": 9816
    },
    {
      "epoch": 3.09977897063467,
      "grad_norm": 3.1459667682647705,
      "learning_rate": 2.111430776787594e-05,
      "loss": 0.6457,
      "step": 9817
    },
    {
      "epoch": 3.100094726870856,
      "grad_norm": 2.9497106075286865,
      "learning_rate": 2.1110799242158448e-05,
      "loss": 0.6324,
      "step": 9818
    },
    {
      "epoch": 3.1004104831070416,
      "grad_norm": 2.588430881500244,
      "learning_rate": 2.1107290716440954e-05,
      "loss": 0.6112,
      "step": 9819
    },
    {
      "epoch": 3.100726239343227,
      "grad_norm": 3.32712721824646,
      "learning_rate": 2.1103782190723458e-05,
      "loss": 0.6598,
      "step": 9820
    },
    {
      "epoch": 3.1010419955794126,
      "grad_norm": 2.4801535606384277,
      "learning_rate": 2.1100273665005965e-05,
      "loss": 0.6099,
      "step": 9821
    },
    {
      "epoch": 3.1013577518155984,
      "grad_norm": 2.8005881309509277,
      "learning_rate": 2.109676513928847e-05,
      "loss": 0.6095,
      "step": 9822
    },
    {
      "epoch": 3.101673508051784,
      "grad_norm": 3.1924686431884766,
      "learning_rate": 2.109325661357098e-05,
      "loss": 0.6936,
      "step": 9823
    },
    {
      "epoch": 3.10198926428797,
      "grad_norm": 3.41544246673584,
      "learning_rate": 2.1089748087853485e-05,
      "loss": 0.6313,
      "step": 9824
    },
    {
      "epoch": 3.102305020524155,
      "grad_norm": 3.5018081665039062,
      "learning_rate": 2.1086239562135992e-05,
      "loss": 0.5927,
      "step": 9825
    },
    {
      "epoch": 3.102620776760341,
      "grad_norm": 3.5582990646362305,
      "learning_rate": 2.10827310364185e-05,
      "loss": 0.618,
      "step": 9826
    },
    {
      "epoch": 3.1029365329965266,
      "grad_norm": 2.6714961528778076,
      "learning_rate": 2.1079222510701006e-05,
      "loss": 0.619,
      "step": 9827
    },
    {
      "epoch": 3.1032522892327123,
      "grad_norm": 3.2026078701019287,
      "learning_rate": 2.1075713984983512e-05,
      "loss": 0.6557,
      "step": 9828
    },
    {
      "epoch": 3.103568045468898,
      "grad_norm": 2.961639642715454,
      "learning_rate": 2.1072205459266016e-05,
      "loss": 0.5813,
      "step": 9829
    },
    {
      "epoch": 3.103883801705084,
      "grad_norm": 3.2244224548339844,
      "learning_rate": 2.1068696933548523e-05,
      "loss": 0.5785,
      "step": 9830
    },
    {
      "epoch": 3.1041995579412696,
      "grad_norm": 3.1014702320098877,
      "learning_rate": 2.106518840783103e-05,
      "loss": 0.6855,
      "step": 9831
    },
    {
      "epoch": 3.104515314177455,
      "grad_norm": 2.8737056255340576,
      "learning_rate": 2.1061679882113536e-05,
      "loss": 0.6874,
      "step": 9832
    },
    {
      "epoch": 3.1048310704136406,
      "grad_norm": 3.0578908920288086,
      "learning_rate": 2.1058171356396046e-05,
      "loss": 0.6459,
      "step": 9833
    },
    {
      "epoch": 3.1051468266498263,
      "grad_norm": 3.052163600921631,
      "learning_rate": 2.105466283067855e-05,
      "loss": 0.7082,
      "step": 9834
    },
    {
      "epoch": 3.105462582886012,
      "grad_norm": 3.2787790298461914,
      "learning_rate": 2.1051154304961057e-05,
      "loss": 0.7361,
      "step": 9835
    },
    {
      "epoch": 3.105778339122198,
      "grad_norm": 2.8697710037231445,
      "learning_rate": 2.1047645779243564e-05,
      "loss": 0.6653,
      "step": 9836
    },
    {
      "epoch": 3.106094095358383,
      "grad_norm": 2.9709014892578125,
      "learning_rate": 2.104413725352607e-05,
      "loss": 0.6065,
      "step": 9837
    },
    {
      "epoch": 3.106409851594569,
      "grad_norm": 2.9452784061431885,
      "learning_rate": 2.1040628727808577e-05,
      "loss": 0.7056,
      "step": 9838
    },
    {
      "epoch": 3.1067256078307546,
      "grad_norm": 3.967409133911133,
      "learning_rate": 2.103712020209108e-05,
      "loss": 0.7421,
      "step": 9839
    },
    {
      "epoch": 3.1070413640669403,
      "grad_norm": 2.9937517642974854,
      "learning_rate": 2.1033611676373587e-05,
      "loss": 0.6347,
      "step": 9840
    },
    {
      "epoch": 3.107357120303126,
      "grad_norm": 3.7934389114379883,
      "learning_rate": 2.1030103150656094e-05,
      "loss": 0.6919,
      "step": 9841
    },
    {
      "epoch": 3.107672876539312,
      "grad_norm": 3.090160369873047,
      "learning_rate": 2.1026594624938604e-05,
      "loss": 0.7226,
      "step": 9842
    },
    {
      "epoch": 3.1079886327754975,
      "grad_norm": 3.019634246826172,
      "learning_rate": 2.1023086099221108e-05,
      "loss": 0.6835,
      "step": 9843
    },
    {
      "epoch": 3.108304389011683,
      "grad_norm": 3.1089508533477783,
      "learning_rate": 2.1019577573503615e-05,
      "loss": 0.6543,
      "step": 9844
    },
    {
      "epoch": 3.1086201452478686,
      "grad_norm": 3.3918094635009766,
      "learning_rate": 2.101606904778612e-05,
      "loss": 0.6369,
      "step": 9845
    },
    {
      "epoch": 3.1089359014840543,
      "grad_norm": 3.190687894821167,
      "learning_rate": 2.1012560522068628e-05,
      "loss": 0.6011,
      "step": 9846
    },
    {
      "epoch": 3.10925165772024,
      "grad_norm": 2.982286214828491,
      "learning_rate": 2.1009051996351135e-05,
      "loss": 0.5886,
      "step": 9847
    },
    {
      "epoch": 3.1095674139564258,
      "grad_norm": 3.092322826385498,
      "learning_rate": 2.100554347063364e-05,
      "loss": 0.7236,
      "step": 9848
    },
    {
      "epoch": 3.1098831701926115,
      "grad_norm": 2.8250362873077393,
      "learning_rate": 2.1002034944916145e-05,
      "loss": 0.6024,
      "step": 9849
    },
    {
      "epoch": 3.110198926428797,
      "grad_norm": 3.056272029876709,
      "learning_rate": 2.0998526419198652e-05,
      "loss": 0.6168,
      "step": 9850
    },
    {
      "epoch": 3.1105146826649825,
      "grad_norm": 3.3272695541381836,
      "learning_rate": 2.0995017893481162e-05,
      "loss": 0.6896,
      "step": 9851
    },
    {
      "epoch": 3.1108304389011683,
      "grad_norm": 3.2697012424468994,
      "learning_rate": 2.099150936776367e-05,
      "loss": 0.6301,
      "step": 9852
    },
    {
      "epoch": 3.111146195137354,
      "grad_norm": 3.1081202030181885,
      "learning_rate": 2.0988000842046173e-05,
      "loss": 0.6344,
      "step": 9853
    },
    {
      "epoch": 3.1114619513735398,
      "grad_norm": 3.206862688064575,
      "learning_rate": 2.098449231632868e-05,
      "loss": 0.6098,
      "step": 9854
    },
    {
      "epoch": 3.1117777076097255,
      "grad_norm": 2.9984047412872314,
      "learning_rate": 2.0980983790611186e-05,
      "loss": 0.6531,
      "step": 9855
    },
    {
      "epoch": 3.112093463845911,
      "grad_norm": 3.2040576934814453,
      "learning_rate": 2.0977475264893693e-05,
      "loss": 0.658,
      "step": 9856
    },
    {
      "epoch": 3.1124092200820965,
      "grad_norm": 2.8558201789855957,
      "learning_rate": 2.09739667391762e-05,
      "loss": 0.6924,
      "step": 9857
    },
    {
      "epoch": 3.1127249763182823,
      "grad_norm": 2.873162269592285,
      "learning_rate": 2.0970458213458703e-05,
      "loss": 0.6357,
      "step": 9858
    },
    {
      "epoch": 3.113040732554468,
      "grad_norm": 3.360114812850952,
      "learning_rate": 2.096694968774121e-05,
      "loss": 0.7033,
      "step": 9859
    },
    {
      "epoch": 3.1133564887906537,
      "grad_norm": 2.610403060913086,
      "learning_rate": 2.096344116202372e-05,
      "loss": 0.6405,
      "step": 9860
    },
    {
      "epoch": 3.1136722450268395,
      "grad_norm": 3.439767599105835,
      "learning_rate": 2.0959932636306227e-05,
      "loss": 0.7525,
      "step": 9861
    },
    {
      "epoch": 3.1139880012630248,
      "grad_norm": 2.454904317855835,
      "learning_rate": 2.095642411058873e-05,
      "loss": 0.6633,
      "step": 9862
    },
    {
      "epoch": 3.1143037574992105,
      "grad_norm": 3.261481523513794,
      "learning_rate": 2.0952915584871237e-05,
      "loss": 0.7026,
      "step": 9863
    },
    {
      "epoch": 3.1146195137353963,
      "grad_norm": 2.7357094287872314,
      "learning_rate": 2.0949407059153744e-05,
      "loss": 0.6567,
      "step": 9864
    },
    {
      "epoch": 3.114935269971582,
      "grad_norm": 2.7203547954559326,
      "learning_rate": 2.094589853343625e-05,
      "loss": 0.6457,
      "step": 9865
    },
    {
      "epoch": 3.1152510262077677,
      "grad_norm": 3.3417887687683105,
      "learning_rate": 2.0942390007718758e-05,
      "loss": 0.7386,
      "step": 9866
    },
    {
      "epoch": 3.1155667824439535,
      "grad_norm": 4.29460334777832,
      "learning_rate": 2.0938881482001265e-05,
      "loss": 0.6536,
      "step": 9867
    },
    {
      "epoch": 3.1158825386801388,
      "grad_norm": 3.2124083042144775,
      "learning_rate": 2.093537295628377e-05,
      "loss": 0.7741,
      "step": 9868
    },
    {
      "epoch": 3.1161982949163245,
      "grad_norm": 3.508098602294922,
      "learning_rate": 2.0931864430566278e-05,
      "loss": 0.7308,
      "step": 9869
    },
    {
      "epoch": 3.1165140511525102,
      "grad_norm": 3.006117105484009,
      "learning_rate": 2.0928355904848785e-05,
      "loss": 0.591,
      "step": 9870
    },
    {
      "epoch": 3.116829807388696,
      "grad_norm": 3.1906323432922363,
      "learning_rate": 2.0924847379131292e-05,
      "loss": 0.6426,
      "step": 9871
    },
    {
      "epoch": 3.1171455636248817,
      "grad_norm": 2.9937326908111572,
      "learning_rate": 2.0921338853413795e-05,
      "loss": 0.6043,
      "step": 9872
    },
    {
      "epoch": 3.1174613198610674,
      "grad_norm": 2.875303268432617,
      "learning_rate": 2.0917830327696302e-05,
      "loss": 0.6844,
      "step": 9873
    },
    {
      "epoch": 3.1177770760972527,
      "grad_norm": 3.2032523155212402,
      "learning_rate": 2.091432180197881e-05,
      "loss": 0.5687,
      "step": 9874
    },
    {
      "epoch": 3.1180928323334385,
      "grad_norm": 3.140916109085083,
      "learning_rate": 2.0910813276261316e-05,
      "loss": 0.6012,
      "step": 9875
    },
    {
      "epoch": 3.118408588569624,
      "grad_norm": 2.8851654529571533,
      "learning_rate": 2.0907304750543822e-05,
      "loss": 0.6414,
      "step": 9876
    },
    {
      "epoch": 3.11872434480581,
      "grad_norm": 3.000746250152588,
      "learning_rate": 2.090379622482633e-05,
      "loss": 0.7028,
      "step": 9877
    },
    {
      "epoch": 3.1190401010419957,
      "grad_norm": 3.0701122283935547,
      "learning_rate": 2.0900287699108836e-05,
      "loss": 0.7094,
      "step": 9878
    },
    {
      "epoch": 3.1193558572781814,
      "grad_norm": 3.1003201007843018,
      "learning_rate": 2.0896779173391343e-05,
      "loss": 0.6568,
      "step": 9879
    },
    {
      "epoch": 3.1196716135143667,
      "grad_norm": 2.7249629497528076,
      "learning_rate": 2.089327064767385e-05,
      "loss": 0.5647,
      "step": 9880
    },
    {
      "epoch": 3.1199873697505525,
      "grad_norm": 3.3172638416290283,
      "learning_rate": 2.0889762121956357e-05,
      "loss": 0.5571,
      "step": 9881
    },
    {
      "epoch": 3.120303125986738,
      "grad_norm": 2.8174099922180176,
      "learning_rate": 2.088625359623886e-05,
      "loss": 0.6845,
      "step": 9882
    },
    {
      "epoch": 3.120618882222924,
      "grad_norm": 3.0689878463745117,
      "learning_rate": 2.0882745070521367e-05,
      "loss": 0.6843,
      "step": 9883
    },
    {
      "epoch": 3.1209346384591097,
      "grad_norm": 3.5928616523742676,
      "learning_rate": 2.0879236544803874e-05,
      "loss": 0.6527,
      "step": 9884
    },
    {
      "epoch": 3.1212503946952954,
      "grad_norm": 3.124638080596924,
      "learning_rate": 2.087572801908638e-05,
      "loss": 0.6737,
      "step": 9885
    },
    {
      "epoch": 3.1215661509314807,
      "grad_norm": 3.375126361846924,
      "learning_rate": 2.0872219493368887e-05,
      "loss": 0.6341,
      "step": 9886
    },
    {
      "epoch": 3.1218819071676664,
      "grad_norm": 3.18485951423645,
      "learning_rate": 2.0868710967651394e-05,
      "loss": 0.6771,
      "step": 9887
    },
    {
      "epoch": 3.122197663403852,
      "grad_norm": 2.5952610969543457,
      "learning_rate": 2.08652024419339e-05,
      "loss": 0.6163,
      "step": 9888
    },
    {
      "epoch": 3.122513419640038,
      "grad_norm": 2.7608230113983154,
      "learning_rate": 2.0861693916216408e-05,
      "loss": 0.6668,
      "step": 9889
    },
    {
      "epoch": 3.1228291758762237,
      "grad_norm": 3.0087878704071045,
      "learning_rate": 2.0858185390498914e-05,
      "loss": 0.6482,
      "step": 9890
    },
    {
      "epoch": 3.1231449321124094,
      "grad_norm": 3.0400500297546387,
      "learning_rate": 2.0854676864781418e-05,
      "loss": 0.7125,
      "step": 9891
    },
    {
      "epoch": 3.1234606883485947,
      "grad_norm": 3.4284396171569824,
      "learning_rate": 2.0851168339063925e-05,
      "loss": 0.6798,
      "step": 9892
    },
    {
      "epoch": 3.1237764445847804,
      "grad_norm": 2.7082669734954834,
      "learning_rate": 2.084765981334643e-05,
      "loss": 0.6103,
      "step": 9893
    },
    {
      "epoch": 3.124092200820966,
      "grad_norm": 3.0831682682037354,
      "learning_rate": 2.084415128762894e-05,
      "loss": 0.6382,
      "step": 9894
    },
    {
      "epoch": 3.124407957057152,
      "grad_norm": 3.0426764488220215,
      "learning_rate": 2.084064276191145e-05,
      "loss": 0.6994,
      "step": 9895
    },
    {
      "epoch": 3.1247237132933376,
      "grad_norm": 2.6934609413146973,
      "learning_rate": 2.0837134236193952e-05,
      "loss": 0.6675,
      "step": 9896
    },
    {
      "epoch": 3.1250394695295234,
      "grad_norm": 2.659111738204956,
      "learning_rate": 2.083362571047646e-05,
      "loss": 0.6806,
      "step": 9897
    },
    {
      "epoch": 3.1253552257657087,
      "grad_norm": 2.891388416290283,
      "learning_rate": 2.0830117184758966e-05,
      "loss": 0.6136,
      "step": 9898
    },
    {
      "epoch": 3.1256709820018944,
      "grad_norm": 3.173365831375122,
      "learning_rate": 2.0826608659041472e-05,
      "loss": 0.7215,
      "step": 9899
    },
    {
      "epoch": 3.12598673823808,
      "grad_norm": 3.193814516067505,
      "learning_rate": 2.082310013332398e-05,
      "loss": 0.664,
      "step": 9900
    },
    {
      "epoch": 3.126302494474266,
      "grad_norm": 3.3351075649261475,
      "learning_rate": 2.0819591607606483e-05,
      "loss": 0.7072,
      "step": 9901
    },
    {
      "epoch": 3.1266182507104516,
      "grad_norm": 2.9465043544769287,
      "learning_rate": 2.081608308188899e-05,
      "loss": 0.6788,
      "step": 9902
    },
    {
      "epoch": 3.1269340069466374,
      "grad_norm": 2.717120885848999,
      "learning_rate": 2.08125745561715e-05,
      "loss": 0.685,
      "step": 9903
    },
    {
      "epoch": 3.1272497631828227,
      "grad_norm": 3.3156259059906006,
      "learning_rate": 2.0809066030454006e-05,
      "loss": 0.7792,
      "step": 9904
    },
    {
      "epoch": 3.1275655194190084,
      "grad_norm": 2.912389039993286,
      "learning_rate": 2.080555750473651e-05,
      "loss": 0.639,
      "step": 9905
    },
    {
      "epoch": 3.127881275655194,
      "grad_norm": 3.163630485534668,
      "learning_rate": 2.0802048979019017e-05,
      "loss": 0.7506,
      "step": 9906
    },
    {
      "epoch": 3.12819703189138,
      "grad_norm": 3.143641948699951,
      "learning_rate": 2.0798540453301524e-05,
      "loss": 0.643,
      "step": 9907
    },
    {
      "epoch": 3.1285127881275656,
      "grad_norm": 3.4117400646209717,
      "learning_rate": 2.079503192758403e-05,
      "loss": 0.7165,
      "step": 9908
    },
    {
      "epoch": 3.1288285443637514,
      "grad_norm": 3.201970100402832,
      "learning_rate": 2.0791523401866537e-05,
      "loss": 0.6794,
      "step": 9909
    },
    {
      "epoch": 3.1291443005999366,
      "grad_norm": 2.9712119102478027,
      "learning_rate": 2.0788014876149044e-05,
      "loss": 0.7254,
      "step": 9910
    },
    {
      "epoch": 3.1294600568361224,
      "grad_norm": 3.3224778175354004,
      "learning_rate": 2.0784506350431547e-05,
      "loss": 0.6801,
      "step": 9911
    },
    {
      "epoch": 3.129775813072308,
      "grad_norm": 3.096626043319702,
      "learning_rate": 2.0780997824714058e-05,
      "loss": 0.6913,
      "step": 9912
    },
    {
      "epoch": 3.130091569308494,
      "grad_norm": 3.1332597732543945,
      "learning_rate": 2.0777489298996564e-05,
      "loss": 0.6498,
      "step": 9913
    },
    {
      "epoch": 3.1304073255446796,
      "grad_norm": 2.923128128051758,
      "learning_rate": 2.077398077327907e-05,
      "loss": 0.7254,
      "step": 9914
    },
    {
      "epoch": 3.1307230817808653,
      "grad_norm": 2.8849985599517822,
      "learning_rate": 2.0770472247561575e-05,
      "loss": 0.6348,
      "step": 9915
    },
    {
      "epoch": 3.1310388380170506,
      "grad_norm": 3.175339937210083,
      "learning_rate": 2.076696372184408e-05,
      "loss": 0.7145,
      "step": 9916
    },
    {
      "epoch": 3.1313545942532364,
      "grad_norm": 2.8040010929107666,
      "learning_rate": 2.0763455196126588e-05,
      "loss": 0.629,
      "step": 9917
    },
    {
      "epoch": 3.131670350489422,
      "grad_norm": 3.0326743125915527,
      "learning_rate": 2.0759946670409095e-05,
      "loss": 0.7514,
      "step": 9918
    },
    {
      "epoch": 3.131986106725608,
      "grad_norm": 2.9722275733947754,
      "learning_rate": 2.0756438144691602e-05,
      "loss": 0.6543,
      "step": 9919
    },
    {
      "epoch": 3.1323018629617936,
      "grad_norm": 2.7883992195129395,
      "learning_rate": 2.0752929618974105e-05,
      "loss": 0.6424,
      "step": 9920
    },
    {
      "epoch": 3.1326176191979793,
      "grad_norm": 2.873659610748291,
      "learning_rate": 2.0749421093256615e-05,
      "loss": 0.6183,
      "step": 9921
    },
    {
      "epoch": 3.132933375434165,
      "grad_norm": 2.7246100902557373,
      "learning_rate": 2.0745912567539122e-05,
      "loss": 0.6436,
      "step": 9922
    },
    {
      "epoch": 3.1332491316703504,
      "grad_norm": 3.206864595413208,
      "learning_rate": 2.074240404182163e-05,
      "loss": 0.637,
      "step": 9923
    },
    {
      "epoch": 3.133564887906536,
      "grad_norm": 2.827866315841675,
      "learning_rate": 2.0738895516104136e-05,
      "loss": 0.6738,
      "step": 9924
    },
    {
      "epoch": 3.133880644142722,
      "grad_norm": 3.397625207901001,
      "learning_rate": 2.073538699038664e-05,
      "loss": 0.7622,
      "step": 9925
    },
    {
      "epoch": 3.1341964003789076,
      "grad_norm": 2.894932746887207,
      "learning_rate": 2.0731878464669146e-05,
      "loss": 0.7291,
      "step": 9926
    },
    {
      "epoch": 3.1345121566150933,
      "grad_norm": 2.9668800830841064,
      "learning_rate": 2.0728369938951653e-05,
      "loss": 0.6826,
      "step": 9927
    },
    {
      "epoch": 3.1348279128512786,
      "grad_norm": 2.7061190605163574,
      "learning_rate": 2.072486141323416e-05,
      "loss": 0.6009,
      "step": 9928
    },
    {
      "epoch": 3.1351436690874643,
      "grad_norm": 2.8950531482696533,
      "learning_rate": 2.0721352887516667e-05,
      "loss": 0.652,
      "step": 9929
    },
    {
      "epoch": 3.13545942532365,
      "grad_norm": 2.7124414443969727,
      "learning_rate": 2.0717844361799173e-05,
      "loss": 0.5884,
      "step": 9930
    },
    {
      "epoch": 3.135775181559836,
      "grad_norm": 3.082362413406372,
      "learning_rate": 2.071433583608168e-05,
      "loss": 0.7774,
      "step": 9931
    },
    {
      "epoch": 3.1360909377960215,
      "grad_norm": 2.9908437728881836,
      "learning_rate": 2.0710827310364187e-05,
      "loss": 0.6801,
      "step": 9932
    },
    {
      "epoch": 3.1364066940322073,
      "grad_norm": 3.011679172515869,
      "learning_rate": 2.0707318784646694e-05,
      "loss": 0.6581,
      "step": 9933
    },
    {
      "epoch": 3.136722450268393,
      "grad_norm": 2.7633471488952637,
      "learning_rate": 2.0703810258929197e-05,
      "loss": 0.6182,
      "step": 9934
    },
    {
      "epoch": 3.1370382065045783,
      "grad_norm": 2.7719104290008545,
      "learning_rate": 2.0700301733211704e-05,
      "loss": 0.5813,
      "step": 9935
    },
    {
      "epoch": 3.137353962740764,
      "grad_norm": 3.368525266647339,
      "learning_rate": 2.069679320749421e-05,
      "loss": 0.6011,
      "step": 9936
    },
    {
      "epoch": 3.13766971897695,
      "grad_norm": 2.544389009475708,
      "learning_rate": 2.0693284681776718e-05,
      "loss": 0.6727,
      "step": 9937
    },
    {
      "epoch": 3.1379854752131355,
      "grad_norm": 2.7705540657043457,
      "learning_rate": 2.0689776156059228e-05,
      "loss": 0.6001,
      "step": 9938
    },
    {
      "epoch": 3.1383012314493213,
      "grad_norm": 2.800987482070923,
      "learning_rate": 2.068626763034173e-05,
      "loss": 0.6661,
      "step": 9939
    },
    {
      "epoch": 3.1386169876855066,
      "grad_norm": 2.7937169075012207,
      "learning_rate": 2.0682759104624238e-05,
      "loss": 0.7042,
      "step": 9940
    },
    {
      "epoch": 3.1389327439216923,
      "grad_norm": 2.947192907333374,
      "learning_rate": 2.0679250578906745e-05,
      "loss": 0.7074,
      "step": 9941
    },
    {
      "epoch": 3.139248500157878,
      "grad_norm": 3.0660250186920166,
      "learning_rate": 2.0675742053189252e-05,
      "loss": 0.6752,
      "step": 9942
    },
    {
      "epoch": 3.139564256394064,
      "grad_norm": 3.041609764099121,
      "learning_rate": 2.067223352747176e-05,
      "loss": 0.713,
      "step": 9943
    },
    {
      "epoch": 3.1398800126302495,
      "grad_norm": 2.9115312099456787,
      "learning_rate": 2.0668725001754262e-05,
      "loss": 0.6197,
      "step": 9944
    },
    {
      "epoch": 3.1401957688664353,
      "grad_norm": 2.8147096633911133,
      "learning_rate": 2.066521647603677e-05,
      "loss": 0.6153,
      "step": 9945
    },
    {
      "epoch": 3.140511525102621,
      "grad_norm": 3.132982015609741,
      "learning_rate": 2.0661707950319276e-05,
      "loss": 0.701,
      "step": 9946
    },
    {
      "epoch": 3.1408272813388063,
      "grad_norm": 2.7531044483184814,
      "learning_rate": 2.0658199424601786e-05,
      "loss": 0.5584,
      "step": 9947
    },
    {
      "epoch": 3.141143037574992,
      "grad_norm": 4.29916524887085,
      "learning_rate": 2.065469089888429e-05,
      "loss": 0.7674,
      "step": 9948
    },
    {
      "epoch": 3.1414587938111778,
      "grad_norm": 3.3893704414367676,
      "learning_rate": 2.0651182373166796e-05,
      "loss": 0.6792,
      "step": 9949
    },
    {
      "epoch": 3.1417745500473635,
      "grad_norm": 3.06136155128479,
      "learning_rate": 2.0647673847449303e-05,
      "loss": 0.669,
      "step": 9950
    },
    {
      "epoch": 3.1420903062835492,
      "grad_norm": 2.8994669914245605,
      "learning_rate": 2.064416532173181e-05,
      "loss": 0.5893,
      "step": 9951
    },
    {
      "epoch": 3.1424060625197345,
      "grad_norm": 5.879748821258545,
      "learning_rate": 2.0640656796014317e-05,
      "loss": 0.6351,
      "step": 9952
    },
    {
      "epoch": 3.1427218187559203,
      "grad_norm": 2.994997501373291,
      "learning_rate": 2.063714827029682e-05,
      "loss": 0.7154,
      "step": 9953
    },
    {
      "epoch": 3.143037574992106,
      "grad_norm": 2.989177942276001,
      "learning_rate": 2.0633639744579327e-05,
      "loss": 0.7133,
      "step": 9954
    },
    {
      "epoch": 3.1433533312282917,
      "grad_norm": 3.228724479675293,
      "learning_rate": 2.0630131218861834e-05,
      "loss": 0.7402,
      "step": 9955
    },
    {
      "epoch": 3.1436690874644775,
      "grad_norm": 2.861231565475464,
      "learning_rate": 2.0626622693144344e-05,
      "loss": 0.6726,
      "step": 9956
    },
    {
      "epoch": 3.1439848437006632,
      "grad_norm": 3.462773561477661,
      "learning_rate": 2.062311416742685e-05,
      "loss": 0.6215,
      "step": 9957
    },
    {
      "epoch": 3.144300599936849,
      "grad_norm": 3.212916612625122,
      "learning_rate": 2.0619605641709354e-05,
      "loss": 0.6537,
      "step": 9958
    },
    {
      "epoch": 3.1446163561730343,
      "grad_norm": 2.961494207382202,
      "learning_rate": 2.061609711599186e-05,
      "loss": 0.5821,
      "step": 9959
    },
    {
      "epoch": 3.14493211240922,
      "grad_norm": 2.852304458618164,
      "learning_rate": 2.0612588590274368e-05,
      "loss": 0.705,
      "step": 9960
    },
    {
      "epoch": 3.1452478686454057,
      "grad_norm": 3.0457170009613037,
      "learning_rate": 2.0609080064556874e-05,
      "loss": 0.602,
      "step": 9961
    },
    {
      "epoch": 3.1455636248815915,
      "grad_norm": 3.00461483001709,
      "learning_rate": 2.060557153883938e-05,
      "loss": 0.6402,
      "step": 9962
    },
    {
      "epoch": 3.145879381117777,
      "grad_norm": 3.6848928928375244,
      "learning_rate": 2.0602063013121885e-05,
      "loss": 0.6503,
      "step": 9963
    },
    {
      "epoch": 3.146195137353963,
      "grad_norm": 3.104623317718506,
      "learning_rate": 2.059855448740439e-05,
      "loss": 0.6517,
      "step": 9964
    },
    {
      "epoch": 3.1465108935901482,
      "grad_norm": 3.2505314350128174,
      "learning_rate": 2.0595045961686902e-05,
      "loss": 0.5848,
      "step": 9965
    },
    {
      "epoch": 3.146826649826334,
      "grad_norm": 2.8963675498962402,
      "learning_rate": 2.059153743596941e-05,
      "loss": 0.6654,
      "step": 9966
    },
    {
      "epoch": 3.1471424060625197,
      "grad_norm": 3.137101888656616,
      "learning_rate": 2.0588028910251912e-05,
      "loss": 0.5854,
      "step": 9967
    },
    {
      "epoch": 3.1474581622987055,
      "grad_norm": 3.711698293685913,
      "learning_rate": 2.058452038453442e-05,
      "loss": 0.7232,
      "step": 9968
    },
    {
      "epoch": 3.147773918534891,
      "grad_norm": 3.4076287746429443,
      "learning_rate": 2.0581011858816926e-05,
      "loss": 0.6544,
      "step": 9969
    },
    {
      "epoch": 3.148089674771077,
      "grad_norm": 3.2192962169647217,
      "learning_rate": 2.0577503333099432e-05,
      "loss": 0.6755,
      "step": 9970
    },
    {
      "epoch": 3.1484054310072622,
      "grad_norm": 3.289911985397339,
      "learning_rate": 2.057399480738194e-05,
      "loss": 0.6998,
      "step": 9971
    },
    {
      "epoch": 3.148721187243448,
      "grad_norm": 3.0648722648620605,
      "learning_rate": 2.0570486281664446e-05,
      "loss": 0.6642,
      "step": 9972
    },
    {
      "epoch": 3.1490369434796337,
      "grad_norm": 3.640836000442505,
      "learning_rate": 2.0566977755946953e-05,
      "loss": 0.7106,
      "step": 9973
    },
    {
      "epoch": 3.1493526997158194,
      "grad_norm": 2.994544267654419,
      "learning_rate": 2.056346923022946e-05,
      "loss": 0.5754,
      "step": 9974
    },
    {
      "epoch": 3.149668455952005,
      "grad_norm": 3.3112025260925293,
      "learning_rate": 2.0559960704511966e-05,
      "loss": 0.7685,
      "step": 9975
    },
    {
      "epoch": 3.149984212188191,
      "grad_norm": 2.9635236263275146,
      "learning_rate": 2.0556452178794473e-05,
      "loss": 0.678,
      "step": 9976
    },
    {
      "epoch": 3.150299968424376,
      "grad_norm": 3.038672924041748,
      "learning_rate": 2.0552943653076977e-05,
      "loss": 0.5737,
      "step": 9977
    },
    {
      "epoch": 3.150615724660562,
      "grad_norm": 3.3223917484283447,
      "learning_rate": 2.0549435127359484e-05,
      "loss": 0.7313,
      "step": 9978
    },
    {
      "epoch": 3.1509314808967477,
      "grad_norm": 3.3666303157806396,
      "learning_rate": 2.054592660164199e-05,
      "loss": 0.5931,
      "step": 9979
    },
    {
      "epoch": 3.1512472371329334,
      "grad_norm": 2.8728957176208496,
      "learning_rate": 2.0542418075924497e-05,
      "loss": 0.7019,
      "step": 9980
    },
    {
      "epoch": 3.151562993369119,
      "grad_norm": 3.525953769683838,
      "learning_rate": 2.0538909550207004e-05,
      "loss": 0.7174,
      "step": 9981
    },
    {
      "epoch": 3.151878749605305,
      "grad_norm": 3.145404815673828,
      "learning_rate": 2.053540102448951e-05,
      "loss": 0.7105,
      "step": 9982
    },
    {
      "epoch": 3.15219450584149,
      "grad_norm": 2.902930974960327,
      "learning_rate": 2.0531892498772018e-05,
      "loss": 0.6061,
      "step": 9983
    },
    {
      "epoch": 3.152510262077676,
      "grad_norm": 2.8869237899780273,
      "learning_rate": 2.0528383973054524e-05,
      "loss": 0.6245,
      "step": 9984
    },
    {
      "epoch": 3.1528260183138617,
      "grad_norm": 2.7118446826934814,
      "learning_rate": 2.052487544733703e-05,
      "loss": 0.6872,
      "step": 9985
    },
    {
      "epoch": 3.1531417745500474,
      "grad_norm": 3.26855731010437,
      "learning_rate": 2.0521366921619538e-05,
      "loss": 0.7078,
      "step": 9986
    },
    {
      "epoch": 3.153457530786233,
      "grad_norm": 2.9175820350646973,
      "learning_rate": 2.051785839590204e-05,
      "loss": 0.6091,
      "step": 9987
    },
    {
      "epoch": 3.153773287022419,
      "grad_norm": 3.3248157501220703,
      "learning_rate": 2.0514349870184548e-05,
      "loss": 0.6375,
      "step": 9988
    },
    {
      "epoch": 3.154089043258604,
      "grad_norm": 3.1453497409820557,
      "learning_rate": 2.0510841344467055e-05,
      "loss": 0.7215,
      "step": 9989
    },
    {
      "epoch": 3.15440479949479,
      "grad_norm": 3.4283607006073,
      "learning_rate": 2.0507332818749562e-05,
      "loss": 0.7271,
      "step": 9990
    },
    {
      "epoch": 3.1547205557309757,
      "grad_norm": 3.4530270099639893,
      "learning_rate": 2.050382429303207e-05,
      "loss": 0.6475,
      "step": 9991
    },
    {
      "epoch": 3.1550363119671614,
      "grad_norm": 2.949171304702759,
      "learning_rate": 2.0500315767314576e-05,
      "loss": 0.6053,
      "step": 9992
    },
    {
      "epoch": 3.155352068203347,
      "grad_norm": 3.290522336959839,
      "learning_rate": 2.0496807241597082e-05,
      "loss": 0.5982,
      "step": 9993
    },
    {
      "epoch": 3.155667824439533,
      "grad_norm": 2.7903552055358887,
      "learning_rate": 2.049329871587959e-05,
      "loss": 0.5189,
      "step": 9994
    },
    {
      "epoch": 3.155983580675718,
      "grad_norm": 3.2393062114715576,
      "learning_rate": 2.0489790190162096e-05,
      "loss": 0.7244,
      "step": 9995
    },
    {
      "epoch": 3.156299336911904,
      "grad_norm": 3.036447048187256,
      "learning_rate": 2.04862816644446e-05,
      "loss": 0.6654,
      "step": 9996
    },
    {
      "epoch": 3.1566150931480896,
      "grad_norm": 3.297752618789673,
      "learning_rate": 2.0482773138727106e-05,
      "loss": 0.6712,
      "step": 9997
    },
    {
      "epoch": 3.1569308493842754,
      "grad_norm": 3.0079891681671143,
      "learning_rate": 2.0479264613009613e-05,
      "loss": 0.6004,
      "step": 9998
    },
    {
      "epoch": 3.157246605620461,
      "grad_norm": 3.099893569946289,
      "learning_rate": 2.047575608729212e-05,
      "loss": 0.6777,
      "step": 9999
    },
    {
      "epoch": 3.157562361856647,
      "grad_norm": 2.994950294494629,
      "learning_rate": 2.047224756157463e-05,
      "loss": 0.6786,
      "step": 10000
    },
    {
      "epoch": 3.157562361856647,
      "eval_loss": 0.7991039156913757,
      "eval_runtime": 4.5872,
      "eval_samples_per_second": 542.816,
      "eval_steps_per_second": 4.578,
      "step": 10000
    },
    {
      "epoch": 3.157878118092832,
      "grad_norm": 3.1866953372955322,
      "learning_rate": 2.0468739035857133e-05,
      "loss": 0.7308,
      "step": 10001
    },
    {
      "epoch": 3.158193874329018,
      "grad_norm": 2.8809077739715576,
      "learning_rate": 2.046523051013964e-05,
      "loss": 0.6683,
      "step": 10002
    },
    {
      "epoch": 3.1585096305652036,
      "grad_norm": 3.192746877670288,
      "learning_rate": 2.0461721984422147e-05,
      "loss": 0.6352,
      "step": 10003
    },
    {
      "epoch": 3.1588253868013894,
      "grad_norm": 3.708557367324829,
      "learning_rate": 2.0458213458704654e-05,
      "loss": 0.7459,
      "step": 10004
    },
    {
      "epoch": 3.159141143037575,
      "grad_norm": 3.6972339153289795,
      "learning_rate": 2.045470493298716e-05,
      "loss": 0.7587,
      "step": 10005
    },
    {
      "epoch": 3.159456899273761,
      "grad_norm": 3.233579397201538,
      "learning_rate": 2.0451196407269664e-05,
      "loss": 0.6384,
      "step": 10006
    },
    {
      "epoch": 3.159772655509946,
      "grad_norm": 2.938000440597534,
      "learning_rate": 2.044768788155217e-05,
      "loss": 0.6382,
      "step": 10007
    },
    {
      "epoch": 3.160088411746132,
      "grad_norm": 2.9316792488098145,
      "learning_rate": 2.044417935583468e-05,
      "loss": 0.5976,
      "step": 10008
    },
    {
      "epoch": 3.1604041679823176,
      "grad_norm": 3.0482423305511475,
      "learning_rate": 2.0440670830117188e-05,
      "loss": 0.6618,
      "step": 10009
    },
    {
      "epoch": 3.1607199242185033,
      "grad_norm": 2.6690893173217773,
      "learning_rate": 2.043716230439969e-05,
      "loss": 0.653,
      "step": 10010
    },
    {
      "epoch": 3.161035680454689,
      "grad_norm": 2.9223861694335938,
      "learning_rate": 2.0433653778682198e-05,
      "loss": 0.6565,
      "step": 10011
    },
    {
      "epoch": 3.161351436690875,
      "grad_norm": 3.027972459793091,
      "learning_rate": 2.0430145252964705e-05,
      "loss": 0.632,
      "step": 10012
    },
    {
      "epoch": 3.16166719292706,
      "grad_norm": 3.0980753898620605,
      "learning_rate": 2.0426636727247212e-05,
      "loss": 0.6069,
      "step": 10013
    },
    {
      "epoch": 3.161982949163246,
      "grad_norm": 2.3402092456817627,
      "learning_rate": 2.042312820152972e-05,
      "loss": 0.5579,
      "step": 10014
    },
    {
      "epoch": 3.1622987053994316,
      "grad_norm": 3.404210329055786,
      "learning_rate": 2.0419619675812225e-05,
      "loss": 0.6866,
      "step": 10015
    },
    {
      "epoch": 3.1626144616356173,
      "grad_norm": 3.588735342025757,
      "learning_rate": 2.041611115009473e-05,
      "loss": 0.6869,
      "step": 10016
    },
    {
      "epoch": 3.162930217871803,
      "grad_norm": 3.8224592208862305,
      "learning_rate": 2.041260262437724e-05,
      "loss": 0.6949,
      "step": 10017
    },
    {
      "epoch": 3.163245974107989,
      "grad_norm": 2.88494873046875,
      "learning_rate": 2.0409094098659746e-05,
      "loss": 0.6653,
      "step": 10018
    },
    {
      "epoch": 3.163561730344174,
      "grad_norm": 3.075571298599243,
      "learning_rate": 2.0405585572942253e-05,
      "loss": 0.7124,
      "step": 10019
    },
    {
      "epoch": 3.16387748658036,
      "grad_norm": 3.6645567417144775,
      "learning_rate": 2.0402077047224756e-05,
      "loss": 0.6975,
      "step": 10020
    },
    {
      "epoch": 3.1641932428165456,
      "grad_norm": 2.759296178817749,
      "learning_rate": 2.0398568521507263e-05,
      "loss": 0.5972,
      "step": 10021
    },
    {
      "epoch": 3.1645089990527313,
      "grad_norm": 3.31764817237854,
      "learning_rate": 2.039505999578977e-05,
      "loss": 0.6397,
      "step": 10022
    },
    {
      "epoch": 3.164824755288917,
      "grad_norm": 3.2692782878875732,
      "learning_rate": 2.0391551470072277e-05,
      "loss": 0.6392,
      "step": 10023
    },
    {
      "epoch": 3.165140511525103,
      "grad_norm": 3.139922857284546,
      "learning_rate": 2.0388042944354783e-05,
      "loss": 0.6832,
      "step": 10024
    },
    {
      "epoch": 3.165456267761288,
      "grad_norm": 3.512040853500366,
      "learning_rate": 2.0384534418637287e-05,
      "loss": 0.6681,
      "step": 10025
    },
    {
      "epoch": 3.165772023997474,
      "grad_norm": 3.2128381729125977,
      "learning_rate": 2.0381025892919797e-05,
      "loss": 0.7187,
      "step": 10026
    },
    {
      "epoch": 3.1660877802336596,
      "grad_norm": 3.1711905002593994,
      "learning_rate": 2.0377517367202304e-05,
      "loss": 0.6087,
      "step": 10027
    },
    {
      "epoch": 3.1664035364698453,
      "grad_norm": 3.7350237369537354,
      "learning_rate": 2.037400884148481e-05,
      "loss": 0.662,
      "step": 10028
    },
    {
      "epoch": 3.166719292706031,
      "grad_norm": 2.90214467048645,
      "learning_rate": 2.0370500315767317e-05,
      "loss": 0.6371,
      "step": 10029
    },
    {
      "epoch": 3.1670350489422168,
      "grad_norm": 3.0017683506011963,
      "learning_rate": 2.036699179004982e-05,
      "loss": 0.6158,
      "step": 10030
    },
    {
      "epoch": 3.167350805178402,
      "grad_norm": 4.369495868682861,
      "learning_rate": 2.0363483264332328e-05,
      "loss": 0.6604,
      "step": 10031
    },
    {
      "epoch": 3.167666561414588,
      "grad_norm": 3.714193105697632,
      "learning_rate": 2.0359974738614834e-05,
      "loss": 0.6974,
      "step": 10032
    },
    {
      "epoch": 3.1679823176507735,
      "grad_norm": 3.714193105697632,
      "learning_rate": 2.035646621289734e-05,
      "loss": 0.658,
      "step": 10033
    },
    {
      "epoch": 3.1682980738869593,
      "grad_norm": 2.725109100341797,
      "learning_rate": 2.0352957687179848e-05,
      "loss": 0.6292,
      "step": 10034
    },
    {
      "epoch": 3.168613830123145,
      "grad_norm": 3.25314998626709,
      "learning_rate": 2.0349449161462355e-05,
      "loss": 0.6771,
      "step": 10035
    },
    {
      "epoch": 3.1689295863593308,
      "grad_norm": 3.092092752456665,
      "learning_rate": 2.0345940635744862e-05,
      "loss": 0.6332,
      "step": 10036
    },
    {
      "epoch": 3.1692453425955165,
      "grad_norm": 2.654160737991333,
      "learning_rate": 2.034243211002737e-05,
      "loss": 0.6474,
      "step": 10037
    },
    {
      "epoch": 3.169561098831702,
      "grad_norm": 3.1095476150512695,
      "learning_rate": 2.0338923584309875e-05,
      "loss": 0.6035,
      "step": 10038
    },
    {
      "epoch": 3.1698768550678875,
      "grad_norm": 2.9968912601470947,
      "learning_rate": 2.033541505859238e-05,
      "loss": 0.625,
      "step": 10039
    },
    {
      "epoch": 3.1701926113040733,
      "grad_norm": 2.9546499252319336,
      "learning_rate": 2.0331906532874886e-05,
      "loss": 0.62,
      "step": 10040
    },
    {
      "epoch": 3.170508367540259,
      "grad_norm": 2.9638919830322266,
      "learning_rate": 2.0328398007157392e-05,
      "loss": 0.6377,
      "step": 10041
    },
    {
      "epoch": 3.1708241237764447,
      "grad_norm": 2.9059481620788574,
      "learning_rate": 2.03248894814399e-05,
      "loss": 0.6521,
      "step": 10042
    },
    {
      "epoch": 3.17113988001263,
      "grad_norm": 2.440622091293335,
      "learning_rate": 2.032138095572241e-05,
      "loss": 0.5628,
      "step": 10043
    },
    {
      "epoch": 3.1714556362488158,
      "grad_norm": 3.3551836013793945,
      "learning_rate": 2.0317872430004913e-05,
      "loss": 0.6845,
      "step": 10044
    },
    {
      "epoch": 3.1717713924850015,
      "grad_norm": 3.389979600906372,
      "learning_rate": 2.031436390428742e-05,
      "loss": 0.7527,
      "step": 10045
    },
    {
      "epoch": 3.1720871487211872,
      "grad_norm": 2.7625367641448975,
      "learning_rate": 2.0310855378569926e-05,
      "loss": 0.7013,
      "step": 10046
    },
    {
      "epoch": 3.172402904957373,
      "grad_norm": 2.7498929500579834,
      "learning_rate": 2.0307346852852433e-05,
      "loss": 0.5867,
      "step": 10047
    },
    {
      "epoch": 3.1727186611935587,
      "grad_norm": 2.5613608360290527,
      "learning_rate": 2.030383832713494e-05,
      "loss": 0.6332,
      "step": 10048
    },
    {
      "epoch": 3.1730344174297445,
      "grad_norm": 2.9709460735321045,
      "learning_rate": 2.0300329801417444e-05,
      "loss": 0.5923,
      "step": 10049
    },
    {
      "epoch": 3.1733501736659298,
      "grad_norm": 3.3903255462646484,
      "learning_rate": 2.029682127569995e-05,
      "loss": 0.688,
      "step": 10050
    },
    {
      "epoch": 3.1736659299021155,
      "grad_norm": 2.994507312774658,
      "learning_rate": 2.0293312749982457e-05,
      "loss": 0.7017,
      "step": 10051
    },
    {
      "epoch": 3.1739816861383012,
      "grad_norm": 2.872572183609009,
      "learning_rate": 2.0289804224264967e-05,
      "loss": 0.6244,
      "step": 10052
    },
    {
      "epoch": 3.174297442374487,
      "grad_norm": 3.2309584617614746,
      "learning_rate": 2.028629569854747e-05,
      "loss": 0.7085,
      "step": 10053
    },
    {
      "epoch": 3.1746131986106727,
      "grad_norm": 3.166962146759033,
      "learning_rate": 2.0282787172829978e-05,
      "loss": 0.6384,
      "step": 10054
    },
    {
      "epoch": 3.174928954846858,
      "grad_norm": 2.4406542778015137,
      "learning_rate": 2.0279278647112484e-05,
      "loss": 0.654,
      "step": 10055
    },
    {
      "epoch": 3.1752447110830437,
      "grad_norm": 3.1541824340820312,
      "learning_rate": 2.027577012139499e-05,
      "loss": 0.6251,
      "step": 10056
    },
    {
      "epoch": 3.1755604673192295,
      "grad_norm": 2.9927749633789062,
      "learning_rate": 2.0272261595677498e-05,
      "loss": 0.6592,
      "step": 10057
    },
    {
      "epoch": 3.175876223555415,
      "grad_norm": 2.70212984085083,
      "learning_rate": 2.026875306996e-05,
      "loss": 0.6509,
      "step": 10058
    },
    {
      "epoch": 3.176191979791601,
      "grad_norm": 3.5990140438079834,
      "learning_rate": 2.0265244544242508e-05,
      "loss": 0.6926,
      "step": 10059
    },
    {
      "epoch": 3.1765077360277867,
      "grad_norm": 2.956237554550171,
      "learning_rate": 2.0261736018525015e-05,
      "loss": 0.6067,
      "step": 10060
    },
    {
      "epoch": 3.1768234922639724,
      "grad_norm": 2.7871320247650146,
      "learning_rate": 2.0258227492807525e-05,
      "loss": 0.6024,
      "step": 10061
    },
    {
      "epoch": 3.1771392485001577,
      "grad_norm": 2.8052241802215576,
      "learning_rate": 2.0254718967090032e-05,
      "loss": 0.6861,
      "step": 10062
    },
    {
      "epoch": 3.1774550047363435,
      "grad_norm": 2.7781081199645996,
      "learning_rate": 2.0251210441372536e-05,
      "loss": 0.607,
      "step": 10063
    },
    {
      "epoch": 3.177770760972529,
      "grad_norm": 3.4224202632904053,
      "learning_rate": 2.0247701915655042e-05,
      "loss": 0.6739,
      "step": 10064
    },
    {
      "epoch": 3.178086517208715,
      "grad_norm": 3.104506731033325,
      "learning_rate": 2.024419338993755e-05,
      "loss": 0.7017,
      "step": 10065
    },
    {
      "epoch": 3.1784022734449007,
      "grad_norm": 3.337785482406616,
      "learning_rate": 2.0240684864220056e-05,
      "loss": 0.6501,
      "step": 10066
    },
    {
      "epoch": 3.178718029681086,
      "grad_norm": 3.3252854347229004,
      "learning_rate": 2.0237176338502563e-05,
      "loss": 0.6836,
      "step": 10067
    },
    {
      "epoch": 3.1790337859172717,
      "grad_norm": 2.9200243949890137,
      "learning_rate": 2.0233667812785066e-05,
      "loss": 0.6589,
      "step": 10068
    },
    {
      "epoch": 3.1793495421534574,
      "grad_norm": 2.9566617012023926,
      "learning_rate": 2.0230159287067573e-05,
      "loss": 0.6697,
      "step": 10069
    },
    {
      "epoch": 3.179665298389643,
      "grad_norm": 2.8611860275268555,
      "learning_rate": 2.0226650761350083e-05,
      "loss": 0.7397,
      "step": 10070
    },
    {
      "epoch": 3.179981054625829,
      "grad_norm": 2.9448254108428955,
      "learning_rate": 2.022314223563259e-05,
      "loss": 0.6656,
      "step": 10071
    },
    {
      "epoch": 3.1802968108620147,
      "grad_norm": 3.0429370403289795,
      "learning_rate": 2.0219633709915093e-05,
      "loss": 0.6061,
      "step": 10072
    },
    {
      "epoch": 3.1806125670982004,
      "grad_norm": 3.440113067626953,
      "learning_rate": 2.02161251841976e-05,
      "loss": 0.7192,
      "step": 10073
    },
    {
      "epoch": 3.1809283233343857,
      "grad_norm": 2.8193235397338867,
      "learning_rate": 2.0212616658480107e-05,
      "loss": 0.694,
      "step": 10074
    },
    {
      "epoch": 3.1812440795705714,
      "grad_norm": 2.7980353832244873,
      "learning_rate": 2.0209108132762614e-05,
      "loss": 0.7307,
      "step": 10075
    },
    {
      "epoch": 3.181559835806757,
      "grad_norm": 3.28839373588562,
      "learning_rate": 2.020559960704512e-05,
      "loss": 0.6756,
      "step": 10076
    },
    {
      "epoch": 3.181875592042943,
      "grad_norm": 2.842825412750244,
      "learning_rate": 2.0202091081327628e-05,
      "loss": 0.5788,
      "step": 10077
    },
    {
      "epoch": 3.1821913482791286,
      "grad_norm": 3.06964373588562,
      "learning_rate": 2.0198582555610134e-05,
      "loss": 0.5906,
      "step": 10078
    },
    {
      "epoch": 3.1825071045153144,
      "grad_norm": 3.123680353164673,
      "learning_rate": 2.019507402989264e-05,
      "loss": 0.6279,
      "step": 10079
    },
    {
      "epoch": 3.1828228607514997,
      "grad_norm": 3.4360733032226562,
      "learning_rate": 2.0191565504175148e-05,
      "loss": 0.6342,
      "step": 10080
    },
    {
      "epoch": 3.1831386169876854,
      "grad_norm": 2.8585970401763916,
      "learning_rate": 2.0188056978457655e-05,
      "loss": 0.5925,
      "step": 10081
    },
    {
      "epoch": 3.183454373223871,
      "grad_norm": 3.0522401332855225,
      "learning_rate": 2.0184548452740158e-05,
      "loss": 0.7268,
      "step": 10082
    },
    {
      "epoch": 3.183770129460057,
      "grad_norm": 2.8555030822753906,
      "learning_rate": 2.0181039927022665e-05,
      "loss": 0.7121,
      "step": 10083
    },
    {
      "epoch": 3.1840858856962426,
      "grad_norm": 3.0889790058135986,
      "learning_rate": 2.0177531401305172e-05,
      "loss": 0.7045,
      "step": 10084
    },
    {
      "epoch": 3.1844016419324284,
      "grad_norm": 3.2817137241363525,
      "learning_rate": 2.017402287558768e-05,
      "loss": 0.6288,
      "step": 10085
    },
    {
      "epoch": 3.1847173981686137,
      "grad_norm": 3.7113566398620605,
      "learning_rate": 2.0170514349870185e-05,
      "loss": 0.7501,
      "step": 10086
    },
    {
      "epoch": 3.1850331544047994,
      "grad_norm": 3.20745849609375,
      "learning_rate": 2.0167005824152692e-05,
      "loss": 0.6764,
      "step": 10087
    },
    {
      "epoch": 3.185348910640985,
      "grad_norm": 2.875607967376709,
      "learning_rate": 2.01634972984352e-05,
      "loss": 0.6267,
      "step": 10088
    },
    {
      "epoch": 3.185664666877171,
      "grad_norm": 3.068671941757202,
      "learning_rate": 2.0159988772717706e-05,
      "loss": 0.7202,
      "step": 10089
    },
    {
      "epoch": 3.1859804231133566,
      "grad_norm": 2.972038507461548,
      "learning_rate": 2.0156480247000213e-05,
      "loss": 0.65,
      "step": 10090
    },
    {
      "epoch": 3.1862961793495423,
      "grad_norm": 3.3918838500976562,
      "learning_rate": 2.015297172128272e-05,
      "loss": 0.7132,
      "step": 10091
    },
    {
      "epoch": 3.1866119355857276,
      "grad_norm": 3.235836982727051,
      "learning_rate": 2.0149463195565223e-05,
      "loss": 0.7315,
      "step": 10092
    },
    {
      "epoch": 3.1869276918219134,
      "grad_norm": 2.9321162700653076,
      "learning_rate": 2.014595466984773e-05,
      "loss": 0.6843,
      "step": 10093
    },
    {
      "epoch": 3.187243448058099,
      "grad_norm": 3.4676356315612793,
      "learning_rate": 2.0142446144130237e-05,
      "loss": 0.6868,
      "step": 10094
    },
    {
      "epoch": 3.187559204294285,
      "grad_norm": 3.6160223484039307,
      "learning_rate": 2.0138937618412743e-05,
      "loss": 0.6773,
      "step": 10095
    },
    {
      "epoch": 3.1878749605304706,
      "grad_norm": 2.626413106918335,
      "learning_rate": 2.013542909269525e-05,
      "loss": 0.554,
      "step": 10096
    },
    {
      "epoch": 3.1881907167666563,
      "grad_norm": 2.9853038787841797,
      "learning_rate": 2.0131920566977757e-05,
      "loss": 0.7109,
      "step": 10097
    },
    {
      "epoch": 3.1885064730028416,
      "grad_norm": 3.500717878341675,
      "learning_rate": 2.0128412041260264e-05,
      "loss": 0.7222,
      "step": 10098
    },
    {
      "epoch": 3.1888222292390274,
      "grad_norm": 3.1335108280181885,
      "learning_rate": 2.012490351554277e-05,
      "loss": 0.7732,
      "step": 10099
    },
    {
      "epoch": 3.189137985475213,
      "grad_norm": 4.073159217834473,
      "learning_rate": 2.0121394989825277e-05,
      "loss": 0.6539,
      "step": 10100
    },
    {
      "epoch": 3.189453741711399,
      "grad_norm": 3.1066999435424805,
      "learning_rate": 2.011788646410778e-05,
      "loss": 0.5889,
      "step": 10101
    },
    {
      "epoch": 3.1897694979475846,
      "grad_norm": 3.3081345558166504,
      "learning_rate": 2.0114377938390288e-05,
      "loss": 0.6454,
      "step": 10102
    },
    {
      "epoch": 3.1900852541837703,
      "grad_norm": 3.1647555828094482,
      "learning_rate": 2.0110869412672794e-05,
      "loss": 0.6637,
      "step": 10103
    },
    {
      "epoch": 3.1904010104199556,
      "grad_norm": 3.1961753368377686,
      "learning_rate": 2.0107360886955305e-05,
      "loss": 0.6922,
      "step": 10104
    },
    {
      "epoch": 3.1907167666561413,
      "grad_norm": 3.1880671977996826,
      "learning_rate": 2.010385236123781e-05,
      "loss": 0.7068,
      "step": 10105
    },
    {
      "epoch": 3.191032522892327,
      "grad_norm": 2.7905237674713135,
      "learning_rate": 2.0100343835520315e-05,
      "loss": 0.5767,
      "step": 10106
    },
    {
      "epoch": 3.191348279128513,
      "grad_norm": 3.195775270462036,
      "learning_rate": 2.0096835309802822e-05,
      "loss": 0.6507,
      "step": 10107
    },
    {
      "epoch": 3.1916640353646986,
      "grad_norm": 3.336928606033325,
      "learning_rate": 2.009332678408533e-05,
      "loss": 0.7121,
      "step": 10108
    },
    {
      "epoch": 3.1919797916008843,
      "grad_norm": 3.0303874015808105,
      "learning_rate": 2.0089818258367835e-05,
      "loss": 0.6406,
      "step": 10109
    },
    {
      "epoch": 3.1922955478370696,
      "grad_norm": 3.067023277282715,
      "learning_rate": 2.0086309732650342e-05,
      "loss": 0.7581,
      "step": 10110
    },
    {
      "epoch": 3.1926113040732553,
      "grad_norm": 3.5239508152008057,
      "learning_rate": 2.0082801206932846e-05,
      "loss": 0.7191,
      "step": 10111
    },
    {
      "epoch": 3.192927060309441,
      "grad_norm": 2.9106388092041016,
      "learning_rate": 2.0079292681215352e-05,
      "loss": 0.6593,
      "step": 10112
    },
    {
      "epoch": 3.193242816545627,
      "grad_norm": 3.64736008644104,
      "learning_rate": 2.0075784155497863e-05,
      "loss": 0.6231,
      "step": 10113
    },
    {
      "epoch": 3.1935585727818125,
      "grad_norm": 4.097681522369385,
      "learning_rate": 2.007227562978037e-05,
      "loss": 0.6356,
      "step": 10114
    },
    {
      "epoch": 3.1938743290179983,
      "grad_norm": 3.091416120529175,
      "learning_rate": 2.0068767104062873e-05,
      "loss": 0.6358,
      "step": 10115
    },
    {
      "epoch": 3.1941900852541836,
      "grad_norm": 3.0053298473358154,
      "learning_rate": 2.006525857834538e-05,
      "loss": 0.6469,
      "step": 10116
    },
    {
      "epoch": 3.1945058414903693,
      "grad_norm": 3.5983362197875977,
      "learning_rate": 2.0061750052627886e-05,
      "loss": 0.7617,
      "step": 10117
    },
    {
      "epoch": 3.194821597726555,
      "grad_norm": 3.2564172744750977,
      "learning_rate": 2.0058241526910393e-05,
      "loss": 0.6963,
      "step": 10118
    },
    {
      "epoch": 3.195137353962741,
      "grad_norm": 2.844454765319824,
      "learning_rate": 2.00547330011929e-05,
      "loss": 0.685,
      "step": 10119
    },
    {
      "epoch": 3.1954531101989265,
      "grad_norm": 3.3221275806427,
      "learning_rate": 2.0051224475475407e-05,
      "loss": 0.7103,
      "step": 10120
    },
    {
      "epoch": 3.1957688664351123,
      "grad_norm": 3.0681793689727783,
      "learning_rate": 2.004771594975791e-05,
      "loss": 0.651,
      "step": 10121
    },
    {
      "epoch": 3.1960846226712976,
      "grad_norm": 3.2778162956237793,
      "learning_rate": 2.004420742404042e-05,
      "loss": 0.5988,
      "step": 10122
    },
    {
      "epoch": 3.1964003789074833,
      "grad_norm": 3.400376796722412,
      "learning_rate": 2.0040698898322927e-05,
      "loss": 0.6966,
      "step": 10123
    },
    {
      "epoch": 3.196716135143669,
      "grad_norm": 3.2823572158813477,
      "learning_rate": 2.0037190372605434e-05,
      "loss": 0.6238,
      "step": 10124
    },
    {
      "epoch": 3.1970318913798548,
      "grad_norm": 2.8461363315582275,
      "learning_rate": 2.0033681846887938e-05,
      "loss": 0.5945,
      "step": 10125
    },
    {
      "epoch": 3.1973476476160405,
      "grad_norm": 3.2443344593048096,
      "learning_rate": 2.0030173321170444e-05,
      "loss": 0.578,
      "step": 10126
    },
    {
      "epoch": 3.1976634038522262,
      "grad_norm": 4.166206359863281,
      "learning_rate": 2.002666479545295e-05,
      "loss": 0.6141,
      "step": 10127
    },
    {
      "epoch": 3.1979791600884115,
      "grad_norm": 3.5565314292907715,
      "learning_rate": 2.0023156269735458e-05,
      "loss": 0.6212,
      "step": 10128
    },
    {
      "epoch": 3.1982949163245973,
      "grad_norm": 2.916330099105835,
      "learning_rate": 2.0019647744017965e-05,
      "loss": 0.6814,
      "step": 10129
    },
    {
      "epoch": 3.198610672560783,
      "grad_norm": 3.0726473331451416,
      "learning_rate": 2.0016139218300468e-05,
      "loss": 0.6755,
      "step": 10130
    },
    {
      "epoch": 3.1989264287969688,
      "grad_norm": 4.09926176071167,
      "learning_rate": 2.001263069258298e-05,
      "loss": 0.7023,
      "step": 10131
    },
    {
      "epoch": 3.1992421850331545,
      "grad_norm": 2.6277451515197754,
      "learning_rate": 2.0009122166865485e-05,
      "loss": 0.6179,
      "step": 10132
    },
    {
      "epoch": 3.1995579412693402,
      "grad_norm": 3.1358654499053955,
      "learning_rate": 2.0005613641147992e-05,
      "loss": 0.6825,
      "step": 10133
    },
    {
      "epoch": 3.1998736975055255,
      "grad_norm": 3.1775901317596436,
      "learning_rate": 2.00021051154305e-05,
      "loss": 0.5994,
      "step": 10134
    },
    {
      "epoch": 3.2001894537417113,
      "grad_norm": 3.4271695613861084,
      "learning_rate": 1.9998596589713002e-05,
      "loss": 0.6322,
      "step": 10135
    },
    {
      "epoch": 3.200505209977897,
      "grad_norm": 3.0483179092407227,
      "learning_rate": 1.999508806399551e-05,
      "loss": 0.6899,
      "step": 10136
    },
    {
      "epoch": 3.2008209662140827,
      "grad_norm": 2.744760036468506,
      "learning_rate": 1.9991579538278016e-05,
      "loss": 0.5849,
      "step": 10137
    },
    {
      "epoch": 3.2011367224502685,
      "grad_norm": 3.0995147228240967,
      "learning_rate": 1.9988071012560523e-05,
      "loss": 0.6779,
      "step": 10138
    },
    {
      "epoch": 3.201452478686454,
      "grad_norm": 3.1508049964904785,
      "learning_rate": 1.998456248684303e-05,
      "loss": 0.59,
      "step": 10139
    },
    {
      "epoch": 3.20176823492264,
      "grad_norm": 3.1453874111175537,
      "learning_rate": 1.9981053961125536e-05,
      "loss": 0.6937,
      "step": 10140
    },
    {
      "epoch": 3.2020839911588252,
      "grad_norm": 3.403275966644287,
      "learning_rate": 1.9977545435408043e-05,
      "loss": 0.7192,
      "step": 10141
    },
    {
      "epoch": 3.202399747395011,
      "grad_norm": 3.573697090148926,
      "learning_rate": 1.997403690969055e-05,
      "loss": 0.6691,
      "step": 10142
    },
    {
      "epoch": 3.2027155036311967,
      "grad_norm": 3.176776170730591,
      "learning_rate": 1.9970528383973057e-05,
      "loss": 0.7188,
      "step": 10143
    },
    {
      "epoch": 3.2030312598673825,
      "grad_norm": 3.590688467025757,
      "learning_rate": 1.996701985825556e-05,
      "loss": 0.6722,
      "step": 10144
    },
    {
      "epoch": 3.203347016103568,
      "grad_norm": 2.8432435989379883,
      "learning_rate": 1.9963511332538067e-05,
      "loss": 0.6663,
      "step": 10145
    },
    {
      "epoch": 3.2036627723397535,
      "grad_norm": 3.4291532039642334,
      "learning_rate": 1.9960002806820574e-05,
      "loss": 0.5997,
      "step": 10146
    },
    {
      "epoch": 3.2039785285759392,
      "grad_norm": 3.3433189392089844,
      "learning_rate": 1.995649428110308e-05,
      "loss": 0.7051,
      "step": 10147
    },
    {
      "epoch": 3.204294284812125,
      "grad_norm": 2.711954355239868,
      "learning_rate": 1.995298575538559e-05,
      "loss": 0.6373,
      "step": 10148
    },
    {
      "epoch": 3.2046100410483107,
      "grad_norm": 3.755852222442627,
      "learning_rate": 1.9949477229668094e-05,
      "loss": 0.6632,
      "step": 10149
    },
    {
      "epoch": 3.2049257972844964,
      "grad_norm": 3.1760311126708984,
      "learning_rate": 1.99459687039506e-05,
      "loss": 0.6436,
      "step": 10150
    },
    {
      "epoch": 3.205241553520682,
      "grad_norm": 3.0289413928985596,
      "learning_rate": 1.9942460178233108e-05,
      "loss": 0.6881,
      "step": 10151
    },
    {
      "epoch": 3.205557309756868,
      "grad_norm": 3.356828212738037,
      "learning_rate": 1.9938951652515615e-05,
      "loss": 0.6635,
      "step": 10152
    },
    {
      "epoch": 3.205873065993053,
      "grad_norm": 3.9425535202026367,
      "learning_rate": 1.993544312679812e-05,
      "loss": 0.7194,
      "step": 10153
    },
    {
      "epoch": 3.206188822229239,
      "grad_norm": 2.92455792427063,
      "learning_rate": 1.9931934601080625e-05,
      "loss": 0.724,
      "step": 10154
    },
    {
      "epoch": 3.2065045784654247,
      "grad_norm": 3.2764084339141846,
      "learning_rate": 1.9928426075363132e-05,
      "loss": 0.6748,
      "step": 10155
    },
    {
      "epoch": 3.2068203347016104,
      "grad_norm": 2.93688702583313,
      "learning_rate": 1.992491754964564e-05,
      "loss": 0.5728,
      "step": 10156
    },
    {
      "epoch": 3.207136090937796,
      "grad_norm": 3.3504834175109863,
      "learning_rate": 1.992140902392815e-05,
      "loss": 0.7212,
      "step": 10157
    },
    {
      "epoch": 3.2074518471739815,
      "grad_norm": 3.6887106895446777,
      "learning_rate": 1.9917900498210652e-05,
      "loss": 0.7015,
      "step": 10158
    },
    {
      "epoch": 3.207767603410167,
      "grad_norm": 3.0869200229644775,
      "learning_rate": 1.991439197249316e-05,
      "loss": 0.6752,
      "step": 10159
    },
    {
      "epoch": 3.208083359646353,
      "grad_norm": 2.8647689819335938,
      "learning_rate": 1.9910883446775666e-05,
      "loss": 0.6431,
      "step": 10160
    },
    {
      "epoch": 3.2083991158825387,
      "grad_norm": 2.8358314037323,
      "learning_rate": 1.9907374921058173e-05,
      "loss": 0.6337,
      "step": 10161
    },
    {
      "epoch": 3.2087148721187244,
      "grad_norm": 3.1287975311279297,
      "learning_rate": 1.990386639534068e-05,
      "loss": 0.6678,
      "step": 10162
    },
    {
      "epoch": 3.20903062835491,
      "grad_norm": 3.444951295852661,
      "learning_rate": 1.9900357869623183e-05,
      "loss": 0.6826,
      "step": 10163
    },
    {
      "epoch": 3.209346384591096,
      "grad_norm": 3.334540605545044,
      "learning_rate": 1.989684934390569e-05,
      "loss": 0.6211,
      "step": 10164
    },
    {
      "epoch": 3.209662140827281,
      "grad_norm": 3.2529079914093018,
      "learning_rate": 1.9893340818188197e-05,
      "loss": 0.6256,
      "step": 10165
    },
    {
      "epoch": 3.209977897063467,
      "grad_norm": 3.13204288482666,
      "learning_rate": 1.9889832292470707e-05,
      "loss": 0.7068,
      "step": 10166
    },
    {
      "epoch": 3.2102936532996527,
      "grad_norm": 3.288771867752075,
      "learning_rate": 1.9886323766753214e-05,
      "loss": 0.814,
      "step": 10167
    },
    {
      "epoch": 3.2106094095358384,
      "grad_norm": 2.8044140338897705,
      "learning_rate": 1.9882815241035717e-05,
      "loss": 0.6276,
      "step": 10168
    },
    {
      "epoch": 3.210925165772024,
      "grad_norm": 2.950979709625244,
      "learning_rate": 1.9879306715318224e-05,
      "loss": 0.596,
      "step": 10169
    },
    {
      "epoch": 3.2112409220082094,
      "grad_norm": 2.824746608734131,
      "learning_rate": 1.987579818960073e-05,
      "loss": 0.6072,
      "step": 10170
    },
    {
      "epoch": 3.211556678244395,
      "grad_norm": 4.151172161102295,
      "learning_rate": 1.9872289663883237e-05,
      "loss": 0.717,
      "step": 10171
    },
    {
      "epoch": 3.211872434480581,
      "grad_norm": 3.461437225341797,
      "learning_rate": 1.9868781138165744e-05,
      "loss": 0.6534,
      "step": 10172
    },
    {
      "epoch": 3.2121881907167666,
      "grad_norm": 3.1469907760620117,
      "learning_rate": 1.9865272612448248e-05,
      "loss": 0.7358,
      "step": 10173
    },
    {
      "epoch": 3.2125039469529524,
      "grad_norm": 2.6596148014068604,
      "learning_rate": 1.9861764086730758e-05,
      "loss": 0.5315,
      "step": 10174
    },
    {
      "epoch": 3.212819703189138,
      "grad_norm": 3.690211057662964,
      "learning_rate": 1.9858255561013265e-05,
      "loss": 0.6669,
      "step": 10175
    },
    {
      "epoch": 3.213135459425324,
      "grad_norm": 3.5954320430755615,
      "learning_rate": 1.985474703529577e-05,
      "loss": 0.6862,
      "step": 10176
    },
    {
      "epoch": 3.213451215661509,
      "grad_norm": 2.566882371902466,
      "learning_rate": 1.9851238509578275e-05,
      "loss": 0.6206,
      "step": 10177
    },
    {
      "epoch": 3.213766971897695,
      "grad_norm": 2.66434383392334,
      "learning_rate": 1.9847729983860782e-05,
      "loss": 0.6192,
      "step": 10178
    },
    {
      "epoch": 3.2140827281338806,
      "grad_norm": 2.9513466358184814,
      "learning_rate": 1.984422145814329e-05,
      "loss": 0.6473,
      "step": 10179
    },
    {
      "epoch": 3.2143984843700664,
      "grad_norm": 2.846134662628174,
      "learning_rate": 1.9840712932425795e-05,
      "loss": 0.5948,
      "step": 10180
    },
    {
      "epoch": 3.214714240606252,
      "grad_norm": 3.391538381576538,
      "learning_rate": 1.9837204406708302e-05,
      "loss": 0.66,
      "step": 10181
    },
    {
      "epoch": 3.215029996842438,
      "grad_norm": 3.1894710063934326,
      "learning_rate": 1.983369588099081e-05,
      "loss": 0.7083,
      "step": 10182
    },
    {
      "epoch": 3.215345753078623,
      "grad_norm": 2.9779820442199707,
      "learning_rate": 1.9830187355273316e-05,
      "loss": 0.6167,
      "step": 10183
    },
    {
      "epoch": 3.215661509314809,
      "grad_norm": 3.0052506923675537,
      "learning_rate": 1.9826678829555823e-05,
      "loss": 0.6019,
      "step": 10184
    },
    {
      "epoch": 3.2159772655509946,
      "grad_norm": 3.191704750061035,
      "learning_rate": 1.982317030383833e-05,
      "loss": 0.6926,
      "step": 10185
    },
    {
      "epoch": 3.2162930217871804,
      "grad_norm": 3.204878091812134,
      "learning_rate": 1.9819661778120836e-05,
      "loss": 0.6513,
      "step": 10186
    },
    {
      "epoch": 3.216608778023366,
      "grad_norm": 3.283317804336548,
      "learning_rate": 1.981615325240334e-05,
      "loss": 0.6001,
      "step": 10187
    },
    {
      "epoch": 3.216924534259552,
      "grad_norm": 2.4706978797912598,
      "learning_rate": 1.9812644726685846e-05,
      "loss": 0.5486,
      "step": 10188
    },
    {
      "epoch": 3.217240290495737,
      "grad_norm": 2.9178597927093506,
      "learning_rate": 1.9809136200968353e-05,
      "loss": 0.565,
      "step": 10189
    },
    {
      "epoch": 3.217556046731923,
      "grad_norm": 3.4158973693847656,
      "learning_rate": 1.980562767525086e-05,
      "loss": 0.6627,
      "step": 10190
    },
    {
      "epoch": 3.2178718029681086,
      "grad_norm": 3.041322708129883,
      "learning_rate": 1.9802119149533367e-05,
      "loss": 0.724,
      "step": 10191
    },
    {
      "epoch": 3.2181875592042943,
      "grad_norm": 3.3688316345214844,
      "learning_rate": 1.9798610623815874e-05,
      "loss": 0.5285,
      "step": 10192
    },
    {
      "epoch": 3.21850331544048,
      "grad_norm": 3.30291485786438,
      "learning_rate": 1.979510209809838e-05,
      "loss": 0.6729,
      "step": 10193
    },
    {
      "epoch": 3.218819071676666,
      "grad_norm": 3.2771999835968018,
      "learning_rate": 1.9791593572380887e-05,
      "loss": 0.6285,
      "step": 10194
    },
    {
      "epoch": 3.219134827912851,
      "grad_norm": 3.4465925693511963,
      "learning_rate": 1.9788085046663394e-05,
      "loss": 0.728,
      "step": 10195
    },
    {
      "epoch": 3.219450584149037,
      "grad_norm": 2.65736722946167,
      "learning_rate": 1.97845765209459e-05,
      "loss": 0.5223,
      "step": 10196
    },
    {
      "epoch": 3.2197663403852226,
      "grad_norm": 2.9903626441955566,
      "learning_rate": 1.9781067995228404e-05,
      "loss": 0.7082,
      "step": 10197
    },
    {
      "epoch": 3.2200820966214083,
      "grad_norm": 2.8633594512939453,
      "learning_rate": 1.977755946951091e-05,
      "loss": 0.6773,
      "step": 10198
    },
    {
      "epoch": 3.220397852857594,
      "grad_norm": 3.0788934230804443,
      "learning_rate": 1.9774050943793418e-05,
      "loss": 0.6586,
      "step": 10199
    },
    {
      "epoch": 3.22071360909378,
      "grad_norm": 2.9566144943237305,
      "learning_rate": 1.9770542418075925e-05,
      "loss": 0.6825,
      "step": 10200
    },
    {
      "epoch": 3.22071360909378,
      "eval_loss": 0.8062103986740112,
      "eval_runtime": 4.6656,
      "eval_samples_per_second": 533.695,
      "eval_steps_per_second": 4.501,
      "step": 10200
    },
    {
      "epoch": 3.221029365329965,
      "grad_norm": 3.424600124359131,
      "learning_rate": 1.976703389235843e-05,
      "loss": 0.6108,
      "step": 10201
    },
    {
      "epoch": 3.221345121566151,
      "grad_norm": 3.1982064247131348,
      "learning_rate": 1.976352536664094e-05,
      "loss": 0.7458,
      "step": 10202
    },
    {
      "epoch": 3.2216608778023366,
      "grad_norm": 3.0389163494110107,
      "learning_rate": 1.9760016840923445e-05,
      "loss": 0.5147,
      "step": 10203
    },
    {
      "epoch": 3.2219766340385223,
      "grad_norm": 3.10367488861084,
      "learning_rate": 1.9756508315205952e-05,
      "loss": 0.6334,
      "step": 10204
    },
    {
      "epoch": 3.222292390274708,
      "grad_norm": 2.939997434616089,
      "learning_rate": 1.975299978948846e-05,
      "loss": 0.5548,
      "step": 10205
    },
    {
      "epoch": 3.222608146510894,
      "grad_norm": 3.7994067668914795,
      "learning_rate": 1.9749491263770962e-05,
      "loss": 0.5993,
      "step": 10206
    },
    {
      "epoch": 3.222923902747079,
      "grad_norm": 2.9017317295074463,
      "learning_rate": 1.974598273805347e-05,
      "loss": 0.6232,
      "step": 10207
    },
    {
      "epoch": 3.223239658983265,
      "grad_norm": 3.5986170768737793,
      "learning_rate": 1.9742474212335976e-05,
      "loss": 0.6259,
      "step": 10208
    },
    {
      "epoch": 3.2235554152194505,
      "grad_norm": 3.6562466621398926,
      "learning_rate": 1.9738965686618486e-05,
      "loss": 0.7376,
      "step": 10209
    },
    {
      "epoch": 3.2238711714556363,
      "grad_norm": 3.9068660736083984,
      "learning_rate": 1.9735457160900993e-05,
      "loss": 0.6577,
      "step": 10210
    },
    {
      "epoch": 3.224186927691822,
      "grad_norm": 3.210594415664673,
      "learning_rate": 1.9731948635183496e-05,
      "loss": 0.5816,
      "step": 10211
    },
    {
      "epoch": 3.2245026839280078,
      "grad_norm": 3.444089412689209,
      "learning_rate": 1.9728440109466003e-05,
      "loss": 0.6179,
      "step": 10212
    },
    {
      "epoch": 3.224818440164193,
      "grad_norm": 3.0061087608337402,
      "learning_rate": 1.972493158374851e-05,
      "loss": 0.6191,
      "step": 10213
    },
    {
      "epoch": 3.225134196400379,
      "grad_norm": 2.9953291416168213,
      "learning_rate": 1.9721423058031017e-05,
      "loss": 0.7046,
      "step": 10214
    },
    {
      "epoch": 3.2254499526365645,
      "grad_norm": 3.2930104732513428,
      "learning_rate": 1.9717914532313524e-05,
      "loss": 0.6144,
      "step": 10215
    },
    {
      "epoch": 3.2257657088727503,
      "grad_norm": 3.4536356925964355,
      "learning_rate": 1.9714406006596027e-05,
      "loss": 0.6839,
      "step": 10216
    },
    {
      "epoch": 3.226081465108936,
      "grad_norm": 3.164811372756958,
      "learning_rate": 1.9710897480878534e-05,
      "loss": 0.6182,
      "step": 10217
    },
    {
      "epoch": 3.2263972213451217,
      "grad_norm": 3.393153667449951,
      "learning_rate": 1.9707388955161044e-05,
      "loss": 0.611,
      "step": 10218
    },
    {
      "epoch": 3.226712977581307,
      "grad_norm": 3.144439220428467,
      "learning_rate": 1.970388042944355e-05,
      "loss": 0.6697,
      "step": 10219
    },
    {
      "epoch": 3.227028733817493,
      "grad_norm": 3.0461740493774414,
      "learning_rate": 1.9700371903726054e-05,
      "loss": 0.6793,
      "step": 10220
    },
    {
      "epoch": 3.2273444900536785,
      "grad_norm": 3.6759490966796875,
      "learning_rate": 1.969686337800856e-05,
      "loss": 0.6589,
      "step": 10221
    },
    {
      "epoch": 3.2276602462898643,
      "grad_norm": 3.355980634689331,
      "learning_rate": 1.9693354852291068e-05,
      "loss": 0.6567,
      "step": 10222
    },
    {
      "epoch": 3.22797600252605,
      "grad_norm": 3.0240478515625,
      "learning_rate": 1.9689846326573575e-05,
      "loss": 0.6501,
      "step": 10223
    },
    {
      "epoch": 3.2282917587622357,
      "grad_norm": 3.478174924850464,
      "learning_rate": 1.968633780085608e-05,
      "loss": 0.7392,
      "step": 10224
    },
    {
      "epoch": 3.228607514998421,
      "grad_norm": 2.8463923931121826,
      "learning_rate": 1.968282927513859e-05,
      "loss": 0.6927,
      "step": 10225
    },
    {
      "epoch": 3.2289232712346068,
      "grad_norm": 3.491426944732666,
      "learning_rate": 1.9679320749421092e-05,
      "loss": 0.6628,
      "step": 10226
    },
    {
      "epoch": 3.2292390274707925,
      "grad_norm": 3.0664846897125244,
      "learning_rate": 1.9675812223703602e-05,
      "loss": 0.6107,
      "step": 10227
    },
    {
      "epoch": 3.2295547837069782,
      "grad_norm": 3.888549327850342,
      "learning_rate": 1.967230369798611e-05,
      "loss": 0.7172,
      "step": 10228
    },
    {
      "epoch": 3.229870539943164,
      "grad_norm": 2.802924156188965,
      "learning_rate": 1.9668795172268616e-05,
      "loss": 0.6355,
      "step": 10229
    },
    {
      "epoch": 3.2301862961793497,
      "grad_norm": 3.0263595581054688,
      "learning_rate": 1.966528664655112e-05,
      "loss": 0.6667,
      "step": 10230
    },
    {
      "epoch": 3.230502052415535,
      "grad_norm": 3.461290121078491,
      "learning_rate": 1.9661778120833626e-05,
      "loss": 0.6273,
      "step": 10231
    },
    {
      "epoch": 3.2308178086517207,
      "grad_norm": 3.2946739196777344,
      "learning_rate": 1.9658269595116133e-05,
      "loss": 0.5757,
      "step": 10232
    },
    {
      "epoch": 3.2311335648879065,
      "grad_norm": 3.5015265941619873,
      "learning_rate": 1.965476106939864e-05,
      "loss": 0.715,
      "step": 10233
    },
    {
      "epoch": 3.2314493211240922,
      "grad_norm": 3.531881809234619,
      "learning_rate": 1.9651252543681146e-05,
      "loss": 0.6964,
      "step": 10234
    },
    {
      "epoch": 3.231765077360278,
      "grad_norm": 3.10319447517395,
      "learning_rate": 1.964774401796365e-05,
      "loss": 0.6068,
      "step": 10235
    },
    {
      "epoch": 3.2320808335964637,
      "grad_norm": 3.18719482421875,
      "learning_rate": 1.964423549224616e-05,
      "loss": 0.6029,
      "step": 10236
    },
    {
      "epoch": 3.232396589832649,
      "grad_norm": 4.167654037475586,
      "learning_rate": 1.9640726966528667e-05,
      "loss": 0.7491,
      "step": 10237
    },
    {
      "epoch": 3.2327123460688347,
      "grad_norm": 3.220613956451416,
      "learning_rate": 1.9637218440811174e-05,
      "loss": 0.6497,
      "step": 10238
    },
    {
      "epoch": 3.2330281023050205,
      "grad_norm": 3.035273313522339,
      "learning_rate": 1.963370991509368e-05,
      "loss": 0.6768,
      "step": 10239
    },
    {
      "epoch": 3.233343858541206,
      "grad_norm": 3.466282606124878,
      "learning_rate": 1.9630201389376184e-05,
      "loss": 0.6454,
      "step": 10240
    },
    {
      "epoch": 3.233659614777392,
      "grad_norm": 3.3244426250457764,
      "learning_rate": 1.962669286365869e-05,
      "loss": 0.7169,
      "step": 10241
    },
    {
      "epoch": 3.2339753710135777,
      "grad_norm": 2.738056182861328,
      "learning_rate": 1.9623184337941197e-05,
      "loss": 0.6724,
      "step": 10242
    },
    {
      "epoch": 3.234291127249763,
      "grad_norm": 3.0037243366241455,
      "learning_rate": 1.9619675812223704e-05,
      "loss": 0.6132,
      "step": 10243
    },
    {
      "epoch": 3.2346068834859487,
      "grad_norm": 3.296419620513916,
      "learning_rate": 1.961616728650621e-05,
      "loss": 0.7462,
      "step": 10244
    },
    {
      "epoch": 3.2349226397221345,
      "grad_norm": 3.0853631496429443,
      "learning_rate": 1.9612658760788718e-05,
      "loss": 0.7371,
      "step": 10245
    },
    {
      "epoch": 3.23523839595832,
      "grad_norm": 2.7838516235351562,
      "learning_rate": 1.9609150235071225e-05,
      "loss": 0.6386,
      "step": 10246
    },
    {
      "epoch": 3.235554152194506,
      "grad_norm": 2.6713011264801025,
      "learning_rate": 1.960564170935373e-05,
      "loss": 0.6288,
      "step": 10247
    },
    {
      "epoch": 3.2358699084306917,
      "grad_norm": 2.7178423404693604,
      "learning_rate": 1.960213318363624e-05,
      "loss": 0.6365,
      "step": 10248
    },
    {
      "epoch": 3.236185664666877,
      "grad_norm": 2.933363914489746,
      "learning_rate": 1.9598624657918742e-05,
      "loss": 0.6373,
      "step": 10249
    },
    {
      "epoch": 3.2365014209030627,
      "grad_norm": 3.9350600242614746,
      "learning_rate": 1.959511613220125e-05,
      "loss": 0.6903,
      "step": 10250
    },
    {
      "epoch": 3.2368171771392484,
      "grad_norm": 3.342601776123047,
      "learning_rate": 1.9591607606483755e-05,
      "loss": 0.6245,
      "step": 10251
    },
    {
      "epoch": 3.237132933375434,
      "grad_norm": 2.759976863861084,
      "learning_rate": 1.9588099080766262e-05,
      "loss": 0.5566,
      "step": 10252
    },
    {
      "epoch": 3.23744868961162,
      "grad_norm": 3.665754795074463,
      "learning_rate": 1.9584590555048772e-05,
      "loss": 0.671,
      "step": 10253
    },
    {
      "epoch": 3.2377644458478056,
      "grad_norm": 3.0804598331451416,
      "learning_rate": 1.9581082029331276e-05,
      "loss": 0.754,
      "step": 10254
    },
    {
      "epoch": 3.2380802020839914,
      "grad_norm": 3.5955421924591064,
      "learning_rate": 1.9577573503613783e-05,
      "loss": 0.6608,
      "step": 10255
    },
    {
      "epoch": 3.2383959583201767,
      "grad_norm": 2.7717528343200684,
      "learning_rate": 1.957406497789629e-05,
      "loss": 0.6578,
      "step": 10256
    },
    {
      "epoch": 3.2387117145563624,
      "grad_norm": 3.3968727588653564,
      "learning_rate": 1.9570556452178796e-05,
      "loss": 0.7224,
      "step": 10257
    },
    {
      "epoch": 3.239027470792548,
      "grad_norm": 3.3622055053710938,
      "learning_rate": 1.9567047926461303e-05,
      "loss": 0.7016,
      "step": 10258
    },
    {
      "epoch": 3.239343227028734,
      "grad_norm": 2.899566888809204,
      "learning_rate": 1.9563539400743806e-05,
      "loss": 0.6065,
      "step": 10259
    },
    {
      "epoch": 3.2396589832649196,
      "grad_norm": 2.8066823482513428,
      "learning_rate": 1.9560030875026313e-05,
      "loss": 0.6307,
      "step": 10260
    },
    {
      "epoch": 3.239974739501105,
      "grad_norm": 3.0988876819610596,
      "learning_rate": 1.955652234930882e-05,
      "loss": 0.6225,
      "step": 10261
    },
    {
      "epoch": 3.2402904957372907,
      "grad_norm": 3.139591932296753,
      "learning_rate": 1.955301382359133e-05,
      "loss": 0.739,
      "step": 10262
    },
    {
      "epoch": 3.2406062519734764,
      "grad_norm": 3.3073410987854004,
      "learning_rate": 1.9549505297873834e-05,
      "loss": 0.6788,
      "step": 10263
    },
    {
      "epoch": 3.240922008209662,
      "grad_norm": 3.165004014968872,
      "learning_rate": 1.954599677215634e-05,
      "loss": 0.6394,
      "step": 10264
    },
    {
      "epoch": 3.241237764445848,
      "grad_norm": 3.578486919403076,
      "learning_rate": 1.9542488246438847e-05,
      "loss": 0.6198,
      "step": 10265
    },
    {
      "epoch": 3.2415535206820336,
      "grad_norm": 2.7205092906951904,
      "learning_rate": 1.9538979720721354e-05,
      "loss": 0.6374,
      "step": 10266
    },
    {
      "epoch": 3.2418692769182194,
      "grad_norm": 3.0383620262145996,
      "learning_rate": 1.953547119500386e-05,
      "loss": 0.6027,
      "step": 10267
    },
    {
      "epoch": 3.2421850331544046,
      "grad_norm": 3.1037023067474365,
      "learning_rate": 1.9531962669286364e-05,
      "loss": 0.6605,
      "step": 10268
    },
    {
      "epoch": 3.2425007893905904,
      "grad_norm": 3.286221504211426,
      "learning_rate": 1.952845414356887e-05,
      "loss": 0.7115,
      "step": 10269
    },
    {
      "epoch": 3.242816545626776,
      "grad_norm": 2.6238057613372803,
      "learning_rate": 1.9524945617851378e-05,
      "loss": 0.6146,
      "step": 10270
    },
    {
      "epoch": 3.243132301862962,
      "grad_norm": 3.360750436782837,
      "learning_rate": 1.9521437092133888e-05,
      "loss": 0.6485,
      "step": 10271
    },
    {
      "epoch": 3.2434480580991476,
      "grad_norm": 3.566032886505127,
      "learning_rate": 1.9517928566416395e-05,
      "loss": 0.6575,
      "step": 10272
    },
    {
      "epoch": 3.243763814335333,
      "grad_norm": 2.734689950942993,
      "learning_rate": 1.95144200406989e-05,
      "loss": 0.6103,
      "step": 10273
    },
    {
      "epoch": 3.2440795705715186,
      "grad_norm": 3.2460410594940186,
      "learning_rate": 1.9510911514981405e-05,
      "loss": 0.7635,
      "step": 10274
    },
    {
      "epoch": 3.2443953268077044,
      "grad_norm": 3.1002752780914307,
      "learning_rate": 1.9507402989263912e-05,
      "loss": 0.7005,
      "step": 10275
    },
    {
      "epoch": 3.24471108304389,
      "grad_norm": 3.3023228645324707,
      "learning_rate": 1.950389446354642e-05,
      "loss": 0.7682,
      "step": 10276
    },
    {
      "epoch": 3.245026839280076,
      "grad_norm": 3.156771421432495,
      "learning_rate": 1.9500385937828926e-05,
      "loss": 0.7121,
      "step": 10277
    },
    {
      "epoch": 3.2453425955162616,
      "grad_norm": 3.549433946609497,
      "learning_rate": 1.949687741211143e-05,
      "loss": 0.8567,
      "step": 10278
    },
    {
      "epoch": 3.2456583517524473,
      "grad_norm": 3.76947283744812,
      "learning_rate": 1.949336888639394e-05,
      "loss": 0.7078,
      "step": 10279
    },
    {
      "epoch": 3.2459741079886326,
      "grad_norm": 3.1559698581695557,
      "learning_rate": 1.9489860360676446e-05,
      "loss": 0.7261,
      "step": 10280
    },
    {
      "epoch": 3.2462898642248184,
      "grad_norm": 2.8311784267425537,
      "learning_rate": 1.9486351834958953e-05,
      "loss": 0.5213,
      "step": 10281
    },
    {
      "epoch": 3.246605620461004,
      "grad_norm": 3.369582414627075,
      "learning_rate": 1.9482843309241456e-05,
      "loss": 0.7155,
      "step": 10282
    },
    {
      "epoch": 3.24692137669719,
      "grad_norm": 3.374375581741333,
      "learning_rate": 1.9479334783523963e-05,
      "loss": 0.668,
      "step": 10283
    },
    {
      "epoch": 3.2472371329333756,
      "grad_norm": 3.3040072917938232,
      "learning_rate": 1.947582625780647e-05,
      "loss": 0.6674,
      "step": 10284
    },
    {
      "epoch": 3.247552889169561,
      "grad_norm": 3.0062344074249268,
      "learning_rate": 1.9472317732088977e-05,
      "loss": 0.6912,
      "step": 10285
    },
    {
      "epoch": 3.2478686454057466,
      "grad_norm": 3.7431766986846924,
      "learning_rate": 1.9468809206371484e-05,
      "loss": 0.6337,
      "step": 10286
    },
    {
      "epoch": 3.2481844016419323,
      "grad_norm": 3.350532293319702,
      "learning_rate": 1.946530068065399e-05,
      "loss": 0.5811,
      "step": 10287
    },
    {
      "epoch": 3.248500157878118,
      "grad_norm": 2.3828046321868896,
      "learning_rate": 1.9461792154936497e-05,
      "loss": 0.566,
      "step": 10288
    },
    {
      "epoch": 3.248815914114304,
      "grad_norm": 3.3396008014678955,
      "learning_rate": 1.9458283629219004e-05,
      "loss": 0.6397,
      "step": 10289
    },
    {
      "epoch": 3.2491316703504896,
      "grad_norm": 2.6237213611602783,
      "learning_rate": 1.945477510350151e-05,
      "loss": 0.6408,
      "step": 10290
    },
    {
      "epoch": 3.2494474265866753,
      "grad_norm": 3.2829647064208984,
      "learning_rate": 1.9451266577784018e-05,
      "loss": 0.6573,
      "step": 10291
    },
    {
      "epoch": 3.2497631828228606,
      "grad_norm": 2.581641674041748,
      "learning_rate": 1.944775805206652e-05,
      "loss": 0.6184,
      "step": 10292
    },
    {
      "epoch": 3.2500789390590463,
      "grad_norm": 2.7341654300689697,
      "learning_rate": 1.9444249526349028e-05,
      "loss": 0.5391,
      "step": 10293
    },
    {
      "epoch": 3.250394695295232,
      "grad_norm": 3.468599796295166,
      "learning_rate": 1.9440741000631535e-05,
      "loss": 0.6505,
      "step": 10294
    },
    {
      "epoch": 3.250710451531418,
      "grad_norm": 3.4017155170440674,
      "learning_rate": 1.943723247491404e-05,
      "loss": 0.6902,
      "step": 10295
    },
    {
      "epoch": 3.2510262077676035,
      "grad_norm": 3.023668050765991,
      "learning_rate": 1.943372394919655e-05,
      "loss": 0.5444,
      "step": 10296
    },
    {
      "epoch": 3.251341964003789,
      "grad_norm": 3.4017951488494873,
      "learning_rate": 1.9430215423479055e-05,
      "loss": 0.6636,
      "step": 10297
    },
    {
      "epoch": 3.2516577202399746,
      "grad_norm": 3.364196538925171,
      "learning_rate": 1.9426706897761562e-05,
      "loss": 0.7135,
      "step": 10298
    },
    {
      "epoch": 3.2519734764761603,
      "grad_norm": 3.2351999282836914,
      "learning_rate": 1.942319837204407e-05,
      "loss": 0.6708,
      "step": 10299
    },
    {
      "epoch": 3.252289232712346,
      "grad_norm": 3.396820068359375,
      "learning_rate": 1.9419689846326576e-05,
      "loss": 0.696,
      "step": 10300
    },
    {
      "epoch": 3.252604988948532,
      "grad_norm": 2.8697452545166016,
      "learning_rate": 1.9416181320609082e-05,
      "loss": 0.5814,
      "step": 10301
    },
    {
      "epoch": 3.2529207451847175,
      "grad_norm": 3.2182717323303223,
      "learning_rate": 1.9412672794891586e-05,
      "loss": 0.6629,
      "step": 10302
    },
    {
      "epoch": 3.2532365014209033,
      "grad_norm": 3.2885639667510986,
      "learning_rate": 1.9409164269174093e-05,
      "loss": 0.7016,
      "step": 10303
    },
    {
      "epoch": 3.2535522576570886,
      "grad_norm": 3.02417254447937,
      "learning_rate": 1.94056557434566e-05,
      "loss": 0.5974,
      "step": 10304
    },
    {
      "epoch": 3.2538680138932743,
      "grad_norm": 2.947120428085327,
      "learning_rate": 1.9402147217739106e-05,
      "loss": 0.6064,
      "step": 10305
    },
    {
      "epoch": 3.25418377012946,
      "grad_norm": 3.6504952907562256,
      "learning_rate": 1.9398638692021613e-05,
      "loss": 0.7051,
      "step": 10306
    },
    {
      "epoch": 3.2544995263656458,
      "grad_norm": 3.6358087062835693,
      "learning_rate": 1.939513016630412e-05,
      "loss": 0.6795,
      "step": 10307
    },
    {
      "epoch": 3.2548152826018315,
      "grad_norm": 3.219451904296875,
      "learning_rate": 1.9391621640586627e-05,
      "loss": 0.5979,
      "step": 10308
    },
    {
      "epoch": 3.2551310388380172,
      "grad_norm": 2.850249767303467,
      "learning_rate": 1.9388113114869134e-05,
      "loss": 0.5764,
      "step": 10309
    },
    {
      "epoch": 3.2554467950742025,
      "grad_norm": 2.477344512939453,
      "learning_rate": 1.938460458915164e-05,
      "loss": 0.6498,
      "step": 10310
    },
    {
      "epoch": 3.2557625513103883,
      "grad_norm": 3.199010133743286,
      "learning_rate": 1.9381096063434144e-05,
      "loss": 0.6201,
      "step": 10311
    },
    {
      "epoch": 3.256078307546574,
      "grad_norm": 2.8916287422180176,
      "learning_rate": 1.937758753771665e-05,
      "loss": 0.5542,
      "step": 10312
    },
    {
      "epoch": 3.2563940637827598,
      "grad_norm": 2.7492127418518066,
      "learning_rate": 1.9374079011999157e-05,
      "loss": 0.5714,
      "step": 10313
    },
    {
      "epoch": 3.2567098200189455,
      "grad_norm": 3.0528032779693604,
      "learning_rate": 1.9370570486281668e-05,
      "loss": 0.6449,
      "step": 10314
    },
    {
      "epoch": 3.2570255762551312,
      "grad_norm": 3.275914192199707,
      "learning_rate": 1.9367061960564174e-05,
      "loss": 0.6092,
      "step": 10315
    },
    {
      "epoch": 3.2573413324913165,
      "grad_norm": 3.1542177200317383,
      "learning_rate": 1.9363553434846678e-05,
      "loss": 0.6401,
      "step": 10316
    },
    {
      "epoch": 3.2576570887275023,
      "grad_norm": 3.798558235168457,
      "learning_rate": 1.9360044909129185e-05,
      "loss": 0.6545,
      "step": 10317
    },
    {
      "epoch": 3.257972844963688,
      "grad_norm": 3.3942224979400635,
      "learning_rate": 1.935653638341169e-05,
      "loss": 0.6139,
      "step": 10318
    },
    {
      "epoch": 3.2582886011998737,
      "grad_norm": 4.289555549621582,
      "learning_rate": 1.93530278576942e-05,
      "loss": 0.6519,
      "step": 10319
    },
    {
      "epoch": 3.2586043574360595,
      "grad_norm": 4.207433700561523,
      "learning_rate": 1.9349519331976705e-05,
      "loss": 0.6654,
      "step": 10320
    },
    {
      "epoch": 3.258920113672245,
      "grad_norm": 2.940674066543579,
      "learning_rate": 1.934601080625921e-05,
      "loss": 0.6465,
      "step": 10321
    },
    {
      "epoch": 3.2592358699084305,
      "grad_norm": 3.3874707221984863,
      "learning_rate": 1.9342502280541715e-05,
      "loss": 0.6484,
      "step": 10322
    },
    {
      "epoch": 3.2595516261446162,
      "grad_norm": 4.059054851531982,
      "learning_rate": 1.9338993754824226e-05,
      "loss": 0.64,
      "step": 10323
    },
    {
      "epoch": 3.259867382380802,
      "grad_norm": 2.802922248840332,
      "learning_rate": 1.9335485229106732e-05,
      "loss": 0.6949,
      "step": 10324
    },
    {
      "epoch": 3.2601831386169877,
      "grad_norm": 3.5021817684173584,
      "learning_rate": 1.9331976703389236e-05,
      "loss": 0.6418,
      "step": 10325
    },
    {
      "epoch": 3.2604988948531735,
      "grad_norm": 2.8240675926208496,
      "learning_rate": 1.9328468177671743e-05,
      "loss": 0.6415,
      "step": 10326
    },
    {
      "epoch": 3.260814651089359,
      "grad_norm": 3.4788122177124023,
      "learning_rate": 1.932495965195425e-05,
      "loss": 0.5932,
      "step": 10327
    },
    {
      "epoch": 3.2611304073255445,
      "grad_norm": 3.657236337661743,
      "learning_rate": 1.9321451126236756e-05,
      "loss": 0.6611,
      "step": 10328
    },
    {
      "epoch": 3.2614461635617302,
      "grad_norm": 3.5622847080230713,
      "learning_rate": 1.9317942600519263e-05,
      "loss": 0.5983,
      "step": 10329
    },
    {
      "epoch": 3.261761919797916,
      "grad_norm": 3.7372844219207764,
      "learning_rate": 1.931443407480177e-05,
      "loss": 0.6892,
      "step": 10330
    },
    {
      "epoch": 3.2620776760341017,
      "grad_norm": 4.315506935119629,
      "learning_rate": 1.9310925549084273e-05,
      "loss": 0.7541,
      "step": 10331
    },
    {
      "epoch": 3.2623934322702874,
      "grad_norm": 4.54455041885376,
      "learning_rate": 1.9307417023366784e-05,
      "loss": 0.6146,
      "step": 10332
    },
    {
      "epoch": 3.262709188506473,
      "grad_norm": 2.85166335105896,
      "learning_rate": 1.930390849764929e-05,
      "loss": 0.6095,
      "step": 10333
    },
    {
      "epoch": 3.263024944742659,
      "grad_norm": 3.7185118198394775,
      "learning_rate": 1.9300399971931797e-05,
      "loss": 0.7473,
      "step": 10334
    },
    {
      "epoch": 3.263340700978844,
      "grad_norm": 4.182090759277344,
      "learning_rate": 1.92968914462143e-05,
      "loss": 0.6294,
      "step": 10335
    },
    {
      "epoch": 3.26365645721503,
      "grad_norm": 4.104657173156738,
      "learning_rate": 1.9293382920496807e-05,
      "loss": 0.5762,
      "step": 10336
    },
    {
      "epoch": 3.2639722134512157,
      "grad_norm": 3.6152493953704834,
      "learning_rate": 1.9289874394779314e-05,
      "loss": 0.6544,
      "step": 10337
    },
    {
      "epoch": 3.2642879696874014,
      "grad_norm": 3.786168098449707,
      "learning_rate": 1.928636586906182e-05,
      "loss": 0.6285,
      "step": 10338
    },
    {
      "epoch": 3.264603725923587,
      "grad_norm": 3.602740526199341,
      "learning_rate": 1.9282857343344328e-05,
      "loss": 0.7228,
      "step": 10339
    },
    {
      "epoch": 3.2649194821597725,
      "grad_norm": 3.3990530967712402,
      "learning_rate": 1.927934881762683e-05,
      "loss": 0.6552,
      "step": 10340
    },
    {
      "epoch": 3.265235238395958,
      "grad_norm": 3.295746326446533,
      "learning_rate": 1.927584029190934e-05,
      "loss": 0.6637,
      "step": 10341
    },
    {
      "epoch": 3.265550994632144,
      "grad_norm": 3.4898841381073,
      "learning_rate": 1.9272331766191848e-05,
      "loss": 0.5311,
      "step": 10342
    },
    {
      "epoch": 3.2658667508683297,
      "grad_norm": 3.7069263458251953,
      "learning_rate": 1.9268823240474355e-05,
      "loss": 0.6741,
      "step": 10343
    },
    {
      "epoch": 3.2661825071045154,
      "grad_norm": 3.3401055335998535,
      "learning_rate": 1.9265314714756862e-05,
      "loss": 0.6517,
      "step": 10344
    },
    {
      "epoch": 3.266498263340701,
      "grad_norm": 3.546398878097534,
      "learning_rate": 1.9261806189039365e-05,
      "loss": 0.6773,
      "step": 10345
    },
    {
      "epoch": 3.266814019576887,
      "grad_norm": 3.4281136989593506,
      "learning_rate": 1.9258297663321872e-05,
      "loss": 0.7361,
      "step": 10346
    },
    {
      "epoch": 3.267129775813072,
      "grad_norm": 2.872624397277832,
      "learning_rate": 1.925478913760438e-05,
      "loss": 0.5668,
      "step": 10347
    },
    {
      "epoch": 3.267445532049258,
      "grad_norm": 3.239637613296509,
      "learning_rate": 1.9251280611886886e-05,
      "loss": 0.6422,
      "step": 10348
    },
    {
      "epoch": 3.2677612882854437,
      "grad_norm": 3.1542234420776367,
      "learning_rate": 1.9247772086169393e-05,
      "loss": 0.6053,
      "step": 10349
    },
    {
      "epoch": 3.2680770445216294,
      "grad_norm": 3.144033193588257,
      "learning_rate": 1.92442635604519e-05,
      "loss": 0.6706,
      "step": 10350
    },
    {
      "epoch": 3.268392800757815,
      "grad_norm": 3.212925910949707,
      "learning_rate": 1.9240755034734406e-05,
      "loss": 0.6296,
      "step": 10351
    },
    {
      "epoch": 3.2687085569940004,
      "grad_norm": 3.0306878089904785,
      "learning_rate": 1.9237246509016913e-05,
      "loss": 0.5748,
      "step": 10352
    },
    {
      "epoch": 3.269024313230186,
      "grad_norm": 3.3200185298919678,
      "learning_rate": 1.923373798329942e-05,
      "loss": 0.6794,
      "step": 10353
    },
    {
      "epoch": 3.269340069466372,
      "grad_norm": 3.8015668392181396,
      "learning_rate": 1.9230229457581923e-05,
      "loss": 0.6523,
      "step": 10354
    },
    {
      "epoch": 3.2696558257025576,
      "grad_norm": 3.100612163543701,
      "learning_rate": 1.922672093186443e-05,
      "loss": 0.6271,
      "step": 10355
    },
    {
      "epoch": 3.2699715819387434,
      "grad_norm": 2.851546287536621,
      "learning_rate": 1.9223212406146937e-05,
      "loss": 0.672,
      "step": 10356
    },
    {
      "epoch": 3.270287338174929,
      "grad_norm": 3.867074728012085,
      "learning_rate": 1.9219703880429444e-05,
      "loss": 0.6686,
      "step": 10357
    },
    {
      "epoch": 3.270603094411115,
      "grad_norm": 3.393667459487915,
      "learning_rate": 1.9216195354711954e-05,
      "loss": 0.6144,
      "step": 10358
    },
    {
      "epoch": 3.2709188506473,
      "grad_norm": 2.7742955684661865,
      "learning_rate": 1.9212686828994457e-05,
      "loss": 0.5786,
      "step": 10359
    },
    {
      "epoch": 3.271234606883486,
      "grad_norm": 3.9373133182525635,
      "learning_rate": 1.9209178303276964e-05,
      "loss": 0.7603,
      "step": 10360
    },
    {
      "epoch": 3.2715503631196716,
      "grad_norm": 3.62406325340271,
      "learning_rate": 1.920566977755947e-05,
      "loss": 0.6845,
      "step": 10361
    },
    {
      "epoch": 3.2718661193558574,
      "grad_norm": 3.7737889289855957,
      "learning_rate": 1.9202161251841978e-05,
      "loss": 0.573,
      "step": 10362
    },
    {
      "epoch": 3.272181875592043,
      "grad_norm": 3.3662972450256348,
      "learning_rate": 1.9198652726124485e-05,
      "loss": 0.7122,
      "step": 10363
    },
    {
      "epoch": 3.2724976318282284,
      "grad_norm": 3.171769857406616,
      "learning_rate": 1.9195144200406988e-05,
      "loss": 0.6137,
      "step": 10364
    },
    {
      "epoch": 3.272813388064414,
      "grad_norm": 2.97861909866333,
      "learning_rate": 1.9191635674689495e-05,
      "loss": 0.6168,
      "step": 10365
    },
    {
      "epoch": 3.2731291443006,
      "grad_norm": 2.943523645401001,
      "learning_rate": 1.9188127148972e-05,
      "loss": 0.6093,
      "step": 10366
    },
    {
      "epoch": 3.2734449005367856,
      "grad_norm": 3.597781181335449,
      "learning_rate": 1.9184618623254512e-05,
      "loss": 0.7559,
      "step": 10367
    },
    {
      "epoch": 3.2737606567729713,
      "grad_norm": 3.543078660964966,
      "learning_rate": 1.9181110097537015e-05,
      "loss": 0.6878,
      "step": 10368
    },
    {
      "epoch": 3.274076413009157,
      "grad_norm": 3.2997264862060547,
      "learning_rate": 1.9177601571819522e-05,
      "loss": 0.6506,
      "step": 10369
    },
    {
      "epoch": 3.274392169245343,
      "grad_norm": 4.122378826141357,
      "learning_rate": 1.917409304610203e-05,
      "loss": 0.7859,
      "step": 10370
    },
    {
      "epoch": 3.274707925481528,
      "grad_norm": 3.5254180431365967,
      "learning_rate": 1.9170584520384536e-05,
      "loss": 0.7039,
      "step": 10371
    },
    {
      "epoch": 3.275023681717714,
      "grad_norm": 3.529256582260132,
      "learning_rate": 1.9167075994667042e-05,
      "loss": 0.6266,
      "step": 10372
    },
    {
      "epoch": 3.2753394379538996,
      "grad_norm": 3.3980464935302734,
      "learning_rate": 1.9163567468949546e-05,
      "loss": 0.597,
      "step": 10373
    },
    {
      "epoch": 3.2756551941900853,
      "grad_norm": 4.055017948150635,
      "learning_rate": 1.9160058943232053e-05,
      "loss": 0.69,
      "step": 10374
    },
    {
      "epoch": 3.275970950426271,
      "grad_norm": 2.759070873260498,
      "learning_rate": 1.915655041751456e-05,
      "loss": 0.6734,
      "step": 10375
    },
    {
      "epoch": 3.2762867066624564,
      "grad_norm": 3.9028372764587402,
      "learning_rate": 1.915304189179707e-05,
      "loss": 0.6078,
      "step": 10376
    },
    {
      "epoch": 3.276602462898642,
      "grad_norm": 3.077742099761963,
      "learning_rate": 1.9149533366079577e-05,
      "loss": 0.6039,
      "step": 10377
    },
    {
      "epoch": 3.276918219134828,
      "grad_norm": 3.0038881301879883,
      "learning_rate": 1.914602484036208e-05,
      "loss": 0.6351,
      "step": 10378
    },
    {
      "epoch": 3.2772339753710136,
      "grad_norm": 3.50582218170166,
      "learning_rate": 1.9142516314644587e-05,
      "loss": 0.5907,
      "step": 10379
    },
    {
      "epoch": 3.2775497316071993,
      "grad_norm": 3.66711688041687,
      "learning_rate": 1.9139007788927094e-05,
      "loss": 0.6477,
      "step": 10380
    },
    {
      "epoch": 3.277865487843385,
      "grad_norm": 2.7749319076538086,
      "learning_rate": 1.91354992632096e-05,
      "loss": 0.6369,
      "step": 10381
    },
    {
      "epoch": 3.278181244079571,
      "grad_norm": 3.1458332538604736,
      "learning_rate": 1.9131990737492107e-05,
      "loss": 0.5761,
      "step": 10382
    },
    {
      "epoch": 3.278497000315756,
      "grad_norm": 3.642118215560913,
      "learning_rate": 1.912848221177461e-05,
      "loss": 0.72,
      "step": 10383
    },
    {
      "epoch": 3.278812756551942,
      "grad_norm": 3.3176352977752686,
      "learning_rate": 1.912497368605712e-05,
      "loss": 0.6343,
      "step": 10384
    },
    {
      "epoch": 3.2791285127881276,
      "grad_norm": 2.662444829940796,
      "learning_rate": 1.9121465160339628e-05,
      "loss": 0.5935,
      "step": 10385
    },
    {
      "epoch": 3.2794442690243133,
      "grad_norm": 3.186878204345703,
      "learning_rate": 1.9117956634622134e-05,
      "loss": 0.7359,
      "step": 10386
    },
    {
      "epoch": 3.279760025260499,
      "grad_norm": 2.991816520690918,
      "learning_rate": 1.9114448108904638e-05,
      "loss": 0.5935,
      "step": 10387
    },
    {
      "epoch": 3.2800757814966843,
      "grad_norm": 3.026256799697876,
      "learning_rate": 1.9110939583187145e-05,
      "loss": 0.6594,
      "step": 10388
    },
    {
      "epoch": 3.28039153773287,
      "grad_norm": 4.766117095947266,
      "learning_rate": 1.910743105746965e-05,
      "loss": 0.5979,
      "step": 10389
    },
    {
      "epoch": 3.280707293969056,
      "grad_norm": 3.540463447570801,
      "learning_rate": 1.910392253175216e-05,
      "loss": 0.5967,
      "step": 10390
    },
    {
      "epoch": 3.2810230502052415,
      "grad_norm": 3.3735220432281494,
      "learning_rate": 1.9100414006034665e-05,
      "loss": 0.787,
      "step": 10391
    },
    {
      "epoch": 3.2813388064414273,
      "grad_norm": 3.4302828311920166,
      "learning_rate": 1.9096905480317172e-05,
      "loss": 0.7389,
      "step": 10392
    },
    {
      "epoch": 3.281654562677613,
      "grad_norm": 3.7126450538635254,
      "learning_rate": 1.909339695459968e-05,
      "loss": 0.7278,
      "step": 10393
    },
    {
      "epoch": 3.2819703189137988,
      "grad_norm": 3.8983347415924072,
      "learning_rate": 1.9089888428882186e-05,
      "loss": 0.6051,
      "step": 10394
    },
    {
      "epoch": 3.282286075149984,
      "grad_norm": 3.7711758613586426,
      "learning_rate": 1.9086379903164692e-05,
      "loss": 0.6713,
      "step": 10395
    },
    {
      "epoch": 3.28260183138617,
      "grad_norm": 3.1386425495147705,
      "learning_rate": 1.90828713774472e-05,
      "loss": 0.5838,
      "step": 10396
    },
    {
      "epoch": 3.2829175876223555,
      "grad_norm": 2.538079261779785,
      "learning_rate": 1.9079362851729703e-05,
      "loss": 0.5761,
      "step": 10397
    },
    {
      "epoch": 3.2832333438585413,
      "grad_norm": 2.9173221588134766,
      "learning_rate": 1.907585432601221e-05,
      "loss": 0.6874,
      "step": 10398
    },
    {
      "epoch": 3.283549100094727,
      "grad_norm": 3.455068349838257,
      "learning_rate": 1.9072345800294716e-05,
      "loss": 0.6789,
      "step": 10399
    },
    {
      "epoch": 3.2838648563309123,
      "grad_norm": 4.0587968826293945,
      "learning_rate": 1.9068837274577223e-05,
      "loss": 0.6936,
      "step": 10400
    },
    {
      "epoch": 3.2838648563309123,
      "eval_loss": 0.8230278491973877,
      "eval_runtime": 4.62,
      "eval_samples_per_second": 538.958,
      "eval_steps_per_second": 4.545,
      "step": 10400
    },
    {
      "epoch": 3.284180612567098,
      "grad_norm": 3.776155471801758,
      "learning_rate": 1.906532874885973e-05,
      "loss": 0.6932,
      "step": 10401
    },
    {
      "epoch": 3.2844963688032838,
      "grad_norm": 4.382483959197998,
      "learning_rate": 1.9061820223142237e-05,
      "loss": 0.6663,
      "step": 10402
    },
    {
      "epoch": 3.2848121250394695,
      "grad_norm": 3.038742780685425,
      "learning_rate": 1.9058311697424744e-05,
      "loss": 0.691,
      "step": 10403
    },
    {
      "epoch": 3.2851278812756552,
      "grad_norm": 4.043774604797363,
      "learning_rate": 1.905480317170725e-05,
      "loss": 0.6809,
      "step": 10404
    },
    {
      "epoch": 3.285443637511841,
      "grad_norm": 3.393939733505249,
      "learning_rate": 1.9051294645989757e-05,
      "loss": 0.6719,
      "step": 10405
    },
    {
      "epoch": 3.2857593937480267,
      "grad_norm": 3.3927531242370605,
      "learning_rate": 1.9047786120272264e-05,
      "loss": 0.5972,
      "step": 10406
    },
    {
      "epoch": 3.286075149984212,
      "grad_norm": 3.1021335124969482,
      "learning_rate": 1.9044277594554767e-05,
      "loss": 0.6338,
      "step": 10407
    },
    {
      "epoch": 3.2863909062203978,
      "grad_norm": 2.953178644180298,
      "learning_rate": 1.9040769068837274e-05,
      "loss": 0.6346,
      "step": 10408
    },
    {
      "epoch": 3.2867066624565835,
      "grad_norm": 3.519547462463379,
      "learning_rate": 1.903726054311978e-05,
      "loss": 0.6886,
      "step": 10409
    },
    {
      "epoch": 3.2870224186927692,
      "grad_norm": 3.399038791656494,
      "learning_rate": 1.9033752017402288e-05,
      "loss": 0.6059,
      "step": 10410
    },
    {
      "epoch": 3.287338174928955,
      "grad_norm": 3.580662250518799,
      "learning_rate": 1.9030243491684795e-05,
      "loss": 0.6801,
      "step": 10411
    },
    {
      "epoch": 3.2876539311651403,
      "grad_norm": 3.4664840698242188,
      "learning_rate": 1.90267349659673e-05,
      "loss": 0.554,
      "step": 10412
    },
    {
      "epoch": 3.287969687401326,
      "grad_norm": 3.689781904220581,
      "learning_rate": 1.9023226440249808e-05,
      "loss": 0.6402,
      "step": 10413
    },
    {
      "epoch": 3.2882854436375117,
      "grad_norm": 3.236832857131958,
      "learning_rate": 1.9019717914532315e-05,
      "loss": 0.5901,
      "step": 10414
    },
    {
      "epoch": 3.2886011998736975,
      "grad_norm": 2.7588064670562744,
      "learning_rate": 1.9016209388814822e-05,
      "loss": 0.568,
      "step": 10415
    },
    {
      "epoch": 3.288916956109883,
      "grad_norm": 2.928332805633545,
      "learning_rate": 1.9012700863097325e-05,
      "loss": 0.6157,
      "step": 10416
    },
    {
      "epoch": 3.289232712346069,
      "grad_norm": 3.6036908626556396,
      "learning_rate": 1.9009192337379832e-05,
      "loss": 0.6262,
      "step": 10417
    },
    {
      "epoch": 3.2895484685822547,
      "grad_norm": 3.4381661415100098,
      "learning_rate": 1.900568381166234e-05,
      "loss": 0.6386,
      "step": 10418
    },
    {
      "epoch": 3.28986422481844,
      "grad_norm": 2.8947770595550537,
      "learning_rate": 1.900217528594485e-05,
      "loss": 0.5977,
      "step": 10419
    },
    {
      "epoch": 3.2901799810546257,
      "grad_norm": 3.495931386947632,
      "learning_rate": 1.8998666760227356e-05,
      "loss": 0.6665,
      "step": 10420
    },
    {
      "epoch": 3.2904957372908115,
      "grad_norm": 4.019857883453369,
      "learning_rate": 1.899515823450986e-05,
      "loss": 0.6576,
      "step": 10421
    },
    {
      "epoch": 3.290811493526997,
      "grad_norm": 3.6135308742523193,
      "learning_rate": 1.8991649708792366e-05,
      "loss": 0.7097,
      "step": 10422
    },
    {
      "epoch": 3.291127249763183,
      "grad_norm": 3.559999704360962,
      "learning_rate": 1.8988141183074873e-05,
      "loss": 0.6537,
      "step": 10423
    },
    {
      "epoch": 3.2914430059993687,
      "grad_norm": 3.1776390075683594,
      "learning_rate": 1.898463265735738e-05,
      "loss": 0.6937,
      "step": 10424
    },
    {
      "epoch": 3.291758762235554,
      "grad_norm": 3.7184865474700928,
      "learning_rate": 1.8981124131639887e-05,
      "loss": 0.6804,
      "step": 10425
    },
    {
      "epoch": 3.2920745184717397,
      "grad_norm": 3.2578036785125732,
      "learning_rate": 1.897761560592239e-05,
      "loss": 0.6513,
      "step": 10426
    },
    {
      "epoch": 3.2923902747079254,
      "grad_norm": 3.7190237045288086,
      "learning_rate": 1.8974107080204897e-05,
      "loss": 0.7191,
      "step": 10427
    },
    {
      "epoch": 3.292706030944111,
      "grad_norm": 3.2457661628723145,
      "learning_rate": 1.8970598554487407e-05,
      "loss": 0.6232,
      "step": 10428
    },
    {
      "epoch": 3.293021787180297,
      "grad_norm": 4.887528419494629,
      "learning_rate": 1.8967090028769914e-05,
      "loss": 0.738,
      "step": 10429
    },
    {
      "epoch": 3.2933375434164827,
      "grad_norm": 3.5538158416748047,
      "learning_rate": 1.8963581503052417e-05,
      "loss": 0.5858,
      "step": 10430
    },
    {
      "epoch": 3.293653299652668,
      "grad_norm": 3.2262871265411377,
      "learning_rate": 1.8960072977334924e-05,
      "loss": 0.6835,
      "step": 10431
    },
    {
      "epoch": 3.2939690558888537,
      "grad_norm": 3.0720834732055664,
      "learning_rate": 1.895656445161743e-05,
      "loss": 0.5845,
      "step": 10432
    },
    {
      "epoch": 3.2942848121250394,
      "grad_norm": 3.3020927906036377,
      "learning_rate": 1.8953055925899938e-05,
      "loss": 0.5638,
      "step": 10433
    },
    {
      "epoch": 3.294600568361225,
      "grad_norm": 3.2553625106811523,
      "learning_rate": 1.8949547400182445e-05,
      "loss": 0.5383,
      "step": 10434
    },
    {
      "epoch": 3.294916324597411,
      "grad_norm": 3.2247159481048584,
      "learning_rate": 1.894603887446495e-05,
      "loss": 0.6763,
      "step": 10435
    },
    {
      "epoch": 3.2952320808335966,
      "grad_norm": 3.1889402866363525,
      "learning_rate": 1.8942530348747455e-05,
      "loss": 0.6437,
      "step": 10436
    },
    {
      "epoch": 3.295547837069782,
      "grad_norm": 3.512279748916626,
      "learning_rate": 1.8939021823029965e-05,
      "loss": 0.748,
      "step": 10437
    },
    {
      "epoch": 3.2958635933059677,
      "grad_norm": 3.761301040649414,
      "learning_rate": 1.8935513297312472e-05,
      "loss": 0.7834,
      "step": 10438
    },
    {
      "epoch": 3.2961793495421534,
      "grad_norm": 3.2527379989624023,
      "learning_rate": 1.893200477159498e-05,
      "loss": 0.6903,
      "step": 10439
    },
    {
      "epoch": 3.296495105778339,
      "grad_norm": 3.591247320175171,
      "learning_rate": 1.8928496245877482e-05,
      "loss": 0.6159,
      "step": 10440
    },
    {
      "epoch": 3.296810862014525,
      "grad_norm": 3.0857372283935547,
      "learning_rate": 1.892498772015999e-05,
      "loss": 0.7043,
      "step": 10441
    },
    {
      "epoch": 3.2971266182507106,
      "grad_norm": 2.982433319091797,
      "learning_rate": 1.8921479194442496e-05,
      "loss": 0.6606,
      "step": 10442
    },
    {
      "epoch": 3.297442374486896,
      "grad_norm": 3.401156425476074,
      "learning_rate": 1.8917970668725002e-05,
      "loss": 0.6295,
      "step": 10443
    },
    {
      "epoch": 3.2977581307230817,
      "grad_norm": 3.0269205570220947,
      "learning_rate": 1.891446214300751e-05,
      "loss": 0.6663,
      "step": 10444
    },
    {
      "epoch": 3.2980738869592674,
      "grad_norm": 3.1610217094421387,
      "learning_rate": 1.8910953617290016e-05,
      "loss": 0.6451,
      "step": 10445
    },
    {
      "epoch": 3.298389643195453,
      "grad_norm": 3.12648868560791,
      "learning_rate": 1.8907445091572523e-05,
      "loss": 0.6822,
      "step": 10446
    },
    {
      "epoch": 3.298705399431639,
      "grad_norm": 3.9271652698516846,
      "learning_rate": 1.890393656585503e-05,
      "loss": 0.6046,
      "step": 10447
    },
    {
      "epoch": 3.2990211556678246,
      "grad_norm": 3.2976744174957275,
      "learning_rate": 1.8900428040137537e-05,
      "loss": 0.7998,
      "step": 10448
    },
    {
      "epoch": 3.2993369119040103,
      "grad_norm": 3.2609610557556152,
      "learning_rate": 1.8896919514420043e-05,
      "loss": 0.6624,
      "step": 10449
    },
    {
      "epoch": 3.2996526681401956,
      "grad_norm": 2.852773427963257,
      "learning_rate": 1.8893410988702547e-05,
      "loss": 0.6884,
      "step": 10450
    },
    {
      "epoch": 3.2999684243763814,
      "grad_norm": 3.6356256008148193,
      "learning_rate": 1.8889902462985054e-05,
      "loss": 0.6846,
      "step": 10451
    },
    {
      "epoch": 3.300284180612567,
      "grad_norm": 3.582871437072754,
      "learning_rate": 1.888639393726756e-05,
      "loss": 0.6939,
      "step": 10452
    },
    {
      "epoch": 3.300599936848753,
      "grad_norm": 3.260270833969116,
      "learning_rate": 1.8882885411550067e-05,
      "loss": 0.7425,
      "step": 10453
    },
    {
      "epoch": 3.3009156930849386,
      "grad_norm": 3.781606674194336,
      "learning_rate": 1.8879376885832574e-05,
      "loss": 0.6893,
      "step": 10454
    },
    {
      "epoch": 3.301231449321124,
      "grad_norm": 3.1881349086761475,
      "learning_rate": 1.887586836011508e-05,
      "loss": 0.662,
      "step": 10455
    },
    {
      "epoch": 3.3015472055573096,
      "grad_norm": 3.6805083751678467,
      "learning_rate": 1.8872359834397588e-05,
      "loss": 0.6357,
      "step": 10456
    },
    {
      "epoch": 3.3018629617934954,
      "grad_norm": 4.168125629425049,
      "learning_rate": 1.8868851308680094e-05,
      "loss": 0.7476,
      "step": 10457
    },
    {
      "epoch": 3.302178718029681,
      "grad_norm": 3.6728882789611816,
      "learning_rate": 1.88653427829626e-05,
      "loss": 0.6418,
      "step": 10458
    },
    {
      "epoch": 3.302494474265867,
      "grad_norm": 3.480102300643921,
      "learning_rate": 1.8861834257245105e-05,
      "loss": 0.6858,
      "step": 10459
    },
    {
      "epoch": 3.3028102305020526,
      "grad_norm": 3.887155532836914,
      "learning_rate": 1.885832573152761e-05,
      "loss": 0.7208,
      "step": 10460
    },
    {
      "epoch": 3.3031259867382383,
      "grad_norm": 3.057934522628784,
      "learning_rate": 1.885481720581012e-05,
      "loss": 0.659,
      "step": 10461
    },
    {
      "epoch": 3.3034417429744236,
      "grad_norm": 3.5910215377807617,
      "learning_rate": 1.8851308680092625e-05,
      "loss": 0.5872,
      "step": 10462
    },
    {
      "epoch": 3.3037574992106093,
      "grad_norm": 4.240668773651123,
      "learning_rate": 1.8847800154375135e-05,
      "loss": 0.6993,
      "step": 10463
    },
    {
      "epoch": 3.304073255446795,
      "grad_norm": 3.504152774810791,
      "learning_rate": 1.884429162865764e-05,
      "loss": 0.6165,
      "step": 10464
    },
    {
      "epoch": 3.304389011682981,
      "grad_norm": 3.8798935413360596,
      "learning_rate": 1.8840783102940146e-05,
      "loss": 0.6567,
      "step": 10465
    },
    {
      "epoch": 3.3047047679191666,
      "grad_norm": 3.129364252090454,
      "learning_rate": 1.8837274577222652e-05,
      "loss": 0.6198,
      "step": 10466
    },
    {
      "epoch": 3.305020524155352,
      "grad_norm": 4.026158332824707,
      "learning_rate": 1.883376605150516e-05,
      "loss": 0.6372,
      "step": 10467
    },
    {
      "epoch": 3.3053362803915376,
      "grad_norm": 2.9339213371276855,
      "learning_rate": 1.8830257525787666e-05,
      "loss": 0.6397,
      "step": 10468
    },
    {
      "epoch": 3.3056520366277233,
      "grad_norm": 3.1807796955108643,
      "learning_rate": 1.882674900007017e-05,
      "loss": 0.5937,
      "step": 10469
    },
    {
      "epoch": 3.305967792863909,
      "grad_norm": 3.688600778579712,
      "learning_rate": 1.8823240474352676e-05,
      "loss": 0.6468,
      "step": 10470
    },
    {
      "epoch": 3.306283549100095,
      "grad_norm": 3.434755802154541,
      "learning_rate": 1.8819731948635183e-05,
      "loss": 0.6927,
      "step": 10471
    },
    {
      "epoch": 3.3065993053362805,
      "grad_norm": 3.819119930267334,
      "learning_rate": 1.8816223422917693e-05,
      "loss": 0.7226,
      "step": 10472
    },
    {
      "epoch": 3.3069150615724663,
      "grad_norm": 3.216155767440796,
      "learning_rate": 1.8812714897200197e-05,
      "loss": 0.6389,
      "step": 10473
    },
    {
      "epoch": 3.3072308178086516,
      "grad_norm": 3.3928565979003906,
      "learning_rate": 1.8809206371482704e-05,
      "loss": 0.6182,
      "step": 10474
    },
    {
      "epoch": 3.3075465740448373,
      "grad_norm": 3.03901743888855,
      "learning_rate": 1.880569784576521e-05,
      "loss": 0.6629,
      "step": 10475
    },
    {
      "epoch": 3.307862330281023,
      "grad_norm": 3.1207544803619385,
      "learning_rate": 1.8802189320047717e-05,
      "loss": 0.557,
      "step": 10476
    },
    {
      "epoch": 3.308178086517209,
      "grad_norm": 4.004955768585205,
      "learning_rate": 1.8798680794330224e-05,
      "loss": 0.6892,
      "step": 10477
    },
    {
      "epoch": 3.3084938427533945,
      "grad_norm": 3.767059326171875,
      "learning_rate": 1.8795172268612727e-05,
      "loss": 0.6453,
      "step": 10478
    },
    {
      "epoch": 3.30880959898958,
      "grad_norm": 3.79787015914917,
      "learning_rate": 1.8791663742895234e-05,
      "loss": 0.7175,
      "step": 10479
    },
    {
      "epoch": 3.3091253552257656,
      "grad_norm": 2.9899797439575195,
      "learning_rate": 1.8788155217177744e-05,
      "loss": 0.6385,
      "step": 10480
    },
    {
      "epoch": 3.3094411114619513,
      "grad_norm": 4.006207466125488,
      "learning_rate": 1.878464669146025e-05,
      "loss": 0.5968,
      "step": 10481
    },
    {
      "epoch": 3.309756867698137,
      "grad_norm": 2.8569912910461426,
      "learning_rate": 1.8781138165742758e-05,
      "loss": 0.5799,
      "step": 10482
    },
    {
      "epoch": 3.3100726239343228,
      "grad_norm": 3.31856632232666,
      "learning_rate": 1.877762964002526e-05,
      "loss": 0.6971,
      "step": 10483
    },
    {
      "epoch": 3.3103883801705085,
      "grad_norm": 2.9327585697174072,
      "learning_rate": 1.8774121114307768e-05,
      "loss": 0.649,
      "step": 10484
    },
    {
      "epoch": 3.3107041364066943,
      "grad_norm": 2.9490747451782227,
      "learning_rate": 1.8770612588590275e-05,
      "loss": 0.6632,
      "step": 10485
    },
    {
      "epoch": 3.3110198926428795,
      "grad_norm": 2.7894232273101807,
      "learning_rate": 1.8767104062872782e-05,
      "loss": 0.6358,
      "step": 10486
    },
    {
      "epoch": 3.3113356488790653,
      "grad_norm": 3.5231800079345703,
      "learning_rate": 1.876359553715529e-05,
      "loss": 0.644,
      "step": 10487
    },
    {
      "epoch": 3.311651405115251,
      "grad_norm": 3.1541454792022705,
      "learning_rate": 1.8760087011437792e-05,
      "loss": 0.6925,
      "step": 10488
    },
    {
      "epoch": 3.3119671613514368,
      "grad_norm": 3.81650447845459,
      "learning_rate": 1.8756578485720302e-05,
      "loss": 0.8031,
      "step": 10489
    },
    {
      "epoch": 3.3122829175876225,
      "grad_norm": 3.1729238033294678,
      "learning_rate": 1.875306996000281e-05,
      "loss": 0.6337,
      "step": 10490
    },
    {
      "epoch": 3.312598673823808,
      "grad_norm": 3.25607967376709,
      "learning_rate": 1.8749561434285316e-05,
      "loss": 0.672,
      "step": 10491
    },
    {
      "epoch": 3.3129144300599935,
      "grad_norm": 3.569378137588501,
      "learning_rate": 1.874605290856782e-05,
      "loss": 0.6841,
      "step": 10492
    },
    {
      "epoch": 3.3132301862961793,
      "grad_norm": 3.3266537189483643,
      "learning_rate": 1.8742544382850326e-05,
      "loss": 0.6711,
      "step": 10493
    },
    {
      "epoch": 3.313545942532365,
      "grad_norm": 3.47688889503479,
      "learning_rate": 1.8739035857132833e-05,
      "loss": 0.6204,
      "step": 10494
    },
    {
      "epoch": 3.3138616987685507,
      "grad_norm": 3.2027156352996826,
      "learning_rate": 1.873552733141534e-05,
      "loss": 0.6066,
      "step": 10495
    },
    {
      "epoch": 3.3141774550047365,
      "grad_norm": 2.842224359512329,
      "learning_rate": 1.8732018805697847e-05,
      "loss": 0.6952,
      "step": 10496
    },
    {
      "epoch": 3.314493211240922,
      "grad_norm": 3.6460001468658447,
      "learning_rate": 1.8728510279980353e-05,
      "loss": 0.698,
      "step": 10497
    },
    {
      "epoch": 3.3148089674771075,
      "grad_norm": 3.576218366622925,
      "learning_rate": 1.872500175426286e-05,
      "loss": 0.6105,
      "step": 10498
    },
    {
      "epoch": 3.3151247237132933,
      "grad_norm": 2.9467873573303223,
      "learning_rate": 1.8721493228545367e-05,
      "loss": 0.6799,
      "step": 10499
    },
    {
      "epoch": 3.315440479949479,
      "grad_norm": 4.000137805938721,
      "learning_rate": 1.8717984702827874e-05,
      "loss": 0.7901,
      "step": 10500
    },
    {
      "epoch": 3.3157562361856647,
      "grad_norm": 3.928105592727661,
      "learning_rate": 1.871447617711038e-05,
      "loss": 0.6239,
      "step": 10501
    },
    {
      "epoch": 3.3160719924218505,
      "grad_norm": 4.604343891143799,
      "learning_rate": 1.8710967651392884e-05,
      "loss": 0.7116,
      "step": 10502
    },
    {
      "epoch": 3.3163877486580358,
      "grad_norm": 3.0322184562683105,
      "learning_rate": 1.870745912567539e-05,
      "loss": 0.6631,
      "step": 10503
    },
    {
      "epoch": 3.3167035048942215,
      "grad_norm": 2.7086265087127686,
      "learning_rate": 1.8703950599957898e-05,
      "loss": 0.6847,
      "step": 10504
    },
    {
      "epoch": 3.3170192611304072,
      "grad_norm": 2.977031946182251,
      "learning_rate": 1.8700442074240405e-05,
      "loss": 0.6032,
      "step": 10505
    },
    {
      "epoch": 3.317335017366593,
      "grad_norm": 3.546543836593628,
      "learning_rate": 1.869693354852291e-05,
      "loss": 0.802,
      "step": 10506
    },
    {
      "epoch": 3.3176507736027787,
      "grad_norm": 3.2068636417388916,
      "learning_rate": 1.8693425022805418e-05,
      "loss": 0.7172,
      "step": 10507
    },
    {
      "epoch": 3.3179665298389645,
      "grad_norm": 3.5113978385925293,
      "learning_rate": 1.8689916497087925e-05,
      "loss": 0.6697,
      "step": 10508
    },
    {
      "epoch": 3.31828228607515,
      "grad_norm": 3.4684815406799316,
      "learning_rate": 1.8686407971370432e-05,
      "loss": 0.6581,
      "step": 10509
    },
    {
      "epoch": 3.3185980423113355,
      "grad_norm": 3.032912015914917,
      "learning_rate": 1.868289944565294e-05,
      "loss": 0.5311,
      "step": 10510
    },
    {
      "epoch": 3.318913798547521,
      "grad_norm": 3.9325692653656006,
      "learning_rate": 1.8679390919935445e-05,
      "loss": 0.7053,
      "step": 10511
    },
    {
      "epoch": 3.319229554783707,
      "grad_norm": 3.368800163269043,
      "learning_rate": 1.867588239421795e-05,
      "loss": 0.6504,
      "step": 10512
    },
    {
      "epoch": 3.3195453110198927,
      "grad_norm": 3.384611129760742,
      "learning_rate": 1.8672373868500456e-05,
      "loss": 0.7398,
      "step": 10513
    },
    {
      "epoch": 3.3198610672560784,
      "grad_norm": 5.2799153327941895,
      "learning_rate": 1.8668865342782962e-05,
      "loss": 0.7427,
      "step": 10514
    },
    {
      "epoch": 3.3201768234922637,
      "grad_norm": 3.0566489696502686,
      "learning_rate": 1.8665356817065473e-05,
      "loss": 0.608,
      "step": 10515
    },
    {
      "epoch": 3.3204925797284495,
      "grad_norm": 3.2417714595794678,
      "learning_rate": 1.8661848291347976e-05,
      "loss": 0.6182,
      "step": 10516
    },
    {
      "epoch": 3.320808335964635,
      "grad_norm": 3.27377986907959,
      "learning_rate": 1.8658339765630483e-05,
      "loss": 0.635,
      "step": 10517
    },
    {
      "epoch": 3.321124092200821,
      "grad_norm": 3.5310723781585693,
      "learning_rate": 1.865483123991299e-05,
      "loss": 0.6863,
      "step": 10518
    },
    {
      "epoch": 3.3214398484370067,
      "grad_norm": 3.6508193016052246,
      "learning_rate": 1.8651322714195497e-05,
      "loss": 0.664,
      "step": 10519
    },
    {
      "epoch": 3.3217556046731924,
      "grad_norm": 3.1734390258789062,
      "learning_rate": 1.8647814188478003e-05,
      "loss": 0.6674,
      "step": 10520
    },
    {
      "epoch": 3.322071360909378,
      "grad_norm": 3.085780620574951,
      "learning_rate": 1.8644305662760507e-05,
      "loss": 0.6563,
      "step": 10521
    },
    {
      "epoch": 3.3223871171455635,
      "grad_norm": 3.0345544815063477,
      "learning_rate": 1.8640797137043014e-05,
      "loss": 0.5641,
      "step": 10522
    },
    {
      "epoch": 3.322702873381749,
      "grad_norm": 2.7303011417388916,
      "learning_rate": 1.863728861132552e-05,
      "loss": 0.5534,
      "step": 10523
    },
    {
      "epoch": 3.323018629617935,
      "grad_norm": 3.3963992595672607,
      "learning_rate": 1.863378008560803e-05,
      "loss": 0.6593,
      "step": 10524
    },
    {
      "epoch": 3.3233343858541207,
      "grad_norm": 3.1082472801208496,
      "learning_rate": 1.8630271559890537e-05,
      "loss": 0.6288,
      "step": 10525
    },
    {
      "epoch": 3.3236501420903064,
      "grad_norm": 3.2934250831604004,
      "learning_rate": 1.862676303417304e-05,
      "loss": 0.6798,
      "step": 10526
    },
    {
      "epoch": 3.323965898326492,
      "grad_norm": 3.1154637336730957,
      "learning_rate": 1.8623254508455548e-05,
      "loss": 0.6707,
      "step": 10527
    },
    {
      "epoch": 3.3242816545626774,
      "grad_norm": 3.080739974975586,
      "learning_rate": 1.8619745982738054e-05,
      "loss": 0.6293,
      "step": 10528
    },
    {
      "epoch": 3.324597410798863,
      "grad_norm": 3.06721830368042,
      "learning_rate": 1.861623745702056e-05,
      "loss": 0.67,
      "step": 10529
    },
    {
      "epoch": 3.324913167035049,
      "grad_norm": 2.9549779891967773,
      "learning_rate": 1.8612728931303068e-05,
      "loss": 0.6129,
      "step": 10530
    },
    {
      "epoch": 3.3252289232712346,
      "grad_norm": 4.002649307250977,
      "learning_rate": 1.860922040558557e-05,
      "loss": 0.6645,
      "step": 10531
    },
    {
      "epoch": 3.3255446795074204,
      "grad_norm": 2.9023423194885254,
      "learning_rate": 1.860571187986808e-05,
      "loss": 0.6116,
      "step": 10532
    },
    {
      "epoch": 3.325860435743606,
      "grad_norm": 3.829345703125,
      "learning_rate": 1.860220335415059e-05,
      "loss": 0.7418,
      "step": 10533
    },
    {
      "epoch": 3.3261761919797914,
      "grad_norm": 3.196843385696411,
      "learning_rate": 1.8598694828433095e-05,
      "loss": 0.628,
      "step": 10534
    },
    {
      "epoch": 3.326491948215977,
      "grad_norm": 3.2763140201568604,
      "learning_rate": 1.85951863027156e-05,
      "loss": 0.579,
      "step": 10535
    },
    {
      "epoch": 3.326807704452163,
      "grad_norm": 3.334104537963867,
      "learning_rate": 1.8591677776998106e-05,
      "loss": 0.6268,
      "step": 10536
    },
    {
      "epoch": 3.3271234606883486,
      "grad_norm": 2.9221808910369873,
      "learning_rate": 1.8588169251280612e-05,
      "loss": 0.5513,
      "step": 10537
    },
    {
      "epoch": 3.3274392169245344,
      "grad_norm": 3.890226125717163,
      "learning_rate": 1.858466072556312e-05,
      "loss": 0.6672,
      "step": 10538
    },
    {
      "epoch": 3.32775497316072,
      "grad_norm": 3.6631436347961426,
      "learning_rate": 1.8581152199845626e-05,
      "loss": 0.6224,
      "step": 10539
    },
    {
      "epoch": 3.3280707293969054,
      "grad_norm": 3.0324063301086426,
      "learning_rate": 1.8577643674128133e-05,
      "loss": 0.6504,
      "step": 10540
    },
    {
      "epoch": 3.328386485633091,
      "grad_norm": 2.939692258834839,
      "learning_rate": 1.8574135148410636e-05,
      "loss": 0.5824,
      "step": 10541
    },
    {
      "epoch": 3.328702241869277,
      "grad_norm": 3.69683837890625,
      "learning_rate": 1.8570626622693146e-05,
      "loss": 0.6995,
      "step": 10542
    },
    {
      "epoch": 3.3290179981054626,
      "grad_norm": 4.381834030151367,
      "learning_rate": 1.8567118096975653e-05,
      "loss": 0.6276,
      "step": 10543
    },
    {
      "epoch": 3.3293337543416484,
      "grad_norm": 3.232088565826416,
      "learning_rate": 1.856360957125816e-05,
      "loss": 0.6849,
      "step": 10544
    },
    {
      "epoch": 3.329649510577834,
      "grad_norm": 3.302194595336914,
      "learning_rate": 1.8560101045540664e-05,
      "loss": 0.7778,
      "step": 10545
    },
    {
      "epoch": 3.3299652668140194,
      "grad_norm": 3.537904739379883,
      "learning_rate": 1.855659251982317e-05,
      "loss": 0.7107,
      "step": 10546
    },
    {
      "epoch": 3.330281023050205,
      "grad_norm": 3.4160642623901367,
      "learning_rate": 1.8553083994105677e-05,
      "loss": 0.6126,
      "step": 10547
    },
    {
      "epoch": 3.330596779286391,
      "grad_norm": 3.993598222732544,
      "learning_rate": 1.8549575468388184e-05,
      "loss": 0.6867,
      "step": 10548
    },
    {
      "epoch": 3.3309125355225766,
      "grad_norm": 3.9470221996307373,
      "learning_rate": 1.854606694267069e-05,
      "loss": 0.634,
      "step": 10549
    },
    {
      "epoch": 3.3312282917587623,
      "grad_norm": 3.9147868156433105,
      "learning_rate": 1.8542558416953198e-05,
      "loss": 0.6406,
      "step": 10550
    },
    {
      "epoch": 3.331544047994948,
      "grad_norm": 3.2672197818756104,
      "learning_rate": 1.8539049891235704e-05,
      "loss": 0.6174,
      "step": 10551
    },
    {
      "epoch": 3.3318598042311334,
      "grad_norm": 3.3762056827545166,
      "learning_rate": 1.853554136551821e-05,
      "loss": 0.6493,
      "step": 10552
    },
    {
      "epoch": 3.332175560467319,
      "grad_norm": 3.630948305130005,
      "learning_rate": 1.8532032839800718e-05,
      "loss": 0.6584,
      "step": 10553
    },
    {
      "epoch": 3.332491316703505,
      "grad_norm": 3.5177571773529053,
      "learning_rate": 1.8528524314083225e-05,
      "loss": 0.655,
      "step": 10554
    },
    {
      "epoch": 3.3328070729396906,
      "grad_norm": 2.845151901245117,
      "learning_rate": 1.8525015788365728e-05,
      "loss": 0.6469,
      "step": 10555
    },
    {
      "epoch": 3.3331228291758763,
      "grad_norm": 4.073164939880371,
      "learning_rate": 1.8521507262648235e-05,
      "loss": 0.717,
      "step": 10556
    },
    {
      "epoch": 3.333438585412062,
      "grad_norm": 3.1467056274414062,
      "learning_rate": 1.8517998736930742e-05,
      "loss": 0.6663,
      "step": 10557
    },
    {
      "epoch": 3.3337543416482474,
      "grad_norm": 3.2521822452545166,
      "learning_rate": 1.851449021121325e-05,
      "loss": 0.6379,
      "step": 10558
    },
    {
      "epoch": 3.334070097884433,
      "grad_norm": 3.289311408996582,
      "learning_rate": 1.8510981685495756e-05,
      "loss": 0.6177,
      "step": 10559
    },
    {
      "epoch": 3.334385854120619,
      "grad_norm": 3.442565679550171,
      "learning_rate": 1.8507473159778262e-05,
      "loss": 0.6106,
      "step": 10560
    },
    {
      "epoch": 3.3347016103568046,
      "grad_norm": 3.7345407009124756,
      "learning_rate": 1.850396463406077e-05,
      "loss": 0.7066,
      "step": 10561
    },
    {
      "epoch": 3.3350173665929903,
      "grad_norm": 3.3387506008148193,
      "learning_rate": 1.8500456108343276e-05,
      "loss": 0.6513,
      "step": 10562
    },
    {
      "epoch": 3.335333122829176,
      "grad_norm": 3.461327314376831,
      "learning_rate": 1.8496947582625783e-05,
      "loss": 0.6741,
      "step": 10563
    },
    {
      "epoch": 3.335648879065362,
      "grad_norm": 3.4333503246307373,
      "learning_rate": 1.8493439056908286e-05,
      "loss": 0.6862,
      "step": 10564
    },
    {
      "epoch": 3.335964635301547,
      "grad_norm": 2.8937947750091553,
      "learning_rate": 1.8489930531190793e-05,
      "loss": 0.5314,
      "step": 10565
    },
    {
      "epoch": 3.336280391537733,
      "grad_norm": 3.0555500984191895,
      "learning_rate": 1.84864220054733e-05,
      "loss": 0.5709,
      "step": 10566
    },
    {
      "epoch": 3.3365961477739186,
      "grad_norm": 3.7016189098358154,
      "learning_rate": 1.8482913479755807e-05,
      "loss": 0.6557,
      "step": 10567
    },
    {
      "epoch": 3.3369119040101043,
      "grad_norm": 3.2567927837371826,
      "learning_rate": 1.8479404954038317e-05,
      "loss": 0.6322,
      "step": 10568
    },
    {
      "epoch": 3.33722766024629,
      "grad_norm": 3.183722734451294,
      "learning_rate": 1.847589642832082e-05,
      "loss": 0.6711,
      "step": 10569
    },
    {
      "epoch": 3.3375434164824753,
      "grad_norm": 3.2397661209106445,
      "learning_rate": 1.8472387902603327e-05,
      "loss": 0.4981,
      "step": 10570
    },
    {
      "epoch": 3.337859172718661,
      "grad_norm": 3.7254693508148193,
      "learning_rate": 1.8468879376885834e-05,
      "loss": 0.7104,
      "step": 10571
    },
    {
      "epoch": 3.338174928954847,
      "grad_norm": 3.466738224029541,
      "learning_rate": 1.846537085116834e-05,
      "loss": 0.5996,
      "step": 10572
    },
    {
      "epoch": 3.3384906851910325,
      "grad_norm": 4.625598430633545,
      "learning_rate": 1.8461862325450848e-05,
      "loss": 0.6438,
      "step": 10573
    },
    {
      "epoch": 3.3388064414272183,
      "grad_norm": 3.357836961746216,
      "learning_rate": 1.845835379973335e-05,
      "loss": 0.5414,
      "step": 10574
    },
    {
      "epoch": 3.339122197663404,
      "grad_norm": 3.7687764167785645,
      "learning_rate": 1.8454845274015858e-05,
      "loss": 0.6959,
      "step": 10575
    },
    {
      "epoch": 3.3394379538995898,
      "grad_norm": 3.620854377746582,
      "learning_rate": 1.8451336748298365e-05,
      "loss": 0.6232,
      "step": 10576
    },
    {
      "epoch": 3.339753710135775,
      "grad_norm": 3.370568037033081,
      "learning_rate": 1.8447828222580875e-05,
      "loss": 0.6626,
      "step": 10577
    },
    {
      "epoch": 3.340069466371961,
      "grad_norm": 2.7180442810058594,
      "learning_rate": 1.8444319696863378e-05,
      "loss": 0.6324,
      "step": 10578
    },
    {
      "epoch": 3.3403852226081465,
      "grad_norm": 4.534849643707275,
      "learning_rate": 1.8440811171145885e-05,
      "loss": 0.6797,
      "step": 10579
    },
    {
      "epoch": 3.3407009788443323,
      "grad_norm": 3.4079482555389404,
      "learning_rate": 1.8437302645428392e-05,
      "loss": 0.7051,
      "step": 10580
    },
    {
      "epoch": 3.341016735080518,
      "grad_norm": 3.1311533451080322,
      "learning_rate": 1.84337941197109e-05,
      "loss": 0.686,
      "step": 10581
    },
    {
      "epoch": 3.3413324913167033,
      "grad_norm": 3.5810749530792236,
      "learning_rate": 1.8430285593993405e-05,
      "loss": 0.6874,
      "step": 10582
    },
    {
      "epoch": 3.341648247552889,
      "grad_norm": 3.600327968597412,
      "learning_rate": 1.842677706827591e-05,
      "loss": 0.5925,
      "step": 10583
    },
    {
      "epoch": 3.3419640037890748,
      "grad_norm": 3.294391632080078,
      "learning_rate": 1.8423268542558416e-05,
      "loss": 0.7174,
      "step": 10584
    },
    {
      "epoch": 3.3422797600252605,
      "grad_norm": 3.9074037075042725,
      "learning_rate": 1.8419760016840926e-05,
      "loss": 0.6353,
      "step": 10585
    },
    {
      "epoch": 3.3425955162614462,
      "grad_norm": 2.9679579734802246,
      "learning_rate": 1.8416251491123433e-05,
      "loss": 0.6273,
      "step": 10586
    },
    {
      "epoch": 3.342911272497632,
      "grad_norm": 3.4267566204071045,
      "learning_rate": 1.841274296540594e-05,
      "loss": 0.7468,
      "step": 10587
    },
    {
      "epoch": 3.3432270287338177,
      "grad_norm": 4.927379608154297,
      "learning_rate": 1.8409234439688443e-05,
      "loss": 0.6922,
      "step": 10588
    },
    {
      "epoch": 3.343542784970003,
      "grad_norm": 3.1100659370422363,
      "learning_rate": 1.840572591397095e-05,
      "loss": 0.7261,
      "step": 10589
    },
    {
      "epoch": 3.3438585412061888,
      "grad_norm": 3.59948468208313,
      "learning_rate": 1.8402217388253457e-05,
      "loss": 0.6736,
      "step": 10590
    },
    {
      "epoch": 3.3441742974423745,
      "grad_norm": 4.565696716308594,
      "learning_rate": 1.8398708862535963e-05,
      "loss": 0.6985,
      "step": 10591
    },
    {
      "epoch": 3.3444900536785602,
      "grad_norm": 3.1507890224456787,
      "learning_rate": 1.839520033681847e-05,
      "loss": 0.6247,
      "step": 10592
    },
    {
      "epoch": 3.344805809914746,
      "grad_norm": 3.60962176322937,
      "learning_rate": 1.8391691811100974e-05,
      "loss": 0.6601,
      "step": 10593
    },
    {
      "epoch": 3.3451215661509313,
      "grad_norm": 3.0187618732452393,
      "learning_rate": 1.8388183285383484e-05,
      "loss": 0.6679,
      "step": 10594
    },
    {
      "epoch": 3.345437322387117,
      "grad_norm": 3.2355258464813232,
      "learning_rate": 1.838467475966599e-05,
      "loss": 0.544,
      "step": 10595
    },
    {
      "epoch": 3.3457530786233027,
      "grad_norm": 3.7711689472198486,
      "learning_rate": 1.8381166233948497e-05,
      "loss": 0.7568,
      "step": 10596
    },
    {
      "epoch": 3.3460688348594885,
      "grad_norm": 3.898469924926758,
      "learning_rate": 1.8377657708231e-05,
      "loss": 0.6837,
      "step": 10597
    },
    {
      "epoch": 3.346384591095674,
      "grad_norm": 3.2430787086486816,
      "learning_rate": 1.8374149182513508e-05,
      "loss": 0.6939,
      "step": 10598
    },
    {
      "epoch": 3.34670034733186,
      "grad_norm": 2.9941813945770264,
      "learning_rate": 1.8370640656796014e-05,
      "loss": 0.5687,
      "step": 10599
    },
    {
      "epoch": 3.3470161035680457,
      "grad_norm": 2.808870792388916,
      "learning_rate": 1.836713213107852e-05,
      "loss": 0.5754,
      "step": 10600
    },
    {
      "epoch": 3.3470161035680457,
      "eval_loss": 0.8120313286781311,
      "eval_runtime": 4.6105,
      "eval_samples_per_second": 540.07,
      "eval_steps_per_second": 4.555,
      "step": 10600
    },
    {
      "epoch": 3.347331859804231,
      "grad_norm": 3.236354351043701,
      "learning_rate": 1.8363623605361028e-05,
      "loss": 0.6789,
      "step": 10601
    },
    {
      "epoch": 3.3476476160404167,
      "grad_norm": 2.7459585666656494,
      "learning_rate": 1.8360115079643535e-05,
      "loss": 0.5766,
      "step": 10602
    },
    {
      "epoch": 3.3479633722766025,
      "grad_norm": 3.369145393371582,
      "learning_rate": 1.8356606553926042e-05,
      "loss": 0.6037,
      "step": 10603
    },
    {
      "epoch": 3.348279128512788,
      "grad_norm": 4.010962009429932,
      "learning_rate": 1.835309802820855e-05,
      "loss": 0.7119,
      "step": 10604
    },
    {
      "epoch": 3.348594884748974,
      "grad_norm": 3.5845096111297607,
      "learning_rate": 1.8349589502491055e-05,
      "loss": 0.6118,
      "step": 10605
    },
    {
      "epoch": 3.3489106409851592,
      "grad_norm": 3.1135671138763428,
      "learning_rate": 1.8346080976773562e-05,
      "loss": 0.571,
      "step": 10606
    },
    {
      "epoch": 3.349226397221345,
      "grad_norm": 3.4090912342071533,
      "learning_rate": 1.8342572451056066e-05,
      "loss": 0.6764,
      "step": 10607
    },
    {
      "epoch": 3.3495421534575307,
      "grad_norm": 2.9554059505462646,
      "learning_rate": 1.8339063925338572e-05,
      "loss": 0.6796,
      "step": 10608
    },
    {
      "epoch": 3.3498579096937164,
      "grad_norm": 2.7910284996032715,
      "learning_rate": 1.833555539962108e-05,
      "loss": 0.6238,
      "step": 10609
    },
    {
      "epoch": 3.350173665929902,
      "grad_norm": 3.5748744010925293,
      "learning_rate": 1.8332046873903586e-05,
      "loss": 0.6897,
      "step": 10610
    },
    {
      "epoch": 3.350489422166088,
      "grad_norm": 3.1803088188171387,
      "learning_rate": 1.8328538348186093e-05,
      "loss": 0.6439,
      "step": 10611
    },
    {
      "epoch": 3.3508051784022737,
      "grad_norm": 2.80715274810791,
      "learning_rate": 1.83250298224686e-05,
      "loss": 0.7044,
      "step": 10612
    },
    {
      "epoch": 3.351120934638459,
      "grad_norm": 3.760317087173462,
      "learning_rate": 1.8321521296751106e-05,
      "loss": 0.7316,
      "step": 10613
    },
    {
      "epoch": 3.3514366908746447,
      "grad_norm": 4.010289669036865,
      "learning_rate": 1.8318012771033613e-05,
      "loss": 0.7508,
      "step": 10614
    },
    {
      "epoch": 3.3517524471108304,
      "grad_norm": 3.356174945831299,
      "learning_rate": 1.831450424531612e-05,
      "loss": 0.6331,
      "step": 10615
    },
    {
      "epoch": 3.352068203347016,
      "grad_norm": 3.7092652320861816,
      "learning_rate": 1.8310995719598627e-05,
      "loss": 0.6756,
      "step": 10616
    },
    {
      "epoch": 3.352383959583202,
      "grad_norm": 3.1385579109191895,
      "learning_rate": 1.830748719388113e-05,
      "loss": 0.6166,
      "step": 10617
    },
    {
      "epoch": 3.352699715819387,
      "grad_norm": 3.269193649291992,
      "learning_rate": 1.8303978668163637e-05,
      "loss": 0.7341,
      "step": 10618
    },
    {
      "epoch": 3.353015472055573,
      "grad_norm": 3.3896889686584473,
      "learning_rate": 1.8300470142446144e-05,
      "loss": 0.7462,
      "step": 10619
    },
    {
      "epoch": 3.3533312282917587,
      "grad_norm": 3.0574686527252197,
      "learning_rate": 1.8296961616728654e-05,
      "loss": 0.5926,
      "step": 10620
    },
    {
      "epoch": 3.3536469845279444,
      "grad_norm": 2.5752811431884766,
      "learning_rate": 1.8293453091011158e-05,
      "loss": 0.6752,
      "step": 10621
    },
    {
      "epoch": 3.35396274076413,
      "grad_norm": 2.747885227203369,
      "learning_rate": 1.8289944565293664e-05,
      "loss": 0.5713,
      "step": 10622
    },
    {
      "epoch": 3.354278497000316,
      "grad_norm": 3.1086747646331787,
      "learning_rate": 1.828643603957617e-05,
      "loss": 0.6687,
      "step": 10623
    },
    {
      "epoch": 3.3545942532365016,
      "grad_norm": 3.336527109146118,
      "learning_rate": 1.8282927513858678e-05,
      "loss": 0.7324,
      "step": 10624
    },
    {
      "epoch": 3.354910009472687,
      "grad_norm": 3.038715362548828,
      "learning_rate": 1.8279418988141185e-05,
      "loss": 0.6091,
      "step": 10625
    },
    {
      "epoch": 3.3552257657088727,
      "grad_norm": 2.8947458267211914,
      "learning_rate": 1.8275910462423688e-05,
      "loss": 0.607,
      "step": 10626
    },
    {
      "epoch": 3.3555415219450584,
      "grad_norm": 3.4409611225128174,
      "learning_rate": 1.8272401936706195e-05,
      "loss": 0.601,
      "step": 10627
    },
    {
      "epoch": 3.355857278181244,
      "grad_norm": 3.24029803276062,
      "learning_rate": 1.8268893410988702e-05,
      "loss": 0.6339,
      "step": 10628
    },
    {
      "epoch": 3.35617303441743,
      "grad_norm": 3.080300807952881,
      "learning_rate": 1.8265384885271212e-05,
      "loss": 0.6279,
      "step": 10629
    },
    {
      "epoch": 3.356488790653615,
      "grad_norm": 3.0492422580718994,
      "learning_rate": 1.826187635955372e-05,
      "loss": 0.6662,
      "step": 10630
    },
    {
      "epoch": 3.356804546889801,
      "grad_norm": 2.917609214782715,
      "learning_rate": 1.8258367833836222e-05,
      "loss": 0.5983,
      "step": 10631
    },
    {
      "epoch": 3.3571203031259866,
      "grad_norm": 3.823194980621338,
      "learning_rate": 1.825485930811873e-05,
      "loss": 0.7206,
      "step": 10632
    },
    {
      "epoch": 3.3574360593621724,
      "grad_norm": 2.959097385406494,
      "learning_rate": 1.8251350782401236e-05,
      "loss": 0.7617,
      "step": 10633
    },
    {
      "epoch": 3.357751815598358,
      "grad_norm": 3.073594331741333,
      "learning_rate": 1.8247842256683743e-05,
      "loss": 0.5869,
      "step": 10634
    },
    {
      "epoch": 3.358067571834544,
      "grad_norm": 3.01485013961792,
      "learning_rate": 1.824433373096625e-05,
      "loss": 0.6328,
      "step": 10635
    },
    {
      "epoch": 3.3583833280707296,
      "grad_norm": 2.938382863998413,
      "learning_rate": 1.8240825205248753e-05,
      "loss": 0.6367,
      "step": 10636
    },
    {
      "epoch": 3.358699084306915,
      "grad_norm": 2.6466245651245117,
      "learning_rate": 1.823731667953126e-05,
      "loss": 0.5652,
      "step": 10637
    },
    {
      "epoch": 3.3590148405431006,
      "grad_norm": 2.909924268722534,
      "learning_rate": 1.823380815381377e-05,
      "loss": 0.7158,
      "step": 10638
    },
    {
      "epoch": 3.3593305967792864,
      "grad_norm": 2.97922682762146,
      "learning_rate": 1.8230299628096277e-05,
      "loss": 0.6256,
      "step": 10639
    },
    {
      "epoch": 3.359646353015472,
      "grad_norm": 2.6529951095581055,
      "learning_rate": 1.822679110237878e-05,
      "loss": 0.6179,
      "step": 10640
    },
    {
      "epoch": 3.359962109251658,
      "grad_norm": 4.090235233306885,
      "learning_rate": 1.8223282576661287e-05,
      "loss": 0.6704,
      "step": 10641
    },
    {
      "epoch": 3.3602778654878436,
      "grad_norm": 3.0606460571289062,
      "learning_rate": 1.8219774050943794e-05,
      "loss": 0.6229,
      "step": 10642
    },
    {
      "epoch": 3.360593621724029,
      "grad_norm": 3.0899739265441895,
      "learning_rate": 1.82162655252263e-05,
      "loss": 0.5619,
      "step": 10643
    },
    {
      "epoch": 3.3609093779602146,
      "grad_norm": 3.5982472896575928,
      "learning_rate": 1.8212756999508808e-05,
      "loss": 0.6838,
      "step": 10644
    },
    {
      "epoch": 3.3612251341964003,
      "grad_norm": 3.3526999950408936,
      "learning_rate": 1.8209248473791314e-05,
      "loss": 0.7244,
      "step": 10645
    },
    {
      "epoch": 3.361540890432586,
      "grad_norm": 3.353415012359619,
      "learning_rate": 1.8205739948073818e-05,
      "loss": 0.7318,
      "step": 10646
    },
    {
      "epoch": 3.361856646668772,
      "grad_norm": 3.100414991378784,
      "learning_rate": 1.8202231422356328e-05,
      "loss": 0.6195,
      "step": 10647
    },
    {
      "epoch": 3.3621724029049576,
      "grad_norm": 3.22916841506958,
      "learning_rate": 1.8198722896638835e-05,
      "loss": 0.6904,
      "step": 10648
    },
    {
      "epoch": 3.362488159141143,
      "grad_norm": 5.071379661560059,
      "learning_rate": 1.819521437092134e-05,
      "loss": 0.6121,
      "step": 10649
    },
    {
      "epoch": 3.3628039153773286,
      "grad_norm": 3.090400457382202,
      "learning_rate": 1.8191705845203845e-05,
      "loss": 0.5755,
      "step": 10650
    },
    {
      "epoch": 3.3631196716135143,
      "grad_norm": 3.186591863632202,
      "learning_rate": 1.8188197319486352e-05,
      "loss": 0.565,
      "step": 10651
    },
    {
      "epoch": 3.3634354278497,
      "grad_norm": 3.5708742141723633,
      "learning_rate": 1.818468879376886e-05,
      "loss": 0.6559,
      "step": 10652
    },
    {
      "epoch": 3.363751184085886,
      "grad_norm": 3.097081422805786,
      "learning_rate": 1.8181180268051365e-05,
      "loss": 0.5897,
      "step": 10653
    },
    {
      "epoch": 3.3640669403220715,
      "grad_norm": 3.0835213661193848,
      "learning_rate": 1.8177671742333872e-05,
      "loss": 0.6201,
      "step": 10654
    },
    {
      "epoch": 3.364382696558257,
      "grad_norm": 3.074803352355957,
      "learning_rate": 1.817416321661638e-05,
      "loss": 0.6624,
      "step": 10655
    },
    {
      "epoch": 3.3646984527944426,
      "grad_norm": 3.086721181869507,
      "learning_rate": 1.8170654690898886e-05,
      "loss": 0.6225,
      "step": 10656
    },
    {
      "epoch": 3.3650142090306283,
      "grad_norm": 3.5034584999084473,
      "learning_rate": 1.8167146165181393e-05,
      "loss": 0.7336,
      "step": 10657
    },
    {
      "epoch": 3.365329965266814,
      "grad_norm": 2.937913179397583,
      "learning_rate": 1.81636376394639e-05,
      "loss": 0.6028,
      "step": 10658
    },
    {
      "epoch": 3.365645721503,
      "grad_norm": 3.871843099594116,
      "learning_rate": 1.8160129113746406e-05,
      "loss": 0.6295,
      "step": 10659
    },
    {
      "epoch": 3.3659614777391855,
      "grad_norm": 3.6043362617492676,
      "learning_rate": 1.815662058802891e-05,
      "loss": 0.6953,
      "step": 10660
    },
    {
      "epoch": 3.366277233975371,
      "grad_norm": 3.74112868309021,
      "learning_rate": 1.8153112062311417e-05,
      "loss": 0.7146,
      "step": 10661
    },
    {
      "epoch": 3.3665929902115566,
      "grad_norm": 3.968684196472168,
      "learning_rate": 1.8149603536593923e-05,
      "loss": 0.7605,
      "step": 10662
    },
    {
      "epoch": 3.3669087464477423,
      "grad_norm": 3.994939088821411,
      "learning_rate": 1.814609501087643e-05,
      "loss": 0.6994,
      "step": 10663
    },
    {
      "epoch": 3.367224502683928,
      "grad_norm": 3.4885921478271484,
      "learning_rate": 1.8142586485158937e-05,
      "loss": 0.6017,
      "step": 10664
    },
    {
      "epoch": 3.3675402589201138,
      "grad_norm": 3.6306331157684326,
      "learning_rate": 1.8139077959441444e-05,
      "loss": 0.6969,
      "step": 10665
    },
    {
      "epoch": 3.3678560151562995,
      "grad_norm": 3.4935693740844727,
      "learning_rate": 1.813556943372395e-05,
      "loss": 0.667,
      "step": 10666
    },
    {
      "epoch": 3.3681717713924852,
      "grad_norm": 3.1619246006011963,
      "learning_rate": 1.8132060908006457e-05,
      "loss": 0.6185,
      "step": 10667
    },
    {
      "epoch": 3.3684875276286705,
      "grad_norm": 2.844815254211426,
      "learning_rate": 1.8128552382288964e-05,
      "loss": 0.6126,
      "step": 10668
    },
    {
      "epoch": 3.3688032838648563,
      "grad_norm": 2.7980782985687256,
      "learning_rate": 1.8125043856571468e-05,
      "loss": 0.6438,
      "step": 10669
    },
    {
      "epoch": 3.369119040101042,
      "grad_norm": 3.0531206130981445,
      "learning_rate": 1.8121535330853974e-05,
      "loss": 0.5975,
      "step": 10670
    },
    {
      "epoch": 3.3694347963372278,
      "grad_norm": 2.6712381839752197,
      "learning_rate": 1.811802680513648e-05,
      "loss": 0.6222,
      "step": 10671
    },
    {
      "epoch": 3.3697505525734135,
      "grad_norm": 3.978222370147705,
      "learning_rate": 1.8114518279418988e-05,
      "loss": 0.671,
      "step": 10672
    },
    {
      "epoch": 3.370066308809599,
      "grad_norm": 3.3003499507904053,
      "learning_rate": 1.81110097537015e-05,
      "loss": 0.6762,
      "step": 10673
    },
    {
      "epoch": 3.3703820650457845,
      "grad_norm": 3.506640911102295,
      "learning_rate": 1.8107501227984002e-05,
      "loss": 0.6059,
      "step": 10674
    },
    {
      "epoch": 3.3706978212819703,
      "grad_norm": 3.8411576747894287,
      "learning_rate": 1.810399270226651e-05,
      "loss": 0.724,
      "step": 10675
    },
    {
      "epoch": 3.371013577518156,
      "grad_norm": 3.2978789806365967,
      "learning_rate": 1.8100484176549015e-05,
      "loss": 0.6275,
      "step": 10676
    },
    {
      "epoch": 3.3713293337543417,
      "grad_norm": 3.6602792739868164,
      "learning_rate": 1.8096975650831522e-05,
      "loss": 0.7275,
      "step": 10677
    },
    {
      "epoch": 3.3716450899905275,
      "grad_norm": 3.3526735305786133,
      "learning_rate": 1.809346712511403e-05,
      "loss": 0.7073,
      "step": 10678
    },
    {
      "epoch": 3.371960846226713,
      "grad_norm": 4.032395362854004,
      "learning_rate": 1.8089958599396532e-05,
      "loss": 0.733,
      "step": 10679
    },
    {
      "epoch": 3.3722766024628985,
      "grad_norm": 3.775667905807495,
      "learning_rate": 1.808645007367904e-05,
      "loss": 0.6312,
      "step": 10680
    },
    {
      "epoch": 3.3725923586990842,
      "grad_norm": 3.751389265060425,
      "learning_rate": 1.8082941547961546e-05,
      "loss": 0.699,
      "step": 10681
    },
    {
      "epoch": 3.37290811493527,
      "grad_norm": 3.686702251434326,
      "learning_rate": 1.8079433022244056e-05,
      "loss": 0.6765,
      "step": 10682
    },
    {
      "epoch": 3.3732238711714557,
      "grad_norm": 3.7738454341888428,
      "learning_rate": 1.807592449652656e-05,
      "loss": 0.7433,
      "step": 10683
    },
    {
      "epoch": 3.3735396274076415,
      "grad_norm": 3.2013370990753174,
      "learning_rate": 1.8072415970809066e-05,
      "loss": 0.5235,
      "step": 10684
    },
    {
      "epoch": 3.3738553836438268,
      "grad_norm": 2.9199931621551514,
      "learning_rate": 1.8068907445091573e-05,
      "loss": 0.6585,
      "step": 10685
    },
    {
      "epoch": 3.3741711398800125,
      "grad_norm": 3.882725954055786,
      "learning_rate": 1.806539891937408e-05,
      "loss": 0.7535,
      "step": 10686
    },
    {
      "epoch": 3.3744868961161982,
      "grad_norm": 3.1705634593963623,
      "learning_rate": 1.8061890393656587e-05,
      "loss": 0.6115,
      "step": 10687
    },
    {
      "epoch": 3.374802652352384,
      "grad_norm": 2.9661715030670166,
      "learning_rate": 1.8058381867939094e-05,
      "loss": 0.6637,
      "step": 10688
    },
    {
      "epoch": 3.3751184085885697,
      "grad_norm": 3.1333181858062744,
      "learning_rate": 1.8054873342221597e-05,
      "loss": 0.7042,
      "step": 10689
    },
    {
      "epoch": 3.3754341648247554,
      "grad_norm": 3.7308757305145264,
      "learning_rate": 1.8051364816504107e-05,
      "loss": 0.7118,
      "step": 10690
    },
    {
      "epoch": 3.375749921060941,
      "grad_norm": 3.588838815689087,
      "learning_rate": 1.8047856290786614e-05,
      "loss": 0.6279,
      "step": 10691
    },
    {
      "epoch": 3.3760656772971265,
      "grad_norm": 3.3485679626464844,
      "learning_rate": 1.804434776506912e-05,
      "loss": 0.5879,
      "step": 10692
    },
    {
      "epoch": 3.376381433533312,
      "grad_norm": 2.968519449234009,
      "learning_rate": 1.8040839239351624e-05,
      "loss": 0.6759,
      "step": 10693
    },
    {
      "epoch": 3.376697189769498,
      "grad_norm": 3.2138428688049316,
      "learning_rate": 1.803733071363413e-05,
      "loss": 0.7007,
      "step": 10694
    },
    {
      "epoch": 3.3770129460056837,
      "grad_norm": 3.6014163494110107,
      "learning_rate": 1.8033822187916638e-05,
      "loss": 0.6437,
      "step": 10695
    },
    {
      "epoch": 3.3773287022418694,
      "grad_norm": 3.1846532821655273,
      "learning_rate": 1.8030313662199145e-05,
      "loss": 0.6458,
      "step": 10696
    },
    {
      "epoch": 3.3776444584780547,
      "grad_norm": 3.2172892093658447,
      "learning_rate": 1.802680513648165e-05,
      "loss": 0.7257,
      "step": 10697
    },
    {
      "epoch": 3.3779602147142405,
      "grad_norm": 3.6480424404144287,
      "learning_rate": 1.8023296610764155e-05,
      "loss": 0.616,
      "step": 10698
    },
    {
      "epoch": 3.378275970950426,
      "grad_norm": 3.325416326522827,
      "learning_rate": 1.8019788085046665e-05,
      "loss": 0.642,
      "step": 10699
    },
    {
      "epoch": 3.378591727186612,
      "grad_norm": 2.9328830242156982,
      "learning_rate": 1.8016279559329172e-05,
      "loss": 0.605,
      "step": 10700
    },
    {
      "epoch": 3.3789074834227977,
      "grad_norm": 3.0764400959014893,
      "learning_rate": 1.801277103361168e-05,
      "loss": 0.6593,
      "step": 10701
    },
    {
      "epoch": 3.3792232396589834,
      "grad_norm": 3.244907855987549,
      "learning_rate": 1.8009262507894182e-05,
      "loss": 0.6959,
      "step": 10702
    },
    {
      "epoch": 3.379538995895169,
      "grad_norm": 3.670992851257324,
      "learning_rate": 1.800575398217669e-05,
      "loss": 0.6445,
      "step": 10703
    },
    {
      "epoch": 3.3798547521313544,
      "grad_norm": 3.129425525665283,
      "learning_rate": 1.8002245456459196e-05,
      "loss": 0.5777,
      "step": 10704
    },
    {
      "epoch": 3.38017050836754,
      "grad_norm": 3.6120553016662598,
      "learning_rate": 1.7998736930741703e-05,
      "loss": 0.6826,
      "step": 10705
    },
    {
      "epoch": 3.380486264603726,
      "grad_norm": 2.5850892066955566,
      "learning_rate": 1.799522840502421e-05,
      "loss": 0.6044,
      "step": 10706
    },
    {
      "epoch": 3.3808020208399117,
      "grad_norm": 3.8446147441864014,
      "learning_rate": 1.7991719879306716e-05,
      "loss": 0.7249,
      "step": 10707
    },
    {
      "epoch": 3.3811177770760974,
      "grad_norm": 2.982072114944458,
      "learning_rate": 1.7988211353589223e-05,
      "loss": 0.7348,
      "step": 10708
    },
    {
      "epoch": 3.3814335333122827,
      "grad_norm": 3.042503833770752,
      "learning_rate": 1.798470282787173e-05,
      "loss": 0.6333,
      "step": 10709
    },
    {
      "epoch": 3.3817492895484684,
      "grad_norm": 2.9468040466308594,
      "learning_rate": 1.7981194302154237e-05,
      "loss": 0.6114,
      "step": 10710
    },
    {
      "epoch": 3.382065045784654,
      "grad_norm": 2.7745471000671387,
      "learning_rate": 1.7977685776436744e-05,
      "loss": 0.6544,
      "step": 10711
    },
    {
      "epoch": 3.38238080202084,
      "grad_norm": 3.376485824584961,
      "learning_rate": 1.7974177250719247e-05,
      "loss": 0.6307,
      "step": 10712
    },
    {
      "epoch": 3.3826965582570256,
      "grad_norm": 3.0492353439331055,
      "learning_rate": 1.7970668725001754e-05,
      "loss": 0.6298,
      "step": 10713
    },
    {
      "epoch": 3.3830123144932114,
      "grad_norm": 2.8492531776428223,
      "learning_rate": 1.796716019928426e-05,
      "loss": 0.6056,
      "step": 10714
    },
    {
      "epoch": 3.383328070729397,
      "grad_norm": 3.3377974033355713,
      "learning_rate": 1.7963651673566768e-05,
      "loss": 0.727,
      "step": 10715
    },
    {
      "epoch": 3.3836438269655824,
      "grad_norm": 3.7363641262054443,
      "learning_rate": 1.7960143147849274e-05,
      "loss": 0.7464,
      "step": 10716
    },
    {
      "epoch": 3.383959583201768,
      "grad_norm": 2.639075517654419,
      "learning_rate": 1.795663462213178e-05,
      "loss": 0.6552,
      "step": 10717
    },
    {
      "epoch": 3.384275339437954,
      "grad_norm": 3.0926876068115234,
      "learning_rate": 1.7953126096414288e-05,
      "loss": 0.6314,
      "step": 10718
    },
    {
      "epoch": 3.3845910956741396,
      "grad_norm": 3.1249470710754395,
      "learning_rate": 1.7949617570696795e-05,
      "loss": 0.6783,
      "step": 10719
    },
    {
      "epoch": 3.3849068519103254,
      "grad_norm": 3.3502676486968994,
      "learning_rate": 1.79461090449793e-05,
      "loss": 0.6571,
      "step": 10720
    },
    {
      "epoch": 3.3852226081465107,
      "grad_norm": 3.690816879272461,
      "learning_rate": 1.794260051926181e-05,
      "loss": 0.6949,
      "step": 10721
    },
    {
      "epoch": 3.3855383643826964,
      "grad_norm": 3.1850109100341797,
      "learning_rate": 1.7939091993544312e-05,
      "loss": 0.6306,
      "step": 10722
    },
    {
      "epoch": 3.385854120618882,
      "grad_norm": 3.5533175468444824,
      "learning_rate": 1.793558346782682e-05,
      "loss": 0.6615,
      "step": 10723
    },
    {
      "epoch": 3.386169876855068,
      "grad_norm": 3.681251287460327,
      "learning_rate": 1.7932074942109325e-05,
      "loss": 0.6592,
      "step": 10724
    },
    {
      "epoch": 3.3864856330912536,
      "grad_norm": 3.671785354614258,
      "learning_rate": 1.7928566416391836e-05,
      "loss": 0.7091,
      "step": 10725
    },
    {
      "epoch": 3.3868013893274393,
      "grad_norm": 3.2827696800231934,
      "learning_rate": 1.792505789067434e-05,
      "loss": 0.6593,
      "step": 10726
    },
    {
      "epoch": 3.387117145563625,
      "grad_norm": 3.5641562938690186,
      "learning_rate": 1.7921549364956846e-05,
      "loss": 0.6861,
      "step": 10727
    },
    {
      "epoch": 3.3874329017998104,
      "grad_norm": 2.8513741493225098,
      "learning_rate": 1.7918040839239353e-05,
      "loss": 0.6191,
      "step": 10728
    },
    {
      "epoch": 3.387748658035996,
      "grad_norm": 3.247087001800537,
      "learning_rate": 1.791453231352186e-05,
      "loss": 0.7386,
      "step": 10729
    },
    {
      "epoch": 3.388064414272182,
      "grad_norm": 3.280559539794922,
      "learning_rate": 1.7911023787804366e-05,
      "loss": 0.6971,
      "step": 10730
    },
    {
      "epoch": 3.3883801705083676,
      "grad_norm": 2.9775795936584473,
      "learning_rate": 1.790751526208687e-05,
      "loss": 0.6549,
      "step": 10731
    },
    {
      "epoch": 3.3886959267445533,
      "grad_norm": 3.4588518142700195,
      "learning_rate": 1.7904006736369377e-05,
      "loss": 0.5992,
      "step": 10732
    },
    {
      "epoch": 3.3890116829807386,
      "grad_norm": 3.397331714630127,
      "learning_rate": 1.7900498210651883e-05,
      "loss": 0.6758,
      "step": 10733
    },
    {
      "epoch": 3.3893274392169244,
      "grad_norm": 3.9390931129455566,
      "learning_rate": 1.7896989684934394e-05,
      "loss": 0.6999,
      "step": 10734
    },
    {
      "epoch": 3.38964319545311,
      "grad_norm": 3.4072444438934326,
      "learning_rate": 1.78934811592169e-05,
      "loss": 0.6573,
      "step": 10735
    },
    {
      "epoch": 3.389958951689296,
      "grad_norm": 3.6506896018981934,
      "learning_rate": 1.7889972633499404e-05,
      "loss": 0.5759,
      "step": 10736
    },
    {
      "epoch": 3.3902747079254816,
      "grad_norm": 3.860459089279175,
      "learning_rate": 1.788646410778191e-05,
      "loss": 0.6591,
      "step": 10737
    },
    {
      "epoch": 3.3905904641616673,
      "grad_norm": 3.3200266361236572,
      "learning_rate": 1.7882955582064417e-05,
      "loss": 0.6809,
      "step": 10738
    },
    {
      "epoch": 3.390906220397853,
      "grad_norm": 2.9921607971191406,
      "learning_rate": 1.7879447056346924e-05,
      "loss": 0.6524,
      "step": 10739
    },
    {
      "epoch": 3.3912219766340383,
      "grad_norm": 3.121206760406494,
      "learning_rate": 1.787593853062943e-05,
      "loss": 0.6473,
      "step": 10740
    },
    {
      "epoch": 3.391537732870224,
      "grad_norm": 3.356400489807129,
      "learning_rate": 1.7872430004911934e-05,
      "loss": 0.6952,
      "step": 10741
    },
    {
      "epoch": 3.39185348910641,
      "grad_norm": 2.917024850845337,
      "learning_rate": 1.786892147919444e-05,
      "loss": 0.6044,
      "step": 10742
    },
    {
      "epoch": 3.3921692453425956,
      "grad_norm": 3.2951536178588867,
      "learning_rate": 1.786541295347695e-05,
      "loss": 0.6985,
      "step": 10743
    },
    {
      "epoch": 3.3924850015787813,
      "grad_norm": 3.4492406845092773,
      "learning_rate": 1.786190442775946e-05,
      "loss": 0.6582,
      "step": 10744
    },
    {
      "epoch": 3.392800757814967,
      "grad_norm": 2.9136059284210205,
      "learning_rate": 1.7858395902041962e-05,
      "loss": 0.615,
      "step": 10745
    },
    {
      "epoch": 3.3931165140511523,
      "grad_norm": 3.226088047027588,
      "learning_rate": 1.785488737632447e-05,
      "loss": 0.6768,
      "step": 10746
    },
    {
      "epoch": 3.393432270287338,
      "grad_norm": 3.3350608348846436,
      "learning_rate": 1.7851378850606975e-05,
      "loss": 0.7149,
      "step": 10747
    },
    {
      "epoch": 3.393748026523524,
      "grad_norm": 2.814138412475586,
      "learning_rate": 1.7847870324889482e-05,
      "loss": 0.6521,
      "step": 10748
    },
    {
      "epoch": 3.3940637827597095,
      "grad_norm": 3.664487361907959,
      "learning_rate": 1.784436179917199e-05,
      "loss": 0.7333,
      "step": 10749
    },
    {
      "epoch": 3.3943795389958953,
      "grad_norm": 3.6578822135925293,
      "learning_rate": 1.7840853273454496e-05,
      "loss": 0.6291,
      "step": 10750
    },
    {
      "epoch": 3.394695295232081,
      "grad_norm": 2.7831668853759766,
      "learning_rate": 1.7837344747737003e-05,
      "loss": 0.6167,
      "step": 10751
    },
    {
      "epoch": 3.3950110514682663,
      "grad_norm": 3.339482307434082,
      "learning_rate": 1.783383622201951e-05,
      "loss": 0.6427,
      "step": 10752
    },
    {
      "epoch": 3.395326807704452,
      "grad_norm": 4.240697860717773,
      "learning_rate": 1.7830327696302016e-05,
      "loss": 0.6886,
      "step": 10753
    },
    {
      "epoch": 3.395642563940638,
      "grad_norm": 3.0731165409088135,
      "learning_rate": 1.7826819170584523e-05,
      "loss": 0.6584,
      "step": 10754
    },
    {
      "epoch": 3.3959583201768235,
      "grad_norm": 3.8139901161193848,
      "learning_rate": 1.7823310644867026e-05,
      "loss": 0.6852,
      "step": 10755
    },
    {
      "epoch": 3.3962740764130093,
      "grad_norm": 3.1256306171417236,
      "learning_rate": 1.7819802119149533e-05,
      "loss": 0.7026,
      "step": 10756
    },
    {
      "epoch": 3.396589832649195,
      "grad_norm": 3.235645055770874,
      "learning_rate": 1.781629359343204e-05,
      "loss": 0.6492,
      "step": 10757
    },
    {
      "epoch": 3.3969055888853803,
      "grad_norm": 2.9192538261413574,
      "learning_rate": 1.7812785067714547e-05,
      "loss": 0.5843,
      "step": 10758
    },
    {
      "epoch": 3.397221345121566,
      "grad_norm": 3.229281425476074,
      "learning_rate": 1.7809276541997054e-05,
      "loss": 0.6236,
      "step": 10759
    },
    {
      "epoch": 3.3975371013577518,
      "grad_norm": 3.187283992767334,
      "learning_rate": 1.780576801627956e-05,
      "loss": 0.7102,
      "step": 10760
    },
    {
      "epoch": 3.3978528575939375,
      "grad_norm": 2.9070396423339844,
      "learning_rate": 1.7802259490562067e-05,
      "loss": 0.6692,
      "step": 10761
    },
    {
      "epoch": 3.3981686138301233,
      "grad_norm": 2.5708577632904053,
      "learning_rate": 1.7798750964844574e-05,
      "loss": 0.6775,
      "step": 10762
    },
    {
      "epoch": 3.398484370066309,
      "grad_norm": 3.284026861190796,
      "learning_rate": 1.779524243912708e-05,
      "loss": 0.6889,
      "step": 10763
    },
    {
      "epoch": 3.3988001263024943,
      "grad_norm": 3.2338473796844482,
      "learning_rate": 1.7791733913409588e-05,
      "loss": 0.6608,
      "step": 10764
    },
    {
      "epoch": 3.39911588253868,
      "grad_norm": 3.3047173023223877,
      "learning_rate": 1.778822538769209e-05,
      "loss": 0.7256,
      "step": 10765
    },
    {
      "epoch": 3.3994316387748658,
      "grad_norm": 2.824486255645752,
      "learning_rate": 1.7784716861974598e-05,
      "loss": 0.6252,
      "step": 10766
    },
    {
      "epoch": 3.3997473950110515,
      "grad_norm": 2.9810402393341064,
      "learning_rate": 1.7781208336257105e-05,
      "loss": 0.676,
      "step": 10767
    },
    {
      "epoch": 3.4000631512472372,
      "grad_norm": 3.567582845687866,
      "learning_rate": 1.777769981053961e-05,
      "loss": 0.7046,
      "step": 10768
    },
    {
      "epoch": 3.400378907483423,
      "grad_norm": 3.1992311477661133,
      "learning_rate": 1.777419128482212e-05,
      "loss": 0.6197,
      "step": 10769
    },
    {
      "epoch": 3.4006946637196083,
      "grad_norm": 2.945446491241455,
      "learning_rate": 1.7770682759104625e-05,
      "loss": 0.6383,
      "step": 10770
    },
    {
      "epoch": 3.401010419955794,
      "grad_norm": 2.801640748977661,
      "learning_rate": 1.7767174233387132e-05,
      "loss": 0.5255,
      "step": 10771
    },
    {
      "epoch": 3.4013261761919797,
      "grad_norm": 2.8864660263061523,
      "learning_rate": 1.776366570766964e-05,
      "loss": 0.6663,
      "step": 10772
    },
    {
      "epoch": 3.4016419324281655,
      "grad_norm": 2.948115110397339,
      "learning_rate": 1.7760157181952146e-05,
      "loss": 0.5387,
      "step": 10773
    },
    {
      "epoch": 3.401957688664351,
      "grad_norm": 3.592280864715576,
      "learning_rate": 1.775664865623465e-05,
      "loss": 0.7109,
      "step": 10774
    },
    {
      "epoch": 3.402273444900537,
      "grad_norm": 3.046926498413086,
      "learning_rate": 1.7753140130517156e-05,
      "loss": 0.5967,
      "step": 10775
    },
    {
      "epoch": 3.4025892011367223,
      "grad_norm": 3.213756561279297,
      "learning_rate": 1.7749631604799663e-05,
      "loss": 0.6549,
      "step": 10776
    },
    {
      "epoch": 3.402904957372908,
      "grad_norm": 3.2844808101654053,
      "learning_rate": 1.774612307908217e-05,
      "loss": 0.5835,
      "step": 10777
    },
    {
      "epoch": 3.4032207136090937,
      "grad_norm": 2.976494073867798,
      "learning_rate": 1.774261455336468e-05,
      "loss": 0.6417,
      "step": 10778
    },
    {
      "epoch": 3.4035364698452795,
      "grad_norm": 3.541499376296997,
      "learning_rate": 1.7739106027647183e-05,
      "loss": 0.7534,
      "step": 10779
    },
    {
      "epoch": 3.403852226081465,
      "grad_norm": 4.145462989807129,
      "learning_rate": 1.773559750192969e-05,
      "loss": 0.6588,
      "step": 10780
    },
    {
      "epoch": 3.404167982317651,
      "grad_norm": 3.388291120529175,
      "learning_rate": 1.7732088976212197e-05,
      "loss": 0.7952,
      "step": 10781
    },
    {
      "epoch": 3.4044837385538367,
      "grad_norm": 3.3382887840270996,
      "learning_rate": 1.7728580450494704e-05,
      "loss": 0.6459,
      "step": 10782
    },
    {
      "epoch": 3.404799494790022,
      "grad_norm": 3.467722177505493,
      "learning_rate": 1.772507192477721e-05,
      "loss": 0.6268,
      "step": 10783
    },
    {
      "epoch": 3.4051152510262077,
      "grad_norm": 3.517265558242798,
      "learning_rate": 1.7721563399059714e-05,
      "loss": 0.6568,
      "step": 10784
    },
    {
      "epoch": 3.4054310072623935,
      "grad_norm": 3.4717493057250977,
      "learning_rate": 1.771805487334222e-05,
      "loss": 0.6866,
      "step": 10785
    },
    {
      "epoch": 3.405746763498579,
      "grad_norm": 3.996614694595337,
      "learning_rate": 1.771454634762473e-05,
      "loss": 0.7198,
      "step": 10786
    },
    {
      "epoch": 3.406062519734765,
      "grad_norm": 3.2908053398132324,
      "learning_rate": 1.7711037821907238e-05,
      "loss": 0.6239,
      "step": 10787
    },
    {
      "epoch": 3.40637827597095,
      "grad_norm": 3.398688554763794,
      "learning_rate": 1.770752929618974e-05,
      "loss": 0.6488,
      "step": 10788
    },
    {
      "epoch": 3.406694032207136,
      "grad_norm": 3.508096694946289,
      "learning_rate": 1.7704020770472248e-05,
      "loss": 0.6292,
      "step": 10789
    },
    {
      "epoch": 3.4070097884433217,
      "grad_norm": 3.73262357711792,
      "learning_rate": 1.7700512244754755e-05,
      "loss": 0.7632,
      "step": 10790
    },
    {
      "epoch": 3.4073255446795074,
      "grad_norm": 2.9160497188568115,
      "learning_rate": 1.769700371903726e-05,
      "loss": 0.6847,
      "step": 10791
    },
    {
      "epoch": 3.407641300915693,
      "grad_norm": 3.3678841590881348,
      "learning_rate": 1.769349519331977e-05,
      "loss": 0.7273,
      "step": 10792
    },
    {
      "epoch": 3.407957057151879,
      "grad_norm": 3.821854829788208,
      "learning_rate": 1.7689986667602275e-05,
      "loss": 0.6594,
      "step": 10793
    },
    {
      "epoch": 3.4082728133880646,
      "grad_norm": 4.071545124053955,
      "learning_rate": 1.768647814188478e-05,
      "loss": 0.6523,
      "step": 10794
    },
    {
      "epoch": 3.40858856962425,
      "grad_norm": 3.7879478931427,
      "learning_rate": 1.768296961616729e-05,
      "loss": 0.7292,
      "step": 10795
    },
    {
      "epoch": 3.4089043258604357,
      "grad_norm": 3.6782991886138916,
      "learning_rate": 1.7679461090449796e-05,
      "loss": 0.673,
      "step": 10796
    },
    {
      "epoch": 3.4092200820966214,
      "grad_norm": 3.127251386642456,
      "learning_rate": 1.7675952564732302e-05,
      "loss": 0.6022,
      "step": 10797
    },
    {
      "epoch": 3.409535838332807,
      "grad_norm": 3.759247064590454,
      "learning_rate": 1.7672444039014806e-05,
      "loss": 0.7009,
      "step": 10798
    },
    {
      "epoch": 3.409851594568993,
      "grad_norm": 3.693213701248169,
      "learning_rate": 1.7668935513297313e-05,
      "loss": 0.6815,
      "step": 10799
    },
    {
      "epoch": 3.410167350805178,
      "grad_norm": 3.4865880012512207,
      "learning_rate": 1.766542698757982e-05,
      "loss": 0.7085,
      "step": 10800
    },
    {
      "epoch": 3.410167350805178,
      "eval_loss": 0.8121626377105713,
      "eval_runtime": 4.59,
      "eval_samples_per_second": 542.483,
      "eval_steps_per_second": 4.575,
      "step": 10800
    },
    {
      "epoch": 3.410483107041364,
      "grad_norm": 3.1485509872436523,
      "learning_rate": 1.7661918461862326e-05,
      "loss": 0.6659,
      "step": 10801
    },
    {
      "epoch": 3.4107988632775497,
      "grad_norm": 3.7710540294647217,
      "learning_rate": 1.7658409936144833e-05,
      "loss": 0.6626,
      "step": 10802
    },
    {
      "epoch": 3.4111146195137354,
      "grad_norm": 5.843705177307129,
      "learning_rate": 1.7654901410427337e-05,
      "loss": 0.6569,
      "step": 10803
    },
    {
      "epoch": 3.411430375749921,
      "grad_norm": 3.075765609741211,
      "learning_rate": 1.7651392884709847e-05,
      "loss": 0.5757,
      "step": 10804
    },
    {
      "epoch": 3.411746131986107,
      "grad_norm": 3.3466596603393555,
      "learning_rate": 1.7647884358992354e-05,
      "loss": 0.7251,
      "step": 10805
    },
    {
      "epoch": 3.4120618882222926,
      "grad_norm": 2.919421672821045,
      "learning_rate": 1.764437583327486e-05,
      "loss": 0.6241,
      "step": 10806
    },
    {
      "epoch": 3.412377644458478,
      "grad_norm": 3.1073153018951416,
      "learning_rate": 1.7640867307557364e-05,
      "loss": 0.6167,
      "step": 10807
    },
    {
      "epoch": 3.4126934006946636,
      "grad_norm": 3.7895262241363525,
      "learning_rate": 1.763735878183987e-05,
      "loss": 0.7285,
      "step": 10808
    },
    {
      "epoch": 3.4130091569308494,
      "grad_norm": 3.5394608974456787,
      "learning_rate": 1.7633850256122377e-05,
      "loss": 0.6442,
      "step": 10809
    },
    {
      "epoch": 3.413324913167035,
      "grad_norm": 3.0816688537597656,
      "learning_rate": 1.7630341730404884e-05,
      "loss": 0.5965,
      "step": 10810
    },
    {
      "epoch": 3.413640669403221,
      "grad_norm": 3.523531436920166,
      "learning_rate": 1.762683320468739e-05,
      "loss": 0.7616,
      "step": 10811
    },
    {
      "epoch": 3.413956425639406,
      "grad_norm": 3.511901617050171,
      "learning_rate": 1.7623324678969898e-05,
      "loss": 0.8199,
      "step": 10812
    },
    {
      "epoch": 3.414272181875592,
      "grad_norm": 3.1779885292053223,
      "learning_rate": 1.7619816153252405e-05,
      "loss": 0.663,
      "step": 10813
    },
    {
      "epoch": 3.4145879381117776,
      "grad_norm": 2.943019390106201,
      "learning_rate": 1.761630762753491e-05,
      "loss": 0.6069,
      "step": 10814
    },
    {
      "epoch": 3.4149036943479634,
      "grad_norm": 3.2894351482391357,
      "learning_rate": 1.761279910181742e-05,
      "loss": 0.6294,
      "step": 10815
    },
    {
      "epoch": 3.415219450584149,
      "grad_norm": 3.229998826980591,
      "learning_rate": 1.7609290576099925e-05,
      "loss": 0.6657,
      "step": 10816
    },
    {
      "epoch": 3.415535206820335,
      "grad_norm": 3.46951961517334,
      "learning_rate": 1.760578205038243e-05,
      "loss": 0.7263,
      "step": 10817
    },
    {
      "epoch": 3.4158509630565206,
      "grad_norm": 3.3153128623962402,
      "learning_rate": 1.7602273524664935e-05,
      "loss": 0.6925,
      "step": 10818
    },
    {
      "epoch": 3.416166719292706,
      "grad_norm": 4.158696174621582,
      "learning_rate": 1.7598764998947442e-05,
      "loss": 0.8557,
      "step": 10819
    },
    {
      "epoch": 3.4164824755288916,
      "grad_norm": 3.4720466136932373,
      "learning_rate": 1.759525647322995e-05,
      "loss": 0.6422,
      "step": 10820
    },
    {
      "epoch": 3.4167982317650774,
      "grad_norm": 3.2740042209625244,
      "learning_rate": 1.7591747947512456e-05,
      "loss": 0.6894,
      "step": 10821
    },
    {
      "epoch": 3.417113988001263,
      "grad_norm": 3.5485785007476807,
      "learning_rate": 1.7588239421794963e-05,
      "loss": 0.6307,
      "step": 10822
    },
    {
      "epoch": 3.417429744237449,
      "grad_norm": 2.938966751098633,
      "learning_rate": 1.758473089607747e-05,
      "loss": 0.5232,
      "step": 10823
    },
    {
      "epoch": 3.417745500473634,
      "grad_norm": 3.6971631050109863,
      "learning_rate": 1.7581222370359976e-05,
      "loss": 0.6527,
      "step": 10824
    },
    {
      "epoch": 3.41806125670982,
      "grad_norm": 3.3638226985931396,
      "learning_rate": 1.7577713844642483e-05,
      "loss": 0.6579,
      "step": 10825
    },
    {
      "epoch": 3.4183770129460056,
      "grad_norm": 4.095026016235352,
      "learning_rate": 1.757420531892499e-05,
      "loss": 0.8092,
      "step": 10826
    },
    {
      "epoch": 3.4186927691821913,
      "grad_norm": 3.050865411758423,
      "learning_rate": 1.7570696793207493e-05,
      "loss": 0.5591,
      "step": 10827
    },
    {
      "epoch": 3.419008525418377,
      "grad_norm": 3.3547751903533936,
      "learning_rate": 1.756718826749e-05,
      "loss": 0.6715,
      "step": 10828
    },
    {
      "epoch": 3.419324281654563,
      "grad_norm": 3.1174702644348145,
      "learning_rate": 1.7563679741772507e-05,
      "loss": 0.6696,
      "step": 10829
    },
    {
      "epoch": 3.4196400378907486,
      "grad_norm": 3.24348783493042,
      "learning_rate": 1.7560171216055017e-05,
      "loss": 0.7173,
      "step": 10830
    },
    {
      "epoch": 3.419955794126934,
      "grad_norm": 3.018301248550415,
      "learning_rate": 1.755666269033752e-05,
      "loss": 0.6567,
      "step": 10831
    },
    {
      "epoch": 3.4202715503631196,
      "grad_norm": 3.0455222129821777,
      "learning_rate": 1.7553154164620027e-05,
      "loss": 0.6948,
      "step": 10832
    },
    {
      "epoch": 3.4205873065993053,
      "grad_norm": 3.0714099407196045,
      "learning_rate": 1.7549645638902534e-05,
      "loss": 0.7585,
      "step": 10833
    },
    {
      "epoch": 3.420903062835491,
      "grad_norm": 3.0684778690338135,
      "learning_rate": 1.754613711318504e-05,
      "loss": 0.6433,
      "step": 10834
    },
    {
      "epoch": 3.421218819071677,
      "grad_norm": 2.973555088043213,
      "learning_rate": 1.7542628587467548e-05,
      "loss": 0.6604,
      "step": 10835
    },
    {
      "epoch": 3.421534575307862,
      "grad_norm": 3.533559560775757,
      "learning_rate": 1.753912006175005e-05,
      "loss": 0.7505,
      "step": 10836
    },
    {
      "epoch": 3.421850331544048,
      "grad_norm": 3.002312183380127,
      "learning_rate": 1.7535611536032558e-05,
      "loss": 0.7379,
      "step": 10837
    },
    {
      "epoch": 3.4221660877802336,
      "grad_norm": 3.668161392211914,
      "learning_rate": 1.7532103010315065e-05,
      "loss": 0.5865,
      "step": 10838
    },
    {
      "epoch": 3.4224818440164193,
      "grad_norm": 3.365370512008667,
      "learning_rate": 1.7528594484597575e-05,
      "loss": 0.7057,
      "step": 10839
    },
    {
      "epoch": 3.422797600252605,
      "grad_norm": 3.5447943210601807,
      "learning_rate": 1.7525085958880082e-05,
      "loss": 0.674,
      "step": 10840
    },
    {
      "epoch": 3.423113356488791,
      "grad_norm": 3.2397398948669434,
      "learning_rate": 1.7521577433162585e-05,
      "loss": 0.602,
      "step": 10841
    },
    {
      "epoch": 3.4234291127249765,
      "grad_norm": 3.186833381652832,
      "learning_rate": 1.7518068907445092e-05,
      "loss": 0.6377,
      "step": 10842
    },
    {
      "epoch": 3.423744868961162,
      "grad_norm": 2.982551336288452,
      "learning_rate": 1.75145603817276e-05,
      "loss": 0.7315,
      "step": 10843
    },
    {
      "epoch": 3.4240606251973476,
      "grad_norm": 3.3653440475463867,
      "learning_rate": 1.7511051856010106e-05,
      "loss": 0.6333,
      "step": 10844
    },
    {
      "epoch": 3.4243763814335333,
      "grad_norm": 2.7714812755584717,
      "learning_rate": 1.7507543330292613e-05,
      "loss": 0.5754,
      "step": 10845
    },
    {
      "epoch": 3.424692137669719,
      "grad_norm": 3.0397982597351074,
      "learning_rate": 1.7504034804575116e-05,
      "loss": 0.5882,
      "step": 10846
    },
    {
      "epoch": 3.4250078939059048,
      "grad_norm": 3.2751550674438477,
      "learning_rate": 1.7500526278857623e-05,
      "loss": 0.7333,
      "step": 10847
    },
    {
      "epoch": 3.42532365014209,
      "grad_norm": 3.7351927757263184,
      "learning_rate": 1.7497017753140133e-05,
      "loss": 0.5566,
      "step": 10848
    },
    {
      "epoch": 3.425639406378276,
      "grad_norm": 3.0762829780578613,
      "learning_rate": 1.749350922742264e-05,
      "loss": 0.6614,
      "step": 10849
    },
    {
      "epoch": 3.4259551626144615,
      "grad_norm": 3.8248417377471924,
      "learning_rate": 1.7490000701705143e-05,
      "loss": 0.705,
      "step": 10850
    },
    {
      "epoch": 3.4262709188506473,
      "grad_norm": 3.253849744796753,
      "learning_rate": 1.748649217598765e-05,
      "loss": 0.6326,
      "step": 10851
    },
    {
      "epoch": 3.426586675086833,
      "grad_norm": 3.2644143104553223,
      "learning_rate": 1.7482983650270157e-05,
      "loss": 0.7299,
      "step": 10852
    },
    {
      "epoch": 3.4269024313230187,
      "grad_norm": 3.824763298034668,
      "learning_rate": 1.7479475124552664e-05,
      "loss": 0.7086,
      "step": 10853
    },
    {
      "epoch": 3.4272181875592045,
      "grad_norm": 3.581543445587158,
      "learning_rate": 1.747596659883517e-05,
      "loss": 0.7106,
      "step": 10854
    },
    {
      "epoch": 3.42753394379539,
      "grad_norm": 2.8856091499328613,
      "learning_rate": 1.7472458073117677e-05,
      "loss": 0.6217,
      "step": 10855
    },
    {
      "epoch": 3.4278497000315755,
      "grad_norm": 3.2608327865600586,
      "learning_rate": 1.7468949547400184e-05,
      "loss": 0.6851,
      "step": 10856
    },
    {
      "epoch": 3.4281654562677613,
      "grad_norm": 3.1034798622131348,
      "learning_rate": 1.746544102168269e-05,
      "loss": 0.63,
      "step": 10857
    },
    {
      "epoch": 3.428481212503947,
      "grad_norm": 3.126819610595703,
      "learning_rate": 1.7461932495965198e-05,
      "loss": 0.6159,
      "step": 10858
    },
    {
      "epoch": 3.4287969687401327,
      "grad_norm": 2.9842090606689453,
      "learning_rate": 1.7458423970247705e-05,
      "loss": 0.6873,
      "step": 10859
    },
    {
      "epoch": 3.4291127249763185,
      "grad_norm": 2.9090023040771484,
      "learning_rate": 1.7454915444530208e-05,
      "loss": 0.6071,
      "step": 10860
    },
    {
      "epoch": 3.4294284812125038,
      "grad_norm": 3.2995362281799316,
      "learning_rate": 1.7451406918812715e-05,
      "loss": 0.6259,
      "step": 10861
    },
    {
      "epoch": 3.4297442374486895,
      "grad_norm": 4.111922740936279,
      "learning_rate": 1.744789839309522e-05,
      "loss": 0.6603,
      "step": 10862
    },
    {
      "epoch": 3.4300599936848752,
      "grad_norm": 3.424929141998291,
      "learning_rate": 1.744438986737773e-05,
      "loss": 0.5768,
      "step": 10863
    },
    {
      "epoch": 3.430375749921061,
      "grad_norm": 3.3669490814208984,
      "learning_rate": 1.7440881341660235e-05,
      "loss": 0.6547,
      "step": 10864
    },
    {
      "epoch": 3.4306915061572467,
      "grad_norm": 3.391850471496582,
      "learning_rate": 1.7437372815942742e-05,
      "loss": 0.707,
      "step": 10865
    },
    {
      "epoch": 3.4310072623934325,
      "grad_norm": 2.8255608081817627,
      "learning_rate": 1.743386429022525e-05,
      "loss": 0.6046,
      "step": 10866
    },
    {
      "epoch": 3.4313230186296177,
      "grad_norm": 3.735029458999634,
      "learning_rate": 1.7430355764507756e-05,
      "loss": 0.6019,
      "step": 10867
    },
    {
      "epoch": 3.4316387748658035,
      "grad_norm": 3.007288694381714,
      "learning_rate": 1.7426847238790262e-05,
      "loss": 0.6076,
      "step": 10868
    },
    {
      "epoch": 3.4319545311019892,
      "grad_norm": 3.4638891220092773,
      "learning_rate": 1.742333871307277e-05,
      "loss": 0.6596,
      "step": 10869
    },
    {
      "epoch": 3.432270287338175,
      "grad_norm": 3.0549755096435547,
      "learning_rate": 1.7419830187355273e-05,
      "loss": 0.5873,
      "step": 10870
    },
    {
      "epoch": 3.4325860435743607,
      "grad_norm": 3.4267117977142334,
      "learning_rate": 1.741632166163778e-05,
      "loss": 0.6256,
      "step": 10871
    },
    {
      "epoch": 3.4329017998105464,
      "grad_norm": 3.789893388748169,
      "learning_rate": 1.7412813135920286e-05,
      "loss": 0.5959,
      "step": 10872
    },
    {
      "epoch": 3.4332175560467317,
      "grad_norm": 3.2164926528930664,
      "learning_rate": 1.7409304610202793e-05,
      "loss": 0.6886,
      "step": 10873
    },
    {
      "epoch": 3.4335333122829175,
      "grad_norm": 3.2603466510772705,
      "learning_rate": 1.74057960844853e-05,
      "loss": 0.6834,
      "step": 10874
    },
    {
      "epoch": 3.433849068519103,
      "grad_norm": 3.469538927078247,
      "learning_rate": 1.7402287558767807e-05,
      "loss": 0.7126,
      "step": 10875
    },
    {
      "epoch": 3.434164824755289,
      "grad_norm": 3.534820556640625,
      "learning_rate": 1.7398779033050314e-05,
      "loss": 0.6466,
      "step": 10876
    },
    {
      "epoch": 3.4344805809914747,
      "grad_norm": 3.2057056427001953,
      "learning_rate": 1.739527050733282e-05,
      "loss": 0.6317,
      "step": 10877
    },
    {
      "epoch": 3.4347963372276604,
      "grad_norm": 3.9841268062591553,
      "learning_rate": 1.7391761981615327e-05,
      "loss": 0.7125,
      "step": 10878
    },
    {
      "epoch": 3.4351120934638457,
      "grad_norm": 3.381772041320801,
      "learning_rate": 1.738825345589783e-05,
      "loss": 0.6884,
      "step": 10879
    },
    {
      "epoch": 3.4354278497000315,
      "grad_norm": 3.742152214050293,
      "learning_rate": 1.7384744930180337e-05,
      "loss": 0.7005,
      "step": 10880
    },
    {
      "epoch": 3.435743605936217,
      "grad_norm": 2.9787404537200928,
      "learning_rate": 1.7381236404462844e-05,
      "loss": 0.6026,
      "step": 10881
    },
    {
      "epoch": 3.436059362172403,
      "grad_norm": 3.0405490398406982,
      "learning_rate": 1.737772787874535e-05,
      "loss": 0.6815,
      "step": 10882
    },
    {
      "epoch": 3.4363751184085887,
      "grad_norm": 3.641967535018921,
      "learning_rate": 1.737421935302786e-05,
      "loss": 0.6188,
      "step": 10883
    },
    {
      "epoch": 3.4366908746447744,
      "grad_norm": 3.2886672019958496,
      "learning_rate": 1.7370710827310365e-05,
      "loss": 0.6233,
      "step": 10884
    },
    {
      "epoch": 3.43700663088096,
      "grad_norm": 3.433485269546509,
      "learning_rate": 1.736720230159287e-05,
      "loss": 0.5515,
      "step": 10885
    },
    {
      "epoch": 3.4373223871171454,
      "grad_norm": 2.617820978164673,
      "learning_rate": 1.736369377587538e-05,
      "loss": 0.5995,
      "step": 10886
    },
    {
      "epoch": 3.437638143353331,
      "grad_norm": 3.728477716445923,
      "learning_rate": 1.7360185250157885e-05,
      "loss": 0.7021,
      "step": 10887
    },
    {
      "epoch": 3.437953899589517,
      "grad_norm": 3.791322708129883,
      "learning_rate": 1.7356676724440392e-05,
      "loss": 0.6287,
      "step": 10888
    },
    {
      "epoch": 3.4382696558257027,
      "grad_norm": 3.5113909244537354,
      "learning_rate": 1.7353168198722895e-05,
      "loss": 0.6493,
      "step": 10889
    },
    {
      "epoch": 3.4385854120618884,
      "grad_norm": 2.8088557720184326,
      "learning_rate": 1.7349659673005402e-05,
      "loss": 0.6923,
      "step": 10890
    },
    {
      "epoch": 3.4389011682980737,
      "grad_norm": 3.4434478282928467,
      "learning_rate": 1.7346151147287912e-05,
      "loss": 0.6861,
      "step": 10891
    },
    {
      "epoch": 3.4392169245342594,
      "grad_norm": 3.67815899848938,
      "learning_rate": 1.734264262157042e-05,
      "loss": 0.6972,
      "step": 10892
    },
    {
      "epoch": 3.439532680770445,
      "grad_norm": 4.111649513244629,
      "learning_rate": 1.7339134095852923e-05,
      "loss": 0.753,
      "step": 10893
    },
    {
      "epoch": 3.439848437006631,
      "grad_norm": 3.46172833442688,
      "learning_rate": 1.733562557013543e-05,
      "loss": 0.6589,
      "step": 10894
    },
    {
      "epoch": 3.4401641932428166,
      "grad_norm": 3.022158622741699,
      "learning_rate": 1.7332117044417936e-05,
      "loss": 0.6513,
      "step": 10895
    },
    {
      "epoch": 3.4404799494790024,
      "grad_norm": 2.9805054664611816,
      "learning_rate": 1.7328608518700443e-05,
      "loss": 0.6252,
      "step": 10896
    },
    {
      "epoch": 3.440795705715188,
      "grad_norm": 3.483023166656494,
      "learning_rate": 1.732509999298295e-05,
      "loss": 0.6746,
      "step": 10897
    },
    {
      "epoch": 3.4411114619513734,
      "grad_norm": 2.7258355617523193,
      "learning_rate": 1.7321591467265457e-05,
      "loss": 0.5845,
      "step": 10898
    },
    {
      "epoch": 3.441427218187559,
      "grad_norm": 3.9557723999023438,
      "learning_rate": 1.731808294154796e-05,
      "loss": 0.7756,
      "step": 10899
    },
    {
      "epoch": 3.441742974423745,
      "grad_norm": 3.1786415576934814,
      "learning_rate": 1.731457441583047e-05,
      "loss": 0.6268,
      "step": 10900
    },
    {
      "epoch": 3.4420587306599306,
      "grad_norm": 4.371512413024902,
      "learning_rate": 1.7311065890112977e-05,
      "loss": 0.6298,
      "step": 10901
    },
    {
      "epoch": 3.4423744868961164,
      "grad_norm": 3.5524404048919678,
      "learning_rate": 1.7307557364395484e-05,
      "loss": 0.746,
      "step": 10902
    },
    {
      "epoch": 3.4426902431323017,
      "grad_norm": 3.618720531463623,
      "learning_rate": 1.7304048838677987e-05,
      "loss": 0.6818,
      "step": 10903
    },
    {
      "epoch": 3.4430059993684874,
      "grad_norm": 3.9297690391540527,
      "learning_rate": 1.7300540312960494e-05,
      "loss": 0.6693,
      "step": 10904
    },
    {
      "epoch": 3.443321755604673,
      "grad_norm": 3.6650710105895996,
      "learning_rate": 1.7297031787243e-05,
      "loss": 0.6956,
      "step": 10905
    },
    {
      "epoch": 3.443637511840859,
      "grad_norm": 3.4237232208251953,
      "learning_rate": 1.7293523261525508e-05,
      "loss": 0.7059,
      "step": 10906
    },
    {
      "epoch": 3.4439532680770446,
      "grad_norm": 3.083308696746826,
      "learning_rate": 1.7290014735808015e-05,
      "loss": 0.5524,
      "step": 10907
    },
    {
      "epoch": 3.4442690243132303,
      "grad_norm": 3.711758613586426,
      "learning_rate": 1.7286506210090518e-05,
      "loss": 0.6907,
      "step": 10908
    },
    {
      "epoch": 3.444584780549416,
      "grad_norm": 3.3640170097351074,
      "learning_rate": 1.7282997684373028e-05,
      "loss": 0.7236,
      "step": 10909
    },
    {
      "epoch": 3.4449005367856014,
      "grad_norm": 4.182497024536133,
      "learning_rate": 1.7279489158655535e-05,
      "loss": 0.7128,
      "step": 10910
    },
    {
      "epoch": 3.445216293021787,
      "grad_norm": 3.623549699783325,
      "learning_rate": 1.7275980632938042e-05,
      "loss": 0.7324,
      "step": 10911
    },
    {
      "epoch": 3.445532049257973,
      "grad_norm": 2.7592551708221436,
      "learning_rate": 1.7272472107220545e-05,
      "loss": 0.7093,
      "step": 10912
    },
    {
      "epoch": 3.4458478054941586,
      "grad_norm": 2.9713146686553955,
      "learning_rate": 1.7268963581503052e-05,
      "loss": 0.6324,
      "step": 10913
    },
    {
      "epoch": 3.4461635617303443,
      "grad_norm": 3.058828592300415,
      "learning_rate": 1.726545505578556e-05,
      "loss": 0.6483,
      "step": 10914
    },
    {
      "epoch": 3.4464793179665296,
      "grad_norm": 3.4931602478027344,
      "learning_rate": 1.7261946530068066e-05,
      "loss": 0.675,
      "step": 10915
    },
    {
      "epoch": 3.4467950742027154,
      "grad_norm": 3.187948703765869,
      "learning_rate": 1.7258438004350573e-05,
      "loss": 0.6741,
      "step": 10916
    },
    {
      "epoch": 3.447110830438901,
      "grad_norm": 3.168229103088379,
      "learning_rate": 1.725492947863308e-05,
      "loss": 0.6518,
      "step": 10917
    },
    {
      "epoch": 3.447426586675087,
      "grad_norm": 3.3023176193237305,
      "learning_rate": 1.7251420952915586e-05,
      "loss": 0.6513,
      "step": 10918
    },
    {
      "epoch": 3.4477423429112726,
      "grad_norm": 3.5291495323181152,
      "learning_rate": 1.7247912427198093e-05,
      "loss": 0.6532,
      "step": 10919
    },
    {
      "epoch": 3.4480580991474583,
      "grad_norm": 3.4008357524871826,
      "learning_rate": 1.72444039014806e-05,
      "loss": 0.623,
      "step": 10920
    },
    {
      "epoch": 3.448373855383644,
      "grad_norm": 3.997269630432129,
      "learning_rate": 1.7240895375763107e-05,
      "loss": 0.6799,
      "step": 10921
    },
    {
      "epoch": 3.4486896116198293,
      "grad_norm": 3.8382673263549805,
      "learning_rate": 1.723738685004561e-05,
      "loss": 0.6503,
      "step": 10922
    },
    {
      "epoch": 3.449005367856015,
      "grad_norm": 2.8568952083587646,
      "learning_rate": 1.7233878324328117e-05,
      "loss": 0.5842,
      "step": 10923
    },
    {
      "epoch": 3.449321124092201,
      "grad_norm": 3.5136466026306152,
      "learning_rate": 1.7230369798610624e-05,
      "loss": 0.6914,
      "step": 10924
    },
    {
      "epoch": 3.4496368803283866,
      "grad_norm": 3.3356456756591797,
      "learning_rate": 1.722686127289313e-05,
      "loss": 0.7068,
      "step": 10925
    },
    {
      "epoch": 3.4499526365645723,
      "grad_norm": 3.5044589042663574,
      "learning_rate": 1.7223352747175637e-05,
      "loss": 0.667,
      "step": 10926
    },
    {
      "epoch": 3.4502683928007576,
      "grad_norm": 3.1601955890655518,
      "learning_rate": 1.7219844221458144e-05,
      "loss": 0.5981,
      "step": 10927
    },
    {
      "epoch": 3.4505841490369433,
      "grad_norm": 3.2508881092071533,
      "learning_rate": 1.721633569574065e-05,
      "loss": 0.6342,
      "step": 10928
    },
    {
      "epoch": 3.450899905273129,
      "grad_norm": 3.307138681411743,
      "learning_rate": 1.7212827170023158e-05,
      "loss": 0.6448,
      "step": 10929
    },
    {
      "epoch": 3.451215661509315,
      "grad_norm": 4.002537250518799,
      "learning_rate": 1.7209318644305665e-05,
      "loss": 0.6631,
      "step": 10930
    },
    {
      "epoch": 3.4515314177455005,
      "grad_norm": 2.818739414215088,
      "learning_rate": 1.720581011858817e-05,
      "loss": 0.634,
      "step": 10931
    },
    {
      "epoch": 3.4518471739816863,
      "grad_norm": 3.2950894832611084,
      "learning_rate": 1.7202301592870675e-05,
      "loss": 0.6966,
      "step": 10932
    },
    {
      "epoch": 3.452162930217872,
      "grad_norm": 2.9659600257873535,
      "learning_rate": 1.719879306715318e-05,
      "loss": 0.5705,
      "step": 10933
    },
    {
      "epoch": 3.4524786864540573,
      "grad_norm": 3.1803202629089355,
      "learning_rate": 1.719528454143569e-05,
      "loss": 0.6528,
      "step": 10934
    },
    {
      "epoch": 3.452794442690243,
      "grad_norm": 3.936784029006958,
      "learning_rate": 1.71917760157182e-05,
      "loss": 0.6881,
      "step": 10935
    },
    {
      "epoch": 3.453110198926429,
      "grad_norm": 2.6531600952148438,
      "learning_rate": 1.7188267490000702e-05,
      "loss": 0.5948,
      "step": 10936
    },
    {
      "epoch": 3.4534259551626145,
      "grad_norm": 3.98130202293396,
      "learning_rate": 1.718475896428321e-05,
      "loss": 0.7335,
      "step": 10937
    },
    {
      "epoch": 3.4537417113988003,
      "grad_norm": 3.219512939453125,
      "learning_rate": 1.7181250438565716e-05,
      "loss": 0.7401,
      "step": 10938
    },
    {
      "epoch": 3.4540574676349856,
      "grad_norm": 3.883681297302246,
      "learning_rate": 1.7177741912848222e-05,
      "loss": 0.6818,
      "step": 10939
    },
    {
      "epoch": 3.4543732238711713,
      "grad_norm": 3.0407967567443848,
      "learning_rate": 1.717423338713073e-05,
      "loss": 0.6144,
      "step": 10940
    },
    {
      "epoch": 3.454688980107357,
      "grad_norm": 3.3930184841156006,
      "learning_rate": 1.7170724861413233e-05,
      "loss": 0.7087,
      "step": 10941
    },
    {
      "epoch": 3.4550047363435428,
      "grad_norm": 2.5969715118408203,
      "learning_rate": 1.716721633569574e-05,
      "loss": 0.5785,
      "step": 10942
    },
    {
      "epoch": 3.4553204925797285,
      "grad_norm": 3.53922438621521,
      "learning_rate": 1.7163707809978246e-05,
      "loss": 0.6756,
      "step": 10943
    },
    {
      "epoch": 3.4556362488159142,
      "grad_norm": 3.229889392852783,
      "learning_rate": 1.7160199284260757e-05,
      "loss": 0.6556,
      "step": 10944
    },
    {
      "epoch": 3.4559520050521,
      "grad_norm": 3.3473689556121826,
      "learning_rate": 1.7156690758543263e-05,
      "loss": 0.5899,
      "step": 10945
    },
    {
      "epoch": 3.4562677612882853,
      "grad_norm": 2.723381757736206,
      "learning_rate": 1.7153182232825767e-05,
      "loss": 0.6598,
      "step": 10946
    },
    {
      "epoch": 3.456583517524471,
      "grad_norm": 3.9834418296813965,
      "learning_rate": 1.7149673707108274e-05,
      "loss": 0.7335,
      "step": 10947
    },
    {
      "epoch": 3.4568992737606568,
      "grad_norm": 2.9716529846191406,
      "learning_rate": 1.714616518139078e-05,
      "loss": 0.638,
      "step": 10948
    },
    {
      "epoch": 3.4572150299968425,
      "grad_norm": 3.731632709503174,
      "learning_rate": 1.7142656655673287e-05,
      "loss": 0.772,
      "step": 10949
    },
    {
      "epoch": 3.4575307862330282,
      "grad_norm": 2.939565658569336,
      "learning_rate": 1.7139148129955794e-05,
      "loss": 0.6423,
      "step": 10950
    },
    {
      "epoch": 3.4578465424692135,
      "grad_norm": 3.4792685508728027,
      "learning_rate": 1.7135639604238297e-05,
      "loss": 0.7115,
      "step": 10951
    },
    {
      "epoch": 3.4581622987053993,
      "grad_norm": 3.223607063293457,
      "learning_rate": 1.7132131078520804e-05,
      "loss": 0.5906,
      "step": 10952
    },
    {
      "epoch": 3.458478054941585,
      "grad_norm": 3.3450675010681152,
      "learning_rate": 1.7128622552803314e-05,
      "loss": 0.7142,
      "step": 10953
    },
    {
      "epoch": 3.4587938111777707,
      "grad_norm": 3.786221742630005,
      "learning_rate": 1.712511402708582e-05,
      "loss": 0.6619,
      "step": 10954
    },
    {
      "epoch": 3.4591095674139565,
      "grad_norm": 3.813049793243408,
      "learning_rate": 1.7121605501368325e-05,
      "loss": 0.744,
      "step": 10955
    },
    {
      "epoch": 3.459425323650142,
      "grad_norm": 3.1537115573883057,
      "learning_rate": 1.711809697565083e-05,
      "loss": 0.6876,
      "step": 10956
    },
    {
      "epoch": 3.459741079886328,
      "grad_norm": 3.2042977809906006,
      "learning_rate": 1.711458844993334e-05,
      "loss": 0.6919,
      "step": 10957
    },
    {
      "epoch": 3.4600568361225132,
      "grad_norm": 3.230433702468872,
      "learning_rate": 1.7111079924215845e-05,
      "loss": 0.6557,
      "step": 10958
    },
    {
      "epoch": 3.460372592358699,
      "grad_norm": 2.8674681186676025,
      "learning_rate": 1.7107571398498352e-05,
      "loss": 0.6241,
      "step": 10959
    },
    {
      "epoch": 3.4606883485948847,
      "grad_norm": 3.0334248542785645,
      "learning_rate": 1.710406287278086e-05,
      "loss": 0.6762,
      "step": 10960
    },
    {
      "epoch": 3.4610041048310705,
      "grad_norm": 3.44022274017334,
      "learning_rate": 1.7100554347063366e-05,
      "loss": 0.7347,
      "step": 10961
    },
    {
      "epoch": 3.461319861067256,
      "grad_norm": 3.3160152435302734,
      "learning_rate": 1.7097045821345872e-05,
      "loss": 0.6578,
      "step": 10962
    },
    {
      "epoch": 3.461635617303442,
      "grad_norm": 3.4180147647857666,
      "learning_rate": 1.709353729562838e-05,
      "loss": 0.7121,
      "step": 10963
    },
    {
      "epoch": 3.4619513735396272,
      "grad_norm": 3.3760268688201904,
      "learning_rate": 1.7090028769910886e-05,
      "loss": 0.5795,
      "step": 10964
    },
    {
      "epoch": 3.462267129775813,
      "grad_norm": 3.8677818775177,
      "learning_rate": 1.708652024419339e-05,
      "loss": 0.7069,
      "step": 10965
    },
    {
      "epoch": 3.4625828860119987,
      "grad_norm": 3.6363210678100586,
      "learning_rate": 1.7083011718475896e-05,
      "loss": 0.5842,
      "step": 10966
    },
    {
      "epoch": 3.4628986422481844,
      "grad_norm": 3.997339963912964,
      "learning_rate": 1.7079503192758403e-05,
      "loss": 0.5902,
      "step": 10967
    },
    {
      "epoch": 3.46321439848437,
      "grad_norm": 3.9911816120147705,
      "learning_rate": 1.707599466704091e-05,
      "loss": 0.597,
      "step": 10968
    },
    {
      "epoch": 3.463530154720556,
      "grad_norm": 3.2527072429656982,
      "learning_rate": 1.7072486141323417e-05,
      "loss": 0.6536,
      "step": 10969
    },
    {
      "epoch": 3.463845910956741,
      "grad_norm": 3.530552864074707,
      "learning_rate": 1.7068977615605924e-05,
      "loss": 0.7172,
      "step": 10970
    },
    {
      "epoch": 3.464161667192927,
      "grad_norm": 2.990114212036133,
      "learning_rate": 1.706546908988843e-05,
      "loss": 0.6561,
      "step": 10971
    },
    {
      "epoch": 3.4644774234291127,
      "grad_norm": 4.0941162109375,
      "learning_rate": 1.7061960564170937e-05,
      "loss": 0.619,
      "step": 10972
    },
    {
      "epoch": 3.4647931796652984,
      "grad_norm": 3.6413145065307617,
      "learning_rate": 1.7058452038453444e-05,
      "loss": 0.5749,
      "step": 10973
    },
    {
      "epoch": 3.465108935901484,
      "grad_norm": 3.2345473766326904,
      "learning_rate": 1.705494351273595e-05,
      "loss": 0.6364,
      "step": 10974
    },
    {
      "epoch": 3.46542469213767,
      "grad_norm": 3.350444793701172,
      "learning_rate": 1.7051434987018454e-05,
      "loss": 0.675,
      "step": 10975
    },
    {
      "epoch": 3.465740448373855,
      "grad_norm": 2.9709768295288086,
      "learning_rate": 1.704792646130096e-05,
      "loss": 0.5687,
      "step": 10976
    },
    {
      "epoch": 3.466056204610041,
      "grad_norm": 3.521899700164795,
      "learning_rate": 1.7044417935583468e-05,
      "loss": 0.6802,
      "step": 10977
    },
    {
      "epoch": 3.4663719608462267,
      "grad_norm": 3.4392037391662598,
      "learning_rate": 1.7040909409865975e-05,
      "loss": 0.7577,
      "step": 10978
    },
    {
      "epoch": 3.4666877170824124,
      "grad_norm": 3.4350318908691406,
      "learning_rate": 1.703740088414848e-05,
      "loss": 0.6847,
      "step": 10979
    },
    {
      "epoch": 3.467003473318598,
      "grad_norm": 3.104896068572998,
      "learning_rate": 1.7033892358430988e-05,
      "loss": 0.6558,
      "step": 10980
    },
    {
      "epoch": 3.467319229554784,
      "grad_norm": 4.060919761657715,
      "learning_rate": 1.7030383832713495e-05,
      "loss": 0.7343,
      "step": 10981
    },
    {
      "epoch": 3.467634985790969,
      "grad_norm": 3.381657600402832,
      "learning_rate": 1.7026875306996002e-05,
      "loss": 0.601,
      "step": 10982
    },
    {
      "epoch": 3.467950742027155,
      "grad_norm": 3.171273946762085,
      "learning_rate": 1.702336678127851e-05,
      "loss": 0.6526,
      "step": 10983
    },
    {
      "epoch": 3.4682664982633407,
      "grad_norm": 2.9454407691955566,
      "learning_rate": 1.7019858255561012e-05,
      "loss": 0.5824,
      "step": 10984
    },
    {
      "epoch": 3.4685822544995264,
      "grad_norm": 3.3407785892486572,
      "learning_rate": 1.701634972984352e-05,
      "loss": 0.5533,
      "step": 10985
    },
    {
      "epoch": 3.468898010735712,
      "grad_norm": 2.9367659091949463,
      "learning_rate": 1.7012841204126026e-05,
      "loss": 0.6413,
      "step": 10986
    },
    {
      "epoch": 3.469213766971898,
      "grad_norm": 2.8219525814056396,
      "learning_rate": 1.7009332678408533e-05,
      "loss": 0.6311,
      "step": 10987
    },
    {
      "epoch": 3.469529523208083,
      "grad_norm": 3.2193071842193604,
      "learning_rate": 1.7005824152691043e-05,
      "loss": 0.5976,
      "step": 10988
    },
    {
      "epoch": 3.469845279444269,
      "grad_norm": 3.5683321952819824,
      "learning_rate": 1.7002315626973546e-05,
      "loss": 0.6858,
      "step": 10989
    },
    {
      "epoch": 3.4701610356804546,
      "grad_norm": 4.437849998474121,
      "learning_rate": 1.6998807101256053e-05,
      "loss": 0.6612,
      "step": 10990
    },
    {
      "epoch": 3.4704767919166404,
      "grad_norm": 3.4619550704956055,
      "learning_rate": 1.699529857553856e-05,
      "loss": 0.727,
      "step": 10991
    },
    {
      "epoch": 3.470792548152826,
      "grad_norm": 3.2150166034698486,
      "learning_rate": 1.6991790049821067e-05,
      "loss": 0.6389,
      "step": 10992
    },
    {
      "epoch": 3.471108304389012,
      "grad_norm": 3.5177721977233887,
      "learning_rate": 1.6988281524103573e-05,
      "loss": 0.6299,
      "step": 10993
    },
    {
      "epoch": 3.471424060625197,
      "grad_norm": 3.8911690711975098,
      "learning_rate": 1.6984772998386077e-05,
      "loss": 0.7137,
      "step": 10994
    },
    {
      "epoch": 3.471739816861383,
      "grad_norm": 3.5510342121124268,
      "learning_rate": 1.6981264472668584e-05,
      "loss": 0.5784,
      "step": 10995
    },
    {
      "epoch": 3.4720555730975686,
      "grad_norm": 3.476609230041504,
      "learning_rate": 1.6977755946951094e-05,
      "loss": 0.6567,
      "step": 10996
    },
    {
      "epoch": 3.4723713293337544,
      "grad_norm": 3.2870984077453613,
      "learning_rate": 1.69742474212336e-05,
      "loss": 0.6682,
      "step": 10997
    },
    {
      "epoch": 3.47268708556994,
      "grad_norm": 3.4055399894714355,
      "learning_rate": 1.6970738895516104e-05,
      "loss": 0.6382,
      "step": 10998
    },
    {
      "epoch": 3.473002841806126,
      "grad_norm": 4.0175018310546875,
      "learning_rate": 1.696723036979861e-05,
      "loss": 0.6621,
      "step": 10999
    },
    {
      "epoch": 3.4733185980423116,
      "grad_norm": 3.3122851848602295,
      "learning_rate": 1.6963721844081118e-05,
      "loss": 0.6891,
      "step": 11000
    },
    {
      "epoch": 3.4733185980423116,
      "eval_loss": 0.8077431321144104,
      "eval_runtime": 4.5998,
      "eval_samples_per_second": 541.325,
      "eval_steps_per_second": 4.565,
      "step": 11000
    },
    {
      "epoch": 3.473634354278497,
      "grad_norm": 3.0931801795959473,
      "learning_rate": 1.6960213318363625e-05,
      "loss": 0.5982,
      "step": 11001
    },
    {
      "epoch": 3.4739501105146826,
      "grad_norm": 3.261590003967285,
      "learning_rate": 1.695670479264613e-05,
      "loss": 0.7181,
      "step": 11002
    },
    {
      "epoch": 3.4742658667508683,
      "grad_norm": 3.157425880432129,
      "learning_rate": 1.6953196266928638e-05,
      "loss": 0.6085,
      "step": 11003
    },
    {
      "epoch": 3.474581622987054,
      "grad_norm": 3.493412733078003,
      "learning_rate": 1.694968774121114e-05,
      "loss": 0.6803,
      "step": 11004
    },
    {
      "epoch": 3.47489737922324,
      "grad_norm": 3.5137898921966553,
      "learning_rate": 1.6946179215493652e-05,
      "loss": 0.6056,
      "step": 11005
    },
    {
      "epoch": 3.475213135459425,
      "grad_norm": 3.630342721939087,
      "learning_rate": 1.694267068977616e-05,
      "loss": 0.6471,
      "step": 11006
    },
    {
      "epoch": 3.475528891695611,
      "grad_norm": 3.5394198894500732,
      "learning_rate": 1.6939162164058665e-05,
      "loss": 0.7358,
      "step": 11007
    },
    {
      "epoch": 3.4758446479317966,
      "grad_norm": 3.61875581741333,
      "learning_rate": 1.693565363834117e-05,
      "loss": 0.7189,
      "step": 11008
    },
    {
      "epoch": 3.4761604041679823,
      "grad_norm": 3.200622797012329,
      "learning_rate": 1.6932145112623676e-05,
      "loss": 0.6226,
      "step": 11009
    },
    {
      "epoch": 3.476476160404168,
      "grad_norm": 2.9310173988342285,
      "learning_rate": 1.6928636586906182e-05,
      "loss": 0.5862,
      "step": 11010
    },
    {
      "epoch": 3.476791916640354,
      "grad_norm": 3.325031042098999,
      "learning_rate": 1.692512806118869e-05,
      "loss": 0.6929,
      "step": 11011
    },
    {
      "epoch": 3.4771076728765395,
      "grad_norm": 2.9298908710479736,
      "learning_rate": 1.6921619535471196e-05,
      "loss": 0.5908,
      "step": 11012
    },
    {
      "epoch": 3.477423429112725,
      "grad_norm": 3.0236566066741943,
      "learning_rate": 1.69181110097537e-05,
      "loss": 0.6328,
      "step": 11013
    },
    {
      "epoch": 3.4777391853489106,
      "grad_norm": 3.8804237842559814,
      "learning_rate": 1.691460248403621e-05,
      "loss": 0.6484,
      "step": 11014
    },
    {
      "epoch": 3.4780549415850963,
      "grad_norm": 2.991586923599243,
      "learning_rate": 1.6911093958318717e-05,
      "loss": 0.6156,
      "step": 11015
    },
    {
      "epoch": 3.478370697821282,
      "grad_norm": 2.9969193935394287,
      "learning_rate": 1.6907585432601223e-05,
      "loss": 0.5894,
      "step": 11016
    },
    {
      "epoch": 3.478686454057468,
      "grad_norm": 3.9185492992401123,
      "learning_rate": 1.6904076906883727e-05,
      "loss": 0.6653,
      "step": 11017
    },
    {
      "epoch": 3.479002210293653,
      "grad_norm": 3.2216696739196777,
      "learning_rate": 1.6900568381166234e-05,
      "loss": 0.7083,
      "step": 11018
    },
    {
      "epoch": 3.479317966529839,
      "grad_norm": 3.606889486312866,
      "learning_rate": 1.689705985544874e-05,
      "loss": 0.6432,
      "step": 11019
    },
    {
      "epoch": 3.4796337227660246,
      "grad_norm": 2.87955904006958,
      "learning_rate": 1.6893551329731247e-05,
      "loss": 0.6061,
      "step": 11020
    },
    {
      "epoch": 3.4799494790022103,
      "grad_norm": 3.5579216480255127,
      "learning_rate": 1.6890042804013754e-05,
      "loss": 0.5659,
      "step": 11021
    },
    {
      "epoch": 3.480265235238396,
      "grad_norm": 3.5585601329803467,
      "learning_rate": 1.688653427829626e-05,
      "loss": 0.6688,
      "step": 11022
    },
    {
      "epoch": 3.4805809914745818,
      "grad_norm": 3.6131696701049805,
      "learning_rate": 1.6883025752578768e-05,
      "loss": 0.6652,
      "step": 11023
    },
    {
      "epoch": 3.4808967477107675,
      "grad_norm": 3.8272457122802734,
      "learning_rate": 1.6879517226861274e-05,
      "loss": 0.6492,
      "step": 11024
    },
    {
      "epoch": 3.481212503946953,
      "grad_norm": 2.9895732402801514,
      "learning_rate": 1.687600870114378e-05,
      "loss": 0.7091,
      "step": 11025
    },
    {
      "epoch": 3.4815282601831385,
      "grad_norm": 3.8943872451782227,
      "learning_rate": 1.6872500175426288e-05,
      "loss": 0.6901,
      "step": 11026
    },
    {
      "epoch": 3.4818440164193243,
      "grad_norm": 2.9849295616149902,
      "learning_rate": 1.686899164970879e-05,
      "loss": 0.583,
      "step": 11027
    },
    {
      "epoch": 3.48215977265551,
      "grad_norm": 3.378389358520508,
      "learning_rate": 1.68654831239913e-05,
      "loss": 0.7061,
      "step": 11028
    },
    {
      "epoch": 3.4824755288916958,
      "grad_norm": 3.832916259765625,
      "learning_rate": 1.6861974598273805e-05,
      "loss": 0.6641,
      "step": 11029
    },
    {
      "epoch": 3.482791285127881,
      "grad_norm": 3.4475295543670654,
      "learning_rate": 1.6858466072556312e-05,
      "loss": 0.6805,
      "step": 11030
    },
    {
      "epoch": 3.483107041364067,
      "grad_norm": 3.5560450553894043,
      "learning_rate": 1.685495754683882e-05,
      "loss": 0.7105,
      "step": 11031
    },
    {
      "epoch": 3.4834227976002525,
      "grad_norm": 3.4886088371276855,
      "learning_rate": 1.6851449021121326e-05,
      "loss": 0.6513,
      "step": 11032
    },
    {
      "epoch": 3.4837385538364383,
      "grad_norm": 4.20126485824585,
      "learning_rate": 1.6847940495403832e-05,
      "loss": 0.7118,
      "step": 11033
    },
    {
      "epoch": 3.484054310072624,
      "grad_norm": 3.7077691555023193,
      "learning_rate": 1.684443196968634e-05,
      "loss": 0.7149,
      "step": 11034
    },
    {
      "epoch": 3.4843700663088097,
      "grad_norm": 3.3299808502197266,
      "learning_rate": 1.6840923443968846e-05,
      "loss": 0.6802,
      "step": 11035
    },
    {
      "epoch": 3.4846858225449955,
      "grad_norm": 3.2088146209716797,
      "learning_rate": 1.6837414918251353e-05,
      "loss": 0.6403,
      "step": 11036
    },
    {
      "epoch": 3.4850015787811808,
      "grad_norm": 3.1107289791107178,
      "learning_rate": 1.6833906392533856e-05,
      "loss": 0.6699,
      "step": 11037
    },
    {
      "epoch": 3.4853173350173665,
      "grad_norm": 3.4821341037750244,
      "learning_rate": 1.6830397866816363e-05,
      "loss": 0.5789,
      "step": 11038
    },
    {
      "epoch": 3.4856330912535523,
      "grad_norm": 3.269038200378418,
      "learning_rate": 1.682688934109887e-05,
      "loss": 0.606,
      "step": 11039
    },
    {
      "epoch": 3.485948847489738,
      "grad_norm": 3.605886459350586,
      "learning_rate": 1.682338081538138e-05,
      "loss": 0.6285,
      "step": 11040
    },
    {
      "epoch": 3.4862646037259237,
      "grad_norm": 3.4286949634552,
      "learning_rate": 1.6819872289663884e-05,
      "loss": 0.6361,
      "step": 11041
    },
    {
      "epoch": 3.486580359962109,
      "grad_norm": 3.9582159519195557,
      "learning_rate": 1.681636376394639e-05,
      "loss": 0.6187,
      "step": 11042
    },
    {
      "epoch": 3.4868961161982948,
      "grad_norm": 3.342491388320923,
      "learning_rate": 1.6812855238228897e-05,
      "loss": 0.7607,
      "step": 11043
    },
    {
      "epoch": 3.4872118724344805,
      "grad_norm": 3.878185272216797,
      "learning_rate": 1.6809346712511404e-05,
      "loss": 0.7581,
      "step": 11044
    },
    {
      "epoch": 3.4875276286706662,
      "grad_norm": 3.6073200702667236,
      "learning_rate": 1.680583818679391e-05,
      "loss": 0.6001,
      "step": 11045
    },
    {
      "epoch": 3.487843384906852,
      "grad_norm": 3.5153920650482178,
      "learning_rate": 1.6802329661076414e-05,
      "loss": 0.6652,
      "step": 11046
    },
    {
      "epoch": 3.4881591411430377,
      "grad_norm": 4.0302815437316895,
      "learning_rate": 1.679882113535892e-05,
      "loss": 0.7323,
      "step": 11047
    },
    {
      "epoch": 3.4884748973792234,
      "grad_norm": 2.958167314529419,
      "learning_rate": 1.6795312609641428e-05,
      "loss": 0.6741,
      "step": 11048
    },
    {
      "epoch": 3.4887906536154087,
      "grad_norm": 3.7407498359680176,
      "learning_rate": 1.6791804083923938e-05,
      "loss": 0.6247,
      "step": 11049
    },
    {
      "epoch": 3.4891064098515945,
      "grad_norm": 2.95448899269104,
      "learning_rate": 1.6788295558206445e-05,
      "loss": 0.5107,
      "step": 11050
    },
    {
      "epoch": 3.48942216608778,
      "grad_norm": 3.45683217048645,
      "learning_rate": 1.6784787032488948e-05,
      "loss": 0.6867,
      "step": 11051
    },
    {
      "epoch": 3.489737922323966,
      "grad_norm": 3.622227668762207,
      "learning_rate": 1.6781278506771455e-05,
      "loss": 0.6517,
      "step": 11052
    },
    {
      "epoch": 3.4900536785601517,
      "grad_norm": 3.5397372245788574,
      "learning_rate": 1.6777769981053962e-05,
      "loss": 0.6667,
      "step": 11053
    },
    {
      "epoch": 3.490369434796337,
      "grad_norm": 3.055077314376831,
      "learning_rate": 1.677426145533647e-05,
      "loss": 0.5597,
      "step": 11054
    },
    {
      "epoch": 3.4906851910325227,
      "grad_norm": 3.1466755867004395,
      "learning_rate": 1.6770752929618976e-05,
      "loss": 0.787,
      "step": 11055
    },
    {
      "epoch": 3.4910009472687085,
      "grad_norm": 3.0389647483825684,
      "learning_rate": 1.676724440390148e-05,
      "loss": 0.6684,
      "step": 11056
    },
    {
      "epoch": 3.491316703504894,
      "grad_norm": 3.6801390647888184,
      "learning_rate": 1.676373587818399e-05,
      "loss": 0.6878,
      "step": 11057
    },
    {
      "epoch": 3.49163245974108,
      "grad_norm": 3.156642436981201,
      "learning_rate": 1.6760227352466496e-05,
      "loss": 0.6391,
      "step": 11058
    },
    {
      "epoch": 3.4919482159772657,
      "grad_norm": 3.4186339378356934,
      "learning_rate": 1.6756718826749003e-05,
      "loss": 0.6888,
      "step": 11059
    },
    {
      "epoch": 3.4922639722134514,
      "grad_norm": 3.0553081035614014,
      "learning_rate": 1.6753210301031506e-05,
      "loss": 0.6334,
      "step": 11060
    },
    {
      "epoch": 3.4925797284496367,
      "grad_norm": 3.7120063304901123,
      "learning_rate": 1.6749701775314013e-05,
      "loss": 0.6438,
      "step": 11061
    },
    {
      "epoch": 3.4928954846858224,
      "grad_norm": 3.846867084503174,
      "learning_rate": 1.674619324959652e-05,
      "loss": 0.7107,
      "step": 11062
    },
    {
      "epoch": 3.493211240922008,
      "grad_norm": 3.715925931930542,
      "learning_rate": 1.6742684723879027e-05,
      "loss": 0.6326,
      "step": 11063
    },
    {
      "epoch": 3.493526997158194,
      "grad_norm": 3.3669369220733643,
      "learning_rate": 1.6739176198161533e-05,
      "loss": 0.636,
      "step": 11064
    },
    {
      "epoch": 3.4938427533943797,
      "grad_norm": 3.384234666824341,
      "learning_rate": 1.673566767244404e-05,
      "loss": 0.6731,
      "step": 11065
    },
    {
      "epoch": 3.494158509630565,
      "grad_norm": 3.963071346282959,
      "learning_rate": 1.6732159146726547e-05,
      "loss": 0.6394,
      "step": 11066
    },
    {
      "epoch": 3.4944742658667507,
      "grad_norm": 3.2542476654052734,
      "learning_rate": 1.6728650621009054e-05,
      "loss": 0.6153,
      "step": 11067
    },
    {
      "epoch": 3.4947900221029364,
      "grad_norm": 3.5152502059936523,
      "learning_rate": 1.672514209529156e-05,
      "loss": 0.6865,
      "step": 11068
    },
    {
      "epoch": 3.495105778339122,
      "grad_norm": 2.88407039642334,
      "learning_rate": 1.6721633569574068e-05,
      "loss": 0.65,
      "step": 11069
    },
    {
      "epoch": 3.495421534575308,
      "grad_norm": 3.475231647491455,
      "learning_rate": 1.671812504385657e-05,
      "loss": 0.6585,
      "step": 11070
    },
    {
      "epoch": 3.4957372908114936,
      "grad_norm": 3.8415753841400146,
      "learning_rate": 1.6714616518139078e-05,
      "loss": 0.6875,
      "step": 11071
    },
    {
      "epoch": 3.4960530470476794,
      "grad_norm": 4.105625152587891,
      "learning_rate": 1.6711107992421585e-05,
      "loss": 0.6897,
      "step": 11072
    },
    {
      "epoch": 3.4963688032838647,
      "grad_norm": 2.9715473651885986,
      "learning_rate": 1.670759946670409e-05,
      "loss": 0.6115,
      "step": 11073
    },
    {
      "epoch": 3.4966845595200504,
      "grad_norm": 2.684156656265259,
      "learning_rate": 1.6704090940986598e-05,
      "loss": 0.5854,
      "step": 11074
    },
    {
      "epoch": 3.497000315756236,
      "grad_norm": 3.466517448425293,
      "learning_rate": 1.6700582415269105e-05,
      "loss": 0.6947,
      "step": 11075
    },
    {
      "epoch": 3.497316071992422,
      "grad_norm": 3.1180200576782227,
      "learning_rate": 1.6697073889551612e-05,
      "loss": 0.6549,
      "step": 11076
    },
    {
      "epoch": 3.4976318282286076,
      "grad_norm": 3.650902032852173,
      "learning_rate": 1.669356536383412e-05,
      "loss": 0.6229,
      "step": 11077
    },
    {
      "epoch": 3.4979475844647934,
      "grad_norm": 3.8134889602661133,
      "learning_rate": 1.6690056838116625e-05,
      "loss": 0.6879,
      "step": 11078
    },
    {
      "epoch": 3.4982633407009787,
      "grad_norm": 3.2432861328125,
      "learning_rate": 1.6686548312399132e-05,
      "loss": 0.7392,
      "step": 11079
    },
    {
      "epoch": 3.4985790969371644,
      "grad_norm": 3.5456531047821045,
      "learning_rate": 1.6683039786681636e-05,
      "loss": 0.6888,
      "step": 11080
    },
    {
      "epoch": 3.49889485317335,
      "grad_norm": 3.924268960952759,
      "learning_rate": 1.6679531260964142e-05,
      "loss": 0.7625,
      "step": 11081
    },
    {
      "epoch": 3.499210609409536,
      "grad_norm": 3.2482082843780518,
      "learning_rate": 1.667602273524665e-05,
      "loss": 0.6137,
      "step": 11082
    },
    {
      "epoch": 3.4995263656457216,
      "grad_norm": 3.993535041809082,
      "learning_rate": 1.6672514209529156e-05,
      "loss": 0.7076,
      "step": 11083
    },
    {
      "epoch": 3.4998421218819074,
      "grad_norm": 3.1718149185180664,
      "learning_rate": 1.6669005683811663e-05,
      "loss": 0.6811,
      "step": 11084
    },
    {
      "epoch": 3.5001578781180926,
      "grad_norm": 2.7658004760742188,
      "learning_rate": 1.666549715809417e-05,
      "loss": 0.6564,
      "step": 11085
    },
    {
      "epoch": 3.5004736343542784,
      "grad_norm": 3.567655563354492,
      "learning_rate": 1.6661988632376677e-05,
      "loss": 0.5778,
      "step": 11086
    },
    {
      "epoch": 3.500789390590464,
      "grad_norm": 3.0784924030303955,
      "learning_rate": 1.6658480106659183e-05,
      "loss": 0.6766,
      "step": 11087
    },
    {
      "epoch": 3.50110514682665,
      "grad_norm": 3.785479784011841,
      "learning_rate": 1.665497158094169e-05,
      "loss": 0.6954,
      "step": 11088
    },
    {
      "epoch": 3.5014209030628356,
      "grad_norm": 3.390955924987793,
      "learning_rate": 1.6651463055224194e-05,
      "loss": 0.6257,
      "step": 11089
    },
    {
      "epoch": 3.501736659299021,
      "grad_norm": 3.3429462909698486,
      "learning_rate": 1.66479545295067e-05,
      "loss": 0.6247,
      "step": 11090
    },
    {
      "epoch": 3.502052415535207,
      "grad_norm": 3.5769712924957275,
      "learning_rate": 1.6644446003789207e-05,
      "loss": 0.7465,
      "step": 11091
    },
    {
      "epoch": 3.5023681717713924,
      "grad_norm": 3.408376455307007,
      "learning_rate": 1.6640937478071717e-05,
      "loss": 0.7122,
      "step": 11092
    },
    {
      "epoch": 3.502683928007578,
      "grad_norm": 3.5856354236602783,
      "learning_rate": 1.6637428952354224e-05,
      "loss": 0.7073,
      "step": 11093
    },
    {
      "epoch": 3.502999684243764,
      "grad_norm": 3.559098243713379,
      "learning_rate": 1.6633920426636728e-05,
      "loss": 0.6638,
      "step": 11094
    },
    {
      "epoch": 3.5033154404799496,
      "grad_norm": 3.5454721450805664,
      "learning_rate": 1.6630411900919234e-05,
      "loss": 0.7198,
      "step": 11095
    },
    {
      "epoch": 3.5036311967161353,
      "grad_norm": 3.685922622680664,
      "learning_rate": 1.662690337520174e-05,
      "loss": 0.582,
      "step": 11096
    },
    {
      "epoch": 3.5039469529523206,
      "grad_norm": 3.532014846801758,
      "learning_rate": 1.6623394849484248e-05,
      "loss": 0.6195,
      "step": 11097
    },
    {
      "epoch": 3.5042627091885064,
      "grad_norm": 3.486556053161621,
      "learning_rate": 1.6619886323766755e-05,
      "loss": 0.702,
      "step": 11098
    },
    {
      "epoch": 3.504578465424692,
      "grad_norm": 3.213782787322998,
      "learning_rate": 1.661637779804926e-05,
      "loss": 0.6421,
      "step": 11099
    },
    {
      "epoch": 3.504894221660878,
      "grad_norm": 3.2618141174316406,
      "learning_rate": 1.6612869272331765e-05,
      "loss": 0.6029,
      "step": 11100
    },
    {
      "epoch": 3.5052099778970636,
      "grad_norm": 3.576859474182129,
      "learning_rate": 1.6609360746614275e-05,
      "loss": 0.722,
      "step": 11101
    },
    {
      "epoch": 3.505525734133249,
      "grad_norm": 3.9395158290863037,
      "learning_rate": 1.6605852220896782e-05,
      "loss": 0.7027,
      "step": 11102
    },
    {
      "epoch": 3.505841490369435,
      "grad_norm": 3.2192294597625732,
      "learning_rate": 1.6602343695179286e-05,
      "loss": 0.6315,
      "step": 11103
    },
    {
      "epoch": 3.5061572466056203,
      "grad_norm": 3.305506706237793,
      "learning_rate": 1.6598835169461792e-05,
      "loss": 0.66,
      "step": 11104
    },
    {
      "epoch": 3.506473002841806,
      "grad_norm": 3.9050538539886475,
      "learning_rate": 1.65953266437443e-05,
      "loss": 0.7405,
      "step": 11105
    },
    {
      "epoch": 3.506788759077992,
      "grad_norm": 2.9228973388671875,
      "learning_rate": 1.6591818118026806e-05,
      "loss": 0.5346,
      "step": 11106
    },
    {
      "epoch": 3.5071045153141776,
      "grad_norm": 3.217148542404175,
      "learning_rate": 1.6588309592309313e-05,
      "loss": 0.6265,
      "step": 11107
    },
    {
      "epoch": 3.5074202715503633,
      "grad_norm": 3.659910202026367,
      "learning_rate": 1.658480106659182e-05,
      "loss": 0.7334,
      "step": 11108
    },
    {
      "epoch": 3.5077360277865486,
      "grad_norm": 3.609339475631714,
      "learning_rate": 1.6581292540874323e-05,
      "loss": 0.7381,
      "step": 11109
    },
    {
      "epoch": 3.5080517840227343,
      "grad_norm": 3.449075937271118,
      "learning_rate": 1.6577784015156833e-05,
      "loss": 0.5654,
      "step": 11110
    },
    {
      "epoch": 3.50836754025892,
      "grad_norm": 3.1720242500305176,
      "learning_rate": 1.657427548943934e-05,
      "loss": 0.6039,
      "step": 11111
    },
    {
      "epoch": 3.508683296495106,
      "grad_norm": 3.6864612102508545,
      "learning_rate": 1.6570766963721847e-05,
      "loss": 0.6053,
      "step": 11112
    },
    {
      "epoch": 3.5089990527312915,
      "grad_norm": 3.3461053371429443,
      "learning_rate": 1.656725843800435e-05,
      "loss": 0.7062,
      "step": 11113
    },
    {
      "epoch": 3.5093148089674773,
      "grad_norm": 3.4790685176849365,
      "learning_rate": 1.6563749912286857e-05,
      "loss": 0.6784,
      "step": 11114
    },
    {
      "epoch": 3.509630565203663,
      "grad_norm": 3.3573460578918457,
      "learning_rate": 1.6560241386569364e-05,
      "loss": 0.6242,
      "step": 11115
    },
    {
      "epoch": 3.5099463214398483,
      "grad_norm": 3.1283559799194336,
      "learning_rate": 1.655673286085187e-05,
      "loss": 0.6455,
      "step": 11116
    },
    {
      "epoch": 3.510262077676034,
      "grad_norm": 3.3084685802459717,
      "learning_rate": 1.6553224335134378e-05,
      "loss": 0.643,
      "step": 11117
    },
    {
      "epoch": 3.51057783391222,
      "grad_norm": 3.657017230987549,
      "learning_rate": 1.654971580941688e-05,
      "loss": 0.6461,
      "step": 11118
    },
    {
      "epoch": 3.5108935901484055,
      "grad_norm": 3.1298062801361084,
      "learning_rate": 1.654620728369939e-05,
      "loss": 0.6664,
      "step": 11119
    },
    {
      "epoch": 3.5112093463845913,
      "grad_norm": 3.074571371078491,
      "learning_rate": 1.6542698757981898e-05,
      "loss": 0.6568,
      "step": 11120
    },
    {
      "epoch": 3.5115251026207766,
      "grad_norm": 4.026172637939453,
      "learning_rate": 1.6539190232264405e-05,
      "loss": 0.7344,
      "step": 11121
    },
    {
      "epoch": 3.5118408588569623,
      "grad_norm": 3.193603754043579,
      "learning_rate": 1.6535681706546908e-05,
      "loss": 0.6646,
      "step": 11122
    },
    {
      "epoch": 3.512156615093148,
      "grad_norm": 3.6645684242248535,
      "learning_rate": 1.6532173180829415e-05,
      "loss": 0.7242,
      "step": 11123
    },
    {
      "epoch": 3.5124723713293338,
      "grad_norm": 3.0314741134643555,
      "learning_rate": 1.6528664655111922e-05,
      "loss": 0.6452,
      "step": 11124
    },
    {
      "epoch": 3.5127881275655195,
      "grad_norm": 3.624519109725952,
      "learning_rate": 1.652515612939443e-05,
      "loss": 0.6563,
      "step": 11125
    },
    {
      "epoch": 3.5131038838017052,
      "grad_norm": 2.857668876647949,
      "learning_rate": 1.6521647603676936e-05,
      "loss": 0.6202,
      "step": 11126
    },
    {
      "epoch": 3.513419640037891,
      "grad_norm": 3.206674575805664,
      "learning_rate": 1.6518139077959442e-05,
      "loss": 0.5275,
      "step": 11127
    },
    {
      "epoch": 3.5137353962740763,
      "grad_norm": 3.518874406814575,
      "learning_rate": 1.651463055224195e-05,
      "loss": 0.693,
      "step": 11128
    },
    {
      "epoch": 3.514051152510262,
      "grad_norm": 3.382927417755127,
      "learning_rate": 1.6511122026524456e-05,
      "loss": 0.7266,
      "step": 11129
    },
    {
      "epoch": 3.5143669087464477,
      "grad_norm": 3.9453213214874268,
      "learning_rate": 1.6507613500806963e-05,
      "loss": 0.7002,
      "step": 11130
    },
    {
      "epoch": 3.5146826649826335,
      "grad_norm": 3.8236892223358154,
      "learning_rate": 1.650410497508947e-05,
      "loss": 0.6336,
      "step": 11131
    },
    {
      "epoch": 3.5149984212188192,
      "grad_norm": 3.918107748031616,
      "learning_rate": 1.6500596449371973e-05,
      "loss": 0.701,
      "step": 11132
    },
    {
      "epoch": 3.5153141774550045,
      "grad_norm": 3.5351297855377197,
      "learning_rate": 1.649708792365448e-05,
      "loss": 0.6855,
      "step": 11133
    },
    {
      "epoch": 3.5156299336911903,
      "grad_norm": 3.356334924697876,
      "learning_rate": 1.6493579397936987e-05,
      "loss": 0.7195,
      "step": 11134
    },
    {
      "epoch": 3.515945689927376,
      "grad_norm": 3.878934860229492,
      "learning_rate": 1.6490070872219493e-05,
      "loss": 0.701,
      "step": 11135
    },
    {
      "epoch": 3.5162614461635617,
      "grad_norm": 3.2408347129821777,
      "learning_rate": 1.6486562346502e-05,
      "loss": 0.6617,
      "step": 11136
    },
    {
      "epoch": 3.5165772023997475,
      "grad_norm": 3.7098190784454346,
      "learning_rate": 1.6483053820784507e-05,
      "loss": 0.6877,
      "step": 11137
    },
    {
      "epoch": 3.516892958635933,
      "grad_norm": 3.662022590637207,
      "learning_rate": 1.6479545295067014e-05,
      "loss": 0.6466,
      "step": 11138
    },
    {
      "epoch": 3.517208714872119,
      "grad_norm": 3.239654064178467,
      "learning_rate": 1.647603676934952e-05,
      "loss": 0.6552,
      "step": 11139
    },
    {
      "epoch": 3.5175244711083042,
      "grad_norm": 3.4012656211853027,
      "learning_rate": 1.6472528243632028e-05,
      "loss": 0.6352,
      "step": 11140
    },
    {
      "epoch": 3.51784022734449,
      "grad_norm": 3.7201175689697266,
      "learning_rate": 1.6469019717914534e-05,
      "loss": 0.673,
      "step": 11141
    },
    {
      "epoch": 3.5181559835806757,
      "grad_norm": 3.0843355655670166,
      "learning_rate": 1.6465511192197038e-05,
      "loss": 0.6358,
      "step": 11142
    },
    {
      "epoch": 3.5184717398168615,
      "grad_norm": 3.2423293590545654,
      "learning_rate": 1.6462002666479545e-05,
      "loss": 0.6497,
      "step": 11143
    },
    {
      "epoch": 3.518787496053047,
      "grad_norm": 3.2872676849365234,
      "learning_rate": 1.645849414076205e-05,
      "loss": 0.67,
      "step": 11144
    },
    {
      "epoch": 3.5191032522892325,
      "grad_norm": 2.7042136192321777,
      "learning_rate": 1.645498561504456e-05,
      "loss": 0.6113,
      "step": 11145
    },
    {
      "epoch": 3.5194190085254182,
      "grad_norm": 3.0880179405212402,
      "learning_rate": 1.6451477089327065e-05,
      "loss": 0.5819,
      "step": 11146
    },
    {
      "epoch": 3.519734764761604,
      "grad_norm": 3.2445292472839355,
      "learning_rate": 1.6447968563609572e-05,
      "loss": 0.6454,
      "step": 11147
    },
    {
      "epoch": 3.5200505209977897,
      "grad_norm": 3.4249088764190674,
      "learning_rate": 1.644446003789208e-05,
      "loss": 0.6233,
      "step": 11148
    },
    {
      "epoch": 3.5203662772339754,
      "grad_norm": 2.825683355331421,
      "learning_rate": 1.6440951512174585e-05,
      "loss": 0.6238,
      "step": 11149
    },
    {
      "epoch": 3.520682033470161,
      "grad_norm": 3.9101216793060303,
      "learning_rate": 1.6437442986457092e-05,
      "loss": 0.6368,
      "step": 11150
    },
    {
      "epoch": 3.520997789706347,
      "grad_norm": 3.622138023376465,
      "learning_rate": 1.6433934460739596e-05,
      "loss": 0.6993,
      "step": 11151
    },
    {
      "epoch": 3.521313545942532,
      "grad_norm": 3.062842845916748,
      "learning_rate": 1.6430425935022102e-05,
      "loss": 0.623,
      "step": 11152
    },
    {
      "epoch": 3.521629302178718,
      "grad_norm": 3.2843997478485107,
      "learning_rate": 1.642691740930461e-05,
      "loss": 0.8004,
      "step": 11153
    },
    {
      "epoch": 3.5219450584149037,
      "grad_norm": 3.8123860359191895,
      "learning_rate": 1.642340888358712e-05,
      "loss": 0.7144,
      "step": 11154
    },
    {
      "epoch": 3.5222608146510894,
      "grad_norm": 3.3838016986846924,
      "learning_rate": 1.6419900357869626e-05,
      "loss": 0.6683,
      "step": 11155
    },
    {
      "epoch": 3.522576570887275,
      "grad_norm": 3.5189929008483887,
      "learning_rate": 1.641639183215213e-05,
      "loss": 0.68,
      "step": 11156
    },
    {
      "epoch": 3.5228923271234605,
      "grad_norm": 3.957282543182373,
      "learning_rate": 1.6412883306434637e-05,
      "loss": 0.7028,
      "step": 11157
    },
    {
      "epoch": 3.523208083359646,
      "grad_norm": 3.143799304962158,
      "learning_rate": 1.6409374780717143e-05,
      "loss": 0.6613,
      "step": 11158
    },
    {
      "epoch": 3.523523839595832,
      "grad_norm": 3.002070903778076,
      "learning_rate": 1.640586625499965e-05,
      "loss": 0.6347,
      "step": 11159
    },
    {
      "epoch": 3.5238395958320177,
      "grad_norm": 3.323556661605835,
      "learning_rate": 1.6402357729282157e-05,
      "loss": 0.6213,
      "step": 11160
    },
    {
      "epoch": 3.5241553520682034,
      "grad_norm": 3.0116920471191406,
      "learning_rate": 1.639884920356466e-05,
      "loss": 0.67,
      "step": 11161
    },
    {
      "epoch": 3.524471108304389,
      "grad_norm": 3.8900086879730225,
      "learning_rate": 1.639534067784717e-05,
      "loss": 0.6865,
      "step": 11162
    },
    {
      "epoch": 3.524786864540575,
      "grad_norm": 3.572848320007324,
      "learning_rate": 1.6391832152129677e-05,
      "loss": 0.7292,
      "step": 11163
    },
    {
      "epoch": 3.52510262077676,
      "grad_norm": 3.275193214416504,
      "learning_rate": 1.6388323626412184e-05,
      "loss": 0.6193,
      "step": 11164
    },
    {
      "epoch": 3.525418377012946,
      "grad_norm": 3.6699984073638916,
      "learning_rate": 1.6384815100694688e-05,
      "loss": 0.6379,
      "step": 11165
    },
    {
      "epoch": 3.5257341332491317,
      "grad_norm": 3.5780465602874756,
      "learning_rate": 1.6381306574977194e-05,
      "loss": 0.6117,
      "step": 11166
    },
    {
      "epoch": 3.5260498894853174,
      "grad_norm": 3.761927366256714,
      "learning_rate": 1.63777980492597e-05,
      "loss": 0.693,
      "step": 11167
    },
    {
      "epoch": 3.526365645721503,
      "grad_norm": 3.093329429626465,
      "learning_rate": 1.6374289523542208e-05,
      "loss": 0.5993,
      "step": 11168
    },
    {
      "epoch": 3.5266814019576884,
      "grad_norm": 3.360136032104492,
      "learning_rate": 1.6370780997824715e-05,
      "loss": 0.6954,
      "step": 11169
    },
    {
      "epoch": 3.5269971581938746,
      "grad_norm": 3.1095502376556396,
      "learning_rate": 1.6367272472107222e-05,
      "loss": 0.6241,
      "step": 11170
    },
    {
      "epoch": 3.52731291443006,
      "grad_norm": 2.8769004344940186,
      "learning_rate": 1.636376394638973e-05,
      "loss": 0.6223,
      "step": 11171
    },
    {
      "epoch": 3.5276286706662456,
      "grad_norm": 3.5836851596832275,
      "learning_rate": 1.6360255420672235e-05,
      "loss": 0.71,
      "step": 11172
    },
    {
      "epoch": 3.5279444269024314,
      "grad_norm": 3.1280834674835205,
      "learning_rate": 1.6356746894954742e-05,
      "loss": 0.6491,
      "step": 11173
    },
    {
      "epoch": 3.528260183138617,
      "grad_norm": 2.990049362182617,
      "learning_rate": 1.635323836923725e-05,
      "loss": 0.6153,
      "step": 11174
    },
    {
      "epoch": 3.528575939374803,
      "grad_norm": 3.094022750854492,
      "learning_rate": 1.6349729843519752e-05,
      "loss": 0.7439,
      "step": 11175
    },
    {
      "epoch": 3.528891695610988,
      "grad_norm": 3.15055775642395,
      "learning_rate": 1.634622131780226e-05,
      "loss": 0.6887,
      "step": 11176
    },
    {
      "epoch": 3.529207451847174,
      "grad_norm": 3.5489847660064697,
      "learning_rate": 1.6342712792084766e-05,
      "loss": 0.6823,
      "step": 11177
    },
    {
      "epoch": 3.5295232080833596,
      "grad_norm": 3.0386996269226074,
      "learning_rate": 1.6339204266367273e-05,
      "loss": 0.6068,
      "step": 11178
    },
    {
      "epoch": 3.5298389643195454,
      "grad_norm": 3.6416561603546143,
      "learning_rate": 1.633569574064978e-05,
      "loss": 0.6991,
      "step": 11179
    },
    {
      "epoch": 3.530154720555731,
      "grad_norm": 3.288677930831909,
      "learning_rate": 1.6332187214932286e-05,
      "loss": 0.6796,
      "step": 11180
    },
    {
      "epoch": 3.5304704767919164,
      "grad_norm": 3.0337159633636475,
      "learning_rate": 1.6328678689214793e-05,
      "loss": 0.558,
      "step": 11181
    },
    {
      "epoch": 3.5307862330281026,
      "grad_norm": 4.697006702423096,
      "learning_rate": 1.63251701634973e-05,
      "loss": 0.6101,
      "step": 11182
    },
    {
      "epoch": 3.531101989264288,
      "grad_norm": 3.5687592029571533,
      "learning_rate": 1.6321661637779807e-05,
      "loss": 0.6297,
      "step": 11183
    },
    {
      "epoch": 3.5314177455004736,
      "grad_norm": 2.8903114795684814,
      "learning_rate": 1.6318153112062314e-05,
      "loss": 0.6862,
      "step": 11184
    },
    {
      "epoch": 3.5317335017366593,
      "grad_norm": 3.5976014137268066,
      "learning_rate": 1.6314644586344817e-05,
      "loss": 0.64,
      "step": 11185
    },
    {
      "epoch": 3.532049257972845,
      "grad_norm": 3.077756404876709,
      "learning_rate": 1.6311136060627324e-05,
      "loss": 0.6942,
      "step": 11186
    },
    {
      "epoch": 3.532365014209031,
      "grad_norm": 3.0688905715942383,
      "learning_rate": 1.630762753490983e-05,
      "loss": 0.6138,
      "step": 11187
    },
    {
      "epoch": 3.532680770445216,
      "grad_norm": 3.4497287273406982,
      "learning_rate": 1.6304119009192338e-05,
      "loss": 0.6655,
      "step": 11188
    },
    {
      "epoch": 3.532996526681402,
      "grad_norm": 4.790315628051758,
      "learning_rate": 1.6300610483474844e-05,
      "loss": 0.715,
      "step": 11189
    },
    {
      "epoch": 3.5333122829175876,
      "grad_norm": 3.926713228225708,
      "learning_rate": 1.629710195775735e-05,
      "loss": 0.6242,
      "step": 11190
    },
    {
      "epoch": 3.5336280391537733,
      "grad_norm": 3.837965726852417,
      "learning_rate": 1.6293593432039858e-05,
      "loss": 0.6223,
      "step": 11191
    },
    {
      "epoch": 3.533943795389959,
      "grad_norm": 3.3842570781707764,
      "learning_rate": 1.6290084906322365e-05,
      "loss": 0.6081,
      "step": 11192
    },
    {
      "epoch": 3.5342595516261444,
      "grad_norm": 2.7762653827667236,
      "learning_rate": 1.628657638060487e-05,
      "loss": 0.509,
      "step": 11193
    },
    {
      "epoch": 3.5345753078623305,
      "grad_norm": 3.4395666122436523,
      "learning_rate": 1.6283067854887375e-05,
      "loss": 0.5609,
      "step": 11194
    },
    {
      "epoch": 3.534891064098516,
      "grad_norm": 3.4435253143310547,
      "learning_rate": 1.6279559329169882e-05,
      "loss": 0.7259,
      "step": 11195
    },
    {
      "epoch": 3.5352068203347016,
      "grad_norm": 3.625293493270874,
      "learning_rate": 1.627605080345239e-05,
      "loss": 0.6907,
      "step": 11196
    },
    {
      "epoch": 3.5355225765708873,
      "grad_norm": 3.0035176277160645,
      "learning_rate": 1.62725422777349e-05,
      "loss": 0.6315,
      "step": 11197
    },
    {
      "epoch": 3.535838332807073,
      "grad_norm": 3.4043288230895996,
      "learning_rate": 1.6269033752017406e-05,
      "loss": 0.6878,
      "step": 11198
    },
    {
      "epoch": 3.536154089043259,
      "grad_norm": 3.4660933017730713,
      "learning_rate": 1.626552522629991e-05,
      "loss": 0.64,
      "step": 11199
    },
    {
      "epoch": 3.536469845279444,
      "grad_norm": 4.791501045227051,
      "learning_rate": 1.6262016700582416e-05,
      "loss": 0.5761,
      "step": 11200
    },
    {
      "epoch": 3.536469845279444,
      "eval_loss": 0.8159790635108948,
      "eval_runtime": 4.6763,
      "eval_samples_per_second": 532.477,
      "eval_steps_per_second": 4.491,
      "step": 11200
    },
    {
      "epoch": 3.53678560151563,
      "grad_norm": 3.1431946754455566,
      "learning_rate": 1.6258508174864923e-05,
      "loss": 0.6629,
      "step": 11201
    },
    {
      "epoch": 3.5371013577518156,
      "grad_norm": 3.36265230178833,
      "learning_rate": 1.625499964914743e-05,
      "loss": 0.6039,
      "step": 11202
    },
    {
      "epoch": 3.5374171139880013,
      "grad_norm": 4.016852378845215,
      "learning_rate": 1.6251491123429936e-05,
      "loss": 0.7824,
      "step": 11203
    },
    {
      "epoch": 3.537732870224187,
      "grad_norm": 3.712906837463379,
      "learning_rate": 1.624798259771244e-05,
      "loss": 0.5787,
      "step": 11204
    },
    {
      "epoch": 3.5380486264603723,
      "grad_norm": 3.6293673515319824,
      "learning_rate": 1.6244474071994947e-05,
      "loss": 0.6861,
      "step": 11205
    },
    {
      "epoch": 3.5383643826965585,
      "grad_norm": 3.670063018798828,
      "learning_rate": 1.6240965546277457e-05,
      "loss": 0.593,
      "step": 11206
    },
    {
      "epoch": 3.538680138932744,
      "grad_norm": 3.8779196739196777,
      "learning_rate": 1.6237457020559964e-05,
      "loss": 0.6463,
      "step": 11207
    },
    {
      "epoch": 3.5389958951689295,
      "grad_norm": 3.3574273586273193,
      "learning_rate": 1.6233948494842467e-05,
      "loss": 0.6614,
      "step": 11208
    },
    {
      "epoch": 3.5393116514051153,
      "grad_norm": 3.4441022872924805,
      "learning_rate": 1.6230439969124974e-05,
      "loss": 0.7422,
      "step": 11209
    },
    {
      "epoch": 3.539627407641301,
      "grad_norm": 4.360506057739258,
      "learning_rate": 1.622693144340748e-05,
      "loss": 0.6505,
      "step": 11210
    },
    {
      "epoch": 3.5399431638774868,
      "grad_norm": 4.055713176727295,
      "learning_rate": 1.6223422917689988e-05,
      "loss": 0.7706,
      "step": 11211
    },
    {
      "epoch": 3.540258920113672,
      "grad_norm": 3.339996099472046,
      "learning_rate": 1.6219914391972494e-05,
      "loss": 0.6822,
      "step": 11212
    },
    {
      "epoch": 3.540574676349858,
      "grad_norm": 3.112762928009033,
      "learning_rate": 1.6216405866255e-05,
      "loss": 0.588,
      "step": 11213
    },
    {
      "epoch": 3.5408904325860435,
      "grad_norm": 3.0348339080810547,
      "learning_rate": 1.6212897340537505e-05,
      "loss": 0.6264,
      "step": 11214
    },
    {
      "epoch": 3.5412061888222293,
      "grad_norm": 3.642599582672119,
      "learning_rate": 1.6209388814820015e-05,
      "loss": 0.7555,
      "step": 11215
    },
    {
      "epoch": 3.541521945058415,
      "grad_norm": 3.122034788131714,
      "learning_rate": 1.620588028910252e-05,
      "loss": 0.6041,
      "step": 11216
    },
    {
      "epoch": 3.5418377012946003,
      "grad_norm": 3.769130229949951,
      "learning_rate": 1.620237176338503e-05,
      "loss": 0.6528,
      "step": 11217
    },
    {
      "epoch": 3.5421534575307865,
      "grad_norm": 3.514854907989502,
      "learning_rate": 1.6198863237667532e-05,
      "loss": 0.6229,
      "step": 11218
    },
    {
      "epoch": 3.5424692137669718,
      "grad_norm": 3.2264888286590576,
      "learning_rate": 1.619535471195004e-05,
      "loss": 0.6576,
      "step": 11219
    },
    {
      "epoch": 3.5427849700031575,
      "grad_norm": 3.368638038635254,
      "learning_rate": 1.6191846186232545e-05,
      "loss": 0.698,
      "step": 11220
    },
    {
      "epoch": 3.5431007262393432,
      "grad_norm": 3.2528700828552246,
      "learning_rate": 1.6188337660515052e-05,
      "loss": 0.6071,
      "step": 11221
    },
    {
      "epoch": 3.543416482475529,
      "grad_norm": 3.5138142108917236,
      "learning_rate": 1.618482913479756e-05,
      "loss": 0.7163,
      "step": 11222
    },
    {
      "epoch": 3.5437322387117147,
      "grad_norm": 3.8066914081573486,
      "learning_rate": 1.6181320609080062e-05,
      "loss": 0.6936,
      "step": 11223
    },
    {
      "epoch": 3.5440479949479,
      "grad_norm": 3.309109687805176,
      "learning_rate": 1.6177812083362573e-05,
      "loss": 0.6037,
      "step": 11224
    },
    {
      "epoch": 3.5443637511840858,
      "grad_norm": 3.3634705543518066,
      "learning_rate": 1.617430355764508e-05,
      "loss": 0.6196,
      "step": 11225
    },
    {
      "epoch": 3.5446795074202715,
      "grad_norm": 3.7398500442504883,
      "learning_rate": 1.6170795031927586e-05,
      "loss": 0.6885,
      "step": 11226
    },
    {
      "epoch": 3.5449952636564572,
      "grad_norm": 3.340686559677124,
      "learning_rate": 1.6167286506210093e-05,
      "loss": 0.6707,
      "step": 11227
    },
    {
      "epoch": 3.545311019892643,
      "grad_norm": 3.5352070331573486,
      "learning_rate": 1.6163777980492597e-05,
      "loss": 0.6138,
      "step": 11228
    },
    {
      "epoch": 3.5456267761288287,
      "grad_norm": 3.336604118347168,
      "learning_rate": 1.6160269454775103e-05,
      "loss": 0.6574,
      "step": 11229
    },
    {
      "epoch": 3.5459425323650144,
      "grad_norm": 3.525078773498535,
      "learning_rate": 1.615676092905761e-05,
      "loss": 0.6132,
      "step": 11230
    },
    {
      "epoch": 3.5462582886011997,
      "grad_norm": 3.538679361343384,
      "learning_rate": 1.6153252403340117e-05,
      "loss": 0.6655,
      "step": 11231
    },
    {
      "epoch": 3.5465740448373855,
      "grad_norm": 2.8886098861694336,
      "learning_rate": 1.6149743877622624e-05,
      "loss": 0.5934,
      "step": 11232
    },
    {
      "epoch": 3.546889801073571,
      "grad_norm": 3.6965553760528564,
      "learning_rate": 1.614623535190513e-05,
      "loss": 0.6937,
      "step": 11233
    },
    {
      "epoch": 3.547205557309757,
      "grad_norm": 3.2444405555725098,
      "learning_rate": 1.6142726826187637e-05,
      "loss": 0.6332,
      "step": 11234
    },
    {
      "epoch": 3.5475213135459427,
      "grad_norm": 3.8328499794006348,
      "learning_rate": 1.6139218300470144e-05,
      "loss": 0.7386,
      "step": 11235
    },
    {
      "epoch": 3.547837069782128,
      "grad_norm": 3.459275484085083,
      "learning_rate": 1.613570977475265e-05,
      "loss": 0.6283,
      "step": 11236
    },
    {
      "epoch": 3.5481528260183137,
      "grad_norm": 3.213700532913208,
      "learning_rate": 1.6132201249035154e-05,
      "loss": 0.5876,
      "step": 11237
    },
    {
      "epoch": 3.5484685822544995,
      "grad_norm": 3.495206356048584,
      "learning_rate": 1.612869272331766e-05,
      "loss": 0.6797,
      "step": 11238
    },
    {
      "epoch": 3.548784338490685,
      "grad_norm": 3.847261667251587,
      "learning_rate": 1.6125184197600168e-05,
      "loss": 0.6566,
      "step": 11239
    },
    {
      "epoch": 3.549100094726871,
      "grad_norm": 3.5034687519073486,
      "learning_rate": 1.6121675671882675e-05,
      "loss": 0.6952,
      "step": 11240
    },
    {
      "epoch": 3.5494158509630567,
      "grad_norm": 3.101628065109253,
      "learning_rate": 1.6118167146165182e-05,
      "loss": 0.6785,
      "step": 11241
    },
    {
      "epoch": 3.5497316071992424,
      "grad_norm": 3.452638626098633,
      "learning_rate": 1.611465862044769e-05,
      "loss": 0.7195,
      "step": 11242
    },
    {
      "epoch": 3.5500473634354277,
      "grad_norm": 3.2568440437316895,
      "learning_rate": 1.6111150094730195e-05,
      "loss": 0.6474,
      "step": 11243
    },
    {
      "epoch": 3.5503631196716134,
      "grad_norm": 3.3641769886016846,
      "learning_rate": 1.6107641569012702e-05,
      "loss": 0.6657,
      "step": 11244
    },
    {
      "epoch": 3.550678875907799,
      "grad_norm": 3.6918866634368896,
      "learning_rate": 1.610413304329521e-05,
      "loss": 0.6458,
      "step": 11245
    },
    {
      "epoch": 3.550994632143985,
      "grad_norm": 4.141355514526367,
      "learning_rate": 1.6100624517577716e-05,
      "loss": 0.7301,
      "step": 11246
    },
    {
      "epoch": 3.5513103883801707,
      "grad_norm": 3.275641679763794,
      "learning_rate": 1.609711599186022e-05,
      "loss": 0.735,
      "step": 11247
    },
    {
      "epoch": 3.551626144616356,
      "grad_norm": 3.3903017044067383,
      "learning_rate": 1.6093607466142726e-05,
      "loss": 0.6066,
      "step": 11248
    },
    {
      "epoch": 3.5519419008525417,
      "grad_norm": 5.425891876220703,
      "learning_rate": 1.6090098940425233e-05,
      "loss": 0.6649,
      "step": 11249
    },
    {
      "epoch": 3.5522576570887274,
      "grad_norm": 3.725731611251831,
      "learning_rate": 1.6086590414707743e-05,
      "loss": 0.6639,
      "step": 11250
    },
    {
      "epoch": 3.552573413324913,
      "grad_norm": 3.3732964992523193,
      "learning_rate": 1.6083081888990246e-05,
      "loss": 0.7042,
      "step": 11251
    },
    {
      "epoch": 3.552889169561099,
      "grad_norm": 3.309019088745117,
      "learning_rate": 1.6079573363272753e-05,
      "loss": 0.5797,
      "step": 11252
    },
    {
      "epoch": 3.5532049257972846,
      "grad_norm": 3.28131365776062,
      "learning_rate": 1.607606483755526e-05,
      "loss": 0.6967,
      "step": 11253
    },
    {
      "epoch": 3.5535206820334704,
      "grad_norm": 3.469968318939209,
      "learning_rate": 1.6072556311837767e-05,
      "loss": 0.5284,
      "step": 11254
    },
    {
      "epoch": 3.5538364382696557,
      "grad_norm": 3.006596565246582,
      "learning_rate": 1.6069047786120274e-05,
      "loss": 0.6949,
      "step": 11255
    },
    {
      "epoch": 3.5541521945058414,
      "grad_norm": 2.953831911087036,
      "learning_rate": 1.6065539260402777e-05,
      "loss": 0.5901,
      "step": 11256
    },
    {
      "epoch": 3.554467950742027,
      "grad_norm": 2.9684040546417236,
      "learning_rate": 1.6062030734685284e-05,
      "loss": 0.6685,
      "step": 11257
    },
    {
      "epoch": 3.554783706978213,
      "grad_norm": 2.981417655944824,
      "learning_rate": 1.605852220896779e-05,
      "loss": 0.6877,
      "step": 11258
    },
    {
      "epoch": 3.5550994632143986,
      "grad_norm": 3.1837751865386963,
      "learning_rate": 1.60550136832503e-05,
      "loss": 0.717,
      "step": 11259
    },
    {
      "epoch": 3.555415219450584,
      "grad_norm": 3.839395761489868,
      "learning_rate": 1.6051505157532808e-05,
      "loss": 0.6025,
      "step": 11260
    },
    {
      "epoch": 3.5557309756867697,
      "grad_norm": 2.973132610321045,
      "learning_rate": 1.604799663181531e-05,
      "loss": 0.5068,
      "step": 11261
    },
    {
      "epoch": 3.5560467319229554,
      "grad_norm": 2.8268003463745117,
      "learning_rate": 1.6044488106097818e-05,
      "loss": 0.6991,
      "step": 11262
    },
    {
      "epoch": 3.556362488159141,
      "grad_norm": 3.1941592693328857,
      "learning_rate": 1.6040979580380325e-05,
      "loss": 0.6221,
      "step": 11263
    },
    {
      "epoch": 3.556678244395327,
      "grad_norm": 3.536370038986206,
      "learning_rate": 1.603747105466283e-05,
      "loss": 0.5881,
      "step": 11264
    },
    {
      "epoch": 3.5569940006315126,
      "grad_norm": 2.792495012283325,
      "learning_rate": 1.603396252894534e-05,
      "loss": 0.5884,
      "step": 11265
    },
    {
      "epoch": 3.5573097568676983,
      "grad_norm": 2.9789533615112305,
      "learning_rate": 1.6030454003227842e-05,
      "loss": 0.6966,
      "step": 11266
    },
    {
      "epoch": 3.5576255131038836,
      "grad_norm": 3.65002179145813,
      "learning_rate": 1.6026945477510352e-05,
      "loss": 0.6358,
      "step": 11267
    },
    {
      "epoch": 3.5579412693400694,
      "grad_norm": 3.8390767574310303,
      "learning_rate": 1.602343695179286e-05,
      "loss": 0.6712,
      "step": 11268
    },
    {
      "epoch": 3.558257025576255,
      "grad_norm": 3.6071720123291016,
      "learning_rate": 1.6019928426075366e-05,
      "loss": 0.6216,
      "step": 11269
    },
    {
      "epoch": 3.558572781812441,
      "grad_norm": 3.4924862384796143,
      "learning_rate": 1.601641990035787e-05,
      "loss": 0.7431,
      "step": 11270
    },
    {
      "epoch": 3.5588885380486266,
      "grad_norm": 3.5291311740875244,
      "learning_rate": 1.6012911374640376e-05,
      "loss": 0.6679,
      "step": 11271
    },
    {
      "epoch": 3.559204294284812,
      "grad_norm": 3.5294511318206787,
      "learning_rate": 1.6009402848922883e-05,
      "loss": 0.6832,
      "step": 11272
    },
    {
      "epoch": 3.5595200505209976,
      "grad_norm": 3.1167821884155273,
      "learning_rate": 1.600589432320539e-05,
      "loss": 0.6515,
      "step": 11273
    },
    {
      "epoch": 3.5598358067571834,
      "grad_norm": 3.6394264698028564,
      "learning_rate": 1.6002385797487896e-05,
      "loss": 0.6408,
      "step": 11274
    },
    {
      "epoch": 3.560151562993369,
      "grad_norm": 3.2634506225585938,
      "learning_rate": 1.5998877271770403e-05,
      "loss": 0.6077,
      "step": 11275
    },
    {
      "epoch": 3.560467319229555,
      "grad_norm": 3.8405613899230957,
      "learning_rate": 1.599536874605291e-05,
      "loss": 0.7273,
      "step": 11276
    },
    {
      "epoch": 3.5607830754657406,
      "grad_norm": 3.4023172855377197,
      "learning_rate": 1.5991860220335417e-05,
      "loss": 0.6423,
      "step": 11277
    },
    {
      "epoch": 3.5610988317019263,
      "grad_norm": 3.5693166255950928,
      "learning_rate": 1.5988351694617924e-05,
      "loss": 0.6112,
      "step": 11278
    },
    {
      "epoch": 3.5614145879381116,
      "grad_norm": 3.636981725692749,
      "learning_rate": 1.598484316890043e-05,
      "loss": 0.646,
      "step": 11279
    },
    {
      "epoch": 3.5617303441742973,
      "grad_norm": 2.9732608795166016,
      "learning_rate": 1.5981334643182934e-05,
      "loss": 0.673,
      "step": 11280
    },
    {
      "epoch": 3.562046100410483,
      "grad_norm": 3.6868762969970703,
      "learning_rate": 1.597782611746544e-05,
      "loss": 0.6561,
      "step": 11281
    },
    {
      "epoch": 3.562361856646669,
      "grad_norm": 3.1099419593811035,
      "learning_rate": 1.5974317591747948e-05,
      "loss": 0.5759,
      "step": 11282
    },
    {
      "epoch": 3.5626776128828546,
      "grad_norm": 3.1751441955566406,
      "learning_rate": 1.5970809066030454e-05,
      "loss": 0.602,
      "step": 11283
    },
    {
      "epoch": 3.56299336911904,
      "grad_norm": 3.4569833278656006,
      "learning_rate": 1.596730054031296e-05,
      "loss": 0.7306,
      "step": 11284
    },
    {
      "epoch": 3.563309125355226,
      "grad_norm": 3.0437605381011963,
      "learning_rate": 1.5963792014595468e-05,
      "loss": 0.6177,
      "step": 11285
    },
    {
      "epoch": 3.5636248815914113,
      "grad_norm": 3.6844568252563477,
      "learning_rate": 1.5960283488877975e-05,
      "loss": 0.705,
      "step": 11286
    },
    {
      "epoch": 3.563940637827597,
      "grad_norm": 4.079415321350098,
      "learning_rate": 1.595677496316048e-05,
      "loss": 0.6896,
      "step": 11287
    },
    {
      "epoch": 3.564256394063783,
      "grad_norm": 3.4513721466064453,
      "learning_rate": 1.595326643744299e-05,
      "loss": 0.5776,
      "step": 11288
    },
    {
      "epoch": 3.5645721502999685,
      "grad_norm": 3.147061824798584,
      "learning_rate": 1.5949757911725495e-05,
      "loss": 0.7546,
      "step": 11289
    },
    {
      "epoch": 3.5648879065361543,
      "grad_norm": 3.8717305660247803,
      "learning_rate": 1.5946249386008e-05,
      "loss": 0.748,
      "step": 11290
    },
    {
      "epoch": 3.5652036627723396,
      "grad_norm": 2.9034435749053955,
      "learning_rate": 1.5942740860290505e-05,
      "loss": 0.6406,
      "step": 11291
    },
    {
      "epoch": 3.5655194190085253,
      "grad_norm": 3.5426533222198486,
      "learning_rate": 1.5939232334573012e-05,
      "loss": 0.69,
      "step": 11292
    },
    {
      "epoch": 3.565835175244711,
      "grad_norm": 3.3353867530822754,
      "learning_rate": 1.593572380885552e-05,
      "loss": 0.649,
      "step": 11293
    },
    {
      "epoch": 3.566150931480897,
      "grad_norm": 3.579448699951172,
      "learning_rate": 1.5932215283138026e-05,
      "loss": 0.6606,
      "step": 11294
    },
    {
      "epoch": 3.5664666877170825,
      "grad_norm": 3.3267719745635986,
      "learning_rate": 1.5928706757420533e-05,
      "loss": 0.6442,
      "step": 11295
    },
    {
      "epoch": 3.566782443953268,
      "grad_norm": 3.1840081214904785,
      "learning_rate": 1.592519823170304e-05,
      "loss": 0.663,
      "step": 11296
    },
    {
      "epoch": 3.567098200189454,
      "grad_norm": 3.2759850025177,
      "learning_rate": 1.5921689705985546e-05,
      "loss": 0.6484,
      "step": 11297
    },
    {
      "epoch": 3.5674139564256393,
      "grad_norm": 3.038137912750244,
      "learning_rate": 1.5918181180268053e-05,
      "loss": 0.6063,
      "step": 11298
    },
    {
      "epoch": 3.567729712661825,
      "grad_norm": 3.457204580307007,
      "learning_rate": 1.5914672654550557e-05,
      "loss": 0.666,
      "step": 11299
    },
    {
      "epoch": 3.5680454688980108,
      "grad_norm": 3.019976854324341,
      "learning_rate": 1.5911164128833063e-05,
      "loss": 0.5871,
      "step": 11300
    },
    {
      "epoch": 3.5683612251341965,
      "grad_norm": 3.121910333633423,
      "learning_rate": 1.590765560311557e-05,
      "loss": 0.6234,
      "step": 11301
    },
    {
      "epoch": 3.5686769813703823,
      "grad_norm": 2.729423999786377,
      "learning_rate": 1.590414707739808e-05,
      "loss": 0.6918,
      "step": 11302
    },
    {
      "epoch": 3.5689927376065675,
      "grad_norm": 3.1678550243377686,
      "learning_rate": 1.5900638551680587e-05,
      "loss": 0.6605,
      "step": 11303
    },
    {
      "epoch": 3.5693084938427533,
      "grad_norm": 3.2394046783447266,
      "learning_rate": 1.589713002596309e-05,
      "loss": 0.6221,
      "step": 11304
    },
    {
      "epoch": 3.569624250078939,
      "grad_norm": 3.0176870822906494,
      "learning_rate": 1.5893621500245597e-05,
      "loss": 0.7348,
      "step": 11305
    },
    {
      "epoch": 3.5699400063151248,
      "grad_norm": 3.4487855434417725,
      "learning_rate": 1.5890112974528104e-05,
      "loss": 0.7217,
      "step": 11306
    },
    {
      "epoch": 3.5702557625513105,
      "grad_norm": 3.1449103355407715,
      "learning_rate": 1.588660444881061e-05,
      "loss": 0.6233,
      "step": 11307
    },
    {
      "epoch": 3.570571518787496,
      "grad_norm": 3.5192131996154785,
      "learning_rate": 1.5883095923093118e-05,
      "loss": 0.7057,
      "step": 11308
    },
    {
      "epoch": 3.570887275023682,
      "grad_norm": 3.4298815727233887,
      "learning_rate": 1.587958739737562e-05,
      "loss": 0.6081,
      "step": 11309
    },
    {
      "epoch": 3.5712030312598673,
      "grad_norm": 3.4708809852600098,
      "learning_rate": 1.5876078871658128e-05,
      "loss": 0.6561,
      "step": 11310
    },
    {
      "epoch": 3.571518787496053,
      "grad_norm": 3.805821418762207,
      "learning_rate": 1.587257034594064e-05,
      "loss": 0.6809,
      "step": 11311
    },
    {
      "epoch": 3.5718345437322387,
      "grad_norm": 3.044438362121582,
      "learning_rate": 1.5869061820223145e-05,
      "loss": 0.7086,
      "step": 11312
    },
    {
      "epoch": 3.5721502999684245,
      "grad_norm": 3.2095158100128174,
      "learning_rate": 1.586555329450565e-05,
      "loss": 0.606,
      "step": 11313
    },
    {
      "epoch": 3.57246605620461,
      "grad_norm": 3.416576385498047,
      "learning_rate": 1.5862044768788155e-05,
      "loss": 0.6422,
      "step": 11314
    },
    {
      "epoch": 3.5727818124407955,
      "grad_norm": 2.9619157314300537,
      "learning_rate": 1.5858536243070662e-05,
      "loss": 0.6557,
      "step": 11315
    },
    {
      "epoch": 3.5730975686769813,
      "grad_norm": 3.2083566188812256,
      "learning_rate": 1.585502771735317e-05,
      "loss": 0.6729,
      "step": 11316
    },
    {
      "epoch": 3.573413324913167,
      "grad_norm": 3.3169500827789307,
      "learning_rate": 1.5851519191635676e-05,
      "loss": 0.6489,
      "step": 11317
    },
    {
      "epoch": 3.5737290811493527,
      "grad_norm": 3.7348999977111816,
      "learning_rate": 1.5848010665918183e-05,
      "loss": 0.6721,
      "step": 11318
    },
    {
      "epoch": 3.5740448373855385,
      "grad_norm": 2.3872036933898926,
      "learning_rate": 1.5844502140200686e-05,
      "loss": 0.4829,
      "step": 11319
    },
    {
      "epoch": 3.5743605936217238,
      "grad_norm": 3.3146653175354004,
      "learning_rate": 1.5840993614483196e-05,
      "loss": 0.8304,
      "step": 11320
    },
    {
      "epoch": 3.57467634985791,
      "grad_norm": 3.600048780441284,
      "learning_rate": 1.5837485088765703e-05,
      "loss": 0.7052,
      "step": 11321
    },
    {
      "epoch": 3.5749921060940952,
      "grad_norm": 3.012051820755005,
      "learning_rate": 1.583397656304821e-05,
      "loss": 0.6529,
      "step": 11322
    },
    {
      "epoch": 3.575307862330281,
      "grad_norm": 3.2775518894195557,
      "learning_rate": 1.5830468037330713e-05,
      "loss": 0.6761,
      "step": 11323
    },
    {
      "epoch": 3.5756236185664667,
      "grad_norm": 3.162821054458618,
      "learning_rate": 1.582695951161322e-05,
      "loss": 0.6683,
      "step": 11324
    },
    {
      "epoch": 3.5759393748026524,
      "grad_norm": 3.6653144359588623,
      "learning_rate": 1.5823450985895727e-05,
      "loss": 0.6964,
      "step": 11325
    },
    {
      "epoch": 3.576255131038838,
      "grad_norm": 3.457273483276367,
      "learning_rate": 1.5819942460178234e-05,
      "loss": 0.6114,
      "step": 11326
    },
    {
      "epoch": 3.5765708872750235,
      "grad_norm": 3.009519577026367,
      "learning_rate": 1.581643393446074e-05,
      "loss": 0.6967,
      "step": 11327
    },
    {
      "epoch": 3.576886643511209,
      "grad_norm": 3.810201644897461,
      "learning_rate": 1.5812925408743244e-05,
      "loss": 0.7076,
      "step": 11328
    },
    {
      "epoch": 3.577202399747395,
      "grad_norm": 3.2101657390594482,
      "learning_rate": 1.5809416883025754e-05,
      "loss": 0.7278,
      "step": 11329
    },
    {
      "epoch": 3.5775181559835807,
      "grad_norm": 3.3454291820526123,
      "learning_rate": 1.580590835730826e-05,
      "loss": 0.6686,
      "step": 11330
    },
    {
      "epoch": 3.5778339122197664,
      "grad_norm": 3.1581950187683105,
      "learning_rate": 1.5802399831590768e-05,
      "loss": 0.6639,
      "step": 11331
    },
    {
      "epoch": 3.578149668455952,
      "grad_norm": 3.3738510608673096,
      "learning_rate": 1.5798891305873275e-05,
      "loss": 0.6442,
      "step": 11332
    },
    {
      "epoch": 3.578465424692138,
      "grad_norm": 3.4143290519714355,
      "learning_rate": 1.5795382780155778e-05,
      "loss": 0.6182,
      "step": 11333
    },
    {
      "epoch": 3.578781180928323,
      "grad_norm": 2.9932422637939453,
      "learning_rate": 1.5791874254438285e-05,
      "loss": 0.563,
      "step": 11334
    },
    {
      "epoch": 3.579096937164509,
      "grad_norm": 3.8473758697509766,
      "learning_rate": 1.578836572872079e-05,
      "loss": 0.7287,
      "step": 11335
    },
    {
      "epoch": 3.5794126934006947,
      "grad_norm": 3.6068954467773438,
      "learning_rate": 1.57848572030033e-05,
      "loss": 0.6687,
      "step": 11336
    },
    {
      "epoch": 3.5797284496368804,
      "grad_norm": 3.4944963455200195,
      "learning_rate": 1.5781348677285805e-05,
      "loss": 0.6403,
      "step": 11337
    },
    {
      "epoch": 3.580044205873066,
      "grad_norm": 3.606445789337158,
      "learning_rate": 1.5777840151568312e-05,
      "loss": 0.6914,
      "step": 11338
    },
    {
      "epoch": 3.5803599621092514,
      "grad_norm": 3.3470144271850586,
      "learning_rate": 1.577433162585082e-05,
      "loss": 0.6242,
      "step": 11339
    },
    {
      "epoch": 3.580675718345437,
      "grad_norm": 3.3678510189056396,
      "learning_rate": 1.5770823100133326e-05,
      "loss": 0.661,
      "step": 11340
    },
    {
      "epoch": 3.580991474581623,
      "grad_norm": 3.872586727142334,
      "learning_rate": 1.5767314574415833e-05,
      "loss": 0.6572,
      "step": 11341
    },
    {
      "epoch": 3.5813072308178087,
      "grad_norm": 3.1426074504852295,
      "learning_rate": 1.5763806048698336e-05,
      "loss": 0.6102,
      "step": 11342
    },
    {
      "epoch": 3.5816229870539944,
      "grad_norm": 2.8549506664276123,
      "learning_rate": 1.5760297522980843e-05,
      "loss": 0.5757,
      "step": 11343
    },
    {
      "epoch": 3.58193874329018,
      "grad_norm": 2.9890730381011963,
      "learning_rate": 1.575678899726335e-05,
      "loss": 0.6395,
      "step": 11344
    },
    {
      "epoch": 3.582254499526366,
      "grad_norm": 3.2396419048309326,
      "learning_rate": 1.5753280471545856e-05,
      "loss": 0.6264,
      "step": 11345
    },
    {
      "epoch": 3.582570255762551,
      "grad_norm": 4.537791728973389,
      "learning_rate": 1.5749771945828363e-05,
      "loss": 0.7243,
      "step": 11346
    },
    {
      "epoch": 3.582886011998737,
      "grad_norm": 3.1564078330993652,
      "learning_rate": 1.574626342011087e-05,
      "loss": 0.6247,
      "step": 11347
    },
    {
      "epoch": 3.5832017682349226,
      "grad_norm": 3.0953829288482666,
      "learning_rate": 1.5742754894393377e-05,
      "loss": 0.6045,
      "step": 11348
    },
    {
      "epoch": 3.5835175244711084,
      "grad_norm": 3.5581164360046387,
      "learning_rate": 1.5739246368675884e-05,
      "loss": 0.6575,
      "step": 11349
    },
    {
      "epoch": 3.583833280707294,
      "grad_norm": 3.7783610820770264,
      "learning_rate": 1.573573784295839e-05,
      "loss": 0.5878,
      "step": 11350
    },
    {
      "epoch": 3.5841490369434794,
      "grad_norm": 3.1103744506835938,
      "learning_rate": 1.5732229317240897e-05,
      "loss": 0.6262,
      "step": 11351
    },
    {
      "epoch": 3.584464793179665,
      "grad_norm": 3.5473015308380127,
      "learning_rate": 1.57287207915234e-05,
      "loss": 0.6468,
      "step": 11352
    },
    {
      "epoch": 3.584780549415851,
      "grad_norm": 3.482638359069824,
      "learning_rate": 1.5725212265805908e-05,
      "loss": 0.6635,
      "step": 11353
    },
    {
      "epoch": 3.5850963056520366,
      "grad_norm": 3.274735689163208,
      "learning_rate": 1.5721703740088414e-05,
      "loss": 0.5863,
      "step": 11354
    },
    {
      "epoch": 3.5854120618882224,
      "grad_norm": 4.327759742736816,
      "learning_rate": 1.5718195214370925e-05,
      "loss": 0.7579,
      "step": 11355
    },
    {
      "epoch": 3.585727818124408,
      "grad_norm": 3.201484441757202,
      "learning_rate": 1.5714686688653428e-05,
      "loss": 0.613,
      "step": 11356
    },
    {
      "epoch": 3.586043574360594,
      "grad_norm": 3.597480535507202,
      "learning_rate": 1.5711178162935935e-05,
      "loss": 0.6861,
      "step": 11357
    },
    {
      "epoch": 3.586359330596779,
      "grad_norm": 3.4165804386138916,
      "learning_rate": 1.570766963721844e-05,
      "loss": 0.7168,
      "step": 11358
    },
    {
      "epoch": 3.586675086832965,
      "grad_norm": 3.2366394996643066,
      "learning_rate": 1.570416111150095e-05,
      "loss": 0.6314,
      "step": 11359
    },
    {
      "epoch": 3.5869908430691506,
      "grad_norm": 3.3451075553894043,
      "learning_rate": 1.5700652585783455e-05,
      "loss": 0.7452,
      "step": 11360
    },
    {
      "epoch": 3.5873065993053364,
      "grad_norm": 2.9148411750793457,
      "learning_rate": 1.569714406006596e-05,
      "loss": 0.7164,
      "step": 11361
    },
    {
      "epoch": 3.587622355541522,
      "grad_norm": 2.8269171714782715,
      "learning_rate": 1.5693635534348465e-05,
      "loss": 0.6334,
      "step": 11362
    },
    {
      "epoch": 3.5879381117777074,
      "grad_norm": 2.709667444229126,
      "learning_rate": 1.5690127008630976e-05,
      "loss": 0.6644,
      "step": 11363
    },
    {
      "epoch": 3.588253868013893,
      "grad_norm": 3.2296032905578613,
      "learning_rate": 1.5686618482913482e-05,
      "loss": 0.626,
      "step": 11364
    },
    {
      "epoch": 3.588569624250079,
      "grad_norm": 3.075040340423584,
      "learning_rate": 1.568310995719599e-05,
      "loss": 0.7005,
      "step": 11365
    },
    {
      "epoch": 3.5888853804862646,
      "grad_norm": 3.3735668659210205,
      "learning_rate": 1.5679601431478493e-05,
      "loss": 0.5987,
      "step": 11366
    },
    {
      "epoch": 3.5892011367224503,
      "grad_norm": 4.618563175201416,
      "learning_rate": 1.5676092905761e-05,
      "loss": 0.7489,
      "step": 11367
    },
    {
      "epoch": 3.589516892958636,
      "grad_norm": 3.2954277992248535,
      "learning_rate": 1.5672584380043506e-05,
      "loss": 0.6448,
      "step": 11368
    },
    {
      "epoch": 3.589832649194822,
      "grad_norm": 2.7883388996124268,
      "learning_rate": 1.5669075854326013e-05,
      "loss": 0.6181,
      "step": 11369
    },
    {
      "epoch": 3.590148405431007,
      "grad_norm": 3.0740926265716553,
      "learning_rate": 1.566556732860852e-05,
      "loss": 0.6295,
      "step": 11370
    },
    {
      "epoch": 3.590464161667193,
      "grad_norm": 3.482356071472168,
      "learning_rate": 1.5662058802891023e-05,
      "loss": 0.6521,
      "step": 11371
    },
    {
      "epoch": 3.5907799179033786,
      "grad_norm": 3.6274755001068115,
      "learning_rate": 1.5658550277173534e-05,
      "loss": 0.7017,
      "step": 11372
    },
    {
      "epoch": 3.5910956741395643,
      "grad_norm": 3.22469425201416,
      "learning_rate": 1.565504175145604e-05,
      "loss": 0.6207,
      "step": 11373
    },
    {
      "epoch": 3.59141143037575,
      "grad_norm": 3.381944179534912,
      "learning_rate": 1.5651533225738547e-05,
      "loss": 0.6394,
      "step": 11374
    },
    {
      "epoch": 3.5917271866119354,
      "grad_norm": 3.0559654235839844,
      "learning_rate": 1.564802470002105e-05,
      "loss": 0.6519,
      "step": 11375
    },
    {
      "epoch": 3.592042942848121,
      "grad_norm": 3.4605252742767334,
      "learning_rate": 1.5644516174303557e-05,
      "loss": 0.6037,
      "step": 11376
    },
    {
      "epoch": 3.592358699084307,
      "grad_norm": 3.0419247150421143,
      "learning_rate": 1.5641007648586064e-05,
      "loss": 0.6736,
      "step": 11377
    },
    {
      "epoch": 3.5926744553204926,
      "grad_norm": 3.0725438594818115,
      "learning_rate": 1.563749912286857e-05,
      "loss": 0.6079,
      "step": 11378
    },
    {
      "epoch": 3.5929902115566783,
      "grad_norm": 2.733231544494629,
      "learning_rate": 1.5633990597151078e-05,
      "loss": 0.6041,
      "step": 11379
    },
    {
      "epoch": 3.593305967792864,
      "grad_norm": 3.174769878387451,
      "learning_rate": 1.5630482071433585e-05,
      "loss": 0.5974,
      "step": 11380
    },
    {
      "epoch": 3.59362172402905,
      "grad_norm": 3.12007474899292,
      "learning_rate": 1.562697354571609e-05,
      "loss": 0.5867,
      "step": 11381
    },
    {
      "epoch": 3.593937480265235,
      "grad_norm": 2.9222335815429688,
      "learning_rate": 1.56234650199986e-05,
      "loss": 0.627,
      "step": 11382
    },
    {
      "epoch": 3.594253236501421,
      "grad_norm": 3.7436373233795166,
      "learning_rate": 1.5619956494281105e-05,
      "loss": 0.6523,
      "step": 11383
    },
    {
      "epoch": 3.5945689927376065,
      "grad_norm": 3.6928963661193848,
      "learning_rate": 1.5616447968563612e-05,
      "loss": 0.7212,
      "step": 11384
    },
    {
      "epoch": 3.5948847489737923,
      "grad_norm": 3.2166881561279297,
      "learning_rate": 1.5612939442846115e-05,
      "loss": 0.6781,
      "step": 11385
    },
    {
      "epoch": 3.595200505209978,
      "grad_norm": 3.460841655731201,
      "learning_rate": 1.5609430917128622e-05,
      "loss": 0.6528,
      "step": 11386
    },
    {
      "epoch": 3.5955162614461633,
      "grad_norm": 2.9865381717681885,
      "learning_rate": 1.560592239141113e-05,
      "loss": 0.7173,
      "step": 11387
    },
    {
      "epoch": 3.5958320176823495,
      "grad_norm": 3.276909828186035,
      "learning_rate": 1.5602413865693636e-05,
      "loss": 0.6553,
      "step": 11388
    },
    {
      "epoch": 3.596147773918535,
      "grad_norm": 3.460465431213379,
      "learning_rate": 1.5598905339976143e-05,
      "loss": 0.5831,
      "step": 11389
    },
    {
      "epoch": 3.5964635301547205,
      "grad_norm": 3.6629621982574463,
      "learning_rate": 1.559539681425865e-05,
      "loss": 0.678,
      "step": 11390
    },
    {
      "epoch": 3.5967792863909063,
      "grad_norm": 3.531895160675049,
      "learning_rate": 1.5591888288541156e-05,
      "loss": 0.7038,
      "step": 11391
    },
    {
      "epoch": 3.597095042627092,
      "grad_norm": 3.2570197582244873,
      "learning_rate": 1.5588379762823663e-05,
      "loss": 0.6921,
      "step": 11392
    },
    {
      "epoch": 3.5974107988632777,
      "grad_norm": 3.300534725189209,
      "learning_rate": 1.558487123710617e-05,
      "loss": 0.6777,
      "step": 11393
    },
    {
      "epoch": 3.597726555099463,
      "grad_norm": 3.8221333026885986,
      "learning_rate": 1.5581362711388677e-05,
      "loss": 0.7571,
      "step": 11394
    },
    {
      "epoch": 3.598042311335649,
      "grad_norm": 3.3482561111450195,
      "learning_rate": 1.557785418567118e-05,
      "loss": 0.6999,
      "step": 11395
    },
    {
      "epoch": 3.5983580675718345,
      "grad_norm": 3.3909711837768555,
      "learning_rate": 1.5574345659953687e-05,
      "loss": 0.6101,
      "step": 11396
    },
    {
      "epoch": 3.5986738238080203,
      "grad_norm": 3.585820198059082,
      "learning_rate": 1.5570837134236194e-05,
      "loss": 0.6938,
      "step": 11397
    },
    {
      "epoch": 3.598989580044206,
      "grad_norm": 3.412998914718628,
      "learning_rate": 1.5567328608518704e-05,
      "loss": 0.7756,
      "step": 11398
    },
    {
      "epoch": 3.5993053362803913,
      "grad_norm": 4.481986999511719,
      "learning_rate": 1.5563820082801207e-05,
      "loss": 0.6622,
      "step": 11399
    },
    {
      "epoch": 3.5996210925165775,
      "grad_norm": 3.0723612308502197,
      "learning_rate": 1.5560311557083714e-05,
      "loss": 0.6967,
      "step": 11400
    },
    {
      "epoch": 3.5996210925165775,
      "eval_loss": 0.8081884980201721,
      "eval_runtime": 4.6112,
      "eval_samples_per_second": 539.987,
      "eval_steps_per_second": 4.554,
      "step": 11400
    },
    {
      "epoch": 3.5999368487527628,
      "grad_norm": 3.9348950386047363,
      "learning_rate": 1.555680303136622e-05,
      "loss": 0.7247,
      "step": 11401
    },
    {
      "epoch": 3.6002526049889485,
      "grad_norm": 3.2316229343414307,
      "learning_rate": 1.5553294505648728e-05,
      "loss": 0.6283,
      "step": 11402
    },
    {
      "epoch": 3.6005683612251342,
      "grad_norm": 2.932386636734009,
      "learning_rate": 1.5549785979931235e-05,
      "loss": 0.6451,
      "step": 11403
    },
    {
      "epoch": 3.60088411746132,
      "grad_norm": 3.9553725719451904,
      "learning_rate": 1.5546277454213738e-05,
      "loss": 0.6552,
      "step": 11404
    },
    {
      "epoch": 3.6011998736975057,
      "grad_norm": 3.1594958305358887,
      "learning_rate": 1.5542768928496245e-05,
      "loss": 0.5748,
      "step": 11405
    },
    {
      "epoch": 3.601515629933691,
      "grad_norm": 3.5878262519836426,
      "learning_rate": 1.553926040277875e-05,
      "loss": 0.6325,
      "step": 11406
    },
    {
      "epoch": 3.6018313861698767,
      "grad_norm": 3.4313149452209473,
      "learning_rate": 1.5535751877061262e-05,
      "loss": 0.6724,
      "step": 11407
    },
    {
      "epoch": 3.6021471424060625,
      "grad_norm": 4.070974826812744,
      "learning_rate": 1.553224335134377e-05,
      "loss": 0.7181,
      "step": 11408
    },
    {
      "epoch": 3.6024628986422482,
      "grad_norm": 2.852520227432251,
      "learning_rate": 1.5528734825626272e-05,
      "loss": 0.6077,
      "step": 11409
    },
    {
      "epoch": 3.602778654878434,
      "grad_norm": 3.589107036590576,
      "learning_rate": 1.552522629990878e-05,
      "loss": 0.6204,
      "step": 11410
    },
    {
      "epoch": 3.6030944111146193,
      "grad_norm": 3.405235767364502,
      "learning_rate": 1.5521717774191286e-05,
      "loss": 0.6782,
      "step": 11411
    },
    {
      "epoch": 3.6034101673508054,
      "grad_norm": 3.539077043533325,
      "learning_rate": 1.5518209248473793e-05,
      "loss": 0.7507,
      "step": 11412
    },
    {
      "epoch": 3.6037259235869907,
      "grad_norm": 3.615304946899414,
      "learning_rate": 1.55147007227563e-05,
      "loss": 0.6395,
      "step": 11413
    },
    {
      "epoch": 3.6040416798231765,
      "grad_norm": 3.5092382431030273,
      "learning_rate": 1.5511192197038803e-05,
      "loss": 0.6358,
      "step": 11414
    },
    {
      "epoch": 3.604357436059362,
      "grad_norm": 3.0722320079803467,
      "learning_rate": 1.550768367132131e-05,
      "loss": 0.646,
      "step": 11415
    },
    {
      "epoch": 3.604673192295548,
      "grad_norm": 3.7937381267547607,
      "learning_rate": 1.550417514560382e-05,
      "loss": 0.6757,
      "step": 11416
    },
    {
      "epoch": 3.6049889485317337,
      "grad_norm": 3.7068309783935547,
      "learning_rate": 1.5500666619886327e-05,
      "loss": 0.6896,
      "step": 11417
    },
    {
      "epoch": 3.605304704767919,
      "grad_norm": 4.192250728607178,
      "learning_rate": 1.549715809416883e-05,
      "loss": 0.6652,
      "step": 11418
    },
    {
      "epoch": 3.6056204610041047,
      "grad_norm": 3.9750335216522217,
      "learning_rate": 1.5493649568451337e-05,
      "loss": 0.7669,
      "step": 11419
    },
    {
      "epoch": 3.6059362172402905,
      "grad_norm": 3.7463300228118896,
      "learning_rate": 1.5490141042733844e-05,
      "loss": 0.7405,
      "step": 11420
    },
    {
      "epoch": 3.606251973476476,
      "grad_norm": 3.162339448928833,
      "learning_rate": 1.548663251701635e-05,
      "loss": 0.5866,
      "step": 11421
    },
    {
      "epoch": 3.606567729712662,
      "grad_norm": 3.5329713821411133,
      "learning_rate": 1.5483123991298857e-05,
      "loss": 0.612,
      "step": 11422
    },
    {
      "epoch": 3.6068834859488472,
      "grad_norm": 3.2393906116485596,
      "learning_rate": 1.5479615465581364e-05,
      "loss": 0.6385,
      "step": 11423
    },
    {
      "epoch": 3.6071992421850334,
      "grad_norm": 4.197107315063477,
      "learning_rate": 1.5476106939863868e-05,
      "loss": 0.6217,
      "step": 11424
    },
    {
      "epoch": 3.6075149984212187,
      "grad_norm": 3.212458610534668,
      "learning_rate": 1.5472598414146378e-05,
      "loss": 0.6742,
      "step": 11425
    },
    {
      "epoch": 3.6078307546574044,
      "grad_norm": 3.1798324584960938,
      "learning_rate": 1.5469089888428885e-05,
      "loss": 0.6954,
      "step": 11426
    },
    {
      "epoch": 3.60814651089359,
      "grad_norm": 3.2270328998565674,
      "learning_rate": 1.546558136271139e-05,
      "loss": 0.6442,
      "step": 11427
    },
    {
      "epoch": 3.608462267129776,
      "grad_norm": 3.6869888305664062,
      "learning_rate": 1.5462072836993895e-05,
      "loss": 0.6657,
      "step": 11428
    },
    {
      "epoch": 3.6087780233659617,
      "grad_norm": 2.9688773155212402,
      "learning_rate": 1.54585643112764e-05,
      "loss": 0.6632,
      "step": 11429
    },
    {
      "epoch": 3.609093779602147,
      "grad_norm": 3.050793409347534,
      "learning_rate": 1.545505578555891e-05,
      "loss": 0.6655,
      "step": 11430
    },
    {
      "epoch": 3.6094095358383327,
      "grad_norm": 2.923922300338745,
      "learning_rate": 1.5451547259841415e-05,
      "loss": 0.5758,
      "step": 11431
    },
    {
      "epoch": 3.6097252920745184,
      "grad_norm": 3.3755393028259277,
      "learning_rate": 1.5448038734123922e-05,
      "loss": 0.653,
      "step": 11432
    },
    {
      "epoch": 3.610041048310704,
      "grad_norm": 2.784141778945923,
      "learning_rate": 1.544453020840643e-05,
      "loss": 0.5773,
      "step": 11433
    },
    {
      "epoch": 3.61035680454689,
      "grad_norm": 3.781937599182129,
      "learning_rate": 1.5441021682688936e-05,
      "loss": 0.7014,
      "step": 11434
    },
    {
      "epoch": 3.610672560783075,
      "grad_norm": 3.7478597164154053,
      "learning_rate": 1.5437513156971442e-05,
      "loss": 0.6718,
      "step": 11435
    },
    {
      "epoch": 3.6109883170192614,
      "grad_norm": 3.4649882316589355,
      "learning_rate": 1.543400463125395e-05,
      "loss": 0.6846,
      "step": 11436
    },
    {
      "epoch": 3.6113040732554467,
      "grad_norm": 2.926755666732788,
      "learning_rate": 1.5430496105536456e-05,
      "loss": 0.645,
      "step": 11437
    },
    {
      "epoch": 3.6116198294916324,
      "grad_norm": 3.5817978382110596,
      "learning_rate": 1.542698757981896e-05,
      "loss": 0.6633,
      "step": 11438
    },
    {
      "epoch": 3.611935585727818,
      "grad_norm": 3.8372254371643066,
      "learning_rate": 1.5423479054101466e-05,
      "loss": 0.6792,
      "step": 11439
    },
    {
      "epoch": 3.612251341964004,
      "grad_norm": 3.3355231285095215,
      "learning_rate": 1.5419970528383973e-05,
      "loss": 0.6555,
      "step": 11440
    },
    {
      "epoch": 3.6125670982001896,
      "grad_norm": 3.4222311973571777,
      "learning_rate": 1.541646200266648e-05,
      "loss": 0.7234,
      "step": 11441
    },
    {
      "epoch": 3.612882854436375,
      "grad_norm": 2.9169929027557373,
      "learning_rate": 1.5412953476948987e-05,
      "loss": 0.5546,
      "step": 11442
    },
    {
      "epoch": 3.6131986106725607,
      "grad_norm": 3.476050615310669,
      "learning_rate": 1.5409444951231494e-05,
      "loss": 0.6343,
      "step": 11443
    },
    {
      "epoch": 3.6135143669087464,
      "grad_norm": 3.6316516399383545,
      "learning_rate": 1.5405936425514e-05,
      "loss": 0.6332,
      "step": 11444
    },
    {
      "epoch": 3.613830123144932,
      "grad_norm": 3.4741995334625244,
      "learning_rate": 1.5402427899796507e-05,
      "loss": 0.6581,
      "step": 11445
    },
    {
      "epoch": 3.614145879381118,
      "grad_norm": 3.5796444416046143,
      "learning_rate": 1.5398919374079014e-05,
      "loss": 0.7157,
      "step": 11446
    },
    {
      "epoch": 3.6144616356173036,
      "grad_norm": 3.256854295730591,
      "learning_rate": 1.5395410848361517e-05,
      "loss": 0.6209,
      "step": 11447
    },
    {
      "epoch": 3.6147773918534893,
      "grad_norm": 3.703381299972534,
      "learning_rate": 1.5391902322644024e-05,
      "loss": 0.6734,
      "step": 11448
    },
    {
      "epoch": 3.6150931480896746,
      "grad_norm": 3.382267713546753,
      "learning_rate": 1.538839379692653e-05,
      "loss": 0.7245,
      "step": 11449
    },
    {
      "epoch": 3.6154089043258604,
      "grad_norm": 3.0131070613861084,
      "learning_rate": 1.5384885271209038e-05,
      "loss": 0.6398,
      "step": 11450
    },
    {
      "epoch": 3.615724660562046,
      "grad_norm": 3.586822271347046,
      "learning_rate": 1.5381376745491545e-05,
      "loss": 0.5853,
      "step": 11451
    },
    {
      "epoch": 3.616040416798232,
      "grad_norm": 3.628537178039551,
      "learning_rate": 1.537786821977405e-05,
      "loss": 0.8405,
      "step": 11452
    },
    {
      "epoch": 3.6163561730344176,
      "grad_norm": 3.429979085922241,
      "learning_rate": 1.537435969405656e-05,
      "loss": 0.6549,
      "step": 11453
    },
    {
      "epoch": 3.616671929270603,
      "grad_norm": 3.7804856300354004,
      "learning_rate": 1.5370851168339065e-05,
      "loss": 0.714,
      "step": 11454
    },
    {
      "epoch": 3.6169876855067886,
      "grad_norm": 3.1204335689544678,
      "learning_rate": 1.5367342642621572e-05,
      "loss": 0.6243,
      "step": 11455
    },
    {
      "epoch": 3.6173034417429744,
      "grad_norm": 3.772275924682617,
      "learning_rate": 1.536383411690408e-05,
      "loss": 0.7176,
      "step": 11456
    },
    {
      "epoch": 3.61761919797916,
      "grad_norm": 4.288338661193848,
      "learning_rate": 1.5360325591186582e-05,
      "loss": 0.6907,
      "step": 11457
    },
    {
      "epoch": 3.617934954215346,
      "grad_norm": 3.6600444316864014,
      "learning_rate": 1.535681706546909e-05,
      "loss": 0.6788,
      "step": 11458
    },
    {
      "epoch": 3.6182507104515316,
      "grad_norm": 3.9698517322540283,
      "learning_rate": 1.5353308539751596e-05,
      "loss": 0.6158,
      "step": 11459
    },
    {
      "epoch": 3.6185664666877173,
      "grad_norm": 3.4142916202545166,
      "learning_rate": 1.5349800014034106e-05,
      "loss": 0.663,
      "step": 11460
    },
    {
      "epoch": 3.6188822229239026,
      "grad_norm": 3.0470166206359863,
      "learning_rate": 1.534629148831661e-05,
      "loss": 0.6374,
      "step": 11461
    },
    {
      "epoch": 3.6191979791600883,
      "grad_norm": 3.1018598079681396,
      "learning_rate": 1.5342782962599116e-05,
      "loss": 0.6451,
      "step": 11462
    },
    {
      "epoch": 3.619513735396274,
      "grad_norm": 3.013949155807495,
      "learning_rate": 1.5339274436881623e-05,
      "loss": 0.6728,
      "step": 11463
    },
    {
      "epoch": 3.61982949163246,
      "grad_norm": 3.7572884559631348,
      "learning_rate": 1.533576591116413e-05,
      "loss": 0.5997,
      "step": 11464
    },
    {
      "epoch": 3.6201452478686456,
      "grad_norm": 3.4742424488067627,
      "learning_rate": 1.5332257385446637e-05,
      "loss": 0.5611,
      "step": 11465
    },
    {
      "epoch": 3.620461004104831,
      "grad_norm": 3.857118606567383,
      "learning_rate": 1.532874885972914e-05,
      "loss": 0.721,
      "step": 11466
    },
    {
      "epoch": 3.6207767603410166,
      "grad_norm": 3.368194580078125,
      "learning_rate": 1.5325240334011647e-05,
      "loss": 0.6284,
      "step": 11467
    },
    {
      "epoch": 3.6210925165772023,
      "grad_norm": 2.6883840560913086,
      "learning_rate": 1.5321731808294157e-05,
      "loss": 0.5951,
      "step": 11468
    },
    {
      "epoch": 3.621408272813388,
      "grad_norm": 2.940788507461548,
      "learning_rate": 1.5318223282576664e-05,
      "loss": 0.5526,
      "step": 11469
    },
    {
      "epoch": 3.621724029049574,
      "grad_norm": 3.3932039737701416,
      "learning_rate": 1.531471475685917e-05,
      "loss": 0.5592,
      "step": 11470
    },
    {
      "epoch": 3.6220397852857595,
      "grad_norm": 3.3074798583984375,
      "learning_rate": 1.5311206231141674e-05,
      "loss": 0.7208,
      "step": 11471
    },
    {
      "epoch": 3.6223555415219453,
      "grad_norm": 3.310594081878662,
      "learning_rate": 1.530769770542418e-05,
      "loss": 0.6336,
      "step": 11472
    },
    {
      "epoch": 3.6226712977581306,
      "grad_norm": 3.1956546306610107,
      "learning_rate": 1.5304189179706688e-05,
      "loss": 0.5287,
      "step": 11473
    },
    {
      "epoch": 3.6229870539943163,
      "grad_norm": 3.059802293777466,
      "learning_rate": 1.5300680653989195e-05,
      "loss": 0.6883,
      "step": 11474
    },
    {
      "epoch": 3.623302810230502,
      "grad_norm": 3.9198172092437744,
      "learning_rate": 1.52971721282717e-05,
      "loss": 0.7744,
      "step": 11475
    },
    {
      "epoch": 3.623618566466688,
      "grad_norm": 3.754243850708008,
      "learning_rate": 1.5293663602554205e-05,
      "loss": 0.7128,
      "step": 11476
    },
    {
      "epoch": 3.6239343227028735,
      "grad_norm": 2.914020538330078,
      "learning_rate": 1.5290155076836715e-05,
      "loss": 0.5081,
      "step": 11477
    },
    {
      "epoch": 3.624250078939059,
      "grad_norm": 5.736383438110352,
      "learning_rate": 1.5286646551119222e-05,
      "loss": 0.6902,
      "step": 11478
    },
    {
      "epoch": 3.6245658351752446,
      "grad_norm": 3.394261598587036,
      "learning_rate": 1.528313802540173e-05,
      "loss": 0.697,
      "step": 11479
    },
    {
      "epoch": 3.6248815914114303,
      "grad_norm": 3.3720462322235107,
      "learning_rate": 1.5279629499684232e-05,
      "loss": 0.6989,
      "step": 11480
    },
    {
      "epoch": 3.625197347647616,
      "grad_norm": 3.418668508529663,
      "learning_rate": 1.527612097396674e-05,
      "loss": 0.6441,
      "step": 11481
    },
    {
      "epoch": 3.6255131038838018,
      "grad_norm": 3.5232372283935547,
      "learning_rate": 1.5272612448249246e-05,
      "loss": 0.6451,
      "step": 11482
    },
    {
      "epoch": 3.6258288601199875,
      "grad_norm": 3.333495855331421,
      "learning_rate": 1.5269103922531753e-05,
      "loss": 0.6643,
      "step": 11483
    },
    {
      "epoch": 3.6261446163561732,
      "grad_norm": 3.7718923091888428,
      "learning_rate": 1.526559539681426e-05,
      "loss": 0.7143,
      "step": 11484
    },
    {
      "epoch": 3.6264603725923585,
      "grad_norm": 3.224731206893921,
      "learning_rate": 1.5262086871096766e-05,
      "loss": 0.6707,
      "step": 11485
    },
    {
      "epoch": 3.6267761288285443,
      "grad_norm": 3.5944337844848633,
      "learning_rate": 1.5258578345379273e-05,
      "loss": 0.6903,
      "step": 11486
    },
    {
      "epoch": 3.62709188506473,
      "grad_norm": 3.225039482116699,
      "learning_rate": 1.525506981966178e-05,
      "loss": 0.6187,
      "step": 11487
    },
    {
      "epoch": 3.6274076413009158,
      "grad_norm": 3.3957276344299316,
      "learning_rate": 1.5251561293944287e-05,
      "loss": 0.648,
      "step": 11488
    },
    {
      "epoch": 3.6277233975371015,
      "grad_norm": 3.7076528072357178,
      "learning_rate": 1.5248052768226792e-05,
      "loss": 0.6232,
      "step": 11489
    },
    {
      "epoch": 3.628039153773287,
      "grad_norm": 2.9913690090179443,
      "learning_rate": 1.5244544242509299e-05,
      "loss": 0.6582,
      "step": 11490
    },
    {
      "epoch": 3.6283549100094725,
      "grad_norm": 3.060661792755127,
      "learning_rate": 1.5241035716791804e-05,
      "loss": 0.677,
      "step": 11491
    },
    {
      "epoch": 3.6286706662456583,
      "grad_norm": 3.344942331314087,
      "learning_rate": 1.523752719107431e-05,
      "loss": 0.6197,
      "step": 11492
    },
    {
      "epoch": 3.628986422481844,
      "grad_norm": 3.2670130729675293,
      "learning_rate": 1.5234018665356817e-05,
      "loss": 0.5748,
      "step": 11493
    },
    {
      "epoch": 3.6293021787180297,
      "grad_norm": 4.357150554656982,
      "learning_rate": 1.5230510139639322e-05,
      "loss": 0.7267,
      "step": 11494
    },
    {
      "epoch": 3.6296179349542155,
      "grad_norm": 3.0924196243286133,
      "learning_rate": 1.5227001613921833e-05,
      "loss": 0.6679,
      "step": 11495
    },
    {
      "epoch": 3.629933691190401,
      "grad_norm": 3.0124449729919434,
      "learning_rate": 1.5223493088204338e-05,
      "loss": 0.6468,
      "step": 11496
    },
    {
      "epoch": 3.6302494474265865,
      "grad_norm": 3.044858932495117,
      "learning_rate": 1.5219984562486845e-05,
      "loss": 0.6465,
      "step": 11497
    },
    {
      "epoch": 3.6305652036627722,
      "grad_norm": 2.9565236568450928,
      "learning_rate": 1.521647603676935e-05,
      "loss": 0.6338,
      "step": 11498
    },
    {
      "epoch": 3.630880959898958,
      "grad_norm": 3.2690622806549072,
      "learning_rate": 1.5212967511051856e-05,
      "loss": 0.6741,
      "step": 11499
    },
    {
      "epoch": 3.6311967161351437,
      "grad_norm": 3.261425733566284,
      "learning_rate": 1.5209458985334363e-05,
      "loss": 0.6658,
      "step": 11500
    },
    {
      "epoch": 3.6315124723713295,
      "grad_norm": 3.7508020401000977,
      "learning_rate": 1.5205950459616868e-05,
      "loss": 0.769,
      "step": 11501
    },
    {
      "epoch": 3.6318282286075148,
      "grad_norm": 3.51450252532959,
      "learning_rate": 1.5202441933899375e-05,
      "loss": 0.6694,
      "step": 11502
    },
    {
      "epoch": 3.632143984843701,
      "grad_norm": 3.8118810653686523,
      "learning_rate": 1.5198933408181884e-05,
      "loss": 0.7135,
      "step": 11503
    },
    {
      "epoch": 3.6324597410798862,
      "grad_norm": 3.660470724105835,
      "learning_rate": 1.519542488246439e-05,
      "loss": 0.7064,
      "step": 11504
    },
    {
      "epoch": 3.632775497316072,
      "grad_norm": 3.2022054195404053,
      "learning_rate": 1.5191916356746896e-05,
      "loss": 0.632,
      "step": 11505
    },
    {
      "epoch": 3.6330912535522577,
      "grad_norm": 3.2238237857818604,
      "learning_rate": 1.5188407831029402e-05,
      "loss": 0.6309,
      "step": 11506
    },
    {
      "epoch": 3.6334070097884434,
      "grad_norm": 3.6397271156311035,
      "learning_rate": 1.518489930531191e-05,
      "loss": 0.6712,
      "step": 11507
    },
    {
      "epoch": 3.633722766024629,
      "grad_norm": 3.174198865890503,
      "learning_rate": 1.5181390779594414e-05,
      "loss": 0.5938,
      "step": 11508
    },
    {
      "epoch": 3.6340385222608145,
      "grad_norm": 3.162435293197632,
      "learning_rate": 1.5177882253876921e-05,
      "loss": 0.686,
      "step": 11509
    },
    {
      "epoch": 3.634354278497,
      "grad_norm": 3.680974245071411,
      "learning_rate": 1.5174373728159428e-05,
      "loss": 0.7105,
      "step": 11510
    },
    {
      "epoch": 3.634670034733186,
      "grad_norm": 4.077892303466797,
      "learning_rate": 1.5170865202441933e-05,
      "loss": 0.5975,
      "step": 11511
    },
    {
      "epoch": 3.6349857909693717,
      "grad_norm": 3.3922953605651855,
      "learning_rate": 1.5167356676724442e-05,
      "loss": 0.5943,
      "step": 11512
    },
    {
      "epoch": 3.6353015472055574,
      "grad_norm": 3.114417791366577,
      "learning_rate": 1.5163848151006948e-05,
      "loss": 0.6716,
      "step": 11513
    },
    {
      "epoch": 3.6356173034417427,
      "grad_norm": 3.6331968307495117,
      "learning_rate": 1.5160339625289455e-05,
      "loss": 0.7423,
      "step": 11514
    },
    {
      "epoch": 3.635933059677929,
      "grad_norm": 3.189899206161499,
      "learning_rate": 1.515683109957196e-05,
      "loss": 0.6224,
      "step": 11515
    },
    {
      "epoch": 3.636248815914114,
      "grad_norm": 3.5510900020599365,
      "learning_rate": 1.5153322573854467e-05,
      "loss": 0.7067,
      "step": 11516
    },
    {
      "epoch": 3.6365645721503,
      "grad_norm": 3.650773286819458,
      "learning_rate": 1.5149814048136974e-05,
      "loss": 0.7399,
      "step": 11517
    },
    {
      "epoch": 3.6368803283864857,
      "grad_norm": 3.298037528991699,
      "learning_rate": 1.514630552241948e-05,
      "loss": 0.6724,
      "step": 11518
    },
    {
      "epoch": 3.6371960846226714,
      "grad_norm": 3.7417287826538086,
      "learning_rate": 1.5142796996701986e-05,
      "loss": 0.5907,
      "step": 11519
    },
    {
      "epoch": 3.637511840858857,
      "grad_norm": 3.111612319946289,
      "learning_rate": 1.5139288470984491e-05,
      "loss": 0.6432,
      "step": 11520
    },
    {
      "epoch": 3.6378275970950424,
      "grad_norm": 3.479597806930542,
      "learning_rate": 1.5135779945267001e-05,
      "loss": 0.7537,
      "step": 11521
    },
    {
      "epoch": 3.638143353331228,
      "grad_norm": 3.4449002742767334,
      "learning_rate": 1.5132271419549506e-05,
      "loss": 0.7493,
      "step": 11522
    },
    {
      "epoch": 3.638459109567414,
      "grad_norm": 3.7044527530670166,
      "learning_rate": 1.5128762893832013e-05,
      "loss": 0.6964,
      "step": 11523
    },
    {
      "epoch": 3.6387748658035997,
      "grad_norm": 3.0231428146362305,
      "learning_rate": 1.512525436811452e-05,
      "loss": 0.6065,
      "step": 11524
    },
    {
      "epoch": 3.6390906220397854,
      "grad_norm": 3.367809295654297,
      "learning_rate": 1.5121745842397025e-05,
      "loss": 0.692,
      "step": 11525
    },
    {
      "epoch": 3.6394063782759707,
      "grad_norm": 3.3492698669433594,
      "learning_rate": 1.5118237316679532e-05,
      "loss": 0.6283,
      "step": 11526
    },
    {
      "epoch": 3.639722134512157,
      "grad_norm": 3.99851131439209,
      "learning_rate": 1.5114728790962037e-05,
      "loss": 0.7018,
      "step": 11527
    },
    {
      "epoch": 3.640037890748342,
      "grad_norm": 3.378363847732544,
      "learning_rate": 1.5111220265244544e-05,
      "loss": 0.7057,
      "step": 11528
    },
    {
      "epoch": 3.640353646984528,
      "grad_norm": 3.359572649002075,
      "learning_rate": 1.510771173952705e-05,
      "loss": 0.5908,
      "step": 11529
    },
    {
      "epoch": 3.6406694032207136,
      "grad_norm": 3.1804850101470947,
      "learning_rate": 1.510420321380956e-05,
      "loss": 0.6302,
      "step": 11530
    },
    {
      "epoch": 3.6409851594568994,
      "grad_norm": 3.3238887786865234,
      "learning_rate": 1.5100694688092066e-05,
      "loss": 0.596,
      "step": 11531
    },
    {
      "epoch": 3.641300915693085,
      "grad_norm": 3.2626166343688965,
      "learning_rate": 1.5097186162374571e-05,
      "loss": 0.5586,
      "step": 11532
    },
    {
      "epoch": 3.6416166719292704,
      "grad_norm": 3.2850258350372314,
      "learning_rate": 1.5093677636657078e-05,
      "loss": 0.6451,
      "step": 11533
    },
    {
      "epoch": 3.641932428165456,
      "grad_norm": 3.144026756286621,
      "learning_rate": 1.5090169110939583e-05,
      "loss": 0.6215,
      "step": 11534
    },
    {
      "epoch": 3.642248184401642,
      "grad_norm": 2.990375280380249,
      "learning_rate": 1.508666058522209e-05,
      "loss": 0.6601,
      "step": 11535
    },
    {
      "epoch": 3.6425639406378276,
      "grad_norm": 2.8907482624053955,
      "learning_rate": 1.5083152059504597e-05,
      "loss": 0.571,
      "step": 11536
    },
    {
      "epoch": 3.6428796968740134,
      "grad_norm": 2.993227958679199,
      "learning_rate": 1.5079643533787102e-05,
      "loss": 0.6624,
      "step": 11537
    },
    {
      "epoch": 3.6431954531101987,
      "grad_norm": 3.982077121734619,
      "learning_rate": 1.5076135008069612e-05,
      "loss": 0.7352,
      "step": 11538
    },
    {
      "epoch": 3.643511209346385,
      "grad_norm": 3.3617632389068604,
      "learning_rate": 1.5072626482352117e-05,
      "loss": 0.6742,
      "step": 11539
    },
    {
      "epoch": 3.64382696558257,
      "grad_norm": 3.861236095428467,
      "learning_rate": 1.5069117956634624e-05,
      "loss": 0.6801,
      "step": 11540
    },
    {
      "epoch": 3.644142721818756,
      "grad_norm": 3.159623861312866,
      "learning_rate": 1.5065609430917129e-05,
      "loss": 0.6353,
      "step": 11541
    },
    {
      "epoch": 3.6444584780549416,
      "grad_norm": 3.438638687133789,
      "learning_rate": 1.5062100905199636e-05,
      "loss": 0.7106,
      "step": 11542
    },
    {
      "epoch": 3.6447742342911273,
      "grad_norm": 3.6695637702941895,
      "learning_rate": 1.5058592379482143e-05,
      "loss": 0.6478,
      "step": 11543
    },
    {
      "epoch": 3.645089990527313,
      "grad_norm": 3.2782235145568848,
      "learning_rate": 1.5055083853764648e-05,
      "loss": 0.6367,
      "step": 11544
    },
    {
      "epoch": 3.6454057467634984,
      "grad_norm": 3.035642385482788,
      "learning_rate": 1.5051575328047155e-05,
      "loss": 0.6783,
      "step": 11545
    },
    {
      "epoch": 3.645721502999684,
      "grad_norm": 3.122437000274658,
      "learning_rate": 1.504806680232966e-05,
      "loss": 0.6096,
      "step": 11546
    },
    {
      "epoch": 3.64603725923587,
      "grad_norm": 3.4534668922424316,
      "learning_rate": 1.504455827661217e-05,
      "loss": 0.6475,
      "step": 11547
    },
    {
      "epoch": 3.6463530154720556,
      "grad_norm": 3.4425718784332275,
      "learning_rate": 1.5041049750894675e-05,
      "loss": 0.6593,
      "step": 11548
    },
    {
      "epoch": 3.6466687717082413,
      "grad_norm": 3.144728422164917,
      "learning_rate": 1.5037541225177182e-05,
      "loss": 0.6358,
      "step": 11549
    },
    {
      "epoch": 3.646984527944427,
      "grad_norm": 3.2238123416900635,
      "learning_rate": 1.5034032699459689e-05,
      "loss": 0.6608,
      "step": 11550
    },
    {
      "epoch": 3.647300284180613,
      "grad_norm": 4.2336106300354,
      "learning_rate": 1.5030524173742194e-05,
      "loss": 0.6829,
      "step": 11551
    },
    {
      "epoch": 3.647616040416798,
      "grad_norm": 3.2864468097686768,
      "learning_rate": 1.50270156480247e-05,
      "loss": 0.6517,
      "step": 11552
    },
    {
      "epoch": 3.647931796652984,
      "grad_norm": 3.158784866333008,
      "learning_rate": 1.5023507122307206e-05,
      "loss": 0.5695,
      "step": 11553
    },
    {
      "epoch": 3.6482475528891696,
      "grad_norm": 3.3760557174682617,
      "learning_rate": 1.5019998596589713e-05,
      "loss": 0.6192,
      "step": 11554
    },
    {
      "epoch": 3.6485633091253553,
      "grad_norm": 4.1279072761535645,
      "learning_rate": 1.501649007087222e-05,
      "loss": 0.7834,
      "step": 11555
    },
    {
      "epoch": 3.648879065361541,
      "grad_norm": 3.374866485595703,
      "learning_rate": 1.5012981545154728e-05,
      "loss": 0.7088,
      "step": 11556
    },
    {
      "epoch": 3.6491948215977263,
      "grad_norm": 3.716108798980713,
      "learning_rate": 1.5009473019437235e-05,
      "loss": 0.5984,
      "step": 11557
    },
    {
      "epoch": 3.649510577833912,
      "grad_norm": 3.289592742919922,
      "learning_rate": 1.500596449371974e-05,
      "loss": 0.7002,
      "step": 11558
    },
    {
      "epoch": 3.649826334070098,
      "grad_norm": 3.3678946495056152,
      "learning_rate": 1.5002455968002247e-05,
      "loss": 0.7254,
      "step": 11559
    },
    {
      "epoch": 3.6501420903062836,
      "grad_norm": 3.460540771484375,
      "learning_rate": 1.4998947442284752e-05,
      "loss": 0.5834,
      "step": 11560
    },
    {
      "epoch": 3.6504578465424693,
      "grad_norm": 3.700723171234131,
      "learning_rate": 1.4995438916567259e-05,
      "loss": 0.5409,
      "step": 11561
    },
    {
      "epoch": 3.650773602778655,
      "grad_norm": 3.70369291305542,
      "learning_rate": 1.4991930390849765e-05,
      "loss": 0.7447,
      "step": 11562
    },
    {
      "epoch": 3.6510893590148408,
      "grad_norm": 3.491884708404541,
      "learning_rate": 1.498842186513227e-05,
      "loss": 0.6612,
      "step": 11563
    },
    {
      "epoch": 3.651405115251026,
      "grad_norm": 3.637310028076172,
      "learning_rate": 1.4984913339414777e-05,
      "loss": 0.641,
      "step": 11564
    },
    {
      "epoch": 3.651720871487212,
      "grad_norm": 3.3955907821655273,
      "learning_rate": 1.4981404813697286e-05,
      "loss": 0.6225,
      "step": 11565
    },
    {
      "epoch": 3.6520366277233975,
      "grad_norm": 3.3498849868774414,
      "learning_rate": 1.4977896287979793e-05,
      "loss": 0.5774,
      "step": 11566
    },
    {
      "epoch": 3.6523523839595833,
      "grad_norm": 3.257174253463745,
      "learning_rate": 1.4974387762262298e-05,
      "loss": 0.5815,
      "step": 11567
    },
    {
      "epoch": 3.652668140195769,
      "grad_norm": 3.0453970432281494,
      "learning_rate": 1.4970879236544805e-05,
      "loss": 0.6098,
      "step": 11568
    },
    {
      "epoch": 3.6529838964319543,
      "grad_norm": 4.21284294128418,
      "learning_rate": 1.4967370710827311e-05,
      "loss": 0.6386,
      "step": 11569
    },
    {
      "epoch": 3.65329965266814,
      "grad_norm": 3.4765868186950684,
      "learning_rate": 1.4963862185109816e-05,
      "loss": 0.6114,
      "step": 11570
    },
    {
      "epoch": 3.653615408904326,
      "grad_norm": 3.4247138500213623,
      "learning_rate": 1.4960353659392323e-05,
      "loss": 0.7021,
      "step": 11571
    },
    {
      "epoch": 3.6539311651405115,
      "grad_norm": 3.937391519546509,
      "learning_rate": 1.495684513367483e-05,
      "loss": 0.6555,
      "step": 11572
    },
    {
      "epoch": 3.6542469213766973,
      "grad_norm": 3.60113263130188,
      "learning_rate": 1.4953336607957339e-05,
      "loss": 0.624,
      "step": 11573
    },
    {
      "epoch": 3.654562677612883,
      "grad_norm": 2.8042142391204834,
      "learning_rate": 1.4949828082239844e-05,
      "loss": 0.5842,
      "step": 11574
    },
    {
      "epoch": 3.6548784338490687,
      "grad_norm": 3.724480152130127,
      "learning_rate": 1.494631955652235e-05,
      "loss": 0.6292,
      "step": 11575
    },
    {
      "epoch": 3.655194190085254,
      "grad_norm": 4.247527122497559,
      "learning_rate": 1.4942811030804857e-05,
      "loss": 0.6605,
      "step": 11576
    },
    {
      "epoch": 3.6555099463214398,
      "grad_norm": 2.9525227546691895,
      "learning_rate": 1.4939302505087362e-05,
      "loss": 0.6462,
      "step": 11577
    },
    {
      "epoch": 3.6558257025576255,
      "grad_norm": 3.5285727977752686,
      "learning_rate": 1.493579397936987e-05,
      "loss": 0.7046,
      "step": 11578
    },
    {
      "epoch": 3.6561414587938112,
      "grad_norm": 3.380282402038574,
      "learning_rate": 1.4932285453652376e-05,
      "loss": 0.6313,
      "step": 11579
    },
    {
      "epoch": 3.656457215029997,
      "grad_norm": 3.7653660774230957,
      "learning_rate": 1.4928776927934881e-05,
      "loss": 0.6239,
      "step": 11580
    },
    {
      "epoch": 3.6567729712661823,
      "grad_norm": 3.60205340385437,
      "learning_rate": 1.4925268402217388e-05,
      "loss": 0.7159,
      "step": 11581
    },
    {
      "epoch": 3.657088727502368,
      "grad_norm": 3.4314582347869873,
      "learning_rate": 1.4921759876499897e-05,
      "loss": 0.6911,
      "step": 11582
    },
    {
      "epoch": 3.6574044837385538,
      "grad_norm": 3.5783843994140625,
      "learning_rate": 1.4918251350782403e-05,
      "loss": 0.6895,
      "step": 11583
    },
    {
      "epoch": 3.6577202399747395,
      "grad_norm": 3.6851613521575928,
      "learning_rate": 1.4914742825064908e-05,
      "loss": 0.6712,
      "step": 11584
    },
    {
      "epoch": 3.6580359962109252,
      "grad_norm": 3.4243226051330566,
      "learning_rate": 1.4911234299347415e-05,
      "loss": 0.8035,
      "step": 11585
    },
    {
      "epoch": 3.658351752447111,
      "grad_norm": 4.0622782707214355,
      "learning_rate": 1.4907725773629922e-05,
      "loss": 0.7851,
      "step": 11586
    },
    {
      "epoch": 3.6586675086832967,
      "grad_norm": 2.800527572631836,
      "learning_rate": 1.4904217247912427e-05,
      "loss": 0.6573,
      "step": 11587
    },
    {
      "epoch": 3.658983264919482,
      "grad_norm": 4.091883182525635,
      "learning_rate": 1.4900708722194934e-05,
      "loss": 0.6312,
      "step": 11588
    },
    {
      "epoch": 3.6592990211556677,
      "grad_norm": 3.586977243423462,
      "learning_rate": 1.489720019647744e-05,
      "loss": 0.7,
      "step": 11589
    },
    {
      "epoch": 3.6596147773918535,
      "grad_norm": 3.498284101486206,
      "learning_rate": 1.4893691670759946e-05,
      "loss": 0.6797,
      "step": 11590
    },
    {
      "epoch": 3.659930533628039,
      "grad_norm": 3.45766544342041,
      "learning_rate": 1.4890183145042454e-05,
      "loss": 0.6214,
      "step": 11591
    },
    {
      "epoch": 3.660246289864225,
      "grad_norm": 3.8867220878601074,
      "learning_rate": 1.4886674619324961e-05,
      "loss": 0.6088,
      "step": 11592
    },
    {
      "epoch": 3.6605620461004102,
      "grad_norm": 3.2781856060028076,
      "learning_rate": 1.4883166093607468e-05,
      "loss": 0.6645,
      "step": 11593
    },
    {
      "epoch": 3.660877802336596,
      "grad_norm": 4.327938079833984,
      "learning_rate": 1.4879657567889973e-05,
      "loss": 0.6237,
      "step": 11594
    },
    {
      "epoch": 3.6611935585727817,
      "grad_norm": 3.113131523132324,
      "learning_rate": 1.487614904217248e-05,
      "loss": 0.615,
      "step": 11595
    },
    {
      "epoch": 3.6615093148089675,
      "grad_norm": 3.2788431644439697,
      "learning_rate": 1.4872640516454985e-05,
      "loss": 0.6514,
      "step": 11596
    },
    {
      "epoch": 3.661825071045153,
      "grad_norm": 2.987942934036255,
      "learning_rate": 1.4869131990737492e-05,
      "loss": 0.5477,
      "step": 11597
    },
    {
      "epoch": 3.662140827281339,
      "grad_norm": 2.4783120155334473,
      "learning_rate": 1.4865623465019999e-05,
      "loss": 0.4992,
      "step": 11598
    },
    {
      "epoch": 3.6624565835175247,
      "grad_norm": 3.52730393409729,
      "learning_rate": 1.4862114939302504e-05,
      "loss": 0.7785,
      "step": 11599
    },
    {
      "epoch": 3.66277233975371,
      "grad_norm": 3.1417129039764404,
      "learning_rate": 1.4858606413585014e-05,
      "loss": 0.6574,
      "step": 11600
    },
    {
      "epoch": 3.66277233975371,
      "eval_loss": 0.8032662868499756,
      "eval_runtime": 4.6253,
      "eval_samples_per_second": 538.344,
      "eval_steps_per_second": 4.54,
      "step": 11600
    },
    {
      "epoch": 3.6630880959898957,
      "grad_norm": 3.7663440704345703,
      "learning_rate": 1.485509788786752e-05,
      "loss": 0.6845,
      "step": 11601
    },
    {
      "epoch": 3.6634038522260814,
      "grad_norm": 3.5016729831695557,
      "learning_rate": 1.4851589362150026e-05,
      "loss": 0.5656,
      "step": 11602
    },
    {
      "epoch": 3.663719608462267,
      "grad_norm": 3.375134229660034,
      "learning_rate": 1.4848080836432531e-05,
      "loss": 0.6436,
      "step": 11603
    },
    {
      "epoch": 3.664035364698453,
      "grad_norm": 3.2233009338378906,
      "learning_rate": 1.4844572310715038e-05,
      "loss": 0.6428,
      "step": 11604
    },
    {
      "epoch": 3.664351120934638,
      "grad_norm": 3.4321391582489014,
      "learning_rate": 1.4841063784997545e-05,
      "loss": 0.7403,
      "step": 11605
    },
    {
      "epoch": 3.6646668771708244,
      "grad_norm": 3.5823068618774414,
      "learning_rate": 1.483755525928005e-05,
      "loss": 0.7532,
      "step": 11606
    },
    {
      "epoch": 3.6649826334070097,
      "grad_norm": 3.474163770675659,
      "learning_rate": 1.4834046733562557e-05,
      "loss": 0.6849,
      "step": 11607
    },
    {
      "epoch": 3.6652983896431954,
      "grad_norm": 3.1653902530670166,
      "learning_rate": 1.4830538207845065e-05,
      "loss": 0.6503,
      "step": 11608
    },
    {
      "epoch": 3.665614145879381,
      "grad_norm": 3.607128858566284,
      "learning_rate": 1.4827029682127572e-05,
      "loss": 0.6122,
      "step": 11609
    },
    {
      "epoch": 3.665929902115567,
      "grad_norm": 3.349156141281128,
      "learning_rate": 1.4823521156410077e-05,
      "loss": 0.6724,
      "step": 11610
    },
    {
      "epoch": 3.6662456583517526,
      "grad_norm": 3.2905166149139404,
      "learning_rate": 1.4820012630692584e-05,
      "loss": 0.6572,
      "step": 11611
    },
    {
      "epoch": 3.666561414587938,
      "grad_norm": 3.2714617252349854,
      "learning_rate": 1.481650410497509e-05,
      "loss": 0.6012,
      "step": 11612
    },
    {
      "epoch": 3.6668771708241237,
      "grad_norm": 3.4373950958251953,
      "learning_rate": 1.4812995579257596e-05,
      "loss": 0.6902,
      "step": 11613
    },
    {
      "epoch": 3.6671929270603094,
      "grad_norm": 3.1285617351531982,
      "learning_rate": 1.4809487053540103e-05,
      "loss": 0.6089,
      "step": 11614
    },
    {
      "epoch": 3.667508683296495,
      "grad_norm": 3.6130053997039795,
      "learning_rate": 1.480597852782261e-05,
      "loss": 0.7779,
      "step": 11615
    },
    {
      "epoch": 3.667824439532681,
      "grad_norm": 3.292147397994995,
      "learning_rate": 1.4802470002105115e-05,
      "loss": 0.6688,
      "step": 11616
    },
    {
      "epoch": 3.668140195768866,
      "grad_norm": 3.7714130878448486,
      "learning_rate": 1.4798961476387623e-05,
      "loss": 0.5347,
      "step": 11617
    },
    {
      "epoch": 3.6684559520050524,
      "grad_norm": 3.5336592197418213,
      "learning_rate": 1.479545295067013e-05,
      "loss": 0.6496,
      "step": 11618
    },
    {
      "epoch": 3.6687717082412377,
      "grad_norm": 3.16691517829895,
      "learning_rate": 1.4791944424952637e-05,
      "loss": 0.679,
      "step": 11619
    },
    {
      "epoch": 3.6690874644774234,
      "grad_norm": 3.5320236682891846,
      "learning_rate": 1.4788435899235142e-05,
      "loss": 0.7396,
      "step": 11620
    },
    {
      "epoch": 3.669403220713609,
      "grad_norm": 4.016422748565674,
      "learning_rate": 1.4784927373517649e-05,
      "loss": 0.6943,
      "step": 11621
    },
    {
      "epoch": 3.669718976949795,
      "grad_norm": 3.146573543548584,
      "learning_rate": 1.4781418847800156e-05,
      "loss": 0.6969,
      "step": 11622
    },
    {
      "epoch": 3.6700347331859806,
      "grad_norm": 3.553978443145752,
      "learning_rate": 1.477791032208266e-05,
      "loss": 0.6564,
      "step": 11623
    },
    {
      "epoch": 3.670350489422166,
      "grad_norm": 3.5833752155303955,
      "learning_rate": 1.4774401796365167e-05,
      "loss": 0.7087,
      "step": 11624
    },
    {
      "epoch": 3.6706662456583516,
      "grad_norm": 3.9065725803375244,
      "learning_rate": 1.4770893270647673e-05,
      "loss": 0.7055,
      "step": 11625
    },
    {
      "epoch": 3.6709820018945374,
      "grad_norm": 3.373941659927368,
      "learning_rate": 1.4767384744930183e-05,
      "loss": 0.6852,
      "step": 11626
    },
    {
      "epoch": 3.671297758130723,
      "grad_norm": 2.9183948040008545,
      "learning_rate": 1.4763876219212688e-05,
      "loss": 0.554,
      "step": 11627
    },
    {
      "epoch": 3.671613514366909,
      "grad_norm": 3.029010772705078,
      "learning_rate": 1.4760367693495195e-05,
      "loss": 0.5624,
      "step": 11628
    },
    {
      "epoch": 3.671929270603094,
      "grad_norm": 3.006350040435791,
      "learning_rate": 1.4756859167777702e-05,
      "loss": 0.6557,
      "step": 11629
    },
    {
      "epoch": 3.6722450268392803,
      "grad_norm": 3.2168099880218506,
      "learning_rate": 1.4753350642060207e-05,
      "loss": 0.6648,
      "step": 11630
    },
    {
      "epoch": 3.6725607830754656,
      "grad_norm": 3.0749094486236572,
      "learning_rate": 1.4749842116342713e-05,
      "loss": 0.6101,
      "step": 11631
    },
    {
      "epoch": 3.6728765393116514,
      "grad_norm": 3.7163186073303223,
      "learning_rate": 1.4746333590625219e-05,
      "loss": 0.6634,
      "step": 11632
    },
    {
      "epoch": 3.673192295547837,
      "grad_norm": 3.3246591091156006,
      "learning_rate": 1.4742825064907725e-05,
      "loss": 0.6532,
      "step": 11633
    },
    {
      "epoch": 3.673508051784023,
      "grad_norm": 4.323907852172852,
      "learning_rate": 1.4739316539190232e-05,
      "loss": 0.5927,
      "step": 11634
    },
    {
      "epoch": 3.6738238080202086,
      "grad_norm": 3.5079877376556396,
      "learning_rate": 1.473580801347274e-05,
      "loss": 0.6674,
      "step": 11635
    },
    {
      "epoch": 3.674139564256394,
      "grad_norm": 3.1434171199798584,
      "learning_rate": 1.4732299487755248e-05,
      "loss": 0.6443,
      "step": 11636
    },
    {
      "epoch": 3.6744553204925796,
      "grad_norm": 2.70865535736084,
      "learning_rate": 1.4728790962037753e-05,
      "loss": 0.5931,
      "step": 11637
    },
    {
      "epoch": 3.6747710767287654,
      "grad_norm": 3.4824986457824707,
      "learning_rate": 1.472528243632026e-05,
      "loss": 0.562,
      "step": 11638
    },
    {
      "epoch": 3.675086832964951,
      "grad_norm": 3.1064019203186035,
      "learning_rate": 1.4721773910602765e-05,
      "loss": 0.6073,
      "step": 11639
    },
    {
      "epoch": 3.675402589201137,
      "grad_norm": 3.889298439025879,
      "learning_rate": 1.4718265384885271e-05,
      "loss": 0.7468,
      "step": 11640
    },
    {
      "epoch": 3.675718345437322,
      "grad_norm": 4.039004802703857,
      "learning_rate": 1.4714756859167778e-05,
      "loss": 0.5958,
      "step": 11641
    },
    {
      "epoch": 3.6760341016735083,
      "grad_norm": 3.2001893520355225,
      "learning_rate": 1.4711248333450283e-05,
      "loss": 0.6378,
      "step": 11642
    },
    {
      "epoch": 3.6763498579096936,
      "grad_norm": 2.9955077171325684,
      "learning_rate": 1.4707739807732794e-05,
      "loss": 0.6001,
      "step": 11643
    },
    {
      "epoch": 3.6766656141458793,
      "grad_norm": 3.1322596073150635,
      "learning_rate": 1.4704231282015299e-05,
      "loss": 0.7181,
      "step": 11644
    },
    {
      "epoch": 3.676981370382065,
      "grad_norm": 3.116060495376587,
      "learning_rate": 1.4700722756297805e-05,
      "loss": 0.6132,
      "step": 11645
    },
    {
      "epoch": 3.677297126618251,
      "grad_norm": 3.3676137924194336,
      "learning_rate": 1.469721423058031e-05,
      "loss": 0.5718,
      "step": 11646
    },
    {
      "epoch": 3.6776128828544365,
      "grad_norm": 3.4835970401763916,
      "learning_rate": 1.4693705704862817e-05,
      "loss": 0.6422,
      "step": 11647
    },
    {
      "epoch": 3.677928639090622,
      "grad_norm": 3.0935792922973633,
      "learning_rate": 1.4690197179145324e-05,
      "loss": 0.68,
      "step": 11648
    },
    {
      "epoch": 3.6782443953268076,
      "grad_norm": 3.7680375576019287,
      "learning_rate": 1.468668865342783e-05,
      "loss": 0.5998,
      "step": 11649
    },
    {
      "epoch": 3.6785601515629933,
      "grad_norm": 3.367079257965088,
      "learning_rate": 1.4683180127710336e-05,
      "loss": 0.5657,
      "step": 11650
    },
    {
      "epoch": 3.678875907799179,
      "grad_norm": 4.100010395050049,
      "learning_rate": 1.4679671601992843e-05,
      "loss": 0.7069,
      "step": 11651
    },
    {
      "epoch": 3.679191664035365,
      "grad_norm": 3.7417848110198975,
      "learning_rate": 1.4676163076275351e-05,
      "loss": 0.6216,
      "step": 11652
    },
    {
      "epoch": 3.67950742027155,
      "grad_norm": 4.155975818634033,
      "learning_rate": 1.4672654550557857e-05,
      "loss": 0.7782,
      "step": 11653
    },
    {
      "epoch": 3.6798231765077363,
      "grad_norm": 3.3169684410095215,
      "learning_rate": 1.4669146024840363e-05,
      "loss": 0.6924,
      "step": 11654
    },
    {
      "epoch": 3.6801389327439216,
      "grad_norm": 3.405287504196167,
      "learning_rate": 1.466563749912287e-05,
      "loss": 0.7084,
      "step": 11655
    },
    {
      "epoch": 3.6804546889801073,
      "grad_norm": 3.534900188446045,
      "learning_rate": 1.4662128973405375e-05,
      "loss": 0.5789,
      "step": 11656
    },
    {
      "epoch": 3.680770445216293,
      "grad_norm": 3.2930569648742676,
      "learning_rate": 1.4658620447687882e-05,
      "loss": 0.6184,
      "step": 11657
    },
    {
      "epoch": 3.681086201452479,
      "grad_norm": 3.1932730674743652,
      "learning_rate": 1.4655111921970387e-05,
      "loss": 0.619,
      "step": 11658
    },
    {
      "epoch": 3.6814019576886645,
      "grad_norm": 4.076601505279541,
      "learning_rate": 1.4651603396252894e-05,
      "loss": 0.7354,
      "step": 11659
    },
    {
      "epoch": 3.68171771392485,
      "grad_norm": 3.9013512134552,
      "learning_rate": 1.4648094870535401e-05,
      "loss": 0.6747,
      "step": 11660
    },
    {
      "epoch": 3.6820334701610355,
      "grad_norm": 2.88830304145813,
      "learning_rate": 1.464458634481791e-05,
      "loss": 0.653,
      "step": 11661
    },
    {
      "epoch": 3.6823492263972213,
      "grad_norm": 3.4624881744384766,
      "learning_rate": 1.4641077819100416e-05,
      "loss": 0.5848,
      "step": 11662
    },
    {
      "epoch": 3.682664982633407,
      "grad_norm": 3.5645744800567627,
      "learning_rate": 1.4637569293382921e-05,
      "loss": 0.6994,
      "step": 11663
    },
    {
      "epoch": 3.6829807388695928,
      "grad_norm": 3.58674693107605,
      "learning_rate": 1.4634060767665428e-05,
      "loss": 0.6928,
      "step": 11664
    },
    {
      "epoch": 3.6832964951057785,
      "grad_norm": 3.7707650661468506,
      "learning_rate": 1.4630552241947933e-05,
      "loss": 0.6281,
      "step": 11665
    },
    {
      "epoch": 3.6836122513419642,
      "grad_norm": 2.6775171756744385,
      "learning_rate": 1.462704371623044e-05,
      "loss": 0.691,
      "step": 11666
    },
    {
      "epoch": 3.6839280075781495,
      "grad_norm": 2.993889093399048,
      "learning_rate": 1.4623535190512947e-05,
      "loss": 0.6912,
      "step": 11667
    },
    {
      "epoch": 3.6842437638143353,
      "grad_norm": 2.6669440269470215,
      "learning_rate": 1.4620026664795452e-05,
      "loss": 0.6406,
      "step": 11668
    },
    {
      "epoch": 3.684559520050521,
      "grad_norm": 3.80379581451416,
      "learning_rate": 1.4616518139077962e-05,
      "loss": 0.5896,
      "step": 11669
    },
    {
      "epoch": 3.6848752762867067,
      "grad_norm": 3.5302844047546387,
      "learning_rate": 1.4613009613360467e-05,
      "loss": 0.6855,
      "step": 11670
    },
    {
      "epoch": 3.6851910325228925,
      "grad_norm": 2.88794207572937,
      "learning_rate": 1.4609501087642974e-05,
      "loss": 0.6024,
      "step": 11671
    },
    {
      "epoch": 3.685506788759078,
      "grad_norm": 3.618891477584839,
      "learning_rate": 1.460599256192548e-05,
      "loss": 0.7257,
      "step": 11672
    },
    {
      "epoch": 3.6858225449952635,
      "grad_norm": 3.2354512214660645,
      "learning_rate": 1.4602484036207986e-05,
      "loss": 0.6535,
      "step": 11673
    },
    {
      "epoch": 3.6861383012314493,
      "grad_norm": 3.2231154441833496,
      "learning_rate": 1.4598975510490493e-05,
      "loss": 0.6499,
      "step": 11674
    },
    {
      "epoch": 3.686454057467635,
      "grad_norm": 2.9155287742614746,
      "learning_rate": 1.4595466984772998e-05,
      "loss": 0.6217,
      "step": 11675
    },
    {
      "epoch": 3.6867698137038207,
      "grad_norm": 3.0319395065307617,
      "learning_rate": 1.4591958459055505e-05,
      "loss": 0.632,
      "step": 11676
    },
    {
      "epoch": 3.6870855699400065,
      "grad_norm": 3.4318690299987793,
      "learning_rate": 1.4588449933338012e-05,
      "loss": 0.6306,
      "step": 11677
    },
    {
      "epoch": 3.687401326176192,
      "grad_norm": 3.5801923274993896,
      "learning_rate": 1.458494140762052e-05,
      "loss": 0.6105,
      "step": 11678
    },
    {
      "epoch": 3.6877170824123775,
      "grad_norm": 3.249390125274658,
      "learning_rate": 1.4581432881903025e-05,
      "loss": 0.6477,
      "step": 11679
    },
    {
      "epoch": 3.6880328386485632,
      "grad_norm": 3.212002992630005,
      "learning_rate": 1.4577924356185532e-05,
      "loss": 0.6806,
      "step": 11680
    },
    {
      "epoch": 3.688348594884749,
      "grad_norm": 2.908480405807495,
      "learning_rate": 1.4574415830468039e-05,
      "loss": 0.6936,
      "step": 11681
    },
    {
      "epoch": 3.6886643511209347,
      "grad_norm": 3.2813167572021484,
      "learning_rate": 1.4570907304750544e-05,
      "loss": 0.6893,
      "step": 11682
    },
    {
      "epoch": 3.6889801073571205,
      "grad_norm": 3.742219924926758,
      "learning_rate": 1.456739877903305e-05,
      "loss": 0.7497,
      "step": 11683
    },
    {
      "epoch": 3.6892958635933057,
      "grad_norm": 3.82529616355896,
      "learning_rate": 1.4563890253315558e-05,
      "loss": 0.5728,
      "step": 11684
    },
    {
      "epoch": 3.6896116198294915,
      "grad_norm": 3.7744648456573486,
      "learning_rate": 1.4560381727598063e-05,
      "loss": 0.7395,
      "step": 11685
    },
    {
      "epoch": 3.6899273760656772,
      "grad_norm": 3.7782440185546875,
      "learning_rate": 1.455687320188057e-05,
      "loss": 0.6367,
      "step": 11686
    },
    {
      "epoch": 3.690243132301863,
      "grad_norm": 3.3413288593292236,
      "learning_rate": 1.4553364676163078e-05,
      "loss": 0.6164,
      "step": 11687
    },
    {
      "epoch": 3.6905588885380487,
      "grad_norm": 3.216228723526001,
      "learning_rate": 1.4549856150445585e-05,
      "loss": 0.68,
      "step": 11688
    },
    {
      "epoch": 3.6908746447742344,
      "grad_norm": 3.448438882827759,
      "learning_rate": 1.454634762472809e-05,
      "loss": 0.7906,
      "step": 11689
    },
    {
      "epoch": 3.69119040101042,
      "grad_norm": 2.814653158187866,
      "learning_rate": 1.4542839099010597e-05,
      "loss": 0.636,
      "step": 11690
    },
    {
      "epoch": 3.6915061572466055,
      "grad_norm": 3.7990176677703857,
      "learning_rate": 1.4539330573293104e-05,
      "loss": 0.7788,
      "step": 11691
    },
    {
      "epoch": 3.691821913482791,
      "grad_norm": 3.5382585525512695,
      "learning_rate": 1.4535822047575609e-05,
      "loss": 0.7086,
      "step": 11692
    },
    {
      "epoch": 3.692137669718977,
      "grad_norm": 3.0234432220458984,
      "learning_rate": 1.4532313521858116e-05,
      "loss": 0.6988,
      "step": 11693
    },
    {
      "epoch": 3.6924534259551627,
      "grad_norm": 3.3249449729919434,
      "learning_rate": 1.452880499614062e-05,
      "loss": 0.5987,
      "step": 11694
    },
    {
      "epoch": 3.6927691821913484,
      "grad_norm": 3.3646063804626465,
      "learning_rate": 1.4525296470423127e-05,
      "loss": 0.64,
      "step": 11695
    },
    {
      "epoch": 3.6930849384275337,
      "grad_norm": 3.378734827041626,
      "learning_rate": 1.4521787944705636e-05,
      "loss": 0.6618,
      "step": 11696
    },
    {
      "epoch": 3.6934006946637195,
      "grad_norm": 3.729719877243042,
      "learning_rate": 1.4518279418988143e-05,
      "loss": 0.623,
      "step": 11697
    },
    {
      "epoch": 3.693716450899905,
      "grad_norm": 3.2486369609832764,
      "learning_rate": 1.451477089327065e-05,
      "loss": 0.7251,
      "step": 11698
    },
    {
      "epoch": 3.694032207136091,
      "grad_norm": 3.6627109050750732,
      "learning_rate": 1.4511262367553155e-05,
      "loss": 0.7117,
      "step": 11699
    },
    {
      "epoch": 3.6943479633722767,
      "grad_norm": 3.0583832263946533,
      "learning_rate": 1.4507753841835662e-05,
      "loss": 0.531,
      "step": 11700
    },
    {
      "epoch": 3.6946637196084624,
      "grad_norm": 3.2922520637512207,
      "learning_rate": 1.4504245316118167e-05,
      "loss": 0.6826,
      "step": 11701
    },
    {
      "epoch": 3.694979475844648,
      "grad_norm": 3.100393772125244,
      "learning_rate": 1.4500736790400673e-05,
      "loss": 0.6491,
      "step": 11702
    },
    {
      "epoch": 3.6952952320808334,
      "grad_norm": 3.0910818576812744,
      "learning_rate": 1.449722826468318e-05,
      "loss": 0.6658,
      "step": 11703
    },
    {
      "epoch": 3.695610988317019,
      "grad_norm": 3.9943227767944336,
      "learning_rate": 1.4493719738965689e-05,
      "loss": 0.6291,
      "step": 11704
    },
    {
      "epoch": 3.695926744553205,
      "grad_norm": 2.929502487182617,
      "learning_rate": 1.4490211213248196e-05,
      "loss": 0.649,
      "step": 11705
    },
    {
      "epoch": 3.6962425007893907,
      "grad_norm": 3.4531092643737793,
      "learning_rate": 1.44867026875307e-05,
      "loss": 0.6312,
      "step": 11706
    },
    {
      "epoch": 3.6965582570255764,
      "grad_norm": 3.422649621963501,
      "learning_rate": 1.4483194161813208e-05,
      "loss": 0.7135,
      "step": 11707
    },
    {
      "epoch": 3.6968740132617617,
      "grad_norm": 3.4267454147338867,
      "learning_rate": 1.4479685636095713e-05,
      "loss": 0.6412,
      "step": 11708
    },
    {
      "epoch": 3.6971897694979474,
      "grad_norm": 2.966229200363159,
      "learning_rate": 1.447617711037822e-05,
      "loss": 0.6265,
      "step": 11709
    },
    {
      "epoch": 3.697505525734133,
      "grad_norm": 2.8101308345794678,
      "learning_rate": 1.4472668584660726e-05,
      "loss": 0.6121,
      "step": 11710
    },
    {
      "epoch": 3.697821281970319,
      "grad_norm": 3.2329697608947754,
      "learning_rate": 1.4469160058943231e-05,
      "loss": 0.629,
      "step": 11711
    },
    {
      "epoch": 3.6981370382065046,
      "grad_norm": 3.3341660499572754,
      "learning_rate": 1.4465651533225738e-05,
      "loss": 0.6542,
      "step": 11712
    },
    {
      "epoch": 3.6984527944426904,
      "grad_norm": 3.1766817569732666,
      "learning_rate": 1.4462143007508247e-05,
      "loss": 0.6036,
      "step": 11713
    },
    {
      "epoch": 3.698768550678876,
      "grad_norm": 3.1117606163024902,
      "learning_rate": 1.4458634481790754e-05,
      "loss": 0.6354,
      "step": 11714
    },
    {
      "epoch": 3.6990843069150614,
      "grad_norm": 2.921504497528076,
      "learning_rate": 1.4455125956073259e-05,
      "loss": 0.6169,
      "step": 11715
    },
    {
      "epoch": 3.699400063151247,
      "grad_norm": 3.222784996032715,
      "learning_rate": 1.4451617430355765e-05,
      "loss": 0.6824,
      "step": 11716
    },
    {
      "epoch": 3.699715819387433,
      "grad_norm": 3.740973711013794,
      "learning_rate": 1.4448108904638272e-05,
      "loss": 0.6237,
      "step": 11717
    },
    {
      "epoch": 3.7000315756236186,
      "grad_norm": 3.349503993988037,
      "learning_rate": 1.4444600378920777e-05,
      "loss": 0.727,
      "step": 11718
    },
    {
      "epoch": 3.7003473318598044,
      "grad_norm": 3.6996278762817383,
      "learning_rate": 1.4441091853203284e-05,
      "loss": 0.7214,
      "step": 11719
    },
    {
      "epoch": 3.7006630880959897,
      "grad_norm": 3.257884979248047,
      "learning_rate": 1.4437583327485791e-05,
      "loss": 0.5908,
      "step": 11720
    },
    {
      "epoch": 3.700978844332176,
      "grad_norm": 3.766512870788574,
      "learning_rate": 1.4434074801768296e-05,
      "loss": 0.643,
      "step": 11721
    },
    {
      "epoch": 3.701294600568361,
      "grad_norm": 3.0341145992279053,
      "learning_rate": 1.4430566276050805e-05,
      "loss": 0.6077,
      "step": 11722
    },
    {
      "epoch": 3.701610356804547,
      "grad_norm": 3.432422399520874,
      "learning_rate": 1.4427057750333311e-05,
      "loss": 0.6031,
      "step": 11723
    },
    {
      "epoch": 3.7019261130407326,
      "grad_norm": 3.4611406326293945,
      "learning_rate": 1.4423549224615818e-05,
      "loss": 0.6454,
      "step": 11724
    },
    {
      "epoch": 3.7022418692769183,
      "grad_norm": 3.5616097450256348,
      "learning_rate": 1.4420040698898323e-05,
      "loss": 0.6684,
      "step": 11725
    },
    {
      "epoch": 3.702557625513104,
      "grad_norm": 3.486243963241577,
      "learning_rate": 1.441653217318083e-05,
      "loss": 0.6302,
      "step": 11726
    },
    {
      "epoch": 3.7028733817492894,
      "grad_norm": 3.2464871406555176,
      "learning_rate": 1.4413023647463337e-05,
      "loss": 0.6372,
      "step": 11727
    },
    {
      "epoch": 3.703189137985475,
      "grad_norm": 3.087242841720581,
      "learning_rate": 1.4409515121745842e-05,
      "loss": 0.7263,
      "step": 11728
    },
    {
      "epoch": 3.703504894221661,
      "grad_norm": 3.3403608798980713,
      "learning_rate": 1.4406006596028349e-05,
      "loss": 0.6244,
      "step": 11729
    },
    {
      "epoch": 3.7038206504578466,
      "grad_norm": 3.637336015701294,
      "learning_rate": 1.4402498070310854e-05,
      "loss": 0.6244,
      "step": 11730
    },
    {
      "epoch": 3.7041364066940323,
      "grad_norm": 3.1502275466918945,
      "learning_rate": 1.4398989544593364e-05,
      "loss": 0.6366,
      "step": 11731
    },
    {
      "epoch": 3.7044521629302176,
      "grad_norm": 3.5024054050445557,
      "learning_rate": 1.439548101887587e-05,
      "loss": 0.647,
      "step": 11732
    },
    {
      "epoch": 3.704767919166404,
      "grad_norm": 3.5941309928894043,
      "learning_rate": 1.4391972493158376e-05,
      "loss": 0.6932,
      "step": 11733
    },
    {
      "epoch": 3.705083675402589,
      "grad_norm": 3.463111639022827,
      "learning_rate": 1.4388463967440883e-05,
      "loss": 0.6471,
      "step": 11734
    },
    {
      "epoch": 3.705399431638775,
      "grad_norm": 3.544321298599243,
      "learning_rate": 1.4384955441723388e-05,
      "loss": 0.6543,
      "step": 11735
    },
    {
      "epoch": 3.7057151878749606,
      "grad_norm": 3.380706548690796,
      "learning_rate": 1.4381446916005895e-05,
      "loss": 0.68,
      "step": 11736
    },
    {
      "epoch": 3.7060309441111463,
      "grad_norm": 3.8086867332458496,
      "learning_rate": 1.43779383902884e-05,
      "loss": 0.7562,
      "step": 11737
    },
    {
      "epoch": 3.706346700347332,
      "grad_norm": 2.9010751247406006,
      "learning_rate": 1.4374429864570907e-05,
      "loss": 0.5671,
      "step": 11738
    },
    {
      "epoch": 3.7066624565835173,
      "grad_norm": 3.0761003494262695,
      "learning_rate": 1.4370921338853415e-05,
      "loss": 0.6167,
      "step": 11739
    },
    {
      "epoch": 3.706978212819703,
      "grad_norm": 3.6186532974243164,
      "learning_rate": 1.4367412813135922e-05,
      "loss": 0.6591,
      "step": 11740
    },
    {
      "epoch": 3.707293969055889,
      "grad_norm": 3.5029635429382324,
      "learning_rate": 1.4363904287418429e-05,
      "loss": 0.7364,
      "step": 11741
    },
    {
      "epoch": 3.7076097252920746,
      "grad_norm": 6.886469841003418,
      "learning_rate": 1.4360395761700934e-05,
      "loss": 0.6534,
      "step": 11742
    },
    {
      "epoch": 3.7079254815282603,
      "grad_norm": 3.7819995880126953,
      "learning_rate": 1.4356887235983441e-05,
      "loss": 0.6031,
      "step": 11743
    },
    {
      "epoch": 3.7082412377644456,
      "grad_norm": 2.9902663230895996,
      "learning_rate": 1.4353378710265946e-05,
      "loss": 0.6583,
      "step": 11744
    },
    {
      "epoch": 3.7085569940006318,
      "grad_norm": 3.364858388900757,
      "learning_rate": 1.4349870184548453e-05,
      "loss": 0.692,
      "step": 11745
    },
    {
      "epoch": 3.708872750236817,
      "grad_norm": 4.544112682342529,
      "learning_rate": 1.434636165883096e-05,
      "loss": 0.5474,
      "step": 11746
    },
    {
      "epoch": 3.709188506473003,
      "grad_norm": 3.8423373699188232,
      "learning_rate": 1.4342853133113465e-05,
      "loss": 0.7634,
      "step": 11747
    },
    {
      "epoch": 3.7095042627091885,
      "grad_norm": 3.1117637157440186,
      "learning_rate": 1.4339344607395975e-05,
      "loss": 0.5989,
      "step": 11748
    },
    {
      "epoch": 3.7098200189453743,
      "grad_norm": 3.733996868133545,
      "learning_rate": 1.433583608167848e-05,
      "loss": 0.58,
      "step": 11749
    },
    {
      "epoch": 3.71013577518156,
      "grad_norm": 3.406794548034668,
      "learning_rate": 1.4332327555960987e-05,
      "loss": 0.6627,
      "step": 11750
    },
    {
      "epoch": 3.7104515314177453,
      "grad_norm": 3.040355682373047,
      "learning_rate": 1.4328819030243492e-05,
      "loss": 0.6945,
      "step": 11751
    },
    {
      "epoch": 3.710767287653931,
      "grad_norm": 3.278654098510742,
      "learning_rate": 1.4325310504525999e-05,
      "loss": 0.6593,
      "step": 11752
    },
    {
      "epoch": 3.711083043890117,
      "grad_norm": 3.3134310245513916,
      "learning_rate": 1.4321801978808506e-05,
      "loss": 0.7348,
      "step": 11753
    },
    {
      "epoch": 3.7113988001263025,
      "grad_norm": 3.983328342437744,
      "learning_rate": 1.431829345309101e-05,
      "loss": 0.7285,
      "step": 11754
    },
    {
      "epoch": 3.7117145563624883,
      "grad_norm": 3.3416030406951904,
      "learning_rate": 1.4314784927373518e-05,
      "loss": 0.7033,
      "step": 11755
    },
    {
      "epoch": 3.7120303125986736,
      "grad_norm": 3.5336759090423584,
      "learning_rate": 1.4311276401656024e-05,
      "loss": 0.6513,
      "step": 11756
    },
    {
      "epoch": 3.7123460688348597,
      "grad_norm": 5.187918663024902,
      "learning_rate": 1.4307767875938533e-05,
      "loss": 0.7123,
      "step": 11757
    },
    {
      "epoch": 3.712661825071045,
      "grad_norm": 2.928943634033203,
      "learning_rate": 1.4304259350221038e-05,
      "loss": 0.6562,
      "step": 11758
    },
    {
      "epoch": 3.7129775813072308,
      "grad_norm": 2.7515995502471924,
      "learning_rate": 1.4300750824503545e-05,
      "loss": 0.6057,
      "step": 11759
    },
    {
      "epoch": 3.7132933375434165,
      "grad_norm": 3.592884063720703,
      "learning_rate": 1.4297242298786052e-05,
      "loss": 0.7215,
      "step": 11760
    },
    {
      "epoch": 3.7136090937796022,
      "grad_norm": 3.5901238918304443,
      "learning_rate": 1.4293733773068557e-05,
      "loss": 0.6043,
      "step": 11761
    },
    {
      "epoch": 3.713924850015788,
      "grad_norm": 3.351957321166992,
      "learning_rate": 1.4290225247351064e-05,
      "loss": 0.5773,
      "step": 11762
    },
    {
      "epoch": 3.7142406062519733,
      "grad_norm": 2.969334602355957,
      "learning_rate": 1.4286716721633569e-05,
      "loss": 0.6122,
      "step": 11763
    },
    {
      "epoch": 3.714556362488159,
      "grad_norm": 3.6606266498565674,
      "learning_rate": 1.4283208195916076e-05,
      "loss": 0.6816,
      "step": 11764
    },
    {
      "epoch": 3.7148721187243448,
      "grad_norm": 3.0750186443328857,
      "learning_rate": 1.4279699670198582e-05,
      "loss": 0.611,
      "step": 11765
    },
    {
      "epoch": 3.7151878749605305,
      "grad_norm": 4.096304893493652,
      "learning_rate": 1.4276191144481091e-05,
      "loss": 0.7227,
      "step": 11766
    },
    {
      "epoch": 3.7155036311967162,
      "grad_norm": 3.2301459312438965,
      "learning_rate": 1.4272682618763598e-05,
      "loss": 0.6179,
      "step": 11767
    },
    {
      "epoch": 3.715819387432902,
      "grad_norm": 3.143395185470581,
      "learning_rate": 1.4269174093046103e-05,
      "loss": 0.629,
      "step": 11768
    },
    {
      "epoch": 3.7161351436690877,
      "grad_norm": 3.030776262283325,
      "learning_rate": 1.426566556732861e-05,
      "loss": 0.6581,
      "step": 11769
    },
    {
      "epoch": 3.716450899905273,
      "grad_norm": 96.79324340820312,
      "learning_rate": 1.4262157041611115e-05,
      "loss": 0.6075,
      "step": 11770
    },
    {
      "epoch": 3.7167666561414587,
      "grad_norm": 3.942471742630005,
      "learning_rate": 1.4258648515893622e-05,
      "loss": 0.7602,
      "step": 11771
    },
    {
      "epoch": 3.7170824123776445,
      "grad_norm": 3.6216683387756348,
      "learning_rate": 1.4255139990176128e-05,
      "loss": 0.5679,
      "step": 11772
    },
    {
      "epoch": 3.71739816861383,
      "grad_norm": 4.023557662963867,
      "learning_rate": 1.4251631464458633e-05,
      "loss": 0.6631,
      "step": 11773
    },
    {
      "epoch": 3.717713924850016,
      "grad_norm": 3.517063617706299,
      "learning_rate": 1.4248122938741144e-05,
      "loss": 0.6301,
      "step": 11774
    },
    {
      "epoch": 3.7180296810862012,
      "grad_norm": 3.263460159301758,
      "learning_rate": 1.4244614413023649e-05,
      "loss": 0.6633,
      "step": 11775
    },
    {
      "epoch": 3.718345437322387,
      "grad_norm": 4.1915602684021,
      "learning_rate": 1.4241105887306156e-05,
      "loss": 0.7128,
      "step": 11776
    },
    {
      "epoch": 3.7186611935585727,
      "grad_norm": 3.623955726623535,
      "learning_rate": 1.423759736158866e-05,
      "loss": 0.5316,
      "step": 11777
    },
    {
      "epoch": 3.7189769497947585,
      "grad_norm": 3.3745055198669434,
      "learning_rate": 1.4234088835871168e-05,
      "loss": 0.6007,
      "step": 11778
    },
    {
      "epoch": 3.719292706030944,
      "grad_norm": 3.642367362976074,
      "learning_rate": 1.4230580310153674e-05,
      "loss": 0.6787,
      "step": 11779
    },
    {
      "epoch": 3.71960846226713,
      "grad_norm": 4.346708297729492,
      "learning_rate": 1.422707178443618e-05,
      "loss": 0.7481,
      "step": 11780
    },
    {
      "epoch": 3.7199242185033157,
      "grad_norm": 3.375683546066284,
      "learning_rate": 1.4223563258718686e-05,
      "loss": 0.6501,
      "step": 11781
    },
    {
      "epoch": 3.720239974739501,
      "grad_norm": 3.3897225856781006,
      "learning_rate": 1.4220054733001193e-05,
      "loss": 0.6423,
      "step": 11782
    },
    {
      "epoch": 3.7205557309756867,
      "grad_norm": 3.5246059894561768,
      "learning_rate": 1.4216546207283702e-05,
      "loss": 0.6244,
      "step": 11783
    },
    {
      "epoch": 3.7208714872118724,
      "grad_norm": 3.7105839252471924,
      "learning_rate": 1.4213037681566207e-05,
      "loss": 0.6413,
      "step": 11784
    },
    {
      "epoch": 3.721187243448058,
      "grad_norm": 5.185129165649414,
      "learning_rate": 1.4209529155848714e-05,
      "loss": 0.6464,
      "step": 11785
    },
    {
      "epoch": 3.721502999684244,
      "grad_norm": 3.700852155685425,
      "learning_rate": 1.420602063013122e-05,
      "loss": 0.7265,
      "step": 11786
    },
    {
      "epoch": 3.721818755920429,
      "grad_norm": 3.5299952030181885,
      "learning_rate": 1.4202512104413725e-05,
      "loss": 0.6855,
      "step": 11787
    },
    {
      "epoch": 3.722134512156615,
      "grad_norm": 2.8682312965393066,
      "learning_rate": 1.4199003578696232e-05,
      "loss": 0.7364,
      "step": 11788
    },
    {
      "epoch": 3.7224502683928007,
      "grad_norm": 3.579347610473633,
      "learning_rate": 1.4195495052978739e-05,
      "loss": 0.6736,
      "step": 11789
    },
    {
      "epoch": 3.7227660246289864,
      "grad_norm": 3.916919469833374,
      "learning_rate": 1.4191986527261244e-05,
      "loss": 0.6716,
      "step": 11790
    },
    {
      "epoch": 3.723081780865172,
      "grad_norm": 3.383573055267334,
      "learning_rate": 1.4188478001543751e-05,
      "loss": 0.7567,
      "step": 11791
    },
    {
      "epoch": 3.723397537101358,
      "grad_norm": 3.529515027999878,
      "learning_rate": 1.418496947582626e-05,
      "loss": 0.7363,
      "step": 11792
    },
    {
      "epoch": 3.7237132933375436,
      "grad_norm": 2.903275966644287,
      "learning_rate": 1.4181460950108766e-05,
      "loss": 0.6676,
      "step": 11793
    },
    {
      "epoch": 3.724029049573729,
      "grad_norm": 3.5818381309509277,
      "learning_rate": 1.4177952424391271e-05,
      "loss": 0.6459,
      "step": 11794
    },
    {
      "epoch": 3.7243448058099147,
      "grad_norm": 3.087620258331299,
      "learning_rate": 1.4174443898673778e-05,
      "loss": 0.5323,
      "step": 11795
    },
    {
      "epoch": 3.7246605620461004,
      "grad_norm": 3.124783992767334,
      "learning_rate": 1.4170935372956285e-05,
      "loss": 0.742,
      "step": 11796
    },
    {
      "epoch": 3.724976318282286,
      "grad_norm": 2.9961893558502197,
      "learning_rate": 1.416742684723879e-05,
      "loss": 0.6667,
      "step": 11797
    },
    {
      "epoch": 3.725292074518472,
      "grad_norm": 3.4051573276519775,
      "learning_rate": 1.4163918321521297e-05,
      "loss": 0.6933,
      "step": 11798
    },
    {
      "epoch": 3.725607830754657,
      "grad_norm": 3.318897008895874,
      "learning_rate": 1.4160409795803802e-05,
      "loss": 0.7253,
      "step": 11799
    },
    {
      "epoch": 3.725923586990843,
      "grad_norm": 4.862437725067139,
      "learning_rate": 1.4156901270086309e-05,
      "loss": 0.7162,
      "step": 11800
    },
    {
      "epoch": 3.725923586990843,
      "eval_loss": 0.8130756616592407,
      "eval_runtime": 4.5974,
      "eval_samples_per_second": 541.605,
      "eval_steps_per_second": 4.568,
      "step": 11800
    },
    {
      "epoch": 3.7262393432270287,
      "grad_norm": 3.2982418537139893,
      "learning_rate": 1.4153392744368817e-05,
      "loss": 0.6751,
      "step": 11801
    },
    {
      "epoch": 3.7265550994632144,
      "grad_norm": 3.3837075233459473,
      "learning_rate": 1.4149884218651324e-05,
      "loss": 0.7454,
      "step": 11802
    },
    {
      "epoch": 3.7268708556994,
      "grad_norm": 3.019216775894165,
      "learning_rate": 1.4146375692933831e-05,
      "loss": 0.7665,
      "step": 11803
    },
    {
      "epoch": 3.727186611935586,
      "grad_norm": 3.546259641647339,
      "learning_rate": 1.4142867167216336e-05,
      "loss": 0.5623,
      "step": 11804
    },
    {
      "epoch": 3.7275023681717716,
      "grad_norm": 3.201746702194214,
      "learning_rate": 1.4139358641498843e-05,
      "loss": 0.6915,
      "step": 11805
    },
    {
      "epoch": 3.727818124407957,
      "grad_norm": 4.125923156738281,
      "learning_rate": 1.4135850115781348e-05,
      "loss": 0.7078,
      "step": 11806
    },
    {
      "epoch": 3.7281338806441426,
      "grad_norm": 3.3439509868621826,
      "learning_rate": 1.4132341590063855e-05,
      "loss": 0.6408,
      "step": 11807
    },
    {
      "epoch": 3.7284496368803284,
      "grad_norm": 3.2063660621643066,
      "learning_rate": 1.4128833064346362e-05,
      "loss": 0.7223,
      "step": 11808
    },
    {
      "epoch": 3.728765393116514,
      "grad_norm": 3.5878665447235107,
      "learning_rate": 1.412532453862887e-05,
      "loss": 0.6915,
      "step": 11809
    },
    {
      "epoch": 3.7290811493527,
      "grad_norm": 4.185541152954102,
      "learning_rate": 1.4121816012911377e-05,
      "loss": 0.6972,
      "step": 11810
    },
    {
      "epoch": 3.729396905588885,
      "grad_norm": 3.441371440887451,
      "learning_rate": 1.4118307487193882e-05,
      "loss": 0.6857,
      "step": 11811
    },
    {
      "epoch": 3.729712661825071,
      "grad_norm": 3.5213561058044434,
      "learning_rate": 1.4114798961476389e-05,
      "loss": 0.6358,
      "step": 11812
    },
    {
      "epoch": 3.7300284180612566,
      "grad_norm": 4.020318984985352,
      "learning_rate": 1.4111290435758894e-05,
      "loss": 0.6602,
      "step": 11813
    },
    {
      "epoch": 3.7303441742974424,
      "grad_norm": 3.271916151046753,
      "learning_rate": 1.4107781910041401e-05,
      "loss": 0.59,
      "step": 11814
    },
    {
      "epoch": 3.730659930533628,
      "grad_norm": 3.489464282989502,
      "learning_rate": 1.4104273384323908e-05,
      "loss": 0.6755,
      "step": 11815
    },
    {
      "epoch": 3.730975686769814,
      "grad_norm": 2.485746383666992,
      "learning_rate": 1.4100764858606413e-05,
      "loss": 0.5589,
      "step": 11816
    },
    {
      "epoch": 3.7312914430059996,
      "grad_norm": 3.1134986877441406,
      "learning_rate": 1.409725633288892e-05,
      "loss": 0.63,
      "step": 11817
    },
    {
      "epoch": 3.731607199242185,
      "grad_norm": 3.649486780166626,
      "learning_rate": 1.4093747807171428e-05,
      "loss": 0.614,
      "step": 11818
    },
    {
      "epoch": 3.7319229554783706,
      "grad_norm": 3.0625267028808594,
      "learning_rate": 1.4090239281453935e-05,
      "loss": 0.5884,
      "step": 11819
    },
    {
      "epoch": 3.7322387117145563,
      "grad_norm": 3.342928171157837,
      "learning_rate": 1.408673075573644e-05,
      "loss": 0.6075,
      "step": 11820
    },
    {
      "epoch": 3.732554467950742,
      "grad_norm": 3.0703511238098145,
      "learning_rate": 1.4083222230018947e-05,
      "loss": 0.6843,
      "step": 11821
    },
    {
      "epoch": 3.732870224186928,
      "grad_norm": 3.6117846965789795,
      "learning_rate": 1.4079713704301454e-05,
      "loss": 0.628,
      "step": 11822
    },
    {
      "epoch": 3.733185980423113,
      "grad_norm": 3.183870792388916,
      "learning_rate": 1.4076205178583959e-05,
      "loss": 0.623,
      "step": 11823
    },
    {
      "epoch": 3.7335017366592993,
      "grad_norm": 3.230999231338501,
      "learning_rate": 1.4072696652866466e-05,
      "loss": 0.6358,
      "step": 11824
    },
    {
      "epoch": 3.7338174928954846,
      "grad_norm": 3.192481756210327,
      "learning_rate": 1.4069188127148972e-05,
      "loss": 0.5888,
      "step": 11825
    },
    {
      "epoch": 3.7341332491316703,
      "grad_norm": 2.8043484687805176,
      "learning_rate": 1.4065679601431478e-05,
      "loss": 0.5857,
      "step": 11826
    },
    {
      "epoch": 3.734449005367856,
      "grad_norm": 3.3474998474121094,
      "learning_rate": 1.4062171075713986e-05,
      "loss": 0.676,
      "step": 11827
    },
    {
      "epoch": 3.734764761604042,
      "grad_norm": 4.643821716308594,
      "learning_rate": 1.4058662549996493e-05,
      "loss": 0.6407,
      "step": 11828
    },
    {
      "epoch": 3.7350805178402275,
      "grad_norm": 3.1820571422576904,
      "learning_rate": 1.4055154024279e-05,
      "loss": 0.6155,
      "step": 11829
    },
    {
      "epoch": 3.735396274076413,
      "grad_norm": 3.53653621673584,
      "learning_rate": 1.4051645498561505e-05,
      "loss": 0.585,
      "step": 11830
    },
    {
      "epoch": 3.7357120303125986,
      "grad_norm": 3.4999849796295166,
      "learning_rate": 1.4048136972844012e-05,
      "loss": 0.6694,
      "step": 11831
    },
    {
      "epoch": 3.7360277865487843,
      "grad_norm": 2.797762632369995,
      "learning_rate": 1.4044628447126518e-05,
      "loss": 0.6048,
      "step": 11832
    },
    {
      "epoch": 3.73634354278497,
      "grad_norm": 3.6550564765930176,
      "learning_rate": 1.4041119921409024e-05,
      "loss": 0.6454,
      "step": 11833
    },
    {
      "epoch": 3.736659299021156,
      "grad_norm": 3.3680062294006348,
      "learning_rate": 1.403761139569153e-05,
      "loss": 0.6353,
      "step": 11834
    },
    {
      "epoch": 3.736975055257341,
      "grad_norm": 3.376297950744629,
      "learning_rate": 1.4034102869974036e-05,
      "loss": 0.6452,
      "step": 11835
    },
    {
      "epoch": 3.7372908114935273,
      "grad_norm": 3.2628085613250732,
      "learning_rate": 1.4030594344256546e-05,
      "loss": 0.7189,
      "step": 11836
    },
    {
      "epoch": 3.7376065677297126,
      "grad_norm": 3.236349582672119,
      "learning_rate": 1.4027085818539051e-05,
      "loss": 0.5906,
      "step": 11837
    },
    {
      "epoch": 3.7379223239658983,
      "grad_norm": 3.822105884552002,
      "learning_rate": 1.4023577292821558e-05,
      "loss": 0.6452,
      "step": 11838
    },
    {
      "epoch": 3.738238080202084,
      "grad_norm": 4.334226608276367,
      "learning_rate": 1.4020068767104064e-05,
      "loss": 0.7633,
      "step": 11839
    },
    {
      "epoch": 3.7385538364382698,
      "grad_norm": 3.2457234859466553,
      "learning_rate": 1.401656024138657e-05,
      "loss": 0.6048,
      "step": 11840
    },
    {
      "epoch": 3.7388695926744555,
      "grad_norm": 3.4611499309539795,
      "learning_rate": 1.4013051715669076e-05,
      "loss": 0.6352,
      "step": 11841
    },
    {
      "epoch": 3.739185348910641,
      "grad_norm": 3.517929792404175,
      "learning_rate": 1.4009543189951582e-05,
      "loss": 0.6201,
      "step": 11842
    },
    {
      "epoch": 3.7395011051468265,
      "grad_norm": 3.468125820159912,
      "learning_rate": 1.4006034664234088e-05,
      "loss": 0.5956,
      "step": 11843
    },
    {
      "epoch": 3.7398168613830123,
      "grad_norm": 3.571413040161133,
      "learning_rate": 1.4002526138516597e-05,
      "loss": 0.6095,
      "step": 11844
    },
    {
      "epoch": 3.740132617619198,
      "grad_norm": 3.2642030715942383,
      "learning_rate": 1.3999017612799104e-05,
      "loss": 0.6598,
      "step": 11845
    },
    {
      "epoch": 3.7404483738553838,
      "grad_norm": 2.8041484355926514,
      "learning_rate": 1.399550908708161e-05,
      "loss": 0.5935,
      "step": 11846
    },
    {
      "epoch": 3.740764130091569,
      "grad_norm": 3.5171544551849365,
      "learning_rate": 1.3992000561364116e-05,
      "loss": 0.6249,
      "step": 11847
    },
    {
      "epoch": 3.7410798863277552,
      "grad_norm": 3.7552437782287598,
      "learning_rate": 1.3988492035646622e-05,
      "loss": 0.6364,
      "step": 11848
    },
    {
      "epoch": 3.7413956425639405,
      "grad_norm": 3.4947803020477295,
      "learning_rate": 1.3984983509929128e-05,
      "loss": 0.6351,
      "step": 11849
    },
    {
      "epoch": 3.7417113988001263,
      "grad_norm": 3.75815749168396,
      "learning_rate": 1.3981474984211634e-05,
      "loss": 0.7365,
      "step": 11850
    },
    {
      "epoch": 3.742027155036312,
      "grad_norm": 4.042494297027588,
      "learning_rate": 1.3977966458494141e-05,
      "loss": 0.6749,
      "step": 11851
    },
    {
      "epoch": 3.7423429112724977,
      "grad_norm": 3.373100996017456,
      "learning_rate": 1.3974457932776646e-05,
      "loss": 0.4928,
      "step": 11852
    },
    {
      "epoch": 3.7426586675086835,
      "grad_norm": 3.971116304397583,
      "learning_rate": 1.3970949407059156e-05,
      "loss": 0.606,
      "step": 11853
    },
    {
      "epoch": 3.7429744237448688,
      "grad_norm": 3.3935015201568604,
      "learning_rate": 1.3967440881341662e-05,
      "loss": 0.5574,
      "step": 11854
    },
    {
      "epoch": 3.7432901799810545,
      "grad_norm": 2.8863706588745117,
      "learning_rate": 1.3963932355624168e-05,
      "loss": 0.5568,
      "step": 11855
    },
    {
      "epoch": 3.7436059362172402,
      "grad_norm": 4.473369121551514,
      "learning_rate": 1.3960423829906674e-05,
      "loss": 0.6766,
      "step": 11856
    },
    {
      "epoch": 3.743921692453426,
      "grad_norm": 3.5020601749420166,
      "learning_rate": 1.395691530418918e-05,
      "loss": 0.6735,
      "step": 11857
    },
    {
      "epoch": 3.7442374486896117,
      "grad_norm": 3.4790537357330322,
      "learning_rate": 1.3953406778471687e-05,
      "loss": 0.6302,
      "step": 11858
    },
    {
      "epoch": 3.744553204925797,
      "grad_norm": 3.5089919567108154,
      "learning_rate": 1.3949898252754192e-05,
      "loss": 0.5943,
      "step": 11859
    },
    {
      "epoch": 3.744868961161983,
      "grad_norm": 4.272492408752441,
      "learning_rate": 1.3946389727036699e-05,
      "loss": 0.6168,
      "step": 11860
    },
    {
      "epoch": 3.7451847173981685,
      "grad_norm": 4.344923496246338,
      "learning_rate": 1.3942881201319206e-05,
      "loss": 0.6838,
      "step": 11861
    },
    {
      "epoch": 3.7455004736343542,
      "grad_norm": 3.3233892917633057,
      "learning_rate": 1.3939372675601714e-05,
      "loss": 0.5984,
      "step": 11862
    },
    {
      "epoch": 3.74581622987054,
      "grad_norm": 3.049208879470825,
      "learning_rate": 1.393586414988422e-05,
      "loss": 0.6446,
      "step": 11863
    },
    {
      "epoch": 3.7461319861067257,
      "grad_norm": 3.42910099029541,
      "learning_rate": 1.3932355624166726e-05,
      "loss": 0.6114,
      "step": 11864
    },
    {
      "epoch": 3.7464477423429114,
      "grad_norm": 3.2915332317352295,
      "learning_rate": 1.3928847098449233e-05,
      "loss": 0.5805,
      "step": 11865
    },
    {
      "epoch": 3.7467634985790967,
      "grad_norm": 3.229726552963257,
      "learning_rate": 1.3925338572731738e-05,
      "loss": 0.5526,
      "step": 11866
    },
    {
      "epoch": 3.7470792548152825,
      "grad_norm": 3.4758622646331787,
      "learning_rate": 1.3921830047014245e-05,
      "loss": 0.6774,
      "step": 11867
    },
    {
      "epoch": 3.747395011051468,
      "grad_norm": 2.7833569049835205,
      "learning_rate": 1.391832152129675e-05,
      "loss": 0.6332,
      "step": 11868
    },
    {
      "epoch": 3.747710767287654,
      "grad_norm": 3.5200283527374268,
      "learning_rate": 1.3914812995579257e-05,
      "loss": 0.7153,
      "step": 11869
    },
    {
      "epoch": 3.7480265235238397,
      "grad_norm": 4.215363025665283,
      "learning_rate": 1.3911304469861764e-05,
      "loss": 0.7652,
      "step": 11870
    },
    {
      "epoch": 3.748342279760025,
      "grad_norm": 3.5763797760009766,
      "learning_rate": 1.3907795944144272e-05,
      "loss": 0.6917,
      "step": 11871
    },
    {
      "epoch": 3.748658035996211,
      "grad_norm": 3.8154497146606445,
      "learning_rate": 1.3904287418426779e-05,
      "loss": 0.6942,
      "step": 11872
    },
    {
      "epoch": 3.7489737922323965,
      "grad_norm": 3.529417037963867,
      "learning_rate": 1.3900778892709284e-05,
      "loss": 0.6763,
      "step": 11873
    },
    {
      "epoch": 3.749289548468582,
      "grad_norm": 3.5389504432678223,
      "learning_rate": 1.3897270366991791e-05,
      "loss": 0.6258,
      "step": 11874
    },
    {
      "epoch": 3.749605304704768,
      "grad_norm": 3.341639995574951,
      "learning_rate": 1.3893761841274296e-05,
      "loss": 0.739,
      "step": 11875
    },
    {
      "epoch": 3.7499210609409537,
      "grad_norm": 3.432187080383301,
      "learning_rate": 1.3890253315556803e-05,
      "loss": 0.5823,
      "step": 11876
    },
    {
      "epoch": 3.7502368171771394,
      "grad_norm": 3.9948067665100098,
      "learning_rate": 1.388674478983931e-05,
      "loss": 0.6508,
      "step": 11877
    },
    {
      "epoch": 3.7505525734133247,
      "grad_norm": 3.40948748588562,
      "learning_rate": 1.3883236264121815e-05,
      "loss": 0.6217,
      "step": 11878
    },
    {
      "epoch": 3.7508683296495104,
      "grad_norm": 3.74528169631958,
      "learning_rate": 1.3879727738404325e-05,
      "loss": 0.6881,
      "step": 11879
    },
    {
      "epoch": 3.751184085885696,
      "grad_norm": 3.588254928588867,
      "learning_rate": 1.387621921268683e-05,
      "loss": 0.6808,
      "step": 11880
    },
    {
      "epoch": 3.751499842121882,
      "grad_norm": 3.0798697471618652,
      "learning_rate": 1.3872710686969337e-05,
      "loss": 0.6304,
      "step": 11881
    },
    {
      "epoch": 3.7518155983580677,
      "grad_norm": 3.7490506172180176,
      "learning_rate": 1.3869202161251842e-05,
      "loss": 0.6607,
      "step": 11882
    },
    {
      "epoch": 3.7521313545942534,
      "grad_norm": 3.6325266361236572,
      "learning_rate": 1.3865693635534349e-05,
      "loss": 0.694,
      "step": 11883
    },
    {
      "epoch": 3.752447110830439,
      "grad_norm": 4.015353202819824,
      "learning_rate": 1.3862185109816856e-05,
      "loss": 0.6698,
      "step": 11884
    },
    {
      "epoch": 3.7527628670666244,
      "grad_norm": 3.589782953262329,
      "learning_rate": 1.3858676584099361e-05,
      "loss": 0.6082,
      "step": 11885
    },
    {
      "epoch": 3.75307862330281,
      "grad_norm": 3.4827358722686768,
      "learning_rate": 1.3855168058381868e-05,
      "loss": 0.6849,
      "step": 11886
    },
    {
      "epoch": 3.753394379538996,
      "grad_norm": 3.1258134841918945,
      "learning_rate": 1.3851659532664375e-05,
      "loss": 0.6901,
      "step": 11887
    },
    {
      "epoch": 3.7537101357751816,
      "grad_norm": 3.8559916019439697,
      "learning_rate": 1.3848151006946883e-05,
      "loss": 0.6732,
      "step": 11888
    },
    {
      "epoch": 3.7540258920113674,
      "grad_norm": 3.7251598834991455,
      "learning_rate": 1.3844642481229388e-05,
      "loss": 0.6855,
      "step": 11889
    },
    {
      "epoch": 3.7543416482475527,
      "grad_norm": 3.311872720718384,
      "learning_rate": 1.3841133955511895e-05,
      "loss": 0.5902,
      "step": 11890
    },
    {
      "epoch": 3.7546574044837384,
      "grad_norm": 3.6662395000457764,
      "learning_rate": 1.3837625429794402e-05,
      "loss": 0.6964,
      "step": 11891
    },
    {
      "epoch": 3.754973160719924,
      "grad_norm": 3.123975992202759,
      "learning_rate": 1.3834116904076907e-05,
      "loss": 0.6244,
      "step": 11892
    },
    {
      "epoch": 3.75528891695611,
      "grad_norm": 3.421643018722534,
      "learning_rate": 1.3830608378359414e-05,
      "loss": 0.6894,
      "step": 11893
    },
    {
      "epoch": 3.7556046731922956,
      "grad_norm": 3.3815581798553467,
      "learning_rate": 1.382709985264192e-05,
      "loss": 0.6159,
      "step": 11894
    },
    {
      "epoch": 3.7559204294284814,
      "grad_norm": 3.854343891143799,
      "learning_rate": 1.3823591326924426e-05,
      "loss": 0.7048,
      "step": 11895
    },
    {
      "epoch": 3.756236185664667,
      "grad_norm": 3.1025795936584473,
      "learning_rate": 1.3820082801206932e-05,
      "loss": 0.7268,
      "step": 11896
    },
    {
      "epoch": 3.7565519419008524,
      "grad_norm": 3.1971728801727295,
      "learning_rate": 1.3816574275489441e-05,
      "loss": 0.6439,
      "step": 11897
    },
    {
      "epoch": 3.756867698137038,
      "grad_norm": 3.187647581100464,
      "learning_rate": 1.3813065749771948e-05,
      "loss": 0.6837,
      "step": 11898
    },
    {
      "epoch": 3.757183454373224,
      "grad_norm": 2.863187551498413,
      "learning_rate": 1.3809557224054453e-05,
      "loss": 0.64,
      "step": 11899
    },
    {
      "epoch": 3.7574992106094096,
      "grad_norm": 3.4472668170928955,
      "learning_rate": 1.380604869833696e-05,
      "loss": 0.6659,
      "step": 11900
    },
    {
      "epoch": 3.7578149668455954,
      "grad_norm": 3.6409053802490234,
      "learning_rate": 1.3802540172619467e-05,
      "loss": 0.7235,
      "step": 11901
    },
    {
      "epoch": 3.7581307230817806,
      "grad_norm": 3.148333787918091,
      "learning_rate": 1.3799031646901972e-05,
      "loss": 0.6001,
      "step": 11902
    },
    {
      "epoch": 3.7584464793179664,
      "grad_norm": 3.2393388748168945,
      "learning_rate": 1.3795523121184478e-05,
      "loss": 0.7342,
      "step": 11903
    },
    {
      "epoch": 3.758762235554152,
      "grad_norm": 2.9780943393707275,
      "learning_rate": 1.3792014595466984e-05,
      "loss": 0.6352,
      "step": 11904
    },
    {
      "epoch": 3.759077991790338,
      "grad_norm": 3.6292989253997803,
      "learning_rate": 1.378850606974949e-05,
      "loss": 0.715,
      "step": 11905
    },
    {
      "epoch": 3.7593937480265236,
      "grad_norm": 3.622738838195801,
      "learning_rate": 1.3784997544031999e-05,
      "loss": 0.6288,
      "step": 11906
    },
    {
      "epoch": 3.7597095042627093,
      "grad_norm": 3.596932888031006,
      "learning_rate": 1.3781489018314506e-05,
      "loss": 0.6681,
      "step": 11907
    },
    {
      "epoch": 3.760025260498895,
      "grad_norm": 3.720738172531128,
      "learning_rate": 1.3777980492597013e-05,
      "loss": 0.6472,
      "step": 11908
    },
    {
      "epoch": 3.7603410167350804,
      "grad_norm": 3.1751832962036133,
      "learning_rate": 1.3774471966879518e-05,
      "loss": 0.6156,
      "step": 11909
    },
    {
      "epoch": 3.760656772971266,
      "grad_norm": 3.582226514816284,
      "learning_rate": 1.3770963441162024e-05,
      "loss": 0.6637,
      "step": 11910
    },
    {
      "epoch": 3.760972529207452,
      "grad_norm": 3.163022041320801,
      "learning_rate": 1.376745491544453e-05,
      "loss": 0.676,
      "step": 11911
    },
    {
      "epoch": 3.7612882854436376,
      "grad_norm": 4.226089000701904,
      "learning_rate": 1.3763946389727036e-05,
      "loss": 0.7566,
      "step": 11912
    },
    {
      "epoch": 3.7616040416798233,
      "grad_norm": 4.290429592132568,
      "learning_rate": 1.3760437864009543e-05,
      "loss": 0.6922,
      "step": 11913
    },
    {
      "epoch": 3.7619197979160086,
      "grad_norm": 3.5971624851226807,
      "learning_rate": 1.3756929338292052e-05,
      "loss": 0.6459,
      "step": 11914
    },
    {
      "epoch": 3.7622355541521944,
      "grad_norm": 3.4207561016082764,
      "learning_rate": 1.3753420812574559e-05,
      "loss": 0.7022,
      "step": 11915
    },
    {
      "epoch": 3.76255131038838,
      "grad_norm": 3.2826879024505615,
      "learning_rate": 1.3749912286857064e-05,
      "loss": 0.6436,
      "step": 11916
    },
    {
      "epoch": 3.762867066624566,
      "grad_norm": 3.3623993396759033,
      "learning_rate": 1.374640376113957e-05,
      "loss": 0.6014,
      "step": 11917
    },
    {
      "epoch": 3.7631828228607516,
      "grad_norm": 3.6361303329467773,
      "learning_rate": 1.3742895235422076e-05,
      "loss": 0.6883,
      "step": 11918
    },
    {
      "epoch": 3.7634985790969373,
      "grad_norm": 2.907144784927368,
      "learning_rate": 1.3739386709704582e-05,
      "loss": 0.5588,
      "step": 11919
    },
    {
      "epoch": 3.763814335333123,
      "grad_norm": 3.2467541694641113,
      "learning_rate": 1.373587818398709e-05,
      "loss": 0.6838,
      "step": 11920
    },
    {
      "epoch": 3.7641300915693083,
      "grad_norm": 3.81398606300354,
      "learning_rate": 1.3732369658269594e-05,
      "loss": 0.6892,
      "step": 11921
    },
    {
      "epoch": 3.764445847805494,
      "grad_norm": 4.989102363586426,
      "learning_rate": 1.3728861132552101e-05,
      "loss": 0.7597,
      "step": 11922
    },
    {
      "epoch": 3.76476160404168,
      "grad_norm": 3.373180389404297,
      "learning_rate": 1.372535260683461e-05,
      "loss": 0.6763,
      "step": 11923
    },
    {
      "epoch": 3.7650773602778655,
      "grad_norm": 4.115656852722168,
      "learning_rate": 1.3721844081117116e-05,
      "loss": 0.7355,
      "step": 11924
    },
    {
      "epoch": 3.7653931165140513,
      "grad_norm": 3.355008602142334,
      "learning_rate": 1.3718335555399622e-05,
      "loss": 0.634,
      "step": 11925
    },
    {
      "epoch": 3.7657088727502366,
      "grad_norm": 3.4906005859375,
      "learning_rate": 1.3714827029682128e-05,
      "loss": 0.6394,
      "step": 11926
    },
    {
      "epoch": 3.7660246289864223,
      "grad_norm": 3.416020154953003,
      "learning_rate": 1.3711318503964635e-05,
      "loss": 0.6723,
      "step": 11927
    },
    {
      "epoch": 3.766340385222608,
      "grad_norm": 2.9585978984832764,
      "learning_rate": 1.370780997824714e-05,
      "loss": 0.6013,
      "step": 11928
    },
    {
      "epoch": 3.766656141458794,
      "grad_norm": 2.647291898727417,
      "learning_rate": 1.3704301452529647e-05,
      "loss": 0.5966,
      "step": 11929
    },
    {
      "epoch": 3.7669718976949795,
      "grad_norm": 3.8930792808532715,
      "learning_rate": 1.3700792926812154e-05,
      "loss": 0.6465,
      "step": 11930
    },
    {
      "epoch": 3.7672876539311653,
      "grad_norm": 3.195244789123535,
      "learning_rate": 1.3697284401094659e-05,
      "loss": 0.6298,
      "step": 11931
    },
    {
      "epoch": 3.767603410167351,
      "grad_norm": 3.5947256088256836,
      "learning_rate": 1.3693775875377168e-05,
      "loss": 0.5559,
      "step": 11932
    },
    {
      "epoch": 3.7679191664035363,
      "grad_norm": 3.4875564575195312,
      "learning_rate": 1.3690267349659674e-05,
      "loss": 0.6218,
      "step": 11933
    },
    {
      "epoch": 3.768234922639722,
      "grad_norm": 3.8960044384002686,
      "learning_rate": 1.3686758823942181e-05,
      "loss": 0.5673,
      "step": 11934
    },
    {
      "epoch": 3.7685506788759078,
      "grad_norm": 3.30354642868042,
      "learning_rate": 1.3683250298224686e-05,
      "loss": 0.6629,
      "step": 11935
    },
    {
      "epoch": 3.7688664351120935,
      "grad_norm": 3.1575498580932617,
      "learning_rate": 1.3679741772507193e-05,
      "loss": 0.7345,
      "step": 11936
    },
    {
      "epoch": 3.7691821913482793,
      "grad_norm": 3.4753687381744385,
      "learning_rate": 1.36762332467897e-05,
      "loss": 0.7039,
      "step": 11937
    },
    {
      "epoch": 3.7694979475844645,
      "grad_norm": 3.5427536964416504,
      "learning_rate": 1.3672724721072205e-05,
      "loss": 0.7217,
      "step": 11938
    },
    {
      "epoch": 3.7698137038206507,
      "grad_norm": 3.4528229236602783,
      "learning_rate": 1.3669216195354712e-05,
      "loss": 0.6935,
      "step": 11939
    },
    {
      "epoch": 3.770129460056836,
      "grad_norm": 3.167511224746704,
      "learning_rate": 1.3665707669637217e-05,
      "loss": 0.6447,
      "step": 11940
    },
    {
      "epoch": 3.7704452162930218,
      "grad_norm": 3.439257860183716,
      "learning_rate": 1.3662199143919727e-05,
      "loss": 0.6981,
      "step": 11941
    },
    {
      "epoch": 3.7707609725292075,
      "grad_norm": 3.601348638534546,
      "learning_rate": 1.3658690618202232e-05,
      "loss": 0.5812,
      "step": 11942
    },
    {
      "epoch": 3.7710767287653932,
      "grad_norm": 2.9741036891937256,
      "learning_rate": 1.3655182092484739e-05,
      "loss": 0.6273,
      "step": 11943
    },
    {
      "epoch": 3.771392485001579,
      "grad_norm": 3.1530237197875977,
      "learning_rate": 1.3651673566767246e-05,
      "loss": 0.5813,
      "step": 11944
    },
    {
      "epoch": 3.7717082412377643,
      "grad_norm": 4.453663349151611,
      "learning_rate": 1.3648165041049751e-05,
      "loss": 0.6429,
      "step": 11945
    },
    {
      "epoch": 3.77202399747395,
      "grad_norm": 3.5640408992767334,
      "learning_rate": 1.3644656515332258e-05,
      "loss": 0.6471,
      "step": 11946
    },
    {
      "epoch": 3.7723397537101357,
      "grad_norm": 3.5740063190460205,
      "learning_rate": 1.3641147989614763e-05,
      "loss": 0.6205,
      "step": 11947
    },
    {
      "epoch": 3.7726555099463215,
      "grad_norm": 3.4652884006500244,
      "learning_rate": 1.363763946389727e-05,
      "loss": 0.7204,
      "step": 11948
    },
    {
      "epoch": 3.772971266182507,
      "grad_norm": 3.3052163124084473,
      "learning_rate": 1.3634130938179778e-05,
      "loss": 0.6539,
      "step": 11949
    },
    {
      "epoch": 3.7732870224186925,
      "grad_norm": 3.8745369911193848,
      "learning_rate": 1.3630622412462285e-05,
      "loss": 0.7394,
      "step": 11950
    },
    {
      "epoch": 3.7736027786548787,
      "grad_norm": 2.7580108642578125,
      "learning_rate": 1.3627113886744792e-05,
      "loss": 0.7125,
      "step": 11951
    },
    {
      "epoch": 3.773918534891064,
      "grad_norm": 3.044476270675659,
      "learning_rate": 1.3623605361027297e-05,
      "loss": 0.7278,
      "step": 11952
    },
    {
      "epoch": 3.7742342911272497,
      "grad_norm": 3.077692985534668,
      "learning_rate": 1.3620096835309804e-05,
      "loss": 0.6411,
      "step": 11953
    },
    {
      "epoch": 3.7745500473634355,
      "grad_norm": 2.9574756622314453,
      "learning_rate": 1.3616588309592309e-05,
      "loss": 0.6344,
      "step": 11954
    },
    {
      "epoch": 3.774865803599621,
      "grad_norm": 4.030107021331787,
      "learning_rate": 1.3613079783874816e-05,
      "loss": 0.7039,
      "step": 11955
    },
    {
      "epoch": 3.775181559835807,
      "grad_norm": 2.8090431690216064,
      "learning_rate": 1.3609571258157323e-05,
      "loss": 0.6645,
      "step": 11956
    },
    {
      "epoch": 3.7754973160719922,
      "grad_norm": 3.0689098834991455,
      "learning_rate": 1.3606062732439828e-05,
      "loss": 0.6094,
      "step": 11957
    },
    {
      "epoch": 3.775813072308178,
      "grad_norm": 3.5997354984283447,
      "learning_rate": 1.3602554206722338e-05,
      "loss": 0.5916,
      "step": 11958
    },
    {
      "epoch": 3.7761288285443637,
      "grad_norm": 3.817105531692505,
      "learning_rate": 1.3599045681004843e-05,
      "loss": 0.6857,
      "step": 11959
    },
    {
      "epoch": 3.7764445847805495,
      "grad_norm": 3.340334177017212,
      "learning_rate": 1.359553715528735e-05,
      "loss": 0.7373,
      "step": 11960
    },
    {
      "epoch": 3.776760341016735,
      "grad_norm": 2.9831314086914062,
      "learning_rate": 1.3592028629569855e-05,
      "loss": 0.5929,
      "step": 11961
    },
    {
      "epoch": 3.7770760972529205,
      "grad_norm": 3.8207476139068604,
      "learning_rate": 1.3588520103852362e-05,
      "loss": 0.6019,
      "step": 11962
    },
    {
      "epoch": 3.7773918534891067,
      "grad_norm": 2.939073324203491,
      "learning_rate": 1.3585011578134869e-05,
      "loss": 0.6304,
      "step": 11963
    },
    {
      "epoch": 3.777707609725292,
      "grad_norm": 3.3074798583984375,
      "learning_rate": 1.3581503052417374e-05,
      "loss": 0.6584,
      "step": 11964
    },
    {
      "epoch": 3.7780233659614777,
      "grad_norm": 3.0884358882904053,
      "learning_rate": 1.357799452669988e-05,
      "loss": 0.6603,
      "step": 11965
    },
    {
      "epoch": 3.7783391221976634,
      "grad_norm": 2.965087652206421,
      "learning_rate": 1.3574486000982387e-05,
      "loss": 0.668,
      "step": 11966
    },
    {
      "epoch": 3.778654878433849,
      "grad_norm": 3.736393690109253,
      "learning_rate": 1.3570977475264896e-05,
      "loss": 0.5697,
      "step": 11967
    },
    {
      "epoch": 3.778970634670035,
      "grad_norm": 3.1521174907684326,
      "learning_rate": 1.3567468949547401e-05,
      "loss": 0.6291,
      "step": 11968
    },
    {
      "epoch": 3.77928639090622,
      "grad_norm": 3.043025493621826,
      "learning_rate": 1.3563960423829908e-05,
      "loss": 0.6572,
      "step": 11969
    },
    {
      "epoch": 3.779602147142406,
      "grad_norm": 3.7786169052124023,
      "learning_rate": 1.3560451898112415e-05,
      "loss": 0.697,
      "step": 11970
    },
    {
      "epoch": 3.7799179033785917,
      "grad_norm": 3.319929361343384,
      "learning_rate": 1.355694337239492e-05,
      "loss": 0.6125,
      "step": 11971
    },
    {
      "epoch": 3.7802336596147774,
      "grad_norm": 3.6686439514160156,
      "learning_rate": 1.3553434846677427e-05,
      "loss": 0.5646,
      "step": 11972
    },
    {
      "epoch": 3.780549415850963,
      "grad_norm": 3.607499837875366,
      "learning_rate": 1.3549926320959933e-05,
      "loss": 0.7096,
      "step": 11973
    },
    {
      "epoch": 3.7808651720871485,
      "grad_norm": 4.801855087280273,
      "learning_rate": 1.3546417795242438e-05,
      "loss": 0.6077,
      "step": 11974
    },
    {
      "epoch": 3.7811809283233346,
      "grad_norm": 3.893686532974243,
      "learning_rate": 1.3542909269524947e-05,
      "loss": 0.65,
      "step": 11975
    },
    {
      "epoch": 3.78149668455952,
      "grad_norm": 4.2268595695495605,
      "learning_rate": 1.3539400743807454e-05,
      "loss": 0.7207,
      "step": 11976
    },
    {
      "epoch": 3.7818124407957057,
      "grad_norm": 3.0394480228424072,
      "learning_rate": 1.353589221808996e-05,
      "loss": 0.6037,
      "step": 11977
    },
    {
      "epoch": 3.7821281970318914,
      "grad_norm": 3.405853509902954,
      "learning_rate": 1.3532383692372466e-05,
      "loss": 0.6451,
      "step": 11978
    },
    {
      "epoch": 3.782443953268077,
      "grad_norm": 3.477038621902466,
      "learning_rate": 1.3528875166654973e-05,
      "loss": 0.713,
      "step": 11979
    },
    {
      "epoch": 3.782759709504263,
      "grad_norm": 3.8095600605010986,
      "learning_rate": 1.3525366640937478e-05,
      "loss": 0.6108,
      "step": 11980
    },
    {
      "epoch": 3.783075465740448,
      "grad_norm": 3.467895030975342,
      "learning_rate": 1.3521858115219984e-05,
      "loss": 0.6309,
      "step": 11981
    },
    {
      "epoch": 3.783391221976634,
      "grad_norm": 3.865335702896118,
      "learning_rate": 1.3518349589502491e-05,
      "loss": 0.5952,
      "step": 11982
    },
    {
      "epoch": 3.7837069782128196,
      "grad_norm": 3.7475504875183105,
      "learning_rate": 1.3514841063784996e-05,
      "loss": 0.7082,
      "step": 11983
    },
    {
      "epoch": 3.7840227344490054,
      "grad_norm": 3.496267557144165,
      "learning_rate": 1.3511332538067507e-05,
      "loss": 0.6715,
      "step": 11984
    },
    {
      "epoch": 3.784338490685191,
      "grad_norm": 3.9610962867736816,
      "learning_rate": 1.3507824012350012e-05,
      "loss": 0.6128,
      "step": 11985
    },
    {
      "epoch": 3.784654246921377,
      "grad_norm": 3.123345136642456,
      "learning_rate": 1.3504315486632519e-05,
      "loss": 0.6562,
      "step": 11986
    },
    {
      "epoch": 3.7849700031575626,
      "grad_norm": 3.6490392684936523,
      "learning_rate": 1.3500806960915024e-05,
      "loss": 0.6872,
      "step": 11987
    },
    {
      "epoch": 3.785285759393748,
      "grad_norm": 3.2117044925689697,
      "learning_rate": 1.349729843519753e-05,
      "loss": 0.7401,
      "step": 11988
    },
    {
      "epoch": 3.7856015156299336,
      "grad_norm": 3.601372003555298,
      "learning_rate": 1.3493789909480037e-05,
      "loss": 0.6465,
      "step": 11989
    },
    {
      "epoch": 3.7859172718661194,
      "grad_norm": 3.8819117546081543,
      "learning_rate": 1.3490281383762542e-05,
      "loss": 0.6939,
      "step": 11990
    },
    {
      "epoch": 3.786233028102305,
      "grad_norm": 3.267878770828247,
      "learning_rate": 1.348677285804505e-05,
      "loss": 0.6274,
      "step": 11991
    },
    {
      "epoch": 3.786548784338491,
      "grad_norm": 2.893080234527588,
      "learning_rate": 1.3483264332327556e-05,
      "loss": 0.6538,
      "step": 11992
    },
    {
      "epoch": 3.786864540574676,
      "grad_norm": 3.0496246814727783,
      "learning_rate": 1.3479755806610065e-05,
      "loss": 0.6439,
      "step": 11993
    },
    {
      "epoch": 3.787180296810862,
      "grad_norm": 3.116791248321533,
      "learning_rate": 1.347624728089257e-05,
      "loss": 0.6503,
      "step": 11994
    },
    {
      "epoch": 3.7874960530470476,
      "grad_norm": 3.309582471847534,
      "learning_rate": 1.3472738755175076e-05,
      "loss": 0.615,
      "step": 11995
    },
    {
      "epoch": 3.7878118092832334,
      "grad_norm": 3.1620895862579346,
      "learning_rate": 1.3469230229457583e-05,
      "loss": 0.7409,
      "step": 11996
    },
    {
      "epoch": 3.788127565519419,
      "grad_norm": 3.801072597503662,
      "learning_rate": 1.3465721703740088e-05,
      "loss": 0.7095,
      "step": 11997
    },
    {
      "epoch": 3.788443321755605,
      "grad_norm": 3.6890900135040283,
      "learning_rate": 1.3462213178022595e-05,
      "loss": 0.659,
      "step": 11998
    },
    {
      "epoch": 3.7887590779917906,
      "grad_norm": 3.0371243953704834,
      "learning_rate": 1.3458704652305102e-05,
      "loss": 0.5955,
      "step": 11999
    },
    {
      "epoch": 3.789074834227976,
      "grad_norm": 3.383124828338623,
      "learning_rate": 1.3455196126587607e-05,
      "loss": 0.6931,
      "step": 12000
    },
    {
      "epoch": 3.789074834227976,
      "eval_loss": 0.8041130900382996,
      "eval_runtime": 4.6014,
      "eval_samples_per_second": 541.142,
      "eval_steps_per_second": 4.564,
      "step": 12000
    },
    {
      "epoch": 3.7893905904641616,
      "grad_norm": 3.424323320388794,
      "learning_rate": 1.3451687600870114e-05,
      "loss": 0.5819,
      "step": 12001
    },
    {
      "epoch": 3.7897063467003473,
