{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.026143702519938035,
  "eval_steps": 73,
  "global_step": 3869,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 7.270200174029405,
      "learning_rate": 2e-07,
      "loss": 1.5886,
      "step": 1
    },
    {
      "epoch": 0.0,
      "eval_loss": 1.8586688041687012,
      "eval_runtime": 4.4089,
      "eval_samples_per_second": 2.041,
      "eval_steps_per_second": 1.134,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.088138615023864,
      "learning_rate": 4e-07,
      "loss": 1.4809,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.409449066839206,
      "learning_rate": 6e-07,
      "loss": 1.6585,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.366466325593624,
      "learning_rate": 8e-07,
      "loss": 1.6965,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.247338916406195,
      "learning_rate": 1e-06,
      "loss": 1.532,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.049902060629968,
      "learning_rate": 1.2e-06,
      "loss": 1.596,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.1820303407795745,
      "learning_rate": 1.4e-06,
      "loss": 1.5625,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.7480024222126165,
      "learning_rate": 1.6e-06,
      "loss": 1.4882,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.09119480827286,
      "learning_rate": 1.8e-06,
      "loss": 1.5345,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.701895875897247,
      "learning_rate": 2e-06,
      "loss": 1.5403,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.444180689164233,
      "learning_rate": 1.9999999999859167e-06,
      "loss": 1.7593,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.671206404255702,
      "learning_rate": 1.9999999999436667e-06,
      "loss": 1.5666,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.879938934255732,
      "learning_rate": 1.99999999987325e-06,
      "loss": 1.5774,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.598165698155055,
      "learning_rate": 1.9999999997746663e-06,
      "loss": 1.3795,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.241997738434293,
      "learning_rate": 1.9999999996479163e-06,
      "loss": 1.5449,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.92151785969858,
      "learning_rate": 1.999999999493e-06,
      "loss": 1.6951,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.539525947721768,
      "learning_rate": 1.9999999993099162e-06,
      "loss": 1.5409,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.351267308329018,
      "learning_rate": 1.9999999990986662e-06,
      "loss": 1.426,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.705234009184205,
      "learning_rate": 1.99999999885925e-06,
      "loss": 1.5517,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.5021957856088175,
      "learning_rate": 1.999999998591666e-06,
      "loss": 1.6403,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.230793498668246,
      "learning_rate": 1.999999998295916e-06,
      "loss": 1.3916,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.632016493197849,
      "learning_rate": 1.9999999979719993e-06,
      "loss": 1.5029,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.224094987664232,
      "learning_rate": 1.999999997619916e-06,
      "loss": 1.5172,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.9052975256388995,
      "learning_rate": 1.9999999972396656e-06,
      "loss": 1.2809,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.749261607817126,
      "learning_rate": 1.9999999968312492e-06,
      "loss": 1.682,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.255433017333782,
      "learning_rate": 1.9999999963946656e-06,
      "loss": 1.5094,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.093790043356135,
      "learning_rate": 1.9999999959299155e-06,
      "loss": 1.4338,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.001582030581129,
      "learning_rate": 1.9999999954369987e-06,
      "loss": 1.5456,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.863618311056822,
      "learning_rate": 1.999999994915915e-06,
      "loss": 1.5252,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.926702449626208,
      "learning_rate": 1.9999999943666654e-06,
      "loss": 1.5579,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.798968821403948,
      "learning_rate": 1.9999999937892486e-06,
      "loss": 1.5616,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.306082122633555,
      "learning_rate": 1.999999993183665e-06,
      "loss": 1.6885,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.324360862135967,
      "learning_rate": 1.999999992549915e-06,
      "loss": 1.6009,
      "step": 33
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.712084687468965,
      "learning_rate": 1.9999999918879976e-06,
      "loss": 1.5022,
      "step": 34
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.562761534033398,
      "learning_rate": 1.9999999911979143e-06,
      "loss": 1.7051,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.4469400814171,
      "learning_rate": 1.9999999904796643e-06,
      "loss": 1.5599,
      "step": 36
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.361519499368759,
      "learning_rate": 1.9999999897332474e-06,
      "loss": 1.5086,
      "step": 37
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.9815078648545965,
      "learning_rate": 1.9999999889586637e-06,
      "loss": 1.5386,
      "step": 38
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.162059963508055,
      "learning_rate": 1.9999999881559136e-06,
      "loss": 1.4769,
      "step": 39
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.122293609109329,
      "learning_rate": 1.9999999873249968e-06,
      "loss": 1.561,
      "step": 40
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.152152754881128,
      "learning_rate": 1.999999986465913e-06,
      "loss": 1.5329,
      "step": 41
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.991233383631504,
      "learning_rate": 1.999999985578663e-06,
      "loss": 1.6348,
      "step": 42
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.1180827908562545,
      "learning_rate": 1.999999984663246e-06,
      "loss": 1.5922,
      "step": 43
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.447368695015986,
      "learning_rate": 1.9999999837196624e-06,
      "loss": 1.3119,
      "step": 44
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.058434450512851,
      "learning_rate": 1.9999999827479124e-06,
      "loss": 1.4098,
      "step": 45
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.792299679667143,
      "learning_rate": 1.9999999817479955e-06,
      "loss": 1.427,
      "step": 46
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.86507140594805,
      "learning_rate": 1.9999999807199118e-06,
      "loss": 1.6528,
      "step": 47
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.121726652742493,
      "learning_rate": 1.9999999796636617e-06,
      "loss": 1.4907,
      "step": 48
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.897225770537551,
      "learning_rate": 1.9999999785792444e-06,
      "loss": 1.5129,
      "step": 49
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.447285160025109,
      "learning_rate": 1.999999977466661e-06,
      "loss": 1.4732,
      "step": 50
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.435992424750984,
      "learning_rate": 1.9999999763259106e-06,
      "loss": 1.5531,
      "step": 51
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.126344987263065,
      "learning_rate": 1.9999999751569937e-06,
      "loss": 1.4008,
      "step": 52
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.003394959358567,
      "learning_rate": 1.99999997395991e-06,
      "loss": 1.5076,
      "step": 53
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.375009945600638,
      "learning_rate": 1.99999997273466e-06,
      "loss": 1.3866,
      "step": 54
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.849278700300698,
      "learning_rate": 1.999999971481243e-06,
      "loss": 1.427,
      "step": 55
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.532239141247253,
      "learning_rate": 1.9999999701996592e-06,
      "loss": 1.6364,
      "step": 56
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.438678986247806,
      "learning_rate": 1.999999968889909e-06,
      "loss": 1.4648,
      "step": 57
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.393102110640265,
      "learning_rate": 1.9999999675519918e-06,
      "loss": 1.5226,
      "step": 58
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.602009023037625,
      "learning_rate": 1.999999966185908e-06,
      "loss": 1.745,
      "step": 59
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.30661712292468,
      "learning_rate": 1.999999964791658e-06,
      "loss": 1.5059,
      "step": 60
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.15102423620726,
      "learning_rate": 1.999999963369241e-06,
      "loss": 1.3974,
      "step": 61
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.870501279592248,
      "learning_rate": 1.9999999619186573e-06,
      "loss": 1.5026,
      "step": 62
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.270272741292025,
      "learning_rate": 1.9999999604399067e-06,
      "loss": 1.6544,
      "step": 63
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.886919913741941,
      "learning_rate": 1.99999995893299e-06,
      "loss": 1.5181,
      "step": 64
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.05398621579166,
      "learning_rate": 1.999999957397906e-06,
      "loss": 1.5616,
      "step": 65
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.836134104240407,
      "learning_rate": 1.9999999558346555e-06,
      "loss": 1.54,
      "step": 66
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.082059874604205,
      "learning_rate": 1.9999999542432386e-06,
      "loss": 1.5239,
      "step": 67
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.874373987895782,
      "learning_rate": 1.999999952623655e-06,
      "loss": 1.5878,
      "step": 68
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.946163953606462,
      "learning_rate": 1.9999999509759043e-06,
      "loss": 1.4434,
      "step": 69
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.876405244918954,
      "learning_rate": 1.9999999492999873e-06,
      "loss": 1.4608,
      "step": 70
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.5778213104294005,
      "learning_rate": 1.9999999475959036e-06,
      "loss": 1.5291,
      "step": 71
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.99889754273985,
      "learning_rate": 1.9999999458636535e-06,
      "loss": 1.3987,
      "step": 72
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.130434143656013,
      "learning_rate": 1.999999944103236e-06,
      "loss": 1.6093,
      "step": 73
    },
    {
      "epoch": 0.0,
      "eval_loss": 1.7611451148986816,
      "eval_runtime": 4.6217,
      "eval_samples_per_second": 1.947,
      "eval_steps_per_second": 1.082,
      "step": 73
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.212510949695165,
      "learning_rate": 1.9999999423146523e-06,
      "loss": 1.7044,
      "step": 74
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.2823971109355705,
      "learning_rate": 1.999999940497902e-06,
      "loss": 1.7013,
      "step": 75
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.045619546308826,
      "learning_rate": 1.999999938652985e-06,
      "loss": 1.5755,
      "step": 76
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.87457170524847,
      "learning_rate": 1.999999936779901e-06,
      "loss": 1.6232,
      "step": 77
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.082677620793644,
      "learning_rate": 1.999999934878651e-06,
      "loss": 1.6031,
      "step": 78
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.872910823814863,
      "learning_rate": 1.9999999329492335e-06,
      "loss": 1.5715,
      "step": 79
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.244165093478454,
      "learning_rate": 1.9999999309916497e-06,
      "loss": 1.3898,
      "step": 80
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.113300123505611,
      "learning_rate": 1.999999929005899e-06,
      "loss": 1.5425,
      "step": 81
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.028851212735335,
      "learning_rate": 1.999999926991982e-06,
      "loss": 1.5812,
      "step": 82
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.11348885382359,
      "learning_rate": 1.9999999249498984e-06,
      "loss": 1.4328,
      "step": 83
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.535550818095342,
      "learning_rate": 1.9999999228796482e-06,
      "loss": 1.6637,
      "step": 84
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.007537107122481,
      "learning_rate": 1.999999920781231e-06,
      "loss": 1.5714,
      "step": 85
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.267096315507558,
      "learning_rate": 1.9999999186546466e-06,
      "loss": 1.5447,
      "step": 86
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.796533721972671,
      "learning_rate": 1.9999999164998964e-06,
      "loss": 1.5971,
      "step": 87
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.228074731568103,
      "learning_rate": 1.999999914316979e-06,
      "loss": 1.7089,
      "step": 88
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.7122834140340295,
      "learning_rate": 1.9999999121058956e-06,
      "loss": 1.2932,
      "step": 89
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.2806919781906965,
      "learning_rate": 1.999999909866645e-06,
      "loss": 1.653,
      "step": 90
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.368043140767265,
      "learning_rate": 1.9999999075992276e-06,
      "loss": 1.6132,
      "step": 91
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.76243769319436,
      "learning_rate": 1.999999905303644e-06,
      "loss": 1.6384,
      "step": 92
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.672061728664766,
      "learning_rate": 1.9999999029798932e-06,
      "loss": 1.4254,
      "step": 93
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.002515171971739,
      "learning_rate": 1.9999999006279762e-06,
      "loss": 1.4739,
      "step": 94
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.2077060992013084,
      "learning_rate": 1.9999998982478924e-06,
      "loss": 1.4923,
      "step": 95
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.167170360061274,
      "learning_rate": 1.999999895839642e-06,
      "loss": 1.6129,
      "step": 96
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.886352788159953,
      "learning_rate": 1.9999998934032244e-06,
      "loss": 1.4177,
      "step": 97
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.408894422523236,
      "learning_rate": 1.999999890938641e-06,
      "loss": 1.6144,
      "step": 98
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.490500856061367,
      "learning_rate": 1.9999998884458904e-06,
      "loss": 1.5951,
      "step": 99
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.1356644085000065,
      "learning_rate": 1.9999998859249733e-06,
      "loss": 1.587,
      "step": 100
    },
    {
      "epoch": 0.0,
      "grad_norm": 14.038552559799147,
      "learning_rate": 1.999999883375889e-06,
      "loss": 1.2883,
      "step": 101
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.226960512898749,
      "learning_rate": 1.999999880798639e-06,
      "loss": 1.593,
      "step": 102
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.414801916717659,
      "learning_rate": 1.9999998781932214e-06,
      "loss": 1.444,
      "step": 103
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.966168140745907,
      "learning_rate": 1.9999998755596376e-06,
      "loss": 1.4407,
      "step": 104
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.159957368813785,
      "learning_rate": 1.999999872897887e-06,
      "loss": 1.465,
      "step": 105
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.6281356469376655,
      "learning_rate": 1.9999998702079695e-06,
      "loss": 1.5183,
      "step": 106
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.823355492437726,
      "learning_rate": 1.9999998674898857e-06,
      "loss": 1.3937,
      "step": 107
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.900910155703639,
      "learning_rate": 1.999999864743635e-06,
      "loss": 1.5294,
      "step": 108
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.025636152106037,
      "learning_rate": 1.999999861969218e-06,
      "loss": 1.6249,
      "step": 109
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.116169125449381,
      "learning_rate": 1.999999859166634e-06,
      "loss": 1.4079,
      "step": 110
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.778904603716487,
      "learning_rate": 1.999999856335883e-06,
      "loss": 1.5865,
      "step": 111
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.756222368715884,
      "learning_rate": 1.999999853476966e-06,
      "loss": 1.7328,
      "step": 112
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.412891077813045,
      "learning_rate": 1.9999998505898822e-06,
      "loss": 1.6096,
      "step": 113
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.918651260002986,
      "learning_rate": 1.9999998476746316e-06,
      "loss": 1.4487,
      "step": 114
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.5264820158788845,
      "learning_rate": 1.9999998447312145e-06,
      "loss": 1.5762,
      "step": 115
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.9444017993692375,
      "learning_rate": 1.9999998417596302e-06,
      "loss": 1.6108,
      "step": 116
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.3741051561343065,
      "learning_rate": 1.9999998387598796e-06,
      "loss": 1.4945,
      "step": 117
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.086038907073674,
      "learning_rate": 1.9999998357319625e-06,
      "loss": 1.4727,
      "step": 118
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.374970852866187,
      "learning_rate": 1.9999998326758787e-06,
      "loss": 1.5108,
      "step": 119
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.046226292058247,
      "learning_rate": 1.999999829591628e-06,
      "loss": 1.547,
      "step": 120
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.920252413086519,
      "learning_rate": 1.9999998264792105e-06,
      "loss": 1.5885,
      "step": 121
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.9199662171825,
      "learning_rate": 1.9999998233386266e-06,
      "loss": 1.5461,
      "step": 122
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.849224216338868,
      "learning_rate": 1.9999998201698764e-06,
      "loss": 1.3918,
      "step": 123
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.122690985334921,
      "learning_rate": 1.999999816972959e-06,
      "loss": 1.7063,
      "step": 124
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.4642737814450495,
      "learning_rate": 1.999999813747875e-06,
      "loss": 1.4925,
      "step": 125
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.985722953022008,
      "learning_rate": 1.9999998104946243e-06,
      "loss": 1.6928,
      "step": 126
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.95531201412864,
      "learning_rate": 1.999999807213207e-06,
      "loss": 1.3453,
      "step": 127
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.9615775535071025,
      "learning_rate": 1.999999803903623e-06,
      "loss": 1.4873,
      "step": 128
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.995275322635195,
      "learning_rate": 1.9999998005658727e-06,
      "loss": 1.4477,
      "step": 129
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.042565222809337,
      "learning_rate": 1.999999797199955e-06,
      "loss": 1.4494,
      "step": 130
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.123208053454304,
      "learning_rate": 1.999999793805871e-06,
      "loss": 1.6236,
      "step": 131
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.125753408156544,
      "learning_rate": 1.9999997903836206e-06,
      "loss": 1.4867,
      "step": 132
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.953051829249889,
      "learning_rate": 1.999999786933203e-06,
      "loss": 1.4471,
      "step": 133
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.01487688139561,
      "learning_rate": 1.999999783454619e-06,
      "loss": 1.5202,
      "step": 134
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.9862855831810355,
      "learning_rate": 1.9999997799478685e-06,
      "loss": 1.2726,
      "step": 135
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.664158110626264,
      "learning_rate": 1.9999997764129514e-06,
      "loss": 1.453,
      "step": 136
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.515072838378242,
      "learning_rate": 1.9999997728498675e-06,
      "loss": 1.4493,
      "step": 137
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.638297190169102,
      "learning_rate": 1.9999997692586168e-06,
      "loss": 1.4888,
      "step": 138
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.88971951842883,
      "learning_rate": 1.9999997656391992e-06,
      "loss": 1.4683,
      "step": 139
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.907276772043835,
      "learning_rate": 1.9999997619916153e-06,
      "loss": 1.4255,
      "step": 140
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.130307226886763,
      "learning_rate": 1.9999997583158646e-06,
      "loss": 1.4295,
      "step": 141
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.878724833576819,
      "learning_rate": 1.9999997546119475e-06,
      "loss": 1.6857,
      "step": 142
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.889147537162318,
      "learning_rate": 1.9999997508798636e-06,
      "loss": 1.2668,
      "step": 143
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.079666872019813,
      "learning_rate": 1.9999997471196124e-06,
      "loss": 1.4661,
      "step": 144
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.896421399772935,
      "learning_rate": 1.9999997433311953e-06,
      "loss": 1.4462,
      "step": 145
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.904873841992958,
      "learning_rate": 1.9999997395146114e-06,
      "loss": 1.5005,
      "step": 146
    },
    {
      "epoch": 0.0,
      "eval_loss": 1.7462446689605713,
      "eval_runtime": 4.5857,
      "eval_samples_per_second": 1.963,
      "eval_steps_per_second": 1.09,
      "step": 146
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.780194658732313,
      "learning_rate": 1.9999997356698607e-06,
      "loss": 1.2281,
      "step": 147
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.947954070905395,
      "learning_rate": 1.9999997317969435e-06,
      "loss": 1.5465,
      "step": 148
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.262663754638768,
      "learning_rate": 1.9999997278958596e-06,
      "loss": 1.7203,
      "step": 149
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.980503880353371,
      "learning_rate": 1.999999723966609e-06,
      "loss": 1.5602,
      "step": 150
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.070738329007044,
      "learning_rate": 1.9999997200091917e-06,
      "loss": 1.4116,
      "step": 151
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.443381763060975,
      "learning_rate": 1.999999716023608e-06,
      "loss": 1.6409,
      "step": 152
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.836870306399049,
      "learning_rate": 1.999999712009857e-06,
      "loss": 1.3384,
      "step": 153
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.427614614695787,
      "learning_rate": 1.9999997079679395e-06,
      "loss": 1.4659,
      "step": 154
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.172785821462023,
      "learning_rate": 1.999999703897856e-06,
      "loss": 1.4818,
      "step": 155
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.300155696147022,
      "learning_rate": 1.9999996997996052e-06,
      "loss": 1.4057,
      "step": 156
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.918600876049223,
      "learning_rate": 1.9999996956731877e-06,
      "loss": 1.5059,
      "step": 157
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.916667098440986,
      "learning_rate": 1.999999691518604e-06,
      "loss": 1.4173,
      "step": 158
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.8728284717219905,
      "learning_rate": 1.999999687335853e-06,
      "loss": 1.3795,
      "step": 159
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.287532891218524,
      "learning_rate": 1.999999683124936e-06,
      "loss": 1.4764,
      "step": 160
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.987191884182559,
      "learning_rate": 1.999999678885852e-06,
      "loss": 1.5181,
      "step": 161
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.186035530995563,
      "learning_rate": 1.999999674618601e-06,
      "loss": 1.4576,
      "step": 162
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.918294071223459,
      "learning_rate": 1.999999670323184e-06,
      "loss": 1.4588,
      "step": 163
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.065760262070265,
      "learning_rate": 1.9999996659996e-06,
      "loss": 1.5348,
      "step": 164
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.017202795011469,
      "learning_rate": 1.999999661647849e-06,
      "loss": 1.5338,
      "step": 165
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.6648484254216225,
      "learning_rate": 1.999999657267932e-06,
      "loss": 1.6196,
      "step": 166
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.4750502865711965,
      "learning_rate": 1.999999652859848e-06,
      "loss": 1.5665,
      "step": 167
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.586077526270329,
      "learning_rate": 1.9999996484235977e-06,
      "loss": 1.5768,
      "step": 168
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.064098969414124,
      "learning_rate": 1.9999996439591805e-06,
      "loss": 1.5485,
      "step": 169
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.831991975556203,
      "learning_rate": 1.999999639466596e-06,
      "loss": 1.6605,
      "step": 170
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.386693105255357,
      "learning_rate": 1.9999996349458458e-06,
      "loss": 1.5403,
      "step": 171
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.383469172835463,
      "learning_rate": 1.9999996303969286e-06,
      "loss": 1.4123,
      "step": 172
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.136249870917203,
      "learning_rate": 1.9999996258198446e-06,
      "loss": 1.5223,
      "step": 173
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.497494715675666,
      "learning_rate": 1.999999621214594e-06,
      "loss": 1.4759,
      "step": 174
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.359298992694515,
      "learning_rate": 1.9999996165811766e-06,
      "loss": 1.5216,
      "step": 175
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.554159062585914,
      "learning_rate": 1.9999996119195926e-06,
      "loss": 1.3022,
      "step": 176
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.655783926270953,
      "learning_rate": 1.9999996072298423e-06,
      "loss": 1.6985,
      "step": 177
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.815355669323773,
      "learning_rate": 1.999999602511925e-06,
      "loss": 1.6323,
      "step": 178
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.090330865005731,
      "learning_rate": 1.999999597765841e-06,
      "loss": 1.4405,
      "step": 179
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.453237104330645,
      "learning_rate": 1.9999995929915903e-06,
      "loss": 1.6569,
      "step": 180
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.140949180379015,
      "learning_rate": 1.999999588189173e-06,
      "loss": 1.5043,
      "step": 181
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.312297253169766,
      "learning_rate": 1.999999583358589e-06,
      "loss": 1.4393,
      "step": 182
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.5602461468607745,
      "learning_rate": 1.9999995784998387e-06,
      "loss": 1.5034,
      "step": 183
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.984563051637153,
      "learning_rate": 1.9999995736129215e-06,
      "loss": 1.5951,
      "step": 184
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.210447188187252,
      "learning_rate": 1.9999995686978374e-06,
      "loss": 1.4994,
      "step": 185
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.986374115612521,
      "learning_rate": 1.999999563754587e-06,
      "loss": 1.3567,
      "step": 186
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.868262699823536,
      "learning_rate": 1.99999955878317e-06,
      "loss": 1.5591,
      "step": 187
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.532695287284847,
      "learning_rate": 1.999999553783586e-06,
      "loss": 1.6909,
      "step": 188
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.064305786007753,
      "learning_rate": 1.9999995487558354e-06,
      "loss": 1.6499,
      "step": 189
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.3615235293142645,
      "learning_rate": 1.999999543699918e-06,
      "loss": 1.5116,
      "step": 190
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.820360691431748,
      "learning_rate": 1.999999538615834e-06,
      "loss": 1.585,
      "step": 191
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.848927324752224,
      "learning_rate": 1.9999995335035838e-06,
      "loss": 1.5208,
      "step": 192
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.008810784644451,
      "learning_rate": 1.999999528363167e-06,
      "loss": 1.4648,
      "step": 193
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.012428143646969,
      "learning_rate": 1.999999523194583e-06,
      "loss": 1.5394,
      "step": 194
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.014801660701698,
      "learning_rate": 1.9999995179978325e-06,
      "loss": 1.5681,
      "step": 195
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.044808653366915,
      "learning_rate": 1.9999995127729153e-06,
      "loss": 1.5071,
      "step": 196
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.138850239659894,
      "learning_rate": 1.9999995075198317e-06,
      "loss": 1.5622,
      "step": 197
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.942376781806205,
      "learning_rate": 1.999999502238581e-06,
      "loss": 1.4336,
      "step": 198
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.454348532164084,
      "learning_rate": 1.999999496929164e-06,
      "loss": 1.5379,
      "step": 199
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.515385692801986,
      "learning_rate": 1.99999949159158e-06,
      "loss": 1.3406,
      "step": 200
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.124129865470126,
      "learning_rate": 1.9999994862258295e-06,
      "loss": 1.3091,
      "step": 201
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.04812044323597,
      "learning_rate": 1.9999994808319127e-06,
      "loss": 1.4656,
      "step": 202
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.832399296186657,
      "learning_rate": 1.9999994754098286e-06,
      "loss": 1.4201,
      "step": 203
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.180488033599416,
      "learning_rate": 1.999999469959578e-06,
      "loss": 1.385,
      "step": 204
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.190369524582554,
      "learning_rate": 1.9999994644811614e-06,
      "loss": 1.561,
      "step": 205
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.554453318862103,
      "learning_rate": 1.9999994589745773e-06,
      "loss": 1.67,
      "step": 206
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.035958799888537,
      "learning_rate": 1.999999453439827e-06,
      "loss": 1.3742,
      "step": 207
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.829678242084621,
      "learning_rate": 1.99999944787691e-06,
      "loss": 1.5053,
      "step": 208
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.67709513251063,
      "learning_rate": 1.999999442285826e-06,
      "loss": 1.6848,
      "step": 209
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.611739029374384,
      "learning_rate": 1.999999436666576e-06,
      "loss": 1.534,
      "step": 210
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.303224421644464,
      "learning_rate": 1.9999994310191587e-06,
      "loss": 1.4115,
      "step": 211
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.739064239711381,
      "learning_rate": 1.999999425343575e-06,
      "loss": 1.5154,
      "step": 212
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.998482123577443,
      "learning_rate": 1.999999419639825e-06,
      "loss": 1.4069,
      "step": 213
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.777605807898373,
      "learning_rate": 1.9999994139079077e-06,
      "loss": 1.4424,
      "step": 214
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.91838714700508,
      "learning_rate": 1.999999408147824e-06,
      "loss": 1.4662,
      "step": 215
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.080194035605747,
      "learning_rate": 1.999999402359574e-06,
      "loss": 1.5027,
      "step": 216
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.028785668898977,
      "learning_rate": 1.9999993965431567e-06,
      "loss": 1.4228,
      "step": 217
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.840444295707605,
      "learning_rate": 1.999999390698573e-06,
      "loss": 1.415,
      "step": 218
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.684443121932812,
      "learning_rate": 1.9999993848258226e-06,
      "loss": 1.5323,
      "step": 219
    },
    {
      "epoch": 0.0,
      "eval_loss": 1.7417716979980469,
      "eval_runtime": 4.6034,
      "eval_samples_per_second": 1.955,
      "eval_steps_per_second": 1.086,
      "step": 219
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.888070245726793,
      "learning_rate": 1.9999993789249058e-06,
      "loss": 1.3426,
      "step": 220
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.874791209837943,
      "learning_rate": 1.999999372995822e-06,
      "loss": 1.5123,
      "step": 221
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.351391729703106,
      "learning_rate": 1.999999367038572e-06,
      "loss": 1.511,
      "step": 222
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.252575209784133,
      "learning_rate": 1.999999361053155e-06,
      "loss": 1.5191,
      "step": 223
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.0084979782736765,
      "learning_rate": 1.9999993550395715e-06,
      "loss": 1.4319,
      "step": 224
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.032795555975619,
      "learning_rate": 1.999999348997821e-06,
      "loss": 1.5326,
      "step": 225
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.882732392699978,
      "learning_rate": 1.9999993429279045e-06,
      "loss": 1.5562,
      "step": 226
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.5857954829159,
      "learning_rate": 1.999999336829821e-06,
      "loss": 1.5575,
      "step": 227
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.368713396037795,
      "learning_rate": 1.9999993307035704e-06,
      "loss": 1.5341,
      "step": 228
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.564926482513187,
      "learning_rate": 1.999999324549154e-06,
      "loss": 1.5136,
      "step": 229
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.87983027120232,
      "learning_rate": 1.9999993183665702e-06,
      "loss": 1.5424,
      "step": 230
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.130333500849336,
      "learning_rate": 1.99999931215582e-06,
      "loss": 1.4305,
      "step": 231
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.810216498710757,
      "learning_rate": 1.9999993059169032e-06,
      "loss": 1.3897,
      "step": 232
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.75131060987687,
      "learning_rate": 1.99999929964982e-06,
      "loss": 1.4905,
      "step": 233
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.9417367696494185,
      "learning_rate": 1.99999929335457e-06,
      "loss": 1.5133,
      "step": 234
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.129642959459062,
      "learning_rate": 1.999999287031153e-06,
      "loss": 1.4786,
      "step": 235
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.589926026863031,
      "learning_rate": 1.9999992806795697e-06,
      "loss": 1.4966,
      "step": 236
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.509054909009049,
      "learning_rate": 1.999999274299819e-06,
      "loss": 1.4371,
      "step": 237
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.6496971486682135,
      "learning_rate": 1.9999992678919027e-06,
      "loss": 1.4595,
      "step": 238
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.832332761745741,
      "learning_rate": 1.999999261455819e-06,
      "loss": 1.5185,
      "step": 239
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.257353906344085,
      "learning_rate": 1.9999992549915693e-06,
      "loss": 1.5056,
      "step": 240
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.37222280076076,
      "learning_rate": 1.9999992484991524e-06,
      "loss": 1.504,
      "step": 241
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.824423532897361,
      "learning_rate": 1.999999241978569e-06,
      "loss": 1.4981,
      "step": 242
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.9091796444197575,
      "learning_rate": 1.999999235429819e-06,
      "loss": 1.4832,
      "step": 243
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.748583674899383,
      "learning_rate": 1.9999992288529025e-06,
      "loss": 1.3462,
      "step": 244
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.125022502509408,
      "learning_rate": 1.9999992222478192e-06,
      "loss": 1.4612,
      "step": 245
    },
    {
      "epoch": 0.0,
      "grad_norm": 32.53860605587701,
      "learning_rate": 1.999999215614569e-06,
      "loss": 1.4692,
      "step": 246
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.052996211809871,
      "learning_rate": 1.9999992089531526e-06,
      "loss": 1.4238,
      "step": 247
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.542095695226945,
      "learning_rate": 1.9999992022635693e-06,
      "loss": 1.4174,
      "step": 248
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.686250494352825,
      "learning_rate": 1.999999195545819e-06,
      "loss": 1.6754,
      "step": 249
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.6359874929300915,
      "learning_rate": 1.9999991887999027e-06,
      "loss": 1.5792,
      "step": 250
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.0936043548140795,
      "learning_rate": 1.9999991820258194e-06,
      "loss": 1.4941,
      "step": 251
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.406141836123647,
      "learning_rate": 1.9999991752235697e-06,
      "loss": 1.5123,
      "step": 252
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.199896874003351,
      "learning_rate": 1.999999168393153e-06,
      "loss": 1.476,
      "step": 253
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.66865264194522,
      "learning_rate": 1.99999916153457e-06,
      "loss": 1.5464,
      "step": 254
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.021383409806155,
      "learning_rate": 1.99999915464782e-06,
      "loss": 1.5831,
      "step": 255
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.615349724765676,
      "learning_rate": 1.9999991477329036e-06,
      "loss": 1.6743,
      "step": 256
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.024216598608762,
      "learning_rate": 1.9999991407898203e-06,
      "loss": 1.5263,
      "step": 257
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.352770335446249,
      "learning_rate": 1.9999991338185706e-06,
      "loss": 1.6316,
      "step": 258
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.472919888558574,
      "learning_rate": 1.9999991268191545e-06,
      "loss": 1.4915,
      "step": 259
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.46803445951313,
      "learning_rate": 1.999999119791571e-06,
      "loss": 1.5079,
      "step": 260
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.154033494012857,
      "learning_rate": 1.9999991127358214e-06,
      "loss": 1.5658,
      "step": 261
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.032995728347807,
      "learning_rate": 1.9999991056519053e-06,
      "loss": 1.5069,
      "step": 262
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.916728694225458,
      "learning_rate": 1.9999990985398224e-06,
      "loss": 1.5897,
      "step": 263
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.075661387114088,
      "learning_rate": 1.9999990913995727e-06,
      "loss": 1.5507,
      "step": 264
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.051445604812477,
      "learning_rate": 1.999999084231156e-06,
      "loss": 1.6803,
      "step": 265
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.971780062105953,
      "learning_rate": 1.999999077034573e-06,
      "loss": 1.4367,
      "step": 266
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.934330515559447,
      "learning_rate": 1.9999990698098235e-06,
      "loss": 1.356,
      "step": 267
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.5083027384301015,
      "learning_rate": 1.9999990625569073e-06,
      "loss": 1.4274,
      "step": 268
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.109968483536287,
      "learning_rate": 1.9999990552758244e-06,
      "loss": 1.4728,
      "step": 269
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.775385612499377,
      "learning_rate": 1.999999047966575e-06,
      "loss": 1.5146,
      "step": 270
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.948468121480936,
      "learning_rate": 1.9999990406291585e-06,
      "loss": 1.4915,
      "step": 271
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.965168560424121,
      "learning_rate": 1.999999033263576e-06,
      "loss": 1.4582,
      "step": 272
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.442431763765304,
      "learning_rate": 1.9999990258698267e-06,
      "loss": 1.4613,
      "step": 273
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.7776067027301545,
      "learning_rate": 1.99999901844791e-06,
      "loss": 1.503,
      "step": 274
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.178892269148893,
      "learning_rate": 1.999999010997827e-06,
      "loss": 1.4328,
      "step": 275
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.939222731616965,
      "learning_rate": 1.9999990035195783e-06,
      "loss": 1.4259,
      "step": 276
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.900888732024022,
      "learning_rate": 1.9999989960131617e-06,
      "loss": 1.4883,
      "step": 277
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.480965646206889,
      "learning_rate": 1.999998988478579e-06,
      "loss": 1.4998,
      "step": 278
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.387613428010502,
      "learning_rate": 1.99999898091583e-06,
      "loss": 1.5284,
      "step": 279
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.93375972130229,
      "learning_rate": 1.9999989733249137e-06,
      "loss": 1.5188,
      "step": 280
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.887766431921025,
      "learning_rate": 1.999998965705831e-06,
      "loss": 1.4875,
      "step": 281
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.872411589272914,
      "learning_rate": 1.999998958058582e-06,
      "loss": 1.4924,
      "step": 282
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.8912427815939825,
      "learning_rate": 1.999998950383166e-06,
      "loss": 1.3855,
      "step": 283
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.849266593984057,
      "learning_rate": 1.9999989426795835e-06,
      "loss": 1.252,
      "step": 284
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.642331959124537,
      "learning_rate": 1.999998934947834e-06,
      "loss": 1.4346,
      "step": 285
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.824479351331792,
      "learning_rate": 1.999998927187918e-06,
      "loss": 1.3413,
      "step": 286
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.088834406666165,
      "learning_rate": 1.999998919399836e-06,
      "loss": 1.5249,
      "step": 287
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.380115793381058,
      "learning_rate": 1.9999989115835865e-06,
      "loss": 1.6594,
      "step": 288
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.560993714860048,
      "learning_rate": 1.9999989037391708e-06,
      "loss": 1.378,
      "step": 289
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.285453090943186,
      "learning_rate": 1.9999988958665882e-06,
      "loss": 1.5047,
      "step": 290
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.939941853112505,
      "learning_rate": 1.9999988879658393e-06,
      "loss": 1.4266,
      "step": 291
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.842983984638459,
      "learning_rate": 1.9999988800369235e-06,
      "loss": 1.4714,
      "step": 292
    },
    {
      "epoch": 0.0,
      "eval_loss": 1.7301859855651855,
      "eval_runtime": 4.6111,
      "eval_samples_per_second": 1.952,
      "eval_steps_per_second": 1.084,
      "step": 292
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.779278420529169,
      "learning_rate": 1.999998872079841e-06,
      "loss": 1.59,
      "step": 293
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.66455555017548,
      "learning_rate": 1.999998864094592e-06,
      "loss": 1.5995,
      "step": 294
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.542518840920991,
      "learning_rate": 1.9999988560811762e-06,
      "loss": 1.3785,
      "step": 295
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.741540663363095,
      "learning_rate": 1.999998848039594e-06,
      "loss": 1.3725,
      "step": 296
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.958319396026163,
      "learning_rate": 1.999998839969845e-06,
      "loss": 1.4389,
      "step": 297
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.858392697584495,
      "learning_rate": 1.99999883187193e-06,
      "loss": 1.4766,
      "step": 298
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.77843632873252,
      "learning_rate": 1.9999988237458472e-06,
      "loss": 1.4784,
      "step": 299
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.166101873536924,
      "learning_rate": 1.9999988155915983e-06,
      "loss": 1.5636,
      "step": 300
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.541352000059288,
      "learning_rate": 1.999998807409183e-06,
      "loss": 1.3294,
      "step": 301
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.738532728807263,
      "learning_rate": 1.9999987991986007e-06,
      "loss": 1.3765,
      "step": 302
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.12811751220863,
      "learning_rate": 1.9999987909598518e-06,
      "loss": 1.6032,
      "step": 303
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.897961031920527,
      "learning_rate": 1.9999987826929364e-06,
      "loss": 1.5486,
      "step": 304
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.108120627298669,
      "learning_rate": 1.9999987743978542e-06,
      "loss": 1.5455,
      "step": 305
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.654624244259197,
      "learning_rate": 1.9999987660746057e-06,
      "loss": 1.563,
      "step": 306
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.32748407900337,
      "learning_rate": 1.9999987577231903e-06,
      "loss": 1.5657,
      "step": 307
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.643090403562411,
      "learning_rate": 1.9999987493436086e-06,
      "loss": 1.4312,
      "step": 308
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.16915443300704,
      "learning_rate": 1.99999874093586e-06,
      "loss": 1.4448,
      "step": 309
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.277599731706892,
      "learning_rate": 1.9999987324999446e-06,
      "loss": 1.6571,
      "step": 310
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.2349500478147215,
      "learning_rate": 1.9999987240358625e-06,
      "loss": 1.4609,
      "step": 311
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.248179668341996,
      "learning_rate": 1.9999987155436143e-06,
      "loss": 1.6418,
      "step": 312
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.223835299607056,
      "learning_rate": 1.9999987070231985e-06,
      "loss": 1.495,
      "step": 313
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.202665887318379,
      "learning_rate": 1.999998698474617e-06,
      "loss": 1.5282,
      "step": 314
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.938213910400009,
      "learning_rate": 1.9999986898978686e-06,
      "loss": 1.4169,
      "step": 315
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.193414540211652,
      "learning_rate": 1.9999986812929536e-06,
      "loss": 1.3747,
      "step": 316
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.870538982226375,
      "learning_rate": 1.999998672659872e-06,
      "loss": 1.3901,
      "step": 317
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.491542785298596,
      "learning_rate": 1.9999986639986233e-06,
      "loss": 1.5331,
      "step": 318
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.500143758119544,
      "learning_rate": 1.9999986553092083e-06,
      "loss": 1.6651,
      "step": 319
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.286451328879394,
      "learning_rate": 1.9999986465916265e-06,
      "loss": 1.6562,
      "step": 320
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.19858420548839,
      "learning_rate": 1.9999986378458784e-06,
      "loss": 1.4969,
      "step": 321
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.524050041906956,
      "learning_rate": 1.9999986290719634e-06,
      "loss": 1.4227,
      "step": 322
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.65324091844115,
      "learning_rate": 1.999998620269882e-06,
      "loss": 1.5175,
      "step": 323
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.955466116973382,
      "learning_rate": 1.999998611439634e-06,
      "loss": 1.3881,
      "step": 324
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.98640055971755,
      "learning_rate": 1.9999986025812193e-06,
      "loss": 1.6281,
      "step": 325
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.372570880498082,
      "learning_rate": 1.9999985936946375e-06,
      "loss": 1.4439,
      "step": 326
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.307836526026473,
      "learning_rate": 1.9999985847798893e-06,
      "loss": 1.5315,
      "step": 327
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.967657152244879,
      "learning_rate": 1.9999985758369748e-06,
      "loss": 1.5171,
      "step": 328
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.189830333449287,
      "learning_rate": 1.9999985668658934e-06,
      "loss": 1.5468,
      "step": 329
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.903849880093316,
      "learning_rate": 1.9999985578666452e-06,
      "loss": 1.6242,
      "step": 330
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.784805252959356,
      "learning_rate": 1.999998548839231e-06,
      "loss": 1.4167,
      "step": 331
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.196004132700368,
      "learning_rate": 1.9999985397836497e-06,
      "loss": 1.5647,
      "step": 332
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.103521406136126,
      "learning_rate": 1.999998530699902e-06,
      "loss": 1.4414,
      "step": 333
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.570598621348803,
      "learning_rate": 1.9999985215879873e-06,
      "loss": 1.3089,
      "step": 334
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.777301480121715,
      "learning_rate": 1.999998512447906e-06,
      "loss": 1.5408,
      "step": 335
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.997876127994012,
      "learning_rate": 1.9999985032796586e-06,
      "loss": 1.4277,
      "step": 336
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.633061500236633,
      "learning_rate": 1.999998494083244e-06,
      "loss": 1.6895,
      "step": 337
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.834617334661858,
      "learning_rate": 1.999998484858663e-06,
      "loss": 1.426,
      "step": 338
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.993606113796682,
      "learning_rate": 1.9999984756059153e-06,
      "loss": 1.3854,
      "step": 339
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.101502771813274,
      "learning_rate": 1.999998466325001e-06,
      "loss": 1.6404,
      "step": 340
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.089236533265877,
      "learning_rate": 1.9999984570159197e-06,
      "loss": 1.4611,
      "step": 341
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.206319847977299,
      "learning_rate": 1.9999984476786723e-06,
      "loss": 1.3741,
      "step": 342
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.331926916079775,
      "learning_rate": 1.999998438313258e-06,
      "loss": 1.5426,
      "step": 343
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.3624137401783445,
      "learning_rate": 1.9999984289196776e-06,
      "loss": 1.6399,
      "step": 344
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.918510892290634,
      "learning_rate": 1.99999841949793e-06,
      "loss": 1.4012,
      "step": 345
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.5972792983175665,
      "learning_rate": 1.999998410048016e-06,
      "loss": 1.4264,
      "step": 346
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.916021369254276,
      "learning_rate": 1.999998400569935e-06,
      "loss": 1.5523,
      "step": 347
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.5315684631864,
      "learning_rate": 1.9999983910636877e-06,
      "loss": 1.5335,
      "step": 348
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.656760353737745,
      "learning_rate": 1.999998381529274e-06,
      "loss": 1.4952,
      "step": 349
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.583073365446635,
      "learning_rate": 1.9999983719666933e-06,
      "loss": 1.4312,
      "step": 350
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.293004596545783,
      "learning_rate": 1.999998362375946e-06,
      "loss": 1.5153,
      "step": 351
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.998189663482702,
      "learning_rate": 1.999998352757032e-06,
      "loss": 1.5468,
      "step": 352
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.704748816779011,
      "learning_rate": 1.9999983431099516e-06,
      "loss": 1.4175,
      "step": 353
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.816706794317481,
      "learning_rate": 1.9999983334347046e-06,
      "loss": 1.5202,
      "step": 354
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.973712391629575,
      "learning_rate": 1.999998323731291e-06,
      "loss": 1.4952,
      "step": 355
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.4799024863000305,
      "learning_rate": 1.9999983139997107e-06,
      "loss": 1.6129,
      "step": 356
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.773895048702807,
      "learning_rate": 1.9999983042399637e-06,
      "loss": 1.5075,
      "step": 357
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.786637205360941,
      "learning_rate": 1.99999829445205e-06,
      "loss": 1.5998,
      "step": 358
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.417325890208001,
      "learning_rate": 1.9999982846359697e-06,
      "loss": 1.491,
      "step": 359
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.441996636074185,
      "learning_rate": 1.9999982747917228e-06,
      "loss": 1.452,
      "step": 360
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.484858685210035,
      "learning_rate": 1.9999982649193094e-06,
      "loss": 1.5286,
      "step": 361
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.304285744838827,
      "learning_rate": 1.999998255018729e-06,
      "loss": 1.261,
      "step": 362
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.00715136108103,
      "learning_rate": 1.9999982450899826e-06,
      "loss": 1.5213,
      "step": 363
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.9421250934625744,
      "learning_rate": 1.9999982351330693e-06,
      "loss": 1.5245,
      "step": 364
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.249784350703182,
      "learning_rate": 1.999998225147989e-06,
      "loss": 1.4959,
      "step": 365
    },
    {
      "epoch": 0.0,
      "eval_loss": 1.7152172327041626,
      "eval_runtime": 4.5993,
      "eval_samples_per_second": 1.957,
      "eval_steps_per_second": 1.087,
      "step": 365
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.950903475262251,
      "learning_rate": 1.9999982151347425e-06,
      "loss": 1.5132,
      "step": 366
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.001256076322384,
      "learning_rate": 1.999998205093329e-06,
      "loss": 1.4178,
      "step": 367
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.145343234500232,
      "learning_rate": 1.9999981950237494e-06,
      "loss": 1.6928,
      "step": 368
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.926796515825282,
      "learning_rate": 1.999998184926003e-06,
      "loss": 1.4183,
      "step": 369
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.205577250811191,
      "learning_rate": 1.99999817480009e-06,
      "loss": 1.4067,
      "step": 370
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.181548092407497,
      "learning_rate": 1.99999816464601e-06,
      "loss": 1.3839,
      "step": 371
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.8304254951229355,
      "learning_rate": 1.9999981544637634e-06,
      "loss": 1.3712,
      "step": 372
    },
    {
      "epoch": 0.0,
      "grad_norm": 12.315569431338385,
      "learning_rate": 1.9999981442533505e-06,
      "loss": 1.5337,
      "step": 373
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.969790738605887,
      "learning_rate": 1.999998134014771e-06,
      "loss": 1.5229,
      "step": 374
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.073608092190098,
      "learning_rate": 1.999998123748025e-06,
      "loss": 1.579,
      "step": 375
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.111176704390247,
      "learning_rate": 1.999998113453112e-06,
      "loss": 1.4727,
      "step": 376
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.404278529357815,
      "learning_rate": 1.9999981031300326e-06,
      "loss": 1.3756,
      "step": 377
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.66133544977496,
      "learning_rate": 1.9999980927787864e-06,
      "loss": 1.3958,
      "step": 378
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.811042205639708,
      "learning_rate": 1.999998082399374e-06,
      "loss": 1.4358,
      "step": 379
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.082316240684894,
      "learning_rate": 1.9999980719917945e-06,
      "loss": 1.5925,
      "step": 380
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.316025491311415,
      "learning_rate": 1.9999980615560487e-06,
      "loss": 1.3726,
      "step": 381
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.922423456498873,
      "learning_rate": 1.9999980510921357e-06,
      "loss": 1.3629,
      "step": 382
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.966088631224505,
      "learning_rate": 1.9999980406000568e-06,
      "loss": 1.4834,
      "step": 383
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.8135477821063475,
      "learning_rate": 1.999998030079811e-06,
      "loss": 1.5864,
      "step": 384
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.2288779061936035,
      "learning_rate": 1.999998019531399e-06,
      "loss": 1.5929,
      "step": 385
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.583909712278632,
      "learning_rate": 1.9999980089548195e-06,
      "loss": 1.4818,
      "step": 386
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.161770592187534,
      "learning_rate": 1.999997998350074e-06,
      "loss": 1.7042,
      "step": 387
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.53786371138208,
      "learning_rate": 1.999997987717162e-06,
      "loss": 1.4949,
      "step": 388
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.547955232044561,
      "learning_rate": 1.999997977056083e-06,
      "loss": 1.5075,
      "step": 389
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.443846222937035,
      "learning_rate": 1.999997966366837e-06,
      "loss": 1.5852,
      "step": 390
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.948645185345369,
      "learning_rate": 1.9999979556494255e-06,
      "loss": 1.3367,
      "step": 391
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.580424596028237,
      "learning_rate": 1.9999979449038465e-06,
      "loss": 1.2409,
      "step": 392
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.704258128824565,
      "learning_rate": 1.999997934130101e-06,
      "loss": 1.4568,
      "step": 393
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.441676548521125,
      "learning_rate": 1.9999979233281894e-06,
      "loss": 1.5953,
      "step": 394
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.912142407168767,
      "learning_rate": 1.999997912498111e-06,
      "loss": 1.54,
      "step": 395
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.1706944951472895,
      "learning_rate": 1.9999979016398654e-06,
      "loss": 1.3123,
      "step": 396
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.26460889745062,
      "learning_rate": 1.9999978907534537e-06,
      "loss": 1.5813,
      "step": 397
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.564397081017155,
      "learning_rate": 1.999997879838875e-06,
      "loss": 1.3763,
      "step": 398
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.9015032586368905,
      "learning_rate": 1.99999786889613e-06,
      "loss": 1.4756,
      "step": 399
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.917803358556267,
      "learning_rate": 1.9999978579252184e-06,
      "loss": 1.2537,
      "step": 400
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.5181742977699555,
      "learning_rate": 1.9999978469261402e-06,
      "loss": 1.4328,
      "step": 401
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.660663659308694,