      "grad_norm": 3.30248761177063,
      "learning_rate": 1.3448179075152622e-05,
      "loss": 0.5847,
      "step": 12002
    },
    {
      "epoch": 3.790022102936533,
      "grad_norm": 3.3242015838623047,
      "learning_rate": 1.344467054943513e-05,
      "loss": 0.7233,
      "step": 12003
    },
    {
      "epoch": 3.790337859172719,
      "grad_norm": 3.0512640476226807,
      "learning_rate": 1.3441162023717634e-05,
      "loss": 0.6307,
      "step": 12004
    },
    {
      "epoch": 3.790653615408904,
      "grad_norm": 3.481388807296753,
      "learning_rate": 1.3437653498000141e-05,
      "loss": 0.6504,
      "step": 12005
    },
    {
      "epoch": 3.79096937164509,
      "grad_norm": 3.162278652191162,
      "learning_rate": 1.3434144972282648e-05,
      "loss": 0.6186,
      "step": 12006
    },
    {
      "epoch": 3.7912851278812756,
      "grad_norm": 3.0203583240509033,
      "learning_rate": 1.3430636446565153e-05,
      "loss": 0.5609,
      "step": 12007
    },
    {
      "epoch": 3.7916008841174613,
      "grad_norm": 2.6294004917144775,
      "learning_rate": 1.342712792084766e-05,
      "loss": 0.6268,
      "step": 12008
    },
    {
      "epoch": 3.791916640353647,
      "grad_norm": 2.9064085483551025,
      "learning_rate": 1.3423619395130165e-05,
      "loss": 0.5434,
      "step": 12009
    },
    {
      "epoch": 3.792232396589833,
      "grad_norm": 3.119554042816162,
      "learning_rate": 1.3420110869412675e-05,
      "loss": 0.6039,
      "step": 12010
    },
    {
      "epoch": 3.7925481528260185,
      "grad_norm": 3.9986538887023926,
      "learning_rate": 1.341660234369518e-05,
      "loss": 0.6541,
      "step": 12011
    },
    {
      "epoch": 3.792863909062204,
      "grad_norm": 3.6055235862731934,
      "learning_rate": 1.3413093817977687e-05,
      "loss": 0.8042,
      "step": 12012
    },
    {
      "epoch": 3.7931796652983896,
      "grad_norm": 3.089738130569458,
      "learning_rate": 1.3409585292260194e-05,
      "loss": 0.5993,
      "step": 12013
    },
    {
      "epoch": 3.7934954215345753,
      "grad_norm": 2.8862977027893066,
      "learning_rate": 1.34060767665427e-05,
      "loss": 0.5663,
      "step": 12014
    },
    {
      "epoch": 3.793811177770761,
      "grad_norm": 3.992905855178833,
      "learning_rate": 1.3402568240825206e-05,
      "loss": 0.6547,
      "step": 12015
    },
    {
      "epoch": 3.794126934006947,
      "grad_norm": 3.2863738536834717,
      "learning_rate": 1.3399059715107711e-05,
      "loss": 0.6195,
      "step": 12016
    },
    {
      "epoch": 3.794442690243132,
      "grad_norm": 5.251259803771973,
      "learning_rate": 1.3395551189390218e-05,
      "loss": 0.6489,
      "step": 12017
    },
    {
      "epoch": 3.794758446479318,
      "grad_norm": 3.393339157104492,
      "learning_rate": 1.3392042663672725e-05,
      "loss": 0.5927,
      "step": 12018
    },
    {
      "epoch": 3.7950742027155036,
      "grad_norm": 3.1785707473754883,
      "learning_rate": 1.3388534137955233e-05,
      "loss": 0.5349,
      "step": 12019
    },
    {
      "epoch": 3.7953899589516893,
      "grad_norm": 3.816955804824829,
      "learning_rate": 1.338502561223774e-05,
      "loss": 0.6731,
      "step": 12020
    },
    {
      "epoch": 3.795705715187875,
      "grad_norm": 5.76211404800415,
      "learning_rate": 1.3381517086520245e-05,
      "loss": 0.7138,
      "step": 12021
    },
    {
      "epoch": 3.7960214714240608,
      "grad_norm": 3.891666889190674,
      "learning_rate": 1.3378008560802752e-05,
      "loss": 0.7405,
      "step": 12022
    },
    {
      "epoch": 3.7963372276602465,
      "grad_norm": 3.086029052734375,
      "learning_rate": 1.3374500035085257e-05,
      "loss": 0.6173,
      "step": 12023
    },
    {
      "epoch": 3.796652983896432,
      "grad_norm": 3.399629831314087,
      "learning_rate": 1.3370991509367764e-05,
      "loss": 0.6425,
      "step": 12024
    },
    {
      "epoch": 3.7969687401326175,
      "grad_norm": 3.118185043334961,
      "learning_rate": 1.336748298365027e-05,
      "loss": 0.6106,
      "step": 12025
    },
    {
      "epoch": 3.7972844963688033,
      "grad_norm": 3.189640998840332,
      "learning_rate": 1.3363974457932776e-05,
      "loss": 0.5234,
      "step": 12026
    },
    {
      "epoch": 3.797600252604989,
      "grad_norm": 3.9893455505371094,
      "learning_rate": 1.3360465932215283e-05,
      "loss": 0.6171,
      "step": 12027
    },
    {
      "epoch": 3.7979160088411748,
      "grad_norm": 3.1049859523773193,
      "learning_rate": 1.3356957406497791e-05,
      "loss": 0.7094,
      "step": 12028
    },
    {
      "epoch": 3.79823176507736,
      "grad_norm": 4.012876033782959,
      "learning_rate": 1.3353448880780298e-05,
      "loss": 0.6868,
      "step": 12029
    },
    {
      "epoch": 3.798547521313546,
      "grad_norm": 3.622222661972046,
      "learning_rate": 1.3349940355062803e-05,
      "loss": 0.6043,
      "step": 12030
    },
    {
      "epoch": 3.7988632775497315,
      "grad_norm": 3.370173692703247,
      "learning_rate": 1.334643182934531e-05,
      "loss": 0.6809,
      "step": 12031
    },
    {
      "epoch": 3.7991790337859173,
      "grad_norm": 3.8174660205841064,
      "learning_rate": 1.3342923303627817e-05,
      "loss": 0.6719,
      "step": 12032
    },
    {
      "epoch": 3.799494790022103,
      "grad_norm": 3.4642391204833984,
      "learning_rate": 1.3339414777910322e-05,
      "loss": 0.6297,
      "step": 12033
    },
    {
      "epoch": 3.7998105462582887,
      "grad_norm": 3.2128653526306152,
      "learning_rate": 1.3335906252192829e-05,
      "loss": 0.6553,
      "step": 12034
    },
    {
      "epoch": 3.8001263024944745,
      "grad_norm": 3.323014736175537,
      "learning_rate": 1.3332397726475335e-05,
      "loss": 0.7177,
      "step": 12035
    },
    {
      "epoch": 3.8004420587306598,
      "grad_norm": 3.2773399353027344,
      "learning_rate": 1.332888920075784e-05,
      "loss": 0.6583,
      "step": 12036
    },
    {
      "epoch": 3.8007578149668455,
      "grad_norm": 3.593698740005493,
      "learning_rate": 1.3325380675040349e-05,
      "loss": 0.6908,
      "step": 12037
    },
    {
      "epoch": 3.8010735712030312,
      "grad_norm": 3.498917579650879,
      "learning_rate": 1.3321872149322856e-05,
      "loss": 0.6267,
      "step": 12038
    },
    {
      "epoch": 3.801389327439217,
      "grad_norm": 3.3154568672180176,
      "learning_rate": 1.3318363623605363e-05,
      "loss": 0.6473,
      "step": 12039
    },
    {
      "epoch": 3.8017050836754027,
      "grad_norm": 3.8972368240356445,
      "learning_rate": 1.3314855097887868e-05,
      "loss": 0.6576,
      "step": 12040
    },
    {
      "epoch": 3.802020839911588,
      "grad_norm": 2.933389663696289,
      "learning_rate": 1.3311346572170375e-05,
      "loss": 0.5846,
      "step": 12041
    },
    {
      "epoch": 3.802336596147774,
      "grad_norm": 3.6077256202697754,
      "learning_rate": 1.3307838046452881e-05,
      "loss": 0.7147,
      "step": 12042
    },
    {
      "epoch": 3.8026523523839595,
      "grad_norm": 3.6218934059143066,
      "learning_rate": 1.3304329520735387e-05,
      "loss": 0.5891,
      "step": 12043
    },
    {
      "epoch": 3.8029681086201452,
      "grad_norm": 3.20620059967041,
      "learning_rate": 1.3300820995017893e-05,
      "loss": 0.6189,
      "step": 12044
    },
    {
      "epoch": 3.803283864856331,
      "grad_norm": 3.2024989128112793,
      "learning_rate": 1.3297312469300402e-05,
      "loss": 0.6262,
      "step": 12045
    },
    {
      "epoch": 3.8035996210925167,
      "grad_norm": 3.663097858428955,
      "learning_rate": 1.3293803943582909e-05,
      "loss": 0.7232,
      "step": 12046
    },
    {
      "epoch": 3.8039153773287024,
      "grad_norm": 3.3136045932769775,
      "learning_rate": 1.3290295417865414e-05,
      "loss": 0.5733,
      "step": 12047
    },
    {
      "epoch": 3.8042311335648877,
      "grad_norm": 3.436595916748047,
      "learning_rate": 1.328678689214792e-05,
      "loss": 0.6358,
      "step": 12048
    },
    {
      "epoch": 3.8045468898010735,
      "grad_norm": 3.3977909088134766,
      "learning_rate": 1.3283278366430427e-05,
      "loss": 0.6249,
      "step": 12049
    },
    {
      "epoch": 3.804862646037259,
      "grad_norm": 5.095193386077881,
      "learning_rate": 1.3279769840712933e-05,
      "loss": 0.6379,
      "step": 12050
    },
    {
      "epoch": 3.805178402273445,
      "grad_norm": 3.3259975910186768,
      "learning_rate": 1.327626131499544e-05,
      "loss": 0.6593,
      "step": 12051
    },
    {
      "epoch": 3.8054941585096307,
      "grad_norm": 3.7332208156585693,
      "learning_rate": 1.3272752789277944e-05,
      "loss": 0.7822,
      "step": 12052
    },
    {
      "epoch": 3.805809914745816,
      "grad_norm": 3.4079596996307373,
      "learning_rate": 1.3269244263560451e-05,
      "loss": 0.5199,
      "step": 12053
    },
    {
      "epoch": 3.806125670982002,
      "grad_norm": 3.1295883655548096,
      "learning_rate": 1.326573573784296e-05,
      "loss": 0.7006,
      "step": 12054
    },
    {
      "epoch": 3.8064414272181875,
      "grad_norm": 3.3579647541046143,
      "learning_rate": 1.3262227212125467e-05,
      "loss": 0.6464,
      "step": 12055
    },
    {
      "epoch": 3.806757183454373,
      "grad_norm": 2.9558727741241455,
      "learning_rate": 1.3258718686407973e-05,
      "loss": 0.7188,
      "step": 12056
    },
    {
      "epoch": 3.807072939690559,
      "grad_norm": 3.2478997707366943,
      "learning_rate": 1.3255210160690479e-05,
      "loss": 0.6055,
      "step": 12057
    },
    {
      "epoch": 3.8073886959267447,
      "grad_norm": 4.050525188446045,
      "learning_rate": 1.3251701634972985e-05,
      "loss": 0.7467,
      "step": 12058
    },
    {
      "epoch": 3.8077044521629304,
      "grad_norm": 3.333617925643921,
      "learning_rate": 1.324819310925549e-05,
      "loss": 0.5704,
      "step": 12059
    },
    {
      "epoch": 3.8080202083991157,
      "grad_norm": 3.4918570518493652,
      "learning_rate": 1.3244684583537997e-05,
      "loss": 0.6486,
      "step": 12060
    },
    {
      "epoch": 3.8083359646353014,
      "grad_norm": 3.0984671115875244,
      "learning_rate": 1.3241176057820504e-05,
      "loss": 0.63,
      "step": 12061
    },
    {
      "epoch": 3.808651720871487,
      "grad_norm": 2.7510759830474854,
      "learning_rate": 1.323766753210301e-05,
      "loss": 0.5814,
      "step": 12062
    },
    {
      "epoch": 3.808967477107673,
      "grad_norm": 3.352442979812622,
      "learning_rate": 1.323415900638552e-05,
      "loss": 0.6831,
      "step": 12063
    },
    {
      "epoch": 3.8092832333438587,
      "grad_norm": 3.85740327835083,
      "learning_rate": 1.3230650480668025e-05,
      "loss": 0.6617,
      "step": 12064
    },
    {
      "epoch": 3.809598989580044,
      "grad_norm": 2.964197874069214,
      "learning_rate": 1.3227141954950531e-05,
      "loss": 0.602,
      "step": 12065
    },
    {
      "epoch": 3.80991474581623,
      "grad_norm": 3.903512477874756,
      "learning_rate": 1.3223633429233036e-05,
      "loss": 0.5769,
      "step": 12066
    },
    {
      "epoch": 3.8102305020524154,
      "grad_norm": 3.0309486389160156,
      "learning_rate": 1.3220124903515543e-05,
      "loss": 0.6271,
      "step": 12067
    },
    {
      "epoch": 3.810546258288601,
      "grad_norm": 4.006217002868652,
      "learning_rate": 1.321661637779805e-05,
      "loss": 0.7223,
      "step": 12068
    },
    {
      "epoch": 3.810862014524787,
      "grad_norm": 3.957686185836792,
      "learning_rate": 1.3213107852080555e-05,
      "loss": 0.6914,
      "step": 12069
    },
    {
      "epoch": 3.8111777707609726,
      "grad_norm": 3.3031508922576904,
      "learning_rate": 1.3209599326363062e-05,
      "loss": 0.6328,
      "step": 12070
    },
    {
      "epoch": 3.8114935269971584,
      "grad_norm": 3.2352757453918457,
      "learning_rate": 1.3206090800645569e-05,
      "loss": 0.6655,
      "step": 12071
    },
    {
      "epoch": 3.8118092832333437,
      "grad_norm": 4.722445964813232,
      "learning_rate": 1.3202582274928077e-05,
      "loss": 0.8569,
      "step": 12072
    },
    {
      "epoch": 3.8121250394695294,
      "grad_norm": 4.388420581817627,
      "learning_rate": 1.3199073749210582e-05,
      "loss": 0.7241,
      "step": 12073
    },
    {
      "epoch": 3.812440795705715,
      "grad_norm": 3.6641674041748047,
      "learning_rate": 1.319556522349309e-05,
      "loss": 0.6088,
      "step": 12074
    },
    {
      "epoch": 3.812756551941901,
      "grad_norm": 3.858689308166504,
      "learning_rate": 1.3192056697775596e-05,
      "loss": 0.7002,
      "step": 12075
    },
    {
      "epoch": 3.8130723081780866,
      "grad_norm": 4.211068153381348,
      "learning_rate": 1.3188548172058101e-05,
      "loss": 0.6526,
      "step": 12076
    },
    {
      "epoch": 3.813388064414272,
      "grad_norm": 3.5107526779174805,
      "learning_rate": 1.3185039646340608e-05,
      "loss": 0.649,
      "step": 12077
    },
    {
      "epoch": 3.813703820650458,
      "grad_norm": 2.819239616394043,
      "learning_rate": 1.3181531120623115e-05,
      "loss": 0.579,
      "step": 12078
    },
    {
      "epoch": 3.8140195768866434,
      "grad_norm": 3.758366584777832,
      "learning_rate": 1.317802259490562e-05,
      "loss": 0.7042,
      "step": 12079
    },
    {
      "epoch": 3.814335333122829,
      "grad_norm": 3.893425941467285,
      "learning_rate": 1.3174514069188128e-05,
      "loss": 0.7233,
      "step": 12080
    },
    {
      "epoch": 3.814651089359015,
      "grad_norm": 4.5694193840026855,
      "learning_rate": 1.3171005543470635e-05,
      "loss": 0.6495,
      "step": 12081
    },
    {
      "epoch": 3.8149668455952006,
      "grad_norm": 3.782301187515259,
      "learning_rate": 1.3167497017753142e-05,
      "loss": 0.6625,
      "step": 12082
    },
    {
      "epoch": 3.8152826018313863,
      "grad_norm": 3.642054557800293,
      "learning_rate": 1.3163988492035647e-05,
      "loss": 0.6261,
      "step": 12083
    },
    {
      "epoch": 3.8155983580675716,
      "grad_norm": 3.2396135330200195,
      "learning_rate": 1.3160479966318154e-05,
      "loss": 0.6738,
      "step": 12084
    },
    {
      "epoch": 3.8159141143037574,
      "grad_norm": 3.563232660293579,
      "learning_rate": 1.315697144060066e-05,
      "loss": 0.6133,
      "step": 12085
    },
    {
      "epoch": 3.816229870539943,
      "grad_norm": 3.0170795917510986,
      "learning_rate": 1.3153462914883166e-05,
      "loss": 0.5839,
      "step": 12086
    },
    {
      "epoch": 3.816545626776129,
      "grad_norm": 3.2516748905181885,
      "learning_rate": 1.3149954389165673e-05,
      "loss": 0.7174,
      "step": 12087
    },
    {
      "epoch": 3.8168613830123146,
      "grad_norm": 3.7435810565948486,
      "learning_rate": 1.3146445863448178e-05,
      "loss": 0.6961,
      "step": 12088
    },
    {
      "epoch": 3.8171771392485,
      "grad_norm": 3.0025928020477295,
      "learning_rate": 1.3142937337730688e-05,
      "loss": 0.6067,
      "step": 12089
    },
    {
      "epoch": 3.817492895484686,
      "grad_norm": 4.308679580688477,
      "learning_rate": 1.3139428812013193e-05,
      "loss": 0.6585,
      "step": 12090
    },
    {
      "epoch": 3.8178086517208714,
      "grad_norm": 3.8546059131622314,
      "learning_rate": 1.31359202862957e-05,
      "loss": 0.6857,
      "step": 12091
    },
    {
      "epoch": 3.818124407957057,
      "grad_norm": 3.9507510662078857,
      "learning_rate": 1.3132411760578205e-05,
      "loss": 0.7643,
      "step": 12092
    },
    {
      "epoch": 3.818440164193243,
      "grad_norm": 3.3365769386291504,
      "learning_rate": 1.3128903234860712e-05,
      "loss": 0.5804,
      "step": 12093
    },
    {
      "epoch": 3.8187559204294286,
      "grad_norm": 3.2132253646850586,
      "learning_rate": 1.3125394709143219e-05,
      "loss": 0.629,
      "step": 12094
    },
    {
      "epoch": 3.8190716766656143,
      "grad_norm": 3.6941986083984375,
      "learning_rate": 1.3121886183425724e-05,
      "loss": 0.6403,
      "step": 12095
    },
    {
      "epoch": 3.8193874329017996,
      "grad_norm": 2.947622776031494,
      "learning_rate": 1.311837765770823e-05,
      "loss": 0.6381,
      "step": 12096
    },
    {
      "epoch": 3.8197031891379853,
      "grad_norm": 3.1621007919311523,
      "learning_rate": 1.3114869131990738e-05,
      "loss": 0.5484,
      "step": 12097
    },
    {
      "epoch": 3.820018945374171,
      "grad_norm": 3.539337396621704,
      "learning_rate": 1.3111360606273246e-05,
      "loss": 0.6474,
      "step": 12098
    },
    {
      "epoch": 3.820334701610357,
      "grad_norm": 3.7524197101593018,
      "learning_rate": 1.3107852080555751e-05,
      "loss": 0.7335,
      "step": 12099
    },
    {
      "epoch": 3.8206504578465426,
      "grad_norm": 3.6979992389678955,
      "learning_rate": 1.3104343554838258e-05,
      "loss": 0.6858,
      "step": 12100
    },
    {
      "epoch": 3.8209662140827283,
      "grad_norm": 3.407636880874634,
      "learning_rate": 1.3100835029120765e-05,
      "loss": 0.712,
      "step": 12101
    },
    {
      "epoch": 3.821281970318914,
      "grad_norm": 3.7942001819610596,
      "learning_rate": 1.309732650340327e-05,
      "loss": 0.6396,
      "step": 12102
    },
    {
      "epoch": 3.8215977265550993,
      "grad_norm": 3.2794349193573,
      "learning_rate": 1.3093817977685777e-05,
      "loss": 0.5441,
      "step": 12103
    },
    {
      "epoch": 3.821913482791285,
      "grad_norm": 3.5022103786468506,
      "learning_rate": 1.3090309451968284e-05,
      "loss": 0.5545,
      "step": 12104
    },
    {
      "epoch": 3.822229239027471,
      "grad_norm": 3.1023192405700684,
      "learning_rate": 1.3086800926250789e-05,
      "loss": 0.6516,
      "step": 12105
    },
    {
      "epoch": 3.8225449952636565,
      "grad_norm": 3.834571361541748,
      "learning_rate": 1.3083292400533295e-05,
      "loss": 0.6644,
      "step": 12106
    },
    {
      "epoch": 3.8228607514998423,
      "grad_norm": 3.0783896446228027,
      "learning_rate": 1.3079783874815804e-05,
      "loss": 0.5846,
      "step": 12107
    },
    {
      "epoch": 3.8231765077360276,
      "grad_norm": 3.62800669670105,
      "learning_rate": 1.307627534909831e-05,
      "loss": 0.7004,
      "step": 12108
    },
    {
      "epoch": 3.8234922639722133,
      "grad_norm": 3.478991746902466,
      "learning_rate": 1.3072766823380816e-05,
      "loss": 0.7326,
      "step": 12109
    },
    {
      "epoch": 3.823808020208399,
      "grad_norm": 2.971705198287964,
      "learning_rate": 1.3069258297663323e-05,
      "loss": 0.5955,
      "step": 12110
    },
    {
      "epoch": 3.824123776444585,
      "grad_norm": 3.1475019454956055,
      "learning_rate": 1.306574977194583e-05,
      "loss": 0.6004,
      "step": 12111
    },
    {
      "epoch": 3.8244395326807705,
      "grad_norm": 3.4347140789031982,
      "learning_rate": 1.3062241246228335e-05,
      "loss": 0.6739,
      "step": 12112
    },
    {
      "epoch": 3.8247552889169563,
      "grad_norm": 3.3277945518493652,
      "learning_rate": 1.3058732720510841e-05,
      "loss": 0.6984,
      "step": 12113
    },
    {
      "epoch": 3.825071045153142,
      "grad_norm": 3.554630994796753,
      "learning_rate": 1.3055224194793347e-05,
      "loss": 0.5351,
      "step": 12114
    },
    {
      "epoch": 3.8253868013893273,
      "grad_norm": 3.307309627532959,
      "learning_rate": 1.3051715669075857e-05,
      "loss": 0.616,
      "step": 12115
    },
    {
      "epoch": 3.825702557625513,
      "grad_norm": 3.4910848140716553,
      "learning_rate": 1.3048207143358362e-05,
      "loss": 0.6564,
      "step": 12116
    },
    {
      "epoch": 3.8260183138616988,
      "grad_norm": 3.064741849899292,
      "learning_rate": 1.3044698617640869e-05,
      "loss": 0.605,
      "step": 12117
    },
    {
      "epoch": 3.8263340700978845,
      "grad_norm": 3.433183431625366,
      "learning_rate": 1.3041190091923376e-05,
      "loss": 0.6912,
      "step": 12118
    },
    {
      "epoch": 3.8266498263340702,
      "grad_norm": 3.7884292602539062,
      "learning_rate": 1.303768156620588e-05,
      "loss": 0.6611,
      "step": 12119
    },
    {
      "epoch": 3.8269655825702555,
      "grad_norm": 3.4173474311828613,
      "learning_rate": 1.3034173040488387e-05,
      "loss": 0.6638,
      "step": 12120
    },
    {
      "epoch": 3.8272813388064413,
      "grad_norm": 3.248460531234741,
      "learning_rate": 1.3030664514770893e-05,
      "loss": 0.698,
      "step": 12121
    },
    {
      "epoch": 3.827597095042627,
      "grad_norm": 3.1754839420318604,
      "learning_rate": 1.30271559890534e-05,
      "loss": 0.744,
      "step": 12122
    },
    {
      "epoch": 3.8279128512788128,
      "grad_norm": 3.348600387573242,
      "learning_rate": 1.3023647463335906e-05,
      "loss": 0.65,
      "step": 12123
    },
    {
      "epoch": 3.8282286075149985,
      "grad_norm": 3.61194109916687,
      "learning_rate": 1.3020138937618415e-05,
      "loss": 0.6211,
      "step": 12124
    },
    {
      "epoch": 3.8285443637511842,
      "grad_norm": 3.514749765396118,
      "learning_rate": 1.3016630411900922e-05,
      "loss": 0.665,
      "step": 12125
    },
    {
      "epoch": 3.82886011998737,
      "grad_norm": 3.1799662113189697,
      "learning_rate": 1.3013121886183427e-05,
      "loss": 0.5982,
      "step": 12126
    },
    {
      "epoch": 3.8291758762235553,
      "grad_norm": 3.5440170764923096,
      "learning_rate": 1.3009613360465933e-05,
      "loss": 0.6987,
      "step": 12127
    },
    {
      "epoch": 3.829491632459741,
      "grad_norm": 3.5473952293395996,
      "learning_rate": 1.3006104834748439e-05,
      "loss": 0.593,
      "step": 12128
    },
    {
      "epoch": 3.8298073886959267,
      "grad_norm": 2.952334403991699,
      "learning_rate": 1.3002596309030945e-05,
      "loss": 0.5987,
      "step": 12129
    },
    {
      "epoch": 3.8301231449321125,
      "grad_norm": 3.879283905029297,
      "learning_rate": 1.2999087783313452e-05,
      "loss": 0.6676,
      "step": 12130
    },
    {
      "epoch": 3.830438901168298,
      "grad_norm": 3.034092426300049,
      "learning_rate": 1.2995579257595957e-05,
      "loss": 0.5744,
      "step": 12131
    },
    {
      "epoch": 3.8307546574044835,
      "grad_norm": 3.6778626441955566,
      "learning_rate": 1.2992070731878464e-05,
      "loss": 0.6432,
      "step": 12132
    },
    {
      "epoch": 3.8310704136406692,
      "grad_norm": 2.9692625999450684,
      "learning_rate": 1.2988562206160973e-05,
      "loss": 0.5507,
      "step": 12133
    },
    {
      "epoch": 3.831386169876855,
      "grad_norm": 3.2437901496887207,
      "learning_rate": 1.298505368044348e-05,
      "loss": 0.6341,
      "step": 12134
    },
    {
      "epoch": 3.8317019261130407,
      "grad_norm": 3.182020425796509,
      "learning_rate": 1.2981545154725985e-05,
      "loss": 0.644,
      "step": 12135
    },
    {
      "epoch": 3.8320176823492265,
      "grad_norm": 3.635453224182129,
      "learning_rate": 1.2978036629008491e-05,
      "loss": 0.7351,
      "step": 12136
    },
    {
      "epoch": 3.832333438585412,
      "grad_norm": 3.9717307090759277,
      "learning_rate": 1.2974528103290998e-05,
      "loss": 0.5828,
      "step": 12137
    },
    {
      "epoch": 3.832649194821598,
      "grad_norm": 3.4143662452697754,
      "learning_rate": 1.2971019577573503e-05,
      "loss": 0.6616,
      "step": 12138
    },
    {
      "epoch": 3.8329649510577832,
      "grad_norm": 3.673901319503784,
      "learning_rate": 1.296751105185601e-05,
      "loss": 0.6865,
      "step": 12139
    },
    {
      "epoch": 3.833280707293969,
      "grad_norm": 3.6799745559692383,
      "learning_rate": 1.2964002526138517e-05,
      "loss": 0.6012,
      "step": 12140
    },
    {
      "epoch": 3.8335964635301547,
      "grad_norm": 3.4271721839904785,
      "learning_rate": 1.2960494000421022e-05,
      "loss": 0.7269,
      "step": 12141
    },
    {
      "epoch": 3.8339122197663404,
      "grad_norm": 3.022494077682495,
      "learning_rate": 1.295698547470353e-05,
      "loss": 0.5424,
      "step": 12142
    },
    {
      "epoch": 3.834227976002526,
      "grad_norm": 3.0106277465820312,
      "learning_rate": 1.2953476948986037e-05,
      "loss": 0.7242,
      "step": 12143
    },
    {
      "epoch": 3.8345437322387115,
      "grad_norm": 3.403193235397339,
      "learning_rate": 1.2949968423268544e-05,
      "loss": 0.5658,
      "step": 12144
    },
    {
      "epoch": 3.834859488474897,
      "grad_norm": 3.678013563156128,
      "learning_rate": 1.294645989755105e-05,
      "loss": 0.7133,
      "step": 12145
    },
    {
      "epoch": 3.835175244711083,
      "grad_norm": 3.4676756858825684,
      "learning_rate": 1.2942951371833556e-05,
      "loss": 0.7232,
      "step": 12146
    },
    {
      "epoch": 3.8354910009472687,
      "grad_norm": 3.4319517612457275,
      "learning_rate": 1.2939442846116063e-05,
      "loss": 0.6491,
      "step": 12147
    },
    {
      "epoch": 3.8358067571834544,
      "grad_norm": 3.6992955207824707,
      "learning_rate": 1.2935934320398568e-05,
      "loss": 0.6982,
      "step": 12148
    },
    {
      "epoch": 3.83612251341964,
      "grad_norm": 3.2809057235717773,
      "learning_rate": 1.2932425794681075e-05,
      "loss": 0.5905,
      "step": 12149
    },
    {
      "epoch": 3.836438269655826,
      "grad_norm": 3.3505945205688477,
      "learning_rate": 1.2928917268963583e-05,
      "loss": 0.7073,
      "step": 12150
    },
    {
      "epoch": 3.836754025892011,
      "grad_norm": 4.079291820526123,
      "learning_rate": 1.292540874324609e-05,
      "loss": 0.6369,
      "step": 12151
    },
    {
      "epoch": 3.837069782128197,
      "grad_norm": 3.084132194519043,
      "learning_rate": 1.2921900217528595e-05,
      "loss": 0.6245,
      "step": 12152
    },
    {
      "epoch": 3.8373855383643827,
      "grad_norm": 3.558659315109253,
      "learning_rate": 1.2918391691811102e-05,
      "loss": 0.638,
      "step": 12153
    },
    {
      "epoch": 3.8377012946005684,
      "grad_norm": 3.846536874771118,
      "learning_rate": 1.2914883166093609e-05,
      "loss": 0.7008,
      "step": 12154
    },
    {
      "epoch": 3.838017050836754,
      "grad_norm": 3.6474647521972656,
      "learning_rate": 1.2911374640376114e-05,
      "loss": 0.6406,
      "step": 12155
    },
    {
      "epoch": 3.8383328070729394,
      "grad_norm": 3.509274482727051,
      "learning_rate": 1.2907866114658621e-05,
      "loss": 0.6986,
      "step": 12156
    },
    {
      "epoch": 3.8386485633091256,
      "grad_norm": 3.8898653984069824,
      "learning_rate": 1.2904357588941126e-05,
      "loss": 0.6424,
      "step": 12157
    },
    {
      "epoch": 3.838964319545311,
      "grad_norm": 2.873096466064453,
      "learning_rate": 1.2900849063223633e-05,
      "loss": 0.5616,
      "step": 12158
    },
    {
      "epoch": 3.8392800757814967,
      "grad_norm": 3.0742440223693848,
      "learning_rate": 1.2897340537506141e-05,
      "loss": 0.6348,
      "step": 12159
    },
    {
      "epoch": 3.8395958320176824,
      "grad_norm": 3.1837716102600098,
      "learning_rate": 1.2893832011788648e-05,
      "loss": 0.6803,
      "step": 12160
    },
    {
      "epoch": 3.839911588253868,
      "grad_norm": 3.3016250133514404,
      "learning_rate": 1.2890323486071155e-05,
      "loss": 0.5983,
      "step": 12161
    },
    {
      "epoch": 3.840227344490054,
      "grad_norm": 3.3553884029388428,
      "learning_rate": 1.288681496035366e-05,
      "loss": 0.6703,
      "step": 12162
    },
    {
      "epoch": 3.840543100726239,
      "grad_norm": 3.7665090560913086,
      "learning_rate": 1.2883306434636167e-05,
      "loss": 0.6337,
      "step": 12163
    },
    {
      "epoch": 3.840858856962425,
      "grad_norm": 3.143357753753662,
      "learning_rate": 1.2879797908918672e-05,
      "loss": 0.5738,
      "step": 12164
    },
    {
      "epoch": 3.8411746131986106,
      "grad_norm": 3.4302377700805664,
      "learning_rate": 1.2876289383201179e-05,
      "loss": 0.5587,
      "step": 12165
    },
    {
      "epoch": 3.8414903694347964,
      "grad_norm": 3.2635176181793213,
      "learning_rate": 1.2872780857483686e-05,
      "loss": 0.6138,
      "step": 12166
    },
    {
      "epoch": 3.841806125670982,
      "grad_norm": 2.967313051223755,
      "learning_rate": 1.286927233176619e-05,
      "loss": 0.7073,
      "step": 12167
    },
    {
      "epoch": 3.8421218819071674,
      "grad_norm": 3.9767210483551025,
      "learning_rate": 1.2865763806048701e-05,
      "loss": 0.6912,
      "step": 12168
    },
    {
      "epoch": 3.8424376381433536,
      "grad_norm": 3.137108087539673,
      "learning_rate": 1.2862255280331206e-05,
      "loss": 0.6017,
      "step": 12169
    },
    {
      "epoch": 3.842753394379539,
      "grad_norm": 3.2752749919891357,
      "learning_rate": 1.2858746754613713e-05,
      "loss": 0.6499,
      "step": 12170
    },
    {
      "epoch": 3.8430691506157246,
      "grad_norm": 3.2832398414611816,
      "learning_rate": 1.2855238228896218e-05,
      "loss": 0.7106,
      "step": 12171
    },
    {
      "epoch": 3.8433849068519104,
      "grad_norm": 3.09153413772583,
      "learning_rate": 1.2851729703178725e-05,
      "loss": 0.5905,
      "step": 12172
    },
    {
      "epoch": 3.843700663088096,
      "grad_norm": 3.8320705890655518,
      "learning_rate": 1.2848221177461232e-05,
      "loss": 0.6748,
      "step": 12173
    },
    {
      "epoch": 3.844016419324282,
      "grad_norm": 3.3023273944854736,
      "learning_rate": 1.2844712651743737e-05,
      "loss": 0.6228,
      "step": 12174
    },
    {
      "epoch": 3.844332175560467,
      "grad_norm": 4.100564479827881,
      "learning_rate": 1.2841204126026244e-05,
      "loss": 0.6333,
      "step": 12175
    },
    {
      "epoch": 3.844647931796653,
      "grad_norm": 3.2683424949645996,
      "learning_rate": 1.283769560030875e-05,
      "loss": 0.5353,
      "step": 12176
    },
    {
      "epoch": 3.8449636880328386,
      "grad_norm": 3.111232280731201,
      "learning_rate": 1.2834187074591259e-05,
      "loss": 0.6816,
      "step": 12177
    },
    {
      "epoch": 3.8452794442690243,
      "grad_norm": 3.8555335998535156,
      "learning_rate": 1.2830678548873764e-05,
      "loss": 0.6894,
      "step": 12178
    },
    {
      "epoch": 3.84559520050521,
      "grad_norm": 3.4811246395111084,
      "learning_rate": 1.282717002315627e-05,
      "loss": 0.7629,
      "step": 12179
    },
    {
      "epoch": 3.8459109567413954,
      "grad_norm": 3.466578960418701,
      "learning_rate": 1.2823661497438778e-05,
      "loss": 0.6489,
      "step": 12180
    },
    {
      "epoch": 3.8462267129775816,
      "grad_norm": 4.095888137817383,
      "learning_rate": 1.2820152971721283e-05,
      "loss": 0.7243,
      "step": 12181
    },
    {
      "epoch": 3.846542469213767,
      "grad_norm": 3.630539655685425,
      "learning_rate": 1.281664444600379e-05,
      "loss": 0.6612,
      "step": 12182
    },
    {
      "epoch": 3.8468582254499526,
      "grad_norm": 3.056018829345703,
      "learning_rate": 1.2813135920286296e-05,
      "loss": 0.6155,
      "step": 12183
    },
    {
      "epoch": 3.8471739816861383,
      "grad_norm": 3.4249963760375977,
      "learning_rate": 1.2809627394568801e-05,
      "loss": 0.6825,
      "step": 12184
    },
    {
      "epoch": 3.847489737922324,
      "grad_norm": 3.2710959911346436,
      "learning_rate": 1.280611886885131e-05,
      "loss": 0.6061,
      "step": 12185
    },
    {
      "epoch": 3.84780549415851,
      "grad_norm": 3.6119799613952637,
      "learning_rate": 1.2802610343133817e-05,
      "loss": 0.6118,
      "step": 12186
    },
    {
      "epoch": 3.848121250394695,
      "grad_norm": 4.517830848693848,
      "learning_rate": 1.2799101817416324e-05,
      "loss": 0.7153,
      "step": 12187
    },
    {
      "epoch": 3.848437006630881,
      "grad_norm": 3.3101983070373535,
      "learning_rate": 1.2795593291698829e-05,
      "loss": 0.6509,
      "step": 12188
    },
    {
      "epoch": 3.8487527628670666,
      "grad_norm": 3.8462653160095215,
      "learning_rate": 1.2792084765981336e-05,
      "loss": 0.572,
      "step": 12189
    },
    {
      "epoch": 3.8490685191032523,
      "grad_norm": 3.032336473464966,
      "learning_rate": 1.2788576240263842e-05,
      "loss": 0.5894,
      "step": 12190
    },
    {
      "epoch": 3.849384275339438,
      "grad_norm": 3.2366538047790527,
      "learning_rate": 1.2785067714546347e-05,
      "loss": 0.5676,
      "step": 12191
    },
    {
      "epoch": 3.8497000315756233,
      "grad_norm": 3.8913984298706055,
      "learning_rate": 1.2781559188828854e-05,
      "loss": 0.664,
      "step": 12192
    },
    {
      "epoch": 3.8500157878118095,
      "grad_norm": 3.4992971420288086,
      "learning_rate": 1.277805066311136e-05,
      "loss": 0.6165,
      "step": 12193
    },
    {
      "epoch": 3.850331544047995,
      "grad_norm": 2.7994468212127686,
      "learning_rate": 1.277454213739387e-05,
      "loss": 0.6509,
      "step": 12194
    },
    {
      "epoch": 3.8506473002841806,
      "grad_norm": 3.6505367755889893,
      "learning_rate": 1.2771033611676375e-05,
      "loss": 0.6327,
      "step": 12195
    },
    {
      "epoch": 3.8509630565203663,
      "grad_norm": 3.2449910640716553,
      "learning_rate": 1.2767525085958882e-05,
      "loss": 0.6109,
      "step": 12196
    },
    {
      "epoch": 3.851278812756552,
      "grad_norm": 3.0583386421203613,
      "learning_rate": 1.2764016560241387e-05,
      "loss": 0.6573,
      "step": 12197
    },
    {
      "epoch": 3.8515945689927378,
      "grad_norm": 3.1844663619995117,
      "learning_rate": 1.2760508034523893e-05,
      "loss": 0.7019,
      "step": 12198
    },
    {
      "epoch": 3.851910325228923,
      "grad_norm": 3.5051791667938232,
      "learning_rate": 1.27569995088064e-05,
      "loss": 0.596,
      "step": 12199
    },
    {
      "epoch": 3.852226081465109,
      "grad_norm": 3.423168659210205,
      "learning_rate": 1.2753490983088905e-05,
      "loss": 0.7317,
      "step": 12200
    },
    {
      "epoch": 3.852226081465109,
      "eval_loss": 0.8044021725654602,
      "eval_runtime": 4.7644,
      "eval_samples_per_second": 522.627,
      "eval_steps_per_second": 4.408,
      "step": 12200
    },
    {
      "epoch": 3.8525418377012945,
      "grad_norm": 3.4506611824035645,
      "learning_rate": 1.2749982457371412e-05,
      "loss": 0.6738,
      "step": 12201
    },
    {
      "epoch": 3.8528575939374803,
      "grad_norm": 4.1222333908081055,
      "learning_rate": 1.2746473931653919e-05,
      "loss": 0.7,
      "step": 12202
    },
    {
      "epoch": 3.853173350173666,
      "grad_norm": 3.4180805683135986,
      "learning_rate": 1.2742965405936428e-05,
      "loss": 0.7517,
      "step": 12203
    },
    {
      "epoch": 3.8534891064098518,
      "grad_norm": 3.0248396396636963,
      "learning_rate": 1.2739456880218933e-05,
      "loss": 0.6063,
      "step": 12204
    },
    {
      "epoch": 3.8538048626460375,
      "grad_norm": 2.8446848392486572,
      "learning_rate": 1.273594835450144e-05,
      "loss": 0.6428,
      "step": 12205
    },
    {
      "epoch": 3.854120618882223,
      "grad_norm": 2.9548087120056152,
      "learning_rate": 1.2732439828783946e-05,
      "loss": 0.7454,
      "step": 12206
    },
    {
      "epoch": 3.8544363751184085,
      "grad_norm": 4.393414497375488,
      "learning_rate": 1.2728931303066451e-05,
      "loss": 0.6502,
      "step": 12207
    },
    {
      "epoch": 3.8547521313545943,
      "grad_norm": 3.1456446647644043,
      "learning_rate": 1.2725422777348958e-05,
      "loss": 0.6625,
      "step": 12208
    },
    {
      "epoch": 3.85506788759078,
      "grad_norm": 3.866157054901123,
      "learning_rate": 1.2721914251631465e-05,
      "loss": 0.5641,
      "step": 12209
    },
    {
      "epoch": 3.8553836438269657,
      "grad_norm": 3.2817978858947754,
      "learning_rate": 1.271840572591397e-05,
      "loss": 0.6901,
      "step": 12210
    },
    {
      "epoch": 3.855699400063151,
      "grad_norm": 3.294797897338867,
      "learning_rate": 1.2714897200196477e-05,
      "loss": 0.6208,
      "step": 12211
    },
    {
      "epoch": 3.8560151562993368,
      "grad_norm": 3.364048480987549,
      "learning_rate": 1.2711388674478985e-05,
      "loss": 0.6459,
      "step": 12212
    },
    {
      "epoch": 3.8563309125355225,
      "grad_norm": 3.114083766937256,
      "learning_rate": 1.2707880148761492e-05,
      "loss": 0.7044,
      "step": 12213
    },
    {
      "epoch": 3.8566466687717083,
      "grad_norm": 3.3668978214263916,
      "learning_rate": 1.2704371623043997e-05,
      "loss": 0.7081,
      "step": 12214
    },
    {
      "epoch": 3.856962425007894,
      "grad_norm": 3.740549087524414,
      "learning_rate": 1.2700863097326504e-05,
      "loss": 0.6275,
      "step": 12215
    },
    {
      "epoch": 3.8572781812440797,
      "grad_norm": 3.903395891189575,
      "learning_rate": 1.2697354571609011e-05,
      "loss": 0.6661,
      "step": 12216
    },
    {
      "epoch": 3.8575939374802655,
      "grad_norm": 3.183821201324463,
      "learning_rate": 1.2693846045891516e-05,
      "loss": 0.641,
      "step": 12217
    },
    {
      "epoch": 3.8579096937164508,
      "grad_norm": 3.709459066390991,
      "learning_rate": 1.2690337520174023e-05,
      "loss": 0.7121,
      "step": 12218
    },
    {
      "epoch": 3.8582254499526365,
      "grad_norm": 3.4442057609558105,
      "learning_rate": 1.2686828994456528e-05,
      "loss": 0.6567,
      "step": 12219
    },
    {
      "epoch": 3.8585412061888222,
      "grad_norm": 3.1393964290618896,
      "learning_rate": 1.2683320468739038e-05,
      "loss": 0.6119,
      "step": 12220
    },
    {
      "epoch": 3.858856962425008,
      "grad_norm": 3.3024003505706787,
      "learning_rate": 1.2679811943021543e-05,
      "loss": 0.6319,
      "step": 12221
    },
    {
      "epoch": 3.8591727186611937,
      "grad_norm": 3.5723352432250977,
      "learning_rate": 1.267630341730405e-05,
      "loss": 0.6709,
      "step": 12222
    },
    {
      "epoch": 3.859488474897379,
      "grad_norm": 3.228372573852539,
      "learning_rate": 1.2672794891586557e-05,
      "loss": 0.6452,
      "step": 12223
    },
    {
      "epoch": 3.8598042311335647,
      "grad_norm": 2.8515563011169434,
      "learning_rate": 1.2669286365869062e-05,
      "loss": 0.6073,
      "step": 12224
    },
    {
      "epoch": 3.8601199873697505,
      "grad_norm": 3.4136064052581787,
      "learning_rate": 1.2665777840151569e-05,
      "loss": 0.6393,
      "step": 12225
    },
    {
      "epoch": 3.860435743605936,
      "grad_norm": 3.0850319862365723,
      "learning_rate": 1.2662269314434074e-05,
      "loss": 0.6174,
      "step": 12226
    },
    {
      "epoch": 3.860751499842122,
      "grad_norm": 3.0015556812286377,
      "learning_rate": 1.2658760788716581e-05,
      "loss": 0.7257,
      "step": 12227
    },
    {
      "epoch": 3.8610672560783077,
      "grad_norm": 3.795605182647705,
      "learning_rate": 1.2655252262999088e-05,
      "loss": 0.654,
      "step": 12228
    },
    {
      "epoch": 3.8613830123144934,
      "grad_norm": 3.6657536029815674,
      "learning_rate": 1.2651743737281596e-05,
      "loss": 0.6885,
      "step": 12229
    },
    {
      "epoch": 3.8616987685506787,
      "grad_norm": 3.7848587036132812,
      "learning_rate": 1.2648235211564103e-05,
      "loss": 0.683,
      "step": 12230
    },
    {
      "epoch": 3.8620145247868645,
      "grad_norm": 2.6214733123779297,
      "learning_rate": 1.2644726685846608e-05,
      "loss": 0.575,
      "step": 12231
    },
    {
      "epoch": 3.86233028102305,
      "grad_norm": 2.7973341941833496,
      "learning_rate": 1.2641218160129115e-05,
      "loss": 0.6389,
      "step": 12232
    },
    {
      "epoch": 3.862646037259236,
      "grad_norm": 3.4022741317749023,
      "learning_rate": 1.263770963441162e-05,
      "loss": 0.5967,
      "step": 12233
    },
    {
      "epoch": 3.8629617934954217,
      "grad_norm": 3.265174388885498,
      "learning_rate": 1.2634201108694127e-05,
      "loss": 0.5994,
      "step": 12234
    },
    {
      "epoch": 3.863277549731607,
      "grad_norm": 3.5299012660980225,
      "learning_rate": 1.2630692582976634e-05,
      "loss": 0.7212,
      "step": 12235
    },
    {
      "epoch": 3.8635933059677927,
      "grad_norm": 3.405001640319824,
      "learning_rate": 1.2627184057259139e-05,
      "loss": 0.8147,
      "step": 12236
    },
    {
      "epoch": 3.8639090622039785,
      "grad_norm": 3.644469976425171,
      "learning_rate": 1.2623675531541646e-05,
      "loss": 0.6457,
      "step": 12237
    },
    {
      "epoch": 3.864224818440164,
      "grad_norm": 3.099736452102661,
      "learning_rate": 1.2620167005824154e-05,
      "loss": 0.6544,
      "step": 12238
    },
    {
      "epoch": 3.86454057467635,
      "grad_norm": 3.6141092777252197,
      "learning_rate": 1.2616658480106661e-05,
      "loss": 0.8045,
      "step": 12239
    },
    {
      "epoch": 3.8648563309125357,
      "grad_norm": 3.2897567749023438,
      "learning_rate": 1.2613149954389166e-05,
      "loss": 0.5642,
      "step": 12240
    },
    {
      "epoch": 3.8651720871487214,
      "grad_norm": 3.3450260162353516,
      "learning_rate": 1.2609641428671673e-05,
      "loss": 0.5989,
      "step": 12241
    },
    {
      "epoch": 3.8654878433849067,
      "grad_norm": 3.1129961013793945,
      "learning_rate": 1.260613290295418e-05,
      "loss": 0.6634,
      "step": 12242
    },
    {
      "epoch": 3.8658035996210924,
      "grad_norm": 3.163390636444092,
      "learning_rate": 1.2602624377236685e-05,
      "loss": 0.6849,
      "step": 12243
    },
    {
      "epoch": 3.866119355857278,
      "grad_norm": 3.3265318870544434,
      "learning_rate": 1.2599115851519192e-05,
      "loss": 0.7307,
      "step": 12244
    },
    {
      "epoch": 3.866435112093464,
      "grad_norm": 3.4765124320983887,
      "learning_rate": 1.2595607325801698e-05,
      "loss": 0.6437,
      "step": 12245
    },
    {
      "epoch": 3.8667508683296496,
      "grad_norm": 3.0626890659332275,
      "learning_rate": 1.2592098800084204e-05,
      "loss": 0.5926,
      "step": 12246
    },
    {
      "epoch": 3.867066624565835,
      "grad_norm": 3.4165360927581787,
      "learning_rate": 1.2588590274366712e-05,
      "loss": 0.6657,
      "step": 12247
    },
    {
      "epoch": 3.8673823808020207,
      "grad_norm": 3.162835121154785,
      "learning_rate": 1.2585081748649219e-05,
      "loss": 0.591,
      "step": 12248
    },
    {
      "epoch": 3.8676981370382064,
      "grad_norm": 3.4445579051971436,
      "learning_rate": 1.2581573222931726e-05,
      "loss": 0.7381,
      "step": 12249
    },
    {
      "epoch": 3.868013893274392,
      "grad_norm": 3.6065146923065186,
      "learning_rate": 1.257806469721423e-05,
      "loss": 0.6712,
      "step": 12250
    },
    {
      "epoch": 3.868329649510578,
      "grad_norm": 3.083275318145752,
      "learning_rate": 1.2574556171496738e-05,
      "loss": 0.6355,
      "step": 12251
    },
    {
      "epoch": 3.8686454057467636,
      "grad_norm": 3.718498945236206,
      "learning_rate": 1.2571047645779244e-05,
      "loss": 0.651,
      "step": 12252
    },
    {
      "epoch": 3.8689611619829494,
      "grad_norm": 3.3122799396514893,
      "learning_rate": 1.256753912006175e-05,
      "loss": 0.6675,
      "step": 12253
    },
    {
      "epoch": 3.8692769182191347,
      "grad_norm": 3.4888792037963867,
      "learning_rate": 1.2564030594344256e-05,
      "loss": 0.5988,
      "step": 12254
    },
    {
      "epoch": 3.8695926744553204,
      "grad_norm": 3.429995536804199,
      "learning_rate": 1.2560522068626765e-05,
      "loss": 0.6019,
      "step": 12255
    },
    {
      "epoch": 3.869908430691506,
      "grad_norm": 3.374730348587036,
      "learning_rate": 1.2557013542909272e-05,
      "loss": 0.626,
      "step": 12256
    },
    {
      "epoch": 3.870224186927692,
      "grad_norm": 2.797635555267334,
      "learning_rate": 1.2553505017191777e-05,
      "loss": 0.593,
      "step": 12257
    },
    {
      "epoch": 3.8705399431638776,
      "grad_norm": 3.3603765964508057,
      "learning_rate": 1.2549996491474284e-05,
      "loss": 0.5633,
      "step": 12258
    },
    {
      "epoch": 3.870855699400063,
      "grad_norm": 3.269404411315918,
      "learning_rate": 1.254648796575679e-05,
      "loss": 0.6973,
      "step": 12259
    },
    {
      "epoch": 3.871171455636249,
      "grad_norm": 3.5354714393615723,
      "learning_rate": 1.2542979440039296e-05,
      "loss": 0.6758,
      "step": 12260
    },
    {
      "epoch": 3.8714872118724344,
      "grad_norm": 3.738459825515747,
      "learning_rate": 1.2539470914321802e-05,
      "loss": 0.7159,
      "step": 12261
    },
    {
      "epoch": 3.87180296810862,
      "grad_norm": 3.486656427383423,
      "learning_rate": 1.2535962388604307e-05,
      "loss": 0.7345,
      "step": 12262
    },
    {
      "epoch": 3.872118724344806,
      "grad_norm": 3.345792055130005,
      "learning_rate": 1.2532453862886814e-05,
      "loss": 0.6604,
      "step": 12263
    },
    {
      "epoch": 3.8724344805809916,
      "grad_norm": 2.9513161182403564,
      "learning_rate": 1.2528945337169323e-05,
      "loss": 0.5697,
      "step": 12264
    },
    {
      "epoch": 3.8727502368171773,
      "grad_norm": 3.2717297077178955,
      "learning_rate": 1.252543681145183e-05,
      "loss": 0.6831,
      "step": 12265
    },
    {
      "epoch": 3.8730659930533626,
      "grad_norm": 3.1485002040863037,
      "learning_rate": 1.2521928285734336e-05,
      "loss": 0.6782,
      "step": 12266
    },
    {
      "epoch": 3.8733817492895484,
      "grad_norm": 3.6390249729156494,
      "learning_rate": 1.2518419760016842e-05,
      "loss": 0.65,
      "step": 12267
    },
    {
      "epoch": 3.873697505525734,
      "grad_norm": 3.095287322998047,
      "learning_rate": 1.2514911234299348e-05,
      "loss": 0.702,
      "step": 12268
    },
    {
      "epoch": 3.87401326176192,
      "grad_norm": 4.362957954406738,
      "learning_rate": 1.2511402708581853e-05,
      "loss": 0.7687,
      "step": 12269
    },
    {
      "epoch": 3.8743290179981056,
      "grad_norm": 3.3087689876556396,
      "learning_rate": 1.250789418286436e-05,
      "loss": 0.653,
      "step": 12270
    },
    {
      "epoch": 3.874644774234291,
      "grad_norm": 3.3509154319763184,
      "learning_rate": 1.2504385657146867e-05,
      "loss": 0.6784,
      "step": 12271
    },
    {
      "epoch": 3.874960530470477,
      "grad_norm": 4.033886909484863,
      "learning_rate": 1.2500877131429372e-05,
      "loss": 0.7688,
      "step": 12272
    },
    {
      "epoch": 3.8752762867066624,
      "grad_norm": 4.559109687805176,
      "learning_rate": 1.249736860571188e-05,
      "loss": 0.7914,
      "step": 12273
    },
    {
      "epoch": 3.875592042942848,
      "grad_norm": 3.085862398147583,
      "learning_rate": 1.2493860079994386e-05,
      "loss": 0.5638,
      "step": 12274
    },
    {
      "epoch": 3.875907799179034,
      "grad_norm": 3.893747329711914,
      "learning_rate": 1.2490351554276894e-05,
      "loss": 0.677,
      "step": 12275
    },
    {
      "epoch": 3.8762235554152196,
      "grad_norm": 3.403141498565674,
      "learning_rate": 1.24868430285594e-05,
      "loss": 0.6097,
      "step": 12276
    },
    {
      "epoch": 3.8765393116514053,
      "grad_norm": 4.2677741050720215,
      "learning_rate": 1.2483334502841906e-05,
      "loss": 0.6399,
      "step": 12277
    },
    {
      "epoch": 3.8768550678875906,
      "grad_norm": 3.3412961959838867,
      "learning_rate": 1.2479825977124413e-05,
      "loss": 0.6346,
      "step": 12278
    },
    {
      "epoch": 3.8771708241237763,
      "grad_norm": 3.09281587600708,
      "learning_rate": 1.247631745140692e-05,
      "loss": 0.6039,
      "step": 12279
    },
    {
      "epoch": 3.877486580359962,
      "grad_norm": 3.630636692047119,
      "learning_rate": 1.2472808925689427e-05,
      "loss": 0.7381,
      "step": 12280
    },
    {
      "epoch": 3.877802336596148,
      "grad_norm": 3.301727533340454,
      "learning_rate": 1.2469300399971932e-05,
      "loss": 0.7494,
      "step": 12281
    },
    {
      "epoch": 3.8781180928323336,
      "grad_norm": 3.785487413406372,
      "learning_rate": 1.2465791874254439e-05,
      "loss": 0.6395,
      "step": 12282
    },
    {
      "epoch": 3.878433849068519,
      "grad_norm": 3.857924222946167,
      "learning_rate": 1.2462283348536945e-05,
      "loss": 0.6361,
      "step": 12283
    },
    {
      "epoch": 3.878749605304705,
      "grad_norm": 2.5044922828674316,
      "learning_rate": 1.2458774822819452e-05,
      "loss": 0.6131,
      "step": 12284
    },
    {
      "epoch": 3.8790653615408903,
      "grad_norm": 3.904142141342163,
      "learning_rate": 1.2455266297101959e-05,
      "loss": 0.6167,
      "step": 12285
    },
    {
      "epoch": 3.879381117777076,
      "grad_norm": 3.394141674041748,
      "learning_rate": 1.2451757771384464e-05,
      "loss": 0.5955,
      "step": 12286
    },
    {
      "epoch": 3.879696874013262,
      "grad_norm": 4.023738861083984,
      "learning_rate": 1.2448249245666971e-05,
      "loss": 0.6278,
      "step": 12287
    },
    {
      "epoch": 3.8800126302494475,
      "grad_norm": 3.6414108276367188,
      "learning_rate": 1.2444740719949478e-05,
      "loss": 0.702,
      "step": 12288
    },
    {
      "epoch": 3.8803283864856333,
      "grad_norm": 3.8249869346618652,
      "learning_rate": 1.2441232194231985e-05,
      "loss": 0.6679,
      "step": 12289
    },
    {
      "epoch": 3.8806441427218186,
      "grad_norm": 3.12278413772583,
      "learning_rate": 1.2437723668514491e-05,
      "loss": 0.5928,
      "step": 12290
    },
    {
      "epoch": 3.8809598989580043,
      "grad_norm": 3.389355182647705,
      "learning_rate": 1.2434215142796997e-05,
      "loss": 0.6369,
      "step": 12291
    },
    {
      "epoch": 3.88127565519419,
      "grad_norm": 3.2207016944885254,
      "learning_rate": 1.2430706617079505e-05,
      "loss": 0.6577,
      "step": 12292
    },
    {
      "epoch": 3.881591411430376,
      "grad_norm": 3.7949788570404053,
      "learning_rate": 1.242719809136201e-05,
      "loss": 0.6849,
      "step": 12293
    },
    {
      "epoch": 3.8819071676665615,
      "grad_norm": 2.9784648418426514,
      "learning_rate": 1.2423689565644517e-05,
      "loss": 0.6079,
      "step": 12294
    },
    {
      "epoch": 3.882222923902747,
      "grad_norm": 3.584777355194092,
      "learning_rate": 1.2420181039927024e-05,
      "loss": 0.7401,
      "step": 12295
    },
    {
      "epoch": 3.882538680138933,
      "grad_norm": 3.7699716091156006,
      "learning_rate": 1.2416672514209529e-05,
      "loss": 0.6774,
      "step": 12296
    },
    {
      "epoch": 3.8828544363751183,
      "grad_norm": 3.5171093940734863,
      "learning_rate": 1.2413163988492037e-05,
      "loss": 0.6451,
      "step": 12297
    },
    {
      "epoch": 3.883170192611304,
      "grad_norm": 3.2546188831329346,
      "learning_rate": 1.2409655462774543e-05,
      "loss": 0.6616,
      "step": 12298
    },
    {
      "epoch": 3.8834859488474898,
      "grad_norm": 4.121408939361572,
      "learning_rate": 1.240614693705705e-05,
      "loss": 0.6687,
      "step": 12299
    },
    {
      "epoch": 3.8838017050836755,
      "grad_norm": 3.542795419692993,
      "learning_rate": 1.2402638411339554e-05,
      "loss": 0.6084,
      "step": 12300
    },
    {
      "epoch": 3.8841174613198612,
      "grad_norm": 3.2266063690185547,
      "learning_rate": 1.2399129885622063e-05,
      "loss": 0.6233,
      "step": 12301
    },
    {
      "epoch": 3.8844332175560465,
      "grad_norm": 3.784442901611328,
      "learning_rate": 1.2395621359904568e-05,
      "loss": 0.7266,
      "step": 12302
    },
    {
      "epoch": 3.8847489737922323,
      "grad_norm": 3.1782379150390625,
      "learning_rate": 1.2392112834187075e-05,
      "loss": 0.596,
      "step": 12303
    },
    {
      "epoch": 3.885064730028418,
      "grad_norm": 3.8499960899353027,
      "learning_rate": 1.2388604308469582e-05,
      "loss": 0.7647,
      "step": 12304
    },
    {
      "epoch": 3.8853804862646037,
      "grad_norm": 2.984421730041504,
      "learning_rate": 1.2385095782752087e-05,
      "loss": 0.511,
      "step": 12305
    },
    {
      "epoch": 3.8856962425007895,
      "grad_norm": 3.8840935230255127,
      "learning_rate": 1.2381587257034595e-05,
      "loss": 0.7759,
      "step": 12306
    },
    {
      "epoch": 3.886011998736975,
      "grad_norm": 3.647003173828125,
      "learning_rate": 1.23780787313171e-05,
      "loss": 0.6699,
      "step": 12307
    },
    {
      "epoch": 3.886327754973161,
      "grad_norm": 3.7421247959136963,
      "learning_rate": 1.2374570205599607e-05,
      "loss": 0.6068,
      "step": 12308
    },
    {
      "epoch": 3.8866435112093463,
      "grad_norm": 3.966416597366333,
      "learning_rate": 1.2371061679882114e-05,
      "loss": 0.7196,
      "step": 12309
    },
    {
      "epoch": 3.886959267445532,
      "grad_norm": 3.3996315002441406,
      "learning_rate": 1.2367553154164621e-05,
      "loss": 0.6764,
      "step": 12310
    },
    {
      "epoch": 3.8872750236817177,
      "grad_norm": 4.433112621307373,
      "learning_rate": 1.2364044628447128e-05,
      "loss": 0.7015,
      "step": 12311
    },
    {
      "epoch": 3.8875907799179035,
      "grad_norm": 2.6593804359436035,
      "learning_rate": 1.2360536102729633e-05,
      "loss": 0.5867,
      "step": 12312
    },
    {
      "epoch": 3.887906536154089,
      "grad_norm": 3.5630910396575928,
      "learning_rate": 1.235702757701214e-05,
      "loss": 0.7,
      "step": 12313
    },
    {
      "epoch": 3.8882222923902745,
      "grad_norm": 3.968508720397949,
      "learning_rate": 1.2353519051294646e-05,
      "loss": 0.6894,
      "step": 12314
    },
    {
      "epoch": 3.8885380486264602,
      "grad_norm": 3.4537980556488037,
      "learning_rate": 1.2350010525577153e-05,
      "loss": 0.666,
      "step": 12315
    },
    {
      "epoch": 3.888853804862646,
      "grad_norm": 3.934107542037964,
      "learning_rate": 1.234650199985966e-05,
      "loss": 0.6401,
      "step": 12316
    },
    {
      "epoch": 3.8891695610988317,
      "grad_norm": 3.6070077419281006,
      "learning_rate": 1.2342993474142165e-05,
      "loss": 0.6825,
      "step": 12317
    },
    {
      "epoch": 3.8894853173350175,
      "grad_norm": 3.5176961421966553,
      "learning_rate": 1.2339484948424672e-05,
      "loss": 0.7126,
      "step": 12318
    },
    {
      "epoch": 3.889801073571203,
      "grad_norm": 3.076857328414917,
      "learning_rate": 1.2335976422707179e-05,
      "loss": 0.6553,
      "step": 12319
    },
    {
      "epoch": 3.890116829807389,
      "grad_norm": 3.514991283416748,
      "learning_rate": 1.2332467896989686e-05,
      "loss": 0.679,
      "step": 12320
    },
    {
      "epoch": 3.8904325860435742,
      "grad_norm": 3.587110757827759,
      "learning_rate": 1.2328959371272192e-05,
      "loss": 0.6902,
      "step": 12321
    },
    {
      "epoch": 3.89074834227976,
      "grad_norm": 3.736053705215454,
      "learning_rate": 1.2325450845554698e-05,
      "loss": 0.6493,
      "step": 12322
    },
    {
      "epoch": 3.8910640985159457,
      "grad_norm": 2.8785855770111084,
      "learning_rate": 1.2321942319837206e-05,
      "loss": 0.6265,
      "step": 12323
    },
    {
      "epoch": 3.8913798547521314,
      "grad_norm": 3.293612241744995,
      "learning_rate": 1.2318433794119711e-05,
      "loss": 0.5711,
      "step": 12324
    },
    {
      "epoch": 3.891695610988317,
      "grad_norm": 3.456897258758545,
      "learning_rate": 1.2314925268402218e-05,
      "loss": 0.5906,
      "step": 12325
    },
    {
      "epoch": 3.8920113672245025,
      "grad_norm": 4.128951072692871,
      "learning_rate": 1.2311416742684725e-05,
      "loss": 0.6444,
      "step": 12326
    },
    {
      "epoch": 3.892327123460688,
      "grad_norm": 3.221144437789917,
      "learning_rate": 1.2307908216967232e-05,
      "loss": 0.6285,
      "step": 12327
    },
    {
      "epoch": 3.892642879696874,
      "grad_norm": 3.541043519973755,
      "learning_rate": 1.2304399691249738e-05,
      "loss": 0.6958,
      "step": 12328
    },
    {
      "epoch": 3.8929586359330597,
      "grad_norm": 4.212547302246094,
      "learning_rate": 1.2300891165532244e-05,
      "loss": 0.7095,
      "step": 12329
    },
    {
      "epoch": 3.8932743921692454,
      "grad_norm": 2.9771568775177,
      "learning_rate": 1.229738263981475e-05,
      "loss": 0.6848,
      "step": 12330
    },
    {
      "epoch": 3.893590148405431,
      "grad_norm": 4.271871089935303,
      "learning_rate": 1.2293874114097256e-05,
      "loss": 0.7882,
      "step": 12331
    },
    {
      "epoch": 3.893905904641617,
      "grad_norm": 3.560016393661499,
      "learning_rate": 1.2290365588379764e-05,
      "loss": 0.521,
      "step": 12332
    },
    {
      "epoch": 3.894221660877802,
      "grad_norm": 3.580913782119751,
      "learning_rate": 1.2286857062662271e-05,
      "loss": 0.6934,
      "step": 12333
    },
    {
      "epoch": 3.894537417113988,
      "grad_norm": 3.9874956607818604,
      "learning_rate": 1.2283348536944776e-05,
      "loss": 0.6755,
      "step": 12334
    },
    {
      "epoch": 3.8948531733501737,
      "grad_norm": 3.3803038597106934,
      "learning_rate": 1.2279840011227283e-05,
      "loss": 0.6237,
      "step": 12335
    },
    {
      "epoch": 3.8951689295863594,
      "grad_norm": 3.4696598052978516,
      "learning_rate": 1.227633148550979e-05,
      "loss": 0.5334,
      "step": 12336
    },
    {
      "epoch": 3.895484685822545,
      "grad_norm": 4.555788993835449,
      "learning_rate": 1.2272822959792296e-05,
      "loss": 0.7485,
      "step": 12337
    },
    {
      "epoch": 3.8958004420587304,
      "grad_norm": 3.5624876022338867,
      "learning_rate": 1.2269314434074802e-05,
      "loss": 0.7393,
      "step": 12338
    },
    {
      "epoch": 3.896116198294916,
      "grad_norm": 3.120274782180786,
      "learning_rate": 1.2265805908357308e-05,
      "loss": 0.6129,
      "step": 12339
    },
    {
      "epoch": 3.896431954531102,
      "grad_norm": 3.587146043777466,
      "learning_rate": 1.2262297382639815e-05,
      "loss": 0.6387,
      "step": 12340
    },
    {
      "epoch": 3.8967477107672877,
      "grad_norm": 2.9740560054779053,
      "learning_rate": 1.2258788856922322e-05,
      "loss": 0.721,
      "step": 12341
    },
    {
      "epoch": 3.8970634670034734,
      "grad_norm": 3.983314275741577,
      "learning_rate": 1.2255280331204829e-05,
      "loss": 0.6365,
      "step": 12342
    },
    {
      "epoch": 3.897379223239659,
      "grad_norm": 3.3676931858062744,
      "learning_rate": 1.2251771805487334e-05,
      "loss": 0.7634,
      "step": 12343
    },
    {
      "epoch": 3.897694979475845,
      "grad_norm": 4.046625137329102,
      "learning_rate": 1.224826327976984e-05,
      "loss": 0.6438,
      "step": 12344
    },
    {
      "epoch": 3.89801073571203,
      "grad_norm": 4.534131050109863,
      "learning_rate": 1.2244754754052348e-05,
      "loss": 0.7868,
      "step": 12345
    },
    {
      "epoch": 3.898326491948216,
      "grad_norm": 3.8582746982574463,
      "learning_rate": 1.2241246228334854e-05,
      "loss": 0.7245,
      "step": 12346
    },
    {
      "epoch": 3.8986422481844016,
      "grad_norm": 2.922698736190796,
      "learning_rate": 1.2237737702617361e-05,
      "loss": 0.592,
      "step": 12347
    },
    {
      "epoch": 3.8989580044205874,
      "grad_norm": 3.3390591144561768,
      "learning_rate": 1.2234229176899866e-05,
      "loss": 0.7095,
      "step": 12348
    },
    {
      "epoch": 3.899273760656773,
      "grad_norm": 2.924816608428955,
      "learning_rate": 1.2230720651182375e-05,
      "loss": 0.6006,
      "step": 12349
    },
    {
      "epoch": 3.8995895168929584,
      "grad_norm": 3.237454414367676,
      "learning_rate": 1.222721212546488e-05,
      "loss": 0.6475,
      "step": 12350
    },
    {
      "epoch": 3.899905273129144,
      "grad_norm": 3.6324045658111572,
      "learning_rate": 1.2223703599747387e-05,
      "loss": 0.6729,
      "step": 12351
    },
    {
      "epoch": 3.90022102936533,
      "grad_norm": 3.7986721992492676,
      "learning_rate": 1.2220195074029894e-05,
      "loss": 0.7125,
      "step": 12352
    },
    {
      "epoch": 3.9005367856015156,
      "grad_norm": 3.707918167114258,
      "learning_rate": 1.2216686548312399e-05,
      "loss": 0.7304,
      "step": 12353
    },
    {
      "epoch": 3.9008525418377014,
      "grad_norm": 2.969672918319702,
      "learning_rate": 1.2213178022594907e-05,
      "loss": 0.6612,
      "step": 12354
    },
    {
      "epoch": 3.901168298073887,
      "grad_norm": 3.442828893661499,
      "learning_rate": 1.2209669496877412e-05,
      "loss": 0.5361,
      "step": 12355
    },
    {
      "epoch": 3.901484054310073,
      "grad_norm": 3.9379773139953613,
      "learning_rate": 1.2206160971159919e-05,
      "loss": 0.6744,
      "step": 12356
    },
    {
      "epoch": 3.901799810546258,
      "grad_norm": 3.9942708015441895,
      "learning_rate": 1.2202652445442426e-05,
      "loss": 0.6766,
      "step": 12357
    },
    {
      "epoch": 3.902115566782444,
      "grad_norm": 3.613953113555908,
      "learning_rate": 1.2199143919724933e-05,
      "loss": 0.574,
      "step": 12358
    },
    {
      "epoch": 3.9024313230186296,
      "grad_norm": 3.704000473022461,
      "learning_rate": 1.219563539400744e-05,
      "loss": 0.7339,
      "step": 12359
    },
    {
      "epoch": 3.9027470792548153,
      "grad_norm": 2.755598306655884,
      "learning_rate": 1.2192126868289945e-05,
      "loss": 0.5716,
      "step": 12360
    },
    {
      "epoch": 3.903062835491001,
      "grad_norm": 3.4095728397369385,
      "learning_rate": 1.2188618342572451e-05,
      "loss": 0.6401,
      "step": 12361
    },
    {
      "epoch": 3.9033785917271864,
      "grad_norm": 3.1983489990234375,
      "learning_rate": 1.2185109816854958e-05,
      "loss": 0.6711,
      "step": 12362
    },
    {
      "epoch": 3.903694347963372,
      "grad_norm": 3.3366215229034424,
      "learning_rate": 1.2181601291137465e-05,
      "loss": 0.5758,
      "step": 12363
    },
    {
      "epoch": 3.904010104199558,
      "grad_norm": 3.927976608276367,
      "learning_rate": 1.2178092765419972e-05,
      "loss": 0.6719,
      "step": 12364
    },
    {
      "epoch": 3.9043258604357436,
      "grad_norm": 2.8691861629486084,
      "learning_rate": 1.2174584239702477e-05,
      "loss": 0.5828,
      "step": 12365
    },
    {
      "epoch": 3.9046416166719293,
      "grad_norm": 4.33198881149292,
      "learning_rate": 1.2171075713984984e-05,
      "loss": 0.5803,
      "step": 12366
    },
    {
      "epoch": 3.904957372908115,
      "grad_norm": 2.8690834045410156,
      "learning_rate": 1.216756718826749e-05,
      "loss": 0.6024,
      "step": 12367
    },
    {
      "epoch": 3.905273129144301,
      "grad_norm": 3.347651958465576,
      "learning_rate": 1.2164058662549997e-05,
      "loss": 0.5994,
      "step": 12368
    },
    {
      "epoch": 3.905588885380486,
      "grad_norm": 3.219097375869751,
      "learning_rate": 1.2160550136832503e-05,
      "loss": 0.6148,
      "step": 12369
    },
    {
      "epoch": 3.905904641616672,
      "grad_norm": 3.517646074295044,
      "learning_rate": 1.215704161111501e-05,
      "loss": 0.6096,
      "step": 12370
    },
    {
      "epoch": 3.9062203978528576,
      "grad_norm": 3.1044116020202637,
      "learning_rate": 1.2153533085397518e-05,
      "loss": 0.6383,
      "step": 12371
    },
    {
      "epoch": 3.9065361540890433,
      "grad_norm": 3.030268669128418,
      "learning_rate": 1.2150024559680023e-05,
      "loss": 0.6348,
      "step": 12372
    },
    {
      "epoch": 3.906851910325229,
      "grad_norm": 3.6704483032226562,
      "learning_rate": 1.214651603396253e-05,
      "loss": 0.6612,
      "step": 12373
    },
    {
      "epoch": 3.9071676665614143,
      "grad_norm": 2.6560556888580322,
      "learning_rate": 1.2143007508245035e-05,
      "loss": 0.6037,
      "step": 12374
    },
    {
      "epoch": 3.9074834227976005,
      "grad_norm": 3.391373872756958,
      "learning_rate": 1.2139498982527543e-05,
      "loss": 0.6462,
      "step": 12375
    },
    {
      "epoch": 3.907799179033786,
      "grad_norm": 3.189055919647217,
      "learning_rate": 1.2135990456810049e-05,
      "loss": 0.6258,
      "step": 12376
    },
    {
      "epoch": 3.9081149352699716,
      "grad_norm": 3.91792893409729,
      "learning_rate": 1.2132481931092555e-05,
      "loss": 0.7348,
      "step": 12377
    },
    {
      "epoch": 3.9084306915061573,
      "grad_norm": 3.8553574085235596,
      "learning_rate": 1.2128973405375062e-05,
      "loss": 0.65,
      "step": 12378
    },
    {
      "epoch": 3.908746447742343,
      "grad_norm": 3.2794876098632812,
      "learning_rate": 1.2125464879657567e-05,
      "loss": 0.5886,
      "step": 12379
    },
    {
      "epoch": 3.9090622039785288,
      "grad_norm": 3.497248411178589,
      "learning_rate": 1.2121956353940076e-05,
      "loss": 0.6684,
      "step": 12380
    },
    {
      "epoch": 3.909377960214714,
      "grad_norm": 2.696685791015625,
      "learning_rate": 1.2118447828222581e-05,
      "loss": 0.6495,
      "step": 12381
    },
    {
      "epoch": 3.9096937164509,
      "grad_norm": 3.127255916595459,
      "learning_rate": 1.2114939302505088e-05,
      "loss": 0.5934,
      "step": 12382
    },
    {
      "epoch": 3.9100094726870855,
      "grad_norm": 3.2497313022613525,
      "learning_rate": 1.2111430776787595e-05,
      "loss": 0.7226,
      "step": 12383
    },
    {
      "epoch": 3.9103252289232713,
      "grad_norm": 2.981867551803589,
      "learning_rate": 1.2107922251070101e-05,
      "loss": 0.603,
      "step": 12384
    },
    {
      "epoch": 3.910640985159457,
      "grad_norm": 3.316481828689575,
      "learning_rate": 1.2104413725352608e-05,
      "loss": 0.7268,
      "step": 12385
    },
    {
      "epoch": 3.9109567413956423,
      "grad_norm": 3.474031686782837,
      "learning_rate": 1.2100905199635113e-05,
      "loss": 0.6929,
      "step": 12386
    },
    {
      "epoch": 3.9112724976318285,
      "grad_norm": 3.092456817626953,
      "learning_rate": 1.209739667391762e-05,
      "loss": 0.6696,
      "step": 12387
    },
    {
      "epoch": 3.911588253868014,
      "grad_norm": 3.6459109783172607,
      "learning_rate": 1.2093888148200127e-05,
      "loss": 0.6529,
      "step": 12388
    },
    {
      "epoch": 3.9119040101041995,
      "grad_norm": 3.4140777587890625,
      "learning_rate": 1.2090379622482634e-05,
      "loss": 0.5969,
      "step": 12389
    },
    {
      "epoch": 3.9122197663403853,
      "grad_norm": 3.485368013381958,
      "learning_rate": 1.208687109676514e-05,
      "loss": 0.689,
      "step": 12390
    },
    {
      "epoch": 3.912535522576571,
      "grad_norm": 3.337118148803711,
      "learning_rate": 1.2083362571047646e-05,
      "loss": 0.6577,
      "step": 12391
    },
    {
      "epoch": 3.9128512788127567,
      "grad_norm": 3.354783535003662,
      "learning_rate": 1.2079854045330152e-05,
      "loss": 0.6749,
      "step": 12392
    },
    {
      "epoch": 3.913167035048942,
      "grad_norm": 4.245083332061768,
      "learning_rate": 1.207634551961266e-05,
      "loss": 0.6805,
      "step": 12393
    },
    {
      "epoch": 3.9134827912851278,
      "grad_norm": 3.3749003410339355,
      "learning_rate": 1.2072836993895166e-05,
      "loss": 0.7145,
      "step": 12394
    },
    {
      "epoch": 3.9137985475213135,
      "grad_norm": 4.189378261566162,
      "learning_rate": 1.2069328468177673e-05,
      "loss": 0.7836,
      "step": 12395
    },
    {
      "epoch": 3.9141143037574992,
      "grad_norm": 3.3660786151885986,
      "learning_rate": 1.2065819942460178e-05,
      "loss": 0.764,
      "step": 12396
    },
    {
      "epoch": 3.914430059993685,
      "grad_norm": 3.3834152221679688,
      "learning_rate": 1.2062311416742687e-05,
      "loss": 0.6965,
      "step": 12397
    },
    {
      "epoch": 3.9147458162298703,
      "grad_norm": 3.551560401916504,
      "learning_rate": 1.2058802891025192e-05,
      "loss": 0.7239,
      "step": 12398
    },
    {
      "epoch": 3.9150615724660565,
      "grad_norm": 2.9926376342773438,
      "learning_rate": 1.2055294365307698e-05,
      "loss": 0.6273,
      "step": 12399
    },
    {
      "epoch": 3.9153773287022418,
      "grad_norm": 3.4816415309906006,
      "learning_rate": 1.2051785839590205e-05,
      "loss": 0.6609,
      "step": 12400
    },
    {
      "epoch": 3.9153773287022418,
      "eval_loss": 0.7988930344581604,
      "eval_runtime": 4.5858,
      "eval_samples_per_second": 542.978,
      "eval_steps_per_second": 4.579,
      "step": 12400
    },
    {
      "epoch": 3.9156930849384275,
      "grad_norm": 3.369330883026123,
      "learning_rate": 1.204827731387271e-05,
      "loss": 0.7189,
      "step": 12401
    },
    {
      "epoch": 3.9160088411746132,
      "grad_norm": 3.1369941234588623,
      "learning_rate": 1.2044768788155219e-05,
      "loss": 0.6564,
      "step": 12402
    },
    {
      "epoch": 3.916324597410799,
      "grad_norm": 3.342503786087036,
      "learning_rate": 1.2041260262437724e-05,
      "loss": 0.62,
      "step": 12403
    },
    {
      "epoch": 3.9166403536469847,
      "grad_norm": 3.6443963050842285,
      "learning_rate": 1.2037751736720231e-05,
      "loss": 0.6908,
      "step": 12404
    },
    {
      "epoch": 3.91695610988317,
      "grad_norm": 3.1608433723449707,
      "learning_rate": 1.2034243211002736e-05,
      "loss": 0.7028,
      "step": 12405
    },
    {
      "epoch": 3.9172718661193557,
      "grad_norm": 3.490210771560669,
      "learning_rate": 1.2030734685285244e-05,
      "loss": 0.713,
      "step": 12406
    },
    {
      "epoch": 3.9175876223555415,
      "grad_norm": 3.0001490116119385,
      "learning_rate": 1.2027226159567751e-05,
      "loss": 0.7039,
      "step": 12407
    },
    {
      "epoch": 3.917903378591727,
      "grad_norm": 3.058274984359741,
      "learning_rate": 1.2023717633850256e-05,
      "loss": 0.6302,
      "step": 12408
    },
    {
      "epoch": 3.918219134827913,
      "grad_norm": 3.515065908432007,
      "learning_rate": 1.2020209108132763e-05,
      "loss": 0.776,
      "step": 12409
    },
    {
      "epoch": 3.9185348910640982,
      "grad_norm": 3.082343816757202,
      "learning_rate": 1.201670058241527e-05,
      "loss": 0.7224,
      "step": 12410
    },
    {
      "epoch": 3.9188506473002844,
      "grad_norm": 3.3313417434692383,
      "learning_rate": 1.2013192056697777e-05,
      "loss": 0.6298,
      "step": 12411
    },
    {
      "epoch": 3.9191664035364697,
      "grad_norm": 2.9389073848724365,
      "learning_rate": 1.2009683530980282e-05,
      "loss": 0.7272,
      "step": 12412
    },
    {
      "epoch": 3.9194821597726555,
      "grad_norm": 2.880592107772827,
      "learning_rate": 1.2006175005262789e-05,
      "loss": 0.5951,
      "step": 12413
    },
    {
      "epoch": 3.919797916008841,
      "grad_norm": 3.236215591430664,
      "learning_rate": 1.2002666479545296e-05,
      "loss": 0.7024,
      "step": 12414
    },
    {
      "epoch": 3.920113672245027,
      "grad_norm": 3.966114044189453,
      "learning_rate": 1.1999157953827802e-05,
      "loss": 0.6968,
      "step": 12415
    },
    {
      "epoch": 3.9204294284812127,
      "grad_norm": 4.0284905433654785,
      "learning_rate": 1.199564942811031e-05,
      "loss": 0.7185,
      "step": 12416
    },
    {
      "epoch": 3.920745184717398,
      "grad_norm": 3.4008421897888184,
      "learning_rate": 1.1992140902392814e-05,
      "loss": 0.674,
      "step": 12417
    },
    {
      "epoch": 3.9210609409535837,
      "grad_norm": 3.3256871700286865,
      "learning_rate": 1.1988632376675321e-05,
      "loss": 0.6137,
      "step": 12418
    },
    {
      "epoch": 3.9213766971897694,
      "grad_norm": 3.3534586429595947,
      "learning_rate": 1.1985123850957828e-05,
      "loss": 0.7604,
      "step": 12419
    },
    {
      "epoch": 3.921692453425955,
      "grad_norm": 3.2772257328033447,
      "learning_rate": 1.1981615325240335e-05,
      "loss": 0.6424,
      "step": 12420
    },
    {
      "epoch": 3.922008209662141,
      "grad_norm": 3.2137062549591064,
      "learning_rate": 1.1978106799522842e-05,
      "loss": 0.6131,
      "step": 12421
    },
    {
      "epoch": 3.922323965898326,
      "grad_norm": 2.9763100147247314,
      "learning_rate": 1.1974598273805347e-05,
      "loss": 0.6134,
      "step": 12422
    },
    {
      "epoch": 3.9226397221345124,
      "grad_norm": 3.323406219482422,
      "learning_rate": 1.1971089748087854e-05,
      "loss": 0.661,
      "step": 12423
    },
    {
      "epoch": 3.9229554783706977,
      "grad_norm": 3.2110934257507324,
      "learning_rate": 1.196758122237036e-05,
      "loss": 0.6649,
      "step": 12424
    },
    {
      "epoch": 3.9232712346068834,
      "grad_norm": 3.6021006107330322,
      "learning_rate": 1.1964072696652867e-05,
      "loss": 0.5911,
      "step": 12425
    },
    {
      "epoch": 3.923586990843069,
      "grad_norm": 3.341427803039551,
      "learning_rate": 1.1960564170935374e-05,
      "loss": 0.6483,
      "step": 12426
    },
    {
      "epoch": 3.923902747079255,
      "grad_norm": 2.9510674476623535,
      "learning_rate": 1.1957055645217879e-05,
      "loss": 0.5461,
      "step": 12427
    },
    {
      "epoch": 3.9242185033154406,
      "grad_norm": 3.075239658355713,
      "learning_rate": 1.1953547119500388e-05,
      "loss": 0.7304,
      "step": 12428
    },
    {
      "epoch": 3.924534259551626,
      "grad_norm": 3.593196153640747,
      "learning_rate": 1.1950038593782893e-05,
      "loss": 0.6764,
      "step": 12429
    },
    {
      "epoch": 3.9248500157878117,
      "grad_norm": 3.2835354804992676,
      "learning_rate": 1.19465300680654e-05,
      "loss": 0.6742,
      "step": 12430
    },
    {
      "epoch": 3.9251657720239974,
      "grad_norm": 3.9206719398498535,
      "learning_rate": 1.1943021542347906e-05,
      "loss": 0.5706,
      "step": 12431
    },
    {
      "epoch": 3.925481528260183,
      "grad_norm": 3.3412914276123047,
      "learning_rate": 1.1939513016630413e-05,
      "loss": 0.6515,
      "step": 12432
    },
    {
      "epoch": 3.925797284496369,
      "grad_norm": 3.1478824615478516,
      "learning_rate": 1.193600449091292e-05,
      "loss": 0.6178,
      "step": 12433
    },
    {
      "epoch": 3.9261130407325546,
      "grad_norm": 4.278934001922607,
      "learning_rate": 1.1932495965195425e-05,
      "loss": 0.615,
      "step": 12434
    },
    {
      "epoch": 3.9264287969687404,
      "grad_norm": 3.2004542350769043,
      "learning_rate": 1.1928987439477932e-05,
      "loss": 0.5792,
      "step": 12435
    },
    {
      "epoch": 3.9267445532049257,
      "grad_norm": 3.02255916595459,
      "learning_rate": 1.1925478913760437e-05,
      "loss": 0.5563,
      "step": 12436
    },
    {
      "epoch": 3.9270603094411114,
      "grad_norm": 3.6626017093658447,
      "learning_rate": 1.1921970388042946e-05,
      "loss": 0.6718,
      "step": 12437
    },
    {
      "epoch": 3.927376065677297,
      "grad_norm": 4.1770148277282715,
      "learning_rate": 1.1918461862325452e-05,
      "loss": 0.669,
      "step": 12438
    },
    {
      "epoch": 3.927691821913483,
      "grad_norm": 3.606010675430298,
      "learning_rate": 1.1914953336607957e-05,
      "loss": 0.7252,
      "step": 12439
    },
    {
      "epoch": 3.9280075781496686,
      "grad_norm": 3.479403018951416,
      "learning_rate": 1.1911444810890464e-05,
      "loss": 0.7064,
      "step": 12440
    },
    {
      "epoch": 3.928323334385854,
      "grad_norm": 2.9479634761810303,
      "learning_rate": 1.1907936285172971e-05,
      "loss": 0.5843,
      "step": 12441
    },
    {
      "epoch": 3.9286390906220396,
      "grad_norm": 3.1038222312927246,
      "learning_rate": 1.1904427759455478e-05,
      "loss": 0.7224,
      "step": 12442
    },
    {
      "epoch": 3.9289548468582254,
      "grad_norm": 3.804750442504883,
      "learning_rate": 1.1900919233737983e-05,
      "loss": 0.701,
      "step": 12443
    },
    {
      "epoch": 3.929270603094411,
      "grad_norm": 3.246626377105713,
      "learning_rate": 1.189741070802049e-05,
      "loss": 0.5516,
      "step": 12444
    },
    {
      "epoch": 3.929586359330597,
      "grad_norm": 3.79679536819458,
      "learning_rate": 1.1893902182302998e-05,
      "loss": 0.6986,
      "step": 12445
    },
    {
      "epoch": 3.9299021155667826,
      "grad_norm": 3.605259895324707,
      "learning_rate": 1.1890393656585503e-05,
      "loss": 0.6453,
      "step": 12446
    },
    {
      "epoch": 3.9302178718029683,
      "grad_norm": 3.4389517307281494,
      "learning_rate": 1.188688513086801e-05,
      "loss": 0.7126,
      "step": 12447
    },
    {
      "epoch": 3.9305336280391536,
      "grad_norm": 3.7385356426239014,
      "learning_rate": 1.1883376605150515e-05,
      "loss": 0.6533,
      "step": 12448
    },
    {
      "epoch": 3.9308493842753394,
      "grad_norm": 3.951639413833618,
      "learning_rate": 1.1879868079433022e-05,
      "loss": 0.8231,
      "step": 12449
    },
    {
      "epoch": 3.931165140511525,
      "grad_norm": 3.2116219997406006,
      "learning_rate": 1.1876359553715529e-05,
      "loss": 0.6617,
      "step": 12450
    },
    {
      "epoch": 3.931480896747711,
      "grad_norm": 3.254253387451172,
      "learning_rate": 1.1872851027998036e-05,
      "loss": 0.6242,
      "step": 12451
    },
    {
      "epoch": 3.9317966529838966,
      "grad_norm": 3.6249606609344482,
      "learning_rate": 1.1869342502280543e-05,
      "loss": 0.7168,
      "step": 12452
    },
    {
      "epoch": 3.932112409220082,
      "grad_norm": 3.6538801193237305,
      "learning_rate": 1.1865833976563048e-05,
      "loss": 0.6855,
      "step": 12453
    },
    {
      "epoch": 3.9324281654562676,
      "grad_norm": 3.8086133003234863,
      "learning_rate": 1.1862325450845556e-05,
      "loss": 0.5779,
      "step": 12454
    },
    {
      "epoch": 3.9327439216924533,
      "grad_norm": 4.298183441162109,
      "learning_rate": 1.1858816925128061e-05,
      "loss": 0.6509,
      "step": 12455
    },
    {
      "epoch": 3.933059677928639,
      "grad_norm": 3.0016934871673584,
      "learning_rate": 1.1855308399410568e-05,
      "loss": 0.5932,
      "step": 12456
    },
    {
      "epoch": 3.933375434164825,
      "grad_norm": 3.211652994155884,
      "learning_rate": 1.1851799873693075e-05,
      "loss": 0.5693,
      "step": 12457
    },
    {
      "epoch": 3.9336911904010106,
      "grad_norm": 3.6644434928894043,
      "learning_rate": 1.184829134797558e-05,
      "loss": 0.6915,
      "step": 12458
    },
    {
      "epoch": 3.9340069466371963,
      "grad_norm": 3.7097697257995605,
      "learning_rate": 1.1844782822258089e-05,
      "loss": 0.7091,
      "step": 12459
    },
    {
      "epoch": 3.9343227028733816,
      "grad_norm": 4.102896213531494,
      "learning_rate": 1.1841274296540594e-05,
      "loss": 0.6461,
      "step": 12460
    },
    {
      "epoch": 3.9346384591095673,
      "grad_norm": 3.1921422481536865,
      "learning_rate": 1.18377657708231e-05,
      "loss": 0.6061,
      "step": 12461
    },
    {
      "epoch": 3.934954215345753,
      "grad_norm": 2.8673317432403564,
      "learning_rate": 1.1834257245105607e-05,
      "loss": 0.5569,
      "step": 12462
    },
    {
      "epoch": 3.935269971581939,
      "grad_norm": 3.371962070465088,
      "learning_rate": 1.1830748719388114e-05,
      "loss": 0.6315,
      "step": 12463
    },
    {
      "epoch": 3.9355857278181245,
      "grad_norm": 3.321746826171875,
      "learning_rate": 1.1827240193670621e-05,
      "loss": 0.6963,
      "step": 12464
    },
    {
      "epoch": 3.93590148405431,
      "grad_norm": 4.0998053550720215,
      "learning_rate": 1.1823731667953126e-05,
      "loss": 0.6903,
      "step": 12465
    },
    {
      "epoch": 3.9362172402904956,
      "grad_norm": 3.6661643981933594,
      "learning_rate": 1.1820223142235633e-05,
      "loss": 0.6817,
      "step": 12466
    },
    {
      "epoch": 3.9365329965266813,
      "grad_norm": 4.026695728302002,
      "learning_rate": 1.181671461651814e-05,
      "loss": 0.6484,
      "step": 12467
    },
    {
      "epoch": 3.936848752762867,
      "grad_norm": 3.4931015968322754,
      "learning_rate": 1.1813206090800647e-05,
      "loss": 0.6663,
      "step": 12468
    },
    {
      "epoch": 3.937164508999053,
      "grad_norm": 3.2404134273529053,
      "learning_rate": 1.1809697565083153e-05,
      "loss": 0.6095,
      "step": 12469
    },
    {
      "epoch": 3.9374802652352385,
      "grad_norm": 3.371272087097168,
      "learning_rate": 1.1806189039365658e-05,
      "loss": 0.6704,
      "step": 12470
    },
    {
      "epoch": 3.9377960214714243,
      "grad_norm": 3.1426289081573486,
      "learning_rate": 1.1802680513648165e-05,
      "loss": 0.6857,
      "step": 12471
    },
    {
      "epoch": 3.9381117777076096,
      "grad_norm": 3.0865767002105713,
      "learning_rate": 1.1799171987930672e-05,
      "loss": 0.5916,
      "step": 12472
    },
    {
      "epoch": 3.9384275339437953,
      "grad_norm": 3.2213542461395264,
      "learning_rate": 1.1795663462213179e-05,
      "loss": 0.5899,
      "step": 12473
    },
    {
      "epoch": 3.938743290179981,
      "grad_norm": 2.8739631175994873,
      "learning_rate": 1.1792154936495684e-05,
      "loss": 0.571,
      "step": 12474
    },
    {
      "epoch": 3.9390590464161668,
      "grad_norm": 3.379096031188965,
      "learning_rate": 1.1788646410778191e-05,
      "loss": 0.5914,
      "step": 12475
    },
    {
      "epoch": 3.9393748026523525,
      "grad_norm": 3.054727554321289,
      "learning_rate": 1.17851378850607e-05,
      "loss": 0.6161,
      "step": 12476
    },
    {
      "epoch": 3.939690558888538,
      "grad_norm": 3.2323825359344482,
      "learning_rate": 1.1781629359343204e-05,
      "loss": 0.6341,
      "step": 12477
    },
    {
      "epoch": 3.940006315124724,
      "grad_norm": 3.8684675693511963,
      "learning_rate": 1.1778120833625711e-05,
      "loss": 0.6,
      "step": 12478
    },
    {
      "epoch": 3.9403220713609093,
      "grad_norm": 3.498840570449829,
      "learning_rate": 1.1774612307908216e-05,
      "loss": 0.7838,
      "step": 12479
    },
    {
      "epoch": 3.940637827597095,
      "grad_norm": 3.6991848945617676,
      "learning_rate": 1.1771103782190725e-05,
      "loss": 0.6436,
      "step": 12480
    },
    {
      "epoch": 3.9409535838332808,
      "grad_norm": 3.48394775390625,
      "learning_rate": 1.176759525647323e-05,
      "loss": 0.6583,
      "step": 12481
    },
    {
      "epoch": 3.9412693400694665,
      "grad_norm": 3.144134044647217,
      "learning_rate": 1.1764086730755737e-05,
      "loss": 0.6016,
      "step": 12482
    },
    {
      "epoch": 3.9415850963056522,
      "grad_norm": 2.981212854385376,
      "learning_rate": 1.1760578205038244e-05,
      "loss": 0.6192,
      "step": 12483
    },
    {
      "epoch": 3.9419008525418375,
      "grad_norm": 4.237020015716553,
      "learning_rate": 1.1757069679320749e-05,
      "loss": 0.7504,
      "step": 12484
    },
    {
      "epoch": 3.9422166087780233,
      "grad_norm": 3.3510873317718506,
      "learning_rate": 1.1753561153603257e-05,
      "loss": 0.7193,
      "step": 12485
    },
    {
      "epoch": 3.942532365014209,
      "grad_norm": 3.9141337871551514,
      "learning_rate": 1.1750052627885762e-05,
      "loss": 0.6038,
      "step": 12486
    },
    {
      "epoch": 3.9428481212503947,
      "grad_norm": 3.3458034992218018,
      "learning_rate": 1.174654410216827e-05,
      "loss": 0.6386,
      "step": 12487
    },
    {
      "epoch": 3.9431638774865805,
      "grad_norm": 3.746422529220581,
      "learning_rate": 1.1743035576450776e-05,
      "loss": 0.6546,
      "step": 12488
    },
    {
      "epoch": 3.9434796337227658,
      "grad_norm": 3.2426912784576416,
      "learning_rate": 1.1739527050733283e-05,
      "loss": 0.6076,
      "step": 12489
    },
    {
      "epoch": 3.943795389958952,
      "grad_norm": 3.5003254413604736,
      "learning_rate": 1.173601852501579e-05,
      "loss": 0.6451,
      "step": 12490
    },
    {
      "epoch": 3.9441111461951373,
      "grad_norm": 2.4587903022766113,
      "learning_rate": 1.1732509999298295e-05,
      "loss": 0.4638,
      "step": 12491
    },
    {
      "epoch": 3.944426902431323,
      "grad_norm": 3.484926223754883,
      "learning_rate": 1.1729001473580802e-05,
      "loss": 0.672,
      "step": 12492
    },
    {
      "epoch": 3.9447426586675087,
      "grad_norm": 2.8731942176818848,
      "learning_rate": 1.1725492947863308e-05,
      "loss": 0.7261,
      "step": 12493
    },
    {
      "epoch": 3.9450584149036945,
      "grad_norm": 3.574281692504883,
      "learning_rate": 1.1721984422145815e-05,
      "loss": 0.6741,
      "step": 12494
    },
    {
      "epoch": 3.94537417113988,
      "grad_norm": 3.28407621383667,
      "learning_rate": 1.1718475896428322e-05,
      "loss": 0.5696,
      "step": 12495
    },
    {
      "epoch": 3.9456899273760655,
      "grad_norm": 3.175814151763916,
      "learning_rate": 1.1714967370710827e-05,
      "loss": 0.5712,
      "step": 12496
    },
    {
      "epoch": 3.9460056836122512,
      "grad_norm": 3.5940845012664795,
      "learning_rate": 1.1711458844993334e-05,
      "loss": 0.5832,
      "step": 12497
    },
    {
      "epoch": 3.946321439848437,
      "grad_norm": 3.4801950454711914,
      "learning_rate": 1.170795031927584e-05,
      "loss": 0.6588,
      "step": 12498
    },
    {
      "epoch": 3.9466371960846227,
      "grad_norm": 3.950594902038574,
      "learning_rate": 1.1704441793558348e-05,
      "loss": 0.6341,
      "step": 12499
    },
    {
      "epoch": 3.9469529523208084,
      "grad_norm": 3.2720117568969727,
      "learning_rate": 1.1700933267840854e-05,
      "loss": 0.5459,
      "step": 12500
    },
    {
      "epoch": 3.9472687085569937,
      "grad_norm": 3.445195436477661,
      "learning_rate": 1.169742474212336e-05,
      "loss": 0.6644,
      "step": 12501
    },
    {
      "epoch": 3.94758446479318,
      "grad_norm": 3.5872387886047363,
      "learning_rate": 1.1693916216405868e-05,
      "loss": 0.6978,
      "step": 12502
    },
    {
      "epoch": 3.947900221029365,
      "grad_norm": 3.7630867958068848,
      "learning_rate": 1.1690407690688373e-05,
      "loss": 0.6129,
      "step": 12503
    },
    {
      "epoch": 3.948215977265551,
      "grad_norm": 3.6818318367004395,
      "learning_rate": 1.168689916497088e-05,
      "loss": 0.6489,
      "step": 12504
    },
    {
      "epoch": 3.9485317335017367,
      "grad_norm": 4.212806701660156,
      "learning_rate": 1.1683390639253387e-05,
      "loss": 0.7221,
      "step": 12505
    },
    {
      "epoch": 3.9488474897379224,
      "grad_norm": 2.8436577320098877,
      "learning_rate": 1.1679882113535892e-05,
      "loss": 0.5472,
      "step": 12506
    },
    {
      "epoch": 3.949163245974108,
      "grad_norm": 3.0437240600585938,
      "learning_rate": 1.16763735878184e-05,
      "loss": 0.696,
      "step": 12507
    },
    {
      "epoch": 3.9494790022102935,
      "grad_norm": 3.426847457885742,
      "learning_rate": 1.1672865062100906e-05,
      "loss": 0.6264,
      "step": 12508
    },
    {
      "epoch": 3.949794758446479,
      "grad_norm": 3.4097752571105957,
      "learning_rate": 1.1669356536383412e-05,
      "loss": 0.6982,
      "step": 12509
    },
    {
      "epoch": 3.950110514682665,
      "grad_norm": 3.071432113647461,
      "learning_rate": 1.1665848010665917e-05,
      "loss": 0.6573,
      "step": 12510
    },
    {
      "epoch": 3.9504262709188507,
      "grad_norm": 3.9434354305267334,
      "learning_rate": 1.1662339484948426e-05,
      "loss": 0.6573,
      "step": 12511
    },
    {
      "epoch": 3.9507420271550364,
      "grad_norm": 3.4993536472320557,
      "learning_rate": 1.1658830959230933e-05,
      "loss": 0.6416,
      "step": 12512
    },
    {
      "epoch": 3.9510577833912217,
      "grad_norm": 3.1131021976470947,
      "learning_rate": 1.1655322433513438e-05,
      "loss": 0.6275,
      "step": 12513
    },
    {
      "epoch": 3.951373539627408,
      "grad_norm": 3.208244800567627,
      "learning_rate": 1.1651813907795945e-05,
      "loss": 0.7322,
      "step": 12514
    },
    {
      "epoch": 3.951689295863593,
      "grad_norm": 4.038217067718506,
      "learning_rate": 1.1648305382078452e-05,
      "loss": 0.5809,
      "step": 12515
    },
    {
      "epoch": 3.952005052099779,
      "grad_norm": 3.749344825744629,
      "learning_rate": 1.1644796856360958e-05,
      "loss": 0.6258,
      "step": 12516
    },
    {
      "epoch": 3.9523208083359647,
      "grad_norm": 4.615625381469727,
      "learning_rate": 1.1641288330643463e-05,
      "loss": 0.631,
      "step": 12517
    },
    {
      "epoch": 3.9526365645721504,
      "grad_norm": 3.816838264465332,
      "learning_rate": 1.163777980492597e-05,
      "loss": 0.6153,
      "step": 12518
    },
    {
      "epoch": 3.952952320808336,
      "grad_norm": 3.822800874710083,
      "learning_rate": 1.1634271279208477e-05,
      "loss": 0.7072,
      "step": 12519
    },
    {
      "epoch": 3.9532680770445214,
      "grad_norm": 3.895200729370117,
      "learning_rate": 1.1630762753490984e-05,
      "loss": 0.6806,
      "step": 12520
    },
    {
      "epoch": 3.953583833280707,
      "grad_norm": 2.886019229888916,
      "learning_rate": 1.162725422777349e-05,
      "loss": 0.5417,
      "step": 12521
    },
    {
      "epoch": 3.953899589516893,
      "grad_norm": 4.1317830085754395,
      "learning_rate": 1.1623745702055996e-05,
      "loss": 0.627,
      "step": 12522
    },
    {
      "epoch": 3.9542153457530786,
      "grad_norm": 2.8918650150299072,
      "learning_rate": 1.1620237176338503e-05,
      "loss": 0.6005,
      "step": 12523
    },
    {
      "epoch": 3.9545311019892644,
      "grad_norm": 3.971795082092285,
      "learning_rate": 1.161672865062101e-05,
      "loss": 0.6557,
      "step": 12524
    },
    {
      "epoch": 3.9548468582254497,
      "grad_norm": 4.57589864730835,
      "learning_rate": 1.1613220124903516e-05,
      "loss": 0.6654,
      "step": 12525
    },
    {
      "epoch": 3.955162614461636,
      "grad_norm": 3.197775363922119,
      "learning_rate": 1.1609711599186023e-05,
      "loss": 0.6414,
      "step": 12526
    },
    {
      "epoch": 3.955478370697821,
      "grad_norm": 3.2411272525787354,
      "learning_rate": 1.1606203073468528e-05,
      "loss": 0.5707,
      "step": 12527
    },
    {
      "epoch": 3.955794126934007,
      "grad_norm": 4.0516276359558105,
      "learning_rate": 1.1602694547751037e-05,
      "loss": 0.7483,
      "step": 12528
    },
    {
      "epoch": 3.9561098831701926,
      "grad_norm": 3.5781376361846924,
      "learning_rate": 1.1599186022033542e-05,
      "loss": 0.7139,
      "step": 12529
    },
    {
      "epoch": 3.9564256394063784,
      "grad_norm": 4.151267051696777,
      "learning_rate": 1.1595677496316049e-05,
      "loss": 0.7127,
      "step": 12530
    },
    {
      "epoch": 3.956741395642564,
      "grad_norm": 3.893772602081299,
      "learning_rate": 1.1592168970598555e-05,
      "loss": 0.6903,
      "step": 12531
    },
    {
      "epoch": 3.9570571518787494,
      "grad_norm": 4.359489917755127,
      "learning_rate": 1.158866044488106e-05,
      "loss": 0.8842,
      "step": 12532
    },
    {
      "epoch": 3.957372908114935,
      "grad_norm": 3.241065502166748,
      "learning_rate": 1.1585151919163569e-05,
      "loss": 0.6081,
      "step": 12533
    },
    {
      "epoch": 3.957688664351121,
      "grad_norm": 3.5938754081726074,
      "learning_rate": 1.1581643393446074e-05,
      "loss": 0.57,
      "step": 12534
    },
    {
      "epoch": 3.9580044205873066,
      "grad_norm": 4.018194198608398,
      "learning_rate": 1.1578134867728581e-05,
      "loss": 0.6808,
      "step": 12535
    },
    {
      "epoch": 3.9583201768234924,
      "grad_norm": 4.111396312713623,
      "learning_rate": 1.1574626342011088e-05,
      "loss": 0.7457,
      "step": 12536
    },
    {
      "epoch": 3.958635933059678,
      "grad_norm": 3.331589698791504,
      "learning_rate": 1.1571117816293595e-05,
      "loss": 0.6229,
      "step": 12537
    },
    {
      "epoch": 3.958951689295864,
      "grad_norm": 3.285391330718994,
      "learning_rate": 1.1567609290576101e-05,
      "loss": 0.6631,
      "step": 12538
    },
    {
      "epoch": 3.959267445532049,
      "grad_norm": 3.4138073921203613,
      "learning_rate": 1.1564100764858607e-05,
      "loss": 0.6047,
      "step": 12539
    },
    {
      "epoch": 3.959583201768235,
      "grad_norm": 3.649979591369629,
      "learning_rate": 1.1560592239141113e-05,
      "loss": 0.6132,
      "step": 12540
    },
    {
      "epoch": 3.9598989580044206,
      "grad_norm": 3.745137929916382,
      "learning_rate": 1.1557083713423618e-05,
      "loss": 0.6031,
      "step": 12541
    },
    {
      "epoch": 3.9602147142406063,
      "grad_norm": 3.468522071838379,
      "learning_rate": 1.1553575187706127e-05,
      "loss": 0.7126,
      "step": 12542
    },
    {
      "epoch": 3.960530470476792,
      "grad_norm": 3.7503013610839844,
      "learning_rate": 1.1550066661988634e-05,
      "loss": 0.6828,
      "step": 12543
    },
    {
      "epoch": 3.9608462267129774,
      "grad_norm": 3.7817020416259766,
      "learning_rate": 1.1546558136271139e-05,
      "loss": 0.7311,
      "step": 12544
    },
    {
      "epoch": 3.961161982949163,
      "grad_norm": 3.527782440185547,
      "learning_rate": 1.1543049610553646e-05,
      "loss": 0.5819,
      "step": 12545
    },
    {
      "epoch": 3.961477739185349,
      "grad_norm": 3.516068696975708,
      "learning_rate": 1.1539541084836153e-05,
      "loss": 0.6482,
      "step": 12546
    },
    {
      "epoch": 3.9617934954215346,
      "grad_norm": 3.4059226512908936,
      "learning_rate": 1.153603255911866e-05,
      "loss": 0.6268,
      "step": 12547
    },
    {
      "epoch": 3.9621092516577203,
      "grad_norm": 4.365283489227295,
      "learning_rate": 1.1532524033401164e-05,
      "loss": 0.6478,
      "step": 12548
    },
    {
      "epoch": 3.962425007893906,
      "grad_norm": 2.873833179473877,
      "learning_rate": 1.1529015507683671e-05,
      "loss": 0.5608,
      "step": 12549
    },
    {
      "epoch": 3.962740764130092,
      "grad_norm": 3.484707832336426,
      "learning_rate": 1.152550698196618e-05,
      "loss": 0.6327,
      "step": 12550
    },
    {
      "epoch": 3.963056520366277,
      "grad_norm": 3.1219992637634277,
      "learning_rate": 1.1521998456248685e-05,
      "loss": 0.537,
      "step": 12551
    },
    {
      "epoch": 3.963372276602463,
      "grad_norm": 3.533700942993164,
      "learning_rate": 1.1518489930531192e-05,
      "loss": 0.6317,
      "step": 12552
    },
    {
      "epoch": 3.9636880328386486,
      "grad_norm": 3.697464942932129,
      "learning_rate": 1.1514981404813697e-05,
      "loss": 0.5887,
      "step": 12553
    },
    {
      "epoch": 3.9640037890748343,
      "grad_norm": 3.6293106079101562,
      "learning_rate": 1.1511472879096204e-05,
      "loss": 0.7483,
      "step": 12554
    },
    {
      "epoch": 3.96431954531102,
      "grad_norm": 4.082148551940918,
      "learning_rate": 1.150796435337871e-05,
      "loss": 0.692,
      "step": 12555
    },
    {
      "epoch": 3.9646353015472053,
      "grad_norm": 3.705061674118042,
      "learning_rate": 1.1504455827661217e-05,
      "loss": 0.6391,
      "step": 12556
    },
    {
      "epoch": 3.964951057783391,
      "grad_norm": 4.012524127960205,
      "learning_rate": 1.1500947301943724e-05,
      "loss": 0.661,
      "step": 12557
    },
    {
      "epoch": 3.965266814019577,
      "grad_norm": 4.01923942565918,
      "learning_rate": 1.149743877622623e-05,
      "loss": 0.6595,
      "step": 12558
    },
    {
      "epoch": 3.9655825702557626,
      "grad_norm": 3.791785478591919,
      "learning_rate": 1.1493930250508738e-05,
      "loss": 0.6926,
      "step": 12559
    },
    {
      "epoch": 3.9658983264919483,
      "grad_norm": 3.4058330059051514,
      "learning_rate": 1.1490421724791243e-05,
      "loss": 0.6979,
      "step": 12560
    },
    {
      "epoch": 3.966214082728134,
      "grad_norm": 4.217468738555908,
      "learning_rate": 1.148691319907375e-05,
      "loss": 0.6799,
      "step": 12561
    },
    {
      "epoch": 3.9665298389643198,
      "grad_norm": 3.232717514038086,
      "learning_rate": 1.1483404673356256e-05,
      "loss": 0.5694,
      "step": 12562
    },
    {
      "epoch": 3.966845595200505,
      "grad_norm": 3.9962644577026367,
      "learning_rate": 1.1479896147638763e-05,
      "loss": 0.6463,
      "step": 12563
    },
    {
      "epoch": 3.967161351436691,
      "grad_norm": 3.701664924621582,
      "learning_rate": 1.147638762192127e-05,
      "loss": 0.6772,
      "step": 12564
    },
    {
      "epoch": 3.9674771076728765,
      "grad_norm": 3.7774481773376465,
      "learning_rate": 1.1472879096203775e-05,
      "loss": 0.7258,
      "step": 12565
    },
    {
      "epoch": 3.9677928639090623,
      "grad_norm": 2.9486749172210693,
      "learning_rate": 1.1469370570486282e-05,
      "loss": 0.5693,
      "step": 12566
    },
    {
      "epoch": 3.968108620145248,
      "grad_norm": 3.5235588550567627,
      "learning_rate": 1.1465862044768789e-05,
      "loss": 0.6215,
      "step": 12567
    },
    {
      "epoch": 3.9684243763814333,
      "grad_norm": 2.903944492340088,
      "learning_rate": 1.1462353519051296e-05,
      "loss": 0.5996,
      "step": 12568
    },
    {
      "epoch": 3.968740132617619,
      "grad_norm": 3.7823092937469482,
      "learning_rate": 1.1458844993333802e-05,
      "loss": 0.5842,
      "step": 12569
    },
    {
      "epoch": 3.969055888853805,
      "grad_norm": 3.2714178562164307,
      "learning_rate": 1.1455336467616308e-05,
      "loss": 0.5706,
      "step": 12570
    },
    {
      "epoch": 3.9693716450899905,
      "grad_norm": 2.895709753036499,
      "learning_rate": 1.1451827941898814e-05,
      "loss": 0.7526,
      "step": 12571
    },
    {
      "epoch": 3.9696874013261763,
      "grad_norm": 2.9918103218078613,
      "learning_rate": 1.1448319416181321e-05,
      "loss": 0.6211,
      "step": 12572
    },
    {
      "epoch": 3.970003157562362,
      "grad_norm": 3.3261191844940186,
      "learning_rate": 1.1444810890463828e-05,
      "loss": 0.6147,
      "step": 12573
    },
    {
      "epoch": 3.9703189137985477,
      "grad_norm": 4.732696056365967,
      "learning_rate": 1.1441302364746335e-05,
      "loss": 0.6174,
      "step": 12574
    },
    {
      "epoch": 3.970634670034733,
      "grad_norm": 3.334048271179199,
      "learning_rate": 1.143779383902884e-05,
      "loss": 0.7012,
      "step": 12575
    },
    {
      "epoch": 3.9709504262709188,
      "grad_norm": 3.620296001434326,
      "learning_rate": 1.1434285313311347e-05,
      "loss": 0.6837,
      "step": 12576
    },
    {
      "epoch": 3.9712661825071045,
      "grad_norm": 3.289883613586426,
      "learning_rate": 1.1430776787593854e-05,
      "loss": 0.6196,
      "step": 12577
    },
    {
      "epoch": 3.9715819387432902,
      "grad_norm": 3.1937201023101807,
      "learning_rate": 1.142726826187636e-05,
      "loss": 0.6495,
      "step": 12578
    },
    {
      "epoch": 3.971897694979476,
      "grad_norm": 4.055854320526123,
      "learning_rate": 1.1423759736158866e-05,
      "loss": 0.6697,
      "step": 12579
    },
    {
      "epoch": 3.9722134512156613,
      "grad_norm": 3.8151698112487793,
      "learning_rate": 1.1420251210441372e-05,
      "loss": 0.6749,
      "step": 12580
    },
    {
      "epoch": 3.972529207451847,
      "grad_norm": 3.5425174236297607,
      "learning_rate": 1.1416742684723881e-05,
      "loss": 0.6429,
      "step": 12581
    },
    {
      "epoch": 3.9728449636880327,
      "grad_norm": 3.699122190475464,
      "learning_rate": 1.1413234159006386e-05,
      "loss": 0.6692,
      "step": 12582
    },
    {
      "epoch": 3.9731607199242185,
      "grad_norm": 4.166167259216309,
      "learning_rate": 1.1409725633288893e-05,
      "loss": 0.7008,
      "step": 12583
    },
    {
      "epoch": 3.9734764761604042,
      "grad_norm": 3.1891212463378906,
      "learning_rate": 1.1406217107571398e-05,
      "loss": 0.6677,
      "step": 12584
    },
    {
      "epoch": 3.97379223239659,
      "grad_norm": 3.2921347618103027,
      "learning_rate": 1.1402708581853906e-05,
      "loss": 0.6158,
      "step": 12585
    },
    {
      "epoch": 3.9741079886327757,
      "grad_norm": 4.202845573425293,
      "learning_rate": 1.1399200056136412e-05,
      "loss": 0.706,
      "step": 12586
    },
    {
      "epoch": 3.974423744868961,
      "grad_norm": 3.576167345046997,
      "learning_rate": 1.1395691530418918e-05,
      "loss": 0.6772,
      "step": 12587
    },
    {
      "epoch": 3.9747395011051467,
      "grad_norm": 3.7175276279449463,
      "learning_rate": 1.1392183004701425e-05,
      "loss": 0.5903,
      "step": 12588
    },
    {
      "epoch": 3.9750552573413325,
      "grad_norm": 4.776944160461426,
      "learning_rate": 1.138867447898393e-05,
      "loss": 0.6721,
      "step": 12589
    },
    {
      "epoch": 3.975371013577518,
      "grad_norm": 3.4488143920898438,
      "learning_rate": 1.1385165953266439e-05,
      "loss": 0.5496,
      "step": 12590
    },
    {
      "epoch": 3.975686769813704,
      "grad_norm": 3.6502344608306885,
      "learning_rate": 1.1381657427548944e-05,
      "loss": 0.6084,
      "step": 12591
    },
    {
      "epoch": 3.9760025260498892,
      "grad_norm": 3.9135501384735107,
      "learning_rate": 1.137814890183145e-05,
      "loss": 0.6911,
      "step": 12592
    },
    {
      "epoch": 3.9763182822860754,
      "grad_norm": 3.3278849124908447,
      "learning_rate": 1.1374640376113958e-05,
      "loss": 0.6112,
      "step": 12593
    },
    {
      "epoch": 3.9766340385222607,
      "grad_norm": 4.190391540527344,
      "learning_rate": 1.1371131850396464e-05,
      "loss": 0.6879,
      "step": 12594
    },
    {
      "epoch": 3.9769497947584465,
      "grad_norm": 3.3106300830841064,
      "learning_rate": 1.1367623324678971e-05,
      "loss": 0.7176,
      "step": 12595
    },
    {
      "epoch": 3.977265550994632,
      "grad_norm": 3.475543737411499,
      "learning_rate": 1.1364114798961476e-05,
      "loss": 0.5722,
      "step": 12596
    },
    {
      "epoch": 3.977581307230818,
      "grad_norm": 4.132209300994873,
      "learning_rate": 1.1360606273243983e-05,
      "loss": 0.5676,
      "step": 12597
    },
    {
      "epoch": 3.9778970634670037,
      "grad_norm": 3.5442042350769043,
      "learning_rate": 1.135709774752649e-05,
      "loss": 0.7401,
      "step": 12598
    },
    {
      "epoch": 3.978212819703189,
      "grad_norm": 3.070645809173584,
      "learning_rate": 1.1353589221808997e-05,
      "loss": 0.6127,
      "step": 12599
    },
    {
      "epoch": 3.9785285759393747,
      "grad_norm": 2.9654200077056885,
      "learning_rate": 1.1350080696091504e-05,
      "loss": 0.5978,
      "step": 12600
    },
    {
      "epoch": 3.9785285759393747,
      "eval_loss": 0.8125180006027222,
      "eval_runtime": 4.5996,
      "eval_samples_per_second": 541.347,
      "eval_steps_per_second": 4.566,
      "step": 12600
    },
    {
      "epoch": 3.9788443321755604,
      "grad_norm": 3.602083683013916,
      "learning_rate": 1.1346572170374009e-05,
      "loss": 0.6341,
      "step": 12601
    },
    {
      "epoch": 3.979160088411746,
      "grad_norm": 3.541171073913574,
      "learning_rate": 1.1343063644656515e-05,
      "loss": 0.6349,
      "step": 12602
    },
    {
      "epoch": 3.979475844647932,
      "grad_norm": 3.7022435665130615,
      "learning_rate": 1.1339555118939022e-05,
      "loss": 0.6484,
      "step": 12603
    },
    {
      "epoch": 3.979791600884117,
      "grad_norm": 3.9238574504852295,
      "learning_rate": 1.1336046593221529e-05,
      "loss": 0.8045,
      "step": 12604
    },
    {
      "epoch": 3.9801073571203034,
      "grad_norm": 3.434730291366577,
      "learning_rate": 1.1332538067504036e-05,
      "loss": 0.6717,
      "step": 12605
    },
    {
      "epoch": 3.9804231133564887,
      "grad_norm": 4.355294227600098,
      "learning_rate": 1.1329029541786541e-05,
      "loss": 0.7252,
      "step": 12606
    },
    {
      "epoch": 3.9807388695926744,
      "grad_norm": 3.7730872631073,
      "learning_rate": 1.132552101606905e-05,
      "loss": 0.585,
      "step": 12607
    },
    {
      "epoch": 3.98105462582886,
      "grad_norm": 4.420048713684082,
      "learning_rate": 1.1322012490351555e-05,
      "loss": 0.6698,
      "step": 12608
    },
    {
      "epoch": 3.981370382065046,
      "grad_norm": 3.5266075134277344,
      "learning_rate": 1.1318503964634061e-05,
      "loss": 0.6563,
      "step": 12609
    },
    {
      "epoch": 3.9816861383012316,
      "grad_norm": 3.4641153812408447,
      "learning_rate": 1.1314995438916568e-05,
      "loss": 0.5781,
      "step": 12610
    },
    {
      "epoch": 3.982001894537417,
      "grad_norm": 4.1811909675598145,
      "learning_rate": 1.1311486913199073e-05,
      "loss": 0.7624,
      "step": 12611
    },
    {
      "epoch": 3.9823176507736027,
      "grad_norm": 3.7103583812713623,
      "learning_rate": 1.1307978387481582e-05,
      "loss": 0.6836,
      "step": 12612
    },
    {
      "epoch": 3.9826334070097884,
      "grad_norm": 3.4620485305786133,
      "learning_rate": 1.1304469861764087e-05,
      "loss": 0.6305,
      "step": 12613
    },
    {
      "epoch": 3.982949163245974,
      "grad_norm": 3.633160352706909,
      "learning_rate": 1.1300961336046594e-05,
      "loss": 0.5792,
      "step": 12614
    },
    {
      "epoch": 3.98326491948216,
      "grad_norm": 3.2880024909973145,
      "learning_rate": 1.1297452810329099e-05,
      "loss": 0.6933,
      "step": 12615
    },
    {
      "epoch": 3.983580675718345,
      "grad_norm": 3.437446355819702,
      "learning_rate": 1.1293944284611607e-05,
      "loss": 0.711,
      "step": 12616
    },
    {
      "epoch": 3.9838964319545314,
      "grad_norm": 3.265329122543335,
      "learning_rate": 1.1290435758894114e-05,
      "loss": 0.6089,
      "step": 12617
    },
    {
      "epoch": 3.9842121881907167,
      "grad_norm": 3.2858452796936035,
      "learning_rate": 1.128692723317662e-05,
      "loss": 0.6599,
      "step": 12618
    },
    {
      "epoch": 3.9845279444269024,
      "grad_norm": 3.455120086669922,
      "learning_rate": 1.1283418707459126e-05,
      "loss": 0.7175,
      "step": 12619
    },
    {
      "epoch": 3.984843700663088,
      "grad_norm": 4.482129096984863,
      "learning_rate": 1.1279910181741633e-05,
      "loss": 0.6611,
      "step": 12620
    },
    {
      "epoch": 3.985159456899274,
      "grad_norm": 3.227004051208496,
      "learning_rate": 1.127640165602414e-05,
      "loss": 0.5344,
      "step": 12621
    },
    {
      "epoch": 3.9854752131354596,
      "grad_norm": 3.114774465560913,
      "learning_rate": 1.1272893130306645e-05,
      "loss": 0.6012,
      "step": 12622
    },
    {
      "epoch": 3.985790969371645,
      "grad_norm": 3.1386125087738037,
      "learning_rate": 1.1269384604589152e-05,
      "loss": 0.5999,
      "step": 12623
    },
    {
      "epoch": 3.9861067256078306,
      "grad_norm": 3.91757869720459,
      "learning_rate": 1.1265876078871659e-05,
      "loss": 0.6961,
      "step": 12624
    },
    {
      "epoch": 3.9864224818440164,
      "grad_norm": 3.8936209678649902,
      "learning_rate": 1.1262367553154165e-05,
      "loss": 0.5563,
      "step": 12625
    },
    {
      "epoch": 3.986738238080202,
      "grad_norm": 3.3614003658294678,
      "learning_rate": 1.1258859027436672e-05,
      "loss": 0.6077,
      "step": 12626
    },
    {
      "epoch": 3.987053994316388,
      "grad_norm": 3.5670151710510254,
      "learning_rate": 1.1255350501719177e-05,
      "loss": 0.8372,
      "step": 12627
    },
    {
      "epoch": 3.987369750552573,
      "grad_norm": 3.3404700756073,
      "learning_rate": 1.1251841976001684e-05,
      "loss": 0.5779,
      "step": 12628
    },
    {
      "epoch": 3.9876855067887593,
      "grad_norm": 3.7667899131774902,
      "learning_rate": 1.1248333450284191e-05,
      "loss": 0.7041,
      "step": 12629
    },
    {
      "epoch": 3.9880012630249446,
      "grad_norm": 3.4718289375305176,
      "learning_rate": 1.1244824924566698e-05,
      "loss": 0.6794,
      "step": 12630
    },
    {
      "epoch": 3.9883170192611304,
      "grad_norm": 3.941533327102661,
      "learning_rate": 1.1241316398849205e-05,
      "loss": 0.7219,
      "step": 12631
    },
    {
      "epoch": 3.988632775497316,
      "grad_norm": 2.994957685470581,
      "learning_rate": 1.123780787313171e-05,
      "loss": 0.5827,
      "step": 12632
    },
    {
      "epoch": 3.988948531733502,
      "grad_norm": 3.4329471588134766,
      "learning_rate": 1.1234299347414218e-05,
      "loss": 0.652,
      "step": 12633
    },
    {
      "epoch": 3.9892642879696876,
      "grad_norm": 3.0539910793304443,
      "learning_rate": 1.1230790821696723e-05,
      "loss": 0.589,
      "step": 12634
    },
    {
      "epoch": 3.989580044205873,
      "grad_norm": 3.717594861984253,
      "learning_rate": 1.122728229597923e-05,
      "loss": 0.604,
      "step": 12635
    },
    {
      "epoch": 3.9898958004420586,
      "grad_norm": 3.4047515392303467,
      "learning_rate": 1.1223773770261737e-05,
      "loss": 0.5967,
      "step": 12636
    },
    {
      "epoch": 3.9902115566782443,
      "grad_norm": 3.395397186279297,
      "learning_rate": 1.1220265244544242e-05,
      "loss": 0.6417,
      "step": 12637
    },
    {
      "epoch": 3.99052731291443,
      "grad_norm": 3.5412049293518066,
      "learning_rate": 1.121675671882675e-05,
      "loss": 0.7475,
      "step": 12638
    },
    {
      "epoch": 3.990843069150616,
      "grad_norm": 3.6377499103546143,
      "learning_rate": 1.1213248193109256e-05,
      "loss": 0.7041,
      "step": 12639
    },
    {
      "epoch": 3.991158825386801,
      "grad_norm": 2.897925853729248,
      "learning_rate": 1.1209739667391762e-05,
      "loss": 0.6157,
      "step": 12640
    },
    {
      "epoch": 3.9914745816229873,
      "grad_norm": 3.419616222381592,
      "learning_rate": 1.120623114167427e-05,
      "loss": 0.5686,
      "step": 12641
    },
    {
      "epoch": 3.9917903378591726,
      "grad_norm": 2.967576742172241,
      "learning_rate": 1.1202722615956776e-05,
      "loss": 0.6792,
      "step": 12642
    },
    {
      "epoch": 3.9921060940953583,
      "grad_norm": 3.9787938594818115,
      "learning_rate": 1.1199214090239283e-05,
      "loss": 0.7448,
      "step": 12643
    },
    {
      "epoch": 3.992421850331544,
      "grad_norm": 3.4411795139312744,
      "learning_rate": 1.1195705564521788e-05,
      "loss": 0.5635,
      "step": 12644
    },
    {
      "epoch": 3.99273760656773,
      "grad_norm": 3.3634986877441406,
      "learning_rate": 1.1192197038804295e-05,
      "loss": 0.6074,
      "step": 12645
    },
    {
      "epoch": 3.9930533628039155,
      "grad_norm": 4.811158657073975,
      "learning_rate": 1.11886885130868e-05,
      "loss": 0.5339,
      "step": 12646
    },
    {
      "epoch": 3.993369119040101,
      "grad_norm": 3.5190820693969727,
      "learning_rate": 1.1185179987369308e-05,
      "loss": 0.5366,
      "step": 12647
    },
    {
      "epoch": 3.9936848752762866,
      "grad_norm": 3.251641273498535,
      "learning_rate": 1.1181671461651815e-05,
      "loss": 0.5068,
      "step": 12648
    },
    {
      "epoch": 3.9940006315124723,
      "grad_norm": 3.2824759483337402,
      "learning_rate": 1.117816293593432e-05,
      "loss": 0.6544,
      "step": 12649
    },
    {
      "epoch": 3.994316387748658,
      "grad_norm": 3.6831789016723633,
      "learning_rate": 1.1174654410216827e-05,
      "loss": 0.6697,
      "step": 12650
    },
    {
      "epoch": 3.994632143984844,
      "grad_norm": 4.011529445648193,
      "learning_rate": 1.1171145884499334e-05,
      "loss": 0.6479,
      "step": 12651
    },
    {
      "epoch": 3.9949479002210295,
      "grad_norm": 3.422590970993042,
      "learning_rate": 1.1167637358781841e-05,
      "loss": 0.6648,
      "step": 12652
    },
    {
      "epoch": 3.9952636564572153,
      "grad_norm": 3.950239896774292,
      "learning_rate": 1.1164128833064346e-05,
      "loss": 0.6392,
      "step": 12653
    },
    {
      "epoch": 3.9955794126934006,
      "grad_norm": 3.4363129138946533,
      "learning_rate": 1.1160620307346853e-05,
      "loss": 0.6753,
      "step": 12654
    },
    {
      "epoch": 3.9958951689295863,
      "grad_norm": 3.944061756134033,
      "learning_rate": 1.1157111781629361e-05,
      "loss": 0.6935,
      "step": 12655
    },
    {
      "epoch": 3.996210925165772,
      "grad_norm": 3.5943782329559326,
      "learning_rate": 1.1153603255911866e-05,
      "loss": 0.6584,
      "step": 12656
    },
    {
      "epoch": 3.9965266814019578,
      "grad_norm": 3.8615896701812744,
      "learning_rate": 1.1150094730194373e-05,
      "loss": 0.7124,
      "step": 12657
    },
    {
      "epoch": 3.9968424376381435,
      "grad_norm": 3.761383295059204,
      "learning_rate": 1.1146586204476878e-05,
      "loss": 0.6798,
      "step": 12658
    },
    {
      "epoch": 3.997158193874329,
      "grad_norm": 3.605548858642578,
      "learning_rate": 1.1143077678759385e-05,
      "loss": 0.6677,
      "step": 12659
    },
    {
      "epoch": 3.9974739501105145,
      "grad_norm": 3.43194842338562,
      "learning_rate": 1.1139569153041892e-05,
      "loss": 0.594,
      "step": 12660
    },
    {
      "epoch": 3.9977897063467003,
      "grad_norm": 3.8301889896392822,
      "learning_rate": 1.1136060627324399e-05,
      "loss": 0.709,
      "step": 12661
    },
    {
      "epoch": 3.998105462582886,
      "grad_norm": 3.156282424926758,
      "learning_rate": 1.1132552101606906e-05,
      "loss": 0.6494,
      "step": 12662
    },
    {
      "epoch": 3.9984212188190718,
      "grad_norm": 2.8771283626556396,
      "learning_rate": 1.112904357588941e-05,
      "loss": 0.6184,
      "step": 12663
    },
    {
      "epoch": 3.9987369750552575,
      "grad_norm": 4.034496307373047,
      "learning_rate": 1.112553505017192e-05,
      "loss": 0.6624,
      "step": 12664
    },
    {
      "epoch": 3.9990527312914432,
      "grad_norm": 3.4145846366882324,
      "learning_rate": 1.1122026524454424e-05,
      "loss": 0.6567,
      "step": 12665
    },
    {
      "epoch": 3.9993684875276285,
      "grad_norm": 3.8775129318237305,
      "learning_rate": 1.1118517998736931e-05,
      "loss": 0.562,
      "step": 12666
    },
    {
      "epoch": 3.9996842437638143,
      "grad_norm": 4.587353706359863,
      "learning_rate": 1.1115009473019438e-05,
      "loss": 0.7555,
      "step": 12667
    },
    {
      "epoch": 4.0,
      "grad_norm": 3.8451716899871826,
      "learning_rate": 1.1111500947301945e-05,
      "loss": 0.6877,
      "step": 12668
    },
    {
      "epoch": 4.000315756236185,
      "grad_norm": 3.498352289199829,
      "learning_rate": 1.1107992421584452e-05,
      "loss": 0.6223,
      "step": 12669
    },
    {
      "epoch": 4.0006315124723715,
      "grad_norm": 3.0701184272766113,
      "learning_rate": 1.1104483895866957e-05,
      "loss": 0.6111,
      "step": 12670
    },
    {
      "epoch": 4.000947268708557,
      "grad_norm": 3.7047007083892822,
      "learning_rate": 1.1100975370149464e-05,
      "loss": 0.641,
      "step": 12671
    },
    {
      "epoch": 4.001263024944743,
      "grad_norm": 2.895069122314453,
      "learning_rate": 1.109746684443197e-05,
      "loss": 0.5681,
      "step": 12672
    },
    {
      "epoch": 4.001578781180928,
      "grad_norm": 3.2834863662719727,
      "learning_rate": 1.1093958318714477e-05,
      "loss": 0.5786,
      "step": 12673
    },
    {
      "epoch": 4.001894537417114,
      "grad_norm": 2.887382984161377,
      "learning_rate": 1.1090449792996984e-05,
      "loss": 0.609,
      "step": 12674
    },
    {
      "epoch": 4.0022102936533,
      "grad_norm": 3.5482795238494873,
      "learning_rate": 1.1086941267279489e-05,
      "loss": 0.5624,
      "step": 12675
    },
    {
      "epoch": 4.002526049889485,
      "grad_norm": 3.0152711868286133,
      "learning_rate": 1.1083432741561996e-05,
      "loss": 0.5991,
      "step": 12676
    },
    {
      "epoch": 4.002841806125671,
      "grad_norm": 4.022271633148193,
      "learning_rate": 1.1079924215844503e-05,
      "loss": 0.6158,
      "step": 12677
    },
    {
      "epoch": 4.0031575623618565,
      "grad_norm": 4.354745864868164,
      "learning_rate": 1.107641569012701e-05,
      "loss": 0.7162,
      "step": 12678
    },
    {
      "epoch": 4.003473318598043,
      "grad_norm": 3.063743829727173,
      "learning_rate": 1.1072907164409516e-05,
      "loss": 0.6087,
      "step": 12679
    },
    {
      "epoch": 4.003789074834228,
      "grad_norm": 3.2119529247283936,
      "learning_rate": 1.1069398638692021e-05,
      "loss": 0.5748,
      "step": 12680
    },
    {
      "epoch": 4.004104831070413,
      "grad_norm": 3.0369932651519775,
      "learning_rate": 1.106589011297453e-05,
      "loss": 0.6049,
      "step": 12681
    },
    {
      "epoch": 4.004420587306599,
      "grad_norm": 3.135789394378662,
      "learning_rate": 1.1062381587257035e-05,
      "loss": 0.5797,
      "step": 12682
    },
    {
      "epoch": 4.004736343542785,
      "grad_norm": 3.344867467880249,
      "learning_rate": 1.1058873061539542e-05,
      "loss": 0.5729,
      "step": 12683
    },
    {
      "epoch": 4.005052099778971,
      "grad_norm": 3.4584293365478516,
      "learning_rate": 1.1055364535822047e-05,
      "loss": 0.6217,
      "step": 12684
    },
    {
      "epoch": 4.005367856015156,
      "grad_norm": 3.2549755573272705,
      "learning_rate": 1.1051856010104554e-05,
      "loss": 0.6355,
      "step": 12685
    },
    {
      "epoch": 4.005683612251342,
      "grad_norm": 3.354215145111084,
      "learning_rate": 1.1048347484387062e-05,
      "loss": 0.6307,
      "step": 12686
    },
    {
      "epoch": 4.005999368487528,
      "grad_norm": 3.288109540939331,
      "learning_rate": 1.1044838958669567e-05,
      "loss": 0.631,
      "step": 12687
    },
    {
      "epoch": 4.006315124723713,
      "grad_norm": 3.9629926681518555,
      "learning_rate": 1.1041330432952074e-05,
      "loss": 0.5262,
      "step": 12688
    },
    {
      "epoch": 4.006630880959899,
      "grad_norm": 4.3964457511901855,
      "learning_rate": 1.103782190723458e-05,
      "loss": 0.6832,
      "step": 12689
    },
    {
      "epoch": 4.0069466371960845,
      "grad_norm": 3.112879991531372,
      "learning_rate": 1.1034313381517088e-05,
      "loss": 0.5384,
      "step": 12690
    },
    {
      "epoch": 4.007262393432271,
      "grad_norm": 3.542675495147705,
      "learning_rate": 1.1030804855799593e-05,
      "loss": 0.6826,
      "step": 12691
    },
    {
      "epoch": 4.007578149668456,
      "grad_norm": 3.3860890865325928,
      "learning_rate": 1.10272963300821e-05,
      "loss": 0.662,
      "step": 12692
    },
    {
      "epoch": 4.007893905904641,
      "grad_norm": 3.7033166885375977,
      "learning_rate": 1.1023787804364607e-05,
      "loss": 0.5893,
      "step": 12693
    },
    {
      "epoch": 4.008209662140827,
      "grad_norm": 3.1826493740081787,
      "learning_rate": 1.1020279278647112e-05,
      "loss": 0.5419,
      "step": 12694
    },
    {
      "epoch": 4.008525418377013,
      "grad_norm": 3.702674150466919,
      "learning_rate": 1.101677075292962e-05,
      "loss": 0.5427,
      "step": 12695
    },
    {
      "epoch": 4.008841174613199,
      "grad_norm": 3.2112855911254883,
      "learning_rate": 1.1013262227212125e-05,
      "loss": 0.5574,
      "step": 12696
    },
    {
      "epoch": 4.009156930849384,
      "grad_norm": 3.850823402404785,
      "learning_rate": 1.1009753701494632e-05,
      "loss": 0.6573,
      "step": 12697
    },
    {
      "epoch": 4.00947268708557,
      "grad_norm": 2.6448230743408203,
      "learning_rate": 1.1006245175777139e-05,
      "loss": 0.5772,
      "step": 12698
    },
    {
      "epoch": 4.009788443321756,
      "grad_norm": 3.5038681030273438,
      "learning_rate": 1.1002736650059646e-05,
      "loss": 0.5781,
      "step": 12699
    },
    {
      "epoch": 4.010104199557941,
      "grad_norm": 2.778935432434082,
      "learning_rate": 1.0999228124342153e-05,
      "loss": 0.5245,
      "step": 12700
    },
    {
      "epoch": 4.010419955794127,
      "grad_norm": 3.4741039276123047,
      "learning_rate": 1.0995719598624658e-05,
      "loss": 0.5555,
      "step": 12701
    },
    {
      "epoch": 4.010735712030312,
      "grad_norm": 3.4967780113220215,
      "learning_rate": 1.0992211072907165e-05,
      "loss": 0.6157,
      "step": 12702
    },
    {
      "epoch": 4.011051468266499,
      "grad_norm": 3.4289097785949707,
      "learning_rate": 1.0988702547189671e-05,
      "loss": 0.6449,
      "step": 12703
    },
    {
      "epoch": 4.011367224502684,
      "grad_norm": 3.5277037620544434,
      "learning_rate": 1.0985194021472178e-05,
      "loss": 0.585,
      "step": 12704
    },
    {
      "epoch": 4.011682980738869,
      "grad_norm": 4.127574443817139,
      "learning_rate": 1.0981685495754685e-05,
      "loss": 0.6283,
      "step": 12705
    },
    {
      "epoch": 4.011998736975055,
      "grad_norm": 3.5630698204040527,
      "learning_rate": 1.097817697003719e-05,
      "loss": 0.5783,
      "step": 12706
    },
    {
      "epoch": 4.012314493211241,
      "grad_norm": 3.645472288131714,
      "learning_rate": 1.0974668444319697e-05,
      "loss": 0.6863,
      "step": 12707
    },
    {
      "epoch": 4.012630249447427,
      "grad_norm": 3.218378782272339,
      "learning_rate": 1.0971159918602204e-05,
      "loss": 0.6862,
      "step": 12708
    },
    {
      "epoch": 4.012946005683612,
      "grad_norm": 3.2739973068237305,
      "learning_rate": 1.096765139288471e-05,
      "loss": 0.7185,
      "step": 12709
    },
    {
      "epoch": 4.013261761919798,
      "grad_norm": 2.9435324668884277,
      "learning_rate": 1.0964142867167217e-05,
      "loss": 0.5311,
      "step": 12710
    },
    {
      "epoch": 4.013577518155984,
      "grad_norm": 3.539716958999634,
      "learning_rate": 1.0960634341449722e-05,
      "loss": 0.6026,
      "step": 12711
    },
    {
      "epoch": 4.013893274392169,
      "grad_norm": 2.7812225818634033,
      "learning_rate": 1.0957125815732231e-05,
      "loss": 0.4997,
      "step": 12712
    },
    {
      "epoch": 4.014209030628355,
      "grad_norm": 3.4592230319976807,
      "learning_rate": 1.0953617290014736e-05,
      "loss": 0.701,
      "step": 12713
    },
    {
      "epoch": 4.01452478686454,
      "grad_norm": 4.060657501220703,
      "learning_rate": 1.0950108764297243e-05,
      "loss": 0.6018,
      "step": 12714
    },
    {
      "epoch": 4.014840543100727,
      "grad_norm": 3.774932861328125,
      "learning_rate": 1.094660023857975e-05,
      "loss": 0.5762,
      "step": 12715
    },
    {
      "epoch": 4.015156299336912,
      "grad_norm": 3.284881830215454,
      "learning_rate": 1.0943091712862257e-05,
      "loss": 0.5632,
      "step": 12716
    },
    {
      "epoch": 4.015472055573097,
      "grad_norm": 3.7064707279205322,
      "learning_rate": 1.0939583187144763e-05,
      "loss": 0.6013,
      "step": 12717
    },
    {
      "epoch": 4.015787811809283,
      "grad_norm": 3.4411914348602295,
      "learning_rate": 1.0936074661427268e-05,
      "loss": 0.4967,
      "step": 12718
    },
    {
      "epoch": 4.016103568045469,
      "grad_norm": 3.1748738288879395,
      "learning_rate": 1.0932566135709775e-05,
      "loss": 0.558,
      "step": 12719
    },
    {
      "epoch": 4.016419324281655,
      "grad_norm": 3.6820623874664307,
      "learning_rate": 1.092905760999228e-05,
      "loss": 0.5612,
      "step": 12720
    },
    {
      "epoch": 4.01673508051784,
      "grad_norm": 3.627262830734253,
      "learning_rate": 1.0925549084274789e-05,
      "loss": 0.6087,
      "step": 12721
    },
    {
      "epoch": 4.017050836754026,
      "grad_norm": 3.0180749893188477,
      "learning_rate": 1.0922040558557296e-05,
      "loss": 0.5869,
      "step": 12722
    },
    {
      "epoch": 4.017366592990212,
      "grad_norm": 3.916923999786377,
      "learning_rate": 1.0918532032839801e-05,
      "loss": 0.6093,
      "step": 12723
    },
    {
      "epoch": 4.017682349226397,
      "grad_norm": 4.08051872253418,
      "learning_rate": 1.0915023507122308e-05,
      "loss": 0.6344,
      "step": 12724
    },
    {
      "epoch": 4.017998105462583,
      "grad_norm": 3.5765457153320312,
      "learning_rate": 1.0911514981404814e-05,
      "loss": 0.6311,
      "step": 12725
    },
    {
      "epoch": 4.018313861698768,
      "grad_norm": 3.647326707839966,
      "learning_rate": 1.0908006455687321e-05,
      "loss": 0.5652,
      "step": 12726
    },
    {
      "epoch": 4.0186296179349545,
      "grad_norm": 3.4636073112487793,
      "learning_rate": 1.0904497929969826e-05,
      "loss": 0.5852,
      "step": 12727
    },
    {
      "epoch": 4.01894537417114,
      "grad_norm": 4.414907455444336,
      "learning_rate": 1.0900989404252333e-05,
      "loss": 0.565,
      "step": 12728
    },
    {
      "epoch": 4.019261130407325,
      "grad_norm": 3.1755661964416504,
      "learning_rate": 1.089748087853484e-05,
      "loss": 0.5055,
      "step": 12729
    },
    {
      "epoch": 4.019576886643511,
      "grad_norm": 4.5708327293396,
      "learning_rate": 1.0893972352817347e-05,
      "loss": 0.5848,
      "step": 12730
    },
    {
      "epoch": 4.019892642879697,
      "grad_norm": 3.19378399848938,
      "learning_rate": 1.0890463827099854e-05,
      "loss": 0.6126,
      "step": 12731
    },
    {
      "epoch": 4.020208399115883,
      "grad_norm": 3.7164154052734375,
      "learning_rate": 1.0886955301382359e-05,
      "loss": 0.5786,
      "step": 12732
    },
    {
      "epoch": 4.020524155352068,
      "grad_norm": 3.873389720916748,
      "learning_rate": 1.0883446775664866e-05,
      "loss": 0.6376,
      "step": 12733
    },
    {
      "epoch": 4.020839911588254,
      "grad_norm": 3.714343547821045,
      "learning_rate": 1.0879938249947372e-05,
      "loss": 0.6043,
      "step": 12734
    },
    {
      "epoch": 4.02115566782444,
      "grad_norm": 3.0889737606048584,
      "learning_rate": 1.087642972422988e-05,
      "loss": 0.5293,
      "step": 12735
    },
    {
      "epoch": 4.021471424060625,
      "grad_norm": 3.7682816982269287,
      "learning_rate": 1.0872921198512386e-05,
      "loss": 0.6465,
      "step": 12736
    },
    {
      "epoch": 4.021787180296811,
      "grad_norm": 3.2600841522216797,
      "learning_rate": 1.0869412672794891e-05,
      "loss": 0.6012,
      "step": 12737
    },
    {
      "epoch": 4.022102936532996,
      "grad_norm": 3.7530903816223145,
      "learning_rate": 1.08659041470774e-05,
      "loss": 0.5433,
      "step": 12738
    },
    {
      "epoch": 4.0224186927691825,
      "grad_norm": 3.466834306716919,
      "learning_rate": 1.0862395621359905e-05,
      "loss": 0.5486,
      "step": 12739
    },
    {
      "epoch": 4.022734449005368,
      "grad_norm": 4.639438629150391,
      "learning_rate": 1.0858887095642412e-05,
      "loss": 0.6394,
      "step": 12740
    },
    {
      "epoch": 4.023050205241553,
      "grad_norm": 3.455615282058716,
      "learning_rate": 1.0855378569924918e-05,
      "loss": 0.5451,
      "step": 12741
    },
    {
      "epoch": 4.023365961477739,
      "grad_norm": 3.44549298286438,
      "learning_rate": 1.0851870044207424e-05,
      "loss": 0.6423,
      "step": 12742
    },
    {
      "epoch": 4.023681717713925,
      "grad_norm": 3.610877752304077,
      "learning_rate": 1.0848361518489932e-05,
      "loss": 0.6055,
      "step": 12743
    },
    {
      "epoch": 4.023997473950111,
      "grad_norm": 4.286133289337158,
      "learning_rate": 1.0844852992772437e-05,
      "loss": 0.5808,
      "step": 12744
    },
    {
      "epoch": 4.024313230186296,
      "grad_norm": 4.317264080047607,
      "learning_rate": 1.0841344467054944e-05,
      "loss": 0.6652,
      "step": 12745
    },
    {
      "epoch": 4.024628986422482,
      "grad_norm": 3.937243700027466,
      "learning_rate": 1.083783594133745e-05,
      "loss": 0.6204,
      "step": 12746
    },
    {
      "epoch": 4.0249447426586675,
      "grad_norm": 3.831433057785034,
      "learning_rate": 1.0834327415619958e-05,
      "loss": 0.6158,
      "step": 12747
    },
    {
      "epoch": 4.025260498894853,
      "grad_norm": 3.30081844329834,
      "learning_rate": 1.0830818889902464e-05,
      "loss": 0.5499,
      "step": 12748
    },
    {
      "epoch": 4.025576255131039,
      "grad_norm": 3.2610812187194824,
      "learning_rate": 1.082731036418497e-05,
      "loss": 0.565,
      "step": 12749
    },
    {
      "epoch": 4.025892011367224,
      "grad_norm": 4.374965190887451,
      "learning_rate": 1.0823801838467476e-05,
      "loss": 0.7204,
      "step": 12750
    },
    {
      "epoch": 4.0262077676034105,
      "grad_norm": 3.8482601642608643,
      "learning_rate": 1.0820293312749983e-05,
      "loss": 0.6194,
      "step": 12751
    },
    {
      "epoch": 4.026523523839596,
      "grad_norm": 3.7645723819732666,
      "learning_rate": 1.081678478703249e-05,
      "loss": 0.6516,
      "step": 12752
    },
    {
      "epoch": 4.026839280075781,
      "grad_norm": 3.6806583404541016,
      "learning_rate": 1.0813276261314997e-05,
      "loss": 0.6193,
      "step": 12753
    },
    {
      "epoch": 4.027155036311967,
      "grad_norm": 3.5496938228607178,
      "learning_rate": 1.0809767735597502e-05,
      "loss": 0.5181,
      "step": 12754
    },
    {
      "epoch": 4.0274707925481525,
      "grad_norm": 2.7792623043060303,
      "learning_rate": 1.0806259209880009e-05,
      "loss": 0.5122,
      "step": 12755
    },
    {
      "epoch": 4.027786548784339,
      "grad_norm": 3.013671636581421,
      "learning_rate": 1.0802750684162516e-05,
      "loss": 0.5204,
      "step": 12756
    },
    {
      "epoch": 4.028102305020524,
      "grad_norm": 3.3898212909698486,
      "learning_rate": 1.0799242158445022e-05,
      "loss": 0.6295,
      "step": 12757
    },
    {
      "epoch": 4.02841806125671,
      "grad_norm": 3.179232120513916,
      "learning_rate": 1.0795733632727527e-05,
      "loss": 0.6109,
      "step": 12758
    },
    {
      "epoch": 4.0287338174928955,
      "grad_norm": 4.234936237335205,
      "learning_rate": 1.0792225107010034e-05,
      "loss": 0.5809,
      "step": 12759
    },
    {
      "epoch": 4.029049573729081,
      "grad_norm": 3.334294080734253,
      "learning_rate": 1.0788716581292543e-05,
      "loss": 0.5651,
      "step": 12760
    },
    {
      "epoch": 4.029365329965267,
      "grad_norm": 3.7866899967193604,
      "learning_rate": 1.0785208055575048e-05,
      "loss": 0.6742,
      "step": 12761
    },
    {
      "epoch": 4.029681086201452,
      "grad_norm": 4.536848068237305,
      "learning_rate": 1.0781699529857555e-05,
      "loss": 0.6373,
      "step": 12762
    },
    {
      "epoch": 4.0299968424376384,
      "grad_norm": 3.3534820079803467,
      "learning_rate": 1.077819100414006e-05,
      "loss": 0.4649,
      "step": 12763
    },
    {
      "epoch": 4.030312598673824,
      "grad_norm": 3.430274724960327,
      "learning_rate": 1.0774682478422567e-05,
      "loss": 0.6357,
      "step": 12764
    },
    {
      "epoch": 4.030628354910009,
      "grad_norm": 4.393438816070557,
      "learning_rate": 1.0771173952705073e-05,
      "loss": 0.6922,
      "step": 12765
    },
    {
      "epoch": 4.030944111146195,
      "grad_norm": 3.3549137115478516,
      "learning_rate": 1.076766542698758e-05,
      "loss": 0.5085,
      "step": 12766
    },
    {
      "epoch": 4.0312598673823805,
      "grad_norm": 3.4478838443756104,
      "learning_rate": 1.0764156901270087e-05,
      "loss": 0.6338,
      "step": 12767
    },
    {
      "epoch": 4.031575623618567,
      "grad_norm": 3.304953098297119,
      "learning_rate": 1.0760648375552592e-05,
      "loss": 0.6627,
      "step": 12768
    },
    {
      "epoch": 4.031891379854752,
      "grad_norm": 3.4524905681610107,
      "learning_rate": 1.07571398498351e-05,
      "loss": 0.5904,
      "step": 12769
    },
    {
      "epoch": 4.032207136090938,
      "grad_norm": 4.918898105621338,
      "learning_rate": 1.0753631324117606e-05,
      "loss": 0.6804,
      "step": 12770
    },
    {
      "epoch": 4.0325228923271235,
      "grad_norm": 3.254105806350708,
      "learning_rate": 1.0750122798400113e-05,
      "loss": 0.6479,
      "step": 12771
    },
    {
      "epoch": 4.032838648563309,
      "grad_norm": 3.4445724487304688,
      "learning_rate": 1.074661427268262e-05,
      "loss": 0.5836,
      "step": 12772
    },
    {
      "epoch": 4.033154404799495,
      "grad_norm": 3.412180185317993,
      "learning_rate": 1.0743105746965126e-05,
      "loss": 0.5699,
      "step": 12773
    },
    {
      "epoch": 4.03347016103568,
      "grad_norm": 3.1919217109680176,
      "learning_rate": 1.0739597221247633e-05,
      "loss": 0.5676,
      "step": 12774
    },
    {
      "epoch": 4.033785917271866,
      "grad_norm": 3.1155385971069336,
      "learning_rate": 1.0736088695530138e-05,
      "loss": 0.5449,
      "step": 12775
    },
    {
      "epoch": 4.034101673508052,
      "grad_norm": 3.5524487495422363,
      "learning_rate": 1.0732580169812645e-05,
      "loss": 0.621,
      "step": 12776
    },
    {
      "epoch": 4.034417429744238,
      "grad_norm": 4.570554733276367,
      "learning_rate": 1.0729071644095152e-05,
      "loss": 0.5773,
      "step": 12777
    },
    {
      "epoch": 4.034733185980423,
      "grad_norm": 3.4698357582092285,
      "learning_rate": 1.0725563118377659e-05,
      "loss": 0.6972,
      "step": 12778
    },
    {
      "epoch": 4.0350489422166085,
      "grad_norm": 3.9322776794433594,
      "learning_rate": 1.0722054592660165e-05,
      "loss": 0.6713,
      "step": 12779
    },
    {
      "epoch": 4.035364698452795,
      "grad_norm": 3.345472812652588,
      "learning_rate": 1.071854606694267e-05,
      "loss": 0.6079,
      "step": 12780
    },
    {
      "epoch": 4.03568045468898,
      "grad_norm": 4.050349235534668,
      "learning_rate": 1.0715037541225177e-05,
      "loss": 0.5605,
      "step": 12781
    },
    {
      "epoch": 4.035996210925166,
      "grad_norm": 3.861011266708374,
      "learning_rate": 1.0711529015507684e-05,
      "loss": 0.6208,
      "step": 12782
    },
    {
      "epoch": 4.036311967161351,
      "grad_norm": 4.069151401519775,
      "learning_rate": 1.0708020489790191e-05,
      "loss": 0.686,
      "step": 12783
    },
    {
      "epoch": 4.036627723397537,
      "grad_norm": 4.891412734985352,
      "learning_rate": 1.0704511964072698e-05,
      "loss": 0.5758,
      "step": 12784
    },
    {
      "epoch": 4.036943479633723,
      "grad_norm": 4.195761203765869,
      "learning_rate": 1.0701003438355203e-05,
      "loss": 0.689,
      "step": 12785
    },
    {
      "epoch": 4.037259235869908,
      "grad_norm": 3.26564621925354,
      "learning_rate": 1.0697494912637711e-05,
      "loss": 0.5251,
      "step": 12786
    },
    {
      "epoch": 4.037574992106094,
      "grad_norm": 3.4234213829040527,
      "learning_rate": 1.0693986386920217e-05,
      "loss": 0.6299,
      "step": 12787
    },
    {
      "epoch": 4.03789074834228,
      "grad_norm": 3.3485829830169678,
      "learning_rate": 1.0690477861202723e-05,
      "loss": 0.6115,
      "step": 12788
    },
    {
      "epoch": 4.038206504578466,
      "grad_norm": 3.61692476272583,
      "learning_rate": 1.0686969335485228e-05,
      "loss": 0.5743,
      "step": 12789
    },
    {
      "epoch": 4.038522260814651,
      "grad_norm": 3.920994758605957,
      "learning_rate": 1.0683460809767735e-05,
      "loss": 0.5745,
      "step": 12790
    },
    {
      "epoch": 4.0388380170508364,
      "grad_norm": 3.7648613452911377,
      "learning_rate": 1.0679952284050244e-05,
      "loss": 0.6316,
      "step": 12791
    },
    {
      "epoch": 4.039153773287023,
      "grad_norm": 3.6753218173980713,
      "learning_rate": 1.0676443758332749e-05,
      "loss": 0.6148,
      "step": 12792
    },
    {
      "epoch": 4.039469529523208,
      "grad_norm": 3.9529428482055664,
      "learning_rate": 1.0672935232615256e-05,
      "loss": 0.6258,
      "step": 12793
    },
    {
      "epoch": 4.039785285759394,
      "grad_norm": 3.1885204315185547,
      "learning_rate": 1.0669426706897761e-05,
      "loss": 0.5269,
      "step": 12794
    },
    {
      "epoch": 4.040101041995579,
      "grad_norm": 3.778282642364502,
      "learning_rate": 1.066591818118027e-05,
      "loss": 0.5922,
      "step": 12795
    },
    {
      "epoch": 4.040416798231765,
      "grad_norm": 3.8185360431671143,
      "learning_rate": 1.0662409655462774e-05,
      "loss": 0.5765,
      "step": 12796
    },
    {
      "epoch": 4.040732554467951,
      "grad_norm": 4.111957550048828,
      "learning_rate": 1.0658901129745281e-05,
      "loss": 0.5956,
      "step": 12797
    },
    {
      "epoch": 4.041048310704136,
      "grad_norm": 3.827563762664795,
      "learning_rate": 1.0655392604027788e-05,
      "loss": 0.6613,
      "step": 12798
    },
    {
      "epoch": 4.041364066940322,
      "grad_norm": 3.137237071990967,
      "learning_rate": 1.0651884078310293e-05,
      "loss": 0.5437,
      "step": 12799
    },
    {
      "epoch": 4.041679823176508,
      "grad_norm": 3.8641517162323,
      "learning_rate": 1.0648375552592802e-05,
      "loss": 0.5783,
      "step": 12800
    },
    {
      "epoch": 4.041679823176508,
      "eval_loss": 0.8196029663085938,
      "eval_runtime": 4.7358,
      "eval_samples_per_second": 525.788,
      "eval_steps_per_second": 4.434,
      "step": 12800
    },
    {
      "epoch": 4.041995579412694,
      "grad_norm": 3.1684036254882812,
      "learning_rate": 1.0644867026875307e-05,
      "loss": 0.6482,
      "step": 12801
    },
    {
      "epoch": 4.042311335648879,
      "grad_norm": 4.489711284637451,
      "learning_rate": 1.0641358501157814e-05,
      "loss": 0.6082,
      "step": 12802
    },
    {
      "epoch": 4.042627091885064,
      "grad_norm": 3.808777093887329,
      "learning_rate": 1.063784997544032e-05,
      "loss": 0.6165,
      "step": 12803
    },
    {
      "epoch": 4.042942848121251,
      "grad_norm": 3.7158453464508057,
      "learning_rate": 1.0634341449722827e-05,
      "loss": 0.6416,
      "step": 12804
    },
    {
      "epoch": 4.043258604357436,
      "grad_norm": 3.911083936691284,
      "learning_rate": 1.0630832924005334e-05,
      "loss": 0.757,
      "step": 12805
    },
    {
      "epoch": 4.043574360593622,
      "grad_norm": 3.8452444076538086,
      "learning_rate": 1.062732439828784e-05,
      "loss": 0.6048,
      "step": 12806
    },
    {
      "epoch": 4.043890116829807,
      "grad_norm": 4.065791606903076,
      "learning_rate": 1.0623815872570346e-05,
      "loss": 0.5691,
      "step": 12807
    },
    {
      "epoch": 4.044205873065993,
      "grad_norm": 3.2883214950561523,
      "learning_rate": 1.0620307346852853e-05,
      "loss": 0.5971,
      "step": 12808
    },
    {
      "epoch": 4.044521629302179,
      "grad_norm": 3.090543270111084,
      "learning_rate": 1.061679882113536e-05,
      "loss": 0.577,
      "step": 12809
    },
    {
      "epoch": 4.044837385538364,
      "grad_norm": 3.937971830368042,
      "learning_rate": 1.0613290295417866e-05,
      "loss": 0.6256,
      "step": 12810
    },
    {
      "epoch": 4.04515314177455,
      "grad_norm": 4.085069179534912,
      "learning_rate": 1.0609781769700372e-05,
      "loss": 0.5708,
      "step": 12811
    },
    {
      "epoch": 4.045468898010736,
      "grad_norm": 4.47459602355957,
      "learning_rate": 1.0606273243982878e-05,
      "loss": 0.6215,
      "step": 12812
    },
    {
      "epoch": 4.045784654246922,
      "grad_norm": 3.4844164848327637,
      "learning_rate": 1.0602764718265385e-05,
      "loss": 0.5966,
      "step": 12813
    },
    {
      "epoch": 4.046100410483107,
      "grad_norm": 4.207738399505615,
      "learning_rate": 1.0599256192547892e-05,
      "loss": 0.5846,
      "step": 12814
    },
    {
      "epoch": 4.046416166719292,
      "grad_norm": 4.039090156555176,
      "learning_rate": 1.0595747666830399e-05,
      "loss": 0.6104,
      "step": 12815
    },
    {
      "epoch": 4.046731922955479,
      "grad_norm": 3.670464038848877,
      "learning_rate": 1.0592239141112904e-05,
      "loss": 0.6829,
      "step": 12816
    },
    {
      "epoch": 4.047047679191664,
      "grad_norm": 3.2991857528686523,
      "learning_rate": 1.0588730615395412e-05,
      "loss": 0.5705,
      "step": 12817
    },
    {
      "epoch": 4.04736343542785,
      "grad_norm": 3.3016488552093506,
      "learning_rate": 1.0585222089677918e-05,
      "loss": 0.5177,
      "step": 12818
    },
    {
      "epoch": 4.047679191664035,
      "grad_norm": 4.176823616027832,
      "learning_rate": 1.0581713563960424e-05,
      "loss": 0.5586,
      "step": 12819
    },
    {
      "epoch": 4.047994947900221,
      "grad_norm": 3.1053669452667236,
      "learning_rate": 1.0578205038242931e-05,
      "loss": 0.6036,
      "step": 12820
    },
    {
      "epoch": 4.048310704136407,
      "grad_norm": 3.7118797302246094,
      "learning_rate": 1.0574696512525438e-05,
      "loss": 0.5665,
      "step": 12821
    },
    {
      "epoch": 4.048626460372592,
      "grad_norm": 3.7808403968811035,
      "learning_rate": 1.0571187986807945e-05,
      "loss": 0.5985,
      "step": 12822
    },
    {
      "epoch": 4.048942216608778,
      "grad_norm": 3.5596072673797607,
      "learning_rate": 1.056767946109045e-05,
      "loss": 0.5903,
      "step": 12823
    },
    {
      "epoch": 4.049257972844964,
      "grad_norm": 4.521047592163086,
      "learning_rate": 1.0564170935372957e-05,
      "loss": 0.6648,
      "step": 12824
    },
    {
      "epoch": 4.04957372908115,
      "grad_norm": 5.101657390594482,
      "learning_rate": 1.0560662409655462e-05,
      "loss": 0.5161,
      "step": 12825
    },
    {
      "epoch": 4.049889485317335,
      "grad_norm": 3.7092020511627197,
      "learning_rate": 1.055715388393797e-05,
      "loss": 0.6315,
      "step": 12826
    },
    {
      "epoch": 4.05020524155352,
      "grad_norm": 3.922889232635498,
      "learning_rate": 1.0553645358220477e-05,
      "loss": 0.5261,
      "step": 12827
    },
    {
      "epoch": 4.0505209977897065,
      "grad_norm": 3.141770601272583,
      "learning_rate": 1.0550136832502982e-05,
      "loss": 0.5085,
      "step": 12828
    },
    {
      "epoch": 4.050836754025892,
      "grad_norm": 3.210174322128296,
      "learning_rate": 1.054662830678549e-05,
      "loss": 0.5115,
      "step": 12829
    },
    {
      "epoch": 4.051152510262078,
      "grad_norm": 3.3138368129730225,
      "learning_rate": 1.0543119781067996e-05,
      "loss": 0.5911,
      "step": 12830
    },
    {
      "epoch": 4.051468266498263,
      "grad_norm": 4.613288402557373,
      "learning_rate": 1.0539611255350503e-05,
      "loss": 0.5391,
      "step": 12831
    },
    {
      "epoch": 4.051784022734449,
      "grad_norm": 3.303922176361084,
      "learning_rate": 1.0536102729633008e-05,
      "loss": 0.5462,
      "step": 12832
    },
    {
      "epoch": 4.052099778970635,
      "grad_norm": 4.408054351806641,
      "learning_rate": 1.0532594203915515e-05,
      "loss": 0.6983,
      "step": 12833
    },
    {
      "epoch": 4.05241553520682,
      "grad_norm": 3.8117003440856934,
      "learning_rate": 1.0529085678198023e-05,
      "loss": 0.5955,
      "step": 12834
    },
    {
      "epoch": 4.052731291443006,
      "grad_norm": 3.745318651199341,
      "learning_rate": 1.0525577152480528e-05,
      "loss": 0.6486,
      "step": 12835
    },
    {
      "epoch": 4.0530470476791916,
      "grad_norm": 3.734248638153076,
      "learning_rate": 1.0522068626763035e-05,
      "loss": 0.5049,
      "step": 12836
    },
    {
      "epoch": 4.053362803915378,
      "grad_norm": 3.2087063789367676,
      "learning_rate": 1.051856010104554e-05,
      "loss": 0.5398,
      "step": 12837
    },
    {
      "epoch": 4.053678560151563,
      "grad_norm": 3.334118604660034,
      "learning_rate": 1.0515051575328047e-05,
      "loss": 0.6008,
      "step": 12838
    },
    {
      "epoch": 4.053994316387748,
      "grad_norm": 3.5832531452178955,
      "learning_rate": 1.0511543049610554e-05,
      "loss": 0.5969,
      "step": 12839
    },
    {
      "epoch": 4.0543100726239345,
      "grad_norm": 3.2711238861083984,
      "learning_rate": 1.050803452389306e-05,
      "loss": 0.5385,
      "step": 12840
    },
    {
      "epoch": 4.05462582886012,
      "grad_norm": 3.3619463443756104,
      "learning_rate": 1.0504525998175568e-05,
      "loss": 0.6852,
      "step": 12841
    },
    {
      "epoch": 4.054941585096306,
      "grad_norm": 4.03378438949585,
      "learning_rate": 1.0501017472458073e-05,
      "loss": 0.6688,
      "step": 12842
    },
    {
      "epoch": 4.055257341332491,
      "grad_norm": 4.180802822113037,
      "learning_rate": 1.0497508946740581e-05,
      "loss": 0.6606,
      "step": 12843
    },
    {
      "epoch": 4.055573097568677,
      "grad_norm": 3.049896717071533,
      "learning_rate": 1.0494000421023086e-05,
      "loss": 0.5322,
      "step": 12844
    },
    {
      "epoch": 4.055888853804863,
      "grad_norm": 3.7042834758758545,
      "learning_rate": 1.0490491895305593e-05,
      "loss": 0.5052,
      "step": 12845
    },
    {
      "epoch": 4.056204610041048,
      "grad_norm": 3.6380903720855713,
      "learning_rate": 1.04869833695881e-05,
      "loss": 0.5789,
      "step": 12846
    },
    {
      "epoch": 4.056520366277234,
      "grad_norm": 3.0472195148468018,
      "learning_rate": 1.0483474843870605e-05,
      "loss": 0.5394,
      "step": 12847
    },
    {
      "epoch": 4.0568361225134195,
      "grad_norm": 4.606112480163574,
      "learning_rate": 1.0479966318153114e-05,
      "loss": 0.6796,
      "step": 12848
    },
    {
      "epoch": 4.057151878749606,
      "grad_norm": 3.673748016357422,
      "learning_rate": 1.0476457792435619e-05,
      "loss": 0.6336,
      "step": 12849
    },
    {
      "epoch": 4.057467634985791,
      "grad_norm": 3.859055995941162,
      "learning_rate": 1.0472949266718125e-05,
      "loss": 0.516,
      "step": 12850
    },
    {
      "epoch": 4.057783391221976,
      "grad_norm": 4.011866092681885,
      "learning_rate": 1.0469440741000632e-05,
      "loss": 0.6297,
      "step": 12851
    },
    {
      "epoch": 4.0580991474581625,
      "grad_norm": 3.3985910415649414,
      "learning_rate": 1.0465932215283139e-05,
      "loss": 0.5623,
      "step": 12852
    },
    {
      "epoch": 4.058414903694348,
      "grad_norm": 3.322442054748535,
      "learning_rate": 1.0462423689565646e-05,
      "loss": 0.5656,
      "step": 12853
    },
    {
      "epoch": 4.058730659930534,
      "grad_norm": 2.9951367378234863,
      "learning_rate": 1.0458915163848151e-05,
      "loss": 0.5012,
      "step": 12854
    },
    {
      "epoch": 4.059046416166719,
      "grad_norm": 3.7227492332458496,
      "learning_rate": 1.0455406638130658e-05,
      "loss": 0.6017,
      "step": 12855
    },
    {
      "epoch": 4.0593621724029045,
      "grad_norm": 3.435910940170288,
      "learning_rate": 1.0451898112413165e-05,
      "loss": 0.5121,
      "step": 12856
    },
    {
      "epoch": 4.059677928639091,
      "grad_norm": 3.835968255996704,
      "learning_rate": 1.0448389586695671e-05,
      "loss": 0.6696,
      "step": 12857
    },
    {
      "epoch": 4.059993684875276,
      "grad_norm": 3.837050676345825,
      "learning_rate": 1.0444881060978178e-05,
      "loss": 0.6168,
      "step": 12858
    },
    {
      "epoch": 4.060309441111462,
      "grad_norm": 3.9661142826080322,
      "learning_rate": 1.0441372535260683e-05,
      "loss": 0.5597,
      "step": 12859
    },
    {
      "epoch": 4.0606251973476475,
      "grad_norm": 3.0696463584899902,
      "learning_rate": 1.043786400954319e-05,
      "loss": 0.5716,
      "step": 12860
    },
    {
      "epoch": 4.060940953583834,
      "grad_norm": 3.8747193813323975,
      "learning_rate": 1.0434355483825697e-05,
      "loss": 0.6515,
      "step": 12861
    },
    {
      "epoch": 4.061256709820019,
      "grad_norm": 4.539163589477539,
      "learning_rate": 1.0430846958108204e-05,
      "loss": 0.6192,
      "step": 12862
    },
    {
      "epoch": 4.061572466056204,
      "grad_norm": 4.048236846923828,
      "learning_rate": 1.0427338432390709e-05,
      "loss": 0.6729,
      "step": 12863
    },
    {
      "epoch": 4.06188822229239,
      "grad_norm": 3.8031699657440186,
      "learning_rate": 1.0423829906673216e-05,
      "loss": 0.5872,
      "step": 12864
    },
    {
      "epoch": 4.062203978528576,
      "grad_norm": 4.188300609588623,
      "learning_rate": 1.0420321380955724e-05,
      "loss": 0.6336,
      "step": 12865
    },
    {
      "epoch": 4.062519734764762,
      "grad_norm": 3.564072370529175,
      "learning_rate": 1.041681285523823e-05,
      "loss": 0.6844,
      "step": 12866
    },
    {
      "epoch": 4.062835491000947,
      "grad_norm": 3.9812119007110596,
      "learning_rate": 1.0413304329520736e-05,
      "loss": 0.6124,
      "step": 12867
    },
    {
      "epoch": 4.0631512472371325,
      "grad_norm": 4.057307243347168,
      "learning_rate": 1.0409795803803241e-05,
      "loss": 0.5533,
      "step": 12868
    },
    {
      "epoch": 4.063467003473319,
      "grad_norm": 3.669809341430664,
      "learning_rate": 1.040628727808575e-05,
      "loss": 0.5347,
      "step": 12869
    },
    {
      "epoch": 4.063782759709504,
      "grad_norm": 4.087433815002441,
      "learning_rate": 1.0402778752368255e-05,
      "loss": 0.608,
      "step": 12870
    },
    {
      "epoch": 4.06409851594569,
      "grad_norm": 3.3369035720825195,
      "learning_rate": 1.0399270226650762e-05,
      "loss": 0.5144,
      "step": 12871
    },
    {
      "epoch": 4.0644142721818755,
      "grad_norm": 3.729434013366699,
      "learning_rate": 1.0395761700933269e-05,
      "loss": 0.6089,
      "step": 12872
    },
    {
      "epoch": 4.064730028418062,
      "grad_norm": 4.464353561401367,
      "learning_rate": 1.0392253175215774e-05,
      "loss": 0.6164,
      "step": 12873
    },
    {
      "epoch": 4.065045784654247,
      "grad_norm": 4.8875412940979,
      "learning_rate": 1.0388744649498282e-05,
      "loss": 0.6089,
      "step": 12874
    },
    {
      "epoch": 4.065361540890432,
      "grad_norm": 3.4778544902801514,
      "learning_rate": 1.0385236123780787e-05,
      "loss": 0.5718,
      "step": 12875
    },
    {
      "epoch": 4.065677297126618,
      "grad_norm": 4.03305196762085,
      "learning_rate": 1.0381727598063294e-05,
      "loss": 0.4962,
      "step": 12876
    },
    {
      "epoch": 4.065993053362804,
      "grad_norm": 3.496553421020508,
      "learning_rate": 1.0378219072345801e-05,
      "loss": 0.6244,
      "step": 12877
    },
    {
      "epoch": 4.06630880959899,
      "grad_norm": 3.872631072998047,
      "learning_rate": 1.0374710546628308e-05,
      "loss": 0.5873,
      "step": 12878
    },
    {
      "epoch": 4.066624565835175,
      "grad_norm": 3.559168577194214,
      "learning_rate": 1.0371202020910815e-05,
      "loss": 0.6616,
      "step": 12879
    },
    {
      "epoch": 4.0669403220713605,
      "grad_norm": 4.205992221832275,
      "learning_rate": 1.036769349519332e-05,
      "loss": 0.6333,
      "step": 12880
    },
    {
      "epoch": 4.067256078307547,
      "grad_norm": 3.4708144664764404,
      "learning_rate": 1.0364184969475826e-05,
      "loss": 0.5775,
      "step": 12881
    },
    {
      "epoch": 4.067571834543732,
      "grad_norm": 3.513157367706299,
      "learning_rate": 1.0360676443758333e-05,
      "loss": 0.6622,
      "step": 12882
    },
    {
      "epoch": 4.067887590779918,
      "grad_norm": 3.34436297416687,
      "learning_rate": 1.035716791804084e-05,
      "loss": 0.5108,
      "step": 12883
    },
    {
      "epoch": 4.068203347016103,
      "grad_norm": 3.359483480453491,
      "learning_rate": 1.0353659392323347e-05,
      "loss": 0.4972,
      "step": 12884
    },
    {
      "epoch": 4.06851910325229,
      "grad_norm": 3.878629207611084,
      "learning_rate": 1.0350150866605852e-05,
      "loss": 0.5862,
      "step": 12885
    },
    {
      "epoch": 4.068834859488475,
      "grad_norm": 2.7553534507751465,
      "learning_rate": 1.0346642340888359e-05,
      "loss": 0.5765,
      "step": 12886
    },
    {
      "epoch": 4.06915061572466,
      "grad_norm": 3.8739476203918457,
      "learning_rate": 1.0343133815170866e-05,
      "loss": 0.605,
      "step": 12887
    },
    {
      "epoch": 4.069466371960846,
      "grad_norm": 4.0728254318237305,
      "learning_rate": 1.0339625289453372e-05,
      "loss": 0.6059,
      "step": 12888
    },
    {
      "epoch": 4.069782128197032,
      "grad_norm": 3.3264732360839844,
      "learning_rate": 1.033611676373588e-05,
      "loss": 0.5279,
      "step": 12889
    },
    {
      "epoch": 4.070097884433218,
      "grad_norm": 3.4087233543395996,
      "learning_rate": 1.0332608238018384e-05,
      "loss": 0.5984,
      "step": 12890
    },
    {
      "epoch": 4.070413640669403,
      "grad_norm": 3.630932331085205,
      "learning_rate": 1.0329099712300893e-05,
      "loss": 0.5671,
      "step": 12891
    },
    {
      "epoch": 4.070729396905589,
      "grad_norm": 3.7048306465148926,
      "learning_rate": 1.0325591186583398e-05,
      "loss": 0.6228,
      "step": 12892
    },
    {
      "epoch": 4.071045153141775,
      "grad_norm": 4.152629852294922,
      "learning_rate": 1.0322082660865905e-05,
      "loss": 0.636,
      "step": 12893
    },
    {
      "epoch": 4.07136090937796,
      "grad_norm": 3.5921554565429688,
      "learning_rate": 1.031857413514841e-05,
      "loss": 0.5017,
      "step": 12894
    },
    {
      "epoch": 4.071676665614146,
      "grad_norm": 3.76459002494812,
      "learning_rate": 1.0315065609430917e-05,
      "loss": 0.5912,
      "step": 12895
    },
    {
      "epoch": 4.071992421850331,
      "grad_norm": 3.3819806575775146,
      "learning_rate": 1.0311557083713425e-05,
      "loss": 0.5736,
      "step": 12896
    },
    {
      "epoch": 4.072308178086518,
      "grad_norm": 3.786701202392578,
      "learning_rate": 1.030804855799593e-05,
      "loss": 0.6035,
      "step": 12897
    },
    {
      "epoch": 4.072623934322703,
      "grad_norm": 3.644134283065796,
      "learning_rate": 1.0304540032278437e-05,
      "loss": 0.5959,
      "step": 12898
    },
    {
      "epoch": 4.072939690558888,
      "grad_norm": 3.059605121612549,
      "learning_rate": 1.0301031506560942e-05,
      "loss": 0.6058,
      "step": 12899
    },
    {
      "epoch": 4.073255446795074,
      "grad_norm": 3.8056650161743164,
      "learning_rate": 1.0297522980843451e-05,
      "loss": 0.5125,
      "step": 12900
    },
    {
      "epoch": 4.07357120303126,
      "grad_norm": 3.603909492492676,
      "learning_rate": 1.0294014455125956e-05,
      "loss": 0.5623,
      "step": 12901
    },
    {
      "epoch": 4.073886959267446,
      "grad_norm": 3.0317769050598145,
      "learning_rate": 1.0290505929408463e-05,
      "loss": 0.586,
      "step": 12902
    },
    {
      "epoch": 4.074202715503631,
      "grad_norm": 3.458388566970825,
      "learning_rate": 1.028699740369097e-05,
      "loss": 0.5998,
      "step": 12903
    },
    {
      "epoch": 4.074518471739817,
      "grad_norm": 4.353362083435059,
      "learning_rate": 1.0283488877973476e-05,
      "loss": 0.5453,
      "step": 12904
    },
    {
      "epoch": 4.074834227976003,
      "grad_norm": 3.4379894733428955,
      "learning_rate": 1.0279980352255983e-05,
      "loss": 0.5694,
      "step": 12905
    },
    {
      "epoch": 4.075149984212188,
      "grad_norm": 3.3759820461273193,
      "learning_rate": 1.0276471826538488e-05,
      "loss": 0.6477,
      "step": 12906
    },
    {
      "epoch": 4.075465740448374,
      "grad_norm": 4.012436866760254,
      "learning_rate": 1.0272963300820995e-05,
      "loss": 0.5919,
      "step": 12907
    },
    {
      "epoch": 4.075781496684559,
      "grad_norm": 3.7738592624664307,
      "learning_rate": 1.0269454775103502e-05,
      "loss": 0.6764,
      "step": 12908
    },
    {
      "epoch": 4.0760972529207455,
      "grad_norm": 3.7703287601470947,
      "learning_rate": 1.0265946249386009e-05,
      "loss": 0.5961,
      "step": 12909
    },
    {
      "epoch": 4.076413009156931,
      "grad_norm": 3.7859041690826416,
      "learning_rate": 1.0262437723668516e-05,
      "loss": 0.5917,
      "step": 12910
    },
    {
      "epoch": 4.076728765393116,
      "grad_norm": 3.464873790740967,
      "learning_rate": 1.025892919795102e-05,
      "loss": 0.5951,
      "step": 12911
    },
    {
      "epoch": 4.077044521629302,
      "grad_norm": 3.655517816543579,
      "learning_rate": 1.0255420672233528e-05,
      "loss": 0.5218,
      "step": 12912
    },
    {
      "epoch": 4.077360277865488,
      "grad_norm": 5.170120716094971,
      "learning_rate": 1.0251912146516034e-05,
      "loss": 0.6305,
      "step": 12913
    },
    {
      "epoch": 4.077676034101674,
      "grad_norm": 4.136005401611328,
      "learning_rate": 1.0248403620798541e-05,
      "loss": 0.5143,
      "step": 12914
    },
    {
      "epoch": 4.077991790337859,
      "grad_norm": 5.779212474822998,
      "learning_rate": 1.0244895095081048e-05,
      "loss": 0.635,
      "step": 12915
    },
    {
      "epoch": 4.078307546574045,
      "grad_norm": 3.523510694503784,
      "learning_rate": 1.0241386569363553e-05,
      "loss": 0.6417,
      "step": 12916
    },
    {
      "epoch": 4.0786233028102306,
      "grad_norm": 3.7380282878875732,
      "learning_rate": 1.023787804364606e-05,
      "loss": 0.5543,
      "step": 12917
    },
    {
      "epoch": 4.078939059046416,
      "grad_norm": 3.619640350341797,
      "learning_rate": 1.0234369517928567e-05,
      "loss": 0.5511,
      "step": 12918
    },
    {
      "epoch": 4.079254815282602,
      "grad_norm": 3.2790114879608154,
      "learning_rate": 1.0230860992211074e-05,
      "loss": 0.5663,
      "step": 12919
    },
    {
      "epoch": 4.079570571518787,
      "grad_norm": 3.7538530826568604,
      "learning_rate": 1.022735246649358e-05,
      "loss": 0.6771,
      "step": 12920
    },
    {
      "epoch": 4.0798863277549735,
      "grad_norm": 2.9966518878936768,
      "learning_rate": 1.0223843940776085e-05,
      "loss": 0.5058,
      "step": 12921
    },
    {
      "epoch": 4.080202083991159,
      "grad_norm": 3.269751787185669,
      "learning_rate": 1.0220335415058594e-05,
      "loss": 0.5297,
      "step": 12922
    },
    {
      "epoch": 4.080517840227344,
      "grad_norm": 4.018555641174316,
      "learning_rate": 1.0216826889341099e-05,
      "loss": 0.486,
      "step": 12923
    },
    {
      "epoch": 4.08083359646353,
      "grad_norm": 4.007140159606934,
      "learning_rate": 1.0213318363623606e-05,
      "loss": 0.5627,
      "step": 12924
    },
    {
      "epoch": 4.081149352699716,
      "grad_norm": 4.317376136779785,
      "learning_rate": 1.0209809837906113e-05,
      "loss": 0.5756,
      "step": 12925
    },
    {
      "epoch": 4.081465108935902,
      "grad_norm": 3.858318328857422,
      "learning_rate": 1.020630131218862e-05,
      "loss": 0.6242,
      "step": 12926
    },
    {
      "epoch": 4.081780865172087,
      "grad_norm": 3.707872152328491,
      "learning_rate": 1.0202792786471126e-05,
      "loss": 0.6414,
      "step": 12927
    },
    {
      "epoch": 4.082096621408273,
      "grad_norm": 3.5883703231811523,
      "learning_rate": 1.0199284260753631e-05,
      "loss": 0.6003,
      "step": 12928
    },
    {
      "epoch": 4.0824123776444585,
      "grad_norm": 4.101365566253662,
      "learning_rate": 1.0195775735036138e-05,
      "loss": 0.6326,
      "step": 12929
    },
    {
      "epoch": 4.082728133880644,
      "grad_norm": 2.6620845794677734,
      "learning_rate": 1.0192267209318643e-05,
      "loss": 0.5493,
      "step": 12930
    },
    {
      "epoch": 4.08304389011683,
      "grad_norm": 4.495128154754639,
      "learning_rate": 1.0188758683601152e-05,
      "loss": 0.619,
      "step": 12931
    },
    {
      "epoch": 4.083359646353015,
      "grad_norm": 4.443236827850342,
      "learning_rate": 1.0185250157883659e-05,
      "loss": 0.6222,
      "step": 12932
    },
    {
      "epoch": 4.0836754025892015,
      "grad_norm": 3.553365707397461,
      "learning_rate": 1.0181741632166164e-05,
      "loss": 0.6628,
      "step": 12933
    },
    {
      "epoch": 4.083991158825387,
      "grad_norm": 3.941387891769409,
      "learning_rate": 1.017823310644867e-05,
      "loss": 0.5348,
      "step": 12934
    },
    {
      "epoch": 4.084306915061572,
      "grad_norm": 4.088018894195557,
      "learning_rate": 1.0174724580731177e-05,
      "loss": 0.6891,
      "step": 12935
    },
    {
      "epoch": 4.084622671297758,
      "grad_norm": 4.5657124519348145,
      "learning_rate": 1.0171216055013684e-05,
      "loss": 0.6142,
      "step": 12936
    },
    {
      "epoch": 4.0849384275339435,
      "grad_norm": 3.3059301376342773,
      "learning_rate": 1.016770752929619e-05,
      "loss": 0.5544,
      "step": 12937
    },
    {
      "epoch": 4.08525418377013,
      "grad_norm": 3.508803367614746,
      "learning_rate": 1.0164199003578696e-05,
      "loss": 0.5273,
      "step": 12938
    },
    {
      "epoch": 4.085569940006315,
      "grad_norm": 3.962054491043091,
      "learning_rate": 1.0160690477861205e-05,
      "loss": 0.5838,
      "step": 12939
    },
    {
      "epoch": 4.085885696242501,
      "grad_norm": 3.8161041736602783,
      "learning_rate": 1.015718195214371e-05,
      "loss": 0.5629,
      "step": 12940
    },
    {
      "epoch": 4.0862014524786865,
      "grad_norm": 3.985208034515381,
      "learning_rate": 1.0153673426426217e-05,
      "loss": 0.5741,
      "step": 12941
    },
    {
      "epoch": 4.086517208714872,
      "grad_norm": 4.160126686096191,
      "learning_rate": 1.0150164900708722e-05,
      "loss": 0.5985,
      "step": 12942
    },
    {
      "epoch": 4.086832964951058,
      "grad_norm": 4.114832878112793,
      "learning_rate": 1.0146656374991229e-05,
      "loss": 0.7426,
      "step": 12943
    },
    {
      "epoch": 4.087148721187243,
      "grad_norm": 3.879436492919922,
      "learning_rate": 1.0143147849273735e-05,
      "loss": 0.5604,
      "step": 12944
    },
    {
      "epoch": 4.087464477423429,
      "grad_norm": 4.057650566101074,
      "learning_rate": 1.0139639323556242e-05,
      "loss": 0.5682,
      "step": 12945
    },
    {
      "epoch": 4.087780233659615,
      "grad_norm": 3.8411917686462402,
      "learning_rate": 1.0136130797838749e-05,
      "loss": 0.5715,
      "step": 12946
    },
    {
      "epoch": 4.0880959898958,
      "grad_norm": 4.330343723297119,
      "learning_rate": 1.0132622272121254e-05,
      "loss": 0.583,
      "step": 12947
    },
    {
      "epoch": 4.088411746131986,
      "grad_norm": 3.7931618690490723,
      "learning_rate": 1.0129113746403763e-05,
      "loss": 0.597,
      "step": 12948
    },
    {
      "epoch": 4.0887275023681715,
      "grad_norm": 3.4784791469573975,
      "learning_rate": 1.0125605220686268e-05,
      "loss": 0.5967,
      "step": 12949
    },
    {
      "epoch": 4.089043258604358,
      "grad_norm": 3.4613752365112305,
      "learning_rate": 1.0122096694968775e-05,
      "loss": 0.5026,
      "step": 12950
    },
    {
      "epoch": 4.089359014840543,
      "grad_norm": 4.429986953735352,
      "learning_rate": 1.0118588169251281e-05,
      "loss": 0.6148,
      "step": 12951
    },
    {
      "epoch": 4.089674771076729,
      "grad_norm": 4.763156890869141,
      "learning_rate": 1.0115079643533786e-05,
      "loss": 0.6072,
      "step": 12952
    },
    {
      "epoch": 4.0899905273129145,
      "grad_norm": 3.8502554893493652,
      "learning_rate": 1.0111571117816295e-05,
      "loss": 0.5591,
      "step": 12953
    },
    {
      "epoch": 4.0903062835491,
      "grad_norm": 3.918788194656372,
      "learning_rate": 1.01080625920988e-05,
      "loss": 0.5038,
      "step": 12954
    },
    {
      "epoch": 4.090622039785286,
      "grad_norm": 4.323645114898682,
      "learning_rate": 1.0104554066381307e-05,
      "loss": 0.5936,
      "step": 12955
    },
    {
      "epoch": 4.090937796021471,
      "grad_norm": 3.4729788303375244,
      "learning_rate": 1.0101045540663814e-05,
      "loss": 0.5206,
      "step": 12956
    },
    {
      "epoch": 4.091253552257657,
      "grad_norm": 4.053699493408203,
      "learning_rate": 1.009753701494632e-05,
      "loss": 0.5864,
      "step": 12957
    },
    {
      "epoch": 4.091569308493843,
      "grad_norm": 3.7024178504943848,
      "learning_rate": 1.0094028489228827e-05,
      "loss": 0.547,
      "step": 12958
    },
    {
      "epoch": 4.091885064730028,
      "grad_norm": 3.5911099910736084,
      "learning_rate": 1.0090519963511332e-05,
      "loss": 0.5068,
      "step": 12959
    },
    {
      "epoch": 4.092200820966214,
      "grad_norm": 3.1703691482543945,
      "learning_rate": 1.008701143779384e-05,
      "loss": 0.62,
      "step": 12960
    },
    {
      "epoch": 4.0925165772023995,
      "grad_norm": 4.172727584838867,
      "learning_rate": 1.0083502912076346e-05,
      "loss": 0.5826,
      "step": 12961
    },
    {
      "epoch": 4.092832333438586,
      "grad_norm": 4.006683349609375,
      "learning_rate": 1.0079994386358853e-05,
      "loss": 0.6134,
      "step": 12962
    },
    {
      "epoch": 4.093148089674771,
      "grad_norm": 3.7655510902404785,
      "learning_rate": 1.007648586064136e-05,
      "loss": 0.6639,
      "step": 12963
    },
    {
      "epoch": 4.093463845910957,
      "grad_norm": 4.149551868438721,
      "learning_rate": 1.0072977334923865e-05,
      "loss": 0.558,
      "step": 12964
    },
    {
      "epoch": 4.093779602147142,
      "grad_norm": 3.3447482585906982,
      "learning_rate": 1.0069468809206372e-05,
      "loss": 0.5199,
      "step": 12965
    },
    {
      "epoch": 4.094095358383328,
      "grad_norm": 3.6987693309783936,
      "learning_rate": 1.0065960283488878e-05,
      "loss": 0.5981,
      "step": 12966
    },
    {
      "epoch": 4.094411114619514,
      "grad_norm": 3.655163049697876,
      "learning_rate": 1.0062451757771385e-05,
      "loss": 0.5588,
      "step": 12967
    },
    {
      "epoch": 4.094726870855699,
      "grad_norm": 3.4375407695770264,
      "learning_rate": 1.005894323205389e-05,
      "loss": 0.5263,
      "step": 12968
    },
    {
      "epoch": 4.095042627091885,
      "grad_norm": 3.62575626373291,
      "learning_rate": 1.0055434706336397e-05,
      "loss": 0.6263,
      "step": 12969
    },
    {
      "epoch": 4.095358383328071,
      "grad_norm": 5.023072242736816,
      "learning_rate": 1.0051926180618906e-05,
      "loss": 0.5662,
      "step": 12970
    },
    {
      "epoch": 4.095674139564256,
      "grad_norm": 4.126835346221924,
      "learning_rate": 1.0048417654901411e-05,
      "loss": 0.6358,
      "step": 12971
    },
    {
      "epoch": 4.095989895800442,
      "grad_norm": 5.055971145629883,
      "learning_rate": 1.0044909129183918e-05,
      "loss": 0.6084,
      "step": 12972
    },
    {
      "epoch": 4.096305652036627,
      "grad_norm": 3.335303544998169,
      "learning_rate": 1.0041400603466423e-05,
      "loss": 0.5255,
      "step": 12973
    },
    {
      "epoch": 4.096621408272814,
      "grad_norm": 3.3699069023132324,
      "learning_rate": 1.0037892077748931e-05,
      "loss": 0.5737,
      "step": 12974
    },
    {
      "epoch": 4.096937164508999,
      "grad_norm": 3.436927080154419,
      "learning_rate": 1.0034383552031436e-05,
      "loss": 0.5274,
      "step": 12975
    },
    {
      "epoch": 4.097252920745185,
      "grad_norm": 3.90643572807312,
      "learning_rate": 1.0030875026313943e-05,
      "loss": 0.583,
      "step": 12976
    },
    {
      "epoch": 4.09756867698137,
      "grad_norm": 4.44471549987793,
      "learning_rate": 1.002736650059645e-05,
      "loss": 0.6478,
      "step": 12977
    },
    {
      "epoch": 4.097884433217556,
      "grad_norm": 4.295309066772461,
      "learning_rate": 1.0023857974878955e-05,
      "loss": 0.6612,
      "step": 12978
    },
    {
      "epoch": 4.098200189453742,
      "grad_norm": 4.060950756072998,
      "learning_rate": 1.0020349449161464e-05,
      "loss": 0.6267,
      "step": 12979
    },
    {
      "epoch": 4.098515945689927,
      "grad_norm": 3.062206745147705,
      "learning_rate": 1.0016840923443969e-05,
      "loss": 0.5437,
      "step": 12980
    },
    {
      "epoch": 4.098831701926113,
      "grad_norm": 4.055416584014893,
      "learning_rate": 1.0013332397726476e-05,
      "loss": 0.5295,
      "step": 12981
    },
    {
      "epoch": 4.099147458162299,
      "grad_norm": 4.054298400878906,
      "learning_rate": 1.0009823872008982e-05,
      "loss": 0.5662,
      "step": 12982
    },
    {
      "epoch": 4.099463214398484,
      "grad_norm": 3.465101480484009,
      "learning_rate": 1.000631534629149e-05,
      "loss": 0.5239,
      "step": 12983
    },
    {
      "epoch": 4.09977897063467,
      "grad_norm": 3.4558937549591064,
      "learning_rate": 1.0002806820573996e-05,
      "loss": 0.5914,
      "step": 12984
    },
    {
      "epoch": 4.100094726870855,
      "grad_norm": 3.9110591411590576,
      "learning_rate": 9.999298294856501e-06,
      "loss": 0.5894,
      "step": 12985
    },
    {
      "epoch": 4.100410483107042,
      "grad_norm": 3.6499292850494385,
      "learning_rate": 9.995789769139008e-06,
      "loss": 0.554,
      "step": 12986
    },
    {
      "epoch": 4.100726239343227,
      "grad_norm": 5.198665618896484,
      "learning_rate": 9.992281243421515e-06,
      "loss": 0.6132,
      "step": 12987
    },
    {
      "epoch": 4.101041995579413,
      "grad_norm": 3.4791367053985596,
      "learning_rate": 9.988772717704022e-06,
      "loss": 0.5205,
      "step": 12988
    },
    {
      "epoch": 4.101357751815598,
      "grad_norm": 3.9647347927093506,
      "learning_rate": 9.985264191986528e-06,
      "loss": 0.5486,
      "step": 12989
    },
    {
      "epoch": 4.101673508051784,
      "grad_norm": 4.185471534729004,
      "learning_rate": 9.981755666269034e-06,
      "loss": 0.5645,
      "step": 12990
    },
    {
      "epoch": 4.10198926428797,
      "grad_norm": 4.3772664070129395,
      "learning_rate": 9.97824714055154e-06,
      "loss": 0.7261,
      "step": 12991
    },
    {
      "epoch": 4.102305020524155,
      "grad_norm": 4.004677772521973,
      "learning_rate": 9.974738614834047e-06,
      "loss": 0.6386,
      "step": 12992
    },
    {
      "epoch": 4.102620776760341,
      "grad_norm": 3.840394973754883,
      "learning_rate": 9.971230089116554e-06,
      "loss": 0.5648,
      "step": 12993
    },
    {
      "epoch": 4.102936532996527,
      "grad_norm": 4.3086957931518555,
      "learning_rate": 9.96772156339906e-06,
      "loss": 0.5967,
      "step": 12994
    },
    {
      "epoch": 4.103252289232712,
      "grad_norm": 4.348057270050049,
      "learning_rate": 9.964213037681566e-06,
      "loss": 0.728,
      "step": 12995
    },
    {
      "epoch": 4.103568045468898,
      "grad_norm": 3.9251294136047363,
      "learning_rate": 9.960704511964074e-06,
      "loss": 0.57,
      "step": 12996
    },
    {
      "epoch": 4.103883801705083,
      "grad_norm": 4.047271251678467,
      "learning_rate": 9.95719598624658e-06,
      "loss": 0.6206,
      "step": 12997
    },
    {
      "epoch": 4.10419955794127,
      "grad_norm": 3.515030860900879,
      "learning_rate": 9.953687460529086e-06,
      "loss": 0.625,
      "step": 12998
    },
    {
      "epoch": 4.104515314177455,
      "grad_norm": 3.626920223236084,
      "learning_rate": 9.950178934811591e-06,
      "loss": 0.5484,
      "step": 12999
    },
    {
      "epoch": 4.104831070413641,
      "grad_norm": 3.472233533859253,
      "learning_rate": 9.946670409094098e-06,
      "loss": 0.552,
      "step": 13000
    },
    {
      "epoch": 4.104831070413641,
      "eval_loss": 0.8345271944999695,
      "eval_runtime": 4.5978,
      "eval_samples_per_second": 541.561,
      "eval_steps_per_second": 4.567,
      "step": 13000
    },
    {
      "epoch": 4.105146826649826,
      "grad_norm": 4.434135437011719,
      "learning_rate": 9.943161883376607e-06,
      "loss": 0.5451,
      "step": 13001
    },
    {
      "epoch": 4.105462582886012,
      "grad_norm": 3.8763339519500732,
      "learning_rate": 9.939653357659112e-06,
      "loss": 0.5718,
      "step": 13002
    },
    {
      "epoch": 4.105778339122198,
      "grad_norm": 4.784031867980957,
      "learning_rate": 9.936144831941619e-06,
      "loss": 0.6028,
      "step": 13003
    },
    {
      "epoch": 4.106094095358383,
      "grad_norm": 3.3705034255981445,
      "learning_rate": 9.932636306224124e-06,
      "loss": 0.5522,
      "step": 13004
    },
    {
      "epoch": 4.106409851594569,
      "grad_norm": 3.4715499877929688,
      "learning_rate": 9.929127780506632e-06,
      "loss": 0.5605,
      "step": 13005
    },
    {
      "epoch": 4.106725607830755,
      "grad_norm": 3.4505069255828857,
      "learning_rate": 9.925619254789137e-06,
      "loss": 0.5816,
      "step": 13006
    },
    {
      "epoch": 4.107041364066941,
      "grad_norm": 4.189521312713623,
      "learning_rate": 9.922110729071644e-06,
      "loss": 0.6555,
      "step": 13007
    },
    {
      "epoch": 4.107357120303126,
      "grad_norm": 3.6706230640411377,
      "learning_rate": 9.918602203354151e-06,
      "loss": 0.5288,
      "step": 13008
    },
    {
      "epoch": 4.107672876539311,
      "grad_norm": 4.61895751953125,
      "learning_rate": 9.915093677636658e-06,
      "loss": 0.6174,
      "step": 13009
    },
    {
      "epoch": 4.1079886327754975,
      "grad_norm": 3.5695931911468506,
      "learning_rate": 9.911585151919165e-06,
      "loss": 0.5595,
      "step": 13010
    },
    {
      "epoch": 4.108304389011683,
      "grad_norm": 3.39674973487854,
      "learning_rate": 9.90807662620167e-06,
      "loss": 0.6432,
      "step": 13011
    },
    {
      "epoch": 4.108620145247869,
      "grad_norm": 3.3579187393188477,
      "learning_rate": 9.904568100484177e-06,
      "loss": 0.5854,
      "step": 13012
    },
    {
      "epoch": 4.108935901484054,
      "grad_norm": 5.92067813873291,
      "learning_rate": 9.901059574766683e-06,
      "loss": 0.6909,
      "step": 13013
    },
    {
      "epoch": 4.10925165772024,
      "grad_norm": 3.988516330718994,
      "learning_rate": 9.89755104904919e-06,
      "loss": 0.6262,
      "step": 13014
    },
    {
      "epoch": 4.109567413956426,
      "grad_norm": 3.5225515365600586,
      "learning_rate": 9.894042523331697e-06,
      "loss": 0.5051,
      "step": 13015
    },
    {
      "epoch": 4.109883170192611,
      "grad_norm": 3.740993022918701,
      "learning_rate": 9.890533997614202e-06,
      "loss": 0.6282,
      "step": 13016
    },
    {
      "epoch": 4.110198926428797,
      "grad_norm": 3.069096803665161,
      "learning_rate": 9.887025471896709e-06,
      "loss": 0.5851,
      "step": 13017
    },
    {
      "epoch": 4.1105146826649825,
      "grad_norm": 4.7527570724487305,
      "learning_rate": 9.883516946179216e-06,
      "loss": 0.5808,
      "step": 13018
    },
    {
      "epoch": 4.110830438901169,
      "grad_norm": 3.504377841949463,
      "learning_rate": 9.880008420461723e-06,
      "loss": 0.6742,
      "step": 13019
    },
    {
      "epoch": 4.111146195137354,
      "grad_norm": 3.7932300567626953,
      "learning_rate": 9.87649989474423e-06,
      "loss": 0.5801,
      "step": 13020
    },
    {
      "epoch": 4.111461951373539,
      "grad_norm": 3.6429171562194824,
      "learning_rate": 9.872991369026735e-06,
      "loss": 0.5377,
      "step": 13021
    },
    {
      "epoch": 4.1117777076097255,
      "grad_norm": 4.2909040451049805,
      "learning_rate": 9.869482843309243e-06,
      "loss": 0.7757,
      "step": 13022
    },
    {
      "epoch": 4.112093463845911,
      "grad_norm": 3.5942630767822266,
      "learning_rate": 9.865974317591748e-06,
      "loss": 0.5592,
      "step": 13023
    },
    {
      "epoch": 4.112409220082097,
      "grad_norm": 2.8002536296844482,
      "learning_rate": 9.862465791874255e-06,
      "loss": 0.5404,
      "step": 13024
    },
    {
      "epoch": 4.112724976318282,
      "grad_norm": 4.778739929199219,
      "learning_rate": 9.858957266156762e-06,
      "loss": 0.593,
      "step": 13025
    },
    {
      "epoch": 4.113040732554468,
      "grad_norm": 3.928291082382202,
      "learning_rate": 9.855448740439267e-06,
      "loss": 0.6046,
      "step": 13026
    },
    {
      "epoch": 4.113356488790654,
      "grad_norm": 4.398281097412109,
      "learning_rate": 9.851940214721775e-06,
      "loss": 0.7443,
      "step": 13027
    },
    {
      "epoch": 4.113672245026839,
      "grad_norm": 4.364375114440918,
      "learning_rate": 9.84843168900428e-06,
      "loss": 0.6035,
      "step": 13028
    },
    {
      "epoch": 4.113988001263025,
      "grad_norm": 3.4041435718536377,
      "learning_rate": 9.844923163286787e-06,
      "loss": 0.5224,
      "step": 13029
    },
    {
      "epoch": 4.1143037574992105,
      "grad_norm": 3.6316049098968506,
      "learning_rate": 9.841414637569294e-06,
      "loss": 0.6155,
      "step": 13030
    },
    {
      "epoch": 4.114619513735397,
      "grad_norm": 4.116806983947754,
      "learning_rate": 9.837906111851801e-06,
      "loss": 0.6538,
      "step": 13031
    },
    {
      "epoch": 4.114935269971582,
      "grad_norm": 3.190012216567993,
      "learning_rate": 9.834397586134308e-06,
      "loss": 0.6139,
      "step": 13032
    },
    {
      "epoch": 4.115251026207767,
      "grad_norm": 3.8511719703674316,
      "learning_rate": 9.830889060416813e-06,
      "loss": 0.5581,
      "step": 13033
    },
    {
      "epoch": 4.1155667824439535,
      "grad_norm": 4.229973793029785,
      "learning_rate": 9.82738053469932e-06,
      "loss": 0.6498,
      "step": 13034
    },
    {
      "epoch": 4.115882538680139,
      "grad_norm": 3.5255331993103027,
      "learning_rate": 9.823872008981825e-06,
      "loss": 0.6158,
      "step": 13035
    },
    {
      "epoch": 4.116198294916325,
      "grad_norm": 3.5286529064178467,
      "learning_rate": 9.820363483264333e-06,
      "loss": 0.6146,
      "step": 13036
    },
    {
      "epoch": 4.11651405115251,
      "grad_norm": 3.7035114765167236,
      "learning_rate": 9.81685495754684e-06,
      "loss": 0.6237,
      "step": 13037
    },
    {
      "epoch": 4.1168298073886955,
      "grad_norm": 3.792667865753174,
      "learning_rate": 9.813346431829345e-06,
      "loss": 0.6435,
      "step": 13038
    },
    {
      "epoch": 4.117145563624882,
      "grad_norm": 3.4394595623016357,
      "learning_rate": 9.809837906111852e-06,
      "loss": 0.4986,
      "step": 13039
    },
    {
      "epoch": 4.117461319861067,
      "grad_norm": 3.361748456954956,
      "learning_rate": 9.806329380394359e-06,
      "loss": 0.523,
      "step": 13040
    },
    {
      "epoch": 4.117777076097253,
      "grad_norm": 3.616441249847412,
      "learning_rate": 9.802820854676866e-06,
      "loss": 0.5846,
      "step": 13041
    },
    {
      "epoch": 4.1180928323334385,
      "grad_norm": 3.438509941101074,
      "learning_rate": 9.799312328959371e-06,
      "loss": 0.5369,
      "step": 13042
    },
    {
      "epoch": 4.118408588569625,
      "grad_norm": 3.5774929523468018,
      "learning_rate": 9.795803803241878e-06,
      "loss": 0.6277,
      "step": 13043
    },
    {
      "epoch": 4.11872434480581,
      "grad_norm": 3.436385154724121,
      "learning_rate": 9.792295277524386e-06,
      "loss": 0.5738,
      "step": 13044
    },
    {
      "epoch": 4.119040101041995,
      "grad_norm": 3.6479625701904297,
      "learning_rate": 9.788786751806891e-06,
      "loss": 0.5742,
      "step": 13045
    },
    {
      "epoch": 4.119355857278181,
      "grad_norm": 3.367802619934082,
      "learning_rate": 9.785278226089398e-06,
      "loss": 0.6044,
      "step": 13046
    },
    {
      "epoch": 4.119671613514367,
      "grad_norm": 3.9155068397521973,
      "learning_rate": 9.781769700371903e-06,
      "loss": 0.5298,
      "step": 13047
    },
    {
      "epoch": 4.119987369750553,
      "grad_norm": 4.084733486175537,
      "learning_rate": 9.77826117465441e-06,
      "loss": 0.557,
      "step": 13048
    },
    {
      "epoch": 4.120303125986738,
      "grad_norm": 3.5657355785369873,
      "learning_rate": 9.774752648936917e-06,
      "loss": 0.5177,
      "step": 13049
    },
    {
      "epoch": 4.1206188822229235,
      "grad_norm": 4.112415790557861,
      "learning_rate": 9.771244123219424e-06,
      "loss": 0.65,
      "step": 13050
    },
    {
      "epoch": 4.12093463845911,
      "grad_norm": 3.760838031768799,
      "learning_rate": 9.76773559750193e-06,
      "loss": 0.673,
      "step": 13051
    },
    {
      "epoch": 4.121250394695295,
      "grad_norm": 3.2642688751220703,
      "learning_rate": 9.764227071784436e-06,
      "loss": 0.6545,
      "step": 13052
    },
    {
      "epoch": 4.121566150931481,
      "grad_norm": 4.229506015777588,
      "learning_rate": 9.760718546066944e-06,
      "loss": 0.5598,
      "step": 13053
    },
    {
      "epoch": 4.1218819071676664,
      "grad_norm": 3.6929562091827393,
      "learning_rate": 9.75721002034945e-06,
      "loss": 0.5556,
      "step": 13054
    },
    {
      "epoch": 4.122197663403853,
      "grad_norm": 3.706700086593628,
      "learning_rate": 9.753701494631956e-06,
      "loss": 0.5953,
      "step": 13055
    },
    {
      "epoch": 4.122513419640038,
      "grad_norm": 3.87666392326355,
      "learning_rate": 9.750192968914463e-06,
      "loss": 0.6058,
      "step": 13056
    },
    {
      "epoch": 4.122829175876223,
      "grad_norm": 3.8191215991973877,
      "learning_rate": 9.74668444319697e-06,
      "loss": 0.5056,
      "step": 13057
    },
    {
      "epoch": 4.123144932112409,
      "grad_norm": 4.348040580749512,
      "learning_rate": 9.743175917479476e-06,
      "loss": 0.5971,
      "step": 13058
    },
    {
      "epoch": 4.123460688348595,
      "grad_norm": 3.597182035446167,
      "learning_rate": 9.739667391761982e-06,
      "loss": 0.6263,
      "step": 13059
    },
    {
      "epoch": 4.123776444584781,
      "grad_norm": 4.054574489593506,
      "learning_rate": 9.736158866044488e-06,
      "loss": 0.5991,
      "step": 13060
    },
    {
      "epoch": 4.124092200820966,
      "grad_norm": 3.8878209590911865,
      "learning_rate": 9.732650340326995e-06,
      "loss": 0.6553,
      "step": 13061
    },
    {
      "epoch": 4.1244079570571515,
      "grad_norm": 4.8367767333984375,
      "learning_rate": 9.729141814609502e-06,
      "loss": 0.665,
      "step": 13062
    },
    {
      "epoch": 4.124723713293338,
      "grad_norm": 3.3343117237091064,
      "learning_rate": 9.725633288892009e-06,
      "loss": 0.545,
      "step": 13063
    },
    {
      "epoch": 4.125039469529523,
      "grad_norm": 4.328374862670898,
      "learning_rate": 9.722124763174514e-06,
      "loss": 0.685,
      "step": 13064
    },
    {
      "epoch": 4.125355225765709,
      "grad_norm": 3.5883820056915283,
      "learning_rate": 9.71861623745702e-06,
      "loss": 0.5835,
      "step": 13065
    },
    {
      "epoch": 4.125670982001894,
      "grad_norm": 3.527397394180298,
      "learning_rate": 9.715107711739528e-06,
      "loss": 0.6062,
      "step": 13066
    },
    {
      "epoch": 4.125986738238081,
      "grad_norm": 3.3799400329589844,
      "learning_rate": 9.711599186022034e-06,
      "loss": 0.6502,
      "step": 13067
    },
    {
      "epoch": 4.126302494474266,
      "grad_norm": 3.409735918045044,
      "learning_rate": 9.708090660304541e-06,
      "loss": 0.6129,
      "step": 13068
    },
    {
      "epoch": 4.126618250710451,
      "grad_norm": 4.308603763580322,
      "learning_rate": 9.704582134587046e-06,
      "loss": 0.5832,
      "step": 13069
    },
    {
      "epoch": 4.126934006946637,
      "grad_norm": 4.968552112579346,
      "learning_rate": 9.701073608869553e-06,
      "loss": 0.633,
      "step": 13070
    },
    {
      "epoch": 4.127249763182823,
      "grad_norm": 3.6622097492218018,
      "learning_rate": 9.69756508315206e-06,
      "loss": 0.5996,
      "step": 13071
    },
    {
      "epoch": 4.127565519419009,
      "grad_norm": 4.0067949295043945,
      "learning_rate": 9.694056557434567e-06,
      "loss": 0.5469,
      "step": 13072
    },
    {
      "epoch": 4.127881275655194,
      "grad_norm": 3.5334086418151855,
      "learning_rate": 9.690548031717072e-06,
      "loss": 0.5527,
      "step": 13073
    },
    {
      "epoch": 4.128197031891379,
      "grad_norm": 4.131180286407471,
      "learning_rate": 9.687039505999579e-06,
      "loss": 0.6947,
      "step": 13074
    },
    {
      "epoch": 4.128512788127566,
      "grad_norm": 3.779233694076538,
      "learning_rate": 9.683530980282087e-06,
      "loss": 0.567,
      "step": 13075
    },
    {
      "epoch": 4.128828544363751,
      "grad_norm": 3.8281562328338623,
      "learning_rate": 9.680022454564592e-06,
      "loss": 0.5467,
      "step": 13076
    },
    {
      "epoch": 4.129144300599937,
      "grad_norm": 3.8224780559539795,
      "learning_rate": 9.6765139288471e-06,
      "loss": 0.5367,
      "step": 13077
    },
    {
      "epoch": 4.129460056836122,
      "grad_norm": 4.02067756652832,
      "learning_rate": 9.673005403129604e-06,
      "loss": 0.5332,
      "step": 13078
    },
    {
      "epoch": 4.129775813072309,
      "grad_norm": 3.7920806407928467,
      "learning_rate": 9.669496877412113e-06,
      "loss": 0.6301,
      "step": 13079
    },
    {
      "epoch": 4.130091569308494,
      "grad_norm": 3.3705873489379883,
      "learning_rate": 9.665988351694618e-06,
      "loss": 0.5341,
      "step": 13080
    },
    {
      "epoch": 4.130407325544679,
      "grad_norm": 4.4546027183532715,
      "learning_rate": 9.662479825977125e-06,
      "loss": 0.5365,
      "step": 13081
    },
    {
      "epoch": 4.130723081780865,
      "grad_norm": 3.5848774909973145,
      "learning_rate": 9.658971300259632e-06,
      "loss": 0.5707,
      "step": 13082
    },
    {
      "epoch": 4.131038838017051,
      "grad_norm": 4.587492942810059,
      "learning_rate": 9.655462774542137e-06,
      "loss": 0.6556,
      "step": 13083
    },
    {
      "epoch": 4.131354594253237,
      "grad_norm": 3.552299737930298,
      "learning_rate": 9.651954248824645e-06,
      "loss": 0.5588,
      "step": 13084
    },
    {
      "epoch": 4.131670350489422,
      "grad_norm": 3.1284356117248535,
      "learning_rate": 9.64844572310715e-06,
      "loss": 0.5764,
      "step": 13085
    },
    {
      "epoch": 4.131986106725607,
      "grad_norm": 2.990338087081909,
      "learning_rate": 9.644937197389657e-06,
      "loss": 0.537,
      "step": 13086
    },
    {
      "epoch": 4.132301862961794,
      "grad_norm": 3.4774210453033447,
      "learning_rate": 9.641428671672164e-06,
      "loss": 0.6098,
      "step": 13087
    },
    {
      "epoch": 4.132617619197979,
      "grad_norm": 3.324706792831421,
      "learning_rate": 9.63792014595467e-06,
      "loss": 0.4688,
      "step": 13088
    },
    {
      "epoch": 4.132933375434165,
      "grad_norm": 4.349649429321289,
      "learning_rate": 9.634411620237178e-06,
      "loss": 0.5787,
      "step": 13089
    },
    {
      "epoch": 4.13324913167035,
      "grad_norm": 3.4564478397369385,
      "learning_rate": 9.630903094519683e-06,
      "loss": 0.5946,
      "step": 13090
    },
    {
      "epoch": 4.1335648879065365,
      "grad_norm": 4.074779033660889,
      "learning_rate": 9.62739456880219e-06,
      "loss": 0.5499,
      "step": 13091
    },
    {
      "epoch": 4.133880644142722,
      "grad_norm": 5.121905326843262,
      "learning_rate": 9.623886043084696e-06,
      "loss": 0.5779,
      "step": 13092
    },
    {
      "epoch": 4.134196400378907,
      "grad_norm": 4.036141872406006,
      "learning_rate": 9.620377517367203e-06,
      "loss": 0.5529,
      "step": 13093
    },
    {
      "epoch": 4.134512156615093,
      "grad_norm": 3.467603921890259,
      "learning_rate": 9.61686899164971e-06,
      "loss": 0.6314,
      "step": 13094
    },
    {
      "epoch": 4.134827912851279,
      "grad_norm": 3.902013063430786,
      "learning_rate": 9.613360465932215e-06,
      "loss": 0.6458,
      "step": 13095
    },
    {
      "epoch": 4.135143669087465,
      "grad_norm": 4.133418083190918,
      "learning_rate": 9.609851940214722e-06,
      "loss": 0.6774,
      "step": 13096
    },
    {
      "epoch": 4.13545942532365,
      "grad_norm": 3.645681619644165,
      "learning_rate": 9.606343414497229e-06,
      "loss": 0.617,
      "step": 13097
    },
    {
      "epoch": 4.135775181559836,
      "grad_norm": 3.989633798599243,
      "learning_rate": 9.602834888779735e-06,
      "loss": 0.5546,
      "step": 13098
    },
    {
      "epoch": 4.1360909377960215,
      "grad_norm": 4.226532459259033,
      "learning_rate": 9.599326363062242e-06,
      "loss": 0.6339,
      "step": 13099
    },
    {
      "epoch": 4.136406694032207,
      "grad_norm": 4.11239767074585,
      "learning_rate": 9.595817837344747e-06,
      "loss": 0.8348,
      "step": 13100
    },
    {
      "epoch": 4.136722450268393,
      "grad_norm": 4.218986988067627,
      "learning_rate": 9.592309311627256e-06,
      "loss": 0.5355,
      "step": 13101
    },
    {
      "epoch": 4.137038206504578,
      "grad_norm": 3.60444974899292,
      "learning_rate": 9.588800785909761e-06,
      "loss": 0.5832,
      "step": 13102
    },
    {
      "epoch": 4.1373539627407645,
      "grad_norm": 3.4290831089019775,
      "learning_rate": 9.585292260192268e-06,
      "loss": 0.6612,
      "step": 13103
    },
    {
      "epoch": 4.13766971897695,
      "grad_norm": 4.141811370849609,
      "learning_rate": 9.581783734474773e-06,
      "loss": 0.6232,
      "step": 13104
    },
    {
      "epoch": 4.137985475213135,
      "grad_norm": 3.595335006713867,
      "learning_rate": 9.57827520875728e-06,
      "loss": 0.6569,
      "step": 13105
    },
    {
      "epoch": 4.138301231449321,
      "grad_norm": 4.218543529510498,
      "learning_rate": 9.574766683039788e-06,
      "loss": 0.6851,
      "step": 13106
    },
    {
      "epoch": 4.138616987685507,
      "grad_norm": 3.4940009117126465,
      "learning_rate": 9.571258157322293e-06,
      "loss": 0.6294,
      "step": 13107
    },
    {
      "epoch": 4.138932743921693,
      "grad_norm": 3.802295684814453,
      "learning_rate": 9.5677496316048e-06,
      "loss": 0.5446,
      "step": 13108
    },
    {
      "epoch": 4.139248500157878,
      "grad_norm": 4.524115085601807,
      "learning_rate": 9.564241105887305e-06,
      "loss": 0.5506,
      "step": 13109
    },
    {
      "epoch": 4.139564256394063,
      "grad_norm": 3.8529586791992188,
      "learning_rate": 9.560732580169814e-06,
      "loss": 0.6456,
      "step": 13110
    },
    {
      "epoch": 4.1398800126302495,
      "grad_norm": 4.1948676109313965,
      "learning_rate": 9.557224054452319e-06,
      "loss": 0.5971,
      "step": 13111
    },
    {
      "epoch": 4.140195768866435,
      "grad_norm": 4.015960216522217,
      "learning_rate": 9.553715528734826e-06,
      "loss": 0.6122,
      "step": 13112
    },
    {
      "epoch": 4.140511525102621,
      "grad_norm": 4.034053325653076,
      "learning_rate": 9.550207003017333e-06,
      "loss": 0.5699,
      "step": 13113
    },
    {
      "epoch": 4.140827281338806,
      "grad_norm": 4.561849117279053,
      "learning_rate": 9.54669847729984e-06,
      "loss": 0.6164,
      "step": 13114
    },
    {
      "epoch": 4.1411430375749925,
      "grad_norm": 4.508412837982178,
      "learning_rate": 9.543189951582346e-06,
      "loss": 0.6163,
      "step": 13115
    },
    {
      "epoch": 4.141458793811178,
      "grad_norm": 4.494898796081543,
      "learning_rate": 9.539681425864851e-06,
      "loss": 0.5871,
      "step": 13116
    },
    {
      "epoch": 4.141774550047363,
      "grad_norm": 3.761096715927124,
      "learning_rate": 9.536172900147358e-06,
      "loss": 0.5694,
      "step": 13117
    },
    {
      "epoch": 4.142090306283549,
      "grad_norm": 3.5469303131103516,
      "learning_rate": 9.532664374429865e-06,
      "loss": 0.5975,
      "step": 13118
    },
    {
      "epoch": 4.1424060625197345,
      "grad_norm": 4.435953617095947,
      "learning_rate": 9.529155848712372e-06,
      "loss": 0.6207,
      "step": 13119
    },
    {
      "epoch": 4.142721818755921,
      "grad_norm": 4.472350120544434,
      "learning_rate": 9.525647322994879e-06,
      "loss": 0.6434,
      "step": 13120
    },
    {
      "epoch": 4.143037574992106,
      "grad_norm": 3.704484224319458,
      "learning_rate": 9.522138797277384e-06,
      "loss": 0.632,
      "step": 13121
    },
    {
      "epoch": 4.143353331228292,
      "grad_norm": 3.3620331287384033,
      "learning_rate": 9.51863027155989e-06,
      "loss": 0.5663,
      "step": 13122
    },
    {
      "epoch": 4.1436690874644775,
      "grad_norm": 4.7181572914123535,
      "learning_rate": 9.515121745842397e-06,
      "loss": 0.5291,
      "step": 13123
    },
    {
      "epoch": 4.143984843700663,
      "grad_norm": 3.709859848022461,
      "learning_rate": 9.511613220124904e-06,
      "loss": 0.6412,
      "step": 13124
    },
    {
      "epoch": 4.144300599936849,
      "grad_norm": 3.6794140338897705,
      "learning_rate": 9.508104694407411e-06,
      "loss": 0.5252,
      "step": 13125
    },
    {
      "epoch": 4.144616356173034,
      "grad_norm": 3.4111950397491455,
      "learning_rate": 9.504596168689916e-06,
      "loss": 0.5187,
      "step": 13126
    },
    {
      "epoch": 4.14493211240922,
      "grad_norm": 3.654911756515503,
      "learning_rate": 9.501087642972425e-06,
      "loss": 0.5772,
      "step": 13127
    },
    {
      "epoch": 4.145247868645406,
      "grad_norm": 3.6392552852630615,
      "learning_rate": 9.49757911725493e-06,
      "loss": 0.6491,
      "step": 13128
    },
    {
      "epoch": 4.145563624881591,
      "grad_norm": 3.390056610107422,
      "learning_rate": 9.494070591537436e-06,
      "loss": 0.5067,
      "step": 13129
    },
    {
      "epoch": 4.145879381117777,
      "grad_norm": 3.3854126930236816,
      "learning_rate": 9.490562065819943e-06,
      "loss": 0.6653,
      "step": 13130
    },
    {
      "epoch": 4.1461951373539625,
      "grad_norm": 4.121694564819336,
      "learning_rate": 9.487053540102448e-06,
      "loss": 0.5444,
      "step": 13131
    },
    {
      "epoch": 4.146510893590149,
      "grad_norm": 3.5618059635162354,
      "learning_rate": 9.483545014384957e-06,
      "loss": 0.5941,
      "step": 13132
    },
    {
      "epoch": 4.146826649826334,
      "grad_norm": 5.100751876831055,
      "learning_rate": 9.480036488667462e-06,
      "loss": 0.6735,
      "step": 13133
    },
    {
      "epoch": 4.14714240606252,
      "grad_norm": 4.161484718322754,
      "learning_rate": 9.476527962949969e-06,
      "loss": 0.5999,
      "step": 13134
    },
    {
      "epoch": 4.1474581622987055,
      "grad_norm": 4.835129261016846,
      "learning_rate": 9.473019437232476e-06,
      "loss": 0.6491,
      "step": 13135
    },
    {
      "epoch": 4.147773918534891,
      "grad_norm": 3.248861312866211,
      "learning_rate": 9.469510911514982e-06,
      "loss": 0.5213,
      "step": 13136
    },
    {
      "epoch": 4.148089674771077,
      "grad_norm": 4.069717884063721,
      "learning_rate": 9.46600238579749e-06,
      "loss": 0.5806,
      "step": 13137
    },
    {
      "epoch": 4.148405431007262,
      "grad_norm": 4.200521945953369,
      "learning_rate": 9.462493860079994e-06,
      "loss": 0.586,
      "step": 13138
    },
    {
      "epoch": 4.148721187243448,
      "grad_norm": 3.6916916370391846,
      "learning_rate": 9.458985334362501e-06,
      "loss": 0.5315,
      "step": 13139
    },
    {
      "epoch": 4.149036943479634,
      "grad_norm": 3.893061399459839,
      "learning_rate": 9.455476808645008e-06,
      "loss": 0.5933,
      "step": 13140
    },
    {
      "epoch": 4.149352699715819,
      "grad_norm": 4.12485408782959,
      "learning_rate": 9.451968282927515e-06,
      "loss": 0.5614,
      "step": 13141
    },
    {
      "epoch": 4.149668455952005,
      "grad_norm": 3.2155814170837402,
      "learning_rate": 9.448459757210022e-06,
      "loss": 0.5987,
      "step": 13142
    },
    {
      "epoch": 4.1499842121881905,
      "grad_norm": 3.4885764122009277,
      "learning_rate": 9.444951231492527e-06,
      "loss": 0.6292,
      "step": 13143
    },
    {
      "epoch": 4.150299968424377,
      "grad_norm": 3.3725690841674805,
      "learning_rate": 9.441442705775034e-06,
      "loss": 0.6119,
      "step": 13144
    },
    {
      "epoch": 4.150615724660562,
      "grad_norm": 4.544161796569824,
      "learning_rate": 9.43793418005754e-06,
      "loss": 0.4842,
      "step": 13145
    },
    {
      "epoch": 4.150931480896748,
      "grad_norm": 3.8931307792663574,
      "learning_rate": 9.434425654340047e-06,
      "loss": 0.6674,
      "step": 13146
    },
    {
      "epoch": 4.151247237132933,
      "grad_norm": 4.443154811859131,
      "learning_rate": 9.430917128622552e-06,
      "loss": 0.6764,
      "step": 13147
    },
    {
      "epoch": 4.151562993369119,
      "grad_norm": 3.9369137287139893,
      "learning_rate": 9.42740860290506e-06,
      "loss": 0.6791,
      "step": 13148
    },
    {
      "epoch": 4.151878749605305,
      "grad_norm": 3.902777910232544,
      "learning_rate": 9.423900077187568e-06,
      "loss": 0.6053,
      "step": 13149
    },
    {
      "epoch": 4.15219450584149,
      "grad_norm": 3.6284422874450684,
      "learning_rate": 9.420391551470073e-06,
      "loss": 0.6421,
      "step": 13150
    },
    {
      "epoch": 4.152510262077676,
      "grad_norm": 3.997068166732788,
      "learning_rate": 9.41688302575258e-06,
      "loss": 0.6251,
      "step": 13151
    },
    {
      "epoch": 4.152826018313862,
      "grad_norm": 4.80547571182251,
      "learning_rate": 9.413374500035085e-06,
      "loss": 0.5,
      "step": 13152
    },
    {
      "epoch": 4.153141774550047,
      "grad_norm": 4.345376014709473,
      "learning_rate": 9.409865974317592e-06,
      "loss": 0.6303,
      "step": 13153
    },
    {
      "epoch": 4.153457530786233,
      "grad_norm": 4.052127838134766,
      "learning_rate": 9.406357448600098e-06,
      "loss": 0.6695,
      "step": 13154
    },
    {
      "epoch": 4.153773287022418,
      "grad_norm": 4.594107627868652,
      "learning_rate": 9.402848922882605e-06,
      "loss": 0.5957,
      "step": 13155
    },
    {
      "epoch": 4.154089043258605,
      "grad_norm": 3.7985005378723145,
      "learning_rate": 9.399340397165112e-06,
      "loss": 0.6052,
      "step": 13156
    },
    {
      "epoch": 4.15440479949479,
      "grad_norm": 4.058386325836182,
      "learning_rate": 9.395831871447617e-06,
      "loss": 0.6843,
      "step": 13157
    },
    {
      "epoch": 4.154720555730976,
      "grad_norm": 4.053072452545166,
      "learning_rate": 9.392323345730126e-06,
      "loss": 0.592,
      "step": 13158
    },
    {
      "epoch": 4.155036311967161,
      "grad_norm": 4.394092559814453,
      "learning_rate": 9.38881482001263e-06,
      "loss": 0.6227,
      "step": 13159
    },
    {
      "epoch": 4.155352068203347,
      "grad_norm": 3.8959975242614746,
      "learning_rate": 9.385306294295138e-06,
      "loss": 0.5227,
      "step": 13160
    },
    {
      "epoch": 4.155667824439533,
      "grad_norm": 3.666536808013916,
      "learning_rate": 9.381797768577644e-06,
      "loss": 0.5697,
      "step": 13161
    },
    {
      "epoch": 4.155983580675718,
      "grad_norm": 3.6940839290618896,
      "learning_rate": 9.378289242860151e-06,
      "loss": 0.6205,
      "step": 13162
    },
    {
      "epoch": 4.156299336911904,
      "grad_norm": 4.211671829223633,
      "learning_rate": 9.374780717142658e-06,
      "loss": 0.6426,
      "step": 13163
    },
    {
      "epoch": 4.15661509314809,
      "grad_norm": 3.923919916152954,
      "learning_rate": 9.371272191425163e-06,
      "loss": 0.6157,
      "step": 13164
    },
    {
      "epoch": 4.156930849384275,
      "grad_norm": 3.7362167835235596,
      "learning_rate": 9.36776366570767e-06,
      "loss": 0.5838,
      "step": 13165
    },
    {
      "epoch": 4.157246605620461,
      "grad_norm": 3.849860191345215,
      "learning_rate": 9.364255139990177e-06,
      "loss": 0.6274,
      "step": 13166
    },
    {
      "epoch": 4.157562361856646,
      "grad_norm": 4.39780855178833,
      "learning_rate": 9.360746614272684e-06,
      "loss": 0.576,
      "step": 13167
    },
    {
      "epoch": 4.157878118092833,
      "grad_norm": 3.666774272918701,
      "learning_rate": 9.35723808855519e-06,
      "loss": 0.6482,
      "step": 13168
    },
    {
      "epoch": 4.158193874329018,
      "grad_norm": 4.272344589233398,
      "learning_rate": 9.353729562837695e-06,
      "loss": 0.6233,
      "step": 13169
    },
    {
      "epoch": 4.158509630565204,
      "grad_norm": 4.544512748718262,
      "learning_rate": 9.350221037120202e-06,
      "loss": 0.5483,
      "step": 13170
    },
    {
      "epoch": 4.158825386801389,
      "grad_norm": 3.939328670501709,
      "learning_rate": 9.346712511402709e-06,
      "loss": 0.5903,
      "step": 13171
    },
    {
      "epoch": 4.159141143037575,
      "grad_norm": 3.3485517501831055,
      "learning_rate": 9.343203985685216e-06,
      "loss": 0.5424,
      "step": 13172
    },
    {
      "epoch": 4.159456899273761,
      "grad_norm": 3.6056854724884033,
      "learning_rate": 9.339695459967723e-06,
      "loss": 0.5845,
      "step": 13173
    },
    {
      "epoch": 4.159772655509946,
      "grad_norm": 4.720902442932129,
      "learning_rate": 9.336186934250228e-06,
      "loss": 0.6015,
      "step": 13174
    },
    {
      "epoch": 4.160088411746132,
      "grad_norm": 4.480081558227539,
      "learning_rate": 9.332678408532736e-06,
      "loss": 0.6227,
      "step": 13175
    },
    {
      "epoch": 4.160404167982318,
      "grad_norm": 4.1729207038879395,
      "learning_rate": 9.329169882815241e-06,
      "loss": 0.5632,
      "step": 13176
    },
    {
      "epoch": 4.160719924218503,
      "grad_norm": 4.682089328765869,
      "learning_rate": 9.325661357097748e-06,
      "loss": 0.5913,
      "step": 13177
    },
    {
      "epoch": 4.161035680454689,
      "grad_norm": 3.954040050506592,
      "learning_rate": 9.322152831380253e-06,
      "loss": 0.5906,
      "step": 13178
    },
    {
      "epoch": 4.161351436690874,
      "grad_norm": 3.8964991569519043,
      "learning_rate": 9.31864430566276e-06,
      "loss": 0.5606,
      "step": 13179
    },
    {
      "epoch": 4.1616671929270606,
      "grad_norm": 3.5477569103240967,
      "learning_rate": 9.315135779945269e-06,
      "loss": 0.5403,
      "step": 13180
    },
    {
      "epoch": 4.161982949163246,
      "grad_norm": 3.391254425048828,
      "learning_rate": 9.311627254227774e-06,
      "loss": 0.5828,
      "step": 13181
    },
    {
      "epoch": 4.162298705399432,
      "grad_norm": 4.158604621887207,
      "learning_rate": 9.30811872851028e-06,
      "loss": 0.6775,
      "step": 13182
    },
    {
      "epoch": 4.162614461635617,
      "grad_norm": 3.619924545288086,
      "learning_rate": 9.304610202792786e-06,
      "loss": 0.6438,
      "step": 13183
    },
    {
      "epoch": 4.162930217871803,
      "grad_norm": 3.2936553955078125,
      "learning_rate": 9.301101677075294e-06,
      "loss": 0.5659,
      "step": 13184
    },
    {
      "epoch": 4.163245974107989,
      "grad_norm": 3.3762006759643555,
      "learning_rate": 9.2975931513578e-06,
      "loss": 0.5825,
      "step": 13185
    },
    {
      "epoch": 4.163561730344174,
      "grad_norm": 3.392023801803589,
      "learning_rate": 9.294084625640306e-06,
      "loss": 0.5538,
      "step": 13186
    },
    {
      "epoch": 4.16387748658036,
      "grad_norm": 3.3321006298065186,
      "learning_rate": 9.290576099922813e-06,
      "loss": 0.5179,
      "step": 13187
    },
    {
      "epoch": 4.164193242816546,
      "grad_norm": 4.231475830078125,
      "learning_rate": 9.287067574205318e-06,
      "loss": 0.6954,
      "step": 13188
    },
    {
      "epoch": 4.164508999052731,
      "grad_norm": 4.588836193084717,
      "learning_rate": 9.283559048487827e-06,
      "loss": 0.6463,
      "step": 13189
    },
    {
      "epoch": 4.164824755288917,
      "grad_norm": 3.2821855545043945,
      "learning_rate": 9.280050522770332e-06,
      "loss": 0.5723,
      "step": 13190
    },
    {
      "epoch": 4.165140511525102,
      "grad_norm": 3.6443142890930176,
      "learning_rate": 9.276541997052839e-06,
      "loss": 0.5253,
      "step": 13191
    },
    {
      "epoch": 4.1654562677612885,
      "grad_norm": 4.020927906036377,
      "learning_rate": 9.273033471335345e-06,
      "loss": 0.6089,
      "step": 13192
    },
    {
      "epoch": 4.165772023997474,
      "grad_norm": 3.050379753112793,
      "learning_rate": 9.269524945617852e-06,
      "loss": 0.5771,
      "step": 13193
    },
    {
      "epoch": 4.16608778023366,
      "grad_norm": 3.965176820755005,
      "learning_rate": 9.266016419900359e-06,
      "loss": 0.6868,
      "step": 13194
    },
    {
      "epoch": 4.166403536469845,
      "grad_norm": 3.1741225719451904,
      "learning_rate": 9.262507894182864e-06,
      "loss": 0.5694,
      "step": 13195
    },
    {
      "epoch": 4.166719292706031,
      "grad_norm": 3.2530996799468994,
      "learning_rate": 9.258999368465371e-06,
      "loss": 0.4499,
      "step": 13196
    },
    {
      "epoch": 4.167035048942217,
      "grad_norm": 3.071261405944824,
      "learning_rate": 9.255490842747878e-06,
      "loss": 0.4998,
      "step": 13197
    },
    {
      "epoch": 4.167350805178402,
      "grad_norm": 4.121120929718018,
      "learning_rate": 9.251982317030385e-06,
      "loss": 0.5541,
      "step": 13198
    },
    {
      "epoch": 4.167666561414588,
      "grad_norm": 4.0834550857543945,
      "learning_rate": 9.248473791312891e-06,
      "loss": 0.5409,
      "step": 13199
    },
    {
      "epoch": 4.1679823176507735,
      "grad_norm": 3.3793883323669434,
      "learning_rate": 9.244965265595397e-06,
      "loss": 0.5807,
      "step": 13200
    },
    {
      "epoch": 4.1679823176507735,
      "eval_loss": 0.8258863687515259,
      "eval_runtime": 4.5965,
      "eval_samples_per_second": 541.713,
      "eval_steps_per_second": 4.569,
      "step": 13200
    },
    {
      "epoch": 4.168298073886959,
      "grad_norm": 4.380119800567627,
      "learning_rate": 9.241456739877903e-06,
      "loss": 0.63,
      "step": 13201
    },
    {
      "epoch": 4.168613830123145,
      "grad_norm": 4.45753288269043,
      "learning_rate": 9.23794821416041e-06,
      "loss": 0.5328,
      "step": 13202
    },
    {
      "epoch": 4.16892958635933,
      "grad_norm": 4.039679527282715,
      "learning_rate": 9.234439688442917e-06,
      "loss": 0.63,
      "step": 13203
    },
    {
      "epoch": 4.1692453425955165,
      "grad_norm": 4.509425640106201,
      "learning_rate": 9.230931162725424e-06,
      "loss": 0.5914,
      "step": 13204
    },
    {
      "epoch": 4.169561098831702,
      "grad_norm": 4.1169843673706055,
      "learning_rate": 9.227422637007929e-06,
      "loss": 0.6777,
      "step": 13205
    },
    {
      "epoch": 4.169876855067888,
      "grad_norm": 2.753063440322876,
      "learning_rate": 9.223914111290437e-06,
      "loss": 0.5835,
      "step": 13206
    },
    {
      "epoch": 4.170192611304073,
      "grad_norm": 4.468491554260254,
      "learning_rate": 9.220405585572942e-06,
      "loss": 0.6938,
      "step": 13207
    },
    {
      "epoch": 4.1705083675402586,
      "grad_norm": 3.1015233993530273,
      "learning_rate": 9.21689705985545e-06,
      "loss": 0.4966,
      "step": 13208
    },
    {
      "epoch": 4.170824123776445,
      "grad_norm": 3.3352982997894287,
      "learning_rate": 9.213388534137954e-06,
      "loss": 0.5334,
      "step": 13209
    },
    {
      "epoch": 4.17113988001263,
      "grad_norm": 4.549609184265137,
      "learning_rate": 9.209880008420463e-06,
      "loss": 0.7223,
      "step": 13210
    },
    {
      "epoch": 4.171455636248816,
      "grad_norm": 3.9659461975097656,
      "learning_rate": 9.20637148270297e-06,
      "loss": 0.6204,
      "step": 13211
    },
    {
      "epoch": 4.1717713924850015,
      "grad_norm": 3.5965170860290527,
      "learning_rate": 9.202862956985475e-06,
      "loss": 0.561,
      "step": 13212
    },
    {
      "epoch": 4.172087148721188,
      "grad_norm": 3.4638853073120117,
      "learning_rate": 9.199354431267982e-06,
      "loss": 0.5904,
      "step": 13213
    },
    {
      "epoch": 4.172402904957373,
      "grad_norm": 3.8234543800354004,
      "learning_rate": 9.195845905550487e-06,
      "loss": 0.5557,
      "step": 13214
    },
    {
      "epoch": 4.172718661193558,
      "grad_norm": 3.6297287940979004,
      "learning_rate": 9.192337379832995e-06,
      "loss": 0.5663,
      "step": 13215
    },
    {
      "epoch": 4.1730344174297445,
      "grad_norm": 4.820063591003418,
      "learning_rate": 9.1888288541155e-06,
      "loss": 0.6033,
      "step": 13216
    },
    {
      "epoch": 4.17335017366593,
      "grad_norm": 4.837833404541016,
      "learning_rate": 9.185320328398007e-06,
      "loss": 0.6243,
      "step": 13217
    },
    {
      "epoch": 4.173665929902116,
      "grad_norm": 3.9706387519836426,
      "learning_rate": 9.181811802680514e-06,
      "loss": 0.5395,
      "step": 13218
    },
    {
      "epoch": 4.173981686138301,
      "grad_norm": 3.8177998065948486,
      "learning_rate": 9.178303276963021e-06,
      "loss": 0.5296,
      "step": 13219
    },
    {
      "epoch": 4.1742974423744865,
      "grad_norm": 4.6618475914001465,
      "learning_rate": 9.174794751245528e-06,
      "loss": 0.5673,
      "step": 13220
    },
    {
      "epoch": 4.174613198610673,
      "grad_norm": 3.1927545070648193,
      "learning_rate": 9.171286225528033e-06,
      "loss": 0.5666,
      "step": 13221
    },
    {
      "epoch": 4.174928954846858,
      "grad_norm": 4.01462459564209,
      "learning_rate": 9.16777769981054e-06,
      "loss": 0.6164,
      "step": 13222
    },
    {
      "epoch": 4.175244711083044,
      "grad_norm": 3.316972494125366,
      "learning_rate": 9.164269174093046e-06,
      "loss": 0.609,
      "step": 13223
    },
    {
      "epoch": 4.1755604673192295,
      "grad_norm": 3.9702961444854736,
      "learning_rate": 9.160760648375553e-06,
      "loss": 0.6069,
      "step": 13224
    },
    {
      "epoch": 4.175876223555415,
      "grad_norm": 3.8899128437042236,
      "learning_rate": 9.15725212265806e-06,
      "loss": 0.6147,
      "step": 13225
    },
    {
      "epoch": 4.176191979791601,
      "grad_norm": 3.374861717224121,
      "learning_rate": 9.153743596940565e-06,
      "loss": 0.5817,
      "step": 13226
    },
    {
      "epoch": 4.176507736027786,
      "grad_norm": 3.741532802581787,
      "learning_rate": 9.150235071223072e-06,
      "loss": 0.5833,
      "step": 13227
    },
    {
      "epoch": 4.176823492263972,
      "grad_norm": 3.302669048309326,
      "learning_rate": 9.146726545505579e-06,
      "loss": 0.5313,
      "step": 13228
    },
    {
      "epoch": 4.177139248500158,
      "grad_norm": 4.906132221221924,
      "learning_rate": 9.143218019788086e-06,
      "loss": 0.7431,
      "step": 13229
    },
    {
      "epoch": 4.177455004736344,
      "grad_norm": 4.2698822021484375,
      "learning_rate": 9.139709494070592e-06,
      "loss": 0.6478,
      "step": 13230
    },
    {
      "epoch": 4.177770760972529,
      "grad_norm": 4.0077972412109375,
      "learning_rate": 9.136200968353098e-06,
      "loss": 0.5751,
      "step": 13231
    },
    {
      "epoch": 4.1780865172087145,
      "grad_norm": 4.2923994064331055,
      "learning_rate": 9.132692442635606e-06,
      "loss": 0.6648,
      "step": 13232
    },
    {
      "epoch": 4.178402273444901,
      "grad_norm": 3.656795024871826,
      "learning_rate": 9.129183916918111e-06,
      "loss": 0.6768,
      "step": 13233
    },
    {
      "epoch": 4.178718029681086,
      "grad_norm": 3.9526429176330566,
      "learning_rate": 9.125675391200618e-06,
      "loss": 0.5577,
      "step": 13234
    },
    {
      "epoch": 4.179033785917272,
      "grad_norm": 4.089728832244873,
      "learning_rate": 9.122166865483125e-06,
      "loss": 0.6456,
      "step": 13235
    },
    {
      "epoch": 4.179349542153457,
      "grad_norm": 3.638162612915039,
      "learning_rate": 9.11865833976563e-06,
      "loss": 0.5296,
      "step": 13236
    },
    {
      "epoch": 4.179665298389644,
      "grad_norm": 3.9197583198547363,
      "learning_rate": 9.115149814048138e-06,
      "loss": 0.5833,
      "step": 13237
    },
    {
      "epoch": 4.179981054625829,
      "grad_norm": 4.799055099487305,
      "learning_rate": 9.111641288330644e-06,
      "loss": 0.5666,
      "step": 13238
    },
    {
      "epoch": 4.180296810862014,
      "grad_norm": 4.179142951965332,
      "learning_rate": 9.10813276261315e-06,
      "loss": 0.5618,
      "step": 13239
    },
    {
      "epoch": 4.1806125670982,
      "grad_norm": 4.2940874099731445,
      "learning_rate": 9.104624236895657e-06,
      "loss": 0.5877,
      "step": 13240
    },
    {
      "epoch": 4.180928323334386,
      "grad_norm": 4.377900123596191,
      "learning_rate": 9.101115711178164e-06,
      "loss": 0.6014,
      "step": 13241
    },
    {
      "epoch": 4.181244079570572,
      "grad_norm": 3.6176764965057373,
      "learning_rate": 9.09760718546067e-06,
      "loss": 0.6134,
      "step": 13242
    },
    {
      "epoch": 4.181559835806757,
      "grad_norm": 3.2915921211242676,
      "learning_rate": 9.094098659743176e-06,
      "loss": 0.5678,
      "step": 13243
    },
    {
      "epoch": 4.1818755920429425,
      "grad_norm": 3.8639471530914307,
      "learning_rate": 9.090590134025683e-06,
      "loss": 0.6482,
      "step": 13244
    },
    {
      "epoch": 4.182191348279129,
      "grad_norm": 4.0617170333862305,
      "learning_rate": 9.08708160830819e-06,
      "loss": 0.6357,
      "step": 13245
    },
    {
      "epoch": 4.182507104515314,
      "grad_norm": 3.584977149963379,
      "learning_rate": 9.083573082590696e-06,
      "loss": 0.505,
      "step": 13246
    },
    {
      "epoch": 4.1828228607515,
      "grad_norm": 3.9811015129089355,
      "learning_rate": 9.080064556873203e-06,
      "loss": 0.6046,
      "step": 13247
    },
    {
      "epoch": 4.183138616987685,
      "grad_norm": 4.620965480804443,
      "learning_rate": 9.076556031155708e-06,
      "loss": 0.655,
      "step": 13248
    },
    {
      "epoch": 4.183454373223872,
      "grad_norm": 6.323416233062744,
      "learning_rate": 9.073047505438215e-06,
      "loss": 0.6036,
      "step": 13249
    },
    {
      "epoch": 4.183770129460057,
      "grad_norm": 3.304724931716919,
      "learning_rate": 9.069538979720722e-06,
      "loss": 0.6994,
      "step": 13250
    },
    {
      "epoch": 4.184085885696242,
      "grad_norm": 3.8142402172088623,
      "learning_rate": 9.066030454003229e-06,
      "loss": 0.6956,
      "step": 13251
    },
    {
      "epoch": 4.184401641932428,
      "grad_norm": 3.5037753582000732,
      "learning_rate": 9.062521928285734e-06,
      "loss": 0.6418,
      "step": 13252
    },
    {
      "epoch": 4.184717398168614,
      "grad_norm": 3.6398088932037354,
      "learning_rate": 9.05901340256824e-06,
      "loss": 0.6121,
      "step": 13253
    },
    {
      "epoch": 4.1850331544048,
      "grad_norm": 3.478318691253662,
      "learning_rate": 9.05550487685075e-06,
      "loss": 0.5598,
      "step": 13254
    },
    {
      "epoch": 4.185348910640985,
      "grad_norm": 3.6915347576141357,
      "learning_rate": 9.051996351133254e-06,
      "loss": 0.6299,
      "step": 13255
    },
    {
      "epoch": 4.18566466687717,
      "grad_norm": 4.389513969421387,
      "learning_rate": 9.048487825415761e-06,
      "loss": 0.6375,
      "step": 13256
    },
    {
      "epoch": 4.185980423113357,
      "grad_norm": 3.4549853801727295,
      "learning_rate": 9.044979299698266e-06,
      "loss": 0.5239,
      "step": 13257
    },
    {
      "epoch": 4.186296179349542,
      "grad_norm": 3.9641215801239014,
      "learning_rate": 9.041470773980773e-06,
      "loss": 0.6226,
      "step": 13258
    },
    {
      "epoch": 4.186611935585728,
      "grad_norm": 3.736107110977173,
      "learning_rate": 9.03796224826328e-06,
      "loss": 0.7149,
      "step": 13259
    },
    {
      "epoch": 4.186927691821913,
      "grad_norm": 3.809396743774414,
      "learning_rate": 9.034453722545787e-06,
      "loss": 0.5588,
      "step": 13260
    },
    {
      "epoch": 4.1872434480581,
      "grad_norm": 3.4001450538635254,
      "learning_rate": 9.030945196828293e-06,
      "loss": 0.5782,
      "step": 13261
    },
    {
      "epoch": 4.187559204294285,
      "grad_norm": 4.493775844573975,
      "learning_rate": 9.027436671110799e-06,
      "loss": 0.6641,
      "step": 13262
    },
    {
      "epoch": 4.18787496053047,
      "grad_norm": 2.9258077144622803,
      "learning_rate": 9.023928145393307e-06,
      "loss": 0.5302,
      "step": 13263
    },
    {
      "epoch": 4.188190716766656,
      "grad_norm": 3.9770872592926025,
      "learning_rate": 9.020419619675812e-06,
      "loss": 0.5547,
      "step": 13264
    },
    {
      "epoch": 4.188506473002842,
      "grad_norm": 4.6189188957214355,
      "learning_rate": 9.016911093958319e-06,
      "loss": 0.5686,
      "step": 13265
    },
    {
      "epoch": 4.188822229239028,
      "grad_norm": 5.422255992889404,
      "learning_rate": 9.013402568240826e-06,
      "loss": 0.6111,
      "step": 13266
    },
    {
      "epoch": 4.189137985475213,
      "grad_norm": 6.927164554595947,
      "learning_rate": 9.009894042523333e-06,
      "loss": 0.5877,
      "step": 13267
    },
    {
      "epoch": 4.189453741711398,
      "grad_norm": 3.6284968852996826,
      "learning_rate": 9.00638551680584e-06,
      "loss": 0.5421,
      "step": 13268
    },
    {
      "epoch": 4.189769497947585,
      "grad_norm": 3.6799888610839844,
      "learning_rate": 9.002876991088345e-06,
      "loss": 0.5901,
      "step": 13269
    },
    {
      "epoch": 4.19008525418377,
      "grad_norm": 4.795427322387695,
      "learning_rate": 8.999368465370851e-06,
      "loss": 0.685,
      "step": 13270
    },
    {
      "epoch": 4.190401010419956,
      "grad_norm": 4.738039970397949,
      "learning_rate": 8.995859939653358e-06,
      "loss": 0.5415,
      "step": 13271
    },
    {
      "epoch": 4.190716766656141,
      "grad_norm": 3.954944372177124,
      "learning_rate": 8.992351413935865e-06,
      "loss": 0.6208,
      "step": 13272
    },
    {
      "epoch": 4.1910325228923275,
      "grad_norm": 3.912309408187866,
      "learning_rate": 8.988842888218372e-06,
      "loss": 0.6555,
      "step": 13273
    },
    {
      "epoch": 4.191348279128513,
      "grad_norm": 3.765951633453369,
      "learning_rate": 8.985334362500877e-06,
      "loss": 0.5629,
      "step": 13274
    },
    {
      "epoch": 4.191664035364698,
      "grad_norm": 4.0727362632751465,
      "learning_rate": 8.981825836783384e-06,
      "loss": 0.629,
      "step": 13275
    },
    {
      "epoch": 4.191979791600884,
      "grad_norm": 3.8749842643737793,
      "learning_rate": 8.97831731106589e-06,
      "loss": 0.699,
      "step": 13276
    },
    {
      "epoch": 4.19229554783707,
      "grad_norm": 3.335573434829712,
      "learning_rate": 8.974808785348397e-06,
      "loss": 0.5135,
      "step": 13277
    },
    {
      "epoch": 4.192611304073256,
      "grad_norm": 4.533804893493652,
      "learning_rate": 8.971300259630904e-06,
      "loss": 0.5704,
      "step": 13278
    },
    {
      "epoch": 4.192927060309441,
      "grad_norm": 3.9923574924468994,
      "learning_rate": 8.96779173391341e-06,
      "loss": 0.6259,
      "step": 13279
    },
    {
      "epoch": 4.193242816545626,
      "grad_norm": 4.388030052185059,
      "learning_rate": 8.964283208195918e-06,
      "loss": 0.6128,
      "step": 13280
    },
    {
      "epoch": 4.1935585727818125,
      "grad_norm": 4.9322428703308105,
      "learning_rate": 8.960774682478423e-06,
      "loss": 0.5662,
      "step": 13281
    },
    {
      "epoch": 4.193874329017998,
      "grad_norm": 3.3763177394866943,
      "learning_rate": 8.95726615676093e-06,
      "loss": 0.5723,
      "step": 13282
    },
    {
      "epoch": 4.194190085254184,
      "grad_norm": 3.7748022079467773,
      "learning_rate": 8.953757631043435e-06,
      "loss": 0.5087,
      "step": 13283
    },
    {
      "epoch": 4.194505841490369,
      "grad_norm": 3.6509556770324707,
      "learning_rate": 8.950249105325942e-06,
      "loss": 0.5606,
      "step": 13284
    },
    {
      "epoch": 4.1948215977265555,
      "grad_norm": 3.829855442047119,
      "learning_rate": 8.94674057960845e-06,
      "loss": 0.6667,
      "step": 13285
    },
    {
      "epoch": 4.195137353962741,
      "grad_norm": 3.8830478191375732,
      "learning_rate": 8.943232053890955e-06,
      "loss": 0.5945,
      "step": 13286
    },
    {
      "epoch": 4.195453110198926,
      "grad_norm": 4.579442977905273,
      "learning_rate": 8.939723528173462e-06,
      "loss": 0.6092,
      "step": 13287
    },
    {
      "epoch": 4.195768866435112,
      "grad_norm": 3.960144281387329,
      "learning_rate": 8.936215002455967e-06,
      "loss": 0.5439,
      "step": 13288
    },
    {
      "epoch": 4.196084622671298,
      "grad_norm": 4.3179097175598145,
      "learning_rate": 8.932706476738476e-06,
      "loss": 0.7038,
      "step": 13289
    },
    {
      "epoch": 4.196400378907484,
      "grad_norm": 4.517404556274414,
      "learning_rate": 8.929197951020981e-06,
      "loss": 0.6433,
      "step": 13290
    },
    {
      "epoch": 4.196716135143669,
      "grad_norm": 4.134110450744629,
      "learning_rate": 8.925689425303488e-06,
      "loss": 0.599,
      "step": 13291
    },
    {
      "epoch": 4.197031891379854,
      "grad_norm": 4.287231922149658,
      "learning_rate": 8.922180899585994e-06,
      "loss": 0.5577,
      "step": 13292
    },
    {
      "epoch": 4.1973476476160405,
      "grad_norm": 4.089583396911621,
      "learning_rate": 8.918672373868501e-06,
      "loss": 0.6148,
      "step": 13293
    },
    {
      "epoch": 4.197663403852226,
      "grad_norm": 3.841571807861328,
      "learning_rate": 8.915163848151008e-06,
      "loss": 0.617,
      "step": 13294
    },
    {
      "epoch": 4.197979160088412,
      "grad_norm": 3.8352534770965576,
      "learning_rate": 8.911655322433513e-06,
      "loss": 0.4922,
      "step": 13295
    },
    {
      "epoch": 4.198294916324597,
      "grad_norm": 4.218916893005371,
      "learning_rate": 8.90814679671602e-06,
      "loss": 0.6787,
      "step": 13296
    },
    {
      "epoch": 4.1986106725607835,
      "grad_norm": 3.9588122367858887,
      "learning_rate": 8.904638270998527e-06,
      "loss": 0.6488,
      "step": 13297
    },
    {
      "epoch": 4.198926428796969,
      "grad_norm": 3.752455711364746,
      "learning_rate": 8.901129745281034e-06,
      "loss": 0.5754,
      "step": 13298
    },
    {
      "epoch": 4.199242185033154,
      "grad_norm": 4.327669143676758,
      "learning_rate": 8.89762121956354e-06,
      "loss": 0.7098,
      "step": 13299
    },
    {
      "epoch": 4.19955794126934,
      "grad_norm": 4.8741044998168945,
      "learning_rate": 8.894112693846046e-06,
      "loss": 0.6625,
      "step": 13300
    },
    {
      "epoch": 4.1998736975055255,
      "grad_norm": 3.2557966709136963,
      "learning_rate": 8.890604168128552e-06,
      "loss": 0.5581,
      "step": 13301
    },
    {
      "epoch": 4.200189453741712,
      "grad_norm": 4.024673938751221,
      "learning_rate": 8.88709564241106e-06,
      "loss": 0.5535,
      "step": 13302
    },
    {
      "epoch": 4.200505209977897,
      "grad_norm": 3.4251790046691895,
      "learning_rate": 8.883587116693566e-06,
      "loss": 0.5311,
      "step": 13303
    },
    {
      "epoch": 4.200820966214083,
      "grad_norm": 4.101785659790039,
      "learning_rate": 8.880078590976073e-06,
      "loss": 0.6647,
      "step": 13304
    },
    {
      "epoch": 4.2011367224502685,
      "grad_norm": 3.6630868911743164,
      "learning_rate": 8.876570065258578e-06,
      "loss": 0.5465,
      "step": 13305
    },
    {
      "epoch": 4.201452478686454,
      "grad_norm": 3.9493703842163086,
      "learning_rate": 8.873061539541085e-06,
      "loss": 0.6013,
      "step": 13306
    },
    {
      "epoch": 4.20176823492264,
      "grad_norm": 4.041006565093994,
      "learning_rate": 8.869553013823592e-06,
      "loss": 0.6156,
      "step": 13307
    },
    {
      "epoch": 4.202083991158825,
      "grad_norm": 3.4645283222198486,
      "learning_rate": 8.866044488106098e-06,
      "loss": 0.55,
      "step": 13308
    },
    {
      "epoch": 4.202399747395011,
      "grad_norm": 3.173319101333618,
      "learning_rate": 8.862535962388605e-06,
      "loss": 0.4855,
      "step": 13309
    },
    {
      "epoch": 4.202715503631197,
      "grad_norm": 3.9602510929107666,
      "learning_rate": 8.85902743667111e-06,
      "loss": 0.6386,
      "step": 13310
    },
    {
      "epoch": 4.203031259867382,
      "grad_norm": 3.6392011642456055,
      "learning_rate": 8.855518910953619e-06,
      "loss": 0.5366,
      "step": 13311
    },
    {
      "epoch": 4.203347016103568,
      "grad_norm": 4.267539024353027,
      "learning_rate": 8.852010385236124e-06,
      "loss": 0.6803,
      "step": 13312
    },
    {
      "epoch": 4.2036627723397535,
      "grad_norm": 3.5552289485931396,
      "learning_rate": 8.84850185951863e-06,
      "loss": 0.5764,
      "step": 13313
    },
    {
      "epoch": 4.20397852857594,
      "grad_norm": 4.130265712738037,
      "learning_rate": 8.844993333801138e-06,
      "loss": 0.687,
      "step": 13314
    },
    {
      "epoch": 4.204294284812125,
      "grad_norm": 3.3401389122009277,
      "learning_rate": 8.841484808083644e-06,
      "loss": 0.5031,
      "step": 13315
    },
    {
      "epoch": 4.20461004104831,
      "grad_norm": 4.4103264808654785,
      "learning_rate": 8.837976282366151e-06,
      "loss": 0.6366,
      "step": 13316
    },
    {
      "epoch": 4.2049257972844964,
      "grad_norm": 3.9613094329833984,
      "learning_rate": 8.834467756648656e-06,
      "loss": 0.5711,
      "step": 13317
    },
    {
      "epoch": 4.205241553520682,
      "grad_norm": 3.4956047534942627,
      "learning_rate": 8.830959230931163e-06,
      "loss": 0.5924,
      "step": 13318
    },
    {
      "epoch": 4.205557309756868,
      "grad_norm": 3.9005887508392334,
      "learning_rate": 8.827450705213668e-06,
      "loss": 0.6138,
      "step": 13319
    },
    {
      "epoch": 4.205873065993053,
      "grad_norm": 3.698354959487915,
      "learning_rate": 8.823942179496177e-06,
      "loss": 0.5868,
      "step": 13320
    },
    {
      "epoch": 4.206188822229239,
      "grad_norm": 3.237820863723755,
      "learning_rate": 8.820433653778682e-06,
      "loss": 0.5561,
      "step": 13321
    },
    {
      "epoch": 4.206504578465425,
      "grad_norm": 3.411783456802368,
      "learning_rate": 8.816925128061189e-06,
      "loss": 0.5246,
      "step": 13322
    },
    {
      "epoch": 4.20682033470161,
      "grad_norm": 3.5763864517211914,
      "learning_rate": 8.813416602343696e-06,
      "loss": 0.5579,
      "step": 13323
    },
    {
      "epoch": 4.207136090937796,
      "grad_norm": 3.214820384979248,
      "learning_rate": 8.809908076626202e-06,
      "loss": 0.5992,
      "step": 13324
    },
    {
      "epoch": 4.2074518471739815,
      "grad_norm": 4.0378265380859375,
      "learning_rate": 8.80639955090871e-06,
      "loss": 0.6098,
      "step": 13325
    },
    {
      "epoch": 4.207767603410168,
      "grad_norm": 3.9793636798858643,
      "learning_rate": 8.802891025191214e-06,
      "loss": 0.5861,
      "step": 13326
    },
    {
      "epoch": 4.208083359646353,
      "grad_norm": 4.470062732696533,
      "learning_rate": 8.799382499473721e-06,
      "loss": 0.5801,
      "step": 13327
    },
    {
      "epoch": 4.208399115882539,
      "grad_norm": 3.7887368202209473,
      "learning_rate": 8.795873973756228e-06,
      "loss": 0.5588,
      "step": 13328
    },
    {
      "epoch": 4.208714872118724,
      "grad_norm": 3.859149932861328,
      "learning_rate": 8.792365448038735e-06,
      "loss": 0.5279,
      "step": 13329
    },
    {
      "epoch": 4.20903062835491,
      "grad_norm": 3.5134358406066895,
      "learning_rate": 8.788856922321242e-06,
      "loss": 0.5286,
      "step": 13330
    },
    {
      "epoch": 4.209346384591096,
      "grad_norm": 3.6522457599639893,
      "learning_rate": 8.785348396603747e-06,
      "loss": 0.5698,
      "step": 13331
    },
    {
      "epoch": 4.209662140827281,
      "grad_norm": 3.9419264793395996,
      "learning_rate": 8.781839870886253e-06,
      "loss": 0.5512,
      "step": 13332
    },
    {
      "epoch": 4.209977897063467,
      "grad_norm": 3.5287435054779053,
      "learning_rate": 8.77833134516876e-06,
      "loss": 0.5361,
      "step": 13333
    },
    {
      "epoch": 4.210293653299653,
      "grad_norm": 3.8705010414123535,
      "learning_rate": 8.774822819451267e-06,
      "loss": 0.6,
      "step": 13334
    },
    {
      "epoch": 4.210609409535838,
      "grad_norm": 3.8985862731933594,
      "learning_rate": 8.771314293733774e-06,
      "loss": 0.6676,
      "step": 13335
    },
    {
      "epoch": 4.210925165772024,
      "grad_norm": 3.79203724861145,
      "learning_rate": 8.767805768016279e-06,
      "loss": 0.5757,
      "step": 13336
    },
    {
      "epoch": 4.211240922008209,
      "grad_norm": 3.2123847007751465,
      "learning_rate": 8.764297242298788e-06,
      "loss": 0.5309,
      "step": 13337
    },
    {
      "epoch": 4.211556678244396,
      "grad_norm": 4.412103652954102,
      "learning_rate": 8.760788716581293e-06,
      "loss": 0.644,
      "step": 13338
    },
    {
      "epoch": 4.211872434480581,
      "grad_norm": 3.756176710128784,
      "learning_rate": 8.7572801908638e-06,
      "loss": 0.5879,
      "step": 13339
    },
    {
      "epoch": 4.212188190716766,
      "grad_norm": 3.778385877609253,
      "learning_rate": 8.753771665146306e-06,
      "loss": 0.5953,
      "step": 13340
    },
    {
      "epoch": 4.212503946952952,
      "grad_norm": 4.368036270141602,
      "learning_rate": 8.750263139428811e-06,
      "loss": 0.5358,
      "step": 13341
    },
    {
      "epoch": 4.212819703189138,
      "grad_norm": 4.677297592163086,
      "learning_rate": 8.74675461371132e-06,
      "loss": 0.6159,
      "step": 13342
    },
    {
      "epoch": 4.213135459425324,
      "grad_norm": 3.971494197845459,
      "learning_rate": 8.743246087993825e-06,
      "loss": 0.5413,
      "step": 13343
    },
    {
      "epoch": 4.213451215661509,
      "grad_norm": 3.886733293533325,
      "learning_rate": 8.739737562276332e-06,
      "loss": 0.5519,
      "step": 13344
    },
    {
      "epoch": 4.213766971897695,
      "grad_norm": 4.225671291351318,
      "learning_rate": 8.736229036558839e-06,
      "loss": 0.6241,
      "step": 13345
    },
    {
      "epoch": 4.214082728133881,
      "grad_norm": 2.928591012954712,
      "learning_rate": 8.732720510841345e-06,
      "loss": 0.4789,
      "step": 13346
    },
    {
      "epoch": 4.214398484370066,
      "grad_norm": 3.8700780868530273,
      "learning_rate": 8.729211985123852e-06,
      "loss": 0.5886,
      "step": 13347
    },
    {
      "epoch": 4.214714240606252,
      "grad_norm": 4.630222320556641,
      "learning_rate": 8.725703459406357e-06,
      "loss": 0.6026,
      "step": 13348
    },
    {
      "epoch": 4.215029996842437,
      "grad_norm": 3.935255765914917,
      "learning_rate": 8.722194933688864e-06,
      "loss": 0.5475,
      "step": 13349
    },
    {
      "epoch": 4.215345753078624,
      "grad_norm": 3.405916213989258,
      "learning_rate": 8.718686407971371e-06,
      "loss": 0.5664,
      "step": 13350
    },
    {
      "epoch": 4.215661509314809,
      "grad_norm": 5.09451961517334,
      "learning_rate": 8.715177882253878e-06,
      "loss": 0.6057,
      "step": 13351
    },
    {
      "epoch": 4.215977265550995,
      "grad_norm": 3.886192798614502,
      "learning_rate": 8.711669356536385e-06,
      "loss": 0.5674,
      "step": 13352
    },
    {
      "epoch": 4.21629302178718,
      "grad_norm": 4.456512451171875,
      "learning_rate": 8.70816083081889e-06,
      "loss": 0.623,
      "step": 13353
    },
    {
      "epoch": 4.216608778023366,
      "grad_norm": 3.3747806549072266,
      "learning_rate": 8.704652305101397e-06,
      "loss": 0.5128,
      "step": 13354
    },
    {
      "epoch": 4.216924534259552,
      "grad_norm": 4.245760440826416,
      "learning_rate": 8.701143779383903e-06,
      "loss": 0.6246,
      "step": 13355
    },
    {
      "epoch": 4.217240290495737,
      "grad_norm": 3.6395375728607178,
      "learning_rate": 8.69763525366641e-06,
      "loss": 0.5429,
      "step": 13356
    },
    {
      "epoch": 4.217556046731923,
      "grad_norm": 3.668518304824829,
      "learning_rate": 8.694126727948915e-06,
      "loss": 0.4901,
      "step": 13357
    },
    {
      "epoch": 4.217871802968109,
      "grad_norm": 3.7381558418273926,
      "learning_rate": 8.690618202231422e-06,
      "loss": 0.5567,
      "step": 13358
    },
    {
      "epoch": 4.218187559204294,
      "grad_norm": 4.186226844787598,
      "learning_rate": 8.68710967651393e-06,
      "loss": 0.5351,
      "step": 13359
    },
    {
      "epoch": 4.21850331544048,
      "grad_norm": 4.507049560546875,
      "learning_rate": 8.683601150796436e-06,
      "loss": 0.7112,
      "step": 13360
    },
    {
      "epoch": 4.218819071676665,
      "grad_norm": 3.6134121417999268,
      "learning_rate": 8.680092625078943e-06,
      "loss": 0.5557,
      "step": 13361
    },
    {
      "epoch": 4.2191348279128515,
      "grad_norm": 3.859985589981079,
      "learning_rate": 8.676584099361448e-06,
      "loss": 0.5495,
      "step": 13362
    },
    {
      "epoch": 4.219450584149037,
      "grad_norm": 4.517104148864746,
      "learning_rate": 8.673075573643956e-06,
      "loss": 0.723,
      "step": 13363
    },
    {
      "epoch": 4.219766340385223,
      "grad_norm": 3.9854025840759277,
      "learning_rate": 8.669567047926461e-06,
      "loss": 0.5882,
      "step": 13364
    },
    {
      "epoch": 4.220082096621408,
      "grad_norm": 3.6409060955047607,
      "learning_rate": 8.666058522208968e-06,
      "loss": 0.5467,
      "step": 13365
    },
    {
      "epoch": 4.220397852857594,
      "grad_norm": 3.571489095687866,
      "learning_rate": 8.662549996491475e-06,
      "loss": 0.5426,
      "step": 13366
    },
    {
      "epoch": 4.22071360909378,
      "grad_norm": 4.1069440841674805,
      "learning_rate": 8.65904147077398e-06,
      "loss": 0.6123,
      "step": 13367
    },
    {
      "epoch": 4.221029365329965,
      "grad_norm": 3.6020359992980957,
      "learning_rate": 8.655532945056489e-06,
      "loss": 0.4846,
      "step": 13368
    },
    {
      "epoch": 4.221345121566151,
      "grad_norm": 5.894413471221924,
      "learning_rate": 8.652024419338994e-06,
      "loss": 0.4904,
      "step": 13369
    },
    {
      "epoch": 4.221660877802337,
      "grad_norm": 4.041855335235596,
      "learning_rate": 8.6485158936215e-06,
      "loss": 0.6018,
      "step": 13370
    },
    {
      "epoch": 4.221976634038522,
      "grad_norm": 4.27958869934082,
      "learning_rate": 8.645007367904007e-06,
      "loss": 0.6756,
      "step": 13371
    },
    {
      "epoch": 4.222292390274708,
      "grad_norm": 3.9828951358795166,
      "learning_rate": 8.641498842186514e-06,
      "loss": 0.5508,
      "step": 13372
    },
    {
      "epoch": 4.222608146510893,
      "grad_norm": 5.034933090209961,
      "learning_rate": 8.637990316469021e-06,
      "loss": 0.5607,
      "step": 13373
    },
    {
      "epoch": 4.2229239027470795,
      "grad_norm": 5.060827255249023,
      "learning_rate": 8.634481790751526e-06,
      "loss": 0.5275,
      "step": 13374
    },
    {
      "epoch": 4.223239658983265,
      "grad_norm": 4.544882297515869,
      "learning_rate": 8.630973265034033e-06,
      "loss": 0.6401,
      "step": 13375
    },
    {
      "epoch": 4.223555415219451,
      "grad_norm": 4.358157157897949,
      "learning_rate": 8.62746473931654e-06,
      "loss": 0.6519,
      "step": 13376
    },
    {
      "epoch": 4.223871171455636,
      "grad_norm": 3.691176176071167,
      "learning_rate": 8.623956213599046e-06,
      "loss": 0.5511,
      "step": 13377
    },
    {
      "epoch": 4.224186927691822,
      "grad_norm": 3.5703718662261963,
      "learning_rate": 8.620447687881553e-06,
      "loss": 0.6508,
      "step": 13378
    },
    {
      "epoch": 4.224502683928008,
      "grad_norm": 3.333462953567505,
      "learning_rate": 8.616939162164058e-06,
      "loss": 0.6244,
      "step": 13379
    },
    {
      "epoch": 4.224818440164193,
      "grad_norm": 3.7921764850616455,
      "learning_rate": 8.613430636446565e-06,
      "loss": 0.6795,
      "step": 13380
    },
    {
      "epoch": 4.225134196400379,
      "grad_norm": 3.360894203186035,
      "learning_rate": 8.609922110729072e-06,
      "loss": 0.6237,
      "step": 13381
    },
    {
      "epoch": 4.2254499526365645,
      "grad_norm": 3.5062999725341797,
      "learning_rate": 8.606413585011579e-06,
      "loss": 0.5371,
      "step": 13382
    },
    {
      "epoch": 4.22576570887275,
      "grad_norm": 3.784379720687866,
      "learning_rate": 8.602905059294086e-06,
      "loss": 0.5493,
      "step": 13383
    },
    {
      "epoch": 4.226081465108936,
      "grad_norm": 4.538633346557617,
      "learning_rate": 8.59939653357659e-06,
      "loss": 0.6288,
      "step": 13384
    },
    {
      "epoch": 4.226397221345121,
      "grad_norm": 4.356693267822266,
      "learning_rate": 8.5958880078591e-06,
      "loss": 0.6104,
      "step": 13385
    },
    {
      "epoch": 4.2267129775813075,
      "grad_norm": 3.503157615661621,
      "learning_rate": 8.592379482141604e-06,
      "loss": 0.5814,
      "step": 13386
    },
    {
      "epoch": 4.227028733817493,
      "grad_norm": 4.72169303894043,
      "learning_rate": 8.588870956424111e-06,
      "loss": 0.5719,
      "step": 13387
    },
    {
      "epoch": 4.227344490053679,
      "grad_norm": 5.948413372039795,
      "learning_rate": 8.585362430706616e-06,
      "loss": 0.6015,
      "step": 13388
    },
    {
      "epoch": 4.227660246289864,
      "grad_norm": 4.338027477264404,
      "learning_rate": 8.581853904989123e-06,
      "loss": 0.5841,
      "step": 13389
    },
    {
      "epoch": 4.2279760025260495,
      "grad_norm": 4.736668586730957,
      "learning_rate": 8.578345379271632e-06,
      "loss": 0.5883,
      "step": 13390
    },
    {
      "epoch": 4.228291758762236,
      "grad_norm": 4.103479862213135,
      "learning_rate": 8.574836853554137e-06,
      "loss": 0.5819,
      "step": 13391
    },
    {
      "epoch": 4.228607514998421,
      "grad_norm": 3.72198748588562,
      "learning_rate": 8.571328327836644e-06,
      "loss": 0.5387,
      "step": 13392
    },
    {
      "epoch": 4.228923271234607,
      "grad_norm": 3.7174150943756104,
      "learning_rate": 8.567819802119149e-06,
      "loss": 0.5647,
      "step": 13393
    },
    {
      "epoch": 4.2292390274707925,
      "grad_norm": 4.311753273010254,
      "learning_rate": 8.564311276401657e-06,
      "loss": 0.5841,
      "step": 13394
    },
    {
      "epoch": 4.229554783706978,
      "grad_norm": 3.673064947128296,
      "learning_rate": 8.560802750684162e-06,
      "loss": 0.5878,
      "step": 13395
    },
    {
      "epoch": 4.229870539943164,
      "grad_norm": 3.4796366691589355,
      "learning_rate": 8.55729422496667e-06,
      "loss": 0.5484,
      "step": 13396
    },
    {
      "epoch": 4.230186296179349,
      "grad_norm": 4.285023212432861,
      "learning_rate": 8.553785699249176e-06,
      "loss": 0.6448,
      "step": 13397
    },
    {
      "epoch": 4.2305020524155355,
      "grad_norm": 3.7233192920684814,
      "learning_rate": 8.550277173531683e-06,
      "loss": 0.4772,
      "step": 13398
    },
    {
      "epoch": 4.230817808651721,
      "grad_norm": 2.969416618347168,
      "learning_rate": 8.54676864781419e-06,
      "loss": 0.517,
      "step": 13399
    },
    {
      "epoch": 4.231133564887907,
      "grad_norm": 3.5583932399749756,
      "learning_rate": 8.543260122096695e-06,
      "loss": 0.5095,
      "step": 13400
    },
    {
      "epoch": 4.231133564887907,
      "eval_loss": 0.8307303190231323,
      "eval_runtime": 4.6197,
      "eval_samples_per_second": 538.992,
      "eval_steps_per_second": 4.546,
      "step": 13400
    },
    {
      "epoch": 4.231449321124092,
      "grad_norm": 4.613730430603027,
      "learning_rate": 8.539751596379202e-06,
      "loss": 0.7103,
      "step": 13401
    },
    {
      "epoch": 4.2317650773602775,
      "grad_norm": 3.4746785163879395,
      "learning_rate": 8.536243070661708e-06,
      "loss": 0.6389,
      "step": 13402
    },
    {
      "epoch": 4.232080833596464,
      "grad_norm": 5.147790908813477,
      "learning_rate": 8.532734544944215e-06,
      "loss": 0.6805,
      "step": 13403
    },
    {
      "epoch": 4.232396589832649,
      "grad_norm": 4.683647632598877,
      "learning_rate": 8.529226019226722e-06,
      "loss": 0.6061,
      "step": 13404
    },
    {
      "epoch": 4.232712346068835,
      "grad_norm": 3.991853713989258,
      "learning_rate": 8.525717493509227e-06,
      "loss": 0.6103,
      "step": 13405
    },
    {
      "epoch": 4.2330281023050205,
      "grad_norm": 3.1275999546051025,
      "learning_rate": 8.522208967791734e-06,
      "loss": 0.6389,
      "step": 13406
    },
    {
      "epoch": 4.233343858541206,
      "grad_norm": 4.242898464202881,
      "learning_rate": 8.51870044207424e-06,
      "loss": 0.6452,
      "step": 13407
    },
    {
      "epoch": 4.233659614777392,
      "grad_norm": 4.5794219970703125,
      "learning_rate": 8.515191916356748e-06,
      "loss": 0.6094,
      "step": 13408
    },
    {
      "epoch": 4.233975371013577,
      "grad_norm": 3.688737630844116,
      "learning_rate": 8.511683390639254e-06,
      "loss": 0.5435,
      "step": 13409
    },
    {
      "epoch": 4.234291127249763,
      "grad_norm": 3.353196144104004,
      "learning_rate": 8.50817486492176e-06,
      "loss": 0.5291,
      "step": 13410
    },
    {
      "epoch": 4.234606883485949,
      "grad_norm": 3.7754409313201904,
      "learning_rate": 8.504666339204266e-06,
      "loss": 0.488,
      "step": 13411
    },
    {
      "epoch": 4.234922639722135,
      "grad_norm": 3.754847288131714,
      "learning_rate": 8.501157813486773e-06,
      "loss": 0.5323,
      "step": 13412
    },
    {
      "epoch": 4.23523839595832,
      "grad_norm": 3.27445387840271,
      "learning_rate": 8.49764928776928e-06,
      "loss": 0.4917,
      "step": 13413
    },
    {
      "epoch": 4.2355541521945055,
      "grad_norm": 5.332182884216309,
      "learning_rate": 8.494140762051787e-06,
      "loss": 0.6151,
      "step": 13414
    },
    {
      "epoch": 4.235869908430692,
      "grad_norm": 3.897245407104492,
      "learning_rate": 8.490632236334292e-06,
      "loss": 0.5971,
      "step": 13415
    },
    {
      "epoch": 4.236185664666877,
      "grad_norm": 3.477829933166504,
      "learning_rate": 8.4871237106168e-06,
      "loss": 0.5639,
      "step": 13416
    },
    {
      "epoch": 4.236501420903063,
      "grad_norm": 3.5839180946350098,
      "learning_rate": 8.483615184899305e-06,
      "loss": 0.6288,
      "step": 13417
    },
    {
      "epoch": 4.236817177139248,
      "grad_norm": 3.832740545272827,
      "learning_rate": 8.480106659181812e-06,
      "loss": 0.5414,
      "step": 13418
    },
    {
      "epoch": 4.237132933375435,
      "grad_norm": 3.83854603767395,
      "learning_rate": 8.476598133464319e-06,
      "loss": 0.5753,
      "step": 13419
    },
    {
      "epoch": 4.23744868961162,
      "grad_norm": 4.070790767669678,
      "learning_rate": 8.473089607746826e-06,
      "loss": 0.5271,
      "step": 13420
    },
    {
      "epoch": 4.237764445847805,
      "grad_norm": 3.5222275257110596,
      "learning_rate": 8.469581082029333e-06,
      "loss": 0.5385,
      "step": 13421
    },
    {
      "epoch": 4.238080202083991,
      "grad_norm": 4.272184371948242,
      "learning_rate": 8.466072556311838e-06,
      "loss": 0.5852,
      "step": 13422
    },
    {
      "epoch": 4.238395958320177,
      "grad_norm": 3.861262321472168,
      "learning_rate": 8.462564030594345e-06,
      "loss": 0.5405,
      "step": 13423
    },
    {
      "epoch": 4.238711714556363,
      "grad_norm": 4.040457248687744,
      "learning_rate": 8.45905550487685e-06,
      "loss": 0.5278,
      "step": 13424
    },
    {
      "epoch": 4.239027470792548,
      "grad_norm": 4.309025764465332,
      "learning_rate": 8.455546979159358e-06,
      "loss": 0.5717,
      "step": 13425
    },
    {
      "epoch": 4.2393432270287335,
      "grad_norm": 4.201138019561768,
      "learning_rate": 8.452038453441863e-06,
      "loss": 0.6864,
      "step": 13426
    },
    {
      "epoch": 4.23965898326492,
      "grad_norm": 4.223470687866211,
      "learning_rate": 8.44852992772437e-06,
      "loss": 0.6028,
      "step": 13427
    },
    {
      "epoch": 4.239974739501105,
      "grad_norm": 4.8117475509643555,
      "learning_rate": 8.445021402006877e-06,
      "loss": 0.6104,
      "step": 13428
    },
    {
      "epoch": 4.240290495737291,
      "grad_norm": 4.072078227996826,
      "learning_rate": 8.441512876289384e-06,
      "loss": 0.6171,
      "step": 13429
    },
    {
      "epoch": 4.240606251973476,
      "grad_norm": 4.230397701263428,
      "learning_rate": 8.43800435057189e-06,
      "loss": 0.6292,
      "step": 13430
    },
    {
      "epoch": 4.240922008209662,
      "grad_norm": 5.106565475463867,
      "learning_rate": 8.434495824854396e-06,
      "loss": 0.5931,
      "step": 13431
    },
    {
      "epoch": 4.241237764445848,
      "grad_norm": 4.424509048461914,
      "learning_rate": 8.430987299136903e-06,
      "loss": 0.6202,
      "step": 13432
    },
    {
      "epoch": 4.241553520682033,
      "grad_norm": 3.745445728302002,
      "learning_rate": 8.42747877341941e-06,
      "loss": 0.6357,
      "step": 13433
    },
    {
      "epoch": 4.241869276918219,
      "grad_norm": 4.434472560882568,
      "learning_rate": 8.423970247701916e-06,
      "loss": 0.6294,
      "step": 13434
    },
    {
      "epoch": 4.242185033154405,
      "grad_norm": 3.452451229095459,
      "learning_rate": 8.420461721984423e-06,
      "loss": 0.5092,
      "step": 13435
    },
    {
      "epoch": 4.242500789390591,
      "grad_norm": 3.8173611164093018,
      "learning_rate": 8.416953196266928e-06,
      "loss": 0.6004,
      "step": 13436
    },
    {
      "epoch": 4.242816545626776,
      "grad_norm": 3.5692176818847656,
      "learning_rate": 8.413444670549435e-06,
      "loss": 0.5157,
      "step": 13437
    },
    {
      "epoch": 4.243132301862961,
      "grad_norm": 3.533590316772461,
      "learning_rate": 8.409936144831942e-06,
      "loss": 0.618,
      "step": 13438
    },
    {
      "epoch": 4.243448058099148,
      "grad_norm": 4.412570476531982,
      "learning_rate": 8.406427619114449e-06,
      "loss": 0.6091,
      "step": 13439
    },
    {
      "epoch": 4.243763814335333,
      "grad_norm": 4.691736221313477,
      "learning_rate": 8.402919093396955e-06,
      "loss": 0.5429,
      "step": 13440
    },
    {
      "epoch": 4.244079570571519,
      "grad_norm": 3.3391776084899902,
      "learning_rate": 8.39941056767946e-06,
      "loss": 0.5881,
      "step": 13441
    },
    {
      "epoch": 4.244395326807704,
      "grad_norm": 3.5716757774353027,
      "learning_rate": 8.395902041961969e-06,
      "loss": 0.5866,
      "step": 13442
    },
    {
      "epoch": 4.2447110830438906,
      "grad_norm": 3.469858169555664,
      "learning_rate": 8.392393516244474e-06,
      "loss": 0.5117,
      "step": 13443
    },
    {
      "epoch": 4.245026839280076,
      "grad_norm": 4.701933860778809,
      "learning_rate": 8.388884990526981e-06,
      "loss": 0.6627,
      "step": 13444
    },
    {
      "epoch": 4.245342595516261,
      "grad_norm": 3.406439781188965,
      "learning_rate": 8.385376464809488e-06,
      "loss": 0.5141,
      "step": 13445
    },
    {
      "epoch": 4.245658351752447,
      "grad_norm": 4.107078552246094,
      "learning_rate": 8.381867939091995e-06,
      "loss": 0.6188,
      "step": 13446
    },
    {
      "epoch": 4.245974107988633,
      "grad_norm": 3.3285324573516846,
      "learning_rate": 8.378359413374501e-06,
      "loss": 0.5263,
      "step": 13447
    },
    {
      "epoch": 4.246289864224819,
      "grad_norm": 4.264531135559082,
      "learning_rate": 8.374850887657007e-06,
      "loss": 0.6949,
      "step": 13448
    },
    {
      "epoch": 4.246605620461004,
      "grad_norm": 5.400569438934326,
      "learning_rate": 8.371342361939513e-06,
      "loss": 0.5988,
      "step": 13449
    },
    {
      "epoch": 4.246921376697189,
      "grad_norm": 3.29426908493042,
      "learning_rate": 8.36783383622202e-06,
      "loss": 0.5714,
      "step": 13450
    },
    {
      "epoch": 4.247237132933376,
      "grad_norm": 3.3690032958984375,
      "learning_rate": 8.364325310504527e-06,
      "loss": 0.6035,
      "step": 13451
    },
    {
      "epoch": 4.247552889169561,
      "grad_norm": 4.013195037841797,
      "learning_rate": 8.360816784787034e-06,
      "loss": 0.5301,
      "step": 13452
    },
    {
      "epoch": 4.247868645405747,
      "grad_norm": 4.1940412521362305,
      "learning_rate": 8.357308259069539e-06,
      "loss": 0.5842,
      "step": 13453
    },
    {
      "epoch": 4.248184401641932,
      "grad_norm": 4.348827362060547,
      "learning_rate": 8.353799733352046e-06,
      "loss": 0.6571,
      "step": 13454
    },
    {
      "epoch": 4.2485001578781185,
      "grad_norm": 3.7601442337036133,
      "learning_rate": 8.350291207634552e-06,
      "loss": 0.6663,
      "step": 13455
    },
    {
      "epoch": 4.248815914114304,
      "grad_norm": 4.417657852172852,
      "learning_rate": 8.34678268191706e-06,
      "loss": 0.7368,
      "step": 13456
    },
    {
      "epoch": 4.249131670350489,
      "grad_norm": 3.5818424224853516,
      "learning_rate": 8.343274156199566e-06,
      "loss": 0.5221,
      "step": 13457
    },
    {
      "epoch": 4.249447426586675,
      "grad_norm": 4.382074356079102,
      "learning_rate": 8.339765630482071e-06,
      "loss": 0.6187,
      "step": 13458
    },
    {
      "epoch": 4.249763182822861,
      "grad_norm": 4.2770915031433105,
      "learning_rate": 8.336257104764578e-06,
      "loss": 0.5136,
      "step": 13459
    },
    {
      "epoch": 4.250078939059047,
      "grad_norm": 4.0055694580078125,
      "learning_rate": 8.332748579047085e-06,
      "loss": 0.6556,
      "step": 13460
    },
    {
      "epoch": 4.250394695295232,
      "grad_norm": 4.338499069213867,
      "learning_rate": 8.329240053329592e-06,
      "loss": 0.5604,
      "step": 13461
    },
    {
      "epoch": 4.250710451531417,
      "grad_norm": 3.3924591541290283,
      "learning_rate": 8.325731527612097e-06,
      "loss": 0.5731,
      "step": 13462
    },
    {
      "epoch": 4.2510262077676035,
      "grad_norm": 4.4381890296936035,
      "learning_rate": 8.322223001894604e-06,
      "loss": 0.6051,
      "step": 13463
    },
    {
      "epoch": 4.251341964003789,
      "grad_norm": 3.606114387512207,
      "learning_rate": 8.318714476177112e-06,
      "loss": 0.5776,
      "step": 13464
    },
    {
      "epoch": 4.251657720239975,
      "grad_norm": 4.650366306304932,
      "learning_rate": 8.315205950459617e-06,
      "loss": 0.6403,
      "step": 13465
    },
    {
      "epoch": 4.25197347647616,
      "grad_norm": 4.265519142150879,
      "learning_rate": 8.311697424742124e-06,
      "loss": 0.6095,
      "step": 13466
    },
    {
      "epoch": 4.2522892327123465,
      "grad_norm": 4.80231237411499,
      "learning_rate": 8.30818889902463e-06,
      "loss": 0.661,
      "step": 13467
    },
    {
      "epoch": 4.252604988948532,
      "grad_norm": 4.308374404907227,
      "learning_rate": 8.304680373307138e-06,
      "loss": 0.5442,
      "step": 13468
    },
    {
      "epoch": 4.252920745184717,
      "grad_norm": 3.454303026199341,
      "learning_rate": 8.301171847589643e-06,
      "loss": 0.5214,
      "step": 13469
    },
    {
      "epoch": 4.253236501420903,
      "grad_norm": 3.8169405460357666,
      "learning_rate": 8.29766332187215e-06,
      "loss": 0.6473,
      "step": 13470
    },
    {
      "epoch": 4.2535522576570886,
      "grad_norm": 5.1722731590271,
      "learning_rate": 8.294154796154656e-06,
      "loss": 0.6175,
      "step": 13471
    },
    {
      "epoch": 4.253868013893275,
      "grad_norm": 5.180493354797363,
      "learning_rate": 8.290646270437162e-06,
      "loss": 0.653,
      "step": 13472
    },
    {
      "epoch": 4.25418377012946,
      "grad_norm": 3.0565953254699707,
      "learning_rate": 8.28713774471967e-06,
      "loss": 0.4778,
      "step": 13473
    },
    {
      "epoch": 4.254499526365645,
      "grad_norm": 4.7613844871521,
      "learning_rate": 8.283629219002175e-06,
      "loss": 0.6736,
      "step": 13474
    },
    {
      "epoch": 4.2548152826018315,
      "grad_norm": 4.588294506072998,
      "learning_rate": 8.280120693284682e-06,
      "loss": 0.6009,
      "step": 13475
    },
    {
      "epoch": 4.255131038838017,
      "grad_norm": 4.2102885246276855,
      "learning_rate": 8.276612167567189e-06,
      "loss": 0.6011,
      "step": 13476
    },
    {
      "epoch": 4.255446795074203,
      "grad_norm": 4.747203350067139,
      "learning_rate": 8.273103641849696e-06,
      "loss": 0.5679,
      "step": 13477
    },
    {
      "epoch": 4.255762551310388,
      "grad_norm": 3.722111463546753,
      "learning_rate": 8.269595116132202e-06,
      "loss": 0.5994,
      "step": 13478
    },
    {
      "epoch": 4.2560783075465745,
      "grad_norm": 3.912118434906006,
      "learning_rate": 8.266086590414708e-06,
      "loss": 0.6056,
      "step": 13479
    },
    {
      "epoch": 4.25639406378276,
      "grad_norm": 4.190140724182129,
      "learning_rate": 8.262578064697214e-06,
      "loss": 0.6097,
      "step": 13480
    },
    {
      "epoch": 4.256709820018945,
      "grad_norm": 3.4932103157043457,
      "learning_rate": 8.259069538979721e-06,
      "loss": 0.6886,
      "step": 13481
    },
    {
      "epoch": 4.257025576255131,
      "grad_norm": 3.7552566528320312,
      "learning_rate": 8.255561013262228e-06,
      "loss": 0.5881,
      "step": 13482
    },
    {
      "epoch": 4.2573413324913165,
      "grad_norm": 4.2325263023376465,
      "learning_rate": 8.252052487544735e-06,
      "loss": 0.583,
      "step": 13483
    },
    {
      "epoch": 4.257657088727503,
      "grad_norm": 3.8765032291412354,
      "learning_rate": 8.24854396182724e-06,
      "loss": 0.4794,
      "step": 13484
    },
    {
      "epoch": 4.257972844963688,
      "grad_norm": 3.67315936088562,
      "learning_rate": 8.245035436109747e-06,
      "loss": 0.5354,
      "step": 13485
    },
    {
      "epoch": 4.258288601199873,
      "grad_norm": 4.231958389282227,
      "learning_rate": 8.241526910392254e-06,
      "loss": 0.6049,
      "step": 13486
    },
    {
      "epoch": 4.2586043574360595,
      "grad_norm": 3.8550753593444824,
      "learning_rate": 8.23801838467476e-06,
      "loss": 0.5794,
      "step": 13487
    },
    {
      "epoch": 4.258920113672245,
      "grad_norm": 5.591867923736572,
      "learning_rate": 8.234509858957267e-06,
      "loss": 0.6351,
      "step": 13488
    },
    {
      "epoch": 4.259235869908431,
      "grad_norm": 3.5406789779663086,
      "learning_rate": 8.231001333239772e-06,
      "loss": 0.5925,
      "step": 13489
    },
    {
      "epoch": 4.259551626144616,
      "grad_norm": 3.9467315673828125,
      "learning_rate": 8.22749280752228e-06,
      "loss": 0.581,
      "step": 13490
    },
    {
      "epoch": 4.259867382380802,
      "grad_norm": 3.9261882305145264,
      "learning_rate": 8.223984281804786e-06,
      "loss": 0.6346,
      "step": 13491
    },
    {
      "epoch": 4.260183138616988,
      "grad_norm": 4.376494884490967,
      "learning_rate": 8.220475756087293e-06,
      "loss": 0.5877,
      "step": 13492
    },
    {
      "epoch": 4.260498894853173,
      "grad_norm": 3.543789863586426,
      "learning_rate": 8.216967230369798e-06,
      "loss": 0.5973,
      "step": 13493
    },
    {
      "epoch": 4.260814651089359,
      "grad_norm": 4.866093158721924,
      "learning_rate": 8.213458704652305e-06,
      "loss": 0.5415,
      "step": 13494
    },
    {
      "epoch": 4.2611304073255445,
      "grad_norm": 4.099491119384766,
      "learning_rate": 8.209950178934813e-06,
      "loss": 0.6107,
      "step": 13495
    },
    {
      "epoch": 4.261446163561731,
      "grad_norm": 3.8874642848968506,
      "learning_rate": 8.206441653217318e-06,
      "loss": 0.6486,
      "step": 13496
    },
    {
      "epoch": 4.261761919797916,
      "grad_norm": 3.7367167472839355,
      "learning_rate": 8.202933127499825e-06,
      "loss": 0.5141,
      "step": 13497
    },
    {
      "epoch": 4.262077676034101,
      "grad_norm": 3.630465507507324,
      "learning_rate": 8.19942460178233e-06,
      "loss": 0.6051,
      "step": 13498
    },
    {
      "epoch": 4.262393432270287,
      "grad_norm": 3.6760356426239014,
      "learning_rate": 8.195916076064839e-06,
      "loss": 0.5312,
      "step": 13499
    },
    {
      "epoch": 4.262709188506473,
      "grad_norm": 4.0348944664001465,
      "learning_rate": 8.192407550347344e-06,
      "loss": 0.512,
      "step": 13500
    },
    {
      "epoch": 4.263024944742659,
      "grad_norm": 3.634824514389038,
      "learning_rate": 8.18889902462985e-06,
      "loss": 0.6238,
      "step": 13501
    },
    {
      "epoch": 4.263340700978844,
      "grad_norm": 3.481510877609253,
      "learning_rate": 8.185390498912357e-06,
      "loss": 0.5423,
      "step": 13502
    },
    {
      "epoch": 4.26365645721503,
      "grad_norm": 3.8677287101745605,
      "learning_rate": 8.181881973194864e-06,
      "loss": 0.5695,
      "step": 13503
    },
    {
      "epoch": 4.263972213451216,
      "grad_norm": 4.156032562255859,
      "learning_rate": 8.178373447477371e-06,
      "loss": 0.5811,
      "step": 13504
    },
    {
      "epoch": 4.264287969687401,
      "grad_norm": 4.072870254516602,
      "learning_rate": 8.174864921759876e-06,
      "loss": 0.5125,
      "step": 13505
    },
    {
      "epoch": 4.264603725923587,
      "grad_norm": 4.258820533752441,
      "learning_rate": 8.171356396042383e-06,
      "loss": 0.6302,
      "step": 13506
    },
    {
      "epoch": 4.2649194821597725,
      "grad_norm": 3.848672866821289,
      "learning_rate": 8.16784787032489e-06,
      "loss": 0.5848,
      "step": 13507
    },
    {
      "epoch": 4.265235238395959,
      "grad_norm": 3.6388778686523438,
      "learning_rate": 8.164339344607397e-06,
      "loss": 0.5662,
      "step": 13508
    },
    {
      "epoch": 4.265550994632144,
      "grad_norm": 4.893404006958008,
      "learning_rate": 8.160830818889903e-06,
      "loss": 0.6171,
      "step": 13509
    },
    {
      "epoch": 4.26586675086833,
      "grad_norm": 3.406329870223999,
      "learning_rate": 8.157322293172409e-06,
      "loss": 0.539,
      "step": 13510
    },
    {
      "epoch": 4.266182507104515,
      "grad_norm": 3.9524776935577393,
      "learning_rate": 8.153813767454915e-06,
      "loss": 0.534,
      "step": 13511
    },
    {
      "epoch": 4.266498263340701,
      "grad_norm": 5.025753498077393,
      "learning_rate": 8.150305241737422e-06,
      "loss": 0.6277,
      "step": 13512
    },
    {
      "epoch": 4.266814019576887,
      "grad_norm": 3.820065498352051,
      "learning_rate": 8.146796716019929e-06,
      "loss": 0.5888,
      "step": 13513
    },
    {
      "epoch": 4.267129775813072,
      "grad_norm": 4.392144680023193,
      "learning_rate": 8.143288190302436e-06,
      "loss": 0.6451,
      "step": 13514
    },
    {
      "epoch": 4.267445532049258,
      "grad_norm": 4.445508003234863,
      "learning_rate": 8.139779664584941e-06,
      "loss": 0.6194,
      "step": 13515
    },
    {
      "epoch": 4.267761288285444,
      "grad_norm": 3.782801866531372,
      "learning_rate": 8.13627113886745e-06,
      "loss": 0.5452,
      "step": 13516
    },
    {
      "epoch": 4.268077044521629,
      "grad_norm": 4.212785243988037,
      "learning_rate": 8.132762613149955e-06,
      "loss": 0.634,
      "step": 13517
    },
    {
      "epoch": 4.268392800757815,
      "grad_norm": 4.234776496887207,
      "learning_rate": 8.129254087432461e-06,
      "loss": 0.5512,
      "step": 13518
    },
    {
      "epoch": 4.268708556994,
      "grad_norm": 4.428403854370117,
      "learning_rate": 8.125745561714968e-06,
      "loss": 0.5812,
      "step": 13519
    },
    {
      "epoch": 4.269024313230187,
      "grad_norm": 4.588593006134033,
      "learning_rate": 8.122237035997473e-06,
      "loss": 0.6468,
      "step": 13520
    },
    {
      "epoch": 4.269340069466372,
      "grad_norm": 5.374019622802734,
      "learning_rate": 8.118728510279982e-06,
      "loss": 0.5355,
      "step": 13521
    },
    {
      "epoch": 4.269655825702557,
      "grad_norm": 3.766190767288208,
      "learning_rate": 8.115219984562487e-06,
      "loss": 0.6429,
      "step": 13522
    },
    {
      "epoch": 4.269971581938743,
      "grad_norm": 3.754957437515259,
      "learning_rate": 8.111711458844994e-06,
      "loss": 0.4862,
      "step": 13523
    },
    {
      "epoch": 4.270287338174929,
      "grad_norm": 4.098648548126221,
      "learning_rate": 8.1082029331275e-06,
      "loss": 0.6527,
      "step": 13524
    },
    {
      "epoch": 4.270603094411115,
      "grad_norm": 3.3806004524230957,
      "learning_rate": 8.104694407410007e-06,
      "loss": 0.5827,
      "step": 13525
    },
    {
      "epoch": 4.2709188506473,
      "grad_norm": 4.489842891693115,
      "learning_rate": 8.101185881692514e-06,
      "loss": 0.6069,
      "step": 13526
    },
    {
      "epoch": 4.271234606883486,
      "grad_norm": 3.7871017456054688,
      "learning_rate": 8.09767735597502e-06,
      "loss": 0.6007,
      "step": 13527
    },
    {
      "epoch": 4.271550363119672,
      "grad_norm": 3.7227768898010254,
      "learning_rate": 8.094168830257526e-06,
      "loss": 0.5822,
      "step": 13528
    },
    {
      "epoch": 4.271866119355857,
      "grad_norm": 3.5275752544403076,
      "learning_rate": 8.090660304540031e-06,
      "loss": 0.6111,
      "step": 13529
    },
    {
      "epoch": 4.272181875592043,
      "grad_norm": 3.8012707233428955,
      "learning_rate": 8.08715177882254e-06,
      "loss": 0.6523,
      "step": 13530
    },
    {
      "epoch": 4.272497631828228,
      "grad_norm": 4.022378921508789,
      "learning_rate": 8.083643253105047e-06,
      "loss": 0.5349,
      "step": 13531
    },
    {
      "epoch": 4.272813388064415,
      "grad_norm": 3.1205978393554688,
      "learning_rate": 8.080134727387552e-06,
      "loss": 0.5321,
      "step": 13532
    },
    {
      "epoch": 4.2731291443006,
      "grad_norm": 4.392327308654785,
      "learning_rate": 8.076626201670058e-06,
      "loss": 0.5892,
      "step": 13533
    },
    {
      "epoch": 4.273444900536786,
      "grad_norm": 4.691727161407471,
      "learning_rate": 8.073117675952565e-06,
      "loss": 0.6314,
      "step": 13534
    },
    {
      "epoch": 4.273760656772971,
      "grad_norm": 3.8589606285095215,
      "learning_rate": 8.069609150235072e-06,
      "loss": 0.6017,
      "step": 13535
    },
    {
      "epoch": 4.274076413009157,
      "grad_norm": 3.7494113445281982,
      "learning_rate": 8.066100624517577e-06,
      "loss": 0.6925,
      "step": 13536
    },
    {
      "epoch": 4.274392169245343,
      "grad_norm": 3.8569416999816895,
      "learning_rate": 8.062592098800084e-06,
      "loss": 0.6414,
      "step": 13537
    },
    {
      "epoch": 4.274707925481528,
      "grad_norm": 4.009058475494385,
      "learning_rate": 8.059083573082591e-06,
      "loss": 0.6048,
      "step": 13538
    },
    {
      "epoch": 4.275023681717714,
      "grad_norm": 3.556849718093872,
      "learning_rate": 8.055575047365098e-06,
      "loss": 0.5993,
      "step": 13539
    },
    {
      "epoch": 4.2753394379539,
      "grad_norm": 4.9555559158325195,
      "learning_rate": 8.052066521647604e-06,
      "loss": 0.5741,
      "step": 13540
    },
    {
      "epoch": 4.275655194190085,
      "grad_norm": 4.019534587860107,
      "learning_rate": 8.04855799593011e-06,
      "loss": 0.7088,
      "step": 13541
    },
    {
      "epoch": 4.275970950426271,
      "grad_norm": 3.130488634109497,
      "learning_rate": 8.045049470212616e-06,
      "loss": 0.5233,
      "step": 13542
    },
    {
      "epoch": 4.276286706662456,
      "grad_norm": 3.959458827972412,
      "learning_rate": 8.041540944495123e-06,
      "loss": 0.6184,
      "step": 13543
    },
    {
      "epoch": 4.2766024628986425,
      "grad_norm": 3.5664775371551514,
      "learning_rate": 8.03803241877763e-06,
      "loss": 0.6452,
      "step": 13544
    },
    {
      "epoch": 4.276918219134828,
      "grad_norm": 3.8567955493927,
      "learning_rate": 8.034523893060137e-06,
      "loss": 0.5441,
      "step": 13545
    },
    {
      "epoch": 4.277233975371013,
      "grad_norm": 3.7413837909698486,
      "learning_rate": 8.031015367342642e-06,
      "loss": 0.598,
      "step": 13546
    },
    {
      "epoch": 4.277549731607199,
      "grad_norm": 3.625981330871582,
      "learning_rate": 8.02750684162515e-06,
      "loss": 0.5806,
      "step": 13547
    },
    {
      "epoch": 4.277865487843385,
      "grad_norm": 3.088932514190674,
      "learning_rate": 8.023998315907656e-06,
      "loss": 0.617,
      "step": 13548
    },
    {
      "epoch": 4.278181244079571,
      "grad_norm": 3.389742612838745,
      "learning_rate": 8.020489790190162e-06,
      "loss": 0.5739,
      "step": 13549
    },
    {
      "epoch": 4.278497000315756,
      "grad_norm": 3.827775001525879,
      "learning_rate": 8.01698126447267e-06,
      "loss": 0.5343,
      "step": 13550
    },
    {
      "epoch": 4.278812756551942,
      "grad_norm": 4.2682204246521,
      "learning_rate": 8.013472738755176e-06,
      "loss": 0.6133,
      "step": 13551
    },
    {
      "epoch": 4.279128512788128,
      "grad_norm": 3.5421135425567627,
      "learning_rate": 8.009964213037683e-06,
      "loss": 0.6814,
      "step": 13552
    },
    {
      "epoch": 4.279444269024313,
      "grad_norm": 4.818976402282715,
      "learning_rate": 8.006455687320188e-06,
      "loss": 0.6353,
      "step": 13553
    },
    {
      "epoch": 4.279760025260499,
      "grad_norm": 3.7310988903045654,
      "learning_rate": 8.002947161602695e-06,
      "loss": 0.5137,
      "step": 13554
    },
    {
      "epoch": 4.280075781496684,
      "grad_norm": 3.45259165763855,
      "learning_rate": 7.999438635885202e-06,
      "loss": 0.5151,
      "step": 13555
    },
    {
      "epoch": 4.2803915377328705,
      "grad_norm": 3.631525993347168,
      "learning_rate": 7.995930110167708e-06,
      "loss": 0.6209,
      "step": 13556
    },
    {
      "epoch": 4.280707293969056,
      "grad_norm": 3.4666733741760254,
      "learning_rate": 7.992421584450215e-06,
      "loss": 0.6172,
      "step": 13557
    },
    {
      "epoch": 4.281023050205242,
      "grad_norm": 3.486823320388794,
      "learning_rate": 7.98891305873272e-06,
      "loss": 0.5654,
      "step": 13558
    },
    {
      "epoch": 4.281338806441427,
      "grad_norm": 3.874401569366455,
      "learning_rate": 7.985404533015227e-06,
      "loss": 0.7198,
      "step": 13559
    },
    {
      "epoch": 4.281654562677613,
      "grad_norm": 3.10567045211792,
      "learning_rate": 7.981896007297734e-06,
      "loss": 0.5596,
      "step": 13560
    },
    {
      "epoch": 4.281970318913799,
      "grad_norm": 3.5181119441986084,
      "learning_rate": 7.97838748158024e-06,
      "loss": 0.518,
      "step": 13561
    },
    {
      "epoch": 4.282286075149984,
      "grad_norm": 3.8191661834716797,
      "learning_rate": 7.974878955862748e-06,
      "loss": 0.6073,
      "step": 13562
    },
    {
      "epoch": 4.28260183138617,
      "grad_norm": 3.7988429069519043,
      "learning_rate": 7.971370430145253e-06,
      "loss": 0.524,
      "step": 13563
    },
    {
      "epoch": 4.2829175876223555,
      "grad_norm": 4.491408348083496,
      "learning_rate": 7.96786190442776e-06,
      "loss": 0.5902,
      "step": 13564
    },
    {
      "epoch": 4.283233343858541,
      "grad_norm": 4.123913764953613,
      "learning_rate": 7.964353378710266e-06,
      "loss": 0.704,
      "step": 13565
    },
    {
      "epoch": 4.283549100094727,
      "grad_norm": 4.068829536437988,
      "learning_rate": 7.960844852992773e-06,
      "loss": 0.5758,
      "step": 13566
    },
    {
      "epoch": 4.283864856330912,
      "grad_norm": 4.30580472946167,
      "learning_rate": 7.957336327275278e-06,
      "loss": 0.6226,
      "step": 13567
    },
    {
      "epoch": 4.2841806125670985,
      "grad_norm": 3.436098575592041,
      "learning_rate": 7.953827801557785e-06,
      "loss": 0.6165,
      "step": 13568
    },
    {
      "epoch": 4.284496368803284,
      "grad_norm": 3.4780492782592773,
      "learning_rate": 7.950319275840294e-06,
      "loss": 0.4898,
      "step": 13569
    },
    {
      "epoch": 4.284812125039469,
      "grad_norm": 4.699763298034668,
      "learning_rate": 7.946810750122799e-06,
      "loss": 0.562,
      "step": 13570
    },
    {
      "epoch": 4.285127881275655,
      "grad_norm": 5.99202823638916,
      "learning_rate": 7.943302224405306e-06,
      "loss": 0.637,
      "step": 13571
    },
    {
      "epoch": 4.2854436375118405,
      "grad_norm": 3.9178175926208496,
      "learning_rate": 7.93979369868781e-06,
      "loss": 0.5592,
      "step": 13572
    },
    {
      "epoch": 4.285759393748027,
      "grad_norm": 6.154036998748779,
      "learning_rate": 7.93628517297032e-06,
      "loss": 0.5101,
      "step": 13573
    },
    {
      "epoch": 4.286075149984212,
      "grad_norm": 3.9283034801483154,
      "learning_rate": 7.932776647252824e-06,
      "loss": 0.5873,
      "step": 13574
    },
    {
      "epoch": 4.286390906220398,
      "grad_norm": 3.6850643157958984,
      "learning_rate": 7.929268121535331e-06,
      "loss": 0.5449,
      "step": 13575
    },
    {
      "epoch": 4.2867066624565835,
      "grad_norm": 3.4549942016601562,
      "learning_rate": 7.925759595817838e-06,
      "loss": 0.6214,
      "step": 13576
    },
    {
      "epoch": 4.287022418692769,
      "grad_norm": 3.8457722663879395,
      "learning_rate": 7.922251070100343e-06,
      "loss": 0.5837,
      "step": 13577
    },
    {
      "epoch": 4.287338174928955,
      "grad_norm": 5.072366714477539,
      "learning_rate": 7.918742544382852e-06,
      "loss": 0.6107,
      "step": 13578
    },
    {
      "epoch": 4.28765393116514,
      "grad_norm": 4.269505023956299,
      "learning_rate": 7.915234018665357e-06,
      "loss": 0.6068,
      "step": 13579
    },
    {
      "epoch": 4.2879696874013264,
      "grad_norm": 5.563798427581787,
      "learning_rate": 7.911725492947863e-06,
      "loss": 0.5823,
      "step": 13580
    },
    {
      "epoch": 4.288285443637512,
      "grad_norm": 3.564063787460327,
      "learning_rate": 7.90821696723037e-06,
      "loss": 0.6293,
      "step": 13581
    },
    {
      "epoch": 4.288601199873698,
      "grad_norm": 3.6824944019317627,
      "learning_rate": 7.904708441512877e-06,
      "loss": 0.6627,
      "step": 13582
    },
    {
      "epoch": 4.288916956109883,
      "grad_norm": 3.8542962074279785,
      "learning_rate": 7.901199915795384e-06,
      "loss": 0.6449,
      "step": 13583
    },
    {
      "epoch": 4.2892327123460685,
      "grad_norm": 4.011915683746338,
      "learning_rate": 7.897691390077889e-06,
      "loss": 0.55,
      "step": 13584
    },
    {
      "epoch": 4.289548468582255,
      "grad_norm": 3.663766384124756,
      "learning_rate": 7.894182864360396e-06,
      "loss": 0.56,
      "step": 13585
    },
    {
      "epoch": 4.28986422481844,
      "grad_norm": 3.6883010864257812,
      "learning_rate": 7.890674338642903e-06,
      "loss": 0.5916,
      "step": 13586
    },
    {
      "epoch": 4.290179981054626,
      "grad_norm": 3.8010668754577637,
      "learning_rate": 7.88716581292541e-06,
      "loss": 0.5529,
      "step": 13587
    },
    {
      "epoch": 4.2904957372908115,
      "grad_norm": 4.396087169647217,
      "learning_rate": 7.883657287207916e-06,
      "loss": 0.6356,
      "step": 13588
    },
    {
      "epoch": 4.290811493526997,
      "grad_norm": 3.355600357055664,
      "learning_rate": 7.880148761490421e-06,
      "loss": 0.554,
      "step": 13589
    },
    {
      "epoch": 4.291127249763183,
      "grad_norm": 3.8097333908081055,
      "learning_rate": 7.876640235772928e-06,
      "loss": 0.6264,
      "step": 13590
    },
    {
      "epoch": 4.291443005999368,
      "grad_norm": 4.2411932945251465,
      "learning_rate": 7.873131710055435e-06,
      "loss": 0.5636,
      "step": 13591
    },
    {
      "epoch": 4.291758762235554,
      "grad_norm": 4.447266578674316,
      "learning_rate": 7.869623184337942e-06,
      "loss": 0.6647,
      "step": 13592
    },
    {
      "epoch": 4.29207451847174,
      "grad_norm": 4.3569016456604,
      "learning_rate": 7.866114658620449e-06,
      "loss": 0.5919,
      "step": 13593
    },
    {
      "epoch": 4.292390274707926,
      "grad_norm": 3.7181804180145264,
      "learning_rate": 7.862606132902954e-06,
      "loss": 0.5283,
      "step": 13594
    },
    {
      "epoch": 4.292706030944111,
      "grad_norm": 4.2975687980651855,
      "learning_rate": 7.859097607185462e-06,
      "loss": 0.5977,
      "step": 13595
    },
    {
      "epoch": 4.2930217871802965,
      "grad_norm": 4.130538463592529,
      "learning_rate": 7.855589081467967e-06,
      "loss": 0.5519,
      "step": 13596
    },
    {
      "epoch": 4.293337543416483,
      "grad_norm": 3.626002550125122,
      "learning_rate": 7.852080555750474e-06,
      "loss": 0.5801,
      "step": 13597
    },
    {
      "epoch": 4.293653299652668,
      "grad_norm": 3.6749801635742188,
      "learning_rate": 7.84857203003298e-06,
      "loss": 0.6073,
      "step": 13598
    },
    {
      "epoch": 4.293969055888854,
      "grad_norm": 3.2015938758850098,
      "learning_rate": 7.845063504315488e-06,
      "loss": 0.5559,
      "step": 13599
    },
    {
      "epoch": 4.294284812125039,
      "grad_norm": 3.9494712352752686,
      "learning_rate": 7.841554978597995e-06,
      "loss": 0.5512,
      "step": 13600
    },
    {
      "epoch": 4.294284812125039,
      "eval_loss": 0.8338437676429749,
      "eval_runtime": 4.5979,
      "eval_samples_per_second": 541.554,
      "eval_steps_per_second": 4.567,
      "step": 13600
    },
    {
      "epoch": 4.294600568361225,
      "grad_norm": 3.763012170791626,
      "learning_rate": 7.8380464528805e-06,
      "loss": 0.5424,
      "step": 13601
    },
    {
      "epoch": 4.294916324597411,
      "grad_norm": 4.457612037658691,
      "learning_rate": 7.834537927163007e-06,
      "loss": 0.6535,
      "step": 13602
    },
    {
      "epoch": 4.295232080833596,
      "grad_norm": 4.470544815063477,
      "learning_rate": 7.831029401445512e-06,
      "loss": 0.6403,
      "step": 13603
    },
    {
      "epoch": 4.295547837069782,
      "grad_norm": 3.4875080585479736,
      "learning_rate": 7.82752087572802e-06,
      "loss": 0.5573,
      "step": 13604
    },
    {
      "epoch": 4.295863593305968,
      "grad_norm": 4.640260219573975,
      "learning_rate": 7.824012350010525e-06,
      "loss": 0.5597,
      "step": 13605
    },
    {
      "epoch": 4.296179349542154,
      "grad_norm": 4.2911763191223145,
      "learning_rate": 7.820503824293032e-06,
      "loss": 0.5726,
      "step": 13606
    },
    {
      "epoch": 4.296495105778339,
      "grad_norm": 3.6096386909484863,
      "learning_rate": 7.816995298575539e-06,
      "loss": 0.5909,
      "step": 13607
    },
    {
      "epoch": 4.2968108620145244,
      "grad_norm": 3.2879481315612793,
      "learning_rate": 7.813486772858046e-06,
      "loss": 0.5314,
      "step": 13608
    },
    {
      "epoch": 4.297126618250711,
      "grad_norm": 3.611537456512451,
      "learning_rate": 7.809978247140553e-06,
      "loss": 0.6195,
      "step": 13609
    },
    {
      "epoch": 4.297442374486896,
      "grad_norm": 3.3129658699035645,
      "learning_rate": 7.806469721423058e-06,
      "loss": 0.5478,
      "step": 13610
    },
    {
      "epoch": 4.297758130723082,
      "grad_norm": 4.868332386016846,
      "learning_rate": 7.802961195705565e-06,
      "loss": 0.5491,
      "step": 13611
    },
    {
      "epoch": 4.298073886959267,
      "grad_norm": 4.190394878387451,
      "learning_rate": 7.799452669988071e-06,
      "loss": 0.6402,
      "step": 13612
    },
    {
      "epoch": 4.298389643195453,
      "grad_norm": 4.518367290496826,
      "learning_rate": 7.795944144270578e-06,
      "loss": 0.6399,
      "step": 13613
    },
    {
      "epoch": 4.298705399431639,
      "grad_norm": 3.8542535305023193,
      "learning_rate": 7.792435618553085e-06,
      "loss": 0.5563,
      "step": 13614
    },
    {
      "epoch": 4.299021155667824,
      "grad_norm": 3.784154176712036,
      "learning_rate": 7.78892709283559e-06,
      "loss": 0.5391,
      "step": 13615
    },
    {
      "epoch": 4.29933691190401,
      "grad_norm": 2.70310378074646,
      "learning_rate": 7.785418567118097e-06,
      "loss": 0.5465,
      "step": 13616
    },
    {
      "epoch": 4.299652668140196,
      "grad_norm": 3.8941285610198975,
      "learning_rate": 7.781910041400604e-06,
      "loss": 0.6476,
      "step": 13617
    },
    {
      "epoch": 4.299968424376382,
      "grad_norm": 3.365556240081787,
      "learning_rate": 7.77840151568311e-06,
      "loss": 0.5833,
      "step": 13618
    },
    {
      "epoch": 4.300284180612567,
      "grad_norm": 4.000344753265381,
      "learning_rate": 7.774892989965617e-06,
      "loss": 0.6414,
      "step": 13619
    },
    {
      "epoch": 4.300599936848752,
      "grad_norm": 4.31934928894043,
      "learning_rate": 7.771384464248122e-06,
      "loss": 0.5926,
      "step": 13620
    },
    {
      "epoch": 4.300915693084939,
      "grad_norm": 4.001494884490967,
      "learning_rate": 7.767875938530631e-06,
      "loss": 0.605,
      "step": 13621
    },
    {
      "epoch": 4.301231449321124,
      "grad_norm": 3.078806161880493,
      "learning_rate": 7.764367412813136e-06,
      "loss": 0.5201,
      "step": 13622
    },
    {
      "epoch": 4.30154720555731,
      "grad_norm": 4.046755313873291,
      "learning_rate": 7.760858887095643e-06,
      "loss": 0.5982,
      "step": 13623
    },
    {
      "epoch": 4.301862961793495,
      "grad_norm": 3.93432879447937,
      "learning_rate": 7.75735036137815e-06,
      "loss": 0.5261,
      "step": 13624
    },
    {
      "epoch": 4.3021787180296815,
      "grad_norm": 4.0841450691223145,
      "learning_rate": 7.753841835660655e-06,
      "loss": 0.6441,
      "step": 13625
    },
    {
      "epoch": 4.302494474265867,
      "grad_norm": 3.8237719535827637,
      "learning_rate": 7.750333309943163e-06,
      "loss": 0.6025,
      "step": 13626
    },
    {
      "epoch": 4.302810230502052,
      "grad_norm": 3.1337926387786865,
      "learning_rate": 7.746824784225668e-06,
      "loss": 0.5776,
      "step": 13627
    },
    {
      "epoch": 4.303125986738238,
      "grad_norm": 3.651897668838501,
      "learning_rate": 7.743316258508175e-06,
      "loss": 0.6674,
      "step": 13628
    },
    {
      "epoch": 4.303441742974424,
      "grad_norm": 3.9322195053100586,
      "learning_rate": 7.739807732790682e-06,
      "loss": 0.5865,
      "step": 13629
    },
    {
      "epoch": 4.30375749921061,
      "grad_norm": 4.353007793426514,
      "learning_rate": 7.736299207073189e-06,
      "loss": 0.6231,
      "step": 13630
    },
    {
      "epoch": 4.304073255446795,
      "grad_norm": 4.09999418258667,
      "learning_rate": 7.732790681355696e-06,
      "loss": 0.6452,
      "step": 13631
    },
    {
      "epoch": 4.30438901168298,
      "grad_norm": 3.3755524158477783,
      "learning_rate": 7.7292821556382e-06,
      "loss": 0.5636,
      "step": 13632
    },
    {
      "epoch": 4.304704767919167,
      "grad_norm": 3.9505369663238525,
      "learning_rate": 7.725773629920708e-06,
      "loss": 0.559,
      "step": 13633
    },
    {
      "epoch": 4.305020524155352,
      "grad_norm": 3.436039924621582,
      "learning_rate": 7.722265104203214e-06,
      "loss": 0.486,
      "step": 13634
    },
    {
      "epoch": 4.305336280391538,
      "grad_norm": 3.8145053386688232,
      "learning_rate": 7.718756578485721e-06,
      "loss": 0.6389,
      "step": 13635
    },
    {
      "epoch": 4.305652036627723,
      "grad_norm": 3.6372053623199463,
      "learning_rate": 7.715248052768228e-06,
      "loss": 0.5792,
      "step": 13636
    },
    {
      "epoch": 4.305967792863909,
      "grad_norm": 3.2554216384887695,
      "learning_rate": 7.711739527050733e-06,
      "loss": 0.588,
      "step": 13637
    },
    {
      "epoch": 4.306283549100095,
      "grad_norm": 6.028339862823486,
      "learning_rate": 7.70823100133324e-06,
      "loss": 0.6499,
      "step": 13638
    },
    {
      "epoch": 4.30659930533628,
      "grad_norm": 3.827035665512085,
      "learning_rate": 7.704722475615747e-06,
      "loss": 0.5849,
      "step": 13639
    },
    {
      "epoch": 4.306915061572466,
      "grad_norm": 4.978625774383545,
      "learning_rate": 7.701213949898254e-06,
      "loss": 0.6339,
      "step": 13640
    },
    {
      "epoch": 4.307230817808652,
      "grad_norm": 3.4726200103759766,
      "learning_rate": 7.697705424180759e-06,
      "loss": 0.5472,
      "step": 13641
    },
    {
      "epoch": 4.307546574044838,
      "grad_norm": 4.020182132720947,
      "learning_rate": 7.694196898463266e-06,
      "loss": 0.56,
      "step": 13642
    },
    {
      "epoch": 4.307862330281023,
      "grad_norm": 3.873933792114258,
      "learning_rate": 7.690688372745772e-06,
      "loss": 0.5576,
      "step": 13643
    },
    {
      "epoch": 4.308178086517208,
      "grad_norm": 4.522529125213623,
      "learning_rate": 7.68717984702828e-06,
      "loss": 0.5956,
      "step": 13644
    },
    {
      "epoch": 4.3084938427533945,
      "grad_norm": 4.6743268966674805,
      "learning_rate": 7.683671321310786e-06,
      "loss": 0.6149,
      "step": 13645
    },
    {
      "epoch": 4.30880959898958,
      "grad_norm": 3.534234046936035,
      "learning_rate": 7.680162795593291e-06,
      "loss": 0.5821,
      "step": 13646
    },
    {
      "epoch": 4.309125355225766,
      "grad_norm": 4.042041301727295,
      "learning_rate": 7.676654269875798e-06,
      "loss": 0.5284,
      "step": 13647
    },
    {
      "epoch": 4.309441111461951,
      "grad_norm": 3.678900718688965,
      "learning_rate": 7.673145744158305e-06,
      "loss": 0.622,
      "step": 13648
    },
    {
      "epoch": 4.3097568676981375,
      "grad_norm": 4.305042743682861,
      "learning_rate": 7.669637218440812e-06,
      "loss": 0.62,
      "step": 13649
    },
    {
      "epoch": 4.310072623934323,
      "grad_norm": 5.274497985839844,
      "learning_rate": 7.666128692723318e-06,
      "loss": 0.6152,
      "step": 13650
    },
    {
      "epoch": 4.310388380170508,
      "grad_norm": 4.145227432250977,
      "learning_rate": 7.662620167005823e-06,
      "loss": 0.6228,
      "step": 13651
    },
    {
      "epoch": 4.310704136406694,
      "grad_norm": 3.6548659801483154,
      "learning_rate": 7.659111641288332e-06,
      "loss": 0.5995,
      "step": 13652
    },
    {
      "epoch": 4.3110198926428795,
      "grad_norm": 4.55280065536499,
      "learning_rate": 7.655603115570837e-06,
      "loss": 0.5537,
      "step": 13653
    },
    {
      "epoch": 4.311335648879066,
      "grad_norm": 3.612886667251587,
      "learning_rate": 7.652094589853344e-06,
      "loss": 0.5856,
      "step": 13654
    },
    {
      "epoch": 4.311651405115251,
      "grad_norm": 3.641418218612671,
      "learning_rate": 7.64858606413585e-06,
      "loss": 0.5647,
      "step": 13655
    },
    {
      "epoch": 4.311967161351436,
      "grad_norm": 4.641690731048584,
      "learning_rate": 7.645077538418358e-06,
      "loss": 0.5179,
      "step": 13656
    },
    {
      "epoch": 4.3122829175876225,
      "grad_norm": 4.20522403717041,
      "learning_rate": 7.641569012700864e-06,
      "loss": 0.6053,
      "step": 13657
    },
    {
      "epoch": 4.312598673823808,
      "grad_norm": 3.89835262298584,
      "learning_rate": 7.63806048698337e-06,
      "loss": 0.6074,
      "step": 13658
    },
    {
      "epoch": 4.312914430059994,
      "grad_norm": 4.665827751159668,
      "learning_rate": 7.634551961265876e-06,
      "loss": 0.578,
      "step": 13659
    },
    {
      "epoch": 4.313230186296179,
      "grad_norm": 4.077165126800537,
      "learning_rate": 7.631043435548383e-06,
      "loss": 0.5881,
      "step": 13660
    },
    {
      "epoch": 4.313545942532365,
      "grad_norm": 4.081027030944824,
      "learning_rate": 7.62753490983089e-06,
      "loss": 0.6336,
      "step": 13661
    },
    {
      "epoch": 4.313861698768551,
      "grad_norm": 4.202767372131348,
      "learning_rate": 7.624026384113396e-06,
      "loss": 0.4809,
      "step": 13662
    },
    {
      "epoch": 4.314177455004736,
      "grad_norm": 4.000243663787842,
      "learning_rate": 7.620517858395902e-06,
      "loss": 0.6486,
      "step": 13663
    },
    {
      "epoch": 4.314493211240922,
      "grad_norm": 3.777336597442627,
      "learning_rate": 7.617009332678409e-06,
      "loss": 0.677,
      "step": 13664
    },
    {
      "epoch": 4.3148089674771075,
      "grad_norm": 4.232002258300781,
      "learning_rate": 7.613500806960916e-06,
      "loss": 0.5964,
      "step": 13665
    },
    {
      "epoch": 4.315124723713294,
      "grad_norm": 3.226146697998047,
      "learning_rate": 7.609992281243422e-06,
      "loss": 0.6148,
      "step": 13666
    },
    {
      "epoch": 4.315440479949479,
      "grad_norm": 3.4408352375030518,
      "learning_rate": 7.606483755525928e-06,
      "loss": 0.5721,
      "step": 13667
    },
    {
      "epoch": 4.315756236185664,
      "grad_norm": 3.737375497817993,
      "learning_rate": 7.602975229808434e-06,
      "loss": 0.5289,
      "step": 13668
    },
    {
      "epoch": 4.3160719924218505,
      "grad_norm": 4.817880153656006,
      "learning_rate": 7.599466704090942e-06,
      "loss": 0.594,
      "step": 13669
    },
    {
      "epoch": 4.316387748658036,
      "grad_norm": 4.114897727966309,
      "learning_rate": 7.595958178373448e-06,
      "loss": 0.644,
      "step": 13670
    },
    {
      "epoch": 4.316703504894222,
      "grad_norm": 4.198027610778809,
      "learning_rate": 7.592449652655955e-06,
      "loss": 0.6367,
      "step": 13671
    },
    {
      "epoch": 4.317019261130407,
      "grad_norm": 3.4378304481506348,
      "learning_rate": 7.588941126938461e-06,
      "loss": 0.5325,
      "step": 13672
    },
    {
      "epoch": 4.317335017366593,
      "grad_norm": 3.474663734436035,
      "learning_rate": 7.585432601220967e-06,
      "loss": 0.5581,
      "step": 13673
    },
    {
      "epoch": 4.317650773602779,
      "grad_norm": 3.664127826690674,
      "learning_rate": 7.581924075503474e-06,
      "loss": 0.5746,
      "step": 13674
    },
    {
      "epoch": 4.317966529838964,
      "grad_norm": 4.428838729858398,
      "learning_rate": 7.57841554978598e-06,
      "loss": 0.6435,
      "step": 13675
    },
    {
      "epoch": 4.31828228607515,
      "grad_norm": 3.8535168170928955,
      "learning_rate": 7.574907024068487e-06,
      "loss": 0.622,
      "step": 13676
    },
    {
      "epoch": 4.3185980423113355,
      "grad_norm": 3.1210813522338867,
      "learning_rate": 7.571398498350993e-06,
      "loss": 0.6096,
      "step": 13677
    },
    {
      "epoch": 4.318913798547522,
      "grad_norm": 4.64290189743042,
      "learning_rate": 7.567889972633501e-06,
      "loss": 0.5625,
      "step": 13678
    },
    {
      "epoch": 4.319229554783707,
      "grad_norm": 4.020349979400635,
      "learning_rate": 7.564381446916007e-06,
      "loss": 0.5765,
      "step": 13679
    },
    {
      "epoch": 4.319545311019892,
      "grad_norm": 3.3800971508026123,
      "learning_rate": 7.5608729211985126e-06,
      "loss": 0.6462,
      "step": 13680
    },
    {
      "epoch": 4.319861067256078,
      "grad_norm": 4.160780429840088,
      "learning_rate": 7.5573643954810185e-06,
      "loss": 0.6305,
      "step": 13681
    },
    {
      "epoch": 4.320176823492264,
      "grad_norm": 4.1056718826293945,
      "learning_rate": 7.553855869763525e-06,
      "loss": 0.5515,
      "step": 13682
    },
    {
      "epoch": 4.32049257972845,
      "grad_norm": 3.9701669216156006,
      "learning_rate": 7.550347344046033e-06,
      "loss": 0.6799,
      "step": 13683
    },
    {
      "epoch": 4.320808335964635,
      "grad_norm": 3.1030662059783936,
      "learning_rate": 7.546838818328539e-06,
      "loss": 0.5319,
      "step": 13684
    },
    {
      "epoch": 4.3211240922008205,
      "grad_norm": 3.8828160762786865,
      "learning_rate": 7.543330292611045e-06,
      "loss": 0.646,
      "step": 13685
    },
    {
      "epoch": 4.321439848437007,
      "grad_norm": 4.09431266784668,
      "learning_rate": 7.539821766893551e-06,
      "loss": 0.6886,
      "step": 13686
    },
    {
      "epoch": 4.321755604673192,
      "grad_norm": 4.732672691345215,
      "learning_rate": 7.5363132411760586e-06,
      "loss": 0.6063,
      "step": 13687
    },
    {
      "epoch": 4.322071360909378,
      "grad_norm": 3.745711326599121,
      "learning_rate": 7.5328047154585645e-06,
      "loss": 0.5641,
      "step": 13688
    },
    {
      "epoch": 4.3223871171455635,
      "grad_norm": 4.456174373626709,
      "learning_rate": 7.529296189741071e-06,
      "loss": 0.501,
      "step": 13689
    },
    {
      "epoch": 4.32270287338175,
      "grad_norm": 3.656130075454712,
      "learning_rate": 7.525787664023577e-06,
      "loss": 0.5938,
      "step": 13690
    },
    {
      "epoch": 4.323018629617935,
      "grad_norm": 3.8021981716156006,
      "learning_rate": 7.522279138306085e-06,
      "loss": 0.5931,
      "step": 13691
    },
    {
      "epoch": 4.32333438585412,
      "grad_norm": 3.3085579872131348,
      "learning_rate": 7.518770612588591e-06,
      "loss": 0.4972,
      "step": 13692
    },
    {
      "epoch": 4.323650142090306,
      "grad_norm": 5.3167548179626465,
      "learning_rate": 7.515262086871097e-06,
      "loss": 0.5871,
      "step": 13693
    },
    {
      "epoch": 4.323965898326492,
      "grad_norm": 4.160200119018555,
      "learning_rate": 7.511753561153603e-06,
      "loss": 0.545,
      "step": 13694
    },
    {
      "epoch": 4.324281654562678,
      "grad_norm": 4.142629623413086,
      "learning_rate": 7.50824503543611e-06,
      "loss": 0.5987,
      "step": 13695
    },
    {
      "epoch": 4.324597410798863,
      "grad_norm": 4.194700717926025,
      "learning_rate": 7.504736509718617e-06,
      "loss": 0.639,
      "step": 13696
    },
    {
      "epoch": 4.324913167035049,
      "grad_norm": 3.0952389240264893,
      "learning_rate": 7.501227984001123e-06,
      "loss": 0.5613,
      "step": 13697
    },
    {
      "epoch": 4.325228923271235,
      "grad_norm": 3.9800922870635986,
      "learning_rate": 7.497719458283629e-06,
      "loss": 0.6438,
      "step": 13698
    },
    {
      "epoch": 4.32554467950742,
      "grad_norm": 4.483695983886719,
      "learning_rate": 7.494210932566135e-06,
      "loss": 0.6951,
      "step": 13699
    },
    {
      "epoch": 4.325860435743606,
      "grad_norm": 3.5693418979644775,
      "learning_rate": 7.490702406848643e-06,
      "loss": 0.5593,
      "step": 13700
    },
    {
      "epoch": 4.326176191979791,
      "grad_norm": 3.8294436931610107,
      "learning_rate": 7.487193881131149e-06,
      "loss": 0.5405,
      "step": 13701
    },
    {
      "epoch": 4.326491948215978,
      "grad_norm": 4.137505531311035,
      "learning_rate": 7.483685355413656e-06,
      "loss": 0.5619,
      "step": 13702
    },
    {
      "epoch": 4.326807704452163,
      "grad_norm": 3.7997887134552,
      "learning_rate": 7.480176829696162e-06,
      "loss": 0.5882,
      "step": 13703
    },
    {
      "epoch": 4.327123460688348,
      "grad_norm": 4.265361309051514,
      "learning_rate": 7.476668303978669e-06,
      "loss": 0.6622,
      "step": 13704
    },
    {
      "epoch": 4.327439216924534,
      "grad_norm": 3.90901255607605,
      "learning_rate": 7.473159778261175e-06,
      "loss": 0.5571,
      "step": 13705
    },
    {
      "epoch": 4.32775497316072,
      "grad_norm": 3.8078930377960205,
      "learning_rate": 7.469651252543681e-06,
      "loss": 0.7252,
      "step": 13706
    },
    {
      "epoch": 4.328070729396906,
      "grad_norm": 4.31292200088501,
      "learning_rate": 7.466142726826188e-06,
      "loss": 0.5316,
      "step": 13707
    },
    {
      "epoch": 4.328386485633091,
      "grad_norm": 3.8867125511169434,
      "learning_rate": 7.462634201108694e-06,
      "loss": 0.6033,
      "step": 13708
    },
    {
      "epoch": 4.328702241869277,
      "grad_norm": 3.581094980239868,
      "learning_rate": 7.459125675391202e-06,
      "loss": 0.5658,
      "step": 13709
    },
    {
      "epoch": 4.329017998105463,
      "grad_norm": 4.583252906799316,
      "learning_rate": 7.455617149673708e-06,
      "loss": 0.6611,
      "step": 13710
    },
    {
      "epoch": 4.329333754341648,
      "grad_norm": 3.6868300437927246,
      "learning_rate": 7.452108623956214e-06,
      "loss": 0.63,
      "step": 13711
    },
    {
      "epoch": 4.329649510577834,
      "grad_norm": 4.003661632537842,
      "learning_rate": 7.44860009823872e-06,
      "loss": 0.6384,
      "step": 13712
    },
    {
      "epoch": 4.329965266814019,
      "grad_norm": 3.368704080581665,
      "learning_rate": 7.445091572521227e-06,
      "loss": 0.4884,
      "step": 13713
    },
    {
      "epoch": 4.330281023050206,
      "grad_norm": 3.925326108932495,
      "learning_rate": 7.441583046803734e-06,
      "loss": 0.525,
      "step": 13714
    },
    {
      "epoch": 4.330596779286391,
      "grad_norm": 3.9159018993377686,
      "learning_rate": 7.43807452108624e-06,
      "loss": 0.5346,
      "step": 13715
    },
    {
      "epoch": 4.330912535522576,
      "grad_norm": 3.7866227626800537,
      "learning_rate": 7.434565995368746e-06,
      "loss": 0.535,
      "step": 13716
    },
    {
      "epoch": 4.331228291758762,
      "grad_norm": 4.210192680358887,
      "learning_rate": 7.431057469651252e-06,
      "loss": 0.6269,
      "step": 13717
    },
    {
      "epoch": 4.331544047994948,
      "grad_norm": 3.4422965049743652,
      "learning_rate": 7.42754894393376e-06,
      "loss": 0.5711,
      "step": 13718
    },
    {
      "epoch": 4.331859804231134,
      "grad_norm": 3.735043525695801,
      "learning_rate": 7.4240404182162656e-06,
      "loss": 0.604,
      "step": 13719
    },
    {
      "epoch": 4.332175560467319,
      "grad_norm": 4.506010055541992,
      "learning_rate": 7.420531892498772e-06,
      "loss": 0.6077,
      "step": 13720
    },
    {
      "epoch": 4.332491316703505,
      "grad_norm": 4.7124481201171875,
      "learning_rate": 7.417023366781278e-06,
      "loss": 0.5318,
      "step": 13721
    },
    {
      "epoch": 4.332807072939691,
      "grad_norm": 4.9691243171691895,
      "learning_rate": 7.413514841063786e-06,
      "loss": 0.4932,
      "step": 13722
    },
    {
      "epoch": 4.333122829175876,
      "grad_norm": 4.035671710968018,
      "learning_rate": 7.410006315346292e-06,
      "loss": 0.5924,
      "step": 13723
    },
    {
      "epoch": 4.333438585412062,
      "grad_norm": 3.134058713912964,
      "learning_rate": 7.406497789628798e-06,
      "loss": 0.6459,
      "step": 13724
    },
    {
      "epoch": 4.333754341648247,
      "grad_norm": 3.4828124046325684,
      "learning_rate": 7.402989263911305e-06,
      "loss": 0.5576,
      "step": 13725
    },
    {
      "epoch": 4.3340700978844335,
      "grad_norm": 3.500694513320923,
      "learning_rate": 7.3994807381938116e-06,
      "loss": 0.5986,
      "step": 13726
    },
    {
      "epoch": 4.334385854120619,
      "grad_norm": 4.008816719055176,
      "learning_rate": 7.395972212476318e-06,
      "loss": 0.5877,
      "step": 13727
    },
    {
      "epoch": 4.334701610356804,
      "grad_norm": 4.193948745727539,
      "learning_rate": 7.392463686758824e-06,
      "loss": 0.6065,
      "step": 13728
    },
    {
      "epoch": 4.33501736659299,
      "grad_norm": 4.084538459777832,
      "learning_rate": 7.38895516104133e-06,
      "loss": 0.6139,
      "step": 13729
    },
    {
      "epoch": 4.335333122829176,
      "grad_norm": 4.175406455993652,
      "learning_rate": 7.385446635323836e-06,
      "loss": 0.607,
      "step": 13730
    },
    {
      "epoch": 4.335648879065362,
      "grad_norm": 3.881751537322998,
      "learning_rate": 7.381938109606344e-06,
      "loss": 0.6016,
      "step": 13731
    },
    {
      "epoch": 4.335964635301547,
      "grad_norm": 4.0575737953186035,
      "learning_rate": 7.378429583888851e-06,
      "loss": 0.6352,
      "step": 13732
    },
    {
      "epoch": 4.336280391537733,
      "grad_norm": 4.0851240158081055,
      "learning_rate": 7.374921058171357e-06,
      "loss": 0.6302,
      "step": 13733
    },
    {
      "epoch": 4.3365961477739186,
      "grad_norm": 3.373943328857422,
      "learning_rate": 7.371412532453863e-06,
      "loss": 0.5857,
      "step": 13734
    },
    {
      "epoch": 4.336911904010104,
      "grad_norm": 3.9122042655944824,
      "learning_rate": 7.36790400673637e-06,
      "loss": 0.5532,
      "step": 13735
    },
    {
      "epoch": 4.33722766024629,
      "grad_norm": 3.878856897354126,
      "learning_rate": 7.364395481018876e-06,
      "loss": 0.4941,
      "step": 13736
    },
    {
      "epoch": 4.337543416482475,
      "grad_norm": 3.477933645248413,
      "learning_rate": 7.360886955301382e-06,
      "loss": 0.5682,
      "step": 13737
    },
    {
      "epoch": 4.3378591727186615,
      "grad_norm": 3.433196544647217,
      "learning_rate": 7.357378429583889e-06,
      "loss": 0.624,
      "step": 13738
    },
    {
      "epoch": 4.338174928954847,
      "grad_norm": 5.043248176574707,
      "learning_rate": 7.353869903866397e-06,
      "loss": 0.6563,
      "step": 13739
    },
    {
      "epoch": 4.338490685191033,
      "grad_norm": 3.5357303619384766,
      "learning_rate": 7.350361378148903e-06,
      "loss": 0.6473,
      "step": 13740
    },
    {
      "epoch": 4.338806441427218,
      "grad_norm": 4.06584358215332,
      "learning_rate": 7.346852852431409e-06,
      "loss": 0.56,
      "step": 13741
    },
    {
      "epoch": 4.339122197663404,
      "grad_norm": 4.657565116882324,
      "learning_rate": 7.343344326713915e-06,
      "loss": 0.6414,
      "step": 13742
    },
    {
      "epoch": 4.33943795389959,
      "grad_norm": 4.374388217926025,
      "learning_rate": 7.3398358009964215e-06,
      "loss": 0.6189,
      "step": 13743
    },
    {
      "epoch": 4.339753710135775,
      "grad_norm": 3.804744005203247,
      "learning_rate": 7.336327275278928e-06,
      "loss": 0.5775,
      "step": 13744
    },
    {
      "epoch": 4.340069466371961,
      "grad_norm": 4.5036139488220215,
      "learning_rate": 7.332818749561435e-06,
      "loss": 0.6559,
      "step": 13745
    },
    {
      "epoch": 4.3403852226081465,
      "grad_norm": 3.3291308879852295,
      "learning_rate": 7.329310223843941e-06,
      "loss": 0.5669,
      "step": 13746
    },
    {
      "epoch": 4.340700978844332,
      "grad_norm": 4.69490909576416,
      "learning_rate": 7.325801698126447e-06,
      "loss": 0.5168,
      "step": 13747
    },
    {
      "epoch": 4.341016735080518,
      "grad_norm": 4.100136756896973,
      "learning_rate": 7.322293172408955e-06,
      "loss": 0.6427,
      "step": 13748
    },
    {
      "epoch": 4.341332491316703,
      "grad_norm": 3.901536703109741,
      "learning_rate": 7.318784646691461e-06,
      "loss": 0.6523,
      "step": 13749
    },
    {
      "epoch": 4.3416482475528895,
      "grad_norm": 3.8058183193206787,
      "learning_rate": 7.315276120973967e-06,
      "loss": 0.5792,
      "step": 13750
    },
    {
      "epoch": 4.341964003789075,
      "grad_norm": 3.6107707023620605,
      "learning_rate": 7.3117675952564734e-06,
      "loss": 0.6427,
      "step": 13751
    },
    {
      "epoch": 4.34227976002526,
      "grad_norm": 5.371987819671631,
      "learning_rate": 7.308259069538981e-06,
      "loss": 0.6791,
      "step": 13752
    },
    {
      "epoch": 4.342595516261446,
      "grad_norm": 4.119262218475342,
      "learning_rate": 7.304750543821487e-06,
      "loss": 0.6002,
      "step": 13753
    },
    {
      "epoch": 4.3429112724976315,
      "grad_norm": 3.138409376144409,
      "learning_rate": 7.301242018103993e-06,
      "loss": 0.5429,
      "step": 13754
    },
    {
      "epoch": 4.343227028733818,
      "grad_norm": 3.6865174770355225,
      "learning_rate": 7.297733492386499e-06,
      "loss": 0.6475,
      "step": 13755
    },
    {
      "epoch": 4.343542784970003,
      "grad_norm": 3.4304447174072266,
      "learning_rate": 7.294224966669006e-06,
      "loss": 0.5859,
      "step": 13756
    },
    {
      "epoch": 4.343858541206189,
      "grad_norm": 4.42329740524292,
      "learning_rate": 7.290716440951513e-06,
      "loss": 0.7408,
      "step": 13757
    },
    {
      "epoch": 4.3441742974423745,
      "grad_norm": 3.6782772541046143,
      "learning_rate": 7.2872079152340194e-06,
      "loss": 0.5303,
      "step": 13758
    },
    {
      "epoch": 4.34449005367856,
      "grad_norm": 3.835970878601074,
      "learning_rate": 7.283699389516525e-06,
      "loss": 0.5822,
      "step": 13759
    },
    {
      "epoch": 4.344805809914746,
      "grad_norm": 3.3986830711364746,
      "learning_rate": 7.280190863799031e-06,
      "loss": 0.5588,
      "step": 13760
    },
    {
      "epoch": 4.345121566150931,
      "grad_norm": 5.280947685241699,
      "learning_rate": 7.276682338081539e-06,
      "loss": 0.5059,
      "step": 13761
    },
    {
      "epoch": 4.345437322387117,
      "grad_norm": 3.3811047077178955,
      "learning_rate": 7.273173812364045e-06,
      "loss": 0.5728,
      "step": 13762
    },
    {
      "epoch": 4.345753078623303,
      "grad_norm": 3.752411365509033,
      "learning_rate": 7.269665286646552e-06,
      "loss": 0.697,
      "step": 13763
    },
    {
      "epoch": 4.346068834859489,
      "grad_norm": 4.42244291305542,
      "learning_rate": 7.266156760929058e-06,
      "loss": 0.6764,
      "step": 13764
    },
    {
      "epoch": 4.346384591095674,
      "grad_norm": 4.63844633102417,
      "learning_rate": 7.262648235211564e-06,
      "loss": 0.584,
      "step": 13765
    },
    {
      "epoch": 4.3467003473318595,
      "grad_norm": 3.8091108798980713,
      "learning_rate": 7.259139709494071e-06,
      "loss": 0.5644,
      "step": 13766
    },
    {
      "epoch": 4.347016103568046,
      "grad_norm": 3.5110504627227783,
      "learning_rate": 7.255631183776577e-06,
      "loss": 0.594,
      "step": 13767
    },
    {
      "epoch": 4.347331859804231,
      "grad_norm": 3.541015148162842,
      "learning_rate": 7.252122658059083e-06,
      "loss": 0.493,
      "step": 13768
    },
    {
      "epoch": 4.347647616040417,
      "grad_norm": 3.7458794116973877,
      "learning_rate": 7.24861413234159e-06,
      "loss": 0.5828,
      "step": 13769
    },
    {
      "epoch": 4.3479633722766025,
      "grad_norm": 4.4990644454956055,
      "learning_rate": 7.245105606624098e-06,
      "loss": 0.5888,
      "step": 13770
    },
    {
      "epoch": 4.348279128512788,
      "grad_norm": 3.6456568241119385,
      "learning_rate": 7.241597080906604e-06,
      "loss": 0.5001,
      "step": 13771
    },
    {
      "epoch": 4.348594884748974,
      "grad_norm": 4.016480922698975,
      "learning_rate": 7.23808855518911e-06,
      "loss": 0.5999,
      "step": 13772
    },
    {
      "epoch": 4.348910640985159,
      "grad_norm": 4.14406156539917,
      "learning_rate": 7.234580029471616e-06,
      "loss": 0.6032,
      "step": 13773
    },
    {
      "epoch": 4.349226397221345,
      "grad_norm": 3.878826856613159,
      "learning_rate": 7.231071503754123e-06,
      "loss": 0.5447,
      "step": 13774
    },
    {
      "epoch": 4.349542153457531,
      "grad_norm": 3.611837863922119,
      "learning_rate": 7.227562978036629e-06,
      "loss": 0.4975,
      "step": 13775
    },
    {
      "epoch": 4.349857909693716,
      "grad_norm": 3.8923628330230713,
      "learning_rate": 7.224054452319136e-06,
      "loss": 0.5496,
      "step": 13776
    },
    {
      "epoch": 4.350173665929902,
      "grad_norm": 4.405738830566406,
      "learning_rate": 7.220545926601642e-06,
      "loss": 0.6136,
      "step": 13777
    },
    {
      "epoch": 4.3504894221660875,
      "grad_norm": 3.8562521934509277,
      "learning_rate": 7.217037400884148e-06,
      "loss": 0.6524,
      "step": 13778
    },
    {
      "epoch": 4.350805178402274,
      "grad_norm": 3.3811774253845215,
      "learning_rate": 7.213528875166656e-06,
      "loss": 0.5657,
      "step": 13779
    },
    {
      "epoch": 4.351120934638459,
      "grad_norm": 4.524863243103027,
      "learning_rate": 7.210020349449162e-06,
      "loss": 0.5612,
      "step": 13780
    },
    {
      "epoch": 4.351436690874645,
      "grad_norm": 4.993488788604736,
      "learning_rate": 7.2065118237316685e-06,
      "loss": 0.581,
      "step": 13781
    },
    {
      "epoch": 4.35175244711083,
      "grad_norm": 4.21853494644165,
      "learning_rate": 7.2030032980141745e-06,
      "loss": 0.5903,
      "step": 13782
    },
    {
      "epoch": 4.352068203347016,
      "grad_norm": 4.064789772033691,
      "learning_rate": 7.199494772296682e-06,
      "loss": 0.5921,
      "step": 13783
    },
    {
      "epoch": 4.352383959583202,
      "grad_norm": 5.184011459350586,
      "learning_rate": 7.195986246579188e-06,
      "loss": 0.6057,
      "step": 13784
    },
    {
      "epoch": 4.352699715819387,
      "grad_norm": 4.175948143005371,
      "learning_rate": 7.192477720861694e-06,
      "loss": 0.5935,
      "step": 13785
    },
    {
      "epoch": 4.353015472055573,
      "grad_norm": 4.211118221282959,
      "learning_rate": 7.1889691951442e-06,
      "loss": 0.5732,
      "step": 13786
    },
    {
      "epoch": 4.353331228291759,
      "grad_norm": 3.8663086891174316,
      "learning_rate": 7.185460669426708e-06,
      "loss": 0.6036,
      "step": 13787
    },
    {
      "epoch": 4.353646984527945,
      "grad_norm": 4.318302154541016,
      "learning_rate": 7.1819521437092145e-06,
      "loss": 0.6315,
      "step": 13788
    },
    {
      "epoch": 4.35396274076413,
      "grad_norm": 3.7615087032318115,
      "learning_rate": 7.1784436179917205e-06,
      "loss": 0.588,
      "step": 13789
    },
    {
      "epoch": 4.354278497000315,
      "grad_norm": 3.6750714778900146,
      "learning_rate": 7.1749350922742264e-06,
      "loss": 0.5802,
      "step": 13790
    },
    {
      "epoch": 4.354594253236502,
      "grad_norm": 3.9863903522491455,
      "learning_rate": 7.171426566556732e-06,
      "loss": 0.6466,
      "step": 13791
    },
    {
      "epoch": 4.354910009472687,
      "grad_norm": 4.8481855392456055,
      "learning_rate": 7.16791804083924e-06,
      "loss": 0.6092,
      "step": 13792
    },
    {
      "epoch": 4.355225765708873,
      "grad_norm": 4.41338586807251,
      "learning_rate": 7.164409515121746e-06,
      "loss": 0.6262,
      "step": 13793
    },
    {
      "epoch": 4.355541521945058,
      "grad_norm": 5.775998592376709,
      "learning_rate": 7.160900989404253e-06,
      "loss": 0.6221,
      "step": 13794
    },
    {
      "epoch": 4.355857278181244,
      "grad_norm": 3.8789095878601074,
      "learning_rate": 7.157392463686759e-06,
      "loss": 0.645,
      "step": 13795
    },
    {
      "epoch": 4.35617303441743,
      "grad_norm": 4.234213829040527,
      "learning_rate": 7.1538839379692665e-06,
      "loss": 0.5293,
      "step": 13796
    },
    {
      "epoch": 4.356488790653615,
      "grad_norm": 3.6454708576202393,
      "learning_rate": 7.1503754122517724e-06,
      "loss": 0.5582,
      "step": 13797
    },
    {
      "epoch": 4.356804546889801,
      "grad_norm": 4.7591552734375,
      "learning_rate": 7.146866886534278e-06,
      "loss": 0.57,
      "step": 13798
    },
    {
      "epoch": 4.357120303125987,
      "grad_norm": 3.844585657119751,
      "learning_rate": 7.143358360816784e-06,
      "loss": 0.5755,
      "step": 13799
    },
    {
      "epoch": 4.357436059362172,
      "grad_norm": 4.047397613525391,
      "learning_rate": 7.139849835099291e-06,
      "loss": 0.6175,
      "step": 13800
    },
    {
      "epoch": 4.357436059362172,
      "eval_loss": 0.833608090877533,
      "eval_runtime": 4.5903,
      "eval_samples_per_second": 542.447,
      "eval_steps_per_second": 4.575,
      "step": 13800
    },
    {
      "epoch": 4.357751815598358,
      "grad_norm": 3.823026657104492,
      "learning_rate": 7.136341309381799e-06,
      "loss": 0.6146,
      "step": 13801
    },
    {
      "epoch": 4.358067571834543,
      "grad_norm": 4.336236476898193,
      "learning_rate": 7.132832783664305e-06,
      "loss": 0.6213,
      "step": 13802
    },
    {
      "epoch": 4.35838332807073,
      "grad_norm": 4.71596622467041,
      "learning_rate": 7.129324257946811e-06,
      "loss": 0.7443,
      "step": 13803
    },
    {
      "epoch": 4.358699084306915,
      "grad_norm": 4.136136054992676,
      "learning_rate": 7.125815732229317e-06,
      "loss": 0.49,
      "step": 13804
    },
    {
      "epoch": 4.359014840543101,
      "grad_norm": 5.32167387008667,
      "learning_rate": 7.122307206511824e-06,
      "loss": 0.5063,
      "step": 13805
    },
    {
      "epoch": 4.359330596779286,
      "grad_norm": 4.853860855102539,
      "learning_rate": 7.11879868079433e-06,
      "loss": 0.7113,
      "step": 13806
    },
    {
      "epoch": 4.359646353015472,
      "grad_norm": 3.7057487964630127,
      "learning_rate": 7.115290155076837e-06,
      "loss": 0.6684,
      "step": 13807
    },
    {
      "epoch": 4.359962109251658,
      "grad_norm": 3.563870429992676,
      "learning_rate": 7.111781629359343e-06,
      "loss": 0.6005,
      "step": 13808
    },
    {
      "epoch": 4.360277865487843,
      "grad_norm": 4.808894634246826,
      "learning_rate": 7.108273103641851e-06,
      "loss": 0.6895,
      "step": 13809
    },
    {
      "epoch": 4.360593621724029,
      "grad_norm": 3.998504638671875,
      "learning_rate": 7.104764577924357e-06,
      "loss": 0.6113,
      "step": 13810
    },
    {
      "epoch": 4.360909377960215,
      "grad_norm": 3.8785624504089355,
      "learning_rate": 7.101256052206863e-06,
      "loss": 0.574,
      "step": 13811
    },
    {
      "epoch": 4.361225134196401,
      "grad_norm": 4.1587815284729,
      "learning_rate": 7.0977475264893695e-06,
      "loss": 0.551,
      "step": 13812
    },
    {
      "epoch": 4.361540890432586,
      "grad_norm": 5.974610805511475,
      "learning_rate": 7.0942390007718755e-06,
      "loss": 0.6766,
      "step": 13813
    },
    {
      "epoch": 4.361856646668771,
      "grad_norm": 3.033799886703491,
      "learning_rate": 7.090730475054383e-06,
      "loss": 0.5423,
      "step": 13814
    },
    {
      "epoch": 4.362172402904958,
      "grad_norm": 3.6185481548309326,
      "learning_rate": 7.087221949336889e-06,
      "loss": 0.5329,
      "step": 13815
    },
    {
      "epoch": 4.362488159141143,
      "grad_norm": 3.6126766204833984,
      "learning_rate": 7.083713423619395e-06,
      "loss": 0.6108,
      "step": 13816
    },
    {
      "epoch": 4.362803915377329,
      "grad_norm": 4.424356460571289,
      "learning_rate": 7.080204897901901e-06,
      "loss": 0.6684,
      "step": 13817
    },
    {
      "epoch": 4.363119671613514,
      "grad_norm": 3.7979490756988525,
      "learning_rate": 7.076696372184409e-06,
      "loss": 0.5799,
      "step": 13818
    },
    {
      "epoch": 4.3634354278497,
      "grad_norm": 4.923998832702637,
      "learning_rate": 7.0731878464669155e-06,
      "loss": 0.5934,
      "step": 13819
    },
    {
      "epoch": 4.363751184085886,
      "grad_norm": 4.199371337890625,
      "learning_rate": 7.0696793207494215e-06,
      "loss": 0.6349,
      "step": 13820
    },
    {
      "epoch": 4.364066940322071,
      "grad_norm": 3.7669830322265625,
      "learning_rate": 7.0661707950319275e-06,
      "loss": 0.5714,
      "step": 13821
    },
    {
      "epoch": 4.364382696558257,
      "grad_norm": 3.874544620513916,
      "learning_rate": 7.062662269314435e-06,
      "loss": 0.6132,
      "step": 13822
    },
    {
      "epoch": 4.364698452794443,
      "grad_norm": 2.9147632122039795,
      "learning_rate": 7.059153743596941e-06,
      "loss": 0.537,
      "step": 13823
    },
    {
      "epoch": 4.365014209030629,
      "grad_norm": 3.9572534561157227,
      "learning_rate": 7.055645217879447e-06,
      "loss": 0.6439,
      "step": 13824
    },
    {
      "epoch": 4.365329965266814,
      "grad_norm": 3.9441192150115967,
      "learning_rate": 7.052136692161954e-06,
      "loss": 0.5481,
      "step": 13825
    },
    {
      "epoch": 4.365645721502999,
      "grad_norm": 4.219788551330566,
      "learning_rate": 7.04862816644446e-06,
      "loss": 0.6285,
      "step": 13826
    },
    {
      "epoch": 4.3659614777391855,
      "grad_norm": 4.181378364562988,
      "learning_rate": 7.0451196407269675e-06,
      "loss": 0.6062,
      "step": 13827
    },
    {
      "epoch": 4.366277233975371,
      "grad_norm": 4.131841659545898,
      "learning_rate": 7.0416111150094735e-06,
      "loss": 0.5978,
      "step": 13828
    },
    {
      "epoch": 4.366592990211557,
      "grad_norm": 4.228878021240234,
      "learning_rate": 7.0381025892919794e-06,
      "loss": 0.5515,
      "step": 13829
    },
    {
      "epoch": 4.366908746447742,
      "grad_norm": 3.587437391281128,
      "learning_rate": 7.034594063574486e-06,
      "loss": 0.5722,
      "step": 13830
    },
    {
      "epoch": 4.3672245026839285,
      "grad_norm": 3.927903175354004,
      "learning_rate": 7.031085537856993e-06,
      "loss": 0.607,
      "step": 13831
    },
    {
      "epoch": 4.367540258920114,
      "grad_norm": 3.5823447704315186,
      "learning_rate": 7.0275770121395e-06,
      "loss": 0.5741,
      "step": 13832
    },
    {
      "epoch": 4.367856015156299,
      "grad_norm": 4.324742794036865,
      "learning_rate": 7.024068486422006e-06,
      "loss": 0.5113,
      "step": 13833
    },
    {
      "epoch": 4.368171771392485,
      "grad_norm": 3.437570810317993,
      "learning_rate": 7.020559960704512e-06,
      "loss": 0.5323,
      "step": 13834
    },
    {
      "epoch": 4.3684875276286705,
      "grad_norm": 3.5931622982025146,
      "learning_rate": 7.017051434987018e-06,
      "loss": 0.4529,
      "step": 13835
    },
    {
      "epoch": 4.368803283864857,
      "grad_norm": 4.057287216186523,
      "learning_rate": 7.0135429092695254e-06,
      "loss": 0.5787,
      "step": 13836
    },
    {
      "epoch": 4.369119040101042,
      "grad_norm": 4.105081081390381,
      "learning_rate": 7.010034383552032e-06,
      "loss": 0.6086,
      "step": 13837
    },
    {
      "epoch": 4.369434796337227,
      "grad_norm": 3.556492805480957,
      "learning_rate": 7.006525857834538e-06,
      "loss": 0.5208,
      "step": 13838
    },
    {
      "epoch": 4.3697505525734135,
      "grad_norm": 3.8421761989593506,
      "learning_rate": 7.003017332117044e-06,
      "loss": 0.5797,
      "step": 13839
    },
    {
      "epoch": 4.370066308809599,
      "grad_norm": 3.566617727279663,
      "learning_rate": 6.999508806399552e-06,
      "loss": 0.5571,
      "step": 13840
    },
    {
      "epoch": 4.370382065045785,
      "grad_norm": 3.272052526473999,
      "learning_rate": 6.996000280682058e-06,
      "loss": 0.635,
      "step": 13841
    },
    {
      "epoch": 4.37069782128197,
      "grad_norm": 4.884022235870361,
      "learning_rate": 6.992491754964564e-06,
      "loss": 0.7059,
      "step": 13842
    },
    {
      "epoch": 4.371013577518156,
      "grad_norm": 3.5893261432647705,
      "learning_rate": 6.9889832292470706e-06,
      "loss": 0.5177,
      "step": 13843
    },
    {
      "epoch": 4.371329333754342,
      "grad_norm": 4.150694847106934,
      "learning_rate": 6.985474703529578e-06,
      "loss": 0.5739,
      "step": 13844
    },
    {
      "epoch": 4.371645089990527,
      "grad_norm": 3.9897866249084473,
      "learning_rate": 6.981966177812084e-06,
      "loss": 0.5786,
      "step": 13845
    },
    {
      "epoch": 4.371960846226713,
      "grad_norm": 4.1873459815979,
      "learning_rate": 6.97845765209459e-06,
      "loss": 0.5615,
      "step": 13846
    },
    {
      "epoch": 4.3722766024628985,
      "grad_norm": 3.8536691665649414,
      "learning_rate": 6.974949126377096e-06,
      "loss": 0.5226,
      "step": 13847
    },
    {
      "epoch": 4.372592358699085,
      "grad_norm": 3.8302743434906006,
      "learning_rate": 6.971440600659603e-06,
      "loss": 0.5806,
      "step": 13848
    },
    {
      "epoch": 4.37290811493527,
      "grad_norm": 4.02713680267334,
      "learning_rate": 6.96793207494211e-06,
      "loss": 0.6112,
      "step": 13849
    },
    {
      "epoch": 4.373223871171455,
      "grad_norm": 4.722591400146484,
      "learning_rate": 6.9644235492246166e-06,
      "loss": 0.6303,
      "step": 13850
    },
    {
      "epoch": 4.3735396274076415,
      "grad_norm": 3.8132920265197754,
      "learning_rate": 6.9609150235071225e-06,
      "loss": 0.6585,
      "step": 13851
    },
    {
      "epoch": 4.373855383643827,
      "grad_norm": 3.5658810138702393,
      "learning_rate": 6.9574064977896285e-06,
      "loss": 0.5275,
      "step": 13852
    },
    {
      "epoch": 4.374171139880013,
      "grad_norm": 3.5912373065948486,
      "learning_rate": 6.953897972072136e-06,
      "loss": 0.5604,
      "step": 13853
    },
    {
      "epoch": 4.374486896116198,
      "grad_norm": 4.086442470550537,
      "learning_rate": 6.950389446354642e-06,
      "loss": 0.5639,
      "step": 13854
    },
    {
      "epoch": 4.374802652352384,
      "grad_norm": 4.224209785461426,
      "learning_rate": 6.946880920637148e-06,
      "loss": 0.6628,
      "step": 13855
    },
    {
      "epoch": 4.37511840858857,
      "grad_norm": 4.161101341247559,
      "learning_rate": 6.943372394919655e-06,
      "loss": 0.7003,
      "step": 13856
    },
    {
      "epoch": 4.375434164824755,
      "grad_norm": 4.040572166442871,
      "learning_rate": 6.9398638692021626e-06,
      "loss": 0.6207,
      "step": 13857
    },
    {
      "epoch": 4.375749921060941,
      "grad_norm": 3.9136605262756348,
      "learning_rate": 6.9363553434846685e-06,
      "loss": 0.6115,
      "step": 13858
    },
    {
      "epoch": 4.3760656772971265,
      "grad_norm": 4.372791290283203,
      "learning_rate": 6.9328468177671745e-06,
      "loss": 0.6231,
      "step": 13859
    },
    {
      "epoch": 4.376381433533313,
      "grad_norm": 3.927671194076538,
      "learning_rate": 6.9293382920496805e-06,
      "loss": 0.61,
      "step": 13860
    },
    {
      "epoch": 4.376697189769498,
      "grad_norm": 3.7705023288726807,
      "learning_rate": 6.925829766332187e-06,
      "loss": 0.5217,
      "step": 13861
    },
    {
      "epoch": 4.377012946005683,
      "grad_norm": 3.719646692276001,
      "learning_rate": 6.922321240614694e-06,
      "loss": 0.6554,
      "step": 13862
    },
    {
      "epoch": 4.377328702241869,
      "grad_norm": 4.244670867919922,
      "learning_rate": 6.918812714897201e-06,
      "loss": 0.5585,
      "step": 13863
    },
    {
      "epoch": 4.377644458478055,
      "grad_norm": 3.8242619037628174,
      "learning_rate": 6.915304189179707e-06,
      "loss": 0.5887,
      "step": 13864
    },
    {
      "epoch": 4.377960214714241,
      "grad_norm": 4.268172264099121,
      "learning_rate": 6.911795663462213e-06,
      "loss": 0.6366,
      "step": 13865
    },
    {
      "epoch": 4.378275970950426,
      "grad_norm": 3.734527111053467,
      "learning_rate": 6.9082871377447205e-06,
      "loss": 0.5511,
      "step": 13866
    },
    {
      "epoch": 4.3785917271866115,
      "grad_norm": 4.14034366607666,
      "learning_rate": 6.9047786120272265e-06,
      "loss": 0.6942,
      "step": 13867
    },
    {
      "epoch": 4.378907483422798,
      "grad_norm": 3.926957130432129,
      "learning_rate": 6.901270086309733e-06,
      "loss": 0.6655,
      "step": 13868
    },
    {
      "epoch": 4.379223239658983,
      "grad_norm": 4.638382911682129,
      "learning_rate": 6.897761560592239e-06,
      "loss": 0.5421,
      "step": 13869
    },
    {
      "epoch": 4.379538995895169,
      "grad_norm": 3.2870426177978516,
      "learning_rate": 6.894253034874745e-06,
      "loss": 0.5326,
      "step": 13870
    },
    {
      "epoch": 4.3798547521313544,
      "grad_norm": 4.210941314697266,
      "learning_rate": 6.890744509157253e-06,
      "loss": 0.643,
      "step": 13871
    },
    {
      "epoch": 4.380170508367541,
      "grad_norm": 3.623236894607544,
      "learning_rate": 6.887235983439759e-06,
      "loss": 0.5853,
      "step": 13872
    },
    {
      "epoch": 4.380486264603726,
      "grad_norm": 3.8162424564361572,
      "learning_rate": 6.883727457722265e-06,
      "loss": 0.6109,
      "step": 13873
    },
    {
      "epoch": 4.380802020839911,
      "grad_norm": 3.4464142322540283,
      "learning_rate": 6.880218932004772e-06,
      "loss": 0.5999,
      "step": 13874
    },
    {
      "epoch": 4.381117777076097,
      "grad_norm": 3.6691598892211914,
      "learning_rate": 6.876710406287279e-06,
      "loss": 0.5516,
      "step": 13875
    },
    {
      "epoch": 4.381433533312283,
      "grad_norm": 4.474390983581543,
      "learning_rate": 6.873201880569785e-06,
      "loss": 0.6069,
      "step": 13876
    },
    {
      "epoch": 4.381749289548469,
      "grad_norm": 3.582098960876465,
      "learning_rate": 6.869693354852291e-06,
      "loss": 0.5717,
      "step": 13877
    },
    {
      "epoch": 4.382065045784654,
      "grad_norm": 4.348203182220459,
      "learning_rate": 6.866184829134797e-06,
      "loss": 0.6025,
      "step": 13878
    },
    {
      "epoch": 4.38238080202084,
      "grad_norm": 5.3792595863342285,
      "learning_rate": 6.862676303417305e-06,
      "loss": 0.6114,
      "step": 13879
    },
    {
      "epoch": 4.382696558257026,
      "grad_norm": 3.553147554397583,
      "learning_rate": 6.859167777699811e-06,
      "loss": 0.5549,
      "step": 13880
    },
    {
      "epoch": 4.383012314493211,
      "grad_norm": 3.8588690757751465,
      "learning_rate": 6.855659251982318e-06,
      "loss": 0.5409,
      "step": 13881
    },
    {
      "epoch": 4.383328070729397,
      "grad_norm": 3.7743310928344727,
      "learning_rate": 6.852150726264824e-06,
      "loss": 0.6523,
      "step": 13882
    },
    {
      "epoch": 4.383643826965582,
      "grad_norm": 3.973538637161255,
      "learning_rate": 6.8486422005473295e-06,
      "loss": 0.7216,
      "step": 13883
    },
    {
      "epoch": 4.383959583201769,
      "grad_norm": 3.660987138748169,
      "learning_rate": 6.845133674829837e-06,
      "loss": 0.5303,
      "step": 13884
    },
    {
      "epoch": 4.384275339437954,
      "grad_norm": 4.012722015380859,
      "learning_rate": 6.841625149112343e-06,
      "loss": 0.6211,
      "step": 13885
    },
    {
      "epoch": 4.384591095674139,
      "grad_norm": 4.239403247833252,
      "learning_rate": 6.83811662339485e-06,
      "loss": 0.5545,
      "step": 13886
    },
    {
      "epoch": 4.384906851910325,
      "grad_norm": 4.167466640472412,
      "learning_rate": 6.834608097677356e-06,
      "loss": 0.5975,
      "step": 13887
    },
    {
      "epoch": 4.385222608146511,
      "grad_norm": 4.299217224121094,
      "learning_rate": 6.831099571959864e-06,
      "loss": 0.517,
      "step": 13888
    },
    {
      "epoch": 4.385538364382697,
      "grad_norm": 4.23861026763916,
      "learning_rate": 6.8275910462423696e-06,
      "loss": 0.5572,
      "step": 13889
    },
    {
      "epoch": 4.385854120618882,
      "grad_norm": 4.778448104858398,
      "learning_rate": 6.8240825205248755e-06,
      "loss": 0.6512,
      "step": 13890
    },
    {
      "epoch": 4.386169876855067,
      "grad_norm": 3.5765724182128906,
      "learning_rate": 6.8205739948073815e-06,
      "loss": 0.5858,
      "step": 13891
    },
    {
      "epoch": 4.386485633091254,
      "grad_norm": 4.5256547927856445,
      "learning_rate": 6.817065469089889e-06,
      "loss": 0.6202,
      "step": 13892
    },
    {
      "epoch": 4.386801389327439,
      "grad_norm": 3.540769100189209,
      "learning_rate": 6.813556943372396e-06,
      "loss": 0.6121,
      "step": 13893
    },
    {
      "epoch": 4.387117145563625,
      "grad_norm": 4.952817916870117,
      "learning_rate": 6.810048417654902e-06,
      "loss": 0.6064,
      "step": 13894
    },
    {
      "epoch": 4.38743290179981,
      "grad_norm": 3.7730066776275635,
      "learning_rate": 6.806539891937408e-06,
      "loss": 0.6482,
      "step": 13895
    },
    {
      "epoch": 4.387748658035997,
      "grad_norm": 4.799103736877441,
      "learning_rate": 6.803031366219914e-06,
      "loss": 0.5122,
      "step": 13896
    },
    {
      "epoch": 4.388064414272182,
      "grad_norm": 3.73386812210083,
      "learning_rate": 6.7995228405024215e-06,
      "loss": 0.6193,
      "step": 13897
    },
    {
      "epoch": 4.388380170508367,
      "grad_norm": 4.258853912353516,
      "learning_rate": 6.7960143147849275e-06,
      "loss": 0.5274,
      "step": 13898
    },
    {
      "epoch": 4.388695926744553,
      "grad_norm": 3.1401567459106445,
      "learning_rate": 6.792505789067434e-06,
      "loss": 0.4435,
      "step": 13899
    },
    {
      "epoch": 4.389011682980739,
      "grad_norm": 3.395289182662964,
      "learning_rate": 6.78899726334994e-06,
      "loss": 0.59,
      "step": 13900
    },
    {
      "epoch": 4.389327439216925,
      "grad_norm": 3.6812703609466553,
      "learning_rate": 6.785488737632448e-06,
      "loss": 0.685,
      "step": 13901
    },
    {
      "epoch": 4.38964319545311,
      "grad_norm": 3.5006399154663086,
      "learning_rate": 6.781980211914954e-06,
      "loss": 0.5835,
      "step": 13902
    },
    {
      "epoch": 4.389958951689296,
      "grad_norm": 4.172393798828125,
      "learning_rate": 6.77847168619746e-06,
      "loss": 0.5273,
      "step": 13903
    },
    {
      "epoch": 4.390274707925482,
      "grad_norm": 4.149204254150391,
      "learning_rate": 6.774963160479967e-06,
      "loss": 0.6135,
      "step": 13904
    },
    {
      "epoch": 4.390590464161667,
      "grad_norm": 3.5979180335998535,
      "learning_rate": 6.7714546347624735e-06,
      "loss": 0.594,
      "step": 13905
    },
    {
      "epoch": 4.390906220397853,
      "grad_norm": 5.125931739807129,
      "learning_rate": 6.76794610904498e-06,
      "loss": 0.6329,
      "step": 13906
    },
    {
      "epoch": 4.391221976634038,
      "grad_norm": 3.7344164848327637,
      "learning_rate": 6.764437583327486e-06,
      "loss": 0.4824,
      "step": 13907
    },
    {
      "epoch": 4.3915377328702245,
      "grad_norm": 3.1013622283935547,
      "learning_rate": 6.760929057609992e-06,
      "loss": 0.5044,
      "step": 13908
    },
    {
      "epoch": 4.39185348910641,
      "grad_norm": 3.953991651535034,
      "learning_rate": 6.757420531892498e-06,
      "loss": 0.5678,
      "step": 13909
    },
    {
      "epoch": 4.392169245342595,
      "grad_norm": 4.078730583190918,
      "learning_rate": 6.753912006175006e-06,
      "loss": 0.6667,
      "step": 13910
    },
    {
      "epoch": 4.392485001578781,
      "grad_norm": 3.614490509033203,
      "learning_rate": 6.750403480457512e-06,
      "loss": 0.5594,
      "step": 13911
    },
    {
      "epoch": 4.392800757814967,
      "grad_norm": 4.144519805908203,
      "learning_rate": 6.746894954740019e-06,
      "loss": 0.6171,
      "step": 13912
    },
    {
      "epoch": 4.393116514051153,
      "grad_norm": 3.7098255157470703,
      "learning_rate": 6.743386429022525e-06,
      "loss": 0.5624,
      "step": 13913
    },
    {
      "epoch": 4.393432270287338,
      "grad_norm": 3.7600784301757812,
      "learning_rate": 6.739877903305032e-06,
      "loss": 0.6055,
      "step": 13914
    },
    {
      "epoch": 4.393748026523524,
      "grad_norm": 3.808090925216675,
      "learning_rate": 6.736369377587538e-06,
      "loss": 0.5667,
      "step": 13915
    },
    {
      "epoch": 4.3940637827597095,
      "grad_norm": 3.853174924850464,
      "learning_rate": 6.732860851870044e-06,
      "loss": 0.6219,
      "step": 13916
    },
    {
      "epoch": 4.394379538995895,
      "grad_norm": 4.754101276397705,
      "learning_rate": 6.729352326152551e-06,
      "loss": 0.5863,
      "step": 13917
    },
    {
      "epoch": 4.394695295232081,
      "grad_norm": 3.620459794998169,
      "learning_rate": 6.725843800435057e-06,
      "loss": 0.5572,
      "step": 13918
    },
    {
      "epoch": 4.395011051468266,
      "grad_norm": 4.847146511077881,
      "learning_rate": 6.722335274717565e-06,
      "loss": 0.5751,
      "step": 13919
    },
    {
      "epoch": 4.3953268077044525,
      "grad_norm": 4.918185234069824,
      "learning_rate": 6.718826749000071e-06,
      "loss": 0.6051,
      "step": 13920
    },
    {
      "epoch": 4.395642563940638,
      "grad_norm": 3.5488548278808594,
      "learning_rate": 6.715318223282577e-06,
      "loss": 0.5136,
      "step": 13921
    },
    {
      "epoch": 4.395958320176823,
      "grad_norm": 3.433000326156616,
      "learning_rate": 6.7118096975650825e-06,
      "loss": 0.5651,
      "step": 13922
    },
    {
      "epoch": 4.396274076413009,
      "grad_norm": 4.889477252960205,
      "learning_rate": 6.70830117184759e-06,
      "loss": 0.6868,
      "step": 13923
    },
    {
      "epoch": 4.396589832649195,
      "grad_norm": 2.761348009109497,
      "learning_rate": 6.704792646130097e-06,
      "loss": 0.5034,
      "step": 13924
    },
    {
      "epoch": 4.396905588885381,
      "grad_norm": 3.8364317417144775,
      "learning_rate": 6.701284120412603e-06,
      "loss": 0.5762,
      "step": 13925
    },
    {
      "epoch": 4.397221345121566,
      "grad_norm": 4.013941764831543,
      "learning_rate": 6.697775594695109e-06,
      "loss": 0.6074,
      "step": 13926
    },
    {
      "epoch": 4.397537101357752,
      "grad_norm": 4.620070934295654,
      "learning_rate": 6.694267068977617e-06,
      "loss": 0.607,
      "step": 13927
    },
    {
      "epoch": 4.3978528575939375,
      "grad_norm": 3.6870651245117188,
      "learning_rate": 6.6907585432601226e-06,
      "loss": 0.5678,
      "step": 13928
    },
    {
      "epoch": 4.398168613830123,
      "grad_norm": 3.4185891151428223,
      "learning_rate": 6.6872500175426285e-06,
      "loss": 0.5166,
      "step": 13929
    },
    {
      "epoch": 4.398484370066309,
      "grad_norm": 3.3807454109191895,
      "learning_rate": 6.683741491825135e-06,
      "loss": 0.5137,
      "step": 13930
    },
    {
      "epoch": 4.398800126302494,
      "grad_norm": 4.36657190322876,
      "learning_rate": 6.680232966107641e-06,
      "loss": 0.5788,
      "step": 13931
    },
    {
      "epoch": 4.3991158825386805,
      "grad_norm": 4.204716205596924,
      "learning_rate": 6.676724440390149e-06,
      "loss": 0.614,
      "step": 13932
    },
    {
      "epoch": 4.399431638774866,
      "grad_norm": 3.709012031555176,
      "learning_rate": 6.673215914672655e-06,
      "loss": 0.6176,
      "step": 13933
    },
    {
      "epoch": 4.399747395011051,
      "grad_norm": 3.8040380477905273,
      "learning_rate": 6.669707388955161e-06,
      "loss": 0.6285,
      "step": 13934
    },
    {
      "epoch": 4.400063151247237,
      "grad_norm": 3.73775053024292,
      "learning_rate": 6.666198863237668e-06,
      "loss": 0.5728,
      "step": 13935
    },
    {
      "epoch": 4.4003789074834225,
      "grad_norm": 4.070060729980469,
      "learning_rate": 6.6626903375201745e-06,
      "loss": 0.662,
      "step": 13936
    },
    {
      "epoch": 4.400694663719609,
      "grad_norm": 4.453070163726807,
      "learning_rate": 6.659181811802681e-06,
      "loss": 0.5732,
      "step": 13937
    },
    {
      "epoch": 4.401010419955794,
      "grad_norm": 4.945810794830322,
      "learning_rate": 6.655673286085187e-06,
      "loss": 0.6054,
      "step": 13938
    },
    {
      "epoch": 4.40132617619198,
      "grad_norm": 4.056966781616211,
      "learning_rate": 6.652164760367693e-06,
      "loss": 0.5389,
      "step": 13939
    },
    {
      "epoch": 4.4016419324281655,
      "grad_norm": 3.4394664764404297,
      "learning_rate": 6.648656234650201e-06,
      "loss": 0.6715,
      "step": 13940
    },
    {
      "epoch": 4.401957688664351,
      "grad_norm": 3.4986536502838135,
      "learning_rate": 6.645147708932707e-06,
      "loss": 0.64,
      "step": 13941
    },
    {
      "epoch": 4.402273444900537,
      "grad_norm": 4.712590217590332,
      "learning_rate": 6.641639183215214e-06,
      "loss": 0.6166,
      "step": 13942
    },
    {
      "epoch": 4.402589201136722,
      "grad_norm": 4.404989719390869,
      "learning_rate": 6.63813065749772e-06,
      "loss": 0.5847,
      "step": 13943
    },
    {
      "epoch": 4.402904957372908,
      "grad_norm": 3.8112220764160156,
      "learning_rate": 6.634622131780226e-06,
      "loss": 0.5827,
      "step": 13944
    },
    {
      "epoch": 4.403220713609094,
      "grad_norm": 4.159569263458252,
      "learning_rate": 6.631113606062733e-06,
      "loss": 0.5995,
      "step": 13945
    },
    {
      "epoch": 4.40353646984528,
      "grad_norm": 3.8278074264526367,
      "learning_rate": 6.627605080345239e-06,
      "loss": 0.5728,
      "step": 13946
    },
    {
      "epoch": 4.403852226081465,
      "grad_norm": 4.252699375152588,
      "learning_rate": 6.624096554627745e-06,
      "loss": 0.6047,
      "step": 13947
    },
    {
      "epoch": 4.4041679823176505,
      "grad_norm": 3.491117238998413,
      "learning_rate": 6.620588028910252e-06,
      "loss": 0.5132,
      "step": 13948
    },
    {
      "epoch": 4.404483738553837,
      "grad_norm": 3.9515044689178467,
      "learning_rate": 6.61707950319276e-06,
      "loss": 0.5644,
      "step": 13949
    },
    {
      "epoch": 4.404799494790022,
      "grad_norm": 4.239861965179443,
      "learning_rate": 6.613570977475266e-06,
      "loss": 0.6655,
      "step": 13950
    },
    {
      "epoch": 4.405115251026208,
      "grad_norm": 4.092355251312256,
      "learning_rate": 6.610062451757772e-06,
      "loss": 0.5644,
      "step": 13951
    },
    {
      "epoch": 4.4054310072623935,
      "grad_norm": 3.7808399200439453,
      "learning_rate": 6.606553926040278e-06,
      "loss": 0.6884,
      "step": 13952
    },
    {
      "epoch": 4.405746763498579,
      "grad_norm": 3.4375836849212646,
      "learning_rate": 6.6030454003227844e-06,
      "loss": 0.4918,
      "step": 13953
    },
    {
      "epoch": 4.406062519734765,
      "grad_norm": 4.429372310638428,
      "learning_rate": 6.599536874605291e-06,
      "loss": 0.6999,
      "step": 13954
    },
    {
      "epoch": 4.40637827597095,
      "grad_norm": 4.254056453704834,
      "learning_rate": 6.596028348887798e-06,
      "loss": 0.5189,
      "step": 13955
    },
    {
      "epoch": 4.406694032207136,
      "grad_norm": 2.9879469871520996,
      "learning_rate": 6.592519823170304e-06,
      "loss": 0.5112,
      "step": 13956
    },
    {
      "epoch": 4.407009788443322,
      "grad_norm": 3.9440841674804688,
      "learning_rate": 6.58901129745281e-06,
      "loss": 0.4724,
      "step": 13957
    },
    {
      "epoch": 4.407325544679507,
      "grad_norm": 3.5521464347839355,
      "learning_rate": 6.585502771735318e-06,
      "loss": 0.531,
      "step": 13958
    },
    {
      "epoch": 4.407641300915693,
      "grad_norm": 3.6158154010772705,
      "learning_rate": 6.581994246017824e-06,
      "loss": 0.5256,
      "step": 13959
    },
    {
      "epoch": 4.4079570571518785,
      "grad_norm": 3.6433587074279785,
      "learning_rate": 6.57848572030033e-06,
      "loss": 0.4905,
      "step": 13960
    },
    {
      "epoch": 4.408272813388065,
      "grad_norm": 3.228469133377075,
      "learning_rate": 6.574977194582836e-06,
      "loss": 0.558,
      "step": 13961
    },
    {
      "epoch": 4.40858856962425,
      "grad_norm": 3.495993137359619,
      "learning_rate": 6.571468668865344e-06,
      "loss": 0.5598,
      "step": 13962
    },
    {
      "epoch": 4.408904325860436,
      "grad_norm": 3.145636558532715,
      "learning_rate": 6.56796014314785e-06,
      "loss": 0.5272,
      "step": 13963
    },
    {
      "epoch": 4.409220082096621,
      "grad_norm": 4.370978832244873,
      "learning_rate": 6.564451617430356e-06,
      "loss": 0.5713,
      "step": 13964
    },
    {
      "epoch": 4.409535838332807,
      "grad_norm": 3.7091376781463623,
      "learning_rate": 6.560943091712862e-06,
      "loss": 0.5376,
      "step": 13965
    },
    {
      "epoch": 4.409851594568993,
      "grad_norm": 4.694751262664795,
      "learning_rate": 6.557434565995369e-06,
      "loss": 0.6514,
      "step": 13966
    },
    {
      "epoch": 4.410167350805178,
      "grad_norm": 4.241053581237793,
      "learning_rate": 6.5539260402778756e-06,
      "loss": 0.592,
      "step": 13967
    },
    {
      "epoch": 4.410483107041364,
      "grad_norm": 3.937969207763672,
      "learning_rate": 6.550417514560382e-06,
      "loss": 0.6249,
      "step": 13968
    },
    {
      "epoch": 4.41079886327755,
      "grad_norm": 4.2025322914123535,
      "learning_rate": 6.546908988842888e-06,
      "loss": 0.6627,
      "step": 13969
    },
    {
      "epoch": 4.411114619513736,
      "grad_norm": 4.058900833129883,
      "learning_rate": 6.543400463125394e-06,
      "loss": 0.5364,
      "step": 13970
    },
    {
      "epoch": 4.411430375749921,
      "grad_norm": 4.047272682189941,
      "learning_rate": 6.539891937407902e-06,
      "loss": 0.6448,
      "step": 13971
    },
    {
      "epoch": 4.411746131986106,
      "grad_norm": 4.324413299560547,
      "learning_rate": 6.536383411690408e-06,
      "loss": 0.5737,
      "step": 13972
    },
    {
      "epoch": 4.412061888222293,
      "grad_norm": 4.630064487457275,
      "learning_rate": 6.532874885972915e-06,
      "loss": 0.6152,
      "step": 13973
    },
    {
      "epoch": 4.412377644458478,
      "grad_norm": 3.513805389404297,
      "learning_rate": 6.529366360255421e-06,
      "loss": 0.4723,
      "step": 13974
    },
    {
      "epoch": 4.412693400694664,
      "grad_norm": 4.47792911529541,
      "learning_rate": 6.525857834537928e-06,
      "loss": 0.588,
      "step": 13975
    },
    {
      "epoch": 4.413009156930849,
      "grad_norm": 4.4538798332214355,
      "learning_rate": 6.522349308820434e-06,
      "loss": 0.6189,
      "step": 13976
    },
    {
      "epoch": 4.413324913167035,
      "grad_norm": 3.2687602043151855,
      "learning_rate": 6.51884078310294e-06,
      "loss": 0.5626,
      "step": 13977
    },
    {
      "epoch": 4.413640669403221,
      "grad_norm": 4.974945068359375,
      "learning_rate": 6.515332257385446e-06,
      "loss": 0.7233,
      "step": 13978
    },
    {
      "epoch": 4.413956425639406,
      "grad_norm": 4.341612815856934,
      "learning_rate": 6.511823731667953e-06,
      "loss": 0.6556,
      "step": 13979
    },
    {
      "epoch": 4.414272181875592,
      "grad_norm": 3.9882891178131104,
      "learning_rate": 6.508315205950461e-06,
      "loss": 0.5745,
      "step": 13980
    },
    {
      "epoch": 4.414587938111778,
      "grad_norm": 3.645519495010376,
      "learning_rate": 6.504806680232967e-06,
      "loss": 0.5188,
      "step": 13981
    },
    {
      "epoch": 4.414903694347963,
      "grad_norm": 4.379815101623535,
      "learning_rate": 6.501298154515473e-06,
      "loss": 0.6236,
      "step": 13982
    },
    {
      "epoch": 4.415219450584149,
      "grad_norm": 3.9360995292663574,
      "learning_rate": 6.497789628797979e-06,
      "loss": 0.5691,
      "step": 13983
    },
    {
      "epoch": 4.415535206820334,
      "grad_norm": 3.9962215423583984,
      "learning_rate": 6.494281103080486e-06,
      "loss": 0.5292,
      "step": 13984
    },
    {
      "epoch": 4.415850963056521,
      "grad_norm": 4.38306999206543,
      "learning_rate": 6.490772577362992e-06,
      "loss": 0.5761,
      "step": 13985
    },
    {
      "epoch": 4.416166719292706,
      "grad_norm": 4.476346015930176,
      "learning_rate": 6.487264051645499e-06,
      "loss": 0.5444,
      "step": 13986
    },
    {
      "epoch": 4.416482475528892,
      "grad_norm": 4.080635070800781,
      "learning_rate": 6.483755525928005e-06,
      "loss": 0.6284,
      "step": 13987
    },
    {
      "epoch": 4.416798231765077,
      "grad_norm": 3.8484554290771484,
      "learning_rate": 6.480247000210511e-06,
      "loss": 0.5969,
      "step": 13988
    },
    {
      "epoch": 4.417113988001263,
      "grad_norm": 3.7474381923675537,
      "learning_rate": 6.476738474493019e-06,
      "loss": 0.5933,
      "step": 13989
    },
    {
      "epoch": 4.417429744237449,
      "grad_norm": 4.497233867645264,
      "learning_rate": 6.473229948775525e-06,
      "loss": 0.5303,
      "step": 13990
    },
    {
      "epoch": 4.417745500473634,
      "grad_norm": 4.4034857749938965,
      "learning_rate": 6.4697214230580315e-06,
      "loss": 0.5233,
      "step": 13991
    },
    {
      "epoch": 4.41806125670982,
      "grad_norm": 3.9659335613250732,
      "learning_rate": 6.4662128973405374e-06,
      "loss": 0.6721,
      "step": 13992
    },
    {
      "epoch": 4.418377012946006,
      "grad_norm": 3.9101157188415527,
      "learning_rate": 6.462704371623045e-06,
      "loss": 0.5619,
      "step": 13993
    },
    {
      "epoch": 4.418692769182192,
      "grad_norm": 3.945535659790039,
      "learning_rate": 6.459195845905551e-06,
      "loss": 0.6313,
      "step": 13994
    },
    {
      "epoch": 4.419008525418377,
      "grad_norm": 4.779336452484131,
      "learning_rate": 6.455687320188057e-06,
      "loss": 0.6896,
      "step": 13995
    },
    {
      "epoch": 4.419324281654562,
      "grad_norm": 3.928875684738159,
      "learning_rate": 6.452178794470563e-06,
      "loss": 0.6042,
      "step": 13996
    },
    {
      "epoch": 4.4196400378907486,
      "grad_norm": 2.9767520427703857,
      "learning_rate": 6.448670268753071e-06,
      "loss": 0.535,
      "step": 13997
    },
    {
      "epoch": 4.419955794126934,
      "grad_norm": 3.4739787578582764,
      "learning_rate": 6.4451617430355775e-06,
      "loss": 0.499,
      "step": 13998
    },
    {
      "epoch": 4.42027155036312,
      "grad_norm": 4.174763202667236,
      "learning_rate": 6.4416532173180834e-06,
      "loss": 0.5741,
      "step": 13999
    },
    {
      "epoch": 4.420587306599305,
      "grad_norm": 3.900331735610962,
      "learning_rate": 6.438144691600589e-06,
      "loss": 0.5707,
      "step": 14000
    },
    {
      "epoch": 4.420587306599305,
      "eval_loss": 0.840876579284668,
      "eval_runtime": 4.5667,
      "eval_samples_per_second": 545.254,
      "eval_steps_per_second": 4.599,
      "step": 14000
    },
    {
      "epoch": 4.420903062835491,
      "grad_norm": 4.262493133544922,
      "learning_rate": 6.434636165883095e-06,
      "loss": 0.6851,
      "step": 14001
    },
    {
      "epoch": 4.421218819071677,
      "grad_norm": 4.134298801422119,
      "learning_rate": 6.431127640165603e-06,
      "loss": 0.5689,
      "step": 14002
    },
    {
      "epoch": 4.421534575307862,
      "grad_norm": 4.418160438537598,
      "learning_rate": 6.427619114448109e-06,
      "loss": 0.5842,
      "step": 14003
    },
    {
      "epoch": 4.421850331544048,
      "grad_norm": 3.4833719730377197,
      "learning_rate": 6.424110588730616e-06,
      "loss": 0.5978,
      "step": 14004
    },
    {
      "epoch": 4.422166087780234,
      "grad_norm": 5.079946994781494,
      "learning_rate": 6.420602063013122e-06,
      "loss": 0.6596,
      "step": 14005
    },
    {
      "epoch": 4.422481844016419,
      "grad_norm": 3.388326644897461,
      "learning_rate": 6.4170935372956294e-06,
      "loss": 0.4932,
      "step": 14006
    },
    {
      "epoch": 4.422797600252605,
      "grad_norm": 6.099284648895264,
      "learning_rate": 6.413585011578135e-06,
      "loss": 0.6853,
      "step": 14007
    },
    {
      "epoch": 4.42311335648879,
      "grad_norm": 4.35194730758667,
      "learning_rate": 6.410076485860641e-06,
      "loss": 0.5665,
      "step": 14008
    },
    {
      "epoch": 4.4234291127249765,
      "grad_norm": 3.044832706451416,
      "learning_rate": 6.406567960143148e-06,
      "loss": 0.5037,
      "step": 14009
    },
    {
      "epoch": 4.423744868961162,
      "grad_norm": 4.251094818115234,
      "learning_rate": 6.403059434425655e-06,
      "loss": 0.5508,
      "step": 14010
    },
    {
      "epoch": 4.424060625197348,
      "grad_norm": 4.5503621101379395,
      "learning_rate": 6.399550908708162e-06,
      "loss": 0.6009,
      "step": 14011
    },
    {
      "epoch": 4.424376381433533,
      "grad_norm": 3.9205124378204346,
      "learning_rate": 6.396042382990668e-06,
      "loss": 0.6814,
      "step": 14012
    },
    {
      "epoch": 4.424692137669719,
      "grad_norm": 4.061188220977783,
      "learning_rate": 6.392533857273174e-06,
      "loss": 0.6931,
      "step": 14013
    },
    {
      "epoch": 4.425007893905905,
      "grad_norm": 3.7340683937072754,
      "learning_rate": 6.38902533155568e-06,
      "loss": 0.5785,
      "step": 14014
    },
    {
      "epoch": 4.42532365014209,
      "grad_norm": 4.012341499328613,
      "learning_rate": 6.385516805838187e-06,
      "loss": 0.4686,
      "step": 14015
    },
    {
      "epoch": 4.425639406378276,
      "grad_norm": 3.6157917976379395,
      "learning_rate": 6.382008280120693e-06,
      "loss": 0.6235,
      "step": 14016
    },
    {
      "epoch": 4.4259551626144615,
      "grad_norm": 3.9349396228790283,
      "learning_rate": 6.3784997544032e-06,
      "loss": 0.5991,
      "step": 14017
    },
    {
      "epoch": 4.426270918850648,
      "grad_norm": 3.7923245429992676,
      "learning_rate": 6.374991228685706e-06,
      "loss": 0.5057,
      "step": 14018
    },
    {
      "epoch": 4.426586675086833,
      "grad_norm": 3.3391361236572266,
      "learning_rate": 6.371482702968214e-06,
      "loss": 0.5575,
      "step": 14019
    },
    {
      "epoch": 4.426902431323018,
      "grad_norm": 3.5642879009246826,
      "learning_rate": 6.36797417725072e-06,
      "loss": 0.5649,
      "step": 14020
    },
    {
      "epoch": 4.4272181875592045,
      "grad_norm": 3.14259934425354,
      "learning_rate": 6.364465651533226e-06,
      "loss": 0.5366,
      "step": 14021
    },
    {
      "epoch": 4.42753394379539,
      "grad_norm": 3.413356065750122,
      "learning_rate": 6.3609571258157325e-06,
      "loss": 0.5285,
      "step": 14022
    },
    {
      "epoch": 4.427849700031576,
      "grad_norm": 3.726457118988037,
      "learning_rate": 6.3574486000982385e-06,
      "loss": 0.6283,
      "step": 14023
    },
    {
      "epoch": 4.428165456267761,
      "grad_norm": 4.154478073120117,
      "learning_rate": 6.353940074380746e-06,
      "loss": 0.5759,
      "step": 14024
    },
    {
      "epoch": 4.4284812125039466,
      "grad_norm": 3.93721866607666,
      "learning_rate": 6.350431548663252e-06,
      "loss": 0.6053,
      "step": 14025
    },
    {
      "epoch": 4.428796968740133,
      "grad_norm": 4.629896640777588,
      "learning_rate": 6.346923022945758e-06,
      "loss": 0.6498,
      "step": 14026
    },
    {
      "epoch": 4.429112724976318,
      "grad_norm": 3.8559741973876953,
      "learning_rate": 6.343414497228264e-06,
      "loss": 0.4948,
      "step": 14027
    },
    {
      "epoch": 4.429428481212504,
      "grad_norm": 3.3787600994110107,
      "learning_rate": 6.339905971510772e-06,
      "loss": 0.5996,
      "step": 14028
    },
    {
      "epoch": 4.4297442374486895,
      "grad_norm": 4.007665634155273,
      "learning_rate": 6.3363974457932785e-06,
      "loss": 0.5198,
      "step": 14029
    },
    {
      "epoch": 4.430059993684876,
      "grad_norm": 3.144374370574951,
      "learning_rate": 6.3328889200757845e-06,
      "loss": 0.5039,
      "step": 14030
    },
    {
      "epoch": 4.430375749921061,
      "grad_norm": 3.773310422897339,
      "learning_rate": 6.3293803943582904e-06,
      "loss": 0.6339,
      "step": 14031
    },
    {
      "epoch": 4.430691506157246,
      "grad_norm": 3.3960702419281006,
      "learning_rate": 6.325871868640798e-06,
      "loss": 0.546,
      "step": 14032
    },
    {
      "epoch": 4.4310072623934325,
      "grad_norm": 4.534151554107666,
      "learning_rate": 6.322363342923304e-06,
      "loss": 0.608,
      "step": 14033
    },
    {
      "epoch": 4.431323018629618,
      "grad_norm": 3.98264479637146,
      "learning_rate": 6.31885481720581e-06,
      "loss": 0.5109,
      "step": 14034
    },
    {
      "epoch": 4.431638774865804,
      "grad_norm": 4.405881881713867,
      "learning_rate": 6.315346291488317e-06,
      "loss": 0.5618,
      "step": 14035
    },
    {
      "epoch": 4.431954531101989,
      "grad_norm": 4.589343070983887,
      "learning_rate": 6.311837765770823e-06,
      "loss": 0.6537,
      "step": 14036
    },
    {
      "epoch": 4.4322702873381745,
      "grad_norm": 3.855437755584717,
      "learning_rate": 6.3083292400533305e-06,
      "loss": 0.6597,
      "step": 14037
    },
    {
      "epoch": 4.432586043574361,
      "grad_norm": 3.812405586242676,
      "learning_rate": 6.3048207143358364e-06,
      "loss": 0.6282,
      "step": 14038
    },
    {
      "epoch": 4.432901799810546,
      "grad_norm": 4.213221073150635,
      "learning_rate": 6.301312188618342e-06,
      "loss": 0.5903,
      "step": 14039
    },
    {
      "epoch": 4.433217556046732,
      "grad_norm": 4.055720806121826,
      "learning_rate": 6.297803662900849e-06,
      "loss": 0.5651,
      "step": 14040
    },
    {
      "epoch": 4.4335333122829175,
      "grad_norm": 3.9310083389282227,
      "learning_rate": 6.294295137183356e-06,
      "loss": 0.5183,
      "step": 14041
    },
    {
      "epoch": 4.433849068519104,
      "grad_norm": 5.136168479919434,
      "learning_rate": 6.290786611465863e-06,
      "loss": 0.5254,
      "step": 14042
    },
    {
      "epoch": 4.434164824755289,
      "grad_norm": 3.6717300415039062,
      "learning_rate": 6.287278085748369e-06,
      "loss": 0.6181,
      "step": 14043
    },
    {
      "epoch": 4.434480580991474,
      "grad_norm": 3.420698642730713,
      "learning_rate": 6.283769560030875e-06,
      "loss": 0.5859,
      "step": 14044
    },
    {
      "epoch": 4.43479633722766,
      "grad_norm": 3.303269624710083,
      "learning_rate": 6.2802610343133824e-06,
      "loss": 0.5214,
      "step": 14045
    },
    {
      "epoch": 4.435112093463846,
      "grad_norm": 3.7506959438323975,
      "learning_rate": 6.276752508595888e-06,
      "loss": 0.6245,
      "step": 14046
    },
    {
      "epoch": 4.435427849700032,
      "grad_norm": 3.6569509506225586,
      "learning_rate": 6.273243982878395e-06,
      "loss": 0.641,
      "step": 14047
    },
    {
      "epoch": 4.435743605936217,
      "grad_norm": 4.636500835418701,
      "learning_rate": 6.269735457160901e-06,
      "loss": 0.5827,
      "step": 14048
    },
    {
      "epoch": 4.4360593621724025,
      "grad_norm": 4.752800941467285,
      "learning_rate": 6.266226931443407e-06,
      "loss": 0.679,
      "step": 14049
    },
    {
      "epoch": 4.436375118408589,
      "grad_norm": 3.897911787033081,
      "learning_rate": 6.262718405725915e-06,
      "loss": 0.6093,
      "step": 14050
    },
    {
      "epoch": 4.436690874644774,
      "grad_norm": 4.957756996154785,
      "learning_rate": 6.259209880008421e-06,
      "loss": 0.5847,
      "step": 14051
    },
    {
      "epoch": 4.43700663088096,
      "grad_norm": 5.22084903717041,
      "learning_rate": 6.255701354290927e-06,
      "loss": 0.4735,
      "step": 14052
    },
    {
      "epoch": 4.437322387117145,
      "grad_norm": 4.089072227478027,
      "learning_rate": 6.2521928285734335e-06,
      "loss": 0.5782,
      "step": 14053
    },
    {
      "epoch": 4.437638143353332,
      "grad_norm": 4.290731906890869,
      "learning_rate": 6.24868430285594e-06,
      "loss": 0.5999,
      "step": 14054
    },
    {
      "epoch": 4.437953899589517,
      "grad_norm": 3.8810765743255615,
      "learning_rate": 6.245175777138447e-06,
      "loss": 0.556,
      "step": 14055
    },
    {
      "epoch": 4.438269655825702,
      "grad_norm": 3.96453857421875,
      "learning_rate": 6.241667251420953e-06,
      "loss": 0.6572,
      "step": 14056
    },
    {
      "epoch": 4.438585412061888,
      "grad_norm": 3.5164849758148193,
      "learning_rate": 6.23815872570346e-06,
      "loss": 0.6149,
      "step": 14057
    },
    {
      "epoch": 4.438901168298074,
      "grad_norm": 3.363508939743042,
      "learning_rate": 6.234650199985966e-06,
      "loss": 0.5147,
      "step": 14058
    },
    {
      "epoch": 4.43921692453426,
      "grad_norm": 4.452375411987305,
      "learning_rate": 6.231141674268473e-06,
      "loss": 0.5638,
      "step": 14059
    },
    {
      "epoch": 4.439532680770445,
      "grad_norm": 3.9452342987060547,
      "learning_rate": 6.2276331485509795e-06,
      "loss": 0.6003,
      "step": 14060
    },
    {
      "epoch": 4.439848437006631,
      "grad_norm": 4.265529155731201,
      "learning_rate": 6.2241246228334855e-06,
      "loss": 0.5471,
      "step": 14061
    },
    {
      "epoch": 4.440164193242817,
      "grad_norm": 4.435832977294922,
      "learning_rate": 6.220616097115992e-06,
      "loss": 0.6372,
      "step": 14062
    },
    {
      "epoch": 4.440479949479002,
      "grad_norm": 3.9483518600463867,
      "learning_rate": 6.217107571398498e-06,
      "loss": 0.5413,
      "step": 14063
    },
    {
      "epoch": 4.440795705715188,
      "grad_norm": 4.160062313079834,
      "learning_rate": 6.213599045681005e-06,
      "loss": 0.5565,
      "step": 14064
    },
    {
      "epoch": 4.441111461951373,
      "grad_norm": 4.6746110916137695,
      "learning_rate": 6.210090519963512e-06,
      "loss": 0.5184,
      "step": 14065
    },
    {
      "epoch": 4.44142721818756,
      "grad_norm": 3.4372005462646484,
      "learning_rate": 6.206581994246019e-06,
      "loss": 0.5828,
      "step": 14066
    },
    {
      "epoch": 4.441742974423745,
      "grad_norm": 5.3773651123046875,
      "learning_rate": 6.203073468528525e-06,
      "loss": 0.6271,
      "step": 14067
    },
    {
      "epoch": 4.44205873065993,
      "grad_norm": 3.862733840942383,
      "learning_rate": 6.1995649428110315e-06,
      "loss": 0.625,
      "step": 14068
    },
    {
      "epoch": 4.442374486896116,
      "grad_norm": 3.756121873855591,
      "learning_rate": 6.1960564170935375e-06,
      "loss": 0.5749,
      "step": 14069
    },
    {
      "epoch": 4.442690243132302,
      "grad_norm": 3.8409175872802734,
      "learning_rate": 6.1925478913760434e-06,
      "loss": 0.6443,
      "step": 14070
    },
    {
      "epoch": 4.443005999368488,
      "grad_norm": 3.363300323486328,
      "learning_rate": 6.18903936565855e-06,
      "loss": 0.6371,
      "step": 14071
    },
    {
      "epoch": 4.443321755604673,
      "grad_norm": 4.161463260650635,
      "learning_rate": 6.185530839941057e-06,
      "loss": 0.5705,
      "step": 14072
    },
    {
      "epoch": 4.443637511840858,
      "grad_norm": 3.8517725467681885,
      "learning_rate": 6.182022314223564e-06,
      "loss": 0.6237,
      "step": 14073
    },
    {
      "epoch": 4.443953268077045,
      "grad_norm": 4.022441387176514,
      "learning_rate": 6.17851378850607e-06,
      "loss": 0.6156,
      "step": 14074
    },
    {
      "epoch": 4.44426902431323,
      "grad_norm": 4.498863697052002,
      "learning_rate": 6.175005262788577e-06,
      "loss": 0.6147,
      "step": 14075
    },
    {
      "epoch": 4.444584780549416,
      "grad_norm": 3.9183080196380615,
      "learning_rate": 6.171496737071083e-06,
      "loss": 0.5609,
      "step": 14076
    },
    {
      "epoch": 4.444900536785601,
      "grad_norm": 4.746714115142822,
      "learning_rate": 6.1679882113535894e-06,
      "loss": 0.6543,
      "step": 14077
    },
    {
      "epoch": 4.445216293021788,
      "grad_norm": 4.065096378326416,
      "learning_rate": 6.164479685636096e-06,
      "loss": 0.6038,
      "step": 14078
    },
    {
      "epoch": 4.445532049257973,
      "grad_norm": 4.259418964385986,
      "learning_rate": 6.160971159918603e-06,
      "loss": 0.6817,
      "step": 14079
    },
    {
      "epoch": 4.445847805494158,
      "grad_norm": 4.111633777618408,
      "learning_rate": 6.157462634201109e-06,
      "loss": 0.5918,
      "step": 14080
    },
    {
      "epoch": 4.446163561730344,
      "grad_norm": 3.767512083053589,
      "learning_rate": 6.153954108483616e-06,
      "loss": 0.5658,
      "step": 14081
    },
    {
      "epoch": 4.44647931796653,
      "grad_norm": 3.3376805782318115,
      "learning_rate": 6.150445582766122e-06,
      "loss": 0.553,
      "step": 14082
    },
    {
      "epoch": 4.446795074202716,
      "grad_norm": 4.174273490905762,
      "learning_rate": 6.146937057048628e-06,
      "loss": 0.6398,
      "step": 14083
    },
    {
      "epoch": 4.447110830438901,
      "grad_norm": 4.992193698883057,
      "learning_rate": 6.1434285313311354e-06,
      "loss": 0.5941,
      "step": 14084
    },
    {
      "epoch": 4.447426586675087,
      "grad_norm": 3.6890101432800293,
      "learning_rate": 6.139920005613641e-06,
      "loss": 0.545,
      "step": 14085
    },
    {
      "epoch": 4.447742342911273,
      "grad_norm": 4.438060760498047,
      "learning_rate": 6.136411479896148e-06,
      "loss": 0.6309,
      "step": 14086
    },
    {
      "epoch": 4.448058099147458,
      "grad_norm": 3.5384247303009033,
      "learning_rate": 6.132902954178654e-06,
      "loss": 0.5243,
      "step": 14087
    },
    {
      "epoch": 4.448373855383644,
      "grad_norm": 3.8756277561187744,
      "learning_rate": 6.129394428461161e-06,
      "loss": 0.5229,
      "step": 14088
    },
    {
      "epoch": 4.448689611619829,
      "grad_norm": 3.77768874168396,
      "learning_rate": 6.125885902743667e-06,
      "loss": 0.6194,
      "step": 14089
    },
    {
      "epoch": 4.4490053678560155,
      "grad_norm": 3.7471566200256348,
      "learning_rate": 6.122377377026174e-06,
      "loss": 0.6546,
      "step": 14090
    },
    {
      "epoch": 4.449321124092201,
      "grad_norm": 3.7231853008270264,
      "learning_rate": 6.118868851308681e-06,
      "loss": 0.5497,
      "step": 14091
    },
    {
      "epoch": 4.449636880328386,
      "grad_norm": 3.7262866497039795,
      "learning_rate": 6.115360325591187e-06,
      "loss": 0.5906,
      "step": 14092
    },
    {
      "epoch": 4.449952636564572,
      "grad_norm": 3.8434207439422607,
      "learning_rate": 6.111851799873693e-06,
      "loss": 0.5769,
      "step": 14093
    },
    {
      "epoch": 4.450268392800758,
      "grad_norm": 4.075684547424316,
      "learning_rate": 6.108343274156199e-06,
      "loss": 0.5499,
      "step": 14094
    },
    {
      "epoch": 4.450584149036944,
      "grad_norm": 4.475438594818115,
      "learning_rate": 6.104834748438706e-06,
      "loss": 0.6283,
      "step": 14095
    },
    {
      "epoch": 4.450899905273129,
      "grad_norm": 4.034211158752441,
      "learning_rate": 6.101326222721213e-06,
      "loss": 0.5862,
      "step": 14096
    },
    {
      "epoch": 4.451215661509314,
      "grad_norm": 3.630013942718506,
      "learning_rate": 6.09781769700372e-06,
      "loss": 0.5724,
      "step": 14097
    },
    {
      "epoch": 4.4515314177455005,
      "grad_norm": 3.966083526611328,
      "learning_rate": 6.094309171286226e-06,
      "loss": 0.6143,
      "step": 14098
    },
    {
      "epoch": 4.451847173981686,
      "grad_norm": 3.976228952407837,
      "learning_rate": 6.0908006455687325e-06,
      "loss": 0.5524,
      "step": 14099
    },
    {
      "epoch": 4.452162930217872,
      "grad_norm": 3.595181465148926,
      "learning_rate": 6.0872921198512385e-06,
      "loss": 0.5757,
      "step": 14100
    },
    {
      "epoch": 4.452478686454057,
      "grad_norm": 4.189726829528809,
      "learning_rate": 6.083783594133745e-06,
      "loss": 0.7044,
      "step": 14101
    },
    {
      "epoch": 4.4527944426902435,
      "grad_norm": 4.484888553619385,
      "learning_rate": 6.080275068416251e-06,
      "loss": 0.6041,
      "step": 14102
    },
    {
      "epoch": 4.453110198926429,
      "grad_norm": 3.9221081733703613,
      "learning_rate": 6.076766542698759e-06,
      "loss": 0.5303,
      "step": 14103
    },
    {
      "epoch": 4.453425955162614,
      "grad_norm": 3.7687273025512695,
      "learning_rate": 6.073258016981265e-06,
      "loss": 0.654,
      "step": 14104
    },
    {
      "epoch": 4.4537417113988,
      "grad_norm": 4.733712196350098,
      "learning_rate": 6.069749491263772e-06,
      "loss": 0.5825,
      "step": 14105
    },
    {
      "epoch": 4.454057467634986,
      "grad_norm": 3.8109660148620605,
      "learning_rate": 6.066240965546278e-06,
      "loss": 0.4975,
      "step": 14106
    },
    {
      "epoch": 4.454373223871172,
      "grad_norm": 5.52777099609375,
      "learning_rate": 6.062732439828784e-06,
      "loss": 0.6085,
      "step": 14107
    },
    {
      "epoch": 4.454688980107357,
      "grad_norm": 3.900256395339966,
      "learning_rate": 6.0592239141112905e-06,
      "loss": 0.5647,
      "step": 14108
    },
    {
      "epoch": 4.455004736343543,
      "grad_norm": 3.914034605026245,
      "learning_rate": 6.055715388393797e-06,
      "loss": 0.4867,
      "step": 14109
    },
    {
      "epoch": 4.4553204925797285,
      "grad_norm": 3.4085018634796143,
      "learning_rate": 6.052206862676304e-06,
      "loss": 0.5163,
      "step": 14110
    },
    {
      "epoch": 4.455636248815914,
      "grad_norm": 4.066985130310059,
      "learning_rate": 6.04869833695881e-06,
      "loss": 0.6256,
      "step": 14111
    },
    {
      "epoch": 4.4559520050521,
      "grad_norm": 4.9976630210876465,
      "learning_rate": 6.045189811241317e-06,
      "loss": 0.6366,
      "step": 14112
    },
    {
      "epoch": 4.456267761288285,
      "grad_norm": 3.205570697784424,
      "learning_rate": 6.041681285523823e-06,
      "loss": 0.6075,
      "step": 14113
    },
    {
      "epoch": 4.4565835175244715,
      "grad_norm": 6.04815149307251,
      "learning_rate": 6.03817275980633e-06,
      "loss": 0.7357,
      "step": 14114
    },
    {
      "epoch": 4.456899273760657,
      "grad_norm": 5.101078510284424,
      "learning_rate": 6.0346642340888365e-06,
      "loss": 0.6208,
      "step": 14115
    },
    {
      "epoch": 4.457215029996842,
      "grad_norm": 3.7468440532684326,
      "learning_rate": 6.031155708371343e-06,
      "loss": 0.5564,
      "step": 14116
    },
    {
      "epoch": 4.457530786233028,
      "grad_norm": 4.029965400695801,
      "learning_rate": 6.027647182653849e-06,
      "loss": 0.5895,
      "step": 14117
    },
    {
      "epoch": 4.4578465424692135,
      "grad_norm": 4.411448955535889,
      "learning_rate": 6.024138656936355e-06,
      "loss": 0.683,
      "step": 14118
    },
    {
      "epoch": 4.4581622987054,
      "grad_norm": 4.548821926116943,
      "learning_rate": 6.020630131218862e-06,
      "loss": 0.6008,
      "step": 14119
    },
    {
      "epoch": 4.458478054941585,
      "grad_norm": 3.7208774089813232,
      "learning_rate": 6.017121605501368e-06,
      "loss": 0.5418,
      "step": 14120
    },
    {
      "epoch": 4.45879381117777,
      "grad_norm": 4.085855007171631,
      "learning_rate": 6.013613079783876e-06,
      "loss": 0.6004,
      "step": 14121
    },
    {
      "epoch": 4.4591095674139565,
      "grad_norm": 4.611964702606201,
      "learning_rate": 6.010104554066382e-06,
      "loss": 0.706,
      "step": 14122
    },
    {
      "epoch": 4.459425323650142,
      "grad_norm": 3.8009860515594482,
      "learning_rate": 6.0065960283488884e-06,
      "loss": 0.6058,
      "step": 14123
    },
    {
      "epoch": 4.459741079886328,
      "grad_norm": 3.976193904876709,
      "learning_rate": 6.003087502631394e-06,
      "loss": 0.6171,
      "step": 14124
    },
    {
      "epoch": 4.460056836122513,
      "grad_norm": 4.203454971313477,
      "learning_rate": 5.999578976913901e-06,
      "loss": 0.6662,
      "step": 14125
    },
    {
      "epoch": 4.460372592358699,
      "grad_norm": 3.6372947692871094,
      "learning_rate": 5.996070451196407e-06,
      "loss": 0.5755,
      "step": 14126
    },
    {
      "epoch": 4.460688348594885,
      "grad_norm": 4.232392311096191,
      "learning_rate": 5.992561925478914e-06,
      "loss": 0.5287,
      "step": 14127
    },
    {
      "epoch": 4.46100410483107,
      "grad_norm": 4.279481887817383,
      "learning_rate": 5.989053399761421e-06,
      "loss": 0.5736,
      "step": 14128
    },
    {
      "epoch": 4.461319861067256,
      "grad_norm": 4.078268051147461,
      "learning_rate": 5.985544874043927e-06,
      "loss": 0.4889,
      "step": 14129
    },
    {
      "epoch": 4.4616356173034415,
      "grad_norm": 3.835078716278076,
      "learning_rate": 5.982036348326434e-06,
      "loss": 0.6074,
      "step": 14130
    },
    {
      "epoch": 4.461951373539628,
      "grad_norm": 4.351586818695068,
      "learning_rate": 5.9785278226089395e-06,
      "loss": 0.6164,
      "step": 14131
    },
    {
      "epoch": 4.462267129775813,
      "grad_norm": 3.795492649078369,
      "learning_rate": 5.975019296891446e-06,
      "loss": 0.5911,
      "step": 14132
    },
    {
      "epoch": 4.462582886011999,
      "grad_norm": 3.8909242153167725,
      "learning_rate": 5.971510771173953e-06,
      "loss": 0.5577,
      "step": 14133
    },
    {
      "epoch": 4.4628986422481844,
      "grad_norm": 3.3730599880218506,
      "learning_rate": 5.96800224545646e-06,
      "loss": 0.5239,
      "step": 14134
    },
    {
      "epoch": 4.46321439848437,
      "grad_norm": 5.440845489501953,
      "learning_rate": 5.964493719738966e-06,
      "loss": 0.5171,
      "step": 14135
    },
    {
      "epoch": 4.463530154720556,
      "grad_norm": 4.043196678161621,
      "learning_rate": 5.960985194021473e-06,
      "loss": 0.6598,
      "step": 14136
    },
    {
      "epoch": 4.463845910956741,
      "grad_norm": 4.90252161026001,
      "learning_rate": 5.957476668303979e-06,
      "loss": 0.5446,
      "step": 14137
    },
    {
      "epoch": 4.464161667192927,
      "grad_norm": 4.0738115310668945,
      "learning_rate": 5.9539681425864855e-06,
      "loss": 0.5536,
      "step": 14138
    },
    {
      "epoch": 4.464477423429113,
      "grad_norm": 4.491530895233154,
      "learning_rate": 5.9504596168689915e-06,
      "loss": 0.5528,
      "step": 14139
    },
    {
      "epoch": 4.464793179665298,
      "grad_norm": 4.959558010101318,
      "learning_rate": 5.946951091151499e-06,
      "loss": 0.6007,
      "step": 14140
    },
    {
      "epoch": 4.465108935901484,
      "grad_norm": 4.034335613250732,
      "learning_rate": 5.943442565434005e-06,
      "loss": 0.6074,
      "step": 14141
    },
    {
      "epoch": 4.4654246921376695,
      "grad_norm": 3.6205875873565674,
      "learning_rate": 5.939934039716511e-06,
      "loss": 0.6127,
      "step": 14142
    },
    {
      "epoch": 4.465740448373856,
      "grad_norm": 3.847388505935669,
      "learning_rate": 5.936425513999018e-06,
      "loss": 0.515,
      "step": 14143
    },
    {
      "epoch": 4.466056204610041,
      "grad_norm": 4.077198505401611,
      "learning_rate": 5.932916988281524e-06,
      "loss": 0.5947,
      "step": 14144
    },
    {
      "epoch": 4.466371960846227,
      "grad_norm": 4.28098726272583,
      "learning_rate": 5.929408462564031e-06,
      "loss": 0.5174,
      "step": 14145
    },
    {
      "epoch": 4.466687717082412,
      "grad_norm": 3.3999505043029785,
      "learning_rate": 5.9258999368465375e-06,
      "loss": 0.629,
      "step": 14146
    },
    {
      "epoch": 4.467003473318598,
      "grad_norm": 3.569457530975342,
      "learning_rate": 5.922391411129044e-06,
      "loss": 0.5468,
      "step": 14147
    },
    {
      "epoch": 4.467319229554784,
      "grad_norm": 3.6818270683288574,
      "learning_rate": 5.91888288541155e-06,
      "loss": 0.5336,
      "step": 14148
    },
    {
      "epoch": 4.467634985790969,
      "grad_norm": 3.8758466243743896,
      "learning_rate": 5.915374359694057e-06,
      "loss": 0.5736,
      "step": 14149
    },
    {
      "epoch": 4.467950742027155,
      "grad_norm": 4.262108325958252,
      "learning_rate": 5.911865833976563e-06,
      "loss": 0.5752,
      "step": 14150
    },
    {
      "epoch": 4.468266498263341,
      "grad_norm": 4.884130477905273,
      "learning_rate": 5.90835730825907e-06,
      "loss": 0.6572,
      "step": 14151
    },
    {
      "epoch": 4.468582254499526,
      "grad_norm": 4.337226867675781,
      "learning_rate": 5.904848782541577e-06,
      "loss": 0.6091,
      "step": 14152
    },
    {
      "epoch": 4.468898010735712,
      "grad_norm": 3.40773344039917,
      "learning_rate": 5.901340256824083e-06,
      "loss": 0.5598,
      "step": 14153
    },
    {
      "epoch": 4.469213766971897,
      "grad_norm": 4.4664154052734375,
      "learning_rate": 5.8978317311065895e-06,
      "loss": 0.5752,
      "step": 14154
    },
    {
      "epoch": 4.469529523208084,
      "grad_norm": 3.8124072551727295,
      "learning_rate": 5.8943232053890954e-06,
      "loss": 0.5855,
      "step": 14155
    },
    {
      "epoch": 4.469845279444269,
      "grad_norm": 4.218572616577148,
      "learning_rate": 5.890814679671602e-06,
      "loss": 0.672,
      "step": 14156
    },
    {
      "epoch": 4.470161035680455,
      "grad_norm": 3.7588162422180176,
      "learning_rate": 5.887306153954108e-06,
      "loss": 0.6203,
      "step": 14157
    },
    {
      "epoch": 4.47047679191664,
      "grad_norm": 5.460412979125977,
      "learning_rate": 5.883797628236615e-06,
      "loss": 0.575,
      "step": 14158
    },
    {
      "epoch": 4.470792548152826,
      "grad_norm": 4.095670223236084,
      "learning_rate": 5.880289102519122e-06,
      "loss": 0.681,
      "step": 14159
    },
    {
      "epoch": 4.471108304389012,
      "grad_norm": 6.307929992675781,
      "learning_rate": 5.876780576801629e-06,
      "loss": 0.6805,
      "step": 14160
    },
    {
      "epoch": 4.471424060625197,
      "grad_norm": 4.340616703033447,
      "learning_rate": 5.873272051084135e-06,
      "loss": 0.6426,
      "step": 14161
    },
    {
      "epoch": 4.471739816861383,
      "grad_norm": 4.744844913482666,
      "learning_rate": 5.8697635253666414e-06,
      "loss": 0.6777,
      "step": 14162
    },
    {
      "epoch": 4.472055573097569,
      "grad_norm": 4.1812920570373535,
      "learning_rate": 5.866254999649147e-06,
      "loss": 0.6853,
      "step": 14163
    },
    {
      "epoch": 4.472371329333754,
      "grad_norm": 3.976661205291748,
      "learning_rate": 5.862746473931654e-06,
      "loss": 0.577,
      "step": 14164
    },
    {
      "epoch": 4.47268708556994,
      "grad_norm": 4.581111907958984,
      "learning_rate": 5.859237948214161e-06,
      "loss": 0.6707,
      "step": 14165
    },
    {
      "epoch": 4.473002841806125,
      "grad_norm": 3.69863224029541,
      "learning_rate": 5.855729422496667e-06,
      "loss": 0.5685,
      "step": 14166
    },
    {
      "epoch": 4.473318598042312,
      "grad_norm": 4.336990833282471,
      "learning_rate": 5.852220896779174e-06,
      "loss": 0.7037,
      "step": 14167
    },
    {
      "epoch": 4.473634354278497,
      "grad_norm": 4.530046463012695,
      "learning_rate": 5.84871237106168e-06,
      "loss": 0.6897,
      "step": 14168
    },
    {
      "epoch": 4.473950110514683,
      "grad_norm": 4.801065444946289,
      "learning_rate": 5.845203845344187e-06,
      "loss": 0.6405,
      "step": 14169
    },
    {
      "epoch": 4.474265866750868,
      "grad_norm": 3.9356281757354736,
      "learning_rate": 5.841695319626693e-06,
      "loss": 0.5873,
      "step": 14170
    },
    {
      "epoch": 4.474581622987054,
      "grad_norm": 5.590521812438965,
      "learning_rate": 5.8381867939092e-06,
      "loss": 0.5646,
      "step": 14171
    },
    {
      "epoch": 4.47489737922324,
      "grad_norm": 4.460421085357666,
      "learning_rate": 5.834678268191706e-06,
      "loss": 0.56,
      "step": 14172
    },
    {
      "epoch": 4.475213135459425,
      "grad_norm": 3.5441787242889404,
      "learning_rate": 5.831169742474213e-06,
      "loss": 0.5667,
      "step": 14173
    },
    {
      "epoch": 4.475528891695611,
      "grad_norm": 4.029053688049316,
      "learning_rate": 5.827661216756719e-06,
      "loss": 0.5691,
      "step": 14174
    },
    {
      "epoch": 4.475844647931797,
      "grad_norm": 3.9714126586914062,
      "learning_rate": 5.824152691039226e-06,
      "loss": 0.5843,
      "step": 14175
    },
    {
      "epoch": 4.476160404167983,
      "grad_norm": 4.987524509429932,
      "learning_rate": 5.820644165321732e-06,
      "loss": 0.6057,
      "step": 14176
    },
    {
      "epoch": 4.476476160404168,
      "grad_norm": 3.820176362991333,
      "learning_rate": 5.8171356396042385e-06,
      "loss": 0.5609,
      "step": 14177
    },
    {
      "epoch": 4.476791916640353,
      "grad_norm": 3.851840019226074,
      "learning_rate": 5.813627113886745e-06,
      "loss": 0.667,
      "step": 14178
    },
    {
      "epoch": 4.4771076728765395,
      "grad_norm": 4.435564041137695,
      "learning_rate": 5.810118588169251e-06,
      "loss": 0.6207,
      "step": 14179
    },
    {
      "epoch": 4.477423429112725,
      "grad_norm": 3.6548027992248535,
      "learning_rate": 5.806610062451758e-06,
      "loss": 0.5553,
      "step": 14180
    },
    {
      "epoch": 4.477739185348911,
      "grad_norm": 4.9721269607543945,
      "learning_rate": 5.803101536734264e-06,
      "loss": 0.5619,
      "step": 14181
    },
    {
      "epoch": 4.478054941585096,
      "grad_norm": 3.578169584274292,
      "learning_rate": 5.799593011016771e-06,
      "loss": 0.5892,
      "step": 14182
    },
    {
      "epoch": 4.478370697821282,
      "grad_norm": 3.1525161266326904,
      "learning_rate": 5.796084485299278e-06,
      "loss": 0.5379,
      "step": 14183
    },
    {
      "epoch": 4.478686454057468,
      "grad_norm": 3.5741238594055176,
      "learning_rate": 5.7925759595817845e-06,
      "loss": 0.5414,
      "step": 14184
    },
    {
      "epoch": 4.479002210293653,
      "grad_norm": 3.761068105697632,
      "learning_rate": 5.7890674338642905e-06,
      "loss": 0.4641,
      "step": 14185
    },
    {
      "epoch": 4.479317966529839,
      "grad_norm": 3.8546342849731445,
      "learning_rate": 5.785558908146797e-06,
      "loss": 0.7594,
      "step": 14186
    },
    {
      "epoch": 4.479633722766025,
      "grad_norm": 3.731111526489258,
      "learning_rate": 5.782050382429303e-06,
      "loss": 0.5355,
      "step": 14187
    },
    {
      "epoch": 4.47994947900221,
      "grad_norm": 4.304540634155273,
      "learning_rate": 5.778541856711809e-06,
      "loss": 0.582,
      "step": 14188
    },
    {
      "epoch": 4.480265235238396,
      "grad_norm": 4.106048583984375,
      "learning_rate": 5.775033330994317e-06,
      "loss": 0.4972,
      "step": 14189
    },
    {
      "epoch": 4.480580991474581,
      "grad_norm": 5.193219184875488,
      "learning_rate": 5.771524805276823e-06,
      "loss": 0.6756,
      "step": 14190
    },
    {
      "epoch": 4.4808967477107675,
      "grad_norm": 4.186735153198242,
      "learning_rate": 5.76801627955933e-06,
      "loss": 0.5592,
      "step": 14191
    },
    {
      "epoch": 4.481212503946953,
      "grad_norm": 5.337577819824219,
      "learning_rate": 5.764507753841836e-06,
      "loss": 0.6483,
      "step": 14192
    },
    {
      "epoch": 4.481528260183139,
      "grad_norm": 3.75608491897583,
      "learning_rate": 5.7609992281243425e-06,
      "loss": 0.4993,
      "step": 14193
    },
    {
      "epoch": 4.481844016419324,
      "grad_norm": 3.8313121795654297,
      "learning_rate": 5.7574907024068484e-06,
      "loss": 0.498,
      "step": 14194
    },
    {
      "epoch": 4.48215977265551,
      "grad_norm": 3.792085647583008,
      "learning_rate": 5.753982176689355e-06,
      "loss": 0.5624,
      "step": 14195
    },
    {
      "epoch": 4.482475528891696,
      "grad_norm": 4.231002330780029,
      "learning_rate": 5.750473650971862e-06,
      "loss": 0.5704,
      "step": 14196
    },
    {
      "epoch": 4.482791285127881,
      "grad_norm": 4.175355911254883,
      "learning_rate": 5.746965125254369e-06,
      "loss": 0.6679,
      "step": 14197