      "learning_rate": 1.9999978358988953e-06,
      "loss": 1.4581,
      "step": 402
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.0474102698666385,
      "learning_rate": 1.999997824843484e-06,
      "loss": 1.4002,
      "step": 403
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.834707543136538,
      "learning_rate": 1.9999978137599058e-06,
      "loss": 1.4236,
      "step": 404
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.168260756918322,
      "learning_rate": 1.9999978026481612e-06,
      "loss": 1.5046,
      "step": 405
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.035029265083008,
      "learning_rate": 1.99999779150825e-06,
      "loss": 1.3636,
      "step": 406
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.28716001715498,
      "learning_rate": 1.999997780340172e-06,
      "loss": 1.4576,
      "step": 407
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.8852077758822015,
      "learning_rate": 1.999997769143927e-06,
      "loss": 1.4481,
      "step": 408
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.232260975458147,
      "learning_rate": 1.999997757919516e-06,
      "loss": 1.5196,
      "step": 409
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.264068624514251,
      "learning_rate": 1.9999977466669385e-06,
      "loss": 1.4934,
      "step": 410
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.469538493605676,
      "learning_rate": 1.999997735386194e-06,
      "loss": 1.505,
      "step": 411
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.999328912549101,
      "learning_rate": 1.999997724077283e-06,
      "loss": 1.4756,
      "step": 412
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.076217295124521,
      "learning_rate": 1.9999977127402057e-06,
      "loss": 1.5694,
      "step": 413
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.226359452459192,
      "learning_rate": 1.999997701374961e-06,
      "loss": 1.4211,
      "step": 414
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.819718845384241,
      "learning_rate": 1.99999768998155e-06,
      "loss": 1.5322,
      "step": 415
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.122043265582762,
      "learning_rate": 1.9999976785599732e-06,
      "loss": 1.4893,
      "step": 416
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.297154589932777,
      "learning_rate": 1.9999976671102287e-06,
      "loss": 1.5449,
      "step": 417
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.890157121974458,
      "learning_rate": 1.999997655632318e-06,
      "loss": 1.4776,
      "step": 418
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.589475875269356,
      "learning_rate": 1.9999976441262408e-06,
      "loss": 1.4896,
      "step": 419
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.277939766849428,
      "learning_rate": 1.999997632591997e-06,
      "loss": 1.5828,
      "step": 420
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.470441711000115,
      "learning_rate": 1.9999976210295865e-06,
      "loss": 1.3217,
      "step": 421
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.995769226585237,
      "learning_rate": 1.9999976094390096e-06,
      "loss": 1.5426,
      "step": 422
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.160211017382697,
      "learning_rate": 1.999997597820266e-06,
      "loss": 1.5534,
      "step": 423
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.8941787869537325,
      "learning_rate": 1.9999975861733557e-06,
      "loss": 1.4683,
      "step": 424
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.213637265959196,
      "learning_rate": 1.9999975744982788e-06,
      "loss": 1.7138,
      "step": 425
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.62520919093602,
      "learning_rate": 1.999997562795035e-06,
      "loss": 1.4748,
      "step": 426
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.836310097570104,
      "learning_rate": 1.999997551063625e-06,
      "loss": 1.5654,
      "step": 427
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.4455635910745555,
      "learning_rate": 1.9999975393040484e-06,
      "loss": 1.3387,
      "step": 428
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.820553071368411,
      "learning_rate": 1.999997527516305e-06,
      "loss": 1.4821,
      "step": 429
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.97958037440551,
      "learning_rate": 1.9999975157003953e-06,
      "loss": 1.5386,
      "step": 430
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.218479803099335,
      "learning_rate": 1.9999975038563184e-06,
      "loss": 1.5253,
      "step": 431
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.817934555265224,
      "learning_rate": 1.9999974919840755e-06,
      "loss": 1.4677,
      "step": 432
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.04962891837311,
      "learning_rate": 1.9999974800836658e-06,
      "loss": 1.4149,
      "step": 433
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.977731627262668,
      "learning_rate": 1.9999974681550892e-06,
      "loss": 1.4752,
      "step": 434
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.593445830695836,
      "learning_rate": 1.9999974561983467e-06,
      "loss": 1.317,
      "step": 435
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.061091292458903,
      "learning_rate": 1.999997444213437e-06,
      "loss": 1.486,
      "step": 436
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.608804550083991,
      "learning_rate": 1.999997432200361e-06,
      "loss": 1.3927,
      "step": 437
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.670859026053921,
      "learning_rate": 1.999997420159118e-06,
      "loss": 1.4257,
      "step": 438
    },
    {
      "epoch": 0.0,
      "eval_loss": 1.7094385623931885,
      "eval_runtime": 4.6238,
      "eval_samples_per_second": 1.946,
      "eval_steps_per_second": 1.081,
      "step": 438
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.859564981544054,
      "learning_rate": 1.9999974080897087e-06,
      "loss": 1.3354,
      "step": 439
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.52294064206441,
      "learning_rate": 1.999997395992133e-06,
      "loss": 1.6065,
      "step": 440
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.424488305091629,
      "learning_rate": 1.99999738386639e-06,
      "loss": 1.4306,
      "step": 441
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.798968031648101,
      "learning_rate": 1.999997371712481e-06,
      "loss": 1.4365,
      "step": 442
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.932924371225362,
      "learning_rate": 1.9999973595304054e-06,
      "loss": 1.3648,
      "step": 443
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.234896304295093,
      "learning_rate": 1.999997347320163e-06,
      "loss": 1.4025,
      "step": 444
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.0542656681469715,
      "learning_rate": 1.9999973350817544e-06,
      "loss": 1.2087,
      "step": 445
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.129429012048715,
      "learning_rate": 1.9999973228151787e-06,
      "loss": 1.4321,
      "step": 446
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.613105028702654,
      "learning_rate": 1.999997310520436e-06,
      "loss": 1.4545,
      "step": 447
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.0864337050088295,
      "learning_rate": 1.9999972981975277e-06,
      "loss": 1.3047,
      "step": 448
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.123209043678265,
      "learning_rate": 1.9999972858464524e-06,
      "loss": 1.4923,
      "step": 449
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.770437981538585,
      "learning_rate": 1.9999972734672107e-06,
      "loss": 1.3795,
      "step": 450
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.66047161014276,
      "learning_rate": 1.999997261059802e-06,
      "loss": 1.4359,
      "step": 451
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.661063937671921,
      "learning_rate": 1.999997248624227e-06,
      "loss": 1.6533,
      "step": 452
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.631944386384085,
      "learning_rate": 1.9999972361604853e-06,
      "loss": 1.4383,
      "step": 453
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.732909501037933,
      "learning_rate": 1.9999972236685768e-06,
      "loss": 1.3426,
      "step": 454
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.460195121758093,
      "learning_rate": 1.999997211148502e-06,
      "loss": 1.5001,
      "step": 455
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.942983319180353,
      "learning_rate": 1.99999719860026e-06,
      "loss": 1.5882,
      "step": 456
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.541373356599531,
      "learning_rate": 1.999997186023852e-06,
      "loss": 1.3362,
      "step": 457
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.6766662288287515,
      "learning_rate": 1.9999971734192776e-06,
      "loss": 1.4117,
      "step": 458
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.240283985252159,
      "learning_rate": 1.9999971607865364e-06,
      "loss": 1.2851,
      "step": 459
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.6165761980037265,
      "learning_rate": 1.9999971481256283e-06,
      "loss": 1.4075,
      "step": 460
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.356136217997657,
      "learning_rate": 1.999997135436554e-06,
      "loss": 1.2614,
      "step": 461
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.235383295905265,
      "learning_rate": 1.999997122719313e-06,
      "loss": 1.5025,
      "step": 462
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.59435469640636,
      "learning_rate": 1.999997109973905e-06,
      "loss": 1.6514,
      "step": 463
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.674661542805518,
      "learning_rate": 1.9999970972003308e-06,
      "loss": 1.5387,
      "step": 464
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.954501972594385,
      "learning_rate": 1.99999708439859e-06,
      "loss": 1.376,
      "step": 465
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.031043958621965,
      "learning_rate": 1.9999970715686827e-06,
      "loss": 1.5981,
      "step": 466
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.849912677371338,
      "learning_rate": 1.9999970587106086e-06,
      "loss": 1.4769,
      "step": 467
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.18062121080789,
      "learning_rate": 1.9999970458243677e-06,
      "loss": 1.5745,
      "step": 468
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.171389079148798,
      "learning_rate": 1.999997032909961e-06,
      "loss": 1.4238,
      "step": 469
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.888754083257385,
      "learning_rate": 1.999997019967387e-06,
      "loss": 1.5669,
      "step": 470
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.661884685242049,
      "learning_rate": 1.9999970069966463e-06,
      "loss": 1.2735,
      "step": 471
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.231399197062376,
      "learning_rate": 1.9999969939977395e-06,
      "loss": 1.4674,
      "step": 472
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.392315151719007,
      "learning_rate": 1.999996980970666e-06,
      "loss": 1.2746,
      "step": 473
    },
    {
      "epoch": 0.0,
      "grad_norm": 14.143521137726825,
      "learning_rate": 1.999996967915426e-06,
      "loss": 1.4949,
      "step": 474
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.609979863275451,
      "learning_rate": 1.999996954832019e-06,
      "loss": 1.5421,
      "step": 475
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.901316944779091,
      "learning_rate": 1.9999969417204457e-06,
      "loss": 1.467,
      "step": 476
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.9893724431034085,
      "learning_rate": 1.999996928580706e-06,
      "loss": 1.5033,
      "step": 477
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.145370058739066,
      "learning_rate": 1.9999969154127992e-06,
      "loss": 1.6459,
      "step": 478
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.6040849521366045,
      "learning_rate": 1.999996902216726e-06,
      "loss": 1.39,
      "step": 479
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.871162617374601,
      "learning_rate": 1.9999968889924863e-06,
      "loss": 1.5008,
      "step": 480
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.041180910719072,
      "learning_rate": 1.9999968757400803e-06,
      "loss": 1.417,
      "step": 481
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.1416379779540415,
      "learning_rate": 1.9999968624595075e-06,
      "loss": 1.3683,
      "step": 482
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.655432088463851,
      "learning_rate": 1.999996849150768e-06,
      "loss": 1.2764,
      "step": 483
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.390658631829691,
      "learning_rate": 1.999996835813862e-06,
      "loss": 1.4183,
      "step": 484
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.139267857868251,
      "learning_rate": 1.9999968224487894e-06,
      "loss": 1.5219,
      "step": 485
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.98265828385138,
      "learning_rate": 1.9999968090555498e-06,
      "loss": 1.5828,
      "step": 486
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.947025615061189,
      "learning_rate": 1.999996795634144e-06,
      "loss": 1.4291,
      "step": 487
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.8361628814359445,
      "learning_rate": 1.999996782184572e-06,
      "loss": 1.5098,
      "step": 488
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.213733320329145,
      "learning_rate": 1.999996768706833e-06,
      "loss": 1.5318,
      "step": 489
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.354255255395692,
      "learning_rate": 1.9999967552009273e-06,
      "loss": 1.5049,
      "step": 490
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.509817618529389,
      "learning_rate": 1.9999967416668553e-06,
      "loss": 1.5207,
      "step": 491
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.820085190259705,
      "learning_rate": 1.9999967281046165e-06,
      "loss": 1.4661,
      "step": 492
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.5842798335932855,
      "learning_rate": 1.9999967145142113e-06,
      "loss": 1.5606,
      "step": 493
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.832919175129925,
      "learning_rate": 1.9999967008956397e-06,
      "loss": 1.5222,
      "step": 494
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.401711540963192,
      "learning_rate": 1.9999966872489013e-06,
      "loss": 1.4282,
      "step": 495
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.352942988278965,
      "learning_rate": 1.999996673573996e-06,
      "loss": 1.4907,
      "step": 496
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.767236823164161,
      "learning_rate": 1.9999966598709245e-06,
      "loss": 1.2979,
      "step": 497
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.7839487752581995,
      "learning_rate": 1.999996646139686e-06,
      "loss": 1.1931,
      "step": 498
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.526284836165884,
      "learning_rate": 1.9999966323802813e-06,
      "loss": 1.4276,
      "step": 499
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.968983179245585,
      "learning_rate": 1.99999661859271e-06,
      "loss": 1.5143,
      "step": 500
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.929416318558409,
      "learning_rate": 1.999996604776972e-06,
      "loss": 1.4274,
      "step": 501
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.319727005695438,
      "learning_rate": 1.999996590933068e-06,
      "loss": 1.4896,
      "step": 502
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.166912706207141,
      "learning_rate": 1.9999965770609966e-06,
      "loss": 1.4971,
      "step": 503
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.724664850608164,
      "learning_rate": 1.999996563160759e-06,
      "loss": 1.314,
      "step": 504
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.033309554063063,
      "learning_rate": 1.9999965492323547e-06,
      "loss": 1.6318,
      "step": 505
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.693812921757236,
      "learning_rate": 1.999996535275784e-06,
      "loss": 1.5998,
      "step": 506
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.027615373760868,
      "learning_rate": 1.9999965212910468e-06,
      "loss": 1.4803,
      "step": 507
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.150014426528141,
      "learning_rate": 1.999996507278143e-06,
      "loss": 1.5533,
      "step": 508
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.251365830445326,
      "learning_rate": 1.999996493237072e-06,
      "loss": 1.5345,
      "step": 509
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.068491757206887,
      "learning_rate": 1.999996479167835e-06,
      "loss": 1.4934,
      "step": 510
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.510297092824997,
      "learning_rate": 1.9999964650704313e-06,
      "loss": 1.4957,
      "step": 511
    },
    {
      "epoch": 0.0,
      "eval_loss": 1.702190637588501,
      "eval_runtime": 4.6182,
      "eval_samples_per_second": 1.949,
      "eval_steps_per_second": 1.083,
      "step": 511
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.755438253224136,
      "learning_rate": 1.999996450944861e-06,
      "loss": 1.4702,
      "step": 512
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.529822887512446,
      "learning_rate": 1.9999964367911242e-06,
      "loss": 1.4433,
      "step": 513
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.923234715248454,
      "learning_rate": 1.9999964226092207e-06,
      "loss": 1.2158,
      "step": 514
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.99275767811329,
      "learning_rate": 1.9999964083991507e-06,
      "loss": 1.4035,
      "step": 515
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.540700855208618,
      "learning_rate": 1.9999963941609144e-06,
      "loss": 1.4089,
      "step": 516
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.423934034968752,
      "learning_rate": 1.9999963798945113e-06,
      "loss": 1.4395,
      "step": 517
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.938286770911095,
      "learning_rate": 1.9999963655999413e-06,
      "loss": 1.5267,
      "step": 518
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.135159232589015,
      "learning_rate": 1.9999963512772054e-06,
      "loss": 1.5788,
      "step": 519
    },
    {
      "epoch": 0.0,
      "grad_norm": 41.00137108815898,
      "learning_rate": 1.9999963369263023e-06,
      "loss": 1.681,
      "step": 520
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.982490976686388,
      "learning_rate": 1.9999963225472327e-06,
      "loss": 1.3849,
      "step": 521
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.200076278688873,
      "learning_rate": 1.999996308139997e-06,
      "loss": 1.4505,
      "step": 522
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.0374386572637935,
      "learning_rate": 1.9999962937045945e-06,
      "loss": 1.6146,
      "step": 523
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.38514239730456,
      "learning_rate": 1.9999962792410254e-06,
      "loss": 1.4862,
      "step": 524
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.278831951738085,
      "learning_rate": 1.9999962647492895e-06,
      "loss": 1.5097,
      "step": 525
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.988682511151002,
      "learning_rate": 1.999996250229387e-06,
      "loss": 1.5579,
      "step": 526
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.6329832866037854,
      "learning_rate": 1.999996235681319e-06,
      "loss": 1.494,
      "step": 527
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.101238841635504,
      "learning_rate": 1.999996221105083e-06,
      "loss": 1.5725,
      "step": 528
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.618252608407,
      "learning_rate": 1.999996206500681e-06,
      "loss": 1.4834,
      "step": 529
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.783224896235011,
      "learning_rate": 1.999996191868113e-06,
      "loss": 1.7878,
      "step": 530
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.039904911976764,
      "learning_rate": 1.9999961772073773e-06,
      "loss": 1.5341,
      "step": 531
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.749210275476864,
      "learning_rate": 1.999996162518476e-06,
      "loss": 1.5232,
      "step": 532
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.75083641214273,
      "learning_rate": 1.999996147801408e-06,
      "loss": 1.4457,
      "step": 533
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.797512529634022,
      "learning_rate": 1.999996133056173e-06,
      "loss": 1.3914,
      "step": 534
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.996223642108566,
      "learning_rate": 1.999996118282772e-06,
      "loss": 1.53,
      "step": 535
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.7829974351407305,
      "learning_rate": 1.9999961034812035e-06,
      "loss": 1.3418,
      "step": 536
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.718076792623708,
      "learning_rate": 1.9999960886514693e-06,
      "loss": 1.3486,
      "step": 537
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.156179924159171,
      "learning_rate": 1.9999960737935678e-06,
      "loss": 1.5375,
      "step": 538
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.3419660684441705,
      "learning_rate": 1.9999960589075003e-06,
      "loss": 1.4394,
      "step": 539
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.969835171700577,
      "learning_rate": 1.999996043993266e-06,
      "loss": 1.4885,
      "step": 540
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.141114094857767,
      "learning_rate": 1.9999960290508654e-06,
      "loss": 1.6062,
      "step": 541
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.732914023923969,
      "learning_rate": 1.9999960140802984e-06,
      "loss": 1.7097,
      "step": 542
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.405085642943276,
      "learning_rate": 1.9999959990815645e-06,
      "loss": 1.5708,
      "step": 543
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.951768383120494,
      "learning_rate": 1.9999959840546643e-06,
      "loss": 1.5517,
      "step": 544
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.805093975628436,
      "learning_rate": 1.9999959689995972e-06,
      "loss": 1.4752,
      "step": 545
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.475429973625488,
      "learning_rate": 1.9999959539163634e-06,
      "loss": 1.3248,
      "step": 546
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.138477383552902,
      "learning_rate": 1.9999959388049636e-06,
      "loss": 1.5739,
      "step": 547
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.991661833650293,
      "learning_rate": 1.9999959236653965e-06,
      "loss": 1.4348,
      "step": 548
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.701391682646717,
      "learning_rate": 1.9999959084976635e-06,
      "loss": 1.3925,
      "step": 549
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.910083477148781,
      "learning_rate": 1.9999958933017637e-06,
      "loss": 1.5543,
      "step": 550
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.027398384260184,
      "learning_rate": 1.9999958780776975e-06,
      "loss": 1.4994,
      "step": 551
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.5869467738586645,
      "learning_rate": 1.9999958628254645e-06,
      "loss": 1.5429,
      "step": 552
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.951650316590327,
      "learning_rate": 1.999995847545065e-06,
      "loss": 1.4204,
      "step": 553
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.779335727504338,
      "learning_rate": 1.999995832236499e-06,
      "loss": 1.5872,
      "step": 554
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.995816902206547,
      "learning_rate": 1.9999958168997667e-06,
      "loss": 1.434,
      "step": 555
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.2068709341822235,
      "learning_rate": 1.9999958015348677e-06,
      "loss": 1.5262,
      "step": 556
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.993826056871235,
      "learning_rate": 1.9999957861418014e-06,
      "loss": 1.4638,
      "step": 557
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.2251786792625206,
      "learning_rate": 1.9999957707205693e-06,
      "loss": 1.508,
      "step": 558
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.073040511213923,
      "learning_rate": 1.9999957552711707e-06,
      "loss": 1.4745,
      "step": 559
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.738891638073386,
      "learning_rate": 1.9999957397936053e-06,
      "loss": 1.6243,
      "step": 560
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.419662722088751,
      "learning_rate": 1.9999957242878735e-06,
      "loss": 1.2774,
      "step": 561
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.122152529238781,
      "learning_rate": 1.999995708753975e-06,
      "loss": 1.4376,
      "step": 562
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.4113125087183205,
      "learning_rate": 1.99999569319191e-06,
      "loss": 1.5806,
      "step": 563
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.379407688726184,
      "learning_rate": 1.999995677601678e-06,
      "loss": 1.4604,
      "step": 564
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.47622178703681,
      "learning_rate": 1.9999956619832805e-06,
      "loss": 1.4344,
      "step": 565
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.117263096048395,
      "learning_rate": 1.999995646336716e-06,
      "loss": 1.4896,
      "step": 566
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.4922547794799,
      "learning_rate": 1.9999956306619846e-06,
      "loss": 1.7709,
      "step": 567
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.489699330478132,
      "learning_rate": 1.999995614959087e-06,
      "loss": 1.5452,
      "step": 568
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.437005541277756,
      "learning_rate": 1.9999955992280223e-06,
      "loss": 1.347,
      "step": 569
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.224691095393556,
      "learning_rate": 1.9999955834687917e-06,
      "loss": 1.5102,
      "step": 570
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.56916203726456,
      "learning_rate": 1.999995567681394e-06,
      "loss": 1.2866,
      "step": 571
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.203471880109494,
      "learning_rate": 1.99999555186583e-06,
      "loss": 1.2939,
      "step": 572
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.295050199339514,
      "learning_rate": 1.9999955360220997e-06,
      "loss": 1.244,
      "step": 573
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.558252219814004,
      "learning_rate": 1.9999955201502024e-06,
      "loss": 1.381,
      "step": 574
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.693879682065885,
      "learning_rate": 1.999995504250139e-06,
      "loss": 1.5172,
      "step": 575
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.054746659752432,
      "learning_rate": 1.9999954883219086e-06,
      "loss": 1.5245,
      "step": 576
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.288250112135058,
      "learning_rate": 1.999995472365512e-06,
      "loss": 1.6557,
      "step": 577
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.7842439911659005,
      "learning_rate": 1.9999954563809488e-06,
      "loss": 1.4294,
      "step": 578
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.995544020303297,
      "learning_rate": 1.999995440368219e-06,
      "loss": 1.3965,
      "step": 579
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.759630572595074,
      "learning_rate": 1.9999954243273226e-06,
      "loss": 1.3996,
      "step": 580
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.712211700693446,
      "learning_rate": 1.99999540825826e-06,
      "loss": 1.4983,
      "step": 581
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.9470327763676405,
      "learning_rate": 1.9999953921610304e-06,
      "loss": 1.355,
      "step": 582
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.063038997344527,
      "learning_rate": 1.9999953760356343e-06,
      "loss": 1.4099,
      "step": 583
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.844081361582967,
      "learning_rate": 1.999995359882072e-06,
      "loss": 1.5044,
      "step": 584
    },
    {
      "epoch": 0.0,
      "eval_loss": 1.696353793144226,
      "eval_runtime": 4.6168,
      "eval_samples_per_second": 1.949,
      "eval_steps_per_second": 1.083,
      "step": 584
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.493708899953276,
      "learning_rate": 1.999995343700343e-06,
      "loss": 1.6627,
      "step": 585
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.954875007169818,
      "learning_rate": 1.9999953274904473e-06,
      "loss": 1.4136,
      "step": 586
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.910325383229665,
      "learning_rate": 1.9999953112523853e-06,
      "loss": 1.4251,
      "step": 587
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.714961872146683,
      "learning_rate": 1.9999952949861564e-06,
      "loss": 1.5491,
      "step": 588
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.2328265295940435,
      "learning_rate": 1.999995278691761e-06,
      "loss": 1.4251,
      "step": 589
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.533584320037035,
      "learning_rate": 1.9999952623691995e-06,
      "loss": 1.4121,
      "step": 590
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.882067176722944,
      "learning_rate": 1.999995246018471e-06,
      "loss": 1.3394,
      "step": 591
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.9509375278675405,
      "learning_rate": 1.9999952296395762e-06,
      "loss": 1.3379,
      "step": 592
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.653441569963669,
      "learning_rate": 1.999995213232515e-06,
      "loss": 1.4337,
      "step": 593
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.426491174400903,
      "learning_rate": 1.999995196797287e-06,
      "loss": 1.2995,
      "step": 594
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.485353755158814,
      "learning_rate": 1.999995180333893e-06,
      "loss": 1.379,
      "step": 595
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.79710641338462,
      "learning_rate": 1.9999951638423317e-06,
      "loss": 1.4514,
      "step": 596
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.92546100162479,
      "learning_rate": 1.999995147322604e-06,
      "loss": 1.4666,
      "step": 597
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.9766360974271056,
      "learning_rate": 1.99999513077471e-06,
      "loss": 1.353,
      "step": 598
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.442918937982415,
      "learning_rate": 1.9999951141986493e-06,
      "loss": 1.6356,
      "step": 599
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.523537086483564,
      "learning_rate": 1.9999950975944225e-06,
      "loss": 1.5245,
      "step": 600
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.955021932634082,
      "learning_rate": 1.9999950809620285e-06,
      "loss": 1.3658,
      "step": 601
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.8036782108306975,
      "learning_rate": 1.999995064301468e-06,
      "loss": 1.4261,
      "step": 602
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.952441088003973,
      "learning_rate": 1.9999950476127418e-06,
      "loss": 1.5592,
      "step": 603
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.712458785442027,
      "learning_rate": 1.999995030895848e-06,
      "loss": 1.6329,
      "step": 604
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.39848861393844,
      "learning_rate": 1.9999950141507886e-06,
      "loss": 1.3637,
      "step": 605
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.565908788566706,
      "learning_rate": 1.9999949973775623e-06,
      "loss": 1.3901,
      "step": 606
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.839741237419302,
      "learning_rate": 1.9999949805761695e-06,
      "loss": 1.5408,
      "step": 607
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.65666455626493,
      "learning_rate": 1.99999496374661e-06,
      "loss": 1.3274,
      "step": 608
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.003729201972364,
      "learning_rate": 1.999994946888884e-06,
      "loss": 1.506,
      "step": 609
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.133503970487184,
      "learning_rate": 1.9999949300029917e-06,
      "loss": 1.5401,
      "step": 610
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.6024698311473795,
      "learning_rate": 1.9999949130889325e-06,
      "loss": 1.6896,
      "step": 611
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.532215717740302,
      "learning_rate": 1.999994896146707e-06,
      "loss": 1.3587,
      "step": 612
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.70340339038857,
      "learning_rate": 1.9999948791763146e-06,
      "loss": 1.577,
      "step": 613
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.082395639793893,
      "learning_rate": 1.9999948621777563e-06,
      "loss": 1.5421,
      "step": 614
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.988618029308243,
      "learning_rate": 1.9999948451510312e-06,
      "loss": 1.5131,
      "step": 615
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.413069306779731,
      "learning_rate": 1.9999948280961393e-06,
      "loss": 1.4763,
      "step": 616
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.723299915394361,
      "learning_rate": 1.999994811013081e-06,
      "loss": 1.4926,
      "step": 617
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.818607426839722,
      "learning_rate": 1.9999947939018567e-06,
      "loss": 1.4112,
      "step": 618
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.090960497753154,
      "learning_rate": 1.9999947767624656e-06,
      "loss": 1.6447,
      "step": 619
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.869884243493298,
      "learning_rate": 1.9999947595949073e-06,
      "loss": 1.4467,
      "step": 620
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.952829084694713,
      "learning_rate": 1.999994742399183e-06,
      "loss": 1.5202,
      "step": 621
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.000534437688967,
      "learning_rate": 1.9999947251752923e-06,
      "loss": 1.569,
      "step": 622
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.891724084780857,
      "learning_rate": 1.9999947079232353e-06,
      "loss": 1.2758,
      "step": 623
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.720768304907384,
      "learning_rate": 1.999994690643011e-06,
      "loss": 1.5005,
      "step": 624
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.8481849499077,
      "learning_rate": 1.9999946733346208e-06,
      "loss": 1.5816,
      "step": 625
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.153018684172472,
      "learning_rate": 1.9999946559980637e-06,
      "loss": 1.6261,
      "step": 626
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.041326934986502,
      "learning_rate": 1.9999946386333402e-06,
      "loss": 1.4737,
      "step": 627
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.615624017109531,
      "learning_rate": 1.9999946212404504e-06,
      "loss": 1.3258,
      "step": 628
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.5498876909447405,
      "learning_rate": 1.999994603819394e-06,
      "loss": 1.6407,
      "step": 629
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.466395618769352,
      "learning_rate": 1.999994586370171e-06,
      "loss": 1.3873,
      "step": 630
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.84924829351922,
      "learning_rate": 1.9999945688927813e-06,
      "loss": 1.4199,
      "step": 631
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.489922257516625,
      "learning_rate": 1.999994551387225e-06,
      "loss": 1.4957,
      "step": 632
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.178583734354207,
      "learning_rate": 1.999994533853503e-06,
      "loss": 1.3008,
      "step": 633
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.631980034454648,
      "learning_rate": 1.9999945162916135e-06,
      "loss": 1.2049,
      "step": 634
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.715631420481108,
      "learning_rate": 1.999994498701558e-06,
      "loss": 1.506,
      "step": 635
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.866461878636338,
      "learning_rate": 1.999994481083336e-06,
      "loss": 1.4913,
      "step": 636
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.81038769741456,
      "learning_rate": 1.9999944634369473e-06,
      "loss": 1.4033,
      "step": 637
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.08078031950503,
      "learning_rate": 1.999994445762392e-06,
      "loss": 1.5048,
      "step": 638
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.047631225535758,
      "learning_rate": 1.9999944280596705e-06,
      "loss": 1.4223,
      "step": 639
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.040805784015986,
      "learning_rate": 1.9999944103287824e-06,
      "loss": 1.5275,
      "step": 640
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.349390518960892,
      "learning_rate": 1.9999943925697274e-06,
      "loss": 1.3653,
      "step": 641
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.378881266474736,
      "learning_rate": 1.9999943747825064e-06,
      "loss": 1.4901,
      "step": 642
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.959820929884786,
      "learning_rate": 1.9999943569671187e-06,
      "loss": 1.5757,
      "step": 643
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.839111038325676,
      "learning_rate": 1.999994339123564e-06,
      "loss": 1.3587,
      "step": 644
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.487918310261601,
      "learning_rate": 1.999994321251843e-06,
      "loss": 1.2875,
      "step": 645
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.622421469454646,
      "learning_rate": 1.9999943033519563e-06,
      "loss": 1.4613,
      "step": 646
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.851420800533353,
      "learning_rate": 1.999994285423902e-06,
      "loss": 1.4116,
      "step": 647
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.13772278531517,
      "learning_rate": 1.999994267467682e-06,
      "loss": 1.503,
      "step": 648
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.995119073749364,
      "learning_rate": 1.9999942494832955e-06,
      "loss": 1.5454,
      "step": 649
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.260803959027427,
      "learning_rate": 1.999994231470742e-06,
      "loss": 1.5893,
      "step": 650
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.836602513098617,
      "learning_rate": 1.999994213430022e-06,
      "loss": 1.5162,
      "step": 651
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.155960472302761,
      "learning_rate": 1.9999941953611356e-06,
      "loss": 1.5372,
      "step": 652
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.746462550620719,
      "learning_rate": 1.999994177264083e-06,
      "loss": 1.4229,
      "step": 653
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.028762735096975,
      "learning_rate": 1.9999941591388634e-06,
      "loss": 1.4108,
      "step": 654
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.8256605605224,
      "learning_rate": 1.9999941409854773e-06,
      "loss": 1.4537,
      "step": 655
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.976124834899918,
      "learning_rate": 1.9999941228039253e-06,
      "loss": 1.4796,
      "step": 656
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.040837341128276,
      "learning_rate": 1.999994104594206e-06,
      "loss": 1.5095,
      "step": 657
    },
    {
      "epoch": 0.0,
      "eval_loss": 1.6910474300384521,
      "eval_runtime": 4.6444,
      "eval_samples_per_second": 1.938,
      "eval_steps_per_second": 1.077,
      "step": 657
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.4913712932783305,
      "learning_rate": 1.9999940863563208e-06,
      "loss": 1.7822,
      "step": 658
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.723500971046963,
      "learning_rate": 1.999994068090269e-06,
      "loss": 1.3798,
      "step": 659
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.445742218350867,
      "learning_rate": 1.9999940497960503e-06,
      "loss": 1.2929,
      "step": 660
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.422756108863784,
      "learning_rate": 1.9999940314736654e-06,
      "loss": 1.3135,
      "step": 661
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.932728683114205,
      "learning_rate": 1.9999940131231142e-06,
      "loss": 1.5933,
      "step": 662
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.725603325024966,
      "learning_rate": 1.999993994744396e-06,
      "loss": 1.4112,
      "step": 663
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.709408296070619,
      "learning_rate": 1.9999939763375114e-06,
      "loss": 1.3498,
      "step": 664
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.936026121310175,
      "learning_rate": 1.9999939579024606e-06,
      "loss": 1.4758,
      "step": 665
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.71294086987452,
      "learning_rate": 1.999993939439243e-06,
      "loss": 1.4319,
      "step": 666
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.066575513242086,
      "learning_rate": 1.999993920947859e-06,
      "loss": 1.6162,
      "step": 667
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.234786123258312,
      "learning_rate": 1.9999939024283086e-06,
      "loss": 1.4918,
      "step": 668
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.8552625381814565,
      "learning_rate": 1.999993883880592e-06,
      "loss": 1.4518,
      "step": 669
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.497782097949712,
      "learning_rate": 1.9999938653047082e-06,
      "loss": 1.394,
      "step": 670
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.189146286839885,
      "learning_rate": 1.9999938467006583e-06,
      "loss": 1.4676,
      "step": 671
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.318939072932948,
      "learning_rate": 1.999993828068442e-06,
      "loss": 1.4845,
      "step": 672
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.231313892198888,
      "learning_rate": 1.9999938094080588e-06,
      "loss": 1.5264,
      "step": 673
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.190824126212358,
      "learning_rate": 1.9999937907195092e-06,
      "loss": 1.5604,
      "step": 674
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.911434823319875,
      "learning_rate": 1.9999937720027937e-06,
      "loss": 1.4599,
      "step": 675
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.527825329074028,
      "learning_rate": 1.999993753257911e-06,
      "loss": 1.4444,
      "step": 676
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.612727754818132,
      "learning_rate": 1.9999937344848622e-06,
      "loss": 1.4543,
      "step": 677
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.912996568592968,
      "learning_rate": 1.9999937156836463e-06,
      "loss": 1.4475,
      "step": 678
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.899245327794625,
      "learning_rate": 1.9999936968542644e-06,
      "loss": 1.4971,
      "step": 679
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.683408208952821,
      "learning_rate": 1.999993677996716e-06,
      "loss": 1.4405,
      "step": 680
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.868881120149723,
      "learning_rate": 1.999993659111001e-06,
      "loss": 1.3217,
      "step": 681
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.725153463356434,
      "learning_rate": 1.9999936401971195e-06,
      "loss": 1.3043,
      "step": 682
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.909021709681444,
      "learning_rate": 1.9999936212550716e-06,
      "loss": 1.4595,
      "step": 683
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.906414442711592,
      "learning_rate": 1.9999936022848574e-06,
      "loss": 1.4195,
      "step": 684
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.200952942030242,
      "learning_rate": 1.9999935832864763e-06,
      "loss": 1.3076,
      "step": 685
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.601568858946815,
      "learning_rate": 1.999993564259929e-06,
      "loss": 1.3687,
      "step": 686
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.694173785690684,
      "learning_rate": 1.999993545205215e-06,
      "loss": 1.3167,
      "step": 687
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.925517734547198,
      "learning_rate": 1.9999935261223344e-06,
      "loss": 1.524,
      "step": 688
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.058783499529197,
      "learning_rate": 1.9999935070112877e-06,
      "loss": 1.3536,
      "step": 689
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.7403320365456185,
      "learning_rate": 1.9999934878720743e-06,
      "loss": 1.45,
      "step": 690
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.216158970114753,
      "learning_rate": 1.9999934687046945e-06,
      "loss": 1.4329,
      "step": 691
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.907151976261744,
      "learning_rate": 1.999993449509148e-06,
      "loss": 1.579,
      "step": 692
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.26679103924561,
      "learning_rate": 1.9999934302854353e-06,
      "loss": 1.6084,
      "step": 693
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.831000211583377,
      "learning_rate": 1.999993411033556e-06,
      "loss": 1.574,
      "step": 694
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.9068622026627065,
      "learning_rate": 1.99999339175351e-06,
      "loss": 1.5251,
      "step": 695
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.824731944389374,
      "learning_rate": 1.9999933724452975e-06,
      "loss": 1.4736,
      "step": 696
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.737465979855922,
      "learning_rate": 1.999993353108919e-06,
      "loss": 1.3993,
      "step": 697
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.601956191342637,
      "learning_rate": 1.9999933337443736e-06,
      "loss": 1.4853,
      "step": 698
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.891756552820562,
      "learning_rate": 1.999993314351662e-06,
      "loss": 1.3733,
      "step": 699
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.343260811587546,
      "learning_rate": 1.9999932949307833e-06,
      "loss": 1.5779,
      "step": 700
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.646914554700779,
      "learning_rate": 1.9999932754817387e-06,
      "loss": 1.4926,
      "step": 701
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.004390716593502,
      "learning_rate": 1.9999932560045274e-06,
      "loss": 1.343,
      "step": 702
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.8435215224734,
      "learning_rate": 1.9999932364991497e-06,
      "loss": 1.5128,
      "step": 703
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.094980526251057,
      "learning_rate": 1.9999932169656056e-06,
      "loss": 1.4424,
      "step": 704
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.71606625952716,
      "learning_rate": 1.9999931974038946e-06,
      "loss": 1.2818,
      "step": 705
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.019306981145199,
      "learning_rate": 1.9999931778140178e-06,
      "loss": 1.3311,
      "step": 706
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.402365143728885,
      "learning_rate": 1.999993158195974e-06,
      "loss": 1.269,
      "step": 707
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.4490937352418065,
      "learning_rate": 1.999993138549764e-06,
      "loss": 1.4177,
      "step": 708
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.710938425019419,
      "learning_rate": 1.999993118875387e-06,
      "loss": 1.3903,
      "step": 709
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.853647057982472,
      "learning_rate": 1.999993099172844e-06,
      "loss": 1.3143,
      "step": 710
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.583393631131937,
      "learning_rate": 1.9999930794421346e-06,
      "loss": 1.5035,
      "step": 711
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.9723347976661625,
      "learning_rate": 1.9999930596832585e-06,
      "loss": 1.4268,
      "step": 712
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.280480245670652,
      "learning_rate": 1.999993039896216e-06,
      "loss": 1.5342,
      "step": 713
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.594520301277882,
      "learning_rate": 1.999993020081007e-06,
      "loss": 1.5057,
      "step": 714
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.0248961526344385,
      "learning_rate": 1.9999930002376316e-06,
      "loss": 1.5489,
      "step": 715
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.710291069541927,
      "learning_rate": 1.999992980366089e-06,
      "loss": 1.4276,
      "step": 716
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.12715968906524,
      "learning_rate": 1.999992960466381e-06,
      "loss": 1.4865,
      "step": 717
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.916214864105662,
      "learning_rate": 1.999992940538506e-06,
      "loss": 1.4147,
      "step": 718
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.383613097319036,
      "learning_rate": 1.999992920582465e-06,
      "loss": 1.3599,
      "step": 719
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.341289838892921,
      "learning_rate": 1.999992900598257e-06,
      "loss": 1.4866,
      "step": 720
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.032528541347688,
      "learning_rate": 1.9999928805858824e-06,
      "loss": 1.5285,
      "step": 721
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.113668938434738,
      "learning_rate": 1.9999928605453417e-06,
      "loss": 1.4797,
      "step": 722
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.891988446714629,
      "learning_rate": 1.9999928404766345e-06,
      "loss": 1.6038,
      "step": 723
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.679783775035857,
      "learning_rate": 1.9999928203797606e-06,
      "loss": 1.4729,
      "step": 724
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.772986476245937,
      "learning_rate": 1.9999928002547202e-06,
      "loss": 1.429,
      "step": 725
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.468260713060045,
      "learning_rate": 1.999992780101514e-06,
      "loss": 1.3496,
      "step": 726
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.7466257837247365,
      "learning_rate": 1.9999927599201404e-06,
      "loss": 1.5408,
      "step": 727
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.584370484620455,
      "learning_rate": 1.999992739710601e-06,
      "loss": 1.4405,
      "step": 728
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.783272641625955,
      "learning_rate": 1.9999927194728945e-06,
      "loss": 1.4221,
      "step": 729
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.52556643023439,
      "learning_rate": 1.9999926992070223e-06,
      "loss": 1.3328,
      "step": 730
    },
    {
      "epoch": 0.0,
      "eval_loss": 1.6846652030944824,
      "eval_runtime": 4.641,
      "eval_samples_per_second": 1.939,
      "eval_steps_per_second": 1.077,
      "step": 730
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.657217221686939,
      "learning_rate": 1.999992678912983e-06,
      "loss": 1.4454,
      "step": 731
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.590847103289048,
      "learning_rate": 1.9999926585907777e-06,
      "loss": 1.3924,
      "step": 732
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.336722077748262,
      "learning_rate": 1.9999926382404054e-06,
      "loss": 1.4235,
      "step": 733
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.450736230018037,
      "learning_rate": 1.999992617861867e-06,
      "loss": 1.3447,
      "step": 734
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.138005519098043,
      "learning_rate": 1.9999925974551625e-06,
      "loss": 1.5023,
      "step": 735
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.723499031096059,
      "learning_rate": 1.9999925770202907e-06,
      "loss": 1.5875,
      "step": 736
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.103542969210838,
      "learning_rate": 1.999992556557253e-06,
      "loss": 1.6389,
      "step": 737
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.358708666669913,
      "learning_rate": 1.9999925360660486e-06,
      "loss": 1.5501,
      "step": 738
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.8728681154106,
      "learning_rate": 1.999992515546678e-06,
      "loss": 1.5199,
      "step": 739
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.254099594123959,
      "learning_rate": 1.9999924949991406e-06,
      "loss": 1.2561,
      "step": 740
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.841976889554841,
      "learning_rate": 1.999992474423437e-06,
      "loss": 1.4955,
      "step": 741
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.3286918973656965,
      "learning_rate": 1.999992453819567e-06,
      "loss": 1.5191,
      "step": 742
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.100923749893232,
      "learning_rate": 1.99999243318753e-06,
      "loss": 1.5339,
      "step": 743
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.937435968281523,
      "learning_rate": 1.999992412527327e-06,
      "loss": 1.5689,
      "step": 744
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.794383156069311,
      "learning_rate": 1.9999923918389573e-06,
      "loss": 1.4932,
      "step": 745
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.328354730784901,
      "learning_rate": 1.9999923711224215e-06,
      "loss": 1.4243,
      "step": 746
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.585973437130427,
      "learning_rate": 1.999992350377719e-06,
      "loss": 1.5275,
      "step": 747
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.252037658328408,
      "learning_rate": 1.99999232960485e-06,
      "loss": 1.6859,
      "step": 748
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.999962938796536,
      "learning_rate": 1.9999923088038147e-06,
      "loss": 1.3763,
      "step": 749
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.6706037411530135,
      "learning_rate": 1.999992287974613e-06,
      "loss": 1.4393,
      "step": 750
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.644118215700577,
      "learning_rate": 1.9999922671172445e-06,
      "loss": 1.3404,
      "step": 751
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.762262865598859,
      "learning_rate": 1.9999922462317096e-06,
      "loss": 1.1912,
      "step": 752
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.509565742435461,
      "learning_rate": 1.9999922253180088e-06,
      "loss": 1.5547,
      "step": 753
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.906236638847442,
      "learning_rate": 1.999992204376141e-06,
      "loss": 1.5497,
      "step": 754
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.389720327672479,
      "learning_rate": 1.999992183406107e-06,
      "loss": 1.5383,
      "step": 755
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.936731396149539,
      "learning_rate": 1.9999921624079066e-06,
      "loss": 1.5598,
      "step": 756
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.891074624493145,
      "learning_rate": 1.9999921413815393e-06,
      "loss": 1.4646,
      "step": 757
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.931816574873316,
      "learning_rate": 1.9999921203270057e-06,
      "loss": 1.5956,
      "step": 758
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.631245635122659,
      "learning_rate": 1.9999920992443056e-06,
      "loss": 1.3381,
      "step": 759
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.69596344452636,
      "learning_rate": 1.9999920781334392e-06,
      "loss": 1.4328,
      "step": 760
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.630359889510624,
      "learning_rate": 1.999992056994407e-06,
      "loss": 1.4445,
      "step": 761
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.4898331473385795,
      "learning_rate": 1.9999920358272076e-06,
      "loss": 1.4693,
      "step": 762
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.976976788582667,
      "learning_rate": 1.9999920146318416e-06,
      "loss": 1.4112,
      "step": 763
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.642721073810829,
      "learning_rate": 1.9999919934083092e-06,
      "loss": 1.507,
      "step": 764
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.938224038260358,
      "learning_rate": 1.999991972156611e-06,
      "loss": 1.4828,
      "step": 765
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.4124846916190545,
      "learning_rate": 1.9999919508767457e-06,
      "loss": 1.5935,
      "step": 766
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.0298956818677665,
      "learning_rate": 1.9999919295687146e-06,
      "loss": 1.5506,
      "step": 767
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.902554353229615,
      "learning_rate": 1.9999919082325163e-06,
      "loss": 1.4076,
      "step": 768
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.046670733247908,
      "learning_rate": 1.999991886868152e-06,
      "loss": 1.5376,
      "step": 769
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.029796705277875,
      "learning_rate": 1.999991865475621e-06,
      "loss": 1.5587,
      "step": 770
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.392708356030036,
      "learning_rate": 1.9999918440549237e-06,
      "loss": 1.5396,
      "step": 771
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.421457119639919,
      "learning_rate": 1.9999918226060602e-06,
      "loss": 1.4929,
      "step": 772
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.752656726954079,
      "learning_rate": 1.99999180112903e-06,
      "loss": 1.1638,
      "step": 773
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.831954478385221,
      "learning_rate": 1.9999917796238333e-06,
      "loss": 1.4306,
      "step": 774
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.101830458146313,
      "learning_rate": 1.99999175809047e-06,
      "loss": 1.4116,
      "step": 775
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.404228318451748,
      "learning_rate": 1.9999917365289403e-06,
      "loss": 1.5313,
      "step": 776
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.736260085200967,
      "learning_rate": 1.9999917149392445e-06,
      "loss": 1.4743,
      "step": 777
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.839914114082853,
      "learning_rate": 1.9999916933213823e-06,
      "loss": 1.4311,
      "step": 778
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.591247776677874,
      "learning_rate": 1.9999916716753536e-06,
      "loss": 1.4233,
      "step": 779
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.6578387294024335,
      "learning_rate": 1.9999916500011582e-06,
      "loss": 1.5085,
      "step": 780
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.93242923456189,
      "learning_rate": 1.9999916282987964e-06,
      "loss": 1.5293,
      "step": 781
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.84156183065501,
      "learning_rate": 1.9999916065682682e-06,
      "loss": 1.4976,
      "step": 782
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.780256207356861,
      "learning_rate": 1.9999915848095736e-06,
      "loss": 1.3376,
      "step": 783
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.519528759054672,
      "learning_rate": 1.9999915630227127e-06,
      "loss": 1.4027,
      "step": 784
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.363927799642175,
      "learning_rate": 1.9999915412076853e-06,
      "loss": 1.4659,
      "step": 785
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.342877391406383,
      "learning_rate": 1.9999915193644916e-06,
      "loss": 1.4013,
      "step": 786
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.813148676207006,
      "learning_rate": 1.999991497493131e-06,
      "loss": 1.4737,
      "step": 787
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.745668191994972,
      "learning_rate": 1.999991475593604e-06,
      "loss": 1.2512,
      "step": 788
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.796669035814396,
      "learning_rate": 1.999991453665911e-06,
      "loss": 1.318,
      "step": 789
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.807933615414672,
      "learning_rate": 1.999991431710051e-06,
      "loss": 1.3862,
      "step": 790
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.092568125176769,
      "learning_rate": 1.9999914097260254e-06,
      "loss": 1.5014,
      "step": 791
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.824966187432624,
      "learning_rate": 1.999991387713833e-06,
      "loss": 1.4623,
      "step": 792
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.0068907060455485,
      "learning_rate": 1.9999913656734736e-06,
      "loss": 1.4753,
      "step": 793
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.838451440188585,
      "learning_rate": 1.9999913436049483e-06,
      "loss": 1.4478,
      "step": 794
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.553033347591773,
      "learning_rate": 1.9999913215082567e-06,
      "loss": 1.5087,
      "step": 795
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.639914766985077,
      "learning_rate": 1.9999912993833987e-06,
      "loss": 1.331,
      "step": 796
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.035741243327864,
      "learning_rate": 1.999991277230374e-06,
      "loss": 1.4473,
      "step": 797
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.0968663437701975,
      "learning_rate": 1.9999912550491826e-06,
      "loss": 1.5815,
      "step": 798
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.730455270173384,
      "learning_rate": 1.999991232839825e-06,
      "loss": 1.4369,
      "step": 799