    },
    {
      "epoch": 4.483107041364067,
      "grad_norm": 4.0341796875,
      "learning_rate": 5.743456599536875e-06,
      "loss": 0.6005,
      "step": 14198
    },
    {
      "epoch": 4.4834227976002525,
      "grad_norm": 3.7948362827301025,
      "learning_rate": 5.739948073819382e-06,
      "loss": 0.5496,
      "step": 14199
    },
    {
      "epoch": 4.483738553836439,
      "grad_norm": 4.2549238204956055,
      "learning_rate": 5.736439548101888e-06,
      "loss": 0.558,
      "step": 14200
    },
    {
      "epoch": 4.483738553836439,
      "eval_loss": 0.8378090262413025,
      "eval_runtime": 4.6762,
      "eval_samples_per_second": 532.48,
      "eval_steps_per_second": 4.491,
      "step": 14200
    },
    {
      "epoch": 4.484054310072624,
      "grad_norm": 3.8738412857055664,
      "learning_rate": 5.7329310223843944e-06,
      "loss": 0.6633,
      "step": 14201
    },
    {
      "epoch": 4.484370066308809,
      "grad_norm": 3.5360653400421143,
      "learning_rate": 5.729422496666901e-06,
      "loss": 0.5536,
      "step": 14202
    },
    {
      "epoch": 4.4846858225449955,
      "grad_norm": 3.955471992492676,
      "learning_rate": 5.725913970949407e-06,
      "loss": 0.5242,
      "step": 14203
    },
    {
      "epoch": 4.485001578781181,
      "grad_norm": 3.5710625648498535,
      "learning_rate": 5.722405445231914e-06,
      "loss": 0.5238,
      "step": 14204
    },
    {
      "epoch": 4.485317335017367,
      "grad_norm": 7.0287580490112305,
      "learning_rate": 5.71889691951442e-06,
      "loss": 0.6566,
      "step": 14205
    },
    {
      "epoch": 4.485633091253552,
      "grad_norm": 3.8899834156036377,
      "learning_rate": 5.715388393796927e-06,
      "loss": 0.6605,
      "step": 14206
    },
    {
      "epoch": 4.4859488474897375,
      "grad_norm": 4.491964817047119,
      "learning_rate": 5.711879868079433e-06,
      "loss": 0.5766,
      "step": 14207
    },
    {
      "epoch": 4.486264603725924,
      "grad_norm": 3.693998098373413,
      "learning_rate": 5.7083713423619404e-06,
      "loss": 0.5597,
      "step": 14208
    },
    {
      "epoch": 4.486580359962109,
      "grad_norm": 3.3438148498535156,
      "learning_rate": 5.704862816644446e-06,
      "loss": 0.5807,
      "step": 14209
    },
    {
      "epoch": 4.486896116198295,
      "grad_norm": 3.940448760986328,
      "learning_rate": 5.701354290926953e-06,
      "loss": 0.5153,
      "step": 14210
    },
    {
      "epoch": 4.4872118724344805,
      "grad_norm": 3.8324334621429443,
      "learning_rate": 5.697845765209459e-06,
      "loss": 0.5021,
      "step": 14211
    },
    {
      "epoch": 4.487527628670666,
      "grad_norm": 5.17466926574707,
      "learning_rate": 5.694337239491965e-06,
      "loss": 0.6295,
      "step": 14212
    },
    {
      "epoch": 4.487843384906852,
      "grad_norm": 4.826691627502441,
      "learning_rate": 5.690828713774472e-06,
      "loss": 0.6224,
      "step": 14213
    },
    {
      "epoch": 4.488159141143037,
      "grad_norm": 4.279946327209473,
      "learning_rate": 5.687320188056979e-06,
      "loss": 0.589,
      "step": 14214
    },
    {
      "epoch": 4.4884748973792234,
      "grad_norm": 4.4188055992126465,
      "learning_rate": 5.683811662339486e-06,
      "loss": 0.7179,
      "step": 14215
    },
    {
      "epoch": 4.488790653615409,
      "grad_norm": 4.593045234680176,
      "learning_rate": 5.6803031366219915e-06,
      "loss": 0.518,
      "step": 14216
    },
    {
      "epoch": 4.489106409851595,
      "grad_norm": 3.243377923965454,
      "learning_rate": 5.676794610904498e-06,
      "loss": 0.5507,
      "step": 14217
    },
    {
      "epoch": 4.48942216608778,
      "grad_norm": 3.7351207733154297,
      "learning_rate": 5.673286085187004e-06,
      "loss": 0.653,
      "step": 14218
    },
    {
      "epoch": 4.4897379223239655,
      "grad_norm": 4.285088539123535,
      "learning_rate": 5.669777559469511e-06,
      "loss": 0.5274,
      "step": 14219
    },
    {
      "epoch": 4.490053678560152,
      "grad_norm": 3.998868703842163,
      "learning_rate": 5.666269033752018e-06,
      "loss": 0.5505,
      "step": 14220
    },
    {
      "epoch": 4.490369434796337,
      "grad_norm": 4.722806930541992,
      "learning_rate": 5.662760508034525e-06,
      "loss": 0.6879,
      "step": 14221
    },
    {
      "epoch": 4.490685191032523,
      "grad_norm": 3.932443380355835,
      "learning_rate": 5.659251982317031e-06,
      "loss": 0.5301,
      "step": 14222
    },
    {
      "epoch": 4.4910009472687085,
      "grad_norm": 4.632800102233887,
      "learning_rate": 5.655743456599537e-06,
      "loss": 0.5663,
      "step": 14223
    },
    {
      "epoch": 4.491316703504895,
      "grad_norm": 4.407804012298584,
      "learning_rate": 5.6522349308820435e-06,
      "loss": 0.6268,
      "step": 14224
    },
    {
      "epoch": 4.49163245974108,
      "grad_norm": 3.998497486114502,
      "learning_rate": 5.6487264051645495e-06,
      "loss": 0.5322,
      "step": 14225
    },
    {
      "epoch": 4.491948215977265,
      "grad_norm": 3.710865020751953,
      "learning_rate": 5.645217879447057e-06,
      "loss": 0.6144,
      "step": 14226
    },
    {
      "epoch": 4.492263972213451,
      "grad_norm": 3.5758769512176514,
      "learning_rate": 5.641709353729563e-06,
      "loss": 0.5495,
      "step": 14227
    },
    {
      "epoch": 4.492579728449637,
      "grad_norm": 4.516882419586182,
      "learning_rate": 5.63820082801207e-06,
      "loss": 0.6204,
      "step": 14228
    },
    {
      "epoch": 4.492895484685823,
      "grad_norm": 3.9574379920959473,
      "learning_rate": 5.634692302294576e-06,
      "loss": 0.5696,
      "step": 14229
    },
    {
      "epoch": 4.493211240922008,
      "grad_norm": 4.412933826446533,
      "learning_rate": 5.631183776577083e-06,
      "loss": 0.6451,
      "step": 14230
    },
    {
      "epoch": 4.4935269971581935,
      "grad_norm": 3.5990352630615234,
      "learning_rate": 5.627675250859589e-06,
      "loss": 0.5227,
      "step": 14231
    },
    {
      "epoch": 4.49384275339438,
      "grad_norm": 3.8105273246765137,
      "learning_rate": 5.6241667251420955e-06,
      "loss": 0.6576,
      "step": 14232
    },
    {
      "epoch": 4.494158509630565,
      "grad_norm": 4.031144618988037,
      "learning_rate": 5.620658199424602e-06,
      "loss": 0.6015,
      "step": 14233
    },
    {
      "epoch": 4.494474265866751,
      "grad_norm": 4.725373268127441,
      "learning_rate": 5.617149673707109e-06,
      "loss": 0.5345,
      "step": 14234
    },
    {
      "epoch": 4.494790022102936,
      "grad_norm": 4.3269147872924805,
      "learning_rate": 5.613641147989615e-06,
      "loss": 0.6844,
      "step": 14235
    },
    {
      "epoch": 4.495105778339122,
      "grad_norm": 4.6483564376831055,
      "learning_rate": 5.610132622272121e-06,
      "loss": 0.5941,
      "step": 14236
    },
    {
      "epoch": 4.495421534575308,
      "grad_norm": 3.882124185562134,
      "learning_rate": 5.606624096554628e-06,
      "loss": 0.6078,
      "step": 14237
    },
    {
      "epoch": 4.495737290811493,
      "grad_norm": 5.251631259918213,
      "learning_rate": 5.603115570837135e-06,
      "loss": 0.7215,
      "step": 14238
    },
    {
      "epoch": 4.496053047047679,
      "grad_norm": 4.265408039093018,
      "learning_rate": 5.5996070451196415e-06,
      "loss": 0.6513,
      "step": 14239
    },
    {
      "epoch": 4.496368803283865,
      "grad_norm": 4.118241310119629,
      "learning_rate": 5.5960985194021474e-06,
      "loss": 0.6228,
      "step": 14240
    },
    {
      "epoch": 4.496684559520051,
      "grad_norm": 5.100881099700928,
      "learning_rate": 5.592589993684654e-06,
      "loss": 0.4323,
      "step": 14241
    },
    {
      "epoch": 4.497000315756236,
      "grad_norm": 3.5171139240264893,
      "learning_rate": 5.58908146796716e-06,
      "loss": 0.6051,
      "step": 14242
    },
    {
      "epoch": 4.4973160719924214,
      "grad_norm": 3.3966643810272217,
      "learning_rate": 5.585572942249667e-06,
      "loss": 0.5118,
      "step": 14243
    },
    {
      "epoch": 4.497631828228608,
      "grad_norm": 3.9803144931793213,
      "learning_rate": 5.582064416532173e-06,
      "loss": 0.4864,
      "step": 14244
    },
    {
      "epoch": 4.497947584464793,
      "grad_norm": 4.067866325378418,
      "learning_rate": 5.578555890814681e-06,
      "loss": 0.6133,
      "step": 14245
    },
    {
      "epoch": 4.498263340700979,
      "grad_norm": 4.703128814697266,
      "learning_rate": 5.575047365097187e-06,
      "loss": 0.5905,
      "step": 14246
    },
    {
      "epoch": 4.498579096937164,
      "grad_norm": 4.113557815551758,
      "learning_rate": 5.571538839379693e-06,
      "loss": 0.5806,
      "step": 14247
    },
    {
      "epoch": 4.498894853173351,
      "grad_norm": 3.5043067932128906,
      "learning_rate": 5.568030313662199e-06,
      "loss": 0.5574,
      "step": 14248
    },
    {
      "epoch": 4.499210609409536,
      "grad_norm": 4.317317008972168,
      "learning_rate": 5.564521787944705e-06,
      "loss": 0.5988,
      "step": 14249
    },
    {
      "epoch": 4.499526365645721,
      "grad_norm": 3.936058759689331,
      "learning_rate": 5.561013262227212e-06,
      "loss": 0.5815,
      "step": 14250
    },
    {
      "epoch": 4.499842121881907,
      "grad_norm": 3.167701244354248,
      "learning_rate": 5.557504736509719e-06,
      "loss": 0.5702,
      "step": 14251
    },
    {
      "epoch": 4.500157878118093,
      "grad_norm": 4.266563415527344,
      "learning_rate": 5.553996210792226e-06,
      "loss": 0.6919,
      "step": 14252
    },
    {
      "epoch": 4.500473634354279,
      "grad_norm": 4.0013227462768555,
      "learning_rate": 5.550487685074732e-06,
      "loss": 0.621,
      "step": 14253
    },
    {
      "epoch": 4.500789390590464,
      "grad_norm": 3.383206367492676,
      "learning_rate": 5.546979159357239e-06,
      "loss": 0.5283,
      "step": 14254
    },
    {
      "epoch": 4.501105146826649,
      "grad_norm": 4.057204723358154,
      "learning_rate": 5.5434706336397445e-06,
      "loss": 0.577,
      "step": 14255
    },
    {
      "epoch": 4.501420903062836,
      "grad_norm": 4.190496921539307,
      "learning_rate": 5.539962107922251e-06,
      "loss": 0.5615,
      "step": 14256
    },
    {
      "epoch": 4.501736659299021,
      "grad_norm": 4.8332905769348145,
      "learning_rate": 5.536453582204758e-06,
      "loss": 0.6203,
      "step": 14257
    },
    {
      "epoch": 4.502052415535207,
      "grad_norm": 4.2772955894470215,
      "learning_rate": 5.532945056487265e-06,
      "loss": 0.5789,
      "step": 14258
    },
    {
      "epoch": 4.502368171771392,
      "grad_norm": 3.6941542625427246,
      "learning_rate": 5.529436530769771e-06,
      "loss": 0.5843,
      "step": 14259
    },
    {
      "epoch": 4.502683928007578,
      "grad_norm": 3.9048619270324707,
      "learning_rate": 5.525928005052277e-06,
      "loss": 0.5347,
      "step": 14260
    },
    {
      "epoch": 4.502999684243764,
      "grad_norm": 4.263216495513916,
      "learning_rate": 5.522419479334784e-06,
      "loss": 0.6798,
      "step": 14261
    },
    {
      "epoch": 4.503315440479949,
      "grad_norm": 4.285156726837158,
      "learning_rate": 5.51891095361729e-06,
      "loss": 0.5849,
      "step": 14262
    },
    {
      "epoch": 4.503631196716135,
      "grad_norm": 4.506689071655273,
      "learning_rate": 5.5154024278997965e-06,
      "loss": 0.6427,
      "step": 14263
    },
    {
      "epoch": 4.503946952952321,
      "grad_norm": 3.939086675643921,
      "learning_rate": 5.511893902182303e-06,
      "loss": 0.4751,
      "step": 14264
    },
    {
      "epoch": 4.504262709188507,
      "grad_norm": 3.7052927017211914,
      "learning_rate": 5.50838537646481e-06,
      "loss": 0.4938,
      "step": 14265
    },
    {
      "epoch": 4.504578465424692,
      "grad_norm": 3.8193697929382324,
      "learning_rate": 5.504876850747316e-06,
      "loss": 0.5672,
      "step": 14266
    },
    {
      "epoch": 4.504894221660878,
      "grad_norm": 4.039775371551514,
      "learning_rate": 5.501368325029823e-06,
      "loss": 0.4747,
      "step": 14267
    },
    {
      "epoch": 4.505209977897064,
      "grad_norm": 2.889030933380127,
      "learning_rate": 5.497859799312329e-06,
      "loss": 0.4573,
      "step": 14268
    },
    {
      "epoch": 4.505525734133249,
      "grad_norm": 4.647496223449707,
      "learning_rate": 5.494351273594836e-06,
      "loss": 0.6321,
      "step": 14269
    },
    {
      "epoch": 4.505841490369435,
      "grad_norm": 3.7861392498016357,
      "learning_rate": 5.4908427478773425e-06,
      "loss": 0.6578,
      "step": 14270
    },
    {
      "epoch": 4.50615724660562,
      "grad_norm": 3.9092156887054443,
      "learning_rate": 5.4873342221598485e-06,
      "loss": 0.6047,
      "step": 14271
    },
    {
      "epoch": 4.5064730028418065,
      "grad_norm": 4.547105312347412,
      "learning_rate": 5.483825696442355e-06,
      "loss": 0.5972,
      "step": 14272
    },
    {
      "epoch": 4.506788759077992,
      "grad_norm": 3.9329848289489746,
      "learning_rate": 5.480317170724861e-06,
      "loss": 0.6848,
      "step": 14273
    },
    {
      "epoch": 4.507104515314177,
      "grad_norm": 3.6053695678710938,
      "learning_rate": 5.476808645007368e-06,
      "loss": 0.5151,
      "step": 14274
    },
    {
      "epoch": 4.507420271550363,
      "grad_norm": 4.2053656578063965,
      "learning_rate": 5.473300119289875e-06,
      "loss": 0.5283,
      "step": 14275
    },
    {
      "epoch": 4.507736027786549,
      "grad_norm": 4.385671615600586,
      "learning_rate": 5.469791593572382e-06,
      "loss": 0.6278,
      "step": 14276
    },
    {
      "epoch": 4.508051784022735,
      "grad_norm": 4.726953029632568,
      "learning_rate": 5.466283067854888e-06,
      "loss": 0.5834,
      "step": 14277
    },
    {
      "epoch": 4.50836754025892,
      "grad_norm": 3.3976452350616455,
      "learning_rate": 5.4627745421373945e-06,
      "loss": 0.5996,
      "step": 14278
    },
    {
      "epoch": 4.508683296495105,
      "grad_norm": 3.8459649085998535,
      "learning_rate": 5.4592660164199004e-06,
      "loss": 0.5683,
      "step": 14279
    },
    {
      "epoch": 4.5089990527312915,
      "grad_norm": 3.704543352127075,
      "learning_rate": 5.455757490702407e-06,
      "loss": 0.5904,
      "step": 14280
    },
    {
      "epoch": 4.509314808967477,
      "grad_norm": 4.728396892547607,
      "learning_rate": 5.452248964984913e-06,
      "loss": 0.6514,
      "step": 14281
    },
    {
      "epoch": 4.509630565203663,
      "grad_norm": 4.376248359680176,
      "learning_rate": 5.44874043926742e-06,
      "loss": 0.6151,
      "step": 14282
    },
    {
      "epoch": 4.509946321439848,
      "grad_norm": 5.067110538482666,
      "learning_rate": 5.445231913549927e-06,
      "loss": 0.5583,
      "step": 14283
    },
    {
      "epoch": 4.5102620776760345,
      "grad_norm": 3.722665309906006,
      "learning_rate": 5.441723387832433e-06,
      "loss": 0.5297,
      "step": 14284
    },
    {
      "epoch": 4.51057783391222,
      "grad_norm": 3.7130959033966064,
      "learning_rate": 5.43821486211494e-06,
      "loss": 0.5725,
      "step": 14285
    },
    {
      "epoch": 4.510893590148405,
      "grad_norm": 3.5378804206848145,
      "learning_rate": 5.434706336397446e-06,
      "loss": 0.5594,
      "step": 14286
    },
    {
      "epoch": 4.511209346384591,
      "grad_norm": 5.702146530151367,
      "learning_rate": 5.431197810679952e-06,
      "loss": 0.6396,
      "step": 14287
    },
    {
      "epoch": 4.5115251026207766,
      "grad_norm": 3.0463616847991943,
      "learning_rate": 5.427689284962459e-06,
      "loss": 0.506,
      "step": 14288
    },
    {
      "epoch": 4.511840858856963,
      "grad_norm": 3.8784067630767822,
      "learning_rate": 5.424180759244966e-06,
      "loss": 0.5997,
      "step": 14289
    },
    {
      "epoch": 4.512156615093148,
      "grad_norm": 3.444248914718628,
      "learning_rate": 5.420672233527472e-06,
      "loss": 0.5221,
      "step": 14290
    },
    {
      "epoch": 4.512472371329334,
      "grad_norm": 4.29467248916626,
      "learning_rate": 5.417163707809979e-06,
      "loss": 0.5868,
      "step": 14291
    },
    {
      "epoch": 4.5127881275655195,
      "grad_norm": 4.7698259353637695,
      "learning_rate": 5.413655182092485e-06,
      "loss": 0.66,
      "step": 14292
    },
    {
      "epoch": 4.513103883801705,
      "grad_norm": 5.02537202835083,
      "learning_rate": 5.410146656374992e-06,
      "loss": 0.5552,
      "step": 14293
    },
    {
      "epoch": 4.513419640037891,
      "grad_norm": 4.080223560333252,
      "learning_rate": 5.406638130657498e-06,
      "loss": 0.5468,
      "step": 14294
    },
    {
      "epoch": 4.513735396274076,
      "grad_norm": 3.9563112258911133,
      "learning_rate": 5.403129604940004e-06,
      "loss": 0.5889,
      "step": 14295
    },
    {
      "epoch": 4.5140511525102625,
      "grad_norm": 5.368015289306641,
      "learning_rate": 5.399621079222511e-06,
      "loss": 0.5459,
      "step": 14296
    },
    {
      "epoch": 4.514366908746448,
      "grad_norm": 3.7417550086975098,
      "learning_rate": 5.396112553505017e-06,
      "loss": 0.5908,
      "step": 14297
    },
    {
      "epoch": 4.514682664982633,
      "grad_norm": 4.345045566558838,
      "learning_rate": 5.392604027787524e-06,
      "loss": 0.5841,
      "step": 14298
    },
    {
      "epoch": 4.514998421218819,
      "grad_norm": 3.8928864002227783,
      "learning_rate": 5.38909550207003e-06,
      "loss": 0.4873,
      "step": 14299
    },
    {
      "epoch": 4.5153141774550045,
      "grad_norm": 5.043690204620361,
      "learning_rate": 5.385586976352537e-06,
      "loss": 0.6559,
      "step": 14300
    },
    {
      "epoch": 4.515629933691191,
      "grad_norm": 4.302140235900879,
      "learning_rate": 5.3820784506350435e-06,
      "loss": 0.6624,
      "step": 14301
    },
    {
      "epoch": 4.515945689927376,
      "grad_norm": 4.237254619598389,
      "learning_rate": 5.37856992491755e-06,
      "loss": 0.6147,
      "step": 14302
    },
    {
      "epoch": 4.516261446163561,
      "grad_norm": 4.292686462402344,
      "learning_rate": 5.375061399200056e-06,
      "loss": 0.5004,
      "step": 14303
    },
    {
      "epoch": 4.5165772023997475,
      "grad_norm": 3.161472797393799,
      "learning_rate": 5.371552873482563e-06,
      "loss": 0.6272,
      "step": 14304
    },
    {
      "epoch": 4.516892958635933,
      "grad_norm": 3.438056707382202,
      "learning_rate": 5.368044347765069e-06,
      "loss": 0.5325,
      "step": 14305
    },
    {
      "epoch": 4.517208714872119,
      "grad_norm": 3.9033944606781006,
      "learning_rate": 5.364535822047576e-06,
      "loss": 0.5441,
      "step": 14306
    },
    {
      "epoch": 4.517524471108304,
      "grad_norm": 4.681055545806885,
      "learning_rate": 5.361027296330083e-06,
      "loss": 0.5778,
      "step": 14307
    },
    {
      "epoch": 4.51784022734449,
      "grad_norm": 4.537674903869629,
      "learning_rate": 5.357518770612589e-06,
      "loss": 0.6882,
      "step": 14308
    },
    {
      "epoch": 4.518155983580676,
      "grad_norm": 3.976126194000244,
      "learning_rate": 5.3540102448950955e-06,
      "loss": 0.6624,
      "step": 14309
    },
    {
      "epoch": 4.518471739816861,
      "grad_norm": 3.6401314735412598,
      "learning_rate": 5.3505017191776015e-06,
      "loss": 0.6365,
      "step": 14310
    },
    {
      "epoch": 4.518787496053047,
      "grad_norm": 3.8265063762664795,
      "learning_rate": 5.346993193460108e-06,
      "loss": 0.5251,
      "step": 14311
    },
    {
      "epoch": 4.5191032522892325,
      "grad_norm": 3.781917095184326,
      "learning_rate": 5.343484667742614e-06,
      "loss": 0.6488,
      "step": 14312
    },
    {
      "epoch": 4.519419008525419,
      "grad_norm": 4.0314741134643555,
      "learning_rate": 5.339976142025122e-06,
      "loss": 0.5691,
      "step": 14313
    },
    {
      "epoch": 4.519734764761604,
      "grad_norm": 3.3283238410949707,
      "learning_rate": 5.336467616307628e-06,
      "loss": 0.5545,
      "step": 14314
    },
    {
      "epoch": 4.52005052099779,
      "grad_norm": 3.745741367340088,
      "learning_rate": 5.332959090590135e-06,
      "loss": 0.5733,
      "step": 14315
    },
    {
      "epoch": 4.520366277233975,
      "grad_norm": 3.970289945602417,
      "learning_rate": 5.329450564872641e-06,
      "loss": 0.5062,
      "step": 14316
    },
    {
      "epoch": 4.520682033470161,
      "grad_norm": 4.116623878479004,
      "learning_rate": 5.325942039155147e-06,
      "loss": 0.4726,
      "step": 14317
    },
    {
      "epoch": 4.520997789706347,
      "grad_norm": 3.3794944286346436,
      "learning_rate": 5.3224335134376534e-06,
      "loss": 0.5403,
      "step": 14318
    },
    {
      "epoch": 4.521313545942532,
      "grad_norm": 4.429907321929932,
      "learning_rate": 5.31892498772016e-06,
      "loss": 0.5191,
      "step": 14319
    },
    {
      "epoch": 4.521629302178718,
      "grad_norm": 3.8933262825012207,
      "learning_rate": 5.315416462002667e-06,
      "loss": 0.6789,
      "step": 14320
    },
    {
      "epoch": 4.521945058414904,
      "grad_norm": 4.324127197265625,
      "learning_rate": 5.311907936285173e-06,
      "loss": 0.7045,
      "step": 14321
    },
    {
      "epoch": 4.522260814651089,
      "grad_norm": 4.084039211273193,
      "learning_rate": 5.30839941056768e-06,
      "loss": 0.6087,
      "step": 14322
    },
    {
      "epoch": 4.522576570887275,
      "grad_norm": 3.439426898956299,
      "learning_rate": 5.304890884850186e-06,
      "loss": 0.5157,
      "step": 14323
    },
    {
      "epoch": 4.5228923271234605,
      "grad_norm": 4.03227424621582,
      "learning_rate": 5.301382359132693e-06,
      "loss": 0.5595,
      "step": 14324
    },
    {
      "epoch": 4.523208083359647,
      "grad_norm": 3.532500982284546,
      "learning_rate": 5.2978738334151994e-06,
      "loss": 0.5003,
      "step": 14325
    },
    {
      "epoch": 4.523523839595832,
      "grad_norm": 4.202742099761963,
      "learning_rate": 5.294365307697706e-06,
      "loss": 0.6784,
      "step": 14326
    },
    {
      "epoch": 4.523839595832017,
      "grad_norm": 5.539073467254639,
      "learning_rate": 5.290856781980212e-06,
      "loss": 0.6005,
      "step": 14327
    },
    {
      "epoch": 4.524155352068203,
      "grad_norm": 3.488359212875366,
      "learning_rate": 5.287348256262719e-06,
      "loss": 0.5757,
      "step": 14328
    },
    {
      "epoch": 4.524471108304389,
      "grad_norm": 3.600961446762085,
      "learning_rate": 5.283839730545225e-06,
      "loss": 0.5319,
      "step": 14329
    },
    {
      "epoch": 4.524786864540575,
      "grad_norm": 4.021499156951904,
      "learning_rate": 5.280331204827731e-06,
      "loss": 0.6473,
      "step": 14330
    },
    {
      "epoch": 4.52510262077676,
      "grad_norm": 3.6970314979553223,
      "learning_rate": 5.276822679110239e-06,
      "loss": 0.5121,
      "step": 14331
    },
    {
      "epoch": 4.525418377012946,
      "grad_norm": 4.675222396850586,
      "learning_rate": 5.273314153392745e-06,
      "loss": 0.6183,
      "step": 14332
    },
    {
      "epoch": 4.525734133249132,
      "grad_norm": 4.268986225128174,
      "learning_rate": 5.269805627675251e-06,
      "loss": 0.6245,
      "step": 14333
    },
    {
      "epoch": 4.526049889485318,
      "grad_norm": 4.199731349945068,
      "learning_rate": 5.266297101957757e-06,
      "loss": 0.6749,
      "step": 14334
    },
    {
      "epoch": 4.526365645721503,
      "grad_norm": 3.8762872219085693,
      "learning_rate": 5.262788576240264e-06,
      "loss": 0.6298,
      "step": 14335
    },
    {
      "epoch": 4.526681401957688,
      "grad_norm": 5.171905994415283,
      "learning_rate": 5.25928005052277e-06,
      "loss": 0.6765,
      "step": 14336
    },
    {
      "epoch": 4.526997158193875,
      "grad_norm": 4.940359115600586,
      "learning_rate": 5.255771524805277e-06,
      "loss": 0.5758,
      "step": 14337
    },
    {
      "epoch": 4.52731291443006,
      "grad_norm": 4.342586994171143,
      "learning_rate": 5.252262999087784e-06,
      "loss": 0.5774,
      "step": 14338
    },
    {
      "epoch": 4.527628670666246,
      "grad_norm": 4.308276653289795,
      "learning_rate": 5.248754473370291e-06,
      "loss": 0.5806,
      "step": 14339
    },
    {
      "epoch": 4.527944426902431,
      "grad_norm": 4.485562324523926,
      "learning_rate": 5.2452459476527965e-06,
      "loss": 0.6025,
      "step": 14340
    },
    {
      "epoch": 4.528260183138617,
      "grad_norm": 3.8767261505126953,
      "learning_rate": 5.2417374219353025e-06,
      "loss": 0.6363,
      "step": 14341
    },
    {
      "epoch": 4.528575939374803,
      "grad_norm": 4.25459623336792,
      "learning_rate": 5.238228896217809e-06,
      "loss": 0.5877,
      "step": 14342
    },
    {
      "epoch": 4.528891695610988,
      "grad_norm": 4.63521146774292,
      "learning_rate": 5.234720370500316e-06,
      "loss": 0.6094,
      "step": 14343
    },
    {
      "epoch": 4.529207451847174,
      "grad_norm": 4.439807415008545,
      "learning_rate": 5.231211844782823e-06,
      "loss": 0.5396,
      "step": 14344
    },
    {
      "epoch": 4.52952320808336,
      "grad_norm": 3.5987510681152344,
      "learning_rate": 5.227703319065329e-06,
      "loss": 0.6403,
      "step": 14345
    },
    {
      "epoch": 4.529838964319545,
      "grad_norm": 3.6369681358337402,
      "learning_rate": 5.224194793347836e-06,
      "loss": 0.6596,
      "step": 14346
    },
    {
      "epoch": 4.530154720555731,
      "grad_norm": 4.231431484222412,
      "learning_rate": 5.220686267630342e-06,
      "loss": 0.5349,
      "step": 14347
    },
    {
      "epoch": 4.530470476791916,
      "grad_norm": 3.845881700515747,
      "learning_rate": 5.2171777419128485e-06,
      "loss": 0.5737,
      "step": 14348
    },
    {
      "epoch": 4.530786233028103,
      "grad_norm": 3.2402658462524414,
      "learning_rate": 5.2136692161953545e-06,
      "loss": 0.5344,
      "step": 14349
    },
    {
      "epoch": 4.531101989264288,
      "grad_norm": 3.2340517044067383,
      "learning_rate": 5.210160690477862e-06,
      "loss": 0.5079,
      "step": 14350
    },
    {
      "epoch": 4.531417745500473,
      "grad_norm": 3.8044614791870117,
      "learning_rate": 5.206652164760368e-06,
      "loss": 0.548,
      "step": 14351
    },
    {
      "epoch": 4.531733501736659,
      "grad_norm": 3.7907216548919678,
      "learning_rate": 5.203143639042875e-06,
      "loss": 0.5789,
      "step": 14352
    },
    {
      "epoch": 4.532049257972845,
      "grad_norm": 3.874142646789551,
      "learning_rate": 5.199635113325381e-06,
      "loss": 0.5915,
      "step": 14353
    },
    {
      "epoch": 4.532365014209031,
      "grad_norm": 4.472775459289551,
      "learning_rate": 5.196126587607887e-06,
      "loss": 0.6585,
      "step": 14354
    },
    {
      "epoch": 4.532680770445216,
      "grad_norm": 3.806968927383423,
      "learning_rate": 5.192618061890394e-06,
      "loss": 0.5069,
      "step": 14355
    },
    {
      "epoch": 4.532996526681402,
      "grad_norm": 3.8913159370422363,
      "learning_rate": 5.1891095361729005e-06,
      "loss": 0.5919,
      "step": 14356
    },
    {
      "epoch": 4.533312282917588,
      "grad_norm": 4.896110534667969,
      "learning_rate": 5.185601010455407e-06,
      "loss": 0.6346,
      "step": 14357
    },
    {
      "epoch": 4.533628039153774,
      "grad_norm": 3.7540900707244873,
      "learning_rate": 5.182092484737913e-06,
      "loss": 0.5923,
      "step": 14358
    },
    {
      "epoch": 4.533943795389959,
      "grad_norm": 3.680590867996216,
      "learning_rate": 5.17858395902042e-06,
      "loss": 0.6054,
      "step": 14359
    },
    {
      "epoch": 4.534259551626144,
      "grad_norm": 3.8537120819091797,
      "learning_rate": 5.175075433302926e-06,
      "loss": 0.6099,
      "step": 14360
    },
    {
      "epoch": 4.5345753078623305,
      "grad_norm": 4.377222537994385,
      "learning_rate": 5.171566907585433e-06,
      "loss": 0.6737,
      "step": 14361
    },
    {
      "epoch": 4.534891064098516,
      "grad_norm": 4.213190078735352,
      "learning_rate": 5.16805838186794e-06,
      "loss": 0.6235,
      "step": 14362
    },
    {
      "epoch": 4.535206820334702,
      "grad_norm": 4.033075332641602,
      "learning_rate": 5.1645498561504465e-06,
      "loss": 0.4994,
      "step": 14363
    },
    {
      "epoch": 4.535522576570887,
      "grad_norm": 5.710567474365234,
      "learning_rate": 5.1610413304329524e-06,
      "loss": 0.6648,
      "step": 14364
    },
    {
      "epoch": 4.535838332807073,
      "grad_norm": 4.043811798095703,
      "learning_rate": 5.157532804715458e-06,
      "loss": 0.606,
      "step": 14365
    },
    {
      "epoch": 4.536154089043259,
      "grad_norm": 3.7211272716522217,
      "learning_rate": 5.154024278997965e-06,
      "loss": 0.6202,
      "step": 14366
    },
    {
      "epoch": 4.536469845279444,
      "grad_norm": 3.431671142578125,
      "learning_rate": 5.150515753280471e-06,
      "loss": 0.5137,
      "step": 14367
    },
    {
      "epoch": 4.53678560151563,
      "grad_norm": 3.7445290088653564,
      "learning_rate": 5.147007227562978e-06,
      "loss": 0.4958,
      "step": 14368
    },
    {
      "epoch": 4.5371013577518156,
      "grad_norm": 4.450698375701904,
      "learning_rate": 5.143498701845485e-06,
      "loss": 0.5955,
      "step": 14369
    },
    {
      "epoch": 4.537417113988001,
      "grad_norm": 4.999032974243164,
      "learning_rate": 5.139990176127992e-06,
      "loss": 0.637,
      "step": 14370
    },
    {
      "epoch": 4.537732870224187,
      "grad_norm": 3.8855600357055664,
      "learning_rate": 5.136481650410498e-06,
      "loss": 0.6325,
      "step": 14371
    },
    {
      "epoch": 4.538048626460372,
      "grad_norm": 3.4889800548553467,
      "learning_rate": 5.132973124693004e-06,
      "loss": 0.5695,
      "step": 14372
    },
    {
      "epoch": 4.5383643826965585,
      "grad_norm": 4.361230373382568,
      "learning_rate": 5.12946459897551e-06,
      "loss": 0.5911,
      "step": 14373
    },
    {
      "epoch": 4.538680138932744,
      "grad_norm": 4.08676815032959,
      "learning_rate": 5.125956073258017e-06,
      "loss": 0.515,
      "step": 14374
    },
    {
      "epoch": 4.538995895168929,
      "grad_norm": 3.5223848819732666,
      "learning_rate": 5.122447547540524e-06,
      "loss": 0.4916,
      "step": 14375
    },
    {
      "epoch": 4.539311651405115,
      "grad_norm": 3.3709824085235596,
      "learning_rate": 5.11893902182303e-06,
      "loss": 0.5551,
      "step": 14376
    },
    {
      "epoch": 4.539627407641301,
      "grad_norm": 4.012054920196533,
      "learning_rate": 5.115430496105537e-06,
      "loss": 0.6204,
      "step": 14377
    },
    {
      "epoch": 4.539943163877487,
      "grad_norm": 3.463705539703369,
      "learning_rate": 5.111921970388043e-06,
      "loss": 0.5915,
      "step": 14378
    },
    {
      "epoch": 4.540258920113672,
      "grad_norm": 3.8689770698547363,
      "learning_rate": 5.1084134446705495e-06,
      "loss": 0.5094,
      "step": 14379
    },
    {
      "epoch": 4.540574676349858,
      "grad_norm": 4.3034868240356445,
      "learning_rate": 5.104904918953056e-06,
      "loss": 0.5738,
      "step": 14380
    },
    {
      "epoch": 4.5408904325860435,
      "grad_norm": 4.197754859924316,
      "learning_rate": 5.101396393235563e-06,
      "loss": 0.5657,
      "step": 14381
    },
    {
      "epoch": 4.54120618882223,
      "grad_norm": 3.696300745010376,
      "learning_rate": 5.097887867518069e-06,
      "loss": 0.5512,
      "step": 14382
    },
    {
      "epoch": 4.541521945058415,
      "grad_norm": 3.994079828262329,
      "learning_rate": 5.094379341800576e-06,
      "loss": 0.632,
      "step": 14383
    },
    {
      "epoch": 4.5418377012946,
      "grad_norm": 4.526368141174316,
      "learning_rate": 5.090870816083082e-06,
      "loss": 0.6577,
      "step": 14384
    },
    {
      "epoch": 4.5421534575307865,
      "grad_norm": 3.5369060039520264,
      "learning_rate": 5.087362290365589e-06,
      "loss": 0.6209,
      "step": 14385
    },
    {
      "epoch": 4.542469213766972,
      "grad_norm": 4.496100902557373,
      "learning_rate": 5.083853764648095e-06,
      "loss": 0.6646,
      "step": 14386
    },
    {
      "epoch": 4.542784970003158,
      "grad_norm": 4.471253871917725,
      "learning_rate": 5.080345238930602e-06,
      "loss": 0.566,
      "step": 14387
    },
    {
      "epoch": 4.543100726239343,
      "grad_norm": 3.9085981845855713,
      "learning_rate": 5.076836713213108e-06,
      "loss": 0.6008,
      "step": 14388
    },
    {
      "epoch": 4.5434164824755285,
      "grad_norm": 4.368932247161865,
      "learning_rate": 5.073328187495614e-06,
      "loss": 0.6241,
      "step": 14389
    },
    {
      "epoch": 4.543732238711715,
      "grad_norm": 4.960752964019775,
      "learning_rate": 5.069819661778121e-06,
      "loss": 0.5542,
      "step": 14390
    },
    {
      "epoch": 4.5440479949479,
      "grad_norm": 3.592060089111328,
      "learning_rate": 5.066311136060627e-06,
      "loss": 0.5025,
      "step": 14391
    },
    {
      "epoch": 4.544363751184086,
      "grad_norm": 4.25108528137207,
      "learning_rate": 5.062802610343134e-06,
      "loss": 0.6275,
      "step": 14392
    },
    {
      "epoch": 4.5446795074202715,
      "grad_norm": 3.6078128814697266,
      "learning_rate": 5.059294084625641e-06,
      "loss": 0.5728,
      "step": 14393
    },
    {
      "epoch": 4.544995263656457,
      "grad_norm": 3.5151734352111816,
      "learning_rate": 5.0557855589081475e-06,
      "loss": 0.6544,
      "step": 14394
    },
    {
      "epoch": 4.545311019892643,
      "grad_norm": 3.801464796066284,
      "learning_rate": 5.0522770331906535e-06,
      "loss": 0.4775,
      "step": 14395
    },
    {
      "epoch": 4.545626776128828,
      "grad_norm": 3.527803897857666,
      "learning_rate": 5.04876850747316e-06,
      "loss": 0.5849,
      "step": 14396
    },
    {
      "epoch": 4.545942532365014,
      "grad_norm": 4.0447468757629395,
      "learning_rate": 5.045259981755666e-06,
      "loss": 0.5499,
      "step": 14397
    },
    {
      "epoch": 4.5462582886012,
      "grad_norm": 4.562722682952881,
      "learning_rate": 5.041751456038173e-06,
      "loss": 0.6447,
      "step": 14398
    },
    {
      "epoch": 4.546574044837386,
      "grad_norm": 3.8704493045806885,
      "learning_rate": 5.03824293032068e-06,
      "loss": 0.5207,
      "step": 14399
    },
    {
      "epoch": 4.546889801073571,
      "grad_norm": 2.9979374408721924,
      "learning_rate": 5.034734404603186e-06,
      "loss": 0.4142,
      "step": 14400
    },
    {
      "epoch": 4.546889801073571,
      "eval_loss": 0.8338869214057922,
      "eval_runtime": 4.5558,
      "eval_samples_per_second": 546.556,
      "eval_steps_per_second": 4.61,
      "step": 14400
    },
    {
      "epoch": 4.5472055573097565,
      "grad_norm": 3.5021045207977295,
      "learning_rate": 5.031225878885693e-06,
      "loss": 0.5389,
      "step": 14401
    },
    {
      "epoch": 4.547521313545943,
      "grad_norm": 4.3052191734313965,
      "learning_rate": 5.027717353168199e-06,
      "loss": 0.6758,
      "step": 14402
    },
    {
      "epoch": 4.547837069782128,
      "grad_norm": 4.548710823059082,
      "learning_rate": 5.0242088274507054e-06,
      "loss": 0.5188,
      "step": 14403
    },
    {
      "epoch": 4.548152826018314,
      "grad_norm": 4.297270774841309,
      "learning_rate": 5.020700301733211e-06,
      "loss": 0.5774,
      "step": 14404
    },
    {
      "epoch": 4.5484685822544995,
      "grad_norm": 4.208998203277588,
      "learning_rate": 5.017191776015718e-06,
      "loss": 0.5891,
      "step": 14405
    },
    {
      "epoch": 4.548784338490686,
      "grad_norm": 4.171866416931152,
      "learning_rate": 5.013683250298225e-06,
      "loss": 0.5113,
      "step": 14406
    },
    {
      "epoch": 4.549100094726871,
      "grad_norm": 3.9988532066345215,
      "learning_rate": 5.010174724580732e-06,
      "loss": 0.6386,
      "step": 14407
    },
    {
      "epoch": 4.549415850963056,
      "grad_norm": 3.795397996902466,
      "learning_rate": 5.006666198863238e-06,
      "loss": 0.5897,
      "step": 14408
    },
    {
      "epoch": 4.549731607199242,
      "grad_norm": 3.805663585662842,
      "learning_rate": 5.003157673145745e-06,
      "loss": 0.5561,
      "step": 14409
    },
    {
      "epoch": 4.550047363435428,
      "grad_norm": 4.242880344390869,
      "learning_rate": 4.999649147428251e-06,
      "loss": 0.5856,
      "step": 14410
    },
    {
      "epoch": 4.550363119671614,
      "grad_norm": 3.65421462059021,
      "learning_rate": 4.996140621710757e-06,
      "loss": 0.5766,
      "step": 14411
    },
    {
      "epoch": 4.550678875907799,
      "grad_norm": 4.051403045654297,
      "learning_rate": 4.992632095993264e-06,
      "loss": 0.6609,
      "step": 14412
    },
    {
      "epoch": 4.5509946321439845,
      "grad_norm": 4.005472660064697,
      "learning_rate": 4.98912357027577e-06,
      "loss": 0.5477,
      "step": 14413
    },
    {
      "epoch": 4.551310388380171,
      "grad_norm": 5.0462141036987305,
      "learning_rate": 4.985615044558277e-06,
      "loss": 0.4957,
      "step": 14414
    },
    {
      "epoch": 4.551626144616356,
      "grad_norm": 4.759757041931152,
      "learning_rate": 4.982106518840783e-06,
      "loss": 0.7022,
      "step": 14415
    },
    {
      "epoch": 4.551941900852542,
      "grad_norm": 4.998772621154785,
      "learning_rate": 4.97859799312329e-06,
      "loss": 0.5741,
      "step": 14416
    },
    {
      "epoch": 4.552257657088727,
      "grad_norm": 4.215261459350586,
      "learning_rate": 4.975089467405796e-06,
      "loss": 0.521,
      "step": 14417
    },
    {
      "epoch": 4.552573413324913,
      "grad_norm": 3.3426201343536377,
      "learning_rate": 4.971580941688303e-06,
      "loss": 0.5772,
      "step": 14418
    },
    {
      "epoch": 4.552889169561099,
      "grad_norm": 4.452219009399414,
      "learning_rate": 4.968072415970809e-06,
      "loss": 0.5235,
      "step": 14419
    },
    {
      "epoch": 4.553204925797284,
      "grad_norm": 4.347962379455566,
      "learning_rate": 4.964563890253316e-06,
      "loss": 0.6106,
      "step": 14420
    },
    {
      "epoch": 4.55352068203347,
      "grad_norm": 4.024838447570801,
      "learning_rate": 4.961055364535822e-06,
      "loss": 0.5467,
      "step": 14421
    },
    {
      "epoch": 4.553836438269656,
      "grad_norm": 3.406282424926758,
      "learning_rate": 4.957546838818329e-06,
      "loss": 0.4728,
      "step": 14422
    },
    {
      "epoch": 4.554152194505842,
      "grad_norm": 3.911388635635376,
      "learning_rate": 4.954038313100835e-06,
      "loss": 0.5965,
      "step": 14423
    },
    {
      "epoch": 4.554467950742027,
      "grad_norm": 3.9516520500183105,
      "learning_rate": 4.950529787383342e-06,
      "loss": 0.587,
      "step": 14424
    },
    {
      "epoch": 4.554783706978212,
      "grad_norm": 3.664400815963745,
      "learning_rate": 4.9470212616658485e-06,
      "loss": 0.5596,
      "step": 14425
    },
    {
      "epoch": 4.555099463214399,
      "grad_norm": 3.269735097885132,
      "learning_rate": 4.9435127359483545e-06,
      "loss": 0.4995,
      "step": 14426
    },
    {
      "epoch": 4.555415219450584,
      "grad_norm": 3.862013578414917,
      "learning_rate": 4.940004210230861e-06,
      "loss": 0.58,
      "step": 14427
    },
    {
      "epoch": 4.55573097568677,
      "grad_norm": 4.402224540710449,
      "learning_rate": 4.936495684513367e-06,
      "loss": 0.5372,
      "step": 14428
    },
    {
      "epoch": 4.556046731922955,
      "grad_norm": 4.215651988983154,
      "learning_rate": 4.932987158795874e-06,
      "loss": 0.552,
      "step": 14429
    },
    {
      "epoch": 4.556362488159142,
      "grad_norm": 4.193905353546143,
      "learning_rate": 4.929478633078381e-06,
      "loss": 0.5597,
      "step": 14430
    },
    {
      "epoch": 4.556678244395327,
      "grad_norm": 5.332514762878418,
      "learning_rate": 4.925970107360888e-06,
      "loss": 0.6624,
      "step": 14431
    },
    {
      "epoch": 4.556994000631512,
      "grad_norm": 3.9586474895477295,
      "learning_rate": 4.922461581643394e-06,
      "loss": 0.5455,
      "step": 14432
    },
    {
      "epoch": 4.557309756867698,
      "grad_norm": 3.381730079650879,
      "learning_rate": 4.9189530559259005e-06,
      "loss": 0.5689,
      "step": 14433
    },
    {
      "epoch": 4.557625513103884,
      "grad_norm": 4.443066596984863,
      "learning_rate": 4.9154445302084065e-06,
      "loss": 0.6003,
      "step": 14434
    },
    {
      "epoch": 4.55794126934007,
      "grad_norm": 4.975936412811279,
      "learning_rate": 4.9119360044909124e-06,
      "loss": 0.6063,
      "step": 14435
    },
    {
      "epoch": 4.558257025576255,
      "grad_norm": 3.3200204372406006,
      "learning_rate": 4.90842747877342e-06,
      "loss": 0.4464,
      "step": 14436
    },
    {
      "epoch": 4.55857278181244,
      "grad_norm": 3.8427071571350098,
      "learning_rate": 4.904918953055926e-06,
      "loss": 0.6497,
      "step": 14437
    },
    {
      "epoch": 4.558888538048627,
      "grad_norm": 4.11784029006958,
      "learning_rate": 4.901410427338433e-06,
      "loss": 0.5825,
      "step": 14438
    },
    {
      "epoch": 4.559204294284812,
      "grad_norm": 4.631167411804199,
      "learning_rate": 4.897901901620939e-06,
      "loss": 0.7304,
      "step": 14439
    },
    {
      "epoch": 4.559520050520998,
      "grad_norm": 4.214578628540039,
      "learning_rate": 4.894393375903446e-06,
      "loss": 0.5897,
      "step": 14440
    },
    {
      "epoch": 4.559835806757183,
      "grad_norm": 5.579919338226318,
      "learning_rate": 4.890884850185952e-06,
      "loss": 0.6871,
      "step": 14441
    },
    {
      "epoch": 4.560151562993369,
      "grad_norm": 4.382658004760742,
      "learning_rate": 4.8873763244684584e-06,
      "loss": 0.6148,
      "step": 14442
    },
    {
      "epoch": 4.560467319229555,
      "grad_norm": 4.183598518371582,
      "learning_rate": 4.883867798750965e-06,
      "loss": 0.5137,
      "step": 14443
    },
    {
      "epoch": 4.56078307546574,
      "grad_norm": 3.8226442337036133,
      "learning_rate": 4.880359273033472e-06,
      "loss": 0.6477,
      "step": 14444
    },
    {
      "epoch": 4.561098831701926,
      "grad_norm": 4.606489181518555,
      "learning_rate": 4.876850747315978e-06,
      "loss": 0.5735,
      "step": 14445
    },
    {
      "epoch": 4.561414587938112,
      "grad_norm": 4.710513591766357,
      "learning_rate": 4.873342221598485e-06,
      "loss": 0.5973,
      "step": 14446
    },
    {
      "epoch": 4.561730344174298,
      "grad_norm": 4.81335973739624,
      "learning_rate": 4.869833695880991e-06,
      "loss": 0.6356,
      "step": 14447
    },
    {
      "epoch": 4.562046100410483,
      "grad_norm": 3.2663021087646484,
      "learning_rate": 4.866325170163498e-06,
      "loss": 0.5532,
      "step": 14448
    },
    {
      "epoch": 4.562361856646669,
      "grad_norm": 4.315070629119873,
      "learning_rate": 4.8628166444460044e-06,
      "loss": 0.6085,
      "step": 14449
    },
    {
      "epoch": 4.562677612882855,
      "grad_norm": 4.153069019317627,
      "learning_rate": 4.85930811872851e-06,
      "loss": 0.626,
      "step": 14450
    },
    {
      "epoch": 4.56299336911904,
      "grad_norm": 3.6024303436279297,
      "learning_rate": 4.855799593011017e-06,
      "loss": 0.5597,
      "step": 14451
    },
    {
      "epoch": 4.563309125355226,
      "grad_norm": 4.611166477203369,
      "learning_rate": 4.852291067293523e-06,
      "loss": 0.6369,
      "step": 14452
    },
    {
      "epoch": 4.563624881591411,
      "grad_norm": 5.4750142097473145,
      "learning_rate": 4.84878254157603e-06,
      "loss": 0.6423,
      "step": 14453
    },
    {
      "epoch": 4.5639406378275975,
      "grad_norm": 4.085116863250732,
      "learning_rate": 4.845274015858536e-06,
      "loss": 0.6299,
      "step": 14454
    },
    {
      "epoch": 4.564256394063783,
      "grad_norm": 4.373213291168213,
      "learning_rate": 4.841765490141044e-06,
      "loss": 0.6327,
      "step": 14455
    },
    {
      "epoch": 4.564572150299968,
      "grad_norm": 3.965573787689209,
      "learning_rate": 4.83825696442355e-06,
      "loss": 0.494,
      "step": 14456
    },
    {
      "epoch": 4.564887906536154,
      "grad_norm": 4.264153957366943,
      "learning_rate": 4.834748438706056e-06,
      "loss": 0.7105,
      "step": 14457
    },
    {
      "epoch": 4.56520366277234,
      "grad_norm": 4.64150333404541,
      "learning_rate": 4.831239912988562e-06,
      "loss": 0.647,
      "step": 14458
    },
    {
      "epoch": 4.565519419008526,
      "grad_norm": 4.397954940795898,
      "learning_rate": 4.827731387271068e-06,
      "loss": 0.5269,
      "step": 14459
    },
    {
      "epoch": 4.565835175244711,
      "grad_norm": 3.867658853530884,
      "learning_rate": 4.824222861553575e-06,
      "loss": 0.5825,
      "step": 14460
    },
    {
      "epoch": 4.566150931480896,
      "grad_norm": 4.138874053955078,
      "learning_rate": 4.820714335836082e-06,
      "loss": 0.6644,
      "step": 14461
    },
    {
      "epoch": 4.5664666877170825,
      "grad_norm": 4.097103118896484,
      "learning_rate": 4.817205810118589e-06,
      "loss": 0.578,
      "step": 14462
    },
    {
      "epoch": 4.566782443953268,
      "grad_norm": 3.6491732597351074,
      "learning_rate": 4.813697284401095e-06,
      "loss": 0.5847,
      "step": 14463
    },
    {
      "epoch": 4.567098200189454,
      "grad_norm": 5.253076076507568,
      "learning_rate": 4.8101887586836015e-06,
      "loss": 0.5698,
      "step": 14464
    },
    {
      "epoch": 4.567413956425639,
      "grad_norm": 3.32509446144104,
      "learning_rate": 4.8066802329661075e-06,
      "loss": 0.5318,
      "step": 14465
    },
    {
      "epoch": 4.567729712661825,
      "grad_norm": 4.2667412757873535,
      "learning_rate": 4.803171707248614e-06,
      "loss": 0.5601,
      "step": 14466
    },
    {
      "epoch": 4.568045468898011,
      "grad_norm": 3.9794602394104004,
      "learning_rate": 4.799663181531121e-06,
      "loss": 0.6027,
      "step": 14467
    },
    {
      "epoch": 4.568361225134196,
      "grad_norm": 3.502735137939453,
      "learning_rate": 4.796154655813628e-06,
      "loss": 0.5258,
      "step": 14468
    },
    {
      "epoch": 4.568676981370382,
      "grad_norm": 4.2695441246032715,
      "learning_rate": 4.792646130096134e-06,
      "loss": 0.6293,
      "step": 14469
    },
    {
      "epoch": 4.5689927376065675,
      "grad_norm": 4.30472469329834,
      "learning_rate": 4.78913760437864e-06,
      "loss": 0.5723,
      "step": 14470
    },
    {
      "epoch": 4.569308493842754,
      "grad_norm": 4.475404262542725,
      "learning_rate": 4.785629078661147e-06,
      "loss": 0.5713,
      "step": 14471
    },
    {
      "epoch": 4.569624250078939,
      "grad_norm": 5.448119163513184,
      "learning_rate": 4.782120552943653e-06,
      "loss": 0.6385,
      "step": 14472
    },
    {
      "epoch": 4.569940006315125,
      "grad_norm": 3.4966318607330322,
      "learning_rate": 4.7786120272261595e-06,
      "loss": 0.5777,
      "step": 14473
    },
    {
      "epoch": 4.5702557625513105,
      "grad_norm": 4.109592914581299,
      "learning_rate": 4.775103501508666e-06,
      "loss": 0.5303,
      "step": 14474
    },
    {
      "epoch": 4.570571518787496,
      "grad_norm": 5.120181083679199,
      "learning_rate": 4.771594975791173e-06,
      "loss": 0.6256,
      "step": 14475
    },
    {
      "epoch": 4.570887275023682,
      "grad_norm": 4.321907997131348,
      "learning_rate": 4.768086450073679e-06,
      "loss": 0.665,
      "step": 14476
    },
    {
      "epoch": 4.571203031259867,
      "grad_norm": 4.137260913848877,
      "learning_rate": 4.764577924356186e-06,
      "loss": 0.5355,
      "step": 14477
    },
    {
      "epoch": 4.5715187874960534,
      "grad_norm": 3.5795419216156006,
      "learning_rate": 4.761069398638692e-06,
      "loss": 0.5732,
      "step": 14478
    },
    {
      "epoch": 4.571834543732239,
      "grad_norm": 3.1220011711120605,
      "learning_rate": 4.757560872921199e-06,
      "loss": 0.4628,
      "step": 14479
    },
    {
      "epoch": 4.572150299968424,
      "grad_norm": 4.293022632598877,
      "learning_rate": 4.7540523472037055e-06,
      "loss": 0.5712,
      "step": 14480
    },
    {
      "epoch": 4.57246605620461,
      "grad_norm": 4.517798900604248,
      "learning_rate": 4.750543821486212e-06,
      "loss": 0.541,
      "step": 14481
    },
    {
      "epoch": 4.5727818124407955,
      "grad_norm": 3.7541663646698,
      "learning_rate": 4.747035295768718e-06,
      "loss": 0.6262,
      "step": 14482
    },
    {
      "epoch": 4.573097568676982,
      "grad_norm": 3.720818281173706,
      "learning_rate": 4.743526770051224e-06,
      "loss": 0.6248,
      "step": 14483
    },
    {
      "epoch": 4.573413324913167,
      "grad_norm": 4.529796123504639,
      "learning_rate": 4.740018244333731e-06,
      "loss": 0.6178,
      "step": 14484
    },
    {
      "epoch": 4.573729081149352,
      "grad_norm": 3.5964300632476807,
      "learning_rate": 4.736509718616238e-06,
      "loss": 0.6166,
      "step": 14485
    },
    {
      "epoch": 4.5740448373855385,
      "grad_norm": 6.828291893005371,
      "learning_rate": 4.733001192898745e-06,
      "loss": 0.5231,
      "step": 14486
    },
    {
      "epoch": 4.574360593621724,
      "grad_norm": 5.054775238037109,
      "learning_rate": 4.729492667181251e-06,
      "loss": 0.7044,
      "step": 14487
    },
    {
      "epoch": 4.57467634985791,
      "grad_norm": 4.188446998596191,
      "learning_rate": 4.7259841414637574e-06,
      "loss": 0.6225,
      "step": 14488
    },
    {
      "epoch": 4.574992106094095,
      "grad_norm": 4.696927547454834,
      "learning_rate": 4.722475615746263e-06,
      "loss": 0.6643,
      "step": 14489
    },
    {
      "epoch": 4.5753078623302805,
      "grad_norm": 4.304292678833008,
      "learning_rate": 4.71896709002877e-06,
      "loss": 0.5903,
      "step": 14490
    },
    {
      "epoch": 4.575623618566467,
      "grad_norm": 3.78909969329834,
      "learning_rate": 4.715458564311276e-06,
      "loss": 0.6009,
      "step": 14491
    },
    {
      "epoch": 4.575939374802652,
      "grad_norm": 4.486238479614258,
      "learning_rate": 4.711950038593784e-06,
      "loss": 0.636,
      "step": 14492
    },
    {
      "epoch": 4.576255131038838,
      "grad_norm": 3.7912404537200928,
      "learning_rate": 4.70844151287629e-06,
      "loss": 0.6054,
      "step": 14493
    },
    {
      "epoch": 4.5765708872750235,
      "grad_norm": 3.3655993938446045,
      "learning_rate": 4.704932987158796e-06,
      "loss": 0.604,
      "step": 14494
    },
    {
      "epoch": 4.57688664351121,
      "grad_norm": 4.769927978515625,
      "learning_rate": 4.701424461441303e-06,
      "loss": 0.596,
      "step": 14495
    },
    {
      "epoch": 4.577202399747395,
      "grad_norm": 5.09766149520874,
      "learning_rate": 4.6979159357238085e-06,
      "loss": 0.5886,
      "step": 14496
    },
    {
      "epoch": 4.577518155983581,
      "grad_norm": 4.044453144073486,
      "learning_rate": 4.694407410006315e-06,
      "loss": 0.5628,
      "step": 14497
    },
    {
      "epoch": 4.577833912219766,
      "grad_norm": 4.130097389221191,
      "learning_rate": 4.690898884288822e-06,
      "loss": 0.57,
      "step": 14498
    },
    {
      "epoch": 4.578149668455952,
      "grad_norm": 4.235357761383057,
      "learning_rate": 4.687390358571329e-06,
      "loss": 0.6151,
      "step": 14499
    },
    {
      "epoch": 4.578465424692138,
      "grad_norm": 3.4842379093170166,
      "learning_rate": 4.683881832853835e-06,
      "loss": 0.5498,
      "step": 14500
    },
    {
      "epoch": 4.578781180928323,
      "grad_norm": 3.596278667449951,
      "learning_rate": 4.680373307136342e-06,
      "loss": 0.527,
      "step": 14501
    },
    {
      "epoch": 4.579096937164509,
      "grad_norm": 4.870327472686768,
      "learning_rate": 4.676864781418848e-06,
      "loss": 0.7278,
      "step": 14502
    },
    {
      "epoch": 4.579412693400695,
      "grad_norm": 4.118122100830078,
      "learning_rate": 4.6733562557013545e-06,
      "loss": 0.5832,
      "step": 14503
    },
    {
      "epoch": 4.57972844963688,
      "grad_norm": 3.918203353881836,
      "learning_rate": 4.669847729983861e-06,
      "loss": 0.6139,
      "step": 14504
    },
    {
      "epoch": 4.580044205873066,
      "grad_norm": 3.6792311668395996,
      "learning_rate": 4.666339204266368e-06,
      "loss": 0.7167,
      "step": 14505
    },
    {
      "epoch": 4.5803599621092514,
      "grad_norm": 4.026276111602783,
      "learning_rate": 4.662830678548874e-06,
      "loss": 0.6104,
      "step": 14506
    },
    {
      "epoch": 4.580675718345438,
      "grad_norm": 4.555048942565918,
      "learning_rate": 4.65932215283138e-06,
      "loss": 0.6072,
      "step": 14507
    },
    {
      "epoch": 4.580991474581623,
      "grad_norm": 3.7660868167877197,
      "learning_rate": 4.655813627113887e-06,
      "loss": 0.5964,
      "step": 14508
    },
    {
      "epoch": 4.581307230817808,
      "grad_norm": 3.741461753845215,
      "learning_rate": 4.652305101396393e-06,
      "loss": 0.4838,
      "step": 14509
    },
    {
      "epoch": 4.581622987053994,
      "grad_norm": 3.5786426067352295,
      "learning_rate": 4.6487965756789e-06,
      "loss": 0.5454,
      "step": 14510
    },
    {
      "epoch": 4.58193874329018,
      "grad_norm": 4.336735248565674,
      "learning_rate": 4.6452880499614065e-06,
      "loss": 0.5219,
      "step": 14511
    },
    {
      "epoch": 4.582254499526366,
      "grad_norm": 4.4844560623168945,
      "learning_rate": 4.641779524243913e-06,
      "loss": 0.5745,
      "step": 14512
    },
    {
      "epoch": 4.582570255762551,
      "grad_norm": 5.309310436248779,
      "learning_rate": 4.638270998526419e-06,
      "loss": 0.5852,
      "step": 14513
    },
    {
      "epoch": 4.582886011998737,
      "grad_norm": 4.478442192077637,
      "learning_rate": 4.634762472808926e-06,
      "loss": 0.5307,
      "step": 14514
    },
    {
      "epoch": 4.583201768234923,
      "grad_norm": 3.9722602367401123,
      "learning_rate": 4.631253947091432e-06,
      "loss": 0.6436,
      "step": 14515
    },
    {
      "epoch": 4.583517524471108,
      "grad_norm": 3.5371859073638916,
      "learning_rate": 4.627745421373939e-06,
      "loss": 0.5646,
      "step": 14516
    },
    {
      "epoch": 4.583833280707294,
      "grad_norm": 3.9055986404418945,
      "learning_rate": 4.624236895656446e-06,
      "loss": 0.6447,
      "step": 14517
    },
    {
      "epoch": 4.584149036943479,
      "grad_norm": 3.491511583328247,
      "learning_rate": 4.620728369938952e-06,
      "loss": 0.5222,
      "step": 14518
    },
    {
      "epoch": 4.584464793179666,
      "grad_norm": 4.601942539215088,
      "learning_rate": 4.6172198442214585e-06,
      "loss": 0.6452,
      "step": 14519
    },
    {
      "epoch": 4.584780549415851,
      "grad_norm": 3.517536163330078,
      "learning_rate": 4.6137113185039644e-06,
      "loss": 0.5564,
      "step": 14520
    },
    {
      "epoch": 4.585096305652037,
      "grad_norm": 3.6437950134277344,
      "learning_rate": 4.610202792786471e-06,
      "loss": 0.5526,
      "step": 14521
    },
    {
      "epoch": 4.585412061888222,
      "grad_norm": 4.2242255210876465,
      "learning_rate": 4.606694267068977e-06,
      "loss": 0.7383,
      "step": 14522
    },
    {
      "epoch": 4.585727818124408,
      "grad_norm": 3.7581210136413574,
      "learning_rate": 4.603185741351485e-06,
      "loss": 0.6504,
      "step": 14523
    },
    {
      "epoch": 4.586043574360594,
      "grad_norm": 3.825834274291992,
      "learning_rate": 4.599677215633991e-06,
      "loss": 0.6385,
      "step": 14524
    },
    {
      "epoch": 4.586359330596779,
      "grad_norm": 3.9630298614501953,
      "learning_rate": 4.596168689916498e-06,
      "loss": 0.5943,
      "step": 14525
    },
    {
      "epoch": 4.586675086832965,
      "grad_norm": 3.5891222953796387,
      "learning_rate": 4.592660164199004e-06,
      "loss": 0.5382,
      "step": 14526
    },
    {
      "epoch": 4.586990843069151,
      "grad_norm": 4.308148384094238,
      "learning_rate": 4.5891516384815104e-06,
      "loss": 0.5768,
      "step": 14527
    },
    {
      "epoch": 4.587306599305336,
      "grad_norm": 4.1843581199646,
      "learning_rate": 4.585643112764016e-06,
      "loss": 0.6783,
      "step": 14528
    },
    {
      "epoch": 4.587622355541522,
      "grad_norm": 4.252618789672852,
      "learning_rate": 4.582134587046523e-06,
      "loss": 0.6521,
      "step": 14529
    },
    {
      "epoch": 4.587938111777707,
      "grad_norm": 4.182275295257568,
      "learning_rate": 4.57862606132903e-06,
      "loss": 0.5726,
      "step": 14530
    },
    {
      "epoch": 4.588253868013894,
      "grad_norm": 3.677412271499634,
      "learning_rate": 4.575117535611536e-06,
      "loss": 0.6371,
      "step": 14531
    },
    {
      "epoch": 4.588569624250079,
      "grad_norm": 4.361705780029297,
      "learning_rate": 4.571609009894043e-06,
      "loss": 0.6781,
      "step": 14532
    },
    {
      "epoch": 4.588885380486264,
      "grad_norm": 4.164811134338379,
      "learning_rate": 4.568100484176549e-06,
      "loss": 0.5349,
      "step": 14533
    },
    {
      "epoch": 4.58920113672245,
      "grad_norm": 3.6414830684661865,
      "learning_rate": 4.564591958459056e-06,
      "loss": 0.5985,
      "step": 14534
    },
    {
      "epoch": 4.589516892958636,
      "grad_norm": 3.7833473682403564,
      "learning_rate": 4.561083432741562e-06,
      "loss": 0.6652,
      "step": 14535
    },
    {
      "epoch": 4.589832649194822,
      "grad_norm": 4.169896125793457,
      "learning_rate": 4.557574907024069e-06,
      "loss": 0.5984,
      "step": 14536
    },
    {
      "epoch": 4.590148405431007,
      "grad_norm": 3.4963507652282715,
      "learning_rate": 4.554066381306575e-06,
      "loss": 0.4843,
      "step": 14537
    },
    {
      "epoch": 4.590464161667193,
      "grad_norm": 3.693838357925415,
      "learning_rate": 4.550557855589082e-06,
      "loss": 0.5534,
      "step": 14538
    },
    {
      "epoch": 4.590779917903379,
      "grad_norm": 3.8137433528900146,
      "learning_rate": 4.547049329871588e-06,
      "loss": 0.6019,
      "step": 14539
    },
    {
      "epoch": 4.591095674139564,
      "grad_norm": 3.9879915714263916,
      "learning_rate": 4.543540804154095e-06,
      "loss": 0.5715,
      "step": 14540
    },
    {
      "epoch": 4.59141143037575,
      "grad_norm": 4.17800235748291,
      "learning_rate": 4.540032278436602e-06,
      "loss": 0.6286,
      "step": 14541
    },
    {
      "epoch": 4.591727186611935,
      "grad_norm": 3.954714775085449,
      "learning_rate": 4.5365237527191075e-06,
      "loss": 0.5475,
      "step": 14542
    },
    {
      "epoch": 4.5920429428481215,
      "grad_norm": 4.396486282348633,
      "learning_rate": 4.533015227001614e-06,
      "loss": 0.6701,
      "step": 14543
    },
    {
      "epoch": 4.592358699084307,
      "grad_norm": 4.810366153717041,
      "learning_rate": 4.52950670128412e-06,
      "loss": 0.5734,
      "step": 14544
    },
    {
      "epoch": 4.592674455320493,
      "grad_norm": 3.953450918197632,
      "learning_rate": 4.525998175566627e-06,
      "loss": 0.6113,
      "step": 14545
    },
    {
      "epoch": 4.592990211556678,
      "grad_norm": 3.4838831424713135,
      "learning_rate": 4.522489649849133e-06,
      "loss": 0.5825,
      "step": 14546
    },
    {
      "epoch": 4.593305967792864,
      "grad_norm": 4.33013391494751,
      "learning_rate": 4.51898112413164e-06,
      "loss": 0.6005,
      "step": 14547
    },
    {
      "epoch": 4.59362172402905,
      "grad_norm": 3.843770980834961,
      "learning_rate": 4.515472598414147e-06,
      "loss": 0.4986,
      "step": 14548
    },
    {
      "epoch": 4.593937480265235,
      "grad_norm": 4.243398666381836,
      "learning_rate": 4.5119640726966535e-06,
      "loss": 0.7292,
      "step": 14549
    },
    {
      "epoch": 4.594253236501421,
      "grad_norm": 3.946629285812378,
      "learning_rate": 4.5084555469791595e-06,
      "loss": 0.5437,
      "step": 14550
    },
    {
      "epoch": 4.5945689927376065,
      "grad_norm": 4.799984931945801,
      "learning_rate": 4.504947021261666e-06,
      "loss": 0.6019,
      "step": 14551
    },
    {
      "epoch": 4.594884748973792,
      "grad_norm": 4.564564228057861,
      "learning_rate": 4.501438495544172e-06,
      "loss": 0.6264,
      "step": 14552
    },
    {
      "epoch": 4.595200505209978,
      "grad_norm": 3.9726064205169678,
      "learning_rate": 4.497929969826679e-06,
      "loss": 0.6654,
      "step": 14553
    },
    {
      "epoch": 4.595516261446163,
      "grad_norm": 4.0092854499816895,
      "learning_rate": 4.494421444109186e-06,
      "loss": 0.5845,
      "step": 14554
    },
    {
      "epoch": 4.5958320176823495,
      "grad_norm": 5.368247985839844,
      "learning_rate": 4.490912918391692e-06,
      "loss": 0.5765,
      "step": 14555
    },
    {
      "epoch": 4.596147773918535,
      "grad_norm": 3.7221603393554688,
      "learning_rate": 4.487404392674199e-06,
      "loss": 0.5465,
      "step": 14556
    },
    {
      "epoch": 4.59646353015472,
      "grad_norm": 3.7133021354675293,
      "learning_rate": 4.483895866956705e-06,
      "loss": 0.6536,
      "step": 14557
    },
    {
      "epoch": 4.596779286390906,
      "grad_norm": 3.730468511581421,
      "learning_rate": 4.4803873412392115e-06,
      "loss": 0.5779,
      "step": 14558
    },
    {
      "epoch": 4.597095042627092,
      "grad_norm": 3.7481865882873535,
      "learning_rate": 4.4768788155217174e-06,
      "loss": 0.5615,
      "step": 14559
    },
    {
      "epoch": 4.597410798863278,
      "grad_norm": 3.4458467960357666,
      "learning_rate": 4.473370289804225e-06,
      "loss": 0.5295,
      "step": 14560
    },
    {
      "epoch": 4.597726555099463,
      "grad_norm": 3.9138734340667725,
      "learning_rate": 4.469861764086731e-06,
      "loss": 0.5476,
      "step": 14561
    },
    {
      "epoch": 4.598042311335649,
      "grad_norm": 4.614127159118652,
      "learning_rate": 4.466353238369238e-06,
      "loss": 0.7533,
      "step": 14562
    },
    {
      "epoch": 4.5983580675718345,
      "grad_norm": 3.8353891372680664,
      "learning_rate": 4.462844712651744e-06,
      "loss": 0.547,
      "step": 14563
    },
    {
      "epoch": 4.598673823808021,
      "grad_norm": 4.129733085632324,
      "learning_rate": 4.459336186934251e-06,
      "loss": 0.6253,
      "step": 14564
    },
    {
      "epoch": 4.598989580044206,
      "grad_norm": 3.764408826828003,
      "learning_rate": 4.455827661216757e-06,
      "loss": 0.6106,
      "step": 14565
    },
    {
      "epoch": 4.599305336280391,
      "grad_norm": 4.568472385406494,
      "learning_rate": 4.4523191354992634e-06,
      "loss": 0.6659,
      "step": 14566
    },
    {
      "epoch": 4.5996210925165775,
      "grad_norm": 3.4093098640441895,
      "learning_rate": 4.44881060978177e-06,
      "loss": 0.541,
      "step": 14567
    },
    {
      "epoch": 4.599936848752763,
      "grad_norm": 3.603628635406494,
      "learning_rate": 4.445302084064276e-06,
      "loss": 0.5556,
      "step": 14568
    },
    {
      "epoch": 4.600252604988949,
      "grad_norm": 4.7711381912231445,
      "learning_rate": 4.441793558346783e-06,
      "loss": 0.7294,
      "step": 14569
    },
    {
      "epoch": 4.600568361225134,
      "grad_norm": 4.0102691650390625,
      "learning_rate": 4.438285032629289e-06,
      "loss": 0.6115,
      "step": 14570
    },
    {
      "epoch": 4.6008841174613195,
      "grad_norm": 3.589184284210205,
      "learning_rate": 4.434776506911796e-06,
      "loss": 0.5686,
      "step": 14571
    },
    {
      "epoch": 4.601199873697506,
      "grad_norm": 3.66504168510437,
      "learning_rate": 4.431267981194303e-06,
      "loss": 0.5583,
      "step": 14572
    },
    {
      "epoch": 4.601515629933691,
      "grad_norm": 5.470980644226074,
      "learning_rate": 4.4277594554768094e-06,
      "loss": 0.5956,
      "step": 14573
    },
    {
      "epoch": 4.601831386169877,
      "grad_norm": 4.333103656768799,
      "learning_rate": 4.424250929759315e-06,
      "loss": 0.6376,
      "step": 14574
    },
    {
      "epoch": 4.6021471424060625,
      "grad_norm": 3.5649051666259766,
      "learning_rate": 4.420742404041822e-06,
      "loss": 0.519,
      "step": 14575
    },
    {
      "epoch": 4.602462898642248,
      "grad_norm": 3.4550254344940186,
      "learning_rate": 4.417233878324328e-06,
      "loss": 0.6286,
      "step": 14576
    },
    {
      "epoch": 4.602778654878434,
      "grad_norm": 3.8158841133117676,
      "learning_rate": 4.413725352606834e-06,
      "loss": 0.5457,
      "step": 14577
    },
    {
      "epoch": 4.603094411114619,
      "grad_norm": 4.164060115814209,
      "learning_rate": 4.410216826889341e-06,
      "loss": 0.5421,
      "step": 14578
    },
    {
      "epoch": 4.603410167350805,
      "grad_norm": 3.6067211627960205,
      "learning_rate": 4.406708301171848e-06,
      "loss": 0.5489,
      "step": 14579
    },
    {
      "epoch": 4.603725923586991,
      "grad_norm": 3.908485174179077,
      "learning_rate": 4.403199775454355e-06,
      "loss": 0.6143,
      "step": 14580
    },
    {
      "epoch": 4.604041679823176,
      "grad_norm": 3.765254259109497,
      "learning_rate": 4.3996912497368605e-06,
      "loss": 0.5678,
      "step": 14581
    },
    {
      "epoch": 4.604357436059362,
      "grad_norm": 4.152533054351807,
      "learning_rate": 4.396182724019367e-06,
      "loss": 0.6257,
      "step": 14582
    },
    {
      "epoch": 4.6046731922955475,
      "grad_norm": 3.8260655403137207,
      "learning_rate": 4.392674198301873e-06,
      "loss": 0.7498,
      "step": 14583
    },
    {
      "epoch": 4.604988948531734,
      "grad_norm": 3.9946982860565186,
      "learning_rate": 4.38916567258438e-06,
      "loss": 0.5894,
      "step": 14584
    },
    {
      "epoch": 4.605304704767919,
      "grad_norm": 3.7573697566986084,
      "learning_rate": 4.385657146866887e-06,
      "loss": 0.6114,
      "step": 14585
    },
    {
      "epoch": 4.605620461004105,
      "grad_norm": 3.8287620544433594,
      "learning_rate": 4.382148621149394e-06,
      "loss": 0.5546,
      "step": 14586
    },
    {
      "epoch": 4.6059362172402905,
      "grad_norm": 3.931596040725708,
      "learning_rate": 4.3786400954319e-06,
      "loss": 0.5085,
      "step": 14587
    },
    {
      "epoch": 4.606251973476477,
      "grad_norm": 4.605862617492676,
      "learning_rate": 4.375131569714406e-06,
      "loss": 0.6132,
      "step": 14588
    },
    {
      "epoch": 4.606567729712662,
      "grad_norm": 4.096085548400879,
      "learning_rate": 4.3716230439969125e-06,
      "loss": 0.6112,
      "step": 14589
    },
    {
      "epoch": 4.606883485948847,
      "grad_norm": 3.5370514392852783,
      "learning_rate": 4.368114518279419e-06,
      "loss": 0.5572,
      "step": 14590
    },
    {
      "epoch": 4.607199242185033,
      "grad_norm": 3.5256526470184326,
      "learning_rate": 4.364605992561926e-06,
      "loss": 0.6928,
      "step": 14591
    },
    {
      "epoch": 4.607514998421219,
      "grad_norm": 3.567399740219116,
      "learning_rate": 4.361097466844432e-06,
      "loss": 0.4904,
      "step": 14592
    },
    {
      "epoch": 4.607830754657405,
      "grad_norm": 4.686964511871338,
      "learning_rate": 4.357588941126939e-06,
      "loss": 0.5807,
      "step": 14593
    },
    {
      "epoch": 4.60814651089359,
      "grad_norm": 4.108820915222168,
      "learning_rate": 4.354080415409445e-06,
      "loss": 0.5653,
      "step": 14594
    },
    {
      "epoch": 4.6084622671297755,
      "grad_norm": 2.944596767425537,
      "learning_rate": 4.350571889691952e-06,
      "loss": 0.5364,
      "step": 14595
    },
    {
      "epoch": 4.608778023365962,
      "grad_norm": 3.218442440032959,
      "learning_rate": 4.347063363974458e-06,
      "loss": 0.4367,
      "step": 14596
    },
    {
      "epoch": 4.609093779602147,
      "grad_norm": 4.033816814422607,
      "learning_rate": 4.343554838256965e-06,
      "loss": 0.6209,
      "step": 14597
    },
    {
      "epoch": 4.609409535838333,
      "grad_norm": 3.742136240005493,
      "learning_rate": 4.340046312539471e-06,
      "loss": 0.6859,
      "step": 14598
    },
    {
      "epoch": 4.609725292074518,
      "grad_norm": 5.3440842628479,
      "learning_rate": 4.336537786821978e-06,
      "loss": 0.588,
      "step": 14599
    },
    {
      "epoch": 4.610041048310704,
      "grad_norm": 3.676731824874878,
      "learning_rate": 4.333029261104484e-06,
      "loss": 0.5378,
      "step": 14600
    },
    {
      "epoch": 4.610041048310704,
      "eval_loss": 0.8337777853012085,
      "eval_runtime": 4.6127,
      "eval_samples_per_second": 539.818,
      "eval_steps_per_second": 4.553,
      "step": 14600
    },
    {
      "epoch": 4.61035680454689,
      "grad_norm": 4.9729228019714355,
      "learning_rate": 4.32952073538699e-06,
      "loss": 0.493,
      "step": 14601
    },
    {
      "epoch": 4.610672560783075,
      "grad_norm": 3.608386993408203,
      "learning_rate": 4.326012209669497e-06,
      "loss": 0.5219,
      "step": 14602
    },
    {
      "epoch": 4.610988317019261,
      "grad_norm": 3.421555757522583,
      "learning_rate": 4.322503683952004e-06,
      "loss": 0.5991,
      "step": 14603
    },
    {
      "epoch": 4.611304073255447,
      "grad_norm": 4.356669902801514,
      "learning_rate": 4.3189951582345105e-06,
      "loss": 0.5699,
      "step": 14604
    },
    {
      "epoch": 4.611619829491632,
      "grad_norm": 4.979928016662598,
      "learning_rate": 4.3154866325170164e-06,
      "loss": 0.6278,
      "step": 14605
    },
    {
      "epoch": 4.611935585727818,
      "grad_norm": 4.600497722625732,
      "learning_rate": 4.311978106799523e-06,
      "loss": 0.5988,
      "step": 14606
    },
    {
      "epoch": 4.612251341964003,
      "grad_norm": 4.938187599182129,
      "learning_rate": 4.308469581082029e-06,
      "loss": 0.6602,
      "step": 14607
    },
    {
      "epoch": 4.61256709820019,
      "grad_norm": 3.793621778488159,
      "learning_rate": 4.304961055364536e-06,
      "loss": 0.5741,
      "step": 14608
    },
    {
      "epoch": 4.612882854436375,
      "grad_norm": 5.360620021820068,
      "learning_rate": 4.301452529647043e-06,
      "loss": 0.6472,
      "step": 14609
    },
    {
      "epoch": 4.613198610672561,
      "grad_norm": 4.389658451080322,
      "learning_rate": 4.29794400392955e-06,
      "loss": 0.5603,
      "step": 14610
    },
    {
      "epoch": 4.613514366908746,
      "grad_norm": 4.464621067047119,
      "learning_rate": 4.294435478212056e-06,
      "loss": 0.5675,
      "step": 14611
    },
    {
      "epoch": 4.613830123144933,
      "grad_norm": 3.9331045150756836,
      "learning_rate": 4.290926952494562e-06,
      "loss": 0.6874,
      "step": 14612
    },
    {
      "epoch": 4.614145879381118,
      "grad_norm": 3.647174596786499,
      "learning_rate": 4.287418426777068e-06,
      "loss": 0.562,
      "step": 14613
    },
    {
      "epoch": 4.614461635617303,
      "grad_norm": 3.9603123664855957,
      "learning_rate": 4.283909901059574e-06,
      "loss": 0.5195,
      "step": 14614
    },
    {
      "epoch": 4.614777391853489,
      "grad_norm": 4.063729763031006,
      "learning_rate": 4.280401375342081e-06,
      "loss": 0.5837,
      "step": 14615
    },
    {
      "epoch": 4.615093148089675,
      "grad_norm": 4.004439353942871,
      "learning_rate": 4.276892849624588e-06,
      "loss": 0.6646,
      "step": 14616
    },
    {
      "epoch": 4.615408904325861,
      "grad_norm": 4.122864723205566,
      "learning_rate": 4.273384323907095e-06,
      "loss": 0.5341,
      "step": 14617
    },
    {
      "epoch": 4.615724660562046,
      "grad_norm": 5.305840492248535,
      "learning_rate": 4.269875798189601e-06,
      "loss": 0.6229,
      "step": 14618
    },
    {
      "epoch": 4.616040416798231,
      "grad_norm": 4.215165615081787,
      "learning_rate": 4.266367272472108e-06,
      "loss": 0.6606,
      "step": 14619
    },
    {
      "epoch": 4.616356173034418,
      "grad_norm": 3.324984550476074,
      "learning_rate": 4.2628587467546135e-06,
      "loss": 0.4688,
      "step": 14620
    },
    {
      "epoch": 4.616671929270603,
      "grad_norm": 3.79229474067688,
      "learning_rate": 4.25935022103712e-06,
      "loss": 0.6124,
      "step": 14621
    },
    {
      "epoch": 4.616987685506789,
      "grad_norm": 3.7881178855895996,
      "learning_rate": 4.255841695319627e-06,
      "loss": 0.5209,
      "step": 14622
    },
    {
      "epoch": 4.617303441742974,
      "grad_norm": 3.777039051055908,
      "learning_rate": 4.252333169602133e-06,
      "loss": 0.5292,
      "step": 14623
    },
    {
      "epoch": 4.61761919797916,
      "grad_norm": 3.8081068992614746,
      "learning_rate": 4.24882464388464e-06,
      "loss": 0.657,
      "step": 14624
    },
    {
      "epoch": 4.617934954215346,
      "grad_norm": 3.685213327407837,
      "learning_rate": 4.245316118167146e-06,
      "loss": 0.496,
      "step": 14625
    },
    {
      "epoch": 4.618250710451531,
      "grad_norm": 3.6250951290130615,
      "learning_rate": 4.241807592449653e-06,
      "loss": 0.6671,
      "step": 14626
    },
    {
      "epoch": 4.618566466687717,
      "grad_norm": 3.989450454711914,
      "learning_rate": 4.2382990667321595e-06,
      "loss": 0.6738,
      "step": 14627
    },
    {
      "epoch": 4.618882222923903,
      "grad_norm": 4.792594909667969,
      "learning_rate": 4.234790541014666e-06,
      "loss": 0.643,
      "step": 14628
    },
    {
      "epoch": 4.619197979160089,
      "grad_norm": 4.132406234741211,
      "learning_rate": 4.231282015297172e-06,
      "loss": 0.7116,
      "step": 14629
    },
    {
      "epoch": 4.619513735396274,
      "grad_norm": 4.407720565795898,
      "learning_rate": 4.227773489579679e-06,
      "loss": 0.6113,
      "step": 14630
    },
    {
      "epoch": 4.619829491632459,
      "grad_norm": 3.7884576320648193,
      "learning_rate": 4.224264963862185e-06,
      "loss": 0.5439,
      "step": 14631
    },
    {
      "epoch": 4.6201452478686456,
      "grad_norm": 4.769278526306152,
      "learning_rate": 4.220756438144692e-06,
      "loss": 0.5359,
      "step": 14632
    },
    {
      "epoch": 4.620461004104831,
      "grad_norm": 3.7528367042541504,
      "learning_rate": 4.217247912427198e-06,
      "loss": 0.6423,
      "step": 14633
    },
    {
      "epoch": 4.620776760341017,
      "grad_norm": 4.21599817276001,
      "learning_rate": 4.213739386709705e-06,
      "loss": 0.6811,
      "step": 14634
    },
    {
      "epoch": 4.621092516577202,
      "grad_norm": 4.449836730957031,
      "learning_rate": 4.2102308609922115e-06,
      "loss": 0.6505,
      "step": 14635
    },
    {
      "epoch": 4.6214082728133885,
      "grad_norm": 3.748671531677246,
      "learning_rate": 4.2067223352747175e-06,
      "loss": 0.4889,
      "step": 14636
    },
    {
      "epoch": 4.621724029049574,
      "grad_norm": 4.5819902420043945,
      "learning_rate": 4.203213809557224e-06,
      "loss": 0.5738,
      "step": 14637
    },
    {
      "epoch": 4.622039785285759,
      "grad_norm": 4.120527744293213,
      "learning_rate": 4.19970528383973e-06,
      "loss": 0.5113,
      "step": 14638
    },
    {
      "epoch": 4.622355541521945,
      "grad_norm": 3.263880729675293,
      "learning_rate": 4.196196758122237e-06,
      "loss": 0.5175,
      "step": 14639
    },
    {
      "epoch": 4.622671297758131,
      "grad_norm": 4.293025016784668,
      "learning_rate": 4.192688232404744e-06,
      "loss": 0.6638,
      "step": 14640
    },
    {
      "epoch": 4.622987053994317,
      "grad_norm": 4.432295799255371,
      "learning_rate": 4.189179706687251e-06,
      "loss": 0.5916,
      "step": 14641
    },
    {
      "epoch": 4.623302810230502,
      "grad_norm": 4.466456890106201,
      "learning_rate": 4.185671180969757e-06,
      "loss": 0.5716,
      "step": 14642
    },
    {
      "epoch": 4.623618566466687,
      "grad_norm": 5.1550421714782715,
      "learning_rate": 4.1821626552522635e-06,
      "loss": 0.6222,
      "step": 14643
    },
    {
      "epoch": 4.6239343227028735,
      "grad_norm": 5.230982303619385,
      "learning_rate": 4.1786541295347694e-06,
      "loss": 0.6278,
      "step": 14644
    },
    {
      "epoch": 4.624250078939059,
      "grad_norm": 4.5976080894470215,
      "learning_rate": 4.175145603817276e-06,
      "loss": 0.7006,
      "step": 14645
    },
    {
      "epoch": 4.624565835175245,
      "grad_norm": 3.8528249263763428,
      "learning_rate": 4.171637078099783e-06,
      "loss": 0.5849,
      "step": 14646
    },
    {
      "epoch": 4.62488159141143,
      "grad_norm": 3.266083002090454,
      "learning_rate": 4.168128552382289e-06,
      "loss": 0.4797,
      "step": 14647
    },
    {
      "epoch": 4.625197347647616,
      "grad_norm": 3.9684386253356934,
      "learning_rate": 4.164620026664796e-06,
      "loss": 0.6227,
      "step": 14648
    },
    {
      "epoch": 4.625513103883802,
      "grad_norm": 3.867509365081787,
      "learning_rate": 4.161111500947302e-06,
      "loss": 0.6006,
      "step": 14649
    },
    {
      "epoch": 4.625828860119987,
      "grad_norm": 3.8468058109283447,
      "learning_rate": 4.157602975229809e-06,
      "loss": 0.6001,
      "step": 14650
    },
    {
      "epoch": 4.626144616356173,
      "grad_norm": 3.5816471576690674,
      "learning_rate": 4.154094449512315e-06,
      "loss": 0.6048,
      "step": 14651
    },
    {
      "epoch": 4.6264603725923585,
      "grad_norm": 5.016561985015869,
      "learning_rate": 4.150585923794821e-06,
      "loss": 0.7813,
      "step": 14652
    },
    {
      "epoch": 4.626776128828545,
      "grad_norm": 3.963261127471924,
      "learning_rate": 4.147077398077328e-06,
      "loss": 0.5912,
      "step": 14653
    },
    {
      "epoch": 4.62709188506473,
      "grad_norm": 5.074929714202881,
      "learning_rate": 4.143568872359835e-06,
      "loss": 0.6342,
      "step": 14654
    },
    {
      "epoch": 4.627407641300915,
      "grad_norm": 4.042488098144531,
      "learning_rate": 4.140060346642341e-06,
      "loss": 0.557,
      "step": 14655
    },
    {
      "epoch": 4.6277233975371015,
      "grad_norm": 3.874706506729126,
      "learning_rate": 4.136551820924848e-06,
      "loss": 0.7695,
      "step": 14656
    },
    {
      "epoch": 4.628039153773287,
      "grad_norm": 4.666489124298096,
      "learning_rate": 4.133043295207354e-06,
      "loss": 0.6405,
      "step": 14657
    },
    {
      "epoch": 4.628354910009473,
      "grad_norm": 4.224051475524902,
      "learning_rate": 4.129534769489861e-06,
      "loss": 0.5704,
      "step": 14658
    },
    {
      "epoch": 4.628670666245658,
      "grad_norm": 4.248129367828369,
      "learning_rate": 4.126026243772367e-06,
      "loss": 0.601,
      "step": 14659
    },
    {
      "epoch": 4.628986422481844,
      "grad_norm": 3.7065422534942627,
      "learning_rate": 4.122517718054873e-06,
      "loss": 0.5902,
      "step": 14660
    },
    {
      "epoch": 4.62930217871803,
      "grad_norm": 3.866407871246338,
      "learning_rate": 4.11900919233738e-06,
      "loss": 0.6302,
      "step": 14661
    },
    {
      "epoch": 4.629617934954215,
      "grad_norm": 5.405722141265869,
      "learning_rate": 4.115500666619886e-06,
      "loss": 0.5813,
      "step": 14662
    },
    {
      "epoch": 4.629933691190401,
      "grad_norm": 3.447232246398926,
      "learning_rate": 4.111992140902393e-06,
      "loss": 0.5641,
      "step": 14663
    },
    {
      "epoch": 4.6302494474265865,
      "grad_norm": 3.7708849906921387,
      "learning_rate": 4.108483615184899e-06,
      "loss": 0.5934,
      "step": 14664
    },
    {
      "epoch": 4.630565203662773,
      "grad_norm": 3.9583523273468018,
      "learning_rate": 4.104975089467407e-06,
      "loss": 0.5088,
      "step": 14665
    },
    {
      "epoch": 4.630880959898958,
      "grad_norm": 3.3780922889709473,
      "learning_rate": 4.1014665637499125e-06,
      "loss": 0.5703,
      "step": 14666
    },
    {
      "epoch": 4.631196716135143,
      "grad_norm": 4.358719825744629,
      "learning_rate": 4.097958038032419e-06,
      "loss": 0.5834,
      "step": 14667
    },
    {
      "epoch": 4.6315124723713295,
      "grad_norm": 4.002765655517578,
      "learning_rate": 4.094449512314925e-06,
      "loss": 0.6414,
      "step": 14668
    },
    {
      "epoch": 4.631828228607515,
      "grad_norm": 3.340843915939331,
      "learning_rate": 4.090940986597432e-06,
      "loss": 0.5621,
      "step": 14669
    },
    {
      "epoch": 4.632143984843701,
      "grad_norm": 3.4833390712738037,
      "learning_rate": 4.087432460879938e-06,
      "loss": 0.5411,
      "step": 14670
    },
    {
      "epoch": 4.632459741079886,
      "grad_norm": 4.296187877655029,
      "learning_rate": 4.083923935162445e-06,
      "loss": 0.6294,
      "step": 14671
    },
    {
      "epoch": 4.6327754973160715,
      "grad_norm": 6.629968643188477,
      "learning_rate": 4.080415409444952e-06,
      "loss": 0.618,
      "step": 14672
    },
    {
      "epoch": 4.633091253552258,
      "grad_norm": 5.263952732086182,
      "learning_rate": 4.076906883727458e-06,
      "loss": 0.608,
      "step": 14673
    },
    {
      "epoch": 4.633407009788443,
      "grad_norm": 4.1362762451171875,
      "learning_rate": 4.0733983580099645e-06,
      "loss": 0.6356,
      "step": 14674
    },
    {
      "epoch": 4.633722766024629,
      "grad_norm": 3.4912397861480713,
      "learning_rate": 4.0698898322924705e-06,
      "loss": 0.5214,
      "step": 14675
    },
    {
      "epoch": 4.6340385222608145,
      "grad_norm": 3.309252977371216,
      "learning_rate": 4.066381306574977e-06,
      "loss": 0.4865,
      "step": 14676
    },
    {
      "epoch": 4.634354278497001,
      "grad_norm": 3.846752882003784,
      "learning_rate": 4.062872780857484e-06,
      "loss": 0.6146,
      "step": 14677
    },
    {
      "epoch": 4.634670034733186,
      "grad_norm": 4.593318462371826,
      "learning_rate": 4.059364255139991e-06,
      "loss": 0.6013,
      "step": 14678
    },
    {
      "epoch": 4.634985790969372,
      "grad_norm": 5.1525373458862305,
      "learning_rate": 4.055855729422497e-06,
      "loss": 0.598,
      "step": 14679
    },
    {
      "epoch": 4.635301547205557,
      "grad_norm": 4.335960388183594,
      "learning_rate": 4.052347203705004e-06,
      "loss": 0.6375,
      "step": 14680
    },
    {
      "epoch": 4.635617303441743,
      "grad_norm": 4.299702167510986,
      "learning_rate": 4.04883867798751e-06,
      "loss": 0.629,
      "step": 14681
    },
    {
      "epoch": 4.635933059677929,
      "grad_norm": 4.9890851974487305,
      "learning_rate": 4.045330152270016e-06,
      "loss": 0.5721,
      "step": 14682
    },
    {
      "epoch": 4.636248815914114,
      "grad_norm": 4.005427837371826,
      "learning_rate": 4.041821626552523e-06,
      "loss": 0.5212,
      "step": 14683
    },
    {
      "epoch": 4.6365645721503,
      "grad_norm": 4.142138957977295,
      "learning_rate": 4.038313100835029e-06,
      "loss": 0.7012,
      "step": 14684
    },
    {
      "epoch": 4.636880328386486,
      "grad_norm": 4.151571273803711,
      "learning_rate": 4.034804575117536e-06,
      "loss": 0.621,
      "step": 14685
    },
    {
      "epoch": 4.637196084622671,
      "grad_norm": 4.355069637298584,
      "learning_rate": 4.031296049400042e-06,
      "loss": 0.5198,
      "step": 14686
    },
    {
      "epoch": 4.637511840858857,
      "grad_norm": 4.015106678009033,
      "learning_rate": 4.027787523682549e-06,
      "loss": 0.5888,
      "step": 14687
    },
    {
      "epoch": 4.637827597095042,
      "grad_norm": 3.8201780319213867,
      "learning_rate": 4.024278997965055e-06,
      "loss": 0.5776,
      "step": 14688
    },
    {
      "epoch": 4.638143353331229,
      "grad_norm": 3.943756103515625,
      "learning_rate": 4.020770472247562e-06,
      "loss": 0.5257,
      "step": 14689
    },
    {
      "epoch": 4.638459109567414,
      "grad_norm": 4.668959140777588,
      "learning_rate": 4.0172619465300684e-06,
      "loss": 0.5306,
      "step": 14690
    },
    {
      "epoch": 4.638774865803599,
      "grad_norm": 3.493609666824341,
      "learning_rate": 4.013753420812575e-06,
      "loss": 0.534,
      "step": 14691
    },
    {
      "epoch": 4.639090622039785,
      "grad_norm": 5.80646276473999,
      "learning_rate": 4.010244895095081e-06,
      "loss": 0.6366,
      "step": 14692
    },
    {
      "epoch": 4.639406378275971,
      "grad_norm": 4.652851581573486,
      "learning_rate": 4.006736369377588e-06,
      "loss": 0.5879,
      "step": 14693
    },
    {
      "epoch": 4.639722134512157,
      "grad_norm": 4.48603630065918,
      "learning_rate": 4.003227843660094e-06,
      "loss": 0.4883,
      "step": 14694
    },
    {
      "epoch": 4.640037890748342,
      "grad_norm": 3.8410632610321045,
      "learning_rate": 3.999719317942601e-06,
      "loss": 0.5946,
      "step": 14695
    },
    {
      "epoch": 4.6403536469845275,
      "grad_norm": 3.487946033477783,
      "learning_rate": 3.996210792225108e-06,
      "loss": 0.4986,
      "step": 14696
    },
    {
      "epoch": 4.640669403220714,
      "grad_norm": 3.8772480487823486,
      "learning_rate": 3.992702266507614e-06,
      "loss": 0.6334,
      "step": 14697
    },
    {
      "epoch": 4.640985159456899,
      "grad_norm": 7.505639553070068,
      "learning_rate": 3.98919374079012e-06,
      "loss": 0.5878,
      "step": 14698
    },
    {
      "epoch": 4.641300915693085,
      "grad_norm": 4.530345439910889,
      "learning_rate": 3.985685215072626e-06,
      "loss": 0.5575,
      "step": 14699
    },
    {
      "epoch": 4.64161667192927,
      "grad_norm": 3.642380714416504,
      "learning_rate": 3.982176689355133e-06,
      "loss": 0.5501,
      "step": 14700
    },
    {
      "epoch": 4.641932428165457,
      "grad_norm": 3.819302558898926,
      "learning_rate": 3.978668163637639e-06,
      "loss": 0.5913,
      "step": 14701
    },
    {
      "epoch": 4.642248184401642,
      "grad_norm": 4.077757358551025,
      "learning_rate": 3.975159637920147e-06,
      "loss": 0.5357,
      "step": 14702
    },
    {
      "epoch": 4.642563940637828,
      "grad_norm": 3.558302402496338,
      "learning_rate": 3.971651112202653e-06,
      "loss": 0.5776,
      "step": 14703
    },
    {
      "epoch": 4.642879696874013,
      "grad_norm": 4.2455644607543945,
      "learning_rate": 3.96814258648516e-06,
      "loss": 0.5965,
      "step": 14704
    },
    {
      "epoch": 4.643195453110199,
      "grad_norm": 4.224452495574951,
      "learning_rate": 3.9646340607676655e-06,
      "loss": 0.544,
      "step": 14705
    },
    {
      "epoch": 4.643511209346385,
      "grad_norm": 4.119162559509277,
      "learning_rate": 3.9611255350501715e-06,
      "loss": 0.5135,
      "step": 14706
    },
    {
      "epoch": 4.64382696558257,
      "grad_norm": 3.7635741233825684,
      "learning_rate": 3.957617009332678e-06,
      "loss": 0.6001,
      "step": 14707
    },
    {
      "epoch": 4.644142721818756,
      "grad_norm": 4.208201885223389,
      "learning_rate": 3.954108483615185e-06,
      "loss": 0.5326,
      "step": 14708
    },
    {
      "epoch": 4.644458478054942,
      "grad_norm": 3.6172356605529785,
      "learning_rate": 3.950599957897692e-06,
      "loss": 0.6776,
      "step": 14709
    },
    {
      "epoch": 4.644774234291127,
      "grad_norm": 3.956591844558716,
      "learning_rate": 3.947091432180198e-06,
      "loss": 0.5238,
      "step": 14710
    },
    {
      "epoch": 4.645089990527313,
      "grad_norm": 3.548848867416382,
      "learning_rate": 3.943582906462705e-06,
      "loss": 0.5295,
      "step": 14711
    },
    {
      "epoch": 4.645405746763498,
      "grad_norm": 4.4314656257629395,
      "learning_rate": 3.940074380745211e-06,
      "loss": 0.6178,
      "step": 14712
    },
    {
      "epoch": 4.645721502999685,
      "grad_norm": 3.7439658641815186,
      "learning_rate": 3.9365658550277175e-06,
      "loss": 0.5745,
      "step": 14713
    },
    {
      "epoch": 4.64603725923587,
      "grad_norm": 4.198018550872803,
      "learning_rate": 3.933057329310224e-06,
      "loss": 0.5244,
      "step": 14714
    },
    {
      "epoch": 4.646353015472055,
      "grad_norm": 3.680821657180786,
      "learning_rate": 3.929548803592731e-06,
      "loss": 0.5293,
      "step": 14715
    },
    {
      "epoch": 4.646668771708241,
      "grad_norm": 4.613710403442383,
      "learning_rate": 3.926040277875237e-06,
      "loss": 0.6448,
      "step": 14716
    },
    {
      "epoch": 4.646984527944427,
      "grad_norm": 3.7959163188934326,
      "learning_rate": 3.922531752157744e-06,
      "loss": 0.57,
      "step": 14717
    },
    {
      "epoch": 4.647300284180613,
      "grad_norm": 3.172006607055664,
      "learning_rate": 3.91902322644025e-06,
      "loss": 0.5789,
      "step": 14718
    },
    {
      "epoch": 4.647616040416798,
      "grad_norm": 4.286215305328369,
      "learning_rate": 3.915514700722756e-06,
      "loss": 0.7163,
      "step": 14719
    },
    {
      "epoch": 4.647931796652984,
      "grad_norm": 4.458975315093994,
      "learning_rate": 3.912006175005263e-06,
      "loss": 0.5394,
      "step": 14720
    },
    {
      "epoch": 4.64824755288917,
      "grad_norm": 3.2604525089263916,
      "learning_rate": 3.9084976492877695e-06,
      "loss": 0.5871,
      "step": 14721
    },
    {
      "epoch": 4.648563309125355,
      "grad_norm": 4.749927520751953,
      "learning_rate": 3.904989123570276e-06,
      "loss": 0.6037,
      "step": 14722
    },
    {
      "epoch": 4.648879065361541,
      "grad_norm": 3.9787142276763916,
      "learning_rate": 3.901480597852782e-06,
      "loss": 0.5735,
      "step": 14723
    },
    {
      "epoch": 4.649194821597726,
      "grad_norm": 4.444304943084717,
      "learning_rate": 3.897972072135289e-06,
      "loss": 0.5803,
      "step": 14724
    },
    {
      "epoch": 4.6495105778339125,
      "grad_norm": 4.878418445587158,
      "learning_rate": 3.894463546417795e-06,
      "loss": 0.568,
      "step": 14725
    },
    {
      "epoch": 4.649826334070098,
      "grad_norm": 4.099466800689697,
      "learning_rate": 3.890955020700302e-06,
      "loss": 0.6309,
      "step": 14726
    },
    {
      "epoch": 4.650142090306284,
      "grad_norm": 4.701895236968994,
      "learning_rate": 3.887446494982809e-06,
      "loss": 0.6662,
      "step": 14727
    },
    {
      "epoch": 4.650457846542469,
      "grad_norm": 3.495816230773926,
      "learning_rate": 3.8839379692653155e-06,
      "loss": 0.621,
      "step": 14728
    },
    {
      "epoch": 4.650773602778655,
      "grad_norm": 4.16528844833374,
      "learning_rate": 3.8804294435478214e-06,
      "loss": 0.5407,
      "step": 14729
    },
    {
      "epoch": 4.651089359014841,
      "grad_norm": 3.9717609882354736,
      "learning_rate": 3.876920917830327e-06,
      "loss": 0.6709,
      "step": 14730
    },
    {
      "epoch": 4.651405115251026,
      "grad_norm": 4.785255432128906,
      "learning_rate": 3.873412392112834e-06,
      "loss": 0.6878,
      "step": 14731
    },
    {
      "epoch": 4.651720871487212,
      "grad_norm": 2.970073938369751,
      "learning_rate": 3.869903866395341e-06,
      "loss": 0.5027,
      "step": 14732
    },
    {
      "epoch": 4.6520366277233975,
      "grad_norm": 4.09835147857666,
      "learning_rate": 3.866395340677848e-06,
      "loss": 0.6881,
      "step": 14733
    },
    {
      "epoch": 4.652352383959583,
      "grad_norm": 3.7483227252960205,
      "learning_rate": 3.862886814960354e-06,
      "loss": 0.5016,
      "step": 14734
    },
    {
      "epoch": 4.652668140195769,
      "grad_norm": 3.645097255706787,
      "learning_rate": 3.859378289242861e-06,
      "loss": 0.5957,
      "step": 14735
    },
    {
      "epoch": 4.652983896431954,
      "grad_norm": 4.058794975280762,
      "learning_rate": 3.855869763525367e-06,
      "loss": 0.6341,
      "step": 14736
    },
    {
      "epoch": 4.6532996526681405,
      "grad_norm": 3.9072115421295166,
      "learning_rate": 3.852361237807873e-06,
      "loss": 0.507,
      "step": 14737
    },
    {
      "epoch": 4.653615408904326,
      "grad_norm": 4.770783424377441,
      "learning_rate": 3.848852712090379e-06,
      "loss": 0.6694,
      "step": 14738
    },
    {
      "epoch": 4.653931165140511,
      "grad_norm": 3.8966054916381836,
      "learning_rate": 3.845344186372886e-06,
      "loss": 0.5688,
      "step": 14739
    },
    {
      "epoch": 4.654246921376697,
      "grad_norm": 4.852351665496826,
      "learning_rate": 3.841835660655393e-06,
      "loss": 0.6789,
      "step": 14740
    },
    {
      "epoch": 4.654562677612883,
      "grad_norm": 3.5674219131469727,
      "learning_rate": 3.838327134937899e-06,
      "loss": 0.4813,
      "step": 14741
    },
    {
      "epoch": 4.654878433849069,
      "grad_norm": 4.824102878570557,
      "learning_rate": 3.834818609220406e-06,
      "loss": 0.6041,
      "step": 14742
    },
    {
      "epoch": 4.655194190085254,
      "grad_norm": 4.422215938568115,
      "learning_rate": 3.831310083502912e-06,
      "loss": 0.544,
      "step": 14743
    },
    {
      "epoch": 4.65550994632144,
      "grad_norm": 3.979238271713257,
      "learning_rate": 3.8278015577854185e-06,
      "loss": 0.557,
      "step": 14744
    },
    {
      "epoch": 4.6558257025576255,
      "grad_norm": 4.112220287322998,
      "learning_rate": 3.824293032067925e-06,
      "loss": 0.5778,
      "step": 14745
    },
    {
      "epoch": 4.656141458793811,
      "grad_norm": 4.21977424621582,
      "learning_rate": 3.820784506350432e-06,
      "loss": 0.6085,
      "step": 14746
    },
    {
      "epoch": 4.656457215029997,
      "grad_norm": 4.686943531036377,
      "learning_rate": 3.817275980632938e-06,
      "loss": 0.5477,
      "step": 14747
    },
    {
      "epoch": 4.656772971266182,
      "grad_norm": 4.564230918884277,
      "learning_rate": 3.813767454915445e-06,
      "loss": 0.6271,
      "step": 14748
    },
    {
      "epoch": 4.6570887275023685,
      "grad_norm": 4.312307834625244,
      "learning_rate": 3.810258929197951e-06,
      "loss": 0.6217,
      "step": 14749
    },
    {
      "epoch": 4.657404483738554,
      "grad_norm": 5.122549533843994,
      "learning_rate": 3.806750403480458e-06,
      "loss": 0.5612,
      "step": 14750
    },
    {
      "epoch": 4.65772023997474,
      "grad_norm": 4.7691731452941895,
      "learning_rate": 3.803241877762964e-06,
      "loss": 0.6364,
      "step": 14751
    },
    {
      "epoch": 4.658035996210925,
      "grad_norm": 4.007424831390381,
      "learning_rate": 3.799733352045471e-06,
      "loss": 0.5678,
      "step": 14752
    },
    {
      "epoch": 4.6583517524471105,
      "grad_norm": 4.862699031829834,
      "learning_rate": 3.7962248263279773e-06,
      "loss": 0.6166,
      "step": 14753
    },
    {
      "epoch": 4.658667508683297,
      "grad_norm": 4.227355003356934,
      "learning_rate": 3.7927163006104833e-06,
      "loss": 0.6656,
      "step": 14754
    },
    {
      "epoch": 4.658983264919482,
      "grad_norm": 4.733191013336182,
      "learning_rate": 3.78920777489299e-06,
      "loss": 0.633,
      "step": 14755
    },
    {
      "epoch": 4.659299021155668,
      "grad_norm": 3.573784828186035,
      "learning_rate": 3.7856992491754965e-06,
      "loss": 0.5914,
      "step": 14756
    },
    {
      "epoch": 4.6596147773918535,
      "grad_norm": 3.8232719898223877,
      "learning_rate": 3.7821907234580033e-06,
      "loss": 0.5789,
      "step": 14757
    },
    {
      "epoch": 4.659930533628039,
      "grad_norm": 4.781667232513428,
      "learning_rate": 3.7786821977405093e-06,
      "loss": 0.5814,
      "step": 14758
    },
    {
      "epoch": 4.660246289864225,
      "grad_norm": 3.865081787109375,
      "learning_rate": 3.7751736720230165e-06,
      "loss": 0.6374,
      "step": 14759
    },
    {
      "epoch": 4.66056204610041,
      "grad_norm": 3.6880922317504883,
      "learning_rate": 3.7716651463055225e-06,
      "loss": 0.6088,
      "step": 14760
    },
    {
      "epoch": 4.660877802336596,
      "grad_norm": 3.533730983734131,
      "learning_rate": 3.7681566205880293e-06,
      "loss": 0.5137,
      "step": 14761
    },
    {
      "epoch": 4.661193558572782,
      "grad_norm": 4.031193256378174,
      "learning_rate": 3.7646480948705357e-06,
      "loss": 0.4786,
      "step": 14762
    },
    {
      "epoch": 4.661509314808967,
      "grad_norm": 9.050088882446289,
      "learning_rate": 3.7611395691530425e-06,
      "loss": 0.6427,
      "step": 14763
    },
    {
      "epoch": 4.661825071045153,
      "grad_norm": 4.156612873077393,
      "learning_rate": 3.7576310434355485e-06,
      "loss": 0.6209,
      "step": 14764
    },
    {
      "epoch": 4.6621408272813385,
      "grad_norm": 4.427711009979248,
      "learning_rate": 3.754122517718055e-06,
      "loss": 0.6392,
      "step": 14765
    },
    {
      "epoch": 4.662456583517525,
      "grad_norm": 3.673203706741333,
      "learning_rate": 3.7506139920005617e-06,
      "loss": 0.4884,
      "step": 14766
    },
    {
      "epoch": 4.66277233975371,
      "grad_norm": 3.50309681892395,
      "learning_rate": 3.7471054662830676e-06,
      "loss": 0.5579,
      "step": 14767
    },
    {
      "epoch": 4.663088095989896,
      "grad_norm": 3.6334519386291504,
      "learning_rate": 3.7435969405655744e-06,
      "loss": 0.5891,
      "step": 14768
    },
    {
      "epoch": 4.6634038522260814,
      "grad_norm": 4.218853950500488,
      "learning_rate": 3.740088414848081e-06,
      "loss": 0.5389,
      "step": 14769
    },
    {
      "epoch": 4.663719608462267,
      "grad_norm": 3.327366590499878,
      "learning_rate": 3.7365798891305876e-06,
      "loss": 0.5398,
      "step": 14770
    },
    {
      "epoch": 4.664035364698453,
      "grad_norm": 4.654067039489746,
      "learning_rate": 3.733071363413094e-06,
      "loss": 0.7797,
      "step": 14771
    },
    {
      "epoch": 4.664351120934638,
      "grad_norm": 4.940653324127197,
      "learning_rate": 3.729562837695601e-06,
      "loss": 0.6213,
      "step": 14772
    },
    {
      "epoch": 4.664666877170824,
      "grad_norm": 4.075131416320801,
      "learning_rate": 3.726054311978107e-06,
      "loss": 0.6229,
      "step": 14773
    },
    {
      "epoch": 4.66498263340701,
      "grad_norm": 4.1929779052734375,
      "learning_rate": 3.7225457862606136e-06,
      "loss": 0.5778,
      "step": 14774
    },
    {
      "epoch": 4.665298389643196,
      "grad_norm": 4.161958694458008,
      "learning_rate": 3.71903726054312e-06,
      "loss": 0.6246,
      "step": 14775
    },
    {
      "epoch": 4.665614145879381,
      "grad_norm": 4.094471454620361,
      "learning_rate": 3.715528734825626e-06,
      "loss": 0.5738,
      "step": 14776
    },
    {
      "epoch": 4.6659299021155665,
      "grad_norm": 4.2926225662231445,
      "learning_rate": 3.7120202091081328e-06,
      "loss": 0.6687,
      "step": 14777
    },
    {
      "epoch": 4.666245658351753,
      "grad_norm": 4.266173362731934,
      "learning_rate": 3.708511683390639e-06,
      "loss": 0.4661,
      "step": 14778
    },
    {
      "epoch": 4.666561414587938,
      "grad_norm": 4.049131393432617,
      "learning_rate": 3.705003157673146e-06,
      "loss": 0.5296,
      "step": 14779
    },
    {
      "epoch": 4.666877170824124,
      "grad_norm": 3.2057366371154785,
      "learning_rate": 3.7014946319556524e-06,
      "loss": 0.5448,
      "step": 14780
    },
    {
      "epoch": 4.667192927060309,
      "grad_norm": 4.433960914611816,
      "learning_rate": 3.697986106238159e-06,
      "loss": 0.5449,
      "step": 14781
    },
    {
      "epoch": 4.667508683296495,
      "grad_norm": 4.074885368347168,
      "learning_rate": 3.694477580520665e-06,
      "loss": 0.5718,
      "step": 14782
    },
    {
      "epoch": 4.667824439532681,
      "grad_norm": 4.443231582641602,
      "learning_rate": 3.690969054803172e-06,
      "loss": 0.5843,
      "step": 14783
    },
    {
      "epoch": 4.668140195768866,
      "grad_norm": 4.175821304321289,
      "learning_rate": 3.6874605290856784e-06,
      "loss": 0.6113,
      "step": 14784
    },
    {
      "epoch": 4.668455952005052,
      "grad_norm": 4.335111141204834,
      "learning_rate": 3.683952003368185e-06,
      "loss": 0.5426,
      "step": 14785
    },
    {
      "epoch": 4.668771708241238,
      "grad_norm": 4.51227331161499,
      "learning_rate": 3.680443477650691e-06,
      "loss": 0.6763,
      "step": 14786
    },
    {
      "epoch": 4.669087464477423,
      "grad_norm": 4.707023620605469,
      "learning_rate": 3.6769349519331984e-06,
      "loss": 0.6003,
      "step": 14787
    },
    {
      "epoch": 4.669403220713609,
      "grad_norm": 3.750417709350586,
      "learning_rate": 3.6734264262157043e-06,
      "loss": 0.5649,
      "step": 14788
    },
    {
      "epoch": 4.669718976949794,
      "grad_norm": 4.098324298858643,
      "learning_rate": 3.6699179004982107e-06,
      "loss": 0.5306,
      "step": 14789
    },
    {
      "epoch": 4.670034733185981,
      "grad_norm": 4.839395046234131,
      "learning_rate": 3.6664093747807175e-06,
      "loss": 0.6164,
      "step": 14790
    },
    {
      "epoch": 4.670350489422166,
      "grad_norm": 3.148343563079834,
      "learning_rate": 3.6629008490632235e-06,
      "loss": 0.5182,
      "step": 14791
    },
    {
      "epoch": 4.670666245658352,
      "grad_norm": 3.82334041595459,
      "learning_rate": 3.6593923233457303e-06,
      "loss": 0.5447,
      "step": 14792
    },
    {
      "epoch": 4.670982001894537,
      "grad_norm": 3.334102153778076,
      "learning_rate": 3.6558837976282367e-06,
      "loss": 0.5432,
      "step": 14793
    },
    {
      "epoch": 4.671297758130724,
      "grad_norm": 4.4676127433776855,
      "learning_rate": 3.6523752719107435e-06,
      "loss": 0.6393,
      "step": 14794
    },
    {
      "epoch": 4.671613514366909,
      "grad_norm": 4.391270160675049,
      "learning_rate": 3.6488667461932495e-06,
      "loss": 0.6532,
      "step": 14795
    },
    {
      "epoch": 4.671929270603094,
      "grad_norm": 4.456188678741455,
      "learning_rate": 3.6453582204757563e-06,
      "loss": 0.577,
      "step": 14796
    },
    {
      "epoch": 4.67224502683928,
      "grad_norm": 4.827840328216553,
      "learning_rate": 3.6418496947582627e-06,
      "loss": 0.6924,
      "step": 14797
    },
    {
      "epoch": 4.672560783075466,
      "grad_norm": 3.953664779663086,
      "learning_rate": 3.6383411690407695e-06,
      "loss": 0.6522,
      "step": 14798
    },
    {
      "epoch": 4.672876539311652,
      "grad_norm": 4.344305515289307,
      "learning_rate": 3.634832643323276e-06,
      "loss": 0.7305,
      "step": 14799
    },
    {
      "epoch": 4.673192295547837,
      "grad_norm": 3.9164066314697266,
      "learning_rate": 3.631324117605782e-06,
      "loss": 0.512,
      "step": 14800
    },
    {
      "epoch": 4.673192295547837,
      "eval_loss": 0.8365557193756104,
      "eval_runtime": 4.6087,
      "eval_samples_per_second": 540.285,
      "eval_steps_per_second": 4.557,
      "step": 14800
    },
    {
      "epoch": 4.673508051784022,
      "grad_norm": 3.437401533126831,
      "learning_rate": 3.6278155918882887e-06,
      "loss": 0.5534,
      "step": 14801
    },
    {
      "epoch": 4.673823808020209,
      "grad_norm": 3.9194037914276123,
      "learning_rate": 3.624307066170795e-06,
      "loss": 0.5342,
      "step": 14802
    },
    {
      "epoch": 4.674139564256394,
      "grad_norm": 4.400981426239014,
      "learning_rate": 3.620798540453302e-06,
      "loss": 0.6361,
      "step": 14803
    },
    {
      "epoch": 4.67445532049258,
      "grad_norm": 5.060398101806641,
      "learning_rate": 3.617290014735808e-06,
      "loss": 0.5942,
      "step": 14804
    },
    {
      "epoch": 4.674771076728765,
      "grad_norm": 4.288723468780518,
      "learning_rate": 3.6137814890183147e-06,
      "loss": 0.6175,
      "step": 14805
    },
    {
      "epoch": 4.675086832964951,
      "grad_norm": 4.246452808380127,
      "learning_rate": 3.610272963300821e-06,
      "loss": 0.5639,
      "step": 14806
    },
    {
      "epoch": 4.675402589201137,
      "grad_norm": 4.550182819366455,
      "learning_rate": 3.606764437583328e-06,
      "loss": 0.5822,
      "step": 14807
    },
    {
      "epoch": 4.675718345437322,
      "grad_norm": 3.8044722080230713,
      "learning_rate": 3.6032559118658342e-06,
      "loss": 0.6525,
      "step": 14808
    },
    {
      "epoch": 4.676034101673508,
      "grad_norm": 4.01586389541626,
      "learning_rate": 3.599747386148341e-06,
      "loss": 0.6536,
      "step": 14809
    },
    {
      "epoch": 4.676349857909694,
      "grad_norm": 3.9294941425323486,
      "learning_rate": 3.596238860430847e-06,
      "loss": 0.62,
      "step": 14810
    },
    {
      "epoch": 4.676665614145879,
      "grad_norm": 5.6557769775390625,
      "learning_rate": 3.592730334713354e-06,
      "loss": 0.633,
      "step": 14811
    },
    {
      "epoch": 4.676981370382065,
      "grad_norm": 3.5762596130371094,
      "learning_rate": 3.5892218089958602e-06,
      "loss": 0.5536,
      "step": 14812
    },
    {
      "epoch": 4.67729712661825,
      "grad_norm": 6.505899429321289,
      "learning_rate": 3.585713283278366e-06,
      "loss": 0.5612,
      "step": 14813
    },
    {
      "epoch": 4.6776128828544365,
      "grad_norm": 4.025099754333496,
      "learning_rate": 3.582204757560873e-06,
      "loss": 0.6512,
      "step": 14814
    },
    {
      "epoch": 4.677928639090622,
      "grad_norm": 4.595228672027588,
      "learning_rate": 3.5786962318433794e-06,
      "loss": 0.6652,
      "step": 14815
    },
    {
      "epoch": 4.678244395326808,
      "grad_norm": 4.80617618560791,
      "learning_rate": 3.5751877061258862e-06,
      "loss": 0.6127,
      "step": 14816
    },
    {
      "epoch": 4.678560151562993,
      "grad_norm": 4.0391693115234375,
      "learning_rate": 3.571679180408392e-06,
      "loss": 0.6079,
      "step": 14817
    },
    {
      "epoch": 4.6788759077991795,
      "grad_norm": 4.838788986206055,
      "learning_rate": 3.5681706546908994e-06,
      "loss": 0.5243,
      "step": 14818
    },
    {
      "epoch": 4.679191664035365,
      "grad_norm": 3.914456605911255,
      "learning_rate": 3.5646621289734054e-06,
      "loss": 0.589,
      "step": 14819
    },
    {
      "epoch": 4.67950742027155,
      "grad_norm": 4.5817790031433105,
      "learning_rate": 3.561153603255912e-06,
      "loss": 0.624,
      "step": 14820
    },
    {
      "epoch": 4.679823176507736,
      "grad_norm": 4.464916706085205,
      "learning_rate": 3.5576450775384186e-06,
      "loss": 0.6047,
      "step": 14821
    },
    {
      "epoch": 4.680138932743922,
      "grad_norm": 3.820011615753174,
      "learning_rate": 3.5541365518209254e-06,
      "loss": 0.5413,
      "step": 14822
    },
    {
      "epoch": 4.680454688980108,
      "grad_norm": 5.884524822235107,
      "learning_rate": 3.5506280261034314e-06,
      "loss": 0.5389,
      "step": 14823
    },
    {
      "epoch": 4.680770445216293,
      "grad_norm": 3.9068644046783447,
      "learning_rate": 3.5471195003859378e-06,
      "loss": 0.5208,
      "step": 14824
    },
    {
      "epoch": 4.681086201452478,
      "grad_norm": 4.365403652191162,
      "learning_rate": 3.5436109746684446e-06,
      "loss": 0.693,
      "step": 14825
    },
    {
      "epoch": 4.6814019576886645,
      "grad_norm": 3.790097951889038,
      "learning_rate": 3.5401024489509505e-06,
      "loss": 0.5255,
      "step": 14826
    },
    {
      "epoch": 4.68171771392485,
      "grad_norm": 3.6951119899749756,
      "learning_rate": 3.5365939232334578e-06,
      "loss": 0.5521,
      "step": 14827
    },
    {
      "epoch": 4.682033470161036,
      "grad_norm": 3.641354560852051,
      "learning_rate": 3.5330853975159637e-06,
      "loss": 0.5704,
      "step": 14828
    },
    {
      "epoch": 4.682349226397221,
      "grad_norm": 3.8731777667999268,
      "learning_rate": 3.5295768717984705e-06,
      "loss": 0.6992,
      "step": 14829
    },
    {
      "epoch": 4.682664982633407,
      "grad_norm": 4.230210781097412,
      "learning_rate": 3.526068346080977e-06,
      "loss": 0.5899,
      "step": 14830
    },
    {
      "epoch": 4.682980738869593,
      "grad_norm": 4.214406967163086,
      "learning_rate": 3.5225598203634837e-06,
      "loss": 0.5977,
      "step": 14831
    },
    {
      "epoch": 4.683296495105778,
      "grad_norm": 5.190972328186035,
      "learning_rate": 3.5190512946459897e-06,
      "loss": 0.6673,
      "step": 14832
    },
    {
      "epoch": 4.683612251341964,
      "grad_norm": 4.545443058013916,
      "learning_rate": 3.5155427689284965e-06,
      "loss": 0.6964,
      "step": 14833
    },
    {
      "epoch": 4.6839280075781495,
      "grad_norm": 5.3796491622924805,
      "learning_rate": 3.512034243211003e-06,
      "loss": 0.7406,
      "step": 14834
    },
    {
      "epoch": 4.684243763814336,
      "grad_norm": 4.246572971343994,
      "learning_rate": 3.508525717493509e-06,
      "loss": 0.5796,
      "step": 14835
    },
    {
      "epoch": 4.684559520050521,
      "grad_norm": 3.3474104404449463,
      "learning_rate": 3.505017191776016e-06,
      "loss": 0.5131,
      "step": 14836
    },
    {
      "epoch": 4.684875276286706,
      "grad_norm": 4.205259799957275,
      "learning_rate": 3.501508666058522e-06,
      "loss": 0.6119,
      "step": 14837
    },
    {
      "epoch": 4.6851910325228925,
      "grad_norm": 3.821805477142334,
      "learning_rate": 3.498000140341029e-06,
      "loss": 0.7126,
      "step": 14838
    },
    {
      "epoch": 4.685506788759078,
      "grad_norm": 3.963472843170166,
      "learning_rate": 3.4944916146235353e-06,
      "loss": 0.4573,
      "step": 14839
    },
    {
      "epoch": 4.685822544995264,
      "grad_norm": 4.122939109802246,
      "learning_rate": 3.490983088906042e-06,
      "loss": 0.4662,
      "step": 14840
    },
    {
      "epoch": 4.686138301231449,
      "grad_norm": 5.0413994789123535,
      "learning_rate": 3.487474563188548e-06,
      "loss": 0.5972,
      "step": 14841
    },
    {
      "epoch": 4.686454057467635,
      "grad_norm": 3.9481749534606934,
      "learning_rate": 3.483966037471055e-06,
      "loss": 0.5924,
      "step": 14842
    },
    {
      "epoch": 4.686769813703821,
      "grad_norm": 4.001840114593506,
      "learning_rate": 3.4804575117535613e-06,
      "loss": 0.4517,
      "step": 14843
    },
    {
      "epoch": 4.687085569940006,
      "grad_norm": 4.729637622833252,
      "learning_rate": 3.476948986036068e-06,
      "loss": 0.6066,
      "step": 14844
    },
    {
      "epoch": 4.687401326176192,
      "grad_norm": 3.2807438373565674,
      "learning_rate": 3.473440460318574e-06,
      "loss": 0.5473,
      "step": 14845
    },
    {
      "epoch": 4.6877170824123775,
      "grad_norm": 4.321842193603516,
      "learning_rate": 3.4699319346010813e-06,
      "loss": 0.5084,
      "step": 14846
    },
    {
      "epoch": 4.688032838648564,
      "grad_norm": 3.324596643447876,
      "learning_rate": 3.4664234088835873e-06,
      "loss": 0.5189,
      "step": 14847
    },
    {
      "epoch": 4.688348594884749,
      "grad_norm": 3.6574578285217285,
      "learning_rate": 3.4629148831660936e-06,
      "loss": 0.5067,
      "step": 14848
    },
    {
      "epoch": 4.688664351120934,
      "grad_norm": 3.622377634048462,
      "learning_rate": 3.4594063574486005e-06,
      "loss": 0.4946,
      "step": 14849
    },
    {
      "epoch": 4.6889801073571205,
      "grad_norm": 3.5056726932525635,
      "learning_rate": 3.4558978317311064e-06,
      "loss": 0.469,
      "step": 14850
    },
    {
      "epoch": 4.689295863593306,
      "grad_norm": 4.29641580581665,
      "learning_rate": 3.4523893060136132e-06,
      "loss": 0.605,
      "step": 14851
    },
    {
      "epoch": 4.689611619829492,
      "grad_norm": 2.9862310886383057,
      "learning_rate": 3.4488807802961196e-06,
      "loss": 0.5816,
      "step": 14852
    },
    {
      "epoch": 4.689927376065677,
      "grad_norm": 3.425429344177246,
      "learning_rate": 3.4453722545786264e-06,
      "loss": 0.5698,
      "step": 14853
    },
    {
      "epoch": 4.6902431323018625,
      "grad_norm": 4.260678291320801,
      "learning_rate": 3.4418637288611324e-06,
      "loss": 0.6923,
      "step": 14854
    },
    {
      "epoch": 4.690558888538049,
      "grad_norm": 4.024963855743408,
      "learning_rate": 3.4383552031436396e-06,
      "loss": 0.5782,
      "step": 14855
    },
    {
      "epoch": 4.690874644774234,
      "grad_norm": 3.8739702701568604,
      "learning_rate": 3.4348466774261456e-06,
      "loss": 0.4533,
      "step": 14856
    },
    {
      "epoch": 4.69119040101042,
      "grad_norm": 4.158016681671143,
      "learning_rate": 3.4313381517086524e-06,
      "loss": 0.6157,
      "step": 14857
    },
    {
      "epoch": 4.6915061572466055,
      "grad_norm": 3.9593076705932617,
      "learning_rate": 3.427829625991159e-06,
      "loss": 0.5983,
      "step": 14858
    },
    {
      "epoch": 4.691821913482792,
      "grad_norm": 4.22740364074707,
      "learning_rate": 3.4243211002736648e-06,
      "loss": 0.6352,
      "step": 14859
    },
    {
      "epoch": 4.692137669718977,
      "grad_norm": 4.988489627838135,
      "learning_rate": 3.4208125745561716e-06,
      "loss": 0.602,
      "step": 14860
    },
    {
      "epoch": 4.692453425955162,
      "grad_norm": 4.204953193664551,
      "learning_rate": 3.417304048838678e-06,
      "loss": 0.538,
      "step": 14861
    },
    {
      "epoch": 4.692769182191348,
      "grad_norm": 3.473691940307617,
      "learning_rate": 3.4137955231211848e-06,
      "loss": 0.4805,
      "step": 14862
    },
    {
      "epoch": 4.693084938427534,
      "grad_norm": 4.077582359313965,
      "learning_rate": 3.4102869974036908e-06,
      "loss": 0.581,
      "step": 14863
    },
    {
      "epoch": 4.69340069466372,
      "grad_norm": 5.312047958374023,
      "learning_rate": 3.406778471686198e-06,
      "loss": 0.6903,
      "step": 14864
    },
    {
      "epoch": 4.693716450899905,
      "grad_norm": 4.089005947113037,
      "learning_rate": 3.403269945968704e-06,
      "loss": 0.6901,
      "step": 14865
    },
    {
      "epoch": 4.694032207136091,
      "grad_norm": 4.064828872680664,
      "learning_rate": 3.3997614202512108e-06,
      "loss": 0.5992,
      "step": 14866
    },
    {
      "epoch": 4.694347963372277,
      "grad_norm": 4.258151054382324,
      "learning_rate": 3.396252894533717e-06,
      "loss": 0.5593,
      "step": 14867
    },
    {
      "epoch": 4.694663719608462,
      "grad_norm": 3.5730340480804443,
      "learning_rate": 3.392744368816224e-06,
      "loss": 0.5604,
      "step": 14868
    },
    {
      "epoch": 4.694979475844648,
      "grad_norm": 3.5754222869873047,
      "learning_rate": 3.38923584309873e-06,
      "loss": 0.6195,
      "step": 14869
    },
    {
      "epoch": 4.695295232080833,
      "grad_norm": 3.830650806427002,
      "learning_rate": 3.3857273173812367e-06,
      "loss": 0.6183,
      "step": 14870
    },
    {
      "epoch": 4.69561098831702,
      "grad_norm": 4.762722492218018,
      "learning_rate": 3.382218791663743e-06,
      "loss": 0.6505,
      "step": 14871
    },
    {
      "epoch": 4.695926744553205,
      "grad_norm": 4.328084945678711,
      "learning_rate": 3.378710265946249e-06,
      "loss": 0.6341,
      "step": 14872
    },
    {
      "epoch": 4.69624250078939,
      "grad_norm": 4.983715534210205,
      "learning_rate": 3.375201740228756e-06,
      "loss": 0.6066,
      "step": 14873
    },
    {
      "epoch": 4.696558257025576,
      "grad_norm": 3.038759708404541,
      "learning_rate": 3.3716932145112623e-06,
      "loss": 0.5386,
      "step": 14874
    },
    {
      "epoch": 4.696874013261762,
      "grad_norm": 3.9126927852630615,
      "learning_rate": 3.368184688793769e-06,
      "loss": 0.5991,
      "step": 14875
    },
    {
      "epoch": 4.697189769497948,
      "grad_norm": 4.2259416580200195,
      "learning_rate": 3.3646761630762755e-06,
      "loss": 0.6763,
      "step": 14876
    },
    {
      "epoch": 4.697505525734133,
      "grad_norm": 4.166280746459961,
      "learning_rate": 3.3611676373587823e-06,
      "loss": 0.5364,
      "step": 14877
    },
    {
      "epoch": 4.6978212819703185,
      "grad_norm": 3.981475591659546,
      "learning_rate": 3.3576591116412883e-06,
      "loss": 0.5744,
      "step": 14878
    },
    {
      "epoch": 4.698137038206505,
      "grad_norm": 3.843625068664551,
      "learning_rate": 3.354150585923795e-06,
      "loss": 0.5983,
      "step": 14879
    },
    {
      "epoch": 4.69845279444269,
      "grad_norm": 4.025243282318115,
      "learning_rate": 3.3506420602063015e-06,
      "loss": 0.5242,
      "step": 14880
    },
    {
      "epoch": 4.698768550678876,
      "grad_norm": 3.4467978477478027,
      "learning_rate": 3.3471335344888083e-06,
      "loss": 0.5239,
      "step": 14881
    },
    {
      "epoch": 4.699084306915061,
      "grad_norm": 3.5186705589294434,
      "learning_rate": 3.3436250087713143e-06,
      "loss": 0.5317,
      "step": 14882
    },
    {
      "epoch": 4.699400063151248,
      "grad_norm": 4.021280288696289,
      "learning_rate": 3.3401164830538207e-06,
      "loss": 0.5646,
      "step": 14883
    },
    {
      "epoch": 4.699715819387433,
      "grad_norm": 4.603508949279785,
      "learning_rate": 3.3366079573363275e-06,
      "loss": 0.5846,
      "step": 14884
    },
    {
      "epoch": 4.700031575623619,
      "grad_norm": 4.204372882843018,
      "learning_rate": 3.333099431618834e-06,
      "loss": 0.5544,
      "step": 14885
    },
    {
      "epoch": 4.700347331859804,
      "grad_norm": 4.020535945892334,
      "learning_rate": 3.3295909059013407e-06,
      "loss": 0.5462,
      "step": 14886
    },
    {
      "epoch": 4.70066308809599,
      "grad_norm": 3.7080793380737305,
      "learning_rate": 3.3260823801838466e-06,
      "loss": 0.6115,
      "step": 14887
    },
    {
      "epoch": 4.700978844332176,
      "grad_norm": 3.795811414718628,
      "learning_rate": 3.3225738544663535e-06,
      "loss": 0.5893,
      "step": 14888
    },
    {
      "epoch": 4.701294600568361,
      "grad_norm": 3.6405563354492188,
      "learning_rate": 3.31906532874886e-06,
      "loss": 0.6048,
      "step": 14889
    },
    {
      "epoch": 4.701610356804547,
      "grad_norm": 5.317929744720459,
      "learning_rate": 3.3155568030313667e-06,
      "loss": 0.6745,
      "step": 14890
    },
    {
      "epoch": 4.701926113040733,
      "grad_norm": 4.00362491607666,
      "learning_rate": 3.3120482773138726e-06,
      "loss": 0.5999,
      "step": 14891
    },
    {
      "epoch": 4.702241869276918,
      "grad_norm": 4.737085342407227,
      "learning_rate": 3.30853975159638e-06,
      "loss": 0.6898,
      "step": 14892
    },
    {
      "epoch": 4.702557625513104,
      "grad_norm": 5.015619277954102,
      "learning_rate": 3.305031225878886e-06,
      "loss": 0.6016,
      "step": 14893
    },
    {
      "epoch": 4.702873381749289,
      "grad_norm": 3.5389764308929443,
      "learning_rate": 3.3015227001613922e-06,
      "loss": 0.5293,
      "step": 14894
    },
    {
      "epoch": 4.7031891379854756,
      "grad_norm": 4.15170955657959,
      "learning_rate": 3.298014174443899e-06,
      "loss": 0.5858,
      "step": 14895
    },
    {
      "epoch": 4.703504894221661,
      "grad_norm": 4.041644096374512,
      "learning_rate": 3.294505648726405e-06,
      "loss": 0.4475,
      "step": 14896
    },
    {
      "epoch": 4.703820650457846,
      "grad_norm": 3.682974100112915,
      "learning_rate": 3.290997123008912e-06,
      "loss": 0.5494,
      "step": 14897
    },
    {
      "epoch": 4.704136406694032,
      "grad_norm": 4.1321187019348145,
      "learning_rate": 3.287488597291418e-06,
      "loss": 0.6294,
      "step": 14898
    },
    {
      "epoch": 4.704452162930218,
      "grad_norm": 4.713131904602051,
      "learning_rate": 3.283980071573925e-06,
      "loss": 0.6343,
      "step": 14899
    },
    {
      "epoch": 4.704767919166404,
      "grad_norm": 4.068668842315674,
      "learning_rate": 3.280471545856431e-06,
      "loss": 0.5399,
      "step": 14900
    },
    {
      "epoch": 4.705083675402589,
      "grad_norm": 5.316087245941162,
      "learning_rate": 3.2769630201389378e-06,
      "loss": 0.5969,
      "step": 14901
    },
    {
      "epoch": 4.705399431638774,
      "grad_norm": 3.6580300331115723,
      "learning_rate": 3.273454494421444e-06,
      "loss": 0.5107,
      "step": 14902
    },
    {
      "epoch": 4.705715187874961,
      "grad_norm": 3.5889458656311035,
      "learning_rate": 3.269945968703951e-06,
      "loss": 0.5687,
      "step": 14903
    },
    {
      "epoch": 4.706030944111146,
      "grad_norm": 4.630335330963135,
      "learning_rate": 3.2664374429864574e-06,
      "loss": 0.5397,
      "step": 14904
    },
    {
      "epoch": 4.706346700347332,
      "grad_norm": 4.799405574798584,
      "learning_rate": 3.262928917268964e-06,
      "loss": 0.5595,
      "step": 14905
    },
    {
      "epoch": 4.706662456583517,
      "grad_norm": 3.230482578277588,
      "learning_rate": 3.25942039155147e-06,
      "loss": 0.5246,
      "step": 14906
    },
    {
      "epoch": 4.7069782128197035,
      "grad_norm": 3.31278395652771,
      "learning_rate": 3.2559118658339765e-06,
      "loss": 0.5132,
      "step": 14907
    },
    {
      "epoch": 4.707293969055889,
      "grad_norm": 3.5990309715270996,
      "learning_rate": 3.2524033401164834e-06,
      "loss": 0.5233,
      "step": 14908
    },
    {
      "epoch": 4.707609725292075,
      "grad_norm": 3.3913180828094482,
      "learning_rate": 3.2488948143989893e-06,
      "loss": 0.5867,
      "step": 14909
    },
    {
      "epoch": 4.70792548152826,
      "grad_norm": 4.358034610748291,
      "learning_rate": 3.245386288681496e-06,
      "loss": 0.5901,
      "step": 14910
    },
    {
      "epoch": 4.708241237764446,
      "grad_norm": 3.80433988571167,
      "learning_rate": 3.2418777629640025e-06,
      "loss": 0.5811,
      "step": 14911
    },
    {
      "epoch": 4.708556994000632,
      "grad_norm": 4.342733383178711,
      "learning_rate": 3.2383692372465093e-06,
      "loss": 0.6176,
      "step": 14912
    },
    {
      "epoch": 4.708872750236817,
      "grad_norm": 4.764159679412842,
      "learning_rate": 3.2348607115290157e-06,
      "loss": 0.6918,
      "step": 14913
    },
    {
      "epoch": 4.709188506473003,
      "grad_norm": 3.384223222732544,
      "learning_rate": 3.2313521858115225e-06,
      "loss": 0.5577,
      "step": 14914
    },
    {
      "epoch": 4.7095042627091885,
      "grad_norm": 5.616222858428955,
      "learning_rate": 3.2278436600940285e-06,
      "loss": 0.6196,
      "step": 14915
    },
    {
      "epoch": 4.709820018945374,
      "grad_norm": 3.9817543029785156,
      "learning_rate": 3.2243351343765353e-06,
      "loss": 0.6024,
      "step": 14916
    },
    {
      "epoch": 4.71013577518156,
      "grad_norm": 4.091338634490967,
      "learning_rate": 3.2208266086590417e-06,
      "loss": 0.6335,
      "step": 14917
    },
    {
      "epoch": 4.710451531417745,
      "grad_norm": 4.318331718444824,
      "learning_rate": 3.2173180829415477e-06,
      "loss": 0.6387,
      "step": 14918
    },
    {
      "epoch": 4.7107672876539315,
      "grad_norm": 3.270026683807373,
      "learning_rate": 3.2138095572240545e-06,
      "loss": 0.6487,
      "step": 14919
    },
    {
      "epoch": 4.711083043890117,
      "grad_norm": 3.568528890609741,
      "learning_rate": 3.210301031506561e-06,
      "loss": 0.5488,
      "step": 14920
    },
    {
      "epoch": 4.711398800126302,
      "grad_norm": 3.662940502166748,
      "learning_rate": 3.2067925057890677e-06,
      "loss": 0.555,
      "step": 14921
    },
    {
      "epoch": 4.711714556362488,
      "grad_norm": 4.167731761932373,
      "learning_rate": 3.203283980071574e-06,
      "loss": 0.5118,
      "step": 14922
    },
    {
      "epoch": 4.7120303125986736,
      "grad_norm": 4.944979667663574,
      "learning_rate": 3.199775454354081e-06,
      "loss": 0.7021,
      "step": 14923
    },
    {
      "epoch": 4.71234606883486,
      "grad_norm": 4.219743728637695,
      "learning_rate": 3.196266928636587e-06,
      "loss": 0.6347,
      "step": 14924
    },
    {
      "epoch": 4.712661825071045,
      "grad_norm": 5.307825565338135,
      "learning_rate": 3.1927584029190937e-06,
      "loss": 0.5629,
      "step": 14925
    },
    {
      "epoch": 4.71297758130723,
      "grad_norm": 3.8231008052825928,
      "learning_rate": 3.1892498772016e-06,
      "loss": 0.534,
      "step": 14926
    },
    {
      "epoch": 4.7132933375434165,
      "grad_norm": 3.747596502304077,
      "learning_rate": 3.185741351484107e-06,
      "loss": 0.5598,
      "step": 14927
    },
    {
      "epoch": 4.713609093779602,
      "grad_norm": 4.377797603607178,
      "learning_rate": 3.182232825766613e-06,
      "loss": 0.688,
      "step": 14928
    },
    {
      "epoch": 4.713924850015788,
      "grad_norm": 2.9494686126708984,
      "learning_rate": 3.1787243000491192e-06,
      "loss": 0.523,
      "step": 14929
    },
    {
      "epoch": 4.714240606251973,
      "grad_norm": 5.259079456329346,
      "learning_rate": 3.175215774331626e-06,
      "loss": 0.4921,
      "step": 14930
    },
    {
      "epoch": 4.7145563624881595,
      "grad_norm": 4.202796459197998,
      "learning_rate": 3.171707248614132e-06,
      "loss": 0.5977,
      "step": 14931
    },
    {
      "epoch": 4.714872118724345,
      "grad_norm": 4.235088348388672,
      "learning_rate": 3.1681987228966392e-06,
      "loss": 0.6717,
      "step": 14932
    },
    {
      "epoch": 4.715187874960531,
      "grad_norm": 3.4327147006988525,
      "learning_rate": 3.1646901971791452e-06,
      "loss": 0.6604,
      "step": 14933
    },
    {
      "epoch": 4.715503631196716,
      "grad_norm": 4.181224822998047,
      "learning_rate": 3.161181671461652e-06,
      "loss": 0.6061,
      "step": 14934
    },
    {
      "epoch": 4.7158193874329015,
      "grad_norm": 4.636715412139893,
      "learning_rate": 3.1576731457441584e-06,
      "loss": 0.6355,
      "step": 14935
    },
    {
      "epoch": 4.716135143669088,
      "grad_norm": 4.881145477294922,
      "learning_rate": 3.1541646200266652e-06,
      "loss": 0.7095,
      "step": 14936
    },
    {
      "epoch": 4.716450899905273,
      "grad_norm": 4.187020778656006,
      "learning_rate": 3.150656094309171e-06,
      "loss": 0.5872,
      "step": 14937
    },
    {
      "epoch": 4.716766656141459,
      "grad_norm": 3.712939739227295,
      "learning_rate": 3.147147568591678e-06,
      "loss": 0.4847,
      "step": 14938
    },
    {
      "epoch": 4.7170824123776445,
      "grad_norm": 4.172686576843262,
      "learning_rate": 3.1436390428741844e-06,
      "loss": 0.5301,
      "step": 14939
    },
    {
      "epoch": 4.71739816861383,
      "grad_norm": 4.1654815673828125,
      "learning_rate": 3.1401305171566912e-06,
      "loss": 0.5734,
      "step": 14940
    },
    {
      "epoch": 4.717713924850016,
      "grad_norm": 5.016501426696777,
      "learning_rate": 3.1366219914391976e-06,
      "loss": 0.6118,
      "step": 14941
    },
    {
      "epoch": 4.718029681086201,
      "grad_norm": 3.522808074951172,
      "learning_rate": 3.1331134657217036e-06,
      "loss": 0.5882,
      "step": 14942
    },
    {
      "epoch": 4.718345437322387,
      "grad_norm": 4.195000171661377,
      "learning_rate": 3.1296049400042104e-06,
      "loss": 0.6265,
      "step": 14943
    },
    {
      "epoch": 4.718661193558573,
      "grad_norm": 4.450666427612305,
      "learning_rate": 3.1260964142867168e-06,
      "loss": 0.5675,
      "step": 14944
    },
    {
      "epoch": 4.718976949794758,
      "grad_norm": 4.916399955749512,
      "learning_rate": 3.1225878885692236e-06,
      "loss": 0.5514,
      "step": 14945
    },
    {
      "epoch": 4.719292706030944,
      "grad_norm": 4.065402030944824,
      "learning_rate": 3.11907936285173e-06,
      "loss": 0.6301,
      "step": 14946
    },
    {
      "epoch": 4.7196084622671295,
      "grad_norm": 3.959419012069702,
      "learning_rate": 3.1155708371342364e-06,
      "loss": 0.6189,
      "step": 14947
    },
    {
      "epoch": 4.719924218503316,
      "grad_norm": 4.073864936828613,
      "learning_rate": 3.1120623114167428e-06,
      "loss": 0.5875,
      "step": 14948
    },
    {
      "epoch": 4.720239974739501,
      "grad_norm": 4.261752605438232,
      "learning_rate": 3.108553785699249e-06,
      "loss": 0.5392,
      "step": 14949
    },
    {
      "epoch": 4.720555730975687,
      "grad_norm": 3.4837582111358643,
      "learning_rate": 3.105045259981756e-06,
      "loss": 0.5045,
      "step": 14950
    },
    {
      "epoch": 4.720871487211872,
      "grad_norm": 4.205557823181152,
      "learning_rate": 3.1015367342642623e-06,
      "loss": 0.5997,
      "step": 14951
    },
    {
      "epoch": 4.721187243448058,
      "grad_norm": 4.035665988922119,
      "learning_rate": 3.0980282085467687e-06,
      "loss": 0.5989,
      "step": 14952
    },
    {
      "epoch": 4.721502999684244,
      "grad_norm": 4.485807418823242,
      "learning_rate": 3.094519682829275e-06,
      "loss": 0.6174,
      "step": 14953
    },
    {
      "epoch": 4.721818755920429,
      "grad_norm": 4.168167591094971,
      "learning_rate": 3.091011157111782e-06,
      "loss": 0.6727,
      "step": 14954
    },
    {
      "epoch": 4.722134512156615,
      "grad_norm": 4.578367233276367,
      "learning_rate": 3.0875026313942883e-06,
      "loss": 0.5887,
      "step": 14955
    },
    {
      "epoch": 4.722450268392801,
      "grad_norm": 5.366611003875732,
      "learning_rate": 3.0839941056767947e-06,
      "loss": 0.6233,
      "step": 14956
    },
    {
      "epoch": 4.722766024628987,
      "grad_norm": 4.0840163230896,
      "learning_rate": 3.0804855799593015e-06,
      "loss": 0.655,
      "step": 14957
    },
    {
      "epoch": 4.723081780865172,
      "grad_norm": 3.841087818145752,
      "learning_rate": 3.076977054241808e-06,
      "loss": 0.6653,
      "step": 14958
    },
    {
      "epoch": 4.7233975371013575,
      "grad_norm": 4.1761932373046875,
      "learning_rate": 3.073468528524314e-06,
      "loss": 0.596,
      "step": 14959
    },
    {
      "epoch": 4.723713293337544,
      "grad_norm": 4.565335750579834,
      "learning_rate": 3.0699600028068207e-06,
      "loss": 0.595,
      "step": 14960
    },
    {
      "epoch": 4.724029049573729,
      "grad_norm": 3.415348768234253,
      "learning_rate": 3.066451477089327e-06,
      "loss": 0.5423,
      "step": 14961
    },
    {
      "epoch": 4.724344805809915,
      "grad_norm": 5.270749568939209,
      "learning_rate": 3.0629429513718335e-06,
      "loss": 0.7062,
      "step": 14962
    },
    {
      "epoch": 4.7246605620461,
      "grad_norm": 4.205593585968018,
      "learning_rate": 3.0594344256543403e-06,
      "loss": 0.6272,
      "step": 14963
    },
    {
      "epoch": 4.724976318282286,
      "grad_norm": 4.244753360748291,
      "learning_rate": 3.0559258999368467e-06,
      "loss": 0.5598,
      "step": 14964
    },
    {
      "epoch": 4.725292074518472,
      "grad_norm": 3.6909701824188232,
      "learning_rate": 3.052417374219353e-06,
      "loss": 0.5971,
      "step": 14965
    },
    {
      "epoch": 4.725607830754657,
      "grad_norm": 4.719940185546875,
      "learning_rate": 3.04890884850186e-06,
      "loss": 0.6103,
      "step": 14966
    },
    {
      "epoch": 4.725923586990843,
      "grad_norm": 4.910126209259033,
      "learning_rate": 3.0454003227843663e-06,
      "loss": 0.5419,
      "step": 14967
    },
    {
      "epoch": 4.726239343227029,
      "grad_norm": 4.5049543380737305,
      "learning_rate": 3.0418917970668727e-06,
      "loss": 0.6214,
      "step": 14968
    },
    {
      "epoch": 4.726555099463214,
      "grad_norm": 4.285879611968994,
      "learning_rate": 3.0383832713493795e-06,
      "loss": 0.6787,
      "step": 14969
    },
    {
      "epoch": 4.7268708556994,
      "grad_norm": 3.9798460006713867,
      "learning_rate": 3.034874745631886e-06,
      "loss": 0.6119,
      "step": 14970
    },
    {
      "epoch": 4.727186611935585,
      "grad_norm": 3.7946271896362305,
      "learning_rate": 3.031366219914392e-06,
      "loss": 0.5841,
      "step": 14971
    },
    {
      "epoch": 4.727502368171772,
      "grad_norm": 4.339445114135742,
      "learning_rate": 3.0278576941968986e-06,
      "loss": 0.6459,
      "step": 14972
    },
    {
      "epoch": 4.727818124407957,
      "grad_norm": 4.723415374755859,
      "learning_rate": 3.024349168479405e-06,
      "loss": 0.6253,
      "step": 14973
    },
    {
      "epoch": 4.728133880644143,
      "grad_norm": 4.580994606018066,
      "learning_rate": 3.0208406427619114e-06,
      "loss": 0.625,
      "step": 14974
    },
    {
      "epoch": 4.728449636880328,
      "grad_norm": 4.202271938323975,
      "learning_rate": 3.0173321170444182e-06,
      "loss": 0.5328,
      "step": 14975
    },
    {
      "epoch": 4.728765393116514,
      "grad_norm": 3.7245278358459473,
      "learning_rate": 3.0138235913269246e-06,
      "loss": 0.5473,
      "step": 14976
    },
    {
      "epoch": 4.7290811493527,
      "grad_norm": 3.6578102111816406,
      "learning_rate": 3.010315065609431e-06,
      "loss": 0.5388,
      "step": 14977
    },
    {
      "epoch": 4.729396905588885,
      "grad_norm": 3.7082581520080566,
      "learning_rate": 3.006806539891938e-06,
      "loss": 0.582,
      "step": 14978
    },
    {
      "epoch": 4.729712661825071,
      "grad_norm": 3.9611077308654785,
      "learning_rate": 3.0032980141744442e-06,
      "loss": 0.538,
      "step": 14979
    },
    {
      "epoch": 4.730028418061257,
      "grad_norm": 3.8652074337005615,
      "learning_rate": 2.9997894884569506e-06,
      "loss": 0.5584,
      "step": 14980
    },
    {
      "epoch": 4.730344174297443,
      "grad_norm": 3.9195544719696045,
      "learning_rate": 2.996280962739457e-06,
      "loss": 0.6046,
      "step": 14981
    },
    {
      "epoch": 4.730659930533628,
      "grad_norm": 3.621807098388672,
      "learning_rate": 2.9927724370219634e-06,
      "loss": 0.5624,
      "step": 14982
    },
    {
      "epoch": 4.730975686769813,
      "grad_norm": 3.6997647285461426,
      "learning_rate": 2.9892639113044698e-06,
      "loss": 0.6062,
      "step": 14983
    },
    {
      "epoch": 4.731291443006,
      "grad_norm": 4.177933216094971,
      "learning_rate": 2.9857553855869766e-06,
      "loss": 0.5611,
      "step": 14984
    },
    {
      "epoch": 4.731607199242185,
      "grad_norm": 4.69673490524292,
      "learning_rate": 2.982246859869483e-06,
      "loss": 0.7181,
      "step": 14985
    },
    {
      "epoch": 4.731922955478371,
      "grad_norm": 4.6733479499816895,
      "learning_rate": 2.9787383341519894e-06,
      "loss": 0.5927,
      "step": 14986
    },
    {
      "epoch": 4.732238711714556,
      "grad_norm": 3.926905393600464,
      "learning_rate": 2.9752298084344958e-06,
      "loss": 0.5882,
      "step": 14987
    },
    {
      "epoch": 4.732554467950742,
      "grad_norm": 3.7041897773742676,
      "learning_rate": 2.9717212827170026e-06,
      "loss": 0.5908,
      "step": 14988
    },
    {
      "epoch": 4.732870224186928,
      "grad_norm": 3.836733102798462,
      "learning_rate": 2.968212756999509e-06,
      "loss": 0.5097,
      "step": 14989
    },
    {
      "epoch": 4.733185980423113,
      "grad_norm": 3.757199287414551,
      "learning_rate": 2.9647042312820153e-06,
      "loss": 0.5856,
      "step": 14990
    },
    {
      "epoch": 4.733501736659299,
      "grad_norm": 3.5690033435821533,
      "learning_rate": 2.961195705564522e-06,
      "loss": 0.6084,
      "step": 14991
    },
    {
      "epoch": 4.733817492895485,
      "grad_norm": 3.941892385482788,
      "learning_rate": 2.9576871798470285e-06,
      "loss": 0.4697,
      "step": 14992
    },
    {
      "epoch": 4.73413324913167,
      "grad_norm": 4.4143877029418945,
      "learning_rate": 2.954178654129535e-06,
      "loss": 0.6068,
      "step": 14993
    },
    {
      "epoch": 4.734449005367856,
      "grad_norm": 4.167727470397949,
      "learning_rate": 2.9506701284120413e-06,
      "loss": 0.6062,
      "step": 14994
    },
    {
      "epoch": 4.734764761604041,
      "grad_norm": 3.8617515563964844,
      "learning_rate": 2.9471616026945477e-06,
      "loss": 0.5755,
      "step": 14995
    },
    {
      "epoch": 4.7350805178402275,
      "grad_norm": 3.670792579650879,
      "learning_rate": 2.943653076977054e-06,
      "loss": 0.5927,
      "step": 14996
    },
    {
      "epoch": 4.735396274076413,
      "grad_norm": 3.9862658977508545,
      "learning_rate": 2.940144551259561e-06,
      "loss": 0.6468,
      "step": 14997
    },
    {
      "epoch": 4.735712030312599,
      "grad_norm": 4.265116214752197,
      "learning_rate": 2.9366360255420673e-06,
      "loss": 0.5008,
      "step": 14998
    },
    {
      "epoch": 4.736027786548784,
      "grad_norm": 3.045544385910034,
      "learning_rate": 2.9331274998245737e-06,
      "loss": 0.5054,
      "step": 14999
    },
    {
      "epoch": 4.7363435427849705,
      "grad_norm": 4.555253028869629,
      "learning_rate": 2.9296189741070805e-06,
      "loss": 0.5735,
      "step": 15000
    },
    {
      "epoch": 4.7363435427849705,
      "eval_loss": 0.8360093832015991,
      "eval_runtime": 4.6125,
      "eval_samples_per_second": 539.841,
      "eval_steps_per_second": 4.553,
      "step": 15000
    },
    {
      "epoch": 4.736659299021156,
      "grad_norm": 3.709766387939453,
      "learning_rate": 2.926110448389587e-06,
      "loss": 0.5124,
      "step": 15001
    },
    {
      "epoch": 4.736975055257341,
      "grad_norm": 4.990177631378174,
      "learning_rate": 2.9226019226720933e-06,
      "loss": 0.655,
      "step": 15002
    },
    {
      "epoch": 4.737290811493527,
      "grad_norm": 3.691666603088379,
      "learning_rate": 2.9190933969546e-06,
      "loss": 0.5865,
      "step": 15003
    },
    {
      "epoch": 4.737606567729713,
      "grad_norm": 3.872912883758545,
      "learning_rate": 2.9155848712371065e-06,
      "loss": 0.5732,
      "step": 15004
    },
    {
      "epoch": 4.737922323965899,
      "grad_norm": 3.3608288764953613,
      "learning_rate": 2.912076345519613e-06,
      "loss": 0.5856,
      "step": 15005
    },
    {
      "epoch": 4.738238080202084,
      "grad_norm": 4.452308177947998,
      "learning_rate": 2.9085678198021193e-06,
      "loss": 0.528,
      "step": 15006
    },
    {
      "epoch": 4.738553836438269,
      "grad_norm": 3.688420534133911,
      "learning_rate": 2.9050592940846257e-06,
      "loss": 0.5765,
      "step": 15007
    },
    {
      "epoch": 4.7388695926744555,
      "grad_norm": 4.135876178741455,
      "learning_rate": 2.901550768367132e-06,
      "loss": 0.5663,
      "step": 15008
    },
    {
      "epoch": 4.739185348910641,
      "grad_norm": 4.570866107940674,
      "learning_rate": 2.898042242649639e-06,
      "loss": 0.58,
      "step": 15009
    },
    {
      "epoch": 4.739501105146827,
      "grad_norm": 4.476876735687256,
      "learning_rate": 2.8945337169321453e-06,
      "loss": 0.6418,
      "step": 15010
    },
    {
      "epoch": 4.739816861383012,
      "grad_norm": 4.2819318771362305,
      "learning_rate": 2.8910251912146516e-06,
      "loss": 0.5944,
      "step": 15011
    },
    {
      "epoch": 4.740132617619198,
      "grad_norm": 3.4321634769439697,
      "learning_rate": 2.8875166654971585e-06,
      "loss": 0.6319,
      "step": 15012
    },
    {
      "epoch": 4.740448373855384,
      "grad_norm": 5.219700336456299,
      "learning_rate": 2.884008139779665e-06,
      "loss": 0.6622,
      "step": 15013
    },
    {
      "epoch": 4.740764130091569,
      "grad_norm": 4.421844959259033,
      "learning_rate": 2.8804996140621712e-06,
      "loss": 0.663,
      "step": 15014
    },
    {
      "epoch": 4.741079886327755,
      "grad_norm": 3.896862030029297,
      "learning_rate": 2.8769910883446776e-06,
      "loss": 0.5621,
      "step": 15015
    },
    {
      "epoch": 4.7413956425639405,
      "grad_norm": 4.362390041351318,
      "learning_rate": 2.8734825626271844e-06,
      "loss": 0.6347,
      "step": 15016
    },
    {
      "epoch": 4.741711398800126,
      "grad_norm": 4.5855631828308105,
      "learning_rate": 2.869974036909691e-06,
      "loss": 0.5913,
      "step": 15017
    },
    {
      "epoch": 4.742027155036312,
      "grad_norm": 4.202632427215576,
      "learning_rate": 2.8664655111921972e-06,
      "loss": 0.5026,
      "step": 15018
    },
    {
      "epoch": 4.742342911272497,
      "grad_norm": 3.2630515098571777,
      "learning_rate": 2.8629569854747036e-06,
      "loss": 0.544,
      "step": 15019
    },
    {
      "epoch": 4.7426586675086835,
      "grad_norm": 3.8587520122528076,
      "learning_rate": 2.85944845975721e-06,
      "loss": 0.6611,
      "step": 15020
    },
    {
      "epoch": 4.742974423744869,
      "grad_norm": 4.053842067718506,
      "learning_rate": 2.8559399340397164e-06,
      "loss": 0.5552,
      "step": 15021
    },
    {
      "epoch": 4.743290179981055,
      "grad_norm": 4.442674160003662,
      "learning_rate": 2.852431408322223e-06,
      "loss": 0.5656,
      "step": 15022
    },
    {
      "epoch": 4.74360593621724,
      "grad_norm": 4.2655510902404785,
      "learning_rate": 2.8489228826047296e-06,
      "loss": 0.6,
      "step": 15023
    },
    {
      "epoch": 4.743921692453426,
      "grad_norm": 3.8527772426605225,
      "learning_rate": 2.845414356887236e-06,
      "loss": 0.5345,
      "step": 15024
    },
    {
      "epoch": 4.744237448689612,
      "grad_norm": 3.8992886543273926,
      "learning_rate": 2.841905831169743e-06,
      "loss": 0.5021,
      "step": 15025
    },
    {
      "epoch": 4.744553204925797,
      "grad_norm": 4.002113342285156,
      "learning_rate": 2.838397305452249e-06,
      "loss": 0.5988,
      "step": 15026
    },
    {
      "epoch": 4.744868961161983,
      "grad_norm": 4.121457099914551,
      "learning_rate": 2.8348887797347556e-06,
      "loss": 0.5484,
      "step": 15027
    },
    {
      "epoch": 4.7451847173981685,
      "grad_norm": 4.09970235824585,
      "learning_rate": 2.8313802540172624e-06,
      "loss": 0.5993,
      "step": 15028
    },
    {
      "epoch": 4.745500473634355,
      "grad_norm": 4.534584045410156,
      "learning_rate": 2.8278717282997683e-06,
      "loss": 0.5776,
      "step": 15029
    },
    {
      "epoch": 4.74581622987054,
      "grad_norm": 5.252788543701172,
      "learning_rate": 2.8243632025822747e-06,
      "loss": 0.4817,
      "step": 15030
    },
    {
      "epoch": 4.746131986106725,
      "grad_norm": 6.186293125152588,
      "learning_rate": 2.8208546768647815e-06,
      "loss": 0.6573,
      "step": 15031
    },
    {
      "epoch": 4.7464477423429114,
      "grad_norm": 3.829972267150879,
      "learning_rate": 2.817346151147288e-06,
      "loss": 0.5689,
      "step": 15032
    },
    {
      "epoch": 4.746763498579097,
      "grad_norm": 3.583416700363159,
      "learning_rate": 2.8138376254297943e-06,
      "loss": 0.6386,
      "step": 15033
    },
    {
      "epoch": 4.747079254815283,
      "grad_norm": 3.793868064880371,
      "learning_rate": 2.810329099712301e-06,
      "loss": 0.6427,
      "step": 15034
    },
    {
      "epoch": 4.747395011051468,
      "grad_norm": 3.9416122436523438,
      "learning_rate": 2.8068205739948075e-06,
      "loss": 0.5067,
      "step": 15035
    },
    {
      "epoch": 4.7477107672876535,
      "grad_norm": 3.630035400390625,
      "learning_rate": 2.803312048277314e-06,
      "loss": 0.5598,
      "step": 15036
    },
    {
      "epoch": 4.74802652352384,
      "grad_norm": 4.466196537017822,
      "learning_rate": 2.7998035225598207e-06,
      "loss": 0.5524,
      "step": 15037
    },
    {
      "epoch": 4.748342279760025,
      "grad_norm": 3.957958698272705,
      "learning_rate": 2.796294996842327e-06,
      "loss": 0.6292,
      "step": 15038
    },
    {
      "epoch": 4.748658035996211,
      "grad_norm": 4.2454304695129395,
      "learning_rate": 2.7927864711248335e-06,
      "loss": 0.5427,
      "step": 15039
    },
    {
      "epoch": 4.7489737922323965,
      "grad_norm": 4.7065043449401855,
      "learning_rate": 2.7892779454073403e-06,
      "loss": 0.643,
      "step": 15040
    },
    {
      "epoch": 4.749289548468582,
      "grad_norm": 4.104641437530518,
      "learning_rate": 2.7857694196898463e-06,
      "loss": 0.6048,
      "step": 15041
    },
    {
      "epoch": 4.749605304704768,
      "grad_norm": 3.913954257965088,
      "learning_rate": 2.7822608939723527e-06,
      "loss": 0.5981,
      "step": 15042
    },
    {
      "epoch": 4.749921060940953,
      "grad_norm": 3.5562663078308105,
      "learning_rate": 2.7787523682548595e-06,
      "loss": 0.5008,
      "step": 15043
    },
    {
      "epoch": 4.750236817177139,
      "grad_norm": 4.678506374359131,
      "learning_rate": 2.775243842537366e-06,
      "loss": 0.5516,
      "step": 15044
    },
    {
      "epoch": 4.750552573413325,
      "grad_norm": 4.692375659942627,
      "learning_rate": 2.7717353168198723e-06,
      "loss": 0.5816,
      "step": 15045
    },
    {
      "epoch": 4.750868329649511,
      "grad_norm": 4.134764671325684,
      "learning_rate": 2.768226791102379e-06,
      "loss": 0.6393,
      "step": 15046
    },
    {
      "epoch": 4.751184085885696,
      "grad_norm": 3.9791972637176514,
      "learning_rate": 2.7647182653848855e-06,
      "loss": 0.5659,
      "step": 15047
    },
    {
      "epoch": 4.751499842121882,
      "grad_norm": 4.109180450439453,
      "learning_rate": 2.761209739667392e-06,
      "loss": 0.5627,
      "step": 15048
    },
    {
      "epoch": 4.751815598358068,
      "grad_norm": 4.391905307769775,
      "learning_rate": 2.7577012139498983e-06,
      "loss": 0.5429,
      "step": 15049
    },
    {
      "epoch": 4.752131354594253,
      "grad_norm": 3.895209312438965,
      "learning_rate": 2.754192688232405e-06,
      "loss": 0.5493,
      "step": 15050
    },
    {
      "epoch": 4.752447110830439,
      "grad_norm": 3.4243297576904297,
      "learning_rate": 2.7506841625149115e-06,
      "loss": 0.5835,
      "step": 15051
    },
    {
      "epoch": 4.752762867066624,
      "grad_norm": 3.868600845336914,
      "learning_rate": 2.747175636797418e-06,
      "loss": 0.5766,
      "step": 15052
    },
    {
      "epoch": 4.753078623302811,
      "grad_norm": 4.570718765258789,
      "learning_rate": 2.7436671110799242e-06,
      "loss": 0.5384,
      "step": 15053
    },
    {
      "epoch": 4.753394379538996,
      "grad_norm": 5.131298065185547,
      "learning_rate": 2.7401585853624306e-06,
      "loss": 0.557,
      "step": 15054
    },
    {
      "epoch": 4.753710135775181,
      "grad_norm": 4.311748504638672,
      "learning_rate": 2.7366500596449374e-06,
      "loss": 0.6628,
      "step": 15055
    },
    {
      "epoch": 4.754025892011367,
      "grad_norm": 4.425333023071289,
      "learning_rate": 2.733141533927444e-06,
      "loss": 0.6155,
      "step": 15056
    },
    {
      "epoch": 4.754341648247553,
      "grad_norm": 3.8290443420410156,
      "learning_rate": 2.7296330082099502e-06,
      "loss": 0.6018,
      "step": 15057
    },
    {
      "epoch": 4.754657404483739,
      "grad_norm": 4.502718925476074,
      "learning_rate": 2.7261244824924566e-06,
      "loss": 0.5954,
      "step": 15058
    },
    {
      "epoch": 4.754973160719924,
      "grad_norm": 3.651845693588257,
      "learning_rate": 2.7226159567749634e-06,
      "loss": 0.4615,
      "step": 15059
    },
    {
      "epoch": 4.7552889169561094,
      "grad_norm": 3.188307046890259,
      "learning_rate": 2.71910743105747e-06,
      "loss": 0.4937,
      "step": 15060
    },
    {
      "epoch": 4.755604673192296,
      "grad_norm": 4.409854412078857,
      "learning_rate": 2.715598905339976e-06,
      "loss": 0.6689,
      "step": 15061
    },
    {
      "epoch": 4.755920429428481,
      "grad_norm": 4.8948798179626465,
      "learning_rate": 2.712090379622483e-06,
      "loss": 0.555,
      "step": 15062
    },
    {
      "epoch": 4.756236185664667,
      "grad_norm": 4.4072160720825195,
      "learning_rate": 2.7085818539049894e-06,
      "loss": 0.681,
      "step": 15063
    },
    {
      "epoch": 4.756551941900852,
      "grad_norm": 4.352787494659424,
      "learning_rate": 2.705073328187496e-06,
      "loss": 0.587,
      "step": 15064
    },
    {
      "epoch": 4.756867698137039,
      "grad_norm": 4.117486000061035,
      "learning_rate": 2.701564802470002e-06,
      "loss": 0.6882,
      "step": 15065
    },
    {
      "epoch": 4.757183454373224,
      "grad_norm": 4.3746538162231445,
      "learning_rate": 2.6980562767525086e-06,
      "loss": 0.6033,
      "step": 15066
    },
    {
      "epoch": 4.757499210609409,
      "grad_norm": 3.892209053039551,
      "learning_rate": 2.694547751035015e-06,
      "loss": 0.546,
      "step": 15067
    },
    {
      "epoch": 4.757814966845595,
      "grad_norm": 3.7621545791625977,
      "learning_rate": 2.6910392253175218e-06,
      "loss": 0.5758,
      "step": 15068
    },
    {
      "epoch": 4.758130723081781,
      "grad_norm": 4.69045877456665,
      "learning_rate": 2.687530699600028e-06,
      "loss": 0.5776,
      "step": 15069
    },
    {
      "epoch": 4.758446479317967,
      "grad_norm": 3.290311574935913,
      "learning_rate": 2.6840221738825345e-06,
      "loss": 0.6248,
      "step": 15070
    },
    {
      "epoch": 4.758762235554152,
      "grad_norm": 3.637934684753418,
      "learning_rate": 2.6805136481650414e-06,
      "loss": 0.652,
      "step": 15071
    },
    {
      "epoch": 4.759077991790338,
      "grad_norm": 5.117332458496094,
      "learning_rate": 2.6770051224475478e-06,
      "loss": 0.5794,
      "step": 15072
    },
    {
      "epoch": 4.759393748026524,
      "grad_norm": 3.7592873573303223,
      "learning_rate": 2.673496596730054e-06,
      "loss": 0.5786,
      "step": 15073
    },
    {
      "epoch": 4.759709504262709,
      "grad_norm": 3.7058751583099365,
      "learning_rate": 2.669988071012561e-06,
      "loss": 0.6177,
      "step": 15074
    },
    {
      "epoch": 4.760025260498895,
      "grad_norm": 4.635594367980957,
      "learning_rate": 2.6664795452950673e-06,
      "loss": 0.5675,
      "step": 15075
    },
    {
      "epoch": 4.76034101673508,
      "grad_norm": 4.150288105010986,
      "learning_rate": 2.6629710195775733e-06,
      "loss": 0.6525,
      "step": 15076
    },
    {
      "epoch": 4.7606567729712665,
      "grad_norm": 4.254905700683594,
      "learning_rate": 2.65946249386008e-06,
      "loss": 0.602,
      "step": 15077
    },
    {
      "epoch": 4.760972529207452,
      "grad_norm": 3.589681625366211,
      "learning_rate": 2.6559539681425865e-06,
      "loss": 0.5891,
      "step": 15078
    },
    {
      "epoch": 4.761288285443637,
      "grad_norm": 4.366471290588379,
      "learning_rate": 2.652445442425093e-06,
      "loss": 0.6738,
      "step": 15079
    },
    {
      "epoch": 4.761604041679823,
      "grad_norm": 3.7954821586608887,
      "learning_rate": 2.6489369167075997e-06,
      "loss": 0.6522,
      "step": 15080
    },
    {
      "epoch": 4.761919797916009,
      "grad_norm": 5.491544723510742,
      "learning_rate": 2.645428390990106e-06,
      "loss": 0.7009,
      "step": 15081
    },
    {
      "epoch": 4.762235554152195,
      "grad_norm": 3.51063871383667,
      "learning_rate": 2.6419198652726125e-06,
      "loss": 0.5107,
      "step": 15082
    },
    {
      "epoch": 4.76255131038838,
      "grad_norm": 4.0281853675842285,
      "learning_rate": 2.6384113395551193e-06,
      "loss": 0.6019,
      "step": 15083
    },
    {
      "epoch": 4.762867066624565,
      "grad_norm": 3.7414164543151855,
      "learning_rate": 2.6349028138376257e-06,
      "loss": 0.6921,
      "step": 15084
    },
    {
      "epoch": 4.763182822860752,
      "grad_norm": 4.1304097175598145,
      "learning_rate": 2.631394288120132e-06,
      "loss": 0.6308,
      "step": 15085
    },
    {
      "epoch": 4.763498579096937,
      "grad_norm": 3.854736328125,
      "learning_rate": 2.6278857624026385e-06,
      "loss": 0.5407,
      "step": 15086
    },
    {
      "epoch": 4.763814335333123,
      "grad_norm": 4.09119987487793,
      "learning_rate": 2.6243772366851453e-06,
      "loss": 0.5892,
      "step": 15087
    },
    {
      "epoch": 4.764130091569308,
      "grad_norm": 3.694086790084839,
      "learning_rate": 2.6208687109676513e-06,
      "loss": 0.6315,
      "step": 15088
    },
    {
      "epoch": 4.7644458478054945,
      "grad_norm": 3.92783260345459,
      "learning_rate": 2.617360185250158e-06,
      "loss": 0.6023,
      "step": 15089
    },
    {
      "epoch": 4.76476160404168,
      "grad_norm": 4.4226837158203125,
      "learning_rate": 2.6138516595326645e-06,
      "loss": 0.6624,
      "step": 15090
    },
    {
      "epoch": 4.765077360277865,
      "grad_norm": 3.655534029006958,
      "learning_rate": 2.610343133815171e-06,
      "loss": 0.6838,
      "step": 15091
    },
    {
      "epoch": 4.765393116514051,
      "grad_norm": 3.8373863697052,
      "learning_rate": 2.6068346080976772e-06,
      "loss": 0.6822,
      "step": 15092
    },
    {
      "epoch": 4.765708872750237,
      "grad_norm": 3.8387198448181152,
      "learning_rate": 2.603326082380184e-06,
      "loss": 0.5091,
      "step": 15093
    },
    {
      "epoch": 4.766024628986423,
      "grad_norm": 4.104846954345703,
      "learning_rate": 2.5998175566626904e-06,
      "loss": 0.4977,
      "step": 15094
    },
    {
      "epoch": 4.766340385222608,
      "grad_norm": 4.58223295211792,
      "learning_rate": 2.596309030945197e-06,
      "loss": 0.6895,
      "step": 15095
    },
    {
      "epoch": 4.766656141458794,
      "grad_norm": 3.543214797973633,
      "learning_rate": 2.5928005052277036e-06,
      "loss": 0.6076,
      "step": 15096
    },
    {
      "epoch": 4.7669718976949795,
      "grad_norm": 4.835012435913086,
      "learning_rate": 2.58929197951021e-06,
      "loss": 0.6068,
      "step": 15097
    },
    {
      "epoch": 4.767287653931165,
      "grad_norm": 3.798400402069092,
      "learning_rate": 2.5857834537927164e-06,
      "loss": 0.596,
      "step": 15098
    },
    {
      "epoch": 4.767603410167351,
      "grad_norm": 4.46333122253418,
      "learning_rate": 2.5822749280752232e-06,
      "loss": 0.5701,
      "step": 15099
    },
    {
      "epoch": 4.767919166403536,
      "grad_norm": 3.945021629333496,
      "learning_rate": 2.578766402357729e-06,
      "loss": 0.6326,
      "step": 15100
    },
    {
      "epoch": 4.7682349226397225,
      "grad_norm": 5.345331192016602,
      "learning_rate": 2.5752578766402356e-06,
      "loss": 0.6027,
      "step": 15101
    },
    {
      "epoch": 4.768550678875908,
      "grad_norm": 4.321534633636475,
      "learning_rate": 2.5717493509227424e-06,
      "loss": 0.6863,
      "step": 15102
    },
    {
      "epoch": 4.768866435112093,
      "grad_norm": 4.005096435546875,
      "learning_rate": 2.568240825205249e-06,
      "loss": 0.6534,
      "step": 15103
    },
    {
      "epoch": 4.769182191348279,
      "grad_norm": 4.328572750091553,
      "learning_rate": 2.564732299487755e-06,
      "loss": 0.6971,
      "step": 15104
    },
    {
      "epoch": 4.7694979475844645,
      "grad_norm": 4.277444362640381,
      "learning_rate": 2.561223773770262e-06,
      "loss": 0.6036,
      "step": 15105
    },
    {
      "epoch": 4.769813703820651,
      "grad_norm": 3.6180217266082764,
      "learning_rate": 2.5577152480527684e-06,
      "loss": 0.6046,
      "step": 15106
    },
    {
      "epoch": 4.770129460056836,
      "grad_norm": 3.6810033321380615,
      "learning_rate": 2.5542067223352748e-06,
      "loss": 0.6004,
      "step": 15107
    },
    {
      "epoch": 4.770445216293021,
      "grad_norm": 3.9152472019195557,
      "learning_rate": 2.5506981966177816e-06,
      "loss": 0.4925,
      "step": 15108
    },
    {
      "epoch": 4.7707609725292075,
      "grad_norm": 4.5655388832092285,
      "learning_rate": 2.547189670900288e-06,
      "loss": 0.5478,
      "step": 15109
    },
    {
      "epoch": 4.771076728765393,
      "grad_norm": 3.5877199172973633,
      "learning_rate": 2.5436811451827944e-06,
      "loss": 0.5281,
      "step": 15110
    },
    {
      "epoch": 4.771392485001579,
      "grad_norm": 4.2165093421936035,
      "learning_rate": 2.540172619465301e-06,
      "loss": 0.5245,
      "step": 15111
    },
    {
      "epoch": 4.771708241237764,
      "grad_norm": 3.828657865524292,
      "learning_rate": 2.536664093747807e-06,
      "loss": 0.5206,
      "step": 15112
    },
    {
      "epoch": 4.7720239974739505,
      "grad_norm": 4.347731590270996,
      "learning_rate": 2.5331555680303135e-06,
      "loss": 0.6312,
      "step": 15113
    },
    {
      "epoch": 4.772339753710136,
      "grad_norm": 3.2996878623962402,
      "learning_rate": 2.5296470423128203e-06,
      "loss": 0.5635,
      "step": 15114
    },
    {
      "epoch": 4.772655509946322,
      "grad_norm": 5.323147296905518,
      "learning_rate": 2.5261385165953267e-06,
      "loss": 0.6815,
      "step": 15115
    },
    {
      "epoch": 4.772971266182507,
      "grad_norm": 3.9451639652252197,
      "learning_rate": 2.522629990877833e-06,
      "loss": 0.5411,
      "step": 15116
    },
    {
      "epoch": 4.7732870224186925,
      "grad_norm": 4.708959102630615,
      "learning_rate": 2.51912146516034e-06,
      "loss": 0.5841,
      "step": 15117
    },
    {
      "epoch": 4.773602778654879,
      "grad_norm": 5.061096668243408,
      "learning_rate": 2.5156129394428463e-06,
      "loss": 0.6465,
      "step": 15118
    },
    {
      "epoch": 4.773918534891064,
      "grad_norm": 4.8090667724609375,
      "learning_rate": 2.5121044137253527e-06,
      "loss": 0.5729,
      "step": 15119
    },
    {
      "epoch": 4.77423429112725,
      "grad_norm": 4.10702657699585,
      "learning_rate": 2.508595888007859e-06,
      "loss": 0.59,
      "step": 15120
    },
    {
      "epoch": 4.7745500473634355,
      "grad_norm": 3.351728677749634,
      "learning_rate": 2.505087362290366e-06,
      "loss": 0.5753,
      "step": 15121
    },
    {
      "epoch": 4.774865803599621,
      "grad_norm": 3.600895404815674,
      "learning_rate": 2.5015788365728723e-06,
      "loss": 0.5251,
      "step": 15122
    },
    {
      "epoch": 4.775181559835807,
      "grad_norm": 3.9884743690490723,
      "learning_rate": 2.4980703108553787e-06,
      "loss": 0.5945,
      "step": 15123
    },
    {
      "epoch": 4.775497316071992,
      "grad_norm": 4.3763017654418945,
      "learning_rate": 2.494561785137885e-06,
      "loss": 0.5564,
      "step": 15124
    },
    {
      "epoch": 4.775813072308178,
      "grad_norm": 3.833498954772949,
      "learning_rate": 2.4910532594203915e-06,
      "loss": 0.5995,
      "step": 15125
    },
    {
      "epoch": 4.776128828544364,
      "grad_norm": 4.166014671325684,
      "learning_rate": 2.487544733702898e-06,
      "loss": 0.5891,
      "step": 15126
    },
    {
      "epoch": 4.776444584780549,
      "grad_norm": 4.616016864776611,
      "learning_rate": 2.4840362079854047e-06,
      "loss": 0.6458,
      "step": 15127
    },
    {
      "epoch": 4.776760341016735,
      "grad_norm": 3.5780880451202393,
      "learning_rate": 2.480527682267911e-06,
      "loss": 0.54,
      "step": 15128
    },
    {
      "epoch": 4.7770760972529205,
      "grad_norm": 4.7628350257873535,
      "learning_rate": 2.4770191565504175e-06,
      "loss": 0.5734,
      "step": 15129
    },
    {
      "epoch": 4.777391853489107,
      "grad_norm": 4.641135215759277,
      "learning_rate": 2.4735106308329243e-06,
      "loss": 0.5748,
      "step": 15130
    },
    {
      "epoch": 4.777707609725292,
      "grad_norm": 2.919358015060425,
      "learning_rate": 2.4700021051154307e-06,
      "loss": 0.5096,
      "step": 15131
    },
    {
      "epoch": 4.778023365961477,
      "grad_norm": 4.709303855895996,
      "learning_rate": 2.466493579397937e-06,
      "loss": 0.5332,
      "step": 15132
    },
    {
      "epoch": 4.778339122197663,
      "grad_norm": 5.20325231552124,
      "learning_rate": 2.462985053680444e-06,
      "loss": 0.7272,
      "step": 15133
    },
    {
      "epoch": 4.778654878433849,
      "grad_norm": 4.292670726776123,
      "learning_rate": 2.4594765279629503e-06,
      "loss": 0.596,
      "step": 15134
    },
    {
      "epoch": 4.778970634670035,
      "grad_norm": 4.599093437194824,
      "learning_rate": 2.4559680022454562e-06,
      "loss": 0.6622,
      "step": 15135
    },
    {
      "epoch": 4.77928639090622,
      "grad_norm": 3.7840025424957275,
      "learning_rate": 2.452459476527963e-06,
      "loss": 0.4964,
      "step": 15136
    },
    {
      "epoch": 4.779602147142406,
      "grad_norm": 4.8684611320495605,
      "learning_rate": 2.4489509508104694e-06,
      "loss": 0.63,
      "step": 15137
    },
    {
      "epoch": 4.779917903378592,
      "grad_norm": 4.628605365753174,
      "learning_rate": 2.445442425092976e-06,
      "loss": 0.6234,
      "step": 15138
    },
    {
      "epoch": 4.780233659614778,
      "grad_norm": 3.25986385345459,
      "learning_rate": 2.4419338993754826e-06,
      "loss": 0.4935,
      "step": 15139
    },
    {
      "epoch": 4.780549415850963,
      "grad_norm": 4.197206020355225,
      "learning_rate": 2.438425373657989e-06,
      "loss": 0.5063,
      "step": 15140
    },
    {
      "epoch": 4.7808651720871485,
      "grad_norm": 4.990467548370361,
      "learning_rate": 2.4349168479404954e-06,
      "loss": 0.6824,
      "step": 15141
    },
    {
      "epoch": 4.781180928323335,
      "grad_norm": 4.800434112548828,
      "learning_rate": 2.4314083222230022e-06,
      "loss": 0.5771,
      "step": 15142
    },
    {
      "epoch": 4.78149668455952,
      "grad_norm": 3.872422218322754,
      "learning_rate": 2.4278997965055086e-06,
      "loss": 0.487,
      "step": 15143
    },
    {
      "epoch": 4.781812440795706,
      "grad_norm": 4.399198055267334,
      "learning_rate": 2.424391270788015e-06,
      "loss": 0.68,
      "step": 15144
    },
    {
      "epoch": 4.782128197031891,
      "grad_norm": 5.60447359085083,
      "learning_rate": 2.420882745070522e-06,
      "loss": 0.6169,
      "step": 15145
    },
    {
      "epoch": 4.782443953268077,
      "grad_norm": 3.773834705352783,
      "learning_rate": 2.417374219353028e-06,
      "loss": 0.6117,
      "step": 15146
    },
    {
      "epoch": 4.782759709504263,
      "grad_norm": 3.678459882736206,
      "learning_rate": 2.413865693635534e-06,
      "loss": 0.5795,
      "step": 15147
    },
    {
      "epoch": 4.783075465740448,
      "grad_norm": 3.704536199569702,
      "learning_rate": 2.410357167918041e-06,
      "loss": 0.6265,
      "step": 15148
    },
    {
      "epoch": 4.783391221976634,
      "grad_norm": 4.609915733337402,
      "learning_rate": 2.4068486422005474e-06,
      "loss": 0.588,
      "step": 15149
    },
    {
      "epoch": 4.78370697821282,
      "grad_norm": 3.4239587783813477,
      "learning_rate": 2.4033401164830538e-06,
      "loss": 0.5175,
      "step": 15150
    },
    {
      "epoch": 4.784022734449005,
      "grad_norm": 4.736837387084961,
      "learning_rate": 2.3998315907655606e-06,
      "loss": 0.6347,
      "step": 15151
    },
    {
      "epoch": 4.784338490685191,
      "grad_norm": 3.5259041786193848,
      "learning_rate": 2.396323065048067e-06,
      "loss": 0.5544,
      "step": 15152
    },
    {
      "epoch": 4.784654246921376,
      "grad_norm": 4.04373025894165,
      "learning_rate": 2.3928145393305733e-06,
      "loss": 0.5381,
      "step": 15153
    },
    {
      "epoch": 4.784970003157563,
      "grad_norm": 3.5573415756225586,
      "learning_rate": 2.3893060136130797e-06,
      "loss": 0.5173,
      "step": 15154
    },
    {
      "epoch": 4.785285759393748,
      "grad_norm": 4.106550693511963,
      "learning_rate": 2.3857974878955865e-06,
      "loss": 0.5479,
      "step": 15155
    },
    {
      "epoch": 4.785601515629934,
      "grad_norm": 4.29001522064209,
      "learning_rate": 2.382288962178093e-06,
      "loss": 0.616,
      "step": 15156
    },
    {
      "epoch": 4.785917271866119,
      "grad_norm": 3.8131797313690186,
      "learning_rate": 2.3787804364605993e-06,
      "loss": 0.4983,
      "step": 15157
    },
    {
      "epoch": 4.786233028102305,
      "grad_norm": 4.6436309814453125,
      "learning_rate": 2.375271910743106e-06,
      "loss": 0.6126,
      "step": 15158
    },
    {
      "epoch": 4.786548784338491,
      "grad_norm": 4.055535316467285,
      "learning_rate": 2.371763385025612e-06,
      "loss": 0.5428,
      "step": 15159
    },
    {
      "epoch": 4.786864540574676,
      "grad_norm": 4.61488676071167,
      "learning_rate": 2.368254859308119e-06,
      "loss": 0.5821,
      "step": 15160
    },
    {
      "epoch": 4.787180296810862,
      "grad_norm": 4.63096809387207,
      "learning_rate": 2.3647463335906253e-06,
      "loss": 0.669,
      "step": 15161
    },
    {
      "epoch": 4.787496053047048,
      "grad_norm": 4.484964370727539,
      "learning_rate": 2.3612378078731317e-06,
      "loss": 0.5927,
      "step": 15162
    },
    {
      "epoch": 4.787811809283234,
      "grad_norm": 3.3727407455444336,
      "learning_rate": 2.357729282155638e-06,
      "loss": 0.5925,
      "step": 15163
    },
    {
      "epoch": 4.788127565519419,
      "grad_norm": 4.203690528869629,
      "learning_rate": 2.354220756438145e-06,
      "loss": 0.6223,
      "step": 15164
    },
    {
      "epoch": 4.788443321755604,
      "grad_norm": 3.7618045806884766,
      "learning_rate": 2.3507122307206513e-06,
      "loss": 0.5764,
      "step": 15165
    },
    {
      "epoch": 4.788759077991791,
      "grad_norm": 4.883883476257324,
      "learning_rate": 2.3472037050031577e-06,
      "loss": 0.5714,
      "step": 15166
    },
    {
      "epoch": 4.789074834227976,
      "grad_norm": 4.039731025695801,
      "learning_rate": 2.3436951792856645e-06,
      "loss": 0.664,
      "step": 15167
    },
    {
      "epoch": 4.789390590464162,
      "grad_norm": 3.5244250297546387,
      "learning_rate": 2.340186653568171e-06,
      "loss": 0.595,
      "step": 15168
    },
    {
      "epoch": 4.789706346700347,
      "grad_norm": 5.686408042907715,
      "learning_rate": 2.3366781278506773e-06,
      "loss": 0.6692,
      "step": 15169
    },
    {
      "epoch": 4.790022102936533,
      "grad_norm": 5.059384822845459,
      "learning_rate": 2.333169602133184e-06,
      "loss": 0.5929,
      "step": 15170
    },
    {
      "epoch": 4.790337859172719,
      "grad_norm": 3.9356298446655273,
      "learning_rate": 2.32966107641569e-06,
      "loss": 0.5445,
      "step": 15171
    },
    {
      "epoch": 4.790653615408904,
      "grad_norm": 5.231904029846191,
      "learning_rate": 2.3261525506981964e-06,
      "loss": 0.6442,
      "step": 15172
    },
    {
      "epoch": 4.79096937164509,
      "grad_norm": 3.732232093811035,
      "learning_rate": 2.3226440249807033e-06,
      "loss": 0.6235,
      "step": 15173
    },
    {
      "epoch": 4.791285127881276,
      "grad_norm": 5.055019855499268,
      "learning_rate": 2.3191354992632096e-06,
      "loss": 0.5586,
      "step": 15174
    },
    {
      "epoch": 4.791600884117461,
      "grad_norm": 4.203725337982178,
      "learning_rate": 2.315626973545716e-06,
      "loss": 0.5982,
      "step": 15175
    },
    {
      "epoch": 4.791916640353647,
      "grad_norm": 3.4111979007720947,
      "learning_rate": 2.312118447828223e-06,
      "loss": 0.5087,
      "step": 15176
    },
    {
      "epoch": 4.792232396589832,
      "grad_norm": 3.6163716316223145,
      "learning_rate": 2.3086099221107292e-06,
      "loss": 0.5816,
      "step": 15177
    },
    {
      "epoch": 4.7925481528260185,
      "grad_norm": 5.266513347625732,
      "learning_rate": 2.3051013963932356e-06,
      "loss": 0.5888,
      "step": 15178
    },
    {
      "epoch": 4.792863909062204,
      "grad_norm": 4.12344217300415,
      "learning_rate": 2.3015928706757424e-06,
      "loss": 0.5433,
      "step": 15179
    },
    {
      "epoch": 4.79317966529839,
      "grad_norm": 3.21382999420166,
      "learning_rate": 2.298084344958249e-06,
      "loss": 0.489,
      "step": 15180
    },
    {
      "epoch": 4.793495421534575,
      "grad_norm": 4.64617395401001,
      "learning_rate": 2.2945758192407552e-06,
      "loss": 0.6384,
      "step": 15181
    },
    {
      "epoch": 4.793811177770761,
      "grad_norm": 4.514238357543945,
      "learning_rate": 2.2910672935232616e-06,
      "loss": 0.5549,
      "step": 15182
    },
    {
      "epoch": 4.794126934006947,
      "grad_norm": 3.883538246154785,
      "learning_rate": 2.287558767805768e-06,
      "loss": 0.5291,
      "step": 15183
    },
    {
      "epoch": 4.794442690243132,
      "grad_norm": 3.8716607093811035,
      "learning_rate": 2.2840502420882744e-06,
      "loss": 0.5852,
      "step": 15184
    },
    {
      "epoch": 4.794758446479318,
      "grad_norm": 4.8446269035339355,
      "learning_rate": 2.280541716370781e-06,
      "loss": 0.4892,
      "step": 15185
    },
    {
      "epoch": 4.7950742027155036,
      "grad_norm": 4.094356536865234,
      "learning_rate": 2.2770331906532876e-06,
      "loss": 0.6381,
      "step": 15186
    },
    {
      "epoch": 4.79538995895169,
      "grad_norm": 4.351500034332275,
      "learning_rate": 2.273524664935794e-06,
      "loss": 0.5738,
      "step": 15187
    },
    {
      "epoch": 4.795705715187875,
      "grad_norm": 4.549954414367676,
      "learning_rate": 2.270016139218301e-06,
      "loss": 0.59,
      "step": 15188
    },
    {
      "epoch": 4.79602147142406,
      "grad_norm": 4.054149627685547,
      "learning_rate": 2.266507613500807e-06,
      "loss": 0.5791,
      "step": 15189
    },
    {
      "epoch": 4.7963372276602465,
      "grad_norm": 4.486202716827393,
      "learning_rate": 2.2629990877833136e-06,
      "loss": 0.5748,
      "step": 15190
    },
    {
      "epoch": 4.796652983896432,
      "grad_norm": 4.251358985900879,
      "learning_rate": 2.25949056206582e-06,
      "loss": 0.6652,
      "step": 15191
    },
    {
      "epoch": 4.796968740132618,
      "grad_norm": 4.325706481933594,
      "learning_rate": 2.2559820363483268e-06,
      "loss": 0.6131,
      "step": 15192
    },
    {
      "epoch": 4.797284496368803,
      "grad_norm": 4.048969268798828,
      "learning_rate": 2.252473510630833e-06,
      "loss": 0.5502,
      "step": 15193
    },
    {
      "epoch": 4.797600252604989,
      "grad_norm": 3.6498923301696777,
      "learning_rate": 2.2489649849133395e-06,
      "loss": 0.485,
      "step": 15194
    },
    {
      "epoch": 4.797916008841175,
      "grad_norm": 3.874384880065918,
      "learning_rate": 2.245456459195846e-06,
      "loss": 0.5206,
      "step": 15195
    },
    {
      "epoch": 4.79823176507736,
      "grad_norm": 3.8351986408233643,
      "learning_rate": 2.2419479334783523e-06,
      "loss": 0.6374,
      "step": 15196
    },
    {
      "epoch": 4.798547521313546,
      "grad_norm": 4.487659454345703,
      "learning_rate": 2.2384394077608587e-06,
      "loss": 0.6023,
      "step": 15197
    },
    {
      "epoch": 4.7988632775497315,
      "grad_norm": 3.599745512008667,
      "learning_rate": 2.2349308820433655e-06,
      "loss": 0.6074,
      "step": 15198
    },
    {
      "epoch": 4.799179033785917,
      "grad_norm": 4.253176212310791,
      "learning_rate": 2.231422356325872e-06,
      "loss": 0.4691,
      "step": 15199
    },
    {
      "epoch": 4.799494790022103,
      "grad_norm": 5.326785087585449,
      "learning_rate": 2.2279138306083783e-06,
      "loss": 0.6403,
      "step": 15200
    },
    {
      "epoch": 4.799494790022103,
      "eval_loss": 0.8384033441543579,
      "eval_runtime": 4.6024,
      "eval_samples_per_second": 541.026,
      "eval_steps_per_second": 4.563,
      "step": 15200
    },
    {
      "epoch": 4.799810546258288,
      "grad_norm": 4.047872543334961,
      "learning_rate": 2.224405304890885e-06,
      "loss": 0.6338,
      "step": 15201
    },
    {
      "epoch": 4.8001263024944745,
      "grad_norm": 4.252284049987793,
      "learning_rate": 2.2208967791733915e-06,
      "loss": 0.5272,
      "step": 15202
    },
    {
      "epoch": 4.80044205873066,
      "grad_norm": 4.549979209899902,
      "learning_rate": 2.217388253455898e-06,
      "loss": 0.548,
      "step": 15203
    },
    {
      "epoch": 4.800757814966846,
      "grad_norm": 3.8344125747680664,
      "learning_rate": 2.2138797277384047e-06,
      "loss": 0.5771,
      "step": 15204
    },
    {
      "epoch": 4.801073571203031,
      "grad_norm": 4.279801845550537,
      "learning_rate": 2.210371202020911e-06,
      "loss": 0.5865,
      "step": 15205
    },
    {
      "epoch": 4.8013893274392165,
      "grad_norm": 3.338080883026123,
      "learning_rate": 2.206862676303417e-06,
      "loss": 0.5286,
      "step": 15206
    },
    {
      "epoch": 4.801705083675403,
      "grad_norm": 4.9928998947143555,
      "learning_rate": 2.203354150585924e-06,
      "loss": 0.5437,
      "step": 15207
    },
    {
      "epoch": 4.802020839911588,
      "grad_norm": 4.068578243255615,
      "learning_rate": 2.1998456248684303e-06,
      "loss": 0.5414,
      "step": 15208
    },
    {
      "epoch": 4.802336596147774,
      "grad_norm": 4.034677028656006,
      "learning_rate": 2.1963370991509367e-06,
      "loss": 0.5567,
      "step": 15209
    },
    {
      "epoch": 4.8026523523839595,
      "grad_norm": 4.540472030639648,
      "learning_rate": 2.1928285734334435e-06,
      "loss": 0.6451,
      "step": 15210
    },
    {
      "epoch": 4.802968108620146,
      "grad_norm": 4.4135050773620605,
      "learning_rate": 2.18932004771595e-06,
      "loss": 0.5964,
      "step": 15211
    },
    {
      "epoch": 4.803283864856331,
      "grad_norm": 4.061688423156738,
      "learning_rate": 2.1858115219984563e-06,
      "loss": 0.5675,
      "step": 15212
    },
    {
      "epoch": 4.803599621092516,
      "grad_norm": 3.796557903289795,
      "learning_rate": 2.182302996280963e-06,
      "loss": 0.5622,
      "step": 15213
    },
    {
      "epoch": 4.803915377328702,
      "grad_norm": 4.010444164276123,
      "learning_rate": 2.1787944705634695e-06,
      "loss": 0.6562,
      "step": 15214
    },
    {
      "epoch": 4.804231133564888,
      "grad_norm": 3.638470411300659,
      "learning_rate": 2.175285944845976e-06,
      "loss": 0.5285,
      "step": 15215
    },
    {
      "epoch": 4.804546889801074,
      "grad_norm": 4.262436866760254,
      "learning_rate": 2.1717774191284827e-06,
      "loss": 0.6019,
      "step": 15216
    },
    {
      "epoch": 4.804862646037259,
      "grad_norm": 4.2963643074035645,
      "learning_rate": 2.168268893410989e-06,
      "loss": 0.5928,
      "step": 15217
    },
    {
      "epoch": 4.8051784022734445,
      "grad_norm": 3.786919593811035,
      "learning_rate": 2.164760367693495e-06,
      "loss": 0.5805,
      "step": 15218
    },
    {
      "epoch": 4.805494158509631,
      "grad_norm": 3.4883625507354736,
      "learning_rate": 2.161251841976002e-06,
      "loss": 0.4524,
      "step": 15219
    },
    {
      "epoch": 4.805809914745816,
      "grad_norm": 4.214141845703125,
      "learning_rate": 2.1577433162585082e-06,
      "loss": 0.6009,
      "step": 15220
    },
    {
      "epoch": 4.806125670982002,
      "grad_norm": 5.037858486175537,
      "learning_rate": 2.1542347905410146e-06,
      "loss": 0.553,
      "step": 15221
    },
    {
      "epoch": 4.8064414272181875,
      "grad_norm": 3.4762327671051025,
      "learning_rate": 2.1507262648235214e-06,
      "loss": 0.5543,
      "step": 15222
    },
    {
      "epoch": 4.806757183454373,
      "grad_norm": 3.8213419914245605,
      "learning_rate": 2.147217739106028e-06,
      "loss": 0.5151,
      "step": 15223
    },
    {
      "epoch": 4.807072939690559,
      "grad_norm": 3.9241623878479004,
      "learning_rate": 2.143709213388534e-06,
      "loss": 0.6421,
      "step": 15224
    },
    {
      "epoch": 4.807388695926744,
      "grad_norm": 4.9333977699279785,
      "learning_rate": 2.1402006876710406e-06,
      "loss": 0.4849,
      "step": 15225
    },
    {
      "epoch": 4.80770445216293,
      "grad_norm": 4.127133369445801,
      "learning_rate": 2.1366921619535474e-06,
      "loss": 0.5572,
      "step": 15226
    },
    {
      "epoch": 4.808020208399116,
      "grad_norm": 3.9848978519439697,
      "learning_rate": 2.133183636236054e-06,
      "loss": 0.5599,
      "step": 15227
    },
    {
      "epoch": 4.808335964635302,
      "grad_norm": 3.6367979049682617,
      "learning_rate": 2.12967511051856e-06,
      "loss": 0.6151,
      "step": 15228
    },
    {
      "epoch": 4.808651720871487,
      "grad_norm": 3.888551950454712,
      "learning_rate": 2.1261665848010666e-06,
      "loss": 0.6695,
      "step": 15229
    },
    {
      "epoch": 4.808967477107673,
      "grad_norm": 5.032436370849609,
      "learning_rate": 2.122658059083573e-06,
      "loss": 0.7212,
      "step": 15230
    },
    {
      "epoch": 4.809283233343859,
      "grad_norm": 4.227202892303467,
      "learning_rate": 2.1191495333660798e-06,
      "loss": 0.644,
      "step": 15231
    },
    {
      "epoch": 4.809598989580044,
      "grad_norm": 3.812777042388916,
      "learning_rate": 2.115641007648586e-06,
      "loss": 0.5951,
      "step": 15232
    },
    {
      "epoch": 4.80991474581623,
      "grad_norm": 4.9225077629089355,
      "learning_rate": 2.1121324819310926e-06,
      "loss": 0.5749,
      "step": 15233
    },
    {
      "epoch": 4.810230502052415,
      "grad_norm": 3.60947847366333,
      "learning_rate": 2.108623956213599e-06,
      "loss": 0.5254,
      "step": 15234
    },
    {
      "epoch": 4.810546258288602,
      "grad_norm": 4.432349681854248,
      "learning_rate": 2.1051154304961058e-06,
      "loss": 0.5467,
      "step": 15235
    },
    {
      "epoch": 4.810862014524787,
      "grad_norm": 3.8317484855651855,
      "learning_rate": 2.101606904778612e-06,
      "loss": 0.497,
      "step": 15236
    },
    {
      "epoch": 4.811177770760972,
      "grad_norm": 4.284629821777344,
      "learning_rate": 2.0980983790611185e-06,
      "loss": 0.5797,
      "step": 15237
    },
    {
      "epoch": 4.811493526997158,
      "grad_norm": 4.104023456573486,
      "learning_rate": 2.0945898533436253e-06,
      "loss": 0.6076,
      "step": 15238
    },
    {
      "epoch": 4.811809283233344,
      "grad_norm": 4.551074981689453,
      "learning_rate": 2.0910813276261317e-06,
      "loss": 0.6774,
      "step": 15239
    },
    {
      "epoch": 4.81212503946953,
      "grad_norm": 3.9705958366394043,
      "learning_rate": 2.087572801908638e-06,
      "loss": 0.5913,
      "step": 15240
    },
    {
      "epoch": 4.812440795705715,
      "grad_norm": 4.330714225769043,
      "learning_rate": 2.0840642761911445e-06,
      "loss": 0.5054,
      "step": 15241
    },
    {
      "epoch": 4.8127565519419,
      "grad_norm": 4.148265838623047,
      "learning_rate": 2.080555750473651e-06,
      "loss": 0.5245,
      "step": 15242
    },
    {
      "epoch": 4.813072308178087,
      "grad_norm": 4.964593410491943,
      "learning_rate": 2.0770472247561573e-06,
      "loss": 0.662,
      "step": 15243
    },
    {
      "epoch": 4.813388064414272,
      "grad_norm": 4.280752658843994,
      "learning_rate": 2.073538699038664e-06,
      "loss": 0.6642,
      "step": 15244
    },
    {
      "epoch": 4.813703820650458,
      "grad_norm": 4.4186882972717285,
      "learning_rate": 2.0700301733211705e-06,
      "loss": 0.5802,
      "step": 15245
    },
    {
      "epoch": 4.814019576886643,
      "grad_norm": 4.976438522338867,
      "learning_rate": 2.066521647603677e-06,
      "loss": 0.6438,
      "step": 15246
    },
    {
      "epoch": 4.814335333122829,
      "grad_norm": 4.252461910247803,
      "learning_rate": 2.0630131218861837e-06,
      "loss": 0.6165,
      "step": 15247
    },
    {
      "epoch": 4.814651089359015,
      "grad_norm": 4.293500900268555,
      "learning_rate": 2.05950459616869e-06,
      "loss": 0.589,
      "step": 15248
    },
    {
      "epoch": 4.8149668455952,
      "grad_norm": 4.061232566833496,
      "learning_rate": 2.0559960704511965e-06,
      "loss": 0.5914,
      "step": 15249
    },
    {
      "epoch": 4.815282601831386,
      "grad_norm": 4.275032043457031,
      "learning_rate": 2.0524875447337033e-06,
      "loss": 0.5889,
      "step": 15250
    },
    {
      "epoch": 4.815598358067572,
      "grad_norm": 4.1021246910095215,
      "learning_rate": 2.0489790190162097e-06,
      "loss": 0.6161,
      "step": 15251
    },
    {
      "epoch": 4.815914114303758,
      "grad_norm": 3.695554733276367,
      "learning_rate": 2.045470493298716e-06,
      "loss": 0.609,
      "step": 15252
    },
    {
      "epoch": 4.816229870539943,
      "grad_norm": 3.922725200653076,
      "learning_rate": 2.0419619675812225e-06,
      "loss": 0.5828,
      "step": 15253
    },
    {
      "epoch": 4.816545626776129,
      "grad_norm": 3.5824429988861084,
      "learning_rate": 2.038453441863729e-06,
      "loss": 0.5345,
      "step": 15254
    },
    {
      "epoch": 4.816861383012315,
      "grad_norm": 3.920938730239868,
      "learning_rate": 2.0349449161462352e-06,
      "loss": 0.5013,
      "step": 15255
    },
    {
      "epoch": 4.8171771392485,
      "grad_norm": 4.49433708190918,
      "learning_rate": 2.031436390428742e-06,
      "loss": 0.5686,
      "step": 15256
    },
    {
      "epoch": 4.817492895484686,
      "grad_norm": 3.2295708656311035,
      "learning_rate": 2.0279278647112484e-06,
      "loss": 0.4958,
      "step": 15257
    },
    {
      "epoch": 4.817808651720871,
      "grad_norm": 3.7608959674835205,
      "learning_rate": 2.024419338993755e-06,
      "loss": 0.5922,
      "step": 15258
    },
    {
      "epoch": 4.8181244079570575,
      "grad_norm": 4.851642608642578,
      "learning_rate": 2.0209108132762616e-06,
      "loss": 0.5951,
      "step": 15259
    },
    {
      "epoch": 4.818440164193243,
      "grad_norm": 4.417990684509277,
      "learning_rate": 2.017402287558768e-06,
      "loss": 0.5908,
      "step": 15260
    },
    {
      "epoch": 4.818755920429428,
      "grad_norm": 3.9772098064422607,
      "learning_rate": 2.0138937618412744e-06,
      "loss": 0.606,
      "step": 15261
    },
    {
      "epoch": 4.819071676665614,
      "grad_norm": 3.6851203441619873,
      "learning_rate": 2.010385236123781e-06,
      "loss": 0.5179,
      "step": 15262
    },
    {
      "epoch": 4.8193874329018,
      "grad_norm": 5.2565741539001465,
      "learning_rate": 2.0068767104062876e-06,
      "loss": 0.5143,
      "step": 15263
    },
    {
      "epoch": 4.819703189137986,
      "grad_norm": 3.82332444190979,
      "learning_rate": 2.003368184688794e-06,
      "loss": 0.4959,
      "step": 15264
    },
    {
      "epoch": 4.820018945374171,
      "grad_norm": 3.9634573459625244,
      "learning_rate": 1.9998596589713004e-06,
      "loss": 0.6163,
      "step": 15265
    },
    {
      "epoch": 4.820334701610356,
      "grad_norm": 4.9279069900512695,
      "learning_rate": 1.996351133253807e-06,
      "loss": 0.5883,
      "step": 15266
    },
    {
      "epoch": 4.820650457846543,
      "grad_norm": 4.8479509353637695,
      "learning_rate": 1.992842607536313e-06,
      "loss": 0.7104,
      "step": 15267
    },
    {
      "epoch": 4.820966214082728,
      "grad_norm": 3.8235273361206055,
      "learning_rate": 1.9893340818188196e-06,
      "loss": 0.6391,
      "step": 15268
    },
    {
      "epoch": 4.821281970318914,
      "grad_norm": 4.4306254386901855,
      "learning_rate": 1.9858255561013264e-06,
      "loss": 0.5771,
      "step": 15269
    },
    {
      "epoch": 4.821597726555099,
      "grad_norm": 5.3823747634887695,
      "learning_rate": 1.9823170303838328e-06,
      "loss": 0.546,
      "step": 15270
    },
    {
      "epoch": 4.8219134827912855,
      "grad_norm": 4.31056022644043,
      "learning_rate": 1.978808504666339e-06,
      "loss": 0.5499,
      "step": 15271
    },
    {
      "epoch": 4.822229239027471,
      "grad_norm": 3.4236695766448975,
      "learning_rate": 1.975299978948846e-06,
      "loss": 0.5345,
      "step": 15272
    },
    {
      "epoch": 4.822544995263656,
      "grad_norm": 4.889766216278076,
      "learning_rate": 1.9717914532313524e-06,
      "loss": 0.6233,
      "step": 15273
    },
    {
      "epoch": 4.822860751499842,
      "grad_norm": 5.2234272956848145,
      "learning_rate": 1.9682829275138588e-06,
      "loss": 0.6115,
      "step": 15274
    },
    {
      "epoch": 4.823176507736028,
      "grad_norm": 5.24628210067749,
      "learning_rate": 1.9647744017963656e-06,
      "loss": 0.5263,
      "step": 15275
    },
    {
      "epoch": 4.823492263972214,
      "grad_norm": 3.5779781341552734,
      "learning_rate": 1.961265876078872e-06,
      "loss": 0.5159,
      "step": 15276
    },
    {
      "epoch": 4.823808020208399,
      "grad_norm": 4.1236186027526855,
      "learning_rate": 1.957757350361378e-06,
      "loss": 0.5856,
      "step": 15277
    },
    {
      "epoch": 4.824123776444585,
      "grad_norm": 4.025196552276611,
      "learning_rate": 1.9542488246438847e-06,
      "loss": 0.6107,
      "step": 15278
    },
    {
      "epoch": 4.8244395326807705,
      "grad_norm": 4.456395626068115,
      "learning_rate": 1.950740298926391e-06,
      "loss": 0.5846,
      "step": 15279
    },
    {
      "epoch": 4.824755288916956,
      "grad_norm": 4.0394134521484375,
      "learning_rate": 1.9472317732088975e-06,
      "loss": 0.6434,
      "step": 15280
    },
    {
      "epoch": 4.825071045153142,
      "grad_norm": 4.323945045471191,
      "learning_rate": 1.9437232474914043e-06,
      "loss": 0.5638,
      "step": 15281
    },
    {
      "epoch": 4.825386801389327,
      "grad_norm": 6.576891899108887,
      "learning_rate": 1.9402147217739107e-06,
      "loss": 0.5908,
      "step": 15282
    },
    {
      "epoch": 4.8257025576255135,
      "grad_norm": 4.149561882019043,
      "learning_rate": 1.936706196056417e-06,
      "loss": 0.5661,
      "step": 15283
    },
    {
      "epoch": 4.826018313861699,
      "grad_norm": 4.551365852355957,
      "learning_rate": 1.933197670338924e-06,
      "loss": 0.5398,
      "step": 15284
    },
    {
      "epoch": 4.826334070097884,
      "grad_norm": 3.7097392082214355,
      "learning_rate": 1.9296891446214303e-06,
      "loss": 0.4631,
      "step": 15285
    },
    {
      "epoch": 4.82664982633407,
      "grad_norm": 4.308566093444824,
      "learning_rate": 1.9261806189039367e-06,
      "loss": 0.5075,
      "step": 15286
    },
    {
      "epoch": 4.8269655825702555,
      "grad_norm": 5.2379279136657715,
      "learning_rate": 1.922672093186443e-06,
      "loss": 0.6505,
      "step": 15287
    },
    {
      "epoch": 4.827281338806442,
      "grad_norm": 3.092250108718872,
      "learning_rate": 1.9191635674689495e-06,
      "loss": 0.5156,
      "step": 15288
    },
    {
      "epoch": 4.827597095042627,
      "grad_norm": 3.9955766201019287,
      "learning_rate": 1.915655041751456e-06,
      "loss": 0.5558,
      "step": 15289
    },
    {
      "epoch": 4.827912851278812,
      "grad_norm": 3.826028823852539,
      "learning_rate": 1.9121465160339627e-06,
      "loss": 0.6378,
      "step": 15290
    },
    {
      "epoch": 4.8282286075149985,
      "grad_norm": 4.145681381225586,
      "learning_rate": 1.908637990316469e-06,
      "loss": 0.5859,
      "step": 15291
    },
    {
      "epoch": 4.828544363751184,
      "grad_norm": 3.8880810737609863,
      "learning_rate": 1.9051294645989755e-06,
      "loss": 0.5912,
      "step": 15292
    },
    {
      "epoch": 4.82886011998737,
      "grad_norm": 4.070903301239014,
      "learning_rate": 1.901620938881482e-06,
      "loss": 0.6342,
      "step": 15293
    },
    {
      "epoch": 4.829175876223555,
      "grad_norm": 4.232814311981201,
      "learning_rate": 1.8981124131639887e-06,
      "loss": 0.6444,
      "step": 15294
    },
    {
      "epoch": 4.8294916324597414,
      "grad_norm": 4.219300746917725,
      "learning_rate": 1.894603887446495e-06,
      "loss": 0.6312,
      "step": 15295
    },
    {
      "epoch": 4.829807388695927,
      "grad_norm": 3.7922885417938232,
      "learning_rate": 1.8910953617290017e-06,
      "loss": 0.5496,
      "step": 15296
    },
    {
      "epoch": 4.830123144932112,
      "grad_norm": 4.046741962432861,
      "learning_rate": 1.8875868360115083e-06,
      "loss": 0.5345,
      "step": 15297
    },
    {
      "epoch": 4.830438901168298,
      "grad_norm": 3.8100974559783936,
      "learning_rate": 1.8840783102940146e-06,
      "loss": 0.6098,
      "step": 15298
    },
    {
      "epoch": 4.8307546574044835,
      "grad_norm": 3.3052308559417725,
      "learning_rate": 1.8805697845765212e-06,
      "loss": 0.5494,
      "step": 15299
    },
    {
      "epoch": 4.83107041364067,
      "grad_norm": 4.751061916351318,
      "learning_rate": 1.8770612588590274e-06,
      "loss": 0.6936,
      "step": 15300
    },
    {
      "epoch": 4.831386169876855,
      "grad_norm": 4.610535144805908,
      "learning_rate": 1.8735527331415338e-06,
      "loss": 0.6219,
      "step": 15301
    },
    {
      "epoch": 4.831701926113041,
      "grad_norm": 4.018260955810547,
      "learning_rate": 1.8700442074240404e-06,
      "loss": 0.4953,
      "step": 15302
    },
    {
      "epoch": 4.8320176823492265,
      "grad_norm": 5.346950531005859,
      "learning_rate": 1.866535681706547e-06,
      "loss": 0.6436,
      "step": 15303
    },
    {
      "epoch": 4.832333438585412,
      "grad_norm": 4.338366508483887,
      "learning_rate": 1.8630271559890534e-06,
      "loss": 0.6476,
      "step": 15304
    },
    {
      "epoch": 4.832649194821598,
      "grad_norm": 4.450331687927246,
      "learning_rate": 1.85951863027156e-06,
      "loss": 0.5998,
      "step": 15305
    },
    {
      "epoch": 4.832964951057783,
      "grad_norm": 4.3482985496521,
      "learning_rate": 1.8560101045540664e-06,
      "loss": 0.5342,
      "step": 15306
    },
    {
      "epoch": 4.833280707293969,
      "grad_norm": 3.7618587017059326,
      "learning_rate": 1.852501578836573e-06,
      "loss": 0.5789,
      "step": 15307
    },
    {
      "epoch": 4.833596463530155,
      "grad_norm": 3.739999294281006,
      "learning_rate": 1.8489930531190796e-06,
      "loss": 0.5798,
      "step": 15308
    },
    {
      "epoch": 4.83391221976634,
      "grad_norm": 4.787474155426025,
      "learning_rate": 1.845484527401586e-06,
      "loss": 0.6067,
      "step": 15309
    },
    {
      "epoch": 4.834227976002526,
      "grad_norm": 4.322257995605469,
      "learning_rate": 1.8419760016840926e-06,
      "loss": 0.7071,
      "step": 15310
    },
    {
      "epoch": 4.8345437322387115,
      "grad_norm": 3.7827701568603516,
      "learning_rate": 1.8384674759665992e-06,
      "loss": 0.546,
      "step": 15311
    },
    {
      "epoch": 4.834859488474898,
      "grad_norm": 3.9460370540618896,
      "learning_rate": 1.8349589502491054e-06,
      "loss": 0.6146,
      "step": 15312
    },
    {
      "epoch": 4.835175244711083,
      "grad_norm": 3.740630865097046,
      "learning_rate": 1.8314504245316118e-06,
      "loss": 0.5578,
      "step": 15313
    },
    {
      "epoch": 4.835491000947268,
      "grad_norm": 4.15948486328125,
      "learning_rate": 1.8279418988141184e-06,
      "loss": 0.5704,
      "step": 15314
    },
    {
      "epoch": 4.835806757183454,
      "grad_norm": 4.683309078216553,
      "learning_rate": 1.8244333730966247e-06,
      "loss": 0.6761,
      "step": 15315
    },
    {
      "epoch": 4.83612251341964,
      "grad_norm": 3.737537384033203,
      "learning_rate": 1.8209248473791313e-06,
      "loss": 0.5608,
      "step": 15316
    },
    {
      "epoch": 4.836438269655826,
      "grad_norm": 4.314555644989014,
      "learning_rate": 1.817416321661638e-06,
      "loss": 0.6154,
      "step": 15317
    },
    {
      "epoch": 4.836754025892011,
      "grad_norm": 5.026854515075684,
      "learning_rate": 1.8139077959441443e-06,
      "loss": 0.6718,
      "step": 15318
    },
    {
      "epoch": 4.837069782128197,
      "grad_norm": 3.981086015701294,
      "learning_rate": 1.810399270226651e-06,
      "loss": 0.5309,
      "step": 15319
    },
    {
      "epoch": 4.837385538364383,
      "grad_norm": 4.152122974395752,
      "learning_rate": 1.8068907445091573e-06,
      "loss": 0.5966,
      "step": 15320
    },
    {
      "epoch": 4.837701294600569,
      "grad_norm": 4.39628791809082,
      "learning_rate": 1.803382218791664e-06,
      "loss": 0.7105,
      "step": 15321
    },
    {
      "epoch": 4.838017050836754,
      "grad_norm": 4.9218573570251465,
      "learning_rate": 1.7998736930741705e-06,
      "loss": 0.5582,
      "step": 15322
    },
    {
      "epoch": 4.8383328070729394,
      "grad_norm": 4.484362602233887,
      "learning_rate": 1.796365167356677e-06,
      "loss": 0.6535,
      "step": 15323
    },
    {
      "epoch": 4.838648563309126,
      "grad_norm": 3.602555274963379,
      "learning_rate": 1.792856641639183e-06,
      "loss": 0.5537,
      "step": 15324
    },
    {
      "epoch": 4.838964319545311,
      "grad_norm": 3.6563265323638916,
      "learning_rate": 1.7893481159216897e-06,
      "loss": 0.5264,
      "step": 15325
    },
    {
      "epoch": 4.839280075781497,
      "grad_norm": 3.265838623046875,
      "learning_rate": 1.785839590204196e-06,
      "loss": 0.5097,
      "step": 15326
    },
    {
      "epoch": 4.839595832017682,
      "grad_norm": 3.5738508701324463,
      "learning_rate": 1.7823310644867027e-06,
      "loss": 0.5264,
      "step": 15327
    },
    {
      "epoch": 4.839911588253868,
      "grad_norm": 4.620321273803711,
      "learning_rate": 1.7788225387692093e-06,
      "loss": 0.66,
      "step": 15328
    },
    {
      "epoch": 4.840227344490054,
      "grad_norm": 3.8160018920898438,
      "learning_rate": 1.7753140130517157e-06,
      "loss": 0.5573,
      "step": 15329
    },
    {
      "epoch": 4.840543100726239,
      "grad_norm": 4.219690322875977,
      "learning_rate": 1.7718054873342223e-06,
      "loss": 0.5459,
      "step": 15330
    },
    {
      "epoch": 4.840858856962425,
      "grad_norm": 4.20612096786499,
      "learning_rate": 1.7682969616167289e-06,
      "loss": 0.5796,
      "step": 15331
    },
    {
      "epoch": 4.841174613198611,
      "grad_norm": 3.572570323944092,
      "learning_rate": 1.7647884358992353e-06,
      "loss": 0.5405,
      "step": 15332
    },
    {
      "epoch": 4.841490369434796,
      "grad_norm": 5.730818748474121,
      "learning_rate": 1.7612799101817419e-06,
      "loss": 0.7035,
      "step": 15333
    },
    {
      "epoch": 4.841806125670982,
      "grad_norm": 4.714581489562988,
      "learning_rate": 1.7577713844642483e-06,
      "loss": 0.6013,
      "step": 15334
    },
    {
      "epoch": 4.842121881907167,
      "grad_norm": 4.743396759033203,
      "learning_rate": 1.7542628587467544e-06,
      "loss": 0.5961,
      "step": 15335
    },
    {
      "epoch": 4.842437638143354,
      "grad_norm": 3.8656399250030518,
      "learning_rate": 1.750754333029261e-06,
      "loss": 0.6071,
      "step": 15336
    },
    {
      "epoch": 4.842753394379539,
      "grad_norm": 4.586223602294922,
      "learning_rate": 1.7472458073117676e-06,
      "loss": 0.5434,
      "step": 15337
    },
    {
      "epoch": 4.843069150615724,
      "grad_norm": 5.231536865234375,
      "learning_rate": 1.743737281594274e-06,
      "loss": 0.5668,
      "step": 15338
    },
    {
      "epoch": 4.84338490685191,
      "grad_norm": 4.383423805236816,
      "learning_rate": 1.7402287558767806e-06,
      "loss": 0.5694,
      "step": 15339
    },
    {
      "epoch": 4.843700663088096,
      "grad_norm": 4.686093807220459,
      "learning_rate": 1.736720230159287e-06,
      "loss": 0.5801,
      "step": 15340
    },
    {
      "epoch": 4.844016419324282,
      "grad_norm": 4.068689823150635,
      "learning_rate": 1.7332117044417936e-06,
      "loss": 0.53,
      "step": 15341
    },
    {
      "epoch": 4.844332175560467,
      "grad_norm": 3.7304301261901855,
      "learning_rate": 1.7297031787243002e-06,
      "loss": 0.5841,
      "step": 15342
    },
    {
      "epoch": 4.844647931796653,
      "grad_norm": 4.155467987060547,
      "learning_rate": 1.7261946530068066e-06,
      "loss": 0.6489,
      "step": 15343
    },
    {
      "epoch": 4.844963688032839,
      "grad_norm": 3.9858992099761963,
      "learning_rate": 1.7226861272893132e-06,
      "loss": 0.5752,
      "step": 15344
    },
    {
      "epoch": 4.845279444269025,
      "grad_norm": 4.287031650543213,
      "learning_rate": 1.7191776015718198e-06,
      "loss": 0.5868,
      "step": 15345
    },
    {
      "epoch": 4.84559520050521,
      "grad_norm": 6.190001487731934,
      "learning_rate": 1.7156690758543262e-06,
      "loss": 0.6575,
      "step": 15346
    },
    {
      "epoch": 4.845910956741395,
      "grad_norm": 3.5481274127960205,
      "learning_rate": 1.7121605501368324e-06,
      "loss": 0.5963,
      "step": 15347
    },
    {
      "epoch": 4.846226712977582,
      "grad_norm": 4.500671863555908,
      "learning_rate": 1.708652024419339e-06,
      "loss": 0.6466,
      "step": 15348
    },
    {
      "epoch": 4.846542469213767,
      "grad_norm": 4.521997928619385,
      "learning_rate": 1.7051434987018454e-06,
      "loss": 0.6052,
      "step": 15349
    },
    {
      "epoch": 4.846858225449953,
      "grad_norm": 4.298007965087891,
      "learning_rate": 1.701634972984352e-06,
      "loss": 0.599,
      "step": 15350
    },
    {
      "epoch": 4.847173981686138,
      "grad_norm": 4.331653594970703,
      "learning_rate": 1.6981264472668586e-06,
      "loss": 0.7057,
      "step": 15351
    },
    {
      "epoch": 4.847489737922324,
      "grad_norm": 4.130436420440674,
      "learning_rate": 1.694617921549365e-06,
      "loss": 0.5485,
      "step": 15352
    },
    {
      "epoch": 4.84780549415851,
      "grad_norm": 4.368161678314209,
      "learning_rate": 1.6911093958318716e-06,
      "loss": 0.6063,
      "step": 15353
    },
    {
      "epoch": 4.848121250394695,
      "grad_norm": 4.02249813079834,
      "learning_rate": 1.687600870114378e-06,
      "loss": 0.6131,
      "step": 15354
    },
    {
      "epoch": 4.848437006630881,
      "grad_norm": 4.831883430480957,
      "learning_rate": 1.6840923443968846e-06,
      "loss": 0.6219,
      "step": 15355
    },
    {
      "epoch": 4.848752762867067,
      "grad_norm": 3.7902402877807617,
      "learning_rate": 1.6805838186793912e-06,
      "loss": 0.5177,
      "step": 15356
    },
    {
      "epoch": 4.849068519103252,
      "grad_norm": 4.471348762512207,
      "learning_rate": 1.6770752929618976e-06,
      "loss": 0.5743,
      "step": 15357
    },
    {
      "epoch": 4.849384275339438,
      "grad_norm": 4.0753631591796875,
      "learning_rate": 1.6735667672444042e-06,
      "loss": 0.6008,
      "step": 15358
    },
    {
      "epoch": 4.849700031575623,
      "grad_norm": 4.162436485290527,
      "learning_rate": 1.6700582415269103e-06,
      "loss": 0.5275,
      "step": 15359
    },
    {
      "epoch": 4.8500157878118095,
      "grad_norm": 3.872540235519409,
      "learning_rate": 1.666549715809417e-06,
      "loss": 0.5925,
      "step": 15360
    },
    {
      "epoch": 4.850331544047995,
      "grad_norm": 4.991692066192627,
      "learning_rate": 1.6630411900919233e-06,
      "loss": 0.5193,
      "step": 15361
    },
    {
      "epoch": 4.85064730028418,
      "grad_norm": 3.6994612216949463,
      "learning_rate": 1.65953266437443e-06,
      "loss": 0.5403,
      "step": 15362
    },
    {
      "epoch": 4.850963056520366,
      "grad_norm": 3.392991781234741,
      "learning_rate": 1.6560241386569363e-06,
      "loss": 0.5409,
      "step": 15363
    },
    {
      "epoch": 4.851278812756552,
      "grad_norm": 3.4618992805480957,
      "learning_rate": 1.652515612939443e-06,
      "loss": 0.5612,
      "step": 15364
    },
    {
      "epoch": 4.851594568992738,
      "grad_norm": 4.136562824249268,
      "learning_rate": 1.6490070872219495e-06,
      "loss": 0.5798,
      "step": 15365
    },
    {
      "epoch": 4.851910325228923,
      "grad_norm": 4.456493854522705,
      "learning_rate": 1.645498561504456e-06,
      "loss": 0.6312,
      "step": 15366
    },
    {
      "epoch": 4.852226081465109,
      "grad_norm": 4.740408897399902,
      "learning_rate": 1.6419900357869625e-06,
      "loss": 0.6292,
      "step": 15367
    },
    {
      "epoch": 4.8525418377012945,
      "grad_norm": 4.419093132019043,
      "learning_rate": 1.6384815100694689e-06,
      "loss": 0.666,
      "step": 15368
    },
    {
      "epoch": 4.852857593937481,
      "grad_norm": 4.339678764343262,
      "learning_rate": 1.6349729843519755e-06,
      "loss": 0.6385,
      "step": 15369
    },
    {
      "epoch": 4.853173350173666,
      "grad_norm": 3.97847580909729,
      "learning_rate": 1.631464458634482e-06,
      "loss": 0.5986,
      "step": 15370
    },
    {
      "epoch": 4.853489106409851,
      "grad_norm": 4.1046142578125,
      "learning_rate": 1.6279559329169883e-06,
      "loss": 0.6344,
      "step": 15371
    },
    {
      "epoch": 4.8538048626460375,
      "grad_norm": 3.749755859375,
      "learning_rate": 1.6244474071994947e-06,
      "loss": 0.5671,
      "step": 15372
    },
    {
      "epoch": 4.854120618882223,
      "grad_norm": 3.366426944732666,
      "learning_rate": 1.6209388814820013e-06,
      "loss": 0.5899,
      "step": 15373
    },
    {
      "epoch": 4.854436375118409,
      "grad_norm": 4.400730133056641,
      "learning_rate": 1.6174303557645079e-06,
      "loss": 0.6057,
      "step": 15374
    },
    {
      "epoch": 4.854752131354594,
      "grad_norm": 4.387651443481445,
      "learning_rate": 1.6139218300470143e-06,
      "loss": 0.5269,
      "step": 15375
    },
    {
      "epoch": 4.85506788759078,
      "grad_norm": 3.5564844608306885,
      "learning_rate": 1.6104133043295209e-06,
      "loss": 0.5559,
      "step": 15376
    },
    {
      "epoch": 4.855383643826966,
      "grad_norm": 3.914191722869873,
      "learning_rate": 1.6069047786120272e-06,
      "loss": 0.6967,
      "step": 15377
    },
    {
      "epoch": 4.855699400063151,
      "grad_norm": 5.211474418640137,
      "learning_rate": 1.6033962528945338e-06,
      "loss": 0.7152,
      "step": 15378
    },
    {
      "epoch": 4.856015156299337,
      "grad_norm": 4.136655807495117,
      "learning_rate": 1.5998877271770404e-06,
      "loss": 0.5589,
      "step": 15379
    },
    {
      "epoch": 4.8563309125355225,
      "grad_norm": 4.333061695098877,
      "learning_rate": 1.5963792014595468e-06,
      "loss": 0.6006,
      "step": 15380
    },
    {
      "epoch": 4.856646668771708,
      "grad_norm": 3.944422721862793,
      "learning_rate": 1.5928706757420534e-06,
      "loss": 0.588,
      "step": 15381
    },
    {
      "epoch": 4.856962425007894,
      "grad_norm": 5.323562145233154,
      "learning_rate": 1.5893621500245596e-06,
      "loss": 0.6068,
      "step": 15382
    },
    {
      "epoch": 4.857278181244079,
      "grad_norm": 4.033251762390137,
      "learning_rate": 1.585853624307066e-06,
      "loss": 0.5281,
      "step": 15383
    },
    {
      "epoch": 4.8575939374802655,
      "grad_norm": 3.9111502170562744,
      "learning_rate": 1.5823450985895726e-06,
      "loss": 0.5579,
      "step": 15384
    },
    {
      "epoch": 4.857909693716451,
      "grad_norm": 3.7338759899139404,
      "learning_rate": 1.5788365728720792e-06,
      "loss": 0.5132,
      "step": 15385
    },
    {
      "epoch": 4.858225449952637,
      "grad_norm": 3.843914747238159,
      "learning_rate": 1.5753280471545856e-06,
      "loss": 0.5597,
      "step": 15386
    },
    {
      "epoch": 4.858541206188822,
      "grad_norm": 3.587995767593384,
      "learning_rate": 1.5718195214370922e-06,
      "loss": 0.4647,
      "step": 15387
    },
    {
      "epoch": 4.8588569624250075,
      "grad_norm": 5.022590160369873,
      "learning_rate": 1.5683109957195988e-06,
      "loss": 0.6491,
      "step": 15388
    },
    {
      "epoch": 4.859172718661194,
      "grad_norm": 4.967799186706543,
      "learning_rate": 1.5648024700021052e-06,
      "loss": 0.5795,
      "step": 15389
    },
    {
      "epoch": 4.859488474897379,
      "grad_norm": 4.563784122467041,
      "learning_rate": 1.5612939442846118e-06,
      "loss": 0.5908,
      "step": 15390
    },
    {
      "epoch": 4.859804231133565,
      "grad_norm": 4.8660125732421875,
      "learning_rate": 1.5577854185671182e-06,
      "loss": 0.6481,
      "step": 15391
    },
    {
      "epoch": 4.8601199873697505,
      "grad_norm": 4.410865783691406,
      "learning_rate": 1.5542768928496246e-06,
      "loss": 0.6666,
      "step": 15392
    },
    {
      "epoch": 4.860435743605937,
      "grad_norm": 3.931494951248169,
      "learning_rate": 1.5507683671321312e-06,
      "loss": 0.5974,
      "step": 15393
    },
    {
      "epoch": 4.860751499842122,
      "grad_norm": 3.787513017654419,
      "learning_rate": 1.5472598414146376e-06,
      "loss": 0.5531,
      "step": 15394
    },
    {
      "epoch": 4.861067256078307,
      "grad_norm": 4.235924243927002,
      "learning_rate": 1.5437513156971442e-06,
      "loss": 0.6393,
      "step": 15395
    },
    {
      "epoch": 4.861383012314493,
      "grad_norm": 4.156938552856445,
      "learning_rate": 1.5402427899796508e-06,
      "loss": 0.5955,
      "step": 15396
    },
    {
      "epoch": 4.861698768550679,
      "grad_norm": 3.6497063636779785,
      "learning_rate": 1.536734264262157e-06,
      "loss": 0.5255,
      "step": 15397
    },
    {
      "epoch": 4.862014524786865,
      "grad_norm": 3.888889789581299,
      "learning_rate": 1.5332257385446635e-06,
      "loss": 0.535,
      "step": 15398
    },
    {
      "epoch": 4.86233028102305,
      "grad_norm": 4.397578716278076,
      "learning_rate": 1.5297172128271701e-06,
      "loss": 0.6148,
      "step": 15399
    },
    {
      "epoch": 4.8626460372592355,
      "grad_norm": 3.7720837593078613,
      "learning_rate": 1.5262086871096765e-06,
      "loss": 0.5861,
      "step": 15400
    },
    {
      "epoch": 4.8626460372592355,
      "eval_loss": 0.8372291922569275,
      "eval_runtime": 4.5992,
      "eval_samples_per_second": 541.396,
      "eval_steps_per_second": 4.566,
      "step": 15400
    },
    {
      "epoch": 4.862961793495422,
      "grad_norm": 3.5074622631073,
      "learning_rate": 1.5227001613921831e-06,
      "loss": 0.551,
      "step": 15401
    },
    {
      "epoch": 4.863277549731607,
      "grad_norm": 4.235756874084473,
      "learning_rate": 1.5191916356746897e-06,
      "loss": 0.6443,
      "step": 15402
    },
    {
      "epoch": 4.863593305967793,
      "grad_norm": 5.23772668838501,
      "learning_rate": 1.515683109957196e-06,
      "loss": 0.6007,
      "step": 15403
    },
    {
      "epoch": 4.8639090622039785,
      "grad_norm": 3.742884635925293,
      "learning_rate": 1.5121745842397025e-06,
      "loss": 0.5688,
      "step": 15404
    },
    {
      "epoch": 4.864224818440164,
      "grad_norm": 4.450394153594971,
      "learning_rate": 1.5086660585222091e-06,
      "loss": 0.4923,
      "step": 15405
    },
    {
      "epoch": 4.86454057467635,
      "grad_norm": 4.3864545822143555,
      "learning_rate": 1.5051575328047155e-06,
      "loss": 0.5739,
      "step": 15406
    },
    {
      "epoch": 4.864856330912535,
      "grad_norm": 4.370642185211182,
      "learning_rate": 1.5016490070872221e-06,
      "loss": 0.694,
      "step": 15407
    },
    {
      "epoch": 4.865172087148721,
      "grad_norm": 3.7343525886535645,
      "learning_rate": 1.4981404813697285e-06,
      "loss": 0.5308,
      "step": 15408
    },
    {
      "epoch": 4.865487843384907,
      "grad_norm": 4.126283168792725,
      "learning_rate": 1.4946319556522349e-06,
      "loss": 0.7154,
      "step": 15409
    },
    {
      "epoch": 4.865803599621093,
      "grad_norm": 4.3616943359375,
      "learning_rate": 1.4911234299347415e-06,
      "loss": 0.5201,
      "step": 15410
    },
    {
      "epoch": 4.866119355857278,
      "grad_norm": 3.611717939376831,
      "learning_rate": 1.4876149042172479e-06,
      "loss": 0.6598,
      "step": 15411
    },
    {
      "epoch": 4.8664351120934635,
      "grad_norm": 4.374582767486572,
      "learning_rate": 1.4841063784997545e-06,
      "loss": 0.5928,
      "step": 15412
    },
    {
      "epoch": 4.86675086832965,
      "grad_norm": 3.887685537338257,
      "learning_rate": 1.480597852782261e-06,
      "loss": 0.5251,
      "step": 15413
    },
    {
      "epoch": 4.867066624565835,
      "grad_norm": 4.01451301574707,
      "learning_rate": 1.4770893270647675e-06,
      "loss": 0.5718,
      "step": 15414
    },
    {
      "epoch": 4.867382380802021,
      "grad_norm": 3.940728187561035,
      "learning_rate": 1.4735808013472739e-06,
      "loss": 0.6735,
      "step": 15415
    },
    {
      "epoch": 4.867698137038206,
      "grad_norm": 3.9932918548583984,
      "learning_rate": 1.4700722756297805e-06,
      "loss": 0.5371,
      "step": 15416
    },
    {
      "epoch": 4.868013893274393,
      "grad_norm": 4.12600564956665,
      "learning_rate": 1.4665637499122868e-06,
      "loss": 0.7117,
      "step": 15417
    },
    {
      "epoch": 4.868329649510578,
      "grad_norm": 4.123805522918701,
      "learning_rate": 1.4630552241947934e-06,
      "loss": 0.576,
      "step": 15418
    },
    {
      "epoch": 4.868645405746763,
      "grad_norm": 3.9289047718048096,
      "learning_rate": 1.4595466984773e-06,
      "loss": 0.6436,
      "step": 15419
    },
    {
      "epoch": 4.868961161982949,
      "grad_norm": 4.664611339569092,
      "learning_rate": 1.4560381727598064e-06,
      "loss": 0.6617,
      "step": 15420
    },
    {
      "epoch": 4.869276918219135,
      "grad_norm": 4.233947277069092,
      "learning_rate": 1.4525296470423128e-06,
      "loss": 0.5593,
      "step": 15421
    },
    {
      "epoch": 4.869592674455321,
      "grad_norm": 4.185688495635986,
      "learning_rate": 1.4490211213248194e-06,
      "loss": 0.4714,
      "step": 15422
    },
    {
      "epoch": 4.869908430691506,
      "grad_norm": 3.980522871017456,
      "learning_rate": 1.4455125956073258e-06,
      "loss": 0.673,
      "step": 15423
    },
    {
      "epoch": 4.870224186927691,
      "grad_norm": 3.9960036277770996,
      "learning_rate": 1.4420040698898324e-06,
      "loss": 0.6161,
      "step": 15424
    },
    {
      "epoch": 4.870539943163878,
      "grad_norm": 3.916792392730713,
      "learning_rate": 1.4384955441723388e-06,
      "loss": 0.5469,
      "step": 15425
    },
    {
      "epoch": 4.870855699400063,
      "grad_norm": 4.646505832672119,
      "learning_rate": 1.4349870184548454e-06,
      "loss": 0.6744,
      "step": 15426
    },
    {
      "epoch": 4.871171455636249,
      "grad_norm": 4.411970615386963,
      "learning_rate": 1.4314784927373518e-06,
      "loss": 0.5915,
      "step": 15427
    },
    {
      "epoch": 4.871487211872434,
      "grad_norm": 4.4406304359436035,
      "learning_rate": 1.4279699670198582e-06,
      "loss": 0.5475,
      "step": 15428
    },
    {
      "epoch": 4.87180296810862,
      "grad_norm": 3.0936219692230225,
      "learning_rate": 1.4244614413023648e-06,
      "loss": 0.5448,
      "step": 15429
    },
    {
      "epoch": 4.872118724344806,
      "grad_norm": 3.7634027004241943,
      "learning_rate": 1.4209529155848714e-06,
      "loss": 0.5555,
      "step": 15430
    },
    {
      "epoch": 4.872434480580991,
      "grad_norm": 3.52605938911438,
      "learning_rate": 1.4174443898673778e-06,
      "loss": 0.5107,
      "step": 15431
    },
    {
      "epoch": 4.872750236817177,
      "grad_norm": 4.463150501251221,
      "learning_rate": 1.4139358641498842e-06,
      "loss": 0.6836,
      "step": 15432
    },
    {
      "epoch": 4.873065993053363,
      "grad_norm": 4.012449264526367,
      "learning_rate": 1.4104273384323908e-06,
      "loss": 0.6192,
      "step": 15433
    },
    {
      "epoch": 4.873381749289549,
      "grad_norm": 3.927941083908081,
      "learning_rate": 1.4069188127148972e-06,
      "loss": 0.7196,
      "step": 15434
    },
    {
      "epoch": 4.873697505525734,
      "grad_norm": 3.598794937133789,
      "learning_rate": 1.4034102869974038e-06,
      "loss": 0.5043,
      "step": 15435
    },
    {
      "epoch": 4.87401326176192,
      "grad_norm": 3.89810848236084,
      "learning_rate": 1.3999017612799104e-06,
      "loss": 0.6067,
      "step": 15436
    },
    {
      "epoch": 4.874329017998106,
      "grad_norm": 5.016238689422607,
      "learning_rate": 1.3963932355624168e-06,
      "loss": 0.6303,
      "step": 15437
    },
    {
      "epoch": 4.874644774234291,
      "grad_norm": 4.900850296020508,
      "learning_rate": 1.3928847098449231e-06,
      "loss": 0.5816,
      "step": 15438
    },
    {
      "epoch": 4.874960530470477,
      "grad_norm": 4.255559921264648,
      "learning_rate": 1.3893761841274297e-06,
      "loss": 0.503,
      "step": 15439
    },
    {
      "epoch": 4.875276286706662,
      "grad_norm": 4.628162384033203,
      "learning_rate": 1.3858676584099361e-06,
      "loss": 0.5842,
      "step": 15440
    },
    {
      "epoch": 4.8755920429428485,
      "grad_norm": 3.865921974182129,
      "learning_rate": 1.3823591326924427e-06,
      "loss": 0.5899,
      "step": 15441
    },
    {
      "epoch": 4.875907799179034,
      "grad_norm": 3.517033100128174,
      "learning_rate": 1.3788506069749491e-06,
      "loss": 0.587,
      "step": 15442
    },
    {
      "epoch": 4.876223555415219,
      "grad_norm": 3.798529624938965,
      "learning_rate": 1.3753420812574557e-06,
      "loss": 0.571,
      "step": 15443
    },
    {
      "epoch": 4.876539311651405,
      "grad_norm": 4.148573875427246,
      "learning_rate": 1.3718335555399621e-06,
      "loss": 0.5641,
      "step": 15444
    },
    {
      "epoch": 4.876855067887591,
      "grad_norm": 4.431605339050293,
      "learning_rate": 1.3683250298224687e-06,
      "loss": 0.5972,
      "step": 15445
    },
    {
      "epoch": 4.877170824123777,
      "grad_norm": 3.7785441875457764,
      "learning_rate": 1.3648165041049751e-06,
      "loss": 0.5444,
      "step": 15446
    },
    {
      "epoch": 4.877486580359962,
      "grad_norm": 4.227197647094727,
      "learning_rate": 1.3613079783874817e-06,
      "loss": 0.6491,
      "step": 15447
    },
    {
      "epoch": 4.877802336596147,
      "grad_norm": 4.026279926300049,
      "learning_rate": 1.357799452669988e-06,
      "loss": 0.5934,
      "step": 15448
    },
    {
      "epoch": 4.8781180928323336,
      "grad_norm": 4.228762626647949,
      "learning_rate": 1.3542909269524947e-06,
      "loss": 0.5594,
      "step": 15449
    },
    {
      "epoch": 4.878433849068519,
      "grad_norm": 4.16111421585083,
      "learning_rate": 1.350782401235001e-06,
      "loss": 0.6465,
      "step": 15450
    },
    {
      "epoch": 4.878749605304705,
      "grad_norm": 4.391034126281738,
      "learning_rate": 1.3472738755175075e-06,
      "loss": 0.7003,
      "step": 15451
    },
    {
      "epoch": 4.87906536154089,
      "grad_norm": 5.405848026275635,
      "learning_rate": 1.343765349800014e-06,
      "loss": 0.67,
      "step": 15452
    },
    {
      "epoch": 4.879381117777076,
      "grad_norm": 3.891871213912964,
      "learning_rate": 1.3402568240825207e-06,
      "loss": 0.5571,
      "step": 15453
    },
    {
      "epoch": 4.879696874013262,
      "grad_norm": 3.374819040298462,
      "learning_rate": 1.336748298365027e-06,
      "loss": 0.4738,
      "step": 15454
    },
    {
      "epoch": 4.880012630249447,
      "grad_norm": 3.6065115928649902,
      "learning_rate": 1.3332397726475337e-06,
      "loss": 0.5335,
      "step": 15455
    },
    {
      "epoch": 4.880328386485633,
      "grad_norm": 3.8647871017456055,
      "learning_rate": 1.32973124693004e-06,
      "loss": 0.6221,
      "step": 15456
    },
    {
      "epoch": 4.880644142721819,
      "grad_norm": 5.431600570678711,
      "learning_rate": 1.3262227212125465e-06,
      "loss": 0.5963,
      "step": 15457
    },
    {
      "epoch": 4.880959898958005,
      "grad_norm": 4.752456188201904,
      "learning_rate": 1.322714195495053e-06,
      "loss": 0.5851,
      "step": 15458
    },
    {
      "epoch": 4.88127565519419,
      "grad_norm": 3.9173800945281982,
      "learning_rate": 1.3192056697775597e-06,
      "loss": 0.5189,
      "step": 15459
    },
    {
      "epoch": 4.881591411430376,
      "grad_norm": 4.635584354400635,
      "learning_rate": 1.315697144060066e-06,
      "loss": 0.5806,
      "step": 15460
    },
    {
      "epoch": 4.8819071676665615,
      "grad_norm": 3.229290723800659,
      "learning_rate": 1.3121886183425726e-06,
      "loss": 0.5125,
      "step": 15461
    },
    {
      "epoch": 4.882222923902747,
      "grad_norm": 3.571394443511963,
      "learning_rate": 1.308680092625079e-06,
      "loss": 0.6332,
      "step": 15462
    },
    {
      "epoch": 4.882538680138933,
      "grad_norm": 3.656883478164673,
      "learning_rate": 1.3051715669075854e-06,
      "loss": 0.5926,
      "step": 15463
    },
    {
      "epoch": 4.882854436375118,
      "grad_norm": 3.787851095199585,
      "learning_rate": 1.301663041190092e-06,
      "loss": 0.6435,
      "step": 15464
    },
    {
      "epoch": 4.8831701926113045,
      "grad_norm": 4.072897911071777,
      "learning_rate": 1.2981545154725984e-06,
      "loss": 0.5833,
      "step": 15465
    },
    {
      "epoch": 4.88348594884749,
      "grad_norm": 4.287287712097168,
      "learning_rate": 1.294645989755105e-06,
      "loss": 0.6229,
      "step": 15466
    },
    {
      "epoch": 4.883801705083675,
      "grad_norm": 3.748347043991089,
      "learning_rate": 1.2911374640376116e-06,
      "loss": 0.5248,
      "step": 15467
    },
    {
      "epoch": 4.884117461319861,
      "grad_norm": 4.7126641273498535,
      "learning_rate": 1.2876289383201178e-06,
      "loss": 0.5823,
      "step": 15468
    },
    {
      "epoch": 4.8844332175560465,
      "grad_norm": 5.080380439758301,
      "learning_rate": 1.2841204126026244e-06,
      "loss": 0.5012,
      "step": 15469
    },
    {
      "epoch": 4.884748973792233,
      "grad_norm": 3.9549639225006104,
      "learning_rate": 1.280611886885131e-06,
      "loss": 0.6112,
      "step": 15470
    },
    {
      "epoch": 4.885064730028418,
      "grad_norm": 4.182096481323242,
      "learning_rate": 1.2771033611676374e-06,
      "loss": 0.6365,
      "step": 15471
    },
    {
      "epoch": 4.885380486264603,
      "grad_norm": 3.651371479034424,
      "learning_rate": 1.273594835450144e-06,
      "loss": 0.6028,
      "step": 15472
    },
    {
      "epoch": 4.8856962425007895,
      "grad_norm": 4.0962724685668945,
      "learning_rate": 1.2700863097326506e-06,
      "loss": 0.5974,
      "step": 15473
    },
    {
      "epoch": 4.886011998736975,
      "grad_norm": 3.951383352279663,
      "learning_rate": 1.2665777840151568e-06,
      "loss": 0.5816,
      "step": 15474
    },
    {
      "epoch": 4.886327754973161,
      "grad_norm": 3.831339120864868,
      "learning_rate": 1.2630692582976634e-06,
      "loss": 0.6652,
      "step": 15475
    },
    {
      "epoch": 4.886643511209346,
      "grad_norm": 3.677882671356201,
      "learning_rate": 1.25956073258017e-06,
      "loss": 0.6043,
      "step": 15476
    },
    {
      "epoch": 4.8869592674455316,
      "grad_norm": 3.5723917484283447,
      "learning_rate": 1.2560522068626764e-06,
      "loss": 0.695,
      "step": 15477
    },
    {
      "epoch": 4.887275023681718,
      "grad_norm": 4.102312088012695,
      "learning_rate": 1.252543681145183e-06,
      "loss": 0.5753,
      "step": 15478
    },
    {
      "epoch": 4.887590779917903,
      "grad_norm": 5.003083229064941,
      "learning_rate": 1.2490351554276893e-06,
      "loss": 0.5798,
      "step": 15479
    },
    {
      "epoch": 4.887906536154089,
      "grad_norm": 3.381455898284912,
      "learning_rate": 1.2455266297101957e-06,
      "loss": 0.4623,
      "step": 15480
    },
    {
      "epoch": 4.8882222923902745,
      "grad_norm": 4.212071418762207,
      "learning_rate": 1.2420181039927023e-06,
      "loss": 0.6191,
      "step": 15481
    },
    {
      "epoch": 4.888538048626461,
      "grad_norm": 3.5699479579925537,
      "learning_rate": 1.2385095782752087e-06,
      "loss": 0.5385,
      "step": 15482
    },
    {
      "epoch": 4.888853804862646,
      "grad_norm": 4.06947135925293,
      "learning_rate": 1.2350010525577153e-06,
      "loss": 0.6558,
      "step": 15483
    },
    {
      "epoch": 4.889169561098832,
      "grad_norm": 4.174320220947266,
      "learning_rate": 1.231492526840222e-06,
      "loss": 0.6671,
      "step": 15484
    },
    {
      "epoch": 4.8894853173350175,
      "grad_norm": 5.142139911651611,
      "learning_rate": 1.2279840011227281e-06,
      "loss": 0.6062,
      "step": 15485
    },
    {
      "epoch": 4.889801073571203,
      "grad_norm": 5.056429386138916,
      "learning_rate": 1.2244754754052347e-06,
      "loss": 0.5617,
      "step": 15486
    },
    {
      "epoch": 4.890116829807389,
      "grad_norm": 3.8185365200042725,
      "learning_rate": 1.2209669496877413e-06,
      "loss": 0.6036,
      "step": 15487
    },
    {
      "epoch": 4.890432586043574,
      "grad_norm": 4.382867336273193,
      "learning_rate": 1.2174584239702477e-06,
      "loss": 0.5914,
      "step": 15488
    },
    {
      "epoch": 4.89074834227976,
      "grad_norm": 4.11430549621582,
      "learning_rate": 1.2139498982527543e-06,
      "loss": 0.4572,
      "step": 15489
    },
    {
      "epoch": 4.891064098515946,
      "grad_norm": 4.693174362182617,
      "learning_rate": 1.210441372535261e-06,
      "loss": 0.5969,
      "step": 15490
    },
    {
      "epoch": 4.891379854752131,
      "grad_norm": 4.12977933883667,
      "learning_rate": 1.206932846817767e-06,
      "loss": 0.5552,
      "step": 15491
    },
    {
      "epoch": 4.891695610988317,
      "grad_norm": 3.375093460083008,
      "learning_rate": 1.2034243211002737e-06,
      "loss": 0.4806,
      "step": 15492
    },
    {
      "epoch": 4.8920113672245025,
      "grad_norm": 3.401186943054199,
      "learning_rate": 1.1999157953827803e-06,
      "loss": 0.5246,
      "step": 15493
    },
    {
      "epoch": 4.892327123460689,
      "grad_norm": 5.30288553237915,
      "learning_rate": 1.1964072696652867e-06,
      "loss": 0.612,
      "step": 15494
    },
    {
      "epoch": 4.892642879696874,
      "grad_norm": 3.6837522983551025,
      "learning_rate": 1.1928987439477933e-06,
      "loss": 0.5656,
      "step": 15495
    },
    {
      "epoch": 4.892958635933059,
      "grad_norm": 4.526764392852783,
      "learning_rate": 1.1893902182302997e-06,
      "loss": 0.6323,
      "step": 15496
    },
    {
      "epoch": 4.893274392169245,
      "grad_norm": 3.5937249660491943,
      "learning_rate": 1.185881692512806e-06,
      "loss": 0.57,
      "step": 15497
    },
    {
      "epoch": 4.893590148405431,
      "grad_norm": 4.499788284301758,
      "learning_rate": 1.1823731667953127e-06,
      "loss": 0.4914,
      "step": 15498
    },
    {
      "epoch": 4.893905904641617,
      "grad_norm": 4.044368743896484,
      "learning_rate": 1.178864641077819e-06,
      "loss": 0.6095,
      "step": 15499
    },
    {
      "epoch": 4.894221660877802,
      "grad_norm": 3.6260921955108643,
      "learning_rate": 1.1753561153603256e-06,
      "loss": 0.6023,
      "step": 15500
    },
    {
      "epoch": 4.894537417113988,
      "grad_norm": 3.9965643882751465,
      "learning_rate": 1.1718475896428322e-06,
      "loss": 0.5903,
      "step": 15501
    },
    {
      "epoch": 4.894853173350174,
      "grad_norm": 4.119852066040039,
      "learning_rate": 1.1683390639253386e-06,
      "loss": 0.5907,
      "step": 15502
    },
    {
      "epoch": 4.895168929586359,
      "grad_norm": 4.189892768859863,
      "learning_rate": 1.164830538207845e-06,
      "loss": 0.5698,
      "step": 15503
    },
    {
      "epoch": 4.895484685822545,
      "grad_norm": 4.235742568969727,
      "learning_rate": 1.1613220124903516e-06,
      "loss": 0.6348,
      "step": 15504
    },
    {
      "epoch": 4.89580044205873,
      "grad_norm": 4.439762592315674,
      "learning_rate": 1.157813486772858e-06,
      "loss": 0.6461,
      "step": 15505
    },
    {
      "epoch": 4.896116198294917,
      "grad_norm": 3.829287528991699,
      "learning_rate": 1.1543049610553646e-06,
      "loss": 0.5773,
      "step": 15506
    },
    {
      "epoch": 4.896431954531102,
      "grad_norm": 5.359795093536377,
      "learning_rate": 1.1507964353378712e-06,
      "loss": 0.5584,
      "step": 15507
    },
    {
      "epoch": 4.896747710767288,
      "grad_norm": 4.848248481750488,
      "learning_rate": 1.1472879096203776e-06,
      "loss": 0.6569,
      "step": 15508
    },
    {
      "epoch": 4.897063467003473,
      "grad_norm": 4.027618885040283,
      "learning_rate": 1.143779383902884e-06,
      "loss": 0.6322,
      "step": 15509
    },
    {
      "epoch": 4.897379223239659,
      "grad_norm": 4.006960868835449,
      "learning_rate": 1.1402708581853906e-06,
      "loss": 0.6205,
      "step": 15510
    },
    {
      "epoch": 4.897694979475845,
      "grad_norm": 4.683032512664795,
      "learning_rate": 1.136762332467897e-06,
      "loss": 0.5764,
      "step": 15511
    },
    {
      "epoch": 4.89801073571203,
      "grad_norm": 3.5236337184906006,
      "learning_rate": 1.1332538067504036e-06,
      "loss": 0.6179,
      "step": 15512
    },
    {
      "epoch": 4.898326491948216,
      "grad_norm": 3.65679669380188,
      "learning_rate": 1.12974528103291e-06,
      "loss": 0.6827,
      "step": 15513
    },
    {
      "epoch": 4.898642248184402,
      "grad_norm": 3.2786452770233154,
      "learning_rate": 1.1262367553154166e-06,
      "loss": 0.574,
      "step": 15514
    },
    {
      "epoch": 4.898958004420587,
      "grad_norm": 4.030442237854004,
      "learning_rate": 1.122728229597923e-06,
      "loss": 0.6145,
      "step": 15515
    },
    {
      "epoch": 4.899273760656773,
      "grad_norm": 4.810882091522217,
      "learning_rate": 1.1192197038804294e-06,
      "loss": 0.6325,
      "step": 15516
    },
    {
      "epoch": 4.899589516892958,
      "grad_norm": 4.602847099304199,
      "learning_rate": 1.115711178162936e-06,
      "loss": 0.6118,
      "step": 15517
    },
    {
      "epoch": 4.899905273129145,
      "grad_norm": 4.0608811378479,
      "learning_rate": 1.1122026524454426e-06,
      "loss": 0.5556,
      "step": 15518
    },
    {
      "epoch": 4.90022102936533,
      "grad_norm": 4.145168304443359,
      "learning_rate": 1.108694126727949e-06,
      "loss": 0.6032,
      "step": 15519
    },
    {
      "epoch": 4.900536785601515,
      "grad_norm": 4.183411121368408,
      "learning_rate": 1.1051856010104556e-06,
      "loss": 0.6431,
      "step": 15520
    },
    {
      "epoch": 4.900852541837701,
      "grad_norm": 4.973640441894531,
      "learning_rate": 1.101677075292962e-06,
      "loss": 0.6233,
      "step": 15521
    },
    {
      "epoch": 4.901168298073887,
      "grad_norm": 5.02140474319458,
      "learning_rate": 1.0981685495754683e-06,
      "loss": 0.6377,
      "step": 15522
    },
    {
      "epoch": 4.901484054310073,
      "grad_norm": 4.618870258331299,
      "learning_rate": 1.094660023857975e-06,
      "loss": 0.6423,
      "step": 15523
    },
    {
      "epoch": 4.901799810546258,
      "grad_norm": 3.6151905059814453,
      "learning_rate": 1.0911514981404815e-06,
      "loss": 0.6273,
      "step": 15524
    },
    {
      "epoch": 4.902115566782444,
      "grad_norm": 3.7987496852874756,
      "learning_rate": 1.087642972422988e-06,
      "loss": 0.6059,
      "step": 15525
    },
    {
      "epoch": 4.90243132301863,
      "grad_norm": 3.3815648555755615,
      "learning_rate": 1.0841344467054945e-06,
      "loss": 0.5397,
      "step": 15526
    },
    {
      "epoch": 4.902747079254815,
      "grad_norm": 3.9253628253936768,
      "learning_rate": 1.080625920988001e-06,
      "loss": 0.6085,
      "step": 15527
    },
    {
      "epoch": 4.903062835491001,
      "grad_norm": 3.744262456893921,
      "learning_rate": 1.0771173952705073e-06,
      "loss": 0.5477,
      "step": 15528
    },
    {
      "epoch": 4.903378591727186,
      "grad_norm": 3.68202805519104,
      "learning_rate": 1.073608869553014e-06,
      "loss": 0.5842,
      "step": 15529
    },
    {
      "epoch": 4.903694347963373,
      "grad_norm": 4.728241443634033,
      "learning_rate": 1.0701003438355203e-06,
      "loss": 0.5681,
      "step": 15530
    },
    {
      "epoch": 4.904010104199558,
      "grad_norm": 3.6476218700408936,
      "learning_rate": 1.066591818118027e-06,
      "loss": 0.5054,
      "step": 15531
    },
    {
      "epoch": 4.904325860435744,
      "grad_norm": 3.7580575942993164,
      "learning_rate": 1.0630832924005333e-06,
      "loss": 0.517,
      "step": 15532
    },
    {
      "epoch": 4.904641616671929,
      "grad_norm": 3.961179256439209,
      "learning_rate": 1.0595747666830399e-06,
      "loss": 0.5181,
      "step": 15533
    },
    {
      "epoch": 4.904957372908115,
      "grad_norm": 4.421803951263428,
      "learning_rate": 1.0560662409655463e-06,
      "loss": 0.6302,
      "step": 15534
    },
    {
      "epoch": 4.905273129144301,
      "grad_norm": 3.987633466720581,
      "learning_rate": 1.0525577152480529e-06,
      "loss": 0.5368,
      "step": 15535
    },
    {
      "epoch": 4.905588885380486,
      "grad_norm": 4.460425853729248,
      "learning_rate": 1.0490491895305593e-06,
      "loss": 0.6253,
      "step": 15536
    },
    {
      "epoch": 4.905904641616672,
      "grad_norm": 3.7898895740509033,
      "learning_rate": 1.0455406638130659e-06,
      "loss": 0.5975,
      "step": 15537
    },
    {
      "epoch": 4.906220397852858,
      "grad_norm": 4.090327262878418,
      "learning_rate": 1.0420321380955723e-06,
      "loss": 0.601,
      "step": 15538
    },
    {
      "epoch": 4.906536154089043,
      "grad_norm": 3.9167027473449707,
      "learning_rate": 1.0385236123780786e-06,
      "loss": 0.5352,
      "step": 15539
    },
    {
      "epoch": 4.906851910325229,
      "grad_norm": 3.735814094543457,
      "learning_rate": 1.0350150866605852e-06,
      "loss": 0.5147,
      "step": 15540
    },
    {
      "epoch": 4.907167666561414,
      "grad_norm": 4.258238315582275,
      "learning_rate": 1.0315065609430918e-06,
      "loss": 0.637,
      "step": 15541
    },
    {
      "epoch": 4.9074834227976005,
      "grad_norm": 3.3638417720794678,
      "learning_rate": 1.0279980352255982e-06,
      "loss": 0.573,
      "step": 15542
    },
    {
      "epoch": 4.907799179033786,
      "grad_norm": 4.3110198974609375,
      "learning_rate": 1.0244895095081048e-06,
      "loss": 0.6033,
      "step": 15543
    },
    {
      "epoch": 4.908114935269971,
      "grad_norm": 4.914641380310059,
      "learning_rate": 1.0209809837906112e-06,
      "loss": 0.6615,
      "step": 15544
    },
    {
      "epoch": 4.908430691506157,
      "grad_norm": 4.230666637420654,
      "learning_rate": 1.0174724580731176e-06,
      "loss": 0.5736,
      "step": 15545
    },
    {
      "epoch": 4.908746447742343,
      "grad_norm": 3.726140022277832,
      "learning_rate": 1.0139639323556242e-06,
      "loss": 0.5872,
      "step": 15546
    },
    {
      "epoch": 4.909062203978529,
      "grad_norm": 4.588929653167725,
      "learning_rate": 1.0104554066381308e-06,
      "loss": 0.5705,
      "step": 15547
    },
    {
      "epoch": 4.909377960214714,
      "grad_norm": 4.280120372772217,
      "learning_rate": 1.0069468809206372e-06,
      "loss": 0.6172,
      "step": 15548
    },
    {
      "epoch": 4.9096937164509,
      "grad_norm": 4.406670093536377,
      "learning_rate": 1.0034383552031438e-06,
      "loss": 0.5607,
      "step": 15549
    },
    {
      "epoch": 4.9100094726870855,
      "grad_norm": 3.996821403503418,
      "learning_rate": 9.999298294856502e-07,
      "loss": 0.5734,
      "step": 15550
    },
    {
      "epoch": 4.910325228923272,
      "grad_norm": 3.552455186843872,
      "learning_rate": 9.964213037681566e-07,
      "loss": 0.5258,
      "step": 15551
    },
    {
      "epoch": 4.910640985159457,
      "grad_norm": 3.7806787490844727,
      "learning_rate": 9.929127780506632e-07,
      "loss": 0.5703,
      "step": 15552
    },
    {
      "epoch": 4.910956741395642,
      "grad_norm": 4.120969295501709,
      "learning_rate": 9.894042523331696e-07,
      "loss": 0.6104,
      "step": 15553
    },
    {
      "epoch": 4.9112724976318285,
      "grad_norm": 3.644610643386841,
      "learning_rate": 9.858957266156762e-07,
      "loss": 0.5116,
      "step": 15554
    },
    {
      "epoch": 4.911588253868014,
      "grad_norm": 3.6984400749206543,
      "learning_rate": 9.823872008981828e-07,
      "loss": 0.489,
      "step": 15555
    },
    {
      "epoch": 4.9119040101042,
      "grad_norm": 3.852180004119873,
      "learning_rate": 9.78878675180689e-07,
      "loss": 0.5512,
      "step": 15556
    },
    {
      "epoch": 4.912219766340385,
      "grad_norm": 4.757859230041504,
      "learning_rate": 9.753701494631956e-07,
      "loss": 0.6487,
      "step": 15557
    },
    {
      "epoch": 4.912535522576571,
      "grad_norm": 3.792208671569824,
      "learning_rate": 9.718616237457022e-07,
      "loss": 0.5542,
      "step": 15558
    },
    {
      "epoch": 4.912851278812757,
      "grad_norm": 3.478660821914673,
      "learning_rate": 9.683530980282086e-07,
      "loss": 0.6049,
      "step": 15559
    },
    {
      "epoch": 4.913167035048942,
      "grad_norm": 3.7399473190307617,
      "learning_rate": 9.648445723107152e-07,
      "loss": 0.5073,
      "step": 15560
    },
    {
      "epoch": 4.913482791285128,
      "grad_norm": 4.204721927642822,
      "learning_rate": 9.613360465932215e-07,
      "loss": 0.5261,
      "step": 15561
    },
    {
      "epoch": 4.9137985475213135,
      "grad_norm": 3.7136027812957764,
      "learning_rate": 9.57827520875728e-07,
      "loss": 0.5789,
      "step": 15562
    },
    {
      "epoch": 4.914114303757499,
      "grad_norm": 3.575047492980957,
      "learning_rate": 9.543189951582345e-07,
      "loss": 0.5276,
      "step": 15563
    },
    {
      "epoch": 4.914430059993685,
      "grad_norm": 4.0631327629089355,
      "learning_rate": 9.50810469440741e-07,
      "loss": 0.5577,
      "step": 15564
    },
    {
      "epoch": 4.91474581622987,
      "grad_norm": 4.093886852264404,
      "learning_rate": 9.473019437232475e-07,
      "loss": 0.6053,
      "step": 15565
    },
    {
      "epoch": 4.9150615724660565,
      "grad_norm": 3.860546827316284,
      "learning_rate": 9.437934180057541e-07,
      "loss": 0.5799,
      "step": 15566
    },
    {
      "epoch": 4.915377328702242,
      "grad_norm": 4.046962738037109,
      "learning_rate": 9.402848922882606e-07,
      "loss": 0.5456,
      "step": 15567
    },
    {
      "epoch": 4.915693084938427,
      "grad_norm": 3.6998629570007324,
      "learning_rate": 9.367763665707669e-07,
      "loss": 0.5559,
      "step": 15568
    },
    {
      "epoch": 4.916008841174613,
      "grad_norm": 4.438398361206055,
      "learning_rate": 9.332678408532735e-07,
      "loss": 0.5749,
      "step": 15569
    },
    {
      "epoch": 4.9163245974107985,
      "grad_norm": 4.55193567276001,
      "learning_rate": 9.2975931513578e-07,
      "loss": 0.6183,
      "step": 15570
    },
    {
      "epoch": 4.916640353646985,
      "grad_norm": 3.47383713722229,
      "learning_rate": 9.262507894182865e-07,
      "loss": 0.4808,
      "step": 15571
    },
    {
      "epoch": 4.91695610988317,
      "grad_norm": 4.073330879211426,
      "learning_rate": 9.22742263700793e-07,
      "loss": 0.5937,
      "step": 15572
    },
    {
      "epoch": 4.917271866119356,
      "grad_norm": 4.223527908325195,
      "learning_rate": 9.192337379832996e-07,
      "loss": 0.5497,
      "step": 15573
    },
    {
      "epoch": 4.9175876223555415,
      "grad_norm": 3.9972805976867676,
      "learning_rate": 9.157252122658059e-07,
      "loss": 0.5835,
      "step": 15574
    },
    {
      "epoch": 4.917903378591728,
      "grad_norm": 4.0320258140563965,
      "learning_rate": 9.122166865483124e-07,
      "loss": 0.5625,
      "step": 15575
    },
    {
      "epoch": 4.918219134827913,
      "grad_norm": 3.5584356784820557,
      "learning_rate": 9.08708160830819e-07,
      "loss": 0.545,
      "step": 15576
    },
    {
      "epoch": 4.918534891064098,
      "grad_norm": 4.096144199371338,
      "learning_rate": 9.051996351133255e-07,
      "loss": 0.553,
      "step": 15577
    },
    {
      "epoch": 4.918850647300284,
      "grad_norm": 4.446168899536133,
      "learning_rate": 9.01691109395832e-07,
      "loss": 0.6158,
      "step": 15578
    },
    {
      "epoch": 4.91916640353647,
      "grad_norm": 4.033273696899414,
      "learning_rate": 8.981825836783385e-07,
      "loss": 0.5192,
      "step": 15579
    },
    {
      "epoch": 4.919482159772656,
      "grad_norm": 5.495368480682373,
      "learning_rate": 8.946740579608448e-07,
      "loss": 0.6826,
      "step": 15580
    },
    {
      "epoch": 4.919797916008841,
      "grad_norm": 4.4708147048950195,
      "learning_rate": 8.911655322433513e-07,
      "loss": 0.6389,
      "step": 15581
    },
    {
      "epoch": 4.9201136722450265,
      "grad_norm": 5.071527004241943,
      "learning_rate": 8.876570065258578e-07,
      "loss": 0.7112,
      "step": 15582
    },
    {
      "epoch": 4.920429428481213,
      "grad_norm": 4.2725300788879395,
      "learning_rate": 8.841484808083644e-07,
      "loss": 0.6549,
      "step": 15583
    },
    {
      "epoch": 4.920745184717398,
      "grad_norm": 3.924741744995117,
      "learning_rate": 8.806399550908709e-07,
      "loss": 0.5692,
      "step": 15584
    },
    {
      "epoch": 4.921060940953584,
      "grad_norm": 4.371431827545166,
      "learning_rate": 8.771314293733772e-07,
      "loss": 0.5695,
      "step": 15585
    },
    {
      "epoch": 4.9213766971897694,
      "grad_norm": 3.95978045463562,
      "learning_rate": 8.736229036558838e-07,
      "loss": 0.5398,
      "step": 15586
    },
    {
      "epoch": 4.921692453425955,
      "grad_norm": 4.431191444396973,
      "learning_rate": 8.701143779383903e-07,
      "loss": 0.6634,
      "step": 15587
    },
    {
      "epoch": 4.922008209662141,
      "grad_norm": 4.075695037841797,
      "learning_rate": 8.666058522208968e-07,
      "loss": 0.5584,
      "step": 15588
    },
    {
      "epoch": 4.922323965898326,
      "grad_norm": 4.373584270477295,
      "learning_rate": 8.630973265034033e-07,
      "loss": 0.509,
      "step": 15589
    },
    {
      "epoch": 4.922639722134512,
      "grad_norm": 4.2283453941345215,
      "learning_rate": 8.595888007859099e-07,
      "loss": 0.6235,
      "step": 15590
    },
    {
      "epoch": 4.922955478370698,
      "grad_norm": 4.53938102722168,
      "learning_rate": 8.560802750684162e-07,
      "loss": 0.5792,
      "step": 15591
    },
    {
      "epoch": 4.923271234606884,
      "grad_norm": 4.440150260925293,
      "learning_rate": 8.525717493509227e-07,
      "loss": 0.5309,
      "step": 15592
    },
    {
      "epoch": 4.923586990843069,
      "grad_norm": 3.7869608402252197,
      "learning_rate": 8.490632236334293e-07,
      "loss": 0.5682,
      "step": 15593
    },
    {
      "epoch": 4.9239027470792545,
      "grad_norm": 3.226496696472168,
      "learning_rate": 8.455546979159358e-07,
      "loss": 0.5898,
      "step": 15594
    },
    {
      "epoch": 4.924218503315441,
      "grad_norm": 4.309359550476074,
      "learning_rate": 8.420461721984423e-07,
      "loss": 0.6974,
      "step": 15595
    },
    {
      "epoch": 4.924534259551626,
      "grad_norm": 3.8244073390960693,
      "learning_rate": 8.385376464809488e-07,
      "loss": 0.5166,
      "step": 15596
    },
    {
      "epoch": 4.924850015787812,
      "grad_norm": 3.5787594318389893,
      "learning_rate": 8.350291207634552e-07,
      "loss": 0.5522,
      "step": 15597
    },
    {
      "epoch": 4.925165772023997,
      "grad_norm": 4.152509689331055,
      "learning_rate": 8.315205950459617e-07,
      "loss": 0.5784,
      "step": 15598
    },
    {
      "epoch": 4.925481528260184,
      "grad_norm": 4.1078033447265625,
      "learning_rate": 8.280120693284682e-07,
      "loss": 0.6029,
      "step": 15599
    },
    {
      "epoch": 4.925797284496369,
      "grad_norm": 4.1709370613098145,
      "learning_rate": 8.245035436109748e-07,
      "loss": 0.6547,
      "step": 15600
    },
    {
      "epoch": 4.925797284496369,
      "eval_loss": 0.8346865177154541,
      "eval_runtime": 4.7669,
      "eval_samples_per_second": 522.349,
      "eval_steps_per_second": 4.405,
      "step": 15600
    },
    {
      "epoch": 4.926113040732554,
      "grad_norm": 4.1083455085754395,
      "learning_rate": 8.209950178934813e-07,
      "loss": 0.5412,
      "step": 15601
    },
    {
      "epoch": 4.92642879696874,
      "grad_norm": 3.769622564315796,
      "learning_rate": 8.174864921759877e-07,
      "loss": 0.5137,
      "step": 15602
    },
    {
      "epoch": 4.926744553204926,
      "grad_norm": 3.9844717979431152,
      "learning_rate": 8.139779664584941e-07,
      "loss": 0.4357,
      "step": 15603
    },
    {
      "epoch": 4.927060309441112,
      "grad_norm": 4.79823637008667,
      "learning_rate": 8.104694407410006e-07,
      "loss": 0.5784,
      "step": 15604
    },
    {
      "epoch": 4.927376065677297,
      "grad_norm": 3.893731117248535,
      "learning_rate": 8.069609150235071e-07,
      "loss": 0.5454,
      "step": 15605
    },
    {
      "epoch": 4.927691821913482,
      "grad_norm": 4.42996883392334,
      "learning_rate": 8.034523893060136e-07,
      "loss": 0.5616,
      "step": 15606
    },
    {
      "epoch": 4.928007578149669,
      "grad_norm": 3.808074712753296,
      "learning_rate": 7.999438635885202e-07,
      "loss": 0.5359,
      "step": 15607
    },
    {
      "epoch": 4.928323334385854,
      "grad_norm": 3.8597803115844727,
      "learning_rate": 7.964353378710267e-07,
      "loss": 0.5309,
      "step": 15608
    },
    {
      "epoch": 4.92863909062204,
      "grad_norm": 4.547677040100098,
      "learning_rate": 7.92926812153533e-07,
      "loss": 0.606,
      "step": 15609
    },
    {
      "epoch": 4.928954846858225,
      "grad_norm": 4.793464183807373,
      "learning_rate": 7.894182864360396e-07,
      "loss": 0.624,
      "step": 15610
    },
    {
      "epoch": 4.929270603094411,
      "grad_norm": 4.044666767120361,
      "learning_rate": 7.859097607185461e-07,
      "loss": 0.5869,
      "step": 15611
    },
    {
      "epoch": 4.929586359330597,
      "grad_norm": 3.9159507751464844,
      "learning_rate": 7.824012350010526e-07,
      "loss": 0.5875,
      "step": 15612
    },
    {
      "epoch": 4.929902115566782,
      "grad_norm": 4.057675838470459,
      "learning_rate": 7.788927092835591e-07,
      "loss": 0.618,
      "step": 15613
    },
    {
      "epoch": 4.930217871802968,
      "grad_norm": 4.1293559074401855,
      "learning_rate": 7.753841835660656e-07,
      "loss": 0.6263,
      "step": 15614
    },
    {
      "epoch": 4.930533628039154,
      "grad_norm": 4.221854209899902,
      "learning_rate": 7.718756578485721e-07,
      "loss": 0.7252,
      "step": 15615
    },
    {
      "epoch": 4.93084938427534,
      "grad_norm": 4.208027362823486,
      "learning_rate": 7.683671321310785e-07,
      "loss": 0.5782,
      "step": 15616
    },
    {
      "epoch": 4.931165140511525,
      "grad_norm": 3.3859167098999023,
      "learning_rate": 7.648586064135851e-07,
      "loss": 0.5407,
      "step": 15617
    },
    {
      "epoch": 4.93148089674771,
      "grad_norm": 4.296416282653809,
      "learning_rate": 7.613500806960916e-07,
      "loss": 0.646,
      "step": 15618
    },
    {
      "epoch": 4.931796652983897,
      "grad_norm": 4.043123722076416,
      "learning_rate": 7.57841554978598e-07,
      "loss": 0.6247,
      "step": 15619
    },
    {
      "epoch": 4.932112409220082,
      "grad_norm": 3.426328659057617,
      "learning_rate": 7.543330292611046e-07,
      "loss": 0.4545,
      "step": 15620
    },
    {
      "epoch": 4.932428165456268,
      "grad_norm": 3.9215569496154785,
      "learning_rate": 7.508245035436111e-07,
      "loss": 0.6031,
      "step": 15621
    },
    {
      "epoch": 4.932743921692453,
      "grad_norm": 5.567298889160156,
      "learning_rate": 7.473159778261174e-07,
      "loss": 0.6279,
      "step": 15622
    },
    {
      "epoch": 4.9330596779286395,
      "grad_norm": 5.586026191711426,
      "learning_rate": 7.438074521086239e-07,
      "loss": 0.6427,
      "step": 15623
    },
    {
      "epoch": 4.933375434164825,
      "grad_norm": 5.021617889404297,
      "learning_rate": 7.402989263911305e-07,
      "loss": 0.573,
      "step": 15624
    },
    {
      "epoch": 4.93369119040101,
      "grad_norm": 4.5124430656433105,
      "learning_rate": 7.367904006736369e-07,
      "loss": 0.5604,
      "step": 15625
    },
    {
      "epoch": 4.934006946637196,
      "grad_norm": 3.014817476272583,
      "learning_rate": 7.332818749561434e-07,
      "loss": 0.5017,
      "step": 15626
    },
    {
      "epoch": 4.934322702873382,
      "grad_norm": 3.7188382148742676,
      "learning_rate": 7.2977334923865e-07,
      "loss": 0.6426,
      "step": 15627
    },
    {
      "epoch": 4.934638459109568,
      "grad_norm": 3.460437536239624,
      "learning_rate": 7.262648235211564e-07,
      "loss": 0.6206,
      "step": 15628
    },
    {
      "epoch": 4.934954215345753,
      "grad_norm": 4.869843482971191,
      "learning_rate": 7.227562978036629e-07,
      "loss": 0.5707,
      "step": 15629
    },
    {
      "epoch": 4.935269971581938,
      "grad_norm": 4.200088024139404,
      "learning_rate": 7.192477720861694e-07,
      "loss": 0.6842,
      "step": 15630
    },
    {
      "epoch": 4.9355857278181245,
      "grad_norm": 3.2385811805725098,
      "learning_rate": 7.157392463686759e-07,
      "loss": 0.4884,
      "step": 15631
    },
    {
      "epoch": 4.93590148405431,
      "grad_norm": 4.24391508102417,
      "learning_rate": 7.122307206511824e-07,
      "loss": 0.5989,
      "step": 15632
    },
    {
      "epoch": 4.936217240290496,
      "grad_norm": 4.570409297943115,
      "learning_rate": 7.087221949336889e-07,
      "loss": 0.5994,
      "step": 15633
    },
    {
      "epoch": 4.936532996526681,
      "grad_norm": 3.9061167240142822,
      "learning_rate": 7.052136692161954e-07,
      "loss": 0.5877,
      "step": 15634
    },
    {
      "epoch": 4.936848752762867,
      "grad_norm": 3.8389358520507812,
      "learning_rate": 7.017051434987019e-07,
      "loss": 0.5504,
      "step": 15635
    },
    {
      "epoch": 4.937164508999053,
      "grad_norm": 3.894704580307007,
      "learning_rate": 6.981966177812084e-07,
      "loss": 0.5153,
      "step": 15636
    },
    {
      "epoch": 4.937480265235238,
      "grad_norm": 5.075299263000488,
      "learning_rate": 6.946880920637149e-07,
      "loss": 0.608,
      "step": 15637
    },
    {
      "epoch": 4.937796021471424,
      "grad_norm": 3.568032741546631,
      "learning_rate": 6.911795663462214e-07,
      "loss": 0.5714,
      "step": 15638
    },
    {
      "epoch": 4.93811177770761,
      "grad_norm": 3.769273519515991,
      "learning_rate": 6.876710406287279e-07,
      "loss": 0.6169,
      "step": 15639
    },
    {
      "epoch": 4.938427533943796,
      "grad_norm": 3.5070841312408447,
      "learning_rate": 6.841625149112344e-07,
      "loss": 0.6433,
      "step": 15640
    },
    {
      "epoch": 4.938743290179981,
      "grad_norm": 4.168747901916504,
      "learning_rate": 6.806539891937409e-07,
      "loss": 0.7254,
      "step": 15641
    },
    {
      "epoch": 4.939059046416166,
      "grad_norm": 3.899411916732788,
      "learning_rate": 6.771454634762473e-07,
      "loss": 0.5217,
      "step": 15642
    },
    {
      "epoch": 4.9393748026523525,
      "grad_norm": 3.5244297981262207,
      "learning_rate": 6.736369377587537e-07,
      "loss": 0.5648,
      "step": 15643
    },
    {
      "epoch": 4.939690558888538,
      "grad_norm": 3.3396384716033936,
      "learning_rate": 6.701284120412603e-07,
      "loss": 0.5011,
      "step": 15644
    },
    {
      "epoch": 4.940006315124724,
      "grad_norm": 3.6081066131591797,
      "learning_rate": 6.666198863237668e-07,
      "loss": 0.5391,
      "step": 15645
    },
    {
      "epoch": 4.940322071360909,
      "grad_norm": 5.242978572845459,
      "learning_rate": 6.631113606062732e-07,
      "loss": 0.6275,
      "step": 15646
    },
    {
      "epoch": 4.9406378275970955,
      "grad_norm": 4.391932010650635,
      "learning_rate": 6.596028348887798e-07,
      "loss": 0.5937,
      "step": 15647
    },
    {
      "epoch": 4.940953583833281,
      "grad_norm": 4.16128396987915,
      "learning_rate": 6.560943091712863e-07,
      "loss": 0.5954,
      "step": 15648
    },
    {
      "epoch": 4.941269340069466,
      "grad_norm": 3.6476917266845703,
      "learning_rate": 6.525857834537927e-07,
      "loss": 0.593,
      "step": 15649
    },
    {
      "epoch": 4.941585096305652,
      "grad_norm": 4.457814693450928,
      "learning_rate": 6.490772577362992e-07,
      "loss": 0.5549,
      "step": 15650
    },
    {
      "epoch": 4.9419008525418375,
      "grad_norm": 3.9893293380737305,
      "learning_rate": 6.455687320188058e-07,
      "loss": 0.6111,
      "step": 15651
    },
    {
      "epoch": 4.942216608778024,
      "grad_norm": 4.04583215713501,
      "learning_rate": 6.420602063013122e-07,
      "loss": 0.5307,
      "step": 15652
    },
    {
      "epoch": 4.942532365014209,
      "grad_norm": 3.6827030181884766,
      "learning_rate": 6.385516805838187e-07,
      "loss": 0.5542,
      "step": 15653
    },
    {
      "epoch": 4.942848121250394,
      "grad_norm": 4.102651119232178,
      "learning_rate": 6.350431548663253e-07,
      "loss": 0.6664,
      "step": 15654
    },
    {
      "epoch": 4.9431638774865805,
      "grad_norm": 3.6802549362182617,
      "learning_rate": 6.315346291488317e-07,
      "loss": 0.5407,
      "step": 15655
    },
    {
      "epoch": 4.943479633722766,
      "grad_norm": 3.9611752033233643,
      "learning_rate": 6.280261034313382e-07,
      "loss": 0.5692,
      "step": 15656
    },
    {
      "epoch": 4.943795389958952,
      "grad_norm": 4.49334716796875,
      "learning_rate": 6.245175777138447e-07,
      "loss": 0.6357,
      "step": 15657
    },
    {
      "epoch": 4.944111146195137,
      "grad_norm": 3.887906312942505,
      "learning_rate": 6.210090519963512e-07,
      "loss": 0.6102,
      "step": 15658
    },
    {
      "epoch": 4.9444269024313225,
      "grad_norm": 3.7063686847686768,
      "learning_rate": 6.175005262788577e-07,
      "loss": 0.5907,
      "step": 15659
    },
    {
      "epoch": 4.944742658667509,
      "grad_norm": 3.36332631111145,
      "learning_rate": 6.139920005613641e-07,
      "loss": 0.5119,
      "step": 15660
    },
    {
      "epoch": 4.945058414903694,
      "grad_norm": 4.883566856384277,
      "learning_rate": 6.104834748438707e-07,
      "loss": 0.592,
      "step": 15661
    },
    {
      "epoch": 4.94537417113988,
      "grad_norm": 3.7755649089813232,
      "learning_rate": 6.069749491263772e-07,
      "loss": 0.6017,
      "step": 15662
    },
    {
      "epoch": 4.9456899273760655,
      "grad_norm": 4.095414161682129,
      "learning_rate": 6.034664234088835e-07,
      "loss": 0.5552,
      "step": 15663
    },
    {
      "epoch": 4.946005683612252,
      "grad_norm": 5.555002212524414,
      "learning_rate": 5.999578976913901e-07,
      "loss": 0.5474,
      "step": 15664
    },
    {
      "epoch": 4.946321439848437,
      "grad_norm": 5.9655632972717285,
      "learning_rate": 5.964493719738966e-07,
      "loss": 0.6243,
      "step": 15665
    },
    {
      "epoch": 4.946637196084623,
      "grad_norm": 3.841780662536621,
      "learning_rate": 5.92940846256403e-07,
      "loss": 0.6319,
      "step": 15666
    },
    {
      "epoch": 4.9469529523208084,
      "grad_norm": 3.3795433044433594,
      "learning_rate": 5.894323205389095e-07,
      "loss": 0.5911,
      "step": 15667
    },
    {
      "epoch": 4.947268708556994,
      "grad_norm": 3.8125710487365723,
      "learning_rate": 5.859237948214161e-07,
      "loss": 0.5991,
      "step": 15668
    },
    {
      "epoch": 4.94758446479318,
      "grad_norm": 3.566707134246826,
      "learning_rate": 5.824152691039225e-07,
      "loss": 0.5656,
      "step": 15669
    },
    {
      "epoch": 4.947900221029365,
      "grad_norm": 4.0996856689453125,
      "learning_rate": 5.78906743386429e-07,
      "loss": 0.5055,
      "step": 15670
    },
    {
      "epoch": 4.948215977265551,
      "grad_norm": 4.1446051597595215,
      "learning_rate": 5.753982176689356e-07,
      "loss": 0.5888,
      "step": 15671
    },
    {
      "epoch": 4.948531733501737,
      "grad_norm": 4.848691940307617,
      "learning_rate": 5.71889691951442e-07,
      "loss": 0.5595,
      "step": 15672
    },
    {
      "epoch": 4.948847489737922,
      "grad_norm": 4.917849540710449,
      "learning_rate": 5.683811662339485e-07,
      "loss": 0.5678,
      "step": 15673
    },
    {
      "epoch": 4.949163245974108,
      "grad_norm": 4.146392345428467,
      "learning_rate": 5.64872640516455e-07,
      "loss": 0.5498,
      "step": 15674
    },
    {
      "epoch": 4.9494790022102935,
      "grad_norm": 4.135636806488037,
      "learning_rate": 5.613641147989615e-07,
      "loss": 0.6379,
      "step": 15675
    },
    {
      "epoch": 4.94979475844648,
      "grad_norm": 3.7478623390197754,
      "learning_rate": 5.57855589081468e-07,
      "loss": 0.5195,
      "step": 15676
    },
    {
      "epoch": 4.950110514682665,
      "grad_norm": 3.288081407546997,
      "learning_rate": 5.543470633639745e-07,
      "loss": 0.5811,
      "step": 15677
    },
    {
      "epoch": 4.95042627091885,
      "grad_norm": 4.217515468597412,
      "learning_rate": 5.50838537646481e-07,
      "loss": 0.6115,
      "step": 15678
    },
    {
      "epoch": 4.950742027155036,
      "grad_norm": 4.138171672821045,
      "learning_rate": 5.473300119289875e-07,
      "loss": 0.5851,
      "step": 15679
    },
    {
      "epoch": 4.951057783391222,
      "grad_norm": 4.927895545959473,
      "learning_rate": 5.43821486211494e-07,
      "loss": 0.6158,
      "step": 15680
    },
    {
      "epoch": 4.951373539627408,
      "grad_norm": 3.9044008255004883,
      "learning_rate": 5.403129604940005e-07,
      "loss": 0.6157,
      "step": 15681
    },
    {
      "epoch": 4.951689295863593,
      "grad_norm": 4.301898002624512,
      "learning_rate": 5.36804434776507e-07,
      "loss": 0.5023,
      "step": 15682
    },
    {
      "epoch": 4.9520050520997785,
      "grad_norm": 4.547629356384277,
      "learning_rate": 5.332959090590134e-07,
      "loss": 0.5308,
      "step": 15683
    },
    {
      "epoch": 4.952320808335965,
      "grad_norm": 4.870757579803467,
      "learning_rate": 5.297873833415199e-07,
      "loss": 0.7084,
      "step": 15684
    },
    {
      "epoch": 4.95263656457215,
      "grad_norm": 3.7655138969421387,
      "learning_rate": 5.262788576240264e-07,
      "loss": 0.5502,
      "step": 15685
    },
    {
      "epoch": 4.952952320808336,
      "grad_norm": 3.952242374420166,
      "learning_rate": 5.227703319065329e-07,
      "loss": 0.6722,
      "step": 15686
    },
    {
      "epoch": 4.953268077044521,
      "grad_norm": 4.426534175872803,
      "learning_rate": 5.192618061890393e-07,
      "loss": 0.6574,
      "step": 15687
    },
    {
      "epoch": 4.953583833280708,
      "grad_norm": 4.343874931335449,
      "learning_rate": 5.157532804715459e-07,
      "loss": 0.5983,
      "step": 15688
    },
    {
      "epoch": 4.953899589516893,
      "grad_norm": 4.663565635681152,
      "learning_rate": 5.122447547540524e-07,
      "loss": 0.6783,
      "step": 15689
    },
    {
      "epoch": 4.954215345753079,
      "grad_norm": 4.424209117889404,
      "learning_rate": 5.087362290365588e-07,
      "loss": 0.5864,
      "step": 15690
    },
    {
      "epoch": 4.954531101989264,
      "grad_norm": 4.48594856262207,
      "learning_rate": 5.052277033190654e-07,
      "loss": 0.5676,
      "step": 15691
    },
    {
      "epoch": 4.95484685822545,
      "grad_norm": 4.176780700683594,
      "learning_rate": 5.017191776015719e-07,
      "loss": 0.6578,
      "step": 15692
    },
    {
      "epoch": 4.955162614461636,
      "grad_norm": 3.8014962673187256,
      "learning_rate": 4.982106518840783e-07,
      "loss": 0.6117,
      "step": 15693
    },
    {
      "epoch": 4.955478370697821,
      "grad_norm": 4.699624538421631,
      "learning_rate": 4.947021261665848e-07,
      "loss": 0.6319,
      "step": 15694
    },
    {
      "epoch": 4.955794126934007,
      "grad_norm": 3.6957826614379883,
      "learning_rate": 4.911936004490914e-07,
      "loss": 0.5774,
      "step": 15695
    },
    {
      "epoch": 4.956109883170193,
      "grad_norm": 5.029584884643555,
      "learning_rate": 4.876850747315978e-07,
      "loss": 0.5742,
      "step": 15696
    },
    {
      "epoch": 4.956425639406378,
      "grad_norm": 4.7381439208984375,
      "learning_rate": 4.841765490141043e-07,
      "loss": 0.6233,
      "step": 15697
    },
    {
      "epoch": 4.956741395642564,
      "grad_norm": 3.9150500297546387,
      "learning_rate": 4.806680232966108e-07,
      "loss": 0.5813,
      "step": 15698
    },
    {
      "epoch": 4.957057151878749,
      "grad_norm": 4.1892852783203125,
      "learning_rate": 4.771594975791173e-07,
      "loss": 0.6538,
      "step": 15699
    },
    {
      "epoch": 4.957372908114936,
      "grad_norm": 3.2063822746276855,
      "learning_rate": 4.7365097186162376e-07,
      "loss": 0.5536,
      "step": 15700
    },
    {
      "epoch": 4.957688664351121,
      "grad_norm": 3.8455445766448975,
      "learning_rate": 4.701424461441303e-07,
      "loss": 0.6584,
      "step": 15701
    },
    {
      "epoch": 4.958004420587306,
      "grad_norm": 4.400425434112549,
      "learning_rate": 4.6663392042663675e-07,
      "loss": 0.612,
      "step": 15702
    },
    {
      "epoch": 4.958320176823492,
      "grad_norm": 4.031698226928711,
      "learning_rate": 4.6312539470914325e-07,
      "loss": 0.6131,
      "step": 15703
    },
    {
      "epoch": 4.958635933059678,
      "grad_norm": 4.804570198059082,
      "learning_rate": 4.596168689916498e-07,
      "loss": 0.6203,
      "step": 15704
    },
    {
      "epoch": 4.958951689295864,
      "grad_norm": 4.108963489532471,
      "learning_rate": 4.561083432741562e-07,
      "loss": 0.6698,
      "step": 15705
    },
    {
      "epoch": 4.959267445532049,
      "grad_norm": 4.266355514526367,
      "learning_rate": 4.5259981755666273e-07,
      "loss": 0.529,
      "step": 15706
    },
    {
      "epoch": 4.959583201768235,
      "grad_norm": 4.486476421356201,
      "learning_rate": 4.4909129183916923e-07,
      "loss": 0.7037,
      "step": 15707
    },
    {
      "epoch": 4.959898958004421,
      "grad_norm": 5.9999823570251465,
      "learning_rate": 4.4558276612167567e-07,
      "loss": 0.6967,
      "step": 15708
    },
    {
      "epoch": 4.960214714240606,
      "grad_norm": 5.03163480758667,
      "learning_rate": 4.420742404041822e-07,
      "loss": 0.6314,
      "step": 15709
    },
    {
      "epoch": 4.960530470476792,
      "grad_norm": 4.392025470733643,
      "learning_rate": 4.385657146866886e-07,
      "loss": 0.624,
      "step": 15710
    },
    {
      "epoch": 4.960846226712977,
      "grad_norm": 4.6912102699279785,
      "learning_rate": 4.3505718896919516e-07,
      "loss": 0.5873,
      "step": 15711
    },
    {
      "epoch": 4.9611619829491636,
      "grad_norm": 4.167947769165039,
      "learning_rate": 4.3154866325170165e-07,
      "loss": 0.6623,
      "step": 15712
    },
    {
      "epoch": 4.961477739185349,
      "grad_norm": 3.5052013397216797,
      "learning_rate": 4.280401375342081e-07,
      "loss": 0.6678,
      "step": 15713
    },
    {
      "epoch": 4.961793495421535,
      "grad_norm": 3.645237684249878,
      "learning_rate": 4.2453161181671464e-07,
      "loss": 0.5421,
      "step": 15714
    },
    {
      "epoch": 4.96210925165772,
      "grad_norm": 3.6420607566833496,
      "learning_rate": 4.2102308609922114e-07,
      "loss": 0.5563,
      "step": 15715
    },
    {
      "epoch": 4.962425007893906,
      "grad_norm": 4.195710182189941,
      "learning_rate": 4.175145603817276e-07,
      "loss": 0.5696,
      "step": 15716
    },
    {
      "epoch": 4.962740764130092,
      "grad_norm": 4.986849308013916,
      "learning_rate": 4.140060346642341e-07,
      "loss": 0.7147,
      "step": 15717
    },
    {
      "epoch": 4.963056520366277,
      "grad_norm": 3.82464337348938,
      "learning_rate": 4.104975089467406e-07,
      "loss": 0.5675,
      "step": 15718
    },
    {
      "epoch": 4.963372276602463,
      "grad_norm": 3.3857014179229736,
      "learning_rate": 4.0698898322924707e-07,
      "loss": 0.5156,
      "step": 15719
    },
    {
      "epoch": 4.963688032838649,
      "grad_norm": 4.1824421882629395,
      "learning_rate": 4.0348045751175356e-07,
      "loss": 0.7089,
      "step": 15720
    },
    {
      "epoch": 4.964003789074834,
      "grad_norm": 3.678845167160034,
      "learning_rate": 3.999719317942601e-07,
      "loss": 0.5859,
      "step": 15721
    },
    {
      "epoch": 4.96431954531102,
      "grad_norm": 4.725507736206055,
      "learning_rate": 3.964634060767665e-07,
      "loss": 0.6785,
      "step": 15722
    },
    {
      "epoch": 4.964635301547205,
      "grad_norm": 3.3101704120635986,
      "learning_rate": 3.9295488035927305e-07,
      "loss": 0.5941,
      "step": 15723
    },
    {
      "epoch": 4.9649510577833915,
      "grad_norm": 3.904279947280884,
      "learning_rate": 3.8944635464177955e-07,
      "loss": 0.5814,
      "step": 15724
    },
    {
      "epoch": 4.965266814019577,
      "grad_norm": 3.5082335472106934,
      "learning_rate": 3.8593782892428604e-07,
      "loss": 0.5355,
      "step": 15725
    },
    {
      "epoch": 4.965582570255762,
      "grad_norm": 4.470569610595703,
      "learning_rate": 3.8242930320679254e-07,
      "loss": 0.6314,
      "step": 15726
    },
    {
      "epoch": 4.965898326491948,
      "grad_norm": 4.307029724121094,
      "learning_rate": 3.78920777489299e-07,
      "loss": 0.6633,
      "step": 15727
    },
    {
      "epoch": 4.966214082728134,
      "grad_norm": 3.6378390789031982,
      "learning_rate": 3.7541225177180553e-07,
      "loss": 0.5301,
      "step": 15728
    },
    {
      "epoch": 4.96652983896432,
      "grad_norm": 4.089347839355469,
      "learning_rate": 3.7190372605431197e-07,
      "loss": 0.5503,
      "step": 15729
    },
    {
      "epoch": 4.966845595200505,
      "grad_norm": 4.060252666473389,
      "learning_rate": 3.6839520033681846e-07,
      "loss": 0.5706,
      "step": 15730
    },
    {
      "epoch": 4.967161351436691,
      "grad_norm": 3.4574368000030518,
      "learning_rate": 3.64886674619325e-07,
      "loss": 0.6427,
      "step": 15731
    },
    {
      "epoch": 4.9674771076728765,
      "grad_norm": 4.117278575897217,
      "learning_rate": 3.6137814890183146e-07,
      "loss": 0.6125,
      "step": 15732
    },
    {
      "epoch": 4.967792863909062,
      "grad_norm": 3.6085922718048096,
      "learning_rate": 3.5786962318433795e-07,
      "loss": 0.5916,
      "step": 15733
    },
    {
      "epoch": 4.968108620145248,
      "grad_norm": 4.103790760040283,
      "learning_rate": 3.5436109746684445e-07,
      "loss": 0.6123,
      "step": 15734
    },
    {
      "epoch": 4.968424376381433,
      "grad_norm": 3.7324371337890625,
      "learning_rate": 3.5085257174935094e-07,
      "loss": 0.5307,
      "step": 15735
    },
    {
      "epoch": 4.9687401326176195,
      "grad_norm": 3.3582072257995605,
      "learning_rate": 3.4734404603185744e-07,
      "loss": 0.5401,
      "step": 15736
    },
    {
      "epoch": 4.969055888853805,
      "grad_norm": 4.499183654785156,
      "learning_rate": 3.4383552031436393e-07,
      "loss": 0.6387,
      "step": 15737
    },
    {
      "epoch": 4.969371645089991,
      "grad_norm": 4.088522434234619,
      "learning_rate": 3.4032699459687043e-07,
      "loss": 0.561,
      "step": 15738
    },
    {
      "epoch": 4.969687401326176,
      "grad_norm": 4.42094612121582,
      "learning_rate": 3.3681846887937687e-07,
      "loss": 0.6677,
      "step": 15739
    },
    {
      "epoch": 4.9700031575623616,
      "grad_norm": 2.661222219467163,
      "learning_rate": 3.333099431618834e-07,
      "loss": 0.452,
      "step": 15740
    },
    {
      "epoch": 4.970318913798548,
      "grad_norm": 4.3406877517700195,
      "learning_rate": 3.298014174443899e-07,
      "loss": 0.6488,
      "step": 15741
    },
    {
      "epoch": 4.970634670034733,
      "grad_norm": 4.019743919372559,
      "learning_rate": 3.2629289172689636e-07,
      "loss": 0.6546,
      "step": 15742
    },
    {
      "epoch": 4.970950426270919,
      "grad_norm": 4.107536792755127,
      "learning_rate": 3.227843660094029e-07,
      "loss": 0.5118,
      "step": 15743
    },
    {
      "epoch": 4.9712661825071045,
      "grad_norm": 4.054830074310303,
      "learning_rate": 3.1927584029190935e-07,
      "loss": 0.5208,
      "step": 15744
    },
    {
      "epoch": 4.97158193874329,
      "grad_norm": 4.018834114074707,
      "learning_rate": 3.1576731457441584e-07,
      "loss": 0.5717,
      "step": 15745
    },
    {
      "epoch": 4.971897694979476,
      "grad_norm": 4.141958236694336,
      "learning_rate": 3.1225878885692234e-07,
      "loss": 0.6904,
      "step": 15746
    },
    {
      "epoch": 4.972213451215661,
      "grad_norm": 4.532526016235352,
      "learning_rate": 3.0875026313942883e-07,
      "loss": 0.57,
      "step": 15747
    },
    {
      "epoch": 4.9725292074518475,
      "grad_norm": 3.266641855239868,
      "learning_rate": 3.0524173742193533e-07,
      "loss": 0.559,
      "step": 15748
    },
    {
      "epoch": 4.972844963688033,
      "grad_norm": 4.597439289093018,
      "learning_rate": 3.0173321170444177e-07,
      "loss": 0.4793,
      "step": 15749
    },
    {
      "epoch": 4.973160719924218,
      "grad_norm": 4.421611785888672,
      "learning_rate": 2.982246859869483e-07,
      "loss": 0.6345,
      "step": 15750
    },
    {
      "epoch": 4.973476476160404,
      "grad_norm": 3.804598093032837,
      "learning_rate": 2.9471616026945476e-07,
      "loss": 0.5029,
      "step": 15751
    },
    {
      "epoch": 4.9737922323965895,
      "grad_norm": 3.8922531604766846,
      "learning_rate": 2.9120763455196126e-07,
      "loss": 0.5818,
      "step": 15752
    },
    {
      "epoch": 4.974107988632776,
      "grad_norm": 3.369830369949341,
      "learning_rate": 2.876991088344678e-07,
      "loss": 0.5203,
      "step": 15753
    },
    {
      "epoch": 4.974423744868961,
      "grad_norm": 4.2907843589782715,
      "learning_rate": 2.8419058311697425e-07,
      "loss": 0.501,
      "step": 15754
    },
    {
      "epoch": 4.974739501105147,
      "grad_norm": 4.505532264709473,
      "learning_rate": 2.8068205739948074e-07,
      "loss": 0.5446,
      "step": 15755
    },
    {
      "epoch": 4.9750552573413325,
      "grad_norm": 4.888215065002441,
      "learning_rate": 2.7717353168198724e-07,
      "loss": 0.5648,
      "step": 15756
    },
    {
      "epoch": 4.975371013577519,
      "grad_norm": 4.312663555145264,
      "learning_rate": 2.7366500596449373e-07,
      "loss": 0.478,
      "step": 15757
    },
    {
      "epoch": 4.975686769813704,
      "grad_norm": 4.133381366729736,
      "learning_rate": 2.7015648024700023e-07,
      "loss": 0.6857,
      "step": 15758
    },
    {
      "epoch": 4.976002526049889,
      "grad_norm": 4.528426647186279,
      "learning_rate": 2.666479545295067e-07,
      "loss": 0.5536,
      "step": 15759
    },
    {
      "epoch": 4.976318282286075,
      "grad_norm": 4.925418376922607,
      "learning_rate": 2.631394288120132e-07,
      "loss": 0.6366,
      "step": 15760
    },
    {
      "epoch": 4.976634038522261,
      "grad_norm": 3.8554532527923584,
      "learning_rate": 2.5963090309451966e-07,
      "loss": 0.5242,
      "step": 15761
    },
    {
      "epoch": 4.976949794758447,
      "grad_norm": 3.6539902687072754,
      "learning_rate": 2.561223773770262e-07,
      "loss": 0.6811,
      "step": 15762
    },
    {
      "epoch": 4.977265550994632,
      "grad_norm": 3.975219488143921,
      "learning_rate": 2.526138516595327e-07,
      "loss": 0.6295,
      "step": 15763
    },
    {
      "epoch": 4.9775813072308175,
      "grad_norm": 5.1707539558410645,
      "learning_rate": 2.4910532594203915e-07,
      "loss": 0.6807,
      "step": 15764
    },
    {
      "epoch": 4.977897063467004,
      "grad_norm": 3.721769332885742,
      "learning_rate": 2.455968002245457e-07,
      "loss": 0.5597,
      "step": 15765
    },
    {
      "epoch": 4.978212819703189,
      "grad_norm": 3.518721342086792,
      "learning_rate": 2.4208827450705214e-07,
      "loss": 0.6414,
      "step": 15766
    },
    {
      "epoch": 4.978528575939375,
      "grad_norm": 4.068751335144043,
      "learning_rate": 2.3857974878955863e-07,
      "loss": 0.6142,
      "step": 15767
    },
    {
      "epoch": 4.97884433217556,
      "grad_norm": 4.347742557525635,
      "learning_rate": 2.3507122307206516e-07,
      "loss": 0.5309,
      "step": 15768
    },
    {
      "epoch": 4.979160088411746,
      "grad_norm": 3.789450168609619,
      "learning_rate": 2.3156269735457162e-07,
      "loss": 0.5132,
      "step": 15769
    },
    {
      "epoch": 4.979475844647932,
      "grad_norm": 4.263712406158447,
      "learning_rate": 2.280541716370781e-07,
      "loss": 0.6336,
      "step": 15770
    },
    {
      "epoch": 4.979791600884117,
      "grad_norm": 3.490591526031494,
      "learning_rate": 2.2454564591958462e-07,
      "loss": 0.4604,
      "step": 15771
    },
    {
      "epoch": 4.980107357120303,
      "grad_norm": 3.336602210998535,
      "learning_rate": 2.210371202020911e-07,
      "loss": 0.586,
      "step": 15772
    },
    {
      "epoch": 4.980423113356489,
      "grad_norm": 4.0830512046813965,
      "learning_rate": 2.1752859448459758e-07,
      "loss": 0.5278,
      "step": 15773
    },
    {
      "epoch": 4.980738869592674,
      "grad_norm": 4.27377462387085,
      "learning_rate": 2.1402006876710405e-07,
      "loss": 0.6205,
      "step": 15774
    },
    {
      "epoch": 4.98105462582886,
      "grad_norm": 3.4312562942504883,
      "learning_rate": 2.1051154304961057e-07,
      "loss": 0.498,
      "step": 15775
    },
    {
      "epoch": 4.9813703820650455,
      "grad_norm": 4.275074005126953,
      "learning_rate": 2.0700301733211704e-07,
      "loss": 0.5968,
      "step": 15776
    },
    {
      "epoch": 4.981686138301232,
      "grad_norm": 4.127798080444336,
      "learning_rate": 2.0349449161462353e-07,
      "loss": 0.4857,
      "step": 15777
    },
    {
      "epoch": 4.982001894537417,
      "grad_norm": 3.0829074382781982,
      "learning_rate": 1.9998596589713006e-07,
      "loss": 0.5862,
      "step": 15778
    },
    {
      "epoch": 4.982317650773603,
      "grad_norm": 4.08339262008667,
      "learning_rate": 1.9647744017963652e-07,
      "loss": 0.5068,
      "step": 15779
    },
    {
      "epoch": 4.982633407009788,
      "grad_norm": 5.596314907073975,
      "learning_rate": 1.9296891446214302e-07,
      "loss": 0.6236,
      "step": 15780
    },
    {
      "epoch": 4.982949163245975,
      "grad_norm": 3.731860876083374,
      "learning_rate": 1.894603887446495e-07,
      "loss": 0.6517,
      "step": 15781
    },
    {
      "epoch": 4.98326491948216,
      "grad_norm": 4.484048366546631,
      "learning_rate": 1.8595186302715598e-07,
      "loss": 0.6243,
      "step": 15782
    },
    {
      "epoch": 4.983580675718345,
      "grad_norm": 3.6723520755767822,
      "learning_rate": 1.824433373096625e-07,
      "loss": 0.5267,
      "step": 15783
    },
    {
      "epoch": 4.983896431954531,
      "grad_norm": 4.2353081703186035,
      "learning_rate": 1.7893481159216898e-07,
      "loss": 0.5391,
      "step": 15784
    },
    {
      "epoch": 4.984212188190717,
      "grad_norm": 4.209038734436035,
      "learning_rate": 1.7542628587467547e-07,
      "loss": 0.5479,
      "step": 15785
    },
    {
      "epoch": 4.984527944426903,
      "grad_norm": 4.447892665863037,
      "learning_rate": 1.7191776015718197e-07,
      "loss": 0.5834,
      "step": 15786
    },
    {
      "epoch": 4.984843700663088,
      "grad_norm": 3.8201355934143066,
      "learning_rate": 1.6840923443968843e-07,
      "loss": 0.514,
      "step": 15787
    },
    {
      "epoch": 4.985159456899273,
      "grad_norm": 5.362665176391602,
      "learning_rate": 1.6490070872219496e-07,
      "loss": 0.5703,
      "step": 15788
    },
    {
      "epoch": 4.98547521313546,
      "grad_norm": 4.708890438079834,
      "learning_rate": 1.6139218300470145e-07,
      "loss": 0.5994,
      "step": 15789
    },
    {
      "epoch": 4.985790969371645,
      "grad_norm": 4.415674686431885,
      "learning_rate": 1.5788365728720792e-07,
      "loss": 0.5461,
      "step": 15790
    },
    {
      "epoch": 4.986106725607831,
      "grad_norm": 4.179316997528076,
      "learning_rate": 1.5437513156971442e-07,
      "loss": 0.5008,
      "step": 15791
    },
    {
      "epoch": 4.986422481844016,
      "grad_norm": 4.250758171081543,
      "learning_rate": 1.5086660585222089e-07,
      "loss": 0.6435,
      "step": 15792
    },
    {
      "epoch": 4.986738238080202,
      "grad_norm": 5.242772102355957,
      "learning_rate": 1.4735808013472738e-07,
      "loss": 0.6482,
      "step": 15793
    },
    {
      "epoch": 4.987053994316388,
      "grad_norm": 3.8340182304382324,
      "learning_rate": 1.438495544172339e-07,
      "loss": 0.5938,
      "step": 15794
    },
    {
      "epoch": 4.987369750552573,
      "grad_norm": 3.969860076904297,
      "learning_rate": 1.4034102869974037e-07,
      "loss": 0.6574,
      "step": 15795
    },
    {
      "epoch": 4.987685506788759,
      "grad_norm": 4.0599541664123535,
      "learning_rate": 1.3683250298224687e-07,
      "loss": 0.5232,
      "step": 15796
    },
    {
      "epoch": 4.988001263024945,
      "grad_norm": 4.0349836349487305,
      "learning_rate": 1.3332397726475336e-07,
      "loss": 0.6565,
      "step": 15797
    },
    {
      "epoch": 4.98831701926113,
      "grad_norm": 4.340076923370361,
      "learning_rate": 1.2981545154725983e-07,
      "loss": 0.7622,
      "step": 15798
    },
    {
      "epoch": 4.988632775497316,
      "grad_norm": 4.438254356384277,
      "learning_rate": 1.2630692582976635e-07,
      "loss": 0.5171,
      "step": 15799
    },
    {
      "epoch": 4.988948531733501,
      "grad_norm": 4.209817409515381,
      "learning_rate": 1.2279840011227285e-07,
      "loss": 0.6222,
      "step": 15800
    },
    {
      "epoch": 4.988948531733501,
      "eval_loss": 0.832170844078064,
      "eval_runtime": 4.628,
      "eval_samples_per_second": 538.031,
      "eval_steps_per_second": 4.538,
      "step": 15800
    },
    {
      "epoch": 4.989264287969688,
      "grad_norm": 0.6176995038986206,
      "learning_rate": 4.169648808897309e-05,
      "loss": 1.1307,
      "step": 15801
    },
    {
      "epoch": 4.989580044205873,
      "grad_norm": 0.6390830278396606,
      "learning_rate": 4.169561098831702e-05,
      "loss": 1.1195,
      "step": 15802
    },
    {
      "epoch": 4.989895800442059,
      "grad_norm": 0.5649023056030273,
      "learning_rate": 4.1694733887660946e-05,
      "loss": 1.1197,
      "step": 15803
    },
    {
      "epoch": 4.990211556678244,
      "grad_norm": 0.5718010663986206,
      "learning_rate": 4.169385678700488e-05,
      "loss": 1.0637,
      "step": 15804
    },
    {
      "epoch": 4.9905273129144305,
      "grad_norm": 0.6463292241096497,
      "learning_rate": 4.169297968634881e-05,
      "loss": 1.0501,
      "step": 15805
    },
    {
      "epoch": 4.990843069150616,
      "grad_norm": 0.7797988057136536,
      "learning_rate": 4.169210258569274e-05,
      "loss": 0.9898,
      "step": 15806
    },
    {
      "epoch": 4.991158825386801,
      "grad_norm": 0.709568440914154,
      "learning_rate": 4.1691225485036666e-05,
      "loss": 0.9666,
      "step": 15807
    },
    {
      "epoch": 4.991474581622987,
      "grad_norm": 1.0644959211349487,
      "learning_rate": 4.16903483843806e-05,
      "loss": 0.9361,
      "step": 15808
    },
    {
      "epoch": 4.991790337859173,
      "grad_norm": 1.1851418018341064,
      "learning_rate": 4.168947128372452e-05,
      "loss": 0.8909,
      "step": 15809
    },
    {
      "epoch": 4.992106094095359,
      "grad_norm": 1.4463562965393066,
      "learning_rate": 4.1688594183068455e-05,
      "loss": 0.8325,
      "step": 15810
    },
    {
      "epoch": 4.992421850331544,
      "grad_norm": 1.6282382011413574,
      "learning_rate": 4.168771708241238e-05,
      "loss": 0.8529,
      "step": 15811
    },
    {
      "epoch": 4.992737606567729,
      "grad_norm": 2.006474494934082,
      "learning_rate": 4.168683998175631e-05,
      "loss": 0.8309,
      "step": 15812
    },
    {
      "epoch": 4.9930533628039155,
      "grad_norm": 1.616233229637146,
      "learning_rate": 4.1685962881100236e-05,
      "loss": 0.7165,
      "step": 15813
    },
    {
      "epoch": 4.993369119040101,
      "grad_norm": 2.57340931892395,
      "learning_rate": 4.168508578044417e-05,
      "loss": 0.8147,
      "step": 15814
    },
    {
      "epoch": 4.993684875276287,
      "grad_norm": 2.7739834785461426,
      "learning_rate": 4.168420867978809e-05,
      "loss": 0.8114,
      "step": 15815
    },
    {
      "epoch": 4.994000631512472,
      "grad_norm": 2.3994905948638916,
      "learning_rate": 4.1683331579132024e-05,
      "loss": 0.6725,
      "step": 15816
    },
    {
      "epoch": 4.994316387748658,
      "grad_norm": 2.822805166244507,
      "learning_rate": 4.168245447847595e-05,
      "loss": 0.792,
      "step": 15817
    },
    {
      "epoch": 4.994632143984844,
      "grad_norm": 2.2244083881378174,
      "learning_rate": 4.168157737781988e-05,
      "loss": 0.7525,
      "step": 15818
    },
    {
      "epoch": 4.994947900221029,
      "grad_norm": 3.9567553997039795,
      "learning_rate": 4.1680700277163806e-05,
      "loss": 0.6984,
      "step": 15819
    },
    {
      "epoch": 4.995263656457215,
      "grad_norm": 2.9668633937835693,
      "learning_rate": 4.167982317650774e-05,
      "loss": 0.6965,
      "step": 15820
    },
    {
      "epoch": 4.995579412693401,
      "grad_norm": 2.531278610229492,
      "learning_rate": 4.167894607585167e-05,
      "loss": 0.6336,
      "step": 15821
    },
    {
      "epoch": 4.995895168929587,
      "grad_norm": 3.0592405796051025,
      "learning_rate": 4.1678068975195594e-05,
      "loss": 0.7219,
      "step": 15822
    },
    {
      "epoch": 4.996210925165772,
      "grad_norm": 3.61221981048584,
      "learning_rate": 4.1677191874539526e-05,
      "loss": 0.648,
      "step": 15823
    },
    {
      "epoch": 4.996526681401957,
      "grad_norm": 4.697492599487305,
      "learning_rate": 4.167631477388345e-05,
      "loss": 0.687,
      "step": 15824
    },
    {
      "epoch": 4.9968424376381435,
      "grad_norm": 4.225047588348389,
      "learning_rate": 4.167543767322738e-05,
      "loss": 0.8192,
      "step": 15825
    },
    {
      "epoch": 4.997158193874329,
      "grad_norm": 3.6752076148986816,
      "learning_rate": 4.167456057257131e-05,
      "loss": 0.7313,
      "step": 15826
    },
    {
      "epoch": 4.997473950110515,
      "grad_norm": 3.8658947944641113,
      "learning_rate": 4.167368347191524e-05,
      "loss": 0.6728,
      "step": 15827
    },
    {
      "epoch": 4.9977897063467,
      "grad_norm": 4.326780796051025,
      "learning_rate": 4.1672806371259164e-05,
      "loss": 0.6837,
      "step": 15828
    },
    {
      "epoch": 4.9981054625828865,
      "grad_norm": 5.751726150512695,
      "learning_rate": 4.1671929270603096e-05,
      "loss": 0.6866,
      "step": 15829
    },
    {
      "epoch": 4.998421218819072,
      "grad_norm": 4.741197109222412,
      "learning_rate": 4.167105216994702e-05,
      "loss": 0.721,
      "step": 15830
    },
    {
      "epoch": 4.998736975055257,
      "grad_norm": 4.350490093231201,
      "learning_rate": 4.167017506929096e-05,
      "loss": 0.6844,
      "step": 15831
    },
    {
      "epoch": 4.999052731291443,
      "grad_norm": 4.173506736755371,
      "learning_rate": 4.1669297968634884e-05,
      "loss": 0.7566,
      "step": 15832
    },
    {
      "epoch": 4.9993684875276285,
      "grad_norm": 3.49896502494812,
      "learning_rate": 4.1668420867978815e-05,
      "loss": 0.6337,
      "step": 15833
    },
    {
      "epoch": 4.999684243763815,
      "grad_norm": 3.900421142578125,
      "learning_rate": 4.166754376732274e-05,
      "loss": 0.6692,
      "step": 15834
    },
    {
      "epoch": 5.0,
      "grad_norm": 3.9290611743927,
      "learning_rate": 4.166666666666667e-05,
      "loss": 0.6492,
      "step": 15835
    },
    {
      "epoch": 5.000315756236185,
      "grad_norm": 3.970100164413452,
      "learning_rate": 4.16657895660106e-05,
      "loss": 0.7393,
      "step": 15836
    },
    {
      "epoch": 5.0006315124723715,
      "grad_norm": 5.025081634521484,
      "learning_rate": 4.166491246535453e-05,
      "loss": 0.672,
      "step": 15837
    },
    {
      "epoch": 5.000947268708557,
      "grad_norm": 4.250788688659668,
      "learning_rate": 4.1664035364698454e-05,
      "loss": 0.7248,
      "step": 15838
    },
    {
      "epoch": 5.001263024944743,
      "grad_norm": 3.9340336322784424,
      "learning_rate": 4.1663158264042385e-05,
      "loss": 0.651,
      "step": 15839
    },
    {
      "epoch": 5.001578781180928,
      "grad_norm": 4.98844051361084,
      "learning_rate": 4.166228116338631e-05,
      "loss": 0.6245,
      "step": 15840
    },
    {
      "epoch": 5.001894537417114,
      "grad_norm": 3.4433786869049072,
      "learning_rate": 4.166140406273024e-05,
      "loss": 0.5954,
      "step": 15841
    },
    {
      "epoch": 5.0022102936533,
      "grad_norm": 3.9336180686950684,
      "learning_rate": 4.166052696207417e-05,
      "loss": 0.6322,
      "step": 15842
    },
    {
      "epoch": 5.002526049889485,
      "grad_norm": 3.7714829444885254,
      "learning_rate": 4.16596498614181e-05,
      "loss": 0.6095,
      "step": 15843
    },
    {
      "epoch": 5.002841806125671,
      "grad_norm": 4.095308780670166,
      "learning_rate": 4.165877276076203e-05,
      "loss": 0.733,
      "step": 15844
    },
    {
      "epoch": 5.0031575623618565,
      "grad_norm": 3.330153703689575,
      "learning_rate": 4.1657895660105955e-05,
      "loss": 0.6364,
      "step": 15845
    },
    {
      "epoch": 5.003473318598043,
      "grad_norm": 4.812596321105957,
      "learning_rate": 4.165701855944989e-05,
      "loss": 0.7936,
      "step": 15846
    },
    {
      "epoch": 5.003789074834228,
      "grad_norm": 3.975982666015625,
      "learning_rate": 4.165614145879381e-05,
      "loss": 0.6406,
      "step": 15847
    },
    {
      "epoch": 5.004104831070413,
      "grad_norm": 3.1379833221435547,
      "learning_rate": 4.165526435813774e-05,
      "loss": 0.7641,
      "step": 15848
    },
    {
      "epoch": 5.004420587306599,
      "grad_norm": 4.4274444580078125,
      "learning_rate": 4.165438725748167e-05,
      "loss": 0.655,
      "step": 15849
    },
    {
      "epoch": 5.004736343542785,
      "grad_norm": 4.33266019821167,
      "learning_rate": 4.16535101568256e-05,
      "loss": 0.6941,
      "step": 15850
    },
    {
      "epoch": 5.005052099778971,
      "grad_norm": 3.515669822692871,
      "learning_rate": 4.1652633056169525e-05,
      "loss": 0.7649,
      "step": 15851
    },
    {
      "epoch": 5.005367856015156,
      "grad_norm": 4.641598224639893,
      "learning_rate": 4.1651755955513456e-05,
      "loss": 0.6368,
      "step": 15852
    },
    {
      "epoch": 5.005683612251342,
      "grad_norm": 3.6283867359161377,
      "learning_rate": 4.165087885485738e-05,
      "loss": 0.7314,
      "step": 15853
    },
    {
      "epoch": 5.005999368487528,
      "grad_norm": 4.152702331542969,
      "learning_rate": 4.165000175420131e-05,
      "loss": 0.7298,
      "step": 15854
    },
    {
      "epoch": 5.006315124723713,
      "grad_norm": 5.292435169219971,
      "learning_rate": 4.164912465354524e-05,
      "loss": 0.6252,
      "step": 15855
    },
    {
      "epoch": 5.006630880959899,
      "grad_norm": 4.206345558166504,
      "learning_rate": 4.164824755288917e-05,
      "loss": 0.6026,
      "step": 15856
    },
    {
      "epoch": 5.0069466371960845,
      "grad_norm": 3.2816731929779053,
      "learning_rate": 4.1647370452233095e-05,
      "loss": 0.6533,
      "step": 15857
    },
    {
      "epoch": 5.007262393432271,
      "grad_norm": 4.553071975708008,
      "learning_rate": 4.164649335157703e-05,
      "loss": 0.6717,
      "step": 15858
    },
    {
      "epoch": 5.007578149668456,
      "grad_norm": 3.8028810024261475,
      "learning_rate": 4.164561625092096e-05,
      "loss": 0.7218,
      "step": 15859
    },
    {
      "epoch": 5.007893905904641,
      "grad_norm": 4.1951904296875,
      "learning_rate": 4.164473915026489e-05,
      "loss": 0.6083,
      "step": 15860
    },
    {
      "epoch": 5.008209662140827,
      "grad_norm": 3.898488998413086,
      "learning_rate": 4.1643862049608814e-05,
      "loss": 0.6594,
      "step": 15861
    },
    {
      "epoch": 5.008525418377013,
      "grad_norm": 3.539797067642212,
      "learning_rate": 4.1642984948952746e-05,
      "loss": 0.6628,
      "step": 15862
    },
    {
      "epoch": 5.008841174613199,
      "grad_norm": 4.105079650878906,
      "learning_rate": 4.164210784829667e-05,
      "loss": 0.5706,
      "step": 15863
    },
    {
      "epoch": 5.009156930849384,
      "grad_norm": 5.0062432289123535,
      "learning_rate": 4.16412307476406e-05,
      "loss": 0.5838,
      "step": 15864
    },
    {
      "epoch": 5.00947268708557,
      "grad_norm": 3.9072303771972656,
      "learning_rate": 4.1640353646984534e-05,
      "loss": 0.6893,
      "step": 15865
    },
    {
      "epoch": 5.009788443321756,
      "grad_norm": 4.173308849334717,
      "learning_rate": 4.163947654632846e-05,
      "loss": 0.5638,
      "step": 15866
    },
    {
      "epoch": 5.010104199557941,
      "grad_norm": 3.3488383293151855,
      "learning_rate": 4.163859944567239e-05,
      "loss": 0.6148,
      "step": 15867
    },
    {
      "epoch": 5.010419955794127,
      "grad_norm": 3.799609422683716,
      "learning_rate": 4.1637722345016316e-05,
      "loss": 0.6759,
      "step": 15868
    },
    {
      "epoch": 5.010735712030312,
      "grad_norm": 3.456880569458008,
      "learning_rate": 4.163684524436025e-05,
      "loss": 0.6355,
      "step": 15869
    },
    {
      "epoch": 5.011051468266499,
      "grad_norm": 4.318525314331055,
      "learning_rate": 4.163596814370417e-05,
      "loss": 0.725,
      "step": 15870
    },
    {
      "epoch": 5.011367224502684,
      "grad_norm": 4.299847602844238,
      "learning_rate": 4.1635091043048104e-05,
      "loss": 0.7011,
      "step": 15871
    },
    {
      "epoch": 5.011682980738869,
      "grad_norm": 3.8795862197875977,
      "learning_rate": 4.163421394239203e-05,
      "loss": 0.5544,
      "step": 15872
    },
    {
      "epoch": 5.011998736975055,
      "grad_norm": 4.467378616333008,
      "learning_rate": 4.163333684173596e-05,
      "loss": 0.7206,
      "step": 15873
    },
    {
      "epoch": 5.012314493211241,
      "grad_norm": 3.727997064590454,
      "learning_rate": 4.1632459741079886e-05,
      "loss": 0.6733,
      "step": 15874
    },
    {
      "epoch": 5.012630249447427,
      "grad_norm": 4.552484512329102,
      "learning_rate": 4.163158264042382e-05,
      "loss": 0.7631,
      "step": 15875
    },
    {
      "epoch": 5.012946005683612,
      "grad_norm": 3.955141305923462,
      "learning_rate": 4.163070553976774e-05,
      "loss": 0.6761,
      "step": 15876
    },
    {
      "epoch": 5.013261761919798,
      "grad_norm": 3.963275194168091,
      "learning_rate": 4.1629828439111674e-05,
      "loss": 0.5704,
      "step": 15877
    },
    {
      "epoch": 5.013577518155984,
      "grad_norm": 3.31327748298645,
      "learning_rate": 4.16289513384556e-05,
      "loss": 0.6621,
      "step": 15878
    },
    {
      "epoch": 5.013893274392169,
      "grad_norm": 3.6611053943634033,
      "learning_rate": 4.162807423779953e-05,
      "loss": 0.6477,
      "step": 15879
    },
    {
      "epoch": 5.014209030628355,
      "grad_norm": 3.4703547954559326,
      "learning_rate": 4.1627197137143455e-05,
      "loss": 0.6552,
      "step": 15880
    },
    {
      "epoch": 5.01452478686454,
      "grad_norm": 3.7211146354675293,
      "learning_rate": 4.162632003648739e-05,
      "loss": 0.6429,
      "step": 15881
    },
    {
      "epoch": 5.014840543100727,
      "grad_norm": 3.8661677837371826,
      "learning_rate": 4.162544293583132e-05,
      "loss": 0.6515,
      "step": 15882
    },
    {
      "epoch": 5.015156299336912,
      "grad_norm": 4.361551761627197,
      "learning_rate": 4.1624565835175244e-05,
      "loss": 0.6791,
      "step": 15883
    },
    {
      "epoch": 5.015472055573097,
      "grad_norm": 5.451143264770508,
      "learning_rate": 4.1623688734519175e-05,
      "loss": 0.6335,
      "step": 15884
    },
    {
      "epoch": 5.015787811809283,
      "grad_norm": 4.511934280395508,
      "learning_rate": 4.162281163386311e-05,
      "loss": 0.6563,
      "step": 15885
    },
    {
      "epoch": 5.016103568045469,
      "grad_norm": 3.6527621746063232,
      "learning_rate": 4.162193453320704e-05,
      "loss": 0.5879,
      "step": 15886
    },
    {
      "epoch": 5.016419324281655,
      "grad_norm": 4.765134334564209,
      "learning_rate": 4.1621057432550964e-05,
      "loss": 0.638,
      "step": 15887
    },
    {
      "epoch": 5.01673508051784,
      "grad_norm": 3.958075761795044,
      "learning_rate": 4.1620180331894895e-05,
      "loss": 0.6093,
      "step": 15888
    },
    {
      "epoch": 5.017050836754026,
      "grad_norm": 3.172123908996582,
      "learning_rate": 4.161930323123882e-05,
      "loss": 0.559,
      "step": 15889
    },
    {
      "epoch": 5.017366592990212,
      "grad_norm": 3.6740965843200684,
      "learning_rate": 4.161842613058275e-05,
      "loss": 0.6309,
      "step": 15890
    },
    {
      "epoch": 5.017682349226397,
      "grad_norm": 4.362498760223389,
      "learning_rate": 4.161754902992668e-05,
      "loss": 0.5407,
      "step": 15891
    },
    {
      "epoch": 5.017998105462583,
      "grad_norm": 3.75304913520813,
      "learning_rate": 4.161667192927061e-05,
      "loss": 0.6464,
      "step": 15892
    },
    {
      "epoch": 5.018313861698768,
      "grad_norm": 3.7987287044525146,
      "learning_rate": 4.1615794828614533e-05,
      "loss": 0.6533,
      "step": 15893
    },
    {
      "epoch": 5.0186296179349545,
      "grad_norm": 3.988351821899414,
      "learning_rate": 4.1614917727958465e-05,
      "loss": 0.7004,
      "step": 15894
    },
    {
      "epoch": 5.01894537417114,
      "grad_norm": 4.7031755447387695,
      "learning_rate": 4.161404062730239e-05,
      "loss": 0.5978,
      "step": 15895
    },
    {
      "epoch": 5.019261130407325,
      "grad_norm": 3.9701449871063232,
      "learning_rate": 4.161316352664632e-05,
      "loss": 0.7072,
      "step": 15896
    },
    {
      "epoch": 5.019576886643511,
      "grad_norm": 4.6607584953308105,
      "learning_rate": 4.161228642599025e-05,
      "loss": 0.7139,
      "step": 15897
    },
    {
      "epoch": 5.019892642879697,
      "grad_norm": 4.2695088386535645,
      "learning_rate": 4.161140932533418e-05,
      "loss": 0.6452,
      "step": 15898
    },
    {
      "epoch": 5.020208399115883,
      "grad_norm": 4.363633632659912,
      "learning_rate": 4.16105322246781e-05,
      "loss": 0.6243,
      "step": 15899
    },
    {
      "epoch": 5.020524155352068,
      "grad_norm": 5.276504993438721,
      "learning_rate": 4.1609655124022035e-05,
      "loss": 0.6032,
      "step": 15900
    },
    {
      "epoch": 5.020839911588254,
      "grad_norm": 5.141308784484863,
      "learning_rate": 4.160877802336596e-05,
      "loss": 0.617,
      "step": 15901
    },
    {
      "epoch": 5.02115566782444,
      "grad_norm": 3.3898682594299316,
      "learning_rate": 4.160790092270989e-05,
      "loss": 0.5598,
      "step": 15902
    },
    {
      "epoch": 5.021471424060625,
      "grad_norm": 4.834097385406494,
      "learning_rate": 4.160702382205382e-05,
      "loss": 0.66,
      "step": 15903
    },
    {
      "epoch": 5.021787180296811,
      "grad_norm": 3.703582763671875,
      "learning_rate": 4.160614672139775e-05,
      "loss": 0.4775,
      "step": 15904
    },
    {
      "epoch": 5.022102936532996,
      "grad_norm": 5.509788990020752,
      "learning_rate": 4.160526962074168e-05,
      "loss": 0.6889,
      "step": 15905
    },
    {
      "epoch": 5.0224186927691825,
      "grad_norm": 5.095499038696289,
      "learning_rate": 4.1604392520085605e-05,
      "loss": 0.6438,
      "step": 15906
    },
    {
      "epoch": 5.022734449005368,
      "grad_norm": 5.616631507873535,
      "learning_rate": 4.1603515419429536e-05,
      "loss": 0.6328,
      "step": 15907
    },
    {
      "epoch": 5.023050205241553,
      "grad_norm": 5.95389461517334,
      "learning_rate": 4.160263831877346e-05,
      "loss": 0.6287,
      "step": 15908
    },
    {
      "epoch": 5.023365961477739,
      "grad_norm": 4.683579921722412,
      "learning_rate": 4.160176121811739e-05,
      "loss": 0.7021,
      "step": 15909
    },
    {
      "epoch": 5.023681717713925,
      "grad_norm": 6.688013553619385,
      "learning_rate": 4.160088411746132e-05,
      "loss": 0.7062,
      "step": 15910
    },
    {
      "epoch": 5.023997473950111,
      "grad_norm": 4.54224967956543,
      "learning_rate": 4.160000701680525e-05,
      "loss": 0.6234,
      "step": 15911
    },
    {
      "epoch": 5.024313230186296,
      "grad_norm": 6.05373477935791,
      "learning_rate": 4.1599129916149174e-05,
      "loss": 0.6493,
      "step": 15912
    },
    {
      "epoch": 5.024628986422482,
      "grad_norm": 4.228238105773926,
      "learning_rate": 4.159825281549311e-05,
      "loss": 0.6411,
      "step": 15913
    },
    {
      "epoch": 5.0249447426586675,
      "grad_norm": 3.9230833053588867,
      "learning_rate": 4.159737571483704e-05,
      "loss": 0.6175,
      "step": 15914
    },
    {
      "epoch": 5.025260498894853,
      "grad_norm": 3.7227792739868164,
      "learning_rate": 4.159649861418097e-05,
      "loss": 0.6512,
      "step": 15915
    },
    {
      "epoch": 5.025576255131039,
      "grad_norm": 3.302989959716797,
      "learning_rate": 4.1595621513524894e-05,
      "loss": 0.6414,
      "step": 15916
    },
    {
      "epoch": 5.025892011367224,
      "grad_norm": 4.0493645668029785,
      "learning_rate": 4.1594744412868826e-05,
      "loss": 0.716,
      "step": 15917
    },
    {
      "epoch": 5.0262077676034105,
      "grad_norm": 3.497485876083374,
      "learning_rate": 4.159386731221275e-05,
      "loss": 0.6538,
      "step": 15918
    },
    {
      "epoch": 5.026523523839596,
      "grad_norm": 3.8327929973602295,
      "learning_rate": 4.159299021155668e-05,
      "loss": 0.6943,
      "step": 15919
    },
    {
      "epoch": 5.026839280075781,
      "grad_norm": 4.773670196533203,
      "learning_rate": 4.159211311090061e-05,
      "loss": 0.6523,
      "step": 15920
    },
    {
      "epoch": 5.027155036311967,
      "grad_norm": 4.139448165893555,
      "learning_rate": 4.159123601024454e-05,
      "loss": 0.6278,
      "step": 15921
    },
    {
      "epoch": 5.0274707925481525,
      "grad_norm": 4.157027721405029,
      "learning_rate": 4.1590358909588464e-05,
      "loss": 0.651,
      "step": 15922
    },
    {
      "epoch": 5.027786548784339,
      "grad_norm": 4.72458553314209,
      "learning_rate": 4.1589481808932396e-05,
      "loss": 0.6955,
      "step": 15923
    },
    {
      "epoch": 5.028102305020524,
      "grad_norm": 4.456912994384766,
      "learning_rate": 4.158860470827633e-05,
      "loss": 0.5965,
      "step": 15924
    },
    {
      "epoch": 5.02841806125671,
      "grad_norm": 3.535371780395508,
      "learning_rate": 4.158772760762025e-05,
      "loss": 0.6407,
      "step": 15925
    },
    {
      "epoch": 5.0287338174928955,
      "grad_norm": 3.60322642326355,
      "learning_rate": 4.1586850506964184e-05,
      "loss": 0.6434,
      "step": 15926
    },
    {
      "epoch": 5.029049573729081,
      "grad_norm": 3.8476662635803223,
      "learning_rate": 4.158597340630811e-05,
      "loss": 0.6348,
      "step": 15927
    },
    {
      "epoch": 5.029365329965267,
      "grad_norm": 3.3100664615631104,
      "learning_rate": 4.158509630565204e-05,
      "loss": 0.6038,
      "step": 15928
    },
    {
      "epoch": 5.029681086201452,
      "grad_norm": 3.8816025257110596,
      "learning_rate": 4.1584219204995966e-05,
      "loss": 0.7033,
      "step": 15929
    },
    {
      "epoch": 5.0299968424376384,
      "grad_norm": 3.759030818939209,
      "learning_rate": 4.15833421043399e-05,
      "loss": 0.6104,
      "step": 15930
    },
    {
      "epoch": 5.030312598673824,
      "grad_norm": 4.537138938903809,
      "learning_rate": 4.158246500368382e-05,
      "loss": 0.6948,
      "step": 15931
    },
    {
      "epoch": 5.030628354910009,
      "grad_norm": 3.941957473754883,
      "learning_rate": 4.1581587903027754e-05,
      "loss": 0.637,
      "step": 15932
    },
    {
      "epoch": 5.030944111146195,
      "grad_norm": 4.093106746673584,
      "learning_rate": 4.158071080237168e-05,
      "loss": 0.6604,
      "step": 15933
    },
    {
      "epoch": 5.0312598673823805,
      "grad_norm": 4.000036239624023,
      "learning_rate": 4.157983370171561e-05,
      "loss": 0.6332,
      "step": 15934
    },
    {
      "epoch": 5.031575623618567,
      "grad_norm": 3.182945728302002,
      "learning_rate": 4.1578956601059535e-05,
      "loss": 0.543,
      "step": 15935
    },
    {
      "epoch": 5.031891379854752,
      "grad_norm": 4.9240546226501465,
      "learning_rate": 4.157807950040347e-05,
      "loss": 0.7226,
      "step": 15936
    },
    {
      "epoch": 5.032207136090938,
      "grad_norm": 3.7360477447509766,
      "learning_rate": 4.157720239974739e-05,
      "loss": 0.5595,
      "step": 15937
    },
    {
      "epoch": 5.0325228923271235,
      "grad_norm": 5.282031059265137,
      "learning_rate": 4.1576325299091324e-05,
      "loss": 0.6171,
      "step": 15938
    },
    {
      "epoch": 5.032838648563309,
      "grad_norm": 3.617358446121216,
      "learning_rate": 4.157544819843525e-05,
      "loss": 0.595,
      "step": 15939
    },
    {
      "epoch": 5.033154404799495,
      "grad_norm": 3.9066386222839355,
      "learning_rate": 4.157457109777919e-05,
      "loss": 0.646,
      "step": 15940
    },
    {
      "epoch": 5.03347016103568,
      "grad_norm": 3.5802581310272217,
      "learning_rate": 4.157369399712311e-05,
      "loss": 0.6471,
      "step": 15941
    },
    {
      "epoch": 5.033785917271866,
      "grad_norm": 3.5186407566070557,
      "learning_rate": 4.1572816896467044e-05,
      "loss": 0.5903,
      "step": 15942
    },
    {
      "epoch": 5.034101673508052,
      "grad_norm": 3.9921584129333496,
      "learning_rate": 4.157193979581097e-05,
      "loss": 0.6046,
      "step": 15943
    },
    {
      "epoch": 5.034417429744238,
      "grad_norm": 4.900243282318115,
      "learning_rate": 4.15710626951549e-05,
      "loss": 0.5806,
      "step": 15944
    },
    {
      "epoch": 5.034733185980423,
      "grad_norm": 8.121490478515625,
      "learning_rate": 4.1570185594498825e-05,
      "loss": 0.5519,
      "step": 15945
    },
    {
      "epoch": 5.0350489422166085,
      "grad_norm": 3.6144397258758545,
      "learning_rate": 4.156930849384276e-05,
      "loss": 0.5998,
      "step": 15946
    },
    {
      "epoch": 5.035364698452795,
      "grad_norm": 4.588873386383057,
      "learning_rate": 4.156843139318669e-05,
      "loss": 0.6804,
      "step": 15947
    },
    {
      "epoch": 5.03568045468898,
      "grad_norm": 4.939698696136475,
      "learning_rate": 4.156755429253061e-05,
      "loss": 0.626,
      "step": 15948
    },
    {
      "epoch": 5.035996210925166,
      "grad_norm": 4.7021002769470215,
      "learning_rate": 4.1566677191874545e-05,
      "loss": 0.7278,
      "step": 15949
    },
    {
      "epoch": 5.036311967161351,
      "grad_norm": 4.225160598754883,
      "learning_rate": 4.156580009121847e-05,
      "loss": 0.6408,
      "step": 15950
    },
    {
      "epoch": 5.036627723397537,
      "grad_norm": 4.888461112976074,
      "learning_rate": 4.15649229905624e-05,
      "loss": 0.6832,
      "step": 15951
    },
    {
      "epoch": 5.036943479633723,
      "grad_norm": 4.044601917266846,
      "learning_rate": 4.1564045889906327e-05,
      "loss": 0.6287,
      "step": 15952
    },
    {
      "epoch": 5.037259235869908,
      "grad_norm": 4.503928184509277,
      "learning_rate": 4.156316878925026e-05,
      "loss": 0.6417,
      "step": 15953
    },
    {
      "epoch": 5.037574992106094,
      "grad_norm": 5.416561603546143,
      "learning_rate": 4.156229168859418e-05,
      "loss": 0.6428,
      "step": 15954
    },
    {
      "epoch": 5.03789074834228,
      "grad_norm": 3.843315362930298,
      "learning_rate": 4.1561414587938115e-05,
      "loss": 0.6903,
      "step": 15955
    },
    {
      "epoch": 5.038206504578466,
      "grad_norm": 4.598977565765381,
      "learning_rate": 4.156053748728204e-05,
      "loss": 0.6934,
      "step": 15956
    },
    {
      "epoch": 5.038522260814651,
      "grad_norm": 4.663852691650391,
      "learning_rate": 4.155966038662597e-05,
      "loss": 0.6141,
      "step": 15957
    },
    {
      "epoch": 5.0388380170508364,
      "grad_norm": 4.043462753295898,
      "learning_rate": 4.1558783285969896e-05,
      "loss": 0.6076,
      "step": 15958
    },
    {
      "epoch": 5.039153773287023,
      "grad_norm": 3.918722629547119,
      "learning_rate": 4.155790618531383e-05,
      "loss": 0.6013,
      "step": 15959
    },
    {
      "epoch": 5.039469529523208,
      "grad_norm": 4.289204120635986,
      "learning_rate": 4.155702908465775e-05,
      "loss": 0.7086,
      "step": 15960
    },
    {
      "epoch": 5.039785285759394,
      "grad_norm": 4.824764728546143,
      "learning_rate": 4.1556151984001685e-05,
      "loss": 0.6702,
      "step": 15961
    },
    {
      "epoch": 5.040101041995579,
      "grad_norm": 3.799672842025757,
      "learning_rate": 4.1555274883345616e-05,
      "loss": 0.6012,
      "step": 15962
    },
    {
      "epoch": 5.040416798231765,
      "grad_norm": 5.684019088745117,
      "learning_rate": 4.155439778268954e-05,
      "loss": 0.7193,
      "step": 15963
    },
    {
      "epoch": 5.040732554467951,
      "grad_norm": 4.039697170257568,
      "learning_rate": 4.155352068203347e-05,
      "loss": 0.6263,
      "step": 15964
    },
    {
      "epoch": 5.041048310704136,
      "grad_norm": 6.165902614593506,
      "learning_rate": 4.15526435813774e-05,
      "loss": 0.6653,
      "step": 15965
    },
    {
      "epoch": 5.041364066940322,
      "grad_norm": 3.93593168258667,
      "learning_rate": 4.155176648072133e-05,
      "loss": 0.5499,
      "step": 15966
    },
    {
      "epoch": 5.041679823176508,
      "grad_norm": 6.0378875732421875,
      "learning_rate": 4.155088938006526e-05,
      "loss": 0.8102,
      "step": 15967
    },
    {
      "epoch": 5.041995579412694,
      "grad_norm": 4.365668296813965,
      "learning_rate": 4.155001227940919e-05,
      "loss": 0.7314,
      "step": 15968
    },
    {
      "epoch": 5.042311335648879,
      "grad_norm": 4.651782035827637,
      "learning_rate": 4.154913517875312e-05,
      "loss": 0.6129,
      "step": 15969
    },
    {
      "epoch": 5.042627091885064,
      "grad_norm": 5.753546714782715,
      "learning_rate": 4.154825807809705e-05,
      "loss": 0.6067,
      "step": 15970
    },
    {
      "epoch": 5.042942848121251,
      "grad_norm": 4.729101657867432,
      "learning_rate": 4.1547380977440974e-05,
      "loss": 0.6859,
      "step": 15971
    },
    {
      "epoch": 5.043258604357436,
      "grad_norm": 4.800994873046875,
      "learning_rate": 4.1546503876784906e-05,
      "loss": 0.6324,
      "step": 15972
    },
    {
      "epoch": 5.043574360593622,
      "grad_norm": 4.655526161193848,
      "learning_rate": 4.154562677612883e-05,
      "loss": 0.6203,
      "step": 15973
    },
    {
      "epoch": 5.043890116829807,
      "grad_norm": 4.395815372467041,
      "learning_rate": 4.154474967547276e-05,
      "loss": 0.6649,
      "step": 15974
    },
    {
      "epoch": 5.044205873065993,
      "grad_norm": 4.30651330947876,
      "learning_rate": 4.154387257481669e-05,
      "loss": 0.7243,
      "step": 15975
    },
    {
      "epoch": 5.044521629302179,
      "grad_norm": 5.43993616104126,
      "learning_rate": 4.154299547416062e-05,
      "loss": 0.6262,
      "step": 15976
    },
    {
      "epoch": 5.044837385538364,
      "grad_norm": 3.8567028045654297,
      "learning_rate": 4.1542118373504544e-05,
      "loss": 0.5481,
      "step": 15977
    },
    {
      "epoch": 5.04515314177455,
      "grad_norm": 4.368499755859375,
      "learning_rate": 4.1541241272848476e-05,
      "loss": 0.6239,
      "step": 15978
    },
    {
      "epoch": 5.045468898010736,
      "grad_norm": 5.101990222930908,
      "learning_rate": 4.15403641721924e-05,
      "loss": 0.6313,
      "step": 15979
    },
    {
      "epoch": 5.045784654246922,
      "grad_norm": 4.612243175506592,
      "learning_rate": 4.153948707153633e-05,
      "loss": 0.6643,
      "step": 15980
    },
    {
      "epoch": 5.046100410483107,
      "grad_norm": 4.5353851318359375,
      "learning_rate": 4.153860997088026e-05,
      "loss": 0.6508,
      "step": 15981
    },
    {
      "epoch": 5.046416166719292,
      "grad_norm": 4.597054481506348,
      "learning_rate": 4.153773287022419e-05,
      "loss": 0.6362,
      "step": 15982
    },
    {
      "epoch": 5.046731922955479,
      "grad_norm": 5.5124430656433105,
      "learning_rate": 4.1536855769568114e-05,
      "loss": 0.6676,
      "step": 15983
    },
    {
      "epoch": 5.047047679191664,
      "grad_norm": 3.1488866806030273,
      "learning_rate": 4.1535978668912045e-05,
      "loss": 0.6766,
      "step": 15984
    },
    {
      "epoch": 5.04736343542785,
      "grad_norm": 3.386162042617798,
      "learning_rate": 4.153510156825598e-05,
      "loss": 0.4995,
      "step": 15985
    },
    {
      "epoch": 5.047679191664035,
      "grad_norm": 3.982584238052368,
      "learning_rate": 4.15342244675999e-05,
      "loss": 0.6816,
      "step": 15986
    },
    {
      "epoch": 5.047994947900221,
      "grad_norm": 4.569609642028809,
      "learning_rate": 4.1533347366943834e-05,
      "loss": 0.5363,
      "step": 15987
    },
    {
      "epoch": 5.048310704136407,
      "grad_norm": 3.6500728130340576,
      "learning_rate": 4.153247026628776e-05,
      "loss": 0.7083,
      "step": 15988
    },
    {
      "epoch": 5.048626460372592,
      "grad_norm": 4.581781387329102,
      "learning_rate": 4.153159316563169e-05,
      "loss": 0.6867,
      "step": 15989
    },
    {
      "epoch": 5.048942216608778,
      "grad_norm": 4.345407962799072,
      "learning_rate": 4.1530716064975615e-05,
      "loss": 0.6444,
      "step": 15990
    },
    {
      "epoch": 5.049257972844964,
      "grad_norm": 5.060337066650391,
      "learning_rate": 4.152983896431955e-05,
      "loss": 0.6045,
      "step": 15991
    },
    {
      "epoch": 5.04957372908115,
      "grad_norm": 3.542840003967285,
      "learning_rate": 4.152896186366347e-05,
      "loss": 0.6234,
      "step": 15992
    },
    {
      "epoch": 5.049889485317335,
      "grad_norm": 4.987048149108887,
      "learning_rate": 4.1528084763007404e-05,
      "loss": 0.6884,
      "step": 15993
    },
    {
      "epoch": 5.05020524155352,
      "grad_norm": 4.053425312042236,
      "learning_rate": 4.1527207662351335e-05,
      "loss": 0.5989,
      "step": 15994
    },
    {
      "epoch": 5.0505209977897065,
      "grad_norm": 4.15158748626709,
      "learning_rate": 4.152633056169527e-05,
      "loss": 0.6015,
      "step": 15995
    },
    {
      "epoch": 5.050836754025892,
      "grad_norm": 3.6278235912323,
      "learning_rate": 4.152545346103919e-05,
      "loss": 0.6535,
      "step": 15996
    },
    {
      "epoch": 5.051152510262078,
      "grad_norm": 4.257264137268066,
      "learning_rate": 4.1524576360383123e-05,
      "loss": 0.6129,
      "step": 15997
    },
    {
      "epoch": 5.051468266498263,
      "grad_norm": 4.351459980010986,
      "learning_rate": 4.152369925972705e-05,
      "loss": 0.6189,
      "step": 15998
    },
    {
      "epoch": 5.051784022734449,
      "grad_norm": 4.075433731079102,
      "learning_rate": 4.152282215907098e-05,
      "loss": 0.6553,
      "step": 15999
    },
    {
      "epoch": 5.052099778970635,
      "grad_norm": 4.376401424407959,
      "learning_rate": 4.1521945058414905e-05,
      "loss": 0.5703,
      "step": 16000
    },
    {
      "epoch": 5.052099778970635,
      "eval_loss": 0.8470284342765808,
      "eval_runtime": 4.6267,
      "eval_samples_per_second": 538.179,
      "eval_steps_per_second": 4.539,
      "step": 16000
    },
    {
      "epoch": 5.05241553520682,
      "grad_norm": 4.959821701049805,
      "learning_rate": 4.152106795775884e-05,
      "loss": 0.7306,
      "step": 16001
    },
    {
      "epoch": 5.052731291443006,
      "grad_norm": 3.690535068511963,
      "learning_rate": 4.152019085710276e-05,
      "loss": 0.5775,
      "step": 16002
    },
    {
      "epoch": 5.0530470476791916,
      "grad_norm": 5.40532922744751,
      "learning_rate": 4.151931375644669e-05,
      "loss": 0.6288,
      "step": 16003
    },
    {
      "epoch": 5.053362803915378,
      "grad_norm": 3.7933733463287354,
      "learning_rate": 4.151843665579062e-05,
      "loss": 0.6148,
      "step": 16004
    },
    {
      "epoch": 5.053678560151563,
      "grad_norm": 4.675110340118408,
      "learning_rate": 4.151755955513455e-05,
      "loss": 0.6581,
      "step": 16005
    },
    {
      "epoch": 5.053994316387748,
      "grad_norm": 4.547968864440918,
      "learning_rate": 4.151668245447848e-05,
      "loss": 0.6782,
      "step": 16006
    },
    {
      "epoch": 5.0543100726239345,
      "grad_norm": 4.672814846038818,
      "learning_rate": 4.1515805353822406e-05,
      "loss": 0.7281,
      "step": 16007
    },
    {
      "epoch": 5.05462582886012,
      "grad_norm": 3.9461727142333984,
      "learning_rate": 4.151492825316634e-05,
      "loss": 0.5736,
      "step": 16008
    },
    {
      "epoch": 5.054941585096306,
      "grad_norm": 4.327916622161865,
      "learning_rate": 4.151405115251026e-05,
      "loss": 0.6948,
      "step": 16009
    },
    {
      "epoch": 5.055257341332491,
      "grad_norm": 4.850953102111816,
      "learning_rate": 4.1513174051854195e-05,
      "loss": 0.6986,
      "step": 16010
    },
    {
      "epoch": 5.055573097568677,
      "grad_norm": 5.572231292724609,
      "learning_rate": 4.151229695119812e-05,
      "loss": 0.6532,
      "step": 16011
    },
    {
      "epoch": 5.055888853804863,
      "grad_norm": 3.930753231048584,
      "learning_rate": 4.151141985054205e-05,
      "loss": 0.652,
      "step": 16012
    },
    {
      "epoch": 5.056204610041048,
      "grad_norm": 4.059998512268066,
      "learning_rate": 4.1510542749885976e-05,
      "loss": 0.5963,
      "step": 16013
    },
    {
      "epoch": 5.056520366277234,
      "grad_norm": 3.8468048572540283,
      "learning_rate": 4.150966564922991e-05,
      "loss": 0.605,
      "step": 16014
    },
    {
      "epoch": 5.0568361225134195,
      "grad_norm": 5.740211009979248,
      "learning_rate": 4.150878854857383e-05,
      "loss": 0.7218,
      "step": 16015
    },
    {
      "epoch": 5.057151878749606,
      "grad_norm": 4.753669738769531,
      "learning_rate": 4.1507911447917764e-05,
      "loss": 0.6464,
      "step": 16016
    },
    {
      "epoch": 5.057467634985791,
      "grad_norm": 4.176849365234375,
      "learning_rate": 4.150703434726169e-05,
      "loss": 0.6925,
      "step": 16017
    },
    {
      "epoch": 5.057783391221976,
      "grad_norm": 3.7367312908172607,
      "learning_rate": 4.150615724660562e-05,
      "loss": 0.5996,
      "step": 16018
    },
    {
      "epoch": 5.0580991474581625,
      "grad_norm": 3.723249673843384,
      "learning_rate": 4.1505280145949546e-05,
      "loss": 0.5935,
      "step": 16019
    },
    {
      "epoch": 5.058414903694348,
      "grad_norm": 5.38348913192749,
      "learning_rate": 4.150440304529348e-05,
      "loss": 0.6801,
      "step": 16020
    },
    {
      "epoch": 5.058730659930534,
      "grad_norm": 4.706558704376221,
      "learning_rate": 4.15035259446374e-05,
      "loss": 0.7,
      "step": 16021
    },
    {
      "epoch": 5.059046416166719,
      "grad_norm": 3.994995355606079,
      "learning_rate": 4.150264884398134e-05,
      "loss": 0.631,
      "step": 16022
    },
    {
      "epoch": 5.0593621724029045,
      "grad_norm": 4.0811767578125,
      "learning_rate": 4.1501771743325266e-05,
      "loss": 0.6427,
      "step": 16023
    },
    {
      "epoch": 5.059677928639091,
      "grad_norm": 4.331816673278809,
      "learning_rate": 4.15008946426692e-05,
      "loss": 0.667,
      "step": 16024
    },
    {
      "epoch": 5.059993684875276,
      "grad_norm": 3.786473035812378,
      "learning_rate": 4.150001754201312e-05,
      "loss": 0.6433,
      "step": 16025
    },
    {
      "epoch": 5.060309441111462,
      "grad_norm": 3.988420009613037,
      "learning_rate": 4.1499140441357054e-05,
      "loss": 0.6324,
      "step": 16026
    },
    {
      "epoch": 5.0606251973476475,
      "grad_norm": 4.084868431091309,
      "learning_rate": 4.149826334070098e-05,
      "loss": 0.5704,
      "step": 16027
    },
    {
      "epoch": 5.060940953583834,
      "grad_norm": 4.558069705963135,
      "learning_rate": 4.149738624004491e-05,
      "loss": 0.6366,
      "step": 16028
    },
    {
      "epoch": 5.061256709820019,
      "grad_norm": 3.930406093597412,
      "learning_rate": 4.149650913938884e-05,
      "loss": 0.7066,
      "step": 16029
    },
    {
      "epoch": 5.061572466056204,
      "grad_norm": 4.237648963928223,
      "learning_rate": 4.149563203873277e-05,
      "loss": 0.6905,
      "step": 16030
    },
    {
      "epoch": 5.06188822229239,
      "grad_norm": 3.2302465438842773,
      "learning_rate": 4.14947549380767e-05,
      "loss": 0.5948,
      "step": 16031
    },
    {
      "epoch": 5.062203978528576,
      "grad_norm": 4.280113697052002,
      "learning_rate": 4.1493877837420624e-05,
      "loss": 0.6684,
      "step": 16032
    },
    {
      "epoch": 5.062519734764762,
      "grad_norm": 3.5832812786102295,
      "learning_rate": 4.1493000736764556e-05,
      "loss": 0.5443,
      "step": 16033
    },
    {
      "epoch": 5.062835491000947,
      "grad_norm": 3.592026710510254,
      "learning_rate": 4.149212363610848e-05,
      "loss": 0.633,
      "step": 16034
    },
    {
      "epoch": 5.0631512472371325,
      "grad_norm": 4.372962474822998,
      "learning_rate": 4.149124653545241e-05,
      "loss": 0.6913,
      "step": 16035
    },
    {
      "epoch": 5.063467003473319,
      "grad_norm": 3.451070785522461,
      "learning_rate": 4.149036943479634e-05,
      "loss": 0.5794,
      "step": 16036
    },
    {
      "epoch": 5.063782759709504,
      "grad_norm": 3.8850929737091064,
      "learning_rate": 4.148949233414027e-05,
      "loss": 0.6053,
      "step": 16037
    },
    {
      "epoch": 5.06409851594569,
      "grad_norm": 4.033620834350586,
      "learning_rate": 4.1488615233484194e-05,
      "loss": 0.6655,
      "step": 16038
    },
    {
      "epoch": 5.0644142721818755,
      "grad_norm": 4.304964542388916,
      "learning_rate": 4.1487738132828125e-05,
      "loss": 0.6029,
      "step": 16039
    },
    {
      "epoch": 5.064730028418062,
      "grad_norm": 4.122103214263916,
      "learning_rate": 4.148686103217205e-05,
      "loss": 0.6652,
      "step": 16040
    },
    {
      "epoch": 5.065045784654247,
      "grad_norm": 3.8706319332122803,
      "learning_rate": 4.148598393151598e-05,
      "loss": 0.5791,
      "step": 16041
    },
    {
      "epoch": 5.065361540890432,
      "grad_norm": 3.65280818939209,
      "learning_rate": 4.148510683085991e-05,
      "loss": 0.6615,
      "step": 16042
    },
    {
      "epoch": 5.065677297126618,
      "grad_norm": 3.420109748840332,
      "learning_rate": 4.148422973020384e-05,
      "loss": 0.5848,
      "step": 16043
    },
    {
      "epoch": 5.065993053362804,
      "grad_norm": 3.9358437061309814,
      "learning_rate": 4.148335262954777e-05,
      "loss": 0.7948,
      "step": 16044
    },
    {
      "epoch": 5.06630880959899,
      "grad_norm": 3.5697319507598877,
      "learning_rate": 4.1482475528891695e-05,
      "loss": 0.6083,
      "step": 16045
    },
    {
      "epoch": 5.066624565835175,
      "grad_norm": 4.20619535446167,
      "learning_rate": 4.148159842823563e-05,
      "loss": 0.6869,
      "step": 16046
    },
    {
      "epoch": 5.0669403220713605,
      "grad_norm": 4.03124475479126,
      "learning_rate": 4.148072132757955e-05,
      "loss": 0.6096,
      "step": 16047
    },
    {
      "epoch": 5.067256078307547,
      "grad_norm": 3.364877939224243,
      "learning_rate": 4.1479844226923483e-05,
      "loss": 0.6053,
      "step": 16048
    },
    {
      "epoch": 5.067571834543732,
      "grad_norm": 3.858346462249756,
      "learning_rate": 4.1478967126267415e-05,
      "loss": 0.645,
      "step": 16049
    },
    {
      "epoch": 5.067887590779918,
      "grad_norm": 4.297241687774658,
      "learning_rate": 4.147809002561135e-05,
      "loss": 0.6222,
      "step": 16050
    },
    {
      "epoch": 5.068203347016103,
      "grad_norm": 5.203825950622559,
      "learning_rate": 4.147721292495527e-05,
      "loss": 0.6898,
      "step": 16051
    },
    {
      "epoch": 5.06851910325229,
      "grad_norm": 5.904796123504639,
      "learning_rate": 4.14763358242992e-05,
      "loss": 0.6763,
      "step": 16052
    },
    {
      "epoch": 5.068834859488475,
      "grad_norm": 4.30973482131958,
      "learning_rate": 4.147545872364313e-05,
      "loss": 0.6241,
      "step": 16053
    },
    {
      "epoch": 5.06915061572466,
      "grad_norm": 4.894724369049072,
      "learning_rate": 4.147458162298706e-05,
      "loss": 0.649,
      "step": 16054
    },
    {
      "epoch": 5.069466371960846,
      "grad_norm": 6.396925926208496,
      "learning_rate": 4.1473704522330985e-05,
      "loss": 0.7125,
      "step": 16055
    },
    {
      "epoch": 5.069782128197032,
      "grad_norm": 6.180676460266113,
      "learning_rate": 4.1472827421674917e-05,
      "loss": 0.7383,
      "step": 16056
    },
    {
      "epoch": 5.070097884433218,
      "grad_norm": 3.5573372840881348,
      "learning_rate": 4.147195032101884e-05,
      "loss": 0.5785,
      "step": 16057
    },
    {
      "epoch": 5.070413640669403,
      "grad_norm": 3.8218185901641846,
      "learning_rate": 4.147107322036277e-05,
      "loss": 0.6348,
      "step": 16058
    },
    {
      "epoch": 5.070729396905589,
      "grad_norm": 4.158382415771484,
      "learning_rate": 4.14701961197067e-05,
      "loss": 0.6502,
      "step": 16059
    },
    {
      "epoch": 5.071045153141775,
      "grad_norm": 3.8233516216278076,
      "learning_rate": 4.146931901905063e-05,
      "loss": 0.6311,
      "step": 16060
    },
    {
      "epoch": 5.07136090937796,
      "grad_norm": 3.795985460281372,
      "learning_rate": 4.1468441918394555e-05,
      "loss": 0.6021,
      "step": 16061
    },
    {
      "epoch": 5.071676665614146,
      "grad_norm": 4.5451250076293945,
      "learning_rate": 4.1467564817738486e-05,
      "loss": 0.6309,
      "step": 16062
    },
    {
      "epoch": 5.071992421850331,
      "grad_norm": 4.034921646118164,
      "learning_rate": 4.146668771708241e-05,
      "loss": 0.6327,
      "step": 16063
    },
    {
      "epoch": 5.072308178086518,
      "grad_norm": 4.436227321624756,
      "learning_rate": 4.146581061642634e-05,
      "loss": 0.6501,
      "step": 16064
    },
    {
      "epoch": 5.072623934322703,
      "grad_norm": 5.109795093536377,
      "learning_rate": 4.1464933515770275e-05,
      "loss": 0.7432,
      "step": 16065
    },
    {
      "epoch": 5.072939690558888,
      "grad_norm": 4.563517093658447,
      "learning_rate": 4.14640564151142e-05,
      "loss": 0.6313,
      "step": 16066
    },
    {
      "epoch": 5.073255446795074,
      "grad_norm": 4.259586334228516,
      "learning_rate": 4.146317931445813e-05,
      "loss": 0.6907,
      "step": 16067
    },
    {
      "epoch": 5.07357120303126,
      "grad_norm": 3.728116989135742,
      "learning_rate": 4.1462302213802056e-05,
      "loss": 0.6499,
      "step": 16068
    },
    {
      "epoch": 5.073886959267446,
      "grad_norm": 3.6832587718963623,
      "learning_rate": 4.146142511314599e-05,
      "loss": 0.6088,
      "step": 16069
    },
    {
      "epoch": 5.074202715503631,
      "grad_norm": 3.8844714164733887,
      "learning_rate": 4.146054801248991e-05,
      "loss": 0.5928,
      "step": 16070
    },
    {
      "epoch": 5.074518471739817,
      "grad_norm": 4.269970417022705,
      "learning_rate": 4.1459670911833844e-05,
      "loss": 0.6297,
      "step": 16071
    },
    {
      "epoch": 5.074834227976003,
      "grad_norm": 3.749929666519165,
      "learning_rate": 4.145879381117777e-05,
      "loss": 0.6267,
      "step": 16072
    },
    {
      "epoch": 5.075149984212188,
      "grad_norm": 4.671297073364258,
      "learning_rate": 4.14579167105217e-05,
      "loss": 0.6445,
      "step": 16073
    },
    {
      "epoch": 5.075465740448374,
      "grad_norm": 3.628896713256836,
      "learning_rate": 4.1457039609865626e-05,
      "loss": 0.6648,
      "step": 16074
    },
    {
      "epoch": 5.075781496684559,
      "grad_norm": 4.499355792999268,
      "learning_rate": 4.145616250920956e-05,
      "loss": 0.7127,
      "step": 16075
    },
    {
      "epoch": 5.0760972529207455,
      "grad_norm": 3.6915266513824463,
      "learning_rate": 4.145528540855349e-05,
      "loss": 0.5874,
      "step": 16076
    },
    {
      "epoch": 5.076413009156931,
      "grad_norm": 5.28945255279541,
      "learning_rate": 4.145440830789742e-05,
      "loss": 0.6705,
      "step": 16077
    },
    {
      "epoch": 5.076728765393116,
      "grad_norm": 4.130647659301758,
      "learning_rate": 4.1453531207241346e-05,
      "loss": 0.5869,
      "step": 16078
    },
    {
      "epoch": 5.077044521629302,
      "grad_norm": 4.12493371963501,
      "learning_rate": 4.145265410658528e-05,
      "loss": 0.6451,
      "step": 16079
    },
    {
      "epoch": 5.077360277865488,
      "grad_norm": 3.96224308013916,
      "learning_rate": 4.14517770059292e-05,
      "loss": 0.6469,
      "step": 16080
    },
    {
      "epoch": 5.077676034101674,
      "grad_norm": 3.7894015312194824,
      "learning_rate": 4.1450899905273134e-05,
      "loss": 0.6145,
      "step": 16081
    },
    {
      "epoch": 5.077991790337859,
      "grad_norm": 3.9703993797302246,
      "learning_rate": 4.145002280461706e-05,
      "loss": 0.6745,
      "step": 16082
    },
    {
      "epoch": 5.078307546574045,
      "grad_norm": 4.109126091003418,
      "learning_rate": 4.144914570396099e-05,
      "loss": 0.6095,
      "step": 16083
    },
    {
      "epoch": 5.0786233028102306,
      "grad_norm": 4.54642391204834,
      "learning_rate": 4.1448268603304916e-05,
      "loss": 0.7205,
      "step": 16084
    },
    {
      "epoch": 5.078939059046416,
      "grad_norm": 4.491814613342285,
      "learning_rate": 4.144739150264885e-05,
      "loss": 0.6105,
      "step": 16085
    },
    {
      "epoch": 5.079254815282602,
      "grad_norm": 3.4553816318511963,
      "learning_rate": 4.144651440199277e-05,
      "loss": 0.6709,
      "step": 16086
    },
    {
      "epoch": 5.079570571518787,
      "grad_norm": 3.3639581203460693,
      "learning_rate": 4.1445637301336704e-05,
      "loss": 0.6147,
      "step": 16087
    },
    {
      "epoch": 5.0798863277549735,
      "grad_norm": 4.573370456695557,
      "learning_rate": 4.1444760200680635e-05,
      "loss": 0.5822,
      "step": 16088
    },
    {
      "epoch": 5.080202083991159,
      "grad_norm": 3.825532913208008,
      "learning_rate": 4.144388310002456e-05,
      "loss": 0.6797,
      "step": 16089
    },
    {
      "epoch": 5.080517840227344,
      "grad_norm": 3.7092795372009277,
      "learning_rate": 4.144300599936849e-05,
      "loss": 0.6282,
      "step": 16090
    },
    {
      "epoch": 5.08083359646353,
      "grad_norm": 4.7371063232421875,
      "learning_rate": 4.144212889871242e-05,
      "loss": 0.5979,
      "step": 16091
    },
    {
      "epoch": 5.081149352699716,
      "grad_norm": 3.7125041484832764,
      "learning_rate": 4.144125179805635e-05,
      "loss": 0.5583,
      "step": 16092
    },
    {
      "epoch": 5.081465108935902,
      "grad_norm": 4.546538829803467,
      "learning_rate": 4.1440374697400274e-05,
      "loss": 0.5996,
      "step": 16093
    },
    {
      "epoch": 5.081780865172087,
      "grad_norm": 4.259352207183838,
      "learning_rate": 4.1439497596744205e-05,
      "loss": 0.5854,
      "step": 16094
    },
    {
      "epoch": 5.082096621408273,
      "grad_norm": 4.3510332107543945,
      "learning_rate": 4.143862049608813e-05,
      "loss": 0.6655,
      "step": 16095
    },
    {
      "epoch": 5.0824123776444585,
      "grad_norm": 4.293037414550781,
      "learning_rate": 4.143774339543206e-05,
      "loss": 0.607,
      "step": 16096
    },
    {
      "epoch": 5.082728133880644,
      "grad_norm": 4.4541335105896,
      "learning_rate": 4.143686629477599e-05,
      "loss": 0.5853,
      "step": 16097
    },
    {
      "epoch": 5.08304389011683,
      "grad_norm": 4.131140232086182,
      "learning_rate": 4.143598919411992e-05,
      "loss": 0.6498,
      "step": 16098
    },
    {
      "epoch": 5.083359646353015,
      "grad_norm": 3.787843704223633,
      "learning_rate": 4.143511209346384e-05,
      "loss": 0.5649,
      "step": 16099
    },
    {
      "epoch": 5.0836754025892015,
      "grad_norm": 4.4755706787109375,
      "learning_rate": 4.1434234992807775e-05,
      "loss": 0.6692,
      "step": 16100
    },
    {
      "epoch": 5.083991158825387,
      "grad_norm": 4.149050712585449,
      "learning_rate": 4.14333578921517e-05,
      "loss": 0.6804,
      "step": 16101
    },
    {
      "epoch": 5.084306915061572,
      "grad_norm": 5.101046085357666,
      "learning_rate": 4.143248079149563e-05,
      "loss": 0.6456,
      "step": 16102
    },
    {
      "epoch": 5.084622671297758,
      "grad_norm": 3.9949796199798584,
      "learning_rate": 4.143160369083956e-05,
      "loss": 0.639,
      "step": 16103
    },
    {
      "epoch": 5.0849384275339435,
      "grad_norm": 4.383967399597168,
      "learning_rate": 4.1430726590183495e-05,
      "loss": 0.6981,
      "step": 16104
    },
    {
      "epoch": 5.08525418377013,
      "grad_norm": 4.388382911682129,
      "learning_rate": 4.142984948952742e-05,
      "loss": 0.6171,
      "step": 16105
    },
    {
      "epoch": 5.085569940006315,
      "grad_norm": 5.002859592437744,
      "learning_rate": 4.142897238887135e-05,
      "loss": 0.6169,
      "step": 16106
    },
    {
      "epoch": 5.085885696242501,
      "grad_norm": 3.186110019683838,
      "learning_rate": 4.1428095288215276e-05,
      "loss": 0.547,
      "step": 16107
    },
    {
      "epoch": 5.0862014524786865,
      "grad_norm": 3.782837390899658,
      "learning_rate": 4.142721818755921e-05,
      "loss": 0.6042,
      "step": 16108
    },
    {
      "epoch": 5.086517208714872,
      "grad_norm": 4.6310224533081055,
      "learning_rate": 4.142634108690314e-05,
      "loss": 0.5853,
      "step": 16109
    },
    {
      "epoch": 5.086832964951058,
      "grad_norm": 5.039865970611572,
      "learning_rate": 4.1425463986247065e-05,
      "loss": 0.6896,
      "step": 16110
    },
    {
      "epoch": 5.087148721187243,
      "grad_norm": 4.057590484619141,
      "learning_rate": 4.1424586885590996e-05,
      "loss": 0.6208,
      "step": 16111
    },
    {
      "epoch": 5.087464477423429,
      "grad_norm": 4.528151988983154,
      "learning_rate": 4.142370978493492e-05,
      "loss": 0.6655,
      "step": 16112
    },
    {
      "epoch": 5.087780233659615,
      "grad_norm": 4.332470417022705,
      "learning_rate": 4.142283268427885e-05,
      "loss": 0.7055,
      "step": 16113
    },
    {
      "epoch": 5.0880959898958,
      "grad_norm": 4.83433723449707,
      "learning_rate": 4.142195558362278e-05,
      "loss": 0.5907,
      "step": 16114
    },
    {
      "epoch": 5.088411746131986,
      "grad_norm": 3.901761531829834,
      "learning_rate": 4.142107848296671e-05,
      "loss": 0.6101,
      "step": 16115
    },
    {
      "epoch": 5.0887275023681715,
      "grad_norm": 4.634686470031738,
      "learning_rate": 4.1420201382310635e-05,
      "loss": 0.6142,
      "step": 16116
    },
    {
      "epoch": 5.089043258604358,
      "grad_norm": 3.941596746444702,
      "learning_rate": 4.1419324281654566e-05,
      "loss": 0.6043,
      "step": 16117
    },
    {
      "epoch": 5.089359014840543,
      "grad_norm": 4.938940525054932,
      "learning_rate": 4.141844718099849e-05,
      "loss": 0.6111,
      "step": 16118
    },
    {
      "epoch": 5.089674771076729,
      "grad_norm": 4.013598918914795,
      "learning_rate": 4.141757008034242e-05,
      "loss": 0.6902,
      "step": 16119
    },
    {
      "epoch": 5.0899905273129145,
      "grad_norm": 3.111401081085205,
      "learning_rate": 4.141669297968635e-05,
      "loss": 0.5878,
      "step": 16120
    },
    {
      "epoch": 5.0903062835491,
      "grad_norm": 4.352319717407227,
      "learning_rate": 4.141581587903028e-05,
      "loss": 0.748,
      "step": 16121
    },
    {
      "epoch": 5.090622039785286,
      "grad_norm": 3.973145008087158,
      "learning_rate": 4.1414938778374204e-05,
      "loss": 0.674,
      "step": 16122
    },
    {
      "epoch": 5.090937796021471,
      "grad_norm": 3.9938483238220215,
      "learning_rate": 4.1414061677718136e-05,
      "loss": 0.617,
      "step": 16123
    },
    {
      "epoch": 5.091253552257657,
      "grad_norm": 4.364981174468994,
      "learning_rate": 4.141318457706206e-05,
      "loss": 0.6115,
      "step": 16124
    },
    {
      "epoch": 5.091569308493843,
      "grad_norm": 4.110325336456299,
      "learning_rate": 4.141230747640599e-05,
      "loss": 0.6074,
      "step": 16125
    },
    {
      "epoch": 5.091885064730028,
      "grad_norm": 3.114992141723633,
      "learning_rate": 4.1411430375749924e-05,
      "loss": 0.5944,
      "step": 16126
    },
    {
      "epoch": 5.092200820966214,
      "grad_norm": 4.523548603057861,
      "learning_rate": 4.141055327509385e-05,
      "loss": 0.687,
      "step": 16127
    },
    {
      "epoch": 5.0925165772023995,
      "grad_norm": 4.36926794052124,
      "learning_rate": 4.140967617443778e-05,
      "loss": 0.678,
      "step": 16128
    },
    {
      "epoch": 5.092832333438586,
      "grad_norm": 4.476365089416504,
      "learning_rate": 4.1408799073781706e-05,
      "loss": 0.6855,
      "step": 16129
    },
    {
      "epoch": 5.093148089674771,
      "grad_norm": 4.146389007568359,
      "learning_rate": 4.140792197312564e-05,
      "loss": 0.563,
      "step": 16130
    },
    {
      "epoch": 5.093463845910957,
      "grad_norm": 4.069138050079346,
      "learning_rate": 4.140704487246957e-05,
      "loss": 0.5958,
      "step": 16131
    },
    {
      "epoch": 5.093779602147142,
      "grad_norm": 3.897189140319824,
      "learning_rate": 4.14061677718135e-05,
      "loss": 0.6301,
      "step": 16132
    },
    {
      "epoch": 5.094095358383328,
      "grad_norm": 3.096625328063965,
      "learning_rate": 4.1405290671157426e-05,
      "loss": 0.6422,
      "step": 16133
    },
    {
      "epoch": 5.094411114619514,
      "grad_norm": 3.9111416339874268,
      "learning_rate": 4.140441357050136e-05,
      "loss": 0.6035,
      "step": 16134
    },
    {
      "epoch": 5.094726870855699,
      "grad_norm": 4.5945143699646,
      "learning_rate": 4.140353646984528e-05,
      "loss": 0.7047,
      "step": 16135
    },
    {
      "epoch": 5.095042627091885,
      "grad_norm": 4.4087090492248535,
      "learning_rate": 4.1402659369189214e-05,
      "loss": 0.64,
      "step": 16136
    },
    {
      "epoch": 5.095358383328071,
      "grad_norm": 3.0120840072631836,
      "learning_rate": 4.140178226853314e-05,
      "loss": 0.5904,
      "step": 16137
    },
    {
      "epoch": 5.095674139564256,
      "grad_norm": 4.0199198722839355,
      "learning_rate": 4.140090516787707e-05,
      "loss": 0.6717,
      "step": 16138
    },
    {
      "epoch": 5.095989895800442,
      "grad_norm": 6.872800350189209,
      "learning_rate": 4.1400028067220995e-05,
      "loss": 0.6491,
      "step": 16139
    },
    {
      "epoch": 5.096305652036627,
      "grad_norm": 4.422708511352539,
      "learning_rate": 4.139915096656493e-05,
      "loss": 0.6442,
      "step": 16140
    },
    {
      "epoch": 5.096621408272814,
      "grad_norm": 4.843491077423096,
      "learning_rate": 4.139827386590885e-05,
      "loss": 0.6869,
      "step": 16141
    },
    {
      "epoch": 5.096937164508999,
      "grad_norm": 3.4497249126434326,
      "learning_rate": 4.1397396765252784e-05,
      "loss": 0.5471,
      "step": 16142
    },
    {
      "epoch": 5.097252920745185,
      "grad_norm": 3.761911153793335,
      "learning_rate": 4.139651966459671e-05,
      "loss": 0.6323,
      "step": 16143
    },
    {
      "epoch": 5.09756867698137,
      "grad_norm": 4.123054027557373,
      "learning_rate": 4.139564256394064e-05,
      "loss": 0.6893,
      "step": 16144
    },
    {
      "epoch": 5.097884433217556,
      "grad_norm": 4.057101249694824,
      "learning_rate": 4.1394765463284565e-05,
      "loss": 0.5499,
      "step": 16145
    },
    {
      "epoch": 5.098200189453742,
      "grad_norm": 3.9696056842803955,
      "learning_rate": 4.13938883626285e-05,
      "loss": 0.6495,
      "step": 16146
    },
    {
      "epoch": 5.098515945689927,
      "grad_norm": 4.211608409881592,
      "learning_rate": 4.139301126197243e-05,
      "loss": 0.5625,
      "step": 16147
    },
    {
      "epoch": 5.098831701926113,
      "grad_norm": 5.169034481048584,
      "learning_rate": 4.1392134161316353e-05,
      "loss": 0.637,
      "step": 16148
    },
    {
      "epoch": 5.099147458162299,
      "grad_norm": 4.044218063354492,
      "learning_rate": 4.1391257060660285e-05,
      "loss": 0.5817,
      "step": 16149
    },
    {
      "epoch": 5.099463214398484,
      "grad_norm": 5.864421844482422,
      "learning_rate": 4.139037996000421e-05,
      "loss": 0.5998,
      "step": 16150
    },
    {
      "epoch": 5.09977897063467,
      "grad_norm": 3.6610896587371826,
      "learning_rate": 4.138950285934814e-05,
      "loss": 0.6284,
      "step": 16151
    },
    {
      "epoch": 5.100094726870855,
      "grad_norm": 4.679323673248291,
      "learning_rate": 4.138862575869207e-05,
      "loss": 0.7216,
      "step": 16152
    },
    {
      "epoch": 5.100410483107042,
      "grad_norm": 4.4304304122924805,
      "learning_rate": 4.1387748658036e-05,
      "loss": 0.6699,
      "step": 16153
    },
    {
      "epoch": 5.100726239343227,
      "grad_norm": 4.696525573730469,
      "learning_rate": 4.138687155737992e-05,
      "loss": 0.6162,
      "step": 16154
    },
    {
      "epoch": 5.101041995579413,
      "grad_norm": 3.903959274291992,
      "learning_rate": 4.1385994456723855e-05,
      "loss": 0.6681,
      "step": 16155
    },
    {
      "epoch": 5.101357751815598,
      "grad_norm": 5.2612624168396,
      "learning_rate": 4.138511735606778e-05,
      "loss": 0.5902,
      "step": 16156
    },
    {
      "epoch": 5.101673508051784,
      "grad_norm": 4.684999465942383,
      "learning_rate": 4.138424025541171e-05,
      "loss": 0.739,
      "step": 16157
    },
    {
      "epoch": 5.10198926428797,
      "grad_norm": 5.69086217880249,
      "learning_rate": 4.138336315475564e-05,
      "loss": 0.6908,
      "step": 16158
    },
    {
      "epoch": 5.102305020524155,
      "grad_norm": 8.81539249420166,
      "learning_rate": 4.1382486054099575e-05,
      "loss": 0.6468,
      "step": 16159
    },
    {
      "epoch": 5.102620776760341,
      "grad_norm": 4.773634433746338,
      "learning_rate": 4.13816089534435e-05,
      "loss": 0.5698,
      "step": 16160
    },
    {
      "epoch": 5.102936532996527,
      "grad_norm": 3.6766483783721924,
      "learning_rate": 4.138073185278743e-05,
      "loss": 0.5949,
      "step": 16161
    },
    {
      "epoch": 5.103252289232712,
      "grad_norm": 4.719614505767822,
      "learning_rate": 4.1379854752131356e-05,
      "loss": 0.6278,
      "step": 16162
    },
    {
      "epoch": 5.103568045468898,
      "grad_norm": 4.316293239593506,
      "learning_rate": 4.137897765147529e-05,
      "loss": 0.6094,
      "step": 16163
    },
    {
      "epoch": 5.103883801705083,
      "grad_norm": 3.5126845836639404,
      "learning_rate": 4.137810055081921e-05,
      "loss": 0.5615,
      "step": 16164
    },
    {
      "epoch": 5.10419955794127,
      "grad_norm": 4.125084400177002,
      "learning_rate": 4.1377223450163145e-05,
      "loss": 0.5648,
      "step": 16165
    },
    {
      "epoch": 5.104515314177455,
      "grad_norm": 3.4453372955322266,
      "learning_rate": 4.137634634950707e-05,
      "loss": 0.5564,
      "step": 16166
    },
    {
      "epoch": 5.104831070413641,
      "grad_norm": 4.39155912399292,
      "learning_rate": 4.1375469248851e-05,
      "loss": 0.5946,
      "step": 16167
    },
    {
      "epoch": 5.105146826649826,
      "grad_norm": 3.3721632957458496,
      "learning_rate": 4.1374592148194926e-05,
      "loss": 0.5547,
      "step": 16168
    },
    {
      "epoch": 5.105462582886012,
      "grad_norm": 3.831167459487915,
      "learning_rate": 4.137371504753886e-05,
      "loss": 0.6752,
      "step": 16169
    },
    {
      "epoch": 5.105778339122198,
      "grad_norm": 4.246902942657471,
      "learning_rate": 4.137283794688279e-05,
      "loss": 0.606,
      "step": 16170
    },
    {
      "epoch": 5.106094095358383,
      "grad_norm": 4.105796813964844,
      "learning_rate": 4.1371960846226714e-05,
      "loss": 0.7034,
      "step": 16171
    },
    {
      "epoch": 5.106409851594569,
      "grad_norm": 5.045901775360107,
      "learning_rate": 4.1371083745570646e-05,
      "loss": 0.759,
      "step": 16172
    },
    {
      "epoch": 5.106725607830755,
      "grad_norm": 4.21860408782959,
      "learning_rate": 4.137020664491457e-05,
      "loss": 0.7052,
      "step": 16173
    },
    {
      "epoch": 5.107041364066941,
      "grad_norm": 3.4559078216552734,
      "learning_rate": 4.13693295442585e-05,
      "loss": 0.5067,
      "step": 16174
    },
    {
      "epoch": 5.107357120303126,
      "grad_norm": 4.111161231994629,
      "learning_rate": 4.136845244360243e-05,
      "loss": 0.5927,
      "step": 16175
    },
    {
      "epoch": 5.107672876539311,
      "grad_norm": 3.7345306873321533,
      "learning_rate": 4.136757534294636e-05,
      "loss": 0.5514,
      "step": 16176
    },
    {
      "epoch": 5.1079886327754975,
      "grad_norm": 3.5891048908233643,
      "learning_rate": 4.1366698242290284e-05,
      "loss": 0.5999,
      "step": 16177
    },
    {
      "epoch": 5.108304389011683,
      "grad_norm": 3.252586841583252,
      "learning_rate": 4.1365821141634216e-05,
      "loss": 0.6167,
      "step": 16178
    },
    {
      "epoch": 5.108620145247869,
      "grad_norm": 3.9738070964813232,
      "learning_rate": 4.136494404097814e-05,
      "loss": 0.5416,
      "step": 16179
    },
    {
      "epoch": 5.108935901484054,
      "grad_norm": 3.922105550765991,
      "learning_rate": 4.136406694032207e-05,
      "loss": 0.6263,
      "step": 16180
    },
    {
      "epoch": 5.10925165772024,
      "grad_norm": 4.334053993225098,
      "learning_rate": 4.1363189839666e-05,
      "loss": 0.6236,
      "step": 16181
    },
    {
      "epoch": 5.109567413956426,
      "grad_norm": 3.5472681522369385,
      "learning_rate": 4.136231273900993e-05,
      "loss": 0.5251,
      "step": 16182
    },
    {
      "epoch": 5.109883170192611,
      "grad_norm": 3.6840426921844482,
      "learning_rate": 4.1361435638353854e-05,
      "loss": 0.6506,
      "step": 16183
    },
    {
      "epoch": 5.110198926428797,
      "grad_norm": 4.54060173034668,
      "learning_rate": 4.1360558537697786e-05,
      "loss": 0.713,
      "step": 16184
    },
    {
      "epoch": 5.1105146826649825,
      "grad_norm": 3.33304762840271,
      "learning_rate": 4.135968143704172e-05,
      "loss": 0.5462,
      "step": 16185
    },
    {
      "epoch": 5.110830438901169,
      "grad_norm": 4.114576816558838,
      "learning_rate": 4.135880433638565e-05,
      "loss": 0.593,
      "step": 16186
    },
    {
      "epoch": 5.111146195137354,
      "grad_norm": 4.333873748779297,
      "learning_rate": 4.1357927235729574e-05,
      "loss": 0.6019,
      "step": 16187
    },
    {
      "epoch": 5.111461951373539,
      "grad_norm": 4.027337551116943,
      "learning_rate": 4.1357050135073506e-05,
      "loss": 0.5935,
      "step": 16188
    },
    {
      "epoch": 5.1117777076097255,
      "grad_norm": 3.4959259033203125,
      "learning_rate": 4.135617303441743e-05,
      "loss": 0.5981,
      "step": 16189
    },
    {
      "epoch": 5.112093463845911,
      "grad_norm": 4.084064960479736,
      "learning_rate": 4.135529593376136e-05,
      "loss": 0.6649,
      "step": 16190
    },
    {
      "epoch": 5.112409220082097,
      "grad_norm": 4.353405952453613,
      "learning_rate": 4.1354418833105294e-05,
      "loss": 0.6906,
      "step": 16191
    },
    {
      "epoch": 5.112724976318282,
      "grad_norm": 3.691603422164917,
      "learning_rate": 4.135354173244922e-05,
      "loss": 0.5807,
      "step": 16192
    },
    {
      "epoch": 5.113040732554468,
      "grad_norm": 5.923047065734863,
      "learning_rate": 4.135266463179315e-05,
      "loss": 0.7728,
      "step": 16193
    },
    {
      "epoch": 5.113356488790654,
      "grad_norm": 4.815986633300781,
      "learning_rate": 4.1351787531137075e-05,
      "loss": 0.6774,
      "step": 16194
    },
    {
      "epoch": 5.113672245026839,
      "grad_norm": 4.071826457977295,
      "learning_rate": 4.135091043048101e-05,
      "loss": 0.665,
      "step": 16195
    },
    {
      "epoch": 5.113988001263025,
      "grad_norm": 3.9902281761169434,
      "learning_rate": 4.135003332982493e-05,
      "loss": 0.5758,
      "step": 16196
    },
    {
      "epoch": 5.1143037574992105,
      "grad_norm": 4.657958030700684,
      "learning_rate": 4.1349156229168864e-05,
      "loss": 0.7115,
      "step": 16197
    },
    {
      "epoch": 5.114619513735397,
      "grad_norm": 4.642496585845947,
      "learning_rate": 4.134827912851279e-05,
      "loss": 0.6381,
      "step": 16198
    },
    {
      "epoch": 5.114935269971582,
      "grad_norm": 3.767570972442627,
      "learning_rate": 4.134740202785672e-05,
      "loss": 0.6363,
      "step": 16199
    },
    {
      "epoch": 5.115251026207767,
      "grad_norm": 4.76125955581665,
      "learning_rate": 4.1346524927200645e-05,
      "loss": 0.693,
      "step": 16200
    },
    {
      "epoch": 5.115251026207767,
      "eval_loss": 0.853821337223053,
      "eval_runtime": 4.5089,
      "eval_samples_per_second": 552.239,
      "eval_steps_per_second": 4.657,
      "step": 16200
    },
    {
      "epoch": 5.1155667824439535,
      "grad_norm": 4.50347900390625,
      "learning_rate": 4.134564782654458e-05,
      "loss": 0.6596,
      "step": 16201
    },
    {
      "epoch": 5.115882538680139,
      "grad_norm": 3.2315852642059326,
      "learning_rate": 4.13447707258885e-05,
      "loss": 0.6302,
      "step": 16202
    },
    {
      "epoch": 5.116198294916325,
      "grad_norm": 3.807058811187744,
      "learning_rate": 4.134389362523243e-05,
      "loss": 0.5348,
      "step": 16203
    },
    {
      "epoch": 5.11651405115251,
      "grad_norm": 4.257772922515869,
      "learning_rate": 4.134301652457636e-05,
      "loss": 0.66,
      "step": 16204
    },
    {
      "epoch": 5.1168298073886955,
      "grad_norm": 4.249604225158691,
      "learning_rate": 4.134213942392029e-05,
      "loss": 0.5712,
      "step": 16205
    },
    {
      "epoch": 5.117145563624882,
      "grad_norm": 3.4919886589050293,
      "learning_rate": 4.1341262323264215e-05,
      "loss": 0.7116,
      "step": 16206
    },
    {
      "epoch": 5.117461319861067,
      "grad_norm": 3.302722454071045,
      "learning_rate": 4.1340385222608147e-05,
      "loss": 0.5822,
      "step": 16207
    },
    {
      "epoch": 5.117777076097253,
      "grad_norm": 4.005092620849609,
      "learning_rate": 4.133950812195208e-05,
      "loss": 0.6959,
      "step": 16208
    },
    {
      "epoch": 5.1180928323334385,
      "grad_norm": 4.530994892120361,
      "learning_rate": 4.1338631021296e-05,
      "loss": 0.6559,
      "step": 16209
    },
    {
      "epoch": 5.118408588569625,
      "grad_norm": 3.66912579536438,
      "learning_rate": 4.1337753920639935e-05,
      "loss": 0.6056,
      "step": 16210
    },
    {
      "epoch": 5.11872434480581,
      "grad_norm": 5.153197288513184,
      "learning_rate": 4.133687681998386e-05,
      "loss": 0.608,
      "step": 16211
    },
    {
      "epoch": 5.119040101041995,
      "grad_norm": 3.946561574935913,
      "learning_rate": 4.13359997193278e-05,
      "loss": 0.5709,
      "step": 16212
    },
    {
      "epoch": 5.119355857278181,
      "grad_norm": 5.008460998535156,
      "learning_rate": 4.133512261867172e-05,
      "loss": 0.5643,
      "step": 16213
    },
    {
      "epoch": 5.119671613514367,
      "grad_norm": 4.122055530548096,
      "learning_rate": 4.1334245518015655e-05,
      "loss": 0.6098,
      "step": 16214
    },
    {
      "epoch": 5.119987369750553,
      "grad_norm": 3.9297893047332764,
      "learning_rate": 4.133336841735958e-05,
      "loss": 0.6615,
      "step": 16215
    },
    {
      "epoch": 5.120303125986738,
      "grad_norm": 3.9641427993774414,
      "learning_rate": 4.133249131670351e-05,
      "loss": 0.6431,
      "step": 16216
    },
    {
      "epoch": 5.1206188822229235,
      "grad_norm": 4.149896621704102,
      "learning_rate": 4.1331614216047436e-05,
      "loss": 0.6297,
      "step": 16217
    },
    {
      "epoch": 5.12093463845911,
      "grad_norm": 4.352819919586182,
      "learning_rate": 4.133073711539137e-05,
      "loss": 0.6944,
      "step": 16218
    },
    {
      "epoch": 5.121250394695295,
      "grad_norm": 4.838973045349121,
      "learning_rate": 4.132986001473529e-05,
      "loss": 0.7116,
      "step": 16219
    },
    {
      "epoch": 5.121566150931481,
      "grad_norm": 4.459479331970215,
      "learning_rate": 4.1328982914079225e-05,
      "loss": 0.7532,
      "step": 16220
    },
    {
      "epoch": 5.1218819071676664,
      "grad_norm": 3.509127140045166,
      "learning_rate": 4.132810581342315e-05,
      "loss": 0.5803,
      "step": 16221
    },
    {
      "epoch": 5.122197663403853,
      "grad_norm": 4.5611958503723145,
      "learning_rate": 4.132722871276708e-05,
      "loss": 0.6394,
      "step": 16222
    },
    {
      "epoch": 5.122513419640038,
      "grad_norm": 4.981406211853027,
      "learning_rate": 4.1326351612111006e-05,
      "loss": 0.609,
      "step": 16223
    },
    {
      "epoch": 5.122829175876223,
      "grad_norm": 4.870112419128418,
      "learning_rate": 4.132547451145494e-05,
      "loss": 0.6044,
      "step": 16224
    },
    {
      "epoch": 5.123144932112409,
      "grad_norm": 3.407484769821167,
      "learning_rate": 4.132459741079886e-05,
      "loss": 0.6412,
      "step": 16225
    },
    {
      "epoch": 5.123460688348595,
      "grad_norm": 3.806211471557617,
      "learning_rate": 4.1323720310142794e-05,
      "loss": 0.6513,
      "step": 16226
    },
    {
      "epoch": 5.123776444584781,
      "grad_norm": 4.178483963012695,
      "learning_rate": 4.132284320948672e-05,
      "loss": 0.607,
      "step": 16227
    },
    {
      "epoch": 5.124092200820966,
      "grad_norm": 4.104130744934082,
      "learning_rate": 4.132196610883065e-05,
      "loss": 0.6644,
      "step": 16228
    },
    {
      "epoch": 5.1244079570571515,
      "grad_norm": 4.311544418334961,
      "learning_rate": 4.132108900817458e-05,
      "loss": 0.6662,
      "step": 16229
    },
    {
      "epoch": 5.124723713293338,
      "grad_norm": 4.235027313232422,
      "learning_rate": 4.132021190751851e-05,
      "loss": 0.6165,
      "step": 16230
    },
    {
      "epoch": 5.125039469529523,
      "grad_norm": 4.445625305175781,
      "learning_rate": 4.131933480686244e-05,
      "loss": 0.6167,
      "step": 16231
    },
    {
      "epoch": 5.125355225765709,
      "grad_norm": 4.334715843200684,
      "learning_rate": 4.1318457706206364e-05,
      "loss": 0.6454,
      "step": 16232
    },
    {
      "epoch": 5.125670982001894,
      "grad_norm": 4.944809913635254,
      "learning_rate": 4.1317580605550296e-05,
      "loss": 0.6827,
      "step": 16233
    },
    {
      "epoch": 5.125986738238081,
      "grad_norm": 4.525068283081055,
      "learning_rate": 4.131670350489422e-05,
      "loss": 0.6308,
      "step": 16234
    },
    {
      "epoch": 5.126302494474266,
      "grad_norm": 4.366648197174072,
      "learning_rate": 4.131582640423815e-05,
      "loss": 0.5349,
      "step": 16235
    },
    {
      "epoch": 5.126618250710451,
      "grad_norm": 4.711081504821777,
      "learning_rate": 4.131494930358208e-05,
      "loss": 0.6135,
      "step": 16236
    },
    {
      "epoch": 5.126934006946637,
      "grad_norm": 4.478187561035156,
      "learning_rate": 4.131407220292601e-05,
      "loss": 0.6633,
      "step": 16237
    },
    {
      "epoch": 5.127249763182823,
      "grad_norm": 3.8939013481140137,
      "learning_rate": 4.1313195102269934e-05,
      "loss": 0.6139,
      "step": 16238
    },
    {
      "epoch": 5.127565519419009,
      "grad_norm": 5.505854606628418,
      "learning_rate": 4.1312318001613866e-05,
      "loss": 0.6289,
      "step": 16239
    },
    {
      "epoch": 5.127881275655194,
      "grad_norm": 4.6911444664001465,
      "learning_rate": 4.13114409009578e-05,
      "loss": 0.6836,
      "step": 16240
    },
    {
      "epoch": 5.128197031891379,
      "grad_norm": 3.3940939903259277,
      "learning_rate": 4.131056380030173e-05,
      "loss": 0.6626,
      "step": 16241
    },
    {
      "epoch": 5.128512788127566,
      "grad_norm": 4.145432949066162,
      "learning_rate": 4.1309686699645654e-05,
      "loss": 0.5843,
      "step": 16242
    },
    {
      "epoch": 5.128828544363751,
      "grad_norm": 4.655227184295654,
      "learning_rate": 4.1308809598989585e-05,
      "loss": 0.6717,
      "step": 16243
    },
    {
      "epoch": 5.129144300599937,
      "grad_norm": 3.965165138244629,
      "learning_rate": 4.130793249833351e-05,
      "loss": 0.5772,
      "step": 16244
    },
    {
      "epoch": 5.129460056836122,
      "grad_norm": 4.688503265380859,
      "learning_rate": 4.130705539767744e-05,
      "loss": 0.6179,
      "step": 16245
    },
    {
      "epoch": 5.129775813072309,
      "grad_norm": 4.471514701843262,
      "learning_rate": 4.130617829702137e-05,
      "loss": 0.6572,
      "step": 16246
    },
    {
      "epoch": 5.130091569308494,
      "grad_norm": 3.932467222213745,
      "learning_rate": 4.13053011963653e-05,
      "loss": 0.6996,
      "step": 16247
    },
    {
      "epoch": 5.130407325544679,
      "grad_norm": 4.049493312835693,
      "learning_rate": 4.1304424095709224e-05,
      "loss": 0.7192,
      "step": 16248
    },
    {
      "epoch": 5.130723081780865,
      "grad_norm": 3.601027250289917,
      "learning_rate": 4.1303546995053155e-05,
      "loss": 0.5755,
      "step": 16249
    },
    {
      "epoch": 5.131038838017051,
      "grad_norm": 3.8247640132904053,
      "learning_rate": 4.130266989439709e-05,
      "loss": 0.5438,
      "step": 16250
    },
    {
      "epoch": 5.131354594253237,
      "grad_norm": 4.858988285064697,
      "learning_rate": 4.130179279374101e-05,
      "loss": 0.6584,
      "step": 16251
    },
    {
      "epoch": 5.131670350489422,
      "grad_norm": 5.3029465675354,
      "learning_rate": 4.1300915693084943e-05,
      "loss": 0.6418,
      "step": 16252
    },
    {
      "epoch": 5.131986106725607,
      "grad_norm": 4.596431732177734,
      "learning_rate": 4.130003859242887e-05,
      "loss": 0.54,
      "step": 16253
    },
    {
      "epoch": 5.132301862961794,
      "grad_norm": 3.9467523097991943,
      "learning_rate": 4.12991614917728e-05,
      "loss": 0.7296,
      "step": 16254
    },
    {
      "epoch": 5.132617619197979,
      "grad_norm": 4.159706115722656,
      "learning_rate": 4.1298284391116725e-05,
      "loss": 0.6336,
      "step": 16255
    },
    {
      "epoch": 5.132933375434165,
      "grad_norm": 3.4580235481262207,
      "learning_rate": 4.129740729046066e-05,
      "loss": 0.557,
      "step": 16256
    },
    {
      "epoch": 5.13324913167035,
      "grad_norm": 3.9024486541748047,
      "learning_rate": 4.129653018980458e-05,
      "loss": 0.5623,
      "step": 16257
    },
    {
      "epoch": 5.1335648879065365,
      "grad_norm": 4.106504917144775,
      "learning_rate": 4.129565308914851e-05,
      "loss": 0.6529,
      "step": 16258
    },
    {
      "epoch": 5.133880644142722,
      "grad_norm": 3.956538200378418,
      "learning_rate": 4.129477598849244e-05,
      "loss": 0.6931,
      "step": 16259
    },
    {
      "epoch": 5.134196400378907,
      "grad_norm": 4.2491655349731445,
      "learning_rate": 4.129389888783637e-05,
      "loss": 0.5612,
      "step": 16260
    },
    {
      "epoch": 5.134512156615093,
      "grad_norm": 4.433248519897461,
      "learning_rate": 4.1293021787180295e-05,
      "loss": 0.6656,
      "step": 16261
    },
    {
      "epoch": 5.134827912851279,
      "grad_norm": 4.111469268798828,
      "learning_rate": 4.1292144686524226e-05,
      "loss": 0.5942,
      "step": 16262
    },
    {
      "epoch": 5.135143669087465,
      "grad_norm": 5.609343528747559,
      "learning_rate": 4.129126758586815e-05,
      "loss": 0.8085,
      "step": 16263
    },
    {
      "epoch": 5.13545942532365,
      "grad_norm": 4.796813488006592,
      "learning_rate": 4.129039048521208e-05,
      "loss": 0.7928,
      "step": 16264
    },
    {
      "epoch": 5.135775181559836,
      "grad_norm": 3.0397820472717285,
      "learning_rate": 4.128951338455601e-05,
      "loss": 0.584,
      "step": 16265
    },
    {
      "epoch": 5.1360909377960215,
      "grad_norm": 3.781229257583618,
      "learning_rate": 4.128863628389994e-05,
      "loss": 0.6455,
      "step": 16266
    },
    {
      "epoch": 5.136406694032207,
      "grad_norm": 4.494765281677246,
      "learning_rate": 4.128775918324387e-05,
      "loss": 0.5942,
      "step": 16267
    },
    {
      "epoch": 5.136722450268393,
      "grad_norm": 4.220853805541992,
      "learning_rate": 4.12868820825878e-05,
      "loss": 0.5547,
      "step": 16268
    },
    {
      "epoch": 5.137038206504578,
      "grad_norm": 5.29337215423584,
      "learning_rate": 4.128600498193173e-05,
      "loss": 0.6823,
      "step": 16269
    },
    {
      "epoch": 5.1373539627407645,
      "grad_norm": 4.312450408935547,
      "learning_rate": 4.128512788127566e-05,
      "loss": 0.6606,
      "step": 16270
    },
    {
      "epoch": 5.13766971897695,
      "grad_norm": 4.065969944000244,
      "learning_rate": 4.1284250780619584e-05,
      "loss": 0.6502,
      "step": 16271
    },
    {
      "epoch": 5.137985475213135,
      "grad_norm": 3.576462745666504,
      "learning_rate": 4.1283373679963516e-05,
      "loss": 0.6754,
      "step": 16272
    },
    {
      "epoch": 5.138301231449321,
      "grad_norm": 4.283768653869629,
      "learning_rate": 4.128249657930745e-05,
      "loss": 0.6647,
      "step": 16273
    },
    {
      "epoch": 5.138616987685507,
      "grad_norm": 3.9878411293029785,
      "learning_rate": 4.128161947865137e-05,
      "loss": 0.6826,
      "step": 16274
    },
    {
      "epoch": 5.138932743921693,
      "grad_norm": 3.582059383392334,
      "learning_rate": 4.1280742377995304e-05,
      "loss": 0.5839,
      "step": 16275
    },
    {
      "epoch": 5.139248500157878,
      "grad_norm": 3.558076858520508,
      "learning_rate": 4.127986527733923e-05,
      "loss": 0.5776,
      "step": 16276
    },
    {
      "epoch": 5.139564256394063,
      "grad_norm": 3.389295816421509,
      "learning_rate": 4.127898817668316e-05,
      "loss": 0.6232,
      "step": 16277
    },
    {
      "epoch": 5.1398800126302495,
      "grad_norm": 4.0574631690979,
      "learning_rate": 4.1278111076027086e-05,
      "loss": 0.6004,
      "step": 16278
    },
    {
      "epoch": 5.140195768866435,
      "grad_norm": 4.427419185638428,
      "learning_rate": 4.127723397537102e-05,
      "loss": 0.6353,
      "step": 16279
    },
    {
      "epoch": 5.140511525102621,
      "grad_norm": 4.171928405761719,
      "learning_rate": 4.127635687471494e-05,
      "loss": 0.6236,
      "step": 16280
    },
    {
      "epoch": 5.140827281338806,
      "grad_norm": 4.294759750366211,
      "learning_rate": 4.1275479774058874e-05,
      "loss": 0.68,
      "step": 16281
    },
    {
      "epoch": 5.1411430375749925,
      "grad_norm": 4.232165336608887,
      "learning_rate": 4.12746026734028e-05,
      "loss": 0.6014,
      "step": 16282
    },
    {
      "epoch": 5.141458793811178,
      "grad_norm": 4.842602252960205,
      "learning_rate": 4.127372557274673e-05,
      "loss": 0.5303,
      "step": 16283
    },
    {
      "epoch": 5.141774550047363,
      "grad_norm": 4.409959316253662,
      "learning_rate": 4.1272848472090656e-05,
      "loss": 0.6807,
      "step": 16284
    },
    {
      "epoch": 5.142090306283549,
      "grad_norm": 5.189877033233643,
      "learning_rate": 4.127197137143459e-05,
      "loss": 0.6212,
      "step": 16285
    },
    {
      "epoch": 5.1424060625197345,
      "grad_norm": 3.734867572784424,
      "learning_rate": 4.127109427077851e-05,
      "loss": 0.5267,
      "step": 16286
    },
    {
      "epoch": 5.142721818755921,
      "grad_norm": 4.815629005432129,
      "learning_rate": 4.1270217170122444e-05,
      "loss": 0.6917,
      "step": 16287
    },
    {
      "epoch": 5.143037574992106,
      "grad_norm": 3.808129072189331,
      "learning_rate": 4.1269340069466376e-05,
      "loss": 0.63,
      "step": 16288
    },
    {
      "epoch": 5.143353331228292,
      "grad_norm": 4.716397285461426,
      "learning_rate": 4.12684629688103e-05,
      "loss": 0.7418,
      "step": 16289
    },
    {
      "epoch": 5.1436690874644775,
      "grad_norm": 4.50485372543335,
      "learning_rate": 4.126758586815423e-05,
      "loss": 0.5887,
      "step": 16290
    },
    {
      "epoch": 5.143984843700663,
      "grad_norm": 4.359118938446045,
      "learning_rate": 4.126670876749816e-05,
      "loss": 0.6204,
      "step": 16291
    },
    {
      "epoch": 5.144300599936849,
      "grad_norm": 3.594965934753418,
      "learning_rate": 4.126583166684209e-05,
      "loss": 0.6176,
      "step": 16292
    },
    {
      "epoch": 5.144616356173034,
      "grad_norm": 4.096256732940674,
      "learning_rate": 4.1264954566186014e-05,
      "loss": 0.5735,
      "step": 16293
    },
    {
      "epoch": 5.14493211240922,
      "grad_norm": 4.483431816101074,
      "learning_rate": 4.126407746552995e-05,
      "loss": 0.583,
      "step": 16294
    },
    {
      "epoch": 5.145247868645406,
      "grad_norm": 4.1757426261901855,
      "learning_rate": 4.126320036487388e-05,
      "loss": 0.5927,
      "step": 16295
    },
    {
      "epoch": 5.145563624881591,
      "grad_norm": 3.502533435821533,
      "learning_rate": 4.126232326421781e-05,
      "loss": 0.5845,
      "step": 16296
    },
    {
      "epoch": 5.145879381117777,
      "grad_norm": 5.139861583709717,
      "learning_rate": 4.1261446163561734e-05,
      "loss": 0.6946,
      "step": 16297
    },
    {
      "epoch": 5.1461951373539625,
      "grad_norm": 4.2180657386779785,
      "learning_rate": 4.1260569062905665e-05,
      "loss": 0.7413,
      "step": 16298
    },
    {
      "epoch": 5.146510893590149,
      "grad_norm": 3.6338694095611572,
      "learning_rate": 4.125969196224959e-05,
      "loss": 0.515,
      "step": 16299
    },
    {
      "epoch": 5.146826649826334,
      "grad_norm": 4.166937351226807,
      "learning_rate": 4.125881486159352e-05,
      "loss": 0.6614,
      "step": 16300
    },
    {
      "epoch": 5.14714240606252,
      "grad_norm": 3.8645620346069336,
      "learning_rate": 4.125793776093745e-05,
      "loss": 0.654,
      "step": 16301
    },
    {
      "epoch": 5.1474581622987055,
      "grad_norm": 3.8647263050079346,
      "learning_rate": 4.125706066028138e-05,
      "loss": 0.5402,
      "step": 16302
    },
    {
      "epoch": 5.147773918534891,
      "grad_norm": 3.346169948577881,
      "learning_rate": 4.1256183559625303e-05,
      "loss": 0.6337,
      "step": 16303
    },
    {
      "epoch": 5.148089674771077,
      "grad_norm": 4.616711616516113,
      "learning_rate": 4.1255306458969235e-05,
      "loss": 0.5429,
      "step": 16304
    },
    {
      "epoch": 5.148405431007262,
      "grad_norm": 3.818052291870117,
      "learning_rate": 4.125442935831316e-05,
      "loss": 0.7279,
      "step": 16305
    },
    {
      "epoch": 5.148721187243448,
      "grad_norm": 3.6740262508392334,
      "learning_rate": 4.125355225765709e-05,
      "loss": 0.6159,
      "step": 16306
    },
    {
      "epoch": 5.149036943479634,
      "grad_norm": 4.2271294593811035,
      "learning_rate": 4.125267515700102e-05,
      "loss": 0.6532,
      "step": 16307
    },
    {
      "epoch": 5.149352699715819,
      "grad_norm": 3.1939077377319336,
      "learning_rate": 4.125179805634495e-05,
      "loss": 0.5465,
      "step": 16308
    },
    {
      "epoch": 5.149668455952005,
      "grad_norm": 5.275920867919922,
      "learning_rate": 4.125092095568887e-05,
      "loss": 0.6704,
      "step": 16309
    },
    {
      "epoch": 5.1499842121881905,
      "grad_norm": 4.160173416137695,
      "learning_rate": 4.1250043855032805e-05,
      "loss": 0.5788,
      "step": 16310
    },
    {
      "epoch": 5.150299968424377,
      "grad_norm": 4.155157566070557,
      "learning_rate": 4.1249166754376737e-05,
      "loss": 0.6513,
      "step": 16311
    },
    {
      "epoch": 5.150615724660562,
      "grad_norm": 4.887603282928467,
      "learning_rate": 4.124828965372066e-05,
      "loss": 0.5715,
      "step": 16312
    },
    {
      "epoch": 5.150931480896748,
      "grad_norm": 3.7344167232513428,
      "learning_rate": 4.124741255306459e-05,
      "loss": 0.749,
      "step": 16313
    },
    {
      "epoch": 5.151247237132933,
      "grad_norm": 5.360400199890137,
      "learning_rate": 4.124653545240852e-05,
      "loss": 0.5676,
      "step": 16314
    },
    {
      "epoch": 5.151562993369119,
      "grad_norm": 4.668440818786621,
      "learning_rate": 4.124565835175245e-05,
      "loss": 0.6689,
      "step": 16315
    },
    {
      "epoch": 5.151878749605305,
      "grad_norm": 4.402480125427246,
      "learning_rate": 4.1244781251096375e-05,
      "loss": 0.5755,
      "step": 16316
    },
    {
      "epoch": 5.15219450584149,
      "grad_norm": 3.7920022010803223,
      "learning_rate": 4.1243904150440306e-05,
      "loss": 0.7183,
      "step": 16317
    },
    {
      "epoch": 5.152510262077676,
      "grad_norm": 4.251483917236328,
      "learning_rate": 4.124302704978423e-05,
      "loss": 0.7187,
      "step": 16318
    },
    {
      "epoch": 5.152826018313862,
      "grad_norm": 4.525269031524658,
      "learning_rate": 4.124214994912816e-05,
      "loss": 0.5763,
      "step": 16319
    },
    {
      "epoch": 5.153141774550047,
      "grad_norm": 4.4178595542907715,
      "learning_rate": 4.124127284847209e-05,
      "loss": 0.6015,
      "step": 16320
    },
    {
      "epoch": 5.153457530786233,
      "grad_norm": 4.34478235244751,
      "learning_rate": 4.1240395747816026e-05,
      "loss": 0.763,
      "step": 16321
    },
    {
      "epoch": 5.153773287022418,
      "grad_norm": 4.630421161651611,
      "learning_rate": 4.123951864715995e-05,
      "loss": 0.5879,
      "step": 16322
    },
    {
      "epoch": 5.154089043258605,
      "grad_norm": 5.137461185455322,
      "learning_rate": 4.123864154650388e-05,
      "loss": 0.6267,
      "step": 16323
    },
    {
      "epoch": 5.15440479949479,
      "grad_norm": 4.190631866455078,
      "learning_rate": 4.123776444584781e-05,
      "loss": 0.6294,
      "step": 16324
    },
    {
      "epoch": 5.154720555730976,
      "grad_norm": 4.579561233520508,
      "learning_rate": 4.123688734519174e-05,
      "loss": 0.6302,
      "step": 16325
    },
    {
      "epoch": 5.155036311967161,
      "grad_norm": 3.8288540840148926,
      "learning_rate": 4.1236010244535664e-05,
      "loss": 0.6772,
      "step": 16326
    },
    {
      "epoch": 5.155352068203347,
      "grad_norm": 4.263469696044922,
      "learning_rate": 4.1235133143879596e-05,
      "loss": 0.5956,
      "step": 16327
    },
    {
      "epoch": 5.155667824439533,
      "grad_norm": 4.5916032791137695,
      "learning_rate": 4.123425604322352e-05,
      "loss": 0.6425,
      "step": 16328
    },
    {
      "epoch": 5.155983580675718,
      "grad_norm": 4.529414653778076,
      "learning_rate": 4.123337894256745e-05,
      "loss": 0.7004,
      "step": 16329
    },
    {
      "epoch": 5.156299336911904,
      "grad_norm": 4.2314653396606445,
      "learning_rate": 4.123250184191138e-05,
      "loss": 0.7123,
      "step": 16330
    },
    {
      "epoch": 5.15661509314809,
      "grad_norm": 5.74763298034668,
      "learning_rate": 4.123162474125531e-05,
      "loss": 0.642,
      "step": 16331
    },
    {
      "epoch": 5.156930849384275,
      "grad_norm": 3.814164161682129,
      "learning_rate": 4.123074764059924e-05,
      "loss": 0.6196,
      "step": 16332
    },
    {
      "epoch": 5.157246605620461,
      "grad_norm": 3.6761281490325928,
      "learning_rate": 4.1229870539943166e-05,
      "loss": 0.5835,
      "step": 16333
    },
    {
      "epoch": 5.157562361856646,
      "grad_norm": 4.499275207519531,
      "learning_rate": 4.12289934392871e-05,
      "loss": 0.5513,
      "step": 16334
    },
    {
      "epoch": 5.157878118092833,
      "grad_norm": 3.961434841156006,
      "learning_rate": 4.122811633863102e-05,
      "loss": 0.5606,
      "step": 16335
    },
    {
      "epoch": 5.158193874329018,
      "grad_norm": 3.6104509830474854,
      "learning_rate": 4.1227239237974954e-05,
      "loss": 0.6054,
      "step": 16336
    },
    {
      "epoch": 5.158509630565204,
      "grad_norm": 5.447559833526611,
      "learning_rate": 4.122636213731888e-05,
      "loss": 0.6216,
      "step": 16337
    },
    {
      "epoch": 5.158825386801389,
      "grad_norm": 4.329579830169678,
      "learning_rate": 4.122548503666281e-05,
      "loss": 0.6318,
      "step": 16338
    },
    {
      "epoch": 5.159141143037575,
      "grad_norm": 4.454065799713135,
      "learning_rate": 4.1224607936006736e-05,
      "loss": 0.713,
      "step": 16339
    },
    {
      "epoch": 5.159456899273761,
      "grad_norm": 3.5196897983551025,
      "learning_rate": 4.122373083535067e-05,
      "loss": 0.5528,
      "step": 16340
    },
    {
      "epoch": 5.159772655509946,
      "grad_norm": 3.694305658340454,
      "learning_rate": 4.122285373469459e-05,
      "loss": 0.6518,
      "step": 16341
    },
    {
      "epoch": 5.160088411746132,
      "grad_norm": 4.175647258758545,
      "learning_rate": 4.1221976634038524e-05,
      "loss": 0.5918,
      "step": 16342
    },
    {
      "epoch": 5.160404167982318,
      "grad_norm": 3.857361078262329,
      "learning_rate": 4.122109953338245e-05,
      "loss": 0.5899,
      "step": 16343
    },
    {
      "epoch": 5.160719924218503,
      "grad_norm": 3.9082164764404297,
      "learning_rate": 4.122022243272638e-05,
      "loss": 0.6285,
      "step": 16344
    },
    {
      "epoch": 5.161035680454689,
      "grad_norm": 4.351558208465576,
      "learning_rate": 4.1219345332070305e-05,
      "loss": 0.6218,
      "step": 16345
    },
    {
      "epoch": 5.161351436690874,
      "grad_norm": 3.9791250228881836,
      "learning_rate": 4.121846823141424e-05,
      "loss": 0.6965,
      "step": 16346
    },
    {
      "epoch": 5.1616671929270606,
      "grad_norm": 4.130126953125,
      "learning_rate": 4.121759113075816e-05,
      "loss": 0.6316,
      "step": 16347
    },
    {
      "epoch": 5.161982949163246,
      "grad_norm": 3.3699920177459717,
      "learning_rate": 4.1216714030102094e-05,
      "loss": 0.5642,
      "step": 16348
    },
    {
      "epoch": 5.162298705399432,
      "grad_norm": 4.527090072631836,
      "learning_rate": 4.1215836929446025e-05,
      "loss": 0.6053,
      "step": 16349
    },
    {
      "epoch": 5.162614461635617,
      "grad_norm": 4.54735803604126,
      "learning_rate": 4.121495982878996e-05,
      "loss": 0.6108,
      "step": 16350
    },
    {
      "epoch": 5.162930217871803,
      "grad_norm": 4.563843250274658,
      "learning_rate": 4.121408272813388e-05,
      "loss": 0.6287,
      "step": 16351
    },
    {
      "epoch": 5.163245974107989,
      "grad_norm": 3.6967947483062744,
      "learning_rate": 4.1213205627477814e-05,
      "loss": 0.5362,
      "step": 16352
    },
    {
      "epoch": 5.163561730344174,
      "grad_norm": 3.7856547832489014,
      "learning_rate": 4.121232852682174e-05,
      "loss": 0.5737,
      "step": 16353
    },
    {
      "epoch": 5.16387748658036,
      "grad_norm": 3.6932449340820312,
      "learning_rate": 4.121145142616567e-05,
      "loss": 0.615,
      "step": 16354
    },
    {
      "epoch": 5.164193242816546,
      "grad_norm": 4.822250843048096,
      "learning_rate": 4.12105743255096e-05,
      "loss": 0.6075,
      "step": 16355
    },
    {
      "epoch": 5.164508999052731,
      "grad_norm": 4.247700214385986,
      "learning_rate": 4.120969722485353e-05,
      "loss": 0.6211,
      "step": 16356
    },
    {
      "epoch": 5.164824755288917,
      "grad_norm": 4.299989223480225,
      "learning_rate": 4.120882012419746e-05,
      "loss": 0.6019,
      "step": 16357
    },
    {
      "epoch": 5.165140511525102,
      "grad_norm": 4.161180019378662,
      "learning_rate": 4.120794302354138e-05,
      "loss": 0.5431,
      "step": 16358
    },
    {
      "epoch": 5.1654562677612885,
      "grad_norm": 3.57777738571167,
      "learning_rate": 4.1207065922885315e-05,
      "loss": 0.5053,
      "step": 16359
    },
    {
      "epoch": 5.165772023997474,
      "grad_norm": 4.610328197479248,
      "learning_rate": 4.120618882222924e-05,
      "loss": 0.5804,
      "step": 16360
    },
    {
      "epoch": 5.16608778023366,
      "grad_norm": 3.557145118713379,
      "learning_rate": 4.120531172157317e-05,
      "loss": 0.6856,
      "step": 16361
    },
    {
      "epoch": 5.166403536469845,
      "grad_norm": 4.350365161895752,
      "learning_rate": 4.1204434620917097e-05,
      "loss": 0.7612,
      "step": 16362
    },
    {
      "epoch": 5.166719292706031,
      "grad_norm": 4.452271938323975,
      "learning_rate": 4.120355752026103e-05,
      "loss": 0.6543,
      "step": 16363
    },
    {
      "epoch": 5.167035048942217,
      "grad_norm": 4.217803478240967,
      "learning_rate": 4.120268041960495e-05,
      "loss": 0.6136,
      "step": 16364
    },
    {
      "epoch": 5.167350805178402,
      "grad_norm": 4.312891960144043,
      "learning_rate": 4.1201803318948885e-05,
      "loss": 0.6019,
      "step": 16365
    },
    {
      "epoch": 5.167666561414588,
      "grad_norm": 4.1008992195129395,
      "learning_rate": 4.120092621829281e-05,
      "loss": 0.5873,
      "step": 16366
    },
    {
      "epoch": 5.1679823176507735,
      "grad_norm": 5.0898847579956055,
      "learning_rate": 4.120004911763674e-05,
      "loss": 0.6642,
      "step": 16367
    },
    {
      "epoch": 5.168298073886959,
      "grad_norm": 3.7935829162597656,
      "learning_rate": 4.1199172016980666e-05,
      "loss": 0.5872,
      "step": 16368
    },
    {
      "epoch": 5.168613830123145,
      "grad_norm": 4.372589111328125,
      "learning_rate": 4.11982949163246e-05,
      "loss": 0.6809,
      "step": 16369
    },
    {
      "epoch": 5.16892958635933,
      "grad_norm": 3.498983860015869,
      "learning_rate": 4.119741781566853e-05,
      "loss": 0.6303,
      "step": 16370
    },
    {
      "epoch": 5.1692453425955165,
      "grad_norm": 3.74419903755188,
      "learning_rate": 4.1196540715012455e-05,
      "loss": 0.6818,
      "step": 16371
    },
    {
      "epoch": 5.169561098831702,
      "grad_norm": 5.218081474304199,
      "learning_rate": 4.1195663614356386e-05,
      "loss": 0.7347,
      "step": 16372
    },
    {
      "epoch": 5.169876855067888,
      "grad_norm": 4.482483386993408,
      "learning_rate": 4.119478651370031e-05,
      "loss": 0.5758,
      "step": 16373
    },
    {
      "epoch": 5.170192611304073,
      "grad_norm": 4.47203254699707,
      "learning_rate": 4.119390941304424e-05,
      "loss": 0.6515,
      "step": 16374
    },
    {
      "epoch": 5.1705083675402586,
      "grad_norm": 4.1167988777160645,
      "learning_rate": 4.119303231238817e-05,
      "loss": 0.5982,
      "step": 16375
    },
    {
      "epoch": 5.170824123776445,
      "grad_norm": 3.907465934753418,
      "learning_rate": 4.1192155211732106e-05,
      "loss": 0.67,
      "step": 16376
    },
    {
      "epoch": 5.17113988001263,
      "grad_norm": 3.3281877040863037,
      "learning_rate": 4.119127811107603e-05,
      "loss": 0.5793,
      "step": 16377
    },
    {
      "epoch": 5.171455636248816,
      "grad_norm": 3.885503053665161,
      "learning_rate": 4.119040101041996e-05,
      "loss": 0.6842,
      "step": 16378
    },
    {
      "epoch": 5.1717713924850015,
      "grad_norm": 4.3706254959106445,
      "learning_rate": 4.118952390976389e-05,
      "loss": 0.6599,
      "step": 16379
    },
    {
      "epoch": 5.172087148721188,
      "grad_norm": 4.787045955657959,
      "learning_rate": 4.118864680910782e-05,
      "loss": 0.6316,
      "step": 16380
    },
    {
      "epoch": 5.172402904957373,
      "grad_norm": 3.642685651779175,
      "learning_rate": 4.1187769708451744e-05,
      "loss": 0.597,
      "step": 16381
    },
    {
      "epoch": 5.172718661193558,
      "grad_norm": 3.422921657562256,
      "learning_rate": 4.1186892607795676e-05,
      "loss": 0.7217,
      "step": 16382
    },
    {
      "epoch": 5.1730344174297445,
      "grad_norm": 3.457763910293579,
      "learning_rate": 4.11860155071396e-05,
      "loss": 0.6264,
      "step": 16383
    },
    {
      "epoch": 5.17335017366593,
      "grad_norm": 3.5780932903289795,
      "learning_rate": 4.118513840648353e-05,
      "loss": 0.5448,
      "step": 16384
    },
    {
      "epoch": 5.173665929902116,
      "grad_norm": 3.3881871700286865,
      "learning_rate": 4.118426130582746e-05,
      "loss": 0.5233,
      "step": 16385
    },
    {
      "epoch": 5.173981686138301,
      "grad_norm": 3.755453586578369,
      "learning_rate": 4.118338420517139e-05,
      "loss": 0.606,
      "step": 16386
    },
    {
      "epoch": 5.1742974423744865,
      "grad_norm": 3.9262609481811523,
      "learning_rate": 4.1182507104515314e-05,
      "loss": 0.6969,
      "step": 16387
    },
    {
      "epoch": 5.174613198610673,
      "grad_norm": 4.137165546417236,
      "learning_rate": 4.1181630003859246e-05,
      "loss": 0.6287,
      "step": 16388
    },
    {
      "epoch": 5.174928954846858,
      "grad_norm": 4.571624755859375,
      "learning_rate": 4.118075290320317e-05,
      "loss": 0.5246,
      "step": 16389
    },
    {
      "epoch": 5.175244711083044,
      "grad_norm": 3.8684606552124023,
      "learning_rate": 4.11798758025471e-05,
      "loss": 0.6479,
      "step": 16390
    },
    {
      "epoch": 5.1755604673192295,
      "grad_norm": 3.787956476211548,
      "learning_rate": 4.1178998701891034e-05,
      "loss": 0.5737,
      "step": 16391
    },
    {
      "epoch": 5.175876223555415,
      "grad_norm": 4.57737922668457,
      "learning_rate": 4.117812160123496e-05,
      "loss": 0.6404,
      "step": 16392
    },
    {
      "epoch": 5.176191979791601,
      "grad_norm": 3.930159330368042,
      "learning_rate": 4.117724450057889e-05,
      "loss": 0.623,
      "step": 16393
    },
    {
      "epoch": 5.176507736027786,
      "grad_norm": 5.730063438415527,
      "learning_rate": 4.1176367399922815e-05,
      "loss": 0.6447,
      "step": 16394
    },
    {
      "epoch": 5.176823492263972,
      "grad_norm": 4.370270729064941,
      "learning_rate": 4.117549029926675e-05,
      "loss": 0.6032,
      "step": 16395
    },
    {
      "epoch": 5.177139248500158,