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.219089987469455,
      "learning_rate": 1.9999912106023014e-06,
      "loss": 1.5169,
      "step": 800
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.750971191979698,
      "learning_rate": 1.999991188336611e-06,
      "loss": 1.4153,
      "step": 801
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.406182419591723,
      "learning_rate": 1.9999911660427543e-06,
      "loss": 1.4437,
      "step": 802
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.063524970869385,
      "learning_rate": 1.9999911437207307e-06,
      "loss": 1.507,
      "step": 803
    },
    {
      "epoch": 0.01,
      "eval_loss": 1.6881712675094604,
      "eval_runtime": 4.626,
      "eval_samples_per_second": 1.946,
      "eval_steps_per_second": 1.081,
      "step": 803
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.467776894504789,
      "learning_rate": 1.999991121370541e-06,
      "loss": 1.522,
      "step": 804
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.971642369149256,
      "learning_rate": 1.999991098992185e-06,
      "loss": 1.5321,
      "step": 805
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.814938121020308,
      "learning_rate": 1.999991076585663e-06,
      "loss": 1.433,
      "step": 806
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.2209822250293065,
      "learning_rate": 1.9999910541509737e-06,
      "loss": 1.335,
      "step": 807
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.296510549803633,
      "learning_rate": 1.9999910316881186e-06,
      "loss": 1.7574,
      "step": 808
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.913991460057143,
      "learning_rate": 1.9999910091970967e-06,
      "loss": 1.5179,
      "step": 809
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.668407347559244,
      "learning_rate": 1.9999909866779085e-06,
      "loss": 1.3712,
      "step": 810
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.453144611957878,
      "learning_rate": 1.9999909641305538e-06,
      "loss": 1.3764,
      "step": 811
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.2650106360957185,
      "learning_rate": 1.9999909415550327e-06,
      "loss": 1.4595,
      "step": 812
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.250833767951291,
      "learning_rate": 1.9999909189513457e-06,
      "loss": 1.3007,
      "step": 813
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.354917246649793,
      "learning_rate": 1.999990896319492e-06,
      "loss": 1.6322,
      "step": 814
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.012749978414504,
      "learning_rate": 1.9999908736594716e-06,
      "loss": 1.5338,
      "step": 815
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.848442817388502,
      "learning_rate": 1.999990850971285e-06,
      "loss": 1.2774,
      "step": 816
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.984924234695706,
      "learning_rate": 1.9999908282549316e-06,
      "loss": 1.4143,
      "step": 817
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.762256555459132,
      "learning_rate": 1.9999908055104122e-06,
      "loss": 1.3228,
      "step": 818
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.844771964077254,
      "learning_rate": 1.9999907827377265e-06,
      "loss": 1.4724,
      "step": 819
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.952200259787622,
      "learning_rate": 1.999990759936874e-06,
      "loss": 1.4344,
      "step": 820
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.711382910161586,
      "learning_rate": 1.999990737107855e-06,
      "loss": 1.5817,
      "step": 821
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.57192836371835,
      "learning_rate": 1.99999071425067e-06,
      "loss": 1.4359,
      "step": 822
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.2612613709184775,
      "learning_rate": 1.9999906913653187e-06,
      "loss": 1.683,
      "step": 823
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.651207872271117,
      "learning_rate": 1.9999906684518005e-06,
      "loss": 1.4202,
      "step": 824
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.33548743621847,
      "learning_rate": 1.999990645510116e-06,
      "loss": 1.2962,
      "step": 825
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.998357754076746,
      "learning_rate": 1.9999906225402655e-06,
      "loss": 1.4576,
      "step": 826
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.936966406696661,
      "learning_rate": 1.9999905995422483e-06,
      "loss": 1.4039,
      "step": 827
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.330845387260364,
      "learning_rate": 1.9999905765160646e-06,
      "loss": 1.5228,
      "step": 828
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8056099201876705,
      "learning_rate": 1.9999905534617145e-06,
      "loss": 1.3755,
      "step": 829
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.029573312366041,
      "learning_rate": 1.999990530379198e-06,
      "loss": 1.5501,
      "step": 830
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.539685125822293,
      "learning_rate": 1.999990507268515e-06,
      "loss": 1.3357,
      "step": 831
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.880928277155648,
      "learning_rate": 1.9999904841296656e-06,
      "loss": 1.6046,
      "step": 832
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.503556293375853,
      "learning_rate": 1.99999046096265e-06,
      "loss": 1.7223,
      "step": 833
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.225897163215507,
      "learning_rate": 1.999990437767468e-06,
      "loss": 1.3195,
      "step": 834
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.690064080550938,
      "learning_rate": 1.9999904145441196e-06,
      "loss": 1.6465,
      "step": 835
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.5459726864219725,
      "learning_rate": 1.999990391292605e-06,
      "loss": 1.4026,
      "step": 836
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.680412720191206,
      "learning_rate": 1.9999903680129237e-06,
      "loss": 1.4353,
      "step": 837
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.622275997037577,
      "learning_rate": 1.9999903447050758e-06,
      "loss": 1.5267,
      "step": 838
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.741060382625769,
      "learning_rate": 1.999990321369062e-06,
      "loss": 1.4955,
      "step": 839
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.875869067274173,
      "learning_rate": 1.999990298004881e-06,
      "loss": 1.5815,
      "step": 840
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.6600192764255395,
      "learning_rate": 1.9999902746125344e-06,
      "loss": 1.3205,
      "step": 841
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.551113040685454,
      "learning_rate": 1.999990251192021e-06,
      "loss": 1.6559,
      "step": 842
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.3125600816949365,
      "learning_rate": 1.9999902277433414e-06,
      "loss": 1.422,
      "step": 843
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.0897853002323155,
      "learning_rate": 1.999990204266495e-06,
      "loss": 1.429,
      "step": 844
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.620329234269747,
      "learning_rate": 1.999990180761483e-06,
      "loss": 1.3901,
      "step": 845
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.952006426703027,
      "learning_rate": 1.999990157228304e-06,
      "loss": 1.4605,
      "step": 846
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.051790815462307,
      "learning_rate": 1.999990133666959e-06,
      "loss": 1.5361,
      "step": 847
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.723454559230461,
      "learning_rate": 1.999990110077447e-06,
      "loss": 1.4063,
      "step": 848
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.714623591910726,
      "learning_rate": 1.999990086459769e-06,
      "loss": 1.4844,
      "step": 849
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.866772682439225,
      "learning_rate": 1.999990062813924e-06,
      "loss": 1.4243,
      "step": 850
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.982727315839908,
      "learning_rate": 1.999990039139913e-06,
      "loss": 1.5169,
      "step": 851
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.88323303732355,
      "learning_rate": 1.9999900154377363e-06,
      "loss": 1.331,
      "step": 852
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.98858734123965,
      "learning_rate": 1.9999899917073925e-06,
      "loss": 1.4386,
      "step": 853
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.7732616301555,
      "learning_rate": 1.9999899679488823e-06,
      "loss": 1.483,
      "step": 854
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.195015855643826,
      "learning_rate": 1.9999899441622062e-06,
      "loss": 1.489,
      "step": 855
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.777377833871161,
      "learning_rate": 1.9999899203473633e-06,
      "loss": 1.3607,
      "step": 856
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.851948399516926,
      "learning_rate": 1.999989896504354e-06,
      "loss": 1.3547,
      "step": 857
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.7729850075029905,
      "learning_rate": 1.9999898726331787e-06,
      "loss": 1.3561,
      "step": 858
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.764811136272429,
      "learning_rate": 1.9999898487338367e-06,
      "loss": 1.3792,
      "step": 859
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.889712073966274,
      "learning_rate": 1.999989824806328e-06,
      "loss": 1.3774,
      "step": 860
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.465422723384496,
      "learning_rate": 1.9999898008506533e-06,
      "loss": 1.3464,
      "step": 861
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.480222802088414,
      "learning_rate": 1.9999897768668125e-06,
      "loss": 1.5737,
      "step": 862
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.492784136252579,
      "learning_rate": 1.999989752854805e-06,
      "loss": 1.5299,
      "step": 863
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.194980321543087,
      "learning_rate": 1.999989728814631e-06,
      "loss": 1.2601,
      "step": 864
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.373628545087017,
      "learning_rate": 1.9999897047462905e-06,
      "loss": 1.6205,
      "step": 865
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.834528620174113,
      "learning_rate": 1.9999896806497837e-06,
      "loss": 1.3969,
      "step": 866
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.365382013388016,
      "learning_rate": 1.999989656525111e-06,
      "loss": 1.4079,
      "step": 867
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.10671741683615,
      "learning_rate": 1.9999896323722714e-06,
      "loss": 1.5323,
      "step": 868
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.794974200564266,
      "learning_rate": 1.9999896081912655e-06,
      "loss": 1.5685,
      "step": 869
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.460786387110753,
      "learning_rate": 1.999989583982093e-06,
      "loss": 1.4878,
      "step": 870
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.954812125157006,
      "learning_rate": 1.999989559744755e-06,
      "loss": 1.6783,
      "step": 871
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.893553227272597,
      "learning_rate": 1.9999895354792497e-06,
      "loss": 1.5179,
      "step": 872
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.971125258081157,
      "learning_rate": 1.9999895111855786e-06,
      "loss": 1.42,
      "step": 873
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.466531237826114,
      "learning_rate": 1.9999894868637408e-06,
      "loss": 1.4594,
      "step": 874
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.643918053264017,
      "learning_rate": 1.9999894625137365e-06,
      "loss": 1.4087,
      "step": 875
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.946505174466154,
      "learning_rate": 1.999989438135566e-06,
      "loss": 1.5287,
      "step": 876
    },
    {
      "epoch": 0.01,
      "eval_loss": 1.6835453510284424,
      "eval_runtime": 4.6442,
      "eval_samples_per_second": 1.938,
      "eval_steps_per_second": 1.077,
      "step": 876
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.239412305756631,
      "learning_rate": 1.9999894137292292e-06,
      "loss": 1.3614,
      "step": 877
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.5395595579858075,
      "learning_rate": 1.999989389294726e-06,
      "loss": 1.5485,
      "step": 878
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.712593321081159,
      "learning_rate": 1.9999893648320564e-06,
      "loss": 1.4338,
      "step": 879
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.458574377375978,
      "learning_rate": 1.9999893403412202e-06,
      "loss": 1.41,
      "step": 880
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.551492101493687,
      "learning_rate": 1.999989315822218e-06,
      "loss": 1.3577,
      "step": 881
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.207575524594766,
      "learning_rate": 1.999989291275049e-06,
      "loss": 1.5347,
      "step": 882
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.6630265828087,
      "learning_rate": 1.999989266699714e-06,
      "loss": 1.5364,
      "step": 883
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.879665992047349,
      "learning_rate": 1.9999892420962124e-06,
      "loss": 1.4612,
      "step": 884
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.561642320920106,
      "learning_rate": 1.9999892174645447e-06,
      "loss": 1.3238,
      "step": 885
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.803746896535269,
      "learning_rate": 1.9999891928047106e-06,
      "loss": 1.3759,
      "step": 886
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.881001838381561,
      "learning_rate": 1.99998916811671e-06,
      "loss": 1.3259,
      "step": 887
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.9185588818445165,
      "learning_rate": 1.9999891434005433e-06,
      "loss": 1.468,
      "step": 888
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.4140094573174045,
      "learning_rate": 1.9999891186562096e-06,
      "loss": 1.5786,
      "step": 889
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.4016730660955625,
      "learning_rate": 1.99998909388371e-06,
      "loss": 1.5057,
      "step": 890
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.4186135870000145,
      "learning_rate": 1.999989069083044e-06,
      "loss": 1.5624,
      "step": 891
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.747116381476095,
      "learning_rate": 1.9999890442542115e-06,
      "loss": 1.4637,
      "step": 892
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.40275388365515,
      "learning_rate": 1.9999890193972127e-06,
      "loss": 1.5146,
      "step": 893
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.991497086501674,
      "learning_rate": 1.9999889945120475e-06,
      "loss": 1.4213,
      "step": 894
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.054323720643013,
      "learning_rate": 1.999988969598716e-06,
      "loss": 1.523,
      "step": 895
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.610288116972087,
      "learning_rate": 1.9999889446572184e-06,
      "loss": 1.351,
      "step": 896
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.3918041618620896,
      "learning_rate": 1.999988919687554e-06,
      "loss": 1.5724,
      "step": 897
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.833705802062882,
      "learning_rate": 1.9999888946897233e-06,
      "loss": 1.572,
      "step": 898
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.166823884179871,
      "learning_rate": 1.999988869663726e-06,
      "loss": 1.3641,
      "step": 899
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.444407021344696,
      "learning_rate": 1.999988844609563e-06,
      "loss": 1.4219,
      "step": 900
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.960252287317098,
      "learning_rate": 1.9999888195272332e-06,
      "loss": 1.4062,
      "step": 901
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.074258501650094,
      "learning_rate": 1.9999887944167374e-06,
      "loss": 1.5055,
      "step": 902
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.815381295386099,
      "learning_rate": 1.9999887692780747e-06,
      "loss": 1.5434,
      "step": 903
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.811556472980446,
      "learning_rate": 1.999988744111246e-06,
      "loss": 1.4531,
      "step": 904
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.5338580513110704,
      "learning_rate": 1.999988718916251e-06,
      "loss": 1.5654,
      "step": 905
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8547503509093985,
      "learning_rate": 1.9999886936930897e-06,
      "loss": 1.4963,
      "step": 906
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.9529825947141655,
      "learning_rate": 1.9999886684417614e-06,
      "loss": 1.424,
      "step": 907
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.992257236115625,
      "learning_rate": 1.9999886431622673e-06,
      "loss": 1.4438,
      "step": 908
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.607129503512642,
      "learning_rate": 1.999988617854607e-06,
      "loss": 1.4505,
      "step": 909
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.157982067067892,
      "learning_rate": 1.99998859251878e-06,
      "loss": 1.5385,
      "step": 910
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.084561010511645,
      "learning_rate": 1.999988567154787e-06,
      "loss": 1.4966,
      "step": 911
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.81877282821209,
      "learning_rate": 1.999988541762627e-06,
      "loss": 1.4731,
      "step": 912
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.009288310745413,
      "learning_rate": 1.9999885163423014e-06,
      "loss": 1.432,
      "step": 913
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.2256414414976184,
      "learning_rate": 1.999988490893809e-06,
      "loss": 1.5411,
      "step": 914
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.261047012186723,
      "learning_rate": 1.99998846541715e-06,
      "loss": 1.6397,
      "step": 915
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.657445695072026,
      "learning_rate": 1.9999884399123252e-06,
      "loss": 1.4764,
      "step": 916
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.668390029076262,
      "learning_rate": 1.999988414379334e-06,
      "loss": 1.5984,
      "step": 917
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.810866617300379,
      "learning_rate": 1.9999883888181764e-06,
      "loss": 1.4278,
      "step": 918
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.825331474808909,
      "learning_rate": 1.9999883632288524e-06,
      "loss": 1.5525,
      "step": 919
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.055755208418581,
      "learning_rate": 1.999988337611362e-06,
      "loss": 1.4505,
      "step": 920
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.750972505525963,
      "learning_rate": 1.999988311965705e-06,
      "loss": 1.486,
      "step": 921
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.217010575239663,
      "learning_rate": 1.999988286291882e-06,
      "loss": 1.7214,
      "step": 922
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.507300171459353,
      "learning_rate": 1.9999882605898925e-06,
      "loss": 1.346,
      "step": 923
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.017321187244063,
      "learning_rate": 1.999988234859737e-06,
      "loss": 1.5818,
      "step": 924
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.028645991367795,
      "learning_rate": 1.999988209101415e-06,
      "loss": 1.5805,
      "step": 925
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.2062677616231765,
      "learning_rate": 1.9999881833149264e-06,
      "loss": 1.6537,
      "step": 926
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.145636588072226,
      "learning_rate": 1.9999881575002717e-06,
      "loss": 1.4681,
      "step": 927
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.828710624451239,
      "learning_rate": 1.99998813165745e-06,
      "loss": 1.496,
      "step": 928
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.549188105899666,
      "learning_rate": 1.9999881057864627e-06,
      "loss": 1.3865,
      "step": 929
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.443520100837284,
      "learning_rate": 1.9999880798873093e-06,
      "loss": 1.2829,
      "step": 930
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.13373771642738,
      "learning_rate": 1.999988053959989e-06,
      "loss": 1.5362,
      "step": 931
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.999494186839008,
      "learning_rate": 1.9999880280045025e-06,
      "loss": 1.3462,
      "step": 932
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.190959037095125,
      "learning_rate": 1.9999880020208495e-06,
      "loss": 1.4841,
      "step": 933
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.780119303450498,
      "learning_rate": 1.9999879760090306e-06,
      "loss": 1.5854,
      "step": 934
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.066755607666266,
      "learning_rate": 1.9999879499690452e-06,
      "loss": 1.3359,
      "step": 935
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.252072684548736,
      "learning_rate": 1.999987923900893e-06,
      "loss": 1.4493,
      "step": 936
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.711513346090706,
      "learning_rate": 1.999987897804575e-06,
      "loss": 1.2207,
      "step": 937
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.728638763732817,
      "learning_rate": 1.9999878716800904e-06,
      "loss": 1.3846,
      "step": 938
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.278677797997787,
      "learning_rate": 1.9999878455274396e-06,
      "loss": 1.336,
      "step": 939
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.0220126825749425,
      "learning_rate": 1.9999878193466223e-06,
      "loss": 1.4228,
      "step": 940
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.034910106052013,
      "learning_rate": 1.999987793137639e-06,
      "loss": 1.3907,
      "step": 941
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.889404163229133,
      "learning_rate": 1.9999877669004894e-06,
      "loss": 1.4144,
      "step": 942
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.162469539125017,
      "learning_rate": 1.999987740635173e-06,
      "loss": 1.5852,
      "step": 943
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.399193356261801,
      "learning_rate": 1.9999877143416906e-06,
      "loss": 1.5513,
      "step": 944
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.138316370644956,
      "learning_rate": 1.9999876880200418e-06,
      "loss": 1.467,
      "step": 945
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.767554379371538,
      "learning_rate": 1.9999876616702266e-06,
      "loss": 1.4015,
      "step": 946
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.432154096584277,
      "learning_rate": 1.9999876352922455e-06,
      "loss": 1.3144,
      "step": 947
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.6901051217643035,
      "learning_rate": 1.9999876088860975e-06,
      "loss": 1.6762,
      "step": 948
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.89608317041777,
      "learning_rate": 1.999987582451783e-06,
      "loss": 1.4469,
      "step": 949
    },
    {
      "epoch": 0.01,
      "eval_loss": 1.6775009632110596,
      "eval_runtime": 4.6401,
      "eval_samples_per_second": 1.94,
      "eval_steps_per_second": 1.078,
      "step": 949
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.685652449039854,
      "learning_rate": 1.999987555989303e-06,
      "loss": 1.3646,
      "step": 950
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.980010862799234,
      "learning_rate": 1.9999875294986562e-06,
      "loss": 1.5188,
      "step": 951
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.218097124666707,
      "learning_rate": 1.999987502979843e-06,
      "loss": 1.4779,
      "step": 952
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.803466971822877,
      "learning_rate": 1.9999874764328637e-06,
      "loss": 1.3129,
      "step": 953
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.729648111188237,
      "learning_rate": 1.999987449857718e-06,
      "loss": 1.4093,
      "step": 954
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.666417668066331,
      "learning_rate": 1.999987423254406e-06,
      "loss": 1.637,
      "step": 955
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.92053994601316,
      "learning_rate": 1.999987396622928e-06,
      "loss": 1.4616,
      "step": 956
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.729794144404417,
      "learning_rate": 1.999987369963283e-06,
      "loss": 1.2795,
      "step": 957
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.48059643960075,
      "learning_rate": 1.999987343275472e-06,
      "loss": 1.4871,
      "step": 958
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.807955183113965,
      "learning_rate": 1.9999873165594945e-06,
      "loss": 1.4542,
      "step": 959
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.3782702475558475,
      "learning_rate": 1.999987289815351e-06,
      "loss": 1.4009,
      "step": 960
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.465383132137198,
      "learning_rate": 1.999987263043041e-06,
      "loss": 1.5234,
      "step": 961
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.863141270288387,
      "learning_rate": 1.9999872362425646e-06,
      "loss": 1.328,
      "step": 962
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.513281389437473,
      "learning_rate": 1.999987209413922e-06,
      "loss": 1.3662,
      "step": 963
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.182406119541521,
      "learning_rate": 1.9999871825571133e-06,
      "loss": 1.4304,
      "step": 964
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.682777815258765,
      "learning_rate": 1.999987155672138e-06,
      "loss": 1.3698,
      "step": 965
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.698433297833218,
      "learning_rate": 1.999987128758997e-06,
      "loss": 1.4941,
      "step": 966
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.122381225960383,
      "learning_rate": 1.9999871018176888e-06,
      "loss": 1.4768,
      "step": 967
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.72681263242357,
      "learning_rate": 1.999987074848215e-06,
      "loss": 1.3749,
      "step": 968
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.507215592270518,
      "learning_rate": 1.999987047850574e-06,
      "loss": 1.5244,
      "step": 969
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.870388076906783,
      "learning_rate": 1.9999870208247677e-06,
      "loss": 1.5131,
      "step": 970
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.513166729875274,
      "learning_rate": 1.999986993770794e-06,
      "loss": 1.4241,
      "step": 971
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.820438122653338,
      "learning_rate": 1.999986966688655e-06,
      "loss": 1.2826,
      "step": 972
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.839651321923382,
      "learning_rate": 1.9999869395783495e-06,
      "loss": 1.2476,
      "step": 973
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.720429959744313,
      "learning_rate": 1.999986912439877e-06,
      "loss": 1.4291,
      "step": 974
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.4799874056887745,
      "learning_rate": 1.999986885273239e-06,
      "loss": 1.3034,
      "step": 975
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.966740554354801,
      "learning_rate": 1.9999868580784343e-06,
      "loss": 1.346,
      "step": 976
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.748631285522209,
      "learning_rate": 1.9999868308554637e-06,
      "loss": 1.3915,
      "step": 977
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.117205503667891,
      "learning_rate": 1.9999868036043262e-06,
      "loss": 1.531,
      "step": 978
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.761977448039064,
      "learning_rate": 1.999986776325023e-06,
      "loss": 1.5092,
      "step": 979
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.383839203745775,
      "learning_rate": 1.999986749017553e-06,
      "loss": 1.4312,
      "step": 980
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.898589027829305,
      "learning_rate": 1.999986721681917e-06,
      "loss": 1.4059,
      "step": 981
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.05243832236956,
      "learning_rate": 1.9999866943181144e-06,
      "loss": 1.4345,
      "step": 982
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.171888830122464,
      "learning_rate": 1.999986666926146e-06,
      "loss": 1.5769,
      "step": 983
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.268910529823187,
      "learning_rate": 1.999986639506011e-06,
      "loss": 1.5645,
      "step": 984
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.1967337373113525,
      "learning_rate": 1.9999866120577097e-06,
      "loss": 1.5157,
      "step": 985
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.42988030195917,
      "learning_rate": 1.999986584581242e-06,
      "loss": 1.5816,
      "step": 986
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.73138904838603,
      "learning_rate": 1.999986557076608e-06,
      "loss": 1.5813,
      "step": 987
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.753580436572943,
      "learning_rate": 1.999986529543808e-06,
      "loss": 1.4126,
      "step": 988
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.7046114145354565,
      "learning_rate": 1.999986501982841e-06,
      "loss": 1.5255,
      "step": 989
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.691510890894805,
      "learning_rate": 1.9999864743937083e-06,
      "loss": 1.4457,
      "step": 990
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.010108387001166,
      "learning_rate": 1.9999864467764095e-06,
      "loss": 1.2545,
      "step": 991
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.100682664462816,
      "learning_rate": 1.9999864191309444e-06,
      "loss": 1.39,
      "step": 992
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.918040690814786,
      "learning_rate": 1.9999863914573124e-06,
      "loss": 1.3931,
      "step": 993
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.32376088870836,
      "learning_rate": 1.9999863637555145e-06,
      "loss": 1.5364,
      "step": 994
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.404294035760159,
      "learning_rate": 1.9999863360255502e-06,
      "loss": 1.6188,
      "step": 995
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.395708410571946,
      "learning_rate": 1.99998630826742e-06,
      "loss": 1.4324,
      "step": 996
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.64843688225223,
      "learning_rate": 1.999986280481123e-06,
      "loss": 1.4221,
      "step": 997
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.483948754696783,
      "learning_rate": 1.99998625266666e-06,
      "loss": 1.4758,
      "step": 998
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.680984970928925,
      "learning_rate": 1.999986224824031e-06,
      "loss": 1.3525,
      "step": 999
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.5483509269337326,
      "learning_rate": 1.9999861969532346e-06,
      "loss": 1.5931,
      "step": 1000
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.685591189533344,
      "learning_rate": 1.999986169054273e-06,
      "loss": 1.4752,
      "step": 1001
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.027111632819377,
      "learning_rate": 1.9999861411271447e-06,
      "loss": 1.4839,
      "step": 1002
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.813630389301654,
      "learning_rate": 1.99998611317185e-06,
      "loss": 1.2419,
      "step": 1003
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8555527438063875,
      "learning_rate": 1.9999860851883896e-06,
      "loss": 1.5733,
      "step": 1004
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.863459059811092,
      "learning_rate": 1.9999860571767623e-06,
      "loss": 1.4519,
      "step": 1005
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.069266897105678,
      "learning_rate": 1.999986029136969e-06,
      "loss": 1.5143,
      "step": 1006
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.396817433935851,
      "learning_rate": 1.9999860010690093e-06,
      "loss": 1.4442,
      "step": 1007
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.846992814886912,
      "learning_rate": 1.9999859729728833e-06,
      "loss": 1.514,
      "step": 1008
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.610189787028616,
      "learning_rate": 1.9999859448485912e-06,
      "loss": 1.3104,
      "step": 1009
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.131296471286178,
      "learning_rate": 1.999985916696133e-06,
      "loss": 1.3135,
      "step": 1010
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.091140293402042,
      "learning_rate": 1.999985888515508e-06,
      "loss": 1.5342,
      "step": 1011
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.120127148783947,
      "learning_rate": 1.999985860306717e-06,
      "loss": 1.5248,
      "step": 1012
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.479120892520487,
      "learning_rate": 1.9999858320697597e-06,
      "loss": 1.5342,
      "step": 1013
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.065215097181712,
      "learning_rate": 1.999985803804636e-06,
      "loss": 1.5236,
      "step": 1014
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.790915278833422,
      "learning_rate": 1.9999857755113463e-06,
      "loss": 1.5493,
      "step": 1015
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.245342572510602,
      "learning_rate": 1.99998574718989e-06,
      "loss": 1.4241,
      "step": 1016
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.450685311013884,
      "learning_rate": 1.9999857188402677e-06,
      "loss": 1.5392,
      "step": 1017
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.167929668678754,
      "learning_rate": 1.9999856904624786e-06,
      "loss": 1.4839,
      "step": 1018
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.655577606549074,
      "learning_rate": 1.999985662056524e-06,
      "loss": 1.3708,
      "step": 1019
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.083589051170835,
      "learning_rate": 1.9999856336224026e-06,
      "loss": 1.4276,
      "step": 1020
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.518979837632458,
      "learning_rate": 1.9999856051601156e-06,
      "loss": 1.5506,
      "step": 1021
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.022963635961508,
      "learning_rate": 1.9999855766696614e-06,
      "loss": 1.503,
      "step": 1022
    },
    {
      "epoch": 0.01,
      "eval_loss": 1.6751856803894043,
      "eval_runtime": 4.6457,
      "eval_samples_per_second": 1.937,
      "eval_steps_per_second": 1.076,
      "step": 1022
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.888363610346068,
      "learning_rate": 1.9999855481510416e-06,
      "loss": 1.3999,
      "step": 1023
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.785844821190048,
      "learning_rate": 1.999985519604255e-06,
      "loss": 1.5704,
      "step": 1024
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.817704365569473,
      "learning_rate": 1.9999854910293026e-06,
      "loss": 1.5568,
      "step": 1025
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.359712088912163,
      "learning_rate": 1.9999854624261837e-06,
      "loss": 1.2568,
      "step": 1026
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.591954237737313,
      "learning_rate": 1.999985433794899e-06,
      "loss": 1.4747,
      "step": 1027
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.102013647933071,
      "learning_rate": 1.9999854051354476e-06,
      "loss": 1.4146,
      "step": 1028
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.931384249700325,
      "learning_rate": 1.99998537644783e-06,
      "loss": 1.4324,
      "step": 1029
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.59036254886226,
      "learning_rate": 1.999985347732046e-06,
      "loss": 1.467,
      "step": 1030
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.994175894124642,
      "learning_rate": 1.999985318988096e-06,
      "loss": 1.3865,
      "step": 1031
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.545053819093002,
      "learning_rate": 1.999985290215979e-06,
      "loss": 1.4351,
      "step": 1032
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.004106893385651,
      "learning_rate": 1.999985261415697e-06,
      "loss": 1.4621,
      "step": 1033
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.279172341870177,
      "learning_rate": 1.9999852325872476e-06,
      "loss": 1.7436,
      "step": 1034
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.343377788723042,
      "learning_rate": 1.9999852037306325e-06,
      "loss": 1.5041,
      "step": 1035
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.268385681335901,
      "learning_rate": 1.999985174845851e-06,
      "loss": 1.5105,
      "step": 1036
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.048875950329858,
      "learning_rate": 1.9999851459329036e-06,
      "loss": 1.5769,
      "step": 1037
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.74228089002837,
      "learning_rate": 1.9999851169917897e-06,
      "loss": 1.5507,
      "step": 1038
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.698397540213852,
      "learning_rate": 1.9999850880225095e-06,
      "loss": 1.3009,
      "step": 1039
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.641564143897353,
      "learning_rate": 1.999985059025063e-06,
      "loss": 1.3415,
      "step": 1040
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.77396271067677,
      "learning_rate": 1.99998502999945e-06,
      "loss": 1.4983,
      "step": 1041
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.772762227532076,
      "learning_rate": 1.9999850009456713e-06,
      "loss": 1.3664,
      "step": 1042
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.914962455888292,
      "learning_rate": 1.999984971863726e-06,
      "loss": 1.4483,
      "step": 1043
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.292636683131925,
      "learning_rate": 1.9999849427536147e-06,
      "loss": 1.614,
      "step": 1044
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.816438676458731,
      "learning_rate": 1.999984913615337e-06,
      "loss": 1.4438,
      "step": 1045
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.61943572442077,
      "learning_rate": 1.999984884448893e-06,
      "loss": 1.3256,
      "step": 1046
    },
    {
      "epoch": 0.01,
      "grad_norm": 16.818243184224574,
      "learning_rate": 1.999984855254283e-06,
      "loss": 1.5516,
      "step": 1047
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.944906131189107,
      "learning_rate": 1.9999848260315064e-06,
      "loss": 1.4782,
      "step": 1048
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.372035876331552,
      "learning_rate": 1.9999847967805635e-06,
      "loss": 1.5814,
      "step": 1049
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.05598468387018,
      "learning_rate": 1.9999847675014548e-06,
      "loss": 1.4904,
      "step": 1050
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.983166835216965,
      "learning_rate": 1.9999847381941796e-06,
      "loss": 1.5657,
      "step": 1051
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.880234134964031,
      "learning_rate": 1.999984708858738e-06,
      "loss": 1.2615,
      "step": 1052
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.383111410300504,
      "learning_rate": 1.9999846794951305e-06,
      "loss": 1.5257,
      "step": 1053
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.319783609516555,
      "learning_rate": 1.9999846501033566e-06,
      "loss": 1.3974,
      "step": 1054
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.276781780830751,
      "learning_rate": 1.9999846206834163e-06,
      "loss": 1.5434,
      "step": 1055
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.704196444791288,
      "learning_rate": 1.99998459123531e-06,
      "loss": 1.4933,
      "step": 1056
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.249291359747204,
      "learning_rate": 1.9999845617590374e-06,
      "loss": 1.4506,
      "step": 1057
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.675801641902775,
      "learning_rate": 1.9999845322545983e-06,
      "loss": 1.582,
      "step": 1058
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.63075829111936,
      "learning_rate": 1.9999845027219933e-06,
      "loss": 1.5674,
      "step": 1059
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.305134362562168,
      "learning_rate": 1.999984473161222e-06,
      "loss": 1.1858,
      "step": 1060
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.923187333888181,
      "learning_rate": 1.999984443572284e-06,
      "loss": 1.5338,
      "step": 1061
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.615589160701651,
      "learning_rate": 1.9999844139551804e-06,
      "loss": 1.3402,
      "step": 1062
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.080506923679196,
      "learning_rate": 1.9999843843099103e-06,
      "loss": 1.4556,
      "step": 1063
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.739895475840536,
      "learning_rate": 1.9999843546364742e-06,
      "loss": 1.4932,
      "step": 1064
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.404861163545689,
      "learning_rate": 1.9999843249348718e-06,
      "loss": 1.3194,
      "step": 1065
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.345308311669248,
      "learning_rate": 1.999984295205103e-06,
      "loss": 1.3354,
      "step": 1066
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.01753123609855,
      "learning_rate": 1.9999842654471677e-06,
      "loss": 1.4866,
      "step": 1067
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.5591172662604595,
      "learning_rate": 1.9999842356610664e-06,
      "loss": 1.3546,
      "step": 1068
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.7557963260332565,
      "learning_rate": 1.999984205846799e-06,
      "loss": 1.4518,
      "step": 1069
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.562175319819017,
      "learning_rate": 1.9999841760043657e-06,
      "loss": 1.3072,
      "step": 1070
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.812788016867621,
      "learning_rate": 1.9999841461337653e-06,
      "loss": 1.371,
      "step": 1071
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.67438764053753,
      "learning_rate": 1.9999841162349994e-06,
      "loss": 1.4096,
      "step": 1072
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.993252181714291,
      "learning_rate": 1.999984086308067e-06,
      "loss": 1.4942,
      "step": 1073
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.910843860455304,
      "learning_rate": 1.9999840563529685e-06,
      "loss": 1.4151,
      "step": 1074
    },
    {
      "epoch": 0.01,
      "grad_norm": 19.5766042127858,
      "learning_rate": 1.999984026369704e-06,
      "loss": 1.4966,
      "step": 1075
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.483294588888528,
      "learning_rate": 1.9999839963582724e-06,
      "loss": 1.4772,
      "step": 1076
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.999652845868447,
      "learning_rate": 1.9999839663186754e-06,
      "loss": 1.496,
      "step": 1077
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.272781233458372,
      "learning_rate": 1.999983936250912e-06,
      "loss": 1.4132,
      "step": 1078
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.980996814545554,
      "learning_rate": 1.9999839061549822e-06,
      "loss": 1.2901,
      "step": 1079
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.953763987949976,
      "learning_rate": 1.999983876030886e-06,
      "loss": 1.5155,
      "step": 1080
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.403058384399181,
      "learning_rate": 1.9999838458786244e-06,
      "loss": 1.5269,
      "step": 1081
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.962939986871904,
      "learning_rate": 1.999983815698196e-06,
      "loss": 1.5753,
      "step": 1082
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.283384276439903,
      "learning_rate": 1.999983785489601e-06,
      "loss": 1.5311,
      "step": 1083
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.614239854059685,
      "learning_rate": 1.9999837552528406e-06,
      "loss": 1.5236,
      "step": 1084
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.5938014662143,
      "learning_rate": 1.9999837249879133e-06,
      "loss": 1.449,
      "step": 1085
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.828053203793281,
      "learning_rate": 1.99998369469482e-06,
      "loss": 1.3608,
      "step": 1086
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.86723210959235,
      "learning_rate": 1.999983664373561e-06,
      "loss": 1.3185,
      "step": 1087
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.15651657506136,
      "learning_rate": 1.999983634024135e-06,
      "loss": 1.3037,
      "step": 1088
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.781841302352801,
      "learning_rate": 1.999983603646543e-06,
      "loss": 1.5157,
      "step": 1089
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.6250573996377256,
      "learning_rate": 1.999983573240785e-06,
      "loss": 1.4612,
      "step": 1090
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.853270627798325,
      "learning_rate": 1.999983542806861e-06,
      "loss": 1.3915,
      "step": 1091
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.416508032931958,
      "learning_rate": 1.99998351234477e-06,
      "loss": 1.4846,
      "step": 1092
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.154167527363172,
      "learning_rate": 1.9999834818545135e-06,
      "loss": 1.4909,
      "step": 1093
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.015726983035818,
      "learning_rate": 1.9999834513360905e-06,
      "loss": 1.4462,
      "step": 1094
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.859626471859029,
      "learning_rate": 1.9999834207895015e-06,
      "loss": 1.5644,
      "step": 1095
    },
    {
      "epoch": 0.01,
      "eval_loss": 1.6735153198242188,
      "eval_runtime": 4.6256,
      "eval_samples_per_second": 1.946,
      "eval_steps_per_second": 1.081,
      "step": 1095
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.566823707743832,
      "learning_rate": 1.999983390214746e-06,
      "loss": 1.4629,
      "step": 1096
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.735335408472524,
      "learning_rate": 1.9999833596118244e-06,
      "loss": 1.4391,
      "step": 1097
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.790636259147894,
      "learning_rate": 1.999983328980737e-06,
      "loss": 1.4981,
      "step": 1098
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.349178643835508,
      "learning_rate": 1.999983298321483e-06,
      "loss": 1.3753,
      "step": 1099
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.263976566555674,
      "learning_rate": 1.9999832676340625e-06,
      "loss": 1.55,
      "step": 1100
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.657912236522682,
      "learning_rate": 1.9999832369184764e-06,
      "loss": 1.4841,
      "step": 1101
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.867208669497604,
      "learning_rate": 1.999983206174724e-06,
      "loss": 1.4229,
      "step": 1102
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.862064804403664,
      "learning_rate": 1.9999831754028052e-06,
      "loss": 1.4944,
      "step": 1103
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.3872647723902585,
      "learning_rate": 1.99998314460272e-06,
      "loss": 1.4938,
      "step": 1104
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.878142365865869,
      "learning_rate": 1.9999831137744693e-06,
      "loss": 1.4478,
      "step": 1105
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.632417620908414,
      "learning_rate": 1.9999830829180517e-06,
      "loss": 1.4551,
      "step": 1106
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.666710624917455,
      "learning_rate": 1.999983052033468e-06,
      "loss": 1.4641,
      "step": 1107
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.732751527383862,
      "learning_rate": 1.9999830211207183e-06,
      "loss": 1.3838,
      "step": 1108
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.192163188860683,
      "learning_rate": 1.9999829901798025e-06,
      "loss": 1.5361,
      "step": 1109
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.029036683181163,
      "learning_rate": 1.9999829592107202e-06,
      "loss": 1.5428,
      "step": 1110
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.235770433822521,
      "learning_rate": 1.999982928213472e-06,
      "loss": 1.4727,
      "step": 1111
    },
    {
      "epoch": 0.01,
      "grad_norm": 17.301123468125486,
      "learning_rate": 1.9999828971880574e-06,
      "loss": 1.3757,
      "step": 1112
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.804238410996875,
      "learning_rate": 1.9999828661344764e-06,
      "loss": 1.3589,
      "step": 1113
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.616566753097097,
      "learning_rate": 1.9999828350527295e-06,
      "loss": 1.3691,
      "step": 1114
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.347116928777432,
      "learning_rate": 1.999982803942817e-06,
      "loss": 1.4647,
      "step": 1115
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.932531917778219,
      "learning_rate": 1.9999827728047373e-06,
      "loss": 1.4675,
      "step": 1116
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.660294529470657,
      "learning_rate": 1.999982741638492e-06,
      "loss": 1.4151,
      "step": 1117
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.620813961955892,
      "learning_rate": 1.99998271044408e-06,
      "loss": 1.6348,
      "step": 1118
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.446077565129274,
      "learning_rate": 1.9999826792215024e-06,
      "loss": 1.3873,
      "step": 1119
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.0629846402349195,
      "learning_rate": 1.9999826479707584e-06,
      "loss": 1.4435,
      "step": 1120
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.390133111401672,
      "learning_rate": 1.999982616691848e-06,
      "loss": 1.5374,
      "step": 1121
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.635104269305596,
      "learning_rate": 1.9999825853847717e-06,
      "loss": 1.4144,
      "step": 1122
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.932450441815705,
      "learning_rate": 1.999982554049529e-06,
      "loss": 1.6113,
      "step": 1123
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.571391891954345,
      "learning_rate": 1.9999825226861202e-06,
      "loss": 1.385,
      "step": 1124
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.463278206524636,
      "learning_rate": 1.999982491294545e-06,
      "loss": 1.3112,
      "step": 1125
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.169524463711344,
      "learning_rate": 1.999982459874804e-06,
      "loss": 1.3639,
      "step": 1126
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.706816189602849,
      "learning_rate": 1.999982428426897e-06,
      "loss": 1.4906,
      "step": 1127
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.4003758033814435,
      "learning_rate": 1.9999823969508233e-06,
      "loss": 1.2288,
      "step": 1128
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.07733560063141,
      "learning_rate": 1.999982365446584e-06,
      "loss": 1.4117,
      "step": 1129
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.969998640669905,
      "learning_rate": 1.9999823339141778e-06,
      "loss": 1.3877,
      "step": 1130
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.142101849234914,
      "learning_rate": 1.9999823023536056e-06,
      "loss": 1.4276,
      "step": 1131
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.223738858689869,
      "learning_rate": 1.9999822707648675e-06,
      "loss": 1.4385,
      "step": 1132
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.955334460812677,
      "learning_rate": 1.999982239147963e-06,
      "loss": 1.4981,
      "step": 1133
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.707569119145383,
      "learning_rate": 1.9999822075028926e-06,
      "loss": 1.5537,
      "step": 1134
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.259173620055015,
      "learning_rate": 1.999982175829656e-06,
      "loss": 1.4162,
      "step": 1135
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.160930511739169,
      "learning_rate": 1.999982144128253e-06,
      "loss": 1.6684,
      "step": 1136
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.436420759166786,
      "learning_rate": 1.999982112398684e-06,
      "loss": 1.2647,
      "step": 1137
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.264705476269943,
      "learning_rate": 1.9999820806409487e-06,
      "loss": 1.5094,
      "step": 1138
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.82448740492418,
      "learning_rate": 1.9999820488550476e-06,
      "loss": 1.4738,
      "step": 1139
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.590049555555406,
      "learning_rate": 1.9999820170409797e-06,
      "loss": 1.4759,
      "step": 1140
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.864218721440189,
      "learning_rate": 1.9999819851987462e-06,
      "loss": 1.5362,
      "step": 1141
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.32497769317765,
      "learning_rate": 1.9999819533283464e-06,
      "loss": 1.4001,
      "step": 1142
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.788543241974733,
      "learning_rate": 1.99998192142978e-06,
      "loss": 1.5497,
      "step": 1143
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.851113964577435,
      "learning_rate": 1.999981889503048e-06,
      "loss": 1.4904,
      "step": 1144
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.107648195516455,
      "learning_rate": 1.9999818575481494e-06,
      "loss": 1.5098,
      "step": 1145
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.583169283696968,
      "learning_rate": 1.999981825565085e-06,
      "loss": 1.3993,
      "step": 1146
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.807776598248037,
      "learning_rate": 1.999981793553854e-06,
      "loss": 1.3587,
      "step": 1147
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.614047081189212,
      "learning_rate": 1.999981761514457e-06,
      "loss": 1.3482,
      "step": 1148
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.35657561717519,
      "learning_rate": 1.999981729446894e-06,
      "loss": 1.5398,
      "step": 1149
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.933553997697057,
      "learning_rate": 1.999981697351165e-06,
      "loss": 1.3589,
      "step": 1150
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.872175454134656,
      "learning_rate": 1.9999816652272697e-06,
      "loss": 1.4732,
      "step": 1151
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.288241183644056,
      "learning_rate": 1.999981633075208e-06,
      "loss": 1.3574,
      "step": 1152
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.009781459825191,
      "learning_rate": 1.99998160089498e-06,
      "loss": 1.394,
      "step": 1153
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.55801046260358,
      "learning_rate": 1.9999815686865867e-06,
      "loss": 1.3084,
      "step": 1154
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.821680285232777,
      "learning_rate": 1.9999815364500263e-06,
      "loss": 1.4516,
      "step": 1155
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.952975805784374,
      "learning_rate": 1.9999815041853005e-06,
      "loss": 1.4385,
      "step": 1156
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.576070368663535,
      "learning_rate": 1.9999814718924082e-06,
      "loss": 1.3069,
      "step": 1157
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.951804621604359,
      "learning_rate": 1.9999814395713496e-06,
      "loss": 1.4112,
      "step": 1158
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.741717751755461,
      "learning_rate": 1.9999814072221254e-06,
      "loss": 1.4899,
      "step": 1159
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.893581963366542,
      "learning_rate": 1.9999813748447344e-06,
      "loss": 1.3654,
      "step": 1160
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.688325242766382,
      "learning_rate": 1.9999813424391775e-06,
      "loss": 1.4792,
      "step": 1161
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.958921548979553,
      "learning_rate": 1.9999813100054546e-06,
      "loss": 1.1737,
      "step": 1162
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.988069853619337,
      "learning_rate": 1.9999812775435653e-06,
      "loss": 1.5248,
      "step": 1163
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.323386669021145,
      "learning_rate": 1.99998124505351e-06,
      "loss": 1.3352,
      "step": 1164
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.863439791281708,
      "learning_rate": 1.9999812125352883e-06,
      "loss": 1.4522,
      "step": 1165
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.50022212533205,
      "learning_rate": 1.9999811799889007e-06,
      "loss": 1.4328,
      "step": 1166
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.283635450013209,
      "learning_rate": 1.9999811474143467e-06,
      "loss": 1.4352,
      "step": 1167
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.613546626420899,
      "learning_rate": 1.999981114811627e-06,
      "loss": 1.4733,
      "step": 1168
    },
    {
      "epoch": 0.01,
      "eval_loss": 1.6666209697723389,
      "eval_runtime": 4.627,
      "eval_samples_per_second": 1.945,
      "eval_steps_per_second": 1.081,
      "step": 1168
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.243713431888752,
      "learning_rate": 1.999981082180741e-06,
      "loss": 1.4229,
      "step": 1169
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.180916470723546,
      "learning_rate": 1.999981049521689e-06,
      "loss": 1.5532,
      "step": 1170
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.634457636781247,
      "learning_rate": 1.9999810168344707e-06,
      "loss": 1.4628,
      "step": 1171
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.841975621960265,
      "learning_rate": 1.999980984119086e-06,
      "loss": 1.4714,
      "step": 1172
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.856407984159355,
      "learning_rate": 1.9999809513755354e-06,
      "loss": 1.3953,
      "step": 1173
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.924593632584216,
      "learning_rate": 1.9999809186038184e-06,
      "loss": 1.4777,
      "step": 1174
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.804521073545573,
      "learning_rate": 1.999980885803936e-06,
      "loss": 1.5477,
      "step": 1175
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.849633875980324,
      "learning_rate": 1.9999808529758865e-06,
      "loss": 1.3616,
      "step": 1176
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.918561776125233,
      "learning_rate": 1.9999808201196716e-06,
      "loss": 1.4449,
      "step": 1177
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.559134203442522,
      "learning_rate": 1.9999807872352903e-06,
      "loss": 1.373,
      "step": 1178
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.7385730138344675,
      "learning_rate": 1.9999807543227426e-06,
      "loss": 1.4561,
      "step": 1179
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.477240093311455,
      "learning_rate": 1.9999807213820294e-06,
      "loss": 1.3705,
      "step": 1180
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.069434481225469,
      "learning_rate": 1.99998068841315e-06,
      "loss": 1.2873,
      "step": 1181
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.697371499630333,
      "learning_rate": 1.999980655416104e-06,
      "loss": 1.4569,
      "step": 1182
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.533455071498911,
      "learning_rate": 1.999980622390892e-06,
      "loss": 1.4891,
      "step": 1183
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.807849990271847,
      "learning_rate": 1.9999805893375135e-06,
      "loss": 1.4576,
      "step": 1184
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.812651120417968,
      "learning_rate": 1.9999805562559696e-06,
      "loss": 1.4451,
      "step": 1185
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.764537937046826,
      "learning_rate": 1.9999805231462594e-06,
      "loss": 1.4257,
      "step": 1186
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.106952053198868,
      "learning_rate": 1.9999804900083827e-06,
      "loss": 1.5414,
      "step": 1187
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.655479441251472,
      "learning_rate": 1.99998045684234e-06,
      "loss": 1.4561,
      "step": 1188
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.473182229797437,
      "learning_rate": 1.9999804236481316e-06,
      "loss": 1.4701,
      "step": 1189
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.065099661197906,
      "learning_rate": 1.9999803904257566e-06,
      "loss": 1.4211,
      "step": 1190
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.538986656004763,
      "learning_rate": 1.9999803571752157e-06,
      "loss": 1.5027,
      "step": 1191
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.4128695506388365,
      "learning_rate": 1.9999803238965084e-06,
      "loss": 1.5027,
      "step": 1192
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.032103944177775,
      "learning_rate": 1.999980290589635e-06,
      "loss": 1.5644,
      "step": 1193
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.682343321057221,
      "learning_rate": 1.9999802572545963e-06,
      "loss": 1.5309,
      "step": 1194
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.343314622774007,
      "learning_rate": 1.9999802238913906e-06,
      "loss": 1.3003,
      "step": 1195
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.630952351276132,
      "learning_rate": 1.999980190500019e-06,
      "loss": 1.4954,
      "step": 1196
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.198619248496427,
      "learning_rate": 1.9999801570804815e-06,
      "loss": 1.397,
      "step": 1197
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.618621734517426,
      "learning_rate": 1.9999801236327776e-06,
      "loss": 1.3981,
      "step": 1198
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.147623660905904,
      "learning_rate": 1.999980090156908e-06,
      "loss": 1.3596,
      "step": 1199
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.946475818289631,
      "learning_rate": 1.999980056652872e-06,
      "loss": 1.5295,
      "step": 1200
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.702884825661346,
      "learning_rate": 1.9999800231206696e-06,
      "loss": 1.4685,
      "step": 1201
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.029321508506549,
      "learning_rate": 1.9999799895603013e-06,
      "loss": 1.2344,
      "step": 1202
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.130362694352546,
      "learning_rate": 1.999979955971767e-06,
      "loss": 1.4078,
      "step": 1203
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.638399765026865,
      "learning_rate": 1.9999799223550666e-06,
      "loss": 1.2877,
      "step": 1204
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.419197466434014,
      "learning_rate": 1.9999798887102e-06,
      "loss": 1.5245,
      "step": 1205
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.048810252708306,
      "learning_rate": 1.999979855037167e-06,
      "loss": 1.6089,
      "step": 1206
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.594986736232871,
      "learning_rate": 1.9999798213359687e-06,
      "loss": 1.4686,
      "step": 1207
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.341694734185055,
      "learning_rate": 1.999979787606604e-06,
      "loss": 1.2083,
      "step": 1208
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.721844492494166,
      "learning_rate": 1.999979753849073e-06,
      "loss": 1.3191,
      "step": 1209
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.660083437293136,
      "learning_rate": 1.9999797200633755e-06,
      "loss": 1.405,
      "step": 1210
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.550917622485102,
      "learning_rate": 1.9999796862495127e-06,
      "loss": 1.392,
      "step": 1211
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.855465281579127,
      "learning_rate": 1.999979652407483e-06,
      "loss": 1.2981,
      "step": 1212
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.7605359185049645,
      "learning_rate": 1.999979618537288e-06,
      "loss": 1.475,
      "step": 1213
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.992826613342054,
      "learning_rate": 1.9999795846389267e-06,
      "loss": 1.309,
      "step": 1214
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.751213987512799,
      "learning_rate": 1.999979550712399e-06,
      "loss": 1.3181,
      "step": 1215
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.56222348172062,
      "learning_rate": 1.999979516757705e-06,
      "loss": 1.4476,
      "step": 1216
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.211498865140055,
      "learning_rate": 1.9999794827748456e-06,
      "loss": 1.6613,
      "step": 1217
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.76843858033082,
      "learning_rate": 1.9999794487638195e-06,
      "loss": 1.455,
      "step": 1218
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.971537042597962,
      "learning_rate": 1.9999794147246278e-06,
      "loss": 1.4828,
      "step": 1219
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.740101198670973,
      "learning_rate": 1.9999793806572697e-06,
      "loss": 1.4563,
      "step": 1220
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.772585640252896,
      "learning_rate": 1.9999793465617452e-06,
      "loss": 1.4642,
      "step": 1221
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.057122514777892,
      "learning_rate": 1.999979312438055e-06,
      "loss": 1.4538,
      "step": 1222
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.39056314627121,
      "learning_rate": 1.999979278286199e-06,
      "loss": 1.547,
      "step": 1223
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.991611184305103,
      "learning_rate": 1.9999792441061764e-06,
      "loss": 1.4423,
      "step": 1224
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.647894296933476,
      "learning_rate": 1.999979209897988e-06,
      "loss": 1.4384,
      "step": 1225
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.99949285749195,
      "learning_rate": 1.9999791756616334e-06,
      "loss": 1.5835,
      "step": 1226
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.567738724749353,
      "learning_rate": 1.9999791413971127e-06,
      "loss": 1.3663,
      "step": 1227
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.394852681906166,
      "learning_rate": 1.9999791071044256e-06,
      "loss": 1.3405,
      "step": 1228
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.3926441113626185,
      "learning_rate": 1.999979072783573e-06,
      "loss": 1.4969,
      "step": 1229
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.406148666652971,
      "learning_rate": 1.999979038434554e-06,
      "loss": 1.5782,
      "step": 1230
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.894426333285634,
      "learning_rate": 1.999979004057369e-06,
      "loss": 1.4171,
      "step": 1231
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.662885913776113,
      "learning_rate": 1.9999789696520178e-06,
      "loss": 1.3504,
      "step": 1232
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.859148631113401,
      "learning_rate": 1.9999789352185005e-06,
      "loss": 1.3593,
      "step": 1233
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.501735921525826,
      "learning_rate": 1.999978900756817e-06,
      "loss": 1.497,
      "step": 1234
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.767981300489952,
      "learning_rate": 1.999978866266968e-06,
      "loss": 1.4222,
      "step": 1235
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.601659778166056,
      "learning_rate": 1.9999788317489523e-06,
      "loss": 1.3368,
      "step": 1236
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.981059658355397,
      "learning_rate": 1.999978797202771e-06,
      "loss": 1.3621,
      "step": 1237
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.128261931608396,
      "learning_rate": 1.9999787626284236e-06,
      "loss": 1.6156,
      "step": 1238
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.881645282270224,
      "learning_rate": 1.9999787280259097e-06,
      "loss": 1.3842,
      "step": 1239
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.893724788852448,
      "learning_rate": 1.9999786933952302e-06,
      "loss": 1.6442,
      "step": 1240
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.222605242350468,
      "learning_rate": 1.9999786587363844e-06,
      "loss": 1.373,
      "step": 1241
    },
    {
      "epoch": 0.01,
      "eval_loss": 1.6639937162399292,
      "eval_runtime": 4.616,
      "eval_samples_per_second": 1.95,
      "eval_steps_per_second": 1.083,
      "step": 1241
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.66908931719967,
      "learning_rate": 1.999978624049372e-06,
      "loss": 1.4993,
      "step": 1242
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.9307281292109035,
      "learning_rate": 1.9999785893341944e-06,
      "loss": 1.5608,
      "step": 1243
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.826571091140564,
      "learning_rate": 1.99997855459085e-06,
      "loss": 1.3162,
      "step": 1244
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8493801546756465,
      "learning_rate": 1.9999785198193405e-06,
      "loss": 1.4282,
      "step": 1245
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.697208964299474,
      "learning_rate": 1.999978485019664e-06,
      "loss": 1.5024,
      "step": 1246
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.919002989067272,
      "learning_rate": 1.999978450191822e-06,
      "loss": 1.5431,
      "step": 1247
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.99900985273033,
      "learning_rate": 1.9999784153358135e-06,
      "loss": 1.4855,
      "step": 1248
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.617964526314292,
      "learning_rate": 1.999978380451639e-06,
      "loss": 1.5545,
      "step": 1249
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.192495194044367,
      "learning_rate": 1.9999783455392987e-06,
      "loss": 1.4625,
      "step": 1250
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.017628302905397,
      "learning_rate": 1.9999783105987924e-06,
      "loss": 1.5347,
      "step": 1251
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.576487300446409,
      "learning_rate": 1.9999782756301197e-06,
      "loss": 1.5079,
      "step": 1252
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.5622763050535085,
      "learning_rate": 1.999978240633281e-06,
      "loss": 1.4778,
      "step": 1253
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.320462635223293,
      "learning_rate": 1.9999782056082763e-06,
      "loss": 1.1527,
      "step": 1254
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.753621834652253,
      "learning_rate": 1.9999781705551057e-06,
      "loss": 1.4019,
      "step": 1255
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.399262354558424,
      "learning_rate": 1.9999781354737687e-06,
      "loss": 1.4149,
      "step": 1256
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.602639150018978,
      "learning_rate": 1.9999781003642658e-06,
      "loss": 1.3364,
      "step": 1257
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.589898719773274,
      "learning_rate": 1.999978065226597e-06,
      "loss": 1.4087,
      "step": 1258
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.536362353334885,
      "learning_rate": 1.999978030060762e-06,
      "loss": 1.4561,
      "step": 1259
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.565066571845603,
      "learning_rate": 1.999977994866761e-06,
      "loss": 1.4477,
      "step": 1260
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.82724395797132,
      "learning_rate": 1.999977959644594e-06,
      "loss": 1.385,
      "step": 1261
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.485711386546417,
      "learning_rate": 1.999977924394261e-06,
      "loss": 1.3381,
      "step": 1262
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.007143806674197,
      "learning_rate": 1.9999778891157615e-06,
      "loss": 1.4557,
      "step": 1263
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.361949819262946,
      "learning_rate": 1.9999778538090964e-06,
      "loss": 1.3389,
      "step": 1264
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.503528627850973,
      "learning_rate": 1.999977818474265e-06,
      "loss": 1.4698,
      "step": 1265
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.233324651443236,
      "learning_rate": 1.999977783111268e-06,
      "loss": 1.3247,
      "step": 1266
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.5046307792398865,
      "learning_rate": 1.9999777477201044e-06,
      "loss": 1.3497,
      "step": 1267
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.645109391595665,
      "learning_rate": 1.999977712300775e-06,
      "loss": 1.3201,
      "step": 1268
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.1446712059940465,
      "learning_rate": 1.9999776768532796e-06,
      "loss": 1.4262,
      "step": 1269
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.28587028516593,
      "learning_rate": 1.999977641377618e-06,
      "loss": 1.6932,
      "step": 1270
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.598379302979389,
      "learning_rate": 1.9999776058737906e-06,
      "loss": 1.4316,
      "step": 1271
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.016449286637428,
      "learning_rate": 1.9999775703417965e-06,
      "loss": 1.5183,
      "step": 1272
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.415644315937934,
      "learning_rate": 1.999977534781637e-06,
      "loss": 1.3129,
      "step": 1273
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.704232037612598,
      "learning_rate": 1.9999774991933113e-06,
      "loss": 1.4044,
      "step": 1274
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.6149493250979,
      "learning_rate": 1.9999774635768197e-06,
      "loss": 1.3827,
      "step": 1275
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.530098577109759,
      "learning_rate": 1.9999774279321617e-06,
      "loss": 1.6327,
      "step": 1276
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.743687061339907,
      "learning_rate": 1.9999773922593383e-06,
      "loss": 1.5278,
      "step": 1277
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8692755474083205,
      "learning_rate": 1.9999773565583484e-06,
      "loss": 1.5159,
      "step": 1278
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.0274537662792556,
      "learning_rate": 1.9999773208291925e-06,
      "loss": 1.6578,
      "step": 1279
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.493325452863065,
      "learning_rate": 1.9999772850718703e-06,
      "loss": 1.3068,
      "step": 1280
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.623446755309784,
      "learning_rate": 1.9999772492863825e-06,
      "loss": 1.3591,
      "step": 1281
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.58471292058994,
      "learning_rate": 1.999977213472729e-06,
      "loss": 1.2837,
      "step": 1282
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.718580482751096,
      "learning_rate": 1.9999771776309083e-06,
      "loss": 1.5876,
      "step": 1283
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.699986818479373,
      "learning_rate": 1.999977141760922e-06,
      "loss": 1.3901,
      "step": 1284
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.60582750403323,
      "learning_rate": 1.99997710586277e-06,
      "loss": 1.4277,
      "step": 1285
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.673408978804247,
      "learning_rate": 1.9999770699364526e-06,
      "loss": 1.4454,
      "step": 1286
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.9768592067046695,
      "learning_rate": 1.999977033981968e-06,
      "loss": 1.3688,
      "step": 1287
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.49039104639861,
      "learning_rate": 1.999976997999318e-06,
      "loss": 1.3585,
      "step": 1288
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.441000592991114,
      "learning_rate": 1.999976961988502e-06,
      "loss": 1.4013,
      "step": 1289
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.248224180467002,
      "learning_rate": 1.9999769259495197e-06,
      "loss": 1.6231,
      "step": 1290
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.967205396228199,
      "learning_rate": 1.9999768898823714e-06,
      "loss": 1.3776,
      "step": 1291
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.786827176028182,
      "learning_rate": 1.999976853787057e-06,
      "loss": 1.5409,
      "step": 1292
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.629480539543215,
      "learning_rate": 1.999976817663577e-06,
      "loss": 1.4354,
      "step": 1293
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.100068802921767,
      "learning_rate": 1.9999767815119305e-06,
      "loss": 1.4632,
      "step": 1294
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.134078229009721,
      "learning_rate": 1.999976745332118e-06,
      "loss": 1.4747,
      "step": 1295
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.039774129292199,
      "learning_rate": 1.99997670912414e-06,
      "loss": 1.5499,
      "step": 1296
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.581809616341501,
      "learning_rate": 1.9999766728879955e-06,
      "loss": 1.5581,
      "step": 1297
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.769459899834711,
      "learning_rate": 1.999976636623685e-06,
      "loss": 1.6019,
      "step": 1298
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.789913009365324,
      "learning_rate": 1.9999766003312087e-06,
      "loss": 1.4384,
      "step": 1299
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.948273546122202,
      "learning_rate": 1.9999765640105663e-06,
      "loss": 1.4859,
      "step": 1300
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.046098251037359,
      "learning_rate": 1.999976527661758e-06,
      "loss": 1.3215,
      "step": 1301
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.759646769803917,
      "learning_rate": 1.9999764912847834e-06,
      "loss": 1.26,
      "step": 1302
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.049312397622781,
      "learning_rate": 1.999976454879643e-06,
      "loss": 1.2966,
      "step": 1303
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.7730400364294985,
      "learning_rate": 1.9999764184463365e-06,
      "loss": 1.5075,
      "step": 1304
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.615090990630283,
      "learning_rate": 1.999976381984864e-06,
      "loss": 1.3308,
      "step": 1305
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.050723699922145,
      "learning_rate": 1.999976345495226e-06,
      "loss": 1.5391,
      "step": 1306
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.2292497493973995,
      "learning_rate": 1.999976308977421e-06,
      "loss": 1.4581,
      "step": 1307
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.484046582647355,
      "learning_rate": 1.9999762724314504e-06,
      "loss": 1.2275,
      "step": 1308
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.7418804258275,
      "learning_rate": 1.9999762358573144e-06,
      "loss": 1.4074,
      "step": 1309
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.656737686412653,
      "learning_rate": 1.9999761992550116e-06,
      "loss": 1.4102,
      "step": 1310
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.913690315220175,
      "learning_rate": 1.9999761626245433e-06,
      "loss": 1.5153,
      "step": 1311
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.631259670029744,
      "learning_rate": 1.9999761259659085e-06,
      "loss": 1.4168,
      "step": 1312
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.352234843978522,
      "learning_rate": 1.9999760892791083e-06,
      "loss": 1.7813,
      "step": 1313
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.806086460573851,
      "learning_rate": 1.9999760525641416e-06,
      "loss": 1.4341,
      "step": 1314
    },
    {
      "epoch": 0.01,
      "eval_loss": 1.664663314819336,
      "eval_runtime": 4.6478,
      "eval_samples_per_second": 1.936,
      "eval_steps_per_second": 1.076,
      "step": 1314
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.865234880117637,
      "learning_rate": 1.999976015821009e-06,
      "loss": 1.5092,
      "step": 1315
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.593356313971813,
      "learning_rate": 1.999975979049711e-06,
      "loss": 1.2732,
      "step": 1316
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.765904676157058,
      "learning_rate": 1.9999759422502462e-06,
      "loss": 1.4894,
      "step": 1317
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.869472003176806,
      "learning_rate": 1.9999759054226157e-06,
      "loss": 1.3162,
      "step": 1318
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.707074378059004,
      "learning_rate": 1.9999758685668192e-06,
      "loss": 1.3817,
      "step": 1319
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.360722245763356,
      "learning_rate": 1.999975831682857e-06,
      "loss": 1.3389,
      "step": 1320
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.432944358728482,
      "learning_rate": 1.9999757947707284e-06,
      "loss": 1.4327,
      "step": 1321
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.82335037337341,
      "learning_rate": 1.9999757578304336e-06,
      "loss": 1.6028,
      "step": 1322
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.585722041095415,
      "learning_rate": 1.9999757208619733e-06,
      "loss": 1.4382,
      "step": 1323
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.668470608848745,
      "learning_rate": 1.999975683865347e-06,
      "loss": 1.2558,
      "step": 1324
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.731181861451524,
      "learning_rate": 1.9999756468405543e-06,
      "loss": 1.4787,
      "step": 1325
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.300289798255382,
      "learning_rate": 1.999975609787596e-06,
      "loss": 1.4782,
      "step": 1326
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.942011068638616,
      "learning_rate": 1.9999755727064715e-06,
      "loss": 1.5667,
      "step": 1327
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.309328004860009,
      "learning_rate": 1.9999755355971813e-06,
      "loss": 1.5442,
      "step": 1328
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.915409373782858,
      "learning_rate": 1.9999754984597248e-06,
      "loss": 1.3834,
      "step": 1329
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.89211942651002,
      "learning_rate": 1.9999754612941023e-06,
      "loss": 1.3551,
      "step": 1330
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.687686388314984,
      "learning_rate": 1.999975424100314e-06,
      "loss": 1.2831,
      "step": 1331
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.582159025430598,
      "learning_rate": 1.9999753868783594e-06,
      "loss": 1.4689,
      "step": 1332
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8213772730034865,
      "learning_rate": 1.9999753496282394e-06,
      "loss": 1.4182,
      "step": 1333
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.5280013653649105,
      "learning_rate": 1.999975312349953e-06,
      "loss": 1.4439,
      "step": 1334
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.893199334359817,
      "learning_rate": 1.9999752750435008e-06,
      "loss": 1.5473,
      "step": 1335
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.869413876075934,
      "learning_rate": 1.9999752377088825e-06,
      "loss": 1.3372,
      "step": 1336
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.9958240347842695,
      "learning_rate": 1.9999752003460982e-06,
      "loss": 1.3613,
      "step": 1337
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.4921822785420655,
      "learning_rate": 1.999975162955148e-06,
      "loss": 1.4645,
      "step": 1338
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.304921830032534,
      "learning_rate": 1.999975125536032e-06,
      "loss": 1.4629,
      "step": 1339
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.649096231830719,
      "learning_rate": 1.9999750880887497e-06,
      "loss": 1.2108,
      "step": 1340
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.420599578392317,
      "learning_rate": 1.9999750506133017e-06,
      "loss": 1.3255,
      "step": 1341
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8401246330468695,
      "learning_rate": 1.9999750131096876e-06,
      "loss": 1.4344,
      "step": 1342
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.15766506468985,
      "learning_rate": 1.9999749755779076e-06,
      "loss": 1.5095,
      "step": 1343
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.838423380109326,
      "learning_rate": 1.999974938017961e-06,
      "loss": 1.4351,
      "step": 1344
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.7944677876331205,
      "learning_rate": 1.9999749004298492e-06,
      "loss": 1.4865,
      "step": 1345
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.857475428823912,
      "learning_rate": 1.9999748628135713e-06,
      "loss": 1.5694,
      "step": 1346
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.117298413799976,
      "learning_rate": 1.9999748251691275e-06,
      "loss": 1.3554,
      "step": 1347
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.729812229132196,
      "learning_rate": 1.9999747874965176e-06,
      "loss": 1.365,
      "step": 1348
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.918522074472892,
      "learning_rate": 1.999974749795742e-06,
      "loss": 1.5329,
      "step": 1349
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.128757606688177,
      "learning_rate": 1.9999747120668e-06,
      "loss": 1.4496,
      "step": 1350
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.512588294782333,
      "learning_rate": 1.9999746743096923e-06,
      "loss": 1.4654,
      "step": 1351
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.932892066430468,
      "learning_rate": 1.9999746365244182e-06,
      "loss": 1.5322,
      "step": 1352
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.9351926580424585,
      "learning_rate": 1.999974598710979e-06,
      "loss": 1.509,
      "step": 1353
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.73360746783656,
      "learning_rate": 1.999974560869373e-06,
      "loss": 1.4276,
      "step": 1354
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.323019209743392,
      "learning_rate": 1.9999745229996014e-06,
      "loss": 1.4279,
      "step": 1355
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.121493184418707,
      "learning_rate": 1.999974485101664e-06,
      "loss": 1.3617,
      "step": 1356
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.924408494387465,
      "learning_rate": 1.9999744471755604e-06,
      "loss": 1.4852,
      "step": 1357
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.577577155791868,
      "learning_rate": 1.999974409221291e-06,
      "loss": 1.463,
      "step": 1358
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.677557438554256,
      "learning_rate": 1.9999743712388555e-06,
      "loss": 1.4702,
      "step": 1359
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.869471619730696,
      "learning_rate": 1.999974333228254e-06,
      "loss": 1.4982,
      "step": 1360
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.23329639876476,
      "learning_rate": 1.9999742951894867e-06,
      "loss": 1.4118,
      "step": 1361
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.664165574708151,
      "learning_rate": 1.9999742571225534e-06,
      "loss": 1.4708,
      "step": 1362
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.5907041789083705,
      "learning_rate": 1.999974219027454e-06,
      "loss": 1.4604,
      "step": 1363
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.940547135047092,
      "learning_rate": 1.999974180904189e-06,
      "loss": 1.4571,
      "step": 1364
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.234304088486649,
      "learning_rate": 1.9999741427527577e-06,
      "loss": 1.4042,
      "step": 1365
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.075734957191831,
      "learning_rate": 1.9999741045731605e-06,
      "loss": 1.4649,
      "step": 1366
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.545302713138621,
      "learning_rate": 1.999974066365398e-06,
      "loss": 1.3497,
      "step": 1367
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.55221088039147,
      "learning_rate": 1.9999740281294687e-06,
      "loss": 1.342,
      "step": 1368
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.6125447731949185,
      "learning_rate": 1.9999739898653736e-06,
      "loss": 1.4971,
      "step": 1369
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.07610846685092,
      "learning_rate": 1.999973951573113e-06,
      "loss": 1.5226,
      "step": 1370
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.318700322087232,
      "learning_rate": 1.999973913252686e-06,
      "loss": 1.4651,
      "step": 1371
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.537300685685728,
      "learning_rate": 1.999973874904093e-06,
      "loss": 1.2609,
      "step": 1372
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.193376063176117,
      "learning_rate": 1.9999738365273346e-06,
      "loss": 1.5545,
      "step": 1373
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.273750459431588,
      "learning_rate": 1.99997379812241e-06,
      "loss": 1.4439,
      "step": 1374
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.595241626337541,
      "learning_rate": 1.9999737596893198e-06,
      "loss": 1.4768,
      "step": 1375
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.259906353516755,
      "learning_rate": 1.9999737212280634e-06,
      "loss": 1.5442,
      "step": 1376
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.201372009358403,
      "learning_rate": 1.999973682738641e-06,
      "loss": 1.368,
      "step": 1377
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.977174887944729,
      "learning_rate": 1.9999736442210523e-06,
      "loss": 1.469,
      "step": 1378
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.782568509322288,
      "learning_rate": 1.999973605675298e-06,
      "loss": 1.4649,
      "step": 1379
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.9003324240880515,
      "learning_rate": 1.999973567101378e-06,
      "loss": 1.4912,
      "step": 1380
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.443690275403413,
      "learning_rate": 1.9999735284992916e-06,
      "loss": 1.3528,
      "step": 1381
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.103063174767599,
      "learning_rate": 1.9999734898690395e-06,
      "loss": 1.5669,
      "step": 1382
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.799160062093434,
      "learning_rate": 1.999973451210622e-06,
      "loss": 1.4682,
      "step": 1383
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.89475639433693,
      "learning_rate": 1.9999734125240377e-06,
      "loss": 1.5212,
      "step": 1384
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.458491045381737,
      "learning_rate": 1.999973373809288e-06,
      "loss": 1.2801,
      "step": 1385
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.308111135518119,
      "learning_rate": 1.999973335066372e-06,
      "loss": 1.4821,
      "step": 1386
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.548350666808942,
      "learning_rate": 1.9999732962952905e-06,
      "loss": 1.4494,
      "step": 1387
    },
    {
      "epoch": 0.01,
      "eval_loss": 1.6605433225631714,
      "eval_runtime": 4.645,
      "eval_samples_per_second": 1.938,
      "eval_steps_per_second": 1.076,
      "step": 1387
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.872935272294134,
      "learning_rate": 1.999973257496043e-06,
      "loss": 1.5121,
      "step": 1388
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.421100493500692,
      "learning_rate": 1.999973218668629e-06,
      "loss": 1.4706,
      "step": 1389
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.651724509052579,
      "learning_rate": 1.99997317981305e-06,
      "loss": 1.3924,
      "step": 1390
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.375754147106289,
      "learning_rate": 1.9999731409293047e-06,
      "loss": 1.6767,
      "step": 1391
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.849899617039911,
      "learning_rate": 1.9999731020173934e-06,
      "loss": 1.4921,
      "step": 1392
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.47624020449905,
      "learning_rate": 1.999973063077316e-06,
      "loss": 1.4762,
      "step": 1393
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.357532784726696,
      "learning_rate": 1.999973024109073e-06,
      "loss": 1.3477,
      "step": 1394
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.627544754203488,
      "learning_rate": 1.999972985112664e-06,
      "loss": 1.4234,
      "step": 1395
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.782545342694373,
      "learning_rate": 1.999972946088089e-06,
      "loss": 1.3345,
      "step": 1396
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.918312434237096,
      "learning_rate": 1.999972907035348e-06,
      "loss": 1.4585,
      "step": 1397
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.74711940647535,
      "learning_rate": 1.9999728679544412e-06,
      "loss": 1.5588,
      "step": 1398
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.756574444935555,
      "learning_rate": 1.999972828845369e-06,
      "loss": 1.5253,
      "step": 1399
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.825155321890688,
      "learning_rate": 1.99997278970813e-06,
      "loss": 1.4386,
      "step": 1400
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.893816643811335,
      "learning_rate": 1.9999727505427257e-06,
      "loss": 1.2703,
      "step": 1401
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.271403904383769,
      "learning_rate": 1.999972711349155e-06,
      "loss": 1.332,
      "step": 1402
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.094872014502599,
      "learning_rate": 1.999972672127419e-06,
      "loss": 1.4158,
      "step": 1403
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.785102973962074,
      "learning_rate": 1.9999726328775166e-06,
      "loss": 1.4165,
      "step": 1404
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.7983947530400535,
      "learning_rate": 1.9999725935994485e-06,
      "loss": 1.4353,
      "step": 1405
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.646416220267023,
      "learning_rate": 1.9999725542932145e-06,
      "loss": 1.3781,
      "step": 1406
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.037078769476571,
      "learning_rate": 1.9999725149588146e-06,
      "loss": 1.3702,
      "step": 1407
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.9356881553322545,
      "learning_rate": 1.9999724755962486e-06,
      "loss": 1.5146,
      "step": 1408
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.717638454921318,
      "learning_rate": 1.999972436205517e-06,
      "loss": 1.2775,
      "step": 1409
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.998493967980832,
      "learning_rate": 1.9999723967866197e-06,
      "loss": 1.3866,
      "step": 1410
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.7378104129164464,
      "learning_rate": 1.9999723573395564e-06,
      "loss": 1.32,
      "step": 1411
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.599759309706441,
      "learning_rate": 1.9999723178643266e-06,
      "loss": 1.5095,
      "step": 1412
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.876824835044584,
      "learning_rate": 1.9999722783609313e-06,
      "loss": 1.283,
      "step": 1413
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.37874420839293,
      "learning_rate": 1.99997223882937e-06,
      "loss": 1.3672,
      "step": 1414
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.233951321610936,
      "learning_rate": 1.999972199269643e-06,
      "loss": 1.5104,
      "step": 1415
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.6266662891163195,
      "learning_rate": 1.99997215968175e-06,
      "loss": 1.422,
      "step": 1416
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.264895342929483,
      "learning_rate": 1.9999721200656912e-06,
      "loss": 1.3256,
      "step": 1417
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.126522085971787,
      "learning_rate": 1.9999720804214665e-06,
      "loss": 1.5388,
      "step": 1418
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.0904993436779655,
      "learning_rate": 1.999972040749076e-06,
      "loss": 1.6427,
      "step": 1419
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.557223446909417,
      "learning_rate": 1.999972001048519e-06,
      "loss": 1.4148,
      "step": 1420
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.7803697348815914,
      "learning_rate": 1.999971961319797e-06,
      "loss": 1.5555,
      "step": 1421
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.800921782137216,
      "learning_rate": 1.9999719215629085e-06,
      "loss": 1.4876,
      "step": 1422
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.9994358843439,
      "learning_rate": 1.999971881777855e-06,
      "loss": 1.4667,
      "step": 1423
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.464016639284502,
      "learning_rate": 1.9999718419646343e-06,
      "loss": 1.2817,
      "step": 1424
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.791881059083396,
      "learning_rate": 1.9999718021232487e-06,
      "loss": 1.3099,
      "step": 1425
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.046707728618333,
      "learning_rate": 1.9999717622536963e-06,
      "loss": 1.3161,
      "step": 1426
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.941858775320615,
      "learning_rate": 1.9999717223559788e-06,
      "loss": 1.4566,
      "step": 1427
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.008676930939444,
      "learning_rate": 1.9999716824300953e-06,
      "loss": 1.4165,
      "step": 1428
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.280497134099123,
      "learning_rate": 1.999971642476046e-06,
      "loss": 1.4435,
      "step": 1429
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.899113042250123,
      "learning_rate": 1.999971602493831e-06,
      "loss": 1.4571,
      "step": 1430
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.250624631469121,
      "learning_rate": 1.9999715624834494e-06,
      "loss": 1.4758,
      "step": 1431
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.646881476231068,
      "learning_rate": 1.999971522444902e-06,
      "loss": 1.581,
      "step": 1432
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.1511243697760705,
      "learning_rate": 1.999971482378189e-06,
      "loss": 1.3673,
      "step": 1433
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.963495678210403,
      "learning_rate": 1.9999714422833108e-06,
      "loss": 1.4672,
      "step": 1434
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.879620074849547,
      "learning_rate": 1.999971402160266e-06,
      "loss": 1.4164,
      "step": 1435
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.393239464097148,
      "learning_rate": 1.9999713620090556e-06,
      "loss": 1.2775,
      "step": 1436
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.371614688959383,
      "learning_rate": 1.999971321829679e-06,
      "loss": 1.3789,
      "step": 1437
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.859029231950951,
      "learning_rate": 1.9999712816221366e-06,
      "loss": 1.2764,
      "step": 1438
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.978190883538404,
      "learning_rate": 1.9999712413864284e-06,
      "loss": 1.5097,
      "step": 1439
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.859269754595811,
      "learning_rate": 1.9999712011225546e-06,
      "loss": 1.3549,
      "step": 1440
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.598800701532648,
      "learning_rate": 1.9999711608305144e-06,
      "loss": 1.4791,
      "step": 1441
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.233668406367012,
      "learning_rate": 1.9999711205103087e-06,
      "loss": 1.5436,
      "step": 1442
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.585328618747414,
      "learning_rate": 1.9999710801619375e-06,
      "loss": 1.4573,
      "step": 1443
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.975282503115869,
      "learning_rate": 1.9999710397854e-06,
      "loss": 1.3631,
      "step": 1444
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.060428307246658,
      "learning_rate": 1.9999709993806967e-06,
      "loss": 1.4522,
      "step": 1445
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.344016376470663,
      "learning_rate": 1.999970958947827e-06,
      "loss": 1.4702,
      "step": 1446
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.153848988906273,
      "learning_rate": 1.999970918486792e-06,
      "loss": 1.5144,
      "step": 1447
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.957762192580373,
      "learning_rate": 1.9999708779975914e-06,
      "loss": 1.5009,
      "step": 1448
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.791562507924204,
      "learning_rate": 1.999970837480225e-06,
      "loss": 1.3169,
      "step": 1449
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.757399799969445,
      "learning_rate": 1.9999707969346922e-06,
      "loss": 1.4207,
      "step": 1450
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.939646292701733,
      "learning_rate": 1.9999707563609937e-06,
      "loss": 1.5557,
      "step": 1451
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.213045670839742,
      "learning_rate": 1.9999707157591296e-06,
      "loss": 1.4483,
      "step": 1452
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.262064916528809,
      "learning_rate": 1.999970675129099e-06,
      "loss": 1.395,
      "step": 1453
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.936947311373785,
      "learning_rate": 1.999970634470903e-06,
      "loss": 1.4634,
      "step": 1454
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.362006992850526,
      "learning_rate": 1.9999705937845413e-06,
      "loss": 1.2776,
      "step": 1455
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.643820891960016,
      "learning_rate": 1.9999705530700138e-06,
      "loss": 1.2722,
      "step": 1456
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.949308239528696,
      "learning_rate": 1.9999705123273203e-06,
      "loss": 1.5015,
      "step": 1457
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.259335084505092,
      "learning_rate": 1.999970471556461e-06,
      "loss": 1.4663,
      "step": 1458
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.267704451871602,
      "learning_rate": 1.9999704307574355e-06,
      "loss": 1.4835,
      "step": 1459
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.251718898458907,
      "learning_rate": 1.9999703899302446e-06,
      "loss": 1.3878,
      "step": 1460
    },
    {
      "epoch": 0.01,
      "eval_loss": 1.6609848737716675,
      "eval_runtime": 4.6315,
      "eval_samples_per_second": 1.943,
      "eval_steps_per_second": 1.08,
      "step": 1460
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.908979692543608,
      "learning_rate": 1.9999703490748877e-06,
      "loss": 1.4374,
      "step": 1461
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.071932947528256,
      "learning_rate": 1.999970308191365e-06,
      "loss": 1.5086,
      "step": 1462
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.519660989077838,
      "learning_rate": 1.9999702672796765e-06,
      "loss": 1.3081,
      "step": 1463
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.423290223247818,
      "learning_rate": 1.999970226339822e-06,
      "loss": 1.5415,
      "step": 1464
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.276573152012557,
      "learning_rate": 1.999970185371802e-06,
      "loss": 1.3497,
      "step": 1465
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.873791431302587,
      "learning_rate": 1.9999701443756155e-06,
      "loss": 1.3745,
      "step": 1466
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.9749684436238315,
      "learning_rate": 1.9999701033512637e-06,
      "loss": 1.4114,
      "step": 1467
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.646373215103361,
      "learning_rate": 1.999970062298746e-06,
      "loss": 1.4112,
      "step": 1468
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.864454376294363,
      "learning_rate": 1.9999700212180626e-06,
      "loss": 1.3469,
      "step": 1469
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.937829789499327,
      "learning_rate": 1.999969980109213e-06,
      "loss": 1.5815,
      "step": 1470
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.639905252344842,
      "learning_rate": 1.9999699389721977e-06,
      "loss": 1.3537,
      "step": 1471
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.78299140918665,
      "learning_rate": 1.999969897807017e-06,
      "loss": 1.4826,
      "step": 1472
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.793192549086234,
      "learning_rate": 1.9999698566136697e-06,
      "loss": 1.436,
      "step": 1473
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.792423303861806,
      "learning_rate": 1.999969815392157e-06,
      "loss": 1.4389,
      "step": 1474
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.817682445317553,
      "learning_rate": 1.9999697741424787e-06,
      "loss": 1.3081,
      "step": 1475
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.763447598313762,
      "learning_rate": 1.9999697328646345e-06,
      "loss": 1.419,
      "step": 1476
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.553254419945732,
      "learning_rate": 1.9999696915586243e-06,
      "loss": 1.5496,
      "step": 1477
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.8449155578371785,
      "learning_rate": 1.999969650224448e-06,
      "loss": 1.4168,
      "step": 1478
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.936042392237835,
      "learning_rate": 1.999969608862106e-06,
      "loss": 1.4295,
      "step": 1479
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.692520927812083,
      "learning_rate": 1.9999695674715985e-06,
      "loss": 1.4999,
      "step": 1480
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.618917093841338,
      "learning_rate": 1.9999695260529253e-06,
      "loss": 1.4672,
      "step": 1481
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.540512172018715,
      "learning_rate": 1.999969484606086e-06,
      "loss": 1.4142,
      "step": 1482
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.257964249055953,
      "learning_rate": 1.999969443131081e-06,
      "loss": 1.5545,
      "step": 1483
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.361199744025823,
      "learning_rate": 1.99996940162791e-06,
      "loss": 1.4301,
      "step": 1484
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.762262068823818,
      "learning_rate": 1.9999693600965733e-06,
      "loss": 1.3145,
      "step": 1485
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.507782598355478,
      "learning_rate": 1.9999693185370708e-06,
      "loss": 1.346,
      "step": 1486
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.20616920405698,
      "learning_rate": 1.9999692769494027e-06,
      "loss": 1.3654,
      "step": 1487
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.9199475980705385,
      "learning_rate": 1.9999692353335686e-06,
      "loss": 1.4154,
      "step": 1488
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.789863379301854,
      "learning_rate": 1.9999691936895685e-06,
      "loss": 1.4946,
      "step": 1489
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.2869321337326225,
      "learning_rate": 1.9999691520174025e-06,
      "loss": 1.3195,
      "step": 1490
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.757207256999973,
      "learning_rate": 1.999969110317071e-06,
      "loss": 1.6459,
      "step": 1491
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.856756819873707,
      "learning_rate": 1.999969068588574e-06,
      "loss": 1.3793,
      "step": 1492
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.1406066928499605,
      "learning_rate": 1.999969026831911e-06,
      "loss": 1.6314,
      "step": 1493
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.339460225757653,
      "learning_rate": 1.9999689850470814e-06,
      "loss": 1.2897,
      "step": 1494
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.118439798561242,
      "learning_rate": 1.999968943234087e-06,
      "loss": 1.3565,
      "step": 1495
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.239267360652984,
      "learning_rate": 1.9999689013929264e-06,
      "loss": 1.3343,
      "step": 1496
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.413381580899369,
      "learning_rate": 1.9999688595236003e-06,
      "loss": 1.3758,
      "step": 1497
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.032731231134136,
      "learning_rate": 1.999968817626108e-06,
      "loss": 1.4616,
      "step": 1498
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.181684854242491,
      "learning_rate": 1.99996877570045e-06,
      "loss": 1.5541,
      "step": 1499
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.529939429132134,
      "learning_rate": 1.9999687337466264e-06,
      "loss": 1.4291,
      "step": 1500
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.083882159448674,
      "learning_rate": 1.9999686917646365e-06,
      "loss": 1.527,
      "step": 1501
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.290245063554081,
      "learning_rate": 1.999968649754481e-06,
      "loss": 1.4742,
      "step": 1502
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.717789412166867,
      "learning_rate": 1.9999686077161605e-06,
      "loss": 1.4765,
      "step": 1503
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.545533558985268,
      "learning_rate": 1.999968565649673e-06,
      "loss": 1.306,
      "step": 1504
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.728947694233556,
      "learning_rate": 1.9999685235550206e-06,
      "loss": 1.3828,
      "step": 1505
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.662992982900369,
      "learning_rate": 1.999968481432202e-06,
      "loss": 1.4146,
      "step": 1506
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.039232945794986,
      "learning_rate": 1.9999684392812178e-06,
      "loss": 1.5415,
      "step": 1507
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.039673874673663,
      "learning_rate": 1.999968397102068e-06,
      "loss": 1.3505,
      "step": 1508
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.7877009049623025,
      "learning_rate": 1.9999683548947523e-06,
      "loss": 1.4874,
      "step": 1509
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.641088327797642,
      "learning_rate": 1.9999683126592705e-06,
      "loss": 1.4175,
      "step": 1510
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.831959829153163,
      "learning_rate": 1.999968270395623e-06,
      "loss": 1.4567,
      "step": 1511
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.699928337044365,
      "learning_rate": 1.99996822810381e-06,
      "loss": 1.392,
      "step": 1512
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.608406926276191,
      "learning_rate": 1.9999681857838308e-06,
      "loss": 1.4505,
      "step": 1513
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.867903533500069,
      "learning_rate": 1.9999681434356863e-06,
      "loss": 1.3949,
      "step": 1514
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.944562213455762,
      "learning_rate": 1.9999681010593754e-06,
      "loss": 1.2331,
      "step": 1515
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.2571821063164474,
      "learning_rate": 1.9999680586548995e-06,
      "loss": 1.5634,
      "step": 1516
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.907465548338297,
      "learning_rate": 1.999968016222257e-06,
      "loss": 1.3562,
      "step": 1517
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.276801096528798,
      "learning_rate": 1.9999679737614493e-06,
      "loss": 1.1667,
      "step": 1518
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.073955336024891,
      "learning_rate": 1.999967931272476e-06,
      "loss": 1.5431,
      "step": 1519
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.183148059254548,
      "learning_rate": 1.9999678887553364e-06,
      "loss": 1.5844,
      "step": 1520
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.591593727529927,
      "learning_rate": 1.9999678462100315e-06,
      "loss": 1.4007,
      "step": 1521
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.83807070074554,
      "learning_rate": 1.99996780363656e-06,
      "loss": 1.4673,
      "step": 1522
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.564201157030677,
      "learning_rate": 1.9999677610349238e-06,
      "loss": 1.4229,
      "step": 1523
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.506652623301389,
      "learning_rate": 1.9999677184051214e-06,
      "loss": 1.4544,
      "step": 1524
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.110861074492502,
      "learning_rate": 1.999967675747153e-06,
      "loss": 1.4452,
      "step": 1525
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.550965773070968,
      "learning_rate": 1.999967633061019e-06,
      "loss": 1.39,
      "step": 1526
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.977881886992321,
      "learning_rate": 1.9999675903467193e-06,
      "loss": 1.5182,
      "step": 1527
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.886960525204347,
      "learning_rate": 1.999967547604254e-06,
      "loss": 1.3334,
      "step": 1528
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.035221458256328,
      "learning_rate": 1.9999675048336225e-06,
      "loss": 1.5484,
      "step": 1529
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.127298418854591,
      "learning_rate": 1.9999674620348256e-06,
      "loss": 1.5482,
      "step": 1530
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.393021516760661,
      "learning_rate": 1.9999674192078627e-06,
      "loss": 1.3923,
      "step": 1531
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.597322785666696,
      "learning_rate": 1.9999673763527343e-06,
      "loss": 1.4507,
      "step": 1532
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.759409590128979,
      "learning_rate": 1.99996733346944e-06,
      "loss": 1.328,
      "step": 1533
    },
    {
      "epoch": 0.01,
      "eval_loss": 1.6537013053894043,
      "eval_runtime": 4.6319,
      "eval_samples_per_second": 1.943,
      "eval_steps_per_second": 1.079,
      "step": 1533
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.004922521706707,
      "learning_rate": 1.99996729055798e-06,
      "loss": 1.3943,
      "step": 1534
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.645165024309459,
      "learning_rate": 1.999967247618354e-06,
      "loss": 1.3322,
      "step": 1535
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.999318863504048,
      "learning_rate": 1.9999672046505628e-06,
      "loss": 1.5395,
      "step": 1536
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.86361645206254,
      "learning_rate": 1.9999671616546054e-06,
      "loss": 1.5212,
      "step": 1537
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.153420847923612,
      "learning_rate": 1.9999671186304825e-06,
      "loss": 1.5427,
      "step": 1538
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.0394644534786135,
      "learning_rate": 1.9999670755781936e-06,
      "loss": 1.2973,
      "step": 1539
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.633634743770092,
      "learning_rate": 1.9999670324977388e-06,
      "loss": 1.5085,
      "step": 1540
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.357519744982609,
      "learning_rate": 1.999966989389119e-06,
      "loss": 1.3645,
      "step": 1541
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.802169847384737,
      "learning_rate": 1.9999669462523325e-06,
      "loss": 1.5779,
      "step": 1542
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.668930926325562,
      "learning_rate": 1.999966903087381e-06,
      "loss": 1.4067,
      "step": 1543
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.43999149148277,
      "learning_rate": 1.9999668598942636e-06,
      "loss": 1.2107,
      "step": 1544
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.802567954958664,
      "learning_rate": 1.99996681667298e-06,
      "loss": 1.4996,
      "step": 1545
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.043050576777043,
      "learning_rate": 1.999966773423531e-06,
      "loss": 1.4864,
      "step": 1546
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.838293228729252,
      "learning_rate": 1.9999667301459164e-06,
      "loss": 1.5079,
      "step": 1547
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8303265370783235,
      "learning_rate": 1.999966686840136e-06,
      "loss": 1.6446,
      "step": 1548
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.4088008516017005,
      "learning_rate": 1.9999666435061896e-06,
      "loss": 1.4987,
      "step": 1549
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.441236329740386,
      "learning_rate": 1.9999666001440776e-06,
      "loss": 1.4233,
      "step": 1550
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.854304337374531,
      "learning_rate": 1.9999665567538e-06,
      "loss": 1.5069,
      "step": 1551
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.651751008584424,
      "learning_rate": 1.9999665133353567e-06,
      "loss": 1.4673,
      "step": 1552
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.650463359683428,
      "learning_rate": 1.9999664698887473e-06,
      "loss": 1.4744,
      "step": 1553
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.294289457136488,
      "learning_rate": 1.999966426413973e-06,
      "loss": 1.7254,
      "step": 1554
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.026568380786172,
      "learning_rate": 1.9999663829110324e-06,
      "loss": 1.574,
      "step": 1555
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.258151411074215,
      "learning_rate": 1.999966339379926e-06,
      "loss": 1.3902,
      "step": 1556
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.636549612306679,
      "learning_rate": 1.999966295820654e-06,
      "loss": 1.3695,
      "step": 1557
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.752515784505921,
      "learning_rate": 1.999966252233216e-06,
      "loss": 1.5815,
      "step": 1558
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.150495198663999,
      "learning_rate": 1.9999662086176125e-06,
      "loss": 1.4708,
      "step": 1559
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.696102556808951,
      "learning_rate": 1.9999661649738435e-06,
      "loss": 1.3237,
      "step": 1560
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.487064828854456,
      "learning_rate": 1.9999661213019085e-06,
      "loss": 1.367,
      "step": 1561
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.674090800118098,
      "learning_rate": 1.9999660776018076e-06,
      "loss": 1.3795,
      "step": 1562
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.884374781206321,
      "learning_rate": 1.9999660338735415e-06,
      "loss": 1.4984,
      "step": 1563
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8404936570617,
      "learning_rate": 1.9999659901171095e-06,
      "loss": 1.4553,
      "step": 1564
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.0291200173626,
      "learning_rate": 1.9999659463325115e-06,
      "loss": 1.4493,
      "step": 1565
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.489038208651526,
      "learning_rate": 1.999965902519748e-06,
      "loss": 1.4916,
      "step": 1566
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.325129603155862,
      "learning_rate": 1.999965858678819e-06,
      "loss": 1.4466,
      "step": 1567
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.845547041345931,
      "learning_rate": 1.999965814809724e-06,
      "loss": 1.4709,
      "step": 1568
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.6323997434080955,
      "learning_rate": 1.999965770912463e-06,
      "loss": 1.2593,
      "step": 1569
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.263469443114738,
      "learning_rate": 1.999965726987037e-06,
      "loss": 1.4309,
      "step": 1570
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.7842493763504,
      "learning_rate": 1.9999656830334447e-06,
      "loss": 1.5784,
      "step": 1571
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.581499884497734,
      "learning_rate": 1.999965639051687e-06,
      "loss": 1.4264,
      "step": 1572
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.713589565139688,
      "learning_rate": 1.9999655950417635e-06,
      "loss": 1.3676,
      "step": 1573
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.224520331971595,
      "learning_rate": 1.9999655510036744e-06,
      "loss": 1.5145,
      "step": 1574
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.044254687476926,
      "learning_rate": 1.9999655069374193e-06,
      "loss": 1.382,
      "step": 1575
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.998067415237716,
      "learning_rate": 1.999965462842999e-06,
      "loss": 1.5149,
      "step": 1576
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.376451146617778,
      "learning_rate": 1.9999654187204126e-06,
      "loss": 1.4605,
      "step": 1577
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.62147315713685,
      "learning_rate": 1.9999653745696605e-06,
      "loss": 1.5111,
      "step": 1578
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.386593885174143,
      "learning_rate": 1.999965330390743e-06,
      "loss": 1.5952,
      "step": 1579
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.6360487402587784,
      "learning_rate": 1.999965286183659e-06,
      "loss": 1.3293,
      "step": 1580
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.759388015554171,
      "learning_rate": 1.9999652419484104e-06,
      "loss": 1.3158,
      "step": 1581
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.720044228281999,
      "learning_rate": 1.9999651976849957e-06,
      "loss": 1.5035,
      "step": 1582
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.571250672292211,
      "learning_rate": 1.999965153393415e-06,
      "loss": 1.3919,
      "step": 1583
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.512671101809202,
      "learning_rate": 1.9999651090736693e-06,
      "loss": 1.3005,
      "step": 1584
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.311600498717284,
      "learning_rate": 1.999965064725757e-06,
      "loss": 1.5076,
      "step": 1585
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.4380918504811815,
      "learning_rate": 1.9999650203496794e-06,
      "loss": 1.5773,
      "step": 1586
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.821728820130227,
      "learning_rate": 1.999964975945436e-06,
      "loss": 1.4923,
      "step": 1587
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.701896651829335,
      "learning_rate": 1.9999649315130274e-06,
      "loss": 1.3324,
      "step": 1588
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.826124943857317,
      "learning_rate": 1.9999648870524526e-06,
      "loss": 1.3942,
      "step": 1589
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.252549552664715,
      "learning_rate": 1.9999648425637123e-06,
      "loss": 1.4413,
      "step": 1590