      "grad_norm": 4.633133411407471,
      "learning_rate": 4.117461319861067e-05,
      "loss": 0.7198,
      "step": 16396
    },
    {
      "epoch": 5.177455004736344,
      "grad_norm": 3.167816400527954,
      "learning_rate": 4.1173736097954604e-05,
      "loss": 0.5738,
      "step": 16397
    },
    {
      "epoch": 5.177770760972529,
      "grad_norm": 4.562131404876709,
      "learning_rate": 4.117285899729853e-05,
      "loss": 0.7024,
      "step": 16398
    },
    {
      "epoch": 5.1780865172087145,
      "grad_norm": 4.22726583480835,
      "learning_rate": 4.117198189664246e-05,
      "loss": 0.633,
      "step": 16399
    },
    {
      "epoch": 5.178402273444901,
      "grad_norm": 4.556908130645752,
      "learning_rate": 4.1171104795986385e-05,
      "loss": 0.7651,
      "step": 16400
    },
    {
      "epoch": 5.178402273444901,
      "eval_loss": 0.8521743416786194,
      "eval_runtime": 4.5221,
      "eval_samples_per_second": 550.628,
      "eval_steps_per_second": 4.644,
      "step": 16400
    },
    {
      "epoch": 5.178718029681086,
      "grad_norm": 4.799262523651123,
      "learning_rate": 4.117022769533032e-05,
      "loss": 0.6828,
      "step": 16401
    },
    {
      "epoch": 5.179033785917272,
      "grad_norm": 3.862684726715088,
      "learning_rate": 4.116935059467424e-05,
      "loss": 0.6281,
      "step": 16402
    },
    {
      "epoch": 5.179349542153457,
      "grad_norm": 3.697821855545044,
      "learning_rate": 4.116847349401818e-05,
      "loss": 0.587,
      "step": 16403
    },
    {
      "epoch": 5.179665298389644,
      "grad_norm": 3.125586986541748,
      "learning_rate": 4.1167596393362105e-05,
      "loss": 0.6333,
      "step": 16404
    },
    {
      "epoch": 5.179981054625829,
      "grad_norm": 4.016829967498779,
      "learning_rate": 4.116671929270604e-05,
      "loss": 0.6801,
      "step": 16405
    },
    {
      "epoch": 5.180296810862014,
      "grad_norm": 4.671111106872559,
      "learning_rate": 4.116584219204996e-05,
      "loss": 0.6537,
      "step": 16406
    },
    {
      "epoch": 5.1806125670982,
      "grad_norm": 4.628978729248047,
      "learning_rate": 4.1164965091393893e-05,
      "loss": 0.7118,
      "step": 16407
    },
    {
      "epoch": 5.180928323334386,
      "grad_norm": 3.7203822135925293,
      "learning_rate": 4.116408799073782e-05,
      "loss": 0.6173,
      "step": 16408
    },
    {
      "epoch": 5.181244079570572,
      "grad_norm": 3.542369842529297,
      "learning_rate": 4.116321089008175e-05,
      "loss": 0.6157,
      "step": 16409
    },
    {
      "epoch": 5.181559835806757,
      "grad_norm": 3.9797658920288086,
      "learning_rate": 4.1162333789425675e-05,
      "loss": 0.6087,
      "step": 16410
    },
    {
      "epoch": 5.1818755920429425,
      "grad_norm": 4.20253849029541,
      "learning_rate": 4.116145668876961e-05,
      "loss": 0.5273,
      "step": 16411
    },
    {
      "epoch": 5.182191348279129,
      "grad_norm": 4.171674728393555,
      "learning_rate": 4.116057958811353e-05,
      "loss": 0.7322,
      "step": 16412
    },
    {
      "epoch": 5.182507104515314,
      "grad_norm": 3.6984214782714844,
      "learning_rate": 4.115970248745746e-05,
      "loss": 0.6313,
      "step": 16413
    },
    {
      "epoch": 5.1828228607515,
      "grad_norm": 3.6592626571655273,
      "learning_rate": 4.1158825386801395e-05,
      "loss": 0.6829,
      "step": 16414
    },
    {
      "epoch": 5.183138616987685,
      "grad_norm": 3.4703681468963623,
      "learning_rate": 4.115794828614532e-05,
      "loss": 0.6162,
      "step": 16415
    },
    {
      "epoch": 5.183454373223872,
      "grad_norm": 3.7799532413482666,
      "learning_rate": 4.115707118548925e-05,
      "loss": 0.613,
      "step": 16416
    },
    {
      "epoch": 5.183770129460057,
      "grad_norm": 4.040790557861328,
      "learning_rate": 4.1156194084833176e-05,
      "loss": 0.7116,
      "step": 16417
    },
    {
      "epoch": 5.184085885696242,
      "grad_norm": 3.8208558559417725,
      "learning_rate": 4.115531698417711e-05,
      "loss": 0.6454,
      "step": 16418
    },
    {
      "epoch": 5.184401641932428,
      "grad_norm": 3.5075995922088623,
      "learning_rate": 4.115443988352103e-05,
      "loss": 0.6007,
      "step": 16419
    },
    {
      "epoch": 5.184717398168614,
      "grad_norm": 3.190035343170166,
      "learning_rate": 4.1153562782864965e-05,
      "loss": 0.5717,
      "step": 16420
    },
    {
      "epoch": 5.1850331544048,
      "grad_norm": 3.091794013977051,
      "learning_rate": 4.115268568220889e-05,
      "loss": 0.5554,
      "step": 16421
    },
    {
      "epoch": 5.185348910640985,
      "grad_norm": 3.3202133178710938,
      "learning_rate": 4.115180858155282e-05,
      "loss": 0.6225,
      "step": 16422
    },
    {
      "epoch": 5.18566466687717,
      "grad_norm": 3.7921056747436523,
      "learning_rate": 4.1150931480896746e-05,
      "loss": 0.627,
      "step": 16423
    },
    {
      "epoch": 5.185980423113357,
      "grad_norm": 3.8686068058013916,
      "learning_rate": 4.115005438024068e-05,
      "loss": 0.6655,
      "step": 16424
    },
    {
      "epoch": 5.186296179349542,
      "grad_norm": 3.6822471618652344,
      "learning_rate": 4.11491772795846e-05,
      "loss": 0.6564,
      "step": 16425
    },
    {
      "epoch": 5.186611935585728,
      "grad_norm": 4.179821968078613,
      "learning_rate": 4.1148300178928534e-05,
      "loss": 0.6475,
      "step": 16426
    },
    {
      "epoch": 5.186927691821913,
      "grad_norm": 4.5358076095581055,
      "learning_rate": 4.114742307827246e-05,
      "loss": 0.6605,
      "step": 16427
    },
    {
      "epoch": 5.1872434480581,
      "grad_norm": 4.42506742477417,
      "learning_rate": 4.114654597761639e-05,
      "loss": 0.6759,
      "step": 16428
    },
    {
      "epoch": 5.187559204294285,
      "grad_norm": 4.348453044891357,
      "learning_rate": 4.114566887696032e-05,
      "loss": 0.6971,
      "step": 16429
    },
    {
      "epoch": 5.18787496053047,
      "grad_norm": 3.694962739944458,
      "learning_rate": 4.1144791776304254e-05,
      "loss": 0.5554,
      "step": 16430
    },
    {
      "epoch": 5.188190716766656,
      "grad_norm": 4.278334140777588,
      "learning_rate": 4.114391467564818e-05,
      "loss": 0.6194,
      "step": 16431
    },
    {
      "epoch": 5.188506473002842,
      "grad_norm": 4.032625675201416,
      "learning_rate": 4.114303757499211e-05,
      "loss": 0.6389,
      "step": 16432
    },
    {
      "epoch": 5.188822229239028,
      "grad_norm": 3.0321500301361084,
      "learning_rate": 4.1142160474336036e-05,
      "loss": 0.6386,
      "step": 16433
    },
    {
      "epoch": 5.189137985475213,
      "grad_norm": 3.4152164459228516,
      "learning_rate": 4.114128337367997e-05,
      "loss": 0.5965,
      "step": 16434
    },
    {
      "epoch": 5.189453741711398,
      "grad_norm": 5.148033618927002,
      "learning_rate": 4.11404062730239e-05,
      "loss": 0.6715,
      "step": 16435
    },
    {
      "epoch": 5.189769497947585,
      "grad_norm": 3.9449102878570557,
      "learning_rate": 4.1139529172367824e-05,
      "loss": 0.603,
      "step": 16436
    },
    {
      "epoch": 5.19008525418377,
      "grad_norm": 4.042813301086426,
      "learning_rate": 4.1138652071711756e-05,
      "loss": 0.642,
      "step": 16437
    },
    {
      "epoch": 5.190401010419956,
      "grad_norm": 3.939582586288452,
      "learning_rate": 4.113777497105568e-05,
      "loss": 0.7198,
      "step": 16438
    },
    {
      "epoch": 5.190716766656141,
      "grad_norm": 4.757686614990234,
      "learning_rate": 4.113689787039961e-05,
      "loss": 0.6989,
      "step": 16439
    },
    {
      "epoch": 5.1910325228923275,
      "grad_norm": 4.260440826416016,
      "learning_rate": 4.113602076974354e-05,
      "loss": 0.7236,
      "step": 16440
    },
    {
      "epoch": 5.191348279128513,
      "grad_norm": 3.3276710510253906,
      "learning_rate": 4.113514366908747e-05,
      "loss": 0.541,
      "step": 16441
    },
    {
      "epoch": 5.191664035364698,
      "grad_norm": 4.194816589355469,
      "learning_rate": 4.1134266568431394e-05,
      "loss": 0.7011,
      "step": 16442
    },
    {
      "epoch": 5.191979791600884,
      "grad_norm": 3.994600296020508,
      "learning_rate": 4.1133389467775326e-05,
      "loss": 0.589,
      "step": 16443
    },
    {
      "epoch": 5.19229554783707,
      "grad_norm": 4.878175735473633,
      "learning_rate": 4.113251236711925e-05,
      "loss": 0.5722,
      "step": 16444
    },
    {
      "epoch": 5.192611304073256,
      "grad_norm": 4.074060916900635,
      "learning_rate": 4.113163526646318e-05,
      "loss": 0.6892,
      "step": 16445
    },
    {
      "epoch": 5.192927060309441,
      "grad_norm": 4.455328464508057,
      "learning_rate": 4.113075816580711e-05,
      "loss": 0.7076,
      "step": 16446
    },
    {
      "epoch": 5.193242816545626,
      "grad_norm": 4.479741096496582,
      "learning_rate": 4.112988106515104e-05,
      "loss": 0.6819,
      "step": 16447
    },
    {
      "epoch": 5.1935585727818125,
      "grad_norm": 3.7801480293273926,
      "learning_rate": 4.1129003964494964e-05,
      "loss": 0.6964,
      "step": 16448
    },
    {
      "epoch": 5.193874329017998,
      "grad_norm": 4.882847309112549,
      "learning_rate": 4.1128126863838895e-05,
      "loss": 0.6933,
      "step": 16449
    },
    {
      "epoch": 5.194190085254184,
      "grad_norm": 3.2703640460968018,
      "learning_rate": 4.112724976318282e-05,
      "loss": 0.5691,
      "step": 16450
    },
    {
      "epoch": 5.194505841490369,
      "grad_norm": 4.4109907150268555,
      "learning_rate": 4.112637266252675e-05,
      "loss": 0.6806,
      "step": 16451
    },
    {
      "epoch": 5.1948215977265555,
      "grad_norm": 5.449517250061035,
      "learning_rate": 4.1125495561870684e-05,
      "loss": 0.6482,
      "step": 16452
    },
    {
      "epoch": 5.195137353962741,
      "grad_norm": 3.3397576808929443,
      "learning_rate": 4.112461846121461e-05,
      "loss": 0.6253,
      "step": 16453
    },
    {
      "epoch": 5.195453110198926,
      "grad_norm": 4.626138210296631,
      "learning_rate": 4.112374136055854e-05,
      "loss": 0.7062,
      "step": 16454
    },
    {
      "epoch": 5.195768866435112,
      "grad_norm": 4.176759243011475,
      "learning_rate": 4.1122864259902465e-05,
      "loss": 0.6562,
      "step": 16455
    },
    {
      "epoch": 5.196084622671298,
      "grad_norm": 3.730133295059204,
      "learning_rate": 4.11219871592464e-05,
      "loss": 0.6304,
      "step": 16456
    },
    {
      "epoch": 5.196400378907484,
      "grad_norm": 3.8393287658691406,
      "learning_rate": 4.112111005859032e-05,
      "loss": 0.7553,
      "step": 16457
    },
    {
      "epoch": 5.196716135143669,
      "grad_norm": 3.5517592430114746,
      "learning_rate": 4.112023295793426e-05,
      "loss": 0.5678,
      "step": 16458
    },
    {
      "epoch": 5.197031891379854,
      "grad_norm": 4.173737049102783,
      "learning_rate": 4.1119355857278185e-05,
      "loss": 0.6054,
      "step": 16459
    },
    {
      "epoch": 5.1973476476160405,
      "grad_norm": 4.093042850494385,
      "learning_rate": 4.111847875662212e-05,
      "loss": 0.6587,
      "step": 16460
    },
    {
      "epoch": 5.197663403852226,
      "grad_norm": 3.9892375469207764,
      "learning_rate": 4.111760165596604e-05,
      "loss": 0.7094,
      "step": 16461
    },
    {
      "epoch": 5.197979160088412,
      "grad_norm": 5.365293025970459,
      "learning_rate": 4.111672455530997e-05,
      "loss": 0.6754,
      "step": 16462
    },
    {
      "epoch": 5.198294916324597,
      "grad_norm": 3.802567481994629,
      "learning_rate": 4.11158474546539e-05,
      "loss": 0.635,
      "step": 16463
    },
    {
      "epoch": 5.1986106725607835,
      "grad_norm": 3.9668257236480713,
      "learning_rate": 4.111497035399783e-05,
      "loss": 0.6167,
      "step": 16464
    },
    {
      "epoch": 5.198926428796969,
      "grad_norm": 5.286586284637451,
      "learning_rate": 4.1114093253341755e-05,
      "loss": 0.6216,
      "step": 16465
    },
    {
      "epoch": 5.199242185033154,
      "grad_norm": 4.246708393096924,
      "learning_rate": 4.1113216152685687e-05,
      "loss": 0.675,
      "step": 16466
    },
    {
      "epoch": 5.19955794126934,
      "grad_norm": 4.181815147399902,
      "learning_rate": 4.111233905202961e-05,
      "loss": 0.7206,
      "step": 16467
    },
    {
      "epoch": 5.1998736975055255,
      "grad_norm": 4.116950988769531,
      "learning_rate": 4.111146195137354e-05,
      "loss": 0.5461,
      "step": 16468
    },
    {
      "epoch": 5.200189453741712,
      "grad_norm": 4.119166374206543,
      "learning_rate": 4.111058485071747e-05,
      "loss": 0.5825,
      "step": 16469
    },
    {
      "epoch": 5.200505209977897,
      "grad_norm": 3.949073314666748,
      "learning_rate": 4.11097077500614e-05,
      "loss": 0.5396,
      "step": 16470
    },
    {
      "epoch": 5.200820966214083,
      "grad_norm": 4.591485977172852,
      "learning_rate": 4.1108830649405325e-05,
      "loss": 0.5682,
      "step": 16471
    },
    {
      "epoch": 5.2011367224502685,
      "grad_norm": 5.043150901794434,
      "learning_rate": 4.1107953548749256e-05,
      "loss": 0.7954,
      "step": 16472
    },
    {
      "epoch": 5.201452478686454,
      "grad_norm": 4.9320969581604,
      "learning_rate": 4.110707644809319e-05,
      "loss": 0.685,
      "step": 16473
    },
    {
      "epoch": 5.20176823492264,
      "grad_norm": 4.453166484832764,
      "learning_rate": 4.110619934743711e-05,
      "loss": 0.7103,
      "step": 16474
    },
    {
      "epoch": 5.202083991158825,
      "grad_norm": 5.799764633178711,
      "learning_rate": 4.1105322246781045e-05,
      "loss": 0.6529,
      "step": 16475
    },
    {
      "epoch": 5.202399747395011,
      "grad_norm": 3.8560950756073,
      "learning_rate": 4.110444514612497e-05,
      "loss": 0.6685,
      "step": 16476
    },
    {
      "epoch": 5.202715503631197,
      "grad_norm": 3.9405088424682617,
      "learning_rate": 4.11035680454689e-05,
      "loss": 0.6347,
      "step": 16477
    },
    {
      "epoch": 5.203031259867382,
      "grad_norm": 4.3499579429626465,
      "learning_rate": 4.1102690944812826e-05,
      "loss": 0.6575,
      "step": 16478
    },
    {
      "epoch": 5.203347016103568,
      "grad_norm": 3.7652478218078613,
      "learning_rate": 4.110181384415676e-05,
      "loss": 0.7315,
      "step": 16479
    },
    {
      "epoch": 5.2036627723397535,
      "grad_norm": 3.39768123626709,
      "learning_rate": 4.110093674350068e-05,
      "loss": 0.5796,
      "step": 16480
    },
    {
      "epoch": 5.20397852857594,
      "grad_norm": 3.498607635498047,
      "learning_rate": 4.1100059642844614e-05,
      "loss": 0.576,
      "step": 16481
    },
    {
      "epoch": 5.204294284812125,
      "grad_norm": 4.193938732147217,
      "learning_rate": 4.109918254218854e-05,
      "loss": 0.6313,
      "step": 16482
    },
    {
      "epoch": 5.20461004104831,
      "grad_norm": 4.949453353881836,
      "learning_rate": 4.109830544153247e-05,
      "loss": 0.628,
      "step": 16483
    },
    {
      "epoch": 5.2049257972844964,
      "grad_norm": 4.169276237487793,
      "learning_rate": 4.1097428340876396e-05,
      "loss": 0.7108,
      "step": 16484
    },
    {
      "epoch": 5.205241553520682,
      "grad_norm": 4.632800102233887,
      "learning_rate": 4.1096551240220334e-05,
      "loss": 0.6638,
      "step": 16485
    },
    {
      "epoch": 5.205557309756868,
      "grad_norm": 4.994633674621582,
      "learning_rate": 4.109567413956426e-05,
      "loss": 0.6591,
      "step": 16486
    },
    {
      "epoch": 5.205873065993053,
      "grad_norm": 4.401902675628662,
      "learning_rate": 4.109479703890819e-05,
      "loss": 0.705,
      "step": 16487
    },
    {
      "epoch": 5.206188822229239,
      "grad_norm": 4.061482906341553,
      "learning_rate": 4.1093919938252116e-05,
      "loss": 0.63,
      "step": 16488
    },
    {
      "epoch": 5.206504578465425,
      "grad_norm": 3.8445849418640137,
      "learning_rate": 4.109304283759605e-05,
      "loss": 0.6584,
      "step": 16489
    },
    {
      "epoch": 5.20682033470161,
      "grad_norm": 3.5856316089630127,
      "learning_rate": 4.109216573693997e-05,
      "loss": 0.5972,
      "step": 16490
    },
    {
      "epoch": 5.207136090937796,
      "grad_norm": 4.517330169677734,
      "learning_rate": 4.1091288636283904e-05,
      "loss": 0.6573,
      "step": 16491
    },
    {
      "epoch": 5.2074518471739815,
      "grad_norm": 3.7370471954345703,
      "learning_rate": 4.109041153562783e-05,
      "loss": 0.6327,
      "step": 16492
    },
    {
      "epoch": 5.207767603410168,
      "grad_norm": 4.296517372131348,
      "learning_rate": 4.108953443497176e-05,
      "loss": 0.6923,
      "step": 16493
    },
    {
      "epoch": 5.208083359646353,
      "grad_norm": 3.2317986488342285,
      "learning_rate": 4.1088657334315686e-05,
      "loss": 0.5413,
      "step": 16494
    },
    {
      "epoch": 5.208399115882539,
      "grad_norm": 3.5874242782592773,
      "learning_rate": 4.108778023365962e-05,
      "loss": 0.626,
      "step": 16495
    },
    {
      "epoch": 5.208714872118724,
      "grad_norm": 3.616105794906616,
      "learning_rate": 4.108690313300355e-05,
      "loss": 0.7253,
      "step": 16496
    },
    {
      "epoch": 5.20903062835491,
      "grad_norm": 3.6187186241149902,
      "learning_rate": 4.1086026032347474e-05,
      "loss": 0.6772,
      "step": 16497
    },
    {
      "epoch": 5.209346384591096,
      "grad_norm": 4.014752388000488,
      "learning_rate": 4.1085148931691405e-05,
      "loss": 0.5951,
      "step": 16498
    },
    {
      "epoch": 5.209662140827281,
      "grad_norm": 4.597178936004639,
      "learning_rate": 4.108427183103533e-05,
      "loss": 0.6806,
      "step": 16499
    },
    {
      "epoch": 5.209977897063467,
      "grad_norm": 4.311672210693359,
      "learning_rate": 4.108339473037926e-05,
      "loss": 0.6522,
      "step": 16500
    },
    {
      "epoch": 5.210293653299653,
      "grad_norm": 4.673101902008057,
      "learning_rate": 4.108251762972319e-05,
      "loss": 0.7668,
      "step": 16501
    },
    {
      "epoch": 5.210609409535838,
      "grad_norm": 4.072361469268799,
      "learning_rate": 4.108164052906712e-05,
      "loss": 0.6297,
      "step": 16502
    },
    {
      "epoch": 5.210925165772024,
      "grad_norm": 4.146743297576904,
      "learning_rate": 4.1080763428411044e-05,
      "loss": 0.6993,
      "step": 16503
    },
    {
      "epoch": 5.211240922008209,
      "grad_norm": 4.153726100921631,
      "learning_rate": 4.1079886327754975e-05,
      "loss": 0.5811,
      "step": 16504
    },
    {
      "epoch": 5.211556678244396,
      "grad_norm": 3.6006171703338623,
      "learning_rate": 4.10790092270989e-05,
      "loss": 0.533,
      "step": 16505
    },
    {
      "epoch": 5.211872434480581,
      "grad_norm": 3.587662935256958,
      "learning_rate": 4.107813212644283e-05,
      "loss": 0.511,
      "step": 16506
    },
    {
      "epoch": 5.212188190716766,
      "grad_norm": 4.57127046585083,
      "learning_rate": 4.107725502578676e-05,
      "loss": 0.802,
      "step": 16507
    },
    {
      "epoch": 5.212503946952952,
      "grad_norm": 4.341665744781494,
      "learning_rate": 4.107637792513069e-05,
      "loss": 0.7113,
      "step": 16508
    },
    {
      "epoch": 5.212819703189138,
      "grad_norm": 4.00150728225708,
      "learning_rate": 4.107550082447461e-05,
      "loss": 0.5767,
      "step": 16509
    },
    {
      "epoch": 5.213135459425324,
      "grad_norm": 5.224425315856934,
      "learning_rate": 4.1074623723818545e-05,
      "loss": 0.6767,
      "step": 16510
    },
    {
      "epoch": 5.213451215661509,
      "grad_norm": 4.2606916427612305,
      "learning_rate": 4.107374662316248e-05,
      "loss": 0.711,
      "step": 16511
    },
    {
      "epoch": 5.213766971897695,
      "grad_norm": 4.946321487426758,
      "learning_rate": 4.107286952250641e-05,
      "loss": 0.7318,
      "step": 16512
    },
    {
      "epoch": 5.214082728133881,
      "grad_norm": 4.2892913818359375,
      "learning_rate": 4.107199242185033e-05,
      "loss": 0.6996,
      "step": 16513
    },
    {
      "epoch": 5.214398484370066,
      "grad_norm": 3.360934257507324,
      "learning_rate": 4.1071115321194265e-05,
      "loss": 0.5888,
      "step": 16514
    },
    {
      "epoch": 5.214714240606252,
      "grad_norm": 4.442338466644287,
      "learning_rate": 4.107023822053819e-05,
      "loss": 0.6839,
      "step": 16515
    },
    {
      "epoch": 5.215029996842437,
      "grad_norm": 4.595097064971924,
      "learning_rate": 4.106936111988212e-05,
      "loss": 0.6516,
      "step": 16516
    },
    {
      "epoch": 5.215345753078624,
      "grad_norm": 4.675769805908203,
      "learning_rate": 4.106848401922605e-05,
      "loss": 0.6494,
      "step": 16517
    },
    {
      "epoch": 5.215661509314809,
      "grad_norm": 3.554835557937622,
      "learning_rate": 4.106760691856998e-05,
      "loss": 0.6758,
      "step": 16518
    },
    {
      "epoch": 5.215977265550995,
      "grad_norm": 4.209473133087158,
      "learning_rate": 4.106672981791391e-05,
      "loss": 0.6586,
      "step": 16519
    },
    {
      "epoch": 5.21629302178718,
      "grad_norm": 4.495378017425537,
      "learning_rate": 4.1065852717257835e-05,
      "loss": 0.6877,
      "step": 16520
    },
    {
      "epoch": 5.216608778023366,
      "grad_norm": 3.4461658000946045,
      "learning_rate": 4.1064975616601766e-05,
      "loss": 0.603,
      "step": 16521
    },
    {
      "epoch": 5.216924534259552,
      "grad_norm": 4.572033405303955,
      "learning_rate": 4.106409851594569e-05,
      "loss": 0.616,
      "step": 16522
    },
    {
      "epoch": 5.217240290495737,
      "grad_norm": 4.346140384674072,
      "learning_rate": 4.106322141528962e-05,
      "loss": 0.6336,
      "step": 16523
    },
    {
      "epoch": 5.217556046731923,
      "grad_norm": 3.5511562824249268,
      "learning_rate": 4.106234431463355e-05,
      "loss": 0.6705,
      "step": 16524
    },
    {
      "epoch": 5.217871802968109,
      "grad_norm": 3.7809054851531982,
      "learning_rate": 4.106146721397748e-05,
      "loss": 0.6281,
      "step": 16525
    },
    {
      "epoch": 5.218187559204294,
      "grad_norm": 5.304078102111816,
      "learning_rate": 4.1060590113321405e-05,
      "loss": 0.6774,
      "step": 16526
    },
    {
      "epoch": 5.21850331544048,
      "grad_norm": 3.6713993549346924,
      "learning_rate": 4.1059713012665336e-05,
      "loss": 0.5907,
      "step": 16527
    },
    {
      "epoch": 5.218819071676665,
      "grad_norm": 3.525177001953125,
      "learning_rate": 4.105883591200926e-05,
      "loss": 0.6489,
      "step": 16528
    },
    {
      "epoch": 5.2191348279128515,
      "grad_norm": 3.9856338500976562,
      "learning_rate": 4.105795881135319e-05,
      "loss": 0.6795,
      "step": 16529
    },
    {
      "epoch": 5.219450584149037,
      "grad_norm": 3.5847671031951904,
      "learning_rate": 4.105708171069712e-05,
      "loss": 0.5958,
      "step": 16530
    },
    {
      "epoch": 5.219766340385223,
      "grad_norm": 3.6469881534576416,
      "learning_rate": 4.105620461004105e-05,
      "loss": 0.6575,
      "step": 16531
    },
    {
      "epoch": 5.220082096621408,
      "grad_norm": 4.136682987213135,
      "learning_rate": 4.1055327509384974e-05,
      "loss": 0.6765,
      "step": 16532
    },
    {
      "epoch": 5.220397852857594,
      "grad_norm": 4.142425537109375,
      "learning_rate": 4.1054450408728906e-05,
      "loss": 0.6223,
      "step": 16533
    },
    {
      "epoch": 5.22071360909378,
      "grad_norm": 3.574613332748413,
      "learning_rate": 4.105357330807284e-05,
      "loss": 0.5732,
      "step": 16534
    },
    {
      "epoch": 5.221029365329965,
      "grad_norm": 3.530374765396118,
      "learning_rate": 4.105269620741676e-05,
      "loss": 0.6077,
      "step": 16535
    },
    {
      "epoch": 5.221345121566151,
      "grad_norm": 4.30519437789917,
      "learning_rate": 4.1051819106760694e-05,
      "loss": 0.6165,
      "step": 16536
    },
    {
      "epoch": 5.221660877802337,
      "grad_norm": 4.231228351593018,
      "learning_rate": 4.105094200610462e-05,
      "loss": 0.6716,
      "step": 16537
    },
    {
      "epoch": 5.221976634038522,
      "grad_norm": 3.718984842300415,
      "learning_rate": 4.105006490544855e-05,
      "loss": 0.5611,
      "step": 16538
    },
    {
      "epoch": 5.222292390274708,
      "grad_norm": 4.757493019104004,
      "learning_rate": 4.104918780479248e-05,
      "loss": 0.6617,
      "step": 16539
    },
    {
      "epoch": 5.222608146510893,
      "grad_norm": 3.421104907989502,
      "learning_rate": 4.1048310704136414e-05,
      "loss": 0.6417,
      "step": 16540
    },
    {
      "epoch": 5.2229239027470795,
      "grad_norm": 3.4130122661590576,
      "learning_rate": 4.104743360348034e-05,
      "loss": 0.5671,
      "step": 16541
    },
    {
      "epoch": 5.223239658983265,
      "grad_norm": 3.508336305618286,
      "learning_rate": 4.104655650282427e-05,
      "loss": 0.5803,
      "step": 16542
    },
    {
      "epoch": 5.223555415219451,
      "grad_norm": 4.023892402648926,
      "learning_rate": 4.1045679402168196e-05,
      "loss": 0.6551,
      "step": 16543
    },
    {
      "epoch": 5.223871171455636,
      "grad_norm": 3.8374738693237305,
      "learning_rate": 4.104480230151213e-05,
      "loss": 0.6536,
      "step": 16544
    },
    {
      "epoch": 5.224186927691822,
      "grad_norm": 4.2203264236450195,
      "learning_rate": 4.104392520085605e-05,
      "loss": 0.5927,
      "step": 16545
    },
    {
      "epoch": 5.224502683928008,
      "grad_norm": 4.584901332855225,
      "learning_rate": 4.1043048100199984e-05,
      "loss": 0.5969,
      "step": 16546
    },
    {
      "epoch": 5.224818440164193,
      "grad_norm": 4.012618064880371,
      "learning_rate": 4.104217099954391e-05,
      "loss": 0.6098,
      "step": 16547
    },
    {
      "epoch": 5.225134196400379,
      "grad_norm": 7.405097484588623,
      "learning_rate": 4.104129389888784e-05,
      "loss": 0.7691,
      "step": 16548
    },
    {
      "epoch": 5.2254499526365645,
      "grad_norm": 4.121578216552734,
      "learning_rate": 4.1040416798231765e-05,
      "loss": 0.6824,
      "step": 16549
    },
    {
      "epoch": 5.22576570887275,
      "grad_norm": 3.2738699913024902,
      "learning_rate": 4.10395396975757e-05,
      "loss": 0.5587,
      "step": 16550
    },
    {
      "epoch": 5.226081465108936,
      "grad_norm": 4.830000877380371,
      "learning_rate": 4.103866259691962e-05,
      "loss": 0.6324,
      "step": 16551
    },
    {
      "epoch": 5.226397221345121,
      "grad_norm": 4.18807315826416,
      "learning_rate": 4.1037785496263554e-05,
      "loss": 0.6448,
      "step": 16552
    },
    {
      "epoch": 5.2267129775813075,
      "grad_norm": 3.9304468631744385,
      "learning_rate": 4.103690839560748e-05,
      "loss": 0.5523,
      "step": 16553
    },
    {
      "epoch": 5.227028733817493,
      "grad_norm": 4.1919474601745605,
      "learning_rate": 4.103603129495141e-05,
      "loss": 0.6692,
      "step": 16554
    },
    {
      "epoch": 5.227344490053679,
      "grad_norm": 4.431512355804443,
      "learning_rate": 4.103515419429534e-05,
      "loss": 0.6604,
      "step": 16555
    },
    {
      "epoch": 5.227660246289864,
      "grad_norm": 3.981093645095825,
      "learning_rate": 4.103427709363927e-05,
      "loss": 0.6331,
      "step": 16556
    },
    {
      "epoch": 5.2279760025260495,
      "grad_norm": 3.437697410583496,
      "learning_rate": 4.10333999929832e-05,
      "loss": 0.6823,
      "step": 16557
    },
    {
      "epoch": 5.228291758762236,
      "grad_norm": 4.3712005615234375,
      "learning_rate": 4.1032522892327123e-05,
      "loss": 0.6824,
      "step": 16558
    },
    {
      "epoch": 5.228607514998421,
      "grad_norm": 4.359858989715576,
      "learning_rate": 4.1031645791671055e-05,
      "loss": 0.6561,
      "step": 16559
    },
    {
      "epoch": 5.228923271234607,
      "grad_norm": 4.076058864593506,
      "learning_rate": 4.103076869101498e-05,
      "loss": 0.5927,
      "step": 16560
    },
    {
      "epoch": 5.2292390274707925,
      "grad_norm": 4.741964340209961,
      "learning_rate": 4.102989159035891e-05,
      "loss": 0.5754,
      "step": 16561
    },
    {
      "epoch": 5.229554783706978,
      "grad_norm": 6.540657997131348,
      "learning_rate": 4.102901448970284e-05,
      "loss": 0.6772,
      "step": 16562
    },
    {
      "epoch": 5.229870539943164,
      "grad_norm": 3.7265594005584717,
      "learning_rate": 4.102813738904677e-05,
      "loss": 0.5981,
      "step": 16563
    },
    {
      "epoch": 5.230186296179349,
      "grad_norm": 3.7010772228240967,
      "learning_rate": 4.102726028839069e-05,
      "loss": 0.5518,
      "step": 16564
    },
    {
      "epoch": 5.2305020524155355,
      "grad_norm": 4.787994861602783,
      "learning_rate": 4.1026383187734625e-05,
      "loss": 0.6387,
      "step": 16565
    },
    {
      "epoch": 5.230817808651721,
      "grad_norm": 5.666694641113281,
      "learning_rate": 4.102550608707855e-05,
      "loss": 0.5342,
      "step": 16566
    },
    {
      "epoch": 5.231133564887907,
      "grad_norm": 3.6581811904907227,
      "learning_rate": 4.102462898642249e-05,
      "loss": 0.5019,
      "step": 16567
    },
    {
      "epoch": 5.231449321124092,
      "grad_norm": 4.7255024909973145,
      "learning_rate": 4.102375188576641e-05,
      "loss": 0.6277,
      "step": 16568
    },
    {
      "epoch": 5.2317650773602775,
      "grad_norm": 4.768283843994141,
      "learning_rate": 4.1022874785110345e-05,
      "loss": 0.698,
      "step": 16569
    },
    {
      "epoch": 5.232080833596464,
      "grad_norm": 4.067939758300781,
      "learning_rate": 4.102199768445427e-05,
      "loss": 0.5661,
      "step": 16570
    },
    {
      "epoch": 5.232396589832649,
      "grad_norm": 4.073996543884277,
      "learning_rate": 4.10211205837982e-05,
      "loss": 0.5732,
      "step": 16571
    },
    {
      "epoch": 5.232712346068835,
      "grad_norm": 4.315242767333984,
      "learning_rate": 4.1020243483142126e-05,
      "loss": 0.7097,
      "step": 16572
    },
    {
      "epoch": 5.2330281023050205,
      "grad_norm": 4.750127792358398,
      "learning_rate": 4.101936638248606e-05,
      "loss": 0.621,
      "step": 16573
    },
    {
      "epoch": 5.233343858541206,
      "grad_norm": 4.6895222663879395,
      "learning_rate": 4.101848928182998e-05,
      "loss": 0.615,
      "step": 16574
    },
    {
      "epoch": 5.233659614777392,
      "grad_norm": 4.112518310546875,
      "learning_rate": 4.1017612181173915e-05,
      "loss": 0.6024,
      "step": 16575
    },
    {
      "epoch": 5.233975371013577,
      "grad_norm": 3.86777925491333,
      "learning_rate": 4.1016735080517846e-05,
      "loss": 0.5272,
      "step": 16576
    },
    {
      "epoch": 5.234291127249763,
      "grad_norm": 4.384182929992676,
      "learning_rate": 4.101585797986177e-05,
      "loss": 0.6562,
      "step": 16577
    },
    {
      "epoch": 5.234606883485949,
      "grad_norm": 4.2793779373168945,
      "learning_rate": 4.10149808792057e-05,
      "loss": 0.622,
      "step": 16578
    },
    {
      "epoch": 5.234922639722135,
      "grad_norm": 5.67141580581665,
      "learning_rate": 4.101410377854963e-05,
      "loss": 0.6832,
      "step": 16579
    },
    {
      "epoch": 5.23523839595832,
      "grad_norm": 5.283713340759277,
      "learning_rate": 4.101322667789356e-05,
      "loss": 0.6316,
      "step": 16580
    },
    {
      "epoch": 5.2355541521945055,
      "grad_norm": 3.7263779640197754,
      "learning_rate": 4.1012349577237484e-05,
      "loss": 0.6188,
      "step": 16581
    },
    {
      "epoch": 5.235869908430692,
      "grad_norm": 3.7652993202209473,
      "learning_rate": 4.1011472476581416e-05,
      "loss": 0.5894,
      "step": 16582
    },
    {
      "epoch": 5.236185664666877,
      "grad_norm": 4.171817302703857,
      "learning_rate": 4.101059537592534e-05,
      "loss": 0.6753,
      "step": 16583
    },
    {
      "epoch": 5.236501420903063,
      "grad_norm": 4.764773845672607,
      "learning_rate": 4.100971827526927e-05,
      "loss": 0.678,
      "step": 16584
    },
    {
      "epoch": 5.236817177139248,
      "grad_norm": 4.747404098510742,
      "learning_rate": 4.10088411746132e-05,
      "loss": 0.6252,
      "step": 16585
    },
    {
      "epoch": 5.237132933375435,
      "grad_norm": 5.620441913604736,
      "learning_rate": 4.100796407395713e-05,
      "loss": 0.7463,
      "step": 16586
    },
    {
      "epoch": 5.23744868961162,
      "grad_norm": 4.074005126953125,
      "learning_rate": 4.1007086973301054e-05,
      "loss": 0.6272,
      "step": 16587
    },
    {
      "epoch": 5.237764445847805,
      "grad_norm": 3.897285223007202,
      "learning_rate": 4.1006209872644986e-05,
      "loss": 0.5889,
      "step": 16588
    },
    {
      "epoch": 5.238080202083991,
      "grad_norm": 4.1850175857543945,
      "learning_rate": 4.100533277198891e-05,
      "loss": 0.651,
      "step": 16589
    },
    {
      "epoch": 5.238395958320177,
      "grad_norm": 3.5368263721466064,
      "learning_rate": 4.100445567133284e-05,
      "loss": 0.5803,
      "step": 16590
    },
    {
      "epoch": 5.238711714556363,
      "grad_norm": 4.7761454582214355,
      "learning_rate": 4.100357857067677e-05,
      "loss": 0.668,
      "step": 16591
    },
    {
      "epoch": 5.239027470792548,
      "grad_norm": 3.503082036972046,
      "learning_rate": 4.10027014700207e-05,
      "loss": 0.5307,
      "step": 16592
    },
    {
      "epoch": 5.2393432270287335,
      "grad_norm": 4.025042533874512,
      "learning_rate": 4.100182436936463e-05,
      "loss": 0.593,
      "step": 16593
    },
    {
      "epoch": 5.23965898326492,
      "grad_norm": 4.373779296875,
      "learning_rate": 4.100094726870856e-05,
      "loss": 0.5799,
      "step": 16594
    },
    {
      "epoch": 5.239974739501105,
      "grad_norm": 4.077778339385986,
      "learning_rate": 4.100007016805249e-05,
      "loss": 0.6925,
      "step": 16595
    },
    {
      "epoch": 5.240290495737291,
      "grad_norm": 4.2585859298706055,
      "learning_rate": 4.099919306739642e-05,
      "loss": 0.6329,
      "step": 16596
    },
    {
      "epoch": 5.240606251973476,
      "grad_norm": 4.740716934204102,
      "learning_rate": 4.0998315966740344e-05,
      "loss": 0.6657,
      "step": 16597
    },
    {
      "epoch": 5.240922008209662,
      "grad_norm": 5.158210754394531,
      "learning_rate": 4.0997438866084276e-05,
      "loss": 0.6676,
      "step": 16598
    },
    {
      "epoch": 5.241237764445848,
      "grad_norm": 3.5666863918304443,
      "learning_rate": 4.099656176542821e-05,
      "loss": 0.6326,
      "step": 16599
    },
    {
      "epoch": 5.241553520682033,
      "grad_norm": 3.8786461353302,
      "learning_rate": 4.099568466477213e-05,
      "loss": 0.6404,
      "step": 16600
    },
    {
      "epoch": 5.241553520682033,
      "eval_loss": 0.8433696627616882,
      "eval_runtime": 4.5539,
      "eval_samples_per_second": 546.778,
      "eval_steps_per_second": 4.611,
      "step": 16600
    },
    {
      "epoch": 5.241869276918219,
      "grad_norm": 3.8527867794036865,
      "learning_rate": 4.0994807564116064e-05,
      "loss": 0.7784,
      "step": 16601
    },
    {
      "epoch": 5.242185033154405,
      "grad_norm": 4.143565654754639,
      "learning_rate": 4.099393046345999e-05,
      "loss": 0.6051,
      "step": 16602
    },
    {
      "epoch": 5.242500789390591,
      "grad_norm": 3.6989080905914307,
      "learning_rate": 4.099305336280392e-05,
      "loss": 0.5704,
      "step": 16603
    },
    {
      "epoch": 5.242816545626776,
      "grad_norm": 3.808652639389038,
      "learning_rate": 4.0992176262147845e-05,
      "loss": 0.6308,
      "step": 16604
    },
    {
      "epoch": 5.243132301862961,
      "grad_norm": 3.9424550533294678,
      "learning_rate": 4.099129916149178e-05,
      "loss": 0.5738,
      "step": 16605
    },
    {
      "epoch": 5.243448058099148,
      "grad_norm": 3.992002487182617,
      "learning_rate": 4.09904220608357e-05,
      "loss": 0.6525,
      "step": 16606
    },
    {
      "epoch": 5.243763814335333,
      "grad_norm": 4.477336406707764,
      "learning_rate": 4.0989544960179634e-05,
      "loss": 0.7463,
      "step": 16607
    },
    {
      "epoch": 5.244079570571519,
      "grad_norm": 3.745393991470337,
      "learning_rate": 4.098866785952356e-05,
      "loss": 0.5631,
      "step": 16608
    },
    {
      "epoch": 5.244395326807704,
      "grad_norm": 5.8545002937316895,
      "learning_rate": 4.098779075886749e-05,
      "loss": 0.7359,
      "step": 16609
    },
    {
      "epoch": 5.2447110830438906,
      "grad_norm": 3.5667734146118164,
      "learning_rate": 4.0986913658211415e-05,
      "loss": 0.7089,
      "step": 16610
    },
    {
      "epoch": 5.245026839280076,
      "grad_norm": 4.336019515991211,
      "learning_rate": 4.098603655755535e-05,
      "loss": 0.6217,
      "step": 16611
    },
    {
      "epoch": 5.245342595516261,
      "grad_norm": 3.6365761756896973,
      "learning_rate": 4.098515945689927e-05,
      "loss": 0.5876,
      "step": 16612
    },
    {
      "epoch": 5.245658351752447,
      "grad_norm": 4.000252723693848,
      "learning_rate": 4.09842823562432e-05,
      "loss": 0.5955,
      "step": 16613
    },
    {
      "epoch": 5.245974107988633,
      "grad_norm": 3.689364194869995,
      "learning_rate": 4.0983405255587135e-05,
      "loss": 0.6809,
      "step": 16614
    },
    {
      "epoch": 5.246289864224819,
      "grad_norm": 4.150262355804443,
      "learning_rate": 4.098252815493106e-05,
      "loss": 0.5892,
      "step": 16615
    },
    {
      "epoch": 5.246605620461004,
      "grad_norm": 3.3624536991119385,
      "learning_rate": 4.098165105427499e-05,
      "loss": 0.5925,
      "step": 16616
    },
    {
      "epoch": 5.246921376697189,
      "grad_norm": 5.643181324005127,
      "learning_rate": 4.0980773953618917e-05,
      "loss": 0.6792,
      "step": 16617
    },
    {
      "epoch": 5.247237132933376,
      "grad_norm": 3.9918954372406006,
      "learning_rate": 4.097989685296285e-05,
      "loss": 0.6102,
      "step": 16618
    },
    {
      "epoch": 5.247552889169561,
      "grad_norm": 5.867069721221924,
      "learning_rate": 4.097901975230677e-05,
      "loss": 0.7624,
      "step": 16619
    },
    {
      "epoch": 5.247868645405747,
      "grad_norm": 3.9231367111206055,
      "learning_rate": 4.0978142651650705e-05,
      "loss": 0.5425,
      "step": 16620
    },
    {
      "epoch": 5.248184401641932,
      "grad_norm": 4.074562072753906,
      "learning_rate": 4.0977265550994636e-05,
      "loss": 0.7498,
      "step": 16621
    },
    {
      "epoch": 5.2485001578781185,
      "grad_norm": 4.155636787414551,
      "learning_rate": 4.097638845033857e-05,
      "loss": 0.5751,
      "step": 16622
    },
    {
      "epoch": 5.248815914114304,
      "grad_norm": 4.428549289703369,
      "learning_rate": 4.097551134968249e-05,
      "loss": 0.6639,
      "step": 16623
    },
    {
      "epoch": 5.249131670350489,
      "grad_norm": 3.7266740798950195,
      "learning_rate": 4.0974634249026425e-05,
      "loss": 0.6348,
      "step": 16624
    },
    {
      "epoch": 5.249447426586675,
      "grad_norm": 3.5851926803588867,
      "learning_rate": 4.097375714837035e-05,
      "loss": 0.6656,
      "step": 16625
    },
    {
      "epoch": 5.249763182822861,
      "grad_norm": 3.9388937950134277,
      "learning_rate": 4.097288004771428e-05,
      "loss": 0.698,
      "step": 16626
    },
    {
      "epoch": 5.250078939059047,
      "grad_norm": 3.6292757987976074,
      "learning_rate": 4.0972002947058206e-05,
      "loss": 0.5723,
      "step": 16627
    },
    {
      "epoch": 5.250394695295232,
      "grad_norm": 3.596069097518921,
      "learning_rate": 4.097112584640214e-05,
      "loss": 0.5833,
      "step": 16628
    },
    {
      "epoch": 5.250710451531417,
      "grad_norm": 3.6981470584869385,
      "learning_rate": 4.097024874574606e-05,
      "loss": 0.6733,
      "step": 16629
    },
    {
      "epoch": 5.2510262077676035,
      "grad_norm": 5.184327602386475,
      "learning_rate": 4.0969371645089995e-05,
      "loss": 0.6372,
      "step": 16630
    },
    {
      "epoch": 5.251341964003789,
      "grad_norm": 4.482119560241699,
      "learning_rate": 4.096849454443392e-05,
      "loss": 0.6708,
      "step": 16631
    },
    {
      "epoch": 5.251657720239975,
      "grad_norm": 3.7918150424957275,
      "learning_rate": 4.096761744377785e-05,
      "loss": 0.5937,
      "step": 16632
    },
    {
      "epoch": 5.25197347647616,
      "grad_norm": 4.577750205993652,
      "learning_rate": 4.0966740343121776e-05,
      "loss": 0.6657,
      "step": 16633
    },
    {
      "epoch": 5.2522892327123465,
      "grad_norm": 3.834829807281494,
      "learning_rate": 4.096586324246571e-05,
      "loss": 0.7032,
      "step": 16634
    },
    {
      "epoch": 5.252604988948532,
      "grad_norm": 3.616323471069336,
      "learning_rate": 4.096498614180963e-05,
      "loss": 0.6577,
      "step": 16635
    },
    {
      "epoch": 5.252920745184717,
      "grad_norm": 4.477343559265137,
      "learning_rate": 4.0964109041153564e-05,
      "loss": 0.7774,
      "step": 16636
    },
    {
      "epoch": 5.253236501420903,
      "grad_norm": 3.5680408477783203,
      "learning_rate": 4.0963231940497496e-05,
      "loss": 0.631,
      "step": 16637
    },
    {
      "epoch": 5.2535522576570886,
      "grad_norm": 4.1316447257995605,
      "learning_rate": 4.096235483984142e-05,
      "loss": 0.6957,
      "step": 16638
    },
    {
      "epoch": 5.253868013893275,
      "grad_norm": 4.856888771057129,
      "learning_rate": 4.096147773918535e-05,
      "loss": 0.6014,
      "step": 16639
    },
    {
      "epoch": 5.25418377012946,
      "grad_norm": 4.138419151306152,
      "learning_rate": 4.096060063852928e-05,
      "loss": 0.6332,
      "step": 16640
    },
    {
      "epoch": 5.254499526365645,
      "grad_norm": 3.8419179916381836,
      "learning_rate": 4.095972353787321e-05,
      "loss": 0.6983,
      "step": 16641
    },
    {
      "epoch": 5.2548152826018315,
      "grad_norm": 4.494497299194336,
      "learning_rate": 4.0958846437217134e-05,
      "loss": 0.6495,
      "step": 16642
    },
    {
      "epoch": 5.255131038838017,
      "grad_norm": 4.469846725463867,
      "learning_rate": 4.0957969336561066e-05,
      "loss": 0.6883,
      "step": 16643
    },
    {
      "epoch": 5.255446795074203,
      "grad_norm": 3.262331008911133,
      "learning_rate": 4.095709223590499e-05,
      "loss": 0.5416,
      "step": 16644
    },
    {
      "epoch": 5.255762551310388,
      "grad_norm": 3.982142448425293,
      "learning_rate": 4.095621513524892e-05,
      "loss": 0.6897,
      "step": 16645
    },
    {
      "epoch": 5.2560783075465745,
      "grad_norm": 3.753021001815796,
      "learning_rate": 4.095533803459285e-05,
      "loss": 0.5498,
      "step": 16646
    },
    {
      "epoch": 5.25639406378276,
      "grad_norm": 4.542710781097412,
      "learning_rate": 4.095446093393678e-05,
      "loss": 0.6263,
      "step": 16647
    },
    {
      "epoch": 5.256709820018945,
      "grad_norm": 5.4104533195495605,
      "learning_rate": 4.095358383328071e-05,
      "loss": 0.739,
      "step": 16648
    },
    {
      "epoch": 5.257025576255131,
      "grad_norm": 4.326671600341797,
      "learning_rate": 4.095270673262464e-05,
      "loss": 0.5829,
      "step": 16649
    },
    {
      "epoch": 5.2573413324913165,
      "grad_norm": 3.543693780899048,
      "learning_rate": 4.095182963196857e-05,
      "loss": 0.5938,
      "step": 16650
    },
    {
      "epoch": 5.257657088727503,
      "grad_norm": 4.4045515060424805,
      "learning_rate": 4.09509525313125e-05,
      "loss": 0.7356,
      "step": 16651
    },
    {
      "epoch": 5.257972844963688,
      "grad_norm": 4.5182204246521,
      "learning_rate": 4.0950075430656424e-05,
      "loss": 0.7608,
      "step": 16652
    },
    {
      "epoch": 5.258288601199873,
      "grad_norm": 4.035547733306885,
      "learning_rate": 4.0949198330000355e-05,
      "loss": 0.6361,
      "step": 16653
    },
    {
      "epoch": 5.2586043574360595,
      "grad_norm": 5.4794816970825195,
      "learning_rate": 4.094832122934428e-05,
      "loss": 0.6611,
      "step": 16654
    },
    {
      "epoch": 5.258920113672245,
      "grad_norm": 3.4729115962982178,
      "learning_rate": 4.094744412868821e-05,
      "loss": 0.5422,
      "step": 16655
    },
    {
      "epoch": 5.259235869908431,
      "grad_norm": 5.118597984313965,
      "learning_rate": 4.094656702803214e-05,
      "loss": 0.627,
      "step": 16656
    },
    {
      "epoch": 5.259551626144616,
      "grad_norm": 4.102987289428711,
      "learning_rate": 4.094568992737607e-05,
      "loss": 0.5377,
      "step": 16657
    },
    {
      "epoch": 5.259867382380802,
      "grad_norm": 3.5813910961151123,
      "learning_rate": 4.094481282672e-05,
      "loss": 0.6817,
      "step": 16658
    },
    {
      "epoch": 5.260183138616988,
      "grad_norm": 3.8839352130889893,
      "learning_rate": 4.0943935726063925e-05,
      "loss": 0.4989,
      "step": 16659
    },
    {
      "epoch": 5.260498894853173,
      "grad_norm": 3.8003146648406982,
      "learning_rate": 4.094305862540786e-05,
      "loss": 0.6991,
      "step": 16660
    },
    {
      "epoch": 5.260814651089359,
      "grad_norm": 4.1706695556640625,
      "learning_rate": 4.094218152475178e-05,
      "loss": 0.6544,
      "step": 16661
    },
    {
      "epoch": 5.2611304073255445,
      "grad_norm": 4.758573055267334,
      "learning_rate": 4.0941304424095713e-05,
      "loss": 0.6485,
      "step": 16662
    },
    {
      "epoch": 5.261446163561731,
      "grad_norm": 4.434588432312012,
      "learning_rate": 4.094042732343964e-05,
      "loss": 0.6581,
      "step": 16663
    },
    {
      "epoch": 5.261761919797916,
      "grad_norm": 5.042057037353516,
      "learning_rate": 4.093955022278357e-05,
      "loss": 0.6912,
      "step": 16664
    },
    {
      "epoch": 5.262077676034101,
      "grad_norm": 3.299248218536377,
      "learning_rate": 4.0938673122127495e-05,
      "loss": 0.5699,
      "step": 16665
    },
    {
      "epoch": 5.262393432270287,
      "grad_norm": 3.5758657455444336,
      "learning_rate": 4.093779602147143e-05,
      "loss": 0.5862,
      "step": 16666
    },
    {
      "epoch": 5.262709188506473,
      "grad_norm": 3.854482889175415,
      "learning_rate": 4.093691892081535e-05,
      "loss": 0.5678,
      "step": 16667
    },
    {
      "epoch": 5.263024944742659,
      "grad_norm": 3.8370461463928223,
      "learning_rate": 4.093604182015928e-05,
      "loss": 0.5382,
      "step": 16668
    },
    {
      "epoch": 5.263340700978844,
      "grad_norm": 4.550622463226318,
      "learning_rate": 4.093516471950321e-05,
      "loss": 0.5536,
      "step": 16669
    },
    {
      "epoch": 5.26365645721503,
      "grad_norm": 4.2605671882629395,
      "learning_rate": 4.093428761884714e-05,
      "loss": 0.7083,
      "step": 16670
    },
    {
      "epoch": 5.263972213451216,
      "grad_norm": 5.038915634155273,
      "learning_rate": 4.0933410518191065e-05,
      "loss": 0.6317,
      "step": 16671
    },
    {
      "epoch": 5.264287969687401,
      "grad_norm": 3.965625762939453,
      "learning_rate": 4.0932533417534996e-05,
      "loss": 0.6647,
      "step": 16672
    },
    {
      "epoch": 5.264603725923587,
      "grad_norm": 3.740978479385376,
      "learning_rate": 4.093165631687892e-05,
      "loss": 0.6493,
      "step": 16673
    },
    {
      "epoch": 5.2649194821597725,
      "grad_norm": 4.049648284912109,
      "learning_rate": 4.093077921622285e-05,
      "loss": 0.57,
      "step": 16674
    },
    {
      "epoch": 5.265235238395959,
      "grad_norm": 4.616310119628906,
      "learning_rate": 4.0929902115566785e-05,
      "loss": 0.7103,
      "step": 16675
    },
    {
      "epoch": 5.265550994632144,
      "grad_norm": 4.307373523712158,
      "learning_rate": 4.0929025014910716e-05,
      "loss": 0.5669,
      "step": 16676
    },
    {
      "epoch": 5.26586675086833,
      "grad_norm": 6.256441593170166,
      "learning_rate": 4.092814791425464e-05,
      "loss": 0.6056,
      "step": 16677
    },
    {
      "epoch": 5.266182507104515,
      "grad_norm": 4.134538650512695,
      "learning_rate": 4.092727081359857e-05,
      "loss": 0.6122,
      "step": 16678
    },
    {
      "epoch": 5.266498263340701,
      "grad_norm": 6.227024078369141,
      "learning_rate": 4.09263937129425e-05,
      "loss": 0.6718,
      "step": 16679
    },
    {
      "epoch": 5.266814019576887,
      "grad_norm": 4.453855514526367,
      "learning_rate": 4.092551661228643e-05,
      "loss": 0.5502,
      "step": 16680
    },
    {
      "epoch": 5.267129775813072,
      "grad_norm": 4.666754722595215,
      "learning_rate": 4.092463951163036e-05,
      "loss": 0.7554,
      "step": 16681
    },
    {
      "epoch": 5.267445532049258,
      "grad_norm": 4.417414665222168,
      "learning_rate": 4.0923762410974286e-05,
      "loss": 0.6008,
      "step": 16682
    },
    {
      "epoch": 5.267761288285444,
      "grad_norm": 3.7759530544281006,
      "learning_rate": 4.092288531031822e-05,
      "loss": 0.6675,
      "step": 16683
    },
    {
      "epoch": 5.268077044521629,
      "grad_norm": 4.453526973724365,
      "learning_rate": 4.092200820966214e-05,
      "loss": 0.6571,
      "step": 16684
    },
    {
      "epoch": 5.268392800757815,
      "grad_norm": 6.900931358337402,
      "learning_rate": 4.0921131109006074e-05,
      "loss": 0.6308,
      "step": 16685
    },
    {
      "epoch": 5.268708556994,
      "grad_norm": 5.121579647064209,
      "learning_rate": 4.092025400835e-05,
      "loss": 0.5839,
      "step": 16686
    },
    {
      "epoch": 5.269024313230187,
      "grad_norm": 4.657153129577637,
      "learning_rate": 4.091937690769393e-05,
      "loss": 0.6304,
      "step": 16687
    },
    {
      "epoch": 5.269340069466372,
      "grad_norm": 4.613969802856445,
      "learning_rate": 4.0918499807037856e-05,
      "loss": 0.6683,
      "step": 16688
    },
    {
      "epoch": 5.269655825702557,
      "grad_norm": 4.652799606323242,
      "learning_rate": 4.091762270638179e-05,
      "loss": 0.6611,
      "step": 16689
    },
    {
      "epoch": 5.269971581938743,
      "grad_norm": 4.009964942932129,
      "learning_rate": 4.091674560572571e-05,
      "loss": 0.6546,
      "step": 16690
    },
    {
      "epoch": 5.270287338174929,
      "grad_norm": 3.249833345413208,
      "learning_rate": 4.0915868505069644e-05,
      "loss": 0.5849,
      "step": 16691
    },
    {
      "epoch": 5.270603094411115,
      "grad_norm": 5.038856029510498,
      "learning_rate": 4.091499140441357e-05,
      "loss": 0.6392,
      "step": 16692
    },
    {
      "epoch": 5.2709188506473,
      "grad_norm": 4.22508430480957,
      "learning_rate": 4.09141143037575e-05,
      "loss": 0.6838,
      "step": 16693
    },
    {
      "epoch": 5.271234606883486,
      "grad_norm": 4.042431831359863,
      "learning_rate": 4.0913237203101426e-05,
      "loss": 0.5775,
      "step": 16694
    },
    {
      "epoch": 5.271550363119672,
      "grad_norm": 4.737951755523682,
      "learning_rate": 4.091236010244536e-05,
      "loss": 0.6724,
      "step": 16695
    },
    {
      "epoch": 5.271866119355857,
      "grad_norm": 4.654538154602051,
      "learning_rate": 4.091148300178929e-05,
      "loss": 0.6053,
      "step": 16696
    },
    {
      "epoch": 5.272181875592043,
      "grad_norm": 4.2324934005737305,
      "learning_rate": 4.0910605901133214e-05,
      "loss": 0.6625,
      "step": 16697
    },
    {
      "epoch": 5.272497631828228,
      "grad_norm": 4.148864269256592,
      "learning_rate": 4.0909728800477146e-05,
      "loss": 0.5339,
      "step": 16698
    },
    {
      "epoch": 5.272813388064415,
      "grad_norm": 3.7752134799957275,
      "learning_rate": 4.090885169982107e-05,
      "loss": 0.6619,
      "step": 16699
    },
    {
      "epoch": 5.2731291443006,
      "grad_norm": 4.537870407104492,
      "learning_rate": 4.0907974599165e-05,
      "loss": 0.5785,
      "step": 16700
    },
    {
      "epoch": 5.273444900536786,
      "grad_norm": 3.4443373680114746,
      "learning_rate": 4.090709749850893e-05,
      "loss": 0.6463,
      "step": 16701
    },
    {
      "epoch": 5.273760656772971,
      "grad_norm": 3.930576801300049,
      "learning_rate": 4.090622039785286e-05,
      "loss": 0.5624,
      "step": 16702
    },
    {
      "epoch": 5.274076413009157,
      "grad_norm": 5.012113571166992,
      "learning_rate": 4.090534329719679e-05,
      "loss": 0.5858,
      "step": 16703
    },
    {
      "epoch": 5.274392169245343,
      "grad_norm": 4.5232133865356445,
      "learning_rate": 4.090446619654072e-05,
      "loss": 0.7593,
      "step": 16704
    },
    {
      "epoch": 5.274707925481528,
      "grad_norm": 3.1954030990600586,
      "learning_rate": 4.090358909588465e-05,
      "loss": 0.6191,
      "step": 16705
    },
    {
      "epoch": 5.275023681717714,
      "grad_norm": 5.291788101196289,
      "learning_rate": 4.090271199522858e-05,
      "loss": 0.7514,
      "step": 16706
    },
    {
      "epoch": 5.2753394379539,
      "grad_norm": 5.128681659698486,
      "learning_rate": 4.0901834894572504e-05,
      "loss": 0.7285,
      "step": 16707
    },
    {
      "epoch": 5.275655194190085,
      "grad_norm": 5.0247721672058105,
      "learning_rate": 4.0900957793916435e-05,
      "loss": 0.638,
      "step": 16708
    },
    {
      "epoch": 5.275970950426271,
      "grad_norm": 3.766420364379883,
      "learning_rate": 4.090008069326036e-05,
      "loss": 0.6187,
      "step": 16709
    },
    {
      "epoch": 5.276286706662456,
      "grad_norm": 4.379584789276123,
      "learning_rate": 4.089920359260429e-05,
      "loss": 0.6246,
      "step": 16710
    },
    {
      "epoch": 5.2766024628986425,
      "grad_norm": 3.901895761489868,
      "learning_rate": 4.089832649194822e-05,
      "loss": 0.5245,
      "step": 16711
    },
    {
      "epoch": 5.276918219134828,
      "grad_norm": 3.9343016147613525,
      "learning_rate": 4.089744939129215e-05,
      "loss": 0.6824,
      "step": 16712
    },
    {
      "epoch": 5.277233975371013,
      "grad_norm": 4.3719587326049805,
      "learning_rate": 4.0896572290636073e-05,
      "loss": 0.7817,
      "step": 16713
    },
    {
      "epoch": 5.277549731607199,
      "grad_norm": 4.299112796783447,
      "learning_rate": 4.0895695189980005e-05,
      "loss": 0.6149,
      "step": 16714
    },
    {
      "epoch": 5.277865487843385,
      "grad_norm": 4.412680149078369,
      "learning_rate": 4.089481808932393e-05,
      "loss": 0.5897,
      "step": 16715
    },
    {
      "epoch": 5.278181244079571,
      "grad_norm": 3.9355545043945312,
      "learning_rate": 4.089394098866786e-05,
      "loss": 0.5844,
      "step": 16716
    },
    {
      "epoch": 5.278497000315756,
      "grad_norm": 4.1584792137146,
      "learning_rate": 4.089306388801179e-05,
      "loss": 0.6236,
      "step": 16717
    },
    {
      "epoch": 5.278812756551942,
      "grad_norm": 3.4423654079437256,
      "learning_rate": 4.089218678735572e-05,
      "loss": 0.5806,
      "step": 16718
    },
    {
      "epoch": 5.279128512788128,
      "grad_norm": 3.789977788925171,
      "learning_rate": 4.089130968669965e-05,
      "loss": 0.5792,
      "step": 16719
    },
    {
      "epoch": 5.279444269024313,
      "grad_norm": 3.505364418029785,
      "learning_rate": 4.0890432586043575e-05,
      "loss": 0.5586,
      "step": 16720
    },
    {
      "epoch": 5.279760025260499,
      "grad_norm": 3.6637706756591797,
      "learning_rate": 4.0889555485387507e-05,
      "loss": 0.5866,
      "step": 16721
    },
    {
      "epoch": 5.280075781496684,
      "grad_norm": 4.09856653213501,
      "learning_rate": 4.088867838473143e-05,
      "loss": 0.6065,
      "step": 16722
    },
    {
      "epoch": 5.2803915377328705,
      "grad_norm": 4.599915981292725,
      "learning_rate": 4.088780128407536e-05,
      "loss": 0.7288,
      "step": 16723
    },
    {
      "epoch": 5.280707293969056,
      "grad_norm": 4.283210754394531,
      "learning_rate": 4.088692418341929e-05,
      "loss": 0.6563,
      "step": 16724
    },
    {
      "epoch": 5.281023050205242,
      "grad_norm": 3.9330923557281494,
      "learning_rate": 4.088604708276322e-05,
      "loss": 0.6778,
      "step": 16725
    },
    {
      "epoch": 5.281338806441427,
      "grad_norm": 4.0741167068481445,
      "learning_rate": 4.0885169982107145e-05,
      "loss": 0.6848,
      "step": 16726
    },
    {
      "epoch": 5.281654562677613,
      "grad_norm": 4.302596569061279,
      "learning_rate": 4.0884292881451076e-05,
      "loss": 0.7247,
      "step": 16727
    },
    {
      "epoch": 5.281970318913799,
      "grad_norm": 4.65435791015625,
      "learning_rate": 4.0883415780795e-05,
      "loss": 0.6614,
      "step": 16728
    },
    {
      "epoch": 5.282286075149984,
      "grad_norm": 5.096517086029053,
      "learning_rate": 4.088253868013893e-05,
      "loss": 0.6373,
      "step": 16729
    },
    {
      "epoch": 5.28260183138617,
      "grad_norm": 4.1864542961120605,
      "learning_rate": 4.0881661579482865e-05,
      "loss": 0.6309,
      "step": 16730
    },
    {
      "epoch": 5.2829175876223555,
      "grad_norm": 3.9655377864837646,
      "learning_rate": 4.0880784478826796e-05,
      "loss": 0.6704,
      "step": 16731
    },
    {
      "epoch": 5.283233343858541,
      "grad_norm": 4.203059673309326,
      "learning_rate": 4.087990737817072e-05,
      "loss": 0.6244,
      "step": 16732
    },
    {
      "epoch": 5.283549100094727,
      "grad_norm": 5.205757141113281,
      "learning_rate": 4.087903027751465e-05,
      "loss": 0.7329,
      "step": 16733
    },
    {
      "epoch": 5.283864856330912,
      "grad_norm": 4.245216369628906,
      "learning_rate": 4.087815317685858e-05,
      "loss": 0.6153,
      "step": 16734
    },
    {
      "epoch": 5.2841806125670985,
      "grad_norm": 3.4437901973724365,
      "learning_rate": 4.087727607620251e-05,
      "loss": 0.5753,
      "step": 16735
    },
    {
      "epoch": 5.284496368803284,
      "grad_norm": 3.9955458641052246,
      "learning_rate": 4.0876398975546434e-05,
      "loss": 0.5525,
      "step": 16736
    },
    {
      "epoch": 5.284812125039469,
      "grad_norm": 4.566498756408691,
      "learning_rate": 4.0875521874890366e-05,
      "loss": 0.6968,
      "step": 16737
    },
    {
      "epoch": 5.285127881275655,
      "grad_norm": 4.29627799987793,
      "learning_rate": 4.087464477423429e-05,
      "loss": 0.6409,
      "step": 16738
    },
    {
      "epoch": 5.2854436375118405,
      "grad_norm": 3.662416696548462,
      "learning_rate": 4.087376767357822e-05,
      "loss": 0.6276,
      "step": 16739
    },
    {
      "epoch": 5.285759393748027,
      "grad_norm": 4.551540374755859,
      "learning_rate": 4.0872890572922154e-05,
      "loss": 0.6846,
      "step": 16740
    },
    {
      "epoch": 5.286075149984212,
      "grad_norm": 4.818722248077393,
      "learning_rate": 4.087201347226608e-05,
      "loss": 0.5973,
      "step": 16741
    },
    {
      "epoch": 5.286390906220398,
      "grad_norm": 4.11458158493042,
      "learning_rate": 4.087113637161001e-05,
      "loss": 0.6965,
      "step": 16742
    },
    {
      "epoch": 5.2867066624565835,
      "grad_norm": 3.070183038711548,
      "learning_rate": 4.0870259270953936e-05,
      "loss": 0.5523,
      "step": 16743
    },
    {
      "epoch": 5.287022418692769,
      "grad_norm": 6.497572898864746,
      "learning_rate": 4.086938217029787e-05,
      "loss": 0.7214,
      "step": 16744
    },
    {
      "epoch": 5.287338174928955,
      "grad_norm": 5.2413482666015625,
      "learning_rate": 4.086850506964179e-05,
      "loss": 0.6892,
      "step": 16745
    },
    {
      "epoch": 5.28765393116514,
      "grad_norm": 3.5925087928771973,
      "learning_rate": 4.0867627968985724e-05,
      "loss": 0.6487,
      "step": 16746
    },
    {
      "epoch": 5.2879696874013264,
      "grad_norm": 3.6147992610931396,
      "learning_rate": 4.086675086832965e-05,
      "loss": 0.5487,
      "step": 16747
    },
    {
      "epoch": 5.288285443637512,
      "grad_norm": 4.125942230224609,
      "learning_rate": 4.086587376767358e-05,
      "loss": 0.6141,
      "step": 16748
    },
    {
      "epoch": 5.288601199873698,
      "grad_norm": 4.667723178863525,
      "learning_rate": 4.0864996667017506e-05,
      "loss": 0.6795,
      "step": 16749
    },
    {
      "epoch": 5.288916956109883,
      "grad_norm": 5.19053316116333,
      "learning_rate": 4.086411956636144e-05,
      "loss": 0.7049,
      "step": 16750
    },
    {
      "epoch": 5.2892327123460685,
      "grad_norm": 4.114137649536133,
      "learning_rate": 4.086324246570536e-05,
      "loss": 0.6501,
      "step": 16751
    },
    {
      "epoch": 5.289548468582255,
      "grad_norm": 4.760214328765869,
      "learning_rate": 4.0862365365049294e-05,
      "loss": 0.668,
      "step": 16752
    },
    {
      "epoch": 5.28986422481844,
      "grad_norm": 4.585320949554443,
      "learning_rate": 4.086148826439322e-05,
      "loss": 0.6918,
      "step": 16753
    },
    {
      "epoch": 5.290179981054626,
      "grad_norm": 4.290403366088867,
      "learning_rate": 4.086061116373715e-05,
      "loss": 0.5737,
      "step": 16754
    },
    {
      "epoch": 5.2904957372908115,
      "grad_norm": 4.800775051116943,
      "learning_rate": 4.085973406308108e-05,
      "loss": 0.7592,
      "step": 16755
    },
    {
      "epoch": 5.290811493526997,
      "grad_norm": 4.820059776306152,
      "learning_rate": 4.085885696242501e-05,
      "loss": 0.6599,
      "step": 16756
    },
    {
      "epoch": 5.291127249763183,
      "grad_norm": 3.8507113456726074,
      "learning_rate": 4.085797986176894e-05,
      "loss": 0.6671,
      "step": 16757
    },
    {
      "epoch": 5.291443005999368,
      "grad_norm": 4.811151027679443,
      "learning_rate": 4.085710276111287e-05,
      "loss": 0.7064,
      "step": 16758
    },
    {
      "epoch": 5.291758762235554,
      "grad_norm": 3.4118235111236572,
      "learning_rate": 4.0856225660456795e-05,
      "loss": 0.6095,
      "step": 16759
    },
    {
      "epoch": 5.29207451847174,
      "grad_norm": 3.7870943546295166,
      "learning_rate": 4.085534855980073e-05,
      "loss": 0.6522,
      "step": 16760
    },
    {
      "epoch": 5.292390274707926,
      "grad_norm": 4.912773132324219,
      "learning_rate": 4.085447145914466e-05,
      "loss": 0.5956,
      "step": 16761
    },
    {
      "epoch": 5.292706030944111,
      "grad_norm": 3.6952836513519287,
      "learning_rate": 4.0853594358488584e-05,
      "loss": 0.6274,
      "step": 16762
    },
    {
      "epoch": 5.2930217871802965,
      "grad_norm": 4.194327354431152,
      "learning_rate": 4.0852717257832515e-05,
      "loss": 0.6532,
      "step": 16763
    },
    {
      "epoch": 5.293337543416483,
      "grad_norm": 3.7562460899353027,
      "learning_rate": 4.085184015717644e-05,
      "loss": 0.6714,
      "step": 16764
    },
    {
      "epoch": 5.293653299652668,
      "grad_norm": 3.6243155002593994,
      "learning_rate": 4.085096305652037e-05,
      "loss": 0.6084,
      "step": 16765
    },
    {
      "epoch": 5.293969055888854,
      "grad_norm": 4.327500343322754,
      "learning_rate": 4.08500859558643e-05,
      "loss": 0.7227,
      "step": 16766
    },
    {
      "epoch": 5.294284812125039,
      "grad_norm": 5.30127477645874,
      "learning_rate": 4.084920885520823e-05,
      "loss": 0.7242,
      "step": 16767
    },
    {
      "epoch": 5.294600568361225,
      "grad_norm": 3.592573642730713,
      "learning_rate": 4.084833175455215e-05,
      "loss": 0.6207,
      "step": 16768
    },
    {
      "epoch": 5.294916324597411,
      "grad_norm": 4.599826812744141,
      "learning_rate": 4.0847454653896085e-05,
      "loss": 0.5538,
      "step": 16769
    },
    {
      "epoch": 5.295232080833596,
      "grad_norm": 4.099644660949707,
      "learning_rate": 4.084657755324001e-05,
      "loss": 0.6466,
      "step": 16770
    },
    {
      "epoch": 5.295547837069782,
      "grad_norm": 4.233729362487793,
      "learning_rate": 4.084570045258394e-05,
      "loss": 0.6936,
      "step": 16771
    },
    {
      "epoch": 5.295863593305968,
      "grad_norm": 5.067417144775391,
      "learning_rate": 4.0844823351927867e-05,
      "loss": 0.6978,
      "step": 16772
    },
    {
      "epoch": 5.296179349542154,
      "grad_norm": 3.631591320037842,
      "learning_rate": 4.08439462512718e-05,
      "loss": 0.641,
      "step": 16773
    },
    {
      "epoch": 5.296495105778339,
      "grad_norm": 3.7450265884399414,
      "learning_rate": 4.084306915061572e-05,
      "loss": 0.5818,
      "step": 16774
    },
    {
      "epoch": 5.2968108620145244,
      "grad_norm": 4.18218994140625,
      "learning_rate": 4.0842192049959655e-05,
      "loss": 0.5203,
      "step": 16775
    },
    {
      "epoch": 5.297126618250711,
      "grad_norm": 3.8516297340393066,
      "learning_rate": 4.084131494930358e-05,
      "loss": 0.5842,
      "step": 16776
    },
    {
      "epoch": 5.297442374486896,
      "grad_norm": 4.551969051361084,
      "learning_rate": 4.084043784864751e-05,
      "loss": 0.6855,
      "step": 16777
    },
    {
      "epoch": 5.297758130723082,
      "grad_norm": 4.6245551109313965,
      "learning_rate": 4.083956074799144e-05,
      "loss": 0.6444,
      "step": 16778
    },
    {
      "epoch": 5.298073886959267,
      "grad_norm": 3.4375534057617188,
      "learning_rate": 4.083868364733537e-05,
      "loss": 0.629,
      "step": 16779
    },
    {
      "epoch": 5.298389643195453,
      "grad_norm": 4.043628692626953,
      "learning_rate": 4.08378065466793e-05,
      "loss": 0.6435,
      "step": 16780
    },
    {
      "epoch": 5.298705399431639,
      "grad_norm": 3.6967525482177734,
      "learning_rate": 4.0836929446023225e-05,
      "loss": 0.5941,
      "step": 16781
    },
    {
      "epoch": 5.299021155667824,
      "grad_norm": 4.769247531890869,
      "learning_rate": 4.0836052345367156e-05,
      "loss": 0.5882,
      "step": 16782
    },
    {
      "epoch": 5.29933691190401,
      "grad_norm": 3.8853752613067627,
      "learning_rate": 4.083517524471108e-05,
      "loss": 0.5192,
      "step": 16783
    },
    {
      "epoch": 5.299652668140196,
      "grad_norm": 4.61496639251709,
      "learning_rate": 4.083429814405501e-05,
      "loss": 0.7551,
      "step": 16784
    },
    {
      "epoch": 5.299968424376382,
      "grad_norm": 4.881227970123291,
      "learning_rate": 4.0833421043398944e-05,
      "loss": 0.674,
      "step": 16785
    },
    {
      "epoch": 5.300284180612567,
      "grad_norm": 4.173529624938965,
      "learning_rate": 4.0832543942742876e-05,
      "loss": 0.6046,
      "step": 16786
    },
    {
      "epoch": 5.300599936848752,
      "grad_norm": 4.114343643188477,
      "learning_rate": 4.08316668420868e-05,
      "loss": 0.6131,
      "step": 16787
    },
    {
      "epoch": 5.300915693084939,
      "grad_norm": 4.8017964363098145,
      "learning_rate": 4.083078974143073e-05,
      "loss": 0.6323,
      "step": 16788
    },
    {
      "epoch": 5.301231449321124,
      "grad_norm": 3.8076486587524414,
      "learning_rate": 4.082991264077466e-05,
      "loss": 0.5751,
      "step": 16789
    },
    {
      "epoch": 5.30154720555731,
      "grad_norm": 4.440614700317383,
      "learning_rate": 4.082903554011859e-05,
      "loss": 0.6302,
      "step": 16790
    },
    {
      "epoch": 5.301862961793495,
      "grad_norm": 4.3936285972595215,
      "learning_rate": 4.0828158439462514e-05,
      "loss": 0.6597,
      "step": 16791
    },
    {
      "epoch": 5.3021787180296815,
      "grad_norm": 5.020362377166748,
      "learning_rate": 4.0827281338806446e-05,
      "loss": 0.6213,
      "step": 16792
    },
    {
      "epoch": 5.302494474265867,
      "grad_norm": 3.937326669692993,
      "learning_rate": 4.082640423815037e-05,
      "loss": 0.5798,
      "step": 16793
    },
    {
      "epoch": 5.302810230502052,
      "grad_norm": 3.676872730255127,
      "learning_rate": 4.08255271374943e-05,
      "loss": 0.6432,
      "step": 16794
    },
    {
      "epoch": 5.303125986738238,
      "grad_norm": 3.9746971130371094,
      "learning_rate": 4.082465003683823e-05,
      "loss": 0.6373,
      "step": 16795
    },
    {
      "epoch": 5.303441742974424,
      "grad_norm": 3.6628546714782715,
      "learning_rate": 4.082377293618216e-05,
      "loss": 0.5676,
      "step": 16796
    },
    {
      "epoch": 5.30375749921061,
      "grad_norm": 3.7743639945983887,
      "learning_rate": 4.0822895835526084e-05,
      "loss": 0.6628,
      "step": 16797
    },
    {
      "epoch": 5.304073255446795,
      "grad_norm": 4.41315221786499,
      "learning_rate": 4.0822018734870016e-05,
      "loss": 0.6606,
      "step": 16798
    },
    {
      "epoch": 5.30438901168298,
      "grad_norm": 4.734318733215332,
      "learning_rate": 4.082114163421395e-05,
      "loss": 0.7094,
      "step": 16799
    },
    {
      "epoch": 5.304704767919167,
      "grad_norm": 3.994837760925293,
      "learning_rate": 4.082026453355787e-05,
      "loss": 0.6852,
      "step": 16800
    },
    {
      "epoch": 5.304704767919167,
      "eval_loss": 0.86509770154953,
      "eval_runtime": 4.5687,
      "eval_samples_per_second": 545.018,
      "eval_steps_per_second": 4.597,
      "step": 16800
    },
    {
      "epoch": 5.305020524155352,
      "grad_norm": 8.731081008911133,
      "learning_rate": 4.0819387432901804e-05,
      "loss": 0.7177,
      "step": 16801
    },
    {
      "epoch": 5.305336280391538,
      "grad_norm": 3.6652283668518066,
      "learning_rate": 4.081851033224573e-05,
      "loss": 0.6493,
      "step": 16802
    },
    {
      "epoch": 5.305652036627723,
      "grad_norm": 3.9270007610321045,
      "learning_rate": 4.081763323158966e-05,
      "loss": 0.5999,
      "step": 16803
    },
    {
      "epoch": 5.305967792863909,
      "grad_norm": 4.399959087371826,
      "learning_rate": 4.0816756130933585e-05,
      "loss": 0.6654,
      "step": 16804
    },
    {
      "epoch": 5.306283549100095,
      "grad_norm": 4.91082239151001,
      "learning_rate": 4.081587903027752e-05,
      "loss": 0.751,
      "step": 16805
    },
    {
      "epoch": 5.30659930533628,
      "grad_norm": 3.8021655082702637,
      "learning_rate": 4.081500192962144e-05,
      "loss": 0.5457,
      "step": 16806
    },
    {
      "epoch": 5.306915061572466,
      "grad_norm": 4.050779819488525,
      "learning_rate": 4.0814124828965374e-05,
      "loss": 0.6754,
      "step": 16807
    },
    {
      "epoch": 5.307230817808652,
      "grad_norm": 4.614618301391602,
      "learning_rate": 4.08132477283093e-05,
      "loss": 0.5777,
      "step": 16808
    },
    {
      "epoch": 5.307546574044838,
      "grad_norm": 3.884620428085327,
      "learning_rate": 4.081237062765323e-05,
      "loss": 0.6431,
      "step": 16809
    },
    {
      "epoch": 5.307862330281023,
      "grad_norm": 3.828251838684082,
      "learning_rate": 4.0811493526997155e-05,
      "loss": 0.6524,
      "step": 16810
    },
    {
      "epoch": 5.308178086517208,
      "grad_norm": 4.571780204772949,
      "learning_rate": 4.081061642634109e-05,
      "loss": 0.6375,
      "step": 16811
    },
    {
      "epoch": 5.3084938427533945,
      "grad_norm": 3.761439085006714,
      "learning_rate": 4.080973932568502e-05,
      "loss": 0.6008,
      "step": 16812
    },
    {
      "epoch": 5.30880959898958,
      "grad_norm": 3.6786911487579346,
      "learning_rate": 4.080886222502895e-05,
      "loss": 0.6886,
      "step": 16813
    },
    {
      "epoch": 5.309125355225766,
      "grad_norm": 4.8380889892578125,
      "learning_rate": 4.0807985124372875e-05,
      "loss": 0.7426,
      "step": 16814
    },
    {
      "epoch": 5.309441111461951,
      "grad_norm": 4.075939178466797,
      "learning_rate": 4.080710802371681e-05,
      "loss": 0.6846,
      "step": 16815
    },
    {
      "epoch": 5.3097568676981375,
      "grad_norm": 4.246726989746094,
      "learning_rate": 4.080623092306073e-05,
      "loss": 0.6027,
      "step": 16816
    },
    {
      "epoch": 5.310072623934323,
      "grad_norm": 3.499716281890869,
      "learning_rate": 4.0805353822404663e-05,
      "loss": 0.6177,
      "step": 16817
    },
    {
      "epoch": 5.310388380170508,
      "grad_norm": 6.158758640289307,
      "learning_rate": 4.080447672174859e-05,
      "loss": 0.5933,
      "step": 16818
    },
    {
      "epoch": 5.310704136406694,
      "grad_norm": 4.0902228355407715,
      "learning_rate": 4.080359962109252e-05,
      "loss": 0.6844,
      "step": 16819
    },
    {
      "epoch": 5.3110198926428795,
      "grad_norm": 4.109443187713623,
      "learning_rate": 4.0802722520436445e-05,
      "loss": 0.6289,
      "step": 16820
    },
    {
      "epoch": 5.311335648879066,
      "grad_norm": 4.5511345863342285,
      "learning_rate": 4.080184541978038e-05,
      "loss": 0.6345,
      "step": 16821
    },
    {
      "epoch": 5.311651405115251,
      "grad_norm": 3.286623954772949,
      "learning_rate": 4.080096831912431e-05,
      "loss": 0.577,
      "step": 16822
    },
    {
      "epoch": 5.311967161351436,
      "grad_norm": 4.853574275970459,
      "learning_rate": 4.080009121846823e-05,
      "loss": 0.6693,
      "step": 16823
    },
    {
      "epoch": 5.3122829175876225,
      "grad_norm": 4.518272399902344,
      "learning_rate": 4.0799214117812165e-05,
      "loss": 0.6524,
      "step": 16824
    },
    {
      "epoch": 5.312598673823808,
      "grad_norm": 3.998763084411621,
      "learning_rate": 4.079833701715609e-05,
      "loss": 0.6758,
      "step": 16825
    },
    {
      "epoch": 5.312914430059994,
      "grad_norm": 4.457012176513672,
      "learning_rate": 4.079745991650002e-05,
      "loss": 0.6232,
      "step": 16826
    },
    {
      "epoch": 5.313230186296179,
      "grad_norm": 4.565929412841797,
      "learning_rate": 4.0796582815843946e-05,
      "loss": 0.547,
      "step": 16827
    },
    {
      "epoch": 5.313545942532365,
      "grad_norm": 3.561520576477051,
      "learning_rate": 4.079570571518788e-05,
      "loss": 0.6542,
      "step": 16828
    },
    {
      "epoch": 5.313861698768551,
      "grad_norm": 3.920802593231201,
      "learning_rate": 4.07948286145318e-05,
      "loss": 0.595,
      "step": 16829
    },
    {
      "epoch": 5.314177455004736,
      "grad_norm": 4.492376804351807,
      "learning_rate": 4.0793951513875735e-05,
      "loss": 0.695,
      "step": 16830
    },
    {
      "epoch": 5.314493211240922,
      "grad_norm": 4.208197116851807,
      "learning_rate": 4.079307441321966e-05,
      "loss": 0.6838,
      "step": 16831
    },
    {
      "epoch": 5.3148089674771075,
      "grad_norm": 4.494638442993164,
      "learning_rate": 4.079219731256359e-05,
      "loss": 0.5608,
      "step": 16832
    },
    {
      "epoch": 5.315124723713294,
      "grad_norm": 4.761181831359863,
      "learning_rate": 4.0791320211907516e-05,
      "loss": 0.6704,
      "step": 16833
    },
    {
      "epoch": 5.315440479949479,
      "grad_norm": 4.6068434715271,
      "learning_rate": 4.079044311125145e-05,
      "loss": 0.6844,
      "step": 16834
    },
    {
      "epoch": 5.315756236185664,
      "grad_norm": 4.649682998657227,
      "learning_rate": 4.078956601059537e-05,
      "loss": 0.7875,
      "step": 16835
    },
    {
      "epoch": 5.3160719924218505,
      "grad_norm": 4.780223369598389,
      "learning_rate": 4.0788688909939304e-05,
      "loss": 0.6217,
      "step": 16836
    },
    {
      "epoch": 5.316387748658036,
      "grad_norm": 4.20621919631958,
      "learning_rate": 4.0787811809283236e-05,
      "loss": 0.625,
      "step": 16837
    },
    {
      "epoch": 5.316703504894222,
      "grad_norm": 4.079632759094238,
      "learning_rate": 4.078693470862716e-05,
      "loss": 0.5824,
      "step": 16838
    },
    {
      "epoch": 5.317019261130407,
      "grad_norm": 5.181775093078613,
      "learning_rate": 4.078605760797109e-05,
      "loss": 0.6197,
      "step": 16839
    },
    {
      "epoch": 5.317335017366593,
      "grad_norm": 4.104080677032471,
      "learning_rate": 4.0785180507315024e-05,
      "loss": 0.6169,
      "step": 16840
    },
    {
      "epoch": 5.317650773602779,
      "grad_norm": 4.226319789886475,
      "learning_rate": 4.078430340665895e-05,
      "loss": 0.6172,
      "step": 16841
    },
    {
      "epoch": 5.317966529838964,
      "grad_norm": 4.443215370178223,
      "learning_rate": 4.078342630600288e-05,
      "loss": 0.6342,
      "step": 16842
    },
    {
      "epoch": 5.31828228607515,
      "grad_norm": 4.397876262664795,
      "learning_rate": 4.078254920534681e-05,
      "loss": 0.5846,
      "step": 16843
    },
    {
      "epoch": 5.3185980423113355,
      "grad_norm": 3.9236867427825928,
      "learning_rate": 4.078167210469074e-05,
      "loss": 0.6138,
      "step": 16844
    },
    {
      "epoch": 5.318913798547522,
      "grad_norm": 3.776745319366455,
      "learning_rate": 4.078079500403467e-05,
      "loss": 0.6753,
      "step": 16845
    },
    {
      "epoch": 5.319229554783707,
      "grad_norm": 4.296751022338867,
      "learning_rate": 4.0779917903378594e-05,
      "loss": 0.5778,
      "step": 16846
    },
    {
      "epoch": 5.319545311019892,
      "grad_norm": 5.008084297180176,
      "learning_rate": 4.0779040802722526e-05,
      "loss": 0.6772,
      "step": 16847
    },
    {
      "epoch": 5.319861067256078,
      "grad_norm": 4.473416328430176,
      "learning_rate": 4.077816370206645e-05,
      "loss": 0.6541,
      "step": 16848
    },
    {
      "epoch": 5.320176823492264,
      "grad_norm": 4.097136497497559,
      "learning_rate": 4.077728660141038e-05,
      "loss": 0.5776,
      "step": 16849
    },
    {
      "epoch": 5.32049257972845,
      "grad_norm": 4.578182697296143,
      "learning_rate": 4.077640950075431e-05,
      "loss": 0.6119,
      "step": 16850
    },
    {
      "epoch": 5.320808335964635,
      "grad_norm": 4.320372104644775,
      "learning_rate": 4.077553240009824e-05,
      "loss": 0.5994,
      "step": 16851
    },
    {
      "epoch": 5.3211240922008205,
      "grad_norm": 3.6371679306030273,
      "learning_rate": 4.0774655299442164e-05,
      "loss": 0.5948,
      "step": 16852
    },
    {
      "epoch": 5.321439848437007,
      "grad_norm": 4.138835906982422,
      "learning_rate": 4.0773778198786096e-05,
      "loss": 0.6133,
      "step": 16853
    },
    {
      "epoch": 5.321755604673192,
      "grad_norm": 4.4117889404296875,
      "learning_rate": 4.077290109813002e-05,
      "loss": 0.5802,
      "step": 16854
    },
    {
      "epoch": 5.322071360909378,
      "grad_norm": 4.444232940673828,
      "learning_rate": 4.077202399747395e-05,
      "loss": 0.6177,
      "step": 16855
    },
    {
      "epoch": 5.3223871171455635,
      "grad_norm": 4.6536478996276855,
      "learning_rate": 4.077114689681788e-05,
      "loss": 0.7031,
      "step": 16856
    },
    {
      "epoch": 5.32270287338175,
      "grad_norm": 4.118711948394775,
      "learning_rate": 4.077026979616181e-05,
      "loss": 0.6576,
      "step": 16857
    },
    {
      "epoch": 5.323018629617935,
      "grad_norm": 4.554242134094238,
      "learning_rate": 4.0769392695505734e-05,
      "loss": 0.6296,
      "step": 16858
    },
    {
      "epoch": 5.32333438585412,
      "grad_norm": 3.9906980991363525,
      "learning_rate": 4.0768515594849665e-05,
      "loss": 0.5133,
      "step": 16859
    },
    {
      "epoch": 5.323650142090306,
      "grad_norm": 4.487895488739014,
      "learning_rate": 4.07676384941936e-05,
      "loss": 0.6632,
      "step": 16860
    },
    {
      "epoch": 5.323965898326492,
      "grad_norm": 5.632083892822266,
      "learning_rate": 4.076676139353752e-05,
      "loss": 0.6851,
      "step": 16861
    },
    {
      "epoch": 5.324281654562678,
      "grad_norm": 5.016716480255127,
      "learning_rate": 4.0765884292881454e-05,
      "loss": 0.6778,
      "step": 16862
    },
    {
      "epoch": 5.324597410798863,
      "grad_norm": 4.16107177734375,
      "learning_rate": 4.076500719222538e-05,
      "loss": 0.6975,
      "step": 16863
    },
    {
      "epoch": 5.324913167035049,
      "grad_norm": 4.424307346343994,
      "learning_rate": 4.076413009156931e-05,
      "loss": 0.6073,
      "step": 16864
    },
    {
      "epoch": 5.325228923271235,
      "grad_norm": 4.0048747062683105,
      "learning_rate": 4.0763252990913235e-05,
      "loss": 0.6022,
      "step": 16865
    },
    {
      "epoch": 5.32554467950742,
      "grad_norm": 4.394938945770264,
      "learning_rate": 4.0762375890257174e-05,
      "loss": 0.6598,
      "step": 16866
    },
    {
      "epoch": 5.325860435743606,
      "grad_norm": 3.898545026779175,
      "learning_rate": 4.07614987896011e-05,
      "loss": 0.5858,
      "step": 16867
    },
    {
      "epoch": 5.326176191979791,
      "grad_norm": 3.529339075088501,
      "learning_rate": 4.076062168894503e-05,
      "loss": 0.5925,
      "step": 16868
    },
    {
      "epoch": 5.326491948215978,
      "grad_norm": 4.303992748260498,
      "learning_rate": 4.0759744588288955e-05,
      "loss": 0.7277,
      "step": 16869
    },
    {
      "epoch": 5.326807704452163,
      "grad_norm": 4.5579514503479,
      "learning_rate": 4.075886748763289e-05,
      "loss": 0.6276,
      "step": 16870
    },
    {
      "epoch": 5.327123460688348,
      "grad_norm": 4.64497184753418,
      "learning_rate": 4.075799038697681e-05,
      "loss": 0.6777,
      "step": 16871
    },
    {
      "epoch": 5.327439216924534,
      "grad_norm": 3.725930690765381,
      "learning_rate": 4.075711328632074e-05,
      "loss": 0.6715,
      "step": 16872
    },
    {
      "epoch": 5.32775497316072,
      "grad_norm": 4.800113201141357,
      "learning_rate": 4.075623618566467e-05,
      "loss": 0.6606,
      "step": 16873
    },
    {
      "epoch": 5.328070729396906,
      "grad_norm": 4.729411602020264,
      "learning_rate": 4.07553590850086e-05,
      "loss": 0.6989,
      "step": 16874
    },
    {
      "epoch": 5.328386485633091,
      "grad_norm": 4.068177700042725,
      "learning_rate": 4.0754481984352525e-05,
      "loss": 0.5448,
      "step": 16875
    },
    {
      "epoch": 5.328702241869277,
      "grad_norm": 3.9476962089538574,
      "learning_rate": 4.0753604883696457e-05,
      "loss": 0.6866,
      "step": 16876
    },
    {
      "epoch": 5.329017998105463,
      "grad_norm": 4.556303977966309,
      "learning_rate": 4.075272778304038e-05,
      "loss": 0.6966,
      "step": 16877
    },
    {
      "epoch": 5.329333754341648,
      "grad_norm": 3.8238556385040283,
      "learning_rate": 4.075185068238431e-05,
      "loss": 0.5586,
      "step": 16878
    },
    {
      "epoch": 5.329649510577834,
      "grad_norm": 2.8769309520721436,
      "learning_rate": 4.075097358172824e-05,
      "loss": 0.5848,
      "step": 16879
    },
    {
      "epoch": 5.329965266814019,
      "grad_norm": 3.262935161590576,
      "learning_rate": 4.075009648107217e-05,
      "loss": 0.6208,
      "step": 16880
    },
    {
      "epoch": 5.330281023050206,
      "grad_norm": 8.371664047241211,
      "learning_rate": 4.07492193804161e-05,
      "loss": 0.6521,
      "step": 16881
    },
    {
      "epoch": 5.330596779286391,
      "grad_norm": 3.4181578159332275,
      "learning_rate": 4.0748342279760026e-05,
      "loss": 0.6271,
      "step": 16882
    },
    {
      "epoch": 5.330912535522576,
      "grad_norm": 3.658787965774536,
      "learning_rate": 4.074746517910396e-05,
      "loss": 0.6388,
      "step": 16883
    },
    {
      "epoch": 5.331228291758762,
      "grad_norm": 3.5830836296081543,
      "learning_rate": 4.074658807844788e-05,
      "loss": 0.6319,
      "step": 16884
    },
    {
      "epoch": 5.331544047994948,
      "grad_norm": 3.7996835708618164,
      "learning_rate": 4.0745710977791815e-05,
      "loss": 0.613,
      "step": 16885
    },
    {
      "epoch": 5.331859804231134,
      "grad_norm": 4.333301544189453,
      "learning_rate": 4.074483387713574e-05,
      "loss": 0.7831,
      "step": 16886
    },
    {
      "epoch": 5.332175560467319,
      "grad_norm": 3.645554542541504,
      "learning_rate": 4.074395677647967e-05,
      "loss": 0.7081,
      "step": 16887
    },
    {
      "epoch": 5.332491316703505,
      "grad_norm": 3.2954466342926025,
      "learning_rate": 4.0743079675823596e-05,
      "loss": 0.548,
      "step": 16888
    },
    {
      "epoch": 5.332807072939691,
      "grad_norm": 4.4217119216918945,
      "learning_rate": 4.074220257516753e-05,
      "loss": 0.6115,
      "step": 16889
    },
    {
      "epoch": 5.333122829175876,
      "grad_norm": 3.656639337539673,
      "learning_rate": 4.074132547451145e-05,
      "loss": 0.5265,
      "step": 16890
    },
    {
      "epoch": 5.333438585412062,
      "grad_norm": 3.7618441581726074,
      "learning_rate": 4.0740448373855384e-05,
      "loss": 0.6403,
      "step": 16891
    },
    {
      "epoch": 5.333754341648247,
      "grad_norm": 3.4224090576171875,
      "learning_rate": 4.073957127319931e-05,
      "loss": 0.6084,
      "step": 16892
    },
    {
      "epoch": 5.3340700978844335,
      "grad_norm": 4.524933815002441,
      "learning_rate": 4.073869417254324e-05,
      "loss": 0.6779,
      "step": 16893
    },
    {
      "epoch": 5.334385854120619,
      "grad_norm": 4.059446334838867,
      "learning_rate": 4.073781707188717e-05,
      "loss": 0.6274,
      "step": 16894
    },
    {
      "epoch": 5.334701610356804,
      "grad_norm": 3.915192127227783,
      "learning_rate": 4.0736939971231104e-05,
      "loss": 0.6834,
      "step": 16895
    },
    {
      "epoch": 5.33501736659299,
      "grad_norm": 4.028563976287842,
      "learning_rate": 4.073606287057503e-05,
      "loss": 0.553,
      "step": 16896
    },
    {
      "epoch": 5.335333122829176,
      "grad_norm": 4.334034442901611,
      "learning_rate": 4.073518576991896e-05,
      "loss": 0.5434,
      "step": 16897
    },
    {
      "epoch": 5.335648879065362,
      "grad_norm": 3.3352890014648438,
      "learning_rate": 4.0734308669262886e-05,
      "loss": 0.5961,
      "step": 16898
    },
    {
      "epoch": 5.335964635301547,
      "grad_norm": 4.888930797576904,
      "learning_rate": 4.073343156860682e-05,
      "loss": 0.682,
      "step": 16899
    },
    {
      "epoch": 5.336280391537733,
      "grad_norm": 3.838960886001587,
      "learning_rate": 4.073255446795074e-05,
      "loss": 0.6862,
      "step": 16900
    },
    {
      "epoch": 5.3365961477739186,
      "grad_norm": 4.134366035461426,
      "learning_rate": 4.0731677367294674e-05,
      "loss": 0.6293,
      "step": 16901
    },
    {
      "epoch": 5.336911904010104,
      "grad_norm": 3.6099274158477783,
      "learning_rate": 4.0730800266638606e-05,
      "loss": 0.6293,
      "step": 16902
    },
    {
      "epoch": 5.33722766024629,
      "grad_norm": 4.718029499053955,
      "learning_rate": 4.072992316598253e-05,
      "loss": 0.6918,
      "step": 16903
    },
    {
      "epoch": 5.337543416482475,
      "grad_norm": 4.365193843841553,
      "learning_rate": 4.072904606532646e-05,
      "loss": 0.5658,
      "step": 16904
    },
    {
      "epoch": 5.3378591727186615,
      "grad_norm": 4.132580280303955,
      "learning_rate": 4.072816896467039e-05,
      "loss": 0.6283,
      "step": 16905
    },
    {
      "epoch": 5.338174928954847,
      "grad_norm": 5.230701923370361,
      "learning_rate": 4.072729186401432e-05,
      "loss": 0.6858,
      "step": 16906
    },
    {
      "epoch": 5.338490685191033,
      "grad_norm": 5.0358967781066895,
      "learning_rate": 4.0726414763358244e-05,
      "loss": 0.5585,
      "step": 16907
    },
    {
      "epoch": 5.338806441427218,
      "grad_norm": 3.540379047393799,
      "learning_rate": 4.0725537662702175e-05,
      "loss": 0.5413,
      "step": 16908
    },
    {
      "epoch": 5.339122197663404,
      "grad_norm": 4.328671455383301,
      "learning_rate": 4.07246605620461e-05,
      "loss": 0.6518,
      "step": 16909
    },
    {
      "epoch": 5.33943795389959,
      "grad_norm": 4.824901103973389,
      "learning_rate": 4.072378346139003e-05,
      "loss": 0.6151,
      "step": 16910
    },
    {
      "epoch": 5.339753710135775,
      "grad_norm": 4.967751979827881,
      "learning_rate": 4.072290636073396e-05,
      "loss": 0.6151,
      "step": 16911
    },
    {
      "epoch": 5.340069466371961,
      "grad_norm": 4.400382041931152,
      "learning_rate": 4.072202926007789e-05,
      "loss": 0.6154,
      "step": 16912
    },
    {
      "epoch": 5.3403852226081465,
      "grad_norm": 4.722268104553223,
      "learning_rate": 4.0721152159421814e-05,
      "loss": 0.5871,
      "step": 16913
    },
    {
      "epoch": 5.340700978844332,
      "grad_norm": 5.061002731323242,
      "learning_rate": 4.0720275058765745e-05,
      "loss": 0.618,
      "step": 16914
    },
    {
      "epoch": 5.341016735080518,
      "grad_norm": 4.288195610046387,
      "learning_rate": 4.071939795810967e-05,
      "loss": 0.6715,
      "step": 16915
    },
    {
      "epoch": 5.341332491316703,
      "grad_norm": 3.8534233570098877,
      "learning_rate": 4.07185208574536e-05,
      "loss": 0.6812,
      "step": 16916
    },
    {
      "epoch": 5.3416482475528895,
      "grad_norm": 3.5379090309143066,
      "learning_rate": 4.071764375679753e-05,
      "loss": 0.5562,
      "step": 16917
    },
    {
      "epoch": 5.341964003789075,
      "grad_norm": 4.278344631195068,
      "learning_rate": 4.071676665614146e-05,
      "loss": 0.7262,
      "step": 16918
    },
    {
      "epoch": 5.34227976002526,
      "grad_norm": 3.964138984680176,
      "learning_rate": 4.071588955548539e-05,
      "loss": 0.6415,
      "step": 16919
    },
    {
      "epoch": 5.342595516261446,
      "grad_norm": 5.612091541290283,
      "learning_rate": 4.0715012454829315e-05,
      "loss": 0.6979,
      "step": 16920
    },
    {
      "epoch": 5.3429112724976315,
      "grad_norm": 5.7634053230285645,
      "learning_rate": 4.071413535417325e-05,
      "loss": 0.6258,
      "step": 16921
    },
    {
      "epoch": 5.343227028733818,
      "grad_norm": 3.907884120941162,
      "learning_rate": 4.071325825351718e-05,
      "loss": 0.5984,
      "step": 16922
    },
    {
      "epoch": 5.343542784970003,
      "grad_norm": 4.078217506408691,
      "learning_rate": 4.07123811528611e-05,
      "loss": 0.6498,
      "step": 16923
    },
    {
      "epoch": 5.343858541206189,
      "grad_norm": 4.4721999168396,
      "learning_rate": 4.0711504052205035e-05,
      "loss": 0.6613,
      "step": 16924
    },
    {
      "epoch": 5.3441742974423745,
      "grad_norm": 3.8779633045196533,
      "learning_rate": 4.071062695154897e-05,
      "loss": 0.6054,
      "step": 16925
    },
    {
      "epoch": 5.34449005367856,
      "grad_norm": 4.225564002990723,
      "learning_rate": 4.070974985089289e-05,
      "loss": 0.7404,
      "step": 16926
    },
    {
      "epoch": 5.344805809914746,
      "grad_norm": 4.592325210571289,
      "learning_rate": 4.070887275023682e-05,
      "loss": 0.6592,
      "step": 16927
    },
    {
      "epoch": 5.345121566150931,
      "grad_norm": 3.984166145324707,
      "learning_rate": 4.070799564958075e-05,
      "loss": 0.6313,
      "step": 16928
    },
    {
      "epoch": 5.345437322387117,
      "grad_norm": 4.400473117828369,
      "learning_rate": 4.070711854892468e-05,
      "loss": 0.7398,
      "step": 16929
    },
    {
      "epoch": 5.345753078623303,
      "grad_norm": 4.807328701019287,
      "learning_rate": 4.0706241448268605e-05,
      "loss": 0.6844,
      "step": 16930
    },
    {
      "epoch": 5.346068834859489,
      "grad_norm": 3.998082399368286,
      "learning_rate": 4.0705364347612536e-05,
      "loss": 0.6618,
      "step": 16931
    },
    {
      "epoch": 5.346384591095674,
      "grad_norm": 4.205057144165039,
      "learning_rate": 4.070448724695646e-05,
      "loss": 0.6359,
      "step": 16932
    },
    {
      "epoch": 5.3467003473318595,
      "grad_norm": 5.052107334136963,
      "learning_rate": 4.070361014630039e-05,
      "loss": 0.6464,
      "step": 16933
    },
    {
      "epoch": 5.347016103568046,
      "grad_norm": 3.7018768787384033,
      "learning_rate": 4.070273304564432e-05,
      "loss": 0.5965,
      "step": 16934
    },
    {
      "epoch": 5.347331859804231,
      "grad_norm": 4.460702896118164,
      "learning_rate": 4.070185594498825e-05,
      "loss": 0.6583,
      "step": 16935
    },
    {
      "epoch": 5.347647616040417,
      "grad_norm": 4.009286403656006,
      "learning_rate": 4.0700978844332175e-05,
      "loss": 0.5752,
      "step": 16936
    },
    {
      "epoch": 5.3479633722766025,
      "grad_norm": 3.490079879760742,
      "learning_rate": 4.0700101743676106e-05,
      "loss": 0.6278,
      "step": 16937
    },
    {
      "epoch": 5.348279128512788,
      "grad_norm": 3.1671218872070312,
      "learning_rate": 4.069922464302003e-05,
      "loss": 0.62,
      "step": 16938
    },
    {
      "epoch": 5.348594884748974,
      "grad_norm": 4.436007022857666,
      "learning_rate": 4.069834754236396e-05,
      "loss": 0.6403,
      "step": 16939
    },
    {
      "epoch": 5.348910640985159,
      "grad_norm": 3.8840370178222656,
      "learning_rate": 4.0697470441707894e-05,
      "loss": 0.6051,
      "step": 16940
    },
    {
      "epoch": 5.349226397221345,
      "grad_norm": 4.665114402770996,
      "learning_rate": 4.069659334105182e-05,
      "loss": 0.5928,
      "step": 16941
    },
    {
      "epoch": 5.349542153457531,
      "grad_norm": 4.638494491577148,
      "learning_rate": 4.069571624039575e-05,
      "loss": 0.6666,
      "step": 16942
    },
    {
      "epoch": 5.349857909693716,
      "grad_norm": 3.8151628971099854,
      "learning_rate": 4.0694839139739676e-05,
      "loss": 0.5871,
      "step": 16943
    },
    {
      "epoch": 5.350173665929902,
      "grad_norm": 5.30582332611084,
      "learning_rate": 4.069396203908361e-05,
      "loss": 0.6837,
      "step": 16944
    },
    {
      "epoch": 5.3504894221660875,
      "grad_norm": 3.246358633041382,
      "learning_rate": 4.069308493842753e-05,
      "loss": 0.6308,
      "step": 16945
    },
    {
      "epoch": 5.350805178402274,
      "grad_norm": 3.7406575679779053,
      "learning_rate": 4.0692207837771464e-05,
      "loss": 0.5436,
      "step": 16946
    },
    {
      "epoch": 5.351120934638459,
      "grad_norm": 3.8298730850219727,
      "learning_rate": 4.069133073711539e-05,
      "loss": 0.5885,
      "step": 16947
    },
    {
      "epoch": 5.351436690874645,
      "grad_norm": 4.007329940795898,
      "learning_rate": 4.069045363645933e-05,
      "loss": 0.4948,
      "step": 16948
    },
    {
      "epoch": 5.35175244711083,
      "grad_norm": 4.096473217010498,
      "learning_rate": 4.068957653580325e-05,
      "loss": 0.5897,
      "step": 16949
    },
    {
      "epoch": 5.352068203347016,
      "grad_norm": 4.6771135330200195,
      "learning_rate": 4.0688699435147184e-05,
      "loss": 0.6149,
      "step": 16950
    },
    {
      "epoch": 5.352383959583202,
      "grad_norm": 4.006464004516602,
      "learning_rate": 4.068782233449111e-05,
      "loss": 0.5976,
      "step": 16951
    },
    {
      "epoch": 5.352699715819387,
      "grad_norm": 4.119725227355957,
      "learning_rate": 4.068694523383504e-05,
      "loss": 0.5964,
      "step": 16952
    },
    {
      "epoch": 5.353015472055573,
      "grad_norm": 4.741885185241699,
      "learning_rate": 4.0686068133178966e-05,
      "loss": 0.6581,
      "step": 16953
    },
    {
      "epoch": 5.353331228291759,
      "grad_norm": 4.3465657234191895,
      "learning_rate": 4.06851910325229e-05,
      "loss": 0.5328,
      "step": 16954
    },
    {
      "epoch": 5.353646984527945,
      "grad_norm": 4.348221302032471,
      "learning_rate": 4.068431393186682e-05,
      "loss": 0.6949,
      "step": 16955
    },
    {
      "epoch": 5.35396274076413,
      "grad_norm": 4.107059955596924,
      "learning_rate": 4.0683436831210754e-05,
      "loss": 0.703,
      "step": 16956
    },
    {
      "epoch": 5.354278497000315,
      "grad_norm": 4.041977882385254,
      "learning_rate": 4.068255973055468e-05,
      "loss": 0.6654,
      "step": 16957
    },
    {
      "epoch": 5.354594253236502,
      "grad_norm": 4.4284987449646,
      "learning_rate": 4.068168262989861e-05,
      "loss": 0.6372,
      "step": 16958
    },
    {
      "epoch": 5.354910009472687,
      "grad_norm": 3.7742443084716797,
      "learning_rate": 4.0680805529242535e-05,
      "loss": 0.5419,
      "step": 16959
    },
    {
      "epoch": 5.355225765708873,
      "grad_norm": 4.770443916320801,
      "learning_rate": 4.067992842858647e-05,
      "loss": 0.7035,
      "step": 16960
    },
    {
      "epoch": 5.355541521945058,
      "grad_norm": 4.291518688201904,
      "learning_rate": 4.067905132793039e-05,
      "loss": 0.6324,
      "step": 16961
    },
    {
      "epoch": 5.355857278181244,
      "grad_norm": 3.7402286529541016,
      "learning_rate": 4.0678174227274324e-05,
      "loss": 0.694,
      "step": 16962
    },
    {
      "epoch": 5.35617303441743,
      "grad_norm": 3.1865017414093018,
      "learning_rate": 4.0677297126618255e-05,
      "loss": 0.5737,
      "step": 16963
    },
    {
      "epoch": 5.356488790653615,
      "grad_norm": 3.71358585357666,
      "learning_rate": 4.067642002596218e-05,
      "loss": 0.5602,
      "step": 16964
    },
    {
      "epoch": 5.356804546889801,
      "grad_norm": 3.996624708175659,
      "learning_rate": 4.067554292530611e-05,
      "loss": 0.637,
      "step": 16965
    },
    {
      "epoch": 5.357120303125987,
      "grad_norm": 4.327112197875977,
      "learning_rate": 4.067466582465004e-05,
      "loss": 0.5474,
      "step": 16966
    },
    {
      "epoch": 5.357436059362172,
      "grad_norm": 3.6007449626922607,
      "learning_rate": 4.067378872399397e-05,
      "loss": 0.586,
      "step": 16967
    },
    {
      "epoch": 5.357751815598358,
      "grad_norm": 4.486313819885254,
      "learning_rate": 4.0672911623337893e-05,
      "loss": 0.708,
      "step": 16968
    },
    {
      "epoch": 5.358067571834543,
      "grad_norm": 4.192931652069092,
      "learning_rate": 4.0672034522681825e-05,
      "loss": 0.642,
      "step": 16969
    },
    {
      "epoch": 5.35838332807073,
      "grad_norm": 4.628664016723633,
      "learning_rate": 4.067115742202575e-05,
      "loss": 0.5932,
      "step": 16970
    },
    {
      "epoch": 5.358699084306915,
      "grad_norm": 4.131717205047607,
      "learning_rate": 4.067028032136968e-05,
      "loss": 0.6055,
      "step": 16971
    },
    {
      "epoch": 5.359014840543101,
      "grad_norm": 3.851019859313965,
      "learning_rate": 4.066940322071361e-05,
      "loss": 0.593,
      "step": 16972
    },
    {
      "epoch": 5.359330596779286,
      "grad_norm": 5.051466464996338,
      "learning_rate": 4.066852612005754e-05,
      "loss": 0.6515,
      "step": 16973
    },
    {
      "epoch": 5.359646353015472,
      "grad_norm": 4.116206169128418,
      "learning_rate": 4.066764901940146e-05,
      "loss": 0.6611,
      "step": 16974
    },
    {
      "epoch": 5.359962109251658,
      "grad_norm": 4.116880893707275,
      "learning_rate": 4.06667719187454e-05,
      "loss": 0.5684,
      "step": 16975
    },
    {
      "epoch": 5.360277865487843,
      "grad_norm": 4.302674293518066,
      "learning_rate": 4.0665894818089327e-05,
      "loss": 0.6199,
      "step": 16976
    },
    {
      "epoch": 5.360593621724029,
      "grad_norm": 4.2996745109558105,
      "learning_rate": 4.066501771743326e-05,
      "loss": 0.6447,
      "step": 16977
    },
    {
      "epoch": 5.360909377960215,
      "grad_norm": 4.732310771942139,
      "learning_rate": 4.066414061677718e-05,
      "loss": 0.5607,
      "step": 16978
    },
    {
      "epoch": 5.361225134196401,
      "grad_norm": 4.2492523193359375,
      "learning_rate": 4.0663263516121115e-05,
      "loss": 0.7184,
      "step": 16979
    },
    {
      "epoch": 5.361540890432586,
      "grad_norm": 4.89603853225708,
      "learning_rate": 4.066238641546504e-05,
      "loss": 0.6325,
      "step": 16980
    },
    {
      "epoch": 5.361856646668771,
      "grad_norm": 3.7941975593566895,
      "learning_rate": 4.066150931480897e-05,
      "loss": 0.6858,
      "step": 16981
    },
    {
      "epoch": 5.362172402904958,
      "grad_norm": 4.1481451988220215,
      "learning_rate": 4.0660632214152896e-05,
      "loss": 0.647,
      "step": 16982
    },
    {
      "epoch": 5.362488159141143,
      "grad_norm": 4.532992839813232,
      "learning_rate": 4.065975511349683e-05,
      "loss": 0.5757,
      "step": 16983
    },
    {
      "epoch": 5.362803915377329,
      "grad_norm": 4.264387607574463,
      "learning_rate": 4.065887801284076e-05,
      "loss": 0.5695,
      "step": 16984
    },
    {
      "epoch": 5.363119671613514,
      "grad_norm": 4.379223346710205,
      "learning_rate": 4.0658000912184685e-05,
      "loss": 0.6506,
      "step": 16985
    },
    {
      "epoch": 5.3634354278497,
      "grad_norm": 4.8605122566223145,
      "learning_rate": 4.0657123811528616e-05,
      "loss": 0.589,
      "step": 16986
    },
    {
      "epoch": 5.363751184085886,
      "grad_norm": 4.6696624755859375,
      "learning_rate": 4.065624671087254e-05,
      "loss": 0.6468,
      "step": 16987
    },
    {
      "epoch": 5.364066940322071,
      "grad_norm": 3.4983365535736084,
      "learning_rate": 4.065536961021647e-05,
      "loss": 0.5862,
      "step": 16988
    },
    {
      "epoch": 5.364382696558257,
      "grad_norm": 4.288956642150879,
      "learning_rate": 4.06544925095604e-05,
      "loss": 0.6221,
      "step": 16989
    },
    {
      "epoch": 5.364698452794443,
      "grad_norm": 4.666842937469482,
      "learning_rate": 4.065361540890433e-05,
      "loss": 0.6585,
      "step": 16990
    },
    {
      "epoch": 5.365014209030629,
      "grad_norm": 4.02744197845459,
      "learning_rate": 4.0652738308248254e-05,
      "loss": 0.5922,
      "step": 16991
    },
    {
      "epoch": 5.365329965266814,
      "grad_norm": 4.776755332946777,
      "learning_rate": 4.0651861207592186e-05,
      "loss": 0.718,
      "step": 16992
    },
    {
      "epoch": 5.365645721502999,
      "grad_norm": 4.924834728240967,
      "learning_rate": 4.065098410693611e-05,
      "loss": 0.5941,
      "step": 16993
    },
    {
      "epoch": 5.3659614777391855,
      "grad_norm": 3.627931833267212,
      "learning_rate": 4.065010700628004e-05,
      "loss": 0.5741,
      "step": 16994
    },
    {
      "epoch": 5.366277233975371,
      "grad_norm": 4.160453796386719,
      "learning_rate": 4.064922990562397e-05,
      "loss": 0.651,
      "step": 16995
    },
    {
      "epoch": 5.366592990211557,
      "grad_norm": 4.685804843902588,
      "learning_rate": 4.06483528049679e-05,
      "loss": 0.6177,
      "step": 16996
    },
    {
      "epoch": 5.366908746447742,
      "grad_norm": 3.93908429145813,
      "learning_rate": 4.0647475704311824e-05,
      "loss": 0.6725,
      "step": 16997
    },
    {
      "epoch": 5.3672245026839285,
      "grad_norm": 3.829242706298828,
      "learning_rate": 4.0646598603655756e-05,
      "loss": 0.5872,
      "step": 16998
    },
    {
      "epoch": 5.367540258920114,
      "grad_norm": 5.503363609313965,
      "learning_rate": 4.064572150299968e-05,
      "loss": 0.6548,
      "step": 16999
    },
    {
      "epoch": 5.367856015156299,
      "grad_norm": 4.340398788452148,
      "learning_rate": 4.064484440234361e-05,
      "loss": 0.5988,
      "step": 17000
    },
    {
      "epoch": 5.367856015156299,
      "eval_loss": 0.8757421970367432,
      "eval_runtime": 4.6147,
      "eval_samples_per_second": 539.577,
      "eval_steps_per_second": 4.551,
      "step": 17000
    },
    {
      "epoch": 5.368171771392485,
      "grad_norm": 5.0376176834106445,
      "learning_rate": 4.0643967301687544e-05,
      "loss": 0.6203,
      "step": 17001
    },
    {
      "epoch": 5.3684875276286705,
      "grad_norm": 4.6939826011657715,
      "learning_rate": 4.064309020103147e-05,
      "loss": 0.6958,
      "step": 17002
    },
    {
      "epoch": 5.368803283864857,
      "grad_norm": 4.292008399963379,
      "learning_rate": 4.06422131003754e-05,
      "loss": 0.6473,
      "step": 17003
    },
    {
      "epoch": 5.369119040101042,
      "grad_norm": 4.6607160568237305,
      "learning_rate": 4.064133599971933e-05,
      "loss": 0.5551,
      "step": 17004
    },
    {
      "epoch": 5.369434796337227,
      "grad_norm": 4.348507881164551,
      "learning_rate": 4.064045889906326e-05,
      "loss": 0.6146,
      "step": 17005
    },
    {
      "epoch": 5.3697505525734135,
      "grad_norm": 4.027072429656982,
      "learning_rate": 4.063958179840719e-05,
      "loss": 0.6985,
      "step": 17006
    },
    {
      "epoch": 5.370066308809599,
      "grad_norm": 3.7925562858581543,
      "learning_rate": 4.063870469775112e-05,
      "loss": 0.6353,
      "step": 17007
    },
    {
      "epoch": 5.370382065045785,
      "grad_norm": 3.7747762203216553,
      "learning_rate": 4.0637827597095046e-05,
      "loss": 0.6068,
      "step": 17008
    },
    {
      "epoch": 5.37069782128197,
      "grad_norm": 4.216203689575195,
      "learning_rate": 4.063695049643898e-05,
      "loss": 0.743,
      "step": 17009
    },
    {
      "epoch": 5.371013577518156,
      "grad_norm": 3.7885591983795166,
      "learning_rate": 4.06360733957829e-05,
      "loss": 0.6175,
      "step": 17010
    },
    {
      "epoch": 5.371329333754342,
      "grad_norm": 5.4527668952941895,
      "learning_rate": 4.0635196295126834e-05,
      "loss": 0.7826,
      "step": 17011
    },
    {
      "epoch": 5.371645089990527,
      "grad_norm": 3.5361714363098145,
      "learning_rate": 4.063431919447076e-05,
      "loss": 0.6023,
      "step": 17012
    },
    {
      "epoch": 5.371960846226713,
      "grad_norm": 4.093204498291016,
      "learning_rate": 4.063344209381469e-05,
      "loss": 0.6018,
      "step": 17013
    },
    {
      "epoch": 5.3722766024628985,
      "grad_norm": 3.3186051845550537,
      "learning_rate": 4.0632564993158615e-05,
      "loss": 0.5953,
      "step": 17014
    },
    {
      "epoch": 5.372592358699085,
      "grad_norm": 3.940688371658325,
      "learning_rate": 4.063168789250255e-05,
      "loss": 0.7629,
      "step": 17015
    },
    {
      "epoch": 5.37290811493527,
      "grad_norm": 3.8000259399414062,
      "learning_rate": 4.063081079184647e-05,
      "loss": 0.612,
      "step": 17016
    },
    {
      "epoch": 5.373223871171455,
      "grad_norm": 4.244765281677246,
      "learning_rate": 4.0629933691190404e-05,
      "loss": 0.6957,
      "step": 17017
    },
    {
      "epoch": 5.3735396274076415,
      "grad_norm": 4.175693511962891,
      "learning_rate": 4.062905659053433e-05,
      "loss": 0.6023,
      "step": 17018
    },
    {
      "epoch": 5.373855383643827,
      "grad_norm": 3.7968392372131348,
      "learning_rate": 4.062817948987826e-05,
      "loss": 0.6298,
      "step": 17019
    },
    {
      "epoch": 5.374171139880013,
      "grad_norm": 3.592618227005005,
      "learning_rate": 4.0627302389222185e-05,
      "loss": 0.5618,
      "step": 17020
    },
    {
      "epoch": 5.374486896116198,
      "grad_norm": 3.687706708908081,
      "learning_rate": 4.062642528856612e-05,
      "loss": 0.6344,
      "step": 17021
    },
    {
      "epoch": 5.374802652352384,
      "grad_norm": 3.6241424083709717,
      "learning_rate": 4.062554818791005e-05,
      "loss": 0.6747,
      "step": 17022
    },
    {
      "epoch": 5.37511840858857,
      "grad_norm": 3.7333881855010986,
      "learning_rate": 4.062467108725397e-05,
      "loss": 0.6333,
      "step": 17023
    },
    {
      "epoch": 5.375434164824755,
      "grad_norm": 3.7295713424682617,
      "learning_rate": 4.0623793986597905e-05,
      "loss": 0.6246,
      "step": 17024
    },
    {
      "epoch": 5.375749921060941,
      "grad_norm": 3.568493366241455,
      "learning_rate": 4.062291688594183e-05,
      "loss": 0.7156,
      "step": 17025
    },
    {
      "epoch": 5.3760656772971265,
      "grad_norm": 4.20283317565918,
      "learning_rate": 4.062203978528576e-05,
      "loss": 0.6154,
      "step": 17026
    },
    {
      "epoch": 5.376381433533313,
      "grad_norm": 3.7798328399658203,
      "learning_rate": 4.0621162684629687e-05,
      "loss": 0.6554,
      "step": 17027
    },
    {
      "epoch": 5.376697189769498,
      "grad_norm": 3.8040528297424316,
      "learning_rate": 4.062028558397362e-05,
      "loss": 0.5863,
      "step": 17028
    },
    {
      "epoch": 5.377012946005683,
      "grad_norm": 4.2422685623168945,
      "learning_rate": 4.061940848331754e-05,
      "loss": 0.6185,
      "step": 17029
    },
    {
      "epoch": 5.377328702241869,
      "grad_norm": 3.443740129470825,
      "learning_rate": 4.061853138266148e-05,
      "loss": 0.601,
      "step": 17030
    },
    {
      "epoch": 5.377644458478055,
      "grad_norm": 4.451894760131836,
      "learning_rate": 4.0617654282005406e-05,
      "loss": 0.5486,
      "step": 17031
    },
    {
      "epoch": 5.377960214714241,
      "grad_norm": 3.664395809173584,
      "learning_rate": 4.061677718134934e-05,
      "loss": 0.7021,
      "step": 17032
    },
    {
      "epoch": 5.378275970950426,
      "grad_norm": 3.8385417461395264,
      "learning_rate": 4.061590008069326e-05,
      "loss": 0.5857,
      "step": 17033
    },
    {
      "epoch": 5.3785917271866115,
      "grad_norm": 4.428713798522949,
      "learning_rate": 4.0615022980037195e-05,
      "loss": 0.6537,
      "step": 17034
    },
    {
      "epoch": 5.378907483422798,
      "grad_norm": 4.539955139160156,
      "learning_rate": 4.061414587938112e-05,
      "loss": 0.5842,
      "step": 17035
    },
    {
      "epoch": 5.379223239658983,
      "grad_norm": 3.866851329803467,
      "learning_rate": 4.061326877872505e-05,
      "loss": 0.6211,
      "step": 17036
    },
    {
      "epoch": 5.379538995895169,
      "grad_norm": 3.670171022415161,
      "learning_rate": 4.0612391678068976e-05,
      "loss": 0.6282,
      "step": 17037
    },
    {
      "epoch": 5.3798547521313544,
      "grad_norm": 3.9432852268218994,
      "learning_rate": 4.061151457741291e-05,
      "loss": 0.726,
      "step": 17038
    },
    {
      "epoch": 5.380170508367541,
      "grad_norm": 4.017283916473389,
      "learning_rate": 4.061063747675683e-05,
      "loss": 0.7129,
      "step": 17039
    },
    {
      "epoch": 5.380486264603726,
      "grad_norm": 4.043154716491699,
      "learning_rate": 4.0609760376100765e-05,
      "loss": 0.6675,
      "step": 17040
    },
    {
      "epoch": 5.380802020839911,
      "grad_norm": 4.938569068908691,
      "learning_rate": 4.060888327544469e-05,
      "loss": 0.6261,
      "step": 17041
    },
    {
      "epoch": 5.381117777076097,
      "grad_norm": 4.101513385772705,
      "learning_rate": 4.060800617478862e-05,
      "loss": 0.6815,
      "step": 17042
    },
    {
      "epoch": 5.381433533312283,
      "grad_norm": 3.456509828567505,
      "learning_rate": 4.0607129074132546e-05,
      "loss": 0.5577,
      "step": 17043
    },
    {
      "epoch": 5.381749289548469,
      "grad_norm": 4.015601634979248,
      "learning_rate": 4.060625197347648e-05,
      "loss": 0.6215,
      "step": 17044
    },
    {
      "epoch": 5.382065045784654,
      "grad_norm": 3.4198145866394043,
      "learning_rate": 4.060537487282041e-05,
      "loss": 0.5285,
      "step": 17045
    },
    {
      "epoch": 5.38238080202084,
      "grad_norm": 4.288279056549072,
      "learning_rate": 4.0604497772164334e-05,
      "loss": 0.7105,
      "step": 17046
    },
    {
      "epoch": 5.382696558257026,
      "grad_norm": 3.5249054431915283,
      "learning_rate": 4.0603620671508266e-05,
      "loss": 0.5967,
      "step": 17047
    },
    {
      "epoch": 5.383012314493211,
      "grad_norm": 3.9934046268463135,
      "learning_rate": 4.060274357085219e-05,
      "loss": 0.5634,
      "step": 17048
    },
    {
      "epoch": 5.383328070729397,
      "grad_norm": 5.327988624572754,
      "learning_rate": 4.060186647019612e-05,
      "loss": 0.6173,
      "step": 17049
    },
    {
      "epoch": 5.383643826965582,
      "grad_norm": 5.823906898498535,
      "learning_rate": 4.060098936954005e-05,
      "loss": 0.7698,
      "step": 17050
    },
    {
      "epoch": 5.383959583201769,
      "grad_norm": 5.383062839508057,
      "learning_rate": 4.060011226888398e-05,
      "loss": 0.7053,
      "step": 17051
    },
    {
      "epoch": 5.384275339437954,
      "grad_norm": 4.216184139251709,
      "learning_rate": 4.0599235168227904e-05,
      "loss": 0.6226,
      "step": 17052
    },
    {
      "epoch": 5.384591095674139,
      "grad_norm": 3.7107222080230713,
      "learning_rate": 4.0598358067571836e-05,
      "loss": 0.5697,
      "step": 17053
    },
    {
      "epoch": 5.384906851910325,
      "grad_norm": 4.197086334228516,
      "learning_rate": 4.059748096691576e-05,
      "loss": 0.6443,
      "step": 17054
    },
    {
      "epoch": 5.385222608146511,
      "grad_norm": 4.3113298416137695,
      "learning_rate": 4.059660386625969e-05,
      "loss": 0.5975,
      "step": 17055
    },
    {
      "epoch": 5.385538364382697,
      "grad_norm": 4.530269145965576,
      "learning_rate": 4.059572676560362e-05,
      "loss": 0.5679,
      "step": 17056
    },
    {
      "epoch": 5.385854120618882,
      "grad_norm": 4.113692283630371,
      "learning_rate": 4.0594849664947556e-05,
      "loss": 0.6517,
      "step": 17057
    },
    {
      "epoch": 5.386169876855067,
      "grad_norm": 4.8499932289123535,
      "learning_rate": 4.059397256429148e-05,
      "loss": 0.646,
      "step": 17058
    },
    {
      "epoch": 5.386485633091254,
      "grad_norm": 4.677331924438477,
      "learning_rate": 4.059309546363541e-05,
      "loss": 0.7265,
      "step": 17059
    },
    {
      "epoch": 5.386801389327439,
      "grad_norm": 5.340413570404053,
      "learning_rate": 4.059221836297934e-05,
      "loss": 0.6913,
      "step": 17060
    },
    {
      "epoch": 5.387117145563625,
      "grad_norm": 4.8908305168151855,
      "learning_rate": 4.059134126232327e-05,
      "loss": 0.7044,
      "step": 17061
    },
    {
      "epoch": 5.38743290179981,
      "grad_norm": 3.711625576019287,
      "learning_rate": 4.0590464161667194e-05,
      "loss": 0.6842,
      "step": 17062
    },
    {
      "epoch": 5.387748658035997,
      "grad_norm": 4.344518184661865,
      "learning_rate": 4.0589587061011125e-05,
      "loss": 0.6365,
      "step": 17063
    },
    {
      "epoch": 5.388064414272182,
      "grad_norm": 3.8032925128936768,
      "learning_rate": 4.058870996035505e-05,
      "loss": 0.5701,
      "step": 17064
    },
    {
      "epoch": 5.388380170508367,
      "grad_norm": 4.547154903411865,
      "learning_rate": 4.058783285969898e-05,
      "loss": 0.6535,
      "step": 17065
    },
    {
      "epoch": 5.388695926744553,
      "grad_norm": 3.64188814163208,
      "learning_rate": 4.0586955759042914e-05,
      "loss": 0.5968,
      "step": 17066
    },
    {
      "epoch": 5.389011682980739,
      "grad_norm": 3.8907887935638428,
      "learning_rate": 4.058607865838684e-05,
      "loss": 0.5259,
      "step": 17067
    },
    {
      "epoch": 5.389327439216925,
      "grad_norm": 4.05265474319458,
      "learning_rate": 4.058520155773077e-05,
      "loss": 0.5868,
      "step": 17068
    },
    {
      "epoch": 5.38964319545311,
      "grad_norm": 3.800945520401001,
      "learning_rate": 4.0584324457074695e-05,
      "loss": 0.61,
      "step": 17069
    },
    {
      "epoch": 5.389958951689296,
      "grad_norm": 3.845470428466797,
      "learning_rate": 4.058344735641863e-05,
      "loss": 0.626,
      "step": 17070
    },
    {
      "epoch": 5.390274707925482,
      "grad_norm": 4.649044513702393,
      "learning_rate": 4.058257025576255e-05,
      "loss": 0.6914,
      "step": 17071
    },
    {
      "epoch": 5.390590464161667,
      "grad_norm": 4.981900691986084,
      "learning_rate": 4.0581693155106483e-05,
      "loss": 0.6822,
      "step": 17072
    },
    {
      "epoch": 5.390906220397853,
      "grad_norm": 3.673478126525879,
      "learning_rate": 4.058081605445041e-05,
      "loss": 0.69,
      "step": 17073
    },
    {
      "epoch": 5.391221976634038,
      "grad_norm": 4.831752777099609,
      "learning_rate": 4.057993895379434e-05,
      "loss": 0.6286,
      "step": 17074
    },
    {
      "epoch": 5.3915377328702245,
      "grad_norm": 3.772160768508911,
      "learning_rate": 4.0579061853138265e-05,
      "loss": 0.6505,
      "step": 17075
    },
    {
      "epoch": 5.39185348910641,
      "grad_norm": 4.480335712432861,
      "learning_rate": 4.05781847524822e-05,
      "loss": 0.5705,
      "step": 17076
    },
    {
      "epoch": 5.392169245342595,
      "grad_norm": 3.9277963638305664,
      "learning_rate": 4.057730765182612e-05,
      "loss": 0.5385,
      "step": 17077
    },
    {
      "epoch": 5.392485001578781,
      "grad_norm": 3.717398166656494,
      "learning_rate": 4.057643055117005e-05,
      "loss": 0.5946,
      "step": 17078
    },
    {
      "epoch": 5.392800757814967,
      "grad_norm": 3.8216583728790283,
      "learning_rate": 4.057555345051398e-05,
      "loss": 0.6316,
      "step": 17079
    },
    {
      "epoch": 5.393116514051153,
      "grad_norm": 4.338808059692383,
      "learning_rate": 4.057467634985791e-05,
      "loss": 0.7481,
      "step": 17080
    },
    {
      "epoch": 5.393432270287338,
      "grad_norm": 5.375535488128662,
      "learning_rate": 4.057379924920184e-05,
      "loss": 0.6725,
      "step": 17081
    },
    {
      "epoch": 5.393748026523524,
      "grad_norm": 4.037476539611816,
      "learning_rate": 4.0572922148545766e-05,
      "loss": 0.6256,
      "step": 17082
    },
    {
      "epoch": 5.3940637827597095,
      "grad_norm": 4.307494163513184,
      "learning_rate": 4.05720450478897e-05,
      "loss": 0.5852,
      "step": 17083
    },
    {
      "epoch": 5.394379538995895,
      "grad_norm": 4.399847507476807,
      "learning_rate": 4.057116794723363e-05,
      "loss": 0.6628,
      "step": 17084
    },
    {
      "epoch": 5.394695295232081,
      "grad_norm": 4.655351161956787,
      "learning_rate": 4.0570290846577555e-05,
      "loss": 0.6278,
      "step": 17085
    },
    {
      "epoch": 5.395011051468266,
      "grad_norm": 4.667290687561035,
      "learning_rate": 4.0569413745921486e-05,
      "loss": 0.6723,
      "step": 17086
    },
    {
      "epoch": 5.3953268077044525,
      "grad_norm": 4.155975818634033,
      "learning_rate": 4.056853664526542e-05,
      "loss": 0.6964,
      "step": 17087
    },
    {
      "epoch": 5.395642563940638,
      "grad_norm": 4.621748447418213,
      "learning_rate": 4.056765954460934e-05,
      "loss": 0.686,
      "step": 17088
    },
    {
      "epoch": 5.395958320176823,
      "grad_norm": 4.464782238006592,
      "learning_rate": 4.0566782443953275e-05,
      "loss": 0.7186,
      "step": 17089
    },
    {
      "epoch": 5.396274076413009,
      "grad_norm": 4.438290596008301,
      "learning_rate": 4.05659053432972e-05,
      "loss": 0.5953,
      "step": 17090
    },
    {
      "epoch": 5.396589832649195,
      "grad_norm": 4.872443199157715,
      "learning_rate": 4.056502824264113e-05,
      "loss": 0.6309,
      "step": 17091
    },
    {
      "epoch": 5.396905588885381,
      "grad_norm": 4.772153377532959,
      "learning_rate": 4.0564151141985056e-05,
      "loss": 0.7675,
      "step": 17092
    },
    {
      "epoch": 5.397221345121566,
      "grad_norm": 4.289575576782227,
      "learning_rate": 4.056327404132899e-05,
      "loss": 0.6613,
      "step": 17093
    },
    {
      "epoch": 5.397537101357752,
      "grad_norm": 3.9319376945495605,
      "learning_rate": 4.056239694067291e-05,
      "loss": 0.623,
      "step": 17094
    },
    {
      "epoch": 5.3978528575939375,
      "grad_norm": 4.503326892852783,
      "learning_rate": 4.0561519840016844e-05,
      "loss": 0.7342,
      "step": 17095
    },
    {
      "epoch": 5.398168613830123,
      "grad_norm": 3.4990234375,
      "learning_rate": 4.056064273936077e-05,
      "loss": 0.6011,
      "step": 17096
    },
    {
      "epoch": 5.398484370066309,
      "grad_norm": 3.753166437149048,
      "learning_rate": 4.05597656387047e-05,
      "loss": 0.6477,
      "step": 17097
    },
    {
      "epoch": 5.398800126302494,
      "grad_norm": 4.253259658813477,
      "learning_rate": 4.0558888538048626e-05,
      "loss": 0.7382,
      "step": 17098
    },
    {
      "epoch": 5.3991158825386805,
      "grad_norm": 4.005855560302734,
      "learning_rate": 4.055801143739256e-05,
      "loss": 0.7358,
      "step": 17099
    },
    {
      "epoch": 5.399431638774866,
      "grad_norm": 4.012662887573242,
      "learning_rate": 4.055713433673648e-05,
      "loss": 0.5371,
      "step": 17100
    },
    {
      "epoch": 5.399747395011051,
      "grad_norm": 3.5452208518981934,
      "learning_rate": 4.0556257236080414e-05,
      "loss": 0.644,
      "step": 17101
    },
    {
      "epoch": 5.400063151247237,
      "grad_norm": 4.085522651672363,
      "learning_rate": 4.055538013542434e-05,
      "loss": 0.6205,
      "step": 17102
    },
    {
      "epoch": 5.4003789074834225,
      "grad_norm": 3.7179486751556396,
      "learning_rate": 4.055450303476827e-05,
      "loss": 0.5849,
      "step": 17103
    },
    {
      "epoch": 5.400694663719609,
      "grad_norm": 4.053463459014893,
      "learning_rate": 4.05536259341122e-05,
      "loss": 0.676,
      "step": 17104
    },
    {
      "epoch": 5.401010419955794,
      "grad_norm": 3.5533759593963623,
      "learning_rate": 4.055274883345613e-05,
      "loss": 0.6389,
      "step": 17105
    },
    {
      "epoch": 5.40132617619198,
      "grad_norm": 4.129241466522217,
      "learning_rate": 4.055187173280006e-05,
      "loss": 0.7517,
      "step": 17106
    },
    {
      "epoch": 5.4016419324281655,
      "grad_norm": 4.337695121765137,
      "learning_rate": 4.0550994632143984e-05,
      "loss": 0.7355,
      "step": 17107
    },
    {
      "epoch": 5.401957688664351,
      "grad_norm": 4.097935676574707,
      "learning_rate": 4.0550117531487916e-05,
      "loss": 0.5573,
      "step": 17108
    },
    {
      "epoch": 5.402273444900537,
      "grad_norm": 4.845083236694336,
      "learning_rate": 4.054924043083184e-05,
      "loss": 0.6506,
      "step": 17109
    },
    {
      "epoch": 5.402589201136722,
      "grad_norm": 3.777449607849121,
      "learning_rate": 4.054836333017577e-05,
      "loss": 0.5692,
      "step": 17110
    },
    {
      "epoch": 5.402904957372908,
      "grad_norm": 4.450705051422119,
      "learning_rate": 4.05474862295197e-05,
      "loss": 0.6179,
      "step": 17111
    },
    {
      "epoch": 5.403220713609094,
      "grad_norm": 3.991100549697876,
      "learning_rate": 4.0546609128863636e-05,
      "loss": 0.6829,
      "step": 17112
    },
    {
      "epoch": 5.40353646984528,
      "grad_norm": 3.7160563468933105,
      "learning_rate": 4.054573202820756e-05,
      "loss": 0.6124,
      "step": 17113
    },
    {
      "epoch": 5.403852226081465,
      "grad_norm": 4.522101879119873,
      "learning_rate": 4.054485492755149e-05,
      "loss": 0.6565,
      "step": 17114
    },
    {
      "epoch": 5.4041679823176505,
      "grad_norm": 4.643537521362305,
      "learning_rate": 4.054397782689542e-05,
      "loss": 0.6195,
      "step": 17115
    },
    {
      "epoch": 5.404483738553837,
      "grad_norm": 3.571653127670288,
      "learning_rate": 4.054310072623935e-05,
      "loss": 0.6089,
      "step": 17116
    },
    {
      "epoch": 5.404799494790022,
      "grad_norm": 4.112448215484619,
      "learning_rate": 4.0542223625583274e-05,
      "loss": 0.6378,
      "step": 17117
    },
    {
      "epoch": 5.405115251026208,
      "grad_norm": 4.18625020980835,
      "learning_rate": 4.0541346524927205e-05,
      "loss": 0.7295,
      "step": 17118
    },
    {
      "epoch": 5.4054310072623935,
      "grad_norm": 3.9891021251678467,
      "learning_rate": 4.054046942427113e-05,
      "loss": 0.6944,
      "step": 17119
    },
    {
      "epoch": 5.405746763498579,
      "grad_norm": 3.2271597385406494,
      "learning_rate": 4.053959232361506e-05,
      "loss": 0.5593,
      "step": 17120
    },
    {
      "epoch": 5.406062519734765,
      "grad_norm": 4.039680004119873,
      "learning_rate": 4.053871522295899e-05,
      "loss": 0.5428,
      "step": 17121
    },
    {
      "epoch": 5.40637827597095,
      "grad_norm": 3.9538748264312744,
      "learning_rate": 4.053783812230292e-05,
      "loss": 0.7564,
      "step": 17122
    },
    {
      "epoch": 5.406694032207136,
      "grad_norm": 3.532883644104004,
      "learning_rate": 4.0536961021646843e-05,
      "loss": 0.5604,
      "step": 17123
    },
    {
      "epoch": 5.407009788443322,
      "grad_norm": 4.537313938140869,
      "learning_rate": 4.0536083920990775e-05,
      "loss": 0.6776,
      "step": 17124
    },
    {
      "epoch": 5.407325544679507,
      "grad_norm": 3.8754262924194336,
      "learning_rate": 4.053520682033471e-05,
      "loss": 0.6427,
      "step": 17125
    },
    {
      "epoch": 5.407641300915693,
      "grad_norm": 3.6120598316192627,
      "learning_rate": 4.053432971967863e-05,
      "loss": 0.5973,
      "step": 17126
    },
    {
      "epoch": 5.4079570571518785,
      "grad_norm": 4.230617046356201,
      "learning_rate": 4.053345261902256e-05,
      "loss": 0.611,
      "step": 17127
    },
    {
      "epoch": 5.408272813388065,
      "grad_norm": 5.292749881744385,
      "learning_rate": 4.053257551836649e-05,
      "loss": 0.6999,
      "step": 17128
    },
    {
      "epoch": 5.40858856962425,
      "grad_norm": 3.9318771362304688,
      "learning_rate": 4.053169841771042e-05,
      "loss": 0.6126,
      "step": 17129
    },
    {
      "epoch": 5.408904325860436,
      "grad_norm": 3.534818172454834,
      "learning_rate": 4.0530821317054345e-05,
      "loss": 0.534,
      "step": 17130
    },
    {
      "epoch": 5.409220082096621,
      "grad_norm": 4.541967391967773,
      "learning_rate": 4.0529944216398277e-05,
      "loss": 0.5897,
      "step": 17131
    },
    {
      "epoch": 5.409535838332807,
      "grad_norm": 5.068105220794678,
      "learning_rate": 4.05290671157422e-05,
      "loss": 0.6499,
      "step": 17132
    },
    {
      "epoch": 5.409851594568993,
      "grad_norm": 4.3497700691223145,
      "learning_rate": 4.052819001508613e-05,
      "loss": 0.5814,
      "step": 17133
    },
    {
      "epoch": 5.410167350805178,
      "grad_norm": 3.551142692565918,
      "learning_rate": 4.052731291443006e-05,
      "loss": 0.6163,
      "step": 17134
    },
    {
      "epoch": 5.410483107041364,
      "grad_norm": 5.405862331390381,
      "learning_rate": 4.052643581377399e-05,
      "loss": 0.6807,
      "step": 17135
    },
    {
      "epoch": 5.41079886327755,
      "grad_norm": 3.4625446796417236,
      "learning_rate": 4.0525558713117915e-05,
      "loss": 0.5723,
      "step": 17136
    },
    {
      "epoch": 5.411114619513736,
      "grad_norm": 5.040912628173828,
      "learning_rate": 4.0524681612461846e-05,
      "loss": 0.5868,
      "step": 17137
    },
    {
      "epoch": 5.411430375749921,
      "grad_norm": 4.364931583404541,
      "learning_rate": 4.052380451180577e-05,
      "loss": 0.7606,
      "step": 17138
    },
    {
      "epoch": 5.411746131986106,
      "grad_norm": 4.115658760070801,
      "learning_rate": 4.052292741114971e-05,
      "loss": 0.647,
      "step": 17139
    },
    {
      "epoch": 5.412061888222293,
      "grad_norm": 4.406895637512207,
      "learning_rate": 4.0522050310493635e-05,
      "loss": 0.6726,
      "step": 17140
    },
    {
      "epoch": 5.412377644458478,
      "grad_norm": 5.312436580657959,
      "learning_rate": 4.0521173209837566e-05,
      "loss": 0.5713,
      "step": 17141
    },
    {
      "epoch": 5.412693400694664,
      "grad_norm": 3.7653865814208984,
      "learning_rate": 4.052029610918149e-05,
      "loss": 0.5569,
      "step": 17142
    },
    {
      "epoch": 5.413009156930849,
      "grad_norm": 3.9725263118743896,
      "learning_rate": 4.051941900852542e-05,
      "loss": 0.6385,
      "step": 17143
    },
    {
      "epoch": 5.413324913167035,
      "grad_norm": 4.543642520904541,
      "learning_rate": 4.051854190786935e-05,
      "loss": 0.6079,
      "step": 17144
    },
    {
      "epoch": 5.413640669403221,
      "grad_norm": 4.3842878341674805,
      "learning_rate": 4.051766480721328e-05,
      "loss": 0.6725,
      "step": 17145
    },
    {
      "epoch": 5.413956425639406,
      "grad_norm": 3.687485456466675,
      "learning_rate": 4.0516787706557204e-05,
      "loss": 0.5465,
      "step": 17146
    },
    {
      "epoch": 5.414272181875592,
      "grad_norm": 3.7386794090270996,
      "learning_rate": 4.0515910605901136e-05,
      "loss": 0.5497,
      "step": 17147
    },
    {
      "epoch": 5.414587938111778,
      "grad_norm": 4.279341697692871,
      "learning_rate": 4.051503350524507e-05,
      "loss": 0.6196,
      "step": 17148
    },
    {
      "epoch": 5.414903694347963,
      "grad_norm": 5.365073204040527,
      "learning_rate": 4.051415640458899e-05,
      "loss": 0.6472,
      "step": 17149
    },
    {
      "epoch": 5.415219450584149,
      "grad_norm": 5.457742691040039,
      "learning_rate": 4.0513279303932924e-05,
      "loss": 0.6257,
      "step": 17150
    },
    {
      "epoch": 5.415535206820334,
      "grad_norm": 3.7083895206451416,
      "learning_rate": 4.051240220327685e-05,
      "loss": 0.5762,
      "step": 17151
    },
    {
      "epoch": 5.415850963056521,
      "grad_norm": 3.5214943885803223,
      "learning_rate": 4.051152510262078e-05,
      "loss": 0.5516,
      "step": 17152
    },
    {
      "epoch": 5.416166719292706,
      "grad_norm": 3.8886165618896484,
      "learning_rate": 4.0510648001964706e-05,
      "loss": 0.657,
      "step": 17153
    },
    {
      "epoch": 5.416482475528892,
      "grad_norm": 3.6558964252471924,
      "learning_rate": 4.050977090130864e-05,
      "loss": 0.6717,
      "step": 17154
    },
    {
      "epoch": 5.416798231765077,
      "grad_norm": 3.9665746688842773,
      "learning_rate": 4.050889380065256e-05,
      "loss": 0.6237,
      "step": 17155
    },
    {
      "epoch": 5.417113988001263,
      "grad_norm": 4.222663402557373,
      "learning_rate": 4.0508016699996494e-05,
      "loss": 0.5695,
      "step": 17156
    },
    {
      "epoch": 5.417429744237449,
      "grad_norm": 5.189784049987793,
      "learning_rate": 4.050713959934042e-05,
      "loss": 0.6855,
      "step": 17157
    },
    {
      "epoch": 5.417745500473634,
      "grad_norm": 5.496017932891846,
      "learning_rate": 4.050626249868435e-05,
      "loss": 0.602,
      "step": 17158
    },
    {
      "epoch": 5.41806125670982,
      "grad_norm": 3.964385747909546,
      "learning_rate": 4.0505385398028276e-05,
      "loss": 0.5894,
      "step": 17159
    },
    {
      "epoch": 5.418377012946006,
      "grad_norm": 4.798521518707275,
      "learning_rate": 4.050450829737221e-05,
      "loss": 0.6995,
      "step": 17160
    },
    {
      "epoch": 5.418692769182192,
      "grad_norm": 3.205268144607544,
      "learning_rate": 4.050363119671613e-05,
      "loss": 0.5823,
      "step": 17161
    },
    {
      "epoch": 5.419008525418377,
      "grad_norm": 3.797407627105713,
      "learning_rate": 4.0502754096060064e-05,
      "loss": 0.7228,
      "step": 17162
    },
    {
      "epoch": 5.419324281654562,
      "grad_norm": 4.935897350311279,
      "learning_rate": 4.0501876995403996e-05,
      "loss": 0.673,
      "step": 17163
    },
    {
      "epoch": 5.4196400378907486,
      "grad_norm": 4.015302658081055,
      "learning_rate": 4.050099989474792e-05,
      "loss": 0.6645,
      "step": 17164
    },
    {
      "epoch": 5.419955794126934,
      "grad_norm": 4.0971574783325195,
      "learning_rate": 4.050012279409185e-05,
      "loss": 0.5874,
      "step": 17165
    },
    {
      "epoch": 5.42027155036312,
      "grad_norm": 3.5566327571868896,
      "learning_rate": 4.0499245693435784e-05,
      "loss": 0.6334,
      "step": 17166
    },
    {
      "epoch": 5.420587306599305,
      "grad_norm": 3.489694118499756,
      "learning_rate": 4.049836859277971e-05,
      "loss": 0.5033,
      "step": 17167
    },
    {
      "epoch": 5.420903062835491,
      "grad_norm": 5.30693244934082,
      "learning_rate": 4.049749149212364e-05,
      "loss": 0.7214,
      "step": 17168
    },
    {
      "epoch": 5.421218819071677,
      "grad_norm": 4.705885887145996,
      "learning_rate": 4.049661439146757e-05,
      "loss": 0.6328,
      "step": 17169
    },
    {
      "epoch": 5.421534575307862,
      "grad_norm": 3.8209545612335205,
      "learning_rate": 4.04957372908115e-05,
      "loss": 0.5962,
      "step": 17170
    },
    {
      "epoch": 5.421850331544048,
      "grad_norm": 3.959399938583374,
      "learning_rate": 4.049486019015543e-05,
      "loss": 0.5786,
      "step": 17171
    },
    {
      "epoch": 5.422166087780234,
      "grad_norm": 4.694903373718262,
      "learning_rate": 4.0493983089499354e-05,
      "loss": 0.7565,
      "step": 17172
    },
    {
      "epoch": 5.422481844016419,
      "grad_norm": 4.119043827056885,
      "learning_rate": 4.0493105988843285e-05,
      "loss": 0.6445,
      "step": 17173
    },
    {
      "epoch": 5.422797600252605,
      "grad_norm": 3.5319416522979736,
      "learning_rate": 4.049222888818721e-05,
      "loss": 0.6374,
      "step": 17174
    },
    {
      "epoch": 5.42311335648879,
      "grad_norm": 4.575221538543701,
      "learning_rate": 4.049135178753114e-05,
      "loss": 0.5917,
      "step": 17175
    },
    {
      "epoch": 5.4234291127249765,
      "grad_norm": 4.449272632598877,
      "learning_rate": 4.049047468687507e-05,
      "loss": 0.6724,
      "step": 17176
    },
    {
      "epoch": 5.423744868961162,
      "grad_norm": 3.5491511821746826,
      "learning_rate": 4.0489597586219e-05,
      "loss": 0.5774,
      "step": 17177
    },
    {
      "epoch": 5.424060625197348,
      "grad_norm": 4.231131553649902,
      "learning_rate": 4.048872048556292e-05,
      "loss": 0.6098,
      "step": 17178
    },
    {
      "epoch": 5.424376381433533,
      "grad_norm": 3.533783435821533,
      "learning_rate": 4.0487843384906855e-05,
      "loss": 0.6336,
      "step": 17179
    },
    {
      "epoch": 5.424692137669719,
      "grad_norm": 4.319758892059326,
      "learning_rate": 4.048696628425078e-05,
      "loss": 0.5649,
      "step": 17180
    },
    {
      "epoch": 5.425007893905905,
      "grad_norm": 3.2686123847961426,
      "learning_rate": 4.048608918359471e-05,
      "loss": 0.5612,
      "step": 17181
    },
    {
      "epoch": 5.42532365014209,
      "grad_norm": 3.80070161819458,
      "learning_rate": 4.0485212082938637e-05,
      "loss": 0.5825,
      "step": 17182
    },
    {
      "epoch": 5.425639406378276,
      "grad_norm": 4.369784832000732,
      "learning_rate": 4.048433498228257e-05,
      "loss": 0.6561,
      "step": 17183
    },
    {
      "epoch": 5.4259551626144615,
      "grad_norm": 4.142406940460205,
      "learning_rate": 4.048345788162649e-05,
      "loss": 0.6194,
      "step": 17184
    },
    {
      "epoch": 5.426270918850648,
      "grad_norm": 3.679347038269043,
      "learning_rate": 4.0482580780970425e-05,
      "loss": 0.5374,
      "step": 17185
    },
    {
      "epoch": 5.426586675086833,
      "grad_norm": 4.171566963195801,
      "learning_rate": 4.0481703680314356e-05,
      "loss": 0.6545,
      "step": 17186
    },
    {
      "epoch": 5.426902431323018,
      "grad_norm": 5.4517059326171875,
      "learning_rate": 4.048082657965828e-05,
      "loss": 0.6891,
      "step": 17187
    },
    {
      "epoch": 5.4272181875592045,
      "grad_norm": 3.9279868602752686,
      "learning_rate": 4.047994947900221e-05,
      "loss": 0.583,
      "step": 17188
    },
    {
      "epoch": 5.42753394379539,
      "grad_norm": 4.100139141082764,
      "learning_rate": 4.047907237834614e-05,
      "loss": 0.6865,
      "step": 17189
    },
    {
      "epoch": 5.427849700031576,
      "grad_norm": 4.890686988830566,
      "learning_rate": 4.047819527769007e-05,
      "loss": 0.6611,
      "step": 17190
    },
    {
      "epoch": 5.428165456267761,
      "grad_norm": 3.992492437362671,
      "learning_rate": 4.0477318177033995e-05,
      "loss": 0.5878,
      "step": 17191
    },
    {
      "epoch": 5.4284812125039466,
      "grad_norm": 3.958901882171631,
      "learning_rate": 4.0476441076377926e-05,
      "loss": 0.6436,
      "step": 17192
    },
    {
      "epoch": 5.428796968740133,
      "grad_norm": 3.9651553630828857,
      "learning_rate": 4.047556397572186e-05,
      "loss": 0.628,
      "step": 17193
    },
    {
      "epoch": 5.429112724976318,
      "grad_norm": 4.020909786224365,
      "learning_rate": 4.047468687506579e-05,
      "loss": 0.5711,
      "step": 17194
    },
    {
      "epoch": 5.429428481212504,
      "grad_norm": 4.857207298278809,
      "learning_rate": 4.0473809774409714e-05,
      "loss": 0.6589,
      "step": 17195
    },
    {
      "epoch": 5.4297442374486895,
      "grad_norm": 3.884277820587158,
      "learning_rate": 4.0472932673753646e-05,
      "loss": 0.6258,
      "step": 17196
    },
    {
      "epoch": 5.430059993684876,
      "grad_norm": 4.666111469268799,
      "learning_rate": 4.047205557309757e-05,
      "loss": 0.7104,
      "step": 17197
    },
    {
      "epoch": 5.430375749921061,
      "grad_norm": 3.7648351192474365,
      "learning_rate": 4.04711784724415e-05,
      "loss": 0.5927,
      "step": 17198
    },
    {
      "epoch": 5.430691506157246,
      "grad_norm": 3.4346060752868652,
      "learning_rate": 4.047030137178543e-05,
      "loss": 0.6397,
      "step": 17199
    },
    {
      "epoch": 5.4310072623934325,
      "grad_norm": 3.721519947052002,
      "learning_rate": 4.046942427112936e-05,
      "loss": 0.6743,
      "step": 17200
    },
    {
      "epoch": 5.4310072623934325,
      "eval_loss": 0.8393192887306213,
      "eval_runtime": 4.5443,
      "eval_samples_per_second": 547.94,
      "eval_steps_per_second": 4.621,
      "step": 17200
    },
    {
      "epoch": 5.431323018629618,
      "grad_norm": 5.7713141441345215,
      "learning_rate": 4.0468547170473284e-05,
      "loss": 0.819,
      "step": 17201
    },
    {
      "epoch": 5.431638774865804,
      "grad_norm": 4.193127632141113,
      "learning_rate": 4.0467670069817216e-05,
      "loss": 0.6962,
      "step": 17202
    },
    {
      "epoch": 5.431954531101989,
      "grad_norm": 4.843364715576172,
      "learning_rate": 4.046679296916114e-05,
      "loss": 0.6947,
      "step": 17203
    },
    {
      "epoch": 5.4322702873381745,
      "grad_norm": 3.5312156677246094,
      "learning_rate": 4.046591586850507e-05,
      "loss": 0.6504,
      "step": 17204
    },
    {
      "epoch": 5.432586043574361,
      "grad_norm": 3.8390157222747803,
      "learning_rate": 4.0465038767849e-05,
      "loss": 0.6143,
      "step": 17205
    },
    {
      "epoch": 5.432901799810546,
      "grad_norm": 4.058398723602295,
      "learning_rate": 4.046416166719293e-05,
      "loss": 0.6077,
      "step": 17206
    },
    {
      "epoch": 5.433217556046732,
      "grad_norm": 3.5968785285949707,
      "learning_rate": 4.046328456653686e-05,
      "loss": 0.5409,
      "step": 17207
    },
    {
      "epoch": 5.4335333122829175,
      "grad_norm": 3.905216693878174,
      "learning_rate": 4.0462407465880786e-05,
      "loss": 0.6876,
      "step": 17208
    },
    {
      "epoch": 5.433849068519104,
      "grad_norm": 4.440425395965576,
      "learning_rate": 4.046153036522472e-05,
      "loss": 0.5407,
      "step": 17209
    },
    {
      "epoch": 5.434164824755289,
      "grad_norm": 4.867194175720215,
      "learning_rate": 4.046065326456864e-05,
      "loss": 0.6317,
      "step": 17210
    },
    {
      "epoch": 5.434480580991474,
      "grad_norm": 4.559325695037842,
      "learning_rate": 4.0459776163912574e-05,
      "loss": 0.5903,
      "step": 17211
    },
    {
      "epoch": 5.43479633722766,
      "grad_norm": 4.527021408081055,
      "learning_rate": 4.04588990632565e-05,
      "loss": 0.6929,
      "step": 17212
    },
    {
      "epoch": 5.435112093463846,
      "grad_norm": 4.194997310638428,
      "learning_rate": 4.045802196260043e-05,
      "loss": 0.6112,
      "step": 17213
    },
    {
      "epoch": 5.435427849700032,
      "grad_norm": 4.136119365692139,
      "learning_rate": 4.0457144861944355e-05,
      "loss": 0.5852,
      "step": 17214
    },
    {
      "epoch": 5.435743605936217,
      "grad_norm": 4.109495639801025,
      "learning_rate": 4.045626776128829e-05,
      "loss": 0.6576,
      "step": 17215
    },
    {
      "epoch": 5.4360593621724025,
      "grad_norm": 3.770564317703247,
      "learning_rate": 4.045539066063221e-05,
      "loss": 0.5327,
      "step": 17216
    },
    {
      "epoch": 5.436375118408589,
      "grad_norm": 4.224137306213379,
      "learning_rate": 4.0454513559976144e-05,
      "loss": 0.5912,
      "step": 17217
    },
    {
      "epoch": 5.436690874644774,
      "grad_norm": 3.6848721504211426,
      "learning_rate": 4.045363645932007e-05,
      "loss": 0.5965,
      "step": 17218
    },
    {
      "epoch": 5.43700663088096,
      "grad_norm": 3.5690534114837646,
      "learning_rate": 4.0452759358664e-05,
      "loss": 0.6555,
      "step": 17219
    },
    {
      "epoch": 5.437322387117145,
      "grad_norm": 4.3890886306762695,
      "learning_rate": 4.0451882258007925e-05,
      "loss": 0.6002,
      "step": 17220
    },
    {
      "epoch": 5.437638143353332,
      "grad_norm": 3.6785597801208496,
      "learning_rate": 4.0451005157351864e-05,
      "loss": 0.6688,
      "step": 17221
    },
    {
      "epoch": 5.437953899589517,
      "grad_norm": 4.30753755569458,
      "learning_rate": 4.045012805669579e-05,
      "loss": 0.6259,
      "step": 17222
    },
    {
      "epoch": 5.438269655825702,
      "grad_norm": 3.2860124111175537,
      "learning_rate": 4.044925095603972e-05,
      "loss": 0.6392,
      "step": 17223
    },
    {
      "epoch": 5.438585412061888,
      "grad_norm": 4.234743595123291,
      "learning_rate": 4.0448373855383645e-05,
      "loss": 0.6121,
      "step": 17224
    },
    {
      "epoch": 5.438901168298074,
      "grad_norm": 3.3613924980163574,
      "learning_rate": 4.044749675472758e-05,
      "loss": 0.5371,
      "step": 17225
    },
    {
      "epoch": 5.43921692453426,
      "grad_norm": 4.352450847625732,
      "learning_rate": 4.04466196540715e-05,
      "loss": 0.659,
      "step": 17226
    },
    {
      "epoch": 5.439532680770445,
      "grad_norm": 4.549707412719727,
      "learning_rate": 4.0445742553415433e-05,
      "loss": 0.6432,
      "step": 17227
    },
    {
      "epoch": 5.439848437006631,
      "grad_norm": 4.498783111572266,
      "learning_rate": 4.0444865452759365e-05,
      "loss": 0.7152,
      "step": 17228
    },
    {
      "epoch": 5.440164193242817,
      "grad_norm": 4.715365409851074,
      "learning_rate": 4.044398835210329e-05,
      "loss": 0.7267,
      "step": 17229
    },
    {
      "epoch": 5.440479949479002,
      "grad_norm": 4.882626533508301,
      "learning_rate": 4.044311125144722e-05,
      "loss": 0.6626,
      "step": 17230
    },
    {
      "epoch": 5.440795705715188,
      "grad_norm": 4.817677974700928,
      "learning_rate": 4.044223415079115e-05,
      "loss": 0.5877,
      "step": 17231
    },
    {
      "epoch": 5.441111461951373,
      "grad_norm": 4.359624862670898,
      "learning_rate": 4.044135705013508e-05,
      "loss": 0.5251,
      "step": 17232
    },
    {
      "epoch": 5.44142721818756,
      "grad_norm": 3.448610305786133,
      "learning_rate": 4.0440479949479e-05,
      "loss": 0.6242,
      "step": 17233
    },
    {
      "epoch": 5.441742974423745,
      "grad_norm": 3.791567087173462,
      "learning_rate": 4.0439602848822935e-05,
      "loss": 0.6198,
      "step": 17234
    },
    {
      "epoch": 5.44205873065993,
      "grad_norm": 4.656086444854736,
      "learning_rate": 4.043872574816686e-05,
      "loss": 0.6487,
      "step": 17235
    },
    {
      "epoch": 5.442374486896116,
      "grad_norm": 4.020218372344971,
      "learning_rate": 4.043784864751079e-05,
      "loss": 0.6859,
      "step": 17236
    },
    {
      "epoch": 5.442690243132302,
      "grad_norm": 5.206175327301025,
      "learning_rate": 4.0436971546854716e-05,
      "loss": 0.6634,
      "step": 17237
    },
    {
      "epoch": 5.443005999368488,
      "grad_norm": 4.250617980957031,
      "learning_rate": 4.043609444619865e-05,
      "loss": 0.6605,
      "step": 17238
    },
    {
      "epoch": 5.443321755604673,
      "grad_norm": 3.552872657775879,
      "learning_rate": 4.043521734554257e-05,
      "loss": 0.594,
      "step": 17239
    },
    {
      "epoch": 5.443637511840858,
      "grad_norm": 4.054955005645752,
      "learning_rate": 4.0434340244886505e-05,
      "loss": 0.6271,
      "step": 17240
    },
    {
      "epoch": 5.443953268077045,
      "grad_norm": 3.3637070655822754,
      "learning_rate": 4.043346314423043e-05,
      "loss": 0.6078,
      "step": 17241
    },
    {
      "epoch": 5.44426902431323,
      "grad_norm": 5.294919013977051,
      "learning_rate": 4.043258604357436e-05,
      "loss": 0.629,
      "step": 17242
    },
    {
      "epoch": 5.444584780549416,
      "grad_norm": 4.07852029800415,
      "learning_rate": 4.0431708942918286e-05,
      "loss": 0.6403,
      "step": 17243
    },
    {
      "epoch": 5.444900536785601,
      "grad_norm": 3.841465950012207,
      "learning_rate": 4.043083184226222e-05,
      "loss": 0.6652,
      "step": 17244
    },
    {
      "epoch": 5.445216293021788,
      "grad_norm": 3.6138434410095215,
      "learning_rate": 4.042995474160615e-05,
      "loss": 0.6216,
      "step": 17245
    },
    {
      "epoch": 5.445532049257973,
      "grad_norm": 3.851633310317993,
      "learning_rate": 4.0429077640950074e-05,
      "loss": 0.5715,
      "step": 17246
    },
    {
      "epoch": 5.445847805494158,
      "grad_norm": 4.455930233001709,
      "learning_rate": 4.0428200540294006e-05,
      "loss": 0.5954,
      "step": 17247
    },
    {
      "epoch": 5.446163561730344,
      "grad_norm": 3.5395681858062744,
      "learning_rate": 4.042732343963794e-05,
      "loss": 0.6305,
      "step": 17248
    },
    {
      "epoch": 5.44647931796653,
      "grad_norm": 3.963043451309204,
      "learning_rate": 4.042644633898186e-05,
      "loss": 0.5938,
      "step": 17249
    },
    {
      "epoch": 5.446795074202716,
      "grad_norm": 3.419343948364258,
      "learning_rate": 4.0425569238325794e-05,
      "loss": 0.5821,
      "step": 17250
    },
    {
      "epoch": 5.447110830438901,
      "grad_norm": 4.932806015014648,
      "learning_rate": 4.0424692137669726e-05,
      "loss": 0.6585,
      "step": 17251
    },
    {
      "epoch": 5.447426586675087,
      "grad_norm": 5.35197114944458,
      "learning_rate": 4.042381503701365e-05,
      "loss": 0.6774,
      "step": 17252
    },
    {
      "epoch": 5.447742342911273,
      "grad_norm": 5.204975605010986,
      "learning_rate": 4.042293793635758e-05,
      "loss": 0.7208,
      "step": 17253
    },
    {
      "epoch": 5.448058099147458,
      "grad_norm": 5.153153896331787,
      "learning_rate": 4.042206083570151e-05,
      "loss": 0.7483,
      "step": 17254
    },
    {
      "epoch": 5.448373855383644,
      "grad_norm": 5.002039432525635,
      "learning_rate": 4.042118373504544e-05,
      "loss": 0.6642,
      "step": 17255
    },
    {
      "epoch": 5.448689611619829,
      "grad_norm": 4.919173240661621,
      "learning_rate": 4.0420306634389364e-05,
      "loss": 0.7057,
      "step": 17256
    },
    {
      "epoch": 5.4490053678560155,
      "grad_norm": 4.347561836242676,
      "learning_rate": 4.0419429533733296e-05,
      "loss": 0.6825,
      "step": 17257
    },
    {
      "epoch": 5.449321124092201,
      "grad_norm": 4.383120536804199,
      "learning_rate": 4.041855243307722e-05,
      "loss": 0.6916,
      "step": 17258
    },
    {
      "epoch": 5.449636880328386,
      "grad_norm": 3.9063031673431396,
      "learning_rate": 4.041767533242115e-05,
      "loss": 0.4854,
      "step": 17259
    },
    {
      "epoch": 5.449952636564572,
      "grad_norm": 4.837193489074707,
      "learning_rate": 4.041679823176508e-05,
      "loss": 0.6552,
      "step": 17260
    },
    {
      "epoch": 5.450268392800758,
      "grad_norm": 4.59175443649292,
      "learning_rate": 4.041592113110901e-05,
      "loss": 0.7398,
      "step": 17261
    },
    {
      "epoch": 5.450584149036944,
      "grad_norm": 3.211017608642578,
      "learning_rate": 4.0415044030452934e-05,
      "loss": 0.6799,
      "step": 17262
    },
    {
      "epoch": 5.450899905273129,
      "grad_norm": 4.445514678955078,
      "learning_rate": 4.0414166929796866e-05,
      "loss": 0.7459,
      "step": 17263
    },
    {
      "epoch": 5.451215661509314,
      "grad_norm": 4.8205180168151855,
      "learning_rate": 4.041328982914079e-05,
      "loss": 0.6265,
      "step": 17264
    },
    {
      "epoch": 5.4515314177455005,
      "grad_norm": 4.651464462280273,
      "learning_rate": 4.041241272848472e-05,
      "loss": 0.5528,
      "step": 17265
    },
    {
      "epoch": 5.451847173981686,
      "grad_norm": 4.068694114685059,
      "learning_rate": 4.0411535627828654e-05,
      "loss": 0.6255,
      "step": 17266
    },
    {
      "epoch": 5.452162930217872,
      "grad_norm": 4.312596797943115,
      "learning_rate": 4.041065852717258e-05,
      "loss": 0.6994,
      "step": 17267
    },
    {
      "epoch": 5.452478686454057,
      "grad_norm": 4.150485992431641,
      "learning_rate": 4.040978142651651e-05,
      "loss": 0.6345,
      "step": 17268
    },
    {
      "epoch": 5.4527944426902435,
      "grad_norm": 3.6361241340637207,
      "learning_rate": 4.0408904325860435e-05,
      "loss": 0.6455,
      "step": 17269
    },
    {
      "epoch": 5.453110198926429,
      "grad_norm": 5.093312740325928,
      "learning_rate": 4.040802722520437e-05,
      "loss": 0.7055,
      "step": 17270
    },
    {
      "epoch": 5.453425955162614,
      "grad_norm": 4.225003719329834,
      "learning_rate": 4.040715012454829e-05,
      "loss": 0.6351,
      "step": 17271
    },
    {
      "epoch": 5.4537417113988,
      "grad_norm": 3.291313409805298,
      "learning_rate": 4.0406273023892224e-05,
      "loss": 0.5954,
      "step": 17272
    },
    {
      "epoch": 5.454057467634986,
      "grad_norm": 4.199373722076416,
      "learning_rate": 4.040539592323615e-05,
      "loss": 0.6006,
      "step": 17273
    },
    {
      "epoch": 5.454373223871172,
      "grad_norm": 3.938286304473877,
      "learning_rate": 4.040451882258008e-05,
      "loss": 0.7074,
      "step": 17274
    },
    {
      "epoch": 5.454688980107357,
      "grad_norm": 5.053944110870361,
      "learning_rate": 4.040364172192401e-05,
      "loss": 0.7477,
      "step": 17275
    },
    {
      "epoch": 5.455004736343543,
      "grad_norm": 4.154024124145508,
      "learning_rate": 4.0402764621267944e-05,
      "loss": 0.5574,
      "step": 17276
    },
    {
      "epoch": 5.4553204925797285,
      "grad_norm": 4.198600769042969,
      "learning_rate": 4.040188752061187e-05,
      "loss": 0.7437,
      "step": 17277
    },
    {
      "epoch": 5.455636248815914,
      "grad_norm": 3.3088066577911377,
      "learning_rate": 4.04010104199558e-05,
      "loss": 0.6381,
      "step": 17278
    },
    {
      "epoch": 5.4559520050521,
      "grad_norm": 4.493430137634277,
      "learning_rate": 4.0400133319299725e-05,
      "loss": 0.6752,
      "step": 17279
    },
    {
      "epoch": 5.456267761288285,
      "grad_norm": 5.016817092895508,
      "learning_rate": 4.039925621864366e-05,
      "loss": 0.6772,
      "step": 17280
    },
    {
      "epoch": 5.4565835175244715,
      "grad_norm": 4.642518520355225,
      "learning_rate": 4.039837911798758e-05,
      "loss": 0.7012,
      "step": 17281
    },
    {
      "epoch": 5.456899273760657,
      "grad_norm": 4.2992424964904785,
      "learning_rate": 4.039750201733151e-05,
      "loss": 0.7711,
      "step": 17282
    },
    {
      "epoch": 5.457215029996842,
      "grad_norm": 5.030874252319336,
      "learning_rate": 4.039662491667544e-05,
      "loss": 0.7433,
      "step": 17283
    },
    {
      "epoch": 5.457530786233028,
      "grad_norm": 3.8211588859558105,
      "learning_rate": 4.039574781601937e-05,
      "loss": 0.6493,
      "step": 17284
    },
    {
      "epoch": 5.4578465424692135,
      "grad_norm": 3.7522478103637695,
      "learning_rate": 4.0394870715363295e-05,
      "loss": 0.6603,
      "step": 17285
    },
    {
      "epoch": 5.4581622987054,
      "grad_norm": 4.120460033416748,
      "learning_rate": 4.0393993614707227e-05,
      "loss": 0.6332,
      "step": 17286
    },
    {
      "epoch": 5.458478054941585,
      "grad_norm": 4.302518367767334,
      "learning_rate": 4.039311651405115e-05,
      "loss": 0.6445,
      "step": 17287
    },
    {
      "epoch": 5.45879381117777,
      "grad_norm": 4.608317852020264,
      "learning_rate": 4.039223941339508e-05,
      "loss": 0.6838,
      "step": 17288
    },
    {
      "epoch": 5.4591095674139565,
      "grad_norm": 3.855051279067993,
      "learning_rate": 4.0391362312739015e-05,
      "loss": 0.7663,
      "step": 17289
    },
    {
      "epoch": 5.459425323650142,
      "grad_norm": 4.408958911895752,
      "learning_rate": 4.039048521208294e-05,
      "loss": 0.6611,
      "step": 17290
    },
    {
      "epoch": 5.459741079886328,
      "grad_norm": 3.4743824005126953,
      "learning_rate": 4.038960811142687e-05,
      "loss": 0.6431,
      "step": 17291
    },
    {
      "epoch": 5.460056836122513,
      "grad_norm": 3.40854811668396,
      "learning_rate": 4.0388731010770796e-05,
      "loss": 0.6045,
      "step": 17292
    },
    {
      "epoch": 5.460372592358699,
      "grad_norm": 3.6060726642608643,
      "learning_rate": 4.038785391011473e-05,
      "loss": 0.5821,
      "step": 17293
    },
    {
      "epoch": 5.460688348594885,
      "grad_norm": 3.7664668560028076,
      "learning_rate": 4.038697680945865e-05,
      "loss": 0.7188,
      "step": 17294
    },
    {
      "epoch": 5.46100410483107,
      "grad_norm": 3.813697576522827,
      "learning_rate": 4.0386099708802585e-05,
      "loss": 0.5792,
      "step": 17295
    },
    {
      "epoch": 5.461319861067256,
      "grad_norm": 3.6714670658111572,
      "learning_rate": 4.038522260814651e-05,
      "loss": 0.5242,
      "step": 17296
    },
    {
      "epoch": 5.4616356173034415,
      "grad_norm": 4.776618003845215,
      "learning_rate": 4.038434550749044e-05,
      "loss": 0.6348,
      "step": 17297
    },
    {
      "epoch": 5.461951373539628,
      "grad_norm": 4.132903575897217,
      "learning_rate": 4.0383468406834366e-05,
      "loss": 0.6528,
      "step": 17298
    },
    {
      "epoch": 5.462267129775813,
      "grad_norm": 3.7833828926086426,
      "learning_rate": 4.03825913061783e-05,
      "loss": 0.6146,
      "step": 17299
    },
    {
      "epoch": 5.462582886011999,
      "grad_norm": 3.506141424179077,
      "learning_rate": 4.038171420552222e-05,
      "loss": 0.5947,
      "step": 17300
    },
    {
      "epoch": 5.4628986422481844,
      "grad_norm": 3.887453079223633,
      "learning_rate": 4.0380837104866154e-05,
      "loss": 0.6073,
      "step": 17301
    },
    {
      "epoch": 5.46321439848437,
      "grad_norm": 4.024008750915527,
      "learning_rate": 4.0379960004210086e-05,
      "loss": 0.6539,
      "step": 17302
    },
    {
      "epoch": 5.463530154720556,
      "grad_norm": 4.199897766113281,
      "learning_rate": 4.037908290355402e-05,
      "loss": 0.738,
      "step": 17303
    },
    {
      "epoch": 5.463845910956741,
      "grad_norm": 3.8655052185058594,
      "learning_rate": 4.037820580289794e-05,
      "loss": 0.6102,
      "step": 17304
    },
    {
      "epoch": 5.464161667192927,
      "grad_norm": 3.737480878829956,
      "learning_rate": 4.0377328702241874e-05,
      "loss": 0.5828,
      "step": 17305
    },
    {
      "epoch": 5.464477423429113,
      "grad_norm": 4.708340167999268,
      "learning_rate": 4.03764516015858e-05,
      "loss": 0.6336,
      "step": 17306
    },
    {
      "epoch": 5.464793179665298,
      "grad_norm": 5.215488433837891,
      "learning_rate": 4.037557450092973e-05,
      "loss": 0.6344,
      "step": 17307
    },
    {
      "epoch": 5.465108935901484,
      "grad_norm": 3.689668655395508,
      "learning_rate": 4.0374697400273656e-05,
      "loss": 0.5454,
      "step": 17308
    },
    {
      "epoch": 5.4654246921376695,
      "grad_norm": 4.118185997009277,
      "learning_rate": 4.037382029961759e-05,
      "loss": 0.6274,
      "step": 17309
    },
    {
      "epoch": 5.465740448373856,
      "grad_norm": 4.9068684577941895,
      "learning_rate": 4.037294319896152e-05,
      "loss": 0.6794,
      "step": 17310
    },
    {
      "epoch": 5.466056204610041,
      "grad_norm": 4.102644443511963,
      "learning_rate": 4.0372066098305444e-05,
      "loss": 0.6092,
      "step": 17311
    },
    {
      "epoch": 5.466371960846227,
      "grad_norm": 4.3936614990234375,
      "learning_rate": 4.0371188997649376e-05,
      "loss": 0.6755,
      "step": 17312
    },
    {
      "epoch": 5.466687717082412,
      "grad_norm": 4.00476598739624,
      "learning_rate": 4.03703118969933e-05,
      "loss": 0.667,
      "step": 17313
    },
    {
      "epoch": 5.467003473318598,
      "grad_norm": 3.5530784130096436,
      "learning_rate": 4.036943479633723e-05,
      "loss": 0.5675,
      "step": 17314
    },
    {
      "epoch": 5.467319229554784,
      "grad_norm": 5.090338706970215,
      "learning_rate": 4.036855769568116e-05,
      "loss": 0.787,
      "step": 17315
    },
    {
      "epoch": 5.467634985790969,
      "grad_norm": 5.02232027053833,
      "learning_rate": 4.036768059502509e-05,
      "loss": 0.5989,
      "step": 17316
    },
    {
      "epoch": 5.467950742027155,
      "grad_norm": 3.8508081436157227,
      "learning_rate": 4.0366803494369014e-05,
      "loss": 0.6709,
      "step": 17317
    },
    {
      "epoch": 5.468266498263341,
      "grad_norm": 4.046529769897461,
      "learning_rate": 4.0365926393712945e-05,
      "loss": 0.5612,
      "step": 17318
    },
    {
      "epoch": 5.468582254499526,
      "grad_norm": 3.6461105346679688,
      "learning_rate": 4.036504929305687e-05,
      "loss": 0.585,
      "step": 17319
    },
    {
      "epoch": 5.468898010735712,
      "grad_norm": 3.868236780166626,
      "learning_rate": 4.03641721924008e-05,
      "loss": 0.613,
      "step": 17320
    },
    {
      "epoch": 5.469213766971897,
      "grad_norm": 4.298393249511719,
      "learning_rate": 4.036329509174473e-05,
      "loss": 0.5842,
      "step": 17321
    },
    {
      "epoch": 5.469529523208084,
      "grad_norm": 4.96052360534668,
      "learning_rate": 4.036241799108866e-05,
      "loss": 0.6073,
      "step": 17322
    },
    {
      "epoch": 5.469845279444269,
      "grad_norm": 4.296792507171631,
      "learning_rate": 4.0361540890432584e-05,
      "loss": 0.7536,
      "step": 17323
    },
    {
      "epoch": 5.470161035680455,
      "grad_norm": 3.991396427154541,
      "learning_rate": 4.0360663789776515e-05,
      "loss": 0.6431,
      "step": 17324
    },
    {
      "epoch": 5.47047679191664,
      "grad_norm": 3.7971041202545166,
      "learning_rate": 4.035978668912044e-05,
      "loss": 0.7028,
      "step": 17325
    },
    {
      "epoch": 5.470792548152826,
      "grad_norm": 4.013018608093262,
      "learning_rate": 4.035890958846437e-05,
      "loss": 0.6451,
      "step": 17326
    },
    {
      "epoch": 5.471108304389012,
      "grad_norm": 4.15641450881958,
      "learning_rate": 4.0358032487808304e-05,
      "loss": 0.5217,
      "step": 17327
    },
    {
      "epoch": 5.471424060625197,
      "grad_norm": 3.8089821338653564,
      "learning_rate": 4.035715538715223e-05,
      "loss": 0.5932,
      "step": 17328
    },
    {
      "epoch": 5.471739816861383,
      "grad_norm": 4.4153828620910645,
      "learning_rate": 4.035627828649616e-05,
      "loss": 0.6084,
      "step": 17329
    },
    {
      "epoch": 5.472055573097569,
      "grad_norm": 4.303684234619141,
      "learning_rate": 4.035540118584009e-05,
      "loss": 0.5324,
      "step": 17330
    },
    {
      "epoch": 5.472371329333754,
      "grad_norm": 3.6606414318084717,
      "learning_rate": 4.035452408518402e-05,
      "loss": 0.6395,
      "step": 17331
    },
    {
      "epoch": 5.47268708556994,
      "grad_norm": 4.601681709289551,
      "learning_rate": 4.035364698452795e-05,
      "loss": 0.6622,
      "step": 17332
    },
    {
      "epoch": 5.473002841806125,
      "grad_norm": 3.1316511631011963,
      "learning_rate": 4.035276988387188e-05,
      "loss": 0.5046,
      "step": 17333
    },
    {
      "epoch": 5.473318598042312,
      "grad_norm": 3.937403917312622,
      "learning_rate": 4.0351892783215805e-05,
      "loss": 0.6903,
      "step": 17334
    },
    {
      "epoch": 5.473634354278497,
      "grad_norm": 4.181463718414307,
      "learning_rate": 4.035101568255974e-05,
      "loss": 0.7265,
      "step": 17335
    },
    {
      "epoch": 5.473950110514683,
      "grad_norm": 4.207149028778076,
      "learning_rate": 4.035013858190366e-05,
      "loss": 0.6347,
      "step": 17336
    },
    {
      "epoch": 5.474265866750868,
      "grad_norm": 4.8834452629089355,
      "learning_rate": 4.034926148124759e-05,
      "loss": 0.6291,
      "step": 17337
    },
    {
      "epoch": 5.474581622987054,
      "grad_norm": 4.199122905731201,
      "learning_rate": 4.034838438059152e-05,
      "loss": 0.684,
      "step": 17338
    },
    {
      "epoch": 5.47489737922324,
      "grad_norm": 4.571414470672607,
      "learning_rate": 4.034750727993545e-05,
      "loss": 0.6839,
      "step": 17339
    },
    {
      "epoch": 5.475213135459425,
      "grad_norm": 5.001892566680908,
      "learning_rate": 4.0346630179279375e-05,
      "loss": 0.6966,
      "step": 17340
    },
    {
      "epoch": 5.475528891695611,
      "grad_norm": 4.068416118621826,
      "learning_rate": 4.0345753078623306e-05,
      "loss": 0.6701,
      "step": 17341
    },
    {
      "epoch": 5.475844647931797,
      "grad_norm": 5.745284557342529,
      "learning_rate": 4.034487597796723e-05,
      "loss": 0.692,
      "step": 17342
    },
    {
      "epoch": 5.476160404167983,
      "grad_norm": 4.3817572593688965,
      "learning_rate": 4.034399887731116e-05,
      "loss": 0.6091,
      "step": 17343
    },
    {
      "epoch": 5.476476160404168,
      "grad_norm": 3.4309804439544678,
      "learning_rate": 4.034312177665509e-05,
      "loss": 0.588,
      "step": 17344
    },
    {
      "epoch": 5.476791916640353,
      "grad_norm": 3.8643436431884766,
      "learning_rate": 4.034224467599902e-05,
      "loss": 0.613,
      "step": 17345
    },
    {
      "epoch": 5.4771076728765395,
      "grad_norm": 3.9981460571289062,
      "learning_rate": 4.0341367575342945e-05,
      "loss": 0.7124,
      "step": 17346
    },
    {
      "epoch": 5.477423429112725,
      "grad_norm": 4.131848335266113,
      "learning_rate": 4.0340490474686876e-05,
      "loss": 0.6149,
      "step": 17347
    },
    {
      "epoch": 5.477739185348911,
      "grad_norm": 4.7070207595825195,
      "learning_rate": 4.033961337403081e-05,
      "loss": 0.6733,
      "step": 17348
    },
    {
      "epoch": 5.478054941585096,
      "grad_norm": 4.347573757171631,
      "learning_rate": 4.033873627337473e-05,
      "loss": 0.7433,
      "step": 17349
    },
    {
      "epoch": 5.478370697821282,
      "grad_norm": 4.4236650466918945,
      "learning_rate": 4.0337859172718664e-05,
      "loss": 0.703,
      "step": 17350
    },
    {
      "epoch": 5.478686454057468,
      "grad_norm": 3.3273258209228516,
      "learning_rate": 4.033698207206259e-05,
      "loss": 0.63,
      "step": 17351
    },
    {
      "epoch": 5.479002210293653,
      "grad_norm": 4.0945329666137695,
      "learning_rate": 4.033610497140652e-05,
      "loss": 0.5967,
      "step": 17352
    },
    {
      "epoch": 5.479317966529839,
      "grad_norm": 4.0021281242370605,
      "learning_rate": 4.0335227870750446e-05,
      "loss": 0.7759,
      "step": 17353
    },
    {
      "epoch": 5.479633722766025,
      "grad_norm": 4.619547367095947,
      "learning_rate": 4.033435077009438e-05,
      "loss": 0.7042,
      "step": 17354
    },
    {
      "epoch": 5.47994947900221,
      "grad_norm": 4.205690383911133,
      "learning_rate": 4.03334736694383e-05,
      "loss": 0.6052,
      "step": 17355
    },
    {
      "epoch": 5.480265235238396,
      "grad_norm": 3.501312732696533,
      "learning_rate": 4.0332596568782234e-05,
      "loss": 0.5314,
      "step": 17356
    },
    {
      "epoch": 5.480580991474581,
      "grad_norm": 4.517881870269775,
      "learning_rate": 4.0331719468126166e-05,
      "loss": 0.5951,
      "step": 17357
    },
    {
      "epoch": 5.4808967477107675,
      "grad_norm": 4.061824321746826,
      "learning_rate": 4.03308423674701e-05,
      "loss": 0.6249,
      "step": 17358
    },
    {
      "epoch": 5.481212503946953,
      "grad_norm": 4.026778697967529,
      "learning_rate": 4.032996526681402e-05,
      "loss": 0.6628,
      "step": 17359
    },
    {
      "epoch": 5.481528260183139,
      "grad_norm": 4.322449207305908,
      "learning_rate": 4.0329088166157954e-05,
      "loss": 0.6737,
      "step": 17360
    },
    {
      "epoch": 5.481844016419324,
      "grad_norm": 6.444361686706543,
      "learning_rate": 4.032821106550188e-05,
      "loss": 0.6653,
      "step": 17361
    },
    {
      "epoch": 5.48215977265551,
      "grad_norm": 4.168694972991943,
      "learning_rate": 4.032733396484581e-05,
      "loss": 0.6432,
      "step": 17362
    },
    {
      "epoch": 5.482475528891696,
      "grad_norm": 3.6820645332336426,
      "learning_rate": 4.0326456864189736e-05,
      "loss": 0.6081,
      "step": 17363
    },
    {
      "epoch": 5.482791285127881,
      "grad_norm": 4.799495220184326,
      "learning_rate": 4.032557976353367e-05,
      "loss": 0.6846,
      "step": 17364
    },
    {
      "epoch": 5.483107041364067,
      "grad_norm": 4.763539791107178,
      "learning_rate": 4.032470266287759e-05,
      "loss": 0.5859,
      "step": 17365
    },
    {
      "epoch": 5.4834227976002525,
      "grad_norm": 3.8208670616149902,
      "learning_rate": 4.0323825562221524e-05,
      "loss": 0.6463,
      "step": 17366
    },
    {
      "epoch": 5.483738553836439,
      "grad_norm": 4.174315929412842,
      "learning_rate": 4.032294846156545e-05,
      "loss": 0.6884,
      "step": 17367
    },
    {
      "epoch": 5.484054310072624,
      "grad_norm": 4.47330904006958,
      "learning_rate": 4.032207136090938e-05,
      "loss": 0.63,
      "step": 17368
    },
    {
      "epoch": 5.484370066308809,
      "grad_norm": 4.5762858390808105,
      "learning_rate": 4.0321194260253305e-05,
      "loss": 0.6127,
      "step": 17369
    },
    {
      "epoch": 5.4846858225449955,
      "grad_norm": 3.9013192653656006,
      "learning_rate": 4.032031715959724e-05,
      "loss": 0.5643,
      "step": 17370
    },
    {
      "epoch": 5.485001578781181,
      "grad_norm": 4.0103759765625,
      "learning_rate": 4.031944005894117e-05,
      "loss": 0.6436,
      "step": 17371
    },
    {
      "epoch": 5.485317335017367,
      "grad_norm": 5.444520473480225,
      "learning_rate": 4.0318562958285094e-05,
      "loss": 0.6892,
      "step": 17372
    },
    {
      "epoch": 5.485633091253552,
      "grad_norm": 4.232804775238037,
      "learning_rate": 4.0317685857629025e-05,
      "loss": 0.5535,
      "step": 17373
    },
    {
      "epoch": 5.4859488474897375,
      "grad_norm": 4.055545806884766,
      "learning_rate": 4.031680875697295e-05,
      "loss": 0.824,
      "step": 17374
    },
    {
      "epoch": 5.486264603725924,
      "grad_norm": 4.951457500457764,
      "learning_rate": 4.031593165631688e-05,
      "loss": 0.753,
      "step": 17375
    },
    {
      "epoch": 5.486580359962109,
      "grad_norm": 3.892266273498535,
      "learning_rate": 4.031505455566081e-05,
      "loss": 0.6124,
      "step": 17376
    },
    {
      "epoch": 5.486896116198295,
      "grad_norm": 4.489002227783203,
      "learning_rate": 4.031417745500474e-05,
      "loss": 0.7293,
      "step": 17377
    },
    {
      "epoch": 5.4872118724344805,
      "grad_norm": 3.938258647918701,
      "learning_rate": 4.0313300354348663e-05,
      "loss": 0.6613,
      "step": 17378
    },
    {
      "epoch": 5.487527628670666,
      "grad_norm": 3.8383474349975586,
      "learning_rate": 4.0312423253692595e-05,
      "loss": 0.5957,
      "step": 17379
    },
    {
      "epoch": 5.487843384906852,
      "grad_norm": 3.6835622787475586,
      "learning_rate": 4.031154615303652e-05,
      "loss": 0.6119,
      "step": 17380
    },
    {
      "epoch": 5.488159141143037,
      "grad_norm": 5.22446346282959,
      "learning_rate": 4.031066905238045e-05,
      "loss": 0.6874,
      "step": 17381
    },
    {
      "epoch": 5.4884748973792234,
      "grad_norm": 3.9996883869171143,
      "learning_rate": 4.030979195172438e-05,
      "loss": 0.5983,
      "step": 17382
    },
    {
      "epoch": 5.488790653615409,
      "grad_norm": 4.453768730163574,
      "learning_rate": 4.030891485106831e-05,
      "loss": 0.7262,
      "step": 17383
    },
    {
      "epoch": 5.489106409851595,
      "grad_norm": 3.346750020980835,
      "learning_rate": 4.030803775041224e-05,
      "loss": 0.6447,
      "step": 17384
    },
    {
      "epoch": 5.48942216608778,
      "grad_norm": 3.6913275718688965,
      "learning_rate": 4.030716064975617e-05,
      "loss": 0.5795,
      "step": 17385
    },
    {
      "epoch": 5.4897379223239655,
      "grad_norm": 3.5014824867248535,
      "learning_rate": 4.0306283549100097e-05,
      "loss": 0.5934,
      "step": 17386
    },
    {
      "epoch": 5.490053678560152,
      "grad_norm": 3.4803855419158936,
      "learning_rate": 4.030540644844403e-05,
      "loss": 0.6734,
      "step": 17387
    },
    {
      "epoch": 5.490369434796337,
      "grad_norm": 4.750236511230469,
      "learning_rate": 4.030452934778795e-05,
      "loss": 0.6975,
      "step": 17388
    },
    {
      "epoch": 5.490685191032523,
      "grad_norm": 4.239320755004883,
      "learning_rate": 4.0303652247131885e-05,
      "loss": 0.671,
      "step": 17389
    },
    {
      "epoch": 5.4910009472687085,
      "grad_norm": 4.0130133628845215,
      "learning_rate": 4.030277514647581e-05,
      "loss": 0.6946,
      "step": 17390
    },
    {
      "epoch": 5.491316703504895,
      "grad_norm": 3.711470365524292,
      "learning_rate": 4.030189804581974e-05,
      "loss": 0.6552,
      "step": 17391
    },
    {
      "epoch": 5.49163245974108,
      "grad_norm": 3.2251908779144287,
      "learning_rate": 4.030102094516367e-05,
      "loss": 0.5755,
      "step": 17392
    },
    {
      "epoch": 5.491948215977265,
      "grad_norm": 4.36562967300415,
      "learning_rate": 4.03001438445076e-05,
      "loss": 0.6209,
      "step": 17393
    },
    {
      "epoch": 5.492263972213451,
      "grad_norm": 3.841106653213501,
      "learning_rate": 4.029926674385153e-05,
      "loss": 0.5979,
      "step": 17394
    },
    {
      "epoch": 5.492579728449637,
      "grad_norm": 3.9540762901306152,
      "learning_rate": 4.0298389643195455e-05,
      "loss": 0.6064,
      "step": 17395
    },
    {
      "epoch": 5.492895484685823,
      "grad_norm": 4.001734256744385,
      "learning_rate": 4.0297512542539386e-05,
      "loss": 0.6985,
      "step": 17396
    },
    {
      "epoch": 5.493211240922008,
      "grad_norm": 3.9307243824005127,
      "learning_rate": 4.029663544188331e-05,
      "loss": 0.7048,
      "step": 17397
    },
    {
      "epoch": 5.4935269971581935,
      "grad_norm": 3.4615142345428467,
      "learning_rate": 4.029575834122724e-05,
      "loss": 0.6033,
      "step": 17398
    },
    {
      "epoch": 5.49384275339438,
      "grad_norm": 3.347269296646118,
      "learning_rate": 4.029488124057117e-05,
      "loss": 0.609,
      "step": 17399
    },
    {
      "epoch": 5.494158509630565,
      "grad_norm": 3.4419641494750977,
      "learning_rate": 4.02940041399151e-05,
      "loss": 0.5609,
      "step": 17400
    },
    {
      "epoch": 5.494158509630565,
      "eval_loss": 0.8540180325508118,
      "eval_runtime": 4.5483,
      "eval_samples_per_second": 547.455,
      "eval_steps_per_second": 4.617,
      "step": 17400
    },
    {
      "epoch": 5.494474265866751,
      "grad_norm": 5.590163230895996,
      "learning_rate": 4.0293127039259024e-05,
      "loss": 0.8161,
      "step": 17401
    },
    {
      "epoch": 5.494790022102936,
      "grad_norm": 3.9665353298187256,
      "learning_rate": 4.0292249938602956e-05,
      "loss": 0.6395,
      "step": 17402
    },
    {
      "epoch": 5.495105778339122,
      "grad_norm": 4.210596561431885,
      "learning_rate": 4.029137283794688e-05,
      "loss": 0.5899,
      "step": 17403
    },
    {
      "epoch": 5.495421534575308,
      "grad_norm": 3.9540038108825684,
      "learning_rate": 4.029049573729081e-05,
      "loss": 0.6583,
      "step": 17404
    },
    {
      "epoch": 5.495737290811493,
      "grad_norm": 3.8597774505615234,
      "learning_rate": 4.028961863663474e-05,
      "loss": 0.5626,
      "step": 17405
    },
    {
      "epoch": 5.496053047047679,
      "grad_norm": 4.00494909286499,
      "learning_rate": 4.028874153597867e-05,
      "loss": 0.6413,
      "step": 17406
    },
    {
      "epoch": 5.496368803283865,
      "grad_norm": 5.375401020050049,
      "learning_rate": 4.02878644353226e-05,
      "loss": 0.6892,
      "step": 17407
    },
    {
      "epoch": 5.496684559520051,
      "grad_norm": 5.202517032623291,
      "learning_rate": 4.0286987334666526e-05,
      "loss": 0.7259,
      "step": 17408
    },
    {
      "epoch": 5.497000315756236,
      "grad_norm": 5.01259183883667,
      "learning_rate": 4.028611023401046e-05,
      "loss": 0.6701,
      "step": 17409
    },
    {
      "epoch": 5.4973160719924214,
      "grad_norm": 3.9362971782684326,
      "learning_rate": 4.028523313335438e-05,
      "loss": 0.638,
      "step": 17410
    },
    {
      "epoch": 5.497631828228608,
      "grad_norm": 3.9946885108947754,
      "learning_rate": 4.0284356032698314e-05,
      "loss": 0.6891,
      "step": 17411
    },
    {
      "epoch": 5.497947584464793,
      "grad_norm": 5.107763767242432,
      "learning_rate": 4.0283478932042246e-05,
      "loss": 0.7349,
      "step": 17412
    },
    {
      "epoch": 5.498263340700979,
      "grad_norm": 4.389832019805908,
      "learning_rate": 4.028260183138618e-05,
      "loss": 0.6585,
      "step": 17413
    },
    {
      "epoch": 5.498579096937164,
      "grad_norm": 3.9229378700256348,
      "learning_rate": 4.02817247307301e-05,
      "loss": 0.6578,
      "step": 17414
    },
    {
      "epoch": 5.498894853173351,
      "grad_norm": 3.7398338317871094,
      "learning_rate": 4.0280847630074034e-05,
      "loss": 0.6324,
      "step": 17415
    },
    {
      "epoch": 5.499210609409536,
      "grad_norm": 4.083322048187256,
      "learning_rate": 4.027997052941796e-05,
      "loss": 0.584,
      "step": 17416
    },
    {
      "epoch": 5.499526365645721,
      "grad_norm": 3.99062442779541,
      "learning_rate": 4.027909342876189e-05,
      "loss": 0.6444,
      "step": 17417
    },
    {
      "epoch": 5.499842121881907,
      "grad_norm": 3.61251163482666,
      "learning_rate": 4.0278216328105816e-05,
      "loss": 0.5444,
      "step": 17418
    },
    {
      "epoch": 5.500157878118093,
      "grad_norm": 4.086933135986328,
      "learning_rate": 4.027733922744975e-05,
      "loss": 0.602,
      "step": 17419
    },
    {
      "epoch": 5.500473634354279,
      "grad_norm": 3.327261447906494,
      "learning_rate": 4.027646212679367e-05,
      "loss": 0.5543,
      "step": 17420
    },
    {
      "epoch": 5.500789390590464,
      "grad_norm": 3.3431122303009033,
      "learning_rate": 4.0275585026137604e-05,
      "loss": 0.551,
      "step": 17421
    },
    {
      "epoch": 5.501105146826649,
      "grad_norm": 3.7337493896484375,
      "learning_rate": 4.027470792548153e-05,
      "loss": 0.6389,
      "step": 17422
    },
    {
      "epoch": 5.501420903062836,
      "grad_norm": 3.87760591506958,
      "learning_rate": 4.027383082482546e-05,
      "loss": 0.6235,
      "step": 17423
    },
    {
      "epoch": 5.501736659299021,
      "grad_norm": 4.46605110168457,
      "learning_rate": 4.0272953724169385e-05,
      "loss": 0.6667,
      "step": 17424
    },
    {
      "epoch": 5.502052415535207,
      "grad_norm": 4.6289801597595215,
      "learning_rate": 4.027207662351332e-05,
      "loss": 0.6915,
      "step": 17425
    },
    {
      "epoch": 5.502368171771392,
      "grad_norm": 2.83990478515625,
      "learning_rate": 4.027119952285724e-05,
      "loss": 0.5351,
      "step": 17426
    },
    {
      "epoch": 5.502683928007578,
      "grad_norm": 4.068299770355225,
      "learning_rate": 4.0270322422201174e-05,
      "loss": 0.75,
      "step": 17427
    },
    {
      "epoch": 5.502999684243764,
      "grad_norm": 4.250922203063965,
      "learning_rate": 4.02694453215451e-05,
      "loss": 0.6535,
      "step": 17428
    },
    {
      "epoch": 5.503315440479949,
      "grad_norm": 3.970231771469116,
      "learning_rate": 4.026856822088903e-05,
      "loss": 0.656,
      "step": 17429
    },
    {
      "epoch": 5.503631196716135,
      "grad_norm": 4.469651222229004,
      "learning_rate": 4.026769112023296e-05,
      "loss": 0.6547,
      "step": 17430
    },
    {
      "epoch": 5.503946952952321,
      "grad_norm": 3.9096758365631104,
      "learning_rate": 4.026681401957689e-05,
      "loss": 0.5823,
      "step": 17431
    },
    {
      "epoch": 5.504262709188507,
      "grad_norm": 4.319040775299072,
      "learning_rate": 4.026593691892082e-05,
      "loss": 0.607,
      "step": 17432
    },
    {
      "epoch": 5.504578465424692,
      "grad_norm": 4.322089672088623,
      "learning_rate": 4.026505981826474e-05,
      "loss": 0.6215,
      "step": 17433
    },
    {
      "epoch": 5.504894221660878,
      "grad_norm": 3.7471776008605957,
      "learning_rate": 4.0264182717608675e-05,
      "loss": 0.6669,
      "step": 17434
    },
    {
      "epoch": 5.505209977897064,
      "grad_norm": 3.7099990844726562,
      "learning_rate": 4.02633056169526e-05,
      "loss": 0.6073,
      "step": 17435
    },
    {
      "epoch": 5.505525734133249,
      "grad_norm": 3.782783269882202,
      "learning_rate": 4.026242851629653e-05,
      "loss": 0.7065,
      "step": 17436
    },
    {
      "epoch": 5.505841490369435,
      "grad_norm": 4.336002349853516,
      "learning_rate": 4.0261551415640457e-05,
      "loss": 0.7748,
      "step": 17437
    },
    {
      "epoch": 5.50615724660562,
      "grad_norm": 5.878425598144531,
      "learning_rate": 4.026067431498439e-05,
      "loss": 0.6478,
      "step": 17438
    },
    {
      "epoch": 5.5064730028418065,
      "grad_norm": 3.395480155944824,
      "learning_rate": 4.025979721432832e-05,
      "loss": 0.6351,
      "step": 17439
    },
    {
      "epoch": 5.506788759077992,
      "grad_norm": 4.522360801696777,
      "learning_rate": 4.025892011367225e-05,
      "loss": 0.6363,
      "step": 17440
    },
    {
      "epoch": 5.507104515314177,
      "grad_norm": 4.685554504394531,
      "learning_rate": 4.0258043013016176e-05,
      "loss": 0.6351,
      "step": 17441
    },
    {
      "epoch": 5.507420271550363,
      "grad_norm": 3.2759151458740234,
      "learning_rate": 4.025716591236011e-05,
      "loss": 0.6261,
      "step": 17442
    },
    {
      "epoch": 5.507736027786549,
      "grad_norm": 4.8905930519104,
      "learning_rate": 4.025628881170403e-05,
      "loss": 0.6984,
      "step": 17443
    },
    {
      "epoch": 5.508051784022735,
      "grad_norm": 3.5291945934295654,
      "learning_rate": 4.0255411711047965e-05,
      "loss": 0.5685,
      "step": 17444
    },
    {
      "epoch": 5.50836754025892,
      "grad_norm": 3.440600872039795,
      "learning_rate": 4.025453461039189e-05,
      "loss": 0.5961,
      "step": 17445
    },
    {
      "epoch": 5.508683296495105,
      "grad_norm": 3.841460943222046,
      "learning_rate": 4.025365750973582e-05,
      "loss": 0.5895,
      "step": 17446
    },
    {
      "epoch": 5.5089990527312915,
      "grad_norm": 4.309482574462891,
      "learning_rate": 4.0252780409079746e-05,
      "loss": 0.7548,
      "step": 17447
    },
    {
      "epoch": 5.509314808967477,
      "grad_norm": 4.702396392822266,
      "learning_rate": 4.025190330842368e-05,
      "loss": 0.6373,
      "step": 17448
    },
    {
      "epoch": 5.509630565203663,
      "grad_norm": 4.0770087242126465,
      "learning_rate": 4.02510262077676e-05,
      "loss": 0.6046,
      "step": 17449
    },
    {
      "epoch": 5.509946321439848,
      "grad_norm": 4.489800930023193,
      "learning_rate": 4.0250149107111535e-05,
      "loss": 0.5609,
      "step": 17450
    },
    {
      "epoch": 5.5102620776760345,
      "grad_norm": 4.039422035217285,
      "learning_rate": 4.0249272006455466e-05,
      "loss": 0.657,
      "step": 17451
    },
    {
      "epoch": 5.51057783391222,
      "grad_norm": 3.9966819286346436,
      "learning_rate": 4.024839490579939e-05,
      "loss": 0.5735,
      "step": 17452
    },
    {
      "epoch": 5.510893590148405,
      "grad_norm": 5.319180965423584,
      "learning_rate": 4.024751780514332e-05,
      "loss": 0.6232,
      "step": 17453
    },
    {
      "epoch": 5.511209346384591,
      "grad_norm": 3.951421022415161,
      "learning_rate": 4.024664070448725e-05,
      "loss": 0.6701,
      "step": 17454
    },
    {
      "epoch": 5.5115251026207766,
      "grad_norm": 4.015212059020996,
      "learning_rate": 4.024576360383118e-05,
      "loss": 0.6298,
      "step": 17455
    },
    {
      "epoch": 5.511840858856963,
      "grad_norm": 3.737746477127075,
      "learning_rate": 4.0244886503175104e-05,
      "loss": 0.7051,
      "step": 17456
    },
    {
      "epoch": 5.512156615093148,
      "grad_norm": 4.511964797973633,
      "learning_rate": 4.0244009402519036e-05,
      "loss": 0.6332,
      "step": 17457
    },
    {
      "epoch": 5.512472371329334,
      "grad_norm": 3.8386876583099365,
      "learning_rate": 4.024313230186296e-05,
      "loss": 0.5648,
      "step": 17458
    },
    {
      "epoch": 5.5127881275655195,
      "grad_norm": 3.6931331157684326,
      "learning_rate": 4.024225520120689e-05,
      "loss": 0.6205,
      "step": 17459
    },
    {
      "epoch": 5.513103883801705,
      "grad_norm": 4.307552337646484,
      "learning_rate": 4.024137810055082e-05,
      "loss": 0.5912,
      "step": 17460
    },
    {
      "epoch": 5.513419640037891,
      "grad_norm": 4.079631805419922,
      "learning_rate": 4.024050099989475e-05,
      "loss": 0.6458,
      "step": 17461
    },
    {
      "epoch": 5.513735396274076,
      "grad_norm": 3.887930154800415,
      "learning_rate": 4.0239623899238674e-05,
      "loss": 0.6422,
      "step": 17462
    },
    {
      "epoch": 5.5140511525102625,
      "grad_norm": 5.177415370941162,
      "learning_rate": 4.0238746798582606e-05,
      "loss": 0.6257,
      "step": 17463
    },
    {
      "epoch": 5.514366908746448,
      "grad_norm": 5.442290306091309,
      "learning_rate": 4.023786969792653e-05,
      "loss": 0.6457,
      "step": 17464
    },
    {
      "epoch": 5.514682664982633,
      "grad_norm": 4.252694606781006,
      "learning_rate": 4.023699259727046e-05,
      "loss": 0.7035,
      "step": 17465
    },
    {
      "epoch": 5.514998421218819,
      "grad_norm": 4.4510369300842285,
      "learning_rate": 4.0236115496614394e-05,
      "loss": 0.6587,
      "step": 17466
    },
    {
      "epoch": 5.5153141774550045,
      "grad_norm": 4.819692611694336,
      "learning_rate": 4.0235238395958326e-05,
      "loss": 0.6004,
      "step": 17467
    },
    {
      "epoch": 5.515629933691191,
      "grad_norm": 4.126108646392822,
      "learning_rate": 4.023436129530225e-05,
      "loss": 0.6281,
      "step": 17468
    },
    {
      "epoch": 5.515945689927376,
      "grad_norm": 4.271968841552734,
      "learning_rate": 4.023348419464618e-05,
      "loss": 0.634,
      "step": 17469
    },
    {
      "epoch": 5.516261446163561,
      "grad_norm": 4.557563304901123,
      "learning_rate": 4.023260709399011e-05,
      "loss": 0.5949,
      "step": 17470
    },
    {
      "epoch": 5.5165772023997475,
      "grad_norm": 4.155574798583984,
      "learning_rate": 4.023172999333404e-05,
      "loss": 0.6749,
      "step": 17471
    },
    {
      "epoch": 5.516892958635933,
      "grad_norm": 3.5941736698150635,
      "learning_rate": 4.0230852892677964e-05,
      "loss": 0.6546,
      "step": 17472
    },
    {
      "epoch": 5.517208714872119,
      "grad_norm": 4.237067699432373,
      "learning_rate": 4.0229975792021895e-05,
      "loss": 0.695,
      "step": 17473
    },
    {
      "epoch": 5.517524471108304,
      "grad_norm": 4.79852294921875,
      "learning_rate": 4.022909869136583e-05,
      "loss": 0.7075,
      "step": 17474
    },
    {
      "epoch": 5.51784022734449,
      "grad_norm": 3.840608835220337,
      "learning_rate": 4.022822159070975e-05,
      "loss": 0.6546,
      "step": 17475
    },
    {
      "epoch": 5.518155983580676,
      "grad_norm": 5.1965813636779785,
      "learning_rate": 4.0227344490053684e-05,
      "loss": 0.7217,
      "step": 17476
    },
    {
      "epoch": 5.518471739816861,
      "grad_norm": 3.83092999458313,
      "learning_rate": 4.022646738939761e-05,
      "loss": 0.5624,
      "step": 17477
    },
    {
      "epoch": 5.518787496053047,
      "grad_norm": 4.259329319000244,
      "learning_rate": 4.022559028874154e-05,
      "loss": 0.6386,
      "step": 17478
    },
    {
      "epoch": 5.5191032522892325,
      "grad_norm": 4.06601619720459,
      "learning_rate": 4.0224713188085465e-05,
      "loss": 0.6352,
      "step": 17479
    },
    {
      "epoch": 5.519419008525419,
      "grad_norm": 4.53019905090332,
      "learning_rate": 4.02238360874294e-05,
      "loss": 0.7116,
      "step": 17480
    },
    {
      "epoch": 5.519734764761604,
      "grad_norm": 3.8394947052001953,
      "learning_rate": 4.022295898677332e-05,
      "loss": 0.6252,
      "step": 17481
    },
    {
      "epoch": 5.52005052099779,
      "grad_norm": 4.203351020812988,
      "learning_rate": 4.0222081886117253e-05,
      "loss": 0.6421,
      "step": 17482
    },
    {
      "epoch": 5.520366277233975,
      "grad_norm": 4.187016010284424,
      "learning_rate": 4.022120478546118e-05,
      "loss": 0.6852,
      "step": 17483
    },
    {
      "epoch": 5.520682033470161,
      "grad_norm": 4.596848964691162,
      "learning_rate": 4.022032768480511e-05,
      "loss": 0.6864,
      "step": 17484
    },
    {
      "epoch": 5.520997789706347,
      "grad_norm": 4.474356174468994,
      "learning_rate": 4.0219450584149035e-05,
      "loss": 0.6361,
      "step": 17485
    },
    {
      "epoch": 5.521313545942532,
      "grad_norm": 4.294147491455078,
      "learning_rate": 4.021857348349297e-05,
      "loss": 0.7018,
      "step": 17486
    },
    {
      "epoch": 5.521629302178718,
      "grad_norm": 3.5370168685913086,
      "learning_rate": 4.021769638283689e-05,
      "loss": 0.6522,
      "step": 17487
    },
    {
      "epoch": 5.521945058414904,
      "grad_norm": 4.421918869018555,
      "learning_rate": 4.021681928218082e-05,
      "loss": 0.6507,
      "step": 17488
    },
    {
      "epoch": 5.522260814651089,
      "grad_norm": 4.4034037590026855,
      "learning_rate": 4.0215942181524755e-05,
      "loss": 0.6854,
      "step": 17489
    },
    {
      "epoch": 5.522576570887275,
      "grad_norm": 4.451444149017334,
      "learning_rate": 4.021506508086868e-05,
      "loss": 0.7045,
      "step": 17490
    },
    {
      "epoch": 5.5228923271234605,
      "grad_norm": 3.451517105102539,
      "learning_rate": 4.021418798021261e-05,
      "loss": 0.6994,
      "step": 17491
    },
    {
      "epoch": 5.523208083359647,
      "grad_norm": 3.5321052074432373,
      "learning_rate": 4.0213310879556536e-05,
      "loss": 0.589,
      "step": 17492
    },
    {
      "epoch": 5.523523839595832,
      "grad_norm": 4.077946662902832,
      "learning_rate": 4.021243377890047e-05,
      "loss": 0.6748,
      "step": 17493
    },
    {
      "epoch": 5.523839595832017,
      "grad_norm": 3.8895187377929688,
      "learning_rate": 4.02115566782444e-05,
      "loss": 0.5853,
      "step": 17494
    },
    {
      "epoch": 5.524155352068203,
      "grad_norm": 3.5960116386413574,
      "learning_rate": 4.021067957758833e-05,
      "loss": 0.5494,
      "step": 17495
    },
    {
      "epoch": 5.524471108304389,
      "grad_norm": 3.5308783054351807,
      "learning_rate": 4.0209802476932256e-05,
      "loss": 0.6203,
      "step": 17496
    },
    {
      "epoch": 5.524786864540575,
      "grad_norm": 3.889200210571289,
      "learning_rate": 4.020892537627619e-05,
      "loss": 0.6588,
      "step": 17497
    },
    {
      "epoch": 5.52510262077676,
      "grad_norm": 4.107665538787842,
      "learning_rate": 4.020804827562011e-05,
      "loss": 0.7451,
      "step": 17498
    },
    {
      "epoch": 5.525418377012946,
      "grad_norm": 3.6260383129119873,
      "learning_rate": 4.0207171174964045e-05,
      "loss": 0.6512,
      "step": 17499
    },
    {
      "epoch": 5.525734133249132,
      "grad_norm": 4.165966510772705,
      "learning_rate": 4.020629407430797e-05,
      "loss": 0.6693,
      "step": 17500
    },
    {
      "epoch": 5.526049889485318,
      "grad_norm": 3.3999133110046387,
      "learning_rate": 4.02054169736519e-05,
      "loss": 0.6398,
      "step": 17501
    },
    {
      "epoch": 5.526365645721503,
      "grad_norm": 4.01585578918457,
      "learning_rate": 4.0204539872995826e-05,
      "loss": 0.6849,
      "step": 17502
    },
    {
      "epoch": 5.526681401957688,
      "grad_norm": 4.359193801879883,
      "learning_rate": 4.020366277233976e-05,
      "loss": 0.7891,
      "step": 17503
    },
    {
      "epoch": 5.526997158193875,
      "grad_norm": 3.4834351539611816,
      "learning_rate": 4.020278567168368e-05,
      "loss": 0.5707,
      "step": 17504
    },
    {
      "epoch": 5.52731291443006,
      "grad_norm": 3.419821262359619,
      "learning_rate": 4.0201908571027614e-05,
      "loss": 0.5374,
      "step": 17505
    },
    {
      "epoch": 5.527628670666246,
      "grad_norm": 4.1736249923706055,
      "learning_rate": 4.020103147037154e-05,
      "loss": 0.7841,
      "step": 17506
    },
    {
      "epoch": 5.527944426902431,
      "grad_norm": 3.7029223442077637,
      "learning_rate": 4.020015436971547e-05,
      "loss": 0.5946,
      "step": 17507
    },
    {
      "epoch": 5.528260183138617,
      "grad_norm": 3.9685754776000977,
      "learning_rate": 4.0199277269059396e-05,
      "loss": 0.6279,
      "step": 17508
    },
    {
      "epoch": 5.528575939374803,
      "grad_norm": 3.3071839809417725,
      "learning_rate": 4.019840016840333e-05,
      "loss": 0.6257,
      "step": 17509
    },
    {
      "epoch": 5.528891695610988,
      "grad_norm": 3.525228500366211,
      "learning_rate": 4.019752306774725e-05,
      "loss": 0.5062,
      "step": 17510
    },
    {
      "epoch": 5.529207451847174,
      "grad_norm": 4.802046775817871,
      "learning_rate": 4.0196645967091184e-05,
      "loss": 0.583,
      "step": 17511
    },
    {
      "epoch": 5.52952320808336,
      "grad_norm": 3.952969551086426,
      "learning_rate": 4.0195768866435116e-05,
      "loss": 0.6096,
      "step": 17512
    },
    {
      "epoch": 5.529838964319545,
      "grad_norm": 4.623702526092529,
      "learning_rate": 4.019489176577904e-05,
      "loss": 0.6127,
      "step": 17513
    },
    {
      "epoch": 5.530154720555731,
      "grad_norm": 3.2017972469329834,
      "learning_rate": 4.019401466512297e-05,
      "loss": 0.6017,
      "step": 17514
    },
    {
      "epoch": 5.530470476791916,
      "grad_norm": 4.0926971435546875,
      "learning_rate": 4.01931375644669e-05,
      "loss": 0.6366,
      "step": 17515
    },
    {
      "epoch": 5.530786233028103,
      "grad_norm": 4.841480731964111,
      "learning_rate": 4.019226046381083e-05,
      "loss": 0.7263,
      "step": 17516
    },
    {
      "epoch": 5.531101989264288,
      "grad_norm": 4.333893299102783,
      "learning_rate": 4.0191383363154754e-05,
      "loss": 0.6347,
      "step": 17517
    },
    {
      "epoch": 5.531417745500473,
      "grad_norm": 4.080610275268555,
      "learning_rate": 4.0190506262498686e-05,
      "loss": 0.5525,
      "step": 17518
    },
    {
      "epoch": 5.531733501736659,
      "grad_norm": 4.091574192047119,
      "learning_rate": 4.018962916184261e-05,
      "loss": 0.5876,
      "step": 17519
    },
    {
      "epoch": 5.532049257972845,
      "grad_norm": 3.3037877082824707,
      "learning_rate": 4.018875206118655e-05,
      "loss": 0.6214,
      "step": 17520
    },
    {
      "epoch": 5.532365014209031,
      "grad_norm": 3.881829261779785,
      "learning_rate": 4.0187874960530474e-05,
      "loss": 0.5976,
      "step": 17521
    },
    {
      "epoch": 5.532680770445216,
      "grad_norm": 3.80407452583313,
      "learning_rate": 4.0186997859874406e-05,
      "loss": 0.5874,
      "step": 17522
    },
    {
      "epoch": 5.532996526681402,
      "grad_norm": 4.3042755126953125,
      "learning_rate": 4.018612075921833e-05,
      "loss": 0.6178,
      "step": 17523
    },
    {
      "epoch": 5.533312282917588,
      "grad_norm": 3.92193341255188,
      "learning_rate": 4.018524365856226e-05,
      "loss": 0.6023,
      "step": 17524
    },
    {
      "epoch": 5.533628039153774,
      "grad_norm": 4.976232528686523,
      "learning_rate": 4.018436655790619e-05,
      "loss": 0.677,
      "step": 17525
    },
    {
      "epoch": 5.533943795389959,
      "grad_norm": 4.225508213043213,
      "learning_rate": 4.018348945725012e-05,
      "loss": 0.6757,
      "step": 17526
    },
    {
      "epoch": 5.534259551626144,
      "grad_norm": 3.5284886360168457,
      "learning_rate": 4.0182612356594044e-05,
      "loss": 0.5721,
      "step": 17527
    },
    {
      "epoch": 5.5345753078623305,
      "grad_norm": 4.490128993988037,
      "learning_rate": 4.0181735255937975e-05,
      "loss": 0.627,
      "step": 17528
    },
    {
      "epoch": 5.534891064098516,
      "grad_norm": 4.37998104095459,
      "learning_rate": 4.01808581552819e-05,
      "loss": 0.6932,
      "step": 17529
    },
    {
      "epoch": 5.535206820334702,
      "grad_norm": 3.4624667167663574,
      "learning_rate": 4.017998105462583e-05,
      "loss": 0.6304,
      "step": 17530
    },
    {
      "epoch": 5.535522576570887,
      "grad_norm": 4.281011581420898,
      "learning_rate": 4.017910395396976e-05,
      "loss": 0.5822,
      "step": 17531
    },
    {
      "epoch": 5.535838332807073,
      "grad_norm": 4.651590347290039,
      "learning_rate": 4.017822685331369e-05,
      "loss": 0.7162,
      "step": 17532
    },
    {
      "epoch": 5.536154089043259,
      "grad_norm": 3.817164897918701,
      "learning_rate": 4.017734975265762e-05,
      "loss": 0.618,
      "step": 17533
    },
    {
      "epoch": 5.536469845279444,
      "grad_norm": 5.200525283813477,
      "learning_rate": 4.0176472652001545e-05,
      "loss": 0.6378,
      "step": 17534
    },
    {
      "epoch": 5.53678560151563,
      "grad_norm": 4.563488483428955,
      "learning_rate": 4.017559555134548e-05,
      "loss": 0.7964,
      "step": 17535
    },
    {
      "epoch": 5.5371013577518156,
      "grad_norm": 4.399718761444092,
      "learning_rate": 4.01747184506894e-05,
      "loss": 0.5638,
      "step": 17536
    },
    {
      "epoch": 5.537417113988001,
      "grad_norm": 3.553579092025757,
      "learning_rate": 4.017384135003333e-05,
      "loss": 0.589,
      "step": 17537
    },
    {
      "epoch": 5.537732870224187,
      "grad_norm": 4.342793941497803,
      "learning_rate": 4.017296424937726e-05,
      "loss": 0.5869,
      "step": 17538
    },
    {
      "epoch": 5.538048626460372,
      "grad_norm": 4.7104339599609375,
      "learning_rate": 4.017208714872119e-05,
      "loss": 0.6547,
      "step": 17539
    },
    {
      "epoch": 5.5383643826965585,
      "grad_norm": 4.809386253356934,
      "learning_rate": 4.0171210048065115e-05,
      "loss": 0.6054,
      "step": 17540
    },
    {
      "epoch": 5.538680138932744,
      "grad_norm": 4.023498058319092,
      "learning_rate": 4.0170332947409047e-05,
      "loss": 0.6093,
      "step": 17541
    },
    {
      "epoch": 5.538995895168929,
      "grad_norm": 4.593772888183594,
      "learning_rate": 4.016945584675297e-05,
      "loss": 0.6738,
      "step": 17542
    },
    {
      "epoch": 5.539311651405115,
      "grad_norm": 3.7588436603546143,
      "learning_rate": 4.01685787460969e-05,
      "loss": 0.5837,
      "step": 17543
    },
    {
      "epoch": 5.539627407641301,
      "grad_norm": 3.425973892211914,
      "learning_rate": 4.016770164544083e-05,
      "loss": 0.55,
      "step": 17544
    },
    {
      "epoch": 5.539943163877487,
      "grad_norm": 5.168362140655518,
      "learning_rate": 4.016682454478476e-05,
      "loss": 0.6765,
      "step": 17545
    },
    {
      "epoch": 5.540258920113672,
      "grad_norm": 3.9698173999786377,
      "learning_rate": 4.0165947444128685e-05,
      "loss": 0.5319,
      "step": 17546
    },
    {
      "epoch": 5.540574676349858,
      "grad_norm": 3.4120407104492188,
      "learning_rate": 4.0165070343472616e-05,
      "loss": 0.6253,
      "step": 17547
    },
    {
      "epoch": 5.5408904325860435,
      "grad_norm": 3.502725601196289,
      "learning_rate": 4.016419324281655e-05,
      "loss": 0.5938,
      "step": 17548
    },
    {
      "epoch": 5.54120618882223,
      "grad_norm": 4.172694206237793,
      "learning_rate": 4.016331614216048e-05,
      "loss": 0.6686,
      "step": 17549
    },
    {
      "epoch": 5.541521945058415,
      "grad_norm": 3.477468967437744,
      "learning_rate": 4.0162439041504405e-05,
      "loss": 0.5404,
      "step": 17550
    },
    {
      "epoch": 5.5418377012946,
      "grad_norm": 4.334348201751709,
      "learning_rate": 4.0161561940848336e-05,
      "loss": 0.7181,
      "step": 17551
    },
    {
      "epoch": 5.5421534575307865,
      "grad_norm": 3.5902931690216064,
      "learning_rate": 4.016068484019226e-05,
      "loss": 0.5814,
      "step": 17552
    },
    {
      "epoch": 5.542469213766972,
      "grad_norm": 4.91360330581665,
      "learning_rate": 4.015980773953619e-05,
      "loss": 0.6842,
      "step": 17553
    },
    {
      "epoch": 5.542784970003158,
      "grad_norm": 4.340245723724365,
      "learning_rate": 4.0158930638880125e-05,
      "loss": 0.6299,
      "step": 17554
    },
    {
      "epoch": 5.543100726239343,
      "grad_norm": 3.9836878776550293,
      "learning_rate": 4.015805353822405e-05,
      "loss": 0.7362,
      "step": 17555
    },
    {
      "epoch": 5.5434164824755285,
      "grad_norm": 5.293654441833496,
      "learning_rate": 4.015717643756798e-05,
      "loss": 0.5886,
      "step": 17556
    },
    {
      "epoch": 5.543732238711715,
      "grad_norm": 4.354039192199707,
      "learning_rate": 4.0156299336911906e-05,
      "loss": 0.6431,
      "step": 17557
    },
    {
      "epoch": 5.5440479949479,
      "grad_norm": 4.390504360198975,
      "learning_rate": 4.015542223625584e-05,
      "loss": 0.649,
      "step": 17558
    },
    {
      "epoch": 5.544363751184086,
      "grad_norm": 6.167964935302734,
      "learning_rate": 4.015454513559976e-05,
      "loss": 0.6912,
      "step": 17559
    },
    {
      "epoch": 5.5446795074202715,
      "grad_norm": 4.52760124206543,
      "learning_rate": 4.0153668034943694e-05,
      "loss": 0.6645,
      "step": 17560
    },
    {
      "epoch": 5.544995263656457,
      "grad_norm": 4.462258338928223,
      "learning_rate": 4.015279093428762e-05,
      "loss": 0.6888,
      "step": 17561
    },
    {
      "epoch": 5.545311019892643,
      "grad_norm": 4.791197776794434,
      "learning_rate": 4.015191383363155e-05,
      "loss": 0.7057,
      "step": 17562
    },
    {
      "epoch": 5.545626776128828,
      "grad_norm": 3.8390893936157227,
      "learning_rate": 4.0151036732975476e-05,
      "loss": 0.611,
      "step": 17563
    },
    {
      "epoch": 5.545942532365014,
      "grad_norm": 3.461966037750244,
      "learning_rate": 4.015015963231941e-05,
      "loss": 0.581,
      "step": 17564
    },
    {
      "epoch": 5.5462582886012,
      "grad_norm": 4.08571720123291,
      "learning_rate": 4.014928253166333e-05,
      "loss": 0.6464,
      "step": 17565
    },
    {
      "epoch": 5.546574044837386,
      "grad_norm": 3.7730712890625,
      "learning_rate": 4.0148405431007264e-05,
      "loss": 0.6091,
      "step": 17566
    },
    {
      "epoch": 5.546889801073571,
      "grad_norm": 4.421401500701904,
      "learning_rate": 4.014752833035119e-05,
      "loss": 0.7619,
      "step": 17567
    },
    {
      "epoch": 5.5472055573097565,
      "grad_norm": 4.959315299987793,
      "learning_rate": 4.014665122969512e-05,
      "loss": 0.7215,
      "step": 17568
    },
    {
      "epoch": 5.547521313545943,
      "grad_norm": 4.037442684173584,
      "learning_rate": 4.0145774129039046e-05,
      "loss": 0.544,
      "step": 17569
    },
    {
      "epoch": 5.547837069782128,
      "grad_norm": 3.9883251190185547,
      "learning_rate": 4.014489702838298e-05,
      "loss": 0.6493,
      "step": 17570
    },
    {
      "epoch": 5.548152826018314,
      "grad_norm": 5.2181878089904785,
      "learning_rate": 4.014401992772691e-05,
      "loss": 0.6323,
      "step": 17571
    },
    {
      "epoch": 5.5484685822544995,
      "grad_norm": 3.7322614192962646,
      "learning_rate": 4.0143142827070834e-05,
      "loss": 0.6363,
      "step": 17572
    },
    {
      "epoch": 5.548784338490686,
      "grad_norm": 3.784435987472534,
      "learning_rate": 4.0142265726414766e-05,
      "loss": 0.6917,
      "step": 17573
    },
    {
      "epoch": 5.549100094726871,
      "grad_norm": 3.243882417678833,
      "learning_rate": 4.014138862575869e-05,
      "loss": 0.5656,
      "step": 17574
    },
    {
      "epoch": 5.549415850963056,
      "grad_norm": 3.695712089538574,
      "learning_rate": 4.014051152510262e-05,
      "loss": 0.5494,
      "step": 17575
    },
    {
      "epoch": 5.549731607199242,
      "grad_norm": 5.335941791534424,
      "learning_rate": 4.0139634424446554e-05,
      "loss": 0.6375,
      "step": 17576
    },
    {
      "epoch": 5.550047363435428,
      "grad_norm": 3.830132007598877,
      "learning_rate": 4.0138757323790485e-05,
      "loss": 0.6659,
      "step": 17577
    },
    {
      "epoch": 5.550363119671614,
      "grad_norm": 3.789980173110962,
      "learning_rate": 4.013788022313441e-05,
      "loss": 0.6155,
      "step": 17578
    },
    {
      "epoch": 5.550678875907799,
      "grad_norm": 4.666609764099121,
      "learning_rate": 4.013700312247834e-05,
      "loss": 0.7194,
      "step": 17579
    },
    {
      "epoch": 5.5509946321439845,
      "grad_norm": 3.2936453819274902,
      "learning_rate": 4.013612602182227e-05,
      "loss": 0.5196,
      "step": 17580
    },
    {
      "epoch": 5.551310388380171,
      "grad_norm": 3.885969877243042,
      "learning_rate": 4.01352489211662e-05,
      "loss": 0.5887,
      "step": 17581
    },
    {
      "epoch": 5.551626144616356,
      "grad_norm": 4.767049312591553,
      "learning_rate": 4.0134371820510124e-05,
      "loss": 0.6277,
      "step": 17582
    },
    {
      "epoch": 5.551941900852542,
      "grad_norm": 4.672749996185303,
      "learning_rate": 4.0133494719854055e-05,
      "loss": 0.6477,
      "step": 17583
    },
    {
      "epoch": 5.552257657088727,
      "grad_norm": 3.993074417114258,
      "learning_rate": 4.013261761919798e-05,
      "loss": 0.6836,
      "step": 17584
    },
    {
      "epoch": 5.552573413324913,
      "grad_norm": 4.151558876037598,
      "learning_rate": 4.013174051854191e-05,
      "loss": 0.504,
      "step": 17585
    },
    {
      "epoch": 5.552889169561099,
      "grad_norm": 3.947516679763794,
      "learning_rate": 4.013086341788584e-05,
      "loss": 0.7105,
      "step": 17586
    },
    {
      "epoch": 5.553204925797284,
      "grad_norm": 3.986144542694092,
      "learning_rate": 4.012998631722977e-05,
      "loss": 0.5492,
      "step": 17587
    },
    {
      "epoch": 5.55352068203347,
      "grad_norm": 4.399022579193115,
      "learning_rate": 4.012910921657369e-05,
      "loss": 0.5878,
      "step": 17588
    },
    {
      "epoch": 5.553836438269656,
      "grad_norm": 3.871674060821533,
      "learning_rate": 4.0128232115917625e-05,
      "loss": 0.6838,
      "step": 17589
    },
    {
      "epoch": 5.554152194505842,
      "grad_norm": 4.2479705810546875,
      "learning_rate": 4.012735501526155e-05,
      "loss": 0.6087,
      "step": 17590
    },
    {
      "epoch": 5.554467950742027,
      "grad_norm": 5.850827217102051,
      "learning_rate": 4.012647791460548e-05,
      "loss": 0.626,
      "step": 17591
    },
    {
      "epoch": 5.554783706978212,
      "grad_norm": 3.545935869216919,
      "learning_rate": 4.012560081394941e-05,
      "loss": 0.5258,
      "step": 17592
    },
    {
      "epoch": 5.555099463214399,
      "grad_norm": 4.244142055511475,
      "learning_rate": 4.012472371329334e-05,
      "loss": 0.6352,
      "step": 17593
    },
    {
      "epoch": 5.555415219450584,
      "grad_norm": 4.571516990661621,
      "learning_rate": 4.012384661263727e-05,
      "loss": 0.6963,
      "step": 17594
    },
    {
      "epoch": 5.55573097568677,
      "grad_norm": 3.4653825759887695,
      "learning_rate": 4.0122969511981195e-05,
      "loss": 0.5428,
      "step": 17595
    },
    {
      "epoch": 5.556046731922955,
      "grad_norm": 5.062540531158447,
      "learning_rate": 4.0122092411325126e-05,
      "loss": 0.6244,
      "step": 17596
    },
    {
      "epoch": 5.556362488159142,
      "grad_norm": 4.156033039093018,
      "learning_rate": 4.012121531066905e-05,
      "loss": 0.6425,
      "step": 17597
    },
    {
      "epoch": 5.556678244395327,
      "grad_norm": 4.631798267364502,
      "learning_rate": 4.012033821001298e-05,
      "loss": 0.6374,
      "step": 17598
    },
    {
      "epoch": 5.556994000631512,
      "grad_norm": 4.0528883934021,
      "learning_rate": 4.011946110935691e-05,
      "loss": 0.6782,
      "step": 17599
    },
    {
      "epoch": 5.557309756867698,
      "grad_norm": 3.875000238418579,
      "learning_rate": 4.011858400870084e-05,
      "loss": 0.6413,
      "step": 17600
    },
    {
      "epoch": 5.557309756867698,
      "eval_loss": 0.8418887257575989,
      "eval_runtime": 4.5462,
      "eval_samples_per_second": 547.715,
      "eval_steps_per_second": 4.619,
      "step": 17600
    },
    {
      "epoch": 5.557625513103884,
      "grad_norm": 3.985647439956665,
      "learning_rate": 4.0117706908044765e-05,
      "loss": 0.6371,
      "step": 17601
    },
    {
      "epoch": 5.55794126934007,
      "grad_norm": 5.257065296173096,
      "learning_rate": 4.01168298073887e-05,
      "loss": 0.566,
      "step": 17602
    },
    {
      "epoch": 5.558257025576255,
      "grad_norm": 3.9744324684143066,
      "learning_rate": 4.011595270673263e-05,
      "loss": 0.6468,
      "step": 17603
    },
    {
      "epoch": 5.55857278181244,
      "grad_norm": 4.068905830383301,
      "learning_rate": 4.011507560607656e-05,
      "loss": 0.658,
      "step": 17604
    },
    {
      "epoch": 5.558888538048627,
      "grad_norm": 4.41050386428833,
      "learning_rate": 4.0114198505420484e-05,
      "loss": 0.6124,
      "step": 17605
    },
    {
      "epoch": 5.559204294284812,
      "grad_norm": 4.709972381591797,
      "learning_rate": 4.0113321404764416e-05,
      "loss": 0.6272,
      "step": 17606
    },
    {
      "epoch": 5.559520050520998,
      "grad_norm": 6.1815266609191895,
      "learning_rate": 4.011244430410834e-05,
      "loss": 0.6283,
      "step": 17607
    },
    {
      "epoch": 5.559835806757183,
      "grad_norm": 4.946155071258545,
      "learning_rate": 4.011156720345227e-05,
      "loss": 0.6757,
      "step": 17608
    },
    {
      "epoch": 5.560151562993369,
      "grad_norm": 4.446709156036377,
      "learning_rate": 4.01106901027962e-05,
      "loss": 0.7295,
      "step": 17609
    },
    {
      "epoch": 5.560467319229555,
      "grad_norm": 3.9425456523895264,
      "learning_rate": 4.010981300214013e-05,
      "loss": 0.5869,
      "step": 17610
    },
    {
      "epoch": 5.56078307546574,
      "grad_norm": 4.449418544769287,
      "learning_rate": 4.0108935901484054e-05,
      "loss": 0.6307,
      "step": 17611
    },
    {
      "epoch": 5.561098831701926,
      "grad_norm": 3.7739880084991455,
      "learning_rate": 4.0108058800827986e-05,
      "loss": 0.6274,
      "step": 17612
    },
    {
      "epoch": 5.561414587938112,
      "grad_norm": 4.626743316650391,
      "learning_rate": 4.010718170017191e-05,
      "loss": 0.7354,
      "step": 17613
    },
    {
      "epoch": 5.561730344174298,
      "grad_norm": 3.1414647102355957,
      "learning_rate": 4.010630459951584e-05,
      "loss": 0.5621,
      "step": 17614
    },
    {
      "epoch": 5.562046100410483,
      "grad_norm": 4.633947372436523,
      "learning_rate": 4.0105427498859774e-05,
      "loss": 0.7182,
      "step": 17615
    },
    {
      "epoch": 5.562361856646669,
      "grad_norm": 5.452261447906494,
      "learning_rate": 4.01045503982037e-05,
      "loss": 0.6978,
      "step": 17616
    },
    {
      "epoch": 5.562677612882855,
      "grad_norm": 4.425944805145264,
      "learning_rate": 4.010367329754763e-05,
      "loss": 0.6656,
      "step": 17617
    },
    {
      "epoch": 5.56299336911904,
      "grad_norm": 3.7388861179351807,
      "learning_rate": 4.0102796196891556e-05,
      "loss": 0.6257,
      "step": 17618
    },
    {
      "epoch": 5.563309125355226,
      "grad_norm": 3.47247314453125,
      "learning_rate": 4.010191909623549e-05,
      "loss": 0.5396,
      "step": 17619
    },
    {
      "epoch": 5.563624881591411,
      "grad_norm": 4.620634078979492,
      "learning_rate": 4.010104199557941e-05,
      "loss": 0.6503,
      "step": 17620
    },
    {
      "epoch": 5.5639406378275975,
      "grad_norm": 4.672364711761475,
      "learning_rate": 4.0100164894923344e-05,
      "loss": 0.727,
      "step": 17621
    },
    {
      "epoch": 5.564256394063783,
      "grad_norm": 3.7466301918029785,
      "learning_rate": 4.009928779426727e-05,
      "loss": 0.6358,
      "step": 17622
    },
    {
      "epoch": 5.564572150299968,
      "grad_norm": 4.449415683746338,
      "learning_rate": 4.00984106936112e-05,
      "loss": 0.7406,
      "step": 17623
    },
    {
      "epoch": 5.564887906536154,
      "grad_norm": 4.328794479370117,
      "learning_rate": 4.0097533592955125e-05,
      "loss": 0.6783,
      "step": 17624
    },
    {
      "epoch": 5.56520366277234,
      "grad_norm": 3.9136886596679688,
      "learning_rate": 4.009665649229906e-05,
      "loss": 0.6816,
      "step": 17625
    },
    {
      "epoch": 5.565519419008526,
      "grad_norm": 5.130120277404785,
      "learning_rate": 4.009577939164298e-05,
      "loss": 0.7148,
      "step": 17626
    },
    {
      "epoch": 5.565835175244711,
      "grad_norm": 3.752737283706665,
      "learning_rate": 4.0094902290986914e-05,
      "loss": 0.6188,
      "step": 17627
    },
    {
      "epoch": 5.566150931480896,
      "grad_norm": 4.145195960998535,
      "learning_rate": 4.009402519033084e-05,
      "loss": 0.5504,
      "step": 17628
    },
    {
      "epoch": 5.5664666877170825,
      "grad_norm": 4.300090789794922,
      "learning_rate": 4.009314808967478e-05,
      "loss": 0.6498,
      "step": 17629
    },
    {
      "epoch": 5.566782443953268,
      "grad_norm": 4.115471363067627,
      "learning_rate": 4.00922709890187e-05,
      "loss": 0.6702,
      "step": 17630
    },
    {
      "epoch": 5.567098200189454,
      "grad_norm": 5.034883499145508,
      "learning_rate": 4.0091393888362634e-05,
      "loss": 0.6074,
      "step": 17631
    },
    {
      "epoch": 5.567413956425639,
      "grad_norm": 4.670183181762695,
      "learning_rate": 4.009051678770656e-05,
      "loss": 0.5769,
      "step": 17632
    },
    {
      "epoch": 5.567729712661825,
      "grad_norm": 3.66048526763916,
      "learning_rate": 4.008963968705049e-05,
      "loss": 0.5878,
      "step": 17633
    },
    {
      "epoch": 5.568045468898011,
      "grad_norm": 3.6845176219940186,
      "learning_rate": 4.0088762586394415e-05,
      "loss": 0.6745,
      "step": 17634
    },
    {
      "epoch": 5.568361225134196,
      "grad_norm": 4.922713756561279,
      "learning_rate": 4.008788548573835e-05,
      "loss": 0.6623,
      "step": 17635
    },
    {
      "epoch": 5.568676981370382,
      "grad_norm": 4.771240711212158,
      "learning_rate": 4.008700838508228e-05,
      "loss": 0.6288,
      "step": 17636
    },
    {
      "epoch": 5.5689927376065675,
      "grad_norm": 4.027492046356201,
      "learning_rate": 4.0086131284426203e-05,
      "loss": 0.6082,
      "step": 17637
    },
    {
      "epoch": 5.569308493842754,
      "grad_norm": 3.7967329025268555,
      "learning_rate": 4.0085254183770135e-05,
      "loss": 0.5871,
      "step": 17638
    },
    {
      "epoch": 5.569624250078939,
      "grad_norm": 4.824473857879639,
      "learning_rate": 4.008437708311406e-05,
      "loss": 0.7269,
      "step": 17639
    },
    {
      "epoch": 5.569940006315125,
      "grad_norm": 3.655327796936035,
      "learning_rate": 4.008349998245799e-05,
      "loss": 0.6248,
      "step": 17640
    },
    {
      "epoch": 5.5702557625513105,
      "grad_norm": 3.7562191486358643,
      "learning_rate": 4.008262288180192e-05,
      "loss": 0.7302,
      "step": 17641
    },
    {
      "epoch": 5.570571518787496,
      "grad_norm": 4.01252555847168,
      "learning_rate": 4.008174578114585e-05,
      "loss": 0.6374,
      "step": 17642
    },
    {
      "epoch": 5.570887275023682,
      "grad_norm": 4.272303104400635,
      "learning_rate": 4.008086868048977e-05,
      "loss": 0.7093,
      "step": 17643
    },
    {
      "epoch": 5.571203031259867,
      "grad_norm": 5.186147212982178,
      "learning_rate": 4.0079991579833705e-05,
      "loss": 0.6657,
      "step": 17644
    },
    {
      "epoch": 5.5715187874960534,
      "grad_norm": 3.6880617141723633,
      "learning_rate": 4.007911447917763e-05,
      "loss": 0.614,
      "step": 17645
    },
    {
      "epoch": 5.571834543732239,
      "grad_norm": 4.465967178344727,
      "learning_rate": 4.007823737852156e-05,
      "loss": 0.5512,
      "step": 17646
    },
    {
      "epoch": 5.572150299968424,
      "grad_norm": 4.442500591278076,
      "learning_rate": 4.0077360277865486e-05,
      "loss": 0.6432,
      "step": 17647
    },
    {
      "epoch": 5.57246605620461,
      "grad_norm": 4.148310661315918,
      "learning_rate": 4.007648317720942e-05,
      "loss": 0.5696,
      "step": 17648
    },
    {
      "epoch": 5.5727818124407955,
      "grad_norm": 3.8516876697540283,
      "learning_rate": 4.007560607655334e-05,
      "loss": 0.6593,
      "step": 17649
    },
    {
      "epoch": 5.573097568676982,
      "grad_norm": 4.109030246734619,
      "learning_rate": 4.0074728975897275e-05,
      "loss": 0.6846,
      "step": 17650
    },
    {
      "epoch": 5.573413324913167,
      "grad_norm": 4.072408676147461,
      "learning_rate": 4.00738518752412e-05,
      "loss": 0.6558,
      "step": 17651
    },
    {
      "epoch": 5.573729081149352,
      "grad_norm": 3.8991236686706543,
      "learning_rate": 4.007297477458513e-05,
      "loss": 0.7474,
      "step": 17652
    },
    {
      "epoch": 5.5740448373855385,
      "grad_norm": 3.2037789821624756,
      "learning_rate": 4.007209767392906e-05,
      "loss": 0.6257,
      "step": 17653
    },
    {
      "epoch": 5.574360593621724,
      "grad_norm": 3.552834987640381,
      "learning_rate": 4.007122057327299e-05,
      "loss": 0.7139,
      "step": 17654
    },
    {
      "epoch": 5.57467634985791,
      "grad_norm": 3.5957250595092773,
      "learning_rate": 4.007034347261692e-05,
      "loss": 0.6194,
      "step": 17655
    },
    {
      "epoch": 5.574992106094095,
      "grad_norm": 4.809767723083496,
      "learning_rate": 4.0069466371960844e-05,
      "loss": 0.6207,
      "step": 17656
    },
    {
      "epoch": 5.5753078623302805,
      "grad_norm": 4.224508762359619,
      "learning_rate": 4.0068589271304776e-05,
      "loss": 0.6319,
      "step": 17657
    },
    {
      "epoch": 5.575623618566467,
      "grad_norm": 3.7581722736358643,
      "learning_rate": 4.006771217064871e-05,
      "loss": 0.6636,
      "step": 17658
    },
    {
      "epoch": 5.575939374802652,
      "grad_norm": 4.124964714050293,
      "learning_rate": 4.006683506999264e-05,
      "loss": 0.5921,
      "step": 17659
    },
    {
      "epoch": 5.576255131038838,
      "grad_norm": 3.91544246673584,
      "learning_rate": 4.0065957969336564e-05,
      "loss": 0.6697,
      "step": 17660
    },
    {
      "epoch": 5.5765708872750235,
      "grad_norm": 4.174775123596191,
      "learning_rate": 4.0065080868680496e-05,
      "loss": 0.687,
      "step": 17661
    },
    {
      "epoch": 5.57688664351121,
      "grad_norm": 3.8506202697753906,
      "learning_rate": 4.006420376802442e-05,
      "loss": 0.6505,
      "step": 17662
    },
    {
      "epoch": 5.577202399747395,
      "grad_norm": 4.614487171173096,
      "learning_rate": 4.006332666736835e-05,
      "loss": 0.7003,
      "step": 17663
    },
    {
      "epoch": 5.577518155983581,
      "grad_norm": 3.510200262069702,
      "learning_rate": 4.006244956671228e-05,
      "loss": 0.5393,
      "step": 17664
    },
    {
      "epoch": 5.577833912219766,
      "grad_norm": 4.523581504821777,
      "learning_rate": 4.006157246605621e-05,
      "loss": 0.6566,
      "step": 17665
    },
    {
      "epoch": 5.578149668455952,
      "grad_norm": 4.588994979858398,
      "learning_rate": 4.0060695365400134e-05,
      "loss": 0.7042,
      "step": 17666
    },
    {
      "epoch": 5.578465424692138,
      "grad_norm": 3.673349380493164,
      "learning_rate": 4.0059818264744066e-05,
      "loss": 0.6213,
      "step": 17667
    },
    {
      "epoch": 5.578781180928323,
      "grad_norm": 4.43607759475708,
      "learning_rate": 4.005894116408799e-05,
      "loss": 0.5821,
      "step": 17668
    },
    {
      "epoch": 5.579096937164509,
      "grad_norm": 3.5886640548706055,
      "learning_rate": 4.005806406343192e-05,
      "loss": 0.6367,
      "step": 17669
    },
    {
      "epoch": 5.579412693400695,
      "grad_norm": 4.033845901489258,
      "learning_rate": 4.005718696277585e-05,
      "loss": 0.6133,
      "step": 17670
    },
    {
      "epoch": 5.57972844963688,
      "grad_norm": 4.285686016082764,
      "learning_rate": 4.005630986211978e-05,
      "loss": 0.5871,
      "step": 17671
    },
    {
      "epoch": 5.580044205873066,
      "grad_norm": 5.744317531585693,
      "learning_rate": 4.0055432761463704e-05,
      "loss": 0.5469,
      "step": 17672
    },
    {
      "epoch": 5.5803599621092514,
      "grad_norm": 4.357353687286377,
      "learning_rate": 4.0054555660807636e-05,
      "loss": 0.6586,
      "step": 17673
    },
    {
      "epoch": 5.580675718345438,
      "grad_norm": 5.024135589599609,
      "learning_rate": 4.005367856015157e-05,
      "loss": 0.6525,
      "step": 17674
    },
    {
      "epoch": 5.580991474581623,
      "grad_norm": 4.215052604675293,
      "learning_rate": 4.005280145949549e-05,
      "loss": 0.6752,
      "step": 17675
    },
    {
      "epoch": 5.581307230817808,
      "grad_norm": 4.619220733642578,
      "learning_rate": 4.0051924358839424e-05,
      "loss": 0.6522,
      "step": 17676
    },
    {
      "epoch": 5.581622987053994,
      "grad_norm": 4.107060432434082,
      "learning_rate": 4.005104725818335e-05,
      "loss": 0.641,
      "step": 17677
    },
    {
      "epoch": 5.58193874329018,
      "grad_norm": 4.387732982635498,
      "learning_rate": 4.005017015752728e-05,
      "loss": 0.5316,
      "step": 17678
    },
    {
      "epoch": 5.582254499526366,
      "grad_norm": 4.099691390991211,
      "learning_rate": 4.0049293056871205e-05,
      "loss": 0.5976,
      "step": 17679
    },
    {
      "epoch": 5.582570255762551,
      "grad_norm": 3.673671007156372,
      "learning_rate": 4.004841595621514e-05,
      "loss": 0.665,
      "step": 17680
    },
    {
      "epoch": 5.582886011998737,
      "grad_norm": 4.250738143920898,
      "learning_rate": 4.004753885555906e-05,
      "loss": 0.6344,
      "step": 17681
    },
    {
      "epoch": 5.583201768234923,
      "grad_norm": 3.6199660301208496,
      "learning_rate": 4.0046661754902994e-05,
      "loss": 0.631,
      "step": 17682
    },
    {
      "epoch": 5.583517524471108,
      "grad_norm": 5.06019401550293,
      "learning_rate": 4.004578465424692e-05,
      "loss": 0.6704,
      "step": 17683
    },
    {
      "epoch": 5.583833280707294,
      "grad_norm": 3.6132752895355225,
      "learning_rate": 4.004490755359086e-05,
      "loss": 0.5376,
      "step": 17684
    },
    {
      "epoch": 5.584149036943479,
      "grad_norm": 3.919264793395996,
      "learning_rate": 4.004403045293478e-05,
      "loss": 0.6766,
      "step": 17685
    },
    {
      "epoch": 5.584464793179666,
      "grad_norm": 4.65828800201416,
      "learning_rate": 4.0043153352278714e-05,
      "loss": 0.7072,
      "step": 17686
    },
    {
      "epoch": 5.584780549415851,
      "grad_norm": 4.555442810058594,
      "learning_rate": 4.004227625162264e-05,
      "loss": 0.6088,
      "step": 17687
    },
    {
      "epoch": 5.585096305652037,
      "grad_norm": 5.0989089012146,
      "learning_rate": 4.004139915096657e-05,
      "loss": 0.7123,
      "step": 17688
    },
    {
      "epoch": 5.585412061888222,
      "grad_norm": 4.9241557121276855,
      "learning_rate": 4.0040522050310495e-05,
      "loss": 0.6025,
      "step": 17689
    },
    {
      "epoch": 5.585727818124408,
      "grad_norm": 4.5822014808654785,
      "learning_rate": 4.003964494965443e-05,
      "loss": 0.6518,
      "step": 17690
    },
    {
      "epoch": 5.586043574360594,
      "grad_norm": 3.657536745071411,
      "learning_rate": 4.003876784899835e-05,
      "loss": 0.5656,
      "step": 17691
    },
    {
      "epoch": 5.586359330596779,
      "grad_norm": 3.942756414413452,
      "learning_rate": 4.003789074834228e-05,
      "loss": 0.6646,
      "step": 17692
    },
    {
      "epoch": 5.586675086832965,
      "grad_norm": 3.746206760406494,
      "learning_rate": 4.003701364768621e-05,
      "loss": 0.6232,
      "step": 17693
    },
    {
      "epoch": 5.586990843069151,
      "grad_norm": 5.42611026763916,
      "learning_rate": 4.003613654703014e-05,
      "loss": 0.7309,
      "step": 17694
    },
    {
      "epoch": 5.587306599305336,
      "grad_norm": 3.814173698425293,
      "learning_rate": 4.0035259446374065e-05,
      "loss": 0.6021,
      "step": 17695
    },
    {
      "epoch": 5.587622355541522,
      "grad_norm": 4.8968186378479,
      "learning_rate": 4.0034382345717997e-05,
      "loss": 0.6154,
      "step": 17696
    },
    {
      "epoch": 5.587938111777707,
      "grad_norm": 3.824902057647705,
      "learning_rate": 4.003350524506193e-05,
      "loss": 0.6483,
      "step": 17697
    },
    {
      "epoch": 5.588253868013894,
      "grad_norm": 3.715557336807251,
      "learning_rate": 4.003262814440585e-05,
      "loss": 0.5922,
      "step": 17698
    },
    {
      "epoch": 5.588569624250079,
      "grad_norm": 4.935893535614014,
      "learning_rate": 4.0031751043749785e-05,
      "loss": 0.7859,
      "step": 17699
    },
    {
      "epoch": 5.588885380486264,
      "grad_norm": 4.537399768829346,
      "learning_rate": 4.003087394309371e-05,
      "loss": 0.667,
      "step": 17700
    },
    {
      "epoch": 5.58920113672245,
      "grad_norm": 4.519407272338867,
      "learning_rate": 4.002999684243764e-05,
      "loss": 0.6838,
      "step": 17701
    },
    {
      "epoch": 5.589516892958636,
      "grad_norm": 3.580709457397461,
      "learning_rate": 4.0029119741781566e-05,
      "loss": 0.5711,
      "step": 17702
    },
    {
      "epoch": 5.589832649194822,
      "grad_norm": 4.036377429962158,
      "learning_rate": 4.00282426411255e-05,
      "loss": 0.5685,
      "step": 17703
    },
    {
      "epoch": 5.590148405431007,
      "grad_norm": 3.3819339275360107,
      "learning_rate": 4.002736554046942e-05,
      "loss": 0.6808,
      "step": 17704
    },
    {
      "epoch": 5.590464161667193,
      "grad_norm": 4.136167526245117,
      "learning_rate": 4.0026488439813355e-05,
      "loss": 0.6692,
      "step": 17705
    },
    {
      "epoch": 5.590779917903379,
      "grad_norm": 3.6721553802490234,
      "learning_rate": 4.002561133915728e-05,
      "loss": 0.5179,
      "step": 17706
    },
    {
      "epoch": 5.591095674139564,
      "grad_norm": 4.237932205200195,
      "learning_rate": 4.002473423850121e-05,
      "loss": 0.7038,
      "step": 17707
    },
    {
      "epoch": 5.59141143037575,
      "grad_norm": 4.228872299194336,
      "learning_rate": 4.0023857137845136e-05,
      "loss": 0.6243,
      "step": 17708
    },
    {
      "epoch": 5.591727186611935,
      "grad_norm": 3.833496332168579,
      "learning_rate": 4.002298003718907e-05,
      "loss": 0.616,
      "step": 17709
    },
    {
      "epoch": 5.5920429428481215,
      "grad_norm": 3.9603450298309326,
      "learning_rate": 4.002210293653299e-05,
      "loss": 0.7025,
      "step": 17710
    },
    {
      "epoch": 5.592358699084307,
      "grad_norm": 3.5454022884368896,
      "learning_rate": 4.002122583587693e-05,
      "loss": 0.5567,
      "step": 17711
    },
    {
      "epoch": 5.592674455320493,
      "grad_norm": 3.8531572818756104,
      "learning_rate": 4.0020348735220856e-05,
      "loss": 0.6104,
      "step": 17712
    },
    {
      "epoch": 5.592990211556678,
      "grad_norm": 3.1480724811553955,
      "learning_rate": 4.001947163456479e-05,
      "loss": 0.5202,
      "step": 17713
    },
    {
      "epoch": 5.593305967792864,
      "grad_norm": 3.81286358833313,
      "learning_rate": 4.001859453390871e-05,
      "loss": 0.5191,
      "step": 17714
    },
    {
      "epoch": 5.59362172402905,
      "grad_norm": 3.3665032386779785,
      "learning_rate": 4.0017717433252644e-05,
      "loss": 0.5728,
      "step": 17715
    },
    {
      "epoch": 5.593937480265235,
      "grad_norm": 4.742079734802246,
      "learning_rate": 4.001684033259657e-05,
      "loss": 0.7063,
      "step": 17716
    },
    {
      "epoch": 5.594253236501421,
      "grad_norm": 4.699954986572266,
      "learning_rate": 4.00159632319405e-05,
      "loss": 0.6915,
      "step": 17717
    },
    {
      "epoch": 5.5945689927376065,
      "grad_norm": 3.7282214164733887,
      "learning_rate": 4.001508613128443e-05,
      "loss": 0.6208,
      "step": 17718
    },
    {
      "epoch": 5.594884748973792,
      "grad_norm": 4.878490924835205,
      "learning_rate": 4.001420903062836e-05,
      "loss": 0.6467,
      "step": 17719
    },
    {
      "epoch": 5.595200505209978,
      "grad_norm": 4.047711372375488,
      "learning_rate": 4.001333192997229e-05,
      "loss": 0.5513,
      "step": 17720
    },
    {
      "epoch": 5.595516261446163,
      "grad_norm": 6.2077202796936035,
      "learning_rate": 4.0012454829316214e-05,
      "loss": 0.6931,
      "step": 17721
    },
    {
      "epoch": 5.5958320176823495,
      "grad_norm": 4.584698677062988,
      "learning_rate": 4.0011577728660146e-05,
      "loss": 0.7636,
      "step": 17722
    },
    {
      "epoch": 5.596147773918535,
      "grad_norm": 4.621293544769287,
      "learning_rate": 4.001070062800407e-05,
      "loss": 0.5934,
      "step": 17723
    },
    {
      "epoch": 5.59646353015472,
      "grad_norm": 4.464272499084473,
      "learning_rate": 4.0009823527348e-05,
      "loss": 0.6158,
      "step": 17724
    },
    {
      "epoch": 5.596779286390906,
      "grad_norm": 4.443140029907227,
      "learning_rate": 4.000894642669193e-05,
      "loss": 0.7235,
      "step": 17725
    },
    {
      "epoch": 5.597095042627092,
      "grad_norm": 4.84433126449585,
      "learning_rate": 4.000806932603586e-05,
      "loss": 0.7134,
      "step": 17726
    },
    {
      "epoch": 5.597410798863278,
      "grad_norm": 4.398155212402344,
      "learning_rate": 4.0007192225379784e-05,
      "loss": 0.5799,
      "step": 17727
    },
    {
      "epoch": 5.597726555099463,
      "grad_norm": 3.508831262588501,
      "learning_rate": 4.0006315124723715e-05,
      "loss": 0.6182,
      "step": 17728
    },
    {
      "epoch": 5.598042311335649,
      "grad_norm": 3.9558510780334473,
      "learning_rate": 4.000543802406764e-05,
      "loss": 0.7517,
      "step": 17729
    },
    {
      "epoch": 5.5983580675718345,
      "grad_norm": 4.68535852432251,
      "learning_rate": 4.000456092341157e-05,
      "loss": 0.601,
      "step": 17730
    },
    {
      "epoch": 5.598673823808021,
      "grad_norm": 3.7114663124084473,
      "learning_rate": 4.00036838227555e-05,
      "loss": 0.6463,
      "step": 17731
    },
    {
      "epoch": 5.598989580044206,
      "grad_norm": 4.118112564086914,
      "learning_rate": 4.000280672209943e-05,
      "loss": 0.6111,
      "step": 17732
    },
    {
      "epoch": 5.599305336280391,
      "grad_norm": 3.7685396671295166,
      "learning_rate": 4.000192962144336e-05,
      "loss": 0.692,
      "step": 17733
    },
    {
      "epoch": 5.5996210925165775,
      "grad_norm": 4.607899188995361,
      "learning_rate": 4.0001052520787285e-05,
      "loss": 0.7051,
      "step": 17734
    },
    {
      "epoch": 5.599936848752763,
      "grad_norm": 4.779252052307129,
      "learning_rate": 4.000017542013122e-05,
      "loss": 0.6326,
      "step": 17735
    },
    {
      "epoch": 5.600252604988949,
      "grad_norm": 4.0758256912231445,
      "learning_rate": 3.999929831947514e-05,
      "loss": 0.62,
      "step": 17736
    },
    {
      "epoch": 5.600568361225134,
      "grad_norm": 3.31093430519104,
      "learning_rate": 3.9998421218819074e-05,
      "loss": 0.5958,
      "step": 17737
    },
    {
      "epoch": 5.6008841174613195,
      "grad_norm": 4.898586750030518,
      "learning_rate": 3.9997544118163005e-05,
      "loss": 0.6883,
      "step": 17738
    },
    {
      "epoch": 5.601199873697506,
      "grad_norm": 4.044790267944336,
      "learning_rate": 3.999666701750694e-05,
      "loss": 0.6532,
      "step": 17739
    },
    {
      "epoch": 5.601515629933691,
      "grad_norm": 4.163332462310791,
      "learning_rate": 3.999578991685086e-05,
      "loss": 0.5997,
      "step": 17740
    },
    {
      "epoch": 5.601831386169877,
      "grad_norm": 3.5056285858154297,
      "learning_rate": 3.9994912816194793e-05,
      "loss": 0.6501,
      "step": 17741
    },
    {
      "epoch": 5.6021471424060625,
      "grad_norm": 3.551973581314087,
      "learning_rate": 3.999403571553872e-05,
      "loss": 0.5806,
      "step": 17742
    },
    {
      "epoch": 5.602462898642248,
      "grad_norm": 3.3258492946624756,
      "learning_rate": 3.999315861488265e-05,
      "loss": 0.5438,
      "step": 17743
    },
    {
      "epoch": 5.602778654878434,
      "grad_norm": 3.8815810680389404,
      "learning_rate": 3.9992281514226575e-05,
      "loss": 0.5754,
      "step": 17744
    },
    {
      "epoch": 5.603094411114619,
      "grad_norm": 4.546191215515137,
      "learning_rate": 3.999140441357051e-05,
      "loss": 0.627,
      "step": 17745
    },
    {
      "epoch": 5.603410167350805,
      "grad_norm": 4.495301723480225,
      "learning_rate": 3.999052731291443e-05,
      "loss": 0.6442,
      "step": 17746
    },
    {
      "epoch": 5.603725923586991,
      "grad_norm": 4.705218315124512,
      "learning_rate": 3.998965021225836e-05,
      "loss": 0.5347,
      "step": 17747
    },
    {
      "epoch": 5.604041679823176,
      "grad_norm": 4.563117027282715,
      "learning_rate": 3.998877311160229e-05,
      "loss": 0.6122,
      "step": 17748
    },
    {
      "epoch": 5.604357436059362,
      "grad_norm": 5.20759391784668,
      "learning_rate": 3.998789601094622e-05,
      "loss": 0.6976,
      "step": 17749
    },
    {
      "epoch": 5.6046731922955475,
      "grad_norm": 3.8738715648651123,
      "learning_rate": 3.9987018910290145e-05,
      "loss": 0.7227,
      "step": 17750
    },
    {
      "epoch": 5.604988948531734,
      "grad_norm": 4.765347003936768,
      "learning_rate": 3.9986141809634076e-05,
      "loss": 0.642,
      "step": 17751
    },
    {
      "epoch": 5.605304704767919,
      "grad_norm": 3.7019102573394775,
      "learning_rate": 3.9985264708978e-05,
      "loss": 0.6431,
      "step": 17752
    },
    {
      "epoch": 5.605620461004105,
      "grad_norm": 4.449183940887451,
      "learning_rate": 3.998438760832193e-05,
      "loss": 0.6203,
      "step": 17753
    },
    {
      "epoch": 5.6059362172402905,
      "grad_norm": 4.974225997924805,
      "learning_rate": 3.998351050766586e-05,
      "loss": 0.593,
      "step": 17754
    },
    {
      "epoch": 5.606251973476477,
      "grad_norm": 4.246240615844727,
      "learning_rate": 3.998263340700979e-05,
      "loss": 0.682,
      "step": 17755
    },
    {
      "epoch": 5.606567729712662,
      "grad_norm": 3.387923240661621,
      "learning_rate": 3.998175630635372e-05,
      "loss": 0.7132,
      "step": 17756
    },
    {
      "epoch": 5.606883485948847,
      "grad_norm": 4.891824245452881,
      "learning_rate": 3.9980879205697646e-05,
      "loss": 0.6486,
      "step": 17757
    },
    {
      "epoch": 5.607199242185033,
      "grad_norm": 4.75840950012207,
      "learning_rate": 3.998000210504158e-05,
      "loss": 0.5604,
      "step": 17758
    },
    {
      "epoch": 5.607514998421219,
      "grad_norm": 4.045520305633545,
      "learning_rate": 3.99791250043855e-05,
      "loss": 0.59,
      "step": 17759
    },
    {
      "epoch": 5.607830754657405,
      "grad_norm": 3.894186019897461,
      "learning_rate": 3.9978247903729434e-05,
      "loss": 0.6298,
      "step": 17760
    },
    {
      "epoch": 5.60814651089359,
      "grad_norm": 4.285950660705566,
      "learning_rate": 3.997737080307336e-05,
      "loss": 0.6779,
      "step": 17761
    },
    {
      "epoch": 5.6084622671297755,
      "grad_norm": 4.679220199584961,
      "learning_rate": 3.997649370241729e-05,
      "loss": 0.6473,
      "step": 17762
    },
    {
      "epoch": 5.608778023365962,
      "grad_norm": 3.906358480453491,
      "learning_rate": 3.9975616601761216e-05,
      "loss": 0.6575,
      "step": 17763
    },
    {
      "epoch": 5.609093779602147,
      "grad_norm": 3.6396279335021973,
      "learning_rate": 3.997473950110515e-05,
      "loss": 0.5865,
      "step": 17764
    },
    {
      "epoch": 5.609409535838333,
      "grad_norm": 3.876821756362915,
      "learning_rate": 3.997386240044907e-05,
      "loss": 0.6645,
      "step": 17765
    },
    {
      "epoch": 5.609725292074518,
      "grad_norm": 3.759852886199951,
      "learning_rate": 3.997298529979301e-05,
      "loss": 0.6587,
      "step": 17766
    },
    {
      "epoch": 5.610041048310704,
      "grad_norm": 4.1666259765625,
      "learning_rate": 3.9972108199136936e-05,
      "loss": 0.6158,
      "step": 17767
    },
    {
      "epoch": 5.61035680454689,
      "grad_norm": 4.6028923988342285,
      "learning_rate": 3.997123109848087e-05,
      "loss": 0.6871,
      "step": 17768
    },
    {
      "epoch": 5.610672560783075,
      "grad_norm": 3.4398117065429688,
      "learning_rate": 3.997035399782479e-05,
      "loss": 0.5959,
      "step": 17769
    },
    {
      "epoch": 5.610988317019261,
      "grad_norm": 3.819021463394165,
      "learning_rate": 3.9969476897168724e-05,
      "loss": 0.7785,
      "step": 17770
    },
    {
      "epoch": 5.611304073255447,
      "grad_norm": 4.146185398101807,
      "learning_rate": 3.996859979651265e-05,
      "loss": 0.5926,
      "step": 17771
    },
    {
      "epoch": 5.611619829491632,
      "grad_norm": 4.225814342498779,
      "learning_rate": 3.996772269585658e-05,
      "loss": 0.6267,
      "step": 17772
    },
    {
      "epoch": 5.611935585727818,
      "grad_norm": 4.231015205383301,
      "learning_rate": 3.9966845595200506e-05,
      "loss": 0.7643,
      "step": 17773
    },
    {
      "epoch": 5.612251341964003,
      "grad_norm": 3.9202497005462646,
      "learning_rate": 3.996596849454444e-05,
      "loss": 0.7019,
      "step": 17774
    },
    {
      "epoch": 5.61256709820019,
      "grad_norm": 4.498490810394287,
      "learning_rate": 3.996509139388836e-05,
      "loss": 0.6896,
      "step": 17775
    },
    {
      "epoch": 5.612882854436375,
      "grad_norm": 3.8748090267181396,
      "learning_rate": 3.9964214293232294e-05,
      "loss": 0.606,
      "step": 17776
    },
    {
      "epoch": 5.613198610672561,
      "grad_norm": 4.6978654861450195,
      "learning_rate": 3.9963337192576226e-05,
      "loss": 0.6638,
      "step": 17777
    },
    {
      "epoch": 5.613514366908746,
      "grad_norm": 3.840816020965576,
      "learning_rate": 3.996246009192015e-05,
      "loss": 0.6312,
      "step": 17778
    },
    {
      "epoch": 5.613830123144933,
      "grad_norm": 4.108292102813721,
      "learning_rate": 3.996158299126408e-05,
      "loss": 0.5922,
      "step": 17779
    },
    {
      "epoch": 5.614145879381118,
      "grad_norm": 4.82448148727417,
      "learning_rate": 3.996070589060801e-05,
      "loss": 0.5863,
      "step": 17780
    },
    {
      "epoch": 5.614461635617303,
      "grad_norm": 4.419853210449219,
      "learning_rate": 3.995982878995194e-05,
      "loss": 0.6257,
      "step": 17781
    },
    {
      "epoch": 5.614777391853489,
      "grad_norm": 3.76102876663208,
      "learning_rate": 3.9958951689295864e-05,
      "loss": 0.5884,
      "step": 17782
    },
    {
      "epoch": 5.615093148089675,
      "grad_norm": 3.836240291595459,
      "learning_rate": 3.9958074588639795e-05,
      "loss": 0.5854,
      "step": 17783
    },
    {
      "epoch": 5.615408904325861,
      "grad_norm": 4.1834330558776855,
      "learning_rate": 3.995719748798372e-05,
      "loss": 0.7279,
      "step": 17784
    },
    {
      "epoch": 5.615724660562046,
      "grad_norm": 4.850983142852783,
      "learning_rate": 3.995632038732765e-05,
      "loss": 0.5539,
      "step": 17785
    },
    {
      "epoch": 5.616040416798231,
      "grad_norm": 4.70145320892334,
      "learning_rate": 3.995544328667158e-05,
      "loss": 0.6677,
      "step": 17786
    },
    {
      "epoch": 5.616356173034418,
      "grad_norm": 4.461780548095703,
      "learning_rate": 3.995456618601551e-05,
      "loss": 0.6341,
      "step": 17787
    },
    {
      "epoch": 5.616671929270603,
      "grad_norm": 5.586582660675049,
      "learning_rate": 3.9953689085359433e-05,
      "loss": 0.7378,
      "step": 17788
    },
    {
      "epoch": 5.616987685506789,
      "grad_norm": 4.306715488433838,
      "learning_rate": 3.9952811984703365e-05,
      "loss": 0.6125,
      "step": 17789
    },
    {
      "epoch": 5.617303441742974,
      "grad_norm": 4.433288097381592,
      "learning_rate": 3.995193488404729e-05,
      "loss": 0.6118,
      "step": 17790
    },
    {
      "epoch": 5.61761919797916,
      "grad_norm": 3.9868175983428955,
      "learning_rate": 3.995105778339122e-05,
      "loss": 0.6123,
      "step": 17791
    },
    {
      "epoch": 5.617934954215346,
      "grad_norm": 3.8358232975006104,
      "learning_rate": 3.995018068273515e-05,
      "loss": 0.6078,
      "step": 17792
    },
    {
      "epoch": 5.618250710451531,
      "grad_norm": 4.031958103179932,
      "learning_rate": 3.9949303582079085e-05,
      "loss": 0.6702,
      "step": 17793
    },
    {
      "epoch": 5.618566466687717,
      "grad_norm": 3.5762107372283936,
      "learning_rate": 3.994842648142301e-05,
      "loss": 0.6144,
      "step": 17794
    },
    {
      "epoch": 5.618882222923903,
      "grad_norm": 3.9471242427825928,
      "learning_rate": 3.994754938076694e-05,
      "loss": 0.6066,
      "step": 17795
    },
    {
      "epoch": 5.619197979160089,
      "grad_norm": 3.5922112464904785,
      "learning_rate": 3.9946672280110867e-05,
      "loss": 0.6546,
      "step": 17796
    },
    {
      "epoch": 5.619513735396274,
      "grad_norm": 4.827828884124756,
      "learning_rate": 3.99457951794548e-05,
      "loss": 0.6849,
      "step": 17797
    },
    {
      "epoch": 5.619829491632459,
      "grad_norm": 3.544612407684326,
      "learning_rate": 3.994491807879872e-05,
      "loss": 0.6288,
      "step": 17798
    },
    {
      "epoch": 5.6201452478686456,
      "grad_norm": 4.6968841552734375,
      "learning_rate": 3.9944040978142655e-05,
      "loss": 0.778,
      "step": 17799
    },
    {
      "epoch": 5.620461004104831,
      "grad_norm": 3.298231601715088,
      "learning_rate": 3.9943163877486587e-05,
      "loss": 0.4938,
      "step": 17800
    },
    {
      "epoch": 5.620461004104831,
      "eval_loss": 0.8310948610305786,
      "eval_runtime": 4.5414,
      "eval_samples_per_second": 548.287,
      "eval_steps_per_second": 4.624,
      "step": 17800
    },
    {
      "epoch": 5.620776760341017,
      "grad_norm": 9.230361938476562,
      "learning_rate": 3.994228677683051e-05,
      "loss": 0.7562,
      "step": 17801
    },
    {
      "epoch": 5.621092516577202,
      "grad_norm": 5.157171726226807,
      "learning_rate": 3.994140967617444e-05,
      "loss": 0.6395,
      "step": 17802
    },
    {
      "epoch": 5.6214082728133885,
      "grad_norm": 3.2725112438201904,
      "learning_rate": 3.994053257551837e-05,
      "loss": 0.6218,
      "step": 17803
    },
    {
      "epoch": 5.621724029049574,
      "grad_norm": 4.777364730834961,
      "learning_rate": 3.99396554748623e-05,
      "loss": 0.687,
      "step": 17804
    },
    {
      "epoch": 5.622039785285759,
      "grad_norm": 3.968876361846924,
      "learning_rate": 3.9938778374206225e-05,
      "loss": 0.6052,
      "step": 17805
    },
    {
      "epoch": 5.622355541521945,
      "grad_norm": 3.6977882385253906,
      "learning_rate": 3.9937901273550156e-05,
      "loss": 0.5521,
      "step": 17806
    },
    {
      "epoch": 5.622671297758131,
      "grad_norm": 3.6665773391723633,
      "learning_rate": 3.993702417289408e-05,
      "loss": 0.6918,
      "step": 17807
    },
    {
      "epoch": 5.622987053994317,
      "grad_norm": 3.7529478073120117,
      "learning_rate": 3.993614707223801e-05,
      "loss": 0.5061,
      "step": 17808
    },
    {
      "epoch": 5.623302810230502,
      "grad_norm": 4.812376976013184,
      "learning_rate": 3.993526997158194e-05,
      "loss": 0.7424,
      "step": 17809
    },
    {
      "epoch": 5.623618566466687,
      "grad_norm": 4.117478370666504,
      "learning_rate": 3.993439287092587e-05,
      "loss": 0.6174,
      "step": 17810
    },
    {
      "epoch": 5.6239343227028735,
      "grad_norm": 3.7535476684570312,
      "learning_rate": 3.9933515770269794e-05,
      "loss": 0.6267,
      "step": 17811
    },
    {
      "epoch": 5.624250078939059,
      "grad_norm": 3.7473583221435547,
      "learning_rate": 3.9932638669613726e-05,
      "loss": 0.5893,
      "step": 17812
    },
    {
      "epoch": 5.624565835175245,
      "grad_norm": 5.074533462524414,
      "learning_rate": 3.993176156895765e-05,
      "loss": 0.5919,
      "step": 17813
    },
    {
      "epoch": 5.62488159141143,
      "grad_norm": 4.231415271759033,
      "learning_rate": 3.993088446830158e-05,
      "loss": 0.701,
      "step": 17814
    },
    {
      "epoch": 5.625197347647616,
      "grad_norm": 4.509829521179199,
      "learning_rate": 3.9930007367645514e-05,
      "loss": 0.6562,
      "step": 17815
    },
    {
      "epoch": 5.625513103883802,
      "grad_norm": 4.698781967163086,
      "learning_rate": 3.992913026698944e-05,
      "loss": 0.7656,
      "step": 17816
    },
    {
      "epoch": 5.625828860119987,
      "grad_norm": 3.656109571456909,
      "learning_rate": 3.992825316633337e-05,
      "loss": 0.6647,
      "step": 17817
    },
    {
      "epoch": 5.626144616356173,
      "grad_norm": 4.5152106285095215,
      "learning_rate": 3.9927376065677296e-05,
      "loss": 0.8122,
      "step": 17818
    },
    {
      "epoch": 5.6264603725923585,
      "grad_norm": 4.474974155426025,
      "learning_rate": 3.992649896502123e-05,
      "loss": 0.5677,
      "step": 17819
    },
    {
      "epoch": 5.626776128828545,
      "grad_norm": 4.026129722595215,
      "learning_rate": 3.992562186436516e-05,
      "loss": 0.706,
      "step": 17820
    },
    {
      "epoch": 5.62709188506473,
      "grad_norm": 4.403326511383057,
      "learning_rate": 3.992474476370909e-05,
      "loss": 0.6924,
      "step": 17821
    },
    {
      "epoch": 5.627407641300915,
      "grad_norm": 3.563333034515381,
      "learning_rate": 3.9923867663053016e-05,
      "loss": 0.6409,
      "step": 17822
    },
    {
      "epoch": 5.6277233975371015,
      "grad_norm": 4.147365093231201,
      "learning_rate": 3.992299056239695e-05,
      "loss": 0.6057,
      "step": 17823
    },
    {
      "epoch": 5.628039153773287,
      "grad_norm": 4.032285213470459,
      "learning_rate": 3.992211346174087e-05,
      "loss": 0.5828,
      "step": 17824
    },
    {
      "epoch": 5.628354910009473,
      "grad_norm": 3.842393398284912,
      "learning_rate": 3.9921236361084804e-05,
      "loss": 0.7693,
      "step": 17825
    },
    {
      "epoch": 5.628670666245658,
      "grad_norm": 3.640045166015625,
      "learning_rate": 3.992035926042873e-05,
      "loss": 0.6741,
      "step": 17826
    },
    {
      "epoch": 5.628986422481844,
      "grad_norm": 4.37045431137085,
      "learning_rate": 3.991948215977266e-05,
      "loss": 0.7671,
      "step": 17827
    },
    {
      "epoch": 5.62930217871803,
      "grad_norm": 3.677345037460327,
      "learning_rate": 3.9918605059116586e-05,
      "loss": 0.6764,
      "step": 17828
    },
    {
      "epoch": 5.629617934954215,
      "grad_norm": 3.5352718830108643,
      "learning_rate": 3.991772795846052e-05,
      "loss": 0.6891,
      "step": 17829
    },
    {
      "epoch": 5.629933691190401,
      "grad_norm": 3.1257803440093994,
      "learning_rate": 3.991685085780444e-05,
      "loss": 0.5223,
      "step": 17830
    },
    {
      "epoch": 5.6302494474265865,
      "grad_norm": 4.791553020477295,
      "learning_rate": 3.9915973757148374e-05,
      "loss": 0.7183,
      "step": 17831
    },
    {
      "epoch": 5.630565203662773,
      "grad_norm": 3.8066117763519287,
      "learning_rate": 3.99150966564923e-05,
      "loss": 0.6073,
      "step": 17832
    },
    {
      "epoch": 5.630880959898958,
      "grad_norm": 3.5837290287017822,
      "learning_rate": 3.991421955583623e-05,
      "loss": 0.6365,
      "step": 17833
    },
    {
      "epoch": 5.631196716135143,
      "grad_norm": 3.6366865634918213,
      "learning_rate": 3.9913342455180155e-05,
      "loss": 0.6366,
      "step": 17834
    },
    {
      "epoch": 5.6315124723713295,
      "grad_norm": 4.540703773498535,
      "learning_rate": 3.991246535452409e-05,
      "loss": 0.637,
      "step": 17835
    },
    {
      "epoch": 5.631828228607515,
      "grad_norm": 3.4671380519866943,
      "learning_rate": 3.991158825386801e-05,
      "loss": 0.6383,
      "step": 17836
    },
    {
      "epoch": 5.632143984843701,
      "grad_norm": 4.010313987731934,
      "learning_rate": 3.9910711153211944e-05,
      "loss": 0.6843,
      "step": 17837
    },
    {
      "epoch": 5.632459741079886,
      "grad_norm": 4.679734230041504,
      "learning_rate": 3.9909834052555875e-05,
      "loss": 0.7309,
      "step": 17838
    },
    {
      "epoch": 5.6327754973160715,
      "grad_norm": 3.7427375316619873,
      "learning_rate": 3.99089569518998e-05,
      "loss": 0.582,
      "step": 17839
    },
    {
      "epoch": 5.633091253552258,
      "grad_norm": 4.206387996673584,
      "learning_rate": 3.990807985124373e-05,
      "loss": 0.6048,
      "step": 17840
    },
    {
      "epoch": 5.633407009788443,
      "grad_norm": 4.907917499542236,
      "learning_rate": 3.990720275058766e-05,
      "loss": 0.6867,
      "step": 17841
    },
    {
      "epoch": 5.633722766024629,
      "grad_norm": 3.5915050506591797,
      "learning_rate": 3.990632564993159e-05,
      "loss": 0.7829,
      "step": 17842
    },
    {
      "epoch": 5.6340385222608145,
      "grad_norm": 3.6049537658691406,
      "learning_rate": 3.990544854927551e-05,
      "loss": 0.6064,
      "step": 17843
    },
    {
      "epoch": 5.634354278497001,
      "grad_norm": 3.5395424365997314,
      "learning_rate": 3.9904571448619445e-05,
      "loss": 0.6892,
      "step": 17844
    },
    {
      "epoch": 5.634670034733186,
      "grad_norm": 3.662487268447876,
      "learning_rate": 3.990369434796337e-05,
      "loss": 0.6451,
      "step": 17845
    },
    {
      "epoch": 5.634985790969372,
      "grad_norm": 4.432030200958252,
      "learning_rate": 3.99028172473073e-05,
      "loss": 0.6553,
      "step": 17846
    },
    {
      "epoch": 5.635301547205557,
      "grad_norm": 4.086714267730713,
      "learning_rate": 3.990194014665123e-05,
      "loss": 0.667,
      "step": 17847
    },
    {
      "epoch": 5.635617303441743,
      "grad_norm": 4.1997785568237305,
      "learning_rate": 3.9901063045995165e-05,
      "loss": 0.6121,
      "step": 17848
    },
    {
      "epoch": 5.635933059677929,
      "grad_norm": 4.481095790863037,
      "learning_rate": 3.990018594533909e-05,
      "loss": 0.6275,
      "step": 17849
    },
    {
      "epoch": 5.636248815914114,
      "grad_norm": 4.0623064041137695,
      "learning_rate": 3.989930884468302e-05,
      "loss": 0.6425,
      "step": 17850
    },
    {
      "epoch": 5.6365645721503,
      "grad_norm": 3.9355084896087646,
      "learning_rate": 3.9898431744026946e-05,
      "loss": 0.6288,
      "step": 17851
    },
    {
      "epoch": 5.636880328386486,
      "grad_norm": 4.390651226043701,
      "learning_rate": 3.989755464337088e-05,
      "loss": 0.7061,
      "step": 17852
    },
    {
      "epoch": 5.637196084622671,
      "grad_norm": 4.516531467437744,
      "learning_rate": 3.98966775427148e-05,
      "loss": 0.6565,
      "step": 17853
    },
    {
      "epoch": 5.637511840858857,
      "grad_norm": 4.392105579376221,
      "learning_rate": 3.9895800442058735e-05,
      "loss": 0.7048,
      "step": 17854
    },
    {
      "epoch": 5.637827597095042,
      "grad_norm": 4.490346908569336,
      "learning_rate": 3.989492334140266e-05,
      "loss": 0.7116,
      "step": 17855
    },
    {
      "epoch": 5.638143353331229,
      "grad_norm": 4.544900894165039,
      "learning_rate": 3.989404624074659e-05,
      "loss": 0.6973,
      "step": 17856
    },
    {
      "epoch": 5.638459109567414,
      "grad_norm": 3.8526201248168945,
      "learning_rate": 3.9893169140090516e-05,
      "loss": 0.7253,
      "step": 17857
    },
    {
      "epoch": 5.638774865803599,
      "grad_norm": 4.710166931152344,
      "learning_rate": 3.989229203943445e-05,
      "loss": 0.5849,
      "step": 17858
    },
    {
      "epoch": 5.639090622039785,
      "grad_norm": 5.306016445159912,
      "learning_rate": 3.989141493877838e-05,
      "loss": 0.6746,
      "step": 17859
    },
    {
      "epoch": 5.639406378275971,
      "grad_norm": 4.714123249053955,
      "learning_rate": 3.9890537838122305e-05,
      "loss": 0.7181,
      "step": 17860
    },
    {
      "epoch": 5.639722134512157,
      "grad_norm": 4.673798084259033,
      "learning_rate": 3.9889660737466236e-05,
      "loss": 0.6116,
      "step": 17861
    },
    {
      "epoch": 5.640037890748342,
      "grad_norm": 3.2932658195495605,
      "learning_rate": 3.988878363681016e-05,
      "loss": 0.5661,
      "step": 17862
    },
    {
      "epoch": 5.6403536469845275,
      "grad_norm": 3.714693546295166,
      "learning_rate": 3.988790653615409e-05,
      "loss": 0.6412,
      "step": 17863
    },
    {
      "epoch": 5.640669403220714,
      "grad_norm": 3.6151199340820312,
      "learning_rate": 3.988702943549802e-05,
      "loss": 0.6234,
      "step": 17864
    },
    {
      "epoch": 5.640985159456899,
      "grad_norm": 3.7845077514648438,
      "learning_rate": 3.988615233484195e-05,
      "loss": 0.608,
      "step": 17865
    },
    {
      "epoch": 5.641300915693085,
      "grad_norm": 3.3371994495391846,
      "learning_rate": 3.9885275234185874e-05,
      "loss": 0.5979,
      "step": 17866
    },
    {
      "epoch": 5.64161667192927,
      "grad_norm": 4.866530895233154,
      "learning_rate": 3.9884398133529806e-05,
      "loss": 0.5971,
      "step": 17867
    },
    {
      "epoch": 5.641932428165457,
      "grad_norm": 3.9624364376068115,
      "learning_rate": 3.988352103287373e-05,
      "loss": 0.6324,
      "step": 17868
    },
    {
      "epoch": 5.642248184401642,
      "grad_norm": 4.018068790435791,
      "learning_rate": 3.988264393221766e-05,
      "loss": 0.5685,
      "step": 17869
    },
    {
      "epoch": 5.642563940637828,
      "grad_norm": 4.152504920959473,
      "learning_rate": 3.988176683156159e-05,
      "loss": 0.704,
      "step": 17870
    },
    {
      "epoch": 5.642879696874013,
      "grad_norm": 4.041078090667725,
      "learning_rate": 3.988088973090552e-05,
      "loss": 0.6203,
      "step": 17871
    },
    {
      "epoch": 5.643195453110199,
      "grad_norm": 4.051408290863037,
      "learning_rate": 3.9880012630249444e-05,
      "loss": 0.6705,
      "step": 17872
    },
    {
      "epoch": 5.643511209346385,
      "grad_norm": 4.161071300506592,
      "learning_rate": 3.9879135529593376e-05,
      "loss": 0.669,
      "step": 17873
    },
    {
      "epoch": 5.64382696558257,
      "grad_norm": 3.7048513889312744,
      "learning_rate": 3.98782584289373e-05,
      "loss": 0.5787,
      "step": 17874
    },
    {
      "epoch": 5.644142721818756,
      "grad_norm": 3.6697616577148438,
      "learning_rate": 3.987738132828124e-05,
      "loss": 0.613,
      "step": 17875
    },
    {
      "epoch": 5.644458478054942,
      "grad_norm": 3.702786684036255,
      "learning_rate": 3.9876504227625164e-05,
      "loss": 0.6533,
      "step": 17876
    },
    {
      "epoch": 5.644774234291127,
      "grad_norm": 4.463025093078613,
      "learning_rate": 3.9875627126969096e-05,
      "loss": 0.6269,
      "step": 17877
    },
    {
      "epoch": 5.645089990527313,
      "grad_norm": 4.650588512420654,
      "learning_rate": 3.987475002631302e-05,
      "loss": 0.693,
      "step": 17878
    },
    {
      "epoch": 5.645405746763498,
      "grad_norm": 4.490218162536621,
      "learning_rate": 3.987387292565695e-05,
      "loss": 0.6203,
      "step": 17879
    },
    {
      "epoch": 5.645721502999685,
      "grad_norm": 4.404046058654785,
      "learning_rate": 3.9872995825000884e-05,
      "loss": 0.6741,
      "step": 17880
    },
    {
      "epoch": 5.64603725923587,
      "grad_norm": 4.8086161613464355,
      "learning_rate": 3.987211872434481e-05,
      "loss": 0.7343,
      "step": 17881
    },
    {
      "epoch": 5.646353015472055,
      "grad_norm": 4.582309722900391,
      "learning_rate": 3.987124162368874e-05,
      "loss": 0.6629,
      "step": 17882
    },
    {
      "epoch": 5.646668771708241,
      "grad_norm": 3.9993720054626465,
      "learning_rate": 3.9870364523032665e-05,
      "loss": 0.5344,
      "step": 17883
    },
    {
      "epoch": 5.646984527944427,
      "grad_norm": 4.380081653594971,
      "learning_rate": 3.98694874223766e-05,
      "loss": 0.6387,
      "step": 17884
    },
    {
      "epoch": 5.647300284180613,
      "grad_norm": 3.6826906204223633,
      "learning_rate": 3.986861032172052e-05,
      "loss": 0.5963,
      "step": 17885
    },
    {
      "epoch": 5.647616040416798,
      "grad_norm": 4.329923152923584,
      "learning_rate": 3.9867733221064454e-05,
      "loss": 0.5527,
      "step": 17886
    },
    {
      "epoch": 5.647931796652984,
      "grad_norm": 5.614543914794922,
      "learning_rate": 3.986685612040838e-05,
      "loss": 0.6357,
      "step": 17887
    },
    {
      "epoch": 5.64824755288917,
      "grad_norm": 4.144800186157227,
      "learning_rate": 3.986597901975231e-05,
      "loss": 0.5846,
      "step": 17888
    },
    {
      "epoch": 5.648563309125355,
      "grad_norm": 4.676774501800537,
      "learning_rate": 3.9865101919096235e-05,
      "loss": 0.6132,
      "step": 17889
    },
    {
      "epoch": 5.648879065361541,
      "grad_norm": 4.244927406311035,
      "learning_rate": 3.986422481844017e-05,
      "loss": 0.6866,
      "step": 17890
    },
    {
      "epoch": 5.649194821597726,
      "grad_norm": 4.226974964141846,
      "learning_rate": 3.986334771778409e-05,
      "loss": 0.6046,
      "step": 17891
    },
    {
      "epoch": 5.6495105778339125,
      "grad_norm": 4.031556129455566,
      "learning_rate": 3.9862470617128023e-05,
      "loss": 0.623,
      "step": 17892
    },
    {
      "epoch": 5.649826334070098,
      "grad_norm": 3.7346198558807373,
      "learning_rate": 3.986159351647195e-05,
      "loss": 0.6316,
      "step": 17893
    },
    {
      "epoch": 5.650142090306284,
      "grad_norm": 4.051063060760498,
      "learning_rate": 3.986071641581588e-05,
      "loss": 0.7304,
      "step": 17894
    },
    {
      "epoch": 5.650457846542469,
      "grad_norm": 3.976651191711426,
      "learning_rate": 3.9859839315159805e-05,
      "loss": 0.6041,
      "step": 17895
    },
    {
      "epoch": 5.650773602778655,
      "grad_norm": 3.7063491344451904,
      "learning_rate": 3.985896221450374e-05,
      "loss": 0.657,
      "step": 17896
    },
    {
      "epoch": 5.651089359014841,
      "grad_norm": 3.8851563930511475,
      "learning_rate": 3.985808511384767e-05,
      "loss": 0.6788,
      "step": 17897
    },
    {
      "epoch": 5.651405115251026,
      "grad_norm": 4.5664567947387695,
      "learning_rate": 3.985720801319159e-05,
      "loss": 0.6149,
      "step": 17898
    },
    {
      "epoch": 5.651720871487212,
      "grad_norm": 3.8499772548675537,
      "learning_rate": 3.9856330912535525e-05,
      "loss": 0.6041,
      "step": 17899
    },
    {
      "epoch": 5.6520366277233975,
      "grad_norm": 3.7922720909118652,
      "learning_rate": 3.985545381187945e-05,
      "loss": 0.6067,
      "step": 17900
    },
    {
      "epoch": 5.652352383959583,
      "grad_norm": 4.030829906463623,
      "learning_rate": 3.985457671122338e-05,
      "loss": 0.7127,
      "step": 17901
    },
    {
      "epoch": 5.652668140195769,
      "grad_norm": 3.4281249046325684,
      "learning_rate": 3.985369961056731e-05,
      "loss": 0.5923,
      "step": 17902
    },
    {
      "epoch": 5.652983896431954,
      "grad_norm": 5.022688388824463,
      "learning_rate": 3.9852822509911245e-05,
      "loss": 0.6055,
      "step": 17903
    },
    {
      "epoch": 5.6532996526681405,
      "grad_norm": 4.164673328399658,
      "learning_rate": 3.985194540925517e-05,
      "loss": 0.6468,
      "step": 17904
    },
    {
      "epoch": 5.653615408904326,
      "grad_norm": 4.79453706741333,
      "learning_rate": 3.98510683085991e-05,
      "loss": 0.5843,
      "step": 17905
    },
    {
      "epoch": 5.653931165140511,
      "grad_norm": 4.744379043579102,
      "learning_rate": 3.9850191207943026e-05,
      "loss": 0.6488,
      "step": 17906
    },
    {
      "epoch": 5.654246921376697,
      "grad_norm": 3.8345000743865967,
      "learning_rate": 3.984931410728696e-05,
      "loss": 0.617,
      "step": 17907
    },
    {
      "epoch": 5.654562677612883,
      "grad_norm": 4.158076286315918,
      "learning_rate": 3.984843700663088e-05,
      "loss": 0.6851,
      "step": 17908
    },
    {
      "epoch": 5.654878433849069,
      "grad_norm": 3.343297004699707,
      "learning_rate": 3.9847559905974815e-05,
      "loss": 0.5857,
      "step": 17909
    },
    {
      "epoch": 5.655194190085254,
      "grad_norm": 4.541953086853027,
      "learning_rate": 3.984668280531874e-05,
      "loss": 0.621,
      "step": 17910
    },
    {
      "epoch": 5.65550994632144,
      "grad_norm": 3.7664599418640137,
      "learning_rate": 3.984580570466267e-05,
      "loss": 0.6918,
      "step": 17911
    },
    {
      "epoch": 5.6558257025576255,
      "grad_norm": 5.011417865753174,
      "learning_rate": 3.9844928604006596e-05,
      "loss": 0.7118,
      "step": 17912
    },
    {
      "epoch": 5.656141458793811,
      "grad_norm": 3.877030372619629,
      "learning_rate": 3.984405150335053e-05,
      "loss": 0.6487,
      "step": 17913
    },
    {
      "epoch": 5.656457215029997,
      "grad_norm": 4.1248931884765625,
      "learning_rate": 3.984317440269445e-05,
      "loss": 0.5928,
      "step": 17914
    },
    {
      "epoch": 5.656772971266182,
      "grad_norm": 3.7799196243286133,
      "learning_rate": 3.9842297302038384e-05,
      "loss": 0.5889,
      "step": 17915
    },
    {
      "epoch": 5.6570887275023685,
      "grad_norm": 3.8671927452087402,
      "learning_rate": 3.984142020138231e-05,
      "loss": 0.6453,
      "step": 17916
    },
    {
      "epoch": 5.657404483738554,
      "grad_norm": 4.294140815734863,
      "learning_rate": 3.984054310072624e-05,
      "loss": 0.6626,
      "step": 17917
    },
    {
      "epoch": 5.65772023997474,
      "grad_norm": 4.053726673126221,
      "learning_rate": 3.983966600007017e-05,
      "loss": 0.6025,
      "step": 17918
    },
    {
      "epoch": 5.658035996210925,
      "grad_norm": 4.0379109382629395,
      "learning_rate": 3.98387888994141e-05,
      "loss": 0.6605,
      "step": 17919
    },
    {
      "epoch": 5.6583517524471105,
      "grad_norm": 4.260637283325195,
      "learning_rate": 3.983791179875803e-05,
      "loss": 0.5755,
      "step": 17920
    },
    {
      "epoch": 5.658667508683297,
      "grad_norm": 3.9936952590942383,
      "learning_rate": 3.9837034698101954e-05,
      "loss": 0.7214,
      "step": 17921
    },
    {
      "epoch": 5.658983264919482,
      "grad_norm": 3.5043785572052,
      "learning_rate": 3.9836157597445886e-05,
      "loss": 0.6185,
      "step": 17922
    },
    {
      "epoch": 5.659299021155668,
      "grad_norm": 3.1441309452056885,
      "learning_rate": 3.983528049678981e-05,
      "loss": 0.5545,
      "step": 17923
    },
    {
      "epoch": 5.6596147773918535,
      "grad_norm": 3.8049070835113525,
      "learning_rate": 3.983440339613374e-05,
      "loss": 0.647,
      "step": 17924
    },
    {
      "epoch": 5.659930533628039,
      "grad_norm": 5.850103378295898,
      "learning_rate": 3.983352629547767e-05,
      "loss": 0.7729,
      "step": 17925
    },
    {
      "epoch": 5.660246289864225,
      "grad_norm": 3.7213330268859863,
      "learning_rate": 3.98326491948216e-05,
      "loss": 0.5796,
      "step": 17926
    },
    {
      "epoch": 5.66056204610041,
      "grad_norm": 4.231651306152344,
      "learning_rate": 3.9831772094165524e-05,
      "loss": 0.5881,
      "step": 17927
    },
    {
      "epoch": 5.660877802336596,
      "grad_norm": 3.808898448944092,
      "learning_rate": 3.9830894993509456e-05,
      "loss": 0.6361,
      "step": 17928
    },
    {
      "epoch": 5.661193558572782,
      "grad_norm": 6.39227294921875,
      "learning_rate": 3.983001789285339e-05,
      "loss": 0.6847,
      "step": 17929
    },
    {
      "epoch": 5.661509314808967,
      "grad_norm": 5.046790599822998,
      "learning_rate": 3.982914079219732e-05,
      "loss": 0.7327,
      "step": 17930
    },
    {
      "epoch": 5.661825071045153,
      "grad_norm": 4.515983581542969,
      "learning_rate": 3.9828263691541244e-05,
      "loss": 0.703,
      "step": 17931
    },
    {
      "epoch": 5.6621408272813385,
      "grad_norm": 4.123304843902588,
      "learning_rate": 3.9827386590885176e-05,
      "loss": 0.6577,
      "step": 17932
    },
    {
      "epoch": 5.662456583517525,
      "grad_norm": 4.286546230316162,
      "learning_rate": 3.98265094902291e-05,
      "loss": 0.6817,
      "step": 17933
    },
    {
      "epoch": 5.66277233975371,
      "grad_norm": 2.913114070892334,
      "learning_rate": 3.982563238957303e-05,
      "loss": 0.5516,
      "step": 17934
    },
    {
      "epoch": 5.663088095989896,
      "grad_norm": 3.92649507522583,
      "learning_rate": 3.982475528891696e-05,
      "loss": 0.6549,
      "step": 17935
    },
    {
      "epoch": 5.6634038522260814,
      "grad_norm": 4.353005886077881,
      "learning_rate": 3.982387818826089e-05,
      "loss": 0.6674,
      "step": 17936
    },
    {
      "epoch": 5.663719608462267,
      "grad_norm": 4.271449089050293,
      "learning_rate": 3.9823001087604814e-05,
      "loss": 0.6464,
      "step": 17937
    },
    {
      "epoch": 5.664035364698453,
      "grad_norm": 4.031500339508057,
      "learning_rate": 3.9822123986948745e-05,
      "loss": 0.6496,
      "step": 17938
    },
    {
      "epoch": 5.664351120934638,
      "grad_norm": 4.437322616577148,
      "learning_rate": 3.982124688629267e-05,
      "loss": 0.5711,
      "step": 17939
    },
    {
      "epoch": 5.664666877170824,
      "grad_norm": 3.500769853591919,
      "learning_rate": 3.98203697856366e-05,
      "loss": 0.705,
      "step": 17940
    },
    {
      "epoch": 5.66498263340701,
      "grad_norm": 4.259260177612305,
      "learning_rate": 3.9819492684980534e-05,
      "loss": 0.5888,
      "step": 17941
    },
    {
      "epoch": 5.665298389643196,
      "grad_norm": 4.103301525115967,
      "learning_rate": 3.981861558432446e-05,
      "loss": 0.5991,
      "step": 17942
    },
    {
      "epoch": 5.665614145879381,
      "grad_norm": 4.447503566741943,
      "learning_rate": 3.981773848366839e-05,
      "loss": 0.6686,
      "step": 17943
    },
    {
      "epoch": 5.6659299021155665,
      "grad_norm": 5.361998081207275,
      "learning_rate": 3.9816861383012315e-05,
      "loss": 0.5832,
      "step": 17944
    },
    {
      "epoch": 5.666245658351753,
      "grad_norm": 3.567985773086548,
      "learning_rate": 3.981598428235625e-05,
      "loss": 0.5568,
      "step": 17945
    },
    {
      "epoch": 5.666561414587938,
      "grad_norm": 5.687725067138672,
      "learning_rate": 3.981510718170017e-05,
      "loss": 0.7722,
      "step": 17946
    },
    {
      "epoch": 5.666877170824124,
      "grad_norm": 3.7892298698425293,
      "learning_rate": 3.98142300810441e-05,
      "loss": 0.4877,
      "step": 17947
    },
    {
      "epoch": 5.667192927060309,
      "grad_norm": 4.278456687927246,
      "learning_rate": 3.981335298038803e-05,
      "loss": 0.7369,
      "step": 17948
    },
    {
      "epoch": 5.667508683296495,
      "grad_norm": 3.430274724960327,
      "learning_rate": 3.981247587973196e-05,
      "loss": 0.5934,
      "step": 17949
    },
    {
      "epoch": 5.667824439532681,
      "grad_norm": 3.963768243789673,
      "learning_rate": 3.9811598779075885e-05,
      "loss": 0.6027,
      "step": 17950
    },
    {
      "epoch": 5.668140195768866,
      "grad_norm": 4.006139278411865,
      "learning_rate": 3.9810721678419817e-05,
      "loss": 0.7377,
      "step": 17951
    },
    {
      "epoch": 5.668455952005052,
      "grad_norm": 3.827768325805664,
      "learning_rate": 3.980984457776374e-05,
      "loss": 0.6171,
      "step": 17952
    },
    {
      "epoch": 5.668771708241238,
      "grad_norm": 3.2213916778564453,
      "learning_rate": 3.980896747710767e-05,
      "loss": 0.6312,
      "step": 17953
    },
    {
      "epoch": 5.669087464477423,
      "grad_norm": 4.145191669464111,
      "learning_rate": 3.98080903764516e-05,
      "loss": 0.6802,
      "step": 17954
    },
    {
      "epoch": 5.669403220713609,
      "grad_norm": 4.045933246612549,
      "learning_rate": 3.980721327579553e-05,
      "loss": 0.6739,
      "step": 17955
    },
    {
      "epoch": 5.669718976949794,
      "grad_norm": 4.070292949676514,
      "learning_rate": 3.980633617513946e-05,
      "loss": 0.6727,
      "step": 17956
    },
    {
      "epoch": 5.670034733185981,
      "grad_norm": 3.9540228843688965,
      "learning_rate": 3.980545907448339e-05,
      "loss": 0.5939,
      "step": 17957
    },
    {
      "epoch": 5.670350489422166,
      "grad_norm": 3.8681743144989014,
      "learning_rate": 3.980458197382732e-05,
      "loss": 0.6081,
      "step": 17958
    },
    {
      "epoch": 5.670666245658352,
      "grad_norm": 3.785067319869995,
      "learning_rate": 3.980370487317125e-05,
      "loss": 0.5577,
      "step": 17959
    },
    {
      "epoch": 5.670982001894537,
      "grad_norm": 3.9342851638793945,
      "learning_rate": 3.9802827772515175e-05,
      "loss": 0.6512,
      "step": 17960
    },
    {
      "epoch": 5.671297758130724,
      "grad_norm": 4.6608123779296875,
      "learning_rate": 3.9801950671859106e-05,
      "loss": 0.6977,
      "step": 17961
    },
    {
      "epoch": 5.671613514366909,
      "grad_norm": 3.679504156112671,
      "learning_rate": 3.980107357120304e-05,
      "loss": 0.5601,
      "step": 17962
    },
    {
      "epoch": 5.671929270603094,
      "grad_norm": 7.2208356857299805,
      "learning_rate": 3.980019647054696e-05,
      "loss": 0.6227,
      "step": 17963
    },
    {
      "epoch": 5.67224502683928,
      "grad_norm": 3.593048572540283,
      "learning_rate": 3.9799319369890895e-05,
      "loss": 0.5562,
      "step": 17964
    },
    {
      "epoch": 5.672560783075466,
      "grad_norm": 3.7550601959228516,
      "learning_rate": 3.979844226923482e-05,
      "loss": 0.5594,
      "step": 17965
    },
    {
      "epoch": 5.672876539311652,
      "grad_norm": 3.133949041366577,
      "learning_rate": 3.979756516857875e-05,
      "loss": 0.5581,
      "step": 17966
    },
    {
      "epoch": 5.673192295547837,
      "grad_norm": 3.9888253211975098,
      "learning_rate": 3.9796688067922676e-05,
      "loss": 0.602,
      "step": 17967
    },
    {
      "epoch": 5.673508051784022,
      "grad_norm": 5.234245777130127,
      "learning_rate": 3.979581096726661e-05,
      "loss": 0.6119,
      "step": 17968
    },
    {
      "epoch": 5.673823808020209,
      "grad_norm": 4.066373348236084,
      "learning_rate": 3.979493386661053e-05,
      "loss": 0.6485,
      "step": 17969
    },
    {
      "epoch": 5.674139564256394,
      "grad_norm": 4.259029865264893,
      "learning_rate": 3.9794056765954464e-05,
      "loss": 0.6814,
      "step": 17970
    },
    {
      "epoch": 5.67445532049258,
      "grad_norm": 4.568853378295898,
      "learning_rate": 3.979317966529839e-05,
      "loss": 0.7688,
      "step": 17971
    },
    {
      "epoch": 5.674771076728765,
      "grad_norm": 3.8966915607452393,
      "learning_rate": 3.979230256464232e-05,
      "loss": 0.6117,
      "step": 17972
    },
    {
      "epoch": 5.675086832964951,
      "grad_norm": 4.627401351928711,
      "learning_rate": 3.9791425463986246e-05,
      "loss": 0.5579,
      "step": 17973
    },
    {
      "epoch": 5.675402589201137,
      "grad_norm": 3.982130765914917,
      "learning_rate": 3.979054836333018e-05,
      "loss": 0.7273,
      "step": 17974
    },
    {
      "epoch": 5.675718345437322,
      "grad_norm": 4.711671829223633,
      "learning_rate": 3.97896712626741e-05,
      "loss": 0.6284,
      "step": 17975
    },
    {
      "epoch": 5.676034101673508,
      "grad_norm": 4.644650936126709,
      "learning_rate": 3.9788794162018034e-05,
      "loss": 0.6718,
      "step": 17976
    },
    {
      "epoch": 5.676349857909694,
      "grad_norm": 5.236063480377197,
      "learning_rate": 3.978791706136196e-05,
      "loss": 0.685,
      "step": 17977
    },
    {
      "epoch": 5.676665614145879,
      "grad_norm": 4.212939739227295,
      "learning_rate": 3.978703996070589e-05,
      "loss": 0.605,
      "step": 17978
    },
    {
      "epoch": 5.676981370382065,
      "grad_norm": 3.798387289047241,
      "learning_rate": 3.978616286004982e-05,
      "loss": 0.6113,
      "step": 17979
    },
    {
      "epoch": 5.67729712661825,
      "grad_norm": 4.151945114135742,
      "learning_rate": 3.978528575939375e-05,
      "loss": 0.6908,
      "step": 17980
    },
    {
      "epoch": 5.6776128828544365,
      "grad_norm": 3.828526496887207,
      "learning_rate": 3.978440865873768e-05,
      "loss": 0.5615,
      "step": 17981
    },
    {
      "epoch": 5.677928639090622,
      "grad_norm": 4.441380500793457,
      "learning_rate": 3.9783531558081604e-05,
      "loss": 0.6734,
      "step": 17982
    },
    {
      "epoch": 5.678244395326808,
      "grad_norm": 5.33909273147583,
      "learning_rate": 3.9782654457425536e-05,
      "loss": 0.7341,
      "step": 17983
    },
    {
      "epoch": 5.678560151562993,
      "grad_norm": 4.390810012817383,
      "learning_rate": 3.978177735676947e-05,
      "loss": 0.6376,
      "step": 17984
    },
    {
      "epoch": 5.6788759077991795,
      "grad_norm": 3.966122627258301,
      "learning_rate": 3.97809002561134e-05,
      "loss": 0.6369,
      "step": 17985
    },
    {
      "epoch": 5.679191664035365,
      "grad_norm": 4.4447021484375,
      "learning_rate": 3.9780023155457324e-05,
      "loss": 0.6646,
      "step": 17986
    },
    {
      "epoch": 5.67950742027155,
      "grad_norm": 4.668946743011475,
      "learning_rate": 3.9779146054801255e-05,
      "loss": 0.6817,
      "step": 17987
    },
    {
      "epoch": 5.679823176507736,
      "grad_norm": 4.006040096282959,
      "learning_rate": 3.977826895414518e-05,
      "loss": 0.6228,
      "step": 17988
    },
    {
      "epoch": 5.680138932743922,
      "grad_norm": 4.210959434509277,
      "learning_rate": 3.977739185348911e-05,
      "loss": 0.6308,
      "step": 17989
    },
    {
      "epoch": 5.680454688980108,
      "grad_norm": 4.815338611602783,
      "learning_rate": 3.977651475283304e-05,
      "loss": 0.5847,
      "step": 17990
    },
    {
      "epoch": 5.680770445216293,
      "grad_norm": 4.454742908477783,
      "learning_rate": 3.977563765217697e-05,
      "loss": 0.7017,
      "step": 17991
    },
    {
      "epoch": 5.681086201452478,
      "grad_norm": 4.520463943481445,
      "learning_rate": 3.9774760551520894e-05,
      "loss": 0.7695,
      "step": 17992
    },
    {
      "epoch": 5.6814019576886645,
      "grad_norm": 4.332448959350586,
      "learning_rate": 3.9773883450864825e-05,
      "loss": 0.6696,
      "step": 17993
    },
    {
      "epoch": 5.68171771392485,
      "grad_norm": 3.577821969985962,
      "learning_rate": 3.977300635020875e-05,
      "loss": 0.6477,
      "step": 17994
    },
    {
      "epoch": 5.682033470161036,
      "grad_norm": 4.057619571685791,
      "learning_rate": 3.977212924955268e-05,
      "loss": 0.6238,
      "step": 17995
    },
    {
      "epoch": 5.682349226397221,
      "grad_norm": 3.7320189476013184,
      "learning_rate": 3.977125214889661e-05,
      "loss": 0.5842,
      "step": 17996
    },
    {
      "epoch": 5.682664982633407,
      "grad_norm": 3.541393280029297,
      "learning_rate": 3.977037504824054e-05,
      "loss": 0.6095,
      "step": 17997
    },
    {
      "epoch": 5.682980738869593,
      "grad_norm": 4.324033260345459,
      "learning_rate": 3.976949794758446e-05,
      "loss": 0.6341,
      "step": 17998
    },
    {
      "epoch": 5.683296495105778,
      "grad_norm": 4.022158145904541,
      "learning_rate": 3.9768620846928395e-05,
      "loss": 0.7505,
      "step": 17999
    },
    {
      "epoch": 5.683612251341964,
      "grad_norm": 3.790184736251831,
      "learning_rate": 3.976774374627233e-05,
      "loss": 0.5935,
      "step": 18000
    },
    {
      "epoch": 5.683612251341964,
      "eval_loss": 0.827370822429657,
      "eval_runtime": 4.6458,
      "eval_samples_per_second": 535.969,
      "eval_steps_per_second": 4.52,
      "step": 18000
    },
    {
      "epoch": 5.6839280075781495,
      "grad_norm": 4.918771266937256,
      "learning_rate": 3.976686664561625e-05,
      "loss": 0.6605,
      "step": 18001
    },
    {
      "epoch": 5.684243763814336,
      "grad_norm": 4.0752058029174805,
      "learning_rate": 3.976598954496018e-05,
      "loss": 0.559,
      "step": 18002
    },
    {
      "epoch": 5.684559520050521,
      "grad_norm": 3.931971549987793,
      "learning_rate": 3.976511244430411e-05,
      "loss": 0.636,
      "step": 18003
    },
    {
      "epoch": 5.684875276286706,
      "grad_norm": 3.8044347763061523,
      "learning_rate": 3.976423534364804e-05,
      "loss": 0.5231,
      "step": 18004
    },
    {
      "epoch": 5.6851910325228925,
      "grad_norm": 3.6874992847442627,
      "learning_rate": 3.9763358242991965e-05,
      "loss": 0.6087,
      "step": 18005
    },
    {
      "epoch": 5.685506788759078,
      "grad_norm": 3.9256432056427,
      "learning_rate": 3.9762481142335896e-05,
      "loss": 0.5648,
      "step": 18006
    },
    {
      "epoch": 5.685822544995264,
      "grad_norm": 5.924459934234619,
      "learning_rate": 3.976160404167982e-05,
      "loss": 0.7205,
      "step": 18007
    },
    {
      "epoch": 5.686138301231449,
      "grad_norm": 3.869635820388794,
      "learning_rate": 3.976072694102375e-05,
      "loss": 0.6405,
      "step": 18008
    },
    {
      "epoch": 5.686454057467635,
      "grad_norm": 4.881205081939697,
      "learning_rate": 3.975984984036768e-05,
      "loss": 0.7713,
      "step": 18009
    },
    {
      "epoch": 5.686769813703821,
      "grad_norm": 4.816073894500732,
      "learning_rate": 3.975897273971161e-05,
      "loss": 0.7055,
      "step": 18010
    },
    {
      "epoch": 5.687085569940006,
      "grad_norm": 3.639308452606201,
      "learning_rate": 3.975809563905554e-05,
      "loss": 0.569,
      "step": 18011
    },
    {
      "epoch": 5.687401326176192,
      "grad_norm": 3.986083984375,
      "learning_rate": 3.975721853839947e-05,
      "loss": 0.5878,
      "step": 18012
    },
    {
      "epoch": 5.6877170824123775,
      "grad_norm": 4.163175106048584,
      "learning_rate": 3.97563414377434e-05,
      "loss": 0.7068,
      "step": 18013
    },
    {
      "epoch": 5.688032838648564,
      "grad_norm": 3.9136478900909424,
      "learning_rate": 3.975546433708733e-05,
      "loss": 0.5776,
      "step": 18014
    },
    {
      "epoch": 5.688348594884749,
      "grad_norm": 4.607946872711182,
      "learning_rate": 3.9754587236431254e-05,
      "loss": 0.647,
      "step": 18015
    },
    {
      "epoch": 5.688664351120934,
      "grad_norm": 4.060205459594727,
      "learning_rate": 3.9753710135775186e-05,
      "loss": 0.5405,
      "step": 18016
    },
    {
      "epoch": 5.6889801073571205,
      "grad_norm": 4.257808685302734,
      "learning_rate": 3.975283303511911e-05,
      "loss": 0.647,
      "step": 18017
    },
    {
      "epoch": 5.689295863593306,
      "grad_norm": 3.907241106033325,
      "learning_rate": 3.975195593446304e-05,
      "loss": 0.5853,
      "step": 18018
    },
    {
      "epoch": 5.689611619829492,
      "grad_norm": 4.331718444824219,
      "learning_rate": 3.975107883380697e-05,
      "loss": 0.6325,
      "step": 18019
    },
    {
      "epoch": 5.689927376065677,
      "grad_norm": 4.124423980712891,
      "learning_rate": 3.97502017331509e-05,
      "loss": 0.7212,
      "step": 18020
    },
    {
      "epoch": 5.6902431323018625,
      "grad_norm": 3.962995767593384,
      "learning_rate": 3.9749324632494824e-05,
      "loss": 0.6007,
      "step": 18021
    },
    {
      "epoch": 5.690558888538049,
      "grad_norm": 3.638333320617676,
      "learning_rate": 3.9748447531838756e-05,
      "loss": 0.709,
      "step": 18022
    },
    {
      "epoch": 5.690874644774234,
      "grad_norm": 4.579648017883301,
      "learning_rate": 3.974757043118269e-05,
      "loss": 0.7716,
      "step": 18023
    },
    {
      "epoch": 5.69119040101042,
      "grad_norm": 5.0504231452941895,
      "learning_rate": 3.974669333052661e-05,
      "loss": 0.6364,
      "step": 18024
    },
    {
      "epoch": 5.6915061572466055,
      "grad_norm": 4.392604827880859,
      "learning_rate": 3.9745816229870544e-05,
      "loss": 0.6759,
      "step": 18025
    },
    {
      "epoch": 5.691821913482792,
      "grad_norm": 3.2425789833068848,
      "learning_rate": 3.974493912921447e-05,
      "loss": 0.5755,
      "step": 18026
    },
    {
      "epoch": 5.692137669718977,
      "grad_norm": 5.192392826080322,
      "learning_rate": 3.97440620285584e-05,
      "loss": 0.6273,
      "step": 18027
    },
    {
      "epoch": 5.692453425955162,
      "grad_norm": 4.3123698234558105,
      "learning_rate": 3.9743184927902326e-05,
      "loss": 0.6513,
      "step": 18028
    },
    {
      "epoch": 5.692769182191348,
      "grad_norm": 4.260972023010254,
      "learning_rate": 3.974230782724626e-05,
      "loss": 0.7807,
      "step": 18029
    },
    {
      "epoch": 5.693084938427534,
      "grad_norm": 4.650468349456787,
      "learning_rate": 3.974143072659018e-05,
      "loss": 0.6278,
      "step": 18030
    },
    {
      "epoch": 5.69340069466372,
      "grad_norm": 3.8666324615478516,
      "learning_rate": 3.9740553625934114e-05,
      "loss": 0.6054,
      "step": 18031
    },
    {
      "epoch": 5.693716450899905,
      "grad_norm": 4.4794135093688965,
      "learning_rate": 3.973967652527804e-05,
      "loss": 0.6904,
      "step": 18032
    },
    {
      "epoch": 5.694032207136091,
      "grad_norm": 3.998015880584717,
      "learning_rate": 3.973879942462197e-05,
      "loss": 0.6265,
      "step": 18033
    },
    {
      "epoch": 5.694347963372277,
      "grad_norm": 4.667994976043701,
      "learning_rate": 3.9737922323965895e-05,
      "loss": 0.5972,
      "step": 18034
    },
    {
      "epoch": 5.694663719608462,
      "grad_norm": 4.202098846435547,
      "learning_rate": 3.973704522330983e-05,
      "loss": 0.6938,
      "step": 18035
    },
    {
      "epoch": 5.694979475844648,
      "grad_norm": 3.9726316928863525,
      "learning_rate": 3.973616812265375e-05,
      "loss": 0.6372,
      "step": 18036
    },
    {
      "epoch": 5.695295232080833,
      "grad_norm": 4.1424384117126465,
      "learning_rate": 3.9735291021997684e-05,
      "loss": 0.6053,
      "step": 18037
    },
    {
      "epoch": 5.69561098831702,
      "grad_norm": 4.7911057472229,
      "learning_rate": 3.9734413921341615e-05,
      "loss": 0.7317,
      "step": 18038
    },
    {
      "epoch": 5.695926744553205,
      "grad_norm": 5.218666076660156,
      "learning_rate": 3.973353682068555e-05,
      "loss": 0.617,
      "step": 18039
    },
    {
      "epoch": 5.69624250078939,
      "grad_norm": 3.5709450244903564,
      "learning_rate": 3.973265972002947e-05,
      "loss": 0.5649,
      "step": 18040
    },
    {
      "epoch": 5.696558257025576,
      "grad_norm": 4.671183109283447,
      "learning_rate": 3.9731782619373404e-05,
      "loss": 0.5941,
      "step": 18041
    },
    {
      "epoch": 5.696874013261762,
      "grad_norm": 4.553045749664307,
      "learning_rate": 3.973090551871733e-05,
      "loss": 0.6654,
      "step": 18042
    },
    {
      "epoch": 5.697189769497948,
      "grad_norm": 4.3641533851623535,
      "learning_rate": 3.973002841806126e-05,
      "loss": 0.5583,
      "step": 18043
    },
    {
      "epoch": 5.697505525734133,
      "grad_norm": 4.226187229156494,
      "learning_rate": 3.972915131740519e-05,
      "loss": 0.5327,
      "step": 18044
    },
    {
      "epoch": 5.6978212819703185,
      "grad_norm": 4.159386157989502,
      "learning_rate": 3.972827421674912e-05,
      "loss": 0.669,
      "step": 18045
    },
    {
      "epoch": 5.698137038206505,
      "grad_norm": 3.6355600357055664,
      "learning_rate": 3.972739711609305e-05,
      "loss": 0.6542,
      "step": 18046
    },
    {
      "epoch": 5.69845279444269,
      "grad_norm": 3.809216260910034,
      "learning_rate": 3.9726520015436973e-05,
      "loss": 0.696,
      "step": 18047
    },
    {
      "epoch": 5.698768550678876,
      "grad_norm": 3.726848602294922,
      "learning_rate": 3.9725642914780905e-05,
      "loss": 0.7419,
      "step": 18048
    },
    {
      "epoch": 5.699084306915061,
      "grad_norm": 3.852327823638916,
      "learning_rate": 3.972476581412483e-05,
      "loss": 0.6763,
      "step": 18049
    },
    {
      "epoch": 5.699400063151248,
      "grad_norm": 3.9865996837615967,
      "learning_rate": 3.972388871346876e-05,
      "loss": 0.6602,
      "step": 18050
    },
    {
      "epoch": 5.699715819387433,
      "grad_norm": 4.003561496734619,
      "learning_rate": 3.972301161281269e-05,
      "loss": 0.6212,
      "step": 18051
    },
    {
      "epoch": 5.700031575623619,
      "grad_norm": 3.836841583251953,
      "learning_rate": 3.972213451215662e-05,
      "loss": 0.652,
      "step": 18052
    },
    {
      "epoch": 5.700347331859804,
      "grad_norm": 4.189659595489502,
      "learning_rate": 3.972125741150054e-05,
      "loss": 0.6759,
      "step": 18053
    },
    {
      "epoch": 5.70066308809599,
      "grad_norm": 3.767085075378418,
      "learning_rate": 3.9720380310844475e-05,
      "loss": 0.7223,
      "step": 18054
    },
    {
      "epoch": 5.700978844332176,
      "grad_norm": 4.830059051513672,
      "learning_rate": 3.97195032101884e-05,
      "loss": 0.6701,
      "step": 18055
    },
    {
      "epoch": 5.701294600568361,
      "grad_norm": 4.524158954620361,
      "learning_rate": 3.971862610953233e-05,
      "loss": 0.6051,
      "step": 18056
    },
    {
      "epoch": 5.701610356804547,
      "grad_norm": 4.330740451812744,
      "learning_rate": 3.9717749008876256e-05,
      "loss": 0.6031,
      "step": 18057
    },
    {
      "epoch": 5.701926113040733,
      "grad_norm": 5.145303249359131,
      "learning_rate": 3.971687190822019e-05,
      "loss": 0.709,
      "step": 18058
    },
    {
      "epoch": 5.702241869276918,
      "grad_norm": 3.9220783710479736,
      "learning_rate": 3.971599480756412e-05,
      "loss": 0.6246,
      "step": 18059
    },
    {
      "epoch": 5.702557625513104,
      "grad_norm": 2.8711934089660645,
      "learning_rate": 3.9715117706908045e-05,
      "loss": 0.5804,
      "step": 18060
    },
    {
      "epoch": 5.702873381749289,
      "grad_norm": 4.765382766723633,
      "learning_rate": 3.9714240606251976e-05,
      "loss": 0.6756,
      "step": 18061
    },
    {
      "epoch": 5.7031891379854756,
      "grad_norm": 3.575773239135742,
      "learning_rate": 3.97133635055959e-05,
      "loss": 0.5294,
      "step": 18062
    },
    {
      "epoch": 5.703504894221661,
      "grad_norm": 4.12835168838501,
      "learning_rate": 3.971248640493983e-05,
      "loss": 0.6255,
      "step": 18063
    },
    {
      "epoch": 5.703820650457846,
      "grad_norm": 5.5377607345581055,
      "learning_rate": 3.971160930428376e-05,
      "loss": 0.6953,
      "step": 18064
    },
    {
      "epoch": 5.704136406694032,
      "grad_norm": 3.720442533493042,
      "learning_rate": 3.9710732203627696e-05,
      "loss": 0.5822,
      "step": 18065
    },
    {
      "epoch": 5.704452162930218,
      "grad_norm": 5.047321319580078,
      "learning_rate": 3.970985510297162e-05,
      "loss": 0.6356,
      "step": 18066
    },
    {
      "epoch": 5.704767919166404,
      "grad_norm": 4.084811687469482,
      "learning_rate": 3.970897800231555e-05,
      "loss": 0.6007,
      "step": 18067
    },
    {
      "epoch": 5.705083675402589,
      "grad_norm": 4.597822666168213,
      "learning_rate": 3.970810090165948e-05,
      "loss": 0.6214,
      "step": 18068
    },
    {
      "epoch": 5.705399431638774,
      "grad_norm": 4.761595249176025,
      "learning_rate": 3.970722380100341e-05,
      "loss": 0.7762,
      "step": 18069
    },
    {
      "epoch": 5.705715187874961,
      "grad_norm": 5.43411922454834,
      "learning_rate": 3.9706346700347334e-05,
      "loss": 0.7344,
      "step": 18070
    },
    {
      "epoch": 5.706030944111146,
      "grad_norm": 4.002972602844238,
      "learning_rate": 3.9705469599691266e-05,
      "loss": 0.673,
      "step": 18071
    },
    {
      "epoch": 5.706346700347332,
      "grad_norm": 4.750976085662842,
      "learning_rate": 3.970459249903519e-05,
      "loss": 0.6891,
      "step": 18072
    },
    {
      "epoch": 5.706662456583517,
      "grad_norm": 4.150424957275391,
      "learning_rate": 3.970371539837912e-05,
      "loss": 0.6144,
      "step": 18073
    },
    {
      "epoch": 5.7069782128197035,
      "grad_norm": 4.240860939025879,
      "learning_rate": 3.970283829772305e-05,
      "loss": 0.5816,
      "step": 18074
    },
    {
      "epoch": 5.707293969055889,
      "grad_norm": 4.469074249267578,
      "learning_rate": 3.970196119706698e-05,
      "loss": 0.6242,
      "step": 18075
    },
    {
      "epoch": 5.707609725292075,
      "grad_norm": 3.92980694770813,
      "learning_rate": 3.9701084096410904e-05,
      "loss": 0.5189,
      "step": 18076
    },
    {
      "epoch": 5.70792548152826,
      "grad_norm": 3.843874216079712,
      "learning_rate": 3.9700206995754836e-05,
      "loss": 0.6838,
      "step": 18077
    },
    {
      "epoch": 5.708241237764446,
      "grad_norm": 3.8174381256103516,
      "learning_rate": 3.969932989509876e-05,
      "loss": 0.6391,
      "step": 18078
    },
    {
      "epoch": 5.708556994000632,
      "grad_norm": 4.351168155670166,
      "learning_rate": 3.969845279444269e-05,
      "loss": 0.6054,
      "step": 18079
    },
    {
      "epoch": 5.708872750236817,
      "grad_norm": 3.814697742462158,
      "learning_rate": 3.969757569378662e-05,
      "loss": 0.5421,
      "step": 18080
    },
    {
      "epoch": 5.709188506473003,
      "grad_norm": 3.2108163833618164,
      "learning_rate": 3.969669859313055e-05,
      "loss": 0.4652,
      "step": 18081
    },
    {
      "epoch": 5.7095042627091885,
      "grad_norm": 4.320159912109375,
      "learning_rate": 3.969582149247448e-05,
      "loss": 0.6368,
      "step": 18082
    },
    {
      "epoch": 5.709820018945374,
      "grad_norm": 4.301817893981934,
      "learning_rate": 3.9694944391818406e-05,
      "loss": 0.677,
      "step": 18083
    },
    {
      "epoch": 5.71013577518156,
      "grad_norm": 3.9977478981018066,
      "learning_rate": 3.969406729116234e-05,
      "loss": 0.5653,
      "step": 18084
    },
    {
      "epoch": 5.710451531417745,
      "grad_norm": 4.475643157958984,
      "learning_rate": 3.969319019050626e-05,
      "loss": 0.6354,
      "step": 18085
    },
    {
      "epoch": 5.7107672876539315,
      "grad_norm": 3.9106884002685547,
      "learning_rate": 3.9692313089850194e-05,
      "loss": 0.5426,
      "step": 18086
    },
    {
      "epoch": 5.711083043890117,
      "grad_norm": 3.7624619007110596,
      "learning_rate": 3.969143598919412e-05,
      "loss": 0.601,
      "step": 18087
    },
    {
      "epoch": 5.711398800126302,
      "grad_norm": 4.094799041748047,
      "learning_rate": 3.969055888853805e-05,
      "loss": 0.6439,
      "step": 18088
    },
    {
      "epoch": 5.711714556362488,
      "grad_norm": 4.019956588745117,
      "learning_rate": 3.9689681787881975e-05,
      "loss": 0.6082,
      "step": 18089
    },
    {
      "epoch": 5.7120303125986736,
      "grad_norm": 4.831859111785889,
      "learning_rate": 3.968880468722591e-05,
      "loss": 0.6988,
      "step": 18090
    },
    {
      "epoch": 5.71234606883486,
      "grad_norm": 5.541704177856445,
      "learning_rate": 3.968792758656983e-05,
      "loss": 0.6188,
      "step": 18091
    },
    {
      "epoch": 5.712661825071045,
      "grad_norm": 4.1610589027404785,
      "learning_rate": 3.9687050485913764e-05,
      "loss": 0.6611,
      "step": 18092
    },
    {
      "epoch": 5.71297758130723,
      "grad_norm": 4.072085857391357,
      "learning_rate": 3.9686173385257695e-05,
      "loss": 0.6196,
      "step": 18093
    },
    {
      "epoch": 5.7132933375434165,
      "grad_norm": 3.3299543857574463,
      "learning_rate": 3.968529628460163e-05,
      "loss": 0.5438,
      "step": 18094
    },
    {
      "epoch": 5.713609093779602,
      "grad_norm": 4.461201190948486,
      "learning_rate": 3.968441918394555e-05,
      "loss": 0.5838,
      "step": 18095
    },
    {
      "epoch": 5.713924850015788,
      "grad_norm": 4.005288600921631,
      "learning_rate": 3.9683542083289484e-05,
      "loss": 0.722,
      "step": 18096
    },
    {
      "epoch": 5.714240606251973,
      "grad_norm": 3.950122594833374,
      "learning_rate": 3.968266498263341e-05,
      "loss": 0.5497,
      "step": 18097
    },
    {
      "epoch": 5.7145563624881595,
      "grad_norm": 3.665900230407715,
      "learning_rate": 3.968178788197734e-05,
      "loss": 0.4916,
      "step": 18098
    },
    {
      "epoch": 5.714872118724345,
      "grad_norm": 4.863375663757324,
      "learning_rate": 3.9680910781321265e-05,
      "loss": 0.6489,
      "step": 18099
    },
    {
      "epoch": 5.715187874960531,
      "grad_norm": 3.880350112915039,
      "learning_rate": 3.96800336806652e-05,
      "loss": 0.5188,
      "step": 18100
    },
    {
      "epoch": 5.715503631196716,
      "grad_norm": 4.36742639541626,
      "learning_rate": 3.967915658000912e-05,
      "loss": 0.6315,
      "step": 18101
    },
    {
      "epoch": 5.7158193874329015,
      "grad_norm": 3.995882511138916,
      "learning_rate": 3.967827947935305e-05,
      "loss": 0.6959,
      "step": 18102
    },
    {
      "epoch": 5.716135143669088,
      "grad_norm": 3.7590599060058594,
      "learning_rate": 3.9677402378696985e-05,
      "loss": 0.6525,
      "step": 18103
    },
    {
      "epoch": 5.716450899905273,
      "grad_norm": 4.481267929077148,
      "learning_rate": 3.967652527804091e-05,
      "loss": 0.7263,
      "step": 18104
    },
    {
      "epoch": 5.716766656141459,
      "grad_norm": 4.806791305541992,
      "learning_rate": 3.967564817738484e-05,
      "loss": 0.6896,
      "step": 18105
    },
    {
      "epoch": 5.7170824123776445,
      "grad_norm": 4.487369060516357,
      "learning_rate": 3.9674771076728767e-05,
      "loss": 0.7175,
      "step": 18106
    },
    {
      "epoch": 5.71739816861383,
      "grad_norm": 4.920586109161377,
      "learning_rate": 3.96738939760727e-05,
      "loss": 0.7392,
      "step": 18107
    },
    {
      "epoch": 5.717713924850016,
      "grad_norm": 4.977016448974609,
      "learning_rate": 3.967301687541662e-05,
      "loss": 0.6035,
      "step": 18108
    },
    {
      "epoch": 5.718029681086201,
      "grad_norm": 3.752465009689331,
      "learning_rate": 3.9672139774760555e-05,
      "loss": 0.6701,
      "step": 18109
    },
    {
      "epoch": 5.718345437322387,
      "grad_norm": 3.667419672012329,
      "learning_rate": 3.967126267410448e-05,
      "loss": 0.5913,
      "step": 18110
    },
    {
      "epoch": 5.718661193558573,
      "grad_norm": 4.167113780975342,
      "learning_rate": 3.967038557344841e-05,
      "loss": 0.6556,
      "step": 18111
    },
    {
      "epoch": 5.718976949794758,
      "grad_norm": 4.500262260437012,
      "learning_rate": 3.9669508472792336e-05,
      "loss": 0.7011,
      "step": 18112
    },
    {
      "epoch": 5.719292706030944,
      "grad_norm": 4.5573272705078125,
      "learning_rate": 3.966863137213627e-05,
      "loss": 0.727,
      "step": 18113
    },
    {
      "epoch": 5.7196084622671295,
      "grad_norm": 5.011991024017334,
      "learning_rate": 3.966775427148019e-05,
      "loss": 0.6631,
      "step": 18114
    },
    {
      "epoch": 5.719924218503316,
      "grad_norm": 4.380434989929199,
      "learning_rate": 3.9666877170824125e-05,
      "loss": 0.6804,
      "step": 18115
    },
    {
      "epoch": 5.720239974739501,
      "grad_norm": 4.237793922424316,
      "learning_rate": 3.966600007016805e-05,
      "loss": 0.6432,
      "step": 18116
    },
    {
      "epoch": 5.720555730975687,
      "grad_norm": 3.439128875732422,
      "learning_rate": 3.966512296951198e-05,
      "loss": 0.6179,
      "step": 18117
    },
    {
      "epoch": 5.720871487211872,
      "grad_norm": 4.921256065368652,
      "learning_rate": 3.9664245868855906e-05,
      "loss": 0.7067,
      "step": 18118
    },
    {
      "epoch": 5.721187243448058,
      "grad_norm": 5.508640766143799,
      "learning_rate": 3.966336876819984e-05,
      "loss": 0.7187,
      "step": 18119
    },
    {
      "epoch": 5.721502999684244,
      "grad_norm": 4.970308780670166,
      "learning_rate": 3.966249166754377e-05,
      "loss": 0.7249,
      "step": 18120
    },
    {
      "epoch": 5.721818755920429,
      "grad_norm": 3.4622974395751953,
      "learning_rate": 3.96616145668877e-05,
      "loss": 0.6303,
      "step": 18121
    },
    {
      "epoch": 5.722134512156615,
      "grad_norm": 4.375384330749512,
      "learning_rate": 3.9660737466231626e-05,
      "loss": 0.7279,
      "step": 18122
    },
    {
      "epoch": 5.722450268392801,
      "grad_norm": 4.037937164306641,
      "learning_rate": 3.965986036557556e-05,
      "loss": 0.6177,
      "step": 18123
    },
    {
      "epoch": 5.722766024628987,
      "grad_norm": 3.3431196212768555,
      "learning_rate": 3.965898326491948e-05,
      "loss": 0.6267,
      "step": 18124
    },
    {
      "epoch": 5.723081780865172,
      "grad_norm": 4.325197219848633,
      "learning_rate": 3.9658106164263414e-05,
      "loss": 0.6944,
      "step": 18125
    },
    {
      "epoch": 5.7233975371013575,
      "grad_norm": 3.3320884704589844,
      "learning_rate": 3.9657229063607346e-05,
      "loss": 0.4958,
      "step": 18126
    },
    {
      "epoch": 5.723713293337544,
      "grad_norm": 4.098400115966797,
      "learning_rate": 3.965635196295127e-05,
      "loss": 0.6481,
      "step": 18127
    },
    {
      "epoch": 5.724029049573729,
      "grad_norm": 3.5600697994232178,
      "learning_rate": 3.96554748622952e-05,
      "loss": 0.6203,
      "step": 18128
    },
    {
      "epoch": 5.724344805809915,
      "grad_norm": 3.7212772369384766,
      "learning_rate": 3.965459776163913e-05,
      "loss": 0.637,
      "step": 18129
    },
    {
      "epoch": 5.7246605620461,
      "grad_norm": 4.756200790405273,
      "learning_rate": 3.965372066098306e-05,
      "loss": 0.6179,
      "step": 18130
    },
    {
      "epoch": 5.724976318282286,
      "grad_norm": 4.701661586761475,
      "learning_rate": 3.9652843560326984e-05,
      "loss": 0.6648,
      "step": 18131
    },
    {
      "epoch": 5.725292074518472,
      "grad_norm": 3.7689554691314697,
      "learning_rate": 3.9651966459670916e-05,
      "loss": 0.609,
      "step": 18132
    },
    {
      "epoch": 5.725607830754657,
      "grad_norm": 3.88350248336792,
      "learning_rate": 3.965108935901484e-05,
      "loss": 0.567,
      "step": 18133
    },
    {
      "epoch": 5.725923586990843,
      "grad_norm": 3.8184080123901367,
      "learning_rate": 3.965021225835877e-05,
      "loss": 0.5947,
      "step": 18134
    },
    {
      "epoch": 5.726239343227029,
      "grad_norm": 4.123953342437744,
      "learning_rate": 3.96493351577027e-05,
      "loss": 0.6394,
      "step": 18135
    },
    {
      "epoch": 5.726555099463214,
      "grad_norm": 4.771387577056885,
      "learning_rate": 3.964845805704663e-05,
      "loss": 0.6728,
      "step": 18136
    },
    {
      "epoch": 5.7268708556994,
      "grad_norm": 4.357162952423096,
      "learning_rate": 3.9647580956390554e-05,
      "loss": 0.7138,
      "step": 18137
    },
    {
      "epoch": 5.727186611935585,
      "grad_norm": 4.381577968597412,
      "learning_rate": 3.9646703855734485e-05,
      "loss": 0.6597,
      "step": 18138
    },
    {
      "epoch": 5.727502368171772,
      "grad_norm": 4.151504993438721,
      "learning_rate": 3.964582675507841e-05,
      "loss": 0.6438,
      "step": 18139
    },
    {
      "epoch": 5.727818124407957,
      "grad_norm": 4.916881084442139,
      "learning_rate": 3.964494965442234e-05,
      "loss": 0.6911,
      "step": 18140
    },
    {
      "epoch": 5.728133880644143,
      "grad_norm": 4.777182102203369,
      "learning_rate": 3.9644072553766274e-05,
      "loss": 0.7838,
      "step": 18141
    },
    {
      "epoch": 5.728449636880328,
      "grad_norm": 4.099913120269775,
      "learning_rate": 3.96431954531102e-05,
      "loss": 0.6358,
      "step": 18142
    },
    {
      "epoch": 5.728765393116514,
      "grad_norm": 3.8852572441101074,
      "learning_rate": 3.964231835245413e-05,
      "loss": 0.6555,
      "step": 18143
    },
    {
      "epoch": 5.7290811493527,
      "grad_norm": 4.12269926071167,
      "learning_rate": 3.9641441251798055e-05,
      "loss": 0.6759,
      "step": 18144
    },
    {
      "epoch": 5.729396905588885,
      "grad_norm": 3.7461495399475098,
      "learning_rate": 3.964056415114199e-05,
      "loss": 0.6237,
      "step": 18145
    },
    {
      "epoch": 5.729712661825071,
      "grad_norm": 4.429337978363037,
      "learning_rate": 3.963968705048591e-05,
      "loss": 0.7025,
      "step": 18146
    },
    {
      "epoch": 5.730028418061257,
      "grad_norm": 3.7193589210510254,
      "learning_rate": 3.963880994982985e-05,
      "loss": 0.6255,
      "step": 18147
    },
    {
      "epoch": 5.730344174297443,
      "grad_norm": 4.363398551940918,
      "learning_rate": 3.9637932849173775e-05,
      "loss": 0.6952,
      "step": 18148
    },
    {
      "epoch": 5.730659930533628,
      "grad_norm": 3.810373067855835,
      "learning_rate": 3.963705574851771e-05,
      "loss": 0.6656,
      "step": 18149
    },
    {
      "epoch": 5.730975686769813,
      "grad_norm": 4.1511030197143555,
      "learning_rate": 3.963617864786163e-05,
      "loss": 0.6918,
      "step": 18150
    },
    {
      "epoch": 5.731291443006,
      "grad_norm": 4.152740001678467,
      "learning_rate": 3.9635301547205563e-05,
      "loss": 0.6955,
      "step": 18151
    },
    {
      "epoch": 5.731607199242185,
      "grad_norm": 4.394556522369385,
      "learning_rate": 3.963442444654949e-05,
      "loss": 0.6851,
      "step": 18152
    },
    {
      "epoch": 5.731922955478371,
      "grad_norm": 3.671177625656128,
      "learning_rate": 3.963354734589342e-05,
      "loss": 0.5883,
      "step": 18153
    },
    {
      "epoch": 5.732238711714556,
      "grad_norm": 3.9377589225769043,
      "learning_rate": 3.9632670245237345e-05,
      "loss": 0.6053,
      "step": 18154
    },
    {
      "epoch": 5.732554467950742,
      "grad_norm": 3.476602792739868,
      "learning_rate": 3.963179314458128e-05,
      "loss": 0.5985,
      "step": 18155
    },
    {
      "epoch": 5.732870224186928,
      "grad_norm": 3.8882644176483154,
      "learning_rate": 3.96309160439252e-05,
      "loss": 0.5873,
      "step": 18156
    },
    {
      "epoch": 5.733185980423113,
      "grad_norm": 4.330459117889404,
      "learning_rate": 3.963003894326913e-05,
      "loss": 0.6516,
      "step": 18157
    },
    {
      "epoch": 5.733501736659299,
      "grad_norm": 3.6954665184020996,
      "learning_rate": 3.962916184261306e-05,
      "loss": 0.6224,
      "step": 18158
    },
    {
      "epoch": 5.733817492895485,
      "grad_norm": 4.414912700653076,
      "learning_rate": 3.962828474195699e-05,
      "loss": 0.8152,
      "step": 18159
    },
    {
      "epoch": 5.73413324913167,
      "grad_norm": 3.670738935470581,
      "learning_rate": 3.9627407641300915e-05,
      "loss": 0.7106,
      "step": 18160
    },
    {
      "epoch": 5.734449005367856,
      "grad_norm": 4.217625141143799,
      "learning_rate": 3.9626530540644846e-05,
      "loss": 0.6805,
      "step": 18161
    },
    {
      "epoch": 5.734764761604041,
      "grad_norm": 3.8234875202178955,
      "learning_rate": 3.962565343998877e-05,
      "loss": 0.5577,
      "step": 18162
    },
    {
      "epoch": 5.7350805178402275,
      "grad_norm": 3.8411877155303955,
      "learning_rate": 3.96247763393327e-05,
      "loss": 0.6251,
      "step": 18163
    },
    {
      "epoch": 5.735396274076413,
      "grad_norm": 4.258761882781982,
      "learning_rate": 3.9623899238676635e-05,
      "loss": 0.6,
      "step": 18164
    },
    {
      "epoch": 5.735712030312599,
      "grad_norm": 3.402988910675049,
      "learning_rate": 3.962302213802056e-05,
      "loss": 0.5848,
      "step": 18165
    },
    {
      "epoch": 5.736027786548784,
      "grad_norm": 3.8912577629089355,
      "learning_rate": 3.962214503736449e-05,
      "loss": 0.6133,
      "step": 18166
    },
    {
      "epoch": 5.7363435427849705,
      "grad_norm": 4.058951377868652,
      "learning_rate": 3.9621267936708416e-05,
      "loss": 0.5974,
      "step": 18167
    },
    {
      "epoch": 5.736659299021156,
      "grad_norm": 3.392603874206543,
      "learning_rate": 3.962039083605235e-05,
      "loss": 0.5842,
      "step": 18168
    },
    {
      "epoch": 5.736975055257341,
      "grad_norm": 4.553102970123291,
      "learning_rate": 3.961951373539627e-05,
      "loss": 0.6354,
      "step": 18169
    },
    {
      "epoch": 5.737290811493527,
      "grad_norm": 3.812333583831787,
      "learning_rate": 3.9618636634740204e-05,
      "loss": 0.6351,
      "step": 18170
    },
    {
      "epoch": 5.737606567729713,
      "grad_norm": 3.67236065864563,
      "learning_rate": 3.961775953408413e-05,
      "loss": 0.6556,
      "step": 18171
    },
    {
      "epoch": 5.737922323965899,
      "grad_norm": 4.235413074493408,
      "learning_rate": 3.961688243342806e-05,
      "loss": 0.6357,
      "step": 18172
    },
    {
      "epoch": 5.738238080202084,
      "grad_norm": 3.3498001098632812,
      "learning_rate": 3.9616005332771986e-05,
      "loss": 0.6465,
      "step": 18173
    },
    {
      "epoch": 5.738553836438269,
      "grad_norm": 4.541024208068848,
      "learning_rate": 3.9615128232115924e-05,
      "loss": 0.6176,
      "step": 18174
    },
    {
      "epoch": 5.7388695926744555,
      "grad_norm": 4.577078342437744,
      "learning_rate": 3.961425113145985e-05,
      "loss": 0.6524,
      "step": 18175
    },
    {
      "epoch": 5.739185348910641,
      "grad_norm": 3.7632839679718018,
      "learning_rate": 3.961337403080378e-05,
      "loss": 0.6201,
      "step": 18176
    },
    {
      "epoch": 5.739501105146827,
      "grad_norm": 5.2779221534729,
      "learning_rate": 3.9612496930147706e-05,
      "loss": 0.7173,
      "step": 18177
    },
    {
      "epoch": 5.739816861383012,
      "grad_norm": 3.993480920791626,
      "learning_rate": 3.961161982949164e-05,
      "loss": 0.7504,
      "step": 18178
    },
    {
      "epoch": 5.740132617619198,
      "grad_norm": 5.174374103546143,
      "learning_rate": 3.961074272883556e-05,
      "loss": 0.6349,
      "step": 18179
    },
    {
      "epoch": 5.740448373855384,
      "grad_norm": 4.913984298706055,
      "learning_rate": 3.9609865628179494e-05,
      "loss": 0.6386,
      "step": 18180
    },
    {
      "epoch": 5.740764130091569,
      "grad_norm": 3.5071868896484375,
      "learning_rate": 3.960898852752342e-05,
      "loss": 0.6529,
      "step": 18181
    },
    {
      "epoch": 5.741079886327755,
      "grad_norm": 4.051624774932861,
      "learning_rate": 3.960811142686735e-05,
      "loss": 0.6623,
      "step": 18182
    },
    {
      "epoch": 5.7413956425639405,
      "grad_norm": 3.9928479194641113,
      "learning_rate": 3.9607234326211276e-05,
      "loss": 0.6788,
      "step": 18183
    },
    {
      "epoch": 5.741711398800126,
      "grad_norm": 4.534483909606934,
      "learning_rate": 3.960635722555521e-05,
      "loss": 0.7475,
      "step": 18184
    },
    {
      "epoch": 5.742027155036312,
      "grad_norm": 4.879164218902588,
      "learning_rate": 3.960548012489914e-05,
      "loss": 0.6068,
      "step": 18185
    },
    {
      "epoch": 5.742342911272497,
      "grad_norm": 3.531724214553833,
      "learning_rate": 3.9604603024243064e-05,
      "loss": 0.5296,
      "step": 18186
    },
    {
      "epoch": 5.7426586675086835,
      "grad_norm": 4.660522937774658,
      "learning_rate": 3.9603725923586996e-05,
      "loss": 0.6016,
      "step": 18187
    },
    {
      "epoch": 5.742974423744869,
      "grad_norm": 3.9720826148986816,
      "learning_rate": 3.960284882293092e-05,
      "loss": 0.5896,
      "step": 18188
    },
    {
      "epoch": 5.743290179981055,
      "grad_norm": 3.8063764572143555,
      "learning_rate": 3.960197172227485e-05,
      "loss": 0.5807,
      "step": 18189
    },
    {
      "epoch": 5.74360593621724,
      "grad_norm": 4.4596076011657715,
      "learning_rate": 3.960109462161878e-05,
      "loss": 0.673,
      "step": 18190
    },
    {
      "epoch": 5.743921692453426,
      "grad_norm": 5.055037498474121,
      "learning_rate": 3.960021752096271e-05,
      "loss": 0.5169,
      "step": 18191
    },
    {
      "epoch": 5.744237448689612,
      "grad_norm": 4.578914642333984,
      "learning_rate": 3.9599340420306634e-05,
      "loss": 0.6893,
      "step": 18192
    },
    {
      "epoch": 5.744553204925797,
      "grad_norm": 4.529940605163574,
      "learning_rate": 3.9598463319650565e-05,
      "loss": 0.6616,
      "step": 18193
    },
    {
      "epoch": 5.744868961161983,
      "grad_norm": 4.0166215896606445,
      "learning_rate": 3.959758621899449e-05,
      "loss": 0.6636,
      "step": 18194
    },
    {
      "epoch": 5.7451847173981685,
      "grad_norm": 4.635612487792969,
      "learning_rate": 3.959670911833842e-05,
      "loss": 0.7068,
      "step": 18195
    },
    {
      "epoch": 5.745500473634355,
      "grad_norm": 3.809089183807373,
      "learning_rate": 3.959583201768235e-05,
      "loss": 0.6065,
      "step": 18196
    },
    {
      "epoch": 5.74581622987054,
      "grad_norm": 3.361077070236206,
      "learning_rate": 3.959495491702628e-05,
      "loss": 0.5722,
      "step": 18197
    },
    {
      "epoch": 5.746131986106725,
      "grad_norm": 3.9462296962738037,
      "learning_rate": 3.9594077816370203e-05,
      "loss": 0.6139,
      "step": 18198
    },
    {
      "epoch": 5.7464477423429114,
      "grad_norm": 3.8928112983703613,
      "learning_rate": 3.9593200715714135e-05,
      "loss": 0.6341,
      "step": 18199
    },
    {
      "epoch": 5.746763498579097,
      "grad_norm": 4.603261947631836,
      "learning_rate": 3.959232361505806e-05,
      "loss": 0.6073,
      "step": 18200
    },
    {
      "epoch": 5.746763498579097,
      "eval_loss": 0.8330137729644775,
      "eval_runtime": 4.551,
      "eval_samples_per_second": 547.127,
      "eval_steps_per_second": 4.614,
      "step": 18200
    },
    {
      "epoch": 5.747079254815283,
      "grad_norm": 3.3232579231262207,
      "learning_rate": 3.959144651440199e-05,
      "loss": 0.6322,
      "step": 18201
    },
    {
      "epoch": 5.747395011051468,
      "grad_norm": 3.5102310180664062,
      "learning_rate": 3.9590569413745923e-05,
      "loss": 0.5721,
      "step": 18202
    },
    {
      "epoch": 5.7477107672876535,
      "grad_norm": 4.112565040588379,
      "learning_rate": 3.9589692313089855e-05,
      "loss": 0.6837,
      "step": 18203
    },
    {
      "epoch": 5.74802652352384,
      "grad_norm": 3.38991641998291,
      "learning_rate": 3.958881521243378e-05,
      "loss": 0.6596,
      "step": 18204
    },
    {
      "epoch": 5.748342279760025,
      "grad_norm": 5.593507289886475,
      "learning_rate": 3.958793811177771e-05,
      "loss": 0.6787,
      "step": 18205
    },
    {
      "epoch": 5.748658035996211,
      "grad_norm": 4.293610095977783,
      "learning_rate": 3.958706101112164e-05,
      "loss": 0.6258,
      "step": 18206
    },
    {
      "epoch": 5.7489737922323965,
      "grad_norm": 4.9554548263549805,
      "learning_rate": 3.958618391046557e-05,
      "loss": 0.7521,
      "step": 18207
    },
    {
      "epoch": 5.749289548468582,
      "grad_norm": 4.325116157531738,
      "learning_rate": 3.95853068098095e-05,
      "loss": 0.5621,
      "step": 18208
    },
    {
      "epoch": 5.749605304704768,
      "grad_norm": 4.682399272918701,
      "learning_rate": 3.9584429709153425e-05,
      "loss": 0.693,
      "step": 18209
    },
    {
      "epoch": 5.749921060940953,
      "grad_norm": 3.987769365310669,
      "learning_rate": 3.9583552608497357e-05,
      "loss": 0.5544,
      "step": 18210
    },
    {
      "epoch": 5.750236817177139,
      "grad_norm": 3.879854202270508,
      "learning_rate": 3.958267550784128e-05,
      "loss": 0.6754,
      "step": 18211
    },
    {
      "epoch": 5.750552573413325,
      "grad_norm": 4.767743110656738,
      "learning_rate": 3.958179840718521e-05,
      "loss": 0.5472,
      "step": 18212
    },
    {
      "epoch": 5.750868329649511,
      "grad_norm": 4.808882713317871,
      "learning_rate": 3.958092130652914e-05,
      "loss": 0.703,
      "step": 18213
    },
    {
      "epoch": 5.751184085885696,
      "grad_norm": 4.986361980438232,
      "learning_rate": 3.958004420587307e-05,
      "loss": 0.6825,
      "step": 18214
    },
    {
      "epoch": 5.751499842121882,
      "grad_norm": 4.439722061157227,
      "learning_rate": 3.9579167105216995e-05,
      "loss": 0.6546,
      "step": 18215
    },
    {
      "epoch": 5.751815598358068,
      "grad_norm": 5.261664390563965,
      "learning_rate": 3.9578290004560926e-05,
      "loss": 0.7169,
      "step": 18216
    },
    {
      "epoch": 5.752131354594253,
      "grad_norm": 4.117123603820801,
      "learning_rate": 3.957741290390485e-05,
      "loss": 0.7084,
      "step": 18217
    },
    {
      "epoch": 5.752447110830439,
      "grad_norm": 4.803516387939453,
      "learning_rate": 3.957653580324878e-05,
      "loss": 0.7104,
      "step": 18218
    },
    {
      "epoch": 5.752762867066624,
      "grad_norm": 4.134155750274658,
      "learning_rate": 3.957565870259271e-05,
      "loss": 0.6384,
      "step": 18219
    },
    {
      "epoch": 5.753078623302811,
      "grad_norm": 5.46153450012207,
      "learning_rate": 3.957478160193664e-05,
      "loss": 0.7231,
      "step": 18220
    },
    {
      "epoch": 5.753394379538996,
      "grad_norm": 3.600588798522949,
      "learning_rate": 3.9573904501280564e-05,
      "loss": 0.6284,
      "step": 18221
    },
    {
      "epoch": 5.753710135775181,
      "grad_norm": 4.025334358215332,
      "learning_rate": 3.9573027400624496e-05,
      "loss": 0.6167,
      "step": 18222
    },
    {
      "epoch": 5.754025892011367,
      "grad_norm": 3.797921657562256,
      "learning_rate": 3.957215029996843e-05,
      "loss": 0.5925,
      "step": 18223
    },
    {
      "epoch": 5.754341648247553,
      "grad_norm": 4.392379283905029,
      "learning_rate": 3.957127319931235e-05,
      "loss": 0.5811,
      "step": 18224
    },
    {
      "epoch": 5.754657404483739,
      "grad_norm": 4.2093610763549805,
      "learning_rate": 3.9570396098656284e-05,
      "loss": 0.5972,
      "step": 18225
    },
    {
      "epoch": 5.754973160719924,
      "grad_norm": 4.389352798461914,
      "learning_rate": 3.956951899800021e-05,
      "loss": 0.6411,
      "step": 18226
    },
    {
      "epoch": 5.7552889169561094,
      "grad_norm": 4.649555683135986,
      "learning_rate": 3.956864189734414e-05,
      "loss": 0.6433,
      "step": 18227
    },
    {
      "epoch": 5.755604673192296,
      "grad_norm": 4.441334247589111,
      "learning_rate": 3.9567764796688066e-05,
      "loss": 0.5591,
      "step": 18228
    },
    {
      "epoch": 5.755920429428481,
      "grad_norm": 5.654735565185547,
      "learning_rate": 3.9566887696032004e-05,
      "loss": 0.6622,
      "step": 18229
    },
    {
      "epoch": 5.756236185664667,
      "grad_norm": 4.171084880828857,
      "learning_rate": 3.956601059537593e-05,
      "loss": 0.7256,
      "step": 18230
    },
    {
      "epoch": 5.756551941900852,
      "grad_norm": 3.91945743560791,
      "learning_rate": 3.956513349471986e-05,
      "loss": 0.5977,
      "step": 18231
    },
    {
      "epoch": 5.756867698137039,
      "grad_norm": 3.965329647064209,
      "learning_rate": 3.9564256394063786e-05,
      "loss": 0.6184,
      "step": 18232
    },
    {
      "epoch": 5.757183454373224,
      "grad_norm": 4.498523235321045,
      "learning_rate": 3.956337929340772e-05,
      "loss": 0.6604,
      "step": 18233
    },
    {
      "epoch": 5.757499210609409,
      "grad_norm": 3.8591794967651367,
      "learning_rate": 3.956250219275164e-05,
      "loss": 0.7016,
      "step": 18234
    },
    {
      "epoch": 5.757814966845595,
      "grad_norm": 3.6903748512268066,
      "learning_rate": 3.9561625092095574e-05,
      "loss": 0.668,
      "step": 18235
    },
    {
      "epoch": 5.758130723081781,
      "grad_norm": 4.507569789886475,
      "learning_rate": 3.95607479914395e-05,
      "loss": 0.6578,
      "step": 18236
    },
    {
      "epoch": 5.758446479317967,
      "grad_norm": 4.041741847991943,
      "learning_rate": 3.955987089078343e-05,
      "loss": 0.6372,
      "step": 18237
    },
    {
      "epoch": 5.758762235554152,
      "grad_norm": 4.741767883300781,
      "learning_rate": 3.9558993790127356e-05,
      "loss": 0.62,
      "step": 18238
    },
    {
      "epoch": 5.759077991790338,
      "grad_norm": 3.9454972743988037,
      "learning_rate": 3.955811668947129e-05,
      "loss": 0.6316,
      "step": 18239
    },
    {
      "epoch": 5.759393748026524,
      "grad_norm": 3.9239063262939453,
      "learning_rate": 3.955723958881521e-05,
      "loss": 0.6646,
      "step": 18240
    },
    {
      "epoch": 5.759709504262709,
      "grad_norm": 4.1037421226501465,
      "learning_rate": 3.9556362488159144e-05,
      "loss": 0.5795,
      "step": 18241
    },
    {
      "epoch": 5.760025260498895,
      "grad_norm": 3.926901340484619,
      "learning_rate": 3.955548538750307e-05,
      "loss": 0.5948,
      "step": 18242
    },
    {
      "epoch": 5.76034101673508,
      "grad_norm": 4.276591777801514,
      "learning_rate": 3.9554608286847e-05,
      "loss": 0.5975,
      "step": 18243
    },
    {
      "epoch": 5.7606567729712665,
      "grad_norm": 4.089577674865723,
      "learning_rate": 3.955373118619093e-05,
      "loss": 0.6296,
      "step": 18244
    },
    {
      "epoch": 5.760972529207452,
      "grad_norm": 3.9713315963745117,
      "learning_rate": 3.955285408553486e-05,
      "loss": 0.6475,
      "step": 18245
    },
    {
      "epoch": 5.761288285443637,
      "grad_norm": 4.316334247589111,
      "learning_rate": 3.955197698487879e-05,
      "loss": 0.6846,
      "step": 18246
    },
    {
      "epoch": 5.761604041679823,
      "grad_norm": 4.232213497161865,
      "learning_rate": 3.9551099884222714e-05,
      "loss": 0.613,
      "step": 18247
    },
    {
      "epoch": 5.761919797916009,
      "grad_norm": 3.795762062072754,
      "learning_rate": 3.9550222783566645e-05,
      "loss": 0.6724,
      "step": 18248
    },
    {
      "epoch": 5.762235554152195,
      "grad_norm": 4.2340803146362305,
      "learning_rate": 3.954934568291057e-05,
      "loss": 0.6788,
      "step": 18249
    },
    {
      "epoch": 5.76255131038838,
      "grad_norm": 4.467743396759033,
      "learning_rate": 3.95484685822545e-05,
      "loss": 0.6466,
      "step": 18250
    },
    {
      "epoch": 5.762867066624565,
      "grad_norm": 4.003578186035156,
      "learning_rate": 3.954759148159843e-05,
      "loss": 0.6645,
      "step": 18251
    },
    {
      "epoch": 5.763182822860752,
      "grad_norm": 3.9237430095672607,
      "learning_rate": 3.954671438094236e-05,
      "loss": 0.6413,
      "step": 18252
    },
    {
      "epoch": 5.763498579096937,
      "grad_norm": 3.9097583293914795,
      "learning_rate": 3.954583728028628e-05,
      "loss": 0.649,
      "step": 18253
    },
    {
      "epoch": 5.763814335333123,
      "grad_norm": 4.699062824249268,
      "learning_rate": 3.9544960179630215e-05,
      "loss": 0.68,
      "step": 18254
    },
    {
      "epoch": 5.764130091569308,
      "grad_norm": 4.681240558624268,
      "learning_rate": 3.954408307897414e-05,
      "loss": 0.6894,
      "step": 18255
    },
    {
      "epoch": 5.7644458478054945,
      "grad_norm": 4.2081732749938965,
      "learning_rate": 3.954320597831808e-05,
      "loss": 0.7028,
      "step": 18256
    },
    {
      "epoch": 5.76476160404168,
      "grad_norm": 3.4076614379882812,
      "learning_rate": 3.9542328877662e-05,
      "loss": 0.6134,
      "step": 18257
    },
    {
      "epoch": 5.765077360277865,
      "grad_norm": 4.254823207855225,
      "learning_rate": 3.9541451777005935e-05,
      "loss": 0.6454,
      "step": 18258
    },
    {
      "epoch": 5.765393116514051,
      "grad_norm": 4.308114528656006,
      "learning_rate": 3.954057467634986e-05,
      "loss": 0.6409,
      "step": 18259
    },
    {
      "epoch": 5.765708872750237,
      "grad_norm": 3.7715961933135986,
      "learning_rate": 3.953969757569379e-05,
      "loss": 0.6074,
      "step": 18260
    },
    {
      "epoch": 5.766024628986423,
      "grad_norm": 3.9140069484710693,
      "learning_rate": 3.9538820475037716e-05,
      "loss": 0.6725,
      "step": 18261
    },
    {
      "epoch": 5.766340385222608,
      "grad_norm": 3.8750524520874023,
      "learning_rate": 3.953794337438165e-05,
      "loss": 0.6408,
      "step": 18262
    },
    {
      "epoch": 5.766656141458794,
      "grad_norm": 3.959057092666626,
      "learning_rate": 3.953706627372557e-05,
      "loss": 0.6284,
      "step": 18263
    },
    {
      "epoch": 5.7669718976949795,
      "grad_norm": 5.323480129241943,
      "learning_rate": 3.9536189173069505e-05,
      "loss": 0.626,
      "step": 18264
    },
    {
      "epoch": 5.767287653931165,
      "grad_norm": 5.0009660720825195,
      "learning_rate": 3.953531207241343e-05,
      "loss": 0.5943,
      "step": 18265
    },
    {
      "epoch": 5.767603410167351,
      "grad_norm": 3.628535270690918,
      "learning_rate": 3.953443497175736e-05,
      "loss": 0.6527,
      "step": 18266
    },
    {
      "epoch": 5.767919166403536,
      "grad_norm": 3.8284854888916016,
      "learning_rate": 3.953355787110129e-05,
      "loss": 0.6348,
      "step": 18267
    },
    {
      "epoch": 5.7682349226397225,
      "grad_norm": 4.451111316680908,
      "learning_rate": 3.953268077044522e-05,
      "loss": 0.6823,
      "step": 18268
    },
    {
      "epoch": 5.768550678875908,
      "grad_norm": 3.834580421447754,
      "learning_rate": 3.953180366978915e-05,
      "loss": 0.5698,
      "step": 18269
    },
    {
      "epoch": 5.768866435112093,
      "grad_norm": 3.963613748550415,
      "learning_rate": 3.9530926569133075e-05,
      "loss": 0.6202,
      "step": 18270
    },
    {
      "epoch": 5.769182191348279,
      "grad_norm": 3.6521167755126953,
      "learning_rate": 3.9530049468477006e-05,
      "loss": 0.6079,
      "step": 18271
    },
    {
      "epoch": 5.7694979475844645,
      "grad_norm": 4.404665946960449,
      "learning_rate": 3.952917236782093e-05,
      "loss": 0.6748,
      "step": 18272
    },
    {
      "epoch": 5.769813703820651,
      "grad_norm": 4.3748064041137695,
      "learning_rate": 3.952829526716486e-05,
      "loss": 0.6499,
      "step": 18273
    },
    {
      "epoch": 5.770129460056836,
      "grad_norm": 2.9991793632507324,
      "learning_rate": 3.952741816650879e-05,
      "loss": 0.5172,
      "step": 18274
    },
    {
      "epoch": 5.770445216293021,
      "grad_norm": 3.5468366146087646,
      "learning_rate": 3.952654106585272e-05,
      "loss": 0.6125,
      "step": 18275
    },
    {
      "epoch": 5.7707609725292075,
      "grad_norm": 4.411474227905273,
      "learning_rate": 3.9525663965196644e-05,
      "loss": 0.7468,
      "step": 18276
    },
    {
      "epoch": 5.771076728765393,
      "grad_norm": 4.113649845123291,
      "learning_rate": 3.9524786864540576e-05,
      "loss": 0.5877,
      "step": 18277
    },
    {
      "epoch": 5.771392485001579,
      "grad_norm": 3.9334542751312256,
      "learning_rate": 3.95239097638845e-05,
      "loss": 0.6386,
      "step": 18278
    },
    {
      "epoch": 5.771708241237764,
      "grad_norm": 5.415665149688721,
      "learning_rate": 3.952303266322843e-05,
      "loss": 0.684,
      "step": 18279
    },
    {
      "epoch": 5.7720239974739505,
      "grad_norm": 4.174541473388672,
      "learning_rate": 3.952215556257236e-05,
      "loss": 0.7027,
      "step": 18280
    },
    {
      "epoch": 5.772339753710136,
      "grad_norm": 4.366758346557617,
      "learning_rate": 3.952127846191629e-05,
      "loss": 0.7304,
      "step": 18281
    },
    {
      "epoch": 5.772655509946322,
      "grad_norm": 4.285162448883057,
      "learning_rate": 3.952040136126022e-05,
      "loss": 0.7605,
      "step": 18282
    },
    {
      "epoch": 5.772971266182507,
      "grad_norm": 4.401092529296875,
      "learning_rate": 3.951952426060415e-05,
      "loss": 0.6992,
      "step": 18283
    },
    {
      "epoch": 5.7732870224186925,
      "grad_norm": 4.71527624130249,
      "learning_rate": 3.951864715994808e-05,
      "loss": 0.6489,
      "step": 18284
    },
    {
      "epoch": 5.773602778654879,
      "grad_norm": 3.883549928665161,
      "learning_rate": 3.951777005929201e-05,
      "loss": 0.6363,
      "step": 18285
    },
    {
      "epoch": 5.773918534891064,
      "grad_norm": 4.1694016456604,
      "learning_rate": 3.9516892958635934e-05,
      "loss": 0.6379,
      "step": 18286
    },
    {
      "epoch": 5.77423429112725,
      "grad_norm": 4.629714488983154,
      "learning_rate": 3.9516015857979866e-05,
      "loss": 0.5593,
      "step": 18287
    },
    {
      "epoch": 5.7745500473634355,
      "grad_norm": 4.019225597381592,
      "learning_rate": 3.95151387573238e-05,
      "loss": 0.649,
      "step": 18288
    },
    {
      "epoch": 5.774865803599621,
      "grad_norm": 4.120412349700928,
      "learning_rate": 3.951426165666772e-05,
      "loss": 0.6418,
      "step": 18289
    },
    {
      "epoch": 5.775181559835807,
      "grad_norm": 4.735018730163574,
      "learning_rate": 3.9513384556011654e-05,
      "loss": 0.7785,
      "step": 18290
    },
    {
      "epoch": 5.775497316071992,
      "grad_norm": 5.892120838165283,
      "learning_rate": 3.951250745535558e-05,
      "loss": 0.7422,
      "step": 18291
    },
    {
      "epoch": 5.775813072308178,
      "grad_norm": 5.003594398498535,
      "learning_rate": 3.951163035469951e-05,
      "loss": 0.7032,
      "step": 18292
    },
    {
      "epoch": 5.776128828544364,
      "grad_norm": 3.538520574569702,
      "learning_rate": 3.9510753254043435e-05,
      "loss": 0.6024,
      "step": 18293
    },
    {
      "epoch": 5.776444584780549,
      "grad_norm": 4.529248237609863,
      "learning_rate": 3.950987615338737e-05,
      "loss": 0.7197,
      "step": 18294
    },
    {
      "epoch": 5.776760341016735,
      "grad_norm": 4.145736217498779,
      "learning_rate": 3.950899905273129e-05,
      "loss": 0.6007,
      "step": 18295
    },
    {
      "epoch": 5.7770760972529205,
      "grad_norm": 4.746877193450928,
      "learning_rate": 3.9508121952075224e-05,
      "loss": 0.7402,
      "step": 18296
    },
    {
      "epoch": 5.777391853489107,
      "grad_norm": 4.901885032653809,
      "learning_rate": 3.950724485141915e-05,
      "loss": 0.6364,
      "step": 18297
    },
    {
      "epoch": 5.777707609725292,
      "grad_norm": 4.180107593536377,
      "learning_rate": 3.950636775076308e-05,
      "loss": 0.6924,
      "step": 18298
    },
    {
      "epoch": 5.778023365961477,
      "grad_norm": 4.6187334060668945,
      "learning_rate": 3.9505490650107005e-05,
      "loss": 0.6073,
      "step": 18299
    },
    {
      "epoch": 5.778339122197663,
      "grad_norm": 4.61751127243042,
      "learning_rate": 3.950461354945094e-05,
      "loss": 0.6366,
      "step": 18300
    },
    {
      "epoch": 5.778654878433849,
      "grad_norm": 3.6768503189086914,
      "learning_rate": 3.950373644879486e-05,
      "loss": 0.5773,
      "step": 18301
    },
    {
      "epoch": 5.778970634670035,
      "grad_norm": 5.525415420532227,
      "learning_rate": 3.9502859348138793e-05,
      "loss": 0.7349,
      "step": 18302
    },
    {
      "epoch": 5.77928639090622,
      "grad_norm": 4.230463027954102,
      "learning_rate": 3.950198224748272e-05,
      "loss": 0.6813,
      "step": 18303
    },
    {
      "epoch": 5.779602147142406,
      "grad_norm": 4.657773494720459,
      "learning_rate": 3.950110514682665e-05,
      "loss": 0.6046,
      "step": 18304
    },
    {
      "epoch": 5.779917903378592,
      "grad_norm": 4.310406684875488,
      "learning_rate": 3.950022804617058e-05,
      "loss": 0.7193,
      "step": 18305
    },
    {
      "epoch": 5.780233659614778,
      "grad_norm": 4.233059406280518,
      "learning_rate": 3.949935094551451e-05,
      "loss": 0.6549,
      "step": 18306
    },
    {
      "epoch": 5.780549415850963,
      "grad_norm": 4.145305633544922,
      "learning_rate": 3.949847384485844e-05,
      "loss": 0.6118,
      "step": 18307
    },
    {
      "epoch": 5.7808651720871485,
      "grad_norm": 4.2327985763549805,
      "learning_rate": 3.949759674420236e-05,
      "loss": 0.7094,
      "step": 18308
    },
    {
      "epoch": 5.781180928323335,
      "grad_norm": 4.418436527252197,
      "learning_rate": 3.9496719643546295e-05,
      "loss": 0.6908,
      "step": 18309
    },
    {
      "epoch": 5.78149668455952,
      "grad_norm": 3.8430511951446533,
      "learning_rate": 3.949584254289022e-05,
      "loss": 0.5873,
      "step": 18310
    },
    {
      "epoch": 5.781812440795706,
      "grad_norm": 3.7083652019500732,
      "learning_rate": 3.949496544223416e-05,
      "loss": 0.5883,
      "step": 18311
    },
    {
      "epoch": 5.782128197031891,
      "grad_norm": 5.397035598754883,
      "learning_rate": 3.949408834157808e-05,
      "loss": 0.7248,
      "step": 18312
    },
    {
      "epoch": 5.782443953268077,
      "grad_norm": 4.277818202972412,
      "learning_rate": 3.9493211240922015e-05,
      "loss": 0.6266,
      "step": 18313
    },
    {
      "epoch": 5.782759709504263,
      "grad_norm": 3.7097952365875244,
      "learning_rate": 3.949233414026594e-05,
      "loss": 0.6794,
      "step": 18314
    },
    {
      "epoch": 5.783075465740448,
      "grad_norm": 4.3787641525268555,
      "learning_rate": 3.949145703960987e-05,
      "loss": 0.5535,
      "step": 18315
    },
    {
      "epoch": 5.783391221976634,
      "grad_norm": 4.351577281951904,
      "learning_rate": 3.9490579938953796e-05,
      "loss": 0.6573,
      "step": 18316
    },
    {
      "epoch": 5.78370697821282,
      "grad_norm": 3.8903591632843018,
      "learning_rate": 3.948970283829773e-05,
      "loss": 0.6222,
      "step": 18317
    },
    {
      "epoch": 5.784022734449005,
      "grad_norm": 3.9594593048095703,
      "learning_rate": 3.948882573764165e-05,
      "loss": 0.6307,
      "step": 18318
    },
    {
      "epoch": 5.784338490685191,
      "grad_norm": 4.064728736877441,
      "learning_rate": 3.9487948636985585e-05,
      "loss": 0.7053,
      "step": 18319
    },
    {
      "epoch": 5.784654246921376,
      "grad_norm": 3.374575614929199,
      "learning_rate": 3.948707153632951e-05,
      "loss": 0.6086,
      "step": 18320
    },
    {
      "epoch": 5.784970003157563,
      "grad_norm": 4.192252159118652,
      "learning_rate": 3.948619443567344e-05,
      "loss": 0.7372,
      "step": 18321
    },
    {
      "epoch": 5.785285759393748,
      "grad_norm": 4.296358585357666,
      "learning_rate": 3.9485317335017366e-05,
      "loss": 0.5169,
      "step": 18322
    },
    {
      "epoch": 5.785601515629934,
      "grad_norm": 3.9654922485351562,
      "learning_rate": 3.94844402343613e-05,
      "loss": 0.704,
      "step": 18323
    },
    {
      "epoch": 5.785917271866119,
      "grad_norm": 4.18842077255249,
      "learning_rate": 3.948356313370522e-05,
      "loss": 0.7248,
      "step": 18324
    },
    {
      "epoch": 5.786233028102305,
      "grad_norm": 4.413466930389404,
      "learning_rate": 3.9482686033049154e-05,
      "loss": 0.6621,
      "step": 18325
    },
    {
      "epoch": 5.786548784338491,
      "grad_norm": 4.661306858062744,
      "learning_rate": 3.9481808932393086e-05,
      "loss": 0.6527,
      "step": 18326
    },
    {
      "epoch": 5.786864540574676,
      "grad_norm": 3.9455184936523438,
      "learning_rate": 3.948093183173701e-05,
      "loss": 0.618,
      "step": 18327
    },
    {
      "epoch": 5.787180296810862,
      "grad_norm": 4.790546894073486,
      "learning_rate": 3.948005473108094e-05,
      "loss": 0.6813,
      "step": 18328
    },
    {
      "epoch": 5.787496053047048,
      "grad_norm": 3.6417617797851562,
      "learning_rate": 3.947917763042487e-05,
      "loss": 0.6332,
      "step": 18329
    },
    {
      "epoch": 5.787811809283234,
      "grad_norm": 3.6173043251037598,
      "learning_rate": 3.94783005297688e-05,
      "loss": 0.6293,
      "step": 18330
    },
    {
      "epoch": 5.788127565519419,
      "grad_norm": 3.931157350540161,
      "learning_rate": 3.9477423429112724e-05,
      "loss": 0.5861,
      "step": 18331
    },
    {
      "epoch": 5.788443321755604,
      "grad_norm": 4.818953514099121,
      "learning_rate": 3.9476546328456656e-05,
      "loss": 0.6358,
      "step": 18332
    },
    {
      "epoch": 5.788759077991791,
      "grad_norm": 5.919155120849609,
      "learning_rate": 3.947566922780058e-05,
      "loss": 0.8126,
      "step": 18333
    },
    {
      "epoch": 5.789074834227976,
      "grad_norm": 4.024837493896484,
      "learning_rate": 3.947479212714451e-05,
      "loss": 0.679,
      "step": 18334
    },
    {
      "epoch": 5.789390590464162,
      "grad_norm": 3.6558046340942383,
      "learning_rate": 3.947391502648844e-05,
      "loss": 0.622,
      "step": 18335
    },
    {
      "epoch": 5.789706346700347,
      "grad_norm": 3.7238152027130127,
      "learning_rate": 3.947303792583237e-05,
      "loss": 0.6302,
      "step": 18336
    },
    {
      "epoch": 5.790022102936533,
      "grad_norm": 4.662126541137695,
      "learning_rate": 3.9472160825176294e-05,
      "loss": 0.6391,
      "step": 18337
    },
    {
      "epoch": 5.790337859172719,
      "grad_norm": 3.391965627670288,
      "learning_rate": 3.947128372452023e-05,
      "loss": 0.7282,
      "step": 18338
    },
    {
      "epoch": 5.790653615408904,
      "grad_norm": 3.869534492492676,
      "learning_rate": 3.947040662386416e-05,
      "loss": 0.6389,
      "step": 18339
    },
    {
      "epoch": 5.79096937164509,
      "grad_norm": 4.176754474639893,
      "learning_rate": 3.946952952320809e-05,
      "loss": 0.6051,
      "step": 18340
    },
    {
      "epoch": 5.791285127881276,
      "grad_norm": 4.317838191986084,
      "learning_rate": 3.9468652422552014e-05,
      "loss": 0.5394,
      "step": 18341
    },
    {
      "epoch": 5.791600884117461,
      "grad_norm": 4.493302345275879,
      "learning_rate": 3.9467775321895946e-05,
      "loss": 0.6823,
      "step": 18342
    },
    {
      "epoch": 5.791916640353647,
      "grad_norm": 4.163460731506348,
      "learning_rate": 3.946689822123987e-05,
      "loss": 0.6681,
      "step": 18343
    },
    {
      "epoch": 5.792232396589832,
      "grad_norm": 3.750497579574585,
      "learning_rate": 3.94660211205838e-05,
      "loss": 0.6379,
      "step": 18344
    },
    {
      "epoch": 5.7925481528260185,
      "grad_norm": 4.73095703125,
      "learning_rate": 3.946514401992773e-05,
      "loss": 0.6769,
      "step": 18345
    },
    {
      "epoch": 5.792863909062204,
      "grad_norm": 4.200215816497803,
      "learning_rate": 3.946426691927166e-05,
      "loss": 0.5557,
      "step": 18346
    },
    {
      "epoch": 5.79317966529839,
      "grad_norm": 3.5673813819885254,
      "learning_rate": 3.9463389818615584e-05,
      "loss": 0.534,
      "step": 18347
    },
    {
      "epoch": 5.793495421534575,
      "grad_norm": 4.083562850952148,
      "learning_rate": 3.9462512717959515e-05,
      "loss": 0.5304,
      "step": 18348
    },
    {
      "epoch": 5.793811177770761,
      "grad_norm": 3.737039566040039,
      "learning_rate": 3.946163561730345e-05,
      "loss": 0.6481,
      "step": 18349
    },
    {
      "epoch": 5.794126934006947,
      "grad_norm": 3.442915201187134,
      "learning_rate": 3.946075851664737e-05,
      "loss": 0.6171,
      "step": 18350
    },
    {
      "epoch": 5.794442690243132,
      "grad_norm": 3.907932758331299,
      "learning_rate": 3.9459881415991304e-05,
      "loss": 0.6004,
      "step": 18351
    },
    {
      "epoch": 5.794758446479318,
      "grad_norm": 4.132491111755371,
      "learning_rate": 3.945900431533523e-05,
      "loss": 0.6849,
      "step": 18352
    },
    {
      "epoch": 5.7950742027155036,
      "grad_norm": 5.3257951736450195,
      "learning_rate": 3.945812721467916e-05,
      "loss": 0.7129,
      "step": 18353
    },
    {
      "epoch": 5.79538995895169,
      "grad_norm": 3.803898334503174,
      "learning_rate": 3.9457250114023085e-05,
      "loss": 0.5314,
      "step": 18354
    },
    {
      "epoch": 5.795705715187875,
      "grad_norm": 4.546090126037598,
      "learning_rate": 3.945637301336702e-05,
      "loss": 0.6547,
      "step": 18355
    },
    {
      "epoch": 5.79602147142406,
      "grad_norm": 3.7796902656555176,
      "learning_rate": 3.945549591271094e-05,
      "loss": 0.5446,
      "step": 18356
    },
    {
      "epoch": 5.7963372276602465,
      "grad_norm": 4.955658435821533,
      "learning_rate": 3.945461881205487e-05,
      "loss": 0.6766,
      "step": 18357
    },
    {
      "epoch": 5.796652983896432,
      "grad_norm": 4.413862705230713,
      "learning_rate": 3.94537417113988e-05,
      "loss": 0.5431,
      "step": 18358
    },
    {
      "epoch": 5.796968740132618,
      "grad_norm": 4.59527587890625,
      "learning_rate": 3.945286461074273e-05,
      "loss": 0.5864,
      "step": 18359
    },
    {
      "epoch": 5.797284496368803,
      "grad_norm": 5.163888454437256,
      "learning_rate": 3.9451987510086655e-05,
      "loss": 0.5794,
      "step": 18360
    },
    {
      "epoch": 5.797600252604989,
      "grad_norm": 4.22941780090332,
      "learning_rate": 3.9451110409430587e-05,
      "loss": 0.6347,
      "step": 18361
    },
    {
      "epoch": 5.797916008841175,
      "grad_norm": 5.412902355194092,
      "learning_rate": 3.945023330877451e-05,
      "loss": 0.6784,
      "step": 18362
    },
    {
      "epoch": 5.79823176507736,
      "grad_norm": 4.332181453704834,
      "learning_rate": 3.944935620811844e-05,
      "loss": 0.5338,
      "step": 18363
    },
    {
      "epoch": 5.798547521313546,
      "grad_norm": 4.5424346923828125,
      "learning_rate": 3.9448479107462375e-05,
      "loss": 0.6741,
      "step": 18364
    },
    {
      "epoch": 5.7988632775497315,
      "grad_norm": 4.372955799102783,
      "learning_rate": 3.9447602006806307e-05,
      "loss": 0.653,
      "step": 18365
    },
    {
      "epoch": 5.799179033785917,
      "grad_norm": 4.199330806732178,
      "learning_rate": 3.944672490615023e-05,
      "loss": 0.6537,
      "step": 18366
    },
    {
      "epoch": 5.799494790022103,
      "grad_norm": 4.162386417388916,
      "learning_rate": 3.944584780549416e-05,
      "loss": 0.5815,
      "step": 18367
    },
    {
      "epoch": 5.799810546258288,
      "grad_norm": 3.6063196659088135,
      "learning_rate": 3.944497070483809e-05,
      "loss": 0.5895,
      "step": 18368
    },
    {
      "epoch": 5.8001263024944745,
      "grad_norm": 5.140671730041504,
      "learning_rate": 3.944409360418202e-05,
      "loss": 0.6427,
      "step": 18369
    },
    {
      "epoch": 5.80044205873066,
      "grad_norm": 4.596963405609131,
      "learning_rate": 3.944321650352595e-05,
      "loss": 0.5516,
      "step": 18370
    },
    {
      "epoch": 5.800757814966846,
      "grad_norm": 3.7571678161621094,
      "learning_rate": 3.9442339402869876e-05,
      "loss": 0.6455,
      "step": 18371
    },
    {
      "epoch": 5.801073571203031,
      "grad_norm": 3.9740922451019287,
      "learning_rate": 3.944146230221381e-05,
      "loss": 0.5872,
      "step": 18372
    },
    {
      "epoch": 5.8013893274392165,
      "grad_norm": 4.585681915283203,
      "learning_rate": 3.944058520155773e-05,
      "loss": 0.6061,
      "step": 18373
    },
    {
      "epoch": 5.801705083675403,
      "grad_norm": 4.780972480773926,
      "learning_rate": 3.9439708100901665e-05,
      "loss": 0.6633,
      "step": 18374
    },
    {
      "epoch": 5.802020839911588,
      "grad_norm": 4.73013973236084,
      "learning_rate": 3.943883100024559e-05,
      "loss": 0.585,
      "step": 18375
    },
    {
      "epoch": 5.802336596147774,
      "grad_norm": 4.4595842361450195,
      "learning_rate": 3.943795389958952e-05,
      "loss": 0.606,
      "step": 18376
    },
    {
      "epoch": 5.8026523523839595,
      "grad_norm": 4.186822891235352,
      "learning_rate": 3.9437076798933446e-05,
      "loss": 0.6813,
      "step": 18377
    },
    {
      "epoch": 5.802968108620146,
      "grad_norm": 3.7136707305908203,
      "learning_rate": 3.943619969827738e-05,
      "loss": 0.6087,
      "step": 18378
    },
    {
      "epoch": 5.803283864856331,
      "grad_norm": 4.431448459625244,
      "learning_rate": 3.94353225976213e-05,
      "loss": 0.5825,
      "step": 18379
    },
    {
      "epoch": 5.803599621092516,
      "grad_norm": 3.6817781925201416,
      "learning_rate": 3.9434445496965234e-05,
      "loss": 0.6864,
      "step": 18380
    },
    {
      "epoch": 5.803915377328702,
      "grad_norm": 5.26390266418457,
      "learning_rate": 3.943356839630916e-05,
      "loss": 0.7167,
      "step": 18381
    },
    {
      "epoch": 5.804231133564888,
      "grad_norm": 4.222229480743408,
      "learning_rate": 3.943269129565309e-05,
      "loss": 0.7029,
      "step": 18382
    },
    {
      "epoch": 5.804546889801074,
      "grad_norm": 4.729080677032471,
      "learning_rate": 3.9431814194997016e-05,
      "loss": 0.6445,
      "step": 18383
    },
    {
      "epoch": 5.804862646037259,
      "grad_norm": 3.7195825576782227,
      "learning_rate": 3.943093709434095e-05,
      "loss": 0.6257,
      "step": 18384
    },
    {
      "epoch": 5.8051784022734445,
      "grad_norm": 4.584759712219238,
      "learning_rate": 3.943005999368488e-05,
      "loss": 0.6567,
      "step": 18385
    },
    {
      "epoch": 5.805494158509631,
      "grad_norm": 4.730149269104004,
      "learning_rate": 3.9429182893028804e-05,
      "loss": 0.6382,
      "step": 18386
    },
    {
      "epoch": 5.805809914745816,
      "grad_norm": 3.92403244972229,
      "learning_rate": 3.9428305792372736e-05,
      "loss": 0.5979,
      "step": 18387
    },
    {
      "epoch": 5.806125670982002,
      "grad_norm": 4.45813512802124,
      "learning_rate": 3.942742869171666e-05,
      "loss": 0.6388,
      "step": 18388
    },
    {
      "epoch": 5.8064414272181875,
      "grad_norm": 4.937664031982422,
      "learning_rate": 3.942655159106059e-05,
      "loss": 0.651,
      "step": 18389
    },
    {
      "epoch": 5.806757183454373,
      "grad_norm": 3.7421584129333496,
      "learning_rate": 3.942567449040452e-05,
      "loss": 0.711,
      "step": 18390
    },
    {
      "epoch": 5.807072939690559,
      "grad_norm": 4.462869644165039,
      "learning_rate": 3.942479738974845e-05,
      "loss": 0.6931,
      "step": 18391
    },
    {
      "epoch": 5.807388695926744,
      "grad_norm": 4.163000106811523,
      "learning_rate": 3.942392028909238e-05,
      "loss": 0.6268,
      "step": 18392
    },
    {
      "epoch": 5.80770445216293,
      "grad_norm": 3.98396897315979,
      "learning_rate": 3.942304318843631e-05,
      "loss": 0.6061,
      "step": 18393
    },
    {
      "epoch": 5.808020208399116,
      "grad_norm": 5.1768903732299805,
      "learning_rate": 3.942216608778024e-05,
      "loss": 0.6194,
      "step": 18394
    },
    {
      "epoch": 5.808335964635302,
      "grad_norm": 3.922752618789673,
      "learning_rate": 3.942128898712417e-05,
      "loss": 0.5979,
      "step": 18395
    },
    {
      "epoch": 5.808651720871487,
      "grad_norm": 3.2911739349365234,
      "learning_rate": 3.9420411886468094e-05,
      "loss": 0.5906,
      "step": 18396
    },
    {
      "epoch": 5.808967477107673,
      "grad_norm": 3.1450650691986084,
      "learning_rate": 3.9419534785812025e-05,
      "loss": 0.6262,
      "step": 18397
    },
    {
      "epoch": 5.809283233343859,
      "grad_norm": 4.546570777893066,
      "learning_rate": 3.941865768515595e-05,
      "loss": 0.6426,
      "step": 18398
    },
    {
      "epoch": 5.809598989580044,
      "grad_norm": 4.0982537269592285,
      "learning_rate": 3.941778058449988e-05,
      "loss": 0.5875,
      "step": 18399
    },
    {
      "epoch": 5.80991474581623,
      "grad_norm": 3.4194302558898926,
      "learning_rate": 3.941690348384381e-05,
      "loss": 0.6502,
      "step": 18400
    },
    {
      "epoch": 5.80991474581623,
      "eval_loss": 0.8325865864753723,
      "eval_runtime": 4.5652,
      "eval_samples_per_second": 545.435,
      "eval_steps_per_second": 4.6,
      "step": 18400
    },
    {
      "epoch": 5.810230502052415,
      "grad_norm": 4.002169609069824,
      "learning_rate": 3.941602638318774e-05,
      "loss": 0.5703,
      "step": 18401
    },
    {
      "epoch": 5.810546258288602,
      "grad_norm": 4.760581016540527,
      "learning_rate": 3.9415149282531664e-05,
      "loss": 0.5968,
      "step": 18402
    },
    {
      "epoch": 5.810862014524787,
      "grad_norm": 4.022073745727539,
      "learning_rate": 3.9414272181875595e-05,
      "loss": 0.7128,
      "step": 18403
    },
    {
      "epoch": 5.811177770760972,
      "grad_norm": 4.565864562988281,
      "learning_rate": 3.941339508121952e-05,
      "loss": 0.6719,
      "step": 18404
    },
    {
      "epoch": 5.811493526997158,
      "grad_norm": 3.967775583267212,
      "learning_rate": 3.941251798056345e-05,
      "loss": 0.5709,
      "step": 18405
    },
    {
      "epoch": 5.811809283233344,
      "grad_norm": 4.0446085929870605,
      "learning_rate": 3.941164087990738e-05,
      "loss": 0.6817,
      "step": 18406
    },
    {
      "epoch": 5.81212503946953,
      "grad_norm": 3.4244792461395264,
      "learning_rate": 3.941076377925131e-05,
      "loss": 0.6171,
      "step": 18407
    },
    {
      "epoch": 5.812440795705715,
      "grad_norm": 3.3710460662841797,
      "learning_rate": 3.940988667859524e-05,
      "loss": 0.5576,
      "step": 18408
    },
    {
      "epoch": 5.8127565519419,
      "grad_norm": 4.254486083984375,
      "learning_rate": 3.9409009577939165e-05,
      "loss": 0.5904,
      "step": 18409
    },
    {
      "epoch": 5.813072308178087,
      "grad_norm": 4.338418006896973,
      "learning_rate": 3.94081324772831e-05,
      "loss": 0.6751,
      "step": 18410
    },
    {
      "epoch": 5.813388064414272,
      "grad_norm": 3.93442702293396,
      "learning_rate": 3.940725537662702e-05,
      "loss": 0.6078,
      "step": 18411
    },
    {
      "epoch": 5.813703820650458,
      "grad_norm": 4.715767860412598,
      "learning_rate": 3.940637827597095e-05,
      "loss": 0.7351,
      "step": 18412
    },
    {
      "epoch": 5.814019576886643,
      "grad_norm": 4.177279472351074,
      "learning_rate": 3.940550117531488e-05,
      "loss": 0.7006,
      "step": 18413
    },
    {
      "epoch": 5.814335333122829,
      "grad_norm": 4.2663140296936035,
      "learning_rate": 3.940462407465881e-05,
      "loss": 0.6723,
      "step": 18414
    },
    {
      "epoch": 5.814651089359015,
      "grad_norm": 4.1032586097717285,
      "learning_rate": 3.9403746974002735e-05,
      "loss": 0.6301,
      "step": 18415
    },
    {
      "epoch": 5.8149668455952,
      "grad_norm": 3.901961326599121,
      "learning_rate": 3.9402869873346666e-05,
      "loss": 0.6245,
      "step": 18416
    },
    {
      "epoch": 5.815282601831386,
      "grad_norm": 3.749701738357544,
      "learning_rate": 3.940199277269059e-05,
      "loss": 0.6354,
      "step": 18417
    },
    {
      "epoch": 5.815598358067572,
      "grad_norm": 3.944575786590576,
      "learning_rate": 3.940111567203452e-05,
      "loss": 0.5656,
      "step": 18418
    },
    {
      "epoch": 5.815914114303758,
      "grad_norm": 4.028590202331543,
      "learning_rate": 3.940023857137845e-05,
      "loss": 0.6772,
      "step": 18419
    },
    {
      "epoch": 5.816229870539943,
      "grad_norm": 4.560065746307373,
      "learning_rate": 3.9399361470722386e-05,
      "loss": 0.662,
      "step": 18420
    },
    {
      "epoch": 5.816545626776129,
      "grad_norm": 4.4679856300354,
      "learning_rate": 3.939848437006631e-05,
      "loss": 0.559,
      "step": 18421
    },
    {
      "epoch": 5.816861383012315,
      "grad_norm": 4.451601982116699,
      "learning_rate": 3.939760726941024e-05,
      "loss": 0.6439,
      "step": 18422
    },
    {
      "epoch": 5.8171771392485,
      "grad_norm": 4.2714033126831055,
      "learning_rate": 3.939673016875417e-05,
      "loss": 0.6494,
      "step": 18423
    },
    {
      "epoch": 5.817492895484686,
      "grad_norm": 3.7965564727783203,
      "learning_rate": 3.93958530680981e-05,
      "loss": 0.586,
      "step": 18424
    },
    {
      "epoch": 5.817808651720871,
      "grad_norm": 3.4124481678009033,
      "learning_rate": 3.9394975967442024e-05,
      "loss": 0.6528,
      "step": 18425
    },
    {
      "epoch": 5.8181244079570575,
      "grad_norm": 4.233684539794922,
      "learning_rate": 3.9394098866785956e-05,
      "loss": 0.7278,
      "step": 18426
    },
    {
      "epoch": 5.818440164193243,
      "grad_norm": 4.542124271392822,
      "learning_rate": 3.939322176612988e-05,
      "loss": 0.5834,
      "step": 18427
    },
    {
      "epoch": 5.818755920429428,
      "grad_norm": 3.3852643966674805,
      "learning_rate": 3.939234466547381e-05,
      "loss": 0.6066,
      "step": 18428
    },
    {
      "epoch": 5.819071676665614,
      "grad_norm": 4.091649055480957,
      "learning_rate": 3.9391467564817744e-05,
      "loss": 0.7142,
      "step": 18429
    },
    {
      "epoch": 5.8193874329018,
      "grad_norm": 4.092207908630371,
      "learning_rate": 3.939059046416167e-05,
      "loss": 0.6549,
      "step": 18430
    },
    {
      "epoch": 5.819703189137986,
      "grad_norm": 4.285260200500488,
      "learning_rate": 3.93897133635056e-05,
      "loss": 0.5806,
      "step": 18431
    },
    {
      "epoch": 5.820018945374171,
      "grad_norm": 4.241168975830078,
      "learning_rate": 3.9388836262849526e-05,
      "loss": 0.597,
      "step": 18432
    },
    {
      "epoch": 5.820334701610356,
      "grad_norm": 5.924184322357178,
      "learning_rate": 3.938795916219346e-05,
      "loss": 0.5523,
      "step": 18433
    },
    {
      "epoch": 5.820650457846543,
      "grad_norm": 4.630560874938965,
      "learning_rate": 3.938708206153738e-05,
      "loss": 0.7147,
      "step": 18434
    },
    {
      "epoch": 5.820966214082728,
      "grad_norm": 4.273985385894775,
      "learning_rate": 3.9386204960881314e-05,
      "loss": 0.6878,
      "step": 18435
    },
    {
      "epoch": 5.821281970318914,
      "grad_norm": 4.28387975692749,
      "learning_rate": 3.938532786022524e-05,
      "loss": 0.6934,
      "step": 18436
    },
    {
      "epoch": 5.821597726555099,
      "grad_norm": 3.9802510738372803,
      "learning_rate": 3.938445075956917e-05,
      "loss": 0.6383,
      "step": 18437
    },
    {
      "epoch": 5.8219134827912855,
      "grad_norm": 4.214570999145508,
      "learning_rate": 3.9383573658913096e-05,
      "loss": 0.6486,
      "step": 18438
    },
    {
      "epoch": 5.822229239027471,
      "grad_norm": 3.4881715774536133,
      "learning_rate": 3.938269655825703e-05,
      "loss": 0.6343,
      "step": 18439
    },
    {
      "epoch": 5.822544995263656,
      "grad_norm": 3.948626756668091,
      "learning_rate": 3.938181945760095e-05,
      "loss": 0.4674,
      "step": 18440
    },
    {
      "epoch": 5.822860751499842,
      "grad_norm": 4.092318534851074,
      "learning_rate": 3.9380942356944884e-05,
      "loss": 0.6504,
      "step": 18441
    },
    {
      "epoch": 5.823176507736028,
      "grad_norm": 3.6594626903533936,
      "learning_rate": 3.938006525628881e-05,
      "loss": 0.6056,
      "step": 18442
    },
    {
      "epoch": 5.823492263972214,
      "grad_norm": 3.5733702182769775,
      "learning_rate": 3.937918815563274e-05,
      "loss": 0.5905,
      "step": 18443
    },
    {
      "epoch": 5.823808020208399,
      "grad_norm": 4.1789774894714355,
      "learning_rate": 3.9378311054976665e-05,
      "loss": 0.5861,
      "step": 18444
    },
    {
      "epoch": 5.824123776444585,
      "grad_norm": 3.9558122158050537,
      "learning_rate": 3.93774339543206e-05,
      "loss": 0.6431,
      "step": 18445
    },
    {
      "epoch": 5.8244395326807705,
      "grad_norm": 4.830594539642334,
      "learning_rate": 3.937655685366453e-05,
      "loss": 0.6684,
      "step": 18446
    },
    {
      "epoch": 5.824755288916956,
      "grad_norm": 4.396152496337891,
      "learning_rate": 3.937567975300846e-05,
      "loss": 0.6204,
      "step": 18447
    },
    {
      "epoch": 5.825071045153142,
      "grad_norm": 4.766666412353516,
      "learning_rate": 3.9374802652352385e-05,
      "loss": 0.6317,
      "step": 18448
    },
    {
      "epoch": 5.825386801389327,
      "grad_norm": 4.978735446929932,
      "learning_rate": 3.937392555169632e-05,
      "loss": 0.6789,
      "step": 18449
    },
    {
      "epoch": 5.8257025576255135,
      "grad_norm": 3.998870849609375,
      "learning_rate": 3.937304845104024e-05,
      "loss": 0.6358,
      "step": 18450
    },
    {
      "epoch": 5.826018313861699,
      "grad_norm": 5.659353256225586,
      "learning_rate": 3.9372171350384174e-05,
      "loss": 0.6288,
      "step": 18451
    },
    {
      "epoch": 5.826334070097884,
      "grad_norm": 3.1891634464263916,
      "learning_rate": 3.9371294249728105e-05,
      "loss": 0.5851,
      "step": 18452
    },
    {
      "epoch": 5.82664982633407,
      "grad_norm": 4.349932670593262,
      "learning_rate": 3.937041714907203e-05,
      "loss": 0.7079,
      "step": 18453
    },
    {
      "epoch": 5.8269655825702555,
      "grad_norm": 10.686745643615723,
      "learning_rate": 3.936954004841596e-05,
      "loss": 0.6654,
      "step": 18454
    },
    {
      "epoch": 5.827281338806442,
      "grad_norm": 4.014056205749512,
      "learning_rate": 3.936866294775989e-05,
      "loss": 0.6073,
      "step": 18455
    },
    {
      "epoch": 5.827597095042627,
      "grad_norm": 4.009698390960693,
      "learning_rate": 3.936778584710382e-05,
      "loss": 0.5871,
      "step": 18456
    },
    {
      "epoch": 5.827912851278812,
      "grad_norm": 4.377124786376953,
      "learning_rate": 3.9366908746447743e-05,
      "loss": 0.6974,
      "step": 18457
    },
    {
      "epoch": 5.8282286075149985,
      "grad_norm": 3.482356071472168,
      "learning_rate": 3.9366031645791675e-05,
      "loss": 0.5197,
      "step": 18458
    },
    {
      "epoch": 5.828544363751184,
      "grad_norm": 4.284902095794678,
      "learning_rate": 3.93651545451356e-05,
      "loss": 0.7272,
      "step": 18459
    },
    {
      "epoch": 5.82886011998737,
      "grad_norm": 4.549561977386475,
      "learning_rate": 3.936427744447953e-05,
      "loss": 0.6162,
      "step": 18460
    },
    {
      "epoch": 5.829175876223555,
      "grad_norm": 4.749250411987305,
      "learning_rate": 3.936340034382346e-05,
      "loss": 0.7704,
      "step": 18461
    },
    {
      "epoch": 5.8294916324597414,
      "grad_norm": 4.4344162940979,
      "learning_rate": 3.936252324316739e-05,
      "loss": 0.5921,
      "step": 18462
    },
    {
      "epoch": 5.829807388695927,
      "grad_norm": 4.059567928314209,
      "learning_rate": 3.936164614251131e-05,
      "loss": 0.6715,
      "step": 18463
    },
    {
      "epoch": 5.830123144932112,
      "grad_norm": 4.01205587387085,
      "learning_rate": 3.9360769041855245e-05,
      "loss": 0.5473,
      "step": 18464
    },
    {
      "epoch": 5.830438901168298,
      "grad_norm": 3.7263355255126953,
      "learning_rate": 3.935989194119917e-05,
      "loss": 0.6312,
      "step": 18465
    },
    {
      "epoch": 5.8307546574044835,
      "grad_norm": 3.716932773590088,
      "learning_rate": 3.93590148405431e-05,
      "loss": 0.6502,
      "step": 18466
    },
    {
      "epoch": 5.83107041364067,
      "grad_norm": 4.809239864349365,
      "learning_rate": 3.935813773988703e-05,
      "loss": 0.7045,
      "step": 18467
    },
    {
      "epoch": 5.831386169876855,
      "grad_norm": 4.050542831420898,
      "learning_rate": 3.935726063923096e-05,
      "loss": 0.6585,
      "step": 18468
    },
    {
      "epoch": 5.831701926113041,
      "grad_norm": 3.542921543121338,
      "learning_rate": 3.935638353857489e-05,
      "loss": 0.5944,
      "step": 18469
    },
    {
      "epoch": 5.8320176823492265,
      "grad_norm": 3.6384356021881104,
      "learning_rate": 3.9355506437918815e-05,
      "loss": 0.5842,
      "step": 18470
    },
    {
      "epoch": 5.832333438585412,
      "grad_norm": 4.050854682922363,
      "learning_rate": 3.9354629337262746e-05,
      "loss": 0.7502,
      "step": 18471
    },
    {
      "epoch": 5.832649194821598,
      "grad_norm": 3.066732883453369,
      "learning_rate": 3.935375223660667e-05,
      "loss": 0.6176,
      "step": 18472
    },
    {
      "epoch": 5.832964951057783,
      "grad_norm": 4.430089473724365,
      "learning_rate": 3.93528751359506e-05,
      "loss": 0.557,
      "step": 18473
    },
    {
      "epoch": 5.833280707293969,
      "grad_norm": 4.486152172088623,
      "learning_rate": 3.9351998035294535e-05,
      "loss": 0.7221,
      "step": 18474
    },
    {
      "epoch": 5.833596463530155,
      "grad_norm": 4.656775951385498,
      "learning_rate": 3.9351120934638466e-05,
      "loss": 0.7381,
      "step": 18475
    },
    {
      "epoch": 5.83391221976634,
      "grad_norm": 4.117346286773682,
      "learning_rate": 3.935024383398239e-05,
      "loss": 0.6864,
      "step": 18476
    },
    {
      "epoch": 5.834227976002526,
      "grad_norm": 4.245609283447266,
      "learning_rate": 3.934936673332632e-05,
      "loss": 0.6364,
      "step": 18477
    },
    {
      "epoch": 5.8345437322387115,
      "grad_norm": 4.316740989685059,
      "learning_rate": 3.934848963267025e-05,
      "loss": 0.6537,
      "step": 18478
    },
    {
      "epoch": 5.834859488474898,
      "grad_norm": 4.115858554840088,
      "learning_rate": 3.934761253201418e-05,
      "loss": 0.6353,
      "step": 18479
    },
    {
      "epoch": 5.835175244711083,
      "grad_norm": 4.697526454925537,
      "learning_rate": 3.9346735431358104e-05,
      "loss": 0.6006,
      "step": 18480
    },
    {
      "epoch": 5.835491000947268,
      "grad_norm": 4.047107219696045,
      "learning_rate": 3.9345858330702036e-05,
      "loss": 0.6662,
      "step": 18481
    },
    {
      "epoch": 5.835806757183454,
      "grad_norm": 3.9391891956329346,
      "learning_rate": 3.934498123004596e-05,
      "loss": 0.5797,
      "step": 18482
    },
    {
      "epoch": 5.83612251341964,
      "grad_norm": 4.483581066131592,
      "learning_rate": 3.934410412938989e-05,
      "loss": 0.5451,
      "step": 18483
    },
    {
      "epoch": 5.836438269655826,
      "grad_norm": 3.5188686847686768,
      "learning_rate": 3.934322702873382e-05,
      "loss": 0.5439,
      "step": 18484
    },
    {
      "epoch": 5.836754025892011,
      "grad_norm": 3.821294069290161,
      "learning_rate": 3.934234992807775e-05,
      "loss": 0.5791,
      "step": 18485
    },
    {
      "epoch": 5.837069782128197,
      "grad_norm": 3.8131210803985596,
      "learning_rate": 3.9341472827421674e-05,
      "loss": 0.6371,
      "step": 18486
    },
    {
      "epoch": 5.837385538364383,
      "grad_norm": 4.271270751953125,
      "learning_rate": 3.9340595726765606e-05,
      "loss": 0.6008,
      "step": 18487
    },
    {
      "epoch": 5.837701294600569,
      "grad_norm": 3.9230265617370605,
      "learning_rate": 3.933971862610953e-05,
      "loss": 0.5234,
      "step": 18488
    },
    {
      "epoch": 5.838017050836754,
      "grad_norm": 4.211228847503662,
      "learning_rate": 3.933884152545346e-05,
      "loss": 0.6176,
      "step": 18489
    },
    {
      "epoch": 5.8383328070729394,
      "grad_norm": 4.616707801818848,
      "learning_rate": 3.9337964424797394e-05,
      "loss": 0.6439,
      "step": 18490
    },
    {
      "epoch": 5.838648563309126,
      "grad_norm": 4.693314552307129,
      "learning_rate": 3.933708732414132e-05,
      "loss": 0.7236,
      "step": 18491
    },
    {
      "epoch": 5.838964319545311,
      "grad_norm": 6.140160083770752,
      "learning_rate": 3.933621022348525e-05,
      "loss": 0.7741,
      "step": 18492
    },
    {
      "epoch": 5.839280075781497,
      "grad_norm": 4.241382122039795,
      "learning_rate": 3.9335333122829176e-05,
      "loss": 0.662,
      "step": 18493
    },
    {
      "epoch": 5.839595832017682,
      "grad_norm": 4.116020679473877,
      "learning_rate": 3.933445602217311e-05,
      "loss": 0.7048,
      "step": 18494
    },
    {
      "epoch": 5.839911588253868,
      "grad_norm": 4.662806987762451,
      "learning_rate": 3.933357892151703e-05,
      "loss": 0.7046,
      "step": 18495
    },
    {
      "epoch": 5.840227344490054,
      "grad_norm": 4.766940593719482,
      "learning_rate": 3.9332701820860964e-05,
      "loss": 0.6178,
      "step": 18496
    },
    {
      "epoch": 5.840543100726239,
      "grad_norm": 5.130269527435303,
      "learning_rate": 3.933182472020489e-05,
      "loss": 0.6588,
      "step": 18497
    },
    {
      "epoch": 5.840858856962425,
      "grad_norm": 5.321107387542725,
      "learning_rate": 3.933094761954882e-05,
      "loss": 0.7207,
      "step": 18498
    },
    {
      "epoch": 5.841174613198611,
      "grad_norm": 5.591015815734863,
      "learning_rate": 3.9330070518892745e-05,
      "loss": 0.7258,
      "step": 18499
    },
    {
      "epoch": 5.841490369434796,
      "grad_norm": 4.456774711608887,
      "learning_rate": 3.932919341823668e-05,
      "loss": 0.6591,
      "step": 18500
    },
    {
      "epoch": 5.841806125670982,
      "grad_norm": 4.136175632476807,
      "learning_rate": 3.932831631758061e-05,
      "loss": 0.6903,
      "step": 18501
    },
    {
      "epoch": 5.842121881907167,
      "grad_norm": 4.511332988739014,
      "learning_rate": 3.932743921692454e-05,
      "loss": 0.7041,
      "step": 18502
    },
    {
      "epoch": 5.842437638143354,
      "grad_norm": 4.241607666015625,
      "learning_rate": 3.9326562116268465e-05,
      "loss": 0.6199,
      "step": 18503
    },
    {
      "epoch": 5.842753394379539,
      "grad_norm": 5.245243072509766,
      "learning_rate": 3.93256850156124e-05,
      "loss": 0.7089,
      "step": 18504
    },
    {
      "epoch": 5.843069150615724,
      "grad_norm": 5.399990081787109,
      "learning_rate": 3.932480791495632e-05,
      "loss": 0.6899,
      "step": 18505
    },
    {
      "epoch": 5.84338490685191,
      "grad_norm": 4.310243129730225,
      "learning_rate": 3.9323930814300254e-05,
      "loss": 0.5867,
      "step": 18506
    },
    {
      "epoch": 5.843700663088096,
      "grad_norm": 5.110260486602783,
      "learning_rate": 3.932305371364418e-05,
      "loss": 0.6177,
      "step": 18507
    },
    {
      "epoch": 5.844016419324282,
      "grad_norm": 3.957411527633667,
      "learning_rate": 3.932217661298811e-05,
      "loss": 0.6717,
      "step": 18508
    },
    {
      "epoch": 5.844332175560467,
      "grad_norm": 3.9865498542785645,
      "learning_rate": 3.9321299512332035e-05,
      "loss": 0.6254,
      "step": 18509
    },
    {
      "epoch": 5.844647931796653,
      "grad_norm": 3.367034912109375,
      "learning_rate": 3.932042241167597e-05,
      "loss": 0.588,
      "step": 18510
    },
    {
      "epoch": 5.844963688032839,
      "grad_norm": 4.300234794616699,
      "learning_rate": 3.93195453110199e-05,
      "loss": 0.6506,
      "step": 18511
    },
    {
      "epoch": 5.845279444269025,
      "grad_norm": 4.1795125007629395,
      "learning_rate": 3.931866821036382e-05,
      "loss": 0.6219,
      "step": 18512
    },
    {
      "epoch": 5.84559520050521,
      "grad_norm": 3.9920146465301514,
      "learning_rate": 3.9317791109707755e-05,
      "loss": 0.6465,
      "step": 18513
    },
    {
      "epoch": 5.845910956741395,
      "grad_norm": 4.2744669914245605,
      "learning_rate": 3.931691400905168e-05,
      "loss": 0.7262,
      "step": 18514
    },
    {
      "epoch": 5.846226712977582,
      "grad_norm": 3.652453899383545,
      "learning_rate": 3.931603690839561e-05,
      "loss": 0.6691,
      "step": 18515
    },
    {
      "epoch": 5.846542469213767,
      "grad_norm": 3.510338068008423,
      "learning_rate": 3.9315159807739537e-05,
      "loss": 0.5797,
      "step": 18516
    },
    {
      "epoch": 5.846858225449953,
      "grad_norm": 4.314855575561523,
      "learning_rate": 3.931428270708347e-05,
      "loss": 0.6978,
      "step": 18517
    },
    {
      "epoch": 5.847173981686138,
      "grad_norm": 3.3733346462249756,
      "learning_rate": 3.931340560642739e-05,
      "loss": 0.5753,
      "step": 18518
    },
    {
      "epoch": 5.847489737922324,
      "grad_norm": 3.3667385578155518,
      "learning_rate": 3.9312528505771325e-05,
      "loss": 0.6646,
      "step": 18519
    },
    {
      "epoch": 5.84780549415851,
      "grad_norm": 4.099564075469971,
      "learning_rate": 3.931165140511525e-05,
      "loss": 0.6342,
      "step": 18520
    },
    {
      "epoch": 5.848121250394695,
      "grad_norm": 3.8015923500061035,
      "learning_rate": 3.931077430445918e-05,
      "loss": 0.6955,
      "step": 18521
    },
    {
      "epoch": 5.848437006630881,
      "grad_norm": 3.350621461868286,
      "learning_rate": 3.9309897203803106e-05,
      "loss": 0.6022,
      "step": 18522
    },
    {
      "epoch": 5.848752762867067,
      "grad_norm": 3.4156243801116943,
      "learning_rate": 3.930902010314704e-05,
      "loss": 0.6277,
      "step": 18523
    },
    {
      "epoch": 5.849068519103252,
      "grad_norm": 3.373542308807373,
      "learning_rate": 3.930814300249096e-05,
      "loss": 0.6863,
      "step": 18524
    },
    {
      "epoch": 5.849384275339438,
      "grad_norm": 3.654245138168335,
      "learning_rate": 3.9307265901834895e-05,
      "loss": 0.5861,
      "step": 18525
    },
    {
      "epoch": 5.849700031575623,
      "grad_norm": 3.8122451305389404,
      "learning_rate": 3.930638880117882e-05,
      "loss": 0.6709,
      "step": 18526
    },
    {
      "epoch": 5.8500157878118095,
      "grad_norm": 3.420836925506592,
      "learning_rate": 3.930551170052275e-05,
      "loss": 0.6766,
      "step": 18527
    },
    {
      "epoch": 5.850331544047995,
      "grad_norm": 3.669346332550049,
      "learning_rate": 3.930463459986668e-05,
      "loss": 0.6024,
      "step": 18528
    },
    {
      "epoch": 5.85064730028418,
      "grad_norm": 3.6391844749450684,
      "learning_rate": 3.9303757499210615e-05,
      "loss": 0.664,
      "step": 18529
    },
    {
      "epoch": 5.850963056520366,
      "grad_norm": 4.787679672241211,
      "learning_rate": 3.930288039855454e-05,
      "loss": 0.7018,
      "step": 18530
    },
    {
      "epoch": 5.851278812756552,
      "grad_norm": 4.015697002410889,
      "learning_rate": 3.930200329789847e-05,
      "loss": 0.6591,
      "step": 18531
    },
    {
      "epoch": 5.851594568992738,
      "grad_norm": 3.977421998977661,
      "learning_rate": 3.93011261972424e-05,
      "loss": 0.6566,
      "step": 18532
    },
    {
      "epoch": 5.851910325228923,
      "grad_norm": 4.221027851104736,
      "learning_rate": 3.930024909658633e-05,
      "loss": 0.641,
      "step": 18533
    },
    {
      "epoch": 5.852226081465109,
      "grad_norm": 3.8770248889923096,
      "learning_rate": 3.929937199593026e-05,
      "loss": 0.6962,
      "step": 18534
    },
    {
      "epoch": 5.8525418377012945,
      "grad_norm": 4.67108154296875,
      "learning_rate": 3.9298494895274184e-05,
      "loss": 0.6393,
      "step": 18535
    },
    {
      "epoch": 5.852857593937481,
      "grad_norm": 3.5691237449645996,
      "learning_rate": 3.9297617794618116e-05,
      "loss": 0.635,
      "step": 18536
    },
    {
      "epoch": 5.853173350173666,
      "grad_norm": 3.35711932182312,
      "learning_rate": 3.929674069396204e-05,
      "loss": 0.6457,
      "step": 18537
    },
    {
      "epoch": 5.853489106409851,
      "grad_norm": 3.455349922180176,
      "learning_rate": 3.929586359330597e-05,
      "loss": 0.6294,
      "step": 18538
    },
    {
      "epoch": 5.8538048626460375,
      "grad_norm": 4.322296142578125,
      "learning_rate": 3.92949864926499e-05,
      "loss": 0.7296,
      "step": 18539
    },
    {
      "epoch": 5.854120618882223,
      "grad_norm": 3.5953359603881836,
      "learning_rate": 3.929410939199383e-05,
      "loss": 0.5201,
      "step": 18540
    },
    {
      "epoch": 5.854436375118409,
      "grad_norm": 3.5701351165771484,
      "learning_rate": 3.9293232291337754e-05,
      "loss": 0.6952,
      "step": 18541
    },
    {
      "epoch": 5.854752131354594,
      "grad_norm": 3.9156219959259033,
      "learning_rate": 3.9292355190681686e-05,
      "loss": 0.6432,
      "step": 18542
    },
    {
      "epoch": 5.85506788759078,
      "grad_norm": 4.358133792877197,
      "learning_rate": 3.929147809002561e-05,
      "loss": 0.6619,
      "step": 18543
    },
    {
      "epoch": 5.855383643826966,
      "grad_norm": 4.6200737953186035,
      "learning_rate": 3.929060098936954e-05,
      "loss": 0.6934,
      "step": 18544
    },
    {
      "epoch": 5.855699400063151,
      "grad_norm": 3.3656930923461914,
      "learning_rate": 3.928972388871347e-05,
      "loss": 0.6484,
      "step": 18545
    },
    {
      "epoch": 5.856015156299337,
      "grad_norm": 3.5904510021209717,
      "learning_rate": 3.92888467880574e-05,
      "loss": 0.5903,
      "step": 18546
    },
    {
      "epoch": 5.8563309125355225,
      "grad_norm": 4.264341354370117,
      "learning_rate": 3.9287969687401324e-05,
      "loss": 0.7076,
      "step": 18547
    },
    {
      "epoch": 5.856646668771708,
      "grad_norm": 4.73774528503418,
      "learning_rate": 3.9287092586745255e-05,
      "loss": 0.6737,
      "step": 18548
    },
    {
      "epoch": 5.856962425007894,
      "grad_norm": 3.5401244163513184,
      "learning_rate": 3.928621548608919e-05,
      "loss": 0.5459,
      "step": 18549
    },
    {
      "epoch": 5.857278181244079,
      "grad_norm": 3.9414262771606445,
      "learning_rate": 3.928533838543311e-05,
      "loss": 0.6214,
      "step": 18550
    },
    {
      "epoch": 5.8575939374802655,
      "grad_norm": 3.8362033367156982,
      "learning_rate": 3.9284461284777044e-05,
      "loss": 0.662,
      "step": 18551
    },
    {
      "epoch": 5.857909693716451,
      "grad_norm": 4.072661399841309,
      "learning_rate": 3.928358418412097e-05,
      "loss": 0.7215,
      "step": 18552
    },
    {
      "epoch": 5.858225449952637,
      "grad_norm": 4.050491809844971,
      "learning_rate": 3.92827070834649e-05,
      "loss": 0.6063,
      "step": 18553
    },
    {
      "epoch": 5.858541206188822,
      "grad_norm": 4.151352882385254,
      "learning_rate": 3.9281829982808825e-05,
      "loss": 0.702,
      "step": 18554
    },
    {
      "epoch": 5.8588569624250075,
      "grad_norm": 4.5883870124816895,
      "learning_rate": 3.928095288215276e-05,
      "loss": 0.6246,
      "step": 18555
    },
    {
      "epoch": 5.859172718661194,
      "grad_norm": 4.704186916351318,
      "learning_rate": 3.928007578149669e-05,
      "loss": 0.745,
      "step": 18556
    },
    {
      "epoch": 5.859488474897379,
      "grad_norm": 4.298678874969482,
      "learning_rate": 3.927919868084062e-05,
      "loss": 0.5873,
      "step": 18557
    },
    {
      "epoch": 5.859804231133565,
      "grad_norm": 3.3250277042388916,
      "learning_rate": 3.9278321580184545e-05,
      "loss": 0.6177,
      "step": 18558
    },
    {
      "epoch": 5.8601199873697505,
      "grad_norm": 4.717106819152832,
      "learning_rate": 3.927744447952848e-05,
      "loss": 0.7303,
      "step": 18559
    },
    {
      "epoch": 5.860435743605937,
      "grad_norm": 5.792118072509766,
      "learning_rate": 3.92765673788724e-05,
      "loss": 0.6698,
      "step": 18560
    },
    {
      "epoch": 5.860751499842122,
      "grad_norm": 3.4065983295440674,
      "learning_rate": 3.9275690278216333e-05,
      "loss": 0.6714,
      "step": 18561
    },
    {
      "epoch": 5.861067256078307,
      "grad_norm": 4.250485420227051,
      "learning_rate": 3.927481317756026e-05,
      "loss": 0.6475,
      "step": 18562
    },
    {
      "epoch": 5.861383012314493,
      "grad_norm": 4.011951446533203,
      "learning_rate": 3.927393607690419e-05,
      "loss": 0.6849,
      "step": 18563
    },
    {
      "epoch": 5.861698768550679,
      "grad_norm": 3.367236614227295,
      "learning_rate": 3.9273058976248115e-05,
      "loss": 0.5842,
      "step": 18564
    },
    {
      "epoch": 5.862014524786865,
      "grad_norm": 3.6030571460723877,
      "learning_rate": 3.927218187559205e-05,
      "loss": 0.6323,
      "step": 18565
    },
    {
      "epoch": 5.86233028102305,
      "grad_norm": 4.058852195739746,
      "learning_rate": 3.927130477493597e-05,
      "loss": 0.6474,
      "step": 18566
    },
    {
      "epoch": 5.8626460372592355,
      "grad_norm": 3.900757074356079,
      "learning_rate": 3.92704276742799e-05,
      "loss": 0.6596,
      "step": 18567
    },
    {
      "epoch": 5.862961793495422,
      "grad_norm": 3.659040927886963,
      "learning_rate": 3.926955057362383e-05,
      "loss": 0.5632,
      "step": 18568
    },
    {
      "epoch": 5.863277549731607,
      "grad_norm": 3.4696102142333984,
      "learning_rate": 3.926867347296776e-05,
      "loss": 0.5061,
      "step": 18569
    },
    {
      "epoch": 5.863593305967793,
      "grad_norm": 4.2557878494262695,
      "learning_rate": 3.926779637231169e-05,
      "loss": 0.7193,
      "step": 18570
    },
    {
      "epoch": 5.8639090622039785,
      "grad_norm": 4.556142330169678,
      "learning_rate": 3.9266919271655616e-05,
      "loss": 0.664,
      "step": 18571
    },
    {
      "epoch": 5.864224818440164,
      "grad_norm": 4.25016450881958,
      "learning_rate": 3.926604217099955e-05,
      "loss": 0.5967,
      "step": 18572
    },
    {
      "epoch": 5.86454057467635,
      "grad_norm": 4.359439373016357,
      "learning_rate": 3.926516507034347e-05,
      "loss": 0.6339,
      "step": 18573
    },
    {
      "epoch": 5.864856330912535,
      "grad_norm": 3.7127490043640137,
      "learning_rate": 3.9264287969687405e-05,
      "loss": 0.5983,
      "step": 18574
    },
    {
      "epoch": 5.865172087148721,
      "grad_norm": 4.259605884552002,
      "learning_rate": 3.926341086903133e-05,
      "loss": 0.7096,
      "step": 18575
    },
    {
      "epoch": 5.865487843384907,
      "grad_norm": 3.748568296432495,
      "learning_rate": 3.926253376837526e-05,
      "loss": 0.631,
      "step": 18576
    },
    {
      "epoch": 5.865803599621093,
      "grad_norm": 4.072754383087158,
      "learning_rate": 3.9261656667719186e-05,
      "loss": 0.6539,
      "step": 18577
    },
    {
      "epoch": 5.866119355857278,
      "grad_norm": 4.190104961395264,
      "learning_rate": 3.926077956706312e-05,
      "loss": 0.7471,
      "step": 18578
    },
    {
      "epoch": 5.8664351120934635,
      "grad_norm": 4.120384216308594,
      "learning_rate": 3.925990246640704e-05,
      "loss": 0.6585,
      "step": 18579
    },
    {
      "epoch": 5.86675086832965,
      "grad_norm": 4.811094760894775,
      "learning_rate": 3.9259025365750974e-05,
      "loss": 0.6027,
      "step": 18580
    },
    {
      "epoch": 5.867066624565835,
      "grad_norm": 4.347559452056885,
      "learning_rate": 3.92581482650949e-05,
      "loss": 0.6333,
      "step": 18581
    },
    {
      "epoch": 5.867382380802021,
      "grad_norm": 3.5635955333709717,
      "learning_rate": 3.925727116443883e-05,
      "loss": 0.5908,
      "step": 18582
    },
    {
      "epoch": 5.867698137038206,
      "grad_norm": 3.758065700531006,
      "learning_rate": 3.925639406378276e-05,
      "loss": 0.609,
      "step": 18583
    },
    {
      "epoch": 5.868013893274393,
      "grad_norm": 3.0644614696502686,
      "learning_rate": 3.9255516963126694e-05,
      "loss": 0.5377,
      "step": 18584
    },
    {
      "epoch": 5.868329649510578,
      "grad_norm": 4.759384632110596,
      "learning_rate": 3.925463986247062e-05,
      "loss": 0.6577,
      "step": 18585
    },
    {
      "epoch": 5.868645405746763,
      "grad_norm": 4.079875469207764,
      "learning_rate": 3.925376276181455e-05,
      "loss": 0.7324,
      "step": 18586
    },
    {
      "epoch": 5.868961161982949,
      "grad_norm": 4.3279128074646,
      "learning_rate": 3.9252885661158476e-05,
      "loss": 0.6353,
      "step": 18587
    },
    {
      "epoch": 5.869276918219135,
      "grad_norm": 4.323702812194824,
      "learning_rate": 3.925200856050241e-05,
      "loss": 0.6714,
      "step": 18588
    },
    {
      "epoch": 5.869592674455321,
      "grad_norm": 4.403369903564453,
      "learning_rate": 3.925113145984633e-05,
      "loss": 0.5693,
      "step": 18589
    },
    {
      "epoch": 5.869908430691506,
      "grad_norm": 3.478569746017456,
      "learning_rate": 3.9250254359190264e-05,
      "loss": 0.5491,
      "step": 18590
    },
    {
      "epoch": 5.870224186927691,
      "grad_norm": 4.3483428955078125,
      "learning_rate": 3.924937725853419e-05,
      "loss": 0.6543,
      "step": 18591
    },
    {
      "epoch": 5.870539943163878,
      "grad_norm": 4.173702716827393,
      "learning_rate": 3.924850015787812e-05,
      "loss": 0.6212,
      "step": 18592
    },
    {
      "epoch": 5.870855699400063,
      "grad_norm": 4.3348894119262695,
      "learning_rate": 3.924762305722205e-05,
      "loss": 0.6468,
      "step": 18593
    },
    {
      "epoch": 5.871171455636249,
      "grad_norm": 3.8643012046813965,
      "learning_rate": 3.924674595656598e-05,
      "loss": 0.6437,
      "step": 18594
    },
    {
      "epoch": 5.871487211872434,