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.33848280401789,
      "learning_rate": 1.9999647980468065e-06,
      "loss": 1.3627,
      "step": 1591
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.232495688477331,
      "learning_rate": 1.999964753501735e-06,
      "loss": 1.371,
      "step": 1592
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.559527771002142,
      "learning_rate": 1.9999647089284974e-06,
      "loss": 1.4434,
      "step": 1593
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.70894162267966,
      "learning_rate": 1.9999646643270945e-06,
      "loss": 1.4673,
      "step": 1594
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.676920877976759,
      "learning_rate": 1.999964619697526e-06,
      "loss": 1.4621,
      "step": 1595
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.232169194197823,
      "learning_rate": 1.9999645750397918e-06,
      "loss": 1.4599,
      "step": 1596
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.693130001219432,
      "learning_rate": 1.9999645303538914e-06,
      "loss": 1.5755,
      "step": 1597
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.616061797152542,
      "learning_rate": 1.9999644856398255e-06,
      "loss": 1.525,
      "step": 1598
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.082035414984437,
      "learning_rate": 1.999964440897594e-06,
      "loss": 1.4748,
      "step": 1599
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.530227379151722,
      "learning_rate": 1.999964396127197e-06,
      "loss": 1.4081,
      "step": 1600
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.714257093847134,
      "learning_rate": 1.9999643513286343e-06,
      "loss": 1.4166,
      "step": 1601
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.188422008714217,
      "learning_rate": 1.999964306501906e-06,
      "loss": 1.3882,
      "step": 1602
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.037793578152508,
      "learning_rate": 1.999964261647012e-06,
      "loss": 1.4275,
      "step": 1603
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.775212123850943,
      "learning_rate": 1.9999642167639523e-06,
      "loss": 1.6035,
      "step": 1604
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.478124459351955,
      "learning_rate": 1.9999641718527268e-06,
      "loss": 1.3335,
      "step": 1605
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.055970312783448,
      "learning_rate": 1.9999641269133357e-06,
      "loss": 1.5906,
      "step": 1606
    },
    {
      "epoch": 0.01,
      "eval_loss": 1.6535860300064087,
      "eval_runtime": 4.6131,
      "eval_samples_per_second": 1.951,
      "eval_steps_per_second": 1.084,
      "step": 1606
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.7695927713899335,
      "learning_rate": 1.9999640819457787e-06,
      "loss": 1.4862,
      "step": 1607
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.631860591367596,
      "learning_rate": 1.9999640369500566e-06,
      "loss": 1.3437,
      "step": 1608
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.184619042074156,
      "learning_rate": 1.9999639919261685e-06,
      "loss": 1.3648,
      "step": 1609
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.273228198303511,
      "learning_rate": 1.999963946874115e-06,
      "loss": 1.2881,
      "step": 1610
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.019924351174469,
      "learning_rate": 1.9999639017938953e-06,
      "loss": 1.4094,
      "step": 1611
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.796825605500974,
      "learning_rate": 1.99996385668551e-06,
      "loss": 1.4356,
      "step": 1612
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.039100317720765,
      "learning_rate": 1.9999638115489595e-06,
      "loss": 1.507,
      "step": 1613
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.760579560458608,
      "learning_rate": 1.9999637663842433e-06,
      "loss": 1.2805,
      "step": 1614
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.289967418587153,
      "learning_rate": 1.999963721191361e-06,
      "loss": 1.5173,
      "step": 1615
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.905881261758454,
      "learning_rate": 1.9999636759703135e-06,
      "loss": 1.4773,
      "step": 1616
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.601006920242404,
      "learning_rate": 1.9999636307211002e-06,
      "loss": 1.4564,
      "step": 1617
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.623950290324913,
      "learning_rate": 1.9999635854437214e-06,
      "loss": 1.5243,
      "step": 1618
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.797135728663394,
      "learning_rate": 1.9999635401381767e-06,
      "loss": 1.6152,
      "step": 1619
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.494921471069805,
      "learning_rate": 1.9999634948044664e-06,
      "loss": 1.3623,
      "step": 1620
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.786655122435106,
      "learning_rate": 1.9999634494425906e-06,
      "loss": 1.3719,
      "step": 1621
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.5772063648225405,
      "learning_rate": 1.9999634040525492e-06,
      "loss": 1.6291,
      "step": 1622
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.797055299395125,
      "learning_rate": 1.999963358634342e-06,
      "loss": 1.4054,
      "step": 1623
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.881210206846928,
      "learning_rate": 1.999963313187969e-06,
      "loss": 1.3922,
      "step": 1624
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.7416575099427405,
      "learning_rate": 1.9999632677134306e-06,
      "loss": 1.3498,
      "step": 1625
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.094353766754562,
      "learning_rate": 1.9999632222107262e-06,
      "loss": 1.4907,
      "step": 1626
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.136116876869801,
      "learning_rate": 1.9999631766798568e-06,
      "loss": 1.4637,
      "step": 1627
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.842724728052074,
      "learning_rate": 1.9999631311208213e-06,
      "loss": 1.5065,
      "step": 1628
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.470556712718268,
      "learning_rate": 1.9999630855336203e-06,
      "loss": 1.4163,
      "step": 1629
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.4357945417775015,
      "learning_rate": 1.999963039918254e-06,
      "loss": 1.5127,
      "step": 1630
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.283784706095175,
      "learning_rate": 1.9999629942747213e-06,
      "loss": 1.5162,
      "step": 1631
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.171430776129922,
      "learning_rate": 1.9999629486030233e-06,
      "loss": 1.4975,
      "step": 1632
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.883035958460619,
      "learning_rate": 1.9999629029031597e-06,
      "loss": 1.5646,
      "step": 1633
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.868003718244226,
      "learning_rate": 1.999962857175131e-06,
      "loss": 1.4711,
      "step": 1634
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.688207116642901,
      "learning_rate": 1.999962811418936e-06,
      "loss": 1.235,
      "step": 1635
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.779397974493194,
      "learning_rate": 1.9999627656345753e-06,
      "loss": 1.4816,
      "step": 1636
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.787958938157594,
      "learning_rate": 1.9999627198220496e-06,
      "loss": 1.3246,
      "step": 1637
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.294916108987848,
      "learning_rate": 1.999962673981358e-06,
      "loss": 1.4679,
      "step": 1638
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.849618123624191,
      "learning_rate": 1.9999626281125002e-06,
      "loss": 1.4614,
      "step": 1639
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.677011611090417,
      "learning_rate": 1.9999625822154775e-06,
      "loss": 1.3398,
      "step": 1640
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.707575677733155,
      "learning_rate": 1.999962536290289e-06,
      "loss": 1.4338,
      "step": 1641
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.1791912413051815,
      "learning_rate": 1.999962490336935e-06,
      "loss": 1.6644,
      "step": 1642
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.636258199528408,
      "learning_rate": 1.999962444355415e-06,
      "loss": 1.3262,
      "step": 1643
    },
    {
      "epoch": 0.01,
      "grad_norm": 9.631481491029284,
      "learning_rate": 1.9999623983457297e-06,
      "loss": 1.3487,
      "step": 1644
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.406573344850598,
      "learning_rate": 1.9999623523078784e-06,
      "loss": 1.4415,
      "step": 1645
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.936420173298707,
      "learning_rate": 1.999962306241862e-06,
      "loss": 1.5284,
      "step": 1646
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.408101415603079,
      "learning_rate": 1.9999622601476796e-06,
      "loss": 1.5724,
      "step": 1647
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.91033489810393,
      "learning_rate": 1.999962214025332e-06,
      "loss": 1.4389,
      "step": 1648
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.7490093326999965,
      "learning_rate": 1.9999621678748186e-06,
      "loss": 1.4127,
      "step": 1649
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.02865055772892,
      "learning_rate": 1.999962121696139e-06,
      "loss": 1.5252,
      "step": 1650
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.941820991121326,
      "learning_rate": 1.9999620754892946e-06,
      "loss": 1.5824,
      "step": 1651
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.584733328893442,
      "learning_rate": 1.9999620292542845e-06,
      "loss": 1.4465,
      "step": 1652
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.668886241818989,
      "learning_rate": 1.9999619829911084e-06,
      "loss": 1.5671,
      "step": 1653
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8595287693689455,
      "learning_rate": 1.9999619366997673e-06,
      "loss": 1.4179,
      "step": 1654
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.505667922340234,
      "learning_rate": 1.99996189038026e-06,
      "loss": 1.3717,
      "step": 1655
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.854733211903465,
      "learning_rate": 1.999961844032587e-06,
      "loss": 1.5526,
      "step": 1656
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.052863076519862,
      "learning_rate": 1.999961797656749e-06,
      "loss": 1.5102,
      "step": 1657
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.391873652615683,
      "learning_rate": 1.999961751252745e-06,
      "loss": 1.313,
      "step": 1658
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.905334859088026,
      "learning_rate": 1.9999617048205758e-06,
      "loss": 1.4915,
      "step": 1659
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.90511142571514,
      "learning_rate": 1.9999616583602405e-06,
      "loss": 1.3705,
      "step": 1660
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.696590131572682,
      "learning_rate": 1.99996161187174e-06,
      "loss": 1.3774,
      "step": 1661
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.901553847339917,
      "learning_rate": 1.999961565355074e-06,
      "loss": 1.4858,
      "step": 1662
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.080014172214551,
      "learning_rate": 1.9999615188102415e-06,
      "loss": 1.3527,
      "step": 1663
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.286819461235812,
      "learning_rate": 1.999961472237244e-06,
      "loss": 1.5443,
      "step": 1664
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.083890130164741,
      "learning_rate": 1.999961425636081e-06,
      "loss": 1.4417,
      "step": 1665
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8993271722773235,
      "learning_rate": 1.9999613790067526e-06,
      "loss": 1.4137,
      "step": 1666
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.130884125485172,
      "learning_rate": 1.9999613323492586e-06,
      "loss": 1.5135,
      "step": 1667
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.560714520466941,
      "learning_rate": 1.9999612856635986e-06,
      "loss": 1.2326,
      "step": 1668
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.469864395669994,
      "learning_rate": 1.9999612389497735e-06,
      "loss": 1.3952,
      "step": 1669
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.0576333320329425,
      "learning_rate": 1.9999611922077824e-06,
      "loss": 1.4062,
      "step": 1670
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.550473014043539,
      "learning_rate": 1.999961145437626e-06,
      "loss": 1.3792,
      "step": 1671
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.749509524045929,
      "learning_rate": 1.9999610986393037e-06,
      "loss": 1.4487,
      "step": 1672
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.365020045883757,
      "learning_rate": 1.999961051812816e-06,
      "loss": 1.4974,
      "step": 1673
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.624114515501705,
      "learning_rate": 1.9999610049581627e-06,
      "loss": 1.388,
      "step": 1674
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.510101567218574,
      "learning_rate": 1.999960958075344e-06,
      "loss": 1.3547,
      "step": 1675
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.07080055515291,
      "learning_rate": 1.9999609111643596e-06,
      "loss": 1.4748,
      "step": 1676
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8404551347295985,
      "learning_rate": 1.9999608642252094e-06,
      "loss": 1.4301,
      "step": 1677
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.636720567872511,
      "learning_rate": 1.999960817257894e-06,
      "loss": 1.351,
      "step": 1678
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.652209867428249,
      "learning_rate": 1.999960770262413e-06,
      "loss": 1.4968,
      "step": 1679
    },
    {
      "epoch": 0.01,
      "eval_loss": 1.6513854265213013,
      "eval_runtime": 4.6169,
      "eval_samples_per_second": 1.949,
      "eval_steps_per_second": 1.083,
      "step": 1679
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.725930228784767,
      "learning_rate": 1.999960723238766e-06,
      "loss": 1.409,
      "step": 1680
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.954953461456659,
      "learning_rate": 1.9999606761869537e-06,
      "loss": 1.4448,
      "step": 1681
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.255376181858065,
      "learning_rate": 1.9999606291069757e-06,
      "loss": 1.3616,
      "step": 1682
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.709113389093899,
      "learning_rate": 1.9999605819988326e-06,
      "loss": 1.4076,
      "step": 1683
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.766680383528532,
      "learning_rate": 1.9999605348625236e-06,
      "loss": 1.2462,
      "step": 1684
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.409591894443744,
      "learning_rate": 1.999960487698049e-06,
      "loss": 1.4091,
      "step": 1685
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.886994914850528,
      "learning_rate": 1.9999604405054093e-06,
      "loss": 1.4295,
      "step": 1686
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.933768715491502,
      "learning_rate": 1.9999603932846036e-06,
      "loss": 1.3117,
      "step": 1687
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.404087310320486,
      "learning_rate": 1.9999603460356324e-06,
      "loss": 1.4082,
      "step": 1688
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.132686826605006,
      "learning_rate": 1.9999602987584952e-06,
      "loss": 1.4659,
      "step": 1689
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.9809185718468205,
      "learning_rate": 1.999960251453193e-06,
      "loss": 1.5435,
      "step": 1690
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.719735058251011,
      "learning_rate": 1.999960204119725e-06,
      "loss": 1.3241,
      "step": 1691
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.016858779358495,
      "learning_rate": 1.9999601567580917e-06,
      "loss": 1.3907,
      "step": 1692
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.518980098450484,
      "learning_rate": 1.999960109368293e-06,
      "loss": 1.5071,
      "step": 1693
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.56680393314666,
      "learning_rate": 1.9999600619503284e-06,
      "loss": 1.277,
      "step": 1694
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.221095667195953,
      "learning_rate": 1.9999600145041984e-06,
      "loss": 1.5565,
      "step": 1695
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.838855837516108,
      "learning_rate": 1.9999599670299024e-06,
      "loss": 1.5075,
      "step": 1696
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.181585087689867,
      "learning_rate": 1.9999599195274414e-06,
      "loss": 1.4358,
      "step": 1697
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.775117548464921,
      "learning_rate": 1.9999598719968148e-06,
      "loss": 1.3676,
      "step": 1698
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.675367411416997,
      "learning_rate": 1.9999598244380226e-06,
      "loss": 1.4902,
      "step": 1699
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.534748298661586,
      "learning_rate": 1.999959776851065e-06,
      "loss": 1.3374,
      "step": 1700
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.092872429354068,
      "learning_rate": 1.9999597292359413e-06,
      "loss": 1.3938,
      "step": 1701
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.1363468139465835,
      "learning_rate": 1.9999596815926525e-06,
      "loss": 1.4641,
      "step": 1702
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.679526003354978,
      "learning_rate": 1.9999596339211982e-06,
      "loss": 1.4826,
      "step": 1703
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.409980918956625,
      "learning_rate": 1.999959586221578e-06,
      "loss": 1.459,
      "step": 1704
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.506106987900174,
      "learning_rate": 1.9999595384937926e-06,
      "loss": 1.4694,
      "step": 1705
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.139999841003741,
      "learning_rate": 1.9999594907378412e-06,
      "loss": 1.497,
      "step": 1706
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.966305341143314,
      "learning_rate": 1.999959442953725e-06,
      "loss": 1.4117,
      "step": 1707
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.81613488843635,
      "learning_rate": 1.9999593951414427e-06,
      "loss": 1.6682,
      "step": 1708
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.377588601757201,
      "learning_rate": 1.999959347300995e-06,
      "loss": 1.3065,
      "step": 1709
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.023397142367646,
      "learning_rate": 1.999959299432382e-06,
      "loss": 1.4304,
      "step": 1710
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.675493936243796,
      "learning_rate": 1.9999592515356035e-06,
      "loss": 1.2623,
      "step": 1711
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.86058388014828,
      "learning_rate": 1.9999592036106593e-06,
      "loss": 1.4823,
      "step": 1712
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.556345518412603,
      "learning_rate": 1.9999591556575496e-06,
      "loss": 1.4409,
      "step": 1713
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.066520926861509,
      "learning_rate": 1.9999591076762744e-06,
      "loss": 1.3931,
      "step": 1714
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.416558606687308,
      "learning_rate": 1.9999590596668336e-06,
      "loss": 1.5513,
      "step": 1715
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.566717284747848,
      "learning_rate": 1.9999590116292273e-06,
      "loss": 1.4196,
      "step": 1716
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.774215759192912,
      "learning_rate": 1.9999589635634554e-06,
      "loss": 1.2179,
      "step": 1717
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.871975120855977,
      "learning_rate": 1.999958915469518e-06,
      "loss": 1.3091,
      "step": 1718
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.667731197924079,
      "learning_rate": 1.999958867347415e-06,
      "loss": 1.4477,
      "step": 1719
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.363688597457996,
      "learning_rate": 1.9999588191971466e-06,
      "loss": 1.4968,
      "step": 1720
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.524755329539832,
      "learning_rate": 1.999958771018713e-06,
      "loss": 1.5737,
      "step": 1721
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.844327612634827,
      "learning_rate": 1.9999587228121135e-06,
      "loss": 1.5503,
      "step": 1722
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.545944405243781,
      "learning_rate": 1.999958674577349e-06,
      "loss": 1.227,
      "step": 1723
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.569019618596105,
      "learning_rate": 1.999958626314418e-06,
      "loss": 1.3704,
      "step": 1724
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.840943336713329,
      "learning_rate": 1.9999585780233224e-06,
      "loss": 1.5026,
      "step": 1725
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.04053411151362,
      "learning_rate": 1.9999585297040607e-06,
      "loss": 1.5642,
      "step": 1726
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.6356258265297985,
      "learning_rate": 1.999958481356634e-06,
      "loss": 1.5226,
      "step": 1727
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.640699731713796,
      "learning_rate": 1.9999584329810415e-06,
      "loss": 1.3299,
      "step": 1728
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.472243462527622,
      "learning_rate": 1.9999583845772836e-06,
      "loss": 1.4519,
      "step": 1729
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.611553303462497,
      "learning_rate": 1.99995833614536e-06,
      "loss": 1.4071,
      "step": 1730
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.944281632533813,
      "learning_rate": 1.999958287685271e-06,
      "loss": 1.2544,
      "step": 1731
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.6772733567861176,
      "learning_rate": 1.9999582391970166e-06,
      "loss": 1.3753,
      "step": 1732
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.263194033542804,
      "learning_rate": 1.999958190680597e-06,
      "loss": 1.2437,
      "step": 1733
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.654617300777728,
      "learning_rate": 1.9999581421360114e-06,
      "loss": 1.3443,
      "step": 1734
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.17256863920859,
      "learning_rate": 1.9999580935632606e-06,
      "loss": 1.5056,
      "step": 1735
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.491095887490403,
      "learning_rate": 1.999958044962344e-06,
      "loss": 1.32,
      "step": 1736
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.391828873655179,
      "learning_rate": 1.999957996333262e-06,
      "loss": 1.342,
      "step": 1737
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.966305374933118,
      "learning_rate": 1.999957947676015e-06,
      "loss": 1.3956,
      "step": 1738
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.995740008299241,
      "learning_rate": 1.999957898990602e-06,
      "loss": 1.5289,
      "step": 1739
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.381812845234866,
      "learning_rate": 1.9999578502770235e-06,
      "loss": 1.2572,
      "step": 1740
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.61281321178712,
      "learning_rate": 1.9999578015352795e-06,
      "loss": 1.2066,
      "step": 1741
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.441293918501137,
      "learning_rate": 1.99995775276537e-06,
      "loss": 1.3348,
      "step": 1742
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.187132615217963,
      "learning_rate": 1.9999577039672954e-06,
      "loss": 1.5402,
      "step": 1743
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.842621185034447,
      "learning_rate": 1.9999576551410553e-06,
      "loss": 1.3723,
      "step": 1744
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.4907842747894895,
      "learning_rate": 1.9999576062866496e-06,
      "loss": 1.3577,
      "step": 1745
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.71227076890854,
      "learning_rate": 1.999957557404078e-06,
      "loss": 1.3866,
      "step": 1746
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.279369312058059,
      "learning_rate": 1.999957508493341e-06,
      "loss": 1.4093,
      "step": 1747
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.646829536444046,
      "learning_rate": 1.9999574595544392e-06,
      "loss": 1.4286,
      "step": 1748
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.593261228259904,
      "learning_rate": 1.9999574105873714e-06,
      "loss": 1.4716,
      "step": 1749
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.0886616589245,
      "learning_rate": 1.999957361592138e-06,
      "loss": 1.6598,
      "step": 1750
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.753829674453155,
      "learning_rate": 1.9999573125687395e-06,
      "loss": 1.3023,
      "step": 1751
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.911294692900917,
      "learning_rate": 1.9999572635171754e-06,
      "loss": 1.4902,
      "step": 1752
    },
    {
      "epoch": 0.01,
      "eval_loss": 1.6478168964385986,
      "eval_runtime": 4.6461,
      "eval_samples_per_second": 1.937,
      "eval_steps_per_second": 1.076,
      "step": 1752
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.700414649164786,
      "learning_rate": 1.999957214437446e-06,
      "loss": 1.482,
      "step": 1753
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.9281821284726615,
      "learning_rate": 1.9999571653295507e-06,
      "loss": 1.306,
      "step": 1754
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.787957118049158,
      "learning_rate": 1.99995711619349e-06,
      "loss": 1.3873,
      "step": 1755
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.631636647001677,
      "learning_rate": 1.999957067029264e-06,
      "loss": 1.4089,
      "step": 1756
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.754702249860543,
      "learning_rate": 1.9999570178368725e-06,
      "loss": 1.4074,
      "step": 1757
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.069526479435708,
      "learning_rate": 1.9999569686163157e-06,
      "loss": 1.4378,
      "step": 1758
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.9557014533023835,
      "learning_rate": 1.9999569193675933e-06,
      "loss": 1.3766,
      "step": 1759
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.889794648818529,
      "learning_rate": 1.9999568700907053e-06,
      "loss": 1.5023,
      "step": 1760
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.6039304576772,
      "learning_rate": 1.9999568207856523e-06,
      "loss": 1.3601,
      "step": 1761
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.626450487742758,
      "learning_rate": 1.9999567714524337e-06,
      "loss": 1.3328,
      "step": 1762
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.935339476011332,
      "learning_rate": 1.999956722091049e-06,
      "loss": 1.4359,
      "step": 1763
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.059440744798255,
      "learning_rate": 1.9999566727014994e-06,
      "loss": 1.4256,
      "step": 1764
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.856890423684793,
      "learning_rate": 1.999956623283784e-06,
      "loss": 1.3171,
      "step": 1765
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.243931477116334,
      "learning_rate": 1.999956573837904e-06,
      "loss": 1.369,
      "step": 1766
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.998539551416535,
      "learning_rate": 1.9999565243638576e-06,
      "loss": 1.439,
      "step": 1767
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.836947003615344,
      "learning_rate": 1.999956474861646e-06,
      "loss": 1.4922,
      "step": 1768
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.000816070522979,
      "learning_rate": 1.999956425331269e-06,
      "loss": 1.5798,
      "step": 1769
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.162103021341265,
      "learning_rate": 1.9999563757727267e-06,
      "loss": 1.4635,
      "step": 1770
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.754537659429389,
      "learning_rate": 1.999956326186019e-06,
      "loss": 1.4111,
      "step": 1771
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.0189050672383,
      "learning_rate": 1.999956276571146e-06,
      "loss": 1.332,
      "step": 1772
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.489911679774594,
      "learning_rate": 1.999956226928107e-06,
      "loss": 1.3697,
      "step": 1773
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.139806714207968,
      "learning_rate": 1.999956177256903e-06,
      "loss": 1.4717,
      "step": 1774
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.74100968652323,
      "learning_rate": 1.9999561275575333e-06,
      "loss": 1.4658,
      "step": 1775
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.879178648686916,
      "learning_rate": 1.9999560778299984e-06,
      "loss": 1.4636,
      "step": 1776
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.891078192372891,
      "learning_rate": 1.999956028074298e-06,
      "loss": 1.5723,
      "step": 1777
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.653962016600203,
      "learning_rate": 1.999955978290432e-06,
      "loss": 1.5495,
      "step": 1778
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.707164030187875,
      "learning_rate": 1.9999559284784005e-06,
      "loss": 1.64,
      "step": 1779
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.454220280449505,
      "learning_rate": 1.999955878638204e-06,
      "loss": 1.336,
      "step": 1780
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.593524682406172,
      "learning_rate": 1.9999558287698418e-06,
      "loss": 1.387,
      "step": 1781
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.265214222108159,
      "learning_rate": 1.999955778873314e-06,
      "loss": 1.5196,
      "step": 1782
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8572705429997525,
      "learning_rate": 1.999955728948621e-06,
      "loss": 1.3624,
      "step": 1783
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.789880139543268,
      "learning_rate": 1.9999556789957625e-06,
      "loss": 1.4488,
      "step": 1784
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.621010536150032,
      "learning_rate": 1.9999556290147386e-06,
      "loss": 1.4625,
      "step": 1785
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.383083285855116,
      "learning_rate": 1.999955579005549e-06,
      "loss": 1.4843,
      "step": 1786
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.724436780612743,
      "learning_rate": 1.9999555289681946e-06,
      "loss": 1.3203,
      "step": 1787
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.410873580372611,
      "learning_rate": 1.999955478902674e-06,
      "loss": 1.5224,
      "step": 1788
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.290692087535022,
      "learning_rate": 1.9999554288089885e-06,
      "loss": 1.6514,
      "step": 1789
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.756689362126989,
      "learning_rate": 1.9999553786871377e-06,
      "loss": 1.4927,
      "step": 1790
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.336049328185806,
      "learning_rate": 1.9999553285371214e-06,
      "loss": 1.4515,
      "step": 1791
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.194031618429659,
      "learning_rate": 1.999955278358939e-06,
      "loss": 1.5161,
      "step": 1792
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.0237033121897845,
      "learning_rate": 1.9999552281525923e-06,
      "loss": 1.4292,
      "step": 1793
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.482776055356631,
      "learning_rate": 1.9999551779180794e-06,
      "loss": 1.402,
      "step": 1794
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.737389022108268,
      "learning_rate": 1.9999551276554014e-06,
      "loss": 1.5099,
      "step": 1795
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.878318638241726,
      "learning_rate": 1.999955077364558e-06,
      "loss": 1.529,
      "step": 1796
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.54561426906885,
      "learning_rate": 1.999955027045549e-06,
      "loss": 1.451,
      "step": 1797
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.560824329023515,
      "learning_rate": 1.999954976698375e-06,
      "loss": 1.3774,
      "step": 1798
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.798803868724221,
      "learning_rate": 1.9999549263230347e-06,
      "loss": 1.4121,
      "step": 1799
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.383273482325183,
      "learning_rate": 1.99995487591953e-06,
      "loss": 1.4391,
      "step": 1800
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.4091547065978425,
      "learning_rate": 1.9999548254878595e-06,
      "loss": 1.2512,
      "step": 1801
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.838374021260141,
      "learning_rate": 1.999954775028023e-06,
      "loss": 1.4316,
      "step": 1802
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.461055735973189,
      "learning_rate": 1.999954724540022e-06,
      "loss": 1.5318,
      "step": 1803
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.633136261316013,
      "learning_rate": 1.9999546740238554e-06,
      "loss": 1.5676,
      "step": 1804
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.303894260809806,
      "learning_rate": 1.9999546234795233e-06,
      "loss": 1.3127,
      "step": 1805
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.072682679306013,
      "learning_rate": 1.9999545729070256e-06,
      "loss": 1.655,
      "step": 1806
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8324040981937815,
      "learning_rate": 1.999954522306363e-06,
      "loss": 1.4801,
      "step": 1807
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.656940444517245,
      "learning_rate": 1.9999544716775345e-06,
      "loss": 1.3215,
      "step": 1808
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.390009789147993,
      "learning_rate": 1.9999544210205406e-06,
      "loss": 1.4592,
      "step": 1809
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.190582805213608,
      "learning_rate": 1.9999543703353816e-06,
      "loss": 1.2857,
      "step": 1810
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.814450823568367,
      "learning_rate": 1.999954319622057e-06,
      "loss": 1.3498,
      "step": 1811
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.868225374359218,
      "learning_rate": 1.9999542688805674e-06,
      "loss": 1.414,
      "step": 1812
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.711417209971132,
      "learning_rate": 1.9999542181109123e-06,
      "loss": 1.355,
      "step": 1813
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.007722725809002,
      "learning_rate": 1.9999541673130915e-06,
      "loss": 1.4713,
      "step": 1814
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8626655741355,
      "learning_rate": 1.9999541164871053e-06,
      "loss": 1.4264,
      "step": 1815
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.723792462194043,
      "learning_rate": 1.999954065632954e-06,
      "loss": 1.4045,
      "step": 1816
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.00100690699963,
      "learning_rate": 1.9999540147506374e-06,
      "loss": 1.4712,
      "step": 1817
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.253942127148356,
      "learning_rate": 1.9999539638401553e-06,
      "loss": 1.4985,
      "step": 1818
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.246538314584338,
      "learning_rate": 1.9999539129015073e-06,
      "loss": 1.4158,
      "step": 1819
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.063213658151458,
      "learning_rate": 1.9999538619346947e-06,
      "loss": 1.4696,
      "step": 1820
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.3593617360186405,
      "learning_rate": 1.9999538109397164e-06,
      "loss": 1.4141,
      "step": 1821
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.648351531479633,
      "learning_rate": 1.9999537599165726e-06,
      "loss": 1.4502,
      "step": 1822
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.134127084206851,
      "learning_rate": 1.9999537088652638e-06,
      "loss": 1.628,
      "step": 1823
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.20685547745487,
      "learning_rate": 1.9999536577857893e-06,
      "loss": 1.2484,
      "step": 1824
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.404803458184888,
      "learning_rate": 1.9999536066781494e-06,
      "loss": 1.4614,
      "step": 1825
    },
    {
      "epoch": 0.01,
      "eval_loss": 1.6452957391738892,
      "eval_runtime": 4.6377,
      "eval_samples_per_second": 1.941,
      "eval_steps_per_second": 1.078,
      "step": 1825
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.325255851791553,
      "learning_rate": 1.9999535555423443e-06,
      "loss": 1.2913,
      "step": 1826
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.812379984138036,
      "learning_rate": 1.9999535043783737e-06,
      "loss": 1.4321,
      "step": 1827
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.574060048853807,
      "learning_rate": 1.999953453186238e-06,
      "loss": 1.4385,
      "step": 1828
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.690167042783861,
      "learning_rate": 1.9999534019659366e-06,
      "loss": 1.4505,
      "step": 1829
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.963037184486775,
      "learning_rate": 1.9999533507174702e-06,
      "loss": 1.3754,
      "step": 1830
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.202540422032654,
      "learning_rate": 1.9999532994408383e-06,
      "loss": 1.5731,
      "step": 1831
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.747566281161776,
      "learning_rate": 1.999953248136041e-06,
      "loss": 1.35,
      "step": 1832
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.016639574119175,
      "learning_rate": 1.9999531968030782e-06,
      "loss": 1.5687,
      "step": 1833
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.195599580383433,
      "learning_rate": 1.99995314544195e-06,
      "loss": 1.3277,
      "step": 1834
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.71924385337544,
      "learning_rate": 1.999953094052657e-06,
      "loss": 1.3144,
      "step": 1835
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.830368171047517,
      "learning_rate": 1.999953042635198e-06,
      "loss": 1.5285,
      "step": 1836
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.34374594117172,
      "learning_rate": 1.999952991189574e-06,
      "loss": 1.4602,
      "step": 1837
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.041136604255877,
      "learning_rate": 1.9999529397157847e-06,
      "loss": 1.4805,
      "step": 1838
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.619671333702356,
      "learning_rate": 1.9999528882138298e-06,
      "loss": 1.4413,
      "step": 1839
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.975416181507071,
      "learning_rate": 1.9999528366837097e-06,
      "loss": 1.395,
      "step": 1840
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.04709900703668,
      "learning_rate": 1.9999527851254245e-06,
      "loss": 1.4027,
      "step": 1841
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.834514593532968,
      "learning_rate": 1.9999527335389733e-06,
      "loss": 1.359,
      "step": 1842
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.770095193729858,
      "learning_rate": 1.9999526819243574e-06,
      "loss": 1.5645,
      "step": 1843
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.082331923840744,
      "learning_rate": 1.9999526302815756e-06,
      "loss": 1.436,
      "step": 1844
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.516471050914241,
      "learning_rate": 1.999952578610629e-06,
      "loss": 1.338,
      "step": 1845
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.952184920350597,
      "learning_rate": 1.999952526911517e-06,
      "loss": 1.4534,
      "step": 1846
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.797375795198566,
      "learning_rate": 1.9999524751842394e-06,
      "loss": 1.3259,
      "step": 1847
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.909335461688805,
      "learning_rate": 1.9999524234287963e-06,
      "loss": 1.464,
      "step": 1848
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.7122725461586,
      "learning_rate": 1.9999523716451884e-06,
      "loss": 1.4756,
      "step": 1849
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.836041938475466,
      "learning_rate": 1.9999523198334146e-06,
      "loss": 1.4444,
      "step": 1850
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.42950126517274,
      "learning_rate": 1.999952267993476e-06,
      "loss": 1.3947,
      "step": 1851
    },
    {
      "epoch": 0.01,
      "grad_norm": 15.611894041325478,
      "learning_rate": 1.9999522161253717e-06,
      "loss": 1.5477,
      "step": 1852
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.816487768685785,
      "learning_rate": 1.999952164229102e-06,
      "loss": 1.4487,
      "step": 1853
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.705134898369493,
      "learning_rate": 1.9999521123046674e-06,
      "loss": 1.2334,
      "step": 1854
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.9204906496645675,
      "learning_rate": 1.999952060352067e-06,
      "loss": 1.367,
      "step": 1855
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.302614564464106,
      "learning_rate": 1.999952008371302e-06,
      "loss": 1.503,
      "step": 1856
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.290262594106217,
      "learning_rate": 1.999951956362371e-06,
      "loss": 1.6158,
      "step": 1857
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.832322054118164,
      "learning_rate": 1.999951904325275e-06,
      "loss": 1.3938,
      "step": 1858
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.918339792868505,
      "learning_rate": 1.9999518522600134e-06,
      "loss": 1.5048,
      "step": 1859
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.590992917749576,
      "learning_rate": 1.999951800166587e-06,
      "loss": 1.3696,
      "step": 1860
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.614974294086556,
      "learning_rate": 1.9999517480449946e-06,
      "loss": 1.5505,
      "step": 1861
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.7241251284973576,
      "learning_rate": 1.9999516958952373e-06,
      "loss": 1.3711,
      "step": 1862
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.122789301908437,
      "learning_rate": 1.999951643717315e-06,
      "loss": 1.5607,
      "step": 1863
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.849382872781319,
      "learning_rate": 1.9999515915112265e-06,
      "loss": 1.1324,
      "step": 1864
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.873163835881678,
      "learning_rate": 1.9999515392769734e-06,
      "loss": 1.5033,
      "step": 1865
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.173654117397512,
      "learning_rate": 1.9999514870145548e-06,
      "loss": 1.5117,
      "step": 1866
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.642950123093243,
      "learning_rate": 1.9999514347239706e-06,
      "loss": 1.4977,
      "step": 1867
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.846117886376132,
      "learning_rate": 1.9999513824052214e-06,
      "loss": 1.4254,
      "step": 1868
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.101156060458363,
      "learning_rate": 1.999951330058307e-06,
      "loss": 1.5828,
      "step": 1869
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.688953673802245,
      "learning_rate": 1.999951277683227e-06,
      "loss": 1.4109,
      "step": 1870
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.703112399070017,
      "learning_rate": 1.999951225279982e-06,
      "loss": 1.4139,
      "step": 1871
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.310901288329353,
      "learning_rate": 1.999951172848572e-06,
      "loss": 1.6931,
      "step": 1872
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.741193705049953,
      "learning_rate": 1.9999511203889957e-06,
      "loss": 1.5699,
      "step": 1873
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.6035588106756835,
      "learning_rate": 1.9999510679012545e-06,
      "loss": 1.4299,
      "step": 1874
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.662657586420871,
      "learning_rate": 1.999951015385348e-06,
      "loss": 1.3702,
      "step": 1875
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8990643238694735,
      "learning_rate": 1.9999509628412766e-06,
      "loss": 1.4812,
      "step": 1876
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.948600983425892,
      "learning_rate": 1.9999509102690396e-06,
      "loss": 1.4146,
      "step": 1877
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.635028291248577,
      "learning_rate": 1.9999508576686375e-06,
      "loss": 1.4875,
      "step": 1878
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.294273694280311,
      "learning_rate": 1.9999508050400702e-06,
      "loss": 1.332,
      "step": 1879
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.184690906135398,
      "learning_rate": 1.999950752383337e-06,
      "loss": 1.3164,
      "step": 1880
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.720673987223821,
      "learning_rate": 1.999950699698439e-06,
      "loss": 1.455,
      "step": 1881
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.367468043629218,
      "learning_rate": 1.9999506469853757e-06,
      "loss": 1.3821,
      "step": 1882
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.599591528503925,
      "learning_rate": 1.999950594244147e-06,
      "loss": 1.4663,
      "step": 1883
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.428585648964182,
      "learning_rate": 1.999950541474753e-06,
      "loss": 1.3065,
      "step": 1884
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.224684072029214,
      "learning_rate": 1.999950488677194e-06,
      "loss": 1.414,
      "step": 1885
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.640404642843782,
      "learning_rate": 1.9999504358514695e-06,
      "loss": 1.4714,
      "step": 1886
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.799354311682633,
      "learning_rate": 1.9999503829975796e-06,
      "loss": 1.3347,
      "step": 1887
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.543001580495537,
      "learning_rate": 1.9999503301155246e-06,
      "loss": 1.3697,
      "step": 1888
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.736298613103733,
      "learning_rate": 1.999950277205304e-06,
      "loss": 1.3371,
      "step": 1889
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.571900592136959,
      "learning_rate": 1.999950224266919e-06,
      "loss": 1.385,
      "step": 1890
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.969693151715899,
      "learning_rate": 1.9999501713003677e-06,
      "loss": 1.3239,
      "step": 1891
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.655737378919456,
      "learning_rate": 1.999950118305652e-06,
      "loss": 1.5423,
      "step": 1892
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.714447970923717,
      "learning_rate": 1.9999500652827704e-06,
      "loss": 1.5264,
      "step": 1893
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.780305107625251,
      "learning_rate": 1.9999500122317235e-06,
      "loss": 1.3597,
      "step": 1894
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.133517088409541,
      "learning_rate": 1.999949959152512e-06,
      "loss": 1.3169,
      "step": 1895
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.924450244844968,
      "learning_rate": 1.9999499060451342e-06,
      "loss": 1.4864,
      "step": 1896
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.275295688447123,
      "learning_rate": 1.9999498529095924e-06,
      "loss": 1.5597,
      "step": 1897
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.7042297343124515,
      "learning_rate": 1.999949799745884e-06,
      "loss": 1.31,
      "step": 1898
    },
    {
      "epoch": 0.01,
      "eval_loss": 1.6416113376617432,
      "eval_runtime": 4.6259,
      "eval_samples_per_second": 1.946,
      "eval_steps_per_second": 1.081,
      "step": 1898
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.423106928459506,
      "learning_rate": 1.9999497465540116e-06,
      "loss": 1.4495,
      "step": 1899
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.0399225789912245,
      "learning_rate": 1.999949693333973e-06,
      "loss": 1.3536,
      "step": 1900
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.928948443819752,
      "learning_rate": 1.9999496400857695e-06,
      "loss": 1.4797,
      "step": 1901
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.67479846437279,
      "learning_rate": 1.9999495868094007e-06,
      "loss": 1.3,
      "step": 1902
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.103095669421286,
      "learning_rate": 1.999949533504867e-06,
      "loss": 1.4622,
      "step": 1903
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.134179165961643,
      "learning_rate": 1.9999494801721675e-06,
      "loss": 1.5459,
      "step": 1904
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.025236385305749,
      "learning_rate": 1.999949426811303e-06,
      "loss": 1.5205,
      "step": 1905
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.22119763423394,
      "learning_rate": 1.9999493734222734e-06,
      "loss": 1.3004,
      "step": 1906
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.669191447439257,
      "learning_rate": 1.9999493200050783e-06,
      "loss": 1.3201,
      "step": 1907
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.968754828299794,
      "learning_rate": 1.999949266559718e-06,
      "loss": 1.4346,
      "step": 1908
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.487597166469777,
      "learning_rate": 1.9999492130861926e-06,
      "loss": 1.4025,
      "step": 1909
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.879843563485657,
      "learning_rate": 1.9999491595845017e-06,
      "loss": 1.3078,
      "step": 1910
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.910078993452133,
      "learning_rate": 1.9999491060546456e-06,
      "loss": 1.5774,
      "step": 1911
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.647585620113467,
      "learning_rate": 1.9999490524966245e-06,
      "loss": 1.5707,
      "step": 1912
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.567026839443758,
      "learning_rate": 1.999948998910438e-06,
      "loss": 1.3951,
      "step": 1913
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.687048241254901,
      "learning_rate": 1.9999489452960864e-06,
      "loss": 1.4327,
      "step": 1914
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.902103912285808,
      "learning_rate": 1.9999488916535695e-06,
      "loss": 1.477,
      "step": 1915
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.57359910064146,
      "learning_rate": 1.9999488379828874e-06,
      "loss": 1.4144,
      "step": 1916
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.105713137074651,
      "learning_rate": 1.99994878428404e-06,
      "loss": 1.4075,
      "step": 1917
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.001514955863734,
      "learning_rate": 1.999948730557027e-06,
      "loss": 1.5653,
      "step": 1918
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.65944563546782,
      "learning_rate": 1.9999486768018493e-06,
      "loss": 1.162,
      "step": 1919
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.019156415872438,
      "learning_rate": 1.999948623018506e-06,
      "loss": 1.6276,
      "step": 1920
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.677363975537384,
      "learning_rate": 1.999948569206998e-06,
      "loss": 1.4125,
      "step": 1921
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.929950546593268,
      "learning_rate": 1.999948515367324e-06,
      "loss": 1.6226,
      "step": 1922
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.6654710600977,
      "learning_rate": 1.9999484614994856e-06,
      "loss": 1.3902,
      "step": 1923
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.450030912502321,
      "learning_rate": 1.9999484076034814e-06,
      "loss": 1.4725,
      "step": 1924
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.994408282983105,
      "learning_rate": 1.9999483536793124e-06,
      "loss": 1.5852,
      "step": 1925
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.714476017763944,
      "learning_rate": 1.999948299726978e-06,
      "loss": 1.4207,
      "step": 1926
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.767073689418852,
      "learning_rate": 1.999948245746478e-06,
      "loss": 1.5189,
      "step": 1927
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.392481238520719,
      "learning_rate": 1.9999481917378133e-06,
      "loss": 1.2504,
      "step": 1928
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.354120579996221,
      "learning_rate": 1.9999481377009834e-06,
      "loss": 1.3287,
      "step": 1929
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.643549574305054,
      "learning_rate": 1.999948083635988e-06,
      "loss": 1.4252,
      "step": 1930
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.6243115760185605,
      "learning_rate": 1.9999480295428276e-06,
      "loss": 1.4642,
      "step": 1931
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.261417138349398,
      "learning_rate": 1.9999479754215016e-06,
      "loss": 1.4865,
      "step": 1932
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.618981173105002,
      "learning_rate": 1.999947921272011e-06,
      "loss": 1.495,
      "step": 1933
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.958757287653689,
      "learning_rate": 1.9999478670943546e-06,
      "loss": 1.4359,
      "step": 1934
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.930067872156233,
      "learning_rate": 1.9999478128885332e-06,
      "loss": 1.4284,
      "step": 1935
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.470181062731635,
      "learning_rate": 1.9999477586545468e-06,
      "loss": 1.2989,
      "step": 1936
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.127129338294685,
      "learning_rate": 1.999947704392395e-06,
      "loss": 1.443,
      "step": 1937
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.127632526299122,
      "learning_rate": 1.999947650102078e-06,
      "loss": 1.3732,
      "step": 1938
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.166443772114098,
      "learning_rate": 1.9999475957835958e-06,
      "loss": 1.4847,
      "step": 1939
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.096431897023338,
      "learning_rate": 1.999947541436949e-06,
      "loss": 1.4193,
      "step": 1940
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.836191132604963,
      "learning_rate": 1.999947487062136e-06,
      "loss": 1.4965,
      "step": 1941
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.925828422364088,
      "learning_rate": 1.9999474326591583e-06,
      "loss": 1.4444,
      "step": 1942
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.667390491716781,
      "learning_rate": 1.999947378228015e-06,
      "loss": 1.5763,
      "step": 1943
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.279038785770578,
      "learning_rate": 1.9999473237687073e-06,
      "loss": 1.5275,
      "step": 1944
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.641029202907494,
      "learning_rate": 1.999947269281234e-06,
      "loss": 1.5775,
      "step": 1945
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.332390674887097,
      "learning_rate": 1.9999472147655955e-06,
      "loss": 1.5664,
      "step": 1946
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.622675952825197,
      "learning_rate": 1.999947160221792e-06,
      "loss": 1.4099,
      "step": 1947
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.613192639788906,
      "learning_rate": 1.9999471056498227e-06,
      "loss": 1.4579,
      "step": 1948
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.879403704365523,
      "learning_rate": 1.999947051049689e-06,
      "loss": 1.4041,
      "step": 1949
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.482648258267666,
      "learning_rate": 1.9999469964213895e-06,
      "loss": 1.317,
      "step": 1950
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.108898563911747,
      "learning_rate": 1.999946941764925e-06,
      "loss": 1.4241,
      "step": 1951
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.568831706403162,
      "learning_rate": 1.9999468870802954e-06,
      "loss": 1.5296,
      "step": 1952
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.786414340270627,
      "learning_rate": 1.9999468323675007e-06,
      "loss": 1.5818,
      "step": 1953
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.78919679188355,
      "learning_rate": 1.999946777626541e-06,
      "loss": 1.2252,
      "step": 1954
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.524987858157338,
      "learning_rate": 1.9999467228574154e-06,
      "loss": 1.4747,
      "step": 1955
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.948724773134819,
      "learning_rate": 1.9999466680601254e-06,
      "loss": 1.3693,
      "step": 1956
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.498995965734504,
      "learning_rate": 1.9999466132346697e-06,
      "loss": 1.3926,
      "step": 1957
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.690852300877899,
      "learning_rate": 1.9999465583810494e-06,
      "loss": 1.5302,
      "step": 1958
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.628604279451408,
      "learning_rate": 1.9999465034992636e-06,
      "loss": 1.406,
      "step": 1959
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.079437190752291,
      "learning_rate": 1.999946448589312e-06,
      "loss": 1.5523,
      "step": 1960
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.966397313206571,
      "learning_rate": 1.9999463936511966e-06,
      "loss": 1.3872,
      "step": 1961
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.652429476663953,
      "learning_rate": 1.999946338684915e-06,
      "loss": 1.4646,
      "step": 1962
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.87657353554175,
      "learning_rate": 1.999946283690468e-06,
      "loss": 1.392,
      "step": 1963
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.917643825546739,
      "learning_rate": 1.999946228667857e-06,
      "loss": 1.3851,
      "step": 1964
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.67340721917725,
      "learning_rate": 1.9999461736170802e-06,
      "loss": 1.5101,
      "step": 1965
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.39919847434518,
      "learning_rate": 1.999946118538138e-06,
      "loss": 1.3595,
      "step": 1966
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.203293308343834,
      "learning_rate": 1.999946063431031e-06,
      "loss": 1.4727,
      "step": 1967
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.678751954911345,
      "learning_rate": 1.9999460082957586e-06,
      "loss": 1.3947,
      "step": 1968
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.569173814364991,
      "learning_rate": 1.999945953132321e-06,
      "loss": 1.4007,
      "step": 1969
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.204406855731333,
      "learning_rate": 1.999945897940719e-06,
      "loss": 1.3463,
      "step": 1970
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.853776529480603,
      "learning_rate": 1.999945842720951e-06,
      "loss": 1.6373,
      "step": 1971
    },
    {
      "epoch": 0.01,
      "eval_loss": 1.6432719230651855,
      "eval_runtime": 4.6393,
      "eval_samples_per_second": 1.94,
      "eval_steps_per_second": 1.078,
      "step": 1971
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.975365230474304,
      "learning_rate": 1.9999457874730182e-06,
      "loss": 1.4259,
      "step": 1972
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.106259928022218,
      "learning_rate": 1.9999457321969203e-06,
      "loss": 1.531,
      "step": 1973
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.805401584156609,
      "learning_rate": 1.999945676892657e-06,
      "loss": 1.4088,
      "step": 1974
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.312695783521927,
      "learning_rate": 1.9999456215602288e-06,
      "loss": 1.572,
      "step": 1975
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.74462546951479,
      "learning_rate": 1.9999455661996355e-06,
      "loss": 1.2957,
      "step": 1976
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.625243006738303,
      "learning_rate": 1.9999455108108767e-06,
      "loss": 1.4034,
      "step": 1977
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.178605492022322,
      "learning_rate": 1.999945455393953e-06,
      "loss": 1.5405,
      "step": 1978
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.655718915504933,
      "learning_rate": 1.999945399948864e-06,
      "loss": 1.5224,
      "step": 1979
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.65830388317099,
      "learning_rate": 1.99994534447561e-06,
      "loss": 1.5357,
      "step": 1980
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.288686031793038,
      "learning_rate": 1.999945288974191e-06,
      "loss": 1.5392,
      "step": 1981
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.43642936625311,
      "learning_rate": 1.999945233444607e-06,
      "loss": 1.488,
      "step": 1982
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.555178280796159,
      "learning_rate": 1.9999451778868575e-06,
      "loss": 1.3151,
      "step": 1983
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.236202376421231,
      "learning_rate": 1.999945122300943e-06,
      "loss": 1.4033,
      "step": 1984
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.887741373566786,