      "grad_norm": 4.682570457458496,
      "learning_rate": 3.924586885590991e-05,
      "loss": 0.5908,
      "step": 18595
    },
    {
      "epoch": 5.87180296810862,
      "grad_norm": 4.3123297691345215,
      "learning_rate": 3.9244991755253834e-05,
      "loss": 0.6003,
      "step": 18596
    },
    {
      "epoch": 5.872118724344806,
      "grad_norm": 4.567746639251709,
      "learning_rate": 3.9244114654597766e-05,
      "loss": 0.5841,
      "step": 18597
    },
    {
      "epoch": 5.872434480580991,
      "grad_norm": 4.038758277893066,
      "learning_rate": 3.924323755394169e-05,
      "loss": 0.6639,
      "step": 18598
    },
    {
      "epoch": 5.872750236817177,
      "grad_norm": 4.391270160675049,
      "learning_rate": 3.924236045328562e-05,
      "loss": 0.6502,
      "step": 18599
    },
    {
      "epoch": 5.873065993053363,
      "grad_norm": 4.044454097747803,
      "learning_rate": 3.924148335262955e-05,
      "loss": 0.6775,
      "step": 18600
    },
    {
      "epoch": 5.873065993053363,
      "eval_loss": 0.8530714511871338,
      "eval_runtime": 4.5309,
      "eval_samples_per_second": 549.559,
      "eval_steps_per_second": 4.635,
      "step": 18600
    },
    {
      "epoch": 5.873381749289549,
      "grad_norm": 4.332732677459717,
      "learning_rate": 3.924060625197348e-05,
      "loss": 0.6973,
      "step": 18601
    },
    {
      "epoch": 5.873697505525734,
      "grad_norm": 5.21673583984375,
      "learning_rate": 3.9239729151317404e-05,
      "loss": 0.7193,
      "step": 18602
    },
    {
      "epoch": 5.87401326176192,
      "grad_norm": 3.69635272026062,
      "learning_rate": 3.9238852050661335e-05,
      "loss": 0.6199,
      "step": 18603
    },
    {
      "epoch": 5.874329017998106,
      "grad_norm": 3.769862174987793,
      "learning_rate": 3.923797495000526e-05,
      "loss": 0.6579,
      "step": 18604
    },
    {
      "epoch": 5.874644774234291,
      "grad_norm": 4.302379608154297,
      "learning_rate": 3.923709784934919e-05,
      "loss": 0.5613,
      "step": 18605
    },
    {
      "epoch": 5.874960530470477,
      "grad_norm": 3.832690477371216,
      "learning_rate": 3.923622074869312e-05,
      "loss": 0.4935,
      "step": 18606
    },
    {
      "epoch": 5.875276286706662,
      "grad_norm": 4.759598255157471,
      "learning_rate": 3.923534364803705e-05,
      "loss": 0.5328,
      "step": 18607
    },
    {
      "epoch": 5.8755920429428485,
      "grad_norm": 4.500927925109863,
      "learning_rate": 3.923446654738098e-05,
      "loss": 0.6741,
      "step": 18608
    },
    {
      "epoch": 5.875907799179034,
      "grad_norm": 4.325838088989258,
      "learning_rate": 3.9233589446724905e-05,
      "loss": 0.6023,
      "step": 18609
    },
    {
      "epoch": 5.876223555415219,
      "grad_norm": 3.9886600971221924,
      "learning_rate": 3.923271234606884e-05,
      "loss": 0.7007,
      "step": 18610
    },
    {
      "epoch": 5.876539311651405,
      "grad_norm": 5.7028656005859375,
      "learning_rate": 3.923183524541277e-05,
      "loss": 0.7702,
      "step": 18611
    },
    {
      "epoch": 5.876855067887591,
      "grad_norm": 3.947697401046753,
      "learning_rate": 3.9230958144756693e-05,
      "loss": 0.5839,
      "step": 18612
    },
    {
      "epoch": 5.877170824123777,
      "grad_norm": 5.103425025939941,
      "learning_rate": 3.9230081044100625e-05,
      "loss": 0.6882,
      "step": 18613
    },
    {
      "epoch": 5.877486580359962,
      "grad_norm": 3.8137335777282715,
      "learning_rate": 3.922920394344456e-05,
      "loss": 0.5921,
      "step": 18614
    },
    {
      "epoch": 5.877802336596147,
      "grad_norm": 4.064662933349609,
      "learning_rate": 3.922832684278848e-05,
      "loss": 0.6245,
      "step": 18615
    },
    {
      "epoch": 5.8781180928323336,
      "grad_norm": 4.170656204223633,
      "learning_rate": 3.922744974213241e-05,
      "loss": 0.6287,
      "step": 18616
    },
    {
      "epoch": 5.878433849068519,
      "grad_norm": 4.2090678215026855,
      "learning_rate": 3.922657264147634e-05,
      "loss": 0.6518,
      "step": 18617
    },
    {
      "epoch": 5.878749605304705,
      "grad_norm": 4.350375175476074,
      "learning_rate": 3.922569554082027e-05,
      "loss": 0.6196,
      "step": 18618
    },
    {
      "epoch": 5.87906536154089,
      "grad_norm": 4.109408855438232,
      "learning_rate": 3.9224818440164195e-05,
      "loss": 0.6368,
      "step": 18619
    },
    {
      "epoch": 5.879381117777076,
      "grad_norm": 4.296733856201172,
      "learning_rate": 3.9223941339508127e-05,
      "loss": 0.6492,
      "step": 18620
    },
    {
      "epoch": 5.879696874013262,
      "grad_norm": 4.288811683654785,
      "learning_rate": 3.922306423885205e-05,
      "loss": 0.681,
      "step": 18621
    },
    {
      "epoch": 5.880012630249447,
      "grad_norm": 3.917095899581909,
      "learning_rate": 3.922218713819598e-05,
      "loss": 0.7232,
      "step": 18622
    },
    {
      "epoch": 5.880328386485633,
      "grad_norm": 4.482824802398682,
      "learning_rate": 3.922131003753991e-05,
      "loss": 0.7148,
      "step": 18623
    },
    {
      "epoch": 5.880644142721819,
      "grad_norm": 3.6397299766540527,
      "learning_rate": 3.922043293688384e-05,
      "loss": 0.6662,
      "step": 18624
    },
    {
      "epoch": 5.880959898958005,
      "grad_norm": 4.63119649887085,
      "learning_rate": 3.9219555836227765e-05,
      "loss": 0.6452,
      "step": 18625
    },
    {
      "epoch": 5.88127565519419,
      "grad_norm": 5.641106605529785,
      "learning_rate": 3.9218678735571696e-05,
      "loss": 0.7475,
      "step": 18626
    },
    {
      "epoch": 5.881591411430376,
      "grad_norm": 4.390200614929199,
      "learning_rate": 3.921780163491562e-05,
      "loss": 0.7118,
      "step": 18627
    },
    {
      "epoch": 5.8819071676665615,
      "grad_norm": 3.900333881378174,
      "learning_rate": 3.921692453425955e-05,
      "loss": 0.5908,
      "step": 18628
    },
    {
      "epoch": 5.882222923902747,
      "grad_norm": 4.400350570678711,
      "learning_rate": 3.921604743360348e-05,
      "loss": 0.7638,
      "step": 18629
    },
    {
      "epoch": 5.882538680138933,
      "grad_norm": 4.918818473815918,
      "learning_rate": 3.921517033294741e-05,
      "loss": 0.6777,
      "step": 18630
    },
    {
      "epoch": 5.882854436375118,
      "grad_norm": 4.867580413818359,
      "learning_rate": 3.921429323229134e-05,
      "loss": 0.7207,
      "step": 18631
    },
    {
      "epoch": 5.8831701926113045,
      "grad_norm": 3.881953477859497,
      "learning_rate": 3.9213416131635266e-05,
      "loss": 0.6446,
      "step": 18632
    },
    {
      "epoch": 5.88348594884749,
      "grad_norm": 4.669314861297607,
      "learning_rate": 3.92125390309792e-05,
      "loss": 0.5699,
      "step": 18633
    },
    {
      "epoch": 5.883801705083675,
      "grad_norm": 3.770421266555786,
      "learning_rate": 3.921166193032312e-05,
      "loss": 0.6754,
      "step": 18634
    },
    {
      "epoch": 5.884117461319861,
      "grad_norm": 4.190687656402588,
      "learning_rate": 3.9210784829667054e-05,
      "loss": 0.6254,
      "step": 18635
    },
    {
      "epoch": 5.8844332175560465,
      "grad_norm": 4.3541436195373535,
      "learning_rate": 3.920990772901098e-05,
      "loss": 0.6256,
      "step": 18636
    },
    {
      "epoch": 5.884748973792233,
      "grad_norm": 4.170764923095703,
      "learning_rate": 3.920903062835491e-05,
      "loss": 0.5794,
      "step": 18637
    },
    {
      "epoch": 5.885064730028418,
      "grad_norm": 4.779078960418701,
      "learning_rate": 3.920815352769884e-05,
      "loss": 0.672,
      "step": 18638
    },
    {
      "epoch": 5.885380486264603,
      "grad_norm": 4.615422248840332,
      "learning_rate": 3.9207276427042774e-05,
      "loss": 0.7411,
      "step": 18639
    },
    {
      "epoch": 5.8856962425007895,
      "grad_norm": 4.684847354888916,
      "learning_rate": 3.92063993263867e-05,
      "loss": 0.6084,
      "step": 18640
    },
    {
      "epoch": 5.886011998736975,
      "grad_norm": 3.348673105239868,
      "learning_rate": 3.920552222573063e-05,
      "loss": 0.5432,
      "step": 18641
    },
    {
      "epoch": 5.886327754973161,
      "grad_norm": 4.867037773132324,
      "learning_rate": 3.9204645125074556e-05,
      "loss": 0.6874,
      "step": 18642
    },
    {
      "epoch": 5.886643511209346,
      "grad_norm": 3.4730169773101807,
      "learning_rate": 3.920376802441849e-05,
      "loss": 0.5246,
      "step": 18643
    },
    {
      "epoch": 5.8869592674455316,
      "grad_norm": 3.6048407554626465,
      "learning_rate": 3.920289092376241e-05,
      "loss": 0.5183,
      "step": 18644
    },
    {
      "epoch": 5.887275023681718,
      "grad_norm": 3.5640902519226074,
      "learning_rate": 3.9202013823106344e-05,
      "loss": 0.6824,
      "step": 18645
    },
    {
      "epoch": 5.887590779917903,
      "grad_norm": 4.379946708679199,
      "learning_rate": 3.920113672245027e-05,
      "loss": 0.6188,
      "step": 18646
    },
    {
      "epoch": 5.887906536154089,
      "grad_norm": 4.256976127624512,
      "learning_rate": 3.92002596217942e-05,
      "loss": 0.6099,
      "step": 18647
    },
    {
      "epoch": 5.8882222923902745,
      "grad_norm": 4.457243919372559,
      "learning_rate": 3.9199382521138126e-05,
      "loss": 0.7321,
      "step": 18648
    },
    {
      "epoch": 5.888538048626461,
      "grad_norm": 3.977973461151123,
      "learning_rate": 3.919850542048206e-05,
      "loss": 0.6277,
      "step": 18649
    },
    {
      "epoch": 5.888853804862646,
      "grad_norm": 4.249192714691162,
      "learning_rate": 3.919762831982598e-05,
      "loss": 0.5626,
      "step": 18650
    },
    {
      "epoch": 5.889169561098832,
      "grad_norm": 4.083840370178223,
      "learning_rate": 3.9196751219169914e-05,
      "loss": 0.5495,
      "step": 18651
    },
    {
      "epoch": 5.8894853173350175,
      "grad_norm": 4.106509208679199,
      "learning_rate": 3.9195874118513846e-05,
      "loss": 0.7127,
      "step": 18652
    },
    {
      "epoch": 5.889801073571203,
      "grad_norm": 3.789071559906006,
      "learning_rate": 3.919499701785777e-05,
      "loss": 0.6025,
      "step": 18653
    },
    {
      "epoch": 5.890116829807389,
      "grad_norm": 3.8189263343811035,
      "learning_rate": 3.91941199172017e-05,
      "loss": 0.6318,
      "step": 18654
    },
    {
      "epoch": 5.890432586043574,
      "grad_norm": 4.1794114112854,
      "learning_rate": 3.919324281654563e-05,
      "loss": 0.6166,
      "step": 18655
    },
    {
      "epoch": 5.89074834227976,
      "grad_norm": 4.517938137054443,
      "learning_rate": 3.919236571588956e-05,
      "loss": 0.6813,
      "step": 18656
    },
    {
      "epoch": 5.891064098515946,
      "grad_norm": 4.994888782501221,
      "learning_rate": 3.9191488615233484e-05,
      "loss": 0.6472,
      "step": 18657
    },
    {
      "epoch": 5.891379854752131,
      "grad_norm": 3.690854549407959,
      "learning_rate": 3.9190611514577415e-05,
      "loss": 0.6491,
      "step": 18658
    },
    {
      "epoch": 5.891695610988317,
      "grad_norm": 3.239459276199341,
      "learning_rate": 3.918973441392134e-05,
      "loss": 0.6129,
      "step": 18659
    },
    {
      "epoch": 5.8920113672245025,
      "grad_norm": 3.5971436500549316,
      "learning_rate": 3.918885731326527e-05,
      "loss": 0.5121,
      "step": 18660
    },
    {
      "epoch": 5.892327123460689,
      "grad_norm": 4.587799549102783,
      "learning_rate": 3.91879802126092e-05,
      "loss": 0.6233,
      "step": 18661
    },
    {
      "epoch": 5.892642879696874,
      "grad_norm": 4.608718395233154,
      "learning_rate": 3.918710311195313e-05,
      "loss": 0.6659,
      "step": 18662
    },
    {
      "epoch": 5.892958635933059,
      "grad_norm": 3.5231826305389404,
      "learning_rate": 3.918622601129705e-05,
      "loss": 0.5382,
      "step": 18663
    },
    {
      "epoch": 5.893274392169245,
      "grad_norm": 4.180911540985107,
      "learning_rate": 3.9185348910640985e-05,
      "loss": 0.6537,
      "step": 18664
    },
    {
      "epoch": 5.893590148405431,
      "grad_norm": 4.5245819091796875,
      "learning_rate": 3.918447180998492e-05,
      "loss": 0.7037,
      "step": 18665
    },
    {
      "epoch": 5.893905904641617,
      "grad_norm": 4.600048542022705,
      "learning_rate": 3.918359470932885e-05,
      "loss": 0.7304,
      "step": 18666
    },
    {
      "epoch": 5.894221660877802,
      "grad_norm": 4.407974720001221,
      "learning_rate": 3.918271760867277e-05,
      "loss": 0.6114,
      "step": 18667
    },
    {
      "epoch": 5.894537417113988,
      "grad_norm": 4.577286720275879,
      "learning_rate": 3.9181840508016705e-05,
      "loss": 0.6492,
      "step": 18668
    },
    {
      "epoch": 5.894853173350174,
      "grad_norm": 3.6962039470672607,
      "learning_rate": 3.918096340736063e-05,
      "loss": 0.643,
      "step": 18669
    },
    {
      "epoch": 5.895168929586359,
      "grad_norm": 3.646303415298462,
      "learning_rate": 3.918008630670456e-05,
      "loss": 0.6152,
      "step": 18670
    },
    {
      "epoch": 5.895484685822545,
      "grad_norm": 6.989019393920898,
      "learning_rate": 3.9179209206048486e-05,
      "loss": 0.599,
      "step": 18671
    },
    {
      "epoch": 5.89580044205873,
      "grad_norm": 4.452682018280029,
      "learning_rate": 3.917833210539242e-05,
      "loss": 0.6453,
      "step": 18672
    },
    {
      "epoch": 5.896116198294917,
      "grad_norm": 4.345503807067871,
      "learning_rate": 3.917745500473634e-05,
      "loss": 0.6909,
      "step": 18673
    },
    {
      "epoch": 5.896431954531102,
      "grad_norm": 3.7032480239868164,
      "learning_rate": 3.9176577904080275e-05,
      "loss": 0.583,
      "step": 18674
    },
    {
      "epoch": 5.896747710767288,
      "grad_norm": 4.014932632446289,
      "learning_rate": 3.9175700803424206e-05,
      "loss": 0.5313,
      "step": 18675
    },
    {
      "epoch": 5.897063467003473,
      "grad_norm": 3.8447179794311523,
      "learning_rate": 3.917482370276813e-05,
      "loss": 0.5867,
      "step": 18676
    },
    {
      "epoch": 5.897379223239659,
      "grad_norm": 3.9306957721710205,
      "learning_rate": 3.917394660211206e-05,
      "loss": 0.6193,
      "step": 18677
    },
    {
      "epoch": 5.897694979475845,
      "grad_norm": 4.270571708679199,
      "learning_rate": 3.917306950145599e-05,
      "loss": 0.6953,
      "step": 18678
    },
    {
      "epoch": 5.89801073571203,
      "grad_norm": 3.847806692123413,
      "learning_rate": 3.917219240079992e-05,
      "loss": 0.6046,
      "step": 18679
    },
    {
      "epoch": 5.898326491948216,
      "grad_norm": 3.3612565994262695,
      "learning_rate": 3.9171315300143845e-05,
      "loss": 0.5622,
      "step": 18680
    },
    {
      "epoch": 5.898642248184402,
      "grad_norm": 3.9874680042266846,
      "learning_rate": 3.9170438199487776e-05,
      "loss": 0.5619,
      "step": 18681
    },
    {
      "epoch": 5.898958004420587,
      "grad_norm": 4.189443111419678,
      "learning_rate": 3.91695610988317e-05,
      "loss": 0.6727,
      "step": 18682
    },
    {
      "epoch": 5.899273760656773,
      "grad_norm": 3.4638566970825195,
      "learning_rate": 3.916868399817563e-05,
      "loss": 0.5965,
      "step": 18683
    },
    {
      "epoch": 5.899589516892958,
      "grad_norm": 4.184054851531982,
      "learning_rate": 3.916780689751956e-05,
      "loss": 0.6389,
      "step": 18684
    },
    {
      "epoch": 5.899905273129145,
      "grad_norm": 4.3374810218811035,
      "learning_rate": 3.916692979686349e-05,
      "loss": 0.7239,
      "step": 18685
    },
    {
      "epoch": 5.90022102936533,
      "grad_norm": 4.1047282218933105,
      "learning_rate": 3.9166052696207414e-05,
      "loss": 0.591,
      "step": 18686
    },
    {
      "epoch": 5.900536785601515,
      "grad_norm": 5.987377166748047,
      "learning_rate": 3.9165175595551346e-05,
      "loss": 0.8037,
      "step": 18687
    },
    {
      "epoch": 5.900852541837701,
      "grad_norm": 4.180068016052246,
      "learning_rate": 3.916429849489527e-05,
      "loss": 0.5931,
      "step": 18688
    },
    {
      "epoch": 5.901168298073887,
      "grad_norm": 4.865485668182373,
      "learning_rate": 3.91634213942392e-05,
      "loss": 0.6289,
      "step": 18689
    },
    {
      "epoch": 5.901484054310073,
      "grad_norm": 4.0731353759765625,
      "learning_rate": 3.9162544293583134e-05,
      "loss": 0.5962,
      "step": 18690
    },
    {
      "epoch": 5.901799810546258,
      "grad_norm": 3.2336413860321045,
      "learning_rate": 3.916166719292706e-05,
      "loss": 0.6052,
      "step": 18691
    },
    {
      "epoch": 5.902115566782444,
      "grad_norm": 3.7845752239227295,
      "learning_rate": 3.916079009227099e-05,
      "loss": 0.6789,
      "step": 18692
    },
    {
      "epoch": 5.90243132301863,
      "grad_norm": 4.224472999572754,
      "learning_rate": 3.915991299161492e-05,
      "loss": 0.593,
      "step": 18693
    },
    {
      "epoch": 5.902747079254815,
      "grad_norm": 4.973026275634766,
      "learning_rate": 3.915903589095885e-05,
      "loss": 0.6642,
      "step": 18694
    },
    {
      "epoch": 5.903062835491001,
      "grad_norm": 3.6813735961914062,
      "learning_rate": 3.915815879030278e-05,
      "loss": 0.6247,
      "step": 18695
    },
    {
      "epoch": 5.903378591727186,
      "grad_norm": 4.375267028808594,
      "learning_rate": 3.915728168964671e-05,
      "loss": 0.68,
      "step": 18696
    },
    {
      "epoch": 5.903694347963373,
      "grad_norm": 3.7050278186798096,
      "learning_rate": 3.9156404588990636e-05,
      "loss": 0.5677,
      "step": 18697
    },
    {
      "epoch": 5.904010104199558,
      "grad_norm": 4.2649617195129395,
      "learning_rate": 3.915552748833457e-05,
      "loss": 0.6379,
      "step": 18698
    },
    {
      "epoch": 5.904325860435744,
      "grad_norm": 3.6951406002044678,
      "learning_rate": 3.915465038767849e-05,
      "loss": 0.6144,
      "step": 18699
    },
    {
      "epoch": 5.904641616671929,
      "grad_norm": 4.79029655456543,
      "learning_rate": 3.9153773287022424e-05,
      "loss": 0.6713,
      "step": 18700
    },
    {
      "epoch": 5.904957372908115,
      "grad_norm": 3.9815361499786377,
      "learning_rate": 3.915289618636635e-05,
      "loss": 0.6459,
      "step": 18701
    },
    {
      "epoch": 5.905273129144301,
      "grad_norm": 5.133005142211914,
      "learning_rate": 3.915201908571028e-05,
      "loss": 0.7452,
      "step": 18702
    },
    {
      "epoch": 5.905588885380486,
      "grad_norm": 4.263973712921143,
      "learning_rate": 3.9151141985054205e-05,
      "loss": 0.6189,
      "step": 18703
    },
    {
      "epoch": 5.905904641616672,
      "grad_norm": 3.6303818225860596,
      "learning_rate": 3.915026488439814e-05,
      "loss": 0.6754,
      "step": 18704
    },
    {
      "epoch": 5.906220397852858,
      "grad_norm": 4.343721389770508,
      "learning_rate": 3.914938778374206e-05,
      "loss": 0.643,
      "step": 18705
    },
    {
      "epoch": 5.906536154089043,
      "grad_norm": 3.886319160461426,
      "learning_rate": 3.9148510683085994e-05,
      "loss": 0.5479,
      "step": 18706
    },
    {
      "epoch": 5.906851910325229,
      "grad_norm": 4.066445350646973,
      "learning_rate": 3.914763358242992e-05,
      "loss": 0.5966,
      "step": 18707
    },
    {
      "epoch": 5.907167666561414,
      "grad_norm": 4.037452697753906,
      "learning_rate": 3.914675648177385e-05,
      "loss": 0.674,
      "step": 18708
    },
    {
      "epoch": 5.9074834227976005,
      "grad_norm": 4.3137736320495605,
      "learning_rate": 3.9145879381117775e-05,
      "loss": 0.635,
      "step": 18709
    },
    {
      "epoch": 5.907799179033786,
      "grad_norm": 3.763498067855835,
      "learning_rate": 3.914500228046171e-05,
      "loss": 0.6187,
      "step": 18710
    },
    {
      "epoch": 5.908114935269971,
      "grad_norm": 3.8022758960723877,
      "learning_rate": 3.914412517980564e-05,
      "loss": 0.6165,
      "step": 18711
    },
    {
      "epoch": 5.908430691506157,
      "grad_norm": 3.4307398796081543,
      "learning_rate": 3.9143248079149563e-05,
      "loss": 0.587,
      "step": 18712
    },
    {
      "epoch": 5.908746447742343,
      "grad_norm": 4.103477954864502,
      "learning_rate": 3.9142370978493495e-05,
      "loss": 0.5801,
      "step": 18713
    },
    {
      "epoch": 5.909062203978529,
      "grad_norm": 3.542296886444092,
      "learning_rate": 3.914149387783742e-05,
      "loss": 0.5768,
      "step": 18714
    },
    {
      "epoch": 5.909377960214714,
      "grad_norm": 4.512630462646484,
      "learning_rate": 3.914061677718135e-05,
      "loss": 0.7066,
      "step": 18715
    },
    {
      "epoch": 5.9096937164509,
      "grad_norm": 4.460239410400391,
      "learning_rate": 3.913973967652528e-05,
      "loss": 0.6321,
      "step": 18716
    },
    {
      "epoch": 5.9100094726870855,
      "grad_norm": 3.9883580207824707,
      "learning_rate": 3.913886257586921e-05,
      "loss": 0.5287,
      "step": 18717
    },
    {
      "epoch": 5.910325228923272,
      "grad_norm": 5.688710689544678,
      "learning_rate": 3.913798547521313e-05,
      "loss": 0.6455,
      "step": 18718
    },
    {
      "epoch": 5.910640985159457,
      "grad_norm": 3.804297924041748,
      "learning_rate": 3.913710837455707e-05,
      "loss": 0.5926,
      "step": 18719
    },
    {
      "epoch": 5.910956741395642,
      "grad_norm": 4.494628429412842,
      "learning_rate": 3.9136231273901e-05,
      "loss": 0.5954,
      "step": 18720
    },
    {
      "epoch": 5.9112724976318285,
      "grad_norm": 4.84122896194458,
      "learning_rate": 3.913535417324493e-05,
      "loss": 0.6688,
      "step": 18721
    },
    {
      "epoch": 5.911588253868014,
      "grad_norm": 5.201327323913574,
      "learning_rate": 3.913447707258885e-05,
      "loss": 0.6238,
      "step": 18722
    },
    {
      "epoch": 5.9119040101042,
      "grad_norm": 4.711786270141602,
      "learning_rate": 3.9133599971932785e-05,
      "loss": 0.6593,
      "step": 18723
    },
    {
      "epoch": 5.912219766340385,
      "grad_norm": 4.021449565887451,
      "learning_rate": 3.913272287127671e-05,
      "loss": 0.6531,
      "step": 18724
    },
    {
      "epoch": 5.912535522576571,
      "grad_norm": 4.875438690185547,
      "learning_rate": 3.913184577062064e-05,
      "loss": 0.6386,
      "step": 18725
    },
    {
      "epoch": 5.912851278812757,
      "grad_norm": 4.11625862121582,
      "learning_rate": 3.9130968669964566e-05,
      "loss": 0.6857,
      "step": 18726
    },
    {
      "epoch": 5.913167035048942,
      "grad_norm": 4.454707145690918,
      "learning_rate": 3.91300915693085e-05,
      "loss": 0.7108,
      "step": 18727
    },
    {
      "epoch": 5.913482791285128,
      "grad_norm": 4.632839679718018,
      "learning_rate": 3.912921446865242e-05,
      "loss": 0.7341,
      "step": 18728
    },
    {
      "epoch": 5.9137985475213135,
      "grad_norm": 4.318973064422607,
      "learning_rate": 3.9128337367996355e-05,
      "loss": 0.6191,
      "step": 18729
    },
    {
      "epoch": 5.914114303757499,
      "grad_norm": 4.272735595703125,
      "learning_rate": 3.912746026734028e-05,
      "loss": 0.6862,
      "step": 18730
    },
    {
      "epoch": 5.914430059993685,
      "grad_norm": 3.601761817932129,
      "learning_rate": 3.912658316668421e-05,
      "loss": 0.6472,
      "step": 18731
    },
    {
      "epoch": 5.91474581622987,
      "grad_norm": 3.549549102783203,
      "learning_rate": 3.9125706066028136e-05,
      "loss": 0.6181,
      "step": 18732
    },
    {
      "epoch": 5.9150615724660565,
      "grad_norm": 4.1623005867004395,
      "learning_rate": 3.912482896537207e-05,
      "loss": 0.6583,
      "step": 18733
    },
    {
      "epoch": 5.915377328702242,
      "grad_norm": 4.858654975891113,
      "learning_rate": 3.9123951864716e-05,
      "loss": 0.7102,
      "step": 18734
    },
    {
      "epoch": 5.915693084938427,
      "grad_norm": 3.8083205223083496,
      "learning_rate": 3.9123074764059924e-05,
      "loss": 0.6806,
      "step": 18735
    },
    {
      "epoch": 5.916008841174613,
      "grad_norm": 4.909088134765625,
      "learning_rate": 3.9122197663403856e-05,
      "loss": 0.6709,
      "step": 18736
    },
    {
      "epoch": 5.9163245974107985,
      "grad_norm": 3.7976326942443848,
      "learning_rate": 3.912132056274778e-05,
      "loss": 0.581,
      "step": 18737
    },
    {
      "epoch": 5.916640353646985,
      "grad_norm": 4.093976020812988,
      "learning_rate": 3.912044346209171e-05,
      "loss": 0.5714,
      "step": 18738
    },
    {
      "epoch": 5.91695610988317,
      "grad_norm": 3.6251566410064697,
      "learning_rate": 3.911956636143564e-05,
      "loss": 0.6408,
      "step": 18739
    },
    {
      "epoch": 5.917271866119356,
      "grad_norm": 3.9367854595184326,
      "learning_rate": 3.911868926077957e-05,
      "loss": 0.6447,
      "step": 18740
    },
    {
      "epoch": 5.9175876223555415,
      "grad_norm": 4.322203159332275,
      "learning_rate": 3.9117812160123494e-05,
      "loss": 0.7159,
      "step": 18741
    },
    {
      "epoch": 5.917903378591728,
      "grad_norm": 3.726069688796997,
      "learning_rate": 3.9116935059467426e-05,
      "loss": 0.6388,
      "step": 18742
    },
    {
      "epoch": 5.918219134827913,
      "grad_norm": 4.1398515701293945,
      "learning_rate": 3.911605795881135e-05,
      "loss": 0.6184,
      "step": 18743
    },
    {
      "epoch": 5.918534891064098,
      "grad_norm": 3.844062089920044,
      "learning_rate": 3.911518085815528e-05,
      "loss": 0.6762,
      "step": 18744
    },
    {
      "epoch": 5.918850647300284,
      "grad_norm": 4.5723347663879395,
      "learning_rate": 3.911430375749921e-05,
      "loss": 0.6522,
      "step": 18745
    },
    {
      "epoch": 5.91916640353647,
      "grad_norm": 4.473435401916504,
      "learning_rate": 3.911342665684314e-05,
      "loss": 0.5715,
      "step": 18746
    },
    {
      "epoch": 5.919482159772656,
      "grad_norm": 4.4924397468566895,
      "learning_rate": 3.911254955618707e-05,
      "loss": 0.6877,
      "step": 18747
    },
    {
      "epoch": 5.919797916008841,
      "grad_norm": 4.204998016357422,
      "learning_rate": 3.9111672455531e-05,
      "loss": 0.6553,
      "step": 18748
    },
    {
      "epoch": 5.9201136722450265,
      "grad_norm": 3.9355552196502686,
      "learning_rate": 3.911079535487493e-05,
      "loss": 0.5821,
      "step": 18749
    },
    {
      "epoch": 5.920429428481213,
      "grad_norm": 4.426124095916748,
      "learning_rate": 3.910991825421886e-05,
      "loss": 0.6445,
      "step": 18750
    },
    {
      "epoch": 5.920745184717398,
      "grad_norm": 3.777705669403076,
      "learning_rate": 3.9109041153562784e-05,
      "loss": 0.6612,
      "step": 18751
    },
    {
      "epoch": 5.921060940953584,
      "grad_norm": 4.138376712799072,
      "learning_rate": 3.9108164052906716e-05,
      "loss": 0.6702,
      "step": 18752
    },
    {
      "epoch": 5.9213766971897694,
      "grad_norm": 3.248401165008545,
      "learning_rate": 3.910728695225064e-05,
      "loss": 0.5936,
      "step": 18753
    },
    {
      "epoch": 5.921692453425955,
      "grad_norm": 4.326920032501221,
      "learning_rate": 3.910640985159457e-05,
      "loss": 0.6447,
      "step": 18754
    },
    {
      "epoch": 5.922008209662141,
      "grad_norm": 3.664341449737549,
      "learning_rate": 3.9105532750938504e-05,
      "loss": 0.6178,
      "step": 18755
    },
    {
      "epoch": 5.922323965898326,
      "grad_norm": 4.214126110076904,
      "learning_rate": 3.910465565028243e-05,
      "loss": 0.645,
      "step": 18756
    },
    {
      "epoch": 5.922639722134512,
      "grad_norm": 4.074782848358154,
      "learning_rate": 3.910377854962636e-05,
      "loss": 0.631,
      "step": 18757
    },
    {
      "epoch": 5.922955478370698,
      "grad_norm": 4.038924217224121,
      "learning_rate": 3.9102901448970285e-05,
      "loss": 0.6083,
      "step": 18758
    },
    {
      "epoch": 5.923271234606884,
      "grad_norm": 3.40693736076355,
      "learning_rate": 3.910202434831422e-05,
      "loss": 0.6179,
      "step": 18759
    },
    {
      "epoch": 5.923586990843069,
      "grad_norm": 3.382539987564087,
      "learning_rate": 3.910114724765814e-05,
      "loss": 0.6472,
      "step": 18760
    },
    {
      "epoch": 5.9239027470792545,
      "grad_norm": 3.831371545791626,
      "learning_rate": 3.9100270147002074e-05,
      "loss": 0.5749,
      "step": 18761
    },
    {
      "epoch": 5.924218503315441,
      "grad_norm": 4.414128303527832,
      "learning_rate": 3.9099393046346e-05,
      "loss": 0.6317,
      "step": 18762
    },
    {
      "epoch": 5.924534259551626,
      "grad_norm": 3.5896522998809814,
      "learning_rate": 3.909851594568993e-05,
      "loss": 0.6506,
      "step": 18763
    },
    {
      "epoch": 5.924850015787812,
      "grad_norm": 4.296501159667969,
      "learning_rate": 3.9097638845033855e-05,
      "loss": 0.6272,
      "step": 18764
    },
    {
      "epoch": 5.925165772023997,
      "grad_norm": 4.089888572692871,
      "learning_rate": 3.909676174437779e-05,
      "loss": 0.5606,
      "step": 18765
    },
    {
      "epoch": 5.925481528260184,
      "grad_norm": 4.065506458282471,
      "learning_rate": 3.909588464372171e-05,
      "loss": 0.6991,
      "step": 18766
    },
    {
      "epoch": 5.925797284496369,
      "grad_norm": 4.160661697387695,
      "learning_rate": 3.909500754306564e-05,
      "loss": 0.6436,
      "step": 18767
    },
    {
      "epoch": 5.926113040732554,
      "grad_norm": 4.860540390014648,
      "learning_rate": 3.909413044240957e-05,
      "loss": 0.7546,
      "step": 18768
    },
    {
      "epoch": 5.92642879696874,
      "grad_norm": 3.5572896003723145,
      "learning_rate": 3.90932533417535e-05,
      "loss": 0.5217,
      "step": 18769
    },
    {
      "epoch": 5.926744553204926,
      "grad_norm": 4.099070072174072,
      "learning_rate": 3.9092376241097425e-05,
      "loss": 0.6719,
      "step": 18770
    },
    {
      "epoch": 5.927060309441112,
      "grad_norm": 5.052351474761963,
      "learning_rate": 3.9091499140441357e-05,
      "loss": 0.734,
      "step": 18771
    },
    {
      "epoch": 5.927376065677297,
      "grad_norm": 3.8566741943359375,
      "learning_rate": 3.909062203978529e-05,
      "loss": 0.6037,
      "step": 18772
    },
    {
      "epoch": 5.927691821913482,
      "grad_norm": 5.178961277008057,
      "learning_rate": 3.908974493912921e-05,
      "loss": 0.6847,
      "step": 18773
    },
    {
      "epoch": 5.928007578149669,
      "grad_norm": 3.816765785217285,
      "learning_rate": 3.9088867838473145e-05,
      "loss": 0.5691,
      "step": 18774
    },
    {
      "epoch": 5.928323334385854,
      "grad_norm": 3.753512144088745,
      "learning_rate": 3.9087990737817077e-05,
      "loss": 0.6273,
      "step": 18775
    },
    {
      "epoch": 5.92863909062204,
      "grad_norm": 4.541024684906006,
      "learning_rate": 3.9087113637161e-05,
      "loss": 0.5747,
      "step": 18776
    },
    {
      "epoch": 5.928954846858225,
      "grad_norm": 4.251530647277832,
      "learning_rate": 3.908623653650493e-05,
      "loss": 0.6372,
      "step": 18777
    },
    {
      "epoch": 5.929270603094411,
      "grad_norm": 5.011879920959473,
      "learning_rate": 3.9085359435848865e-05,
      "loss": 0.7418,
      "step": 18778
    },
    {
      "epoch": 5.929586359330597,
      "grad_norm": 4.052225112915039,
      "learning_rate": 3.908448233519279e-05,
      "loss": 0.6303,
      "step": 18779
    },
    {
      "epoch": 5.929902115566782,
      "grad_norm": 3.019141912460327,
      "learning_rate": 3.908360523453672e-05,
      "loss": 0.5577,
      "step": 18780
    },
    {
      "epoch": 5.930217871802968,
      "grad_norm": 4.044006824493408,
      "learning_rate": 3.9082728133880646e-05,
      "loss": 0.6879,
      "step": 18781
    },
    {
      "epoch": 5.930533628039154,
      "grad_norm": 3.819124460220337,
      "learning_rate": 3.908185103322458e-05,
      "loss": 0.5533,
      "step": 18782
    },
    {
      "epoch": 5.93084938427534,
      "grad_norm": 4.173369884490967,
      "learning_rate": 3.90809739325685e-05,
      "loss": 0.6433,
      "step": 18783
    },
    {
      "epoch": 5.931165140511525,
      "grad_norm": 4.157179832458496,
      "learning_rate": 3.9080096831912435e-05,
      "loss": 0.6397,
      "step": 18784
    },
    {
      "epoch": 5.93148089674771,
      "grad_norm": 4.426692008972168,
      "learning_rate": 3.907921973125636e-05,
      "loss": 0.5833,
      "step": 18785
    },
    {
      "epoch": 5.931796652983897,
      "grad_norm": 3.8783156871795654,
      "learning_rate": 3.907834263060029e-05,
      "loss": 0.7033,
      "step": 18786
    },
    {
      "epoch": 5.932112409220082,
      "grad_norm": 4.8262457847595215,
      "learning_rate": 3.9077465529944216e-05,
      "loss": 0.586,
      "step": 18787
    },
    {
      "epoch": 5.932428165456268,
      "grad_norm": 3.692805290222168,
      "learning_rate": 3.907658842928815e-05,
      "loss": 0.5359,
      "step": 18788
    },
    {
      "epoch": 5.932743921692453,
      "grad_norm": 4.686226844787598,
      "learning_rate": 3.907571132863207e-05,
      "loss": 0.6697,
      "step": 18789
    },
    {
      "epoch": 5.9330596779286395,
      "grad_norm": 4.2579345703125,
      "learning_rate": 3.9074834227976004e-05,
      "loss": 0.5723,
      "step": 18790
    },
    {
      "epoch": 5.933375434164825,
      "grad_norm": 3.7898786067962646,
      "learning_rate": 3.907395712731993e-05,
      "loss": 0.5208,
      "step": 18791
    },
    {
      "epoch": 5.93369119040101,
      "grad_norm": 4.704123020172119,
      "learning_rate": 3.907308002666386e-05,
      "loss": 0.6122,
      "step": 18792
    },
    {
      "epoch": 5.934006946637196,
      "grad_norm": 3.9642062187194824,
      "learning_rate": 3.907220292600779e-05,
      "loss": 0.5735,
      "step": 18793
    },
    {
      "epoch": 5.934322702873382,
      "grad_norm": 4.559034824371338,
      "learning_rate": 3.907132582535172e-05,
      "loss": 0.6851,
      "step": 18794
    },
    {
      "epoch": 5.934638459109568,
      "grad_norm": 4.619158744812012,
      "learning_rate": 3.907044872469565e-05,
      "loss": 0.5849,
      "step": 18795
    },
    {
      "epoch": 5.934954215345753,
      "grad_norm": 4.350985527038574,
      "learning_rate": 3.9069571624039574e-05,
      "loss": 0.6221,
      "step": 18796
    },
    {
      "epoch": 5.935269971581938,
      "grad_norm": 4.069890975952148,
      "learning_rate": 3.9068694523383506e-05,
      "loss": 0.641,
      "step": 18797
    },
    {
      "epoch": 5.9355857278181245,
      "grad_norm": 3.49678897857666,
      "learning_rate": 3.906781742272743e-05,
      "loss": 0.6364,
      "step": 18798
    },
    {
      "epoch": 5.93590148405431,
      "grad_norm": 4.302653789520264,
      "learning_rate": 3.906694032207136e-05,
      "loss": 0.5608,
      "step": 18799
    },
    {
      "epoch": 5.936217240290496,
      "grad_norm": 5.717174053192139,
      "learning_rate": 3.906606322141529e-05,
      "loss": 0.6568,
      "step": 18800
    },
    {
      "epoch": 5.936217240290496,
      "eval_loss": 0.8300952911376953,
      "eval_runtime": 4.5546,
      "eval_samples_per_second": 546.704,
      "eval_steps_per_second": 4.611,
      "step": 18800
    },
    {
      "epoch": 5.936532996526681,
      "grad_norm": 4.6024909019470215,
      "learning_rate": 3.9065186120759226e-05,
      "loss": 0.7161,
      "step": 18801
    },
    {
      "epoch": 5.936848752762867,
      "grad_norm": 3.705082416534424,
      "learning_rate": 3.906430902010315e-05,
      "loss": 0.5402,
      "step": 18802
    },
    {
      "epoch": 5.937164508999053,
      "grad_norm": 4.083139419555664,
      "learning_rate": 3.906343191944708e-05,
      "loss": 0.6379,
      "step": 18803
    },
    {
      "epoch": 5.937480265235238,
      "grad_norm": 4.0777459144592285,
      "learning_rate": 3.906255481879101e-05,
      "loss": 0.5995,
      "step": 18804
    },
    {
      "epoch": 5.937796021471424,
      "grad_norm": 4.24589729309082,
      "learning_rate": 3.906167771813494e-05,
      "loss": 0.6474,
      "step": 18805
    },
    {
      "epoch": 5.93811177770761,
      "grad_norm": 4.374648571014404,
      "learning_rate": 3.9060800617478864e-05,
      "loss": 0.5162,
      "step": 18806
    },
    {
      "epoch": 5.938427533943796,
      "grad_norm": 4.282017230987549,
      "learning_rate": 3.9059923516822795e-05,
      "loss": 0.5774,
      "step": 18807
    },
    {
      "epoch": 5.938743290179981,
      "grad_norm": 4.626824378967285,
      "learning_rate": 3.905904641616672e-05,
      "loss": 0.6586,
      "step": 18808
    },
    {
      "epoch": 5.939059046416166,
      "grad_norm": 4.637310981750488,
      "learning_rate": 3.905816931551065e-05,
      "loss": 0.6212,
      "step": 18809
    },
    {
      "epoch": 5.9393748026523525,
      "grad_norm": 3.7339024543762207,
      "learning_rate": 3.905729221485458e-05,
      "loss": 0.5643,
      "step": 18810
    },
    {
      "epoch": 5.939690558888538,
      "grad_norm": 4.087708473205566,
      "learning_rate": 3.905641511419851e-05,
      "loss": 0.7555,
      "step": 18811
    },
    {
      "epoch": 5.940006315124724,
      "grad_norm": 4.640169620513916,
      "learning_rate": 3.9055538013542434e-05,
      "loss": 0.7027,
      "step": 18812
    },
    {
      "epoch": 5.940322071360909,
      "grad_norm": 4.241949558258057,
      "learning_rate": 3.9054660912886365e-05,
      "loss": 0.5431,
      "step": 18813
    },
    {
      "epoch": 5.9406378275970955,
      "grad_norm": 4.713444709777832,
      "learning_rate": 3.905378381223029e-05,
      "loss": 0.642,
      "step": 18814
    },
    {
      "epoch": 5.940953583833281,
      "grad_norm": 4.61072301864624,
      "learning_rate": 3.905290671157422e-05,
      "loss": 0.6613,
      "step": 18815
    },
    {
      "epoch": 5.941269340069466,
      "grad_norm": 4.088028907775879,
      "learning_rate": 3.9052029610918154e-05,
      "loss": 0.6121,
      "step": 18816
    },
    {
      "epoch": 5.941585096305652,
      "grad_norm": 4.215367794036865,
      "learning_rate": 3.905115251026208e-05,
      "loss": 0.7005,
      "step": 18817
    },
    {
      "epoch": 5.9419008525418375,
      "grad_norm": 4.215217113494873,
      "learning_rate": 3.905027540960601e-05,
      "loss": 0.6235,
      "step": 18818
    },
    {
      "epoch": 5.942216608778024,
      "grad_norm": 4.643846035003662,
      "learning_rate": 3.9049398308949935e-05,
      "loss": 0.5577,
      "step": 18819
    },
    {
      "epoch": 5.942532365014209,
      "grad_norm": 6.016253471374512,
      "learning_rate": 3.904852120829387e-05,
      "loss": 0.6879,
      "step": 18820
    },
    {
      "epoch": 5.942848121250394,
      "grad_norm": 4.6054253578186035,
      "learning_rate": 3.904764410763779e-05,
      "loss": 0.6082,
      "step": 18821
    },
    {
      "epoch": 5.9431638774865805,
      "grad_norm": 3.784531593322754,
      "learning_rate": 3.904676700698172e-05,
      "loss": 0.5669,
      "step": 18822
    },
    {
      "epoch": 5.943479633722766,
      "grad_norm": 4.218311786651611,
      "learning_rate": 3.904588990632565e-05,
      "loss": 0.6562,
      "step": 18823
    },
    {
      "epoch": 5.943795389958952,
      "grad_norm": 4.819983005523682,
      "learning_rate": 3.904501280566958e-05,
      "loss": 0.6978,
      "step": 18824
    },
    {
      "epoch": 5.944111146195137,
      "grad_norm": 4.544519424438477,
      "learning_rate": 3.9044135705013505e-05,
      "loss": 0.7266,
      "step": 18825
    },
    {
      "epoch": 5.9444269024313225,
      "grad_norm": 4.903228759765625,
      "learning_rate": 3.9043258604357436e-05,
      "loss": 0.6446,
      "step": 18826
    },
    {
      "epoch": 5.944742658667509,
      "grad_norm": 4.669501781463623,
      "learning_rate": 3.904238150370136e-05,
      "loss": 0.751,
      "step": 18827
    },
    {
      "epoch": 5.945058414903694,
      "grad_norm": 3.8431460857391357,
      "learning_rate": 3.90415044030453e-05,
      "loss": 0.5964,
      "step": 18828
    },
    {
      "epoch": 5.94537417113988,
      "grad_norm": 4.141731262207031,
      "learning_rate": 3.9040627302389225e-05,
      "loss": 0.697,
      "step": 18829
    },
    {
      "epoch": 5.9456899273760655,
      "grad_norm": 3.830568790435791,
      "learning_rate": 3.9039750201733156e-05,
      "loss": 0.599,
      "step": 18830
    },
    {
      "epoch": 5.946005683612252,
      "grad_norm": 3.399141788482666,
      "learning_rate": 3.903887310107708e-05,
      "loss": 0.5637,
      "step": 18831
    },
    {
      "epoch": 5.946321439848437,
      "grad_norm": 4.143897533416748,
      "learning_rate": 3.903799600042101e-05,
      "loss": 0.6386,
      "step": 18832
    },
    {
      "epoch": 5.946637196084623,
      "grad_norm": 4.790981769561768,
      "learning_rate": 3.903711889976494e-05,
      "loss": 0.658,
      "step": 18833
    },
    {
      "epoch": 5.9469529523208084,
      "grad_norm": 4.325262069702148,
      "learning_rate": 3.903624179910887e-05,
      "loss": 0.6705,
      "step": 18834
    },
    {
      "epoch": 5.947268708556994,
      "grad_norm": 3.5916802883148193,
      "learning_rate": 3.9035364698452794e-05,
      "loss": 0.5954,
      "step": 18835
    },
    {
      "epoch": 5.94758446479318,
      "grad_norm": 4.019606113433838,
      "learning_rate": 3.9034487597796726e-05,
      "loss": 0.6373,
      "step": 18836
    },
    {
      "epoch": 5.947900221029365,
      "grad_norm": 4.5609846115112305,
      "learning_rate": 3.903361049714066e-05,
      "loss": 0.6884,
      "step": 18837
    },
    {
      "epoch": 5.948215977265551,
      "grad_norm": 3.7040905952453613,
      "learning_rate": 3.903273339648458e-05,
      "loss": 0.692,
      "step": 18838
    },
    {
      "epoch": 5.948531733501737,
      "grad_norm": 3.5629689693450928,
      "learning_rate": 3.9031856295828514e-05,
      "loss": 0.6118,
      "step": 18839
    },
    {
      "epoch": 5.948847489737922,
      "grad_norm": 3.8451056480407715,
      "learning_rate": 3.903097919517244e-05,
      "loss": 0.6706,
      "step": 18840
    },
    {
      "epoch": 5.949163245974108,
      "grad_norm": 3.6425113677978516,
      "learning_rate": 3.903010209451637e-05,
      "loss": 0.5768,
      "step": 18841
    },
    {
      "epoch": 5.9494790022102935,
      "grad_norm": 4.9278645515441895,
      "learning_rate": 3.9029224993860296e-05,
      "loss": 0.8576,
      "step": 18842
    },
    {
      "epoch": 5.94979475844648,
      "grad_norm": 5.231960296630859,
      "learning_rate": 3.902834789320423e-05,
      "loss": 0.7184,
      "step": 18843
    },
    {
      "epoch": 5.950110514682665,
      "grad_norm": 3.663738965988159,
      "learning_rate": 3.902747079254815e-05,
      "loss": 0.7413,
      "step": 18844
    },
    {
      "epoch": 5.95042627091885,
      "grad_norm": 3.4977939128875732,
      "learning_rate": 3.9026593691892084e-05,
      "loss": 0.6004,
      "step": 18845
    },
    {
      "epoch": 5.950742027155036,
      "grad_norm": 4.214312553405762,
      "learning_rate": 3.902571659123601e-05,
      "loss": 0.6155,
      "step": 18846
    },
    {
      "epoch": 5.951057783391222,
      "grad_norm": 4.878962516784668,
      "learning_rate": 3.902483949057994e-05,
      "loss": 0.6354,
      "step": 18847
    },
    {
      "epoch": 5.951373539627408,
      "grad_norm": 4.472371578216553,
      "learning_rate": 3.9023962389923866e-05,
      "loss": 0.57,
      "step": 18848
    },
    {
      "epoch": 5.951689295863593,
      "grad_norm": 4.949447154998779,
      "learning_rate": 3.90230852892678e-05,
      "loss": 0.7298,
      "step": 18849
    },
    {
      "epoch": 5.9520050520997785,
      "grad_norm": 4.298762321472168,
      "learning_rate": 3.902220818861172e-05,
      "loss": 0.7154,
      "step": 18850
    },
    {
      "epoch": 5.952320808335965,
      "grad_norm": 2.98333477973938,
      "learning_rate": 3.9021331087955654e-05,
      "loss": 0.5667,
      "step": 18851
    },
    {
      "epoch": 5.95263656457215,
      "grad_norm": 3.1881630420684814,
      "learning_rate": 3.902045398729958e-05,
      "loss": 0.5855,
      "step": 18852
    },
    {
      "epoch": 5.952952320808336,
      "grad_norm": 3.294631242752075,
      "learning_rate": 3.901957688664351e-05,
      "loss": 0.5824,
      "step": 18853
    },
    {
      "epoch": 5.953268077044521,
      "grad_norm": 4.970726490020752,
      "learning_rate": 3.901869978598744e-05,
      "loss": 0.6393,
      "step": 18854
    },
    {
      "epoch": 5.953583833280708,
      "grad_norm": 4.831282615661621,
      "learning_rate": 3.901782268533137e-05,
      "loss": 0.5548,
      "step": 18855
    },
    {
      "epoch": 5.953899589516893,
      "grad_norm": 4.041653633117676,
      "learning_rate": 3.90169455846753e-05,
      "loss": 0.7307,
      "step": 18856
    },
    {
      "epoch": 5.954215345753079,
      "grad_norm": 3.9860076904296875,
      "learning_rate": 3.901606848401923e-05,
      "loss": 0.6561,
      "step": 18857
    },
    {
      "epoch": 5.954531101989264,
      "grad_norm": 4.285194396972656,
      "learning_rate": 3.901519138336316e-05,
      "loss": 0.6252,
      "step": 18858
    },
    {
      "epoch": 5.95484685822545,
      "grad_norm": 3.41084361076355,
      "learning_rate": 3.901431428270709e-05,
      "loss": 0.6669,
      "step": 18859
    },
    {
      "epoch": 5.955162614461636,
      "grad_norm": 3.9280636310577393,
      "learning_rate": 3.901343718205102e-05,
      "loss": 0.6839,
      "step": 18860
    },
    {
      "epoch": 5.955478370697821,
      "grad_norm": 3.7642345428466797,
      "learning_rate": 3.9012560081394944e-05,
      "loss": 0.624,
      "step": 18861
    },
    {
      "epoch": 5.955794126934007,
      "grad_norm": 3.486875057220459,
      "learning_rate": 3.9011682980738875e-05,
      "loss": 0.6255,
      "step": 18862
    },
    {
      "epoch": 5.956109883170193,
      "grad_norm": 4.520173072814941,
      "learning_rate": 3.90108058800828e-05,
      "loss": 0.5987,
      "step": 18863
    },
    {
      "epoch": 5.956425639406378,
      "grad_norm": 3.8460958003997803,
      "learning_rate": 3.900992877942673e-05,
      "loss": 0.6598,
      "step": 18864
    },
    {
      "epoch": 5.956741395642564,
      "grad_norm": 5.1903862953186035,
      "learning_rate": 3.900905167877066e-05,
      "loss": 0.665,
      "step": 18865
    },
    {
      "epoch": 5.957057151878749,
      "grad_norm": 4.34146785736084,
      "learning_rate": 3.900817457811459e-05,
      "loss": 0.6393,
      "step": 18866
    },
    {
      "epoch": 5.957372908114936,
      "grad_norm": 3.4372003078460693,
      "learning_rate": 3.9007297477458513e-05,
      "loss": 0.5085,
      "step": 18867
    },
    {
      "epoch": 5.957688664351121,
      "grad_norm": 4.29457950592041,
      "learning_rate": 3.9006420376802445e-05,
      "loss": 0.7222,
      "step": 18868
    },
    {
      "epoch": 5.958004420587306,
      "grad_norm": 4.185940742492676,
      "learning_rate": 3.900554327614637e-05,
      "loss": 0.5612,
      "step": 18869
    },
    {
      "epoch": 5.958320176823492,
      "grad_norm": 4.962240695953369,
      "learning_rate": 3.90046661754903e-05,
      "loss": 0.7106,
      "step": 18870
    },
    {
      "epoch": 5.958635933059678,
      "grad_norm": 4.290449619293213,
      "learning_rate": 3.900378907483423e-05,
      "loss": 0.7254,
      "step": 18871
    },
    {
      "epoch": 5.958951689295864,
      "grad_norm": 5.265349864959717,
      "learning_rate": 3.900291197417816e-05,
      "loss": 0.6997,
      "step": 18872
    },
    {
      "epoch": 5.959267445532049,
      "grad_norm": 4.272889614105225,
      "learning_rate": 3.900203487352208e-05,
      "loss": 0.6852,
      "step": 18873
    },
    {
      "epoch": 5.959583201768235,
      "grad_norm": 4.189887523651123,
      "learning_rate": 3.9001157772866015e-05,
      "loss": 0.5737,
      "step": 18874
    },
    {
      "epoch": 5.959898958004421,
      "grad_norm": 3.3175222873687744,
      "learning_rate": 3.9000280672209947e-05,
      "loss": 0.7008,
      "step": 18875
    },
    {
      "epoch": 5.960214714240606,
      "grad_norm": 4.552657127380371,
      "learning_rate": 3.899940357155387e-05,
      "loss": 0.7315,
      "step": 18876
    },
    {
      "epoch": 5.960530470476792,
      "grad_norm": 4.664196014404297,
      "learning_rate": 3.89985264708978e-05,
      "loss": 0.6749,
      "step": 18877
    },
    {
      "epoch": 5.960846226712977,
      "grad_norm": 3.758572578430176,
      "learning_rate": 3.899764937024173e-05,
      "loss": 0.5831,
      "step": 18878
    },
    {
      "epoch": 5.9611619829491636,
      "grad_norm": 3.837703227996826,
      "learning_rate": 3.899677226958566e-05,
      "loss": 0.5767,
      "step": 18879
    },
    {
      "epoch": 5.961477739185349,
      "grad_norm": 4.152406215667725,
      "learning_rate": 3.8995895168929585e-05,
      "loss": 0.7552,
      "step": 18880
    },
    {
      "epoch": 5.961793495421535,
      "grad_norm": 4.261078357696533,
      "learning_rate": 3.8995018068273516e-05,
      "loss": 0.6448,
      "step": 18881
    },
    {
      "epoch": 5.96210925165772,
      "grad_norm": 4.176559925079346,
      "learning_rate": 3.899414096761744e-05,
      "loss": 0.7431,
      "step": 18882
    },
    {
      "epoch": 5.962425007893906,
      "grad_norm": 3.757676124572754,
      "learning_rate": 3.899326386696138e-05,
      "loss": 0.6709,
      "step": 18883
    },
    {
      "epoch": 5.962740764130092,
      "grad_norm": 3.8098340034484863,
      "learning_rate": 3.8992386766305305e-05,
      "loss": 0.6243,
      "step": 18884
    },
    {
      "epoch": 5.963056520366277,
      "grad_norm": 4.359403133392334,
      "learning_rate": 3.8991509665649236e-05,
      "loss": 0.6505,
      "step": 18885
    },
    {
      "epoch": 5.963372276602463,
      "grad_norm": 4.229589462280273,
      "learning_rate": 3.899063256499316e-05,
      "loss": 0.739,
      "step": 18886
    },
    {
      "epoch": 5.963688032838649,
      "grad_norm": 4.459061145782471,
      "learning_rate": 3.898975546433709e-05,
      "loss": 0.7639,
      "step": 18887
    },
    {
      "epoch": 5.964003789074834,
      "grad_norm": 3.537951946258545,
      "learning_rate": 3.898887836368102e-05,
      "loss": 0.6505,
      "step": 18888
    },
    {
      "epoch": 5.96431954531102,
      "grad_norm": 3.496297597885132,
      "learning_rate": 3.898800126302495e-05,
      "loss": 0.6579,
      "step": 18889
    },
    {
      "epoch": 5.964635301547205,
      "grad_norm": 3.5102860927581787,
      "learning_rate": 3.8987124162368874e-05,
      "loss": 0.6081,
      "step": 18890
    },
    {
      "epoch": 5.9649510577833915,
      "grad_norm": 3.22346830368042,
      "learning_rate": 3.8986247061712806e-05,
      "loss": 0.6042,
      "step": 18891
    },
    {
      "epoch": 5.965266814019577,
      "grad_norm": 4.178524017333984,
      "learning_rate": 3.898536996105673e-05,
      "loss": 0.6924,
      "step": 18892
    },
    {
      "epoch": 5.965582570255762,
      "grad_norm": 3.528315782546997,
      "learning_rate": 3.898449286040066e-05,
      "loss": 0.6151,
      "step": 18893
    },
    {
      "epoch": 5.965898326491948,
      "grad_norm": 3.3666622638702393,
      "learning_rate": 3.898361575974459e-05,
      "loss": 0.5786,
      "step": 18894
    },
    {
      "epoch": 5.966214082728134,
      "grad_norm": 3.8755762577056885,
      "learning_rate": 3.898273865908852e-05,
      "loss": 0.66,
      "step": 18895
    },
    {
      "epoch": 5.96652983896432,
      "grad_norm": 3.9221560955047607,
      "learning_rate": 3.898186155843245e-05,
      "loss": 0.6255,
      "step": 18896
    },
    {
      "epoch": 5.966845595200505,
      "grad_norm": 3.6047658920288086,
      "learning_rate": 3.8980984457776376e-05,
      "loss": 0.6953,
      "step": 18897
    },
    {
      "epoch": 5.967161351436691,
      "grad_norm": 4.549970626831055,
      "learning_rate": 3.898010735712031e-05,
      "loss": 0.7127,
      "step": 18898
    },
    {
      "epoch": 5.9674771076728765,
      "grad_norm": 3.789274215698242,
      "learning_rate": 3.897923025646423e-05,
      "loss": 0.677,
      "step": 18899
    },
    {
      "epoch": 5.967792863909062,
      "grad_norm": 4.006741523742676,
      "learning_rate": 3.8978353155808164e-05,
      "loss": 0.6589,
      "step": 18900
    },
    {
      "epoch": 5.968108620145248,
      "grad_norm": 4.520087718963623,
      "learning_rate": 3.897747605515209e-05,
      "loss": 0.7401,
      "step": 18901
    },
    {
      "epoch": 5.968424376381433,
      "grad_norm": 4.151277542114258,
      "learning_rate": 3.897659895449602e-05,
      "loss": 0.7658,
      "step": 18902
    },
    {
      "epoch": 5.9687401326176195,
      "grad_norm": 3.732182741165161,
      "learning_rate": 3.8975721853839946e-05,
      "loss": 0.6379,
      "step": 18903
    },
    {
      "epoch": 5.969055888853805,
      "grad_norm": 3.482292413711548,
      "learning_rate": 3.897484475318388e-05,
      "loss": 0.5838,
      "step": 18904
    },
    {
      "epoch": 5.969371645089991,
      "grad_norm": 3.7362356185913086,
      "learning_rate": 3.89739676525278e-05,
      "loss": 0.613,
      "step": 18905
    },
    {
      "epoch": 5.969687401326176,
      "grad_norm": 3.3713531494140625,
      "learning_rate": 3.8973090551871734e-05,
      "loss": 0.5849,
      "step": 18906
    },
    {
      "epoch": 5.9700031575623616,
      "grad_norm": 3.8926310539245605,
      "learning_rate": 3.897221345121566e-05,
      "loss": 0.7345,
      "step": 18907
    },
    {
      "epoch": 5.970318913798548,
      "grad_norm": 3.4860219955444336,
      "learning_rate": 3.897133635055959e-05,
      "loss": 0.5874,
      "step": 18908
    },
    {
      "epoch": 5.970634670034733,
      "grad_norm": 3.8898978233337402,
      "learning_rate": 3.8970459249903515e-05,
      "loss": 0.6792,
      "step": 18909
    },
    {
      "epoch": 5.970950426270919,
      "grad_norm": 6.814505100250244,
      "learning_rate": 3.8969582149247454e-05,
      "loss": 0.6327,
      "step": 18910
    },
    {
      "epoch": 5.9712661825071045,
      "grad_norm": 3.743290662765503,
      "learning_rate": 3.896870504859138e-05,
      "loss": 0.6104,
      "step": 18911
    },
    {
      "epoch": 5.97158193874329,
      "grad_norm": 4.079965114593506,
      "learning_rate": 3.896782794793531e-05,
      "loss": 0.7325,
      "step": 18912
    },
    {
      "epoch": 5.971897694979476,
      "grad_norm": 3.4552860260009766,
      "learning_rate": 3.8966950847279235e-05,
      "loss": 0.5602,
      "step": 18913
    },
    {
      "epoch": 5.972213451215661,
      "grad_norm": 3.9316840171813965,
      "learning_rate": 3.896607374662317e-05,
      "loss": 0.7264,
      "step": 18914
    },
    {
      "epoch": 5.9725292074518475,
      "grad_norm": 3.747856616973877,
      "learning_rate": 3.896519664596709e-05,
      "loss": 0.6086,
      "step": 18915
    },
    {
      "epoch": 5.972844963688033,
      "grad_norm": 4.244474411010742,
      "learning_rate": 3.8964319545311024e-05,
      "loss": 0.5568,
      "step": 18916
    },
    {
      "epoch": 5.973160719924218,
      "grad_norm": 3.8879194259643555,
      "learning_rate": 3.896344244465495e-05,
      "loss": 0.6833,
      "step": 18917
    },
    {
      "epoch": 5.973476476160404,
      "grad_norm": 4.529118537902832,
      "learning_rate": 3.896256534399888e-05,
      "loss": 0.7568,
      "step": 18918
    },
    {
      "epoch": 5.9737922323965895,
      "grad_norm": 3.374763250350952,
      "learning_rate": 3.896168824334281e-05,
      "loss": 0.5423,
      "step": 18919
    },
    {
      "epoch": 5.974107988632776,
      "grad_norm": 3.2759933471679688,
      "learning_rate": 3.896081114268674e-05,
      "loss": 0.68,
      "step": 18920
    },
    {
      "epoch": 5.974423744868961,
      "grad_norm": 4.714364528656006,
      "learning_rate": 3.895993404203067e-05,
      "loss": 0.7045,
      "step": 18921
    },
    {
      "epoch": 5.974739501105147,
      "grad_norm": 3.49896240234375,
      "learning_rate": 3.895905694137459e-05,
      "loss": 0.6424,
      "step": 18922
    },
    {
      "epoch": 5.9750552573413325,
      "grad_norm": 4.126099109649658,
      "learning_rate": 3.8958179840718525e-05,
      "loss": 0.5974,
      "step": 18923
    },
    {
      "epoch": 5.975371013577519,
      "grad_norm": 3.7124316692352295,
      "learning_rate": 3.895730274006245e-05,
      "loss": 0.6326,
      "step": 18924
    },
    {
      "epoch": 5.975686769813704,
      "grad_norm": 4.475553035736084,
      "learning_rate": 3.895642563940638e-05,
      "loss": 0.6638,
      "step": 18925
    },
    {
      "epoch": 5.976002526049889,
      "grad_norm": 4.090775489807129,
      "learning_rate": 3.8955548538750307e-05,
      "loss": 0.551,
      "step": 18926
    },
    {
      "epoch": 5.976318282286075,
      "grad_norm": 4.955806732177734,
      "learning_rate": 3.895467143809424e-05,
      "loss": 0.8358,
      "step": 18927
    },
    {
      "epoch": 5.976634038522261,
      "grad_norm": 4.25096321105957,
      "learning_rate": 3.895379433743816e-05,
      "loss": 0.6556,
      "step": 18928
    },
    {
      "epoch": 5.976949794758447,
      "grad_norm": 3.884253978729248,
      "learning_rate": 3.8952917236782095e-05,
      "loss": 0.6521,
      "step": 18929
    },
    {
      "epoch": 5.977265550994632,
      "grad_norm": 4.031165599822998,
      "learning_rate": 3.895204013612602e-05,
      "loss": 0.6849,
      "step": 18930
    },
    {
      "epoch": 5.9775813072308175,
      "grad_norm": 3.3256595134735107,
      "learning_rate": 3.895116303546995e-05,
      "loss": 0.6039,
      "step": 18931
    },
    {
      "epoch": 5.977897063467004,
      "grad_norm": 4.288661956787109,
      "learning_rate": 3.8950285934813876e-05,
      "loss": 0.6347,
      "step": 18932
    },
    {
      "epoch": 5.978212819703189,
      "grad_norm": 3.7140979766845703,
      "learning_rate": 3.894940883415781e-05,
      "loss": 0.5172,
      "step": 18933
    },
    {
      "epoch": 5.978528575939375,
      "grad_norm": 4.336061954498291,
      "learning_rate": 3.894853173350174e-05,
      "loss": 0.6755,
      "step": 18934
    },
    {
      "epoch": 5.97884433217556,
      "grad_norm": 4.024496555328369,
      "learning_rate": 3.8947654632845665e-05,
      "loss": 0.6485,
      "step": 18935
    },
    {
      "epoch": 5.979160088411746,
      "grad_norm": 4.245024681091309,
      "learning_rate": 3.8946777532189596e-05,
      "loss": 0.6733,
      "step": 18936
    },
    {
      "epoch": 5.979475844647932,
      "grad_norm": 4.061288356781006,
      "learning_rate": 3.894590043153353e-05,
      "loss": 0.7492,
      "step": 18937
    },
    {
      "epoch": 5.979791600884117,
      "grad_norm": 4.518464088439941,
      "learning_rate": 3.894502333087745e-05,
      "loss": 0.6669,
      "step": 18938
    },
    {
      "epoch": 5.980107357120303,
      "grad_norm": 3.7718586921691895,
      "learning_rate": 3.8944146230221385e-05,
      "loss": 0.5586,
      "step": 18939
    },
    {
      "epoch": 5.980423113356489,
      "grad_norm": 3.911531686782837,
      "learning_rate": 3.8943269129565316e-05,
      "loss": 0.5767,
      "step": 18940
    },
    {
      "epoch": 5.980738869592674,
      "grad_norm": 3.4749836921691895,
      "learning_rate": 3.894239202890924e-05,
      "loss": 0.501,
      "step": 18941
    },
    {
      "epoch": 5.98105462582886,
      "grad_norm": 4.117038726806641,
      "learning_rate": 3.894151492825317e-05,
      "loss": 0.5731,
      "step": 18942
    },
    {
      "epoch": 5.9813703820650455,
      "grad_norm": 4.986505508422852,
      "learning_rate": 3.89406378275971e-05,
      "loss": 0.7021,
      "step": 18943
    },
    {
      "epoch": 5.981686138301232,
      "grad_norm": 3.288877487182617,
      "learning_rate": 3.893976072694103e-05,
      "loss": 0.6037,
      "step": 18944
    },
    {
      "epoch": 5.982001894537417,
      "grad_norm": 3.7959442138671875,
      "learning_rate": 3.8938883626284954e-05,
      "loss": 0.5846,
      "step": 18945
    },
    {
      "epoch": 5.982317650773603,
      "grad_norm": 3.9501078128814697,
      "learning_rate": 3.8938006525628886e-05,
      "loss": 0.7354,
      "step": 18946
    },
    {
      "epoch": 5.982633407009788,
      "grad_norm": 3.919703483581543,
      "learning_rate": 3.893712942497281e-05,
      "loss": 0.6975,
      "step": 18947
    },
    {
      "epoch": 5.982949163245975,
      "grad_norm": 3.461223602294922,
      "learning_rate": 3.893625232431674e-05,
      "loss": 0.5868,
      "step": 18948
    },
    {
      "epoch": 5.98326491948216,
      "grad_norm": 3.8120923042297363,
      "learning_rate": 3.893537522366067e-05,
      "loss": 0.611,
      "step": 18949
    },
    {
      "epoch": 5.983580675718345,
      "grad_norm": 4.187610626220703,
      "learning_rate": 3.89344981230046e-05,
      "loss": 0.6885,
      "step": 18950
    },
    {
      "epoch": 5.983896431954531,
      "grad_norm": 4.664694786071777,
      "learning_rate": 3.8933621022348524e-05,
      "loss": 0.719,
      "step": 18951
    },
    {
      "epoch": 5.984212188190717,
      "grad_norm": 3.2492499351501465,
      "learning_rate": 3.8932743921692456e-05,
      "loss": 0.6092,
      "step": 18952
    },
    {
      "epoch": 5.984527944426903,
      "grad_norm": 3.971869945526123,
      "learning_rate": 3.893186682103638e-05,
      "loss": 0.6312,
      "step": 18953
    },
    {
      "epoch": 5.984843700663088,
      "grad_norm": 3.5039002895355225,
      "learning_rate": 3.893098972038031e-05,
      "loss": 0.5788,
      "step": 18954
    },
    {
      "epoch": 5.985159456899273,
      "grad_norm": 3.3783726692199707,
      "learning_rate": 3.893011261972424e-05,
      "loss": 0.6106,
      "step": 18955
    },
    {
      "epoch": 5.98547521313546,
      "grad_norm": 4.033379554748535,
      "learning_rate": 3.892923551906817e-05,
      "loss": 0.6064,
      "step": 18956
    },
    {
      "epoch": 5.985790969371645,
      "grad_norm": 3.7769017219543457,
      "learning_rate": 3.89283584184121e-05,
      "loss": 0.5548,
      "step": 18957
    },
    {
      "epoch": 5.986106725607831,
      "grad_norm": 3.8335189819335938,
      "learning_rate": 3.8927481317756025e-05,
      "loss": 0.6129,
      "step": 18958
    },
    {
      "epoch": 5.986422481844016,
      "grad_norm": 4.313701152801514,
      "learning_rate": 3.892660421709996e-05,
      "loss": 0.6956,
      "step": 18959
    },
    {
      "epoch": 5.986738238080202,
      "grad_norm": 3.4559295177459717,
      "learning_rate": 3.892572711644388e-05,
      "loss": 0.4915,
      "step": 18960
    },
    {
      "epoch": 5.987053994316388,
      "grad_norm": 4.61830997467041,
      "learning_rate": 3.8924850015787814e-05,
      "loss": 0.6364,
      "step": 18961
    },
    {
      "epoch": 5.987369750552573,
      "grad_norm": 3.8133115768432617,
      "learning_rate": 3.892397291513174e-05,
      "loss": 0.6765,
      "step": 18962
    },
    {
      "epoch": 5.987685506788759,
      "grad_norm": 4.1560139656066895,
      "learning_rate": 3.892309581447567e-05,
      "loss": 0.6428,
      "step": 18963
    },
    {
      "epoch": 5.988001263024945,
      "grad_norm": 4.481167793273926,
      "learning_rate": 3.8922218713819595e-05,
      "loss": 0.7521,
      "step": 18964
    },
    {
      "epoch": 5.98831701926113,
      "grad_norm": 3.4522738456726074,
      "learning_rate": 3.8921341613163534e-05,
      "loss": 0.6904,
      "step": 18965
    },
    {
      "epoch": 5.988632775497316,
      "grad_norm": 3.6555335521698,
      "learning_rate": 3.892046451250746e-05,
      "loss": 0.6199,
      "step": 18966
    },
    {
      "epoch": 5.988948531733501,
      "grad_norm": 4.404059886932373,
      "learning_rate": 3.891958741185139e-05,
      "loss": 0.6677,
      "step": 18967
    },
    {
      "epoch": 5.989264287969688,
      "grad_norm": 3.6764934062957764,
      "learning_rate": 3.8918710311195315e-05,
      "loss": 0.645,
      "step": 18968
    },
    {
      "epoch": 5.989580044205873,
      "grad_norm": 3.6261258125305176,
      "learning_rate": 3.891783321053925e-05,
      "loss": 0.6484,
      "step": 18969
    },
    {
      "epoch": 5.989895800442059,
      "grad_norm": 4.255712985992432,
      "learning_rate": 3.891695610988317e-05,
      "loss": 0.6381,
      "step": 18970
    },
    {
      "epoch": 5.990211556678244,
      "grad_norm": 3.5016050338745117,
      "learning_rate": 3.8916079009227103e-05,
      "loss": 0.6487,
      "step": 18971
    },
    {
      "epoch": 5.9905273129144305,
      "grad_norm": 3.8803608417510986,
      "learning_rate": 3.891520190857103e-05,
      "loss": 0.7203,
      "step": 18972
    },
    {
      "epoch": 5.990843069150616,
      "grad_norm": 3.8753106594085693,
      "learning_rate": 3.891432480791496e-05,
      "loss": 0.6381,
      "step": 18973
    },
    {
      "epoch": 5.991158825386801,
      "grad_norm": 4.02841854095459,
      "learning_rate": 3.8913447707258885e-05,
      "loss": 0.681,
      "step": 18974
    },
    {
      "epoch": 5.991474581622987,
      "grad_norm": 3.7408406734466553,
      "learning_rate": 3.891257060660282e-05,
      "loss": 0.7751,
      "step": 18975
    },
    {
      "epoch": 5.991790337859173,
      "grad_norm": 3.762449026107788,
      "learning_rate": 3.891169350594674e-05,
      "loss": 0.6609,
      "step": 18976
    },
    {
      "epoch": 5.992106094095359,
      "grad_norm": 4.192922592163086,
      "learning_rate": 3.891081640529067e-05,
      "loss": 0.6961,
      "step": 18977
    },
    {
      "epoch": 5.992421850331544,
      "grad_norm": 3.8207297325134277,
      "learning_rate": 3.8909939304634605e-05,
      "loss": 0.7133,
      "step": 18978
    },
    {
      "epoch": 5.992737606567729,
      "grad_norm": 3.9674134254455566,
      "learning_rate": 3.890906220397853e-05,
      "loss": 0.733,
      "step": 18979
    },
    {
      "epoch": 5.9930533628039155,
      "grad_norm": 4.779201030731201,
      "learning_rate": 3.890818510332246e-05,
      "loss": 0.7767,
      "step": 18980
    },
    {
      "epoch": 5.993369119040101,
      "grad_norm": 3.4435300827026367,
      "learning_rate": 3.8907308002666386e-05,
      "loss": 0.6391,
      "step": 18981
    },
    {
      "epoch": 5.993684875276287,
      "grad_norm": 4.259660243988037,
      "learning_rate": 3.890643090201032e-05,
      "loss": 0.693,
      "step": 18982
    },
    {
      "epoch": 5.994000631512472,
      "grad_norm": 3.067924737930298,
      "learning_rate": 3.890555380135424e-05,
      "loss": 0.5481,
      "step": 18983
    },
    {
      "epoch": 5.994316387748658,
      "grad_norm": 4.487456321716309,
      "learning_rate": 3.8904676700698175e-05,
      "loss": 0.65,
      "step": 18984
    },
    {
      "epoch": 5.994632143984844,
      "grad_norm": 4.269394874572754,
      "learning_rate": 3.89037996000421e-05,
      "loss": 0.6601,
      "step": 18985
    },
    {
      "epoch": 5.994947900221029,
      "grad_norm": 3.9205403327941895,
      "learning_rate": 3.890292249938603e-05,
      "loss": 0.6402,
      "step": 18986
    },
    {
      "epoch": 5.995263656457215,
      "grad_norm": 5.322890281677246,
      "learning_rate": 3.8902045398729956e-05,
      "loss": 0.7199,
      "step": 18987
    },
    {
      "epoch": 5.995579412693401,
      "grad_norm": 3.437819242477417,
      "learning_rate": 3.890116829807389e-05,
      "loss": 0.5734,
      "step": 18988
    },
    {
      "epoch": 5.995895168929587,
      "grad_norm": 4.410650253295898,
      "learning_rate": 3.890029119741781e-05,
      "loss": 0.7535,
      "step": 18989
    },
    {
      "epoch": 5.996210925165772,
      "grad_norm": 3.8266942501068115,
      "learning_rate": 3.8899414096761744e-05,
      "loss": 0.6327,
      "step": 18990
    },
    {
      "epoch": 5.996526681401957,
      "grad_norm": 4.391696453094482,
      "learning_rate": 3.889853699610567e-05,
      "loss": 0.7043,
      "step": 18991
    },
    {
      "epoch": 5.9968424376381435,
      "grad_norm": 4.7826714515686035,
      "learning_rate": 3.889765989544961e-05,
      "loss": 0.5893,
      "step": 18992
    },
    {
      "epoch": 5.997158193874329,
      "grad_norm": 3.8495311737060547,
      "learning_rate": 3.889678279479353e-05,
      "loss": 0.6492,
      "step": 18993
    },
    {
      "epoch": 5.997473950110515,
      "grad_norm": 3.409900188446045,
      "learning_rate": 3.8895905694137464e-05,
      "loss": 0.627,
      "step": 18994
    },
    {
      "epoch": 5.9977897063467,
      "grad_norm": 4.025269985198975,
      "learning_rate": 3.889502859348139e-05,
      "loss": 0.6625,
      "step": 18995
    },
    {
      "epoch": 5.9981054625828865,
      "grad_norm": 4.023127555847168,
      "learning_rate": 3.889415149282532e-05,
      "loss": 0.6945,
      "step": 18996
    },
    {
      "epoch": 5.998421218819072,
      "grad_norm": 3.3018953800201416,
      "learning_rate": 3.8893274392169246e-05,
      "loss": 0.6535,
      "step": 18997
    },
    {
      "epoch": 5.998736975055257,
      "grad_norm": 3.727832078933716,
      "learning_rate": 3.889239729151318e-05,
      "loss": 0.648,
      "step": 18998
    },
    {
      "epoch": 5.999052731291443,
      "grad_norm": 3.513928174972534,
      "learning_rate": 3.88915201908571e-05,
      "loss": 0.6728,
      "step": 18999
    },
    {
      "epoch": 5.9993684875276285,
      "grad_norm": 3.486708402633667,
      "learning_rate": 3.8890643090201034e-05,
      "loss": 0.6665,
      "step": 19000
    },
    {
      "epoch": 5.9993684875276285,
      "eval_loss": 0.8296696543693542,
      "eval_runtime": 4.5555,
      "eval_samples_per_second": 546.591,
      "eval_steps_per_second": 4.61,
      "step": 19000
    },
    {
      "epoch": 5.999684243763815,
      "grad_norm": 3.540971040725708,
      "learning_rate": 3.8889765989544966e-05,
      "loss": 0.6457,
      "step": 19001
    },
    {
      "epoch": 6.0,
      "grad_norm": 3.4898509979248047,
      "learning_rate": 3.888888888888889e-05,
      "loss": 0.6284,
      "step": 19002
    },
    {
      "epoch": 6.000315756236185,
      "grad_norm": 3.3978753089904785,
      "learning_rate": 3.888801178823282e-05,
      "loss": 0.4928,
      "step": 19003
    },
    {
      "epoch": 6.0006315124723715,
      "grad_norm": 3.045703411102295,
      "learning_rate": 3.888713468757675e-05,
      "loss": 0.6155,
      "step": 19004
    },
    {
      "epoch": 6.000947268708557,
      "grad_norm": 3.5869057178497314,
      "learning_rate": 3.888625758692068e-05,
      "loss": 0.5339,
      "step": 19005
    },
    {
      "epoch": 6.001263024944743,
      "grad_norm": 3.607941150665283,
      "learning_rate": 3.8885380486264604e-05,
      "loss": 0.5633,
      "step": 19006
    },
    {
      "epoch": 6.001578781180928,
      "grad_norm": 3.6718621253967285,
      "learning_rate": 3.8884503385608536e-05,
      "loss": 0.5009,
      "step": 19007
    },
    {
      "epoch": 6.001894537417114,
      "grad_norm": 4.040913105010986,
      "learning_rate": 3.888362628495246e-05,
      "loss": 0.6574,
      "step": 19008
    },
    {
      "epoch": 6.0022102936533,
      "grad_norm": 3.4368326663970947,
      "learning_rate": 3.888274918429639e-05,
      "loss": 0.5201,
      "step": 19009
    },
    {
      "epoch": 6.002526049889485,
      "grad_norm": 3.3413798809051514,
      "learning_rate": 3.888187208364032e-05,
      "loss": 0.5808,
      "step": 19010
    },
    {
      "epoch": 6.002841806125671,
      "grad_norm": 3.398953914642334,
      "learning_rate": 3.888099498298425e-05,
      "loss": 0.589,
      "step": 19011
    },
    {
      "epoch": 6.0031575623618565,
      "grad_norm": 3.5915169715881348,
      "learning_rate": 3.8880117882328174e-05,
      "loss": 0.6572,
      "step": 19012
    },
    {
      "epoch": 6.003473318598043,
      "grad_norm": 3.4691359996795654,
      "learning_rate": 3.8879240781672105e-05,
      "loss": 0.528,
      "step": 19013
    },
    {
      "epoch": 6.003789074834228,
      "grad_norm": 3.637037515640259,
      "learning_rate": 3.887836368101603e-05,
      "loss": 0.5425,
      "step": 19014
    },
    {
      "epoch": 6.004104831070413,
      "grad_norm": 4.859009742736816,
      "learning_rate": 3.887748658035996e-05,
      "loss": 0.4287,
      "step": 19015
    },
    {
      "epoch": 6.004420587306599,
      "grad_norm": 3.7862000465393066,
      "learning_rate": 3.8876609479703894e-05,
      "loss": 0.4747,
      "step": 19016
    },
    {
      "epoch": 6.004736343542785,
      "grad_norm": 4.2946906089782715,
      "learning_rate": 3.887573237904782e-05,
      "loss": 0.5448,
      "step": 19017
    },
    {
      "epoch": 6.005052099778971,
      "grad_norm": 4.979546070098877,
      "learning_rate": 3.887485527839175e-05,
      "loss": 0.5607,
      "step": 19018
    },
    {
      "epoch": 6.005367856015156,
      "grad_norm": 3.818683385848999,
      "learning_rate": 3.887397817773568e-05,
      "loss": 0.5737,
      "step": 19019
    },
    {
      "epoch": 6.005683612251342,
      "grad_norm": 3.6675803661346436,
      "learning_rate": 3.887310107707961e-05,
      "loss": 0.4893,
      "step": 19020
    },
    {
      "epoch": 6.005999368487528,
      "grad_norm": 4.442096710205078,
      "learning_rate": 3.887222397642354e-05,
      "loss": 0.6372,
      "step": 19021
    },
    {
      "epoch": 6.006315124723713,
      "grad_norm": 4.275791645050049,
      "learning_rate": 3.887134687576747e-05,
      "loss": 0.5497,
      "step": 19022
    },
    {
      "epoch": 6.006630880959899,
      "grad_norm": 3.787243366241455,
      "learning_rate": 3.8870469775111395e-05,
      "loss": 0.4726,
      "step": 19023
    },
    {
      "epoch": 6.0069466371960845,
      "grad_norm": 3.662323236465454,
      "learning_rate": 3.886959267445533e-05,
      "loss": 0.5014,
      "step": 19024
    },
    {
      "epoch": 6.007262393432271,
      "grad_norm": 4.017457962036133,
      "learning_rate": 3.886871557379925e-05,
      "loss": 0.5375,
      "step": 19025
    },
    {
      "epoch": 6.007578149668456,
      "grad_norm": 3.984872579574585,
      "learning_rate": 3.886783847314318e-05,
      "loss": 0.3914,
      "step": 19026
    },
    {
      "epoch": 6.007893905904641,
      "grad_norm": 3.7320656776428223,
      "learning_rate": 3.886696137248711e-05,
      "loss": 0.5456,
      "step": 19027
    },
    {
      "epoch": 6.008209662140827,
      "grad_norm": 3.680537223815918,
      "learning_rate": 3.886608427183104e-05,
      "loss": 0.562,
      "step": 19028
    },
    {
      "epoch": 6.008525418377013,
      "grad_norm": 4.195338726043701,
      "learning_rate": 3.8865207171174965e-05,
      "loss": 0.509,
      "step": 19029
    },
    {
      "epoch": 6.008841174613199,
      "grad_norm": 4.55434513092041,
      "learning_rate": 3.8864330070518897e-05,
      "loss": 0.5071,
      "step": 19030
    },
    {
      "epoch": 6.009156930849384,
      "grad_norm": 3.800097703933716,
      "learning_rate": 3.886345296986282e-05,
      "loss": 0.4993,
      "step": 19031
    },
    {
      "epoch": 6.00947268708557,
      "grad_norm": 4.442776203155518,
      "learning_rate": 3.886257586920675e-05,
      "loss": 0.4641,
      "step": 19032
    },
    {
      "epoch": 6.009788443321756,
      "grad_norm": 4.623039245605469,
      "learning_rate": 3.886169876855068e-05,
      "loss": 0.585,
      "step": 19033
    },
    {
      "epoch": 6.010104199557941,
      "grad_norm": 3.3867719173431396,
      "learning_rate": 3.886082166789461e-05,
      "loss": 0.392,
      "step": 19034
    },
    {
      "epoch": 6.010419955794127,
      "grad_norm": 4.510968208312988,
      "learning_rate": 3.8859944567238535e-05,
      "loss": 0.5285,
      "step": 19035
    },
    {
      "epoch": 6.010735712030312,
      "grad_norm": 3.6041882038116455,
      "learning_rate": 3.8859067466582466e-05,
      "loss": 0.4847,
      "step": 19036
    },
    {
      "epoch": 6.011051468266499,
      "grad_norm": 4.909026622772217,
      "learning_rate": 3.88581903659264e-05,
      "loss": 0.4508,
      "step": 19037
    },
    {
      "epoch": 6.011367224502684,
      "grad_norm": 4.679943084716797,
      "learning_rate": 3.885731326527032e-05,
      "loss": 0.6058,
      "step": 19038
    },
    {
      "epoch": 6.011682980738869,
      "grad_norm": 4.886078357696533,
      "learning_rate": 3.8856436164614255e-05,
      "loss": 0.5491,
      "step": 19039
    },
    {
      "epoch": 6.011998736975055,
      "grad_norm": 5.332467555999756,
      "learning_rate": 3.885555906395818e-05,
      "loss": 0.6942,
      "step": 19040
    },
    {
      "epoch": 6.012314493211241,
      "grad_norm": 3.739743232727051,
      "learning_rate": 3.885468196330211e-05,
      "loss": 0.5652,
      "step": 19041
    },
    {
      "epoch": 6.012630249447427,
      "grad_norm": 4.24221134185791,
      "learning_rate": 3.8853804862646036e-05,
      "loss": 0.5761,
      "step": 19042
    },
    {
      "epoch": 6.012946005683612,
      "grad_norm": 4.066910266876221,
      "learning_rate": 3.885292776198997e-05,
      "loss": 0.5593,
      "step": 19043
    },
    {
      "epoch": 6.013261761919798,
      "grad_norm": 3.6443276405334473,
      "learning_rate": 3.885205066133389e-05,
      "loss": 0.5819,
      "step": 19044
    },
    {
      "epoch": 6.013577518155984,
      "grad_norm": 3.903029441833496,
      "learning_rate": 3.8851173560677824e-05,
      "loss": 0.575,
      "step": 19045
    },
    {
      "epoch": 6.013893274392169,
      "grad_norm": 4.38923454284668,
      "learning_rate": 3.8850296460021756e-05,
      "loss": 0.5439,
      "step": 19046
    },
    {
      "epoch": 6.014209030628355,
      "grad_norm": 4.4100422859191895,
      "learning_rate": 3.884941935936569e-05,
      "loss": 0.5838,
      "step": 19047
    },
    {
      "epoch": 6.01452478686454,
      "grad_norm": 4.691032886505127,
      "learning_rate": 3.884854225870961e-05,
      "loss": 0.5553,
      "step": 19048
    },
    {
      "epoch": 6.014840543100727,
      "grad_norm": 3.8978993892669678,
      "learning_rate": 3.8847665158053544e-05,
      "loss": 0.5237,
      "step": 19049
    },
    {
      "epoch": 6.015156299336912,
      "grad_norm": 4.210725784301758,
      "learning_rate": 3.884678805739747e-05,
      "loss": 0.5481,
      "step": 19050
    },
    {
      "epoch": 6.015472055573097,
      "grad_norm": 4.074349880218506,
      "learning_rate": 3.88459109567414e-05,
      "loss": 0.4606,
      "step": 19051
    },
    {
      "epoch": 6.015787811809283,
      "grad_norm": 4.562344074249268,
      "learning_rate": 3.8845033856085326e-05,
      "loss": 0.494,
      "step": 19052
    },
    {
      "epoch": 6.016103568045469,
      "grad_norm": 4.845857620239258,
      "learning_rate": 3.884415675542926e-05,
      "loss": 0.5131,
      "step": 19053
    },
    {
      "epoch": 6.016419324281655,
      "grad_norm": 4.598299980163574,
      "learning_rate": 3.884327965477318e-05,
      "loss": 0.5654,
      "step": 19054
    },
    {
      "epoch": 6.01673508051784,
      "grad_norm": 4.250957012176514,
      "learning_rate": 3.8842402554117114e-05,
      "loss": 0.602,
      "step": 19055
    },
    {
      "epoch": 6.017050836754026,
      "grad_norm": 4.372329235076904,
      "learning_rate": 3.884152545346104e-05,
      "loss": 0.4591,
      "step": 19056
    },
    {
      "epoch": 6.017366592990212,
      "grad_norm": 4.933135986328125,
      "learning_rate": 3.884064835280497e-05,
      "loss": 0.6304,
      "step": 19057
    },
    {
      "epoch": 6.017682349226397,
      "grad_norm": 3.8103713989257812,
      "learning_rate": 3.8839771252148896e-05,
      "loss": 0.4735,
      "step": 19058
    },
    {
      "epoch": 6.017998105462583,
      "grad_norm": 4.198678970336914,
      "learning_rate": 3.883889415149283e-05,
      "loss": 0.4463,
      "step": 19059
    },
    {
      "epoch": 6.018313861698768,
      "grad_norm": 3.959632635116577,
      "learning_rate": 3.883801705083676e-05,
      "loss": 0.6,
      "step": 19060
    },
    {
      "epoch": 6.0186296179349545,
      "grad_norm": 5.253981590270996,
      "learning_rate": 3.8837139950180684e-05,
      "loss": 0.5761,
      "step": 19061
    },
    {
      "epoch": 6.01894537417114,
      "grad_norm": 3.055494546890259,
      "learning_rate": 3.8836262849524616e-05,
      "loss": 0.4109,
      "step": 19062
    },
    {
      "epoch": 6.019261130407325,
      "grad_norm": 4.751966953277588,
      "learning_rate": 3.883538574886854e-05,
      "loss": 0.4727,
      "step": 19063
    },
    {
      "epoch": 6.019576886643511,
      "grad_norm": 3.7804267406463623,
      "learning_rate": 3.883450864821247e-05,
      "loss": 0.4305,
      "step": 19064
    },
    {
      "epoch": 6.019892642879697,
      "grad_norm": 4.213563442230225,
      "learning_rate": 3.88336315475564e-05,
      "loss": 0.5103,
      "step": 19065
    },
    {
      "epoch": 6.020208399115883,
      "grad_norm": 5.444753170013428,
      "learning_rate": 3.883275444690033e-05,
      "loss": 0.5533,
      "step": 19066
    },
    {
      "epoch": 6.020524155352068,
      "grad_norm": 4.1413421630859375,
      "learning_rate": 3.8831877346244254e-05,
      "loss": 0.5331,
      "step": 19067
    },
    {
      "epoch": 6.020839911588254,
      "grad_norm": 4.302812099456787,
      "learning_rate": 3.8831000245588185e-05,
      "loss": 0.5967,
      "step": 19068
    },
    {
      "epoch": 6.02115566782444,
      "grad_norm": 4.110512733459473,
      "learning_rate": 3.883012314493211e-05,
      "loss": 0.5211,
      "step": 19069
    },
    {
      "epoch": 6.021471424060625,
      "grad_norm": 4.964660167694092,
      "learning_rate": 3.882924604427604e-05,
      "loss": 0.5795,
      "step": 19070
    },
    {
      "epoch": 6.021787180296811,
      "grad_norm": 4.288976192474365,
      "learning_rate": 3.882836894361997e-05,
      "loss": 0.5178,
      "step": 19071
    },
    {
      "epoch": 6.022102936532996,
      "grad_norm": 4.818565845489502,
      "learning_rate": 3.88274918429639e-05,
      "loss": 0.4286,
      "step": 19072
    },
    {
      "epoch": 6.0224186927691825,
      "grad_norm": 5.072762489318848,
      "learning_rate": 3.882661474230782e-05,
      "loss": 0.6228,
      "step": 19073
    },
    {
      "epoch": 6.022734449005368,
      "grad_norm": 4.39359188079834,
      "learning_rate": 3.882573764165176e-05,
      "loss": 0.5275,
      "step": 19074
    },
    {
      "epoch": 6.023050205241553,
      "grad_norm": 5.093830585479736,
      "learning_rate": 3.882486054099569e-05,
      "loss": 0.5586,
      "step": 19075
    },
    {
      "epoch": 6.023365961477739,
      "grad_norm": 6.463253974914551,
      "learning_rate": 3.882398344033962e-05,
      "loss": 0.5792,
      "step": 19076
    },
    {
      "epoch": 6.023681717713925,
      "grad_norm": 4.130011081695557,
      "learning_rate": 3.882310633968354e-05,
      "loss": 0.465,
      "step": 19077
    },
    {
      "epoch": 6.023997473950111,
      "grad_norm": 4.350928783416748,
      "learning_rate": 3.8822229239027475e-05,
      "loss": 0.5066,
      "step": 19078
    },
    {
      "epoch": 6.024313230186296,
      "grad_norm": 4.293388366699219,
      "learning_rate": 3.88213521383714e-05,
      "loss": 0.5129,
      "step": 19079
    },
    {
      "epoch": 6.024628986422482,
      "grad_norm": 4.9557366371154785,
      "learning_rate": 3.882047503771533e-05,
      "loss": 0.4491,
      "step": 19080
    },
    {
      "epoch": 6.0249447426586675,
      "grad_norm": 4.171571254730225,
      "learning_rate": 3.881959793705926e-05,
      "loss": 0.4661,
      "step": 19081
    },
    {
      "epoch": 6.025260498894853,
      "grad_norm": 3.5421197414398193,
      "learning_rate": 3.881872083640319e-05,
      "loss": 0.4882,
      "step": 19082
    },
    {
      "epoch": 6.025576255131039,
      "grad_norm": 4.267844200134277,
      "learning_rate": 3.881784373574712e-05,
      "loss": 0.4823,
      "step": 19083
    },
    {
      "epoch": 6.025892011367224,
      "grad_norm": 4.443912982940674,
      "learning_rate": 3.8816966635091045e-05,
      "loss": 0.5512,
      "step": 19084
    },
    {
      "epoch": 6.0262077676034105,
      "grad_norm": 4.23737907409668,
      "learning_rate": 3.8816089534434976e-05,
      "loss": 0.4683,
      "step": 19085
    },
    {
      "epoch": 6.026523523839596,
      "grad_norm": 4.7148237228393555,
      "learning_rate": 3.88152124337789e-05,
      "loss": 0.5202,
      "step": 19086
    },
    {
      "epoch": 6.026839280075781,
      "grad_norm": 4.972785472869873,
      "learning_rate": 3.881433533312283e-05,
      "loss": 0.5542,
      "step": 19087
    },
    {
      "epoch": 6.027155036311967,
      "grad_norm": 4.4483184814453125,
      "learning_rate": 3.881345823246676e-05,
      "loss": 0.5354,
      "step": 19088
    },
    {
      "epoch": 6.0274707925481525,
      "grad_norm": 4.175746917724609,
      "learning_rate": 3.881258113181069e-05,
      "loss": 0.6404,
      "step": 19089
    },
    {
      "epoch": 6.027786548784339,
      "grad_norm": 4.34991455078125,
      "learning_rate": 3.8811704031154615e-05,
      "loss": 0.4926,
      "step": 19090
    },
    {
      "epoch": 6.028102305020524,
      "grad_norm": 4.778984069824219,
      "learning_rate": 3.8810826930498546e-05,
      "loss": 0.5689,
      "step": 19091
    },
    {
      "epoch": 6.02841806125671,
      "grad_norm": 4.186468124389648,
      "learning_rate": 3.880994982984247e-05,
      "loss": 0.4719,
      "step": 19092
    },
    {
      "epoch": 6.0287338174928955,
      "grad_norm": 5.386517524719238,
      "learning_rate": 3.88090727291864e-05,
      "loss": 0.4741,
      "step": 19093
    },
    {
      "epoch": 6.029049573729081,
      "grad_norm": 4.76111364364624,
      "learning_rate": 3.880819562853033e-05,
      "loss": 0.4356,
      "step": 19094
    },
    {
      "epoch": 6.029365329965267,
      "grad_norm": 4.174331188201904,
      "learning_rate": 3.880731852787426e-05,
      "loss": 0.6906,
      "step": 19095
    },
    {
      "epoch": 6.029681086201452,
      "grad_norm": 6.266899108886719,
      "learning_rate": 3.8806441427218184e-05,
      "loss": 0.5375,
      "step": 19096
    },
    {
      "epoch": 6.0299968424376384,
      "grad_norm": 4.959430694580078,
      "learning_rate": 3.8805564326562116e-05,
      "loss": 0.4798,
      "step": 19097
    },
    {
      "epoch": 6.030312598673824,
      "grad_norm": 4.826866149902344,
      "learning_rate": 3.880468722590605e-05,
      "loss": 0.5347,
      "step": 19098
    },
    {
      "epoch": 6.030628354910009,
      "grad_norm": 4.409898281097412,
      "learning_rate": 3.880381012524997e-05,
      "loss": 0.5571,
      "step": 19099
    },
    {
      "epoch": 6.030944111146195,
      "grad_norm": 4.115806579589844,
      "learning_rate": 3.8802933024593904e-05,
      "loss": 0.5299,
      "step": 19100
    },
    {
      "epoch": 6.0312598673823805,
      "grad_norm": 5.586905002593994,
      "learning_rate": 3.8802055923937836e-05,
      "loss": 0.5877,
      "step": 19101
    },
    {
      "epoch": 6.031575623618567,
      "grad_norm": 3.7959303855895996,
      "learning_rate": 3.880117882328176e-05,
      "loss": 0.5496,
      "step": 19102
    },
    {
      "epoch": 6.031891379854752,
      "grad_norm": 5.103792667388916,
      "learning_rate": 3.880030172262569e-05,
      "loss": 0.5543,
      "step": 19103
    },
    {
      "epoch": 6.032207136090938,
      "grad_norm": 4.8230366706848145,
      "learning_rate": 3.8799424621969624e-05,
      "loss": 0.5159,
      "step": 19104
    },
    {
      "epoch": 6.0325228923271235,
      "grad_norm": 5.5385589599609375,
      "learning_rate": 3.879854752131355e-05,
      "loss": 0.6164,
      "step": 19105
    },
    {
      "epoch": 6.032838648563309,
      "grad_norm": 4.994232654571533,
      "learning_rate": 3.879767042065748e-05,
      "loss": 0.5098,
      "step": 19106
    },
    {
      "epoch": 6.033154404799495,
      "grad_norm": 4.430750846862793,
      "learning_rate": 3.8796793320001406e-05,
      "loss": 0.5166,
      "step": 19107
    },
    {
      "epoch": 6.03347016103568,
      "grad_norm": 4.263552188873291,
      "learning_rate": 3.879591621934534e-05,
      "loss": 0.5464,
      "step": 19108
    },
    {
      "epoch": 6.033785917271866,
      "grad_norm": 4.666874885559082,
      "learning_rate": 3.879503911868926e-05,
      "loss": 0.5378,
      "step": 19109
    },
    {
      "epoch": 6.034101673508052,
      "grad_norm": 5.036930084228516,
      "learning_rate": 3.8794162018033194e-05,
      "loss": 0.6353,
      "step": 19110
    },
    {
      "epoch": 6.034417429744238,
      "grad_norm": 3.8861799240112305,
      "learning_rate": 3.879328491737712e-05,
      "loss": 0.5953,
      "step": 19111
    },
    {
      "epoch": 6.034733185980423,
      "grad_norm": 4.548035621643066,
      "learning_rate": 3.879240781672105e-05,
      "loss": 0.513,
      "step": 19112
    },
    {
      "epoch": 6.0350489422166085,
      "grad_norm": 5.292771816253662,
      "learning_rate": 3.8791530716064975e-05,
      "loss": 0.4318,
      "step": 19113
    },
    {
      "epoch": 6.035364698452795,
      "grad_norm": 4.764347553253174,
      "learning_rate": 3.879065361540891e-05,
      "loss": 0.5478,
      "step": 19114
    },
    {
      "epoch": 6.03568045468898,
      "grad_norm": 3.628606081008911,
      "learning_rate": 3.878977651475283e-05,
      "loss": 0.5043,
      "step": 19115
    },
    {
      "epoch": 6.035996210925166,
      "grad_norm": 4.482261657714844,
      "learning_rate": 3.8788899414096764e-05,
      "loss": 0.6053,
      "step": 19116
    },
    {
      "epoch": 6.036311967161351,
      "grad_norm": 4.359251022338867,
      "learning_rate": 3.878802231344069e-05,
      "loss": 0.5232,
      "step": 19117
    },
    {
      "epoch": 6.036627723397537,
      "grad_norm": 5.178619861602783,
      "learning_rate": 3.878714521278462e-05,
      "loss": 0.5664,
      "step": 19118
    },
    {
      "epoch": 6.036943479633723,
      "grad_norm": 4.548579216003418,
      "learning_rate": 3.878626811212855e-05,
      "loss": 0.5819,
      "step": 19119
    },
    {
      "epoch": 6.037259235869908,
      "grad_norm": 5.684675693511963,
      "learning_rate": 3.878539101147248e-05,
      "loss": 0.5607,
      "step": 19120
    },
    {
      "epoch": 6.037574992106094,
      "grad_norm": 4.3295111656188965,
      "learning_rate": 3.878451391081641e-05,
      "loss": 0.5156,
      "step": 19121
    },
    {
      "epoch": 6.03789074834228,
      "grad_norm": 4.702877044677734,
      "learning_rate": 3.8783636810160333e-05,
      "loss": 0.6043,
      "step": 19122
    },
    {
      "epoch": 6.038206504578466,
      "grad_norm": 3.9662017822265625,
      "learning_rate": 3.8782759709504265e-05,
      "loss": 0.496,
      "step": 19123
    },
    {
      "epoch": 6.038522260814651,
      "grad_norm": 4.158941745758057,
      "learning_rate": 3.878188260884819e-05,
      "loss": 0.4933,
      "step": 19124
    },
    {
      "epoch": 6.0388380170508364,
      "grad_norm": 4.374612808227539,
      "learning_rate": 3.878100550819212e-05,
      "loss": 0.5181,
      "step": 19125
    },
    {
      "epoch": 6.039153773287023,
      "grad_norm": 5.0902628898620605,
      "learning_rate": 3.878012840753605e-05,
      "loss": 0.5643,
      "step": 19126
    },
    {
      "epoch": 6.039469529523208,
      "grad_norm": 4.635817527770996,
      "learning_rate": 3.877925130687998e-05,
      "loss": 0.4856,
      "step": 19127
    },
    {
      "epoch": 6.039785285759394,
      "grad_norm": 5.7551188468933105,
      "learning_rate": 3.877837420622391e-05,
      "loss": 0.6887,
      "step": 19128
    },
    {
      "epoch": 6.040101041995579,
      "grad_norm": 4.71423864364624,
      "learning_rate": 3.877749710556784e-05,
      "loss": 0.5024,
      "step": 19129
    },
    {
      "epoch": 6.040416798231765,
      "grad_norm": 4.373601913452148,
      "learning_rate": 3.8776620004911767e-05,
      "loss": 0.4982,
      "step": 19130
    },
    {
      "epoch": 6.040732554467951,
      "grad_norm": 5.296268463134766,
      "learning_rate": 3.87757429042557e-05,
      "loss": 0.4918,
      "step": 19131
    },
    {
      "epoch": 6.041048310704136,
      "grad_norm": 3.7009003162384033,
      "learning_rate": 3.877486580359962e-05,
      "loss": 0.4913,
      "step": 19132
    },
    {
      "epoch": 6.041364066940322,
      "grad_norm": 5.3171305656433105,
      "learning_rate": 3.8773988702943555e-05,
      "loss": 0.5202,
      "step": 19133
    },
    {
      "epoch": 6.041679823176508,
      "grad_norm": 3.786489486694336,
      "learning_rate": 3.877311160228748e-05,
      "loss": 0.4867,
      "step": 19134
    },
    {
      "epoch": 6.041995579412694,
      "grad_norm": 4.69616174697876,
      "learning_rate": 3.877223450163141e-05,
      "loss": 0.4263,
      "step": 19135
    },
    {
      "epoch": 6.042311335648879,
      "grad_norm": 4.257242202758789,
      "learning_rate": 3.8771357400975336e-05,
      "loss": 0.5499,
      "step": 19136
    },
    {
      "epoch": 6.042627091885064,
      "grad_norm": 4.319581508636475,
      "learning_rate": 3.877048030031927e-05,
      "loss": 0.5072,
      "step": 19137
    },
    {
      "epoch": 6.042942848121251,
      "grad_norm": 8.800222396850586,
      "learning_rate": 3.876960319966319e-05,
      "loss": 0.4662,
      "step": 19138
    },
    {
      "epoch": 6.043258604357436,
      "grad_norm": 4.8656511306762695,
      "learning_rate": 3.8768726099007125e-05,
      "loss": 0.5753,
      "step": 19139
    },
    {
      "epoch": 6.043574360593622,
      "grad_norm": 5.495271682739258,
      "learning_rate": 3.876784899835105e-05,
      "loss": 0.5774,
      "step": 19140
    },
    {
      "epoch": 6.043890116829807,
      "grad_norm": 5.673631191253662,
      "learning_rate": 3.876697189769498e-05,
      "loss": 0.6007,
      "step": 19141
    },
    {
      "epoch": 6.044205873065993,
      "grad_norm": 4.513270378112793,
      "learning_rate": 3.876609479703891e-05,
      "loss": 0.4756,
      "step": 19142
    },
    {
      "epoch": 6.044521629302179,
      "grad_norm": 4.368355751037598,
      "learning_rate": 3.876521769638284e-05,
      "loss": 0.4193,
      "step": 19143
    },
    {
      "epoch": 6.044837385538364,
      "grad_norm": 4.440155982971191,
      "learning_rate": 3.876434059572677e-05,
      "loss": 0.4716,
      "step": 19144
    },
    {
      "epoch": 6.04515314177455,
      "grad_norm": 4.936107635498047,
      "learning_rate": 3.8763463495070694e-05,
      "loss": 0.5953,
      "step": 19145
    },
    {
      "epoch": 6.045468898010736,
      "grad_norm": 4.365263938903809,
      "learning_rate": 3.8762586394414626e-05,
      "loss": 0.6305,
      "step": 19146
    },
    {
      "epoch": 6.045784654246922,
      "grad_norm": 5.625818252563477,
      "learning_rate": 3.876170929375855e-05,
      "loss": 0.5502,
      "step": 19147
    },
    {
      "epoch": 6.046100410483107,
      "grad_norm": 4.793545722961426,
      "learning_rate": 3.876083219310248e-05,
      "loss": 0.4831,
      "step": 19148
    },
    {
      "epoch": 6.046416166719292,
      "grad_norm": 4.107318878173828,
      "learning_rate": 3.875995509244641e-05,
      "loss": 0.4471,
      "step": 19149
    },
    {
      "epoch": 6.046731922955479,
      "grad_norm": 5.610276699066162,
      "learning_rate": 3.875907799179034e-05,
      "loss": 0.5656,
      "step": 19150
    },
    {
      "epoch": 6.047047679191664,
      "grad_norm": 4.051235198974609,
      "learning_rate": 3.8758200891134264e-05,
      "loss": 0.5566,
      "step": 19151
    },
    {
      "epoch": 6.04736343542785,
      "grad_norm": 5.011022090911865,
      "learning_rate": 3.8757323790478196e-05,
      "loss": 0.4447,
      "step": 19152
    },
    {
      "epoch": 6.047679191664035,
      "grad_norm": 5.172586441040039,
      "learning_rate": 3.875644668982212e-05,
      "loss": 0.5948,
      "step": 19153
    },
    {
      "epoch": 6.047994947900221,
      "grad_norm": 3.7992191314697266,
      "learning_rate": 3.875556958916605e-05,
      "loss": 0.4334,
      "step": 19154
    },
    {
      "epoch": 6.048310704136407,
      "grad_norm": 4.672800540924072,
      "learning_rate": 3.8754692488509984e-05,
      "loss": 0.5085,
      "step": 19155
    },
    {
      "epoch": 6.048626460372592,
      "grad_norm": 5.357012748718262,
      "learning_rate": 3.8753815387853916e-05,
      "loss": 0.5936,
      "step": 19156
    },
    {
      "epoch": 6.048942216608778,
      "grad_norm": 5.424846649169922,
      "learning_rate": 3.875293828719784e-05,
      "loss": 0.636,
      "step": 19157
    },
    {
      "epoch": 6.049257972844964,
      "grad_norm": 4.066673755645752,
      "learning_rate": 3.875206118654177e-05,
      "loss": 0.5537,
      "step": 19158
    },
    {
      "epoch": 6.04957372908115,
      "grad_norm": 6.260152816772461,
      "learning_rate": 3.87511840858857e-05,
      "loss": 0.5821,
      "step": 19159
    },
    {
      "epoch": 6.049889485317335,
      "grad_norm": 5.729990482330322,
      "learning_rate": 3.875030698522963e-05,
      "loss": 0.5957,
      "step": 19160
    },
    {
      "epoch": 6.05020524155352,
      "grad_norm": 5.209537982940674,
      "learning_rate": 3.8749429884573554e-05,
      "loss": 0.5447,
      "step": 19161
    },
    {
      "epoch": 6.0505209977897065,
      "grad_norm": 4.672971248626709,
      "learning_rate": 3.8748552783917486e-05,
      "loss": 0.5195,
      "step": 19162
    },
    {
      "epoch": 6.050836754025892,
      "grad_norm": 4.688908100128174,
      "learning_rate": 3.874767568326142e-05,
      "loss": 0.4601,
      "step": 19163
    },
    {
      "epoch": 6.051152510262078,
      "grad_norm": 4.465378761291504,
      "learning_rate": 3.874679858260534e-05,
      "loss": 0.5662,
      "step": 19164
    },
    {
      "epoch": 6.051468266498263,
      "grad_norm": 4.879842281341553,
      "learning_rate": 3.8745921481949274e-05,
      "loss": 0.5383,
      "step": 19165
    },
    {
      "epoch": 6.051784022734449,
      "grad_norm": 5.149183750152588,
      "learning_rate": 3.87450443812932e-05,
      "loss": 0.5148,
      "step": 19166
    },
    {
      "epoch": 6.052099778970635,
      "grad_norm": 4.798679828643799,
      "learning_rate": 3.874416728063713e-05,
      "loss": 0.5836,
      "step": 19167
    },
    {
      "epoch": 6.05241553520682,
      "grad_norm": 5.173056125640869,
      "learning_rate": 3.8743290179981055e-05,
      "loss": 0.5721,
      "step": 19168
    },
    {
      "epoch": 6.052731291443006,
      "grad_norm": 4.467877388000488,
      "learning_rate": 3.874241307932499e-05,
      "loss": 0.4291,
      "step": 19169
    },
    {
      "epoch": 6.0530470476791916,
      "grad_norm": 4.804873466491699,
      "learning_rate": 3.874153597866891e-05,
      "loss": 0.5774,
      "step": 19170
    },
    {
      "epoch": 6.053362803915378,
      "grad_norm": 4.695404529571533,
      "learning_rate": 3.8740658878012844e-05,
      "loss": 0.5119,
      "step": 19171
    },
    {
      "epoch": 6.053678560151563,
      "grad_norm": 4.6191534996032715,
      "learning_rate": 3.873978177735677e-05,
      "loss": 0.5338,
      "step": 19172
    },
    {
      "epoch": 6.053994316387748,
      "grad_norm": 4.212815284729004,
      "learning_rate": 3.87389046767007e-05,
      "loss": 0.5257,
      "step": 19173
    },
    {
      "epoch": 6.0543100726239345,
      "grad_norm": 4.930461883544922,
      "learning_rate": 3.8738027576044625e-05,
      "loss": 0.5678,
      "step": 19174
    },
    {
      "epoch": 6.05462582886012,
      "grad_norm": 4.748530864715576,
      "learning_rate": 3.873715047538856e-05,
      "loss": 0.5313,
      "step": 19175
    },
    {
      "epoch": 6.054941585096306,
      "grad_norm": 5.356431007385254,
      "learning_rate": 3.873627337473248e-05,
      "loss": 0.6144,
      "step": 19176
    },
    {
      "epoch": 6.055257341332491,
      "grad_norm": 5.692637920379639,
      "learning_rate": 3.873539627407641e-05,
      "loss": 0.4581,
      "step": 19177
    },
    {
      "epoch": 6.055573097568677,
      "grad_norm": 4.13459587097168,
      "learning_rate": 3.873451917342034e-05,
      "loss": 0.565,
      "step": 19178
    },
    {
      "epoch": 6.055888853804863,
      "grad_norm": 4.214603424072266,
      "learning_rate": 3.873364207276427e-05,
      "loss": 0.425,
      "step": 19179
    },
    {
      "epoch": 6.056204610041048,
      "grad_norm": 4.937498569488525,
      "learning_rate": 3.87327649721082e-05,
      "loss": 0.5296,
      "step": 19180
    },
    {
      "epoch": 6.056520366277234,
      "grad_norm": 5.838397979736328,
      "learning_rate": 3.8731887871452127e-05,
      "loss": 0.561,
      "step": 19181
    },
    {
      "epoch": 6.0568361225134195,
      "grad_norm": 4.259878158569336,
      "learning_rate": 3.873101077079606e-05,
      "loss": 0.4721,
      "step": 19182
    },
    {
      "epoch": 6.057151878749606,
      "grad_norm": 4.318313121795654,
      "learning_rate": 3.873013367013999e-05,
      "loss": 0.5478,
      "step": 19183
    },
    {
      "epoch": 6.057467634985791,
      "grad_norm": 5.029994487762451,
      "learning_rate": 3.872925656948392e-05,
      "loss": 0.6729,
      "step": 19184
    },
    {
      "epoch": 6.057783391221976,
      "grad_norm": 4.521271228790283,
      "learning_rate": 3.8728379468827847e-05,
      "loss": 0.5464,
      "step": 19185
    },
    {
      "epoch": 6.0580991474581625,
      "grad_norm": 4.9603447914123535,
      "learning_rate": 3.872750236817178e-05,
      "loss": 0.5373,
      "step": 19186
    },
    {
      "epoch": 6.058414903694348,
      "grad_norm": 4.040429592132568,
      "learning_rate": 3.87266252675157e-05,
      "loss": 0.5042,
      "step": 19187
    },
    {
      "epoch": 6.058730659930534,
      "grad_norm": 4.964794635772705,
      "learning_rate": 3.8725748166859635e-05,
      "loss": 0.5909,
      "step": 19188
    },
    {
      "epoch": 6.059046416166719,
      "grad_norm": 4.702632904052734,
      "learning_rate": 3.872487106620356e-05,
      "loss": 0.5267,
      "step": 19189
    },
    {
      "epoch": 6.0593621724029045,
      "grad_norm": 6.013421535491943,
      "learning_rate": 3.872399396554749e-05,
      "loss": 0.5785,
      "step": 19190
    },
    {
      "epoch": 6.059677928639091,
      "grad_norm": 5.038196563720703,
      "learning_rate": 3.8723116864891416e-05,
      "loss": 0.5628,
      "step": 19191
    },
    {
      "epoch": 6.059993684875276,
      "grad_norm": 4.777223587036133,
      "learning_rate": 3.872223976423535e-05,
      "loss": 0.533,
      "step": 19192
    },
    {
      "epoch": 6.060309441111462,
      "grad_norm": 4.177000522613525,
      "learning_rate": 3.872136266357927e-05,
      "loss": 0.5338,
      "step": 19193
    },
    {
      "epoch": 6.0606251973476475,
      "grad_norm": 4.035135746002197,
      "learning_rate": 3.8720485562923205e-05,
      "loss": 0.5081,
      "step": 19194
    },
    {
      "epoch": 6.060940953583834,
      "grad_norm": 4.211075782775879,
      "learning_rate": 3.871960846226713e-05,
      "loss": 0.5546,
      "step": 19195
    },
    {
      "epoch": 6.061256709820019,
      "grad_norm": 6.211649417877197,
      "learning_rate": 3.871873136161106e-05,
      "loss": 0.6675,
      "step": 19196
    },
    {
      "epoch": 6.061572466056204,
      "grad_norm": 4.526993274688721,
      "learning_rate": 3.8717854260954986e-05,
      "loss": 0.5444,
      "step": 19197
    },
    {
      "epoch": 6.06188822229239,
      "grad_norm": 4.8501152992248535,
      "learning_rate": 3.871697716029892e-05,
      "loss": 0.5916,
      "step": 19198
    },
    {
      "epoch": 6.062203978528576,
      "grad_norm": 5.1024651527404785,
      "learning_rate": 3.871610005964284e-05,
      "loss": 0.6696,
      "step": 19199
    },
    {
      "epoch": 6.062519734764762,
      "grad_norm": 4.753866672515869,
      "learning_rate": 3.8715222958986774e-05,
      "loss": 0.5316,
      "step": 19200
    },
    {
      "epoch": 6.062519734764762,
      "eval_loss": 0.8785858154296875,
      "eval_runtime": 4.6189,
      "eval_samples_per_second": 539.093,
      "eval_steps_per_second": 4.547,
      "step": 19200
    },
    {
      "epoch": 6.062835491000947,
      "grad_norm": 3.924225091934204,
      "learning_rate": 3.8714345858330706e-05,
      "loss": 0.5139,
      "step": 19201
    },
    {
      "epoch": 6.0631512472371325,
      "grad_norm": 4.039456367492676,
      "learning_rate": 3.871346875767463e-05,
      "loss": 0.5586,
      "step": 19202
    },
    {
      "epoch": 6.063467003473319,
      "grad_norm": 4.682171821594238,
      "learning_rate": 3.871259165701856e-05,
      "loss": 0.534,
      "step": 19203
    },
    {
      "epoch": 6.063782759709504,
      "grad_norm": 5.240368843078613,
      "learning_rate": 3.871171455636249e-05,
      "loss": 0.4287,
      "step": 19204
    },
    {
      "epoch": 6.06409851594569,
      "grad_norm": 3.361088275909424,
      "learning_rate": 3.871083745570642e-05,
      "loss": 0.4817,
      "step": 19205
    },
    {
      "epoch": 6.0644142721818755,
      "grad_norm": 4.103672027587891,
      "learning_rate": 3.8709960355050344e-05,
      "loss": 0.5512,
      "step": 19206
    },
    {
      "epoch": 6.064730028418062,
      "grad_norm": 3.835928201675415,
      "learning_rate": 3.8709083254394276e-05,
      "loss": 0.5525,
      "step": 19207
    },
    {
      "epoch": 6.065045784654247,
      "grad_norm": 5.204422473907471,
      "learning_rate": 3.87082061537382e-05,
      "loss": 0.6372,
      "step": 19208
    },
    {
      "epoch": 6.065361540890432,
      "grad_norm": 4.706571578979492,
      "learning_rate": 3.870732905308213e-05,
      "loss": 0.5339,
      "step": 19209
    },
    {
      "epoch": 6.065677297126618,
      "grad_norm": 4.0920729637146,
      "learning_rate": 3.8706451952426064e-05,
      "loss": 0.5584,
      "step": 19210
    },
    {
      "epoch": 6.065993053362804,
      "grad_norm": 3.842724561691284,
      "learning_rate": 3.8705574851769996e-05,
      "loss": 0.5469,
      "step": 19211
    },
    {
      "epoch": 6.06630880959899,
      "grad_norm": 4.111110687255859,
      "learning_rate": 3.870469775111392e-05,
      "loss": 0.5292,
      "step": 19212
    },
    {
      "epoch": 6.066624565835175,
      "grad_norm": 4.651238918304443,
      "learning_rate": 3.870382065045785e-05,
      "loss": 0.53,
      "step": 19213
    },
    {
      "epoch": 6.0669403220713605,
      "grad_norm": 4.412952423095703,
      "learning_rate": 3.870294354980178e-05,
      "loss": 0.5141,
      "step": 19214
    },
    {
      "epoch": 6.067256078307547,
      "grad_norm": 6.11461067199707,
      "learning_rate": 3.870206644914571e-05,
      "loss": 0.5597,
      "step": 19215
    },
    {
      "epoch": 6.067571834543732,
      "grad_norm": 4.335175514221191,
      "learning_rate": 3.8701189348489634e-05,
      "loss": 0.5543,
      "step": 19216
    },
    {
      "epoch": 6.067887590779918,
      "grad_norm": 4.773852825164795,
      "learning_rate": 3.8700312247833565e-05,
      "loss": 0.5958,
      "step": 19217
    },
    {
      "epoch": 6.068203347016103,
      "grad_norm": 4.757269859313965,
      "learning_rate": 3.869943514717749e-05,
      "loss": 0.5089,
      "step": 19218
    },
    {
      "epoch": 6.06851910325229,
      "grad_norm": 4.614140510559082,
      "learning_rate": 3.869855804652142e-05,
      "loss": 0.5596,
      "step": 19219
    },
    {
      "epoch": 6.068834859488475,
      "grad_norm": 4.39194917678833,
      "learning_rate": 3.869768094586535e-05,
      "loss": 0.614,
      "step": 19220
    },
    {
      "epoch": 6.06915061572466,
      "grad_norm": 3.6514320373535156,
      "learning_rate": 3.869680384520928e-05,
      "loss": 0.5215,
      "step": 19221
    },
    {
      "epoch": 6.069466371960846,
      "grad_norm": 5.663857936859131,
      "learning_rate": 3.869592674455321e-05,
      "loss": 0.4491,
      "step": 19222
    },
    {
      "epoch": 6.069782128197032,
      "grad_norm": 4.904416084289551,
      "learning_rate": 3.8695049643897135e-05,
      "loss": 0.6616,
      "step": 19223
    },
    {
      "epoch": 6.070097884433218,
      "grad_norm": 4.966194152832031,
      "learning_rate": 3.869417254324107e-05,
      "loss": 0.5278,
      "step": 19224
    },
    {
      "epoch": 6.070413640669403,
      "grad_norm": 4.797999382019043,
      "learning_rate": 3.869329544258499e-05,
      "loss": 0.5396,
      "step": 19225
    },
    {
      "epoch": 6.070729396905589,
      "grad_norm": 4.860675811767578,
      "learning_rate": 3.8692418341928924e-05,
      "loss": 0.5814,
      "step": 19226
    },
    {
      "epoch": 6.071045153141775,
      "grad_norm": 3.694331169128418,
      "learning_rate": 3.869154124127285e-05,
      "loss": 0.4909,
      "step": 19227
    },
    {
      "epoch": 6.07136090937796,
      "grad_norm": 4.471615314483643,
      "learning_rate": 3.869066414061678e-05,
      "loss": 0.5238,
      "step": 19228
    },
    {
      "epoch": 6.071676665614146,
      "grad_norm": 5.117315769195557,
      "learning_rate": 3.8689787039960705e-05,
      "loss": 0.5892,
      "step": 19229
    },
    {
      "epoch": 6.071992421850331,
      "grad_norm": 4.924007892608643,
      "learning_rate": 3.868890993930464e-05,
      "loss": 0.4932,
      "step": 19230
    },
    {
      "epoch": 6.072308178086518,
      "grad_norm": 4.768149375915527,
      "learning_rate": 3.868803283864856e-05,
      "loss": 0.523,
      "step": 19231
    },
    {
      "epoch": 6.072623934322703,
      "grad_norm": 5.849143028259277,
      "learning_rate": 3.868715573799249e-05,
      "loss": 0.6634,
      "step": 19232
    },
    {
      "epoch": 6.072939690558888,
      "grad_norm": 4.190755844116211,
      "learning_rate": 3.868627863733642e-05,
      "loss": 0.5367,
      "step": 19233
    },
    {
      "epoch": 6.073255446795074,
      "grad_norm": 4.541486740112305,
      "learning_rate": 3.868540153668035e-05,
      "loss": 0.5655,
      "step": 19234
    },
    {
      "epoch": 6.07357120303126,
      "grad_norm": 4.2751922607421875,
      "learning_rate": 3.8684524436024275e-05,
      "loss": 0.5079,
      "step": 19235
    },
    {
      "epoch": 6.073886959267446,
      "grad_norm": 5.049468994140625,
      "learning_rate": 3.8683647335368206e-05,
      "loss": 0.542,
      "step": 19236
    },
    {
      "epoch": 6.074202715503631,
      "grad_norm": 4.669381618499756,
      "learning_rate": 3.868277023471214e-05,
      "loss": 0.5816,
      "step": 19237
    },
    {
      "epoch": 6.074518471739817,
      "grad_norm": 4.930968761444092,
      "learning_rate": 3.868189313405607e-05,
      "loss": 0.5819,
      "step": 19238
    },
    {
      "epoch": 6.074834227976003,
      "grad_norm": 3.864854574203491,
      "learning_rate": 3.8681016033399995e-05,
      "loss": 0.6188,
      "step": 19239
    },
    {
      "epoch": 6.075149984212188,
      "grad_norm": 5.316928386688232,
      "learning_rate": 3.8680138932743926e-05,
      "loss": 0.5525,
      "step": 19240
    },
    {
      "epoch": 6.075465740448374,
      "grad_norm": 4.301572799682617,
      "learning_rate": 3.867926183208785e-05,
      "loss": 0.4432,
      "step": 19241
    },
    {
      "epoch": 6.075781496684559,
      "grad_norm": 6.232663631439209,
      "learning_rate": 3.867838473143178e-05,
      "loss": 0.5321,
      "step": 19242
    },
    {
      "epoch": 6.0760972529207455,
      "grad_norm": 4.377511978149414,
      "learning_rate": 3.867750763077571e-05,
      "loss": 0.54,
      "step": 19243
    },
    {
      "epoch": 6.076413009156931,
      "grad_norm": 5.360091209411621,
      "learning_rate": 3.867663053011964e-05,
      "loss": 0.6365,
      "step": 19244
    },
    {
      "epoch": 6.076728765393116,
      "grad_norm": 4.945279121398926,
      "learning_rate": 3.867575342946357e-05,
      "loss": 0.5907,
      "step": 19245
    },
    {
      "epoch": 6.077044521629302,
      "grad_norm": 4.29353666305542,
      "learning_rate": 3.8674876328807496e-05,
      "loss": 0.5013,
      "step": 19246
    },
    {
      "epoch": 6.077360277865488,
      "grad_norm": 4.235612869262695,
      "learning_rate": 3.867399922815143e-05,
      "loss": 0.561,
      "step": 19247
    },
    {
      "epoch": 6.077676034101674,
      "grad_norm": 7.912813663482666,
      "learning_rate": 3.867312212749535e-05,
      "loss": 0.5417,
      "step": 19248
    },
    {
      "epoch": 6.077991790337859,
      "grad_norm": 4.415191650390625,
      "learning_rate": 3.8672245026839284e-05,
      "loss": 0.4567,
      "step": 19249
    },
    {
      "epoch": 6.078307546574045,
      "grad_norm": 5.156689643859863,
      "learning_rate": 3.867136792618321e-05,
      "loss": 0.5358,
      "step": 19250
    },
    {
      "epoch": 6.0786233028102306,
      "grad_norm": 4.805480480194092,
      "learning_rate": 3.867049082552714e-05,
      "loss": 0.4411,
      "step": 19251
    },
    {
      "epoch": 6.078939059046416,
      "grad_norm": 5.428408145904541,
      "learning_rate": 3.8669613724871066e-05,
      "loss": 0.5671,
      "step": 19252
    },
    {
      "epoch": 6.079254815282602,
      "grad_norm": 5.359206676483154,
      "learning_rate": 3.8668736624215e-05,
      "loss": 0.5093,
      "step": 19253
    },
    {
      "epoch": 6.079570571518787,
      "grad_norm": 4.436506748199463,
      "learning_rate": 3.866785952355892e-05,
      "loss": 0.5699,
      "step": 19254
    },
    {
      "epoch": 6.0798863277549735,
      "grad_norm": 4.763530731201172,
      "learning_rate": 3.8666982422902854e-05,
      "loss": 0.5359,
      "step": 19255
    },
    {
      "epoch": 6.080202083991159,
      "grad_norm": 5.272853374481201,
      "learning_rate": 3.866610532224678e-05,
      "loss": 0.5574,
      "step": 19256
    },
    {
      "epoch": 6.080517840227344,
      "grad_norm": 4.9555864334106445,
      "learning_rate": 3.866522822159071e-05,
      "loss": 0.5218,
      "step": 19257
    },
    {
      "epoch": 6.08083359646353,
      "grad_norm": 5.869931697845459,
      "learning_rate": 3.8664351120934636e-05,
      "loss": 0.5873,
      "step": 19258
    },
    {
      "epoch": 6.081149352699716,
      "grad_norm": 7.709561824798584,
      "learning_rate": 3.866347402027857e-05,
      "loss": 0.5145,
      "step": 19259
    },
    {
      "epoch": 6.081465108935902,
      "grad_norm": 4.3937458992004395,
      "learning_rate": 3.86625969196225e-05,
      "loss": 0.4719,
      "step": 19260
    },
    {
      "epoch": 6.081780865172087,
      "grad_norm": 5.153443813323975,
      "learning_rate": 3.8661719818966424e-05,
      "loss": 0.4784,
      "step": 19261
    },
    {
      "epoch": 6.082096621408273,
      "grad_norm": 6.281277179718018,
      "learning_rate": 3.8660842718310356e-05,
      "loss": 0.4825,
      "step": 19262
    },
    {
      "epoch": 6.0824123776444585,
      "grad_norm": 3.8105156421661377,
      "learning_rate": 3.865996561765428e-05,
      "loss": 0.4631,
      "step": 19263
    },
    {
      "epoch": 6.082728133880644,
      "grad_norm": 4.102684020996094,
      "learning_rate": 3.865908851699821e-05,
      "loss": 0.5599,
      "step": 19264
    },
    {
      "epoch": 6.08304389011683,
      "grad_norm": 5.217466354370117,
      "learning_rate": 3.8658211416342144e-05,
      "loss": 0.4822,
      "step": 19265
    },
    {
      "epoch": 6.083359646353015,
      "grad_norm": 5.0025177001953125,
      "learning_rate": 3.8657334315686076e-05,
      "loss": 0.567,
      "step": 19266
    },
    {
      "epoch": 6.0836754025892015,
      "grad_norm": 4.769082546234131,
      "learning_rate": 3.865645721503e-05,
      "loss": 0.5331,
      "step": 19267
    },
    {
      "epoch": 6.083991158825387,
      "grad_norm": 3.630565881729126,
      "learning_rate": 3.865558011437393e-05,
      "loss": 0.4053,
      "step": 19268
    },
    {
      "epoch": 6.084306915061572,
      "grad_norm": 5.300273895263672,
      "learning_rate": 3.865470301371786e-05,
      "loss": 0.5219,
      "step": 19269
    },
    {
      "epoch": 6.084622671297758,
      "grad_norm": 3.792370557785034,
      "learning_rate": 3.865382591306179e-05,
      "loss": 0.4373,
      "step": 19270
    },
    {
      "epoch": 6.0849384275339435,
      "grad_norm": 5.393277168273926,
      "learning_rate": 3.8652948812405714e-05,
      "loss": 0.5589,
      "step": 19271
    },
    {
      "epoch": 6.08525418377013,
      "grad_norm": 4.253884315490723,
      "learning_rate": 3.8652071711749645e-05,
      "loss": 0.4986,
      "step": 19272
    },
    {
      "epoch": 6.085569940006315,
      "grad_norm": 4.438899040222168,
      "learning_rate": 3.865119461109357e-05,
      "loss": 0.6422,
      "step": 19273
    },
    {
      "epoch": 6.085885696242501,
      "grad_norm": 5.350765228271484,
      "learning_rate": 3.86503175104375e-05,
      "loss": 0.4833,
      "step": 19274
    },
    {
      "epoch": 6.0862014524786865,
      "grad_norm": 4.829704761505127,
      "learning_rate": 3.864944040978143e-05,
      "loss": 0.4956,
      "step": 19275
    },
    {
      "epoch": 6.086517208714872,
      "grad_norm": 4.90872049331665,
      "learning_rate": 3.864856330912536e-05,
      "loss": 0.5084,
      "step": 19276
    },
    {
      "epoch": 6.086832964951058,
      "grad_norm": 4.884903430938721,
      "learning_rate": 3.8647686208469283e-05,
      "loss": 0.5403,
      "step": 19277
    },
    {
      "epoch": 6.087148721187243,
      "grad_norm": 4.556715965270996,
      "learning_rate": 3.8646809107813215e-05,
      "loss": 0.5947,
      "step": 19278
    },
    {
      "epoch": 6.087464477423429,
      "grad_norm": 5.99237585067749,
      "learning_rate": 3.864593200715714e-05,
      "loss": 0.6243,
      "step": 19279
    },
    {
      "epoch": 6.087780233659615,
      "grad_norm": 10.393939971923828,
      "learning_rate": 3.864505490650107e-05,
      "loss": 0.542,
      "step": 19280
    },
    {
      "epoch": 6.0880959898958,
      "grad_norm": 4.901520729064941,
      "learning_rate": 3.8644177805845e-05,
      "loss": 0.5837,
      "step": 19281
    },
    {
      "epoch": 6.088411746131986,
      "grad_norm": 5.402565002441406,
      "learning_rate": 3.864330070518893e-05,
      "loss": 0.5179,
      "step": 19282
    },
    {
      "epoch": 6.0887275023681715,
      "grad_norm": 5.808990955352783,
      "learning_rate": 3.864242360453286e-05,
      "loss": 0.5494,
      "step": 19283
    },
    {
      "epoch": 6.089043258604358,
      "grad_norm": 5.350639820098877,
      "learning_rate": 3.8641546503876785e-05,
      "loss": 0.5507,
      "step": 19284
    },
    {
      "epoch": 6.089359014840543,
      "grad_norm": 5.9568376541137695,
      "learning_rate": 3.8640669403220717e-05,
      "loss": 0.5878,
      "step": 19285
    },
    {
      "epoch": 6.089674771076729,
      "grad_norm": 6.363183498382568,
      "learning_rate": 3.863979230256464e-05,
      "loss": 0.6104,
      "step": 19286
    },
    {
      "epoch": 6.0899905273129145,
      "grad_norm": 5.491600036621094,
      "learning_rate": 3.863891520190857e-05,
      "loss": 0.6089,
      "step": 19287
    },
    {
      "epoch": 6.0903062835491,
      "grad_norm": 4.814655303955078,
      "learning_rate": 3.86380381012525e-05,
      "loss": 0.6015,
      "step": 19288
    },
    {
      "epoch": 6.090622039785286,
      "grad_norm": 4.065524578094482,
      "learning_rate": 3.863716100059643e-05,
      "loss": 0.4544,
      "step": 19289
    },
    {
      "epoch": 6.090937796021471,
      "grad_norm": 5.704403877258301,
      "learning_rate": 3.8636283899940355e-05,
      "loss": 0.5678,
      "step": 19290
    },
    {
      "epoch": 6.091253552257657,
      "grad_norm": 5.410664081573486,
      "learning_rate": 3.8635406799284286e-05,
      "loss": 0.5487,
      "step": 19291
    },
    {
      "epoch": 6.091569308493843,
      "grad_norm": 4.967745304107666,
      "learning_rate": 3.863452969862822e-05,
      "loss": 0.6208,
      "step": 19292
    },
    {
      "epoch": 6.091885064730028,
      "grad_norm": 5.093880653381348,
      "learning_rate": 3.863365259797215e-05,
      "loss": 0.5953,
      "step": 19293
    },
    {
      "epoch": 6.092200820966214,
      "grad_norm": 4.868369102478027,
      "learning_rate": 3.8632775497316075e-05,
      "loss": 0.5322,
      "step": 19294
    },
    {
      "epoch": 6.0925165772023995,
      "grad_norm": 4.728320121765137,
      "learning_rate": 3.8631898396660006e-05,
      "loss": 0.6451,
      "step": 19295
    },
    {
      "epoch": 6.092832333438586,
      "grad_norm": 4.9820170402526855,
      "learning_rate": 3.863102129600393e-05,
      "loss": 0.5082,
      "step": 19296
    },
    {
      "epoch": 6.093148089674771,
      "grad_norm": 3.5476572513580322,
      "learning_rate": 3.863014419534786e-05,
      "loss": 0.5052,
      "step": 19297
    },
    {
      "epoch": 6.093463845910957,
      "grad_norm": 4.633820056915283,
      "learning_rate": 3.862926709469179e-05,
      "loss": 0.4724,
      "step": 19298
    },
    {
      "epoch": 6.093779602147142,
      "grad_norm": 4.5134735107421875,
      "learning_rate": 3.862838999403572e-05,
      "loss": 0.4784,
      "step": 19299
    },
    {
      "epoch": 6.094095358383328,
      "grad_norm": 4.83816385269165,
      "learning_rate": 3.8627512893379644e-05,
      "loss": 0.5622,
      "step": 19300
    },
    {
      "epoch": 6.094411114619514,
      "grad_norm": 4.774569511413574,
      "learning_rate": 3.8626635792723576e-05,
      "loss": 0.6102,
      "step": 19301
    },
    {
      "epoch": 6.094726870855699,
      "grad_norm": 5.547092437744141,
      "learning_rate": 3.86257586920675e-05,
      "loss": 0.5716,
      "step": 19302
    },
    {
      "epoch": 6.095042627091885,
      "grad_norm": 7.723728179931641,
      "learning_rate": 3.862488159141143e-05,
      "loss": 0.5969,
      "step": 19303
    },
    {
      "epoch": 6.095358383328071,
      "grad_norm": 5.475771903991699,
      "learning_rate": 3.8624004490755364e-05,
      "loss": 0.5738,
      "step": 19304
    },
    {
      "epoch": 6.095674139564256,
      "grad_norm": 5.0401458740234375,
      "learning_rate": 3.862312739009929e-05,
      "loss": 0.5611,
      "step": 19305
    },
    {
      "epoch": 6.095989895800442,
      "grad_norm": 4.424618244171143,
      "learning_rate": 3.862225028944322e-05,
      "loss": 0.5239,
      "step": 19306
    },
    {
      "epoch": 6.096305652036627,
      "grad_norm": 4.552433013916016,
      "learning_rate": 3.8621373188787146e-05,
      "loss": 0.5202,
      "step": 19307
    },
    {
      "epoch": 6.096621408272814,
      "grad_norm": 5.084836483001709,
      "learning_rate": 3.862049608813108e-05,
      "loss": 0.5129,
      "step": 19308
    },
    {
      "epoch": 6.096937164508999,
      "grad_norm": 5.164605617523193,
      "learning_rate": 3.8619618987475e-05,
      "loss": 0.5606,
      "step": 19309
    },
    {
      "epoch": 6.097252920745185,
      "grad_norm": 4.2934041023254395,
      "learning_rate": 3.8618741886818934e-05,
      "loss": 0.5774,
      "step": 19310
    },
    {
      "epoch": 6.09756867698137,
      "grad_norm": 5.661055088043213,
      "learning_rate": 3.861786478616286e-05,
      "loss": 0.5306,
      "step": 19311
    },
    {
      "epoch": 6.097884433217556,
      "grad_norm": 4.292966365814209,
      "learning_rate": 3.861698768550679e-05,
      "loss": 0.4923,
      "step": 19312
    },
    {
      "epoch": 6.098200189453742,
      "grad_norm": 3.9943459033966064,
      "learning_rate": 3.8616110584850716e-05,
      "loss": 0.5672,
      "step": 19313
    },
    {
      "epoch": 6.098515945689927,
      "grad_norm": 5.528664588928223,
      "learning_rate": 3.861523348419465e-05,
      "loss": 0.5271,
      "step": 19314
    },
    {
      "epoch": 6.098831701926113,
      "grad_norm": 3.7775657176971436,
      "learning_rate": 3.861435638353857e-05,
      "loss": 0.556,
      "step": 19315
    },
    {
      "epoch": 6.099147458162299,
      "grad_norm": 6.749685764312744,
      "learning_rate": 3.8613479282882504e-05,
      "loss": 0.5578,
      "step": 19316
    },
    {
      "epoch": 6.099463214398484,
      "grad_norm": 3.944777727127075,
      "learning_rate": 3.861260218222643e-05,
      "loss": 0.4547,
      "step": 19317
    },
    {
      "epoch": 6.09977897063467,
      "grad_norm": 4.51922607421875,
      "learning_rate": 3.861172508157036e-05,
      "loss": 0.5632,
      "step": 19318
    },
    {
      "epoch": 6.100094726870855,
      "grad_norm": 5.653581142425537,
      "learning_rate": 3.861084798091429e-05,
      "loss": 0.612,
      "step": 19319
    },
    {
      "epoch": 6.100410483107042,
      "grad_norm": 3.947051763534546,
      "learning_rate": 3.8609970880258224e-05,
      "loss": 0.5181,
      "step": 19320
    },
    {
      "epoch": 6.100726239343227,
      "grad_norm": 3.714498996734619,
      "learning_rate": 3.860909377960215e-05,
      "loss": 0.5644,
      "step": 19321
    },
    {
      "epoch": 6.101041995579413,
      "grad_norm": 5.6170878410339355,
      "learning_rate": 3.860821667894608e-05,
      "loss": 0.5074,
      "step": 19322
    },
    {
      "epoch": 6.101357751815598,
      "grad_norm": 4.8218841552734375,
      "learning_rate": 3.8607339578290005e-05,
      "loss": 0.5436,
      "step": 19323
    },
    {
      "epoch": 6.101673508051784,
      "grad_norm": 4.7587080001831055,
      "learning_rate": 3.860646247763394e-05,
      "loss": 0.4656,
      "step": 19324
    },
    {
      "epoch": 6.10198926428797,
      "grad_norm": 6.342790603637695,
      "learning_rate": 3.860558537697786e-05,
      "loss": 0.637,
      "step": 19325
    },
    {
      "epoch": 6.102305020524155,
      "grad_norm": 4.374898910522461,
      "learning_rate": 3.8604708276321794e-05,
      "loss": 0.5295,
      "step": 19326
    },
    {
      "epoch": 6.102620776760341,
      "grad_norm": 4.204529762268066,
      "learning_rate": 3.8603831175665725e-05,
      "loss": 0.5696,
      "step": 19327
    },
    {
      "epoch": 6.102936532996527,
      "grad_norm": 6.487770080566406,
      "learning_rate": 3.860295407500965e-05,
      "loss": 0.5132,
      "step": 19328
    },
    {
      "epoch": 6.103252289232712,
      "grad_norm": 4.250041961669922,
      "learning_rate": 3.860207697435358e-05,
      "loss": 0.4406,
      "step": 19329
    },
    {
      "epoch": 6.103568045468898,
      "grad_norm": 4.765947341918945,
      "learning_rate": 3.860119987369751e-05,
      "loss": 0.4957,
      "step": 19330
    },
    {
      "epoch": 6.103883801705083,
      "grad_norm": 4.236738204956055,
      "learning_rate": 3.860032277304144e-05,
      "loss": 0.5207,
      "step": 19331
    },
    {
      "epoch": 6.10419955794127,
      "grad_norm": 5.5435075759887695,
      "learning_rate": 3.859944567238536e-05,
      "loss": 0.6336,
      "step": 19332
    },
    {
      "epoch": 6.104515314177455,
      "grad_norm": 3.9775733947753906,
      "learning_rate": 3.8598568571729295e-05,
      "loss": 0.4889,
      "step": 19333
    },
    {
      "epoch": 6.104831070413641,
      "grad_norm": 4.51759672164917,
      "learning_rate": 3.859769147107322e-05,
      "loss": 0.5593,
      "step": 19334
    },
    {
      "epoch": 6.105146826649826,
      "grad_norm": 5.501622676849365,
      "learning_rate": 3.859681437041715e-05,
      "loss": 0.5544,
      "step": 19335
    },
    {
      "epoch": 6.105462582886012,
      "grad_norm": 5.603848457336426,
      "learning_rate": 3.8595937269761077e-05,
      "loss": 0.5865,
      "step": 19336
    },
    {
      "epoch": 6.105778339122198,
      "grad_norm": 3.644325017929077,
      "learning_rate": 3.859506016910501e-05,
      "loss": 0.4156,
      "step": 19337
    },
    {
      "epoch": 6.106094095358383,
      "grad_norm": 4.723912715911865,
      "learning_rate": 3.859418306844893e-05,
      "loss": 0.5116,
      "step": 19338
    },
    {
      "epoch": 6.106409851594569,
      "grad_norm": 4.008573055267334,
      "learning_rate": 3.8593305967792865e-05,
      "loss": 0.4962,
      "step": 19339
    },
    {
      "epoch": 6.106725607830755,
      "grad_norm": 4.33093786239624,
      "learning_rate": 3.859242886713679e-05,
      "loss": 0.6778,
      "step": 19340
    },
    {
      "epoch": 6.107041364066941,
      "grad_norm": 4.984989166259766,
      "learning_rate": 3.859155176648072e-05,
      "loss": 0.5789,
      "step": 19341
    },
    {
      "epoch": 6.107357120303126,
      "grad_norm": 5.54381799697876,
      "learning_rate": 3.859067466582465e-05,
      "loss": 0.5392,
      "step": 19342
    },
    {
      "epoch": 6.107672876539311,
      "grad_norm": 5.735755443572998,
      "learning_rate": 3.858979756516858e-05,
      "loss": 0.495,
      "step": 19343
    },
    {
      "epoch": 6.1079886327754975,
      "grad_norm": 4.0661301612854,
      "learning_rate": 3.858892046451251e-05,
      "loss": 0.5343,
      "step": 19344
    },
    {
      "epoch": 6.108304389011683,
      "grad_norm": 4.6931586265563965,
      "learning_rate": 3.8588043363856435e-05,
      "loss": 0.5907,
      "step": 19345
    },
    {
      "epoch": 6.108620145247869,
      "grad_norm": 4.353609085083008,
      "learning_rate": 3.8587166263200366e-05,
      "loss": 0.491,
      "step": 19346
    },
    {
      "epoch": 6.108935901484054,
      "grad_norm": 4.9989752769470215,
      "learning_rate": 3.85862891625443e-05,
      "loss": 0.6278,
      "step": 19347
    },
    {
      "epoch": 6.10925165772024,
      "grad_norm": 5.143468379974365,
      "learning_rate": 3.858541206188823e-05,
      "loss": 0.5426,
      "step": 19348
    },
    {
      "epoch": 6.109567413956426,
      "grad_norm": 4.357733249664307,
      "learning_rate": 3.8584534961232155e-05,
      "loss": 0.4356,
      "step": 19349
    },
    {
      "epoch": 6.109883170192611,
      "grad_norm": 4.9929304122924805,
      "learning_rate": 3.8583657860576086e-05,
      "loss": 0.6074,
      "step": 19350
    },
    {
      "epoch": 6.110198926428797,
      "grad_norm": 5.132781505584717,
      "learning_rate": 3.858278075992001e-05,
      "loss": 0.5736,
      "step": 19351
    },
    {
      "epoch": 6.1105146826649825,
      "grad_norm": 4.670170783996582,
      "learning_rate": 3.858190365926394e-05,
      "loss": 0.4755,
      "step": 19352
    },
    {
      "epoch": 6.110830438901169,
      "grad_norm": 3.8742473125457764,
      "learning_rate": 3.858102655860787e-05,
      "loss": 0.557,
      "step": 19353
    },
    {
      "epoch": 6.111146195137354,
      "grad_norm": 5.21669864654541,
      "learning_rate": 3.85801494579518e-05,
      "loss": 0.5734,
      "step": 19354
    },
    {
      "epoch": 6.111461951373539,
      "grad_norm": 6.212246894836426,
      "learning_rate": 3.8579272357295724e-05,
      "loss": 0.5968,
      "step": 19355
    },
    {
      "epoch": 6.1117777076097255,
      "grad_norm": 4.644890785217285,
      "learning_rate": 3.8578395256639656e-05,
      "loss": 0.4472,
      "step": 19356
    },
    {
      "epoch": 6.112093463845911,
      "grad_norm": 4.578973293304443,
      "learning_rate": 3.857751815598358e-05,
      "loss": 0.5469,
      "step": 19357
    },
    {
      "epoch": 6.112409220082097,
      "grad_norm": 5.017146110534668,
      "learning_rate": 3.857664105532751e-05,
      "loss": 0.5264,
      "step": 19358
    },
    {
      "epoch": 6.112724976318282,
      "grad_norm": 3.994487762451172,
      "learning_rate": 3.857576395467144e-05,
      "loss": 0.5145,
      "step": 19359
    },
    {
      "epoch": 6.113040732554468,
      "grad_norm": 5.205955505371094,
      "learning_rate": 3.857488685401537e-05,
      "loss": 0.5367,
      "step": 19360
    },
    {
      "epoch": 6.113356488790654,
      "grad_norm": 6.011228561401367,
      "learning_rate": 3.8574009753359294e-05,
      "loss": 0.4669,
      "step": 19361
    },
    {
      "epoch": 6.113672245026839,
      "grad_norm": 4.241981029510498,
      "learning_rate": 3.8573132652703226e-05,
      "loss": 0.4504,
      "step": 19362
    },
    {
      "epoch": 6.113988001263025,
      "grad_norm": 6.6169586181640625,
      "learning_rate": 3.857225555204716e-05,
      "loss": 0.6116,
      "step": 19363
    },
    {
      "epoch": 6.1143037574992105,
      "grad_norm": 4.871487617492676,
      "learning_rate": 3.857137845139108e-05,
      "loss": 0.5245,
      "step": 19364
    },
    {
      "epoch": 6.114619513735397,
      "grad_norm": 5.623633861541748,
      "learning_rate": 3.8570501350735014e-05,
      "loss": 0.6052,
      "step": 19365
    },
    {
      "epoch": 6.114935269971582,
      "grad_norm": 4.596688270568848,
      "learning_rate": 3.856962425007894e-05,
      "loss": 0.4558,
      "step": 19366
    },
    {
      "epoch": 6.115251026207767,
      "grad_norm": 4.590485572814941,
      "learning_rate": 3.856874714942287e-05,
      "loss": 0.5257,
      "step": 19367
    },
    {
      "epoch": 6.1155667824439535,
      "grad_norm": 4.90207576751709,
      "learning_rate": 3.8567870048766795e-05,
      "loss": 0.5211,
      "step": 19368
    },
    {
      "epoch": 6.115882538680139,
      "grad_norm": 4.658449649810791,
      "learning_rate": 3.856699294811073e-05,
      "loss": 0.5403,
      "step": 19369
    },
    {
      "epoch": 6.116198294916325,
      "grad_norm": 4.4433674812316895,
      "learning_rate": 3.856611584745465e-05,
      "loss": 0.5257,
      "step": 19370
    },
    {
      "epoch": 6.11651405115251,
      "grad_norm": 4.567420959472656,
      "learning_rate": 3.8565238746798584e-05,
      "loss": 0.4997,
      "step": 19371
    },
    {
      "epoch": 6.1168298073886955,
      "grad_norm": 4.182559490203857,
      "learning_rate": 3.856436164614251e-05,
      "loss": 0.5302,
      "step": 19372
    },
    {
      "epoch": 6.117145563624882,
      "grad_norm": 4.037980079650879,
      "learning_rate": 3.856348454548645e-05,
      "loss": 0.5378,
      "step": 19373
    },
    {
      "epoch": 6.117461319861067,
      "grad_norm": 4.6596221923828125,
      "learning_rate": 3.856260744483037e-05,
      "loss": 0.4708,
      "step": 19374
    },
    {
      "epoch": 6.117777076097253,
      "grad_norm": 5.117971420288086,
      "learning_rate": 3.8561730344174304e-05,
      "loss": 0.5291,
      "step": 19375
    },
    {
      "epoch": 6.1180928323334385,
      "grad_norm": 3.7098827362060547,
      "learning_rate": 3.856085324351823e-05,
      "loss": 0.4449,
      "step": 19376
    },
    {
      "epoch": 6.118408588569625,
      "grad_norm": 4.296205997467041,
      "learning_rate": 3.855997614286216e-05,
      "loss": 0.5452,
      "step": 19377
    },
    {
      "epoch": 6.11872434480581,
      "grad_norm": 3.457303047180176,
      "learning_rate": 3.8559099042206085e-05,
      "loss": 0.4704,
      "step": 19378
    },
    {
      "epoch": 6.119040101041995,
      "grad_norm": 5.753150939941406,
      "learning_rate": 3.855822194155002e-05,
      "loss": 0.4806,
      "step": 19379
    },
    {
      "epoch": 6.119355857278181,
      "grad_norm": 8.19874095916748,
      "learning_rate": 3.855734484089394e-05,
      "loss": 0.5961,
      "step": 19380
    },
    {
      "epoch": 6.119671613514367,
      "grad_norm": 5.032822132110596,
      "learning_rate": 3.8556467740237873e-05,
      "loss": 0.5376,
      "step": 19381
    },
    {
      "epoch": 6.119987369750553,
      "grad_norm": 5.792872905731201,
      "learning_rate": 3.85555906395818e-05,
      "loss": 0.5442,
      "step": 19382
    },
    {
      "epoch": 6.120303125986738,
      "grad_norm": 4.073696613311768,
      "learning_rate": 3.855471353892573e-05,
      "loss": 0.484,
      "step": 19383
    },
    {
      "epoch": 6.1206188822229235,
      "grad_norm": 5.425846099853516,
      "learning_rate": 3.8553836438269655e-05,
      "loss": 0.5269,
      "step": 19384
    },
    {
      "epoch": 6.12093463845911,
      "grad_norm": 5.239013195037842,
      "learning_rate": 3.855295933761359e-05,
      "loss": 0.5697,
      "step": 19385
    },
    {
      "epoch": 6.121250394695295,
      "grad_norm": 4.5617995262146,
      "learning_rate": 3.855208223695752e-05,
      "loss": 0.5248,
      "step": 19386
    },
    {
      "epoch": 6.121566150931481,
      "grad_norm": 3.8840720653533936,
      "learning_rate": 3.855120513630144e-05,
      "loss": 0.4719,
      "step": 19387
    },
    {
      "epoch": 6.1218819071676664,
      "grad_norm": 4.293017387390137,
      "learning_rate": 3.8550328035645375e-05,
      "loss": 0.5361,
      "step": 19388
    },
    {
      "epoch": 6.122197663403853,
      "grad_norm": 4.676685333251953,
      "learning_rate": 3.85494509349893e-05,
      "loss": 0.5955,
      "step": 19389
    },
    {
      "epoch": 6.122513419640038,
      "grad_norm": 4.253477096557617,
      "learning_rate": 3.854857383433323e-05,
      "loss": 0.5821,
      "step": 19390
    },
    {
      "epoch": 6.122829175876223,
      "grad_norm": 5.589310646057129,
      "learning_rate": 3.8547696733677156e-05,
      "loss": 0.4812,
      "step": 19391
    },
    {
      "epoch": 6.123144932112409,
      "grad_norm": 5.284753322601318,
      "learning_rate": 3.854681963302109e-05,
      "loss": 0.4799,
      "step": 19392
    },
    {
      "epoch": 6.123460688348595,
      "grad_norm": 4.498991966247559,
      "learning_rate": 3.854594253236501e-05,
      "loss": 0.5185,
      "step": 19393
    },
    {
      "epoch": 6.123776444584781,
      "grad_norm": 5.457430362701416,
      "learning_rate": 3.8545065431708945e-05,
      "loss": 0.6297,
      "step": 19394
    },
    {
      "epoch": 6.124092200820966,
      "grad_norm": 4.916840076446533,
      "learning_rate": 3.854418833105287e-05,
      "loss": 0.5167,
      "step": 19395
    },
    {
      "epoch": 6.1244079570571515,
      "grad_norm": 4.470439434051514,
      "learning_rate": 3.85433112303968e-05,
      "loss": 0.5751,
      "step": 19396
    },
    {
      "epoch": 6.124723713293338,
      "grad_norm": 4.947014331817627,
      "learning_rate": 3.8542434129740726e-05,
      "loss": 0.5656,
      "step": 19397
    },
    {
      "epoch": 6.125039469529523,
      "grad_norm": 5.033401966094971,
      "learning_rate": 3.854155702908466e-05,
      "loss": 0.6204,
      "step": 19398
    },
    {
      "epoch": 6.125355225765709,
      "grad_norm": 4.596342086791992,
      "learning_rate": 3.854067992842858e-05,
      "loss": 0.6319,
      "step": 19399
    },
    {
      "epoch": 6.125670982001894,
      "grad_norm": 4.8719024658203125,
      "learning_rate": 3.8539802827772514e-05,
      "loss": 0.5012,
      "step": 19400
    },
    {
      "epoch": 6.125670982001894,
      "eval_loss": 0.8798548579216003,
      "eval_runtime": 4.5424,
      "eval_samples_per_second": 548.168,
      "eval_steps_per_second": 4.623,
      "step": 19400
    },
    {
      "epoch": 6.125986738238081,
      "grad_norm": 5.103349685668945,
      "learning_rate": 3.8538925727116446e-05,
      "loss": 0.5668,
      "step": 19401
    },
    {
      "epoch": 6.126302494474266,
      "grad_norm": 4.580458641052246,
      "learning_rate": 3.853804862646038e-05,
      "loss": 0.549,
      "step": 19402
    },
    {
      "epoch": 6.126618250710451,
      "grad_norm": 5.335165023803711,
      "learning_rate": 3.85371715258043e-05,
      "loss": 0.5437,
      "step": 19403
    },
    {
      "epoch": 6.126934006946637,
      "grad_norm": 4.941868782043457,
      "learning_rate": 3.8536294425148234e-05,
      "loss": 0.5748,
      "step": 19404
    },
    {
      "epoch": 6.127249763182823,
      "grad_norm": 5.624961853027344,
      "learning_rate": 3.853541732449216e-05,
      "loss": 0.6418,
      "step": 19405
    },
    {
      "epoch": 6.127565519419009,
      "grad_norm": 4.7583327293396,
      "learning_rate": 3.853454022383609e-05,
      "loss": 0.4979,
      "step": 19406
    },
    {
      "epoch": 6.127881275655194,
      "grad_norm": 3.9327220916748047,
      "learning_rate": 3.853366312318002e-05,
      "loss": 0.5439,
      "step": 19407
    },
    {
      "epoch": 6.128197031891379,
      "grad_norm": 5.230981349945068,
      "learning_rate": 3.853278602252395e-05,
      "loss": 0.546,
      "step": 19408
    },
    {
      "epoch": 6.128512788127566,
      "grad_norm": 4.433962345123291,
      "learning_rate": 3.853190892186788e-05,
      "loss": 0.5397,
      "step": 19409
    },
    {
      "epoch": 6.128828544363751,
      "grad_norm": 5.180168628692627,
      "learning_rate": 3.8531031821211804e-05,
      "loss": 0.5937,
      "step": 19410
    },
    {
      "epoch": 6.129144300599937,
      "grad_norm": 5.043923854827881,
      "learning_rate": 3.8530154720555736e-05,
      "loss": 0.4726,
      "step": 19411
    },
    {
      "epoch": 6.129460056836122,
      "grad_norm": 4.089583873748779,
      "learning_rate": 3.852927761989966e-05,
      "loss": 0.5224,
      "step": 19412
    },
    {
      "epoch": 6.129775813072309,
      "grad_norm": 3.8923072814941406,
      "learning_rate": 3.852840051924359e-05,
      "loss": 0.5762,
      "step": 19413
    },
    {
      "epoch": 6.130091569308494,
      "grad_norm": 4.176920413970947,
      "learning_rate": 3.852752341858752e-05,
      "loss": 0.5916,
      "step": 19414
    },
    {
      "epoch": 6.130407325544679,
      "grad_norm": 5.278459548950195,
      "learning_rate": 3.852664631793145e-05,
      "loss": 0.4696,
      "step": 19415
    },
    {
      "epoch": 6.130723081780865,
      "grad_norm": 5.305477619171143,
      "learning_rate": 3.8525769217275374e-05,
      "loss": 0.5357,
      "step": 19416
    },
    {
      "epoch": 6.131038838017051,
      "grad_norm": 5.533986568450928,
      "learning_rate": 3.8524892116619306e-05,
      "loss": 0.5423,
      "step": 19417
    },
    {
      "epoch": 6.131354594253237,
      "grad_norm": 5.0330424308776855,
      "learning_rate": 3.852401501596323e-05,
      "loss": 0.6065,
      "step": 19418
    },
    {
      "epoch": 6.131670350489422,
      "grad_norm": 5.5713419914245605,
      "learning_rate": 3.852313791530716e-05,
      "loss": 0.5526,
      "step": 19419
    },
    {
      "epoch": 6.131986106725607,
      "grad_norm": 4.412535667419434,
      "learning_rate": 3.852226081465109e-05,
      "loss": 0.4935,
      "step": 19420
    },
    {
      "epoch": 6.132301862961794,
      "grad_norm": 5.053867340087891,
      "learning_rate": 3.852138371399502e-05,
      "loss": 0.5824,
      "step": 19421
    },
    {
      "epoch": 6.132617619197979,
      "grad_norm": 5.331984519958496,
      "learning_rate": 3.8520506613338944e-05,
      "loss": 0.5377,
      "step": 19422
    },
    {
      "epoch": 6.132933375434165,
      "grad_norm": 4.6811747550964355,
      "learning_rate": 3.8519629512682875e-05,
      "loss": 0.4492,
      "step": 19423
    },
    {
      "epoch": 6.13324913167035,
      "grad_norm": 4.023958683013916,
      "learning_rate": 3.851875241202681e-05,
      "loss": 0.5575,
      "step": 19424
    },
    {
      "epoch": 6.1335648879065365,
      "grad_norm": 4.496096611022949,
      "learning_rate": 3.851787531137073e-05,
      "loss": 0.4612,
      "step": 19425
    },
    {
      "epoch": 6.133880644142722,
      "grad_norm": 3.2940540313720703,
      "learning_rate": 3.8516998210714664e-05,
      "loss": 0.4188,
      "step": 19426
    },
    {
      "epoch": 6.134196400378907,
      "grad_norm": 4.537318706512451,
      "learning_rate": 3.851612111005859e-05,
      "loss": 0.5641,
      "step": 19427
    },
    {
      "epoch": 6.134512156615093,
      "grad_norm": 4.682945728302002,
      "learning_rate": 3.851524400940252e-05,
      "loss": 0.3982,
      "step": 19428
    },
    {
      "epoch": 6.134827912851279,
      "grad_norm": 6.946424961090088,
      "learning_rate": 3.851436690874645e-05,
      "loss": 0.6101,
      "step": 19429
    },
    {
      "epoch": 6.135143669087465,
      "grad_norm": 6.084756851196289,
      "learning_rate": 3.8513489808090384e-05,
      "loss": 0.7112,
      "step": 19430
    },
    {
      "epoch": 6.13545942532365,
      "grad_norm": 4.466704845428467,
      "learning_rate": 3.851261270743431e-05,
      "loss": 0.5293,
      "step": 19431
    },
    {
      "epoch": 6.135775181559836,
      "grad_norm": 5.640931129455566,
      "learning_rate": 3.851173560677824e-05,
      "loss": 0.5879,
      "step": 19432
    },
    {
      "epoch": 6.1360909377960215,
      "grad_norm": 4.947343349456787,
      "learning_rate": 3.8510858506122165e-05,
      "loss": 0.5092,
      "step": 19433
    },
    {
      "epoch": 6.136406694032207,
      "grad_norm": 5.106344223022461,
      "learning_rate": 3.85099814054661e-05,
      "loss": 0.4795,
      "step": 19434
    },
    {
      "epoch": 6.136722450268393,
      "grad_norm": 5.007274627685547,
      "learning_rate": 3.850910430481002e-05,
      "loss": 0.5233,
      "step": 19435
    },
    {
      "epoch": 6.137038206504578,
      "grad_norm": 4.267892360687256,
      "learning_rate": 3.850822720415395e-05,
      "loss": 0.5684,
      "step": 19436
    },
    {
      "epoch": 6.1373539627407645,
      "grad_norm": 4.664697170257568,
      "learning_rate": 3.850735010349788e-05,
      "loss": 0.5439,
      "step": 19437
    },
    {
      "epoch": 6.13766971897695,
      "grad_norm": 4.794332027435303,
      "learning_rate": 3.850647300284181e-05,
      "loss": 0.5175,
      "step": 19438
    },
    {
      "epoch": 6.137985475213135,
      "grad_norm": 4.52910041809082,
      "learning_rate": 3.8505595902185735e-05,
      "loss": 0.5178,
      "step": 19439
    },
    {
      "epoch": 6.138301231449321,
      "grad_norm": 3.932075262069702,
      "learning_rate": 3.8504718801529667e-05,
      "loss": 0.441,
      "step": 19440
    },
    {
      "epoch": 6.138616987685507,
      "grad_norm": 6.4728593826293945,
      "learning_rate": 3.850384170087359e-05,
      "loss": 0.6397,
      "step": 19441
    },
    {
      "epoch": 6.138932743921693,
      "grad_norm": 5.335690498352051,
      "learning_rate": 3.850296460021752e-05,
      "loss": 0.5666,
      "step": 19442
    },
    {
      "epoch": 6.139248500157878,
      "grad_norm": 4.604469299316406,
      "learning_rate": 3.850208749956145e-05,
      "loss": 0.5716,
      "step": 19443
    },
    {
      "epoch": 6.139564256394063,
      "grad_norm": 5.776535511016846,
      "learning_rate": 3.850121039890538e-05,
      "loss": 0.5847,
      "step": 19444
    },
    {
      "epoch": 6.1398800126302495,
      "grad_norm": 5.1232123374938965,
      "learning_rate": 3.850033329824931e-05,
      "loss": 0.6612,
      "step": 19445
    },
    {
      "epoch": 6.140195768866435,
      "grad_norm": 4.444897651672363,
      "learning_rate": 3.8499456197593236e-05,
      "loss": 0.5221,
      "step": 19446
    },
    {
      "epoch": 6.140511525102621,
      "grad_norm": 4.5114569664001465,
      "learning_rate": 3.849857909693717e-05,
      "loss": 0.5583,
      "step": 19447
    },
    {
      "epoch": 6.140827281338806,
      "grad_norm": 5.553089141845703,
      "learning_rate": 3.849770199628109e-05,
      "loss": 0.5139,
      "step": 19448
    },
    {
      "epoch": 6.1411430375749925,
      "grad_norm": 4.938470840454102,
      "learning_rate": 3.8496824895625025e-05,
      "loss": 0.5914,
      "step": 19449
    },
    {
      "epoch": 6.141458793811178,
      "grad_norm": 5.231996536254883,
      "learning_rate": 3.849594779496895e-05,
      "loss": 0.5673,
      "step": 19450
    },
    {
      "epoch": 6.141774550047363,
      "grad_norm": 5.379411697387695,
      "learning_rate": 3.849507069431288e-05,
      "loss": 0.5618,
      "step": 19451
    },
    {
      "epoch": 6.142090306283549,
      "grad_norm": 4.958529949188232,
      "learning_rate": 3.8494193593656806e-05,
      "loss": 0.5729,
      "step": 19452
    },
    {
      "epoch": 6.1424060625197345,
      "grad_norm": 4.197965621948242,
      "learning_rate": 3.849331649300074e-05,
      "loss": 0.5077,
      "step": 19453
    },
    {
      "epoch": 6.142721818755921,
      "grad_norm": 6.331445217132568,
      "learning_rate": 3.849243939234466e-05,
      "loss": 0.5985,
      "step": 19454
    },
    {
      "epoch": 6.143037574992106,
      "grad_norm": 5.311405658721924,
      "learning_rate": 3.84915622916886e-05,
      "loss": 0.5681,
      "step": 19455
    },
    {
      "epoch": 6.143353331228292,
      "grad_norm": 4.485727787017822,
      "learning_rate": 3.8490685191032526e-05,
      "loss": 0.5746,
      "step": 19456
    },
    {
      "epoch": 6.1436690874644775,
      "grad_norm": 6.19212532043457,
      "learning_rate": 3.848980809037646e-05,
      "loss": 0.4512,
      "step": 19457
    },
    {
      "epoch": 6.143984843700663,
      "grad_norm": 4.867109298706055,
      "learning_rate": 3.848893098972038e-05,
      "loss": 0.519,
      "step": 19458
    },
    {
      "epoch": 6.144300599936849,
      "grad_norm": 4.168518543243408,
      "learning_rate": 3.8488053889064314e-05,
      "loss": 0.5306,
      "step": 19459
    },
    {
      "epoch": 6.144616356173034,
      "grad_norm": 4.107864856719971,
      "learning_rate": 3.848717678840824e-05,
      "loss": 0.618,
      "step": 19460
    },
    {
      "epoch": 6.14493211240922,
      "grad_norm": 4.4371161460876465,
      "learning_rate": 3.848629968775217e-05,
      "loss": 0.6437,
      "step": 19461
    },
    {
      "epoch": 6.145247868645406,
      "grad_norm": 5.158352375030518,
      "learning_rate": 3.8485422587096096e-05,
      "loss": 0.5914,
      "step": 19462
    },
    {
      "epoch": 6.145563624881591,
      "grad_norm": 4.545846939086914,
      "learning_rate": 3.848454548644003e-05,
      "loss": 0.4386,
      "step": 19463
    },
    {
      "epoch": 6.145879381117777,
      "grad_norm": 4.471569061279297,
      "learning_rate": 3.848366838578395e-05,
      "loss": 0.5256,
      "step": 19464
    },
    {
      "epoch": 6.1461951373539625,
      "grad_norm": 4.518599033355713,
      "learning_rate": 3.8482791285127884e-05,
      "loss": 0.4892,
      "step": 19465
    },
    {
      "epoch": 6.146510893590149,
      "grad_norm": 4.528811931610107,
      "learning_rate": 3.848191418447181e-05,
      "loss": 0.5405,
      "step": 19466
    },
    {
      "epoch": 6.146826649826334,
      "grad_norm": 5.390522003173828,
      "learning_rate": 3.848103708381574e-05,
      "loss": 0.5322,
      "step": 19467
    },
    {
      "epoch": 6.14714240606252,
      "grad_norm": 4.110266208648682,
      "learning_rate": 3.848015998315967e-05,
      "loss": 0.5548,
      "step": 19468
    },
    {
      "epoch": 6.1474581622987055,
      "grad_norm": 5.146378517150879,
      "learning_rate": 3.84792828825036e-05,
      "loss": 0.5403,
      "step": 19469
    },
    {
      "epoch": 6.147773918534891,
      "grad_norm": 4.586768627166748,
      "learning_rate": 3.847840578184753e-05,
      "loss": 0.6019,
      "step": 19470
    },
    {
      "epoch": 6.148089674771077,
      "grad_norm": 4.379913806915283,
      "learning_rate": 3.8477528681191454e-05,
      "loss": 0.521,
      "step": 19471
    },
    {
      "epoch": 6.148405431007262,
      "grad_norm": 5.462595462799072,
      "learning_rate": 3.8476651580535386e-05,
      "loss": 0.5798,
      "step": 19472
    },
    {
      "epoch": 6.148721187243448,
      "grad_norm": 5.03527307510376,
      "learning_rate": 3.847577447987931e-05,
      "loss": 0.4993,
      "step": 19473
    },
    {
      "epoch": 6.149036943479634,
      "grad_norm": 5.015912055969238,
      "learning_rate": 3.847489737922324e-05,
      "loss": 0.596,
      "step": 19474
    },
    {
      "epoch": 6.149352699715819,
      "grad_norm": 4.389493465423584,
      "learning_rate": 3.847402027856717e-05,
      "loss": 0.4734,
      "step": 19475
    },
    {
      "epoch": 6.149668455952005,
      "grad_norm": 4.860287666320801,
      "learning_rate": 3.84731431779111e-05,
      "loss": 0.4606,
      "step": 19476
    },
    {
      "epoch": 6.1499842121881905,
      "grad_norm": 4.5777082443237305,
      "learning_rate": 3.8472266077255024e-05,
      "loss": 0.5189,
      "step": 19477
    },
    {
      "epoch": 6.150299968424377,
      "grad_norm": 4.3076982498168945,
      "learning_rate": 3.8471388976598955e-05,
      "loss": 0.5864,
      "step": 19478
    },
    {
      "epoch": 6.150615724660562,
      "grad_norm": 5.868874549865723,
      "learning_rate": 3.847051187594288e-05,
      "loss": 0.6098,
      "step": 19479
    },
    {
      "epoch": 6.150931480896748,
      "grad_norm": 4.7932515144348145,
      "learning_rate": 3.846963477528681e-05,
      "loss": 0.529,
      "step": 19480
    },
    {
      "epoch": 6.151247237132933,
      "grad_norm": 5.598609924316406,
      "learning_rate": 3.846875767463074e-05,
      "loss": 0.5344,
      "step": 19481
    },
    {
      "epoch": 6.151562993369119,
      "grad_norm": 5.0869011878967285,
      "learning_rate": 3.8467880573974675e-05,
      "loss": 0.4717,
      "step": 19482
    },
    {
      "epoch": 6.151878749605305,
      "grad_norm": 5.162927627563477,
      "learning_rate": 3.84670034733186e-05,
      "loss": 0.5596,
      "step": 19483
    },
    {
      "epoch": 6.15219450584149,
      "grad_norm": 4.92724084854126,
      "learning_rate": 3.846612637266253e-05,
      "loss": 0.5266,
      "step": 19484
    },
    {
      "epoch": 6.152510262077676,
      "grad_norm": 5.389407634735107,
      "learning_rate": 3.846524927200646e-05,
      "loss": 0.6065,
      "step": 19485
    },
    {
      "epoch": 6.152826018313862,
      "grad_norm": 6.407480716705322,
      "learning_rate": 3.846437217135039e-05,
      "loss": 0.6459,
      "step": 19486
    },
    {
      "epoch": 6.153141774550047,
      "grad_norm": 3.906205177307129,
      "learning_rate": 3.846349507069431e-05,
      "loss": 0.5137,
      "step": 19487
    },
    {
      "epoch": 6.153457530786233,
      "grad_norm": 4.988410949707031,
      "learning_rate": 3.8462617970038245e-05,
      "loss": 0.5323,
      "step": 19488
    },
    {
      "epoch": 6.153773287022418,
      "grad_norm": 4.865865707397461,
      "learning_rate": 3.846174086938218e-05,
      "loss": 0.5183,
      "step": 19489
    },
    {
      "epoch": 6.154089043258605,
      "grad_norm": 4.77100944519043,
      "learning_rate": 3.84608637687261e-05,
      "loss": 0.4992,
      "step": 19490
    },
    {
      "epoch": 6.15440479949479,
      "grad_norm": 4.187543869018555,
      "learning_rate": 3.845998666807003e-05,
      "loss": 0.5428,
      "step": 19491
    },
    {
      "epoch": 6.154720555730976,
      "grad_norm": 4.887997627258301,
      "learning_rate": 3.845910956741396e-05,
      "loss": 0.65,
      "step": 19492
    },
    {
      "epoch": 6.155036311967161,
      "grad_norm": 5.002127647399902,
      "learning_rate": 3.845823246675789e-05,
      "loss": 0.6339,
      "step": 19493
    },
    {
      "epoch": 6.155352068203347,
      "grad_norm": 4.9695892333984375,
      "learning_rate": 3.8457355366101815e-05,
      "loss": 0.5248,
      "step": 19494
    },
    {
      "epoch": 6.155667824439533,
      "grad_norm": 4.153435230255127,
      "learning_rate": 3.8456478265445746e-05,
      "loss": 0.4835,
      "step": 19495
    },
    {
      "epoch": 6.155983580675718,
      "grad_norm": 4.852488994598389,
      "learning_rate": 3.845560116478967e-05,
      "loss": 0.5545,
      "step": 19496
    },
    {
      "epoch": 6.156299336911904,
      "grad_norm": 5.169232368469238,
      "learning_rate": 3.84547240641336e-05,
      "loss": 0.614,
      "step": 19497
    },
    {
      "epoch": 6.15661509314809,
      "grad_norm": 3.9550437927246094,
      "learning_rate": 3.845384696347753e-05,
      "loss": 0.4791,
      "step": 19498
    },
    {
      "epoch": 6.156930849384275,
      "grad_norm": 3.9263548851013184,
      "learning_rate": 3.845296986282146e-05,
      "loss": 0.4846,
      "step": 19499
    },
    {
      "epoch": 6.157246605620461,
      "grad_norm": 4.615304946899414,
      "learning_rate": 3.8452092762165385e-05,
      "loss": 0.5364,
      "step": 19500
    },
    {
      "epoch": 6.157562361856646,
      "grad_norm": 4.940096378326416,
      "learning_rate": 3.8451215661509316e-05,
      "loss": 0.5288,
      "step": 19501
    },
    {
      "epoch": 6.157878118092833,
      "grad_norm": 5.608459949493408,
      "learning_rate": 3.845033856085324e-05,
      "loss": 0.5587,
      "step": 19502
    },
    {
      "epoch": 6.158193874329018,
      "grad_norm": 4.582699775695801,
      "learning_rate": 3.844946146019717e-05,
      "loss": 0.6035,
      "step": 19503
    },
    {
      "epoch": 6.158509630565204,
      "grad_norm": 4.884612560272217,
      "learning_rate": 3.84485843595411e-05,
      "loss": 0.6622,
      "step": 19504
    },
    {
      "epoch": 6.158825386801389,
      "grad_norm": 4.645603179931641,
      "learning_rate": 3.844770725888503e-05,
      "loss": 0.6148,
      "step": 19505
    },
    {
      "epoch": 6.159141143037575,
      "grad_norm": 4.372463703155518,
      "learning_rate": 3.844683015822896e-05,
      "loss": 0.5731,
      "step": 19506
    },
    {
      "epoch": 6.159456899273761,
      "grad_norm": 4.294646739959717,
      "learning_rate": 3.8445953057572886e-05,
      "loss": 0.5367,
      "step": 19507
    },
    {
      "epoch": 6.159772655509946,
      "grad_norm": 4.616035461425781,
      "learning_rate": 3.844507595691682e-05,
      "loss": 0.5127,
      "step": 19508
    },
    {
      "epoch": 6.160088411746132,
      "grad_norm": 4.651458740234375,
      "learning_rate": 3.844419885626074e-05,
      "loss": 0.4418,
      "step": 19509
    },
    {
      "epoch": 6.160404167982318,
      "grad_norm": 5.990492820739746,
      "learning_rate": 3.844332175560468e-05,
      "loss": 0.669,
      "step": 19510
    },
    {
      "epoch": 6.160719924218503,
      "grad_norm": 3.715195655822754,
      "learning_rate": 3.8442444654948606e-05,
      "loss": 0.5513,
      "step": 19511
    },
    {
      "epoch": 6.161035680454689,
      "grad_norm": 4.561344146728516,
      "learning_rate": 3.844156755429254e-05,
      "loss": 0.5698,
      "step": 19512
    },
    {
      "epoch": 6.161351436690874,
      "grad_norm": 5.895358085632324,
      "learning_rate": 3.844069045363646e-05,
      "loss": 0.54,
      "step": 19513
    },
    {
      "epoch": 6.1616671929270606,
      "grad_norm": 4.409156799316406,
      "learning_rate": 3.8439813352980394e-05,
      "loss": 0.5355,
      "step": 19514
    },
    {
      "epoch": 6.161982949163246,
      "grad_norm": 4.31409215927124,
      "learning_rate": 3.843893625232432e-05,
      "loss": 0.4769,
      "step": 19515
    },
    {
      "epoch": 6.162298705399432,
      "grad_norm": 4.8109450340271,
      "learning_rate": 3.843805915166825e-05,
      "loss": 0.5403,
      "step": 19516
    },
    {
      "epoch": 6.162614461635617,
      "grad_norm": 4.7534403800964355,
      "learning_rate": 3.8437182051012176e-05,
      "loss": 0.5635,
      "step": 19517
    },
    {
      "epoch": 6.162930217871803,
      "grad_norm": 5.106140613555908,
      "learning_rate": 3.843630495035611e-05,
      "loss": 0.5126,
      "step": 19518
    },
    {
      "epoch": 6.163245974107989,
      "grad_norm": 4.272737503051758,
      "learning_rate": 3.843542784970003e-05,
      "loss": 0.5356,
      "step": 19519
    },
    {
      "epoch": 6.163561730344174,
      "grad_norm": 5.043898582458496,
      "learning_rate": 3.8434550749043964e-05,
      "loss": 0.6315,
      "step": 19520
    },
    {
      "epoch": 6.16387748658036,
      "grad_norm": 5.650783538818359,
      "learning_rate": 3.843367364838789e-05,
      "loss": 0.4774,
      "step": 19521
    },
    {
      "epoch": 6.164193242816546,
      "grad_norm": 6.705132007598877,
      "learning_rate": 3.843279654773182e-05,
      "loss": 0.5289,
      "step": 19522
    },
    {
      "epoch": 6.164508999052731,
      "grad_norm": 6.338131904602051,
      "learning_rate": 3.8431919447075745e-05,
      "loss": 0.5538,
      "step": 19523
    },
    {
      "epoch": 6.164824755288917,
      "grad_norm": 4.217652320861816,
      "learning_rate": 3.843104234641968e-05,
      "loss": 0.4691,
      "step": 19524
    },
    {
      "epoch": 6.165140511525102,
      "grad_norm": 5.32870626449585,
      "learning_rate": 3.84301652457636e-05,
      "loss": 0.5282,
      "step": 19525
    },
    {
      "epoch": 6.1654562677612885,
      "grad_norm": 3.5651869773864746,
      "learning_rate": 3.8429288145107534e-05,
      "loss": 0.4553,
      "step": 19526
    },
    {
      "epoch": 6.165772023997474,
      "grad_norm": 5.512937545776367,
      "learning_rate": 3.8428411044451465e-05,
      "loss": 0.4922,
      "step": 19527
    },
    {
      "epoch": 6.16608778023366,
      "grad_norm": 4.31821870803833,
      "learning_rate": 3.842753394379539e-05,
      "loss": 0.429,
      "step": 19528
    },
    {
      "epoch": 6.166403536469845,
      "grad_norm": 4.664306640625,
      "learning_rate": 3.842665684313932e-05,
      "loss": 0.5064,
      "step": 19529
    },
    {
      "epoch": 6.166719292706031,
      "grad_norm": 5.8981709480285645,
      "learning_rate": 3.842577974248325e-05,
      "loss": 0.5153,
      "step": 19530
    },
    {
      "epoch": 6.167035048942217,
      "grad_norm": 5.166636943817139,
      "learning_rate": 3.842490264182718e-05,
      "loss": 0.6178,
      "step": 19531
    },
    {
      "epoch": 6.167350805178402,
      "grad_norm": 5.049495697021484,
      "learning_rate": 3.8424025541171103e-05,
      "loss": 0.569,
      "step": 19532
    },
    {
      "epoch": 6.167666561414588,
      "grad_norm": 4.452121734619141,
      "learning_rate": 3.8423148440515035e-05,
      "loss": 0.5645,
      "step": 19533
    },
    {
      "epoch": 6.1679823176507735,
      "grad_norm": 4.394960403442383,
      "learning_rate": 3.842227133985896e-05,
      "loss": 0.4747,
      "step": 19534
    },
    {
      "epoch": 6.168298073886959,
      "grad_norm": 5.603433609008789,
      "learning_rate": 3.842139423920289e-05,
      "loss": 0.5788,
      "step": 19535
    },
    {
      "epoch": 6.168613830123145,
      "grad_norm": 5.248440265655518,
      "learning_rate": 3.842051713854682e-05,
      "loss": 0.5022,
      "step": 19536
    },
    {
      "epoch": 6.16892958635933,
      "grad_norm": 4.762298583984375,
      "learning_rate": 3.8419640037890755e-05,
      "loss": 0.5723,
      "step": 19537
    },
    {
      "epoch": 6.1692453425955165,
      "grad_norm": 6.306025981903076,
      "learning_rate": 3.841876293723468e-05,
      "loss": 0.5855,
      "step": 19538
    },
    {
      "epoch": 6.169561098831702,
      "grad_norm": 5.55889368057251,
      "learning_rate": 3.841788583657861e-05,
      "loss": 0.5663,
      "step": 19539
    },
    {
      "epoch": 6.169876855067888,
      "grad_norm": 4.6321702003479,
      "learning_rate": 3.8417008735922537e-05,
      "loss": 0.5427,
      "step": 19540
    },
    {
      "epoch": 6.170192611304073,
      "grad_norm": 5.3087286949157715,
      "learning_rate": 3.841613163526647e-05,
      "loss": 0.5945,
      "step": 19541
    },
    {
      "epoch": 6.1705083675402586,
      "grad_norm": 4.082034587860107,
      "learning_rate": 3.841525453461039e-05,
      "loss": 0.5194,
      "step": 19542
    },
    {
      "epoch": 6.170824123776445,
      "grad_norm": 5.01057243347168,
      "learning_rate": 3.8414377433954325e-05,
      "loss": 0.5873,
      "step": 19543
    },
    {
      "epoch": 6.17113988001263,
      "grad_norm": 5.0829877853393555,
      "learning_rate": 3.841350033329825e-05,
      "loss": 0.5879,
      "step": 19544
    },
    {
      "epoch": 6.171455636248816,
      "grad_norm": 5.094765663146973,
      "learning_rate": 3.841262323264218e-05,
      "loss": 0.5279,
      "step": 19545
    },
    {
      "epoch": 6.1717713924850015,
      "grad_norm": 5.958986759185791,
      "learning_rate": 3.8411746131986106e-05,
      "loss": 0.5588,
      "step": 19546
    },
    {
      "epoch": 6.172087148721188,
      "grad_norm": 6.553628921508789,
      "learning_rate": 3.841086903133004e-05,
      "loss": 0.7208,
      "step": 19547
    },
    {
      "epoch": 6.172402904957373,
      "grad_norm": 5.172684669494629,
      "learning_rate": 3.840999193067397e-05,
      "loss": 0.5377,
      "step": 19548
    },
    {
      "epoch": 6.172718661193558,
      "grad_norm": 5.087648868560791,
      "learning_rate": 3.8409114830017895e-05,
      "loss": 0.5522,
      "step": 19549
    },
    {
      "epoch": 6.1730344174297445,
      "grad_norm": 5.581910133361816,
      "learning_rate": 3.8408237729361826e-05,
      "loss": 0.5788,
      "step": 19550
    },
    {
      "epoch": 6.17335017366593,
      "grad_norm": 4.380098819732666,
      "learning_rate": 3.840736062870575e-05,
      "loss": 0.5084,
      "step": 19551
    },
    {
      "epoch": 6.173665929902116,
      "grad_norm": 4.318737030029297,
      "learning_rate": 3.840648352804968e-05,
      "loss": 0.6188,
      "step": 19552
    },
    {
      "epoch": 6.173981686138301,
      "grad_norm": 5.920050621032715,
      "learning_rate": 3.840560642739361e-05,
      "loss": 0.6012,
      "step": 19553
    },
    {
      "epoch": 6.1742974423744865,
      "grad_norm": 4.274579048156738,
      "learning_rate": 3.840472932673754e-05,
      "loss": 0.5644,
      "step": 19554
    },
    {
      "epoch": 6.174613198610673,
      "grad_norm": 5.396795272827148,
      "learning_rate": 3.8403852226081464e-05,
      "loss": 0.6242,
      "step": 19555
    },
    {
      "epoch": 6.174928954846858,
      "grad_norm": 5.1551618576049805,
      "learning_rate": 3.8402975125425396e-05,
      "loss": 0.47,
      "step": 19556
    },
    {
      "epoch": 6.175244711083044,
      "grad_norm": 4.610315799713135,
      "learning_rate": 3.840209802476932e-05,
      "loss": 0.5414,
      "step": 19557
    },
    {
      "epoch": 6.1755604673192295,
      "grad_norm": 4.505842208862305,
      "learning_rate": 3.840122092411325e-05,
      "loss": 0.5894,
      "step": 19558
    },
    {
      "epoch": 6.175876223555415,
      "grad_norm": 4.93370246887207,
      "learning_rate": 3.840034382345718e-05,
      "loss": 0.5672,
      "step": 19559
    },
    {
      "epoch": 6.176191979791601,
      "grad_norm": 4.575852394104004,
      "learning_rate": 3.839946672280111e-05,
      "loss": 0.5639,
      "step": 19560
    },
    {
      "epoch": 6.176507736027786,
      "grad_norm": 5.012190818786621,
      "learning_rate": 3.8398589622145034e-05,
      "loss": 0.6378,
      "step": 19561
    },
    {
      "epoch": 6.176823492263972,
      "grad_norm": 4.42702579498291,
      "learning_rate": 3.8397712521488966e-05,
      "loss": 0.5539,
      "step": 19562
    },
    {
      "epoch": 6.177139248500158,
      "grad_norm": 4.940619468688965,
      "learning_rate": 3.839683542083289e-05,
      "loss": 0.5644,
      "step": 19563
    },
    {
      "epoch": 6.177455004736344,
      "grad_norm": 5.851883411407471,
      "learning_rate": 3.839595832017683e-05,
      "loss": 0.5148,
      "step": 19564
    },
    {
      "epoch": 6.177770760972529,
      "grad_norm": 4.130856037139893,
      "learning_rate": 3.8395081219520754e-05,
      "loss": 0.5799,
      "step": 19565
    },
    {
      "epoch": 6.1780865172087145,
      "grad_norm": 3.7386069297790527,
      "learning_rate": 3.8394204118864686e-05,
      "loss": 0.5107,
      "step": 19566
    },
    {
      "epoch": 6.178402273444901,
      "grad_norm": 4.5685858726501465,
      "learning_rate": 3.839332701820861e-05,
      "loss": 0.5995,
      "step": 19567
    },
    {
      "epoch": 6.178718029681086,
      "grad_norm": 4.28421688079834,
      "learning_rate": 3.839244991755254e-05,
      "loss": 0.5833,
      "step": 19568
    },
    {
      "epoch": 6.179033785917272,
      "grad_norm": 4.701530456542969,
      "learning_rate": 3.839157281689647e-05,
      "loss": 0.6327,
      "step": 19569
    },
    {
      "epoch": 6.179349542153457,
      "grad_norm": 4.348891258239746,
      "learning_rate": 3.83906957162404e-05,
      "loss": 0.5832,
      "step": 19570
    },
    {
      "epoch": 6.179665298389644,
      "grad_norm": 4.612054347991943,
      "learning_rate": 3.838981861558433e-05,
      "loss": 0.4358,
      "step": 19571
    },
    {
      "epoch": 6.179981054625829,
      "grad_norm": 4.32236385345459,
      "learning_rate": 3.8388941514928256e-05,
      "loss": 0.5423,
      "step": 19572
    },
    {
      "epoch": 6.180296810862014,
      "grad_norm": 4.047733783721924,
      "learning_rate": 3.838806441427219e-05,
      "loss": 0.5738,
      "step": 19573
    },
    {
      "epoch": 6.1806125670982,
      "grad_norm": 4.904108047485352,
      "learning_rate": 3.838718731361611e-05,
      "loss": 0.5708,
      "step": 19574
    },
    {
      "epoch": 6.180928323334386,
      "grad_norm": 3.8999338150024414,
      "learning_rate": 3.8386310212960044e-05,
      "loss": 0.454,
      "step": 19575
    },
    {
      "epoch": 6.181244079570572,
      "grad_norm": 5.219375133514404,
      "learning_rate": 3.838543311230397e-05,
      "loss": 0.5486,
      "step": 19576
    },
    {
      "epoch": 6.181559835806757,
      "grad_norm": 5.154388427734375,
      "learning_rate": 3.83845560116479e-05,
      "loss": 0.6019,
      "step": 19577
    },
    {
      "epoch": 6.1818755920429425,
      "grad_norm": 4.222980499267578,
      "learning_rate": 3.8383678910991825e-05,
      "loss": 0.5474,
      "step": 19578
    },
    {
      "epoch": 6.182191348279129,
      "grad_norm": 4.177022457122803,
      "learning_rate": 3.838280181033576e-05,
      "loss": 0.4356,
      "step": 19579
    },
    {
      "epoch": 6.182507104515314,
      "grad_norm": 4.405401229858398,
      "learning_rate": 3.838192470967968e-05,
      "loss": 0.5271,
      "step": 19580
    },
    {
      "epoch": 6.1828228607515,
      "grad_norm": 6.218269348144531,
      "learning_rate": 3.8381047609023614e-05,
      "loss": 0.6251,
      "step": 19581
    },
    {
      "epoch": 6.183138616987685,
      "grad_norm": 4.51823616027832,
      "learning_rate": 3.838017050836754e-05,
      "loss": 0.64,
      "step": 19582
    },
    {
      "epoch": 6.183454373223872,
      "grad_norm": 3.96747088432312,
      "learning_rate": 3.837929340771147e-05,
      "loss": 0.4601,
      "step": 19583
    },
    {
      "epoch": 6.183770129460057,
      "grad_norm": 3.8905107975006104,
      "learning_rate": 3.8378416307055395e-05,
      "loss": 0.5093,
      "step": 19584
    },
    {
      "epoch": 6.184085885696242,
      "grad_norm": 4.312781810760498,
      "learning_rate": 3.837753920639933e-05,
      "loss": 0.5149,
      "step": 19585
    },
    {
      "epoch": 6.184401641932428,
      "grad_norm": 5.126269817352295,
      "learning_rate": 3.837666210574326e-05,
      "loss": 0.6192,
      "step": 19586
    },
    {
      "epoch": 6.184717398168614,
      "grad_norm": 4.821691513061523,
      "learning_rate": 3.837578500508718e-05,
      "loss": 0.5187,
      "step": 19587
    },
    {
      "epoch": 6.1850331544048,
      "grad_norm": 4.4927239418029785,
      "learning_rate": 3.8374907904431115e-05,
      "loss": 0.5575,
      "step": 19588
    },
    {
      "epoch": 6.185348910640985,
      "grad_norm": 4.272158622741699,
      "learning_rate": 3.837403080377504e-05,
      "loss": 0.504,
      "step": 19589
    },
    {
      "epoch": 6.18566466687717,
      "grad_norm": 4.491067886352539,
      "learning_rate": 3.837315370311897e-05,
      "loss": 0.5397,
      "step": 19590
    },
    {
      "epoch": 6.185980423113357,
      "grad_norm": 4.805172443389893,
      "learning_rate": 3.83722766024629e-05,
      "loss": 0.5154,
      "step": 19591
    },
    {
      "epoch": 6.186296179349542,
      "grad_norm": 4.463881015777588,
      "learning_rate": 3.8371399501806835e-05,
      "loss": 0.5117,
      "step": 19592
    },
    {
      "epoch": 6.186611935585728,
      "grad_norm": 4.773512840270996,
      "learning_rate": 3.837052240115076e-05,
      "loss": 0.5959,
      "step": 19593
    },
    {
      "epoch": 6.186927691821913,
      "grad_norm": 4.805020332336426,
      "learning_rate": 3.836964530049469e-05,
      "loss": 0.5353,
      "step": 19594
    },
    {
      "epoch": 6.1872434480581,
      "grad_norm": 5.176672458648682,
      "learning_rate": 3.8368768199838617e-05,
      "loss": 0.6612,
      "step": 19595
    },
    {
      "epoch": 6.187559204294285,
      "grad_norm": 4.892606258392334,
      "learning_rate": 3.836789109918255e-05,
      "loss": 0.4654,
      "step": 19596
    },
    {
      "epoch": 6.18787496053047,
      "grad_norm": 4.715699672698975,
      "learning_rate": 3.836701399852647e-05,
      "loss": 0.6571,
      "step": 19597
    },
    {
      "epoch": 6.188190716766656,
      "grad_norm": 4.25923490524292,
      "learning_rate": 3.8366136897870405e-05,
      "loss": 0.5024,
      "step": 19598
    },
    {
      "epoch": 6.188506473002842,
      "grad_norm": 4.299834728240967,
      "learning_rate": 3.836525979721433e-05,
      "loss": 0.5322,
      "step": 19599
    },
    {
      "epoch": 6.188822229239028,
      "grad_norm": 4.143624305725098,
      "learning_rate": 3.836438269655826e-05,
      "loss": 0.5281,
      "step": 19600
    },
    {
      "epoch": 6.188822229239028,
      "eval_loss": 0.8869750499725342,
      "eval_runtime": 4.5469,
      "eval_samples_per_second": 547.629,
      "eval_steps_per_second": 4.619,
      "step": 19600
    },
    {
      "epoch": 6.189137985475213,
      "grad_norm": 4.815510272979736,
      "learning_rate": 3.8363505595902186e-05,
      "loss": 0.5694,
      "step": 19601
    },
    {
      "epoch": 6.189453741711398,
      "grad_norm": 4.709219932556152,
      "learning_rate": 3.836262849524612e-05,
      "loss": 0.5841,
      "step": 19602
    },
    {
      "epoch": 6.189769497947585,
      "grad_norm": 6.406580448150635,
      "learning_rate": 3.836175139459004e-05,
      "loss": 0.5649,
      "step": 19603
    },
    {
      "epoch": 6.19008525418377,
      "grad_norm": 6.867364406585693,
      "learning_rate": 3.8360874293933975e-05,
      "loss": 0.6053,
      "step": 19604
    },
    {
      "epoch": 6.190401010419956,
      "grad_norm": 4.6610260009765625,
      "learning_rate": 3.83599971932779e-05,
      "loss": 0.5173,
      "step": 19605
    },
    {
      "epoch": 6.190716766656141,
      "grad_norm": 4.536859512329102,
      "learning_rate": 3.835912009262183e-05,
      "loss": 0.5736,
      "step": 19606
    },
    {
      "epoch": 6.1910325228923275,
      "grad_norm": 5.49523401260376,
      "learning_rate": 3.8358242991965756e-05,
      "loss": 0.5446,
      "step": 19607
    },
    {
      "epoch": 6.191348279128513,
      "grad_norm": 4.486024856567383,
      "learning_rate": 3.835736589130969e-05,
      "loss": 0.4695,
      "step": 19608
    },
    {
      "epoch": 6.191664035364698,
      "grad_norm": 4.040520668029785,
      "learning_rate": 3.835648879065362e-05,
      "loss": 0.5174,
      "step": 19609
    },
    {
      "epoch": 6.191979791600884,
      "grad_norm": 4.146780490875244,
      "learning_rate": 3.8355611689997544e-05,
      "loss": 0.5544,
      "step": 19610
    },
    {
      "epoch": 6.19229554783707,
      "grad_norm": 3.999458074569702,
      "learning_rate": 3.8354734589341476e-05,
      "loss": 0.4561,
      "step": 19611
    },
    {
      "epoch": 6.192611304073256,
      "grad_norm": 6.057984352111816,
      "learning_rate": 3.83538574886854e-05,
      "loss": 0.6125,
      "step": 19612
    },
    {
      "epoch": 6.192927060309441,
      "grad_norm": 4.337370872497559,
      "learning_rate": 3.835298038802933e-05,
      "loss": 0.5736,
      "step": 19613
    },
    {
      "epoch": 6.193242816545626,
      "grad_norm": 4.8292412757873535,
      "learning_rate": 3.835210328737326e-05,
      "loss": 0.4867,
      "step": 19614
    },
    {
      "epoch": 6.1935585727818125,
      "grad_norm": 4.206820011138916,
      "learning_rate": 3.835122618671719e-05,
      "loss": 0.5443,
      "step": 19615
    },
    {
      "epoch": 6.193874329017998,
      "grad_norm": 5.378178119659424,
      "learning_rate": 3.8350349086061114e-05,
      "loss": 0.5419,
      "step": 19616
    },
    {
      "epoch": 6.194190085254184,
      "grad_norm": 4.713527202606201,
      "learning_rate": 3.8349471985405046e-05,
      "loss": 0.5135,
      "step": 19617
    },
    {
      "epoch": 6.194505841490369,
      "grad_norm": 4.948395729064941,
      "learning_rate": 3.834859488474897e-05,
      "loss": 0.5373,
      "step": 19618
    },
    {
      "epoch": 6.1948215977265555,
      "grad_norm": 5.342155933380127,
      "learning_rate": 3.834771778409291e-05,
      "loss": 0.5277,
      "step": 19619
    },
    {
      "epoch": 6.195137353962741,
      "grad_norm": 4.878630638122559,
      "learning_rate": 3.8346840683436834e-05,
      "loss": 0.4983,
      "step": 19620
    },
    {
      "epoch": 6.195453110198926,
      "grad_norm": 3.71734356880188,
      "learning_rate": 3.8345963582780766e-05,
      "loss": 0.4354,
      "step": 19621
    },
    {
      "epoch": 6.195768866435112,
      "grad_norm": 5.201793670654297,
      "learning_rate": 3.834508648212469e-05,
      "loss": 0.4892,
      "step": 19622
    },
    {
      "epoch": 6.196084622671298,
      "grad_norm": 5.122434139251709,
      "learning_rate": 3.834420938146862e-05,
      "loss": 0.4747,
      "step": 19623
    },
    {
      "epoch": 6.196400378907484,
      "grad_norm": 5.35713529586792,
      "learning_rate": 3.834333228081255e-05,
      "loss": 0.4872,
      "step": 19624
    },
    {
      "epoch": 6.196716135143669,
      "grad_norm": 4.521585941314697,
      "learning_rate": 3.834245518015648e-05,
      "loss": 0.5805,
      "step": 19625
    },
    {
      "epoch": 6.197031891379854,
      "grad_norm": 5.6529388427734375,
      "learning_rate": 3.8341578079500404e-05,
      "loss": 0.5045,
      "step": 19626
    },
    {
      "epoch": 6.1973476476160405,
      "grad_norm": 5.208528995513916,
      "learning_rate": 3.8340700978844335e-05,
      "loss": 0.5495,
      "step": 19627
    },
    {
      "epoch": 6.197663403852226,
      "grad_norm": 5.29930305480957,
      "learning_rate": 3.833982387818826e-05,
      "loss": 0.5994,
      "step": 19628
    },
    {
      "epoch": 6.197979160088412,
      "grad_norm": 7.678797245025635,
      "learning_rate": 3.833894677753219e-05,
      "loss": 0.5008,
      "step": 19629
    },
    {
      "epoch": 6.198294916324597,
      "grad_norm": 5.409761428833008,
      "learning_rate": 3.8338069676876124e-05,
      "loss": 0.629,
      "step": 19630
    },
    {
      "epoch": 6.1986106725607835,
      "grad_norm": 6.284402847290039,
      "learning_rate": 3.833719257622005e-05,
      "loss": 0.5893,
      "step": 19631
    },
    {
      "epoch": 6.198926428796969,
      "grad_norm": 4.759988307952881,
      "learning_rate": 3.833631547556398e-05,
      "loss": 0.499,
      "step": 19632
    },
    {
      "epoch": 6.199242185033154,
      "grad_norm": 5.50101900100708,
      "learning_rate": 3.8335438374907905e-05,
      "loss": 0.603,
      "step": 19633
    },
    {
      "epoch": 6.19955794126934,
      "grad_norm": 4.501059532165527,
      "learning_rate": 3.833456127425184e-05,
      "loss": 0.5489,
      "step": 19634
    },
    {
      "epoch": 6.1998736975055255,
      "grad_norm": 4.491910457611084,
      "learning_rate": 3.833368417359576e-05,
      "loss": 0.5121,
      "step": 19635
    },
    {
      "epoch": 6.200189453741712,
      "grad_norm": 5.441285133361816,
      "learning_rate": 3.8332807072939694e-05,
      "loss": 0.6937,
      "step": 19636
    },
    {
      "epoch": 6.200505209977897,
      "grad_norm": 5.590004920959473,
      "learning_rate": 3.833192997228362e-05,
      "loss": 0.5844,
      "step": 19637
    },
    {
      "epoch": 6.200820966214083,
      "grad_norm": 5.175129413604736,
      "learning_rate": 3.833105287162755e-05,
      "loss": 0.5003,
      "step": 19638
    },
    {
      "epoch": 6.2011367224502685,
      "grad_norm": 4.504563331604004,
      "learning_rate": 3.8330175770971475e-05,
      "loss": 0.4675,
      "step": 19639
    },
    {
      "epoch": 6.201452478686454,
      "grad_norm": 4.508379936218262,
      "learning_rate": 3.832929867031541e-05,
      "loss": 0.4776,
      "step": 19640
    },
    {
      "epoch": 6.20176823492264,
      "grad_norm": 4.674225807189941,
      "learning_rate": 3.832842156965933e-05,
      "loss": 0.606,
      "step": 19641
    },
    {
      "epoch": 6.202083991158825,
      "grad_norm": 5.170284271240234,
      "learning_rate": 3.832754446900326e-05,
      "loss": 0.4889,
      "step": 19642
    },
    {
      "epoch": 6.202399747395011,
      "grad_norm": 5.249948978424072,
      "learning_rate": 3.832666736834719e-05,
      "loss": 0.5927,
      "step": 19643
    },
    {
      "epoch": 6.202715503631197,
      "grad_norm": 4.602605819702148,
      "learning_rate": 3.832579026769112e-05,
      "loss": 0.4297,
      "step": 19644
    },
    {
      "epoch": 6.203031259867382,
      "grad_norm": 5.1992902755737305,
      "learning_rate": 3.8324913167035045e-05,
      "loss": 0.5731,
      "step": 19645
    },
    {
      "epoch": 6.203347016103568,
      "grad_norm": 4.163872718811035,
      "learning_rate": 3.832403606637898e-05,
      "loss": 0.5808,
      "step": 19646
    },
    {
      "epoch": 6.2036627723397535,
      "grad_norm": 3.667217969894409,
      "learning_rate": 3.832315896572291e-05,
      "loss": 0.4811,
      "step": 19647
    },
    {
      "epoch": 6.20397852857594,
      "grad_norm": 5.304830551147461,
      "learning_rate": 3.832228186506684e-05,
      "loss": 0.5853,
      "step": 19648
    },
    {
      "epoch": 6.204294284812125,
      "grad_norm": 6.529557228088379,
      "learning_rate": 3.8321404764410765e-05,
      "loss": 0.5521,
      "step": 19649
    },
    {
      "epoch": 6.20461004104831,
      "grad_norm": 5.254039764404297,
      "learning_rate": 3.8320527663754696e-05,
      "loss": 0.5522,
      "step": 19650
    },
    {
      "epoch": 6.2049257972844964,
      "grad_norm": 5.667732238769531,
      "learning_rate": 3.831965056309862e-05,
      "loss": 0.5654,
      "step": 19651
    },
    {
      "epoch": 6.205241553520682,
      "grad_norm": 5.821609973907471,
      "learning_rate": 3.831877346244255e-05,
      "loss": 0.4466,
      "step": 19652
    },
    {
      "epoch": 6.205557309756868,
      "grad_norm": 4.807068824768066,
      "learning_rate": 3.8317896361786485e-05,
      "loss": 0.4908,
      "step": 19653
    },
    {
      "epoch": 6.205873065993053,
      "grad_norm": 5.629577159881592,
      "learning_rate": 3.831701926113041e-05,
      "loss": 0.539,
      "step": 19654
    },
    {
      "epoch": 6.206188822229239,
      "grad_norm": 6.771609306335449,
      "learning_rate": 3.831614216047434e-05,
      "loss": 0.5482,
      "step": 19655
    },
    {
      "epoch": 6.206504578465425,
      "grad_norm": 4.904804229736328,
      "learning_rate": 3.8315265059818266e-05,
      "loss": 0.5879,
      "step": 19656
    },
    {
      "epoch": 6.20682033470161,
      "grad_norm": 4.602167129516602,
      "learning_rate": 3.83143879591622e-05,
      "loss": 0.5417,
      "step": 19657
    },
    {
      "epoch": 6.207136090937796,
      "grad_norm": 3.8592405319213867,
      "learning_rate": 3.831351085850612e-05,
      "loss": 0.4737,
      "step": 19658
    },
    {
      "epoch": 6.2074518471739815,
      "grad_norm": 4.675307750701904,
      "learning_rate": 3.8312633757850054e-05,
      "loss": 0.4365,
      "step": 19659
    },
    {
      "epoch": 6.207767603410168,
      "grad_norm": 5.450925350189209,
      "learning_rate": 3.831175665719398e-05,
      "loss": 0.5603,
      "step": 19660
    },
    {
      "epoch": 6.208083359646353,
      "grad_norm": 4.576780796051025,
      "learning_rate": 3.831087955653791e-05,
      "loss": 0.5663,
      "step": 19661
    },
    {
      "epoch": 6.208399115882539,
      "grad_norm": 5.003049373626709,
      "learning_rate": 3.8310002455881836e-05,
      "loss": 0.6094,
      "step": 19662
    },
    {
      "epoch": 6.208714872118724,
      "grad_norm": 5.100141525268555,
      "learning_rate": 3.830912535522577e-05,
      "loss": 0.5406,
      "step": 19663
    },
    {
      "epoch": 6.20903062835491,
      "grad_norm": 4.736326694488525,
      "learning_rate": 3.830824825456969e-05,
      "loss": 0.5641,
      "step": 19664
    },
    {
      "epoch": 6.209346384591096,
      "grad_norm": 4.277807712554932,
      "learning_rate": 3.8307371153913624e-05,
      "loss": 0.5027,
      "step": 19665
    },
    {
      "epoch": 6.209662140827281,
      "grad_norm": 4.845877170562744,
      "learning_rate": 3.830649405325755e-05,
      "loss": 0.6377,
      "step": 19666
    },
    {
      "epoch": 6.209977897063467,
      "grad_norm": 5.45919132232666,
      "learning_rate": 3.830561695260148e-05,
      "loss": 0.6179,
      "step": 19667
    },
    {
      "epoch": 6.210293653299653,
      "grad_norm": 5.231159210205078,
      "learning_rate": 3.830473985194541e-05,
      "loss": 0.6009,
      "step": 19668
    },
    {
      "epoch": 6.210609409535838,
      "grad_norm": 5.54968786239624,
      "learning_rate": 3.830386275128934e-05,
      "loss": 0.6598,
      "step": 19669
    },
    {
      "epoch": 6.210925165772024,
      "grad_norm": 5.566149711608887,
      "learning_rate": 3.830298565063327e-05,
      "loss": 0.6248,
      "step": 19670
    },
    {
      "epoch": 6.211240922008209,
      "grad_norm": 5.3238325119018555,
      "learning_rate": 3.8302108549977194e-05,
      "loss": 0.6121,
      "step": 19671
    },
    {
      "epoch": 6.211556678244396,
      "grad_norm": 4.7946085929870605,
      "learning_rate": 3.8301231449321126e-05,
      "loss": 0.5589,
      "step": 19672
    },
    {
      "epoch": 6.211872434480581,
      "grad_norm": 4.896730422973633,
      "learning_rate": 3.830035434866506e-05,
      "loss": 0.4912,
      "step": 19673
    },
    {
      "epoch": 6.212188190716766,
      "grad_norm": 4.408274173736572,
      "learning_rate": 3.829947724800899e-05,
      "loss": 0.5844,
      "step": 19674
    },
    {
      "epoch": 6.212503946952952,
      "grad_norm": 4.812210559844971,
      "learning_rate": 3.8298600147352914e-05,
      "loss": 0.5602,
      "step": 19675
    },
    {
      "epoch": 6.212819703189138,
      "grad_norm": 5.582644462585449,
      "learning_rate": 3.8297723046696846e-05,
      "loss": 0.6414,
      "step": 19676
    },
    {
      "epoch": 6.213135459425324,
      "grad_norm": 4.428800106048584,
      "learning_rate": 3.829684594604077e-05,
      "loss": 0.5333,
      "step": 19677
    },
    {
      "epoch": 6.213451215661509,
      "grad_norm": 5.126800537109375,
      "learning_rate": 3.82959688453847e-05,
      "loss": 0.5882,
      "step": 19678
    },
    {
      "epoch": 6.213766971897695,
      "grad_norm": 4.439751148223877,
      "learning_rate": 3.829509174472863e-05,
      "loss": 0.5054,
      "step": 19679
    },
    {
      "epoch": 6.214082728133881,
      "grad_norm": 5.686029434204102,
      "learning_rate": 3.829421464407256e-05,
      "loss": 0.6654,
      "step": 19680
    },
    {
      "epoch": 6.214398484370066,
      "grad_norm": 3.6479053497314453,
      "learning_rate": 3.8293337543416484e-05,
      "loss": 0.4577,
      "step": 19681
    },
    {
      "epoch": 6.214714240606252,
      "grad_norm": 5.221221446990967,
      "learning_rate": 3.8292460442760415e-05,
      "loss": 0.576,
      "step": 19682
    },
    {
      "epoch": 6.215029996842437,
      "grad_norm": 4.892724514007568,
      "learning_rate": 3.829158334210434e-05,
      "loss": 0.5454,
      "step": 19683
    },
    {
      "epoch": 6.215345753078624,
      "grad_norm": 5.315215587615967,
      "learning_rate": 3.829070624144827e-05,
      "loss": 0.5572,
      "step": 19684
    },
    {
      "epoch": 6.215661509314809,
      "grad_norm": 5.029303073883057,
      "learning_rate": 3.82898291407922e-05,
      "loss": 0.5195,
      "step": 19685
    },
    {
      "epoch": 6.215977265550995,
      "grad_norm": 4.150119781494141,
      "learning_rate": 3.828895204013613e-05,
      "loss": 0.4946,
      "step": 19686
    },
    {
      "epoch": 6.21629302178718,
      "grad_norm": 4.6147990226745605,
      "learning_rate": 3.8288074939480053e-05,
      "loss": 0.578,
      "step": 19687
    },
    {
      "epoch": 6.216608778023366,
      "grad_norm": 4.3072357177734375,
      "learning_rate": 3.8287197838823985e-05,
      "loss": 0.5385,
      "step": 19688
    },
    {
      "epoch": 6.216924534259552,
      "grad_norm": 4.415019989013672,
      "learning_rate": 3.828632073816792e-05,
      "loss": 0.5201,
      "step": 19689
    },
    {
      "epoch": 6.217240290495737,
      "grad_norm": 5.4477009773254395,
      "learning_rate": 3.828544363751184e-05,
      "loss": 0.6018,
      "step": 19690
    },
    {
      "epoch": 6.217556046731923,
      "grad_norm": 5.366652488708496,
      "learning_rate": 3.8284566536855773e-05,
      "loss": 0.4657,
      "step": 19691
    },
    {
      "epoch": 6.217871802968109,
      "grad_norm": 4.029288291931152,
      "learning_rate": 3.82836894361997e-05,
      "loss": 0.4241,
      "step": 19692
    },
    {
      "epoch": 6.218187559204294,
      "grad_norm": 4.819857597351074,
      "learning_rate": 3.828281233554363e-05,
      "loss": 0.6122,
      "step": 19693
    },
    {
      "epoch": 6.21850331544048,
      "grad_norm": 5.242976188659668,
      "learning_rate": 3.8281935234887555e-05,
      "loss": 0.563,
      "step": 19694
    },
    {
      "epoch": 6.218819071676665,
      "grad_norm": 4.608606815338135,
      "learning_rate": 3.8281058134231487e-05,
      "loss": 0.534,
      "step": 19695
    },
    {
      "epoch": 6.2191348279128515,
      "grad_norm": 5.473269462585449,
      "learning_rate": 3.828018103357541e-05,
      "loss": 0.5386,
      "step": 19696
    },
    {
      "epoch": 6.219450584149037,
      "grad_norm": 4.593085765838623,
      "learning_rate": 3.827930393291934e-05,
      "loss": 0.5532,
      "step": 19697
    },
    {
      "epoch": 6.219766340385223,
      "grad_norm": 4.6271796226501465,
      "learning_rate": 3.827842683226327e-05,
      "loss": 0.4897,
      "step": 19698
    },
    {
      "epoch": 6.220082096621408,
      "grad_norm": 4.254361629486084,
      "learning_rate": 3.82775497316072e-05,
      "loss": 0.4573,
      "step": 19699
    },
    {
      "epoch": 6.220397852857594,
      "grad_norm": 5.723285675048828,
      "learning_rate": 3.827667263095113e-05,
      "loss": 0.5048,
      "step": 19700
    },
    {
      "epoch": 6.22071360909378,
      "grad_norm": 3.584359645843506,
      "learning_rate": 3.827579553029506e-05,
      "loss": 0.4989,
      "step": 19701
    },
    {
      "epoch": 6.221029365329965,
      "grad_norm": 3.9319939613342285,
      "learning_rate": 3.827491842963899e-05,
      "loss": 0.4966,
      "step": 19702
    },
    {
      "epoch": 6.221345121566151,
      "grad_norm": 4.7461676597595215,
      "learning_rate": 3.827404132898292e-05,
      "loss": 0.499,
      "step": 19703
    },
    {
      "epoch": 6.221660877802337,
      "grad_norm": 4.563778400421143,
      "learning_rate": 3.8273164228326845e-05,
      "loss": 0.4471,
      "step": 19704
    },
    {
      "epoch": 6.221976634038522,
      "grad_norm": 4.393778324127197,
      "learning_rate": 3.8272287127670776e-05,
      "loss": 0.495,
      "step": 19705
    },
    {
      "epoch": 6.222292390274708,
      "grad_norm": 4.519150257110596,
      "learning_rate": 3.82714100270147e-05,
      "loss": 0.4115,
      "step": 19706
    },
    {
      "epoch": 6.222608146510893,
      "grad_norm": 5.357801914215088,
      "learning_rate": 3.827053292635863e-05,
      "loss": 0.5327,
      "step": 19707
    },
    {
      "epoch": 6.2229239027470795,
      "grad_norm": 5.787901878356934,
      "learning_rate": 3.826965582570256e-05,
      "loss": 0.5714,
      "step": 19708
    },
    {
      "epoch": 6.223239658983265,
      "grad_norm": 5.327409267425537,
      "learning_rate": 3.826877872504649e-05,
      "loss": 0.5647,
      "step": 19709
    },
    {
      "epoch": 6.223555415219451,
      "grad_norm": 5.089325428009033,
      "learning_rate": 3.8267901624390414e-05,
      "loss": 0.5474,
      "step": 19710
    },
    {
      "epoch": 6.223871171455636,
      "grad_norm": 4.906105041503906,
      "learning_rate": 3.8267024523734346e-05,
      "loss": 0.6006,
      "step": 19711
    },
    {
      "epoch": 6.224186927691822,
      "grad_norm": 4.806793212890625,
      "learning_rate": 3.826614742307828e-05,
      "loss": 0.594,
      "step": 19712
    },
    {
      "epoch": 6.224502683928008,
      "grad_norm": 5.356766223907471,
      "learning_rate": 3.82652703224222e-05,
      "loss": 0.5019,
      "step": 19713
    },
    {
      "epoch": 6.224818440164193,
      "grad_norm": 4.368383407592773,
      "learning_rate": 3.8264393221766134e-05,
      "loss": 0.605,
      "step": 19714
    },
    {
      "epoch": 6.225134196400379,
      "grad_norm": 4.734030723571777,
      "learning_rate": 3.826351612111006e-05,
      "loss": 0.5128,
      "step": 19715
    },
    {
      "epoch": 6.2254499526365645,
      "grad_norm": 6.283407211303711,
      "learning_rate": 3.826263902045399e-05,
      "loss": 0.5657,
      "step": 19716
    },
    {
      "epoch": 6.22576570887275,
      "grad_norm": 5.134721279144287,
      "learning_rate": 3.8261761919797916e-05,
      "loss": 0.6024,
      "step": 19717
    },
    {
      "epoch": 6.226081465108936,
      "grad_norm": 7.429078102111816,
      "learning_rate": 3.826088481914185e-05,
      "loss": 0.5146,
      "step": 19718
    },
    {
      "epoch": 6.226397221345121,
      "grad_norm": 3.8833324909210205,
      "learning_rate": 3.826000771848577e-05,
      "loss": 0.5097,
      "step": 19719
    },
    {
      "epoch": 6.2267129775813075,
      "grad_norm": 4.755387783050537,
      "learning_rate": 3.8259130617829704e-05,
      "loss": 0.6021,
      "step": 19720
    },
    {
      "epoch": 6.227028733817493,
      "grad_norm": 5.025084495544434,
      "learning_rate": 3.825825351717363e-05,
      "loss": 0.5167,
      "step": 19721
    },
    {
      "epoch": 6.227344490053679,
      "grad_norm": 5.309596061706543,
      "learning_rate": 3.825737641651756e-05,
      "loss": 0.544,
      "step": 19722
    },
    {
      "epoch": 6.227660246289864,
      "grad_norm": 4.660036563873291,
      "learning_rate": 3.8256499315861486e-05,
      "loss": 0.5635,
      "step": 19723
    },
    {
      "epoch": 6.2279760025260495,
      "grad_norm": 5.254148483276367,
      "learning_rate": 3.825562221520542e-05,
      "loss": 0.6342,
      "step": 19724
    },
    {
      "epoch": 6.228291758762236,
      "grad_norm": 5.021556377410889,
      "learning_rate": 3.825474511454934e-05,
      "loss": 0.6105,
      "step": 19725
    },
    {
      "epoch": 6.228607514998421,
      "grad_norm": 4.106010913848877,
      "learning_rate": 3.8253868013893274e-05,
      "loss": 0.5783,
      "step": 19726
    },
    {
      "epoch": 6.228923271234607,
      "grad_norm": 4.538952827453613,
      "learning_rate": 3.8252990913237206e-05,
      "loss": 0.5195,
      "step": 19727
    },
    {
      "epoch": 6.2292390274707925,
      "grad_norm": 4.291345596313477,
      "learning_rate": 3.825211381258114e-05,
      "loss": 0.5995,
      "step": 19728
    },
    {
      "epoch": 6.229554783706978,
      "grad_norm": 4.019443511962891,
      "learning_rate": 3.825123671192506e-05,
      "loss": 0.4634,
      "step": 19729
    },
    {
      "epoch": 6.229870539943164,
      "grad_norm": 4.344766616821289,
      "learning_rate": 3.8250359611268994e-05,
      "loss": 0.5119,
      "step": 19730
    },
    {
      "epoch": 6.230186296179349,
      "grad_norm": 5.256058216094971,
      "learning_rate": 3.824948251061292e-05,
      "loss": 0.546,
      "step": 19731
    },
    {
      "epoch": 6.2305020524155355,
      "grad_norm": 5.068062782287598,
      "learning_rate": 3.824860540995685e-05,
      "loss": 0.4972,
      "step": 19732
    },
    {
      "epoch": 6.230817808651721,
      "grad_norm": 4.992603778839111,
      "learning_rate": 3.824772830930078e-05,
      "loss": 0.4995,
      "step": 19733
    },
    {
      "epoch": 6.231133564887907,
      "grad_norm": 4.681967735290527,
      "learning_rate": 3.824685120864471e-05,
      "loss": 0.5081,
      "step": 19734
    },
    {
      "epoch": 6.231449321124092,
      "grad_norm": 4.984151840209961,
      "learning_rate": 3.824597410798864e-05,
      "loss": 0.5456,
      "step": 19735
    },
    {
      "epoch": 6.2317650773602775,
      "grad_norm": 4.0732293128967285,
      "learning_rate": 3.8245097007332564e-05,
      "loss": 0.465,
      "step": 19736
    },
    {
      "epoch": 6.232080833596464,
      "grad_norm": 4.519439220428467,
      "learning_rate": 3.8244219906676495e-05,
      "loss": 0.4691,
      "step": 19737
    },
    {
      "epoch": 6.232396589832649,
      "grad_norm": 5.266888618469238,
      "learning_rate": 3.824334280602042e-05,
      "loss": 0.5658,
      "step": 19738
    },
    {
      "epoch": 6.232712346068835,
      "grad_norm": 4.879373550415039,
      "learning_rate": 3.824246570536435e-05,
      "loss": 0.6225,
      "step": 19739
    },
    {
      "epoch": 6.2330281023050205,
      "grad_norm": 3.5862503051757812,
      "learning_rate": 3.824158860470828e-05,
      "loss": 0.5096,
      "step": 19740
    },
    {
      "epoch": 6.233343858541206,
      "grad_norm": 4.555661678314209,
      "learning_rate": 3.824071150405221e-05,
      "loss": 0.5632,
      "step": 19741
    },
    {
      "epoch": 6.233659614777392,
      "grad_norm": 4.919483661651611,
      "learning_rate": 3.823983440339613e-05,
      "loss": 0.5323,
      "step": 19742
    },
    {
      "epoch": 6.233975371013577,
      "grad_norm": 4.930375576019287,
      "learning_rate": 3.8238957302740065e-05,
      "loss": 0.5404,
      "step": 19743
    },
    {
      "epoch": 6.234291127249763,
      "grad_norm": 5.460573196411133,
      "learning_rate": 3.823808020208399e-05,
      "loss": 0.5667,
      "step": 19744
    },
    {
      "epoch": 6.234606883485949,
      "grad_norm": 5.49251651763916,
      "learning_rate": 3.823720310142792e-05,
      "loss": 0.4949,
      "step": 19745
    },
    {
      "epoch": 6.234922639722135,
      "grad_norm": 6.055002212524414,
      "learning_rate": 3.8236326000771847e-05,
      "loss": 0.5695,
      "step": 19746
    },
    {
      "epoch": 6.23523839595832,
      "grad_norm": 5.766036033630371,
      "learning_rate": 3.823544890011578e-05,
      "loss": 0.5673,
      "step": 19747
    },
    {
      "epoch": 6.2355541521945055,
      "grad_norm": 5.107959747314453,
      "learning_rate": 3.82345717994597e-05,
      "loss": 0.6616,
      "step": 19748
    },
    {
      "epoch": 6.235869908430692,
      "grad_norm": 7.680520534515381,
      "learning_rate": 3.8233694698803635e-05,
      "loss": 0.5668,
      "step": 19749
    },
    {
      "epoch": 6.236185664666877,
      "grad_norm": 4.370698928833008,
      "learning_rate": 3.8232817598147566e-05,
      "loss": 0.5146,
      "step": 19750
    },
    {
      "epoch": 6.236501420903063,
      "grad_norm": 4.304549217224121,
      "learning_rate": 3.823194049749149e-05,
      "loss": 0.4967,
      "step": 19751
    },
    {
      "epoch": 6.236817177139248,
      "grad_norm": 4.803071975708008,
      "learning_rate": 3.823106339683542e-05,
      "loss": 0.4569,
      "step": 19752
    },
    {
      "epoch": 6.237132933375435,
      "grad_norm": 3.496619701385498,
      "learning_rate": 3.823018629617935e-05,
      "loss": 0.4616,
      "step": 19753
    },
    {
      "epoch": 6.23744868961162,
      "grad_norm": 4.8044281005859375,
      "learning_rate": 3.822930919552328e-05,
      "loss": 0.5238,
      "step": 19754
    },
    {
      "epoch": 6.237764445847805,
      "grad_norm": 4.609419822692871,
      "learning_rate": 3.822843209486721e-05,
      "loss": 0.5881,
      "step": 19755
    },
    {
      "epoch": 6.238080202083991,
      "grad_norm": 5.167274475097656,
      "learning_rate": 3.822755499421114e-05,
      "loss": 0.5675,
      "step": 19756
    },
    {
      "epoch": 6.238395958320177,
      "grad_norm": 4.047726154327393,
      "learning_rate": 3.822667789355507e-05,
      "loss": 0.5006,
      "step": 19757
    },
    {
      "epoch": 6.238711714556363,
      "grad_norm": 4.172407150268555,
      "learning_rate": 3.8225800792899e-05,
      "loss": 0.481,
      "step": 19758
    },
    {
      "epoch": 6.239027470792548,
      "grad_norm": 5.245381832122803,
      "learning_rate": 3.8224923692242925e-05,
      "loss": 0.5326,
      "step": 19759
    },
    {
      "epoch": 6.2393432270287335,
      "grad_norm": 4.445018768310547,
      "learning_rate": 3.8224046591586856e-05,
      "loss": 0.5083,
      "step": 19760
    },
    {
      "epoch": 6.23965898326492,
      "grad_norm": 4.669900417327881,
      "learning_rate": 3.822316949093078e-05,
      "loss": 0.5695,
      "step": 19761
    },
    {
      "epoch": 6.239974739501105,
      "grad_norm": 5.304919242858887,
      "learning_rate": 3.822229239027471e-05,
      "loss": 0.6337,
      "step": 19762
    },
    {
      "epoch": 6.240290495737291,
      "grad_norm": 6.287749767303467,
      "learning_rate": 3.822141528961864e-05,
      "loss": 0.5452,
      "step": 19763
    },
    {
      "epoch": 6.240606251973476,
      "grad_norm": 4.740357875823975,
      "learning_rate": 3.822053818896257e-05,
      "loss": 0.5554,
      "step": 19764
    },
    {
      "epoch": 6.240922008209662,
      "grad_norm": 4.097965717315674,
      "learning_rate": 3.8219661088306494e-05,
      "loss": 0.4581,
      "step": 19765
    },
    {
      "epoch": 6.241237764445848,
      "grad_norm": 4.545534133911133,
      "learning_rate": 3.8218783987650426e-05,
      "loss": 0.4943,
      "step": 19766
    },
    {
      "epoch": 6.241553520682033,
      "grad_norm": 4.583906173706055,
      "learning_rate": 3.821790688699435e-05,
      "loss": 0.4927,
      "step": 19767
    },
    {
      "epoch": 6.241869276918219,
      "grad_norm": 4.203279972076416,
      "learning_rate": 3.821702978633828e-05,
      "loss": 0.5073,
      "step": 19768
    },
    {
      "epoch": 6.242185033154405,
      "grad_norm": 5.357064723968506,
      "learning_rate": 3.821615268568221e-05,
      "loss": 0.581,
      "step": 19769
    },
    {
      "epoch": 6.242500789390591,
      "grad_norm": 4.939150810241699,
      "learning_rate": 3.821527558502614e-05,
      "loss": 0.5108,
      "step": 19770
    },
    {
      "epoch": 6.242816545626776,
      "grad_norm": 7.350887298583984,
      "learning_rate": 3.821439848437007e-05,
      "loss": 0.6164,
      "step": 19771
    },
    {
      "epoch": 6.243132301862961,
      "grad_norm": 3.9269189834594727,
      "learning_rate": 3.8213521383713996e-05,
      "loss": 0.5116,
      "step": 19772
    },
    {
      "epoch": 6.243448058099148,
      "grad_norm": 5.8052473068237305,
      "learning_rate": 3.821264428305793e-05,
      "loss": 0.555,
      "step": 19773
    },
    {
      "epoch": 6.243763814335333,
      "grad_norm": 5.088978290557861,
      "learning_rate": 3.821176718240185e-05,
      "loss": 0.5996,
      "step": 19774
    },
    {
      "epoch": 6.244079570571519,
      "grad_norm": 6.073317527770996,
      "learning_rate": 3.8210890081745784e-05,
      "loss": 0.5014,
      "step": 19775
    },
    {
      "epoch": 6.244395326807704,
      "grad_norm": 4.973733425140381,
      "learning_rate": 3.821001298108971e-05,
      "loss": 0.5429,
      "step": 19776
    },
    {
      "epoch": 6.2447110830438906,
      "grad_norm": 4.503368854522705,
      "learning_rate": 3.820913588043364e-05,
      "loss": 0.5199,
      "step": 19777
    },
    {
      "epoch": 6.245026839280076,
      "grad_norm": 6.170614719390869,
      "learning_rate": 3.8208258779777565e-05,
      "loss": 0.5892,
      "step": 19778
    },
    {
      "epoch": 6.245342595516261,
      "grad_norm": 4.1282806396484375,
      "learning_rate": 3.82073816791215e-05,
      "loss": 0.5366,
      "step": 19779
    },
    {
      "epoch": 6.245658351752447,
      "grad_norm": 8.027261734008789,
      "learning_rate": 3.820650457846542e-05,
      "loss": 0.5699,
      "step": 19780
    },
    {
      "epoch": 6.245974107988633,
      "grad_norm": 4.194045543670654,
      "learning_rate": 3.8205627477809354e-05,
      "loss": 0.5338,
      "step": 19781
    },
    {
      "epoch": 6.246289864224819,
      "grad_norm": 4.901054382324219,
      "learning_rate": 3.8204750377153285e-05,
      "loss": 0.5764,
      "step": 19782
    },
    {
      "epoch": 6.246605620461004,
      "grad_norm": 4.391032695770264,
      "learning_rate": 3.820387327649722e-05,
      "loss": 0.4552,
      "step": 19783
    },
    {
      "epoch": 6.246921376697189,
      "grad_norm": 4.1248908042907715,
      "learning_rate": 3.820299617584114e-05,
      "loss": 0.5416,
      "step": 19784
    },
    {
      "epoch": 6.247237132933376,
      "grad_norm": 4.867813587188721,
      "learning_rate": 3.8202119075185074e-05,
      "loss": 0.5448,
      "step": 19785
    },
    {
      "epoch": 6.247552889169561,
      "grad_norm": 4.592299938201904,
      "learning_rate": 3.8201241974529e-05,
      "loss": 0.4561,
      "step": 19786
    },
    {
      "epoch": 6.247868645405747,
      "grad_norm": 5.922141075134277,
      "learning_rate": 3.820036487387293e-05,
      "loss": 0.5943,
      "step": 19787
    },
    {
      "epoch": 6.248184401641932,
      "grad_norm": 4.766730308532715,
      "learning_rate": 3.8199487773216855e-05,
      "loss": 0.5279,
      "step": 19788
    },
    {
      "epoch": 6.2485001578781185,
      "grad_norm": 4.150850296020508,
      "learning_rate": 3.819861067256079e-05,
      "loss": 0.5509,
      "step": 19789
    },
    {
      "epoch": 6.248815914114304,
      "grad_norm": 5.296535968780518,
      "learning_rate": 3.819773357190471e-05,
      "loss": 0.5168,
      "step": 19790
    },
    {
      "epoch": 6.249131670350489,
      "grad_norm": 6.370417594909668,
      "learning_rate": 3.8196856471248643e-05,
      "loss": 0.4989,
      "step": 19791
    },
    {
      "epoch": 6.249447426586675,
      "grad_norm": 4.082113742828369,
      "learning_rate": 3.819597937059257e-05,
      "loss": 0.3879,
      "step": 19792
    },
    {
      "epoch": 6.249763182822861,
      "grad_norm": 5.010664463043213,
      "learning_rate": 3.81951022699365e-05,
      "loss": 0.6641,
      "step": 19793
    },
    {
      "epoch": 6.250078939059047,
      "grad_norm": 6.24601411819458,
      "learning_rate": 3.819422516928043e-05,
      "loss": 0.5442,
      "step": 19794
    },
    {
      "epoch": 6.250394695295232,
      "grad_norm": 4.73787260055542,
      "learning_rate": 3.819334806862436e-05,
      "loss": 0.4616,
      "step": 19795
    },
    {
      "epoch": 6.250710451531417,
      "grad_norm": 5.206791400909424,
      "learning_rate": 3.819247096796829e-05,
      "loss": 0.5542,
      "step": 19796
    },
    {
      "epoch": 6.2510262077676035,
      "grad_norm": 7.436471462249756,
      "learning_rate": 3.819159386731221e-05,
      "loss": 0.5318,
      "step": 19797
    },
    {
      "epoch": 6.251341964003789,
      "grad_norm": 5.541157245635986,
      "learning_rate": 3.8190716766656145e-05,
      "loss": 0.5366,
      "step": 19798
    },
    {
      "epoch": 6.251657720239975,
      "grad_norm": 5.814428806304932,
      "learning_rate": 3.818983966600007e-05,
      "loss": 0.6657,
      "step": 19799
    },
    {
      "epoch": 6.25197347647616,
      "grad_norm": 5.12393045425415,
      "learning_rate": 3.8188962565344e-05,
      "loss": 0.5612,
      "step": 19800
    },
    {
      "epoch": 6.25197347647616,
      "eval_loss": 0.8808771371841431,
      "eval_runtime": 4.547,
      "eval_samples_per_second": 547.62,
      "eval_steps_per_second": 4.618,
      "step": 19800
    },
    {
      "epoch": 6.2522892327123465,
      "grad_norm": 4.90953254699707,
      "learning_rate": 3.8188085464687926e-05,
      "loss": 0.5242,
      "step": 19801
    },
    {
      "epoch": 6.252604988948532,
      "grad_norm": 4.1778669357299805,
      "learning_rate": 3.818720836403186e-05,
      "loss": 0.4861,
      "step": 19802
    },
    {
      "epoch": 6.252920745184717,
      "grad_norm": 4.5341267585754395,
      "learning_rate": 3.818633126337578e-05,
      "loss": 0.5015,
      "step": 19803
    },
    {
      "epoch": 6.253236501420903,
      "grad_norm": 3.835299253463745,
      "learning_rate": 3.8185454162719715e-05,
      "loss": 0.4125,
      "step": 19804
    },
    {
      "epoch": 6.2535522576570886,
      "grad_norm": 6.831423759460449,
      "learning_rate": 3.818457706206364e-05,
      "loss": 0.5811,
      "step": 19805
    },
    {
      "epoch": 6.253868013893275,
      "grad_norm": 5.06610631942749,
      "learning_rate": 3.818369996140757e-05,
      "loss": 0.757,
      "step": 19806
    },
    {
      "epoch": 6.25418377012946,
      "grad_norm": 5.330420970916748,
      "learning_rate": 3.8182822860751496e-05,
      "loss": 0.6168,
      "step": 19807
    },
    {
      "epoch": 6.254499526365645,
      "grad_norm": 4.977745532989502,
      "learning_rate": 3.818194576009543e-05,
      "loss": 0.537,
      "step": 19808
    },
    {
      "epoch": 6.2548152826018315,
      "grad_norm": 4.869994640350342,
      "learning_rate": 3.818106865943936e-05,
      "loss": 0.5279,
      "step": 19809
    },
    {
      "epoch": 6.255131038838017,
      "grad_norm": 4.91919469833374,
      "learning_rate": 3.818019155878329e-05,
      "loss": 0.5295,
      "step": 19810
    },
    {
      "epoch": 6.255446795074203,
      "grad_norm": 4.436304092407227,
      "learning_rate": 3.8179314458127216e-05,
      "loss": 0.5848,
      "step": 19811
    },
    {
      "epoch": 6.255762551310388,
      "grad_norm": 4.942993640899658,
      "learning_rate": 3.817843735747115e-05,
      "loss": 0.5851,
      "step": 19812
    },
    {
      "epoch": 6.2560783075465745,
      "grad_norm": 7.1727776527404785,
      "learning_rate": 3.817756025681507e-05,
      "loss": 0.5937,
      "step": 19813
    },
    {
      "epoch": 6.25639406378276,
      "grad_norm": 4.647177696228027,
      "learning_rate": 3.8176683156159004e-05,
      "loss": 0.5978,
      "step": 19814
    },
    {
      "epoch": 6.256709820018945,
      "grad_norm": 4.438706874847412,
      "learning_rate": 3.8175806055502936e-05,
      "loss": 0.5094,
      "step": 19815
    },
    {
      "epoch": 6.257025576255131,
      "grad_norm": 4.3444600105285645,
      "learning_rate": 3.817492895484686e-05,
      "loss": 0.5297,
      "step": 19816
    },
    {
      "epoch": 6.2573413324913165,
      "grad_norm": 4.61161994934082,
      "learning_rate": 3.817405185419079e-05,
      "loss": 0.5205,
      "step": 19817
    },
    {
      "epoch": 6.257657088727503,
      "grad_norm": 6.481533050537109,
      "learning_rate": 3.817317475353472e-05,
      "loss": 0.5665,
      "step": 19818
    },
    {
      "epoch": 6.257972844963688,
      "grad_norm": 4.263754844665527,
      "learning_rate": 3.817229765287865e-05,
      "loss": 0.5021,
      "step": 19819
    },
    {
      "epoch": 6.258288601199873,
      "grad_norm": 5.181369304656982,
      "learning_rate": 3.8171420552222574e-05,
      "loss": 0.6371,
      "step": 19820
    },
    {
      "epoch": 6.2586043574360595,
      "grad_norm": 4.896982669830322,
      "learning_rate": 3.8170543451566506e-05,
      "loss": 0.5264,
      "step": 19821
    },
    {
      "epoch": 6.258920113672245,
      "grad_norm": 5.1461334228515625,
      "learning_rate": 3.816966635091043e-05,
      "loss": 0.4856,
      "step": 19822
    },
    {
      "epoch": 6.259235869908431,
      "grad_norm": 5.620983600616455,
      "learning_rate": 3.816878925025436e-05,
      "loss": 0.5381,
      "step": 19823
    },
    {
      "epoch": 6.259551626144616,
      "grad_norm": 3.9852921962738037,
      "learning_rate": 3.816791214959829e-05,
      "loss": 0.5075,
      "step": 19824
    },
    {
      "epoch": 6.259867382380802,
      "grad_norm": 5.755980491638184,
      "learning_rate": 3.816703504894222e-05,
      "loss": 0.637,
      "step": 19825
    },
    {
      "epoch": 6.260183138616988,
      "grad_norm": 5.7581939697265625,
      "learning_rate": 3.8166157948286144e-05,
      "loss": 0.5357,
      "step": 19826
    },
    {
      "epoch": 6.260498894853173,
      "grad_norm": 4.326846122741699,
      "learning_rate": 3.8165280847630076e-05,
      "loss": 0.4487,
      "step": 19827
    },
    {
      "epoch": 6.260814651089359,
      "grad_norm": 4.657287120819092,
      "learning_rate": 3.8164403746974e-05,
      "loss": 0.501,
      "step": 19828
    },
    {
      "epoch": 6.2611304073255445,
      "grad_norm": 4.616907596588135,
      "learning_rate": 3.816352664631793e-05,
      "loss": 0.5681,
      "step": 19829
    },
    {
      "epoch": 6.261446163561731,
      "grad_norm": 4.57004976272583,
      "learning_rate": 3.816264954566186e-05,
      "loss": 0.5624,
      "step": 19830
    },
    {
      "epoch": 6.261761919797916,
      "grad_norm": 4.794520854949951,
      "learning_rate": 3.816177244500579e-05,
      "loss": 0.6448,
      "step": 19831
    },
    {
      "epoch": 6.262077676034101,
      "grad_norm": 5.18654727935791,
      "learning_rate": 3.816089534434972e-05,
      "loss": 0.6152,
      "step": 19832
    },
    {
      "epoch": 6.262393432270287,
      "grad_norm": 4.627779960632324,
      "learning_rate": 3.8160018243693645e-05,
      "loss": 0.6181,
      "step": 19833
    },
    {
      "epoch": 6.262709188506473,
      "grad_norm": 5.214789867401123,
      "learning_rate": 3.815914114303758e-05,
      "loss": 0.5191,
      "step": 19834
    },
    {
      "epoch": 6.263024944742659,
      "grad_norm": 4.615248203277588,
      "learning_rate": 3.81582640423815e-05,
      "loss": 0.5229,
      "step": 19835
    },
    {
      "epoch": 6.263340700978844,
      "grad_norm": 5.092837810516357,
      "learning_rate": 3.8157386941725434e-05,
      "loss": 0.6301,
      "step": 19836
    },
    {
      "epoch": 6.26365645721503,
      "grad_norm": 4.392977237701416,
      "learning_rate": 3.8156509841069365e-05,
      "loss": 0.535,
      "step": 19837
    },
    {
      "epoch": 6.263972213451216,
      "grad_norm": 4.873290538787842,
      "learning_rate": 3.81556327404133e-05,
      "loss": 0.5504,
      "step": 19838
    },
    {
      "epoch": 6.264287969687401,
      "grad_norm": 4.982989311218262,
      "learning_rate": 3.815475563975722e-05,
      "loss": 0.5549,
      "step": 19839
    },
    {
      "epoch": 6.264603725923587,
      "grad_norm": 5.017500877380371,
      "learning_rate": 3.8153878539101154e-05,
      "loss": 0.5851,
      "step": 19840
    },
    {
      "epoch": 6.2649194821597725,
      "grad_norm": 4.064390182495117,
      "learning_rate": 3.815300143844508e-05,
      "loss": 0.5905,
      "step": 19841
    },
    {
      "epoch": 6.265235238395959,
      "grad_norm": 5.275705337524414,
      "learning_rate": 3.815212433778901e-05,
      "loss": 0.6922,
      "step": 19842
    },
    {
      "epoch": 6.265550994632144,
      "grad_norm": 4.787642478942871,
      "learning_rate": 3.8151247237132935e-05,
      "loss": 0.6657,
      "step": 19843
    },
    {
      "epoch": 6.26586675086833,
      "grad_norm": 4.363554000854492,
      "learning_rate": 3.815037013647687e-05,
      "loss": 0.6257,
      "step": 19844
    },
    {
      "epoch": 6.266182507104515,
      "grad_norm": 6.448753356933594,
      "learning_rate": 3.814949303582079e-05,
      "loss": 0.549,
      "step": 19845
    },
    {
      "epoch": 6.266498263340701,
      "grad_norm": 3.9045910835266113,
      "learning_rate": 3.814861593516472e-05,
      "loss": 0.5235,
      "step": 19846
    },
    {
      "epoch": 6.266814019576887,
      "grad_norm": 5.438889026641846,
      "learning_rate": 3.814773883450865e-05,
      "loss": 0.6377,
      "step": 19847
    },
    {
      "epoch": 6.267129775813072,
      "grad_norm": 5.982069969177246,
      "learning_rate": 3.814686173385258e-05,
      "loss": 0.6086,
      "step": 19848
    },
    {
      "epoch": 6.267445532049258,
      "grad_norm": 6.372313976287842,
      "learning_rate": 3.8145984633196505e-05,
      "loss": 0.5931,
      "step": 19849
    },
    {
      "epoch": 6.267761288285444,
      "grad_norm": 4.691631317138672,
      "learning_rate": 3.8145107532540437e-05,
      "loss": 0.5942,
      "step": 19850
    },
    {
      "epoch": 6.268077044521629,
      "grad_norm": 6.13710355758667,
      "learning_rate": 3.814423043188436e-05,
      "loss": 0.4985,
      "step": 19851
    },
    {
      "epoch": 6.268392800757815,
      "grad_norm": 4.765895366668701,
      "learning_rate": 3.814335333122829e-05,
      "loss": 0.5086,
      "step": 19852
    },
    {
      "epoch": 6.268708556994,
      "grad_norm": 4.57280969619751,
      "learning_rate": 3.8142476230572225e-05,
      "loss": 0.5838,
      "step": 19853
    },
    {
      "epoch": 6.269024313230187,
      "grad_norm": 5.324800968170166,
      "learning_rate": 3.814159912991615e-05,
      "loss": 0.5379,
      "step": 19854
    },
    {
      "epoch": 6.269340069466372,
      "grad_norm": 4.736273765563965,
      "learning_rate": 3.814072202926008e-05,
      "loss": 0.5341,
      "step": 19855
    },
    {
      "epoch": 6.269655825702557,
      "grad_norm": 3.968961477279663,
      "learning_rate": 3.8139844928604006e-05,
      "loss": 0.455,
      "step": 19856
    },
    {
      "epoch": 6.269971581938743,
      "grad_norm": 5.120563507080078,
      "learning_rate": 3.813896782794794e-05,
      "loss": 0.6212,
      "step": 19857
    },
    {
      "epoch": 6.270287338174929,
      "grad_norm": 4.599905490875244,
      "learning_rate": 3.813809072729186e-05,
      "loss": 0.6419,
      "step": 19858
    },
    {
      "epoch": 6.270603094411115,
      "grad_norm": 4.366239070892334,
      "learning_rate": 3.8137213626635795e-05,
      "loss": 0.5452,
      "step": 19859
    },
    {
      "epoch": 6.2709188506473,
      "grad_norm": 5.615140438079834,
      "learning_rate": 3.813633652597972e-05,
      "loss": 0.5528,
      "step": 19860
    },
    {
      "epoch": 6.271234606883486,
      "grad_norm": 6.14358377456665,
      "learning_rate": 3.813545942532365e-05,
      "loss": 0.6745,
      "step": 19861
    },
    {
      "epoch": 6.271550363119672,
      "grad_norm": 4.597527980804443,
      "learning_rate": 3.8134582324667576e-05,
      "loss": 0.5742,
      "step": 19862
    },
    {
      "epoch": 6.271866119355857,
      "grad_norm": 6.995629787445068,
      "learning_rate": 3.813370522401151e-05,
      "loss": 0.7241,
      "step": 19863
    },
    {
      "epoch": 6.272181875592043,
      "grad_norm": 5.433715343475342,
      "learning_rate": 3.813282812335544e-05,
      "loss": 0.6689,
      "step": 19864
    },
    {
      "epoch": 6.272497631828228,
      "grad_norm": 4.0094499588012695,
      "learning_rate": 3.813195102269937e-05,
      "loss": 0.5564,
      "step": 19865
    },
    {
      "epoch": 6.272813388064415,
      "grad_norm": 4.175488471984863,
      "learning_rate": 3.8131073922043296e-05,
      "loss": 0.5681,
      "step": 19866
    },
    {
      "epoch": 6.2731291443006,
      "grad_norm": 4.360036373138428,
      "learning_rate": 3.813019682138723e-05,
      "loss": 0.4807,
      "step": 19867
    },
    {
      "epoch": 6.273444900536786,
      "grad_norm": 4.552827835083008,
      "learning_rate": 3.812931972073115e-05,
      "loss": 0.4519,
      "step": 19868
    },
    {
      "epoch": 6.273760656772971,
      "grad_norm": 3.6867287158966064,
      "learning_rate": 3.8128442620075084e-05,
      "loss": 0.4956,
      "step": 19869
    },
    {
      "epoch": 6.274076413009157,
      "grad_norm": 5.776064872741699,
      "learning_rate": 3.812756551941901e-05,
      "loss": 0.6038,
      "step": 19870
    },
    {
      "epoch": 6.274392169245343,
      "grad_norm": 3.8056142330169678,
      "learning_rate": 3.812668841876294e-05,
      "loss": 0.4909,
      "step": 19871
    },
    {
      "epoch": 6.274707925481528,
      "grad_norm": 4.735381603240967,
      "learning_rate": 3.8125811318106866e-05,
      "loss": 0.5201,
      "step": 19872
    },
    {
      "epoch": 6.275023681717714,
      "grad_norm": 5.2135443687438965,
      "learning_rate": 3.81249342174508e-05,
      "loss": 0.6126,
      "step": 19873
    },
    {
      "epoch": 6.2753394379539,
      "grad_norm": 6.645788192749023,
      "learning_rate": 3.812405711679473e-05,
      "loss": 0.5969,
      "step": 19874
    },
    {
      "epoch": 6.275655194190085,
      "grad_norm": 4.717860698699951,
      "learning_rate": 3.8123180016138654e-05,
      "loss": 0.6219,
      "step": 19875
    },
    {
      "epoch": 6.275970950426271,
      "grad_norm": 5.2511467933654785,
      "learning_rate": 3.8122302915482586e-05,
      "loss": 0.6306,
      "step": 19876
    },
    {
      "epoch": 6.276286706662456,
      "grad_norm": 4.55771017074585,
      "learning_rate": 3.812142581482651e-05,
      "loss": 0.5336,
      "step": 19877
    },
    {
      "epoch": 6.2766024628986425,
      "grad_norm": 4.747781753540039,
      "learning_rate": 3.812054871417044e-05,
      "loss": 0.5175,
      "step": 19878
    },
    {
      "epoch": 6.276918219134828,
      "grad_norm": 4.560137748718262,
      "learning_rate": 3.811967161351437e-05,
      "loss": 0.5317,
      "step": 19879
    },
    {
      "epoch": 6.277233975371013,
      "grad_norm": 5.347316265106201,
      "learning_rate": 3.81187945128583e-05,
      "loss": 0.6086,
      "step": 19880
    },
    {
      "epoch": 6.277549731607199,
      "grad_norm": 4.521608829498291,
      "learning_rate": 3.8117917412202224e-05,
      "loss": 0.5634,
      "step": 19881
    },
    {
      "epoch": 6.277865487843385,
      "grad_norm": 4.517362594604492,
      "learning_rate": 3.8117040311546156e-05,
      "loss": 0.4943,
      "step": 19882
    },
    {
      "epoch": 6.278181244079571,
      "grad_norm": 4.761142730712891,
      "learning_rate": 3.811616321089008e-05,
      "loss": 0.5827,
      "step": 19883
    },
    {
      "epoch": 6.278497000315756,
      "grad_norm": 5.3329620361328125,
      "learning_rate": 3.811528611023401e-05,
      "loss": 0.5131,
      "step": 19884
    },
    {
      "epoch": 6.278812756551942,
      "grad_norm": 5.9321136474609375,
      "learning_rate": 3.811440900957794e-05,
      "loss": 0.6561,
      "step": 19885
    },
    {
      "epoch": 6.279128512788128,
      "grad_norm": 5.397989749908447,
      "learning_rate": 3.811353190892187e-05,
      "loss": 0.6266,
      "step": 19886
    },
    {
      "epoch": 6.279444269024313,
      "grad_norm": 5.920359134674072,
      "learning_rate": 3.8112654808265794e-05,
      "loss": 0.6481,
      "step": 19887
    },
    {
      "epoch": 6.279760025260499,
      "grad_norm": 4.93864107131958,
      "learning_rate": 3.8111777707609725e-05,
      "loss": 0.5338,
      "step": 19888
    },
    {
      "epoch": 6.280075781496684,
      "grad_norm": 5.705445766448975,
      "learning_rate": 3.811090060695365e-05,
      "loss": 0.6934,
      "step": 19889
    },
    {
      "epoch": 6.2803915377328705,
      "grad_norm": 4.770336151123047,
      "learning_rate": 3.811002350629758e-05,
      "loss": 0.6833,
      "step": 19890
    },
    {
      "epoch": 6.280707293969056,
      "grad_norm": 3.866690158843994,
      "learning_rate": 3.8109146405641514e-05,
      "loss": 0.4683,
      "step": 19891
    },
    {
      "epoch": 6.281023050205242,
      "grad_norm": 4.283205509185791,
      "learning_rate": 3.8108269304985445e-05,
      "loss": 0.5291,
      "step": 19892
    },
    {
      "epoch": 6.281338806441427,
      "grad_norm": 5.196132183074951,
      "learning_rate": 3.810739220432937e-05,
      "loss": 0.5783,
      "step": 19893
    },
    {
      "epoch": 6.281654562677613,
      "grad_norm": 4.405132293701172,
      "learning_rate": 3.81065151036733e-05,
      "loss": 0.4203,
      "step": 19894
    },
    {
      "epoch": 6.281970318913799,
      "grad_norm": 4.613105773925781,
      "learning_rate": 3.810563800301723e-05,
      "loss": 0.5256,
      "step": 19895
    },
    {
      "epoch": 6.282286075149984,
      "grad_norm": 5.025731086730957,
      "learning_rate": 3.810476090236116e-05,
      "loss": 0.5771,
      "step": 19896
    },
    {
      "epoch": 6.28260183138617,
      "grad_norm": 4.289618492126465,
      "learning_rate": 3.810388380170509e-05,
      "loss": 0.5882,
      "step": 19897
    },
    {
      "epoch": 6.2829175876223555,
      "grad_norm": 3.7792606353759766,
      "learning_rate": 3.8103006701049015e-05,
      "loss": 0.3989,
      "step": 19898
    },
    {
      "epoch": 6.283233343858541,
      "grad_norm": 4.376801013946533,
      "learning_rate": 3.810212960039295e-05,
      "loss": 0.5909,
      "step": 19899
    },
    {
      "epoch": 6.283549100094727,
      "grad_norm": 4.870447158813477,
      "learning_rate": 3.810125249973687e-05,
      "loss": 0.5525,
      "step": 19900
    },
    {
      "epoch": 6.283864856330912,
      "grad_norm": 4.672911643981934,
      "learning_rate": 3.81003753990808e-05,
      "loss": 0.6357,
      "step": 19901
    },
    {
      "epoch": 6.2841806125670985,
      "grad_norm": 4.192631721496582,
      "learning_rate": 3.809949829842473e-05,
      "loss": 0.4882,
      "step": 19902
    },
    {
      "epoch": 6.284496368803284,
      "grad_norm": 4.780590057373047,
      "learning_rate": 3.809862119776866e-05,
      "loss": 0.6072,
      "step": 19903
    },
    {
      "epoch": 6.284812125039469,
      "grad_norm": 4.563407897949219,
      "learning_rate": 3.8097744097112585e-05,
      "loss": 0.5263,
      "step": 19904
    },
    {
      "epoch": 6.285127881275655,
      "grad_norm": 4.576781272888184,
      "learning_rate": 3.8096866996456516e-05,
      "loss": 0.5632,
      "step": 19905
    },
    {
      "epoch": 6.2854436375118405,
      "grad_norm": 4.365489482879639,
      "learning_rate": 3.809598989580044e-05,
      "loss": 0.509,
      "step": 19906
    },
    {
      "epoch": 6.285759393748027,
      "grad_norm": 4.186729907989502,
      "learning_rate": 3.809511279514437e-05,
      "loss": 0.5134,
      "step": 19907
    },
    {
      "epoch": 6.286075149984212,
      "grad_norm": 4.2269062995910645,
      "learning_rate": 3.80942356944883e-05,
      "loss": 0.476,
      "step": 19908
    },
    {
      "epoch": 6.286390906220398,
      "grad_norm": 4.39771032333374,
      "learning_rate": 3.809335859383223e-05,
      "loss": 0.5196,
      "step": 19909
    },
    {
      "epoch": 6.2867066624565835,
      "grad_norm": 4.699497222900391,
      "learning_rate": 3.8092481493176155e-05,
      "loss": 0.5347,
      "step": 19910
    },
    {
      "epoch": 6.287022418692769,
      "grad_norm": 4.4113264083862305,
      "learning_rate": 3.8091604392520086e-05,
      "loss": 0.4731,
      "step": 19911
    },
    {
      "epoch": 6.287338174928955,
      "grad_norm": 4.23927116394043,
      "learning_rate": 3.809072729186402e-05,
      "loss": 0.5354,
      "step": 19912
    },
    {
      "epoch": 6.28765393116514,
      "grad_norm": 5.595963478088379,
      "learning_rate": 3.808985019120794e-05,
      "loss": 0.6079,
      "step": 19913
    },
    {
      "epoch": 6.2879696874013264,
      "grad_norm": 5.139841079711914,
      "learning_rate": 3.8088973090551874e-05,
      "loss": 0.5081,
      "step": 19914
    },
    {
      "epoch": 6.288285443637512,
      "grad_norm": 3.815960645675659,
      "learning_rate": 3.80880959898958e-05,
      "loss": 0.5324,
      "step": 19915
    },
    {
      "epoch": 6.288601199873698,
      "grad_norm": 7.660187721252441,
      "learning_rate": 3.808721888923973e-05,
      "loss": 0.6402,
      "step": 19916
    },
    {
      "epoch": 6.288916956109883,
      "grad_norm": 5.327591896057129,
      "learning_rate": 3.8086341788583656e-05,
      "loss": 0.5451,
      "step": 19917
    },
    {
      "epoch": 6.2892327123460685,
      "grad_norm": 4.528990745544434,
      "learning_rate": 3.8085464687927594e-05,
      "loss": 0.5192,
      "step": 19918
    },
    {
      "epoch": 6.289548468582255,
      "grad_norm": 5.103721618652344,
      "learning_rate": 3.808458758727152e-05,
      "loss": 0.654,
      "step": 19919
    },
    {
      "epoch": 6.28986422481844,
      "grad_norm": 5.814479827880859,
      "learning_rate": 3.808371048661545e-05,
      "loss": 0.6521,
      "step": 19920
    },
    {
      "epoch": 6.290179981054626,
      "grad_norm": 4.520328998565674,
      "learning_rate": 3.8082833385959376e-05,
      "loss": 0.5042,
      "step": 19921
    },
    {
      "epoch": 6.2904957372908115,
      "grad_norm": 5.330357551574707,
      "learning_rate": 3.808195628530331e-05,
      "loss": 0.5638,
      "step": 19922
    },
    {
      "epoch": 6.290811493526997,
      "grad_norm": 5.227426052093506,
      "learning_rate": 3.808107918464723e-05,
      "loss": 0.6294,
      "step": 19923
    },
    {
      "epoch": 6.291127249763183,
      "grad_norm": 4.795641899108887,
      "learning_rate": 3.8080202083991164e-05,
      "loss": 0.6059,
      "step": 19924
    },
    {
      "epoch": 6.291443005999368,
      "grad_norm": 5.191168308258057,
      "learning_rate": 3.807932498333509e-05,
      "loss": 0.5816,
      "step": 19925
    },
    {
      "epoch": 6.291758762235554,
      "grad_norm": 5.0446062088012695,
      "learning_rate": 3.807844788267902e-05,
      "loss": 0.7064,
      "step": 19926
    },
    {
      "epoch": 6.29207451847174,
      "grad_norm": 4.085293292999268,
      "learning_rate": 3.8077570782022946e-05,
      "loss": 0.538,
      "step": 19927
    },
    {
      "epoch": 6.292390274707926,
      "grad_norm": 3.819354295730591,
      "learning_rate": 3.807669368136688e-05,
      "loss": 0.4456,
      "step": 19928
    },
    {
      "epoch": 6.292706030944111,
      "grad_norm": 5.290402412414551,
      "learning_rate": 3.80758165807108e-05,
      "loss": 0.5659,
      "step": 19929
    },
    {
      "epoch": 6.2930217871802965,
      "grad_norm": 3.866086006164551,
      "learning_rate": 3.8074939480054734e-05,
      "loss": 0.4779,
      "step": 19930
    },
    {
      "epoch": 6.293337543416483,
      "grad_norm": 5.388267517089844,
      "learning_rate": 3.807406237939866e-05,
      "loss": 0.5101,
      "step": 19931
    },
    {
      "epoch": 6.293653299652668,
      "grad_norm": 5.404120922088623,
      "learning_rate": 3.807318527874259e-05,
      "loss": 0.5898,
      "step": 19932
    },
    {
      "epoch": 6.293969055888854,
      "grad_norm": 4.420909881591797,
      "learning_rate": 3.8072308178086515e-05,
      "loss": 0.5299,
      "step": 19933
    },
    {
      "epoch": 6.294284812125039,
      "grad_norm": 4.354738712310791,
      "learning_rate": 3.807143107743045e-05,
      "loss": 0.5343,
      "step": 19934
    },
    {
      "epoch": 6.294600568361225,
      "grad_norm": 4.109598636627197,
      "learning_rate": 3.807055397677438e-05,
      "loss": 0.5661,
      "step": 19935
    },
    {
      "epoch": 6.294916324597411,
      "grad_norm": 5.396235942840576,
      "learning_rate": 3.8069676876118304e-05,
      "loss": 0.6085,
      "step": 19936
    },
    {
      "epoch": 6.295232080833596,
      "grad_norm": 5.649945259094238,
      "learning_rate": 3.8068799775462235e-05,
      "loss": 0.6454,
      "step": 19937
    },
    {
      "epoch": 6.295547837069782,
      "grad_norm": 4.657750606536865,
      "learning_rate": 3.806792267480616e-05,
      "loss": 0.4627,
      "step": 19938
    },
    {
      "epoch": 6.295863593305968,
      "grad_norm": 4.816073894500732,
      "learning_rate": 3.806704557415009e-05,
      "loss": 0.5659,
      "step": 19939
    },
    {
      "epoch": 6.296179349542154,
      "grad_norm": 4.838050842285156,
      "learning_rate": 3.806616847349402e-05,
      "loss": 0.5553,
      "step": 19940
    },
    {
      "epoch": 6.296495105778339,
      "grad_norm": 4.677206993103027,
      "learning_rate": 3.806529137283795e-05,
      "loss": 0.5632,
      "step": 19941
    },
    {
      "epoch": 6.2968108620145244,
      "grad_norm": 4.900526523590088,
      "learning_rate": 3.8064414272181873e-05,
      "loss": 0.5778,
      "step": 19942
    },
    {
      "epoch": 6.297126618250711,
      "grad_norm": 4.875369548797607,
      "learning_rate": 3.8063537171525805e-05,
      "loss": 0.5526,
      "step": 19943
    },
    {
      "epoch": 6.297442374486896,
      "grad_norm": 4.463021755218506,
      "learning_rate": 3.806266007086973e-05,
      "loss": 0.4854,
      "step": 19944
    },
    {
      "epoch": 6.297758130723082,
      "grad_norm": 4.66146993637085,
      "learning_rate": 3.806178297021366e-05,
      "loss": 0.5907,
      "step": 19945
    },
    {
      "epoch": 6.298073886959267,
      "grad_norm": 4.624927997589111,
      "learning_rate": 3.8060905869557593e-05,
      "loss": 0.5281,
      "step": 19946
    },
    {
      "epoch": 6.298389643195453,
      "grad_norm": 4.046142101287842,
      "learning_rate": 3.8060028768901525e-05,
      "loss": 0.5052,
      "step": 19947
    },
    {
      "epoch": 6.298705399431639,
      "grad_norm": 4.8470139503479,
      "learning_rate": 3.805915166824545e-05,
      "loss": 0.4945,
      "step": 19948
    },
    {
      "epoch": 6.299021155667824,
      "grad_norm": 5.204683780670166,
      "learning_rate": 3.805827456758938e-05,
      "loss": 0.516,
      "step": 19949
    },
    {
      "epoch": 6.29933691190401,
      "grad_norm": 4.071673393249512,
      "learning_rate": 3.8057397466933307e-05,
      "loss": 0.5273,
      "step": 19950
    },
    {
      "epoch": 6.299652668140196,
      "grad_norm": 4.912153244018555,
      "learning_rate": 3.805652036627724e-05,
      "loss": 0.5931,
      "step": 19951
    },
    {
      "epoch": 6.299968424376382,
      "grad_norm": 4.2216877937316895,
      "learning_rate": 3.805564326562116e-05,
      "loss": 0.5147,
      "step": 19952
    },
    {
      "epoch": 6.300284180612567,
      "grad_norm": 4.364152908325195,
      "learning_rate": 3.8054766164965095e-05,
      "loss": 0.6074,
      "step": 19953
    },
    {
      "epoch": 6.300599936848752,
      "grad_norm": 5.035661220550537,
      "learning_rate": 3.805388906430902e-05,
      "loss": 0.5798,
      "step": 19954
    },
    {
      "epoch": 6.300915693084939,
      "grad_norm": 3.90848708152771,
      "learning_rate": 3.805301196365295e-05,
      "loss": 0.5378,
      "step": 19955
    },
    {
      "epoch": 6.301231449321124,
      "grad_norm": 4.663891792297363,
      "learning_rate": 3.805213486299688e-05,
      "loss": 0.6188,
      "step": 19956
    },
    {
      "epoch": 6.30154720555731,
      "grad_norm": 4.129935264587402,
      "learning_rate": 3.805125776234081e-05,
      "loss": 0.5092,
      "step": 19957
    },
    {
      "epoch": 6.301862961793495,
      "grad_norm": 5.33107328414917,
      "learning_rate": 3.805038066168474e-05,
      "loss": 0.5871,
      "step": 19958
    },
    {
      "epoch": 6.3021787180296815,
      "grad_norm": 4.269098281860352,
      "learning_rate": 3.8049503561028665e-05,
      "loss": 0.5214,
      "step": 19959
    },
    {
      "epoch": 6.302494474265867,
      "grad_norm": 4.7573771476745605,
      "learning_rate": 3.8048626460372596e-05,
      "loss": 0.5256,
      "step": 19960
    },
    {
      "epoch": 6.302810230502052,
      "grad_norm": 4.853549957275391,
      "learning_rate": 3.804774935971652e-05,
      "loss": 0.5673,
      "step": 19961
    },
    {
      "epoch": 6.303125986738238,
      "grad_norm": 4.898044586181641,
      "learning_rate": 3.804687225906045e-05,
      "loss": 0.5562,
      "step": 19962
    },
    {
      "epoch": 6.303441742974424,
      "grad_norm": 5.401671886444092,
      "learning_rate": 3.804599515840438e-05,
      "loss": 0.5118,
      "step": 19963
    },
    {
      "epoch": 6.30375749921061,
      "grad_norm": 5.4885101318359375,
      "learning_rate": 3.804511805774831e-05,
      "loss": 0.6563,
      "step": 19964
    },
    {
      "epoch": 6.304073255446795,
      "grad_norm": 4.776140213012695,
      "learning_rate": 3.8044240957092234e-05,
      "loss": 0.5673,
      "step": 19965
    },
    {
      "epoch": 6.30438901168298,
      "grad_norm": 7.872138500213623,
      "learning_rate": 3.8043363856436166e-05,
      "loss": 0.5744,
      "step": 19966
    },
    {
      "epoch": 6.304704767919167,
      "grad_norm": 5.961231708526611,
      "learning_rate": 3.804248675578009e-05,
      "loss": 0.5752,
      "step": 19967
    },
    {
      "epoch": 6.305020524155352,
      "grad_norm": 4.822962760925293,
      "learning_rate": 3.804160965512402e-05,
      "loss": 0.6039,
      "step": 19968
    },
    {
      "epoch": 6.305336280391538,
      "grad_norm": 5.651944160461426,
      "learning_rate": 3.804073255446795e-05,
      "loss": 0.5775,
      "step": 19969
    },
    {
      "epoch": 6.305652036627723,
      "grad_norm": 4.351196765899658,
      "learning_rate": 3.803985545381188e-05,
      "loss": 0.5451,
      "step": 19970
    },
    {
      "epoch": 6.305967792863909,
      "grad_norm": 4.9641594886779785,
      "learning_rate": 3.8038978353155804e-05,
      "loss": 0.61,
      "step": 19971
    },
    {
      "epoch": 6.306283549100095,
      "grad_norm": 5.418788909912109,
      "learning_rate": 3.8038101252499736e-05,
      "loss": 0.5221,
      "step": 19972
    },
    {
      "epoch": 6.30659930533628,
      "grad_norm": 5.482302665710449,
      "learning_rate": 3.803722415184367e-05,
      "loss": 0.5919,
      "step": 19973
    },
    {
      "epoch": 6.306915061572466,
      "grad_norm": 3.9943222999572754,
      "learning_rate": 3.80363470511876e-05,
      "loss": 0.4534,
      "step": 19974
    },
    {
      "epoch": 6.307230817808652,
      "grad_norm": 4.944986820220947,
      "learning_rate": 3.8035469950531524e-05,
      "loss": 0.5649,
      "step": 19975
    },
    {
      "epoch": 6.307546574044838,
      "grad_norm": 5.39808988571167,
      "learning_rate": 3.8034592849875456e-05,
      "loss": 0.5788,
      "step": 19976
    },
    {
      "epoch": 6.307862330281023,
      "grad_norm": 4.3244805335998535,
      "learning_rate": 3.803371574921938e-05,
      "loss": 0.4909,
      "step": 19977
    },
    {
      "epoch": 6.308178086517208,
      "grad_norm": 5.258419036865234,
      "learning_rate": 3.803283864856331e-05,
      "loss": 0.5611,
      "step": 19978
    },
    {
      "epoch": 6.3084938427533945,
      "grad_norm": 4.897175312042236,
      "learning_rate": 3.8031961547907244e-05,
      "loss": 0.6519,
      "step": 19979
    },
    {
      "epoch": 6.30880959898958,
      "grad_norm": 6.089073657989502,
      "learning_rate": 3.803108444725117e-05,
      "loss": 0.5967,
      "step": 19980
    },
    {
      "epoch": 6.309125355225766,
      "grad_norm": 5.114573001861572,
      "learning_rate": 3.80302073465951e-05,
      "loss": 0.4607,
      "step": 19981
    },
    {
      "epoch": 6.309441111461951,
      "grad_norm": 5.3752923011779785,
      "learning_rate": 3.8029330245939026e-05,
      "loss": 0.6251,
      "step": 19982
    },
    {
      "epoch": 6.3097568676981375,
      "grad_norm": 4.430440425872803,
      "learning_rate": 3.802845314528296e-05,
      "loss": 0.6514,
      "step": 19983
    },
    {
      "epoch": 6.310072623934323,
      "grad_norm": 4.474634647369385,
      "learning_rate": 3.802757604462688e-05,
      "loss": 0.4737,
      "step": 19984
    },
    {
      "epoch": 6.310388380170508,
      "grad_norm": 4.1199951171875,
      "learning_rate": 3.8026698943970814e-05,
      "loss": 0.5191,
      "step": 19985
    },
    {
      "epoch": 6.310704136406694,
      "grad_norm": 3.8759214878082275,
      "learning_rate": 3.802582184331474e-05,
      "loss": 0.4596,
      "step": 19986
    },
    {
      "epoch": 6.3110198926428795,
      "grad_norm": 4.576883792877197,
      "learning_rate": 3.802494474265867e-05,
      "loss": 0.6195,
      "step": 19987
    },
    {
      "epoch": 6.311335648879066,
      "grad_norm": 5.139993667602539,
      "learning_rate": 3.8024067642002595e-05,
      "loss": 0.4848,
      "step": 19988
    },
    {
      "epoch": 6.311651405115251,
      "grad_norm": 6.707374095916748,
      "learning_rate": 3.802319054134653e-05,
      "loss": 0.6511,
      "step": 19989
    },
    {
      "epoch": 6.311967161351436,
      "grad_norm": 4.517267227172852,
      "learning_rate": 3.802231344069045e-05,
      "loss": 0.508,
      "step": 19990
    },
    {
      "epoch": 6.3122829175876225,
      "grad_norm": 4.874435901641846,
      "learning_rate": 3.8021436340034384e-05,
      "loss": 0.5619,
      "step": 19991
    },
    {
      "epoch": 6.312598673823808,
      "grad_norm": 5.689874649047852,
      "learning_rate": 3.802055923937831e-05,
      "loss": 0.6565,
      "step": 19992
    },
    {
      "epoch": 6.312914430059994,
      "grad_norm": 4.147286415100098,
      "learning_rate": 3.801968213872224e-05,
      "loss": 0.4772,
      "step": 19993
    },
    {
      "epoch": 6.313230186296179,
      "grad_norm": 6.160012245178223,
      "learning_rate": 3.801880503806617e-05,
      "loss": 0.507,
      "step": 19994
    },
    {
      "epoch": 6.313545942532365,
      "grad_norm": 5.479426383972168,
      "learning_rate": 3.80179279374101e-05,
      "loss": 0.5926,
      "step": 19995
    },
    {
      "epoch": 6.313861698768551,
      "grad_norm": 5.7216949462890625,
      "learning_rate": 3.801705083675403e-05,
      "loss": 0.5485,
      "step": 19996
    },
    {
      "epoch": 6.314177455004736,
      "grad_norm": 5.155942440032959,
      "learning_rate": 3.801617373609795e-05,
      "loss": 0.5552,
      "step": 19997
    },
    {
      "epoch": 6.314493211240922,
      "grad_norm": 4.850526332855225,
      "learning_rate": 3.8015296635441885e-05,
      "loss": 0.6093,
      "step": 19998
    },
    {
      "epoch": 6.3148089674771075,
      "grad_norm": 5.519396781921387,
      "learning_rate": 3.801441953478581e-05,
      "loss": 0.5169,
      "step": 19999
    },
    {
      "epoch": 6.315124723713294,
      "grad_norm": 4.57039213180542,
      "learning_rate": 3.801354243412975e-05,
      "loss": 0.4926,
      "step": 20000
    },
    {
      "epoch": 6.315124723713294,
      "eval_loss": 0.8858704566955566,
      "eval_runtime": 4.529,
      "eval_samples_per_second": 549.795,
      "eval_steps_per_second": 4.637,
      "step": 20000
    },
    {
      "epoch": 6.315440479949479,
      "grad_norm": 6.724633693695068,
      "learning_rate": 3.801266533347367e-05,
      "loss": 0.5483,
      "step": 20001
    },
    {
      "epoch": 6.315756236185664,
      "grad_norm": 6.137542724609375,
      "learning_rate": 3.8011788232817605e-05,
      "loss": 0.6063,
      "step": 20002
    },
    {
      "epoch": 6.3160719924218505,
      "grad_norm": 4.160109043121338,
      "learning_rate": 3.801091113216153e-05,
      "loss": 0.5189,
      "step": 20003
    },
    {
      "epoch": 6.316387748658036,
      "grad_norm": 5.5705060958862305,
      "learning_rate": 3.801003403150546e-05,
      "loss": 0.5905,
      "step": 20004
    },
    {
      "epoch": 6.316703504894222,
      "grad_norm": 4.370935440063477,
      "learning_rate": 3.8009156930849387e-05,
      "loss": 0.5676,
      "step": 20005
    },
    {
      "epoch": 6.317019261130407,
      "grad_norm": 4.635379791259766,
      "learning_rate": 3.800827983019332e-05,
      "loss": 0.4766,
      "step": 20006
    },
    {
      "epoch": 6.317335017366593,
      "grad_norm": 4.778022289276123,
      "learning_rate": 3.800740272953724e-05,
      "loss": 0.5585,
      "step": 20007
    },
    {
      "epoch": 6.317650773602779,
      "grad_norm": 5.436031341552734,
      "learning_rate": 3.8006525628881175e-05,
      "loss": 0.5172,
      "step": 20008
    },
    {
      "epoch": 6.317966529838964,
      "grad_norm": 5.349567890167236,
      "learning_rate": 3.80056485282251e-05,
      "loss": 0.5314,
      "step": 20009
    },
    {
      "epoch": 6.31828228607515,
      "grad_norm": 4.340682029724121,
      "learning_rate": 3.800477142756903e-05,
      "loss": 0.6572,
      "step": 20010
    },
    {
      "epoch": 6.3185980423113355,
      "grad_norm": 5.0304951667785645,
      "learning_rate": 3.8003894326912956e-05,
      "loss": 0.486,
      "step": 20011
    },
    {
      "epoch": 6.318913798547522,
      "grad_norm": 5.9866042137146,
      "learning_rate": 3.800301722625689e-05,
      "loss": 0.6422,
      "step": 20012
    },
    {
      "epoch": 6.319229554783707,
      "grad_norm": 3.970947504043579,
      "learning_rate": 3.800214012560081e-05,
      "loss": 0.4517,
      "step": 20013
    },
    {
      "epoch": 6.319545311019892,
      "grad_norm": 4.496171474456787,
      "learning_rate": 3.8001263024944745e-05,
      "loss": 0.5645,
      "step": 20014
    },
    {
      "epoch": 6.319861067256078,
      "grad_norm": 4.607635974884033,
      "learning_rate": 3.8000385924288676e-05,
      "loss": 0.6265,
      "step": 20015
    },
    {
      "epoch": 6.320176823492264,
      "grad_norm": 4.178745746612549,
      "learning_rate": 3.79995088236326e-05,
      "loss": 0.5157,
      "step": 20016
    },
    {
      "epoch": 6.32049257972845,
      "grad_norm": 4.285256862640381,
      "learning_rate": 3.799863172297653e-05,
      "loss": 0.4732,
      "step": 20017
    },
    {
      "epoch": 6.320808335964635,
      "grad_norm": 5.174956321716309,
      "learning_rate": 3.799775462232046e-05,
      "loss": 0.6093,
      "step": 20018
    },
    {
      "epoch": 6.3211240922008205,
      "grad_norm": 5.914335250854492,
      "learning_rate": 3.799687752166439e-05,
      "loss": 0.5664,
      "step": 20019
    },
    {
      "epoch": 6.321439848437007,
      "grad_norm": 6.110134124755859,
      "learning_rate": 3.7996000421008314e-05,
      "loss": 0.6447,
      "step": 20020
    },
    {
      "epoch": 6.321755604673192,
      "grad_norm": 4.899552345275879,
      "learning_rate": 3.7995123320352246e-05,
      "loss": 0.5436,
      "step": 20021
    },
    {
      "epoch": 6.322071360909378,
      "grad_norm": 4.277342319488525,
      "learning_rate": 3.799424621969617e-05,
      "loss": 0.5407,
      "step": 20022
    },
    {
      "epoch": 6.3223871171455635,
      "grad_norm": 3.705629587173462,
      "learning_rate": 3.79933691190401e-05,
      "loss": 0.5622,
      "step": 20023
    },
    {
      "epoch": 6.32270287338175,
      "grad_norm": 5.431956768035889,
      "learning_rate": 3.799249201838403e-05,
      "loss": 0.533,
      "step": 20024
    },
    {
      "epoch": 6.323018629617935,
      "grad_norm": 4.0084757804870605,
      "learning_rate": 3.799161491772796e-05,
      "loss": 0.4234,
      "step": 20025
    },
    {
      "epoch": 6.32333438585412,
      "grad_norm": 4.43281364440918,
      "learning_rate": 3.7990737817071884e-05,
      "loss": 0.5435,
      "step": 20026
    },
    {
      "epoch": 6.323650142090306,
      "grad_norm": 4.840095520019531,
      "learning_rate": 3.798986071641582e-05,
      "loss": 0.6293,
      "step": 20027
    },
    {
      "epoch": 6.323965898326492,
      "grad_norm": 3.5184900760650635,
      "learning_rate": 3.798898361575975e-05,
      "loss": 0.4924,
      "step": 20028
    },
    {
      "epoch": 6.324281654562678,
      "grad_norm": 4.424259185791016,
      "learning_rate": 3.798810651510368e-05,
      "loss": 0.487,
      "step": 20029
    },
    {
      "epoch": 6.324597410798863,
      "grad_norm": 5.341587066650391,
      "learning_rate": 3.7987229414447604e-05,
      "loss": 0.5485,
      "step": 20030
    },
    {
      "epoch": 6.324913167035049,
      "grad_norm": 4.141517162322998,
      "learning_rate": 3.7986352313791536e-05,
      "loss": 0.6512,
      "step": 20031
    },
    {
      "epoch": 6.325228923271235,
      "grad_norm": 4.280179977416992,
      "learning_rate": 3.798547521313546e-05,
      "loss": 0.444,
      "step": 20032
    },
    {
      "epoch": 6.32554467950742,
      "grad_norm": 5.672369956970215,
      "learning_rate": 3.798459811247939e-05,
      "loss": 0.5068,
      "step": 20033
    },
    {
      "epoch": 6.325860435743606,
      "grad_norm": 5.580483436584473,
      "learning_rate": 3.798372101182332e-05,
      "loss": 0.6058,
      "step": 20034
    },
    {
      "epoch": 6.326176191979791,
      "grad_norm": 4.333906650543213,
      "learning_rate": 3.798284391116725e-05,
      "loss": 0.515,
      "step": 20035
    },
    {
      "epoch": 6.326491948215978,
      "grad_norm": 5.567683219909668,
      "learning_rate": 3.7981966810511174e-05,
      "loss": 0.4816,
      "step": 20036
    },
    {
      "epoch": 6.326807704452163,
      "grad_norm": 5.394087791442871,
      "learning_rate": 3.7981089709855105e-05,
      "loss": 0.4802,
      "step": 20037
    },
    {
      "epoch": 6.327123460688348,
      "grad_norm": 5.883114814758301,
      "learning_rate": 3.798021260919904e-05,
      "loss": 0.6568,
      "step": 20038
    },
    {
      "epoch": 6.327439216924534,
      "grad_norm": 5.112370014190674,
      "learning_rate": 3.797933550854296e-05,
      "loss": 0.5557,
      "step": 20039
    },
    {
      "epoch": 6.32775497316072,
      "grad_norm": 6.208863735198975,
      "learning_rate": 3.7978458407886894e-05,
      "loss": 0.6428,
      "step": 20040
    },
    {
      "epoch": 6.328070729396906,
      "grad_norm": 8.07403564453125,
      "learning_rate": 3.797758130723082e-05,
      "loss": 0.4733,
      "step": 20041
    },
    {
      "epoch": 6.328386485633091,
      "grad_norm": 5.2287678718566895,
      "learning_rate": 3.797670420657475e-05,
      "loss": 0.5462,
      "step": 20042
    },
    {
      "epoch": 6.328702241869277,
      "grad_norm": 5.997861385345459,
      "learning_rate": 3.7975827105918675e-05,
      "loss": 0.495,
      "step": 20043
    },
    {
      "epoch": 6.329017998105463,
      "grad_norm": 5.102799415588379,
      "learning_rate": 3.797495000526261e-05,
      "loss": 0.4744,
      "step": 20044
    },
    {
      "epoch": 6.329333754341648,
      "grad_norm": 4.515707492828369,
      "learning_rate": 3.797407290460653e-05,
      "loss": 0.617,
      "step": 20045
    },
    {
      "epoch": 6.329649510577834,
      "grad_norm": 4.106293201446533,
      "learning_rate": 3.7973195803950463e-05,
      "loss": 0.4503,
      "step": 20046
    },
    {
      "epoch": 6.329965266814019,
      "grad_norm": 5.00166130065918,
      "learning_rate": 3.797231870329439e-05,
      "loss": 0.6045,
      "step": 20047
    },
    {
      "epoch": 6.330281023050206,
      "grad_norm": 5.194067001342773,
      "learning_rate": 3.797144160263832e-05,
      "loss": 0.6024,
      "step": 20048
    },
    {
      "epoch": 6.330596779286391,
      "grad_norm": 4.574303150177002,
      "learning_rate": 3.7970564501982245e-05,
      "loss": 0.5557,
      "step": 20049
    },
    {
      "epoch": 6.330912535522576,
      "grad_norm": 5.634212970733643,
      "learning_rate": 3.796968740132618e-05,
      "loss": 0.5922,
      "step": 20050
    },
    {
      "epoch": 6.331228291758762,
      "grad_norm": 5.506183624267578,
      "learning_rate": 3.79688103006701e-05,
      "loss": 0.5114,
      "step": 20051
    },
    {
      "epoch": 6.331544047994948,
      "grad_norm": 4.983269691467285,
      "learning_rate": 3.796793320001403e-05,
      "loss": 0.5636,
      "step": 20052
    },
    {
      "epoch": 6.331859804231134,
      "grad_norm": 4.489537239074707,
      "learning_rate": 3.7967056099357965e-05,
      "loss": 0.573,
      "step": 20053
    },
    {
      "epoch": 6.332175560467319,
      "grad_norm": 4.894186973571777,
      "learning_rate": 3.796617899870189e-05,
      "loss": 0.5436,
      "step": 20054
    },
    {
      "epoch": 6.332491316703505,
      "grad_norm": 4.619128704071045,
      "learning_rate": 3.796530189804582e-05,
      "loss": 0.5922,
      "step": 20055
    },
    {
      "epoch": 6.332807072939691,
      "grad_norm": 4.002668857574463,
      "learning_rate": 3.796442479738975e-05,
      "loss": 0.5277,
      "step": 20056
    },
    {
      "epoch": 6.333122829175876,
      "grad_norm": 4.447104454040527,
      "learning_rate": 3.796354769673368e-05,
      "loss": 0.6234,
      "step": 20057
    },
    {
      "epoch": 6.333438585412062,
      "grad_norm": 6.319929599761963,
      "learning_rate": 3.796267059607761e-05,
      "loss": 0.5734,
      "step": 20058
    },
    {
      "epoch": 6.333754341648247,
      "grad_norm": 3.6790285110473633,
      "learning_rate": 3.796179349542154e-05,
      "loss": 0.4802,
      "step": 20059
    },
    {
      "epoch": 6.3340700978844335,
      "grad_norm": 5.031989097595215,
      "learning_rate": 3.7960916394765466e-05,
      "loss": 0.5945,
      "step": 20060
    },
    {
      "epoch": 6.334385854120619,
      "grad_norm": 4.297998428344727,
      "learning_rate": 3.79600392941094e-05,
      "loss": 0.4759,
      "step": 20061
    },
    {
      "epoch": 6.334701610356804,
      "grad_norm": 6.980151176452637,
      "learning_rate": 3.795916219345332e-05,
      "loss": 0.4912,
      "step": 20062
    },
    {
      "epoch": 6.33501736659299,
      "grad_norm": 6.45509672164917,
      "learning_rate": 3.7958285092797255e-05,
      "loss": 0.6587,
      "step": 20063
    },
    {
      "epoch": 6.335333122829176,
      "grad_norm": 5.130531311035156,
      "learning_rate": 3.795740799214118e-05,
      "loss": 0.5338,
      "step": 20064
    },
    {
      "epoch": 6.335648879065362,
      "grad_norm": 4.662084579467773,
      "learning_rate": 3.795653089148511e-05,
      "loss": 0.5368,
      "step": 20065
    },
    {
      "epoch": 6.335964635301547,
      "grad_norm": 6.65269660949707,
      "learning_rate": 3.7955653790829036e-05,
      "loss": 0.5495,
      "step": 20066
    },
    {
      "epoch": 6.336280391537733,
      "grad_norm": 4.192251205444336,
      "learning_rate": 3.795477669017297e-05,
      "loss": 0.4491,
      "step": 20067
    },
    {
      "epoch": 6.3365961477739186,
      "grad_norm": 4.46602201461792,
      "learning_rate": 3.795389958951689e-05,
      "loss": 0.4921,
      "step": 20068
    },
    {
      "epoch": 6.336911904010104,
      "grad_norm": 5.307815074920654,
      "learning_rate": 3.7953022488860824e-05,
      "loss": 0.4805,
      "step": 20069
    },
    {
      "epoch": 6.33722766024629,
      "grad_norm": 5.4433183670043945,
      "learning_rate": 3.795214538820475e-05,
      "loss": 0.6056,
      "step": 20070
    },
    {
      "epoch": 6.337543416482475,
      "grad_norm": 5.388355731964111,
      "learning_rate": 3.795126828754868e-05,
      "loss": 0.6155,
      "step": 20071
    },
    {
      "epoch": 6.3378591727186615,
      "grad_norm": 3.970893383026123,
      "learning_rate": 3.7950391186892606e-05,
      "loss": 0.478,
      "step": 20072
    },
    {
      "epoch": 6.338174928954847,
      "grad_norm": 3.1183416843414307,
      "learning_rate": 3.794951408623654e-05,
      "loss": 0.4486,
      "step": 20073
    },
    {
      "epoch": 6.338490685191033,
      "grad_norm": 4.9230241775512695,
      "learning_rate": 3.794863698558046e-05,
      "loss": 0.549,
      "step": 20074
    },
    {
      "epoch": 6.338806441427218,
      "grad_norm": 4.779385089874268,
      "learning_rate": 3.7947759884924394e-05,
      "loss": 0.6405,
      "step": 20075
    },
    {
      "epoch": 6.339122197663404,
      "grad_norm": 5.035606861114502,
      "learning_rate": 3.7946882784268326e-05,
      "loss": 0.5904,
      "step": 20076
    },
    {
      "epoch": 6.33943795389959,
      "grad_norm": 4.453649997711182,
      "learning_rate": 3.794600568361225e-05,
      "loss": 0.5912,
      "step": 20077
    },
    {
      "epoch": 6.339753710135775,
      "grad_norm": 4.966204643249512,
      "learning_rate": 3.794512858295618e-05,
      "loss": 0.5567,
      "step": 20078
    },
    {
      "epoch": 6.340069466371961,
      "grad_norm": 5.678737163543701,
      "learning_rate": 3.794425148230011e-05,
      "loss": 0.5758,
      "step": 20079
    },
    {
      "epoch": 6.3403852226081465,
      "grad_norm": 4.5891337394714355,
      "learning_rate": 3.794337438164404e-05,
      "loss": 0.5272,
      "step": 20080
    },
    {
      "epoch": 6.340700978844332,
      "grad_norm": 4.58020544052124,
      "learning_rate": 3.7942497280987964e-05,
      "loss": 0.5566,
      "step": 20081
    },
    {
      "epoch": 6.341016735080518,
      "grad_norm": 6.153701305389404,
      "learning_rate": 3.79416201803319e-05,
      "loss": 0.685,
      "step": 20082
    },
    {
      "epoch": 6.341332491316703,
      "grad_norm": 4.489994049072266,
      "learning_rate": 3.794074307967583e-05,
      "loss": 0.4589,
      "step": 20083
    },
    {
      "epoch": 6.3416482475528895,
      "grad_norm": 6.741508960723877,
      "learning_rate": 3.793986597901976e-05,
      "loss": 0.5519,
      "step": 20084
    },
    {
      "epoch": 6.341964003789075,
      "grad_norm": 3.8528850078582764,
      "learning_rate": 3.7938988878363684e-05,
      "loss": 0.5398,
      "step": 20085
    },
    {
      "epoch": 6.34227976002526,
      "grad_norm": 5.661656379699707,
      "learning_rate": 3.7938111777707616e-05,
      "loss": 0.612,
      "step": 20086
    },
    {
      "epoch": 6.342595516261446,
      "grad_norm": 4.0140461921691895,
      "learning_rate": 3.793723467705154e-05,
      "loss": 0.5783,
      "step": 20087
    },
    {
      "epoch": 6.3429112724976315,
      "grad_norm": 4.944681644439697,
      "learning_rate": 3.793635757639547e-05,
      "loss": 0.5524,
      "step": 20088
    },
    {
      "epoch": 6.343227028733818,
      "grad_norm": 4.6613450050354,
      "learning_rate": 3.79354804757394e-05,
      "loss": 0.5772,
      "step": 20089
    },
    {
      "epoch": 6.343542784970003,
      "grad_norm": 5.2837748527526855,
      "learning_rate": 3.793460337508333e-05,
      "loss": 0.6238,
      "step": 20090
    },
    {
      "epoch": 6.343858541206189,
      "grad_norm": 5.165663242340088,
      "learning_rate": 3.7933726274427254e-05,
      "loss": 0.6913,
      "step": 20091
    },
    {
      "epoch": 6.3441742974423745,
      "grad_norm": 4.728214263916016,
      "learning_rate": 3.7932849173771185e-05,
      "loss": 0.5715,
      "step": 20092
    },
    {
      "epoch": 6.34449005367856,
      "grad_norm": 5.48568058013916,
      "learning_rate": 3.793197207311511e-05,
      "loss": 0.5637,
      "step": 20093
    },
    {
      "epoch": 6.344805809914746,
      "grad_norm": 5.369919300079346,
      "learning_rate": 3.793109497245904e-05,
      "loss": 0.4905,
      "step": 20094
    },
    {
      "epoch": 6.345121566150931,
      "grad_norm": 5.187039375305176,
      "learning_rate": 3.793021787180297e-05,
      "loss": 0.5258,
      "step": 20095
    },
    {
      "epoch": 6.345437322387117,
      "grad_norm": 4.63800573348999,
      "learning_rate": 3.79293407711469e-05,
      "loss": 0.4689,
      "step": 20096
    },
    {
      "epoch": 6.345753078623303,
      "grad_norm": 5.491262912750244,
      "learning_rate": 3.792846367049083e-05,
      "loss": 0.5014,
      "step": 20097
    },
    {
      "epoch": 6.346068834859489,
      "grad_norm": 5.255179405212402,
      "learning_rate": 3.7927586569834755e-05,
      "loss": 0.5859,
      "step": 20098
    },
    {
      "epoch": 6.346384591095674,
      "grad_norm": 3.9858129024505615,
      "learning_rate": 3.792670946917869e-05,
      "loss": 0.472,
      "step": 20099
    },
    {
      "epoch": 6.3467003473318595,
      "grad_norm": 5.11214017868042,
      "learning_rate": 3.792583236852261e-05,
      "loss": 0.6086,
      "step": 20100
    },
    {
      "epoch": 6.347016103568046,
      "grad_norm": 5.139270782470703,
      "learning_rate": 3.7924955267866543e-05,
      "loss": 0.6464,
      "step": 20101
    },
    {
      "epoch": 6.347331859804231,
      "grad_norm": 4.491567134857178,
      "learning_rate": 3.792407816721047e-05,
      "loss": 0.5545,
      "step": 20102
    },
    {
      "epoch": 6.347647616040417,
      "grad_norm": 4.65859842300415,
      "learning_rate": 3.79232010665544e-05,
      "loss": 0.5431,
      "step": 20103
    },
    {
      "epoch": 6.3479633722766025,
      "grad_norm": 5.791092872619629,
      "learning_rate": 3.7922323965898325e-05,
      "loss": 0.5927,
      "step": 20104
    },
    {
      "epoch": 6.348279128512788,
      "grad_norm": 4.886553764343262,
      "learning_rate": 3.7921446865242257e-05,
      "loss": 0.5031,
      "step": 20105
    },
    {
      "epoch": 6.348594884748974,
      "grad_norm": 4.472190856933594,
      "learning_rate": 3.792056976458618e-05,
      "loss": 0.5978,
      "step": 20106
    },
    {
      "epoch": 6.348910640985159,
      "grad_norm": 4.176294803619385,
      "learning_rate": 3.791969266393011e-05,
      "loss": 0.5187,
      "step": 20107
    },
    {
      "epoch": 6.349226397221345,
      "grad_norm": 4.819422245025635,
      "learning_rate": 3.791881556327404e-05,
      "loss": 0.5354,
      "step": 20108
    },
    {
      "epoch": 6.349542153457531,
      "grad_norm": 4.044826030731201,
      "learning_rate": 3.7917938462617977e-05,
      "loss": 0.5508,
      "step": 20109
    },
    {
      "epoch": 6.349857909693716,
      "grad_norm": 4.504233360290527,
      "learning_rate": 3.79170613619619e-05,
      "loss": 0.5922,
      "step": 20110
    },
    {
      "epoch": 6.350173665929902,
      "grad_norm": 6.179513931274414,
      "learning_rate": 3.791618426130583e-05,
      "loss": 0.6423,
      "step": 20111
    },
    {
      "epoch": 6.3504894221660875,
      "grad_norm": 5.32912540435791,
      "learning_rate": 3.791530716064976e-05,
      "loss": 0.5329,
      "step": 20112
    },
    {
      "epoch": 6.350805178402274,
      "grad_norm": 4.6864333152771,
      "learning_rate": 3.791443005999369e-05,
      "loss": 0.4874,
      "step": 20113
    },
    {
      "epoch": 6.351120934638459,
      "grad_norm": 4.7683796882629395,
      "learning_rate": 3.7913552959337615e-05,
      "loss": 0.5837,
      "step": 20114
    },
    {
      "epoch": 6.351436690874645,
      "grad_norm": 4.873595714569092,
      "learning_rate": 3.7912675858681546e-05,
      "loss": 0.592,
      "step": 20115
    },
    {
      "epoch": 6.35175244711083,
      "grad_norm": 4.362314701080322,
      "learning_rate": 3.791179875802547e-05,
      "loss": 0.6142,
      "step": 20116
    },
    {
      "epoch": 6.352068203347016,
      "grad_norm": 4.4715399742126465,
      "learning_rate": 3.79109216573694e-05,
      "loss": 0.6219,
      "step": 20117
    },
    {
      "epoch": 6.352383959583202,
      "grad_norm": 4.887246131896973,
      "learning_rate": 3.791004455671333e-05,
      "loss": 0.5263,
      "step": 20118
    },
    {
      "epoch": 6.352699715819387,
      "grad_norm": 4.435419082641602,
      "learning_rate": 3.790916745605726e-05,
      "loss": 0.5222,
      "step": 20119
    },
    {
      "epoch": 6.353015472055573,
      "grad_norm": 4.689207077026367,
      "learning_rate": 3.790829035540119e-05,
      "loss": 0.5653,
      "step": 20120
    },
    {
      "epoch": 6.353331228291759,
      "grad_norm": 4.3735833168029785,
      "learning_rate": 3.7907413254745116e-05,
      "loss": 0.5736,
      "step": 20121
    },
    {
      "epoch": 6.353646984527945,
      "grad_norm": 4.574765682220459,
      "learning_rate": 3.790653615408905e-05,
      "loss": 0.5754,
      "step": 20122
    },
    {
      "epoch": 6.35396274076413,
      "grad_norm": 4.753625869750977,
      "learning_rate": 3.790565905343297e-05,
      "loss": 0.5637,
      "step": 20123
    },
    {
      "epoch": 6.354278497000315,
      "grad_norm": 4.291818141937256,
      "learning_rate": 3.7904781952776904e-05,
      "loss": 0.4466,
      "step": 20124
    },
    {
      "epoch": 6.354594253236502,
      "grad_norm": 5.716591835021973,
      "learning_rate": 3.790390485212083e-05,
      "loss": 0.6218,
      "step": 20125
    },
    {
      "epoch": 6.354910009472687,
      "grad_norm": 4.6537628173828125,
      "learning_rate": 3.790302775146476e-05,
      "loss": 0.5719,
      "step": 20126
    },
    {
      "epoch": 6.355225765708873,
      "grad_norm": 4.112664222717285,
      "learning_rate": 3.7902150650808686e-05,
      "loss": 0.489,
      "step": 20127
    },
    {
      "epoch": 6.355541521945058,
      "grad_norm": 4.835784435272217,
      "learning_rate": 3.790127355015262e-05,
      "loss": 0.5783,
      "step": 20128
    },
    {
      "epoch": 6.355857278181244,
      "grad_norm": 4.591182231903076,
      "learning_rate": 3.790039644949654e-05,
      "loss": 0.5183,
      "step": 20129
    },
    {
      "epoch": 6.35617303441743,
      "grad_norm": 4.702629566192627,
      "learning_rate": 3.7899519348840474e-05,
      "loss": 0.5313,
      "step": 20130
    },
    {
      "epoch": 6.356488790653615,
      "grad_norm": 4.623154163360596,
      "learning_rate": 3.78986422481844e-05,
      "loss": 0.5305,
      "step": 20131
    },
    {
      "epoch": 6.356804546889801,
      "grad_norm": 5.628222942352295,
      "learning_rate": 3.789776514752833e-05,
      "loss": 0.5002,
      "step": 20132
    },
    {
      "epoch": 6.357120303125987,
      "grad_norm": 5.348610877990723,
      "learning_rate": 3.7896888046872256e-05,
      "loss": 0.5693,
      "step": 20133
    },
    {
      "epoch": 6.357436059362172,
      "grad_norm": 4.331168174743652,
      "learning_rate": 3.789601094621619e-05,
      "loss": 0.5714,
      "step": 20134
    },
    {
      "epoch": 6.357751815598358,
      "grad_norm": 5.797537326812744,
      "learning_rate": 3.789513384556012e-05,
      "loss": 0.6045,
      "step": 20135
    },
    {
      "epoch": 6.358067571834543,
      "grad_norm": 4.444230556488037,
      "learning_rate": 3.789425674490405e-05,
      "loss": 0.591,
      "step": 20136
    },
    {
      "epoch": 6.35838332807073,
      "grad_norm": 5.3068413734436035,
      "learning_rate": 3.7893379644247976e-05,
      "loss": 0.576,
      "step": 20137
    },
    {
      "epoch": 6.358699084306915,
      "grad_norm": 4.612595558166504,
      "learning_rate": 3.789250254359191e-05,
      "loss": 0.4857,
      "step": 20138
    },
    {
      "epoch": 6.359014840543101,
      "grad_norm": 5.812642574310303,
      "learning_rate": 3.789162544293583e-05,
      "loss": 0.6032,
      "step": 20139
    },
    {
      "epoch": 6.359330596779286,
      "grad_norm": 4.980307102203369,
      "learning_rate": 3.7890748342279764e-05,
      "loss": 0.5114,
      "step": 20140
    },
    {
      "epoch": 6.359646353015472,
      "grad_norm": 4.964173316955566,
      "learning_rate": 3.7889871241623695e-05,
      "loss": 0.5483,
      "step": 20141
    },
    {
      "epoch": 6.359962109251658,
      "grad_norm": 6.230589389801025,
      "learning_rate": 3.788899414096762e-05,
      "loss": 0.6498,
      "step": 20142
    },
    {
      "epoch": 6.360277865487843,
      "grad_norm": 4.088945388793945,
      "learning_rate": 3.788811704031155e-05,
      "loss": 0.4131,
      "step": 20143
    },
    {
      "epoch": 6.360593621724029,
      "grad_norm": 4.757166385650635,
      "learning_rate": 3.788723993965548e-05,
      "loss": 0.5833,
      "step": 20144
    },
    {
      "epoch": 6.360909377960215,
      "grad_norm": 5.364084720611572,
      "learning_rate": 3.788636283899941e-05,
      "loss": 0.501,
      "step": 20145
    },
    {
      "epoch": 6.361225134196401,
      "grad_norm": 6.121432304382324,
      "learning_rate": 3.7885485738343334e-05,
      "loss": 0.6745,
      "step": 20146
    },
    {
      "epoch": 6.361540890432586,
      "grad_norm": 7.390251636505127,
      "learning_rate": 3.7884608637687265e-05,
      "loss": 0.6397,
      "step": 20147
    },
    {
      "epoch": 6.361856646668771,
      "grad_norm": 5.274146556854248,
      "learning_rate": 3.788373153703119e-05,
      "loss": 0.5283,
      "step": 20148
    },
    {
      "epoch": 6.362172402904958,
      "grad_norm": 4.280710220336914,
      "learning_rate": 3.788285443637512e-05,
      "loss": 0.6345,
      "step": 20149
    },
    {
      "epoch": 6.362488159141143,
      "grad_norm": 5.63359260559082,
      "learning_rate": 3.788197733571905e-05,
      "loss": 0.5061,
      "step": 20150
    },
    {
      "epoch": 6.362803915377329,
      "grad_norm": 4.9330525398254395,
      "learning_rate": 3.788110023506298e-05,
      "loss": 0.51,
      "step": 20151
    },
    {
      "epoch": 6.363119671613514,
      "grad_norm": 5.051264762878418,
      "learning_rate": 3.78802231344069e-05,
      "loss": 0.5122,
      "step": 20152
    },
    {
      "epoch": 6.3634354278497,
      "grad_norm": 4.259067535400391,
      "learning_rate": 3.7879346033750835e-05,
      "loss": 0.5356,
      "step": 20153
    },
    {
      "epoch": 6.363751184085886,
      "grad_norm": 4.130179405212402,
      "learning_rate": 3.787846893309476e-05,
      "loss": 0.4493,
      "step": 20154
    },
    {
      "epoch": 6.364066940322071,
      "grad_norm": 4.9466071128845215,
      "learning_rate": 3.787759183243869e-05,
      "loss": 0.568,
      "step": 20155
    },
    {
      "epoch": 6.364382696558257,
      "grad_norm": 4.091452121734619,
      "learning_rate": 3.7876714731782617e-05,
      "loss": 0.4223,
      "step": 20156
    },
    {
      "epoch": 6.364698452794443,
      "grad_norm": 5.684792518615723,
      "learning_rate": 3.787583763112655e-05,
      "loss": 0.5546,
      "step": 20157
    },
    {
      "epoch": 6.365014209030629,
      "grad_norm": 3.8937556743621826,
      "learning_rate": 3.787496053047048e-05,
      "loss": 0.4588,
      "step": 20158
    },
    {
      "epoch": 6.365329965266814,
      "grad_norm": 3.933288335800171,
      "learning_rate": 3.7874083429814405e-05,
      "loss": 0.6137,
      "step": 20159
    },
    {
      "epoch": 6.365645721502999,
      "grad_norm": 6.284936904907227,
      "learning_rate": 3.7873206329158336e-05,
      "loss": 0.4717,
      "step": 20160
    },
    {
      "epoch": 6.3659614777391855,
      "grad_norm": 5.836061477661133,
      "learning_rate": 3.787232922850226e-05,
      "loss": 0.5435,
      "step": 20161
    },
    {
      "epoch": 6.366277233975371,
      "grad_norm": 5.657797336578369,
      "learning_rate": 3.787145212784619e-05,
      "loss": 0.6541,
      "step": 20162
    },
    {
      "epoch": 6.366592990211557,
      "grad_norm": 4.622621059417725,
      "learning_rate": 3.787057502719012e-05,
      "loss": 0.5944,
      "step": 20163
    },
    {
      "epoch": 6.366908746447742,
      "grad_norm": 6.635079383850098,
      "learning_rate": 3.7869697926534056e-05,
      "loss": 0.6882,
      "step": 20164
    },
    {
      "epoch": 6.3672245026839285,
      "grad_norm": 4.475678443908691,
      "learning_rate": 3.786882082587798e-05,
      "loss": 0.4627,
      "step": 20165
    },
    {
      "epoch": 6.367540258920114,
      "grad_norm": 4.7588701248168945,
      "learning_rate": 3.786794372522191e-05,
      "loss": 0.4624,
      "step": 20166
    },
    {
      "epoch": 6.367856015156299,
      "grad_norm": 4.34013032913208,
      "learning_rate": 3.786706662456584e-05,
      "loss": 0.6379,
      "step": 20167
    },
    {
      "epoch": 6.368171771392485,
      "grad_norm": 4.704349040985107,
      "learning_rate": 3.786618952390977e-05,
      "loss": 0.5483,
      "step": 20168
    },
    {
      "epoch": 6.3684875276286705,
      "grad_norm": 4.843536376953125,
      "learning_rate": 3.7865312423253694e-05,
      "loss": 0.5417,
      "step": 20169
    },
    {
      "epoch": 6.368803283864857,
      "grad_norm": 4.821539402008057,
      "learning_rate": 3.7864435322597626e-05,
      "loss": 0.5302,
      "step": 20170
    },
    {
      "epoch": 6.369119040101042,
      "grad_norm": 6.802778244018555,
      "learning_rate": 3.786355822194155e-05,
      "loss": 0.5308,
      "step": 20171
    },
    {
      "epoch": 6.369434796337227,
      "grad_norm": 4.6437530517578125,
      "learning_rate": 3.786268112128548e-05,
      "loss": 0.5325,
      "step": 20172
    },
    {
      "epoch": 6.3697505525734135,
      "grad_norm": 4.5049028396606445,
      "learning_rate": 3.786180402062941e-05,
      "loss": 0.4918,
      "step": 20173
    },
    {
      "epoch": 6.370066308809599,
      "grad_norm": 5.723482131958008,
      "learning_rate": 3.786092691997334e-05,
      "loss": 0.5699,
      "step": 20174
    },
    {
      "epoch": 6.370382065045785,
      "grad_norm": 4.551186561584473,
      "learning_rate": 3.7860049819317264e-05,
      "loss": 0.5248,
      "step": 20175
    },
    {
      "epoch": 6.37069782128197,
      "grad_norm": 5.243914604187012,
      "learning_rate": 3.7859172718661196e-05,
      "loss": 0.604,
      "step": 20176
    },
    {
      "epoch": 6.371013577518156,
      "grad_norm": 4.661707878112793,
      "learning_rate": 3.785829561800512e-05,
      "loss": 0.6105,
      "step": 20177
    },
    {
      "epoch": 6.371329333754342,
      "grad_norm": 5.103991508483887,
      "learning_rate": 3.785741851734905e-05,
      "loss": 0.5118,
      "step": 20178
    },
    {
      "epoch": 6.371645089990527,
      "grad_norm": 4.91917085647583,
      "learning_rate": 3.7856541416692984e-05,
      "loss": 0.6011,
      "step": 20179
    },
    {
      "epoch": 6.371960846226713,
      "grad_norm": 6.313063144683838,
      "learning_rate": 3.785566431603691e-05,
      "loss": 0.5626,
      "step": 20180
    },
    {
      "epoch": 6.3722766024628985,
      "grad_norm": 4.87377405166626,
      "learning_rate": 3.785478721538084e-05,
      "loss": 0.4837,
      "step": 20181
    },
    {
      "epoch": 6.372592358699085,
      "grad_norm": 4.400333404541016,
      "learning_rate": 3.7853910114724766e-05,
      "loss": 0.5211,
      "step": 20182
    },
    {
      "epoch": 6.37290811493527,
      "grad_norm": 4.738274574279785,
      "learning_rate": 3.78530330140687e-05,
      "loss": 0.5637,
      "step": 20183
    },
    {
      "epoch": 6.373223871171455,
      "grad_norm": 5.394996643066406,
      "learning_rate": 3.785215591341262e-05,
      "loss": 0.6005,
      "step": 20184
    },
    {
      "epoch": 6.3735396274076415,
      "grad_norm": 4.99213981628418,
      "learning_rate": 3.7851278812756554e-05,
      "loss": 0.4963,
      "step": 20185
    },
    {
      "epoch": 6.373855383643827,
      "grad_norm": 5.091945648193359,
      "learning_rate": 3.785040171210048e-05,
      "loss": 0.5077,
      "step": 20186
    },
    {
      "epoch": 6.374171139880013,
      "grad_norm": 5.050224781036377,
      "learning_rate": 3.784952461144441e-05,
      "loss": 0.6044,
      "step": 20187
    },
    {
      "epoch": 6.374486896116198,
      "grad_norm": 4.82454776763916,
      "learning_rate": 3.7848647510788335e-05,
      "loss": 0.6533,
      "step": 20188
    },
    {
      "epoch": 6.374802652352384,
      "grad_norm": 4.443662166595459,
      "learning_rate": 3.784777041013227e-05,
      "loss": 0.5123,
      "step": 20189
    },
    {
      "epoch": 6.37511840858857,
      "grad_norm": 5.02208948135376,
      "learning_rate": 3.784689330947619e-05,
      "loss": 0.5147,
      "step": 20190
    },
    {
      "epoch": 6.375434164824755,
      "grad_norm": 4.918454647064209,
      "learning_rate": 3.784601620882013e-05,
      "loss": 0.6169,
      "step": 20191
    },
    {
      "epoch": 6.375749921060941,
      "grad_norm": 4.302917957305908,
      "learning_rate": 3.7845139108164055e-05,
      "loss": 0.5285,
      "step": 20192
    },
    {
      "epoch": 6.3760656772971265,
      "grad_norm": 5.572870254516602,
      "learning_rate": 3.784426200750799e-05,
      "loss": 0.6736,
      "step": 20193
    },
    {
      "epoch": 6.376381433533313,
      "grad_norm": 5.733425140380859,
      "learning_rate": 3.784338490685191e-05,
      "loss": 0.6392,
      "step": 20194
    },
    {
      "epoch": 6.376697189769498,
      "grad_norm": 5.104864120483398,
      "learning_rate": 3.7842507806195844e-05,
      "loss": 0.5069,
      "step": 20195
    },
    {
      "epoch": 6.377012946005683,
      "grad_norm": 5.594916820526123,
      "learning_rate": 3.784163070553977e-05,
      "loss": 0.6678,
      "step": 20196
    },
    {
      "epoch": 6.377328702241869,
      "grad_norm": 4.005313396453857,
      "learning_rate": 3.78407536048837e-05,
      "loss": 0.5248,
      "step": 20197
    },
    {
      "epoch": 6.377644458478055,
      "grad_norm": 4.1358137130737305,
      "learning_rate": 3.7839876504227625e-05,
      "loss": 0.4869,
      "step": 20198
    },
    {
      "epoch": 6.377960214714241,
      "grad_norm": 5.175489902496338,
      "learning_rate": 3.783899940357156e-05,
      "loss": 0.6404,
      "step": 20199
    },
    {
      "epoch": 6.378275970950426,
      "grad_norm": 5.420354843139648,
      "learning_rate": 3.783812230291549e-05,
      "loss": 0.5362,
      "step": 20200
    },
    {
      "epoch": 6.378275970950426,
      "eval_loss": 0.8836385607719421,
      "eval_runtime": 4.5472,
      "eval_samples_per_second": 547.59,
      "eval_steps_per_second": 4.618,
      "step": 20200
    },
    {
      "epoch": 6.3785917271866115,
      "grad_norm": 5.688795566558838,
      "learning_rate": 3.7837245202259413e-05,
      "loss": 0.6696,
      "step": 20201
    },
    {
      "epoch": 6.378907483422798,
      "grad_norm": 6.056127071380615,
      "learning_rate": 3.7836368101603345e-05,
      "loss": 0.5322,
      "step": 20202
    },
    {
      "epoch": 6.379223239658983,
      "grad_norm": 4.529676914215088,
      "learning_rate": 3.783549100094727e-05,
      "loss": 0.5867,
      "step": 20203
    },
    {
      "epoch": 6.379538995895169,
      "grad_norm": 4.44910192489624,
      "learning_rate": 3.78346139002912e-05,
      "loss": 0.5461,
      "step": 20204
    },
    {
      "epoch": 6.3798547521313544,
      "grad_norm": 4.994837760925293,
      "learning_rate": 3.783373679963513e-05,
      "loss": 0.5386,
      "step": 20205
    },
    {
      "epoch": 6.380170508367541,
      "grad_norm": 7.0421223640441895,
      "learning_rate": 3.783285969897906e-05,
      "loss": 0.6023,
      "step": 20206
    },
    {
      "epoch": 6.380486264603726,
      "grad_norm": 6.1839599609375,
      "learning_rate": 3.783198259832298e-05,
      "loss": 0.6506,
      "step": 20207
    },
    {
      "epoch": 6.380802020839911,
      "grad_norm": 5.092772960662842,
      "learning_rate": 3.7831105497666915e-05,
      "loss": 0.5171,
      "step": 20208
    },
    {
      "epoch": 6.381117777076097,
      "grad_norm": 5.934814929962158,
      "learning_rate": 3.783022839701084e-05,
      "loss": 0.644,
      "step": 20209
    },
    {
      "epoch": 6.381433533312283,
      "grad_norm": 5.5996809005737305,
      "learning_rate": 3.782935129635477e-05,
      "loss": 0.5102,
      "step": 20210
    },
    {
      "epoch": 6.381749289548469,
      "grad_norm": 5.784054756164551,
      "learning_rate": 3.7828474195698696e-05,
      "loss": 0.51,
      "step": 20211
    },
    {
      "epoch": 6.382065045784654,
      "grad_norm": 4.416887283325195,
      "learning_rate": 3.782759709504263e-05,
      "loss": 0.6689,
      "step": 20212
    },
    {
      "epoch": 6.38238080202084,
      "grad_norm": 6.026096343994141,
      "learning_rate": 3.782671999438655e-05,
      "loss": 0.6147,
      "step": 20213
    },
    {
      "epoch": 6.382696558257026,
      "grad_norm": 5.996210098266602,
      "learning_rate": 3.7825842893730485e-05,
      "loss": 0.6487,
      "step": 20214
    },
    {
      "epoch": 6.383012314493211,
      "grad_norm": 4.745553493499756,
      "learning_rate": 3.782496579307441e-05,
      "loss": 0.5846,
      "step": 20215
    },
    {
      "epoch": 6.383328070729397,
      "grad_norm": 4.9873366355896,
      "learning_rate": 3.782408869241834e-05,
      "loss": 0.5074,
      "step": 20216
    },
    {
      "epoch": 6.383643826965582,
      "grad_norm": 4.190504550933838,
      "learning_rate": 3.782321159176227e-05,
      "loss": 0.5169,
      "step": 20217
    },
    {
      "epoch": 6.383959583201769,
      "grad_norm": 5.693117618560791,
      "learning_rate": 3.7822334491106205e-05,
      "loss": 0.5374,
      "step": 20218
    },
    {
      "epoch": 6.384275339437954,
      "grad_norm": 5.22679328918457,
      "learning_rate": 3.782145739045013e-05,
      "loss": 0.6892,
      "step": 20219
    },
    {
      "epoch": 6.384591095674139,
      "grad_norm": 4.811350345611572,
      "learning_rate": 3.782058028979406e-05,
      "loss": 0.6131,
      "step": 20220
    },
    {
      "epoch": 6.384906851910325,
      "grad_norm": 4.313206195831299,
      "learning_rate": 3.7819703189137986e-05,
      "loss": 0.548,
      "step": 20221
    },
    {
      "epoch": 6.385222608146511,
      "grad_norm": 5.194201469421387,
      "learning_rate": 3.781882608848192e-05,
      "loss": 0.5715,
      "step": 20222
    },
    {
      "epoch": 6.385538364382697,
      "grad_norm": 4.852151870727539,
      "learning_rate": 3.781794898782585e-05,
      "loss": 0.608,
      "step": 20223
    },
    {
      "epoch": 6.385854120618882,
      "grad_norm": 3.8560781478881836,
      "learning_rate": 3.7817071887169774e-05,
      "loss": 0.4591,
      "step": 20224
    },
    {
      "epoch": 6.386169876855067,
      "grad_norm": 4.895031452178955,
      "learning_rate": 3.7816194786513706e-05,
      "loss": 0.5488,
      "step": 20225
    },
    {
      "epoch": 6.386485633091254,
      "grad_norm": 4.779160976409912,
      "learning_rate": 3.781531768585763e-05,
      "loss": 0.5795,
      "step": 20226
    },
    {
      "epoch": 6.386801389327439,
      "grad_norm": 6.002885818481445,
      "learning_rate": 3.781444058520156e-05,
      "loss": 0.6686,
      "step": 20227
    },
    {
      "epoch": 6.387117145563625,
      "grad_norm": 3.759838581085205,
      "learning_rate": 3.781356348454549e-05,
      "loss": 0.4052,
      "step": 20228
    },
    {
      "epoch": 6.38743290179981,
      "grad_norm": 5.125711917877197,
      "learning_rate": 3.781268638388942e-05,
      "loss": 0.5851,
      "step": 20229
    },
    {
      "epoch": 6.387748658035997,
      "grad_norm": 4.00679874420166,
      "learning_rate": 3.7811809283233344e-05,
      "loss": 0.4627,
      "step": 20230
    },
    {
      "epoch": 6.388064414272182,
      "grad_norm": 5.072666645050049,
      "learning_rate": 3.7810932182577276e-05,
      "loss": 0.6,
      "step": 20231
    },
    {
      "epoch": 6.388380170508367,
      "grad_norm": 5.908910274505615,
      "learning_rate": 3.78100550819212e-05,
      "loss": 0.6244,
      "step": 20232
    },
    {
      "epoch": 6.388695926744553,
      "grad_norm": 6.602437496185303,
      "learning_rate": 3.780917798126513e-05,
      "loss": 0.5056,
      "step": 20233
    },
    {
      "epoch": 6.389011682980739,
      "grad_norm": 5.180014133453369,
      "learning_rate": 3.780830088060906e-05,
      "loss": 0.5925,
      "step": 20234
    },
    {
      "epoch": 6.389327439216925,
      "grad_norm": 4.624292373657227,
      "learning_rate": 3.780742377995299e-05,
      "loss": 0.5403,
      "step": 20235
    },
    {
      "epoch": 6.38964319545311,
      "grad_norm": 4.762679100036621,
      "learning_rate": 3.7806546679296914e-05,
      "loss": 0.4895,
      "step": 20236
    },
    {
      "epoch": 6.389958951689296,
      "grad_norm": 6.198777198791504,
      "learning_rate": 3.7805669578640846e-05,
      "loss": 0.6577,
      "step": 20237
    },
    {
      "epoch": 6.390274707925482,
      "grad_norm": 5.281411647796631,
      "learning_rate": 3.780479247798478e-05,
      "loss": 0.5794,
      "step": 20238
    },
    {
      "epoch": 6.390590464161667,
      "grad_norm": 6.011589527130127,
      "learning_rate": 3.78039153773287e-05,
      "loss": 0.5848,
      "step": 20239
    },
    {
      "epoch": 6.390906220397853,
      "grad_norm": 4.910301685333252,
      "learning_rate": 3.7803038276672634e-05,
      "loss": 0.5406,
      "step": 20240
    },
    {
      "epoch": 6.391221976634038,
      "grad_norm": 4.893643379211426,
      "learning_rate": 3.780216117601656e-05,
      "loss": 0.5918,
      "step": 20241
    },
    {
      "epoch": 6.3915377328702245,
      "grad_norm": 4.869718551635742,
      "learning_rate": 3.780128407536049e-05,
      "loss": 0.4769,
      "step": 20242
    },
    {
      "epoch": 6.39185348910641,
      "grad_norm": 4.080388069152832,
      "learning_rate": 3.7800406974704415e-05,
      "loss": 0.473,
      "step": 20243
    },
    {
      "epoch": 6.392169245342595,
      "grad_norm": 6.362020492553711,
      "learning_rate": 3.779952987404835e-05,
      "loss": 0.562,
      "step": 20244
    },
    {
      "epoch": 6.392485001578781,
      "grad_norm": 5.424895286560059,
      "learning_rate": 3.779865277339228e-05,
      "loss": 0.4961,
      "step": 20245
    },
    {
      "epoch": 6.392800757814967,
      "grad_norm": 4.477762222290039,
      "learning_rate": 3.779777567273621e-05,
      "loss": 0.4723,
      "step": 20246
    },
    {
      "epoch": 6.393116514051153,
      "grad_norm": 6.468374729156494,
      "learning_rate": 3.7796898572080135e-05,
      "loss": 0.5654,
      "step": 20247
    },
    {
      "epoch": 6.393432270287338,
      "grad_norm": 5.915870666503906,
      "learning_rate": 3.779602147142407e-05,
      "loss": 0.4164,
      "step": 20248
    },
    {
      "epoch": 6.393748026523524,
      "grad_norm": 5.563703536987305,
      "learning_rate": 3.779514437076799e-05,
      "loss": 0.5246,
      "step": 20249
    },
    {
      "epoch": 6.3940637827597095,
      "grad_norm": 5.46156120300293,
      "learning_rate": 3.7794267270111924e-05,
      "loss": 0.5902,
      "step": 20250
    },
    {
      "epoch": 6.394379538995895,
      "grad_norm": 4.621615409851074,
      "learning_rate": 3.779339016945585e-05,
      "loss": 0.5585,
      "step": 20251
    },
    {
      "epoch": 6.394695295232081,
      "grad_norm": 5.433607578277588,
      "learning_rate": 3.779251306879978e-05,
      "loss": 0.5819,
      "step": 20252
    },
    {
      "epoch": 6.395011051468266,
      "grad_norm": 5.215586185455322,
      "learning_rate": 3.7791635968143705e-05,
      "loss": 0.5384,
      "step": 20253
    },
    {
      "epoch": 6.3953268077044525,
      "grad_norm": 4.293199062347412,
      "learning_rate": 3.779075886748764e-05,
      "loss": 0.5474,
      "step": 20254
    },
    {
      "epoch": 6.395642563940638,
      "grad_norm": 4.6398024559021,
      "learning_rate": 3.778988176683156e-05,
      "loss": 0.4624,
      "step": 20255
    },
    {
      "epoch": 6.395958320176823,
      "grad_norm": 4.832080364227295,
      "learning_rate": 3.778900466617549e-05,
      "loss": 0.5208,
      "step": 20256
    },
    {
      "epoch": 6.396274076413009,
      "grad_norm": 5.8649210929870605,
      "learning_rate": 3.778812756551942e-05,
      "loss": 0.5161,
      "step": 20257
    },
    {
      "epoch": 6.396589832649195,
      "grad_norm": 5.682287693023682,
      "learning_rate": 3.778725046486335e-05,
      "loss": 0.5249,
      "step": 20258
    },
    {
      "epoch": 6.396905588885381,
      "grad_norm": 6.1044745445251465,
      "learning_rate": 3.7786373364207275e-05,
      "loss": 0.4953,
      "step": 20259
    },
    {
      "epoch": 6.397221345121566,
      "grad_norm": 4.283268451690674,
      "learning_rate": 3.7785496263551207e-05,
      "loss": 0.5303,
      "step": 20260
    },
    {
      "epoch": 6.397537101357752,
      "grad_norm": 4.515781879425049,
      "learning_rate": 3.778461916289514e-05,
      "loss": 0.5091,
      "step": 20261
    },
    {
      "epoch": 6.3978528575939375,
      "grad_norm": 4.966569900512695,
      "learning_rate": 3.778374206223906e-05,
      "loss": 0.6547,
      "step": 20262
    },
    {
      "epoch": 6.398168613830123,
      "grad_norm": 4.668441295623779,
      "learning_rate": 3.7782864961582995e-05,
      "loss": 0.5522,
      "step": 20263
    },
    {
      "epoch": 6.398484370066309,
      "grad_norm": 4.841917991638184,
      "learning_rate": 3.778198786092692e-05,
      "loss": 0.5696,
      "step": 20264
    },
    {
      "epoch": 6.398800126302494,
      "grad_norm": 5.292201995849609,
      "learning_rate": 3.778111076027085e-05,
      "loss": 0.5891,
      "step": 20265
    },
    {
      "epoch": 6.3991158825386805,
      "grad_norm": 4.908858776092529,
      "learning_rate": 3.7780233659614776e-05,
      "loss": 0.5044,
      "step": 20266
    },
    {
      "epoch": 6.399431638774866,
      "grad_norm": 4.629379749298096,
      "learning_rate": 3.777935655895871e-05,
      "loss": 0.5051,
      "step": 20267
    },
    {
      "epoch": 6.399747395011051,
      "grad_norm": 5.3423590660095215,
      "learning_rate": 3.777847945830263e-05,
      "loss": 0.5995,
      "step": 20268
    },
    {
      "epoch": 6.400063151247237,
      "grad_norm": 4.831096649169922,
      "learning_rate": 3.7777602357646565e-05,
      "loss": 0.5767,
      "step": 20269
    },
    {
      "epoch": 6.4003789074834225,
      "grad_norm": 4.466212272644043,
      "learning_rate": 3.777672525699049e-05,
      "loss": 0.6303,
      "step": 20270
    },
    {
      "epoch": 6.400694663719609,
      "grad_norm": 4.357978343963623,
      "learning_rate": 3.777584815633442e-05,
      "loss": 0.5457,
      "step": 20271
    },
    {
      "epoch": 6.401010419955794,
      "grad_norm": 4.546121120452881,
      "learning_rate": 3.7774971055678346e-05,
      "loss": 0.6241,
      "step": 20272
    },
    {
      "epoch": 6.40132617619198,
      "grad_norm": 4.290966987609863,
      "learning_rate": 3.7774093955022285e-05,
      "loss": 0.5944,
      "step": 20273
    },
    {
      "epoch": 6.4016419324281655,
      "grad_norm": 4.31494140625,
      "learning_rate": 3.777321685436621e-05,
      "loss": 0.548,
      "step": 20274
    },
    {
      "epoch": 6.401957688664351,
      "grad_norm": 4.874886989593506,
      "learning_rate": 3.777233975371014e-05,
      "loss": 0.5205,
      "step": 20275
    },
    {
      "epoch": 6.402273444900537,
      "grad_norm": 5.725073337554932,
      "learning_rate": 3.7771462653054066e-05,
      "loss": 0.6195,
      "step": 20276
    },
    {
      "epoch": 6.402589201136722,
      "grad_norm": 4.560368537902832,
      "learning_rate": 3.7770585552398e-05,
      "loss": 0.6371,
      "step": 20277
    },
    {
      "epoch": 6.402904957372908,
      "grad_norm": 4.12470006942749,
      "learning_rate": 3.776970845174192e-05,
      "loss": 0.4705,
      "step": 20278
    },
    {
      "epoch": 6.403220713609094,
      "grad_norm": 3.453202724456787,
      "learning_rate": 3.7768831351085854e-05,
      "loss": 0.4797,
      "step": 20279
    },
    {
      "epoch": 6.40353646984528,
      "grad_norm": 5.205002784729004,
      "learning_rate": 3.776795425042978e-05,
      "loss": 0.5326,
      "step": 20280
    },
    {
      "epoch": 6.403852226081465,
      "grad_norm": 4.9384942054748535,
      "learning_rate": 3.776707714977371e-05,
      "loss": 0.5496,
      "step": 20281
    },
    {
      "epoch": 6.4041679823176505,
      "grad_norm": 4.903839588165283,
      "learning_rate": 3.776620004911764e-05,
      "loss": 0.5063,
      "step": 20282
    },
    {
      "epoch": 6.404483738553837,
      "grad_norm": 5.589667320251465,
      "learning_rate": 3.776532294846157e-05,
      "loss": 0.5892,
      "step": 20283
    },
    {
      "epoch": 6.404799494790022,
      "grad_norm": 5.174784183502197,
      "learning_rate": 3.77644458478055e-05,
      "loss": 0.5546,
      "step": 20284
    },
    {
      "epoch": 6.405115251026208,
      "grad_norm": 4.948694229125977,
      "learning_rate": 3.7763568747149424e-05,
      "loss": 0.5838,
      "step": 20285
    },
    {
      "epoch": 6.4054310072623935,
      "grad_norm": 4.355710506439209,
      "learning_rate": 3.7762691646493356e-05,
      "loss": 0.597,
      "step": 20286
    },
    {
      "epoch": 6.405746763498579,
      "grad_norm": 4.838066101074219,
      "learning_rate": 3.776181454583728e-05,
      "loss": 0.5707,
      "step": 20287
    },
    {
      "epoch": 6.406062519734765,
      "grad_norm": 4.9894232749938965,
      "learning_rate": 3.776093744518121e-05,
      "loss": 0.5425,
      "step": 20288
    },
    {
      "epoch": 6.40637827597095,
      "grad_norm": 4.861741542816162,
      "learning_rate": 3.776006034452514e-05,
      "loss": 0.4857,
      "step": 20289
    },
    {
      "epoch": 6.406694032207136,
      "grad_norm": 4.465909957885742,
      "learning_rate": 3.775918324386907e-05,
      "loss": 0.5443,
      "step": 20290
    },
    {
      "epoch": 6.407009788443322,
      "grad_norm": 4.959743022918701,
      "learning_rate": 3.7758306143212994e-05,
      "loss": 0.5344,
      "step": 20291
    },
    {
      "epoch": 6.407325544679507,
      "grad_norm": 5.22138786315918,
      "learning_rate": 3.7757429042556925e-05,
      "loss": 0.5697,
      "step": 20292
    },
    {
      "epoch": 6.407641300915693,
      "grad_norm": 5.212815761566162,
      "learning_rate": 3.775655194190085e-05,
      "loss": 0.6124,
      "step": 20293
    },
    {
      "epoch": 6.4079570571518785,
      "grad_norm": 5.427818298339844,
      "learning_rate": 3.775567484124478e-05,
      "loss": 0.5576,
      "step": 20294
    },
    {
      "epoch": 6.408272813388065,
      "grad_norm": 4.1251020431518555,
      "learning_rate": 3.775479774058871e-05,
      "loss": 0.4445,
      "step": 20295
    },
    {
      "epoch": 6.40858856962425,
      "grad_norm": 4.725343704223633,
      "learning_rate": 3.775392063993264e-05,
      "loss": 0.4564,
      "step": 20296
    },
    {
      "epoch": 6.408904325860436,
      "grad_norm": 4.652590751647949,
      "learning_rate": 3.7753043539276564e-05,
      "loss": 0.558,
      "step": 20297
    },
    {
      "epoch": 6.409220082096621,
      "grad_norm": 4.669347763061523,
      "learning_rate": 3.7752166438620495e-05,
      "loss": 0.4432,
      "step": 20298
    },
    {
      "epoch": 6.409535838332807,
      "grad_norm": 4.115769863128662,
      "learning_rate": 3.775128933796443e-05,
      "loss": 0.5362,
      "step": 20299
    },
    {
      "epoch": 6.409851594568993,
      "grad_norm": 4.475729942321777,
      "learning_rate": 3.775041223730836e-05,
      "loss": 0.6217,
      "step": 20300
    },
    {
      "epoch": 6.410167350805178,
      "grad_norm": 4.478827953338623,
      "learning_rate": 3.7749535136652284e-05,
      "loss": 0.4327,
      "step": 20301
    },
    {
      "epoch": 6.410483107041364,
      "grad_norm": 5.763742446899414,
      "learning_rate": 3.7748658035996215e-05,
      "loss": 0.585,
      "step": 20302
    },
    {
      "epoch": 6.41079886327755,
      "grad_norm": 5.08411979675293,
      "learning_rate": 3.774778093534014e-05,
      "loss": 0.6013,
      "step": 20303
    },
    {
      "epoch": 6.411114619513736,
      "grad_norm": 4.74250602722168,
      "learning_rate": 3.774690383468407e-05,
      "loss": 0.5986,
      "step": 20304
    },
    {
      "epoch": 6.411430375749921,
      "grad_norm": 5.536261558532715,
      "learning_rate": 3.7746026734028003e-05,
      "loss": 0.573,
      "step": 20305
    },
    {
      "epoch": 6.411746131986106,
      "grad_norm": 4.487751007080078,
      "learning_rate": 3.774514963337193e-05,
      "loss": 0.5971,
      "step": 20306
    },
    {
      "epoch": 6.412061888222293,
      "grad_norm": 4.963351249694824,
      "learning_rate": 3.774427253271586e-05,
      "loss": 0.5965,
      "step": 20307
    },
    {
      "epoch": 6.412377644458478,
      "grad_norm": 4.856590747833252,
      "learning_rate": 3.7743395432059785e-05,
      "loss": 0.6001,
      "step": 20308
    },
    {
      "epoch": 6.412693400694664,
      "grad_norm": 5.050963878631592,
      "learning_rate": 3.774251833140372e-05,
      "loss": 0.4763,
      "step": 20309
    },
    {
      "epoch": 6.413009156930849,
      "grad_norm": 5.962681293487549,
      "learning_rate": 3.774164123074764e-05,
      "loss": 0.6008,
      "step": 20310
    },
    {
      "epoch": 6.413324913167035,
      "grad_norm": 5.940154075622559,
      "learning_rate": 3.774076413009157e-05,
      "loss": 0.5328,
      "step": 20311
    },
    {
      "epoch": 6.413640669403221,
      "grad_norm": 6.157742023468018,
      "learning_rate": 3.77398870294355e-05,
      "loss": 0.7701,
      "step": 20312
    },
    {
      "epoch": 6.413956425639406,
      "grad_norm": 5.593417167663574,
      "learning_rate": 3.773900992877943e-05,
      "loss": 0.5276,
      "step": 20313
    },
    {
      "epoch": 6.414272181875592,
      "grad_norm": 4.525600433349609,
      "learning_rate": 3.7738132828123355e-05,
      "loss": 0.5494,
      "step": 20314
    },
    {
      "epoch": 6.414587938111778,
      "grad_norm": 6.121950149536133,
      "learning_rate": 3.7737255727467286e-05,
      "loss": 0.5627,
      "step": 20315
    },
    {
      "epoch": 6.414903694347963,
      "grad_norm": 4.02796745300293,
      "learning_rate": 3.773637862681121e-05,
      "loss": 0.4731,
      "step": 20316
    },
    {
      "epoch": 6.415219450584149,
      "grad_norm": 4.93748140335083,
      "learning_rate": 3.773550152615514e-05,
      "loss": 0.5857,
      "step": 20317
    },
    {
      "epoch": 6.415535206820334,
      "grad_norm": 5.446627140045166,
      "learning_rate": 3.773462442549907e-05,
      "loss": 0.505,
      "step": 20318
    },
    {
      "epoch": 6.415850963056521,
      "grad_norm": 6.935938358306885,
      "learning_rate": 3.7733747324843e-05,
      "loss": 0.5968,
      "step": 20319
    },
    {
      "epoch": 6.416166719292706,
      "grad_norm": 4.920030117034912,
      "learning_rate": 3.773287022418693e-05,
      "loss": 0.5236,
      "step": 20320
    },
    {
      "epoch": 6.416482475528892,
      "grad_norm": 4.735947132110596,
      "learning_rate": 3.7731993123530856e-05,
      "loss": 0.5584,
      "step": 20321
    },
    {
      "epoch": 6.416798231765077,
      "grad_norm": 7.574570178985596,
      "learning_rate": 3.773111602287479e-05,
      "loss": 0.6641,
      "step": 20322
    },
    {
      "epoch": 6.417113988001263,
      "grad_norm": 4.572808742523193,
      "learning_rate": 3.773023892221871e-05,
      "loss": 0.5524,
      "step": 20323
    },
    {
      "epoch": 6.417429744237449,
      "grad_norm": 6.021241664886475,
      "learning_rate": 3.7729361821562644e-05,
      "loss": 0.4945,
      "step": 20324
    },
    {
      "epoch": 6.417745500473634,
      "grad_norm": 4.673056125640869,
      "learning_rate": 3.772848472090657e-05,
      "loss": 0.445,
      "step": 20325
    },
    {
      "epoch": 6.41806125670982,
      "grad_norm": 5.184945583343506,
      "learning_rate": 3.77276076202505e-05,
      "loss": 0.5792,
      "step": 20326
    },
    {
      "epoch": 6.418377012946006,
      "grad_norm": 4.084338665008545,
      "learning_rate": 3.772673051959443e-05,
      "loss": 0.6001,
      "step": 20327
    },
    {
      "epoch": 6.418692769182192,
      "grad_norm": 3.6483120918273926,
      "learning_rate": 3.7725853418938364e-05,
      "loss": 0.515,
      "step": 20328
    },
    {
      "epoch": 6.419008525418377,
      "grad_norm": 4.04671049118042,
      "learning_rate": 3.772497631828229e-05,
      "loss": 0.5141,
      "step": 20329
    },
    {
      "epoch": 6.419324281654562,
      "grad_norm": 4.56057596206665,
      "learning_rate": 3.772409921762622e-05,
      "loss": 0.545,
      "step": 20330
    },
    {
      "epoch": 6.4196400378907486,
      "grad_norm": 5.891011714935303,
      "learning_rate": 3.7723222116970146e-05,
      "loss": 0.5567,
      "step": 20331
    },
    {
      "epoch": 6.419955794126934,
      "grad_norm": 6.034372329711914,
      "learning_rate": 3.772234501631408e-05,
      "loss": 0.6188,
      "step": 20332
    },
    {
      "epoch": 6.42027155036312,
      "grad_norm": 4.353086948394775,
      "learning_rate": 3.7721467915658e-05,
      "loss": 0.5338,
      "step": 20333
    },
    {
      "epoch": 6.420587306599305,
      "grad_norm": 4.596810340881348,
      "learning_rate": 3.7720590815001934e-05,
      "loss": 0.5452,
      "step": 20334
    },
    {
      "epoch": 6.420903062835491,
      "grad_norm": 4.9087677001953125,
      "learning_rate": 3.771971371434586e-05,
      "loss": 0.6564,
      "step": 20335
    },
    {
      "epoch": 6.421218819071677,
      "grad_norm": 4.465910911560059,
      "learning_rate": 3.771883661368979e-05,
      "loss": 0.5764,
      "step": 20336
    },
    {
      "epoch": 6.421534575307862,
      "grad_norm": 4.625697612762451,
      "learning_rate": 3.7717959513033716e-05,
      "loss": 0.5327,
      "step": 20337
    },
    {
      "epoch": 6.421850331544048,
      "grad_norm": 5.328892230987549,
      "learning_rate": 3.771708241237765e-05,
      "loss": 0.5456,
      "step": 20338
    },
    {
      "epoch": 6.422166087780234,
      "grad_norm": 4.555005073547363,
      "learning_rate": 3.771620531172157e-05,
      "loss": 0.5002,
      "step": 20339
    },
    {
      "epoch": 6.422481844016419,
      "grad_norm": 4.233161926269531,
      "learning_rate": 3.7715328211065504e-05,
      "loss": 0.4995,
      "step": 20340
    },
    {
      "epoch": 6.422797600252605,
      "grad_norm": 4.204377174377441,
      "learning_rate": 3.7714451110409436e-05,
      "loss": 0.488,
      "step": 20341
    },
    {
      "epoch": 6.42311335648879,
      "grad_norm": 5.186302185058594,
      "learning_rate": 3.771357400975336e-05,
      "loss": 0.5502,
      "step": 20342
    },
    {
      "epoch": 6.4234291127249765,
      "grad_norm": 6.1759352684021,
      "learning_rate": 3.771269690909729e-05,
      "loss": 0.5519,
      "step": 20343
    },
    {
      "epoch": 6.423744868961162,
      "grad_norm": 5.080042362213135,
      "learning_rate": 3.771181980844122e-05,
      "loss": 0.5447,
      "step": 20344
    },
    {
      "epoch": 6.424060625197348,
      "grad_norm": 5.562732696533203,
      "learning_rate": 3.771094270778515e-05,
      "loss": 0.6086,
      "step": 20345
    },
    {
      "epoch": 6.424376381433533,
      "grad_norm": 4.508487224578857,
      "learning_rate": 3.7710065607129074e-05,
      "loss": 0.4669,
      "step": 20346
    },
    {
      "epoch": 6.424692137669719,
      "grad_norm": 5.792731285095215,
      "learning_rate": 3.7709188506473005e-05,
      "loss": 0.5743,
      "step": 20347
    },
    {
      "epoch": 6.425007893905905,
      "grad_norm": 5.743950366973877,
      "learning_rate": 3.770831140581693e-05,
      "loss": 0.5465,
      "step": 20348
    },
    {
      "epoch": 6.42532365014209,
      "grad_norm": 5.424438953399658,
      "learning_rate": 3.770743430516086e-05,
      "loss": 0.5793,
      "step": 20349
    },
    {
      "epoch": 6.425639406378276,
      "grad_norm": 4.858100414276123,
      "learning_rate": 3.770655720450479e-05,
      "loss": 0.5724,
      "step": 20350
    },
    {
      "epoch": 6.4259551626144615,
      "grad_norm": 4.217836856842041,
      "learning_rate": 3.770568010384872e-05,
      "loss": 0.4417,
      "step": 20351
    },
    {
      "epoch": 6.426270918850648,
      "grad_norm": 5.063295841217041,
      "learning_rate": 3.7704803003192643e-05,
      "loss": 0.5994,
      "step": 20352
    },
    {
      "epoch": 6.426586675086833,
      "grad_norm": 5.2920074462890625,
      "learning_rate": 3.7703925902536575e-05,
      "loss": 0.6337,
      "step": 20353
    },
    {
      "epoch": 6.426902431323018,
      "grad_norm": 5.333672046661377,
      "learning_rate": 3.770304880188051e-05,
      "loss": 0.5791,
      "step": 20354
    },
    {
      "epoch": 6.4272181875592045,
      "grad_norm": 4.292544841766357,
      "learning_rate": 3.770217170122444e-05,
      "loss": 0.4997,
      "step": 20355
    },
    {
      "epoch": 6.42753394379539,
      "grad_norm": 4.159290790557861,
      "learning_rate": 3.7701294600568363e-05,
      "loss": 0.5594,
      "step": 20356
    },
    {
      "epoch": 6.427849700031576,
      "grad_norm": 6.350479602813721,
      "learning_rate": 3.7700417499912295e-05,
      "loss": 0.5075,
      "step": 20357
    },
    {
      "epoch": 6.428165456267761,
      "grad_norm": 5.279515743255615,
      "learning_rate": 3.769954039925622e-05,
      "loss": 0.531,
      "step": 20358
    },
    {
      "epoch": 6.4284812125039466,
      "grad_norm": 5.669064521789551,
      "learning_rate": 3.769866329860015e-05,
      "loss": 0.5146,
      "step": 20359
    },
    {
      "epoch": 6.428796968740133,
      "grad_norm": 4.432694911956787,
      "learning_rate": 3.7697786197944077e-05,
      "loss": 0.4071,
      "step": 20360
    },
    {
      "epoch": 6.429112724976318,
      "grad_norm": 5.40976095199585,
      "learning_rate": 3.769690909728801e-05,
      "loss": 0.5893,
      "step": 20361
    },
    {
      "epoch": 6.429428481212504,
      "grad_norm": 5.461121082305908,
      "learning_rate": 3.769603199663193e-05,
      "loss": 0.5119,
      "step": 20362
    },
    {
      "epoch": 6.4297442374486895,
      "grad_norm": 6.090808868408203,
      "learning_rate": 3.7695154895975865e-05,
      "loss": 0.6984,
      "step": 20363
    },
    {
      "epoch": 6.430059993684876,
      "grad_norm": 5.318385601043701,
      "learning_rate": 3.7694277795319797e-05,
      "loss": 0.6752,
      "step": 20364
    },
    {
      "epoch": 6.430375749921061,
      "grad_norm": 4.611861228942871,
      "learning_rate": 3.769340069466372e-05,
      "loss": 0.4636,
      "step": 20365
    },
    {
      "epoch": 6.430691506157246,
      "grad_norm": 5.453535556793213,
      "learning_rate": 3.769252359400765e-05,
      "loss": 0.605,
      "step": 20366
    },
    {
      "epoch": 6.4310072623934325,
      "grad_norm": 5.247544765472412,
      "learning_rate": 3.769164649335158e-05,
      "loss": 0.5718,
      "step": 20367
    },
    {
      "epoch": 6.431323018629618,
      "grad_norm": 5.990452766418457,
      "learning_rate": 3.769076939269551e-05,
      "loss": 0.6512,
      "step": 20368
    },
    {
      "epoch": 6.431638774865804,
      "grad_norm": 5.086580276489258,
      "learning_rate": 3.7689892292039435e-05,
      "loss": 0.5228,
      "step": 20369
    },
    {
      "epoch": 6.431954531101989,
      "grad_norm": 5.303379535675049,
      "learning_rate": 3.7689015191383366e-05,
      "loss": 0.5806,
      "step": 20370
    },
    {
      "epoch": 6.4322702873381745,
      "grad_norm": 4.67767333984375,
      "learning_rate": 3.768813809072729e-05,
      "loss": 0.6156,
      "step": 20371
    },
    {
      "epoch": 6.432586043574361,
      "grad_norm": 4.015589714050293,
      "learning_rate": 3.768726099007122e-05,
      "loss": 0.4868,
      "step": 20372
    },
    {
      "epoch": 6.432901799810546,
      "grad_norm": 5.926098823547363,
      "learning_rate": 3.768638388941515e-05,
      "loss": 0.5889,
      "step": 20373
    },
    {
      "epoch": 6.433217556046732,
      "grad_norm": 4.053974151611328,
      "learning_rate": 3.768550678875908e-05,
      "loss": 0.4287,
      "step": 20374
    },
    {
      "epoch": 6.4335333122829175,
      "grad_norm": 4.718062877655029,
      "learning_rate": 3.7684629688103004e-05,
      "loss": 0.5613,
      "step": 20375
    },
    {
      "epoch": 6.433849068519104,
      "grad_norm": 5.797745227813721,
      "learning_rate": 3.7683752587446936e-05,
      "loss": 0.6096,
      "step": 20376
    },
    {
      "epoch": 6.434164824755289,
      "grad_norm": 5.175730228424072,
      "learning_rate": 3.768287548679086e-05,
      "loss": 0.4825,
      "step": 20377
    },
    {
      "epoch": 6.434480580991474,
      "grad_norm": 4.315356731414795,
      "learning_rate": 3.768199838613479e-05,
      "loss": 0.5505,
      "step": 20378
    },
    {
      "epoch": 6.43479633722766,
      "grad_norm": 4.232666969299316,
      "learning_rate": 3.7681121285478724e-05,
      "loss": 0.5875,
      "step": 20379
    },
    {
      "epoch": 6.435112093463846,
      "grad_norm": 5.030745029449463,
      "learning_rate": 3.768024418482265e-05,
      "loss": 0.5794,
      "step": 20380
    },
    {
      "epoch": 6.435427849700032,
      "grad_norm": 5.9591803550720215,
      "learning_rate": 3.767936708416658e-05,
      "loss": 0.5136,
      "step": 20381
    },
    {
      "epoch": 6.435743605936217,
      "grad_norm": 5.6495184898376465,
      "learning_rate": 3.767848998351051e-05,
      "loss": 0.6577,
      "step": 20382
    },
    {
      "epoch": 6.4360593621724025,
      "grad_norm": 4.640996932983398,
      "learning_rate": 3.767761288285444e-05,
      "loss": 0.5356,
      "step": 20383
    },
    {
      "epoch": 6.436375118408589,
      "grad_norm": 4.496543884277344,
      "learning_rate": 3.767673578219837e-05,
      "loss": 0.5222,
      "step": 20384
    },
    {
      "epoch": 6.436690874644774,
      "grad_norm": 4.325969219207764,
      "learning_rate": 3.76758586815423e-05,
      "loss": 0.5593,
      "step": 20385
    },
    {
      "epoch": 6.43700663088096,
      "grad_norm": 5.06735897064209,
      "learning_rate": 3.7674981580886226e-05,
      "loss": 0.5447,
      "step": 20386
    },
    {
      "epoch": 6.437322387117145,
      "grad_norm": 5.696120262145996,
      "learning_rate": 3.767410448023016e-05,
      "loss": 0.6549,
      "step": 20387
    },
    {
      "epoch": 6.437638143353332,
      "grad_norm": 4.589387893676758,
      "learning_rate": 3.767322737957408e-05,
      "loss": 0.5897,
      "step": 20388
    },
    {
      "epoch": 6.437953899589517,
      "grad_norm": 4.3743743896484375,
      "learning_rate": 3.7672350278918014e-05,
      "loss": 0.5574,
      "step": 20389
    },
    {
      "epoch": 6.438269655825702,
      "grad_norm": 5.312495231628418,
      "learning_rate": 3.767147317826194e-05,
      "loss": 0.5868,
      "step": 20390
    },
    {
      "epoch": 6.438585412061888,
      "grad_norm": 5.744365692138672,
      "learning_rate": 3.767059607760587e-05,
      "loss": 0.5921,
      "step": 20391
    },
    {
      "epoch": 6.438901168298074,
      "grad_norm": 4.164633274078369,
      "learning_rate": 3.7669718976949796e-05,
      "loss": 0.5336,
      "step": 20392
    },
    {
      "epoch": 6.43921692453426,
      "grad_norm": 4.34999942779541,
      "learning_rate": 3.766884187629373e-05,
      "loss": 0.6255,
      "step": 20393
    },
    {
      "epoch": 6.439532680770445,
      "grad_norm": 4.908547401428223,
      "learning_rate": 3.766796477563765e-05,
      "loss": 0.5948,
      "step": 20394
    },
    {
      "epoch": 6.439848437006631,
      "grad_norm": 4.922901630401611,
      "learning_rate": 3.7667087674981584e-05,
      "loss": 0.5131,
      "step": 20395
    },
    {
      "epoch": 6.440164193242817,
      "grad_norm": 4.573385715484619,
      "learning_rate": 3.766621057432551e-05,
      "loss": 0.5594,
      "step": 20396
    },
    {
      "epoch": 6.440479949479002,
      "grad_norm": 4.698751926422119,
      "learning_rate": 3.766533347366944e-05,
      "loss": 0.5476,
      "step": 20397
    },
    {
      "epoch": 6.440795705715188,
      "grad_norm": 4.039875507354736,
      "learning_rate": 3.7664456373013365e-05,
      "loss": 0.5171,
      "step": 20398
    },
    {
      "epoch": 6.441111461951373,
      "grad_norm": 4.955499172210693,
      "learning_rate": 3.76635792723573e-05,
      "loss": 0.5353,
      "step": 20399
    },
    {
      "epoch": 6.44142721818756,
      "grad_norm": 3.629940986633301,
      "learning_rate": 3.766270217170122e-05,
      "loss": 0.4516,
      "step": 20400
    },
    {
      "epoch": 6.44142721818756,
      "eval_loss": 0.8579589128494263,
      "eval_runtime": 4.5445,
      "eval_samples_per_second": 547.913,
      "eval_steps_per_second": 4.621,
      "step": 20400
    },
    {
      "epoch": 6.441742974423745,
      "grad_norm": 4.589565753936768,
      "learning_rate": 3.7661825071045154e-05,
      "loss": 0.5069,
      "step": 20401
    },
    {
      "epoch": 6.44205873065993,
      "grad_norm": 4.779140949249268,
      "learning_rate": 3.7660947970389085e-05,
      "loss": 0.5971,
      "step": 20402
    },
    {
      "epoch": 6.442374486896116,
      "grad_norm": 5.1559271812438965,
      "learning_rate": 3.766007086973301e-05,
      "loss": 0.5891,
      "step": 20403
    },
    {
      "epoch": 6.442690243132302,
      "grad_norm": 3.859161853790283,
      "learning_rate": 3.765919376907694e-05,
      "loss": 0.5125,
      "step": 20404
    },
    {
      "epoch": 6.443005999368488,
      "grad_norm": 4.7667036056518555,
      "learning_rate": 3.765831666842087e-05,
      "loss": 0.5085,
      "step": 20405
    },
    {
      "epoch": 6.443321755604673,
      "grad_norm": 4.5870232582092285,
      "learning_rate": 3.76574395677648e-05,
      "loss": 0.5259,
      "step": 20406
    },
    {
      "epoch": 6.443637511840858,
      "grad_norm": 4.1183390617370605,
      "learning_rate": 3.765656246710872e-05,
      "loss": 0.4706,
      "step": 20407
    },
    {
      "epoch": 6.443953268077045,
      "grad_norm": 4.764719486236572,
      "learning_rate": 3.7655685366452655e-05,
      "loss": 0.5963,
      "step": 20408
    },
    {
      "epoch": 6.44426902431323,
      "grad_norm": 5.920827388763428,
      "learning_rate": 3.765480826579659e-05,
      "loss": 0.5399,
      "step": 20409
    },
    {
      "epoch": 6.444584780549416,
      "grad_norm": 4.6814494132995605,
      "learning_rate": 3.765393116514052e-05,
      "loss": 0.5685,
      "step": 20410
    },
    {
      "epoch": 6.444900536785601,
      "grad_norm": 4.110385894775391,
      "learning_rate": 3.765305406448444e-05,
      "loss": 0.4685,
      "step": 20411
    },
    {
      "epoch": 6.445216293021788,
      "grad_norm": 4.077451229095459,
      "learning_rate": 3.7652176963828375e-05,
      "loss": 0.5156,
      "step": 20412
    },
    {
      "epoch": 6.445532049257973,
      "grad_norm": 4.591845989227295,
      "learning_rate": 3.76512998631723e-05,
      "loss": 0.5776,
      "step": 20413
    },
    {
      "epoch": 6.445847805494158,
      "grad_norm": 7.5257568359375,
      "learning_rate": 3.765042276251623e-05,
      "loss": 0.5583,
      "step": 20414
    },
    {
      "epoch": 6.446163561730344,
      "grad_norm": 6.973613262176514,
      "learning_rate": 3.7649545661860156e-05,
      "loss": 0.5564,
      "step": 20415
    },
    {
      "epoch": 6.44647931796653,
      "grad_norm": 4.071589469909668,
      "learning_rate": 3.764866856120409e-05,
      "loss": 0.5,
      "step": 20416
    },
    {
      "epoch": 6.446795074202716,
      "grad_norm": 5.052629470825195,
      "learning_rate": 3.764779146054801e-05,
      "loss": 0.5003,
      "step": 20417
    },
    {
      "epoch": 6.447110830438901,
      "grad_norm": 4.104019641876221,
      "learning_rate": 3.7646914359891945e-05,
      "loss": 0.4798,
      "step": 20418
    },
    {
      "epoch": 6.447426586675087,
      "grad_norm": 6.6772260665893555,
      "learning_rate": 3.764603725923587e-05,
      "loss": 0.5486,
      "step": 20419
    },
    {
      "epoch": 6.447742342911273,
      "grad_norm": 4.864139080047607,
      "learning_rate": 3.76451601585798e-05,
      "loss": 0.4968,
      "step": 20420
    },
    {
      "epoch": 6.448058099147458,
      "grad_norm": 4.391901969909668,
      "learning_rate": 3.7644283057923726e-05,
      "loss": 0.5156,
      "step": 20421
    },
    {
      "epoch": 6.448373855383644,
      "grad_norm": 5.18325138092041,
      "learning_rate": 3.764340595726766e-05,
      "loss": 0.5709,
      "step": 20422
    },
    {
      "epoch": 6.448689611619829,
      "grad_norm": 5.181893348693848,
      "learning_rate": 3.764252885661159e-05,
      "loss": 0.517,
      "step": 20423
    },
    {
      "epoch": 6.4490053678560155,
      "grad_norm": 7.253856658935547,
      "learning_rate": 3.7641651755955515e-05,
      "loss": 0.5866,
      "step": 20424
    },
    {
      "epoch": 6.449321124092201,
      "grad_norm": 4.707892417907715,
      "learning_rate": 3.7640774655299446e-05,
      "loss": 0.5458,
      "step": 20425
    },
    {
      "epoch": 6.449636880328386,
      "grad_norm": 5.018573760986328,
      "learning_rate": 3.763989755464337e-05,
      "loss": 0.5127,
      "step": 20426
    },
    {
      "epoch": 6.449952636564572,
      "grad_norm": 5.146605491638184,
      "learning_rate": 3.76390204539873e-05,
      "loss": 0.5016,
      "step": 20427
    },
    {
      "epoch": 6.450268392800758,
      "grad_norm": 4.917673110961914,
      "learning_rate": 3.763814335333123e-05,
      "loss": 0.5133,
      "step": 20428
    },
    {
      "epoch": 6.450584149036944,
      "grad_norm": 5.697995185852051,
      "learning_rate": 3.763726625267516e-05,
      "loss": 0.6859,
      "step": 20429
    },
    {
      "epoch": 6.450899905273129,
      "grad_norm": 4.975672245025635,
      "learning_rate": 3.7636389152019084e-05,
      "loss": 0.5496,
      "step": 20430
    },
    {
      "epoch": 6.451215661509314,
      "grad_norm": 4.956869125366211,
      "learning_rate": 3.7635512051363016e-05,
      "loss": 0.631,
      "step": 20431
    },
    {
      "epoch": 6.4515314177455005,
      "grad_norm": 5.112936496734619,
      "learning_rate": 3.763463495070694e-05,
      "loss": 0.631,
      "step": 20432
    },
    {
      "epoch": 6.451847173981686,
      "grad_norm": 5.855309963226318,
      "learning_rate": 3.763375785005087e-05,
      "loss": 0.5149,
      "step": 20433
    },
    {
      "epoch": 6.452162930217872,
      "grad_norm": 5.508225917816162,
      "learning_rate": 3.76328807493948e-05,
      "loss": 0.608,
      "step": 20434
    },
    {
      "epoch": 6.452478686454057,
      "grad_norm": 4.847487926483154,
      "learning_rate": 3.763200364873873e-05,
      "loss": 0.6073,
      "step": 20435
    },
    {
      "epoch": 6.4527944426902435,
      "grad_norm": 5.897792816162109,
      "learning_rate": 3.763112654808266e-05,
      "loss": 0.6099,
      "step": 20436
    },
    {
      "epoch": 6.453110198926429,
      "grad_norm": 5.107728004455566,
      "learning_rate": 3.763024944742659e-05,
      "loss": 0.7047,
      "step": 20437
    },
    {
      "epoch": 6.453425955162614,
      "grad_norm": 5.8103413581848145,
      "learning_rate": 3.762937234677052e-05,
      "loss": 0.5684,
      "step": 20438
    },
    {
      "epoch": 6.4537417113988,
      "grad_norm": 4.333243370056152,
      "learning_rate": 3.762849524611445e-05,
      "loss": 0.552,
      "step": 20439
    },
    {
      "epoch": 6.454057467634986,
      "grad_norm": 6.123591899871826,
      "learning_rate": 3.7627618145458374e-05,
      "loss": 0.5618,
      "step": 20440
    },
    {
      "epoch": 6.454373223871172,
      "grad_norm": 5.369500637054443,
      "learning_rate": 3.7626741044802306e-05,
      "loss": 0.5702,
      "step": 20441
    },
    {
      "epoch": 6.454688980107357,
      "grad_norm": 4.822795867919922,
      "learning_rate": 3.762586394414623e-05,
      "loss": 0.5343,
      "step": 20442
    },
    {
      "epoch": 6.455004736343543,
      "grad_norm": 4.328269004821777,
      "learning_rate": 3.762498684349016e-05,
      "loss": 0.5517,
      "step": 20443
    },
    {
      "epoch": 6.4553204925797285,
      "grad_norm": 5.008111476898193,
      "learning_rate": 3.762410974283409e-05,
      "loss": 0.5379,
      "step": 20444
    },
    {
      "epoch": 6.455636248815914,
      "grad_norm": 5.3326334953308105,
      "learning_rate": 3.762323264217802e-05,
      "loss": 0.5419,
      "step": 20445
    },
    {
      "epoch": 6.4559520050521,
      "grad_norm": 3.8575024604797363,
      "learning_rate": 3.762235554152195e-05,
      "loss": 0.5124,
      "step": 20446
    },
    {
      "epoch": 6.456267761288285,
      "grad_norm": 4.328808307647705,
      "learning_rate": 3.7621478440865875e-05,
      "loss": 0.5316,
      "step": 20447
    },
    {
      "epoch": 6.4565835175244715,
      "grad_norm": 6.0826640129089355,
      "learning_rate": 3.762060134020981e-05,
      "loss": 0.6424,
      "step": 20448
    },
    {
      "epoch": 6.456899273760657,
      "grad_norm": 4.7189130783081055,
      "learning_rate": 3.761972423955373e-05,
      "loss": 0.5134,
      "step": 20449
    },
    {
      "epoch": 6.457215029996842,
      "grad_norm": 4.5318284034729,
      "learning_rate": 3.7618847138897664e-05,
      "loss": 0.4796,
      "step": 20450
    },
    {
      "epoch": 6.457530786233028,
      "grad_norm": 4.674173355102539,
      "learning_rate": 3.761797003824159e-05,
      "loss": 0.5441,
      "step": 20451
    },
    {
      "epoch": 6.4578465424692135,
      "grad_norm": 4.875282287597656,
      "learning_rate": 3.761709293758552e-05,
      "loss": 0.5634,
      "step": 20452
    },
    {
      "epoch": 6.4581622987054,
      "grad_norm": 4.590569496154785,
      "learning_rate": 3.7616215836929445e-05,
      "loss": 0.467,
      "step": 20453
    },
    {
      "epoch": 6.458478054941585,
      "grad_norm": 5.0163187980651855,
      "learning_rate": 3.761533873627338e-05,
      "loss": 0.5888,
      "step": 20454
    },
    {
      "epoch": 6.45879381117777,
      "grad_norm": 3.808748245239258,
      "learning_rate": 3.76144616356173e-05,
      "loss": 0.4766,
      "step": 20455
    },
    {
      "epoch": 6.4591095674139565,
      "grad_norm": 5.0087480545043945,
      "learning_rate": 3.7613584534961233e-05,
      "loss": 0.4762,
      "step": 20456
    },
    {
      "epoch": 6.459425323650142,
      "grad_norm": 4.351603031158447,
      "learning_rate": 3.761270743430516e-05,
      "loss": 0.6122,
      "step": 20457
    },
    {
      "epoch": 6.459741079886328,
      "grad_norm": 5.858543872833252,
      "learning_rate": 3.761183033364909e-05,
      "loss": 0.5979,
      "step": 20458
    },
    {
      "epoch": 6.460056836122513,
      "grad_norm": 4.781881809234619,
      "learning_rate": 3.7610953232993015e-05,
      "loss": 0.5274,
      "step": 20459
    },
    {
      "epoch": 6.460372592358699,
      "grad_norm": 5.007981777191162,
      "learning_rate": 3.761007613233695e-05,
      "loss": 0.6541,
      "step": 20460
    },
    {
      "epoch": 6.460688348594885,
      "grad_norm": 5.425254821777344,
      "learning_rate": 3.760919903168088e-05,
      "loss": 0.5615,
      "step": 20461
    },
    {
      "epoch": 6.46100410483107,
      "grad_norm": 4.642821311950684,
      "learning_rate": 3.76083219310248e-05,
      "loss": 0.5531,
      "step": 20462
    },
    {
      "epoch": 6.461319861067256,
      "grad_norm": 4.987969398498535,
      "learning_rate": 3.7607444830368735e-05,
      "loss": 0.4704,
      "step": 20463
    },
    {
      "epoch": 6.4616356173034415,
      "grad_norm": 4.837405681610107,
      "learning_rate": 3.760656772971267e-05,
      "loss": 0.5159,
      "step": 20464
    },
    {
      "epoch": 6.461951373539628,
      "grad_norm": 5.881285667419434,
      "learning_rate": 3.760569062905659e-05,
      "loss": 0.6706,
      "step": 20465
    },
    {
      "epoch": 6.462267129775813,
      "grad_norm": 5.333015441894531,
      "learning_rate": 3.760481352840052e-05,
      "loss": 0.5425,
      "step": 20466
    },
    {
      "epoch": 6.462582886011999,
      "grad_norm": 4.939762592315674,
      "learning_rate": 3.7603936427744455e-05,
      "loss": 0.5701,
      "step": 20467
    },
    {
      "epoch": 6.4628986422481844,
      "grad_norm": 5.397885322570801,
      "learning_rate": 3.760305932708838e-05,
      "loss": 0.6338,
      "step": 20468
    },
    {
      "epoch": 6.46321439848437,
      "grad_norm": 5.217689037322998,
      "learning_rate": 3.760218222643231e-05,
      "loss": 0.506,
      "step": 20469
    },
    {
      "epoch": 6.463530154720556,
      "grad_norm": 5.371678829193115,
      "learning_rate": 3.7601305125776236e-05,
      "loss": 0.5765,
      "step": 20470
    },
    {
      "epoch": 6.463845910956741,
      "grad_norm": 4.787965297698975,
      "learning_rate": 3.760042802512017e-05,
      "loss": 0.61,
      "step": 20471
    },
    {
      "epoch": 6.464161667192927,
      "grad_norm": 3.8604671955108643,
      "learning_rate": 3.759955092446409e-05,
      "loss": 0.4457,
      "step": 20472
    },
    {
      "epoch": 6.464477423429113,
      "grad_norm": 5.231571197509766,
      "learning_rate": 3.7598673823808025e-05,
      "loss": 0.5414,
      "step": 20473
    },
    {
      "epoch": 6.464793179665298,
      "grad_norm": 5.017187118530273,
      "learning_rate": 3.759779672315195e-05,
      "loss": 0.5912,
      "step": 20474
    },
    {
      "epoch": 6.465108935901484,
      "grad_norm": 4.906368732452393,
      "learning_rate": 3.759691962249588e-05,
      "loss": 0.6103,
      "step": 20475
    },
    {
      "epoch": 6.4654246921376695,
      "grad_norm": 5.448594570159912,
      "learning_rate": 3.7596042521839806e-05,
      "loss": 0.5918,
      "step": 20476
    },
    {
      "epoch": 6.465740448373856,
      "grad_norm": 5.636707782745361,
      "learning_rate": 3.759516542118374e-05,
      "loss": 0.5886,
      "step": 20477
    },
    {
      "epoch": 6.466056204610041,
      "grad_norm": 4.497722148895264,
      "learning_rate": 3.759428832052766e-05,
      "loss": 0.5097,
      "step": 20478
    },
    {
      "epoch": 6.466371960846227,
      "grad_norm": 4.606041431427002,
      "learning_rate": 3.7593411219871594e-05,
      "loss": 0.5262,
      "step": 20479
    },
    {
      "epoch": 6.466687717082412,
      "grad_norm": 4.734260559082031,
      "learning_rate": 3.759253411921552e-05,
      "loss": 0.5301,
      "step": 20480
    },
    {
      "epoch": 6.467003473318598,
      "grad_norm": 6.295777320861816,
      "learning_rate": 3.759165701855945e-05,
      "loss": 0.5469,
      "step": 20481
    },
    {
      "epoch": 6.467319229554784,
      "grad_norm": 4.679834842681885,
      "learning_rate": 3.7590779917903376e-05,
      "loss": 0.586,
      "step": 20482
    },
    {
      "epoch": 6.467634985790969,
      "grad_norm": 5.944006443023682,
      "learning_rate": 3.758990281724731e-05,
      "loss": 0.5798,
      "step": 20483
    },
    {
      "epoch": 6.467950742027155,
      "grad_norm": 5.264884948730469,
      "learning_rate": 3.758902571659124e-05,
      "loss": 0.5476,
      "step": 20484
    },
    {
      "epoch": 6.468266498263341,
      "grad_norm": 4.694153785705566,
      "learning_rate": 3.7588148615935164e-05,
      "loss": 0.5875,
      "step": 20485
    },
    {
      "epoch": 6.468582254499526,
      "grad_norm": 4.520950794219971,
      "learning_rate": 3.7587271515279096e-05,
      "loss": 0.5619,
      "step": 20486
    },
    {
      "epoch": 6.468898010735712,
      "grad_norm": 4.897191047668457,
      "learning_rate": 3.758639441462302e-05,
      "loss": 0.6591,
      "step": 20487
    },
    {
      "epoch": 6.469213766971897,
      "grad_norm": 5.077407360076904,
      "learning_rate": 3.758551731396695e-05,
      "loss": 0.6173,
      "step": 20488
    },
    {
      "epoch": 6.469529523208084,
      "grad_norm": 4.413147926330566,
      "learning_rate": 3.758464021331088e-05,
      "loss": 0.4194,
      "step": 20489
    },
    {
      "epoch": 6.469845279444269,
      "grad_norm": 5.681084156036377,
      "learning_rate": 3.758376311265481e-05,
      "loss": 0.6423,
      "step": 20490
    },
    {
      "epoch": 6.470161035680455,
      "grad_norm": 4.518143653869629,
      "learning_rate": 3.758288601199874e-05,
      "loss": 0.4726,
      "step": 20491
    },
    {
      "epoch": 6.47047679191664,
      "grad_norm": 4.706515312194824,
      "learning_rate": 3.758200891134267e-05,
      "loss": 0.5217,
      "step": 20492
    },
    {
      "epoch": 6.470792548152826,
      "grad_norm": 5.446728706359863,
      "learning_rate": 3.75811318106866e-05,
      "loss": 0.6074,
      "step": 20493
    },
    {
      "epoch": 6.471108304389012,
      "grad_norm": 4.36214017868042,
      "learning_rate": 3.758025471003053e-05,
      "loss": 0.5327,
      "step": 20494
    },
    {
      "epoch": 6.471424060625197,
      "grad_norm": 3.7673380374908447,
      "learning_rate": 3.7579377609374454e-05,
      "loss": 0.5005,
      "step": 20495
    },
    {
      "epoch": 6.471739816861383,
      "grad_norm": 5.588095664978027,
      "learning_rate": 3.7578500508718386e-05,
      "loss": 0.6051,
      "step": 20496
    },
    {
      "epoch": 6.472055573097569,
      "grad_norm": 5.138113498687744,
      "learning_rate": 3.757762340806231e-05,
      "loss": 0.5165,
      "step": 20497
    },
    {
      "epoch": 6.472371329333754,
      "grad_norm": 5.413331985473633,
      "learning_rate": 3.757674630740624e-05,
      "loss": 0.5262,
      "step": 20498
    },
    {
      "epoch": 6.47268708556994,
      "grad_norm": 4.562353134155273,
      "learning_rate": 3.757586920675017e-05,
      "loss": 0.5162,
      "step": 20499
    },
    {
      "epoch": 6.473002841806125,
      "grad_norm": 5.7241973876953125,
      "learning_rate": 3.75749921060941e-05,
      "loss": 0.52,
      "step": 20500
    },
    {
      "epoch": 6.473318598042312,
      "grad_norm": 5.498325824737549,
      "learning_rate": 3.7574115005438024e-05,
      "loss": 0.5625,
      "step": 20501
    },
    {
      "epoch": 6.473634354278497,
      "grad_norm": 4.377091884613037,
      "learning_rate": 3.7573237904781955e-05,
      "loss": 0.5397,
      "step": 20502
    },
    {
      "epoch": 6.473950110514683,
      "grad_norm": 5.466963291168213,
      "learning_rate": 3.757236080412588e-05,
      "loss": 0.5317,
      "step": 20503
    },
    {
      "epoch": 6.474265866750868,
      "grad_norm": 5.217645645141602,
      "learning_rate": 3.757148370346981e-05,
      "loss": 0.5311,
      "step": 20504
    },
    {
      "epoch": 6.474581622987054,
      "grad_norm": 5.55272102355957,
      "learning_rate": 3.7570606602813744e-05,
      "loss": 0.5346,
      "step": 20505
    },
    {
      "epoch": 6.47489737922324,
      "grad_norm": 4.9862518310546875,
      "learning_rate": 3.756972950215767e-05,
      "loss": 0.5276,
      "step": 20506
    },
    {
      "epoch": 6.475213135459425,
      "grad_norm": 4.0958991050720215,
      "learning_rate": 3.75688524015016e-05,
      "loss": 0.5009,
      "step": 20507
    },
    {
      "epoch": 6.475528891695611,
      "grad_norm": 4.53579568862915,
      "learning_rate": 3.7567975300845525e-05,
      "loss": 0.4955,
      "step": 20508
    },
    {
      "epoch": 6.475844647931797,
      "grad_norm": 5.111384868621826,
      "learning_rate": 3.756709820018946e-05,
      "loss": 0.6173,
      "step": 20509
    },
    {
      "epoch": 6.476160404167983,
      "grad_norm": 4.550724029541016,
      "learning_rate": 3.756622109953338e-05,
      "loss": 0.5295,
      "step": 20510
    },
    {
      "epoch": 6.476476160404168,
      "grad_norm": 5.01965856552124,
      "learning_rate": 3.7565343998877313e-05,
      "loss": 0.5375,
      "step": 20511
    },
    {
      "epoch": 6.476791916640353,
      "grad_norm": 5.303671836853027,
      "learning_rate": 3.756446689822124e-05,
      "loss": 0.5772,
      "step": 20512
    },
    {
      "epoch": 6.4771076728765395,
      "grad_norm": 3.8861892223358154,
      "learning_rate": 3.756358979756517e-05,
      "loss": 0.4353,
      "step": 20513
    },
    {
      "epoch": 6.477423429112725,
      "grad_norm": 5.516625881195068,
      "learning_rate": 3.7562712696909095e-05,
      "loss": 0.4928,
      "step": 20514
    },
    {
      "epoch": 6.477739185348911,
      "grad_norm": 5.873453140258789,
      "learning_rate": 3.7561835596253027e-05,
      "loss": 0.5263,
      "step": 20515
    },
    {
      "epoch": 6.478054941585096,
      "grad_norm": 4.6880340576171875,
      "learning_rate": 3.756095849559695e-05,
      "loss": 0.5748,
      "step": 20516
    },
    {
      "epoch": 6.478370697821282,
      "grad_norm": 4.141476154327393,
      "learning_rate": 3.756008139494088e-05,
      "loss": 0.5494,
      "step": 20517
    },
    {
      "epoch": 6.478686454057468,
      "grad_norm": 5.636568069458008,
      "learning_rate": 3.7559204294284815e-05,
      "loss": 0.5319,
      "step": 20518
    },
    {
      "epoch": 6.479002210293653,
      "grad_norm": 5.9707722663879395,
      "learning_rate": 3.7558327193628747e-05,
      "loss": 0.5363,
      "step": 20519
    },
    {
      "epoch": 6.479317966529839,
      "grad_norm": 5.022814750671387,
      "learning_rate": 3.755745009297267e-05,
      "loss": 0.6461,
      "step": 20520
    },
    {
      "epoch": 6.479633722766025,
      "grad_norm": 4.830698490142822,
      "learning_rate": 3.75565729923166e-05,
      "loss": 0.643,
      "step": 20521
    },
    {
      "epoch": 6.47994947900221,
      "grad_norm": 5.966024875640869,
      "learning_rate": 3.755569589166053e-05,
      "loss": 0.6849,
      "step": 20522
    },
    {
      "epoch": 6.480265235238396,
      "grad_norm": 4.515352725982666,
      "learning_rate": 3.755481879100446e-05,
      "loss": 0.5213,
      "step": 20523
    },
    {
      "epoch": 6.480580991474581,
      "grad_norm": 4.90864896774292,
      "learning_rate": 3.7553941690348385e-05,
      "loss": 0.5215,
      "step": 20524
    },
    {
      "epoch": 6.4808967477107675,
      "grad_norm": 5.408693790435791,
      "learning_rate": 3.7553064589692316e-05,
      "loss": 0.5859,
      "step": 20525
    },
    {
      "epoch": 6.481212503946953,
      "grad_norm": 4.520186424255371,
      "learning_rate": 3.755218748903625e-05,
      "loss": 0.4595,
      "step": 20526
    },
    {
      "epoch": 6.481528260183139,
      "grad_norm": 4.764204978942871,
      "learning_rate": 3.755131038838017e-05,
      "loss": 0.5565,
      "step": 20527
    },
    {
      "epoch": 6.481844016419324,
      "grad_norm": 5.164906024932861,
      "learning_rate": 3.7550433287724105e-05,
      "loss": 0.5876,
      "step": 20528
    },
    {
      "epoch": 6.48215977265551,
      "grad_norm": 4.185204029083252,
      "learning_rate": 3.754955618706803e-05,
      "loss": 0.4714,
      "step": 20529
    },
    {
      "epoch": 6.482475528891696,
      "grad_norm": 4.630690097808838,
      "learning_rate": 3.754867908641196e-05,
      "loss": 0.4779,
      "step": 20530
    },
    {
      "epoch": 6.482791285127881,
      "grad_norm": 5.627082347869873,
      "learning_rate": 3.7547801985755886e-05,
      "loss": 0.541,
      "step": 20531
    },
    {
      "epoch": 6.483107041364067,
      "grad_norm": 6.112770080566406,
      "learning_rate": 3.754692488509982e-05,
      "loss": 0.5945,
      "step": 20532
    },
    {
      "epoch": 6.4834227976002525,
      "grad_norm": 5.695994853973389,
      "learning_rate": 3.754604778444374e-05,
      "loss": 0.5946,
      "step": 20533
    },
    {
      "epoch": 6.483738553836439,
      "grad_norm": 5.482604026794434,
      "learning_rate": 3.7545170683787674e-05,
      "loss": 0.5994,
      "step": 20534
    },
    {
      "epoch": 6.484054310072624,
      "grad_norm": 5.218924045562744,
      "learning_rate": 3.75442935831316e-05,
      "loss": 0.5222,
      "step": 20535
    },
    {
      "epoch": 6.484370066308809,
      "grad_norm": 5.529560565948486,
      "learning_rate": 3.754341648247553e-05,
      "loss": 0.5205,
      "step": 20536
    },
    {
      "epoch": 6.4846858225449955,
      "grad_norm": 5.867706775665283,
      "learning_rate": 3.7542539381819456e-05,
      "loss": 0.615,
      "step": 20537
    },
    {
      "epoch": 6.485001578781181,
      "grad_norm": 4.941518783569336,
      "learning_rate": 3.754166228116339e-05,
      "loss": 0.6678,
      "step": 20538
    },
    {
      "epoch": 6.485317335017367,
      "grad_norm": 4.971313953399658,
      "learning_rate": 3.754078518050731e-05,
      "loss": 0.5295,
      "step": 20539
    },
    {
      "epoch": 6.485633091253552,
      "grad_norm": 5.037629127502441,
      "learning_rate": 3.7539908079851244e-05,
      "loss": 0.463,
      "step": 20540
    },
    {
      "epoch": 6.4859488474897375,
      "grad_norm": 4.24908447265625,
      "learning_rate": 3.753903097919517e-05,
      "loss": 0.528,
      "step": 20541
    },
    {
      "epoch": 6.486264603725924,
      "grad_norm": 5.90669584274292,
      "learning_rate": 3.75381538785391e-05,
      "loss": 0.5733,
      "step": 20542
    },
    {
      "epoch": 6.486580359962109,
      "grad_norm": 5.2963666915893555,
      "learning_rate": 3.753727677788303e-05,
      "loss": 0.5844,
      "step": 20543
    },
    {
      "epoch": 6.486896116198295,
      "grad_norm": 5.41015100479126,
      "learning_rate": 3.753639967722696e-05,
      "loss": 0.5372,
      "step": 20544
    },
    {
      "epoch": 6.4872118724344805,
      "grad_norm": 5.91697359085083,
      "learning_rate": 3.753552257657089e-05,
      "loss": 0.6616,
      "step": 20545
    },
    {
      "epoch": 6.487527628670666,
      "grad_norm": 4.2377028465271,
      "learning_rate": 3.753464547591482e-05,
      "loss": 0.5392,
      "step": 20546
    },
    {
      "epoch": 6.487843384906852,
      "grad_norm": 5.134120941162109,
      "learning_rate": 3.7533768375258746e-05,
      "loss": 0.53,
      "step": 20547
    },
    {
      "epoch": 6.488159141143037,
      "grad_norm": 5.125763416290283,
      "learning_rate": 3.753289127460268e-05,
      "loss": 0.6208,
      "step": 20548
    },
    {
      "epoch": 6.4884748973792234,
      "grad_norm": 5.806398391723633,
      "learning_rate": 3.753201417394661e-05,
      "loss": 0.652,
      "step": 20549
    },
    {
      "epoch": 6.488790653615409,
      "grad_norm": 5.501012325286865,
      "learning_rate": 3.7531137073290534e-05,
      "loss": 0.6642,
      "step": 20550
    },
    {
      "epoch": 6.489106409851595,
      "grad_norm": 5.66269063949585,
      "learning_rate": 3.7530259972634465e-05,
      "loss": 0.5167,
      "step": 20551
    },
    {
      "epoch": 6.48942216608778,
      "grad_norm": 5.174579620361328,
      "learning_rate": 3.752938287197839e-05,
      "loss": 0.5249,
      "step": 20552
    },
    {
      "epoch": 6.4897379223239655,
      "grad_norm": 5.107693195343018,
      "learning_rate": 3.752850577132232e-05,
      "loss": 0.4782,
      "step": 20553
    },
    {
      "epoch": 6.490053678560152,
      "grad_norm": 3.7230303287506104,
      "learning_rate": 3.752762867066625e-05,
      "loss": 0.5371,
      "step": 20554
    },
    {
      "epoch": 6.490369434796337,
      "grad_norm": 4.307106971740723,
      "learning_rate": 3.752675157001018e-05,
      "loss": 0.4764,
      "step": 20555
    },
    {
      "epoch": 6.490685191032523,
      "grad_norm": 5.555866241455078,
      "learning_rate": 3.7525874469354104e-05,
      "loss": 0.5904,
      "step": 20556
    },
    {
      "epoch": 6.4910009472687085,
      "grad_norm": 4.988597393035889,
      "learning_rate": 3.7524997368698035e-05,
      "loss": 0.5892,
      "step": 20557
    },
    {
      "epoch": 6.491316703504895,
      "grad_norm": 4.822052001953125,
      "learning_rate": 3.752412026804196e-05,
      "loss": 0.6143,
      "step": 20558
    },
    {
      "epoch": 6.49163245974108,
      "grad_norm": 4.661495208740234,
      "learning_rate": 3.752324316738589e-05,
      "loss": 0.5064,
      "step": 20559
    },
    {
      "epoch": 6.491948215977265,
      "grad_norm": 6.496689796447754,
      "learning_rate": 3.752236606672982e-05,
      "loss": 0.6318,
      "step": 20560
    },
    {
      "epoch": 6.492263972213451,
      "grad_norm": 5.126914024353027,
      "learning_rate": 3.752148896607375e-05,
      "loss": 0.5994,
      "step": 20561
    },
    {
      "epoch": 6.492579728449637,
      "grad_norm": 6.639904022216797,
      "learning_rate": 3.752061186541767e-05,
      "loss": 0.5898,
      "step": 20562
    },
    {
      "epoch": 6.492895484685823,
      "grad_norm": 4.499147891998291,
      "learning_rate": 3.7519734764761605e-05,
      "loss": 0.5345,
      "step": 20563
    },
    {
      "epoch": 6.493211240922008,
      "grad_norm": 4.095234394073486,
      "learning_rate": 3.751885766410554e-05,
      "loss": 0.4684,
      "step": 20564
    },
    {
      "epoch": 6.4935269971581935,
      "grad_norm": 4.8247151374816895,
      "learning_rate": 3.751798056344946e-05,
      "loss": 0.5186,
      "step": 20565
    },
    {
      "epoch": 6.49384275339438,
      "grad_norm": 4.684501647949219,
      "learning_rate": 3.751710346279339e-05,
      "loss": 0.536,
      "step": 20566
    },
    {
      "epoch": 6.494158509630565,
      "grad_norm": 5.777320384979248,
      "learning_rate": 3.751622636213732e-05,
      "loss": 0.4544,
      "step": 20567
    },
    {
      "epoch": 6.494474265866751,
      "grad_norm": 5.08710241317749,
      "learning_rate": 3.751534926148125e-05,
      "loss": 0.5279,
      "step": 20568
    },
    {
      "epoch": 6.494790022102936,
      "grad_norm": 5.3394389152526855,
      "learning_rate": 3.7514472160825175e-05,
      "loss": 0.5777,
      "step": 20569
    },
    {
      "epoch": 6.495105778339122,
      "grad_norm": 4.808836460113525,
      "learning_rate": 3.7513595060169106e-05,
      "loss": 0.5264,
      "step": 20570
    },
    {
      "epoch": 6.495421534575308,
      "grad_norm": 5.249978542327881,
      "learning_rate": 3.751271795951303e-05,
      "loss": 0.5549,
      "step": 20571
    },
    {
      "epoch": 6.495737290811493,
      "grad_norm": 4.941359043121338,
      "learning_rate": 3.751184085885697e-05,
      "loss": 0.4722,
      "step": 20572
    },
    {
      "epoch": 6.496053047047679,
      "grad_norm": 6.423171520233154,
      "learning_rate": 3.7510963758200895e-05,
      "loss": 0.6324,
      "step": 20573
    },
    {
      "epoch": 6.496368803283865,
      "grad_norm": 4.849071025848389,
      "learning_rate": 3.7510086657544826e-05,
      "loss": 0.5995,
      "step": 20574
    },
    {
      "epoch": 6.496684559520051,
      "grad_norm": 4.912248611450195,
      "learning_rate": 3.750920955688875e-05,
      "loss": 0.5713,
      "step": 20575
    },
    {
      "epoch": 6.497000315756236,
      "grad_norm": 4.743213653564453,
      "learning_rate": 3.750833245623268e-05,
      "loss": 0.5498,
      "step": 20576
    },
    {
      "epoch": 6.4973160719924214,
      "grad_norm": 5.754642963409424,
      "learning_rate": 3.750745535557661e-05,
      "loss": 0.6737,
      "step": 20577
    },
    {
      "epoch": 6.497631828228608,
      "grad_norm": 5.53098726272583,
      "learning_rate": 3.750657825492054e-05,
      "loss": 0.658,
      "step": 20578
    },
    {
      "epoch": 6.497947584464793,
      "grad_norm": 7.525722980499268,
      "learning_rate": 3.7505701154264464e-05,
      "loss": 0.5799,
      "step": 20579
    },
    {
      "epoch": 6.498263340700979,
      "grad_norm": 6.32658052444458,
      "learning_rate": 3.7504824053608396e-05,
      "loss": 0.6743,
      "step": 20580
    },
    {
      "epoch": 6.498579096937164,
      "grad_norm": 4.199642181396484,
      "learning_rate": 3.750394695295232e-05,
      "loss": 0.5102,
      "step": 20581
    },
    {
      "epoch": 6.498894853173351,
      "grad_norm": 5.265127658843994,
      "learning_rate": 3.750306985229625e-05,
      "loss": 0.4935,
      "step": 20582
    },
    {
      "epoch": 6.499210609409536,
      "grad_norm": 4.856285572052002,
      "learning_rate": 3.750219275164018e-05,
      "loss": 0.5243,
      "step": 20583
    },
    {
      "epoch": 6.499526365645721,
      "grad_norm": 4.100789546966553,
      "learning_rate": 3.750131565098411e-05,
      "loss": 0.4878,
      "step": 20584
    },
    {
      "epoch": 6.499842121881907,
      "grad_norm": 4.845950126647949,
      "learning_rate": 3.7500438550328034e-05,
      "loss": 0.6117,
      "step": 20585
    },
    {
      "epoch": 6.500157878118093,
      "grad_norm": 4.396385192871094,
      "learning_rate": 3.7499561449671966e-05,
      "loss": 0.517,
      "step": 20586
    },
    {
      "epoch": 6.500473634354279,
      "grad_norm": 6.865699291229248,
      "learning_rate": 3.74986843490159e-05,
      "loss": 0.5127,
      "step": 20587
    },
    {
      "epoch": 6.500789390590464,
      "grad_norm": 5.616053104400635,
      "learning_rate": 3.749780724835982e-05,
      "loss": 0.5756,
      "step": 20588
    },
    {
      "epoch": 6.501105146826649,
      "grad_norm": 5.782784461975098,
      "learning_rate": 3.7496930147703754e-05,
      "loss": 0.5573,
      "step": 20589
    },
    {
      "epoch": 6.501420903062836,
      "grad_norm": 4.653481483459473,
      "learning_rate": 3.749605304704768e-05,
      "loss": 0.4863,
      "step": 20590
    },
    {
      "epoch": 6.501736659299021,
      "grad_norm": 4.2240071296691895,
      "learning_rate": 3.749517594639161e-05,
      "loss": 0.5972,
      "step": 20591
    },
    {
      "epoch": 6.502052415535207,
      "grad_norm": 5.622707843780518,
      "learning_rate": 3.7494298845735536e-05,
      "loss": 0.5366,
      "step": 20592
    },
    {
      "epoch": 6.502368171771392,
      "grad_norm": 4.632281303405762,
      "learning_rate": 3.749342174507947e-05,
      "loss": 0.4457,
      "step": 20593
    },
    {
      "epoch": 6.502683928007578,
      "grad_norm": 4.750917434692383,
      "learning_rate": 3.749254464442339e-05,
      "loss": 0.4953,
      "step": 20594
    },
    {
      "epoch": 6.502999684243764,
      "grad_norm": 5.042547702789307,
      "learning_rate": 3.7491667543767324e-05,
      "loss": 0.5362,
      "step": 20595
    },
    {
      "epoch": 6.503315440479949,
      "grad_norm": 4.483210563659668,
      "learning_rate": 3.749079044311125e-05,
      "loss": 0.4627,
      "step": 20596
    },
    {
      "epoch": 6.503631196716135,
      "grad_norm": 5.006889343261719,
      "learning_rate": 3.748991334245518e-05,
      "loss": 0.5646,
      "step": 20597
    },
    {
      "epoch": 6.503946952952321,
      "grad_norm": 4.228079319000244,
      "learning_rate": 3.7489036241799105e-05,
      "loss": 0.5401,
      "step": 20598
    },
    {
      "epoch": 6.504262709188507,
      "grad_norm": 5.545529842376709,
      "learning_rate": 3.748815914114304e-05,
      "loss": 0.5549,
      "step": 20599
    },
    {
      "epoch": 6.504578465424692,
      "grad_norm": 5.618605613708496,
      "learning_rate": 3.748728204048697e-05,
      "loss": 0.5872,
      "step": 20600
    },
    {
      "epoch": 6.504578465424692,
      "eval_loss": 0.8793366551399231,
      "eval_runtime": 4.638,
      "eval_samples_per_second": 536.873,
      "eval_steps_per_second": 4.528,
      "step": 20600
    },
    {
      "epoch": 6.504894221660878,
      "grad_norm": 3.6920270919799805,
      "learning_rate": 3.74864049398309e-05,
      "loss": 0.5115,
      "step": 20601
    },
    {
      "epoch": 6.505209977897064,
      "grad_norm": 4.790878772735596,
      "learning_rate": 3.7485527839174825e-05,
      "loss": 0.6081,
      "step": 20602
    },
    {
      "epoch": 6.505525734133249,
      "grad_norm": 3.9736247062683105,
      "learning_rate": 3.748465073851876e-05,
      "loss": 0.4881,
      "step": 20603
    },
    {
      "epoch": 6.505841490369435,
      "grad_norm": 4.685170650482178,
      "learning_rate": 3.748377363786268e-05,
      "loss": 0.5376,
      "step": 20604
    },
    {
      "epoch": 6.50615724660562,
      "grad_norm": 5.0303730964660645,
      "learning_rate": 3.7482896537206614e-05,
      "loss": 0.6514,
      "step": 20605
    },
    {
      "epoch": 6.5064730028418065,
      "grad_norm": 5.504604339599609,
      "learning_rate": 3.748201943655054e-05,
      "loss": 0.5522,
      "step": 20606
    },
    {
      "epoch": 6.506788759077992,
      "grad_norm": 5.494097709655762,
      "learning_rate": 3.748114233589447e-05,
      "loss": 0.5192,
      "step": 20607
    },
    {
      "epoch": 6.507104515314177,
      "grad_norm": 5.447098731994629,
      "learning_rate": 3.74802652352384e-05,
      "loss": 0.5131,
      "step": 20608
    },
    {
      "epoch": 6.507420271550363,
      "grad_norm": 4.880213737487793,
      "learning_rate": 3.747938813458233e-05,
      "loss": 0.4952,
      "step": 20609
    },
    {
      "epoch": 6.507736027786549,
      "grad_norm": 4.6841864585876465,
      "learning_rate": 3.747851103392626e-05,
      "loss": 0.4656,
      "step": 20610
    },
    {
      "epoch": 6.508051784022735,
      "grad_norm": 4.469210624694824,
      "learning_rate": 3.7477633933270183e-05,
      "loss": 0.496,
      "step": 20611
    },
    {
      "epoch": 6.50836754025892,
      "grad_norm": 6.093766212463379,
      "learning_rate": 3.7476756832614115e-05,
      "loss": 0.5584,
      "step": 20612
    },
    {
      "epoch": 6.508683296495105,
      "grad_norm": 5.006241321563721,
      "learning_rate": 3.747587973195804e-05,
      "loss": 0.5253,
      "step": 20613
    },
    {
      "epoch": 6.5089990527312915,
      "grad_norm": 4.367794513702393,
      "learning_rate": 3.747500263130197e-05,
      "loss": 0.5249,
      "step": 20614
    },
    {
      "epoch": 6.509314808967477,
      "grad_norm": 4.907140254974365,
      "learning_rate": 3.74741255306459e-05,
      "loss": 0.5987,
      "step": 20615
    },
    {
      "epoch": 6.509630565203663,
      "grad_norm": 4.590368747711182,
      "learning_rate": 3.747324842998983e-05,
      "loss": 0.5596,
      "step": 20616
    },
    {
      "epoch": 6.509946321439848,
      "grad_norm": 4.665028095245361,
      "learning_rate": 3.747237132933375e-05,
      "loss": 0.6071,
      "step": 20617
    },
    {
      "epoch": 6.5102620776760345,
      "grad_norm": 4.450924396514893,
      "learning_rate": 3.7471494228677685e-05,
      "loss": 0.5546,
      "step": 20618
    },
    {
      "epoch": 6.51057783391222,
      "grad_norm": 5.376713275909424,
      "learning_rate": 3.747061712802161e-05,
      "loss": 0.5801,
      "step": 20619
    },
    {
      "epoch": 6.510893590148405,
      "grad_norm": 4.49027156829834,
      "learning_rate": 3.746974002736554e-05,
      "loss": 0.5286,
      "step": 20620
    },
    {
      "epoch": 6.511209346384591,
      "grad_norm": 4.8467254638671875,
      "learning_rate": 3.7468862926709466e-05,
      "loss": 0.5898,
      "step": 20621
    },
    {
      "epoch": 6.5115251026207766,
      "grad_norm": 6.613624095916748,
      "learning_rate": 3.74679858260534e-05,
      "loss": 0.619,
      "step": 20622
    },
    {
      "epoch": 6.511840858856963,
      "grad_norm": 3.9662318229675293,
      "learning_rate": 3.746710872539732e-05,
      "loss": 0.4637,
      "step": 20623
    },
    {
      "epoch": 6.512156615093148,
      "grad_norm": 5.165070056915283,
      "learning_rate": 3.7466231624741255e-05,
      "loss": 0.5746,
      "step": 20624
    },
    {
      "epoch": 6.512472371329334,
      "grad_norm": 5.878535270690918,
      "learning_rate": 3.7465354524085186e-05,
      "loss": 0.5143,
      "step": 20625
    },
    {
      "epoch": 6.5127881275655195,
      "grad_norm": 5.599517345428467,
      "learning_rate": 3.746447742342911e-05,
      "loss": 0.5787,
      "step": 20626
    },
    {
      "epoch": 6.513103883801705,
      "grad_norm": 4.734420299530029,
      "learning_rate": 3.746360032277304e-05,
      "loss": 0.5254,
      "step": 20627
    },
    {
      "epoch": 6.513419640037891,
      "grad_norm": 4.442788124084473,
      "learning_rate": 3.7462723222116975e-05,
      "loss": 0.6243,
      "step": 20628
    },
    {
      "epoch": 6.513735396274076,
      "grad_norm": 4.735457897186279,
      "learning_rate": 3.74618461214609e-05,
      "loss": 0.5214,
      "step": 20629
    },
    {
      "epoch": 6.5140511525102625,
      "grad_norm": 5.336484909057617,
      "learning_rate": 3.746096902080483e-05,
      "loss": 0.5166,
      "step": 20630
    },
    {
      "epoch": 6.514366908746448,
      "grad_norm": 6.755926132202148,
      "learning_rate": 3.746009192014876e-05,
      "loss": 0.5703,
      "step": 20631
    },
    {
      "epoch": 6.514682664982633,
      "grad_norm": 5.824974060058594,
      "learning_rate": 3.745921481949269e-05,
      "loss": 0.6984,
      "step": 20632
    },
    {
      "epoch": 6.514998421218819,
      "grad_norm": 5.058187961578369,
      "learning_rate": 3.745833771883662e-05,
      "loss": 0.5553,
      "step": 20633
    },
    {
      "epoch": 6.5153141774550045,
      "grad_norm": 3.899588108062744,
      "learning_rate": 3.7457460618180544e-05,
      "loss": 0.4804,
      "step": 20634
    },
    {
      "epoch": 6.515629933691191,
      "grad_norm": 4.925307750701904,
      "learning_rate": 3.7456583517524476e-05,
      "loss": 0.5777,
      "step": 20635
    },
    {
      "epoch": 6.515945689927376,
      "grad_norm": 6.181602954864502,
      "learning_rate": 3.74557064168684e-05,
      "loss": 0.5483,
      "step": 20636
    },
    {
      "epoch": 6.516261446163561,
      "grad_norm": 4.856261730194092,
      "learning_rate": 3.745482931621233e-05,
      "loss": 0.5712,
      "step": 20637
    },
    {
      "epoch": 6.5165772023997475,
      "grad_norm": 4.772817134857178,
      "learning_rate": 3.745395221555626e-05,
      "loss": 0.506,
      "step": 20638
    },
    {
      "epoch": 6.516892958635933,
      "grad_norm": 6.338009357452393,
      "learning_rate": 3.745307511490019e-05,
      "loss": 0.6135,
      "step": 20639
    },
    {
      "epoch": 6.517208714872119,
      "grad_norm": 4.716266632080078,
      "learning_rate": 3.7452198014244114e-05,
      "loss": 0.4854,
      "step": 20640
    },
    {
      "epoch": 6.517524471108304,
      "grad_norm": 5.3177289962768555,
      "learning_rate": 3.7451320913588046e-05,
      "loss": 0.5598,
      "step": 20641
    },
    {
      "epoch": 6.51784022734449,
      "grad_norm": 4.79631233215332,
      "learning_rate": 3.745044381293197e-05,
      "loss": 0.4936,
      "step": 20642
    },
    {
      "epoch": 6.518155983580676,
      "grad_norm": 4.698276519775391,
      "learning_rate": 3.74495667122759e-05,
      "loss": 0.5109,
      "step": 20643
    },
    {
      "epoch": 6.518471739816861,
      "grad_norm": 4.831854820251465,
      "learning_rate": 3.744868961161983e-05,
      "loss": 0.4378,
      "step": 20644
    },
    {
      "epoch": 6.518787496053047,
      "grad_norm": 5.4673004150390625,
      "learning_rate": 3.744781251096376e-05,
      "loss": 0.4872,
      "step": 20645
    },
    {
      "epoch": 6.5191032522892325,
      "grad_norm": 6.543850898742676,
      "learning_rate": 3.744693541030769e-05,
      "loss": 0.6677,
      "step": 20646
    },
    {
      "epoch": 6.519419008525419,
      "grad_norm": 6.098822116851807,
      "learning_rate": 3.7446058309651616e-05,
      "loss": 0.6794,
      "step": 20647
    },
    {
      "epoch": 6.519734764761604,
      "grad_norm": 5.178183555603027,
      "learning_rate": 3.744518120899555e-05,
      "loss": 0.5158,
      "step": 20648
    },
    {
      "epoch": 6.52005052099779,
      "grad_norm": 6.053361415863037,
      "learning_rate": 3.744430410833947e-05,
      "loss": 0.664,
      "step": 20649
    },
    {
      "epoch": 6.520366277233975,
      "grad_norm": 5.048672676086426,
      "learning_rate": 3.7443427007683404e-05,
      "loss": 0.4861,
      "step": 20650
    },
    {
      "epoch": 6.520682033470161,
      "grad_norm": 6.773439884185791,
      "learning_rate": 3.744254990702733e-05,
      "loss": 0.5982,
      "step": 20651
    },
    {
      "epoch": 6.520997789706347,
      "grad_norm": 5.9695281982421875,
      "learning_rate": 3.744167280637126e-05,
      "loss": 0.6531,
      "step": 20652
    },
    {
      "epoch": 6.521313545942532,
      "grad_norm": 5.663944721221924,
      "learning_rate": 3.7440795705715185e-05,
      "loss": 0.5108,
      "step": 20653
    },
    {
      "epoch": 6.521629302178718,
      "grad_norm": 4.7355499267578125,
      "learning_rate": 3.7439918605059124e-05,
      "loss": 0.5622,
      "step": 20654
    },
    {
      "epoch": 6.521945058414904,
      "grad_norm": 6.37096643447876,
      "learning_rate": 3.743904150440305e-05,
      "loss": 0.5301,
      "step": 20655
    },
    {
      "epoch": 6.522260814651089,
      "grad_norm": 6.325002670288086,
      "learning_rate": 3.743816440374698e-05,
      "loss": 0.572,
      "step": 20656
    },
    {
      "epoch": 6.522576570887275,
      "grad_norm": 4.396462917327881,
      "learning_rate": 3.7437287303090905e-05,
      "loss": 0.4947,
      "step": 20657
    },
    {
      "epoch": 6.5228923271234605,
      "grad_norm": 5.740496635437012,
      "learning_rate": 3.743641020243484e-05,
      "loss": 0.6719,
      "step": 20658
    },
    {
      "epoch": 6.523208083359647,
      "grad_norm": 4.912156581878662,
      "learning_rate": 3.743553310177876e-05,
      "loss": 0.5373,
      "step": 20659
    },
    {
      "epoch": 6.523523839595832,
      "grad_norm": 5.264977931976318,
      "learning_rate": 3.7434656001122694e-05,
      "loss": 0.513,
      "step": 20660
    },
    {
      "epoch": 6.523839595832017,
      "grad_norm": 5.313749313354492,
      "learning_rate": 3.743377890046662e-05,
      "loss": 0.6092,
      "step": 20661
    },
    {
      "epoch": 6.524155352068203,
      "grad_norm": 4.950864315032959,
      "learning_rate": 3.743290179981055e-05,
      "loss": 0.5931,
      "step": 20662
    },
    {
      "epoch": 6.524471108304389,
      "grad_norm": 4.107954025268555,
      "learning_rate": 3.7432024699154475e-05,
      "loss": 0.5187,
      "step": 20663
    },
    {
      "epoch": 6.524786864540575,
      "grad_norm": 7.54472541809082,
      "learning_rate": 3.743114759849841e-05,
      "loss": 0.5194,
      "step": 20664
    },
    {
      "epoch": 6.52510262077676,
      "grad_norm": 5.801425933837891,
      "learning_rate": 3.743027049784233e-05,
      "loss": 0.5322,
      "step": 20665
    },
    {
      "epoch": 6.525418377012946,
      "grad_norm": 5.896122455596924,
      "learning_rate": 3.742939339718626e-05,
      "loss": 0.595,
      "step": 20666
    },
    {
      "epoch": 6.525734133249132,
      "grad_norm": 4.168522834777832,
      "learning_rate": 3.742851629653019e-05,
      "loss": 0.5518,
      "step": 20667
    },
    {
      "epoch": 6.526049889485318,
      "grad_norm": 5.525961399078369,
      "learning_rate": 3.742763919587412e-05,
      "loss": 0.5168,
      "step": 20668
    },
    {
      "epoch": 6.526365645721503,
      "grad_norm": 3.777865171432495,
      "learning_rate": 3.742676209521805e-05,
      "loss": 0.5232,
      "step": 20669
    },
    {
      "epoch": 6.526681401957688,
      "grad_norm": 5.139551639556885,
      "learning_rate": 3.7425884994561977e-05,
      "loss": 0.5339,
      "step": 20670
    },
    {
      "epoch": 6.526997158193875,
      "grad_norm": 5.366314888000488,
      "learning_rate": 3.742500789390591e-05,
      "loss": 0.5902,
      "step": 20671
    },
    {
      "epoch": 6.52731291443006,
      "grad_norm": 5.024820327758789,
      "learning_rate": 3.742413079324983e-05,
      "loss": 0.5393,
      "step": 20672
    },
    {
      "epoch": 6.527628670666246,
      "grad_norm": 4.435372829437256,
      "learning_rate": 3.7423253692593765e-05,
      "loss": 0.6188,
      "step": 20673
    },
    {
      "epoch": 6.527944426902431,
      "grad_norm": 4.688698768615723,
      "learning_rate": 3.742237659193769e-05,
      "loss": 0.5239,
      "step": 20674
    },
    {
      "epoch": 6.528260183138617,
      "grad_norm": 5.9054274559021,
      "learning_rate": 3.742149949128162e-05,
      "loss": 0.5541,
      "step": 20675
    },
    {
      "epoch": 6.528575939374803,
      "grad_norm": 6.636408805847168,
      "learning_rate": 3.7420622390625546e-05,
      "loss": 0.5373,
      "step": 20676
    },
    {
      "epoch": 6.528891695610988,
      "grad_norm": 5.133223533630371,
      "learning_rate": 3.741974528996948e-05,
      "loss": 0.5763,
      "step": 20677
    },
    {
      "epoch": 6.529207451847174,
      "grad_norm": 5.363935947418213,
      "learning_rate": 3.74188681893134e-05,
      "loss": 0.5704,
      "step": 20678
    },
    {
      "epoch": 6.52952320808336,
      "grad_norm": 5.93562650680542,
      "learning_rate": 3.7417991088657335e-05,
      "loss": 0.5061,
      "step": 20679
    },
    {
      "epoch": 6.529838964319545,
      "grad_norm": 4.945465087890625,
      "learning_rate": 3.741711398800126e-05,
      "loss": 0.5962,
      "step": 20680
    },
    {
      "epoch": 6.530154720555731,
      "grad_norm": 5.281923294067383,
      "learning_rate": 3.74162368873452e-05,
      "loss": 0.5867,
      "step": 20681
    },
    {
      "epoch": 6.530470476791916,
      "grad_norm": 5.400125503540039,
      "learning_rate": 3.741535978668912e-05,
      "loss": 0.62,
      "step": 20682
    },
    {
      "epoch": 6.530786233028103,
      "grad_norm": 5.273797988891602,
      "learning_rate": 3.7414482686033055e-05,
      "loss": 0.5968,
      "step": 20683
    },
    {
      "epoch": 6.531101989264288,
      "grad_norm": 5.956919193267822,
      "learning_rate": 3.741360558537698e-05,
      "loss": 0.6791,
      "step": 20684
    },
    {
      "epoch": 6.531417745500473,
      "grad_norm": 6.059340476989746,
      "learning_rate": 3.741272848472091e-05,
      "loss": 0.5782,
      "step": 20685
    },
    {
      "epoch": 6.531733501736659,
      "grad_norm": 5.172131061553955,
      "learning_rate": 3.7411851384064836e-05,
      "loss": 0.6008,
      "step": 20686
    },
    {
      "epoch": 6.532049257972845,
      "grad_norm": 5.463650703430176,
      "learning_rate": 3.741097428340877e-05,
      "loss": 0.5339,
      "step": 20687
    },
    {
      "epoch": 6.532365014209031,
      "grad_norm": 4.729908466339111,
      "learning_rate": 3.741009718275269e-05,
      "loss": 0.4753,
      "step": 20688
    },
    {
      "epoch": 6.532680770445216,
      "grad_norm": 5.146338939666748,
      "learning_rate": 3.7409220082096624e-05,
      "loss": 0.6351,
      "step": 20689
    },
    {
      "epoch": 6.532996526681402,
      "grad_norm": 4.915036678314209,
      "learning_rate": 3.7408342981440556e-05,
      "loss": 0.4452,
      "step": 20690
    },
    {
      "epoch": 6.533312282917588,
      "grad_norm": 5.292022228240967,
      "learning_rate": 3.740746588078448e-05,
      "loss": 0.5364,
      "step": 20691
    },
    {
      "epoch": 6.533628039153774,
      "grad_norm": 5.548783302307129,
      "learning_rate": 3.740658878012841e-05,
      "loss": 0.6537,
      "step": 20692
    },
    {
      "epoch": 6.533943795389959,
      "grad_norm": 5.962210655212402,
      "learning_rate": 3.740571167947234e-05,
      "loss": 0.5827,
      "step": 20693
    },
    {
      "epoch": 6.534259551626144,
      "grad_norm": 4.880260467529297,
      "learning_rate": 3.740483457881627e-05,
      "loss": 0.5952,
      "step": 20694
    },
    {
      "epoch": 6.5345753078623305,
      "grad_norm": 5.637571811676025,
      "learning_rate": 3.7403957478160194e-05,
      "loss": 0.5789,
      "step": 20695
    },
    {
      "epoch": 6.534891064098516,
      "grad_norm": 4.197533130645752,
      "learning_rate": 3.7403080377504126e-05,
      "loss": 0.4272,
      "step": 20696
    },
    {
      "epoch": 6.535206820334702,
      "grad_norm": 4.90061616897583,
      "learning_rate": 3.740220327684805e-05,
      "loss": 0.5079,
      "step": 20697
    },
    {
      "epoch": 6.535522576570887,
      "grad_norm": 5.621953010559082,
      "learning_rate": 3.740132617619198e-05,
      "loss": 0.584,
      "step": 20698
    },
    {
      "epoch": 6.535838332807073,
      "grad_norm": 4.80865478515625,
      "learning_rate": 3.740044907553591e-05,
      "loss": 0.6076,
      "step": 20699
    },
    {
      "epoch": 6.536154089043259,
      "grad_norm": 4.205329895019531,
      "learning_rate": 3.739957197487984e-05,
      "loss": 0.5706,
      "step": 20700
    },
    {
      "epoch": 6.536469845279444,
      "grad_norm": 4.281010150909424,
      "learning_rate": 3.7398694874223764e-05,
      "loss": 0.5094,
      "step": 20701
    },
    {
      "epoch": 6.53678560151563,
      "grad_norm": 4.862522125244141,
      "learning_rate": 3.7397817773567695e-05,
      "loss": 0.5733,
      "step": 20702
    },
    {
      "epoch": 6.5371013577518156,
      "grad_norm": 7.077276229858398,
      "learning_rate": 3.739694067291162e-05,
      "loss": 0.6104,
      "step": 20703
    },
    {
      "epoch": 6.537417113988001,
      "grad_norm": 5.259522914886475,
      "learning_rate": 3.739606357225555e-05,
      "loss": 0.572,
      "step": 20704
    },
    {
      "epoch": 6.537732870224187,
      "grad_norm": 4.390298843383789,
      "learning_rate": 3.7395186471599484e-05,
      "loss": 0.6147,
      "step": 20705
    },
    {
      "epoch": 6.538048626460372,
      "grad_norm": 4.843005180358887,
      "learning_rate": 3.739430937094341e-05,
      "loss": 0.6004,
      "step": 20706
    },
    {
      "epoch": 6.5383643826965585,
      "grad_norm": 4.845381259918213,
      "learning_rate": 3.739343227028734e-05,
      "loss": 0.5298,
      "step": 20707
    },
    {
      "epoch": 6.538680138932744,
      "grad_norm": 4.672707557678223,
      "learning_rate": 3.7392555169631265e-05,
      "loss": 0.4968,
      "step": 20708
    },
    {
      "epoch": 6.538995895168929,
      "grad_norm": 5.019414901733398,
      "learning_rate": 3.73916780689752e-05,
      "loss": 0.5465,
      "step": 20709
    },
    {
      "epoch": 6.539311651405115,
      "grad_norm": 4.838311672210693,
      "learning_rate": 3.739080096831913e-05,
      "loss": 0.5128,
      "step": 20710
    },
    {
      "epoch": 6.539627407641301,
      "grad_norm": 3.978668689727783,
      "learning_rate": 3.738992386766306e-05,
      "loss": 0.4693,
      "step": 20711
    },
    {
      "epoch": 6.539943163877487,
      "grad_norm": 7.012866497039795,
      "learning_rate": 3.7389046767006985e-05,
      "loss": 0.5112,
      "step": 20712
    },
    {
      "epoch": 6.540258920113672,
      "grad_norm": 5.76909065246582,
      "learning_rate": 3.738816966635092e-05,
      "loss": 0.7391,
      "step": 20713
    },
    {
      "epoch": 6.540574676349858,
      "grad_norm": 5.163275241851807,
      "learning_rate": 3.738729256569484e-05,
      "loss": 0.5309,
      "step": 20714
    },
    {
      "epoch": 6.5408904325860435,
      "grad_norm": 4.644142150878906,
      "learning_rate": 3.7386415465038773e-05,
      "loss": 0.5754,
      "step": 20715
    },
    {
      "epoch": 6.54120618882223,
      "grad_norm": 5.4321794509887695,
      "learning_rate": 3.73855383643827e-05,
      "loss": 0.6291,
      "step": 20716
    },
    {
      "epoch": 6.541521945058415,
      "grad_norm": 4.746884822845459,
      "learning_rate": 3.738466126372663e-05,
      "loss": 0.4993,
      "step": 20717
    },
    {
      "epoch": 6.5418377012946,
      "grad_norm": 4.569951057434082,
      "learning_rate": 3.7383784163070555e-05,
      "loss": 0.5196,
      "step": 20718
    },
    {
      "epoch": 6.5421534575307865,
      "grad_norm": 5.122334957122803,
      "learning_rate": 3.738290706241449e-05,
      "loss": 0.6534,
      "step": 20719
    },
    {
      "epoch": 6.542469213766972,
      "grad_norm": 5.244229316711426,
      "learning_rate": 3.738202996175841e-05,
      "loss": 0.6485,
      "step": 20720
    },
    {
      "epoch": 6.542784970003158,
      "grad_norm": 4.441320896148682,
      "learning_rate": 3.738115286110234e-05,
      "loss": 0.6088,
      "step": 20721
    },
    {
      "epoch": 6.543100726239343,
      "grad_norm": 5.45968770980835,
      "learning_rate": 3.738027576044627e-05,
      "loss": 0.5587,
      "step": 20722
    },
    {
      "epoch": 6.5434164824755285,
      "grad_norm": 5.613433361053467,
      "learning_rate": 3.73793986597902e-05,
      "loss": 0.5973,
      "step": 20723
    },
    {
      "epoch": 6.543732238711715,
      "grad_norm": 5.033890724182129,
      "learning_rate": 3.7378521559134125e-05,
      "loss": 0.475,
      "step": 20724
    },
    {
      "epoch": 6.5440479949479,
      "grad_norm": 5.330976486206055,
      "learning_rate": 3.7377644458478056e-05,
      "loss": 0.5287,
      "step": 20725
    },
    {
      "epoch": 6.544363751184086,
      "grad_norm": 5.815401077270508,
      "learning_rate": 3.737676735782198e-05,
      "loss": 0.5674,
      "step": 20726
    },
    {
      "epoch": 6.5446795074202715,
      "grad_norm": 4.8454484939575195,
      "learning_rate": 3.737589025716591e-05,
      "loss": 0.4812,
      "step": 20727
    },
    {
      "epoch": 6.544995263656457,
      "grad_norm": 5.408132553100586,
      "learning_rate": 3.7375013156509845e-05,
      "loss": 0.517,
      "step": 20728
    },
    {
      "epoch": 6.545311019892643,
      "grad_norm": 5.0436811447143555,
      "learning_rate": 3.737413605585377e-05,
      "loss": 0.5125,
      "step": 20729
    },
    {
      "epoch": 6.545626776128828,
      "grad_norm": 6.512438774108887,
      "learning_rate": 3.73732589551977e-05,
      "loss": 0.5933,
      "step": 20730
    },
    {
      "epoch": 6.545942532365014,
      "grad_norm": 5.005454063415527,
      "learning_rate": 3.7372381854541626e-05,
      "loss": 0.6001,
      "step": 20731
    },
    {
      "epoch": 6.5462582886012,
      "grad_norm": 4.666779518127441,
      "learning_rate": 3.737150475388556e-05,
      "loss": 0.5061,
      "step": 20732
    },
    {
      "epoch": 6.546574044837386,
      "grad_norm": 5.164219379425049,
      "learning_rate": 3.737062765322948e-05,
      "loss": 0.5211,
      "step": 20733
    },
    {
      "epoch": 6.546889801073571,
      "grad_norm": 5.314627170562744,
      "learning_rate": 3.7369750552573414e-05,
      "loss": 0.5329,
      "step": 20734
    },
    {
      "epoch": 6.5472055573097565,
      "grad_norm": 4.405649185180664,
      "learning_rate": 3.736887345191734e-05,
      "loss": 0.4787,
      "step": 20735
    },
    {
      "epoch": 6.547521313545943,
      "grad_norm": 4.645783424377441,
      "learning_rate": 3.736799635126128e-05,
      "loss": 0.5669,
      "step": 20736
    },
    {
      "epoch": 6.547837069782128,
      "grad_norm": 4.796234607696533,
      "learning_rate": 3.73671192506052e-05,
      "loss": 0.5294,
      "step": 20737
    },
    {
      "epoch": 6.548152826018314,
      "grad_norm": 4.434930324554443,
      "learning_rate": 3.7366242149949134e-05,
      "loss": 0.4543,
      "step": 20738
    },
    {
      "epoch": 6.5484685822544995,
      "grad_norm": 5.246950626373291,
      "learning_rate": 3.736536504929306e-05,
      "loss": 0.6376,
      "step": 20739
    },
    {
      "epoch": 6.548784338490686,
      "grad_norm": 5.327958583831787,
      "learning_rate": 3.736448794863699e-05,
      "loss": 0.5181,
      "step": 20740
    },
    {
      "epoch": 6.549100094726871,
      "grad_norm": 5.978255748748779,
      "learning_rate": 3.7363610847980916e-05,
      "loss": 0.5653,
      "step": 20741
    },
    {
      "epoch": 6.549415850963056,
      "grad_norm": 4.34398078918457,
      "learning_rate": 3.736273374732485e-05,
      "loss": 0.5591,
      "step": 20742
    },
    {
      "epoch": 6.549731607199242,
      "grad_norm": 5.167438507080078,
      "learning_rate": 3.736185664666877e-05,
      "loss": 0.6024,
      "step": 20743
    },
    {
      "epoch": 6.550047363435428,
      "grad_norm": 5.013480186462402,
      "learning_rate": 3.7360979546012704e-05,
      "loss": 0.5003,
      "step": 20744
    },
    {
      "epoch": 6.550363119671614,
      "grad_norm": 5.39769172668457,
      "learning_rate": 3.736010244535663e-05,
      "loss": 0.5681,
      "step": 20745
    },
    {
      "epoch": 6.550678875907799,
      "grad_norm": 4.778980731964111,
      "learning_rate": 3.735922534470056e-05,
      "loss": 0.503,
      "step": 20746
    },
    {
      "epoch": 6.5509946321439845,
      "grad_norm": 5.287981986999512,
      "learning_rate": 3.7358348244044486e-05,
      "loss": 0.563,
      "step": 20747
    },
    {
      "epoch": 6.551310388380171,
      "grad_norm": 5.676577091217041,
      "learning_rate": 3.735747114338842e-05,
      "loss": 0.6321,
      "step": 20748
    },
    {
      "epoch": 6.551626144616356,
      "grad_norm": 4.112998962402344,
      "learning_rate": 3.735659404273235e-05,
      "loss": 0.5752,
      "step": 20749
    },
    {
      "epoch": 6.551941900852542,
      "grad_norm": 5.484244346618652,
      "learning_rate": 3.7355716942076274e-05,
      "loss": 0.6073,
      "step": 20750
    },
    {
      "epoch": 6.552257657088727,
      "grad_norm": 4.429454803466797,
      "learning_rate": 3.7354839841420206e-05,
      "loss": 0.5021,
      "step": 20751
    },
    {
      "epoch": 6.552573413324913,
      "grad_norm": 5.157445430755615,
      "learning_rate": 3.735396274076413e-05,
      "loss": 0.5605,
      "step": 20752
    },
    {
      "epoch": 6.552889169561099,
      "grad_norm": 6.395700454711914,
      "learning_rate": 3.735308564010806e-05,
      "loss": 0.6236,
      "step": 20753
    },
    {
      "epoch": 6.553204925797284,
      "grad_norm": 5.413058757781982,
      "learning_rate": 3.735220853945199e-05,
      "loss": 0.6578,
      "step": 20754
    },
    {
      "epoch": 6.55352068203347,
      "grad_norm": 5.23375940322876,
      "learning_rate": 3.735133143879592e-05,
      "loss": 0.5863,
      "step": 20755
    },
    {
      "epoch": 6.553836438269656,
      "grad_norm": 4.935790538787842,
      "learning_rate": 3.7350454338139844e-05,
      "loss": 0.5825,
      "step": 20756
    },
    {
      "epoch": 6.554152194505842,
      "grad_norm": 5.167540073394775,
      "learning_rate": 3.7349577237483775e-05,
      "loss": 0.6112,
      "step": 20757
    },
    {
      "epoch": 6.554467950742027,
      "grad_norm": 3.914909601211548,
      "learning_rate": 3.73487001368277e-05,
      "loss": 0.5239,
      "step": 20758
    },
    {
      "epoch": 6.554783706978212,
      "grad_norm": 4.547148704528809,
      "learning_rate": 3.734782303617163e-05,
      "loss": 0.5811,
      "step": 20759
    },
    {
      "epoch": 6.555099463214399,
      "grad_norm": 5.879372596740723,
      "learning_rate": 3.734694593551556e-05,
      "loss": 0.5566,
      "step": 20760
    },
    {
      "epoch": 6.555415219450584,
      "grad_norm": 5.047024250030518,
      "learning_rate": 3.734606883485949e-05,
      "loss": 0.6007,
      "step": 20761
    },
    {
      "epoch": 6.55573097568677,
      "grad_norm": 5.2017693519592285,
      "learning_rate": 3.7345191734203413e-05,
      "loss": 0.6326,
      "step": 20762
    },
    {
      "epoch": 6.556046731922955,
      "grad_norm": 5.181841850280762,
      "learning_rate": 3.734431463354735e-05,
      "loss": 0.4858,
      "step": 20763
    },
    {
      "epoch": 6.556362488159142,
      "grad_norm": 5.390955448150635,
      "learning_rate": 3.734343753289128e-05,
      "loss": 0.6732,
      "step": 20764
    },
    {
      "epoch": 6.556678244395327,
      "grad_norm": 4.624913692474365,
      "learning_rate": 3.734256043223521e-05,
      "loss": 0.7015,
      "step": 20765
    },
    {
      "epoch": 6.556994000631512,
      "grad_norm": 5.706642150878906,
      "learning_rate": 3.7341683331579133e-05,
      "loss": 0.5413,
      "step": 20766
    },
    {
      "epoch": 6.557309756867698,
      "grad_norm": 4.534438133239746,
      "learning_rate": 3.7340806230923065e-05,
      "loss": 0.5048,
      "step": 20767
    },
    {
      "epoch": 6.557625513103884,
      "grad_norm": 4.130267143249512,
      "learning_rate": 3.733992913026699e-05,
      "loss": 0.4945,
      "step": 20768
    },
    {
      "epoch": 6.55794126934007,
      "grad_norm": 4.793159008026123,
      "learning_rate": 3.733905202961092e-05,
      "loss": 0.556,
      "step": 20769
    },
    {
      "epoch": 6.558257025576255,
      "grad_norm": 4.268373012542725,
      "learning_rate": 3.7338174928954847e-05,
      "loss": 0.5014,
      "step": 20770
    },
    {
      "epoch": 6.55857278181244,
      "grad_norm": 4.743824005126953,
      "learning_rate": 3.733729782829878e-05,
      "loss": 0.5544,
      "step": 20771
    },
    {
      "epoch": 6.558888538048627,
      "grad_norm": 6.845080852508545,
      "learning_rate": 3.733642072764271e-05,
      "loss": 0.5677,
      "step": 20772
    },
    {
      "epoch": 6.559204294284812,
      "grad_norm": 5.993923664093018,
      "learning_rate": 3.7335543626986635e-05,
      "loss": 0.6144,
      "step": 20773
    },
    {
      "epoch": 6.559520050520998,
      "grad_norm": 5.655149936676025,
      "learning_rate": 3.7334666526330567e-05,
      "loss": 0.5567,
      "step": 20774
    },
    {
      "epoch": 6.559835806757183,
      "grad_norm": 5.734947681427002,
      "learning_rate": 3.733378942567449e-05,
      "loss": 0.5387,
      "step": 20775
    },
    {
      "epoch": 6.560151562993369,
      "grad_norm": 4.224515914916992,
      "learning_rate": 3.733291232501842e-05,
      "loss": 0.5108,
      "step": 20776
    },
    {
      "epoch": 6.560467319229555,
      "grad_norm": 4.911988258361816,
      "learning_rate": 3.733203522436235e-05,
      "loss": 0.5011,
      "step": 20777
    },
    {
      "epoch": 6.56078307546574,
      "grad_norm": 4.65300989151001,
      "learning_rate": 3.733115812370628e-05,
      "loss": 0.5836,
      "step": 20778
    },
    {
      "epoch": 6.561098831701926,
      "grad_norm": 5.926207065582275,
      "learning_rate": 3.7330281023050205e-05,
      "loss": 0.5465,
      "step": 20779
    },
    {
      "epoch": 6.561414587938112,
      "grad_norm": 5.46043062210083,
      "learning_rate": 3.7329403922394136e-05,
      "loss": 0.5833,
      "step": 20780
    },
    {
      "epoch": 6.561730344174298,
      "grad_norm": 4.848517417907715,
      "learning_rate": 3.732852682173806e-05,
      "loss": 0.4947,
      "step": 20781
    },
    {
      "epoch": 6.562046100410483,
      "grad_norm": 6.109355449676514,
      "learning_rate": 3.732764972108199e-05,
      "loss": 0.5569,
      "step": 20782
    },
    {
      "epoch": 6.562361856646669,
      "grad_norm": 4.77367639541626,
      "learning_rate": 3.732677262042592e-05,
      "loss": 0.5034,
      "step": 20783
    },
    {
      "epoch": 6.562677612882855,
      "grad_norm": 5.658452033996582,
      "learning_rate": 3.732589551976985e-05,
      "loss": 0.5102,
      "step": 20784
    },
    {
      "epoch": 6.56299336911904,
      "grad_norm": 5.580920696258545,
      "learning_rate": 3.7325018419113774e-05,
      "loss": 0.5959,
      "step": 20785
    },
    {
      "epoch": 6.563309125355226,
      "grad_norm": 5.133306503295898,
      "learning_rate": 3.7324141318457706e-05,
      "loss": 0.5428,
      "step": 20786
    },
    {
      "epoch": 6.563624881591411,
      "grad_norm": 5.392627239227295,
      "learning_rate": 3.732326421780164e-05,
      "loss": 0.6417,
      "step": 20787
    },
    {
      "epoch": 6.5639406378275975,
      "grad_norm": 4.605578422546387,
      "learning_rate": 3.732238711714556e-05,
      "loss": 0.5032,
      "step": 20788
    },
    {
      "epoch": 6.564256394063783,
      "grad_norm": 5.232134819030762,
      "learning_rate": 3.7321510016489494e-05,
      "loss": 0.4494,
      "step": 20789
    },
    {
      "epoch": 6.564572150299968,
      "grad_norm": 4.89767599105835,
      "learning_rate": 3.7320632915833426e-05,
      "loss": 0.605,
      "step": 20790
    },
    {
      "epoch": 6.564887906536154,
      "grad_norm": 4.902283668518066,
      "learning_rate": 3.731975581517735e-05,
      "loss": 0.589,
      "step": 20791
    },
    {
      "epoch": 6.56520366277234,
      "grad_norm": 5.823939800262451,
      "learning_rate": 3.731887871452128e-05,
      "loss": 0.6325,
      "step": 20792
    },
    {
      "epoch": 6.565519419008526,
      "grad_norm": 4.896152973175049,
      "learning_rate": 3.7318001613865214e-05,
      "loss": 0.4579,
      "step": 20793
    },
    {
      "epoch": 6.565835175244711,
      "grad_norm": 4.876791954040527,
      "learning_rate": 3.731712451320914e-05,
      "loss": 0.5623,
      "step": 20794
    },
    {
      "epoch": 6.566150931480896,
      "grad_norm": 4.365065574645996,
      "learning_rate": 3.731624741255307e-05,
      "loss": 0.5242,
      "step": 20795
    },
    {
      "epoch": 6.5664666877170825,