      "learning_rate": 1.999945066686863e-06,
      "loss": 1.506,
      "step": 1985
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.9138979134608975,
      "learning_rate": 1.9999450110446183e-06,
      "loss": 1.4881,
      "step": 1986
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.542007061012194,
      "learning_rate": 1.9999449553742084e-06,
      "loss": 1.4365,
      "step": 1987
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.528814673826827,
      "learning_rate": 1.9999448996756337e-06,
      "loss": 1.4825,
      "step": 1988
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.4557691220675375,
      "learning_rate": 1.9999448439488936e-06,
      "loss": 1.4531,
      "step": 1989
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.758397625075794,
      "learning_rate": 1.999944788193988e-06,
      "loss": 1.4119,
      "step": 1990
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.583454128627948,
      "learning_rate": 1.999944732410918e-06,
      "loss": 1.4116,
      "step": 1991
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.532783999892837,
      "learning_rate": 1.999944676599682e-06,
      "loss": 1.2279,
      "step": 1992
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.906450343559313,
      "learning_rate": 1.9999446207602813e-06,
      "loss": 1.4905,
      "step": 1993
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.346899749377525,
      "learning_rate": 1.999944564892716e-06,
      "loss": 1.413,
      "step": 1994
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.523495511164613,
      "learning_rate": 1.999944508996985e-06,
      "loss": 1.5664,
      "step": 1995
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.959351356739527,
      "learning_rate": 1.999944453073089e-06,
      "loss": 1.3644,
      "step": 1996
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.871410142653787,
      "learning_rate": 1.999944397121028e-06,
      "loss": 1.396,
      "step": 1997
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.76408338524292,
      "learning_rate": 1.9999443411408018e-06,
      "loss": 1.4451,
      "step": 1998
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.498293300971013,
      "learning_rate": 1.9999442851324104e-06,
      "loss": 1.397,
      "step": 1999
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.9463867469913,
      "learning_rate": 1.999944229095854e-06,
      "loss": 1.459,
      "step": 2000
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.4931004236648775,
      "learning_rate": 1.9999441730311324e-06,
      "loss": 1.4252,
      "step": 2001
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.718233309528229,
      "learning_rate": 1.999944116938246e-06,
      "loss": 1.4942,
      "step": 2002
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.889040911882571,
      "learning_rate": 1.9999440608171944e-06,
      "loss": 1.5412,
      "step": 2003
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.014283756243838,
      "learning_rate": 1.9999440046679775e-06,
      "loss": 1.5232,
      "step": 2004
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.395377515573,
      "learning_rate": 1.999943948490596e-06,
      "loss": 1.3687,
      "step": 2005
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.1324825308243005,
      "learning_rate": 1.9999438922850487e-06,
      "loss": 1.459,
      "step": 2006
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.302528839998102,
      "learning_rate": 1.9999438360513364e-06,
      "loss": 1.1898,
      "step": 2007
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.921387927464629,
      "learning_rate": 1.9999437797894595e-06,
      "loss": 1.4922,
      "step": 2008
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.66789838005482,
      "learning_rate": 1.9999437234994174e-06,
      "loss": 1.2908,
      "step": 2009
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.466483592815993,
      "learning_rate": 1.99994366718121e-06,
      "loss": 1.3559,
      "step": 2010
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.989407567640141,
      "learning_rate": 1.9999436108348375e-06,
      "loss": 1.3186,
      "step": 2011
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.7584380055697135,
      "learning_rate": 1.9999435544602996e-06,
      "loss": 1.481,
      "step": 2012
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.410113901804026,
      "learning_rate": 1.999943498057597e-06,
      "loss": 1.4516,
      "step": 2013
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.145181164985424,
      "learning_rate": 1.99994344162673e-06,
      "loss": 1.2735,
      "step": 2014
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.997375983666718,
      "learning_rate": 1.9999433851676967e-06,
      "loss": 1.5026,
      "step": 2015
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.501900393004941,
      "learning_rate": 1.9999433286804992e-06,
      "loss": 1.5178,
      "step": 2016
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.183628468382797,
      "learning_rate": 1.9999432721651362e-06,
      "loss": 1.5511,
      "step": 2017
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.615198435575112,
      "learning_rate": 1.999943215621608e-06,
      "loss": 1.4319,
      "step": 2018
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.91098537314749,
      "learning_rate": 1.999943159049915e-06,
      "loss": 1.5055,
      "step": 2019
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.533623257731385,
      "learning_rate": 1.999943102450057e-06,
      "loss": 1.3607,
      "step": 2020
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.823264533879004,
      "learning_rate": 1.9999430458220335e-06,
      "loss": 1.3773,
      "step": 2021
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.610357246741694,
      "learning_rate": 1.9999429891658453e-06,
      "loss": 1.4646,
      "step": 2022
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.279580425295216,
      "learning_rate": 1.999942932481492e-06,
      "loss": 1.3543,
      "step": 2023
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.920902713497585,
      "learning_rate": 1.9999428757689737e-06,
      "loss": 1.4141,
      "step": 2024
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.301852036011902,
      "learning_rate": 1.99994281902829e-06,
      "loss": 1.5213,
      "step": 2025
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.135724505311549,
      "learning_rate": 1.9999427622594415e-06,
      "loss": 1.5948,
      "step": 2026
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.852010935690981,
      "learning_rate": 1.9999427054624282e-06,
      "loss": 1.4197,
      "step": 2027
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.9165669204888625,
      "learning_rate": 1.9999426486372494e-06,
      "loss": 1.5093,
      "step": 2028
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.575212660119817,
      "learning_rate": 1.999942591783906e-06,
      "loss": 1.3396,
      "step": 2029
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.259797384037423,
      "learning_rate": 1.9999425349023967e-06,
      "loss": 1.4153,
      "step": 2030
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.548169399634219,
      "learning_rate": 1.999942477992723e-06,
      "loss": 1.3866,
      "step": 2031
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.282143331135165,
      "learning_rate": 1.999942421054884e-06,
      "loss": 1.4014,
      "step": 2032
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.574333910357711,
      "learning_rate": 1.99994236408888e-06,
      "loss": 1.3933,
      "step": 2033
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.629247995921187,
      "learning_rate": 1.9999423070947114e-06,
      "loss": 1.5145,
      "step": 2034
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.71772592571776,
      "learning_rate": 1.999942250072377e-06,
      "loss": 1.353,
      "step": 2035
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.908977840581366,
      "learning_rate": 1.999942193021878e-06,
      "loss": 1.5628,
      "step": 2036
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.021790328358759,
      "learning_rate": 1.9999421359432137e-06,
      "loss": 1.4148,
      "step": 2037
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.176490897391262,
      "learning_rate": 1.999942078836385e-06,
      "loss": 1.586,
      "step": 2038
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.99627582571255,
      "learning_rate": 1.9999420217013907e-06,
      "loss": 1.5591,
      "step": 2039
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.02431904656082,
      "learning_rate": 1.9999419645382313e-06,
      "loss": 1.4402,
      "step": 2040
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.713468056290315,
      "learning_rate": 1.999941907346907e-06,
      "loss": 1.339,
      "step": 2041
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.7923489185700126,
      "learning_rate": 1.9999418501274176e-06,
      "loss": 1.4056,
      "step": 2042
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.776165026219362,
      "learning_rate": 1.9999417928797632e-06,
      "loss": 1.5542,
      "step": 2043
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.990681472067374,
      "learning_rate": 1.999941735603944e-06,
      "loss": 1.3998,
      "step": 2044
    },
    {
      "epoch": 0.01,
      "eval_loss": 1.6361159086227417,
      "eval_runtime": 4.6533,
      "eval_samples_per_second": 1.934,
      "eval_steps_per_second": 1.075,
      "step": 2044
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.903134049101306,
      "learning_rate": 1.9999416782999592e-06,
      "loss": 1.4561,
      "step": 2045
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.5369642093418285,
      "learning_rate": 1.99994162096781e-06,
      "loss": 1.4306,
      "step": 2046
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.674123365349024,
      "learning_rate": 1.9999415636074956e-06,
      "loss": 1.397,
      "step": 2047
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.020382697207387,
      "learning_rate": 1.9999415062190157e-06,
      "loss": 1.3635,
      "step": 2048
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.402239102827217,
      "learning_rate": 1.9999414488023716e-06,
      "loss": 1.3843,
      "step": 2049
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.367641687630019,
      "learning_rate": 1.999941391357562e-06,
      "loss": 1.4433,
      "step": 2050
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.567561738037097,
      "learning_rate": 1.999941333884587e-06,
      "loss": 1.4309,
      "step": 2051
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.525282756054969,
      "learning_rate": 1.9999412763834475e-06,
      "loss": 1.3293,
      "step": 2052
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.908627704395454,
      "learning_rate": 1.999941218854143e-06,
      "loss": 1.3566,
      "step": 2053
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.432919107784135,
      "learning_rate": 1.999941161296673e-06,
      "loss": 1.2394,
      "step": 2054
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.778341393380081,
      "learning_rate": 1.9999411037110387e-06,
      "loss": 1.3844,
      "step": 2055
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.954440841765967,
      "learning_rate": 1.9999410460972387e-06,
      "loss": 1.4503,
      "step": 2056
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.758941261308998,
      "learning_rate": 1.999940988455274e-06,
      "loss": 1.4572,
      "step": 2057
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.564392211755666,
      "learning_rate": 1.9999409307851446e-06,
      "loss": 1.1965,
      "step": 2058
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.725945941216895,
      "learning_rate": 1.9999408730868497e-06,
      "loss": 1.3813,
      "step": 2059
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.292635293917362,
      "learning_rate": 1.9999408153603897e-06,
      "loss": 1.4697,
      "step": 2060
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.371826291437516,
      "learning_rate": 1.9999407576057654e-06,
      "loss": 1.4433,
      "step": 2061
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.624824451917616,
      "learning_rate": 1.9999406998229756e-06,
      "loss": 1.5372,
      "step": 2062
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.828294055435367,
      "learning_rate": 1.9999406420120206e-06,
      "loss": 1.4827,
      "step": 2063
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.927358950382133,
      "learning_rate": 1.999940584172901e-06,
      "loss": 1.4385,
      "step": 2064
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.4353174579637775,
      "learning_rate": 1.9999405263056162e-06,
      "loss": 1.4795,
      "step": 2065
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.675940989803078,
      "learning_rate": 1.9999404684101663e-06,
      "loss": 1.4257,
      "step": 2066
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.0245884345574225,
      "learning_rate": 1.9999404104865518e-06,
      "loss": 1.2672,
      "step": 2067
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.1119365663762055,
      "learning_rate": 1.999940352534772e-06,
      "loss": 1.251,
      "step": 2068
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.0121477273349315,
      "learning_rate": 1.9999402945548273e-06,
      "loss": 1.426,
      "step": 2069
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.128014707129984,
      "learning_rate": 1.9999402365467174e-06,
      "loss": 1.3379,
      "step": 2070
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.862057104969277,
      "learning_rate": 1.9999401785104428e-06,
      "loss": 1.424,
      "step": 2071
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.873753921801018,
      "learning_rate": 1.999940120446003e-06,
      "loss": 1.3945,
      "step": 2072
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.1012581647028945,
      "learning_rate": 1.999940062353398e-06,
      "loss": 1.5296,
      "step": 2073
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.6157521798841055,
      "learning_rate": 1.9999400042326282e-06,
      "loss": 1.2629,
      "step": 2074
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.803622065399791,
      "learning_rate": 1.999939946083694e-06,
      "loss": 1.408,
      "step": 2075
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.740381722750265,
      "learning_rate": 1.9999398879065943e-06,
      "loss": 1.3605,
      "step": 2076
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.831139523992052,
      "learning_rate": 1.99993982970133e-06,
      "loss": 1.2995,
      "step": 2077
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.829459069984273,
      "learning_rate": 1.9999397714679002e-06,
      "loss": 1.3909,
      "step": 2078
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.827491011717043,
      "learning_rate": 1.9999397132063055e-06,
      "loss": 1.559,
      "step": 2079
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.947759137746265,
      "learning_rate": 1.9999396549165457e-06,
      "loss": 1.4971,
      "step": 2080
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.739533384237727,
      "learning_rate": 1.9999395965986212e-06,
      "loss": 1.3768,
      "step": 2081
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.339254038218756,
      "learning_rate": 1.9999395382525316e-06,
      "loss": 1.4479,
      "step": 2082
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.537488781200089,
      "learning_rate": 1.9999394798782773e-06,
      "loss": 1.3548,
      "step": 2083
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.7315607432035005,
      "learning_rate": 1.999939421475858e-06,
      "loss": 1.3923,
      "step": 2084
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.154239758619694,
      "learning_rate": 1.9999393630452733e-06,
      "loss": 1.3989,
      "step": 2085
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.535664302600297,
      "learning_rate": 1.999939304586524e-06,
      "loss": 1.3719,
      "step": 2086
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.655332915728781,
      "learning_rate": 1.9999392460996098e-06,
      "loss": 1.3278,
      "step": 2087
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.806855739816848,
      "learning_rate": 1.9999391875845303e-06,
      "loss": 1.5083,
      "step": 2088
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.213130734185716,
      "learning_rate": 1.999939129041286e-06,
      "loss": 1.4663,
      "step": 2089
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.229568233154151,
      "learning_rate": 1.999939070469877e-06,
      "loss": 1.3439,
      "step": 2090
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.086635517733998,
      "learning_rate": 1.999939011870303e-06,
      "loss": 1.2555,
      "step": 2091
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.727112810450526,
      "learning_rate": 1.999938953242564e-06,
      "loss": 1.4361,
      "step": 2092
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.052258554851617,
      "learning_rate": 1.9999388945866596e-06,
      "loss": 1.3848,
      "step": 2093
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.88161487935287,
      "learning_rate": 1.9999388359025903e-06,
      "loss": 1.4832,
      "step": 2094
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.6739895014787685,
      "learning_rate": 1.9999387771903563e-06,
      "loss": 1.478,
      "step": 2095
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.728323614320543,
      "learning_rate": 1.9999387184499576e-06,
      "loss": 1.5456,
      "step": 2096
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.765256915622586,
      "learning_rate": 1.999938659681394e-06,
      "loss": 1.407,
      "step": 2097
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.431982504309138,
      "learning_rate": 1.999938600884665e-06,
      "loss": 1.2764,
      "step": 2098
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.902178089633812,
      "learning_rate": 1.999938542059771e-06,
      "loss": 1.5341,
      "step": 2099
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.510226095063566,
      "learning_rate": 1.999938483206712e-06,
      "loss": 1.3526,
      "step": 2100
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.074107216841095,
      "learning_rate": 1.9999384243254887e-06,
      "loss": 1.3247,
      "step": 2101
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.667756343723743,
      "learning_rate": 1.9999383654160997e-06,
      "loss": 1.4049,
      "step": 2102
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.752073321972883,
      "learning_rate": 1.999938306478546e-06,
      "loss": 1.4703,
      "step": 2103
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.742868907788746,
      "learning_rate": 1.9999382475128277e-06,
      "loss": 1.2977,
      "step": 2104
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.886382117755776,
      "learning_rate": 1.999938188518944e-06,
      "loss": 1.4292,
      "step": 2105
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.781559805396283,
      "learning_rate": 1.999938129496896e-06,
      "loss": 1.3088,
      "step": 2106
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.593766296713237,
      "learning_rate": 1.9999380704466828e-06,
      "loss": 1.3449,
      "step": 2107
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.524007597424321,
      "learning_rate": 1.9999380113683048e-06,
      "loss": 1.4613,
      "step": 2108
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.622852323760914,
      "learning_rate": 1.9999379522617613e-06,
      "loss": 1.4825,
      "step": 2109
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.616671774319379,
      "learning_rate": 1.999937893127053e-06,
      "loss": 1.5888,
      "step": 2110
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.481567608026818,
      "learning_rate": 1.99993783396418e-06,
      "loss": 1.4353,
      "step": 2111
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.602477850579874,
      "learning_rate": 1.9999377747731426e-06,
      "loss": 1.3397,
      "step": 2112
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.81449808655875,
      "learning_rate": 1.9999377155539395e-06,
      "loss": 1.4046,
      "step": 2113
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.073953878891228,
      "learning_rate": 1.9999376563065716e-06,
      "loss": 1.4595,
      "step": 2114
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.743457594007675,
      "learning_rate": 1.999937597031039e-06,
      "loss": 1.3975,
      "step": 2115
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.530772105704062,
      "learning_rate": 1.9999375377273415e-06,
      "loss": 1.5028,
      "step": 2116
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.257434903260648,
      "learning_rate": 1.999937478395479e-06,
      "loss": 1.4187,
      "step": 2117
    },
    {
      "epoch": 0.01,
      "eval_loss": 1.6335363388061523,
      "eval_runtime": 4.6337,
      "eval_samples_per_second": 1.942,
      "eval_steps_per_second": 1.079,
      "step": 2117
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.722957307355561,
      "learning_rate": 1.9999374190354517e-06,
      "loss": 1.4206,
      "step": 2118
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.473561914847432,
      "learning_rate": 1.999937359647259e-06,
      "loss": 1.3537,
      "step": 2119
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.580086304567218,
      "learning_rate": 1.999937300230902e-06,
      "loss": 1.3944,
      "step": 2120
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.245944086564923,
      "learning_rate": 1.9999372407863796e-06,
      "loss": 1.4074,
      "step": 2121
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.676469646761249,
      "learning_rate": 1.999937181313693e-06,
      "loss": 1.4051,
      "step": 2122
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.922681086761417,
      "learning_rate": 1.9999371218128408e-06,
      "loss": 1.4609,
      "step": 2123
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.654470497689943,
      "learning_rate": 1.999937062283824e-06,
      "loss": 1.3944,
      "step": 2124
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.961121386583948,
      "learning_rate": 1.9999370027266423e-06,
      "loss": 1.3877,
      "step": 2125
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.590317374681009,
      "learning_rate": 1.9999369431412957e-06,
      "loss": 1.3661,
      "step": 2126
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.530242304003091,
      "learning_rate": 1.9999368835277843e-06,
      "loss": 1.4709,
      "step": 2127
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.619061402439089,
      "learning_rate": 1.999936823886108e-06,
      "loss": 1.3906,
      "step": 2128
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.012913828479663,
      "learning_rate": 1.9999367642162662e-06,
      "loss": 1.5088,
      "step": 2129
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.979952662806363,
      "learning_rate": 1.99993670451826e-06,
      "loss": 1.3039,
      "step": 2130
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.5714181931648445,
      "learning_rate": 1.999936644792089e-06,
      "loss": 1.3549,
      "step": 2131
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.770544347901491,
      "learning_rate": 1.999936585037753e-06,
      "loss": 1.3643,
      "step": 2132
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.401807178237035,
      "learning_rate": 1.999936525255252e-06,
      "loss": 1.2191,
      "step": 2133
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.894088878558767,
      "learning_rate": 1.999936465444586e-06,
      "loss": 1.5738,
      "step": 2134
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.829142286770646,
      "learning_rate": 1.9999364056057555e-06,
      "loss": 1.5857,
      "step": 2135
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.585395214505738,
      "learning_rate": 1.99993634573876e-06,
      "loss": 1.4653,
      "step": 2136
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.752105085256273,
      "learning_rate": 1.9999362858435994e-06,
      "loss": 1.5093,
      "step": 2137
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.567527923711839,
      "learning_rate": 1.9999362259202743e-06,
      "loss": 1.3605,
      "step": 2138
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.928571238482233,
      "learning_rate": 1.999936165968784e-06,
      "loss": 1.4562,
      "step": 2139
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.328826801769719,
      "learning_rate": 1.9999361059891288e-06,
      "loss": 1.347,
      "step": 2140
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.886917145275371,
      "learning_rate": 1.9999360459813087e-06,
      "loss": 1.4164,
      "step": 2141
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.291684965837117,
      "learning_rate": 1.999935985945324e-06,
      "loss": 1.4878,
      "step": 2142
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.683286308224392,
      "learning_rate": 1.999935925881174e-06,
      "loss": 1.4892,
      "step": 2143
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.5644324951779875,
      "learning_rate": 1.9999358657888597e-06,
      "loss": 1.4277,
      "step": 2144
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.719071022798878,
      "learning_rate": 1.99993580566838e-06,
      "loss": 1.3753,
      "step": 2145
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.384593412220494,
      "learning_rate": 1.9999357455197357e-06,
      "loss": 1.4575,
      "step": 2146
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.49104950913911,
      "learning_rate": 1.999935685342927e-06,
      "loss": 1.4833,
      "step": 2147
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.386345133556867,
      "learning_rate": 1.9999356251379525e-06,
      "loss": 1.2825,
      "step": 2148
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.084357385014738,
      "learning_rate": 1.999935564904814e-06,
      "loss": 1.5971,
      "step": 2149
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.483804488981741,
      "learning_rate": 1.99993550464351e-06,
      "loss": 1.3572,
      "step": 2150
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.9800344533105045,
      "learning_rate": 1.999935444354042e-06,
      "loss": 1.4381,
      "step": 2151
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.2020100481802025,
      "learning_rate": 1.999935384036408e-06,
      "loss": 1.5282,
      "step": 2152
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.513002376188213,
      "learning_rate": 1.9999353236906097e-06,
      "loss": 1.4226,
      "step": 2153
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.271851960236858,
      "learning_rate": 1.9999352633166466e-06,
      "loss": 1.4555,
      "step": 2154
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.505644532012899,
      "learning_rate": 1.9999352029145187e-06,
      "loss": 1.3634,
      "step": 2155
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.878236341705409,
      "learning_rate": 1.9999351424842258e-06,
      "loss": 1.3889,
      "step": 2156
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.398260948524359,
      "learning_rate": 1.999935082025768e-06,
      "loss": 1.2472,
      "step": 2157
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.83937816831458,
      "learning_rate": 1.9999350215391454e-06,
      "loss": 1.3623,
      "step": 2158
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.865952677064821,
      "learning_rate": 1.999934961024358e-06,
      "loss": 1.4206,
      "step": 2159
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.5376685001437,
      "learning_rate": 1.9999349004814058e-06,
      "loss": 1.4301,
      "step": 2160
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.823420924569888,
      "learning_rate": 1.999934839910289e-06,
      "loss": 1.3083,
      "step": 2161
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.854880352037673,
      "learning_rate": 1.999934779311007e-06,
      "loss": 1.4037,
      "step": 2162
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.888876031412537,
      "learning_rate": 1.9999347186835603e-06,
      "loss": 1.4077,
      "step": 2163
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.311155169710946,
      "learning_rate": 1.9999346580279485e-06,
      "loss": 1.2698,
      "step": 2164
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.665174656087581,
      "learning_rate": 1.999934597344172e-06,
      "loss": 1.3712,
      "step": 2165
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.162416137409393,
      "learning_rate": 1.999934536632231e-06,
      "loss": 1.4851,
      "step": 2166
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.018889909992565,
      "learning_rate": 1.9999344758921247e-06,
      "loss": 1.452,
      "step": 2167
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.793825358446103,
      "learning_rate": 1.999934415123854e-06,
      "loss": 1.4716,
      "step": 2168
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.377258615565779,
      "learning_rate": 1.999934354327418e-06,
      "loss": 1.3063,
      "step": 2169
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.656539358484254,
      "learning_rate": 1.9999342935028177e-06,
      "loss": 1.3844,
      "step": 2170
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.869287289839831,
      "learning_rate": 1.9999342326500522e-06,
      "loss": 1.44,
      "step": 2171
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.642787385567338,
      "learning_rate": 1.999934171769122e-06,
      "loss": 1.3566,
      "step": 2172
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8759975771352755,
      "learning_rate": 1.999934110860027e-06,
      "loss": 1.4363,
      "step": 2173
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.319719105020677,
      "learning_rate": 1.999934049922767e-06,
      "loss": 1.3242,
      "step": 2174
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.482396020182398,
      "learning_rate": 1.9999339889573426e-06,
      "loss": 1.5786,
      "step": 2175
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8751202038012265,
      "learning_rate": 1.999933927963753e-06,
      "loss": 1.5066,
      "step": 2176
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.557230982959098,
      "learning_rate": 1.9999338669419984e-06,
      "loss": 1.4745,
      "step": 2177
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.424760214415525,
      "learning_rate": 1.9999338058920797e-06,
      "loss": 1.3904,
      "step": 2178
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.892348644844545,
      "learning_rate": 1.999933744813996e-06,
      "loss": 1.4981,
      "step": 2179
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.751102176031589,
      "learning_rate": 1.999933683707747e-06,
      "loss": 1.4348,
      "step": 2180
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.242736776599296,
      "learning_rate": 1.9999336225733336e-06,
      "loss": 1.5039,
      "step": 2181
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.817718856225954,
      "learning_rate": 1.9999335614107553e-06,
      "loss": 1.4643,
      "step": 2182
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8671512340556315,
      "learning_rate": 1.999933500220012e-06,
      "loss": 1.4266,
      "step": 2183
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.643685486109339,
      "learning_rate": 1.999933439001104e-06,
      "loss": 1.5044,
      "step": 2184
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.812633898353339,
      "learning_rate": 1.9999333777540312e-06,
      "loss": 1.5403,
      "step": 2185
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.5061691305089235,
      "learning_rate": 1.9999333164787937e-06,
      "loss": 1.2902,
      "step": 2186
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.888665712532743,
      "learning_rate": 1.9999332551753915e-06,
      "loss": 1.4128,
      "step": 2187
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.587451641455223,
      "learning_rate": 1.9999331938438245e-06,
      "loss": 1.4712,
      "step": 2188
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.71997025307305,
      "learning_rate": 1.9999331324840925e-06,
      "loss": 1.3803,
      "step": 2189
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.756832246172219,
      "learning_rate": 1.9999330710961957e-06,
      "loss": 1.4978,
      "step": 2190
    },
    {
      "epoch": 0.01,
      "eval_loss": 1.6319785118103027,
      "eval_runtime": 4.6267,
      "eval_samples_per_second": 1.945,
      "eval_steps_per_second": 1.081,
      "step": 2190
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.050699124286467,
      "learning_rate": 1.9999330096801343e-06,
      "loss": 1.4176,
      "step": 2191
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.898809642977276,
      "learning_rate": 1.9999329482359078e-06,
      "loss": 1.3222,
      "step": 2192
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.704754463426861,
      "learning_rate": 1.999932886763517e-06,
      "loss": 1.3086,
      "step": 2193
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.924235736627156,
      "learning_rate": 1.999932825262961e-06,
      "loss": 1.4768,
      "step": 2194
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.042352873967065,
      "learning_rate": 1.9999327637342404e-06,
      "loss": 1.6917,
      "step": 2195
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.9553596330060925,
      "learning_rate": 1.999932702177355e-06,
      "loss": 1.4411,
      "step": 2196
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.963498435657598,
      "learning_rate": 1.9999326405923047e-06,
      "loss": 1.395,
      "step": 2197
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.506052602096603,
      "learning_rate": 1.9999325789790895e-06,
      "loss": 1.247,
      "step": 2198
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.278628706173458,
      "learning_rate": 1.99993251733771e-06,
      "loss": 1.4507,
      "step": 2199
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.53129051287259,
      "learning_rate": 1.9999324556681656e-06,
      "loss": 1.4194,
      "step": 2200
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.821670837144131,
      "learning_rate": 1.9999323939704564e-06,
      "loss": 1.3976,
      "step": 2201
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.011321422146149,
      "learning_rate": 1.999932332244582e-06,
      "loss": 1.6157,
      "step": 2202
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.951896374917011,
      "learning_rate": 1.9999322704905435e-06,
      "loss": 1.3892,
      "step": 2203
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.653655462890473,
      "learning_rate": 1.99993220870834e-06,
      "loss": 1.4186,
      "step": 2204
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.9073760614579784,
      "learning_rate": 1.9999321468979714e-06,
      "loss": 1.3918,
      "step": 2205
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.615946898949113,
      "learning_rate": 1.9999320850594383e-06,
      "loss": 1.4513,
      "step": 2206
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.781700938808737,
      "learning_rate": 1.99993202319274e-06,
      "loss": 1.4084,
      "step": 2207
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.919243252625909,
      "learning_rate": 1.9999319612978777e-06,
      "loss": 1.3552,
      "step": 2208
    },
    {
      "epoch": 0.01,
      "grad_norm": 11.350832956279241,
      "learning_rate": 1.9999318993748505e-06,
      "loss": 1.6017,
      "step": 2209
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.273831701204187,
      "learning_rate": 1.9999318374236587e-06,
      "loss": 1.4666,
      "step": 2210
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.29032597581334,
      "learning_rate": 1.9999317754443013e-06,
      "loss": 1.6475,
      "step": 2211
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.679461340547445,
      "learning_rate": 1.99993171343678e-06,
      "loss": 1.4847,
      "step": 2212
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.881691601287856,
      "learning_rate": 1.9999316514010932e-06,
      "loss": 1.6707,
      "step": 2213
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.880588356691005,
      "learning_rate": 1.999931589337242e-06,
      "loss": 1.4591,
      "step": 2214
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.198012469615865,
      "learning_rate": 1.999931527245226e-06,
      "loss": 1.557,
      "step": 2215
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.599284399092693,
      "learning_rate": 1.9999314651250456e-06,
      "loss": 1.4037,
      "step": 2216
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.55023727427313,
      "learning_rate": 1.9999314029767004e-06,
      "loss": 1.3229,
      "step": 2217
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.916366612795883,
      "learning_rate": 1.99993134080019e-06,
      "loss": 1.4152,
      "step": 2218
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.272257622372692,
      "learning_rate": 1.9999312785955153e-06,
      "loss": 1.2069,
      "step": 2219
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.793645780553963,
      "learning_rate": 1.9999312163626756e-06,
      "loss": 1.3437,
      "step": 2220
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.765355672753897,
      "learning_rate": 1.9999311541016713e-06,
      "loss": 1.3828,
      "step": 2221
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.609573170614245,
      "learning_rate": 1.9999310918125023e-06,
      "loss": 1.3513,
      "step": 2222
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.705044702022353,
      "learning_rate": 1.9999310294951686e-06,
      "loss": 1.4292,
      "step": 2223
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.549299970897536,
      "learning_rate": 1.99993096714967e-06,
      "loss": 1.3031,
      "step": 2224
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.685926142426188,
      "learning_rate": 1.9999309047760067e-06,
      "loss": 1.4263,
      "step": 2225
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.892295754105913,
      "learning_rate": 1.999930842374179e-06,
      "loss": 1.5274,
      "step": 2226
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.723207695873954,
      "learning_rate": 1.999930779944186e-06,
      "loss": 1.2833,
      "step": 2227
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.43058673722676,
      "learning_rate": 1.9999307174860284e-06,
      "loss": 1.3916,
      "step": 2228
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.442144662758888,
      "learning_rate": 1.9999306549997065e-06,
      "loss": 1.3429,
      "step": 2229
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.263310396091934,
      "learning_rate": 1.9999305924852196e-06,
      "loss": 1.4077,
      "step": 2230
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.693930449937829,
      "learning_rate": 1.999930529942568e-06,
      "loss": 1.3103,
      "step": 2231
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.394685919756182,
      "learning_rate": 1.9999304673717515e-06,
      "loss": 1.368,
      "step": 2232
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.808214571784509,
      "learning_rate": 1.9999304047727705e-06,
      "loss": 1.5119,
      "step": 2233
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5647319860721165,
      "learning_rate": 1.9999303421456247e-06,
      "loss": 1.4842,
      "step": 2234
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.270790805945803,
      "learning_rate": 1.9999302794903143e-06,
      "loss": 1.5416,
      "step": 2235
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.115712741622331,
      "learning_rate": 1.999930216806839e-06,
      "loss": 1.6271,
      "step": 2236
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.867749398037803,
      "learning_rate": 1.9999301540951993e-06,
      "loss": 1.4521,
      "step": 2237
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.759697223017583,
      "learning_rate": 1.999930091355395e-06,
      "loss": 1.4084,
      "step": 2238
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.910296412113315,
      "learning_rate": 1.999930028587425e-06,
      "loss": 1.3314,
      "step": 2239
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.978260498852945,
      "learning_rate": 1.9999299657912913e-06,
      "loss": 1.389,
      "step": 2240
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.148168402542014,
      "learning_rate": 1.9999299029669927e-06,
      "loss": 1.376,
      "step": 2241
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.290115867763426,
      "learning_rate": 1.9999298401145294e-06,
      "loss": 1.6375,
      "step": 2242
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.97467209793556,
      "learning_rate": 1.999929777233901e-06,
      "loss": 1.2882,
      "step": 2243
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.344906181950449,
      "learning_rate": 1.9999297143251083e-06,
      "loss": 1.4118,
      "step": 2244
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.308318004204632,
      "learning_rate": 1.999929651388151e-06,
      "loss": 1.3563,
      "step": 2245
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.568152176449807,
      "learning_rate": 1.999929588423029e-06,
      "loss": 1.5212,
      "step": 2246
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.601269314744774,
      "learning_rate": 1.9999295254297417e-06,
      "loss": 1.4808,
      "step": 2247
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.854159941498488,
      "learning_rate": 1.9999294624082903e-06,
      "loss": 1.3002,
      "step": 2248
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.718540879766618,
      "learning_rate": 1.9999293993586737e-06,
      "loss": 1.4872,
      "step": 2249
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.7452734133361325,
      "learning_rate": 1.9999293362808933e-06,
      "loss": 1.3559,
      "step": 2250
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.526122852996041,
      "learning_rate": 1.9999292731749473e-06,
      "loss": 1.4187,
      "step": 2251
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.963529673562205,
      "learning_rate": 1.999929210040837e-06,
      "loss": 1.4198,
      "step": 2252
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.77049575346727,
      "learning_rate": 1.999929146878562e-06,
      "loss": 1.2526,
      "step": 2253
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.6993939688150395,
      "learning_rate": 1.999929083688122e-06,
      "loss": 1.5895,
      "step": 2254
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.118619735033707,
      "learning_rate": 1.999929020469518e-06,
      "loss": 1.429,
      "step": 2255
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.6599775134774815,
      "learning_rate": 1.999928957222749e-06,
      "loss": 1.5317,
      "step": 2256
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.7355804324143405,
      "learning_rate": 1.9999288939478153e-06,
      "loss": 1.4884,
      "step": 2257
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.3932795591976275,
      "learning_rate": 1.999928830644717e-06,
      "loss": 1.4564,
      "step": 2258
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.091350488878629,
      "learning_rate": 1.9999287673134542e-06,
      "loss": 1.5135,
      "step": 2259
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.7772806452872265,
      "learning_rate": 1.999928703954026e-06,
      "loss": 1.3913,
      "step": 2260
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.604236988471221,
      "learning_rate": 1.999928640566434e-06,
      "loss": 1.5703,
      "step": 2261
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.270866871119281,
      "learning_rate": 1.999928577150677e-06,
      "loss": 1.3042,
      "step": 2262
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.55553737525265,
      "learning_rate": 1.9999285137067555e-06,
      "loss": 1.3917,
      "step": 2263
    },
    {
      "epoch": 0.02,
      "eval_loss": 1.6318128108978271,
      "eval_runtime": 4.6236,
      "eval_samples_per_second": 1.947,
      "eval_steps_per_second": 1.081,
      "step": 2263
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.57569740895167,
      "learning_rate": 1.999928450234669e-06,
      "loss": 1.3676,
      "step": 2264
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.233907230202161,
      "learning_rate": 1.9999283867344178e-06,
      "loss": 1.4965,
      "step": 2265
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.36719675852659,
      "learning_rate": 1.9999283232060023e-06,
      "loss": 1.2795,
      "step": 2266
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.6618468973225005,
      "learning_rate": 1.999928259649422e-06,
      "loss": 1.5947,
      "step": 2267
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.211675232112218,
      "learning_rate": 1.999928196064677e-06,
      "loss": 1.6732,
      "step": 2268
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.691350080075693,
      "learning_rate": 1.9999281324517672e-06,
      "loss": 1.4264,
      "step": 2269
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.310562396402286,
      "learning_rate": 1.9999280688106934e-06,
      "loss": 1.3275,
      "step": 2270
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.653521933252809,
      "learning_rate": 1.999928005141454e-06,
      "loss": 1.2394,
      "step": 2271
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.251196276782816,
      "learning_rate": 1.999927941444051e-06,
      "loss": 1.4138,
      "step": 2272
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.73612278780909,
      "learning_rate": 1.9999278777184825e-06,
      "loss": 1.3869,
      "step": 2273
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.700582189189819,
      "learning_rate": 1.99992781396475e-06,
      "loss": 1.2867,
      "step": 2274
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.995811821783478,
      "learning_rate": 1.999927750182852e-06,
      "loss": 1.4574,
      "step": 2275
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.7048294811265405,
      "learning_rate": 1.99992768637279e-06,
      "loss": 1.5312,
      "step": 2276
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.015500835378081,
      "learning_rate": 1.9999276225345635e-06,
      "loss": 1.4603,
      "step": 2277
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.038403242723015,
      "learning_rate": 1.9999275586681717e-06,
      "loss": 1.4336,
      "step": 2278
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.94366068107535,
      "learning_rate": 1.9999274947736156e-06,
      "loss": 1.4457,
      "step": 2279
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.853975513709176,
      "learning_rate": 1.9999274308508953e-06,
      "loss": 1.4836,
      "step": 2280
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.250005051833288,
      "learning_rate": 1.99992736690001e-06,
      "loss": 1.4746,
      "step": 2281
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.706101925023441,
      "learning_rate": 1.9999273029209597e-06,
      "loss": 1.5153,
      "step": 2282
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.691916068442643,
      "learning_rate": 1.9999272389137453e-06,
      "loss": 1.6037,
      "step": 2283
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.891521294365354,
      "learning_rate": 1.999927174878366e-06,
      "loss": 1.4662,
      "step": 2284
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.299833570856004,
      "learning_rate": 1.9999271108148224e-06,
      "loss": 1.4442,
      "step": 2285
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.233339122045351,
      "learning_rate": 1.9999270467231143e-06,
      "loss": 1.593,
      "step": 2286
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.876412995521755,
      "learning_rate": 1.999926982603241e-06,
      "loss": 1.4954,
      "step": 2287
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.83398554670689,
      "learning_rate": 1.9999269184552033e-06,
      "loss": 1.4952,
      "step": 2288
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.659977577870166,
      "learning_rate": 1.999926854279001e-06,
      "loss": 1.5399,
      "step": 2289
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.709246532034214,
      "learning_rate": 1.9999267900746343e-06,
      "loss": 1.4852,
      "step": 2290
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.469809734473597,
      "learning_rate": 1.9999267258421028e-06,
      "loss": 1.4133,
      "step": 2291
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.470510103145094,
      "learning_rate": 1.999926661581407e-06,
      "loss": 1.393,
      "step": 2292
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.045637874654517,
      "learning_rate": 1.999926597292546e-06,
      "loss": 1.4533,
      "step": 2293
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.56360911520591,
      "learning_rate": 1.9999265329755204e-06,
      "loss": 1.4627,
      "step": 2294
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.90860484176041,
      "learning_rate": 1.9999264686303306e-06,
      "loss": 1.4674,
      "step": 2295
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.513345359016672,
      "learning_rate": 1.999926404256976e-06,
      "loss": 1.2764,
      "step": 2296
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.223838246271355,
      "learning_rate": 1.9999263398554567e-06,
      "loss": 1.5127,
      "step": 2297
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.195325189394897,
      "learning_rate": 1.999926275425773e-06,
      "loss": 1.3677,
      "step": 2298
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.198912590441455,
      "learning_rate": 1.999926210967925e-06,
      "loss": 1.3427,
      "step": 2299
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.614705914091713,
      "learning_rate": 1.999926146481912e-06,
      "loss": 1.4733,
      "step": 2300
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.975373741434217,
      "learning_rate": 1.9999260819677345e-06,
      "loss": 1.4271,
      "step": 2301
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.970231553749659,
      "learning_rate": 1.9999260174253926e-06,
      "loss": 1.52,
      "step": 2302
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.1470565052529516,
      "learning_rate": 1.9999259528548856e-06,
      "loss": 1.4938,
      "step": 2303
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.804932112180643,
      "learning_rate": 1.9999258882562143e-06,
      "loss": 1.3168,
      "step": 2304
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.117885144420734,
      "learning_rate": 1.9999258236293784e-06,
      "loss": 1.4501,
      "step": 2305
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.891670263126413,
      "learning_rate": 1.999925758974378e-06,
      "loss": 1.4911,
      "step": 2306
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.195863211639049,
      "learning_rate": 1.9999256942912132e-06,
      "loss": 1.4502,
      "step": 2307
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.2574150657434,
      "learning_rate": 1.9999256295798832e-06,
      "loss": 1.5026,
      "step": 2308
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.872600619683232,
      "learning_rate": 1.9999255648403893e-06,
      "loss": 1.5434,
      "step": 2309
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.686714043104333,
      "learning_rate": 1.9999255000727303e-06,
      "loss": 1.6448,
      "step": 2310
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.631001470345563,
      "learning_rate": 1.999925435276907e-06,
      "loss": 1.3702,
      "step": 2311
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.337508348563074,
      "learning_rate": 1.999925370452919e-06,
      "loss": 1.3578,
      "step": 2312
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.65818564635677,
      "learning_rate": 1.999925305600767e-06,
      "loss": 1.4024,
      "step": 2313
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.379822492720928,
      "learning_rate": 1.99992524072045e-06,
      "loss": 1.4887,
      "step": 2314
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.299169016401586,
      "learning_rate": 1.9999251758119684e-06,
      "loss": 1.4817,
      "step": 2315
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.225815129184746,
      "learning_rate": 1.999925110875322e-06,
      "loss": 1.4054,
      "step": 2316
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.568549504456492,
      "learning_rate": 1.999925045910511e-06,
      "loss": 1.4551,
      "step": 2317
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.431025457178463,
      "learning_rate": 1.999924980917536e-06,
      "loss": 1.4775,
      "step": 2318
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.8399275566764235,
      "learning_rate": 1.999924915896396e-06,
      "loss": 1.4959,
      "step": 2319
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.473563523649748,
      "learning_rate": 1.9999248508470912e-06,
      "loss": 1.4546,
      "step": 2320
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.740884057821344,
      "learning_rate": 1.9999247857696227e-06,
      "loss": 1.4305,
      "step": 2321
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.860828036066125,
      "learning_rate": 1.9999247206639887e-06,
      "loss": 1.4348,
      "step": 2322
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.6662009244963,
      "learning_rate": 1.999924655530191e-06,
      "loss": 1.55,
      "step": 2323
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.9234684189906925,
      "learning_rate": 1.9999245903682283e-06,
      "loss": 1.4198,
      "step": 2324
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.492554780845503,
      "learning_rate": 1.999924525178101e-06,
      "loss": 1.3904,
      "step": 2325
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.763408364164771,
      "learning_rate": 1.9999244599598095e-06,
      "loss": 1.2435,
      "step": 2326
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.2225195105349975,
      "learning_rate": 1.9999243947133532e-06,
      "loss": 1.4942,
      "step": 2327
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.06710806212036,
      "learning_rate": 1.9999243294387323e-06,
      "loss": 1.3485,
      "step": 2328
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.97215088823951,
      "learning_rate": 1.9999242641359467e-06,
      "loss": 1.4744,
      "step": 2329
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.085684880572574,
      "learning_rate": 1.999924198804997e-06,
      "loss": 1.347,
      "step": 2330
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.378163474560379,
      "learning_rate": 1.9999241334458823e-06,
      "loss": 1.3741,
      "step": 2331
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.503096219870561,
      "learning_rate": 1.999924068058603e-06,
      "loss": 1.5229,
      "step": 2332
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.36927927637507,
      "learning_rate": 1.99992400264316e-06,
      "loss": 1.207,
      "step": 2333
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.457857920180457,
      "learning_rate": 1.9999239371995517e-06,
      "loss": 1.4385,
      "step": 2334
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.6240106572076956,
      "learning_rate": 1.999923871727779e-06,
      "loss": 1.4643,
      "step": 2335
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.861127113148184,
      "learning_rate": 1.999923806227842e-06,
      "loss": 1.423,
      "step": 2336
    },
    {
      "epoch": 0.02,
      "eval_loss": 1.6273260116577148,
      "eval_runtime": 4.6368,
      "eval_samples_per_second": 1.941,
      "eval_steps_per_second": 1.078,
      "step": 2336
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.9052593670852405,
      "learning_rate": 1.99992374069974e-06,
      "loss": 1.4129,
      "step": 2337
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.35788095136664,
      "learning_rate": 1.9999236751434744e-06,
      "loss": 1.5536,
      "step": 2338
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5895028769970505,
      "learning_rate": 1.9999236095590435e-06,
      "loss": 1.3911,
      "step": 2339
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.831901014472646,
      "learning_rate": 1.999923543946448e-06,
      "loss": 1.4226,
      "step": 2340
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.898166932033894,
      "learning_rate": 1.9999234783056886e-06,
      "loss": 1.5627,
      "step": 2341
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.579327466598639,
      "learning_rate": 1.999923412636764e-06,
      "loss": 1.4576,
      "step": 2342
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.894443631125458,
      "learning_rate": 1.9999233469396754e-06,
      "loss": 1.1901,
      "step": 2343
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.691429649313104,
      "learning_rate": 1.999923281214422e-06,
      "loss": 1.6006,
      "step": 2344
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.688934834826396,
      "learning_rate": 1.999923215461004e-06,
      "loss": 1.562,
      "step": 2345
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.587544882543127,
      "learning_rate": 1.999923149679422e-06,
      "loss": 1.4466,
      "step": 2346
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.412183277646008,
      "learning_rate": 1.999923083869675e-06,
      "loss": 1.4389,
      "step": 2347
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.939665962691669,
      "learning_rate": 1.9999230180317637e-06,
      "loss": 1.3812,
      "step": 2348
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.20037573679403,
      "learning_rate": 1.9999229521656876e-06,
      "loss": 1.3061,
      "step": 2349
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.219055105472269,
      "learning_rate": 1.9999228862714473e-06,
      "loss": 1.5609,
      "step": 2350
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.346948682121618,
      "learning_rate": 1.9999228203490427e-06,
      "loss": 1.3565,
      "step": 2351
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.644800626615502,
      "learning_rate": 1.999922754398473e-06,
      "loss": 1.3957,
      "step": 2352
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.526521338106829,
      "learning_rate": 1.9999226884197394e-06,
      "loss": 1.3943,
      "step": 2353
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.860775731784836,
      "learning_rate": 1.999922622412841e-06,
      "loss": 1.5376,
      "step": 2354
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.987367583243536,
      "learning_rate": 1.999922556377778e-06,
      "loss": 1.1198,
      "step": 2355
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.734455961994287,
      "learning_rate": 1.999922490314551e-06,
      "loss": 1.5315,
      "step": 2356
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.71775855055711,
      "learning_rate": 1.999922424223159e-06,
      "loss": 1.4829,
      "step": 2357
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.340625782014353,
      "learning_rate": 1.999922358103603e-06,
      "loss": 1.3968,
      "step": 2358
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.624947314913078,
      "learning_rate": 1.999922291955882e-06,
      "loss": 1.3363,
      "step": 2359
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.617825324071575,
      "learning_rate": 1.9999222257799964e-06,
      "loss": 1.5,
      "step": 2360
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.485319822796489,
      "learning_rate": 1.999922159575947e-06,
      "loss": 1.4177,
      "step": 2361
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.850158299358528,
      "learning_rate": 1.9999220933437324e-06,
      "loss": 1.3689,
      "step": 2362
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.219052152565596,
      "learning_rate": 1.999922027083354e-06,
      "loss": 1.4092,
      "step": 2363
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.931613378755357,
      "learning_rate": 1.999921960794811e-06,
      "loss": 1.326,
      "step": 2364
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.992683247482379,
      "learning_rate": 1.9999218944781032e-06,
      "loss": 1.4233,
      "step": 2365
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.349977050699164,
      "learning_rate": 1.9999218281332308e-06,
      "loss": 1.3647,
      "step": 2366
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.701955352762745,
      "learning_rate": 1.999921761760194e-06,
      "loss": 1.4519,
      "step": 2367
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.494673397335141,
      "learning_rate": 1.999921695358993e-06,
      "loss": 1.4392,
      "step": 2368
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.7365593408579345,
      "learning_rate": 1.9999216289296273e-06,
      "loss": 1.3825,
      "step": 2369
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.856324736948551,
      "learning_rate": 1.9999215624720974e-06,
      "loss": 1.5478,
      "step": 2370
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.8131720405427645,
      "learning_rate": 1.999921495986403e-06,
      "loss": 1.4974,
      "step": 2371
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.6424157982225935,
      "learning_rate": 1.9999214294725442e-06,
      "loss": 1.5017,
      "step": 2372
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.7994216756118595,
      "learning_rate": 1.9999213629305206e-06,
      "loss": 1.3935,
      "step": 2373
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.161522599209015,
      "learning_rate": 1.9999212963603328e-06,
      "loss": 1.5953,
      "step": 2374
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.663179583938319,
      "learning_rate": 1.9999212297619806e-06,
      "loss": 1.3695,
      "step": 2375
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.992382036334836,
      "learning_rate": 1.9999211631354633e-06,
      "loss": 1.5055,
      "step": 2376
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.8753351925338695,
      "learning_rate": 1.9999210964807822e-06,
      "loss": 1.3948,
      "step": 2377
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.222563145132559,
      "learning_rate": 1.9999210297979364e-06,
      "loss": 1.3826,
      "step": 2378
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.88643245267929,
      "learning_rate": 1.9999209630869264e-06,
      "loss": 1.3931,
      "step": 2379
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.560779789666663,
      "learning_rate": 1.999920896347752e-06,
      "loss": 1.4941,
      "step": 2380
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.296411387581933,
      "learning_rate": 1.999920829580413e-06,
      "loss": 1.5771,
      "step": 2381
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.739960436369839,
      "learning_rate": 1.9999207627849093e-06,
      "loss": 1.402,
      "step": 2382
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.494606888309802,
      "learning_rate": 1.9999206959612413e-06,
      "loss": 1.3666,
      "step": 2383
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.102337146702119,
      "learning_rate": 1.999920629109409e-06,
      "loss": 1.4514,
      "step": 2384
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.2918281175179365,
      "learning_rate": 1.999920562229412e-06,
      "loss": 1.4275,
      "step": 2385
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5428032056841845,
      "learning_rate": 1.999920495321251e-06,
      "loss": 1.4875,
      "step": 2386
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.981403898596884,
      "learning_rate": 1.9999204283849253e-06,
      "loss": 1.4604,
      "step": 2387
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.656946904533867,
      "learning_rate": 1.999920361420435e-06,
      "loss": 1.3959,
      "step": 2388
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.69793455074141,
      "learning_rate": 1.9999202944277807e-06,
      "loss": 1.3798,
      "step": 2389
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.8697187270242415,
      "learning_rate": 1.9999202274069615e-06,
      "loss": 1.5877,
      "step": 2390
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.303206163342291,
      "learning_rate": 1.999920160357978e-06,
      "loss": 1.3649,
      "step": 2391
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.428043235458622,
      "learning_rate": 1.9999200932808304e-06,
      "loss": 1.4989,
      "step": 2392
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.69775284325294,
      "learning_rate": 1.999920026175518e-06,
      "loss": 1.4359,
      "step": 2393
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.468054893971967,
      "learning_rate": 1.9999199590420414e-06,
      "loss": 1.3861,
      "step": 2394
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.1096088777638675,
      "learning_rate": 1.9999198918804005e-06,
      "loss": 1.3725,
      "step": 2395
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.09309736334753,
      "learning_rate": 1.999919824690595e-06,
      "loss": 1.354,
      "step": 2396
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.061146836059262,
      "learning_rate": 1.999919757472625e-06,
      "loss": 1.4047,
      "step": 2397
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.698403918335731,
      "learning_rate": 1.999919690226491e-06,
      "loss": 1.5047,
      "step": 2398
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.722675479564567,
      "learning_rate": 1.999919622952192e-06,
      "loss": 1.455,
      "step": 2399
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.676824374912994,
      "learning_rate": 1.999919555649729e-06,
      "loss": 1.3603,
      "step": 2400
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.660790706723943,
      "learning_rate": 1.999919488319101e-06,
      "loss": 1.4138,
      "step": 2401
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.8886919642276006,
      "learning_rate": 1.9999194209603095e-06,
      "loss": 1.4662,
      "step": 2402
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.728150466755214,
      "learning_rate": 1.999919353573353e-06,
      "loss": 1.3994,
      "step": 2403
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.2893263295633,
      "learning_rate": 1.999919286158232e-06,
      "loss": 1.5503,
      "step": 2404
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.462205296842043,
      "learning_rate": 1.9999192187149468e-06,
      "loss": 1.4898,
      "step": 2405
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.171445485236358,
      "learning_rate": 1.9999191512434972e-06,
      "loss": 1.2432,
      "step": 2406
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.065808066331224,
      "learning_rate": 1.9999190837438834e-06,
      "loss": 1.6226,
      "step": 2407
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.94314214884708,
      "learning_rate": 1.999919016216105e-06,
      "loss": 1.4387,
      "step": 2408
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.321472894774982,
      "learning_rate": 1.9999189486601625e-06,
      "loss": 1.525,
      "step": 2409
    },
    {
      "epoch": 0.02,
      "eval_loss": 1.6314270496368408,
      "eval_runtime": 4.6321,
      "eval_samples_per_second": 1.943,
      "eval_steps_per_second": 1.079,
      "step": 2409
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.2667747042009365,
      "learning_rate": 1.9999188810760554e-06,
      "loss": 1.4374,
      "step": 2410
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.33459038400234,
      "learning_rate": 1.9999188134637836e-06,
      "loss": 1.5239,
      "step": 2411
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.682545029673098,
      "learning_rate": 1.9999187458233476e-06,
      "loss": 1.503,
      "step": 2412
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.067629662325673,
      "learning_rate": 1.9999186781547473e-06,
      "loss": 1.5611,
      "step": 2413
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.885518592851602,
      "learning_rate": 1.9999186104579827e-06,
      "loss": 1.399,
      "step": 2414
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.8623966665305245,
      "learning_rate": 1.9999185427330535e-06,
      "loss": 1.5801,
      "step": 2415
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.4540876235305555,
      "learning_rate": 1.9999184749799604e-06,
      "loss": 1.5821,
      "step": 2416
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.705447443475058,
      "learning_rate": 1.999918407198702e-06,
      "loss": 1.3744,
      "step": 2417
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.51004939941681,
      "learning_rate": 1.99991833938928e-06,
      "loss": 1.5278,
      "step": 2418
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.781795445026661,
      "learning_rate": 1.9999182715516937e-06,
      "loss": 1.5668,
      "step": 2419
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.4984493425238945,
      "learning_rate": 1.9999182036859427e-06,
      "loss": 1.4154,
      "step": 2420
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.6682629310048345,
      "learning_rate": 1.999918135792027e-06,
      "loss": 1.1895,
      "step": 2421
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.157730325288316,
      "learning_rate": 1.9999180678699474e-06,
      "loss": 1.3339,
      "step": 2422
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.318416120479912,
      "learning_rate": 1.999917999919703e-06,
      "loss": 1.2886,
      "step": 2423
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.782830315820668,
      "learning_rate": 1.999917931941295e-06,
      "loss": 1.3462,
      "step": 2424
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.9884978695023685,
      "learning_rate": 1.9999178639347223e-06,
      "loss": 1.4021,
      "step": 2425
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.93256932892002,
      "learning_rate": 1.999917795899985e-06,
      "loss": 1.493,
      "step": 2426
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.556244224765811,
      "learning_rate": 1.9999177278370834e-06,
      "loss": 1.4557,
      "step": 2427
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.629046540813747,
      "learning_rate": 1.999917659746018e-06,
      "loss": 1.5647,
      "step": 2428
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.073793267819167,
      "learning_rate": 1.9999175916267875e-06,
      "loss": 1.5932,
      "step": 2429
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.560810629230434,
      "learning_rate": 1.999917523479393e-06,
      "loss": 1.4904,
      "step": 2430
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.097830006203227,
      "learning_rate": 1.999917455303834e-06,
      "loss": 1.4545,
      "step": 2431
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.589161304086774,
      "learning_rate": 1.999917387100111e-06,
      "loss": 1.4455,
      "step": 2432
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.262666403698471,
      "learning_rate": 1.9999173188682232e-06,
      "loss": 1.5229,
      "step": 2433
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.509535961863751,
      "learning_rate": 1.999917250608171e-06,
      "loss": 1.4361,
      "step": 2434
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.508763347593167,
      "learning_rate": 1.999917182319955e-06,
      "loss": 1.2921,
      "step": 2435
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.40129312190739,
      "learning_rate": 1.999917114003574e-06,
      "loss": 1.3886,
      "step": 2436
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.645761685555377,
      "learning_rate": 1.9999170456590293e-06,
      "loss": 1.3333,
      "step": 2437
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.642316987637496,
      "learning_rate": 1.99991697728632e-06,
      "loss": 1.4236,
      "step": 2438
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.516269375769559,
      "learning_rate": 1.9999169088854464e-06,
      "loss": 1.4596,
      "step": 2439
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.96343766345647,
      "learning_rate": 1.9999168404564082e-06,
      "loss": 1.3929,
      "step": 2440
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.69420089914068,
      "learning_rate": 1.999916771999206e-06,
      "loss": 1.4285,
      "step": 2441
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.942122889945423,
      "learning_rate": 1.9999167035138392e-06,
      "loss": 1.4459,
      "step": 2442
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.716647864815381,
      "learning_rate": 1.9999166350003083e-06,
      "loss": 1.4375,
      "step": 2443
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.558659758531409,
      "learning_rate": 1.999916566458613e-06,
      "loss": 1.5505,
      "step": 2444
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.032420136381452,
      "learning_rate": 1.9999164978887537e-06,
      "loss": 1.3798,
      "step": 2445
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.387046415116763,
      "learning_rate": 1.9999164292907295e-06,
      "loss": 1.4069,
      "step": 2446
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.829694624857794,
      "learning_rate": 1.999916360664541e-06,
      "loss": 1.4942,
      "step": 2447
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.304283704762636,
      "learning_rate": 1.9999162920101885e-06,
      "loss": 1.3792,
      "step": 2448
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.957652008482992,
      "learning_rate": 1.9999162233276715e-06,
      "loss": 1.5595,
      "step": 2449
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.881099312639281,
      "learning_rate": 1.9999161546169903e-06,
      "loss": 1.3463,
      "step": 2450
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.55918290195136,
      "learning_rate": 1.999916085878145e-06,
      "loss": 1.3157,
      "step": 2451
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.933187617415705,
      "learning_rate": 1.999916017111135e-06,
      "loss": 1.325,
      "step": 2452
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.522893367254299,
      "learning_rate": 1.999915948315961e-06,
      "loss": 1.3697,
      "step": 2453
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.753016879907898,
      "learning_rate": 1.9999158794926227e-06,
      "loss": 1.5241,
      "step": 2454
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.738367977115928,
      "learning_rate": 1.9999158106411197e-06,
      "loss": 1.5124,
      "step": 2455
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.259436967977112,
      "learning_rate": 1.999915741761453e-06,
      "loss": 1.1845,
      "step": 2456
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.327412482017732,
      "learning_rate": 1.9999156728536214e-06,
      "loss": 1.4195,
      "step": 2457
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.815255649938258,
      "learning_rate": 1.9999156039176256e-06,
      "loss": 1.4678,
      "step": 2458
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.830224872542197,
      "learning_rate": 1.999915534953466e-06,
      "loss": 1.4407,
      "step": 2459
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.055447650223887,
      "learning_rate": 1.9999154659611416e-06,
      "loss": 1.3783,
      "step": 2460
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.582184501347047,
      "learning_rate": 1.999915396940653e-06,
      "loss": 1.4415,
      "step": 2461
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.691006151625649,
      "learning_rate": 1.999915327892e-06,
      "loss": 1.5197,
      "step": 2462
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.591425262583514,
      "learning_rate": 1.999915258815183e-06,
      "loss": 1.4104,
      "step": 2463
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.871239888750451,
      "learning_rate": 1.9999151897102016e-06,
      "loss": 1.3372,
      "step": 2464
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.330668450921265,
      "learning_rate": 1.999915120577056e-06,
      "loss": 1.4389,
      "step": 2465
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.340186565522926,
      "learning_rate": 1.999915051415746e-06,
      "loss": 1.4118,
      "step": 2466
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.261449629532745,
      "learning_rate": 1.9999149822262718e-06,
      "loss": 1.7365,
      "step": 2467
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.742851588374142,
      "learning_rate": 1.999914913008633e-06,
      "loss": 1.4563,
      "step": 2468
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.025387402777135,
      "learning_rate": 1.99991484376283e-06,
      "loss": 1.4834,
      "step": 2469
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.355185254680489,
      "learning_rate": 1.999914774488863e-06,
      "loss": 1.443,
      "step": 2470
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.520118266978153,
      "learning_rate": 1.999914705186732e-06,
      "loss": 1.2398,
      "step": 2471
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.679385582459207,
      "learning_rate": 1.9999146358564362e-06,
      "loss": 1.3546,
      "step": 2472
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.653878537687508,
      "learning_rate": 1.9999145664979764e-06,
      "loss": 1.3641,
      "step": 2473
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.288484525559772,
      "learning_rate": 1.9999144971113523e-06,
      "loss": 1.4576,
      "step": 2474
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.113711451335311,
      "learning_rate": 1.999914427696564e-06,
      "loss": 1.1885,
      "step": 2475
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.792513033095759,
      "learning_rate": 1.9999143582536113e-06,
      "loss": 1.49,
      "step": 2476
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.916268869163306,
      "learning_rate": 1.999914288782494e-06,
      "loss": 1.4821,
      "step": 2477
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.9324603436046415,
      "learning_rate": 1.999914219283213e-06,
      "loss": 1.5012,
      "step": 2478
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.283635481534406,
      "learning_rate": 1.9999141497557674e-06,
      "loss": 1.3083,
      "step": 2479
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.3789031660655935,
      "learning_rate": 1.9999140802001576e-06,
      "loss": 1.5692,
      "step": 2480
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.496253992299153,
      "learning_rate": 1.9999140106163837e-06,
      "loss": 1.475,
      "step": 2481
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.810348760053238,
      "learning_rate": 1.9999139410044454e-06,
      "loss": 1.3871,
      "step": 2482
    },
    {
      "epoch": 0.02,
      "eval_loss": 1.6317201852798462,
      "eval_runtime": 4.6311,
      "eval_samples_per_second": 1.943,
      "eval_steps_per_second": 1.08,
      "step": 2482
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.936385744644064,
      "learning_rate": 1.999913871364343e-06,
      "loss": 1.5163,
      "step": 2483
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.3932940350781315,
      "learning_rate": 1.999913801696076e-06,
      "loss": 1.2527,
      "step": 2484
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.527212771343054,
      "learning_rate": 1.999913731999645e-06,
      "loss": 1.3808,
      "step": 2485
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.6367894135268415,
      "learning_rate": 1.9999136622750497e-06,
      "loss": 1.4644,
      "step": 2486
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.431008065838847,
      "learning_rate": 1.9999135925222905e-06,
      "loss": 1.4108,
      "step": 2487
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.60955842936837,
      "learning_rate": 1.9999135227413667e-06,
      "loss": 1.5293,
      "step": 2488
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.815249149207515,
      "learning_rate": 1.9999134529322785e-06,
      "loss": 1.4405,
      "step": 2489
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.905408881518468,
      "learning_rate": 1.9999133830950266e-06,
      "loss": 1.5907,
      "step": 2490
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.3654121964040815,
      "learning_rate": 1.99991331322961e-06,
      "loss": 1.5178,
      "step": 2491
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.58569484491462,
      "learning_rate": 1.9999132433360294e-06,
      "loss": 1.3849,
      "step": 2492
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.861754836669881,
      "learning_rate": 1.9999131734142846e-06,
      "loss": 1.5712,
      "step": 2493
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.651684688922827,
      "learning_rate": 1.9999131034643755e-06,
      "loss": 1.4253,
      "step": 2494
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.7784530707603885,
      "learning_rate": 1.9999130334863022e-06,
      "loss": 1.4367,
      "step": 2495
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.616133306685504,
      "learning_rate": 1.9999129634800646e-06,
      "loss": 1.3498,
      "step": 2496
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.119118711647458,
      "learning_rate": 1.9999128934456628e-06,
      "loss": 1.484,
      "step": 2497
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.283146448319006,
      "learning_rate": 1.999912823383097e-06,
      "loss": 1.3398,
      "step": 2498
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.571981013911557,
      "learning_rate": 1.9999127532923667e-06,
      "loss": 1.2996,
      "step": 2499
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.332851655290826,
      "learning_rate": 1.999912683173472e-06,
      "loss": 1.5005,
      "step": 2500
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.210017666226172,
      "learning_rate": 1.9999126130264135e-06,
      "loss": 1.3147,
      "step": 2501
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.862337302086877,
      "learning_rate": 1.9999125428511907e-06,
      "loss": 1.3428,
      "step": 2502
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.576195646780718,
      "learning_rate": 1.9999124726478036e-06,
      "loss": 1.4371,
      "step": 2503
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.339924831875166,
      "learning_rate": 1.9999124024162523e-06,
      "loss": 1.5637,
      "step": 2504
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.7192701371337025,
      "learning_rate": 1.9999123321565367e-06,
      "loss": 1.4242,
      "step": 2505
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.163473966004869,
      "learning_rate": 1.999912261868657e-06,
      "loss": 1.3637,
      "step": 2506
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.337334857544459,
      "learning_rate": 1.999912191552613e-06,
      "loss": 1.2784,
      "step": 2507
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.051444830035015,
      "learning_rate": 1.999912121208405e-06,
      "loss": 1.6266,
      "step": 2508
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.479649509721164,
      "learning_rate": 1.999912050836033e-06,
      "loss": 1.2662,
      "step": 2509
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.472943621796013,
      "learning_rate": 1.9999119804354964e-06,
      "loss": 1.4028,
      "step": 2510
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.710024013128741,
      "learning_rate": 1.9999119100067957e-06,
      "loss": 1.3498,
      "step": 2511
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.970475591078188,
      "learning_rate": 1.9999118395499306e-06,
      "loss": 1.5096,
      "step": 2512
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.75371637466236,
      "learning_rate": 1.9999117690649017e-06,
      "loss": 1.4432,
      "step": 2513
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.468537061799282,
      "learning_rate": 1.9999116985517085e-06,
      "loss": 1.4252,
      "step": 2514
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.49951522854794,
      "learning_rate": 1.999911628010351e-06,
      "loss": 1.3306,
      "step": 2515
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.719020079035169,
      "learning_rate": 1.9999115574408294e-06,
      "loss": 1.3402,
      "step": 2516
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.795670467668128,
      "learning_rate": 1.9999114868431434e-06,
      "loss": 1.4464,
      "step": 2517
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.36164451676725,
      "learning_rate": 1.9999114162172936e-06,
      "loss": 1.3221,
      "step": 2518
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.752857588130247,
      "learning_rate": 1.999911345563279e-06,
      "loss": 1.2821,
      "step": 2519
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.279730766268046,
      "learning_rate": 1.9999112748811008e-06,
      "loss": 1.4016,
      "step": 2520
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.695047237818786,
      "learning_rate": 1.9999112041707586e-06,
      "loss": 1.4613,
      "step": 2521
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.692585340555871,
      "learning_rate": 1.9999111334322517e-06,
      "loss": 1.4694,
      "step": 2522
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.771936002396662,
      "learning_rate": 1.999911062665581e-06,
      "loss": 1.3274,
      "step": 2523
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.220174817764933,
      "learning_rate": 1.999910991870746e-06,
      "loss": 1.571,
      "step": 2524
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.672229165440042,
      "learning_rate": 1.9999109210477466e-06,
      "loss": 1.5377,
      "step": 2525
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.1839178319765935,
      "learning_rate": 1.999910850196583e-06,
      "loss": 1.3993,
      "step": 2526
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.681220552283899,
      "learning_rate": 1.9999107793172557e-06,
      "loss": 1.4291,
      "step": 2527
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.972007639821219,
      "learning_rate": 1.999910708409764e-06,
      "loss": 1.431,
      "step": 2528
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.4861243909511295,
      "learning_rate": 1.999910637474108e-06,
      "loss": 1.4019,
      "step": 2529
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.851005767238794,
      "learning_rate": 1.9999105665102883e-06,
      "loss": 1.4337,
      "step": 2530
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.846084074130138,
      "learning_rate": 1.999910495518304e-06,
      "loss": 1.5946,
      "step": 2531
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.536951575081545,
      "learning_rate": 1.9999104244981556e-06,
      "loss": 1.4743,
      "step": 2532
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.248194758577965,
      "learning_rate": 1.9999103534498434e-06,
      "loss": 1.3195,
      "step": 2533
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.056484295400697,
      "learning_rate": 1.9999102823733665e-06,
      "loss": 1.2914,
      "step": 2534
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.505711846073996,
      "learning_rate": 1.999910211268726e-06,
      "loss": 1.4096,
      "step": 2535
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.9063691562807055,
      "learning_rate": 1.999910140135921e-06,
      "loss": 1.2676,
      "step": 2536
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.783198227824001,
      "learning_rate": 1.9999100689749517e-06,
      "loss": 1.4266,
      "step": 2537
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.4803987517358665,
      "learning_rate": 1.9999099977858186e-06,
      "loss": 1.3878,
      "step": 2538
    },
    {
      "epoch": 0.02,
      "grad_norm": 21.125075865811542,
      "learning_rate": 1.9999099265685212e-06,
      "loss": 1.481,
      "step": 2539
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.551735343167499,
      "learning_rate": 1.99990985532306e-06,
      "loss": 1.3335,
      "step": 2540
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.525931993787852,
      "learning_rate": 1.999909784049434e-06,
      "loss": 1.4145,
      "step": 2541
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.723242340525657,
      "learning_rate": 1.9999097127476444e-06,
      "loss": 1.425,
      "step": 2542
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.52159628767104,
      "learning_rate": 1.9999096414176904e-06,
      "loss": 1.5459,
      "step": 2543
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.499238976068291,
      "learning_rate": 1.9999095700595726e-06,
      "loss": 1.5052,
      "step": 2544
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.3943003554534315,
      "learning_rate": 1.999909498673291e-06,
      "loss": 1.3532,
      "step": 2545
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.384405241916104,
      "learning_rate": 1.9999094272588445e-06,
      "loss": 1.5439,
      "step": 2546
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.401136108906954,
      "learning_rate": 1.999909355816234e-06,
      "loss": 1.4723,
      "step": 2547
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.4369310319106114,
      "learning_rate": 1.9999092843454593e-06,
      "loss": 1.4131,
      "step": 2548
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.975825034724867,
      "learning_rate": 1.999909212846521e-06,
      "loss": 1.5112,
      "step": 2549
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.256239787239192,
      "learning_rate": 1.999909141319418e-06,
      "loss": 1.5494,
      "step": 2550
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.563176202007188,
      "learning_rate": 1.9999090697641515e-06,
      "loss": 1.3635,
      "step": 2551
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.431366662952086,
      "learning_rate": 1.9999089981807203e-06,
      "loss": 1.3677,
      "step": 2552
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.189139491517448,
      "learning_rate": 1.9999089265691253e-06,
      "loss": 1.3525,
      "step": 2553
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.080861898020357,
      "learning_rate": 1.999908854929366e-06,
      "loss": 1.3792,
      "step": 2554
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.3571862989208325,
      "learning_rate": 1.9999087832614425e-06,
      "loss": 1.5974,
      "step": 2555
    },
    {
      "epoch": 0.02,
      "eval_loss": 1.626267433166504,
      "eval_runtime": 4.6223,
      "eval_samples_per_second": 1.947,
      "eval_steps_per_second": 1.082,
      "step": 2555
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.82494449812951,
      "learning_rate": 1.9999087115653556e-06,
      "loss": 1.4692,
      "step": 2556
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.929797834611915,
      "learning_rate": 1.999908639841104e-06,
      "loss": 1.4674,
      "step": 2557
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.589160350904509,
      "learning_rate": 1.9999085680886884e-06,
      "loss": 1.4853,
      "step": 2558
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.657411083617213,
      "learning_rate": 1.9999084963081087e-06,
      "loss": 1.3349,
      "step": 2559
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.635155884205405,
      "learning_rate": 1.999908424499365e-06,
      "loss": 1.3812,
      "step": 2560
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.039436368621972,
      "learning_rate": 1.9999083526624568e-06,
      "loss": 1.4228,
      "step": 2561
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.731509118685059,
      "learning_rate": 1.999908280797385e-06,
      "loss": 1.5188,
      "step": 2562
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.845729221450556,
      "learning_rate": 1.999908208904149e-06,
      "loss": 1.3457,
      "step": 2563
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.6098361015362705,
      "learning_rate": 1.9999081369827488e-06,
      "loss": 1.4256,
      "step": 2564
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.916273960056245,
      "learning_rate": 1.9999080650331847e-06,
      "loss": 1.3279,
      "step": 2565
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.438335756166127,
      "learning_rate": 1.9999079930554563e-06,
      "loss": 1.241,
      "step": 2566
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.817882655917825,
      "learning_rate": 1.999907921049564e-06,
      "loss": 1.4117,
      "step": 2567
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.003123275534057,
      "learning_rate": 1.9999078490155076e-06,
      "loss": 1.2974,
      "step": 2568
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.017000975631727,
      "learning_rate": 1.999907776953287e-06,
      "loss": 1.363,
      "step": 2569
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.600005159137907,
      "learning_rate": 1.9999077048629022e-06,
      "loss": 1.4458,
      "step": 2570
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.9265496357161425,
      "learning_rate": 1.9999076327443538e-06,
      "loss": 1.3172,
      "step": 2571
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.824876884767345,
      "learning_rate": 1.9999075605976406e-06,
      "loss": 1.4476,
      "step": 2572
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.361863130512629,
      "learning_rate": 1.9999074884227636e-06,
      "loss": 1.3836,
      "step": 2573
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.7955376481293115,
      "learning_rate": 1.999907416219723e-06,
      "loss": 1.3473,
      "step": 2574
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.666469831511099,
      "learning_rate": 1.9999073439885177e-06,
      "loss": 1.4382,
      "step": 2575
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.73863410034199,
      "learning_rate": 1.999907271729149e-06,
      "loss": 1.3452,
      "step": 2576
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.535702568291937,
      "learning_rate": 1.999907199441616e-06,
      "loss": 1.433,
      "step": 2577
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.577074935520658,
      "learning_rate": 1.999907127125919e-06,
      "loss": 1.4831,
      "step": 2578
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.722599811411352,
      "learning_rate": 1.9999070547820576e-06,
      "loss": 1.5383,
      "step": 2579
    },
    {
      "epoch": 0.02,
      "grad_norm": 22.655065655121753,
      "learning_rate": 1.999906982410032e-06,
      "loss": 1.3815,
      "step": 2580
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.003296650995224,
      "learning_rate": 1.9999069100098425e-06,
      "loss": 1.5457,
      "step": 2581
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.741277967056868,
      "learning_rate": 1.9999068375814896e-06,
      "loss": 1.5165,
      "step": 2582
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.300018975344748,
      "learning_rate": 1.999906765124972e-06,
      "loss": 1.4036,
      "step": 2583
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.552215789433446,
      "learning_rate": 1.9999066926402907e-06,
      "loss": 1.3368,
      "step": 2584
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.470582591235478,
      "learning_rate": 1.999906620127445e-06,
      "loss": 1.5863,
      "step": 2585
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.100878877106844,
      "learning_rate": 1.9999065475864355e-06,
      "loss": 1.501,
      "step": 2586
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.389860076860747,
      "learning_rate": 1.9999064750172617e-06,
      "loss": 1.4151,
      "step": 2587
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.63645147327467,
      "learning_rate": 1.999906402419924e-06,
      "loss": 1.5282,
      "step": 2588
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.229014935819337,
      "learning_rate": 1.9999063297944226e-06,
      "loss": 1.5142,
      "step": 2589
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.590820624310537,
      "learning_rate": 1.999906257140757e-06,
      "loss": 1.3774,
      "step": 2590
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.724381873263039,
      "learning_rate": 1.9999061844589268e-06,
      "loss": 1.3214,
      "step": 2591
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.484001614536028,
      "learning_rate": 1.999906111748933e-06,
      "loss": 1.3858,
      "step": 2592
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.208403452992326,
      "learning_rate": 1.999906039010775e-06,
      "loss": 1.3302,
      "step": 2593
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.505618278783764,
      "learning_rate": 1.9999059662444537e-06,
      "loss": 1.4443,
      "step": 2594
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.591195710426086,
      "learning_rate": 1.999905893449968e-06,
      "loss": 1.3943,
      "step": 2595
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.696805772544854,
      "learning_rate": 1.9999058206273177e-06,
      "loss": 1.4307,
      "step": 2596
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.021073562891561,
      "learning_rate": 1.999905747776504e-06,
      "loss": 1.5345,
      "step": 2597
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.7598018312861194,
      "learning_rate": 1.999905674897526e-06,
      "loss": 1.6453,
      "step": 2598
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.581994132757299,
      "learning_rate": 1.9999056019903835e-06,
      "loss": 1.4905,
      "step": 2599
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.841310271459569,
      "learning_rate": 1.999905529055078e-06,
      "loss": 1.4203,
      "step": 2600
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.189830617523313,
      "learning_rate": 1.999905456091608e-06,
      "loss": 1.3274,
      "step": 2601
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.720146352163821,
      "learning_rate": 1.999905383099974e-06,
      "loss": 1.4674,
      "step": 2602
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.543963809407407,
      "learning_rate": 1.999905310080176e-06,
      "loss": 1.4255,
      "step": 2603
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.18614032345166,
      "learning_rate": 1.999905237032214e-06,
      "loss": 1.4297,
      "step": 2604
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.465791115799426,
      "learning_rate": 1.999905163956088e-06,
      "loss": 1.4399,
      "step": 2605
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.694449197808517,
      "learning_rate": 1.9999050908517977e-06,
      "loss": 1.4938,
      "step": 2606
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.4688966732418365,
      "learning_rate": 1.9999050177193436e-06,
      "loss": 1.4578,
      "step": 2607
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.10844282586031,
      "learning_rate": 1.9999049445587256e-06,
      "loss": 1.3347,
      "step": 2608
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.762680506088013,
      "learning_rate": 1.999904871369944e-06,
      "loss": 1.4342,
      "step": 2609
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.979551926427287,
      "learning_rate": 1.999904798152998e-06,
      "loss": 1.5078,
      "step": 2610
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.533251775082327,
      "learning_rate": 1.999904724907888e-06,
      "loss": 1.2065,
      "step": 2611
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.121922341187516,
      "learning_rate": 1.9999046516346138e-06,
      "loss": 1.5454,
      "step": 2612
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.781680034327118,
      "learning_rate": 1.9999045783331757e-06,
      "loss": 1.4578,
      "step": 2613
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.826495337980005,
      "learning_rate": 1.999904505003574e-06,
      "loss": 1.4422,
      "step": 2614
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.335079927329629,
      "learning_rate": 1.9999044316458078e-06,
      "loss": 1.3345,
      "step": 2615
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.778422328814399,
      "learning_rate": 1.9999043582598782e-06,
      "loss": 1.3202,
      "step": 2616
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.8282702985765145,
      "learning_rate": 1.999904284845784e-06,
      "loss": 1.407,
      "step": 2617
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.867494484747974,
      "learning_rate": 1.999904211403526e-06,
      "loss": 1.5508,
      "step": 2618
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.781917993855085,
      "learning_rate": 1.9999041379331044e-06,
      "loss": 1.3034,
      "step": 2619
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.712392259942616,
      "learning_rate": 1.9999040644345186e-06,
      "loss": 1.5253,
      "step": 2620
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.030532462772803,
      "learning_rate": 1.9999039909077686e-06,
      "loss": 1.5149,
      "step": 2621
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.621139707916859,
      "learning_rate": 1.999903917352855e-06,
      "loss": 1.5093,
      "step": 2622
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.008006829313361,
      "learning_rate": 1.9999038437697774e-06,
      "loss": 1.4026,
      "step": 2623
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.136006728609445,
      "learning_rate": 1.9999037701585354e-06,
      "loss": 1.5477,
      "step": 2624
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.760430548827488,
      "learning_rate": 1.9999036965191295e-06,
      "loss": 1.4864,
      "step": 2625
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.006634707089645,
      "learning_rate": 1.9999036228515603e-06,
      "loss": 1.4774,
      "step": 2626
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.588292407337218,
      "learning_rate": 1.9999035491558263e-06,
      "loss": 1.3735,
      "step": 2627
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.816326208591829,
      "learning_rate": 1.999903475431929e-06,
      "loss": 1.2208,
      "step": 2628
    },
    {
      "epoch": 0.02,
      "eval_loss": 1.6295270919799805,
      "eval_runtime": 4.6318,
      "eval_samples_per_second": 1.943,
      "eval_steps_per_second": 1.079,
      "step": 2628
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.638403311998232,
      "learning_rate": 1.9999034016798672e-06,
      "loss": 1.4943,
      "step": 2629
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.961202684709964,
      "learning_rate": 1.999903327899642e-06,
      "loss": 1.4069,
      "step": 2630
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.348996064319459,
      "learning_rate": 1.9999032540912524e-06,
      "loss": 1.1698,
      "step": 2631
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.444823433064274,
      "learning_rate": 1.999903180254699e-06,
      "loss": 1.3074,
      "step": 2632
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.965888101259268,
      "learning_rate": 1.9999031063899817e-06,
      "loss": 1.3823,
      "step": 2633
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.453473043817252,
      "learning_rate": 1.999903032497101e-06,
      "loss": 1.4969,
      "step": 2634
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.068247346576678,
      "learning_rate": 1.9999029585760553e-06,
      "loss": 1.5847,
      "step": 2635
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.6600389675468294,
      "learning_rate": 1.9999028846268463e-06,
      "loss": 1.3724,
      "step": 2636
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.374383238249181,
      "learning_rate": 1.9999028106494735e-06,
      "loss": 1.4313,
      "step": 2637
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.680029067986695,
      "learning_rate": 1.999902736643936e-06,
      "loss": 1.3181,
      "step": 2638
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.580712621593848,
      "learning_rate": 1.999902662610235e-06,
      "loss": 1.3987,
      "step": 2639
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.092256320436487,
      "learning_rate": 1.9999025885483706e-06,
      "loss": 1.6909,
      "step": 2640
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.7827050109074385,
      "learning_rate": 1.9999025144583415e-06,
      "loss": 1.517,
      "step": 2641
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.011057538279476,
      "learning_rate": 1.9999024403401486e-06,
      "loss": 1.4349,
      "step": 2642
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.8294099232047785,
      "learning_rate": 1.9999023661937923e-06,
      "loss": 1.4704,
      "step": 2643
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.597159590269809,
      "learning_rate": 1.9999022920192712e-06,
      "loss": 1.3426,
      "step": 2644
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.9150712542829496,
      "learning_rate": 1.9999022178165868e-06,
      "loss": 1.3268,
      "step": 2645
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.027583647063347,
      "learning_rate": 1.9999021435857385e-06,
      "loss": 1.4659,
      "step": 2646
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.68442647731963,
      "learning_rate": 1.9999020693267263e-06,
      "loss": 1.3447,
      "step": 2647
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.9605688635740695,
      "learning_rate": 1.9999019950395503e-06,
      "loss": 1.4796,
      "step": 2648
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.809892082871139,
      "learning_rate": 1.9999019207242097e-06,
      "loss": 1.4104,
      "step": 2649
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.743744228034844,
      "learning_rate": 1.999901846380706e-06,
      "loss": 1.4046,
      "step": 2650
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.269118159347582,
      "learning_rate": 1.999901772009038e-06,
      "loss": 1.2958,
      "step": 2651
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.853949838326959,
      "learning_rate": 1.999901697609206e-06,
      "loss": 1.4926,
      "step": 2652
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.776685722649282,
      "learning_rate": 1.99990162318121e-06,
      "loss": 1.4355,
      "step": 2653
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.208703600062207,
      "learning_rate": 1.9999015487250503e-06,
      "loss": 1.5673,
      "step": 2654
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.7039599749841186,
      "learning_rate": 1.999901474240727e-06,
      "loss": 1.5676,
      "step": 2655
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.417347561144367,
      "learning_rate": 1.9999013997282393e-06,
      "loss": 1.3797,
      "step": 2656
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.6270515827313465,
      "learning_rate": 1.999901325187588e-06,
      "loss": 1.4198,
      "step": 2657
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.644490932379133,
      "learning_rate": 1.9999012506187726e-06,
      "loss": 1.3909,
      "step": 2658
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.766952782820096,
      "learning_rate": 1.9999011760217935e-06,
      "loss": 1.5499,
      "step": 2659
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.024704187001623,
      "learning_rate": 1.99990110139665e-06,
      "loss": 1.5967,
      "step": 2660
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.484156877401999,
      "learning_rate": 1.9999010267433432e-06,
      "loss": 1.3459,
      "step": 2661
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.960235392776955,
      "learning_rate": 1.9999009520618725e-06,
      "loss": 1.47,
      "step": 2662
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.013252259638264,
      "learning_rate": 1.9999008773522376e-06,
      "loss": 1.2694,
      "step": 2663
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.796804229686202,
      "learning_rate": 1.9999008026144392e-06,
      "loss": 1.4616,
      "step": 2664
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.1481769265464825,
      "learning_rate": 1.9999007278484766e-06,
      "loss": 1.494,
      "step": 2665
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.895324247068634,
      "learning_rate": 1.99990065305435e-06,
      "loss": 1.5384,
      "step": 2666
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.655033924444686,
      "learning_rate": 1.99990057823206e-06,
      "loss": 1.427,
      "step": 2667
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.7945063377020425,
      "learning_rate": 1.999900503381606e-06,
      "loss": 1.4635,
      "step": 2668
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.282804236505679,
      "learning_rate": 1.999900428502988e-06,
      "loss": 1.3777,
      "step": 2669
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.956119769772323,
      "learning_rate": 1.9999003535962058e-06,
      "loss": 1.5537,
      "step": 2670
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.744533892549758,
      "learning_rate": 1.99990027866126e-06,
      "loss": 1.411,
      "step": 2671
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.685437895155002,
      "learning_rate": 1.9999002036981505e-06,
      "loss": 1.3089,
      "step": 2672
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.679001375905665,
      "learning_rate": 1.999900128706877e-06,
      "loss": 1.5004,
      "step": 2673
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.816836882292446,
      "learning_rate": 1.99990005368744e-06,
      "loss": 1.4415,
      "step": 2674
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.397353037395355,
      "learning_rate": 1.9998999786398383e-06,
      "loss": 1.457,
      "step": 2675
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.836128128995581,
      "learning_rate": 1.9998999035640734e-06,
      "loss": 1.3645,
      "step": 2676
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.345520428845227,
      "learning_rate": 1.9998998284601446e-06,
      "loss": 1.5424,
      "step": 2677
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.582942854801658,
      "learning_rate": 1.999899753328052e-06,
      "loss": 1.4421,
      "step": 2678
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5571891655235275,
      "learning_rate": 1.999899678167795e-06,
      "loss": 1.3543,
      "step": 2679
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.349049923012939,
      "learning_rate": 1.999899602979375e-06,
      "loss": 1.209,
      "step": 2680
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.584099365206942,
      "learning_rate": 1.9998995277627906e-06,
      "loss": 1.4517,
      "step": 2681
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.851375584879138,
      "learning_rate": 1.9998994525180426e-06,
      "loss": 1.3411,
      "step": 2682
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.6601703251560025,
      "learning_rate": 1.9998993772451304e-06,
      "loss": 1.3298,
      "step": 2683
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.8561771787958365,
      "learning_rate": 1.9998993019440547e-06,
      "loss": 1.4937,
      "step": 2684
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.1790245554459755,
      "learning_rate": 1.999899226614815e-06,
      "loss": 1.471,
      "step": 2685
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.742734422892583,
      "learning_rate": 1.9998991512574113e-06,
      "loss": 1.5035,
      "step": 2686
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.785648242343571,
      "learning_rate": 1.999899075871844e-06,
      "loss": 1.5267,
      "step": 2687
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.057841994753448,
      "learning_rate": 1.999899000458113e-06,
      "loss": 1.3789,
      "step": 2688
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.0118905395912945,
      "learning_rate": 1.999898925016218e-06,
      "loss": 1.3579,
      "step": 2689
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.35192729500763,
      "learning_rate": 1.999898849546159e-06,
      "loss": 1.4838,
      "step": 2690
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.337685327376343,
      "learning_rate": 1.9998987740479367e-06,
      "loss": 1.3297,
      "step": 2691
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.808629946983388,
      "learning_rate": 1.99989869852155e-06,
      "loss": 1.4873,
      "step": 2692
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.2407858594510035,
      "learning_rate": 1.999898622967e-06,
      "loss": 1.5086,
      "step": 2693
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.941322758554874,
      "learning_rate": 1.9998985473842858e-06,
      "loss": 1.3952,
      "step": 2694
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.260790698502375,
      "learning_rate": 1.9998984717734078e-06,
      "loss": 1.3652,
      "step": 2695
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.2005982502957595,
      "learning_rate": 1.999898396134366e-06,
      "loss": 1.5483,
      "step": 2696
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.772216539864026,
      "learning_rate": 1.9998983204671606e-06,
      "loss": 1.2309,
      "step": 2697
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.448621297124078,
      "learning_rate": 1.999898244771791e-06,
      "loss": 1.3967,
      "step": 2698
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.6464418666503295,
      "learning_rate": 1.9998981690482577e-06,
      "loss": 1.2706,
      "step": 2699
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.142174365370682,
      "learning_rate": 1.999898093296561e-06,
      "loss": 1.3629,
      "step": 2700
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.489321589199872,
      "learning_rate": 1.9998980175167003e-06,
      "loss": 1.4768,
      "step": 2701
    },
    {
      "epoch": 0.02,
      "eval_loss": 1.6249644756317139,
      "eval_runtime": 4.6538,
      "eval_samples_per_second": 1.934,
      "eval_steps_per_second": 1.074,
      "step": 2701
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.728941606346881,
      "learning_rate": 1.9998979417086758e-06,
      "loss": 1.4208,
      "step": 2702
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.721435664278658,
      "learning_rate": 1.9998978658724874e-06,
      "loss": 1.6635,
      "step": 2703
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.672729490886054,
      "learning_rate": 1.9998977900081353e-06,
      "loss": 1.3433,
      "step": 2704
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.476896381988856,
      "learning_rate": 1.9998977141156192e-06,
      "loss": 1.4389,
      "step": 2705
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.783782128937111,
      "learning_rate": 1.9998976381949394e-06,
      "loss": 1.4866,
      "step": 2706
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.724902284051836,
      "learning_rate": 1.999897562246096e-06,
      "loss": 1.3447,
      "step": 2707
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.850028686058834,
      "learning_rate": 1.999897486269089e-06,
      "loss": 1.3209,
      "step": 2708
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.83053553627955,
      "learning_rate": 1.9998974102639175e-06,
      "loss": 1.5586,
      "step": 2709
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.712477754987166,
      "learning_rate": 1.9998973342305827e-06,
      "loss": 1.4225,
      "step": 2710
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.54227448455875,
      "learning_rate": 1.999897258169084e-06,
      "loss": 1.3462,
      "step": 2711
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.5411282379403985,
      "learning_rate": 1.9998971820794215e-06,
      "loss": 1.4259,
      "step": 2712
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.739705320723445,
      "learning_rate": 1.999897105961595e-06,
      "loss": 1.457,
      "step": 2713
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.701603405635517,
      "learning_rate": 1.9998970298156053e-06,
      "loss": 1.4297,
      "step": 2714
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.991641204953335,
      "learning_rate": 1.9998969536414512e-06,
      "loss": 1.4822,
      "step": 2715
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.480036657625506,
      "learning_rate": 1.9998968774391338e-06,
      "loss": 1.3061,
      "step": 2716
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.857570508327938,
      "learning_rate": 1.9998968012086524e-06,
      "loss": 1.5446,
      "step": 2717
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.917848715898165,
      "learning_rate": 1.9998967249500073e-06,
      "loss": 1.4604,
      "step": 2718
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.532991154379635,
      "learning_rate": 1.9998966486631987e-06,
      "loss": 1.3497,
      "step": 2719
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.516960068763919,
      "learning_rate": 1.999896572348226e-06,
      "loss": 1.3371,
      "step": 2720
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.4294845884659875,
      "learning_rate": 1.9998964960050895e-06,
      "loss": 1.4655,
      "step": 2721
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.465778207032482,
      "learning_rate": 1.9998964196337894e-06,
      "loss": 1.1239,
      "step": 2722
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.380579112263108,
      "learning_rate": 1.9998963432343254e-06,
      "loss": 1.4072,
      "step": 2723
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.347121577221788,
      "learning_rate": 1.999896266806698e-06,
      "loss": 1.4905,
      "step": 2724
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.091313845837881,
      "learning_rate": 1.9998961903509063e-06,
      "loss": 1.1784,
      "step": 2725
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.9326180566194155,
      "learning_rate": 1.9998961138669512e-06,
      "loss": 1.5945,
      "step": 2726
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.547464104001262,
      "learning_rate": 1.9998960373548323e-06,
      "loss": 1.7427,
      "step": 2727
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.955919797294654,
      "learning_rate": 1.9998959608145495e-06,
      "loss": 1.5829,
      "step": 2728
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.259954508473689,
      "learning_rate": 1.9998958842461033e-06,
      "loss": 1.2882,
      "step": 2729
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.242648662389775,
      "learning_rate": 1.9998958076494933e-06,
      "loss": 1.2997,
      "step": 2730
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.442253262087044,
      "learning_rate": 1.9998957310247194e-06,
      "loss": 1.2864,
      "step": 2731
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.672333737798939,
      "learning_rate": 1.9998956543717816e-06,
      "loss": 1.5532,
      "step": 2732
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.768920768366485,
      "learning_rate": 1.9998955776906805e-06,
      "loss": 1.4113,
      "step": 2733
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.365298258133692,
      "learning_rate": 1.9998955009814155e-06,
      "loss": 1.4779,
      "step": 2734
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.977640813439696,
      "learning_rate": 1.9998954242439866e-06,
      "loss": 1.3955,
      "step": 2735
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.779169735265989,
      "learning_rate": 1.999895347478394e-06,
      "loss": 1.4087,
      "step": 2736
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.7600337936905275,
      "learning_rate": 1.9998952706846378e-06,
      "loss": 1.46,
      "step": 2737
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.654467318840138,
      "learning_rate": 1.999895193862718e-06,
      "loss": 1.3913,
      "step": 2738
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.704589355819833,
      "learning_rate": 1.999895117012634e-06,
      "loss": 1.5544,
      "step": 2739
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.1941897065206595,
      "learning_rate": 1.9998950401343863e-06,
      "loss": 1.273,
      "step": 2740
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.439869003538217,
      "learning_rate": 1.9998949632279753e-06,
      "loss": 1.4237,
      "step": 2741
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.50147659109475,
      "learning_rate": 1.9998948862934008e-06,
      "loss": 1.3462,
      "step": 2742
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.584850957877264,
      "learning_rate": 1.999894809330662e-06,
      "loss": 1.2504,
      "step": 2743
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.042375090827711,
      "learning_rate": 1.99989473233976e-06,
      "loss": 1.3003,
      "step": 2744
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.023492077637436,
      "learning_rate": 1.9998946553206938e-06,
      "loss": 1.4422,
      "step": 2745
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.748553264507999,
      "learning_rate": 1.999894578273464e-06,
      "loss": 1.5238,
      "step": 2746
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.570660659610173,
      "learning_rate": 1.9998945011980706e-06,
      "loss": 1.4269,
      "step": 2747
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.452679880713028,
      "learning_rate": 1.9998944240945134e-06,
      "loss": 1.4164,
      "step": 2748
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.836656376655193,
      "learning_rate": 1.999894346962793e-06,
      "loss": 1.2856,
      "step": 2749
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.798771542642795,
      "learning_rate": 1.999894269802908e-06,
      "loss": 1.3372,
      "step": 2750
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.170216529268188,
      "learning_rate": 1.9998941926148597e-06,
      "loss": 1.4021,
      "step": 2751
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.167356093869082,
      "learning_rate": 1.999894115398648e-06,
      "loss": 1.4013,
      "step": 2752
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.587890795315358,
      "learning_rate": 1.9998940381542725e-06,
      "loss": 1.4176,
      "step": 2753
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.177127842863576,
      "learning_rate": 1.999893960881733e-06,
      "loss": 1.5869,
      "step": 2754
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.574137972828784,
      "learning_rate": 1.9998938835810304e-06,
      "loss": 1.5673,
      "step": 2755
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.778126829684162,
      "learning_rate": 1.9998938062521637e-06,
      "loss": 1.5361,
      "step": 2756
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.851913885459397,
      "learning_rate": 1.9998937288951332e-06,
      "loss": 1.412,
      "step": 2757
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.122129218110728,
      "learning_rate": 1.9998936515099393e-06,
      "loss": 1.4406,
      "step": 2758
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.775466238362049,
      "learning_rate": 1.9998935740965815e-06,
      "loss": 1.3279,
      "step": 2759
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.097236710699216,
      "learning_rate": 1.9998934966550604e-06,
      "loss": 1.5103,
      "step": 2760
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.852445289044562,
      "learning_rate": 1.999893419185375e-06,
      "loss": 1.469,
      "step": 2761
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.912077857670805,
      "learning_rate": 1.9998933416875265e-06,
      "loss": 1.4218,
      "step": 2762
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.689088161317069,
      "learning_rate": 1.999893264161514e-06,
      "loss": 1.3336,
      "step": 2763
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.766112013484904,
      "learning_rate": 1.999893186607338e-06,
      "loss": 1.3066,
      "step": 2764
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.593303699801557,
      "learning_rate": 1.9998931090249985e-06,
      "loss": 1.5657,
      "step": 2765
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.163779733210183,
      "learning_rate": 1.9998930314144947e-06,
      "loss": 1.4395,
      "step": 2766
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.674315487003727,
      "learning_rate": 1.999892953775828e-06,
      "loss": 1.5407,
      "step": 2767
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.676661266930629,
      "learning_rate": 1.999892876108997e-06,
      "loss": 1.5139,
      "step": 2768
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.128307746492878,
      "learning_rate": 1.9998927984140026e-06,
      "loss": 1.3218,
      "step": 2769
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.240640196066125,
      "learning_rate": 1.9998927206908447e-06,
      "loss": 1.2989,
      "step": 2770
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.27816898885737,
      "learning_rate": 1.999892642939523e-06,
      "loss": 1.3487,
      "step": 2771
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.025010385489242,
      "learning_rate": 1.9998925651600377e-06,
      "loss": 1.3706,
      "step": 2772
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.464981368592326,
      "learning_rate": 1.9998924873523886e-06,
      "loss": 1.3258,
      "step": 2773
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.665177992679757,
      "learning_rate": 1.999892409516576e-06,
      "loss": 1.3428,
      "step": 2774
    },
    {
      "epoch": 0.02,
      "eval_loss": 1.6228926181793213,
      "eval_runtime": 4.6339,
      "eval_samples_per_second": 1.942,