      "grad_norm": 5.074807167053223,
      "learning_rate": 3.7315370311896996e-05,
      "loss": 0.5999,
      "step": 20796
    },
    {
      "epoch": 6.566782443953268,
      "grad_norm": 5.1204094886779785,
      "learning_rate": 3.731449321124093e-05,
      "loss": 0.5295,
      "step": 20797
    },
    {
      "epoch": 6.567098200189454,
      "grad_norm": 5.215045928955078,
      "learning_rate": 3.731361611058485e-05,
      "loss": 0.6008,
      "step": 20798
    },
    {
      "epoch": 6.567413956425639,
      "grad_norm": 5.4648284912109375,
      "learning_rate": 3.7312739009928784e-05,
      "loss": 0.5902,
      "step": 20799
    },
    {
      "epoch": 6.567729712661825,
      "grad_norm": 4.919531345367432,
      "learning_rate": 3.731186190927271e-05,
      "loss": 0.5588,
      "step": 20800
    },
    {
      "epoch": 6.567729712661825,
      "eval_loss": 0.8688452839851379,
      "eval_runtime": 4.5555,
      "eval_samples_per_second": 546.597,
      "eval_steps_per_second": 4.61,
      "step": 20800
    },
    {
      "epoch": 6.568045468898011,
      "grad_norm": 4.988650798797607,
      "learning_rate": 3.731098480861664e-05,
      "loss": 0.5682,
      "step": 20801
    },
    {
      "epoch": 6.568361225134196,
      "grad_norm": 5.16914701461792,
      "learning_rate": 3.7310107707960566e-05,
      "loss": 0.5673,
      "step": 20802
    },
    {
      "epoch": 6.568676981370382,
      "grad_norm": 4.815414905548096,
      "learning_rate": 3.73092306073045e-05,
      "loss": 0.5205,
      "step": 20803
    },
    {
      "epoch": 6.5689927376065675,
      "grad_norm": 5.142170429229736,
      "learning_rate": 3.730835350664842e-05,
      "loss": 0.5057,
      "step": 20804
    },
    {
      "epoch": 6.569308493842754,
      "grad_norm": 5.347635746002197,
      "learning_rate": 3.7307476405992354e-05,
      "loss": 0.5334,
      "step": 20805
    },
    {
      "epoch": 6.569624250078939,
      "grad_norm": 4.986781597137451,
      "learning_rate": 3.730659930533628e-05,
      "loss": 0.6822,
      "step": 20806
    },
    {
      "epoch": 6.569940006315125,
      "grad_norm": 5.613041400909424,
      "learning_rate": 3.730572220468021e-05,
      "loss": 0.6178,
      "step": 20807
    },
    {
      "epoch": 6.5702557625513105,
      "grad_norm": 4.714512825012207,
      "learning_rate": 3.7304845104024135e-05,
      "loss": 0.5084,
      "step": 20808
    },
    {
      "epoch": 6.570571518787496,
      "grad_norm": 5.535714149475098,
      "learning_rate": 3.730396800336807e-05,
      "loss": 0.5634,
      "step": 20809
    },
    {
      "epoch": 6.570887275023682,
      "grad_norm": 5.302710056304932,
      "learning_rate": 3.7303090902712e-05,
      "loss": 0.4959,
      "step": 20810
    },
    {
      "epoch": 6.571203031259867,
      "grad_norm": 4.820706367492676,
      "learning_rate": 3.7302213802055924e-05,
      "loss": 0.5753,
      "step": 20811
    },
    {
      "epoch": 6.5715187874960534,
      "grad_norm": 5.00747537612915,
      "learning_rate": 3.7301336701399855e-05,
      "loss": 0.5589,
      "step": 20812
    },
    {
      "epoch": 6.571834543732239,
      "grad_norm": 5.77482795715332,
      "learning_rate": 3.730045960074378e-05,
      "loss": 0.6121,
      "step": 20813
    },
    {
      "epoch": 6.572150299968424,
      "grad_norm": 5.863575458526611,
      "learning_rate": 3.729958250008771e-05,
      "loss": 0.5621,
      "step": 20814
    },
    {
      "epoch": 6.57246605620461,
      "grad_norm": 5.972341537475586,
      "learning_rate": 3.729870539943164e-05,
      "loss": 0.5079,
      "step": 20815
    },
    {
      "epoch": 6.5727818124407955,
      "grad_norm": 6.032840728759766,
      "learning_rate": 3.729782829877557e-05,
      "loss": 0.5183,
      "step": 20816
    },
    {
      "epoch": 6.573097568676982,
      "grad_norm": 4.431295871734619,
      "learning_rate": 3.729695119811949e-05,
      "loss": 0.4273,
      "step": 20817
    },
    {
      "epoch": 6.573413324913167,
      "grad_norm": 4.558930397033691,
      "learning_rate": 3.729607409746343e-05,
      "loss": 0.6196,
      "step": 20818
    },
    {
      "epoch": 6.573729081149352,
      "grad_norm": 5.289233684539795,
      "learning_rate": 3.729519699680736e-05,
      "loss": 0.587,
      "step": 20819
    },
    {
      "epoch": 6.5740448373855385,
      "grad_norm": 5.0924153327941895,
      "learning_rate": 3.729431989615129e-05,
      "loss": 0.6805,
      "step": 20820
    },
    {
      "epoch": 6.574360593621724,
      "grad_norm": 6.4168243408203125,
      "learning_rate": 3.729344279549521e-05,
      "loss": 0.626,
      "step": 20821
    },
    {
      "epoch": 6.57467634985791,
      "grad_norm": 4.553669452667236,
      "learning_rate": 3.7292565694839145e-05,
      "loss": 0.6331,
      "step": 20822
    },
    {
      "epoch": 6.574992106094095,
      "grad_norm": 4.787689685821533,
      "learning_rate": 3.729168859418307e-05,
      "loss": 0.518,
      "step": 20823
    },
    {
      "epoch": 6.5753078623302805,
      "grad_norm": 4.420285701751709,
      "learning_rate": 3.7290811493527e-05,
      "loss": 0.5327,
      "step": 20824
    },
    {
      "epoch": 6.575623618566467,
      "grad_norm": 3.990722179412842,
      "learning_rate": 3.7289934392870926e-05,
      "loss": 0.5453,
      "step": 20825
    },
    {
      "epoch": 6.575939374802652,
      "grad_norm": 5.758113861083984,
      "learning_rate": 3.728905729221486e-05,
      "loss": 0.6587,
      "step": 20826
    },
    {
      "epoch": 6.576255131038838,
      "grad_norm": 4.4411187171936035,
      "learning_rate": 3.728818019155878e-05,
      "loss": 0.466,
      "step": 20827
    },
    {
      "epoch": 6.5765708872750235,
      "grad_norm": 4.6567702293396,
      "learning_rate": 3.7287303090902715e-05,
      "loss": 0.5496,
      "step": 20828
    },
    {
      "epoch": 6.57688664351121,
      "grad_norm": 5.611785411834717,
      "learning_rate": 3.728642599024664e-05,
      "loss": 0.48,
      "step": 20829
    },
    {
      "epoch": 6.577202399747395,
      "grad_norm": 4.9815754890441895,
      "learning_rate": 3.728554888959057e-05,
      "loss": 0.6027,
      "step": 20830
    },
    {
      "epoch": 6.577518155983581,
      "grad_norm": 4.291624069213867,
      "learning_rate": 3.72846717889345e-05,
      "loss": 0.4516,
      "step": 20831
    },
    {
      "epoch": 6.577833912219766,
      "grad_norm": 5.49895715713501,
      "learning_rate": 3.728379468827843e-05,
      "loss": 0.6708,
      "step": 20832
    },
    {
      "epoch": 6.578149668455952,
      "grad_norm": 4.016472339630127,
      "learning_rate": 3.728291758762236e-05,
      "loss": 0.5392,
      "step": 20833
    },
    {
      "epoch": 6.578465424692138,
      "grad_norm": 4.258761405944824,
      "learning_rate": 3.7282040486966285e-05,
      "loss": 0.5138,
      "step": 20834
    },
    {
      "epoch": 6.578781180928323,
      "grad_norm": 4.670604228973389,
      "learning_rate": 3.7281163386310216e-05,
      "loss": 0.5751,
      "step": 20835
    },
    {
      "epoch": 6.579096937164509,
      "grad_norm": 4.897226810455322,
      "learning_rate": 3.728028628565414e-05,
      "loss": 0.578,
      "step": 20836
    },
    {
      "epoch": 6.579412693400695,
      "grad_norm": 5.6565656661987305,
      "learning_rate": 3.727940918499807e-05,
      "loss": 0.6119,
      "step": 20837
    },
    {
      "epoch": 6.57972844963688,
      "grad_norm": 4.659353256225586,
      "learning_rate": 3.7278532084342e-05,
      "loss": 0.5857,
      "step": 20838
    },
    {
      "epoch": 6.580044205873066,
      "grad_norm": 4.275335311889648,
      "learning_rate": 3.727765498368593e-05,
      "loss": 0.5014,
      "step": 20839
    },
    {
      "epoch": 6.5803599621092514,
      "grad_norm": 5.272615909576416,
      "learning_rate": 3.7276777883029854e-05,
      "loss": 0.6114,
      "step": 20840
    },
    {
      "epoch": 6.580675718345438,
      "grad_norm": 3.9714369773864746,
      "learning_rate": 3.7275900782373786e-05,
      "loss": 0.5084,
      "step": 20841
    },
    {
      "epoch": 6.580991474581623,
      "grad_norm": 4.921871662139893,
      "learning_rate": 3.727502368171771e-05,
      "loss": 0.5164,
      "step": 20842
    },
    {
      "epoch": 6.581307230817808,
      "grad_norm": 4.636440277099609,
      "learning_rate": 3.727414658106164e-05,
      "loss": 0.558,
      "step": 20843
    },
    {
      "epoch": 6.581622987053994,
      "grad_norm": 5.1487603187561035,
      "learning_rate": 3.727326948040557e-05,
      "loss": 0.612,
      "step": 20844
    },
    {
      "epoch": 6.58193874329018,
      "grad_norm": 4.928896427154541,
      "learning_rate": 3.7272392379749506e-05,
      "loss": 0.5796,
      "step": 20845
    },
    {
      "epoch": 6.582254499526366,
      "grad_norm": 4.1439080238342285,
      "learning_rate": 3.727151527909343e-05,
      "loss": 0.5568,
      "step": 20846
    },
    {
      "epoch": 6.582570255762551,
      "grad_norm": 5.342835426330566,
      "learning_rate": 3.727063817843736e-05,
      "loss": 0.6005,
      "step": 20847
    },
    {
      "epoch": 6.582886011998737,
      "grad_norm": 4.627394676208496,
      "learning_rate": 3.726976107778129e-05,
      "loss": 0.6467,
      "step": 20848
    },
    {
      "epoch": 6.583201768234923,
      "grad_norm": 5.580790996551514,
      "learning_rate": 3.726888397712522e-05,
      "loss": 0.5529,
      "step": 20849
    },
    {
      "epoch": 6.583517524471108,
      "grad_norm": 4.311312198638916,
      "learning_rate": 3.7268006876469144e-05,
      "loss": 0.5489,
      "step": 20850
    },
    {
      "epoch": 6.583833280707294,
      "grad_norm": 6.12802791595459,
      "learning_rate": 3.7267129775813076e-05,
      "loss": 0.6196,
      "step": 20851
    },
    {
      "epoch": 6.584149036943479,
      "grad_norm": 3.8895132541656494,
      "learning_rate": 3.726625267515701e-05,
      "loss": 0.4597,
      "step": 20852
    },
    {
      "epoch": 6.584464793179666,
      "grad_norm": 7.2533860206604,
      "learning_rate": 3.726537557450093e-05,
      "loss": 0.6152,
      "step": 20853
    },
    {
      "epoch": 6.584780549415851,
      "grad_norm": 4.254279613494873,
      "learning_rate": 3.7264498473844864e-05,
      "loss": 0.4793,
      "step": 20854
    },
    {
      "epoch": 6.585096305652037,
      "grad_norm": 5.518122673034668,
      "learning_rate": 3.726362137318879e-05,
      "loss": 0.5768,
      "step": 20855
    },
    {
      "epoch": 6.585412061888222,
      "grad_norm": 5.152459621429443,
      "learning_rate": 3.726274427253272e-05,
      "loss": 0.5823,
      "step": 20856
    },
    {
      "epoch": 6.585727818124408,
      "grad_norm": 5.170602321624756,
      "learning_rate": 3.7261867171876645e-05,
      "loss": 0.5164,
      "step": 20857
    },
    {
      "epoch": 6.586043574360594,
      "grad_norm": 5.449825286865234,
      "learning_rate": 3.726099007122058e-05,
      "loss": 0.6675,
      "step": 20858
    },
    {
      "epoch": 6.586359330596779,
      "grad_norm": 4.382550239562988,
      "learning_rate": 3.72601129705645e-05,
      "loss": 0.5717,
      "step": 20859
    },
    {
      "epoch": 6.586675086832965,
      "grad_norm": 4.489309787750244,
      "learning_rate": 3.7259235869908434e-05,
      "loss": 0.5268,
      "step": 20860
    },
    {
      "epoch": 6.586990843069151,
      "grad_norm": 5.366863250732422,
      "learning_rate": 3.725835876925236e-05,
      "loss": 0.595,
      "step": 20861
    },
    {
      "epoch": 6.587306599305336,
      "grad_norm": 6.092296123504639,
      "learning_rate": 3.725748166859629e-05,
      "loss": 0.6161,
      "step": 20862
    },
    {
      "epoch": 6.587622355541522,
      "grad_norm": 5.472152233123779,
      "learning_rate": 3.7256604567940215e-05,
      "loss": 0.6025,
      "step": 20863
    },
    {
      "epoch": 6.587938111777707,
      "grad_norm": 4.861530780792236,
      "learning_rate": 3.725572746728415e-05,
      "loss": 0.5472,
      "step": 20864
    },
    {
      "epoch": 6.588253868013894,
      "grad_norm": 4.461004257202148,
      "learning_rate": 3.725485036662807e-05,
      "loss": 0.5452,
      "step": 20865
    },
    {
      "epoch": 6.588569624250079,
      "grad_norm": 5.415480136871338,
      "learning_rate": 3.7253973265972003e-05,
      "loss": 0.5954,
      "step": 20866
    },
    {
      "epoch": 6.588885380486264,
      "grad_norm": 5.070541858673096,
      "learning_rate": 3.725309616531593e-05,
      "loss": 0.6136,
      "step": 20867
    },
    {
      "epoch": 6.58920113672245,
      "grad_norm": 5.067947864532471,
      "learning_rate": 3.725221906465986e-05,
      "loss": 0.5986,
      "step": 20868
    },
    {
      "epoch": 6.589516892958636,
      "grad_norm": 4.699738025665283,
      "learning_rate": 3.725134196400379e-05,
      "loss": 0.5194,
      "step": 20869
    },
    {
      "epoch": 6.589832649194822,
      "grad_norm": 4.961620330810547,
      "learning_rate": 3.725046486334772e-05,
      "loss": 0.5232,
      "step": 20870
    },
    {
      "epoch": 6.590148405431007,
      "grad_norm": 4.592791557312012,
      "learning_rate": 3.724958776269165e-05,
      "loss": 0.4885,
      "step": 20871
    },
    {
      "epoch": 6.590464161667193,
      "grad_norm": 4.090299606323242,
      "learning_rate": 3.724871066203558e-05,
      "loss": 0.3998,
      "step": 20872
    },
    {
      "epoch": 6.590779917903379,
      "grad_norm": 5.288843154907227,
      "learning_rate": 3.7247833561379505e-05,
      "loss": 0.5304,
      "step": 20873
    },
    {
      "epoch": 6.591095674139564,
      "grad_norm": 4.548080921173096,
      "learning_rate": 3.724695646072344e-05,
      "loss": 0.4768,
      "step": 20874
    },
    {
      "epoch": 6.59141143037575,
      "grad_norm": 4.274195194244385,
      "learning_rate": 3.724607936006737e-05,
      "loss": 0.5587,
      "step": 20875
    },
    {
      "epoch": 6.591727186611935,
      "grad_norm": 4.421107769012451,
      "learning_rate": 3.724520225941129e-05,
      "loss": 0.494,
      "step": 20876
    },
    {
      "epoch": 6.5920429428481215,
      "grad_norm": 5.505157947540283,
      "learning_rate": 3.7244325158755225e-05,
      "loss": 0.6373,
      "step": 20877
    },
    {
      "epoch": 6.592358699084307,
      "grad_norm": 5.037569046020508,
      "learning_rate": 3.724344805809915e-05,
      "loss": 0.5644,
      "step": 20878
    },
    {
      "epoch": 6.592674455320493,
      "grad_norm": 4.91302490234375,
      "learning_rate": 3.724257095744308e-05,
      "loss": 0.648,
      "step": 20879
    },
    {
      "epoch": 6.592990211556678,
      "grad_norm": 4.700779438018799,
      "learning_rate": 3.7241693856787006e-05,
      "loss": 0.5245,
      "step": 20880
    },
    {
      "epoch": 6.593305967792864,
      "grad_norm": 5.267431259155273,
      "learning_rate": 3.724081675613094e-05,
      "loss": 0.5356,
      "step": 20881
    },
    {
      "epoch": 6.59362172402905,
      "grad_norm": 4.265410423278809,
      "learning_rate": 3.723993965547486e-05,
      "loss": 0.4997,
      "step": 20882
    },
    {
      "epoch": 6.593937480265235,
      "grad_norm": 5.554671764373779,
      "learning_rate": 3.7239062554818795e-05,
      "loss": 0.5716,
      "step": 20883
    },
    {
      "epoch": 6.594253236501421,
      "grad_norm": 5.415224552154541,
      "learning_rate": 3.723818545416272e-05,
      "loss": 0.5441,
      "step": 20884
    },
    {
      "epoch": 6.5945689927376065,
      "grad_norm": 5.217523097991943,
      "learning_rate": 3.723730835350665e-05,
      "loss": 0.5296,
      "step": 20885
    },
    {
      "epoch": 6.594884748973792,
      "grad_norm": 6.145662784576416,
      "learning_rate": 3.7236431252850576e-05,
      "loss": 0.5026,
      "step": 20886
    },
    {
      "epoch": 6.595200505209978,
      "grad_norm": 6.049778461456299,
      "learning_rate": 3.723555415219451e-05,
      "loss": 0.5428,
      "step": 20887
    },
    {
      "epoch": 6.595516261446163,
      "grad_norm": 5.317209243774414,
      "learning_rate": 3.723467705153843e-05,
      "loss": 0.5221,
      "step": 20888
    },
    {
      "epoch": 6.5958320176823495,
      "grad_norm": 5.369644641876221,
      "learning_rate": 3.7233799950882364e-05,
      "loss": 0.6623,
      "step": 20889
    },
    {
      "epoch": 6.596147773918535,
      "grad_norm": 4.710731029510498,
      "learning_rate": 3.7232922850226296e-05,
      "loss": 0.4707,
      "step": 20890
    },
    {
      "epoch": 6.59646353015472,
      "grad_norm": 5.210209369659424,
      "learning_rate": 3.723204574957022e-05,
      "loss": 0.4746,
      "step": 20891
    },
    {
      "epoch": 6.596779286390906,
      "grad_norm": 6.575198173522949,
      "learning_rate": 3.723116864891415e-05,
      "loss": 0.5466,
      "step": 20892
    },
    {
      "epoch": 6.597095042627092,
      "grad_norm": 5.025267601013184,
      "learning_rate": 3.723029154825808e-05,
      "loss": 0.5785,
      "step": 20893
    },
    {
      "epoch": 6.597410798863278,
      "grad_norm": 4.929516792297363,
      "learning_rate": 3.722941444760201e-05,
      "loss": 0.6171,
      "step": 20894
    },
    {
      "epoch": 6.597726555099463,
      "grad_norm": 6.091232776641846,
      "learning_rate": 3.7228537346945934e-05,
      "loss": 0.5436,
      "step": 20895
    },
    {
      "epoch": 6.598042311335649,
      "grad_norm": 5.57139778137207,
      "learning_rate": 3.7227660246289866e-05,
      "loss": 0.5045,
      "step": 20896
    },
    {
      "epoch": 6.5983580675718345,
      "grad_norm": 4.375288486480713,
      "learning_rate": 3.722678314563379e-05,
      "loss": 0.38,
      "step": 20897
    },
    {
      "epoch": 6.598673823808021,
      "grad_norm": 5.213824272155762,
      "learning_rate": 3.722590604497772e-05,
      "loss": 0.4911,
      "step": 20898
    },
    {
      "epoch": 6.598989580044206,
      "grad_norm": 4.806632995605469,
      "learning_rate": 3.7225028944321654e-05,
      "loss": 0.5714,
      "step": 20899
    },
    {
      "epoch": 6.599305336280391,
      "grad_norm": 5.387782096862793,
      "learning_rate": 3.7224151843665586e-05,
      "loss": 0.5787,
      "step": 20900
    },
    {
      "epoch": 6.5996210925165775,
      "grad_norm": 5.684221267700195,
      "learning_rate": 3.722327474300951e-05,
      "loss": 0.6101,
      "step": 20901
    },
    {
      "epoch": 6.599936848752763,
      "grad_norm": 4.382937431335449,
      "learning_rate": 3.722239764235344e-05,
      "loss": 0.5355,
      "step": 20902
    },
    {
      "epoch": 6.600252604988949,
      "grad_norm": 5.264070987701416,
      "learning_rate": 3.722152054169737e-05,
      "loss": 0.5891,
      "step": 20903
    },
    {
      "epoch": 6.600568361225134,
      "grad_norm": 5.880817413330078,
      "learning_rate": 3.72206434410413e-05,
      "loss": 0.5921,
      "step": 20904
    },
    {
      "epoch": 6.6008841174613195,
      "grad_norm": 6.03432035446167,
      "learning_rate": 3.7219766340385224e-05,
      "loss": 0.6478,
      "step": 20905
    },
    {
      "epoch": 6.601199873697506,
      "grad_norm": 4.714260578155518,
      "learning_rate": 3.7218889239729156e-05,
      "loss": 0.5196,
      "step": 20906
    },
    {
      "epoch": 6.601515629933691,
      "grad_norm": 4.825051784515381,
      "learning_rate": 3.721801213907308e-05,
      "loss": 0.5815,
      "step": 20907
    },
    {
      "epoch": 6.601831386169877,
      "grad_norm": 4.664528846740723,
      "learning_rate": 3.721713503841701e-05,
      "loss": 0.5886,
      "step": 20908
    },
    {
      "epoch": 6.6021471424060625,
      "grad_norm": 5.775585651397705,
      "learning_rate": 3.721625793776094e-05,
      "loss": 0.5858,
      "step": 20909
    },
    {
      "epoch": 6.602462898642248,
      "grad_norm": 5.1164445877075195,
      "learning_rate": 3.721538083710487e-05,
      "loss": 0.6269,
      "step": 20910
    },
    {
      "epoch": 6.602778654878434,
      "grad_norm": 5.860571384429932,
      "learning_rate": 3.7214503736448794e-05,
      "loss": 0.5907,
      "step": 20911
    },
    {
      "epoch": 6.603094411114619,
      "grad_norm": 5.829842567443848,
      "learning_rate": 3.7213626635792725e-05,
      "loss": 0.5297,
      "step": 20912
    },
    {
      "epoch": 6.603410167350805,
      "grad_norm": 6.784411907196045,
      "learning_rate": 3.721274953513666e-05,
      "loss": 0.5467,
      "step": 20913
    },
    {
      "epoch": 6.603725923586991,
      "grad_norm": 4.554996013641357,
      "learning_rate": 3.721187243448058e-05,
      "loss": 0.6461,
      "step": 20914
    },
    {
      "epoch": 6.604041679823176,
      "grad_norm": 5.6519083976745605,
      "learning_rate": 3.7210995333824514e-05,
      "loss": 0.6162,
      "step": 20915
    },
    {
      "epoch": 6.604357436059362,
      "grad_norm": 4.696691036224365,
      "learning_rate": 3.721011823316844e-05,
      "loss": 0.6027,
      "step": 20916
    },
    {
      "epoch": 6.6046731922955475,
      "grad_norm": 4.7890801429748535,
      "learning_rate": 3.720924113251237e-05,
      "loss": 0.6006,
      "step": 20917
    },
    {
      "epoch": 6.604988948531734,
      "grad_norm": 5.482363700866699,
      "learning_rate": 3.7208364031856295e-05,
      "loss": 0.5824,
      "step": 20918
    },
    {
      "epoch": 6.605304704767919,
      "grad_norm": 4.799261569976807,
      "learning_rate": 3.720748693120023e-05,
      "loss": 0.4332,
      "step": 20919
    },
    {
      "epoch": 6.605620461004105,
      "grad_norm": 5.327181816101074,
      "learning_rate": 3.720660983054415e-05,
      "loss": 0.5389,
      "step": 20920
    },
    {
      "epoch": 6.6059362172402905,
      "grad_norm": 4.244172096252441,
      "learning_rate": 3.7205732729888083e-05,
      "loss": 0.4999,
      "step": 20921
    },
    {
      "epoch": 6.606251973476477,
      "grad_norm": 4.349152565002441,
      "learning_rate": 3.720485562923201e-05,
      "loss": 0.4956,
      "step": 20922
    },
    {
      "epoch": 6.606567729712662,
      "grad_norm": 4.628847122192383,
      "learning_rate": 3.720397852857594e-05,
      "loss": 0.4697,
      "step": 20923
    },
    {
      "epoch": 6.606883485948847,
      "grad_norm": 6.189316749572754,
      "learning_rate": 3.7203101427919865e-05,
      "loss": 0.6539,
      "step": 20924
    },
    {
      "epoch": 6.607199242185033,
      "grad_norm": 4.38358736038208,
      "learning_rate": 3.7202224327263797e-05,
      "loss": 0.5313,
      "step": 20925
    },
    {
      "epoch": 6.607514998421219,
      "grad_norm": 4.480069160461426,
      "learning_rate": 3.720134722660772e-05,
      "loss": 0.6065,
      "step": 20926
    },
    {
      "epoch": 6.607830754657405,
      "grad_norm": 4.975532531738281,
      "learning_rate": 3.720047012595166e-05,
      "loss": 0.6752,
      "step": 20927
    },
    {
      "epoch": 6.60814651089359,
      "grad_norm": 4.747028827667236,
      "learning_rate": 3.7199593025295585e-05,
      "loss": 0.4971,
      "step": 20928
    },
    {
      "epoch": 6.6084622671297755,
      "grad_norm": 5.17538595199585,
      "learning_rate": 3.7198715924639517e-05,
      "loss": 0.5028,
      "step": 20929
    },
    {
      "epoch": 6.608778023365962,
      "grad_norm": 5.655113220214844,
      "learning_rate": 3.719783882398344e-05,
      "loss": 0.5813,
      "step": 20930
    },
    {
      "epoch": 6.609093779602147,
      "grad_norm": 4.526494979858398,
      "learning_rate": 3.719696172332737e-05,
      "loss": 0.5621,
      "step": 20931
    },
    {
      "epoch": 6.609409535838333,
      "grad_norm": 4.800119400024414,
      "learning_rate": 3.71960846226713e-05,
      "loss": 0.5686,
      "step": 20932
    },
    {
      "epoch": 6.609725292074518,
      "grad_norm": 4.667267322540283,
      "learning_rate": 3.719520752201523e-05,
      "loss": 0.6032,
      "step": 20933
    },
    {
      "epoch": 6.610041048310704,
      "grad_norm": 4.399743556976318,
      "learning_rate": 3.719433042135916e-05,
      "loss": 0.57,
      "step": 20934
    },
    {
      "epoch": 6.61035680454689,
      "grad_norm": 4.809208869934082,
      "learning_rate": 3.7193453320703086e-05,
      "loss": 0.6859,
      "step": 20935
    },
    {
      "epoch": 6.610672560783075,
      "grad_norm": 4.822405815124512,
      "learning_rate": 3.719257622004702e-05,
      "loss": 0.5461,
      "step": 20936
    },
    {
      "epoch": 6.610988317019261,
      "grad_norm": 4.601648330688477,
      "learning_rate": 3.719169911939094e-05,
      "loss": 0.5171,
      "step": 20937
    },
    {
      "epoch": 6.611304073255447,
      "grad_norm": 5.716212749481201,
      "learning_rate": 3.7190822018734875e-05,
      "loss": 0.6463,
      "step": 20938
    },
    {
      "epoch": 6.611619829491632,
      "grad_norm": 5.694417476654053,
      "learning_rate": 3.71899449180788e-05,
      "loss": 0.6553,
      "step": 20939
    },
    {
      "epoch": 6.611935585727818,
      "grad_norm": 4.517592906951904,
      "learning_rate": 3.718906781742273e-05,
      "loss": 0.514,
      "step": 20940
    },
    {
      "epoch": 6.612251341964003,
      "grad_norm": 5.187070846557617,
      "learning_rate": 3.7188190716766656e-05,
      "loss": 0.6026,
      "step": 20941
    },
    {
      "epoch": 6.61256709820019,
      "grad_norm": 4.496438980102539,
      "learning_rate": 3.718731361611059e-05,
      "loss": 0.5489,
      "step": 20942
    },
    {
      "epoch": 6.612882854436375,
      "grad_norm": 4.611940860748291,
      "learning_rate": 3.718643651545451e-05,
      "loss": 0.5205,
      "step": 20943
    },
    {
      "epoch": 6.613198610672561,
      "grad_norm": 5.052036285400391,
      "learning_rate": 3.7185559414798444e-05,
      "loss": 0.504,
      "step": 20944
    },
    {
      "epoch": 6.613514366908746,
      "grad_norm": 5.370755672454834,
      "learning_rate": 3.718468231414237e-05,
      "loss": 0.674,
      "step": 20945
    },
    {
      "epoch": 6.613830123144933,
      "grad_norm": 4.43596076965332,
      "learning_rate": 3.71838052134863e-05,
      "loss": 0.4401,
      "step": 20946
    },
    {
      "epoch": 6.614145879381118,
      "grad_norm": 5.402243614196777,
      "learning_rate": 3.7182928112830226e-05,
      "loss": 0.5519,
      "step": 20947
    },
    {
      "epoch": 6.614461635617303,
      "grad_norm": 4.418890476226807,
      "learning_rate": 3.718205101217416e-05,
      "loss": 0.5339,
      "step": 20948
    },
    {
      "epoch": 6.614777391853489,
      "grad_norm": 5.016228675842285,
      "learning_rate": 3.718117391151808e-05,
      "loss": 0.5562,
      "step": 20949
    },
    {
      "epoch": 6.615093148089675,
      "grad_norm": 5.205928802490234,
      "learning_rate": 3.7180296810862014e-05,
      "loss": 0.6644,
      "step": 20950
    },
    {
      "epoch": 6.615408904325861,
      "grad_norm": 4.56831169128418,
      "learning_rate": 3.7179419710205946e-05,
      "loss": 0.54,
      "step": 20951
    },
    {
      "epoch": 6.615724660562046,
      "grad_norm": 4.220328330993652,
      "learning_rate": 3.717854260954987e-05,
      "loss": 0.5627,
      "step": 20952
    },
    {
      "epoch": 6.616040416798231,
      "grad_norm": 4.53402853012085,
      "learning_rate": 3.71776655088938e-05,
      "loss": 0.5676,
      "step": 20953
    },
    {
      "epoch": 6.616356173034418,
      "grad_norm": 4.979076862335205,
      "learning_rate": 3.7176788408237734e-05,
      "loss": 0.534,
      "step": 20954
    },
    {
      "epoch": 6.616671929270603,
      "grad_norm": 4.753332138061523,
      "learning_rate": 3.717591130758166e-05,
      "loss": 0.4852,
      "step": 20955
    },
    {
      "epoch": 6.616987685506789,
      "grad_norm": 5.760063171386719,
      "learning_rate": 3.717503420692559e-05,
      "loss": 0.5074,
      "step": 20956
    },
    {
      "epoch": 6.617303441742974,
      "grad_norm": 4.55338716506958,
      "learning_rate": 3.717415710626952e-05,
      "loss": 0.5533,
      "step": 20957
    },
    {
      "epoch": 6.61761919797916,
      "grad_norm": 4.580760955810547,
      "learning_rate": 3.717328000561345e-05,
      "loss": 0.5529,
      "step": 20958
    },
    {
      "epoch": 6.617934954215346,
      "grad_norm": 4.580586910247803,
      "learning_rate": 3.717240290495738e-05,
      "loss": 0.45,
      "step": 20959
    },
    {
      "epoch": 6.618250710451531,
      "grad_norm": 4.858741283416748,
      "learning_rate": 3.7171525804301304e-05,
      "loss": 0.4896,
      "step": 20960
    },
    {
      "epoch": 6.618566466687717,
      "grad_norm": 5.352046966552734,
      "learning_rate": 3.7170648703645235e-05,
      "loss": 0.536,
      "step": 20961
    },
    {
      "epoch": 6.618882222923903,
      "grad_norm": 6.651905536651611,
      "learning_rate": 3.716977160298916e-05,
      "loss": 0.6288,
      "step": 20962
    },
    {
      "epoch": 6.619197979160089,
      "grad_norm": 5.305641174316406,
      "learning_rate": 3.716889450233309e-05,
      "loss": 0.574,
      "step": 20963
    },
    {
      "epoch": 6.619513735396274,
      "grad_norm": 4.94942569732666,
      "learning_rate": 3.716801740167702e-05,
      "loss": 0.6585,
      "step": 20964
    },
    {
      "epoch": 6.619829491632459,
      "grad_norm": 6.493513107299805,
      "learning_rate": 3.716714030102095e-05,
      "loss": 0.6753,
      "step": 20965
    },
    {
      "epoch": 6.6201452478686456,
      "grad_norm": 5.79124116897583,
      "learning_rate": 3.7166263200364874e-05,
      "loss": 0.5523,
      "step": 20966
    },
    {
      "epoch": 6.620461004104831,
      "grad_norm": 4.747196197509766,
      "learning_rate": 3.7165386099708805e-05,
      "loss": 0.6084,
      "step": 20967
    },
    {
      "epoch": 6.620776760341017,
      "grad_norm": 5.778109073638916,
      "learning_rate": 3.716450899905273e-05,
      "loss": 0.5325,
      "step": 20968
    },
    {
      "epoch": 6.621092516577202,
      "grad_norm": 5.721806049346924,
      "learning_rate": 3.716363189839666e-05,
      "loss": 0.5864,
      "step": 20969
    },
    {
      "epoch": 6.6214082728133885,
      "grad_norm": 5.363094806671143,
      "learning_rate": 3.716275479774059e-05,
      "loss": 0.6022,
      "step": 20970
    },
    {
      "epoch": 6.621724029049574,
      "grad_norm": 5.189262866973877,
      "learning_rate": 3.716187769708452e-05,
      "loss": 0.4933,
      "step": 20971
    },
    {
      "epoch": 6.622039785285759,
      "grad_norm": 4.654143333435059,
      "learning_rate": 3.716100059642845e-05,
      "loss": 0.5514,
      "step": 20972
    },
    {
      "epoch": 6.622355541521945,
      "grad_norm": 4.484548091888428,
      "learning_rate": 3.7160123495772375e-05,
      "loss": 0.4811,
      "step": 20973
    },
    {
      "epoch": 6.622671297758131,
      "grad_norm": 4.632720470428467,
      "learning_rate": 3.715924639511631e-05,
      "loss": 0.553,
      "step": 20974
    },
    {
      "epoch": 6.622987053994317,
      "grad_norm": 5.771614074707031,
      "learning_rate": 3.715836929446023e-05,
      "loss": 0.6382,
      "step": 20975
    },
    {
      "epoch": 6.623302810230502,
      "grad_norm": 5.004924297332764,
      "learning_rate": 3.715749219380416e-05,
      "loss": 0.4931,
      "step": 20976
    },
    {
      "epoch": 6.623618566466687,
      "grad_norm": 4.778469562530518,
      "learning_rate": 3.715661509314809e-05,
      "loss": 0.5561,
      "step": 20977
    },
    {
      "epoch": 6.6239343227028735,
      "grad_norm": 15.942957878112793,
      "learning_rate": 3.715573799249202e-05,
      "loss": 0.6267,
      "step": 20978
    },
    {
      "epoch": 6.624250078939059,
      "grad_norm": 4.546548843383789,
      "learning_rate": 3.7154860891835945e-05,
      "loss": 0.5984,
      "step": 20979
    },
    {
      "epoch": 6.624565835175245,
      "grad_norm": 5.13936710357666,
      "learning_rate": 3.7153983791179876e-05,
      "loss": 0.5858,
      "step": 20980
    },
    {
      "epoch": 6.62488159141143,
      "grad_norm": 4.590760231018066,
      "learning_rate": 3.715310669052381e-05,
      "loss": 0.4561,
      "step": 20981
    },
    {
      "epoch": 6.625197347647616,
      "grad_norm": 5.0537428855896,
      "learning_rate": 3.715222958986774e-05,
      "loss": 0.5699,
      "step": 20982
    },
    {
      "epoch": 6.625513103883802,
      "grad_norm": 5.387243747711182,
      "learning_rate": 3.7151352489211665e-05,
      "loss": 0.5594,
      "step": 20983
    },
    {
      "epoch": 6.625828860119987,
      "grad_norm": 5.2621965408325195,
      "learning_rate": 3.7150475388555596e-05,
      "loss": 0.4712,
      "step": 20984
    },
    {
      "epoch": 6.626144616356173,
      "grad_norm": 5.272621154785156,
      "learning_rate": 3.714959828789952e-05,
      "loss": 0.5935,
      "step": 20985
    },
    {
      "epoch": 6.6264603725923585,
      "grad_norm": 5.1778059005737305,
      "learning_rate": 3.714872118724345e-05,
      "loss": 0.5064,
      "step": 20986
    },
    {
      "epoch": 6.626776128828545,
      "grad_norm": 5.284881591796875,
      "learning_rate": 3.714784408658738e-05,
      "loss": 0.5089,
      "step": 20987
    },
    {
      "epoch": 6.62709188506473,
      "grad_norm": 5.080144882202148,
      "learning_rate": 3.714696698593131e-05,
      "loss": 0.6779,
      "step": 20988
    },
    {
      "epoch": 6.627407641300915,
      "grad_norm": 4.249024391174316,
      "learning_rate": 3.7146089885275234e-05,
      "loss": 0.5699,
      "step": 20989
    },
    {
      "epoch": 6.6277233975371015,
      "grad_norm": 4.3223090171813965,
      "learning_rate": 3.7145212784619166e-05,
      "loss": 0.4963,
      "step": 20990
    },
    {
      "epoch": 6.628039153773287,
      "grad_norm": 4.672606468200684,
      "learning_rate": 3.714433568396309e-05,
      "loss": 0.6405,
      "step": 20991
    },
    {
      "epoch": 6.628354910009473,
      "grad_norm": 4.921289920806885,
      "learning_rate": 3.714345858330702e-05,
      "loss": 0.5946,
      "step": 20992
    },
    {
      "epoch": 6.628670666245658,
      "grad_norm": 4.547745704650879,
      "learning_rate": 3.714258148265095e-05,
      "loss": 0.6256,
      "step": 20993
    },
    {
      "epoch": 6.628986422481844,
      "grad_norm": 4.28743314743042,
      "learning_rate": 3.714170438199488e-05,
      "loss": 0.4598,
      "step": 20994
    },
    {
      "epoch": 6.62930217871803,
      "grad_norm": 4.574857234954834,
      "learning_rate": 3.714082728133881e-05,
      "loss": 0.5142,
      "step": 20995
    },
    {
      "epoch": 6.629617934954215,
      "grad_norm": 4.856332302093506,
      "learning_rate": 3.7139950180682736e-05,
      "loss": 0.6567,
      "step": 20996
    },
    {
      "epoch": 6.629933691190401,
      "grad_norm": 4.706521987915039,
      "learning_rate": 3.713907308002667e-05,
      "loss": 0.5671,
      "step": 20997
    },
    {
      "epoch": 6.6302494474265865,
      "grad_norm": 4.536319732666016,
      "learning_rate": 3.713819597937059e-05,
      "loss": 0.6081,
      "step": 20998
    },
    {
      "epoch": 6.630565203662773,
      "grad_norm": 5.973392486572266,
      "learning_rate": 3.7137318878714524e-05,
      "loss": 0.6441,
      "step": 20999
    },
    {
      "epoch": 6.630880959898958,
      "grad_norm": 7.101317405700684,
      "learning_rate": 3.713644177805845e-05,
      "loss": 0.5672,
      "step": 21000
    },
    {
      "epoch": 6.630880959898958,
      "eval_loss": 0.8699281215667725,
      "eval_runtime": 4.5281,
      "eval_samples_per_second": 549.896,
      "eval_steps_per_second": 4.638,
      "step": 21000
    },
    {
      "epoch": 6.631196716135143,
      "grad_norm": 4.641695499420166,
      "learning_rate": 3.713556467740238e-05,
      "loss": 0.5296,
      "step": 21001
    },
    {
      "epoch": 6.6315124723713295,
      "grad_norm": 5.164689540863037,
      "learning_rate": 3.7134687576746306e-05,
      "loss": 0.61,
      "step": 21002
    },
    {
      "epoch": 6.631828228607515,
      "grad_norm": 4.6649394035339355,
      "learning_rate": 3.713381047609024e-05,
      "loss": 0.5281,
      "step": 21003
    },
    {
      "epoch": 6.632143984843701,
      "grad_norm": 6.03128719329834,
      "learning_rate": 3.713293337543416e-05,
      "loss": 0.6089,
      "step": 21004
    },
    {
      "epoch": 6.632459741079886,
      "grad_norm": 4.551717758178711,
      "learning_rate": 3.7132056274778094e-05,
      "loss": 0.598,
      "step": 21005
    },
    {
      "epoch": 6.6327754973160715,
      "grad_norm": 5.499481201171875,
      "learning_rate": 3.713117917412202e-05,
      "loss": 0.5804,
      "step": 21006
    },
    {
      "epoch": 6.633091253552258,
      "grad_norm": 4.0929059982299805,
      "learning_rate": 3.713030207346595e-05,
      "loss": 0.56,
      "step": 21007
    },
    {
      "epoch": 6.633407009788443,
      "grad_norm": 5.943384170532227,
      "learning_rate": 3.712942497280988e-05,
      "loss": 0.63,
      "step": 21008
    },
    {
      "epoch": 6.633722766024629,
      "grad_norm": 5.096606254577637,
      "learning_rate": 3.7128547872153814e-05,
      "loss": 0.5879,
      "step": 21009
    },
    {
      "epoch": 6.6340385222608145,
      "grad_norm": 4.379014492034912,
      "learning_rate": 3.712767077149774e-05,
      "loss": 0.563,
      "step": 21010
    },
    {
      "epoch": 6.634354278497001,
      "grad_norm": 3.800938367843628,
      "learning_rate": 3.712679367084167e-05,
      "loss": 0.4773,
      "step": 21011
    },
    {
      "epoch": 6.634670034733186,
      "grad_norm": 5.125688076019287,
      "learning_rate": 3.7125916570185595e-05,
      "loss": 0.5474,
      "step": 21012
    },
    {
      "epoch": 6.634985790969372,
      "grad_norm": 5.896397590637207,
      "learning_rate": 3.712503946952953e-05,
      "loss": 0.6694,
      "step": 21013
    },
    {
      "epoch": 6.635301547205557,
      "grad_norm": 3.6182799339294434,
      "learning_rate": 3.712416236887345e-05,
      "loss": 0.5743,
      "step": 21014
    },
    {
      "epoch": 6.635617303441743,
      "grad_norm": 4.317422866821289,
      "learning_rate": 3.7123285268217384e-05,
      "loss": 0.5669,
      "step": 21015
    },
    {
      "epoch": 6.635933059677929,
      "grad_norm": 5.767441749572754,
      "learning_rate": 3.7122408167561315e-05,
      "loss": 0.6811,
      "step": 21016
    },
    {
      "epoch": 6.636248815914114,
      "grad_norm": 4.942648410797119,
      "learning_rate": 3.712153106690524e-05,
      "loss": 0.6395,
      "step": 21017
    },
    {
      "epoch": 6.6365645721503,
      "grad_norm": 5.36586856842041,
      "learning_rate": 3.712065396624917e-05,
      "loss": 0.6367,
      "step": 21018
    },
    {
      "epoch": 6.636880328386486,
      "grad_norm": 4.368879795074463,
      "learning_rate": 3.71197768655931e-05,
      "loss": 0.6368,
      "step": 21019
    },
    {
      "epoch": 6.637196084622671,
      "grad_norm": 3.606440305709839,
      "learning_rate": 3.711889976493703e-05,
      "loss": 0.4266,
      "step": 21020
    },
    {
      "epoch": 6.637511840858857,
      "grad_norm": 5.878358364105225,
      "learning_rate": 3.7118022664280953e-05,
      "loss": 0.5113,
      "step": 21021
    },
    {
      "epoch": 6.637827597095042,
      "grad_norm": 5.196922302246094,
      "learning_rate": 3.7117145563624885e-05,
      "loss": 0.6377,
      "step": 21022
    },
    {
      "epoch": 6.638143353331229,
      "grad_norm": 4.507862567901611,
      "learning_rate": 3.711626846296881e-05,
      "loss": 0.5916,
      "step": 21023
    },
    {
      "epoch": 6.638459109567414,
      "grad_norm": 4.79278564453125,
      "learning_rate": 3.711539136231274e-05,
      "loss": 0.5377,
      "step": 21024
    },
    {
      "epoch": 6.638774865803599,
      "grad_norm": 5.360424518585205,
      "learning_rate": 3.711451426165667e-05,
      "loss": 0.591,
      "step": 21025
    },
    {
      "epoch": 6.639090622039785,
      "grad_norm": 4.435812473297119,
      "learning_rate": 3.71136371610006e-05,
      "loss": 0.4383,
      "step": 21026
    },
    {
      "epoch": 6.639406378275971,
      "grad_norm": 3.23567533493042,
      "learning_rate": 3.711276006034452e-05,
      "loss": 0.4481,
      "step": 21027
    },
    {
      "epoch": 6.639722134512157,
      "grad_norm": 4.950713634490967,
      "learning_rate": 3.7111882959688455e-05,
      "loss": 0.5508,
      "step": 21028
    },
    {
      "epoch": 6.640037890748342,
      "grad_norm": 5.181661128997803,
      "learning_rate": 3.711100585903238e-05,
      "loss": 0.5128,
      "step": 21029
    },
    {
      "epoch": 6.6403536469845275,
      "grad_norm": 4.835306644439697,
      "learning_rate": 3.711012875837631e-05,
      "loss": 0.5355,
      "step": 21030
    },
    {
      "epoch": 6.640669403220714,
      "grad_norm": 6.451262474060059,
      "learning_rate": 3.710925165772024e-05,
      "loss": 0.5811,
      "step": 21031
    },
    {
      "epoch": 6.640985159456899,
      "grad_norm": 4.87533712387085,
      "learning_rate": 3.710837455706417e-05,
      "loss": 0.5139,
      "step": 21032
    },
    {
      "epoch": 6.641300915693085,
      "grad_norm": 4.911808967590332,
      "learning_rate": 3.71074974564081e-05,
      "loss": 0.4376,
      "step": 21033
    },
    {
      "epoch": 6.64161667192927,
      "grad_norm": 4.8051533699035645,
      "learning_rate": 3.7106620355752025e-05,
      "loss": 0.5792,
      "step": 21034
    },
    {
      "epoch": 6.641932428165457,
      "grad_norm": 5.360031604766846,
      "learning_rate": 3.7105743255095956e-05,
      "loss": 0.5686,
      "step": 21035
    },
    {
      "epoch": 6.642248184401642,
      "grad_norm": 4.721413612365723,
      "learning_rate": 3.710486615443989e-05,
      "loss": 0.6249,
      "step": 21036
    },
    {
      "epoch": 6.642563940637828,
      "grad_norm": 5.3771891593933105,
      "learning_rate": 3.710398905378382e-05,
      "loss": 0.5165,
      "step": 21037
    },
    {
      "epoch": 6.642879696874013,
      "grad_norm": 4.331657886505127,
      "learning_rate": 3.7103111953127745e-05,
      "loss": 0.4393,
      "step": 21038
    },
    {
      "epoch": 6.643195453110199,
      "grad_norm": 4.551044940948486,
      "learning_rate": 3.7102234852471676e-05,
      "loss": 0.6213,
      "step": 21039
    },
    {
      "epoch": 6.643511209346385,
      "grad_norm": 6.349948883056641,
      "learning_rate": 3.71013577518156e-05,
      "loss": 0.6073,
      "step": 21040
    },
    {
      "epoch": 6.64382696558257,
      "grad_norm": 4.936140060424805,
      "learning_rate": 3.710048065115953e-05,
      "loss": 0.5539,
      "step": 21041
    },
    {
      "epoch": 6.644142721818756,
      "grad_norm": 5.726021766662598,
      "learning_rate": 3.709960355050346e-05,
      "loss": 0.6272,
      "step": 21042
    },
    {
      "epoch": 6.644458478054942,
      "grad_norm": 4.939266204833984,
      "learning_rate": 3.709872644984739e-05,
      "loss": 0.6642,
      "step": 21043
    },
    {
      "epoch": 6.644774234291127,
      "grad_norm": 6.0022358894348145,
      "learning_rate": 3.7097849349191314e-05,
      "loss": 0.6121,
      "step": 21044
    },
    {
      "epoch": 6.645089990527313,
      "grad_norm": 4.677540302276611,
      "learning_rate": 3.7096972248535246e-05,
      "loss": 0.5197,
      "step": 21045
    },
    {
      "epoch": 6.645405746763498,
      "grad_norm": 5.14852237701416,
      "learning_rate": 3.709609514787917e-05,
      "loss": 0.5322,
      "step": 21046
    },
    {
      "epoch": 6.645721502999685,
      "grad_norm": 4.7118330001831055,
      "learning_rate": 3.70952180472231e-05,
      "loss": 0.5712,
      "step": 21047
    },
    {
      "epoch": 6.64603725923587,
      "grad_norm": 4.988056182861328,
      "learning_rate": 3.709434094656703e-05,
      "loss": 0.556,
      "step": 21048
    },
    {
      "epoch": 6.646353015472055,
      "grad_norm": 5.154384613037109,
      "learning_rate": 3.709346384591096e-05,
      "loss": 0.6176,
      "step": 21049
    },
    {
      "epoch": 6.646668771708241,
      "grad_norm": 4.574700355529785,
      "learning_rate": 3.7092586745254884e-05,
      "loss": 0.5697,
      "step": 21050
    },
    {
      "epoch": 6.646984527944427,
      "grad_norm": 4.693670272827148,
      "learning_rate": 3.7091709644598816e-05,
      "loss": 0.6193,
      "step": 21051
    },
    {
      "epoch": 6.647300284180613,
      "grad_norm": 5.094329833984375,
      "learning_rate": 3.709083254394274e-05,
      "loss": 0.6382,
      "step": 21052
    },
    {
      "epoch": 6.647616040416798,
      "grad_norm": 5.8476033210754395,
      "learning_rate": 3.708995544328667e-05,
      "loss": 0.6483,
      "step": 21053
    },
    {
      "epoch": 6.647931796652984,
      "grad_norm": 4.614645004272461,
      "learning_rate": 3.7089078342630604e-05,
      "loss": 0.6058,
      "step": 21054
    },
    {
      "epoch": 6.64824755288917,
      "grad_norm": 4.075576305389404,
      "learning_rate": 3.708820124197453e-05,
      "loss": 0.5027,
      "step": 21055
    },
    {
      "epoch": 6.648563309125355,
      "grad_norm": 5.09025239944458,
      "learning_rate": 3.708732414131846e-05,
      "loss": 0.4773,
      "step": 21056
    },
    {
      "epoch": 6.648879065361541,
      "grad_norm": 4.360922813415527,
      "learning_rate": 3.7086447040662386e-05,
      "loss": 0.525,
      "step": 21057
    },
    {
      "epoch": 6.649194821597726,
      "grad_norm": 4.3209052085876465,
      "learning_rate": 3.708556994000632e-05,
      "loss": 0.5712,
      "step": 21058
    },
    {
      "epoch": 6.6495105778339125,
      "grad_norm": 4.7345709800720215,
      "learning_rate": 3.708469283935024e-05,
      "loss": 0.5655,
      "step": 21059
    },
    {
      "epoch": 6.649826334070098,
      "grad_norm": 5.467195510864258,
      "learning_rate": 3.7083815738694174e-05,
      "loss": 0.5587,
      "step": 21060
    },
    {
      "epoch": 6.650142090306284,
      "grad_norm": 4.482797145843506,
      "learning_rate": 3.70829386380381e-05,
      "loss": 0.4777,
      "step": 21061
    },
    {
      "epoch": 6.650457846542469,
      "grad_norm": 4.812686443328857,
      "learning_rate": 3.708206153738203e-05,
      "loss": 0.5391,
      "step": 21062
    },
    {
      "epoch": 6.650773602778655,
      "grad_norm": 5.8448028564453125,
      "learning_rate": 3.708118443672596e-05,
      "loss": 0.5117,
      "step": 21063
    },
    {
      "epoch": 6.651089359014841,
      "grad_norm": 4.735223770141602,
      "learning_rate": 3.7080307336069894e-05,
      "loss": 0.5644,
      "step": 21064
    },
    {
      "epoch": 6.651405115251026,
      "grad_norm": 3.99214768409729,
      "learning_rate": 3.707943023541382e-05,
      "loss": 0.5849,
      "step": 21065
    },
    {
      "epoch": 6.651720871487212,
      "grad_norm": 4.540950298309326,
      "learning_rate": 3.707855313475775e-05,
      "loss": 0.5735,
      "step": 21066
    },
    {
      "epoch": 6.6520366277233975,
      "grad_norm": 4.944608688354492,
      "learning_rate": 3.7077676034101675e-05,
      "loss": 0.6241,
      "step": 21067
    },
    {
      "epoch": 6.652352383959583,
      "grad_norm": 5.164187431335449,
      "learning_rate": 3.707679893344561e-05,
      "loss": 0.5697,
      "step": 21068
    },
    {
      "epoch": 6.652668140195769,
      "grad_norm": 5.07932186126709,
      "learning_rate": 3.707592183278953e-05,
      "loss": 0.598,
      "step": 21069
    },
    {
      "epoch": 6.652983896431954,
      "grad_norm": 4.605752468109131,
      "learning_rate": 3.7075044732133464e-05,
      "loss": 0.5388,
      "step": 21070
    },
    {
      "epoch": 6.6532996526681405,
      "grad_norm": 5.206037521362305,
      "learning_rate": 3.707416763147739e-05,
      "loss": 0.5962,
      "step": 21071
    },
    {
      "epoch": 6.653615408904326,
      "grad_norm": 4.2231879234313965,
      "learning_rate": 3.707329053082132e-05,
      "loss": 0.5558,
      "step": 21072
    },
    {
      "epoch": 6.653931165140511,
      "grad_norm": 4.648756980895996,
      "learning_rate": 3.7072413430165245e-05,
      "loss": 0.5812,
      "step": 21073
    },
    {
      "epoch": 6.654246921376697,
      "grad_norm": 4.729954242706299,
      "learning_rate": 3.707153632950918e-05,
      "loss": 0.5884,
      "step": 21074
    },
    {
      "epoch": 6.654562677612883,
      "grad_norm": 6.3542304039001465,
      "learning_rate": 3.707065922885311e-05,
      "loss": 0.5838,
      "step": 21075
    },
    {
      "epoch": 6.654878433849069,
      "grad_norm": 5.142889499664307,
      "learning_rate": 3.706978212819703e-05,
      "loss": 0.5535,
      "step": 21076
    },
    {
      "epoch": 6.655194190085254,
      "grad_norm": 6.3294878005981445,
      "learning_rate": 3.7068905027540965e-05,
      "loss": 0.5488,
      "step": 21077
    },
    {
      "epoch": 6.65550994632144,
      "grad_norm": 5.780413627624512,
      "learning_rate": 3.706802792688489e-05,
      "loss": 0.5901,
      "step": 21078
    },
    {
      "epoch": 6.6558257025576255,
      "grad_norm": 5.155593395233154,
      "learning_rate": 3.706715082622882e-05,
      "loss": 0.557,
      "step": 21079
    },
    {
      "epoch": 6.656141458793811,
      "grad_norm": 5.157116413116455,
      "learning_rate": 3.7066273725572747e-05,
      "loss": 0.5946,
      "step": 21080
    },
    {
      "epoch": 6.656457215029997,
      "grad_norm": 5.708528995513916,
      "learning_rate": 3.706539662491668e-05,
      "loss": 0.5128,
      "step": 21081
    },
    {
      "epoch": 6.656772971266182,
      "grad_norm": 5.024064540863037,
      "learning_rate": 3.70645195242606e-05,
      "loss": 0.5759,
      "step": 21082
    },
    {
      "epoch": 6.6570887275023685,
      "grad_norm": 4.448167324066162,
      "learning_rate": 3.7063642423604535e-05,
      "loss": 0.5879,
      "step": 21083
    },
    {
      "epoch": 6.657404483738554,
      "grad_norm": 4.498493194580078,
      "learning_rate": 3.706276532294846e-05,
      "loss": 0.4638,
      "step": 21084
    },
    {
      "epoch": 6.65772023997474,
      "grad_norm": 4.118900299072266,
      "learning_rate": 3.706188822229239e-05,
      "loss": 0.5629,
      "step": 21085
    },
    {
      "epoch": 6.658035996210925,
      "grad_norm": 4.835747718811035,
      "learning_rate": 3.7061011121636316e-05,
      "loss": 0.5131,
      "step": 21086
    },
    {
      "epoch": 6.6583517524471105,
      "grad_norm": 4.041489601135254,
      "learning_rate": 3.706013402098025e-05,
      "loss": 0.5786,
      "step": 21087
    },
    {
      "epoch": 6.658667508683297,
      "grad_norm": 4.138550758361816,
      "learning_rate": 3.705925692032417e-05,
      "loss": 0.4657,
      "step": 21088
    },
    {
      "epoch": 6.658983264919482,
      "grad_norm": 4.031269550323486,
      "learning_rate": 3.7058379819668105e-05,
      "loss": 0.5129,
      "step": 21089
    },
    {
      "epoch": 6.659299021155668,
      "grad_norm": 4.962172985076904,
      "learning_rate": 3.7057502719012036e-05,
      "loss": 0.5704,
      "step": 21090
    },
    {
      "epoch": 6.6596147773918535,
      "grad_norm": 4.912170886993408,
      "learning_rate": 3.705662561835597e-05,
      "loss": 0.5781,
      "step": 21091
    },
    {
      "epoch": 6.659930533628039,
      "grad_norm": 6.517424583435059,
      "learning_rate": 3.705574851769989e-05,
      "loss": 0.4806,
      "step": 21092
    },
    {
      "epoch": 6.660246289864225,
      "grad_norm": 5.706692695617676,
      "learning_rate": 3.7054871417043825e-05,
      "loss": 0.5627,
      "step": 21093
    },
    {
      "epoch": 6.66056204610041,
      "grad_norm": 4.951785087585449,
      "learning_rate": 3.705399431638775e-05,
      "loss": 0.629,
      "step": 21094
    },
    {
      "epoch": 6.660877802336596,
      "grad_norm": 5.3080267906188965,
      "learning_rate": 3.705311721573168e-05,
      "loss": 0.5267,
      "step": 21095
    },
    {
      "epoch": 6.661193558572782,
      "grad_norm": 4.583768367767334,
      "learning_rate": 3.7052240115075606e-05,
      "loss": 0.6255,
      "step": 21096
    },
    {
      "epoch": 6.661509314808967,
      "grad_norm": 4.379088401794434,
      "learning_rate": 3.705136301441954e-05,
      "loss": 0.6252,
      "step": 21097
    },
    {
      "epoch": 6.661825071045153,
      "grad_norm": 5.486194610595703,
      "learning_rate": 3.705048591376347e-05,
      "loss": 0.5095,
      "step": 21098
    },
    {
      "epoch": 6.6621408272813385,
      "grad_norm": 5.592592716217041,
      "learning_rate": 3.7049608813107394e-05,
      "loss": 0.471,
      "step": 21099
    },
    {
      "epoch": 6.662456583517525,
      "grad_norm": 4.701905250549316,
      "learning_rate": 3.7048731712451326e-05,
      "loss": 0.5021,
      "step": 21100
    },
    {
      "epoch": 6.66277233975371,
      "grad_norm": 5.72802734375,
      "learning_rate": 3.704785461179525e-05,
      "loss": 0.5965,
      "step": 21101
    },
    {
      "epoch": 6.663088095989896,
      "grad_norm": 4.77668571472168,
      "learning_rate": 3.704697751113918e-05,
      "loss": 0.5038,
      "step": 21102
    },
    {
      "epoch": 6.6634038522260814,
      "grad_norm": 5.14470100402832,
      "learning_rate": 3.704610041048311e-05,
      "loss": 0.4636,
      "step": 21103
    },
    {
      "epoch": 6.663719608462267,
      "grad_norm": 4.931011199951172,
      "learning_rate": 3.704522330982704e-05,
      "loss": 0.5419,
      "step": 21104
    },
    {
      "epoch": 6.664035364698453,
      "grad_norm": 4.951234817504883,
      "learning_rate": 3.7044346209170964e-05,
      "loss": 0.5314,
      "step": 21105
    },
    {
      "epoch": 6.664351120934638,
      "grad_norm": 4.566330432891846,
      "learning_rate": 3.7043469108514896e-05,
      "loss": 0.5578,
      "step": 21106
    },
    {
      "epoch": 6.664666877170824,
      "grad_norm": 4.906651973724365,
      "learning_rate": 3.704259200785882e-05,
      "loss": 0.5855,
      "step": 21107
    },
    {
      "epoch": 6.66498263340701,
      "grad_norm": 6.128584384918213,
      "learning_rate": 3.704171490720275e-05,
      "loss": 0.6283,
      "step": 21108
    },
    {
      "epoch": 6.665298389643196,
      "grad_norm": 6.33219575881958,
      "learning_rate": 3.704083780654668e-05,
      "loss": 0.5717,
      "step": 21109
    },
    {
      "epoch": 6.665614145879381,
      "grad_norm": 6.492252826690674,
      "learning_rate": 3.703996070589061e-05,
      "loss": 0.6859,
      "step": 21110
    },
    {
      "epoch": 6.6659299021155665,
      "grad_norm": 6.570378303527832,
      "learning_rate": 3.7039083605234534e-05,
      "loss": 0.5932,
      "step": 21111
    },
    {
      "epoch": 6.666245658351753,
      "grad_norm": 6.394201278686523,
      "learning_rate": 3.7038206504578465e-05,
      "loss": 0.5712,
      "step": 21112
    },
    {
      "epoch": 6.666561414587938,
      "grad_norm": 5.446215629577637,
      "learning_rate": 3.70373294039224e-05,
      "loss": 0.6792,
      "step": 21113
    },
    {
      "epoch": 6.666877170824124,
      "grad_norm": 5.063717365264893,
      "learning_rate": 3.703645230326632e-05,
      "loss": 0.5879,
      "step": 21114
    },
    {
      "epoch": 6.667192927060309,
      "grad_norm": 5.587393760681152,
      "learning_rate": 3.7035575202610254e-05,
      "loss": 0.5246,
      "step": 21115
    },
    {
      "epoch": 6.667508683296495,
      "grad_norm": 5.248676300048828,
      "learning_rate": 3.703469810195418e-05,
      "loss": 0.5842,
      "step": 21116
    },
    {
      "epoch": 6.667824439532681,
      "grad_norm": 4.411584854125977,
      "learning_rate": 3.703382100129811e-05,
      "loss": 0.5429,
      "step": 21117
    },
    {
      "epoch": 6.668140195768866,
      "grad_norm": 5.736271858215332,
      "learning_rate": 3.703294390064204e-05,
      "loss": 0.6966,
      "step": 21118
    },
    {
      "epoch": 6.668455952005052,
      "grad_norm": 5.982497215270996,
      "learning_rate": 3.7032066799985974e-05,
      "loss": 0.6259,
      "step": 21119
    },
    {
      "epoch": 6.668771708241238,
      "grad_norm": 5.248878002166748,
      "learning_rate": 3.70311896993299e-05,
      "loss": 0.597,
      "step": 21120
    },
    {
      "epoch": 6.669087464477423,
      "grad_norm": 5.3070783615112305,
      "learning_rate": 3.703031259867383e-05,
      "loss": 0.6821,
      "step": 21121
    },
    {
      "epoch": 6.669403220713609,
      "grad_norm": 5.129964351654053,
      "learning_rate": 3.7029435498017755e-05,
      "loss": 0.6124,
      "step": 21122
    },
    {
      "epoch": 6.669718976949794,
      "grad_norm": 4.713027000427246,
      "learning_rate": 3.702855839736169e-05,
      "loss": 0.6413,
      "step": 21123
    },
    {
      "epoch": 6.670034733185981,
      "grad_norm": 4.440096855163574,
      "learning_rate": 3.702768129670561e-05,
      "loss": 0.5719,
      "step": 21124
    },
    {
      "epoch": 6.670350489422166,
      "grad_norm": 4.707428932189941,
      "learning_rate": 3.7026804196049543e-05,
      "loss": 0.6526,
      "step": 21125
    },
    {
      "epoch": 6.670666245658352,
      "grad_norm": 4.879922866821289,
      "learning_rate": 3.702592709539347e-05,
      "loss": 0.5622,
      "step": 21126
    },
    {
      "epoch": 6.670982001894537,
      "grad_norm": 5.147921085357666,
      "learning_rate": 3.70250499947374e-05,
      "loss": 0.6085,
      "step": 21127
    },
    {
      "epoch": 6.671297758130724,
      "grad_norm": 4.873839378356934,
      "learning_rate": 3.7024172894081325e-05,
      "loss": 0.4992,
      "step": 21128
    },
    {
      "epoch": 6.671613514366909,
      "grad_norm": 5.704240798950195,
      "learning_rate": 3.702329579342526e-05,
      "loss": 0.6857,
      "step": 21129
    },
    {
      "epoch": 6.671929270603094,
      "grad_norm": 5.012824058532715,
      "learning_rate": 3.702241869276918e-05,
      "loss": 0.632,
      "step": 21130
    },
    {
      "epoch": 6.67224502683928,
      "grad_norm": 4.174654006958008,
      "learning_rate": 3.702154159211311e-05,
      "loss": 0.6359,
      "step": 21131
    },
    {
      "epoch": 6.672560783075466,
      "grad_norm": 4.813339710235596,
      "learning_rate": 3.702066449145704e-05,
      "loss": 0.5464,
      "step": 21132
    },
    {
      "epoch": 6.672876539311652,
      "grad_norm": 5.087499618530273,
      "learning_rate": 3.701978739080097e-05,
      "loss": 0.586,
      "step": 21133
    },
    {
      "epoch": 6.673192295547837,
      "grad_norm": 4.790595531463623,
      "learning_rate": 3.7018910290144895e-05,
      "loss": 0.5493,
      "step": 21134
    },
    {
      "epoch": 6.673508051784022,
      "grad_norm": 4.3032941818237305,
      "learning_rate": 3.7018033189488826e-05,
      "loss": 0.6142,
      "step": 21135
    },
    {
      "epoch": 6.673823808020209,
      "grad_norm": 4.239046096801758,
      "learning_rate": 3.701715608883276e-05,
      "loss": 0.6678,
      "step": 21136
    },
    {
      "epoch": 6.674139564256394,
      "grad_norm": 4.613865852355957,
      "learning_rate": 3.701627898817668e-05,
      "loss": 0.6902,
      "step": 21137
    },
    {
      "epoch": 6.67445532049258,
      "grad_norm": 4.748202323913574,
      "learning_rate": 3.7015401887520615e-05,
      "loss": 0.6545,
      "step": 21138
    },
    {
      "epoch": 6.674771076728765,
      "grad_norm": 5.021595478057861,
      "learning_rate": 3.701452478686454e-05,
      "loss": 0.5076,
      "step": 21139
    },
    {
      "epoch": 6.675086832964951,
      "grad_norm": 4.458683013916016,
      "learning_rate": 3.701364768620847e-05,
      "loss": 0.6616,
      "step": 21140
    },
    {
      "epoch": 6.675402589201137,
      "grad_norm": 4.211365699768066,
      "learning_rate": 3.7012770585552396e-05,
      "loss": 0.5354,
      "step": 21141
    },
    {
      "epoch": 6.675718345437322,
      "grad_norm": 4.531017303466797,
      "learning_rate": 3.701189348489633e-05,
      "loss": 0.5259,
      "step": 21142
    },
    {
      "epoch": 6.676034101673508,
      "grad_norm": 4.308961868286133,
      "learning_rate": 3.701101638424025e-05,
      "loss": 0.5479,
      "step": 21143
    },
    {
      "epoch": 6.676349857909694,
      "grad_norm": 5.714181423187256,
      "learning_rate": 3.7010139283584184e-05,
      "loss": 0.5939,
      "step": 21144
    },
    {
      "epoch": 6.676665614145879,
      "grad_norm": 5.123617649078369,
      "learning_rate": 3.7009262182928116e-05,
      "loss": 0.6343,
      "step": 21145
    },
    {
      "epoch": 6.676981370382065,
      "grad_norm": 5.461437225341797,
      "learning_rate": 3.700838508227205e-05,
      "loss": 0.6301,
      "step": 21146
    },
    {
      "epoch": 6.67729712661825,
      "grad_norm": 5.173875331878662,
      "learning_rate": 3.700750798161597e-05,
      "loss": 0.7234,
      "step": 21147
    },
    {
      "epoch": 6.6776128828544365,
      "grad_norm": 4.124063014984131,
      "learning_rate": 3.7006630880959904e-05,
      "loss": 0.4582,
      "step": 21148
    },
    {
      "epoch": 6.677928639090622,
      "grad_norm": 4.52065896987915,
      "learning_rate": 3.700575378030383e-05,
      "loss": 0.5609,
      "step": 21149
    },
    {
      "epoch": 6.678244395326808,
      "grad_norm": 5.246395111083984,
      "learning_rate": 3.700487667964776e-05,
      "loss": 0.5407,
      "step": 21150
    },
    {
      "epoch": 6.678560151562993,
      "grad_norm": 5.170871257781982,
      "learning_rate": 3.7003999578991686e-05,
      "loss": 0.5412,
      "step": 21151
    },
    {
      "epoch": 6.6788759077991795,
      "grad_norm": 3.6363725662231445,
      "learning_rate": 3.700312247833562e-05,
      "loss": 0.505,
      "step": 21152
    },
    {
      "epoch": 6.679191664035365,
      "grad_norm": 4.856725692749023,
      "learning_rate": 3.700224537767954e-05,
      "loss": 0.5796,
      "step": 21153
    },
    {
      "epoch": 6.67950742027155,
      "grad_norm": 6.547928333282471,
      "learning_rate": 3.7001368277023474e-05,
      "loss": 0.62,
      "step": 21154
    },
    {
      "epoch": 6.679823176507736,
      "grad_norm": 4.990301132202148,
      "learning_rate": 3.70004911763674e-05,
      "loss": 0.6103,
      "step": 21155
    },
    {
      "epoch": 6.680138932743922,
      "grad_norm": 4.930610179901123,
      "learning_rate": 3.699961407571133e-05,
      "loss": 0.4322,
      "step": 21156
    },
    {
      "epoch": 6.680454688980108,
      "grad_norm": 4.791330814361572,
      "learning_rate": 3.699873697505526e-05,
      "loss": 0.5645,
      "step": 21157
    },
    {
      "epoch": 6.680770445216293,
      "grad_norm": 4.825796127319336,
      "learning_rate": 3.699785987439919e-05,
      "loss": 0.6295,
      "step": 21158
    },
    {
      "epoch": 6.681086201452478,
      "grad_norm": 5.8522114753723145,
      "learning_rate": 3.699698277374312e-05,
      "loss": 0.5839,
      "step": 21159
    },
    {
      "epoch": 6.6814019576886645,
      "grad_norm": 4.5463433265686035,
      "learning_rate": 3.6996105673087044e-05,
      "loss": 0.5037,
      "step": 21160
    },
    {
      "epoch": 6.68171771392485,
      "grad_norm": 5.92018461227417,
      "learning_rate": 3.6995228572430976e-05,
      "loss": 0.5941,
      "step": 21161
    },
    {
      "epoch": 6.682033470161036,
      "grad_norm": 5.062238693237305,
      "learning_rate": 3.69943514717749e-05,
      "loss": 0.5848,
      "step": 21162
    },
    {
      "epoch": 6.682349226397221,
      "grad_norm": 4.365847110748291,
      "learning_rate": 3.699347437111883e-05,
      "loss": 0.5262,
      "step": 21163
    },
    {
      "epoch": 6.682664982633407,
      "grad_norm": 7.112579822540283,
      "learning_rate": 3.699259727046276e-05,
      "loss": 0.5935,
      "step": 21164
    },
    {
      "epoch": 6.682980738869593,
      "grad_norm": 4.364557266235352,
      "learning_rate": 3.699172016980669e-05,
      "loss": 0.5251,
      "step": 21165
    },
    {
      "epoch": 6.683296495105778,
      "grad_norm": 5.505921840667725,
      "learning_rate": 3.6990843069150614e-05,
      "loss": 0.672,
      "step": 21166
    },
    {
      "epoch": 6.683612251341964,
      "grad_norm": 4.314089298248291,
      "learning_rate": 3.6989965968494545e-05,
      "loss": 0.4673,
      "step": 21167
    },
    {
      "epoch": 6.6839280075781495,
      "grad_norm": 5.014153957366943,
      "learning_rate": 3.698908886783847e-05,
      "loss": 0.5185,
      "step": 21168
    },
    {
      "epoch": 6.684243763814336,
      "grad_norm": 4.769038200378418,
      "learning_rate": 3.69882117671824e-05,
      "loss": 0.6174,
      "step": 21169
    },
    {
      "epoch": 6.684559520050521,
      "grad_norm": 5.3970232009887695,
      "learning_rate": 3.698733466652633e-05,
      "loss": 0.6871,
      "step": 21170
    },
    {
      "epoch": 6.684875276286706,
      "grad_norm": 5.466236591339111,
      "learning_rate": 3.698645756587026e-05,
      "loss": 0.5632,
      "step": 21171
    },
    {
      "epoch": 6.6851910325228925,
      "grad_norm": 6.639613151550293,
      "learning_rate": 3.698558046521419e-05,
      "loss": 0.5427,
      "step": 21172
    },
    {
      "epoch": 6.685506788759078,
      "grad_norm": 5.095594882965088,
      "learning_rate": 3.698470336455812e-05,
      "loss": 0.475,
      "step": 21173
    },
    {
      "epoch": 6.685822544995264,
      "grad_norm": 5.843761444091797,
      "learning_rate": 3.698382626390205e-05,
      "loss": 0.654,
      "step": 21174
    },
    {
      "epoch": 6.686138301231449,
      "grad_norm": 4.7594194412231445,
      "learning_rate": 3.698294916324598e-05,
      "loss": 0.5943,
      "step": 21175
    },
    {
      "epoch": 6.686454057467635,
      "grad_norm": 5.128991603851318,
      "learning_rate": 3.6982072062589903e-05,
      "loss": 0.6707,
      "step": 21176
    },
    {
      "epoch": 6.686769813703821,
      "grad_norm": 4.316229820251465,
      "learning_rate": 3.6981194961933835e-05,
      "loss": 0.5504,
      "step": 21177
    },
    {
      "epoch": 6.687085569940006,
      "grad_norm": 5.767012596130371,
      "learning_rate": 3.698031786127777e-05,
      "loss": 0.6226,
      "step": 21178
    },
    {
      "epoch": 6.687401326176192,
      "grad_norm": 4.799744129180908,
      "learning_rate": 3.697944076062169e-05,
      "loss": 0.6483,
      "step": 21179
    },
    {
      "epoch": 6.6877170824123775,
      "grad_norm": 4.388599395751953,
      "learning_rate": 3.697856365996562e-05,
      "loss": 0.5382,
      "step": 21180
    },
    {
      "epoch": 6.688032838648564,
      "grad_norm": 4.8670878410339355,
      "learning_rate": 3.697768655930955e-05,
      "loss": 0.4328,
      "step": 21181
    },
    {
      "epoch": 6.688348594884749,
      "grad_norm": 3.9994513988494873,
      "learning_rate": 3.697680945865348e-05,
      "loss": 0.5156,
      "step": 21182
    },
    {
      "epoch": 6.688664351120934,
      "grad_norm": 5.234165668487549,
      "learning_rate": 3.6975932357997405e-05,
      "loss": 0.6078,
      "step": 21183
    },
    {
      "epoch": 6.6889801073571205,
      "grad_norm": 4.265045642852783,
      "learning_rate": 3.6975055257341337e-05,
      "loss": 0.4831,
      "step": 21184
    },
    {
      "epoch": 6.689295863593306,
      "grad_norm": 4.836201190948486,
      "learning_rate": 3.697417815668526e-05,
      "loss": 0.652,
      "step": 21185
    },
    {
      "epoch": 6.689611619829492,
      "grad_norm": 5.707010746002197,
      "learning_rate": 3.697330105602919e-05,
      "loss": 0.5945,
      "step": 21186
    },
    {
      "epoch": 6.689927376065677,
      "grad_norm": 4.65079927444458,
      "learning_rate": 3.697242395537312e-05,
      "loss": 0.4679,
      "step": 21187
    },
    {
      "epoch": 6.6902431323018625,
      "grad_norm": 6.088992595672607,
      "learning_rate": 3.697154685471705e-05,
      "loss": 0.6278,
      "step": 21188
    },
    {
      "epoch": 6.690558888538049,
      "grad_norm": 5.290589809417725,
      "learning_rate": 3.6970669754060975e-05,
      "loss": 0.6174,
      "step": 21189
    },
    {
      "epoch": 6.690874644774234,
      "grad_norm": 6.014688968658447,
      "learning_rate": 3.6969792653404906e-05,
      "loss": 0.6089,
      "step": 21190
    },
    {
      "epoch": 6.69119040101042,
      "grad_norm": 6.189300060272217,
      "learning_rate": 3.696891555274883e-05,
      "loss": 0.6092,
      "step": 21191
    },
    {
      "epoch": 6.6915061572466055,
      "grad_norm": 4.486336708068848,
      "learning_rate": 3.696803845209276e-05,
      "loss": 0.5455,
      "step": 21192
    },
    {
      "epoch": 6.691821913482792,
      "grad_norm": 5.563554763793945,
      "learning_rate": 3.696716135143669e-05,
      "loss": 0.5487,
      "step": 21193
    },
    {
      "epoch": 6.692137669718977,
      "grad_norm": 4.797843933105469,
      "learning_rate": 3.696628425078062e-05,
      "loss": 0.538,
      "step": 21194
    },
    {
      "epoch": 6.692453425955162,
      "grad_norm": 3.7647879123687744,
      "learning_rate": 3.696540715012455e-05,
      "loss": 0.4723,
      "step": 21195
    },
    {
      "epoch": 6.692769182191348,
      "grad_norm": 5.8488688468933105,
      "learning_rate": 3.6964530049468476e-05,
      "loss": 0.6076,
      "step": 21196
    },
    {
      "epoch": 6.693084938427534,
      "grad_norm": 5.721691131591797,
      "learning_rate": 3.696365294881241e-05,
      "loss": 0.5828,
      "step": 21197
    },
    {
      "epoch": 6.69340069466372,
      "grad_norm": 5.465809345245361,
      "learning_rate": 3.696277584815633e-05,
      "loss": 0.4815,
      "step": 21198
    },
    {
      "epoch": 6.693716450899905,
      "grad_norm": 5.299170970916748,
      "learning_rate": 3.6961898747500264e-05,
      "loss": 0.5073,
      "step": 21199
    },
    {
      "epoch": 6.694032207136091,
      "grad_norm": 6.107196807861328,
      "learning_rate": 3.6961021646844196e-05,
      "loss": 0.639,
      "step": 21200
    },
    {
      "epoch": 6.694032207136091,
      "eval_loss": 0.8946542143821716,
      "eval_runtime": 4.55,
      "eval_samples_per_second": 547.251,
      "eval_steps_per_second": 4.615,
      "step": 21200
    },
    {
      "epoch": 6.694347963372277,
      "grad_norm": 5.5162034034729,
      "learning_rate": 3.696014454618813e-05,
      "loss": 0.6297,
      "step": 21201
    },
    {
      "epoch": 6.694663719608462,
      "grad_norm": 5.56198263168335,
      "learning_rate": 3.695926744553205e-05,
      "loss": 0.5045,
      "step": 21202
    },
    {
      "epoch": 6.694979475844648,
      "grad_norm": 5.002174377441406,
      "learning_rate": 3.6958390344875984e-05,
      "loss": 0.6458,
      "step": 21203
    },
    {
      "epoch": 6.695295232080833,
      "grad_norm": 4.85468864440918,
      "learning_rate": 3.695751324421991e-05,
      "loss": 0.5446,
      "step": 21204
    },
    {
      "epoch": 6.69561098831702,
      "grad_norm": 4.119270324707031,
      "learning_rate": 3.695663614356384e-05,
      "loss": 0.6121,
      "step": 21205
    },
    {
      "epoch": 6.695926744553205,
      "grad_norm": 5.407104969024658,
      "learning_rate": 3.6955759042907766e-05,
      "loss": 0.6274,
      "step": 21206
    },
    {
      "epoch": 6.69624250078939,
      "grad_norm": 5.513650894165039,
      "learning_rate": 3.69548819422517e-05,
      "loss": 0.5893,
      "step": 21207
    },
    {
      "epoch": 6.696558257025576,
      "grad_norm": 5.791415214538574,
      "learning_rate": 3.695400484159562e-05,
      "loss": 0.5767,
      "step": 21208
    },
    {
      "epoch": 6.696874013261762,
      "grad_norm": 5.3916401863098145,
      "learning_rate": 3.6953127740939554e-05,
      "loss": 0.5884,
      "step": 21209
    },
    {
      "epoch": 6.697189769497948,
      "grad_norm": 9.942463874816895,
      "learning_rate": 3.695225064028348e-05,
      "loss": 0.5554,
      "step": 21210
    },
    {
      "epoch": 6.697505525734133,
      "grad_norm": 4.895512580871582,
      "learning_rate": 3.695137353962741e-05,
      "loss": 0.5423,
      "step": 21211
    },
    {
      "epoch": 6.6978212819703185,
      "grad_norm": 4.542381763458252,
      "learning_rate": 3.6950496438971336e-05,
      "loss": 0.5649,
      "step": 21212
    },
    {
      "epoch": 6.698137038206505,
      "grad_norm": 4.907773971557617,
      "learning_rate": 3.694961933831527e-05,
      "loss": 0.5331,
      "step": 21213
    },
    {
      "epoch": 6.69845279444269,
      "grad_norm": 5.372700214385986,
      "learning_rate": 3.694874223765919e-05,
      "loss": 0.5688,
      "step": 21214
    },
    {
      "epoch": 6.698768550678876,
      "grad_norm": 4.804360389709473,
      "learning_rate": 3.6947865137003124e-05,
      "loss": 0.4846,
      "step": 21215
    },
    {
      "epoch": 6.699084306915061,
      "grad_norm": 5.42339563369751,
      "learning_rate": 3.6946988036347056e-05,
      "loss": 0.6174,
      "step": 21216
    },
    {
      "epoch": 6.699400063151248,
      "grad_norm": 4.70443868637085,
      "learning_rate": 3.694611093569098e-05,
      "loss": 0.577,
      "step": 21217
    },
    {
      "epoch": 6.699715819387433,
      "grad_norm": 4.835148334503174,
      "learning_rate": 3.694523383503491e-05,
      "loss": 0.5732,
      "step": 21218
    },
    {
      "epoch": 6.700031575623619,
      "grad_norm": 4.622591018676758,
      "learning_rate": 3.694435673437884e-05,
      "loss": 0.6003,
      "step": 21219
    },
    {
      "epoch": 6.700347331859804,
      "grad_norm": 4.553218364715576,
      "learning_rate": 3.694347963372277e-05,
      "loss": 0.5301,
      "step": 21220
    },
    {
      "epoch": 6.70066308809599,
      "grad_norm": 4.711370468139648,
      "learning_rate": 3.6942602533066694e-05,
      "loss": 0.4895,
      "step": 21221
    },
    {
      "epoch": 6.700978844332176,
      "grad_norm": 5.3332929611206055,
      "learning_rate": 3.6941725432410625e-05,
      "loss": 0.5294,
      "step": 21222
    },
    {
      "epoch": 6.701294600568361,
      "grad_norm": 4.471475124359131,
      "learning_rate": 3.694084833175455e-05,
      "loss": 0.5165,
      "step": 21223
    },
    {
      "epoch": 6.701610356804547,
      "grad_norm": 4.483168125152588,
      "learning_rate": 3.693997123109848e-05,
      "loss": 0.5328,
      "step": 21224
    },
    {
      "epoch": 6.701926113040733,
      "grad_norm": 5.467512607574463,
      "learning_rate": 3.693909413044241e-05,
      "loss": 0.6288,
      "step": 21225
    },
    {
      "epoch": 6.702241869276918,
      "grad_norm": 4.8437180519104,
      "learning_rate": 3.6938217029786345e-05,
      "loss": 0.5868,
      "step": 21226
    },
    {
      "epoch": 6.702557625513104,
      "grad_norm": 5.214758396148682,
      "learning_rate": 3.693733992913027e-05,
      "loss": 0.5353,
      "step": 21227
    },
    {
      "epoch": 6.702873381749289,
      "grad_norm": 3.988536834716797,
      "learning_rate": 3.69364628284742e-05,
      "loss": 0.3738,
      "step": 21228
    },
    {
      "epoch": 6.7031891379854756,
      "grad_norm": 5.354594707489014,
      "learning_rate": 3.693558572781813e-05,
      "loss": 0.617,
      "step": 21229
    },
    {
      "epoch": 6.703504894221661,
      "grad_norm": 4.785187244415283,
      "learning_rate": 3.693470862716206e-05,
      "loss": 0.5128,
      "step": 21230
    },
    {
      "epoch": 6.703820650457846,
      "grad_norm": 5.4295735359191895,
      "learning_rate": 3.693383152650598e-05,
      "loss": 0.5958,
      "step": 21231
    },
    {
      "epoch": 6.704136406694032,
      "grad_norm": 5.6930975914001465,
      "learning_rate": 3.6932954425849915e-05,
      "loss": 0.5986,
      "step": 21232
    },
    {
      "epoch": 6.704452162930218,
      "grad_norm": 5.276753902435303,
      "learning_rate": 3.693207732519384e-05,
      "loss": 0.6026,
      "step": 21233
    },
    {
      "epoch": 6.704767919166404,
      "grad_norm": 4.692189693450928,
      "learning_rate": 3.693120022453777e-05,
      "loss": 0.4828,
      "step": 21234
    },
    {
      "epoch": 6.705083675402589,
      "grad_norm": 4.829920291900635,
      "learning_rate": 3.6930323123881696e-05,
      "loss": 0.555,
      "step": 21235
    },
    {
      "epoch": 6.705399431638774,
      "grad_norm": 4.314234256744385,
      "learning_rate": 3.692944602322563e-05,
      "loss": 0.4414,
      "step": 21236
    },
    {
      "epoch": 6.705715187874961,
      "grad_norm": 6.333461761474609,
      "learning_rate": 3.692856892256955e-05,
      "loss": 0.5987,
      "step": 21237
    },
    {
      "epoch": 6.706030944111146,
      "grad_norm": 4.891462802886963,
      "learning_rate": 3.6927691821913485e-05,
      "loss": 0.5354,
      "step": 21238
    },
    {
      "epoch": 6.706346700347332,
      "grad_norm": 4.933726787567139,
      "learning_rate": 3.6926814721257416e-05,
      "loss": 0.5378,
      "step": 21239
    },
    {
      "epoch": 6.706662456583517,
      "grad_norm": 5.241252899169922,
      "learning_rate": 3.692593762060134e-05,
      "loss": 0.5647,
      "step": 21240
    },
    {
      "epoch": 6.7069782128197035,
      "grad_norm": 5.196821212768555,
      "learning_rate": 3.692506051994527e-05,
      "loss": 0.5387,
      "step": 21241
    },
    {
      "epoch": 6.707293969055889,
      "grad_norm": 6.765556335449219,
      "learning_rate": 3.69241834192892e-05,
      "loss": 0.5605,
      "step": 21242
    },
    {
      "epoch": 6.707609725292075,
      "grad_norm": 5.173084735870361,
      "learning_rate": 3.692330631863313e-05,
      "loss": 0.6272,
      "step": 21243
    },
    {
      "epoch": 6.70792548152826,
      "grad_norm": 5.876845359802246,
      "learning_rate": 3.6922429217977055e-05,
      "loss": 0.5744,
      "step": 21244
    },
    {
      "epoch": 6.708241237764446,
      "grad_norm": 4.753918170928955,
      "learning_rate": 3.6921552117320986e-05,
      "loss": 0.5163,
      "step": 21245
    },
    {
      "epoch": 6.708556994000632,
      "grad_norm": 4.481765270233154,
      "learning_rate": 3.692067501666491e-05,
      "loss": 0.5929,
      "step": 21246
    },
    {
      "epoch": 6.708872750236817,
      "grad_norm": 5.239047527313232,
      "learning_rate": 3.691979791600884e-05,
      "loss": 0.5386,
      "step": 21247
    },
    {
      "epoch": 6.709188506473003,
      "grad_norm": 5.676278591156006,
      "learning_rate": 3.691892081535277e-05,
      "loss": 0.5626,
      "step": 21248
    },
    {
      "epoch": 6.7095042627091885,
      "grad_norm": 6.392148017883301,
      "learning_rate": 3.69180437146967e-05,
      "loss": 0.6102,
      "step": 21249
    },
    {
      "epoch": 6.709820018945374,
      "grad_norm": 6.1310272216796875,
      "learning_rate": 3.6917166614040624e-05,
      "loss": 0.6044,
      "step": 21250
    },
    {
      "epoch": 6.71013577518156,
      "grad_norm": 4.15265417098999,
      "learning_rate": 3.6916289513384556e-05,
      "loss": 0.4794,
      "step": 21251
    },
    {
      "epoch": 6.710451531417745,
      "grad_norm": 5.187044143676758,
      "learning_rate": 3.691541241272848e-05,
      "loss": 0.5779,
      "step": 21252
    },
    {
      "epoch": 6.7107672876539315,
      "grad_norm": 5.009997844696045,
      "learning_rate": 3.691453531207241e-05,
      "loss": 0.5504,
      "step": 21253
    },
    {
      "epoch": 6.711083043890117,
      "grad_norm": 6.088021755218506,
      "learning_rate": 3.6913658211416344e-05,
      "loss": 0.7233,
      "step": 21254
    },
    {
      "epoch": 6.711398800126302,
      "grad_norm": 4.43092679977417,
      "learning_rate": 3.6912781110760276e-05,
      "loss": 0.4912,
      "step": 21255
    },
    {
      "epoch": 6.711714556362488,
      "grad_norm": 6.2080864906311035,
      "learning_rate": 3.69119040101042e-05,
      "loss": 0.5076,
      "step": 21256
    },
    {
      "epoch": 6.7120303125986736,
      "grad_norm": 6.164026737213135,
      "learning_rate": 3.691102690944813e-05,
      "loss": 0.6454,
      "step": 21257
    },
    {
      "epoch": 6.71234606883486,
      "grad_norm": 5.5030412673950195,
      "learning_rate": 3.691014980879206e-05,
      "loss": 0.5124,
      "step": 21258
    },
    {
      "epoch": 6.712661825071045,
      "grad_norm": 5.461047172546387,
      "learning_rate": 3.690927270813599e-05,
      "loss": 0.6486,
      "step": 21259
    },
    {
      "epoch": 6.71297758130723,
      "grad_norm": 6.393360614776611,
      "learning_rate": 3.690839560747992e-05,
      "loss": 0.5439,
      "step": 21260
    },
    {
      "epoch": 6.7132933375434165,
      "grad_norm": 5.5263237953186035,
      "learning_rate": 3.6907518506823846e-05,
      "loss": 0.6242,
      "step": 21261
    },
    {
      "epoch": 6.713609093779602,
      "grad_norm": 5.152385234832764,
      "learning_rate": 3.690664140616778e-05,
      "loss": 0.6199,
      "step": 21262
    },
    {
      "epoch": 6.713924850015788,
      "grad_norm": 5.6032819747924805,
      "learning_rate": 3.69057643055117e-05,
      "loss": 0.621,
      "step": 21263
    },
    {
      "epoch": 6.714240606251973,
      "grad_norm": 5.388637065887451,
      "learning_rate": 3.6904887204855634e-05,
      "loss": 0.6246,
      "step": 21264
    },
    {
      "epoch": 6.7145563624881595,
      "grad_norm": 4.429262161254883,
      "learning_rate": 3.690401010419956e-05,
      "loss": 0.4494,
      "step": 21265
    },
    {
      "epoch": 6.714872118724345,
      "grad_norm": 5.123185634613037,
      "learning_rate": 3.690313300354349e-05,
      "loss": 0.5436,
      "step": 21266
    },
    {
      "epoch": 6.715187874960531,
      "grad_norm": 4.767491817474365,
      "learning_rate": 3.6902255902887415e-05,
      "loss": 0.4446,
      "step": 21267
    },
    {
      "epoch": 6.715503631196716,
      "grad_norm": 6.628791809082031,
      "learning_rate": 3.690137880223135e-05,
      "loss": 0.6513,
      "step": 21268
    },
    {
      "epoch": 6.7158193874329015,
      "grad_norm": 5.558740615844727,
      "learning_rate": 3.690050170157527e-05,
      "loss": 0.6202,
      "step": 21269
    },
    {
      "epoch": 6.716135143669088,
      "grad_norm": 5.726995944976807,
      "learning_rate": 3.6899624600919204e-05,
      "loss": 0.6574,
      "step": 21270
    },
    {
      "epoch": 6.716450899905273,
      "grad_norm": 5.887655258178711,
      "learning_rate": 3.689874750026313e-05,
      "loss": 0.6282,
      "step": 21271
    },
    {
      "epoch": 6.716766656141459,
      "grad_norm": 5.556155681610107,
      "learning_rate": 3.689787039960706e-05,
      "loss": 0.5751,
      "step": 21272
    },
    {
      "epoch": 6.7170824123776445,
      "grad_norm": 4.876643657684326,
      "learning_rate": 3.6896993298950985e-05,
      "loss": 0.5597,
      "step": 21273
    },
    {
      "epoch": 6.71739816861383,
      "grad_norm": 4.651214599609375,
      "learning_rate": 3.689611619829492e-05,
      "loss": 0.5061,
      "step": 21274
    },
    {
      "epoch": 6.717713924850016,
      "grad_norm": 4.388912200927734,
      "learning_rate": 3.689523909763884e-05,
      "loss": 0.5195,
      "step": 21275
    },
    {
      "epoch": 6.718029681086201,
      "grad_norm": 5.026604175567627,
      "learning_rate": 3.6894361996982773e-05,
      "loss": 0.5068,
      "step": 21276
    },
    {
      "epoch": 6.718345437322387,
      "grad_norm": 5.724648475646973,
      "learning_rate": 3.6893484896326705e-05,
      "loss": 0.7067,
      "step": 21277
    },
    {
      "epoch": 6.718661193558573,
      "grad_norm": 5.88593053817749,
      "learning_rate": 3.689260779567063e-05,
      "loss": 0.6809,
      "step": 21278
    },
    {
      "epoch": 6.718976949794758,
      "grad_norm": 4.342446804046631,
      "learning_rate": 3.689173069501456e-05,
      "loss": 0.5522,
      "step": 21279
    },
    {
      "epoch": 6.719292706030944,
      "grad_norm": 4.489997386932373,
      "learning_rate": 3.689085359435849e-05,
      "loss": 0.5399,
      "step": 21280
    },
    {
      "epoch": 6.7196084622671295,
      "grad_norm": 4.308990001678467,
      "learning_rate": 3.688997649370242e-05,
      "loss": 0.5443,
      "step": 21281
    },
    {
      "epoch": 6.719924218503316,
      "grad_norm": 3.9698750972747803,
      "learning_rate": 3.688909939304635e-05,
      "loss": 0.4854,
      "step": 21282
    },
    {
      "epoch": 6.720239974739501,
      "grad_norm": 5.609236240386963,
      "learning_rate": 3.688822229239028e-05,
      "loss": 0.5733,
      "step": 21283
    },
    {
      "epoch": 6.720555730975687,
      "grad_norm": 4.489475727081299,
      "learning_rate": 3.688734519173421e-05,
      "loss": 0.5508,
      "step": 21284
    },
    {
      "epoch": 6.720871487211872,
      "grad_norm": 4.055568218231201,
      "learning_rate": 3.688646809107814e-05,
      "loss": 0.6143,
      "step": 21285
    },
    {
      "epoch": 6.721187243448058,
      "grad_norm": 5.285406112670898,
      "learning_rate": 3.688559099042206e-05,
      "loss": 0.6198,
      "step": 21286
    },
    {
      "epoch": 6.721502999684244,
      "grad_norm": 4.497157573699951,
      "learning_rate": 3.6884713889765995e-05,
      "loss": 0.4978,
      "step": 21287
    },
    {
      "epoch": 6.721818755920429,
      "grad_norm": 8.950613021850586,
      "learning_rate": 3.688383678910992e-05,
      "loss": 0.5948,
      "step": 21288
    },
    {
      "epoch": 6.722134512156615,
      "grad_norm": 4.458363056182861,
      "learning_rate": 3.688295968845385e-05,
      "loss": 0.5839,
      "step": 21289
    },
    {
      "epoch": 6.722450268392801,
      "grad_norm": 4.431023597717285,
      "learning_rate": 3.6882082587797776e-05,
      "loss": 0.5299,
      "step": 21290
    },
    {
      "epoch": 6.722766024628987,
      "grad_norm": 4.476511001586914,
      "learning_rate": 3.688120548714171e-05,
      "loss": 0.5581,
      "step": 21291
    },
    {
      "epoch": 6.723081780865172,
      "grad_norm": 4.006006717681885,
      "learning_rate": 3.688032838648563e-05,
      "loss": 0.5351,
      "step": 21292
    },
    {
      "epoch": 6.7233975371013575,
      "grad_norm": 5.078372001647949,
      "learning_rate": 3.6879451285829565e-05,
      "loss": 0.6263,
      "step": 21293
    },
    {
      "epoch": 6.723713293337544,
      "grad_norm": 5.428628444671631,
      "learning_rate": 3.687857418517349e-05,
      "loss": 0.5439,
      "step": 21294
    },
    {
      "epoch": 6.724029049573729,
      "grad_norm": 5.082265853881836,
      "learning_rate": 3.687769708451742e-05,
      "loss": 0.5372,
      "step": 21295
    },
    {
      "epoch": 6.724344805809915,
      "grad_norm": 5.619863510131836,
      "learning_rate": 3.6876819983861346e-05,
      "loss": 0.5531,
      "step": 21296
    },
    {
      "epoch": 6.7246605620461,
      "grad_norm": 5.751456260681152,
      "learning_rate": 3.687594288320528e-05,
      "loss": 0.6403,
      "step": 21297
    },
    {
      "epoch": 6.724976318282286,
      "grad_norm": 5.492307662963867,
      "learning_rate": 3.687506578254921e-05,
      "loss": 0.5074,
      "step": 21298
    },
    {
      "epoch": 6.725292074518472,
      "grad_norm": 5.9255452156066895,
      "learning_rate": 3.6874188681893134e-05,
      "loss": 0.5337,
      "step": 21299
    },
    {
      "epoch": 6.725607830754657,
      "grad_norm": 4.4575018882751465,
      "learning_rate": 3.6873311581237066e-05,
      "loss": 0.5093,
      "step": 21300
    },
    {
      "epoch": 6.725923586990843,
      "grad_norm": 5.092848300933838,
      "learning_rate": 3.687243448058099e-05,
      "loss": 0.5916,
      "step": 21301
    },
    {
      "epoch": 6.726239343227029,
      "grad_norm": 4.332483291625977,
      "learning_rate": 3.687155737992492e-05,
      "loss": 0.4886,
      "step": 21302
    },
    {
      "epoch": 6.726555099463214,
      "grad_norm": 3.867645263671875,
      "learning_rate": 3.687068027926885e-05,
      "loss": 0.4841,
      "step": 21303
    },
    {
      "epoch": 6.7268708556994,
      "grad_norm": 4.521482467651367,
      "learning_rate": 3.686980317861278e-05,
      "loss": 0.6005,
      "step": 21304
    },
    {
      "epoch": 6.727186611935585,
      "grad_norm": 5.551029682159424,
      "learning_rate": 3.6868926077956704e-05,
      "loss": 0.5213,
      "step": 21305
    },
    {
      "epoch": 6.727502368171772,
      "grad_norm": 4.689298152923584,
      "learning_rate": 3.6868048977300636e-05,
      "loss": 0.6112,
      "step": 21306
    },
    {
      "epoch": 6.727818124407957,
      "grad_norm": 6.020059585571289,
      "learning_rate": 3.686717187664456e-05,
      "loss": 0.6333,
      "step": 21307
    },
    {
      "epoch": 6.728133880644143,
      "grad_norm": 5.4344072341918945,
      "learning_rate": 3.68662947759885e-05,
      "loss": 0.4964,
      "step": 21308
    },
    {
      "epoch": 6.728449636880328,
      "grad_norm": 6.068309307098389,
      "learning_rate": 3.6865417675332424e-05,
      "loss": 0.6163,
      "step": 21309
    },
    {
      "epoch": 6.728765393116514,
      "grad_norm": 5.131680488586426,
      "learning_rate": 3.6864540574676356e-05,
      "loss": 0.586,
      "step": 21310
    },
    {
      "epoch": 6.7290811493527,
      "grad_norm": 6.190831184387207,
      "learning_rate": 3.686366347402028e-05,
      "loss": 0.5938,
      "step": 21311
    },
    {
      "epoch": 6.729396905588885,
      "grad_norm": 4.204392433166504,
      "learning_rate": 3.686278637336421e-05,
      "loss": 0.4799,
      "step": 21312
    },
    {
      "epoch": 6.729712661825071,
      "grad_norm": 5.017548561096191,
      "learning_rate": 3.686190927270814e-05,
      "loss": 0.6405,
      "step": 21313
    },
    {
      "epoch": 6.730028418061257,
      "grad_norm": 4.446895599365234,
      "learning_rate": 3.686103217205207e-05,
      "loss": 0.4503,
      "step": 21314
    },
    {
      "epoch": 6.730344174297443,
      "grad_norm": 5.836540222167969,
      "learning_rate": 3.6860155071395994e-05,
      "loss": 0.5612,
      "step": 21315
    },
    {
      "epoch": 6.730659930533628,
      "grad_norm": 5.357979774475098,
      "learning_rate": 3.6859277970739926e-05,
      "loss": 0.6502,
      "step": 21316
    },
    {
      "epoch": 6.730975686769813,
      "grad_norm": 5.151202201843262,
      "learning_rate": 3.685840087008385e-05,
      "loss": 0.6471,
      "step": 21317
    },
    {
      "epoch": 6.731291443006,
      "grad_norm": 7.416875839233398,
      "learning_rate": 3.685752376942778e-05,
      "loss": 0.5231,
      "step": 21318
    },
    {
      "epoch": 6.731607199242185,
      "grad_norm": 5.040287017822266,
      "learning_rate": 3.685664666877171e-05,
      "loss": 0.5676,
      "step": 21319
    },
    {
      "epoch": 6.731922955478371,
      "grad_norm": 4.808198928833008,
      "learning_rate": 3.685576956811564e-05,
      "loss": 0.4438,
      "step": 21320
    },
    {
      "epoch": 6.732238711714556,
      "grad_norm": 4.973333358764648,
      "learning_rate": 3.685489246745957e-05,
      "loss": 0.6524,
      "step": 21321
    },
    {
      "epoch": 6.732554467950742,
      "grad_norm": 5.395985126495361,
      "learning_rate": 3.6854015366803495e-05,
      "loss": 0.6252,
      "step": 21322
    },
    {
      "epoch": 6.732870224186928,
      "grad_norm": 4.759546756744385,
      "learning_rate": 3.685313826614743e-05,
      "loss": 0.5463,
      "step": 21323
    },
    {
      "epoch": 6.733185980423113,
      "grad_norm": 4.474764347076416,
      "learning_rate": 3.685226116549135e-05,
      "loss": 0.5307,
      "step": 21324
    },
    {
      "epoch": 6.733501736659299,
      "grad_norm": 5.2847747802734375,
      "learning_rate": 3.6851384064835284e-05,
      "loss": 0.6227,
      "step": 21325
    },
    {
      "epoch": 6.733817492895485,
      "grad_norm": 5.443298816680908,
      "learning_rate": 3.685050696417921e-05,
      "loss": 0.6335,
      "step": 21326
    },
    {
      "epoch": 6.73413324913167,
      "grad_norm": 5.351893424987793,
      "learning_rate": 3.684962986352314e-05,
      "loss": 0.6359,
      "step": 21327
    },
    {
      "epoch": 6.734449005367856,
      "grad_norm": 4.946588039398193,
      "learning_rate": 3.6848752762867065e-05,
      "loss": 0.5981,
      "step": 21328
    },
    {
      "epoch": 6.734764761604041,
      "grad_norm": 6.060284614562988,
      "learning_rate": 3.6847875662211e-05,
      "loss": 0.6426,
      "step": 21329
    },
    {
      "epoch": 6.7350805178402275,
      "grad_norm": 5.726293563842773,
      "learning_rate": 3.684699856155492e-05,
      "loss": 0.5044,
      "step": 21330
    },
    {
      "epoch": 6.735396274076413,
      "grad_norm": 4.785084247589111,
      "learning_rate": 3.684612146089885e-05,
      "loss": 0.6058,
      "step": 21331
    },
    {
      "epoch": 6.735712030312599,
      "grad_norm": 4.9157819747924805,
      "learning_rate": 3.684524436024278e-05,
      "loss": 0.4767,
      "step": 21332
    },
    {
      "epoch": 6.736027786548784,
      "grad_norm": 4.952643394470215,
      "learning_rate": 3.684436725958671e-05,
      "loss": 0.4919,
      "step": 21333
    },
    {
      "epoch": 6.7363435427849705,
      "grad_norm": 4.962165355682373,
      "learning_rate": 3.6843490158930635e-05,
      "loss": 0.5153,
      "step": 21334
    },
    {
      "epoch": 6.736659299021156,
      "grad_norm": 5.18610954284668,
      "learning_rate": 3.684261305827457e-05,
      "loss": 0.5883,
      "step": 21335
    },
    {
      "epoch": 6.736975055257341,
      "grad_norm": 6.112578868865967,
      "learning_rate": 3.68417359576185e-05,
      "loss": 0.5349,
      "step": 21336
    },
    {
      "epoch": 6.737290811493527,
      "grad_norm": 5.343876838684082,
      "learning_rate": 3.684085885696243e-05,
      "loss": 0.5712,
      "step": 21337
    },
    {
      "epoch": 6.737606567729713,
      "grad_norm": 8.296669006347656,
      "learning_rate": 3.6839981756306355e-05,
      "loss": 0.6157,
      "step": 21338
    },
    {
      "epoch": 6.737922323965899,
      "grad_norm": 5.1427226066589355,
      "learning_rate": 3.6839104655650287e-05,
      "loss": 0.5175,
      "step": 21339
    },
    {
      "epoch": 6.738238080202084,
      "grad_norm": 4.902616500854492,
      "learning_rate": 3.683822755499421e-05,
      "loss": 0.6128,
      "step": 21340
    },
    {
      "epoch": 6.738553836438269,
      "grad_norm": 5.6587371826171875,
      "learning_rate": 3.683735045433814e-05,
      "loss": 0.5937,
      "step": 21341
    },
    {
      "epoch": 6.7388695926744555,
      "grad_norm": 5.438665390014648,
      "learning_rate": 3.6836473353682075e-05,
      "loss": 0.5802,
      "step": 21342
    },
    {
      "epoch": 6.739185348910641,
      "grad_norm": 4.966044902801514,
      "learning_rate": 3.6835596253026e-05,
      "loss": 0.5822,
      "step": 21343
    },
    {
      "epoch": 6.739501105146827,
      "grad_norm": 5.264739990234375,
      "learning_rate": 3.683471915236993e-05,
      "loss": 0.5837,
      "step": 21344
    },
    {
      "epoch": 6.739816861383012,
      "grad_norm": 4.683414936065674,
      "learning_rate": 3.6833842051713856e-05,
      "loss": 0.4928,
      "step": 21345
    },
    {
      "epoch": 6.740132617619198,
      "grad_norm": 4.497363090515137,
      "learning_rate": 3.683296495105779e-05,
      "loss": 0.6443,
      "step": 21346
    },
    {
      "epoch": 6.740448373855384,
      "grad_norm": 5.616392135620117,
      "learning_rate": 3.683208785040171e-05,
      "loss": 0.5964,
      "step": 21347
    },
    {
      "epoch": 6.740764130091569,
      "grad_norm": 5.135204315185547,
      "learning_rate": 3.6831210749745645e-05,
      "loss": 0.539,
      "step": 21348
    },
    {
      "epoch": 6.741079886327755,
      "grad_norm": 4.174044132232666,
      "learning_rate": 3.683033364908957e-05,
      "loss": 0.5304,
      "step": 21349
    },
    {
      "epoch": 6.7413956425639405,
      "grad_norm": 4.39529275894165,
      "learning_rate": 3.68294565484335e-05,
      "loss": 0.4436,
      "step": 21350
    },
    {
      "epoch": 6.741711398800126,
      "grad_norm": 3.909923553466797,
      "learning_rate": 3.6828579447777426e-05,
      "loss": 0.5174,
      "step": 21351
    },
    {
      "epoch": 6.742027155036312,
      "grad_norm": 3.8869051933288574,
      "learning_rate": 3.682770234712136e-05,
      "loss": 0.5942,
      "step": 21352
    },
    {
      "epoch": 6.742342911272497,
      "grad_norm": 5.657345294952393,
      "learning_rate": 3.682682524646528e-05,
      "loss": 0.7039,
      "step": 21353
    },
    {
      "epoch": 6.7426586675086835,
      "grad_norm": 4.404757499694824,
      "learning_rate": 3.6825948145809214e-05,
      "loss": 0.5229,
      "step": 21354
    },
    {
      "epoch": 6.742974423744869,
      "grad_norm": 4.68736457824707,
      "learning_rate": 3.682507104515314e-05,
      "loss": 0.5359,
      "step": 21355
    },
    {
      "epoch": 6.743290179981055,
      "grad_norm": 5.422234058380127,
      "learning_rate": 3.682419394449707e-05,
      "loss": 0.6233,
      "step": 21356
    },
    {
      "epoch": 6.74360593621724,
      "grad_norm": 4.30955171585083,
      "learning_rate": 3.6823316843841e-05,
      "loss": 0.5436,
      "step": 21357
    },
    {
      "epoch": 6.743921692453426,
      "grad_norm": 4.664994239807129,
      "learning_rate": 3.682243974318493e-05,
      "loss": 0.4824,
      "step": 21358
    },
    {
      "epoch": 6.744237448689612,
      "grad_norm": 5.713696002960205,
      "learning_rate": 3.682156264252886e-05,
      "loss": 0.5546,
      "step": 21359
    },
    {
      "epoch": 6.744553204925797,
      "grad_norm": 4.677418231964111,
      "learning_rate": 3.6820685541872784e-05,
      "loss": 0.5557,
      "step": 21360
    },
    {
      "epoch": 6.744868961161983,
      "grad_norm": 5.7112321853637695,
      "learning_rate": 3.6819808441216716e-05,
      "loss": 0.534,
      "step": 21361
    },
    {
      "epoch": 6.7451847173981685,
      "grad_norm": 4.62410306930542,
      "learning_rate": 3.681893134056064e-05,
      "loss": 0.5219,
      "step": 21362
    },
    {
      "epoch": 6.745500473634355,
      "grad_norm": 5.061739444732666,
      "learning_rate": 3.681805423990458e-05,
      "loss": 0.5705,
      "step": 21363
    },
    {
      "epoch": 6.74581622987054,
      "grad_norm": 5.325164794921875,
      "learning_rate": 3.6817177139248504e-05,
      "loss": 0.5907,
      "step": 21364
    },
    {
      "epoch": 6.746131986106725,
      "grad_norm": 5.097352027893066,
      "learning_rate": 3.6816300038592436e-05,
      "loss": 0.5599,
      "step": 21365
    },
    {
      "epoch": 6.7464477423429114,
      "grad_norm": 4.386279582977295,
      "learning_rate": 3.681542293793636e-05,
      "loss": 0.5143,
      "step": 21366
    },
    {
      "epoch": 6.746763498579097,
      "grad_norm": 5.052946090698242,
      "learning_rate": 3.681454583728029e-05,
      "loss": 0.5542,
      "step": 21367
    },
    {
      "epoch": 6.747079254815283,
      "grad_norm": 4.832147121429443,
      "learning_rate": 3.681366873662422e-05,
      "loss": 0.5387,
      "step": 21368
    },
    {
      "epoch": 6.747395011051468,
      "grad_norm": 5.410723686218262,
      "learning_rate": 3.681279163596815e-05,
      "loss": 0.5967,
      "step": 21369
    },
    {
      "epoch": 6.7477107672876535,
      "grad_norm": 4.494721412658691,
      "learning_rate": 3.6811914535312074e-05,
      "loss": 0.5523,
      "step": 21370
    },
    {
      "epoch": 6.74802652352384,
      "grad_norm": 4.86842155456543,
      "learning_rate": 3.6811037434656005e-05,
      "loss": 0.6028,
      "step": 21371
    },
    {
      "epoch": 6.748342279760025,
      "grad_norm": 5.648841857910156,
      "learning_rate": 3.681016033399993e-05,
      "loss": 0.5367,
      "step": 21372
    },
    {
      "epoch": 6.748658035996211,
      "grad_norm": 5.3822503089904785,
      "learning_rate": 3.680928323334386e-05,
      "loss": 0.6054,
      "step": 21373
    },
    {
      "epoch": 6.7489737922323965,
      "grad_norm": 6.134334087371826,
      "learning_rate": 3.680840613268779e-05,
      "loss": 0.6495,
      "step": 21374
    },
    {
      "epoch": 6.749289548468582,
      "grad_norm": 3.917837619781494,
      "learning_rate": 3.680752903203172e-05,
      "loss": 0.5455,
      "step": 21375
    },
    {
      "epoch": 6.749605304704768,
      "grad_norm": 4.658244609832764,
      "learning_rate": 3.6806651931375644e-05,
      "loss": 0.5674,
      "step": 21376
    },
    {
      "epoch": 6.749921060940953,
      "grad_norm": 5.327985763549805,
      "learning_rate": 3.6805774830719575e-05,
      "loss": 0.4573,
      "step": 21377
    },
    {
      "epoch": 6.750236817177139,
      "grad_norm": 4.741667747497559,
      "learning_rate": 3.68048977300635e-05,
      "loss": 0.5081,
      "step": 21378
    },
    {
      "epoch": 6.750552573413325,
      "grad_norm": 4.049291610717773,
      "learning_rate": 3.680402062940743e-05,
      "loss": 0.5629,
      "step": 21379
    },
    {
      "epoch": 6.750868329649511,
      "grad_norm": 5.569511890411377,
      "learning_rate": 3.6803143528751364e-05,
      "loss": 0.6812,
      "step": 21380
    },
    {
      "epoch": 6.751184085885696,
      "grad_norm": 5.045998573303223,
      "learning_rate": 3.680226642809529e-05,
      "loss": 0.4607,
      "step": 21381
    },
    {
      "epoch": 6.751499842121882,
      "grad_norm": 4.641425132751465,
      "learning_rate": 3.680138932743922e-05,
      "loss": 0.5114,
      "step": 21382
    },
    {
      "epoch": 6.751815598358068,
      "grad_norm": 4.473003387451172,
      "learning_rate": 3.6800512226783145e-05,
      "loss": 0.5048,
      "step": 21383
    },
    {
      "epoch": 6.752131354594253,
      "grad_norm": 4.896113872528076,
      "learning_rate": 3.679963512612708e-05,
      "loss": 0.4153,
      "step": 21384
    },
    {
      "epoch": 6.752447110830439,
      "grad_norm": 5.766927242279053,
      "learning_rate": 3.6798758025471e-05,
      "loss": 0.6658,
      "step": 21385
    },
    {
      "epoch": 6.752762867066624,
      "grad_norm": 5.612339496612549,
      "learning_rate": 3.679788092481493e-05,
      "loss": 0.5726,
      "step": 21386
    },
    {
      "epoch": 6.753078623302811,
      "grad_norm": 5.051956653594971,
      "learning_rate": 3.679700382415886e-05,
      "loss": 0.487,
      "step": 21387
    },
    {
      "epoch": 6.753394379538996,
      "grad_norm": 4.305765628814697,
      "learning_rate": 3.679612672350279e-05,
      "loss": 0.5521,
      "step": 21388
    },
    {
      "epoch": 6.753710135775181,
      "grad_norm": 5.760560512542725,
      "learning_rate": 3.6795249622846715e-05,
      "loss": 0.6312,
      "step": 21389
    },
    {
      "epoch": 6.754025892011367,
      "grad_norm": 6.698605537414551,
      "learning_rate": 3.679437252219065e-05,
      "loss": 0.7,
      "step": 21390
    },
    {
      "epoch": 6.754341648247553,
      "grad_norm": 6.032281398773193,
      "learning_rate": 3.679349542153458e-05,
      "loss": 0.598,
      "step": 21391
    },
    {
      "epoch": 6.754657404483739,
      "grad_norm": 4.731136322021484,
      "learning_rate": 3.679261832087851e-05,
      "loss": 0.5203,
      "step": 21392
    },
    {
      "epoch": 6.754973160719924,
      "grad_norm": 5.306649208068848,
      "learning_rate": 3.6791741220222435e-05,
      "loss": 0.571,
      "step": 21393
    },
    {
      "epoch": 6.7552889169561094,
      "grad_norm": 4.706419944763184,
      "learning_rate": 3.6790864119566366e-05,
      "loss": 0.4981,
      "step": 21394
    },
    {
      "epoch": 6.755604673192296,
      "grad_norm": 4.276810169219971,
      "learning_rate": 3.678998701891029e-05,
      "loss": 0.4837,
      "step": 21395
    },
    {
      "epoch": 6.755920429428481,
      "grad_norm": 5.360541820526123,
      "learning_rate": 3.678910991825422e-05,
      "loss": 0.5707,
      "step": 21396
    },
    {
      "epoch": 6.756236185664667,
      "grad_norm": 5.481532096862793,
      "learning_rate": 3.678823281759815e-05,
      "loss": 0.5978,
      "step": 21397
    },
    {
      "epoch": 6.756551941900852,
      "grad_norm": 4.548281669616699,
      "learning_rate": 3.678735571694208e-05,
      "loss": 0.5194,
      "step": 21398
    },
    {
      "epoch": 6.756867698137039,
      "grad_norm": 4.89478063583374,
      "learning_rate": 3.6786478616286004e-05,
      "loss": 0.6036,
      "step": 21399
    },
    {
      "epoch": 6.757183454373224,
      "grad_norm": 4.724964618682861,
      "learning_rate": 3.6785601515629936e-05,
      "loss": 0.4427,
      "step": 21400
    },
    {
      "epoch": 6.757183454373224,
      "eval_loss": 0.8826109170913696,
      "eval_runtime": 4.5376,
      "eval_samples_per_second": 548.751,
      "eval_steps_per_second": 4.628,
      "step": 21400
    },
    {
      "epoch": 6.757499210609409,
      "grad_norm": 5.16641092300415,
      "learning_rate": 3.678472441497387e-05,
      "loss": 0.5504,
      "step": 21401
    },
    {
      "epoch": 6.757814966845595,
      "grad_norm": 5.008870601654053,
      "learning_rate": 3.678384731431779e-05,
      "loss": 0.4556,
      "step": 21402
    },
    {
      "epoch": 6.758130723081781,
      "grad_norm": 4.45504093170166,
      "learning_rate": 3.6782970213661724e-05,
      "loss": 0.5547,
      "step": 21403
    },
    {
      "epoch": 6.758446479317967,
      "grad_norm": 4.09901762008667,
      "learning_rate": 3.678209311300565e-05,
      "loss": 0.6013,
      "step": 21404
    },
    {
      "epoch": 6.758762235554152,
      "grad_norm": 4.978071212768555,
      "learning_rate": 3.678121601234958e-05,
      "loss": 0.4665,
      "step": 21405
    },
    {
      "epoch": 6.759077991790338,
      "grad_norm": 5.338733673095703,
      "learning_rate": 3.6780338911693506e-05,
      "loss": 0.6055,
      "step": 21406
    },
    {
      "epoch": 6.759393748026524,
      "grad_norm": 4.9277448654174805,
      "learning_rate": 3.677946181103744e-05,
      "loss": 0.5663,
      "step": 21407
    },
    {
      "epoch": 6.759709504262709,
      "grad_norm": 5.4393768310546875,
      "learning_rate": 3.677858471038136e-05,
      "loss": 0.5855,
      "step": 21408
    },
    {
      "epoch": 6.760025260498895,
      "grad_norm": 6.135983943939209,
      "learning_rate": 3.6777707609725294e-05,
      "loss": 0.5944,
      "step": 21409
    },
    {
      "epoch": 6.76034101673508,
      "grad_norm": 5.385761737823486,
      "learning_rate": 3.677683050906922e-05,
      "loss": 0.538,
      "step": 21410
    },
    {
      "epoch": 6.7606567729712665,
      "grad_norm": 4.520063400268555,
      "learning_rate": 3.677595340841315e-05,
      "loss": 0.5828,
      "step": 21411
    },
    {
      "epoch": 6.760972529207452,
      "grad_norm": 4.703919410705566,
      "learning_rate": 3.6775076307757076e-05,
      "loss": 0.5771,
      "step": 21412
    },
    {
      "epoch": 6.761288285443637,
      "grad_norm": 5.85687255859375,
      "learning_rate": 3.677419920710101e-05,
      "loss": 0.4567,
      "step": 21413
    },
    {
      "epoch": 6.761604041679823,
      "grad_norm": 5.687711715698242,
      "learning_rate": 3.677332210644493e-05,
      "loss": 0.5634,
      "step": 21414
    },
    {
      "epoch": 6.761919797916009,
      "grad_norm": 7.356783390045166,
      "learning_rate": 3.6772445005788864e-05,
      "loss": 0.7122,
      "step": 21415
    },
    {
      "epoch": 6.762235554152195,
      "grad_norm": 4.757278919219971,
      "learning_rate": 3.677156790513279e-05,
      "loss": 0.5623,
      "step": 21416
    },
    {
      "epoch": 6.76255131038838,
      "grad_norm": 5.219235897064209,
      "learning_rate": 3.677069080447673e-05,
      "loss": 0.6892,
      "step": 21417
    },
    {
      "epoch": 6.762867066624565,
      "grad_norm": 6.559325695037842,
      "learning_rate": 3.676981370382065e-05,
      "loss": 0.6313,
      "step": 21418
    },
    {
      "epoch": 6.763182822860752,
      "grad_norm": 6.250279426574707,
      "learning_rate": 3.6768936603164584e-05,
      "loss": 0.6181,
      "step": 21419
    },
    {
      "epoch": 6.763498579096937,
      "grad_norm": 5.177375793457031,
      "learning_rate": 3.676805950250851e-05,
      "loss": 0.6426,
      "step": 21420
    },
    {
      "epoch": 6.763814335333123,
      "grad_norm": 5.320670127868652,
      "learning_rate": 3.676718240185244e-05,
      "loss": 0.6361,
      "step": 21421
    },
    {
      "epoch": 6.764130091569308,
      "grad_norm": 4.5862932205200195,
      "learning_rate": 3.6766305301196365e-05,
      "loss": 0.445,
      "step": 21422
    },
    {
      "epoch": 6.7644458478054945,
      "grad_norm": 5.279445171356201,
      "learning_rate": 3.67654282005403e-05,
      "loss": 0.5966,
      "step": 21423
    },
    {
      "epoch": 6.76476160404168,
      "grad_norm": 5.9509196281433105,
      "learning_rate": 3.676455109988423e-05,
      "loss": 0.5775,
      "step": 21424
    },
    {
      "epoch": 6.765077360277865,
      "grad_norm": 7.758383274078369,
      "learning_rate": 3.6763673999228154e-05,
      "loss": 0.6928,
      "step": 21425
    },
    {
      "epoch": 6.765393116514051,
      "grad_norm": 4.837049961090088,
      "learning_rate": 3.6762796898572085e-05,
      "loss": 0.4841,
      "step": 21426
    },
    {
      "epoch": 6.765708872750237,
      "grad_norm": 4.167712211608887,
      "learning_rate": 3.676191979791601e-05,
      "loss": 0.5155,
      "step": 21427
    },
    {
      "epoch": 6.766024628986423,
      "grad_norm": 4.8641133308410645,
      "learning_rate": 3.676104269725994e-05,
      "loss": 0.6498,
      "step": 21428
    },
    {
      "epoch": 6.766340385222608,
      "grad_norm": 5.3315205574035645,
      "learning_rate": 3.676016559660387e-05,
      "loss": 0.622,
      "step": 21429
    },
    {
      "epoch": 6.766656141458794,
      "grad_norm": 4.972264766693115,
      "learning_rate": 3.67592884959478e-05,
      "loss": 0.5896,
      "step": 21430
    },
    {
      "epoch": 6.7669718976949795,
      "grad_norm": 6.202009677886963,
      "learning_rate": 3.6758411395291723e-05,
      "loss": 0.4975,
      "step": 21431
    },
    {
      "epoch": 6.767287653931165,
      "grad_norm": 5.196760177612305,
      "learning_rate": 3.6757534294635655e-05,
      "loss": 0.5814,
      "step": 21432
    },
    {
      "epoch": 6.767603410167351,
      "grad_norm": 4.493005752563477,
      "learning_rate": 3.675665719397958e-05,
      "loss": 0.5592,
      "step": 21433
    },
    {
      "epoch": 6.767919166403536,
      "grad_norm": 5.2215895652771,
      "learning_rate": 3.675578009332351e-05,
      "loss": 0.5546,
      "step": 21434
    },
    {
      "epoch": 6.7682349226397225,
      "grad_norm": 4.563576698303223,
      "learning_rate": 3.675490299266744e-05,
      "loss": 0.5572,
      "step": 21435
    },
    {
      "epoch": 6.768550678875908,
      "grad_norm": 3.7017834186553955,
      "learning_rate": 3.675402589201137e-05,
      "loss": 0.5001,
      "step": 21436
    },
    {
      "epoch": 6.768866435112093,
      "grad_norm": 4.761131286621094,
      "learning_rate": 3.675314879135529e-05,
      "loss": 0.551,
      "step": 21437
    },
    {
      "epoch": 6.769182191348279,
      "grad_norm": 4.136242389678955,
      "learning_rate": 3.6752271690699225e-05,
      "loss": 0.4811,
      "step": 21438
    },
    {
      "epoch": 6.7694979475844645,
      "grad_norm": 5.389222145080566,
      "learning_rate": 3.6751394590043157e-05,
      "loss": 0.5412,
      "step": 21439
    },
    {
      "epoch": 6.769813703820651,
      "grad_norm": 4.335203647613525,
      "learning_rate": 3.675051748938708e-05,
      "loss": 0.477,
      "step": 21440
    },
    {
      "epoch": 6.770129460056836,
      "grad_norm": 5.608728408813477,
      "learning_rate": 3.674964038873101e-05,
      "loss": 0.5508,
      "step": 21441
    },
    {
      "epoch": 6.770445216293021,
      "grad_norm": 4.70160436630249,
      "learning_rate": 3.674876328807494e-05,
      "loss": 0.4653,
      "step": 21442
    },
    {
      "epoch": 6.7707609725292075,
      "grad_norm": 4.25423526763916,
      "learning_rate": 3.674788618741887e-05,
      "loss": 0.4524,
      "step": 21443
    },
    {
      "epoch": 6.771076728765393,
      "grad_norm": 5.8557257652282715,
      "learning_rate": 3.67470090867628e-05,
      "loss": 0.5353,
      "step": 21444
    },
    {
      "epoch": 6.771392485001579,
      "grad_norm": 5.575436115264893,
      "learning_rate": 3.674613198610673e-05,
      "loss": 0.6298,
      "step": 21445
    },
    {
      "epoch": 6.771708241237764,
      "grad_norm": 6.457493305206299,
      "learning_rate": 3.674525488545066e-05,
      "loss": 0.6817,
      "step": 21446
    },
    {
      "epoch": 6.7720239974739505,
      "grad_norm": 6.7873125076293945,
      "learning_rate": 3.674437778479459e-05,
      "loss": 0.5742,
      "step": 21447
    },
    {
      "epoch": 6.772339753710136,
      "grad_norm": 4.71445369720459,
      "learning_rate": 3.6743500684138515e-05,
      "loss": 0.5202,
      "step": 21448
    },
    {
      "epoch": 6.772655509946322,
      "grad_norm": 4.374838829040527,
      "learning_rate": 3.6742623583482446e-05,
      "loss": 0.513,
      "step": 21449
    },
    {
      "epoch": 6.772971266182507,
      "grad_norm": 4.516148567199707,
      "learning_rate": 3.674174648282637e-05,
      "loss": 0.6232,
      "step": 21450
    },
    {
      "epoch": 6.7732870224186925,
      "grad_norm": 4.939082622528076,
      "learning_rate": 3.67408693821703e-05,
      "loss": 0.5398,
      "step": 21451
    },
    {
      "epoch": 6.773602778654879,
      "grad_norm": 5.515450477600098,
      "learning_rate": 3.673999228151423e-05,
      "loss": 0.5489,
      "step": 21452
    },
    {
      "epoch": 6.773918534891064,
      "grad_norm": 4.755683422088623,
      "learning_rate": 3.673911518085816e-05,
      "loss": 0.5655,
      "step": 21453
    },
    {
      "epoch": 6.77423429112725,
      "grad_norm": 4.604780197143555,
      "learning_rate": 3.6738238080202084e-05,
      "loss": 0.5322,
      "step": 21454
    },
    {
      "epoch": 6.7745500473634355,
      "grad_norm": 5.8408050537109375,
      "learning_rate": 3.6737360979546016e-05,
      "loss": 0.63,
      "step": 21455
    },
    {
      "epoch": 6.774865803599621,
      "grad_norm": 5.849665641784668,
      "learning_rate": 3.673648387888994e-05,
      "loss": 0.6198,
      "step": 21456
    },
    {
      "epoch": 6.775181559835807,
      "grad_norm": 4.989314079284668,
      "learning_rate": 3.673560677823387e-05,
      "loss": 0.5675,
      "step": 21457
    },
    {
      "epoch": 6.775497316071992,
      "grad_norm": 5.179619312286377,
      "learning_rate": 3.67347296775778e-05,
      "loss": 0.5937,
      "step": 21458
    },
    {
      "epoch": 6.775813072308178,
      "grad_norm": 5.669928550720215,
      "learning_rate": 3.673385257692173e-05,
      "loss": 0.646,
      "step": 21459
    },
    {
      "epoch": 6.776128828544364,
      "grad_norm": 4.5068840980529785,
      "learning_rate": 3.6732975476265654e-05,
      "loss": 0.4879,
      "step": 21460
    },
    {
      "epoch": 6.776444584780549,
      "grad_norm": 5.6655707359313965,
      "learning_rate": 3.6732098375609586e-05,
      "loss": 0.5034,
      "step": 21461
    },
    {
      "epoch": 6.776760341016735,
      "grad_norm": 5.232656478881836,
      "learning_rate": 3.673122127495352e-05,
      "loss": 0.6413,
      "step": 21462
    },
    {
      "epoch": 6.7770760972529205,
      "grad_norm": 4.836032390594482,
      "learning_rate": 3.673034417429744e-05,
      "loss": 0.632,
      "step": 21463
    },
    {
      "epoch": 6.777391853489107,
      "grad_norm": 3.783579111099243,
      "learning_rate": 3.6729467073641374e-05,
      "loss": 0.4575,
      "step": 21464
    },
    {
      "epoch": 6.777707609725292,
      "grad_norm": 4.727510929107666,
      "learning_rate": 3.67285899729853e-05,
      "loss": 0.5362,
      "step": 21465
    },
    {
      "epoch": 6.778023365961477,
      "grad_norm": 4.264166831970215,
      "learning_rate": 3.672771287232923e-05,
      "loss": 0.55,
      "step": 21466
    },
    {
      "epoch": 6.778339122197663,
      "grad_norm": 4.960065841674805,
      "learning_rate": 3.6726835771673156e-05,
      "loss": 0.4264,
      "step": 21467
    },
    {
      "epoch": 6.778654878433849,
      "grad_norm": 4.667893409729004,
      "learning_rate": 3.672595867101709e-05,
      "loss": 0.5666,
      "step": 21468
    },
    {
      "epoch": 6.778970634670035,
      "grad_norm": 4.754735946655273,
      "learning_rate": 3.672508157036101e-05,
      "loss": 0.5262,
      "step": 21469
    },
    {
      "epoch": 6.77928639090622,
      "grad_norm": 5.577048301696777,
      "learning_rate": 3.6724204469704944e-05,
      "loss": 0.482,
      "step": 21470
    },
    {
      "epoch": 6.779602147142406,
      "grad_norm": 4.892318248748779,
      "learning_rate": 3.672332736904887e-05,
      "loss": 0.5464,
      "step": 21471
    },
    {
      "epoch": 6.779917903378592,
      "grad_norm": 4.632786750793457,
      "learning_rate": 3.672245026839281e-05,
      "loss": 0.6373,
      "step": 21472
    },
    {
      "epoch": 6.780233659614778,
      "grad_norm": 5.234057426452637,
      "learning_rate": 3.672157316773673e-05,
      "loss": 0.5527,
      "step": 21473
    },
    {
      "epoch": 6.780549415850963,
      "grad_norm": 4.773049354553223,
      "learning_rate": 3.6720696067080664e-05,
      "loss": 0.5548,
      "step": 21474
    },
    {
      "epoch": 6.7808651720871485,
      "grad_norm": 5.118930816650391,
      "learning_rate": 3.671981896642459e-05,
      "loss": 0.4921,
      "step": 21475
    },
    {
      "epoch": 6.781180928323335,
      "grad_norm": 6.098283290863037,
      "learning_rate": 3.671894186576852e-05,
      "loss": 0.6141,
      "step": 21476
    },
    {
      "epoch": 6.78149668455952,
      "grad_norm": 4.305027961730957,
      "learning_rate": 3.6718064765112445e-05,
      "loss": 0.5647,
      "step": 21477
    },
    {
      "epoch": 6.781812440795706,
      "grad_norm": 4.990242958068848,
      "learning_rate": 3.671718766445638e-05,
      "loss": 0.6156,
      "step": 21478
    },
    {
      "epoch": 6.782128197031891,
      "grad_norm": 4.1636457443237305,
      "learning_rate": 3.67163105638003e-05,
      "loss": 0.5088,
      "step": 21479
    },
    {
      "epoch": 6.782443953268077,
      "grad_norm": 4.793511390686035,
      "learning_rate": 3.6715433463144234e-05,
      "loss": 0.5002,
      "step": 21480
    },
    {
      "epoch": 6.782759709504263,
      "grad_norm": 5.569430351257324,
      "learning_rate": 3.671455636248816e-05,
      "loss": 0.6246,
      "step": 21481
    },
    {
      "epoch": 6.783075465740448,
      "grad_norm": 5.054793357849121,
      "learning_rate": 3.671367926183209e-05,
      "loss": 0.5334,
      "step": 21482
    },
    {
      "epoch": 6.783391221976634,
      "grad_norm": 4.646552562713623,
      "learning_rate": 3.671280216117602e-05,
      "loss": 0.554,
      "step": 21483
    },
    {
      "epoch": 6.78370697821282,
      "grad_norm": 6.084089279174805,
      "learning_rate": 3.671192506051995e-05,
      "loss": 0.5272,
      "step": 21484
    },
    {
      "epoch": 6.784022734449005,
      "grad_norm": 5.534051895141602,
      "learning_rate": 3.671104795986388e-05,
      "loss": 0.5797,
      "step": 21485
    },
    {
      "epoch": 6.784338490685191,
      "grad_norm": 5.105374336242676,
      "learning_rate": 3.67101708592078e-05,
      "loss": 0.6141,
      "step": 21486
    },
    {
      "epoch": 6.784654246921376,
      "grad_norm": 4.828289985656738,
      "learning_rate": 3.6709293758551735e-05,
      "loss": 0.515,
      "step": 21487
    },
    {
      "epoch": 6.784970003157563,
      "grad_norm": 4.531754016876221,
      "learning_rate": 3.670841665789566e-05,
      "loss": 0.6237,
      "step": 21488
    },
    {
      "epoch": 6.785285759393748,
      "grad_norm": 6.197349548339844,
      "learning_rate": 3.670753955723959e-05,
      "loss": 0.5355,
      "step": 21489
    },
    {
      "epoch": 6.785601515629934,
      "grad_norm": 4.348756790161133,
      "learning_rate": 3.6706662456583517e-05,
      "loss": 0.5872,
      "step": 21490
    },
    {
      "epoch": 6.785917271866119,
      "grad_norm": 5.574090957641602,
      "learning_rate": 3.670578535592745e-05,
      "loss": 0.6069,
      "step": 21491
    },
    {
      "epoch": 6.786233028102305,
      "grad_norm": 4.734564781188965,
      "learning_rate": 3.670490825527137e-05,
      "loss": 0.5365,
      "step": 21492
    },
    {
      "epoch": 6.786548784338491,
      "grad_norm": 5.1992011070251465,
      "learning_rate": 3.6704031154615305e-05,
      "loss": 0.5073,
      "step": 21493
    },
    {
      "epoch": 6.786864540574676,
      "grad_norm": 4.769522666931152,
      "learning_rate": 3.670315405395923e-05,
      "loss": 0.4913,
      "step": 21494
    },
    {
      "epoch": 6.787180296810862,
      "grad_norm": 3.640864849090576,
      "learning_rate": 3.670227695330316e-05,
      "loss": 0.4493,
      "step": 21495
    },
    {
      "epoch": 6.787496053047048,
      "grad_norm": 5.148050785064697,
      "learning_rate": 3.6701399852647086e-05,
      "loss": 0.542,
      "step": 21496
    },
    {
      "epoch": 6.787811809283234,
      "grad_norm": 6.5962300300598145,
      "learning_rate": 3.670052275199102e-05,
      "loss": 0.6086,
      "step": 21497
    },
    {
      "epoch": 6.788127565519419,
      "grad_norm": 4.430952548980713,
      "learning_rate": 3.669964565133494e-05,
      "loss": 0.5555,
      "step": 21498
    },
    {
      "epoch": 6.788443321755604,
      "grad_norm": 4.0994648933410645,
      "learning_rate": 3.669876855067888e-05,
      "loss": 0.4098,
      "step": 21499
    },
    {
      "epoch": 6.788759077991791,
      "grad_norm": 6.579324722290039,
      "learning_rate": 3.6697891450022806e-05,
      "loss": 0.6279,
      "step": 21500
    },
    {
      "epoch": 6.789074834227976,
      "grad_norm": 5.215522289276123,
      "learning_rate": 3.669701434936674e-05,
      "loss": 0.5634,
      "step": 21501
    },
    {
      "epoch": 6.789390590464162,
      "grad_norm": 4.932587623596191,
      "learning_rate": 3.669613724871066e-05,
      "loss": 0.4284,
      "step": 21502
    },
    {
      "epoch": 6.789706346700347,
      "grad_norm": 5.35141658782959,
      "learning_rate": 3.6695260148054595e-05,
      "loss": 0.5756,
      "step": 21503
    },
    {
      "epoch": 6.790022102936533,
      "grad_norm": 5.17038631439209,
      "learning_rate": 3.6694383047398526e-05,
      "loss": 0.6177,
      "step": 21504
    },
    {
      "epoch": 6.790337859172719,
      "grad_norm": 5.141813278198242,
      "learning_rate": 3.669350594674245e-05,
      "loss": 0.5674,
      "step": 21505
    },
    {
      "epoch": 6.790653615408904,
      "grad_norm": 4.941184997558594,
      "learning_rate": 3.669262884608638e-05,
      "loss": 0.5253,
      "step": 21506
    },
    {
      "epoch": 6.79096937164509,
      "grad_norm": 5.417403221130371,
      "learning_rate": 3.669175174543031e-05,
      "loss": 0.5603,
      "step": 21507
    },
    {
      "epoch": 6.791285127881276,
      "grad_norm": 5.516692638397217,
      "learning_rate": 3.669087464477424e-05,
      "loss": 0.6238,
      "step": 21508
    },
    {
      "epoch": 6.791600884117461,
      "grad_norm": 5.659553527832031,
      "learning_rate": 3.6689997544118164e-05,
      "loss": 0.5187,
      "step": 21509
    },
    {
      "epoch": 6.791916640353647,
      "grad_norm": 6.184413433074951,
      "learning_rate": 3.6689120443462096e-05,
      "loss": 0.7067,
      "step": 21510
    },
    {
      "epoch": 6.792232396589832,
      "grad_norm": 5.745368957519531,
      "learning_rate": 3.668824334280602e-05,
      "loss": 0.5561,
      "step": 21511
    },
    {
      "epoch": 6.7925481528260185,
      "grad_norm": 6.183125019073486,
      "learning_rate": 3.668736624214995e-05,
      "loss": 0.5955,
      "step": 21512
    },
    {
      "epoch": 6.792863909062204,
      "grad_norm": 4.624324798583984,
      "learning_rate": 3.668648914149388e-05,
      "loss": 0.5358,
      "step": 21513
    },
    {
      "epoch": 6.79317966529839,
      "grad_norm": 4.349135875701904,
      "learning_rate": 3.668561204083781e-05,
      "loss": 0.5452,
      "step": 21514
    },
    {
      "epoch": 6.793495421534575,
      "grad_norm": 5.466526031494141,
      "learning_rate": 3.6684734940181734e-05,
      "loss": 0.575,
      "step": 21515
    },
    {
      "epoch": 6.793811177770761,
      "grad_norm": 6.283191680908203,
      "learning_rate": 3.6683857839525666e-05,
      "loss": 0.5033,
      "step": 21516
    },
    {
      "epoch": 6.794126934006947,
      "grad_norm": 5.23251485824585,
      "learning_rate": 3.668298073886959e-05,
      "loss": 0.5258,
      "step": 21517
    },
    {
      "epoch": 6.794442690243132,
      "grad_norm": 5.142614364624023,
      "learning_rate": 3.668210363821352e-05,
      "loss": 0.6076,
      "step": 21518
    },
    {
      "epoch": 6.794758446479318,
      "grad_norm": 6.442278861999512,
      "learning_rate": 3.668122653755745e-05,
      "loss": 0.6455,
      "step": 21519
    },
    {
      "epoch": 6.7950742027155036,
      "grad_norm": 5.008676052093506,
      "learning_rate": 3.668034943690138e-05,
      "loss": 0.6878,
      "step": 21520
    },
    {
      "epoch": 6.79538995895169,
      "grad_norm": 5.816427230834961,
      "learning_rate": 3.667947233624531e-05,
      "loss": 0.5523,
      "step": 21521
    },
    {
      "epoch": 6.795705715187875,
      "grad_norm": 7.704687118530273,
      "learning_rate": 3.6678595235589235e-05,
      "loss": 0.6689,
      "step": 21522
    },
    {
      "epoch": 6.79602147142406,
      "grad_norm": 4.880266189575195,
      "learning_rate": 3.667771813493317e-05,
      "loss": 0.5868,
      "step": 21523
    },
    {
      "epoch": 6.7963372276602465,
      "grad_norm": 3.922844886779785,
      "learning_rate": 3.667684103427709e-05,
      "loss": 0.5254,
      "step": 21524
    },
    {
      "epoch": 6.796652983896432,
      "grad_norm": 4.95676326751709,
      "learning_rate": 3.6675963933621024e-05,
      "loss": 0.577,
      "step": 21525
    },
    {
      "epoch": 6.796968740132618,
      "grad_norm": 5.203453540802002,
      "learning_rate": 3.6675086832964955e-05,
      "loss": 0.4957,
      "step": 21526
    },
    {
      "epoch": 6.797284496368803,
      "grad_norm": 5.428767204284668,
      "learning_rate": 3.667420973230889e-05,
      "loss": 0.6433,
      "step": 21527
    },
    {
      "epoch": 6.797600252604989,
      "grad_norm": 5.009040355682373,
      "learning_rate": 3.667333263165281e-05,
      "loss": 0.4707,
      "step": 21528
    },
    {
      "epoch": 6.797916008841175,
      "grad_norm": 5.319087982177734,
      "learning_rate": 3.6672455530996744e-05,
      "loss": 0.5706,
      "step": 21529
    },
    {
      "epoch": 6.79823176507736,
      "grad_norm": 5.406466484069824,
      "learning_rate": 3.667157843034067e-05,
      "loss": 0.5393,
      "step": 21530
    },
    {
      "epoch": 6.798547521313546,
      "grad_norm": 4.768966197967529,
      "learning_rate": 3.66707013296846e-05,
      "loss": 0.5726,
      "step": 21531
    },
    {
      "epoch": 6.7988632775497315,
      "grad_norm": 5.901907444000244,
      "learning_rate": 3.6669824229028525e-05,
      "loss": 0.5643,
      "step": 21532
    },
    {
      "epoch": 6.799179033785917,
      "grad_norm": 5.2954230308532715,
      "learning_rate": 3.666894712837246e-05,
      "loss": 0.673,
      "step": 21533
    },
    {
      "epoch": 6.799494790022103,
      "grad_norm": 4.821281433105469,
      "learning_rate": 3.666807002771638e-05,
      "loss": 0.5445,
      "step": 21534
    },
    {
      "epoch": 6.799810546258288,
      "grad_norm": 4.440164089202881,
      "learning_rate": 3.6667192927060313e-05,
      "loss": 0.6016,
      "step": 21535
    },
    {
      "epoch": 6.8001263024944745,
      "grad_norm": 5.765815734863281,
      "learning_rate": 3.666631582640424e-05,
      "loss": 0.6518,
      "step": 21536
    },
    {
      "epoch": 6.80044205873066,
      "grad_norm": 4.933181285858154,
      "learning_rate": 3.666543872574817e-05,
      "loss": 0.7141,
      "step": 21537
    },
    {
      "epoch": 6.800757814966846,
      "grad_norm": 5.607255458831787,
      "learning_rate": 3.6664561625092095e-05,
      "loss": 0.5195,
      "step": 21538
    },
    {
      "epoch": 6.801073571203031,
      "grad_norm": 4.9570112228393555,
      "learning_rate": 3.666368452443603e-05,
      "loss": 0.6325,
      "step": 21539
    },
    {
      "epoch": 6.8013893274392165,
      "grad_norm": 5.006957054138184,
      "learning_rate": 3.666280742377995e-05,
      "loss": 0.5201,
      "step": 21540
    },
    {
      "epoch": 6.801705083675403,
      "grad_norm": 4.770092964172363,
      "learning_rate": 3.666193032312388e-05,
      "loss": 0.5909,
      "step": 21541
    },
    {
      "epoch": 6.802020839911588,
      "grad_norm": 5.475422382354736,
      "learning_rate": 3.6661053222467815e-05,
      "loss": 0.5166,
      "step": 21542
    },
    {
      "epoch": 6.802336596147774,
      "grad_norm": 4.9317498207092285,
      "learning_rate": 3.666017612181174e-05,
      "loss": 0.47,
      "step": 21543
    },
    {
      "epoch": 6.8026523523839595,
      "grad_norm": 6.42963981628418,
      "learning_rate": 3.665929902115567e-05,
      "loss": 0.6034,
      "step": 21544
    },
    {
      "epoch": 6.802968108620146,
      "grad_norm": 6.206191062927246,
      "learning_rate": 3.6658421920499596e-05,
      "loss": 0.64,
      "step": 21545
    },
    {
      "epoch": 6.803283864856331,
      "grad_norm": 5.4618401527404785,
      "learning_rate": 3.665754481984353e-05,
      "loss": 0.5741,
      "step": 21546
    },
    {
      "epoch": 6.803599621092516,
      "grad_norm": 5.070817470550537,
      "learning_rate": 3.665666771918745e-05,
      "loss": 0.5527,
      "step": 21547
    },
    {
      "epoch": 6.803915377328702,
      "grad_norm": 4.874840259552002,
      "learning_rate": 3.6655790618531385e-05,
      "loss": 0.5806,
      "step": 21548
    },
    {
      "epoch": 6.804231133564888,
      "grad_norm": 5.209120750427246,
      "learning_rate": 3.665491351787531e-05,
      "loss": 0.5748,
      "step": 21549
    },
    {
      "epoch": 6.804546889801074,
      "grad_norm": 5.348264694213867,
      "learning_rate": 3.665403641721924e-05,
      "loss": 0.6133,
      "step": 21550
    },
    {
      "epoch": 6.804862646037259,
      "grad_norm": 4.672905921936035,
      "learning_rate": 3.6653159316563166e-05,
      "loss": 0.5928,
      "step": 21551
    },
    {
      "epoch": 6.8051784022734445,
      "grad_norm": 4.848036289215088,
      "learning_rate": 3.66522822159071e-05,
      "loss": 0.6619,
      "step": 21552
    },
    {
      "epoch": 6.805494158509631,
      "grad_norm": 4.966507911682129,
      "learning_rate": 3.665140511525103e-05,
      "loss": 0.5754,
      "step": 21553
    },
    {
      "epoch": 6.805809914745816,
      "grad_norm": 5.588746070861816,
      "learning_rate": 3.665052801459496e-05,
      "loss": 0.6035,
      "step": 21554
    },
    {
      "epoch": 6.806125670982002,
      "grad_norm": 5.161073684692383,
      "learning_rate": 3.6649650913938886e-05,
      "loss": 0.5547,
      "step": 21555
    },
    {
      "epoch": 6.8064414272181875,
      "grad_norm": 4.438267707824707,
      "learning_rate": 3.664877381328282e-05,
      "loss": 0.4609,
      "step": 21556
    },
    {
      "epoch": 6.806757183454373,
      "grad_norm": 5.0426344871521,
      "learning_rate": 3.664789671262674e-05,
      "loss": 0.6756,
      "step": 21557
    },
    {
      "epoch": 6.807072939690559,
      "grad_norm": 4.853519916534424,
      "learning_rate": 3.6647019611970674e-05,
      "loss": 0.5912,
      "step": 21558
    },
    {
      "epoch": 6.807388695926744,
      "grad_norm": 4.907668113708496,
      "learning_rate": 3.66461425113146e-05,
      "loss": 0.4597,
      "step": 21559
    },
    {
      "epoch": 6.80770445216293,
      "grad_norm": 5.618820667266846,
      "learning_rate": 3.664526541065853e-05,
      "loss": 0.6281,
      "step": 21560
    },
    {
      "epoch": 6.808020208399116,
      "grad_norm": 5.740634918212891,
      "learning_rate": 3.6644388310002456e-05,
      "loss": 0.6079,
      "step": 21561
    },
    {
      "epoch": 6.808335964635302,
      "grad_norm": 4.402573108673096,
      "learning_rate": 3.664351120934639e-05,
      "loss": 0.5698,
      "step": 21562
    },
    {
      "epoch": 6.808651720871487,
      "grad_norm": 4.607041835784912,
      "learning_rate": 3.664263410869031e-05,
      "loss": 0.5545,
      "step": 21563
    },
    {
      "epoch": 6.808967477107673,
      "grad_norm": 3.8333609104156494,
      "learning_rate": 3.6641757008034244e-05,
      "loss": 0.5577,
      "step": 21564
    },
    {
      "epoch": 6.809283233343859,
      "grad_norm": 5.792619228363037,
      "learning_rate": 3.6640879907378176e-05,
      "loss": 0.6102,
      "step": 21565
    },
    {
      "epoch": 6.809598989580044,
      "grad_norm": 4.314835548400879,
      "learning_rate": 3.66400028067221e-05,
      "loss": 0.5447,
      "step": 21566
    },
    {
      "epoch": 6.80991474581623,
      "grad_norm": 3.908470869064331,
      "learning_rate": 3.663912570606603e-05,
      "loss": 0.5344,
      "step": 21567
    },
    {
      "epoch": 6.810230502052415,
      "grad_norm": 4.342972278594971,
      "learning_rate": 3.663824860540996e-05,
      "loss": 0.5729,
      "step": 21568
    },
    {
      "epoch": 6.810546258288602,
      "grad_norm": 5.1860480308532715,
      "learning_rate": 3.663737150475389e-05,
      "loss": 0.6006,
      "step": 21569
    },
    {
      "epoch": 6.810862014524787,
      "grad_norm": 4.317319869995117,
      "learning_rate": 3.6636494404097814e-05,
      "loss": 0.5783,
      "step": 21570
    },
    {
      "epoch": 6.811177770760972,
      "grad_norm": 4.625183582305908,
      "learning_rate": 3.6635617303441746e-05,
      "loss": 0.5805,
      "step": 21571
    },
    {
      "epoch": 6.811493526997158,
      "grad_norm": 4.384869575500488,
      "learning_rate": 3.663474020278567e-05,
      "loss": 0.5608,
      "step": 21572
    },
    {
      "epoch": 6.811809283233344,
      "grad_norm": 5.2997050285339355,
      "learning_rate": 3.66338631021296e-05,
      "loss": 0.6277,
      "step": 21573
    },
    {
      "epoch": 6.81212503946953,
      "grad_norm": 3.9860570430755615,
      "learning_rate": 3.663298600147353e-05,
      "loss": 0.4285,
      "step": 21574
    },
    {
      "epoch": 6.812440795705715,
      "grad_norm": 4.55446195602417,
      "learning_rate": 3.663210890081746e-05,
      "loss": 0.5593,
      "step": 21575
    },
    {
      "epoch": 6.8127565519419,
      "grad_norm": 4.081612586975098,
      "learning_rate": 3.6631231800161384e-05,
      "loss": 0.5369,
      "step": 21576
    },
    {
      "epoch": 6.813072308178087,
      "grad_norm": 4.20340633392334,
      "learning_rate": 3.6630354699505315e-05,
      "loss": 0.4731,
      "step": 21577
    },
    {
      "epoch": 6.813388064414272,
      "grad_norm": 5.530608654022217,
      "learning_rate": 3.662947759884924e-05,
      "loss": 0.57,
      "step": 21578
    },
    {
      "epoch": 6.813703820650458,
      "grad_norm": 6.871424674987793,
      "learning_rate": 3.662860049819317e-05,
      "loss": 0.5257,
      "step": 21579
    },
    {
      "epoch": 6.814019576886643,
      "grad_norm": 4.713202476501465,
      "learning_rate": 3.6627723397537104e-05,
      "loss": 0.6372,
      "step": 21580
    },
    {
      "epoch": 6.814335333122829,
      "grad_norm": 5.216977119445801,
      "learning_rate": 3.6626846296881035e-05,
      "loss": 0.5739,
      "step": 21581
    },
    {
      "epoch": 6.814651089359015,
      "grad_norm": 4.437427997589111,
      "learning_rate": 3.662596919622496e-05,
      "loss": 0.524,
      "step": 21582
    },
    {
      "epoch": 6.8149668455952,
      "grad_norm": 4.011932373046875,
      "learning_rate": 3.662509209556889e-05,
      "loss": 0.4939,
      "step": 21583
    },
    {
      "epoch": 6.815282601831386,
      "grad_norm": 5.595005035400391,
      "learning_rate": 3.662421499491282e-05,
      "loss": 0.615,
      "step": 21584
    },
    {
      "epoch": 6.815598358067572,
      "grad_norm": 5.221508502960205,
      "learning_rate": 3.662333789425675e-05,
      "loss": 0.521,
      "step": 21585
    },
    {
      "epoch": 6.815914114303758,
      "grad_norm": 6.1662702560424805,
      "learning_rate": 3.662246079360068e-05,
      "loss": 0.5493,
      "step": 21586
    },
    {
      "epoch": 6.816229870539943,
      "grad_norm": 4.720197677612305,
      "learning_rate": 3.6621583692944605e-05,
      "loss": 0.5183,
      "step": 21587
    },
    {
      "epoch": 6.816545626776129,
      "grad_norm": 4.932775497436523,
      "learning_rate": 3.662070659228854e-05,
      "loss": 0.5294,
      "step": 21588
    },
    {
      "epoch": 6.816861383012315,
      "grad_norm": 5.844990253448486,
      "learning_rate": 3.661982949163246e-05,
      "loss": 0.6189,
      "step": 21589
    },
    {
      "epoch": 6.8171771392485,
      "grad_norm": 4.329136848449707,
      "learning_rate": 3.661895239097639e-05,
      "loss": 0.4579,
      "step": 21590
    },
    {
      "epoch": 6.817492895484686,
      "grad_norm": 4.577882766723633,
      "learning_rate": 3.661807529032032e-05,
      "loss": 0.659,
      "step": 21591
    },
    {
      "epoch": 6.817808651720871,
      "grad_norm": 5.430144309997559,
      "learning_rate": 3.661719818966425e-05,
      "loss": 0.5047,
      "step": 21592
    },
    {
      "epoch": 6.8181244079570575,
      "grad_norm": 5.040229797363281,
      "learning_rate": 3.6616321089008175e-05,
      "loss": 0.487,
      "step": 21593
    },
    {
      "epoch": 6.818440164193243,
      "grad_norm": 5.901639461517334,
      "learning_rate": 3.6615443988352107e-05,
      "loss": 0.6507,
      "step": 21594
    },
    {
      "epoch": 6.818755920429428,
      "grad_norm": 5.942808628082275,
      "learning_rate": 3.661456688769603e-05,
      "loss": 0.5346,
      "step": 21595
    },
    {
      "epoch": 6.819071676665614,
      "grad_norm": 4.749720573425293,
      "learning_rate": 3.661368978703996e-05,
      "loss": 0.5214,
      "step": 21596
    },
    {
      "epoch": 6.8193874329018,
      "grad_norm": 6.044249057769775,
      "learning_rate": 3.661281268638389e-05,
      "loss": 0.5896,
      "step": 21597
    },
    {
      "epoch": 6.819703189137986,
      "grad_norm": 6.805647850036621,
      "learning_rate": 3.661193558572782e-05,
      "loss": 0.668,
      "step": 21598
    },
    {
      "epoch": 6.820018945374171,
      "grad_norm": 4.433023452758789,
      "learning_rate": 3.6611058485071745e-05,
      "loss": 0.5151,
      "step": 21599
    },
    {
      "epoch": 6.820334701610356,
      "grad_norm": 5.265380382537842,
      "learning_rate": 3.6610181384415676e-05,
      "loss": 0.5339,
      "step": 21600
    },
    {
      "epoch": 6.820334701610356,
      "eval_loss": 0.8632935881614685,
      "eval_runtime": 4.6066,
      "eval_samples_per_second": 540.528,
      "eval_steps_per_second": 4.559,
      "step": 21600
    },
    {
      "epoch": 6.820650457846543,
      "grad_norm": 4.532557964324951,
      "learning_rate": 3.66093042837596e-05,
      "loss": 0.4377,
      "step": 21601
    },
    {
      "epoch": 6.820966214082728,
      "grad_norm": 4.511231422424316,
      "learning_rate": 3.660842718310353e-05,
      "loss": 0.6079,
      "step": 21602
    },
    {
      "epoch": 6.821281970318914,
      "grad_norm": 5.578283786773682,
      "learning_rate": 3.6607550082447465e-05,
      "loss": 0.6972,
      "step": 21603
    },
    {
      "epoch": 6.821597726555099,
      "grad_norm": 5.17068338394165,
      "learning_rate": 3.660667298179139e-05,
      "loss": 0.592,
      "step": 21604
    },
    {
      "epoch": 6.8219134827912855,
      "grad_norm": 5.850121974945068,
      "learning_rate": 3.660579588113532e-05,
      "loss": 0.6381,
      "step": 21605
    },
    {
      "epoch": 6.822229239027471,
      "grad_norm": 5.156266212463379,
      "learning_rate": 3.6604918780479246e-05,
      "loss": 0.5085,
      "step": 21606
    },
    {
      "epoch": 6.822544995263656,
      "grad_norm": 6.086967945098877,
      "learning_rate": 3.660404167982318e-05,
      "loss": 0.5358,
      "step": 21607
    },
    {
      "epoch": 6.822860751499842,
      "grad_norm": 4.875889778137207,
      "learning_rate": 3.660316457916711e-05,
      "loss": 0.6024,
      "step": 21608
    },
    {
      "epoch": 6.823176507736028,
      "grad_norm": 5.030893802642822,
      "learning_rate": 3.660228747851104e-05,
      "loss": 0.6078,
      "step": 21609
    },
    {
      "epoch": 6.823492263972214,
      "grad_norm": 5.0184125900268555,
      "learning_rate": 3.6601410377854966e-05,
      "loss": 0.4815,
      "step": 21610
    },
    {
      "epoch": 6.823808020208399,
      "grad_norm": 4.2707061767578125,
      "learning_rate": 3.66005332771989e-05,
      "loss": 0.5109,
      "step": 21611
    },
    {
      "epoch": 6.824123776444585,
      "grad_norm": 5.295492649078369,
      "learning_rate": 3.659965617654282e-05,
      "loss": 0.6411,
      "step": 21612
    },
    {
      "epoch": 6.8244395326807705,
      "grad_norm": 6.070176601409912,
      "learning_rate": 3.6598779075886754e-05,
      "loss": 0.5807,
      "step": 21613
    },
    {
      "epoch": 6.824755288916956,
      "grad_norm": 5.537731647491455,
      "learning_rate": 3.659790197523068e-05,
      "loss": 0.6496,
      "step": 21614
    },
    {
      "epoch": 6.825071045153142,
      "grad_norm": 4.857115745544434,
      "learning_rate": 3.659702487457461e-05,
      "loss": 0.4683,
      "step": 21615
    },
    {
      "epoch": 6.825386801389327,
      "grad_norm": 4.804615020751953,
      "learning_rate": 3.6596147773918536e-05,
      "loss": 0.4762,
      "step": 21616
    },
    {
      "epoch": 6.8257025576255135,
      "grad_norm": 4.781126499176025,
      "learning_rate": 3.659527067326247e-05,
      "loss": 0.5928,
      "step": 21617
    },
    {
      "epoch": 6.826018313861699,
      "grad_norm": 4.905951976776123,
      "learning_rate": 3.659439357260639e-05,
      "loss": 0.5332,
      "step": 21618
    },
    {
      "epoch": 6.826334070097884,
      "grad_norm": 5.248310565948486,
      "learning_rate": 3.6593516471950324e-05,
      "loss": 0.5686,
      "step": 21619
    },
    {
      "epoch": 6.82664982633407,
      "grad_norm": 5.375070095062256,
      "learning_rate": 3.659263937129425e-05,
      "loss": 0.5969,
      "step": 21620
    },
    {
      "epoch": 6.8269655825702555,
      "grad_norm": 5.667910099029541,
      "learning_rate": 3.659176227063818e-05,
      "loss": 0.6423,
      "step": 21621
    },
    {
      "epoch": 6.827281338806442,
      "grad_norm": 6.387533187866211,
      "learning_rate": 3.6590885169982106e-05,
      "loss": 0.6215,
      "step": 21622
    },
    {
      "epoch": 6.827597095042627,
      "grad_norm": 4.4768171310424805,
      "learning_rate": 3.659000806932604e-05,
      "loss": 0.577,
      "step": 21623
    },
    {
      "epoch": 6.827912851278812,
      "grad_norm": 6.584080696105957,
      "learning_rate": 3.658913096866997e-05,
      "loss": 0.5613,
      "step": 21624
    },
    {
      "epoch": 6.8282286075149985,
      "grad_norm": 4.809306621551514,
      "learning_rate": 3.6588253868013894e-05,
      "loss": 0.4998,
      "step": 21625
    },
    {
      "epoch": 6.828544363751184,
      "grad_norm": 5.541568279266357,
      "learning_rate": 3.6587376767357826e-05,
      "loss": 0.4431,
      "step": 21626
    },
    {
      "epoch": 6.82886011998737,
      "grad_norm": 6.726812839508057,
      "learning_rate": 3.658649966670175e-05,
      "loss": 0.6154,
      "step": 21627
    },
    {
      "epoch": 6.829175876223555,
      "grad_norm": 5.3802056312561035,
      "learning_rate": 3.658562256604568e-05,
      "loss": 0.5166,
      "step": 21628
    },
    {
      "epoch": 6.8294916324597414,
      "grad_norm": 4.8144450187683105,
      "learning_rate": 3.658474546538961e-05,
      "loss": 0.5749,
      "step": 21629
    },
    {
      "epoch": 6.829807388695927,
      "grad_norm": 4.914788246154785,
      "learning_rate": 3.658386836473354e-05,
      "loss": 0.5283,
      "step": 21630
    },
    {
      "epoch": 6.830123144932112,
      "grad_norm": 5.1182451248168945,
      "learning_rate": 3.6582991264077464e-05,
      "loss": 0.6816,
      "step": 21631
    },
    {
      "epoch": 6.830438901168298,
      "grad_norm": 6.285730361938477,
      "learning_rate": 3.6582114163421395e-05,
      "loss": 0.6748,
      "step": 21632
    },
    {
      "epoch": 6.8307546574044835,
      "grad_norm": 5.092596054077148,
      "learning_rate": 3.658123706276532e-05,
      "loss": 0.6125,
      "step": 21633
    },
    {
      "epoch": 6.83107041364067,
      "grad_norm": 6.191707611083984,
      "learning_rate": 3.658035996210925e-05,
      "loss": 0.6075,
      "step": 21634
    },
    {
      "epoch": 6.831386169876855,
      "grad_norm": 5.212038516998291,
      "learning_rate": 3.6579482861453184e-05,
      "loss": 0.5382,
      "step": 21635
    },
    {
      "epoch": 6.831701926113041,
      "grad_norm": 4.3253679275512695,
      "learning_rate": 3.6578605760797115e-05,
      "loss": 0.5517,
      "step": 21636
    },
    {
      "epoch": 6.8320176823492265,
      "grad_norm": 5.130084037780762,
      "learning_rate": 3.657772866014104e-05,
      "loss": 0.5936,
      "step": 21637
    },
    {
      "epoch": 6.832333438585412,
      "grad_norm": 5.397085189819336,
      "learning_rate": 3.657685155948497e-05,
      "loss": 0.5313,
      "step": 21638
    },
    {
      "epoch": 6.832649194821598,
      "grad_norm": 4.329056262969971,
      "learning_rate": 3.65759744588289e-05,
      "loss": 0.6307,
      "step": 21639
    },
    {
      "epoch": 6.832964951057783,
      "grad_norm": 4.733333587646484,
      "learning_rate": 3.657509735817283e-05,
      "loss": 0.5196,
      "step": 21640
    },
    {
      "epoch": 6.833280707293969,
      "grad_norm": 5.014721393585205,
      "learning_rate": 3.657422025751675e-05,
      "loss": 0.5015,
      "step": 21641
    },
    {
      "epoch": 6.833596463530155,
      "grad_norm": 4.811516284942627,
      "learning_rate": 3.6573343156860685e-05,
      "loss": 0.5488,
      "step": 21642
    },
    {
      "epoch": 6.83391221976634,
      "grad_norm": 5.14726448059082,
      "learning_rate": 3.657246605620461e-05,
      "loss": 0.5721,
      "step": 21643
    },
    {
      "epoch": 6.834227976002526,
      "grad_norm": 5.577309608459473,
      "learning_rate": 3.657158895554854e-05,
      "loss": 0.6217,
      "step": 21644
    },
    {
      "epoch": 6.8345437322387115,
      "grad_norm": 5.016112804412842,
      "learning_rate": 3.6570711854892466e-05,
      "loss": 0.577,
      "step": 21645
    },
    {
      "epoch": 6.834859488474898,
      "grad_norm": 5.588457107543945,
      "learning_rate": 3.65698347542364e-05,
      "loss": 0.5813,
      "step": 21646
    },
    {
      "epoch": 6.835175244711083,
      "grad_norm": 5.032774448394775,
      "learning_rate": 3.656895765358033e-05,
      "loss": 0.4879,
      "step": 21647
    },
    {
      "epoch": 6.835491000947268,
      "grad_norm": 7.579799652099609,
      "learning_rate": 3.6568080552924255e-05,
      "loss": 0.7041,
      "step": 21648
    },
    {
      "epoch": 6.835806757183454,
      "grad_norm": 4.494833946228027,
      "learning_rate": 3.6567203452268186e-05,
      "loss": 0.5362,
      "step": 21649
    },
    {
      "epoch": 6.83612251341964,
      "grad_norm": 5.514266490936279,
      "learning_rate": 3.656632635161211e-05,
      "loss": 0.5996,
      "step": 21650
    },
    {
      "epoch": 6.836438269655826,
      "grad_norm": 4.816939353942871,
      "learning_rate": 3.656544925095604e-05,
      "loss": 0.5591,
      "step": 21651
    },
    {
      "epoch": 6.836754025892011,
      "grad_norm": 4.887154579162598,
      "learning_rate": 3.656457215029997e-05,
      "loss": 0.5709,
      "step": 21652
    },
    {
      "epoch": 6.837069782128197,
      "grad_norm": 4.758841514587402,
      "learning_rate": 3.65636950496439e-05,
      "loss": 0.6454,
      "step": 21653
    },
    {
      "epoch": 6.837385538364383,
      "grad_norm": 5.198714256286621,
      "learning_rate": 3.6562817948987825e-05,
      "loss": 0.5085,
      "step": 21654
    },
    {
      "epoch": 6.837701294600569,
      "grad_norm": 5.652411937713623,
      "learning_rate": 3.6561940848331756e-05,
      "loss": 0.6096,
      "step": 21655
    },
    {
      "epoch": 6.838017050836754,
      "grad_norm": 4.919768810272217,
      "learning_rate": 3.656106374767568e-05,
      "loss": 0.5549,
      "step": 21656
    },
    {
      "epoch": 6.8383328070729394,
      "grad_norm": 4.852294921875,
      "learning_rate": 3.656018664701961e-05,
      "loss": 0.5006,
      "step": 21657
    },
    {
      "epoch": 6.838648563309126,
      "grad_norm": 3.6564643383026123,
      "learning_rate": 3.655930954636354e-05,
      "loss": 0.4303,
      "step": 21658
    },
    {
      "epoch": 6.838964319545311,
      "grad_norm": 3.9987075328826904,
      "learning_rate": 3.655843244570747e-05,
      "loss": 0.4885,
      "step": 21659
    },
    {
      "epoch": 6.839280075781497,
      "grad_norm": 5.129641056060791,
      "learning_rate": 3.6557555345051394e-05,
      "loss": 0.5689,
      "step": 21660
    },
    {
      "epoch": 6.839595832017682,
      "grad_norm": 4.820631980895996,
      "learning_rate": 3.6556678244395326e-05,
      "loss": 0.7023,
      "step": 21661
    },
    {
      "epoch": 6.839911588253868,
      "grad_norm": 4.277462959289551,
      "learning_rate": 3.655580114373926e-05,
      "loss": 0.4979,
      "step": 21662
    },
    {
      "epoch": 6.840227344490054,
      "grad_norm": 4.989129066467285,
      "learning_rate": 3.655492404308319e-05,
      "loss": 0.5789,
      "step": 21663
    },
    {
      "epoch": 6.840543100726239,
      "grad_norm": 6.5839128494262695,
      "learning_rate": 3.6554046942427114e-05,
      "loss": 0.5859,
      "step": 21664
    },
    {
      "epoch": 6.840858856962425,
      "grad_norm": 4.665473461151123,
      "learning_rate": 3.6553169841771046e-05,
      "loss": 0.514,
      "step": 21665
    },
    {
      "epoch": 6.841174613198611,
      "grad_norm": 5.042199611663818,
      "learning_rate": 3.655229274111497e-05,
      "loss": 0.4756,
      "step": 21666
    },
    {
      "epoch": 6.841490369434796,
      "grad_norm": 3.6978507041931152,
      "learning_rate": 3.65514156404589e-05,
      "loss": 0.451,
      "step": 21667
    },
    {
      "epoch": 6.841806125670982,
      "grad_norm": 3.538043975830078,
      "learning_rate": 3.6550538539802834e-05,
      "loss": 0.4442,
      "step": 21668
    },
    {
      "epoch": 6.842121881907167,
      "grad_norm": 5.319309234619141,
      "learning_rate": 3.654966143914676e-05,
      "loss": 0.5236,
      "step": 21669
    },
    {
      "epoch": 6.842437638143354,
      "grad_norm": 4.191411972045898,
      "learning_rate": 3.654878433849069e-05,
      "loss": 0.6017,
      "step": 21670
    },
    {
      "epoch": 6.842753394379539,
      "grad_norm": 4.264934062957764,
      "learning_rate": 3.6547907237834616e-05,
      "loss": 0.5748,
      "step": 21671
    },
    {
      "epoch": 6.843069150615724,
      "grad_norm": 4.772828578948975,
      "learning_rate": 3.654703013717855e-05,
      "loss": 0.5866,
      "step": 21672
    },
    {
      "epoch": 6.84338490685191,
      "grad_norm": 10.083090782165527,
      "learning_rate": 3.654615303652247e-05,
      "loss": 0.5441,
      "step": 21673
    },
    {
      "epoch": 6.843700663088096,
      "grad_norm": 5.930562496185303,
      "learning_rate": 3.6545275935866404e-05,
      "loss": 0.5873,
      "step": 21674
    },
    {
      "epoch": 6.844016419324282,
      "grad_norm": 5.187102317810059,
      "learning_rate": 3.654439883521033e-05,
      "loss": 0.5066,
      "step": 21675
    },
    {
      "epoch": 6.844332175560467,
      "grad_norm": 5.496939659118652,
      "learning_rate": 3.654352173455426e-05,
      "loss": 0.5599,
      "step": 21676
    },
    {
      "epoch": 6.844647931796653,
      "grad_norm": 5.511193752288818,
      "learning_rate": 3.6542644633898185e-05,
      "loss": 0.7094,
      "step": 21677
    },
    {
      "epoch": 6.844963688032839,
      "grad_norm": 5.070833683013916,
      "learning_rate": 3.654176753324212e-05,
      "loss": 0.5539,
      "step": 21678
    },
    {
      "epoch": 6.845279444269025,
      "grad_norm": 5.217886447906494,
      "learning_rate": 3.654089043258604e-05,
      "loss": 0.558,
      "step": 21679
    },
    {
      "epoch": 6.84559520050521,
      "grad_norm": 5.1595282554626465,
      "learning_rate": 3.6540013331929974e-05,
      "loss": 0.5549,
      "step": 21680
    },
    {
      "epoch": 6.845910956741395,
      "grad_norm": 5.075878143310547,
      "learning_rate": 3.65391362312739e-05,
      "loss": 0.5808,
      "step": 21681
    },
    {
      "epoch": 6.846226712977582,
      "grad_norm": 5.520247936248779,
      "learning_rate": 3.653825913061783e-05,
      "loss": 0.564,
      "step": 21682
    },
    {
      "epoch": 6.846542469213767,
      "grad_norm": 4.6376495361328125,
      "learning_rate": 3.653738202996176e-05,
      "loss": 0.5277,
      "step": 21683
    },
    {
      "epoch": 6.846858225449953,
      "grad_norm": 5.5538330078125,
      "learning_rate": 3.653650492930569e-05,
      "loss": 0.6211,
      "step": 21684
    },
    {
      "epoch": 6.847173981686138,
      "grad_norm": 5.25825309753418,
      "learning_rate": 3.653562782864962e-05,
      "loss": 0.6092,
      "step": 21685
    },
    {
      "epoch": 6.847489737922324,
      "grad_norm": 5.389729022979736,
      "learning_rate": 3.6534750727993543e-05,
      "loss": 0.6937,
      "step": 21686
    },
    {
      "epoch": 6.84780549415851,
      "grad_norm": 4.751579761505127,
      "learning_rate": 3.6533873627337475e-05,
      "loss": 0.5411,
      "step": 21687
    },
    {
      "epoch": 6.848121250394695,
      "grad_norm": 5.143507480621338,
      "learning_rate": 3.65329965266814e-05,
      "loss": 0.6482,
      "step": 21688
    },
    {
      "epoch": 6.848437006630881,
      "grad_norm": 5.382545471191406,
      "learning_rate": 3.653211942602533e-05,
      "loss": 0.5976,
      "step": 21689
    },
    {
      "epoch": 6.848752762867067,
      "grad_norm": 4.250127792358398,
      "learning_rate": 3.6531242325369263e-05,
      "loss": 0.5731,
      "step": 21690
    },
    {
      "epoch": 6.849068519103252,
      "grad_norm": 5.294835090637207,
      "learning_rate": 3.6530365224713195e-05,
      "loss": 0.6408,
      "step": 21691
    },
    {
      "epoch": 6.849384275339438,
      "grad_norm": 4.615155220031738,
      "learning_rate": 3.652948812405712e-05,
      "loss": 0.4928,
      "step": 21692
    },
    {
      "epoch": 6.849700031575623,
      "grad_norm": 5.20127534866333,
      "learning_rate": 3.652861102340105e-05,
      "loss": 0.5038,
      "step": 21693
    },
    {
      "epoch": 6.8500157878118095,
      "grad_norm": 4.980279445648193,
      "learning_rate": 3.652773392274498e-05,
      "loss": 0.6575,
      "step": 21694
    },
    {
      "epoch": 6.850331544047995,
      "grad_norm": 5.000631809234619,
      "learning_rate": 3.652685682208891e-05,
      "loss": 0.5934,
      "step": 21695
    },
    {
      "epoch": 6.85064730028418,
      "grad_norm": 4.545907020568848,
      "learning_rate": 3.652597972143283e-05,
      "loss": 0.5075,
      "step": 21696
    },
    {
      "epoch": 6.850963056520366,
      "grad_norm": 4.787064075469971,
      "learning_rate": 3.6525102620776765e-05,
      "loss": 0.6205,
      "step": 21697
    },
    {
      "epoch": 6.851278812756552,
      "grad_norm": 5.127420902252197,
      "learning_rate": 3.652422552012069e-05,
      "loss": 0.6011,
      "step": 21698
    },
    {
      "epoch": 6.851594568992738,
      "grad_norm": 5.562854290008545,
      "learning_rate": 3.652334841946462e-05,
      "loss": 0.5846,
      "step": 21699
    },
    {
      "epoch": 6.851910325228923,
      "grad_norm": 3.687087059020996,
      "learning_rate": 3.6522471318808546e-05,
      "loss": 0.4571,
      "step": 21700
    },
    {
      "epoch": 6.852226081465109,
      "grad_norm": 6.73410177230835,
      "learning_rate": 3.652159421815248e-05,
      "loss": 0.6429,
      "step": 21701
    },
    {
      "epoch": 6.8525418377012945,
      "grad_norm": 4.990422248840332,
      "learning_rate": 3.65207171174964e-05,
      "loss": 0.5114,
      "step": 21702
    },
    {
      "epoch": 6.852857593937481,
      "grad_norm": 9.39694595336914,
      "learning_rate": 3.6519840016840335e-05,
      "loss": 0.5564,
      "step": 21703
    },
    {
      "epoch": 6.853173350173666,
      "grad_norm": 4.211479187011719,
      "learning_rate": 3.651896291618426e-05,
      "loss": 0.557,
      "step": 21704
    },
    {
      "epoch": 6.853489106409851,
      "grad_norm": 3.939023494720459,
      "learning_rate": 3.651808581552819e-05,
      "loss": 0.5026,
      "step": 21705
    },
    {
      "epoch": 6.8538048626460375,
      "grad_norm": 5.449832439422607,
      "learning_rate": 3.651720871487212e-05,
      "loss": 0.6229,
      "step": 21706
    },
    {
      "epoch": 6.854120618882223,
      "grad_norm": 3.964031457901001,
      "learning_rate": 3.651633161421605e-05,
      "loss": 0.4972,
      "step": 21707
    },
    {
      "epoch": 6.854436375118409,
      "grad_norm": 5.3206467628479,
      "learning_rate": 3.651545451355998e-05,
      "loss": 0.504,
      "step": 21708
    },
    {
      "epoch": 6.854752131354594,
      "grad_norm": 4.789079189300537,
      "learning_rate": 3.6514577412903904e-05,
      "loss": 0.6589,
      "step": 21709
    },
    {
      "epoch": 6.85506788759078,
      "grad_norm": 4.886221408843994,
      "learning_rate": 3.6513700312247836e-05,
      "loss": 0.588,
      "step": 21710
    },
    {
      "epoch": 6.855383643826966,
      "grad_norm": 5.67382287979126,
      "learning_rate": 3.651282321159176e-05,
      "loss": 0.6299,
      "step": 21711
    },
    {
      "epoch": 6.855699400063151,
      "grad_norm": 4.646214008331299,
      "learning_rate": 3.651194611093569e-05,
      "loss": 0.6149,
      "step": 21712
    },
    {
      "epoch": 6.856015156299337,
      "grad_norm": 5.329489707946777,
      "learning_rate": 3.651106901027962e-05,
      "loss": 0.4264,
      "step": 21713
    },
    {
      "epoch": 6.8563309125355225,
      "grad_norm": 5.164224147796631,
      "learning_rate": 3.651019190962355e-05,
      "loss": 0.5521,
      "step": 21714
    },
    {
      "epoch": 6.856646668771708,
      "grad_norm": 4.8443827629089355,
      "learning_rate": 3.6509314808967474e-05,
      "loss": 0.5239,
      "step": 21715
    },
    {
      "epoch": 6.856962425007894,
      "grad_norm": 6.838675022125244,
      "learning_rate": 3.6508437708311406e-05,
      "loss": 0.584,
      "step": 21716
    },
    {
      "epoch": 6.857278181244079,
      "grad_norm": 6.560579299926758,
      "learning_rate": 3.650756060765534e-05,
      "loss": 0.6031,
      "step": 21717
    },
    {
      "epoch": 6.8575939374802655,
      "grad_norm": 5.233857154846191,
      "learning_rate": 3.650668350699927e-05,
      "loss": 0.5812,
      "step": 21718
    },
    {
      "epoch": 6.857909693716451,
      "grad_norm": 5.319204330444336,
      "learning_rate": 3.6505806406343194e-05,
      "loss": 0.5833,
      "step": 21719
    },
    {
      "epoch": 6.858225449952637,
      "grad_norm": 6.597043991088867,
      "learning_rate": 3.6504929305687126e-05,
      "loss": 0.5693,
      "step": 21720
    },
    {
      "epoch": 6.858541206188822,
      "grad_norm": 5.594126224517822,
      "learning_rate": 3.650405220503105e-05,
      "loss": 0.6002,
      "step": 21721
    },
    {
      "epoch": 6.8588569624250075,
      "grad_norm": 5.537253379821777,
      "learning_rate": 3.650317510437498e-05,
      "loss": 0.5623,
      "step": 21722
    },
    {
      "epoch": 6.859172718661194,
      "grad_norm": 5.734169960021973,
      "learning_rate": 3.650229800371891e-05,
      "loss": 0.6004,
      "step": 21723
    },
    {
      "epoch": 6.859488474897379,
      "grad_norm": 6.063299179077148,
      "learning_rate": 3.650142090306284e-05,
      "loss": 0.5144,
      "step": 21724
    },
    {
      "epoch": 6.859804231133565,
      "grad_norm": 5.47530460357666,
      "learning_rate": 3.6500543802406764e-05,
      "loss": 0.5378,
      "step": 21725
    },
    {
      "epoch": 6.8601199873697505,
      "grad_norm": 6.177582740783691,
      "learning_rate": 3.6499666701750696e-05,
      "loss": 0.6231,
      "step": 21726
    },
    {
      "epoch": 6.860435743605937,
      "grad_norm": 5.65767240524292,
      "learning_rate": 3.649878960109463e-05,
      "loss": 0.5714,
      "step": 21727
    },
    {
      "epoch": 6.860751499842122,
      "grad_norm": 5.397266387939453,
      "learning_rate": 3.649791250043855e-05,
      "loss": 0.5812,
      "step": 21728
    },
    {
      "epoch": 6.861067256078307,
      "grad_norm": 6.934589385986328,
      "learning_rate": 3.6497035399782484e-05,
      "loss": 0.6529,
      "step": 21729
    },
    {
      "epoch": 6.861383012314493,
      "grad_norm": 4.076581001281738,
      "learning_rate": 3.649615829912641e-05,
      "loss": 0.5393,
      "step": 21730
    },
    {
      "epoch": 6.861698768550679,
      "grad_norm": 5.362076759338379,
      "learning_rate": 3.649528119847034e-05,
      "loss": 0.6374,
      "step": 21731
    },
    {
      "epoch": 6.862014524786865,
      "grad_norm": 6.0318474769592285,
      "learning_rate": 3.6494404097814265e-05,
      "loss": 0.6536,
      "step": 21732
    },
    {
      "epoch": 6.86233028102305,
      "grad_norm": 5.201866626739502,
      "learning_rate": 3.64935269971582e-05,
      "loss": 0.521,
      "step": 21733
    },
    {
      "epoch": 6.8626460372592355,
      "grad_norm": 4.149767875671387,
      "learning_rate": 3.649264989650212e-05,
      "loss": 0.5287,
      "step": 21734
    },
    {
      "epoch": 6.862961793495422,
      "grad_norm": 5.217894077301025,
      "learning_rate": 3.6491772795846054e-05,
      "loss": 0.6489,
      "step": 21735
    },
    {
      "epoch": 6.863277549731607,
      "grad_norm": 4.500101566314697,
      "learning_rate": 3.649089569518998e-05,
      "loss": 0.5248,
      "step": 21736
    },
    {
      "epoch": 6.863593305967793,
      "grad_norm": 4.221009731292725,
      "learning_rate": 3.649001859453391e-05,
      "loss": 0.491,
      "step": 21737
    },
    {
      "epoch": 6.8639090622039785,
      "grad_norm": 5.618288516998291,
      "learning_rate": 3.6489141493877835e-05,
      "loss": 0.4861,
      "step": 21738
    },
    {
      "epoch": 6.864224818440164,
      "grad_norm": 4.953693389892578,
      "learning_rate": 3.648826439322177e-05,
      "loss": 0.5705,
      "step": 21739
    },
    {
      "epoch": 6.86454057467635,
      "grad_norm": 4.522959232330322,
      "learning_rate": 3.648738729256569e-05,
      "loss": 0.5852,
      "step": 21740
    },
    {
      "epoch": 6.864856330912535,
      "grad_norm": 4.879659652709961,
      "learning_rate": 3.648651019190962e-05,
      "loss": 0.5929,
      "step": 21741
    },
    {
      "epoch": 6.865172087148721,
      "grad_norm": 4.536688804626465,
      "learning_rate": 3.648563309125355e-05,
      "loss": 0.5346,
      "step": 21742
    },
    {
      "epoch": 6.865487843384907,
      "grad_norm": 6.875521183013916,
      "learning_rate": 3.648475599059748e-05,
      "loss": 0.5933,
      "step": 21743
    },
    {
      "epoch": 6.865803599621093,
      "grad_norm": 5.674757480621338,
      "learning_rate": 3.648387888994141e-05,
      "loss": 0.7213,
      "step": 21744
    },
    {
      "epoch": 6.866119355857278,
      "grad_norm": 4.921602725982666,
      "learning_rate": 3.648300178928534e-05,
      "loss": 0.5945,
      "step": 21745
    },
    {
      "epoch": 6.8664351120934635,
      "grad_norm": 4.758934020996094,
      "learning_rate": 3.648212468862927e-05,
      "loss": 0.6174,
      "step": 21746
    },
    {
      "epoch": 6.86675086832965,
      "grad_norm": 4.850128173828125,
      "learning_rate": 3.64812475879732e-05,
      "loss": 0.6053,
      "step": 21747
    },
    {
      "epoch": 6.867066624565835,
      "grad_norm": 5.235041618347168,
      "learning_rate": 3.6480370487317125e-05,
      "loss": 0.5742,
      "step": 21748
    },
    {
      "epoch": 6.867382380802021,
      "grad_norm": 5.630012512207031,
      "learning_rate": 3.6479493386661057e-05,
      "loss": 0.5106,
      "step": 21749
    },
    {
      "epoch": 6.867698137038206,
      "grad_norm": 4.726609230041504,
      "learning_rate": 3.647861628600499e-05,
      "loss": 0.5664,
      "step": 21750
    },
    {
      "epoch": 6.868013893274393,
      "grad_norm": 4.751597881317139,
      "learning_rate": 3.647773918534891e-05,
      "loss": 0.6068,
      "step": 21751
    },
    {
      "epoch": 6.868329649510578,
      "grad_norm": 6.546719551086426,
      "learning_rate": 3.6476862084692845e-05,
      "loss": 0.5725,
      "step": 21752
    },
    {
      "epoch": 6.868645405746763,
      "grad_norm": 5.5273118019104,
      "learning_rate": 3.647598498403677e-05,
      "loss": 0.5914,
      "step": 21753
    },
    {
      "epoch": 6.868961161982949,
      "grad_norm": 5.045465469360352,
      "learning_rate": 3.64751078833807e-05,
      "loss": 0.5231,
      "step": 21754
    },
    {
      "epoch": 6.869276918219135,
      "grad_norm": 5.299431324005127,
      "learning_rate": 3.6474230782724626e-05,
      "loss": 0.5041,
      "step": 21755
    },
    {
      "epoch": 6.869592674455321,
      "grad_norm": 5.302980899810791,
      "learning_rate": 3.647335368206856e-05,
      "loss": 0.6339,
      "step": 21756
    },
    {
      "epoch": 6.869908430691506,
      "grad_norm": 4.835616588592529,
      "learning_rate": 3.647247658141248e-05,
      "loss": 0.5406,
      "step": 21757
    },
    {
      "epoch": 6.870224186927691,
      "grad_norm": 4.827569484710693,
      "learning_rate": 3.6471599480756415e-05,
      "loss": 0.5967,
      "step": 21758
    },
    {
      "epoch": 6.870539943163878,
      "grad_norm": 4.268425941467285,
      "learning_rate": 3.647072238010034e-05,
      "loss": 0.4498,
      "step": 21759
    },
    {
      "epoch": 6.870855699400063,
      "grad_norm": 7.052299976348877,
      "learning_rate": 3.646984527944427e-05,
      "loss": 0.6106,
      "step": 21760
    },
    {
      "epoch": 6.871171455636249,
      "grad_norm": 5.808652400970459,
      "learning_rate": 3.6468968178788196e-05,
      "loss": 0.6071,
      "step": 21761
    },
    {
      "epoch": 6.871487211872434,
      "grad_norm": 5.1342926025390625,
      "learning_rate": 3.646809107813213e-05,
      "loss": 0.6317,
      "step": 21762
    },
    {
      "epoch": 6.87180296810862,
      "grad_norm": 4.618950843811035,
      "learning_rate": 3.646721397747605e-05,
      "loss": 0.5945,
      "step": 21763
    },
    {
      "epoch": 6.872118724344806,
      "grad_norm": 4.828756809234619,
      "learning_rate": 3.6466336876819984e-05,
      "loss": 0.5951,
      "step": 21764
    },
    {
      "epoch": 6.872434480580991,
      "grad_norm": 5.561873912811279,
      "learning_rate": 3.6465459776163916e-05,
      "loss": 0.6163,
      "step": 21765
    },
    {
      "epoch": 6.872750236817177,
      "grad_norm": 5.209070205688477,
      "learning_rate": 3.646458267550784e-05,
      "loss": 0.6694,
      "step": 21766
    },
    {
      "epoch": 6.873065993053363,
      "grad_norm": 6.1649298667907715,
      "learning_rate": 3.646370557485177e-05,
      "loss": 0.6773,
      "step": 21767
    },
    {
      "epoch": 6.873381749289549,
      "grad_norm": 5.3105854988098145,
      "learning_rate": 3.64628284741957e-05,
      "loss": 0.5581,
      "step": 21768
    },
    {
      "epoch": 6.873697505525734,
      "grad_norm": 4.318063259124756,
      "learning_rate": 3.646195137353963e-05,
      "loss": 0.5261,
      "step": 21769
    },
    {
      "epoch": 6.87401326176192,
      "grad_norm": 5.852632999420166,
      "learning_rate": 3.6461074272883554e-05,
      "loss": 0.558,
      "step": 21770
    },
    {
      "epoch": 6.874329017998106,
      "grad_norm": 4.975449085235596,
      "learning_rate": 3.646019717222749e-05,
      "loss": 0.5854,
      "step": 21771
    },
    {
      "epoch": 6.874644774234291,
      "grad_norm": 6.295208930969238,
      "learning_rate": 3.645932007157142e-05,
      "loss": 0.5495,
      "step": 21772
    },
    {
      "epoch": 6.874960530470477,
      "grad_norm": 5.951234340667725,
      "learning_rate": 3.645844297091535e-05,
      "loss": 0.5984,
      "step": 21773
    },
    {
      "epoch": 6.875276286706662,
      "grad_norm": 5.106196403503418,
      "learning_rate": 3.6457565870259274e-05,
      "loss": 0.5705,
      "step": 21774
    },
    {
      "epoch": 6.8755920429428485,
      "grad_norm": 6.097422122955322,
      "learning_rate": 3.6456688769603206e-05,
      "loss": 0.6634,
      "step": 21775
    },
    {
      "epoch": 6.875907799179034,
      "grad_norm": 5.069736480712891,
      "learning_rate": 3.645581166894713e-05,
      "loss": 0.7381,
      "step": 21776
    },
    {
      "epoch": 6.876223555415219,
      "grad_norm": 4.514712333679199,
      "learning_rate": 3.645493456829106e-05,
      "loss": 0.558,
      "step": 21777
    },
    {
      "epoch": 6.876539311651405,
      "grad_norm": 4.708218574523926,
      "learning_rate": 3.645405746763499e-05,
      "loss": 0.6208,
      "step": 21778
    },
    {
      "epoch": 6.876855067887591,
      "grad_norm": 5.729693412780762,
      "learning_rate": 3.645318036697892e-05,
      "loss": 0.5424,
      "step": 21779
    },
    {
      "epoch": 6.877170824123777,
      "grad_norm": 6.838181972503662,
      "learning_rate": 3.6452303266322844e-05,
      "loss": 0.5034,
      "step": 21780
    },
    {
      "epoch": 6.877486580359962,
      "grad_norm": 4.741436958312988,
      "learning_rate": 3.6451426165666775e-05,
      "loss": 0.5616,
      "step": 21781
    },
    {
      "epoch": 6.877802336596147,
      "grad_norm": 4.937977313995361,
      "learning_rate": 3.64505490650107e-05,
      "loss": 0.5775,
      "step": 21782
    },
    {
      "epoch": 6.8781180928323336,
      "grad_norm": 4.140738487243652,
      "learning_rate": 3.644967196435463e-05,
      "loss": 0.5114,
      "step": 21783
    },
    {
      "epoch": 6.878433849068519,
      "grad_norm": 5.593808174133301,
      "learning_rate": 3.644879486369856e-05,
      "loss": 0.5602,
      "step": 21784
    },
    {
      "epoch": 6.878749605304705,
      "grad_norm": 4.7749457359313965,
      "learning_rate": 3.644791776304249e-05,
      "loss": 0.5339,
      "step": 21785
    },
    {
      "epoch": 6.87906536154089,
      "grad_norm": 5.71980094909668,
      "learning_rate": 3.6447040662386414e-05,
      "loss": 0.4709,
      "step": 21786
    },
    {
      "epoch": 6.879381117777076,
      "grad_norm": 5.180665493011475,
      "learning_rate": 3.6446163561730345e-05,
      "loss": 0.6428,
      "step": 21787
    },
    {
      "epoch": 6.879696874013262,
      "grad_norm": 4.601988792419434,
      "learning_rate": 3.644528646107428e-05,
      "loss": 0.6258,
      "step": 21788
    },
    {
      "epoch": 6.880012630249447,
      "grad_norm": 3.8552401065826416,
      "learning_rate": 3.64444093604182e-05,
      "loss": 0.5122,
      "step": 21789
    },
    {
      "epoch": 6.880328386485633,
      "grad_norm": 5.448275089263916,
      "learning_rate": 3.6443532259762134e-05,
      "loss": 0.6743,
      "step": 21790
    },
    {
      "epoch": 6.880644142721819,
      "grad_norm": 6.470672130584717,
      "learning_rate": 3.644265515910606e-05,
      "loss": 0.5187,
      "step": 21791
    },
    {
      "epoch": 6.880959898958005,
      "grad_norm": 6.312719345092773,
      "learning_rate": 3.644177805844999e-05,
      "loss": 0.6748,
      "step": 21792
    },
    {
      "epoch": 6.88127565519419,
      "grad_norm": 5.739932537078857,
      "learning_rate": 3.6440900957793915e-05,
      "loss": 0.6946,
      "step": 21793
    },
    {
      "epoch": 6.881591411430376,
      "grad_norm": 5.475170612335205,
      "learning_rate": 3.644002385713785e-05,
      "loss": 0.6037,
      "step": 21794
    },
    {
      "epoch": 6.8819071676665615,
      "grad_norm": 5.037830352783203,
      "learning_rate": 3.643914675648177e-05,
      "loss": 0.5666,
      "step": 21795
    },
    {
      "epoch": 6.882222923902747,
      "grad_norm": 5.045548439025879,
      "learning_rate": 3.64382696558257e-05,
      "loss": 0.5515,
      "step": 21796
    },
    {
      "epoch": 6.882538680138933,
      "grad_norm": 5.20890474319458,
      "learning_rate": 3.643739255516963e-05,
      "loss": 0.6016,
      "step": 21797
    },
    {
      "epoch": 6.882854436375118,
      "grad_norm": 6.609052658081055,
      "learning_rate": 3.643651545451356e-05,
      "loss": 0.605,
      "step": 21798
    },
    {
      "epoch": 6.8831701926113045,
      "grad_norm": 4.7860260009765625,
      "learning_rate": 3.643563835385749e-05,
      "loss": 0.4805,
      "step": 21799
    },
    {
      "epoch": 6.88348594884749,
      "grad_norm": 4.382340431213379,
      "learning_rate": 3.643476125320142e-05,
      "loss": 0.4757,
      "step": 21800
    },
    {
      "epoch": 6.88348594884749,
      "eval_loss": 0.8592398762702942,
      "eval_runtime": 4.5657,
      "eval_samples_per_second": 545.374,
      "eval_steps_per_second": 4.6,
      "step": 21800
    },
    {
      "epoch": 6.883801705083675,
      "grad_norm": 4.5847487449646,
      "learning_rate": 3.643388415254535e-05,
      "loss": 0.599,
      "step": 21801
    },
    {
      "epoch": 6.884117461319861,
      "grad_norm": 4.912693977355957,
      "learning_rate": 3.643300705188928e-05,
      "loss": 0.5221,
      "step": 21802
    },
    {
      "epoch": 6.8844332175560465,
      "grad_norm": 4.36607027053833,
      "learning_rate": 3.6432129951233205e-05,
      "loss": 0.4463,
      "step": 21803
    },
    {
      "epoch": 6.884748973792233,
      "grad_norm": 5.442755699157715,
      "learning_rate": 3.6431252850577136e-05,
      "loss": 0.6055,
      "step": 21804
    },
    {
      "epoch": 6.885064730028418,
      "grad_norm": 5.473182201385498,
      "learning_rate": 3.643037574992106e-05,
      "loss": 0.6009,
      "step": 21805
    },
    {
      "epoch": 6.885380486264603,
      "grad_norm": 5.078603744506836,
      "learning_rate": 3.642949864926499e-05,
      "loss": 0.5523,
      "step": 21806
    },
    {
      "epoch": 6.8856962425007895,
      "grad_norm": 5.566743850708008,
      "learning_rate": 3.642862154860892e-05,
      "loss": 0.5028,
      "step": 21807
    },
    {
      "epoch": 6.886011998736975,
      "grad_norm": 5.735572814941406,
      "learning_rate": 3.642774444795285e-05,
      "loss": 0.6712,
      "step": 21808
    },
    {
      "epoch": 6.886327754973161,
      "grad_norm": 4.9774017333984375,
      "learning_rate": 3.642686734729678e-05,
      "loss": 0.4932,
      "step": 21809
    },
    {
      "epoch": 6.886643511209346,
      "grad_norm": 5.31821870803833,
      "learning_rate": 3.6425990246640706e-05,
      "loss": 0.5096,
      "step": 21810
    },
    {
      "epoch": 6.8869592674455316,
      "grad_norm": 5.005016326904297,
      "learning_rate": 3.642511314598464e-05,
      "loss": 0.5005,
      "step": 21811
    },
    {
      "epoch": 6.887275023681718,
      "grad_norm": 5.738287448883057,
      "learning_rate": 3.642423604532856e-05,
      "loss": 0.5452,
      "step": 21812
    },
    {
      "epoch": 6.887590779917903,
      "grad_norm": 4.519545078277588,
      "learning_rate": 3.6423358944672494e-05,
      "loss": 0.5933,
      "step": 21813
    },
    {
      "epoch": 6.887906536154089,
      "grad_norm": 5.637600898742676,
      "learning_rate": 3.642248184401642e-05,
      "loss": 0.5494,
      "step": 21814
    },
    {
      "epoch": 6.8882222923902745,
      "grad_norm": 5.26143217086792,
      "learning_rate": 3.642160474336035e-05,
      "loss": 0.4996,
      "step": 21815
    },
    {
      "epoch": 6.888538048626461,
      "grad_norm": 5.99439811706543,
      "learning_rate": 3.6420727642704276e-05,
      "loss": 0.692,
      "step": 21816
    },
    {
      "epoch": 6.888853804862646,
      "grad_norm": 4.447551250457764,
      "learning_rate": 3.641985054204821e-05,
      "loss": 0.5377,
      "step": 21817
    },
    {
      "epoch": 6.889169561098832,
      "grad_norm": 4.975083827972412,
      "learning_rate": 3.641897344139213e-05,
      "loss": 0.5488,
      "step": 21818
    },
    {
      "epoch": 6.8894853173350175,
      "grad_norm": 5.721864700317383,
      "learning_rate": 3.6418096340736064e-05,
      "loss": 0.6626,
      "step": 21819
    },
    {
      "epoch": 6.889801073571203,
      "grad_norm": 5.994251728057861,
      "learning_rate": 3.641721924007999e-05,
      "loss": 0.5365,
      "step": 21820
    },
    {
      "epoch": 6.890116829807389,
      "grad_norm": 3.986940860748291,
      "learning_rate": 3.641634213942392e-05,
      "loss": 0.5969,
      "step": 21821
    },
    {
      "epoch": 6.890432586043574,
      "grad_norm": 6.278656959533691,
      "learning_rate": 3.6415465038767846e-05,
      "loss": 0.6752,
      "step": 21822
    },
    {
      "epoch": 6.89074834227976,
      "grad_norm": 5.759134292602539,
      "learning_rate": 3.641458793811178e-05,
      "loss": 0.5901,
      "step": 21823
    },
    {
      "epoch": 6.891064098515946,
      "grad_norm": 6.018701553344727,
      "learning_rate": 3.64137108374557e-05,
      "loss": 0.6796,
      "step": 21824
    },
    {
      "epoch": 6.891379854752131,
      "grad_norm": 5.303421974182129,
      "learning_rate": 3.6412833736799634e-05,
      "loss": 0.5694,
      "step": 21825
    },
    {
      "epoch": 6.891695610988317,
      "grad_norm": 5.434041976928711,
      "learning_rate": 3.6411956636143566e-05,
      "loss": 0.6528,
      "step": 21826
    },
    {
      "epoch": 6.8920113672245025,
      "grad_norm": 6.555203914642334,
      "learning_rate": 3.64110795354875e-05,
      "loss": 0.6036,
      "step": 21827
    },
    {
      "epoch": 6.892327123460689,
      "grad_norm": 4.548198699951172,
      "learning_rate": 3.641020243483142e-05,
      "loss": 0.6063,
      "step": 21828
    },
    {
      "epoch": 6.892642879696874,
      "grad_norm": 4.239800453186035,
      "learning_rate": 3.6409325334175354e-05,
      "loss": 0.6748,
      "step": 21829
    },
    {
      "epoch": 6.892958635933059,
      "grad_norm": 5.041695594787598,
      "learning_rate": 3.6408448233519286e-05,
      "loss": 0.6141,
      "step": 21830
    },
    {
      "epoch": 6.893274392169245,
      "grad_norm": 5.25331449508667,
      "learning_rate": 3.640757113286321e-05,
      "loss": 0.7294,
      "step": 21831
    },
    {
      "epoch": 6.893590148405431,
      "grad_norm": 4.463510036468506,
      "learning_rate": 3.640669403220714e-05,
      "loss": 0.5045,
      "step": 21832
    },
    {
      "epoch": 6.893905904641617,
      "grad_norm": 5.528938293457031,
      "learning_rate": 3.640581693155107e-05,
      "loss": 0.5512,
      "step": 21833
    },
    {
      "epoch": 6.894221660877802,
      "grad_norm": 4.925065994262695,
      "learning_rate": 3.6404939830895e-05,
      "loss": 0.5386,
      "step": 21834
    },
    {
      "epoch": 6.894537417113988,
      "grad_norm": 3.781599760055542,
      "learning_rate": 3.6404062730238924e-05,
      "loss": 0.4239,
      "step": 21835
    },
    {
      "epoch": 6.894853173350174,
      "grad_norm": 4.502293109893799,
      "learning_rate": 3.6403185629582855e-05,
      "loss": 0.5948,
      "step": 21836
    },
    {
      "epoch": 6.895168929586359,
      "grad_norm": 4.3464789390563965,
      "learning_rate": 3.640230852892678e-05,
      "loss": 0.5341,
      "step": 21837
    },
    {
      "epoch": 6.895484685822545,
      "grad_norm": 5.059599876403809,
      "learning_rate": 3.640143142827071e-05,
      "loss": 0.6071,
      "step": 21838
    },
    {
      "epoch": 6.89580044205873,
      "grad_norm": 4.6074628829956055,
      "learning_rate": 3.640055432761464e-05,
      "loss": 0.5615,
      "step": 21839
    },
    {
      "epoch": 6.896116198294917,
      "grad_norm": 3.642857074737549,
      "learning_rate": 3.639967722695857e-05,
      "loss": 0.4787,
      "step": 21840
    },
    {
      "epoch": 6.896431954531102,
      "grad_norm": 4.725266933441162,
      "learning_rate": 3.6398800126302493e-05,
      "loss": 0.6636,
      "step": 21841
    },
    {
      "epoch": 6.896747710767288,
      "grad_norm": 4.338469982147217,
      "learning_rate": 3.6397923025646425e-05,
      "loss": 0.5949,
      "step": 21842
    },
    {
      "epoch": 6.897063467003473,
      "grad_norm": 5.492008686065674,
      "learning_rate": 3.639704592499035e-05,
      "loss": 0.5741,
      "step": 21843
    },
    {
      "epoch": 6.897379223239659,
      "grad_norm": 4.46917200088501,
      "learning_rate": 3.639616882433428e-05,
      "loss": 0.5421,
      "step": 21844
    },
    {
      "epoch": 6.897694979475845,
      "grad_norm": 4.156988620758057,
      "learning_rate": 3.639529172367821e-05,
      "loss": 0.4525,
      "step": 21845
    },
    {
      "epoch": 6.89801073571203,
      "grad_norm": 4.845644474029541,
      "learning_rate": 3.639441462302214e-05,
      "loss": 0.7173,
      "step": 21846
    },
    {
      "epoch": 6.898326491948216,
      "grad_norm": 3.930727958679199,
      "learning_rate": 3.639353752236607e-05,
      "loss": 0.5935,
      "step": 21847
    },
    {
      "epoch": 6.898642248184402,
      "grad_norm": 4.6002655029296875,
      "learning_rate": 3.6392660421709995e-05,
      "loss": 0.6506,
      "step": 21848
    },
    {
      "epoch": 6.898958004420587,
      "grad_norm": 5.405893802642822,
      "learning_rate": 3.6391783321053927e-05,
      "loss": 0.6392,
      "step": 21849
    },
    {
      "epoch": 6.899273760656773,
      "grad_norm": 4.75107479095459,
      "learning_rate": 3.639090622039785e-05,
      "loss": 0.5263,
      "step": 21850
    },
    {
      "epoch": 6.899589516892958,
      "grad_norm": 5.498317241668701,
      "learning_rate": 3.639002911974178e-05,
      "loss": 0.6343,
      "step": 21851
    },
    {
      "epoch": 6.899905273129145,
      "grad_norm": 5.1029744148254395,
      "learning_rate": 3.638915201908571e-05,
      "loss": 0.5349,
      "step": 21852
    },
    {
      "epoch": 6.90022102936533,
      "grad_norm": 4.064058303833008,
      "learning_rate": 3.6388274918429647e-05,
      "loss": 0.5396,
      "step": 21853
    },
    {
      "epoch": 6.900536785601515,
      "grad_norm": 5.548067569732666,
      "learning_rate": 3.638739781777357e-05,
      "loss": 0.6,
      "step": 21854
    },
    {
      "epoch": 6.900852541837701,
      "grad_norm": 5.731992244720459,
      "learning_rate": 3.63865207171175e-05,
      "loss": 0.6491,
      "step": 21855
    },
    {
      "epoch": 6.901168298073887,
      "grad_norm": 4.553394317626953,
      "learning_rate": 3.638564361646143e-05,
      "loss": 0.5366,
      "step": 21856
    },
    {
      "epoch": 6.901484054310073,
      "grad_norm": 4.776022434234619,
      "learning_rate": 3.638476651580536e-05,
      "loss": 0.5941,
      "step": 21857
    },
    {
      "epoch": 6.901799810546258,
      "grad_norm": 5.68233060836792,
      "learning_rate": 3.6383889415149285e-05,
      "loss": 0.7505,
      "step": 21858
    },
    {
      "epoch": 6.902115566782444,
      "grad_norm": 4.362742900848389,
      "learning_rate": 3.6383012314493216e-05,
      "loss": 0.4917,
      "step": 21859
    },
    {
      "epoch": 6.90243132301863,
      "grad_norm": 5.417472839355469,
      "learning_rate": 3.638213521383714e-05,
      "loss": 0.6177,
      "step": 21860
    },
    {
      "epoch": 6.902747079254815,
      "grad_norm": 4.858581066131592,
      "learning_rate": 3.638125811318107e-05,
      "loss": 0.4515,
      "step": 21861
    },
    {
      "epoch": 6.903062835491001,
      "grad_norm": 3.4614670276641846,
      "learning_rate": 3.6380381012525e-05,
      "loss": 0.4288,
      "step": 21862
    },
    {
      "epoch": 6.903378591727186,
      "grad_norm": 4.588822841644287,
      "learning_rate": 3.637950391186893e-05,
      "loss": 0.5555,
      "step": 21863
    },
    {
      "epoch": 6.903694347963373,
      "grad_norm": 4.23451042175293,
      "learning_rate": 3.6378626811212854e-05,
      "loss": 0.5539,
      "step": 21864
    },
    {
      "epoch": 6.904010104199558,
      "grad_norm": 4.050284385681152,
      "learning_rate": 3.6377749710556786e-05,
      "loss": 0.5355,
      "step": 21865
    },
    {
      "epoch": 6.904325860435744,
      "grad_norm": 5.0850510597229,
      "learning_rate": 3.637687260990071e-05,
      "loss": 0.5699,
      "step": 21866
    },
    {
      "epoch": 6.904641616671929,
      "grad_norm": 5.377569198608398,
      "learning_rate": 3.637599550924464e-05,
      "loss": 0.5661,
      "step": 21867
    },
    {
      "epoch": 6.904957372908115,
      "grad_norm": 5.148188591003418,
      "learning_rate": 3.6375118408588574e-05,
      "loss": 0.5429,
      "step": 21868
    },
    {
      "epoch": 6.905273129144301,
      "grad_norm": 4.979276180267334,
      "learning_rate": 3.63742413079325e-05,
      "loss": 0.5168,
      "step": 21869
    },
    {
      "epoch": 6.905588885380486,
      "grad_norm": 6.6657514572143555,
      "learning_rate": 3.637336420727643e-05,
      "loss": 0.5487,
      "step": 21870
    },
    {
      "epoch": 6.905904641616672,
      "grad_norm": 5.076171875,
      "learning_rate": 3.6372487106620356e-05,
      "loss": 0.5332,
      "step": 21871
    },
    {
      "epoch": 6.906220397852858,
      "grad_norm": 4.257256507873535,
      "learning_rate": 3.637161000596429e-05,
      "loss": 0.5461,
      "step": 21872
    },
    {
      "epoch": 6.906536154089043,
      "grad_norm": 6.292064666748047,
      "learning_rate": 3.637073290530821e-05,
      "loss": 0.658,
      "step": 21873
    },
    {
      "epoch": 6.906851910325229,
      "grad_norm": 7.14929723739624,
      "learning_rate": 3.6369855804652144e-05,
      "loss": 0.6322,
      "step": 21874
    },
    {
      "epoch": 6.907167666561414,
      "grad_norm": 5.245968341827393,
      "learning_rate": 3.636897870399607e-05,
      "loss": 0.6422,
      "step": 21875
    },
    {
      "epoch": 6.9074834227976005,
      "grad_norm": 4.092431545257568,
      "learning_rate": 3.636810160334e-05,
      "loss": 0.5193,
      "step": 21876
    },
    {
      "epoch": 6.907799179033786,
      "grad_norm": 5.32529354095459,
      "learning_rate": 3.6367224502683926e-05,
      "loss": 0.6092,
      "step": 21877
    },
    {
      "epoch": 6.908114935269971,
      "grad_norm": 4.450199604034424,
      "learning_rate": 3.636634740202786e-05,
      "loss": 0.5777,
      "step": 21878
    },
    {
      "epoch": 6.908430691506157,
      "grad_norm": 4.980392932891846,
      "learning_rate": 3.636547030137178e-05,
      "loss": 0.4591,
      "step": 21879
    },
    {
      "epoch": 6.908746447742343,
      "grad_norm": 4.641604900360107,
      "learning_rate": 3.636459320071572e-05,
      "loss": 0.5831,
      "step": 21880
    },
    {
      "epoch": 6.909062203978529,
      "grad_norm": 5.866537094116211,
      "learning_rate": 3.6363716100059646e-05,
      "loss": 0.6602,
      "step": 21881
    },
    {
      "epoch": 6.909377960214714,
      "grad_norm": 5.6747307777404785,
      "learning_rate": 3.636283899940358e-05,
      "loss": 0.641,
      "step": 21882
    },
    {
      "epoch": 6.9096937164509,
      "grad_norm": 4.937130451202393,
      "learning_rate": 3.63619618987475e-05,
      "loss": 0.5453,
      "step": 21883
    },
    {
      "epoch": 6.9100094726870855,
      "grad_norm": 6.857262134552002,
      "learning_rate": 3.6361084798091434e-05,
      "loss": 0.6128,
      "step": 21884
    },
    {
      "epoch": 6.910325228923272,
      "grad_norm": 5.9212822914123535,
      "learning_rate": 3.636020769743536e-05,
      "loss": 0.5091,
      "step": 21885
    },
    {
      "epoch": 6.910640985159457,
      "grad_norm": 4.9696455001831055,
      "learning_rate": 3.635933059677929e-05,
      "loss": 0.5754,
      "step": 21886
    },
    {
      "epoch": 6.910956741395642,
      "grad_norm": 6.00742769241333,
      "learning_rate": 3.6358453496123215e-05,
      "loss": 0.6303,
      "step": 21887
    },
    {
      "epoch": 6.9112724976318285,
      "grad_norm": 5.256567478179932,
      "learning_rate": 3.635757639546715e-05,
      "loss": 0.5392,
      "step": 21888
    },
    {
      "epoch": 6.911588253868014,
      "grad_norm": 5.825445175170898,
      "learning_rate": 3.635669929481107e-05,
      "loss": 0.6311,
      "step": 21889
    },
    {
      "epoch": 6.9119040101042,
      "grad_norm": 4.968553066253662,
      "learning_rate": 3.6355822194155004e-05,
      "loss": 0.5075,
      "step": 21890
    },
    {
      "epoch": 6.912219766340385,
      "grad_norm": 4.40940523147583,
      "learning_rate": 3.6354945093498935e-05,
      "loss": 0.4817,
      "step": 21891
    },
    {
      "epoch": 6.912535522576571,
      "grad_norm": 4.373195171356201,
      "learning_rate": 3.635406799284286e-05,
      "loss": 0.4856,
      "step": 21892
    },
    {
      "epoch": 6.912851278812757,
      "grad_norm": 4.199926376342773,
      "learning_rate": 3.635319089218679e-05,
      "loss": 0.6976,
      "step": 21893
    },
    {
      "epoch": 6.913167035048942,
      "grad_norm": 5.464949131011963,
      "learning_rate": 3.635231379153072e-05,
      "loss": 0.6738,
      "step": 21894
    },
    {
      "epoch": 6.913482791285128,
      "grad_norm": 4.330084800720215,
      "learning_rate": 3.635143669087465e-05,
      "loss": 0.5807,
      "step": 21895
    },
    {
      "epoch": 6.9137985475213135,
      "grad_norm": 4.616462707519531,
      "learning_rate": 3.635055959021857e-05,
      "loss": 0.5174,
      "step": 21896
    },
    {
      "epoch": 6.914114303757499,
      "grad_norm": 4.693736553192139,
      "learning_rate": 3.6349682489562505e-05,
      "loss": 0.5564,
      "step": 21897
    },
    {
      "epoch": 6.914430059993685,
      "grad_norm": 5.10884428024292,
      "learning_rate": 3.634880538890643e-05,
      "loss": 0.6604,
      "step": 21898
    },
    {
      "epoch": 6.91474581622987,
      "grad_norm": 9.296088218688965,
      "learning_rate": 3.634792828825036e-05,
      "loss": 0.6648,
      "step": 21899
    },
    {
      "epoch": 6.9150615724660565,
      "grad_norm": 4.741483688354492,
      "learning_rate": 3.6347051187594287e-05,
      "loss": 0.6462,
      "step": 21900
    },
    {
      "epoch": 6.915377328702242,
      "grad_norm": 3.986636161804199,
      "learning_rate": 3.634617408693822e-05,
      "loss": 0.5588,
      "step": 21901
    },
    {
      "epoch": 6.915693084938427,
      "grad_norm": 4.434941291809082,
      "learning_rate": 3.634529698628214e-05,
      "loss": 0.6077,
      "step": 21902
    },
    {
      "epoch": 6.916008841174613,
      "grad_norm": 4.862707614898682,
      "learning_rate": 3.6344419885626075e-05,
      "loss": 0.6626,
      "step": 21903
    },
    {
      "epoch": 6.9163245974107985,
      "grad_norm": 4.26932954788208,
      "learning_rate": 3.634354278497e-05,
      "loss": 0.6534,
      "step": 21904
    },
    {
      "epoch": 6.916640353646985,
      "grad_norm": 4.416993141174316,
      "learning_rate": 3.634266568431393e-05,
      "loss": 0.4911,
      "step": 21905
    },
    {
      "epoch": 6.91695610988317,
      "grad_norm": 5.447985649108887,
      "learning_rate": 3.634178858365786e-05,
      "loss": 0.7142,
      "step": 21906
    },
    {
      "epoch": 6.917271866119356,
      "grad_norm": 4.647870063781738,
      "learning_rate": 3.634091148300179e-05,
      "loss": 0.59,
      "step": 21907
    },
    {
      "epoch": 6.9175876223555415,
      "grad_norm": 4.442900657653809,
      "learning_rate": 3.634003438234572e-05,
      "loss": 0.5379,
      "step": 21908
    },
    {
      "epoch": 6.917903378591728,
      "grad_norm": 4.470151901245117,
      "learning_rate": 3.633915728168965e-05,
      "loss": 0.5069,
      "step": 21909
    },
    {
      "epoch": 6.918219134827913,
      "grad_norm": 4.8759284019470215,
      "learning_rate": 3.6338280181033576e-05,
      "loss": 0.4894,
      "step": 21910
    },
    {
      "epoch": 6.918534891064098,
      "grad_norm": 4.454814434051514,
      "learning_rate": 3.633740308037751e-05,
      "loss": 0.5057,
      "step": 21911
    },
    {
      "epoch": 6.918850647300284,
      "grad_norm": 5.286508083343506,
      "learning_rate": 3.633652597972144e-05,
      "loss": 0.7049,
      "step": 21912
    },
    {
      "epoch": 6.91916640353647,
      "grad_norm": 4.665655612945557,
      "learning_rate": 3.6335648879065365e-05,
      "loss": 0.6646,
      "step": 21913
    },
    {
      "epoch": 6.919482159772656,
      "grad_norm": 4.933284282684326,
      "learning_rate": 3.6334771778409296e-05,
      "loss": 0.604,
      "step": 21914
    },
    {
      "epoch": 6.919797916008841,
      "grad_norm": 4.319310665130615,
      "learning_rate": 3.633389467775322e-05,
      "loss": 0.5568,
      "step": 21915
    },
    {
      "epoch": 6.9201136722450265,
      "grad_norm": 4.449550151824951,
      "learning_rate": 3.633301757709715e-05,
      "loss": 0.5577,
      "step": 21916
    },
    {
      "epoch": 6.920429428481213,
      "grad_norm": 3.961425304412842,
      "learning_rate": 3.633214047644108e-05,
      "loss": 0.6016,
      "step": 21917
    },
    {
      "epoch": 6.920745184717398,
      "grad_norm": 5.136697292327881,
      "learning_rate": 3.633126337578501e-05,
      "loss": 0.6138,
      "step": 21918
    },
    {
      "epoch": 6.921060940953584,
      "grad_norm": 4.2330451011657715,
      "learning_rate": 3.6330386275128934e-05,
      "loss": 0.4783,
      "step": 21919
    },
    {
      "epoch": 6.9213766971897694,
      "grad_norm": 4.474733829498291,
      "learning_rate": 3.6329509174472866e-05,
      "loss": 0.651,
      "step": 21920
    },
    {
      "epoch": 6.921692453425955,
      "grad_norm": 4.738875389099121,
      "learning_rate": 3.632863207381679e-05,
      "loss": 0.6142,
      "step": 21921
    },
    {
      "epoch": 6.922008209662141,
      "grad_norm": 5.471715927124023,
      "learning_rate": 3.632775497316072e-05,
      "loss": 0.6167,
      "step": 21922
    },
    {
      "epoch": 6.922323965898326,
      "grad_norm": 5.625910758972168,
      "learning_rate": 3.632687787250465e-05,
      "loss": 0.5655,
      "step": 21923
    },
    {
      "epoch": 6.922639722134512,
      "grad_norm": 4.819077968597412,
      "learning_rate": 3.632600077184858e-05,
      "loss": 0.5585,
      "step": 21924
    },
    {
      "epoch": 6.922955478370698,
      "grad_norm": 6.841338634490967,
      "learning_rate": 3.6325123671192504e-05,
      "loss": 0.5802,
      "step": 21925
    },
    {
      "epoch": 6.923271234606884,
      "grad_norm": 4.649377346038818,
      "learning_rate": 3.6324246570536436e-05,
      "loss": 0.558,
      "step": 21926
    },
    {
      "epoch": 6.923586990843069,
      "grad_norm": 6.376389503479004,
      "learning_rate": 3.632336946988036e-05,
      "loss": 0.5562,
      "step": 21927
    },
    {
      "epoch": 6.9239027470792545,
      "grad_norm": 4.429003715515137,
      "learning_rate": 3.632249236922429e-05,
      "loss": 0.5664,
      "step": 21928
    },
    {
      "epoch": 6.924218503315441,
      "grad_norm": 5.3257737159729,
      "learning_rate": 3.6321615268568224e-05,
      "loss": 0.6854,
      "step": 21929
    },
    {
      "epoch": 6.924534259551626,
      "grad_norm": 4.860147953033447,
      "learning_rate": 3.632073816791215e-05,
      "loss": 0.4793,
      "step": 21930
    },
    {
      "epoch": 6.924850015787812,
      "grad_norm": 5.650786399841309,
      "learning_rate": 3.631986106725608e-05,
      "loss": 0.5379,
      "step": 21931
    },
    {
      "epoch": 6.925165772023997,
      "grad_norm": 5.278338432312012,
      "learning_rate": 3.6318983966600005e-05,
      "loss": 0.5705,
      "step": 21932
    },
    {
      "epoch": 6.925481528260184,
      "grad_norm": 5.003152847290039,
      "learning_rate": 3.631810686594394e-05,
      "loss": 0.6178,
      "step": 21933
    },
    {
      "epoch": 6.925797284496369,
      "grad_norm": 5.206007480621338,
      "learning_rate": 3.631722976528786e-05,
      "loss": 0.4665,
      "step": 21934
    },
    {
      "epoch": 6.926113040732554,
      "grad_norm": 5.320176601409912,
      "learning_rate": 3.63163526646318e-05,
      "loss": 0.4961,
      "step": 21935
    },
    {
      "epoch": 6.92642879696874,
      "grad_norm": 4.952146530151367,
      "learning_rate": 3.6315475563975725e-05,
      "loss": 0.6688,
      "step": 21936
    },
    {
      "epoch": 6.926744553204926,
      "grad_norm": 4.903549671173096,
      "learning_rate": 3.631459846331966e-05,
      "loss": 0.6182,
      "step": 21937
    },
    {
      "epoch": 6.927060309441112,
      "grad_norm": 4.918035507202148,
      "learning_rate": 3.631372136266358e-05,
      "loss": 0.4812,
      "step": 21938
    },
    {
      "epoch": 6.927376065677297,
      "grad_norm": 4.230896472930908,
      "learning_rate": 3.6312844262007514e-05,
      "loss": 0.56,
      "step": 21939
    },
    {
      "epoch": 6.927691821913482,
      "grad_norm": 5.1922478675842285,
      "learning_rate": 3.631196716135144e-05,
      "loss": 0.5761,
      "step": 21940
    },
    {
      "epoch": 6.928007578149669,
      "grad_norm": 4.4462761878967285,
      "learning_rate": 3.631109006069537e-05,
      "loss": 0.5037,
      "step": 21941
    },
    {
      "epoch": 6.928323334385854,
      "grad_norm": 6.301415920257568,
      "learning_rate": 3.6310212960039295e-05,
      "loss": 0.6342,
      "step": 21942
    },
    {
      "epoch": 6.92863909062204,
      "grad_norm": 5.342058181762695,
      "learning_rate": 3.630933585938323e-05,
      "loss": 0.5559,
      "step": 21943
    },
    {
      "epoch": 6.928954846858225,
      "grad_norm": 6.912765979766846,
      "learning_rate": 3.630845875872715e-05,
      "loss": 0.5692,
      "step": 21944
    },
    {
      "epoch": 6.929270603094411,
      "grad_norm": 5.986969470977783,
      "learning_rate": 3.6307581658071083e-05,
      "loss": 0.7964,
      "step": 21945
    },
    {
      "epoch": 6.929586359330597,
      "grad_norm": 4.805359363555908,
      "learning_rate": 3.630670455741501e-05,
      "loss": 0.5213,
      "step": 21946
    },
    {
      "epoch": 6.929902115566782,
      "grad_norm": 5.232640266418457,
      "learning_rate": 3.630582745675894e-05,
      "loss": 0.5703,
      "step": 21947
    },
    {
      "epoch": 6.930217871802968,
      "grad_norm": 5.575284004211426,
      "learning_rate": 3.6304950356102865e-05,
      "loss": 0.5253,
      "step": 21948
    },
    {
      "epoch": 6.930533628039154,
      "grad_norm": 5.9751482009887695,
      "learning_rate": 3.63040732554468e-05,
      "loss": 0.5535,
      "step": 21949
    },
    {
      "epoch": 6.93084938427534,
      "grad_norm": 5.3636274337768555,
      "learning_rate": 3.630319615479073e-05,
      "loss": 0.5612,
      "step": 21950
    },
    {
      "epoch": 6.931165140511525,
      "grad_norm": 5.052586078643799,
      "learning_rate": 3.630231905413465e-05,
      "loss": 0.5138,
      "step": 21951
    },
    {
      "epoch": 6.93148089674771,
      "grad_norm": 5.5589280128479,
      "learning_rate": 3.6301441953478585e-05,
      "loss": 0.5488,
      "step": 21952
    },
    {
      "epoch": 6.931796652983897,
      "grad_norm": 4.853449821472168,
      "learning_rate": 3.630056485282251e-05,
      "loss": 0.4716,
      "step": 21953
    },
    {
      "epoch": 6.932112409220082,
      "grad_norm": 6.143102645874023,
      "learning_rate": 3.629968775216644e-05,
      "loss": 0.5856,
      "step": 21954
    },
    {
      "epoch": 6.932428165456268,
      "grad_norm": 5.490749359130859,
      "learning_rate": 3.6298810651510366e-05,
      "loss": 0.5997,
      "step": 21955
    },
    {
      "epoch": 6.932743921692453,
      "grad_norm": 5.2011308670043945,
      "learning_rate": 3.62979335508543e-05,
      "loss": 0.5716,
      "step": 21956
    },
    {
      "epoch": 6.9330596779286395,
      "grad_norm": 4.6473259925842285,
      "learning_rate": 3.629705645019822e-05,
      "loss": 0.5727,
      "step": 21957
    },
    {
      "epoch": 6.933375434164825,
      "grad_norm": 4.489608287811279,
      "learning_rate": 3.6296179349542155e-05,
      "loss": 0.5827,
      "step": 21958
    },
    {
      "epoch": 6.93369119040101,
      "grad_norm": 5.559496879577637,
      "learning_rate": 3.629530224888608e-05,
      "loss": 0.6189,
      "step": 21959
    },
    {
      "epoch": 6.934006946637196,
      "grad_norm": 6.811140537261963,
      "learning_rate": 3.629442514823001e-05,
      "loss": 0.7605,
      "step": 21960
    },
    {
      "epoch": 6.934322702873382,
      "grad_norm": 5.115991592407227,
      "learning_rate": 3.6293548047573936e-05,
      "loss": 0.5771,
      "step": 21961
    },
    {
      "epoch": 6.934638459109568,
      "grad_norm": 6.177570819854736,
      "learning_rate": 3.6292670946917875e-05,
      "loss": 0.5848,
      "step": 21962
    },
    {
      "epoch": 6.934954215345753,
      "grad_norm": 5.624081134796143,
      "learning_rate": 3.62917938462618e-05,
      "loss": 0.5575,
      "step": 21963
    },
    {
      "epoch": 6.935269971581938,
      "grad_norm": 5.072790145874023,
      "learning_rate": 3.629091674560573e-05,
      "loss": 0.5853,
      "step": 21964
    },
    {
      "epoch": 6.9355857278181245,
      "grad_norm": 5.223048686981201,
      "learning_rate": 3.6290039644949656e-05,
      "loss": 0.4909,
      "step": 21965
    },
    {
      "epoch": 6.93590148405431,
      "grad_norm": 6.141612529754639,
      "learning_rate": 3.628916254429359e-05,
      "loss": 0.584,
      "step": 21966
    },
    {
      "epoch": 6.936217240290496,
      "grad_norm": 4.890366554260254,
      "learning_rate": 3.628828544363751e-05,
      "loss": 0.5612,
      "step": 21967
    },
    {
      "epoch": 6.936532996526681,
      "grad_norm": 6.288575649261475,
      "learning_rate": 3.6287408342981444e-05,
      "loss": 0.5973,
      "step": 21968
    },
    {
      "epoch": 6.936848752762867,
      "grad_norm": 5.99875020980835,
      "learning_rate": 3.628653124232537e-05,
      "loss": 0.5899,
      "step": 21969
    },
    {
      "epoch": 6.937164508999053,
      "grad_norm": 5.365102291107178,
      "learning_rate": 3.62856541416693e-05,
      "loss": 0.6027,
      "step": 21970
    },
    {
      "epoch": 6.937480265235238,
      "grad_norm": 4.414927005767822,
      "learning_rate": 3.6284777041013226e-05,
      "loss": 0.5412,
      "step": 21971
    },
    {
      "epoch": 6.937796021471424,
      "grad_norm": 4.914824485778809,
      "learning_rate": 3.628389994035716e-05,
      "loss": 0.5689,
      "step": 21972
    },
    {
      "epoch": 6.93811177770761,
      "grad_norm": 5.300017356872559,
      "learning_rate": 3.628302283970109e-05,
      "loss": 0.5949,
      "step": 21973
    },
    {
      "epoch": 6.938427533943796,
      "grad_norm": 4.771495342254639,
      "learning_rate": 3.6282145739045014e-05,
      "loss": 0.6209,
      "step": 21974
    },
    {
      "epoch": 6.938743290179981,
      "grad_norm": 4.7735371589660645,
      "learning_rate": 3.6281268638388946e-05,
      "loss": 0.5763,
      "step": 21975
    },
    {
      "epoch": 6.939059046416166,
      "grad_norm": 5.8882832527160645,
      "learning_rate": 3.628039153773287e-05,
      "loss": 0.5944,
      "step": 21976
    },
    {
      "epoch": 6.9393748026523525,
      "grad_norm": 4.108493804931641,
      "learning_rate": 3.62795144370768e-05,
      "loss": 0.5405,
      "step": 21977
    },
    {
      "epoch": 6.939690558888538,
      "grad_norm": 4.931944847106934,
      "learning_rate": 3.627863733642073e-05,
      "loss": 0.6966,
      "step": 21978
    },
    {
      "epoch": 6.940006315124724,
      "grad_norm": 4.469751834869385,
      "learning_rate": 3.627776023576466e-05,
      "loss": 0.5405,
      "step": 21979
    },
    {
      "epoch": 6.940322071360909,
      "grad_norm": 5.739583492279053,
      "learning_rate": 3.6276883135108584e-05,
      "loss": 0.5551,
      "step": 21980
    },
    {
      "epoch": 6.9406378275970955,
      "grad_norm": 4.249877452850342,
      "learning_rate": 3.6276006034452516e-05,
      "loss": 0.656,
      "step": 21981
    },
    {
      "epoch": 6.940953583833281,
      "grad_norm": 4.937161445617676,
      "learning_rate": 3.627512893379644e-05,
      "loss": 0.535,
      "step": 21982
    },
    {
      "epoch": 6.941269340069466,
      "grad_norm": 5.6894683837890625,
      "learning_rate": 3.627425183314037e-05,
      "loss": 0.6415,
      "step": 21983
    },
    {
      "epoch": 6.941585096305652,
      "grad_norm": 4.587116718292236,
      "learning_rate": 3.62733747324843e-05,
      "loss": 0.4752,
      "step": 21984
    },
    {
      "epoch": 6.9419008525418375,
      "grad_norm": 4.534526348114014,
      "learning_rate": 3.627249763182823e-05,
      "loss": 0.5874,
      "step": 21985
    },
    {
      "epoch": 6.942216608778024,
      "grad_norm": 6.222031593322754,
      "learning_rate": 3.6271620531172154e-05,
      "loss": 0.5844,
      "step": 21986
    },
    {
      "epoch": 6.942532365014209,
      "grad_norm": 4.897909641265869,
      "learning_rate": 3.6270743430516085e-05,
      "loss": 0.5891,
      "step": 21987
    },
    {
      "epoch": 6.942848121250394,
      "grad_norm": 5.481843948364258,
      "learning_rate": 3.626986632986002e-05,
      "loss": 0.5863,
      "step": 21988
    },
    {
      "epoch": 6.9431638774865805,
      "grad_norm": 5.648734092712402,
      "learning_rate": 3.626898922920395e-05,
      "loss": 0.5223,
      "step": 21989
    },
    {
      "epoch": 6.943479633722766,
      "grad_norm": 5.176648139953613,
      "learning_rate": 3.6268112128547874e-05,
      "loss": 0.644,
      "step": 21990
    },
    {
      "epoch": 6.943795389958952,
      "grad_norm": 4.720588207244873,
      "learning_rate": 3.6267235027891805e-05,
      "loss": 0.5565,
      "step": 21991
    },
    {
      "epoch": 6.944111146195137,
      "grad_norm": 4.493261814117432,
      "learning_rate": 3.626635792723573e-05,
      "loss": 0.5454,
      "step": 21992
    },
    {
      "epoch": 6.9444269024313225,
      "grad_norm": 4.950875759124756,
      "learning_rate": 3.626548082657966e-05,
      "loss": 0.6023,
      "step": 21993
    },
    {
      "epoch": 6.944742658667509,
      "grad_norm": 4.920722484588623,
      "learning_rate": 3.6264603725923594e-05,
      "loss": 0.5466,
      "step": 21994
    },
    {
      "epoch": 6.945058414903694,
      "grad_norm": 4.441908836364746,
      "learning_rate": 3.626372662526752e-05,
      "loss": 0.5646,
      "step": 21995
    },
    {
      "epoch": 6.94537417113988,
      "grad_norm": 4.6790361404418945,
      "learning_rate": 3.626284952461145e-05,
      "loss": 0.6578,
      "step": 21996
    },
    {
      "epoch": 6.9456899273760655,
      "grad_norm": 4.435924530029297,
      "learning_rate": 3.6261972423955375e-05,
      "loss": 0.5859,
      "step": 21997
    },
    {
      "epoch": 6.946005683612252,
      "grad_norm": 4.794007778167725,
      "learning_rate": 3.626109532329931e-05,
      "loss": 0.4954,
      "step": 21998
    },
    {
      "epoch": 6.946321439848437,
      "grad_norm": 5.042635917663574,
      "learning_rate": 3.626021822264323e-05,
      "loss": 0.5986,
      "step": 21999
    },
    {
      "epoch": 6.946637196084623,
      "grad_norm": 4.829497814178467,
      "learning_rate": 3.625934112198716e-05,
      "loss": 0.5853,
      "step": 22000
    },
    {
      "epoch": 6.946637196084623,
      "eval_loss": 0.8561819791793823,
      "eval_runtime": 4.536,
      "eval_samples_per_second": 548.942,
      "eval_steps_per_second": 4.63,
      "step": 22000
    },
    {
      "epoch": 6.9469529523208084,
      "grad_norm": 4.41313362121582,
      "learning_rate": 3.625846402133109e-05,
      "loss": 0.6347,
      "step": 22001
    },
    {
      "epoch": 6.947268708556994,
      "grad_norm": 5.982255458831787,
      "learning_rate": 3.625758692067502e-05,
      "loss": 0.5836,
      "step": 22002
    },
    {
      "epoch": 6.94758446479318,
      "grad_norm": 4.259904384613037,
      "learning_rate": 3.6256709820018945e-05,
      "loss": 0.5997,
      "step": 22003
    },
    {
      "epoch": 6.947900221029365,
      "grad_norm": 5.225088119506836,
      "learning_rate": 3.6255832719362877e-05,
      "loss": 0.6396,
      "step": 22004
    },
    {
      "epoch": 6.948215977265551,
      "grad_norm": 5.654023170471191,
      "learning_rate": 3.62549556187068e-05,
      "loss": 0.6087,
      "step": 22005
    },
    {
      "epoch": 6.948531733501737,
      "grad_norm": 5.617895603179932,
      "learning_rate": 3.625407851805073e-05,
      "loss": 0.6487,
      "step": 22006
    },
    {
      "epoch": 6.948847489737922,
      "grad_norm": 4.644101142883301,
      "learning_rate": 3.625320141739466e-05,
      "loss": 0.541,
      "step": 22007
    },
    {
      "epoch": 6.949163245974108,
      "grad_norm": 4.598981857299805,
      "learning_rate": 3.625232431673859e-05,
      "loss": 0.549,
      "step": 22008
    },
    {
      "epoch": 6.9494790022102935,
      "grad_norm": 4.8030219078063965,
      "learning_rate": 3.625144721608252e-05,
      "loss": 0.5505,
      "step": 22009
    },
    {
      "epoch": 6.94979475844648,
      "grad_norm": 4.617120742797852,
      "learning_rate": 3.6250570115426446e-05,
      "loss": 0.6638,
      "step": 22010
    },
    {
      "epoch": 6.950110514682665,
      "grad_norm": 5.081666469573975,
      "learning_rate": 3.624969301477038e-05,
      "loss": 0.632,
      "step": 22011
    },
    {
      "epoch": 6.95042627091885,
      "grad_norm": 4.871155738830566,
      "learning_rate": 3.62488159141143e-05,
      "loss": 0.5329,
      "step": 22012
    },
    {
      "epoch": 6.950742027155036,
      "grad_norm": 7.595811367034912,
      "learning_rate": 3.6247938813458235e-05,
      "loss": 0.6387,
      "step": 22013
    },
    {
      "epoch": 6.951057783391222,
      "grad_norm": 5.114306926727295,
      "learning_rate": 3.624706171280216e-05,
      "loss": 0.6157,
      "step": 22014
    },
    {
      "epoch": 6.951373539627408,
      "grad_norm": 6.0673627853393555,
      "learning_rate": 3.624618461214609e-05,
      "loss": 0.5873,
      "step": 22015
    },
    {
      "epoch": 6.951689295863593,
      "grad_norm": 5.231578350067139,
      "learning_rate": 3.6245307511490016e-05,
      "loss": 0.5713,
      "step": 22016
    },
    {
      "epoch": 6.9520050520997785,
      "grad_norm": 4.924066543579102,
      "learning_rate": 3.6244430410833955e-05,
      "loss": 0.5498,
      "step": 22017
    },
    {
      "epoch": 6.952320808335965,
      "grad_norm": 4.306492805480957,
      "learning_rate": 3.624355331017788e-05,
      "loss": 0.5295,
      "step": 22018
    },
    {
      "epoch": 6.95263656457215,
      "grad_norm": 5.979563236236572,
      "learning_rate": 3.624267620952181e-05,
      "loss": 0.5648,
      "step": 22019
    },
    {
      "epoch": 6.952952320808336,
      "grad_norm": 5.213050842285156,
      "learning_rate": 3.6241799108865736e-05,
      "loss": 0.5528,
      "step": 22020
    },
    {
      "epoch": 6.953268077044521,
      "grad_norm": 4.84235954284668,
      "learning_rate": 3.624092200820967e-05,
      "loss": 0.5912,
      "step": 22021
    },
    {
      "epoch": 6.953583833280708,
      "grad_norm": 4.471217155456543,
      "learning_rate": 3.624004490755359e-05,
      "loss": 0.5477,
      "step": 22022
    },
    {
      "epoch": 6.953899589516893,
      "grad_norm": 4.956441879272461,
      "learning_rate": 3.6239167806897524e-05,
      "loss": 0.5755,
      "step": 22023
    },
    {
      "epoch": 6.954215345753079,
      "grad_norm": 6.376102924346924,
      "learning_rate": 3.623829070624145e-05,
      "loss": 0.6252,
      "step": 22024
    },
    {
      "epoch": 6.954531101989264,
      "grad_norm": 4.716926097869873,
      "learning_rate": 3.623741360558538e-05,
      "loss": 0.585,
      "step": 22025
    },
    {
      "epoch": 6.95484685822545,
      "grad_norm": 4.359919548034668,
      "learning_rate": 3.6236536504929306e-05,
      "loss": 0.6566,
      "step": 22026
    },
    {
      "epoch": 6.955162614461636,
      "grad_norm": 5.3678297996521,
      "learning_rate": 3.623565940427324e-05,
      "loss": 0.5203,
      "step": 22027
    },
    {
      "epoch": 6.955478370697821,
      "grad_norm": 4.630148410797119,
      "learning_rate": 3.623478230361716e-05,
      "loss": 0.5415,
      "step": 22028
    },
    {
      "epoch": 6.955794126934007,
      "grad_norm": 5.129802703857422,
      "learning_rate": 3.6233905202961094e-05,
      "loss": 0.5841,
      "step": 22029
    },
    {
      "epoch": 6.956109883170193,
      "grad_norm": 6.261389255523682,
      "learning_rate": 3.623302810230502e-05,
      "loss": 0.6164,
      "step": 22030
    },
    {
      "epoch": 6.956425639406378,
      "grad_norm": 4.554428577423096,
      "learning_rate": 3.623215100164895e-05,
      "loss": 0.4711,
      "step": 22031
    },
    {
      "epoch": 6.956741395642564,
      "grad_norm": 4.96378755569458,
      "learning_rate": 3.623127390099288e-05,
      "loss": 0.5667,
      "step": 22032
    },
    {
      "epoch": 6.957057151878749,
      "grad_norm": 5.083214282989502,
      "learning_rate": 3.623039680033681e-05,
      "loss": 0.4344,
      "step": 22033
    },
    {
      "epoch": 6.957372908114936,
      "grad_norm": 4.126733779907227,
      "learning_rate": 3.622951969968074e-05,
      "loss": 0.5738,
      "step": 22034
    },
    {
      "epoch": 6.957688664351121,
      "grad_norm": 5.498283863067627,
      "learning_rate": 3.6228642599024664e-05,
      "loss": 0.6143,
      "step": 22035
    },
    {
      "epoch": 6.958004420587306,
      "grad_norm": 4.354509353637695,
      "learning_rate": 3.6227765498368596e-05,
      "loss": 0.4588,
      "step": 22036
    },
    {
      "epoch": 6.958320176823492,
      "grad_norm": 4.1070942878723145,
      "learning_rate": 3.622688839771252e-05,
      "loss": 0.4547,
      "step": 22037
    },
    {
      "epoch": 6.958635933059678,
      "grad_norm": 4.2584333419799805,
      "learning_rate": 3.622601129705645e-05,
      "loss": 0.5067,
      "step": 22038
    },
    {
      "epoch": 6.958951689295864,
      "grad_norm": 5.956662654876709,
      "learning_rate": 3.622513419640038e-05,
      "loss": 0.6052,
      "step": 22039
    },
    {
      "epoch": 6.959267445532049,
      "grad_norm": 5.202266693115234,
      "learning_rate": 3.622425709574431e-05,
      "loss": 0.5372,
      "step": 22040
    },
    {
      "epoch": 6.959583201768235,
      "grad_norm": 5.274975299835205,
      "learning_rate": 3.6223379995088234e-05,
      "loss": 0.6898,
      "step": 22041
    },
    {
      "epoch": 6.959898958004421,
      "grad_norm": 5.610268592834473,
      "learning_rate": 3.6222502894432165e-05,
      "loss": 0.5487,
      "step": 22042
    },
    {
      "epoch": 6.960214714240606,
      "grad_norm": 6.122384548187256,
      "learning_rate": 3.622162579377609e-05,
      "loss": 0.6508,
      "step": 22043
    },
    {
      "epoch": 6.960530470476792,
      "grad_norm": 5.24529504776001,
      "learning_rate": 3.622074869312003e-05,
      "loss": 0.6618,
      "step": 22044
    },
    {
      "epoch": 6.960846226712977,
      "grad_norm": 4.7757062911987305,
      "learning_rate": 3.6219871592463954e-05,
      "loss": 0.5382,
      "step": 22045
    },
    {
      "epoch": 6.9611619829491636,
      "grad_norm": 5.588476181030273,
      "learning_rate": 3.6218994491807885e-05,
      "loss": 0.5528,
      "step": 22046
    },
    {
      "epoch": 6.961477739185349,
      "grad_norm": 6.023371696472168,
      "learning_rate": 3.621811739115181e-05,
      "loss": 0.5954,
      "step": 22047
    },
    {
      "epoch": 6.961793495421535,
      "grad_norm": 5.962533950805664,
      "learning_rate": 3.621724029049574e-05,
      "loss": 0.7099,
      "step": 22048
    },
    {
      "epoch": 6.96210925165772,
      "grad_norm": 5.79345703125,
      "learning_rate": 3.621636318983967e-05,
      "loss": 0.5853,
      "step": 22049
    },
    {
      "epoch": 6.962425007893906,
      "grad_norm": 4.873682022094727,
      "learning_rate": 3.62154860891836e-05,
      "loss": 0.6281,
      "step": 22050
    },
    {
      "epoch": 6.962740764130092,
      "grad_norm": 5.544485569000244,
      "learning_rate": 3.621460898852752e-05,
      "loss": 0.556,
      "step": 22051
    },
    {
      "epoch": 6.963056520366277,
      "grad_norm": 5.521617889404297,
      "learning_rate": 3.6213731887871455e-05,
      "loss": 0.6678,
      "step": 22052
    },
    {
      "epoch": 6.963372276602463,
      "grad_norm": 4.442016124725342,
      "learning_rate": 3.621285478721539e-05,
      "loss": 0.5817,
      "step": 22053
    },
    {
      "epoch": 6.963688032838649,
      "grad_norm": 4.102853298187256,
      "learning_rate": 3.621197768655931e-05,
      "loss": 0.507,
      "step": 22054
    },
    {
      "epoch": 6.964003789074834,
      "grad_norm": 4.802496433258057,
      "learning_rate": 3.621110058590324e-05,
      "loss": 0.5393,
      "step": 22055
    },
    {
      "epoch": 6.96431954531102,
      "grad_norm": 4.716209888458252,
      "learning_rate": 3.621022348524717e-05,
      "loss": 0.6135,
      "step": 22056
    },
    {
      "epoch": 6.964635301547205,
      "grad_norm": 4.840910911560059,
      "learning_rate": 3.62093463845911e-05,
      "loss": 0.6039,
      "step": 22057
    },
    {
      "epoch": 6.9649510577833915,
      "grad_norm": 5.796439170837402,
      "learning_rate": 3.6208469283935025e-05,
      "loss": 0.5302,
      "step": 22058
    },
    {
      "epoch": 6.965266814019577,
      "grad_norm": 4.791380405426025,
      "learning_rate": 3.6207592183278956e-05,
      "loss": 0.5455,
      "step": 22059
    },
    {
      "epoch": 6.965582570255762,
      "grad_norm": 5.1400370597839355,
      "learning_rate": 3.620671508262288e-05,
      "loss": 0.5928,
      "step": 22060
    },
    {
      "epoch": 6.965898326491948,
      "grad_norm": 5.334714889526367,
      "learning_rate": 3.620583798196681e-05,
      "loss": 0.6087,
      "step": 22061
    },
    {
      "epoch": 6.966214082728134,
      "grad_norm": 4.644920349121094,
      "learning_rate": 3.620496088131074e-05,
      "loss": 0.5645,
      "step": 22062
    },
    {
      "epoch": 6.96652983896432,
      "grad_norm": 4.698676109313965,
      "learning_rate": 3.620408378065467e-05,
      "loss": 0.5248,
      "step": 22063
    },
    {
      "epoch": 6.966845595200505,
      "grad_norm": 4.268430709838867,
      "learning_rate": 3.6203206679998595e-05,
      "loss": 0.5781,
      "step": 22064
    },
    {
      "epoch": 6.967161351436691,
      "grad_norm": 4.989902496337891,
      "learning_rate": 3.6202329579342526e-05,
      "loss": 0.6237,
      "step": 22065
    },
    {
      "epoch": 6.9674771076728765,
      "grad_norm": 5.499139308929443,
      "learning_rate": 3.620145247868645e-05,
      "loss": 0.5317,
      "step": 22066
    },
    {
      "epoch": 6.967792863909062,
      "grad_norm": 5.749910354614258,
      "learning_rate": 3.620057537803038e-05,
      "loss": 0.6221,
      "step": 22067
    },
    {
      "epoch": 6.968108620145248,
      "grad_norm": 4.93072509765625,
      "learning_rate": 3.619969827737431e-05,
      "loss": 0.6257,
      "step": 22068
    },
    {
      "epoch": 6.968424376381433,
      "grad_norm": 3.8937599658966064,
      "learning_rate": 3.619882117671824e-05,
      "loss": 0.4919,
      "step": 22069
    },
    {
      "epoch": 6.9687401326176195,
      "grad_norm": 4.290457725524902,
      "learning_rate": 3.619794407606217e-05,
      "loss": 0.6411,
      "step": 22070
    },
    {
      "epoch": 6.969055888853805,
      "grad_norm": 4.484784126281738,
      "learning_rate": 3.61970669754061e-05,
      "loss": 0.5727,
      "step": 22071
    },
    {
      "epoch": 6.969371645089991,
      "grad_norm": 4.5852952003479,
      "learning_rate": 3.619618987475003e-05,
      "loss": 0.5575,
      "step": 22072
    },
    {
      "epoch": 6.969687401326176,
      "grad_norm": 5.434322834014893,
      "learning_rate": 3.619531277409396e-05,
      "loss": 0.5506,
      "step": 22073
    },
    {
      "epoch": 6.9700031575623616,
      "grad_norm": 4.8988447189331055,
      "learning_rate": 3.6194435673437884e-05,
      "loss": 0.7049,
      "step": 22074
    },
    {
      "epoch": 6.970318913798548,
      "grad_norm": 4.471123218536377,
      "learning_rate": 3.6193558572781816e-05,
      "loss": 0.6913,
      "step": 22075
    },
    {
      "epoch": 6.970634670034733,
      "grad_norm": 4.121973514556885,
      "learning_rate": 3.619268147212575e-05,
      "loss": 0.5995,
      "step": 22076
    },
    {
      "epoch": 6.970950426270919,
      "grad_norm": 4.093874454498291,
      "learning_rate": 3.619180437146967e-05,
      "loss": 0.4406,
      "step": 22077
    },
    {
      "epoch": 6.9712661825071045,
      "grad_norm": 5.763881206512451,
      "learning_rate": 3.6190927270813604e-05,
      "loss": 0.4952,
      "step": 22078
    },
    {
      "epoch": 6.97158193874329,
      "grad_norm": 4.985153675079346,
      "learning_rate": 3.619005017015753e-05,
      "loss": 0.5736,
      "step": 22079
    },
    {
      "epoch": 6.971897694979476,
      "grad_norm": 3.98976469039917,
      "learning_rate": 3.618917306950146e-05,
      "loss": 0.5431,
      "step": 22080
    },
    {
      "epoch": 6.972213451215661,
      "grad_norm": 4.5564703941345215,
      "learning_rate": 3.6188295968845386e-05,
      "loss": 0.4889,
      "step": 22081
    },
    {
      "epoch": 6.9725292074518475,
      "grad_norm": 4.149458885192871,
      "learning_rate": 3.618741886818932e-05,
      "loss": 0.6214,
      "step": 22082
    },
    {
      "epoch": 6.972844963688033,
      "grad_norm": 4.84595251083374,
      "learning_rate": 3.618654176753324e-05,
      "loss": 0.5308,
      "step": 22083
    },
    {
      "epoch": 6.973160719924218,
      "grad_norm": 6.817205429077148,
      "learning_rate": 3.6185664666877174e-05,
      "loss": 0.5746,
      "step": 22084
    },
    {
      "epoch": 6.973476476160404,
      "grad_norm": 5.901490688323975,
      "learning_rate": 3.61847875662211e-05,
      "loss": 0.5129,
      "step": 22085
    },
    {
      "epoch": 6.9737922323965895,
      "grad_norm": 4.408807277679443,
      "learning_rate": 3.618391046556503e-05,
      "loss": 0.6919,
      "step": 22086
    },
    {
      "epoch": 6.974107988632776,
      "grad_norm": 4.958324432373047,
      "learning_rate": 3.6183033364908955e-05,
      "loss": 0.5739,
      "step": 22087
    },
    {
      "epoch": 6.974423744868961,
      "grad_norm": 5.70616340637207,
      "learning_rate": 3.618215626425289e-05,
      "loss": 0.5743,
      "step": 22088
    },
    {
      "epoch": 6.974739501105147,
      "grad_norm": 6.8489298820495605,
      "learning_rate": 3.618127916359681e-05,
      "loss": 0.5172,
      "step": 22089
    },
    {
      "epoch": 6.9750552573413325,
      "grad_norm": 5.098927974700928,
      "learning_rate": 3.6180402062940744e-05,
      "loss": 0.5544,
      "step": 22090
    },
    {
      "epoch": 6.975371013577519,
      "grad_norm": 5.447993755340576,
      "learning_rate": 3.6179524962284675e-05,
      "loss": 0.6219,
      "step": 22091
    },
    {
      "epoch": 6.975686769813704,
      "grad_norm": 5.703955173492432,
      "learning_rate": 3.61786478616286e-05,
      "loss": 0.6053,
      "step": 22092
    },
    {
      "epoch": 6.976002526049889,
      "grad_norm": 5.668084621429443,
      "learning_rate": 3.617777076097253e-05,
      "loss": 0.5808,
      "step": 22093
    },
    {
      "epoch": 6.976318282286075,
      "grad_norm": 5.390112400054932,
      "learning_rate": 3.617689366031646e-05,
      "loss": 0.694,
      "step": 22094
    },
    {
      "epoch": 6.976634038522261,
      "grad_norm": 4.953259468078613,
      "learning_rate": 3.617601655966039e-05,
      "loss": 0.5726,
      "step": 22095
    },
    {
      "epoch": 6.976949794758447,
      "grad_norm": 5.639485836029053,
      "learning_rate": 3.6175139459004313e-05,
      "loss": 0.5185,
      "step": 22096
    },
    {
      "epoch": 6.977265550994632,
      "grad_norm": 5.941633701324463,
      "learning_rate": 3.6174262358348245e-05,
      "loss": 0.554,
      "step": 22097
    },
    {
      "epoch": 6.9775813072308175,
      "grad_norm": 6.360679626464844,
      "learning_rate": 3.617338525769218e-05,
      "loss": 0.6718,
      "step": 22098
    },
    {
      "epoch": 6.977897063467004,
      "grad_norm": 4.838873863220215,
      "learning_rate": 3.617250815703611e-05,
      "loss": 0.4878,
      "step": 22099
    },
    {
      "epoch": 6.978212819703189,
      "grad_norm": 5.640377521514893,
      "learning_rate": 3.6171631056380033e-05,
      "loss": 0.6695,
      "step": 22100
    },
    {
      "epoch": 6.978528575939375,
      "grad_norm": 5.2996907234191895,
      "learning_rate": 3.6170753955723965e-05,
      "loss": 0.6554,
      "step": 22101
    },
    {
      "epoch": 6.97884433217556,
      "grad_norm": 4.842787265777588,
      "learning_rate": 3.616987685506789e-05,
      "loss": 0.4733,
      "step": 22102
    },
    {
      "epoch": 6.979160088411746,
      "grad_norm": 4.831332206726074,
      "learning_rate": 3.616899975441182e-05,
      "loss": 0.5899,
      "step": 22103
    },
    {
      "epoch": 6.979475844647932,
      "grad_norm": 4.707373142242432,
      "learning_rate": 3.616812265375575e-05,
      "loss": 0.5826,
      "step": 22104
    },
    {
      "epoch": 6.979791600884117,
      "grad_norm": 5.172704696655273,
      "learning_rate": 3.616724555309968e-05,
      "loss": 0.5717,
      "step": 22105
    },
    {
      "epoch": 6.980107357120303,
      "grad_norm": 4.29000997543335,
      "learning_rate": 3.61663684524436e-05,
      "loss": 0.4464,
      "step": 22106
    },
    {
      "epoch": 6.980423113356489,
      "grad_norm": 5.227535724639893,
      "learning_rate": 3.6165491351787535e-05,
      "loss": 0.5883,
      "step": 22107
    },
    {
      "epoch": 6.980738869592674,
      "grad_norm": 5.169153690338135,
      "learning_rate": 3.616461425113146e-05,
      "loss": 0.55,
      "step": 22108
    },
    {
      "epoch": 6.98105462582886,
      "grad_norm": 5.606324672698975,
      "learning_rate": 3.616373715047539e-05,
      "loss": 0.593,
      "step": 22109
    },
    {
      "epoch": 6.9813703820650455,
      "grad_norm": 4.778304100036621,
      "learning_rate": 3.6162860049819316e-05,
      "loss": 0.5748,
      "step": 22110
    },
    {
      "epoch": 6.981686138301232,
      "grad_norm": 4.626556873321533,
      "learning_rate": 3.616198294916325e-05,
      "loss": 0.565,
      "step": 22111
    },
    {
      "epoch": 6.982001894537417,
      "grad_norm": 5.125060081481934,
      "learning_rate": 3.616110584850717e-05,
      "loss": 0.6656,
      "step": 22112
    },
    {
      "epoch": 6.982317650773603,
      "grad_norm": 5.446906089782715,
      "learning_rate": 3.6160228747851105e-05,
      "loss": 0.6335,
      "step": 22113
    },
    {
      "epoch": 6.982633407009788,
      "grad_norm": 6.274146556854248,
      "learning_rate": 3.6159351647195036e-05,
      "loss": 0.6671,
      "step": 22114
    },
    {
      "epoch": 6.982949163245975,
      "grad_norm": 5.08788537979126,
      "learning_rate": 3.615847454653896e-05,
      "loss": 0.5402,
      "step": 22115
    },
    {
      "epoch": 6.98326491948216,
      "grad_norm": 6.11372184753418,
      "learning_rate": 3.615759744588289e-05,
      "loss": 0.5774,
      "step": 22116
    },
    {
      "epoch": 6.983580675718345,
      "grad_norm": 4.643338680267334,
      "learning_rate": 3.615672034522682e-05,
      "loss": 0.5957,
      "step": 22117
    },
    {
      "epoch": 6.983896431954531,
      "grad_norm": 5.168761730194092,
      "learning_rate": 3.615584324457075e-05,
      "loss": 0.5382,
      "step": 22118
    },
    {
      "epoch": 6.984212188190717,
      "grad_norm": 5.4062066078186035,
      "learning_rate": 3.6154966143914674e-05,
      "loss": 0.7194,
      "step": 22119
    },
    {
      "epoch": 6.984527944426903,
      "grad_norm": 3.9344162940979004,
      "learning_rate": 3.6154089043258606e-05,
      "loss": 0.5827,
      "step": 22120
    },
    {
      "epoch": 6.984843700663088,
      "grad_norm": 5.538644790649414,
      "learning_rate": 3.615321194260253e-05,
      "loss": 0.7336,
      "step": 22121
    },
    {
      "epoch": 6.985159456899273,
      "grad_norm": 4.8563032150268555,
      "learning_rate": 3.615233484194646e-05,
      "loss": 0.5587,
      "step": 22122
    },
    {
      "epoch": 6.98547521313546,
      "grad_norm": 3.9849038124084473,
      "learning_rate": 3.615145774129039e-05,
      "loss": 0.5201,
      "step": 22123
    },
    {
      "epoch": 6.985790969371645,
      "grad_norm": 4.249179840087891,
      "learning_rate": 3.615058064063432e-05,
      "loss": 0.5871,
      "step": 22124
    },
    {
      "epoch": 6.986106725607831,
      "grad_norm": 4.364628314971924,
      "learning_rate": 3.6149703539978244e-05,
      "loss": 0.5418,
      "step": 22125
    },
    {
      "epoch": 6.986422481844016,
      "grad_norm": 4.829502105712891,
      "learning_rate": 3.614882643932218e-05,
      "loss": 0.5821,
      "step": 22126
    },
    {
      "epoch": 6.986738238080202,
      "grad_norm": 6.123438835144043,
      "learning_rate": 3.614794933866611e-05,
      "loss": 0.598,
      "step": 22127
    },
    {
      "epoch": 6.987053994316388,
      "grad_norm": 4.111670017242432,
      "learning_rate": 3.614707223801004e-05,
      "loss": 0.5743,
      "step": 22128
    },
    {
      "epoch": 6.987369750552573,
      "grad_norm": 4.1264729499816895,
      "learning_rate": 3.6146195137353964e-05,
      "loss": 0.5406,
      "step": 22129
    },
    {
      "epoch": 6.987685506788759,
      "grad_norm": 5.2442946434021,
      "learning_rate": 3.6145318036697896e-05,
      "loss": 0.6231,
      "step": 22130
    },
    {
      "epoch": 6.988001263024945,
      "grad_norm": 4.917387962341309,
      "learning_rate": 3.614444093604182e-05,
      "loss": 0.5566,
      "step": 22131
    },
    {
      "epoch": 6.98831701926113,
      "grad_norm": 3.9482507705688477,
      "learning_rate": 3.614356383538575e-05,
      "loss": 0.5467,
      "step": 22132
    },
    {
      "epoch": 6.988632775497316,
      "grad_norm": 5.101865291595459,
      "learning_rate": 3.614268673472968e-05,
      "loss": 0.6072,
      "step": 22133
    },
    {
      "epoch": 6.988948531733501,
      "grad_norm": 4.589438438415527,
      "learning_rate": 3.614180963407361e-05,
      "loss": 0.5461,
      "step": 22134
    },
    {
      "epoch": 6.989264287969688,
      "grad_norm": 4.402433395385742,
      "learning_rate": 3.614093253341754e-05,
      "loss": 0.5881,
      "step": 22135
    },
    {
      "epoch": 6.989580044205873,
      "grad_norm": 4.437968730926514,
      "learning_rate": 3.6140055432761466e-05,
      "loss": 0.5195,
      "step": 22136
    },
    {
      "epoch": 6.989895800442059,
      "grad_norm": 4.451127529144287,
      "learning_rate": 3.61391783321054e-05,
      "loss": 0.5936,
      "step": 22137
    },
    {
      "epoch": 6.990211556678244,
      "grad_norm": 4.660393238067627,
      "learning_rate": 3.613830123144932e-05,
      "loss": 0.5726,
      "step": 22138
    },
    {
      "epoch": 6.9905273129144305,
      "grad_norm": 4.647471904754639,
      "learning_rate": 3.6137424130793254e-05,
      "loss": 0.4883,
      "step": 22139
    },
    {
      "epoch": 6.990843069150616,
      "grad_norm": 3.8384299278259277,
      "learning_rate": 3.613654703013718e-05,
      "loss": 0.4622,
      "step": 22140
    },
    {
      "epoch": 6.991158825386801,
      "grad_norm": 7.770935535430908,
      "learning_rate": 3.613566992948111e-05,
      "loss": 0.6322,
      "step": 22141
    },
    {
      "epoch": 6.991474581622987,
      "grad_norm": 4.852118492126465,
      "learning_rate": 3.6134792828825035e-05,
      "loss": 0.5413,
      "step": 22142
    },
    {
      "epoch": 6.991790337859173,
      "grad_norm": 4.690256118774414,
      "learning_rate": 3.613391572816897e-05,
      "loss": 0.4617,
      "step": 22143
    },
    {
      "epoch": 6.992106094095359,
      "grad_norm": 4.635382652282715,
      "learning_rate": 3.613303862751289e-05,
      "loss": 0.5722,
      "step": 22144
    },
    {
      "epoch": 6.992421850331544,
      "grad_norm": 4.499157905578613,
      "learning_rate": 3.6132161526856824e-05,
      "loss": 0.4587,
      "step": 22145
    },
    {
      "epoch": 6.992737606567729,
      "grad_norm": 6.0409040451049805,
      "learning_rate": 3.613128442620075e-05,
      "loss": 0.6968,
      "step": 22146
    },
    {
      "epoch": 6.9930533628039155,
      "grad_norm": 5.684030055999756,
      "learning_rate": 3.613040732554468e-05,
      "loss": 0.6313,
      "step": 22147
    },
    {
      "epoch": 6.993369119040101,
      "grad_norm": 4.70092248916626,
      "learning_rate": 3.6129530224888605e-05,
      "loss": 0.4983,
      "step": 22148
    },
    {
      "epoch": 6.993684875276287,
      "grad_norm": 4.553476810455322,
      "learning_rate": 3.612865312423254e-05,
      "loss": 0.5716,
      "step": 22149
    },
    {
      "epoch": 6.994000631512472,
      "grad_norm": 4.955400466918945,
      "learning_rate": 3.612777602357646e-05,
      "loss": 0.6128,
      "step": 22150
    },
    {
      "epoch": 6.994316387748658,
      "grad_norm": 4.936614990234375,
      "learning_rate": 3.612689892292039e-05,
      "loss": 0.6406,
      "step": 22151
    },
    {
      "epoch": 6.994632143984844,
      "grad_norm": 4.8133864402771,
      "learning_rate": 3.6126021822264325e-05,
      "loss": 0.4317,
      "step": 22152
    },
    {
      "epoch": 6.994947900221029,
      "grad_norm": 4.9877119064331055,
      "learning_rate": 3.612514472160826e-05,
      "loss": 0.5705,
      "step": 22153
    },
    {
      "epoch": 6.995263656457215,
      "grad_norm": 5.143065929412842,
      "learning_rate": 3.612426762095218e-05,
      "loss": 0.4974,
      "step": 22154
    },
    {
      "epoch": 6.995579412693401,
      "grad_norm": 4.997758388519287,
      "learning_rate": 3.612339052029611e-05,
      "loss": 0.5656,
      "step": 22155
    },
    {
      "epoch": 6.995895168929587,
      "grad_norm": 4.428543567657471,
      "learning_rate": 3.6122513419640045e-05,
      "loss": 0.5122,
      "step": 22156
    },
    {
      "epoch": 6.996210925165772,
      "grad_norm": 5.026739597320557,
      "learning_rate": 3.612163631898397e-05,
      "loss": 0.5665,
      "step": 22157
    },
    {
      "epoch": 6.996526681401957,
      "grad_norm": 4.745314121246338,
      "learning_rate": 3.61207592183279e-05,
      "loss": 0.602,
      "step": 22158
    },
    {
      "epoch": 6.9968424376381435,
      "grad_norm": 5.263713359832764,
      "learning_rate": 3.6119882117671827e-05,
      "loss": 0.496,
      "step": 22159
    },
    {
      "epoch": 6.997158193874329,
      "grad_norm": 5.843262195587158,
      "learning_rate": 3.611900501701576e-05,
      "loss": 0.705,
      "step": 22160
    },
    {
      "epoch": 6.997473950110515,
      "grad_norm": 4.540343761444092,
      "learning_rate": 3.611812791635968e-05,
      "loss": 0.5674,
      "step": 22161
    },
    {
      "epoch": 6.9977897063467,
      "grad_norm": 5.682435512542725,
      "learning_rate": 3.6117250815703615e-05,
      "loss": 0.5955,
      "step": 22162
    },
    {
      "epoch": 6.9981054625828865,
      "grad_norm": 4.900399684906006,
      "learning_rate": 3.611637371504754e-05,
      "loss": 0.5198,
      "step": 22163
    },
    {
      "epoch": 6.998421218819072,
      "grad_norm": 4.251272678375244,
      "learning_rate": 3.611549661439147e-05,
      "loss": 0.4854,
      "step": 22164
    },
    {
      "epoch": 6.998736975055257,
      "grad_norm": 5.129830837249756,
      "learning_rate": 3.6114619513735396e-05,
      "loss": 0.5044,
      "step": 22165
    },
    {
      "epoch": 6.999052731291443,
      "grad_norm": 4.993007183074951,
      "learning_rate": 3.611374241307933e-05,
      "loss": 0.5562,
      "step": 22166
    },
    {
      "epoch": 6.9993684875276285,
      "grad_norm": 4.81241512298584,
      "learning_rate": 3.611286531242325e-05,
      "loss": 0.5861,
      "step": 22167
    },
    {
      "epoch": 6.999684243763815,
      "grad_norm": 4.619251728057861,
      "learning_rate": 3.6111988211767185e-05,
      "loss": 0.6025,
      "step": 22168
    },
    {
      "epoch": 7.0,
      "grad_norm": 5.146200180053711,
      "learning_rate": 3.611111111111111e-05,
      "loss": 0.5603,
      "step": 22169
    },
    {
      "epoch": 7.000315756236185,
      "grad_norm": 4.811770915985107,
      "learning_rate": 3.611023401045504e-05,
      "loss": 0.5685,
      "step": 22170
    },
    {
      "epoch": 7.0006315124723715,
      "grad_norm": 3.4534571170806885,
      "learning_rate": 3.6109356909798966e-05,
      "loss": 0.4284,
      "step": 22171
    },
    {
      "epoch": 7.000947268708557,
      "grad_norm": 4.338915824890137,
      "learning_rate": 3.61084798091429e-05,
      "loss": 0.4313,
      "step": 22172
    },
    {
      "epoch": 7.001263024944743,
      "grad_norm": 5.102180004119873,
      "learning_rate": 3.610760270848683e-05,
      "loss": 0.6068,
      "step": 22173
    },
    {
      "epoch": 7.001578781180928,
      "grad_norm": 5.177271842956543,
      "learning_rate": 3.6106725607830754e-05,
      "loss": 0.5233,
      "step": 22174
    },
    {
      "epoch": 7.001894537417114,
      "grad_norm": 3.8336310386657715,
      "learning_rate": 3.6105848507174686e-05,
      "loss": 0.4741,
      "step": 22175
    },
    {
      "epoch": 7.0022102936533,
      "grad_norm": 4.494474411010742,
      "learning_rate": 3.610497140651861e-05,
      "loss": 0.4868,
      "step": 22176
    },
    {
      "epoch": 7.002526049889485,
      "grad_norm": 3.676522731781006,
      "learning_rate": 3.610409430586254e-05,
      "loss": 0.4289,
      "step": 22177
    },
    {
      "epoch": 7.002841806125671,
      "grad_norm": 4.406410217285156,
      "learning_rate": 3.610321720520647e-05,
      "loss": 0.4103,
      "step": 22178
    },
    {
      "epoch": 7.0031575623618565,
      "grad_norm": 4.891279697418213,
      "learning_rate": 3.61023401045504e-05,
      "loss": 0.4388,
      "step": 22179
    },
    {
      "epoch": 7.003473318598043,
      "grad_norm": 4.773900985717773,
      "learning_rate": 3.610146300389433e-05,
      "loss": 0.4388,
      "step": 22180
    },
    {
      "epoch": 7.003789074834228,
      "grad_norm": 4.63384485244751,
      "learning_rate": 3.610058590323826e-05,
      "loss": 0.5466,
      "step": 22181
    },
    {
      "epoch": 7.004104831070413,
      "grad_norm": 4.627593994140625,
      "learning_rate": 3.609970880258219e-05,
      "loss": 0.5458,
      "step": 22182
    },
    {
      "epoch": 7.004420587306599,
      "grad_norm": 4.427268028259277,
      "learning_rate": 3.609883170192612e-05,
      "loss": 0.4424,
      "step": 22183
    },
    {
      "epoch": 7.004736343542785,
      "grad_norm": 3.9524528980255127,
      "learning_rate": 3.6097954601270044e-05,
      "loss": 0.4054,
      "step": 22184
    },
    {
      "epoch": 7.005052099778971,
      "grad_norm": 5.35416841506958,
      "learning_rate": 3.6097077500613976e-05,
      "loss": 0.5335,
      "step": 22185
    },
    {
      "epoch": 7.005367856015156,
      "grad_norm": 4.573141098022461,
      "learning_rate": 3.60962003999579e-05,
      "loss": 0.3771,
      "step": 22186
    },
    {
      "epoch": 7.005683612251342,
      "grad_norm": 4.866006851196289,
      "learning_rate": 3.609532329930183e-05,
      "loss": 0.4558,
      "step": 22187
    },
    {
      "epoch": 7.005999368487528,
      "grad_norm": 5.247819900512695,
      "learning_rate": 3.609444619864576e-05,
      "loss": 0.5305,
      "step": 22188
    },
    {
      "epoch": 7.006315124723713,
      "grad_norm": 4.462770938873291,
      "learning_rate": 3.609356909798969e-05,
      "loss": 0.4475,
      "step": 22189
    },
    {
      "epoch": 7.006630880959899,
      "grad_norm": 4.276569366455078,
      "learning_rate": 3.6092691997333614e-05,
      "loss": 0.3492,
      "step": 22190
    },
    {
      "epoch": 7.0069466371960845,
      "grad_norm": 5.455066204071045,
      "learning_rate": 3.6091814896677545e-05,
      "loss": 0.5101,
      "step": 22191
    },
    {
      "epoch": 7.007262393432271,
      "grad_norm": 5.182767391204834,
      "learning_rate": 3.609093779602147e-05,
      "loss": 0.4175,
      "step": 22192
    },
    {
      "epoch": 7.007578149668456,
      "grad_norm": 5.602136611938477,
      "learning_rate": 3.60900606953654e-05,
      "loss": 0.437,
      "step": 22193
    },
    {
      "epoch": 7.007893905904641,
      "grad_norm": 4.641940116882324,
      "learning_rate": 3.6089183594709334e-05,
      "loss": 0.4501,
      "step": 22194
    },
    {
      "epoch": 7.008209662140827,
      "grad_norm": 4.597947120666504,
      "learning_rate": 3.608830649405326e-05,
      "loss": 0.4705,
      "step": 22195
    },
    {
      "epoch": 7.008525418377013,
      "grad_norm": 5.084550380706787,
      "learning_rate": 3.608742939339719e-05,
      "loss": 0.4876,
      "step": 22196
    },
    {
      "epoch": 7.008841174613199,
      "grad_norm": 4.375195503234863,
      "learning_rate": 3.6086552292741115e-05,
      "loss": 0.4669,
      "step": 22197
    },
    {
      "epoch": 7.009156930849384,
      "grad_norm": 6.755227565765381,
      "learning_rate": 3.608567519208505e-05,
      "loss": 0.5237,
      "step": 22198
    },
    {
      "epoch": 7.00947268708557,
      "grad_norm": 5.142900466918945,
      "learning_rate": 3.608479809142897e-05,
      "loss": 0.437,
      "step": 22199
    },
    {
      "epoch": 7.009788443321756,
      "grad_norm": 5.781574249267578,
      "learning_rate": 3.6083920990772904e-05,
      "loss": 0.4464,
      "step": 22200
    },
    {
      "epoch": 7.009788443321756,
      "eval_loss": 0.9121037125587463,
      "eval_runtime": 4.7039,
      "eval_samples_per_second": 529.349,
      "eval_steps_per_second": 4.464,
      "step": 22200
    },
    {
      "epoch": 7.010104199557941,
      "grad_norm": 4.838747024536133,
      "learning_rate": 3.608304389011683e-05,
      "loss": 0.3884,
      "step": 22201
    },
    {
      "epoch": 7.010419955794127,
      "grad_norm": 6.846618175506592,
      "learning_rate": 3.608216678946076e-05,
      "loss": 0.5567,
      "step": 22202
    },
    {
      "epoch": 7.010735712030312,
      "grad_norm": 5.168092727661133,
      "learning_rate": 3.6081289688804685e-05,
      "loss": 0.4941,
      "step": 22203
    },
    {
      "epoch": 7.011051468266499,
      "grad_norm": 5.036281108856201,
      "learning_rate": 3.608041258814862e-05,
      "loss": 0.4412,
      "step": 22204
    },
    {
      "epoch": 7.011367224502684,
      "grad_norm": 4.907787799835205,
      "learning_rate": 3.607953548749254e-05,
      "loss": 0.4892,
      "step": 22205
    },
    {
      "epoch": 7.011682980738869,
      "grad_norm": 5.942442417144775,
      "learning_rate": 3.607865838683647e-05,
      "loss": 0.5489,
      "step": 22206
    },
    {
      "epoch": 7.011998736975055,
      "grad_norm": 4.9739251136779785,
      "learning_rate": 3.6077781286180405e-05,
      "loss": 0.5401,
      "step": 22207
    },
    {
      "epoch": 7.012314493211241,
      "grad_norm": 5.257174491882324,
      "learning_rate": 3.607690418552434e-05,
      "loss": 0.4358,
      "step": 22208
    },
    {
      "epoch": 7.012630249447427,
      "grad_norm": 4.4504594802856445,
      "learning_rate": 3.607602708486826e-05,
      "loss": 0.4525,
      "step": 22209
    },
    {
      "epoch": 7.012946005683612,
      "grad_norm": 5.618514537811279,
      "learning_rate": 3.607514998421219e-05,
      "loss": 0.433,
      "step": 22210
    },
    {
      "epoch": 7.013261761919798,
      "grad_norm": 4.7885050773620605,
      "learning_rate": 3.607427288355612e-05,
      "loss": 0.4681,
      "step": 22211
    },
    {
      "epoch": 7.013577518155984,
      "grad_norm": 6.884038925170898,
      "learning_rate": 3.607339578290005e-05,
      "loss": 0.4517,
      "step": 22212
    },
    {
      "epoch": 7.013893274392169,
      "grad_norm": 4.283053398132324,
      "learning_rate": 3.6072518682243975e-05,
      "loss": 0.4081,
      "step": 22213
    },
    {
      "epoch": 7.014209030628355,
      "grad_norm": 4.686877250671387,
      "learning_rate": 3.6071641581587906e-05,
      "loss": 0.5223,
      "step": 22214
    },
    {
      "epoch": 7.01452478686454,
      "grad_norm": 5.0121965408325195,
      "learning_rate": 3.607076448093183e-05,
      "loss": 0.433,
      "step": 22215
    },
    {
      "epoch": 7.014840543100727,
      "grad_norm": 4.900450229644775,
      "learning_rate": 3.606988738027576e-05,
      "loss": 0.4386,
      "step": 22216
    },
    {
      "epoch": 7.015156299336912,
      "grad_norm": 4.718852519989014,
      "learning_rate": 3.6069010279619695e-05,
      "loss": 0.4735,
      "step": 22217
    },
    {
      "epoch": 7.015472055573097,
      "grad_norm": 7.0347137451171875,
      "learning_rate": 3.606813317896362e-05,
      "loss": 0.4774,
      "step": 22218
    },
    {
      "epoch": 7.015787811809283,
      "grad_norm": 5.175618648529053,
      "learning_rate": 3.606725607830755e-05,
      "loss": 0.4933,
      "step": 22219
    },
    {
      "epoch": 7.016103568045469,
      "grad_norm": 3.788339853286743,
      "learning_rate": 3.6066378977651476e-05,
      "loss": 0.3663,
      "step": 22220
    },
    {
      "epoch": 7.016419324281655,
      "grad_norm": 5.65502405166626,
      "learning_rate": 3.606550187699541e-05,
      "loss": 0.4387,
      "step": 22221
    },
    {
      "epoch": 7.01673508051784,
      "grad_norm": 4.6724629402160645,
      "learning_rate": 3.606462477633933e-05,
      "loss": 0.3589,
      "step": 22222
    },
    {
      "epoch": 7.017050836754026,
      "grad_norm": 5.904171943664551,
      "learning_rate": 3.6063747675683264e-05,
      "loss": 0.4196,
      "step": 22223
    },
    {
      "epoch": 7.017366592990212,
      "grad_norm": 4.817084312438965,
      "learning_rate": 3.606287057502719e-05,
      "loss": 0.4844,
      "step": 22224
    },
    {
      "epoch": 7.017682349226397,
      "grad_norm": 5.056992053985596,
      "learning_rate": 3.606199347437112e-05,
      "loss": 0.4335,
      "step": 22225
    },
    {
      "epoch": 7.017998105462583,
      "grad_norm": 6.259918689727783,
      "learning_rate": 3.6061116373715046e-05,
      "loss": 0.4613,
      "step": 22226
    },
    {
      "epoch": 7.018313861698768,
      "grad_norm": 5.649210453033447,
      "learning_rate": 3.606023927305898e-05,
      "loss": 0.5014,
      "step": 22227
    },
    {
      "epoch": 7.0186296179349545,
      "grad_norm": 4.71779203414917,
      "learning_rate": 3.60593621724029e-05,
      "loss": 0.3768,
      "step": 22228
    },
    {
      "epoch": 7.01894537417114,
      "grad_norm": 5.150831699371338,
      "learning_rate": 3.6058485071746834e-05,
      "loss": 0.4526,
      "step": 22229
    },
    {
      "epoch": 7.019261130407325,
      "grad_norm": 6.507899761199951,
      "learning_rate": 3.605760797109076e-05,
      "loss": 0.449,
      "step": 22230
    },
    {
      "epoch": 7.019576886643511,
      "grad_norm": 3.916746139526367,
      "learning_rate": 3.605673087043469e-05,
      "loss": 0.3779,
      "step": 22231
    },
    {
      "epoch": 7.019892642879697,
      "grad_norm": 5.454901218414307,
      "learning_rate": 3.605585376977862e-05,
      "loss": 0.6087,
      "step": 22232
    },
    {
      "epoch": 7.020208399115883,
      "grad_norm": 6.0900983810424805,
      "learning_rate": 3.605497666912255e-05,
      "loss": 0.4879,
      "step": 22233
    },
    {
      "epoch": 7.020524155352068,
      "grad_norm": 4.463115215301514,
      "learning_rate": 3.605409956846648e-05,
      "loss": 0.3658,
      "step": 22234
    },
    {
      "epoch": 7.020839911588254,
      "grad_norm": 5.053996562957764,
      "learning_rate": 3.605322246781041e-05,
      "loss": 0.4837,
      "step": 22235
    },
    {
      "epoch": 7.02115566782444,
      "grad_norm": 3.9846956729888916,
      "learning_rate": 3.6052345367154336e-05,
      "loss": 0.3635,
      "step": 22236
    },
    {
      "epoch": 7.021471424060625,
      "grad_norm": 5.767993927001953,
      "learning_rate": 3.605146826649827e-05,
      "loss": 0.4093,
      "step": 22237
    },
    {
      "epoch": 7.021787180296811,
      "grad_norm": 5.698065757751465,
      "learning_rate": 3.60505911658422e-05,
      "loss": 0.5015,
      "step": 22238
    },
    {
      "epoch": 7.022102936532996,
      "grad_norm": 4.999824047088623,
      "learning_rate": 3.6049714065186124e-05,
      "loss": 0.5143,
      "step": 22239
    },
    {
      "epoch": 7.0224186927691825,
      "grad_norm": 6.471340179443359,
      "learning_rate": 3.6048836964530056e-05,
      "loss": 0.53,
      "step": 22240
    },
    {
      "epoch": 7.022734449005368,
      "grad_norm": 6.469850540161133,
      "learning_rate": 3.604795986387398e-05,
      "loss": 0.4832,
      "step": 22241
    },
    {
      "epoch": 7.023050205241553,
      "grad_norm": 3.4533934593200684,
      "learning_rate": 3.604708276321791e-05,
      "loss": 0.3091,
      "step": 22242
    },
    {
      "epoch": 7.023365961477739,
      "grad_norm": 5.118527412414551,
      "learning_rate": 3.604620566256184e-05,
      "loss": 0.4557,
      "step": 22243
    },
    {
      "epoch": 7.023681717713925,
      "grad_norm": 3.2801427841186523,
      "learning_rate": 3.604532856190577e-05,
      "loss": 0.3172,
      "step": 22244
    },
    {
      "epoch": 7.023997473950111,
      "grad_norm": 7.287694454193115,
      "learning_rate": 3.6044451461249694e-05,
      "loss": 0.5244,
      "step": 22245
    },
    {
      "epoch": 7.024313230186296,
      "grad_norm": 6.925536155700684,
      "learning_rate": 3.6043574360593625e-05,
      "loss": 0.5782,
      "step": 22246
    },
    {
      "epoch": 7.024628986422482,
      "grad_norm": 4.985787868499756,
      "learning_rate": 3.604269725993755e-05,
      "loss": 0.3932,
      "step": 22247
    },
    {
      "epoch": 7.0249447426586675,
      "grad_norm": 6.319300174713135,
      "learning_rate": 3.604182015928148e-05,
      "loss": 0.5406,
      "step": 22248
    },
    {
      "epoch": 7.025260498894853,
      "grad_norm": 4.286462783813477,
      "learning_rate": 3.604094305862541e-05,
      "loss": 0.4065,
      "step": 22249
    },
    {
      "epoch": 7.025576255131039,
      "grad_norm": 6.7775349617004395,
      "learning_rate": 3.604006595796934e-05,
      "loss": 0.5628,
      "step": 22250
    },
    {
      "epoch": 7.025892011367224,
      "grad_norm": 6.476775169372559,
      "learning_rate": 3.6039188857313263e-05,
      "loss": 0.6141,
      "step": 22251
    },
    {
      "epoch": 7.0262077676034105,
      "grad_norm": 6.1819539070129395,
      "learning_rate": 3.6038311756657195e-05,
      "loss": 0.4979,
      "step": 22252
    },
    {
      "epoch": 7.026523523839596,
      "grad_norm": 7.016693592071533,
      "learning_rate": 3.603743465600112e-05,
      "loss": 0.4981,
      "step": 22253
    },
    {
      "epoch": 7.026839280075781,
      "grad_norm": 4.3792853355407715,
      "learning_rate": 3.603655755534505e-05,
      "loss": 0.4057,
      "step": 22254
    },
    {
      "epoch": 7.027155036311967,
      "grad_norm": 6.27034330368042,
      "learning_rate": 3.6035680454688983e-05,
      "loss": 0.5349,
      "step": 22255
    },
    {
      "epoch": 7.0274707925481525,
      "grad_norm": 5.304362773895264,
      "learning_rate": 3.603480335403291e-05,
      "loss": 0.4515,
      "step": 22256
    },
    {
      "epoch": 7.027786548784339,
      "grad_norm": 6.699838638305664,
      "learning_rate": 3.603392625337684e-05,
      "loss": 0.4665,
      "step": 22257
    },
    {
      "epoch": 7.028102305020524,
      "grad_norm": 4.858046531677246,
      "learning_rate": 3.6033049152720765e-05,
      "loss": 0.4704,
      "step": 22258
    },
    {
      "epoch": 7.02841806125671,
      "grad_norm": 4.3791656494140625,
      "learning_rate": 3.6032172052064697e-05,
      "loss": 0.405,
      "step": 22259
    },
    {
      "epoch": 7.0287338174928955,
      "grad_norm": 5.116612434387207,
      "learning_rate": 3.603129495140862e-05,
      "loss": 0.4421,
      "step": 22260
    },
    {
      "epoch": 7.029049573729081,
      "grad_norm": 5.783936500549316,
      "learning_rate": 3.603041785075255e-05,
      "loss": 0.5231,
      "step": 22261
    },
    {
      "epoch": 7.029365329965267,
      "grad_norm": 4.243934631347656,
      "learning_rate": 3.6029540750096485e-05,
      "loss": 0.3186,
      "step": 22262
    },
    {
      "epoch": 7.029681086201452,
      "grad_norm": 5.8851518630981445,
      "learning_rate": 3.6028663649440417e-05,
      "loss": 0.5191,
      "step": 22263
    },
    {
      "epoch": 7.0299968424376384,
      "grad_norm": 4.4610395431518555,
      "learning_rate": 3.602778654878434e-05,
      "loss": 0.36,
      "step": 22264
    },
    {
      "epoch": 7.030312598673824,
      "grad_norm": 6.8099541664123535,
      "learning_rate": 3.602690944812827e-05,
      "loss": 0.4185,
      "step": 22265
    },
    {
      "epoch": 7.030628354910009,
      "grad_norm": 7.22166109085083,
      "learning_rate": 3.60260323474722e-05,
      "loss": 0.4093,
      "step": 22266
    },
    {
      "epoch": 7.030944111146195,
      "grad_norm": 5.897063255310059,
      "learning_rate": 3.602515524681613e-05,
      "loss": 0.4302,
      "step": 22267
    },
    {
      "epoch": 7.0312598673823805,
      "grad_norm": 5.804470062255859,
      "learning_rate": 3.6024278146160055e-05,
      "loss": 0.3931,
      "step": 22268
    },
    {
      "epoch": 7.031575623618567,
      "grad_norm": 5.009557723999023,
      "learning_rate": 3.6023401045503986e-05,
      "loss": 0.4669,
      "step": 22269
    },
    {
      "epoch": 7.031891379854752,
      "grad_norm": 4.7615532875061035,
      "learning_rate": 3.602252394484791e-05,
      "loss": 0.4789,
      "step": 22270
    },
    {
      "epoch": 7.032207136090938,
      "grad_norm": 9.225977897644043,
      "learning_rate": 3.602164684419184e-05,
      "loss": 0.524,
      "step": 22271
    },
    {
      "epoch": 7.0325228923271235,
      "grad_norm": 4.917749404907227,
      "learning_rate": 3.602076974353577e-05,
      "loss": 0.3726,
      "step": 22272
    },
    {
      "epoch": 7.032838648563309,
      "grad_norm": 4.9206671714782715,
      "learning_rate": 3.60198926428797e-05,
      "loss": 0.455,
      "step": 22273
    },
    {
      "epoch": 7.033154404799495,
      "grad_norm": 4.92777156829834,
      "learning_rate": 3.6019015542223624e-05,
      "loss": 0.4953,
      "step": 22274
    },
    {
      "epoch": 7.03347016103568,
      "grad_norm": 4.922708034515381,
      "learning_rate": 3.6018138441567556e-05,
      "loss": 0.446,
      "step": 22275
    },
    {
      "epoch": 7.033785917271866,
      "grad_norm": 5.617835521697998,
      "learning_rate": 3.601726134091149e-05,
      "loss": 0.4119,
      "step": 22276
    },
    {
      "epoch": 7.034101673508052,
      "grad_norm": 5.458741664886475,
      "learning_rate": 3.601638424025541e-05,
      "loss": 0.6373,
      "step": 22277
    },
    {
      "epoch": 7.034417429744238,
      "grad_norm": 4.873772144317627,
      "learning_rate": 3.6015507139599344e-05,
      "loss": 0.4759,
      "step": 22278
    },
    {
      "epoch": 7.034733185980423,
      "grad_norm": 5.3311848640441895,
      "learning_rate": 3.601463003894327e-05,
      "loss": 0.4269,
      "step": 22279
    },
    {
      "epoch": 7.0350489422166085,
      "grad_norm": 4.063807010650635,
      "learning_rate": 3.60137529382872e-05,
      "loss": 0.4234,
      "step": 22280
    },
    {
      "epoch": 7.035364698452795,
      "grad_norm": 7.708060264587402,
      "learning_rate": 3.6012875837631126e-05,
      "loss": 0.4328,
      "step": 22281
    },
    {
      "epoch": 7.03568045468898,
      "grad_norm": 5.687057971954346,
      "learning_rate": 3.601199873697506e-05,
      "loss": 0.4608,
      "step": 22282
    },
    {
      "epoch": 7.035996210925166,
      "grad_norm": 7.400481700897217,
      "learning_rate": 3.601112163631898e-05,
      "loss": 0.5381,
      "step": 22283
    },
    {
      "epoch": 7.036311967161351,
      "grad_norm": 4.905197620391846,
      "learning_rate": 3.6010244535662914e-05,
      "loss": 0.4363,
      "step": 22284
    },
    {
      "epoch": 7.036627723397537,
      "grad_norm": 4.078649997711182,
      "learning_rate": 3.600936743500684e-05,
      "loss": 0.3314,
      "step": 22285
    },
    {
      "epoch": 7.036943479633723,
      "grad_norm": 4.855475902557373,
      "learning_rate": 3.600849033435077e-05,
      "loss": 0.443,
      "step": 22286
    },
    {
      "epoch": 7.037259235869908,
      "grad_norm": 5.995343208312988,
      "learning_rate": 3.6007613233694696e-05,
      "loss": 0.4451,
      "step": 22287
    },
    {
      "epoch": 7.037574992106094,
      "grad_norm": 6.130988597869873,
      "learning_rate": 3.600673613303863e-05,
      "loss": 0.4448,
      "step": 22288
    },
    {
      "epoch": 7.03789074834228,
      "grad_norm": 4.769659519195557,
      "learning_rate": 3.600585903238256e-05,
      "loss": 0.5268,
      "step": 22289
    },
    {
      "epoch": 7.038206504578466,
      "grad_norm": 5.90848445892334,
      "learning_rate": 3.600498193172649e-05,
      "loss": 0.4517,
      "step": 22290
    },
    {
      "epoch": 7.038522260814651,
      "grad_norm": 4.566667079925537,
      "learning_rate": 3.6004104831070416e-05,
      "loss": 0.4056,
      "step": 22291
    },
    {
      "epoch": 7.0388380170508364,
      "grad_norm": 4.4019694328308105,
      "learning_rate": 3.600322773041435e-05,
      "loss": 0.3435,
      "step": 22292
    },
    {
      "epoch": 7.039153773287023,
      "grad_norm": 5.387929916381836,
      "learning_rate": 3.600235062975827e-05,
      "loss": 0.4849,
      "step": 22293
    },
    {
      "epoch": 7.039469529523208,
      "grad_norm": 5.73419713973999,
      "learning_rate": 3.6001473529102204e-05,
      "loss": 0.4235,
      "step": 22294
    },
    {
      "epoch": 7.039785285759394,
      "grad_norm": 4.866235733032227,
      "learning_rate": 3.600059642844613e-05,
      "loss": 0.437,
      "step": 22295
    },
    {
      "epoch": 7.040101041995579,
      "grad_norm": 5.300231456756592,
      "learning_rate": 3.599971932779006e-05,
      "loss": 0.4151,
      "step": 22296
    },
    {
      "epoch": 7.040416798231765,
      "grad_norm": 5.227283000946045,
      "learning_rate": 3.5998842227133985e-05,
      "loss": 0.484,
      "step": 22297
    },
    {
      "epoch": 7.040732554467951,
      "grad_norm": 5.496938228607178,
      "learning_rate": 3.599796512647792e-05,
      "loss": 0.4923,
      "step": 22298
    },
    {
      "epoch": 7.041048310704136,
      "grad_norm": 4.9756999015808105,
      "learning_rate": 3.599708802582185e-05,
      "loss": 0.4494,
      "step": 22299
    },
    {
      "epoch": 7.041364066940322,
      "grad_norm": 5.167628288269043,
      "learning_rate": 3.5996210925165774e-05,
      "loss": 0.386,
      "step": 22300
    },
    {
      "epoch": 7.041679823176508,
      "grad_norm": 5.100106239318848,
      "learning_rate": 3.5995333824509705e-05,
      "loss": 0.4682,
      "step": 22301
    },
    {
      "epoch": 7.041995579412694,
      "grad_norm": 4.756168365478516,
      "learning_rate": 3.599445672385363e-05,
      "loss": 0.43,
      "step": 22302
    },
    {
      "epoch": 7.042311335648879,
      "grad_norm": 5.099382400512695,
      "learning_rate": 3.599357962319756e-05,
      "loss": 0.4792,
      "step": 22303
    },
    {
      "epoch": 7.042627091885064,
      "grad_norm": 5.959208965301514,
      "learning_rate": 3.599270252254149e-05,
      "loss": 0.5346,
      "step": 22304
    },
    {
      "epoch": 7.042942848121251,
      "grad_norm": 4.969681739807129,
      "learning_rate": 3.599182542188542e-05,
      "loss": 0.3156,
      "step": 22305
    },
    {
      "epoch": 7.043258604357436,
      "grad_norm": 4.816198348999023,
      "learning_rate": 3.599094832122934e-05,
      "loss": 0.4624,
      "step": 22306
    },
    {
      "epoch": 7.043574360593622,
      "grad_norm": 4.6841559410095215,
      "learning_rate": 3.5990071220573275e-05,
      "loss": 0.3977,
      "step": 22307
    },
    {
      "epoch": 7.043890116829807,
      "grad_norm": 6.481935977935791,
      "learning_rate": 3.59891941199172e-05,
      "loss": 0.4352,
      "step": 22308
    },
    {
      "epoch": 7.044205873065993,
      "grad_norm": 5.0252885818481445,
      "learning_rate": 3.598831701926113e-05,
      "loss": 0.4573,
      "step": 22309
    },
    {
      "epoch": 7.044521629302179,
      "grad_norm": 5.08314323425293,
      "learning_rate": 3.5987439918605057e-05,
      "loss": 0.428,
      "step": 22310
    },
    {
      "epoch": 7.044837385538364,
      "grad_norm": 5.868117809295654,
      "learning_rate": 3.598656281794899e-05,
      "loss": 0.4988,
      "step": 22311
    },
    {
      "epoch": 7.04515314177455,
      "grad_norm": 4.224593639373779,
      "learning_rate": 3.598568571729291e-05,
      "loss": 0.4366,
      "step": 22312
    },
    {
      "epoch": 7.045468898010736,
      "grad_norm": 5.816057205200195,
      "learning_rate": 3.5984808616636845e-05,
      "loss": 0.4711,
      "step": 22313
    },
    {
      "epoch": 7.045784654246922,
      "grad_norm": 6.980649948120117,
      "learning_rate": 3.5983931515980776e-05,
      "loss": 0.4494,
      "step": 22314
    },
    {
      "epoch": 7.046100410483107,
      "grad_norm": 5.2229180335998535,
      "learning_rate": 3.59830544153247e-05,
      "loss": 0.4162,
      "step": 22315
    },
    {
      "epoch": 7.046416166719292,
      "grad_norm": 4.843115329742432,
      "learning_rate": 3.598217731466863e-05,
      "loss": 0.4018,
      "step": 22316
    },
    {
      "epoch": 7.046731922955479,
      "grad_norm": 3.7030158042907715,
      "learning_rate": 3.5981300214012565e-05,
      "loss": 0.4263,
      "step": 22317
    },
    {
      "epoch": 7.047047679191664,
      "grad_norm": 4.130303382873535,
      "learning_rate": 3.598042311335649e-05,
      "loss": 0.386,
      "step": 22318
    },
    {
      "epoch": 7.04736343542785,
      "grad_norm": 5.373084545135498,
      "learning_rate": 3.597954601270042e-05,
      "loss": 0.481,
      "step": 22319
    },
    {
      "epoch": 7.047679191664035,
      "grad_norm": 5.685662269592285,
      "learning_rate": 3.597866891204435e-05,
      "loss": 0.482,
      "step": 22320
    },
    {
      "epoch": 7.047994947900221,
      "grad_norm": 6.130359649658203,
      "learning_rate": 3.597779181138828e-05,
      "loss": 0.5995,
      "step": 22321
    },
    {
      "epoch": 7.048310704136407,
      "grad_norm": 4.602170944213867,
      "learning_rate": 3.597691471073221e-05,
      "loss": 0.3688,
      "step": 22322
    },
    {
      "epoch": 7.048626460372592,
      "grad_norm": 9.011584281921387,
      "learning_rate": 3.5976037610076135e-05,
      "loss": 0.4399,
      "step": 22323
    },
    {
      "epoch": 7.048942216608778,
      "grad_norm": 5.231954574584961,
      "learning_rate": 3.5975160509420066e-05,
      "loss": 0.3925,
      "step": 22324
    },
    {
      "epoch": 7.049257972844964,
      "grad_norm": 5.912276268005371,
      "learning_rate": 3.597428340876399e-05,
      "loss": 0.4598,
      "step": 22325
    },
    {
      "epoch": 7.04957372908115,
      "grad_norm": 5.855095386505127,
      "learning_rate": 3.597340630810792e-05,
      "loss": 0.4731,
      "step": 22326
    },
    {
      "epoch": 7.049889485317335,
      "grad_norm": 5.663570404052734,
      "learning_rate": 3.597252920745185e-05,
      "loss": 0.4706,
      "step": 22327
    },
    {
      "epoch": 7.05020524155352,
      "grad_norm": 4.679487705230713,
      "learning_rate": 3.597165210679578e-05,
      "loss": 0.3669,
      "step": 22328
    },
    {
      "epoch": 7.0505209977897065,
      "grad_norm": 6.764662265777588,
      "learning_rate": 3.5970775006139704e-05,
      "loss": 0.5803,
      "step": 22329
    },
    {
      "epoch": 7.050836754025892,
      "grad_norm": 7.565079689025879,
      "learning_rate": 3.5969897905483636e-05,
      "loss": 0.5426,
      "step": 22330
    },
    {
      "epoch": 7.051152510262078,
      "grad_norm": 5.697246074676514,
      "learning_rate": 3.596902080482756e-05,
      "loss": 0.5199,
      "step": 22331
    },
    {
      "epoch": 7.051468266498263,
      "grad_norm": 5.440536022186279,
      "learning_rate": 3.596814370417149e-05,
      "loss": 0.4563,
      "step": 22332
    },
    {
      "epoch": 7.051784022734449,
      "grad_norm": 4.33384895324707,
      "learning_rate": 3.596726660351542e-05,
      "loss": 0.4298,
      "step": 22333
    },
    {
      "epoch": 7.052099778970635,
      "grad_norm": 7.395542144775391,
      "learning_rate": 3.596638950285935e-05,
      "loss": 0.5495,
      "step": 22334
    },
    {
      "epoch": 7.05241553520682,
      "grad_norm": 4.415109157562256,
      "learning_rate": 3.596551240220328e-05,
      "loss": 0.3983,
      "step": 22335
    },
    {
      "epoch": 7.052731291443006,
      "grad_norm": 5.431436061859131,
      "learning_rate": 3.5964635301547206e-05,
      "loss": 0.4376,
      "step": 22336
    },
    {
      "epoch": 7.0530470476791916,
      "grad_norm": 5.266690731048584,
      "learning_rate": 3.596375820089114e-05,
      "loss": 0.4056,
      "step": 22337
    },
    {
      "epoch": 7.053362803915378,
      "grad_norm": 4.499402046203613,
      "learning_rate": 3.596288110023506e-05,
      "loss": 0.4387,
      "step": 22338
    },
    {
      "epoch": 7.053678560151563,
      "grad_norm": 4.4716410636901855,
      "learning_rate": 3.5962003999578994e-05,
      "loss": 0.5319,
      "step": 22339
    },
    {
      "epoch": 7.053994316387748,
      "grad_norm": 4.741358280181885,
      "learning_rate": 3.596112689892292e-05,
      "loss": 0.3568,
      "step": 22340
    },
    {
      "epoch": 7.0543100726239345,
      "grad_norm": 4.674736976623535,
      "learning_rate": 3.596024979826685e-05,
      "loss": 0.4169,
      "step": 22341
    },
    {
      "epoch": 7.05462582886012,
      "grad_norm": 5.534772872924805,
      "learning_rate": 3.5959372697610775e-05,
      "loss": 0.4417,
      "step": 22342
    },
    {
      "epoch": 7.054941585096306,
      "grad_norm": 5.583869457244873,
      "learning_rate": 3.595849559695471e-05,
      "loss": 0.4021,
      "step": 22343
    },
    {
      "epoch": 7.055257341332491,
      "grad_norm": 5.0885910987854,
      "learning_rate": 3.595761849629864e-05,
      "loss": 0.4936,
      "step": 22344
    },
    {
      "epoch": 7.055573097568677,
      "grad_norm": 5.0338006019592285,
      "learning_rate": 3.595674139564257e-05,
      "loss": 0.55,
      "step": 22345
    },
    {
      "epoch": 7.055888853804863,
      "grad_norm": 6.09207820892334,
      "learning_rate": 3.5955864294986495e-05,
      "loss": 0.4599,
      "step": 22346
    },
    {
      "epoch": 7.056204610041048,
      "grad_norm": 6.146852016448975,
      "learning_rate": 3.595498719433043e-05,
      "loss": 0.5842,
      "step": 22347
    },
    {
      "epoch": 7.056520366277234,
      "grad_norm": 6.275496006011963,
      "learning_rate": 3.595411009367435e-05,
      "loss": 0.3981,
      "step": 22348
    },
    {
      "epoch": 7.0568361225134195,
      "grad_norm": 5.790773868560791,
      "learning_rate": 3.5953232993018284e-05,
      "loss": 0.4419,
      "step": 22349
    },
    {
      "epoch": 7.057151878749606,
      "grad_norm": 4.62076997756958,
      "learning_rate": 3.595235589236221e-05,
      "loss": 0.4338,
      "step": 22350
    },
    {
      "epoch": 7.057467634985791,
      "grad_norm": 4.869534969329834,
      "learning_rate": 3.595147879170614e-05,
      "loss": 0.4435,
      "step": 22351
    },
    {
      "epoch": 7.057783391221976,
      "grad_norm": 4.445836544036865,
      "learning_rate": 3.5950601691050065e-05,
      "loss": 0.4304,
      "step": 22352
    },
    {
      "epoch": 7.0580991474581625,
      "grad_norm": 7.724268913269043,
      "learning_rate": 3.5949724590394e-05,
      "loss": 0.4882,
      "step": 22353
    },
    {
      "epoch": 7.058414903694348,
      "grad_norm": 4.435666084289551,
      "learning_rate": 3.594884748973792e-05,
      "loss": 0.4256,
      "step": 22354
    },
    {
      "epoch": 7.058730659930534,
      "grad_norm": 5.064178943634033,
      "learning_rate": 3.5947970389081853e-05,
      "loss": 0.4408,
      "step": 22355
    },
    {
      "epoch": 7.059046416166719,
      "grad_norm": 5.903217315673828,
      "learning_rate": 3.594709328842578e-05,
      "loss": 0.4196,
      "step": 22356
    },
    {
      "epoch": 7.0593621724029045,
      "grad_norm": 6.529848575592041,
      "learning_rate": 3.594621618776971e-05,
      "loss": 0.4293,
      "step": 22357
    },
    {
      "epoch": 7.059677928639091,
      "grad_norm": 4.4652862548828125,
      "learning_rate": 3.594533908711364e-05,
      "loss": 0.5391,
      "step": 22358
    },
    {
      "epoch": 7.059993684875276,
      "grad_norm": 5.491406440734863,
      "learning_rate": 3.594446198645757e-05,
      "loss": 0.4966,
      "step": 22359
    },
    {
      "epoch": 7.060309441111462,
      "grad_norm": 8.079123497009277,
      "learning_rate": 3.59435848858015e-05,
      "loss": 0.4974,
      "step": 22360
    },
    {
      "epoch": 7.0606251973476475,
      "grad_norm": 5.738033771514893,
      "learning_rate": 3.594270778514542e-05,
      "loss": 0.4722,
      "step": 22361
    },
    {
      "epoch": 7.060940953583834,
      "grad_norm": 6.220080375671387,
      "learning_rate": 3.5941830684489355e-05,
      "loss": 0.398,
      "step": 22362
    },
    {
      "epoch": 7.061256709820019,
      "grad_norm": 5.485145092010498,
      "learning_rate": 3.594095358383328e-05,
      "loss": 0.4786,
      "step": 22363
    },
    {
      "epoch": 7.061572466056204,
      "grad_norm": 4.414881706237793,
      "learning_rate": 3.594007648317721e-05,
      "loss": 0.475,
      "step": 22364
    },
    {
      "epoch": 7.06188822229239,
      "grad_norm": 4.200226306915283,
      "learning_rate": 3.5939199382521136e-05,
      "loss": 0.4145,
      "step": 22365
    },
    {
      "epoch": 7.062203978528576,
      "grad_norm": 5.436367511749268,
      "learning_rate": 3.593832228186507e-05,
      "loss": 0.5207,
      "step": 22366
    },
    {
      "epoch": 7.062519734764762,
      "grad_norm": 3.927725315093994,
      "learning_rate": 3.593744518120899e-05,
      "loss": 0.3921,
      "step": 22367
    },
    {
      "epoch": 7.062835491000947,
      "grad_norm": 5.366022109985352,
      "learning_rate": 3.5936568080552925e-05,
      "loss": 0.5124,
      "step": 22368
    },
    {
      "epoch": 7.0631512472371325,
      "grad_norm": 4.872943878173828,
      "learning_rate": 3.593569097989685e-05,
      "loss": 0.4794,
      "step": 22369
    },
    {
      "epoch": 7.063467003473319,
      "grad_norm": 4.848676681518555,
      "learning_rate": 3.593481387924078e-05,
      "loss": 0.4306,
      "step": 22370
    },
    {
      "epoch": 7.063782759709504,
      "grad_norm": 5.825948715209961,
      "learning_rate": 3.593393677858471e-05,
      "loss": 0.4681,
      "step": 22371
    },
    {
      "epoch": 7.06409851594569,
      "grad_norm": 4.439289569854736,
      "learning_rate": 3.5933059677928645e-05,
      "loss": 0.342,
      "step": 22372
    },
    {
      "epoch": 7.0644142721818755,
      "grad_norm": 5.424180030822754,
      "learning_rate": 3.593218257727257e-05,
      "loss": 0.4517,
      "step": 22373
    },
    {
      "epoch": 7.064730028418062,
      "grad_norm": 6.007444858551025,
      "learning_rate": 3.59313054766165e-05,
      "loss": 0.4343,
      "step": 22374
    },
    {
      "epoch": 7.065045784654247,
      "grad_norm": 6.259256362915039,
      "learning_rate": 3.5930428375960426e-05,
      "loss": 0.4352,
      "step": 22375
    },
    {
      "epoch": 7.065361540890432,
      "grad_norm": 4.554975986480713,
      "learning_rate": 3.592955127530436e-05,
      "loss": 0.4475,
      "step": 22376
    },
    {
      "epoch": 7.065677297126618,
      "grad_norm": 5.375634670257568,
      "learning_rate": 3.592867417464828e-05,
      "loss": 0.4199,
      "step": 22377
    },
    {
      "epoch": 7.065993053362804,
      "grad_norm": 5.270655155181885,
      "learning_rate": 3.5927797073992214e-05,
      "loss": 0.3594,
      "step": 22378
    },
    {
      "epoch": 7.06630880959899,
      "grad_norm": 6.780384063720703,
      "learning_rate": 3.5926919973336146e-05,
      "loss": 0.4056,
      "step": 22379
    },
    {
      "epoch": 7.066624565835175,
      "grad_norm": 4.5633931159973145,
      "learning_rate": 3.592604287268007e-05,
      "loss": 0.4707,
      "step": 22380
    },
    {
      "epoch": 7.0669403220713605,
      "grad_norm": 6.364652633666992,
      "learning_rate": 3.5925165772024e-05,
      "loss": 0.5091,
      "step": 22381
    },
    {
      "epoch": 7.067256078307547,
      "grad_norm": 5.970709800720215,
      "learning_rate": 3.592428867136793e-05,
      "loss": 0.3979,
      "step": 22382
    },
    {
      "epoch": 7.067571834543732,
      "grad_norm": 5.396983623504639,
      "learning_rate": 3.592341157071186e-05,
      "loss": 0.5312,
      "step": 22383
    },
    {
      "epoch": 7.067887590779918,
      "grad_norm": 6.079472541809082,
      "learning_rate": 3.5922534470055784e-05,
      "loss": 0.4508,
      "step": 22384
    },
    {
      "epoch": 7.068203347016103,
      "grad_norm": 8.003192901611328,
      "learning_rate": 3.5921657369399716e-05,
      "loss": 0.4967,
      "step": 22385
    },
    {
      "epoch": 7.06851910325229,
      "grad_norm": 5.3119964599609375,
      "learning_rate": 3.592078026874364e-05,
      "loss": 0.4183,
      "step": 22386
    },
    {
      "epoch": 7.068834859488475,
      "grad_norm": 5.856378078460693,
      "learning_rate": 3.591990316808757e-05,
      "loss": 0.4813,
      "step": 22387
    },
    {
      "epoch": 7.06915061572466,
      "grad_norm": 4.034310340881348,
      "learning_rate": 3.59190260674315e-05,
      "loss": 0.3879,
      "step": 22388
    },
    {
      "epoch": 7.069466371960846,
      "grad_norm": 5.3182806968688965,
      "learning_rate": 3.591814896677543e-05,
      "loss": 0.4375,
      "step": 22389
    },
    {
      "epoch": 7.069782128197032,
      "grad_norm": 5.194889545440674,
      "learning_rate": 3.5917271866119354e-05,
      "loss": 0.4228,
      "step": 22390
    },
    {
      "epoch": 7.070097884433218,
      "grad_norm": 4.905822277069092,
      "learning_rate": 3.5916394765463286e-05,
      "loss": 0.4406,
      "step": 22391
    },
    {
      "epoch": 7.070413640669403,
      "grad_norm": 6.447943687438965,
      "learning_rate": 3.591551766480721e-05,
      "loss": 0.5334,
      "step": 22392
    },
    {
      "epoch": 7.070729396905589,
      "grad_norm": 5.762459754943848,
      "learning_rate": 3.591464056415114e-05,
      "loss": 0.3967,
      "step": 22393
    },
    {
      "epoch": 7.071045153141775,
      "grad_norm": 6.824715614318848,
      "learning_rate": 3.591376346349507e-05,
      "loss": 0.5183,
      "step": 22394
    },
    {
      "epoch": 7.07136090937796,
      "grad_norm": 5.888587951660156,
      "learning_rate": 3.5912886362839e-05,
      "loss": 0.4213,
      "step": 22395
    },
    {
      "epoch": 7.071676665614146,
      "grad_norm": 5.501126766204834,
      "learning_rate": 3.591200926218293e-05,
      "loss": 0.4676,
      "step": 22396
    },
    {
      "epoch": 7.071992421850331,
      "grad_norm": 4.55050802230835,
      "learning_rate": 3.5911132161526855e-05,
      "loss": 0.4744,
      "step": 22397
    },
    {
      "epoch": 7.072308178086518,
      "grad_norm": 5.293005466461182,
      "learning_rate": 3.591025506087079e-05,
      "loss": 0.5037,
      "step": 22398
    },
    {
      "epoch": 7.072623934322703,
      "grad_norm": 6.939387321472168,
      "learning_rate": 3.590937796021472e-05,
      "loss": 0.5383,
      "step": 22399
    },
    {
      "epoch": 7.072939690558888,
      "grad_norm": 4.485966682434082,
      "learning_rate": 3.5908500859558644e-05,
      "loss": 0.3556,
      "step": 22400
    },
    {
      "epoch": 7.072939690558888,
      "eval_loss": 0.9429607391357422,
      "eval_runtime": 4.5424,
      "eval_samples_per_second": 548.168,
      "eval_steps_per_second": 4.623,
      "step": 22400
    },
    {
      "epoch": 7.073255446795074,
      "grad_norm": 5.612006664276123,
      "learning_rate": 3.5907623758902575e-05,
      "loss": 0.3682,
      "step": 22401
    },
    {
      "epoch": 7.07357120303126,
      "grad_norm": 5.43302059173584,
      "learning_rate": 3.590674665824651e-05,
      "loss": 0.4744,
      "step": 22402
    },
    {
      "epoch": 7.073886959267446,
      "grad_norm": 6.373016357421875,
      "learning_rate": 3.590586955759043e-05,
      "loss": 0.4459,
      "step": 22403
    },
    {
      "epoch": 7.074202715503631,
      "grad_norm": 6.986098289489746,
      "learning_rate": 3.5904992456934364e-05,
      "loss": 0.4767,
      "step": 22404
    },
    {
      "epoch": 7.074518471739817,
      "grad_norm": 4.6520490646362305,
      "learning_rate": 3.590411535627829e-05,
      "loss": 0.4015,
      "step": 22405
    },
    {
      "epoch": 7.074834227976003,
      "grad_norm": 6.259149551391602,
      "learning_rate": 3.590323825562222e-05,
      "loss": 0.5238,
      "step": 22406
    },
    {
      "epoch": 7.075149984212188,
      "grad_norm": 6.1135053634643555,
      "learning_rate": 3.5902361154966145e-05,
      "loss": 0.4315,
      "step": 22407
    },
    {
      "epoch": 7.075465740448374,
      "grad_norm": 4.644266605377197,
      "learning_rate": 3.590148405431008e-05,
      "loss": 0.4307,
      "step": 22408
    },
    {
      "epoch": 7.075781496684559,
      "grad_norm": 5.705076217651367,
      "learning_rate": 3.5900606953654e-05,
      "loss": 0.4811,
      "step": 22409
    },
    {
      "epoch": 7.0760972529207455,
      "grad_norm": 4.890546798706055,
      "learning_rate": 3.589972985299793e-05,
      "loss": 0.4687,
      "step": 22410
    },
    {
      "epoch": 7.076413009156931,
      "grad_norm": 5.381651401519775,
      "learning_rate": 3.589885275234186e-05,
      "loss": 0.435,
      "step": 22411
    },
    {
      "epoch": 7.076728765393116,
      "grad_norm": 5.851934909820557,
      "learning_rate": 3.589797565168579e-05,
      "loss": 0.5179,
      "step": 22412
    },
    {
      "epoch": 7.077044521629302,
      "grad_norm": 5.346093654632568,
      "learning_rate": 3.5897098551029715e-05,
      "loss": 0.3918,
      "step": 22413
    },
    {
      "epoch": 7.077360277865488,
      "grad_norm": 4.687443256378174,
      "learning_rate": 3.5896221450373647e-05,
      "loss": 0.4151,
      "step": 22414
    },
    {
      "epoch": 7.077676034101674,
      "grad_norm": 5.486275672912598,
      "learning_rate": 3.589534434971757e-05,
      "loss": 0.4254,
      "step": 22415
    },
    {
      "epoch": 7.077991790337859,
      "grad_norm": 7.371901988983154,
      "learning_rate": 3.58944672490615e-05,
      "loss": 0.4655,
      "step": 22416
    },
    {
      "epoch": 7.078307546574045,
      "grad_norm": 5.791675090789795,
      "learning_rate": 3.5893590148405435e-05,
      "loss": 0.4886,
      "step": 22417
    },
    {
      "epoch": 7.0786233028102306,
      "grad_norm": 6.710893154144287,
      "learning_rate": 3.589271304774936e-05,
      "loss": 0.4143,
      "step": 22418
    },
    {
      "epoch": 7.078939059046416,
      "grad_norm": 5.227053165435791,
      "learning_rate": 3.589183594709329e-05,
      "loss": 0.5062,
      "step": 22419
    },
    {
      "epoch": 7.079254815282602,
      "grad_norm": 6.132260799407959,
      "learning_rate": 3.5890958846437216e-05,
      "loss": 0.4157,
      "step": 22420
    },
    {
      "epoch": 7.079570571518787,
      "grad_norm": 6.126026153564453,
      "learning_rate": 3.589008174578115e-05,
      "loss": 0.4825,
      "step": 22421
    },
    {
      "epoch": 7.0798863277549735,
      "grad_norm": 6.276448726654053,
      "learning_rate": 3.588920464512507e-05,
      "loss": 0.4923,
      "step": 22422
    },
    {
      "epoch": 7.080202083991159,
      "grad_norm": 3.8659932613372803,
      "learning_rate": 3.5888327544469005e-05,
      "loss": 0.3551,
      "step": 22423
    },
    {
      "epoch": 7.080517840227344,
      "grad_norm": 5.970307350158691,
      "learning_rate": 3.588745044381293e-05,
      "loss": 0.6146,
      "step": 22424
    },
    {
      "epoch": 7.08083359646353,
      "grad_norm": 5.690118312835693,
      "learning_rate": 3.588657334315687e-05,
      "loss": 0.5261,
      "step": 22425
    },
    {
      "epoch": 7.081149352699716,
      "grad_norm": 5.9520649909973145,
      "learning_rate": 3.588569624250079e-05,
      "loss": 0.4326,
      "step": 22426
    },
    {
      "epoch": 7.081465108935902,
      "grad_norm": 5.977876663208008,
      "learning_rate": 3.5884819141844725e-05,
      "loss": 0.4784,
      "step": 22427
    },
    {
      "epoch": 7.081780865172087,
      "grad_norm": 5.019123554229736,
      "learning_rate": 3.588394204118865e-05,
      "loss": 0.4246,
      "step": 22428
    },
    {
      "epoch": 7.082096621408273,
      "grad_norm": 8.287189483642578,
      "learning_rate": 3.588306494053258e-05,
      "loss": 0.4745,
      "step": 22429
    },
    {
      "epoch": 7.0824123776444585,
      "grad_norm": 5.58215856552124,
      "learning_rate": 3.5882187839876506e-05,
      "loss": 0.4398,
      "step": 22430
    },
    {
      "epoch": 7.082728133880644,
      "grad_norm": 4.620900630950928,
      "learning_rate": 3.588131073922044e-05,
      "loss": 0.4127,
      "step": 22431
    },
    {
      "epoch": 7.08304389011683,
      "grad_norm": 5.842019557952881,
      "learning_rate": 3.588043363856436e-05,
      "loss": 0.4608,
      "step": 22432
    },
    {
      "epoch": 7.083359646353015,
      "grad_norm": 6.038783073425293,
      "learning_rate": 3.5879556537908294e-05,
      "loss": 0.4344,
      "step": 22433
    },
    {
      "epoch": 7.0836754025892015,
      "grad_norm": 5.044764995574951,
      "learning_rate": 3.587867943725222e-05,
      "loss": 0.4025,
      "step": 22434
    },
    {
      "epoch": 7.083991158825387,
      "grad_norm": 5.809607982635498,
      "learning_rate": 3.587780233659615e-05,
      "loss": 0.4823,
      "step": 22435
    },
    {
      "epoch": 7.084306915061572,
      "grad_norm": 4.8887457847595215,
      "learning_rate": 3.5876925235940076e-05,
      "loss": 0.3598,
      "step": 22436
    },
    {
      "epoch": 7.084622671297758,
      "grad_norm": 5.095202445983887,
      "learning_rate": 3.587604813528401e-05,
      "loss": 0.4386,
      "step": 22437
    },
    {
      "epoch": 7.0849384275339435,
      "grad_norm": 5.7813544273376465,
      "learning_rate": 3.587517103462793e-05,
      "loss": 0.5187,
      "step": 22438
    },
    {
      "epoch": 7.08525418377013,
      "grad_norm": 6.813140869140625,
      "learning_rate": 3.5874293933971864e-05,
      "loss": 0.5355,
      "step": 22439
    },
    {
      "epoch": 7.085569940006315,
      "grad_norm": 5.285849094390869,
      "learning_rate": 3.5873416833315796e-05,
      "loss": 0.453,
      "step": 22440
    },
    {
      "epoch": 7.085885696242501,
      "grad_norm": 7.3211493492126465,
      "learning_rate": 3.587253973265972e-05,
      "loss": 0.5082,
      "step": 22441
    },
    {
      "epoch": 7.0862014524786865,
      "grad_norm": 5.77404260635376,
      "learning_rate": 3.587166263200365e-05,
      "loss": 0.5092,
      "step": 22442
    },
    {
      "epoch": 7.086517208714872,
      "grad_norm": 5.135067462921143,
      "learning_rate": 3.587078553134758e-05,
      "loss": 0.4897,
      "step": 22443
    },
    {
      "epoch": 7.086832964951058,
      "grad_norm": 6.189635753631592,
      "learning_rate": 3.586990843069151e-05,
      "loss": 0.5154,
      "step": 22444
    },
    {
      "epoch": 7.087148721187243,
      "grad_norm": 5.0377278327941895,
      "learning_rate": 3.5869031330035434e-05,
      "loss": 0.4629,
      "step": 22445
    },
    {
      "epoch": 7.087464477423429,
      "grad_norm": 5.628221035003662,
      "learning_rate": 3.5868154229379366e-05,
      "loss": 0.4292,
      "step": 22446
    },
    {
      "epoch": 7.087780233659615,
      "grad_norm": 4.944143295288086,
      "learning_rate": 3.586727712872329e-05,
      "loss": 0.4764,
      "step": 22447
    },
    {
      "epoch": 7.0880959898958,
      "grad_norm": 4.268307209014893,
      "learning_rate": 3.586640002806722e-05,
      "loss": 0.2895,
      "step": 22448
    },
    {
      "epoch": 7.088411746131986,
      "grad_norm": 5.3330397605896,
      "learning_rate": 3.586552292741115e-05,
      "loss": 0.4795,
      "step": 22449
    },
    {
      "epoch": 7.0887275023681715,
      "grad_norm": 6.095956802368164,
      "learning_rate": 3.586464582675508e-05,
      "loss": 0.469,
      "step": 22450
    },
    {
      "epoch": 7.089043258604358,
      "grad_norm": 6.028151512145996,
      "learning_rate": 3.5863768726099004e-05,
      "loss": 0.4525,
      "step": 22451
    },
    {
      "epoch": 7.089359014840543,
      "grad_norm": 6.474027156829834,
      "learning_rate": 3.5862891625442935e-05,
      "loss": 0.501,
      "step": 22452
    },
    {
      "epoch": 7.089674771076729,
      "grad_norm": 6.783787727355957,
      "learning_rate": 3.586201452478687e-05,
      "loss": 0.4914,
      "step": 22453
    },
    {
      "epoch": 7.0899905273129145,
      "grad_norm": 4.502157211303711,
      "learning_rate": 3.58611374241308e-05,
      "loss": 0.4332,
      "step": 22454
    },
    {
      "epoch": 7.0903062835491,
      "grad_norm": 5.61667013168335,
      "learning_rate": 3.5860260323474724e-05,
      "loss": 0.3666,
      "step": 22455
    },
    {
      "epoch": 7.090622039785286,
      "grad_norm": 5.084000587463379,
      "learning_rate": 3.5859383222818655e-05,
      "loss": 0.4215,
      "step": 22456
    },
    {
      "epoch": 7.090937796021471,
      "grad_norm": 5.000969886779785,
      "learning_rate": 3.585850612216258e-05,
      "loss": 0.441,
      "step": 22457
    },
    {
      "epoch": 7.091253552257657,
      "grad_norm": 5.9274187088012695,
      "learning_rate": 3.585762902150651e-05,
      "loss": 0.4805,
      "step": 22458
    },
    {
      "epoch": 7.091569308493843,
      "grad_norm": 6.806347370147705,
      "learning_rate": 3.585675192085044e-05,
      "loss": 0.5035,
      "step": 22459
    },
    {
      "epoch": 7.091885064730028,
      "grad_norm": 7.1001458168029785,
      "learning_rate": 3.585587482019437e-05,
      "loss": 0.471,
      "step": 22460
    },
    {
      "epoch": 7.092200820966214,
      "grad_norm": 4.422188758850098,
      "learning_rate": 3.58549977195383e-05,
      "loss": 0.458,
      "step": 22461
    },
    {
      "epoch": 7.0925165772023995,
      "grad_norm": 5.363656997680664,
      "learning_rate": 3.5854120618882225e-05,
      "loss": 0.4689,
      "step": 22462
    },
    {
      "epoch": 7.092832333438586,
      "grad_norm": 4.735030651092529,
      "learning_rate": 3.585324351822616e-05,
      "loss": 0.4986,
      "step": 22463
    },
    {
      "epoch": 7.093148089674771,
      "grad_norm": 5.025501251220703,
      "learning_rate": 3.585236641757008e-05,
      "loss": 0.4714,
      "step": 22464
    },
    {
      "epoch": 7.093463845910957,
      "grad_norm": 5.525783538818359,
      "learning_rate": 3.585148931691401e-05,
      "loss": 0.4315,
      "step": 22465
    },
    {
      "epoch": 7.093779602147142,
      "grad_norm": 4.8275370597839355,
      "learning_rate": 3.585061221625794e-05,
      "loss": 0.4266,
      "step": 22466
    },
    {
      "epoch": 7.094095358383328,
      "grad_norm": 5.638674259185791,
      "learning_rate": 3.584973511560187e-05,
      "loss": 0.4728,
      "step": 22467
    },
    {
      "epoch": 7.094411114619514,
      "grad_norm": 5.954931735992432,
      "learning_rate": 3.5848858014945795e-05,
      "loss": 0.4918,
      "step": 22468
    },
    {
      "epoch": 7.094726870855699,
      "grad_norm": 6.051211833953857,
      "learning_rate": 3.5847980914289726e-05,
      "loss": 0.4509,
      "step": 22469
    },
    {
      "epoch": 7.095042627091885,
      "grad_norm": 5.006488800048828,
      "learning_rate": 3.584710381363365e-05,
      "loss": 0.4715,
      "step": 22470
    },
    {
      "epoch": 7.095358383328071,
      "grad_norm": 5.932934284210205,
      "learning_rate": 3.584622671297758e-05,
      "loss": 0.5658,
      "step": 22471
    },
    {
      "epoch": 7.095674139564256,
      "grad_norm": 4.14774751663208,
      "learning_rate": 3.584534961232151e-05,
      "loss": 0.4129,
      "step": 22472
    },
    {
      "epoch": 7.095989895800442,
      "grad_norm": 5.902124404907227,
      "learning_rate": 3.584447251166544e-05,
      "loss": 0.5107,
      "step": 22473
    },
    {
      "epoch": 7.096305652036627,
      "grad_norm": 5.445435047149658,
      "learning_rate": 3.5843595411009365e-05,
      "loss": 0.4658,
      "step": 22474
    },
    {
      "epoch": 7.096621408272814,
      "grad_norm": 4.9049882888793945,
      "learning_rate": 3.5842718310353296e-05,
      "loss": 0.4848,
      "step": 22475
    },
    {
      "epoch": 7.096937164508999,
      "grad_norm": 3.906215190887451,
      "learning_rate": 3.584184120969722e-05,
      "loss": 0.3698,
      "step": 22476
    },
    {
      "epoch": 7.097252920745185,
      "grad_norm": 5.891743183135986,
      "learning_rate": 3.584096410904115e-05,
      "loss": 0.5584,
      "step": 22477
    },
    {
      "epoch": 7.09756867698137,
      "grad_norm": 4.99963903427124,
      "learning_rate": 3.5840087008385084e-05,
      "loss": 0.4445,
      "step": 22478
    },
    {
      "epoch": 7.097884433217556,
      "grad_norm": 4.86058235168457,
      "learning_rate": 3.583920990772901e-05,
      "loss": 0.4133,
      "step": 22479
    },
    {
      "epoch": 7.098200189453742,
      "grad_norm": 5.144009113311768,
      "learning_rate": 3.583833280707294e-05,
      "loss": 0.4795,
      "step": 22480
    },
    {
      "epoch": 7.098515945689927,
      "grad_norm": 5.41950798034668,
      "learning_rate": 3.583745570641687e-05,
      "loss": 0.5009,
      "step": 22481
    },
    {
      "epoch": 7.098831701926113,
      "grad_norm": 5.689339637756348,
      "learning_rate": 3.5836578605760804e-05,
      "loss": 0.4409,
      "step": 22482
    },
    {
      "epoch": 7.099147458162299,
      "grad_norm": 5.302387237548828,
      "learning_rate": 3.583570150510473e-05,
      "loss": 0.516,
      "step": 22483
    },
    {
      "epoch": 7.099463214398484,
      "grad_norm": 4.341753005981445,
      "learning_rate": 3.583482440444866e-05,
      "loss": 0.4031,
      "step": 22484
    },
    {
      "epoch": 7.09977897063467,
      "grad_norm": 5.815405368804932,
      "learning_rate": 3.5833947303792586e-05,
      "loss": 0.4489,
      "step": 22485
    },
    {
      "epoch": 7.100094726870855,
      "grad_norm": 4.54805326461792,
      "learning_rate": 3.583307020313652e-05,
      "loss": 0.3473,
      "step": 22486
    },
    {
      "epoch": 7.100410483107042,
      "grad_norm": 5.7638936042785645,
      "learning_rate": 3.583219310248044e-05,
      "loss": 0.4403,
      "step": 22487
    },
    {
      "epoch": 7.100726239343227,
      "grad_norm": 6.046411037445068,
      "learning_rate": 3.5831316001824374e-05,
      "loss": 0.5101,
      "step": 22488
    },
    {
      "epoch": 7.101041995579413,
      "grad_norm": 5.701701641082764,
      "learning_rate": 3.58304389011683e-05,
      "loss": 0.4466,
      "step": 22489
    },
    {
      "epoch": 7.101357751815598,
      "grad_norm": 4.6068220138549805,
      "learning_rate": 3.582956180051223e-05,
      "loss": 0.4275,
      "step": 22490
    },
    {
      "epoch": 7.101673508051784,
      "grad_norm": 4.826099395751953,
      "learning_rate": 3.5828684699856156e-05,
      "loss": 0.459,
      "step": 22491
    },
    {
      "epoch": 7.10198926428797,
      "grad_norm": 5.711246013641357,
      "learning_rate": 3.582780759920009e-05,
      "loss": 0.4117,
      "step": 22492
    },
    {
      "epoch": 7.102305020524155,
      "grad_norm": 5.4856038093566895,
      "learning_rate": 3.582693049854401e-05,
      "loss": 0.3372,
      "step": 22493
    },
    {
      "epoch": 7.102620776760341,
      "grad_norm": 6.183712959289551,
      "learning_rate": 3.5826053397887944e-05,
      "loss": 0.4112,
      "step": 22494
    },
    {
      "epoch": 7.102936532996527,
      "grad_norm": 5.170934200286865,
      "learning_rate": 3.582517629723187e-05,
      "loss": 0.4053,
      "step": 22495
    },
    {
      "epoch": 7.103252289232712,
      "grad_norm": 5.139697551727295,
      "learning_rate": 3.58242991965758e-05,
      "loss": 0.5043,
      "step": 22496
    },
    {
      "epoch": 7.103568045468898,
      "grad_norm": 5.360268592834473,
      "learning_rate": 3.5823422095919725e-05,
      "loss": 0.4446,
      "step": 22497
    },
    {
      "epoch": 7.103883801705083,
      "grad_norm": 4.98378849029541,
      "learning_rate": 3.582254499526366e-05,
      "loss": 0.3841,
      "step": 22498
    },
    {
      "epoch": 7.10419955794127,
      "grad_norm": 4.8479323387146,
      "learning_rate": 3.582166789460759e-05,
      "loss": 0.5022,
      "step": 22499
    },
    {
      "epoch": 7.104515314177455,
      "grad_norm": 5.992180347442627,
      "learning_rate": 3.5820790793951514e-05,
      "loss": 0.4481,
      "step": 22500
    },
    {
      "epoch": 7.104831070413641,
      "grad_norm": 5.253616809844971,
      "learning_rate": 3.5819913693295445e-05,
      "loss": 0.4144,
      "step": 22501
    },
    {
      "epoch": 7.105146826649826,
      "grad_norm": 5.263510227203369,
      "learning_rate": 3.581903659263937e-05,
      "loss": 0.4968,
      "step": 22502
    },
    {
      "epoch": 7.105462582886012,
      "grad_norm": 5.192197799682617,
      "learning_rate": 3.58181594919833e-05,
      "loss": 0.4737,
      "step": 22503
    },
    {
      "epoch": 7.105778339122198,
      "grad_norm": 5.483333110809326,
      "learning_rate": 3.581728239132723e-05,
      "loss": 0.4027,
      "step": 22504
    },
    {
      "epoch": 7.106094095358383,
      "grad_norm": 4.938979625701904,
      "learning_rate": 3.581640529067116e-05,
      "loss": 0.4531,
      "step": 22505
    },
    {
      "epoch": 7.106409851594569,
      "grad_norm": 5.63877534866333,
      "learning_rate": 3.5815528190015083e-05,
      "loss": 0.545,
      "step": 22506
    },
    {
      "epoch": 7.106725607830755,
      "grad_norm": 6.8120927810668945,
      "learning_rate": 3.581465108935902e-05,
      "loss": 0.4662,
      "step": 22507
    },
    {
      "epoch": 7.107041364066941,
      "grad_norm": 6.199398040771484,
      "learning_rate": 3.581377398870295e-05,
      "loss": 0.425,
      "step": 22508
    },
    {
      "epoch": 7.107357120303126,
      "grad_norm": 5.8281025886535645,
      "learning_rate": 3.581289688804688e-05,
      "loss": 0.454,
      "step": 22509
    },
    {
      "epoch": 7.107672876539311,
      "grad_norm": 5.770891189575195,
      "learning_rate": 3.5812019787390803e-05,
      "loss": 0.4739,
      "step": 22510
    },
    {
      "epoch": 7.1079886327754975,
      "grad_norm": 5.915074348449707,
      "learning_rate": 3.5811142686734735e-05,
      "loss": 0.5448,
      "step": 22511
    },
    {
      "epoch": 7.108304389011683,
      "grad_norm": 5.178533554077148,
      "learning_rate": 3.581026558607866e-05,
      "loss": 0.4663,
      "step": 22512
    },
    {
      "epoch": 7.108620145247869,
      "grad_norm": 4.812094688415527,
      "learning_rate": 3.580938848542259e-05,
      "loss": 0.3464,
      "step": 22513
    },
    {
      "epoch": 7.108935901484054,
      "grad_norm": 4.68075704574585,
      "learning_rate": 3.580851138476652e-05,
      "loss": 0.352,
      "step": 22514
    },
    {
      "epoch": 7.10925165772024,
      "grad_norm": 4.75660514831543,
      "learning_rate": 3.580763428411045e-05,
      "loss": 0.4431,
      "step": 22515
    },
    {
      "epoch": 7.109567413956426,
      "grad_norm": 5.631868362426758,
      "learning_rate": 3.580675718345437e-05,
      "loss": 0.5086,
      "step": 22516
    },
    {
      "epoch": 7.109883170192611,
      "grad_norm": 4.99567985534668,
      "learning_rate": 3.5805880082798305e-05,
      "loss": 0.4182,
      "step": 22517
    },
    {
      "epoch": 7.110198926428797,
      "grad_norm": 5.059821128845215,
      "learning_rate": 3.580500298214223e-05,
      "loss": 0.4278,
      "step": 22518
    },
    {
      "epoch": 7.1105146826649825,
      "grad_norm": 5.782153606414795,
      "learning_rate": 3.580412588148616e-05,
      "loss": 0.4909,
      "step": 22519
    },
    {
      "epoch": 7.110830438901169,
      "grad_norm": 6.027443885803223,
      "learning_rate": 3.580324878083009e-05,
      "loss": 0.3884,
      "step": 22520
    },
    {
      "epoch": 7.111146195137354,
      "grad_norm": 5.8474907875061035,
      "learning_rate": 3.580237168017402e-05,
      "loss": 0.4336,
      "step": 22521
    },
    {
      "epoch": 7.111461951373539,
      "grad_norm": 7.212499141693115,
      "learning_rate": 3.580149457951795e-05,
      "loss": 0.4455,
      "step": 22522
    },
    {
      "epoch": 7.1117777076097255,
      "grad_norm": 4.5240864753723145,
      "learning_rate": 3.5800617478861875e-05,
      "loss": 0.4647,
      "step": 22523
    },
    {
      "epoch": 7.112093463845911,
      "grad_norm": 4.971889972686768,
      "learning_rate": 3.5799740378205806e-05,
      "loss": 0.4656,
      "step": 22524
    },
    {
      "epoch": 7.112409220082097,
      "grad_norm": 6.0788726806640625,
      "learning_rate": 3.579886327754973e-05,
      "loss": 0.497,
      "step": 22525
    },
    {
      "epoch": 7.112724976318282,
      "grad_norm": 7.1846418380737305,
      "learning_rate": 3.579798617689366e-05,
      "loss": 0.4709,
      "step": 22526
    },
    {
      "epoch": 7.113040732554468,
      "grad_norm": 6.146310806274414,
      "learning_rate": 3.579710907623759e-05,
      "loss": 0.5169,
      "step": 22527
    },
    {
      "epoch": 7.113356488790654,
      "grad_norm": 4.410887718200684,
      "learning_rate": 3.579623197558152e-05,
      "loss": 0.4522,
      "step": 22528
    },
    {
      "epoch": 7.113672245026839,
      "grad_norm": 5.133411407470703,
      "learning_rate": 3.5795354874925444e-05,
      "loss": 0.5232,
      "step": 22529
    },
    {
      "epoch": 7.113988001263025,
      "grad_norm": 6.325957298278809,
      "learning_rate": 3.5794477774269376e-05,
      "loss": 0.4503,
      "step": 22530
    },
    {
      "epoch": 7.1143037574992105,
      "grad_norm": 5.6483893394470215,
      "learning_rate": 3.57936006736133e-05,
      "loss": 0.4232,
      "step": 22531
    },
    {
      "epoch": 7.114619513735397,
      "grad_norm": 5.476306438446045,
      "learning_rate": 3.579272357295723e-05,
      "loss": 0.4591,
      "step": 22532
    },
    {
      "epoch": 7.114935269971582,
      "grad_norm": 5.503390312194824,
      "learning_rate": 3.579184647230116e-05,
      "loss": 0.4819,
      "step": 22533
    },
    {
      "epoch": 7.115251026207767,
      "grad_norm": 5.611538887023926,
      "learning_rate": 3.5790969371645096e-05,
      "loss": 0.528,
      "step": 22534
    },
    {
      "epoch": 7.1155667824439535,
      "grad_norm": 5.610389232635498,
      "learning_rate": 3.579009227098902e-05,
      "loss": 0.4652,
      "step": 22535
    },
    {
      "epoch": 7.115882538680139,
      "grad_norm": 5.739768981933594,
      "learning_rate": 3.578921517033295e-05,
      "loss": 0.5954,
      "step": 22536
    },
    {
      "epoch": 7.116198294916325,
      "grad_norm": 4.838098526000977,
      "learning_rate": 3.578833806967688e-05,
      "loss": 0.4501,
      "step": 22537
    },
    {
      "epoch": 7.11651405115251,
      "grad_norm": 5.1631669998168945,
      "learning_rate": 3.578746096902081e-05,
      "loss": 0.332,
      "step": 22538
    },
    {
      "epoch": 7.1168298073886955,
      "grad_norm": 6.04219388961792,
      "learning_rate": 3.5786583868364734e-05,
      "loss": 0.4839,
      "step": 22539
    },
    {
      "epoch": 7.117145563624882,
      "grad_norm": 4.6470561027526855,
      "learning_rate": 3.5785706767708666e-05,
      "loss": 0.3687,
      "step": 22540
    },
    {
      "epoch": 7.117461319861067,
      "grad_norm": 4.924928665161133,
      "learning_rate": 3.578482966705259e-05,
      "loss": 0.4593,
      "step": 22541
    },
    {
      "epoch": 7.117777076097253,
      "grad_norm": 4.638423919677734,
      "learning_rate": 3.578395256639652e-05,
      "loss": 0.4189,
      "step": 22542
    },
    {
      "epoch": 7.1180928323334385,
      "grad_norm": 5.53619909286499,
      "learning_rate": 3.5783075465740454e-05,
      "loss": 0.42,
      "step": 22543
    },
    {
      "epoch": 7.118408588569625,
      "grad_norm": 5.9631757736206055,
      "learning_rate": 3.578219836508438e-05,
      "loss": 0.4834,
      "step": 22544
    },
    {
      "epoch": 7.11872434480581,
      "grad_norm": 5.304795742034912,
      "learning_rate": 3.578132126442831e-05,
      "loss": 0.5008,
      "step": 22545
    },
    {
      "epoch": 7.119040101041995,
      "grad_norm": 7.342537879943848,
      "learning_rate": 3.5780444163772236e-05,
      "loss": 0.5928,
      "step": 22546
    },
    {
      "epoch": 7.119355857278181,
      "grad_norm": 5.1660356521606445,
      "learning_rate": 3.577956706311617e-05,
      "loss": 0.4508,
      "step": 22547
    },
    {
      "epoch": 7.119671613514367,
      "grad_norm": 5.951986312866211,
      "learning_rate": 3.577868996246009e-05,
      "loss": 0.3693,
      "step": 22548
    },
    {
      "epoch": 7.119987369750553,
      "grad_norm": 4.975528240203857,
      "learning_rate": 3.5777812861804024e-05,
      "loss": 0.3831,
      "step": 22549
    },
    {
      "epoch": 7.120303125986738,
      "grad_norm": 5.320348739624023,
      "learning_rate": 3.577693576114795e-05,
      "loss": 0.52,
      "step": 22550
    },
    {
      "epoch": 7.1206188822229235,
      "grad_norm": 6.537148952484131,
      "learning_rate": 3.577605866049188e-05,
      "loss": 0.5774,
      "step": 22551
    },
    {
      "epoch": 7.12093463845911,
      "grad_norm": 5.804482936859131,
      "learning_rate": 3.5775181559835805e-05,
      "loss": 0.4681,
      "step": 22552
    },
    {
      "epoch": 7.121250394695295,
      "grad_norm": 5.434585094451904,
      "learning_rate": 3.577430445917974e-05,
      "loss": 0.467,
      "step": 22553
    },
    {
      "epoch": 7.121566150931481,
      "grad_norm": 5.424333572387695,
      "learning_rate": 3.577342735852366e-05,
      "loss": 0.3809,
      "step": 22554
    },
    {
      "epoch": 7.1218819071676664,
      "grad_norm": 5.695471286773682,
      "learning_rate": 3.5772550257867594e-05,
      "loss": 0.4436,
      "step": 22555
    },
    {
      "epoch": 7.122197663403853,
      "grad_norm": 7.112077236175537,
      "learning_rate": 3.577167315721152e-05,
      "loss": 0.4985,
      "step": 22556
    },
    {
      "epoch": 7.122513419640038,
      "grad_norm": 5.097160339355469,
      "learning_rate": 3.577079605655545e-05,
      "loss": 0.4931,
      "step": 22557
    },
    {
      "epoch": 7.122829175876223,
      "grad_norm": 7.015291213989258,
      "learning_rate": 3.576991895589938e-05,
      "loss": 0.4568,
      "step": 22558
    },
    {
      "epoch": 7.123144932112409,
      "grad_norm": 8.10346508026123,
      "learning_rate": 3.576904185524331e-05,
      "loss": 0.4399,
      "step": 22559
    },
    {
      "epoch": 7.123460688348595,
      "grad_norm": 6.358394145965576,
      "learning_rate": 3.576816475458724e-05,
      "loss": 0.4205,
      "step": 22560
    },
    {
      "epoch": 7.123776444584781,
      "grad_norm": 4.70474910736084,
      "learning_rate": 3.576728765393116e-05,
      "loss": 0.4627,
      "step": 22561
    },
    {
      "epoch": 7.124092200820966,
      "grad_norm": 6.046079635620117,
      "learning_rate": 3.5766410553275095e-05,
      "loss": 0.4588,
      "step": 22562
    },
    {
      "epoch": 7.1244079570571515,
      "grad_norm": 7.560370445251465,
      "learning_rate": 3.576553345261903e-05,
      "loss": 0.47,
      "step": 22563
    },
    {
      "epoch": 7.124723713293338,
      "grad_norm": 5.379199504852295,
      "learning_rate": 3.576465635196296e-05,
      "loss": 0.4469,
      "step": 22564
    },
    {
      "epoch": 7.125039469529523,
      "grad_norm": 4.95309591293335,
      "learning_rate": 3.576377925130688e-05,
      "loss": 0.5043,
      "step": 22565
    },
    {
      "epoch": 7.125355225765709,
      "grad_norm": 5.776133060455322,
      "learning_rate": 3.5762902150650815e-05,
      "loss": 0.4269,
      "step": 22566
    },
    {
      "epoch": 7.125670982001894,
      "grad_norm": 5.18507194519043,
      "learning_rate": 3.576202504999474e-05,
      "loss": 0.525,
      "step": 22567
    },
    {
      "epoch": 7.125986738238081,
      "grad_norm": 4.761876106262207,
      "learning_rate": 3.576114794933867e-05,
      "loss": 0.4147,
      "step": 22568
    },
    {
      "epoch": 7.126302494474266,
      "grad_norm": 5.312589645385742,
      "learning_rate": 3.5760270848682597e-05,
      "loss": 0.5063,
      "step": 22569
    },
    {
      "epoch": 7.126618250710451,
      "grad_norm": 4.276181697845459,
      "learning_rate": 3.575939374802653e-05,
      "loss": 0.4371,
      "step": 22570
    },
    {
      "epoch": 7.126934006946637,
      "grad_norm": 5.231050491333008,
      "learning_rate": 3.575851664737045e-05,
      "loss": 0.4513,
      "step": 22571
    },
    {
      "epoch": 7.127249763182823,
      "grad_norm": 7.405262470245361,
      "learning_rate": 3.5757639546714385e-05,
      "loss": 0.5847,
      "step": 22572
    },
    {
      "epoch": 7.127565519419009,
      "grad_norm": 4.882863521575928,
      "learning_rate": 3.575676244605831e-05,
      "loss": 0.3809,
      "step": 22573
    },
    {
      "epoch": 7.127881275655194,
      "grad_norm": 6.989027500152588,
      "learning_rate": 3.575588534540224e-05,
      "loss": 0.572,
      "step": 22574
    },
    {
      "epoch": 7.128197031891379,
      "grad_norm": 4.398602485656738,
      "learning_rate": 3.5755008244746166e-05,
      "loss": 0.4365,
      "step": 22575
    },
    {
      "epoch": 7.128512788127566,
      "grad_norm": 4.354330539703369,
      "learning_rate": 3.57541311440901e-05,
      "loss": 0.4644,
      "step": 22576
    },
    {
      "epoch": 7.128828544363751,
      "grad_norm": 5.863006591796875,
      "learning_rate": 3.575325404343402e-05,
      "loss": 0.481,
      "step": 22577
    },
    {
      "epoch": 7.129144300599937,
      "grad_norm": 5.987093925476074,
      "learning_rate": 3.5752376942777955e-05,
      "loss": 0.4363,
      "step": 22578
    },
    {
      "epoch": 7.129460056836122,
      "grad_norm": 6.025120735168457,
      "learning_rate": 3.575149984212188e-05,
      "loss": 0.4702,
      "step": 22579
    },
    {
      "epoch": 7.129775813072309,
      "grad_norm": 4.563098430633545,
      "learning_rate": 3.575062274146581e-05,
      "loss": 0.4439,
      "step": 22580
    },
    {
      "epoch": 7.130091569308494,
      "grad_norm": 5.648772716522217,
      "learning_rate": 3.574974564080974e-05,
      "loss": 0.4354,
      "step": 22581
    },
    {
      "epoch": 7.130407325544679,
      "grad_norm": 6.001322269439697,
      "learning_rate": 3.574886854015367e-05,
      "loss": 0.4646,
      "step": 22582
    },
    {
      "epoch": 7.130723081780865,
      "grad_norm": 3.8447930812835693,
      "learning_rate": 3.57479914394976e-05,
      "loss": 0.4235,
      "step": 22583
    },
    {
      "epoch": 7.131038838017051,
      "grad_norm": 4.988368988037109,
      "learning_rate": 3.5747114338841524e-05,
      "loss": 0.4827,
      "step": 22584
    },
    {
      "epoch": 7.131354594253237,
      "grad_norm": 3.9988698959350586,
      "learning_rate": 3.5746237238185456e-05,
      "loss": 0.3931,
      "step": 22585
    },
    {
      "epoch": 7.131670350489422,
      "grad_norm": 5.012807846069336,
      "learning_rate": 3.574536013752938e-05,
      "loss": 0.4572,
      "step": 22586
    },
    {
      "epoch": 7.131986106725607,
      "grad_norm": 6.2266435623168945,
      "learning_rate": 3.574448303687331e-05,
      "loss": 0.4863,
      "step": 22587
    },
    {
      "epoch": 7.132301862961794,
      "grad_norm": 5.6272478103637695,
      "learning_rate": 3.574360593621724e-05,
      "loss": 0.4669,
      "step": 22588
    },
    {
      "epoch": 7.132617619197979,
      "grad_norm": 6.70484733581543,
      "learning_rate": 3.5742728835561176e-05,
      "loss": 0.547,
      "step": 22589
    },
    {
      "epoch": 7.132933375434165,
      "grad_norm": 7.081356525421143,
      "learning_rate": 3.57418517349051e-05,
      "loss": 0.4786,
      "step": 22590
    },
    {
      "epoch": 7.13324913167035,
      "grad_norm": 4.954072952270508,
      "learning_rate": 3.574097463424903e-05,
      "loss": 0.4597,
      "step": 22591
    },
    {
      "epoch": 7.1335648879065365,
      "grad_norm": 6.521296501159668,
      "learning_rate": 3.574009753359296e-05,
      "loss": 0.4256,
      "step": 22592
    },
    {
      "epoch": 7.133880644142722,
      "grad_norm": 6.451456546783447,
      "learning_rate": 3.573922043293689e-05,
      "loss": 0.5609,
      "step": 22593
    },
    {
      "epoch": 7.134196400378907,
      "grad_norm": 6.365875720977783,
      "learning_rate": 3.5738343332280814e-05,
      "loss": 0.4863,
      "step": 22594
    },
    {
      "epoch": 7.134512156615093,
      "grad_norm": 5.8962321281433105,
      "learning_rate": 3.5737466231624746e-05,
      "loss": 0.4785,
      "step": 22595
    },
    {
      "epoch": 7.134827912851279,
      "grad_norm": 4.815359115600586,
      "learning_rate": 3.573658913096867e-05,
      "loss": 0.4454,
      "step": 22596
    },
    {
      "epoch": 7.135143669087465,
      "grad_norm": 5.394891738891602,
      "learning_rate": 3.57357120303126e-05,
      "loss": 0.4209,
      "step": 22597
    },
    {
      "epoch": 7.13545942532365,
      "grad_norm": 5.30246114730835,
      "learning_rate": 3.573483492965653e-05,
      "loss": 0.4162,
      "step": 22598
    },
    {
      "epoch": 7.135775181559836,
      "grad_norm": 5.554043769836426,
      "learning_rate": 3.573395782900046e-05,
      "loss": 0.5766,
      "step": 22599
    },
    {
      "epoch": 7.1360909377960215,
      "grad_norm": 6.043001174926758,
      "learning_rate": 3.5733080728344384e-05,
      "loss": 0.4788,
      "step": 22600
    },
    {
      "epoch": 7.1360909377960215,
      "eval_loss": 0.9226821064949036,
      "eval_runtime": 4.5305,
      "eval_samples_per_second": 549.613,
      "eval_steps_per_second": 4.635,
      "step": 22600
    },
    {
      "epoch": 7.136406694032207,
      "grad_norm": 5.138669490814209,
      "learning_rate": 3.5732203627688315e-05,
      "loss": 0.4711,
      "step": 22601
    },
    {
      "epoch": 7.136722450268393,
      "grad_norm": 5.603199005126953,
      "learning_rate": 3.573132652703225e-05,
      "loss": 0.4641,
      "step": 22602
    },
    {
      "epoch": 7.137038206504578,
      "grad_norm": 5.040750503540039,
      "learning_rate": 3.573044942637617e-05,
      "loss": 0.3995,
      "step": 22603
    },
    {
      "epoch": 7.1373539627407645,
      "grad_norm": 5.997057914733887,
      "learning_rate": 3.5729572325720104e-05,
      "loss": 0.4015,
      "step": 22604
    },
    {
      "epoch": 7.13766971897695,
      "grad_norm": 4.616168975830078,
      "learning_rate": 3.572869522506403e-05,
      "loss": 0.5001,
      "step": 22605
    },
    {
      "epoch": 7.137985475213135,
      "grad_norm": 5.386143207550049,
      "learning_rate": 3.572781812440796e-05,
      "loss": 0.4463,
      "step": 22606
    },
    {
      "epoch": 7.138301231449321,
      "grad_norm": 4.898292541503906,
      "learning_rate": 3.5726941023751885e-05,
      "loss": 0.4976,
      "step": 22607
    },
    {
      "epoch": 7.138616987685507,
      "grad_norm": 5.058248519897461,
      "learning_rate": 3.572606392309582e-05,
      "loss": 0.5184,
      "step": 22608
    },
    {
      "epoch": 7.138932743921693,
      "grad_norm": 5.937067031860352,
      "learning_rate": 3.572518682243974e-05,
      "loss": 0.4378,
      "step": 22609
    },
    {
      "epoch": 7.139248500157878,
      "grad_norm": 6.418275356292725,
      "learning_rate": 3.5724309721783674e-05,
      "loss": 0.477,
      "step": 22610
    },
    {
      "epoch": 7.139564256394063,
      "grad_norm": 5.076213359832764,
      "learning_rate": 3.57234326211276e-05,
      "loss": 0.5577,
      "step": 22611
    },
    {
      "epoch": 7.1398800126302495,
      "grad_norm": 5.948283672332764,
      "learning_rate": 3.572255552047153e-05,
      "loss": 0.4244,
      "step": 22612
    },
    {
      "epoch": 7.140195768866435,
      "grad_norm": 4.97308874130249,
      "learning_rate": 3.5721678419815455e-05,
      "loss": 0.4354,
      "step": 22613
    },
    {
      "epoch": 7.140511525102621,
      "grad_norm": 10.228257179260254,
      "learning_rate": 3.572080131915939e-05,
      "loss": 0.5925,
      "step": 22614
    },
    {
      "epoch": 7.140827281338806,
      "grad_norm": 5.083061218261719,
      "learning_rate": 3.571992421850331e-05,
      "loss": 0.451,
      "step": 22615
    },
    {
      "epoch": 7.1411430375749925,
      "grad_norm": 6.693957328796387,
      "learning_rate": 3.571904711784725e-05,
      "loss": 0.5049,
      "step": 22616
    },
    {
      "epoch": 7.141458793811178,
      "grad_norm": 5.628998756408691,
      "learning_rate": 3.5718170017191175e-05,
      "loss": 0.4251,
      "step": 22617
    },
    {
      "epoch": 7.141774550047363,
      "grad_norm": 4.418317794799805,
      "learning_rate": 3.571729291653511e-05,
      "loss": 0.535,
      "step": 22618
    },
    {
      "epoch": 7.142090306283549,
      "grad_norm": 6.020055770874023,
      "learning_rate": 3.571641581587903e-05,
      "loss": 0.3887,
      "step": 22619
    },
    {
      "epoch": 7.1424060625197345,
      "grad_norm": 6.869390964508057,
      "learning_rate": 3.571553871522296e-05,
      "loss": 0.4863,
      "step": 22620
    },
    {
      "epoch": 7.142721818755921,
      "grad_norm": 5.61269474029541,
      "learning_rate": 3.571466161456689e-05,
      "loss": 0.5019,
      "step": 22621
    },
    {
      "epoch": 7.143037574992106,
      "grad_norm": 5.890473365783691,
      "learning_rate": 3.571378451391082e-05,
      "loss": 0.5196,
      "step": 22622
    },
    {
      "epoch": 7.143353331228292,
      "grad_norm": 8.53084945678711,
      "learning_rate": 3.5712907413254745e-05,
      "loss": 0.6181,
      "step": 22623
    },
    {
      "epoch": 7.1436690874644775,
      "grad_norm": 5.3287200927734375,
      "learning_rate": 3.5712030312598676e-05,
      "loss": 0.5146,
      "step": 22624
    },
    {
      "epoch": 7.143984843700663,
      "grad_norm": 6.0062127113342285,
      "learning_rate": 3.571115321194261e-05,
      "loss": 0.3852,
      "step": 22625
    },
    {
      "epoch": 7.144300599936849,
      "grad_norm": 5.868476390838623,
      "learning_rate": 3.571027611128653e-05,
      "loss": 0.4791,
      "step": 22626
    },
    {
      "epoch": 7.144616356173034,
      "grad_norm": 3.74953556060791,
      "learning_rate": 3.5709399010630465e-05,
      "loss": 0.3614,
      "step": 22627
    },
    {
      "epoch": 7.14493211240922,
      "grad_norm": 6.332742214202881,
      "learning_rate": 3.570852190997439e-05,
      "loss": 0.4923,
      "step": 22628
    },
    {
      "epoch": 7.145247868645406,
      "grad_norm": 4.85554838180542,
      "learning_rate": 3.570764480931832e-05,
      "loss": 0.4301,
      "step": 22629
    },
    {
      "epoch": 7.145563624881591,
      "grad_norm": 4.567058086395264,
      "learning_rate": 3.5706767708662246e-05,
      "loss": 0.4268,
      "step": 22630
    },
    {
      "epoch": 7.145879381117777,
      "grad_norm": 5.2620439529418945,
      "learning_rate": 3.570589060800618e-05,
      "loss": 0.4645,
      "step": 22631
    },
    {
      "epoch": 7.1461951373539625,
      "grad_norm": 5.6009087562561035,
      "learning_rate": 3.57050135073501e-05,
      "loss": 0.492,
      "step": 22632
    },
    {
      "epoch": 7.146510893590149,
      "grad_norm": 5.489243030548096,
      "learning_rate": 3.5704136406694034e-05,
      "loss": 0.4923,
      "step": 22633
    },
    {
      "epoch": 7.146826649826334,
      "grad_norm": 6.065682888031006,
      "learning_rate": 3.570325930603796e-05,
      "loss": 0.5502,
      "step": 22634
    },
    {
      "epoch": 7.14714240606252,
      "grad_norm": 4.359348297119141,
      "learning_rate": 3.570238220538189e-05,
      "loss": 0.4027,
      "step": 22635
    },
    {
      "epoch": 7.1474581622987055,
      "grad_norm": 5.207051753997803,
      "learning_rate": 3.5701505104725816e-05,
      "loss": 0.4982,
      "step": 22636
    },
    {
      "epoch": 7.147773918534891,
      "grad_norm": 5.761163234710693,
      "learning_rate": 3.570062800406975e-05,
      "loss": 0.425,
      "step": 22637
    },
    {
      "epoch": 7.148089674771077,
      "grad_norm": 4.674661159515381,
      "learning_rate": 3.569975090341367e-05,
      "loss": 0.4097,
      "step": 22638
    },
    {
      "epoch": 7.148405431007262,
      "grad_norm": 5.402137756347656,
      "learning_rate": 3.5698873802757604e-05,
      "loss": 0.512,
      "step": 22639
    },
    {
      "epoch": 7.148721187243448,
      "grad_norm": 6.41784143447876,
      "learning_rate": 3.5697996702101536e-05,
      "loss": 0.4872,
      "step": 22640
    },
    {
      "epoch": 7.149036943479634,
      "grad_norm": 5.029235363006592,
      "learning_rate": 3.569711960144546e-05,
      "loss": 0.4465,
      "step": 22641
    },
    {
      "epoch": 7.149352699715819,
      "grad_norm": 5.918269157409668,
      "learning_rate": 3.569624250078939e-05,
      "loss": 0.4064,
      "step": 22642
    },
    {
      "epoch": 7.149668455952005,
      "grad_norm": 5.928116798400879,
      "learning_rate": 3.5695365400133324e-05,
      "loss": 0.5079,
      "step": 22643
    },
    {
      "epoch": 7.1499842121881905,
      "grad_norm": 4.76165771484375,
      "learning_rate": 3.569448829947725e-05,
      "loss": 0.4731,
      "step": 22644
    },
    {
      "epoch": 7.150299968424377,
      "grad_norm": 4.443127632141113,
      "learning_rate": 3.569361119882118e-05,
      "loss": 0.4173,
      "step": 22645
    },
    {
      "epoch": 7.150615724660562,
      "grad_norm": 7.387996673583984,
      "learning_rate": 3.569273409816511e-05,
      "loss": 0.4934,
      "step": 22646
    },
    {
      "epoch": 7.150931480896748,
      "grad_norm": 4.6947503089904785,
      "learning_rate": 3.569185699750904e-05,
      "loss": 0.3735,
      "step": 22647
    },
    {
      "epoch": 7.151247237132933,
      "grad_norm": 5.013918399810791,
      "learning_rate": 3.569097989685297e-05,
      "loss": 0.4602,
      "step": 22648
    },
    {
      "epoch": 7.151562993369119,
      "grad_norm": 5.490453720092773,
      "learning_rate": 3.5690102796196894e-05,
      "loss": 0.4778,
      "step": 22649
    },
    {
      "epoch": 7.151878749605305,
      "grad_norm": 5.663905620574951,
      "learning_rate": 3.5689225695540826e-05,
      "loss": 0.5597,
      "step": 22650
    },
    {
      "epoch": 7.15219450584149,
      "grad_norm": 5.260549545288086,
      "learning_rate": 3.568834859488475e-05,
      "loss": 0.374,
      "step": 22651
    },
    {
      "epoch": 7.152510262077676,
      "grad_norm": 5.348943710327148,
      "learning_rate": 3.568747149422868e-05,
      "loss": 0.4862,
      "step": 22652
    },
    {
      "epoch": 7.152826018313862,
      "grad_norm": 4.439618110656738,
      "learning_rate": 3.568659439357261e-05,
      "loss": 0.4761,
      "step": 22653
    },
    {
      "epoch": 7.153141774550047,
      "grad_norm": 5.790163040161133,
      "learning_rate": 3.568571729291654e-05,
      "loss": 0.5926,
      "step": 22654
    },
    {
      "epoch": 7.153457530786233,
      "grad_norm": 6.85952091217041,
      "learning_rate": 3.5684840192260464e-05,
      "loss": 0.4716,
      "step": 22655
    },
    {
      "epoch": 7.153773287022418,
      "grad_norm": 4.488359451293945,
      "learning_rate": 3.5683963091604395e-05,
      "loss": 0.4015,
      "step": 22656
    },
    {
      "epoch": 7.154089043258605,
      "grad_norm": 5.608605861663818,
      "learning_rate": 3.568308599094832e-05,
      "loss": 0.4673,
      "step": 22657
    },
    {
      "epoch": 7.15440479949479,
      "grad_norm": 6.806878566741943,
      "learning_rate": 3.568220889029225e-05,
      "loss": 0.4782,
      "step": 22658
    },
    {
      "epoch": 7.154720555730976,
      "grad_norm": 5.821547985076904,
      "learning_rate": 3.568133178963618e-05,
      "loss": 0.4506,
      "step": 22659
    },
    {
      "epoch": 7.155036311967161,
      "grad_norm": 5.739768028259277,
      "learning_rate": 3.568045468898011e-05,
      "loss": 0.4283,
      "step": 22660
    },
    {
      "epoch": 7.155352068203347,
      "grad_norm": 5.180433750152588,
      "learning_rate": 3.567957758832404e-05,
      "loss": 0.4059,
      "step": 22661
    },
    {
      "epoch": 7.155667824439533,
      "grad_norm": 5.384390830993652,
      "learning_rate": 3.5678700487667965e-05,
      "loss": 0.3883,
      "step": 22662
    },
    {
      "epoch": 7.155983580675718,
      "grad_norm": 6.536964416503906,
      "learning_rate": 3.56778233870119e-05,
      "loss": 0.5387,
      "step": 22663
    },
    {
      "epoch": 7.156299336911904,
      "grad_norm": 5.590348720550537,
      "learning_rate": 3.567694628635582e-05,
      "loss": 0.5675,
      "step": 22664
    },
    {
      "epoch": 7.15661509314809,
      "grad_norm": 6.603347301483154,
      "learning_rate": 3.5676069185699753e-05,
      "loss": 0.4771,
      "step": 22665
    },
    {
      "epoch": 7.156930849384275,
      "grad_norm": 4.629136085510254,
      "learning_rate": 3.567519208504368e-05,
      "loss": 0.3743,
      "step": 22666
    },
    {
      "epoch": 7.157246605620461,
      "grad_norm": 6.026752948760986,
      "learning_rate": 3.567431498438761e-05,
      "loss": 0.4985,
      "step": 22667
    },
    {
      "epoch": 7.157562361856646,
      "grad_norm": 5.039644241333008,
      "learning_rate": 3.5673437883731535e-05,
      "loss": 0.4783,
      "step": 22668
    },
    {
      "epoch": 7.157878118092833,
      "grad_norm": 4.859165668487549,
      "learning_rate": 3.5672560783075467e-05,
      "loss": 0.5038,
      "step": 22669
    },
    {
      "epoch": 7.158193874329018,
      "grad_norm": 5.679750919342041,
      "learning_rate": 3.567168368241939e-05,
      "loss": 0.4609,
      "step": 22670
    },
    {
      "epoch": 7.158509630565204,
      "grad_norm": 5.401156425476074,
      "learning_rate": 3.567080658176333e-05,
      "loss": 0.4398,
      "step": 22671
    },
    {
      "epoch": 7.158825386801389,
      "grad_norm": 5.079134941101074,
      "learning_rate": 3.5669929481107255e-05,
      "loss": 0.5214,
      "step": 22672
    },
    {
      "epoch": 7.159141143037575,
      "grad_norm": 6.49338436126709,
      "learning_rate": 3.5669052380451187e-05,
      "loss": 0.6041,
      "step": 22673
    },
    {
      "epoch": 7.159456899273761,
      "grad_norm": 5.981334209442139,
      "learning_rate": 3.566817527979511e-05,
      "loss": 0.4322,
      "step": 22674
    },
    {
      "epoch": 7.159772655509946,
      "grad_norm": 4.905630111694336,
      "learning_rate": 3.566729817913904e-05,
      "loss": 0.4362,
      "step": 22675
    },
    {
      "epoch": 7.160088411746132,
      "grad_norm": 5.608448028564453,
      "learning_rate": 3.566642107848297e-05,
      "loss": 0.3621,
      "step": 22676
    },
    {
      "epoch": 7.160404167982318,
      "grad_norm": 5.315598011016846,
      "learning_rate": 3.56655439778269e-05,
      "loss": 0.4603,
      "step": 22677
    },
    {
      "epoch": 7.160719924218503,
      "grad_norm": 4.543768882751465,
      "learning_rate": 3.5664666877170825e-05,
      "loss": 0.4605,
      "step": 22678
    },
    {
      "epoch": 7.161035680454689,
      "grad_norm": 5.702667236328125,
      "learning_rate": 3.5663789776514756e-05,
      "loss": 0.4668,
      "step": 22679
    },
    {
      "epoch": 7.161351436690874,
      "grad_norm": 5.687383651733398,
      "learning_rate": 3.566291267585868e-05,
      "loss": 0.4494,
      "step": 22680
    },
    {
      "epoch": 7.1616671929270606,
      "grad_norm": 6.100212097167969,
      "learning_rate": 3.566203557520261e-05,
      "loss": 0.4315,
      "step": 22681
    },
    {
      "epoch": 7.161982949163246,
      "grad_norm": 4.465134620666504,
      "learning_rate": 3.566115847454654e-05,
      "loss": 0.3759,
      "step": 22682
    },
    {
      "epoch": 7.162298705399432,
      "grad_norm": 5.822551250457764,
      "learning_rate": 3.566028137389047e-05,
      "loss": 0.5942,
      "step": 22683
    },
    {
      "epoch": 7.162614461635617,
      "grad_norm": 6.078052520751953,
      "learning_rate": 3.56594042732344e-05,
      "loss": 0.478,
      "step": 22684
    },
    {
      "epoch": 7.162930217871803,
      "grad_norm": 5.350008964538574,
      "learning_rate": 3.5658527172578326e-05,
      "loss": 0.4584,
      "step": 22685
    },
    {
      "epoch": 7.163245974107989,
      "grad_norm": 6.128155708312988,
      "learning_rate": 3.565765007192226e-05,
      "loss": 0.4868,
      "step": 22686
    },
    {
      "epoch": 7.163561730344174,
      "grad_norm": 5.1897735595703125,
      "learning_rate": 3.565677297126618e-05,
      "loss": 0.4669,
      "step": 22687
    },
    {
      "epoch": 7.16387748658036,
      "grad_norm": 5.51115608215332,
      "learning_rate": 3.5655895870610114e-05,
      "loss": 0.5518,
      "step": 22688
    },
    {
      "epoch": 7.164193242816546,
      "grad_norm": 5.2586350440979,
      "learning_rate": 3.565501876995404e-05,
      "loss": 0.4186,
      "step": 22689
    },
    {
      "epoch": 7.164508999052731,
      "grad_norm": 5.996283531188965,
      "learning_rate": 3.565414166929797e-05,
      "loss": 0.4577,
      "step": 22690
    },
    {
      "epoch": 7.164824755288917,
      "grad_norm": 5.596292972564697,
      "learning_rate": 3.5653264568641896e-05,
      "loss": 0.4265,
      "step": 22691
    },
    {
      "epoch": 7.165140511525102,
      "grad_norm": 5.3326029777526855,
      "learning_rate": 3.565238746798583e-05,
      "loss": 0.5248,
      "step": 22692
    },
    {
      "epoch": 7.1654562677612885,
      "grad_norm": 6.4575018882751465,
      "learning_rate": 3.565151036732975e-05,
      "loss": 0.4236,
      "step": 22693
    },
    {
      "epoch": 7.165772023997474,
      "grad_norm": 5.314996242523193,
      "learning_rate": 3.5650633266673684e-05,
      "loss": 0.528,
      "step": 22694
    },
    {
      "epoch": 7.16608778023366,
      "grad_norm": 6.134920597076416,
      "learning_rate": 3.564975616601761e-05,
      "loss": 0.4833,
      "step": 22695
    },
    {
      "epoch": 7.166403536469845,
      "grad_norm": 5.858377456665039,
      "learning_rate": 3.564887906536154e-05,
      "loss": 0.4683,
      "step": 22696
    },
    {
      "epoch": 7.166719292706031,
      "grad_norm": 5.628345966339111,
      "learning_rate": 3.5648001964705466e-05,
      "loss": 0.471,
      "step": 22697
    },
    {
      "epoch": 7.167035048942217,
      "grad_norm": 6.471242427825928,
      "learning_rate": 3.5647124864049404e-05,
      "loss": 0.4053,
      "step": 22698
    },
    {
      "epoch": 7.167350805178402,
      "grad_norm": 5.9004082679748535,
      "learning_rate": 3.564624776339333e-05,
      "loss": 0.4931,
      "step": 22699
    },
    {
      "epoch": 7.167666561414588,
      "grad_norm": 9.622363090515137,
      "learning_rate": 3.564537066273726e-05,
      "loss": 0.5431,
      "step": 22700
    },
    {
      "epoch": 7.1679823176507735,
      "grad_norm": 7.0330705642700195,
      "learning_rate": 3.5644493562081186e-05,
      "loss": 0.4766,
      "step": 22701
    },
    {
      "epoch": 7.168298073886959,
      "grad_norm": 5.389521598815918,
      "learning_rate": 3.564361646142512e-05,
      "loss": 0.4403,
      "step": 22702
    },
    {
      "epoch": 7.168613830123145,
      "grad_norm": 4.377397060394287,
      "learning_rate": 3.564273936076904e-05,
      "loss": 0.418,
      "step": 22703
    },
    {
      "epoch": 7.16892958635933,
      "grad_norm": 4.533183574676514,
      "learning_rate": 3.5641862260112974e-05,
      "loss": 0.473,
      "step": 22704
    },
    {
      "epoch": 7.1692453425955165,
      "grad_norm": 7.852151870727539,
      "learning_rate": 3.5640985159456905e-05,
      "loss": 0.4928,
      "step": 22705
    },
    {
      "epoch": 7.169561098831702,
      "grad_norm": 4.393639087677002,
      "learning_rate": 3.564010805880083e-05,
      "loss": 0.3729,
      "step": 22706
    },
    {
      "epoch": 7.169876855067888,
      "grad_norm": 6.892276287078857,
      "learning_rate": 3.563923095814476e-05,
      "loss": 0.534,
      "step": 22707
    },
    {
      "epoch": 7.170192611304073,
      "grad_norm": 5.423511028289795,
      "learning_rate": 3.563835385748869e-05,
      "loss": 0.4232,
      "step": 22708
    },
    {
      "epoch": 7.1705083675402586,
      "grad_norm": 5.447515487670898,
      "learning_rate": 3.563747675683262e-05,
      "loss": 0.4834,
      "step": 22709
    },
    {
      "epoch": 7.170824123776445,
      "grad_norm": 4.877754211425781,
      "learning_rate": 3.5636599656176544e-05,
      "loss": 0.4402,
      "step": 22710
    },
    {
      "epoch": 7.17113988001263,
      "grad_norm": 3.934969663619995,
      "learning_rate": 3.5635722555520475e-05,
      "loss": 0.4151,
      "step": 22711
    },
    {
      "epoch": 7.171455636248816,
      "grad_norm": 6.532742500305176,
      "learning_rate": 3.56348454548644e-05,
      "loss": 0.5628,
      "step": 22712
    },
    {
      "epoch": 7.1717713924850015,
      "grad_norm": 6.346555709838867,
      "learning_rate": 3.563396835420833e-05,
      "loss": 0.5552,
      "step": 22713
    },
    {
      "epoch": 7.172087148721188,
      "grad_norm": 6.2441229820251465,
      "learning_rate": 3.563309125355226e-05,
      "loss": 0.4692,
      "step": 22714
    },
    {
      "epoch": 7.172402904957373,
      "grad_norm": 3.8521604537963867,
      "learning_rate": 3.563221415289619e-05,
      "loss": 0.3999,
      "step": 22715
    },
    {
      "epoch": 7.172718661193558,
      "grad_norm": 6.21467399597168,
      "learning_rate": 3.563133705224011e-05,
      "loss": 0.396,
      "step": 22716
    },
    {
      "epoch": 7.1730344174297445,
      "grad_norm": 4.174829006195068,
      "learning_rate": 3.5630459951584045e-05,
      "loss": 0.391,
      "step": 22717
    },
    {
      "epoch": 7.17335017366593,
      "grad_norm": 5.368980884552002,
      "learning_rate": 3.562958285092797e-05,
      "loss": 0.4562,
      "step": 22718
    },
    {
      "epoch": 7.173665929902116,
      "grad_norm": 5.825230598449707,
      "learning_rate": 3.56287057502719e-05,
      "loss": 0.445,
      "step": 22719
    },
    {
      "epoch": 7.173981686138301,
      "grad_norm": 4.476372718811035,
      "learning_rate": 3.5627828649615827e-05,
      "loss": 0.3964,
      "step": 22720
    },
    {
      "epoch": 7.1742974423744865,
      "grad_norm": 5.45636510848999,
      "learning_rate": 3.562695154895976e-05,
      "loss": 0.4526,
      "step": 22721
    },
    {
      "epoch": 7.174613198610673,
      "grad_norm": 5.179783821105957,
      "learning_rate": 3.562607444830369e-05,
      "loss": 0.4162,
      "step": 22722
    },
    {
      "epoch": 7.174928954846858,
      "grad_norm": 4.8379950523376465,
      "learning_rate": 3.5625197347647615e-05,
      "loss": 0.4624,
      "step": 22723
    },
    {
      "epoch": 7.175244711083044,
      "grad_norm": 7.236095428466797,
      "learning_rate": 3.5624320246991546e-05,
      "loss": 0.4654,
      "step": 22724
    },
    {
      "epoch": 7.1755604673192295,
      "grad_norm": 5.3097100257873535,
      "learning_rate": 3.562344314633548e-05,
      "loss": 0.4678,
      "step": 22725
    },
    {
      "epoch": 7.175876223555415,
      "grad_norm": 5.062762260437012,
      "learning_rate": 3.56225660456794e-05,
      "loss": 0.5235,
      "step": 22726
    },
    {
      "epoch": 7.176191979791601,
      "grad_norm": 5.028376579284668,
      "learning_rate": 3.5621688945023335e-05,
      "loss": 0.4338,
      "step": 22727
    },
    {
      "epoch": 7.176507736027786,
      "grad_norm": 6.335750102996826,
      "learning_rate": 3.5620811844367266e-05,
      "loss": 0.4456,
      "step": 22728
    },
    {
      "epoch": 7.176823492263972,
      "grad_norm": 4.53285551071167,
      "learning_rate": 3.561993474371119e-05,
      "loss": 0.4232,
      "step": 22729
    },
    {
      "epoch": 7.177139248500158,
      "grad_norm": 6.782638072967529,
      "learning_rate": 3.561905764305512e-05,
      "loss": 0.4962,
      "step": 22730
    },
    {
      "epoch": 7.177455004736344,
      "grad_norm": 6.661136627197266,
      "learning_rate": 3.561818054239905e-05,
      "loss": 0.4301,
      "step": 22731
    },
    {
      "epoch": 7.177770760972529,
      "grad_norm": 5.984202861785889,
      "learning_rate": 3.561730344174298e-05,
      "loss": 0.4808,
      "step": 22732
    },
    {
      "epoch": 7.1780865172087145,
      "grad_norm": 5.522934913635254,
      "learning_rate": 3.5616426341086905e-05,
      "loss": 0.4759,
      "step": 22733
    },
    {
      "epoch": 7.178402273444901,
      "grad_norm": 5.377166271209717,
      "learning_rate": 3.5615549240430836e-05,
      "loss": 0.4211,
      "step": 22734
    },
    {
      "epoch": 7.178718029681086,
      "grad_norm": 5.420454025268555,
      "learning_rate": 3.561467213977476e-05,
      "loss": 0.4594,
      "step": 22735
    },
    {
      "epoch": 7.179033785917272,
      "grad_norm": 5.079048156738281,
      "learning_rate": 3.561379503911869e-05,
      "loss": 0.4488,
      "step": 22736
    },
    {
      "epoch": 7.179349542153457,
      "grad_norm": 6.62606143951416,
      "learning_rate": 3.561291793846262e-05,
      "loss": 0.5094,
      "step": 22737
    },
    {
      "epoch": 7.179665298389644,
      "grad_norm": 5.754552364349365,
      "learning_rate": 3.561204083780655e-05,
      "loss": 0.5374,
      "step": 22738
    },
    {
      "epoch": 7.179981054625829,
      "grad_norm": 4.634807109832764,
      "learning_rate": 3.5611163737150474e-05,
      "loss": 0.5021,
      "step": 22739
    },
    {
      "epoch": 7.180296810862014,
      "grad_norm": 5.697965621948242,
      "learning_rate": 3.5610286636494406e-05,
      "loss": 0.4103,
      "step": 22740
    },
    {
      "epoch": 7.1806125670982,
      "grad_norm": 6.45551061630249,
      "learning_rate": 3.560940953583833e-05,
      "loss": 0.5699,
      "step": 22741
    },
    {
      "epoch": 7.180928323334386,
      "grad_norm": 6.375174522399902,
      "learning_rate": 3.560853243518226e-05,
      "loss": 0.5682,
      "step": 22742
    },
    {
      "epoch": 7.181244079570572,
      "grad_norm": 6.026449203491211,
      "learning_rate": 3.5607655334526194e-05,
      "loss": 0.5665,
      "step": 22743
    },
    {
      "epoch": 7.181559835806757,
      "grad_norm": 5.641820907592773,
      "learning_rate": 3.560677823387012e-05,
      "loss": 0.4612,
      "step": 22744
    },
    {
      "epoch": 7.1818755920429425,
      "grad_norm": 4.157271862030029,
      "learning_rate": 3.560590113321405e-05,
      "loss": 0.3031,
      "step": 22745
    },
    {
      "epoch": 7.182191348279129,
      "grad_norm": 4.491852760314941,
      "learning_rate": 3.5605024032557976e-05,
      "loss": 0.3926,
      "step": 22746
    },
    {
      "epoch": 7.182507104515314,
      "grad_norm": 5.708003520965576,
      "learning_rate": 3.560414693190191e-05,
      "loss": 0.4581,
      "step": 22747
    },
    {
      "epoch": 7.1828228607515,
      "grad_norm": 5.081571102142334,
      "learning_rate": 3.560326983124583e-05,
      "loss": 0.4043,
      "step": 22748
    },
    {
      "epoch": 7.183138616987685,
      "grad_norm": 5.256591320037842,
      "learning_rate": 3.5602392730589764e-05,
      "loss": 0.426,
      "step": 22749
    },
    {
      "epoch": 7.183454373223872,
      "grad_norm": 5.795232772827148,
      "learning_rate": 3.560151562993369e-05,
      "loss": 0.5143,
      "step": 22750
    },
    {
      "epoch": 7.183770129460057,
      "grad_norm": 5.457809925079346,
      "learning_rate": 3.560063852927762e-05,
      "loss": 0.6135,
      "step": 22751
    },
    {
      "epoch": 7.184085885696242,
      "grad_norm": 5.4665141105651855,
      "learning_rate": 3.559976142862155e-05,
      "loss": 0.619,
      "step": 22752
    },
    {
      "epoch": 7.184401641932428,
      "grad_norm": 5.128470420837402,
      "learning_rate": 3.5598884327965484e-05,
      "loss": 0.3871,
      "step": 22753
    },
    {
      "epoch": 7.184717398168614,
      "grad_norm": 5.8470306396484375,
      "learning_rate": 3.559800722730941e-05,
      "loss": 0.5457,
      "step": 22754
    },
    {
      "epoch": 7.1850331544048,
      "grad_norm": 6.658123016357422,
      "learning_rate": 3.559713012665334e-05,
      "loss": 0.5855,
      "step": 22755
    },
    {
      "epoch": 7.185348910640985,
      "grad_norm": 5.125473499298096,
      "learning_rate": 3.5596253025997265e-05,
      "loss": 0.412,
      "step": 22756
    },
    {
      "epoch": 7.18566466687717,
      "grad_norm": 6.281113147735596,
      "learning_rate": 3.55953759253412e-05,
      "loss": 0.5595,
      "step": 22757
    },
    {
      "epoch": 7.185980423113357,
      "grad_norm": 4.981464385986328,
      "learning_rate": 3.559449882468512e-05,
      "loss": 0.4434,
      "step": 22758
    },
    {
      "epoch": 7.186296179349542,
      "grad_norm": 4.748308181762695,
      "learning_rate": 3.5593621724029054e-05,
      "loss": 0.4595,
      "step": 22759
    },
    {
      "epoch": 7.186611935585728,
      "grad_norm": 6.665553092956543,
      "learning_rate": 3.559274462337298e-05,
      "loss": 0.5813,
      "step": 22760
    },
    {
      "epoch": 7.186927691821913,
      "grad_norm": 5.901778697967529,
      "learning_rate": 3.559186752271691e-05,
      "loss": 0.5654,
      "step": 22761
    },
    {
      "epoch": 7.1872434480581,
      "grad_norm": 5.261004447937012,
      "learning_rate": 3.5590990422060835e-05,
      "loss": 0.5042,
      "step": 22762
    },
    {
      "epoch": 7.187559204294285,
      "grad_norm": 5.919010162353516,
      "learning_rate": 3.559011332140477e-05,
      "loss": 0.4909,
      "step": 22763
    },
    {
      "epoch": 7.18787496053047,
      "grad_norm": 5.194098472595215,
      "learning_rate": 3.558923622074869e-05,
      "loss": 0.4017,
      "step": 22764
    },
    {
      "epoch": 7.188190716766656,
      "grad_norm": 5.113267421722412,
      "learning_rate": 3.5588359120092623e-05,
      "loss": 0.4955,
      "step": 22765
    },
    {
      "epoch": 7.188506473002842,
      "grad_norm": 6.367583274841309,
      "learning_rate": 3.5587482019436555e-05,
      "loss": 0.5252,
      "step": 22766
    },
    {
      "epoch": 7.188822229239028,
      "grad_norm": 5.8898091316223145,
      "learning_rate": 3.558660491878048e-05,
      "loss": 0.6121,
      "step": 22767
    },
    {
      "epoch": 7.189137985475213,
      "grad_norm": 5.957082748413086,
      "learning_rate": 3.558572781812441e-05,
      "loss": 0.4674,
      "step": 22768
    },
    {
      "epoch": 7.189453741711398,
      "grad_norm": 4.8705668449401855,
      "learning_rate": 3.558485071746834e-05,
      "loss": 0.4108,
      "step": 22769
    },
    {
      "epoch": 7.189769497947585,
      "grad_norm": 6.112671375274658,
      "learning_rate": 3.558397361681227e-05,
      "loss": 0.4155,
      "step": 22770
    },
    {
      "epoch": 7.19008525418377,
      "grad_norm": 7.051662921905518,
      "learning_rate": 3.558309651615619e-05,
      "loss": 0.5575,
      "step": 22771
    },
    {
      "epoch": 7.190401010419956,
      "grad_norm": 5.048953056335449,
      "learning_rate": 3.5582219415500125e-05,
      "loss": 0.488,
      "step": 22772
    },
    {
      "epoch": 7.190716766656141,
      "grad_norm": 8.32802677154541,
      "learning_rate": 3.558134231484405e-05,
      "loss": 0.5535,
      "step": 22773
    },
    {
      "epoch": 7.1910325228923275,
      "grad_norm": 5.06299352645874,
      "learning_rate": 3.558046521418798e-05,
      "loss": 0.4027,
      "step": 22774
    },
    {
      "epoch": 7.191348279128513,
      "grad_norm": 4.8015456199646,
      "learning_rate": 3.5579588113531906e-05,
      "loss": 0.4226,
      "step": 22775
    },
    {
      "epoch": 7.191664035364698,
      "grad_norm": 7.216803550720215,
      "learning_rate": 3.557871101287584e-05,
      "loss": 0.5223,
      "step": 22776
    },
    {
      "epoch": 7.191979791600884,
      "grad_norm": 4.480638027191162,
      "learning_rate": 3.557783391221976e-05,
      "loss": 0.4594,
      "step": 22777
    },
    {
      "epoch": 7.19229554783707,
      "grad_norm": 7.541806697845459,
      "learning_rate": 3.5576956811563695e-05,
      "loss": 0.5978,
      "step": 22778
    },
    {
      "epoch": 7.192611304073256,
      "grad_norm": 7.765865325927734,
      "learning_rate": 3.557607971090762e-05,
      "loss": 0.5492,
      "step": 22779
    },
    {
      "epoch": 7.192927060309441,
      "grad_norm": 6.038012981414795,
      "learning_rate": 3.557520261025156e-05,
      "loss": 0.5161,
      "step": 22780
    },
    {
      "epoch": 7.193242816545626,
      "grad_norm": 5.69826602935791,
      "learning_rate": 3.557432550959548e-05,
      "loss": 0.4488,
      "step": 22781
    },
    {
      "epoch": 7.1935585727818125,
      "grad_norm": 5.400811672210693,
      "learning_rate": 3.5573448408939415e-05,
      "loss": 0.4699,
      "step": 22782
    },
    {
      "epoch": 7.193874329017998,
      "grad_norm": 5.347658157348633,
      "learning_rate": 3.557257130828334e-05,
      "loss": 0.4295,
      "step": 22783
    },
    {
      "epoch": 7.194190085254184,
      "grad_norm": 5.737813949584961,
      "learning_rate": 3.557169420762727e-05,
      "loss": 0.4767,
      "step": 22784
    },
    {
      "epoch": 7.194505841490369,
      "grad_norm": 5.269777774810791,
      "learning_rate": 3.5570817106971196e-05,
      "loss": 0.5506,
      "step": 22785
    },
    {
      "epoch": 7.1948215977265555,
      "grad_norm": 5.636567115783691,
      "learning_rate": 3.556994000631513e-05,
      "loss": 0.4724,
      "step": 22786
    },
    {
      "epoch": 7.195137353962741,
      "grad_norm": 4.900243759155273,
      "learning_rate": 3.556906290565906e-05,
      "loss": 0.544,
      "step": 22787
    },
    {
      "epoch": 7.195453110198926,
      "grad_norm": 6.052870273590088,
      "learning_rate": 3.5568185805002984e-05,
      "loss": 0.5015,
      "step": 22788
    },
    {
      "epoch": 7.195768866435112,
      "grad_norm": 3.9968762397766113,
      "learning_rate": 3.5567308704346916e-05,
      "loss": 0.4175,
      "step": 22789
    },
    {
      "epoch": 7.196084622671298,
      "grad_norm": 8.029719352722168,
      "learning_rate": 3.556643160369084e-05,
      "loss": 0.4068,
      "step": 22790
    },
    {
      "epoch": 7.196400378907484,
      "grad_norm": 5.966735363006592,
      "learning_rate": 3.556555450303477e-05,
      "loss": 0.5384,
      "step": 22791
    },
    {
      "epoch": 7.196716135143669,
      "grad_norm": 4.617420673370361,
      "learning_rate": 3.55646774023787e-05,
      "loss": 0.4043,
      "step": 22792
    },
    {
      "epoch": 7.197031891379854,
      "grad_norm": 5.130821704864502,
      "learning_rate": 3.556380030172263e-05,
      "loss": 0.4427,
      "step": 22793
    },
    {
      "epoch": 7.1973476476160405,
      "grad_norm": 6.036555767059326,
      "learning_rate": 3.5562923201066554e-05,
      "loss": 0.4407,
      "step": 22794
    },
    {
      "epoch": 7.197663403852226,
      "grad_norm": 6.297396183013916,
      "learning_rate": 3.5562046100410486e-05,
      "loss": 0.5187,
      "step": 22795
    },
    {
      "epoch": 7.197979160088412,
      "grad_norm": 5.166051387786865,
      "learning_rate": 3.556116899975441e-05,
      "loss": 0.405,
      "step": 22796
    },
    {
      "epoch": 7.198294916324597,
      "grad_norm": 5.76202917098999,
      "learning_rate": 3.556029189909834e-05,
      "loss": 0.4934,
      "step": 22797
    },
    {
      "epoch": 7.1986106725607835,
      "grad_norm": 5.428977966308594,
      "learning_rate": 3.555941479844227e-05,
      "loss": 0.5178,
      "step": 22798
    },
    {
      "epoch": 7.198926428796969,
      "grad_norm": 5.786489963531494,
      "learning_rate": 3.55585376977862e-05,
      "loss": 0.4566,
      "step": 22799
    },
    {
      "epoch": 7.199242185033154,
      "grad_norm": 4.533040523529053,
      "learning_rate": 3.5557660597130124e-05,
      "loss": 0.3428,
      "step": 22800
    },
    {
      "epoch": 7.199242185033154,
      "eval_loss": 0.9229885339736938,
      "eval_runtime": 4.5428,
      "eval_samples_per_second": 548.116,
      "eval_steps_per_second": 4.623,
      "step": 22800
    },
    {
      "epoch": 7.19955794126934,
      "grad_norm": 5.287024021148682,
      "learning_rate": 3.5556783496474056e-05,
      "loss": 0.4489,
      "step": 22801
    },
    {
      "epoch": 7.1998736975055255,
      "grad_norm": 5.227520942687988,
      "learning_rate": 3.555590639581798e-05,
      "loss": 0.4147,
      "step": 22802
    },
    {
      "epoch": 7.200189453741712,
      "grad_norm": 5.093524932861328,
      "learning_rate": 3.555502929516191e-05,
      "loss": 0.4187,
      "step": 22803
    },
    {
      "epoch": 7.200505209977897,
      "grad_norm": 5.0652313232421875,
      "learning_rate": 3.5554152194505844e-05,
      "loss": 0.4936,
      "step": 22804
    },
    {
      "epoch": 7.200820966214083,
      "grad_norm": 7.18691349029541,
      "learning_rate": 3.555327509384977e-05,
      "loss": 0.412,
      "step": 22805
    },
    {
      "epoch": 7.2011367224502685,
      "grad_norm": 4.311334609985352,
      "learning_rate": 3.55523979931937e-05,
      "loss": 0.4868,
      "step": 22806
    },
    {
      "epoch": 7.201452478686454,
      "grad_norm": 6.2922492027282715,
      "learning_rate": 3.555152089253763e-05,
      "loss": 0.4146,
      "step": 22807
    },
    {
      "epoch": 7.20176823492264,
      "grad_norm": 4.520719528198242,
      "learning_rate": 3.5550643791881564e-05,
      "loss": 0.5181,
      "step": 22808
    },
    {
      "epoch": 7.202083991158825,
      "grad_norm": 6.000689506530762,
      "learning_rate": 3.554976669122549e-05,
      "loss": 0.5164,
      "step": 22809
    },
    {
      "epoch": 7.202399747395011,
      "grad_norm": 5.553770542144775,
      "learning_rate": 3.554888959056942e-05,
      "loss": 0.4444,
      "step": 22810
    },
    {
      "epoch": 7.202715503631197,
      "grad_norm": 10.408552169799805,
      "learning_rate": 3.5548012489913345e-05,
      "loss": 0.4348,
      "step": 22811
    },
    {
      "epoch": 7.203031259867382,
      "grad_norm": 5.122759819030762,
      "learning_rate": 3.554713538925728e-05,
      "loss": 0.4314,
      "step": 22812
    },
    {
      "epoch": 7.203347016103568,
      "grad_norm": 4.4916300773620605,
      "learning_rate": 3.55462582886012e-05,
      "loss": 0.4494,
      "step": 22813
    },
    {
      "epoch": 7.2036627723397535,
      "grad_norm": 5.581121444702148,
      "learning_rate": 3.5545381187945134e-05,
      "loss": 0.4111,
      "step": 22814
    },
    {
      "epoch": 7.20397852857594,
      "grad_norm": 5.214041709899902,
      "learning_rate": 3.554450408728906e-05,
      "loss": 0.4925,
      "step": 22815
    },
    {
      "epoch": 7.204294284812125,
      "grad_norm": 7.139715194702148,
      "learning_rate": 3.554362698663299e-05,
      "loss": 0.5018,
      "step": 22816
    },
    {
      "epoch": 7.20461004104831,
      "grad_norm": 4.550910472869873,
      "learning_rate": 3.5542749885976915e-05,
      "loss": 0.4124,
      "step": 22817
    },
    {
      "epoch": 7.2049257972844964,
      "grad_norm": 5.137755870819092,
      "learning_rate": 3.554187278532085e-05,
      "loss": 0.4146,
      "step": 22818
    },
    {
      "epoch": 7.205241553520682,
      "grad_norm": 4.438161849975586,
      "learning_rate": 3.554099568466477e-05,
      "loss": 0.5072,
      "step": 22819
    },
    {
      "epoch": 7.205557309756868,
      "grad_norm": 6.25473165512085,
      "learning_rate": 3.55401185840087e-05,
      "loss": 0.5167,
      "step": 22820
    },
    {
      "epoch": 7.205873065993053,
      "grad_norm": 5.060571670532227,
      "learning_rate": 3.553924148335263e-05,
      "loss": 0.4955,
      "step": 22821
    },
    {
      "epoch": 7.206188822229239,
      "grad_norm": 4.543044567108154,
      "learning_rate": 3.553836438269656e-05,
      "loss": 0.5533,
      "step": 22822
    },
    {
      "epoch": 7.206504578465425,
      "grad_norm": 5.3217339515686035,
      "learning_rate": 3.5537487282040485e-05,
      "loss": 0.4862,
      "step": 22823
    },
    {
      "epoch": 7.20682033470161,
      "grad_norm": 5.228382587432861,
      "learning_rate": 3.5536610181384417e-05,
      "loss": 0.3302,
      "step": 22824
    },
    {
      "epoch": 7.207136090937796,
      "grad_norm": 4.647213935852051,
      "learning_rate": 3.553573308072835e-05,
      "loss": 0.4929,
      "step": 22825
    },
    {
      "epoch": 7.2074518471739815,
      "grad_norm": 6.695091724395752,
      "learning_rate": 3.553485598007227e-05,
      "loss": 0.5496,
      "step": 22826
    },
    {
      "epoch": 7.207767603410168,
      "grad_norm": 7.871787071228027,
      "learning_rate": 3.5533978879416205e-05,
      "loss": 0.5508,
      "step": 22827
    },
    {
      "epoch": 7.208083359646353,
      "grad_norm": 5.104682445526123,
      "learning_rate": 3.553310177876013e-05,
      "loss": 0.3896,
      "step": 22828
    },
    {
      "epoch": 7.208399115882539,
      "grad_norm": 6.6069512367248535,
      "learning_rate": 3.553222467810406e-05,
      "loss": 0.4461,
      "step": 22829
    },
    {
      "epoch": 7.208714872118724,
      "grad_norm": 5.607176303863525,
      "learning_rate": 3.5531347577447986e-05,
      "loss": 0.4908,
      "step": 22830
    },
    {
      "epoch": 7.20903062835491,
      "grad_norm": 4.72003173828125,
      "learning_rate": 3.553047047679192e-05,
      "loss": 0.5476,
      "step": 22831
    },
    {
      "epoch": 7.209346384591096,
      "grad_norm": 5.52592658996582,
      "learning_rate": 3.552959337613584e-05,
      "loss": 0.4342,
      "step": 22832
    },
    {
      "epoch": 7.209662140827281,
      "grad_norm": 4.981156349182129,
      "learning_rate": 3.5528716275479775e-05,
      "loss": 0.4545,
      "step": 22833
    },
    {
      "epoch": 7.209977897063467,
      "grad_norm": 5.972721576690674,
      "learning_rate": 3.5527839174823706e-05,
      "loss": 0.579,
      "step": 22834
    },
    {
      "epoch": 7.210293653299653,
      "grad_norm": 6.604201793670654,
      "learning_rate": 3.552696207416764e-05,
      "loss": 0.6518,
      "step": 22835
    },
    {
      "epoch": 7.210609409535838,
      "grad_norm": 4.479339122772217,
      "learning_rate": 3.552608497351156e-05,
      "loss": 0.3753,
      "step": 22836
    },
    {
      "epoch": 7.210925165772024,
      "grad_norm": 7.466172695159912,
      "learning_rate": 3.5525207872855495e-05,
      "loss": 0.4559,
      "step": 22837
    },
    {
      "epoch": 7.211240922008209,
      "grad_norm": 5.975768089294434,
      "learning_rate": 3.552433077219942e-05,
      "loss": 0.4089,
      "step": 22838
    },
    {
      "epoch": 7.211556678244396,
      "grad_norm": 5.474720478057861,
      "learning_rate": 3.552345367154335e-05,
      "loss": 0.4833,
      "step": 22839
    },
    {
      "epoch": 7.211872434480581,
      "grad_norm": 7.075077533721924,
      "learning_rate": 3.5522576570887276e-05,
      "loss": 0.501,
      "step": 22840
    },
    {
      "epoch": 7.212188190716766,
      "grad_norm": 5.740599155426025,
      "learning_rate": 3.552169947023121e-05,
      "loss": 0.632,
      "step": 22841
    },
    {
      "epoch": 7.212503946952952,
      "grad_norm": 5.1631340980529785,
      "learning_rate": 3.552082236957513e-05,
      "loss": 0.4358,
      "step": 22842
    },
    {
      "epoch": 7.212819703189138,
      "grad_norm": 5.308266639709473,
      "learning_rate": 3.5519945268919064e-05,
      "loss": 0.5105,
      "step": 22843
    },
    {
      "epoch": 7.213135459425324,
      "grad_norm": 5.157332420349121,
      "learning_rate": 3.551906816826299e-05,
      "loss": 0.4316,
      "step": 22844
    },
    {
      "epoch": 7.213451215661509,
      "grad_norm": 5.742983341217041,
      "learning_rate": 3.551819106760692e-05,
      "loss": 0.4494,
      "step": 22845
    },
    {
      "epoch": 7.213766971897695,
      "grad_norm": 5.955564498901367,
      "learning_rate": 3.551731396695085e-05,
      "loss": 0.458,
      "step": 22846
    },
    {
      "epoch": 7.214082728133881,
      "grad_norm": 4.34698486328125,
      "learning_rate": 3.551643686629478e-05,
      "loss": 0.4178,
      "step": 22847
    },
    {
      "epoch": 7.214398484370066,
      "grad_norm": 5.320803642272949,
      "learning_rate": 3.551555976563871e-05,
      "loss": 0.4424,
      "step": 22848
    },
    {
      "epoch": 7.214714240606252,
      "grad_norm": 4.875733852386475,
      "learning_rate": 3.5514682664982634e-05,
      "loss": 0.3802,
      "step": 22849
    },
    {
      "epoch": 7.215029996842437,
      "grad_norm": 6.124943733215332,
      "learning_rate": 3.5513805564326566e-05,
      "loss": 0.5239,
      "step": 22850
    },
    {
      "epoch": 7.215345753078624,
      "grad_norm": 7.171687126159668,
      "learning_rate": 3.551292846367049e-05,
      "loss": 0.4366,
      "step": 22851
    },
    {
      "epoch": 7.215661509314809,
      "grad_norm": 5.127191066741943,
      "learning_rate": 3.551205136301442e-05,
      "loss": 0.4339,
      "step": 22852
    },
    {
      "epoch": 7.215977265550995,
      "grad_norm": 4.8373284339904785,
      "learning_rate": 3.551117426235835e-05,
      "loss": 0.4782,
      "step": 22853
    },
    {
      "epoch": 7.21629302178718,
      "grad_norm": 5.338231086730957,
      "learning_rate": 3.551029716170228e-05,
      "loss": 0.5159,
      "step": 22854
    },
    {
      "epoch": 7.216608778023366,
      "grad_norm": 5.64408540725708,
      "learning_rate": 3.5509420061046204e-05,
      "loss": 0.4281,
      "step": 22855
    },
    {
      "epoch": 7.216924534259552,
      "grad_norm": 4.187222003936768,
      "learning_rate": 3.5508542960390136e-05,
      "loss": 0.4091,
      "step": 22856
    },
    {
      "epoch": 7.217240290495737,
      "grad_norm": 4.187217712402344,
      "learning_rate": 3.550766585973406e-05,
      "loss": 0.4381,
      "step": 22857
    },
    {
      "epoch": 7.217556046731923,
      "grad_norm": 5.730690002441406,
      "learning_rate": 3.550678875907799e-05,
      "loss": 0.4867,
      "step": 22858
    },
    {
      "epoch": 7.217871802968109,
      "grad_norm": 6.272022247314453,
      "learning_rate": 3.550591165842192e-05,
      "loss": 0.5046,
      "step": 22859
    },
    {
      "epoch": 7.218187559204294,
      "grad_norm": 5.623830795288086,
      "learning_rate": 3.550503455776585e-05,
      "loss": 0.4024,
      "step": 22860
    },
    {
      "epoch": 7.21850331544048,
      "grad_norm": 5.516840934753418,
      "learning_rate": 3.550415745710978e-05,
      "loss": 0.4758,
      "step": 22861
    },
    {
      "epoch": 7.218819071676665,
      "grad_norm": 4.589058876037598,
      "learning_rate": 3.550328035645371e-05,
      "loss": 0.4285,
      "step": 22862
    },
    {
      "epoch": 7.2191348279128515,
      "grad_norm": 4.773648262023926,
      "learning_rate": 3.550240325579764e-05,
      "loss": 0.503,
      "step": 22863
    },
    {
      "epoch": 7.219450584149037,
      "grad_norm": 4.255792140960693,
      "learning_rate": 3.550152615514157e-05,
      "loss": 0.3308,
      "step": 22864
    },
    {
      "epoch": 7.219766340385223,
      "grad_norm": 5.006243705749512,
      "learning_rate": 3.5500649054485494e-05,
      "loss": 0.4816,
      "step": 22865
    },
    {
      "epoch": 7.220082096621408,
      "grad_norm": 5.98381233215332,
      "learning_rate": 3.5499771953829425e-05,
      "loss": 0.4742,
      "step": 22866
    },
    {
      "epoch": 7.220397852857594,
      "grad_norm": 5.874510288238525,
      "learning_rate": 3.549889485317335e-05,
      "loss": 0.5444,
      "step": 22867
    },
    {
      "epoch": 7.22071360909378,
      "grad_norm": 4.68148136138916,
      "learning_rate": 3.549801775251728e-05,
      "loss": 0.5353,
      "step": 22868
    },
    {
      "epoch": 7.221029365329965,
      "grad_norm": 5.194401741027832,
      "learning_rate": 3.5497140651861213e-05,
      "loss": 0.3865,
      "step": 22869
    },
    {
      "epoch": 7.221345121566151,
      "grad_norm": 5.349534511566162,
      "learning_rate": 3.549626355120514e-05,
      "loss": 0.5512,
      "step": 22870
    },
    {
      "epoch": 7.221660877802337,
      "grad_norm": 5.361007213592529,
      "learning_rate": 3.549538645054907e-05,
      "loss": 0.3515,
      "step": 22871
    },
    {
      "epoch": 7.221976634038522,
      "grad_norm": 5.27130651473999,
      "learning_rate": 3.5494509349892995e-05,
      "loss": 0.3786,
      "step": 22872
    },
    {
      "epoch": 7.222292390274708,
      "grad_norm": 5.355329513549805,
      "learning_rate": 3.549363224923693e-05,
      "loss": 0.3351,
      "step": 22873
    },
    {
      "epoch": 7.222608146510893,
      "grad_norm": 8.856934547424316,
      "learning_rate": 3.549275514858085e-05,
      "loss": 0.4053,
      "step": 22874
    },
    {
      "epoch": 7.2229239027470795,
      "grad_norm": 6.8539204597473145,
      "learning_rate": 3.549187804792478e-05,
      "loss": 0.4591,
      "step": 22875
    },
    {
      "epoch": 7.223239658983265,
      "grad_norm": 7.226263999938965,
      "learning_rate": 3.549100094726871e-05,
      "loss": 0.5046,
      "step": 22876
    },
    {
      "epoch": 7.223555415219451,
      "grad_norm": 7.483605861663818,
      "learning_rate": 3.549012384661264e-05,
      "loss": 0.5737,
      "step": 22877
    },
    {
      "epoch": 7.223871171455636,
      "grad_norm": 6.340723991394043,
      "learning_rate": 3.5489246745956565e-05,
      "loss": 0.478,
      "step": 22878
    },
    {
      "epoch": 7.224186927691822,
      "grad_norm": 6.109525680541992,
      "learning_rate": 3.5488369645300496e-05,
      "loss": 0.3558,
      "step": 22879
    },
    {
      "epoch": 7.224502683928008,
      "grad_norm": 5.697549343109131,
      "learning_rate": 3.548749254464442e-05,
      "loss": 0.4033,
      "step": 22880
    },
    {
      "epoch": 7.224818440164193,
      "grad_norm": 4.487539768218994,
      "learning_rate": 3.548661544398835e-05,
      "loss": 0.3925,
      "step": 22881
    },
    {
      "epoch": 7.225134196400379,
      "grad_norm": 5.284371376037598,
      "learning_rate": 3.548573834333228e-05,
      "loss": 0.4848,
      "step": 22882
    },
    {
      "epoch": 7.2254499526365645,
      "grad_norm": 5.930054664611816,
      "learning_rate": 3.548486124267621e-05,
      "loss": 0.3997,
      "step": 22883
    },
    {
      "epoch": 7.22576570887275,
      "grad_norm": 5.419722557067871,
      "learning_rate": 3.548398414202014e-05,
      "loss": 0.4203,
      "step": 22884
    },
    {
      "epoch": 7.226081465108936,
      "grad_norm": 5.246829986572266,
      "learning_rate": 3.5483107041364066e-05,
      "loss": 0.4253,
      "step": 22885
    },
    {
      "epoch": 7.226397221345121,
      "grad_norm": 7.312710762023926,
      "learning_rate": 3.5482229940708e-05,
      "loss": 0.448,
      "step": 22886
    },
    {
      "epoch": 7.2267129775813075,
      "grad_norm": 5.119015693664551,
      "learning_rate": 3.548135284005192e-05,
      "loss": 0.5187,
      "step": 22887
    },
    {
      "epoch": 7.227028733817493,
      "grad_norm": 6.468506336212158,
      "learning_rate": 3.5480475739395854e-05,
      "loss": 0.5554,
      "step": 22888
    },
    {
      "epoch": 7.227344490053679,
      "grad_norm": 6.150180816650391,
      "learning_rate": 3.5479598638739786e-05,
      "loss": 0.4674,
      "step": 22889
    },
    {
      "epoch": 7.227660246289864,
      "grad_norm": 5.655854225158691,
      "learning_rate": 3.547872153808372e-05,
      "loss": 0.3731,
      "step": 22890
    },
    {
      "epoch": 7.2279760025260495,
      "grad_norm": 5.672179222106934,
      "learning_rate": 3.547784443742764e-05,
      "loss": 0.515,
      "step": 22891
    },
    {
      "epoch": 7.228291758762236,
      "grad_norm": 5.03962516784668,
      "learning_rate": 3.5476967336771574e-05,
      "loss": 0.6132,
      "step": 22892
    },
    {
      "epoch": 7.228607514998421,
      "grad_norm": 6.354611873626709,
      "learning_rate": 3.54760902361155e-05,
      "loss": 0.4857,
      "step": 22893
    },
    {
      "epoch": 7.228923271234607,
      "grad_norm": 7.78801155090332,
      "learning_rate": 3.547521313545943e-05,
      "loss": 0.5275,
      "step": 22894
    },
    {
      "epoch": 7.2292390274707925,
      "grad_norm": 5.20448637008667,
      "learning_rate": 3.5474336034803356e-05,
      "loss": 0.4171,
      "step": 22895
    },
    {
      "epoch": 7.229554783706978,
      "grad_norm": 4.854667663574219,
      "learning_rate": 3.547345893414729e-05,
      "loss": 0.4851,
      "step": 22896
    },
    {
      "epoch": 7.229870539943164,
      "grad_norm": 5.356016635894775,
      "learning_rate": 3.547258183349121e-05,
      "loss": 0.4564,
      "step": 22897
    },
    {
      "epoch": 7.230186296179349,
      "grad_norm": 5.884738922119141,
      "learning_rate": 3.5471704732835144e-05,
      "loss": 0.5323,
      "step": 22898
    },
    {
      "epoch": 7.2305020524155355,
      "grad_norm": 5.7651286125183105,
      "learning_rate": 3.547082763217907e-05,
      "loss": 0.4541,
      "step": 22899
    },
    {
      "epoch": 7.230817808651721,
      "grad_norm": 5.154694080352783,
      "learning_rate": 3.5469950531523e-05,
      "loss": 0.4049,
      "step": 22900
    },
    {
      "epoch": 7.231133564887907,
      "grad_norm": 5.156009674072266,
      "learning_rate": 3.5469073430866926e-05,
      "loss": 0.4232,
      "step": 22901
    },
    {
      "epoch": 7.231449321124092,
      "grad_norm": 6.095948219299316,
      "learning_rate": 3.546819633021086e-05,
      "loss": 0.4803,
      "step": 22902
    },
    {
      "epoch": 7.2317650773602775,
      "grad_norm": 5.974354267120361,
      "learning_rate": 3.546731922955478e-05,
      "loss": 0.4184,
      "step": 22903
    },
    {
      "epoch": 7.232080833596464,
      "grad_norm": 5.3525567054748535,
      "learning_rate": 3.5466442128898714e-05,
      "loss": 0.3806,
      "step": 22904
    },
    {
      "epoch": 7.232396589832649,
      "grad_norm": 5.453007698059082,
      "learning_rate": 3.546556502824264e-05,
      "loss": 0.5553,
      "step": 22905
    },
    {
      "epoch": 7.232712346068835,
      "grad_norm": 6.802423000335693,
      "learning_rate": 3.546468792758657e-05,
      "loss": 0.4982,
      "step": 22906
    },
    {
      "epoch": 7.2330281023050205,
      "grad_norm": 6.751460552215576,
      "learning_rate": 3.54638108269305e-05,
      "loss": 0.4961,
      "step": 22907
    },
    {
      "epoch": 7.233343858541206,
      "grad_norm": 5.922307968139648,
      "learning_rate": 3.546293372627443e-05,
      "loss": 0.5055,
      "step": 22908
    },
    {
      "epoch": 7.233659614777392,
      "grad_norm": 7.372158050537109,
      "learning_rate": 3.546205662561836e-05,
      "loss": 0.5324,
      "step": 22909
    },
    {
      "epoch": 7.233975371013577,
      "grad_norm": 5.8294501304626465,
      "learning_rate": 3.5461179524962284e-05,
      "loss": 0.4751,
      "step": 22910
    },
    {
      "epoch": 7.234291127249763,
      "grad_norm": 4.040860176086426,
      "learning_rate": 3.5460302424306215e-05,
      "loss": 0.3987,
      "step": 22911
    },
    {
      "epoch": 7.234606883485949,
      "grad_norm": 5.710515022277832,
      "learning_rate": 3.545942532365014e-05,
      "loss": 0.5315,
      "step": 22912
    },
    {
      "epoch": 7.234922639722135,
      "grad_norm": 5.414395809173584,
      "learning_rate": 3.545854822299407e-05,
      "loss": 0.5042,
      "step": 22913
    },
    {
      "epoch": 7.23523839595832,
      "grad_norm": 5.107513904571533,
      "learning_rate": 3.5457671122338e-05,
      "loss": 0.5074,
      "step": 22914
    },
    {
      "epoch": 7.2355541521945055,
      "grad_norm": 6.033810615539551,
      "learning_rate": 3.545679402168193e-05,
      "loss": 0.4327,
      "step": 22915
    },
    {
      "epoch": 7.235869908430692,
      "grad_norm": 4.41677188873291,
      "learning_rate": 3.545591692102586e-05,
      "loss": 0.3911,
      "step": 22916
    },
    {
      "epoch": 7.236185664666877,
      "grad_norm": 5.90000581741333,
      "learning_rate": 3.545503982036979e-05,
      "loss": 0.4824,
      "step": 22917
    },
    {
      "epoch": 7.236501420903063,
      "grad_norm": 6.966625690460205,
      "learning_rate": 3.545416271971372e-05,
      "loss": 0.4607,
      "step": 22918
    },
    {
      "epoch": 7.236817177139248,
      "grad_norm": 6.24226188659668,
      "learning_rate": 3.545328561905765e-05,
      "loss": 0.4949,
      "step": 22919
    },
    {
      "epoch": 7.237132933375435,
      "grad_norm": 3.977998733520508,
      "learning_rate": 3.5452408518401573e-05,
      "loss": 0.3615,
      "step": 22920
    },
    {
      "epoch": 7.23744868961162,
      "grad_norm": 5.2261271476745605,
      "learning_rate": 3.5451531417745505e-05,
      "loss": 0.4581,
      "step": 22921
    },
    {
      "epoch": 7.237764445847805,
      "grad_norm": 6.377143859863281,
      "learning_rate": 3.545065431708943e-05,
      "loss": 0.5165,
      "step": 22922
    },
    {
      "epoch": 7.238080202083991,
      "grad_norm": 5.704044342041016,
      "learning_rate": 3.544977721643336e-05,
      "loss": 0.4623,
      "step": 22923
    },
    {
      "epoch": 7.238395958320177,
      "grad_norm": 4.604532241821289,
      "learning_rate": 3.544890011577729e-05,
      "loss": 0.4744,
      "step": 22924
    },
    {
      "epoch": 7.238711714556363,
      "grad_norm": 4.901187419891357,
      "learning_rate": 3.544802301512122e-05,
      "loss": 0.5148,
      "step": 22925
    },
    {
      "epoch": 7.239027470792548,
      "grad_norm": 5.217576503753662,
      "learning_rate": 3.544714591446514e-05,
      "loss": 0.4529,
      "step": 22926
    },
    {
      "epoch": 7.2393432270287335,
      "grad_norm": 6.192583084106445,
      "learning_rate": 3.5446268813809075e-05,
      "loss": 0.4571,
      "step": 22927
    },
    {
      "epoch": 7.23965898326492,
      "grad_norm": 5.1198811531066895,
      "learning_rate": 3.5445391713153007e-05,
      "loss": 0.4284,
      "step": 22928
    },
    {
      "epoch": 7.239974739501105,
      "grad_norm": 5.14911413192749,
      "learning_rate": 3.544451461249693e-05,
      "loss": 0.4404,
      "step": 22929
    },
    {
      "epoch": 7.240290495737291,
      "grad_norm": 6.390934944152832,
      "learning_rate": 3.544363751184086e-05,
      "loss": 0.5161,
      "step": 22930
    },
    {
      "epoch": 7.240606251973476,
      "grad_norm": 5.958348274230957,
      "learning_rate": 3.544276041118479e-05,
      "loss": 0.5451,
      "step": 22931
    },
    {
      "epoch": 7.240922008209662,
      "grad_norm": 5.999059200286865,
      "learning_rate": 3.544188331052872e-05,
      "loss": 0.4739,
      "step": 22932
    },
    {
      "epoch": 7.241237764445848,
      "grad_norm": 5.873910903930664,
      "learning_rate": 3.5441006209872645e-05,
      "loss": 0.3699,
      "step": 22933
    },
    {
      "epoch": 7.241553520682033,
      "grad_norm": 5.6207356452941895,
      "learning_rate": 3.5440129109216576e-05,
      "loss": 0.4909,
      "step": 22934
    },
    {
      "epoch": 7.241869276918219,
      "grad_norm": 5.409909248352051,
      "learning_rate": 3.54392520085605e-05,
      "loss": 0.4264,
      "step": 22935
    },
    {
      "epoch": 7.242185033154405,
      "grad_norm": 6.175648212432861,
      "learning_rate": 3.543837490790443e-05,
      "loss": 0.4478,
      "step": 22936
    },
    {
      "epoch": 7.242500789390591,
      "grad_norm": 5.212260723114014,
      "learning_rate": 3.543749780724836e-05,
      "loss": 0.5595,
      "step": 22937
    },
    {
      "epoch": 7.242816545626776,
      "grad_norm": 5.241497039794922,
      "learning_rate": 3.543662070659229e-05,
      "loss": 0.4416,
      "step": 22938
    },
    {
      "epoch": 7.243132301862961,
      "grad_norm": 6.665451526641846,
      "learning_rate": 3.5435743605936214e-05,
      "loss": 0.5296,
      "step": 22939
    },
    {
      "epoch": 7.243448058099148,
      "grad_norm": 6.0650458335876465,
      "learning_rate": 3.5434866505280146e-05,
      "loss": 0.4994,
      "step": 22940
    },
    {
      "epoch": 7.243763814335333,
      "grad_norm": 6.7911376953125,
      "learning_rate": 3.543398940462407e-05,
      "loss": 0.4389,
      "step": 22941
    },
    {
      "epoch": 7.244079570571519,
      "grad_norm": 5.120075225830078,
      "learning_rate": 3.5433112303968e-05,
      "loss": 0.5525,
      "step": 22942
    },
    {
      "epoch": 7.244395326807704,
      "grad_norm": 5.79120397567749,
      "learning_rate": 3.5432235203311934e-05,
      "loss": 0.5038,
      "step": 22943
    },
    {
      "epoch": 7.2447110830438906,
      "grad_norm": 5.709569454193115,
      "learning_rate": 3.5431358102655866e-05,
      "loss": 0.498,
      "step": 22944
    },
    {
      "epoch": 7.245026839280076,
      "grad_norm": 6.7858076095581055,
      "learning_rate": 3.543048100199979e-05,
      "loss": 0.5741,
      "step": 22945
    },
    {
      "epoch": 7.245342595516261,
      "grad_norm": 6.534295082092285,
      "learning_rate": 3.542960390134372e-05,
      "loss": 0.6797,
      "step": 22946
    },
    {
      "epoch": 7.245658351752447,
      "grad_norm": 6.618473052978516,
      "learning_rate": 3.542872680068765e-05,
      "loss": 0.4928,
      "step": 22947
    },
    {
      "epoch": 7.245974107988633,
      "grad_norm": 5.413226127624512,
      "learning_rate": 3.542784970003158e-05,
      "loss": 0.5926,
      "step": 22948
    },
    {
      "epoch": 7.246289864224819,
      "grad_norm": 5.887002944946289,
      "learning_rate": 3.5426972599375504e-05,
      "loss": 0.3627,
      "step": 22949
    },
    {
      "epoch": 7.246605620461004,
      "grad_norm": 5.63049840927124,
      "learning_rate": 3.5426095498719436e-05,
      "loss": 0.5078,
      "step": 22950
    },
    {
      "epoch": 7.246921376697189,
      "grad_norm": 5.306499004364014,
      "learning_rate": 3.542521839806337e-05,
      "loss": 0.4324,
      "step": 22951
    },
    {
      "epoch": 7.247237132933376,
      "grad_norm": 5.458683490753174,
      "learning_rate": 3.542434129740729e-05,
      "loss": 0.5102,
      "step": 22952
    },
    {
      "epoch": 7.247552889169561,
      "grad_norm": 5.250447750091553,
      "learning_rate": 3.5423464196751224e-05,
      "loss": 0.5058,
      "step": 22953
    },
    {
      "epoch": 7.247868645405747,
      "grad_norm": 6.035377502441406,
      "learning_rate": 3.542258709609515e-05,
      "loss": 0.4303,
      "step": 22954
    },
    {
      "epoch": 7.248184401641932,
      "grad_norm": 6.541157245635986,
      "learning_rate": 3.542170999543908e-05,
      "loss": 0.3794,
      "step": 22955
    },
    {
      "epoch": 7.2485001578781185,
      "grad_norm": 6.098724842071533,
      "learning_rate": 3.5420832894783006e-05,
      "loss": 0.4439,
      "step": 22956
    },
    {
      "epoch": 7.248815914114304,
      "grad_norm": 4.675187110900879,
      "learning_rate": 3.541995579412694e-05,
      "loss": 0.4161,
      "step": 22957
    },
    {
      "epoch": 7.249131670350489,
      "grad_norm": 6.0889716148376465,
      "learning_rate": 3.541907869347086e-05,
      "loss": 0.4935,
      "step": 22958
    },
    {
      "epoch": 7.249447426586675,
      "grad_norm": 4.539684772491455,
      "learning_rate": 3.5418201592814794e-05,
      "loss": 0.4633,
      "step": 22959
    },
    {
      "epoch": 7.249763182822861,
      "grad_norm": 5.649630546569824,
      "learning_rate": 3.541732449215872e-05,
      "loss": 0.4942,
      "step": 22960
    },
    {
      "epoch": 7.250078939059047,
      "grad_norm": 4.394096374511719,
      "learning_rate": 3.541644739150265e-05,
      "loss": 0.3614,
      "step": 22961
    },
    {
      "epoch": 7.250394695295232,
      "grad_norm": 5.174762725830078,
      "learning_rate": 3.5415570290846575e-05,
      "loss": 0.4271,
      "step": 22962
    },
    {
      "epoch": 7.250710451531417,
      "grad_norm": 4.6917619705200195,
      "learning_rate": 3.541469319019051e-05,
      "loss": 0.4291,
      "step": 22963
    },
    {
      "epoch": 7.2510262077676035,
      "grad_norm": 4.665046215057373,
      "learning_rate": 3.541381608953443e-05,
      "loss": 0.3496,
      "step": 22964
    },
    {
      "epoch": 7.251341964003789,
      "grad_norm": 3.9081687927246094,
      "learning_rate": 3.5412938988878364e-05,
      "loss": 0.4595,
      "step": 22965
    },
    {
      "epoch": 7.251657720239975,
      "grad_norm": 5.764062881469727,
      "learning_rate": 3.5412061888222295e-05,
      "loss": 0.5169,
      "step": 22966
    },
    {
      "epoch": 7.25197347647616,
      "grad_norm": 5.805572509765625,
      "learning_rate": 3.541118478756622e-05,
      "loss": 0.4538,
      "step": 22967
    },
    {
      "epoch": 7.2522892327123465,
      "grad_norm": 6.787797451019287,
      "learning_rate": 3.541030768691015e-05,
      "loss": 0.5794,
      "step": 22968
    },
    {
      "epoch": 7.252604988948532,
      "grad_norm": 5.371757984161377,
      "learning_rate": 3.540943058625408e-05,
      "loss": 0.469,
      "step": 22969
    },
    {
      "epoch": 7.252920745184717,
      "grad_norm": 4.542624473571777,
      "learning_rate": 3.540855348559801e-05,
      "loss": 0.4682,
      "step": 22970
    },
    {
      "epoch": 7.253236501420903,
      "grad_norm": 5.609297275543213,
      "learning_rate": 3.540767638494194e-05,
      "loss": 0.5102,
      "step": 22971
    },
    {
      "epoch": 7.2535522576570886,
      "grad_norm": 5.146568298339844,
      "learning_rate": 3.540679928428587e-05,
      "loss": 0.3601,
      "step": 22972
    },
    {
      "epoch": 7.253868013893275,
      "grad_norm": 4.95853328704834,
      "learning_rate": 3.54059221836298e-05,
      "loss": 0.4364,
      "step": 22973
    },
    {
      "epoch": 7.25418377012946,
      "grad_norm": 4.9397125244140625,
      "learning_rate": 3.540504508297373e-05,
      "loss": 0.544,
      "step": 22974
    },
    {
      "epoch": 7.254499526365645,
      "grad_norm": 5.7981743812561035,
      "learning_rate": 3.540416798231765e-05,
      "loss": 0.5044,
      "step": 22975
    },
    {
      "epoch": 7.2548152826018315,
      "grad_norm": 5.713344097137451,
      "learning_rate": 3.5403290881661585e-05,
      "loss": 0.4286,
      "step": 22976
    },
    {
      "epoch": 7.255131038838017,
      "grad_norm": 6.014986991882324,
      "learning_rate": 3.540241378100551e-05,
      "loss": 0.417,
      "step": 22977
    },
    {
      "epoch": 7.255446795074203,
      "grad_norm": 4.127355575561523,
      "learning_rate": 3.540153668034944e-05,
      "loss": 0.4736,
      "step": 22978
    },
    {
      "epoch": 7.255762551310388,
      "grad_norm": 8.398299217224121,
      "learning_rate": 3.5400659579693367e-05,
      "loss": 0.4963,
      "step": 22979
    },
    {
      "epoch": 7.2560783075465745,
      "grad_norm": 6.344231128692627,
      "learning_rate": 3.53997824790373e-05,
      "loss": 0.403,
      "step": 22980
    },
    {
      "epoch": 7.25639406378276,
      "grad_norm": 4.492664337158203,
      "learning_rate": 3.539890537838122e-05,
      "loss": 0.3845,
      "step": 22981
    },
    {
      "epoch": 7.256709820018945,
      "grad_norm": 6.2412567138671875,
      "learning_rate": 3.5398028277725155e-05,
      "loss": 0.4739,
      "step": 22982
    },
    {
      "epoch": 7.257025576255131,
      "grad_norm": 5.633768558502197,
      "learning_rate": 3.539715117706908e-05,
      "loss": 0.4428,
      "step": 22983
    },
    {
      "epoch": 7.2573413324913165,
      "grad_norm": 6.704285144805908,
      "learning_rate": 3.539627407641301e-05,
      "loss": 0.448,
      "step": 22984
    },
    {
      "epoch": 7.257657088727503,
      "grad_norm": 5.169938087463379,
      "learning_rate": 3.5395396975756936e-05,
      "loss": 0.4316,
      "step": 22985
    },
    {
      "epoch": 7.257972844963688,
      "grad_norm": 4.7060866355896,
      "learning_rate": 3.539451987510087e-05,
      "loss": 0.4146,
      "step": 22986
    },
    {
      "epoch": 7.258288601199873,
      "grad_norm": 4.915277004241943,
      "learning_rate": 3.53936427744448e-05,
      "loss": 0.4311,
      "step": 22987
    },
    {
      "epoch": 7.2586043574360595,
      "grad_norm": 7.674006462097168,
      "learning_rate": 3.5392765673788725e-05,
      "loss": 0.5103,
      "step": 22988
    },
    {
      "epoch": 7.258920113672245,
      "grad_norm": 4.979389190673828,
      "learning_rate": 3.5391888573132656e-05,
      "loss": 0.4478,
      "step": 22989
    },
    {
      "epoch": 7.259235869908431,
      "grad_norm": 7.501252174377441,
      "learning_rate": 3.539101147247658e-05,
      "loss": 0.4436,
      "step": 22990
    },
    {
      "epoch": 7.259551626144616,
      "grad_norm": 7.235522270202637,
      "learning_rate": 3.539013437182051e-05,
      "loss": 0.446,
      "step": 22991
    },
    {
      "epoch": 7.259867382380802,
      "grad_norm": 5.1144185066223145,
      "learning_rate": 3.538925727116444e-05,
      "loss": 0.5415,
      "step": 22992
    },
    {
      "epoch": 7.260183138616988,
      "grad_norm": 4.779536724090576,
      "learning_rate": 3.538838017050837e-05,
      "loss": 0.4683,
      "step": 22993
    },
    {
      "epoch": 7.260498894853173,
      "grad_norm": 4.568725109100342,
      "learning_rate": 3.5387503069852294e-05,
      "loss": 0.5252,
      "step": 22994
    },
    {
      "epoch": 7.260814651089359,
      "grad_norm": 4.6597771644592285,
      "learning_rate": 3.5386625969196226e-05,
      "loss": 0.3911,
      "step": 22995
    },
    {
      "epoch": 7.2611304073255445,
      "grad_norm": 7.239017963409424,
      "learning_rate": 3.538574886854015e-05,
      "loss": 0.4679,
      "step": 22996
    },
    {
      "epoch": 7.261446163561731,
      "grad_norm": 4.090285778045654,
      "learning_rate": 3.538487176788408e-05,
      "loss": 0.4017,
      "step": 22997
    },
    {
      "epoch": 7.261761919797916,
      "grad_norm": 5.24363374710083,
      "learning_rate": 3.5383994667228014e-05,
      "loss": 0.5566,
      "step": 22998
    },
    {
      "epoch": 7.262077676034101,
      "grad_norm": 4.9766693115234375,
      "learning_rate": 3.5383117566571946e-05,
      "loss": 0.4986,
      "step": 22999
    },
    {
      "epoch": 7.262393432270287,
      "grad_norm": 5.788210391998291,
      "learning_rate": 3.538224046591587e-05,
      "loss": 0.5044,
      "step": 23000
    },
    {
      "epoch": 7.262393432270287,
      "eval_loss": 0.9590301513671875,
      "eval_runtime": 4.5292,
      "eval_samples_per_second": 549.767,
      "eval_steps_per_second": 4.637,
      "step": 23000
    },
    {
      "epoch": 7.262709188506473,
      "grad_norm": 5.89581298828125,
      "learning_rate": 3.53813633652598e-05,
      "loss": 0.4433,
      "step": 23001
    },
    {
      "epoch": 7.263024944742659,
      "grad_norm": 6.096266746520996,
      "learning_rate": 3.538048626460373e-05,
      "loss": 0.4982,
      "step": 23002
    },
    {
      "epoch": 7.263340700978844,
      "grad_norm": 6.978667736053467,
      "learning_rate": 3.537960916394766e-05,
      "loss": 0.4363,
      "step": 23003
    },
    {
      "epoch": 7.26365645721503,
      "grad_norm": 5.248255729675293,
      "learning_rate": 3.5378732063291584e-05,
      "loss": 0.3456,
      "step": 23004
    },
    {
      "epoch": 7.263972213451216,
      "grad_norm": 6.1576247215271,
      "learning_rate": 3.5377854962635516e-05,
      "loss": 0.487,
      "step": 23005
    },
    {
      "epoch": 7.264287969687401,
      "grad_norm": 6.948305130004883,
      "learning_rate": 3.537697786197944e-05,
      "loss": 0.5447,
      "step": 23006
    },
    {
      "epoch": 7.264603725923587,
      "grad_norm": 6.720223903656006,
      "learning_rate": 3.537610076132337e-05,
      "loss": 0.5267,
      "step": 23007
    },
    {
      "epoch": 7.2649194821597725,
      "grad_norm": 5.753552436828613,
      "learning_rate": 3.53752236606673e-05,
      "loss": 0.4936,
      "step": 23008
    },
    {
      "epoch": 7.265235238395959,
      "grad_norm": 5.9787421226501465,
      "learning_rate": 3.537434656001123e-05,
      "loss": 0.5135,
      "step": 23009
    },
    {
      "epoch": 7.265550994632144,
      "grad_norm": 6.611588478088379,
      "learning_rate": 3.537346945935516e-05,
      "loss": 0.3832,
      "step": 23010
    },
    {
      "epoch": 7.26586675086833,
      "grad_norm": 3.8540751934051514,
      "learning_rate": 3.5372592358699085e-05,
      "loss": 0.3882,
      "step": 23011
    },
    {
      "epoch": 7.266182507104515,
      "grad_norm": 8.088850975036621,
      "learning_rate": 3.537171525804302e-05,
      "loss": 0.464,
      "step": 23012
    },
    {
      "epoch": 7.266498263340701,
      "grad_norm": 6.650431156158447,
      "learning_rate": 3.537083815738694e-05,
      "loss": 0.5502,
      "step": 23013
    },
    {
      "epoch": 7.266814019576887,
      "grad_norm": 5.8567891120910645,
      "learning_rate": 3.5369961056730874e-05,
      "loss": 0.5298,
      "step": 23014
    },
    {
      "epoch": 7.267129775813072,
      "grad_norm": 5.878270149230957,
      "learning_rate": 3.53690839560748e-05,
      "loss": 0.4149,
      "step": 23015
    },
    {
      "epoch": 7.267445532049258,
      "grad_norm": 5.794994831085205,
      "learning_rate": 3.536820685541873e-05,
      "loss": 0.5114,
      "step": 23016
    },
    {
      "epoch": 7.267761288285444,
      "grad_norm": 6.977208137512207,
      "learning_rate": 3.5367329754762655e-05,
      "loss": 0.457,
      "step": 23017
    },
    {
      "epoch": 7.268077044521629,
      "grad_norm": 5.894834995269775,
      "learning_rate": 3.536645265410659e-05,
      "loss": 0.4969,
      "step": 23018
    },
    {
      "epoch": 7.268392800757815,
      "grad_norm": 4.914224147796631,
      "learning_rate": 3.536557555345051e-05,
      "loss": 0.4674,
      "step": 23019
    },
    {
      "epoch": 7.268708556994,
      "grad_norm": 6.177527904510498,
      "learning_rate": 3.5364698452794444e-05,
      "loss": 0.4077,
      "step": 23020
    },
    {
      "epoch": 7.269024313230187,
      "grad_norm": 5.071863174438477,
      "learning_rate": 3.536382135213837e-05,
      "loss": 0.402,
      "step": 23021
    },
    {
      "epoch": 7.269340069466372,
      "grad_norm": 6.154098033905029,
      "learning_rate": 3.53629442514823e-05,
      "loss": 0.4933,
      "step": 23022
    },
    {
      "epoch": 7.269655825702557,
      "grad_norm": 4.0718817710876465,
      "learning_rate": 3.5362067150826225e-05,
      "loss": 0.4506,
      "step": 23023
    },
    {
      "epoch": 7.269971581938743,
      "grad_norm": 6.231880187988281,
      "learning_rate": 3.536119005017016e-05,
      "loss": 0.5719,
      "step": 23024
    },
    {
      "epoch": 7.270287338174929,
      "grad_norm": 5.308575630187988,
      "learning_rate": 3.536031294951409e-05,
      "loss": 0.4586,
      "step": 23025
    },
    {
      "epoch": 7.270603094411115,
      "grad_norm": 5.031778812408447,
      "learning_rate": 3.535943584885802e-05,
      "loss": 0.418,
      "step": 23026
    },
    {
      "epoch": 7.2709188506473,
      "grad_norm": 5.006193161010742,
      "learning_rate": 3.5358558748201945e-05,
      "loss": 0.4003,
      "step": 23027
    },
    {
      "epoch": 7.271234606883486,
      "grad_norm": 5.814734935760498,
      "learning_rate": 3.535768164754588e-05,
      "loss": 0.538,
      "step": 23028
    },
    {
      "epoch": 7.271550363119672,
      "grad_norm": 6.235574722290039,
      "learning_rate": 3.53568045468898e-05,
      "loss": 0.4186,
      "step": 23029
    },
    {
      "epoch": 7.271866119355857,
      "grad_norm": 6.137280464172363,
      "learning_rate": 3.535592744623373e-05,
      "loss": 0.4078,
      "step": 23030
    },
    {
      "epoch": 7.272181875592043,
      "grad_norm": 5.498967170715332,
      "learning_rate": 3.5355050345577665e-05,
      "loss": 0.4761,
      "step": 23031
    },
    {
      "epoch": 7.272497631828228,
      "grad_norm": 4.781091213226318,
      "learning_rate": 3.535417324492159e-05,
      "loss": 0.4234,
      "step": 23032
    },
    {
      "epoch": 7.272813388064415,
      "grad_norm": 6.216729640960693,
      "learning_rate": 3.535329614426552e-05,
      "loss": 0.4744,
      "step": 23033
    },
    {
      "epoch": 7.2731291443006,
      "grad_norm": 7.08507776260376,
      "learning_rate": 3.5352419043609446e-05,
      "loss": 0.461,
      "step": 23034
    },
    {
      "epoch": 7.273444900536786,
      "grad_norm": 5.848609447479248,
      "learning_rate": 3.535154194295338e-05,
      "loss": 0.4827,
      "step": 23035
    },
    {
      "epoch": 7.273760656772971,
      "grad_norm": 4.923248767852783,
      "learning_rate": 3.53506648422973e-05,
      "loss": 0.4457,
      "step": 23036
    },
    {
      "epoch": 7.274076413009157,
      "grad_norm": 7.096045970916748,
      "learning_rate": 3.5349787741641235e-05,
      "loss": 0.5502,
      "step": 23037
    },
    {
      "epoch": 7.274392169245343,
      "grad_norm": 8.385225296020508,
      "learning_rate": 3.534891064098516e-05,
      "loss": 0.6322,
      "step": 23038
    },
    {
      "epoch": 7.274707925481528,
      "grad_norm": 5.425881385803223,
      "learning_rate": 3.534803354032909e-05,
      "loss": 0.5135,
      "step": 23039
    },
    {
      "epoch": 7.275023681717714,
      "grad_norm": 4.912808895111084,
      "learning_rate": 3.5347156439673016e-05,
      "loss": 0.4021,
      "step": 23040
    },
    {
      "epoch": 7.2753394379539,
      "grad_norm": 5.620456218719482,
      "learning_rate": 3.534627933901695e-05,
      "loss": 0.3991,
      "step": 23041
    },
    {
      "epoch": 7.275655194190085,
      "grad_norm": 7.760956764221191,
      "learning_rate": 3.534540223836087e-05,
      "loss": 0.5123,
      "step": 23042
    },
    {
      "epoch": 7.275970950426271,
      "grad_norm": 4.384390830993652,
      "learning_rate": 3.5344525137704804e-05,
      "loss": 0.5061,
      "step": 23043
    },
    {
      "epoch": 7.276286706662456,
      "grad_norm": 4.874774932861328,
      "learning_rate": 3.534364803704873e-05,
      "loss": 0.434,
      "step": 23044
    },
    {
      "epoch": 7.2766024628986425,
      "grad_norm": 6.164465427398682,
      "learning_rate": 3.534277093639266e-05,
      "loss": 0.4966,
      "step": 23045
    },
    {
      "epoch": 7.276918219134828,
      "grad_norm": 4.667741775512695,
      "learning_rate": 3.5341893835736586e-05,
      "loss": 0.4042,
      "step": 23046
    },
    {
      "epoch": 7.277233975371013,
      "grad_norm": 5.891613483428955,
      "learning_rate": 3.534101673508052e-05,
      "loss": 0.3734,
      "step": 23047
    },
    {
      "epoch": 7.277549731607199,
      "grad_norm": 5.310751438140869,
      "learning_rate": 3.534013963442445e-05,
      "loss": 0.439,
      "step": 23048
    },
    {
      "epoch": 7.277865487843385,
      "grad_norm": 5.414514064788818,
      "learning_rate": 3.5339262533768374e-05,
      "loss": 0.4333,
      "step": 23049
    },
    {
      "epoch": 7.278181244079571,
      "grad_norm": 4.89165735244751,
      "learning_rate": 3.5338385433112306e-05,
      "loss": 0.3784,
      "step": 23050
    },
    {
      "epoch": 7.278497000315756,
      "grad_norm": 6.414231777191162,
      "learning_rate": 3.533750833245623e-05,
      "loss": 0.4914,
      "step": 23051
    },
    {
      "epoch": 7.278812756551942,
      "grad_norm": 6.093448162078857,
      "learning_rate": 3.533663123180016e-05,
      "loss": 0.4686,
      "step": 23052
    },
    {
      "epoch": 7.279128512788128,
      "grad_norm": 6.1447672843933105,
      "learning_rate": 3.5335754131144094e-05,
      "loss": 0.4286,
      "step": 23053
    },
    {
      "epoch": 7.279444269024313,
      "grad_norm": 5.999521255493164,
      "learning_rate": 3.5334877030488026e-05,
      "loss": 0.5486,
      "step": 23054
    },
    {
      "epoch": 7.279760025260499,
      "grad_norm": 4.3441009521484375,
      "learning_rate": 3.533399992983195e-05,
      "loss": 0.3806,
      "step": 23055
    },
    {
      "epoch": 7.280075781496684,
      "grad_norm": 5.57848596572876,
      "learning_rate": 3.533312282917588e-05,
      "loss": 0.4864,
      "step": 23056
    },
    {
      "epoch": 7.2803915377328705,
      "grad_norm": 4.293830871582031,
      "learning_rate": 3.533224572851981e-05,
      "loss": 0.4743,
      "step": 23057
    },
    {
      "epoch": 7.280707293969056,
      "grad_norm": 5.668155670166016,
      "learning_rate": 3.533136862786374e-05,
      "loss": 0.4435,
      "step": 23058
    },
    {
      "epoch": 7.281023050205242,
      "grad_norm": 7.264223098754883,
      "learning_rate": 3.5330491527207664e-05,
      "loss": 0.5944,
      "step": 23059
    },
    {
      "epoch": 7.281338806441427,
      "grad_norm": 5.6368889808654785,
      "learning_rate": 3.5329614426551596e-05,
      "loss": 0.5319,
      "step": 23060
    },
    {
      "epoch": 7.281654562677613,
      "grad_norm": 5.113399028778076,
      "learning_rate": 3.532873732589552e-05,
      "loss": 0.4819,
      "step": 23061
    },
    {
      "epoch": 7.281970318913799,
      "grad_norm": 5.8428850173950195,
      "learning_rate": 3.532786022523945e-05,
      "loss": 0.4293,
      "step": 23062
    },
    {
      "epoch": 7.282286075149984,
      "grad_norm": 5.950947284698486,
      "learning_rate": 3.532698312458338e-05,
      "loss": 0.568,
      "step": 23063
    },
    {
      "epoch": 7.28260183138617,
      "grad_norm": 4.333217144012451,
      "learning_rate": 3.532610602392731e-05,
      "loss": 0.3746,
      "step": 23064
    },
    {
      "epoch": 7.2829175876223555,
      "grad_norm": 4.856252193450928,
      "learning_rate": 3.5325228923271234e-05,
      "loss": 0.4049,
      "step": 23065
    },
    {
      "epoch": 7.283233343858541,
      "grad_norm": 6.001450061798096,
      "learning_rate": 3.5324351822615165e-05,
      "loss": 0.5236,
      "step": 23066
    },
    {
      "epoch": 7.283549100094727,
      "grad_norm": 5.712525844573975,
      "learning_rate": 3.532347472195909e-05,
      "loss": 0.4884,
      "step": 23067
    },
    {
      "epoch": 7.283864856330912,
      "grad_norm": 6.342641353607178,
      "learning_rate": 3.532259762130302e-05,
      "loss": 0.4912,
      "step": 23068
    },
    {
      "epoch": 7.2841806125670985,
      "grad_norm": 6.036072254180908,
      "learning_rate": 3.5321720520646954e-05,
      "loss": 0.4873,
      "step": 23069
    },
    {
      "epoch": 7.284496368803284,
      "grad_norm": 8.339876174926758,
      "learning_rate": 3.532084341999088e-05,
      "loss": 0.4526,
      "step": 23070
    },
    {
      "epoch": 7.284812125039469,
      "grad_norm": 4.936834335327148,
      "learning_rate": 3.531996631933481e-05,
      "loss": 0.4984,
      "step": 23071
    },
    {
      "epoch": 7.285127881275655,
      "grad_norm": 5.2698822021484375,
      "learning_rate": 3.5319089218678735e-05,
      "loss": 0.4467,
      "step": 23072
    },
    {
      "epoch": 7.2854436375118405,
      "grad_norm": 6.833589553833008,
      "learning_rate": 3.531821211802267e-05,
      "loss": 0.5435,
      "step": 23073
    },
    {
      "epoch": 7.285759393748027,
      "grad_norm": 5.663983345031738,
      "learning_rate": 3.531733501736659e-05,
      "loss": 0.4355,
      "step": 23074
    },
    {
      "epoch": 7.286075149984212,
      "grad_norm": 5.405688285827637,
      "learning_rate": 3.5316457916710523e-05,
      "loss": 0.4448,
      "step": 23075
    },
    {
      "epoch": 7.286390906220398,
      "grad_norm": 5.617763042449951,
      "learning_rate": 3.531558081605445e-05,
      "loss": 0.4596,
      "step": 23076
    },
    {
      "epoch": 7.2867066624565835,
      "grad_norm": 6.248106479644775,
      "learning_rate": 3.531470371539838e-05,
      "loss": 0.5017,
      "step": 23077
    },
    {
      "epoch": 7.287022418692769,
      "grad_norm": 6.386375904083252,
      "learning_rate": 3.5313826614742305e-05,
      "loss": 0.5146,
      "step": 23078
    },
    {
      "epoch": 7.287338174928955,
      "grad_norm": 7.733115196228027,
      "learning_rate": 3.531294951408624e-05,
      "loss": 0.5375,
      "step": 23079
    },
    {
      "epoch": 7.28765393116514,
      "grad_norm": 4.652679920196533,
      "learning_rate": 3.531207241343017e-05,
      "loss": 0.304,
      "step": 23080
    },
    {
      "epoch": 7.2879696874013264,
      "grad_norm": 5.518210411071777,
      "learning_rate": 3.53111953127741e-05,
      "loss": 0.4167,
      "step": 23081
    },
    {
      "epoch": 7.288285443637512,
      "grad_norm": 5.3524885177612305,
      "learning_rate": 3.5310318212118025e-05,
      "loss": 0.4063,
      "step": 23082
    },
    {
      "epoch": 7.288601199873698,
      "grad_norm": 5.121555805206299,
      "learning_rate": 3.5309441111461957e-05,
      "loss": 0.4323,
      "step": 23083
    },
    {
      "epoch": 7.288916956109883,
      "grad_norm": 5.562990188598633,
      "learning_rate": 3.530856401080588e-05,
      "loss": 0.4705,
      "step": 23084
    },
    {
      "epoch": 7.2892327123460685,
      "grad_norm": 6.609281063079834,
      "learning_rate": 3.530768691014981e-05,
      "loss": 0.4475,
      "step": 23085
    },
    {
      "epoch": 7.289548468582255,
      "grad_norm": 4.687401294708252,
      "learning_rate": 3.530680980949374e-05,
      "loss": 0.4635,
      "step": 23086
    },
    {
      "epoch": 7.28986422481844,
      "grad_norm": 5.523831844329834,
      "learning_rate": 3.530593270883767e-05,
      "loss": 0.3887,
      "step": 23087
    },
    {
      "epoch": 7.290179981054626,
      "grad_norm": 4.480413913726807,
      "learning_rate": 3.5305055608181595e-05,
      "loss": 0.4463,
      "step": 23088
    },
    {
      "epoch": 7.2904957372908115,
      "grad_norm": 5.898940086364746,
      "learning_rate": 3.5304178507525526e-05,
      "loss": 0.4617,
      "step": 23089
    },
    {
      "epoch": 7.290811493526997,
      "grad_norm": 5.597124099731445,
      "learning_rate": 3.530330140686945e-05,
      "loss": 0.4867,
      "step": 23090
    },
    {
      "epoch": 7.291127249763183,
      "grad_norm": 6.286357402801514,
      "learning_rate": 3.530242430621338e-05,
      "loss": 0.5473,
      "step": 23091
    },
    {
      "epoch": 7.291443005999368,
      "grad_norm": 5.4518632888793945,
      "learning_rate": 3.5301547205557315e-05,
      "loss": 0.5069,
      "step": 23092
    },
    {
      "epoch": 7.291758762235554,
      "grad_norm": 6.773139953613281,
      "learning_rate": 3.530067010490124e-05,
      "loss": 0.4744,
      "step": 23093
    },
    {
      "epoch": 7.29207451847174,
      "grad_norm": 5.988358497619629,
      "learning_rate": 3.529979300424517e-05,
      "loss": 0.459,
      "step": 23094
    },
    {
      "epoch": 7.292390274707926,
      "grad_norm": 4.930373191833496,
      "learning_rate": 3.5298915903589096e-05,
      "loss": 0.4199,
      "step": 23095
    },
    {
      "epoch": 7.292706030944111,
      "grad_norm": 5.072908878326416,
      "learning_rate": 3.529803880293303e-05,
      "loss": 0.494,
      "step": 23096
    },
    {
      "epoch": 7.2930217871802965,
      "grad_norm": 5.752626419067383,
      "learning_rate": 3.529716170227695e-05,
      "loss": 0.5317,
      "step": 23097
    },
    {
      "epoch": 7.293337543416483,
      "grad_norm": 5.07020902633667,
      "learning_rate": 3.5296284601620884e-05,
      "loss": 0.4893,
      "step": 23098
    },
    {
      "epoch": 7.293653299652668,
      "grad_norm": 5.730283737182617,
      "learning_rate": 3.529540750096481e-05,
      "loss": 0.4836,
      "step": 23099
    },
    {
      "epoch": 7.293969055888854,
      "grad_norm": 5.596224784851074,
      "learning_rate": 3.529453040030874e-05,
      "loss": 0.4502,
      "step": 23100
    },
    {
      "epoch": 7.294284812125039,
      "grad_norm": 4.391265869140625,
      "learning_rate": 3.5293653299652666e-05,
      "loss": 0.3687,
      "step": 23101
    },
    {
      "epoch": 7.294600568361225,
      "grad_norm": 6.007000923156738,
      "learning_rate": 3.52927761989966e-05,
      "loss": 0.407,
      "step": 23102
    },
    {
      "epoch": 7.294916324597411,
      "grad_norm": 6.911380767822266,
      "learning_rate": 3.529189909834052e-05,
      "loss": 0.5106,
      "step": 23103
    },
    {
      "epoch": 7.295232080833596,
      "grad_norm": 5.281103610992432,
      "learning_rate": 3.5291021997684454e-05,
      "loss": 0.4721,
      "step": 23104
    },
    {
      "epoch": 7.295547837069782,
      "grad_norm": 6.611018657684326,
      "learning_rate": 3.529014489702838e-05,
      "loss": 0.5224,
      "step": 23105
    },
    {
      "epoch": 7.295863593305968,
      "grad_norm": 7.710254192352295,
      "learning_rate": 3.528926779637231e-05,
      "loss": 0.5527,
      "step": 23106
    },
    {
      "epoch": 7.296179349542154,
      "grad_norm": 6.520291805267334,
      "learning_rate": 3.528839069571624e-05,
      "loss": 0.5343,
      "step": 23107
    },
    {
      "epoch": 7.296495105778339,
      "grad_norm": 7.410635948181152,
      "learning_rate": 3.5287513595060174e-05,
      "loss": 0.4846,
      "step": 23108
    },
    {
      "epoch": 7.2968108620145244,
      "grad_norm": 6.173615455627441,
      "learning_rate": 3.52866364944041e-05,
      "loss": 0.5275,
      "step": 23109
    },
    {
      "epoch": 7.297126618250711,
      "grad_norm": 5.8911967277526855,
      "learning_rate": 3.528575939374803e-05,
      "loss": 0.529,
      "step": 23110
    },
    {
      "epoch": 7.297442374486896,
      "grad_norm": 6.723394870758057,
      "learning_rate": 3.5284882293091956e-05,
      "loss": 0.4937,
      "step": 23111
    },
    {
      "epoch": 7.297758130723082,
      "grad_norm": 6.7318949699401855,
      "learning_rate": 3.528400519243589e-05,
      "loss": 0.5426,
      "step": 23112
    },
    {
      "epoch": 7.298073886959267,
      "grad_norm": 5.845340251922607,
      "learning_rate": 3.528312809177982e-05,
      "loss": 0.4809,
      "step": 23113
    },
    {
      "epoch": 7.298389643195453,
      "grad_norm": 4.559391498565674,
      "learning_rate": 3.5282250991123744e-05,
      "loss": 0.4843,
      "step": 23114
    },
    {
      "epoch": 7.298705399431639,
      "grad_norm": 5.900968074798584,
      "learning_rate": 3.5281373890467675e-05,
      "loss": 0.3672,
      "step": 23115
    },
    {
      "epoch": 7.299021155667824,
      "grad_norm": 5.907999515533447,
      "learning_rate": 3.52804967898116e-05,
      "loss": 0.6047,
      "step": 23116
    },
    {
      "epoch": 7.29933691190401,
      "grad_norm": 4.524370193481445,
      "learning_rate": 3.527961968915553e-05,
      "loss": 0.5464,
      "step": 23117
    },
    {
      "epoch": 7.299652668140196,
      "grad_norm": 5.536247253417969,
      "learning_rate": 3.527874258849946e-05,
      "loss": 0.5649,
      "step": 23118
    },
    {
      "epoch": 7.299968424376382,
      "grad_norm": 8.349148750305176,
      "learning_rate": 3.527786548784339e-05,
      "loss": 0.4875,
      "step": 23119
    },
    {
      "epoch": 7.300284180612567,
      "grad_norm": 5.8294878005981445,
      "learning_rate": 3.5276988387187314e-05,
      "loss": 0.5155,
      "step": 23120
    },
    {
      "epoch": 7.300599936848752,
      "grad_norm": 5.068504333496094,
      "learning_rate": 3.5276111286531245e-05,
      "loss": 0.4764,
      "step": 23121
    },
    {
      "epoch": 7.300915693084939,
      "grad_norm": 6.519462585449219,
      "learning_rate": 3.527523418587517e-05,
      "loss": 0.4475,
      "step": 23122
    },
    {
      "epoch": 7.301231449321124,
      "grad_norm": 5.53420877456665,
      "learning_rate": 3.52743570852191e-05,
      "loss": 0.5143,
      "step": 23123
    },
    {
      "epoch": 7.30154720555731,
      "grad_norm": 6.073249816894531,
      "learning_rate": 3.527347998456303e-05,
      "loss": 0.4041,
      "step": 23124
    },
    {
      "epoch": 7.301862961793495,
      "grad_norm": 6.229330539703369,
      "learning_rate": 3.527260288390696e-05,
      "loss": 0.5247,
      "step": 23125
    },
    {
      "epoch": 7.3021787180296815,
      "grad_norm": 7.22527551651001,
      "learning_rate": 3.527172578325088e-05,
      "loss": 0.5501,
      "step": 23126
    },
    {
      "epoch": 7.302494474265867,
      "grad_norm": 6.1092329025268555,
      "learning_rate": 3.5270848682594815e-05,
      "loss": 0.5077,
      "step": 23127
    },
    {
      "epoch": 7.302810230502052,
      "grad_norm": 5.579975128173828,
      "learning_rate": 3.526997158193874e-05,
      "loss": 0.5844,
      "step": 23128
    },
    {
      "epoch": 7.303125986738238,
      "grad_norm": 5.912816524505615,
      "learning_rate": 3.526909448128267e-05,
      "loss": 0.5176,
      "step": 23129
    },
    {
      "epoch": 7.303441742974424,
      "grad_norm": 4.98961877822876,
      "learning_rate": 3.52682173806266e-05,
      "loss": 0.4116,
      "step": 23130
    },
    {
      "epoch": 7.30375749921061,
      "grad_norm": 6.071042060852051,
      "learning_rate": 3.526734027997053e-05,
      "loss": 0.3821,
      "step": 23131
    },
    {
      "epoch": 7.304073255446795,
      "grad_norm": 4.900140762329102,
      "learning_rate": 3.526646317931446e-05,
      "loss": 0.4474,
      "step": 23132
    },
    {
      "epoch": 7.30438901168298,
      "grad_norm": 4.880713939666748,
      "learning_rate": 3.5265586078658385e-05,
      "loss": 0.387,
      "step": 23133
    },
    {
      "epoch": 7.304704767919167,
      "grad_norm": 5.4647369384765625,
      "learning_rate": 3.526470897800232e-05,
      "loss": 0.4607,
      "step": 23134
    },
    {
      "epoch": 7.305020524155352,
      "grad_norm": 5.53627347946167,
      "learning_rate": 3.526383187734625e-05,
      "loss": 0.4044,
      "step": 23135
    },
    {
      "epoch": 7.305336280391538,
      "grad_norm": 6.079113483428955,
      "learning_rate": 3.526295477669018e-05,
      "loss": 0.5103,
      "step": 23136
    },
    {
      "epoch": 7.305652036627723,
      "grad_norm": 5.3552961349487305,
      "learning_rate": 3.5262077676034105e-05,
      "loss": 0.3452,
      "step": 23137
    },
    {
      "epoch": 7.305967792863909,
      "grad_norm": 5.838961124420166,
      "learning_rate": 3.5261200575378036e-05,
      "loss": 0.4374,
      "step": 23138
    },
    {
      "epoch": 7.306283549100095,
      "grad_norm": 5.934899806976318,
      "learning_rate": 3.526032347472196e-05,
      "loss": 0.4573,
      "step": 23139
    },
    {
      "epoch": 7.30659930533628,
      "grad_norm": 6.4725847244262695,
      "learning_rate": 3.525944637406589e-05,
      "loss": 0.4804,
      "step": 23140
    },
    {
      "epoch": 7.306915061572466,
      "grad_norm": 5.089081287384033,
      "learning_rate": 3.525856927340982e-05,
      "loss": 0.4608,
      "step": 23141
    },
    {
      "epoch": 7.307230817808652,
      "grad_norm": 5.453039169311523,
      "learning_rate": 3.525769217275375e-05,
      "loss": 0.471,
      "step": 23142
    },
    {
      "epoch": 7.307546574044838,
      "grad_norm": 5.2683258056640625,
      "learning_rate": 3.5256815072097675e-05,
      "loss": 0.4684,
      "step": 23143
    },
    {
      "epoch": 7.307862330281023,
      "grad_norm": 5.47334623336792,
      "learning_rate": 3.5255937971441606e-05,
      "loss": 0.3989,
      "step": 23144
    },
    {
      "epoch": 7.308178086517208,
      "grad_norm": 6.096369743347168,
      "learning_rate": 3.525506087078553e-05,
      "loss": 0.4917,
      "step": 23145
    },
    {
      "epoch": 7.3084938427533945,
      "grad_norm": 4.916029453277588,
      "learning_rate": 3.525418377012946e-05,
      "loss": 0.479,
      "step": 23146
    },
    {
      "epoch": 7.30880959898958,
      "grad_norm": 5.721418857574463,
      "learning_rate": 3.525330666947339e-05,
      "loss": 0.4695,
      "step": 23147
    },
    {
      "epoch": 7.309125355225766,
      "grad_norm": 6.646650314331055,
      "learning_rate": 3.525242956881732e-05,
      "loss": 0.5534,
      "step": 23148
    },
    {
      "epoch": 7.309441111461951,
      "grad_norm": 4.7062249183654785,
      "learning_rate": 3.5251552468161244e-05,
      "loss": 0.4227,
      "step": 23149
    },
    {
      "epoch": 7.3097568676981375,
      "grad_norm": 5.887301445007324,
      "learning_rate": 3.5250675367505176e-05,
      "loss": 0.4241,
      "step": 23150
    },
    {
      "epoch": 7.310072623934323,
      "grad_norm": 6.662050724029541,
      "learning_rate": 3.524979826684911e-05,
      "loss": 0.4868,
      "step": 23151
    },
    {
      "epoch": 7.310388380170508,
      "grad_norm": 4.809962749481201,
      "learning_rate": 3.524892116619303e-05,
      "loss": 0.4465,
      "step": 23152
    },
    {
      "epoch": 7.310704136406694,
      "grad_norm": 4.351973533630371,
      "learning_rate": 3.5248044065536964e-05,
      "loss": 0.3897,
      "step": 23153
    },
    {
      "epoch": 7.3110198926428795,
      "grad_norm": 6.171013832092285,
      "learning_rate": 3.524716696488089e-05,
      "loss": 0.4422,
      "step": 23154
    },
    {
      "epoch": 7.311335648879066,
      "grad_norm": 5.671016693115234,
      "learning_rate": 3.524628986422482e-05,
      "loss": 0.4659,
      "step": 23155
    },
    {
      "epoch": 7.311651405115251,
      "grad_norm": 5.189286231994629,
      "learning_rate": 3.5245412763568746e-05,
      "loss": 0.5387,
      "step": 23156
    },
    {
      "epoch": 7.311967161351436,
      "grad_norm": 4.965160846710205,
      "learning_rate": 3.524453566291268e-05,
      "loss": 0.5751,
      "step": 23157
    },
    {
      "epoch": 7.3122829175876225,
      "grad_norm": 4.21790885925293,
      "learning_rate": 3.52436585622566e-05,
      "loss": 0.5088,
      "step": 23158
    },
    {
      "epoch": 7.312598673823808,
      "grad_norm": 5.594479560852051,
      "learning_rate": 3.5242781461600534e-05,
      "loss": 0.4859,
      "step": 23159
    },
    {
      "epoch": 7.312914430059994,
      "grad_norm": 5.736493110656738,
      "learning_rate": 3.524190436094446e-05,
      "loss": 0.5075,
      "step": 23160
    },
    {
      "epoch": 7.313230186296179,
      "grad_norm": 5.411075592041016,
      "learning_rate": 3.52410272602884e-05,
      "loss": 0.5156,
      "step": 23161
    },
    {
      "epoch": 7.313545942532365,
      "grad_norm": 4.868038177490234,
      "learning_rate": 3.524015015963232e-05,
      "loss": 0.4485,
      "step": 23162
    },
    {
      "epoch": 7.313861698768551,
      "grad_norm": 5.705883026123047,
      "learning_rate": 3.5239273058976254e-05,
      "loss": 0.4692,
      "step": 23163
    },
    {
      "epoch": 7.314177455004736,
      "grad_norm": 5.628185272216797,
      "learning_rate": 3.523839595832018e-05,
      "loss": 0.4828,
      "step": 23164
    },
    {
      "epoch": 7.314493211240922,
      "grad_norm": 6.969752788543701,
      "learning_rate": 3.523751885766411e-05,
      "loss": 0.5351,
      "step": 23165
    },
    {
      "epoch": 7.3148089674771075,
      "grad_norm": 4.192901611328125,
      "learning_rate": 3.5236641757008035e-05,
      "loss": 0.4058,
      "step": 23166
    },
    {
      "epoch": 7.315124723713294,
      "grad_norm": 6.900545597076416,
      "learning_rate": 3.523576465635197e-05,
      "loss": 0.5507,
      "step": 23167
    },
    {
      "epoch": 7.315440479949479,
      "grad_norm": 6.268242835998535,
      "learning_rate": 3.523488755569589e-05,
      "loss": 0.5314,
      "step": 23168
    },
    {
      "epoch": 7.315756236185664,
      "grad_norm": 5.920867919921875,
      "learning_rate": 3.5234010455039824e-05,
      "loss": 0.4628,
      "step": 23169
    },
    {
      "epoch": 7.3160719924218505,
      "grad_norm": 5.557586193084717,
      "learning_rate": 3.523313335438375e-05,
      "loss": 0.5174,
      "step": 23170
    },
    {
      "epoch": 7.316387748658036,
      "grad_norm": 5.078808307647705,
      "learning_rate": 3.523225625372768e-05,
      "loss": 0.5023,
      "step": 23171
    },
    {
      "epoch": 7.316703504894222,
      "grad_norm": 5.054447650909424,
      "learning_rate": 3.523137915307161e-05,
      "loss": 0.4864,
      "step": 23172
    },
    {
      "epoch": 7.317019261130407,
      "grad_norm": 5.960123062133789,
      "learning_rate": 3.523050205241554e-05,
      "loss": 0.4491,
      "step": 23173
    },
    {
      "epoch": 7.317335017366593,
      "grad_norm": 5.844254970550537,
      "learning_rate": 3.522962495175947e-05,
      "loss": 0.6562,
      "step": 23174
    },
    {
      "epoch": 7.317650773602779,
      "grad_norm": 5.268911838531494,
      "learning_rate": 3.5228747851103393e-05,
      "loss": 0.5116,
      "step": 23175
    },
    {
      "epoch": 7.317966529838964,
      "grad_norm": 7.148914813995361,
      "learning_rate": 3.5227870750447325e-05,
      "loss": 0.46,
      "step": 23176
    },
    {
      "epoch": 7.31828228607515,
      "grad_norm": 6.211874961853027,
      "learning_rate": 3.522699364979125e-05,
      "loss": 0.4596,
      "step": 23177
    },
    {
      "epoch": 7.3185980423113355,
      "grad_norm": 6.029203414916992,
      "learning_rate": 3.522611654913518e-05,
      "loss": 0.4317,
      "step": 23178
    },
    {
      "epoch": 7.318913798547522,
      "grad_norm": 5.751341819763184,
      "learning_rate": 3.522523944847911e-05,
      "loss": 0.5128,
      "step": 23179
    },
    {
      "epoch": 7.319229554783707,
      "grad_norm": 4.915703296661377,
      "learning_rate": 3.522436234782304e-05,
      "loss": 0.3919,
      "step": 23180
    },
    {
      "epoch": 7.319545311019892,
      "grad_norm": 5.961087703704834,
      "learning_rate": 3.522348524716696e-05,
      "loss": 0.573,
      "step": 23181
    },
    {
      "epoch": 7.319861067256078,
      "grad_norm": 5.9663920402526855,
      "learning_rate": 3.5222608146510895e-05,
      "loss": 0.54,
      "step": 23182
    },
    {
      "epoch": 7.320176823492264,
      "grad_norm": 4.791313171386719,
      "learning_rate": 3.522173104585482e-05,
      "loss": 0.4417,
      "step": 23183
    },
    {
      "epoch": 7.32049257972845,
      "grad_norm": 5.5576276779174805,
      "learning_rate": 3.522085394519875e-05,
      "loss": 0.3402,
      "step": 23184
    },
    {
      "epoch": 7.320808335964635,
      "grad_norm": 5.25117301940918,
      "learning_rate": 3.5219976844542676e-05,
      "loss": 0.4364,
      "step": 23185
    },
    {
      "epoch": 7.3211240922008205,
      "grad_norm": 6.964932918548584,
      "learning_rate": 3.521909974388661e-05,
      "loss": 0.5449,
      "step": 23186
    },
    {
      "epoch": 7.321439848437007,
      "grad_norm": 5.199363708496094,
      "learning_rate": 3.521822264323053e-05,
      "loss": 0.4551,
      "step": 23187
    },
    {
      "epoch": 7.321755604673192,
      "grad_norm": 5.671660900115967,
      "learning_rate": 3.521734554257447e-05,
      "loss": 0.4734,
      "step": 23188
    },
    {
      "epoch": 7.322071360909378,
      "grad_norm": 5.048471927642822,
      "learning_rate": 3.5216468441918396e-05,
      "loss": 0.472,
      "step": 23189
    },
    {
      "epoch": 7.3223871171455635,
      "grad_norm": 5.077535152435303,
      "learning_rate": 3.521559134126233e-05,
      "loss": 0.345,
      "step": 23190
    },
    {
      "epoch": 7.32270287338175,
      "grad_norm": 4.549695014953613,
      "learning_rate": 3.521471424060625e-05,
      "loss": 0.4092,
      "step": 23191
    },
    {
      "epoch": 7.323018629617935,
      "grad_norm": 5.321345329284668,
      "learning_rate": 3.5213837139950185e-05,
      "loss": 0.4986,
      "step": 23192
    },
    {
      "epoch": 7.32333438585412,
      "grad_norm": 4.3877482414245605,
      "learning_rate": 3.521296003929411e-05,
      "loss": 0.4188,
      "step": 23193
    },
    {
      "epoch": 7.323650142090306,
      "grad_norm": 5.8376665115356445,
      "learning_rate": 3.521208293863804e-05,
      "loss": 0.429,
      "step": 23194
    },
    {
      "epoch": 7.323965898326492,
      "grad_norm": 8.089978218078613,
      "learning_rate": 3.521120583798197e-05,
      "loss": 0.5394,
      "step": 23195
    },
    {
      "epoch": 7.324281654562678,
      "grad_norm": 5.237843036651611,
      "learning_rate": 3.52103287373259e-05,
      "loss": 0.48,
      "step": 23196
    },
    {
      "epoch": 7.324597410798863,
      "grad_norm": 6.847440719604492,
      "learning_rate": 3.520945163666983e-05,
      "loss": 0.5237,
      "step": 23197
    },
    {
      "epoch": 7.324913167035049,
      "grad_norm": 5.7885541915893555,
      "learning_rate": 3.5208574536013754e-05,
      "loss": 0.4282,
      "step": 23198
    },
    {
      "epoch": 7.325228923271235,
      "grad_norm": 5.1299357414245605,
      "learning_rate": 3.5207697435357686e-05,
      "loss": 0.379,
      "step": 23199
    },
    {
      "epoch": 7.32554467950742,
      "grad_norm": 6.398616790771484,
      "learning_rate": 3.520682033470161e-05,
      "loss": 0.5024,
      "step": 23200
    },
    {
      "epoch": 7.32554467950742,
      "eval_loss": 0.9576037526130676,
      "eval_runtime": 4.6429,
      "eval_samples_per_second": 536.304,
      "eval_steps_per_second": 4.523,
      "step": 23200
    },
    {
      "epoch": 7.325860435743606,
      "grad_norm": 8.113249778747559,
      "learning_rate": 3.520594323404554e-05,
      "loss": 0.5111,
      "step": 23201
    },
    {
      "epoch": 7.326176191979791,
      "grad_norm": 5.622810363769531,
      "learning_rate": 3.520506613338947e-05,
      "loss": 0.5213,
      "step": 23202
    },
    {
      "epoch": 7.326491948215978,
      "grad_norm": 4.770864009857178,
      "learning_rate": 3.52041890327334e-05,
      "loss": 0.4196,
      "step": 23203
    },
    {
      "epoch": 7.326807704452163,
      "grad_norm": 5.522693634033203,
      "learning_rate": 3.5203311932077324e-05,
      "loss": 0.4526,
      "step": 23204
    },
    {
      "epoch": 7.327123460688348,
      "grad_norm": 5.180981636047363,
      "learning_rate": 3.5202434831421256e-05,
      "loss": 0.3808,
      "step": 23205
    },
    {
      "epoch": 7.327439216924534,
      "grad_norm": 5.677543640136719,
      "learning_rate": 3.520155773076518e-05,
      "loss": 0.464,
      "step": 23206
    },
    {
      "epoch": 7.32775497316072,
      "grad_norm": 4.606962203979492,
      "learning_rate": 3.520068063010911e-05,
      "loss": 0.4267,
      "step": 23207
    },
    {
      "epoch": 7.328070729396906,
      "grad_norm": 5.712975025177002,
      "learning_rate": 3.519980352945304e-05,
      "loss": 0.3995,
      "step": 23208
    },
    {
      "epoch": 7.328386485633091,
      "grad_norm": 5.192389488220215,
      "learning_rate": 3.519892642879697e-05,
      "loss": 0.4071,
      "step": 23209
    },
    {
      "epoch": 7.328702241869277,
      "grad_norm": 5.328757286071777,
      "learning_rate": 3.51980493281409e-05,
      "loss": 0.4574,
      "step": 23210
    },
    {
      "epoch": 7.329017998105463,
      "grad_norm": 6.227831840515137,
      "learning_rate": 3.5197172227484826e-05,
      "loss": 0.4151,
      "step": 23211
    },
    {
      "epoch": 7.329333754341648,
      "grad_norm": 6.690010070800781,
      "learning_rate": 3.519629512682876e-05,
      "loss": 0.4498,
      "step": 23212
    },
    {
      "epoch": 7.329649510577834,
      "grad_norm": 5.357117652893066,
      "learning_rate": 3.519541802617268e-05,
      "loss": 0.6157,
      "step": 23213
    },
    {
      "epoch": 7.329965266814019,
      "grad_norm": 6.5299153327941895,
      "learning_rate": 3.5194540925516614e-05,
      "loss": 0.6252,
      "step": 23214
    },
    {
      "epoch": 7.330281023050206,
      "grad_norm": 4.971800804138184,
      "learning_rate": 3.519366382486054e-05,
      "loss": 0.4445,
      "step": 23215
    },
    {
      "epoch": 7.330596779286391,
      "grad_norm": 5.476785182952881,
      "learning_rate": 3.519278672420448e-05,
      "loss": 0.5137,
      "step": 23216
    },
    {
      "epoch": 7.330912535522576,
      "grad_norm": 7.736101150512695,
      "learning_rate": 3.51919096235484e-05,
      "loss": 0.4454,
      "step": 23217
    },
    {
      "epoch": 7.331228291758762,
      "grad_norm": 5.665488243103027,
      "learning_rate": 3.5191032522892334e-05,
      "loss": 0.4663,
      "step": 23218
    },
    {
      "epoch": 7.331544047994948,
      "grad_norm": 7.222865104675293,
      "learning_rate": 3.519015542223626e-05,
      "loss": 0.4417,
      "step": 23219
    },
    {
      "epoch": 7.331859804231134,
      "grad_norm": 6.8460211753845215,
      "learning_rate": 3.518927832158019e-05,
      "loss": 0.5243,
      "step": 23220
    },
    {
      "epoch": 7.332175560467319,
      "grad_norm": 6.116392135620117,
      "learning_rate": 3.5188401220924115e-05,
      "loss": 0.4403,
      "step": 23221
    },
    {
      "epoch": 7.332491316703505,
      "grad_norm": 9.579723358154297,
      "learning_rate": 3.518752412026805e-05,
      "loss": 0.5845,
      "step": 23222
    },
    {
      "epoch": 7.332807072939691,
      "grad_norm": 4.9232611656188965,
      "learning_rate": 3.518664701961197e-05,
      "loss": 0.5003,
      "step": 23223
    },
    {
      "epoch": 7.333122829175876,
      "grad_norm": 5.811882019042969,
      "learning_rate": 3.5185769918955904e-05,
      "loss": 0.5452,
      "step": 23224
    },
    {
      "epoch": 7.333438585412062,
      "grad_norm": 7.676072120666504,
      "learning_rate": 3.518489281829983e-05,
      "loss": 0.5141,
      "step": 23225
    },
    {
      "epoch": 7.333754341648247,
      "grad_norm": 5.794415473937988,
      "learning_rate": 3.518401571764376e-05,
      "loss": 0.5127,
      "step": 23226
    },
    {
      "epoch": 7.3340700978844335,
      "grad_norm": 3.9927546977996826,
      "learning_rate": 3.5183138616987685e-05,
      "loss": 0.5244,
      "step": 23227
    },
    {
      "epoch": 7.334385854120619,
      "grad_norm": 5.861671447753906,
      "learning_rate": 3.518226151633162e-05,
      "loss": 0.4558,
      "step": 23228
    },
    {
      "epoch": 7.334701610356804,
      "grad_norm": 5.546461582183838,
      "learning_rate": 3.518138441567554e-05,
      "loss": 0.53,
      "step": 23229
    },
    {
      "epoch": 7.33501736659299,
      "grad_norm": 6.410208702087402,
      "learning_rate": 3.518050731501947e-05,
      "loss": 0.5113,
      "step": 23230
    },
    {
      "epoch": 7.335333122829176,
      "grad_norm": 6.063647747039795,
      "learning_rate": 3.51796302143634e-05,
      "loss": 0.4734,
      "step": 23231
    },
    {
      "epoch": 7.335648879065362,
      "grad_norm": 6.291217803955078,
      "learning_rate": 3.517875311370733e-05,
      "loss": 0.5166,
      "step": 23232
    },
    {
      "epoch": 7.335964635301547,
      "grad_norm": 5.231227874755859,
      "learning_rate": 3.517787601305126e-05,
      "loss": 0.4344,
      "step": 23233
    },
    {
      "epoch": 7.336280391537733,
      "grad_norm": 4.666931629180908,
      "learning_rate": 3.5176998912395187e-05,
      "loss": 0.4068,
      "step": 23234
    },
    {
      "epoch": 7.3365961477739186,
      "grad_norm": 5.5091986656188965,
      "learning_rate": 3.517612181173912e-05,
      "loss": 0.5116,
      "step": 23235
    },
    {
      "epoch": 7.336911904010104,
      "grad_norm": 4.1442155838012695,
      "learning_rate": 3.517524471108304e-05,
      "loss": 0.4206,
      "step": 23236
    },
    {
      "epoch": 7.33722766024629,
      "grad_norm": 4.809803485870361,
      "learning_rate": 3.5174367610426975e-05,
      "loss": 0.4974,
      "step": 23237
    },
    {
      "epoch": 7.337543416482475,
      "grad_norm": 4.908082485198975,
      "learning_rate": 3.51734905097709e-05,
      "loss": 0.4041,
      "step": 23238
    },
    {
      "epoch": 7.3378591727186615,
      "grad_norm": 6.181991100311279,
      "learning_rate": 3.517261340911483e-05,
      "loss": 0.5331,
      "step": 23239
    },
    {
      "epoch": 7.338174928954847,
      "grad_norm": 5.158658027648926,
      "learning_rate": 3.5171736308458756e-05,
      "loss": 0.4325,
      "step": 23240
    },
    {
      "epoch": 7.338490685191033,
      "grad_norm": 4.7731733322143555,
      "learning_rate": 3.517085920780269e-05,
      "loss": 0.424,
      "step": 23241
    },
    {
      "epoch": 7.338806441427218,
      "grad_norm": 5.554385185241699,
      "learning_rate": 3.516998210714661e-05,
      "loss": 0.3725,
      "step": 23242
    },
    {
      "epoch": 7.339122197663404,
      "grad_norm": 5.192575931549072,
      "learning_rate": 3.516910500649055e-05,
      "loss": 0.4915,
      "step": 23243
    },
    {
      "epoch": 7.33943795389959,
      "grad_norm": 6.2238054275512695,
      "learning_rate": 3.5168227905834476e-05,
      "loss": 0.4214,
      "step": 23244
    },
    {
      "epoch": 7.339753710135775,
      "grad_norm": 5.240410804748535,
      "learning_rate": 3.516735080517841e-05,
      "loss": 0.4597,
      "step": 23245
    },
    {
      "epoch": 7.340069466371961,
      "grad_norm": 5.541134357452393,
      "learning_rate": 3.516647370452233e-05,
      "loss": 0.5212,
      "step": 23246
    },
    {
      "epoch": 7.3403852226081465,
      "grad_norm": 6.633351802825928,
      "learning_rate": 3.5165596603866265e-05,
      "loss": 0.4879,
      "step": 23247
    },
    {
      "epoch": 7.340700978844332,
      "grad_norm": 6.982244968414307,
      "learning_rate": 3.516471950321019e-05,
      "loss": 0.475,
      "step": 23248
    },
    {
      "epoch": 7.341016735080518,
      "grad_norm": 5.919989585876465,
      "learning_rate": 3.516384240255412e-05,
      "loss": 0.4728,
      "step": 23249
    },
    {
      "epoch": 7.341332491316703,
      "grad_norm": 5.9993438720703125,
      "learning_rate": 3.5162965301898046e-05,
      "loss": 0.5255,
      "step": 23250
    },
    {
      "epoch": 7.3416482475528895,
      "grad_norm": 5.497433185577393,
      "learning_rate": 3.516208820124198e-05,
      "loss": 0.4294,
      "step": 23251
    },
    {
      "epoch": 7.341964003789075,
      "grad_norm": 5.841831684112549,
      "learning_rate": 3.51612111005859e-05,
      "loss": 0.5658,
      "step": 23252
    },
    {
      "epoch": 7.34227976002526,
      "grad_norm": 6.24370813369751,
      "learning_rate": 3.5160333999929834e-05,
      "loss": 0.4411,
      "step": 23253
    },
    {
      "epoch": 7.342595516261446,
      "grad_norm": 5.521726131439209,
      "learning_rate": 3.5159456899273766e-05,
      "loss": 0.4257,
      "step": 23254
    },
    {
      "epoch": 7.3429112724976315,
      "grad_norm": 6.16361141204834,
      "learning_rate": 3.515857979861769e-05,
      "loss": 0.5224,
      "step": 23255
    },
    {
      "epoch": 7.343227028733818,
      "grad_norm": 6.217735290527344,
      "learning_rate": 3.515770269796162e-05,
      "loss": 0.481,
      "step": 23256
    },
    {
      "epoch": 7.343542784970003,
      "grad_norm": 6.908052921295166,
      "learning_rate": 3.515682559730555e-05,
      "loss": 0.5196,
      "step": 23257
    },
    {
      "epoch": 7.343858541206189,
      "grad_norm": 5.710214614868164,
      "learning_rate": 3.515594849664948e-05,
      "loss": 0.4862,
      "step": 23258
    },
    {
      "epoch": 7.3441742974423745,
      "grad_norm": 5.651183128356934,
      "learning_rate": 3.5155071395993404e-05,
      "loss": 0.5495,
      "step": 23259
    },
    {
      "epoch": 7.34449005367856,
      "grad_norm": 5.030603408813477,
      "learning_rate": 3.5154194295337336e-05,
      "loss": 0.4915,
      "step": 23260
    },
    {
      "epoch": 7.344805809914746,
      "grad_norm": 4.331315994262695,
      "learning_rate": 3.515331719468126e-05,
      "loss": 0.4842,
      "step": 23261
    },
    {
      "epoch": 7.345121566150931,
      "grad_norm": 4.9422149658203125,
      "learning_rate": 3.515244009402519e-05,
      "loss": 0.4951,
      "step": 23262
    },
    {
      "epoch": 7.345437322387117,
      "grad_norm": 5.485778331756592,
      "learning_rate": 3.515156299336912e-05,
      "loss": 0.4424,
      "step": 23263
    },
    {
      "epoch": 7.345753078623303,
      "grad_norm": 5.205381393432617,
      "learning_rate": 3.515068589271305e-05,
      "loss": 0.5318,
      "step": 23264
    },
    {
      "epoch": 7.346068834859489,
      "grad_norm": 5.153028964996338,
      "learning_rate": 3.5149808792056974e-05,
      "loss": 0.449,
      "step": 23265
    },
    {
      "epoch": 7.346384591095674,
      "grad_norm": 5.525399208068848,
      "learning_rate": 3.5148931691400906e-05,
      "loss": 0.538,
      "step": 23266
    },
    {
      "epoch": 7.3467003473318595,
      "grad_norm": 6.407515525817871,
      "learning_rate": 3.514805459074483e-05,
      "loss": 0.3813,
      "step": 23267
    },
    {
      "epoch": 7.347016103568046,
      "grad_norm": 5.042416095733643,
      "learning_rate": 3.514717749008876e-05,
      "loss": 0.432,
      "step": 23268
    },
    {
      "epoch": 7.347331859804231,
      "grad_norm": 5.108482837677002,
      "learning_rate": 3.514630038943269e-05,
      "loss": 0.4948,
      "step": 23269
    },
    {
      "epoch": 7.347647616040417,
      "grad_norm": 4.310375213623047,
      "learning_rate": 3.5145423288776625e-05,
      "loss": 0.4003,
      "step": 23270
    },
    {
      "epoch": 7.3479633722766025,
      "grad_norm": 4.6493964195251465,
      "learning_rate": 3.514454618812055e-05,
      "loss": 0.3701,
      "step": 23271
    },
    {
      "epoch": 7.348279128512788,
      "grad_norm": 5.041058540344238,
      "learning_rate": 3.514366908746448e-05,
      "loss": 0.4756,
      "step": 23272
    },
    {
      "epoch": 7.348594884748974,
      "grad_norm": 6.915468692779541,
      "learning_rate": 3.514279198680841e-05,
      "loss": 0.4017,
      "step": 23273
    },
    {
      "epoch": 7.348910640985159,
      "grad_norm": 5.70331335067749,
      "learning_rate": 3.514191488615234e-05,
      "loss": 0.3848,
      "step": 23274
    },
    {
      "epoch": 7.349226397221345,
      "grad_norm": 7.420917987823486,
      "learning_rate": 3.5141037785496264e-05,
      "loss": 0.5406,
      "step": 23275
    },
    {
      "epoch": 7.349542153457531,
      "grad_norm": 5.047117233276367,
      "learning_rate": 3.5140160684840195e-05,
      "loss": 0.5279,
      "step": 23276
    },
    {
      "epoch": 7.349857909693716,
      "grad_norm": 5.298003673553467,
      "learning_rate": 3.513928358418413e-05,
      "loss": 0.4783,
      "step": 23277
    },
    {
      "epoch": 7.350173665929902,
      "grad_norm": 4.843113422393799,
      "learning_rate": 3.513840648352805e-05,
      "loss": 0.4188,
      "step": 23278
    },
    {
      "epoch": 7.3504894221660875,
      "grad_norm": 6.048373699188232,
      "learning_rate": 3.5137529382871983e-05,
      "loss": 0.429,
      "step": 23279
    },
    {
      "epoch": 7.350805178402274,
      "grad_norm": 6.297832489013672,
      "learning_rate": 3.513665228221591e-05,
      "loss": 0.4987,
      "step": 23280
    },
    {
      "epoch": 7.351120934638459,
      "grad_norm": 6.566612243652344,
      "learning_rate": 3.513577518155984e-05,
      "loss": 0.5224,
      "step": 23281
    },
    {
      "epoch": 7.351436690874645,
      "grad_norm": 7.403131484985352,
      "learning_rate": 3.5134898080903765e-05,
      "loss": 0.5518,
      "step": 23282
    },
    {
      "epoch": 7.35175244711083,
      "grad_norm": 5.2338547706604,
      "learning_rate": 3.51340209802477e-05,
      "loss": 0.4339,
      "step": 23283
    },
    {
      "epoch": 7.352068203347016,
      "grad_norm": 5.27277135848999,
      "learning_rate": 3.513314387959162e-05,
      "loss": 0.4062,
      "step": 23284
    },
    {
      "epoch": 7.352383959583202,
      "grad_norm": 6.599747180938721,
      "learning_rate": 3.513226677893555e-05,
      "loss": 0.5155,
      "step": 23285
    },
    {
      "epoch": 7.352699715819387,
      "grad_norm": 4.843359470367432,
      "learning_rate": 3.513138967827948e-05,
      "loss": 0.5301,
      "step": 23286
    },
    {
      "epoch": 7.353015472055573,
      "grad_norm": 6.204495906829834,
      "learning_rate": 3.513051257762341e-05,
      "loss": 0.5393,
      "step": 23287
    },
    {
      "epoch": 7.353331228291759,
      "grad_norm": 7.214910507202148,
      "learning_rate": 3.5129635476967335e-05,
      "loss": 0.405,
      "step": 23288
    },
    {
      "epoch": 7.353646984527945,
      "grad_norm": 5.416095733642578,
      "learning_rate": 3.5128758376311266e-05,
      "loss": 0.5077,
      "step": 23289
    },
    {
      "epoch": 7.35396274076413,
      "grad_norm": 6.846424579620361,
      "learning_rate": 3.512788127565519e-05,
      "loss": 0.4738,
      "step": 23290
    },
    {
      "epoch": 7.354278497000315,
      "grad_norm": 5.350990295410156,
      "learning_rate": 3.512700417499912e-05,
      "loss": 0.5123,
      "step": 23291
    },
    {
      "epoch": 7.354594253236502,
      "grad_norm": 5.23291015625,
      "learning_rate": 3.5126127074343055e-05,
      "loss": 0.5272,
      "step": 23292
    },
    {
      "epoch": 7.354910009472687,
      "grad_norm": 5.458196640014648,
      "learning_rate": 3.512524997368698e-05,
      "loss": 0.4304,
      "step": 23293
    },
    {
      "epoch": 7.355225765708873,
      "grad_norm": 5.160516262054443,
      "learning_rate": 3.512437287303091e-05,
      "loss": 0.4562,
      "step": 23294
    },
    {
      "epoch": 7.355541521945058,
      "grad_norm": 3.9558393955230713,
      "learning_rate": 3.5123495772374836e-05,
      "loss": 0.3634,
      "step": 23295
    },
    {
      "epoch": 7.355857278181244,
      "grad_norm": 4.261580467224121,
      "learning_rate": 3.512261867171877e-05,
      "loss": 0.4634,
      "step": 23296
    },
    {
      "epoch": 7.35617303441743,
      "grad_norm": 5.3524088859558105,
      "learning_rate": 3.51217415710627e-05,
      "loss": 0.5125,
      "step": 23297
    },
    {
      "epoch": 7.356488790653615,
      "grad_norm": 4.967280864715576,
      "learning_rate": 3.512086447040663e-05,
      "loss": 0.4673,
      "step": 23298
    },
    {
      "epoch": 7.356804546889801,
      "grad_norm": 4.6671013832092285,
      "learning_rate": 3.5119987369750556e-05,
      "loss": 0.4625,
      "step": 23299
    },
    {
      "epoch": 7.357120303125987,
      "grad_norm": 6.952759265899658,
      "learning_rate": 3.511911026909449e-05,
      "loss": 0.5415,
      "step": 23300
    },
    {
      "epoch": 7.357436059362172,
      "grad_norm": 6.094881534576416,
      "learning_rate": 3.511823316843841e-05,
      "loss": 0.5564,
      "step": 23301
    },
    {
      "epoch": 7.357751815598358,
      "grad_norm": 4.3261566162109375,
      "learning_rate": 3.5117356067782344e-05,
      "loss": 0.3742,
      "step": 23302
    },
    {
      "epoch": 7.358067571834543,
      "grad_norm": 5.0647077560424805,
      "learning_rate": 3.511647896712627e-05,
      "loss": 0.4722,
      "step": 23303
    },
    {
      "epoch": 7.35838332807073,
      "grad_norm": 5.374771595001221,
      "learning_rate": 3.51156018664702e-05,
      "loss": 0.4268,
      "step": 23304
    },
    {
      "epoch": 7.358699084306915,
      "grad_norm": 5.451280117034912,
      "learning_rate": 3.5114724765814126e-05,
      "loss": 0.4419,
      "step": 23305
    },
    {
      "epoch": 7.359014840543101,
      "grad_norm": 5.621084213256836,
      "learning_rate": 3.511384766515806e-05,
      "loss": 0.5536,
      "step": 23306
    },
    {
      "epoch": 7.359330596779286,
      "grad_norm": 6.213857173919678,
      "learning_rate": 3.511297056450198e-05,
      "loss": 0.4399,
      "step": 23307
    },
    {
      "epoch": 7.359646353015472,
      "grad_norm": 5.591876029968262,
      "learning_rate": 3.5112093463845914e-05,
      "loss": 0.4795,
      "step": 23308
    },
    {
      "epoch": 7.359962109251658,
      "grad_norm": 6.307544708251953,
      "learning_rate": 3.511121636318984e-05,
      "loss": 0.5192,
      "step": 23309
    },
    {
      "epoch": 7.360277865487843,
      "grad_norm": 4.717374324798584,
      "learning_rate": 3.511033926253377e-05,
      "loss": 0.4032,
      "step": 23310
    },
    {
      "epoch": 7.360593621724029,
      "grad_norm": 5.312800884246826,
      "learning_rate": 3.5109462161877696e-05,
      "loss": 0.394,
      "step": 23311
    },
    {
      "epoch": 7.360909377960215,
      "grad_norm": 6.541585445404053,
      "learning_rate": 3.510858506122163e-05,
      "loss": 0.4811,
      "step": 23312
    },
    {
      "epoch": 7.361225134196401,
      "grad_norm": 5.846630096435547,
      "learning_rate": 3.510770796056556e-05,
      "loss": 0.5542,
      "step": 23313
    },
    {
      "epoch": 7.361540890432586,
      "grad_norm": 6.000417709350586,
      "learning_rate": 3.5106830859909484e-05,
      "loss": 0.5338,
      "step": 23314
    },
    {
      "epoch": 7.361856646668771,
      "grad_norm": 5.564364433288574,
      "learning_rate": 3.5105953759253416e-05,
      "loss": 0.483,
      "step": 23315
    },
    {
      "epoch": 7.362172402904958,
      "grad_norm": 5.965891361236572,
      "learning_rate": 3.510507665859734e-05,
      "loss": 0.4908,
      "step": 23316
    },
    {
      "epoch": 7.362488159141143,
      "grad_norm": 6.56425142288208,
      "learning_rate": 3.510419955794127e-05,
      "loss": 0.5129,
      "step": 23317
    },
    {
      "epoch": 7.362803915377329,
      "grad_norm": 5.146793365478516,
      "learning_rate": 3.51033224572852e-05,
      "loss": 0.3921,
      "step": 23318
    },
    {
      "epoch": 7.363119671613514,
      "grad_norm": 6.874203205108643,
      "learning_rate": 3.510244535662913e-05,
      "loss": 0.4089,
      "step": 23319
    },
    {
      "epoch": 7.3634354278497,
      "grad_norm": 5.08351993560791,
      "learning_rate": 3.5101568255973054e-05,
      "loss": 0.4442,
      "step": 23320
    },
    {
      "epoch": 7.363751184085886,
      "grad_norm": 4.899547100067139,
      "learning_rate": 3.5100691155316985e-05,
      "loss": 0.4293,
      "step": 23321
    },
    {
      "epoch": 7.364066940322071,
      "grad_norm": 5.672946453094482,
      "learning_rate": 3.509981405466091e-05,
      "loss": 0.4914,
      "step": 23322
    },
    {
      "epoch": 7.364382696558257,
      "grad_norm": 4.6142964363098145,
      "learning_rate": 3.509893695400484e-05,
      "loss": 0.4633,
      "step": 23323
    },
    {
      "epoch": 7.364698452794443,
      "grad_norm": 5.374863624572754,
      "learning_rate": 3.509805985334877e-05,
      "loss": 0.488,
      "step": 23324
    },
    {
      "epoch": 7.365014209030629,
      "grad_norm": 7.732840061187744,
      "learning_rate": 3.5097182752692705e-05,
      "loss": 0.5623,
      "step": 23325
    },
    {
      "epoch": 7.365329965266814,
      "grad_norm": 6.224309921264648,
      "learning_rate": 3.509630565203663e-05,
      "loss": 0.4605,
      "step": 23326
    },
    {
      "epoch": 7.365645721502999,
      "grad_norm": 4.959959030151367,
      "learning_rate": 3.509542855138056e-05,
      "loss": 0.3847,
      "step": 23327
    },
    {
      "epoch": 7.3659614777391855,
      "grad_norm": 5.873134136199951,
      "learning_rate": 3.509455145072449e-05,
      "loss": 0.5525,
      "step": 23328
    },
    {
      "epoch": 7.366277233975371,
      "grad_norm": 6.0909423828125,
      "learning_rate": 3.509367435006842e-05,
      "loss": 0.4677,
      "step": 23329
    },
    {
      "epoch": 7.366592990211557,
      "grad_norm": 5.748619079589844,
      "learning_rate": 3.5092797249412343e-05,
      "loss": 0.5085,
      "step": 23330
    },
    {
      "epoch": 7.366908746447742,
      "grad_norm": 5.76438570022583,
      "learning_rate": 3.5091920148756275e-05,
      "loss": 0.4696,
      "step": 23331
    },
    {
      "epoch": 7.3672245026839285,
      "grad_norm": 5.136319160461426,
      "learning_rate": 3.50910430481002e-05,
      "loss": 0.347,
      "step": 23332
    },
    {
      "epoch": 7.367540258920114,
      "grad_norm": 5.697685241699219,
      "learning_rate": 3.509016594744413e-05,
      "loss": 0.447,
      "step": 23333
    },
    {
      "epoch": 7.367856015156299,
      "grad_norm": 4.910641670227051,
      "learning_rate": 3.508928884678806e-05,
      "loss": 0.5099,
      "step": 23334
    },
    {
      "epoch": 7.368171771392485,
      "grad_norm": 4.48844575881958,
      "learning_rate": 3.508841174613199e-05,
      "loss": 0.4083,
      "step": 23335
    },
    {
      "epoch": 7.3684875276286705,
      "grad_norm": 5.907402992248535,
      "learning_rate": 3.508753464547592e-05,
      "loss": 0.5662,
      "step": 23336
    },
    {
      "epoch": 7.368803283864857,
      "grad_norm": 6.5952043533325195,
      "learning_rate": 3.5086657544819845e-05,
      "loss": 0.5198,
      "step": 23337
    },
    {
      "epoch": 7.369119040101042,
      "grad_norm": 5.9136433601379395,
      "learning_rate": 3.5085780444163777e-05,
      "loss": 0.4059,
      "step": 23338
    },
    {
      "epoch": 7.369434796337227,
      "grad_norm": 6.743380069732666,
      "learning_rate": 3.50849033435077e-05,
      "loss": 0.4923,
      "step": 23339
    },
    {
      "epoch": 7.3697505525734135,
      "grad_norm": 5.144359588623047,
      "learning_rate": 3.508402624285163e-05,
      "loss": 0.4232,
      "step": 23340
    },
    {
      "epoch": 7.370066308809599,
      "grad_norm": 5.386120319366455,
      "learning_rate": 3.508314914219556e-05,
      "loss": 0.4572,
      "step": 23341
    },
    {
      "epoch": 7.370382065045785,
      "grad_norm": 6.4060797691345215,
      "learning_rate": 3.508227204153949e-05,
      "loss": 0.3614,
      "step": 23342
    },
    {
      "epoch": 7.37069782128197,
      "grad_norm": 4.976682186126709,
      "learning_rate": 3.5081394940883415e-05,
      "loss": 0.4856,
      "step": 23343
    },
    {
      "epoch": 7.371013577518156,
      "grad_norm": 6.657920837402344,
      "learning_rate": 3.5080517840227346e-05,
      "loss": 0.4708,
      "step": 23344
    },
    {
      "epoch": 7.371329333754342,
      "grad_norm": 5.917774677276611,
      "learning_rate": 3.507964073957127e-05,
      "loss": 0.5534,
      "step": 23345
    },
    {
      "epoch": 7.371645089990527,
      "grad_norm": 5.629793643951416,
      "learning_rate": 3.50787636389152e-05,
      "loss": 0.5026,
      "step": 23346
    },
    {
      "epoch": 7.371960846226713,
      "grad_norm": 6.611233711242676,
      "learning_rate": 3.507788653825913e-05,
      "loss": 0.5234,
      "step": 23347
    },
    {
      "epoch": 7.3722766024628985,
      "grad_norm": 5.250622749328613,
      "learning_rate": 3.507700943760306e-05,
      "loss": 0.4799,
      "step": 23348
    },
    {
      "epoch": 7.372592358699085,
      "grad_norm": 6.91294527053833,
      "learning_rate": 3.5076132336946984e-05,
      "loss": 0.4953,
      "step": 23349
    },
    {
      "epoch": 7.37290811493527,
      "grad_norm": 5.80117130279541,
      "learning_rate": 3.5075255236290916e-05,
      "loss": 0.5123,
      "step": 23350
    },
    {
      "epoch": 7.373223871171455,
      "grad_norm": 5.731978893280029,
      "learning_rate": 3.507437813563485e-05,
      "loss": 0.5981,
      "step": 23351
    },
    {
      "epoch": 7.3735396274076415,
      "grad_norm": 4.962331295013428,
      "learning_rate": 3.507350103497878e-05,
      "loss": 0.3652,
      "step": 23352
    },
    {
      "epoch": 7.373855383643827,
      "grad_norm": 5.34874153137207,
      "learning_rate": 3.5072623934322704e-05,
      "loss": 0.4764,
      "step": 23353
    },
    {
      "epoch": 7.374171139880013,
      "grad_norm": 4.9400200843811035,
      "learning_rate": 3.5071746833666636e-05,
      "loss": 0.5126,
      "step": 23354
    },
    {
      "epoch": 7.374486896116198,
      "grad_norm": 5.208404064178467,
      "learning_rate": 3.507086973301056e-05,
      "loss": 0.4753,
      "step": 23355
    },
    {
      "epoch": 7.374802652352384,
      "grad_norm": 5.311972618103027,
      "learning_rate": 3.506999263235449e-05,
      "loss": 0.4974,
      "step": 23356
    },
    {
      "epoch": 7.37511840858857,
      "grad_norm": 5.420474529266357,
      "learning_rate": 3.5069115531698424e-05,
      "loss": 0.4849,
      "step": 23357
    },
    {
      "epoch": 7.375434164824755,
      "grad_norm": 5.273365020751953,
      "learning_rate": 3.506823843104235e-05,
      "loss": 0.4748,
      "step": 23358
    },
    {
      "epoch": 7.375749921060941,
      "grad_norm": 6.0491533279418945,
      "learning_rate": 3.506736133038628e-05,
      "loss": 0.5252,
      "step": 23359
    },
    {
      "epoch": 7.3760656772971265,
      "grad_norm": 5.152035236358643,
      "learning_rate": 3.5066484229730206e-05,
      "loss": 0.4409,
      "step": 23360
    },
    {
      "epoch": 7.376381433533313,
      "grad_norm": 4.938866138458252,
      "learning_rate": 3.506560712907414e-05,
      "loss": 0.5212,
      "step": 23361
    },
    {
      "epoch": 7.376697189769498,
      "grad_norm": 5.887657165527344,
      "learning_rate": 3.506473002841806e-05,
      "loss": 0.5773,
      "step": 23362
    },
    {
      "epoch": 7.377012946005683,
      "grad_norm": 5.639051914215088,
      "learning_rate": 3.5063852927761994e-05,
      "loss": 0.4469,
      "step": 23363
    },
    {
      "epoch": 7.377328702241869,
      "grad_norm": 5.545628070831299,
      "learning_rate": 3.506297582710592e-05,
      "loss": 0.5559,
      "step": 23364
    },
    {
      "epoch": 7.377644458478055,
      "grad_norm": 5.378509521484375,
      "learning_rate": 3.506209872644985e-05,
      "loss": 0.5771,
      "step": 23365
    },
    {
      "epoch": 7.377960214714241,
      "grad_norm": 5.041875839233398,
      "learning_rate": 3.5061221625793776e-05,
      "loss": 0.4162,
      "step": 23366
    },
    {
      "epoch": 7.378275970950426,
      "grad_norm": 4.993914604187012,
      "learning_rate": 3.506034452513771e-05,
      "loss": 0.4455,
      "step": 23367
    },
    {
      "epoch": 7.3785917271866115,
      "grad_norm": 5.023383140563965,
      "learning_rate": 3.505946742448163e-05,
      "loss": 0.5194,
      "step": 23368
    },
    {
      "epoch": 7.378907483422798,
      "grad_norm": 5.084901809692383,
      "learning_rate": 3.5058590323825564e-05,
      "loss": 0.3964,
      "step": 23369
    },
    {
      "epoch": 7.379223239658983,
      "grad_norm": 4.801366329193115,
      "learning_rate": 3.505771322316949e-05,
      "loss": 0.4784,
      "step": 23370
    },
    {
      "epoch": 7.379538995895169,
      "grad_norm": 5.413127899169922,
      "learning_rate": 3.505683612251342e-05,
      "loss": 0.4846,
      "step": 23371
    },
    {
      "epoch": 7.3798547521313544,
      "grad_norm": 5.458218097686768,
      "learning_rate": 3.5055959021857345e-05,
      "loss": 0.4535,
      "step": 23372
    },
    {
      "epoch": 7.380170508367541,
      "grad_norm": 6.7610063552856445,
      "learning_rate": 3.505508192120128e-05,
      "loss": 0.4818,
      "step": 23373
    },
    {
      "epoch": 7.380486264603726,
      "grad_norm": 4.24107027053833,
      "learning_rate": 3.505420482054521e-05,
      "loss": 0.4769,
      "step": 23374
    },
    {
      "epoch": 7.380802020839911,
      "grad_norm": 5.673830509185791,
      "learning_rate": 3.5053327719889134e-05,
      "loss": 0.5693,
      "step": 23375
    },
    {
      "epoch": 7.381117777076097,
      "grad_norm": 6.906373500823975,
      "learning_rate": 3.5052450619233065e-05,
      "loss": 0.6099,
      "step": 23376
    },
    {
      "epoch": 7.381433533312283,
      "grad_norm": 4.477110385894775,
      "learning_rate": 3.505157351857699e-05,
      "loss": 0.4986,
      "step": 23377
    },
    {
      "epoch": 7.381749289548469,
      "grad_norm": 5.198461532592773,
      "learning_rate": 3.505069641792092e-05,
      "loss": 0.4616,
      "step": 23378
    },
    {
      "epoch": 7.382065045784654,
      "grad_norm": 4.3399152755737305,
      "learning_rate": 3.5049819317264854e-05,
      "loss": 0.4093,
      "step": 23379
    },
    {
      "epoch": 7.38238080202084,
      "grad_norm": 6.171956539154053,
      "learning_rate": 3.5048942216608785e-05,
      "loss": 0.4627,
      "step": 23380
    },
    {
      "epoch": 7.382696558257026,
      "grad_norm": 4.819649696350098,
      "learning_rate": 3.504806511595271e-05,
      "loss": 0.4538,
      "step": 23381
    },
    {
      "epoch": 7.383012314493211,
      "grad_norm": 4.171048641204834,
      "learning_rate": 3.504718801529664e-05,
      "loss": 0.4182,
      "step": 23382
    },
    {
      "epoch": 7.383328070729397,
      "grad_norm": 4.947778701782227,
      "learning_rate": 3.504631091464057e-05,
      "loss": 0.5675,
      "step": 23383
    },
    {
      "epoch": 7.383643826965582,
      "grad_norm": 4.797527313232422,
      "learning_rate": 3.50454338139845e-05,
      "loss": 0.5172,
      "step": 23384
    },
    {
      "epoch": 7.383959583201769,
      "grad_norm": 3.8610103130340576,
      "learning_rate": 3.504455671332842e-05,
      "loss": 0.4188,
      "step": 23385
    },
    {
      "epoch": 7.384275339437954,
      "grad_norm": 6.228464603424072,
      "learning_rate": 3.5043679612672355e-05,
      "loss": 0.3957,
      "step": 23386
    },
    {
      "epoch": 7.384591095674139,
      "grad_norm": 6.725024223327637,
      "learning_rate": 3.504280251201628e-05,
      "loss": 0.5176,
      "step": 23387
    },
    {
      "epoch": 7.384906851910325,
      "grad_norm": 4.3560919761657715,
      "learning_rate": 3.504192541136021e-05,
      "loss": 0.4506,
      "step": 23388
    },
    {
      "epoch": 7.385222608146511,
      "grad_norm": 6.2890777587890625,
      "learning_rate": 3.5041048310704137e-05,
      "loss": 0.5535,
      "step": 23389
    },
    {
      "epoch": 7.385538364382697,
      "grad_norm": 5.894309997558594,
      "learning_rate": 3.504017121004807e-05,
      "loss": 0.409,
      "step": 23390
    },
    {
      "epoch": 7.385854120618882,
      "grad_norm": 5.987422466278076,
      "learning_rate": 3.503929410939199e-05,
      "loss": 0.5064,
      "step": 23391
    },
    {
      "epoch": 7.386169876855067,
      "grad_norm": 7.115739345550537,
      "learning_rate": 3.5038417008735925e-05,
      "loss": 0.4961,
      "step": 23392
    },
    {
      "epoch": 7.386485633091254,
      "grad_norm": 6.4153594970703125,
      "learning_rate": 3.503753990807985e-05,
      "loss": 0.4955,
      "step": 23393
    },
    {
      "epoch": 7.386801389327439,
      "grad_norm": 6.540898323059082,
      "learning_rate": 3.503666280742378e-05,
      "loss": 0.5455,
      "step": 23394
    },
    {
      "epoch": 7.387117145563625,
      "grad_norm": 5.630197048187256,
      "learning_rate": 3.503578570676771e-05,
      "loss": 0.4505,
      "step": 23395
    },
    {
      "epoch": 7.38743290179981,
      "grad_norm": 5.57900857925415,
      "learning_rate": 3.503490860611164e-05,
      "loss": 0.4636,
      "step": 23396
    },
    {
      "epoch": 7.387748658035997,
      "grad_norm": 5.966159343719482,
      "learning_rate": 3.503403150545557e-05,
      "loss": 0.5421,
      "step": 23397
    },
    {
      "epoch": 7.388064414272182,
      "grad_norm": 6.831244945526123,
      "learning_rate": 3.5033154404799495e-05,
      "loss": 0.5946,
      "step": 23398
    },
    {
      "epoch": 7.388380170508367,
      "grad_norm": 4.739068508148193,
      "learning_rate": 3.5032277304143426e-05,
      "loss": 0.4795,
      "step": 23399
    },
    {
      "epoch": 7.388695926744553,
      "grad_norm": 6.544963359832764,
      "learning_rate": 3.503140020348735e-05,
      "loss": 0.4483,
      "step": 23400
    },
    {
      "epoch": 7.388695926744553,
      "eval_loss": 0.9449676871299744,
      "eval_runtime": 4.5341,
      "eval_samples_per_second": 549.177,
      "eval_steps_per_second": 4.632,
      "step": 23400
    },
    {
      "epoch": 7.389011682980739,
      "grad_norm": 6.463405132293701,
      "learning_rate": 3.503052310283128e-05,
      "loss": 0.5556,
      "step": 23401
    },
    {
      "epoch": 7.389327439216925,
      "grad_norm": 5.794002532958984,
      "learning_rate": 3.502964600217521e-05,
      "loss": 0.5195,
      "step": 23402
    },
    {
      "epoch": 7.38964319545311,
      "grad_norm": 5.131077766418457,
      "learning_rate": 3.502876890151914e-05,
      "loss": 0.439,
      "step": 23403
    },
    {
      "epoch": 7.389958951689296,
      "grad_norm": 5.288796901702881,
      "learning_rate": 3.5027891800863064e-05,
      "loss": 0.3728,
      "step": 23404
    },
    {
      "epoch": 7.390274707925482,
      "grad_norm": 5.939654350280762,
      "learning_rate": 3.5027014700206996e-05,
      "loss": 0.4996,
      "step": 23405
    },
    {
      "epoch": 7.390590464161667,
      "grad_norm": 5.685515880584717,
      "learning_rate": 3.502613759955093e-05,
      "loss": 0.4625,
      "step": 23406
    },
    {
      "epoch": 7.390906220397853,
      "grad_norm": 5.99751091003418,
      "learning_rate": 3.502526049889486e-05,
      "loss": 0.503,
      "step": 23407
    },
    {
      "epoch": 7.391221976634038,
      "grad_norm": 4.6789774894714355,
      "learning_rate": 3.5024383398238784e-05,
      "loss": 0.411,
      "step": 23408
    },
    {
      "epoch": 7.3915377328702245,
      "grad_norm": 4.908172607421875,
      "learning_rate": 3.5023506297582716e-05,
      "loss": 0.5194,
      "step": 23409
    },
    {
      "epoch": 7.39185348910641,
      "grad_norm": 6.599206924438477,
      "learning_rate": 3.502262919692664e-05,
      "loss": 0.468,
      "step": 23410
    },
    {
      "epoch": 7.392169245342595,
      "grad_norm": 5.508307456970215,
      "learning_rate": 3.502175209627057e-05,
      "loss": 0.4549,
      "step": 23411
    },
    {
      "epoch": 7.392485001578781,
      "grad_norm": 5.415604114532471,
      "learning_rate": 3.50208749956145e-05,
      "loss": 0.4148,
      "step": 23412
    },
    {
      "epoch": 7.392800757814967,
      "grad_norm": 6.475123882293701,
      "learning_rate": 3.501999789495843e-05,
      "loss": 0.4996,
      "step": 23413
    },
    {
      "epoch": 7.393116514051153,
      "grad_norm": 4.783874988555908,
      "learning_rate": 3.5019120794302354e-05,
      "loss": 0.4823,
      "step": 23414
    },
    {
      "epoch": 7.393432270287338,
      "grad_norm": 5.853965759277344,
      "learning_rate": 3.5018243693646286e-05,
      "loss": 0.4938,
      "step": 23415
    },
    {
      "epoch": 7.393748026523524,
      "grad_norm": 4.643636703491211,
      "learning_rate": 3.501736659299021e-05,
      "loss": 0.4829,
      "step": 23416
    },
    {
      "epoch": 7.3940637827597095,
      "grad_norm": 6.161108016967773,
      "learning_rate": 3.501648949233414e-05,
      "loss": 0.5447,
      "step": 23417
    },
    {
      "epoch": 7.394379538995895,
      "grad_norm": 5.385837078094482,
      "learning_rate": 3.5015612391678074e-05,
      "loss": 0.4056,
      "step": 23418
    },
    {
      "epoch": 7.394695295232081,
      "grad_norm": 4.949510097503662,
      "learning_rate": 3.5014735291022e-05,
      "loss": 0.4743,
      "step": 23419
    },
    {
      "epoch": 7.395011051468266,
      "grad_norm": 5.337611198425293,
      "learning_rate": 3.501385819036593e-05,
      "loss": 0.4518,
      "step": 23420
    },
    {
      "epoch": 7.3953268077044525,
      "grad_norm": 6.398224830627441,
      "learning_rate": 3.5012981089709855e-05,
      "loss": 0.4903,
      "step": 23421
    },
    {
      "epoch": 7.395642563940638,
      "grad_norm": 6.575761318206787,
      "learning_rate": 3.501210398905379e-05,
      "loss": 0.4402,
      "step": 23422
    },
    {
      "epoch": 7.395958320176823,
      "grad_norm": 4.621384143829346,
      "learning_rate": 3.501122688839771e-05,
      "loss": 0.4695,
      "step": 23423
    },
    {
      "epoch": 7.396274076413009,
      "grad_norm": 5.193436622619629,
      "learning_rate": 3.5010349787741644e-05,
      "loss": 0.4856,
      "step": 23424
    },
    {
      "epoch": 7.396589832649195,
      "grad_norm": 6.253477573394775,
      "learning_rate": 3.500947268708557e-05,
      "loss": 0.5037,
      "step": 23425
    },
    {
      "epoch": 7.396905588885381,
      "grad_norm": 4.736871242523193,
      "learning_rate": 3.50085955864295e-05,
      "loss": 0.3467,
      "step": 23426
    },
    {
      "epoch": 7.397221345121566,
      "grad_norm": 5.977992057800293,
      "learning_rate": 3.5007718485773425e-05,
      "loss": 0.4614,
      "step": 23427
    },
    {
      "epoch": 7.397537101357752,
      "grad_norm": 5.281662464141846,
      "learning_rate": 3.500684138511736e-05,
      "loss": 0.5248,
      "step": 23428
    },
    {
      "epoch": 7.3978528575939375,
      "grad_norm": 4.709213733673096,
      "learning_rate": 3.500596428446128e-05,
      "loss": 0.5074,
      "step": 23429
    },
    {
      "epoch": 7.398168613830123,
      "grad_norm": 5.802579879760742,
      "learning_rate": 3.5005087183805214e-05,
      "loss": 0.5672,
      "step": 23430
    },
    {
      "epoch": 7.398484370066309,
      "grad_norm": 4.7113237380981445,
      "learning_rate": 3.500421008314914e-05,
      "loss": 0.4596,
      "step": 23431
    },
    {
      "epoch": 7.398800126302494,
      "grad_norm": 7.486184120178223,
      "learning_rate": 3.500333298249307e-05,
      "loss": 0.5293,
      "step": 23432
    },
    {
      "epoch": 7.3991158825386805,
      "grad_norm": 7.377758979797363,
      "learning_rate": 3.5002455881837e-05,
      "loss": 0.5605,
      "step": 23433
    },
    {
      "epoch": 7.399431638774866,
      "grad_norm": 5.406240463256836,
      "learning_rate": 3.5001578781180933e-05,
      "loss": 0.4575,
      "step": 23434
    },
    {
      "epoch": 7.399747395011051,
      "grad_norm": 5.578146457672119,
      "learning_rate": 3.500070168052486e-05,
      "loss": 0.509,
      "step": 23435
    },
    {
      "epoch": 7.400063151247237,
      "grad_norm": 5.559585094451904,
      "learning_rate": 3.499982457986879e-05,
      "loss": 0.5288,
      "step": 23436
    },
    {
      "epoch": 7.4003789074834225,
      "grad_norm": 5.225754737854004,
      "learning_rate": 3.4998947479212715e-05,
      "loss": 0.4473,
      "step": 23437
    },
    {
      "epoch": 7.400694663719609,
      "grad_norm": 6.168183326721191,
      "learning_rate": 3.499807037855665e-05,
      "loss": 0.4844,
      "step": 23438
    },
    {
      "epoch": 7.401010419955794,
      "grad_norm": 4.569505214691162,
      "learning_rate": 3.499719327790058e-05,
      "loss": 0.3304,
      "step": 23439
    },
    {
      "epoch": 7.40132617619198,
      "grad_norm": 5.359486103057861,
      "learning_rate": 3.49963161772445e-05,
      "loss": 0.5245,
      "step": 23440
    },
    {
      "epoch": 7.4016419324281655,
      "grad_norm": 4.702516555786133,
      "learning_rate": 3.4995439076588435e-05,
      "loss": 0.4041,
      "step": 23441
    },
    {
      "epoch": 7.401957688664351,
      "grad_norm": 5.5519866943359375,
      "learning_rate": 3.499456197593236e-05,
      "loss": 0.5422,
      "step": 23442
    },
    {
      "epoch": 7.402273444900537,
      "grad_norm": 5.825502872467041,
      "learning_rate": 3.499368487527629e-05,
      "loss": 0.5011,
      "step": 23443
    },
    {
      "epoch": 7.402589201136722,
      "grad_norm": 6.4857916831970215,
      "learning_rate": 3.4992807774620216e-05,
      "loss": 0.4578,
      "step": 23444
    },
    {
      "epoch": 7.402904957372908,
      "grad_norm": 5.491199493408203,
      "learning_rate": 3.499193067396415e-05,
      "loss": 0.3542,
      "step": 23445
    },
    {
      "epoch": 7.403220713609094,
      "grad_norm": 6.589781284332275,
      "learning_rate": 3.499105357330807e-05,
      "loss": 0.4901,
      "step": 23446
    },
    {
      "epoch": 7.40353646984528,
      "grad_norm": 6.5076446533203125,
      "learning_rate": 3.4990176472652005e-05,
      "loss": 0.5939,
      "step": 23447
    },
    {
      "epoch": 7.403852226081465,
      "grad_norm": 7.636565208435059,
      "learning_rate": 3.498929937199593e-05,
      "loss": 0.4854,
      "step": 23448
    },
    {
      "epoch": 7.4041679823176505,
      "grad_norm": 4.753048419952393,
      "learning_rate": 3.498842227133986e-05,
      "loss": 0.539,
      "step": 23449
    },
    {
      "epoch": 7.404483738553837,
      "grad_norm": 5.682832717895508,
      "learning_rate": 3.4987545170683786e-05,
      "loss": 0.4745,
      "step": 23450
    },
    {
      "epoch": 7.404799494790022,
      "grad_norm": 5.433321475982666,
      "learning_rate": 3.498666807002772e-05,
      "loss": 0.4286,
      "step": 23451
    },
    {
      "epoch": 7.405115251026208,
      "grad_norm": 5.773873329162598,
      "learning_rate": 3.498579096937164e-05,
      "loss": 0.475,
      "step": 23452
    },
    {
      "epoch": 7.4054310072623935,
      "grad_norm": 7.320188045501709,
      "learning_rate": 3.4984913868715574e-05,
      "loss": 0.4669,
      "step": 23453
    },
    {
      "epoch": 7.405746763498579,
      "grad_norm": 6.271556377410889,
      "learning_rate": 3.49840367680595e-05,
      "loss": 0.4534,
      "step": 23454
    },
    {
      "epoch": 7.406062519734765,
      "grad_norm": 5.289532661437988,
      "learning_rate": 3.498315966740343e-05,
      "loss": 0.4688,
      "step": 23455
    },
    {
      "epoch": 7.40637827597095,
      "grad_norm": 4.482885837554932,
      "learning_rate": 3.498228256674736e-05,
      "loss": 0.4599,
      "step": 23456
    },
    {
      "epoch": 7.406694032207136,
      "grad_norm": 5.103977203369141,
      "learning_rate": 3.498140546609129e-05,
      "loss": 0.488,
      "step": 23457
    },
    {
      "epoch": 7.407009788443322,
      "grad_norm": 5.474236011505127,
      "learning_rate": 3.498052836543522e-05,
      "loss": 0.5776,
      "step": 23458
    },
    {
      "epoch": 7.407325544679507,
      "grad_norm": 10.811422348022461,
      "learning_rate": 3.4979651264779144e-05,
      "loss": 0.5584,
      "step": 23459
    },
    {
      "epoch": 7.407641300915693,
      "grad_norm": 6.308874607086182,
      "learning_rate": 3.4978774164123076e-05,
      "loss": 0.5528,
      "step": 23460
    },
    {
      "epoch": 7.4079570571518785,
      "grad_norm": 8.552297592163086,
      "learning_rate": 3.497789706346701e-05,
      "loss": 0.3813,
      "step": 23461
    },
    {
      "epoch": 7.408272813388065,
      "grad_norm": 5.5949296951293945,
      "learning_rate": 3.497701996281094e-05,
      "loss": 0.4958,
      "step": 23462
    },
    {
      "epoch": 7.40858856962425,
      "grad_norm": 5.913780212402344,
      "learning_rate": 3.4976142862154864e-05,
      "loss": 0.5233,
      "step": 23463
    },
    {
      "epoch": 7.408904325860436,
      "grad_norm": 6.2590155601501465,
      "learning_rate": 3.4975265761498796e-05,
      "loss": 0.5723,
      "step": 23464
    },
    {
      "epoch": 7.409220082096621,
      "grad_norm": 5.3733134269714355,
      "learning_rate": 3.497438866084272e-05,
      "loss": 0.4695,
      "step": 23465
    },
    {
      "epoch": 7.409535838332807,
      "grad_norm": 7.4643683433532715,
      "learning_rate": 3.497351156018665e-05,
      "loss": 0.5627,
      "step": 23466
    },
    {
      "epoch": 7.409851594568993,
      "grad_norm": 8.138741493225098,
      "learning_rate": 3.497263445953058e-05,
      "loss": 0.5265,
      "step": 23467
    },
    {
      "epoch": 7.410167350805178,
      "grad_norm": 5.718353748321533,
      "learning_rate": 3.497175735887451e-05,
      "loss": 0.5776,
      "step": 23468
    },
    {
      "epoch": 7.410483107041364,
      "grad_norm": 5.292324066162109,
      "learning_rate": 3.4970880258218434e-05,
      "loss": 0.4549,
      "step": 23469
    },
    {
      "epoch": 7.41079886327755,
      "grad_norm": 6.278627872467041,
      "learning_rate": 3.4970003157562366e-05,
      "loss": 0.5668,
      "step": 23470
    },
    {
      "epoch": 7.411114619513736,
      "grad_norm": 6.572845935821533,
      "learning_rate": 3.496912605690629e-05,
      "loss": 0.4877,
      "step": 23471
    },
    {
      "epoch": 7.411430375749921,
      "grad_norm": 7.04388427734375,
      "learning_rate": 3.496824895625022e-05,
      "loss": 0.4349,
      "step": 23472
    },
    {
      "epoch": 7.411746131986106,
      "grad_norm": 5.810563564300537,
      "learning_rate": 3.496737185559415e-05,
      "loss": 0.571,
      "step": 23473
    },
    {
      "epoch": 7.412061888222293,
      "grad_norm": 4.803008556365967,
      "learning_rate": 3.496649475493808e-05,
      "loss": 0.4519,
      "step": 23474
    },
    {
      "epoch": 7.412377644458478,
      "grad_norm": 4.831153869628906,
      "learning_rate": 3.4965617654282004e-05,
      "loss": 0.4188,
      "step": 23475
    },
    {
      "epoch": 7.412693400694664,
      "grad_norm": 5.782501220703125,
      "learning_rate": 3.4964740553625935e-05,
      "loss": 0.5811,
      "step": 23476
    },
    {
      "epoch": 7.413009156930849,
      "grad_norm": 6.066633224487305,
      "learning_rate": 3.496386345296987e-05,
      "loss": 0.5228,
      "step": 23477
    },
    {
      "epoch": 7.413324913167035,
      "grad_norm": 5.053969383239746,
      "learning_rate": 3.496298635231379e-05,
      "loss": 0.4455,
      "step": 23478
    },
    {
      "epoch": 7.413640669403221,
      "grad_norm": 4.303134918212891,
      "learning_rate": 3.4962109251657724e-05,
      "loss": 0.5308,
      "step": 23479
    },
    {
      "epoch": 7.413956425639406,
      "grad_norm": 7.227946758270264,
      "learning_rate": 3.496123215100165e-05,
      "loss": 0.6322,
      "step": 23480
    },
    {
      "epoch": 7.414272181875592,
      "grad_norm": 5.913692951202393,
      "learning_rate": 3.496035505034558e-05,
      "loss": 0.5197,
      "step": 23481
    },
    {
      "epoch": 7.414587938111778,
      "grad_norm": 4.782135009765625,
      "learning_rate": 3.4959477949689505e-05,
      "loss": 0.4601,
      "step": 23482
    },
    {
      "epoch": 7.414903694347963,
      "grad_norm": 6.299245834350586,
      "learning_rate": 3.495860084903344e-05,
      "loss": 0.4197,
      "step": 23483
    },
    {
      "epoch": 7.415219450584149,
      "grad_norm": 4.934723854064941,
      "learning_rate": 3.495772374837736e-05,
      "loss": 0.4284,
      "step": 23484
    },
    {
      "epoch": 7.415535206820334,
      "grad_norm": 4.26083517074585,
      "learning_rate": 3.4956846647721293e-05,
      "loss": 0.4808,
      "step": 23485
    },
    {
      "epoch": 7.415850963056521,
      "grad_norm": 5.903797149658203,
      "learning_rate": 3.495596954706522e-05,
      "loss": 0.5679,
      "step": 23486
    },
    {
      "epoch": 7.416166719292706,
      "grad_norm": 7.318061828613281,
      "learning_rate": 3.495509244640915e-05,
      "loss": 0.5971,
      "step": 23487
    },
    {
      "epoch": 7.416482475528892,
      "grad_norm": 5.056309223175049,
      "learning_rate": 3.495421534575308e-05,
      "loss": 0.4055,
      "step": 23488
    },
    {
      "epoch": 7.416798231765077,
      "grad_norm": 3.901266098022461,
      "learning_rate": 3.495333824509701e-05,
      "loss": 0.4433,
      "step": 23489
    },
    {
      "epoch": 7.417113988001263,
      "grad_norm": 5.261961460113525,
      "learning_rate": 3.495246114444094e-05,
      "loss": 0.4165,
      "step": 23490
    },
    {
      "epoch": 7.417429744237449,
      "grad_norm": 4.493224620819092,
      "learning_rate": 3.495158404378487e-05,
      "loss": 0.3335,
      "step": 23491
    },
    {
      "epoch": 7.417745500473634,
      "grad_norm": 5.818722724914551,
      "learning_rate": 3.4950706943128795e-05,
      "loss": 0.4753,
      "step": 23492
    },
    {
      "epoch": 7.41806125670982,
      "grad_norm": 5.659721851348877,
      "learning_rate": 3.4949829842472727e-05,
      "loss": 0.4794,
      "step": 23493
    },
    {
      "epoch": 7.418377012946006,
      "grad_norm": 6.034130096435547,
      "learning_rate": 3.494895274181665e-05,
      "loss": 0.5258,
      "step": 23494
    },
    {
      "epoch": 7.418692769182192,
      "grad_norm": 4.511250972747803,
      "learning_rate": 3.494807564116058e-05,
      "loss": 0.3595,
      "step": 23495
    },
    {
      "epoch": 7.419008525418377,
      "grad_norm": 5.348628044128418,
      "learning_rate": 3.494719854050451e-05,
      "loss": 0.4665,
      "step": 23496
    },
    {
      "epoch": 7.419324281654562,
      "grad_norm": 5.618111610412598,
      "learning_rate": 3.494632143984844e-05,
      "loss": 0.4357,
      "step": 23497
    },
    {
      "epoch": 7.4196400378907486,
      "grad_norm": 5.890662670135498,
      "learning_rate": 3.494544433919237e-05,
      "loss": 0.5956,
      "step": 23498
    },
    {
      "epoch": 7.419955794126934,
      "grad_norm": 5.6634321212768555,
      "learning_rate": 3.4944567238536296e-05,
      "loss": 0.5492,
      "step": 23499
    },
    {
      "epoch": 7.42027155036312,
      "grad_norm": 6.859167575836182,
      "learning_rate": 3.494369013788023e-05,
      "loss": 0.4431,
      "step": 23500
    },
    {
      "epoch": 7.420587306599305,
      "grad_norm": 5.514556407928467,
      "learning_rate": 3.494281303722415e-05,
      "loss": 0.5503,
      "step": 23501
    },
    {
      "epoch": 7.420903062835491,
      "grad_norm": 5.0465407371521,
      "learning_rate": 3.4941935936568085e-05,
      "loss": 0.3726,
      "step": 23502
    },
    {
      "epoch": 7.421218819071677,
      "grad_norm": 5.1948628425598145,
      "learning_rate": 3.494105883591201e-05,
      "loss": 0.4814,
      "step": 23503
    },
    {
      "epoch": 7.421534575307862,
      "grad_norm": 7.096360683441162,
      "learning_rate": 3.494018173525594e-05,
      "loss": 0.5364,
      "step": 23504
    },
    {
      "epoch": 7.421850331544048,
      "grad_norm": 6.058749675750732,
      "learning_rate": 3.4939304634599866e-05,
      "loss": 0.5731,
      "step": 23505
    },
    {
      "epoch": 7.422166087780234,
      "grad_norm": 5.755774974822998,
      "learning_rate": 3.49384275339438e-05,
      "loss": 0.4306,
      "step": 23506
    },
    {
      "epoch": 7.422481844016419,
      "grad_norm": 5.494649410247803,
      "learning_rate": 3.493755043328772e-05,
      "loss": 0.5609,
      "step": 23507
    },
    {
      "epoch": 7.422797600252605,
      "grad_norm": 4.68192720413208,
      "learning_rate": 3.4936673332631654e-05,
      "loss": 0.4323,
      "step": 23508
    },
    {
      "epoch": 7.42311335648879,
      "grad_norm": 6.16024923324585,
      "learning_rate": 3.493579623197558e-05,
      "loss": 0.4389,
      "step": 23509
    },
    {
      "epoch": 7.4234291127249765,
      "grad_norm": 7.7645955085754395,
      "learning_rate": 3.493491913131951e-05,
      "loss": 0.5865,
      "step": 23510
    },
    {
      "epoch": 7.423744868961162,
      "grad_norm": 6.418163299560547,
      "learning_rate": 3.4934042030663436e-05,
      "loss": 0.5266,
      "step": 23511
    },
    {
      "epoch": 7.424060625197348,
      "grad_norm": 5.998364448547363,
      "learning_rate": 3.493316493000737e-05,
      "loss": 0.4702,
      "step": 23512
    },
    {
      "epoch": 7.424376381433533,
      "grad_norm": 5.431270122528076,
      "learning_rate": 3.493228782935129e-05,
      "loss": 0.3799,
      "step": 23513
    },
    {
      "epoch": 7.424692137669719,
      "grad_norm": 6.5997633934021,
      "learning_rate": 3.4931410728695224e-05,
      "loss": 0.4719,
      "step": 23514
    },
    {
      "epoch": 7.425007893905905,
      "grad_norm": 6.370941638946533,
      "learning_rate": 3.4930533628039156e-05,
      "loss": 0.5496,
      "step": 23515
    },
    {
      "epoch": 7.42532365014209,
      "grad_norm": 4.806075096130371,
      "learning_rate": 3.492965652738309e-05,
      "loss": 0.3625,
      "step": 23516
    },
    {
      "epoch": 7.425639406378276,
      "grad_norm": 5.211930751800537,
      "learning_rate": 3.492877942672701e-05,
      "loss": 0.5041,
      "step": 23517
    },
    {
      "epoch": 7.4259551626144615,
      "grad_norm": 4.293539524078369,
      "learning_rate": 3.4927902326070944e-05,
      "loss": 0.4768,
      "step": 23518
    },
    {
      "epoch": 7.426270918850648,
      "grad_norm": 5.373445987701416,
      "learning_rate": 3.492702522541487e-05,
      "loss": 0.4611,
      "step": 23519
    },
    {
      "epoch": 7.426586675086833,
      "grad_norm": 5.655552387237549,
      "learning_rate": 3.49261481247588e-05,
      "loss": 0.6192,
      "step": 23520
    },
    {
      "epoch": 7.426902431323018,
      "grad_norm": 5.442737579345703,
      "learning_rate": 3.492527102410273e-05,
      "loss": 0.4775,
      "step": 23521
    },
    {
      "epoch": 7.4272181875592045,
      "grad_norm": 5.191556930541992,
      "learning_rate": 3.492439392344666e-05,
      "loss": 0.4947,
      "step": 23522
    },
    {
      "epoch": 7.42753394379539,
      "grad_norm": 4.312471866607666,
      "learning_rate": 3.492351682279059e-05,
      "loss": 0.4868,
      "step": 23523
    },
    {
      "epoch": 7.427849700031576,
      "grad_norm": 4.743610858917236,
      "learning_rate": 3.4922639722134514e-05,
      "loss": 0.5467,
      "step": 23524
    },
    {
      "epoch": 7.428165456267761,
      "grad_norm": 5.28191614151001,
      "learning_rate": 3.4921762621478445e-05,
      "loss": 0.473,
      "step": 23525
    },
    {
      "epoch": 7.4284812125039466,
      "grad_norm": 5.488137245178223,
      "learning_rate": 3.492088552082237e-05,
      "loss": 0.5046,
      "step": 23526
    },
    {
      "epoch": 7.428796968740133,
      "grad_norm": 5.458948612213135,
      "learning_rate": 3.49200084201663e-05,
      "loss": 0.5552,
      "step": 23527
    },
    {
      "epoch": 7.429112724976318,
      "grad_norm": 5.078760147094727,
      "learning_rate": 3.491913131951023e-05,
      "loss": 0.3985,
      "step": 23528
    },
    {
      "epoch": 7.429428481212504,
      "grad_norm": 5.974579811096191,
      "learning_rate": 3.491825421885416e-05,
      "loss": 0.4769,
      "step": 23529
    },
    {
      "epoch": 7.4297442374486895,
      "grad_norm": 7.872140407562256,
      "learning_rate": 3.4917377118198084e-05,
      "loss": 0.5137,
      "step": 23530
    },
    {
      "epoch": 7.430059993684876,
      "grad_norm": 4.697173118591309,
      "learning_rate": 3.4916500017542015e-05,
      "loss": 0.4066,
      "step": 23531
    },
    {
      "epoch": 7.430375749921061,
      "grad_norm": 5.160194396972656,
      "learning_rate": 3.491562291688594e-05,
      "loss": 0.4622,
      "step": 23532
    },
    {
      "epoch": 7.430691506157246,
      "grad_norm": 5.094390869140625,
      "learning_rate": 3.491474581622987e-05,
      "loss": 0.5161,
      "step": 23533
    },
    {
      "epoch": 7.4310072623934325,
      "grad_norm": 4.41209602355957,
      "learning_rate": 3.49138687155738e-05,
      "loss": 0.4089,
      "step": 23534
    },
    {
      "epoch": 7.431323018629618,
      "grad_norm": 5.339850902557373,
      "learning_rate": 3.491299161491773e-05,
      "loss": 0.4638,
      "step": 23535
    },
    {
      "epoch": 7.431638774865804,
      "grad_norm": 6.212796211242676,
      "learning_rate": 3.491211451426166e-05,
      "loss": 0.4669,
      "step": 23536
    },
    {
      "epoch": 7.431954531101989,
      "grad_norm": 6.491491317749023,
      "learning_rate": 3.4911237413605585e-05,
      "loss": 0.4874,
      "step": 23537
    },
    {
      "epoch": 7.4322702873381745,
      "grad_norm": 4.712969779968262,
      "learning_rate": 3.491036031294952e-05,
      "loss": 0.4359,
      "step": 23538
    },
    {
      "epoch": 7.432586043574361,
      "grad_norm": 6.1781439781188965,
      "learning_rate": 3.490948321229344e-05,
      "loss": 0.4467,
      "step": 23539
    },
    {
      "epoch": 7.432901799810546,
      "grad_norm": 4.747338771820068,
      "learning_rate": 3.490860611163737e-05,
      "loss": 0.3748,
      "step": 23540
    },
    {
      "epoch": 7.433217556046732,
      "grad_norm": 6.396127223968506,
      "learning_rate": 3.49077290109813e-05,
      "loss": 0.5735,
      "step": 23541
    },
    {
      "epoch": 7.4335333122829175,
      "grad_norm": 5.308441162109375,
      "learning_rate": 3.490685191032523e-05,
      "loss": 0.4745,
      "step": 23542
    },
    {
      "epoch": 7.433849068519104,
      "grad_norm": 5.035839557647705,
      "learning_rate": 3.490597480966916e-05,
      "loss": 0.4826,
      "step": 23543
    },
    {
      "epoch": 7.434164824755289,
      "grad_norm": 4.759677886962891,
      "learning_rate": 3.490509770901309e-05,
      "loss": 0.3654,
      "step": 23544
    },
    {
      "epoch": 7.434480580991474,
      "grad_norm": 5.868936061859131,
      "learning_rate": 3.490422060835702e-05,
      "loss": 0.4815,
      "step": 23545
    },
    {
      "epoch": 7.43479633722766,
      "grad_norm": 5.715188026428223,
      "learning_rate": 3.490334350770095e-05,
      "loss": 0.4474,
      "step": 23546
    },
    {
      "epoch": 7.435112093463846,
      "grad_norm": 5.497004985809326,
      "learning_rate": 3.4902466407044875e-05,
      "loss": 0.5407,
      "step": 23547
    },
    {
      "epoch": 7.435427849700032,
      "grad_norm": 4.808941841125488,
      "learning_rate": 3.4901589306388806e-05,
      "loss": 0.5046,
      "step": 23548
    },
    {
      "epoch": 7.435743605936217,
      "grad_norm": 5.074974536895752,
      "learning_rate": 3.490071220573273e-05,
      "loss": 0.4043,
      "step": 23549
    },
    {
      "epoch": 7.4360593621724025,
      "grad_norm": 5.22758674621582,
      "learning_rate": 3.489983510507666e-05,
      "loss": 0.5207,
      "step": 23550
    },
    {
      "epoch": 7.436375118408589,
      "grad_norm": 5.560478210449219,
      "learning_rate": 3.489895800442059e-05,
      "loss": 0.4593,
      "step": 23551
    },
    {
      "epoch": 7.436690874644774,
      "grad_norm": 6.183051586151123,
      "learning_rate": 3.489808090376452e-05,
      "loss": 0.4937,
      "step": 23552
    },
    {
      "epoch": 7.43700663088096,
      "grad_norm": 5.333226203918457,
      "learning_rate": 3.4897203803108445e-05,
      "loss": 0.3899,
      "step": 23553
    },
    {
      "epoch": 7.437322387117145,
      "grad_norm": 5.365571022033691,
      "learning_rate": 3.4896326702452376e-05,
      "loss": 0.5121,
      "step": 23554
    },
    {
      "epoch": 7.437638143353332,
      "grad_norm": 4.890918731689453,
      "learning_rate": 3.48954496017963e-05,
      "loss": 0.4288,
      "step": 23555
    },
    {
      "epoch": 7.437953899589517,
      "grad_norm": 6.907557964324951,
      "learning_rate": 3.489457250114023e-05,
      "loss": 0.4569,
      "step": 23556
    },
    {
      "epoch": 7.438269655825702,
      "grad_norm": 5.478210926055908,
      "learning_rate": 3.489369540048416e-05,
      "loss": 0.4686,
      "step": 23557
    },
    {
      "epoch": 7.438585412061888,
      "grad_norm": 4.307276248931885,
      "learning_rate": 3.489281829982809e-05,
      "loss": 0.4534,
      "step": 23558
    },
    {
      "epoch": 7.438901168298074,
      "grad_norm": 6.388429164886475,
      "learning_rate": 3.489194119917202e-05,
      "loss": 0.5501,
      "step": 23559
    },
    {
      "epoch": 7.43921692453426,
      "grad_norm": 5.5984296798706055,
      "learning_rate": 3.4891064098515946e-05,
      "loss": 0.4355,
      "step": 23560
    },
    {
      "epoch": 7.439532680770445,
      "grad_norm": 7.207472324371338,
      "learning_rate": 3.489018699785988e-05,
      "loss": 0.5968,
      "step": 23561
    },
    {
      "epoch": 7.439848437006631,
      "grad_norm": 5.267454147338867,
      "learning_rate": 3.48893098972038e-05,
      "loss": 0.4184,
      "step": 23562
    },
    {
      "epoch": 7.440164193242817,
      "grad_norm": 5.077747344970703,
      "learning_rate": 3.4888432796547734e-05,
      "loss": 0.4637,
      "step": 23563
    },
    {
      "epoch": 7.440479949479002,
      "grad_norm": 4.539336204528809,
      "learning_rate": 3.488755569589166e-05,
      "loss": 0.4202,
      "step": 23564
    },
    {
      "epoch": 7.440795705715188,
      "grad_norm": 5.905402660369873,
      "learning_rate": 3.488667859523559e-05,
      "loss": 0.479,
      "step": 23565
    },
    {
      "epoch": 7.441111461951373,
      "grad_norm": 5.365485191345215,
      "learning_rate": 3.4885801494579516e-05,
      "loss": 0.4667,
      "step": 23566
    },
    {
      "epoch": 7.44142721818756,
      "grad_norm": 6.092545032501221,
      "learning_rate": 3.488492439392345e-05,
      "loss": 0.5634,
      "step": 23567
    },
    {
      "epoch": 7.441742974423745,
      "grad_norm": 5.670989990234375,
      "learning_rate": 3.488404729326737e-05,
      "loss": 0.47,
      "step": 23568
    },
    {
      "epoch": 7.44205873065993,
      "grad_norm": 5.41681432723999,
      "learning_rate": 3.4883170192611304e-05,
      "loss": 0.4785,
      "step": 23569
    },
    {
      "epoch": 7.442374486896116,
      "grad_norm": 5.624789714813232,
      "learning_rate": 3.4882293091955236e-05,
      "loss": 0.5524,
      "step": 23570
    },
    {
      "epoch": 7.442690243132302,
      "grad_norm": 6.5911407470703125,
      "learning_rate": 3.488141599129917e-05,
      "loss": 0.5409,
      "step": 23571
    },
    {
      "epoch": 7.443005999368488,
      "grad_norm": 5.123415470123291,
      "learning_rate": 3.488053889064309e-05,
      "loss": 0.4623,
      "step": 23572
    },
    {
      "epoch": 7.443321755604673,
      "grad_norm": 5.628080368041992,
      "learning_rate": 3.4879661789987024e-05,
      "loss": 0.4996,
      "step": 23573
    },
    {
      "epoch": 7.443637511840858,
      "grad_norm": 6.2170939445495605,
      "learning_rate": 3.487878468933095e-05,
      "loss": 0.5429,
      "step": 23574
    },
    {
      "epoch": 7.443953268077045,
      "grad_norm": 5.578004837036133,
      "learning_rate": 3.487790758867488e-05,
      "loss": 0.3899,
      "step": 23575
    },
    {
      "epoch": 7.44426902431323,
      "grad_norm": 7.722846984863281,
      "learning_rate": 3.4877030488018805e-05,
      "loss": 0.5233,
      "step": 23576
    },
    {
      "epoch": 7.444584780549416,
      "grad_norm": 6.829883575439453,
      "learning_rate": 3.487615338736274e-05,
      "loss": 0.5751,
      "step": 23577
    },
    {
      "epoch": 7.444900536785601,
      "grad_norm": 6.695561408996582,
      "learning_rate": 3.487527628670666e-05,
      "loss": 0.5424,
      "step": 23578
    },
    {
      "epoch": 7.445216293021788,
      "grad_norm": 6.043259620666504,
      "learning_rate": 3.4874399186050594e-05,
      "loss": 0.5233,
      "step": 23579
    },
    {
      "epoch": 7.445532049257973,
      "grad_norm": 4.780274391174316,
      "learning_rate": 3.4873522085394525e-05,
      "loss": 0.5841,
      "step": 23580
    },
    {
      "epoch": 7.445847805494158,
      "grad_norm": 5.939475059509277,
      "learning_rate": 3.487264498473845e-05,
      "loss": 0.4269,
      "step": 23581
    },
    {
      "epoch": 7.446163561730344,
      "grad_norm": 4.6664509773254395,
      "learning_rate": 3.487176788408238e-05,
      "loss": 0.4352,
      "step": 23582
    },
    {
      "epoch": 7.44647931796653,
      "grad_norm": 5.249161243438721,
      "learning_rate": 3.487089078342631e-05,
      "loss": 0.4492,
      "step": 23583
    },
    {
      "epoch": 7.446795074202716,
      "grad_norm": 7.107351779937744,
      "learning_rate": 3.487001368277024e-05,
      "loss": 0.544,
      "step": 23584
    },
    {
      "epoch": 7.447110830438901,
      "grad_norm": 6.142548561096191,
      "learning_rate": 3.4869136582114163e-05,
      "loss": 0.5171,
      "step": 23585
    },
    {
      "epoch": 7.447426586675087,
      "grad_norm": 5.570342540740967,
      "learning_rate": 3.4868259481458095e-05,
      "loss": 0.5256,
      "step": 23586
    },
    {
      "epoch": 7.447742342911273,
      "grad_norm": 6.0853590965271,
      "learning_rate": 3.486738238080202e-05,
      "loss": 0.4217,
      "step": 23587
    },
    {
      "epoch": 7.448058099147458,
      "grad_norm": 6.053009510040283,
      "learning_rate": 3.486650528014595e-05,
      "loss": 0.4592,
      "step": 23588
    },
    {
      "epoch": 7.448373855383644,
      "grad_norm": 5.985400676727295,
      "learning_rate": 3.486562817948988e-05,
      "loss": 0.4629,
      "step": 23589
    },
    {
      "epoch": 7.448689611619829,
      "grad_norm": 6.264589786529541,
      "learning_rate": 3.486475107883381e-05,
      "loss": 0.4648,
      "step": 23590
    },
    {
      "epoch": 7.4490053678560155,
      "grad_norm": 5.46030330657959,
      "learning_rate": 3.486387397817773e-05,
      "loss": 0.4137,
      "step": 23591
    },
    {
      "epoch": 7.449321124092201,
      "grad_norm": 5.905811309814453,
      "learning_rate": 3.4862996877521665e-05,
      "loss": 0.5306,
      "step": 23592
    },
    {
      "epoch": 7.449636880328386,
      "grad_norm": 5.040370941162109,
      "learning_rate": 3.486211977686559e-05,
      "loss": 0.5486,
      "step": 23593
    },
    {
      "epoch": 7.449952636564572,
      "grad_norm": 7.161888122558594,
      "learning_rate": 3.486124267620952e-05,
      "loss": 0.4454,
      "step": 23594
    },
    {
      "epoch": 7.450268392800758,
      "grad_norm": 5.789646148681641,
      "learning_rate": 3.4860365575553446e-05,
      "loss": 0.4613,
      "step": 23595
    },
    {
      "epoch": 7.450584149036944,
      "grad_norm": 5.295777320861816,
      "learning_rate": 3.485948847489738e-05,
      "loss": 0.4808,
      "step": 23596
    },
    {
      "epoch": 7.450899905273129,
      "grad_norm": 4.427700996398926,
      "learning_rate": 3.485861137424131e-05,
      "loss": 0.3879,
      "step": 23597
    },
    {
      "epoch": 7.451215661509314,
      "grad_norm": 5.456850051879883,
      "learning_rate": 3.485773427358524e-05,
      "loss": 0.5289,
      "step": 23598
    },
    {
      "epoch": 7.4515314177455005,
      "grad_norm": 6.247129917144775,
      "learning_rate": 3.4856857172929166e-05,
      "loss": 0.5075,
      "step": 23599
    },
    {
      "epoch": 7.451847173981686,
      "grad_norm": 4.647772312164307,
      "learning_rate": 3.48559800722731e-05,
      "loss": 0.4789,
      "step": 23600
    },
    {
      "epoch": 7.451847173981686,
      "eval_loss": 0.9366036653518677,
      "eval_runtime": 4.6241,
      "eval_samples_per_second": 538.483,
      "eval_steps_per_second": 4.541,
      "step": 23600
    },
    {
      "epoch": 7.452162930217872,
      "grad_norm": 5.184392929077148,
      "learning_rate": 3.485510297161702e-05,
      "loss": 0.3842,
      "step": 23601
    },
    {
      "epoch": 7.452478686454057,
      "grad_norm": 5.597607135772705,
      "learning_rate": 3.4854225870960955e-05,
      "loss": 0.4723,
      "step": 23602
    },
    {
      "epoch": 7.4527944426902435,
      "grad_norm": 5.787070274353027,
      "learning_rate": 3.4853348770304886e-05,
      "loss": 0.4826,
      "step": 23603
    },
    {
      "epoch": 7.453110198926429,
      "grad_norm": 4.9068193435668945,
      "learning_rate": 3.485247166964881e-05,
      "loss": 0.434,
      "step": 23604
    },
    {
      "epoch": 7.453425955162614,
      "grad_norm": 6.291075229644775,
      "learning_rate": 3.485159456899274e-05,
      "loss": 0.5516,
      "step": 23605
    },
    {
      "epoch": 7.4537417113988,
      "grad_norm": 6.053097248077393,
      "learning_rate": 3.485071746833667e-05,
      "loss": 0.5688,
      "step": 23606
    },
    {
      "epoch": 7.454057467634986,
      "grad_norm": 5.539705276489258,
      "learning_rate": 3.48498403676806e-05,
      "loss": 0.4533,
      "step": 23607
    },
    {
      "epoch": 7.454373223871172,
      "grad_norm": 4.514556407928467,
      "learning_rate": 3.4848963267024524e-05,
      "loss": 0.4521,
      "step": 23608
    },
    {
      "epoch": 7.454688980107357,
      "grad_norm": 5.825781345367432,
      "learning_rate": 3.4848086166368456e-05,
      "loss": 0.5494,
      "step": 23609
    },
    {
      "epoch": 7.455004736343543,
      "grad_norm": 4.8812432289123535,
      "learning_rate": 3.484720906571238e-05,
      "loss": 0.4317,
      "step": 23610
    },
    {
      "epoch": 7.4553204925797285,
      "grad_norm": 5.291028022766113,
      "learning_rate": 3.484633196505631e-05,
      "loss": 0.5167,
      "step": 23611
    },
    {
      "epoch": 7.455636248815914,
      "grad_norm": 6.266304016113281,
      "learning_rate": 3.484545486440024e-05,
      "loss": 0.5004,
      "step": 23612
    },
    {
      "epoch": 7.4559520050521,
      "grad_norm": 5.805854320526123,
      "learning_rate": 3.484457776374417e-05,
      "loss": 0.4156,
      "step": 23613
    },
    {
      "epoch": 7.456267761288285,
      "grad_norm": 5.796006202697754,
      "learning_rate": 3.4843700663088094e-05,
      "loss": 0.473,
      "step": 23614
    },
    {
      "epoch": 7.4565835175244715,
      "grad_norm": 9.711774826049805,
      "learning_rate": 3.4842823562432026e-05,
      "loss": 0.6111,
      "step": 23615
    },
    {
      "epoch": 7.456899273760657,
      "grad_norm": 5.373473167419434,
      "learning_rate": 3.484194646177595e-05,
      "loss": 0.4497,
      "step": 23616
    },
    {
      "epoch": 7.457215029996842,
      "grad_norm": 7.250823020935059,
      "learning_rate": 3.484106936111988e-05,
      "loss": 0.5598,
      "step": 23617
    },
    {
      "epoch": 7.457530786233028,
      "grad_norm": 6.676865100860596,
      "learning_rate": 3.4840192260463814e-05,
      "loss": 0.5178,
      "step": 23618
    },
    {
      "epoch": 7.4578465424692135,
      "grad_norm": 6.803675174713135,
      "learning_rate": 3.483931515980774e-05,
      "loss": 0.5502,
      "step": 23619
    },
    {
      "epoch": 7.4581622987054,
      "grad_norm": 6.696410179138184,
      "learning_rate": 3.483843805915167e-05,
      "loss": 0.4665,
      "step": 23620
    },
    {
      "epoch": 7.458478054941585,
      "grad_norm": 7.884701251983643,
      "learning_rate": 3.4837560958495596e-05,
      "loss": 0.5518,
      "step": 23621
    },
    {
      "epoch": 7.45879381117777,
      "grad_norm": 7.609029293060303,
      "learning_rate": 3.483668385783953e-05,
      "loss": 0.4538,
      "step": 23622
    },
    {
      "epoch": 7.4591095674139565,
      "grad_norm": 7.58545446395874,
      "learning_rate": 3.483580675718345e-05,
      "loss": 0.5626,
      "step": 23623
    },
    {
      "epoch": 7.459425323650142,
      "grad_norm": 6.175451278686523,
      "learning_rate": 3.483492965652739e-05,
      "loss": 0.5757,
      "step": 23624
    },
    {
      "epoch": 7.459741079886328,
      "grad_norm": 4.753048419952393,
      "learning_rate": 3.4834052555871316e-05,
      "loss": 0.4766,
      "step": 23625
    },
    {
      "epoch": 7.460056836122513,
      "grad_norm": 5.0271382331848145,
      "learning_rate": 3.483317545521525e-05,
      "loss": 0.4245,
      "step": 23626
    },
    {
      "epoch": 7.460372592358699,
      "grad_norm": 6.28743314743042,
      "learning_rate": 3.483229835455917e-05,
      "loss": 0.5033,
      "step": 23627
    },
    {
      "epoch": 7.460688348594885,
      "grad_norm": 5.632485389709473,
      "learning_rate": 3.4831421253903104e-05,
      "loss": 0.5025,
      "step": 23628
    },
    {
      "epoch": 7.46100410483107,
      "grad_norm": 6.681432723999023,
      "learning_rate": 3.483054415324703e-05,
      "loss": 0.5958,
      "step": 23629
    },
    {
      "epoch": 7.461319861067256,
      "grad_norm": 7.17874813079834,
      "learning_rate": 3.482966705259096e-05,
      "loss": 0.497,
      "step": 23630
    },
    {
      "epoch": 7.4616356173034415,
      "grad_norm": 7.0234479904174805,
      "learning_rate": 3.4828789951934885e-05,
      "loss": 0.538,
      "step": 23631
    },
    {
      "epoch": 7.461951373539628,
      "grad_norm": 5.292324066162109,
      "learning_rate": 3.482791285127882e-05,
      "loss": 0.4285,
      "step": 23632
    },
    {
      "epoch": 7.462267129775813,
      "grad_norm": 6.600733280181885,
      "learning_rate": 3.482703575062274e-05,
      "loss": 0.5863,
      "step": 23633
    },
    {
      "epoch": 7.462582886011999,
      "grad_norm": 6.1741862297058105,
      "learning_rate": 3.4826158649966674e-05,
      "loss": 0.4329,
      "step": 23634
    },
    {
      "epoch": 7.4628986422481844,
      "grad_norm": 5.482278347015381,
      "learning_rate": 3.48252815493106e-05,
      "loss": 0.3789,
      "step": 23635
    },
    {
      "epoch": 7.46321439848437,
      "grad_norm": 4.903439998626709,
      "learning_rate": 3.482440444865453e-05,
      "loss": 0.3979,
      "step": 23636
    },
    {
      "epoch": 7.463530154720556,
      "grad_norm": 6.076566219329834,
      "learning_rate": 3.4823527347998455e-05,
      "loss": 0.5256,
      "step": 23637
    },
    {
      "epoch": 7.463845910956741,
      "grad_norm": 6.551605701446533,
      "learning_rate": 3.482265024734239e-05,
      "loss": 0.5729,
      "step": 23638
    },
    {
      "epoch": 7.464161667192927,
      "grad_norm": 6.005155563354492,
      "learning_rate": 3.482177314668632e-05,
      "loss": 0.5393,
      "step": 23639
    },
    {
      "epoch": 7.464477423429113,
      "grad_norm": 5.751317024230957,
      "learning_rate": 3.482089604603024e-05,
      "loss": 0.3849,
      "step": 23640
    },
    {
      "epoch": 7.464793179665298,
      "grad_norm": 5.493991374969482,
      "learning_rate": 3.4820018945374175e-05,
      "loss": 0.4033,
      "step": 23641
    },
    {
      "epoch": 7.465108935901484,
      "grad_norm": 4.986715793609619,
      "learning_rate": 3.48191418447181e-05,
      "loss": 0.4242,
      "step": 23642
    },
    {
      "epoch": 7.4654246921376695,
      "grad_norm": 7.419501781463623,
      "learning_rate": 3.481826474406203e-05,
      "loss": 0.5085,
      "step": 23643
    },
    {
      "epoch": 7.465740448373856,
      "grad_norm": 5.170459270477295,
      "learning_rate": 3.4817387643405957e-05,
      "loss": 0.3895,
      "step": 23644
    },
    {
      "epoch": 7.466056204610041,
      "grad_norm": 6.289670467376709,
      "learning_rate": 3.481651054274989e-05,
      "loss": 0.5349,
      "step": 23645
    },
    {
      "epoch": 7.466371960846227,
      "grad_norm": 4.9133524894714355,
      "learning_rate": 3.481563344209381e-05,
      "loss": 0.5251,
      "step": 23646
    },
    {
      "epoch": 7.466687717082412,
      "grad_norm": 5.111903667449951,
      "learning_rate": 3.4814756341437745e-05,
      "loss": 0.5425,
      "step": 23647
    },
    {
      "epoch": 7.467003473318598,
      "grad_norm": 6.920496940612793,
      "learning_rate": 3.481387924078167e-05,
      "loss": 0.479,
      "step": 23648
    },
    {
      "epoch": 7.467319229554784,
      "grad_norm": 6.3311638832092285,
      "learning_rate": 3.48130021401256e-05,
      "loss": 0.5451,
      "step": 23649
    },
    {
      "epoch": 7.467634985790969,
      "grad_norm": 7.1512651443481445,
      "learning_rate": 3.4812125039469526e-05,
      "loss": 0.4559,
      "step": 23650
    },
    {
      "epoch": 7.467950742027155,
      "grad_norm": 4.79568338394165,
      "learning_rate": 3.481124793881346e-05,
      "loss": 0.4952,
      "step": 23651
    },
    {
      "epoch": 7.468266498263341,
      "grad_norm": 4.821611404418945,
      "learning_rate": 3.481037083815739e-05,
      "loss": 0.3821,
      "step": 23652
    },
    {
      "epoch": 7.468582254499526,
      "grad_norm": 4.32114315032959,
      "learning_rate": 3.480949373750132e-05,
      "loss": 0.3798,
      "step": 23653
    },
    {
      "epoch": 7.468898010735712,
      "grad_norm": 6.185199737548828,
      "learning_rate": 3.4808616636845246e-05,
      "loss": 0.5118,
      "step": 23654
    },
    {
      "epoch": 7.469213766971897,
      "grad_norm": 4.550906658172607,
      "learning_rate": 3.480773953618918e-05,
      "loss": 0.4655,
      "step": 23655
    },
    {
      "epoch": 7.469529523208084,
      "grad_norm": 4.884733200073242,
      "learning_rate": 3.48068624355331e-05,
      "loss": 0.3828,
      "step": 23656
    },
    {
      "epoch": 7.469845279444269,
      "grad_norm": 6.392996311187744,
      "learning_rate": 3.4805985334877035e-05,
      "loss": 0.5528,
      "step": 23657
    },
    {
      "epoch": 7.470161035680455,
      "grad_norm": 6.8869853019714355,
      "learning_rate": 3.480510823422096e-05,
      "loss": 0.4381,
      "step": 23658
    },
    {
      "epoch": 7.47047679191664,
      "grad_norm": 5.999274730682373,
      "learning_rate": 3.480423113356489e-05,
      "loss": 0.5822,
      "step": 23659
    },
    {
      "epoch": 7.470792548152826,
      "grad_norm": 5.112720012664795,
      "learning_rate": 3.4803354032908816e-05,
      "loss": 0.4488,
      "step": 23660
    },
    {
      "epoch": 7.471108304389012,
      "grad_norm": 6.164953708648682,
      "learning_rate": 3.480247693225275e-05,
      "loss": 0.5045,
      "step": 23661
    },
    {
      "epoch": 7.471424060625197,
      "grad_norm": 6.698678493499756,
      "learning_rate": 3.480159983159668e-05,
      "loss": 0.5401,
      "step": 23662
    },
    {
      "epoch": 7.471739816861383,
      "grad_norm": 4.678776264190674,
      "learning_rate": 3.4800722730940604e-05,
      "loss": 0.4423,
      "step": 23663
    },
    {
      "epoch": 7.472055573097569,
      "grad_norm": 5.3711838722229,
      "learning_rate": 3.4799845630284536e-05,
      "loss": 0.5326,
      "step": 23664
    },
    {
      "epoch": 7.472371329333754,
      "grad_norm": 5.670684337615967,
      "learning_rate": 3.479896852962846e-05,
      "loss": 0.4515,
      "step": 23665
    },
    {
      "epoch": 7.47268708556994,
      "grad_norm": 6.395492076873779,
      "learning_rate": 3.479809142897239e-05,
      "loss": 0.4235,
      "step": 23666
    },
    {
      "epoch": 7.473002841806125,
      "grad_norm": 7.125784397125244,
      "learning_rate": 3.479721432831632e-05,
      "loss": 0.4664,
      "step": 23667
    },
    {
      "epoch": 7.473318598042312,
      "grad_norm": 4.808650970458984,
      "learning_rate": 3.479633722766025e-05,
      "loss": 0.462,
      "step": 23668
    },
    {
      "epoch": 7.473634354278497,
      "grad_norm": 6.539163589477539,
      "learning_rate": 3.4795460127004174e-05,
      "loss": 0.4805,
      "step": 23669
    },
    {
      "epoch": 7.473950110514683,
      "grad_norm": 5.469463348388672,
      "learning_rate": 3.4794583026348106e-05,
      "loss": 0.5177,
      "step": 23670
    },
    {
      "epoch": 7.474265866750868,
      "grad_norm": 9.526811599731445,
      "learning_rate": 3.479370592569203e-05,
      "loss": 0.4478,
      "step": 23671
    },
    {
      "epoch": 7.474581622987054,
      "grad_norm": 7.546627044677734,
      "learning_rate": 3.479282882503596e-05,
      "loss": 0.5025,
      "step": 23672
    },
    {
      "epoch": 7.47489737922324,
      "grad_norm": 6.255486011505127,
      "learning_rate": 3.479195172437989e-05,
      "loss": 0.4008,
      "step": 23673
    },
    {
      "epoch": 7.475213135459425,
      "grad_norm": 8.197165489196777,
      "learning_rate": 3.479107462372382e-05,
      "loss": 0.4241,
      "step": 23674
    },
    {
      "epoch": 7.475528891695611,
      "grad_norm": 4.871253490447998,
      "learning_rate": 3.4790197523067744e-05,
      "loss": 0.4629,
      "step": 23675
    },
    {
      "epoch": 7.475844647931797,
      "grad_norm": 5.173619270324707,
      "learning_rate": 3.4789320422411676e-05,
      "loss": 0.5044,
      "step": 23676
    },
    {
      "epoch": 7.476160404167983,
      "grad_norm": 5.06015682220459,
      "learning_rate": 3.478844332175561e-05,
      "loss": 0.4526,
      "step": 23677
    },
    {
      "epoch": 7.476476160404168,
      "grad_norm": 4.372437953948975,
      "learning_rate": 3.478756622109953e-05,
      "loss": 0.3858,
      "step": 23678
    },
    {
      "epoch": 7.476791916640353,
      "grad_norm": 4.67030668258667,
      "learning_rate": 3.4786689120443464e-05,
      "loss": 0.512,
      "step": 23679
    },
    {
      "epoch": 7.4771076728765395,
      "grad_norm": 5.529179096221924,
      "learning_rate": 3.4785812019787395e-05,
      "loss": 0.4262,
      "step": 23680
    },
    {
      "epoch": 7.477423429112725,
      "grad_norm": 5.288912296295166,
      "learning_rate": 3.478493491913132e-05,
      "loss": 0.4475,
      "step": 23681
    },
    {
      "epoch": 7.477739185348911,
      "grad_norm": 5.58001184463501,
      "learning_rate": 3.478405781847525e-05,
      "loss": 0.4706,
      "step": 23682
    },
    {
      "epoch": 7.478054941585096,
      "grad_norm": 5.66654634475708,
      "learning_rate": 3.4783180717819184e-05,
      "loss": 0.4882,
      "step": 23683
    },
    {
      "epoch": 7.478370697821282,
      "grad_norm": 4.8870158195495605,
      "learning_rate": 3.478230361716311e-05,
      "loss": 0.4077,
      "step": 23684
    },
    {
      "epoch": 7.478686454057468,
      "grad_norm": 5.7750163078308105,
      "learning_rate": 3.478142651650704e-05,
      "loss": 0.5026,
      "step": 23685
    },
    {
      "epoch": 7.479002210293653,
      "grad_norm": 5.609231472015381,
      "learning_rate": 3.4780549415850965e-05,
      "loss": 0.4539,
      "step": 23686
    },
    {
      "epoch": 7.479317966529839,
      "grad_norm": 7.45710563659668,
      "learning_rate": 3.47796723151949e-05,
      "loss": 0.4945,
      "step": 23687
    },
    {
      "epoch": 7.479633722766025,
      "grad_norm": 4.635588645935059,
      "learning_rate": 3.477879521453882e-05,
      "loss": 0.4514,
      "step": 23688
    },
    {
      "epoch": 7.47994947900221,
      "grad_norm": 4.476492404937744,
      "learning_rate": 3.4777918113882753e-05,
      "loss": 0.3844,
      "step": 23689
    },
    {
      "epoch": 7.480265235238396,
      "grad_norm": 4.341561317443848,
      "learning_rate": 3.477704101322668e-05,
      "loss": 0.4908,
      "step": 23690
    },
    {
      "epoch": 7.480580991474581,
      "grad_norm": 7.908227920532227,
      "learning_rate": 3.477616391257061e-05,
      "loss": 0.6123,
      "step": 23691
    },
    {
      "epoch": 7.4808967477107675,
      "grad_norm": 6.404728412628174,
      "learning_rate": 3.4775286811914535e-05,
      "loss": 0.451,
      "step": 23692
    },
    {
      "epoch": 7.481212503946953,
      "grad_norm": 5.785565376281738,
      "learning_rate": 3.477440971125847e-05,
      "loss": 0.5327,
      "step": 23693
    },
    {
      "epoch": 7.481528260183139,
      "grad_norm": 6.858173847198486,
      "learning_rate": 3.477353261060239e-05,
      "loss": 0.5788,
      "step": 23694
    },
    {
      "epoch": 7.481844016419324,
      "grad_norm": 6.796960353851318,
      "learning_rate": 3.477265550994632e-05,
      "loss": 0.434,
      "step": 23695
    },
    {
      "epoch": 7.48215977265551,
      "grad_norm": 6.724356651306152,
      "learning_rate": 3.477177840929025e-05,
      "loss": 0.5878,
      "step": 23696
    },
    {
      "epoch": 7.482475528891696,
      "grad_norm": 7.505291938781738,
      "learning_rate": 3.477090130863418e-05,
      "loss": 0.4677,
      "step": 23697
    },
    {
      "epoch": 7.482791285127881,
      "grad_norm": 5.696653842926025,
      "learning_rate": 3.4770024207978105e-05,
      "loss": 0.5039,
      "step": 23698
    },
    {
      "epoch": 7.483107041364067,
      "grad_norm": 6.19357967376709,
      "learning_rate": 3.4769147107322036e-05,
      "loss": 0.5959,
      "step": 23699
    },
    {
      "epoch": 7.4834227976002525,
      "grad_norm": 5.433896541595459,
      "learning_rate": 3.476827000666597e-05,
      "loss": 0.4592,
      "step": 23700
    },
    {
      "epoch": 7.483738553836439,
      "grad_norm": 8.089799880981445,
      "learning_rate": 3.476739290600989e-05,
      "loss": 0.5759,
      "step": 23701
    },
    {
      "epoch": 7.484054310072624,
      "grad_norm": 5.425144672393799,
      "learning_rate": 3.4766515805353825e-05,
      "loss": 0.5189,
      "step": 23702
    },
    {
      "epoch": 7.484370066308809,
      "grad_norm": 5.824973106384277,
      "learning_rate": 3.476563870469775e-05,
      "loss": 0.5546,
      "step": 23703
    },
    {
      "epoch": 7.4846858225449955,
      "grad_norm": 5.762998580932617,
      "learning_rate": 3.476476160404168e-05,
      "loss": 0.4679,
      "step": 23704
    },
    {
      "epoch": 7.485001578781181,
      "grad_norm": 6.557353973388672,
      "learning_rate": 3.4763884503385606e-05,
      "loss": 0.4684,
      "step": 23705
    },
    {
      "epoch": 7.485317335017367,
      "grad_norm": 4.260290145874023,
      "learning_rate": 3.4763007402729545e-05,
      "loss": 0.413,
      "step": 23706
    },
    {
      "epoch": 7.485633091253552,
      "grad_norm": 5.704799175262451,
      "learning_rate": 3.476213030207347e-05,
      "loss": 0.4515,
      "step": 23707
    },
    {
      "epoch": 7.4859488474897375,
      "grad_norm": 5.103714942932129,
      "learning_rate": 3.47612532014174e-05,
      "loss": 0.4638,
      "step": 23708
    },
    {
      "epoch": 7.486264603725924,
      "grad_norm": 5.090917587280273,
      "learning_rate": 3.4760376100761326e-05,
      "loss": 0.4377,
      "step": 23709
    },
    {
      "epoch": 7.486580359962109,
      "grad_norm": 4.765336513519287,
      "learning_rate": 3.475949900010526e-05,
      "loss": 0.4159,
      "step": 23710
    },
    {
      "epoch": 7.486896116198295,
      "grad_norm": 7.291837692260742,
      "learning_rate": 3.475862189944918e-05,
      "loss": 0.609,
      "step": 23711
    },
    {
      "epoch": 7.4872118724344805,
      "grad_norm": 6.227534770965576,
      "learning_rate": 3.4757744798793114e-05,
      "loss": 0.578,
      "step": 23712
    },
    {
      "epoch": 7.487527628670666,
      "grad_norm": 6.178685665130615,
      "learning_rate": 3.475686769813704e-05,
      "loss": 0.4612,
      "step": 23713
    },
    {
      "epoch": 7.487843384906852,
      "grad_norm": 5.261553764343262,
      "learning_rate": 3.475599059748097e-05,
      "loss": 0.4199,
      "step": 23714
    },
    {
      "epoch": 7.488159141143037,
      "grad_norm": 6.719128131866455,
      "learning_rate": 3.4755113496824896e-05,
      "loss": 0.6251,
      "step": 23715
    },
    {
      "epoch": 7.4884748973792234,
      "grad_norm": 5.055933475494385,
      "learning_rate": 3.475423639616883e-05,
      "loss": 0.541,
      "step": 23716
    },
    {
      "epoch": 7.488790653615409,
      "grad_norm": 5.498513698577881,
      "learning_rate": 3.475335929551275e-05,
      "loss": 0.5057,
      "step": 23717
    },
    {
      "epoch": 7.489106409851595,
      "grad_norm": 5.9618449211120605,
      "learning_rate": 3.4752482194856684e-05,
      "loss": 0.3517,
      "step": 23718
    },
    {
      "epoch": 7.48942216608778,
      "grad_norm": 6.291383743286133,
      "learning_rate": 3.475160509420061e-05,
      "loss": 0.4886,
      "step": 23719
    },
    {
      "epoch": 7.4897379223239655,
      "grad_norm": 5.65578031539917,
      "learning_rate": 3.475072799354454e-05,
      "loss": 0.5305,
      "step": 23720
    },
    {
      "epoch": 7.490053678560152,
      "grad_norm": 5.609557628631592,
      "learning_rate": 3.474985089288847e-05,
      "loss": 0.4939,
      "step": 23721
    },
    {
      "epoch": 7.490369434796337,
      "grad_norm": 4.931384086608887,
      "learning_rate": 3.47489737922324e-05,
      "loss": 0.4291,
      "step": 23722
    },
    {
      "epoch": 7.490685191032523,
      "grad_norm": 4.83317232131958,
      "learning_rate": 3.474809669157633e-05,
      "loss": 0.6031,
      "step": 23723
    },
    {
      "epoch": 7.4910009472687085,
      "grad_norm": 6.8340067863464355,
      "learning_rate": 3.4747219590920254e-05,
      "loss": 0.5502,
      "step": 23724
    },
    {
      "epoch": 7.491316703504895,
      "grad_norm": 5.376275539398193,
      "learning_rate": 3.4746342490264186e-05,
      "loss": 0.5404,
      "step": 23725
    },
    {
      "epoch": 7.49163245974108,
      "grad_norm": 5.647120952606201,
      "learning_rate": 3.474546538960811e-05,
      "loss": 0.4296,
      "step": 23726
    },
    {
      "epoch": 7.491948215977265,
      "grad_norm": 5.500165939331055,
      "learning_rate": 3.474458828895204e-05,
      "loss": 0.4872,
      "step": 23727
    },
    {
      "epoch": 7.492263972213451,
      "grad_norm": 6.453808784484863,
      "learning_rate": 3.474371118829597e-05,
      "loss": 0.4964,
      "step": 23728
    },
    {
      "epoch": 7.492579728449637,
      "grad_norm": 6.446281433105469,
      "learning_rate": 3.47428340876399e-05,
      "loss": 0.5273,
      "step": 23729
    },
    {
      "epoch": 7.492895484685823,
      "grad_norm": 6.588396072387695,
      "learning_rate": 3.4741956986983824e-05,
      "loss": 0.4552,
      "step": 23730
    },
    {
      "epoch": 7.493211240922008,
      "grad_norm": 5.101980686187744,
      "learning_rate": 3.4741079886327755e-05,
      "loss": 0.4779,
      "step": 23731
    },
    {
      "epoch": 7.4935269971581935,
      "grad_norm": 4.547389984130859,
      "learning_rate": 3.474020278567168e-05,
      "loss": 0.4517,
      "step": 23732
    },
    {
      "epoch": 7.49384275339438,
      "grad_norm": 5.955738544464111,
      "learning_rate": 3.473932568501562e-05,
      "loss": 0.4826,
      "step": 23733
    },
    {
      "epoch": 7.494158509630565,
      "grad_norm": 6.276697635650635,
      "learning_rate": 3.4738448584359544e-05,
      "loss": 0.4747,
      "step": 23734
    },
    {
      "epoch": 7.494474265866751,
      "grad_norm": 5.047400951385498,
      "learning_rate": 3.4737571483703475e-05,
      "loss": 0.4522,
      "step": 23735
    },
    {
      "epoch": 7.494790022102936,
      "grad_norm": 6.485739707946777,
      "learning_rate": 3.47366943830474e-05,
      "loss": 0.4589,
      "step": 23736
    },
    {
      "epoch": 7.495105778339122,
      "grad_norm": 5.60633659362793,
      "learning_rate": 3.473581728239133e-05,
      "loss": 0.4223,
      "step": 23737
    },
    {
      "epoch": 7.495421534575308,
      "grad_norm": 5.2866387367248535,
      "learning_rate": 3.473494018173526e-05,
      "loss": 0.5262,
      "step": 23738
    },
    {
      "epoch": 7.495737290811493,
      "grad_norm": 7.120434284210205,
      "learning_rate": 3.473406308107919e-05,
      "loss": 0.5131,
      "step": 23739
    },
    {
      "epoch": 7.496053047047679,
      "grad_norm": 5.872015476226807,
      "learning_rate": 3.4733185980423113e-05,
      "loss": 0.4468,
      "step": 23740
    },
    {
      "epoch": 7.496368803283865,
      "grad_norm": 5.422430038452148,
      "learning_rate": 3.4732308879767045e-05,
      "loss": 0.4207,
      "step": 23741
    },
    {
      "epoch": 7.496684559520051,
      "grad_norm": 5.4299726486206055,
      "learning_rate": 3.473143177911097e-05,
      "loss": 0.4713,
      "step": 23742
    },
    {
      "epoch": 7.497000315756236,
      "grad_norm": 5.815830230712891,
      "learning_rate": 3.47305546784549e-05,
      "loss": 0.5307,
      "step": 23743
    },
    {
      "epoch": 7.4973160719924214,
      "grad_norm": 5.536947250366211,
      "learning_rate": 3.472967757779883e-05,
      "loss": 0.5341,
      "step": 23744
    },
    {
      "epoch": 7.497631828228608,
      "grad_norm": 7.727229118347168,
      "learning_rate": 3.472880047714276e-05,
      "loss": 0.5183,
      "step": 23745
    },
    {
      "epoch": 7.497947584464793,
      "grad_norm": 10.455164909362793,
      "learning_rate": 3.472792337648669e-05,
      "loss": 0.4778,
      "step": 23746
    },
    {
      "epoch": 7.498263340700979,
      "grad_norm": 5.4779744148254395,
      "learning_rate": 3.4727046275830615e-05,
      "loss": 0.5668,
      "step": 23747
    },
    {
      "epoch": 7.498579096937164,
      "grad_norm": 5.913113594055176,
      "learning_rate": 3.4726169175174547e-05,
      "loss": 0.4901,
      "step": 23748
    },
    {
      "epoch": 7.498894853173351,
      "grad_norm": 6.560209274291992,
      "learning_rate": 3.472529207451847e-05,
      "loss": 0.5985,
      "step": 23749
    },
    {
      "epoch": 7.499210609409536,
      "grad_norm": 6.260003089904785,
      "learning_rate": 3.47244149738624e-05,
      "loss": 0.4211,
      "step": 23750
    },
    {
      "epoch": 7.499526365645721,
      "grad_norm": 6.576989650726318,
      "learning_rate": 3.472353787320633e-05,
      "loss": 0.5708,
      "step": 23751
    },
    {
      "epoch": 7.499842121881907,
      "grad_norm": 5.0833940505981445,
      "learning_rate": 3.472266077255026e-05,
      "loss": 0.4721,
      "step": 23752
    },
    {
      "epoch": 7.500157878118093,
      "grad_norm": 6.5243988037109375,
      "learning_rate": 3.4721783671894185e-05,
      "loss": 0.5336,
      "step": 23753
    },
    {
      "epoch": 7.500473634354279,
      "grad_norm": 5.982131004333496,
      "learning_rate": 3.4720906571238116e-05,
      "loss": 0.3972,
      "step": 23754
    },
    {
      "epoch": 7.500789390590464,
      "grad_norm": 6.969998359680176,
      "learning_rate": 3.472002947058204e-05,
      "loss": 0.6057,
      "step": 23755
    },
    {
      "epoch": 7.501105146826649,
      "grad_norm": 4.4341864585876465,
      "learning_rate": 3.471915236992597e-05,
      "loss": 0.5265,
      "step": 23756
    },
    {
      "epoch": 7.501420903062836,
      "grad_norm": 6.187628269195557,
      "learning_rate": 3.47182752692699e-05,
      "loss": 0.4209,
      "step": 23757
    },
    {
      "epoch": 7.501736659299021,
      "grad_norm": 5.661796569824219,
      "learning_rate": 3.471739816861383e-05,
      "loss": 0.4617,
      "step": 23758
    },
    {
      "epoch": 7.502052415535207,
      "grad_norm": 6.281247615814209,
      "learning_rate": 3.471652106795776e-05,
      "loss": 0.5441,
      "step": 23759
    },
    {
      "epoch": 7.502368171771392,
      "grad_norm": 7.008938312530518,
      "learning_rate": 3.4715643967301686e-05,
      "loss": 0.4964,
      "step": 23760
    },
    {
      "epoch": 7.502683928007578,
      "grad_norm": 5.41003942489624,
      "learning_rate": 3.471476686664562e-05,
      "loss": 0.4772,
      "step": 23761
    },
    {
      "epoch": 7.502999684243764,
      "grad_norm": 5.307648658752441,
      "learning_rate": 3.471388976598955e-05,
      "loss": 0.4799,
      "step": 23762
    },
    {
      "epoch": 7.503315440479949,
      "grad_norm": 5.580920696258545,
      "learning_rate": 3.4713012665333474e-05,
      "loss": 0.5692,
      "step": 23763
    },
    {
      "epoch": 7.503631196716135,
      "grad_norm": 5.067877769470215,
      "learning_rate": 3.4712135564677406e-05,
      "loss": 0.4788,
      "step": 23764
    },
    {
      "epoch": 7.503946952952321,
      "grad_norm": 7.581140041351318,
      "learning_rate": 3.471125846402134e-05,
      "loss": 0.5165,
      "step": 23765
    },
    {
      "epoch": 7.504262709188507,
      "grad_norm": 6.801863670349121,
      "learning_rate": 3.471038136336526e-05,
      "loss": 0.5192,
      "step": 23766
    },
    {
      "epoch": 7.504578465424692,
      "grad_norm": 7.090400695800781,
      "learning_rate": 3.4709504262709194e-05,
      "loss": 0.5549,
      "step": 23767
    },
    {
      "epoch": 7.504894221660878,
      "grad_norm": 7.132610321044922,
      "learning_rate": 3.470862716205312e-05,
      "loss": 0.5697,
      "step": 23768
    },
    {
      "epoch": 7.505209977897064,
      "grad_norm": 6.5366902351379395,
      "learning_rate": 3.470775006139705e-05,
      "loss": 0.4955,
      "step": 23769
    },
    {
      "epoch": 7.505525734133249,
      "grad_norm": 6.227426052093506,
      "learning_rate": 3.4706872960740976e-05,
      "loss": 0.5462,
      "step": 23770
    },
    {
      "epoch": 7.505841490369435,
      "grad_norm": 5.984681129455566,
      "learning_rate": 3.470599586008491e-05,
      "loss": 0.4541,
      "step": 23771
    },
    {
      "epoch": 7.50615724660562,
      "grad_norm": 5.805799961090088,
      "learning_rate": 3.470511875942883e-05,
      "loss": 0.5434,
      "step": 23772
    },
    {
      "epoch": 7.5064730028418065,
      "grad_norm": 6.139660358428955,
      "learning_rate": 3.4704241658772764e-05,
      "loss": 0.5704,
      "step": 23773
    },
    {
      "epoch": 7.506788759077992,
      "grad_norm": 5.073254108428955,
      "learning_rate": 3.470336455811669e-05,
      "loss": 0.4447,
      "step": 23774
    },
    {
      "epoch": 7.507104515314177,
      "grad_norm": 4.672746658325195,
      "learning_rate": 3.470248745746062e-05,
      "loss": 0.455,
      "step": 23775
    },
    {
      "epoch": 7.507420271550363,
      "grad_norm": 5.312446594238281,
      "learning_rate": 3.4701610356804546e-05,
      "loss": 0.4902,
      "step": 23776
    },
    {
      "epoch": 7.507736027786549,
      "grad_norm": 5.758530139923096,
      "learning_rate": 3.470073325614848e-05,
      "loss": 0.5861,
      "step": 23777
    },
    {
      "epoch": 7.508051784022735,
      "grad_norm": 4.632449626922607,
      "learning_rate": 3.46998561554924e-05,
      "loss": 0.3496,
      "step": 23778
    },
    {
      "epoch": 7.50836754025892,
      "grad_norm": 5.527921676635742,
      "learning_rate": 3.4698979054836334e-05,
      "loss": 0.4791,
      "step": 23779
    },
    {
      "epoch": 7.508683296495105,
      "grad_norm": 5.228327751159668,
      "learning_rate": 3.469810195418026e-05,
      "loss": 0.4098,
      "step": 23780
    },
    {
      "epoch": 7.5089990527312915,
      "grad_norm": 6.094189643859863,
      "learning_rate": 3.469722485352419e-05,
      "loss": 0.4955,
      "step": 23781
    },
    {
      "epoch": 7.509314808967477,
      "grad_norm": 6.677829265594482,
      "learning_rate": 3.469634775286812e-05,
      "loss": 0.5565,
      "step": 23782
    },
    {
      "epoch": 7.509630565203663,
      "grad_norm": 4.555818557739258,
      "learning_rate": 3.469547065221205e-05,
      "loss": 0.4196,
      "step": 23783
    },
    {
      "epoch": 7.509946321439848,
      "grad_norm": 5.200562953948975,
      "learning_rate": 3.469459355155598e-05,
      "loss": 0.4666,
      "step": 23784
    },
    {
      "epoch": 7.5102620776760345,
      "grad_norm": 4.236753940582275,
      "learning_rate": 3.4693716450899904e-05,
      "loss": 0.4605,
      "step": 23785
    },
    {
      "epoch": 7.51057783391222,
      "grad_norm": 4.127358913421631,
      "learning_rate": 3.4692839350243835e-05,
      "loss": 0.4218,
      "step": 23786
    },
    {
      "epoch": 7.510893590148405,
      "grad_norm": 4.319839954376221,
      "learning_rate": 3.469196224958776e-05,
      "loss": 0.468,
      "step": 23787
    },
    {
      "epoch": 7.511209346384591,
      "grad_norm": 4.181540489196777,
      "learning_rate": 3.46910851489317e-05,
      "loss": 0.4514,
      "step": 23788
    },
    {
      "epoch": 7.5115251026207766,
      "grad_norm": 5.603110313415527,
      "learning_rate": 3.4690208048275624e-05,
      "loss": 0.549,
      "step": 23789
    },
    {
      "epoch": 7.511840858856963,
      "grad_norm": 6.454756259918213,
      "learning_rate": 3.4689330947619555e-05,
      "loss": 0.4895,
      "step": 23790
    },
    {
      "epoch": 7.512156615093148,
      "grad_norm": 6.083267688751221,
      "learning_rate": 3.468845384696348e-05,
      "loss": 0.4974,
      "step": 23791
    },
    {
      "epoch": 7.512472371329334,
      "grad_norm": 5.127468585968018,
      "learning_rate": 3.468757674630741e-05,
      "loss": 0.5362,
      "step": 23792
    },
    {
      "epoch": 7.5127881275655195,
      "grad_norm": 5.303302764892578,
      "learning_rate": 3.468669964565134e-05,
      "loss": 0.5102,
      "step": 23793
    },
    {
      "epoch": 7.513103883801705,
      "grad_norm": 5.739170551300049,
      "learning_rate": 3.468582254499527e-05,
      "loss": 0.414,
      "step": 23794
    },
    {
      "epoch": 7.513419640037891,
      "grad_norm": 4.881004810333252,
      "learning_rate": 3.468494544433919e-05,
      "loss": 0.4118,
      "step": 23795
    },
    {
      "epoch": 7.513735396274076,
      "grad_norm": 6.089118480682373,
      "learning_rate": 3.4684068343683125e-05,
      "loss": 0.5054,
      "step": 23796
    },
    {
      "epoch": 7.5140511525102625,
      "grad_norm": 5.715967655181885,
      "learning_rate": 3.468319124302705e-05,
      "loss": 0.5553,
      "step": 23797
    },
    {
      "epoch": 7.514366908746448,
      "grad_norm": 6.870791912078857,
      "learning_rate": 3.468231414237098e-05,
      "loss": 0.5276,
      "step": 23798
    },
    {
      "epoch": 7.514682664982633,
      "grad_norm": 4.987016201019287,
      "learning_rate": 3.4681437041714907e-05,
      "loss": 0.4514,
      "step": 23799
    },
    {
      "epoch": 7.514998421218819,
      "grad_norm": 7.3851470947265625,
      "learning_rate": 3.468055994105884e-05,
      "loss": 0.5073,
      "step": 23800
    },
    {
      "epoch": 7.514998421218819,
      "eval_loss": 0.9365304708480835,
      "eval_runtime": 4.5334,
      "eval_samples_per_second": 549.261,
      "eval_steps_per_second": 4.632,
      "step": 23800
    },
    {
      "epoch": 7.5153141774550045,
      "grad_norm": 7.180239200592041,
      "learning_rate": 3.467968284040276e-05,
      "loss": 0.4805,
      "step": 23801
    },
    {
      "epoch": 7.515629933691191,
      "grad_norm": 5.601720809936523,
      "learning_rate": 3.4678805739746695e-05,
      "loss": 0.4997,
      "step": 23802
    },
    {
      "epoch": 7.515945689927376,
      "grad_norm": 5.677399635314941,
      "learning_rate": 3.4677928639090626e-05,
      "loss": 0.4329,
      "step": 23803
    },
    {
      "epoch": 7.516261446163561,
      "grad_norm": 5.89307165145874,
      "learning_rate": 3.467705153843455e-05,
      "loss": 0.5151,
      "step": 23804
    },
    {
      "epoch": 7.5165772023997475,
      "grad_norm": 5.65185546875,
      "learning_rate": 3.467617443777848e-05,
      "loss": 0.4827,
      "step": 23805
    },
    {
      "epoch": 7.516892958635933,
      "grad_norm": 6.527129650115967,
      "learning_rate": 3.467529733712241e-05,
      "loss": 0.4466,
      "step": 23806
    },
    {
      "epoch": 7.517208714872119,
      "grad_norm": 4.289577007293701,
      "learning_rate": 3.467442023646634e-05,
      "loss": 0.4606,
      "step": 23807
    },
    {
      "epoch": 7.517524471108304,
      "grad_norm": 6.096625804901123,
      "learning_rate": 3.4673543135810265e-05,
      "loss": 0.4796,
      "step": 23808
    },
    {
      "epoch": 7.51784022734449,
      "grad_norm": 6.790071487426758,
      "learning_rate": 3.4672666035154196e-05,
      "loss": 0.4945,
      "step": 23809
    },
    {
      "epoch": 7.518155983580676,
      "grad_norm": 5.041111469268799,
      "learning_rate": 3.467178893449812e-05,
      "loss": 0.4749,
      "step": 23810
    },
    {
      "epoch": 7.518471739816861,
      "grad_norm": 5.421207427978516,
      "learning_rate": 3.467091183384205e-05,
      "loss": 0.5056,
      "step": 23811
    },
    {
      "epoch": 7.518787496053047,
      "grad_norm": 5.745044231414795,
      "learning_rate": 3.467003473318598e-05,
      "loss": 0.4997,
      "step": 23812
    },
    {
      "epoch": 7.5191032522892325,
      "grad_norm": 5.262537479400635,
      "learning_rate": 3.466915763252991e-05,
      "loss": 0.4967,
      "step": 23813
    },
    {
      "epoch": 7.519419008525419,
      "grad_norm": 4.561779499053955,
      "learning_rate": 3.4668280531873834e-05,
      "loss": 0.3874,
      "step": 23814
    },
    {
      "epoch": 7.519734764761604,
      "grad_norm": 5.751625061035156,
      "learning_rate": 3.466740343121777e-05,
      "loss": 0.4082,
      "step": 23815
    },
    {
      "epoch": 7.52005052099779,
      "grad_norm": 6.371181964874268,
      "learning_rate": 3.46665263305617e-05,
      "loss": 0.4348,
      "step": 23816
    },
    {
      "epoch": 7.520366277233975,
      "grad_norm": 5.141236305236816,
      "learning_rate": 3.466564922990563e-05,
      "loss": 0.4962,
      "step": 23817
    },
    {
      "epoch": 7.520682033470161,
      "grad_norm": 6.761542797088623,
      "learning_rate": 3.4664772129249554e-05,
      "loss": 0.4992,
      "step": 23818
    },
    {
      "epoch": 7.520997789706347,
      "grad_norm": 5.370831489562988,
      "learning_rate": 3.4663895028593486e-05,
      "loss": 0.5285,
      "step": 23819
    },
    {
      "epoch": 7.521313545942532,
      "grad_norm": 5.294628143310547,
      "learning_rate": 3.466301792793741e-05,
      "loss": 0.4176,
      "step": 23820
    },
    {
      "epoch": 7.521629302178718,
      "grad_norm": 6.262225151062012,
      "learning_rate": 3.466214082728134e-05,
      "loss": 0.6233,
      "step": 23821
    },
    {
      "epoch": 7.521945058414904,
      "grad_norm": 7.379168510437012,
      "learning_rate": 3.466126372662527e-05,
      "loss": 0.5392,
      "step": 23822
    },
    {
      "epoch": 7.522260814651089,
      "grad_norm": 8.231832504272461,
      "learning_rate": 3.46603866259692e-05,
      "loss": 0.5049,
      "step": 23823
    },
    {
      "epoch": 7.522576570887275,
      "grad_norm": 5.662160396575928,
      "learning_rate": 3.465950952531313e-05,
      "loss": 0.4491,
      "step": 23824
    },
    {
      "epoch": 7.5228923271234605,
      "grad_norm": 5.455349445343018,
      "learning_rate": 3.4658632424657056e-05,
      "loss": 0.5535,
      "step": 23825
    },
    {
      "epoch": 7.523208083359647,
      "grad_norm": 4.474159240722656,
      "learning_rate": 3.465775532400099e-05,
      "loss": 0.485,
      "step": 23826
    },
    {
      "epoch": 7.523523839595832,
      "grad_norm": 4.539592266082764,
      "learning_rate": 3.465687822334491e-05,
      "loss": 0.4246,
      "step": 23827
    },
    {
      "epoch": 7.523839595832017,
      "grad_norm": 4.489493370056152,
      "learning_rate": 3.4656001122688844e-05,
      "loss": 0.4384,
      "step": 23828
    },
    {
      "epoch": 7.524155352068203,
      "grad_norm": 5.399844646453857,
      "learning_rate": 3.465512402203277e-05,
      "loss": 0.5497,
      "step": 23829
    },
    {
      "epoch": 7.524471108304389,
      "grad_norm": 6.37155294418335,
      "learning_rate": 3.46542469213767e-05,
      "loss": 0.5088,
      "step": 23830
    },
    {
      "epoch": 7.524786864540575,
      "grad_norm": 6.942257881164551,
      "learning_rate": 3.4653369820720625e-05,
      "loss": 0.5502,
      "step": 23831
    },
    {
      "epoch": 7.52510262077676,
      "grad_norm": 5.60867166519165,
      "learning_rate": 3.465249272006456e-05,
      "loss": 0.4774,
      "step": 23832
    },
    {
      "epoch": 7.525418377012946,
      "grad_norm": 5.814034938812256,
      "learning_rate": 3.465161561940848e-05,
      "loss": 0.4873,
      "step": 23833
    },
    {
      "epoch": 7.525734133249132,
      "grad_norm": 4.814878463745117,
      "learning_rate": 3.4650738518752414e-05,
      "loss": 0.4232,
      "step": 23834
    },
    {
      "epoch": 7.526049889485318,
      "grad_norm": 5.73795747756958,
      "learning_rate": 3.464986141809634e-05,
      "loss": 0.5326,
      "step": 23835
    },
    {
      "epoch": 7.526365645721503,
      "grad_norm": 5.271274566650391,
      "learning_rate": 3.464898431744027e-05,
      "loss": 0.401,
      "step": 23836
    },
    {
      "epoch": 7.526681401957688,
      "grad_norm": 6.506627559661865,
      "learning_rate": 3.4648107216784195e-05,
      "loss": 0.4994,
      "step": 23837
    },
    {
      "epoch": 7.526997158193875,
      "grad_norm": 4.964332580566406,
      "learning_rate": 3.464723011612813e-05,
      "loss": 0.4293,
      "step": 23838
    },
    {
      "epoch": 7.52731291443006,
      "grad_norm": 5.553145885467529,
      "learning_rate": 3.464635301547205e-05,
      "loss": 0.4341,
      "step": 23839
    },
    {
      "epoch": 7.527628670666246,
      "grad_norm": 5.132135391235352,
      "learning_rate": 3.4645475914815984e-05,
      "loss": 0.5058,
      "step": 23840
    },
    {
      "epoch": 7.527944426902431,
      "grad_norm": 5.237412929534912,
      "learning_rate": 3.4644598814159915e-05,
      "loss": 0.46,
      "step": 23841
    },
    {
      "epoch": 7.528260183138617,
      "grad_norm": 4.974216938018799,
      "learning_rate": 3.464372171350385e-05,
      "loss": 0.4581,
      "step": 23842
    },
    {
      "epoch": 7.528575939374803,
      "grad_norm": 6.280976295471191,
      "learning_rate": 3.464284461284777e-05,
      "loss": 0.5318,
      "step": 23843
    },
    {
      "epoch": 7.528891695610988,
      "grad_norm": 4.606147289276123,
      "learning_rate": 3.4641967512191703e-05,
      "loss": 0.4882,
      "step": 23844
    },
    {
      "epoch": 7.529207451847174,
      "grad_norm": 5.472376823425293,
      "learning_rate": 3.464109041153563e-05,
      "loss": 0.4534,
      "step": 23845
    },
    {
      "epoch": 7.52952320808336,
      "grad_norm": 5.2219085693359375,
      "learning_rate": 3.464021331087956e-05,
      "loss": 0.4443,
      "step": 23846
    },
    {
      "epoch": 7.529838964319545,
      "grad_norm": 4.6290388107299805,
      "learning_rate": 3.463933621022349e-05,
      "loss": 0.4731,
      "step": 23847
    },
    {
      "epoch": 7.530154720555731,
      "grad_norm": 4.779530048370361,
      "learning_rate": 3.463845910956742e-05,
      "loss": 0.4759,
      "step": 23848
    },
    {
      "epoch": 7.530470476791916,
      "grad_norm": 5.702219009399414,
      "learning_rate": 3.463758200891135e-05,
      "loss": 0.4126,
      "step": 23849
    },
    {
      "epoch": 7.530786233028103,
      "grad_norm": 5.273904323577881,
      "learning_rate": 3.463670490825527e-05,
      "loss": 0.3596,
      "step": 23850
    },
    {
      "epoch": 7.531101989264288,
      "grad_norm": 6.2230305671691895,
      "learning_rate": 3.4635827807599205e-05,
      "loss": 0.4051,
      "step": 23851
    },
    {
      "epoch": 7.531417745500473,
      "grad_norm": 5.309213161468506,
      "learning_rate": 3.463495070694313e-05,
      "loss": 0.5603,
      "step": 23852
    },
    {
      "epoch": 7.531733501736659,
      "grad_norm": 4.69700813293457,
      "learning_rate": 3.463407360628706e-05,
      "loss": 0.4474,
      "step": 23853
    },
    {
      "epoch": 7.532049257972845,
      "grad_norm": 7.342020511627197,
      "learning_rate": 3.4633196505630986e-05,
      "loss": 0.5368,
      "step": 23854
    },
    {
      "epoch": 7.532365014209031,
      "grad_norm": 6.325178146362305,
      "learning_rate": 3.463231940497492e-05,
      "loss": 0.4402,
      "step": 23855
    },
    {
      "epoch": 7.532680770445216,
      "grad_norm": 6.732951641082764,
      "learning_rate": 3.463144230431884e-05,
      "loss": 0.5314,
      "step": 23856
    },
    {
      "epoch": 7.532996526681402,
      "grad_norm": 6.837629795074463,
      "learning_rate": 3.4630565203662775e-05,
      "loss": 0.5333,
      "step": 23857
    },
    {
      "epoch": 7.533312282917588,
      "grad_norm": 6.191009044647217,
      "learning_rate": 3.46296881030067e-05,
      "loss": 0.4973,
      "step": 23858
    },
    {
      "epoch": 7.533628039153774,
      "grad_norm": 6.047267913818359,
      "learning_rate": 3.462881100235063e-05,
      "loss": 0.454,
      "step": 23859
    },
    {
      "epoch": 7.533943795389959,
      "grad_norm": 6.211548805236816,
      "learning_rate": 3.4627933901694556e-05,
      "loss": 0.3936,
      "step": 23860
    },
    {
      "epoch": 7.534259551626144,
      "grad_norm": 4.568721294403076,
      "learning_rate": 3.462705680103849e-05,
      "loss": 0.3374,
      "step": 23861
    },
    {
      "epoch": 7.5345753078623305,
      "grad_norm": 5.649531841278076,
      "learning_rate": 3.462617970038242e-05,
      "loss": 0.3669,
      "step": 23862
    },
    {
      "epoch": 7.534891064098516,
      "grad_norm": 8.198132514953613,
      "learning_rate": 3.4625302599726344e-05,
      "loss": 0.5875,
      "step": 23863
    },
    {
      "epoch": 7.535206820334702,
      "grad_norm": 5.499093055725098,
      "learning_rate": 3.4624425499070276e-05,
      "loss": 0.5269,
      "step": 23864
    },
    {
      "epoch": 7.535522576570887,
      "grad_norm": 5.21810245513916,
      "learning_rate": 3.46235483984142e-05,
      "loss": 0.4622,
      "step": 23865
    },
    {
      "epoch": 7.535838332807073,
      "grad_norm": 5.878455638885498,
      "learning_rate": 3.462267129775813e-05,
      "loss": 0.5326,
      "step": 23866
    },
    {
      "epoch": 7.536154089043259,
      "grad_norm": 5.189730644226074,
      "learning_rate": 3.462179419710206e-05,
      "loss": 0.4426,
      "step": 23867
    },
    {
      "epoch": 7.536469845279444,
      "grad_norm": 4.933535099029541,
      "learning_rate": 3.462091709644599e-05,
      "loss": 0.4202,
      "step": 23868
    },
    {
      "epoch": 7.53678560151563,
      "grad_norm": 5.258825778961182,
      "learning_rate": 3.4620039995789914e-05,
      "loss": 0.4925,
      "step": 23869
    },
    {
      "epoch": 7.5371013577518156,
      "grad_norm": 5.343934535980225,
      "learning_rate": 3.461916289513385e-05,
      "loss": 0.4189,
      "step": 23870
    },
    {
      "epoch": 7.537417113988001,
      "grad_norm": 6.456643581390381,
      "learning_rate": 3.461828579447778e-05,
      "loss": 0.5971,
      "step": 23871
    },
    {
      "epoch": 7.537732870224187,
      "grad_norm": 5.1798996925354,
      "learning_rate": 3.461740869382171e-05,
      "loss": 0.4015,
      "step": 23872
    },
    {
      "epoch": 7.538048626460372,
      "grad_norm": 5.735329627990723,
      "learning_rate": 3.4616531593165634e-05,
      "loss": 0.5167,
      "step": 23873
    },
    {
      "epoch": 7.5383643826965585,
      "grad_norm": 4.7088117599487305,
      "learning_rate": 3.4615654492509566e-05,
      "loss": 0.4156,
      "step": 23874
    },
    {
      "epoch": 7.538680138932744,
      "grad_norm": 6.635647773742676,
      "learning_rate": 3.461477739185349e-05,
      "loss": 0.5406,
      "step": 23875
    },
    {
      "epoch": 7.538995895168929,
      "grad_norm": 5.945004463195801,
      "learning_rate": 3.461390029119742e-05,
      "loss": 0.4582,
      "step": 23876
    },
    {
      "epoch": 7.539311651405115,
      "grad_norm": 6.657668590545654,
      "learning_rate": 3.461302319054135e-05,
      "loss": 0.5163,
      "step": 23877
    },
    {
      "epoch": 7.539627407641301,
      "grad_norm": 5.98282527923584,
      "learning_rate": 3.461214608988528e-05,
      "loss": 0.5375,
      "step": 23878
    },
    {
      "epoch": 7.539943163877487,
      "grad_norm": 5.797229290008545,
      "learning_rate": 3.4611268989229204e-05,
      "loss": 0.5136,
      "step": 23879
    },
    {
      "epoch": 7.540258920113672,
      "grad_norm": 5.313650131225586,
      "learning_rate": 3.4610391888573136e-05,
      "loss": 0.5755,
      "step": 23880
    },
    {
      "epoch": 7.540574676349858,
      "grad_norm": 5.160175323486328,
      "learning_rate": 3.460951478791706e-05,
      "loss": 0.4342,
      "step": 23881
    },
    {
      "epoch": 7.5408904325860435,
      "grad_norm": 7.436343193054199,
      "learning_rate": 3.460863768726099e-05,
      "loss": 0.4913,
      "step": 23882
    },
    {
      "epoch": 7.54120618882223,
      "grad_norm": 5.83310604095459,
      "learning_rate": 3.460776058660492e-05,
      "loss": 0.5417,
      "step": 23883
    },
    {
      "epoch": 7.541521945058415,
      "grad_norm": 5.315715789794922,
      "learning_rate": 3.460688348594885e-05,
      "loss": 0.3864,
      "step": 23884
    },
    {
      "epoch": 7.5418377012946,
      "grad_norm": 5.257015228271484,
      "learning_rate": 3.460600638529278e-05,
      "loss": 0.4666,
      "step": 23885
    },
    {
      "epoch": 7.5421534575307865,
      "grad_norm": 5.092372417449951,
      "learning_rate": 3.4605129284636705e-05,
      "loss": 0.4219,
      "step": 23886
    },
    {
      "epoch": 7.542469213766972,
      "grad_norm": 7.11730432510376,
      "learning_rate": 3.460425218398064e-05,
      "loss": 0.5011,
      "step": 23887
    },
    {
      "epoch": 7.542784970003158,
      "grad_norm": 6.1827616691589355,
      "learning_rate": 3.460337508332456e-05,
      "loss": 0.5293,
      "step": 23888
    },
    {
      "epoch": 7.543100726239343,
      "grad_norm": 5.554017066955566,
      "learning_rate": 3.4602497982668494e-05,
      "loss": 0.4618,
      "step": 23889
    },
    {
      "epoch": 7.5434164824755285,
      "grad_norm": 6.098814010620117,
      "learning_rate": 3.460162088201242e-05,
      "loss": 0.5301,
      "step": 23890
    },
    {
      "epoch": 7.543732238711715,
      "grad_norm": 6.161600112915039,
      "learning_rate": 3.460074378135635e-05,
      "loss": 0.5209,
      "step": 23891
    },
    {
      "epoch": 7.5440479949479,
      "grad_norm": 6.397136211395264,
      "learning_rate": 3.4599866680700275e-05,
      "loss": 0.4835,
      "step": 23892
    },
    {
      "epoch": 7.544363751184086,
      "grad_norm": 6.343918800354004,
      "learning_rate": 3.459898958004421e-05,
      "loss": 0.5748,
      "step": 23893
    },
    {
      "epoch": 7.5446795074202715,
      "grad_norm": 6.029354572296143,
      "learning_rate": 3.459811247938813e-05,
      "loss": 0.4806,
      "step": 23894
    },
    {
      "epoch": 7.544995263656457,
      "grad_norm": 5.719078063964844,
      "learning_rate": 3.4597235378732063e-05,
      "loss": 0.4707,
      "step": 23895
    },
    {
      "epoch": 7.545311019892643,
      "grad_norm": 4.213464736938477,
      "learning_rate": 3.459635827807599e-05,
      "loss": 0.3941,
      "step": 23896
    },
    {
      "epoch": 7.545626776128828,
      "grad_norm": 6.120460510253906,
      "learning_rate": 3.459548117741993e-05,
      "loss": 0.5559,
      "step": 23897
    },
    {
      "epoch": 7.545942532365014,
      "grad_norm": 5.40773868560791,
      "learning_rate": 3.459460407676385e-05,
      "loss": 0.4386,
      "step": 23898
    },
    {
      "epoch": 7.5462582886012,
      "grad_norm": 7.177593231201172,
      "learning_rate": 3.459372697610778e-05,
      "loss": 0.5024,
      "step": 23899
    },
    {
      "epoch": 7.546574044837386,
      "grad_norm": 6.365858554840088,
      "learning_rate": 3.459284987545171e-05,
      "loss": 0.3771,
      "step": 23900
    },
    {
      "epoch": 7.546889801073571,
      "grad_norm": 5.284698963165283,
      "learning_rate": 3.459197277479564e-05,
      "loss": 0.4465,
      "step": 23901
    },
    {
      "epoch": 7.5472055573097565,
      "grad_norm": 6.577608108520508,
      "learning_rate": 3.4591095674139565e-05,
      "loss": 0.4529,
      "step": 23902
    },
    {
      "epoch": 7.547521313545943,
      "grad_norm": 6.509596347808838,
      "learning_rate": 3.4590218573483497e-05,
      "loss": 0.5257,
      "step": 23903
    },
    {
      "epoch": 7.547837069782128,
      "grad_norm": 5.864702224731445,
      "learning_rate": 3.458934147282742e-05,
      "loss": 0.4654,
      "step": 23904
    },
    {
      "epoch": 7.548152826018314,
      "grad_norm": 5.8312458992004395,
      "learning_rate": 3.458846437217135e-05,
      "loss": 0.4703,
      "step": 23905
    },
    {
      "epoch": 7.5484685822544995,
      "grad_norm": 6.4782795906066895,
      "learning_rate": 3.4587587271515285e-05,
      "loss": 0.4696,
      "step": 23906
    },
    {
      "epoch": 7.548784338490686,
      "grad_norm": 6.217597961425781,
      "learning_rate": 3.458671017085921e-05,
      "loss": 0.5183,
      "step": 23907
    },
    {
      "epoch": 7.549100094726871,
      "grad_norm": 6.493387222290039,
      "learning_rate": 3.458583307020314e-05,
      "loss": 0.5188,
      "step": 23908
    },
    {
      "epoch": 7.549415850963056,
      "grad_norm": 6.378678321838379,
      "learning_rate": 3.4584955969547066e-05,
      "loss": 0.5093,
      "step": 23909
    },
    {
      "epoch": 7.549731607199242,
      "grad_norm": 8.446389198303223,
      "learning_rate": 3.4584078868891e-05,
      "loss": 0.5405,
      "step": 23910
    },
    {
      "epoch": 7.550047363435428,
      "grad_norm": 6.874025821685791,
      "learning_rate": 3.458320176823492e-05,
      "loss": 0.459,
      "step": 23911
    },
    {
      "epoch": 7.550363119671614,
      "grad_norm": 5.497379302978516,
      "learning_rate": 3.4582324667578855e-05,
      "loss": 0.509,
      "step": 23912
    },
    {
      "epoch": 7.550678875907799,
      "grad_norm": 5.411712169647217,
      "learning_rate": 3.458144756692278e-05,
      "loss": 0.4793,
      "step": 23913
    },
    {
      "epoch": 7.5509946321439845,
      "grad_norm": 5.764706134796143,
      "learning_rate": 3.458057046626671e-05,
      "loss": 0.6124,
      "step": 23914
    },
    {
      "epoch": 7.551310388380171,
      "grad_norm": 5.947817325592041,
      "learning_rate": 3.4579693365610636e-05,
      "loss": 0.5877,
      "step": 23915
    },
    {
      "epoch": 7.551626144616356,
      "grad_norm": 5.106548309326172,
      "learning_rate": 3.457881626495457e-05,
      "loss": 0.4134,
      "step": 23916
    },
    {
      "epoch": 7.551941900852542,
      "grad_norm": 5.569694995880127,
      "learning_rate": 3.457793916429849e-05,
      "loss": 0.485,
      "step": 23917
    },
    {
      "epoch": 7.552257657088727,
      "grad_norm": 5.489075183868408,
      "learning_rate": 3.4577062063642424e-05,
      "loss": 0.4959,
      "step": 23918
    },
    {
      "epoch": 7.552573413324913,
      "grad_norm": 6.534596920013428,
      "learning_rate": 3.457618496298635e-05,
      "loss": 0.6063,
      "step": 23919
    },
    {
      "epoch": 7.552889169561099,
      "grad_norm": 5.205744743347168,
      "learning_rate": 3.457530786233028e-05,
      "loss": 0.4005,
      "step": 23920
    },
    {
      "epoch": 7.553204925797284,
      "grad_norm": 5.92136287689209,
      "learning_rate": 3.4574430761674206e-05,
      "loss": 0.5369,
      "step": 23921
    },
    {
      "epoch": 7.55352068203347,
      "grad_norm": 5.1785407066345215,
      "learning_rate": 3.457355366101814e-05,
      "loss": 0.3519,
      "step": 23922
    },
    {
      "epoch": 7.553836438269656,
      "grad_norm": 5.695662975311279,
      "learning_rate": 3.457267656036207e-05,
      "loss": 0.5712,
      "step": 23923
    },
    {
      "epoch": 7.554152194505842,
      "grad_norm": 4.079189777374268,
      "learning_rate": 3.4571799459706e-05,
      "loss": 0.3751,
      "step": 23924
    },
    {
      "epoch": 7.554467950742027,
      "grad_norm": 6.0180134773254395,
      "learning_rate": 3.4570922359049926e-05,
      "loss": 0.5594,
      "step": 23925
    },
    {
      "epoch": 7.554783706978212,
      "grad_norm": 4.810778617858887,
      "learning_rate": 3.457004525839386e-05,
      "loss": 0.5118,
      "step": 23926
    },
    {
      "epoch": 7.555099463214399,
      "grad_norm": 5.849157810211182,
      "learning_rate": 3.456916815773778e-05,
      "loss": 0.5177,
      "step": 23927
    },
    {
      "epoch": 7.555415219450584,
      "grad_norm": 5.885441780090332,
      "learning_rate": 3.4568291057081714e-05,
      "loss": 0.4408,
      "step": 23928
    },
    {
      "epoch": 7.55573097568677,
      "grad_norm": 6.006387710571289,
      "learning_rate": 3.4567413956425646e-05,
      "loss": 0.3755,
      "step": 23929
    },
    {
      "epoch": 7.556046731922955,
      "grad_norm": 5.112981796264648,
      "learning_rate": 3.456653685576957e-05,
      "loss": 0.5093,
      "step": 23930
    },
    {
      "epoch": 7.556362488159142,
      "grad_norm": 4.894474029541016,
      "learning_rate": 3.45656597551135e-05,
      "loss": 0.4297,
      "step": 23931
    },
    {
      "epoch": 7.556678244395327,
      "grad_norm": 6.824679851531982,
      "learning_rate": 3.456478265445743e-05,
      "loss": 0.396,
      "step": 23932
    },
    {
      "epoch": 7.556994000631512,
      "grad_norm": 6.399559497833252,
      "learning_rate": 3.456390555380136e-05,
      "loss": 0.5663,
      "step": 23933
    },
    {
      "epoch": 7.557309756867698,
      "grad_norm": 6.163623809814453,
      "learning_rate": 3.4563028453145284e-05,
      "loss": 0.5718,
      "step": 23934
    },
    {
      "epoch": 7.557625513103884,
      "grad_norm": 6.895425796508789,
      "learning_rate": 3.4562151352489215e-05,
      "loss": 0.4851,
      "step": 23935
    },
    {
      "epoch": 7.55794126934007,
      "grad_norm": 5.435280799865723,
      "learning_rate": 3.456127425183314e-05,
      "loss": 0.4511,
      "step": 23936
    },
    {
      "epoch": 7.558257025576255,
      "grad_norm": 7.628452777862549,
      "learning_rate": 3.456039715117707e-05,
      "loss": 0.4831,
      "step": 23937
    },
    {
      "epoch": 7.55857278181244,
      "grad_norm": 7.8928751945495605,
      "learning_rate": 3.4559520050521e-05,
      "loss": 0.6151,
      "step": 23938
    },
    {
      "epoch": 7.558888538048627,
      "grad_norm": 5.433314323425293,
      "learning_rate": 3.455864294986493e-05,
      "loss": 0.44,
      "step": 23939
    },
    {
      "epoch": 7.559204294284812,
      "grad_norm": 5.616591453552246,
      "learning_rate": 3.4557765849208854e-05,
      "loss": 0.4882,
      "step": 23940
    },
    {
      "epoch": 7.559520050520998,
      "grad_norm": 6.332393169403076,
      "learning_rate": 3.4556888748552785e-05,
      "loss": 0.4747,
      "step": 23941
    },
    {
      "epoch": 7.559835806757183,
      "grad_norm": 5.3129563331604,
      "learning_rate": 3.455601164789671e-05,
      "loss": 0.4573,
      "step": 23942
    },
    {
      "epoch": 7.560151562993369,
      "grad_norm": 5.504077434539795,
      "learning_rate": 3.455513454724064e-05,
      "loss": 0.4495,
      "step": 23943
    },
    {
      "epoch": 7.560467319229555,
      "grad_norm": 5.211007595062256,
      "learning_rate": 3.4554257446584574e-05,
      "loss": 0.5275,
      "step": 23944
    },
    {
      "epoch": 7.56078307546574,
      "grad_norm": 5.255282878875732,
      "learning_rate": 3.45533803459285e-05,
      "loss": 0.4963,
      "step": 23945
    },
    {
      "epoch": 7.561098831701926,
      "grad_norm": 5.462660312652588,
      "learning_rate": 3.455250324527243e-05,
      "loss": 0.4322,
      "step": 23946
    },
    {
      "epoch": 7.561414587938112,
      "grad_norm": 6.754168510437012,
      "learning_rate": 3.4551626144616355e-05,
      "loss": 0.5319,
      "step": 23947
    },
    {
      "epoch": 7.561730344174298,
      "grad_norm": 4.584634304046631,
      "learning_rate": 3.455074904396029e-05,
      "loss": 0.4723,
      "step": 23948
    },
    {
      "epoch": 7.562046100410483,
      "grad_norm": 6.524698257446289,
      "learning_rate": 3.454987194330421e-05,
      "loss": 0.4917,
      "step": 23949
    },
    {
      "epoch": 7.562361856646669,
      "grad_norm": 5.621714115142822,
      "learning_rate": 3.454899484264814e-05,
      "loss": 0.4361,
      "step": 23950
    },
    {
      "epoch": 7.562677612882855,
      "grad_norm": 4.8745012283325195,
      "learning_rate": 3.4548117741992075e-05,
      "loss": 0.4235,
      "step": 23951
    },
    {
      "epoch": 7.56299336911904,
      "grad_norm": 5.344397068023682,
      "learning_rate": 3.454724064133601e-05,
      "loss": 0.5874,
      "step": 23952
    },
    {
      "epoch": 7.563309125355226,
      "grad_norm": 5.500855445861816,
      "learning_rate": 3.454636354067993e-05,
      "loss": 0.4995,
      "step": 23953
    },
    {
      "epoch": 7.563624881591411,
      "grad_norm": 5.622689723968506,
      "learning_rate": 3.454548644002386e-05,
      "loss": 0.4666,
      "step": 23954
    },
    {
      "epoch": 7.5639406378275975,
      "grad_norm": 6.632812976837158,
      "learning_rate": 3.454460933936779e-05,
      "loss": 0.4557,
      "step": 23955
    },
    {
      "epoch": 7.564256394063783,
      "grad_norm": 4.8448615074157715,
      "learning_rate": 3.454373223871172e-05,
      "loss": 0.3684,
      "step": 23956
    },
    {
      "epoch": 7.564572150299968,
      "grad_norm": 5.441974639892578,
      "learning_rate": 3.4542855138055645e-05,
      "loss": 0.5589,
      "step": 23957
    },
    {
      "epoch": 7.564887906536154,
      "grad_norm": 5.41159725189209,
      "learning_rate": 3.4541978037399576e-05,
      "loss": 0.529,
      "step": 23958
    },
    {
      "epoch": 7.56520366277234,
      "grad_norm": 5.3168625831604,
      "learning_rate": 3.45411009367435e-05,
      "loss": 0.4965,
      "step": 23959
    },
    {
      "epoch": 7.565519419008526,
      "grad_norm": 5.885646343231201,
      "learning_rate": 3.454022383608743e-05,
      "loss": 0.523,
      "step": 23960
    },
    {
      "epoch": 7.565835175244711,
      "grad_norm": 5.759611129760742,
      "learning_rate": 3.453934673543136e-05,
      "loss": 0.4615,
      "step": 23961
    },
    {
      "epoch": 7.566150931480896,
      "grad_norm": 6.237578392028809,
      "learning_rate": 3.453846963477529e-05,
      "loss": 0.5844,
      "step": 23962
    },
    {
      "epoch": 7.5664666877170825,
      "grad_norm": 4.717855930328369,
      "learning_rate": 3.4537592534119215e-05,
      "loss": 0.4916,
      "step": 23963
    },
    {
      "epoch": 7.566782443953268,
      "grad_norm": 6.014033794403076,
      "learning_rate": 3.4536715433463146e-05,
      "loss": 0.394,
      "step": 23964
    },
    {
      "epoch": 7.567098200189454,
      "grad_norm": 5.484015941619873,
      "learning_rate": 3.453583833280708e-05,
      "loss": 0.4581,
      "step": 23965
    },
    {
      "epoch": 7.567413956425639,
      "grad_norm": 5.327047348022461,
      "learning_rate": 3.4534961232151e-05,
      "loss": 0.4287,
      "step": 23966
    },
    {
      "epoch": 7.567729712661825,
      "grad_norm": 5.093648433685303,
      "learning_rate": 3.4534084131494934e-05,
      "loss": 0.4413,
      "step": 23967
    },
    {
      "epoch": 7.568045468898011,
      "grad_norm": 5.055325984954834,
      "learning_rate": 3.453320703083886e-05,
      "loss": 0.3514,
      "step": 23968
    },
    {
      "epoch": 7.568361225134196,
      "grad_norm": 5.386600971221924,
      "learning_rate": 3.453232993018279e-05,
      "loss": 0.4381,
      "step": 23969
    },
    {
      "epoch": 7.568676981370382,
      "grad_norm": 4.979127407073975,
      "learning_rate": 3.4531452829526716e-05,
      "loss": 0.5192,
      "step": 23970
    },
    {
      "epoch": 7.5689927376065675,
      "grad_norm": 4.699256420135498,
      "learning_rate": 3.453057572887065e-05,
      "loss": 0.5145,
      "step": 23971
    },
    {
      "epoch": 7.569308493842754,
      "grad_norm": 4.973723411560059,
      "learning_rate": 3.452969862821457e-05,
      "loss": 0.4195,
      "step": 23972
    },
    {
      "epoch": 7.569624250078939,
      "grad_norm": 5.514206409454346,
      "learning_rate": 3.4528821527558504e-05,
      "loss": 0.5797,
      "step": 23973
    },
    {
      "epoch": 7.569940006315125,
      "grad_norm": 5.097771167755127,
      "learning_rate": 3.452794442690243e-05,
      "loss": 0.5219,
      "step": 23974
    },
    {
      "epoch": 7.5702557625513105,
      "grad_norm": 5.3795485496521,
      "learning_rate": 3.452706732624636e-05,
      "loss": 0.4609,
      "step": 23975
    },
    {
      "epoch": 7.570571518787496,
      "grad_norm": 5.822205543518066,
      "learning_rate": 3.4526190225590286e-05,
      "loss": 0.5329,
      "step": 23976
    },
    {
      "epoch": 7.570887275023682,
      "grad_norm": 5.41278076171875,
      "learning_rate": 3.452531312493422e-05,
      "loss": 0.454,
      "step": 23977
    },
    {
      "epoch": 7.571203031259867,
      "grad_norm": 5.82535982131958,
      "learning_rate": 3.452443602427814e-05,
      "loss": 0.4851,
      "step": 23978
    },
    {
      "epoch": 7.5715187874960534,
      "grad_norm": 4.315293788909912,
      "learning_rate": 3.452355892362208e-05,
      "loss": 0.4398,
      "step": 23979
    },
    {
      "epoch": 7.571834543732239,
      "grad_norm": 5.8367085456848145,
      "learning_rate": 3.4522681822966006e-05,
      "loss": 0.5562,
      "step": 23980
    },
    {
      "epoch": 7.572150299968424,
      "grad_norm": 5.32027006149292,
      "learning_rate": 3.452180472230994e-05,
      "loss": 0.3789,
      "step": 23981
    },
    {
      "epoch": 7.57246605620461,
      "grad_norm": 5.880457401275635,
      "learning_rate": 3.452092762165386e-05,
      "loss": 0.4504,
      "step": 23982
    },
    {
      "epoch": 7.5727818124407955,
      "grad_norm": 6.5490875244140625,
      "learning_rate": 3.4520050520997794e-05,
      "loss": 0.4434,
      "step": 23983
    },
    {
      "epoch": 7.573097568676982,
      "grad_norm": 5.811532974243164,
      "learning_rate": 3.451917342034172e-05,
      "loss": 0.5141,
      "step": 23984
    },
    {
      "epoch": 7.573413324913167,
      "grad_norm": 5.4566569328308105,
      "learning_rate": 3.451829631968565e-05,
      "loss": 0.5046,
      "step": 23985
    },
    {
      "epoch": 7.573729081149352,
      "grad_norm": 6.986639499664307,
      "learning_rate": 3.4517419219029575e-05,
      "loss": 0.4716,
      "step": 23986
    },
    {
      "epoch": 7.5740448373855385,
      "grad_norm": 7.518964767456055,
      "learning_rate": 3.451654211837351e-05,
      "loss": 0.4283,
      "step": 23987
    },
    {
      "epoch": 7.574360593621724,
      "grad_norm": 6.550130844116211,
      "learning_rate": 3.451566501771744e-05,
      "loss": 0.6807,
      "step": 23988
    },
    {
      "epoch": 7.57467634985791,
      "grad_norm": 6.88748025894165,
      "learning_rate": 3.4514787917061364e-05,
      "loss": 0.5137,
      "step": 23989
    },
    {
      "epoch": 7.574992106094095,
      "grad_norm": 6.803426265716553,
      "learning_rate": 3.4513910816405295e-05,
      "loss": 0.4966,
      "step": 23990
    },
    {
      "epoch": 7.5753078623302805,
      "grad_norm": 4.46879243850708,
      "learning_rate": 3.451303371574922e-05,
      "loss": 0.4754,
      "step": 23991
    },
    {
      "epoch": 7.575623618566467,
      "grad_norm": 6.029566764831543,
      "learning_rate": 3.451215661509315e-05,
      "loss": 0.571,
      "step": 23992
    },
    {
      "epoch": 7.575939374802652,
      "grad_norm": 5.920723915100098,
      "learning_rate": 3.451127951443708e-05,
      "loss": 0.491,
      "step": 23993
    },
    {
      "epoch": 7.576255131038838,
      "grad_norm": 5.847009181976318,
      "learning_rate": 3.451040241378101e-05,
      "loss": 0.4157,
      "step": 23994
    },
    {
      "epoch": 7.5765708872750235,
      "grad_norm": 6.1261186599731445,
      "learning_rate": 3.4509525313124933e-05,
      "loss": 0.4884,
      "step": 23995
    },
    {
      "epoch": 7.57688664351121,
      "grad_norm": 5.682343006134033,
      "learning_rate": 3.4508648212468865e-05,
      "loss": 0.5508,
      "step": 23996
    },
    {
      "epoch": 7.577202399747395,
      "grad_norm": 6.271111965179443,
      "learning_rate": 3.450777111181279e-05,
      "loss": 0.5685,
      "step": 23997
    },
    {
      "epoch": 7.577518155983581,
      "grad_norm": 5.3527069091796875,
      "learning_rate": 3.450689401115672e-05,
      "loss": 0.4822,
      "step": 23998
    },
    {
      "epoch": 7.577833912219766,
      "grad_norm": 5.49171257019043,
      "learning_rate": 3.450601691050065e-05,
      "loss": 0.5575,
      "step": 23999
    },
    {
      "epoch": 7.578149668455952,
      "grad_norm": 6.605659484863281,
      "learning_rate": 3.450513980984458e-05,
      "loss": 0.5035,
      "step": 24000
    },
    {
      "epoch": 7.578149668455952,
      "eval_loss": 0.9310944080352783,
      "eval_runtime": 4.5395,
      "eval_samples_per_second": 548.519,
      "eval_steps_per_second": 4.626,
      "step": 24000
    },
    {
      "epoch": 7.578465424692138,
      "grad_norm": 6.634766578674316,
      "learning_rate": 3.45042627091885e-05,
      "loss": 0.4414,
      "step": 24001
    },
    {
      "epoch": 7.578781180928323,
      "grad_norm": 4.5177321434021,
      "learning_rate": 3.4503385608532435e-05,
      "loss": 0.3909,
      "step": 24002
    },
    {
      "epoch": 7.579096937164509,
      "grad_norm": 5.925997734069824,
      "learning_rate": 3.4502508507876367e-05,
      "loss": 0.517,
      "step": 24003
    },
    {
      "epoch": 7.579412693400695,
      "grad_norm": 6.505760192871094,
      "learning_rate": 3.450163140722029e-05,
      "loss": 0.5255,
      "step": 24004
    },
    {
      "epoch": 7.57972844963688,
      "grad_norm": 4.641974925994873,
      "learning_rate": 3.450075430656422e-05,
      "loss": 0.4766,
      "step": 24005
    },
    {
      "epoch": 7.580044205873066,
      "grad_norm": 7.492923736572266,
      "learning_rate": 3.4499877205908155e-05,
      "loss": 0.5119,
      "step": 24006
    },
    {
      "epoch": 7.5803599621092514,
      "grad_norm": 5.068393230438232,
      "learning_rate": 3.449900010525208e-05,
      "loss": 0.4574,
      "step": 24007
    },
    {
      "epoch": 7.580675718345438,
      "grad_norm": 5.841287136077881,
      "learning_rate": 3.449812300459601e-05,
      "loss": 0.4899,
      "step": 24008
    },
    {
      "epoch": 7.580991474581623,
      "grad_norm": 4.866267681121826,
      "learning_rate": 3.449724590393994e-05,
      "loss": 0.4069,
      "step": 24009
    },
    {
      "epoch": 7.581307230817808,
      "grad_norm": 5.504018783569336,
      "learning_rate": 3.449636880328387e-05,
      "loss": 0.4248,
      "step": 24010
    },
    {
      "epoch": 7.581622987053994,
      "grad_norm": 4.045665264129639,
      "learning_rate": 3.44954917026278e-05,
      "loss": 0.4098,
      "step": 24011
    },
    {
      "epoch": 7.58193874329018,
      "grad_norm": 4.304111003875732,
      "learning_rate": 3.4494614601971725e-05,
      "loss": 0.4649,
      "step": 24012
    },
    {
      "epoch": 7.582254499526366,
      "grad_norm": 5.47765588760376,
      "learning_rate": 3.4493737501315656e-05,
      "loss": 0.4218,
      "step": 24013
    },
    {
      "epoch": 7.582570255762551,
      "grad_norm": 5.9150543212890625,
      "learning_rate": 3.449286040065958e-05,
      "loss": 0.5005,
      "step": 24014
    },
    {
      "epoch": 7.582886011998737,
      "grad_norm": 6.7191057205200195,
      "learning_rate": 3.449198330000351e-05,
      "loss": 0.4402,
      "step": 24015
    },
    {
      "epoch": 7.583201768234923,
      "grad_norm": 5.64273738861084,
      "learning_rate": 3.449110619934744e-05,
      "loss": 0.4453,
      "step": 24016
    },
    {
      "epoch": 7.583517524471108,
      "grad_norm": 4.553785800933838,
      "learning_rate": 3.449022909869137e-05,
      "loss": 0.45,
      "step": 24017
    },
    {
      "epoch": 7.583833280707294,
      "grad_norm": 5.741030216217041,
      "learning_rate": 3.4489351998035294e-05,
      "loss": 0.4648,
      "step": 24018
    },
    {
      "epoch": 7.584149036943479,
      "grad_norm": 4.55269193649292,
      "learning_rate": 3.4488474897379226e-05,
      "loss": 0.4294,
      "step": 24019
    },
    {
      "epoch": 7.584464793179666,
      "grad_norm": 6.184386253356934,
      "learning_rate": 3.448759779672315e-05,
      "loss": 0.5262,
      "step": 24020
    },
    {
      "epoch": 7.584780549415851,
      "grad_norm": 4.884706974029541,
      "learning_rate": 3.448672069606708e-05,
      "loss": 0.4318,
      "step": 24021
    },
    {
      "epoch": 7.585096305652037,
      "grad_norm": 6.591823577880859,
      "learning_rate": 3.448584359541101e-05,
      "loss": 0.4976,
      "step": 24022
    },
    {
      "epoch": 7.585412061888222,
      "grad_norm": 6.802276611328125,
      "learning_rate": 3.448496649475494e-05,
      "loss": 0.6032,
      "step": 24023
    },
    {
      "epoch": 7.585727818124408,
      "grad_norm": 5.697005271911621,
      "learning_rate": 3.4484089394098864e-05,
      "loss": 0.4369,
      "step": 24024
    },
    {
      "epoch": 7.586043574360594,
      "grad_norm": 5.360586166381836,
      "learning_rate": 3.4483212293442796e-05,
      "loss": 0.4712,
      "step": 24025
    },
    {
      "epoch": 7.586359330596779,
      "grad_norm": 6.111663818359375,
      "learning_rate": 3.448233519278673e-05,
      "loss": 0.5832,
      "step": 24026
    },
    {
      "epoch": 7.586675086832965,
      "grad_norm": 5.384833812713623,
      "learning_rate": 3.448145809213065e-05,
      "loss": 0.463,
      "step": 24027
    },
    {
      "epoch": 7.586990843069151,
      "grad_norm": 4.954183101654053,
      "learning_rate": 3.4480580991474584e-05,
      "loss": 0.5202,
      "step": 24028
    },
    {
      "epoch": 7.587306599305336,
      "grad_norm": 5.622772216796875,
      "learning_rate": 3.447970389081851e-05,
      "loss": 0.516,
      "step": 24029
    },
    {
      "epoch": 7.587622355541522,
      "grad_norm": 4.413492202758789,
      "learning_rate": 3.447882679016244e-05,
      "loss": 0.4541,
      "step": 24030
    },
    {
      "epoch": 7.587938111777707,
      "grad_norm": 5.753299713134766,
      "learning_rate": 3.4477949689506366e-05,
      "loss": 0.5481,
      "step": 24031
    },
    {
      "epoch": 7.588253868013894,
      "grad_norm": 6.148152828216553,
      "learning_rate": 3.44770725888503e-05,
      "loss": 0.5075,
      "step": 24032
    },
    {
      "epoch": 7.588569624250079,
      "grad_norm": 5.998506546020508,
      "learning_rate": 3.447619548819423e-05,
      "loss": 0.5659,
      "step": 24033
    },
    {
      "epoch": 7.588885380486264,
      "grad_norm": 7.530089855194092,
      "learning_rate": 3.447531838753816e-05,
      "loss": 0.5026,
      "step": 24034
    },
    {
      "epoch": 7.58920113672245,
      "grad_norm": 5.48845100402832,
      "learning_rate": 3.4474441286882086e-05,
      "loss": 0.5653,
      "step": 24035
    },
    {
      "epoch": 7.589516892958636,
      "grad_norm": 4.874171257019043,
      "learning_rate": 3.447356418622602e-05,
      "loss": 0.4919,
      "step": 24036
    },
    {
      "epoch": 7.589832649194822,
      "grad_norm": 5.506930828094482,
      "learning_rate": 3.447268708556994e-05,
      "loss": 0.5174,
      "step": 24037
    },
    {
      "epoch": 7.590148405431007,
      "grad_norm": 4.634296894073486,
      "learning_rate": 3.4471809984913874e-05,
      "loss": 0.4725,
      "step": 24038
    },
    {
      "epoch": 7.590464161667193,
      "grad_norm": 6.088910102844238,
      "learning_rate": 3.44709328842578e-05,
      "loss": 0.4775,
      "step": 24039
    },
    {
      "epoch": 7.590779917903379,
      "grad_norm": 7.68007755279541,
      "learning_rate": 3.447005578360173e-05,
      "loss": 0.5086,
      "step": 24040
    },
    {
      "epoch": 7.591095674139564,
      "grad_norm": 6.623610973358154,
      "learning_rate": 3.4469178682945655e-05,
      "loss": 0.4465,
      "step": 24041
    },
    {
      "epoch": 7.59141143037575,
      "grad_norm": 5.511361122131348,
      "learning_rate": 3.446830158228959e-05,
      "loss": 0.4808,
      "step": 24042
    },
    {
      "epoch": 7.591727186611935,
      "grad_norm": 5.180079936981201,
      "learning_rate": 3.446742448163351e-05,
      "loss": 0.5444,
      "step": 24043
    },
    {
      "epoch": 7.5920429428481215,
      "grad_norm": 6.134136199951172,
      "learning_rate": 3.4466547380977444e-05,
      "loss": 0.4357,
      "step": 24044
    },
    {
      "epoch": 7.592358699084307,
      "grad_norm": 5.177584171295166,
      "learning_rate": 3.446567028032137e-05,
      "loss": 0.4914,
      "step": 24045
    },
    {
      "epoch": 7.592674455320493,
      "grad_norm": 4.806359767913818,
      "learning_rate": 3.44647931796653e-05,
      "loss": 0.4613,
      "step": 24046
    },
    {
      "epoch": 7.592990211556678,
      "grad_norm": 5.762784481048584,
      "learning_rate": 3.446391607900923e-05,
      "loss": 0.4612,
      "step": 24047
    },
    {
      "epoch": 7.593305967792864,
      "grad_norm": 4.8347015380859375,
      "learning_rate": 3.446303897835316e-05,
      "loss": 0.4551,
      "step": 24048
    },
    {
      "epoch": 7.59362172402905,
      "grad_norm": 5.231103420257568,
      "learning_rate": 3.446216187769709e-05,
      "loss": 0.4603,
      "step": 24049
    },
    {
      "epoch": 7.593937480265235,
      "grad_norm": 5.149458885192871,
      "learning_rate": 3.446128477704101e-05,
      "loss": 0.478,
      "step": 24050
    },
    {
      "epoch": 7.594253236501421,
      "grad_norm": 6.282290458679199,
      "learning_rate": 3.4460407676384945e-05,
      "loss": 0.4499,
      "step": 24051
    },
    {
      "epoch": 7.5945689927376065,
      "grad_norm": 4.945362567901611,
      "learning_rate": 3.445953057572887e-05,
      "loss": 0.5096,
      "step": 24052
    },
    {
      "epoch": 7.594884748973792,
      "grad_norm": 5.635825157165527,
      "learning_rate": 3.44586534750728e-05,
      "loss": 0.5058,
      "step": 24053
    },
    {
      "epoch": 7.595200505209978,
      "grad_norm": 5.496294975280762,
      "learning_rate": 3.4457776374416727e-05,
      "loss": 0.4796,
      "step": 24054
    },
    {
      "epoch": 7.595516261446163,
      "grad_norm": 5.942119121551514,
      "learning_rate": 3.445689927376066e-05,
      "loss": 0.5206,
      "step": 24055
    },
    {
      "epoch": 7.5958320176823495,
      "grad_norm": 5.379788398742676,
      "learning_rate": 3.445602217310458e-05,
      "loss": 0.4696,
      "step": 24056
    },
    {
      "epoch": 7.596147773918535,
      "grad_norm": 3.94467830657959,
      "learning_rate": 3.4455145072448515e-05,
      "loss": 0.3757,
      "step": 24057
    },
    {
      "epoch": 7.59646353015472,
      "grad_norm": 6.214574813842773,
      "learning_rate": 3.445426797179244e-05,
      "loss": 0.4839,
      "step": 24058
    },
    {
      "epoch": 7.596779286390906,
      "grad_norm": 4.592757701873779,
      "learning_rate": 3.445339087113637e-05,
      "loss": 0.5147,
      "step": 24059
    },
    {
      "epoch": 7.597095042627092,
      "grad_norm": 4.286338806152344,
      "learning_rate": 3.44525137704803e-05,
      "loss": 0.5127,
      "step": 24060
    },
    {
      "epoch": 7.597410798863278,
      "grad_norm": 4.353888988494873,
      "learning_rate": 3.4451636669824235e-05,
      "loss": 0.4809,
      "step": 24061
    },
    {
      "epoch": 7.597726555099463,
      "grad_norm": 6.809926986694336,
      "learning_rate": 3.445075956916816e-05,
      "loss": 0.6437,
      "step": 24062
    },
    {
      "epoch": 7.598042311335649,
      "grad_norm": 5.629578113555908,
      "learning_rate": 3.444988246851209e-05,
      "loss": 0.4487,
      "step": 24063
    },
    {
      "epoch": 7.5983580675718345,
      "grad_norm": 4.457120895385742,
      "learning_rate": 3.4449005367856016e-05,
      "loss": 0.4336,
      "step": 24064
    },
    {
      "epoch": 7.598673823808021,
      "grad_norm": 4.704092025756836,
      "learning_rate": 3.444812826719995e-05,
      "loss": 0.4774,
      "step": 24065
    },
    {
      "epoch": 7.598989580044206,
      "grad_norm": 5.903083801269531,
      "learning_rate": 3.444725116654387e-05,
      "loss": 0.6125,
      "step": 24066
    },
    {
      "epoch": 7.599305336280391,
      "grad_norm": 6.274083137512207,
      "learning_rate": 3.4446374065887805e-05,
      "loss": 0.4882,
      "step": 24067
    },
    {
      "epoch": 7.5996210925165775,
      "grad_norm": 5.744048595428467,
      "learning_rate": 3.444549696523173e-05,
      "loss": 0.5184,
      "step": 24068
    },
    {
      "epoch": 7.599936848752763,
      "grad_norm": 7.234991073608398,
      "learning_rate": 3.444461986457566e-05,
      "loss": 0.5621,
      "step": 24069
    },
    {
      "epoch": 7.600252604988949,
      "grad_norm": 5.211428642272949,
      "learning_rate": 3.444374276391959e-05,
      "loss": 0.4928,
      "step": 24070
    },
    {
      "epoch": 7.600568361225134,
      "grad_norm": 5.551784515380859,
      "learning_rate": 3.444286566326352e-05,
      "loss": 0.4644,
      "step": 24071
    },
    {
      "epoch": 7.6008841174613195,
      "grad_norm": 6.091910362243652,
      "learning_rate": 3.444198856260745e-05,
      "loss": 0.5413,
      "step": 24072
    },
    {
      "epoch": 7.601199873697506,
      "grad_norm": 4.484320163726807,
      "learning_rate": 3.4441111461951374e-05,
      "loss": 0.4287,
      "step": 24073
    },
    {
      "epoch": 7.601515629933691,
      "grad_norm": 5.785983562469482,
      "learning_rate": 3.4440234361295306e-05,
      "loss": 0.4671,
      "step": 24074
    },
    {
      "epoch": 7.601831386169877,
      "grad_norm": 9.916656494140625,
      "learning_rate": 3.443935726063923e-05,
      "loss": 0.4532,
      "step": 24075
    },
    {
      "epoch": 7.6021471424060625,
      "grad_norm": 5.653786659240723,
      "learning_rate": 3.443848015998316e-05,
      "loss": 0.5736,
      "step": 24076
    },
    {
      "epoch": 7.602462898642248,
      "grad_norm": 5.0966105461120605,
      "learning_rate": 3.443760305932709e-05,
      "loss": 0.4237,
      "step": 24077
    },
    {
      "epoch": 7.602778654878434,
      "grad_norm": 5.633697986602783,
      "learning_rate": 3.443672595867102e-05,
      "loss": 0.4812,
      "step": 24078
    },
    {
      "epoch": 7.603094411114619,
      "grad_norm": 5.558786869049072,
      "learning_rate": 3.4435848858014944e-05,
      "loss": 0.4074,
      "step": 24079
    },
    {
      "epoch": 7.603410167350805,
      "grad_norm": 6.431313514709473,
      "learning_rate": 3.4434971757358876e-05,
      "loss": 0.6286,
      "step": 24080
    },
    {
      "epoch": 7.603725923586991,
      "grad_norm": 5.6727752685546875,
      "learning_rate": 3.44340946567028e-05,
      "loss": 0.4315,
      "step": 24081
    },
    {
      "epoch": 7.604041679823176,
      "grad_norm": 6.2559685707092285,
      "learning_rate": 3.443321755604673e-05,
      "loss": 0.4897,
      "step": 24082
    },
    {
      "epoch": 7.604357436059362,
      "grad_norm": 5.113485336303711,
      "learning_rate": 3.443234045539066e-05,
      "loss": 0.4754,
      "step": 24083
    },
    {
      "epoch": 7.6046731922955475,
      "grad_norm": 5.957728385925293,
      "learning_rate": 3.443146335473459e-05,
      "loss": 0.549,
      "step": 24084
    },
    {
      "epoch": 7.604988948531734,
      "grad_norm": 6.739568710327148,
      "learning_rate": 3.443058625407852e-05,
      "loss": 0.5572,
      "step": 24085
    },
    {
      "epoch": 7.605304704767919,
      "grad_norm": 5.745264530181885,
      "learning_rate": 3.4429709153422446e-05,
      "loss": 0.4931,
      "step": 24086
    },
    {
      "epoch": 7.605620461004105,
      "grad_norm": 5.903539180755615,
      "learning_rate": 3.442883205276638e-05,
      "loss": 0.6718,
      "step": 24087
    },
    {
      "epoch": 7.6059362172402905,
      "grad_norm": 6.480398178100586,
      "learning_rate": 3.442795495211031e-05,
      "loss": 0.5476,
      "step": 24088
    },
    {
      "epoch": 7.606251973476477,
      "grad_norm": 6.079104423522949,
      "learning_rate": 3.4427077851454234e-05,
      "loss": 0.4395,
      "step": 24089
    },
    {
      "epoch": 7.606567729712662,
      "grad_norm": 7.493100643157959,
      "learning_rate": 3.4426200750798165e-05,
      "loss": 0.4411,
      "step": 24090
    },
    {
      "epoch": 7.606883485948847,
      "grad_norm": 6.174495220184326,
      "learning_rate": 3.44253236501421e-05,
      "loss": 0.5162,
      "step": 24091
    },
    {
      "epoch": 7.607199242185033,
      "grad_norm": 5.584884166717529,
      "learning_rate": 3.442444654948602e-05,
      "loss": 0.4732,
      "step": 24092
    },
    {
      "epoch": 7.607514998421219,
      "grad_norm": 6.0397562980651855,
      "learning_rate": 3.4423569448829954e-05,
      "loss": 0.4792,
      "step": 24093
    },
    {
      "epoch": 7.607830754657405,
      "grad_norm": 5.010749340057373,
      "learning_rate": 3.442269234817388e-05,
      "loss": 0.4914,
      "step": 24094
    },
    {
      "epoch": 7.60814651089359,
      "grad_norm": 7.036653518676758,
      "learning_rate": 3.442181524751781e-05,
      "loss": 0.5394,
      "step": 24095
    },
    {
      "epoch": 7.6084622671297755,
      "grad_norm": 5.863152980804443,
      "learning_rate": 3.4420938146861735e-05,
      "loss": 0.4924,
      "step": 24096
    },
    {
      "epoch": 7.608778023365962,
      "grad_norm": 4.2306413650512695,
      "learning_rate": 3.442006104620567e-05,
      "loss": 0.4758,
      "step": 24097
    },
    {
      "epoch": 7.609093779602147,
      "grad_norm": 6.244081497192383,
      "learning_rate": 3.441918394554959e-05,
      "loss": 0.5184,
      "step": 24098
    },
    {
      "epoch": 7.609409535838333,
      "grad_norm": 4.856373310089111,
      "learning_rate": 3.4418306844893523e-05,
      "loss": 0.3977,
      "step": 24099
    },
    {
      "epoch": 7.609725292074518,
      "grad_norm": 5.692080497741699,
      "learning_rate": 3.441742974423745e-05,
      "loss": 0.5294,
      "step": 24100
    },
    {
      "epoch": 7.610041048310704,
      "grad_norm": 5.8491950035095215,
      "learning_rate": 3.441655264358138e-05,
      "loss": 0.432,
      "step": 24101
    },
    {
      "epoch": 7.61035680454689,
      "grad_norm": 5.481725215911865,
      "learning_rate": 3.4415675542925305e-05,
      "loss": 0.4546,
      "step": 24102
    },
    {
      "epoch": 7.610672560783075,
      "grad_norm": 4.506287097930908,
      "learning_rate": 3.441479844226924e-05,
      "loss": 0.4617,
      "step": 24103
    },
    {
      "epoch": 7.610988317019261,
      "grad_norm": 5.473780155181885,
      "learning_rate": 3.441392134161316e-05,
      "loss": 0.4703,
      "step": 24104
    },
    {
      "epoch": 7.611304073255447,
      "grad_norm": 5.984524726867676,
      "learning_rate": 3.441304424095709e-05,
      "loss": 0.5256,
      "step": 24105
    },
    {
      "epoch": 7.611619829491632,
      "grad_norm": 6.432440280914307,
      "learning_rate": 3.441216714030102e-05,
      "loss": 0.5082,
      "step": 24106
    },
    {
      "epoch": 7.611935585727818,
      "grad_norm": 6.1906023025512695,
      "learning_rate": 3.441129003964495e-05,
      "loss": 0.5332,
      "step": 24107
    },
    {
      "epoch": 7.612251341964003,
      "grad_norm": 5.395929336547852,
      "learning_rate": 3.441041293898888e-05,
      "loss": 0.4661,
      "step": 24108
    },
    {
      "epoch": 7.61256709820019,
      "grad_norm": 6.209550380706787,
      "learning_rate": 3.4409535838332806e-05,
      "loss": 0.5313,
      "step": 24109
    },
    {
      "epoch": 7.612882854436375,
      "grad_norm": 4.679808139801025,
      "learning_rate": 3.440865873767674e-05,
      "loss": 0.4084,
      "step": 24110
    },
    {
      "epoch": 7.613198610672561,
      "grad_norm": 4.651838302612305,
      "learning_rate": 3.440778163702066e-05,
      "loss": 0.4628,
      "step": 24111
    },
    {
      "epoch": 7.613514366908746,
      "grad_norm": 6.679104804992676,
      "learning_rate": 3.4406904536364595e-05,
      "loss": 0.4732,
      "step": 24112
    },
    {
      "epoch": 7.613830123144933,
      "grad_norm": 5.079807758331299,
      "learning_rate": 3.440602743570852e-05,
      "loss": 0.4814,
      "step": 24113
    },
    {
      "epoch": 7.614145879381118,
      "grad_norm": 5.623930931091309,
      "learning_rate": 3.440515033505245e-05,
      "loss": 0.5604,
      "step": 24114
    },
    {
      "epoch": 7.614461635617303,
      "grad_norm": 4.209829807281494,
      "learning_rate": 3.440427323439638e-05,
      "loss": 0.4263,
      "step": 24115
    },
    {
      "epoch": 7.614777391853489,
      "grad_norm": 4.9599289894104,
      "learning_rate": 3.4403396133740315e-05,
      "loss": 0.5196,
      "step": 24116
    },
    {
      "epoch": 7.615093148089675,
      "grad_norm": 5.5529327392578125,
      "learning_rate": 3.440251903308424e-05,
      "loss": 0.5767,
      "step": 24117
    },
    {
      "epoch": 7.615408904325861,
      "grad_norm": 8.43179988861084,
      "learning_rate": 3.440164193242817e-05,
      "loss": 0.4908,
      "step": 24118
    },
    {
      "epoch": 7.615724660562046,
      "grad_norm": 6.338954448699951,
      "learning_rate": 3.4400764831772096e-05,
      "loss": 0.4911,
      "step": 24119
    },
    {
      "epoch": 7.616040416798231,
      "grad_norm": 5.6171722412109375,
      "learning_rate": 3.439988773111603e-05,
      "loss": 0.5951,
      "step": 24120
    },
    {
      "epoch": 7.616356173034418,
      "grad_norm": 5.489980220794678,
      "learning_rate": 3.439901063045995e-05,
      "loss": 0.4985,
      "step": 24121
    },
    {
      "epoch": 7.616671929270603,
      "grad_norm": 8.208892822265625,
      "learning_rate": 3.4398133529803884e-05,
      "loss": 0.5529,
      "step": 24122
    },
    {
      "epoch": 7.616987685506789,
      "grad_norm": 6.477116584777832,
      "learning_rate": 3.439725642914781e-05,
      "loss": 0.4671,
      "step": 24123
    },
    {
      "epoch": 7.617303441742974,
      "grad_norm": 4.33837890625,
      "learning_rate": 3.439637932849174e-05,
      "loss": 0.5119,
      "step": 24124
    },
    {
      "epoch": 7.61761919797916,
      "grad_norm": 5.524838447570801,
      "learning_rate": 3.4395502227835666e-05,
      "loss": 0.506,
      "step": 24125
    },
    {
      "epoch": 7.617934954215346,
      "grad_norm": 5.372618198394775,
      "learning_rate": 3.43946251271796e-05,
      "loss": 0.4858,
      "step": 24126
    },
    {
      "epoch": 7.618250710451531,
      "grad_norm": 4.747220993041992,
      "learning_rate": 3.439374802652352e-05,
      "loss": 0.4847,
      "step": 24127
    },
    {
      "epoch": 7.618566466687717,
      "grad_norm": 6.231636047363281,
      "learning_rate": 3.4392870925867454e-05,
      "loss": 0.5498,
      "step": 24128
    },
    {
      "epoch": 7.618882222923903,
      "grad_norm": 5.669899940490723,
      "learning_rate": 3.4391993825211386e-05,
      "loss": 0.5182,
      "step": 24129
    },
    {
      "epoch": 7.619197979160089,
      "grad_norm": 6.194028377532959,
      "learning_rate": 3.439111672455531e-05,
      "loss": 0.4793,
      "step": 24130
    },
    {
      "epoch": 7.619513735396274,
      "grad_norm": 5.574909210205078,
      "learning_rate": 3.439023962389924e-05,
      "loss": 0.5043,
      "step": 24131
    },
    {
      "epoch": 7.619829491632459,
      "grad_norm": 5.408613681793213,
      "learning_rate": 3.438936252324317e-05,
      "loss": 0.441,
      "step": 24132
    },
    {
      "epoch": 7.6201452478686456,
      "grad_norm": 5.390230655670166,
      "learning_rate": 3.43884854225871e-05,
      "loss": 0.4134,
      "step": 24133
    },
    {
      "epoch": 7.620461004104831,
      "grad_norm": 5.910552978515625,
      "learning_rate": 3.4387608321931024e-05,
      "loss": 0.6006,
      "step": 24134
    },
    {
      "epoch": 7.620776760341017,
      "grad_norm": 5.932424545288086,
      "learning_rate": 3.4386731221274956e-05,
      "loss": 0.4678,
      "step": 24135
    },
    {
      "epoch": 7.621092516577202,
      "grad_norm": 5.084651947021484,
      "learning_rate": 3.438585412061888e-05,
      "loss": 0.4169,
      "step": 24136
    },
    {
      "epoch": 7.6214082728133885,
      "grad_norm": 5.655557155609131,
      "learning_rate": 3.438497701996281e-05,
      "loss": 0.3792,
      "step": 24137
    },
    {
      "epoch": 7.621724029049574,
      "grad_norm": 4.725211143493652,
      "learning_rate": 3.438409991930674e-05,
      "loss": 0.3967,
      "step": 24138
    },
    {
      "epoch": 7.622039785285759,
      "grad_norm": 4.768216609954834,
      "learning_rate": 3.438322281865067e-05,
      "loss": 0.5017,
      "step": 24139
    },
    {
      "epoch": 7.622355541521945,
      "grad_norm": 4.970634460449219,
      "learning_rate": 3.4382345717994594e-05,
      "loss": 0.4698,
      "step": 24140
    },
    {
      "epoch": 7.622671297758131,
      "grad_norm": 3.8050925731658936,
      "learning_rate": 3.4381468617338525e-05,
      "loss": 0.3637,
      "step": 24141
    },
    {
      "epoch": 7.622987053994317,
      "grad_norm": 7.231938362121582,
      "learning_rate": 3.438059151668246e-05,
      "loss": 0.5443,
      "step": 24142
    },
    {
      "epoch": 7.623302810230502,
      "grad_norm": 5.613000392913818,
      "learning_rate": 3.437971441602639e-05,
      "loss": 0.5531,
      "step": 24143
    },
    {
      "epoch": 7.623618566466687,
      "grad_norm": 6.028076171875,
      "learning_rate": 3.4378837315370314e-05,
      "loss": 0.5684,
      "step": 24144
    },
    {
      "epoch": 7.6239343227028735,
      "grad_norm": 3.9762816429138184,
      "learning_rate": 3.4377960214714245e-05,
      "loss": 0.3578,
      "step": 24145
    },
    {
      "epoch": 7.624250078939059,
      "grad_norm": 5.566035270690918,
      "learning_rate": 3.437708311405817e-05,
      "loss": 0.549,
      "step": 24146
    },
    {
      "epoch": 7.624565835175245,
      "grad_norm": 5.3228960037231445,
      "learning_rate": 3.43762060134021e-05,
      "loss": 0.4932,
      "step": 24147
    },
    {
      "epoch": 7.62488159141143,
      "grad_norm": 5.0347771644592285,
      "learning_rate": 3.437532891274603e-05,
      "loss": 0.4884,
      "step": 24148
    },
    {
      "epoch": 7.625197347647616,
      "grad_norm": 7.7511305809021,
      "learning_rate": 3.437445181208996e-05,
      "loss": 0.4693,
      "step": 24149
    },
    {
      "epoch": 7.625513103883802,
      "grad_norm": 4.616234302520752,
      "learning_rate": 3.437357471143389e-05,
      "loss": 0.4693,
      "step": 24150
    },
    {
      "epoch": 7.625828860119987,
      "grad_norm": 5.929224967956543,
      "learning_rate": 3.4372697610777815e-05,
      "loss": 0.4863,
      "step": 24151
    },
    {
      "epoch": 7.626144616356173,
      "grad_norm": 5.266802787780762,
      "learning_rate": 3.437182051012175e-05,
      "loss": 0.584,
      "step": 24152
    },
    {
      "epoch": 7.6264603725923585,
      "grad_norm": 5.351457118988037,
      "learning_rate": 3.437094340946567e-05,
      "loss": 0.4693,
      "step": 24153
    },
    {
      "epoch": 7.626776128828545,
      "grad_norm": 5.4480791091918945,
      "learning_rate": 3.43700663088096e-05,
      "loss": 0.4939,
      "step": 24154
    },
    {
      "epoch": 7.62709188506473,
      "grad_norm": 5.647205352783203,
      "learning_rate": 3.436918920815353e-05,
      "loss": 0.5035,
      "step": 24155
    },
    {
      "epoch": 7.627407641300915,
      "grad_norm": 5.241983890533447,
      "learning_rate": 3.436831210749746e-05,
      "loss": 0.4775,
      "step": 24156
    },
    {
      "epoch": 7.6277233975371015,
      "grad_norm": 5.559214115142822,
      "learning_rate": 3.4367435006841385e-05,
      "loss": 0.4136,
      "step": 24157
    },
    {
      "epoch": 7.628039153773287,
      "grad_norm": 4.45959997177124,
      "learning_rate": 3.4366557906185317e-05,
      "loss": 0.4969,
      "step": 24158
    },
    {
      "epoch": 7.628354910009473,
      "grad_norm": 4.9543352127075195,
      "learning_rate": 3.436568080552924e-05,
      "loss": 0.5186,
      "step": 24159
    },
    {
      "epoch": 7.628670666245658,
      "grad_norm": 6.559137344360352,
      "learning_rate": 3.436480370487317e-05,
      "loss": 0.4792,
      "step": 24160
    },
    {
      "epoch": 7.628986422481844,
      "grad_norm": 5.204360008239746,
      "learning_rate": 3.43639266042171e-05,
      "loss": 0.4439,
      "step": 24161
    },
    {
      "epoch": 7.62930217871803,
      "grad_norm": 5.475893020629883,
      "learning_rate": 3.436304950356103e-05,
      "loss": 0.4936,
      "step": 24162
    },
    {
      "epoch": 7.629617934954215,
      "grad_norm": 4.824962615966797,
      "learning_rate": 3.4362172402904955e-05,
      "loss": 0.5028,
      "step": 24163
    },
    {
      "epoch": 7.629933691190401,
      "grad_norm": 4.792305946350098,
      "learning_rate": 3.4361295302248886e-05,
      "loss": 0.443,
      "step": 24164
    },
    {
      "epoch": 7.6302494474265865,
      "grad_norm": 5.70532751083374,
      "learning_rate": 3.436041820159281e-05,
      "loss": 0.456,
      "step": 24165
    },
    {
      "epoch": 7.630565203662773,
      "grad_norm": 6.148447036743164,
      "learning_rate": 3.435954110093674e-05,
      "loss": 0.4726,
      "step": 24166
    },
    {
      "epoch": 7.630880959898958,
      "grad_norm": 5.770519733428955,
      "learning_rate": 3.4358664000280675e-05,
      "loss": 0.5443,
      "step": 24167
    },
    {
      "epoch": 7.631196716135143,
      "grad_norm": 5.2280802726745605,
      "learning_rate": 3.43577868996246e-05,
      "loss": 0.4796,
      "step": 24168
    },
    {
      "epoch": 7.6315124723713295,
      "grad_norm": 6.397102355957031,
      "learning_rate": 3.435690979896853e-05,
      "loss": 0.5499,
      "step": 24169
    },
    {
      "epoch": 7.631828228607515,
      "grad_norm": 5.106385231018066,
      "learning_rate": 3.435603269831246e-05,
      "loss": 0.5056,
      "step": 24170
    },
    {
      "epoch": 7.632143984843701,
      "grad_norm": 6.510078430175781,
      "learning_rate": 3.435515559765639e-05,
      "loss": 0.5766,
      "step": 24171
    },
    {
      "epoch": 7.632459741079886,
      "grad_norm": 5.269069671630859,
      "learning_rate": 3.435427849700032e-05,
      "loss": 0.5144,
      "step": 24172
    },
    {
      "epoch": 7.6327754973160715,
      "grad_norm": 5.480001449584961,
      "learning_rate": 3.435340139634425e-05,
      "loss": 0.5063,
      "step": 24173
    },
    {
      "epoch": 7.633091253552258,
      "grad_norm": 5.7997894287109375,
      "learning_rate": 3.4352524295688176e-05,
      "loss": 0.4604,
      "step": 24174
    },
    {
      "epoch": 7.633407009788443,
      "grad_norm": 6.892055034637451,
      "learning_rate": 3.435164719503211e-05,
      "loss": 0.597,
      "step": 24175
    },
    {
      "epoch": 7.633722766024629,
      "grad_norm": 4.20391321182251,
      "learning_rate": 3.435077009437603e-05,
      "loss": 0.3856,
      "step": 24176
    },
    {
      "epoch": 7.6340385222608145,
      "grad_norm": 5.052927017211914,
      "learning_rate": 3.4349892993719964e-05,
      "loss": 0.4889,
      "step": 24177
    },
    {
      "epoch": 7.634354278497001,
      "grad_norm": 4.624734878540039,
      "learning_rate": 3.434901589306389e-05,
      "loss": 0.5177,
      "step": 24178
    },
    {
      "epoch": 7.634670034733186,
      "grad_norm": 5.608930587768555,
      "learning_rate": 3.434813879240782e-05,
      "loss": 0.4983,
      "step": 24179
    },
    {
      "epoch": 7.634985790969372,
      "grad_norm": 6.54060697555542,
      "learning_rate": 3.4347261691751746e-05,
      "loss": 0.4826,
      "step": 24180
    },
    {
      "epoch": 7.635301547205557,
      "grad_norm": 6.3136444091796875,
      "learning_rate": 3.434638459109568e-05,
      "loss": 0.5353,
      "step": 24181
    },
    {
      "epoch": 7.635617303441743,
      "grad_norm": 5.732267379760742,
      "learning_rate": 3.43455074904396e-05,
      "loss": 0.5543,
      "step": 24182
    },
    {
      "epoch": 7.635933059677929,
      "grad_norm": 5.561709403991699,
      "learning_rate": 3.4344630389783534e-05,
      "loss": 0.4049,
      "step": 24183
    },
    {
      "epoch": 7.636248815914114,
      "grad_norm": 5.347296714782715,
      "learning_rate": 3.434375328912746e-05,
      "loss": 0.5362,
      "step": 24184
    },
    {
      "epoch": 7.6365645721503,
      "grad_norm": 7.483136177062988,
      "learning_rate": 3.434287618847139e-05,
      "loss": 0.5452,
      "step": 24185
    },
    {
      "epoch": 7.636880328386486,
      "grad_norm": 5.314676761627197,
      "learning_rate": 3.4341999087815316e-05,
      "loss": 0.4884,
      "step": 24186
    },
    {
      "epoch": 7.637196084622671,
      "grad_norm": 5.689643859863281,
      "learning_rate": 3.434112198715925e-05,
      "loss": 0.4184,
      "step": 24187
    },
    {
      "epoch": 7.637511840858857,
      "grad_norm": 5.019592761993408,
      "learning_rate": 3.434024488650318e-05,
      "loss": 0.5303,
      "step": 24188
    },
    {
      "epoch": 7.637827597095042,
      "grad_norm": 7.37350606918335,
      "learning_rate": 3.4339367785847104e-05,
      "loss": 0.6631,
      "step": 24189
    },
    {
      "epoch": 7.638143353331229,
      "grad_norm": 6.079105377197266,
      "learning_rate": 3.4338490685191036e-05,
      "loss": 0.5655,
      "step": 24190
    },
    {
      "epoch": 7.638459109567414,
      "grad_norm": 5.5957159996032715,
      "learning_rate": 3.433761358453496e-05,
      "loss": 0.4864,
      "step": 24191
    },
    {
      "epoch": 7.638774865803599,
      "grad_norm": 5.455521583557129,
      "learning_rate": 3.433673648387889e-05,
      "loss": 0.4714,
      "step": 24192
    },
    {
      "epoch": 7.639090622039785,
      "grad_norm": 5.0861687660217285,
      "learning_rate": 3.433585938322282e-05,
      "loss": 0.4711,
      "step": 24193
    },
    {
      "epoch": 7.639406378275971,
      "grad_norm": 7.365078926086426,
      "learning_rate": 3.433498228256675e-05,
      "loss": 0.4897,
      "step": 24194
    },
    {
      "epoch": 7.639722134512157,
      "grad_norm": 4.303738594055176,
      "learning_rate": 3.4334105181910674e-05,
      "loss": 0.4213,
      "step": 24195
    },
    {
      "epoch": 7.640037890748342,
      "grad_norm": 4.8698506355285645,
      "learning_rate": 3.4333228081254605e-05,
      "loss": 0.5489,
      "step": 24196
    },
    {
      "epoch": 7.6403536469845275,
      "grad_norm": 6.141714096069336,
      "learning_rate": 3.433235098059854e-05,
      "loss": 0.473,
      "step": 24197
    },
    {
      "epoch": 7.640669403220714,
      "grad_norm": 4.748095512390137,
      "learning_rate": 3.433147387994247e-05,
      "loss": 0.4632,
      "step": 24198
    },
    {
      "epoch": 7.640985159456899,
      "grad_norm": 4.167973041534424,
      "learning_rate": 3.4330596779286394e-05,
      "loss": 0.4497,
      "step": 24199
    },
    {
      "epoch": 7.641300915693085,
      "grad_norm": 5.190948009490967,
      "learning_rate": 3.4329719678630325e-05,
      "loss": 0.414,
      "step": 24200
    },
    {
      "epoch": 7.641300915693085,
      "eval_loss": 0.918535590171814,
      "eval_runtime": 4.5497,
      "eval_samples_per_second": 547.287,
      "eval_steps_per_second": 4.616,
      "step": 24200
    },
    {
      "epoch": 7.64161667192927,
      "grad_norm": 5.037040710449219,
      "learning_rate": 3.432884257797425e-05,
      "loss": 0.5162,
      "step": 24201
    },
    {
      "epoch": 7.641932428165457,
      "grad_norm": 5.576385974884033,
      "learning_rate": 3.432796547731818e-05,
      "loss": 0.4486,
      "step": 24202
    },
    {
      "epoch": 7.642248184401642,
      "grad_norm": 4.704209327697754,
      "learning_rate": 3.432708837666211e-05,
      "loss": 0.4892,
      "step": 24203
    },
    {
      "epoch": 7.642563940637828,
      "grad_norm": 7.487070560455322,
      "learning_rate": 3.432621127600604e-05,
      "loss": 0.4146,
      "step": 24204
    },
    {
      "epoch": 7.642879696874013,
      "grad_norm": 4.533637523651123,
      "learning_rate": 3.432533417534996e-05,
      "loss": 0.5052,
      "step": 24205
    },
    {
      "epoch": 7.643195453110199,
      "grad_norm": 6.311003684997559,
      "learning_rate": 3.4324457074693895e-05,
      "loss": 0.4843,
      "step": 24206
    },
    {
      "epoch": 7.643511209346385,
      "grad_norm": 6.72305154800415,
      "learning_rate": 3.432357997403782e-05,
      "loss": 0.5809,
      "step": 24207
    },
    {
      "epoch": 7.64382696558257,
      "grad_norm": 6.108185291290283,
      "learning_rate": 3.432270287338175e-05,
      "loss": 0.522,
      "step": 24208
    },
    {
      "epoch": 7.644142721818756,
      "grad_norm": 5.237762928009033,
      "learning_rate": 3.4321825772725677e-05,
      "loss": 0.4597,
      "step": 24209
    },
    {
      "epoch": 7.644458478054942,
      "grad_norm": 6.906371116638184,
      "learning_rate": 3.432094867206961e-05,
      "loss": 0.6089,
      "step": 24210
    },
    {
      "epoch": 7.644774234291127,
      "grad_norm": 6.225135326385498,
      "learning_rate": 3.432007157141354e-05,
      "loss": 0.4237,
      "step": 24211
    },
    {
      "epoch": 7.645089990527313,
      "grad_norm": 5.658829212188721,
      "learning_rate": 3.4319194470757465e-05,
      "loss": 0.4633,
      "step": 24212
    },
    {
      "epoch": 7.645405746763498,
      "grad_norm": 5.008937358856201,
      "learning_rate": 3.4318317370101396e-05,
      "loss": 0.5064,
      "step": 24213
    },
    {
      "epoch": 7.645721502999685,
      "grad_norm": 6.538280487060547,
      "learning_rate": 3.431744026944532e-05,
      "loss": 0.5196,
      "step": 24214
    },
    {
      "epoch": 7.64603725923587,
      "grad_norm": 5.133860111236572,
      "learning_rate": 3.431656316878925e-05,
      "loss": 0.4218,
      "step": 24215
    },
    {
      "epoch": 7.646353015472055,
      "grad_norm": 6.286319732666016,
      "learning_rate": 3.431568606813318e-05,
      "loss": 0.3962,
      "step": 24216
    },
    {
      "epoch": 7.646668771708241,
      "grad_norm": 7.108813762664795,
      "learning_rate": 3.431480896747711e-05,
      "loss": 0.528,
      "step": 24217
    },
    {
      "epoch": 7.646984527944427,
      "grad_norm": 7.497303009033203,
      "learning_rate": 3.4313931866821035e-05,
      "loss": 0.5225,
      "step": 24218
    },
    {
      "epoch": 7.647300284180613,
      "grad_norm": 5.5937066078186035,
      "learning_rate": 3.4313054766164966e-05,
      "loss": 0.4312,
      "step": 24219
    },
    {
      "epoch": 7.647616040416798,
      "grad_norm": 7.694912910461426,
      "learning_rate": 3.431217766550889e-05,
      "loss": 0.4869,
      "step": 24220
    },
    {
      "epoch": 7.647931796652984,
      "grad_norm": 4.114481449127197,
      "learning_rate": 3.431130056485282e-05,
      "loss": 0.3651,
      "step": 24221
    },
    {
      "epoch": 7.64824755288917,
      "grad_norm": 6.0757598876953125,
      "learning_rate": 3.431042346419675e-05,
      "loss": 0.4856,
      "step": 24222
    },
    {
      "epoch": 7.648563309125355,
      "grad_norm": 7.158519268035889,
      "learning_rate": 3.430954636354068e-05,
      "loss": 0.6129,
      "step": 24223
    },
    {
      "epoch": 7.648879065361541,
      "grad_norm": 5.031703948974609,
      "learning_rate": 3.430866926288461e-05,
      "loss": 0.4816,
      "step": 24224
    },
    {
      "epoch": 7.649194821597726,
      "grad_norm": 5.928289413452148,
      "learning_rate": 3.430779216222854e-05,
      "loss": 0.389,
      "step": 24225
    },
    {
      "epoch": 7.6495105778339125,
      "grad_norm": 7.254131317138672,
      "learning_rate": 3.430691506157247e-05,
      "loss": 0.4849,
      "step": 24226
    },
    {
      "epoch": 7.649826334070098,
      "grad_norm": 6.692073822021484,
      "learning_rate": 3.43060379609164e-05,
      "loss": 0.5516,
      "step": 24227
    },
    {
      "epoch": 7.650142090306284,
      "grad_norm": 5.915390968322754,
      "learning_rate": 3.4305160860260324e-05,
      "loss": 0.4358,
      "step": 24228
    },
    {
      "epoch": 7.650457846542469,
      "grad_norm": 4.521118640899658,
      "learning_rate": 3.4304283759604256e-05,
      "loss": 0.3159,
      "step": 24229
    },
    {
      "epoch": 7.650773602778655,
      "grad_norm": 6.032784938812256,
      "learning_rate": 3.430340665894818e-05,
      "loss": 0.5103,
      "step": 24230
    },
    {
      "epoch": 7.651089359014841,
      "grad_norm": 6.3423566818237305,
      "learning_rate": 3.430252955829211e-05,
      "loss": 0.5514,
      "step": 24231
    },
    {
      "epoch": 7.651405115251026,
      "grad_norm": 5.30274772644043,
      "learning_rate": 3.4301652457636044e-05,
      "loss": 0.479,
      "step": 24232
    },
    {
      "epoch": 7.651720871487212,
      "grad_norm": 5.960014820098877,
      "learning_rate": 3.430077535697997e-05,
      "loss": 0.4349,
      "step": 24233
    },
    {
      "epoch": 7.6520366277233975,
      "grad_norm": 6.295951843261719,
      "learning_rate": 3.42998982563239e-05,
      "loss": 0.404,
      "step": 24234
    },
    {
      "epoch": 7.652352383959583,
      "grad_norm": 6.825462341308594,
      "learning_rate": 3.4299021155667826e-05,
      "loss": 0.4329,
      "step": 24235
    },
    {
      "epoch": 7.652668140195769,
      "grad_norm": 5.768239498138428,
      "learning_rate": 3.429814405501176e-05,
      "loss": 0.4904,
      "step": 24236
    },
    {
      "epoch": 7.652983896431954,
      "grad_norm": 7.035669326782227,
      "learning_rate": 3.429726695435568e-05,
      "loss": 0.4746,
      "step": 24237
    },
    {
      "epoch": 7.6532996526681405,
      "grad_norm": 6.761988639831543,
      "learning_rate": 3.4296389853699614e-05,
      "loss": 0.4266,
      "step": 24238
    },
    {
      "epoch": 7.653615408904326,
      "grad_norm": 6.260768890380859,
      "learning_rate": 3.429551275304354e-05,
      "loss": 0.4361,
      "step": 24239
    },
    {
      "epoch": 7.653931165140511,
      "grad_norm": 5.165891647338867,
      "learning_rate": 3.429463565238747e-05,
      "loss": 0.5215,
      "step": 24240
    },
    {
      "epoch": 7.654246921376697,
      "grad_norm": 5.135268211364746,
      "learning_rate": 3.4293758551731395e-05,
      "loss": 0.4463,
      "step": 24241
    },
    {
      "epoch": 7.654562677612883,
      "grad_norm": 5.271297931671143,
      "learning_rate": 3.429288145107533e-05,
      "loss": 0.4567,
      "step": 24242
    },
    {
      "epoch": 7.654878433849069,
      "grad_norm": 7.8354949951171875,
      "learning_rate": 3.429200435041925e-05,
      "loss": 0.5024,
      "step": 24243
    },
    {
      "epoch": 7.655194190085254,
      "grad_norm": 6.305018424987793,
      "learning_rate": 3.4291127249763184e-05,
      "loss": 0.5231,
      "step": 24244
    },
    {
      "epoch": 7.65550994632144,
      "grad_norm": 7.732828140258789,
      "learning_rate": 3.429025014910711e-05,
      "loss": 0.5975,
      "step": 24245
    },
    {
      "epoch": 7.6558257025576255,
      "grad_norm": 5.670164585113525,
      "learning_rate": 3.428937304845104e-05,
      "loss": 0.4426,
      "step": 24246
    },
    {
      "epoch": 7.656141458793811,
      "grad_norm": 7.820401191711426,
      "learning_rate": 3.4288495947794965e-05,
      "loss": 0.5917,
      "step": 24247
    },
    {
      "epoch": 7.656457215029997,
      "grad_norm": 6.204827785491943,
      "learning_rate": 3.42876188471389e-05,
      "loss": 0.4535,
      "step": 24248
    },
    {
      "epoch": 7.656772971266182,
      "grad_norm": 5.7760186195373535,
      "learning_rate": 3.428674174648283e-05,
      "loss": 0.5392,
      "step": 24249
    },
    {
      "epoch": 7.6570887275023685,
      "grad_norm": 5.43991231918335,
      "learning_rate": 3.4285864645826754e-05,
      "loss": 0.4679,
      "step": 24250
    },
    {
      "epoch": 7.657404483738554,
      "grad_norm": 6.575455665588379,
      "learning_rate": 3.4284987545170685e-05,
      "loss": 0.54,
      "step": 24251
    },
    {
      "epoch": 7.65772023997474,
      "grad_norm": 5.2912468910217285,
      "learning_rate": 3.428411044451462e-05,
      "loss": 0.5331,
      "step": 24252
    },
    {
      "epoch": 7.658035996210925,
      "grad_norm": 6.2274489402771,
      "learning_rate": 3.428323334385854e-05,
      "loss": 0.598,
      "step": 24253
    },
    {
      "epoch": 7.6583517524471105,
      "grad_norm": 6.67497444152832,
      "learning_rate": 3.4282356243202473e-05,
      "loss": 0.6257,
      "step": 24254
    },
    {
      "epoch": 7.658667508683297,
      "grad_norm": 4.488787651062012,
      "learning_rate": 3.4281479142546405e-05,
      "loss": 0.5567,
      "step": 24255
    },
    {
      "epoch": 7.658983264919482,
      "grad_norm": 5.647314548492432,
      "learning_rate": 3.428060204189033e-05,
      "loss": 0.4517,
      "step": 24256
    },
    {
      "epoch": 7.659299021155668,
      "grad_norm": 6.181623458862305,
      "learning_rate": 3.427972494123426e-05,
      "loss": 0.5257,
      "step": 24257
    },
    {
      "epoch": 7.6596147773918535,
      "grad_norm": 4.711131572723389,
      "learning_rate": 3.427884784057819e-05,
      "loss": 0.4341,
      "step": 24258
    },
    {
      "epoch": 7.659930533628039,
      "grad_norm": 5.748852252960205,
      "learning_rate": 3.427797073992212e-05,
      "loss": 0.4496,
      "step": 24259
    },
    {
      "epoch": 7.660246289864225,
      "grad_norm": 6.712473392486572,
      "learning_rate": 3.427709363926604e-05,
      "loss": 0.463,
      "step": 24260
    },
    {
      "epoch": 7.66056204610041,
      "grad_norm": 4.93590784072876,
      "learning_rate": 3.4276216538609975e-05,
      "loss": 0.513,
      "step": 24261
    },
    {
      "epoch": 7.660877802336596,
      "grad_norm": 4.842931747436523,
      "learning_rate": 3.42753394379539e-05,
      "loss": 0.3918,
      "step": 24262
    },
    {
      "epoch": 7.661193558572782,
      "grad_norm": 4.286365985870361,
      "learning_rate": 3.427446233729783e-05,
      "loss": 0.4105,
      "step": 24263
    },
    {
      "epoch": 7.661509314808967,
      "grad_norm": 5.507359027862549,
      "learning_rate": 3.4273585236641756e-05,
      "loss": 0.4753,
      "step": 24264
    },
    {
      "epoch": 7.661825071045153,
      "grad_norm": 5.469542980194092,
      "learning_rate": 3.427270813598569e-05,
      "loss": 0.4162,
      "step": 24265
    },
    {
      "epoch": 7.6621408272813385,
      "grad_norm": 4.649666786193848,
      "learning_rate": 3.427183103532961e-05,
      "loss": 0.3737,
      "step": 24266
    },
    {
      "epoch": 7.662456583517525,
      "grad_norm": 7.512542247772217,
      "learning_rate": 3.4270953934673545e-05,
      "loss": 0.6589,
      "step": 24267
    },
    {
      "epoch": 7.66277233975371,
      "grad_norm": 5.412355422973633,
      "learning_rate": 3.427007683401747e-05,
      "loss": 0.5272,
      "step": 24268
    },
    {
      "epoch": 7.663088095989896,
      "grad_norm": 5.652482509613037,
      "learning_rate": 3.42691997333614e-05,
      "loss": 0.5035,
      "step": 24269
    },
    {
      "epoch": 7.6634038522260814,
      "grad_norm": 5.144406318664551,
      "learning_rate": 3.426832263270533e-05,
      "loss": 0.414,
      "step": 24270
    },
    {
      "epoch": 7.663719608462267,
      "grad_norm": 4.952641487121582,
      "learning_rate": 3.426744553204926e-05,
      "loss": 0.4241,
      "step": 24271
    },
    {
      "epoch": 7.664035364698453,
      "grad_norm": 6.250527858734131,
      "learning_rate": 3.426656843139319e-05,
      "loss": 0.5568,
      "step": 24272
    },
    {
      "epoch": 7.664351120934638,
      "grad_norm": 8.87890625,
      "learning_rate": 3.4265691330737114e-05,
      "loss": 0.6322,
      "step": 24273
    },
    {
      "epoch": 7.664666877170824,
      "grad_norm": 8.094497680664062,
      "learning_rate": 3.4264814230081046e-05,
      "loss": 0.5398,
      "step": 24274
    },
    {
      "epoch": 7.66498263340701,
      "grad_norm": 5.204880714416504,
      "learning_rate": 3.426393712942497e-05,
      "loss": 0.4777,
      "step": 24275
    },
    {
      "epoch": 7.665298389643196,
      "grad_norm": 7.08947229385376,
      "learning_rate": 3.42630600287689e-05,
      "loss": 0.5255,
      "step": 24276
    },
    {
      "epoch": 7.665614145879381,
      "grad_norm": 4.728302955627441,
      "learning_rate": 3.426218292811283e-05,
      "loss": 0.4179,
      "step": 24277
    },
    {
      "epoch": 7.6659299021155665,
      "grad_norm": 6.20871639251709,
      "learning_rate": 3.4261305827456766e-05,
      "loss": 0.5089,
      "step": 24278
    },
    {
      "epoch": 7.666245658351753,
      "grad_norm": 5.236501693725586,
      "learning_rate": 3.426042872680069e-05,
      "loss": 0.5493,
      "step": 24279
    },
    {
      "epoch": 7.666561414587938,
      "grad_norm": 6.006330490112305,
      "learning_rate": 3.425955162614462e-05,
      "loss": 0.4924,
      "step": 24280
    },
    {
      "epoch": 7.666877170824124,
      "grad_norm": 7.084575176239014,
      "learning_rate": 3.425867452548855e-05,
      "loss": 0.5471,
      "step": 24281
    },
    {
      "epoch": 7.667192927060309,
      "grad_norm": 4.356810569763184,
      "learning_rate": 3.425779742483248e-05,
      "loss": 0.4463,
      "step": 24282
    },
    {
      "epoch": 7.667508683296495,
      "grad_norm": 5.1671648025512695,
      "learning_rate": 3.4256920324176404e-05,
      "loss": 0.5575,
      "step": 24283
    },
    {
      "epoch": 7.667824439532681,
      "grad_norm": 5.642789840698242,
      "learning_rate": 3.4256043223520336e-05,
      "loss": 0.5902,
      "step": 24284
    },
    {
      "epoch": 7.668140195768866,
      "grad_norm": 5.965258598327637,
      "learning_rate": 3.425516612286426e-05,
      "loss": 0.4743,
      "step": 24285
    },
    {
      "epoch": 7.668455952005052,
      "grad_norm": 5.170958518981934,
      "learning_rate": 3.425428902220819e-05,
      "loss": 0.4903,
      "step": 24286
    },
    {
      "epoch": 7.668771708241238,
      "grad_norm": 5.930522918701172,
      "learning_rate": 3.425341192155212e-05,
      "loss": 0.637,
      "step": 24287
    },
    {
      "epoch": 7.669087464477423,
      "grad_norm": 5.633864879608154,
      "learning_rate": 3.425253482089605e-05,
      "loss": 0.5923,
      "step": 24288
    },
    {
      "epoch": 7.669403220713609,
      "grad_norm": 5.752825736999512,
      "learning_rate": 3.4251657720239974e-05,
      "loss": 0.4446,
      "step": 24289
    },
    {
      "epoch": 7.669718976949794,
      "grad_norm": 4.699370861053467,
      "learning_rate": 3.4250780619583906e-05,
      "loss": 0.5188,
      "step": 24290
    },
    {
      "epoch": 7.670034733185981,
      "grad_norm": 5.590835094451904,
      "learning_rate": 3.424990351892783e-05,
      "loss": 0.5263,
      "step": 24291
    },
    {
      "epoch": 7.670350489422166,
      "grad_norm": 6.102434158325195,
      "learning_rate": 3.424902641827176e-05,
      "loss": 0.6096,
      "step": 24292
    },
    {
      "epoch": 7.670666245658352,
      "grad_norm": 5.659481048583984,
      "learning_rate": 3.4248149317615694e-05,
      "loss": 0.5097,
      "step": 24293
    },
    {
      "epoch": 7.670982001894537,
      "grad_norm": 6.12909460067749,
      "learning_rate": 3.424727221695962e-05,
      "loss": 0.5488,
      "step": 24294
    },
    {
      "epoch": 7.671297758130724,
      "grad_norm": 4.862268447875977,
      "learning_rate": 3.424639511630355e-05,
      "loss": 0.3885,
      "step": 24295
    },
    {
      "epoch": 7.671613514366909,
      "grad_norm": 3.2866783142089844,
      "learning_rate": 3.4245518015647475e-05,
      "loss": 0.3372,
      "step": 24296
    },
    {
      "epoch": 7.671929270603094,
      "grad_norm": 5.574714183807373,
      "learning_rate": 3.424464091499141e-05,
      "loss": 0.5051,
      "step": 24297
    },
    {
      "epoch": 7.67224502683928,
      "grad_norm": 4.562542915344238,
      "learning_rate": 3.424376381433533e-05,
      "loss": 0.5014,
      "step": 24298
    },
    {
      "epoch": 7.672560783075466,
      "grad_norm": 4.740236759185791,
      "learning_rate": 3.4242886713679264e-05,
      "loss": 0.459,
      "step": 24299
    },
    {
      "epoch": 7.672876539311652,
      "grad_norm": 5.7317280769348145,
      "learning_rate": 3.424200961302319e-05,
      "loss": 0.4554,
      "step": 24300
    },
    {
      "epoch": 7.673192295547837,
      "grad_norm": 7.160970211029053,
      "learning_rate": 3.424113251236712e-05,
      "loss": 0.5578,
      "step": 24301
    },
    {
      "epoch": 7.673508051784022,
      "grad_norm": 6.0421271324157715,
      "learning_rate": 3.4240255411711045e-05,
      "loss": 0.4616,
      "step": 24302
    },
    {
      "epoch": 7.673823808020209,
      "grad_norm": 5.844336986541748,
      "learning_rate": 3.423937831105498e-05,
      "loss": 0.5292,
      "step": 24303
    },
    {
      "epoch": 7.674139564256394,
      "grad_norm": 5.051977634429932,
      "learning_rate": 3.42385012103989e-05,
      "loss": 0.4725,
      "step": 24304
    },
    {
      "epoch": 7.67445532049258,
      "grad_norm": 4.693570613861084,
      "learning_rate": 3.4237624109742833e-05,
      "loss": 0.4493,
      "step": 24305
    },
    {
      "epoch": 7.674771076728765,
      "grad_norm": 5.563476085662842,
      "learning_rate": 3.4236747009086765e-05,
      "loss": 0.4578,
      "step": 24306
    },
    {
      "epoch": 7.675086832964951,
      "grad_norm": 4.444282054901123,
      "learning_rate": 3.42358699084307e-05,
      "loss": 0.4339,
      "step": 24307
    },
    {
      "epoch": 7.675402589201137,
      "grad_norm": 5.206998348236084,
      "learning_rate": 3.423499280777462e-05,
      "loss": 0.5579,
      "step": 24308
    },
    {
      "epoch": 7.675718345437322,
      "grad_norm": 7.479743957519531,
      "learning_rate": 3.423411570711855e-05,
      "loss": 0.4874,
      "step": 24309
    },
    {
      "epoch": 7.676034101673508,
      "grad_norm": 5.3731865882873535,
      "learning_rate": 3.423323860646248e-05,
      "loss": 0.5287,
      "step": 24310
    },
    {
      "epoch": 7.676349857909694,
      "grad_norm": 5.22590446472168,
      "learning_rate": 3.423236150580641e-05,
      "loss": 0.3958,
      "step": 24311
    },
    {
      "epoch": 7.676665614145879,
      "grad_norm": 7.211216449737549,
      "learning_rate": 3.4231484405150335e-05,
      "loss": 0.5461,
      "step": 24312
    },
    {
      "epoch": 7.676981370382065,
      "grad_norm": 5.894041061401367,
      "learning_rate": 3.4230607304494267e-05,
      "loss": 0.4722,
      "step": 24313
    },
    {
      "epoch": 7.67729712661825,
      "grad_norm": 5.784677505493164,
      "learning_rate": 3.42297302038382e-05,
      "loss": 0.4057,
      "step": 24314
    },
    {
      "epoch": 7.6776128828544365,
      "grad_norm": 6.906167507171631,
      "learning_rate": 3.422885310318212e-05,
      "loss": 0.5224,
      "step": 24315
    },
    {
      "epoch": 7.677928639090622,
      "grad_norm": 5.105172634124756,
      "learning_rate": 3.4227976002526055e-05,
      "loss": 0.4726,
      "step": 24316
    },
    {
      "epoch": 7.678244395326808,
      "grad_norm": 8.185453414916992,
      "learning_rate": 3.422709890186998e-05,
      "loss": 0.4614,
      "step": 24317
    },
    {
      "epoch": 7.678560151562993,
      "grad_norm": 6.0350542068481445,
      "learning_rate": 3.422622180121391e-05,
      "loss": 0.3643,
      "step": 24318
    },
    {
      "epoch": 7.6788759077991795,
      "grad_norm": 7.520066738128662,
      "learning_rate": 3.4225344700557836e-05,
      "loss": 0.5375,
      "step": 24319
    },
    {
      "epoch": 7.679191664035365,
      "grad_norm": 5.776121139526367,
      "learning_rate": 3.422446759990177e-05,
      "loss": 0.4999,
      "step": 24320
    },
    {
      "epoch": 7.67950742027155,
      "grad_norm": 10.439702987670898,
      "learning_rate": 3.422359049924569e-05,
      "loss": 0.6084,
      "step": 24321
    },
    {
      "epoch": 7.679823176507736,
      "grad_norm": 6.617347240447998,
      "learning_rate": 3.4222713398589625e-05,
      "loss": 0.5256,
      "step": 24322
    },
    {
      "epoch": 7.680138932743922,
      "grad_norm": 5.4843950271606445,
      "learning_rate": 3.422183629793355e-05,
      "loss": 0.5337,
      "step": 24323
    },
    {
      "epoch": 7.680454688980108,
      "grad_norm": 4.862438201904297,
      "learning_rate": 3.422095919727748e-05,
      "loss": 0.4401,
      "step": 24324
    },
    {
      "epoch": 7.680770445216293,
      "grad_norm": 5.154290199279785,
      "learning_rate": 3.4220082096621406e-05,
      "loss": 0.4474,
      "step": 24325
    },
    {
      "epoch": 7.681086201452478,
      "grad_norm": 4.751481056213379,
      "learning_rate": 3.421920499596534e-05,
      "loss": 0.4228,
      "step": 24326
    },
    {
      "epoch": 7.6814019576886645,
      "grad_norm": 5.684267997741699,
      "learning_rate": 3.421832789530926e-05,
      "loss": 0.5442,
      "step": 24327
    },
    {
      "epoch": 7.68171771392485,
      "grad_norm": 5.727933883666992,
      "learning_rate": 3.4217450794653194e-05,
      "loss": 0.45,
      "step": 24328
    },
    {
      "epoch": 7.682033470161036,
      "grad_norm": 5.280784606933594,
      "learning_rate": 3.4216573693997126e-05,
      "loss": 0.4406,
      "step": 24329
    },
    {
      "epoch": 7.682349226397221,
      "grad_norm": 5.4570770263671875,
      "learning_rate": 3.421569659334105e-05,
      "loss": 0.4525,
      "step": 24330
    },
    {
      "epoch": 7.682664982633407,
      "grad_norm": 5.1986403465271,
      "learning_rate": 3.421481949268498e-05,
      "loss": 0.5184,
      "step": 24331
    },
    {
      "epoch": 7.682980738869593,
      "grad_norm": 5.577568531036377,
      "learning_rate": 3.421394239202891e-05,
      "loss": 0.5118,
      "step": 24332
    },
    {
      "epoch": 7.683296495105778,
      "grad_norm": 5.625468730926514,
      "learning_rate": 3.421306529137284e-05,
      "loss": 0.4382,
      "step": 24333
    },
    {
      "epoch": 7.683612251341964,
      "grad_norm": 4.591849327087402,
      "learning_rate": 3.421218819071677e-05,
      "loss": 0.545,
      "step": 24334
    },
    {
      "epoch": 7.6839280075781495,
      "grad_norm": 6.586806774139404,
      "learning_rate": 3.42113110900607e-05,
      "loss": 0.6258,
      "step": 24335
    },
    {
      "epoch": 7.684243763814336,
      "grad_norm": 6.056931018829346,
      "learning_rate": 3.421043398940463e-05,
      "loss": 0.5133,
      "step": 24336
    },
    {
      "epoch": 7.684559520050521,
      "grad_norm": 6.193394184112549,
      "learning_rate": 3.420955688874856e-05,
      "loss": 0.4597,
      "step": 24337
    },
    {
      "epoch": 7.684875276286706,
      "grad_norm": 5.1980414390563965,
      "learning_rate": 3.4208679788092484e-05,
      "loss": 0.5197,
      "step": 24338
    },
    {
      "epoch": 7.6851910325228925,
      "grad_norm": 5.72150182723999,
      "learning_rate": 3.4207802687436416e-05,
      "loss": 0.4413,
      "step": 24339
    },
    {
      "epoch": 7.685506788759078,
      "grad_norm": 6.481607437133789,
      "learning_rate": 3.420692558678034e-05,
      "loss": 0.5566,
      "step": 24340
    },
    {
      "epoch": 7.685822544995264,
      "grad_norm": 6.157076358795166,
      "learning_rate": 3.420604848612427e-05,
      "loss": 0.5613,
      "step": 24341
    },
    {
      "epoch": 7.686138301231449,
      "grad_norm": 5.191776752471924,
      "learning_rate": 3.42051713854682e-05,
      "loss": 0.4398,
      "step": 24342
    },
    {
      "epoch": 7.686454057467635,
      "grad_norm": 6.068154335021973,
      "learning_rate": 3.420429428481213e-05,
      "loss": 0.487,
      "step": 24343
    },
    {
      "epoch": 7.686769813703821,
      "grad_norm": 5.825100421905518,
      "learning_rate": 3.4203417184156054e-05,
      "loss": 0.4194,
      "step": 24344
    },
    {
      "epoch": 7.687085569940006,
      "grad_norm": 5.120146751403809,
      "learning_rate": 3.4202540083499985e-05,
      "loss": 0.4869,
      "step": 24345
    },
    {
      "epoch": 7.687401326176192,
      "grad_norm": 6.224878787994385,
      "learning_rate": 3.420166298284391e-05,
      "loss": 0.4633,
      "step": 24346
    },
    {
      "epoch": 7.6877170824123775,
      "grad_norm": 5.488560199737549,
      "learning_rate": 3.420078588218784e-05,
      "loss": 0.5026,
      "step": 24347
    },
    {
      "epoch": 7.688032838648564,
      "grad_norm": 4.9311203956604,
      "learning_rate": 3.419990878153177e-05,
      "loss": 0.5074,
      "step": 24348
    },
    {
      "epoch": 7.688348594884749,
      "grad_norm": 6.090887546539307,
      "learning_rate": 3.41990316808757e-05,
      "loss": 0.5951,
      "step": 24349
    },
    {
      "epoch": 7.688664351120934,
      "grad_norm": 6.436315059661865,
      "learning_rate": 3.4198154580219624e-05,
      "loss": 0.5789,
      "step": 24350
    },
    {
      "epoch": 7.6889801073571205,
      "grad_norm": 5.260501861572266,
      "learning_rate": 3.4197277479563555e-05,
      "loss": 0.5431,
      "step": 24351
    },
    {
      "epoch": 7.689295863593306,
      "grad_norm": 5.3501458168029785,
      "learning_rate": 3.419640037890749e-05,
      "loss": 0.4355,
      "step": 24352
    },
    {
      "epoch": 7.689611619829492,
      "grad_norm": 5.343894958496094,
      "learning_rate": 3.419552327825141e-05,
      "loss": 0.541,
      "step": 24353
    },
    {
      "epoch": 7.689927376065677,
      "grad_norm": 4.72637939453125,
      "learning_rate": 3.4194646177595344e-05,
      "loss": 0.4702,
      "step": 24354
    },
    {
      "epoch": 7.6902431323018625,
      "grad_norm": 6.433797359466553,
      "learning_rate": 3.419376907693927e-05,
      "loss": 0.4535,
      "step": 24355
    },
    {
      "epoch": 7.690558888538049,
      "grad_norm": 6.147392749786377,
      "learning_rate": 3.41928919762832e-05,
      "loss": 0.5205,
      "step": 24356
    },
    {
      "epoch": 7.690874644774234,
      "grad_norm": 6.260871887207031,
      "learning_rate": 3.4192014875627125e-05,
      "loss": 0.6348,
      "step": 24357
    },
    {
      "epoch": 7.69119040101042,
      "grad_norm": 5.772577285766602,
      "learning_rate": 3.419113777497106e-05,
      "loss": 0.4743,
      "step": 24358
    },
    {
      "epoch": 7.6915061572466055,
      "grad_norm": 6.765781402587891,
      "learning_rate": 3.419026067431498e-05,
      "loss": 0.5824,
      "step": 24359
    },
    {
      "epoch": 7.691821913482792,
      "grad_norm": 5.6390814781188965,
      "learning_rate": 3.418938357365892e-05,
      "loss": 0.4888,
      "step": 24360
    },
    {
      "epoch": 7.692137669718977,
      "grad_norm": 5.368392467498779,
      "learning_rate": 3.4188506473002845e-05,
      "loss": 0.6052,
      "step": 24361
    },
    {
      "epoch": 7.692453425955162,
      "grad_norm": 5.514351844787598,
      "learning_rate": 3.418762937234678e-05,
      "loss": 0.5114,
      "step": 24362
    },
    {
      "epoch": 7.692769182191348,
      "grad_norm": 5.351858615875244,
      "learning_rate": 3.41867522716907e-05,
      "loss": 0.4854,
      "step": 24363
    },
    {
      "epoch": 7.693084938427534,
      "grad_norm": 6.855733394622803,
      "learning_rate": 3.418587517103463e-05,
      "loss": 0.5595,
      "step": 24364
    },
    {
      "epoch": 7.69340069466372,
      "grad_norm": 6.179750919342041,
      "learning_rate": 3.418499807037856e-05,
      "loss": 0.468,
      "step": 24365
    },
    {
      "epoch": 7.693716450899905,
      "grad_norm": 5.861949443817139,
      "learning_rate": 3.418412096972249e-05,
      "loss": 0.5594,
      "step": 24366
    },
    {
      "epoch": 7.694032207136091,
      "grad_norm": 5.73521614074707,
      "learning_rate": 3.4183243869066415e-05,
      "loss": 0.539,
      "step": 24367
    },
    {
      "epoch": 7.694347963372277,
      "grad_norm": 6.466289043426514,
      "learning_rate": 3.4182366768410346e-05,
      "loss": 0.3695,
      "step": 24368
    },
    {
      "epoch": 7.694663719608462,
      "grad_norm": 4.676049709320068,
      "learning_rate": 3.418148966775427e-05,
      "loss": 0.4531,
      "step": 24369
    },
    {
      "epoch": 7.694979475844648,
      "grad_norm": 6.589642524719238,
      "learning_rate": 3.41806125670982e-05,
      "loss": 0.483,
      "step": 24370
    },
    {
      "epoch": 7.695295232080833,
      "grad_norm": 5.240423679351807,
      "learning_rate": 3.417973546644213e-05,
      "loss": 0.4927,
      "step": 24371
    },
    {
      "epoch": 7.69561098831702,
      "grad_norm": 5.408514976501465,
      "learning_rate": 3.417885836578606e-05,
      "loss": 0.4757,
      "step": 24372
    },
    {
      "epoch": 7.695926744553205,
      "grad_norm": 4.993497848510742,
      "learning_rate": 3.417798126512999e-05,
      "loss": 0.52,
      "step": 24373
    },
    {
      "epoch": 7.69624250078939,
      "grad_norm": 6.562189102172852,
      "learning_rate": 3.4177104164473916e-05,
      "loss": 0.5057,
      "step": 24374
    },
    {
      "epoch": 7.696558257025576,
      "grad_norm": 6.021998405456543,
      "learning_rate": 3.417622706381785e-05,
      "loss": 0.4909,
      "step": 24375
    },
    {
      "epoch": 7.696874013261762,
      "grad_norm": 5.625779628753662,
      "learning_rate": 3.417534996316177e-05,
      "loss": 0.4928,
      "step": 24376
    },
    {
      "epoch": 7.697189769497948,
      "grad_norm": 5.594884395599365,
      "learning_rate": 3.4174472862505704e-05,
      "loss": 0.457,
      "step": 24377
    },
    {
      "epoch": 7.697505525734133,
      "grad_norm": 4.914128303527832,
      "learning_rate": 3.417359576184963e-05,
      "loss": 0.4385,
      "step": 24378
    },
    {
      "epoch": 7.6978212819703185,
      "grad_norm": 6.905771255493164,
      "learning_rate": 3.417271866119356e-05,
      "loss": 0.4749,
      "step": 24379
    },
    {
      "epoch": 7.698137038206505,
      "grad_norm": 5.925443172454834,
      "learning_rate": 3.4171841560537486e-05,
      "loss": 0.5181,
      "step": 24380
    },
    {
      "epoch": 7.69845279444269,
      "grad_norm": 6.788637161254883,
      "learning_rate": 3.417096445988142e-05,
      "loss": 0.4532,
      "step": 24381
    },
    {
      "epoch": 7.698768550678876,
      "grad_norm": 7.124380588531494,
      "learning_rate": 3.417008735922534e-05,
      "loss": 0.4652,
      "step": 24382
    },
    {
      "epoch": 7.699084306915061,
      "grad_norm": 5.286342144012451,
      "learning_rate": 3.4169210258569274e-05,
      "loss": 0.4552,
      "step": 24383
    },
    {
      "epoch": 7.699400063151248,
      "grad_norm": 6.263749599456787,
      "learning_rate": 3.41683331579132e-05,
      "loss": 0.5883,
      "step": 24384
    },
    {
      "epoch": 7.699715819387433,
      "grad_norm": 6.0721235275268555,
      "learning_rate": 3.416745605725713e-05,
      "loss": 0.4717,
      "step": 24385
    },
    {
      "epoch": 7.700031575623619,
      "grad_norm": 6.3482890129089355,
      "learning_rate": 3.4166578956601056e-05,
      "loss": 0.4797,
      "step": 24386
    },
    {
      "epoch": 7.700347331859804,
      "grad_norm": 8.326505661010742,
      "learning_rate": 3.4165701855944994e-05,
      "loss": 0.5569,
      "step": 24387
    },
    {
      "epoch": 7.70066308809599,
      "grad_norm": 5.721224308013916,
      "learning_rate": 3.416482475528892e-05,
      "loss": 0.4966,
      "step": 24388
    },
    {
      "epoch": 7.700978844332176,
      "grad_norm": 10.857797622680664,
      "learning_rate": 3.416394765463285e-05,
      "loss": 0.515,
      "step": 24389
    },
    {
      "epoch": 7.701294600568361,
      "grad_norm": 6.4245195388793945,
      "learning_rate": 3.4163070553976776e-05,
      "loss": 0.4893,
      "step": 24390
    },
    {
      "epoch": 7.701610356804547,
      "grad_norm": 6.100274562835693,
      "learning_rate": 3.416219345332071e-05,
      "loss": 0.4637,
      "step": 24391
    },
    {
      "epoch": 7.701926113040733,
      "grad_norm": 6.61467981338501,
      "learning_rate": 3.416131635266463e-05,
      "loss": 0.515,
      "step": 24392
    },
    {
      "epoch": 7.702241869276918,
      "grad_norm": 9.715614318847656,
      "learning_rate": 3.4160439252008564e-05,
      "loss": 0.5117,
      "step": 24393
    },
    {
      "epoch": 7.702557625513104,
      "grad_norm": 6.8328728675842285,
      "learning_rate": 3.415956215135249e-05,
      "loss": 0.4714,
      "step": 24394
    },
    {
      "epoch": 7.702873381749289,
      "grad_norm": 4.826258659362793,
      "learning_rate": 3.415868505069642e-05,
      "loss": 0.5025,
      "step": 24395
    },
    {
      "epoch": 7.7031891379854756,
      "grad_norm": 6.403159141540527,
      "learning_rate": 3.415780795004035e-05,
      "loss": 0.5734,
      "step": 24396
    },
    {
      "epoch": 7.703504894221661,
      "grad_norm": 6.020610809326172,
      "learning_rate": 3.415693084938428e-05,
      "loss": 0.4257,
      "step": 24397
    },
    {
      "epoch": 7.703820650457846,
      "grad_norm": 5.66478157043457,
      "learning_rate": 3.415605374872821e-05,
      "loss": 0.4813,
      "step": 24398
    },
    {
      "epoch": 7.704136406694032,
      "grad_norm": 4.49939489364624,
      "learning_rate": 3.4155176648072134e-05,
      "loss": 0.4602,
      "step": 24399
    },
    {
      "epoch": 7.704452162930218,
      "grad_norm": 5.541858673095703,
      "learning_rate": 3.4154299547416065e-05,
      "loss": 0.4685,
      "step": 24400
    },
    {
      "epoch": 7.704452162930218,
      "eval_loss": 0.9414507746696472,
      "eval_runtime": 4.5429,
      "eval_samples_per_second": 548.109,
      "eval_steps_per_second": 4.623,
      "step": 24400
    },
    {
      "epoch": 7.704767919166404,
      "grad_norm": 5.022629261016846,
      "learning_rate": 3.415342244675999e-05,
      "loss": 0.4633,
      "step": 24401
    },
    {
      "epoch": 7.705083675402589,
      "grad_norm": 5.545980930328369,
      "learning_rate": 3.415254534610392e-05,
      "loss": 0.4573,
      "step": 24402
    },
    {
      "epoch": 7.705399431638774,
      "grad_norm": 4.358457565307617,
      "learning_rate": 3.415166824544785e-05,
      "loss": 0.4211,
      "step": 24403
    },
    {
      "epoch": 7.705715187874961,
      "grad_norm": 6.521570682525635,
      "learning_rate": 3.415079114479178e-05,
      "loss": 0.5169,
      "step": 24404
    },
    {
      "epoch": 7.706030944111146,
      "grad_norm": 5.7252607345581055,
      "learning_rate": 3.4149914044135703e-05,
      "loss": 0.6114,
      "step": 24405
    },
    {
      "epoch": 7.706346700347332,
      "grad_norm": 6.100417613983154,
      "learning_rate": 3.4149036943479635e-05,
      "loss": 0.4466,
      "step": 24406
    },
    {
      "epoch": 7.706662456583517,
      "grad_norm": 5.722185134887695,
      "learning_rate": 3.414815984282356e-05,
      "loss": 0.4433,
      "step": 24407
    },
    {
      "epoch": 7.7069782128197035,
      "grad_norm": 6.558138847351074,
      "learning_rate": 3.414728274216749e-05,
      "loss": 0.4622,
      "step": 24408
    },
    {
      "epoch": 7.707293969055889,
      "grad_norm": 5.380725383758545,
      "learning_rate": 3.414640564151142e-05,
      "loss": 0.5418,
      "step": 24409
    },
    {
      "epoch": 7.707609725292075,
      "grad_norm": 6.496612071990967,
      "learning_rate": 3.414552854085535e-05,
      "loss": 0.5127,
      "step": 24410
    },
    {
      "epoch": 7.70792548152826,
      "grad_norm": 4.6657867431640625,
      "learning_rate": 3.414465144019928e-05,
      "loss": 0.4275,
      "step": 24411
    },
    {
      "epoch": 7.708241237764446,
      "grad_norm": 6.780081272125244,
      "learning_rate": 3.4143774339543205e-05,
      "loss": 0.5613,
      "step": 24412
    },
    {
      "epoch": 7.708556994000632,
      "grad_norm": 5.005739688873291,
      "learning_rate": 3.4142897238887137e-05,
      "loss": 0.4747,
      "step": 24413
    },
    {
      "epoch": 7.708872750236817,
      "grad_norm": 7.711823463439941,
      "learning_rate": 3.414202013823106e-05,
      "loss": 0.4091,
      "step": 24414
    },
    {
      "epoch": 7.709188506473003,
      "grad_norm": 5.675817966461182,
      "learning_rate": 3.414114303757499e-05,
      "loss": 0.5827,
      "step": 24415
    },
    {
      "epoch": 7.7095042627091885,
      "grad_norm": 5.241769790649414,
      "learning_rate": 3.4140265936918925e-05,
      "loss": 0.4358,
      "step": 24416
    },
    {
      "epoch": 7.709820018945374,
      "grad_norm": 7.421395301818848,
      "learning_rate": 3.4139388836262857e-05,
      "loss": 0.5238,
      "step": 24417
    },
    {
      "epoch": 7.71013577518156,
      "grad_norm": 5.82530403137207,
      "learning_rate": 3.413851173560678e-05,
      "loss": 0.5609,
      "step": 24418
    },
    {
      "epoch": 7.710451531417745,
      "grad_norm": 4.89836311340332,
      "learning_rate": 3.413763463495071e-05,
      "loss": 0.4917,
      "step": 24419
    },
    {
      "epoch": 7.7107672876539315,
      "grad_norm": 5.151973724365234,
      "learning_rate": 3.413675753429464e-05,
      "loss": 0.4762,
      "step": 24420
    },
    {
      "epoch": 7.711083043890117,
      "grad_norm": 5.503655433654785,
      "learning_rate": 3.413588043363857e-05,
      "loss": 0.5617,
      "step": 24421
    },
    {
      "epoch": 7.711398800126302,
      "grad_norm": 5.879724502563477,
      "learning_rate": 3.4135003332982495e-05,
      "loss": 0.533,
      "step": 24422
    },
    {
      "epoch": 7.711714556362488,
      "grad_norm": 5.14723539352417,
      "learning_rate": 3.4134126232326426e-05,
      "loss": 0.4829,
      "step": 24423
    },
    {
      "epoch": 7.7120303125986736,
      "grad_norm": 4.165321350097656,
      "learning_rate": 3.413324913167035e-05,
      "loss": 0.4798,
      "step": 24424
    },
    {
      "epoch": 7.71234606883486,
      "grad_norm": 6.431370735168457,
      "learning_rate": 3.413237203101428e-05,
      "loss": 0.5002,
      "step": 24425
    },
    {
      "epoch": 7.712661825071045,
      "grad_norm": 6.064390182495117,
      "learning_rate": 3.413149493035821e-05,
      "loss": 0.5278,
      "step": 24426
    },
    {
      "epoch": 7.71297758130723,
      "grad_norm": 4.830724716186523,
      "learning_rate": 3.413061782970214e-05,
      "loss": 0.4906,
      "step": 24427
    },
    {
      "epoch": 7.7132933375434165,
      "grad_norm": 7.761414051055908,
      "learning_rate": 3.4129740729046064e-05,
      "loss": 0.5792,
      "step": 24428
    },
    {
      "epoch": 7.713609093779602,
      "grad_norm": 5.273749351501465,
      "learning_rate": 3.4128863628389996e-05,
      "loss": 0.5072,
      "step": 24429
    },
    {
      "epoch": 7.713924850015788,
      "grad_norm": 6.198923110961914,
      "learning_rate": 3.412798652773392e-05,
      "loss": 0.4657,
      "step": 24430
    },
    {
      "epoch": 7.714240606251973,
      "grad_norm": 4.770660877227783,
      "learning_rate": 3.412710942707785e-05,
      "loss": 0.428,
      "step": 24431
    },
    {
      "epoch": 7.7145563624881595,
      "grad_norm": 5.818572998046875,
      "learning_rate": 3.412623232642178e-05,
      "loss": 0.4939,
      "step": 24432
    },
    {
      "epoch": 7.714872118724345,
      "grad_norm": 5.645635604858398,
      "learning_rate": 3.412535522576571e-05,
      "loss": 0.4149,
      "step": 24433
    },
    {
      "epoch": 7.715187874960531,
      "grad_norm": 6.7259955406188965,
      "learning_rate": 3.412447812510964e-05,
      "loss": 0.5731,
      "step": 24434
    },
    {
      "epoch": 7.715503631196716,
      "grad_norm": 5.797380447387695,
      "learning_rate": 3.4123601024453566e-05,
      "loss": 0.494,
      "step": 24435
    },
    {
      "epoch": 7.7158193874329015,
      "grad_norm": 6.989369869232178,
      "learning_rate": 3.41227239237975e-05,
      "loss": 0.4618,
      "step": 24436
    },
    {
      "epoch": 7.716135143669088,
      "grad_norm": 7.273431777954102,
      "learning_rate": 3.412184682314142e-05,
      "loss": 0.4966,
      "step": 24437
    },
    {
      "epoch": 7.716450899905273,
      "grad_norm": 7.03981351852417,
      "learning_rate": 3.4120969722485354e-05,
      "loss": 0.4897,
      "step": 24438
    },
    {
      "epoch": 7.716766656141459,
      "grad_norm": 7.077220916748047,
      "learning_rate": 3.412009262182928e-05,
      "loss": 0.5051,
      "step": 24439
    },
    {
      "epoch": 7.7170824123776445,
      "grad_norm": 6.364189147949219,
      "learning_rate": 3.411921552117321e-05,
      "loss": 0.4321,
      "step": 24440
    },
    {
      "epoch": 7.71739816861383,
      "grad_norm": 4.7846174240112305,
      "learning_rate": 3.4118338420517136e-05,
      "loss": 0.5137,
      "step": 24441
    },
    {
      "epoch": 7.717713924850016,
      "grad_norm": 6.015773296356201,
      "learning_rate": 3.4117461319861074e-05,
      "loss": 0.5495,
      "step": 24442
    },
    {
      "epoch": 7.718029681086201,
      "grad_norm": 6.333199977874756,
      "learning_rate": 3.4116584219205e-05,
      "loss": 0.517,
      "step": 24443
    },
    {
      "epoch": 7.718345437322387,
      "grad_norm": 6.68567419052124,
      "learning_rate": 3.411570711854893e-05,
      "loss": 0.5297,
      "step": 24444
    },
    {
      "epoch": 7.718661193558573,
      "grad_norm": 5.9542765617370605,
      "learning_rate": 3.4114830017892856e-05,
      "loss": 0.4061,
      "step": 24445
    },
    {
      "epoch": 7.718976949794758,
      "grad_norm": 5.123924732208252,
      "learning_rate": 3.411395291723679e-05,
      "loss": 0.4384,
      "step": 24446
    },
    {
      "epoch": 7.719292706030944,
      "grad_norm": 7.430655002593994,
      "learning_rate": 3.411307581658071e-05,
      "loss": 0.4884,
      "step": 24447
    },
    {
      "epoch": 7.7196084622671295,
      "grad_norm": 8.344282150268555,
      "learning_rate": 3.4112198715924644e-05,
      "loss": 0.5576,
      "step": 24448
    },
    {
      "epoch": 7.719924218503316,
      "grad_norm": 4.753937721252441,
      "learning_rate": 3.411132161526857e-05,
      "loss": 0.5214,
      "step": 24449
    },
    {
      "epoch": 7.720239974739501,
      "grad_norm": 6.682652950286865,
      "learning_rate": 3.41104445146125e-05,
      "loss": 0.4819,
      "step": 24450
    },
    {
      "epoch": 7.720555730975687,
      "grad_norm": 4.664402961730957,
      "learning_rate": 3.4109567413956425e-05,
      "loss": 0.4127,
      "step": 24451
    },
    {
      "epoch": 7.720871487211872,
      "grad_norm": 5.850580215454102,
      "learning_rate": 3.410869031330036e-05,
      "loss": 0.44,
      "step": 24452
    },
    {
      "epoch": 7.721187243448058,
      "grad_norm": 5.2330756187438965,
      "learning_rate": 3.410781321264428e-05,
      "loss": 0.5503,
      "step": 24453
    },
    {
      "epoch": 7.721502999684244,
      "grad_norm": 6.313613414764404,
      "learning_rate": 3.4106936111988214e-05,
      "loss": 0.5896,
      "step": 24454
    },
    {
      "epoch": 7.721818755920429,
      "grad_norm": 6.21881628036499,
      "learning_rate": 3.4106059011332145e-05,
      "loss": 0.4597,
      "step": 24455
    },
    {
      "epoch": 7.722134512156615,
      "grad_norm": 5.240547180175781,
      "learning_rate": 3.410518191067607e-05,
      "loss": 0.5992,
      "step": 24456
    },
    {
      "epoch": 7.722450268392801,
      "grad_norm": 4.850319862365723,
      "learning_rate": 3.410430481002e-05,
      "loss": 0.5359,
      "step": 24457
    },
    {
      "epoch": 7.722766024628987,
      "grad_norm": 5.362825870513916,
      "learning_rate": 3.410342770936393e-05,
      "loss": 0.5322,
      "step": 24458
    },
    {
      "epoch": 7.723081780865172,
      "grad_norm": 6.992894649505615,
      "learning_rate": 3.410255060870786e-05,
      "loss": 0.5727,
      "step": 24459
    },
    {
      "epoch": 7.7233975371013575,
      "grad_norm": 7.871829509735107,
      "learning_rate": 3.410167350805178e-05,
      "loss": 0.5935,
      "step": 24460
    },
    {
      "epoch": 7.723713293337544,
      "grad_norm": 5.375301837921143,
      "learning_rate": 3.4100796407395715e-05,
      "loss": 0.48,
      "step": 24461
    },
    {
      "epoch": 7.724029049573729,
      "grad_norm": 5.941113471984863,
      "learning_rate": 3.409991930673964e-05,
      "loss": 0.5126,
      "step": 24462
    },
    {
      "epoch": 7.724344805809915,
      "grad_norm": 10.364253997802734,
      "learning_rate": 3.409904220608357e-05,
      "loss": 0.4521,
      "step": 24463
    },
    {
      "epoch": 7.7246605620461,
      "grad_norm": 5.79645299911499,
      "learning_rate": 3.4098165105427497e-05,
      "loss": 0.5795,
      "step": 24464
    },
    {
      "epoch": 7.724976318282286,
      "grad_norm": 5.257101058959961,
      "learning_rate": 3.409728800477143e-05,
      "loss": 0.5276,
      "step": 24465
    },
    {
      "epoch": 7.725292074518472,
      "grad_norm": 5.40331506729126,
      "learning_rate": 3.409641090411535e-05,
      "loss": 0.5151,
      "step": 24466
    },
    {
      "epoch": 7.725607830754657,
      "grad_norm": 6.102080345153809,
      "learning_rate": 3.4095533803459285e-05,
      "loss": 0.588,
      "step": 24467
    },
    {
      "epoch": 7.725923586990843,
      "grad_norm": 6.280014514923096,
      "learning_rate": 3.409465670280321e-05,
      "loss": 0.5493,
      "step": 24468
    },
    {
      "epoch": 7.726239343227029,
      "grad_norm": 5.36525821685791,
      "learning_rate": 3.409377960214715e-05,
      "loss": 0.4979,
      "step": 24469
    },
    {
      "epoch": 7.726555099463214,
      "grad_norm": 5.532746315002441,
      "learning_rate": 3.409290250149107e-05,
      "loss": 0.5189,
      "step": 24470
    },
    {
      "epoch": 7.7268708556994,
      "grad_norm": 6.455307483673096,
      "learning_rate": 3.4092025400835005e-05,
      "loss": 0.5872,
      "step": 24471
    },
    {
      "epoch": 7.727186611935585,
      "grad_norm": 4.949612140655518,
      "learning_rate": 3.409114830017893e-05,
      "loss": 0.4201,
      "step": 24472
    },
    {
      "epoch": 7.727502368171772,
      "grad_norm": 6.128682613372803,
      "learning_rate": 3.409027119952286e-05,
      "loss": 0.6007,
      "step": 24473
    },
    {
      "epoch": 7.727818124407957,
      "grad_norm": 4.988081932067871,
      "learning_rate": 3.4089394098866786e-05,
      "loss": 0.5467,
      "step": 24474
    },
    {
      "epoch": 7.728133880644143,
      "grad_norm": 5.5936079025268555,
      "learning_rate": 3.408851699821072e-05,
      "loss": 0.4724,
      "step": 24475
    },
    {
      "epoch": 7.728449636880328,
      "grad_norm": 5.148110389709473,
      "learning_rate": 3.408763989755465e-05,
      "loss": 0.4432,
      "step": 24476
    },
    {
      "epoch": 7.728765393116514,
      "grad_norm": 6.057583332061768,
      "learning_rate": 3.4086762796898575e-05,
      "loss": 0.5101,
      "step": 24477
    },
    {
      "epoch": 7.7290811493527,
      "grad_norm": 5.29234504699707,
      "learning_rate": 3.4085885696242506e-05,
      "loss": 0.4816,
      "step": 24478
    },
    {
      "epoch": 7.729396905588885,
      "grad_norm": 6.11430025100708,
      "learning_rate": 3.408500859558643e-05,
      "loss": 0.5189,
      "step": 24479
    },
    {
      "epoch": 7.729712661825071,
      "grad_norm": 5.844046592712402,
      "learning_rate": 3.408413149493036e-05,
      "loss": 0.3984,
      "step": 24480
    },
    {
      "epoch": 7.730028418061257,
      "grad_norm": 8.168408393859863,
      "learning_rate": 3.408325439427429e-05,
      "loss": 0.4709,
      "step": 24481
    },
    {
      "epoch": 7.730344174297443,
      "grad_norm": 6.129782676696777,
      "learning_rate": 3.408237729361822e-05,
      "loss": 0.4122,
      "step": 24482
    },
    {
      "epoch": 7.730659930533628,
      "grad_norm": 5.227976322174072,
      "learning_rate": 3.4081500192962144e-05,
      "loss": 0.4988,
      "step": 24483
    },
    {
      "epoch": 7.730975686769813,
      "grad_norm": 5.621649742126465,
      "learning_rate": 3.4080623092306076e-05,
      "loss": 0.4175,
      "step": 24484
    },
    {
      "epoch": 7.731291443006,
      "grad_norm": 6.549520015716553,
      "learning_rate": 3.407974599165e-05,
      "loss": 0.4797,
      "step": 24485
    },
    {
      "epoch": 7.731607199242185,
      "grad_norm": 10.339485168457031,
      "learning_rate": 3.407886889099393e-05,
      "loss": 0.4222,
      "step": 24486
    },
    {
      "epoch": 7.731922955478371,
      "grad_norm": 4.874812126159668,
      "learning_rate": 3.407799179033786e-05,
      "loss": 0.434,
      "step": 24487
    },
    {
      "epoch": 7.732238711714556,
      "grad_norm": 5.90730619430542,
      "learning_rate": 3.407711468968179e-05,
      "loss": 0.4563,
      "step": 24488
    },
    {
      "epoch": 7.732554467950742,
      "grad_norm": 5.8303608894348145,
      "learning_rate": 3.4076237589025714e-05,
      "loss": 0.5575,
      "step": 24489
    },
    {
      "epoch": 7.732870224186928,
      "grad_norm": 6.471522808074951,
      "learning_rate": 3.4075360488369646e-05,
      "loss": 0.4806,
      "step": 24490
    },
    {
      "epoch": 7.733185980423113,
      "grad_norm": 5.937220096588135,
      "learning_rate": 3.407448338771357e-05,
      "loss": 0.4586,
      "step": 24491
    },
    {
      "epoch": 7.733501736659299,
      "grad_norm": 5.7959442138671875,
      "learning_rate": 3.40736062870575e-05,
      "loss": 0.4799,
      "step": 24492
    },
    {
      "epoch": 7.733817492895485,
      "grad_norm": 4.627164363861084,
      "learning_rate": 3.4072729186401434e-05,
      "loss": 0.4612,
      "step": 24493
    },
    {
      "epoch": 7.73413324913167,
      "grad_norm": 5.4366068840026855,
      "learning_rate": 3.407185208574536e-05,
      "loss": 0.4857,
      "step": 24494
    },
    {
      "epoch": 7.734449005367856,
      "grad_norm": 8.147612571716309,
      "learning_rate": 3.407097498508929e-05,
      "loss": 0.5021,
      "step": 24495
    },
    {
      "epoch": 7.734764761604041,
      "grad_norm": 6.166742324829102,
      "learning_rate": 3.407009788443322e-05,
      "loss": 0.4573,
      "step": 24496
    },
    {
      "epoch": 7.7350805178402275,
      "grad_norm": 5.36167573928833,
      "learning_rate": 3.406922078377715e-05,
      "loss": 0.4339,
      "step": 24497
    },
    {
      "epoch": 7.735396274076413,
      "grad_norm": 5.170472621917725,
      "learning_rate": 3.406834368312108e-05,
      "loss": 0.3699,
      "step": 24498
    },
    {
      "epoch": 7.735712030312599,
      "grad_norm": 8.181207656860352,
      "learning_rate": 3.406746658246501e-05,
      "loss": 0.4764,
      "step": 24499
    },
    {
      "epoch": 7.736027786548784,
      "grad_norm": 5.151822090148926,
      "learning_rate": 3.4066589481808935e-05,
      "loss": 0.4954,
      "step": 24500
    },
    {
      "epoch": 7.7363435427849705,
      "grad_norm": 6.31390380859375,
      "learning_rate": 3.406571238115287e-05,
      "loss": 0.4852,
      "step": 24501
    },
    {
      "epoch": 7.736659299021156,
      "grad_norm": 4.734913349151611,
      "learning_rate": 3.406483528049679e-05,
      "loss": 0.4759,
      "step": 24502
    },
    {
      "epoch": 7.736975055257341,
      "grad_norm": 6.5715813636779785,
      "learning_rate": 3.4063958179840724e-05,
      "loss": 0.5106,
      "step": 24503
    },
    {
      "epoch": 7.737290811493527,
      "grad_norm": 6.519681453704834,
      "learning_rate": 3.406308107918465e-05,
      "loss": 0.6008,
      "step": 24504
    },
    {
      "epoch": 7.737606567729713,
      "grad_norm": 5.596598148345947,
      "learning_rate": 3.406220397852858e-05,
      "loss": 0.4524,
      "step": 24505
    },
    {
      "epoch": 7.737922323965899,
      "grad_norm": 4.637294292449951,
      "learning_rate": 3.4061326877872505e-05,
      "loss": 0.4198,
      "step": 24506
    },
    {
      "epoch": 7.738238080202084,
      "grad_norm": 5.028743743896484,
      "learning_rate": 3.406044977721644e-05,
      "loss": 0.5836,
      "step": 24507
    },
    {
      "epoch": 7.738553836438269,
      "grad_norm": 6.161675930023193,
      "learning_rate": 3.405957267656036e-05,
      "loss": 0.4351,
      "step": 24508
    },
    {
      "epoch": 7.7388695926744555,
      "grad_norm": 5.92026424407959,
      "learning_rate": 3.4058695575904293e-05,
      "loss": 0.4508,
      "step": 24509
    },
    {
      "epoch": 7.739185348910641,
      "grad_norm": 5.07465124130249,
      "learning_rate": 3.405781847524822e-05,
      "loss": 0.4454,
      "step": 24510
    },
    {
      "epoch": 7.739501105146827,
      "grad_norm": 6.700721740722656,
      "learning_rate": 3.405694137459215e-05,
      "loss": 0.4423,
      "step": 24511
    },
    {
      "epoch": 7.739816861383012,
      "grad_norm": 5.72750997543335,
      "learning_rate": 3.4056064273936075e-05,
      "loss": 0.4473,
      "step": 24512
    },
    {
      "epoch": 7.740132617619198,
      "grad_norm": 5.072987079620361,
      "learning_rate": 3.405518717328001e-05,
      "loss": 0.4772,
      "step": 24513
    },
    {
      "epoch": 7.740448373855384,
      "grad_norm": 5.32208776473999,
      "learning_rate": 3.405431007262394e-05,
      "loss": 0.4791,
      "step": 24514
    },
    {
      "epoch": 7.740764130091569,
      "grad_norm": 5.375752925872803,
      "learning_rate": 3.405343297196786e-05,
      "loss": 0.4485,
      "step": 24515
    },
    {
      "epoch": 7.741079886327755,
      "grad_norm": 6.540480613708496,
      "learning_rate": 3.4052555871311795e-05,
      "loss": 0.5419,
      "step": 24516
    },
    {
      "epoch": 7.7413956425639405,
      "grad_norm": 6.417546272277832,
      "learning_rate": 3.405167877065572e-05,
      "loss": 0.5253,
      "step": 24517
    },
    {
      "epoch": 7.741711398800126,
      "grad_norm": 6.2002153396606445,
      "learning_rate": 3.405080166999965e-05,
      "loss": 0.5495,
      "step": 24518
    },
    {
      "epoch": 7.742027155036312,
      "grad_norm": 5.179248809814453,
      "learning_rate": 3.4049924569343576e-05,
      "loss": 0.4462,
      "step": 24519
    },
    {
      "epoch": 7.742342911272497,
      "grad_norm": 5.738667964935303,
      "learning_rate": 3.404904746868751e-05,
      "loss": 0.3828,
      "step": 24520
    },
    {
      "epoch": 7.7426586675086835,
      "grad_norm": 5.861727237701416,
      "learning_rate": 3.404817036803143e-05,
      "loss": 0.4531,
      "step": 24521
    },
    {
      "epoch": 7.742974423744869,
      "grad_norm": 5.972433567047119,
      "learning_rate": 3.4047293267375365e-05,
      "loss": 0.4657,
      "step": 24522
    },
    {
      "epoch": 7.743290179981055,
      "grad_norm": 4.892493724822998,
      "learning_rate": 3.404641616671929e-05,
      "loss": 0.3797,
      "step": 24523
    },
    {
      "epoch": 7.74360593621724,
      "grad_norm": 6.244939804077148,
      "learning_rate": 3.404553906606323e-05,
      "loss": 0.5482,
      "step": 24524
    },
    {
      "epoch": 7.743921692453426,
      "grad_norm": 5.863813877105713,
      "learning_rate": 3.404466196540715e-05,
      "loss": 0.466,
      "step": 24525
    },
    {
      "epoch": 7.744237448689612,
      "grad_norm": 6.714974403381348,
      "learning_rate": 3.4043784864751085e-05,
      "loss": 0.508,
      "step": 24526
    },
    {
      "epoch": 7.744553204925797,
      "grad_norm": 5.951221466064453,
      "learning_rate": 3.404290776409501e-05,
      "loss": 0.4619,
      "step": 24527
    },
    {
      "epoch": 7.744868961161983,
      "grad_norm": 5.400632381439209,
      "learning_rate": 3.404203066343894e-05,
      "loss": 0.472,
      "step": 24528
    },
    {
      "epoch": 7.7451847173981685,
      "grad_norm": 5.331352233886719,
      "learning_rate": 3.4041153562782866e-05,
      "loss": 0.4859,
      "step": 24529
    },
    {
      "epoch": 7.745500473634355,
      "grad_norm": 5.136883735656738,
      "learning_rate": 3.40402764621268e-05,
      "loss": 0.4428,
      "step": 24530
    },
    {
      "epoch": 7.74581622987054,
      "grad_norm": 5.429234504699707,
      "learning_rate": 3.403939936147072e-05,
      "loss": 0.56,
      "step": 24531
    },
    {
      "epoch": 7.746131986106725,
      "grad_norm": 5.059053421020508,
      "learning_rate": 3.4038522260814654e-05,
      "loss": 0.4412,
      "step": 24532
    },
    {
      "epoch": 7.7464477423429114,
      "grad_norm": 5.292763710021973,
      "learning_rate": 3.403764516015858e-05,
      "loss": 0.4059,
      "step": 24533
    },
    {
      "epoch": 7.746763498579097,
      "grad_norm": 5.584027290344238,
      "learning_rate": 3.403676805950251e-05,
      "loss": 0.4557,
      "step": 24534
    },
    {
      "epoch": 7.747079254815283,
      "grad_norm": 5.016859531402588,
      "learning_rate": 3.4035890958846436e-05,
      "loss": 0.4685,
      "step": 24535
    },
    {
      "epoch": 7.747395011051468,
      "grad_norm": 4.2933173179626465,
      "learning_rate": 3.403501385819037e-05,
      "loss": 0.4633,
      "step": 24536
    },
    {
      "epoch": 7.7477107672876535,
      "grad_norm": 6.906801700592041,
      "learning_rate": 3.40341367575343e-05,
      "loss": 0.6299,
      "step": 24537
    },
    {
      "epoch": 7.74802652352384,
      "grad_norm": 5.750057220458984,
      "learning_rate": 3.4033259656878224e-05,
      "loss": 0.6063,
      "step": 24538
    },
    {
      "epoch": 7.748342279760025,
      "grad_norm": 5.4603071212768555,
      "learning_rate": 3.4032382556222156e-05,
      "loss": 0.5631,
      "step": 24539
    },
    {
      "epoch": 7.748658035996211,
      "grad_norm": 5.534727096557617,
      "learning_rate": 3.403150545556608e-05,
      "loss": 0.4956,
      "step": 24540
    },
    {
      "epoch": 7.7489737922323965,
      "grad_norm": 5.432828426361084,
      "learning_rate": 3.403062835491001e-05,
      "loss": 0.5209,
      "step": 24541
    },
    {
      "epoch": 7.749289548468582,
      "grad_norm": 6.168618679046631,
      "learning_rate": 3.402975125425394e-05,
      "loss": 0.4934,
      "step": 24542
    },
    {
      "epoch": 7.749605304704768,
      "grad_norm": 5.106008052825928,
      "learning_rate": 3.402887415359787e-05,
      "loss": 0.5662,
      "step": 24543
    },
    {
      "epoch": 7.749921060940953,
      "grad_norm": 3.6278862953186035,
      "learning_rate": 3.4027997052941794e-05,
      "loss": 0.3892,
      "step": 24544
    },
    {
      "epoch": 7.750236817177139,
      "grad_norm": 6.8927001953125,
      "learning_rate": 3.4027119952285726e-05,
      "loss": 0.5443,
      "step": 24545
    },
    {
      "epoch": 7.750552573413325,
      "grad_norm": 4.858646869659424,
      "learning_rate": 3.402624285162965e-05,
      "loss": 0.4658,
      "step": 24546
    },
    {
      "epoch": 7.750868329649511,
      "grad_norm": 5.315151214599609,
      "learning_rate": 3.402536575097358e-05,
      "loss": 0.4706,
      "step": 24547
    },
    {
      "epoch": 7.751184085885696,
      "grad_norm": 5.215938568115234,
      "learning_rate": 3.402448865031751e-05,
      "loss": 0.4838,
      "step": 24548
    },
    {
      "epoch": 7.751499842121882,
      "grad_norm": 5.470386028289795,
      "learning_rate": 3.402361154966144e-05,
      "loss": 0.4434,
      "step": 24549
    },
    {
      "epoch": 7.751815598358068,
      "grad_norm": 7.531530380249023,
      "learning_rate": 3.4022734449005364e-05,
      "loss": 0.4718,
      "step": 24550
    },
    {
      "epoch": 7.752131354594253,
      "grad_norm": 6.42430305480957,
      "learning_rate": 3.40218573483493e-05,
      "loss": 0.5196,
      "step": 24551
    },
    {
      "epoch": 7.752447110830439,
      "grad_norm": 5.67386531829834,
      "learning_rate": 3.402098024769323e-05,
      "loss": 0.513,
      "step": 24552
    },
    {
      "epoch": 7.752762867066624,
      "grad_norm": 5.776220321655273,
      "learning_rate": 3.402010314703716e-05,
      "loss": 0.4635,
      "step": 24553
    },
    {
      "epoch": 7.753078623302811,
      "grad_norm": 7.585020542144775,
      "learning_rate": 3.4019226046381084e-05,
      "loss": 0.5375,
      "step": 24554
    },
    {
      "epoch": 7.753394379538996,
      "grad_norm": 6.618837833404541,
      "learning_rate": 3.4018348945725015e-05,
      "loss": 0.555,
      "step": 24555
    },
    {
      "epoch": 7.753710135775181,
      "grad_norm": 5.060762882232666,
      "learning_rate": 3.401747184506894e-05,
      "loss": 0.5189,
      "step": 24556
    },
    {
      "epoch": 7.754025892011367,
      "grad_norm": 7.056665420532227,
      "learning_rate": 3.401659474441287e-05,
      "loss": 0.5621,
      "step": 24557
    },
    {
      "epoch": 7.754341648247553,
      "grad_norm": 5.102815628051758,
      "learning_rate": 3.4015717643756804e-05,
      "loss": 0.4283,
      "step": 24558
    },
    {
      "epoch": 7.754657404483739,
      "grad_norm": 7.053191661834717,
      "learning_rate": 3.401484054310073e-05,
      "loss": 0.5656,
      "step": 24559
    },
    {
      "epoch": 7.754973160719924,
      "grad_norm": 4.600368976593018,
      "learning_rate": 3.401396344244466e-05,
      "loss": 0.4129,
      "step": 24560
    },
    {
      "epoch": 7.7552889169561094,
      "grad_norm": 7.760107517242432,
      "learning_rate": 3.4013086341788585e-05,
      "loss": 0.5957,
      "step": 24561
    },
    {
      "epoch": 7.755604673192296,
      "grad_norm": 4.565169811248779,
      "learning_rate": 3.401220924113252e-05,
      "loss": 0.4267,
      "step": 24562
    },
    {
      "epoch": 7.755920429428481,
      "grad_norm": 5.683244228363037,
      "learning_rate": 3.401133214047644e-05,
      "loss": 0.4665,
      "step": 24563
    },
    {
      "epoch": 7.756236185664667,
      "grad_norm": 6.157104015350342,
      "learning_rate": 3.401045503982037e-05,
      "loss": 0.5941,
      "step": 24564
    },
    {
      "epoch": 7.756551941900852,
      "grad_norm": 6.222358703613281,
      "learning_rate": 3.40095779391643e-05,
      "loss": 0.5092,
      "step": 24565
    },
    {
      "epoch": 7.756867698137039,
      "grad_norm": 5.892089366912842,
      "learning_rate": 3.400870083850823e-05,
      "loss": 0.5587,
      "step": 24566
    },
    {
      "epoch": 7.757183454373224,
      "grad_norm": 5.182835102081299,
      "learning_rate": 3.4007823737852155e-05,
      "loss": 0.4794,
      "step": 24567
    },
    {
      "epoch": 7.757499210609409,
      "grad_norm": 6.335943698883057,
      "learning_rate": 3.4006946637196087e-05,
      "loss": 0.5694,
      "step": 24568
    },
    {
      "epoch": 7.757814966845595,
      "grad_norm": 7.175065040588379,
      "learning_rate": 3.400606953654001e-05,
      "loss": 0.6554,
      "step": 24569
    },
    {
      "epoch": 7.758130723081781,
      "grad_norm": 5.196305751800537,
      "learning_rate": 3.400519243588394e-05,
      "loss": 0.4792,
      "step": 24570
    },
    {
      "epoch": 7.758446479317967,
      "grad_norm": 5.6143035888671875,
      "learning_rate": 3.400431533522787e-05,
      "loss": 0.5638,
      "step": 24571
    },
    {
      "epoch": 7.758762235554152,
      "grad_norm": 6.40297269821167,
      "learning_rate": 3.40034382345718e-05,
      "loss": 0.5043,
      "step": 24572
    },
    {
      "epoch": 7.759077991790338,
      "grad_norm": 4.955352783203125,
      "learning_rate": 3.4002561133915725e-05,
      "loss": 0.5304,
      "step": 24573
    },
    {
      "epoch": 7.759393748026524,
      "grad_norm": 6.341638088226318,
      "learning_rate": 3.4001684033259656e-05,
      "loss": 0.4482,
      "step": 24574
    },
    {
      "epoch": 7.759709504262709,
      "grad_norm": 5.400972366333008,
      "learning_rate": 3.400080693260359e-05,
      "loss": 0.5435,
      "step": 24575
    },
    {
      "epoch": 7.760025260498895,
      "grad_norm": 5.155022144317627,
      "learning_rate": 3.399992983194751e-05,
      "loss": 0.41,
      "step": 24576
    },
    {
      "epoch": 7.76034101673508,
      "grad_norm": 3.9372875690460205,
      "learning_rate": 3.3999052731291445e-05,
      "loss": 0.3882,
      "step": 24577
    },
    {
      "epoch": 7.7606567729712665,
      "grad_norm": 5.2138190269470215,
      "learning_rate": 3.3998175630635376e-05,
      "loss": 0.5485,
      "step": 24578
    },
    {
      "epoch": 7.760972529207452,
      "grad_norm": 4.6481757164001465,
      "learning_rate": 3.39972985299793e-05,
      "loss": 0.556,
      "step": 24579
    },
    {
      "epoch": 7.761288285443637,
      "grad_norm": 6.437254905700684,
      "learning_rate": 3.399642142932323e-05,
      "loss": 0.5806,
      "step": 24580
    },
    {
      "epoch": 7.761604041679823,
      "grad_norm": 6.543833255767822,
      "learning_rate": 3.3995544328667165e-05,
      "loss": 0.4937,
      "step": 24581
    },
    {
      "epoch": 7.761919797916009,
      "grad_norm": 4.845306396484375,
      "learning_rate": 3.399466722801109e-05,
      "loss": 0.443,
      "step": 24582
    },
    {
      "epoch": 7.762235554152195,
      "grad_norm": 4.973429203033447,
      "learning_rate": 3.399379012735502e-05,
      "loss": 0.4877,
      "step": 24583
    },
    {
      "epoch": 7.76255131038838,
      "grad_norm": 4.930348873138428,
      "learning_rate": 3.3992913026698946e-05,
      "loss": 0.476,
      "step": 24584
    },
    {
      "epoch": 7.762867066624565,
      "grad_norm": 5.294396877288818,
      "learning_rate": 3.399203592604288e-05,
      "loss": 0.4366,
      "step": 24585
    },
    {
      "epoch": 7.763182822860752,
      "grad_norm": 6.271330833435059,
      "learning_rate": 3.39911588253868e-05,
      "loss": 0.4946,
      "step": 24586
    },
    {
      "epoch": 7.763498579096937,
      "grad_norm": 4.545307159423828,
      "learning_rate": 3.3990281724730734e-05,
      "loss": 0.419,
      "step": 24587
    },
    {
      "epoch": 7.763814335333123,
      "grad_norm": 6.761368751525879,
      "learning_rate": 3.398940462407466e-05,
      "loss": 0.4929,
      "step": 24588
    },
    {
      "epoch": 7.764130091569308,
      "grad_norm": 5.574032306671143,
      "learning_rate": 3.398852752341859e-05,
      "loss": 0.4516,
      "step": 24589
    },
    {
      "epoch": 7.7644458478054945,
      "grad_norm": 4.856653690338135,
      "learning_rate": 3.3987650422762516e-05,
      "loss": 0.5128,
      "step": 24590
    },
    {
      "epoch": 7.76476160404168,
      "grad_norm": 5.683958053588867,
      "learning_rate": 3.398677332210645e-05,
      "loss": 0.4805,
      "step": 24591
    },
    {
      "epoch": 7.765077360277865,
      "grad_norm": 8.232451438903809,
      "learning_rate": 3.398589622145037e-05,
      "loss": 0.5463,
      "step": 24592
    },
    {
      "epoch": 7.765393116514051,
      "grad_norm": 6.311675071716309,
      "learning_rate": 3.3985019120794304e-05,
      "loss": 0.4975,
      "step": 24593
    },
    {
      "epoch": 7.765708872750237,
      "grad_norm": 6.925487995147705,
      "learning_rate": 3.398414202013823e-05,
      "loss": 0.4813,
      "step": 24594
    },
    {
      "epoch": 7.766024628986423,
      "grad_norm": 6.008655071258545,
      "learning_rate": 3.398326491948216e-05,
      "loss": 0.5604,
      "step": 24595
    },
    {
      "epoch": 7.766340385222608,
      "grad_norm": 4.447588920593262,
      "learning_rate": 3.398238781882609e-05,
      "loss": 0.4286,
      "step": 24596
    },
    {
      "epoch": 7.766656141458794,
      "grad_norm": 4.9045515060424805,
      "learning_rate": 3.398151071817002e-05,
      "loss": 0.4847,
      "step": 24597
    },
    {
      "epoch": 7.7669718976949795,
      "grad_norm": 5.764402389526367,
      "learning_rate": 3.398063361751395e-05,
      "loss": 0.5248,
      "step": 24598
    },
    {
      "epoch": 7.767287653931165,
      "grad_norm": 5.206434726715088,
      "learning_rate": 3.3979756516857874e-05,
      "loss": 0.4521,
      "step": 24599
    },
    {
      "epoch": 7.767603410167351,
      "grad_norm": 4.888866901397705,
      "learning_rate": 3.3978879416201806e-05,
      "loss": 0.3644,
      "step": 24600
    },
    {
      "epoch": 7.767603410167351,
      "eval_loss": 0.9399959444999695,
      "eval_runtime": 4.683,
      "eval_samples_per_second": 531.713,
      "eval_steps_per_second": 4.484,
      "step": 24600
    },
    {
      "epoch": 7.767919166403536,
      "grad_norm": 4.839745044708252,
      "learning_rate": 3.397800231554573e-05,
      "loss": 0.4456,
      "step": 24601
    },
    {
      "epoch": 7.7682349226397225,
      "grad_norm": 7.153501510620117,
      "learning_rate": 3.397712521488966e-05,
      "loss": 0.4725,
      "step": 24602
    },
    {
      "epoch": 7.768550678875908,
      "grad_norm": 6.822513103485107,
      "learning_rate": 3.397624811423359e-05,
      "loss": 0.4715,
      "step": 24603
    },
    {
      "epoch": 7.768866435112093,
      "grad_norm": 5.958865165710449,
      "learning_rate": 3.397537101357752e-05,
      "loss": 0.6054,
      "step": 24604
    },
    {
      "epoch": 7.769182191348279,
      "grad_norm": 5.919800758361816,
      "learning_rate": 3.397449391292145e-05,
      "loss": 0.5374,
      "step": 24605
    },
    {
      "epoch": 7.7694979475844645,
      "grad_norm": 5.395540237426758,
      "learning_rate": 3.397361681226538e-05,
      "loss": 0.5671,
      "step": 24606
    },
    {
      "epoch": 7.769813703820651,
      "grad_norm": 6.416505813598633,
      "learning_rate": 3.397273971160931e-05,
      "loss": 0.4974,
      "step": 24607
    },
    {
      "epoch": 7.770129460056836,
      "grad_norm": 5.354897975921631,
      "learning_rate": 3.397186261095324e-05,
      "loss": 0.4665,
      "step": 24608
    },
    {
      "epoch": 7.770445216293021,
      "grad_norm": 8.000927925109863,
      "learning_rate": 3.3970985510297164e-05,
      "loss": 0.613,
      "step": 24609
    },
    {
      "epoch": 7.7707609725292075,
      "grad_norm": 6.029479026794434,
      "learning_rate": 3.3970108409641095e-05,
      "loss": 0.4088,
      "step": 24610
    },
    {
      "epoch": 7.771076728765393,
      "grad_norm": 5.687119007110596,
      "learning_rate": 3.396923130898502e-05,
      "loss": 0.47,
      "step": 24611
    },
    {
      "epoch": 7.771392485001579,
      "grad_norm": 5.377256870269775,
      "learning_rate": 3.396835420832895e-05,
      "loss": 0.4158,
      "step": 24612
    },
    {
      "epoch": 7.771708241237764,
      "grad_norm": 5.267536640167236,
      "learning_rate": 3.396747710767288e-05,
      "loss": 0.44,
      "step": 24613
    },
    {
      "epoch": 7.7720239974739505,
      "grad_norm": 5.880112171173096,
      "learning_rate": 3.396660000701681e-05,
      "loss": 0.4561,
      "step": 24614
    },
    {
      "epoch": 7.772339753710136,
      "grad_norm": 6.055591106414795,
      "learning_rate": 3.396572290636073e-05,
      "loss": 0.4663,
      "step": 24615
    },
    {
      "epoch": 7.772655509946322,
      "grad_norm": 4.849295616149902,
      "learning_rate": 3.3964845805704665e-05,
      "loss": 0.5778,
      "step": 24616
    },
    {
      "epoch": 7.772971266182507,
      "grad_norm": 5.830443382263184,
      "learning_rate": 3.396396870504859e-05,
      "loss": 0.4861,
      "step": 24617
    },
    {
      "epoch": 7.7732870224186925,
      "grad_norm": 5.312346935272217,
      "learning_rate": 3.396309160439252e-05,
      "loss": 0.4912,
      "step": 24618
    },
    {
      "epoch": 7.773602778654879,
      "grad_norm": 9.152440071105957,
      "learning_rate": 3.396221450373645e-05,
      "loss": 0.5968,
      "step": 24619
    },
    {
      "epoch": 7.773918534891064,
      "grad_norm": 6.71992826461792,
      "learning_rate": 3.396133740308038e-05,
      "loss": 0.6175,
      "step": 24620
    },
    {
      "epoch": 7.77423429112725,
      "grad_norm": 6.026232719421387,
      "learning_rate": 3.396046030242431e-05,
      "loss": 0.4371,
      "step": 24621
    },
    {
      "epoch": 7.7745500473634355,
      "grad_norm": 18.760986328125,
      "learning_rate": 3.3959583201768235e-05,
      "loss": 0.4235,
      "step": 24622
    },
    {
      "epoch": 7.774865803599621,
      "grad_norm": 5.3109354972839355,
      "learning_rate": 3.3958706101112166e-05,
      "loss": 0.5494,
      "step": 24623
    },
    {
      "epoch": 7.775181559835807,
      "grad_norm": 5.552474021911621,
      "learning_rate": 3.395782900045609e-05,
      "loss": 0.5226,
      "step": 24624
    },
    {
      "epoch": 7.775497316071992,
      "grad_norm": 5.4737372398376465,
      "learning_rate": 3.395695189980002e-05,
      "loss": 0.4321,
      "step": 24625
    },
    {
      "epoch": 7.775813072308178,
      "grad_norm": 4.962453842163086,
      "learning_rate": 3.395607479914395e-05,
      "loss": 0.427,
      "step": 24626
    },
    {
      "epoch": 7.776128828544364,
      "grad_norm": 5.150883674621582,
      "learning_rate": 3.395519769848788e-05,
      "loss": 0.5206,
      "step": 24627
    },
    {
      "epoch": 7.776444584780549,
      "grad_norm": 5.6212921142578125,
      "learning_rate": 3.3954320597831805e-05,
      "loss": 0.5501,
      "step": 24628
    },
    {
      "epoch": 7.776760341016735,
      "grad_norm": 5.471866607666016,
      "learning_rate": 3.3953443497175736e-05,
      "loss": 0.4772,
      "step": 24629
    },
    {
      "epoch": 7.7770760972529205,
      "grad_norm": 5.750648021697998,
      "learning_rate": 3.395256639651966e-05,
      "loss": 0.5106,
      "step": 24630
    },
    {
      "epoch": 7.777391853489107,
      "grad_norm": 5.008350849151611,
      "learning_rate": 3.395168929586359e-05,
      "loss": 0.49,
      "step": 24631
    },
    {
      "epoch": 7.777707609725292,
      "grad_norm": 6.152757167816162,
      "learning_rate": 3.395081219520752e-05,
      "loss": 0.4942,
      "step": 24632
    },
    {
      "epoch": 7.778023365961477,
      "grad_norm": 5.239198207855225,
      "learning_rate": 3.3949935094551456e-05,
      "loss": 0.4581,
      "step": 24633
    },
    {
      "epoch": 7.778339122197663,
      "grad_norm": 5.756923198699951,
      "learning_rate": 3.394905799389538e-05,
      "loss": 0.4631,
      "step": 24634
    },
    {
      "epoch": 7.778654878433849,
      "grad_norm": 5.425830841064453,
      "learning_rate": 3.394818089323931e-05,
      "loss": 0.4329,
      "step": 24635
    },
    {
      "epoch": 7.778970634670035,
      "grad_norm": 5.194479465484619,
      "learning_rate": 3.394730379258324e-05,
      "loss": 0.456,
      "step": 24636
    },
    {
      "epoch": 7.77928639090622,
      "grad_norm": 5.183365821838379,
      "learning_rate": 3.394642669192717e-05,
      "loss": 0.523,
      "step": 24637
    },
    {
      "epoch": 7.779602147142406,
      "grad_norm": 6.5022125244140625,
      "learning_rate": 3.3945549591271094e-05,
      "loss": 0.5511,
      "step": 24638
    },
    {
      "epoch": 7.779917903378592,
      "grad_norm": 6.995261192321777,
      "learning_rate": 3.3944672490615026e-05,
      "loss": 0.4998,
      "step": 24639
    },
    {
      "epoch": 7.780233659614778,
      "grad_norm": 5.039374828338623,
      "learning_rate": 3.394379538995896e-05,
      "loss": 0.3515,
      "step": 24640
    },
    {
      "epoch": 7.780549415850963,
      "grad_norm": 6.597656726837158,
      "learning_rate": 3.394291828930288e-05,
      "loss": 0.6038,
      "step": 24641
    },
    {
      "epoch": 7.7808651720871485,
      "grad_norm": 6.532550811767578,
      "learning_rate": 3.3942041188646814e-05,
      "loss": 0.4757,
      "step": 24642
    },
    {
      "epoch": 7.781180928323335,
      "grad_norm": 4.920984268188477,
      "learning_rate": 3.394116408799074e-05,
      "loss": 0.5096,
      "step": 24643
    },
    {
      "epoch": 7.78149668455952,
      "grad_norm": 6.752944469451904,
      "learning_rate": 3.394028698733467e-05,
      "loss": 0.5725,
      "step": 24644
    },
    {
      "epoch": 7.781812440795706,
      "grad_norm": 4.0070061683654785,
      "learning_rate": 3.3939409886678596e-05,
      "loss": 0.4035,
      "step": 24645
    },
    {
      "epoch": 7.782128197031891,
      "grad_norm": 6.61279296875,
      "learning_rate": 3.393853278602253e-05,
      "loss": 0.535,
      "step": 24646
    },
    {
      "epoch": 7.782443953268077,
      "grad_norm": 5.845382213592529,
      "learning_rate": 3.393765568536645e-05,
      "loss": 0.5751,
      "step": 24647
    },
    {
      "epoch": 7.782759709504263,
      "grad_norm": 5.870367527008057,
      "learning_rate": 3.3936778584710384e-05,
      "loss": 0.4758,
      "step": 24648
    },
    {
      "epoch": 7.783075465740448,
      "grad_norm": 7.9758381843566895,
      "learning_rate": 3.393590148405431e-05,
      "loss": 0.5198,
      "step": 24649
    },
    {
      "epoch": 7.783391221976634,
      "grad_norm": 6.966758728027344,
      "learning_rate": 3.393502438339824e-05,
      "loss": 0.5036,
      "step": 24650
    },
    {
      "epoch": 7.78370697821282,
      "grad_norm": 6.355992794036865,
      "learning_rate": 3.3934147282742165e-05,
      "loss": 0.5616,
      "step": 24651
    },
    {
      "epoch": 7.784022734449005,
      "grad_norm": 7.3413801193237305,
      "learning_rate": 3.39332701820861e-05,
      "loss": 0.5604,
      "step": 24652
    },
    {
      "epoch": 7.784338490685191,
      "grad_norm": 4.751183986663818,
      "learning_rate": 3.393239308143002e-05,
      "loss": 0.4838,
      "step": 24653
    },
    {
      "epoch": 7.784654246921376,
      "grad_norm": 6.978358268737793,
      "learning_rate": 3.3931515980773954e-05,
      "loss": 0.5181,
      "step": 24654
    },
    {
      "epoch": 7.784970003157563,
      "grad_norm": 6.924454689025879,
      "learning_rate": 3.3930638880117885e-05,
      "loss": 0.5918,
      "step": 24655
    },
    {
      "epoch": 7.785285759393748,
      "grad_norm": 5.855520725250244,
      "learning_rate": 3.392976177946181e-05,
      "loss": 0.5632,
      "step": 24656
    },
    {
      "epoch": 7.785601515629934,
      "grad_norm": 5.435208797454834,
      "learning_rate": 3.392888467880574e-05,
      "loss": 0.5746,
      "step": 24657
    },
    {
      "epoch": 7.785917271866119,
      "grad_norm": 6.3730058670043945,
      "learning_rate": 3.392800757814967e-05,
      "loss": 0.5128,
      "step": 24658
    },
    {
      "epoch": 7.786233028102305,
      "grad_norm": 6.3485236167907715,
      "learning_rate": 3.39271304774936e-05,
      "loss": 0.4878,
      "step": 24659
    },
    {
      "epoch": 7.786548784338491,
      "grad_norm": 4.869632244110107,
      "learning_rate": 3.392625337683753e-05,
      "loss": 0.4852,
      "step": 24660
    },
    {
      "epoch": 7.786864540574676,
      "grad_norm": 5.306459903717041,
      "learning_rate": 3.392537627618146e-05,
      "loss": 0.4987,
      "step": 24661
    },
    {
      "epoch": 7.787180296810862,
      "grad_norm": 6.354001998901367,
      "learning_rate": 3.392449917552539e-05,
      "loss": 0.4804,
      "step": 24662
    },
    {
      "epoch": 7.787496053047048,
      "grad_norm": 6.176353454589844,
      "learning_rate": 3.392362207486932e-05,
      "loss": 0.5338,
      "step": 24663
    },
    {
      "epoch": 7.787811809283234,
      "grad_norm": 7.823150634765625,
      "learning_rate": 3.3922744974213243e-05,
      "loss": 0.5971,
      "step": 24664
    },
    {
      "epoch": 7.788127565519419,
      "grad_norm": 5.115209579467773,
      "learning_rate": 3.3921867873557175e-05,
      "loss": 0.4824,
      "step": 24665
    },
    {
      "epoch": 7.788443321755604,
      "grad_norm": 6.836040496826172,
      "learning_rate": 3.39209907729011e-05,
      "loss": 0.5176,
      "step": 24666
    },
    {
      "epoch": 7.788759077991791,
      "grad_norm": 6.002504825592041,
      "learning_rate": 3.392011367224503e-05,
      "loss": 0.434,
      "step": 24667
    },
    {
      "epoch": 7.789074834227976,
      "grad_norm": 5.762038707733154,
      "learning_rate": 3.391923657158896e-05,
      "loss": 0.54,
      "step": 24668
    },
    {
      "epoch": 7.789390590464162,
      "grad_norm": 4.50288200378418,
      "learning_rate": 3.391835947093289e-05,
      "loss": 0.4429,
      "step": 24669
    },
    {
      "epoch": 7.789706346700347,
      "grad_norm": 6.145411968231201,
      "learning_rate": 3.391748237027681e-05,
      "loss": 0.4925,
      "step": 24670
    },
    {
      "epoch": 7.790022102936533,
      "grad_norm": 5.533825874328613,
      "learning_rate": 3.3916605269620745e-05,
      "loss": 0.5194,
      "step": 24671
    },
    {
      "epoch": 7.790337859172719,
      "grad_norm": 6.266472339630127,
      "learning_rate": 3.391572816896467e-05,
      "loss": 0.4859,
      "step": 24672
    },
    {
      "epoch": 7.790653615408904,
      "grad_norm": 5.502071380615234,
      "learning_rate": 3.39148510683086e-05,
      "loss": 0.5171,
      "step": 24673
    },
    {
      "epoch": 7.79096937164509,
      "grad_norm": 5.652970314025879,
      "learning_rate": 3.3913973967652526e-05,
      "loss": 0.475,
      "step": 24674
    },
    {
      "epoch": 7.791285127881276,
      "grad_norm": 4.479231357574463,
      "learning_rate": 3.391309686699646e-05,
      "loss": 0.3923,
      "step": 24675
    },
    {
      "epoch": 7.791600884117461,
      "grad_norm": 7.613482475280762,
      "learning_rate": 3.391221976634038e-05,
      "loss": 0.5197,
      "step": 24676
    },
    {
      "epoch": 7.791916640353647,
      "grad_norm": 4.6443867683410645,
      "learning_rate": 3.3911342665684315e-05,
      "loss": 0.3825,
      "step": 24677
    },
    {
      "epoch": 7.792232396589832,
      "grad_norm": 4.8895978927612305,
      "learning_rate": 3.3910465565028246e-05,
      "loss": 0.4532,
      "step": 24678
    },
    {
      "epoch": 7.7925481528260185,
      "grad_norm": 6.584475517272949,
      "learning_rate": 3.390958846437217e-05,
      "loss": 0.5127,
      "step": 24679
    },
    {
      "epoch": 7.792863909062204,
      "grad_norm": 7.414073944091797,
      "learning_rate": 3.39087113637161e-05,
      "loss": 0.6236,
      "step": 24680
    },
    {
      "epoch": 7.79317966529839,
      "grad_norm": 5.147064208984375,
      "learning_rate": 3.390783426306003e-05,
      "loss": 0.4345,
      "step": 24681
    },
    {
      "epoch": 7.793495421534575,
      "grad_norm": 6.856391429901123,
      "learning_rate": 3.390695716240396e-05,
      "loss": 0.5034,
      "step": 24682
    },
    {
      "epoch": 7.793811177770761,
      "grad_norm": 5.684432506561279,
      "learning_rate": 3.3906080061747884e-05,
      "loss": 0.5213,
      "step": 24683
    },
    {
      "epoch": 7.794126934006947,
      "grad_norm": 6.958639621734619,
      "learning_rate": 3.3905202961091816e-05,
      "loss": 0.5025,
      "step": 24684
    },
    {
      "epoch": 7.794442690243132,
      "grad_norm": 5.301943302154541,
      "learning_rate": 3.390432586043574e-05,
      "loss": 0.497,
      "step": 24685
    },
    {
      "epoch": 7.794758446479318,
      "grad_norm": 4.241085529327393,
      "learning_rate": 3.390344875977967e-05,
      "loss": 0.3941,
      "step": 24686
    },
    {
      "epoch": 7.7950742027155036,
      "grad_norm": 7.330477237701416,
      "learning_rate": 3.3902571659123604e-05,
      "loss": 0.4692,
      "step": 24687
    },
    {
      "epoch": 7.79538995895169,
      "grad_norm": 7.368915557861328,
      "learning_rate": 3.3901694558467536e-05,
      "loss": 0.5514,
      "step": 24688
    },
    {
      "epoch": 7.795705715187875,
      "grad_norm": 5.952671527862549,
      "learning_rate": 3.390081745781146e-05,
      "loss": 0.544,
      "step": 24689
    },
    {
      "epoch": 7.79602147142406,
      "grad_norm": 5.5764336585998535,
      "learning_rate": 3.389994035715539e-05,
      "loss": 0.5189,
      "step": 24690
    },
    {
      "epoch": 7.7963372276602465,
      "grad_norm": 5.712540149688721,
      "learning_rate": 3.389906325649932e-05,
      "loss": 0.4996,
      "step": 24691
    },
    {
      "epoch": 7.796652983896432,
      "grad_norm": 5.486278057098389,
      "learning_rate": 3.389818615584325e-05,
      "loss": 0.4872,
      "step": 24692
    },
    {
      "epoch": 7.796968740132618,
      "grad_norm": 4.011038780212402,
      "learning_rate": 3.3897309055187174e-05,
      "loss": 0.4513,
      "step": 24693
    },
    {
      "epoch": 7.797284496368803,
      "grad_norm": 4.370597839355469,
      "learning_rate": 3.3896431954531106e-05,
      "loss": 0.3366,
      "step": 24694
    },
    {
      "epoch": 7.797600252604989,
      "grad_norm": 5.022006034851074,
      "learning_rate": 3.389555485387503e-05,
      "loss": 0.4997,
      "step": 24695
    },
    {
      "epoch": 7.797916008841175,
      "grad_norm": 11.172831535339355,
      "learning_rate": 3.389467775321896e-05,
      "loss": 0.5586,
      "step": 24696
    },
    {
      "epoch": 7.79823176507736,
      "grad_norm": 5.3118791580200195,
      "learning_rate": 3.389380065256289e-05,
      "loss": 0.5631,
      "step": 24697
    },
    {
      "epoch": 7.798547521313546,
      "grad_norm": 5.783919334411621,
      "learning_rate": 3.389292355190682e-05,
      "loss": 0.4576,
      "step": 24698
    },
    {
      "epoch": 7.7988632775497315,
      "grad_norm": 4.630356311798096,
      "learning_rate": 3.389204645125075e-05,
      "loss": 0.4438,
      "step": 24699
    },
    {
      "epoch": 7.799179033785917,
      "grad_norm": 5.198412895202637,
      "learning_rate": 3.3891169350594676e-05,
      "loss": 0.5972,
      "step": 24700
    },
    {
      "epoch": 7.799494790022103,
      "grad_norm": 5.913772106170654,
      "learning_rate": 3.389029224993861e-05,
      "loss": 0.4845,
      "step": 24701
    },
    {
      "epoch": 7.799810546258288,
      "grad_norm": 5.463473796844482,
      "learning_rate": 3.388941514928253e-05,
      "loss": 0.4953,
      "step": 24702
    },
    {
      "epoch": 7.8001263024944745,
      "grad_norm": 6.432437419891357,
      "learning_rate": 3.3888538048626464e-05,
      "loss": 0.5328,
      "step": 24703
    },
    {
      "epoch": 7.80044205873066,
      "grad_norm": 4.4138665199279785,
      "learning_rate": 3.388766094797039e-05,
      "loss": 0.4786,
      "step": 24704
    },
    {
      "epoch": 7.800757814966846,
      "grad_norm": 6.855269908905029,
      "learning_rate": 3.388678384731432e-05,
      "loss": 0.5839,
      "step": 24705
    },
    {
      "epoch": 7.801073571203031,
      "grad_norm": 5.444900035858154,
      "learning_rate": 3.3885906746658245e-05,
      "loss": 0.4559,
      "step": 24706
    },
    {
      "epoch": 7.8013893274392165,
      "grad_norm": 5.360102653503418,
      "learning_rate": 3.388502964600218e-05,
      "loss": 0.481,
      "step": 24707
    },
    {
      "epoch": 7.801705083675403,
      "grad_norm": 5.459222793579102,
      "learning_rate": 3.38841525453461e-05,
      "loss": 0.4616,
      "step": 24708
    },
    {
      "epoch": 7.802020839911588,
      "grad_norm": 5.778487682342529,
      "learning_rate": 3.3883275444690034e-05,
      "loss": 0.4811,
      "step": 24709
    },
    {
      "epoch": 7.802336596147774,
      "grad_norm": 6.805906295776367,
      "learning_rate": 3.388239834403396e-05,
      "loss": 0.5557,
      "step": 24710
    },
    {
      "epoch": 7.8026523523839595,
      "grad_norm": 5.529179573059082,
      "learning_rate": 3.388152124337789e-05,
      "loss": 0.3974,
      "step": 24711
    },
    {
      "epoch": 7.802968108620146,
      "grad_norm": 5.4365763664245605,
      "learning_rate": 3.3880644142721815e-05,
      "loss": 0.4482,
      "step": 24712
    },
    {
      "epoch": 7.803283864856331,
      "grad_norm": 6.915795803070068,
      "learning_rate": 3.387976704206575e-05,
      "loss": 0.5287,
      "step": 24713
    },
    {
      "epoch": 7.803599621092516,
      "grad_norm": 6.771910190582275,
      "learning_rate": 3.387888994140968e-05,
      "loss": 0.662,
      "step": 24714
    },
    {
      "epoch": 7.803915377328702,
      "grad_norm": 5.474401950836182,
      "learning_rate": 3.387801284075361e-05,
      "loss": 0.4165,
      "step": 24715
    },
    {
      "epoch": 7.804231133564888,
      "grad_norm": 5.937501907348633,
      "learning_rate": 3.3877135740097535e-05,
      "loss": 0.5078,
      "step": 24716
    },
    {
      "epoch": 7.804546889801074,
      "grad_norm": 4.740969657897949,
      "learning_rate": 3.387625863944147e-05,
      "loss": 0.4109,
      "step": 24717
    },
    {
      "epoch": 7.804862646037259,
      "grad_norm": 5.214254856109619,
      "learning_rate": 3.387538153878539e-05,
      "loss": 0.4518,
      "step": 24718
    },
    {
      "epoch": 7.8051784022734445,
      "grad_norm": 4.9680280685424805,
      "learning_rate": 3.387450443812932e-05,
      "loss": 0.4638,
      "step": 24719
    },
    {
      "epoch": 7.805494158509631,
      "grad_norm": 5.4902873039245605,
      "learning_rate": 3.387362733747325e-05,
      "loss": 0.5805,
      "step": 24720
    },
    {
      "epoch": 7.805809914745816,
      "grad_norm": 5.9207305908203125,
      "learning_rate": 3.387275023681718e-05,
      "loss": 0.3592,
      "step": 24721
    },
    {
      "epoch": 7.806125670982002,
      "grad_norm": 5.404660701751709,
      "learning_rate": 3.387187313616111e-05,
      "loss": 0.4938,
      "step": 24722
    },
    {
      "epoch": 7.8064414272181875,
      "grad_norm": 5.031489372253418,
      "learning_rate": 3.3870996035505037e-05,
      "loss": 0.4382,
      "step": 24723
    },
    {
      "epoch": 7.806757183454373,
      "grad_norm": 5.622481346130371,
      "learning_rate": 3.387011893484897e-05,
      "loss": 0.5907,
      "step": 24724
    },
    {
      "epoch": 7.807072939690559,
      "grad_norm": 5.017211437225342,
      "learning_rate": 3.386924183419289e-05,
      "loss": 0.4571,
      "step": 24725
    },
    {
      "epoch": 7.807388695926744,
      "grad_norm": 7.583398342132568,
      "learning_rate": 3.3868364733536825e-05,
      "loss": 0.6053,
      "step": 24726
    },
    {
      "epoch": 7.80770445216293,
      "grad_norm": 8.136631965637207,
      "learning_rate": 3.386748763288075e-05,
      "loss": 0.4242,
      "step": 24727
    },
    {
      "epoch": 7.808020208399116,
      "grad_norm": 5.453639984130859,
      "learning_rate": 3.386661053222468e-05,
      "loss": 0.447,
      "step": 24728
    },
    {
      "epoch": 7.808335964635302,
      "grad_norm": 5.127136707305908,
      "learning_rate": 3.3865733431568606e-05,
      "loss": 0.5899,
      "step": 24729
    },
    {
      "epoch": 7.808651720871487,
      "grad_norm": 4.671282768249512,
      "learning_rate": 3.386485633091254e-05,
      "loss": 0.4036,
      "step": 24730
    },
    {
      "epoch": 7.808967477107673,
      "grad_norm": 5.284394264221191,
      "learning_rate": 3.386397923025646e-05,
      "loss": 0.4997,
      "step": 24731
    },
    {
      "epoch": 7.809283233343859,
      "grad_norm": 7.2989726066589355,
      "learning_rate": 3.3863102129600395e-05,
      "loss": 0.5522,
      "step": 24732
    },
    {
      "epoch": 7.809598989580044,
      "grad_norm": 6.341584205627441,
      "learning_rate": 3.386222502894432e-05,
      "loss": 0.4679,
      "step": 24733
    },
    {
      "epoch": 7.80991474581623,
      "grad_norm": 8.70380973815918,
      "learning_rate": 3.386134792828825e-05,
      "loss": 0.4886,
      "step": 24734
    },
    {
      "epoch": 7.810230502052415,
      "grad_norm": 6.125075340270996,
      "learning_rate": 3.3860470827632176e-05,
      "loss": 0.4688,
      "step": 24735
    },
    {
      "epoch": 7.810546258288602,
      "grad_norm": 7.233050346374512,
      "learning_rate": 3.385959372697611e-05,
      "loss": 0.5916,
      "step": 24736
    },
    {
      "epoch": 7.810862014524787,
      "grad_norm": 6.130439758300781,
      "learning_rate": 3.385871662632004e-05,
      "loss": 0.5723,
      "step": 24737
    },
    {
      "epoch": 7.811177770760972,
      "grad_norm": 6.494647979736328,
      "learning_rate": 3.3857839525663964e-05,
      "loss": 0.4929,
      "step": 24738
    },
    {
      "epoch": 7.811493526997158,
      "grad_norm": 5.89881706237793,
      "learning_rate": 3.3856962425007896e-05,
      "loss": 0.5314,
      "step": 24739
    },
    {
      "epoch": 7.811809283233344,
      "grad_norm": 6.280475616455078,
      "learning_rate": 3.385608532435182e-05,
      "loss": 0.6199,
      "step": 24740
    },
    {
      "epoch": 7.81212503946953,
      "grad_norm": 6.834736347198486,
      "learning_rate": 3.385520822369575e-05,
      "loss": 0.5157,
      "step": 24741
    },
    {
      "epoch": 7.812440795705715,
      "grad_norm": 10.442535400390625,
      "learning_rate": 3.3854331123039684e-05,
      "loss": 0.4358,
      "step": 24742
    },
    {
      "epoch": 7.8127565519419,
      "grad_norm": 7.053411960601807,
      "learning_rate": 3.3853454022383616e-05,
      "loss": 0.5672,
      "step": 24743
    },
    {
      "epoch": 7.813072308178087,
      "grad_norm": 4.860915184020996,
      "learning_rate": 3.385257692172754e-05,
      "loss": 0.4699,
      "step": 24744
    },
    {
      "epoch": 7.813388064414272,
      "grad_norm": 6.6999311447143555,
      "learning_rate": 3.385169982107147e-05,
      "loss": 0.484,
      "step": 24745
    },
    {
      "epoch": 7.813703820650458,
      "grad_norm": 6.537802219390869,
      "learning_rate": 3.38508227204154e-05,
      "loss": 0.4646,
      "step": 24746
    },
    {
      "epoch": 7.814019576886643,
      "grad_norm": 6.630086898803711,
      "learning_rate": 3.384994561975933e-05,
      "loss": 0.5957,
      "step": 24747
    },
    {
      "epoch": 7.814335333122829,
      "grad_norm": 6.560672760009766,
      "learning_rate": 3.3849068519103254e-05,
      "loss": 0.5983,
      "step": 24748
    },
    {
      "epoch": 7.814651089359015,
      "grad_norm": 5.019637584686279,
      "learning_rate": 3.3848191418447186e-05,
      "loss": 0.4759,
      "step": 24749
    },
    {
      "epoch": 7.8149668455952,
      "grad_norm": 5.265598773956299,
      "learning_rate": 3.384731431779111e-05,
      "loss": 0.476,
      "step": 24750
    },
    {
      "epoch": 7.815282601831386,
      "grad_norm": 6.268020153045654,
      "learning_rate": 3.384643721713504e-05,
      "loss": 0.5218,
      "step": 24751
    },
    {
      "epoch": 7.815598358067572,
      "grad_norm": 5.29644250869751,
      "learning_rate": 3.384556011647897e-05,
      "loss": 0.4924,
      "step": 24752
    },
    {
      "epoch": 7.815914114303758,
      "grad_norm": 3.841235399246216,
      "learning_rate": 3.38446830158229e-05,
      "loss": 0.4119,
      "step": 24753
    },
    {
      "epoch": 7.816229870539943,
      "grad_norm": 4.178685665130615,
      "learning_rate": 3.3843805915166824e-05,
      "loss": 0.396,
      "step": 24754
    },
    {
      "epoch": 7.816545626776129,
      "grad_norm": 5.398375511169434,
      "learning_rate": 3.3842928814510755e-05,
      "loss": 0.4722,
      "step": 24755
    },
    {
      "epoch": 7.816861383012315,
      "grad_norm": 6.232877731323242,
      "learning_rate": 3.384205171385468e-05,
      "loss": 0.4804,
      "step": 24756
    },
    {
      "epoch": 7.8171771392485,
      "grad_norm": 5.494775295257568,
      "learning_rate": 3.384117461319861e-05,
      "loss": 0.5422,
      "step": 24757
    },
    {
      "epoch": 7.817492895484686,
      "grad_norm": 5.416357517242432,
      "learning_rate": 3.384029751254254e-05,
      "loss": 0.5199,
      "step": 24758
    },
    {
      "epoch": 7.817808651720871,
      "grad_norm": 5.479986190795898,
      "learning_rate": 3.383942041188647e-05,
      "loss": 0.4412,
      "step": 24759
    },
    {
      "epoch": 7.8181244079570575,
      "grad_norm": 5.694607257843018,
      "learning_rate": 3.38385433112304e-05,
      "loss": 0.5779,
      "step": 24760
    },
    {
      "epoch": 7.818440164193243,
      "grad_norm": 5.1116251945495605,
      "learning_rate": 3.3837666210574325e-05,
      "loss": 0.4701,
      "step": 24761
    },
    {
      "epoch": 7.818755920429428,
      "grad_norm": 7.088943958282471,
      "learning_rate": 3.383678910991826e-05,
      "loss": 0.5552,
      "step": 24762
    },
    {
      "epoch": 7.819071676665614,
      "grad_norm": 6.7899346351623535,
      "learning_rate": 3.383591200926218e-05,
      "loss": 0.4966,
      "step": 24763
    },
    {
      "epoch": 7.8193874329018,
      "grad_norm": 5.325791358947754,
      "learning_rate": 3.3835034908606114e-05,
      "loss": 0.5756,
      "step": 24764
    },
    {
      "epoch": 7.819703189137986,
      "grad_norm": 5.763094902038574,
      "learning_rate": 3.383415780795004e-05,
      "loss": 0.4483,
      "step": 24765
    },
    {
      "epoch": 7.820018945374171,
      "grad_norm": 6.225139617919922,
      "learning_rate": 3.383328070729397e-05,
      "loss": 0.4572,
      "step": 24766
    },
    {
      "epoch": 7.820334701610356,
      "grad_norm": 5.428426265716553,
      "learning_rate": 3.3832403606637895e-05,
      "loss": 0.5772,
      "step": 24767
    },
    {
      "epoch": 7.820650457846543,
      "grad_norm": 5.7235589027404785,
      "learning_rate": 3.383152650598183e-05,
      "loss": 0.5605,
      "step": 24768
    },
    {
      "epoch": 7.820966214082728,
      "grad_norm": 5.267340660095215,
      "learning_rate": 3.383064940532576e-05,
      "loss": 0.4767,
      "step": 24769
    },
    {
      "epoch": 7.821281970318914,
      "grad_norm": 5.936751365661621,
      "learning_rate": 3.382977230466969e-05,
      "loss": 0.4715,
      "step": 24770
    },
    {
      "epoch": 7.821597726555099,
      "grad_norm": 4.638950824737549,
      "learning_rate": 3.3828895204013615e-05,
      "loss": 0.4144,
      "step": 24771
    },
    {
      "epoch": 7.8219134827912855,
      "grad_norm": 5.364846706390381,
      "learning_rate": 3.382801810335755e-05,
      "loss": 0.3563,
      "step": 24772
    },
    {
      "epoch": 7.822229239027471,
      "grad_norm": 6.599632740020752,
      "learning_rate": 3.382714100270147e-05,
      "loss": 0.5007,
      "step": 24773
    },
    {
      "epoch": 7.822544995263656,
      "grad_norm": 6.218525409698486,
      "learning_rate": 3.38262639020454e-05,
      "loss": 0.513,
      "step": 24774
    },
    {
      "epoch": 7.822860751499842,
      "grad_norm": 6.352238178253174,
      "learning_rate": 3.382538680138933e-05,
      "loss": 0.4454,
      "step": 24775
    },
    {
      "epoch": 7.823176507736028,
      "grad_norm": 4.958086013793945,
      "learning_rate": 3.382450970073326e-05,
      "loss": 0.4991,
      "step": 24776
    },
    {
      "epoch": 7.823492263972214,
      "grad_norm": 6.516744136810303,
      "learning_rate": 3.3823632600077185e-05,
      "loss": 0.5787,
      "step": 24777
    },
    {
      "epoch": 7.823808020208399,
      "grad_norm": 6.364917755126953,
      "learning_rate": 3.3822755499421116e-05,
      "loss": 0.4459,
      "step": 24778
    },
    {
      "epoch": 7.824123776444585,
      "grad_norm": 5.958503723144531,
      "learning_rate": 3.382187839876504e-05,
      "loss": 0.525,
      "step": 24779
    },
    {
      "epoch": 7.8244395326807705,
      "grad_norm": 5.841684341430664,
      "learning_rate": 3.382100129810897e-05,
      "loss": 0.558,
      "step": 24780
    },
    {
      "epoch": 7.824755288916956,
      "grad_norm": 4.761292457580566,
      "learning_rate": 3.3820124197452905e-05,
      "loss": 0.4982,
      "step": 24781
    },
    {
      "epoch": 7.825071045153142,
      "grad_norm": 4.744578838348389,
      "learning_rate": 3.381924709679683e-05,
      "loss": 0.3377,
      "step": 24782
    },
    {
      "epoch": 7.825386801389327,
      "grad_norm": 7.077661991119385,
      "learning_rate": 3.381836999614076e-05,
      "loss": 0.4929,
      "step": 24783
    },
    {
      "epoch": 7.8257025576255135,
      "grad_norm": 5.694118976593018,
      "learning_rate": 3.3817492895484686e-05,
      "loss": 0.5605,
      "step": 24784
    },
    {
      "epoch": 7.826018313861699,
      "grad_norm": 4.9689040184021,
      "learning_rate": 3.381661579482862e-05,
      "loss": 0.4001,
      "step": 24785
    },
    {
      "epoch": 7.826334070097884,
      "grad_norm": 7.0207037925720215,
      "learning_rate": 3.381573869417254e-05,
      "loss": 0.5274,
      "step": 24786
    },
    {
      "epoch": 7.82664982633407,
      "grad_norm": 5.955965042114258,
      "learning_rate": 3.3814861593516474e-05,
      "loss": 0.5138,
      "step": 24787
    },
    {
      "epoch": 7.8269655825702555,
      "grad_norm": 6.633127689361572,
      "learning_rate": 3.38139844928604e-05,
      "loss": 0.5574,
      "step": 24788
    },
    {
      "epoch": 7.827281338806442,
      "grad_norm": 5.480309009552002,
      "learning_rate": 3.381310739220433e-05,
      "loss": 0.4666,
      "step": 24789
    },
    {
      "epoch": 7.827597095042627,
      "grad_norm": 5.211432456970215,
      "learning_rate": 3.3812230291548256e-05,
      "loss": 0.4993,
      "step": 24790
    },
    {
      "epoch": 7.827912851278812,
      "grad_norm": 5.221668720245361,
      "learning_rate": 3.381135319089219e-05,
      "loss": 0.5211,
      "step": 24791
    },
    {
      "epoch": 7.8282286075149985,
      "grad_norm": 5.549527168273926,
      "learning_rate": 3.381047609023611e-05,
      "loss": 0.514,
      "step": 24792
    },
    {
      "epoch": 7.828544363751184,
      "grad_norm": 7.008903980255127,
      "learning_rate": 3.3809598989580044e-05,
      "loss": 0.5489,
      "step": 24793
    },
    {
      "epoch": 7.82886011998737,
      "grad_norm": 5.596511363983154,
      "learning_rate": 3.380872188892397e-05,
      "loss": 0.4587,
      "step": 24794
    },
    {
      "epoch": 7.829175876223555,
      "grad_norm": 5.297616004943848,
      "learning_rate": 3.38078447882679e-05,
      "loss": 0.5561,
      "step": 24795
    },
    {
      "epoch": 7.8294916324597414,
      "grad_norm": 4.647331237792969,
      "learning_rate": 3.380696768761183e-05,
      "loss": 0.385,
      "step": 24796
    },
    {
      "epoch": 7.829807388695927,
      "grad_norm": 6.055208683013916,
      "learning_rate": 3.3806090586955764e-05,
      "loss": 0.4512,
      "step": 24797
    },
    {
      "epoch": 7.830123144932112,
      "grad_norm": 5.020087242126465,
      "learning_rate": 3.380521348629969e-05,
      "loss": 0.4859,
      "step": 24798
    },
    {
      "epoch": 7.830438901168298,
      "grad_norm": 5.329738616943359,
      "learning_rate": 3.380433638564362e-05,
      "loss": 0.4297,
      "step": 24799
    },
    {
      "epoch": 7.8307546574044835,
      "grad_norm": 5.850117206573486,
      "learning_rate": 3.3803459284987546e-05,
      "loss": 0.4886,
      "step": 24800
    },
    {
      "epoch": 7.8307546574044835,
      "eval_loss": 0.9271837472915649,
      "eval_runtime": 4.5619,
      "eval_samples_per_second": 545.824,
      "eval_steps_per_second": 4.603,
      "step": 24800
    },
    {
      "epoch": 7.83107041364067,
      "grad_norm": 4.975038051605225,
      "learning_rate": 3.380258218433148e-05,
      "loss": 0.4231,
      "step": 24801
    },
    {
      "epoch": 7.831386169876855,
      "grad_norm": 6.849006652832031,
      "learning_rate": 3.380170508367541e-05,
      "loss": 0.5006,
      "step": 24802
    },
    {
      "epoch": 7.831701926113041,
      "grad_norm": 4.935678482055664,
      "learning_rate": 3.3800827983019334e-05,
      "loss": 0.4978,
      "step": 24803
    },
    {
      "epoch": 7.8320176823492265,
      "grad_norm": 5.186914443969727,
      "learning_rate": 3.3799950882363266e-05,
      "loss": 0.5664,
      "step": 24804
    },
    {
      "epoch": 7.832333438585412,
      "grad_norm": 6.171753406524658,
      "learning_rate": 3.379907378170719e-05,
      "loss": 0.4581,
      "step": 24805
    },
    {
      "epoch": 7.832649194821598,
      "grad_norm": 4.7191290855407715,
      "learning_rate": 3.379819668105112e-05,
      "loss": 0.4395,
      "step": 24806
    },
    {
      "epoch": 7.832964951057783,
      "grad_norm": 4.864372253417969,
      "learning_rate": 3.379731958039505e-05,
      "loss": 0.4632,
      "step": 24807
    },
    {
      "epoch": 7.833280707293969,
      "grad_norm": 5.1694512367248535,
      "learning_rate": 3.379644247973898e-05,
      "loss": 0.422,
      "step": 24808
    },
    {
      "epoch": 7.833596463530155,
      "grad_norm": 6.486904621124268,
      "learning_rate": 3.3795565379082904e-05,
      "loss": 0.6227,
      "step": 24809
    },
    {
      "epoch": 7.83391221976634,
      "grad_norm": 6.141509056091309,
      "learning_rate": 3.3794688278426835e-05,
      "loss": 0.3902,
      "step": 24810
    },
    {
      "epoch": 7.834227976002526,
      "grad_norm": 6.268927574157715,
      "learning_rate": 3.379381117777076e-05,
      "loss": 0.5598,
      "step": 24811
    },
    {
      "epoch": 7.8345437322387115,
      "grad_norm": 5.7303595542907715,
      "learning_rate": 3.379293407711469e-05,
      "loss": 0.4119,
      "step": 24812
    },
    {
      "epoch": 7.834859488474898,
      "grad_norm": 7.066562652587891,
      "learning_rate": 3.379205697645862e-05,
      "loss": 0.5901,
      "step": 24813
    },
    {
      "epoch": 7.835175244711083,
      "grad_norm": 6.961780071258545,
      "learning_rate": 3.379117987580255e-05,
      "loss": 0.5664,
      "step": 24814
    },
    {
      "epoch": 7.835491000947268,
      "grad_norm": 5.660167694091797,
      "learning_rate": 3.3790302775146473e-05,
      "loss": 0.4469,
      "step": 24815
    },
    {
      "epoch": 7.835806757183454,
      "grad_norm": 6.363842487335205,
      "learning_rate": 3.3789425674490405e-05,
      "loss": 0.5567,
      "step": 24816
    },
    {
      "epoch": 7.83612251341964,
      "grad_norm": 4.5658183097839355,
      "learning_rate": 3.378854857383433e-05,
      "loss": 0.3656,
      "step": 24817
    },
    {
      "epoch": 7.836438269655826,
      "grad_norm": 6.04235315322876,
      "learning_rate": 3.378767147317826e-05,
      "loss": 0.5802,
      "step": 24818
    },
    {
      "epoch": 7.836754025892011,
      "grad_norm": 6.725184440612793,
      "learning_rate": 3.3786794372522193e-05,
      "loss": 0.4639,
      "step": 24819
    },
    {
      "epoch": 7.837069782128197,
      "grad_norm": 5.042815208435059,
      "learning_rate": 3.378591727186612e-05,
      "loss": 0.4671,
      "step": 24820
    },
    {
      "epoch": 7.837385538364383,
      "grad_norm": 6.334475517272949,
      "learning_rate": 3.378504017121005e-05,
      "loss": 0.5129,
      "step": 24821
    },
    {
      "epoch": 7.837701294600569,
      "grad_norm": 8.027141571044922,
      "learning_rate": 3.3784163070553975e-05,
      "loss": 0.586,
      "step": 24822
    },
    {
      "epoch": 7.838017050836754,
      "grad_norm": 6.117138385772705,
      "learning_rate": 3.3783285969897907e-05,
      "loss": 0.4781,
      "step": 24823
    },
    {
      "epoch": 7.8383328070729394,
      "grad_norm": 5.664443492889404,
      "learning_rate": 3.378240886924184e-05,
      "loss": 0.5254,
      "step": 24824
    },
    {
      "epoch": 7.838648563309126,
      "grad_norm": 5.573575019836426,
      "learning_rate": 3.378153176858577e-05,
      "loss": 0.4646,
      "step": 24825
    },
    {
      "epoch": 7.838964319545311,
      "grad_norm": 4.994751930236816,
      "learning_rate": 3.3780654667929695e-05,
      "loss": 0.4274,
      "step": 24826
    },
    {
      "epoch": 7.839280075781497,
      "grad_norm": 5.463135242462158,
      "learning_rate": 3.3779777567273627e-05,
      "loss": 0.4724,
      "step": 24827
    },
    {
      "epoch": 7.839595832017682,
      "grad_norm": 5.733376502990723,
      "learning_rate": 3.377890046661755e-05,
      "loss": 0.5023,
      "step": 24828
    },
    {
      "epoch": 7.839911588253868,
      "grad_norm": 4.184042930603027,
      "learning_rate": 3.377802336596148e-05,
      "loss": 0.427,
      "step": 24829
    },
    {
      "epoch": 7.840227344490054,
      "grad_norm": 5.200512409210205,
      "learning_rate": 3.377714626530541e-05,
      "loss": 0.5177,
      "step": 24830
    },
    {
      "epoch": 7.840543100726239,
      "grad_norm": 5.994185924530029,
      "learning_rate": 3.377626916464934e-05,
      "loss": 0.5102,
      "step": 24831
    },
    {
      "epoch": 7.840858856962425,
      "grad_norm": 5.492584705352783,
      "learning_rate": 3.3775392063993265e-05,
      "loss": 0.4277,
      "step": 24832
    },
    {
      "epoch": 7.841174613198611,
      "grad_norm": 5.044182300567627,
      "learning_rate": 3.3774514963337196e-05,
      "loss": 0.405,
      "step": 24833
    },
    {
      "epoch": 7.841490369434796,
      "grad_norm": 5.985719680786133,
      "learning_rate": 3.377363786268112e-05,
      "loss": 0.4798,
      "step": 24834
    },
    {
      "epoch": 7.841806125670982,
      "grad_norm": 5.396574020385742,
      "learning_rate": 3.377276076202505e-05,
      "loss": 0.594,
      "step": 24835
    },
    {
      "epoch": 7.842121881907167,
      "grad_norm": 5.987430095672607,
      "learning_rate": 3.377188366136898e-05,
      "loss": 0.5188,
      "step": 24836
    },
    {
      "epoch": 7.842437638143354,
      "grad_norm": 5.249615669250488,
      "learning_rate": 3.377100656071291e-05,
      "loss": 0.5141,
      "step": 24837
    },
    {
      "epoch": 7.842753394379539,
      "grad_norm": 5.4259934425354,
      "learning_rate": 3.3770129460056834e-05,
      "loss": 0.4952,
      "step": 24838
    },
    {
      "epoch": 7.843069150615724,
      "grad_norm": 5.0027666091918945,
      "learning_rate": 3.3769252359400766e-05,
      "loss": 0.4653,
      "step": 24839
    },
    {
      "epoch": 7.84338490685191,
      "grad_norm": 6.306177616119385,
      "learning_rate": 3.37683752587447e-05,
      "loss": 0.4863,
      "step": 24840
    },
    {
      "epoch": 7.843700663088096,
      "grad_norm": 4.819715976715088,
      "learning_rate": 3.376749815808862e-05,
      "loss": 0.4435,
      "step": 24841
    },
    {
      "epoch": 7.844016419324282,
      "grad_norm": 3.808598041534424,
      "learning_rate": 3.3766621057432554e-05,
      "loss": 0.38,
      "step": 24842
    },
    {
      "epoch": 7.844332175560467,
      "grad_norm": 5.038935661315918,
      "learning_rate": 3.376574395677648e-05,
      "loss": 0.4949,
      "step": 24843
    },
    {
      "epoch": 7.844647931796653,
      "grad_norm": 3.821040153503418,
      "learning_rate": 3.376486685612041e-05,
      "loss": 0.2895,
      "step": 24844
    },
    {
      "epoch": 7.844963688032839,
      "grad_norm": 5.725983142852783,
      "learning_rate": 3.3763989755464336e-05,
      "loss": 0.5158,
      "step": 24845
    },
    {
      "epoch": 7.845279444269025,
      "grad_norm": 6.071750640869141,
      "learning_rate": 3.376311265480827e-05,
      "loss": 0.5199,
      "step": 24846
    },
    {
      "epoch": 7.84559520050521,
      "grad_norm": 5.3569111824035645,
      "learning_rate": 3.376223555415219e-05,
      "loss": 0.5526,
      "step": 24847
    },
    {
      "epoch": 7.845910956741395,
      "grad_norm": 6.350406169891357,
      "learning_rate": 3.3761358453496124e-05,
      "loss": 0.5564,
      "step": 24848
    },
    {
      "epoch": 7.846226712977582,
      "grad_norm": 5.648291110992432,
      "learning_rate": 3.376048135284005e-05,
      "loss": 0.5686,
      "step": 24849
    },
    {
      "epoch": 7.846542469213767,
      "grad_norm": 4.654673099517822,
      "learning_rate": 3.375960425218398e-05,
      "loss": 0.4743,
      "step": 24850
    },
    {
      "epoch": 7.846858225449953,
      "grad_norm": 6.110404968261719,
      "learning_rate": 3.375872715152791e-05,
      "loss": 0.459,
      "step": 24851
    },
    {
      "epoch": 7.847173981686138,
      "grad_norm": 4.893057823181152,
      "learning_rate": 3.3757850050871844e-05,
      "loss": 0.4618,
      "step": 24852
    },
    {
      "epoch": 7.847489737922324,
      "grad_norm": 6.347972393035889,
      "learning_rate": 3.375697295021577e-05,
      "loss": 0.5038,
      "step": 24853
    },
    {
      "epoch": 7.84780549415851,
      "grad_norm": 5.7932233810424805,
      "learning_rate": 3.37560958495597e-05,
      "loss": 0.452,
      "step": 24854
    },
    {
      "epoch": 7.848121250394695,
      "grad_norm": 6.645558834075928,
      "learning_rate": 3.3755218748903626e-05,
      "loss": 0.5931,
      "step": 24855
    },
    {
      "epoch": 7.848437006630881,
      "grad_norm": 5.272090435028076,
      "learning_rate": 3.375434164824756e-05,
      "loss": 0.5027,
      "step": 24856
    },
    {
      "epoch": 7.848752762867067,
      "grad_norm": 6.275309085845947,
      "learning_rate": 3.375346454759148e-05,
      "loss": 0.4284,
      "step": 24857
    },
    {
      "epoch": 7.849068519103252,
      "grad_norm": 4.83613395690918,
      "learning_rate": 3.3752587446935414e-05,
      "loss": 0.4289,
      "step": 24858
    },
    {
      "epoch": 7.849384275339438,
      "grad_norm": 5.5248589515686035,
      "learning_rate": 3.375171034627934e-05,
      "loss": 0.5074,
      "step": 24859
    },
    {
      "epoch": 7.849700031575623,
      "grad_norm": 6.28485631942749,
      "learning_rate": 3.375083324562327e-05,
      "loss": 0.5301,
      "step": 24860
    },
    {
      "epoch": 7.8500157878118095,
      "grad_norm": 5.7042999267578125,
      "learning_rate": 3.3749956144967195e-05,
      "loss": 0.4635,
      "step": 24861
    },
    {
      "epoch": 7.850331544047995,
      "grad_norm": 4.692979335784912,
      "learning_rate": 3.374907904431113e-05,
      "loss": 0.4688,
      "step": 24862
    },
    {
      "epoch": 7.85064730028418,
      "grad_norm": 3.8760147094726562,
      "learning_rate": 3.374820194365506e-05,
      "loss": 0.4233,
      "step": 24863
    },
    {
      "epoch": 7.850963056520366,
      "grad_norm": 7.569521427154541,
      "learning_rate": 3.3747324842998984e-05,
      "loss": 0.6329,
      "step": 24864
    },
    {
      "epoch": 7.851278812756552,
      "grad_norm": 7.895748615264893,
      "learning_rate": 3.3746447742342915e-05,
      "loss": 0.5414,
      "step": 24865
    },
    {
      "epoch": 7.851594568992738,
      "grad_norm": 5.648393154144287,
      "learning_rate": 3.374557064168684e-05,
      "loss": 0.564,
      "step": 24866
    },
    {
      "epoch": 7.851910325228923,
      "grad_norm": 5.595354080200195,
      "learning_rate": 3.374469354103077e-05,
      "loss": 0.4238,
      "step": 24867
    },
    {
      "epoch": 7.852226081465109,
      "grad_norm": 5.76796817779541,
      "learning_rate": 3.37438164403747e-05,
      "loss": 0.4995,
      "step": 24868
    },
    {
      "epoch": 7.8525418377012945,
      "grad_norm": 5.502867698669434,
      "learning_rate": 3.374293933971863e-05,
      "loss": 0.499,
      "step": 24869
    },
    {
      "epoch": 7.852857593937481,
      "grad_norm": 11.176486015319824,
      "learning_rate": 3.374206223906255e-05,
      "loss": 0.5272,
      "step": 24870
    },
    {
      "epoch": 7.853173350173666,
      "grad_norm": 5.456272125244141,
      "learning_rate": 3.3741185138406485e-05,
      "loss": 0.5873,
      "step": 24871
    },
    {
      "epoch": 7.853489106409851,
      "grad_norm": 5.149500370025635,
      "learning_rate": 3.374030803775041e-05,
      "loss": 0.4663,
      "step": 24872
    },
    {
      "epoch": 7.8538048626460375,
      "grad_norm": 5.501880168914795,
      "learning_rate": 3.373943093709434e-05,
      "loss": 0.5225,
      "step": 24873
    },
    {
      "epoch": 7.854120618882223,
      "grad_norm": 6.543102741241455,
      "learning_rate": 3.3738553836438267e-05,
      "loss": 0.6159,
      "step": 24874
    },
    {
      "epoch": 7.854436375118409,
      "grad_norm": 5.171684741973877,
      "learning_rate": 3.37376767357822e-05,
      "loss": 0.5165,
      "step": 24875
    },
    {
      "epoch": 7.854752131354594,
      "grad_norm": 5.449423789978027,
      "learning_rate": 3.373679963512612e-05,
      "loss": 0.5779,
      "step": 24876
    },
    {
      "epoch": 7.85506788759078,
      "grad_norm": 5.589585781097412,
      "learning_rate": 3.3735922534470055e-05,
      "loss": 0.3807,
      "step": 24877
    },
    {
      "epoch": 7.855383643826966,
      "grad_norm": 5.983016490936279,
      "learning_rate": 3.3735045433813986e-05,
      "loss": 0.5313,
      "step": 24878
    },
    {
      "epoch": 7.855699400063151,
      "grad_norm": 5.049971103668213,
      "learning_rate": 3.373416833315792e-05,
      "loss": 0.5031,
      "step": 24879
    },
    {
      "epoch": 7.856015156299337,
      "grad_norm": 5.20465087890625,
      "learning_rate": 3.373329123250184e-05,
      "loss": 0.5146,
      "step": 24880
    },
    {
      "epoch": 7.8563309125355225,
      "grad_norm": 8.939661979675293,
      "learning_rate": 3.3732414131845775e-05,
      "loss": 0.6005,
      "step": 24881
    },
    {
      "epoch": 7.856646668771708,
      "grad_norm": 5.521458625793457,
      "learning_rate": 3.37315370311897e-05,
      "loss": 0.5418,
      "step": 24882
    },
    {
      "epoch": 7.856962425007894,
      "grad_norm": 5.552394866943359,
      "learning_rate": 3.373065993053363e-05,
      "loss": 0.4298,
      "step": 24883
    },
    {
      "epoch": 7.857278181244079,
      "grad_norm": 6.871307373046875,
      "learning_rate": 3.372978282987756e-05,
      "loss": 0.5033,
      "step": 24884
    },
    {
      "epoch": 7.8575939374802655,
      "grad_norm": 4.920716285705566,
      "learning_rate": 3.372890572922149e-05,
      "loss": 0.5331,
      "step": 24885
    },
    {
      "epoch": 7.857909693716451,
      "grad_norm": 6.537604808807373,
      "learning_rate": 3.372802862856542e-05,
      "loss": 0.5333,
      "step": 24886
    },
    {
      "epoch": 7.858225449952637,
      "grad_norm": 5.030524253845215,
      "learning_rate": 3.3727151527909345e-05,
      "loss": 0.5063,
      "step": 24887
    },
    {
      "epoch": 7.858541206188822,
      "grad_norm": 6.026093006134033,
      "learning_rate": 3.3726274427253276e-05,
      "loss": 0.5317,
      "step": 24888
    },
    {
      "epoch": 7.8588569624250075,
      "grad_norm": 6.059146881103516,
      "learning_rate": 3.37253973265972e-05,
      "loss": 0.4684,
      "step": 24889
    },
    {
      "epoch": 7.859172718661194,
      "grad_norm": 6.0551862716674805,
      "learning_rate": 3.372452022594113e-05,
      "loss": 0.5214,
      "step": 24890
    },
    {
      "epoch": 7.859488474897379,
      "grad_norm": 5.941253185272217,
      "learning_rate": 3.372364312528506e-05,
      "loss": 0.5362,
      "step": 24891
    },
    {
      "epoch": 7.859804231133565,
      "grad_norm": 5.062766075134277,
      "learning_rate": 3.372276602462899e-05,
      "loss": 0.4544,
      "step": 24892
    },
    {
      "epoch": 7.8601199873697505,
      "grad_norm": 4.656501770019531,
      "learning_rate": 3.3721888923972914e-05,
      "loss": 0.4914,
      "step": 24893
    },
    {
      "epoch": 7.860435743605937,
      "grad_norm": 5.540139198303223,
      "learning_rate": 3.3721011823316846e-05,
      "loss": 0.5011,
      "step": 24894
    },
    {
      "epoch": 7.860751499842122,
      "grad_norm": 4.649083137512207,
      "learning_rate": 3.372013472266077e-05,
      "loss": 0.5551,
      "step": 24895
    },
    {
      "epoch": 7.861067256078307,
      "grad_norm": 7.196763515472412,
      "learning_rate": 3.37192576220047e-05,
      "loss": 0.5335,
      "step": 24896
    },
    {
      "epoch": 7.861383012314493,
      "grad_norm": 6.354470729827881,
      "learning_rate": 3.371838052134863e-05,
      "loss": 0.5479,
      "step": 24897
    },
    {
      "epoch": 7.861698768550679,
      "grad_norm": 5.366375923156738,
      "learning_rate": 3.371750342069256e-05,
      "loss": 0.5141,
      "step": 24898
    },
    {
      "epoch": 7.862014524786865,
      "grad_norm": 5.375630855560303,
      "learning_rate": 3.3716626320036484e-05,
      "loss": 0.4527,
      "step": 24899
    },
    {
      "epoch": 7.86233028102305,
      "grad_norm": 5.314986228942871,
      "learning_rate": 3.3715749219380416e-05,
      "loss": 0.5568,
      "step": 24900
    },
    {
      "epoch": 7.8626460372592355,
      "grad_norm": 5.003788948059082,
      "learning_rate": 3.371487211872435e-05,
      "loss": 0.4471,
      "step": 24901
    },
    {
      "epoch": 7.862961793495422,
      "grad_norm": 5.28352689743042,
      "learning_rate": 3.371399501806827e-05,
      "loss": 0.497,
      "step": 24902
    },
    {
      "epoch": 7.863277549731607,
      "grad_norm": 5.758516788482666,
      "learning_rate": 3.3713117917412204e-05,
      "loss": 0.5068,
      "step": 24903
    },
    {
      "epoch": 7.863593305967793,
      "grad_norm": 5.496701717376709,
      "learning_rate": 3.371224081675613e-05,
      "loss": 0.6451,
      "step": 24904
    },
    {
      "epoch": 7.8639090622039785,
      "grad_norm": 5.9160308837890625,
      "learning_rate": 3.371136371610006e-05,
      "loss": 0.5059,
      "step": 24905
    },
    {
      "epoch": 7.864224818440164,
      "grad_norm": 4.5844855308532715,
      "learning_rate": 3.371048661544399e-05,
      "loss": 0.4676,
      "step": 24906
    },
    {
      "epoch": 7.86454057467635,
      "grad_norm": 6.986740589141846,
      "learning_rate": 3.3709609514787924e-05,
      "loss": 0.6234,
      "step": 24907
    },
    {
      "epoch": 7.864856330912535,
      "grad_norm": 6.082756042480469,
      "learning_rate": 3.370873241413185e-05,
      "loss": 0.4743,
      "step": 24908
    },
    {
      "epoch": 7.865172087148721,
      "grad_norm": 6.35115385055542,
      "learning_rate": 3.370785531347578e-05,
      "loss": 0.5034,
      "step": 24909
    },
    {
      "epoch": 7.865487843384907,
      "grad_norm": 6.069050312042236,
      "learning_rate": 3.3706978212819705e-05,
      "loss": 0.5056,
      "step": 24910
    },
    {
      "epoch": 7.865803599621093,
      "grad_norm": 4.635684967041016,
      "learning_rate": 3.370610111216364e-05,
      "loss": 0.5726,
      "step": 24911
    },
    {
      "epoch": 7.866119355857278,
      "grad_norm": 5.0902791023254395,
      "learning_rate": 3.370522401150756e-05,
      "loss": 0.6036,
      "step": 24912
    },
    {
      "epoch": 7.8664351120934635,
      "grad_norm": 4.542932510375977,
      "learning_rate": 3.3704346910851494e-05,
      "loss": 0.4665,
      "step": 24913
    },
    {
      "epoch": 7.86675086832965,
      "grad_norm": 4.426031112670898,
      "learning_rate": 3.370346981019542e-05,
      "loss": 0.4558,
      "step": 24914
    },
    {
      "epoch": 7.867066624565835,
      "grad_norm": 4.696286678314209,
      "learning_rate": 3.370259270953935e-05,
      "loss": 0.505,
      "step": 24915
    },
    {
      "epoch": 7.867382380802021,
      "grad_norm": 4.808368682861328,
      "learning_rate": 3.3701715608883275e-05,
      "loss": 0.4631,
      "step": 24916
    },
    {
      "epoch": 7.867698137038206,
      "grad_norm": 5.462060928344727,
      "learning_rate": 3.370083850822721e-05,
      "loss": 0.5738,
      "step": 24917
    },
    {
      "epoch": 7.868013893274393,
      "grad_norm": 4.758191108703613,
      "learning_rate": 3.369996140757113e-05,
      "loss": 0.5065,
      "step": 24918
    },
    {
      "epoch": 7.868329649510578,
      "grad_norm": 6.190764427185059,
      "learning_rate": 3.3699084306915063e-05,
      "loss": 0.486,
      "step": 24919
    },
    {
      "epoch": 7.868645405746763,
      "grad_norm": 5.185075759887695,
      "learning_rate": 3.369820720625899e-05,
      "loss": 0.4265,
      "step": 24920
    },
    {
      "epoch": 7.868961161982949,
      "grad_norm": 6.211611270904541,
      "learning_rate": 3.369733010560292e-05,
      "loss": 0.5337,
      "step": 24921
    },
    {
      "epoch": 7.869276918219135,
      "grad_norm": 6.016597270965576,
      "learning_rate": 3.369645300494685e-05,
      "loss": 0.5676,
      "step": 24922
    },
    {
      "epoch": 7.869592674455321,
      "grad_norm": 4.786180019378662,
      "learning_rate": 3.369557590429078e-05,
      "loss": 0.416,
      "step": 24923
    },
    {
      "epoch": 7.869908430691506,
      "grad_norm": 7.319511890411377,
      "learning_rate": 3.369469880363471e-05,
      "loss": 0.5064,
      "step": 24924
    },
    {
      "epoch": 7.870224186927691,
      "grad_norm": 4.662413120269775,
      "learning_rate": 3.369382170297863e-05,
      "loss": 0.4659,
      "step": 24925
    },
    {
      "epoch": 7.870539943163878,
      "grad_norm": 6.744253635406494,
      "learning_rate": 3.3692944602322565e-05,
      "loss": 0.5055,
      "step": 24926
    },
    {
      "epoch": 7.870855699400063,
      "grad_norm": 6.551508903503418,
      "learning_rate": 3.369206750166649e-05,
      "loss": 0.5777,
      "step": 24927
    },
    {
      "epoch": 7.871171455636249,
      "grad_norm": 7.315989017486572,
      "learning_rate": 3.369119040101042e-05,
      "loss": 0.5856,
      "step": 24928
    },
    {
      "epoch": 7.871487211872434,
      "grad_norm": 6.3915114402771,
      "learning_rate": 3.3690313300354346e-05,
      "loss": 0.5773,
      "step": 24929
    },
    {
      "epoch": 7.87180296810862,
      "grad_norm": 5.440644264221191,
      "learning_rate": 3.368943619969828e-05,
      "loss": 0.5513,
      "step": 24930
    },
    {
      "epoch": 7.872118724344806,
      "grad_norm": 4.894985198974609,
      "learning_rate": 3.36885590990422e-05,
      "loss": 0.4885,
      "step": 24931
    },
    {
      "epoch": 7.872434480580991,
      "grad_norm": 4.429284572601318,
      "learning_rate": 3.368768199838614e-05,
      "loss": 0.3569,
      "step": 24932
    },
    {
      "epoch": 7.872750236817177,
      "grad_norm": 6.7623138427734375,
      "learning_rate": 3.3686804897730066e-05,
      "loss": 0.5648,
      "step": 24933
    },
    {
      "epoch": 7.873065993053363,
      "grad_norm": 5.288488388061523,
      "learning_rate": 3.3685927797074e-05,
      "loss": 0.4062,
      "step": 24934
    },
    {
      "epoch": 7.873381749289549,
      "grad_norm": 6.476733207702637,
      "learning_rate": 3.368505069641792e-05,
      "loss": 0.4782,
      "step": 24935
    },
    {
      "epoch": 7.873697505525734,
      "grad_norm": 4.753860950469971,
      "learning_rate": 3.3684173595761855e-05,
      "loss": 0.6065,
      "step": 24936
    },
    {
      "epoch": 7.87401326176192,
      "grad_norm": 6.3705267906188965,
      "learning_rate": 3.368329649510578e-05,
      "loss": 0.5555,
      "step": 24937
    },
    {
      "epoch": 7.874329017998106,
      "grad_norm": 5.837637424468994,
      "learning_rate": 3.368241939444971e-05,
      "loss": 0.5308,
      "step": 24938
    },
    {
      "epoch": 7.874644774234291,
      "grad_norm": 5.333331108093262,
      "learning_rate": 3.3681542293793636e-05,
      "loss": 0.5165,
      "step": 24939
    },
    {
      "epoch": 7.874960530470477,
      "grad_norm": 7.217324733734131,
      "learning_rate": 3.368066519313757e-05,
      "loss": 0.5628,
      "step": 24940
    },
    {
      "epoch": 7.875276286706662,
      "grad_norm": 6.117918968200684,
      "learning_rate": 3.367978809248149e-05,
      "loss": 0.6391,
      "step": 24941
    },
    {
      "epoch": 7.8755920429428485,
      "grad_norm": 4.660076141357422,
      "learning_rate": 3.3678910991825424e-05,
      "loss": 0.3864,
      "step": 24942
    },
    {
      "epoch": 7.875907799179034,
      "grad_norm": 7.7794013023376465,
      "learning_rate": 3.367803389116935e-05,
      "loss": 0.5262,
      "step": 24943
    },
    {
      "epoch": 7.876223555415219,
      "grad_norm": 5.415255069732666,
      "learning_rate": 3.367715679051328e-05,
      "loss": 0.4584,
      "step": 24944
    },
    {
      "epoch": 7.876539311651405,
      "grad_norm": 9.297484397888184,
      "learning_rate": 3.367627968985721e-05,
      "loss": 0.4942,
      "step": 24945
    },
    {
      "epoch": 7.876855067887591,
      "grad_norm": 5.69495153427124,
      "learning_rate": 3.367540258920114e-05,
      "loss": 0.4817,
      "step": 24946
    },
    {
      "epoch": 7.877170824123777,
      "grad_norm": 6.472632884979248,
      "learning_rate": 3.367452548854507e-05,
      "loss": 0.4649,
      "step": 24947
    },
    {
      "epoch": 7.877486580359962,
      "grad_norm": 7.034738540649414,
      "learning_rate": 3.3673648387888994e-05,
      "loss": 0.5827,
      "step": 24948
    },
    {
      "epoch": 7.877802336596147,
      "grad_norm": 7.670749187469482,
      "learning_rate": 3.3672771287232926e-05,
      "loss": 0.6246,
      "step": 24949
    },
    {
      "epoch": 7.8781180928323336,
      "grad_norm": 7.202738285064697,
      "learning_rate": 3.367189418657685e-05,
      "loss": 0.4953,
      "step": 24950
    },
    {
      "epoch": 7.878433849068519,
      "grad_norm": 5.976434230804443,
      "learning_rate": 3.367101708592078e-05,
      "loss": 0.4881,
      "step": 24951
    },
    {
      "epoch": 7.878749605304705,
      "grad_norm": 5.92603063583374,
      "learning_rate": 3.367013998526471e-05,
      "loss": 0.4765,
      "step": 24952
    },
    {
      "epoch": 7.87906536154089,
      "grad_norm": 5.90154504776001,
      "learning_rate": 3.366926288460864e-05,
      "loss": 0.463,
      "step": 24953
    },
    {
      "epoch": 7.879381117777076,
      "grad_norm": 5.3124518394470215,
      "learning_rate": 3.3668385783952564e-05,
      "loss": 0.4942,
      "step": 24954
    },
    {
      "epoch": 7.879696874013262,
      "grad_norm": 5.178063869476318,
      "learning_rate": 3.3667508683296496e-05,
      "loss": 0.4934,
      "step": 24955
    },
    {
      "epoch": 7.880012630249447,
      "grad_norm": 5.852679252624512,
      "learning_rate": 3.366663158264042e-05,
      "loss": 0.5567,
      "step": 24956
    },
    {
      "epoch": 7.880328386485633,
      "grad_norm": 5.236523628234863,
      "learning_rate": 3.366575448198435e-05,
      "loss": 0.5047,
      "step": 24957
    },
    {
      "epoch": 7.880644142721819,
      "grad_norm": 7.942060947418213,
      "learning_rate": 3.366487738132828e-05,
      "loss": 0.4428,
      "step": 24958
    },
    {
      "epoch": 7.880959898958005,
      "grad_norm": 5.3399834632873535,
      "learning_rate": 3.366400028067221e-05,
      "loss": 0.4572,
      "step": 24959
    },
    {
      "epoch": 7.88127565519419,
      "grad_norm": 5.4910783767700195,
      "learning_rate": 3.366312318001614e-05,
      "loss": 0.4859,
      "step": 24960
    },
    {
      "epoch": 7.881591411430376,
      "grad_norm": 6.032798767089844,
      "learning_rate": 3.366224607936007e-05,
      "loss": 0.5431,
      "step": 24961
    },
    {
      "epoch": 7.8819071676665615,
      "grad_norm": 5.170261859893799,
      "learning_rate": 3.3661368978704e-05,
      "loss": 0.5406,
      "step": 24962
    },
    {
      "epoch": 7.882222923902747,
      "grad_norm": 5.605739116668701,
      "learning_rate": 3.366049187804793e-05,
      "loss": 0.5922,
      "step": 24963
    },
    {
      "epoch": 7.882538680138933,
      "grad_norm": 7.121800422668457,
      "learning_rate": 3.3659614777391854e-05,
      "loss": 0.6267,
      "step": 24964
    },
    {
      "epoch": 7.882854436375118,
      "grad_norm": 6.459033012390137,
      "learning_rate": 3.3658737676735785e-05,
      "loss": 0.4228,
      "step": 24965
    },
    {
      "epoch": 7.8831701926113045,
      "grad_norm": 5.634371280670166,
      "learning_rate": 3.365786057607972e-05,
      "loss": 0.5499,
      "step": 24966
    },
    {
      "epoch": 7.88348594884749,
      "grad_norm": 5.732072353363037,
      "learning_rate": 3.365698347542364e-05,
      "loss": 0.5138,
      "step": 24967
    },
    {
      "epoch": 7.883801705083675,
      "grad_norm": 6.178797721862793,
      "learning_rate": 3.3656106374767574e-05,
      "loss": 0.5276,
      "step": 24968
    },
    {
      "epoch": 7.884117461319861,
      "grad_norm": 5.372343063354492,
      "learning_rate": 3.36552292741115e-05,
      "loss": 0.5153,
      "step": 24969
    },
    {
      "epoch": 7.8844332175560465,
      "grad_norm": 5.786337375640869,
      "learning_rate": 3.365435217345543e-05,
      "loss": 0.5855,
      "step": 24970
    },
    {
      "epoch": 7.884748973792233,
      "grad_norm": 5.082566261291504,
      "learning_rate": 3.3653475072799355e-05,
      "loss": 0.4091,
      "step": 24971
    },
    {
      "epoch": 7.885064730028418,
      "grad_norm": 5.700841426849365,
      "learning_rate": 3.365259797214329e-05,
      "loss": 0.5603,
      "step": 24972
    },
    {
      "epoch": 7.885380486264603,
      "grad_norm": 4.659460067749023,
      "learning_rate": 3.365172087148721e-05,
      "loss": 0.3966,
      "step": 24973
    },
    {
      "epoch": 7.8856962425007895,
      "grad_norm": 5.221163272857666,
      "learning_rate": 3.365084377083114e-05,
      "loss": 0.5756,
      "step": 24974
    },
    {
      "epoch": 7.886011998736975,
      "grad_norm": 5.498287200927734,
      "learning_rate": 3.364996667017507e-05,
      "loss": 0.5142,
      "step": 24975
    },
    {
      "epoch": 7.886327754973161,
      "grad_norm": 5.669528484344482,
      "learning_rate": 3.3649089569519e-05,
      "loss": 0.4602,
      "step": 24976
    },
    {
      "epoch": 7.886643511209346,
      "grad_norm": 6.1849870681762695,
      "learning_rate": 3.3648212468862925e-05,
      "loss": 0.5569,
      "step": 24977
    },
    {
      "epoch": 7.8869592674455316,
      "grad_norm": 5.274498462677002,
      "learning_rate": 3.3647335368206857e-05,
      "loss": 0.4784,
      "step": 24978
    },
    {
      "epoch": 7.887275023681718,
      "grad_norm": 4.807825565338135,
      "learning_rate": 3.364645826755078e-05,
      "loss": 0.4411,
      "step": 24979
    },
    {
      "epoch": 7.887590779917903,
      "grad_norm": 5.249710559844971,
      "learning_rate": 3.364558116689471e-05,
      "loss": 0.499,
      "step": 24980
    },
    {
      "epoch": 7.887906536154089,
      "grad_norm": 5.911553859710693,
      "learning_rate": 3.3644704066238645e-05,
      "loss": 0.565,
      "step": 24981
    },
    {
      "epoch": 7.8882222923902745,
      "grad_norm": 5.786334037780762,
      "learning_rate": 3.364382696558257e-05,
      "loss": 0.506,
      "step": 24982
    },
    {
      "epoch": 7.888538048626461,
      "grad_norm": 5.894332408905029,
      "learning_rate": 3.36429498649265e-05,
      "loss": 0.4938,
      "step": 24983
    },
    {
      "epoch": 7.888853804862646,
      "grad_norm": 4.568839073181152,
      "learning_rate": 3.3642072764270426e-05,
      "loss": 0.4949,
      "step": 24984
    },
    {
      "epoch": 7.889169561098832,
      "grad_norm": 5.904520034790039,
      "learning_rate": 3.364119566361436e-05,
      "loss": 0.5912,
      "step": 24985
    },
    {
      "epoch": 7.8894853173350175,
      "grad_norm": 5.6128106117248535,
      "learning_rate": 3.364031856295828e-05,
      "loss": 0.523,
      "step": 24986
    },
    {
      "epoch": 7.889801073571203,
      "grad_norm": 5.841679573059082,
      "learning_rate": 3.363944146230222e-05,
      "loss": 0.4901,
      "step": 24987
    },
    {
      "epoch": 7.890116829807389,
      "grad_norm": 5.553699970245361,
      "learning_rate": 3.3638564361646146e-05,
      "loss": 0.504,
      "step": 24988
    },
    {
      "epoch": 7.890432586043574,
      "grad_norm": 5.860404968261719,
      "learning_rate": 3.363768726099008e-05,
      "loss": 0.617,
      "step": 24989
    },
    {
      "epoch": 7.89074834227976,
      "grad_norm": 6.473987579345703,
      "learning_rate": 3.3636810160334e-05,
      "loss": 0.5606,
      "step": 24990
    },
    {
      "epoch": 7.891064098515946,
      "grad_norm": 6.443248271942139,
      "learning_rate": 3.3635933059677935e-05,
      "loss": 0.4779,
      "step": 24991
    },
    {
      "epoch": 7.891379854752131,
      "grad_norm": 5.141778945922852,
      "learning_rate": 3.363505595902186e-05,
      "loss": 0.454,
      "step": 24992
    },
    {
      "epoch": 7.891695610988317,
      "grad_norm": 6.404646873474121,
      "learning_rate": 3.363417885836579e-05,
      "loss": 0.4672,
      "step": 24993
    },
    {
      "epoch": 7.8920113672245025,
      "grad_norm": 5.592427730560303,
      "learning_rate": 3.3633301757709716e-05,
      "loss": 0.5129,
      "step": 24994
    },
    {
      "epoch": 7.892327123460689,
      "grad_norm": 5.74976110458374,
      "learning_rate": 3.363242465705365e-05,
      "loss": 0.5259,
      "step": 24995
    },
    {
      "epoch": 7.892642879696874,
      "grad_norm": 5.985231399536133,
      "learning_rate": 3.363154755639757e-05,
      "loss": 0.6363,
      "step": 24996
    },
    {
      "epoch": 7.892958635933059,
      "grad_norm": 5.157309532165527,
      "learning_rate": 3.3630670455741504e-05,
      "loss": 0.533,
      "step": 24997
    },
    {
      "epoch": 7.893274392169245,
      "grad_norm": 6.181020259857178,
      "learning_rate": 3.362979335508543e-05,
      "loss": 0.5909,
      "step": 24998
    },
    {
      "epoch": 7.893590148405431,
      "grad_norm": 5.340603351593018,
      "learning_rate": 3.362891625442936e-05,
      "loss": 0.5928,
      "step": 24999
    },
    {
      "epoch": 7.893905904641617,
      "grad_norm": 5.466520309448242,
      "learning_rate": 3.3628039153773286e-05,
      "loss": 0.3836,
      "step": 25000
    },
    {
      "epoch": 7.893905904641617,
      "eval_loss": 0.9125893115997314,
      "eval_runtime": 4.5888,
      "eval_samples_per_second": 542.626,
      "eval_steps_per_second": 4.576,
      "step": 25000
    },
    {
      "epoch": 7.894221660877802,
      "grad_norm": 5.564343452453613,
      "learning_rate": 3.362716205311722e-05,
      "loss": 0.5943,
      "step": 25001
    },
    {
      "epoch": 7.894537417113988,
      "grad_norm": 5.772937774658203,
      "learning_rate": 3.362628495246114e-05,
      "loss": 0.5004,
      "step": 25002
    },
    {
      "epoch": 7.894853173350174,
      "grad_norm": 5.204169750213623,
      "learning_rate": 3.3625407851805074e-05,
      "loss": 0.3828,
      "step": 25003
    },
    {
      "epoch": 7.895168929586359,
      "grad_norm": 4.160699367523193,
      "learning_rate": 3.3624530751149006e-05,
      "loss": 0.3558,
      "step": 25004
    },
    {
      "epoch": 7.895484685822545,
      "grad_norm": 6.021625995635986,
      "learning_rate": 3.362365365049293e-05,
      "loss": 0.5638,
      "step": 25005
    },
    {
      "epoch": 7.89580044205873,
      "grad_norm": 7.6077141761779785,
      "learning_rate": 3.362277654983686e-05,
      "loss": 0.5911,
      "step": 25006
    },
    {
      "epoch": 7.896116198294917,
      "grad_norm": 5.256026268005371,
      "learning_rate": 3.362189944918079e-05,
      "loss": 0.5076,
      "step": 25007
    },
    {
      "epoch": 7.896431954531102,
      "grad_norm": 5.206358909606934,
      "learning_rate": 3.362102234852472e-05,
      "loss": 0.426,
      "step": 25008
    },
    {
      "epoch": 7.896747710767288,
      "grad_norm": 5.6435747146606445,
      "learning_rate": 3.3620145247868644e-05,
      "loss": 0.5875,
      "step": 25009
    },
    {
      "epoch": 7.897063467003473,
      "grad_norm": 4.813241004943848,
      "learning_rate": 3.3619268147212576e-05,
      "loss": 0.4816,
      "step": 25010
    },
    {
      "epoch": 7.897379223239659,
      "grad_norm": 5.464188575744629,
      "learning_rate": 3.36183910465565e-05,
      "loss": 0.6042,
      "step": 25011
    },
    {
      "epoch": 7.897694979475845,
      "grad_norm": 6.0693559646606445,
      "learning_rate": 3.361751394590043e-05,
      "loss": 0.4518,
      "step": 25012
    },
    {
      "epoch": 7.89801073571203,
      "grad_norm": 5.659135818481445,
      "learning_rate": 3.361663684524436e-05,
      "loss": 0.621,
      "step": 25013
    },
    {
      "epoch": 7.898326491948216,
      "grad_norm": 4.831893444061279,
      "learning_rate": 3.3615759744588295e-05,
      "loss": 0.4851,
      "step": 25014
    },
    {
      "epoch": 7.898642248184402,
      "grad_norm": 4.946932315826416,
      "learning_rate": 3.361488264393222e-05,
      "loss": 0.4362,
      "step": 25015
    },
    {
      "epoch": 7.898958004420587,
      "grad_norm": 7.5716142654418945,
      "learning_rate": 3.361400554327615e-05,
      "loss": 0.5908,
      "step": 25016
    },
    {
      "epoch": 7.899273760656773,
      "grad_norm": 5.662925720214844,
      "learning_rate": 3.361312844262008e-05,
      "loss": 0.4882,
      "step": 25017
    },
    {
      "epoch": 7.899589516892958,
      "grad_norm": 5.138887882232666,
      "learning_rate": 3.361225134196401e-05,
      "loss": 0.5243,
      "step": 25018
    },
    {
      "epoch": 7.899905273129145,
      "grad_norm": 5.504673957824707,
      "learning_rate": 3.3611374241307934e-05,
      "loss": 0.5208,
      "step": 25019
    },
    {
      "epoch": 7.90022102936533,
      "grad_norm": 4.959921360015869,
      "learning_rate": 3.3610497140651865e-05,
      "loss": 0.5474,
      "step": 25020
    },
    {
      "epoch": 7.900536785601515,
      "grad_norm": 6.01669979095459,
      "learning_rate": 3.360962003999579e-05,
      "loss": 0.5341,
      "step": 25021
    },
    {
      "epoch": 7.900852541837701,
      "grad_norm": 5.479823589324951,
      "learning_rate": 3.360874293933972e-05,
      "loss": 0.5005,
      "step": 25022
    },
    {
      "epoch": 7.901168298073887,
      "grad_norm": 4.595181465148926,
      "learning_rate": 3.360786583868365e-05,
      "loss": 0.3607,
      "step": 25023
    },
    {
      "epoch": 7.901484054310073,
      "grad_norm": 5.82223653793335,
      "learning_rate": 3.360698873802758e-05,
      "loss": 0.496,
      "step": 25024
    },
    {
      "epoch": 7.901799810546258,
      "grad_norm": 5.948606491088867,
      "learning_rate": 3.360611163737151e-05,
      "loss": 0.5735,
      "step": 25025
    },
    {
      "epoch": 7.902115566782444,
      "grad_norm": 4.684382915496826,
      "learning_rate": 3.3605234536715435e-05,
      "loss": 0.4405,
      "step": 25026
    },
    {
      "epoch": 7.90243132301863,
      "grad_norm": 5.629249572753906,
      "learning_rate": 3.360435743605937e-05,
      "loss": 0.5765,
      "step": 25027
    },
    {
      "epoch": 7.902747079254815,
      "grad_norm": 5.315922737121582,
      "learning_rate": 3.360348033540329e-05,
      "loss": 0.4923,
      "step": 25028
    },
    {
      "epoch": 7.903062835491001,
      "grad_norm": 6.609760761260986,
      "learning_rate": 3.360260323474722e-05,
      "loss": 0.5596,
      "step": 25029
    },
    {
      "epoch": 7.903378591727186,
      "grad_norm": 5.7799553871154785,
      "learning_rate": 3.360172613409115e-05,
      "loss": 0.5414,
      "step": 25030
    },
    {
      "epoch": 7.903694347963373,
      "grad_norm": 4.438065528869629,
      "learning_rate": 3.360084903343508e-05,
      "loss": 0.4156,
      "step": 25031
    },
    {
      "epoch": 7.904010104199558,
      "grad_norm": 5.652108669281006,
      "learning_rate": 3.3599971932779005e-05,
      "loss": 0.5237,
      "step": 25032
    },
    {
      "epoch": 7.904325860435744,
      "grad_norm": 4.175403594970703,
      "learning_rate": 3.3599094832122936e-05,
      "loss": 0.4295,
      "step": 25033
    },
    {
      "epoch": 7.904641616671929,
      "grad_norm": 8.260732650756836,
      "learning_rate": 3.359821773146686e-05,
      "loss": 0.4777,
      "step": 25034
    },
    {
      "epoch": 7.904957372908115,
      "grad_norm": 5.706547737121582,
      "learning_rate": 3.359734063081079e-05,
      "loss": 0.5322,
      "step": 25035
    },
    {
      "epoch": 7.905273129144301,
      "grad_norm": 6.0232367515563965,
      "learning_rate": 3.359646353015472e-05,
      "loss": 0.4583,
      "step": 25036
    },
    {
      "epoch": 7.905588885380486,
      "grad_norm": 4.618127822875977,
      "learning_rate": 3.359558642949865e-05,
      "loss": 0.4027,
      "step": 25037
    },
    {
      "epoch": 7.905904641616672,
      "grad_norm": 5.138893127441406,
      "learning_rate": 3.3594709328842575e-05,
      "loss": 0.4303,
      "step": 25038
    },
    {
      "epoch": 7.906220397852858,
      "grad_norm": 4.5350260734558105,
      "learning_rate": 3.3593832228186506e-05,
      "loss": 0.4461,
      "step": 25039
    },
    {
      "epoch": 7.906536154089043,
      "grad_norm": 4.613184452056885,
      "learning_rate": 3.359295512753043e-05,
      "loss": 0.4372,
      "step": 25040
    },
    {
      "epoch": 7.906851910325229,
      "grad_norm": 6.0315985679626465,
      "learning_rate": 3.359207802687437e-05,
      "loss": 0.4711,
      "step": 25041
    },
    {
      "epoch": 7.907167666561414,
      "grad_norm": 5.9076056480407715,
      "learning_rate": 3.3591200926218294e-05,
      "loss": 0.4832,
      "step": 25042
    },
    {
      "epoch": 7.9074834227976005,
      "grad_norm": 5.416701316833496,
      "learning_rate": 3.3590323825562226e-05,
      "loss": 0.5122,
      "step": 25043
    },
    {
      "epoch": 7.907799179033786,
      "grad_norm": 5.685701370239258,
      "learning_rate": 3.358944672490615e-05,
      "loss": 0.4318,
      "step": 25044
    },
    {
      "epoch": 7.908114935269971,
      "grad_norm": 6.602685451507568,
      "learning_rate": 3.358856962425008e-05,
      "loss": 0.5381,
      "step": 25045
    },
    {
      "epoch": 7.908430691506157,
      "grad_norm": 4.8457417488098145,
      "learning_rate": 3.358769252359401e-05,
      "loss": 0.4542,
      "step": 25046
    },
    {
      "epoch": 7.908746447742343,
      "grad_norm": 4.307812213897705,
      "learning_rate": 3.358681542293794e-05,
      "loss": 0.3958,
      "step": 25047
    },
    {
      "epoch": 7.909062203978529,
      "grad_norm": 4.547451972961426,
      "learning_rate": 3.358593832228187e-05,
      "loss": 0.4947,
      "step": 25048
    },
    {
      "epoch": 7.909377960214714,
      "grad_norm": 6.9182939529418945,
      "learning_rate": 3.3585061221625796e-05,
      "loss": 0.5534,
      "step": 25049
    },
    {
      "epoch": 7.9096937164509,
      "grad_norm": 4.353264808654785,
      "learning_rate": 3.358418412096973e-05,
      "loss": 0.4084,
      "step": 25050
    },
    {
      "epoch": 7.9100094726870855,
      "grad_norm": 7.839415073394775,
      "learning_rate": 3.358330702031365e-05,
      "loss": 0.58,
      "step": 25051
    },
    {
      "epoch": 7.910325228923272,
      "grad_norm": 4.234469890594482,
      "learning_rate": 3.3582429919657584e-05,
      "loss": 0.379,
      "step": 25052
    },
    {
      "epoch": 7.910640985159457,
      "grad_norm": 7.015510082244873,
      "learning_rate": 3.358155281900151e-05,
      "loss": 0.6208,
      "step": 25053
    },
    {
      "epoch": 7.910956741395642,
      "grad_norm": 5.820189476013184,
      "learning_rate": 3.358067571834544e-05,
      "loss": 0.5123,
      "step": 25054
    },
    {
      "epoch": 7.9112724976318285,
      "grad_norm": 5.417384147644043,
      "learning_rate": 3.3579798617689366e-05,
      "loss": 0.5346,
      "step": 25055
    },
    {
      "epoch": 7.911588253868014,
      "grad_norm": 4.873292446136475,
      "learning_rate": 3.35789215170333e-05,
      "loss": 0.4923,
      "step": 25056
    },
    {
      "epoch": 7.9119040101042,
      "grad_norm": 6.065768241882324,
      "learning_rate": 3.357804441637722e-05,
      "loss": 0.4853,
      "step": 25057
    },
    {
      "epoch": 7.912219766340385,
      "grad_norm": 6.043734073638916,
      "learning_rate": 3.3577167315721154e-05,
      "loss": 0.5945,
      "step": 25058
    },
    {
      "epoch": 7.912535522576571,
      "grad_norm": 5.178821563720703,
      "learning_rate": 3.357629021506508e-05,
      "loss": 0.5264,
      "step": 25059
    },
    {
      "epoch": 7.912851278812757,
      "grad_norm": 6.009006500244141,
      "learning_rate": 3.357541311440901e-05,
      "loss": 0.5389,
      "step": 25060
    },
    {
      "epoch": 7.913167035048942,
      "grad_norm": 5.23153829574585,
      "learning_rate": 3.3574536013752935e-05,
      "loss": 0.4887,
      "step": 25061
    },
    {
      "epoch": 7.913482791285128,
      "grad_norm": 6.623554706573486,
      "learning_rate": 3.357365891309687e-05,
      "loss": 0.6387,
      "step": 25062
    },
    {
      "epoch": 7.9137985475213135,
      "grad_norm": 5.187828063964844,
      "learning_rate": 3.35727818124408e-05,
      "loss": 0.6353,
      "step": 25063
    },
    {
      "epoch": 7.914114303757499,
      "grad_norm": 6.341704368591309,
      "learning_rate": 3.3571904711784724e-05,
      "loss": 0.6837,
      "step": 25064
    },
    {
      "epoch": 7.914430059993685,
      "grad_norm": 5.321035385131836,
      "learning_rate": 3.3571027611128655e-05,
      "loss": 0.5712,
      "step": 25065
    },
    {
      "epoch": 7.91474581622987,
      "grad_norm": 5.290307998657227,
      "learning_rate": 3.357015051047258e-05,
      "loss": 0.5364,
      "step": 25066
    },
    {
      "epoch": 7.9150615724660565,
      "grad_norm": 6.540792942047119,
      "learning_rate": 3.356927340981651e-05,
      "loss": 0.5622,
      "step": 25067
    },
    {
      "epoch": 7.915377328702242,
      "grad_norm": 4.766602993011475,
      "learning_rate": 3.356839630916044e-05,
      "loss": 0.4543,
      "step": 25068
    },
    {
      "epoch": 7.915693084938427,
      "grad_norm": 5.38555908203125,
      "learning_rate": 3.3567519208504375e-05,
      "loss": 0.5342,
      "step": 25069
    },
    {
      "epoch": 7.916008841174613,
      "grad_norm": 5.037910461425781,
      "learning_rate": 3.35666421078483e-05,
      "loss": 0.5168,
      "step": 25070
    },
    {
      "epoch": 7.9163245974107985,
      "grad_norm": 6.640056610107422,
      "learning_rate": 3.356576500719223e-05,
      "loss": 0.5018,
      "step": 25071
    },
    {
      "epoch": 7.916640353646985,
      "grad_norm": 4.718790054321289,
      "learning_rate": 3.356488790653616e-05,
      "loss": 0.5285,
      "step": 25072
    },
    {
      "epoch": 7.91695610988317,
      "grad_norm": 5.854372978210449,
      "learning_rate": 3.356401080588009e-05,
      "loss": 0.575,
      "step": 25073
    },
    {
      "epoch": 7.917271866119356,
      "grad_norm": 4.890913963317871,
      "learning_rate": 3.3563133705224013e-05,
      "loss": 0.5344,
      "step": 25074
    },
    {
      "epoch": 7.9175876223555415,
      "grad_norm": 5.088609218597412,
      "learning_rate": 3.3562256604567945e-05,
      "loss": 0.5382,
      "step": 25075
    },
    {
      "epoch": 7.917903378591728,
      "grad_norm": 3.9169442653656006,
      "learning_rate": 3.356137950391187e-05,
      "loss": 0.456,
      "step": 25076
    },
    {
      "epoch": 7.918219134827913,
      "grad_norm": 6.825075626373291,
      "learning_rate": 3.35605024032558e-05,
      "loss": 0.617,
      "step": 25077
    },
    {
      "epoch": 7.918534891064098,
      "grad_norm": 4.437727928161621,
      "learning_rate": 3.355962530259973e-05,
      "loss": 0.5117,
      "step": 25078
    },
    {
      "epoch": 7.918850647300284,
      "grad_norm": 5.1579060554504395,
      "learning_rate": 3.355874820194366e-05,
      "loss": 0.4976,
      "step": 25079
    },
    {
      "epoch": 7.91916640353647,
      "grad_norm": 4.911123275756836,
      "learning_rate": 3.355787110128758e-05,
      "loss": 0.5818,
      "step": 25080
    },
    {
      "epoch": 7.919482159772656,
      "grad_norm": 5.396620273590088,
      "learning_rate": 3.3556994000631515e-05,
      "loss": 0.5023,
      "step": 25081
    },
    {
      "epoch": 7.919797916008841,
      "grad_norm": 5.099360466003418,
      "learning_rate": 3.355611689997544e-05,
      "loss": 0.5972,
      "step": 25082
    },
    {
      "epoch": 7.9201136722450265,
      "grad_norm": 5.930096626281738,
      "learning_rate": 3.355523979931937e-05,
      "loss": 0.5954,
      "step": 25083
    },
    {
      "epoch": 7.920429428481213,
      "grad_norm": 5.280816555023193,
      "learning_rate": 3.3554362698663296e-05,
      "loss": 0.5073,
      "step": 25084
    },
    {
      "epoch": 7.920745184717398,
      "grad_norm": 7.573934078216553,
      "learning_rate": 3.355348559800723e-05,
      "loss": 0.5528,
      "step": 25085
    },
    {
      "epoch": 7.921060940953584,
      "grad_norm": 4.714738845825195,
      "learning_rate": 3.355260849735116e-05,
      "loss": 0.4467,
      "step": 25086
    },
    {
      "epoch": 7.9213766971897694,
      "grad_norm": 6.548955917358398,
      "learning_rate": 3.3551731396695085e-05,
      "loss": 0.5369,
      "step": 25087
    },
    {
      "epoch": 7.921692453425955,
      "grad_norm": 6.1396708488464355,
      "learning_rate": 3.3550854296039016e-05,
      "loss": 0.4964,
      "step": 25088
    },
    {
      "epoch": 7.922008209662141,
      "grad_norm": 6.869327545166016,
      "learning_rate": 3.354997719538294e-05,
      "loss": 0.4907,
      "step": 25089
    },
    {
      "epoch": 7.922323965898326,
      "grad_norm": 4.90052604675293,
      "learning_rate": 3.354910009472687e-05,
      "loss": 0.4189,
      "step": 25090
    },
    {
      "epoch": 7.922639722134512,
      "grad_norm": 5.665724754333496,
      "learning_rate": 3.35482229940708e-05,
      "loss": 0.5726,
      "step": 25091
    },
    {
      "epoch": 7.922955478370698,
      "grad_norm": 5.595396995544434,
      "learning_rate": 3.354734589341473e-05,
      "loss": 0.4759,
      "step": 25092
    },
    {
      "epoch": 7.923271234606884,
      "grad_norm": 5.038853168487549,
      "learning_rate": 3.3546468792758654e-05,
      "loss": 0.4132,
      "step": 25093
    },
    {
      "epoch": 7.923586990843069,
      "grad_norm": 5.710946083068848,
      "learning_rate": 3.3545591692102586e-05,
      "loss": 0.4712,
      "step": 25094
    },
    {
      "epoch": 7.9239027470792545,
      "grad_norm": 5.132606029510498,
      "learning_rate": 3.354471459144651e-05,
      "loss": 0.4442,
      "step": 25095
    },
    {
      "epoch": 7.924218503315441,
      "grad_norm": 9.559606552124023,
      "learning_rate": 3.354383749079045e-05,
      "loss": 0.6637,
      "step": 25096
    },
    {
      "epoch": 7.924534259551626,
      "grad_norm": 6.359515190124512,
      "learning_rate": 3.3542960390134374e-05,
      "loss": 0.4998,
      "step": 25097
    },
    {
      "epoch": 7.924850015787812,
      "grad_norm": 5.688345432281494,
      "learning_rate": 3.3542083289478306e-05,
      "loss": 0.5576,
      "step": 25098
    },
    {
      "epoch": 7.925165772023997,
      "grad_norm": 5.242861747741699,
      "learning_rate": 3.354120618882223e-05,
      "loss": 0.3839,
      "step": 25099
    },
    {
      "epoch": 7.925481528260184,
      "grad_norm": 5.136758327484131,
      "learning_rate": 3.354032908816616e-05,
      "loss": 0.4769,
      "step": 25100
    },
    {
      "epoch": 7.925797284496369,
      "grad_norm": 6.127936363220215,
      "learning_rate": 3.353945198751009e-05,
      "loss": 0.503,
      "step": 25101
    },
    {
      "epoch": 7.926113040732554,
      "grad_norm": 4.678566932678223,
      "learning_rate": 3.353857488685402e-05,
      "loss": 0.491,
      "step": 25102
    },
    {
      "epoch": 7.92642879696874,
      "grad_norm": 4.872808933258057,
      "learning_rate": 3.3537697786197944e-05,
      "loss": 0.4234,
      "step": 25103
    },
    {
      "epoch": 7.926744553204926,
      "grad_norm": 4.980273723602295,
      "learning_rate": 3.3536820685541876e-05,
      "loss": 0.4099,
      "step": 25104
    },
    {
      "epoch": 7.927060309441112,
      "grad_norm": 5.393723487854004,
      "learning_rate": 3.35359435848858e-05,
      "loss": 0.5212,
      "step": 25105
    },
    {
      "epoch": 7.927376065677297,
      "grad_norm": 6.3653244972229,
      "learning_rate": 3.353506648422973e-05,
      "loss": 0.5367,
      "step": 25106
    },
    {
      "epoch": 7.927691821913482,
      "grad_norm": 5.489541530609131,
      "learning_rate": 3.3534189383573664e-05,
      "loss": 0.537,
      "step": 25107
    },
    {
      "epoch": 7.928007578149669,
      "grad_norm": 5.7740302085876465,
      "learning_rate": 3.353331228291759e-05,
      "loss": 0.5806,
      "step": 25108
    },
    {
      "epoch": 7.928323334385854,
      "grad_norm": 5.950094699859619,
      "learning_rate": 3.353243518226152e-05,
      "loss": 0.5501,
      "step": 25109
    },
    {
      "epoch": 7.92863909062204,
      "grad_norm": 4.9236578941345215,
      "learning_rate": 3.3531558081605446e-05,
      "loss": 0.5856,
      "step": 25110
    },
    {
      "epoch": 7.928954846858225,
      "grad_norm": 5.61505126953125,
      "learning_rate": 3.353068098094938e-05,
      "loss": 0.4861,
      "step": 25111
    },
    {
      "epoch": 7.929270603094411,
      "grad_norm": 6.306419372558594,
      "learning_rate": 3.35298038802933e-05,
      "loss": 0.3393,
      "step": 25112
    },
    {
      "epoch": 7.929586359330597,
      "grad_norm": 7.776026248931885,
      "learning_rate": 3.3528926779637234e-05,
      "loss": 0.6085,
      "step": 25113
    },
    {
      "epoch": 7.929902115566782,
      "grad_norm": 5.840158939361572,
      "learning_rate": 3.352804967898116e-05,
      "loss": 0.4602,
      "step": 25114
    },
    {
      "epoch": 7.930217871802968,
      "grad_norm": 5.908775329589844,
      "learning_rate": 3.352717257832509e-05,
      "loss": 0.5036,
      "step": 25115
    },
    {
      "epoch": 7.930533628039154,
      "grad_norm": 4.817406177520752,
      "learning_rate": 3.3526295477669015e-05,
      "loss": 0.5234,
      "step": 25116
    },
    {
      "epoch": 7.93084938427534,
      "grad_norm": 6.5421295166015625,
      "learning_rate": 3.352541837701295e-05,
      "loss": 0.6948,
      "step": 25117
    },
    {
      "epoch": 7.931165140511525,
      "grad_norm": 5.2360148429870605,
      "learning_rate": 3.352454127635687e-05,
      "loss": 0.5409,
      "step": 25118
    },
    {
      "epoch": 7.93148089674771,
      "grad_norm": 4.443299293518066,
      "learning_rate": 3.3523664175700804e-05,
      "loss": 0.4338,
      "step": 25119
    },
    {
      "epoch": 7.931796652983897,
      "grad_norm": 5.3176960945129395,
      "learning_rate": 3.352278707504473e-05,
      "loss": 0.5887,
      "step": 25120
    },
    {
      "epoch": 7.932112409220082,
      "grad_norm": 6.828208923339844,
      "learning_rate": 3.352190997438866e-05,
      "loss": 0.616,
      "step": 25121
    },
    {
      "epoch": 7.932428165456268,
      "grad_norm": 6.330726623535156,
      "learning_rate": 3.3521032873732585e-05,
      "loss": 0.4277,
      "step": 25122
    },
    {
      "epoch": 7.932743921692453,
      "grad_norm": 6.0665082931518555,
      "learning_rate": 3.3520155773076524e-05,
      "loss": 0.5533,
      "step": 25123
    },
    {
      "epoch": 7.9330596779286395,
      "grad_norm": 5.732019424438477,
      "learning_rate": 3.351927867242045e-05,
      "loss": 0.4644,
      "step": 25124
    },
    {
      "epoch": 7.933375434164825,
      "grad_norm": 5.404148101806641,
      "learning_rate": 3.351840157176438e-05,
      "loss": 0.4525,
      "step": 25125
    },
    {
      "epoch": 7.93369119040101,
      "grad_norm": 6.322234630584717,
      "learning_rate": 3.3517524471108305e-05,
      "loss": 0.5377,
      "step": 25126
    },
    {
      "epoch": 7.934006946637196,
      "grad_norm": 4.584230422973633,
      "learning_rate": 3.351664737045224e-05,
      "loss": 0.4948,
      "step": 25127
    },
    {
      "epoch": 7.934322702873382,
      "grad_norm": 8.080741882324219,
      "learning_rate": 3.351577026979617e-05,
      "loss": 0.5461,
      "step": 25128
    },
    {
      "epoch": 7.934638459109568,
      "grad_norm": 6.604861736297607,
      "learning_rate": 3.351489316914009e-05,
      "loss": 0.4405,
      "step": 25129
    },
    {
      "epoch": 7.934954215345753,
      "grad_norm": 6.2335686683654785,
      "learning_rate": 3.3514016068484025e-05,
      "loss": 0.4912,
      "step": 25130
    },
    {
      "epoch": 7.935269971581938,
      "grad_norm": 6.107553482055664,
      "learning_rate": 3.351313896782795e-05,
      "loss": 0.47,
      "step": 25131
    },
    {
      "epoch": 7.9355857278181245,
      "grad_norm": 5.5950703620910645,
      "learning_rate": 3.351226186717188e-05,
      "loss": 0.5511,
      "step": 25132
    },
    {
      "epoch": 7.93590148405431,
      "grad_norm": 5.27675199508667,
      "learning_rate": 3.3511384766515807e-05,
      "loss": 0.5081,
      "step": 25133
    },
    {
      "epoch": 7.936217240290496,
      "grad_norm": 6.322755813598633,
      "learning_rate": 3.351050766585974e-05,
      "loss": 0.5276,
      "step": 25134
    },
    {
      "epoch": 7.936532996526681,
      "grad_norm": 5.800239562988281,
      "learning_rate": 3.350963056520366e-05,
      "loss": 0.5296,
      "step": 25135
    },
    {
      "epoch": 7.936848752762867,
      "grad_norm": 6.019006729125977,
      "learning_rate": 3.3508753464547595e-05,
      "loss": 0.6109,
      "step": 25136
    },
    {
      "epoch": 7.937164508999053,
      "grad_norm": 4.983691692352295,
      "learning_rate": 3.350787636389152e-05,
      "loss": 0.4616,
      "step": 25137
    },
    {
      "epoch": 7.937480265235238,
      "grad_norm": 6.088138580322266,
      "learning_rate": 3.350699926323545e-05,
      "loss": 0.5211,
      "step": 25138
    },
    {
      "epoch": 7.937796021471424,
      "grad_norm": 5.498225212097168,
      "learning_rate": 3.3506122162579376e-05,
      "loss": 0.5677,
      "step": 25139
    },
    {
      "epoch": 7.93811177770761,
      "grad_norm": 5.353864669799805,
      "learning_rate": 3.350524506192331e-05,
      "loss": 0.5118,
      "step": 25140
    },
    {
      "epoch": 7.938427533943796,
      "grad_norm": 5.030052661895752,
      "learning_rate": 3.350436796126723e-05,
      "loss": 0.5071,
      "step": 25141
    },
    {
      "epoch": 7.938743290179981,
      "grad_norm": 4.494981288909912,
      "learning_rate": 3.3503490860611165e-05,
      "loss": 0.4234,
      "step": 25142
    },
    {
      "epoch": 7.939059046416166,
      "grad_norm": 6.197914123535156,
      "learning_rate": 3.350261375995509e-05,
      "loss": 0.4122,
      "step": 25143
    },
    {
      "epoch": 7.9393748026523525,
      "grad_norm": 5.437439441680908,
      "learning_rate": 3.350173665929902e-05,
      "loss": 0.4709,
      "step": 25144
    },
    {
      "epoch": 7.939690558888538,
      "grad_norm": 5.202510356903076,
      "learning_rate": 3.350085955864295e-05,
      "loss": 0.516,
      "step": 25145
    },
    {
      "epoch": 7.940006315124724,
      "grad_norm": 4.9398908615112305,
      "learning_rate": 3.349998245798688e-05,
      "loss": 0.4347,
      "step": 25146
    },
    {
      "epoch": 7.940322071360909,
      "grad_norm": 6.126698017120361,
      "learning_rate": 3.349910535733081e-05,
      "loss": 0.6821,
      "step": 25147
    },
    {
      "epoch": 7.9406378275970955,
      "grad_norm": 6.28717565536499,
      "learning_rate": 3.3498228256674734e-05,
      "loss": 0.5431,
      "step": 25148
    },
    {
      "epoch": 7.940953583833281,
      "grad_norm": 5.712449073791504,
      "learning_rate": 3.3497351156018666e-05,
      "loss": 0.5228,
      "step": 25149
    },
    {
      "epoch": 7.941269340069466,
      "grad_norm": 4.456228256225586,
      "learning_rate": 3.34964740553626e-05,
      "loss": 0.38,
      "step": 25150
    },
    {
      "epoch": 7.941585096305652,
      "grad_norm": 5.946821212768555,
      "learning_rate": 3.349559695470653e-05,
      "loss": 0.5067,
      "step": 25151
    },
    {
      "epoch": 7.9419008525418375,
      "grad_norm": 7.323879241943359,
      "learning_rate": 3.3494719854050454e-05,
      "loss": 0.6285,
      "step": 25152
    },
    {
      "epoch": 7.942216608778024,
      "grad_norm": 5.431787490844727,
      "learning_rate": 3.3493842753394386e-05,
      "loss": 0.5391,
      "step": 25153
    },
    {
      "epoch": 7.942532365014209,
      "grad_norm": 6.010695934295654,
      "learning_rate": 3.349296565273831e-05,
      "loss": 0.4915,
      "step": 25154
    },
    {
      "epoch": 7.942848121250394,
      "grad_norm": 4.907790184020996,
      "learning_rate": 3.349208855208224e-05,
      "loss": 0.4381,
      "step": 25155
    },
    {
      "epoch": 7.9431638774865805,
      "grad_norm": 6.699649810791016,
      "learning_rate": 3.349121145142617e-05,
      "loss": 0.5239,
      "step": 25156
    },
    {
      "epoch": 7.943479633722766,
      "grad_norm": 5.933403968811035,
      "learning_rate": 3.34903343507701e-05,
      "loss": 0.4685,
      "step": 25157
    },
    {
      "epoch": 7.943795389958952,
      "grad_norm": 5.177937030792236,
      "learning_rate": 3.3489457250114024e-05,
      "loss": 0.484,
      "step": 25158
    },
    {
      "epoch": 7.944111146195137,
      "grad_norm": 4.694788932800293,
      "learning_rate": 3.3488580149457956e-05,
      "loss": 0.3262,
      "step": 25159
    },
    {
      "epoch": 7.9444269024313225,
      "grad_norm": 5.379736423492432,
      "learning_rate": 3.348770304880188e-05,
      "loss": 0.4651,
      "step": 25160
    },
    {
      "epoch": 7.944742658667509,
      "grad_norm": 4.903967380523682,
      "learning_rate": 3.348682594814581e-05,
      "loss": 0.5329,
      "step": 25161
    },
    {
      "epoch": 7.945058414903694,
      "grad_norm": 6.746187686920166,
      "learning_rate": 3.348594884748974e-05,
      "loss": 0.5271,
      "step": 25162
    },
    {
      "epoch": 7.94537417113988,
      "grad_norm": 5.5535688400268555,
      "learning_rate": 3.348507174683367e-05,
      "loss": 0.4268,
      "step": 25163
    },
    {
      "epoch": 7.9456899273760655,
      "grad_norm": 7.531042098999023,
      "learning_rate": 3.3484194646177594e-05,
      "loss": 0.4846,
      "step": 25164
    },
    {
      "epoch": 7.946005683612252,
      "grad_norm": 5.209039211273193,
      "learning_rate": 3.3483317545521525e-05,
      "loss": 0.4472,
      "step": 25165
    },
    {
      "epoch": 7.946321439848437,
      "grad_norm": 5.739217281341553,
      "learning_rate": 3.348244044486546e-05,
      "loss": 0.5512,
      "step": 25166
    },
    {
      "epoch": 7.946637196084623,
      "grad_norm": 6.530091285705566,
      "learning_rate": 3.348156334420938e-05,
      "loss": 0.5285,
      "step": 25167
    },
    {
      "epoch": 7.9469529523208084,
      "grad_norm": 5.607644557952881,
      "learning_rate": 3.3480686243553314e-05,
      "loss": 0.4283,
      "step": 25168
    },
    {
      "epoch": 7.947268708556994,
      "grad_norm": 5.035081386566162,
      "learning_rate": 3.347980914289724e-05,
      "loss": 0.4655,
      "step": 25169
    },
    {
      "epoch": 7.94758446479318,
      "grad_norm": 6.839024543762207,
      "learning_rate": 3.347893204224117e-05,
      "loss": 0.5166,
      "step": 25170
    },
    {
      "epoch": 7.947900221029365,
      "grad_norm": 5.788182258605957,
      "learning_rate": 3.3478054941585095e-05,
      "loss": 0.4669,
      "step": 25171
    },
    {
      "epoch": 7.948215977265551,
      "grad_norm": 6.71367883682251,
      "learning_rate": 3.347717784092903e-05,
      "loss": 0.5233,
      "step": 25172
    },
    {
      "epoch": 7.948531733501737,
      "grad_norm": 6.3876237869262695,
      "learning_rate": 3.347630074027295e-05,
      "loss": 0.5648,
      "step": 25173
    },
    {
      "epoch": 7.948847489737922,
      "grad_norm": 5.17983341217041,
      "learning_rate": 3.3475423639616884e-05,
      "loss": 0.4106,
      "step": 25174
    },
    {
      "epoch": 7.949163245974108,
      "grad_norm": 5.403903007507324,
      "learning_rate": 3.347454653896081e-05,
      "loss": 0.5019,
      "step": 25175
    },
    {
      "epoch": 7.9494790022102935,
      "grad_norm": 6.282931327819824,
      "learning_rate": 3.347366943830474e-05,
      "loss": 0.4682,
      "step": 25176
    },
    {
      "epoch": 7.94979475844648,
      "grad_norm": 6.363588809967041,
      "learning_rate": 3.3472792337648665e-05,
      "loss": 0.5323,
      "step": 25177
    },
    {
      "epoch": 7.950110514682665,
      "grad_norm": 5.338317394256592,
      "learning_rate": 3.3471915236992603e-05,
      "loss": 0.4978,
      "step": 25178
    },
    {
      "epoch": 7.95042627091885,
      "grad_norm": 6.994160175323486,
      "learning_rate": 3.347103813633653e-05,
      "loss": 0.5334,
      "step": 25179
    },
    {
      "epoch": 7.950742027155036,
      "grad_norm": 5.1748857498168945,
      "learning_rate": 3.347016103568046e-05,
      "loss": 0.4516,
      "step": 25180
    },
    {
      "epoch": 7.951057783391222,
      "grad_norm": 4.781492233276367,
      "learning_rate": 3.3469283935024385e-05,
      "loss": 0.4812,
      "step": 25181
    },
    {
      "epoch": 7.951373539627408,
      "grad_norm": 6.3312087059021,
      "learning_rate": 3.346840683436832e-05,
      "loss": 0.6138,
      "step": 25182
    },
    {
      "epoch": 7.951689295863593,
      "grad_norm": 6.559086322784424,
      "learning_rate": 3.346752973371224e-05,
      "loss": 0.4048,
      "step": 25183
    },
    {
      "epoch": 7.9520050520997785,
      "grad_norm": 5.7274322509765625,
      "learning_rate": 3.346665263305617e-05,
      "loss": 0.5106,
      "step": 25184
    },
    {
      "epoch": 7.952320808335965,
      "grad_norm": 7.507129669189453,
      "learning_rate": 3.34657755324001e-05,
      "loss": 0.4973,
      "step": 25185
    },
    {
      "epoch": 7.95263656457215,
      "grad_norm": 6.608728408813477,
      "learning_rate": 3.346489843174403e-05,
      "loss": 0.5211,
      "step": 25186
    },
    {
      "epoch": 7.952952320808336,
      "grad_norm": 4.979165554046631,
      "learning_rate": 3.3464021331087955e-05,
      "loss": 0.5058,
      "step": 25187
    },
    {
      "epoch": 7.953268077044521,
      "grad_norm": 5.9378814697265625,
      "learning_rate": 3.3463144230431886e-05,
      "loss": 0.4987,
      "step": 25188
    },
    {
      "epoch": 7.953583833280708,
      "grad_norm": 5.018572807312012,
      "learning_rate": 3.346226712977582e-05,
      "loss": 0.4154,
      "step": 25189
    },
    {
      "epoch": 7.953899589516893,
      "grad_norm": 5.44279146194458,
      "learning_rate": 3.346139002911974e-05,
      "loss": 0.4552,
      "step": 25190
    },
    {
      "epoch": 7.954215345753079,
      "grad_norm": 4.887214660644531,
      "learning_rate": 3.3460512928463675e-05,
      "loss": 0.5078,
      "step": 25191
    },
    {
      "epoch": 7.954531101989264,
      "grad_norm": 6.957798480987549,
      "learning_rate": 3.34596358278076e-05,
      "loss": 0.5745,
      "step": 25192
    },
    {
      "epoch": 7.95484685822545,
      "grad_norm": 4.993614196777344,
      "learning_rate": 3.345875872715153e-05,
      "loss": 0.564,
      "step": 25193
    },
    {
      "epoch": 7.955162614461636,
      "grad_norm": 5.332834720611572,
      "learning_rate": 3.3457881626495456e-05,
      "loss": 0.4558,
      "step": 25194
    },
    {
      "epoch": 7.955478370697821,
      "grad_norm": 5.78618860244751,
      "learning_rate": 3.345700452583939e-05,
      "loss": 0.4652,
      "step": 25195
    },
    {
      "epoch": 7.955794126934007,
      "grad_norm": 5.723803520202637,
      "learning_rate": 3.345612742518331e-05,
      "loss": 0.5987,
      "step": 25196
    },
    {
      "epoch": 7.956109883170193,
      "grad_norm": 6.417935371398926,
      "learning_rate": 3.3455250324527244e-05,
      "loss": 0.5181,
      "step": 25197
    },
    {
      "epoch": 7.956425639406378,
      "grad_norm": 6.038443565368652,
      "learning_rate": 3.345437322387117e-05,
      "loss": 0.5319,
      "step": 25198
    },
    {
      "epoch": 7.956741395642564,
      "grad_norm": 5.821143627166748,
      "learning_rate": 3.34534961232151e-05,
      "loss": 0.5091,
      "step": 25199
    },
    {
      "epoch": 7.957057151878749,
      "grad_norm": 5.85775089263916,
      "learning_rate": 3.3452619022559026e-05,
      "loss": 0.574,
      "step": 25200
    },
    {
      "epoch": 7.957057151878749,
      "eval_loss": 0.9111206531524658,
      "eval_runtime": 4.6099,
      "eval_samples_per_second": 540.142,
      "eval_steps_per_second": 4.555,
      "step": 25200
    },
    {
      "epoch": 7.957372908114936,
      "grad_norm": 6.595793724060059,
      "learning_rate": 3.345174192190296e-05,
      "loss": 0.4984,
      "step": 25201
    },
    {
      "epoch": 7.957688664351121,
      "grad_norm": 4.935374736785889,
      "learning_rate": 3.345086482124688e-05,
      "loss": 0.4299,
      "step": 25202
    },
    {
      "epoch": 7.958004420587306,
      "grad_norm": 6.300991535186768,
      "learning_rate": 3.3449987720590814e-05,
      "loss": 0.521,
      "step": 25203
    },
    {
      "epoch": 7.958320176823492,
      "grad_norm": 6.469217777252197,
      "learning_rate": 3.3449110619934746e-05,
      "loss": 0.6184,
      "step": 25204
    },
    {
      "epoch": 7.958635933059678,
      "grad_norm": 4.504117488861084,
      "learning_rate": 3.344823351927868e-05,
      "loss": 0.3856,
      "step": 25205
    },
    {
      "epoch": 7.958951689295864,
      "grad_norm": 5.164139747619629,
      "learning_rate": 3.34473564186226e-05,
      "loss": 0.5579,
      "step": 25206
    },
    {
      "epoch": 7.959267445532049,
      "grad_norm": 5.543693542480469,
      "learning_rate": 3.3446479317966534e-05,
      "loss": 0.5324,
      "step": 25207
    },
    {
      "epoch": 7.959583201768235,
      "grad_norm": 6.781177043914795,
      "learning_rate": 3.344560221731046e-05,
      "loss": 0.5218,
      "step": 25208
    },
    {
      "epoch": 7.959898958004421,
      "grad_norm": 6.269225120544434,
      "learning_rate": 3.344472511665439e-05,
      "loss": 0.49,
      "step": 25209
    },
    {
      "epoch": 7.960214714240606,
      "grad_norm": 5.027641773223877,
      "learning_rate": 3.344384801599832e-05,
      "loss": 0.5342,
      "step": 25210
    },
    {
      "epoch": 7.960530470476792,
      "grad_norm": 4.808901786804199,
      "learning_rate": 3.344297091534225e-05,
      "loss": 0.5188,
      "step": 25211
    },
    {
      "epoch": 7.960846226712977,
      "grad_norm": 8.388445854187012,
      "learning_rate": 3.344209381468618e-05,
      "loss": 0.619,
      "step": 25212
    },
    {
      "epoch": 7.9611619829491636,
      "grad_norm": 6.41982364654541,
      "learning_rate": 3.3441216714030104e-05,
      "loss": 0.4321,
      "step": 25213
    },
    {
      "epoch": 7.961477739185349,
      "grad_norm": 4.574724197387695,
      "learning_rate": 3.3440339613374036e-05,
      "loss": 0.4616,
      "step": 25214
    },
    {
      "epoch": 7.961793495421535,
      "grad_norm": 5.455817222595215,
      "learning_rate": 3.343946251271796e-05,
      "loss": 0.4626,
      "step": 25215
    },
    {
      "epoch": 7.96210925165772,
      "grad_norm": 5.886575222015381,
      "learning_rate": 3.343858541206189e-05,
      "loss": 0.5623,
      "step": 25216
    },
    {
      "epoch": 7.962425007893906,
      "grad_norm": 5.541059970855713,
      "learning_rate": 3.343770831140582e-05,
      "loss": 0.4365,
      "step": 25217
    },
    {
      "epoch": 7.962740764130092,
      "grad_norm": 6.9691386222839355,
      "learning_rate": 3.343683121074975e-05,
      "loss": 0.5845,
      "step": 25218
    },
    {
      "epoch": 7.963056520366277,
      "grad_norm": 4.830489158630371,
      "learning_rate": 3.3435954110093674e-05,
      "loss": 0.4529,
      "step": 25219
    },
    {
      "epoch": 7.963372276602463,
      "grad_norm": 5.271068572998047,
      "learning_rate": 3.3435077009437605e-05,
      "loss": 0.4856,
      "step": 25220
    },
    {
      "epoch": 7.963688032838649,
      "grad_norm": 4.900447368621826,
      "learning_rate": 3.343419990878153e-05,
      "loss": 0.3945,
      "step": 25221
    },
    {
      "epoch": 7.964003789074834,
      "grad_norm": 6.475874423980713,
      "learning_rate": 3.343332280812546e-05,
      "loss": 0.6546,
      "step": 25222
    },
    {
      "epoch": 7.96431954531102,
      "grad_norm": 7.020246982574463,
      "learning_rate": 3.343244570746939e-05,
      "loss": 0.5928,
      "step": 25223
    },
    {
      "epoch": 7.964635301547205,
      "grad_norm": 4.961475849151611,
      "learning_rate": 3.343156860681332e-05,
      "loss": 0.4955,
      "step": 25224
    },
    {
      "epoch": 7.9649510577833915,
      "grad_norm": 6.622122764587402,
      "learning_rate": 3.3430691506157243e-05,
      "loss": 0.5882,
      "step": 25225
    },
    {
      "epoch": 7.965266814019577,
      "grad_norm": 6.905985355377197,
      "learning_rate": 3.3429814405501175e-05,
      "loss": 0.5305,
      "step": 25226
    },
    {
      "epoch": 7.965582570255762,
      "grad_norm": 5.865495681762695,
      "learning_rate": 3.342893730484511e-05,
      "loss": 0.5242,
      "step": 25227
    },
    {
      "epoch": 7.965898326491948,
      "grad_norm": 5.594522476196289,
      "learning_rate": 3.342806020418903e-05,
      "loss": 0.5582,
      "step": 25228
    },
    {
      "epoch": 7.966214082728134,
      "grad_norm": 5.773643970489502,
      "learning_rate": 3.3427183103532963e-05,
      "loss": 0.5143,
      "step": 25229
    },
    {
      "epoch": 7.96652983896432,
      "grad_norm": 5.245715141296387,
      "learning_rate": 3.342630600287689e-05,
      "loss": 0.4975,
      "step": 25230
    },
    {
      "epoch": 7.966845595200505,
      "grad_norm": 6.43018913269043,
      "learning_rate": 3.342542890222082e-05,
      "loss": 0.4835,
      "step": 25231
    },
    {
      "epoch": 7.967161351436691,
      "grad_norm": 5.7346720695495605,
      "learning_rate": 3.342455180156475e-05,
      "loss": 0.4962,
      "step": 25232
    },
    {
      "epoch": 7.9674771076728765,
      "grad_norm": 5.493285655975342,
      "learning_rate": 3.342367470090868e-05,
      "loss": 0.5543,
      "step": 25233
    },
    {
      "epoch": 7.967792863909062,
      "grad_norm": 5.378920555114746,
      "learning_rate": 3.342279760025261e-05,
      "loss": 0.5198,
      "step": 25234
    },
    {
      "epoch": 7.968108620145248,
      "grad_norm": 6.782376766204834,
      "learning_rate": 3.342192049959654e-05,
      "loss": 0.5233,
      "step": 25235
    },
    {
      "epoch": 7.968424376381433,
      "grad_norm": 4.531407833099365,
      "learning_rate": 3.3421043398940465e-05,
      "loss": 0.4825,
      "step": 25236
    },
    {
      "epoch": 7.9687401326176195,
      "grad_norm": 4.057629108428955,
      "learning_rate": 3.3420166298284397e-05,
      "loss": 0.4014,
      "step": 25237
    },
    {
      "epoch": 7.969055888853805,
      "grad_norm": 4.0626630783081055,
      "learning_rate": 3.341928919762832e-05,
      "loss": 0.351,
      "step": 25238
    },
    {
      "epoch": 7.969371645089991,
      "grad_norm": 5.181682586669922,
      "learning_rate": 3.341841209697225e-05,
      "loss": 0.6394,
      "step": 25239
    },
    {
      "epoch": 7.969687401326176,
      "grad_norm": 6.103519916534424,
      "learning_rate": 3.341753499631618e-05,
      "loss": 0.544,
      "step": 25240
    },
    {
      "epoch": 7.9700031575623616,
      "grad_norm": 4.409172534942627,
      "learning_rate": 3.341665789566011e-05,
      "loss": 0.4843,
      "step": 25241
    },
    {
      "epoch": 7.970318913798548,
      "grad_norm": 4.64935302734375,
      "learning_rate": 3.3415780795004035e-05,
      "loss": 0.359,
      "step": 25242
    },
    {
      "epoch": 7.970634670034733,
      "grad_norm": 5.097007751464844,
      "learning_rate": 3.3414903694347966e-05,
      "loss": 0.5599,
      "step": 25243
    },
    {
      "epoch": 7.970950426270919,
      "grad_norm": 6.120127201080322,
      "learning_rate": 3.341402659369189e-05,
      "loss": 0.5076,
      "step": 25244
    },
    {
      "epoch": 7.9712661825071045,
      "grad_norm": 6.443177223205566,
      "learning_rate": 3.341314949303582e-05,
      "loss": 0.526,
      "step": 25245
    },
    {
      "epoch": 7.97158193874329,
      "grad_norm": 4.791345596313477,
      "learning_rate": 3.341227239237975e-05,
      "loss": 0.4765,
      "step": 25246
    },
    {
      "epoch": 7.971897694979476,
      "grad_norm": 5.9066386222839355,
      "learning_rate": 3.341139529172368e-05,
      "loss": 0.5058,
      "step": 25247
    },
    {
      "epoch": 7.972213451215661,
      "grad_norm": 6.769766330718994,
      "learning_rate": 3.341051819106761e-05,
      "loss": 0.608,
      "step": 25248
    },
    {
      "epoch": 7.9725292074518475,
      "grad_norm": 7.619885444641113,
      "learning_rate": 3.3409641090411536e-05,
      "loss": 0.5738,
      "step": 25249
    },
    {
      "epoch": 7.972844963688033,
      "grad_norm": 6.4831013679504395,
      "learning_rate": 3.340876398975547e-05,
      "loss": 0.5741,
      "step": 25250
    },
    {
      "epoch": 7.973160719924218,
      "grad_norm": 5.212691783905029,
      "learning_rate": 3.340788688909939e-05,
      "loss": 0.5346,
      "step": 25251
    },
    {
      "epoch": 7.973476476160404,
      "grad_norm": 7.917845726013184,
      "learning_rate": 3.3407009788443324e-05,
      "loss": 0.5517,
      "step": 25252
    },
    {
      "epoch": 7.9737922323965895,
      "grad_norm": 5.183855056762695,
      "learning_rate": 3.340613268778725e-05,
      "loss": 0.5092,
      "step": 25253
    },
    {
      "epoch": 7.974107988632776,
      "grad_norm": 5.610924243927002,
      "learning_rate": 3.340525558713118e-05,
      "loss": 0.5179,
      "step": 25254
    },
    {
      "epoch": 7.974423744868961,
      "grad_norm": 5.821362018585205,
      "learning_rate": 3.3404378486475106e-05,
      "loss": 0.455,
      "step": 25255
    },
    {
      "epoch": 7.974739501105147,
      "grad_norm": 4.595622539520264,
      "learning_rate": 3.340350138581904e-05,
      "loss": 0.4217,
      "step": 25256
    },
    {
      "epoch": 7.9750552573413325,
      "grad_norm": 6.531248092651367,
      "learning_rate": 3.340262428516296e-05,
      "loss": 0.4931,
      "step": 25257
    },
    {
      "epoch": 7.975371013577519,
      "grad_norm": 4.652591705322266,
      "learning_rate": 3.3401747184506894e-05,
      "loss": 0.3946,
      "step": 25258
    },
    {
      "epoch": 7.975686769813704,
      "grad_norm": 6.061920166015625,
      "learning_rate": 3.3400870083850826e-05,
      "loss": 0.5418,
      "step": 25259
    },
    {
      "epoch": 7.976002526049889,
      "grad_norm": 7.521754741668701,
      "learning_rate": 3.339999298319476e-05,
      "loss": 0.4875,
      "step": 25260
    },
    {
      "epoch": 7.976318282286075,
      "grad_norm": 4.868498802185059,
      "learning_rate": 3.339911588253868e-05,
      "loss": 0.4417,
      "step": 25261
    },
    {
      "epoch": 7.976634038522261,
      "grad_norm": 5.627039432525635,
      "learning_rate": 3.3398238781882614e-05,
      "loss": 0.5259,
      "step": 25262
    },
    {
      "epoch": 7.976949794758447,
      "grad_norm": 6.54042387008667,
      "learning_rate": 3.339736168122654e-05,
      "loss": 0.6684,
      "step": 25263
    },
    {
      "epoch": 7.977265550994632,
      "grad_norm": 5.4080810546875,
      "learning_rate": 3.339648458057047e-05,
      "loss": 0.4454,
      "step": 25264
    },
    {
      "epoch": 7.9775813072308175,
      "grad_norm": 5.893425941467285,
      "learning_rate": 3.3395607479914396e-05,
      "loss": 0.4416,
      "step": 25265
    },
    {
      "epoch": 7.977897063467004,
      "grad_norm": 5.812243938446045,
      "learning_rate": 3.339473037925833e-05,
      "loss": 0.6534,
      "step": 25266
    },
    {
      "epoch": 7.978212819703189,
      "grad_norm": 5.657212257385254,
      "learning_rate": 3.339385327860225e-05,
      "loss": 0.5046,
      "step": 25267
    },
    {
      "epoch": 7.978528575939375,
      "grad_norm": 5.087594509124756,
      "learning_rate": 3.3392976177946184e-05,
      "loss": 0.5209,
      "step": 25268
    },
    {
      "epoch": 7.97884433217556,
      "grad_norm": 5.079836845397949,
      "learning_rate": 3.339209907729011e-05,
      "loss": 0.4747,
      "step": 25269
    },
    {
      "epoch": 7.979160088411746,
      "grad_norm": 5.164479732513428,
      "learning_rate": 3.339122197663404e-05,
      "loss": 0.4719,
      "step": 25270
    },
    {
      "epoch": 7.979475844647932,
      "grad_norm": 5.3838396072387695,
      "learning_rate": 3.339034487597797e-05,
      "loss": 0.4766,
      "step": 25271
    },
    {
      "epoch": 7.979791600884117,
      "grad_norm": 5.851734161376953,
      "learning_rate": 3.33894677753219e-05,
      "loss": 0.6679,
      "step": 25272
    },
    {
      "epoch": 7.980107357120303,
      "grad_norm": 5.552962303161621,
      "learning_rate": 3.338859067466583e-05,
      "loss": 0.4976,
      "step": 25273
    },
    {
      "epoch": 7.980423113356489,
      "grad_norm": 6.957499027252197,
      "learning_rate": 3.3387713574009754e-05,
      "loss": 0.6286,
      "step": 25274
    },
    {
      "epoch": 7.980738869592674,
      "grad_norm": 5.456080913543701,
      "learning_rate": 3.3386836473353685e-05,
      "loss": 0.4372,
      "step": 25275
    },
    {
      "epoch": 7.98105462582886,
      "grad_norm": 5.821036338806152,
      "learning_rate": 3.338595937269761e-05,
      "loss": 0.5008,
      "step": 25276
    },
    {
      "epoch": 7.9813703820650455,
      "grad_norm": 5.638287544250488,
      "learning_rate": 3.338508227204154e-05,
      "loss": 0.488,
      "step": 25277
    },
    {
      "epoch": 7.981686138301232,
      "grad_norm": 6.811020374298096,
      "learning_rate": 3.338420517138547e-05,
      "loss": 0.3859,
      "step": 25278
    },
    {
      "epoch": 7.982001894537417,
      "grad_norm": 5.930424690246582,
      "learning_rate": 3.33833280707294e-05,
      "loss": 0.5125,
      "step": 25279
    },
    {
      "epoch": 7.982317650773603,
      "grad_norm": 5.274215221405029,
      "learning_rate": 3.338245097007332e-05,
      "loss": 0.5166,
      "step": 25280
    },
    {
      "epoch": 7.982633407009788,
      "grad_norm": 7.23173189163208,
      "learning_rate": 3.3381573869417255e-05,
      "loss": 0.5545,
      "step": 25281
    },
    {
      "epoch": 7.982949163245975,
      "grad_norm": 5.297033786773682,
      "learning_rate": 3.338069676876118e-05,
      "loss": 0.4423,
      "step": 25282
    },
    {
      "epoch": 7.98326491948216,
      "grad_norm": 5.376876354217529,
      "learning_rate": 3.337981966810511e-05,
      "loss": 0.5305,
      "step": 25283
    },
    {
      "epoch": 7.983580675718345,
      "grad_norm": 5.464635372161865,
      "learning_rate": 3.3378942567449037e-05,
      "loss": 0.5631,
      "step": 25284
    },
    {
      "epoch": 7.983896431954531,
      "grad_norm": 7.443291187286377,
      "learning_rate": 3.337806546679297e-05,
      "loss": 0.4519,
      "step": 25285
    },
    {
      "epoch": 7.984212188190717,
      "grad_norm": 5.54359245300293,
      "learning_rate": 3.33771883661369e-05,
      "loss": 0.5361,
      "step": 25286
    },
    {
      "epoch": 7.984527944426903,
      "grad_norm": 5.340054512023926,
      "learning_rate": 3.337631126548083e-05,
      "loss": 0.4777,
      "step": 25287
    },
    {
      "epoch": 7.984843700663088,
      "grad_norm": 5.54115629196167,
      "learning_rate": 3.3375434164824756e-05,
      "loss": 0.6303,
      "step": 25288
    },
    {
      "epoch": 7.985159456899273,
      "grad_norm": 5.2541422843933105,
      "learning_rate": 3.337455706416869e-05,
      "loss": 0.3873,
      "step": 25289
    },
    {
      "epoch": 7.98547521313546,
      "grad_norm": 6.488763332366943,
      "learning_rate": 3.337367996351261e-05,
      "loss": 0.4888,
      "step": 25290
    },
    {
      "epoch": 7.985790969371645,
      "grad_norm": 6.905250549316406,
      "learning_rate": 3.3372802862856545e-05,
      "loss": 0.5104,
      "step": 25291
    },
    {
      "epoch": 7.986106725607831,
      "grad_norm": 7.227046489715576,
      "learning_rate": 3.3371925762200476e-05,
      "loss": 0.5138,
      "step": 25292
    },
    {
      "epoch": 7.986422481844016,
      "grad_norm": 6.34807014465332,
      "learning_rate": 3.33710486615444e-05,
      "loss": 0.4461,
      "step": 25293
    },
    {
      "epoch": 7.986738238080202,
      "grad_norm": 4.4152326583862305,
      "learning_rate": 3.337017156088833e-05,
      "loss": 0.4183,
      "step": 25294
    },
    {
      "epoch": 7.987053994316388,
      "grad_norm": 5.583194732666016,
      "learning_rate": 3.336929446023226e-05,
      "loss": 0.6554,
      "step": 25295
    },
    {
      "epoch": 7.987369750552573,
      "grad_norm": 4.925952911376953,
      "learning_rate": 3.336841735957619e-05,
      "loss": 0.4339,
      "step": 25296
    },
    {
      "epoch": 7.987685506788759,
      "grad_norm": 5.430541038513184,
      "learning_rate": 3.3367540258920115e-05,
      "loss": 0.4969,
      "step": 25297
    },
    {
      "epoch": 7.988001263024945,
      "grad_norm": 7.194968223571777,
      "learning_rate": 3.3366663158264046e-05,
      "loss": 0.4917,
      "step": 25298
    },
    {
      "epoch": 7.98831701926113,
      "grad_norm": 7.763960838317871,
      "learning_rate": 3.336578605760797e-05,
      "loss": 0.5781,
      "step": 25299
    },
    {
      "epoch": 7.988632775497316,
      "grad_norm": 4.901101112365723,
      "learning_rate": 3.33649089569519e-05,
      "loss": 0.4691,
      "step": 25300
    },
    {
      "epoch": 7.988948531733501,
      "grad_norm": 5.744428634643555,
      "learning_rate": 3.336403185629583e-05,
      "loss": 0.5311,
      "step": 25301
    },
    {
      "epoch": 7.989264287969688,
      "grad_norm": 7.036128997802734,
      "learning_rate": 3.336315475563976e-05,
      "loss": 0.4934,
      "step": 25302
    },
    {
      "epoch": 7.989580044205873,
      "grad_norm": 6.111357688903809,
      "learning_rate": 3.3362277654983684e-05,
      "loss": 0.5226,
      "step": 25303
    },
    {
      "epoch": 7.989895800442059,
      "grad_norm": 6.745138168334961,
      "learning_rate": 3.3361400554327616e-05,
      "loss": 0.4327,
      "step": 25304
    },
    {
      "epoch": 7.990211556678244,
      "grad_norm": 5.966529846191406,
      "learning_rate": 3.336052345367154e-05,
      "loss": 0.4995,
      "step": 25305
    },
    {
      "epoch": 7.9905273129144305,
      "grad_norm": 6.542387962341309,
      "learning_rate": 3.335964635301547e-05,
      "loss": 0.5361,
      "step": 25306
    },
    {
      "epoch": 7.990843069150616,
      "grad_norm": 5.1625776290893555,
      "learning_rate": 3.3358769252359404e-05,
      "loss": 0.574,
      "step": 25307
    },
    {
      "epoch": 7.991158825386801,
      "grad_norm": 4.456498622894287,
      "learning_rate": 3.335789215170333e-05,
      "loss": 0.4745,
      "step": 25308
    },
    {
      "epoch": 7.991474581622987,
      "grad_norm": 5.416578769683838,
      "learning_rate": 3.335701505104726e-05,
      "loss": 0.5086,
      "step": 25309
    },
    {
      "epoch": 7.991790337859173,
      "grad_norm": 5.51607084274292,
      "learning_rate": 3.3356137950391186e-05,
      "loss": 0.5302,
      "step": 25310
    },
    {
      "epoch": 7.992106094095359,
      "grad_norm": 5.258428573608398,
      "learning_rate": 3.335526084973512e-05,
      "loss": 0.581,
      "step": 25311
    },
    {
      "epoch": 7.992421850331544,
      "grad_norm": 3.962549924850464,
      "learning_rate": 3.335438374907904e-05,
      "loss": 0.4437,
      "step": 25312
    },
    {
      "epoch": 7.992737606567729,
      "grad_norm": 6.096534252166748,
      "learning_rate": 3.3353506648422974e-05,
      "loss": 0.5438,
      "step": 25313
    },
    {
      "epoch": 7.9930533628039155,
      "grad_norm": 5.062343120574951,
      "learning_rate": 3.3352629547766906e-05,
      "loss": 0.4135,
      "step": 25314
    },
    {
      "epoch": 7.993369119040101,
      "grad_norm": 5.066215991973877,
      "learning_rate": 3.335175244711084e-05,
      "loss": 0.4144,
      "step": 25315
    },
    {
      "epoch": 7.993684875276287,
      "grad_norm": 5.419503211975098,
      "learning_rate": 3.335087534645476e-05,
      "loss": 0.4492,
      "step": 25316
    },
    {
      "epoch": 7.994000631512472,
      "grad_norm": 8.137663841247559,
      "learning_rate": 3.3349998245798694e-05,
      "loss": 0.5787,
      "step": 25317
    },
    {
      "epoch": 7.994316387748658,
      "grad_norm": 6.25880765914917,
      "learning_rate": 3.334912114514262e-05,
      "loss": 0.5452,
      "step": 25318
    },
    {
      "epoch": 7.994632143984844,
      "grad_norm": 7.034809589385986,
      "learning_rate": 3.334824404448655e-05,
      "loss": 0.5821,
      "step": 25319
    },
    {
      "epoch": 7.994947900221029,
      "grad_norm": 6.821280479431152,
      "learning_rate": 3.3347366943830475e-05,
      "loss": 0.5684,
      "step": 25320
    },
    {
      "epoch": 7.995263656457215,
      "grad_norm": 5.767335891723633,
      "learning_rate": 3.334648984317441e-05,
      "loss": 0.4287,
      "step": 25321
    },
    {
      "epoch": 7.995579412693401,
      "grad_norm": 5.1493659019470215,
      "learning_rate": 3.334561274251833e-05,
      "loss": 0.4445,
      "step": 25322
    },
    {
      "epoch": 7.995895168929587,
      "grad_norm": 6.04866886138916,
      "learning_rate": 3.3344735641862264e-05,
      "loss": 0.4948,
      "step": 25323
    },
    {
      "epoch": 7.996210925165772,
      "grad_norm": 9.115906715393066,
      "learning_rate": 3.334385854120619e-05,
      "loss": 0.4917,
      "step": 25324
    },
    {
      "epoch": 7.996526681401957,
      "grad_norm": 7.315640926361084,
      "learning_rate": 3.334298144055012e-05,
      "loss": 0.5583,
      "step": 25325
    },
    {
      "epoch": 7.9968424376381435,
      "grad_norm": 6.318122863769531,
      "learning_rate": 3.3342104339894045e-05,
      "loss": 0.5524,
      "step": 25326
    },
    {
      "epoch": 7.997158193874329,
      "grad_norm": 6.915225982666016,
      "learning_rate": 3.334122723923798e-05,
      "loss": 0.5631,
      "step": 25327
    },
    {
      "epoch": 7.997473950110515,
      "grad_norm": 6.195561408996582,
      "learning_rate": 3.33403501385819e-05,
      "loss": 0.4915,
      "step": 25328
    },
    {
      "epoch": 7.9977897063467,
      "grad_norm": 5.388145923614502,
      "learning_rate": 3.3339473037925833e-05,
      "loss": 0.5786,
      "step": 25329
    },
    {
      "epoch": 7.9981054625828865,
      "grad_norm": 6.311473846435547,
      "learning_rate": 3.3338595937269765e-05,
      "loss": 0.4974,
      "step": 25330
    },
    {
      "epoch": 7.998421218819072,
      "grad_norm": 7.908024311065674,
      "learning_rate": 3.333771883661369e-05,
      "loss": 0.4861,
      "step": 25331
    },
    {
      "epoch": 7.998736975055257,
      "grad_norm": 6.595334053039551,
      "learning_rate": 3.333684173595762e-05,
      "loss": 0.5424,
      "step": 25332
    },
    {
      "epoch": 7.999052731291443,
      "grad_norm": 5.756217956542969,
      "learning_rate": 3.333596463530155e-05,
      "loss": 0.479,
      "step": 25333
    },
    {
      "epoch": 7.9993684875276285,
      "grad_norm": 4.681522846221924,
      "learning_rate": 3.333508753464548e-05,
      "loss": 0.3613,
      "step": 25334
    },
    {
      "epoch": 7.999684243763815,
      "grad_norm": 6.579835414886475,
      "learning_rate": 3.33342104339894e-05,
      "loss": 0.5949,
      "step": 25335
    },
    {
      "epoch": 8.0,
      "grad_norm": 5.860681056976318,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.4875,
      "step": 25336
    },
    {
      "epoch": 8.000315756236185,
      "grad_norm": 4.837512969970703,
      "learning_rate": 3.333245623267726e-05,
      "loss": 0.4406,
      "step": 25337
    },
    {
      "epoch": 8.00063151247237,
      "grad_norm": 4.5498785972595215,
      "learning_rate": 3.333157913202119e-05,
      "loss": 0.4015,
      "step": 25338
    },
    {
      "epoch": 8.000947268708558,
      "grad_norm": 4.5034098625183105,
      "learning_rate": 3.3330702031365116e-05,
      "loss": 0.408,
      "step": 25339
    },
    {
      "epoch": 8.001263024944743,
      "grad_norm": 4.091699123382568,
      "learning_rate": 3.332982493070905e-05,
      "loss": 0.441,
      "step": 25340
    },
    {
      "epoch": 8.001578781180928,
      "grad_norm": 3.3259434700012207,
      "learning_rate": 3.332894783005298e-05,
      "loss": 0.3164,
      "step": 25341
    },
    {
      "epoch": 8.001894537417114,
      "grad_norm": 4.5760111808776855,
      "learning_rate": 3.332807072939691e-05,
      "loss": 0.4125,
      "step": 25342
    },
    {
      "epoch": 8.002210293653299,
      "grad_norm": 4.739762783050537,
      "learning_rate": 3.3327193628740836e-05,
      "loss": 0.3982,
      "step": 25343
    },
    {
      "epoch": 8.002526049889486,
      "grad_norm": 4.630195617675781,
      "learning_rate": 3.332631652808477e-05,
      "loss": 0.4164,
      "step": 25344
    },
    {
      "epoch": 8.002841806125671,
      "grad_norm": 5.191437721252441,
      "learning_rate": 3.332543942742869e-05,
      "loss": 0.3691,
      "step": 25345
    },
    {
      "epoch": 8.003157562361856,
      "grad_norm": 5.661723613739014,
      "learning_rate": 3.3324562326772625e-05,
      "loss": 0.4975,
      "step": 25346
    },
    {
      "epoch": 8.003473318598042,
      "grad_norm": 4.711858749389648,
      "learning_rate": 3.332368522611655e-05,
      "loss": 0.3577,
      "step": 25347
    },
    {
      "epoch": 8.003789074834229,
      "grad_norm": 5.105906963348389,
      "learning_rate": 3.332280812546048e-05,
      "loss": 0.4088,
      "step": 25348
    },
    {
      "epoch": 8.004104831070414,
      "grad_norm": 5.137452602386475,
      "learning_rate": 3.3321931024804406e-05,
      "loss": 0.3672,
      "step": 25349
    },
    {
      "epoch": 8.0044205873066,
      "grad_norm": 6.165457725524902,
      "learning_rate": 3.332105392414834e-05,
      "loss": 0.3997,
      "step": 25350
    },
    {
      "epoch": 8.004736343542785,
      "grad_norm": 5.381542682647705,
      "learning_rate": 3.332017682349227e-05,
      "loss": 0.3634,
      "step": 25351
    },
    {
      "epoch": 8.00505209977897,
      "grad_norm": 4.140751361846924,
      "learning_rate": 3.3319299722836194e-05,
      "loss": 0.3346,
      "step": 25352
    },
    {
      "epoch": 8.005367856015157,
      "grad_norm": 7.090455532073975,
      "learning_rate": 3.3318422622180126e-05,
      "loss": 0.4544,
      "step": 25353
    },
    {
      "epoch": 8.005683612251342,
      "grad_norm": 4.579223155975342,
      "learning_rate": 3.331754552152405e-05,
      "loss": 0.3826,
      "step": 25354
    },
    {
      "epoch": 8.005999368487528,
      "grad_norm": 4.299929141998291,
      "learning_rate": 3.331666842086798e-05,
      "loss": 0.422,
      "step": 25355
    },
    {
      "epoch": 8.006315124723713,
      "grad_norm": 4.85601806640625,
      "learning_rate": 3.331579132021191e-05,
      "loss": 0.3691,
      "step": 25356
    },
    {
      "epoch": 8.006630880959898,
      "grad_norm": 5.4140448570251465,
      "learning_rate": 3.331491421955584e-05,
      "loss": 0.4066,
      "step": 25357
    },
    {
      "epoch": 8.006946637196085,
      "grad_norm": 4.502094268798828,
      "learning_rate": 3.3314037118899764e-05,
      "loss": 0.3146,
      "step": 25358
    },
    {
      "epoch": 8.00726239343227,
      "grad_norm": 5.772599220275879,
      "learning_rate": 3.3313160018243696e-05,
      "loss": 0.4673,
      "step": 25359
    },
    {
      "epoch": 8.007578149668456,
      "grad_norm": 7.042736530303955,
      "learning_rate": 3.331228291758762e-05,
      "loss": 0.4373,
      "step": 25360
    },
    {
      "epoch": 8.007893905904641,
      "grad_norm": 5.094088077545166,
      "learning_rate": 3.331140581693155e-05,
      "loss": 0.3753,
      "step": 25361
    },
    {
      "epoch": 8.008209662140827,
      "grad_norm": 4.96348237991333,
      "learning_rate": 3.331052871627548e-05,
      "loss": 0.3387,
      "step": 25362
    },
    {
      "epoch": 8.008525418377014,
      "grad_norm": 5.248043060302734,
      "learning_rate": 3.330965161561941e-05,
      "loss": 0.3891,
      "step": 25363
    },
    {
      "epoch": 8.008841174613199,
      "grad_norm": 5.143496513366699,
      "learning_rate": 3.3308774514963334e-05,
      "loss": 0.3795,
      "step": 25364
    },
    {
      "epoch": 8.009156930849384,
      "grad_norm": 5.052576065063477,
      "learning_rate": 3.3307897414307266e-05,
      "loss": 0.4593,
      "step": 25365
    },
    {
      "epoch": 8.00947268708557,
      "grad_norm": 4.9333648681640625,
      "learning_rate": 3.330702031365119e-05,
      "loss": 0.4233,
      "step": 25366
    },
    {
      "epoch": 8.009788443321755,
      "grad_norm": 5.5848493576049805,
      "learning_rate": 3.330614321299512e-05,
      "loss": 0.3946,
      "step": 25367
    },
    {
      "epoch": 8.010104199557942,
      "grad_norm": 4.34079647064209,
      "learning_rate": 3.3305266112339054e-05,
      "loss": 0.3152,
      "step": 25368
    },
    {
      "epoch": 8.010419955794127,
      "grad_norm": 4.7676191329956055,
      "learning_rate": 3.3304389011682986e-05,
      "loss": 0.3141,
      "step": 25369
    },
    {
      "epoch": 8.010735712030312,
      "grad_norm": 6.570499897003174,
      "learning_rate": 3.330351191102691e-05,
      "loss": 0.3915,
      "step": 25370
    },
    {
      "epoch": 8.011051468266498,
      "grad_norm": 6.396994113922119,
      "learning_rate": 3.330263481037084e-05,
      "loss": 0.442,
      "step": 25371
    },
    {
      "epoch": 8.011367224502685,
      "grad_norm": 4.249238014221191,
      "learning_rate": 3.330175770971477e-05,
      "loss": 0.3211,
      "step": 25372
    },
    {
      "epoch": 8.01168298073887,
      "grad_norm": 7.474259376525879,
      "learning_rate": 3.33008806090587e-05,
      "loss": 0.418,
      "step": 25373
    },
    {
      "epoch": 8.011998736975055,
      "grad_norm": 6.542509078979492,
      "learning_rate": 3.330000350840263e-05,
      "loss": 0.4407,
      "step": 25374
    },
    {
      "epoch": 8.01231449321124,
      "grad_norm": 6.301709175109863,
      "learning_rate": 3.3299126407746555e-05,
      "loss": 0.494,
      "step": 25375
    },
    {
      "epoch": 8.012630249447426,
      "grad_norm": 5.2169389724731445,
      "learning_rate": 3.329824930709049e-05,
      "loss": 0.3998,
      "step": 25376
    },
    {
      "epoch": 8.012946005683613,
      "grad_norm": 6.3245062828063965,
      "learning_rate": 3.329737220643441e-05,
      "loss": 0.5319,
      "step": 25377
    },
    {
      "epoch": 8.013261761919798,
      "grad_norm": 6.000221252441406,
      "learning_rate": 3.3296495105778344e-05,
      "loss": 0.4152,
      "step": 25378
    },
    {
      "epoch": 8.013577518155984,
      "grad_norm": 6.178860664367676,
      "learning_rate": 3.329561800512227e-05,
      "loss": 0.4122,
      "step": 25379
    },
    {
      "epoch": 8.013893274392169,
      "grad_norm": 5.515242099761963,
      "learning_rate": 3.32947409044662e-05,
      "loss": 0.3507,
      "step": 25380
    },
    {
      "epoch": 8.014209030628354,
      "grad_norm": 5.784034729003906,
      "learning_rate": 3.3293863803810125e-05,
      "loss": 0.4416,
      "step": 25381
    },
    {
      "epoch": 8.014524786864541,
      "grad_norm": 6.542550086975098,
      "learning_rate": 3.329298670315406e-05,
      "loss": 0.4001,
      "step": 25382
    },
    {
      "epoch": 8.014840543100727,
      "grad_norm": 4.728296279907227,
      "learning_rate": 3.329210960249798e-05,
      "loss": 0.3806,
      "step": 25383
    },
    {
      "epoch": 8.015156299336912,
      "grad_norm": 5.6387739181518555,
      "learning_rate": 3.329123250184191e-05,
      "loss": 0.359,
      "step": 25384
    },
    {
      "epoch": 8.015472055573097,
      "grad_norm": 5.009937763214111,
      "learning_rate": 3.329035540118584e-05,
      "loss": 0.3992,
      "step": 25385
    },
    {
      "epoch": 8.015787811809282,
      "grad_norm": 4.495936870574951,
      "learning_rate": 3.328947830052977e-05,
      "loss": 0.3062,
      "step": 25386
    },
    {
      "epoch": 8.01610356804547,
      "grad_norm": 4.117343902587891,
      "learning_rate": 3.3288601199873695e-05,
      "loss": 0.3583,
      "step": 25387
    },
    {
      "epoch": 8.016419324281655,
      "grad_norm": 6.755296230316162,
      "learning_rate": 3.3287724099217627e-05,
      "loss": 0.4567,
      "step": 25388
    },
    {
      "epoch": 8.01673508051784,
      "grad_norm": 6.122472286224365,
      "learning_rate": 3.328684699856156e-05,
      "loss": 0.4117,
      "step": 25389
    },
    {
      "epoch": 8.017050836754025,
      "grad_norm": 8.789450645446777,
      "learning_rate": 3.328596989790548e-05,
      "loss": 0.4473,
      "step": 25390
    },
    {
      "epoch": 8.01736659299021,
      "grad_norm": 7.41603422164917,
      "learning_rate": 3.3285092797249415e-05,
      "loss": 0.4963,
      "step": 25391
    },
    {
      "epoch": 8.017682349226398,
      "grad_norm": 8.249232292175293,
      "learning_rate": 3.328421569659334e-05,
      "loss": 0.4941,
      "step": 25392
    },
    {
      "epoch": 8.017998105462583,
      "grad_norm": 5.957586765289307,
      "learning_rate": 3.328333859593727e-05,
      "loss": 0.4364,
      "step": 25393
    },
    {
      "epoch": 8.018313861698768,
      "grad_norm": 7.321343898773193,
      "learning_rate": 3.3282461495281196e-05,
      "loss": 0.4188,
      "step": 25394
    },
    {
      "epoch": 8.018629617934954,
      "grad_norm": 6.997237682342529,
      "learning_rate": 3.328158439462513e-05,
      "loss": 0.4146,
      "step": 25395
    },
    {
      "epoch": 8.01894537417114,
      "grad_norm": 6.128035068511963,
      "learning_rate": 3.328070729396906e-05,
      "loss": 0.3404,
      "step": 25396
    },
    {
      "epoch": 8.019261130407326,
      "grad_norm": 5.74287748336792,
      "learning_rate": 3.327983019331299e-05,
      "loss": 0.3363,
      "step": 25397
    },
    {
      "epoch": 8.019576886643511,
      "grad_norm": 6.341186046600342,
      "learning_rate": 3.3278953092656916e-05,
      "loss": 0.3802,
      "step": 25398
    },
    {
      "epoch": 8.019892642879697,
      "grad_norm": 6.193948268890381,
      "learning_rate": 3.327807599200085e-05,
      "loss": 0.3929,
      "step": 25399
    },
    {
      "epoch": 8.020208399115882,
      "grad_norm": 5.906247138977051,
      "learning_rate": 3.327719889134477e-05,
      "loss": 0.3584,
      "step": 25400
    },
    {
      "epoch": 8.020208399115882,
      "eval_loss": 0.980355978012085,
      "eval_runtime": 4.5985,
      "eval_samples_per_second": 541.484,
      "eval_steps_per_second": 4.567,
      "step": 25400
    },
    {
      "epoch": 8.020524155352069,
      "grad_norm": 4.898906230926514,
      "learning_rate": 3.3276321790688705e-05,
      "loss": 0.4571,
      "step": 25401
    },
    {
      "epoch": 8.020839911588254,
      "grad_norm": 5.371743202209473,
      "learning_rate": 3.327544469003263e-05,
      "loss": 0.3292,
      "step": 25402
    },
    {
      "epoch": 8.02115566782444,
      "grad_norm": 4.580621719360352,
      "learning_rate": 3.327456758937656e-05,
      "loss": 0.3729,
      "step": 25403
    },
    {
      "epoch": 8.021471424060625,
      "grad_norm": 6.046999454498291,
      "learning_rate": 3.3273690488720486e-05,
      "loss": 0.4348,
      "step": 25404
    },
    {
      "epoch": 8.02178718029681,
      "grad_norm": 6.725062847137451,
      "learning_rate": 3.327281338806442e-05,
      "loss": 0.4675,
      "step": 25405
    },
    {
      "epoch": 8.022102936532997,
      "grad_norm": 5.135830879211426,
      "learning_rate": 3.327193628740834e-05,
      "loss": 0.2883,
      "step": 25406
    },
    {
      "epoch": 8.022418692769183,
      "grad_norm": 5.768505573272705,
      "learning_rate": 3.3271059186752274e-05,
      "loss": 0.371,
      "step": 25407
    },
    {
      "epoch": 8.022734449005368,
      "grad_norm": 7.192331314086914,
      "learning_rate": 3.32701820860962e-05,
      "loss": 0.3714,
      "step": 25408
    },
    {
      "epoch": 8.023050205241553,
      "grad_norm": 6.757851600646973,
      "learning_rate": 3.326930498544013e-05,
      "loss": 0.401,
      "step": 25409
    },
    {
      "epoch": 8.023365961477738,
      "grad_norm": 6.296507358551025,
      "learning_rate": 3.3268427884784056e-05,
      "loss": 0.5021,
      "step": 25410
    },
    {
      "epoch": 8.023681717713925,
      "grad_norm": 5.334496974945068,
      "learning_rate": 3.326755078412799e-05,
      "loss": 0.3536,
      "step": 25411
    },
    {
      "epoch": 8.02399747395011,
      "grad_norm": 4.728350639343262,
      "learning_rate": 3.326667368347192e-05,
      "loss": 0.4206,
      "step": 25412
    },
    {
      "epoch": 8.024313230186296,
      "grad_norm": 7.175784587860107,
      "learning_rate": 3.3265796582815844e-05,
      "loss": 0.4023,
      "step": 25413
    },
    {
      "epoch": 8.024628986422481,
      "grad_norm": 5.967340469360352,
      "learning_rate": 3.3264919482159776e-05,
      "loss": 0.4061,
      "step": 25414
    },
    {
      "epoch": 8.024944742658667,
      "grad_norm": 4.034686088562012,
      "learning_rate": 3.32640423815037e-05,
      "loss": 0.3109,
      "step": 25415
    },
    {
      "epoch": 8.025260498894854,
      "grad_norm": 4.644827842712402,
      "learning_rate": 3.326316528084763e-05,
      "loss": 0.3275,
      "step": 25416
    },
    {
      "epoch": 8.025576255131039,
      "grad_norm": 7.392913341522217,
      "learning_rate": 3.326228818019156e-05,
      "loss": 0.4389,
      "step": 25417
    },
    {
      "epoch": 8.025892011367224,
      "grad_norm": 5.196203231811523,
      "learning_rate": 3.326141107953549e-05,
      "loss": 0.4218,
      "step": 25418
    },
    {
      "epoch": 8.02620776760341,
      "grad_norm": 5.424672603607178,
      "learning_rate": 3.3260533978879414e-05,
      "loss": 0.4188,
      "step": 25419
    },
    {
      "epoch": 8.026523523839597,
      "grad_norm": 5.486503601074219,
      "learning_rate": 3.3259656878223346e-05,
      "loss": 0.4071,
      "step": 25420
    },
    {
      "epoch": 8.026839280075782,
      "grad_norm": 5.179703235626221,
      "learning_rate": 3.325877977756727e-05,
      "loss": 0.3542,
      "step": 25421
    },
    {
      "epoch": 8.027155036311967,
      "grad_norm": 5.094493389129639,
      "learning_rate": 3.32579026769112e-05,
      "loss": 0.3414,
      "step": 25422
    },
    {
      "epoch": 8.027470792548153,
      "grad_norm": 6.9491682052612305,
      "learning_rate": 3.3257025576255134e-05,
      "loss": 0.3479,
      "step": 25423
    },
    {
      "epoch": 8.027786548784338,
      "grad_norm": 4.768853187561035,
      "learning_rate": 3.3256148475599065e-05,
      "loss": 0.3076,
      "step": 25424
    },
    {
      "epoch": 8.028102305020525,
      "grad_norm": 5.819995403289795,
      "learning_rate": 3.325527137494299e-05,
      "loss": 0.4476,
      "step": 25425
    },
    {
      "epoch": 8.02841806125671,
      "grad_norm": 5.2154860496521,
      "learning_rate": 3.325439427428692e-05,
      "loss": 0.4067,
      "step": 25426
    },
    {
      "epoch": 8.028733817492895,
      "grad_norm": 5.236615180969238,
      "learning_rate": 3.325351717363085e-05,
      "loss": 0.3492,
      "step": 25427
    },
    {
      "epoch": 8.02904957372908,
      "grad_norm": 5.380213260650635,
      "learning_rate": 3.325264007297478e-05,
      "loss": 0.3937,
      "step": 25428
    },
    {
      "epoch": 8.029365329965266,
      "grad_norm": 5.482794761657715,
      "learning_rate": 3.3251762972318704e-05,
      "loss": 0.3652,
      "step": 25429
    },
    {
      "epoch": 8.029681086201453,
      "grad_norm": 6.986404895782471,
      "learning_rate": 3.3250885871662635e-05,
      "loss": 0.4994,
      "step": 25430
    },
    {
      "epoch": 8.029996842437638,
      "grad_norm": 5.636473655700684,
      "learning_rate": 3.325000877100656e-05,
      "loss": 0.2899,
      "step": 25431
    },
    {
      "epoch": 8.030312598673824,
      "grad_norm": 5.331287860870361,
      "learning_rate": 3.324913167035049e-05,
      "loss": 0.4757,
      "step": 25432
    },
    {
      "epoch": 8.030628354910009,
      "grad_norm": 7.800785064697266,
      "learning_rate": 3.3248254569694424e-05,
      "loss": 0.4677,
      "step": 25433
    },
    {
      "epoch": 8.030944111146194,
      "grad_norm": 4.779550075531006,
      "learning_rate": 3.324737746903835e-05,
      "loss": 0.3072,
      "step": 25434
    },
    {
      "epoch": 8.031259867382381,
      "grad_norm": 5.569695949554443,
      "learning_rate": 3.324650036838228e-05,
      "loss": 0.3958,
      "step": 25435
    },
    {
      "epoch": 8.031575623618567,
      "grad_norm": 6.889438629150391,
      "learning_rate": 3.3245623267726205e-05,
      "loss": 0.4469,
      "step": 25436
    },
    {
      "epoch": 8.031891379854752,
      "grad_norm": 5.310381889343262,
      "learning_rate": 3.324474616707014e-05,
      "loss": 0.4085,
      "step": 25437
    },
    {
      "epoch": 8.032207136090937,
      "grad_norm": 5.745708465576172,
      "learning_rate": 3.324386906641406e-05,
      "loss": 0.4387,
      "step": 25438
    },
    {
      "epoch": 8.032522892327123,
      "grad_norm": 6.023251533508301,
      "learning_rate": 3.324299196575799e-05,
      "loss": 0.3958,
      "step": 25439
    },
    {
      "epoch": 8.03283864856331,
      "grad_norm": 4.728291988372803,
      "learning_rate": 3.324211486510192e-05,
      "loss": 0.3507,
      "step": 25440
    },
    {
      "epoch": 8.033154404799495,
      "grad_norm": 4.798748970031738,
      "learning_rate": 3.324123776444585e-05,
      "loss": 0.3472,
      "step": 25441
    },
    {
      "epoch": 8.03347016103568,
      "grad_norm": 6.399054527282715,
      "learning_rate": 3.3240360663789775e-05,
      "loss": 0.4349,
      "step": 25442
    },
    {
      "epoch": 8.033785917271866,
      "grad_norm": 5.648554801940918,
      "learning_rate": 3.3239483563133706e-05,
      "loss": 0.3647,
      "step": 25443
    },
    {
      "epoch": 8.034101673508053,
      "grad_norm": 4.372868061065674,
      "learning_rate": 3.323860646247763e-05,
      "loss": 0.3547,
      "step": 25444
    },
    {
      "epoch": 8.034417429744238,
      "grad_norm": 6.9545722007751465,
      "learning_rate": 3.323772936182156e-05,
      "loss": 0.3151,
      "step": 25445
    },
    {
      "epoch": 8.034733185980423,
      "grad_norm": 6.250122547149658,
      "learning_rate": 3.323685226116549e-05,
      "loss": 0.4034,
      "step": 25446
    },
    {
      "epoch": 8.035048942216608,
      "grad_norm": 5.621817111968994,
      "learning_rate": 3.323597516050942e-05,
      "loss": 0.3996,
      "step": 25447
    },
    {
      "epoch": 8.035364698452794,
      "grad_norm": 5.599437236785889,
      "learning_rate": 3.3235098059853345e-05,
      "loss": 0.4225,
      "step": 25448
    },
    {
      "epoch": 8.03568045468898,
      "grad_norm": 5.68863582611084,
      "learning_rate": 3.3234220959197276e-05,
      "loss": 0.3483,
      "step": 25449
    },
    {
      "epoch": 8.035996210925166,
      "grad_norm": 6.314027309417725,
      "learning_rate": 3.323334385854121e-05,
      "loss": 0.438,
      "step": 25450
    },
    {
      "epoch": 8.036311967161351,
      "grad_norm": 6.648703575134277,
      "learning_rate": 3.323246675788514e-05,
      "loss": 0.3929,
      "step": 25451
    },
    {
      "epoch": 8.036627723397537,
      "grad_norm": 5.963940143585205,
      "learning_rate": 3.3231589657229064e-05,
      "loss": 0.3834,
      "step": 25452
    },
    {
      "epoch": 8.036943479633722,
      "grad_norm": 5.554579257965088,
      "learning_rate": 3.3230712556572996e-05,
      "loss": 0.4784,
      "step": 25453
    },
    {
      "epoch": 8.037259235869909,
      "grad_norm": 4.891551971435547,
      "learning_rate": 3.322983545591693e-05,
      "loss": 0.3387,
      "step": 25454
    },
    {
      "epoch": 8.037574992106094,
      "grad_norm": 6.64348840713501,
      "learning_rate": 3.322895835526085e-05,
      "loss": 0.4599,
      "step": 25455
    },
    {
      "epoch": 8.03789074834228,
      "grad_norm": 6.838260650634766,
      "learning_rate": 3.3228081254604784e-05,
      "loss": 0.5257,
      "step": 25456
    },
    {
      "epoch": 8.038206504578465,
      "grad_norm": 5.111423492431641,
      "learning_rate": 3.322720415394871e-05,
      "loss": 0.3716,
      "step": 25457
    },
    {
      "epoch": 8.03852226081465,
      "grad_norm": 4.839354515075684,
      "learning_rate": 3.322632705329264e-05,
      "loss": 0.434,
      "step": 25458
    },
    {
      "epoch": 8.038838017050837,
      "grad_norm": 7.542938232421875,
      "learning_rate": 3.3225449952636566e-05,
      "loss": 0.448,
      "step": 25459
    },
    {
      "epoch": 8.039153773287023,
      "grad_norm": 4.459887981414795,
      "learning_rate": 3.32245728519805e-05,
      "loss": 0.4062,
      "step": 25460
    },
    {
      "epoch": 8.039469529523208,
      "grad_norm": 4.219688415527344,
      "learning_rate": 3.322369575132442e-05,
      "loss": 0.3506,
      "step": 25461
    },
    {
      "epoch": 8.039785285759393,
      "grad_norm": 6.009738445281982,
      "learning_rate": 3.3222818650668354e-05,
      "loss": 0.3631,
      "step": 25462
    },
    {
      "epoch": 8.04010104199558,
      "grad_norm": 4.800045013427734,
      "learning_rate": 3.322194155001228e-05,
      "loss": 0.3781,
      "step": 25463
    },
    {
      "epoch": 8.040416798231766,
      "grad_norm": 5.158969879150391,
      "learning_rate": 3.322106444935621e-05,
      "loss": 0.3278,
      "step": 25464
    },
    {
      "epoch": 8.04073255446795,
      "grad_norm": 6.200314521789551,
      "learning_rate": 3.3220187348700136e-05,
      "loss": 0.373,
      "step": 25465
    },
    {
      "epoch": 8.041048310704136,
      "grad_norm": 5.652410507202148,
      "learning_rate": 3.321931024804407e-05,
      "loss": 0.4237,
      "step": 25466
    },
    {
      "epoch": 8.041364066940321,
      "grad_norm": 5.1048784255981445,
      "learning_rate": 3.321843314738799e-05,
      "loss": 0.3534,
      "step": 25467
    },
    {
      "epoch": 8.041679823176509,
      "grad_norm": 5.286774635314941,
      "learning_rate": 3.3217556046731924e-05,
      "loss": 0.3233,
      "step": 25468
    },
    {
      "epoch": 8.041995579412694,
      "grad_norm": 4.996139049530029,
      "learning_rate": 3.321667894607585e-05,
      "loss": 0.4721,
      "step": 25469
    },
    {
      "epoch": 8.04231133564888,
      "grad_norm": 6.562928199768066,
      "learning_rate": 3.321580184541978e-05,
      "loss": 0.3603,
      "step": 25470
    },
    {
      "epoch": 8.042627091885064,
      "grad_norm": 7.425041198730469,
      "learning_rate": 3.321492474476371e-05,
      "loss": 0.4397,
      "step": 25471
    },
    {
      "epoch": 8.04294284812125,
      "grad_norm": 5.564579963684082,
      "learning_rate": 3.321404764410764e-05,
      "loss": 0.3632,
      "step": 25472
    },
    {
      "epoch": 8.043258604357437,
      "grad_norm": 7.07814884185791,
      "learning_rate": 3.321317054345157e-05,
      "loss": 0.6182,
      "step": 25473
    },
    {
      "epoch": 8.043574360593622,
      "grad_norm": 7.9804911613464355,
      "learning_rate": 3.3212293442795494e-05,
      "loss": 0.3445,
      "step": 25474
    },
    {
      "epoch": 8.043890116829807,
      "grad_norm": 5.570679187774658,
      "learning_rate": 3.3211416342139425e-05,
      "loss": 0.337,
      "step": 25475
    },
    {
      "epoch": 8.044205873065993,
      "grad_norm": 5.837382793426514,
      "learning_rate": 3.321053924148335e-05,
      "loss": 0.4083,
      "step": 25476
    },
    {
      "epoch": 8.044521629302178,
      "grad_norm": 6.860426902770996,
      "learning_rate": 3.320966214082729e-05,
      "loss": 0.4309,
      "step": 25477
    },
    {
      "epoch": 8.044837385538365,
      "grad_norm": 5.8649468421936035,
      "learning_rate": 3.3208785040171214e-05,
      "loss": 0.4477,
      "step": 25478
    },
    {
      "epoch": 8.04515314177455,
      "grad_norm": 4.921430587768555,
      "learning_rate": 3.3207907939515145e-05,
      "loss": 0.351,
      "step": 25479
    },
    {
      "epoch": 8.045468898010736,
      "grad_norm": 6.216490745544434,
      "learning_rate": 3.320703083885907e-05,
      "loss": 0.4524,
      "step": 25480
    },
    {
      "epoch": 8.045784654246921,
      "grad_norm": 5.357871055603027,
      "learning_rate": 3.3206153738203e-05,
      "loss": 0.3197,
      "step": 25481
    },
    {
      "epoch": 8.046100410483106,
      "grad_norm": 5.363649845123291,
      "learning_rate": 3.320527663754693e-05,
      "loss": 0.3629,
      "step": 25482
    },
    {
      "epoch": 8.046416166719293,
      "grad_norm": 5.0341386795043945,
      "learning_rate": 3.320439953689086e-05,
      "loss": 0.3883,
      "step": 25483
    },
    {
      "epoch": 8.046731922955479,
      "grad_norm": 6.213036060333252,
      "learning_rate": 3.3203522436234783e-05,
      "loss": 0.5549,
      "step": 25484
    },
    {
      "epoch": 8.047047679191664,
      "grad_norm": 4.45267915725708,
      "learning_rate": 3.3202645335578715e-05,
      "loss": 0.351,
      "step": 25485
    },
    {
      "epoch": 8.04736343542785,
      "grad_norm": 3.566558361053467,
      "learning_rate": 3.320176823492264e-05,
      "loss": 0.2731,
      "step": 25486
    },
    {
      "epoch": 8.047679191664036,
      "grad_norm": 7.8765788078308105,
      "learning_rate": 3.320089113426657e-05,
      "loss": 0.4435,
      "step": 25487
    },
    {
      "epoch": 8.047994947900222,
      "grad_norm": 4.741966724395752,
      "learning_rate": 3.32000140336105e-05,
      "loss": 0.405,
      "step": 25488
    },
    {
      "epoch": 8.048310704136407,
      "grad_norm": 5.665288925170898,
      "learning_rate": 3.319913693295443e-05,
      "loss": 0.3794,
      "step": 25489
    },
    {
      "epoch": 8.048626460372592,
      "grad_norm": 6.6667070388793945,
      "learning_rate": 3.319825983229835e-05,
      "loss": 0.4461,
      "step": 25490
    },
    {
      "epoch": 8.048942216608777,
      "grad_norm": 6.132197856903076,
      "learning_rate": 3.3197382731642285e-05,
      "loss": 0.4016,
      "step": 25491
    },
    {
      "epoch": 8.049257972844964,
      "grad_norm": 8.33215618133545,
      "learning_rate": 3.3196505630986217e-05,
      "loss": 0.5063,
      "step": 25492
    },
    {
      "epoch": 8.04957372908115,
      "grad_norm": 4.298281669616699,
      "learning_rate": 3.319562853033014e-05,
      "loss": 0.3296,
      "step": 25493
    },
    {
      "epoch": 8.049889485317335,
      "grad_norm": 4.93661642074585,
      "learning_rate": 3.319475142967407e-05,
      "loss": 0.3906,
      "step": 25494
    },
    {
      "epoch": 8.05020524155352,
      "grad_norm": 6.088298797607422,
      "learning_rate": 3.3193874329018e-05,
      "loss": 0.3497,
      "step": 25495
    },
    {
      "epoch": 8.050520997789706,
      "grad_norm": 5.111875534057617,
      "learning_rate": 3.319299722836193e-05,
      "loss": 0.3196,
      "step": 25496
    },
    {
      "epoch": 8.050836754025893,
      "grad_norm": 4.6368513107299805,
      "learning_rate": 3.3192120127705855e-05,
      "loss": 0.322,
      "step": 25497
    },
    {
      "epoch": 8.051152510262078,
      "grad_norm": 4.5363030433654785,
      "learning_rate": 3.3191243027049786e-05,
      "loss": 0.3585,
      "step": 25498
    },
    {
      "epoch": 8.051468266498263,
      "grad_norm": 5.555912494659424,
      "learning_rate": 3.319036592639371e-05,
      "loss": 0.3086,
      "step": 25499
    },
    {
      "epoch": 8.051784022734449,
      "grad_norm": 6.048706531524658,
      "learning_rate": 3.318948882573764e-05,
      "loss": 0.3855,
      "step": 25500
    },
    {
      "epoch": 8.052099778970634,
      "grad_norm": 6.444277286529541,
      "learning_rate": 3.318861172508157e-05,
      "loss": 0.4118,
      "step": 25501
    },
    {
      "epoch": 8.052415535206821,
      "grad_norm": 5.75486421585083,
      "learning_rate": 3.31877346244255e-05,
      "loss": 0.3818,
      "step": 25502
    },
    {
      "epoch": 8.052731291443006,
      "grad_norm": 5.525105953216553,
      "learning_rate": 3.3186857523769424e-05,
      "loss": 0.4349,
      "step": 25503
    },
    {
      "epoch": 8.053047047679192,
      "grad_norm": 4.803842067718506,
      "learning_rate": 3.3185980423113356e-05,
      "loss": 0.2952,
      "step": 25504
    },
    {
      "epoch": 8.053362803915377,
      "grad_norm": 6.391342639923096,
      "learning_rate": 3.318510332245729e-05,
      "loss": 0.4042,
      "step": 25505
    },
    {
      "epoch": 8.053678560151562,
      "grad_norm": 5.820845127105713,
      "learning_rate": 3.318422622180122e-05,
      "loss": 0.4983,
      "step": 25506
    },
    {
      "epoch": 8.05399431638775,
      "grad_norm": 4.576479434967041,
      "learning_rate": 3.3183349121145144e-05,
      "loss": 0.3794,
      "step": 25507
    },
    {
      "epoch": 8.054310072623935,
      "grad_norm": 6.489149570465088,
      "learning_rate": 3.3182472020489076e-05,
      "loss": 0.4129,
      "step": 25508
    },
    {
      "epoch": 8.05462582886012,
      "grad_norm": 6.17751932144165,
      "learning_rate": 3.3181594919833e-05,
      "loss": 0.4843,
      "step": 25509
    },
    {
      "epoch": 8.054941585096305,
      "grad_norm": 6.751262187957764,
      "learning_rate": 3.318071781917693e-05,
      "loss": 0.4402,
      "step": 25510
    },
    {
      "epoch": 8.055257341332492,
      "grad_norm": 7.636989593505859,
      "learning_rate": 3.317984071852086e-05,
      "loss": 0.478,
      "step": 25511
    },
    {
      "epoch": 8.055573097568677,
      "grad_norm": 8.334954261779785,
      "learning_rate": 3.317896361786479e-05,
      "loss": 0.376,
      "step": 25512
    },
    {
      "epoch": 8.055888853804863,
      "grad_norm": 5.298713684082031,
      "learning_rate": 3.3178086517208714e-05,
      "loss": 0.3683,
      "step": 25513
    },
    {
      "epoch": 8.056204610041048,
      "grad_norm": 5.459591865539551,
      "learning_rate": 3.3177209416552646e-05,
      "loss": 0.3555,
      "step": 25514
    },
    {
      "epoch": 8.056520366277233,
      "grad_norm": 5.838832378387451,
      "learning_rate": 3.317633231589658e-05,
      "loss": 0.4517,
      "step": 25515
    },
    {
      "epoch": 8.05683612251342,
      "grad_norm": 5.052845478057861,
      "learning_rate": 3.31754552152405e-05,
      "loss": 0.3922,
      "step": 25516
    },
    {
      "epoch": 8.057151878749606,
      "grad_norm": 4.798679351806641,
      "learning_rate": 3.3174578114584434e-05,
      "loss": 0.3774,
      "step": 25517
    },
    {
      "epoch": 8.057467634985791,
      "grad_norm": 5.982198715209961,
      "learning_rate": 3.317370101392836e-05,
      "loss": 0.3393,
      "step": 25518
    },
    {
      "epoch": 8.057783391221976,
      "grad_norm": 6.099907875061035,
      "learning_rate": 3.317282391327229e-05,
      "loss": 0.3696,
      "step": 25519
    },
    {
      "epoch": 8.058099147458162,
      "grad_norm": 5.765919208526611,
      "learning_rate": 3.3171946812616216e-05,
      "loss": 0.3866,
      "step": 25520
    },
    {
      "epoch": 8.058414903694349,
      "grad_norm": 4.691653251647949,
      "learning_rate": 3.317106971196015e-05,
      "loss": 0.3886,
      "step": 25521
    },
    {
      "epoch": 8.058730659930534,
      "grad_norm": 4.8636393547058105,
      "learning_rate": 3.317019261130407e-05,
      "loss": 0.3358,
      "step": 25522
    },
    {
      "epoch": 8.05904641616672,
      "grad_norm": 4.757317543029785,
      "learning_rate": 3.3169315510648004e-05,
      "loss": 0.3476,
      "step": 25523
    },
    {
      "epoch": 8.059362172402905,
      "grad_norm": 6.203282356262207,
      "learning_rate": 3.316843840999193e-05,
      "loss": 0.3476,
      "step": 25524
    },
    {
      "epoch": 8.05967792863909,
      "grad_norm": 4.286862373352051,
      "learning_rate": 3.316756130933586e-05,
      "loss": 0.3969,
      "step": 25525
    },
    {
      "epoch": 8.059993684875277,
      "grad_norm": 5.681732177734375,
      "learning_rate": 3.3166684208679785e-05,
      "loss": 0.3845,
      "step": 25526
    },
    {
      "epoch": 8.060309441111462,
      "grad_norm": 19.01719856262207,
      "learning_rate": 3.316580710802372e-05,
      "loss": 0.3249,
      "step": 25527
    },
    {
      "epoch": 8.060625197347647,
      "grad_norm": 5.861850261688232,
      "learning_rate": 3.316493000736764e-05,
      "loss": 0.3705,
      "step": 25528
    },
    {
      "epoch": 8.060940953583833,
      "grad_norm": 6.665703773498535,
      "learning_rate": 3.3164052906711574e-05,
      "loss": 0.4338,
      "step": 25529
    },
    {
      "epoch": 8.061256709820018,
      "grad_norm": 5.127597332000732,
      "learning_rate": 3.3163175806055505e-05,
      "loss": 0.344,
      "step": 25530
    },
    {
      "epoch": 8.061572466056205,
      "grad_norm": 6.720476150512695,
      "learning_rate": 3.316229870539943e-05,
      "loss": 0.3844,
      "step": 25531
    },
    {
      "epoch": 8.06188822229239,
      "grad_norm": 5.765183448791504,
      "learning_rate": 3.316142160474336e-05,
      "loss": 0.3335,
      "step": 25532
    },
    {
      "epoch": 8.062203978528576,
      "grad_norm": 7.7730207443237305,
      "learning_rate": 3.3160544504087294e-05,
      "loss": 0.3801,
      "step": 25533
    },
    {
      "epoch": 8.062519734764761,
      "grad_norm": 5.231419563293457,
      "learning_rate": 3.315966740343122e-05,
      "loss": 0.4329,
      "step": 25534
    },
    {
      "epoch": 8.062835491000948,
      "grad_norm": 5.637840747833252,
      "learning_rate": 3.315879030277515e-05,
      "loss": 0.3773,
      "step": 25535
    },
    {
      "epoch": 8.063151247237133,
      "grad_norm": 6.138518810272217,
      "learning_rate": 3.315791320211908e-05,
      "loss": 0.3429,
      "step": 25536
    },
    {
      "epoch": 8.063467003473319,
      "grad_norm": 4.972481727600098,
      "learning_rate": 3.315703610146301e-05,
      "loss": 0.3283,
      "step": 25537
    },
    {
      "epoch": 8.063782759709504,
      "grad_norm": 9.189970970153809,
      "learning_rate": 3.315615900080694e-05,
      "loss": 0.526,
      "step": 25538
    },
    {
      "epoch": 8.06409851594569,
      "grad_norm": 4.507514953613281,
      "learning_rate": 3.315528190015086e-05,
      "loss": 0.3153,
      "step": 25539
    },
    {
      "epoch": 8.064414272181876,
      "grad_norm": 4.991957664489746,
      "learning_rate": 3.3154404799494795e-05,
      "loss": 0.3325,
      "step": 25540
    },
    {
      "epoch": 8.064730028418062,
      "grad_norm": 5.824626445770264,
      "learning_rate": 3.315352769883872e-05,
      "loss": 0.4202,
      "step": 25541
    },
    {
      "epoch": 8.065045784654247,
      "grad_norm": 4.494315147399902,
      "learning_rate": 3.315265059818265e-05,
      "loss": 0.3922,
      "step": 25542
    },
    {
      "epoch": 8.065361540890432,
      "grad_norm": 5.595474720001221,
      "learning_rate": 3.3151773497526577e-05,
      "loss": 0.3464,
      "step": 25543
    },
    {
      "epoch": 8.065677297126618,
      "grad_norm": 5.925870895385742,
      "learning_rate": 3.315089639687051e-05,
      "loss": 0.3847,
      "step": 25544
    },
    {
      "epoch": 8.065993053362805,
      "grad_norm": 6.0551018714904785,
      "learning_rate": 3.315001929621443e-05,
      "loss": 0.4217,
      "step": 25545
    },
    {
      "epoch": 8.06630880959899,
      "grad_norm": 7.722222328186035,
      "learning_rate": 3.3149142195558365e-05,
      "loss": 0.3542,
      "step": 25546
    },
    {
      "epoch": 8.066624565835175,
      "grad_norm": 6.70393180847168,
      "learning_rate": 3.314826509490229e-05,
      "loss": 0.4974,
      "step": 25547
    },
    {
      "epoch": 8.06694032207136,
      "grad_norm": 6.199000835418701,
      "learning_rate": 3.314738799424622e-05,
      "loss": 0.3859,
      "step": 25548
    },
    {
      "epoch": 8.067256078307546,
      "grad_norm": 5.625103950500488,
      "learning_rate": 3.3146510893590146e-05,
      "loss": 0.3988,
      "step": 25549
    },
    {
      "epoch": 8.067571834543733,
      "grad_norm": 7.2414703369140625,
      "learning_rate": 3.314563379293408e-05,
      "loss": 0.4594,
      "step": 25550
    },
    {
      "epoch": 8.067887590779918,
      "grad_norm": 5.002753734588623,
      "learning_rate": 3.3144756692278e-05,
      "loss": 0.2944,
      "step": 25551
    },
    {
      "epoch": 8.068203347016103,
      "grad_norm": 4.226338863372803,
      "learning_rate": 3.3143879591621935e-05,
      "loss": 0.2727,
      "step": 25552
    },
    {
      "epoch": 8.068519103252289,
      "grad_norm": 5.797687530517578,
      "learning_rate": 3.3143002490965866e-05,
      "loss": 0.4151,
      "step": 25553
    },
    {
      "epoch": 8.068834859488476,
      "grad_norm": 4.048707962036133,
      "learning_rate": 3.314212539030979e-05,
      "loss": 0.3811,
      "step": 25554
    },
    {
      "epoch": 8.069150615724661,
      "grad_norm": 6.4094462394714355,
      "learning_rate": 3.314124828965372e-05,
      "loss": 0.4336,
      "step": 25555
    },
    {
      "epoch": 8.069466371960846,
      "grad_norm": 6.685544490814209,
      "learning_rate": 3.314037118899765e-05,
      "loss": 0.3845,
      "step": 25556
    },
    {
      "epoch": 8.069782128197032,
      "grad_norm": 5.451541900634766,
      "learning_rate": 3.313949408834158e-05,
      "loss": 0.3356,
      "step": 25557
    },
    {
      "epoch": 8.070097884433217,
      "grad_norm": 6.43076229095459,
      "learning_rate": 3.3138616987685504e-05,
      "loss": 0.435,
      "step": 25558
    },
    {
      "epoch": 8.070413640669404,
      "grad_norm": 6.703387260437012,
      "learning_rate": 3.313773988702944e-05,
      "loss": 0.3902,
      "step": 25559
    },
    {
      "epoch": 8.07072939690559,
      "grad_norm": 6.122992992401123,
      "learning_rate": 3.313686278637337e-05,
      "loss": 0.3996,
      "step": 25560
    },
    {
      "epoch": 8.071045153141775,
      "grad_norm": 8.486872673034668,
      "learning_rate": 3.31359856857173e-05,
      "loss": 0.4821,
      "step": 25561
    },
    {
      "epoch": 8.07136090937796,
      "grad_norm": 5.628899097442627,
      "learning_rate": 3.3135108585061224e-05,
      "loss": 0.313,
      "step": 25562
    },
    {
      "epoch": 8.071676665614145,
      "grad_norm": 8.006911277770996,
      "learning_rate": 3.3134231484405156e-05,
      "loss": 0.3822,
      "step": 25563
    },
    {
      "epoch": 8.071992421850332,
      "grad_norm": 5.907465934753418,
      "learning_rate": 3.313335438374908e-05,
      "loss": 0.3601,
      "step": 25564
    },
    {
      "epoch": 8.072308178086518,
      "grad_norm": 6.311863422393799,
      "learning_rate": 3.313247728309301e-05,
      "loss": 0.4003,
      "step": 25565
    },
    {
      "epoch": 8.072623934322703,
      "grad_norm": 5.534120082855225,
      "learning_rate": 3.313160018243694e-05,
      "loss": 0.383,
      "step": 25566
    },
    {
      "epoch": 8.072939690558888,
      "grad_norm": 5.205721855163574,
      "learning_rate": 3.313072308178087e-05,
      "loss": 0.4416,
      "step": 25567
    },
    {
      "epoch": 8.073255446795073,
      "grad_norm": 5.431152820587158,
      "learning_rate": 3.3129845981124794e-05,
      "loss": 0.3572,
      "step": 25568
    },
    {
      "epoch": 8.07357120303126,
      "grad_norm": 5.500993728637695,
      "learning_rate": 3.3128968880468726e-05,
      "loss": 0.3582,
      "step": 25569
    },
    {
      "epoch": 8.073886959267446,
      "grad_norm": 4.840610980987549,
      "learning_rate": 3.312809177981265e-05,
      "loss": 0.3937,
      "step": 25570
    },
    {
      "epoch": 8.074202715503631,
      "grad_norm": 5.994827747344971,
      "learning_rate": 3.312721467915658e-05,
      "loss": 0.418,
      "step": 25571
    },
    {
      "epoch": 8.074518471739816,
      "grad_norm": 5.478878021240234,
      "learning_rate": 3.312633757850051e-05,
      "loss": 0.382,
      "step": 25572
    },
    {
      "epoch": 8.074834227976002,
      "grad_norm": 5.536559581756592,
      "learning_rate": 3.312546047784444e-05,
      "loss": 0.3887,
      "step": 25573
    },
    {
      "epoch": 8.075149984212189,
      "grad_norm": 5.97597599029541,
      "learning_rate": 3.312458337718837e-05,
      "loss": 0.4326,
      "step": 25574
    },
    {
      "epoch": 8.075465740448374,
      "grad_norm": 5.416113376617432,
      "learning_rate": 3.3123706276532295e-05,
      "loss": 0.3795,
      "step": 25575
    },
    {
      "epoch": 8.07578149668456,
      "grad_norm": 6.034222602844238,
      "learning_rate": 3.312282917587623e-05,
      "loss": 0.474,
      "step": 25576
    },
    {
      "epoch": 8.076097252920745,
      "grad_norm": 5.848433494567871,
      "learning_rate": 3.312195207522015e-05,
      "loss": 0.4439,
      "step": 25577
    },
    {
      "epoch": 8.076413009156932,
      "grad_norm": 7.055784225463867,
      "learning_rate": 3.3121074974564084e-05,
      "loss": 0.3969,
      "step": 25578
    },
    {
      "epoch": 8.076728765393117,
      "grad_norm": 6.277595520019531,
      "learning_rate": 3.312019787390801e-05,
      "loss": 0.4041,
      "step": 25579
    },
    {
      "epoch": 8.077044521629302,
      "grad_norm": 5.306163787841797,
      "learning_rate": 3.311932077325194e-05,
      "loss": 0.3079,
      "step": 25580
    },
    {
      "epoch": 8.077360277865488,
      "grad_norm": 6.412067890167236,
      "learning_rate": 3.3118443672595865e-05,
      "loss": 0.4346,
      "step": 25581
    },
    {
      "epoch": 8.077676034101673,
      "grad_norm": 5.105767250061035,
      "learning_rate": 3.31175665719398e-05,
      "loss": 0.2743,
      "step": 25582
    },
    {
      "epoch": 8.07799179033786,
      "grad_norm": 5.547919750213623,
      "learning_rate": 3.311668947128372e-05,
      "loss": 0.4636,
      "step": 25583
    },
    {
      "epoch": 8.078307546574045,
      "grad_norm": 4.68227481842041,
      "learning_rate": 3.3115812370627654e-05,
      "loss": 0.376,
      "step": 25584
    },
    {
      "epoch": 8.07862330281023,
      "grad_norm": 6.379084587097168,
      "learning_rate": 3.311493526997158e-05,
      "loss": 0.3539,
      "step": 25585
    },
    {
      "epoch": 8.078939059046416,
      "grad_norm": 4.611754417419434,
      "learning_rate": 3.311405816931552e-05,
      "loss": 0.3665,
      "step": 25586
    },
    {
      "epoch": 8.079254815282601,
      "grad_norm": 5.052892208099365,
      "learning_rate": 3.311318106865944e-05,
      "loss": 0.4017,
      "step": 25587
    },
    {
      "epoch": 8.079570571518788,
      "grad_norm": 6.233636379241943,
      "learning_rate": 3.3112303968003373e-05,
      "loss": 0.4307,
      "step": 25588
    },
    {
      "epoch": 8.079886327754974,
      "grad_norm": 6.740291595458984,
      "learning_rate": 3.31114268673473e-05,
      "loss": 0.3606,
      "step": 25589
    },
    {
      "epoch": 8.080202083991159,
      "grad_norm": 4.837064266204834,
      "learning_rate": 3.311054976669123e-05,
      "loss": 0.3284,
      "step": 25590
    },
    {
      "epoch": 8.080517840227344,
      "grad_norm": 8.39039134979248,
      "learning_rate": 3.3109672666035155e-05,
      "loss": 0.3533,
      "step": 25591
    },
    {
      "epoch": 8.08083359646353,
      "grad_norm": 4.832098007202148,
      "learning_rate": 3.310879556537909e-05,
      "loss": 0.3529,
      "step": 25592
    },
    {
      "epoch": 8.081149352699716,
      "grad_norm": 7.204036712646484,
      "learning_rate": 3.310791846472301e-05,
      "loss": 0.3069,
      "step": 25593
    },
    {
      "epoch": 8.081465108935902,
      "grad_norm": 6.734086036682129,
      "learning_rate": 3.310704136406694e-05,
      "loss": 0.4835,
      "step": 25594
    },
    {
      "epoch": 8.081780865172087,
      "grad_norm": 4.714554309844971,
      "learning_rate": 3.310616426341087e-05,
      "loss": 0.3272,
      "step": 25595
    },
    {
      "epoch": 8.082096621408272,
      "grad_norm": 9.236308097839355,
      "learning_rate": 3.31052871627548e-05,
      "loss": 0.3997,
      "step": 25596
    },
    {
      "epoch": 8.082412377644458,
      "grad_norm": 6.332578659057617,
      "learning_rate": 3.310441006209873e-05,
      "loss": 0.397,
      "step": 25597
    },
    {
      "epoch": 8.082728133880645,
      "grad_norm": 4.37985897064209,
      "learning_rate": 3.3103532961442656e-05,
      "loss": 0.39,
      "step": 25598
    },
    {
      "epoch": 8.08304389011683,
      "grad_norm": 6.986300945281982,
      "learning_rate": 3.310265586078659e-05,
      "loss": 0.4121,
      "step": 25599
    },
    {
      "epoch": 8.083359646353015,
      "grad_norm": 6.716485023498535,
      "learning_rate": 3.310177876013051e-05,
      "loss": 0.4094,
      "step": 25600
    },
    {
      "epoch": 8.083359646353015,
      "eval_loss": 0.9855818748474121,
      "eval_runtime": 4.6153,
      "eval_samples_per_second": 539.506,
      "eval_steps_per_second": 4.55,
      "step": 25600
    },
    {
      "epoch": 8.0836754025892,
      "grad_norm": 6.214844226837158,
      "learning_rate": 3.3100901659474445e-05,
      "loss": 0.4477,
      "step": 25601
    },
    {
      "epoch": 8.083991158825388,
      "grad_norm": 5.459610462188721,
      "learning_rate": 3.310002455881837e-05,
      "loss": 0.3424,
      "step": 25602
    },
    {
      "epoch": 8.084306915061573,
      "grad_norm": 5.332514762878418,
      "learning_rate": 3.30991474581623e-05,
      "loss": 0.431,
      "step": 25603
    },
    {
      "epoch": 8.084622671297758,
      "grad_norm": 5.787818431854248,
      "learning_rate": 3.3098270357506226e-05,
      "loss": 0.376,
      "step": 25604
    },
    {
      "epoch": 8.084938427533944,
      "grad_norm": 4.672502517700195,
      "learning_rate": 3.309739325685016e-05,
      "loss": 0.3672,
      "step": 25605
    },
    {
      "epoch": 8.085254183770129,
      "grad_norm": 5.4536309242248535,
      "learning_rate": 3.309651615619408e-05,
      "loss": 0.3781,
      "step": 25606
    },
    {
      "epoch": 8.085569940006316,
      "grad_norm": 5.733331680297852,
      "learning_rate": 3.3095639055538014e-05,
      "loss": 0.3529,
      "step": 25607
    },
    {
      "epoch": 8.085885696242501,
      "grad_norm": 8.248607635498047,
      "learning_rate": 3.309476195488194e-05,
      "loss": 0.4125,
      "step": 25608
    },
    {
      "epoch": 8.086201452478686,
      "grad_norm": 5.399105548858643,
      "learning_rate": 3.309388485422587e-05,
      "loss": 0.3317,
      "step": 25609
    },
    {
      "epoch": 8.086517208714872,
      "grad_norm": 7.0760016441345215,
      "learning_rate": 3.3093007753569796e-05,
      "loss": 0.3949,
      "step": 25610
    },
    {
      "epoch": 8.086832964951057,
      "grad_norm": 5.7316484451293945,
      "learning_rate": 3.309213065291373e-05,
      "loss": 0.4208,
      "step": 25611
    },
    {
      "epoch": 8.087148721187244,
      "grad_norm": 7.016031265258789,
      "learning_rate": 3.309125355225766e-05,
      "loss": 0.4285,
      "step": 25612
    },
    {
      "epoch": 8.08746447742343,
      "grad_norm": 6.8776984214782715,
      "learning_rate": 3.3090376451601584e-05,
      "loss": 0.3645,
      "step": 25613
    },
    {
      "epoch": 8.087780233659615,
      "grad_norm": 7.062067985534668,
      "learning_rate": 3.3089499350945516e-05,
      "loss": 0.3851,
      "step": 25614
    },
    {
      "epoch": 8.0880959898958,
      "grad_norm": 7.372175693511963,
      "learning_rate": 3.308862225028945e-05,
      "loss": 0.4174,
      "step": 25615
    },
    {
      "epoch": 8.088411746131985,
      "grad_norm": 7.057225704193115,
      "learning_rate": 3.308774514963337e-05,
      "loss": 0.3687,
      "step": 25616
    },
    {
      "epoch": 8.088727502368172,
      "grad_norm": 6.0559611320495605,
      "learning_rate": 3.3086868048977304e-05,
      "loss": 0.3972,
      "step": 25617
    },
    {
      "epoch": 8.089043258604358,
      "grad_norm": 5.40281867980957,
      "learning_rate": 3.3085990948321236e-05,
      "loss": 0.3577,
      "step": 25618
    },
    {
      "epoch": 8.089359014840543,
      "grad_norm": 6.46671199798584,
      "learning_rate": 3.308511384766516e-05,
      "loss": 0.384,
      "step": 25619
    },
    {
      "epoch": 8.089674771076728,
      "grad_norm": 6.364355564117432,
      "learning_rate": 3.308423674700909e-05,
      "loss": 0.4429,
      "step": 25620
    },
    {
      "epoch": 8.089990527312914,
      "grad_norm": 4.263622283935547,
      "learning_rate": 3.308335964635302e-05,
      "loss": 0.303,
      "step": 25621
    },
    {
      "epoch": 8.0903062835491,
      "grad_norm": 5.455845832824707,
      "learning_rate": 3.308248254569695e-05,
      "loss": 0.3675,
      "step": 25622
    },
    {
      "epoch": 8.090622039785286,
      "grad_norm": 6.713884353637695,
      "learning_rate": 3.3081605445040874e-05,
      "loss": 0.3671,
      "step": 25623
    },
    {
      "epoch": 8.090937796021471,
      "grad_norm": 5.2243123054504395,
      "learning_rate": 3.3080728344384806e-05,
      "loss": 0.3609,
      "step": 25624
    },
    {
      "epoch": 8.091253552257657,
      "grad_norm": 6.368896007537842,
      "learning_rate": 3.307985124372873e-05,
      "loss": 0.3812,
      "step": 25625
    },
    {
      "epoch": 8.091569308493844,
      "grad_norm": 6.286235332489014,
      "learning_rate": 3.307897414307266e-05,
      "loss": 0.3462,
      "step": 25626
    },
    {
      "epoch": 8.091885064730029,
      "grad_norm": 6.466432094573975,
      "learning_rate": 3.307809704241659e-05,
      "loss": 0.3219,
      "step": 25627
    },
    {
      "epoch": 8.092200820966214,
      "grad_norm": 8.552482604980469,
      "learning_rate": 3.307721994176052e-05,
      "loss": 0.4631,
      "step": 25628
    },
    {
      "epoch": 8.0925165772024,
      "grad_norm": 4.97003173828125,
      "learning_rate": 3.3076342841104444e-05,
      "loss": 0.3438,
      "step": 25629
    },
    {
      "epoch": 8.092832333438585,
      "grad_norm": 7.701997756958008,
      "learning_rate": 3.3075465740448375e-05,
      "loss": 0.5101,
      "step": 25630
    },
    {
      "epoch": 8.093148089674772,
      "grad_norm": 6.646937847137451,
      "learning_rate": 3.30745886397923e-05,
      "loss": 0.4269,
      "step": 25631
    },
    {
      "epoch": 8.093463845910957,
      "grad_norm": 5.572614669799805,
      "learning_rate": 3.307371153913623e-05,
      "loss": 0.3949,
      "step": 25632
    },
    {
      "epoch": 8.093779602147142,
      "grad_norm": 4.554136276245117,
      "learning_rate": 3.3072834438480164e-05,
      "loss": 0.3221,
      "step": 25633
    },
    {
      "epoch": 8.094095358383328,
      "grad_norm": 6.308282852172852,
      "learning_rate": 3.307195733782409e-05,
      "loss": 0.416,
      "step": 25634
    },
    {
      "epoch": 8.094411114619513,
      "grad_norm": 5.485428810119629,
      "learning_rate": 3.307108023716802e-05,
      "loss": 0.4269,
      "step": 25635
    },
    {
      "epoch": 8.0947268708557,
      "grad_norm": 5.133833885192871,
      "learning_rate": 3.3070203136511945e-05,
      "loss": 0.3141,
      "step": 25636
    },
    {
      "epoch": 8.095042627091885,
      "grad_norm": 6.50888729095459,
      "learning_rate": 3.306932603585588e-05,
      "loss": 0.4189,
      "step": 25637
    },
    {
      "epoch": 8.09535838332807,
      "grad_norm": 7.1065287590026855,
      "learning_rate": 3.30684489351998e-05,
      "loss": 0.3569,
      "step": 25638
    },
    {
      "epoch": 8.095674139564256,
      "grad_norm": 6.628368377685547,
      "learning_rate": 3.3067571834543733e-05,
      "loss": 0.4284,
      "step": 25639
    },
    {
      "epoch": 8.095989895800441,
      "grad_norm": 5.704347133636475,
      "learning_rate": 3.306669473388766e-05,
      "loss": 0.4324,
      "step": 25640
    },
    {
      "epoch": 8.096305652036628,
      "grad_norm": 6.380210876464844,
      "learning_rate": 3.30658176332316e-05,
      "loss": 0.3851,
      "step": 25641
    },
    {
      "epoch": 8.096621408272814,
      "grad_norm": 6.755177974700928,
      "learning_rate": 3.306494053257552e-05,
      "loss": 0.4151,
      "step": 25642
    },
    {
      "epoch": 8.096937164508999,
      "grad_norm": 5.228434085845947,
      "learning_rate": 3.306406343191945e-05,
      "loss": 0.3507,
      "step": 25643
    },
    {
      "epoch": 8.097252920745184,
      "grad_norm": 7.010218143463135,
      "learning_rate": 3.306318633126338e-05,
      "loss": 0.4147,
      "step": 25644
    },
    {
      "epoch": 8.09756867698137,
      "grad_norm": 6.497062683105469,
      "learning_rate": 3.306230923060731e-05,
      "loss": 0.4631,
      "step": 25645
    },
    {
      "epoch": 8.097884433217557,
      "grad_norm": 6.290494918823242,
      "learning_rate": 3.3061432129951235e-05,
      "loss": 0.4428,
      "step": 25646
    },
    {
      "epoch": 8.098200189453742,
      "grad_norm": 10.232769966125488,
      "learning_rate": 3.3060555029295167e-05,
      "loss": 0.4081,
      "step": 25647
    },
    {
      "epoch": 8.098515945689927,
      "grad_norm": 5.113853931427002,
      "learning_rate": 3.305967792863909e-05,
      "loss": 0.3772,
      "step": 25648
    },
    {
      "epoch": 8.098831701926112,
      "grad_norm": 5.776847839355469,
      "learning_rate": 3.305880082798302e-05,
      "loss": 0.34,
      "step": 25649
    },
    {
      "epoch": 8.0991474581623,
      "grad_norm": 6.3422417640686035,
      "learning_rate": 3.305792372732695e-05,
      "loss": 0.3998,
      "step": 25650
    },
    {
      "epoch": 8.099463214398485,
      "grad_norm": 7.159027099609375,
      "learning_rate": 3.305704662667088e-05,
      "loss": 0.3515,
      "step": 25651
    },
    {
      "epoch": 8.09977897063467,
      "grad_norm": 5.505999565124512,
      "learning_rate": 3.3056169526014805e-05,
      "loss": 0.4574,
      "step": 25652
    },
    {
      "epoch": 8.100094726870855,
      "grad_norm": 6.646686553955078,
      "learning_rate": 3.3055292425358736e-05,
      "loss": 0.451,
      "step": 25653
    },
    {
      "epoch": 8.10041048310704,
      "grad_norm": 4.985410213470459,
      "learning_rate": 3.305441532470266e-05,
      "loss": 0.3389,
      "step": 25654
    },
    {
      "epoch": 8.100726239343228,
      "grad_norm": 6.943836688995361,
      "learning_rate": 3.305353822404659e-05,
      "loss": 0.3873,
      "step": 25655
    },
    {
      "epoch": 8.101041995579413,
      "grad_norm": 4.973147392272949,
      "learning_rate": 3.3052661123390525e-05,
      "loss": 0.3152,
      "step": 25656
    },
    {
      "epoch": 8.101357751815598,
      "grad_norm": 6.381163597106934,
      "learning_rate": 3.305178402273445e-05,
      "loss": 0.4158,
      "step": 25657
    },
    {
      "epoch": 8.101673508051784,
      "grad_norm": 4.591521739959717,
      "learning_rate": 3.305090692207838e-05,
      "loss": 0.3586,
      "step": 25658
    },
    {
      "epoch": 8.101989264287969,
      "grad_norm": 6.439090251922607,
      "learning_rate": 3.3050029821422306e-05,
      "loss": 0.4658,
      "step": 25659
    },
    {
      "epoch": 8.102305020524156,
      "grad_norm": 6.867161273956299,
      "learning_rate": 3.304915272076624e-05,
      "loss": 0.4086,
      "step": 25660
    },
    {
      "epoch": 8.102620776760341,
      "grad_norm": 5.675534725189209,
      "learning_rate": 3.304827562011016e-05,
      "loss": 0.3863,
      "step": 25661
    },
    {
      "epoch": 8.102936532996527,
      "grad_norm": 5.411419868469238,
      "learning_rate": 3.3047398519454094e-05,
      "loss": 0.3588,
      "step": 25662
    },
    {
      "epoch": 8.103252289232712,
      "grad_norm": 6.000138282775879,
      "learning_rate": 3.304652141879802e-05,
      "loss": 0.4458,
      "step": 25663
    },
    {
      "epoch": 8.103568045468897,
      "grad_norm": 5.784897804260254,
      "learning_rate": 3.304564431814195e-05,
      "loss": 0.3863,
      "step": 25664
    },
    {
      "epoch": 8.103883801705084,
      "grad_norm": 6.76482629776001,
      "learning_rate": 3.3044767217485876e-05,
      "loss": 0.4196,
      "step": 25665
    },
    {
      "epoch": 8.10419955794127,
      "grad_norm": 6.491652965545654,
      "learning_rate": 3.304389011682981e-05,
      "loss": 0.4531,
      "step": 25666
    },
    {
      "epoch": 8.104515314177455,
      "grad_norm": 6.2500834465026855,
      "learning_rate": 3.304301301617373e-05,
      "loss": 0.3086,
      "step": 25667
    },
    {
      "epoch": 8.10483107041364,
      "grad_norm": 5.701425075531006,
      "learning_rate": 3.304213591551767e-05,
      "loss": 0.3484,
      "step": 25668
    },
    {
      "epoch": 8.105146826649825,
      "grad_norm": 5.801767826080322,
      "learning_rate": 3.3041258814861596e-05,
      "loss": 0.3571,
      "step": 25669
    },
    {
      "epoch": 8.105462582886013,
      "grad_norm": 8.527530670166016,
      "learning_rate": 3.304038171420553e-05,
      "loss": 0.4219,
      "step": 25670
    },
    {
      "epoch": 8.105778339122198,
      "grad_norm": 8.009488105773926,
      "learning_rate": 3.303950461354945e-05,
      "loss": 0.351,
      "step": 25671
    },
    {
      "epoch": 8.106094095358383,
      "grad_norm": 5.937117099761963,
      "learning_rate": 3.3038627512893384e-05,
      "loss": 0.3601,
      "step": 25672
    },
    {
      "epoch": 8.106409851594568,
      "grad_norm": 5.836513519287109,
      "learning_rate": 3.303775041223731e-05,
      "loss": 0.4304,
      "step": 25673
    },
    {
      "epoch": 8.106725607830755,
      "grad_norm": 6.768901824951172,
      "learning_rate": 3.303687331158124e-05,
      "loss": 0.4247,
      "step": 25674
    },
    {
      "epoch": 8.10704136406694,
      "grad_norm": 6.006324291229248,
      "learning_rate": 3.3035996210925166e-05,
      "loss": 0.4365,
      "step": 25675
    },
    {
      "epoch": 8.107357120303126,
      "grad_norm": 8.101696014404297,
      "learning_rate": 3.30351191102691e-05,
      "loss": 0.413,
      "step": 25676
    },
    {
      "epoch": 8.107672876539311,
      "grad_norm": 7.174811363220215,
      "learning_rate": 3.303424200961303e-05,
      "loss": 0.3944,
      "step": 25677
    },
    {
      "epoch": 8.107988632775497,
      "grad_norm": 3.85343337059021,
      "learning_rate": 3.3033364908956954e-05,
      "loss": 0.3005,
      "step": 25678
    },
    {
      "epoch": 8.108304389011684,
      "grad_norm": 6.8314595222473145,
      "learning_rate": 3.3032487808300886e-05,
      "loss": 0.4704,
      "step": 25679
    },
    {
      "epoch": 8.108620145247869,
      "grad_norm": 5.870666027069092,
      "learning_rate": 3.303161070764481e-05,
      "loss": 0.4045,
      "step": 25680
    },
    {
      "epoch": 8.108935901484054,
      "grad_norm": 7.555366039276123,
      "learning_rate": 3.303073360698874e-05,
      "loss": 0.4275,
      "step": 25681
    },
    {
      "epoch": 8.10925165772024,
      "grad_norm": 6.561972618103027,
      "learning_rate": 3.302985650633267e-05,
      "loss": 0.4638,
      "step": 25682
    },
    {
      "epoch": 8.109567413956425,
      "grad_norm": 4.891112327575684,
      "learning_rate": 3.30289794056766e-05,
      "loss": 0.3368,
      "step": 25683
    },
    {
      "epoch": 8.109883170192612,
      "grad_norm": 5.6732611656188965,
      "learning_rate": 3.3028102305020524e-05,
      "loss": 0.3563,
      "step": 25684
    },
    {
      "epoch": 8.110198926428797,
      "grad_norm": 4.875546455383301,
      "learning_rate": 3.3027225204364455e-05,
      "loss": 0.3707,
      "step": 25685
    },
    {
      "epoch": 8.110514682664983,
      "grad_norm": 8.435469627380371,
      "learning_rate": 3.302634810370838e-05,
      "loss": 0.384,
      "step": 25686
    },
    {
      "epoch": 8.110830438901168,
      "grad_norm": 5.929880619049072,
      "learning_rate": 3.302547100305231e-05,
      "loss": 0.3658,
      "step": 25687
    },
    {
      "epoch": 8.111146195137353,
      "grad_norm": 6.640378475189209,
      "learning_rate": 3.302459390239624e-05,
      "loss": 0.4099,
      "step": 25688
    },
    {
      "epoch": 8.11146195137354,
      "grad_norm": 5.995486736297607,
      "learning_rate": 3.302371680174017e-05,
      "loss": 0.3718,
      "step": 25689
    },
    {
      "epoch": 8.111777707609725,
      "grad_norm": 6.373023986816406,
      "learning_rate": 3.302283970108409e-05,
      "loss": 0.4561,
      "step": 25690
    },
    {
      "epoch": 8.11209346384591,
      "grad_norm": 4.393394947052002,
      "learning_rate": 3.3021962600428025e-05,
      "loss": 0.3232,
      "step": 25691
    },
    {
      "epoch": 8.112409220082096,
      "grad_norm": 4.793882846832275,
      "learning_rate": 3.302108549977195e-05,
      "loss": 0.3565,
      "step": 25692
    },
    {
      "epoch": 8.112724976318283,
      "grad_norm": 6.157295227050781,
      "learning_rate": 3.302020839911588e-05,
      "loss": 0.3216,
      "step": 25693
    },
    {
      "epoch": 8.113040732554468,
      "grad_norm": 6.315619945526123,
      "learning_rate": 3.301933129845981e-05,
      "loss": 0.3928,
      "step": 25694
    },
    {
      "epoch": 8.113356488790654,
      "grad_norm": 7.850151062011719,
      "learning_rate": 3.3018454197803745e-05,
      "loss": 0.4945,
      "step": 25695
    },
    {
      "epoch": 8.113672245026839,
      "grad_norm": 6.670854568481445,
      "learning_rate": 3.301757709714767e-05,
      "loss": 0.4112,
      "step": 25696
    },
    {
      "epoch": 8.113988001263024,
      "grad_norm": 6.470516681671143,
      "learning_rate": 3.30166999964916e-05,
      "loss": 0.4064,
      "step": 25697
    },
    {
      "epoch": 8.114303757499211,
      "grad_norm": 5.227546691894531,
      "learning_rate": 3.3015822895835526e-05,
      "loss": 0.4363,
      "step": 25698
    },
    {
      "epoch": 8.114619513735397,
      "grad_norm": 5.352411270141602,
      "learning_rate": 3.301494579517946e-05,
      "loss": 0.401,
      "step": 25699
    },
    {
      "epoch": 8.114935269971582,
      "grad_norm": 6.358745098114014,
      "learning_rate": 3.301406869452339e-05,
      "loss": 0.4031,
      "step": 25700
    },
    {
      "epoch": 8.115251026207767,
      "grad_norm": 6.590022563934326,
      "learning_rate": 3.3013191593867315e-05,
      "loss": 0.3898,
      "step": 25701
    },
    {
      "epoch": 8.115566782443953,
      "grad_norm": 5.8647685050964355,
      "learning_rate": 3.3012314493211246e-05,
      "loss": 0.4335,
      "step": 25702
    },
    {
      "epoch": 8.11588253868014,
      "grad_norm": 7.98635196685791,
      "learning_rate": 3.301143739255517e-05,
      "loss": 0.4657,
      "step": 25703
    },
    {
      "epoch": 8.116198294916325,
      "grad_norm": 6.452792644500732,
      "learning_rate": 3.30105602918991e-05,
      "loss": 0.4053,
      "step": 25704
    },
    {
      "epoch": 8.11651405115251,
      "grad_norm": 5.488797664642334,
      "learning_rate": 3.300968319124303e-05,
      "loss": 0.41,
      "step": 25705
    },
    {
      "epoch": 8.116829807388696,
      "grad_norm": 6.788732528686523,
      "learning_rate": 3.300880609058696e-05,
      "loss": 0.4259,
      "step": 25706
    },
    {
      "epoch": 8.11714556362488,
      "grad_norm": 6.370469093322754,
      "learning_rate": 3.3007928989930885e-05,
      "loss": 0.3348,
      "step": 25707
    },
    {
      "epoch": 8.117461319861068,
      "grad_norm": 3.8511595726013184,
      "learning_rate": 3.3007051889274816e-05,
      "loss": 0.3229,
      "step": 25708
    },
    {
      "epoch": 8.117777076097253,
      "grad_norm": 5.6676225662231445,
      "learning_rate": 3.300617478861874e-05,
      "loss": 0.4085,
      "step": 25709
    },
    {
      "epoch": 8.118092832333438,
      "grad_norm": 5.915192604064941,
      "learning_rate": 3.300529768796267e-05,
      "loss": 0.3816,
      "step": 25710
    },
    {
      "epoch": 8.118408588569624,
      "grad_norm": 6.15069055557251,
      "learning_rate": 3.30044205873066e-05,
      "loss": 0.4327,
      "step": 25711
    },
    {
      "epoch": 8.118724344805809,
      "grad_norm": 6.330173492431641,
      "learning_rate": 3.300354348665053e-05,
      "loss": 0.3554,
      "step": 25712
    },
    {
      "epoch": 8.119040101041996,
      "grad_norm": 6.5759501457214355,
      "learning_rate": 3.3002666385994454e-05,
      "loss": 0.3849,
      "step": 25713
    },
    {
      "epoch": 8.119355857278181,
      "grad_norm": 5.22178316116333,
      "learning_rate": 3.3001789285338386e-05,
      "loss": 0.4241,
      "step": 25714
    },
    {
      "epoch": 8.119671613514367,
      "grad_norm": 6.206790924072266,
      "learning_rate": 3.300091218468232e-05,
      "loss": 0.3636,
      "step": 25715
    },
    {
      "epoch": 8.119987369750552,
      "grad_norm": 5.7786712646484375,
      "learning_rate": 3.300003508402624e-05,
      "loss": 0.3216,
      "step": 25716
    },
    {
      "epoch": 8.120303125986739,
      "grad_norm": 5.989114761352539,
      "learning_rate": 3.2999157983370174e-05,
      "loss": 0.3831,
      "step": 25717
    },
    {
      "epoch": 8.120618882222924,
      "grad_norm": 4.838812351226807,
      "learning_rate": 3.29982808827141e-05,
      "loss": 0.4026,
      "step": 25718
    },
    {
      "epoch": 8.12093463845911,
      "grad_norm": 4.409646987915039,
      "learning_rate": 3.299740378205803e-05,
      "loss": 0.4573,
      "step": 25719
    },
    {
      "epoch": 8.121250394695295,
      "grad_norm": 5.503533840179443,
      "learning_rate": 3.2996526681401956e-05,
      "loss": 0.5104,
      "step": 25720
    },
    {
      "epoch": 8.12156615093148,
      "grad_norm": 7.305179595947266,
      "learning_rate": 3.299564958074589e-05,
      "loss": 0.3189,
      "step": 25721
    },
    {
      "epoch": 8.121881907167667,
      "grad_norm": 6.500208377838135,
      "learning_rate": 3.299477248008981e-05,
      "loss": 0.3619,
      "step": 25722
    },
    {
      "epoch": 8.122197663403853,
      "grad_norm": 6.126001834869385,
      "learning_rate": 3.299389537943375e-05,
      "loss": 0.3827,
      "step": 25723
    },
    {
      "epoch": 8.122513419640038,
      "grad_norm": 5.822574615478516,
      "learning_rate": 3.2993018278777676e-05,
      "loss": 0.4082,
      "step": 25724
    },
    {
      "epoch": 8.122829175876223,
      "grad_norm": 4.6705474853515625,
      "learning_rate": 3.299214117812161e-05,
      "loss": 0.3005,
      "step": 25725
    },
    {
      "epoch": 8.123144932112409,
      "grad_norm": 7.8872199058532715,
      "learning_rate": 3.299126407746553e-05,
      "loss": 0.4136,
      "step": 25726
    },
    {
      "epoch": 8.123460688348596,
      "grad_norm": 6.862853050231934,
      "learning_rate": 3.2990386976809464e-05,
      "loss": 0.4209,
      "step": 25727
    },
    {
      "epoch": 8.12377644458478,
      "grad_norm": 6.776836395263672,
      "learning_rate": 3.298950987615339e-05,
      "loss": 0.3796,
      "step": 25728
    },
    {
      "epoch": 8.124092200820966,
      "grad_norm": 6.692903518676758,
      "learning_rate": 3.298863277549732e-05,
      "loss": 0.4751,
      "step": 25729
    },
    {
      "epoch": 8.124407957057151,
      "grad_norm": 7.012763023376465,
      "learning_rate": 3.2987755674841245e-05,
      "loss": 0.4226,
      "step": 25730
    },
    {
      "epoch": 8.124723713293337,
      "grad_norm": 4.574400901794434,
      "learning_rate": 3.298687857418518e-05,
      "loss": 0.2868,
      "step": 25731
    },
    {
      "epoch": 8.125039469529524,
      "grad_norm": 5.0979766845703125,
      "learning_rate": 3.29860014735291e-05,
      "loss": 0.4386,
      "step": 25732
    },
    {
      "epoch": 8.12535522576571,
      "grad_norm": 7.1220622062683105,
      "learning_rate": 3.2985124372873034e-05,
      "loss": 0.4077,
      "step": 25733
    },
    {
      "epoch": 8.125670982001894,
      "grad_norm": 5.422706127166748,
      "learning_rate": 3.298424727221696e-05,
      "loss": 0.3522,
      "step": 25734
    },
    {
      "epoch": 8.12598673823808,
      "grad_norm": 5.117888927459717,
      "learning_rate": 3.298337017156089e-05,
      "loss": 0.4038,
      "step": 25735
    },
    {
      "epoch": 8.126302494474265,
      "grad_norm": 4.765973091125488,
      "learning_rate": 3.2982493070904815e-05,
      "loss": 0.4882,
      "step": 25736
    },
    {
      "epoch": 8.126618250710452,
      "grad_norm": 7.349185466766357,
      "learning_rate": 3.298161597024875e-05,
      "loss": 0.3998,
      "step": 25737
    },
    {
      "epoch": 8.126934006946637,
      "grad_norm": 6.991906642913818,
      "learning_rate": 3.298073886959268e-05,
      "loss": 0.4199,
      "step": 25738
    },
    {
      "epoch": 8.127249763182823,
      "grad_norm": 4.97905158996582,
      "learning_rate": 3.2979861768936603e-05,
      "loss": 0.347,
      "step": 25739
    },
    {
      "epoch": 8.127565519419008,
      "grad_norm": 4.167076587677002,
      "learning_rate": 3.2978984668280535e-05,
      "loss": 0.3299,
      "step": 25740
    },
    {
      "epoch": 8.127881275655195,
      "grad_norm": 4.22838020324707,
      "learning_rate": 3.297810756762446e-05,
      "loss": 0.3492,
      "step": 25741
    },
    {
      "epoch": 8.12819703189138,
      "grad_norm": 4.8728766441345215,
      "learning_rate": 3.297723046696839e-05,
      "loss": 0.4416,
      "step": 25742
    },
    {
      "epoch": 8.128512788127566,
      "grad_norm": 6.2070722579956055,
      "learning_rate": 3.297635336631232e-05,
      "loss": 0.3794,
      "step": 25743
    },
    {
      "epoch": 8.128828544363751,
      "grad_norm": 5.038743019104004,
      "learning_rate": 3.297547626565625e-05,
      "loss": 0.3371,
      "step": 25744
    },
    {
      "epoch": 8.129144300599936,
      "grad_norm": 5.886146068572998,
      "learning_rate": 3.297459916500017e-05,
      "loss": 0.3595,
      "step": 25745
    },
    {
      "epoch": 8.129460056836123,
      "grad_norm": 5.867348670959473,
      "learning_rate": 3.2973722064344105e-05,
      "loss": 0.2909,
      "step": 25746
    },
    {
      "epoch": 8.129775813072309,
      "grad_norm": 5.557319641113281,
      "learning_rate": 3.297284496368803e-05,
      "loss": 0.4684,
      "step": 25747
    },
    {
      "epoch": 8.130091569308494,
      "grad_norm": 5.996189117431641,
      "learning_rate": 3.297196786303196e-05,
      "loss": 0.3218,
      "step": 25748
    },
    {
      "epoch": 8.13040732554468,
      "grad_norm": 4.890168190002441,
      "learning_rate": 3.2971090762375886e-05,
      "loss": 0.3356,
      "step": 25749
    },
    {
      "epoch": 8.130723081780864,
      "grad_norm": 5.588528156280518,
      "learning_rate": 3.2970213661719825e-05,
      "loss": 0.3899,
      "step": 25750
    },
    {
      "epoch": 8.131038838017052,
      "grad_norm": 8.77872085571289,
      "learning_rate": 3.296933656106375e-05,
      "loss": 0.5158,
      "step": 25751
    },
    {
      "epoch": 8.131354594253237,
      "grad_norm": 4.532521724700928,
      "learning_rate": 3.296845946040768e-05,
      "loss": 0.3144,
      "step": 25752
    },
    {
      "epoch": 8.131670350489422,
      "grad_norm": 7.151022911071777,
      "learning_rate": 3.2967582359751606e-05,
      "loss": 0.3673,
      "step": 25753
    },
    {
      "epoch": 8.131986106725607,
      "grad_norm": 6.90761661529541,
      "learning_rate": 3.296670525909554e-05,
      "loss": 0.422,
      "step": 25754
    },
    {
      "epoch": 8.132301862961793,
      "grad_norm": 4.867072105407715,
      "learning_rate": 3.296582815843946e-05,
      "loss": 0.3934,
      "step": 25755
    },
    {
      "epoch": 8.13261761919798,
      "grad_norm": 3.912346124649048,
      "learning_rate": 3.2964951057783395e-05,
      "loss": 0.2846,
      "step": 25756
    },
    {
      "epoch": 8.132933375434165,
      "grad_norm": 6.365638732910156,
      "learning_rate": 3.296407395712732e-05,
      "loss": 0.3804,
      "step": 25757
    },
    {
      "epoch": 8.13324913167035,
      "grad_norm": 5.837403297424316,
      "learning_rate": 3.296319685647125e-05,
      "loss": 0.3583,
      "step": 25758
    },
    {
      "epoch": 8.133564887906536,
      "grad_norm": 5.060723781585693,
      "learning_rate": 3.296231975581518e-05,
      "loss": 0.3549,
      "step": 25759
    },
    {
      "epoch": 8.133880644142721,
      "grad_norm": 4.880369186401367,
      "learning_rate": 3.296144265515911e-05,
      "loss": 0.3468,
      "step": 25760
    },
    {
      "epoch": 8.134196400378908,
      "grad_norm": 4.808518886566162,
      "learning_rate": 3.296056555450304e-05,
      "loss": 0.3333,
      "step": 25761
    },
    {
      "epoch": 8.134512156615093,
      "grad_norm": 4.654268741607666,
      "learning_rate": 3.2959688453846964e-05,
      "loss": 0.4201,
      "step": 25762
    },
    {
      "epoch": 8.134827912851279,
      "grad_norm": 7.034278869628906,
      "learning_rate": 3.2958811353190896e-05,
      "loss": 0.5134,
      "step": 25763
    },
    {
      "epoch": 8.135143669087464,
      "grad_norm": 6.083137035369873,
      "learning_rate": 3.295793425253482e-05,
      "loss": 0.5316,
      "step": 25764
    },
    {
      "epoch": 8.135459425323651,
      "grad_norm": 5.255063533782959,
      "learning_rate": 3.295705715187875e-05,
      "loss": 0.4168,
      "step": 25765
    },
    {
      "epoch": 8.135775181559836,
      "grad_norm": 6.790687084197998,
      "learning_rate": 3.295618005122268e-05,
      "loss": 0.3561,
      "step": 25766
    },
    {
      "epoch": 8.136090937796022,
      "grad_norm": 7.721762180328369,
      "learning_rate": 3.295530295056661e-05,
      "loss": 0.4864,
      "step": 25767
    },
    {
      "epoch": 8.136406694032207,
      "grad_norm": 6.923054218292236,
      "learning_rate": 3.2954425849910534e-05,
      "loss": 0.4518,
      "step": 25768
    },
    {
      "epoch": 8.136722450268392,
      "grad_norm": 4.782064437866211,
      "learning_rate": 3.2953548749254466e-05,
      "loss": 0.3142,
      "step": 25769
    },
    {
      "epoch": 8.13703820650458,
      "grad_norm": 5.304203987121582,
      "learning_rate": 3.295267164859839e-05,
      "loss": 0.4229,
      "step": 25770
    },
    {
      "epoch": 8.137353962740765,
      "grad_norm": 5.242931842803955,
      "learning_rate": 3.295179454794232e-05,
      "loss": 0.4544,
      "step": 25771
    },
    {
      "epoch": 8.13766971897695,
      "grad_norm": 6.178976058959961,
      "learning_rate": 3.295091744728625e-05,
      "loss": 0.3947,
      "step": 25772
    },
    {
      "epoch": 8.137985475213135,
      "grad_norm": 5.156245708465576,
      "learning_rate": 3.295004034663018e-05,
      "loss": 0.463,
      "step": 25773
    },
    {
      "epoch": 8.13830123144932,
      "grad_norm": 5.4412384033203125,
      "learning_rate": 3.2949163245974104e-05,
      "loss": 0.4896,
      "step": 25774
    },
    {
      "epoch": 8.138616987685507,
      "grad_norm": 6.227180480957031,
      "learning_rate": 3.2948286145318036e-05,
      "loss": 0.3829,
      "step": 25775
    },
    {
      "epoch": 8.138932743921693,
      "grad_norm": 6.226616859436035,
      "learning_rate": 3.294740904466197e-05,
      "loss": 0.2994,
      "step": 25776
    },
    {
      "epoch": 8.139248500157878,
      "grad_norm": 5.734079837799072,
      "learning_rate": 3.29465319440059e-05,
      "loss": 0.3444,
      "step": 25777
    },
    {
      "epoch": 8.139564256394063,
      "grad_norm": 5.688495635986328,
      "learning_rate": 3.2945654843349824e-05,
      "loss": 0.4438,
      "step": 25778
    },
    {
      "epoch": 8.139880012630249,
      "grad_norm": 5.122275352478027,
      "learning_rate": 3.2944777742693756e-05,
      "loss": 0.3666,
      "step": 25779
    },
    {
      "epoch": 8.140195768866436,
      "grad_norm": 5.132413387298584,
      "learning_rate": 3.294390064203769e-05,
      "loss": 0.3726,
      "step": 25780
    },
    {
      "epoch": 8.140511525102621,
      "grad_norm": 7.756196022033691,
      "learning_rate": 3.294302354138161e-05,
      "loss": 0.4166,
      "step": 25781
    },
    {
      "epoch": 8.140827281338806,
      "grad_norm": 6.074362754821777,
      "learning_rate": 3.2942146440725544e-05,
      "loss": 0.3921,
      "step": 25782
    },
    {
      "epoch": 8.141143037574992,
      "grad_norm": 6.310527801513672,
      "learning_rate": 3.294126934006947e-05,
      "loss": 0.3863,
      "step": 25783
    },
    {
      "epoch": 8.141458793811179,
      "grad_norm": 6.229923725128174,
      "learning_rate": 3.29403922394134e-05,
      "loss": 0.3353,
      "step": 25784
    },
    {
      "epoch": 8.141774550047364,
      "grad_norm": 6.2732391357421875,
      "learning_rate": 3.2939515138757325e-05,
      "loss": 0.3734,
      "step": 25785
    },
    {
      "epoch": 8.14209030628355,
      "grad_norm": 5.3954596519470215,
      "learning_rate": 3.293863803810126e-05,
      "loss": 0.4192,
      "step": 25786
    },
    {
      "epoch": 8.142406062519735,
      "grad_norm": 8.753178596496582,
      "learning_rate": 3.293776093744518e-05,
      "loss": 0.4595,
      "step": 25787
    },
    {
      "epoch": 8.14272181875592,
      "grad_norm": 7.228723049163818,
      "learning_rate": 3.2936883836789114e-05,
      "loss": 0.4409,
      "step": 25788
    },
    {
      "epoch": 8.143037574992107,
      "grad_norm": 7.022678375244141,
      "learning_rate": 3.293600673613304e-05,
      "loss": 0.4562,
      "step": 25789
    },
    {
      "epoch": 8.143353331228292,
      "grad_norm": 4.981781959533691,
      "learning_rate": 3.293512963547697e-05,
      "loss": 0.4246,
      "step": 25790
    },
    {
      "epoch": 8.143669087464477,
      "grad_norm": 5.775618553161621,
      "learning_rate": 3.2934252534820895e-05,
      "loss": 0.3779,
      "step": 25791
    },
    {
      "epoch": 8.143984843700663,
      "grad_norm": 6.853745937347412,
      "learning_rate": 3.293337543416483e-05,
      "loss": 0.4707,
      "step": 25792
    },
    {
      "epoch": 8.144300599936848,
      "grad_norm": 6.9800801277160645,
      "learning_rate": 3.293249833350875e-05,
      "loss": 0.3143,
      "step": 25793
    },
    {
      "epoch": 8.144616356173035,
      "grad_norm": 6.249140739440918,
      "learning_rate": 3.293162123285268e-05,
      "loss": 0.2969,
      "step": 25794
    },
    {
      "epoch": 8.14493211240922,
      "grad_norm": 5.080460548400879,
      "learning_rate": 3.293074413219661e-05,
      "loss": 0.3127,
      "step": 25795
    },
    {
      "epoch": 8.145247868645406,
      "grad_norm": 5.728504657745361,
      "learning_rate": 3.292986703154054e-05,
      "loss": 0.4843,
      "step": 25796
    },
    {
      "epoch": 8.145563624881591,
      "grad_norm": 4.627720832824707,
      "learning_rate": 3.292898993088447e-05,
      "loss": 0.3169,
      "step": 25797
    },
    {
      "epoch": 8.145879381117776,
      "grad_norm": 5.480630874633789,
      "learning_rate": 3.2928112830228397e-05,
      "loss": 0.4407,
      "step": 25798
    },
    {
      "epoch": 8.146195137353963,
      "grad_norm": 5.064578056335449,
      "learning_rate": 3.292723572957233e-05,
      "loss": 0.3639,
      "step": 25799
    },
    {
      "epoch": 8.146510893590149,
      "grad_norm": 6.474903583526611,
      "learning_rate": 3.292635862891625e-05,
      "loss": 0.4005,
      "step": 25800
    },
    {
      "epoch": 8.146510893590149,
      "eval_loss": 1.0216947793960571,
      "eval_runtime": 4.5864,
      "eval_samples_per_second": 542.908,
      "eval_steps_per_second": 4.579,
      "step": 25800
    },
    {
      "epoch": 8.146826649826334,
      "grad_norm": 6.390689373016357,
      "learning_rate": 3.2925481528260185e-05,
      "loss": 0.4616,
      "step": 25801
    },
    {
      "epoch": 8.14714240606252,
      "grad_norm": 6.827483177185059,
      "learning_rate": 3.292460442760411e-05,
      "loss": 0.4185,
      "step": 25802
    },
    {
      "epoch": 8.147458162298705,
      "grad_norm": 7.0497941970825195,
      "learning_rate": 3.292372732694804e-05,
      "loss": 0.4117,
      "step": 25803
    },
    {
      "epoch": 8.147773918534892,
      "grad_norm": 5.480169296264648,
      "learning_rate": 3.292285022629197e-05,
      "loss": 0.3411,
      "step": 25804
    },
    {
      "epoch": 8.148089674771077,
      "grad_norm": 6.272717475891113,
      "learning_rate": 3.2921973125635905e-05,
      "loss": 0.3576,
      "step": 25805
    },
    {
      "epoch": 8.148405431007262,
      "grad_norm": 7.265142440795898,
      "learning_rate": 3.292109602497983e-05,
      "loss": 0.3918,
      "step": 25806
    },
    {
      "epoch": 8.148721187243448,
      "grad_norm": 6.1168718338012695,
      "learning_rate": 3.292021892432376e-05,
      "loss": 0.4002,
      "step": 25807
    },
    {
      "epoch": 8.149036943479635,
      "grad_norm": 5.747483253479004,
      "learning_rate": 3.2919341823667686e-05,
      "loss": 0.3874,
      "step": 25808
    },
    {
      "epoch": 8.14935269971582,
      "grad_norm": 5.364072799682617,
      "learning_rate": 3.291846472301162e-05,
      "loss": 0.3862,
      "step": 25809
    },
    {
      "epoch": 8.149668455952005,
      "grad_norm": 5.956439971923828,
      "learning_rate": 3.291758762235554e-05,
      "loss": 0.479,
      "step": 25810
    },
    {
      "epoch": 8.14998421218819,
      "grad_norm": 7.272150039672852,
      "learning_rate": 3.2916710521699475e-05,
      "loss": 0.4461,
      "step": 25811
    },
    {
      "epoch": 8.150299968424376,
      "grad_norm": 6.323802471160889,
      "learning_rate": 3.29158334210434e-05,
      "loss": 0.4985,
      "step": 25812
    },
    {
      "epoch": 8.150615724660563,
      "grad_norm": 6.363986015319824,
      "learning_rate": 3.291495632038733e-05,
      "loss": 0.2438,
      "step": 25813
    },
    {
      "epoch": 8.150931480896748,
      "grad_norm": 6.015542507171631,
      "learning_rate": 3.2914079219731256e-05,
      "loss": 0.439,
      "step": 25814
    },
    {
      "epoch": 8.151247237132933,
      "grad_norm": 6.789657115936279,
      "learning_rate": 3.291320211907519e-05,
      "loss": 0.5152,
      "step": 25815
    },
    {
      "epoch": 8.151562993369119,
      "grad_norm": 7.203004837036133,
      "learning_rate": 3.291232501841911e-05,
      "loss": 0.4774,
      "step": 25816
    },
    {
      "epoch": 8.151878749605304,
      "grad_norm": 6.175904273986816,
      "learning_rate": 3.2911447917763044e-05,
      "loss": 0.3786,
      "step": 25817
    },
    {
      "epoch": 8.152194505841491,
      "grad_norm": 7.14473819732666,
      "learning_rate": 3.2910570817106976e-05,
      "loss": 0.4682,
      "step": 25818
    },
    {
      "epoch": 8.152510262077676,
      "grad_norm": 5.709969997406006,
      "learning_rate": 3.29096937164509e-05,
      "loss": 0.473,
      "step": 25819
    },
    {
      "epoch": 8.152826018313862,
      "grad_norm": 6.22105598449707,
      "learning_rate": 3.290881661579483e-05,
      "loss": 0.3358,
      "step": 25820
    },
    {
      "epoch": 8.153141774550047,
      "grad_norm": 6.557440280914307,
      "learning_rate": 3.290793951513876e-05,
      "loss": 0.4048,
      "step": 25821
    },
    {
      "epoch": 8.153457530786232,
      "grad_norm": 7.556938648223877,
      "learning_rate": 3.290706241448269e-05,
      "loss": 0.4858,
      "step": 25822
    },
    {
      "epoch": 8.15377328702242,
      "grad_norm": 5.083742618560791,
      "learning_rate": 3.2906185313826614e-05,
      "loss": 0.3469,
      "step": 25823
    },
    {
      "epoch": 8.154089043258605,
      "grad_norm": 5.966939449310303,
      "learning_rate": 3.2905308213170546e-05,
      "loss": 0.3916,
      "step": 25824
    },
    {
      "epoch": 8.15440479949479,
      "grad_norm": 6.718000411987305,
      "learning_rate": 3.290443111251447e-05,
      "loss": 0.4988,
      "step": 25825
    },
    {
      "epoch": 8.154720555730975,
      "grad_norm": 5.102184772491455,
      "learning_rate": 3.29035540118584e-05,
      "loss": 0.3474,
      "step": 25826
    },
    {
      "epoch": 8.15503631196716,
      "grad_norm": 5.8494553565979,
      "learning_rate": 3.290267691120233e-05,
      "loss": 0.4166,
      "step": 25827
    },
    {
      "epoch": 8.155352068203348,
      "grad_norm": 6.688383102416992,
      "learning_rate": 3.290179981054626e-05,
      "loss": 0.3151,
      "step": 25828
    },
    {
      "epoch": 8.155667824439533,
      "grad_norm": 6.018033027648926,
      "learning_rate": 3.2900922709890184e-05,
      "loss": 0.4019,
      "step": 25829
    },
    {
      "epoch": 8.155983580675718,
      "grad_norm": 5.129284858703613,
      "learning_rate": 3.2900045609234116e-05,
      "loss": 0.4031,
      "step": 25830
    },
    {
      "epoch": 8.156299336911903,
      "grad_norm": 8.460700035095215,
      "learning_rate": 3.289916850857804e-05,
      "loss": 0.414,
      "step": 25831
    },
    {
      "epoch": 8.15661509314809,
      "grad_norm": 6.617900371551514,
      "learning_rate": 3.289829140792198e-05,
      "loss": 0.426,
      "step": 25832
    },
    {
      "epoch": 8.156930849384276,
      "grad_norm": 8.006307601928711,
      "learning_rate": 3.2897414307265904e-05,
      "loss": 0.4096,
      "step": 25833
    },
    {
      "epoch": 8.157246605620461,
      "grad_norm": 5.997675895690918,
      "learning_rate": 3.2896537206609835e-05,
      "loss": 0.4053,
      "step": 25834
    },
    {
      "epoch": 8.157562361856646,
      "grad_norm": 7.819699287414551,
      "learning_rate": 3.289566010595376e-05,
      "loss": 0.3677,
      "step": 25835
    },
    {
      "epoch": 8.157878118092832,
      "grad_norm": 6.272213459014893,
      "learning_rate": 3.289478300529769e-05,
      "loss": 0.4438,
      "step": 25836
    },
    {
      "epoch": 8.158193874329019,
      "grad_norm": 5.666982650756836,
      "learning_rate": 3.289390590464162e-05,
      "loss": 0.4228,
      "step": 25837
    },
    {
      "epoch": 8.158509630565204,
      "grad_norm": 6.354024410247803,
      "learning_rate": 3.289302880398555e-05,
      "loss": 0.322,
      "step": 25838
    },
    {
      "epoch": 8.15882538680139,
      "grad_norm": 5.3909912109375,
      "learning_rate": 3.2892151703329474e-05,
      "loss": 0.4683,
      "step": 25839
    },
    {
      "epoch": 8.159141143037575,
      "grad_norm": 4.835480690002441,
      "learning_rate": 3.2891274602673405e-05,
      "loss": 0.3551,
      "step": 25840
    },
    {
      "epoch": 8.15945689927376,
      "grad_norm": 5.334651947021484,
      "learning_rate": 3.289039750201734e-05,
      "loss": 0.3788,
      "step": 25841
    },
    {
      "epoch": 8.159772655509947,
      "grad_norm": 5.590600490570068,
      "learning_rate": 3.288952040136126e-05,
      "loss": 0.3257,
      "step": 25842
    },
    {
      "epoch": 8.160088411746132,
      "grad_norm": 5.477522373199463,
      "learning_rate": 3.2888643300705194e-05,
      "loss": 0.372,
      "step": 25843
    },
    {
      "epoch": 8.160404167982318,
      "grad_norm": 5.987421035766602,
      "learning_rate": 3.288776620004912e-05,
      "loss": 0.3749,
      "step": 25844
    },
    {
      "epoch": 8.160719924218503,
      "grad_norm": 6.253908634185791,
      "learning_rate": 3.288688909939305e-05,
      "loss": 0.4446,
      "step": 25845
    },
    {
      "epoch": 8.161035680454688,
      "grad_norm": 3.9673430919647217,
      "learning_rate": 3.2886011998736975e-05,
      "loss": 0.3688,
      "step": 25846
    },
    {
      "epoch": 8.161351436690875,
      "grad_norm": 6.08974027633667,
      "learning_rate": 3.288513489808091e-05,
      "loss": 0.404,
      "step": 25847
    },
    {
      "epoch": 8.16166719292706,
      "grad_norm": 5.703006267547607,
      "learning_rate": 3.288425779742483e-05,
      "loss": 0.3626,
      "step": 25848
    },
    {
      "epoch": 8.161982949163246,
      "grad_norm": 5.289701461791992,
      "learning_rate": 3.288338069676876e-05,
      "loss": 0.3876,
      "step": 25849
    },
    {
      "epoch": 8.162298705399431,
      "grad_norm": 8.278870582580566,
      "learning_rate": 3.288250359611269e-05,
      "loss": 0.4881,
      "step": 25850
    },
    {
      "epoch": 8.162614461635616,
      "grad_norm": 4.399487018585205,
      "learning_rate": 3.288162649545662e-05,
      "loss": 0.4163,
      "step": 25851
    },
    {
      "epoch": 8.162930217871804,
      "grad_norm": 6.585099220275879,
      "learning_rate": 3.2880749394800545e-05,
      "loss": 0.3633,
      "step": 25852
    },
    {
      "epoch": 8.163245974107989,
      "grad_norm": 6.2538676261901855,
      "learning_rate": 3.2879872294144476e-05,
      "loss": 0.4055,
      "step": 25853
    },
    {
      "epoch": 8.163561730344174,
      "grad_norm": 7.149543285369873,
      "learning_rate": 3.28789951934884e-05,
      "loss": 0.3274,
      "step": 25854
    },
    {
      "epoch": 8.16387748658036,
      "grad_norm": 5.835623264312744,
      "learning_rate": 3.287811809283233e-05,
      "loss": 0.3497,
      "step": 25855
    },
    {
      "epoch": 8.164193242816546,
      "grad_norm": 5.929296016693115,
      "learning_rate": 3.2877240992176265e-05,
      "loss": 0.4815,
      "step": 25856
    },
    {
      "epoch": 8.164508999052732,
      "grad_norm": 5.829503059387207,
      "learning_rate": 3.287636389152019e-05,
      "loss": 0.4049,
      "step": 25857
    },
    {
      "epoch": 8.164824755288917,
      "grad_norm": 4.981118679046631,
      "learning_rate": 3.287548679086412e-05,
      "loss": 0.402,
      "step": 25858
    },
    {
      "epoch": 8.165140511525102,
      "grad_norm": 5.35775089263916,
      "learning_rate": 3.287460969020805e-05,
      "loss": 0.4067,
      "step": 25859
    },
    {
      "epoch": 8.165456267761288,
      "grad_norm": 4.805267810821533,
      "learning_rate": 3.287373258955198e-05,
      "loss": 0.3916,
      "step": 25860
    },
    {
      "epoch": 8.165772023997475,
      "grad_norm": 5.879353046417236,
      "learning_rate": 3.287285548889591e-05,
      "loss": 0.458,
      "step": 25861
    },
    {
      "epoch": 8.16608778023366,
      "grad_norm": 4.633039951324463,
      "learning_rate": 3.287197838823984e-05,
      "loss": 0.4549,
      "step": 25862
    },
    {
      "epoch": 8.166403536469845,
      "grad_norm": 4.664970874786377,
      "learning_rate": 3.2871101287583766e-05,
      "loss": 0.3893,
      "step": 25863
    },
    {
      "epoch": 8.16671929270603,
      "grad_norm": 5.824599742889404,
      "learning_rate": 3.28702241869277e-05,
      "loss": 0.2682,
      "step": 25864
    },
    {
      "epoch": 8.167035048942216,
      "grad_norm": 5.089544773101807,
      "learning_rate": 3.286934708627162e-05,
      "loss": 0.3367,
      "step": 25865
    },
    {
      "epoch": 8.167350805178403,
      "grad_norm": 5.3793745040893555,
      "learning_rate": 3.2868469985615554e-05,
      "loss": 0.3058,
      "step": 25866
    },
    {
      "epoch": 8.167666561414588,
      "grad_norm": 6.33621883392334,
      "learning_rate": 3.286759288495948e-05,
      "loss": 0.42,
      "step": 25867
    },
    {
      "epoch": 8.167982317650774,
      "grad_norm": 4.828815937042236,
      "learning_rate": 3.286671578430341e-05,
      "loss": 0.3491,
      "step": 25868
    },
    {
      "epoch": 8.168298073886959,
      "grad_norm": 6.2159810066223145,
      "learning_rate": 3.2865838683647336e-05,
      "loss": 0.3974,
      "step": 25869
    },
    {
      "epoch": 8.168613830123144,
      "grad_norm": 5.000025272369385,
      "learning_rate": 3.286496158299127e-05,
      "loss": 0.3002,
      "step": 25870
    },
    {
      "epoch": 8.168929586359331,
      "grad_norm": 6.6531782150268555,
      "learning_rate": 3.286408448233519e-05,
      "loss": 0.4554,
      "step": 25871
    },
    {
      "epoch": 8.169245342595516,
      "grad_norm": 6.209685802459717,
      "learning_rate": 3.2863207381679124e-05,
      "loss": 0.3902,
      "step": 25872
    },
    {
      "epoch": 8.169561098831702,
      "grad_norm": 4.715317249298096,
      "learning_rate": 3.286233028102305e-05,
      "loss": 0.4305,
      "step": 25873
    },
    {
      "epoch": 8.169876855067887,
      "grad_norm": 5.505527973175049,
      "learning_rate": 3.286145318036698e-05,
      "loss": 0.4601,
      "step": 25874
    },
    {
      "epoch": 8.170192611304074,
      "grad_norm": 7.608757972717285,
      "learning_rate": 3.2860576079710906e-05,
      "loss": 0.5125,
      "step": 25875
    },
    {
      "epoch": 8.17050836754026,
      "grad_norm": 6.936064720153809,
      "learning_rate": 3.285969897905484e-05,
      "loss": 0.3418,
      "step": 25876
    },
    {
      "epoch": 8.170824123776445,
      "grad_norm": 5.909269332885742,
      "learning_rate": 3.285882187839876e-05,
      "loss": 0.3458,
      "step": 25877
    },
    {
      "epoch": 8.17113988001263,
      "grad_norm": 6.00295877456665,
      "learning_rate": 3.2857944777742694e-05,
      "loss": 0.512,
      "step": 25878
    },
    {
      "epoch": 8.171455636248815,
      "grad_norm": 7.988621711730957,
      "learning_rate": 3.2857067677086626e-05,
      "loss": 0.4475,
      "step": 25879
    },
    {
      "epoch": 8.171771392485002,
      "grad_norm": 6.713517189025879,
      "learning_rate": 3.285619057643055e-05,
      "loss": 0.4493,
      "step": 25880
    },
    {
      "epoch": 8.172087148721188,
      "grad_norm": 6.469115734100342,
      "learning_rate": 3.285531347577448e-05,
      "loss": 0.4271,
      "step": 25881
    },
    {
      "epoch": 8.172402904957373,
      "grad_norm": 5.3970537185668945,
      "learning_rate": 3.285443637511841e-05,
      "loss": 0.3163,
      "step": 25882
    },
    {
      "epoch": 8.172718661193558,
      "grad_norm": 4.705235958099365,
      "learning_rate": 3.285355927446234e-05,
      "loss": 0.3385,
      "step": 25883
    },
    {
      "epoch": 8.173034417429744,
      "grad_norm": 4.990393161773682,
      "learning_rate": 3.2852682173806264e-05,
      "loss": 0.4055,
      "step": 25884
    },
    {
      "epoch": 8.17335017366593,
      "grad_norm": 6.245896339416504,
      "learning_rate": 3.2851805073150195e-05,
      "loss": 0.3916,
      "step": 25885
    },
    {
      "epoch": 8.173665929902116,
      "grad_norm": 5.402750492095947,
      "learning_rate": 3.285092797249413e-05,
      "loss": 0.3331,
      "step": 25886
    },
    {
      "epoch": 8.173981686138301,
      "grad_norm": 6.770425319671631,
      "learning_rate": 3.285005087183806e-05,
      "loss": 0.4063,
      "step": 25887
    },
    {
      "epoch": 8.174297442374487,
      "grad_norm": 6.3772807121276855,
      "learning_rate": 3.2849173771181984e-05,
      "loss": 0.4138,
      "step": 25888
    },
    {
      "epoch": 8.174613198610672,
      "grad_norm": 7.039432048797607,
      "learning_rate": 3.2848296670525915e-05,
      "loss": 0.4254,
      "step": 25889
    },
    {
      "epoch": 8.174928954846859,
      "grad_norm": 6.590437412261963,
      "learning_rate": 3.284741956986984e-05,
      "loss": 0.4403,
      "step": 25890
    },
    {
      "epoch": 8.175244711083044,
      "grad_norm": 6.061173915863037,
      "learning_rate": 3.284654246921377e-05,
      "loss": 0.4716,
      "step": 25891
    },
    {
      "epoch": 8.17556046731923,
      "grad_norm": 7.5544586181640625,
      "learning_rate": 3.28456653685577e-05,
      "loss": 0.4553,
      "step": 25892
    },
    {
      "epoch": 8.175876223555415,
      "grad_norm": 6.44222354888916,
      "learning_rate": 3.284478826790163e-05,
      "loss": 0.4512,
      "step": 25893
    },
    {
      "epoch": 8.1761919797916,
      "grad_norm": 7.450717449188232,
      "learning_rate": 3.2843911167245553e-05,
      "loss": 0.4003,
      "step": 25894
    },
    {
      "epoch": 8.176507736027787,
      "grad_norm": 6.61693811416626,
      "learning_rate": 3.2843034066589485e-05,
      "loss": 0.4607,
      "step": 25895
    },
    {
      "epoch": 8.176823492263972,
      "grad_norm": 5.572332859039307,
      "learning_rate": 3.284215696593341e-05,
      "loss": 0.3263,
      "step": 25896
    },
    {
      "epoch": 8.177139248500158,
      "grad_norm": 5.152011871337891,
      "learning_rate": 3.284127986527734e-05,
      "loss": 0.4968,
      "step": 25897
    },
    {
      "epoch": 8.177455004736343,
      "grad_norm": 5.634253978729248,
      "learning_rate": 3.284040276462127e-05,
      "loss": 0.3984,
      "step": 25898
    },
    {
      "epoch": 8.177770760972528,
      "grad_norm": 7.349892616271973,
      "learning_rate": 3.28395256639652e-05,
      "loss": 0.3983,
      "step": 25899
    },
    {
      "epoch": 8.178086517208715,
      "grad_norm": 6.4336652755737305,
      "learning_rate": 3.283864856330913e-05,
      "loss": 0.4428,
      "step": 25900
    },
    {
      "epoch": 8.1784022734449,
      "grad_norm": 7.951562404632568,
      "learning_rate": 3.2837771462653055e-05,
      "loss": 0.4456,
      "step": 25901
    },
    {
      "epoch": 8.178718029681086,
      "grad_norm": 5.3889031410217285,
      "learning_rate": 3.2836894361996987e-05,
      "loss": 0.365,
      "step": 25902
    },
    {
      "epoch": 8.179033785917271,
      "grad_norm": 5.380388259887695,
      "learning_rate": 3.283601726134091e-05,
      "loss": 0.4503,
      "step": 25903
    },
    {
      "epoch": 8.179349542153458,
      "grad_norm": 6.234136581420898,
      "learning_rate": 3.283514016068484e-05,
      "loss": 0.3517,
      "step": 25904
    },
    {
      "epoch": 8.179665298389644,
      "grad_norm": 6.259365558624268,
      "learning_rate": 3.283426306002877e-05,
      "loss": 0.3844,
      "step": 25905
    },
    {
      "epoch": 8.179981054625829,
      "grad_norm": 6.465071678161621,
      "learning_rate": 3.28333859593727e-05,
      "loss": 0.4137,
      "step": 25906
    },
    {
      "epoch": 8.180296810862014,
      "grad_norm": 5.335835933685303,
      "learning_rate": 3.2832508858716625e-05,
      "loss": 0.3241,
      "step": 25907
    },
    {
      "epoch": 8.1806125670982,
      "grad_norm": 6.548624038696289,
      "learning_rate": 3.2831631758060556e-05,
      "loss": 0.4179,
      "step": 25908
    },
    {
      "epoch": 8.180928323334387,
      "grad_norm": 6.323729038238525,
      "learning_rate": 3.283075465740448e-05,
      "loss": 0.3478,
      "step": 25909
    },
    {
      "epoch": 8.181244079570572,
      "grad_norm": 5.593034744262695,
      "learning_rate": 3.282987755674841e-05,
      "loss": 0.4107,
      "step": 25910
    },
    {
      "epoch": 8.181559835806757,
      "grad_norm": 5.888402938842773,
      "learning_rate": 3.282900045609234e-05,
      "loss": 0.4519,
      "step": 25911
    },
    {
      "epoch": 8.181875592042942,
      "grad_norm": 6.911390781402588,
      "learning_rate": 3.282812335543627e-05,
      "loss": 0.4224,
      "step": 25912
    },
    {
      "epoch": 8.182191348279128,
      "grad_norm": 5.265666484832764,
      "learning_rate": 3.28272462547802e-05,
      "loss": 0.4099,
      "step": 25913
    },
    {
      "epoch": 8.182507104515315,
      "grad_norm": 4.24911642074585,
      "learning_rate": 3.282636915412413e-05,
      "loss": 0.3247,
      "step": 25914
    },
    {
      "epoch": 8.1828228607515,
      "grad_norm": 7.25627326965332,
      "learning_rate": 3.282549205346806e-05,
      "loss": 0.3421,
      "step": 25915
    },
    {
      "epoch": 8.183138616987685,
      "grad_norm": 6.18070650100708,
      "learning_rate": 3.282461495281199e-05,
      "loss": 0.4065,
      "step": 25916
    },
    {
      "epoch": 8.18345437322387,
      "grad_norm": 4.655592441558838,
      "learning_rate": 3.2823737852155914e-05,
      "loss": 0.3501,
      "step": 25917
    },
    {
      "epoch": 8.183770129460056,
      "grad_norm": 5.791947841644287,
      "learning_rate": 3.2822860751499846e-05,
      "loss": 0.464,
      "step": 25918
    },
    {
      "epoch": 8.184085885696243,
      "grad_norm": 5.483467102050781,
      "learning_rate": 3.282198365084377e-05,
      "loss": 0.4533,
      "step": 25919
    },
    {
      "epoch": 8.184401641932428,
      "grad_norm": 4.624977111816406,
      "learning_rate": 3.28211065501877e-05,
      "loss": 0.428,
      "step": 25920
    },
    {
      "epoch": 8.184717398168614,
      "grad_norm": 5.824831485748291,
      "learning_rate": 3.282022944953163e-05,
      "loss": 0.3977,
      "step": 25921
    },
    {
      "epoch": 8.185033154404799,
      "grad_norm": 5.31365966796875,
      "learning_rate": 3.281935234887556e-05,
      "loss": 0.4594,
      "step": 25922
    },
    {
      "epoch": 8.185348910640986,
      "grad_norm": 6.222723484039307,
      "learning_rate": 3.281847524821949e-05,
      "loss": 0.4748,
      "step": 25923
    },
    {
      "epoch": 8.185664666877171,
      "grad_norm": 6.010831356048584,
      "learning_rate": 3.2817598147563416e-05,
      "loss": 0.4688,
      "step": 25924
    },
    {
      "epoch": 8.185980423113357,
      "grad_norm": 5.247861385345459,
      "learning_rate": 3.281672104690735e-05,
      "loss": 0.3472,
      "step": 25925
    },
    {
      "epoch": 8.186296179349542,
      "grad_norm": 5.28084659576416,
      "learning_rate": 3.281584394625127e-05,
      "loss": 0.3331,
      "step": 25926
    },
    {
      "epoch": 8.186611935585727,
      "grad_norm": 6.409111022949219,
      "learning_rate": 3.2814966845595204e-05,
      "loss": 0.538,
      "step": 25927
    },
    {
      "epoch": 8.186927691821914,
      "grad_norm": 6.562021732330322,
      "learning_rate": 3.281408974493913e-05,
      "loss": 0.3735,
      "step": 25928
    },
    {
      "epoch": 8.1872434480581,
      "grad_norm": 4.738245010375977,
      "learning_rate": 3.281321264428306e-05,
      "loss": 0.3796,
      "step": 25929
    },
    {
      "epoch": 8.187559204294285,
      "grad_norm": 5.449736595153809,
      "learning_rate": 3.2812335543626986e-05,
      "loss": 0.4979,
      "step": 25930
    },
    {
      "epoch": 8.18787496053047,
      "grad_norm": 5.353478908538818,
      "learning_rate": 3.281145844297092e-05,
      "loss": 0.3297,
      "step": 25931
    },
    {
      "epoch": 8.188190716766655,
      "grad_norm": 7.105810165405273,
      "learning_rate": 3.281058134231484e-05,
      "loss": 0.3873,
      "step": 25932
    },
    {
      "epoch": 8.188506473002843,
      "grad_norm": 5.043678283691406,
      "learning_rate": 3.2809704241658774e-05,
      "loss": 0.3343,
      "step": 25933
    },
    {
      "epoch": 8.188822229239028,
      "grad_norm": 5.84392786026001,
      "learning_rate": 3.28088271410027e-05,
      "loss": 0.3967,
      "step": 25934
    },
    {
      "epoch": 8.189137985475213,
      "grad_norm": 6.102219581604004,
      "learning_rate": 3.280795004034663e-05,
      "loss": 0.5087,
      "step": 25935
    },
    {
      "epoch": 8.189453741711398,
      "grad_norm": 8.220527648925781,
      "learning_rate": 3.2807072939690555e-05,
      "loss": 0.362,
      "step": 25936
    },
    {
      "epoch": 8.189769497947584,
      "grad_norm": 6.393778324127197,
      "learning_rate": 3.280619583903449e-05,
      "loss": 0.3733,
      "step": 25937
    },
    {
      "epoch": 8.19008525418377,
      "grad_norm": 4.782287120819092,
      "learning_rate": 3.280531873837842e-05,
      "loss": 0.4128,
      "step": 25938
    },
    {
      "epoch": 8.190401010419956,
      "grad_norm": 6.176413536071777,
      "learning_rate": 3.2804441637722344e-05,
      "loss": 0.3082,
      "step": 25939
    },
    {
      "epoch": 8.190716766656141,
      "grad_norm": 6.146662712097168,
      "learning_rate": 3.2803564537066275e-05,
      "loss": 0.451,
      "step": 25940
    },
    {
      "epoch": 8.191032522892327,
      "grad_norm": 7.887166500091553,
      "learning_rate": 3.280268743641021e-05,
      "loss": 0.4376,
      "step": 25941
    },
    {
      "epoch": 8.191348279128512,
      "grad_norm": 5.564674377441406,
      "learning_rate": 3.280181033575413e-05,
      "loss": 0.3332,
      "step": 25942
    },
    {
      "epoch": 8.191664035364699,
      "grad_norm": 6.673681735992432,
      "learning_rate": 3.2800933235098064e-05,
      "loss": 0.4332,
      "step": 25943
    },
    {
      "epoch": 8.191979791600884,
      "grad_norm": 6.209815502166748,
      "learning_rate": 3.2800056134441995e-05,
      "loss": 0.4518,
      "step": 25944
    },
    {
      "epoch": 8.19229554783707,
      "grad_norm": 6.683390140533447,
      "learning_rate": 3.279917903378592e-05,
      "loss": 0.4314,
      "step": 25945
    },
    {
      "epoch": 8.192611304073255,
      "grad_norm": 7.68327522277832,
      "learning_rate": 3.279830193312985e-05,
      "loss": 0.4104,
      "step": 25946
    },
    {
      "epoch": 8.192927060309442,
      "grad_norm": 6.110327243804932,
      "learning_rate": 3.279742483247378e-05,
      "loss": 0.4582,
      "step": 25947
    },
    {
      "epoch": 8.193242816545627,
      "grad_norm": 6.867270469665527,
      "learning_rate": 3.279654773181771e-05,
      "loss": 0.3919,
      "step": 25948
    },
    {
      "epoch": 8.193558572781813,
      "grad_norm": 4.916481971740723,
      "learning_rate": 3.279567063116163e-05,
      "loss": 0.3192,
      "step": 25949
    },
    {
      "epoch": 8.193874329017998,
      "grad_norm": 8.65770149230957,
      "learning_rate": 3.2794793530505565e-05,
      "loss": 0.3898,
      "step": 25950
    },
    {
      "epoch": 8.194190085254183,
      "grad_norm": 4.629174709320068,
      "learning_rate": 3.279391642984949e-05,
      "loss": 0.3112,
      "step": 25951
    },
    {
      "epoch": 8.19450584149037,
      "grad_norm": 6.880432605743408,
      "learning_rate": 3.279303932919342e-05,
      "loss": 0.3975,
      "step": 25952
    },
    {
      "epoch": 8.194821597726555,
      "grad_norm": 6.312780380249023,
      "learning_rate": 3.2792162228537347e-05,
      "loss": 0.4327,
      "step": 25953
    },
    {
      "epoch": 8.19513735396274,
      "grad_norm": 6.488605499267578,
      "learning_rate": 3.279128512788128e-05,
      "loss": 0.3503,
      "step": 25954
    },
    {
      "epoch": 8.195453110198926,
      "grad_norm": 4.416995525360107,
      "learning_rate": 3.27904080272252e-05,
      "loss": 0.3241,
      "step": 25955
    },
    {
      "epoch": 8.195768866435111,
      "grad_norm": 3.982886791229248,
      "learning_rate": 3.2789530926569135e-05,
      "loss": 0.317,
      "step": 25956
    },
    {
      "epoch": 8.196084622671298,
      "grad_norm": 6.5237135887146,
      "learning_rate": 3.278865382591306e-05,
      "loss": 0.4532,
      "step": 25957
    },
    {
      "epoch": 8.196400378907484,
      "grad_norm": 7.483749866485596,
      "learning_rate": 3.278777672525699e-05,
      "loss": 0.3923,
      "step": 25958
    },
    {
      "epoch": 8.196716135143669,
      "grad_norm": 6.005581378936768,
      "learning_rate": 3.278689962460092e-05,
      "loss": 0.3665,
      "step": 25959
    },
    {
      "epoch": 8.197031891379854,
      "grad_norm": 5.777531623840332,
      "learning_rate": 3.278602252394485e-05,
      "loss": 0.3597,
      "step": 25960
    },
    {
      "epoch": 8.19734764761604,
      "grad_norm": 5.091935157775879,
      "learning_rate": 3.278514542328878e-05,
      "loss": 0.4126,
      "step": 25961
    },
    {
      "epoch": 8.197663403852227,
      "grad_norm": 5.412898063659668,
      "learning_rate": 3.2784268322632705e-05,
      "loss": 0.3747,
      "step": 25962
    },
    {
      "epoch": 8.197979160088412,
      "grad_norm": 6.122927665710449,
      "learning_rate": 3.2783391221976636e-05,
      "loss": 0.4155,
      "step": 25963
    },
    {
      "epoch": 8.198294916324597,
      "grad_norm": 6.729282855987549,
      "learning_rate": 3.278251412132056e-05,
      "loss": 0.4125,
      "step": 25964
    },
    {
      "epoch": 8.198610672560783,
      "grad_norm": 4.853704452514648,
      "learning_rate": 3.278163702066449e-05,
      "loss": 0.4388,
      "step": 25965
    },
    {
      "epoch": 8.198926428796968,
      "grad_norm": 4.510544776916504,
      "learning_rate": 3.278075992000842e-05,
      "loss": 0.4168,
      "step": 25966
    },
    {
      "epoch": 8.199242185033155,
      "grad_norm": 7.1315789222717285,
      "learning_rate": 3.277988281935235e-05,
      "loss": 0.4639,
      "step": 25967
    },
    {
      "epoch": 8.19955794126934,
      "grad_norm": 8.460555076599121,
      "learning_rate": 3.277900571869628e-05,
      "loss": 0.4829,
      "step": 25968
    },
    {
      "epoch": 8.199873697505526,
      "grad_norm": 5.556898593902588,
      "learning_rate": 3.277812861804021e-05,
      "loss": 0.3837,
      "step": 25969
    },
    {
      "epoch": 8.20018945374171,
      "grad_norm": 7.0535807609558105,
      "learning_rate": 3.277725151738414e-05,
      "loss": 0.4246,
      "step": 25970
    },
    {
      "epoch": 8.200505209977898,
      "grad_norm": 6.437206745147705,
      "learning_rate": 3.277637441672807e-05,
      "loss": 0.3651,
      "step": 25971
    },
    {
      "epoch": 8.200820966214083,
      "grad_norm": 5.9922308921813965,
      "learning_rate": 3.2775497316071994e-05,
      "loss": 0.4772,
      "step": 25972
    },
    {
      "epoch": 8.201136722450268,
      "grad_norm": 6.817697525024414,
      "learning_rate": 3.2774620215415926e-05,
      "loss": 0.3419,
      "step": 25973
    },
    {
      "epoch": 8.201452478686454,
      "grad_norm": 5.6442060470581055,
      "learning_rate": 3.277374311475985e-05,
      "loss": 0.3864,
      "step": 25974
    },
    {
      "epoch": 8.201768234922639,
      "grad_norm": 6.636898994445801,
      "learning_rate": 3.277286601410378e-05,
      "loss": 0.422,
      "step": 25975
    },
    {
      "epoch": 8.202083991158826,
      "grad_norm": 5.481508731842041,
      "learning_rate": 3.277198891344771e-05,
      "loss": 0.3265,
      "step": 25976
    },
    {
      "epoch": 8.202399747395011,
      "grad_norm": 6.205074310302734,
      "learning_rate": 3.277111181279164e-05,
      "loss": 0.392,
      "step": 25977
    },
    {
      "epoch": 8.202715503631197,
      "grad_norm": 5.847962379455566,
      "learning_rate": 3.2770234712135564e-05,
      "loss": 0.3934,
      "step": 25978
    },
    {
      "epoch": 8.203031259867382,
      "grad_norm": 5.604267597198486,
      "learning_rate": 3.2769357611479496e-05,
      "loss": 0.4186,
      "step": 25979
    },
    {
      "epoch": 8.203347016103567,
      "grad_norm": 5.934741020202637,
      "learning_rate": 3.276848051082342e-05,
      "loss": 0.4685,
      "step": 25980
    },
    {
      "epoch": 8.203662772339754,
      "grad_norm": 4.953113079071045,
      "learning_rate": 3.276760341016735e-05,
      "loss": 0.3683,
      "step": 25981
    },
    {
      "epoch": 8.20397852857594,
      "grad_norm": 6.443305969238281,
      "learning_rate": 3.2766726309511284e-05,
      "loss": 0.485,
      "step": 25982
    },
    {
      "epoch": 8.204294284812125,
      "grad_norm": 7.719144344329834,
      "learning_rate": 3.276584920885521e-05,
      "loss": 0.402,
      "step": 25983
    },
    {
      "epoch": 8.20461004104831,
      "grad_norm": 7.64414119720459,
      "learning_rate": 3.276497210819914e-05,
      "loss": 0.4379,
      "step": 25984
    },
    {
      "epoch": 8.204925797284496,
      "grad_norm": 6.1206159591674805,
      "learning_rate": 3.2764095007543065e-05,
      "loss": 0.3413,
      "step": 25985
    },
    {
      "epoch": 8.205241553520683,
      "grad_norm": 6.332059860229492,
      "learning_rate": 3.2763217906887e-05,
      "loss": 0.4508,
      "step": 25986
    },
    {
      "epoch": 8.205557309756868,
      "grad_norm": 4.8321380615234375,
      "learning_rate": 3.276234080623092e-05,
      "loss": 0.3662,
      "step": 25987
    },
    {
      "epoch": 8.205873065993053,
      "grad_norm": 5.052680969238281,
      "learning_rate": 3.2761463705574854e-05,
      "loss": 0.3514,
      "step": 25988
    },
    {
      "epoch": 8.206188822229239,
      "grad_norm": 6.769752502441406,
      "learning_rate": 3.276058660491878e-05,
      "loss": 0.4435,
      "step": 25989
    },
    {
      "epoch": 8.206504578465424,
      "grad_norm": 5.378875255584717,
      "learning_rate": 3.275970950426271e-05,
      "loss": 0.3667,
      "step": 25990
    },
    {
      "epoch": 8.20682033470161,
      "grad_norm": 6.195512771606445,
      "learning_rate": 3.2758832403606635e-05,
      "loss": 0.3651,
      "step": 25991
    },
    {
      "epoch": 8.207136090937796,
      "grad_norm": 7.790572643280029,
      "learning_rate": 3.275795530295057e-05,
      "loss": 0.4551,
      "step": 25992
    },
    {
      "epoch": 8.207451847173981,
      "grad_norm": 5.451693058013916,
      "learning_rate": 3.275707820229449e-05,
      "loss": 0.4285,
      "step": 25993
    },
    {
      "epoch": 8.207767603410167,
      "grad_norm": 7.8020453453063965,
      "learning_rate": 3.2756201101638424e-05,
      "loss": 0.3535,
      "step": 25994
    },
    {
      "epoch": 8.208083359646354,
      "grad_norm": 6.021552085876465,
      "learning_rate": 3.2755324000982355e-05,
      "loss": 0.3861,
      "step": 25995
    },
    {
      "epoch": 8.20839911588254,
      "grad_norm": 7.001058101654053,
      "learning_rate": 3.275444690032629e-05,
      "loss": 0.3552,
      "step": 25996
    },
    {
      "epoch": 8.208714872118724,
      "grad_norm": 4.322706699371338,
      "learning_rate": 3.275356979967021e-05,
      "loss": 0.2723,
      "step": 25997
    },
    {
      "epoch": 8.20903062835491,
      "grad_norm": 6.514237880706787,
      "learning_rate": 3.2752692699014143e-05,
      "loss": 0.3198,
      "step": 25998
    },
    {
      "epoch": 8.209346384591095,
      "grad_norm": 6.02908992767334,
      "learning_rate": 3.275181559835807e-05,
      "loss": 0.3761,
      "step": 25999
    },
    {
      "epoch": 8.209662140827282,
      "grad_norm": 5.953617095947266,
      "learning_rate": 3.2750938497702e-05,
      "loss": 0.3869,
      "step": 26000
    },
    {
      "epoch": 8.209662140827282,
      "eval_loss": 0.9912862181663513,
      "eval_runtime": 4.5865,
      "eval_samples_per_second": 542.896,
      "eval_steps_per_second": 4.579,
      "step": 26000
    },
    {
      "epoch": 8.209977897063467,
      "grad_norm": 6.510524749755859,
      "learning_rate": 3.2750061397045925e-05,
      "loss": 0.3138,
      "step": 26001
    },
    {
      "epoch": 8.210293653299653,
      "grad_norm": 7.271490573883057,
      "learning_rate": 3.274918429638986e-05,
      "loss": 0.4047,
      "step": 26002
    },
    {
      "epoch": 8.210609409535838,
      "grad_norm": 6.900976181030273,
      "learning_rate": 3.274830719573379e-05,
      "loss": 0.521,
      "step": 26003
    },
    {
      "epoch": 8.210925165772023,
      "grad_norm": 7.1865339279174805,
      "learning_rate": 3.274743009507771e-05,
      "loss": 0.345,
      "step": 26004
    },
    {
      "epoch": 8.21124092200821,
      "grad_norm": 6.691831588745117,
      "learning_rate": 3.2746552994421645e-05,
      "loss": 0.3657,
      "step": 26005
    },
    {
      "epoch": 8.211556678244396,
      "grad_norm": 8.103604316711426,
      "learning_rate": 3.274567589376557e-05,
      "loss": 0.475,
      "step": 26006
    },
    {
      "epoch": 8.211872434480581,
      "grad_norm": 6.948406219482422,
      "learning_rate": 3.27447987931095e-05,
      "loss": 0.3969,
      "step": 26007
    },
    {
      "epoch": 8.212188190716766,
      "grad_norm": 7.320162296295166,
      "learning_rate": 3.2743921692453426e-05,
      "loss": 0.3751,
      "step": 26008
    },
    {
      "epoch": 8.212503946952951,
      "grad_norm": 7.252155780792236,
      "learning_rate": 3.274304459179736e-05,
      "loss": 0.3059,
      "step": 26009
    },
    {
      "epoch": 8.212819703189139,
      "grad_norm": 5.445494651794434,
      "learning_rate": 3.274216749114128e-05,
      "loss": 0.3361,
      "step": 26010
    },
    {
      "epoch": 8.213135459425324,
      "grad_norm": 5.982250690460205,
      "learning_rate": 3.2741290390485215e-05,
      "loss": 0.3598,
      "step": 26011
    },
    {
      "epoch": 8.21345121566151,
      "grad_norm": 6.217508792877197,
      "learning_rate": 3.274041328982914e-05,
      "loss": 0.4704,
      "step": 26012
    },
    {
      "epoch": 8.213766971897694,
      "grad_norm": 5.425208568572998,
      "learning_rate": 3.273953618917307e-05,
      "loss": 0.435,
      "step": 26013
    },
    {
      "epoch": 8.214082728133882,
      "grad_norm": 6.573391437530518,
      "learning_rate": 3.2738659088516996e-05,
      "loss": 0.3419,
      "step": 26014
    },
    {
      "epoch": 8.214398484370067,
      "grad_norm": 7.105493545532227,
      "learning_rate": 3.273778198786093e-05,
      "loss": 0.3645,
      "step": 26015
    },
    {
      "epoch": 8.214714240606252,
      "grad_norm": 6.586075305938721,
      "learning_rate": 3.273690488720485e-05,
      "loss": 0.3586,
      "step": 26016
    },
    {
      "epoch": 8.215029996842437,
      "grad_norm": 5.088505744934082,
      "learning_rate": 3.2736027786548784e-05,
      "loss": 0.3285,
      "step": 26017
    },
    {
      "epoch": 8.215345753078623,
      "grad_norm": 5.42401647567749,
      "learning_rate": 3.273515068589271e-05,
      "loss": 0.4848,
      "step": 26018
    },
    {
      "epoch": 8.21566150931481,
      "grad_norm": 5.403776168823242,
      "learning_rate": 3.273427358523664e-05,
      "loss": 0.3585,
      "step": 26019
    },
    {
      "epoch": 8.215977265550995,
      "grad_norm": 5.241400241851807,
      "learning_rate": 3.273339648458057e-05,
      "loss": 0.3689,
      "step": 26020
    },
    {
      "epoch": 8.21629302178718,
      "grad_norm": 5.523692607879639,
      "learning_rate": 3.27325193839245e-05,
      "loss": 0.3592,
      "step": 26021
    },
    {
      "epoch": 8.216608778023366,
      "grad_norm": 6.109517574310303,
      "learning_rate": 3.273164228326843e-05,
      "loss": 0.3846,
      "step": 26022
    },
    {
      "epoch": 8.216924534259551,
      "grad_norm": 5.010289192199707,
      "learning_rate": 3.273076518261236e-05,
      "loss": 0.4014,
      "step": 26023
    },
    {
      "epoch": 8.217240290495738,
      "grad_norm": 6.006879806518555,
      "learning_rate": 3.2729888081956286e-05,
      "loss": 0.3626,
      "step": 26024
    },
    {
      "epoch": 8.217556046731923,
      "grad_norm": 6.302420616149902,
      "learning_rate": 3.272901098130022e-05,
      "loss": 0.3189,
      "step": 26025
    },
    {
      "epoch": 8.217871802968109,
      "grad_norm": 4.904211044311523,
      "learning_rate": 3.272813388064415e-05,
      "loss": 0.2962,
      "step": 26026
    },
    {
      "epoch": 8.218187559204294,
      "grad_norm": 6.3721113204956055,
      "learning_rate": 3.2727256779988074e-05,
      "loss": 0.3617,
      "step": 26027
    },
    {
      "epoch": 8.21850331544048,
      "grad_norm": 7.262555122375488,
      "learning_rate": 3.2726379679332006e-05,
      "loss": 0.537,
      "step": 26028
    },
    {
      "epoch": 8.218819071676666,
      "grad_norm": 5.609904766082764,
      "learning_rate": 3.272550257867593e-05,
      "loss": 0.3404,
      "step": 26029
    },
    {
      "epoch": 8.219134827912852,
      "grad_norm": 7.324138164520264,
      "learning_rate": 3.272462547801986e-05,
      "loss": 0.3341,
      "step": 26030
    },
    {
      "epoch": 8.219450584149037,
      "grad_norm": 6.371659755706787,
      "learning_rate": 3.272374837736379e-05,
      "loss": 0.5418,
      "step": 26031
    },
    {
      "epoch": 8.219766340385222,
      "grad_norm": 6.24009895324707,
      "learning_rate": 3.272287127670772e-05,
      "loss": 0.4206,
      "step": 26032
    },
    {
      "epoch": 8.220082096621407,
      "grad_norm": 5.409339427947998,
      "learning_rate": 3.2721994176051644e-05,
      "loss": 0.3939,
      "step": 26033
    },
    {
      "epoch": 8.220397852857595,
      "grad_norm": 7.121970176696777,
      "learning_rate": 3.2721117075395576e-05,
      "loss": 0.3973,
      "step": 26034
    },
    {
      "epoch": 8.22071360909378,
      "grad_norm": 6.1265668869018555,
      "learning_rate": 3.27202399747395e-05,
      "loss": 0.3912,
      "step": 26035
    },
    {
      "epoch": 8.221029365329965,
      "grad_norm": 5.630855560302734,
      "learning_rate": 3.271936287408343e-05,
      "loss": 0.3432,
      "step": 26036
    },
    {
      "epoch": 8.22134512156615,
      "grad_norm": 6.086127281188965,
      "learning_rate": 3.271848577342736e-05,
      "loss": 0.3437,
      "step": 26037
    },
    {
      "epoch": 8.221660877802337,
      "grad_norm": 4.850333213806152,
      "learning_rate": 3.271760867277129e-05,
      "loss": 0.3587,
      "step": 26038
    },
    {
      "epoch": 8.221976634038523,
      "grad_norm": 6.2832560539245605,
      "learning_rate": 3.2716731572115214e-05,
      "loss": 0.4265,
      "step": 26039
    },
    {
      "epoch": 8.222292390274708,
      "grad_norm": 4.7078857421875,
      "learning_rate": 3.2715854471459145e-05,
      "loss": 0.365,
      "step": 26040
    },
    {
      "epoch": 8.222608146510893,
      "grad_norm": 6.5404052734375,
      "learning_rate": 3.271497737080308e-05,
      "loss": 0.341,
      "step": 26041
    },
    {
      "epoch": 8.222923902747079,
      "grad_norm": 6.716195106506348,
      "learning_rate": 3.2714100270147e-05,
      "loss": 0.3141,
      "step": 26042
    },
    {
      "epoch": 8.223239658983266,
      "grad_norm": 6.745774745941162,
      "learning_rate": 3.2713223169490934e-05,
      "loss": 0.411,
      "step": 26043
    },
    {
      "epoch": 8.223555415219451,
      "grad_norm": 5.772684097290039,
      "learning_rate": 3.271234606883486e-05,
      "loss": 0.5056,
      "step": 26044
    },
    {
      "epoch": 8.223871171455636,
      "grad_norm": 22.5662841796875,
      "learning_rate": 3.271146896817879e-05,
      "loss": 0.486,
      "step": 26045
    },
    {
      "epoch": 8.224186927691822,
      "grad_norm": 6.674353122711182,
      "learning_rate": 3.2710591867522715e-05,
      "loss": 0.5218,
      "step": 26046
    },
    {
      "epoch": 8.224502683928007,
      "grad_norm": 6.324740409851074,
      "learning_rate": 3.270971476686665e-05,
      "loss": 0.3833,
      "step": 26047
    },
    {
      "epoch": 8.224818440164194,
      "grad_norm": 5.492585182189941,
      "learning_rate": 3.270883766621057e-05,
      "loss": 0.5527,
      "step": 26048
    },
    {
      "epoch": 8.22513419640038,
      "grad_norm": 5.139349460601807,
      "learning_rate": 3.2707960565554503e-05,
      "loss": 0.4519,
      "step": 26049
    },
    {
      "epoch": 8.225449952636565,
      "grad_norm": 5.031941890716553,
      "learning_rate": 3.2707083464898435e-05,
      "loss": 0.3378,
      "step": 26050
    },
    {
      "epoch": 8.22576570887275,
      "grad_norm": 8.798274993896484,
      "learning_rate": 3.270620636424237e-05,
      "loss": 0.3675,
      "step": 26051
    },
    {
      "epoch": 8.226081465108935,
      "grad_norm": 5.903726577758789,
      "learning_rate": 3.270532926358629e-05,
      "loss": 0.3866,
      "step": 26052
    },
    {
      "epoch": 8.226397221345122,
      "grad_norm": 5.448929786682129,
      "learning_rate": 3.270445216293022e-05,
      "loss": 0.3736,
      "step": 26053
    },
    {
      "epoch": 8.226712977581307,
      "grad_norm": 5.006805896759033,
      "learning_rate": 3.270357506227415e-05,
      "loss": 0.3615,
      "step": 26054
    },
    {
      "epoch": 8.227028733817493,
      "grad_norm": 4.787117004394531,
      "learning_rate": 3.270269796161808e-05,
      "loss": 0.3807,
      "step": 26055
    },
    {
      "epoch": 8.227344490053678,
      "grad_norm": 6.418102264404297,
      "learning_rate": 3.2701820860962005e-05,
      "loss": 0.4818,
      "step": 26056
    },
    {
      "epoch": 8.227660246289863,
      "grad_norm": 6.072298049926758,
      "learning_rate": 3.2700943760305937e-05,
      "loss": 0.4184,
      "step": 26057
    },
    {
      "epoch": 8.22797600252605,
      "grad_norm": 6.109467506408691,
      "learning_rate": 3.270006665964986e-05,
      "loss": 0.4474,
      "step": 26058
    },
    {
      "epoch": 8.228291758762236,
      "grad_norm": 7.469594955444336,
      "learning_rate": 3.269918955899379e-05,
      "loss": 0.4082,
      "step": 26059
    },
    {
      "epoch": 8.228607514998421,
      "grad_norm": 8.827743530273438,
      "learning_rate": 3.269831245833772e-05,
      "loss": 0.3832,
      "step": 26060
    },
    {
      "epoch": 8.228923271234606,
      "grad_norm": 6.346507549285889,
      "learning_rate": 3.269743535768165e-05,
      "loss": 0.4023,
      "step": 26061
    },
    {
      "epoch": 8.229239027470793,
      "grad_norm": 5.371551513671875,
      "learning_rate": 3.2696558257025575e-05,
      "loss": 0.3255,
      "step": 26062
    },
    {
      "epoch": 8.229554783706979,
      "grad_norm": 5.364680290222168,
      "learning_rate": 3.2695681156369506e-05,
      "loss": 0.4709,
      "step": 26063
    },
    {
      "epoch": 8.229870539943164,
      "grad_norm": 5.905700206756592,
      "learning_rate": 3.269480405571344e-05,
      "loss": 0.3464,
      "step": 26064
    },
    {
      "epoch": 8.23018629617935,
      "grad_norm": 5.74087381362915,
      "learning_rate": 3.269392695505736e-05,
      "loss": 0.4197,
      "step": 26065
    },
    {
      "epoch": 8.230502052415535,
      "grad_norm": 4.936283111572266,
      "learning_rate": 3.2693049854401295e-05,
      "loss": 0.3489,
      "step": 26066
    },
    {
      "epoch": 8.230817808651722,
      "grad_norm": 5.036103248596191,
      "learning_rate": 3.269217275374522e-05,
      "loss": 0.3157,
      "step": 26067
    },
    {
      "epoch": 8.231133564887907,
      "grad_norm": 6.903654098510742,
      "learning_rate": 3.269129565308915e-05,
      "loss": 0.3481,
      "step": 26068
    },
    {
      "epoch": 8.231449321124092,
      "grad_norm": 6.604973316192627,
      "learning_rate": 3.2690418552433076e-05,
      "loss": 0.483,
      "step": 26069
    },
    {
      "epoch": 8.231765077360278,
      "grad_norm": 5.922421932220459,
      "learning_rate": 3.268954145177701e-05,
      "loss": 0.4708,
      "step": 26070
    },
    {
      "epoch": 8.232080833596463,
      "grad_norm": 7.49973726272583,
      "learning_rate": 3.268866435112093e-05,
      "loss": 0.4643,
      "step": 26071
    },
    {
      "epoch": 8.23239658983265,
      "grad_norm": 5.255098342895508,
      "learning_rate": 3.2687787250464864e-05,
      "loss": 0.3709,
      "step": 26072
    },
    {
      "epoch": 8.232712346068835,
      "grad_norm": 6.246613502502441,
      "learning_rate": 3.268691014980879e-05,
      "loss": 0.4313,
      "step": 26073
    },
    {
      "epoch": 8.23302810230502,
      "grad_norm": 5.726693630218506,
      "learning_rate": 3.268603304915272e-05,
      "loss": 0.4719,
      "step": 26074
    },
    {
      "epoch": 8.233343858541206,
      "grad_norm": 7.066476345062256,
      "learning_rate": 3.2685155948496646e-05,
      "loss": 0.4791,
      "step": 26075
    },
    {
      "epoch": 8.233659614777391,
      "grad_norm": 7.284405708312988,
      "learning_rate": 3.268427884784058e-05,
      "loss": 0.4081,
      "step": 26076
    },
    {
      "epoch": 8.233975371013578,
      "grad_norm": 6.2858076095581055,
      "learning_rate": 3.268340174718451e-05,
      "loss": 0.4349,
      "step": 26077
    },
    {
      "epoch": 8.234291127249763,
      "grad_norm": 7.071824550628662,
      "learning_rate": 3.268252464652844e-05,
      "loss": 0.4301,
      "step": 26078
    },
    {
      "epoch": 8.234606883485949,
      "grad_norm": 4.783841133117676,
      "learning_rate": 3.2681647545872366e-05,
      "loss": 0.3019,
      "step": 26079
    },
    {
      "epoch": 8.234922639722134,
      "grad_norm": 6.114084243774414,
      "learning_rate": 3.26807704452163e-05,
      "loss": 0.3713,
      "step": 26080
    },
    {
      "epoch": 8.23523839595832,
      "grad_norm": 4.503635406494141,
      "learning_rate": 3.267989334456022e-05,
      "loss": 0.385,
      "step": 26081
    },
    {
      "epoch": 8.235554152194506,
      "grad_norm": 6.559690475463867,
      "learning_rate": 3.2679016243904154e-05,
      "loss": 0.4168,
      "step": 26082
    },
    {
      "epoch": 8.235869908430692,
      "grad_norm": 6.76216459274292,
      "learning_rate": 3.267813914324808e-05,
      "loss": 0.3628,
      "step": 26083
    },
    {
      "epoch": 8.236185664666877,
      "grad_norm": 3.838977336883545,
      "learning_rate": 3.267726204259201e-05,
      "loss": 0.4176,
      "step": 26084
    },
    {
      "epoch": 8.236501420903062,
      "grad_norm": 5.481271743774414,
      "learning_rate": 3.267638494193594e-05,
      "loss": 0.4137,
      "step": 26085
    },
    {
      "epoch": 8.23681717713925,
      "grad_norm": 5.345052719116211,
      "learning_rate": 3.267550784127987e-05,
      "loss": 0.3683,
      "step": 26086
    },
    {
      "epoch": 8.237132933375435,
      "grad_norm": 7.442383289337158,
      "learning_rate": 3.26746307406238e-05,
      "loss": 0.497,
      "step": 26087
    },
    {
      "epoch": 8.23744868961162,
      "grad_norm": 6.2952656745910645,
      "learning_rate": 3.2673753639967724e-05,
      "loss": 0.4279,
      "step": 26088
    },
    {
      "epoch": 8.237764445847805,
      "grad_norm": 6.6755242347717285,
      "learning_rate": 3.2672876539311656e-05,
      "loss": 0.3983,
      "step": 26089
    },
    {
      "epoch": 8.23808020208399,
      "grad_norm": 5.681911468505859,
      "learning_rate": 3.267199943865558e-05,
      "loss": 0.3557,
      "step": 26090
    },
    {
      "epoch": 8.238395958320178,
      "grad_norm": 4.8427534103393555,
      "learning_rate": 3.267112233799951e-05,
      "loss": 0.3761,
      "step": 26091
    },
    {
      "epoch": 8.238711714556363,
      "grad_norm": 7.640591144561768,
      "learning_rate": 3.267024523734344e-05,
      "loss": 0.3833,
      "step": 26092
    },
    {
      "epoch": 8.239027470792548,
      "grad_norm": 6.682860374450684,
      "learning_rate": 3.266936813668737e-05,
      "loss": 0.3514,
      "step": 26093
    },
    {
      "epoch": 8.239343227028733,
      "grad_norm": 5.3267741203308105,
      "learning_rate": 3.2668491036031294e-05,
      "loss": 0.4556,
      "step": 26094
    },
    {
      "epoch": 8.239658983264919,
      "grad_norm": 6.189447402954102,
      "learning_rate": 3.2667613935375225e-05,
      "loss": 0.4139,
      "step": 26095
    },
    {
      "epoch": 8.239974739501106,
      "grad_norm": 5.377887725830078,
      "learning_rate": 3.266673683471915e-05,
      "loss": 0.3857,
      "step": 26096
    },
    {
      "epoch": 8.240290495737291,
      "grad_norm": 5.414181232452393,
      "learning_rate": 3.266585973406308e-05,
      "loss": 0.3941,
      "step": 26097
    },
    {
      "epoch": 8.240606251973476,
      "grad_norm": 5.90826940536499,
      "learning_rate": 3.266498263340701e-05,
      "loss": 0.4386,
      "step": 26098
    },
    {
      "epoch": 8.240922008209662,
      "grad_norm": 5.377837181091309,
      "learning_rate": 3.266410553275094e-05,
      "loss": 0.3679,
      "step": 26099
    },
    {
      "epoch": 8.241237764445847,
      "grad_norm": 5.38578987121582,
      "learning_rate": 3.266322843209486e-05,
      "loss": 0.4231,
      "step": 26100
    },
    {
      "epoch": 8.241553520682034,
      "grad_norm": 6.946491241455078,
      "learning_rate": 3.2662351331438795e-05,
      "loss": 0.4178,
      "step": 26101
    },
    {
      "epoch": 8.24186927691822,
      "grad_norm": 6.11614990234375,
      "learning_rate": 3.266147423078273e-05,
      "loss": 0.4234,
      "step": 26102
    },
    {
      "epoch": 8.242185033154405,
      "grad_norm": 4.517524242401123,
      "learning_rate": 3.266059713012665e-05,
      "loss": 0.2897,
      "step": 26103
    },
    {
      "epoch": 8.24250078939059,
      "grad_norm": 6.476355075836182,
      "learning_rate": 3.265972002947058e-05,
      "loss": 0.4431,
      "step": 26104
    },
    {
      "epoch": 8.242816545626777,
      "grad_norm": 4.980909824371338,
      "learning_rate": 3.2658842928814515e-05,
      "loss": 0.3291,
      "step": 26105
    },
    {
      "epoch": 8.243132301862962,
      "grad_norm": 6.373166084289551,
      "learning_rate": 3.265796582815845e-05,
      "loss": 0.3817,
      "step": 26106
    },
    {
      "epoch": 8.243448058099148,
      "grad_norm": 7.289117336273193,
      "learning_rate": 3.265708872750237e-05,
      "loss": 0.4658,
      "step": 26107
    },
    {
      "epoch": 8.243763814335333,
      "grad_norm": 5.3622565269470215,
      "learning_rate": 3.26562116268463e-05,
      "loss": 0.336,
      "step": 26108
    },
    {
      "epoch": 8.244079570571518,
      "grad_norm": 6.293656826019287,
      "learning_rate": 3.265533452619023e-05,
      "loss": 0.4114,
      "step": 26109
    },
    {
      "epoch": 8.244395326807705,
      "grad_norm": 5.839355945587158,
      "learning_rate": 3.265445742553416e-05,
      "loss": 0.3638,
      "step": 26110
    },
    {
      "epoch": 8.24471108304389,
      "grad_norm": 5.915808200836182,
      "learning_rate": 3.2653580324878085e-05,
      "loss": 0.3691,
      "step": 26111
    },
    {
      "epoch": 8.245026839280076,
      "grad_norm": 6.5315117835998535,
      "learning_rate": 3.2652703224222016e-05,
      "loss": 0.4376,
      "step": 26112
    },
    {
      "epoch": 8.245342595516261,
      "grad_norm": 5.559147834777832,
      "learning_rate": 3.265182612356594e-05,
      "loss": 0.3911,
      "step": 26113
    },
    {
      "epoch": 8.245658351752446,
      "grad_norm": 4.373565673828125,
      "learning_rate": 3.265094902290987e-05,
      "loss": 0.3391,
      "step": 26114
    },
    {
      "epoch": 8.245974107988634,
      "grad_norm": 6.041567325592041,
      "learning_rate": 3.26500719222538e-05,
      "loss": 0.3248,
      "step": 26115
    },
    {
      "epoch": 8.246289864224819,
      "grad_norm": 7.752862930297852,
      "learning_rate": 3.264919482159773e-05,
      "loss": 0.4485,
      "step": 26116
    },
    {
      "epoch": 8.246605620461004,
      "grad_norm": 6.132533073425293,
      "learning_rate": 3.2648317720941655e-05,
      "loss": 0.405,
      "step": 26117
    },
    {
      "epoch": 8.24692137669719,
      "grad_norm": 5.606957912445068,
      "learning_rate": 3.2647440620285586e-05,
      "loss": 0.469,
      "step": 26118
    },
    {
      "epoch": 8.247237132933375,
      "grad_norm": 6.18855094909668,
      "learning_rate": 3.264656351962951e-05,
      "loss": 0.4267,
      "step": 26119
    },
    {
      "epoch": 8.247552889169562,
      "grad_norm": 7.088528633117676,
      "learning_rate": 3.264568641897344e-05,
      "loss": 0.3654,
      "step": 26120
    },
    {
      "epoch": 8.247868645405747,
      "grad_norm": 5.886485576629639,
      "learning_rate": 3.264480931831737e-05,
      "loss": 0.3676,
      "step": 26121
    },
    {
      "epoch": 8.248184401641932,
      "grad_norm": 6.748645782470703,
      "learning_rate": 3.26439322176613e-05,
      "loss": 0.4378,
      "step": 26122
    },
    {
      "epoch": 8.248500157878118,
      "grad_norm": 7.509986877441406,
      "learning_rate": 3.264305511700523e-05,
      "loss": 0.4855,
      "step": 26123
    },
    {
      "epoch": 8.248815914114303,
      "grad_norm": 7.868681907653809,
      "learning_rate": 3.2642178016349156e-05,
      "loss": 0.5624,
      "step": 26124
    },
    {
      "epoch": 8.24913167035049,
      "grad_norm": 6.0161871910095215,
      "learning_rate": 3.264130091569309e-05,
      "loss": 0.3734,
      "step": 26125
    },
    {
      "epoch": 8.249447426586675,
      "grad_norm": 5.218997478485107,
      "learning_rate": 3.264042381503701e-05,
      "loss": 0.3886,
      "step": 26126
    },
    {
      "epoch": 8.24976318282286,
      "grad_norm": 5.341125011444092,
      "learning_rate": 3.2639546714380944e-05,
      "loss": 0.3665,
      "step": 26127
    },
    {
      "epoch": 8.250078939059046,
      "grad_norm": 6.948755264282227,
      "learning_rate": 3.263866961372487e-05,
      "loss": 0.4335,
      "step": 26128
    },
    {
      "epoch": 8.250394695295231,
      "grad_norm": 4.230447769165039,
      "learning_rate": 3.26377925130688e-05,
      "loss": 0.3552,
      "step": 26129
    },
    {
      "epoch": 8.250710451531418,
      "grad_norm": 6.075901031494141,
      "learning_rate": 3.2636915412412726e-05,
      "loss": 0.4257,
      "step": 26130
    },
    {
      "epoch": 8.251026207767604,
      "grad_norm": 6.264490127563477,
      "learning_rate": 3.2636038311756664e-05,
      "loss": 0.4576,
      "step": 26131
    },
    {
      "epoch": 8.251341964003789,
      "grad_norm": 6.780104637145996,
      "learning_rate": 3.263516121110059e-05,
      "loss": 0.3614,
      "step": 26132
    },
    {
      "epoch": 8.251657720239974,
      "grad_norm": 5.682136535644531,
      "learning_rate": 3.263428411044452e-05,
      "loss": 0.4972,
      "step": 26133
    },
    {
      "epoch": 8.251973476476161,
      "grad_norm": 5.279261112213135,
      "learning_rate": 3.2633407009788446e-05,
      "loss": 0.3627,
      "step": 26134
    },
    {
      "epoch": 8.252289232712346,
      "grad_norm": 6.918431282043457,
      "learning_rate": 3.263252990913238e-05,
      "loss": 0.3766,
      "step": 26135
    },
    {
      "epoch": 8.252604988948532,
      "grad_norm": 6.125942230224609,
      "learning_rate": 3.26316528084763e-05,
      "loss": 0.373,
      "step": 26136
    },
    {
      "epoch": 8.252920745184717,
      "grad_norm": 7.732995510101318,
      "learning_rate": 3.2630775707820234e-05,
      "loss": 0.4159,
      "step": 26137
    },
    {
      "epoch": 8.253236501420902,
      "grad_norm": 8.014564514160156,
      "learning_rate": 3.262989860716416e-05,
      "loss": 0.4425,
      "step": 26138
    },
    {
      "epoch": 8.25355225765709,
      "grad_norm": 8.9927339553833,
      "learning_rate": 3.262902150650809e-05,
      "loss": 0.4056,
      "step": 26139
    },
    {
      "epoch": 8.253868013893275,
      "grad_norm": 5.117563247680664,
      "learning_rate": 3.2628144405852015e-05,
      "loss": 0.4435,
      "step": 26140
    },
    {
      "epoch": 8.25418377012946,
      "grad_norm": 6.544697284698486,
      "learning_rate": 3.262726730519595e-05,
      "loss": 0.3252,
      "step": 26141
    },
    {
      "epoch": 8.254499526365645,
      "grad_norm": 6.026458740234375,
      "learning_rate": 3.262639020453987e-05,
      "loss": 0.3896,
      "step": 26142
    },
    {
      "epoch": 8.25481528260183,
      "grad_norm": 5.748053073883057,
      "learning_rate": 3.2625513103883804e-05,
      "loss": 0.4464,
      "step": 26143
    },
    {
      "epoch": 8.255131038838018,
      "grad_norm": 5.952536582946777,
      "learning_rate": 3.2624636003227735e-05,
      "loss": 0.4768,
      "step": 26144
    },
    {
      "epoch": 8.255446795074203,
      "grad_norm": 4.2908616065979,
      "learning_rate": 3.262375890257166e-05,
      "loss": 0.3425,
      "step": 26145
    },
    {
      "epoch": 8.255762551310388,
      "grad_norm": 5.835071563720703,
      "learning_rate": 3.262288180191559e-05,
      "loss": 0.4245,
      "step": 26146
    },
    {
      "epoch": 8.256078307546574,
      "grad_norm": 6.567018985748291,
      "learning_rate": 3.262200470125952e-05,
      "loss": 0.3967,
      "step": 26147
    },
    {
      "epoch": 8.256394063782759,
      "grad_norm": 7.024683475494385,
      "learning_rate": 3.262112760060345e-05,
      "loss": 0.4938,
      "step": 26148
    },
    {
      "epoch": 8.256709820018946,
      "grad_norm": 5.891550540924072,
      "learning_rate": 3.2620250499947373e-05,
      "loss": 0.4516,
      "step": 26149
    },
    {
      "epoch": 8.257025576255131,
      "grad_norm": 6.9412360191345215,
      "learning_rate": 3.2619373399291305e-05,
      "loss": 0.5275,
      "step": 26150
    },
    {
      "epoch": 8.257341332491317,
      "grad_norm": 7.025164604187012,
      "learning_rate": 3.261849629863523e-05,
      "loss": 0.4023,
      "step": 26151
    },
    {
      "epoch": 8.257657088727502,
      "grad_norm": 4.310440540313721,
      "learning_rate": 3.261761919797916e-05,
      "loss": 0.33,
      "step": 26152
    },
    {
      "epoch": 8.257972844963689,
      "grad_norm": 4.938216686248779,
      "learning_rate": 3.261674209732309e-05,
      "loss": 0.3001,
      "step": 26153
    },
    {
      "epoch": 8.258288601199874,
      "grad_norm": 7.383237361907959,
      "learning_rate": 3.261586499666702e-05,
      "loss": 0.4594,
      "step": 26154
    },
    {
      "epoch": 8.25860435743606,
      "grad_norm": 7.945244789123535,
      "learning_rate": 3.261498789601094e-05,
      "loss": 0.493,
      "step": 26155
    },
    {
      "epoch": 8.258920113672245,
      "grad_norm": 6.880100250244141,
      "learning_rate": 3.2614110795354875e-05,
      "loss": 0.4429,
      "step": 26156
    },
    {
      "epoch": 8.25923586990843,
      "grad_norm": 4.778326511383057,
      "learning_rate": 3.26132336946988e-05,
      "loss": 0.3865,
      "step": 26157
    },
    {
      "epoch": 8.259551626144617,
      "grad_norm": 6.629410743713379,
      "learning_rate": 3.261235659404273e-05,
      "loss": 0.3624,
      "step": 26158
    },
    {
      "epoch": 8.259867382380802,
      "grad_norm": 5.377645969390869,
      "learning_rate": 3.261147949338666e-05,
      "loss": 0.4075,
      "step": 26159
    },
    {
      "epoch": 8.260183138616988,
      "grad_norm": 8.320406913757324,
      "learning_rate": 3.2610602392730595e-05,
      "loss": 0.4132,
      "step": 26160
    },
    {
      "epoch": 8.260498894853173,
      "grad_norm": 7.711541652679443,
      "learning_rate": 3.260972529207452e-05,
      "loss": 0.4767,
      "step": 26161
    },
    {
      "epoch": 8.260814651089358,
      "grad_norm": 4.677454948425293,
      "learning_rate": 3.260884819141845e-05,
      "loss": 0.327,
      "step": 26162
    },
    {
      "epoch": 8.261130407325545,
      "grad_norm": 4.856757640838623,
      "learning_rate": 3.2607971090762376e-05,
      "loss": 0.4103,
      "step": 26163
    },
    {
      "epoch": 8.26144616356173,
      "grad_norm": 8.383312225341797,
      "learning_rate": 3.260709399010631e-05,
      "loss": 0.4796,
      "step": 26164
    },
    {
      "epoch": 8.261761919797916,
      "grad_norm": 4.148979663848877,
      "learning_rate": 3.260621688945023e-05,
      "loss": 0.3235,
      "step": 26165
    },
    {
      "epoch": 8.262077676034101,
      "grad_norm": 6.515260219573975,
      "learning_rate": 3.2605339788794165e-05,
      "loss": 0.4868,
      "step": 26166
    },
    {
      "epoch": 8.262393432270287,
      "grad_norm": 4.5557475090026855,
      "learning_rate": 3.2604462688138096e-05,
      "loss": 0.3557,
      "step": 26167
    },
    {
      "epoch": 8.262709188506474,
      "grad_norm": 5.649381160736084,
      "learning_rate": 3.260358558748202e-05,
      "loss": 0.337,
      "step": 26168
    },
    {
      "epoch": 8.263024944742659,
      "grad_norm": 5.174560070037842,
      "learning_rate": 3.260270848682595e-05,
      "loss": 0.4634,
      "step": 26169
    },
    {
      "epoch": 8.263340700978844,
      "grad_norm": 6.643150806427002,
      "learning_rate": 3.260183138616988e-05,
      "loss": 0.3803,
      "step": 26170
    },
    {
      "epoch": 8.26365645721503,
      "grad_norm": 7.455817699432373,
      "learning_rate": 3.260095428551381e-05,
      "loss": 0.4495,
      "step": 26171
    },
    {
      "epoch": 8.263972213451215,
      "grad_norm": 4.980039596557617,
      "learning_rate": 3.2600077184857734e-05,
      "loss": 0.3357,
      "step": 26172
    },
    {
      "epoch": 8.264287969687402,
      "grad_norm": 5.757614612579346,
      "learning_rate": 3.2599200084201666e-05,
      "loss": 0.372,
      "step": 26173
    },
    {
      "epoch": 8.264603725923587,
      "grad_norm": 6.586486339569092,
      "learning_rate": 3.259832298354559e-05,
      "loss": 0.4389,
      "step": 26174
    },
    {
      "epoch": 8.264919482159772,
      "grad_norm": 5.629745960235596,
      "learning_rate": 3.259744588288952e-05,
      "loss": 0.3943,
      "step": 26175
    },
    {
      "epoch": 8.265235238395958,
      "grad_norm": 6.182028293609619,
      "learning_rate": 3.259656878223345e-05,
      "loss": 0.3745,
      "step": 26176
    },
    {
      "epoch": 8.265550994632145,
      "grad_norm": 5.578061103820801,
      "learning_rate": 3.259569168157738e-05,
      "loss": 0.3983,
      "step": 26177
    },
    {
      "epoch": 8.26586675086833,
      "grad_norm": 5.769967079162598,
      "learning_rate": 3.2594814580921304e-05,
      "loss": 0.3563,
      "step": 26178
    },
    {
      "epoch": 8.266182507104515,
      "grad_norm": 6.232571125030518,
      "learning_rate": 3.2593937480265236e-05,
      "loss": 0.3693,
      "step": 26179
    },
    {
      "epoch": 8.2664982633407,
      "grad_norm": 5.303381443023682,
      "learning_rate": 3.259306037960916e-05,
      "loss": 0.3617,
      "step": 26180
    },
    {
      "epoch": 8.266814019576886,
      "grad_norm": 4.992043495178223,
      "learning_rate": 3.259218327895309e-05,
      "loss": 0.3741,
      "step": 26181
    },
    {
      "epoch": 8.267129775813073,
      "grad_norm": 6.929670810699463,
      "learning_rate": 3.2591306178297024e-05,
      "loss": 0.4876,
      "step": 26182
    },
    {
      "epoch": 8.267445532049258,
      "grad_norm": 7.850572109222412,
      "learning_rate": 3.259042907764095e-05,
      "loss": 0.4177,
      "step": 26183
    },
    {
      "epoch": 8.267761288285444,
      "grad_norm": 5.478618621826172,
      "learning_rate": 3.258955197698488e-05,
      "loss": 0.3343,
      "step": 26184
    },
    {
      "epoch": 8.268077044521629,
      "grad_norm": 6.715656757354736,
      "learning_rate": 3.2588674876328806e-05,
      "loss": 0.4047,
      "step": 26185
    },
    {
      "epoch": 8.268392800757814,
      "grad_norm": 5.338883399963379,
      "learning_rate": 3.258779777567274e-05,
      "loss": 0.3505,
      "step": 26186
    },
    {
      "epoch": 8.268708556994001,
      "grad_norm": 7.408463001251221,
      "learning_rate": 3.258692067501667e-05,
      "loss": 0.3536,
      "step": 26187
    },
    {
      "epoch": 8.269024313230187,
      "grad_norm": 7.781749248504639,
      "learning_rate": 3.25860435743606e-05,
      "loss": 0.4639,
      "step": 26188
    },
    {
      "epoch": 8.269340069466372,
      "grad_norm": 12.781335830688477,
      "learning_rate": 3.2585166473704526e-05,
      "loss": 0.3109,
      "step": 26189
    },
    {
      "epoch": 8.269655825702557,
      "grad_norm": 5.794554710388184,
      "learning_rate": 3.258428937304846e-05,
      "loss": 0.4416,
      "step": 26190
    },
    {
      "epoch": 8.269971581938742,
      "grad_norm": 6.974004745483398,
      "learning_rate": 3.258341227239238e-05,
      "loss": 0.3208,
      "step": 26191
    },
    {
      "epoch": 8.27028733817493,
      "grad_norm": 5.687220096588135,
      "learning_rate": 3.2582535171736314e-05,
      "loss": 0.4837,
      "step": 26192
    },
    {
      "epoch": 8.270603094411115,
      "grad_norm": 5.159718990325928,
      "learning_rate": 3.258165807108024e-05,
      "loss": 0.387,
      "step": 26193
    },
    {
      "epoch": 8.2709188506473,
      "grad_norm": 6.160956859588623,
      "learning_rate": 3.258078097042417e-05,
      "loss": 0.3576,
      "step": 26194
    },
    {
      "epoch": 8.271234606883485,
      "grad_norm": 5.096312999725342,
      "learning_rate": 3.2579903869768095e-05,
      "loss": 0.3575,
      "step": 26195
    },
    {
      "epoch": 8.271550363119673,
      "grad_norm": 5.65768575668335,
      "learning_rate": 3.257902676911203e-05,
      "loss": 0.4653,
      "step": 26196
    },
    {
      "epoch": 8.271866119355858,
      "grad_norm": 7.7237701416015625,
      "learning_rate": 3.257814966845595e-05,
      "loss": 0.5066,
      "step": 26197
    },
    {
      "epoch": 8.272181875592043,
      "grad_norm": 6.4172773361206055,
      "learning_rate": 3.2577272567799884e-05,
      "loss": 0.4432,
      "step": 26198
    },
    {
      "epoch": 8.272497631828228,
      "grad_norm": 5.568349838256836,
      "learning_rate": 3.257639546714381e-05,
      "loss": 0.3358,
      "step": 26199
    },
    {
      "epoch": 8.272813388064414,
      "grad_norm": 5.2790350914001465,
      "learning_rate": 3.257551836648774e-05,
      "loss": 0.3714,
      "step": 26200
    },
    {
      "epoch": 8.272813388064414,
      "eval_loss": 0.9815741777420044,
      "eval_runtime": 4.7649,
      "eval_samples_per_second": 522.568,
      "eval_steps_per_second": 4.407,
      "step": 26200
    },
    {
      "epoch": 8.2731291443006,
      "grad_norm": 4.477412223815918,
      "learning_rate": 3.2574641265831665e-05,
      "loss": 0.359,
      "step": 26201
    },
    {
      "epoch": 8.273444900536786,
      "grad_norm": 6.703719615936279,
      "learning_rate": 3.25737641651756e-05,
      "loss": 0.394,
      "step": 26202
    },
    {
      "epoch": 8.273760656772971,
      "grad_norm": 5.024260997772217,
      "learning_rate": 3.257288706451952e-05,
      "loss": 0.3635,
      "step": 26203
    },
    {
      "epoch": 8.274076413009157,
      "grad_norm": 5.947843074798584,
      "learning_rate": 3.257200996386345e-05,
      "loss": 0.4884,
      "step": 26204
    },
    {
      "epoch": 8.274392169245342,
      "grad_norm": 7.7029266357421875,
      "learning_rate": 3.2571132863207385e-05,
      "loss": 0.4292,
      "step": 26205
    },
    {
      "epoch": 8.274707925481529,
      "grad_norm": 6.946162223815918,
      "learning_rate": 3.257025576255131e-05,
      "loss": 0.4238,
      "step": 26206
    },
    {
      "epoch": 8.275023681717714,
      "grad_norm": 6.59590482711792,
      "learning_rate": 3.256937866189524e-05,
      "loss": 0.4541,
      "step": 26207
    },
    {
      "epoch": 8.2753394379539,
      "grad_norm": 5.918127059936523,
      "learning_rate": 3.2568501561239167e-05,
      "loss": 0.4051,
      "step": 26208
    },
    {
      "epoch": 8.275655194190085,
      "grad_norm": 8.349365234375,
      "learning_rate": 3.25676244605831e-05,
      "loss": 0.5146,
      "step": 26209
    },
    {
      "epoch": 8.27597095042627,
      "grad_norm": 7.6465888023376465,
      "learning_rate": 3.256674735992702e-05,
      "loss": 0.352,
      "step": 26210
    },
    {
      "epoch": 8.276286706662457,
      "grad_norm": 7.192287445068359,
      "learning_rate": 3.2565870259270955e-05,
      "loss": 0.3832,
      "step": 26211
    },
    {
      "epoch": 8.276602462898643,
      "grad_norm": 6.067083358764648,
      "learning_rate": 3.256499315861488e-05,
      "loss": 0.4817,
      "step": 26212
    },
    {
      "epoch": 8.276918219134828,
      "grad_norm": 8.469970703125,
      "learning_rate": 3.256411605795882e-05,
      "loss": 0.4045,
      "step": 26213
    },
    {
      "epoch": 8.277233975371013,
      "grad_norm": 8.286454200744629,
      "learning_rate": 3.256323895730274e-05,
      "loss": 0.4718,
      "step": 26214
    },
    {
      "epoch": 8.277549731607198,
      "grad_norm": 6.06212043762207,
      "learning_rate": 3.2562361856646675e-05,
      "loss": 0.3917,
      "step": 26215
    },
    {
      "epoch": 8.277865487843385,
      "grad_norm": 5.743088245391846,
      "learning_rate": 3.25614847559906e-05,
      "loss": 0.4831,
      "step": 26216
    },
    {
      "epoch": 8.27818124407957,
      "grad_norm": 5.511542320251465,
      "learning_rate": 3.256060765533453e-05,
      "loss": 0.422,
      "step": 26217
    },
    {
      "epoch": 8.278497000315756,
      "grad_norm": 5.490318298339844,
      "learning_rate": 3.2559730554678456e-05,
      "loss": 0.3457,
      "step": 26218
    },
    {
      "epoch": 8.278812756551941,
      "grad_norm": 6.828299522399902,
      "learning_rate": 3.255885345402239e-05,
      "loss": 0.421,
      "step": 26219
    },
    {
      "epoch": 8.279128512788127,
      "grad_norm": 5.387667655944824,
      "learning_rate": 3.255797635336631e-05,
      "loss": 0.4905,
      "step": 26220
    },
    {
      "epoch": 8.279444269024314,
      "grad_norm": 7.656128883361816,
      "learning_rate": 3.2557099252710245e-05,
      "loss": 0.3617,
      "step": 26221
    },
    {
      "epoch": 8.279760025260499,
      "grad_norm": 5.192109107971191,
      "learning_rate": 3.255622215205417e-05,
      "loss": 0.3256,
      "step": 26222
    },
    {
      "epoch": 8.280075781496684,
      "grad_norm": 8.225024223327637,
      "learning_rate": 3.25553450513981e-05,
      "loss": 0.4247,
      "step": 26223
    },
    {
      "epoch": 8.28039153773287,
      "grad_norm": 7.017269611358643,
      "learning_rate": 3.2554467950742026e-05,
      "loss": 0.4715,
      "step": 26224
    },
    {
      "epoch": 8.280707293969057,
      "grad_norm": 5.335146903991699,
      "learning_rate": 3.255359085008596e-05,
      "loss": 0.3598,
      "step": 26225
    },
    {
      "epoch": 8.281023050205242,
      "grad_norm": 6.056997776031494,
      "learning_rate": 3.255271374942989e-05,
      "loss": 0.4543,
      "step": 26226
    },
    {
      "epoch": 8.281338806441427,
      "grad_norm": 7.224231719970703,
      "learning_rate": 3.2551836648773814e-05,
      "loss": 0.4208,
      "step": 26227
    },
    {
      "epoch": 8.281654562677613,
      "grad_norm": 5.596956729888916,
      "learning_rate": 3.2550959548117746e-05,
      "loss": 0.4263,
      "step": 26228
    },
    {
      "epoch": 8.281970318913798,
      "grad_norm": 5.29574728012085,
      "learning_rate": 3.255008244746167e-05,
      "loss": 0.3597,
      "step": 26229
    },
    {
      "epoch": 8.282286075149985,
      "grad_norm": 5.990054607391357,
      "learning_rate": 3.25492053468056e-05,
      "loss": 0.4089,
      "step": 26230
    },
    {
      "epoch": 8.28260183138617,
      "grad_norm": 5.648138999938965,
      "learning_rate": 3.254832824614953e-05,
      "loss": 0.3957,
      "step": 26231
    },
    {
      "epoch": 8.282917587622356,
      "grad_norm": 5.3705549240112305,
      "learning_rate": 3.254745114549346e-05,
      "loss": 0.4362,
      "step": 26232
    },
    {
      "epoch": 8.28323334385854,
      "grad_norm": 5.4879255294799805,
      "learning_rate": 3.2546574044837384e-05,
      "loss": 0.5313,
      "step": 26233
    },
    {
      "epoch": 8.283549100094726,
      "grad_norm": 6.445348739624023,
      "learning_rate": 3.2545696944181316e-05,
      "loss": 0.4898,
      "step": 26234
    },
    {
      "epoch": 8.283864856330913,
      "grad_norm": 7.185102939605713,
      "learning_rate": 3.254481984352524e-05,
      "loss": 0.4483,
      "step": 26235
    },
    {
      "epoch": 8.284180612567098,
      "grad_norm": 8.625907897949219,
      "learning_rate": 3.254394274286917e-05,
      "loss": 0.5078,
      "step": 26236
    },
    {
      "epoch": 8.284496368803284,
      "grad_norm": 6.905599117279053,
      "learning_rate": 3.25430656422131e-05,
      "loss": 0.3364,
      "step": 26237
    },
    {
      "epoch": 8.284812125039469,
      "grad_norm": 5.465630531311035,
      "learning_rate": 3.254218854155703e-05,
      "loss": 0.3238,
      "step": 26238
    },
    {
      "epoch": 8.285127881275654,
      "grad_norm": 7.0126214027404785,
      "learning_rate": 3.2541311440900954e-05,
      "loss": 0.406,
      "step": 26239
    },
    {
      "epoch": 8.285443637511841,
      "grad_norm": 6.609295845031738,
      "learning_rate": 3.254043434024489e-05,
      "loss": 0.411,
      "step": 26240
    },
    {
      "epoch": 8.285759393748027,
      "grad_norm": 7.026000022888184,
      "learning_rate": 3.253955723958882e-05,
      "loss": 0.3441,
      "step": 26241
    },
    {
      "epoch": 8.286075149984212,
      "grad_norm": 5.792336940765381,
      "learning_rate": 3.253868013893275e-05,
      "loss": 0.4098,
      "step": 26242
    },
    {
      "epoch": 8.286390906220397,
      "grad_norm": 5.741052150726318,
      "learning_rate": 3.2537803038276674e-05,
      "loss": 0.3756,
      "step": 26243
    },
    {
      "epoch": 8.286706662456584,
      "grad_norm": 6.12492036819458,
      "learning_rate": 3.2536925937620605e-05,
      "loss": 0.4697,
      "step": 26244
    },
    {
      "epoch": 8.28702241869277,
      "grad_norm": 5.91534423828125,
      "learning_rate": 3.253604883696453e-05,
      "loss": 0.3762,
      "step": 26245
    },
    {
      "epoch": 8.287338174928955,
      "grad_norm": 5.256624698638916,
      "learning_rate": 3.253517173630846e-05,
      "loss": 0.3599,
      "step": 26246
    },
    {
      "epoch": 8.28765393116514,
      "grad_norm": 4.707192897796631,
      "learning_rate": 3.253429463565239e-05,
      "loss": 0.4297,
      "step": 26247
    },
    {
      "epoch": 8.287969687401326,
      "grad_norm": 6.043311595916748,
      "learning_rate": 3.253341753499632e-05,
      "loss": 0.4265,
      "step": 26248
    },
    {
      "epoch": 8.288285443637513,
      "grad_norm": 6.099456310272217,
      "learning_rate": 3.253254043434025e-05,
      "loss": 0.4383,
      "step": 26249
    },
    {
      "epoch": 8.288601199873698,
      "grad_norm": 6.619207382202148,
      "learning_rate": 3.2531663333684175e-05,
      "loss": 0.4476,
      "step": 26250
    },
    {
      "epoch": 8.288916956109883,
      "grad_norm": 6.220597267150879,
      "learning_rate": 3.253078623302811e-05,
      "loss": 0.4093,
      "step": 26251
    },
    {
      "epoch": 8.289232712346069,
      "grad_norm": 5.849869251251221,
      "learning_rate": 3.252990913237203e-05,
      "loss": 0.3758,
      "step": 26252
    },
    {
      "epoch": 8.289548468582254,
      "grad_norm": 5.208134174346924,
      "learning_rate": 3.2529032031715964e-05,
      "loss": 0.3709,
      "step": 26253
    },
    {
      "epoch": 8.28986422481844,
      "grad_norm": 6.47114896774292,
      "learning_rate": 3.252815493105989e-05,
      "loss": 0.4077,
      "step": 26254
    },
    {
      "epoch": 8.290179981054626,
      "grad_norm": 4.991421222686768,
      "learning_rate": 3.252727783040382e-05,
      "loss": 0.3404,
      "step": 26255
    },
    {
      "epoch": 8.290495737290811,
      "grad_norm": 6.455019950866699,
      "learning_rate": 3.2526400729747745e-05,
      "loss": 0.4598,
      "step": 26256
    },
    {
      "epoch": 8.290811493526997,
      "grad_norm": 8.181947708129883,
      "learning_rate": 3.252552362909168e-05,
      "loss": 0.3971,
      "step": 26257
    },
    {
      "epoch": 8.291127249763182,
      "grad_norm": 6.091250896453857,
      "learning_rate": 3.25246465284356e-05,
      "loss": 0.4723,
      "step": 26258
    },
    {
      "epoch": 8.29144300599937,
      "grad_norm": 5.90443754196167,
      "learning_rate": 3.252376942777953e-05,
      "loss": 0.3766,
      "step": 26259
    },
    {
      "epoch": 8.291758762235554,
      "grad_norm": 6.289364814758301,
      "learning_rate": 3.252289232712346e-05,
      "loss": 0.4803,
      "step": 26260
    },
    {
      "epoch": 8.29207451847174,
      "grad_norm": 6.463775157928467,
      "learning_rate": 3.252201522646739e-05,
      "loss": 0.3914,
      "step": 26261
    },
    {
      "epoch": 8.292390274707925,
      "grad_norm": 6.01702356338501,
      "learning_rate": 3.2521138125811315e-05,
      "loss": 0.3599,
      "step": 26262
    },
    {
      "epoch": 8.29270603094411,
      "grad_norm": 7.906797885894775,
      "learning_rate": 3.2520261025155246e-05,
      "loss": 0.3814,
      "step": 26263
    },
    {
      "epoch": 8.293021787180297,
      "grad_norm": 6.563748359680176,
      "learning_rate": 3.251938392449918e-05,
      "loss": 0.3845,
      "step": 26264
    },
    {
      "epoch": 8.293337543416483,
      "grad_norm": 5.393807411193848,
      "learning_rate": 3.25185068238431e-05,
      "loss": 0.4463,
      "step": 26265
    },
    {
      "epoch": 8.293653299652668,
      "grad_norm": 5.190562725067139,
      "learning_rate": 3.2517629723187035e-05,
      "loss": 0.4386,
      "step": 26266
    },
    {
      "epoch": 8.293969055888853,
      "grad_norm": 4.5614213943481445,
      "learning_rate": 3.251675262253096e-05,
      "loss": 0.3653,
      "step": 26267
    },
    {
      "epoch": 8.29428481212504,
      "grad_norm": 5.425395488739014,
      "learning_rate": 3.251587552187489e-05,
      "loss": 0.4139,
      "step": 26268
    },
    {
      "epoch": 8.294600568361226,
      "grad_norm": 6.099277973175049,
      "learning_rate": 3.251499842121882e-05,
      "loss": 0.3741,
      "step": 26269
    },
    {
      "epoch": 8.294916324597411,
      "grad_norm": 6.011996746063232,
      "learning_rate": 3.2514121320562755e-05,
      "loss": 0.4146,
      "step": 26270
    },
    {
      "epoch": 8.295232080833596,
      "grad_norm": 5.333427429199219,
      "learning_rate": 3.251324421990668e-05,
      "loss": 0.4177,
      "step": 26271
    },
    {
      "epoch": 8.295547837069781,
      "grad_norm": 4.486510276794434,
      "learning_rate": 3.251236711925061e-05,
      "loss": 0.4302,
      "step": 26272
    },
    {
      "epoch": 8.295863593305969,
      "grad_norm": 8.951364517211914,
      "learning_rate": 3.2511490018594536e-05,
      "loss": 0.4203,
      "step": 26273
    },
    {
      "epoch": 8.296179349542154,
      "grad_norm": 5.615066051483154,
      "learning_rate": 3.251061291793847e-05,
      "loss": 0.3667,
      "step": 26274
    },
    {
      "epoch": 8.29649510577834,
      "grad_norm": 6.329992294311523,
      "learning_rate": 3.250973581728239e-05,
      "loss": 0.3808,
      "step": 26275
    },
    {
      "epoch": 8.296810862014524,
      "grad_norm": 5.929025650024414,
      "learning_rate": 3.2508858716626324e-05,
      "loss": 0.4032,
      "step": 26276
    },
    {
      "epoch": 8.29712661825071,
      "grad_norm": 5.932660102844238,
      "learning_rate": 3.250798161597025e-05,
      "loss": 0.4659,
      "step": 26277
    },
    {
      "epoch": 8.297442374486897,
      "grad_norm": 5.041687965393066,
      "learning_rate": 3.250710451531418e-05,
      "loss": 0.3476,
      "step": 26278
    },
    {
      "epoch": 8.297758130723082,
      "grad_norm": 3.9739747047424316,
      "learning_rate": 3.2506227414658106e-05,
      "loss": 0.3061,
      "step": 26279
    },
    {
      "epoch": 8.298073886959267,
      "grad_norm": 4.772372722625732,
      "learning_rate": 3.250535031400204e-05,
      "loss": 0.4743,
      "step": 26280
    },
    {
      "epoch": 8.298389643195453,
      "grad_norm": 6.384881019592285,
      "learning_rate": 3.250447321334596e-05,
      "loss": 0.3845,
      "step": 26281
    },
    {
      "epoch": 8.298705399431638,
      "grad_norm": 7.368159770965576,
      "learning_rate": 3.2503596112689894e-05,
      "loss": 0.427,
      "step": 26282
    },
    {
      "epoch": 8.299021155667825,
      "grad_norm": 4.5824785232543945,
      "learning_rate": 3.250271901203382e-05,
      "loss": 0.3485,
      "step": 26283
    },
    {
      "epoch": 8.29933691190401,
      "grad_norm": 5.033443450927734,
      "learning_rate": 3.250184191137775e-05,
      "loss": 0.4236,
      "step": 26284
    },
    {
      "epoch": 8.299652668140196,
      "grad_norm": 7.937896728515625,
      "learning_rate": 3.250096481072168e-05,
      "loss": 0.4345,
      "step": 26285
    },
    {
      "epoch": 8.299968424376381,
      "grad_norm": 4.002184867858887,
      "learning_rate": 3.250008771006561e-05,
      "loss": 0.3992,
      "step": 26286
    },
    {
      "epoch": 8.300284180612568,
      "grad_norm": 6.197635173797607,
      "learning_rate": 3.249921060940954e-05,
      "loss": 0.408,
      "step": 26287
    },
    {
      "epoch": 8.300599936848753,
      "grad_norm": 6.284440994262695,
      "learning_rate": 3.2498333508753464e-05,
      "loss": 0.4418,
      "step": 26288
    },
    {
      "epoch": 8.300915693084939,
      "grad_norm": 5.567397594451904,
      "learning_rate": 3.2497456408097396e-05,
      "loss": 0.4181,
      "step": 26289
    },
    {
      "epoch": 8.301231449321124,
      "grad_norm": 6.087873458862305,
      "learning_rate": 3.249657930744132e-05,
      "loss": 0.4043,
      "step": 26290
    },
    {
      "epoch": 8.30154720555731,
      "grad_norm": 6.067837715148926,
      "learning_rate": 3.249570220678525e-05,
      "loss": 0.4017,
      "step": 26291
    },
    {
      "epoch": 8.301862961793496,
      "grad_norm": 5.584583759307861,
      "learning_rate": 3.249482510612918e-05,
      "loss": 0.3355,
      "step": 26292
    },
    {
      "epoch": 8.302178718029682,
      "grad_norm": 7.668466567993164,
      "learning_rate": 3.249394800547311e-05,
      "loss": 0.4646,
      "step": 26293
    },
    {
      "epoch": 8.302494474265867,
      "grad_norm": 8.293437004089355,
      "learning_rate": 3.2493070904817034e-05,
      "loss": 0.367,
      "step": 26294
    },
    {
      "epoch": 8.302810230502052,
      "grad_norm": 6.70164680480957,
      "learning_rate": 3.249219380416097e-05,
      "loss": 0.4223,
      "step": 26295
    },
    {
      "epoch": 8.303125986738237,
      "grad_norm": 5.939335823059082,
      "learning_rate": 3.24913167035049e-05,
      "loss": 0.4464,
      "step": 26296
    },
    {
      "epoch": 8.303441742974424,
      "grad_norm": 6.745326519012451,
      "learning_rate": 3.249043960284883e-05,
      "loss": 0.4478,
      "step": 26297
    },
    {
      "epoch": 8.30375749921061,
      "grad_norm": 5.898866653442383,
      "learning_rate": 3.2489562502192754e-05,
      "loss": 0.358,
      "step": 26298
    },
    {
      "epoch": 8.304073255446795,
      "grad_norm": 9.899209022521973,
      "learning_rate": 3.2488685401536685e-05,
      "loss": 0.433,
      "step": 26299
    },
    {
      "epoch": 8.30438901168298,
      "grad_norm": 5.998438835144043,
      "learning_rate": 3.248780830088061e-05,
      "loss": 0.3616,
      "step": 26300
    },
    {
      "epoch": 8.304704767919166,
      "grad_norm": 8.148626327514648,
      "learning_rate": 3.248693120022454e-05,
      "loss": 0.3905,
      "step": 26301
    },
    {
      "epoch": 8.305020524155353,
      "grad_norm": 7.878254413604736,
      "learning_rate": 3.248605409956847e-05,
      "loss": 0.3517,
      "step": 26302
    },
    {
      "epoch": 8.305336280391538,
      "grad_norm": 6.957207679748535,
      "learning_rate": 3.24851769989124e-05,
      "loss": 0.5001,
      "step": 26303
    },
    {
      "epoch": 8.305652036627723,
      "grad_norm": 6.003876209259033,
      "learning_rate": 3.2484299898256323e-05,
      "loss": 0.4483,
      "step": 26304
    },
    {
      "epoch": 8.305967792863909,
      "grad_norm": 7.048958778381348,
      "learning_rate": 3.2483422797600255e-05,
      "loss": 0.4304,
      "step": 26305
    },
    {
      "epoch": 8.306283549100094,
      "grad_norm": 7.224407196044922,
      "learning_rate": 3.248254569694418e-05,
      "loss": 0.4673,
      "step": 26306
    },
    {
      "epoch": 8.306599305336281,
      "grad_norm": 6.414976119995117,
      "learning_rate": 3.248166859628811e-05,
      "loss": 0.3854,
      "step": 26307
    },
    {
      "epoch": 8.306915061572466,
      "grad_norm": 5.93223237991333,
      "learning_rate": 3.2480791495632043e-05,
      "loss": 0.3916,
      "step": 26308
    },
    {
      "epoch": 8.307230817808652,
      "grad_norm": 5.410266399383545,
      "learning_rate": 3.247991439497597e-05,
      "loss": 0.3543,
      "step": 26309
    },
    {
      "epoch": 8.307546574044837,
      "grad_norm": 6.821674346923828,
      "learning_rate": 3.24790372943199e-05,
      "loss": 0.4032,
      "step": 26310
    },
    {
      "epoch": 8.307862330281022,
      "grad_norm": 6.477256774902344,
      "learning_rate": 3.2478160193663825e-05,
      "loss": 0.4089,
      "step": 26311
    },
    {
      "epoch": 8.30817808651721,
      "grad_norm": 7.16783332824707,
      "learning_rate": 3.2477283093007757e-05,
      "loss": 0.3722,
      "step": 26312
    },
    {
      "epoch": 8.308493842753395,
      "grad_norm": 8.057267189025879,
      "learning_rate": 3.247640599235168e-05,
      "loss": 0.4638,
      "step": 26313
    },
    {
      "epoch": 8.30880959898958,
      "grad_norm": 17.951492309570312,
      "learning_rate": 3.247552889169561e-05,
      "loss": 0.4532,
      "step": 26314
    },
    {
      "epoch": 8.309125355225765,
      "grad_norm": 6.114940643310547,
      "learning_rate": 3.247465179103954e-05,
      "loss": 0.3999,
      "step": 26315
    },
    {
      "epoch": 8.309441111461952,
      "grad_norm": 6.749838352203369,
      "learning_rate": 3.247377469038347e-05,
      "loss": 0.413,
      "step": 26316
    },
    {
      "epoch": 8.309756867698137,
      "grad_norm": 5.552931308746338,
      "learning_rate": 3.2472897589727395e-05,
      "loss": 0.4708,
      "step": 26317
    },
    {
      "epoch": 8.310072623934323,
      "grad_norm": 5.871334075927734,
      "learning_rate": 3.2472020489071326e-05,
      "loss": 0.4237,
      "step": 26318
    },
    {
      "epoch": 8.310388380170508,
      "grad_norm": 6.444705009460449,
      "learning_rate": 3.247114338841525e-05,
      "loss": 0.4919,
      "step": 26319
    },
    {
      "epoch": 8.310704136406693,
      "grad_norm": 6.735838890075684,
      "learning_rate": 3.247026628775918e-05,
      "loss": 0.4717,
      "step": 26320
    },
    {
      "epoch": 8.31101989264288,
      "grad_norm": 8.532402992248535,
      "learning_rate": 3.246938918710311e-05,
      "loss": 0.4614,
      "step": 26321
    },
    {
      "epoch": 8.311335648879066,
      "grad_norm": 5.122908592224121,
      "learning_rate": 3.2468512086447046e-05,
      "loss": 0.4248,
      "step": 26322
    },
    {
      "epoch": 8.311651405115251,
      "grad_norm": 6.054628372192383,
      "learning_rate": 3.246763498579097e-05,
      "loss": 0.337,
      "step": 26323
    },
    {
      "epoch": 8.311967161351436,
      "grad_norm": 9.38193416595459,
      "learning_rate": 3.24667578851349e-05,
      "loss": 0.4057,
      "step": 26324
    },
    {
      "epoch": 8.312282917587622,
      "grad_norm": 6.3078436851501465,
      "learning_rate": 3.246588078447883e-05,
      "loss": 0.461,
      "step": 26325
    },
    {
      "epoch": 8.312598673823809,
      "grad_norm": 6.620100021362305,
      "learning_rate": 3.246500368382276e-05,
      "loss": 0.462,
      "step": 26326
    },
    {
      "epoch": 8.312914430059994,
      "grad_norm": 6.530341148376465,
      "learning_rate": 3.2464126583166684e-05,
      "loss": 0.3614,
      "step": 26327
    },
    {
      "epoch": 8.31323018629618,
      "grad_norm": 5.346355438232422,
      "learning_rate": 3.2463249482510616e-05,
      "loss": 0.3646,
      "step": 26328
    },
    {
      "epoch": 8.313545942532365,
      "grad_norm": 6.911806106567383,
      "learning_rate": 3.246237238185455e-05,
      "loss": 0.4194,
      "step": 26329
    },
    {
      "epoch": 8.31386169876855,
      "grad_norm": 6.441895961761475,
      "learning_rate": 3.246149528119847e-05,
      "loss": 0.3246,
      "step": 26330
    },
    {
      "epoch": 8.314177455004737,
      "grad_norm": 7.101787567138672,
      "learning_rate": 3.2460618180542404e-05,
      "loss": 0.4751,
      "step": 26331
    },
    {
      "epoch": 8.314493211240922,
      "grad_norm": 6.475310325622559,
      "learning_rate": 3.245974107988633e-05,
      "loss": 0.4562,
      "step": 26332
    },
    {
      "epoch": 8.314808967477108,
      "grad_norm": 7.735019683837891,
      "learning_rate": 3.245886397923026e-05,
      "loss": 0.4493,
      "step": 26333
    },
    {
      "epoch": 8.315124723713293,
      "grad_norm": 5.789551258087158,
      "learning_rate": 3.2457986878574186e-05,
      "loss": 0.4421,
      "step": 26334
    },
    {
      "epoch": 8.31544047994948,
      "grad_norm": 5.602222442626953,
      "learning_rate": 3.245710977791812e-05,
      "loss": 0.4423,
      "step": 26335
    },
    {
      "epoch": 8.315756236185665,
      "grad_norm": 5.320356845855713,
      "learning_rate": 3.245623267726204e-05,
      "loss": 0.3593,
      "step": 26336
    },
    {
      "epoch": 8.31607199242185,
      "grad_norm": 7.294353485107422,
      "learning_rate": 3.2455355576605974e-05,
      "loss": 0.3231,
      "step": 26337
    },
    {
      "epoch": 8.316387748658036,
      "grad_norm": 6.373568058013916,
      "learning_rate": 3.24544784759499e-05,
      "loss": 0.4396,
      "step": 26338
    },
    {
      "epoch": 8.316703504894221,
      "grad_norm": 6.9785332679748535,
      "learning_rate": 3.245360137529383e-05,
      "loss": 0.5114,
      "step": 26339
    },
    {
      "epoch": 8.317019261130408,
      "grad_norm": 4.56117057800293,
      "learning_rate": 3.2452724274637756e-05,
      "loss": 0.354,
      "step": 26340
    },
    {
      "epoch": 8.317335017366593,
      "grad_norm": 4.503605365753174,
      "learning_rate": 3.245184717398169e-05,
      "loss": 0.3367,
      "step": 26341
    },
    {
      "epoch": 8.317650773602779,
      "grad_norm": 7.469731330871582,
      "learning_rate": 3.245097007332561e-05,
      "loss": 0.3797,
      "step": 26342
    },
    {
      "epoch": 8.317966529838964,
      "grad_norm": 7.143353462219238,
      "learning_rate": 3.2450092972669544e-05,
      "loss": 0.4817,
      "step": 26343
    },
    {
      "epoch": 8.31828228607515,
      "grad_norm": 5.121073246002197,
      "learning_rate": 3.244921587201347e-05,
      "loss": 0.4298,
      "step": 26344
    },
    {
      "epoch": 8.318598042311336,
      "grad_norm": 5.819303512573242,
      "learning_rate": 3.24483387713574e-05,
      "loss": 0.4655,
      "step": 26345
    },
    {
      "epoch": 8.318913798547522,
      "grad_norm": 9.502305030822754,
      "learning_rate": 3.244746167070133e-05,
      "loss": 0.365,
      "step": 26346
    },
    {
      "epoch": 8.319229554783707,
      "grad_norm": 6.149625301361084,
      "learning_rate": 3.244658457004526e-05,
      "loss": 0.3783,
      "step": 26347
    },
    {
      "epoch": 8.319545311019892,
      "grad_norm": 7.31165885925293,
      "learning_rate": 3.244570746938919e-05,
      "loss": 0.5262,
      "step": 26348
    },
    {
      "epoch": 8.319861067256078,
      "grad_norm": 5.586194038391113,
      "learning_rate": 3.244483036873312e-05,
      "loss": 0.4278,
      "step": 26349
    },
    {
      "epoch": 8.320176823492265,
      "grad_norm": 6.300291538238525,
      "learning_rate": 3.2443953268077045e-05,
      "loss": 0.3145,
      "step": 26350
    },
    {
      "epoch": 8.32049257972845,
      "grad_norm": 6.670507431030273,
      "learning_rate": 3.244307616742098e-05,
      "loss": 0.4852,
      "step": 26351
    },
    {
      "epoch": 8.320808335964635,
      "grad_norm": 5.339749336242676,
      "learning_rate": 3.244219906676491e-05,
      "loss": 0.3098,
      "step": 26352
    },
    {
      "epoch": 8.32112409220082,
      "grad_norm": 7.069468975067139,
      "learning_rate": 3.2441321966108834e-05,
      "loss": 0.4926,
      "step": 26353
    },
    {
      "epoch": 8.321439848437006,
      "grad_norm": 5.709237098693848,
      "learning_rate": 3.2440444865452765e-05,
      "loss": 0.4611,
      "step": 26354
    },
    {
      "epoch": 8.321755604673193,
      "grad_norm": 7.983531951904297,
      "learning_rate": 3.243956776479669e-05,
      "loss": 0.4086,
      "step": 26355
    },
    {
      "epoch": 8.322071360909378,
      "grad_norm": 5.8586225509643555,
      "learning_rate": 3.243869066414062e-05,
      "loss": 0.4371,
      "step": 26356
    },
    {
      "epoch": 8.322387117145563,
      "grad_norm": 5.1370391845703125,
      "learning_rate": 3.243781356348455e-05,
      "loss": 0.3547,
      "step": 26357
    },
    {
      "epoch": 8.322702873381749,
      "grad_norm": 5.698273181915283,
      "learning_rate": 3.243693646282848e-05,
      "loss": 0.4529,
      "step": 26358
    },
    {
      "epoch": 8.323018629617934,
      "grad_norm": 6.058475494384766,
      "learning_rate": 3.24360593621724e-05,
      "loss": 0.3722,
      "step": 26359
    },
    {
      "epoch": 8.323334385854121,
      "grad_norm": 5.137777328491211,
      "learning_rate": 3.2435182261516335e-05,
      "loss": 0.3356,
      "step": 26360
    },
    {
      "epoch": 8.323650142090306,
      "grad_norm": 5.379976272583008,
      "learning_rate": 3.243430516086026e-05,
      "loss": 0.3904,
      "step": 26361
    },
    {
      "epoch": 8.323965898326492,
      "grad_norm": 8.453716278076172,
      "learning_rate": 3.243342806020419e-05,
      "loss": 0.3807,
      "step": 26362
    },
    {
      "epoch": 8.324281654562677,
      "grad_norm": 6.558168888092041,
      "learning_rate": 3.2432550959548117e-05,
      "loss": 0.4345,
      "step": 26363
    },
    {
      "epoch": 8.324597410798864,
      "grad_norm": 8.375739097595215,
      "learning_rate": 3.243167385889205e-05,
      "loss": 0.4987,
      "step": 26364
    },
    {
      "epoch": 8.32491316703505,
      "grad_norm": 5.091122627258301,
      "learning_rate": 3.243079675823597e-05,
      "loss": 0.4179,
      "step": 26365
    },
    {
      "epoch": 8.325228923271235,
      "grad_norm": 6.937322616577148,
      "learning_rate": 3.2429919657579905e-05,
      "loss": 0.455,
      "step": 26366
    },
    {
      "epoch": 8.32554467950742,
      "grad_norm": 5.231779098510742,
      "learning_rate": 3.2429042556923836e-05,
      "loss": 0.4458,
      "step": 26367
    },
    {
      "epoch": 8.325860435743605,
      "grad_norm": 4.677241325378418,
      "learning_rate": 3.242816545626776e-05,
      "loss": 0.3898,
      "step": 26368
    },
    {
      "epoch": 8.326176191979792,
      "grad_norm": 5.69813346862793,
      "learning_rate": 3.242728835561169e-05,
      "loss": 0.3604,
      "step": 26369
    },
    {
      "epoch": 8.326491948215978,
      "grad_norm": 5.431032180786133,
      "learning_rate": 3.242641125495562e-05,
      "loss": 0.4165,
      "step": 26370
    },
    {
      "epoch": 8.326807704452163,
      "grad_norm": 6.412964820861816,
      "learning_rate": 3.242553415429955e-05,
      "loss": 0.4464,
      "step": 26371
    },
    {
      "epoch": 8.327123460688348,
      "grad_norm": 4.548511981964111,
      "learning_rate": 3.2424657053643475e-05,
      "loss": 0.4816,
      "step": 26372
    },
    {
      "epoch": 8.327439216924533,
      "grad_norm": 5.72243070602417,
      "learning_rate": 3.2423779952987406e-05,
      "loss": 0.3747,
      "step": 26373
    },
    {
      "epoch": 8.32775497316072,
      "grad_norm": 6.021235942840576,
      "learning_rate": 3.242290285233133e-05,
      "loss": 0.5721,
      "step": 26374
    },
    {
      "epoch": 8.328070729396906,
      "grad_norm": 6.443791389465332,
      "learning_rate": 3.242202575167526e-05,
      "loss": 0.3938,
      "step": 26375
    },
    {
      "epoch": 8.328386485633091,
      "grad_norm": 6.917316436767578,
      "learning_rate": 3.242114865101919e-05,
      "loss": 0.4223,
      "step": 26376
    },
    {
      "epoch": 8.328702241869276,
      "grad_norm": 5.050969123840332,
      "learning_rate": 3.2420271550363126e-05,
      "loss": 0.3618,
      "step": 26377
    },
    {
      "epoch": 8.329017998105462,
      "grad_norm": 6.7018232345581055,
      "learning_rate": 3.241939444970705e-05,
      "loss": 0.4415,
      "step": 26378
    },
    {
      "epoch": 8.329333754341649,
      "grad_norm": 6.588116645812988,
      "learning_rate": 3.241851734905098e-05,
      "loss": 0.4705,
      "step": 26379
    },
    {
      "epoch": 8.329649510577834,
      "grad_norm": 7.8248491287231445,
      "learning_rate": 3.241764024839491e-05,
      "loss": 0.5753,
      "step": 26380
    },
    {
      "epoch": 8.32996526681402,
      "grad_norm": 5.303781032562256,
      "learning_rate": 3.241676314773884e-05,
      "loss": 0.3953,
      "step": 26381
    },
    {
      "epoch": 8.330281023050205,
      "grad_norm": 7.9980669021606445,
      "learning_rate": 3.2415886047082764e-05,
      "loss": 0.3997,
      "step": 26382
    },
    {
      "epoch": 8.330596779286392,
      "grad_norm": 5.018172264099121,
      "learning_rate": 3.2415008946426696e-05,
      "loss": 0.3517,
      "step": 26383
    },
    {
      "epoch": 8.330912535522577,
      "grad_norm": 5.290098667144775,
      "learning_rate": 3.241413184577062e-05,
      "loss": 0.3316,
      "step": 26384
    },
    {
      "epoch": 8.331228291758762,
      "grad_norm": 6.093502998352051,
      "learning_rate": 3.241325474511455e-05,
      "loss": 0.4048,
      "step": 26385
    },
    {
      "epoch": 8.331544047994948,
      "grad_norm": 6.614793300628662,
      "learning_rate": 3.241237764445848e-05,
      "loss": 0.3498,
      "step": 26386
    },
    {
      "epoch": 8.331859804231133,
      "grad_norm": 5.713740825653076,
      "learning_rate": 3.241150054380241e-05,
      "loss": 0.4777,
      "step": 26387
    },
    {
      "epoch": 8.33217556046732,
      "grad_norm": 5.645822048187256,
      "learning_rate": 3.2410623443146334e-05,
      "loss": 0.4092,
      "step": 26388
    },
    {
      "epoch": 8.332491316703505,
      "grad_norm": 5.414953231811523,
      "learning_rate": 3.2409746342490266e-05,
      "loss": 0.4056,
      "step": 26389
    },
    {
      "epoch": 8.33280707293969,
      "grad_norm": 5.843465805053711,
      "learning_rate": 3.24088692418342e-05,
      "loss": 0.3419,
      "step": 26390
    },
    {
      "epoch": 8.333122829175876,
      "grad_norm": 5.377943992614746,
      "learning_rate": 3.240799214117812e-05,
      "loss": 0.3866,
      "step": 26391
    },
    {
      "epoch": 8.333438585412061,
      "grad_norm": 6.5560503005981445,
      "learning_rate": 3.2407115040522054e-05,
      "loss": 0.4866,
      "step": 26392
    },
    {
      "epoch": 8.333754341648248,
      "grad_norm": 7.399163722991943,
      "learning_rate": 3.240623793986598e-05,
      "loss": 0.4428,
      "step": 26393
    },
    {
      "epoch": 8.334070097884434,
      "grad_norm": 7.8384222984313965,
      "learning_rate": 3.240536083920991e-05,
      "loss": 0.4621,
      "step": 26394
    },
    {
      "epoch": 8.334385854120619,
      "grad_norm": 5.048452377319336,
      "learning_rate": 3.2404483738553835e-05,
      "loss": 0.3358,
      "step": 26395
    },
    {
      "epoch": 8.334701610356804,
      "grad_norm": 7.089169502258301,
      "learning_rate": 3.240360663789777e-05,
      "loss": 0.4046,
      "step": 26396
    },
    {
      "epoch": 8.33501736659299,
      "grad_norm": 5.805610179901123,
      "learning_rate": 3.240272953724169e-05,
      "loss": 0.4371,
      "step": 26397
    },
    {
      "epoch": 8.335333122829176,
      "grad_norm": 4.930132865905762,
      "learning_rate": 3.2401852436585624e-05,
      "loss": 0.4015,
      "step": 26398
    },
    {
      "epoch": 8.335648879065362,
      "grad_norm": 4.998363494873047,
      "learning_rate": 3.240097533592955e-05,
      "loss": 0.4229,
      "step": 26399
    },
    {
      "epoch": 8.335964635301547,
      "grad_norm": 5.082121849060059,
      "learning_rate": 3.240009823527348e-05,
      "loss": 0.4485,
      "step": 26400
    },
    {
      "epoch": 8.335964635301547,
      "eval_loss": 0.9460871815681458,
      "eval_runtime": 4.6402,
      "eval_samples_per_second": 536.618,
      "eval_steps_per_second": 4.526,
      "step": 26400
    },
    {
      "epoch": 8.336280391537732,
      "grad_norm": 5.562905788421631,
      "learning_rate": 3.2399221134617405e-05,
      "loss": 0.4838,
      "step": 26401
    },
    {
      "epoch": 8.336596147773918,
      "grad_norm": 4.7983527183532715,
      "learning_rate": 3.239834403396134e-05,
      "loss": 0.3598,
      "step": 26402
    },
    {
      "epoch": 8.336911904010105,
      "grad_norm": 5.432160377502441,
      "learning_rate": 3.239746693330526e-05,
      "loss": 0.4029,
      "step": 26403
    },
    {
      "epoch": 8.33722766024629,
      "grad_norm": 4.655757904052734,
      "learning_rate": 3.23965898326492e-05,
      "loss": 0.337,
      "step": 26404
    },
    {
      "epoch": 8.337543416482475,
      "grad_norm": 6.364091873168945,
      "learning_rate": 3.2395712731993125e-05,
      "loss": 0.4227,
      "step": 26405
    },
    {
      "epoch": 8.33785917271866,
      "grad_norm": 5.590327262878418,
      "learning_rate": 3.239483563133706e-05,
      "loss": 0.5087,
      "step": 26406
    },
    {
      "epoch": 8.338174928954848,
      "grad_norm": 5.2392578125,
      "learning_rate": 3.239395853068098e-05,
      "loss": 0.4762,
      "step": 26407
    },
    {
      "epoch": 8.338490685191033,
      "grad_norm": 5.44436502456665,
      "learning_rate": 3.2393081430024913e-05,
      "loss": 0.4513,
      "step": 26408
    },
    {
      "epoch": 8.338806441427218,
      "grad_norm": 5.652056694030762,
      "learning_rate": 3.239220432936884e-05,
      "loss": 0.3798,
      "step": 26409
    },
    {
      "epoch": 8.339122197663404,
      "grad_norm": 5.421526908874512,
      "learning_rate": 3.239132722871277e-05,
      "loss": 0.3778,
      "step": 26410
    },
    {
      "epoch": 8.339437953899589,
      "grad_norm": 5.6162519454956055,
      "learning_rate": 3.23904501280567e-05,
      "loss": 0.4347,
      "step": 26411
    },
    {
      "epoch": 8.339753710135776,
      "grad_norm": 6.273671627044678,
      "learning_rate": 3.238957302740063e-05,
      "loss": 0.4065,
      "step": 26412
    },
    {
      "epoch": 8.340069466371961,
      "grad_norm": 5.747858047485352,
      "learning_rate": 3.238869592674456e-05,
      "loss": 0.5132,
      "step": 26413
    },
    {
      "epoch": 8.340385222608147,
      "grad_norm": 5.483492851257324,
      "learning_rate": 3.238781882608848e-05,
      "loss": 0.4121,
      "step": 26414
    },
    {
      "epoch": 8.340700978844332,
      "grad_norm": 3.718472957611084,
      "learning_rate": 3.2386941725432415e-05,
      "loss": 0.2812,
      "step": 26415
    },
    {
      "epoch": 8.341016735080517,
      "grad_norm": 4.86176872253418,
      "learning_rate": 3.238606462477634e-05,
      "loss": 0.3937,
      "step": 26416
    },
    {
      "epoch": 8.341332491316704,
      "grad_norm": 7.533679962158203,
      "learning_rate": 3.238518752412027e-05,
      "loss": 0.4656,
      "step": 26417
    },
    {
      "epoch": 8.34164824755289,
      "grad_norm": 4.241954326629639,
      "learning_rate": 3.2384310423464196e-05,
      "loss": 0.368,
      "step": 26418
    },
    {
      "epoch": 8.341964003789075,
      "grad_norm": 6.381752014160156,
      "learning_rate": 3.238343332280813e-05,
      "loss": 0.3647,
      "step": 26419
    },
    {
      "epoch": 8.34227976002526,
      "grad_norm": 6.716231346130371,
      "learning_rate": 3.238255622215205e-05,
      "loss": 0.5396,
      "step": 26420
    },
    {
      "epoch": 8.342595516261445,
      "grad_norm": 5.8343095779418945,
      "learning_rate": 3.2381679121495985e-05,
      "loss": 0.4883,
      "step": 26421
    },
    {
      "epoch": 8.342911272497632,
      "grad_norm": 7.276100158691406,
      "learning_rate": 3.238080202083991e-05,
      "loss": 0.4593,
      "step": 26422
    },
    {
      "epoch": 8.343227028733818,
      "grad_norm": 6.483922481536865,
      "learning_rate": 3.237992492018384e-05,
      "loss": 0.4945,
      "step": 26423
    },
    {
      "epoch": 8.343542784970003,
      "grad_norm": 5.678557395935059,
      "learning_rate": 3.2379047819527766e-05,
      "loss": 0.3816,
      "step": 26424
    },
    {
      "epoch": 8.343858541206188,
      "grad_norm": 6.4158124923706055,
      "learning_rate": 3.23781707188717e-05,
      "loss": 0.4708,
      "step": 26425
    },
    {
      "epoch": 8.344174297442375,
      "grad_norm": 7.570247173309326,
      "learning_rate": 3.237729361821562e-05,
      "loss": 0.3624,
      "step": 26426
    },
    {
      "epoch": 8.34449005367856,
      "grad_norm": 6.1028008460998535,
      "learning_rate": 3.2376416517559554e-05,
      "loss": 0.4295,
      "step": 26427
    },
    {
      "epoch": 8.344805809914746,
      "grad_norm": 5.773728370666504,
      "learning_rate": 3.2375539416903486e-05,
      "loss": 0.4399,
      "step": 26428
    },
    {
      "epoch": 8.345121566150931,
      "grad_norm": 17.971118927001953,
      "learning_rate": 3.237466231624741e-05,
      "loss": 0.3519,
      "step": 26429
    },
    {
      "epoch": 8.345437322387117,
      "grad_norm": 4.843010902404785,
      "learning_rate": 3.237378521559134e-05,
      "loss": 0.3583,
      "step": 26430
    },
    {
      "epoch": 8.345753078623304,
      "grad_norm": 6.159945964813232,
      "learning_rate": 3.2372908114935274e-05,
      "loss": 0.5014,
      "step": 26431
    },
    {
      "epoch": 8.346068834859489,
      "grad_norm": 4.673200607299805,
      "learning_rate": 3.2372031014279206e-05,
      "loss": 0.4802,
      "step": 26432
    },
    {
      "epoch": 8.346384591095674,
      "grad_norm": 5.634931564331055,
      "learning_rate": 3.237115391362313e-05,
      "loss": 0.4029,
      "step": 26433
    },
    {
      "epoch": 8.34670034733186,
      "grad_norm": 5.137386798858643,
      "learning_rate": 3.237027681296706e-05,
      "loss": 0.3938,
      "step": 26434
    },
    {
      "epoch": 8.347016103568045,
      "grad_norm": 5.28087854385376,
      "learning_rate": 3.236939971231099e-05,
      "loss": 0.3869,
      "step": 26435
    },
    {
      "epoch": 8.347331859804232,
      "grad_norm": 4.9505486488342285,
      "learning_rate": 3.236852261165492e-05,
      "loss": 0.3102,
      "step": 26436
    },
    {
      "epoch": 8.347647616040417,
      "grad_norm": 6.907150745391846,
      "learning_rate": 3.2367645510998844e-05,
      "loss": 0.495,
      "step": 26437
    },
    {
      "epoch": 8.347963372276602,
      "grad_norm": 6.258215427398682,
      "learning_rate": 3.2366768410342776e-05,
      "loss": 0.4288,
      "step": 26438
    },
    {
      "epoch": 8.348279128512788,
      "grad_norm": 5.488489151000977,
      "learning_rate": 3.23658913096867e-05,
      "loss": 0.3416,
      "step": 26439
    },
    {
      "epoch": 8.348594884748973,
      "grad_norm": 4.9236321449279785,
      "learning_rate": 3.236501420903063e-05,
      "loss": 0.4528,
      "step": 26440
    },
    {
      "epoch": 8.34891064098516,
      "grad_norm": 7.2241716384887695,
      "learning_rate": 3.236413710837456e-05,
      "loss": 0.4712,
      "step": 26441
    },
    {
      "epoch": 8.349226397221345,
      "grad_norm": 5.276041507720947,
      "learning_rate": 3.236326000771849e-05,
      "loss": 0.3786,
      "step": 26442
    },
    {
      "epoch": 8.34954215345753,
      "grad_norm": 4.935580253601074,
      "learning_rate": 3.2362382907062414e-05,
      "loss": 0.3199,
      "step": 26443
    },
    {
      "epoch": 8.349857909693716,
      "grad_norm": 6.205041408538818,
      "learning_rate": 3.2361505806406346e-05,
      "loss": 0.3928,
      "step": 26444
    },
    {
      "epoch": 8.350173665929901,
      "grad_norm": 5.209482669830322,
      "learning_rate": 3.236062870575027e-05,
      "loss": 0.42,
      "step": 26445
    },
    {
      "epoch": 8.350489422166088,
      "grad_norm": 6.067549228668213,
      "learning_rate": 3.23597516050942e-05,
      "loss": 0.5149,
      "step": 26446
    },
    {
      "epoch": 8.350805178402274,
      "grad_norm": 5.042770862579346,
      "learning_rate": 3.235887450443813e-05,
      "loss": 0.3932,
      "step": 26447
    },
    {
      "epoch": 8.351120934638459,
      "grad_norm": 6.576039791107178,
      "learning_rate": 3.235799740378206e-05,
      "loss": 0.376,
      "step": 26448
    },
    {
      "epoch": 8.351436690874644,
      "grad_norm": 6.033763408660889,
      "learning_rate": 3.235712030312599e-05,
      "loss": 0.4092,
      "step": 26449
    },
    {
      "epoch": 8.35175244711083,
      "grad_norm": 4.5485382080078125,
      "learning_rate": 3.2356243202469915e-05,
      "loss": 0.4465,
      "step": 26450
    },
    {
      "epoch": 8.352068203347017,
      "grad_norm": 6.381410121917725,
      "learning_rate": 3.235536610181385e-05,
      "loss": 0.4345,
      "step": 26451
    },
    {
      "epoch": 8.352383959583202,
      "grad_norm": 7.239903450012207,
      "learning_rate": 3.235448900115777e-05,
      "loss": 0.3817,
      "step": 26452
    },
    {
      "epoch": 8.352699715819387,
      "grad_norm": 5.689253807067871,
      "learning_rate": 3.2353611900501704e-05,
      "loss": 0.3988,
      "step": 26453
    },
    {
      "epoch": 8.353015472055572,
      "grad_norm": 6.243356227874756,
      "learning_rate": 3.235273479984563e-05,
      "loss": 0.407,
      "step": 26454
    },
    {
      "epoch": 8.35333122829176,
      "grad_norm": 5.436314105987549,
      "learning_rate": 3.235185769918956e-05,
      "loss": 0.4157,
      "step": 26455
    },
    {
      "epoch": 8.353646984527945,
      "grad_norm": 7.033570766448975,
      "learning_rate": 3.2350980598533485e-05,
      "loss": 0.4338,
      "step": 26456
    },
    {
      "epoch": 8.35396274076413,
      "grad_norm": 7.907810211181641,
      "learning_rate": 3.235010349787742e-05,
      "loss": 0.4192,
      "step": 26457
    },
    {
      "epoch": 8.354278497000315,
      "grad_norm": 7.2640790939331055,
      "learning_rate": 3.234922639722135e-05,
      "loss": 0.4371,
      "step": 26458
    },
    {
      "epoch": 8.3545942532365,
      "grad_norm": 5.58031702041626,
      "learning_rate": 3.234834929656528e-05,
      "loss": 0.4689,
      "step": 26459
    },
    {
      "epoch": 8.354910009472688,
      "grad_norm": 7.060235977172852,
      "learning_rate": 3.2347472195909205e-05,
      "loss": 0.4402,
      "step": 26460
    },
    {
      "epoch": 8.355225765708873,
      "grad_norm": 6.1162943840026855,
      "learning_rate": 3.234659509525314e-05,
      "loss": 0.3945,
      "step": 26461
    },
    {
      "epoch": 8.355541521945058,
      "grad_norm": 6.811753273010254,
      "learning_rate": 3.234571799459706e-05,
      "loss": 0.334,
      "step": 26462
    },
    {
      "epoch": 8.355857278181244,
      "grad_norm": 6.782186985015869,
      "learning_rate": 3.234484089394099e-05,
      "loss": 0.4819,
      "step": 26463
    },
    {
      "epoch": 8.356173034417429,
      "grad_norm": 8.110562324523926,
      "learning_rate": 3.234396379328492e-05,
      "loss": 0.4291,
      "step": 26464
    },
    {
      "epoch": 8.356488790653616,
      "grad_norm": 5.766965389251709,
      "learning_rate": 3.234308669262885e-05,
      "loss": 0.4108,
      "step": 26465
    },
    {
      "epoch": 8.356804546889801,
      "grad_norm": 6.764031410217285,
      "learning_rate": 3.2342209591972775e-05,
      "loss": 0.3693,
      "step": 26466
    },
    {
      "epoch": 8.357120303125987,
      "grad_norm": 6.272158145904541,
      "learning_rate": 3.2341332491316707e-05,
      "loss": 0.4141,
      "step": 26467
    },
    {
      "epoch": 8.357436059362172,
      "grad_norm": 6.9777655601501465,
      "learning_rate": 3.234045539066063e-05,
      "loss": 0.4341,
      "step": 26468
    },
    {
      "epoch": 8.357751815598357,
      "grad_norm": 6.972018241882324,
      "learning_rate": 3.233957829000456e-05,
      "loss": 0.4115,
      "step": 26469
    },
    {
      "epoch": 8.358067571834544,
      "grad_norm": 7.1896281242370605,
      "learning_rate": 3.2338701189348495e-05,
      "loss": 0.3659,
      "step": 26470
    },
    {
      "epoch": 8.35838332807073,
      "grad_norm": 9.077176094055176,
      "learning_rate": 3.233782408869242e-05,
      "loss": 0.4362,
      "step": 26471
    },
    {
      "epoch": 8.358699084306915,
      "grad_norm": 5.439827919006348,
      "learning_rate": 3.233694698803635e-05,
      "loss": 0.3328,
      "step": 26472
    },
    {
      "epoch": 8.3590148405431,
      "grad_norm": 4.764092922210693,
      "learning_rate": 3.2336069887380276e-05,
      "loss": 0.3676,
      "step": 26473
    },
    {
      "epoch": 8.359330596779287,
      "grad_norm": 6.4184064865112305,
      "learning_rate": 3.233519278672421e-05,
      "loss": 0.4727,
      "step": 26474
    },
    {
      "epoch": 8.359646353015473,
      "grad_norm": 6.69465446472168,
      "learning_rate": 3.233431568606813e-05,
      "loss": 0.4768,
      "step": 26475
    },
    {
      "epoch": 8.359962109251658,
      "grad_norm": 6.485175132751465,
      "learning_rate": 3.2333438585412065e-05,
      "loss": 0.465,
      "step": 26476
    },
    {
      "epoch": 8.360277865487843,
      "grad_norm": 6.178873062133789,
      "learning_rate": 3.233256148475599e-05,
      "loss": 0.4834,
      "step": 26477
    },
    {
      "epoch": 8.360593621724028,
      "grad_norm": 6.925860404968262,
      "learning_rate": 3.233168438409992e-05,
      "loss": 0.4536,
      "step": 26478
    },
    {
      "epoch": 8.360909377960215,
      "grad_norm": 5.564880847930908,
      "learning_rate": 3.2330807283443846e-05,
      "loss": 0.3772,
      "step": 26479
    },
    {
      "epoch": 8.3612251341964,
      "grad_norm": 6.208123207092285,
      "learning_rate": 3.232993018278778e-05,
      "loss": 0.4461,
      "step": 26480
    },
    {
      "epoch": 8.361540890432586,
      "grad_norm": 5.6933088302612305,
      "learning_rate": 3.23290530821317e-05,
      "loss": 0.4227,
      "step": 26481
    },
    {
      "epoch": 8.361856646668771,
      "grad_norm": 5.289669513702393,
      "learning_rate": 3.2328175981475634e-05,
      "loss": 0.3307,
      "step": 26482
    },
    {
      "epoch": 8.362172402904957,
      "grad_norm": 5.490790367126465,
      "learning_rate": 3.232729888081956e-05,
      "loss": 0.3623,
      "step": 26483
    },
    {
      "epoch": 8.362488159141144,
      "grad_norm": 5.983233451843262,
      "learning_rate": 3.232642178016349e-05,
      "loss": 0.4142,
      "step": 26484
    },
    {
      "epoch": 8.362803915377329,
      "grad_norm": 6.33003568649292,
      "learning_rate": 3.2325544679507416e-05,
      "loss": 0.4858,
      "step": 26485
    },
    {
      "epoch": 8.363119671613514,
      "grad_norm": 5.319845676422119,
      "learning_rate": 3.2324667578851354e-05,
      "loss": 0.4073,
      "step": 26486
    },
    {
      "epoch": 8.3634354278497,
      "grad_norm": 7.324123859405518,
      "learning_rate": 3.232379047819528e-05,
      "loss": 0.51,
      "step": 26487
    },
    {
      "epoch": 8.363751184085885,
      "grad_norm": 7.194279193878174,
      "learning_rate": 3.232291337753921e-05,
      "loss": 0.4516,
      "step": 26488
    },
    {
      "epoch": 8.364066940322072,
      "grad_norm": 5.55577278137207,
      "learning_rate": 3.2322036276883136e-05,
      "loss": 0.3338,
      "step": 26489
    },
    {
      "epoch": 8.364382696558257,
      "grad_norm": 5.698799133300781,
      "learning_rate": 3.232115917622707e-05,
      "loss": 0.4386,
      "step": 26490
    },
    {
      "epoch": 8.364698452794443,
      "grad_norm": 5.261393070220947,
      "learning_rate": 3.232028207557099e-05,
      "loss": 0.3933,
      "step": 26491
    },
    {
      "epoch": 8.365014209030628,
      "grad_norm": 5.775566577911377,
      "learning_rate": 3.2319404974914924e-05,
      "loss": 0.5222,
      "step": 26492
    },
    {
      "epoch": 8.365329965266813,
      "grad_norm": 7.023841381072998,
      "learning_rate": 3.2318527874258856e-05,
      "loss": 0.4105,
      "step": 26493
    },
    {
      "epoch": 8.365645721503,
      "grad_norm": 6.5969977378845215,
      "learning_rate": 3.231765077360278e-05,
      "loss": 0.4597,
      "step": 26494
    },
    {
      "epoch": 8.365961477739186,
      "grad_norm": 5.812811374664307,
      "learning_rate": 3.231677367294671e-05,
      "loss": 0.4627,
      "step": 26495
    },
    {
      "epoch": 8.36627723397537,
      "grad_norm": 5.719029903411865,
      "learning_rate": 3.231589657229064e-05,
      "loss": 0.3552,
      "step": 26496
    },
    {
      "epoch": 8.366592990211556,
      "grad_norm": 4.624356269836426,
      "learning_rate": 3.231501947163457e-05,
      "loss": 0.3169,
      "step": 26497
    },
    {
      "epoch": 8.366908746447743,
      "grad_norm": 5.1189165115356445,
      "learning_rate": 3.2314142370978494e-05,
      "loss": 0.37,
      "step": 26498
    },
    {
      "epoch": 8.367224502683928,
      "grad_norm": 5.603985786437988,
      "learning_rate": 3.2313265270322426e-05,
      "loss": 0.3416,
      "step": 26499
    },
    {
      "epoch": 8.367540258920114,
      "grad_norm": 6.290078639984131,
      "learning_rate": 3.231238816966635e-05,
      "loss": 0.3984,
      "step": 26500
    },
    {
      "epoch": 8.367856015156299,
      "grad_norm": 6.348243236541748,
      "learning_rate": 3.231151106901028e-05,
      "loss": 0.4488,
      "step": 26501
    },
    {
      "epoch": 8.368171771392484,
      "grad_norm": 8.466463088989258,
      "learning_rate": 3.231063396835421e-05,
      "loss": 0.4693,
      "step": 26502
    },
    {
      "epoch": 8.368487527628671,
      "grad_norm": 4.872462749481201,
      "learning_rate": 3.230975686769814e-05,
      "loss": 0.285,
      "step": 26503
    },
    {
      "epoch": 8.368803283864857,
      "grad_norm": 6.015044689178467,
      "learning_rate": 3.2308879767042064e-05,
      "loss": 0.4235,
      "step": 26504
    },
    {
      "epoch": 8.369119040101042,
      "grad_norm": 6.214395046234131,
      "learning_rate": 3.2308002666385995e-05,
      "loss": 0.4378,
      "step": 26505
    },
    {
      "epoch": 8.369434796337227,
      "grad_norm": 7.40894889831543,
      "learning_rate": 3.230712556572992e-05,
      "loss": 0.418,
      "step": 26506
    },
    {
      "epoch": 8.369750552573413,
      "grad_norm": 5.309204578399658,
      "learning_rate": 3.230624846507385e-05,
      "loss": 0.4217,
      "step": 26507
    },
    {
      "epoch": 8.3700663088096,
      "grad_norm": 5.841280460357666,
      "learning_rate": 3.2305371364417784e-05,
      "loss": 0.325,
      "step": 26508
    },
    {
      "epoch": 8.370382065045785,
      "grad_norm": 5.661202907562256,
      "learning_rate": 3.230449426376171e-05,
      "loss": 0.4613,
      "step": 26509
    },
    {
      "epoch": 8.37069782128197,
      "grad_norm": 5.964596748352051,
      "learning_rate": 3.230361716310564e-05,
      "loss": 0.531,
      "step": 26510
    },
    {
      "epoch": 8.371013577518156,
      "grad_norm": 6.561283111572266,
      "learning_rate": 3.2302740062449565e-05,
      "loss": 0.357,
      "step": 26511
    },
    {
      "epoch": 8.37132933375434,
      "grad_norm": 4.727676868438721,
      "learning_rate": 3.23018629617935e-05,
      "loss": 0.3523,
      "step": 26512
    },
    {
      "epoch": 8.371645089990528,
      "grad_norm": 7.490167617797852,
      "learning_rate": 3.230098586113743e-05,
      "loss": 0.4491,
      "step": 26513
    },
    {
      "epoch": 8.371960846226713,
      "grad_norm": 4.762760162353516,
      "learning_rate": 3.230010876048136e-05,
      "loss": 0.3379,
      "step": 26514
    },
    {
      "epoch": 8.372276602462899,
      "grad_norm": 7.341203689575195,
      "learning_rate": 3.2299231659825285e-05,
      "loss": 0.3648,
      "step": 26515
    },
    {
      "epoch": 8.372592358699084,
      "grad_norm": 5.498849868774414,
      "learning_rate": 3.229835455916922e-05,
      "loss": 0.4287,
      "step": 26516
    },
    {
      "epoch": 8.37290811493527,
      "grad_norm": 6.065944671630859,
      "learning_rate": 3.229747745851314e-05,
      "loss": 0.4131,
      "step": 26517
    },
    {
      "epoch": 8.373223871171456,
      "grad_norm": 5.9979753494262695,
      "learning_rate": 3.229660035785707e-05,
      "loss": 0.4304,
      "step": 26518
    },
    {
      "epoch": 8.373539627407641,
      "grad_norm": 6.07566499710083,
      "learning_rate": 3.2295723257201e-05,
      "loss": 0.4315,
      "step": 26519
    },
    {
      "epoch": 8.373855383643827,
      "grad_norm": 5.425323486328125,
      "learning_rate": 3.229484615654493e-05,
      "loss": 0.3838,
      "step": 26520
    },
    {
      "epoch": 8.374171139880012,
      "grad_norm": 5.303096771240234,
      "learning_rate": 3.2293969055888855e-05,
      "loss": 0.3582,
      "step": 26521
    },
    {
      "epoch": 8.3744868961162,
      "grad_norm": 6.231930255889893,
      "learning_rate": 3.2293091955232786e-05,
      "loss": 0.4309,
      "step": 26522
    },
    {
      "epoch": 8.374802652352384,
      "grad_norm": 5.969395637512207,
      "learning_rate": 3.229221485457671e-05,
      "loss": 0.457,
      "step": 26523
    },
    {
      "epoch": 8.37511840858857,
      "grad_norm": 5.519982814788818,
      "learning_rate": 3.229133775392064e-05,
      "loss": 0.5012,
      "step": 26524
    },
    {
      "epoch": 8.375434164824755,
      "grad_norm": 6.11760950088501,
      "learning_rate": 3.229046065326457e-05,
      "loss": 0.4207,
      "step": 26525
    },
    {
      "epoch": 8.37574992106094,
      "grad_norm": 6.020134925842285,
      "learning_rate": 3.22895835526085e-05,
      "loss": 0.4618,
      "step": 26526
    },
    {
      "epoch": 8.376065677297127,
      "grad_norm": 6.081425666809082,
      "learning_rate": 3.2288706451952425e-05,
      "loss": 0.3908,
      "step": 26527
    },
    {
      "epoch": 8.376381433533313,
      "grad_norm": 6.074560642242432,
      "learning_rate": 3.2287829351296356e-05,
      "loss": 0.4039,
      "step": 26528
    },
    {
      "epoch": 8.376697189769498,
      "grad_norm": 5.614276885986328,
      "learning_rate": 3.228695225064028e-05,
      "loss": 0.4276,
      "step": 26529
    },
    {
      "epoch": 8.377012946005683,
      "grad_norm": 5.885659217834473,
      "learning_rate": 3.228607514998421e-05,
      "loss": 0.4527,
      "step": 26530
    },
    {
      "epoch": 8.377328702241869,
      "grad_norm": 6.576430320739746,
      "learning_rate": 3.2285198049328144e-05,
      "loss": 0.3543,
      "step": 26531
    },
    {
      "epoch": 8.377644458478056,
      "grad_norm": 6.561387062072754,
      "learning_rate": 3.228432094867207e-05,
      "loss": 0.3578,
      "step": 26532
    },
    {
      "epoch": 8.377960214714241,
      "grad_norm": 5.934525012969971,
      "learning_rate": 3.2283443848016e-05,
      "loss": 0.4189,
      "step": 26533
    },
    {
      "epoch": 8.378275970950426,
      "grad_norm": 5.2094902992248535,
      "learning_rate": 3.2282566747359926e-05,
      "loss": 0.3618,
      "step": 26534
    },
    {
      "epoch": 8.378591727186611,
      "grad_norm": 6.1681928634643555,
      "learning_rate": 3.228168964670386e-05,
      "loss": 0.4839,
      "step": 26535
    },
    {
      "epoch": 8.378907483422797,
      "grad_norm": 5.1261420249938965,
      "learning_rate": 3.228081254604778e-05,
      "loss": 0.4549,
      "step": 26536
    },
    {
      "epoch": 8.379223239658984,
      "grad_norm": 5.572480201721191,
      "learning_rate": 3.2279935445391714e-05,
      "loss": 0.2976,
      "step": 26537
    },
    {
      "epoch": 8.37953899589517,
      "grad_norm": 4.952466011047363,
      "learning_rate": 3.227905834473564e-05,
      "loss": 0.337,
      "step": 26538
    },
    {
      "epoch": 8.379854752131354,
      "grad_norm": 10.683074951171875,
      "learning_rate": 3.227818124407957e-05,
      "loss": 0.4269,
      "step": 26539
    },
    {
      "epoch": 8.38017050836754,
      "grad_norm": 5.314829349517822,
      "learning_rate": 3.22773041434235e-05,
      "loss": 0.385,
      "step": 26540
    },
    {
      "epoch": 8.380486264603725,
      "grad_norm": 5.317951202392578,
      "learning_rate": 3.2276427042767434e-05,
      "loss": 0.4007,
      "step": 26541
    },
    {
      "epoch": 8.380802020839912,
      "grad_norm": 7.839385986328125,
      "learning_rate": 3.227554994211136e-05,
      "loss": 0.4434,
      "step": 26542
    },
    {
      "epoch": 8.381117777076097,
      "grad_norm": 5.0720906257629395,
      "learning_rate": 3.227467284145529e-05,
      "loss": 0.3137,
      "step": 26543
    },
    {
      "epoch": 8.381433533312283,
      "grad_norm": 4.367034435272217,
      "learning_rate": 3.2273795740799216e-05,
      "loss": 0.3178,
      "step": 26544
    },
    {
      "epoch": 8.381749289548468,
      "grad_norm": 5.099503993988037,
      "learning_rate": 3.227291864014315e-05,
      "loss": 0.3595,
      "step": 26545
    },
    {
      "epoch": 8.382065045784655,
      "grad_norm": 7.634732723236084,
      "learning_rate": 3.227204153948707e-05,
      "loss": 0.3977,
      "step": 26546
    },
    {
      "epoch": 8.38238080202084,
      "grad_norm": 5.932504177093506,
      "learning_rate": 3.2271164438831004e-05,
      "loss": 0.4153,
      "step": 26547
    },
    {
      "epoch": 8.382696558257026,
      "grad_norm": 5.981707572937012,
      "learning_rate": 3.227028733817493e-05,
      "loss": 0.3884,
      "step": 26548
    },
    {
      "epoch": 8.383012314493211,
      "grad_norm": 7.42318868637085,
      "learning_rate": 3.226941023751886e-05,
      "loss": 0.3966,
      "step": 26549
    },
    {
      "epoch": 8.383328070729396,
      "grad_norm": 5.404091835021973,
      "learning_rate": 3.2268533136862785e-05,
      "loss": 0.4297,
      "step": 26550
    },
    {
      "epoch": 8.383643826965583,
      "grad_norm": 5.230757713317871,
      "learning_rate": 3.226765603620672e-05,
      "loss": 0.4828,
      "step": 26551
    },
    {
      "epoch": 8.383959583201769,
      "grad_norm": 6.942090034484863,
      "learning_rate": 3.226677893555065e-05,
      "loss": 0.4445,
      "step": 26552
    },
    {
      "epoch": 8.384275339437954,
      "grad_norm": 6.5907979011535645,
      "learning_rate": 3.2265901834894574e-05,
      "loss": 0.409,
      "step": 26553
    },
    {
      "epoch": 8.38459109567414,
      "grad_norm": 5.531621932983398,
      "learning_rate": 3.2265024734238505e-05,
      "loss": 0.3503,
      "step": 26554
    },
    {
      "epoch": 8.384906851910324,
      "grad_norm": 7.066463470458984,
      "learning_rate": 3.226414763358243e-05,
      "loss": 0.3941,
      "step": 26555
    },
    {
      "epoch": 8.385222608146512,
      "grad_norm": 6.097995758056641,
      "learning_rate": 3.226327053292636e-05,
      "loss": 0.4161,
      "step": 26556
    },
    {
      "epoch": 8.385538364382697,
      "grad_norm": 6.559021472930908,
      "learning_rate": 3.226239343227029e-05,
      "loss": 0.357,
      "step": 26557
    },
    {
      "epoch": 8.385854120618882,
      "grad_norm": 6.4495110511779785,
      "learning_rate": 3.226151633161422e-05,
      "loss": 0.3939,
      "step": 26558
    },
    {
      "epoch": 8.386169876855067,
      "grad_norm": 7.111776828765869,
      "learning_rate": 3.2260639230958143e-05,
      "loss": 0.4278,
      "step": 26559
    },
    {
      "epoch": 8.386485633091253,
      "grad_norm": 6.344408988952637,
      "learning_rate": 3.2259762130302075e-05,
      "loss": 0.3737,
      "step": 26560
    },
    {
      "epoch": 8.38680138932744,
      "grad_norm": 5.917508602142334,
      "learning_rate": 3.2258885029646e-05,
      "loss": 0.4002,
      "step": 26561
    },
    {
      "epoch": 8.387117145563625,
      "grad_norm": 6.053769111633301,
      "learning_rate": 3.225800792898993e-05,
      "loss": 0.4099,
      "step": 26562
    },
    {
      "epoch": 8.38743290179981,
      "grad_norm": 10.149825096130371,
      "learning_rate": 3.225713082833386e-05,
      "loss": 0.4927,
      "step": 26563
    },
    {
      "epoch": 8.387748658035996,
      "grad_norm": 10.846261024475098,
      "learning_rate": 3.225625372767779e-05,
      "loss": 0.3174,
      "step": 26564
    },
    {
      "epoch": 8.388064414272183,
      "grad_norm": 5.03440523147583,
      "learning_rate": 3.225537662702171e-05,
      "loss": 0.379,
      "step": 26565
    },
    {
      "epoch": 8.388380170508368,
      "grad_norm": 8.13483715057373,
      "learning_rate": 3.2254499526365645e-05,
      "loss": 0.3516,
      "step": 26566
    },
    {
      "epoch": 8.388695926744553,
      "grad_norm": 5.140583038330078,
      "learning_rate": 3.2253622425709577e-05,
      "loss": 0.2332,
      "step": 26567
    },
    {
      "epoch": 8.389011682980739,
      "grad_norm": 6.6971917152404785,
      "learning_rate": 3.225274532505351e-05,
      "loss": 0.4468,
      "step": 26568
    },
    {
      "epoch": 8.389327439216924,
      "grad_norm": 7.362797260284424,
      "learning_rate": 3.225186822439743e-05,
      "loss": 0.4871,
      "step": 26569
    },
    {
      "epoch": 8.389643195453111,
      "grad_norm": 6.2902913093566895,
      "learning_rate": 3.2250991123741365e-05,
      "loss": 0.4038,
      "step": 26570
    },
    {
      "epoch": 8.389958951689296,
      "grad_norm": 5.1121506690979,
      "learning_rate": 3.225011402308529e-05,
      "loss": 0.3323,
      "step": 26571
    },
    {
      "epoch": 8.390274707925482,
      "grad_norm": 8.588970184326172,
      "learning_rate": 3.224923692242922e-05,
      "loss": 0.4866,
      "step": 26572
    },
    {
      "epoch": 8.390590464161667,
      "grad_norm": 5.593449592590332,
      "learning_rate": 3.2248359821773146e-05,
      "loss": 0.4509,
      "step": 26573
    },
    {
      "epoch": 8.390906220397852,
      "grad_norm": 7.1175217628479,
      "learning_rate": 3.224748272111708e-05,
      "loss": 0.5227,
      "step": 26574
    },
    {
      "epoch": 8.39122197663404,
      "grad_norm": 6.337218761444092,
      "learning_rate": 3.224660562046101e-05,
      "loss": 0.3278,
      "step": 26575
    },
    {
      "epoch": 8.391537732870225,
      "grad_norm": 5.982528209686279,
      "learning_rate": 3.2245728519804935e-05,
      "loss": 0.3765,
      "step": 26576
    },
    {
      "epoch": 8.39185348910641,
      "grad_norm": 6.328372478485107,
      "learning_rate": 3.2244851419148866e-05,
      "loss": 0.4487,
      "step": 26577
    },
    {
      "epoch": 8.392169245342595,
      "grad_norm": 5.708611965179443,
      "learning_rate": 3.224397431849279e-05,
      "loss": 0.4183,
      "step": 26578
    },
    {
      "epoch": 8.39248500157878,
      "grad_norm": 5.3305344581604,
      "learning_rate": 3.224309721783672e-05,
      "loss": 0.4167,
      "step": 26579
    },
    {
      "epoch": 8.392800757814967,
      "grad_norm": 5.421535491943359,
      "learning_rate": 3.224222011718065e-05,
      "loss": 0.44,
      "step": 26580
    },
    {
      "epoch": 8.393116514051153,
      "grad_norm": 6.617343902587891,
      "learning_rate": 3.224134301652458e-05,
      "loss": 0.421,
      "step": 26581
    },
    {
      "epoch": 8.393432270287338,
      "grad_norm": 5.525852680206299,
      "learning_rate": 3.2240465915868504e-05,
      "loss": 0.4692,
      "step": 26582
    },
    {
      "epoch": 8.393748026523523,
      "grad_norm": 6.499417304992676,
      "learning_rate": 3.2239588815212436e-05,
      "loss": 0.3511,
      "step": 26583
    },
    {
      "epoch": 8.394063782759709,
      "grad_norm": 6.345253944396973,
      "learning_rate": 3.223871171455636e-05,
      "loss": 0.4319,
      "step": 26584
    },
    {
      "epoch": 8.394379538995896,
      "grad_norm": 5.3525166511535645,
      "learning_rate": 3.223783461390029e-05,
      "loss": 0.354,
      "step": 26585
    },
    {
      "epoch": 8.394695295232081,
      "grad_norm": 5.095176696777344,
      "learning_rate": 3.223695751324422e-05,
      "loss": 0.3549,
      "step": 26586
    },
    {
      "epoch": 8.395011051468266,
      "grad_norm": 6.596391201019287,
      "learning_rate": 3.223608041258815e-05,
      "loss": 0.4288,
      "step": 26587
    },
    {
      "epoch": 8.395326807704452,
      "grad_norm": 6.1202898025512695,
      "learning_rate": 3.2235203311932074e-05,
      "loss": 0.4039,
      "step": 26588
    },
    {
      "epoch": 8.395642563940639,
      "grad_norm": 5.545119762420654,
      "learning_rate": 3.2234326211276006e-05,
      "loss": 0.3533,
      "step": 26589
    },
    {
      "epoch": 8.395958320176824,
      "grad_norm": 5.215873718261719,
      "learning_rate": 3.223344911061994e-05,
      "loss": 0.4247,
      "step": 26590
    },
    {
      "epoch": 8.39627407641301,
      "grad_norm": 7.980874538421631,
      "learning_rate": 3.223257200996386e-05,
      "loss": 0.4959,
      "step": 26591
    },
    {
      "epoch": 8.396589832649195,
      "grad_norm": 4.774885177612305,
      "learning_rate": 3.2231694909307794e-05,
      "loss": 0.3418,
      "step": 26592
    },
    {
      "epoch": 8.39690558888538,
      "grad_norm": 5.918025016784668,
      "learning_rate": 3.223081780865172e-05,
      "loss": 0.4326,
      "step": 26593
    },
    {
      "epoch": 8.397221345121567,
      "grad_norm": 6.45418119430542,
      "learning_rate": 3.222994070799565e-05,
      "loss": 0.4255,
      "step": 26594
    },
    {
      "epoch": 8.397537101357752,
      "grad_norm": 5.9285197257995605,
      "learning_rate": 3.222906360733958e-05,
      "loss": 0.3974,
      "step": 26595
    },
    {
      "epoch": 8.397852857593938,
      "grad_norm": 4.928992748260498,
      "learning_rate": 3.2228186506683514e-05,
      "loss": 0.3442,
      "step": 26596
    },
    {
      "epoch": 8.398168613830123,
      "grad_norm": 6.20465612411499,
      "learning_rate": 3.222730940602744e-05,
      "loss": 0.3886,
      "step": 26597
    },
    {
      "epoch": 8.398484370066308,
      "grad_norm": 5.8290605545043945,
      "learning_rate": 3.222643230537137e-05,
      "loss": 0.3631,
      "step": 26598
    },
    {
      "epoch": 8.398800126302495,
      "grad_norm": 6.835604667663574,
      "learning_rate": 3.2225555204715296e-05,
      "loss": 0.4931,
      "step": 26599
    },
    {
      "epoch": 8.39911588253868,
      "grad_norm": 7.417283535003662,
      "learning_rate": 3.222467810405923e-05,
      "loss": 0.427,
      "step": 26600
    },
    {
      "epoch": 8.39911588253868,
      "eval_loss": 0.9763776063919067,
      "eval_runtime": 4.7112,
      "eval_samples_per_second": 528.525,
      "eval_steps_per_second": 4.457,
      "step": 26600
    },
    {
      "epoch": 8.399431638774866,
      "grad_norm": 4.474654197692871,
      "learning_rate": 3.222380100340315e-05,
      "loss": 0.4287,
      "step": 26601
    },
    {
      "epoch": 8.399747395011051,
      "grad_norm": 5.9277663230896,
      "learning_rate": 3.2222923902747084e-05,
      "loss": 0.4829,
      "step": 26602
    },
    {
      "epoch": 8.400063151247236,
      "grad_norm": 5.96812629699707,
      "learning_rate": 3.222204680209101e-05,
      "loss": 0.3557,
      "step": 26603
    },
    {
      "epoch": 8.400378907483423,
      "grad_norm": 7.294458866119385,
      "learning_rate": 3.222116970143494e-05,
      "loss": 0.4193,
      "step": 26604
    },
    {
      "epoch": 8.400694663719609,
      "grad_norm": 5.85269021987915,
      "learning_rate": 3.2220292600778865e-05,
      "loss": 0.3499,
      "step": 26605
    },
    {
      "epoch": 8.401010419955794,
      "grad_norm": 5.752410888671875,
      "learning_rate": 3.22194155001228e-05,
      "loss": 0.3797,
      "step": 26606
    },
    {
      "epoch": 8.40132617619198,
      "grad_norm": 5.973392009735107,
      "learning_rate": 3.221853839946672e-05,
      "loss": 0.3911,
      "step": 26607
    },
    {
      "epoch": 8.401641932428166,
      "grad_norm": 7.938413619995117,
      "learning_rate": 3.2217661298810654e-05,
      "loss": 0.4707,
      "step": 26608
    },
    {
      "epoch": 8.401957688664352,
      "grad_norm": 5.42623233795166,
      "learning_rate": 3.221678419815458e-05,
      "loss": 0.4852,
      "step": 26609
    },
    {
      "epoch": 8.402273444900537,
      "grad_norm": 5.90358304977417,
      "learning_rate": 3.221590709749851e-05,
      "loss": 0.4035,
      "step": 26610
    },
    {
      "epoch": 8.402589201136722,
      "grad_norm": 5.880504131317139,
      "learning_rate": 3.221502999684244e-05,
      "loss": 0.4285,
      "step": 26611
    },
    {
      "epoch": 8.402904957372908,
      "grad_norm": 6.170522212982178,
      "learning_rate": 3.221415289618637e-05,
      "loss": 0.3701,
      "step": 26612
    },
    {
      "epoch": 8.403220713609095,
      "grad_norm": 5.612435817718506,
      "learning_rate": 3.22132757955303e-05,
      "loss": 0.3864,
      "step": 26613
    },
    {
      "epoch": 8.40353646984528,
      "grad_norm": 6.034356594085693,
      "learning_rate": 3.221239869487422e-05,
      "loss": 0.3988,
      "step": 26614
    },
    {
      "epoch": 8.403852226081465,
      "grad_norm": 5.174301624298096,
      "learning_rate": 3.2211521594218155e-05,
      "loss": 0.3153,
      "step": 26615
    },
    {
      "epoch": 8.40416798231765,
      "grad_norm": 4.7434868812561035,
      "learning_rate": 3.221064449356208e-05,
      "loss": 0.3029,
      "step": 26616
    },
    {
      "epoch": 8.404483738553836,
      "grad_norm": 6.684299945831299,
      "learning_rate": 3.220976739290601e-05,
      "loss": 0.3797,
      "step": 26617
    },
    {
      "epoch": 8.404799494790023,
      "grad_norm": 6.981339931488037,
      "learning_rate": 3.2208890292249937e-05,
      "loss": 0.4256,
      "step": 26618
    },
    {
      "epoch": 8.405115251026208,
      "grad_norm": 5.178604602813721,
      "learning_rate": 3.220801319159387e-05,
      "loss": 0.3778,
      "step": 26619
    },
    {
      "epoch": 8.405431007262393,
      "grad_norm": 6.170984745025635,
      "learning_rate": 3.220713609093779e-05,
      "loss": 0.4441,
      "step": 26620
    },
    {
      "epoch": 8.405746763498579,
      "grad_norm": 4.735987663269043,
      "learning_rate": 3.2206258990281725e-05,
      "loss": 0.2876,
      "step": 26621
    },
    {
      "epoch": 8.406062519734764,
      "grad_norm": 7.6410040855407715,
      "learning_rate": 3.2205381889625657e-05,
      "loss": 0.4621,
      "step": 26622
    },
    {
      "epoch": 8.406378275970951,
      "grad_norm": 6.317324638366699,
      "learning_rate": 3.220450478896959e-05,
      "loss": 0.4259,
      "step": 26623
    },
    {
      "epoch": 8.406694032207136,
      "grad_norm": 6.05602502822876,
      "learning_rate": 3.220362768831351e-05,
      "loss": 0.3847,
      "step": 26624
    },
    {
      "epoch": 8.407009788443322,
      "grad_norm": 7.364024639129639,
      "learning_rate": 3.2202750587657445e-05,
      "loss": 0.3564,
      "step": 26625
    },
    {
      "epoch": 8.407325544679507,
      "grad_norm": 10.483257293701172,
      "learning_rate": 3.220187348700137e-05,
      "loss": 0.4062,
      "step": 26626
    },
    {
      "epoch": 8.407641300915692,
      "grad_norm": 6.09731912612915,
      "learning_rate": 3.22009963863453e-05,
      "loss": 0.3644,
      "step": 26627
    },
    {
      "epoch": 8.40795705715188,
      "grad_norm": 5.795336723327637,
      "learning_rate": 3.2200119285689226e-05,
      "loss": 0.3204,
      "step": 26628
    },
    {
      "epoch": 8.408272813388065,
      "grad_norm": 4.577847957611084,
      "learning_rate": 3.219924218503316e-05,
      "loss": 0.3819,
      "step": 26629
    },
    {
      "epoch": 8.40858856962425,
      "grad_norm": 5.44804048538208,
      "learning_rate": 3.219836508437708e-05,
      "loss": 0.3524,
      "step": 26630
    },
    {
      "epoch": 8.408904325860435,
      "grad_norm": 7.602144718170166,
      "learning_rate": 3.2197487983721015e-05,
      "loss": 0.3976,
      "step": 26631
    },
    {
      "epoch": 8.40922008209662,
      "grad_norm": 6.48422908782959,
      "learning_rate": 3.219661088306494e-05,
      "loss": 0.3759,
      "step": 26632
    },
    {
      "epoch": 8.409535838332808,
      "grad_norm": 6.93110466003418,
      "learning_rate": 3.219573378240887e-05,
      "loss": 0.3495,
      "step": 26633
    },
    {
      "epoch": 8.409851594568993,
      "grad_norm": 6.325238227844238,
      "learning_rate": 3.21948566817528e-05,
      "loss": 0.4191,
      "step": 26634
    },
    {
      "epoch": 8.410167350805178,
      "grad_norm": 4.260667324066162,
      "learning_rate": 3.219397958109673e-05,
      "loss": 0.3405,
      "step": 26635
    },
    {
      "epoch": 8.410483107041363,
      "grad_norm": 9.422945022583008,
      "learning_rate": 3.219310248044066e-05,
      "loss": 0.5196,
      "step": 26636
    },
    {
      "epoch": 8.41079886327755,
      "grad_norm": 5.865710258483887,
      "learning_rate": 3.2192225379784584e-05,
      "loss": 0.4632,
      "step": 26637
    },
    {
      "epoch": 8.411114619513736,
      "grad_norm": 6.593166351318359,
      "learning_rate": 3.2191348279128516e-05,
      "loss": 0.3774,
      "step": 26638
    },
    {
      "epoch": 8.411430375749921,
      "grad_norm": 6.9304728507995605,
      "learning_rate": 3.219047117847244e-05,
      "loss": 0.4979,
      "step": 26639
    },
    {
      "epoch": 8.411746131986106,
      "grad_norm": 8.212808609008789,
      "learning_rate": 3.218959407781637e-05,
      "loss": 0.441,
      "step": 26640
    },
    {
      "epoch": 8.412061888222292,
      "grad_norm": 6.3962907791137695,
      "learning_rate": 3.21887169771603e-05,
      "loss": 0.4268,
      "step": 26641
    },
    {
      "epoch": 8.412377644458479,
      "grad_norm": 6.2927937507629395,
      "learning_rate": 3.218783987650423e-05,
      "loss": 0.3176,
      "step": 26642
    },
    {
      "epoch": 8.412693400694664,
      "grad_norm": 6.830533027648926,
      "learning_rate": 3.2186962775848154e-05,
      "loss": 0.4247,
      "step": 26643
    },
    {
      "epoch": 8.41300915693085,
      "grad_norm": 7.772685527801514,
      "learning_rate": 3.2186085675192086e-05,
      "loss": 0.4846,
      "step": 26644
    },
    {
      "epoch": 8.413324913167035,
      "grad_norm": 5.341597557067871,
      "learning_rate": 3.218520857453601e-05,
      "loss": 0.3651,
      "step": 26645
    },
    {
      "epoch": 8.41364066940322,
      "grad_norm": 5.393550872802734,
      "learning_rate": 3.218433147387994e-05,
      "loss": 0.5696,
      "step": 26646
    },
    {
      "epoch": 8.413956425639407,
      "grad_norm": 5.412946701049805,
      "learning_rate": 3.218345437322387e-05,
      "loss": 0.4161,
      "step": 26647
    },
    {
      "epoch": 8.414272181875592,
      "grad_norm": 6.714267730712891,
      "learning_rate": 3.21825772725678e-05,
      "loss": 0.3609,
      "step": 26648
    },
    {
      "epoch": 8.414587938111778,
      "grad_norm": 5.943161487579346,
      "learning_rate": 3.218170017191173e-05,
      "loss": 0.3446,
      "step": 26649
    },
    {
      "epoch": 8.414903694347963,
      "grad_norm": 5.5546183586120605,
      "learning_rate": 3.218082307125566e-05,
      "loss": 0.478,
      "step": 26650
    },
    {
      "epoch": 8.415219450584148,
      "grad_norm": 5.676325798034668,
      "learning_rate": 3.217994597059959e-05,
      "loss": 0.3298,
      "step": 26651
    },
    {
      "epoch": 8.415535206820335,
      "grad_norm": 5.892094135284424,
      "learning_rate": 3.217906886994352e-05,
      "loss": 0.3674,
      "step": 26652
    },
    {
      "epoch": 8.41585096305652,
      "grad_norm": 8.179658889770508,
      "learning_rate": 3.2178191769287444e-05,
      "loss": 0.4335,
      "step": 26653
    },
    {
      "epoch": 8.416166719292706,
      "grad_norm": 5.3346428871154785,
      "learning_rate": 3.2177314668631375e-05,
      "loss": 0.3887,
      "step": 26654
    },
    {
      "epoch": 8.416482475528891,
      "grad_norm": 5.9004435539245605,
      "learning_rate": 3.217643756797531e-05,
      "loss": 0.4235,
      "step": 26655
    },
    {
      "epoch": 8.416798231765078,
      "grad_norm": 5.6682448387146,
      "learning_rate": 3.217556046731923e-05,
      "loss": 0.4076,
      "step": 26656
    },
    {
      "epoch": 8.417113988001264,
      "grad_norm": 6.38143253326416,
      "learning_rate": 3.2174683366663164e-05,
      "loss": 0.4156,
      "step": 26657
    },
    {
      "epoch": 8.417429744237449,
      "grad_norm": 7.073275566101074,
      "learning_rate": 3.217380626600709e-05,
      "loss": 0.3666,
      "step": 26658
    },
    {
      "epoch": 8.417745500473634,
      "grad_norm": 5.21414041519165,
      "learning_rate": 3.217292916535102e-05,
      "loss": 0.3585,
      "step": 26659
    },
    {
      "epoch": 8.41806125670982,
      "grad_norm": 7.695603847503662,
      "learning_rate": 3.2172052064694945e-05,
      "loss": 0.4958,
      "step": 26660
    },
    {
      "epoch": 8.418377012946006,
      "grad_norm": 6.525541305541992,
      "learning_rate": 3.217117496403888e-05,
      "loss": 0.4677,
      "step": 26661
    },
    {
      "epoch": 8.418692769182192,
      "grad_norm": 6.012521266937256,
      "learning_rate": 3.21702978633828e-05,
      "loss": 0.4613,
      "step": 26662
    },
    {
      "epoch": 8.419008525418377,
      "grad_norm": 7.2208685874938965,
      "learning_rate": 3.2169420762726734e-05,
      "loss": 0.4986,
      "step": 26663
    },
    {
      "epoch": 8.419324281654562,
      "grad_norm": 7.476418972015381,
      "learning_rate": 3.216854366207066e-05,
      "loss": 0.4224,
      "step": 26664
    },
    {
      "epoch": 8.419640037890748,
      "grad_norm": 5.305358409881592,
      "learning_rate": 3.216766656141459e-05,
      "loss": 0.4423,
      "step": 26665
    },
    {
      "epoch": 8.419955794126935,
      "grad_norm": 4.573252201080322,
      "learning_rate": 3.2166789460758515e-05,
      "loss": 0.3289,
      "step": 26666
    },
    {
      "epoch": 8.42027155036312,
      "grad_norm": 5.845496654510498,
      "learning_rate": 3.216591236010245e-05,
      "loss": 0.3307,
      "step": 26667
    },
    {
      "epoch": 8.420587306599305,
      "grad_norm": 5.740060806274414,
      "learning_rate": 3.216503525944637e-05,
      "loss": 0.3972,
      "step": 26668
    },
    {
      "epoch": 8.42090306283549,
      "grad_norm": 8.65478801727295,
      "learning_rate": 3.21641581587903e-05,
      "loss": 0.4904,
      "step": 26669
    },
    {
      "epoch": 8.421218819071676,
      "grad_norm": 5.041930198669434,
      "learning_rate": 3.216328105813423e-05,
      "loss": 0.3717,
      "step": 26670
    },
    {
      "epoch": 8.421534575307863,
      "grad_norm": 8.28792667388916,
      "learning_rate": 3.216240395747816e-05,
      "loss": 0.4016,
      "step": 26671
    },
    {
      "epoch": 8.421850331544048,
      "grad_norm": 6.508860111236572,
      "learning_rate": 3.216152685682209e-05,
      "loss": 0.433,
      "step": 26672
    },
    {
      "epoch": 8.422166087780234,
      "grad_norm": 7.3018903732299805,
      "learning_rate": 3.2160649756166016e-05,
      "loss": 0.409,
      "step": 26673
    },
    {
      "epoch": 8.422481844016419,
      "grad_norm": 4.114863395690918,
      "learning_rate": 3.215977265550995e-05,
      "loss": 0.2981,
      "step": 26674
    },
    {
      "epoch": 8.422797600252604,
      "grad_norm": 7.973171234130859,
      "learning_rate": 3.215889555485387e-05,
      "loss": 0.456,
      "step": 26675
    },
    {
      "epoch": 8.423113356488791,
      "grad_norm": 6.263343811035156,
      "learning_rate": 3.2158018454197805e-05,
      "loss": 0.3866,
      "step": 26676
    },
    {
      "epoch": 8.423429112724977,
      "grad_norm": 4.622921466827393,
      "learning_rate": 3.2157141353541736e-05,
      "loss": 0.2836,
      "step": 26677
    },
    {
      "epoch": 8.423744868961162,
      "grad_norm": 7.974309921264648,
      "learning_rate": 3.215626425288567e-05,
      "loss": 0.4009,
      "step": 26678
    },
    {
      "epoch": 8.424060625197347,
      "grad_norm": 6.212564468383789,
      "learning_rate": 3.215538715222959e-05,
      "loss": 0.3818,
      "step": 26679
    },
    {
      "epoch": 8.424376381433532,
      "grad_norm": 6.335330963134766,
      "learning_rate": 3.2154510051573525e-05,
      "loss": 0.424,
      "step": 26680
    },
    {
      "epoch": 8.42469213766972,
      "grad_norm": 7.5257368087768555,
      "learning_rate": 3.215363295091745e-05,
      "loss": 0.5168,
      "step": 26681
    },
    {
      "epoch": 8.425007893905905,
      "grad_norm": 8.036944389343262,
      "learning_rate": 3.215275585026138e-05,
      "loss": 0.4183,
      "step": 26682
    },
    {
      "epoch": 8.42532365014209,
      "grad_norm": 5.284348487854004,
      "learning_rate": 3.2151878749605306e-05,
      "loss": 0.2959,
      "step": 26683
    },
    {
      "epoch": 8.425639406378275,
      "grad_norm": 7.244270324707031,
      "learning_rate": 3.215100164894924e-05,
      "loss": 0.4484,
      "step": 26684
    },
    {
      "epoch": 8.425955162614462,
      "grad_norm": 6.493074893951416,
      "learning_rate": 3.215012454829316e-05,
      "loss": 0.4596,
      "step": 26685
    },
    {
      "epoch": 8.426270918850648,
      "grad_norm": 4.0962138175964355,
      "learning_rate": 3.2149247447637094e-05,
      "loss": 0.3098,
      "step": 26686
    },
    {
      "epoch": 8.426586675086833,
      "grad_norm": 6.185085773468018,
      "learning_rate": 3.214837034698102e-05,
      "loss": 0.3787,
      "step": 26687
    },
    {
      "epoch": 8.426902431323018,
      "grad_norm": 5.999283313751221,
      "learning_rate": 3.214749324632495e-05,
      "loss": 0.4202,
      "step": 26688
    },
    {
      "epoch": 8.427218187559204,
      "grad_norm": 6.359715461730957,
      "learning_rate": 3.2146616145668876e-05,
      "loss": 0.366,
      "step": 26689
    },
    {
      "epoch": 8.42753394379539,
      "grad_norm": 5.365255832672119,
      "learning_rate": 3.214573904501281e-05,
      "loss": 0.3191,
      "step": 26690
    },
    {
      "epoch": 8.427849700031576,
      "grad_norm": 7.606934070587158,
      "learning_rate": 3.214486194435673e-05,
      "loss": 0.4323,
      "step": 26691
    },
    {
      "epoch": 8.428165456267761,
      "grad_norm": 4.796069622039795,
      "learning_rate": 3.2143984843700664e-05,
      "loss": 0.3823,
      "step": 26692
    },
    {
      "epoch": 8.428481212503947,
      "grad_norm": 5.786350727081299,
      "learning_rate": 3.2143107743044596e-05,
      "loss": 0.3768,
      "step": 26693
    },
    {
      "epoch": 8.428796968740132,
      "grad_norm": 6.6486077308654785,
      "learning_rate": 3.214223064238852e-05,
      "loss": 0.4056,
      "step": 26694
    },
    {
      "epoch": 8.429112724976319,
      "grad_norm": 6.8684186935424805,
      "learning_rate": 3.214135354173245e-05,
      "loss": 0.3989,
      "step": 26695
    },
    {
      "epoch": 8.429428481212504,
      "grad_norm": 5.729847431182861,
      "learning_rate": 3.214047644107638e-05,
      "loss": 0.4263,
      "step": 26696
    },
    {
      "epoch": 8.42974423744869,
      "grad_norm": 5.844830513000488,
      "learning_rate": 3.213959934042031e-05,
      "loss": 0.425,
      "step": 26697
    },
    {
      "epoch": 8.430059993684875,
      "grad_norm": 6.614610195159912,
      "learning_rate": 3.2138722239764234e-05,
      "loss": 0.3729,
      "step": 26698
    },
    {
      "epoch": 8.43037574992106,
      "grad_norm": 5.384809494018555,
      "learning_rate": 3.2137845139108166e-05,
      "loss": 0.4319,
      "step": 26699
    },
    {
      "epoch": 8.430691506157247,
      "grad_norm": 9.215229988098145,
      "learning_rate": 3.213696803845209e-05,
      "loss": 0.4427,
      "step": 26700
    },
    {
      "epoch": 8.431007262393432,
      "grad_norm": 4.475106239318848,
      "learning_rate": 3.213609093779602e-05,
      "loss": 0.3099,
      "step": 26701
    },
    {
      "epoch": 8.431323018629618,
      "grad_norm": 6.32620906829834,
      "learning_rate": 3.213521383713995e-05,
      "loss": 0.3012,
      "step": 26702
    },
    {
      "epoch": 8.431638774865803,
      "grad_norm": 7.965640068054199,
      "learning_rate": 3.213433673648388e-05,
      "loss": 0.3964,
      "step": 26703
    },
    {
      "epoch": 8.43195453110199,
      "grad_norm": 6.027582168579102,
      "learning_rate": 3.213345963582781e-05,
      "loss": 0.4348,
      "step": 26704
    },
    {
      "epoch": 8.432270287338175,
      "grad_norm": 6.869719982147217,
      "learning_rate": 3.213258253517174e-05,
      "loss": 0.4437,
      "step": 26705
    },
    {
      "epoch": 8.43258604357436,
      "grad_norm": 6.093422889709473,
      "learning_rate": 3.213170543451567e-05,
      "loss": 0.408,
      "step": 26706
    },
    {
      "epoch": 8.432901799810546,
      "grad_norm": 5.972864627838135,
      "learning_rate": 3.21308283338596e-05,
      "loss": 0.4778,
      "step": 26707
    },
    {
      "epoch": 8.433217556046731,
      "grad_norm": 4.541996479034424,
      "learning_rate": 3.2129951233203524e-05,
      "loss": 0.3776,
      "step": 26708
    },
    {
      "epoch": 8.433533312282918,
      "grad_norm": 5.617953777313232,
      "learning_rate": 3.2129074132547455e-05,
      "loss": 0.3117,
      "step": 26709
    },
    {
      "epoch": 8.433849068519104,
      "grad_norm": 5.973137378692627,
      "learning_rate": 3.212819703189138e-05,
      "loss": 0.3291,
      "step": 26710
    },
    {
      "epoch": 8.434164824755289,
      "grad_norm": 8.582623481750488,
      "learning_rate": 3.212731993123531e-05,
      "loss": 0.4352,
      "step": 26711
    },
    {
      "epoch": 8.434480580991474,
      "grad_norm": 5.768676280975342,
      "learning_rate": 3.212644283057924e-05,
      "loss": 0.3992,
      "step": 26712
    },
    {
      "epoch": 8.43479633722766,
      "grad_norm": 7.347864151000977,
      "learning_rate": 3.212556572992317e-05,
      "loss": 0.3438,
      "step": 26713
    },
    {
      "epoch": 8.435112093463847,
      "grad_norm": 5.8663482666015625,
      "learning_rate": 3.2124688629267093e-05,
      "loss": 0.379,
      "step": 26714
    },
    {
      "epoch": 8.435427849700032,
      "grad_norm": 5.889820098876953,
      "learning_rate": 3.2123811528611025e-05,
      "loss": 0.4795,
      "step": 26715
    },
    {
      "epoch": 8.435743605936217,
      "grad_norm": 7.126771450042725,
      "learning_rate": 3.212293442795496e-05,
      "loss": 0.4138,
      "step": 26716
    },
    {
      "epoch": 8.436059362172402,
      "grad_norm": 6.342787265777588,
      "learning_rate": 3.212205732729888e-05,
      "loss": 0.4807,
      "step": 26717
    },
    {
      "epoch": 8.436375118408588,
      "grad_norm": 5.250584125518799,
      "learning_rate": 3.2121180226642813e-05,
      "loss": 0.4034,
      "step": 26718
    },
    {
      "epoch": 8.436690874644775,
      "grad_norm": 6.304867267608643,
      "learning_rate": 3.212030312598674e-05,
      "loss": 0.3678,
      "step": 26719
    },
    {
      "epoch": 8.43700663088096,
      "grad_norm": 5.7507476806640625,
      "learning_rate": 3.211942602533067e-05,
      "loss": 0.3047,
      "step": 26720
    },
    {
      "epoch": 8.437322387117145,
      "grad_norm": 6.468947410583496,
      "learning_rate": 3.2118548924674595e-05,
      "loss": 0.4244,
      "step": 26721
    },
    {
      "epoch": 8.43763814335333,
      "grad_norm": 5.703767776489258,
      "learning_rate": 3.2117671824018527e-05,
      "loss": 0.4297,
      "step": 26722
    },
    {
      "epoch": 8.437953899589516,
      "grad_norm": 5.670670509338379,
      "learning_rate": 3.211679472336245e-05,
      "loss": 0.389,
      "step": 26723
    },
    {
      "epoch": 8.438269655825703,
      "grad_norm": 5.627203941345215,
      "learning_rate": 3.211591762270638e-05,
      "loss": 0.4379,
      "step": 26724
    },
    {
      "epoch": 8.438585412061888,
      "grad_norm": 5.2351298332214355,
      "learning_rate": 3.211504052205031e-05,
      "loss": 0.3771,
      "step": 26725
    },
    {
      "epoch": 8.438901168298074,
      "grad_norm": 5.959540367126465,
      "learning_rate": 3.211416342139424e-05,
      "loss": 0.3594,
      "step": 26726
    },
    {
      "epoch": 8.439216924534259,
      "grad_norm": 11.650930404663086,
      "learning_rate": 3.2113286320738165e-05,
      "loss": 0.3351,
      "step": 26727
    },
    {
      "epoch": 8.439532680770446,
      "grad_norm": 6.255973815917969,
      "learning_rate": 3.2112409220082096e-05,
      "loss": 0.4182,
      "step": 26728
    },
    {
      "epoch": 8.439848437006631,
      "grad_norm": 8.73956298828125,
      "learning_rate": 3.211153211942602e-05,
      "loss": 0.432,
      "step": 26729
    },
    {
      "epoch": 8.440164193242817,
      "grad_norm": 6.234027862548828,
      "learning_rate": 3.211065501876995e-05,
      "loss": 0.4132,
      "step": 26730
    },
    {
      "epoch": 8.440479949479002,
      "grad_norm": 5.857088565826416,
      "learning_rate": 3.2109777918113885e-05,
      "loss": 0.3357,
      "step": 26731
    },
    {
      "epoch": 8.440795705715187,
      "grad_norm": 8.064187049865723,
      "learning_rate": 3.2108900817457816e-05,
      "loss": 0.4129,
      "step": 26732
    },
    {
      "epoch": 8.441111461951374,
      "grad_norm": 8.104366302490234,
      "learning_rate": 3.210802371680174e-05,
      "loss": 0.361,
      "step": 26733
    },
    {
      "epoch": 8.44142721818756,
      "grad_norm": 7.280031204223633,
      "learning_rate": 3.210714661614567e-05,
      "loss": 0.4845,
      "step": 26734
    },
    {
      "epoch": 8.441742974423745,
      "grad_norm": 7.288888454437256,
      "learning_rate": 3.21062695154896e-05,
      "loss": 0.4392,
      "step": 26735
    },
    {
      "epoch": 8.44205873065993,
      "grad_norm": 6.024000644683838,
      "learning_rate": 3.210539241483353e-05,
      "loss": 0.4799,
      "step": 26736
    },
    {
      "epoch": 8.442374486896115,
      "grad_norm": 7.499197006225586,
      "learning_rate": 3.210451531417746e-05,
      "loss": 0.355,
      "step": 26737
    },
    {
      "epoch": 8.442690243132303,
      "grad_norm": 5.308246612548828,
      "learning_rate": 3.2103638213521386e-05,
      "loss": 0.4679,
      "step": 26738
    },
    {
      "epoch": 8.443005999368488,
      "grad_norm": 5.710867881774902,
      "learning_rate": 3.210276111286532e-05,
      "loss": 0.491,
      "step": 26739
    },
    {
      "epoch": 8.443321755604673,
      "grad_norm": 6.8823628425598145,
      "learning_rate": 3.210188401220924e-05,
      "loss": 0.3485,
      "step": 26740
    },
    {
      "epoch": 8.443637511840858,
      "grad_norm": 8.664093971252441,
      "learning_rate": 3.2101006911553174e-05,
      "loss": 0.4372,
      "step": 26741
    },
    {
      "epoch": 8.443953268077044,
      "grad_norm": 5.515536308288574,
      "learning_rate": 3.21001298108971e-05,
      "loss": 0.3712,
      "step": 26742
    },
    {
      "epoch": 8.44426902431323,
      "grad_norm": 6.112565994262695,
      "learning_rate": 3.209925271024103e-05,
      "loss": 0.4048,
      "step": 26743
    },
    {
      "epoch": 8.444584780549416,
      "grad_norm": 5.9295196533203125,
      "learning_rate": 3.2098375609584956e-05,
      "loss": 0.3784,
      "step": 26744
    },
    {
      "epoch": 8.444900536785601,
      "grad_norm": 8.327250480651855,
      "learning_rate": 3.209749850892889e-05,
      "loss": 0.4687,
      "step": 26745
    },
    {
      "epoch": 8.445216293021787,
      "grad_norm": 5.737680435180664,
      "learning_rate": 3.209662140827281e-05,
      "loss": 0.4707,
      "step": 26746
    },
    {
      "epoch": 8.445532049257974,
      "grad_norm": 8.182941436767578,
      "learning_rate": 3.2095744307616744e-05,
      "loss": 0.4775,
      "step": 26747
    },
    {
      "epoch": 8.445847805494159,
      "grad_norm": 5.075246810913086,
      "learning_rate": 3.209486720696067e-05,
      "loss": 0.4486,
      "step": 26748
    },
    {
      "epoch": 8.446163561730344,
      "grad_norm": 6.087188720703125,
      "learning_rate": 3.20939901063046e-05,
      "loss": 0.4469,
      "step": 26749
    },
    {
      "epoch": 8.44647931796653,
      "grad_norm": 5.105904579162598,
      "learning_rate": 3.2093113005648526e-05,
      "loss": 0.319,
      "step": 26750
    },
    {
      "epoch": 8.446795074202715,
      "grad_norm": 7.321711540222168,
      "learning_rate": 3.209223590499246e-05,
      "loss": 0.5036,
      "step": 26751
    },
    {
      "epoch": 8.447110830438902,
      "grad_norm": 5.37716817855835,
      "learning_rate": 3.209135880433638e-05,
      "loss": 0.3506,
      "step": 26752
    },
    {
      "epoch": 8.447426586675087,
      "grad_norm": 5.734228610992432,
      "learning_rate": 3.2090481703680314e-05,
      "loss": 0.3788,
      "step": 26753
    },
    {
      "epoch": 8.447742342911273,
      "grad_norm": 5.196010589599609,
      "learning_rate": 3.2089604603024246e-05,
      "loss": 0.4168,
      "step": 26754
    },
    {
      "epoch": 8.448058099147458,
      "grad_norm": 6.652707099914551,
      "learning_rate": 3.208872750236817e-05,
      "loss": 0.3603,
      "step": 26755
    },
    {
      "epoch": 8.448373855383643,
      "grad_norm": 6.138591289520264,
      "learning_rate": 3.20878504017121e-05,
      "loss": 0.4058,
      "step": 26756
    },
    {
      "epoch": 8.44868961161983,
      "grad_norm": 5.97739839553833,
      "learning_rate": 3.208697330105603e-05,
      "loss": 0.4325,
      "step": 26757
    },
    {
      "epoch": 8.449005367856016,
      "grad_norm": 6.56978702545166,
      "learning_rate": 3.2086096200399965e-05,
      "loss": 0.4461,
      "step": 26758
    },
    {
      "epoch": 8.4493211240922,
      "grad_norm": 5.586521625518799,
      "learning_rate": 3.208521909974389e-05,
      "loss": 0.3329,
      "step": 26759
    },
    {
      "epoch": 8.449636880328386,
      "grad_norm": 7.246405601501465,
      "learning_rate": 3.208434199908782e-05,
      "loss": 0.386,
      "step": 26760
    },
    {
      "epoch": 8.449952636564571,
      "grad_norm": 6.565619468688965,
      "learning_rate": 3.208346489843175e-05,
      "loss": 0.4205,
      "step": 26761
    },
    {
      "epoch": 8.450268392800758,
      "grad_norm": 6.884802341461182,
      "learning_rate": 3.208258779777568e-05,
      "loss": 0.4039,
      "step": 26762
    },
    {
      "epoch": 8.450584149036944,
      "grad_norm": 7.31093168258667,
      "learning_rate": 3.2081710697119604e-05,
      "loss": 0.4695,
      "step": 26763
    },
    {
      "epoch": 8.450899905273129,
      "grad_norm": 6.595949172973633,
      "learning_rate": 3.2080833596463535e-05,
      "loss": 0.3833,
      "step": 26764
    },
    {
      "epoch": 8.451215661509314,
      "grad_norm": 8.60678768157959,
      "learning_rate": 3.207995649580746e-05,
      "loss": 0.3839,
      "step": 26765
    },
    {
      "epoch": 8.4515314177455,
      "grad_norm": 4.889914512634277,
      "learning_rate": 3.207907939515139e-05,
      "loss": 0.4016,
      "step": 26766
    },
    {
      "epoch": 8.451847173981687,
      "grad_norm": 4.554956912994385,
      "learning_rate": 3.207820229449532e-05,
      "loss": 0.3019,
      "step": 26767
    },
    {
      "epoch": 8.452162930217872,
      "grad_norm": 6.212258338928223,
      "learning_rate": 3.207732519383925e-05,
      "loss": 0.3752,
      "step": 26768
    },
    {
      "epoch": 8.452478686454057,
      "grad_norm": 7.499239444732666,
      "learning_rate": 3.207644809318317e-05,
      "loss": 0.4244,
      "step": 26769
    },
    {
      "epoch": 8.452794442690243,
      "grad_norm": 5.789496421813965,
      "learning_rate": 3.2075570992527105e-05,
      "loss": 0.408,
      "step": 26770
    },
    {
      "epoch": 8.453110198926428,
      "grad_norm": 4.485988616943359,
      "learning_rate": 3.207469389187103e-05,
      "loss": 0.2601,
      "step": 26771
    },
    {
      "epoch": 8.453425955162615,
      "grad_norm": 6.393985271453857,
      "learning_rate": 3.207381679121496e-05,
      "loss": 0.5537,
      "step": 26772
    },
    {
      "epoch": 8.4537417113988,
      "grad_norm": 5.58629846572876,
      "learning_rate": 3.2072939690558887e-05,
      "loss": 0.3739,
      "step": 26773
    },
    {
      "epoch": 8.454057467634986,
      "grad_norm": 6.6361494064331055,
      "learning_rate": 3.207206258990282e-05,
      "loss": 0.3733,
      "step": 26774
    },
    {
      "epoch": 8.45437322387117,
      "grad_norm": 5.589397430419922,
      "learning_rate": 3.207118548924675e-05,
      "loss": 0.4024,
      "step": 26775
    },
    {
      "epoch": 8.454688980107358,
      "grad_norm": 6.380389213562012,
      "learning_rate": 3.2070308388590675e-05,
      "loss": 0.4212,
      "step": 26776
    },
    {
      "epoch": 8.455004736343543,
      "grad_norm": 7.396324157714844,
      "learning_rate": 3.2069431287934606e-05,
      "loss": 0.387,
      "step": 26777
    },
    {
      "epoch": 8.455320492579729,
      "grad_norm": 5.625674247741699,
      "learning_rate": 3.206855418727853e-05,
      "loss": 0.3044,
      "step": 26778
    },
    {
      "epoch": 8.455636248815914,
      "grad_norm": 6.40407657623291,
      "learning_rate": 3.206767708662246e-05,
      "loss": 0.454,
      "step": 26779
    },
    {
      "epoch": 8.455952005052099,
      "grad_norm": 7.090536117553711,
      "learning_rate": 3.206679998596639e-05,
      "loss": 0.4534,
      "step": 26780
    },
    {
      "epoch": 8.456267761288286,
      "grad_norm": 7.682063579559326,
      "learning_rate": 3.206592288531032e-05,
      "loss": 0.4924,
      "step": 26781
    },
    {
      "epoch": 8.456583517524471,
      "grad_norm": 5.8010711669921875,
      "learning_rate": 3.2065045784654245e-05,
      "loss": 0.4945,
      "step": 26782
    },
    {
      "epoch": 8.456899273760657,
      "grad_norm": 6.461574554443359,
      "learning_rate": 3.2064168683998176e-05,
      "loss": 0.4043,
      "step": 26783
    },
    {
      "epoch": 8.457215029996842,
      "grad_norm": 13.604671478271484,
      "learning_rate": 3.20632915833421e-05,
      "loss": 0.3391,
      "step": 26784
    },
    {
      "epoch": 8.457530786233027,
      "grad_norm": 6.482261657714844,
      "learning_rate": 3.206241448268604e-05,
      "loss": 0.4738,
      "step": 26785
    },
    {
      "epoch": 8.457846542469214,
      "grad_norm": 7.636562347412109,
      "learning_rate": 3.2061537382029964e-05,
      "loss": 0.5428,
      "step": 26786
    },
    {
      "epoch": 8.4581622987054,
      "grad_norm": 7.213345527648926,
      "learning_rate": 3.2060660281373896e-05,
      "loss": 0.3963,
      "step": 26787
    },
    {
      "epoch": 8.458478054941585,
      "grad_norm": 7.388049602508545,
      "learning_rate": 3.205978318071782e-05,
      "loss": 0.3467,
      "step": 26788
    },
    {
      "epoch": 8.45879381117777,
      "grad_norm": 6.019649982452393,
      "learning_rate": 3.205890608006175e-05,
      "loss": 0.3977,
      "step": 26789
    },
    {
      "epoch": 8.459109567413956,
      "grad_norm": 7.4052653312683105,
      "learning_rate": 3.205802897940568e-05,
      "loss": 0.4565,
      "step": 26790
    },
    {
      "epoch": 8.459425323650143,
      "grad_norm": 8.349105834960938,
      "learning_rate": 3.205715187874961e-05,
      "loss": 0.4633,
      "step": 26791
    },
    {
      "epoch": 8.459741079886328,
      "grad_norm": 7.017873287200928,
      "learning_rate": 3.2056274778093534e-05,
      "loss": 0.4878,
      "step": 26792
    },
    {
      "epoch": 8.460056836122513,
      "grad_norm": 8.152048110961914,
      "learning_rate": 3.2055397677437466e-05,
      "loss": 0.511,
      "step": 26793
    },
    {
      "epoch": 8.460372592358699,
      "grad_norm": 4.52070951461792,
      "learning_rate": 3.205452057678139e-05,
      "loss": 0.4016,
      "step": 26794
    },
    {
      "epoch": 8.460688348594886,
      "grad_norm": 5.594266891479492,
      "learning_rate": 3.205364347612532e-05,
      "loss": 0.3423,
      "step": 26795
    },
    {
      "epoch": 8.461004104831071,
      "grad_norm": 7.450305938720703,
      "learning_rate": 3.2052766375469254e-05,
      "loss": 0.3962,
      "step": 26796
    },
    {
      "epoch": 8.461319861067256,
      "grad_norm": 5.852030277252197,
      "learning_rate": 3.205188927481318e-05,
      "loss": 0.3326,
      "step": 26797
    },
    {
      "epoch": 8.461635617303441,
      "grad_norm": 5.322719573974609,
      "learning_rate": 3.205101217415711e-05,
      "loss": 0.4535,
      "step": 26798
    },
    {
      "epoch": 8.461951373539627,
      "grad_norm": 5.921181678771973,
      "learning_rate": 3.2050135073501036e-05,
      "loss": 0.328,
      "step": 26799
    },
    {
      "epoch": 8.462267129775814,
      "grad_norm": 6.5212721824646,
      "learning_rate": 3.204925797284497e-05,
      "loss": 0.4549,
      "step": 26800
    },
    {
      "epoch": 8.462267129775814,
      "eval_loss": 0.9955306649208069,
      "eval_runtime": 4.5896,
      "eval_samples_per_second": 542.535,
      "eval_steps_per_second": 4.576,
      "step": 26800
    },
    {
      "epoch": 8.462582886012,
      "grad_norm": 7.819137096405029,
      "learning_rate": 3.204838087218889e-05,
      "loss": 0.3874,
      "step": 26801
    },
    {
      "epoch": 8.462898642248184,
      "grad_norm": 7.053864002227783,
      "learning_rate": 3.2047503771532824e-05,
      "loss": 0.4386,
      "step": 26802
    },
    {
      "epoch": 8.46321439848437,
      "grad_norm": 6.224266052246094,
      "learning_rate": 3.204662667087675e-05,
      "loss": 0.3746,
      "step": 26803
    },
    {
      "epoch": 8.463530154720555,
      "grad_norm": 7.066861152648926,
      "learning_rate": 3.204574957022068e-05,
      "loss": 0.5082,
      "step": 26804
    },
    {
      "epoch": 8.463845910956742,
      "grad_norm": 6.626469135284424,
      "learning_rate": 3.2044872469564605e-05,
      "loss": 0.3706,
      "step": 26805
    },
    {
      "epoch": 8.464161667192927,
      "grad_norm": 6.197044372558594,
      "learning_rate": 3.204399536890854e-05,
      "loss": 0.3665,
      "step": 26806
    },
    {
      "epoch": 8.464477423429113,
      "grad_norm": 7.510491847991943,
      "learning_rate": 3.204311826825246e-05,
      "loss": 0.417,
      "step": 26807
    },
    {
      "epoch": 8.464793179665298,
      "grad_norm": 5.19160270690918,
      "learning_rate": 3.2042241167596394e-05,
      "loss": 0.3429,
      "step": 26808
    },
    {
      "epoch": 8.465108935901483,
      "grad_norm": 8.435209274291992,
      "learning_rate": 3.204136406694032e-05,
      "loss": 0.5352,
      "step": 26809
    },
    {
      "epoch": 8.46542469213767,
      "grad_norm": 6.961883544921875,
      "learning_rate": 3.204048696628425e-05,
      "loss": 0.4576,
      "step": 26810
    },
    {
      "epoch": 8.465740448373856,
      "grad_norm": 4.746642112731934,
      "learning_rate": 3.2039609865628175e-05,
      "loss": 0.371,
      "step": 26811
    },
    {
      "epoch": 8.466056204610041,
      "grad_norm": 6.24619722366333,
      "learning_rate": 3.203873276497211e-05,
      "loss": 0.426,
      "step": 26812
    },
    {
      "epoch": 8.466371960846226,
      "grad_norm": 7.857232093811035,
      "learning_rate": 3.203785566431604e-05,
      "loss": 0.3542,
      "step": 26813
    },
    {
      "epoch": 8.466687717082412,
      "grad_norm": 6.565585136413574,
      "learning_rate": 3.203697856365997e-05,
      "loss": 0.5147,
      "step": 26814
    },
    {
      "epoch": 8.467003473318599,
      "grad_norm": 4.814857006072998,
      "learning_rate": 3.2036101463003895e-05,
      "loss": 0.3791,
      "step": 26815
    },
    {
      "epoch": 8.467319229554784,
      "grad_norm": 6.289206027984619,
      "learning_rate": 3.203522436234783e-05,
      "loss": 0.3834,
      "step": 26816
    },
    {
      "epoch": 8.46763498579097,
      "grad_norm": 6.764106750488281,
      "learning_rate": 3.203434726169175e-05,
      "loss": 0.4101,
      "step": 26817
    },
    {
      "epoch": 8.467950742027154,
      "grad_norm": 7.037728786468506,
      "learning_rate": 3.2033470161035683e-05,
      "loss": 0.4081,
      "step": 26818
    },
    {
      "epoch": 8.468266498263342,
      "grad_norm": 6.449165344238281,
      "learning_rate": 3.2032593060379615e-05,
      "loss": 0.4655,
      "step": 26819
    },
    {
      "epoch": 8.468582254499527,
      "grad_norm": 18.275705337524414,
      "learning_rate": 3.203171595972354e-05,
      "loss": 0.4428,
      "step": 26820
    },
    {
      "epoch": 8.468898010735712,
      "grad_norm": 6.3480377197265625,
      "learning_rate": 3.203083885906747e-05,
      "loss": 0.3898,
      "step": 26821
    },
    {
      "epoch": 8.469213766971897,
      "grad_norm": 6.4554362297058105,
      "learning_rate": 3.20299617584114e-05,
      "loss": 0.4022,
      "step": 26822
    },
    {
      "epoch": 8.469529523208083,
      "grad_norm": 5.382813930511475,
      "learning_rate": 3.202908465775533e-05,
      "loss": 0.3871,
      "step": 26823
    },
    {
      "epoch": 8.46984527944427,
      "grad_norm": 6.438390254974365,
      "learning_rate": 3.202820755709925e-05,
      "loss": 0.5304,
      "step": 26824
    },
    {
      "epoch": 8.470161035680455,
      "grad_norm": 5.527815818786621,
      "learning_rate": 3.2027330456443185e-05,
      "loss": 0.4778,
      "step": 26825
    },
    {
      "epoch": 8.47047679191664,
      "grad_norm": 5.606126308441162,
      "learning_rate": 3.202645335578711e-05,
      "loss": 0.3508,
      "step": 26826
    },
    {
      "epoch": 8.470792548152826,
      "grad_norm": 6.786189079284668,
      "learning_rate": 3.202557625513104e-05,
      "loss": 0.6006,
      "step": 26827
    },
    {
      "epoch": 8.471108304389011,
      "grad_norm": 6.629604339599609,
      "learning_rate": 3.2024699154474966e-05,
      "loss": 0.5145,
      "step": 26828
    },
    {
      "epoch": 8.471424060625198,
      "grad_norm": 6.195775032043457,
      "learning_rate": 3.20238220538189e-05,
      "loss": 0.494,
      "step": 26829
    },
    {
      "epoch": 8.471739816861383,
      "grad_norm": 6.004810810089111,
      "learning_rate": 3.202294495316282e-05,
      "loss": 0.4544,
      "step": 26830
    },
    {
      "epoch": 8.472055573097569,
      "grad_norm": 4.677374362945557,
      "learning_rate": 3.2022067852506755e-05,
      "loss": 0.4463,
      "step": 26831
    },
    {
      "epoch": 8.472371329333754,
      "grad_norm": 5.303272724151611,
      "learning_rate": 3.202119075185068e-05,
      "loss": 0.3369,
      "step": 26832
    },
    {
      "epoch": 8.47268708556994,
      "grad_norm": 4.873710632324219,
      "learning_rate": 3.202031365119461e-05,
      "loss": 0.434,
      "step": 26833
    },
    {
      "epoch": 8.473002841806126,
      "grad_norm": 6.521552085876465,
      "learning_rate": 3.201943655053854e-05,
      "loss": 0.3648,
      "step": 26834
    },
    {
      "epoch": 8.473318598042312,
      "grad_norm": 5.844810485839844,
      "learning_rate": 3.201855944988247e-05,
      "loss": 0.4719,
      "step": 26835
    },
    {
      "epoch": 8.473634354278497,
      "grad_norm": 6.359070301055908,
      "learning_rate": 3.20176823492264e-05,
      "loss": 0.4736,
      "step": 26836
    },
    {
      "epoch": 8.473950110514682,
      "grad_norm": 5.883760452270508,
      "learning_rate": 3.2016805248570324e-05,
      "loss": 0.3619,
      "step": 26837
    },
    {
      "epoch": 8.47426586675087,
      "grad_norm": 6.345056056976318,
      "learning_rate": 3.2015928147914256e-05,
      "loss": 0.3927,
      "step": 26838
    },
    {
      "epoch": 8.474581622987055,
      "grad_norm": 5.671914577484131,
      "learning_rate": 3.201505104725818e-05,
      "loss": 0.4126,
      "step": 26839
    },
    {
      "epoch": 8.47489737922324,
      "grad_norm": 4.744419097900391,
      "learning_rate": 3.201417394660212e-05,
      "loss": 0.2973,
      "step": 26840
    },
    {
      "epoch": 8.475213135459425,
      "grad_norm": 5.529316425323486,
      "learning_rate": 3.2013296845946044e-05,
      "loss": 0.4062,
      "step": 26841
    },
    {
      "epoch": 8.47552889169561,
      "grad_norm": 5.025306224822998,
      "learning_rate": 3.2012419745289976e-05,
      "loss": 0.3707,
      "step": 26842
    },
    {
      "epoch": 8.475844647931797,
      "grad_norm": 7.147665500640869,
      "learning_rate": 3.20115426446339e-05,
      "loss": 0.3937,
      "step": 26843
    },
    {
      "epoch": 8.476160404167983,
      "grad_norm": 4.973987579345703,
      "learning_rate": 3.201066554397783e-05,
      "loss": 0.4137,
      "step": 26844
    },
    {
      "epoch": 8.476476160404168,
      "grad_norm": 6.267867565155029,
      "learning_rate": 3.200978844332176e-05,
      "loss": 0.3272,
      "step": 26845
    },
    {
      "epoch": 8.476791916640353,
      "grad_norm": 5.783503532409668,
      "learning_rate": 3.200891134266569e-05,
      "loss": 0.4721,
      "step": 26846
    },
    {
      "epoch": 8.477107672876539,
      "grad_norm": 8.924861907958984,
      "learning_rate": 3.2008034242009614e-05,
      "loss": 0.4299,
      "step": 26847
    },
    {
      "epoch": 8.477423429112726,
      "grad_norm": 6.933912754058838,
      "learning_rate": 3.2007157141353546e-05,
      "loss": 0.4034,
      "step": 26848
    },
    {
      "epoch": 8.477739185348911,
      "grad_norm": 5.606997966766357,
      "learning_rate": 3.200628004069747e-05,
      "loss": 0.3872,
      "step": 26849
    },
    {
      "epoch": 8.478054941585096,
      "grad_norm": 7.3025431632995605,
      "learning_rate": 3.20054029400414e-05,
      "loss": 0.4747,
      "step": 26850
    },
    {
      "epoch": 8.478370697821282,
      "grad_norm": 6.296989917755127,
      "learning_rate": 3.200452583938533e-05,
      "loss": 0.4101,
      "step": 26851
    },
    {
      "epoch": 8.478686454057467,
      "grad_norm": 6.1062140464782715,
      "learning_rate": 3.200364873872926e-05,
      "loss": 0.3717,
      "step": 26852
    },
    {
      "epoch": 8.479002210293654,
      "grad_norm": 4.700623512268066,
      "learning_rate": 3.2002771638073184e-05,
      "loss": 0.2514,
      "step": 26853
    },
    {
      "epoch": 8.47931796652984,
      "grad_norm": 6.438424110412598,
      "learning_rate": 3.2001894537417116e-05,
      "loss": 0.5673,
      "step": 26854
    },
    {
      "epoch": 8.479633722766025,
      "grad_norm": 7.443210601806641,
      "learning_rate": 3.200101743676104e-05,
      "loss": 0.4348,
      "step": 26855
    },
    {
      "epoch": 8.47994947900221,
      "grad_norm": 6.752701759338379,
      "learning_rate": 3.200014033610497e-05,
      "loss": 0.3533,
      "step": 26856
    },
    {
      "epoch": 8.480265235238395,
      "grad_norm": 4.998376369476318,
      "learning_rate": 3.1999263235448904e-05,
      "loss": 0.3405,
      "step": 26857
    },
    {
      "epoch": 8.480580991474582,
      "grad_norm": 6.515280723571777,
      "learning_rate": 3.199838613479283e-05,
      "loss": 0.4444,
      "step": 26858
    },
    {
      "epoch": 8.480896747710768,
      "grad_norm": 6.780737400054932,
      "learning_rate": 3.199750903413676e-05,
      "loss": 0.3681,
      "step": 26859
    },
    {
      "epoch": 8.481212503946953,
      "grad_norm": 6.280923366546631,
      "learning_rate": 3.1996631933480685e-05,
      "loss": 0.4727,
      "step": 26860
    },
    {
      "epoch": 8.481528260183138,
      "grad_norm": 7.283169746398926,
      "learning_rate": 3.199575483282462e-05,
      "loss": 0.3689,
      "step": 26861
    },
    {
      "epoch": 8.481844016419323,
      "grad_norm": 7.904890537261963,
      "learning_rate": 3.199487773216854e-05,
      "loss": 0.3812,
      "step": 26862
    },
    {
      "epoch": 8.48215977265551,
      "grad_norm": 6.21160888671875,
      "learning_rate": 3.1994000631512474e-05,
      "loss": 0.4052,
      "step": 26863
    },
    {
      "epoch": 8.482475528891696,
      "grad_norm": 6.390812397003174,
      "learning_rate": 3.19931235308564e-05,
      "loss": 0.4518,
      "step": 26864
    },
    {
      "epoch": 8.482791285127881,
      "grad_norm": 8.528725624084473,
      "learning_rate": 3.199224643020033e-05,
      "loss": 0.545,
      "step": 26865
    },
    {
      "epoch": 8.483107041364066,
      "grad_norm": 8.003607749938965,
      "learning_rate": 3.1991369329544255e-05,
      "loss": 0.4468,
      "step": 26866
    },
    {
      "epoch": 8.483422797600253,
      "grad_norm": 6.301816940307617,
      "learning_rate": 3.1990492228888194e-05,
      "loss": 0.3736,
      "step": 26867
    },
    {
      "epoch": 8.483738553836439,
      "grad_norm": 5.586493492126465,
      "learning_rate": 3.198961512823212e-05,
      "loss": 0.385,
      "step": 26868
    },
    {
      "epoch": 8.484054310072624,
      "grad_norm": 6.9058051109313965,
      "learning_rate": 3.198873802757605e-05,
      "loss": 0.4545,
      "step": 26869
    },
    {
      "epoch": 8.48437006630881,
      "grad_norm": 5.346429824829102,
      "learning_rate": 3.1987860926919975e-05,
      "loss": 0.3818,
      "step": 26870
    },
    {
      "epoch": 8.484685822544995,
      "grad_norm": 6.836484432220459,
      "learning_rate": 3.198698382626391e-05,
      "loss": 0.4457,
      "step": 26871
    },
    {
      "epoch": 8.485001578781182,
      "grad_norm": 5.801908493041992,
      "learning_rate": 3.198610672560783e-05,
      "loss": 0.3984,
      "step": 26872
    },
    {
      "epoch": 8.485317335017367,
      "grad_norm": 6.105436325073242,
      "learning_rate": 3.198522962495176e-05,
      "loss": 0.4242,
      "step": 26873
    },
    {
      "epoch": 8.485633091253552,
      "grad_norm": 4.733102321624756,
      "learning_rate": 3.198435252429569e-05,
      "loss": 0.4352,
      "step": 26874
    },
    {
      "epoch": 8.485948847489738,
      "grad_norm": 6.055638313293457,
      "learning_rate": 3.198347542363962e-05,
      "loss": 0.4146,
      "step": 26875
    },
    {
      "epoch": 8.486264603725923,
      "grad_norm": 4.555523872375488,
      "learning_rate": 3.1982598322983545e-05,
      "loss": 0.3427,
      "step": 26876
    },
    {
      "epoch": 8.48658035996211,
      "grad_norm": 5.482748031616211,
      "learning_rate": 3.1981721222327477e-05,
      "loss": 0.4476,
      "step": 26877
    },
    {
      "epoch": 8.486896116198295,
      "grad_norm": 6.009481430053711,
      "learning_rate": 3.198084412167141e-05,
      "loss": 0.3964,
      "step": 26878
    },
    {
      "epoch": 8.48721187243448,
      "grad_norm": 6.441323280334473,
      "learning_rate": 3.197996702101533e-05,
      "loss": 0.3752,
      "step": 26879
    },
    {
      "epoch": 8.487527628670666,
      "grad_norm": 6.815606594085693,
      "learning_rate": 3.1979089920359265e-05,
      "loss": 0.4992,
      "step": 26880
    },
    {
      "epoch": 8.487843384906851,
      "grad_norm": 5.040600776672363,
      "learning_rate": 3.197821281970319e-05,
      "loss": 0.4047,
      "step": 26881
    },
    {
      "epoch": 8.488159141143038,
      "grad_norm": 5.367273807525635,
      "learning_rate": 3.197733571904712e-05,
      "loss": 0.3799,
      "step": 26882
    },
    {
      "epoch": 8.488474897379223,
      "grad_norm": 6.421723365783691,
      "learning_rate": 3.1976458618391046e-05,
      "loss": 0.4682,
      "step": 26883
    },
    {
      "epoch": 8.488790653615409,
      "grad_norm": 6.364774703979492,
      "learning_rate": 3.197558151773498e-05,
      "loss": 0.3435,
      "step": 26884
    },
    {
      "epoch": 8.489106409851594,
      "grad_norm": 6.4030866622924805,
      "learning_rate": 3.19747044170789e-05,
      "loss": 0.4004,
      "step": 26885
    },
    {
      "epoch": 8.489422166087781,
      "grad_norm": 6.354109287261963,
      "learning_rate": 3.1973827316422835e-05,
      "loss": 0.4453,
      "step": 26886
    },
    {
      "epoch": 8.489737922323966,
      "grad_norm": 5.666796684265137,
      "learning_rate": 3.197295021576676e-05,
      "loss": 0.3157,
      "step": 26887
    },
    {
      "epoch": 8.490053678560152,
      "grad_norm": 8.539009094238281,
      "learning_rate": 3.197207311511069e-05,
      "loss": 0.3746,
      "step": 26888
    },
    {
      "epoch": 8.490369434796337,
      "grad_norm": 9.38416576385498,
      "learning_rate": 3.1971196014454616e-05,
      "loss": 0.5501,
      "step": 26889
    },
    {
      "epoch": 8.490685191032522,
      "grad_norm": 5.795144557952881,
      "learning_rate": 3.197031891379855e-05,
      "loss": 0.3978,
      "step": 26890
    },
    {
      "epoch": 8.49100094726871,
      "grad_norm": 6.267282485961914,
      "learning_rate": 3.196944181314247e-05,
      "loss": 0.3415,
      "step": 26891
    },
    {
      "epoch": 8.491316703504895,
      "grad_norm": 9.898223876953125,
      "learning_rate": 3.1968564712486404e-05,
      "loss": 0.4859,
      "step": 26892
    },
    {
      "epoch": 8.49163245974108,
      "grad_norm": 7.072644233703613,
      "learning_rate": 3.196768761183033e-05,
      "loss": 0.405,
      "step": 26893
    },
    {
      "epoch": 8.491948215977265,
      "grad_norm": 8.318324089050293,
      "learning_rate": 3.196681051117427e-05,
      "loss": 0.5231,
      "step": 26894
    },
    {
      "epoch": 8.49226397221345,
      "grad_norm": 6.123703479766846,
      "learning_rate": 3.196593341051819e-05,
      "loss": 0.4646,
      "step": 26895
    },
    {
      "epoch": 8.492579728449638,
      "grad_norm": 5.891982078552246,
      "learning_rate": 3.1965056309862124e-05,
      "loss": 0.5016,
      "step": 26896
    },
    {
      "epoch": 8.492895484685823,
      "grad_norm": 6.106083393096924,
      "learning_rate": 3.196417920920605e-05,
      "loss": 0.388,
      "step": 26897
    },
    {
      "epoch": 8.493211240922008,
      "grad_norm": 5.726653575897217,
      "learning_rate": 3.196330210854998e-05,
      "loss": 0.4406,
      "step": 26898
    },
    {
      "epoch": 8.493526997158193,
      "grad_norm": 5.017943859100342,
      "learning_rate": 3.1962425007893906e-05,
      "loss": 0.3379,
      "step": 26899
    },
    {
      "epoch": 8.493842753394379,
      "grad_norm": 7.511658191680908,
      "learning_rate": 3.196154790723784e-05,
      "loss": 0.4595,
      "step": 26900
    },
    {
      "epoch": 8.494158509630566,
      "grad_norm": 6.671798229217529,
      "learning_rate": 3.196067080658177e-05,
      "loss": 0.4849,
      "step": 26901
    },
    {
      "epoch": 8.494474265866751,
      "grad_norm": 5.960197448730469,
      "learning_rate": 3.1959793705925694e-05,
      "loss": 0.3415,
      "step": 26902
    },
    {
      "epoch": 8.494790022102936,
      "grad_norm": 6.289115905761719,
      "learning_rate": 3.1958916605269626e-05,
      "loss": 0.4544,
      "step": 26903
    },
    {
      "epoch": 8.495105778339122,
      "grad_norm": 6.3165483474731445,
      "learning_rate": 3.195803950461355e-05,
      "loss": 0.4215,
      "step": 26904
    },
    {
      "epoch": 8.495421534575307,
      "grad_norm": 6.421543598175049,
      "learning_rate": 3.195716240395748e-05,
      "loss": 0.3954,
      "step": 26905
    },
    {
      "epoch": 8.495737290811494,
      "grad_norm": 5.130736827850342,
      "learning_rate": 3.195628530330141e-05,
      "loss": 0.4778,
      "step": 26906
    },
    {
      "epoch": 8.49605304704768,
      "grad_norm": 5.735800266265869,
      "learning_rate": 3.195540820264534e-05,
      "loss": 0.4673,
      "step": 26907
    },
    {
      "epoch": 8.496368803283865,
      "grad_norm": 5.6314568519592285,
      "learning_rate": 3.1954531101989264e-05,
      "loss": 0.3957,
      "step": 26908
    },
    {
      "epoch": 8.49668455952005,
      "grad_norm": 8.007589340209961,
      "learning_rate": 3.1953654001333195e-05,
      "loss": 0.3742,
      "step": 26909
    },
    {
      "epoch": 8.497000315756237,
      "grad_norm": 6.445249080657959,
      "learning_rate": 3.195277690067712e-05,
      "loss": 0.5674,
      "step": 26910
    },
    {
      "epoch": 8.497316071992422,
      "grad_norm": 5.299530982971191,
      "learning_rate": 3.195189980002105e-05,
      "loss": 0.3991,
      "step": 26911
    },
    {
      "epoch": 8.497631828228608,
      "grad_norm": 5.555357933044434,
      "learning_rate": 3.195102269936498e-05,
      "loss": 0.3375,
      "step": 26912
    },
    {
      "epoch": 8.497947584464793,
      "grad_norm": 6.933319091796875,
      "learning_rate": 3.195014559870891e-05,
      "loss": 0.4541,
      "step": 26913
    },
    {
      "epoch": 8.498263340700978,
      "grad_norm": 5.786975383758545,
      "learning_rate": 3.1949268498052834e-05,
      "loss": 0.4363,
      "step": 26914
    },
    {
      "epoch": 8.498579096937165,
      "grad_norm": 6.617677688598633,
      "learning_rate": 3.1948391397396765e-05,
      "loss": 0.4284,
      "step": 26915
    },
    {
      "epoch": 8.49889485317335,
      "grad_norm": 5.800714015960693,
      "learning_rate": 3.19475142967407e-05,
      "loss": 0.3774,
      "step": 26916
    },
    {
      "epoch": 8.499210609409536,
      "grad_norm": 5.134852409362793,
      "learning_rate": 3.194663719608462e-05,
      "loss": 0.4139,
      "step": 26917
    },
    {
      "epoch": 8.499526365645721,
      "grad_norm": 5.396790027618408,
      "learning_rate": 3.1945760095428554e-05,
      "loss": 0.4108,
      "step": 26918
    },
    {
      "epoch": 8.499842121881906,
      "grad_norm": 4.607370853424072,
      "learning_rate": 3.194488299477248e-05,
      "loss": 0.3973,
      "step": 26919
    },
    {
      "epoch": 8.500157878118094,
      "grad_norm": 6.748005390167236,
      "learning_rate": 3.194400589411641e-05,
      "loss": 0.4765,
      "step": 26920
    },
    {
      "epoch": 8.500473634354279,
      "grad_norm": 5.273657321929932,
      "learning_rate": 3.1943128793460335e-05,
      "loss": 0.375,
      "step": 26921
    },
    {
      "epoch": 8.500789390590464,
      "grad_norm": 6.565196990966797,
      "learning_rate": 3.1942251692804273e-05,
      "loss": 0.422,
      "step": 26922
    },
    {
      "epoch": 8.50110514682665,
      "grad_norm": 5.596762180328369,
      "learning_rate": 3.19413745921482e-05,
      "loss": 0.4226,
      "step": 26923
    },
    {
      "epoch": 8.501420903062835,
      "grad_norm": 9.018922805786133,
      "learning_rate": 3.194049749149213e-05,
      "loss": 0.416,
      "step": 26924
    },
    {
      "epoch": 8.501736659299022,
      "grad_norm": 6.138383388519287,
      "learning_rate": 3.1939620390836055e-05,
      "loss": 0.4219,
      "step": 26925
    },
    {
      "epoch": 8.502052415535207,
      "grad_norm": 5.7220025062561035,
      "learning_rate": 3.193874329017999e-05,
      "loss": 0.3909,
      "step": 26926
    },
    {
      "epoch": 8.502368171771392,
      "grad_norm": 6.521768093109131,
      "learning_rate": 3.193786618952391e-05,
      "loss": 0.3815,
      "step": 26927
    },
    {
      "epoch": 8.502683928007578,
      "grad_norm": 7.582303524017334,
      "learning_rate": 3.193698908886784e-05,
      "loss": 0.4745,
      "step": 26928
    },
    {
      "epoch": 8.502999684243765,
      "grad_norm": 9.159095764160156,
      "learning_rate": 3.193611198821177e-05,
      "loss": 0.461,
      "step": 26929
    },
    {
      "epoch": 8.50331544047995,
      "grad_norm": 5.341193675994873,
      "learning_rate": 3.19352348875557e-05,
      "loss": 0.4226,
      "step": 26930
    },
    {
      "epoch": 8.503631196716135,
      "grad_norm": 5.835306167602539,
      "learning_rate": 3.1934357786899625e-05,
      "loss": 0.4411,
      "step": 26931
    },
    {
      "epoch": 8.50394695295232,
      "grad_norm": 5.777665615081787,
      "learning_rate": 3.1933480686243556e-05,
      "loss": 0.3649,
      "step": 26932
    },
    {
      "epoch": 8.504262709188506,
      "grad_norm": 5.860259532928467,
      "learning_rate": 3.193260358558748e-05,
      "loss": 0.3094,
      "step": 26933
    },
    {
      "epoch": 8.504578465424693,
      "grad_norm": 6.423087120056152,
      "learning_rate": 3.193172648493141e-05,
      "loss": 0.4558,
      "step": 26934
    },
    {
      "epoch": 8.504894221660878,
      "grad_norm": 5.134546279907227,
      "learning_rate": 3.193084938427534e-05,
      "loss": 0.3264,
      "step": 26935
    },
    {
      "epoch": 8.505209977897064,
      "grad_norm": 6.485703945159912,
      "learning_rate": 3.192997228361927e-05,
      "loss": 0.3216,
      "step": 26936
    },
    {
      "epoch": 8.505525734133249,
      "grad_norm": 5.465254306793213,
      "learning_rate": 3.19290951829632e-05,
      "loss": 0.3616,
      "step": 26937
    },
    {
      "epoch": 8.505841490369434,
      "grad_norm": 6.160281181335449,
      "learning_rate": 3.1928218082307126e-05,
      "loss": 0.4633,
      "step": 26938
    },
    {
      "epoch": 8.506157246605621,
      "grad_norm": 6.348958969116211,
      "learning_rate": 3.192734098165106e-05,
      "loss": 0.4394,
      "step": 26939
    },
    {
      "epoch": 8.506473002841807,
      "grad_norm": 6.001226425170898,
      "learning_rate": 3.192646388099498e-05,
      "loss": 0.3989,
      "step": 26940
    },
    {
      "epoch": 8.506788759077992,
      "grad_norm": 5.02800989151001,
      "learning_rate": 3.1925586780338914e-05,
      "loss": 0.4572,
      "step": 26941
    },
    {
      "epoch": 8.507104515314177,
      "grad_norm": 5.077153205871582,
      "learning_rate": 3.192470967968284e-05,
      "loss": 0.3434,
      "step": 26942
    },
    {
      "epoch": 8.507420271550362,
      "grad_norm": 5.653006076812744,
      "learning_rate": 3.192383257902677e-05,
      "loss": 0.3352,
      "step": 26943
    },
    {
      "epoch": 8.50773602778655,
      "grad_norm": 7.172675132751465,
      "learning_rate": 3.1922955478370696e-05,
      "loss": 0.477,
      "step": 26944
    },
    {
      "epoch": 8.508051784022735,
      "grad_norm": 7.2850117683410645,
      "learning_rate": 3.192207837771463e-05,
      "loss": 0.4219,
      "step": 26945
    },
    {
      "epoch": 8.50836754025892,
      "grad_norm": 6.636894702911377,
      "learning_rate": 3.192120127705855e-05,
      "loss": 0.4697,
      "step": 26946
    },
    {
      "epoch": 8.508683296495105,
      "grad_norm": 6.592832088470459,
      "learning_rate": 3.1920324176402484e-05,
      "loss": 0.4583,
      "step": 26947
    },
    {
      "epoch": 8.50899905273129,
      "grad_norm": 5.856575965881348,
      "learning_rate": 3.191944707574641e-05,
      "loss": 0.4257,
      "step": 26948
    },
    {
      "epoch": 8.509314808967478,
      "grad_norm": 7.735716819763184,
      "learning_rate": 3.191856997509035e-05,
      "loss": 0.435,
      "step": 26949
    },
    {
      "epoch": 8.509630565203663,
      "grad_norm": 5.412051677703857,
      "learning_rate": 3.191769287443427e-05,
      "loss": 0.4478,
      "step": 26950
    },
    {
      "epoch": 8.509946321439848,
      "grad_norm": 6.032618045806885,
      "learning_rate": 3.1916815773778204e-05,
      "loss": 0.3599,
      "step": 26951
    },
    {
      "epoch": 8.510262077676034,
      "grad_norm": 4.8339056968688965,
      "learning_rate": 3.191593867312213e-05,
      "loss": 0.3445,
      "step": 26952
    },
    {
      "epoch": 8.510577833912219,
      "grad_norm": 5.865135192871094,
      "learning_rate": 3.191506157246606e-05,
      "loss": 0.4301,
      "step": 26953
    },
    {
      "epoch": 8.510893590148406,
      "grad_norm": 8.377622604370117,
      "learning_rate": 3.1914184471809986e-05,
      "loss": 0.4258,
      "step": 26954
    },
    {
      "epoch": 8.511209346384591,
      "grad_norm": 5.653964042663574,
      "learning_rate": 3.191330737115392e-05,
      "loss": 0.4786,
      "step": 26955
    },
    {
      "epoch": 8.511525102620777,
      "grad_norm": 5.587865352630615,
      "learning_rate": 3.191243027049784e-05,
      "loss": 0.3133,
      "step": 26956
    },
    {
      "epoch": 8.511840858856962,
      "grad_norm": 8.0905122756958,
      "learning_rate": 3.1911553169841774e-05,
      "loss": 0.4854,
      "step": 26957
    },
    {
      "epoch": 8.512156615093149,
      "grad_norm": 6.50285005569458,
      "learning_rate": 3.19106760691857e-05,
      "loss": 0.3544,
      "step": 26958
    },
    {
      "epoch": 8.512472371329334,
      "grad_norm": 5.404611110687256,
      "learning_rate": 3.190979896852963e-05,
      "loss": 0.3822,
      "step": 26959
    },
    {
      "epoch": 8.51278812756552,
      "grad_norm": 5.864920139312744,
      "learning_rate": 3.190892186787356e-05,
      "loss": 0.4506,
      "step": 26960
    },
    {
      "epoch": 8.513103883801705,
      "grad_norm": 6.12653923034668,
      "learning_rate": 3.190804476721749e-05,
      "loss": 0.3983,
      "step": 26961
    },
    {
      "epoch": 8.51341964003789,
      "grad_norm": 5.889952659606934,
      "learning_rate": 3.190716766656142e-05,
      "loss": 0.3918,
      "step": 26962
    },
    {
      "epoch": 8.513735396274077,
      "grad_norm": 6.290416717529297,
      "learning_rate": 3.1906290565905344e-05,
      "loss": 0.4001,
      "step": 26963
    },
    {
      "epoch": 8.514051152510262,
      "grad_norm": 6.920464515686035,
      "learning_rate": 3.1905413465249275e-05,
      "loss": 0.416,
      "step": 26964
    },
    {
      "epoch": 8.514366908746448,
      "grad_norm": 6.904560089111328,
      "learning_rate": 3.19045363645932e-05,
      "loss": 0.4519,
      "step": 26965
    },
    {
      "epoch": 8.514682664982633,
      "grad_norm": 8.057205200195312,
      "learning_rate": 3.190365926393713e-05,
      "loss": 0.4429,
      "step": 26966
    },
    {
      "epoch": 8.514998421218818,
      "grad_norm": 4.5249786376953125,
      "learning_rate": 3.190278216328106e-05,
      "loss": 0.3246,
      "step": 26967
    },
    {
      "epoch": 8.515314177455005,
      "grad_norm": 7.219845294952393,
      "learning_rate": 3.190190506262499e-05,
      "loss": 0.4555,
      "step": 26968
    },
    {
      "epoch": 8.51562993369119,
      "grad_norm": 6.783864498138428,
      "learning_rate": 3.1901027961968913e-05,
      "loss": 0.4846,
      "step": 26969
    },
    {
      "epoch": 8.515945689927376,
      "grad_norm": 7.254941940307617,
      "learning_rate": 3.1900150861312845e-05,
      "loss": 0.3961,
      "step": 26970
    },
    {
      "epoch": 8.516261446163561,
      "grad_norm": 6.420502662658691,
      "learning_rate": 3.189927376065677e-05,
      "loss": 0.3411,
      "step": 26971
    },
    {
      "epoch": 8.516577202399747,
      "grad_norm": 7.477523326873779,
      "learning_rate": 3.18983966600007e-05,
      "loss": 0.4899,
      "step": 26972
    },
    {
      "epoch": 8.516892958635934,
      "grad_norm": 5.014457702636719,
      "learning_rate": 3.189751955934463e-05,
      "loss": 0.3654,
      "step": 26973
    },
    {
      "epoch": 8.517208714872119,
      "grad_norm": 4.937280178070068,
      "learning_rate": 3.189664245868856e-05,
      "loss": 0.3409,
      "step": 26974
    },
    {
      "epoch": 8.517524471108304,
      "grad_norm": 6.091954708099365,
      "learning_rate": 3.189576535803249e-05,
      "loss": 0.4595,
      "step": 26975
    },
    {
      "epoch": 8.51784022734449,
      "grad_norm": 9.409509658813477,
      "learning_rate": 3.189488825737642e-05,
      "loss": 0.4777,
      "step": 26976
    },
    {
      "epoch": 8.518155983580677,
      "grad_norm": 6.510448455810547,
      "learning_rate": 3.1894011156720347e-05,
      "loss": 0.4633,
      "step": 26977
    },
    {
      "epoch": 8.518471739816862,
      "grad_norm": 5.227116107940674,
      "learning_rate": 3.189313405606428e-05,
      "loss": 0.4595,
      "step": 26978
    },
    {
      "epoch": 8.518787496053047,
      "grad_norm": 6.233577728271484,
      "learning_rate": 3.18922569554082e-05,
      "loss": 0.4152,
      "step": 26979
    },
    {
      "epoch": 8.519103252289232,
      "grad_norm": 5.9908366203308105,
      "learning_rate": 3.1891379854752135e-05,
      "loss": 0.4134,
      "step": 26980
    },
    {
      "epoch": 8.519419008525418,
      "grad_norm": 5.711000919342041,
      "learning_rate": 3.1890502754096067e-05,
      "loss": 0.4139,
      "step": 26981
    },
    {
      "epoch": 8.519734764761605,
      "grad_norm": 5.521933555603027,
      "learning_rate": 3.188962565343999e-05,
      "loss": 0.4182,
      "step": 26982
    },
    {
      "epoch": 8.52005052099779,
      "grad_norm": 10.517498970031738,
      "learning_rate": 3.188874855278392e-05,
      "loss": 0.3753,
      "step": 26983
    },
    {
      "epoch": 8.520366277233975,
      "grad_norm": 4.798395156860352,
      "learning_rate": 3.188787145212785e-05,
      "loss": 0.3038,
      "step": 26984
    },
    {
      "epoch": 8.52068203347016,
      "grad_norm": 5.137973785400391,
      "learning_rate": 3.188699435147178e-05,
      "loss": 0.256,
      "step": 26985
    },
    {
      "epoch": 8.520997789706346,
      "grad_norm": 4.675364017486572,
      "learning_rate": 3.1886117250815705e-05,
      "loss": 0.3526,
      "step": 26986
    },
    {
      "epoch": 8.521313545942533,
      "grad_norm": 4.5440449714660645,
      "learning_rate": 3.1885240150159636e-05,
      "loss": 0.4017,
      "step": 26987
    },
    {
      "epoch": 8.521629302178718,
      "grad_norm": 6.392312049865723,
      "learning_rate": 3.188436304950356e-05,
      "loss": 0.5533,
      "step": 26988
    },
    {
      "epoch": 8.521945058414904,
      "grad_norm": 5.689052104949951,
      "learning_rate": 3.188348594884749e-05,
      "loss": 0.5505,
      "step": 26989
    },
    {
      "epoch": 8.522260814651089,
      "grad_norm": 5.442564487457275,
      "learning_rate": 3.188260884819142e-05,
      "loss": 0.3977,
      "step": 26990
    },
    {
      "epoch": 8.522576570887274,
      "grad_norm": 6.0533599853515625,
      "learning_rate": 3.188173174753535e-05,
      "loss": 0.4049,
      "step": 26991
    },
    {
      "epoch": 8.522892327123461,
      "grad_norm": 5.197484970092773,
      "learning_rate": 3.1880854646879274e-05,
      "loss": 0.3933,
      "step": 26992
    },
    {
      "epoch": 8.523208083359647,
      "grad_norm": 7.796316623687744,
      "learning_rate": 3.1879977546223206e-05,
      "loss": 0.3917,
      "step": 26993
    },
    {
      "epoch": 8.523523839595832,
      "grad_norm": 7.257989883422852,
      "learning_rate": 3.187910044556713e-05,
      "loss": 0.5683,
      "step": 26994
    },
    {
      "epoch": 8.523839595832017,
      "grad_norm": 5.042630195617676,
      "learning_rate": 3.187822334491106e-05,
      "loss": 0.3703,
      "step": 26995
    },
    {
      "epoch": 8.524155352068203,
      "grad_norm": 5.309036731719971,
      "learning_rate": 3.187734624425499e-05,
      "loss": 0.3379,
      "step": 26996
    },
    {
      "epoch": 8.52447110830439,
      "grad_norm": 7.101490497589111,
      "learning_rate": 3.187646914359892e-05,
      "loss": 0.4522,
      "step": 26997
    },
    {
      "epoch": 8.524786864540575,
      "grad_norm": 5.8307600021362305,
      "learning_rate": 3.187559204294285e-05,
      "loss": 0.5124,
      "step": 26998
    },
    {
      "epoch": 8.52510262077676,
      "grad_norm": 5.072119235992432,
      "learning_rate": 3.1874714942286776e-05,
      "loss": 0.4034,
      "step": 26999
    },
    {
      "epoch": 8.525418377012945,
      "grad_norm": 11.718917846679688,
      "learning_rate": 3.187383784163071e-05,
      "loss": 0.4358,
      "step": 27000
    },
    {
      "epoch": 8.525418377012945,
      "eval_loss": 0.9763853549957275,
      "eval_runtime": 4.5934,
      "eval_samples_per_second": 542.082,
      "eval_steps_per_second": 4.572,
      "step": 27000
    },
    {
      "epoch": 8.52573413324913,
      "grad_norm": 4.546372890472412,
      "learning_rate": 3.187296074097463e-05,
      "loss": 0.3735,
      "step": 27001
    },
    {
      "epoch": 8.526049889485318,
      "grad_norm": 6.794504642486572,
      "learning_rate": 3.1872083640318564e-05,
      "loss": 0.4544,
      "step": 27002
    },
    {
      "epoch": 8.526365645721503,
      "grad_norm": 5.676342964172363,
      "learning_rate": 3.1871206539662496e-05,
      "loss": 0.4399,
      "step": 27003
    },
    {
      "epoch": 8.526681401957688,
      "grad_norm": 5.649836540222168,
      "learning_rate": 3.187032943900643e-05,
      "loss": 0.4726,
      "step": 27004
    },
    {
      "epoch": 8.526997158193874,
      "grad_norm": 7.2005085945129395,
      "learning_rate": 3.186945233835035e-05,
      "loss": 0.4073,
      "step": 27005
    },
    {
      "epoch": 8.52731291443006,
      "grad_norm": 6.509398937225342,
      "learning_rate": 3.1868575237694284e-05,
      "loss": 0.3976,
      "step": 27006
    },
    {
      "epoch": 8.527628670666246,
      "grad_norm": 6.624650478363037,
      "learning_rate": 3.186769813703821e-05,
      "loss": 0.3985,
      "step": 27007
    },
    {
      "epoch": 8.527944426902431,
      "grad_norm": 5.122860908508301,
      "learning_rate": 3.186682103638214e-05,
      "loss": 0.4082,
      "step": 27008
    },
    {
      "epoch": 8.528260183138617,
      "grad_norm": 7.910267353057861,
      "learning_rate": 3.1865943935726066e-05,
      "loss": 0.4774,
      "step": 27009
    },
    {
      "epoch": 8.528575939374802,
      "grad_norm": 5.403974533081055,
      "learning_rate": 3.186506683507e-05,
      "loss": 0.408,
      "step": 27010
    },
    {
      "epoch": 8.528891695610989,
      "grad_norm": 6.463896751403809,
      "learning_rate": 3.186418973441392e-05,
      "loss": 0.482,
      "step": 27011
    },
    {
      "epoch": 8.529207451847174,
      "grad_norm": 7.057119846343994,
      "learning_rate": 3.1863312633757854e-05,
      "loss": 0.3369,
      "step": 27012
    },
    {
      "epoch": 8.52952320808336,
      "grad_norm": 5.602597713470459,
      "learning_rate": 3.186243553310178e-05,
      "loss": 0.4649,
      "step": 27013
    },
    {
      "epoch": 8.529838964319545,
      "grad_norm": 6.029576778411865,
      "learning_rate": 3.186155843244571e-05,
      "loss": 0.476,
      "step": 27014
    },
    {
      "epoch": 8.53015472055573,
      "grad_norm": 5.1142072677612305,
      "learning_rate": 3.1860681331789635e-05,
      "loss": 0.3628,
      "step": 27015
    },
    {
      "epoch": 8.530470476791917,
      "grad_norm": 5.15899658203125,
      "learning_rate": 3.185980423113357e-05,
      "loss": 0.3755,
      "step": 27016
    },
    {
      "epoch": 8.530786233028103,
      "grad_norm": 6.6963419914245605,
      "learning_rate": 3.185892713047749e-05,
      "loss": 0.4049,
      "step": 27017
    },
    {
      "epoch": 8.531101989264288,
      "grad_norm": 4.657416343688965,
      "learning_rate": 3.1858050029821424e-05,
      "loss": 0.2765,
      "step": 27018
    },
    {
      "epoch": 8.531417745500473,
      "grad_norm": 7.715137481689453,
      "learning_rate": 3.1857172929165355e-05,
      "loss": 0.3833,
      "step": 27019
    },
    {
      "epoch": 8.53173350173666,
      "grad_norm": 5.553157329559326,
      "learning_rate": 3.185629582850928e-05,
      "loss": 0.3557,
      "step": 27020
    },
    {
      "epoch": 8.532049257972846,
      "grad_norm": 7.4745025634765625,
      "learning_rate": 3.185541872785321e-05,
      "loss": 0.4478,
      "step": 27021
    },
    {
      "epoch": 8.53236501420903,
      "grad_norm": 7.961978912353516,
      "learning_rate": 3.185454162719714e-05,
      "loss": 0.578,
      "step": 27022
    },
    {
      "epoch": 8.532680770445216,
      "grad_norm": 5.729761600494385,
      "learning_rate": 3.185366452654107e-05,
      "loss": 0.2959,
      "step": 27023
    },
    {
      "epoch": 8.532996526681401,
      "grad_norm": 7.51733922958374,
      "learning_rate": 3.185278742588499e-05,
      "loss": 0.4744,
      "step": 27024
    },
    {
      "epoch": 8.533312282917588,
      "grad_norm": 5.378239631652832,
      "learning_rate": 3.1851910325228925e-05,
      "loss": 0.4534,
      "step": 27025
    },
    {
      "epoch": 8.533628039153774,
      "grad_norm": 6.180099010467529,
      "learning_rate": 3.185103322457285e-05,
      "loss": 0.3757,
      "step": 27026
    },
    {
      "epoch": 8.533943795389959,
      "grad_norm": 5.35819673538208,
      "learning_rate": 3.185015612391678e-05,
      "loss": 0.4637,
      "step": 27027
    },
    {
      "epoch": 8.534259551626144,
      "grad_norm": 6.9206132888793945,
      "learning_rate": 3.1849279023260707e-05,
      "loss": 0.4866,
      "step": 27028
    },
    {
      "epoch": 8.53457530786233,
      "grad_norm": 8.849153518676758,
      "learning_rate": 3.184840192260464e-05,
      "loss": 0.4799,
      "step": 27029
    },
    {
      "epoch": 8.534891064098517,
      "grad_norm": 6.89426326751709,
      "learning_rate": 3.184752482194856e-05,
      "loss": 0.5203,
      "step": 27030
    },
    {
      "epoch": 8.535206820334702,
      "grad_norm": 4.853050231933594,
      "learning_rate": 3.18466477212925e-05,
      "loss": 0.2911,
      "step": 27031
    },
    {
      "epoch": 8.535522576570887,
      "grad_norm": 5.4326934814453125,
      "learning_rate": 3.1845770620636426e-05,
      "loss": 0.423,
      "step": 27032
    },
    {
      "epoch": 8.535838332807073,
      "grad_norm": 5.378769874572754,
      "learning_rate": 3.184489351998036e-05,
      "loss": 0.3903,
      "step": 27033
    },
    {
      "epoch": 8.536154089043258,
      "grad_norm": 4.379167079925537,
      "learning_rate": 3.184401641932428e-05,
      "loss": 0.342,
      "step": 27034
    },
    {
      "epoch": 8.536469845279445,
      "grad_norm": 6.019814968109131,
      "learning_rate": 3.1843139318668215e-05,
      "loss": 0.3329,
      "step": 27035
    },
    {
      "epoch": 8.53678560151563,
      "grad_norm": 6.013745307922363,
      "learning_rate": 3.184226221801214e-05,
      "loss": 0.4221,
      "step": 27036
    },
    {
      "epoch": 8.537101357751816,
      "grad_norm": 9.154380798339844,
      "learning_rate": 3.184138511735607e-05,
      "loss": 0.4444,
      "step": 27037
    },
    {
      "epoch": 8.537417113988,
      "grad_norm": 5.5310959815979,
      "learning_rate": 3.1840508016699996e-05,
      "loss": 0.422,
      "step": 27038
    },
    {
      "epoch": 8.537732870224186,
      "grad_norm": 6.284863471984863,
      "learning_rate": 3.183963091604393e-05,
      "loss": 0.4255,
      "step": 27039
    },
    {
      "epoch": 8.538048626460373,
      "grad_norm": 6.37358283996582,
      "learning_rate": 3.183875381538785e-05,
      "loss": 0.4352,
      "step": 27040
    },
    {
      "epoch": 8.538364382696559,
      "grad_norm": 7.35109806060791,
      "learning_rate": 3.1837876714731785e-05,
      "loss": 0.5132,
      "step": 27041
    },
    {
      "epoch": 8.538680138932744,
      "grad_norm": 5.8659844398498535,
      "learning_rate": 3.1836999614075716e-05,
      "loss": 0.3466,
      "step": 27042
    },
    {
      "epoch": 8.538995895168929,
      "grad_norm": 8.717597007751465,
      "learning_rate": 3.183612251341964e-05,
      "loss": 0.3766,
      "step": 27043
    },
    {
      "epoch": 8.539311651405114,
      "grad_norm": 5.6986918449401855,
      "learning_rate": 3.183524541276357e-05,
      "loss": 0.4249,
      "step": 27044
    },
    {
      "epoch": 8.539627407641301,
      "grad_norm": 10.974926948547363,
      "learning_rate": 3.18343683121075e-05,
      "loss": 0.5356,
      "step": 27045
    },
    {
      "epoch": 8.539943163877487,
      "grad_norm": 7.080690860748291,
      "learning_rate": 3.183349121145143e-05,
      "loss": 0.4032,
      "step": 27046
    },
    {
      "epoch": 8.540258920113672,
      "grad_norm": 5.28264045715332,
      "learning_rate": 3.1832614110795354e-05,
      "loss": 0.3581,
      "step": 27047
    },
    {
      "epoch": 8.540574676349857,
      "grad_norm": 6.043037414550781,
      "learning_rate": 3.1831737010139286e-05,
      "loss": 0.4021,
      "step": 27048
    },
    {
      "epoch": 8.540890432586043,
      "grad_norm": 6.790359020233154,
      "learning_rate": 3.183085990948321e-05,
      "loss": 0.3984,
      "step": 27049
    },
    {
      "epoch": 8.54120618882223,
      "grad_norm": 6.532845973968506,
      "learning_rate": 3.182998280882714e-05,
      "loss": 0.4356,
      "step": 27050
    },
    {
      "epoch": 8.541521945058415,
      "grad_norm": 5.33720064163208,
      "learning_rate": 3.182910570817107e-05,
      "loss": 0.4824,
      "step": 27051
    },
    {
      "epoch": 8.5418377012946,
      "grad_norm": 7.773157119750977,
      "learning_rate": 3.1828228607515e-05,
      "loss": 0.4681,
      "step": 27052
    },
    {
      "epoch": 8.542153457530786,
      "grad_norm": 5.3948235511779785,
      "learning_rate": 3.1827351506858924e-05,
      "loss": 0.4356,
      "step": 27053
    },
    {
      "epoch": 8.542469213766973,
      "grad_norm": 7.151463031768799,
      "learning_rate": 3.1826474406202856e-05,
      "loss": 0.4939,
      "step": 27054
    },
    {
      "epoch": 8.542784970003158,
      "grad_norm": 6.903818607330322,
      "learning_rate": 3.182559730554678e-05,
      "loss": 0.3879,
      "step": 27055
    },
    {
      "epoch": 8.543100726239343,
      "grad_norm": 4.232204914093018,
      "learning_rate": 3.182472020489071e-05,
      "loss": 0.3468,
      "step": 27056
    },
    {
      "epoch": 8.543416482475529,
      "grad_norm": 4.8670973777771,
      "learning_rate": 3.1823843104234644e-05,
      "loss": 0.413,
      "step": 27057
    },
    {
      "epoch": 8.543732238711714,
      "grad_norm": 5.5602264404296875,
      "learning_rate": 3.1822966003578576e-05,
      "loss": 0.3909,
      "step": 27058
    },
    {
      "epoch": 8.544047994947901,
      "grad_norm": 4.365481376647949,
      "learning_rate": 3.18220889029225e-05,
      "loss": 0.3733,
      "step": 27059
    },
    {
      "epoch": 8.544363751184086,
      "grad_norm": 7.601342678070068,
      "learning_rate": 3.182121180226643e-05,
      "loss": 0.455,
      "step": 27060
    },
    {
      "epoch": 8.544679507420271,
      "grad_norm": 5.41341495513916,
      "learning_rate": 3.182033470161036e-05,
      "loss": 0.4287,
      "step": 27061
    },
    {
      "epoch": 8.544995263656457,
      "grad_norm": 5.476365566253662,
      "learning_rate": 3.181945760095429e-05,
      "loss": 0.5099,
      "step": 27062
    },
    {
      "epoch": 8.545311019892642,
      "grad_norm": 7.166950225830078,
      "learning_rate": 3.181858050029822e-05,
      "loss": 0.3719,
      "step": 27063
    },
    {
      "epoch": 8.54562677612883,
      "grad_norm": 8.258638381958008,
      "learning_rate": 3.1817703399642145e-05,
      "loss": 0.4398,
      "step": 27064
    },
    {
      "epoch": 8.545942532365014,
      "grad_norm": 6.336374759674072,
      "learning_rate": 3.181682629898608e-05,
      "loss": 0.4133,
      "step": 27065
    },
    {
      "epoch": 8.5462582886012,
      "grad_norm": 7.2806396484375,
      "learning_rate": 3.181594919833e-05,
      "loss": 0.3698,
      "step": 27066
    },
    {
      "epoch": 8.546574044837385,
      "grad_norm": 6.28342342376709,
      "learning_rate": 3.1815072097673934e-05,
      "loss": 0.3424,
      "step": 27067
    },
    {
      "epoch": 8.546889801073572,
      "grad_norm": 6.720498085021973,
      "learning_rate": 3.181419499701786e-05,
      "loss": 0.2915,
      "step": 27068
    },
    {
      "epoch": 8.547205557309757,
      "grad_norm": 6.000845432281494,
      "learning_rate": 3.181331789636179e-05,
      "loss": 0.4447,
      "step": 27069
    },
    {
      "epoch": 8.547521313545943,
      "grad_norm": 8.714201927185059,
      "learning_rate": 3.1812440795705715e-05,
      "loss": 0.4646,
      "step": 27070
    },
    {
      "epoch": 8.547837069782128,
      "grad_norm": 7.898458957672119,
      "learning_rate": 3.181156369504965e-05,
      "loss": 0.4184,
      "step": 27071
    },
    {
      "epoch": 8.548152826018313,
      "grad_norm": 7.422436237335205,
      "learning_rate": 3.181068659439357e-05,
      "loss": 0.3979,
      "step": 27072
    },
    {
      "epoch": 8.5484685822545,
      "grad_norm": 6.273396015167236,
      "learning_rate": 3.1809809493737503e-05,
      "loss": 0.4109,
      "step": 27073
    },
    {
      "epoch": 8.548784338490686,
      "grad_norm": 5.3957600593566895,
      "learning_rate": 3.180893239308143e-05,
      "loss": 0.2921,
      "step": 27074
    },
    {
      "epoch": 8.549100094726871,
      "grad_norm": 6.384585380554199,
      "learning_rate": 3.180805529242536e-05,
      "loss": 0.4961,
      "step": 27075
    },
    {
      "epoch": 8.549415850963056,
      "grad_norm": 5.4996490478515625,
      "learning_rate": 3.1807178191769285e-05,
      "loss": 0.3816,
      "step": 27076
    },
    {
      "epoch": 8.549731607199242,
      "grad_norm": 3.8404667377471924,
      "learning_rate": 3.180630109111322e-05,
      "loss": 0.3364,
      "step": 27077
    },
    {
      "epoch": 8.550047363435429,
      "grad_norm": 6.62678861618042,
      "learning_rate": 3.180542399045714e-05,
      "loss": 0.4296,
      "step": 27078
    },
    {
      "epoch": 8.550363119671614,
      "grad_norm": 8.834723472595215,
      "learning_rate": 3.180454688980107e-05,
      "loss": 0.4001,
      "step": 27079
    },
    {
      "epoch": 8.5506788759078,
      "grad_norm": 6.111433982849121,
      "learning_rate": 3.1803669789145005e-05,
      "loss": 0.4676,
      "step": 27080
    },
    {
      "epoch": 8.550994632143984,
      "grad_norm": 5.162250995635986,
      "learning_rate": 3.180279268848893e-05,
      "loss": 0.3561,
      "step": 27081
    },
    {
      "epoch": 8.55131038838017,
      "grad_norm": 8.032641410827637,
      "learning_rate": 3.180191558783286e-05,
      "loss": 0.3615,
      "step": 27082
    },
    {
      "epoch": 8.551626144616357,
      "grad_norm": 7.85264253616333,
      "learning_rate": 3.1801038487176786e-05,
      "loss": 0.4863,
      "step": 27083
    },
    {
      "epoch": 8.551941900852542,
      "grad_norm": 6.521717548370361,
      "learning_rate": 3.180016138652072e-05,
      "loss": 0.299,
      "step": 27084
    },
    {
      "epoch": 8.552257657088727,
      "grad_norm": 6.786249160766602,
      "learning_rate": 3.179928428586465e-05,
      "loss": 0.423,
      "step": 27085
    },
    {
      "epoch": 8.552573413324913,
      "grad_norm": 5.935614109039307,
      "learning_rate": 3.179840718520858e-05,
      "loss": 0.3558,
      "step": 27086
    },
    {
      "epoch": 8.552889169561098,
      "grad_norm": 6.575183868408203,
      "learning_rate": 3.1797530084552506e-05,
      "loss": 0.3881,
      "step": 27087
    },
    {
      "epoch": 8.553204925797285,
      "grad_norm": 6.791321277618408,
      "learning_rate": 3.179665298389644e-05,
      "loss": 0.4622,
      "step": 27088
    },
    {
      "epoch": 8.55352068203347,
      "grad_norm": 7.81744384765625,
      "learning_rate": 3.179577588324036e-05,
      "loss": 0.4118,
      "step": 27089
    },
    {
      "epoch": 8.553836438269656,
      "grad_norm": 9.235757827758789,
      "learning_rate": 3.1794898782584295e-05,
      "loss": 0.3573,
      "step": 27090
    },
    {
      "epoch": 8.554152194505841,
      "grad_norm": 6.0892534255981445,
      "learning_rate": 3.179402168192822e-05,
      "loss": 0.4493,
      "step": 27091
    },
    {
      "epoch": 8.554467950742026,
      "grad_norm": 6.248452663421631,
      "learning_rate": 3.179314458127215e-05,
      "loss": 0.4126,
      "step": 27092
    },
    {
      "epoch": 8.554783706978213,
      "grad_norm": 6.91897439956665,
      "learning_rate": 3.1792267480616076e-05,
      "loss": 0.3815,
      "step": 27093
    },
    {
      "epoch": 8.555099463214399,
      "grad_norm": 7.387355327606201,
      "learning_rate": 3.179139037996001e-05,
      "loss": 0.4113,
      "step": 27094
    },
    {
      "epoch": 8.555415219450584,
      "grad_norm": 7.408442497253418,
      "learning_rate": 3.179051327930393e-05,
      "loss": 0.4225,
      "step": 27095
    },
    {
      "epoch": 8.55573097568677,
      "grad_norm": 8.187637329101562,
      "learning_rate": 3.1789636178647864e-05,
      "loss": 0.3478,
      "step": 27096
    },
    {
      "epoch": 8.556046731922956,
      "grad_norm": 4.425787925720215,
      "learning_rate": 3.178875907799179e-05,
      "loss": 0.3164,
      "step": 27097
    },
    {
      "epoch": 8.556362488159142,
      "grad_norm": 6.706385612487793,
      "learning_rate": 3.178788197733572e-05,
      "loss": 0.3108,
      "step": 27098
    },
    {
      "epoch": 8.556678244395327,
      "grad_norm": 7.4120659828186035,
      "learning_rate": 3.1787004876679646e-05,
      "loss": 0.48,
      "step": 27099
    },
    {
      "epoch": 8.556994000631512,
      "grad_norm": 5.144829273223877,
      "learning_rate": 3.178612777602358e-05,
      "loss": 0.3896,
      "step": 27100
    },
    {
      "epoch": 8.557309756867697,
      "grad_norm": 6.097249984741211,
      "learning_rate": 3.178525067536751e-05,
      "loss": 0.4228,
      "step": 27101
    },
    {
      "epoch": 8.557625513103885,
      "grad_norm": 5.609243869781494,
      "learning_rate": 3.1784373574711434e-05,
      "loss": 0.4432,
      "step": 27102
    },
    {
      "epoch": 8.55794126934007,
      "grad_norm": 6.976555824279785,
      "learning_rate": 3.1783496474055366e-05,
      "loss": 0.4485,
      "step": 27103
    },
    {
      "epoch": 8.558257025576255,
      "grad_norm": 5.695809364318848,
      "learning_rate": 3.178261937339929e-05,
      "loss": 0.3141,
      "step": 27104
    },
    {
      "epoch": 8.55857278181244,
      "grad_norm": 8.067859649658203,
      "learning_rate": 3.178174227274322e-05,
      "loss": 0.4792,
      "step": 27105
    },
    {
      "epoch": 8.558888538048626,
      "grad_norm": 6.292194843292236,
      "learning_rate": 3.178086517208715e-05,
      "loss": 0.377,
      "step": 27106
    },
    {
      "epoch": 8.559204294284813,
      "grad_norm": 5.602575302124023,
      "learning_rate": 3.177998807143108e-05,
      "loss": 0.4794,
      "step": 27107
    },
    {
      "epoch": 8.559520050520998,
      "grad_norm": 8.147098541259766,
      "learning_rate": 3.1779110970775004e-05,
      "loss": 0.5313,
      "step": 27108
    },
    {
      "epoch": 8.559835806757183,
      "grad_norm": 7.325850963592529,
      "learning_rate": 3.1778233870118936e-05,
      "loss": 0.5751,
      "step": 27109
    },
    {
      "epoch": 8.560151562993369,
      "grad_norm": 7.277541160583496,
      "learning_rate": 3.177735676946286e-05,
      "loss": 0.463,
      "step": 27110
    },
    {
      "epoch": 8.560467319229556,
      "grad_norm": 5.365611553192139,
      "learning_rate": 3.177647966880679e-05,
      "loss": 0.3028,
      "step": 27111
    },
    {
      "epoch": 8.560783075465741,
      "grad_norm": 9.07190990447998,
      "learning_rate": 3.1775602568150724e-05,
      "loss": 0.4913,
      "step": 27112
    },
    {
      "epoch": 8.561098831701926,
      "grad_norm": 5.772093772888184,
      "learning_rate": 3.1774725467494656e-05,
      "loss": 0.4088,
      "step": 27113
    },
    {
      "epoch": 8.561414587938112,
      "grad_norm": 6.171682357788086,
      "learning_rate": 3.177384836683858e-05,
      "loss": 0.4113,
      "step": 27114
    },
    {
      "epoch": 8.561730344174297,
      "grad_norm": 5.795095920562744,
      "learning_rate": 3.177297126618251e-05,
      "loss": 0.4468,
      "step": 27115
    },
    {
      "epoch": 8.562046100410484,
      "grad_norm": 5.163541793823242,
      "learning_rate": 3.177209416552644e-05,
      "loss": 0.4244,
      "step": 27116
    },
    {
      "epoch": 8.56236185664667,
      "grad_norm": 5.740020275115967,
      "learning_rate": 3.177121706487037e-05,
      "loss": 0.4627,
      "step": 27117
    },
    {
      "epoch": 8.562677612882855,
      "grad_norm": 7.121267318725586,
      "learning_rate": 3.1770339964214294e-05,
      "loss": 0.4312,
      "step": 27118
    },
    {
      "epoch": 8.56299336911904,
      "grad_norm": 5.268569469451904,
      "learning_rate": 3.1769462863558225e-05,
      "loss": 0.4093,
      "step": 27119
    },
    {
      "epoch": 8.563309125355225,
      "grad_norm": 5.359701633453369,
      "learning_rate": 3.176858576290215e-05,
      "loss": 0.381,
      "step": 27120
    },
    {
      "epoch": 8.563624881591412,
      "grad_norm": 6.9574079513549805,
      "learning_rate": 3.176770866224608e-05,
      "loss": 0.4332,
      "step": 27121
    },
    {
      "epoch": 8.563940637827598,
      "grad_norm": 6.9371771812438965,
      "learning_rate": 3.1766831561590014e-05,
      "loss": 0.5256,
      "step": 27122
    },
    {
      "epoch": 8.564256394063783,
      "grad_norm": 6.7801618576049805,
      "learning_rate": 3.176595446093394e-05,
      "loss": 0.4391,
      "step": 27123
    },
    {
      "epoch": 8.564572150299968,
      "grad_norm": 7.465712070465088,
      "learning_rate": 3.176507736027787e-05,
      "loss": 0.3843,
      "step": 27124
    },
    {
      "epoch": 8.564887906536153,
      "grad_norm": 8.668693542480469,
      "learning_rate": 3.1764200259621795e-05,
      "loss": 0.4774,
      "step": 27125
    },
    {
      "epoch": 8.56520366277234,
      "grad_norm": 7.690810680389404,
      "learning_rate": 3.176332315896573e-05,
      "loss": 0.4064,
      "step": 27126
    },
    {
      "epoch": 8.565519419008526,
      "grad_norm": 6.9591851234436035,
      "learning_rate": 3.176244605830965e-05,
      "loss": 0.4774,
      "step": 27127
    },
    {
      "epoch": 8.565835175244711,
      "grad_norm": 10.306963920593262,
      "learning_rate": 3.1761568957653583e-05,
      "loss": 0.3906,
      "step": 27128
    },
    {
      "epoch": 8.566150931480896,
      "grad_norm": 6.56127405166626,
      "learning_rate": 3.176069185699751e-05,
      "loss": 0.4241,
      "step": 27129
    },
    {
      "epoch": 8.566466687717082,
      "grad_norm": 5.653336048126221,
      "learning_rate": 3.175981475634144e-05,
      "loss": 0.4576,
      "step": 27130
    },
    {
      "epoch": 8.566782443953269,
      "grad_norm": 6.26597261428833,
      "learning_rate": 3.1758937655685365e-05,
      "loss": 0.391,
      "step": 27131
    },
    {
      "epoch": 8.567098200189454,
      "grad_norm": 7.229394912719727,
      "learning_rate": 3.1758060555029297e-05,
      "loss": 0.3842,
      "step": 27132
    },
    {
      "epoch": 8.56741395642564,
      "grad_norm": 6.194643497467041,
      "learning_rate": 3.175718345437322e-05,
      "loss": 0.402,
      "step": 27133
    },
    {
      "epoch": 8.567729712661825,
      "grad_norm": 6.174442291259766,
      "learning_rate": 3.175630635371715e-05,
      "loss": 0.4042,
      "step": 27134
    },
    {
      "epoch": 8.56804546889801,
      "grad_norm": 6.305044651031494,
      "learning_rate": 3.175542925306108e-05,
      "loss": 0.475,
      "step": 27135
    },
    {
      "epoch": 8.568361225134197,
      "grad_norm": 6.218270301818848,
      "learning_rate": 3.175455215240501e-05,
      "loss": 0.3767,
      "step": 27136
    },
    {
      "epoch": 8.568676981370382,
      "grad_norm": 6.1601643562316895,
      "learning_rate": 3.1753675051748935e-05,
      "loss": 0.4784,
      "step": 27137
    },
    {
      "epoch": 8.568992737606568,
      "grad_norm": 6.811288833618164,
      "learning_rate": 3.1752797951092866e-05,
      "loss": 0.4731,
      "step": 27138
    },
    {
      "epoch": 8.569308493842753,
      "grad_norm": 5.804346561431885,
      "learning_rate": 3.17519208504368e-05,
      "loss": 0.4166,
      "step": 27139
    },
    {
      "epoch": 8.569624250078938,
      "grad_norm": 10.208517074584961,
      "learning_rate": 3.175104374978073e-05,
      "loss": 0.4247,
      "step": 27140
    },
    {
      "epoch": 8.569940006315125,
      "grad_norm": 6.132707595825195,
      "learning_rate": 3.1750166649124655e-05,
      "loss": 0.4028,
      "step": 27141
    },
    {
      "epoch": 8.57025576255131,
      "grad_norm": 7.1010589599609375,
      "learning_rate": 3.1749289548468586e-05,
      "loss": 0.3714,
      "step": 27142
    },
    {
      "epoch": 8.570571518787496,
      "grad_norm": 6.445137023925781,
      "learning_rate": 3.174841244781251e-05,
      "loss": 0.4549,
      "step": 27143
    },
    {
      "epoch": 8.570887275023681,
      "grad_norm": 7.190310478210449,
      "learning_rate": 3.174753534715644e-05,
      "loss": 0.473,
      "step": 27144
    },
    {
      "epoch": 8.571203031259868,
      "grad_norm": 5.364979267120361,
      "learning_rate": 3.1746658246500375e-05,
      "loss": 0.3824,
      "step": 27145
    },
    {
      "epoch": 8.571518787496053,
      "grad_norm": 4.407707691192627,
      "learning_rate": 3.17457811458443e-05,
      "loss": 0.3517,
      "step": 27146
    },
    {
      "epoch": 8.571834543732239,
      "grad_norm": 6.4855217933654785,
      "learning_rate": 3.174490404518823e-05,
      "loss": 0.3598,
      "step": 27147
    },
    {
      "epoch": 8.572150299968424,
      "grad_norm": 4.955365180969238,
      "learning_rate": 3.1744026944532156e-05,
      "loss": 0.4033,
      "step": 27148
    },
    {
      "epoch": 8.57246605620461,
      "grad_norm": 5.335076808929443,
      "learning_rate": 3.174314984387609e-05,
      "loss": 0.4231,
      "step": 27149
    },
    {
      "epoch": 8.572781812440796,
      "grad_norm": 7.051120281219482,
      "learning_rate": 3.174227274322001e-05,
      "loss": 0.4207,
      "step": 27150
    },
    {
      "epoch": 8.573097568676982,
      "grad_norm": 5.658889293670654,
      "learning_rate": 3.1741395642563944e-05,
      "loss": 0.4442,
      "step": 27151
    },
    {
      "epoch": 8.573413324913167,
      "grad_norm": 9.262431144714355,
      "learning_rate": 3.174051854190787e-05,
      "loss": 0.4661,
      "step": 27152
    },
    {
      "epoch": 8.573729081149352,
      "grad_norm": 5.85286283493042,
      "learning_rate": 3.17396414412518e-05,
      "loss": 0.4945,
      "step": 27153
    },
    {
      "epoch": 8.574044837385538,
      "grad_norm": 5.577611923217773,
      "learning_rate": 3.1738764340595726e-05,
      "loss": 0.3593,
      "step": 27154
    },
    {
      "epoch": 8.574360593621725,
      "grad_norm": 5.8026251792907715,
      "learning_rate": 3.173788723993966e-05,
      "loss": 0.4551,
      "step": 27155
    },
    {
      "epoch": 8.57467634985791,
      "grad_norm": 6.232577800750732,
      "learning_rate": 3.173701013928358e-05,
      "loss": 0.4795,
      "step": 27156
    },
    {
      "epoch": 8.574992106094095,
      "grad_norm": 5.510627746582031,
      "learning_rate": 3.1736133038627514e-05,
      "loss": 0.4187,
      "step": 27157
    },
    {
      "epoch": 8.57530786233028,
      "grad_norm": 6.43461275100708,
      "learning_rate": 3.173525593797144e-05,
      "loss": 0.4083,
      "step": 27158
    },
    {
      "epoch": 8.575623618566468,
      "grad_norm": 5.685737609863281,
      "learning_rate": 3.173437883731537e-05,
      "loss": 0.4083,
      "step": 27159
    },
    {
      "epoch": 8.575939374802653,
      "grad_norm": 6.1396636962890625,
      "learning_rate": 3.17335017366593e-05,
      "loss": 0.4466,
      "step": 27160
    },
    {
      "epoch": 8.576255131038838,
      "grad_norm": 5.158979892730713,
      "learning_rate": 3.173262463600323e-05,
      "loss": 0.4222,
      "step": 27161
    },
    {
      "epoch": 8.576570887275023,
      "grad_norm": 6.616439342498779,
      "learning_rate": 3.173174753534716e-05,
      "loss": 0.4494,
      "step": 27162
    },
    {
      "epoch": 8.576886643511209,
      "grad_norm": 8.907816886901855,
      "learning_rate": 3.1730870434691084e-05,
      "loss": 0.5383,
      "step": 27163
    },
    {
      "epoch": 8.577202399747396,
      "grad_norm": 5.099236965179443,
      "learning_rate": 3.1729993334035016e-05,
      "loss": 0.435,
      "step": 27164
    },
    {
      "epoch": 8.577518155983581,
      "grad_norm": 5.4197211265563965,
      "learning_rate": 3.172911623337894e-05,
      "loss": 0.3743,
      "step": 27165
    },
    {
      "epoch": 8.577833912219766,
      "grad_norm": 6.364943027496338,
      "learning_rate": 3.172823913272287e-05,
      "loss": 0.3995,
      "step": 27166
    },
    {
      "epoch": 8.578149668455952,
      "grad_norm": 5.5960845947265625,
      "learning_rate": 3.1727362032066804e-05,
      "loss": 0.4453,
      "step": 27167
    },
    {
      "epoch": 8.578465424692137,
      "grad_norm": 5.142599582672119,
      "learning_rate": 3.1726484931410735e-05,
      "loss": 0.4119,
      "step": 27168
    },
    {
      "epoch": 8.578781180928324,
      "grad_norm": 4.372696399688721,
      "learning_rate": 3.172560783075466e-05,
      "loss": 0.3586,
      "step": 27169
    },
    {
      "epoch": 8.57909693716451,
      "grad_norm": 5.5083909034729,
      "learning_rate": 3.172473073009859e-05,
      "loss": 0.4215,
      "step": 27170
    },
    {
      "epoch": 8.579412693400695,
      "grad_norm": 7.164194107055664,
      "learning_rate": 3.172385362944252e-05,
      "loss": 0.4363,
      "step": 27171
    },
    {
      "epoch": 8.57972844963688,
      "grad_norm": 5.305843353271484,
      "learning_rate": 3.172297652878645e-05,
      "loss": 0.3975,
      "step": 27172
    },
    {
      "epoch": 8.580044205873065,
      "grad_norm": 6.034931659698486,
      "learning_rate": 3.1722099428130374e-05,
      "loss": 0.4751,
      "step": 27173
    },
    {
      "epoch": 8.580359962109252,
      "grad_norm": 8.915412902832031,
      "learning_rate": 3.1721222327474305e-05,
      "loss": 0.4516,
      "step": 27174
    },
    {
      "epoch": 8.580675718345438,
      "grad_norm": 6.1713361740112305,
      "learning_rate": 3.172034522681823e-05,
      "loss": 0.377,
      "step": 27175
    },
    {
      "epoch": 8.580991474581623,
      "grad_norm": 6.266157627105713,
      "learning_rate": 3.171946812616216e-05,
      "loss": 0.43,
      "step": 27176
    },
    {
      "epoch": 8.581307230817808,
      "grad_norm": 5.623977184295654,
      "learning_rate": 3.171859102550609e-05,
      "loss": 0.3562,
      "step": 27177
    },
    {
      "epoch": 8.581622987053994,
      "grad_norm": 6.00996208190918,
      "learning_rate": 3.171771392485002e-05,
      "loss": 0.4435,
      "step": 27178
    },
    {
      "epoch": 8.58193874329018,
      "grad_norm": 5.190196990966797,
      "learning_rate": 3.171683682419394e-05,
      "loss": 0.3133,
      "step": 27179
    },
    {
      "epoch": 8.582254499526366,
      "grad_norm": 6.185795783996582,
      "learning_rate": 3.1715959723537875e-05,
      "loss": 0.3755,
      "step": 27180
    },
    {
      "epoch": 8.582570255762551,
      "grad_norm": 7.029810905456543,
      "learning_rate": 3.17150826228818e-05,
      "loss": 0.3616,
      "step": 27181
    },
    {
      "epoch": 8.582886011998736,
      "grad_norm": 5.64406156539917,
      "learning_rate": 3.171420552222573e-05,
      "loss": 0.3513,
      "step": 27182
    },
    {
      "epoch": 8.583201768234922,
      "grad_norm": 6.5110273361206055,
      "learning_rate": 3.171332842156966e-05,
      "loss": 0.457,
      "step": 27183
    },
    {
      "epoch": 8.583517524471109,
      "grad_norm": 4.685999393463135,
      "learning_rate": 3.171245132091359e-05,
      "loss": 0.3696,
      "step": 27184
    },
    {
      "epoch": 8.583833280707294,
      "grad_norm": 8.270379066467285,
      "learning_rate": 3.171157422025752e-05,
      "loss": 0.4939,
      "step": 27185
    },
    {
      "epoch": 8.58414903694348,
      "grad_norm": 6.486666679382324,
      "learning_rate": 3.1710697119601445e-05,
      "loss": 0.3744,
      "step": 27186
    },
    {
      "epoch": 8.584464793179665,
      "grad_norm": 5.681169509887695,
      "learning_rate": 3.1709820018945376e-05,
      "loss": 0.3975,
      "step": 27187
    },
    {
      "epoch": 8.584780549415852,
      "grad_norm": 5.8278608322143555,
      "learning_rate": 3.17089429182893e-05,
      "loss": 0.404,
      "step": 27188
    },
    {
      "epoch": 8.585096305652037,
      "grad_norm": 8.05742073059082,
      "learning_rate": 3.170806581763323e-05,
      "loss": 0.4537,
      "step": 27189
    },
    {
      "epoch": 8.585412061888222,
      "grad_norm": 6.7489333152771,
      "learning_rate": 3.170718871697716e-05,
      "loss": 0.5351,
      "step": 27190
    },
    {
      "epoch": 8.585727818124408,
      "grad_norm": 6.456061363220215,
      "learning_rate": 3.170631161632109e-05,
      "loss": 0.4506,
      "step": 27191
    },
    {
      "epoch": 8.586043574360593,
      "grad_norm": 5.879573345184326,
      "learning_rate": 3.1705434515665015e-05,
      "loss": 0.4421,
      "step": 27192
    },
    {
      "epoch": 8.58635933059678,
      "grad_norm": 6.060600757598877,
      "learning_rate": 3.1704557415008946e-05,
      "loss": 0.4173,
      "step": 27193
    },
    {
      "epoch": 8.586675086832965,
      "grad_norm": 5.852120399475098,
      "learning_rate": 3.170368031435288e-05,
      "loss": 0.374,
      "step": 27194
    },
    {
      "epoch": 8.58699084306915,
      "grad_norm": 5.966066360473633,
      "learning_rate": 3.170280321369681e-05,
      "loss": 0.3944,
      "step": 27195
    },
    {
      "epoch": 8.587306599305336,
      "grad_norm": 7.2953290939331055,
      "learning_rate": 3.1701926113040734e-05,
      "loss": 0.4582,
      "step": 27196
    },
    {
      "epoch": 8.587622355541521,
      "grad_norm": 5.434734344482422,
      "learning_rate": 3.1701049012384666e-05,
      "loss": 0.4247,
      "step": 27197
    },
    {
      "epoch": 8.587938111777708,
      "grad_norm": 5.701608180999756,
      "learning_rate": 3.170017191172859e-05,
      "loss": 0.4286,
      "step": 27198
    },
    {
      "epoch": 8.588253868013894,
      "grad_norm": 5.205108165740967,
      "learning_rate": 3.169929481107252e-05,
      "loss": 0.4718,
      "step": 27199
    },
    {
      "epoch": 8.588569624250079,
      "grad_norm": 7.971468925476074,
      "learning_rate": 3.169841771041645e-05,
      "loss": 0.5094,
      "step": 27200
    },
    {
      "epoch": 8.588569624250079,
      "eval_loss": 0.9699059724807739,
      "eval_runtime": 4.6665,
      "eval_samples_per_second": 533.596,
      "eval_steps_per_second": 4.5,
      "step": 27200
    },
    {
      "epoch": 8.588885380486264,
      "grad_norm": 6.058443069458008,
      "learning_rate": 3.169754060976038e-05,
      "loss": 0.3474,
      "step": 27201
    },
    {
      "epoch": 8.58920113672245,
      "grad_norm": 5.804065704345703,
      "learning_rate": 3.1696663509104304e-05,
      "loss": 0.4666,
      "step": 27202
    },
    {
      "epoch": 8.589516892958637,
      "grad_norm": 5.67666482925415,
      "learning_rate": 3.1695786408448236e-05,
      "loss": 0.5196,
      "step": 27203
    },
    {
      "epoch": 8.589832649194822,
      "grad_norm": 6.97169828414917,
      "learning_rate": 3.169490930779217e-05,
      "loss": 0.4334,
      "step": 27204
    },
    {
      "epoch": 8.590148405431007,
      "grad_norm": 5.613837242126465,
      "learning_rate": 3.169403220713609e-05,
      "loss": 0.3397,
      "step": 27205
    },
    {
      "epoch": 8.590464161667192,
      "grad_norm": 6.44287633895874,
      "learning_rate": 3.1693155106480024e-05,
      "loss": 0.3817,
      "step": 27206
    },
    {
      "epoch": 8.59077991790338,
      "grad_norm": 6.118575572967529,
      "learning_rate": 3.169227800582395e-05,
      "loss": 0.3502,
      "step": 27207
    },
    {
      "epoch": 8.591095674139565,
      "grad_norm": 7.092749118804932,
      "learning_rate": 3.169140090516788e-05,
      "loss": 0.4352,
      "step": 27208
    },
    {
      "epoch": 8.59141143037575,
      "grad_norm": 7.046749114990234,
      "learning_rate": 3.1690523804511806e-05,
      "loss": 0.5278,
      "step": 27209
    },
    {
      "epoch": 8.591727186611935,
      "grad_norm": 5.520078182220459,
      "learning_rate": 3.168964670385574e-05,
      "loss": 0.3697,
      "step": 27210
    },
    {
      "epoch": 8.59204294284812,
      "grad_norm": 7.566824913024902,
      "learning_rate": 3.168876960319966e-05,
      "loss": 0.4935,
      "step": 27211
    },
    {
      "epoch": 8.592358699084308,
      "grad_norm": 6.172701358795166,
      "learning_rate": 3.1687892502543594e-05,
      "loss": 0.4715,
      "step": 27212
    },
    {
      "epoch": 8.592674455320493,
      "grad_norm": 5.4308648109436035,
      "learning_rate": 3.168701540188752e-05,
      "loss": 0.3323,
      "step": 27213
    },
    {
      "epoch": 8.592990211556678,
      "grad_norm": 6.883896827697754,
      "learning_rate": 3.168613830123145e-05,
      "loss": 0.4376,
      "step": 27214
    },
    {
      "epoch": 8.593305967792864,
      "grad_norm": 6.2185869216918945,
      "learning_rate": 3.1685261200575375e-05,
      "loss": 0.4122,
      "step": 27215
    },
    {
      "epoch": 8.593621724029049,
      "grad_norm": 5.6626386642456055,
      "learning_rate": 3.168438409991931e-05,
      "loss": 0.2945,
      "step": 27216
    },
    {
      "epoch": 8.593937480265236,
      "grad_norm": 6.341629981994629,
      "learning_rate": 3.168350699926323e-05,
      "loss": 0.5298,
      "step": 27217
    },
    {
      "epoch": 8.594253236501421,
      "grad_norm": 7.060736179351807,
      "learning_rate": 3.1682629898607164e-05,
      "loss": 0.3346,
      "step": 27218
    },
    {
      "epoch": 8.594568992737607,
      "grad_norm": 7.291611194610596,
      "learning_rate": 3.168175279795109e-05,
      "loss": 0.4639,
      "step": 27219
    },
    {
      "epoch": 8.594884748973792,
      "grad_norm": 5.692365646362305,
      "learning_rate": 3.168087569729502e-05,
      "loss": 0.4407,
      "step": 27220
    },
    {
      "epoch": 8.595200505209977,
      "grad_norm": 6.785994529724121,
      "learning_rate": 3.167999859663895e-05,
      "loss": 0.5041,
      "step": 27221
    },
    {
      "epoch": 8.595516261446164,
      "grad_norm": 6.3018693923950195,
      "learning_rate": 3.1679121495982884e-05,
      "loss": 0.3489,
      "step": 27222
    },
    {
      "epoch": 8.59583201768235,
      "grad_norm": 5.751760482788086,
      "learning_rate": 3.167824439532681e-05,
      "loss": 0.3841,
      "step": 27223
    },
    {
      "epoch": 8.596147773918535,
      "grad_norm": 5.810462951660156,
      "learning_rate": 3.167736729467074e-05,
      "loss": 0.3957,
      "step": 27224
    },
    {
      "epoch": 8.59646353015472,
      "grad_norm": 6.444847106933594,
      "learning_rate": 3.1676490194014665e-05,
      "loss": 0.5345,
      "step": 27225
    },
    {
      "epoch": 8.596779286390905,
      "grad_norm": 6.9415693283081055,
      "learning_rate": 3.16756130933586e-05,
      "loss": 0.4709,
      "step": 27226
    },
    {
      "epoch": 8.597095042627092,
      "grad_norm": 4.953574180603027,
      "learning_rate": 3.167473599270253e-05,
      "loss": 0.3137,
      "step": 27227
    },
    {
      "epoch": 8.597410798863278,
      "grad_norm": 5.725205898284912,
      "learning_rate": 3.1673858892046453e-05,
      "loss": 0.3663,
      "step": 27228
    },
    {
      "epoch": 8.597726555099463,
      "grad_norm": 5.705055236816406,
      "learning_rate": 3.1672981791390385e-05,
      "loss": 0.3605,
      "step": 27229
    },
    {
      "epoch": 8.598042311335648,
      "grad_norm": 5.748292446136475,
      "learning_rate": 3.167210469073431e-05,
      "loss": 0.4976,
      "step": 27230
    },
    {
      "epoch": 8.598358067571834,
      "grad_norm": 5.869623184204102,
      "learning_rate": 3.167122759007824e-05,
      "loss": 0.3772,
      "step": 27231
    },
    {
      "epoch": 8.59867382380802,
      "grad_norm": 6.571224689483643,
      "learning_rate": 3.167035048942217e-05,
      "loss": 0.4021,
      "step": 27232
    },
    {
      "epoch": 8.598989580044206,
      "grad_norm": 5.760580062866211,
      "learning_rate": 3.16694733887661e-05,
      "loss": 0.3536,
      "step": 27233
    },
    {
      "epoch": 8.599305336280391,
      "grad_norm": 5.5923967361450195,
      "learning_rate": 3.166859628811002e-05,
      "loss": 0.4273,
      "step": 27234
    },
    {
      "epoch": 8.599621092516577,
      "grad_norm": 4.2207536697387695,
      "learning_rate": 3.1667719187453955e-05,
      "loss": 0.3134,
      "step": 27235
    },
    {
      "epoch": 8.599936848752764,
      "grad_norm": 5.46574592590332,
      "learning_rate": 3.166684208679788e-05,
      "loss": 0.4084,
      "step": 27236
    },
    {
      "epoch": 8.600252604988949,
      "grad_norm": 6.557784080505371,
      "learning_rate": 3.166596498614181e-05,
      "loss": 0.5363,
      "step": 27237
    },
    {
      "epoch": 8.600568361225134,
      "grad_norm": 12.214330673217773,
      "learning_rate": 3.1665087885485736e-05,
      "loss": 0.5541,
      "step": 27238
    },
    {
      "epoch": 8.60088411746132,
      "grad_norm": 5.721279621124268,
      "learning_rate": 3.166421078482967e-05,
      "loss": 0.3924,
      "step": 27239
    },
    {
      "epoch": 8.601199873697505,
      "grad_norm": 7.147789478302002,
      "learning_rate": 3.166333368417359e-05,
      "loss": 0.4005,
      "step": 27240
    },
    {
      "epoch": 8.601515629933692,
      "grad_norm": 5.317654609680176,
      "learning_rate": 3.1662456583517525e-05,
      "loss": 0.4224,
      "step": 27241
    },
    {
      "epoch": 8.601831386169877,
      "grad_norm": 5.2826151847839355,
      "learning_rate": 3.1661579482861456e-05,
      "loss": 0.458,
      "step": 27242
    },
    {
      "epoch": 8.602147142406062,
      "grad_norm": 6.468916416168213,
      "learning_rate": 3.166070238220538e-05,
      "loss": 0.3423,
      "step": 27243
    },
    {
      "epoch": 8.602462898642248,
      "grad_norm": 6.551453590393066,
      "learning_rate": 3.165982528154931e-05,
      "loss": 0.4946,
      "step": 27244
    },
    {
      "epoch": 8.602778654878433,
      "grad_norm": 4.374748229980469,
      "learning_rate": 3.165894818089324e-05,
      "loss": 0.3746,
      "step": 27245
    },
    {
      "epoch": 8.60309441111462,
      "grad_norm": 5.532835960388184,
      "learning_rate": 3.165807108023717e-05,
      "loss": 0.383,
      "step": 27246
    },
    {
      "epoch": 8.603410167350805,
      "grad_norm": 6.440882205963135,
      "learning_rate": 3.1657193979581094e-05,
      "loss": 0.3825,
      "step": 27247
    },
    {
      "epoch": 8.60372592358699,
      "grad_norm": 7.386203289031982,
      "learning_rate": 3.1656316878925026e-05,
      "loss": 0.442,
      "step": 27248
    },
    {
      "epoch": 8.604041679823176,
      "grad_norm": 7.795694351196289,
      "learning_rate": 3.165543977826896e-05,
      "loss": 0.5179,
      "step": 27249
    },
    {
      "epoch": 8.604357436059363,
      "grad_norm": 6.6224799156188965,
      "learning_rate": 3.165456267761289e-05,
      "loss": 0.4633,
      "step": 27250
    },
    {
      "epoch": 8.604673192295548,
      "grad_norm": 6.475003242492676,
      "learning_rate": 3.1653685576956814e-05,
      "loss": 0.6131,
      "step": 27251
    },
    {
      "epoch": 8.604988948531734,
      "grad_norm": 7.836292743682861,
      "learning_rate": 3.1652808476300746e-05,
      "loss": 0.4651,
      "step": 27252
    },
    {
      "epoch": 8.605304704767919,
      "grad_norm": 6.338902473449707,
      "learning_rate": 3.165193137564467e-05,
      "loss": 0.4779,
      "step": 27253
    },
    {
      "epoch": 8.605620461004104,
      "grad_norm": 6.345070838928223,
      "learning_rate": 3.16510542749886e-05,
      "loss": 0.4337,
      "step": 27254
    },
    {
      "epoch": 8.605936217240291,
      "grad_norm": 6.107771396636963,
      "learning_rate": 3.165017717433253e-05,
      "loss": 0.3358,
      "step": 27255
    },
    {
      "epoch": 8.606251973476477,
      "grad_norm": 5.651839733123779,
      "learning_rate": 3.164930007367646e-05,
      "loss": 0.364,
      "step": 27256
    },
    {
      "epoch": 8.606567729712662,
      "grad_norm": 5.229222774505615,
      "learning_rate": 3.1648422973020384e-05,
      "loss": 0.4541,
      "step": 27257
    },
    {
      "epoch": 8.606883485948847,
      "grad_norm": 5.609489917755127,
      "learning_rate": 3.1647545872364316e-05,
      "loss": 0.3672,
      "step": 27258
    },
    {
      "epoch": 8.607199242185033,
      "grad_norm": 5.859641075134277,
      "learning_rate": 3.164666877170824e-05,
      "loss": 0.6741,
      "step": 27259
    },
    {
      "epoch": 8.60751499842122,
      "grad_norm": 4.979299068450928,
      "learning_rate": 3.164579167105217e-05,
      "loss": 0.3161,
      "step": 27260
    },
    {
      "epoch": 8.607830754657405,
      "grad_norm": 6.722628593444824,
      "learning_rate": 3.16449145703961e-05,
      "loss": 0.4991,
      "step": 27261
    },
    {
      "epoch": 8.60814651089359,
      "grad_norm": 5.4018354415893555,
      "learning_rate": 3.164403746974003e-05,
      "loss": 0.3578,
      "step": 27262
    },
    {
      "epoch": 8.608462267129775,
      "grad_norm": 5.627844333648682,
      "learning_rate": 3.164316036908396e-05,
      "loss": 0.43,
      "step": 27263
    },
    {
      "epoch": 8.60877802336596,
      "grad_norm": 4.59666633605957,
      "learning_rate": 3.1642283268427886e-05,
      "loss": 0.3473,
      "step": 27264
    },
    {
      "epoch": 8.609093779602148,
      "grad_norm": 6.588860034942627,
      "learning_rate": 3.164140616777182e-05,
      "loss": 0.4,
      "step": 27265
    },
    {
      "epoch": 8.609409535838333,
      "grad_norm": 5.301826477050781,
      "learning_rate": 3.164052906711574e-05,
      "loss": 0.4367,
      "step": 27266
    },
    {
      "epoch": 8.609725292074518,
      "grad_norm": 5.278087615966797,
      "learning_rate": 3.1639651966459674e-05,
      "loss": 0.397,
      "step": 27267
    },
    {
      "epoch": 8.610041048310704,
      "grad_norm": 5.183888912200928,
      "learning_rate": 3.16387748658036e-05,
      "loss": 0.3319,
      "step": 27268
    },
    {
      "epoch": 8.610356804546889,
      "grad_norm": 4.661459445953369,
      "learning_rate": 3.163789776514753e-05,
      "loss": 0.2939,
      "step": 27269
    },
    {
      "epoch": 8.610672560783076,
      "grad_norm": 4.636306285858154,
      "learning_rate": 3.1637020664491455e-05,
      "loss": 0.3218,
      "step": 27270
    },
    {
      "epoch": 8.610988317019261,
      "grad_norm": 6.752342700958252,
      "learning_rate": 3.163614356383539e-05,
      "loss": 0.5261,
      "step": 27271
    },
    {
      "epoch": 8.611304073255447,
      "grad_norm": 10.26837158203125,
      "learning_rate": 3.163526646317931e-05,
      "loss": 0.458,
      "step": 27272
    },
    {
      "epoch": 8.611619829491632,
      "grad_norm": 8.104742050170898,
      "learning_rate": 3.1634389362523244e-05,
      "loss": 0.5148,
      "step": 27273
    },
    {
      "epoch": 8.611935585727817,
      "grad_norm": 6.350918292999268,
      "learning_rate": 3.163351226186717e-05,
      "loss": 0.4366,
      "step": 27274
    },
    {
      "epoch": 8.612251341964004,
      "grad_norm": 7.1422648429870605,
      "learning_rate": 3.16326351612111e-05,
      "loss": 0.5244,
      "step": 27275
    },
    {
      "epoch": 8.61256709820019,
      "grad_norm": 7.73478889465332,
      "learning_rate": 3.163175806055503e-05,
      "loss": 0.3912,
      "step": 27276
    },
    {
      "epoch": 8.612882854436375,
      "grad_norm": 8.249588012695312,
      "learning_rate": 3.1630880959898964e-05,
      "loss": 0.4731,
      "step": 27277
    },
    {
      "epoch": 8.61319861067256,
      "grad_norm": 4.614843845367432,
      "learning_rate": 3.163000385924289e-05,
      "loss": 0.3577,
      "step": 27278
    },
    {
      "epoch": 8.613514366908747,
      "grad_norm": 5.7795023918151855,
      "learning_rate": 3.162912675858682e-05,
      "loss": 0.4254,
      "step": 27279
    },
    {
      "epoch": 8.613830123144933,
      "grad_norm": 7.3839921951293945,
      "learning_rate": 3.1628249657930745e-05,
      "loss": 0.5411,
      "step": 27280
    },
    {
      "epoch": 8.614145879381118,
      "grad_norm": 7.581859588623047,
      "learning_rate": 3.162737255727468e-05,
      "loss": 0.4287,
      "step": 27281
    },
    {
      "epoch": 8.614461635617303,
      "grad_norm": 6.659029006958008,
      "learning_rate": 3.16264954566186e-05,
      "loss": 0.4295,
      "step": 27282
    },
    {
      "epoch": 8.614777391853488,
      "grad_norm": 6.8021111488342285,
      "learning_rate": 3.162561835596253e-05,
      "loss": 0.3838,
      "step": 27283
    },
    {
      "epoch": 8.615093148089676,
      "grad_norm": 6.544478416442871,
      "learning_rate": 3.162474125530646e-05,
      "loss": 0.4432,
      "step": 27284
    },
    {
      "epoch": 8.61540890432586,
      "grad_norm": 5.797375202178955,
      "learning_rate": 3.162386415465039e-05,
      "loss": 0.4125,
      "step": 27285
    },
    {
      "epoch": 8.615724660562046,
      "grad_norm": 6.063352108001709,
      "learning_rate": 3.162298705399432e-05,
      "loss": 0.3905,
      "step": 27286
    },
    {
      "epoch": 8.616040416798231,
      "grad_norm": 5.578633785247803,
      "learning_rate": 3.1622109953338247e-05,
      "loss": 0.469,
      "step": 27287
    },
    {
      "epoch": 8.616356173034417,
      "grad_norm": 5.56766939163208,
      "learning_rate": 3.162123285268218e-05,
      "loss": 0.3595,
      "step": 27288
    },
    {
      "epoch": 8.616671929270604,
      "grad_norm": 6.320158004760742,
      "learning_rate": 3.16203557520261e-05,
      "loss": 0.4975,
      "step": 27289
    },
    {
      "epoch": 8.616987685506789,
      "grad_norm": 5.765130519866943,
      "learning_rate": 3.1619478651370035e-05,
      "loss": 0.4136,
      "step": 27290
    },
    {
      "epoch": 8.617303441742974,
      "grad_norm": 7.995781421661377,
      "learning_rate": 3.161860155071396e-05,
      "loss": 0.4141,
      "step": 27291
    },
    {
      "epoch": 8.61761919797916,
      "grad_norm": 6.593349933624268,
      "learning_rate": 3.161772445005789e-05,
      "loss": 0.5361,
      "step": 27292
    },
    {
      "epoch": 8.617934954215345,
      "grad_norm": 4.864131927490234,
      "learning_rate": 3.1616847349401816e-05,
      "loss": 0.3194,
      "step": 27293
    },
    {
      "epoch": 8.618250710451532,
      "grad_norm": 6.969161510467529,
      "learning_rate": 3.161597024874575e-05,
      "loss": 0.5116,
      "step": 27294
    },
    {
      "epoch": 8.618566466687717,
      "grad_norm": 8.5075044631958,
      "learning_rate": 3.161509314808967e-05,
      "loss": 0.6136,
      "step": 27295
    },
    {
      "epoch": 8.618882222923903,
      "grad_norm": 5.712631702423096,
      "learning_rate": 3.1614216047433605e-05,
      "loss": 0.4723,
      "step": 27296
    },
    {
      "epoch": 8.619197979160088,
      "grad_norm": 6.916195392608643,
      "learning_rate": 3.161333894677753e-05,
      "loss": 0.483,
      "step": 27297
    },
    {
      "epoch": 8.619513735396275,
      "grad_norm": 7.936459541320801,
      "learning_rate": 3.161246184612146e-05,
      "loss": 0.3746,
      "step": 27298
    },
    {
      "epoch": 8.61982949163246,
      "grad_norm": 6.6040873527526855,
      "learning_rate": 3.1611584745465386e-05,
      "loss": 0.3301,
      "step": 27299
    },
    {
      "epoch": 8.620145247868646,
      "grad_norm": 7.93479061126709,
      "learning_rate": 3.161070764480932e-05,
      "loss": 0.3683,
      "step": 27300
    },
    {
      "epoch": 8.62046100410483,
      "grad_norm": 5.228682994842529,
      "learning_rate": 3.160983054415325e-05,
      "loss": 0.492,
      "step": 27301
    },
    {
      "epoch": 8.620776760341016,
      "grad_norm": 5.125072956085205,
      "learning_rate": 3.1608953443497174e-05,
      "loss": 0.5068,
      "step": 27302
    },
    {
      "epoch": 8.621092516577203,
      "grad_norm": 6.530675888061523,
      "learning_rate": 3.1608076342841106e-05,
      "loss": 0.4894,
      "step": 27303
    },
    {
      "epoch": 8.621408272813389,
      "grad_norm": 8.961101531982422,
      "learning_rate": 3.160719924218504e-05,
      "loss": 0.3869,
      "step": 27304
    },
    {
      "epoch": 8.621724029049574,
      "grad_norm": 6.085288047790527,
      "learning_rate": 3.160632214152896e-05,
      "loss": 0.3559,
      "step": 27305
    },
    {
      "epoch": 8.622039785285759,
      "grad_norm": 5.867044925689697,
      "learning_rate": 3.1605445040872894e-05,
      "loss": 0.4005,
      "step": 27306
    },
    {
      "epoch": 8.622355541521944,
      "grad_norm": 6.674447536468506,
      "learning_rate": 3.1604567940216826e-05,
      "loss": 0.4642,
      "step": 27307
    },
    {
      "epoch": 8.622671297758131,
      "grad_norm": 7.014064311981201,
      "learning_rate": 3.160369083956075e-05,
      "loss": 0.4255,
      "step": 27308
    },
    {
      "epoch": 8.622987053994317,
      "grad_norm": 5.88092041015625,
      "learning_rate": 3.160281373890468e-05,
      "loss": 0.4068,
      "step": 27309
    },
    {
      "epoch": 8.623302810230502,
      "grad_norm": 6.543643951416016,
      "learning_rate": 3.160193663824861e-05,
      "loss": 0.4293,
      "step": 27310
    },
    {
      "epoch": 8.623618566466687,
      "grad_norm": 6.710465431213379,
      "learning_rate": 3.160105953759254e-05,
      "loss": 0.5002,
      "step": 27311
    },
    {
      "epoch": 8.623934322702873,
      "grad_norm": 5.36889123916626,
      "learning_rate": 3.1600182436936464e-05,
      "loss": 0.4948,
      "step": 27312
    },
    {
      "epoch": 8.62425007893906,
      "grad_norm": 5.827184200286865,
      "learning_rate": 3.1599305336280396e-05,
      "loss": 0.4919,
      "step": 27313
    },
    {
      "epoch": 8.624565835175245,
      "grad_norm": 6.771191596984863,
      "learning_rate": 3.159842823562432e-05,
      "loss": 0.4698,
      "step": 27314
    },
    {
      "epoch": 8.62488159141143,
      "grad_norm": 5.264606952667236,
      "learning_rate": 3.159755113496825e-05,
      "loss": 0.4007,
      "step": 27315
    },
    {
      "epoch": 8.625197347647616,
      "grad_norm": 6.045193195343018,
      "learning_rate": 3.159667403431218e-05,
      "loss": 0.5585,
      "step": 27316
    },
    {
      "epoch": 8.6255131038838,
      "grad_norm": 6.689236164093018,
      "learning_rate": 3.159579693365611e-05,
      "loss": 0.5148,
      "step": 27317
    },
    {
      "epoch": 8.625828860119988,
      "grad_norm": 5.699181079864502,
      "learning_rate": 3.1594919833000034e-05,
      "loss": 0.3997,
      "step": 27318
    },
    {
      "epoch": 8.626144616356173,
      "grad_norm": 6.306108474731445,
      "learning_rate": 3.1594042732343965e-05,
      "loss": 0.4277,
      "step": 27319
    },
    {
      "epoch": 8.626460372592359,
      "grad_norm": 5.982632637023926,
      "learning_rate": 3.159316563168789e-05,
      "loss": 0.5496,
      "step": 27320
    },
    {
      "epoch": 8.626776128828544,
      "grad_norm": 5.947962760925293,
      "learning_rate": 3.159228853103182e-05,
      "loss": 0.4898,
      "step": 27321
    },
    {
      "epoch": 8.62709188506473,
      "grad_norm": 6.878478050231934,
      "learning_rate": 3.159141143037575e-05,
      "loss": 0.52,
      "step": 27322
    },
    {
      "epoch": 8.627407641300916,
      "grad_norm": 5.291895866394043,
      "learning_rate": 3.159053432971968e-05,
      "loss": 0.3989,
      "step": 27323
    },
    {
      "epoch": 8.627723397537101,
      "grad_norm": 6.1512770652771,
      "learning_rate": 3.158965722906361e-05,
      "loss": 0.5402,
      "step": 27324
    },
    {
      "epoch": 8.628039153773287,
      "grad_norm": 5.586914539337158,
      "learning_rate": 3.1588780128407535e-05,
      "loss": 0.4879,
      "step": 27325
    },
    {
      "epoch": 8.628354910009472,
      "grad_norm": 6.839482307434082,
      "learning_rate": 3.158790302775147e-05,
      "loss": 0.471,
      "step": 27326
    },
    {
      "epoch": 8.62867066624566,
      "grad_norm": 5.592315196990967,
      "learning_rate": 3.158702592709539e-05,
      "loss": 0.4513,
      "step": 27327
    },
    {
      "epoch": 8.628986422481844,
      "grad_norm": 6.53555965423584,
      "learning_rate": 3.1586148826439324e-05,
      "loss": 0.4599,
      "step": 27328
    },
    {
      "epoch": 8.62930217871803,
      "grad_norm": 5.195647239685059,
      "learning_rate": 3.158527172578325e-05,
      "loss": 0.4608,
      "step": 27329
    },
    {
      "epoch": 8.629617934954215,
      "grad_norm": 5.159511089324951,
      "learning_rate": 3.158439462512719e-05,
      "loss": 0.3189,
      "step": 27330
    },
    {
      "epoch": 8.6299336911904,
      "grad_norm": 7.450114727020264,
      "learning_rate": 3.158351752447111e-05,
      "loss": 0.4824,
      "step": 27331
    },
    {
      "epoch": 8.630249447426587,
      "grad_norm": 5.626876354217529,
      "learning_rate": 3.1582640423815043e-05,
      "loss": 0.429,
      "step": 27332
    },
    {
      "epoch": 8.630565203662773,
      "grad_norm": 5.652380466461182,
      "learning_rate": 3.158176332315897e-05,
      "loss": 0.3932,
      "step": 27333
    },
    {
      "epoch": 8.630880959898958,
      "grad_norm": 5.5681681632995605,
      "learning_rate": 3.15808862225029e-05,
      "loss": 0.3855,
      "step": 27334
    },
    {
      "epoch": 8.631196716135143,
      "grad_norm": 4.7447357177734375,
      "learning_rate": 3.1580009121846825e-05,
      "loss": 0.4019,
      "step": 27335
    },
    {
      "epoch": 8.631512472371329,
      "grad_norm": 6.330123424530029,
      "learning_rate": 3.157913202119076e-05,
      "loss": 0.4832,
      "step": 27336
    },
    {
      "epoch": 8.631828228607516,
      "grad_norm": 5.720104694366455,
      "learning_rate": 3.157825492053468e-05,
      "loss": 0.4621,
      "step": 27337
    },
    {
      "epoch": 8.632143984843701,
      "grad_norm": 6.706615447998047,
      "learning_rate": 3.157737781987861e-05,
      "loss": 0.3389,
      "step": 27338
    },
    {
      "epoch": 8.632459741079886,
      "grad_norm": 4.823704719543457,
      "learning_rate": 3.157650071922254e-05,
      "loss": 0.4002,
      "step": 27339
    },
    {
      "epoch": 8.632775497316072,
      "grad_norm": 6.705187797546387,
      "learning_rate": 3.157562361856647e-05,
      "loss": 0.4956,
      "step": 27340
    },
    {
      "epoch": 8.633091253552259,
      "grad_norm": 6.82246208190918,
      "learning_rate": 3.1574746517910395e-05,
      "loss": 0.4436,
      "step": 27341
    },
    {
      "epoch": 8.633407009788444,
      "grad_norm": 7.464452266693115,
      "learning_rate": 3.1573869417254326e-05,
      "loss": 0.5211,
      "step": 27342
    },
    {
      "epoch": 8.63372276602463,
      "grad_norm": 6.289798736572266,
      "learning_rate": 3.157299231659825e-05,
      "loss": 0.4259,
      "step": 27343
    },
    {
      "epoch": 8.634038522260814,
      "grad_norm": 4.986855983734131,
      "learning_rate": 3.157211521594218e-05,
      "loss": 0.333,
      "step": 27344
    },
    {
      "epoch": 8.634354278497,
      "grad_norm": 5.098315715789795,
      "learning_rate": 3.1571238115286115e-05,
      "loss": 0.4295,
      "step": 27345
    },
    {
      "epoch": 8.634670034733187,
      "grad_norm": 6.888010501861572,
      "learning_rate": 3.157036101463004e-05,
      "loss": 0.4229,
      "step": 27346
    },
    {
      "epoch": 8.634985790969372,
      "grad_norm": 7.023077011108398,
      "learning_rate": 3.156948391397397e-05,
      "loss": 0.3945,
      "step": 27347
    },
    {
      "epoch": 8.635301547205557,
      "grad_norm": 7.394742965698242,
      "learning_rate": 3.1568606813317896e-05,
      "loss": 0.4712,
      "step": 27348
    },
    {
      "epoch": 8.635617303441743,
      "grad_norm": 8.146416664123535,
      "learning_rate": 3.156772971266183e-05,
      "loss": 0.4695,
      "step": 27349
    },
    {
      "epoch": 8.635933059677928,
      "grad_norm": 6.880133152008057,
      "learning_rate": 3.156685261200575e-05,
      "loss": 0.4571,
      "step": 27350
    },
    {
      "epoch": 8.636248815914115,
      "grad_norm": 6.8234171867370605,
      "learning_rate": 3.1565975511349684e-05,
      "loss": 0.4551,
      "step": 27351
    },
    {
      "epoch": 8.6365645721503,
      "grad_norm": 6.246387004852295,
      "learning_rate": 3.156509841069361e-05,
      "loss": 0.4914,
      "step": 27352
    },
    {
      "epoch": 8.636880328386486,
      "grad_norm": 8.244661331176758,
      "learning_rate": 3.156422131003754e-05,
      "loss": 0.4745,
      "step": 27353
    },
    {
      "epoch": 8.637196084622671,
      "grad_norm": 6.324538230895996,
      "learning_rate": 3.1563344209381466e-05,
      "loss": 0.3568,
      "step": 27354
    },
    {
      "epoch": 8.637511840858856,
      "grad_norm": 6.393261432647705,
      "learning_rate": 3.15624671087254e-05,
      "loss": 0.3963,
      "step": 27355
    },
    {
      "epoch": 8.637827597095043,
      "grad_norm": 6.208563804626465,
      "learning_rate": 3.156159000806932e-05,
      "loss": 0.4344,
      "step": 27356
    },
    {
      "epoch": 8.638143353331229,
      "grad_norm": 5.774938583374023,
      "learning_rate": 3.1560712907413254e-05,
      "loss": 0.3436,
      "step": 27357
    },
    {
      "epoch": 8.638459109567414,
      "grad_norm": 6.987234115600586,
      "learning_rate": 3.1559835806757186e-05,
      "loss": 0.4574,
      "step": 27358
    },
    {
      "epoch": 8.6387748658036,
      "grad_norm": 7.060830116271973,
      "learning_rate": 3.155895870610112e-05,
      "loss": 0.3626,
      "step": 27359
    },
    {
      "epoch": 8.639090622039785,
      "grad_norm": 6.945664405822754,
      "learning_rate": 3.155808160544504e-05,
      "loss": 0.4895,
      "step": 27360
    },
    {
      "epoch": 8.639406378275972,
      "grad_norm": 5.972715854644775,
      "learning_rate": 3.1557204504788974e-05,
      "loss": 0.3712,
      "step": 27361
    },
    {
      "epoch": 8.639722134512157,
      "grad_norm": 6.919093608856201,
      "learning_rate": 3.15563274041329e-05,
      "loss": 0.3527,
      "step": 27362
    },
    {
      "epoch": 8.640037890748342,
      "grad_norm": 5.438526153564453,
      "learning_rate": 3.155545030347683e-05,
      "loss": 0.421,
      "step": 27363
    },
    {
      "epoch": 8.640353646984527,
      "grad_norm": 4.846860885620117,
      "learning_rate": 3.1554573202820756e-05,
      "loss": 0.3074,
      "step": 27364
    },
    {
      "epoch": 8.640669403220713,
      "grad_norm": 7.99388313293457,
      "learning_rate": 3.155369610216469e-05,
      "loss": 0.4532,
      "step": 27365
    },
    {
      "epoch": 8.6409851594569,
      "grad_norm": 6.12407922744751,
      "learning_rate": 3.155281900150861e-05,
      "loss": 0.4079,
      "step": 27366
    },
    {
      "epoch": 8.641300915693085,
      "grad_norm": 5.2844557762146,
      "learning_rate": 3.1551941900852544e-05,
      "loss": 0.3289,
      "step": 27367
    },
    {
      "epoch": 8.64161667192927,
      "grad_norm": 6.557043075561523,
      "learning_rate": 3.1551064800196476e-05,
      "loss": 0.4644,
      "step": 27368
    },
    {
      "epoch": 8.641932428165456,
      "grad_norm": 5.5123209953308105,
      "learning_rate": 3.15501876995404e-05,
      "loss": 0.4655,
      "step": 27369
    },
    {
      "epoch": 8.642248184401641,
      "grad_norm": 5.772370338439941,
      "learning_rate": 3.154931059888433e-05,
      "loss": 0.3476,
      "step": 27370
    },
    {
      "epoch": 8.642563940637828,
      "grad_norm": 6.378408908843994,
      "learning_rate": 3.154843349822826e-05,
      "loss": 0.4766,
      "step": 27371
    },
    {
      "epoch": 8.642879696874013,
      "grad_norm": 6.493675708770752,
      "learning_rate": 3.154755639757219e-05,
      "loss": 0.4142,
      "step": 27372
    },
    {
      "epoch": 8.643195453110199,
      "grad_norm": 6.101130962371826,
      "learning_rate": 3.1546679296916114e-05,
      "loss": 0.3792,
      "step": 27373
    },
    {
      "epoch": 8.643511209346384,
      "grad_norm": 5.230980396270752,
      "learning_rate": 3.1545802196260045e-05,
      "loss": 0.348,
      "step": 27374
    },
    {
      "epoch": 8.643826965582571,
      "grad_norm": 7.575222969055176,
      "learning_rate": 3.154492509560397e-05,
      "loss": 0.5083,
      "step": 27375
    },
    {
      "epoch": 8.644142721818756,
      "grad_norm": 6.780185699462891,
      "learning_rate": 3.15440479949479e-05,
      "loss": 0.4474,
      "step": 27376
    },
    {
      "epoch": 8.644458478054942,
      "grad_norm": 8.178790092468262,
      "learning_rate": 3.154317089429183e-05,
      "loss": 0.6068,
      "step": 27377
    },
    {
      "epoch": 8.644774234291127,
      "grad_norm": 5.605228900909424,
      "learning_rate": 3.154229379363576e-05,
      "loss": 0.357,
      "step": 27378
    },
    {
      "epoch": 8.645089990527312,
      "grad_norm": 6.2917070388793945,
      "learning_rate": 3.1541416692979683e-05,
      "loss": 0.3746,
      "step": 27379
    },
    {
      "epoch": 8.6454057467635,
      "grad_norm": 7.668536186218262,
      "learning_rate": 3.1540539592323615e-05,
      "loss": 0.4473,
      "step": 27380
    },
    {
      "epoch": 8.645721502999685,
      "grad_norm": 5.60540246963501,
      "learning_rate": 3.153966249166754e-05,
      "loss": 0.3984,
      "step": 27381
    },
    {
      "epoch": 8.64603725923587,
      "grad_norm": 5.595595836639404,
      "learning_rate": 3.153878539101147e-05,
      "loss": 0.3549,
      "step": 27382
    },
    {
      "epoch": 8.646353015472055,
      "grad_norm": 5.621836185455322,
      "learning_rate": 3.1537908290355403e-05,
      "loss": 0.3995,
      "step": 27383
    },
    {
      "epoch": 8.64666877170824,
      "grad_norm": 5.821145534515381,
      "learning_rate": 3.153703118969933e-05,
      "loss": 0.4252,
      "step": 27384
    },
    {
      "epoch": 8.646984527944428,
      "grad_norm": 6.282121181488037,
      "learning_rate": 3.153615408904326e-05,
      "loss": 0.5032,
      "step": 27385
    },
    {
      "epoch": 8.647300284180613,
      "grad_norm": 4.932888984680176,
      "learning_rate": 3.153527698838719e-05,
      "loss": 0.3813,
      "step": 27386
    },
    {
      "epoch": 8.647616040416798,
      "grad_norm": 8.129733085632324,
      "learning_rate": 3.1534399887731117e-05,
      "loss": 0.4834,
      "step": 27387
    },
    {
      "epoch": 8.647931796652983,
      "grad_norm": 7.307932376861572,
      "learning_rate": 3.153352278707505e-05,
      "loss": 0.3338,
      "step": 27388
    },
    {
      "epoch": 8.64824755288917,
      "grad_norm": 5.236300468444824,
      "learning_rate": 3.153264568641898e-05,
      "loss": 0.4787,
      "step": 27389
    },
    {
      "epoch": 8.648563309125356,
      "grad_norm": 5.820810794830322,
      "learning_rate": 3.1531768585762905e-05,
      "loss": 0.4202,
      "step": 27390
    },
    {
      "epoch": 8.648879065361541,
      "grad_norm": 5.382250785827637,
      "learning_rate": 3.1530891485106837e-05,
      "loss": 0.3839,
      "step": 27391
    },
    {
      "epoch": 8.649194821597726,
      "grad_norm": 6.440791606903076,
      "learning_rate": 3.153001438445076e-05,
      "loss": 0.3696,
      "step": 27392
    },
    {
      "epoch": 8.649510577833912,
      "grad_norm": 6.135524272918701,
      "learning_rate": 3.152913728379469e-05,
      "loss": 0.3749,
      "step": 27393
    },
    {
      "epoch": 8.649826334070099,
      "grad_norm": 7.1851606369018555,
      "learning_rate": 3.152826018313862e-05,
      "loss": 0.4752,
      "step": 27394
    },
    {
      "epoch": 8.650142090306284,
      "grad_norm": 7.521263599395752,
      "learning_rate": 3.152738308248255e-05,
      "loss": 0.4751,
      "step": 27395
    },
    {
      "epoch": 8.65045784654247,
      "grad_norm": 6.659304141998291,
      "learning_rate": 3.1526505981826475e-05,
      "loss": 0.4959,
      "step": 27396
    },
    {
      "epoch": 8.650773602778655,
      "grad_norm": 5.064731121063232,
      "learning_rate": 3.1525628881170406e-05,
      "loss": 0.3663,
      "step": 27397
    },
    {
      "epoch": 8.65108935901484,
      "grad_norm": 7.8676981925964355,
      "learning_rate": 3.152475178051433e-05,
      "loss": 0.4592,
      "step": 27398
    },
    {
      "epoch": 8.651405115251027,
      "grad_norm": 6.174574375152588,
      "learning_rate": 3.152387467985826e-05,
      "loss": 0.5102,
      "step": 27399
    },
    {
      "epoch": 8.651720871487212,
      "grad_norm": 7.454016208648682,
      "learning_rate": 3.152299757920219e-05,
      "loss": 0.3596,
      "step": 27400
    },
    {
      "epoch": 8.651720871487212,
      "eval_loss": 0.9655474424362183,
      "eval_runtime": 4.6666,
      "eval_samples_per_second": 533.584,
      "eval_steps_per_second": 4.5,
      "step": 27400
    }
  ],
  "logging_steps": 1,
  "max_steps": 63340,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 20,
  "save_steps": 200,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 124,
  "trial_name": null,
  "trial_params": null
}