      "eval_steps_per_second": 1.079,
      "step": 2774
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.868974430046027,
      "learning_rate": 1.9998923316526e-06,
      "loss": 1.4719,
      "step": 2775
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.304645907019113,
      "learning_rate": 1.99989225376046e-06,
      "loss": 1.5306,
      "step": 2776
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.387713736296703,
      "learning_rate": 1.9998921758401565e-06,
      "loss": 1.3475,
      "step": 2777
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.442792610487523,
      "learning_rate": 1.9998920978916895e-06,
      "loss": 1.36,
      "step": 2778
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.683582295474543,
      "learning_rate": 1.9998920199150587e-06,
      "loss": 1.4342,
      "step": 2779
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.777504817382186,
      "learning_rate": 1.9998919419102644e-06,
      "loss": 1.3633,
      "step": 2780
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.598408814061361,
      "learning_rate": 1.9998918638773063e-06,
      "loss": 1.3561,
      "step": 2781
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.531887898739241,
      "learning_rate": 1.9998917858161843e-06,
      "loss": 1.4105,
      "step": 2782
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.379373053575446,
      "learning_rate": 1.999891707726899e-06,
      "loss": 1.3188,
      "step": 2783
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.7609097936376195,
      "learning_rate": 1.99989162960945e-06,
      "loss": 1.3523,
      "step": 2784
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.7932273540706545,
      "learning_rate": 1.9998915514638374e-06,
      "loss": 1.4839,
      "step": 2785
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.403408099341777,
      "learning_rate": 1.9998914732900618e-06,
      "loss": 1.3651,
      "step": 2786
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.363493337417038,
      "learning_rate": 1.9998913950881214e-06,
      "loss": 1.3903,
      "step": 2787
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.703872779282757,
      "learning_rate": 1.999891316858018e-06,
      "loss": 1.3602,
      "step": 2788
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.594866601682083,
      "learning_rate": 1.9998912385997513e-06,
      "loss": 1.2933,
      "step": 2789
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.606930331365836,
      "learning_rate": 1.9998911603133207e-06,
      "loss": 1.4318,
      "step": 2790
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.67703505544561,
      "learning_rate": 1.9998910819987262e-06,
      "loss": 1.1987,
      "step": 2791
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.858347802822455,
      "learning_rate": 1.9998910036559688e-06,
      "loss": 1.3342,
      "step": 2792
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.238675085196649,
      "learning_rate": 1.999890925285047e-06,
      "loss": 1.1874,
      "step": 2793
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.636225801921176,
      "learning_rate": 1.999890846885962e-06,
      "loss": 1.3584,
      "step": 2794
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.633817615345395,
      "learning_rate": 1.9998907684587133e-06,
      "loss": 1.4063,
      "step": 2795
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.893305859480175,
      "learning_rate": 1.999890690003301e-06,
      "loss": 1.4289,
      "step": 2796
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.7172189836983796,
      "learning_rate": 1.999890611519725e-06,
      "loss": 1.4481,
      "step": 2797
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.881737643176644,
      "learning_rate": 1.999890533007986e-06,
      "loss": 1.4956,
      "step": 2798
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.986642114604307,
      "learning_rate": 1.9998904544680827e-06,
      "loss": 1.6169,
      "step": 2799
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.2687198972018905,
      "learning_rate": 1.999890375900016e-06,
      "loss": 1.3991,
      "step": 2800
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.692782105304108,
      "learning_rate": 1.9998902973037858e-06,
      "loss": 1.3991,
      "step": 2801
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.509363415701804,
      "learning_rate": 1.999890218679392e-06,
      "loss": 1.4011,
      "step": 2802
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.431731471169426,
      "learning_rate": 1.9998901400268348e-06,
      "loss": 1.3621,
      "step": 2803
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.881655716824346,
      "learning_rate": 1.9998900613461137e-06,
      "loss": 1.4308,
      "step": 2804
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.081486388140841,
      "learning_rate": 1.9998899826372292e-06,
      "loss": 1.3531,
      "step": 2805
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.1521080991642965,
      "learning_rate": 1.9998899039001813e-06,
      "loss": 1.31,
      "step": 2806
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.210301192592582,
      "learning_rate": 1.9998898251349696e-06,
      "loss": 1.5408,
      "step": 2807
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.589178232793303,
      "learning_rate": 1.9998897463415944e-06,
      "loss": 1.3702,
      "step": 2808
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.375935678979782,
      "learning_rate": 1.999889667520056e-06,
      "loss": 1.3905,
      "step": 2809
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.436455510679924,
      "learning_rate": 1.999889588670353e-06,
      "loss": 1.3327,
      "step": 2810
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.29249042088829,
      "learning_rate": 1.9998895097924875e-06,
      "loss": 1.4343,
      "step": 2811
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.784394796780278,
      "learning_rate": 1.9998894308864578e-06,
      "loss": 1.4381,
      "step": 2812
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.946717504980601,
      "learning_rate": 1.9998893519522646e-06,
      "loss": 1.4603,
      "step": 2813
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.766983915055353,
      "learning_rate": 1.999889272989908e-06,
      "loss": 1.5148,
      "step": 2814
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.708307718420187,
      "learning_rate": 1.9998891939993877e-06,
      "loss": 1.3662,
      "step": 2815
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.562866445292407,
      "learning_rate": 1.999889114980704e-06,
      "loss": 1.5425,
      "step": 2816
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.8892392384319745,
      "learning_rate": 1.9998890359338566e-06,
      "loss": 1.5102,
      "step": 2817
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.681445930399619,
      "learning_rate": 1.999888956858846e-06,
      "loss": 1.5407,
      "step": 2818
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.932383185973984,
      "learning_rate": 1.9998888777556714e-06,
      "loss": 1.3519,
      "step": 2819
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.573974144160327,
      "learning_rate": 1.9998887986243335e-06,
      "loss": 1.3792,
      "step": 2820
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.945964809681693,
      "learning_rate": 1.999888719464832e-06,
      "loss": 1.3079,
      "step": 2821
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.6784540645297366,
      "learning_rate": 1.999888640277167e-06,
      "loss": 1.3658,
      "step": 2822
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.4758180832015775,
      "learning_rate": 1.9998885610613383e-06,
      "loss": 1.4252,
      "step": 2823
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.924403479676831,
      "learning_rate": 1.9998884818173467e-06,
      "loss": 1.4362,
      "step": 2824
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.1946491484073745,
      "learning_rate": 1.999888402545191e-06,
      "loss": 1.4022,
      "step": 2825
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.116594932180298,
      "learning_rate": 1.999888323244872e-06,
      "loss": 1.458,
      "step": 2826
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.641238557929875,
      "learning_rate": 1.999888243916389e-06,
      "loss": 1.3547,
      "step": 2827
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.594999462463972,
      "learning_rate": 1.999888164559743e-06,
      "loss": 1.4215,
      "step": 2828
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.047703148420556,
      "learning_rate": 1.9998880851749335e-06,
      "loss": 1.4627,
      "step": 2829
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.422103915078788,
      "learning_rate": 1.99988800576196e-06,
      "loss": 1.5064,
      "step": 2830
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.927470474587449,
      "learning_rate": 1.999887926320823e-06,
      "loss": 1.4422,
      "step": 2831
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.8891825059531095,
      "learning_rate": 1.999887846851523e-06,
      "loss": 1.3049,
      "step": 2832
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.89929973889463,
      "learning_rate": 1.9998877673540592e-06,
      "loss": 1.4641,
      "step": 2833
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.330474087668212,
      "learning_rate": 1.999887687828432e-06,
      "loss": 1.2395,
      "step": 2834
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.689316103906117,
      "learning_rate": 1.999887608274641e-06,
      "loss": 1.5746,
      "step": 2835
    },
    {
      "epoch": 0.02,
      "grad_norm": 30.022522598437778,
      "learning_rate": 1.999887528692687e-06,
      "loss": 1.552,
      "step": 2836
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.979619290046022,
      "learning_rate": 1.999887449082569e-06,
      "loss": 1.4355,
      "step": 2837
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5639834865515185,
      "learning_rate": 1.9998873694442878e-06,
      "loss": 1.383,
      "step": 2838
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.572884425917966,
      "learning_rate": 1.9998872897778427e-06,
      "loss": 1.4367,
      "step": 2839
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.935183007183054,
      "learning_rate": 1.9998872100832346e-06,
      "loss": 1.2201,
      "step": 2840
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.211909738145334,
      "learning_rate": 1.9998871303604627e-06,
      "loss": 1.3496,
      "step": 2841
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5122328941008,
      "learning_rate": 1.9998870506095274e-06,
      "loss": 1.2845,
      "step": 2842
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.924447610654198,
      "learning_rate": 1.9998869708304286e-06,
      "loss": 1.4161,
      "step": 2843
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.2054369143001145,
      "learning_rate": 1.9998868910231665e-06,
      "loss": 1.5187,
      "step": 2844
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.861035173442859,
      "learning_rate": 1.9998868111877404e-06,
      "loss": 1.3305,
      "step": 2845
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.63315608539278,
      "learning_rate": 1.9998867313241514e-06,
      "loss": 1.3722,
      "step": 2846
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.566229003342197,
      "learning_rate": 1.9998866514323985e-06,
      "loss": 1.3612,
      "step": 2847
    },
    {
      "epoch": 0.02,
      "eval_loss": 1.6186871528625488,
      "eval_runtime": 4.6182,
      "eval_samples_per_second": 1.949,
      "eval_steps_per_second": 1.083,
      "step": 2847
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.9773867457031615,
      "learning_rate": 1.9998865715124822e-06,
      "loss": 1.3684,
      "step": 2848
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.776322816477926,
      "learning_rate": 1.9998864915644025e-06,
      "loss": 1.355,
      "step": 2849
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.200825295575172,
      "learning_rate": 1.9998864115881594e-06,
      "loss": 1.4849,
      "step": 2850
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.669774310904668,
      "learning_rate": 1.999886331583753e-06,
      "loss": 1.439,
      "step": 2851
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.961641023055851,
      "learning_rate": 1.9998862515511824e-06,
      "loss": 1.3667,
      "step": 2852
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.822249685134192,
      "learning_rate": 1.999886171490449e-06,
      "loss": 1.3395,
      "step": 2853
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.098165282053863,
      "learning_rate": 1.999886091401552e-06,
      "loss": 1.4478,
      "step": 2854
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.297976415034525,
      "learning_rate": 1.999886011284491e-06,
      "loss": 1.3084,
      "step": 2855
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.60213897121254,
      "learning_rate": 1.9998859311392675e-06,
      "loss": 1.3949,
      "step": 2856
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.128426005267793,
      "learning_rate": 1.99988585096588e-06,
      "loss": 1.5439,
      "step": 2857
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.238279682926694,
      "learning_rate": 1.999885770764329e-06,
      "loss": 1.3442,
      "step": 2858
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.454111710242467,
      "learning_rate": 1.9998856905346142e-06,
      "loss": 1.2708,
      "step": 2859
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.597041686087069,
      "learning_rate": 1.9998856102767365e-06,
      "loss": 1.4927,
      "step": 2860
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.9566004997832325,
      "learning_rate": 1.9998855299906953e-06,
      "loss": 1.4864,
      "step": 2861
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.803022534988008,
      "learning_rate": 1.9998854496764907e-06,
      "loss": 1.4479,
      "step": 2862
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.3284227061764025,
      "learning_rate": 1.9998853693341226e-06,
      "loss": 1.3775,
      "step": 2863
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.449051605434009,
      "learning_rate": 1.9998852889635907e-06,
      "loss": 1.3944,
      "step": 2864
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.524264356204093,
      "learning_rate": 1.999885208564896e-06,
      "loss": 1.3925,
      "step": 2865
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.576738634642325,
      "learning_rate": 1.9998851281380375e-06,
      "loss": 1.4023,
      "step": 2866
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5128730915629305,
      "learning_rate": 1.999885047683015e-06,
      "loss": 1.3308,
      "step": 2867
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.774177800040769,
      "learning_rate": 1.99988496719983e-06,
      "loss": 1.3723,
      "step": 2868
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.841933746799073,
      "learning_rate": 1.999884886688481e-06,
      "loss": 1.4242,
      "step": 2869
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.190890743181518,
      "learning_rate": 1.999884806148969e-06,
      "loss": 1.4457,
      "step": 2870
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.389488660525391,
      "learning_rate": 1.999884725581293e-06,
      "loss": 1.3262,
      "step": 2871
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.9366007381855725,
      "learning_rate": 1.999884644985454e-06,
      "loss": 1.4215,
      "step": 2872
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.51134223312188,
      "learning_rate": 1.9998845643614514e-06,
      "loss": 1.4528,
      "step": 2873
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5911234789565345,
      "learning_rate": 1.9998844837092857e-06,
      "loss": 1.3725,
      "step": 2874
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.714425532161108,
      "learning_rate": 1.9998844030289562e-06,
      "loss": 1.3326,
      "step": 2875
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.194761253541817,
      "learning_rate": 1.9998843223204633e-06,
      "loss": 1.4908,
      "step": 2876
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.504642312252081,
      "learning_rate": 1.999884241583807e-06,
      "loss": 1.3287,
      "step": 2877
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.794699180375288,
      "learning_rate": 1.9998841608189875e-06,
      "loss": 1.4244,
      "step": 2878
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.751999842845239,
      "learning_rate": 1.9998840800260043e-06,
      "loss": 1.454,
      "step": 2879
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.726264353641071,
      "learning_rate": 1.999883999204858e-06,
      "loss": 1.3897,
      "step": 2880
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.729803974265776,
      "learning_rate": 1.999883918355548e-06,
      "loss": 1.5002,
      "step": 2881
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.285212933171045,
      "learning_rate": 1.999883837478075e-06,
      "loss": 1.6833,
      "step": 2882
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.596625945889018,
      "learning_rate": 1.999883756572438e-06,
      "loss": 1.3805,
      "step": 2883
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.0059586218503105,
      "learning_rate": 1.999883675638638e-06,
      "loss": 1.4069,
      "step": 2884
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.8719439751839575,
      "learning_rate": 1.999883594676675e-06,
      "loss": 1.4273,
      "step": 2885
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.836717800773741,
      "learning_rate": 1.999883513686548e-06,
      "loss": 1.5191,
      "step": 2886
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.448400261385887,
      "learning_rate": 1.9998834326682575e-06,
      "loss": 1.5117,
      "step": 2887
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.341830112851468,
      "learning_rate": 1.9998833516218043e-06,
      "loss": 1.3386,
      "step": 2888
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.330108245622447,
      "learning_rate": 1.999883270547187e-06,
      "loss": 1.3865,
      "step": 2889
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.161889492402766,
      "learning_rate": 1.9998831894444064e-06,
      "loss": 1.4923,
      "step": 2890
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.583340804795185,
      "learning_rate": 1.9998831083134626e-06,
      "loss": 1.4128,
      "step": 2891
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.475403361495518,
      "learning_rate": 1.9998830271543557e-06,
      "loss": 1.2633,
      "step": 2892
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.429544299978376,
      "learning_rate": 1.999882945967085e-06,
      "loss": 1.2271,
      "step": 2893
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.924573918325059,
      "learning_rate": 1.999882864751651e-06,
      "loss": 1.5329,
      "step": 2894
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.72164854751247,
      "learning_rate": 1.9998827835080538e-06,
      "loss": 1.3411,
      "step": 2895
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.513620466728661,
      "learning_rate": 1.9998827022362932e-06,
      "loss": 1.557,
      "step": 2896
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.893569608594087,
      "learning_rate": 1.9998826209363693e-06,
      "loss": 1.4545,
      "step": 2897
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.59988609368051,
      "learning_rate": 1.999882539608282e-06,
      "loss": 1.4207,
      "step": 2898
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.289259835995451,
      "learning_rate": 1.9998824582520306e-06,
      "loss": 1.4633,
      "step": 2899
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.600672131161404,
      "learning_rate": 1.999882376867617e-06,
      "loss": 1.2747,
      "step": 2900
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.73209077596222,
      "learning_rate": 1.9998822954550396e-06,
      "loss": 1.4534,
      "step": 2901
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.649416419045721,
      "learning_rate": 1.9998822140142985e-06,
      "loss": 1.423,
      "step": 2902
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.815284149190514,
      "learning_rate": 1.999882132545394e-06,
      "loss": 1.1985,
      "step": 2903
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.993787069518763,
      "learning_rate": 1.999882051048327e-06,
      "loss": 1.3226,
      "step": 2904
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.30556271239256,
      "learning_rate": 1.9998819695230956e-06,
      "loss": 1.6076,
      "step": 2905
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.111676758762733,
      "learning_rate": 1.9998818879697013e-06,
      "loss": 1.4156,
      "step": 2906
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.57532368590634,
      "learning_rate": 1.999881806388144e-06,
      "loss": 1.5127,
      "step": 2907
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.097665766053441,
      "learning_rate": 1.9998817247784227e-06,
      "loss": 1.2153,
      "step": 2908
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.7664796065004404,
      "learning_rate": 1.9998816431405386e-06,
      "loss": 1.4997,
      "step": 2909
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.707849161238309,
      "learning_rate": 1.999881561474491e-06,
      "loss": 1.3614,
      "step": 2910
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.715418586973402,
      "learning_rate": 1.99988147978028e-06,
      "loss": 1.3648,
      "step": 2911
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.698559700737574,
      "learning_rate": 1.9998813980579055e-06,
      "loss": 1.4627,
      "step": 2912
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.492256458663923,
      "learning_rate": 1.999881316307368e-06,
      "loss": 1.3053,
      "step": 2913
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.154059898049573,
      "learning_rate": 1.9998812345286667e-06,
      "loss": 1.1044,
      "step": 2914
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.62208125851778,
      "learning_rate": 1.999881152721803e-06,
      "loss": 1.3578,
      "step": 2915
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.66847610055548,
      "learning_rate": 1.999881070886775e-06,
      "loss": 1.5454,
      "step": 2916
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.317475104898607,
      "learning_rate": 1.999880989023584e-06,
      "loss": 1.3935,
      "step": 2917
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.527930294875366,
      "learning_rate": 1.99988090713223e-06,
      "loss": 1.4274,
      "step": 2918
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.416272106126581,
      "learning_rate": 1.9998808252127123e-06,
      "loss": 1.3224,
      "step": 2919
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.316728315979006,
      "learning_rate": 1.9998807432650313e-06,
      "loss": 1.5215,
      "step": 2920
    },
    {
      "epoch": 0.02,
      "eval_loss": 1.6199383735656738,
      "eval_runtime": 4.6314,
      "eval_samples_per_second": 1.943,
      "eval_steps_per_second": 1.08,
      "step": 2920
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.490493646901724,
      "learning_rate": 1.999880661289187e-06,
      "loss": 1.4416,
      "step": 2921
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.624580242853857,
      "learning_rate": 1.9998805792851795e-06,
      "loss": 1.3179,
      "step": 2922
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.466914354748946,
      "learning_rate": 1.9998804972530087e-06,
      "loss": 1.6161,
      "step": 2923
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.764398192424773,
      "learning_rate": 1.9998804151926745e-06,
      "loss": 1.5612,
      "step": 2924
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.270011750308185,
      "learning_rate": 1.9998803331041772e-06,
      "loss": 1.3574,
      "step": 2925
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.688420881807707,
      "learning_rate": 1.999880250987516e-06,
      "loss": 1.4334,
      "step": 2926
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.649454207981505,
      "learning_rate": 1.999880168842692e-06,
      "loss": 1.3212,
      "step": 2927
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.156111274716404,
      "learning_rate": 1.9998800866697046e-06,
      "loss": 1.1788,
      "step": 2928
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.189510589727771,
      "learning_rate": 1.999880004468554e-06,
      "loss": 1.5181,
      "step": 2929
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.20778314788957,
      "learning_rate": 1.99987992223924e-06,
      "loss": 1.4894,
      "step": 2930
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.613692977472925,
      "learning_rate": 1.9998798399817628e-06,
      "loss": 1.3359,
      "step": 2931
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.45441252637321,
      "learning_rate": 1.999879757696122e-06,
      "loss": 1.3484,
      "step": 2932
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.84796908759116,
      "learning_rate": 1.9998796753823182e-06,
      "loss": 1.3102,
      "step": 2933
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.057071265216301,
      "learning_rate": 1.9998795930403515e-06,
      "loss": 1.4964,
      "step": 2934
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.65737240395791,
      "learning_rate": 1.9998795106702213e-06,
      "loss": 1.3021,
      "step": 2935
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.466035985587055,
      "learning_rate": 1.9998794282719273e-06,
      "loss": 1.5032,
      "step": 2936
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.856160311135987,
      "learning_rate": 1.9998793458454702e-06,
      "loss": 1.5538,
      "step": 2937
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.610346348605172,
      "learning_rate": 1.99987926339085e-06,
      "loss": 1.462,
      "step": 2938
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.928466235921176,
      "learning_rate": 1.9998791809080667e-06,
      "loss": 1.3877,
      "step": 2939
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.973945408707448,
      "learning_rate": 1.99987909839712e-06,
      "loss": 1.3815,
      "step": 2940
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.625537540163681,
      "learning_rate": 1.99987901585801e-06,
      "loss": 1.2582,
      "step": 2941
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.477355848214243,
      "learning_rate": 1.9998789332907363e-06,
      "loss": 1.236,
      "step": 2942
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.492618763572616,
      "learning_rate": 1.9998788506953e-06,
      "loss": 1.3155,
      "step": 2943
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.602003607577262,
      "learning_rate": 1.9998787680717002e-06,
      "loss": 1.2852,
      "step": 2944
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.567502332448367,
      "learning_rate": 1.999878685419937e-06,
      "loss": 1.4841,
      "step": 2945
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.9405542923073815,
      "learning_rate": 1.9998786027400105e-06,
      "loss": 1.3458,
      "step": 2946
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.6929594602807025,
      "learning_rate": 1.999878520031921e-06,
      "loss": 1.4189,
      "step": 2947
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.197148859419166,
      "learning_rate": 1.9998784372956684e-06,
      "loss": 1.3436,
      "step": 2948
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.599441454159472,
      "learning_rate": 1.999878354531252e-06,
      "loss": 1.3806,
      "step": 2949
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.615989461212095,
      "learning_rate": 1.9998782717386726e-06,
      "loss": 1.3673,
      "step": 2950
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.9156780589895295,
      "learning_rate": 1.9998781889179302e-06,
      "loss": 1.3702,
      "step": 2951
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.29684223795888,
      "learning_rate": 1.9998781060690244e-06,
      "loss": 1.3308,
      "step": 2952
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.517575497946087,
      "learning_rate": 1.999878023191955e-06,
      "loss": 1.3706,
      "step": 2953
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.767296353617181,
      "learning_rate": 1.999877940286723e-06,
      "loss": 1.5224,
      "step": 2954
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.751314082046642,
      "learning_rate": 1.9998778573533272e-06,
      "loss": 1.3569,
      "step": 2955
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.40734821128334,
      "learning_rate": 1.9998777743917686e-06,
      "loss": 1.4663,
      "step": 2956
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.350543171262743,
      "learning_rate": 1.9998776914020465e-06,
      "loss": 1.2895,
      "step": 2957
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.6281115386375395,
      "learning_rate": 1.9998776083841614e-06,
      "loss": 1.5124,
      "step": 2958
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.2232144573948185,
      "learning_rate": 1.9998775253381125e-06,
      "loss": 1.2376,
      "step": 2959
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.689282233407712,
      "learning_rate": 1.999877442263901e-06,
      "loss": 1.5472,
      "step": 2960
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.917409633814714,
      "learning_rate": 1.999877359161526e-06,
      "loss": 1.4326,
      "step": 2961
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.994380191725462,
      "learning_rate": 1.9998772760309877e-06,
      "loss": 1.3549,
      "step": 2962
    },
    {
      "epoch": 0.02,
      "grad_norm": 14.47259236356199,
      "learning_rate": 1.9998771928722868e-06,
      "loss": 1.4232,
      "step": 2963
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.656592137584895,
      "learning_rate": 1.999877109685422e-06,
      "loss": 1.3973,
      "step": 2964
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.6307964918603615,
      "learning_rate": 1.9998770264703942e-06,
      "loss": 1.3096,
      "step": 2965
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5164633394118825,
      "learning_rate": 1.9998769432272035e-06,
      "loss": 1.3215,
      "step": 2966
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5103839413771185,
      "learning_rate": 1.9998768599558493e-06,
      "loss": 1.3789,
      "step": 2967
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.2216327904598385,
      "learning_rate": 1.9998767766563317e-06,
      "loss": 1.3297,
      "step": 2968
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.059047029865863,
      "learning_rate": 1.999876693328651e-06,
      "loss": 1.3984,
      "step": 2969
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.700581236103446,
      "learning_rate": 1.9998766099728075e-06,
      "loss": 1.5166,
      "step": 2970
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.854212931369697,
      "learning_rate": 1.9998765265888e-06,
      "loss": 1.4143,
      "step": 2971
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.725119262001126,
      "learning_rate": 1.99987644317663e-06,
      "loss": 1.178,
      "step": 2972
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.370144258521158,
      "learning_rate": 1.9998763597362965e-06,
      "loss": 1.3893,
      "step": 2973
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.638058498542562,
      "learning_rate": 1.9998762762678e-06,
      "loss": 1.4689,
      "step": 2974
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.213106554526003,
      "learning_rate": 1.9998761927711402e-06,
      "loss": 1.3459,
      "step": 2975
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.858477665120681,
      "learning_rate": 1.9998761092463173e-06,
      "loss": 1.4884,
      "step": 2976
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.570024440771522,
      "learning_rate": 1.999876025693331e-06,
      "loss": 1.469,
      "step": 2977
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.16955091971151,
      "learning_rate": 1.9998759421121818e-06,
      "loss": 1.4388,
      "step": 2978
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.903458987120334,
      "learning_rate": 1.999875858502869e-06,
      "loss": 1.3786,
      "step": 2979
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.529203011610496,
      "learning_rate": 1.9998757748653933e-06,
      "loss": 1.2684,
      "step": 2980
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.729523457919252,
      "learning_rate": 1.9998756911997546e-06,
      "loss": 1.5905,
      "step": 2981
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.582684926177018,
      "learning_rate": 1.9998756075059525e-06,
      "loss": 1.598,
      "step": 2982
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.659482522086951,
      "learning_rate": 1.9998755237839874e-06,
      "loss": 1.3603,
      "step": 2983
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.628018340481434,
      "learning_rate": 1.999875440033859e-06,
      "loss": 1.5392,
      "step": 2984
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.03947825852203,
      "learning_rate": 1.9998753562555673e-06,
      "loss": 1.47,
      "step": 2985
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.908146139932012,
      "learning_rate": 1.9998752724491127e-06,
      "loss": 1.5795,
      "step": 2986
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.642150400889997,
      "learning_rate": 1.9998751886144948e-06,
      "loss": 1.4156,
      "step": 2987
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.580016701743678,
      "learning_rate": 1.999875104751714e-06,
      "loss": 1.3989,
      "step": 2988
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.60906478003677,
      "learning_rate": 1.99987502086077e-06,
      "loss": 1.4226,
      "step": 2989
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.563599437751002,
      "learning_rate": 1.9998749369416624e-06,
      "loss": 1.5099,
      "step": 2990
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.223270236357775,
      "learning_rate": 1.9998748529943916e-06,
      "loss": 1.3115,
      "step": 2991
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.633656419315019,
      "learning_rate": 1.9998747690189583e-06,
      "loss": 1.4029,
      "step": 2992
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.630285775575748,
      "learning_rate": 1.9998746850153614e-06,
      "loss": 1.3736,
      "step": 2993
    },
    {
      "epoch": 0.02,
      "eval_loss": 1.6208245754241943,
      "eval_runtime": 4.6165,
      "eval_samples_per_second": 1.95,
      "eval_steps_per_second": 1.083,
      "step": 2993
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.936246851895618,
      "learning_rate": 1.9998746009836016e-06,
      "loss": 1.406,
      "step": 2994
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.149643170506049,
      "learning_rate": 1.9998745169236784e-06,
      "loss": 1.3712,
      "step": 2995
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.788568030128947,
      "learning_rate": 1.999874432835592e-06,
      "loss": 1.4386,
      "step": 2996
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.578352571887644,
      "learning_rate": 1.999874348719343e-06,
      "loss": 1.3946,
      "step": 2997
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.124704832034737,
      "learning_rate": 1.9998742645749303e-06,
      "loss": 1.301,
      "step": 2998
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.405037637799255,
      "learning_rate": 1.999874180402355e-06,
      "loss": 1.4424,
      "step": 2999
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.6389112469188,
      "learning_rate": 1.999874096201616e-06,
      "loss": 1.4466,
      "step": 3000
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.19152230930798,
      "learning_rate": 1.9998740119727144e-06,
      "loss": 1.4456,
      "step": 3001
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.121951342555157,
      "learning_rate": 1.999873927715649e-06,
      "loss": 1.3808,
      "step": 3002
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.607444674665859,
      "learning_rate": 1.999873843430421e-06,
      "loss": 1.4636,
      "step": 3003
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.46848570773085,
      "learning_rate": 1.9998737591170298e-06,
      "loss": 1.3631,
      "step": 3004
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.330823996962009,
      "learning_rate": 1.9998736747754753e-06,
      "loss": 1.3945,
      "step": 3005
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.168688715915758,
      "learning_rate": 1.999873590405758e-06,
      "loss": 1.265,
      "step": 3006
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.21101121326567,
      "learning_rate": 1.9998735060078773e-06,
      "loss": 1.4035,
      "step": 3007
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.9325677986365335,
      "learning_rate": 1.999873421581834e-06,
      "loss": 1.5848,
      "step": 3008
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.726896704178509,
      "learning_rate": 1.999873337127627e-06,
      "loss": 1.3209,
      "step": 3009
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.923881575075372,
      "learning_rate": 1.999873252645257e-06,
      "loss": 1.3872,
      "step": 3010
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.405371972429703,
      "learning_rate": 1.999873168134724e-06,
      "loss": 1.3634,
      "step": 3011
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.953407177301594,
      "learning_rate": 1.999873083596028e-06,
      "loss": 1.4459,
      "step": 3012
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.432312541640287,
      "learning_rate": 1.9998729990291685e-06,
      "loss": 1.36,
      "step": 3013
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.6106671367098855,
      "learning_rate": 1.9998729144341462e-06,
      "loss": 1.3693,
      "step": 3014
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.363880892273669,
      "learning_rate": 1.999872829810961e-06,
      "loss": 1.4125,
      "step": 3015
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.671858356354371,
      "learning_rate": 1.9998727451596126e-06,
      "loss": 1.4628,
      "step": 3016
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.1327904162947835,
      "learning_rate": 1.999872660480101e-06,
      "loss": 1.484,
      "step": 3017
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.726123143448659,
      "learning_rate": 1.999872575772426e-06,
      "loss": 1.3706,
      "step": 3018
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.6201517878177984,
      "learning_rate": 1.9998724910365884e-06,
      "loss": 1.3531,
      "step": 3019
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.7617011200342585,
      "learning_rate": 1.9998724062725877e-06,
      "loss": 1.292,
      "step": 3020
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.684197716362908,
      "learning_rate": 1.999872321480424e-06,
      "loss": 1.4344,
      "step": 3021
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.683892557856816,
      "learning_rate": 1.9998722366600968e-06,
      "loss": 1.5146,
      "step": 3022
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.659566302966707,
      "learning_rate": 1.9998721518116066e-06,
      "loss": 1.4766,
      "step": 3023
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.56223709999532,
      "learning_rate": 1.9998720669349535e-06,
      "loss": 1.4421,
      "step": 3024
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.040663438792937,
      "learning_rate": 1.9998719820301373e-06,
      "loss": 1.4359,
      "step": 3025
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.513110115307069,
      "learning_rate": 1.9998718970971578e-06,
      "loss": 1.4452,
      "step": 3026
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.912192039188993,
      "learning_rate": 1.9998718121360156e-06,
      "loss": 1.4882,
      "step": 3027
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.525564198825131,
      "learning_rate": 1.99987172714671e-06,
      "loss": 1.2955,
      "step": 3028
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.369450073058397,
      "learning_rate": 1.999871642129242e-06,
      "loss": 1.4109,
      "step": 3029
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.399069353447917,
      "learning_rate": 1.9998715570836104e-06,
      "loss": 1.5341,
      "step": 3030
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.863376705625461,
      "learning_rate": 1.999871472009816e-06,
      "loss": 1.3178,
      "step": 3031
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.999004673955154,
      "learning_rate": 1.999871386907858e-06,
      "loss": 1.358,
      "step": 3032
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.667839262748732,
      "learning_rate": 1.9998713017777373e-06,
      "loss": 1.4091,
      "step": 3033
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5743647889890475,
      "learning_rate": 1.9998712166194537e-06,
      "loss": 1.3385,
      "step": 3034
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.975891971893521,
      "learning_rate": 1.9998711314330067e-06,
      "loss": 1.4072,
      "step": 3035
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.830200195962722,
      "learning_rate": 1.999871046218397e-06,
      "loss": 1.4913,
      "step": 3036
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.805815436884879,
      "learning_rate": 1.9998709609756238e-06,
      "loss": 1.4484,
      "step": 3037
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.600176911297759,
      "learning_rate": 1.9998708757046882e-06,
      "loss": 1.2973,
      "step": 3038
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.060093592105702,
      "learning_rate": 1.9998707904055893e-06,
      "loss": 1.1942,
      "step": 3039
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.802674568723147,
      "learning_rate": 1.9998707050783273e-06,
      "loss": 1.5334,
      "step": 3040
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.426696065645248,
      "learning_rate": 1.9998706197229023e-06,
      "loss": 1.4872,
      "step": 3041
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.898642916611915,
      "learning_rate": 1.999870534339314e-06,
      "loss": 1.5561,
      "step": 3042
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5979860791423635,
      "learning_rate": 1.999870448927563e-06,
      "loss": 1.5231,
      "step": 3043
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.535954226116485,
      "learning_rate": 1.999870363487649e-06,
      "loss": 1.3058,
      "step": 3044
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.745795440686125,
      "learning_rate": 1.9998702780195716e-06,
      "loss": 1.4387,
      "step": 3045
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.635960415984835,
      "learning_rate": 1.999870192523332e-06,
      "loss": 1.4911,
      "step": 3046
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.759909878239948,
      "learning_rate": 1.9998701069989287e-06,
      "loss": 1.4024,
      "step": 3047
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.802542836315548,
      "learning_rate": 1.9998700214463623e-06,
      "loss": 1.4512,
      "step": 3048
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.378746183379735,
      "learning_rate": 1.9998699358656334e-06,
      "loss": 1.5516,
      "step": 3049
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.618905225926722,
      "learning_rate": 1.999869850256741e-06,
      "loss": 1.3609,
      "step": 3050
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.905191662886102,
      "learning_rate": 1.9998697646196856e-06,
      "loss": 1.5205,
      "step": 3051
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.391956056937744,
      "learning_rate": 1.9998696789544677e-06,
      "loss": 1.3854,
      "step": 3052
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.57834379820915,
      "learning_rate": 1.9998695932610863e-06,
      "loss": 1.4971,
      "step": 3053
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.9358795449257,
      "learning_rate": 1.9998695075395423e-06,
      "loss": 1.3831,
      "step": 3054
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.168823055094264,
      "learning_rate": 1.999869421789835e-06,
      "loss": 1.2735,
      "step": 3055
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.134532360101292,
      "learning_rate": 1.9998693360119646e-06,
      "loss": 1.3684,
      "step": 3056
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.281772250542135,
      "learning_rate": 1.9998692502059317e-06,
      "loss": 1.25,
      "step": 3057
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.862624797364669,
      "learning_rate": 1.9998691643717353e-06,
      "loss": 1.4888,
      "step": 3058
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5369738915051965,
      "learning_rate": 1.999869078509376e-06,
      "loss": 1.3752,
      "step": 3059
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.264457712237391,
      "learning_rate": 1.999868992618854e-06,
      "loss": 1.3389,
      "step": 3060
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.148997720929055,
      "learning_rate": 1.9998689067001686e-06,
      "loss": 1.4634,
      "step": 3061
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.69514649110256,
      "learning_rate": 1.9998688207533207e-06,
      "loss": 1.4781,
      "step": 3062
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.906133670567751,
      "learning_rate": 1.9998687347783098e-06,
      "loss": 1.2841,
      "step": 3063
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.450305903996531,
      "learning_rate": 1.9998686487751354e-06,
      "loss": 1.3008,
      "step": 3064
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.9761783730067926,
      "learning_rate": 1.9998685627437985e-06,
      "loss": 1.1917,
      "step": 3065
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.90862469089785,
      "learning_rate": 1.9998684766842985e-06,
      "loss": 1.3536,
      "step": 3066
    },
    {
      "epoch": 0.02,
      "eval_loss": 1.614479660987854,
      "eval_runtime": 4.6281,
      "eval_samples_per_second": 1.945,
      "eval_steps_per_second": 1.08,
      "step": 3066
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.514246443648035,
      "learning_rate": 1.999868390596635e-06,
      "loss": 1.46,
      "step": 3067
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.795614462660274,
      "learning_rate": 1.9998683044808097e-06,
      "loss": 1.4595,
      "step": 3068
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.018533958527286,
      "learning_rate": 1.9998682183368203e-06,
      "loss": 1.3904,
      "step": 3069
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.942520732496524,
      "learning_rate": 1.999868132164669e-06,
      "loss": 1.4983,
      "step": 3070
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.777495716431802,
      "learning_rate": 1.999868045964354e-06,
      "loss": 1.4747,
      "step": 3071
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.426971447313713,
      "learning_rate": 1.999867959735876e-06,
      "loss": 1.2999,
      "step": 3072
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.039700517061406,
      "learning_rate": 1.999867873479235e-06,
      "loss": 1.3682,
      "step": 3073
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.018072800299674,
      "learning_rate": 1.9998677871944316e-06,
      "loss": 1.4626,
      "step": 3074
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.620448908678396,
      "learning_rate": 1.9998677008814647e-06,
      "loss": 1.3042,
      "step": 3075
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.309072084405069,
      "learning_rate": 1.999867614540335e-06,
      "loss": 1.3497,
      "step": 3076
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.1018581186592025,
      "learning_rate": 1.9998675281710427e-06,
      "loss": 1.333,
      "step": 3077
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.945567567355229,
      "learning_rate": 1.999867441773587e-06,
      "loss": 1.2968,
      "step": 3078
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.406409232689529,
      "learning_rate": 1.9998673553479687e-06,
      "loss": 1.4256,
      "step": 3079
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.218194975860625,
      "learning_rate": 1.9998672688941872e-06,
      "loss": 1.4264,
      "step": 3080
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.739664546488815,
      "learning_rate": 1.999867182412243e-06,
      "loss": 1.355,
      "step": 3081
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.91149132142301,
      "learning_rate": 1.9998670959021357e-06,
      "loss": 1.524,
      "step": 3082
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.432247494986651,
      "learning_rate": 1.9998670093638656e-06,
      "loss": 1.2699,
      "step": 3083
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.31888360981259,
      "learning_rate": 1.9998669227974326e-06,
      "loss": 1.3485,
      "step": 3084
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.766908478434986,
      "learning_rate": 1.9998668362028365e-06,
      "loss": 1.3453,
      "step": 3085
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.684318593661933,
      "learning_rate": 1.9998667495800775e-06,
      "loss": 1.3912,
      "step": 3086
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.707084972080545,
      "learning_rate": 1.9998666629291554e-06,
      "loss": 1.4999,
      "step": 3087
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.658271583331655,
      "learning_rate": 1.999866576250071e-06,
      "loss": 1.3584,
      "step": 3088
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.8632287556229645,
      "learning_rate": 1.999866489542823e-06,
      "loss": 1.4268,
      "step": 3089
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.305841703345282,
      "learning_rate": 1.9998664028074126e-06,
      "loss": 1.425,
      "step": 3090
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.537852499872765,
      "learning_rate": 1.999866316043839e-06,
      "loss": 1.3343,
      "step": 3091
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.525709370028815,
      "learning_rate": 1.9998662292521023e-06,
      "loss": 1.4023,
      "step": 3092
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.566090943972325,
      "learning_rate": 1.999866142432203e-06,
      "loss": 1.4643,
      "step": 3093
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.2695795542821875,
      "learning_rate": 1.9998660555841405e-06,
      "loss": 1.2825,
      "step": 3094
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.396645953705008,
      "learning_rate": 1.9998659687079157e-06,
      "loss": 1.2208,
      "step": 3095
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.585920026446324,
      "learning_rate": 1.9998658818035275e-06,
      "loss": 1.3192,
      "step": 3096
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.724254453766402,
      "learning_rate": 1.9998657948709768e-06,
      "loss": 1.3259,
      "step": 3097
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.546497072265447,
      "learning_rate": 1.9998657079102626e-06,
      "loss": 1.4462,
      "step": 3098
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.067058100414485,
      "learning_rate": 1.999865620921386e-06,
      "loss": 1.371,
      "step": 3099
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.680817047547335,
      "learning_rate": 1.9998655339043465e-06,
      "loss": 1.382,
      "step": 3100
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.684807070794312,
      "learning_rate": 1.999865446859144e-06,
      "loss": 1.3792,
      "step": 3101
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.108394614810223,
      "learning_rate": 1.999865359785779e-06,
      "loss": 1.3695,
      "step": 3102
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.678775301437812,
      "learning_rate": 1.99986527268425e-06,
      "loss": 1.3234,
      "step": 3103
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.198154854127655,
      "learning_rate": 1.999865185554559e-06,
      "loss": 1.3228,
      "step": 3104
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.954029857664215,
      "learning_rate": 1.9998650983967053e-06,
      "loss": 1.2434,
      "step": 3105
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.716268546944123,
      "learning_rate": 1.999865011210688e-06,
      "loss": 1.3835,
      "step": 3106
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.706544890974926,
      "learning_rate": 1.9998649239965085e-06,
      "loss": 1.4839,
      "step": 3107
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.0878170468870545,
      "learning_rate": 1.9998648367541656e-06,
      "loss": 1.5127,
      "step": 3108
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.546409694838889,
      "learning_rate": 1.99986474948366e-06,
      "loss": 1.4668,
      "step": 3109
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.262790272972708,
      "learning_rate": 1.9998646621849916e-06,
      "loss": 1.3487,
      "step": 3110
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.437228037164254,
      "learning_rate": 1.9998645748581606e-06,
      "loss": 1.4362,
      "step": 3111
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.591524369547653,
      "learning_rate": 1.9998644875031665e-06,
      "loss": 1.3518,
      "step": 3112
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.943081436316897,
      "learning_rate": 1.9998644001200095e-06,
      "loss": 1.2977,
      "step": 3113
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.745398436424019,
      "learning_rate": 1.9998643127086894e-06,
      "loss": 1.2858,
      "step": 3114
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.451084562147302,
      "learning_rate": 1.9998642252692072e-06,
      "loss": 1.4795,
      "step": 3115
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.642520886807483,
      "learning_rate": 1.9998641378015616e-06,
      "loss": 1.5245,
      "step": 3116
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.829976676988873,
      "learning_rate": 1.9998640503057534e-06,
      "loss": 1.3036,
      "step": 3117
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.272610276691716,
      "learning_rate": 1.9998639627817822e-06,
      "loss": 1.4307,
      "step": 3118
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.279012748336821,
      "learning_rate": 1.999863875229648e-06,
      "loss": 1.3246,
      "step": 3119
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.017262661372447,
      "learning_rate": 1.999863787649351e-06,
      "loss": 1.4561,
      "step": 3120
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.159770574845129,
      "learning_rate": 1.9998637000408915e-06,
      "loss": 1.4325,
      "step": 3121
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.138809355786813,
      "learning_rate": 1.9998636124042688e-06,
      "loss": 1.3595,
      "step": 3122
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.411909404561028,
      "learning_rate": 1.9998635247394834e-06,
      "loss": 1.2877,
      "step": 3123
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.8356590420018675,
      "learning_rate": 1.999863437046535e-06,
      "loss": 1.382,
      "step": 3124
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.438539228016791,
      "learning_rate": 1.999863349325424e-06,
      "loss": 1.2727,
      "step": 3125
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.290518205135856,
      "learning_rate": 1.9998632615761503e-06,
      "loss": 1.2096,
      "step": 3126
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.869806082110767,
      "learning_rate": 1.9998631737987134e-06,
      "loss": 1.3335,
      "step": 3127
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.001272070760005,
      "learning_rate": 1.999863085993114e-06,
      "loss": 1.48,
      "step": 3128
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.588174640600946,
      "learning_rate": 1.999862998159352e-06,
      "loss": 1.3918,
      "step": 3129
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.48282878680433,
      "learning_rate": 1.9998629102974263e-06,
      "loss": 1.3621,
      "step": 3130
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.327695042907354,
      "learning_rate": 1.9998628224073387e-06,
      "loss": 1.1881,
      "step": 3131
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.935240520013496,
      "learning_rate": 1.999862734489088e-06,
      "loss": 1.4896,
      "step": 3132
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.762415048958248,
      "learning_rate": 1.999862646542674e-06,
      "loss": 1.3664,
      "step": 3133
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.514983890100395,
      "learning_rate": 1.999862558568098e-06,
      "loss": 1.3976,
      "step": 3134
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.318544973311922,
      "learning_rate": 1.9998624705653586e-06,
      "loss": 1.3479,
      "step": 3135
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.831718076957779,
      "learning_rate": 1.999862382534457e-06,
      "loss": 1.4235,
      "step": 3136
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.289805977486961,
      "learning_rate": 1.9998622944753917e-06,
      "loss": 1.2588,
      "step": 3137
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.49073990213304,
      "learning_rate": 1.9998622063881643e-06,
      "loss": 1.4302,
      "step": 3138
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.548814969360347,
      "learning_rate": 1.999862118272774e-06,
      "loss": 1.3936,
      "step": 3139
    },
    {
      "epoch": 0.02,
      "eval_loss": 1.6162147521972656,
      "eval_runtime": 4.6224,
      "eval_samples_per_second": 1.947,
      "eval_steps_per_second": 1.082,
      "step": 3139
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.563259529239899,
      "learning_rate": 1.999862030129221e-06,
      "loss": 1.4137,
      "step": 3140
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.64213902825873,
      "learning_rate": 1.999861941957505e-06,
      "loss": 1.5853,
      "step": 3141
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.394890330300038,
      "learning_rate": 1.999861853757626e-06,
      "loss": 1.4709,
      "step": 3142
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.911886598072668,
      "learning_rate": 1.9998617655295847e-06,
      "loss": 1.4397,
      "step": 3143
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.65470961214868,
      "learning_rate": 1.9998616772733802e-06,
      "loss": 1.199,
      "step": 3144
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.636379664598463,
      "learning_rate": 1.999861588989013e-06,
      "loss": 1.2613,
      "step": 3145
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.9044581943490275,
      "learning_rate": 1.9998615006764835e-06,
      "loss": 1.4446,
      "step": 3146
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.966547861520061,
      "learning_rate": 1.999861412335791e-06,
      "loss": 1.3623,
      "step": 3147
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.124726053985599,
      "learning_rate": 1.9998613239669356e-06,
      "loss": 1.3474,
      "step": 3148
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.572636835170383,
      "learning_rate": 1.9998612355699174e-06,
      "loss": 1.3531,
      "step": 3149
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.618987542008811,
      "learning_rate": 1.9998611471447362e-06,
      "loss": 1.4206,
      "step": 3150
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.3621396822779746,
      "learning_rate": 1.999861058691393e-06,
      "loss": 1.4164,
      "step": 3151
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.341999064682612,
      "learning_rate": 1.999860970209886e-06,
      "loss": 1.5216,
      "step": 3152
    },
    {
      "epoch": 0.02,
      "grad_norm": 16.061509297130684,
      "learning_rate": 1.999860881700217e-06,
      "loss": 1.3329,
      "step": 3153
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.808958846677696,
      "learning_rate": 1.999860793162385e-06,
      "loss": 1.483,
      "step": 3154
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.114117981524813,
      "learning_rate": 1.9998607045963903e-06,
      "loss": 1.5337,
      "step": 3155
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.295040011290924,
      "learning_rate": 1.9998606160022328e-06,
      "loss": 1.3669,
      "step": 3156
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.704314977626543,
      "learning_rate": 1.9998605273799127e-06,
      "loss": 1.3311,
      "step": 3157
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.488680014172767,
      "learning_rate": 1.9998604387294296e-06,
      "loss": 1.4133,
      "step": 3158
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.104286348143494,
      "learning_rate": 1.999860350050784e-06,
      "loss": 1.5682,
      "step": 3159
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.485660767564034,
      "learning_rate": 1.9998602613439757e-06,
      "loss": 1.5762,
      "step": 3160
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.939741107373614,
      "learning_rate": 1.9998601726090045e-06,
      "loss": 1.4258,
      "step": 3161
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.881029948420793,
      "learning_rate": 1.9998600838458707e-06,
      "loss": 1.4878,
      "step": 3162
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.300753949774209,
      "learning_rate": 1.999859995054574e-06,
      "loss": 1.5591,
      "step": 3163
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.919255455770104,
      "learning_rate": 1.999859906235114e-06,
      "loss": 1.5089,
      "step": 3164
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.903957763395779,
      "learning_rate": 1.9998598173874925e-06,
      "loss": 1.3316,
      "step": 3165
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.786877472597078,
      "learning_rate": 1.9998597285117076e-06,
      "loss": 1.4113,
      "step": 3166
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.642407926930237,
      "learning_rate": 1.99985963960776e-06,
      "loss": 1.4368,
      "step": 3167
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.9722903457303715,
      "learning_rate": 1.9998595506756495e-06,
      "loss": 1.5279,
      "step": 3168
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.735779629402262,
      "learning_rate": 1.999859461715377e-06,
      "loss": 1.4509,
      "step": 3169
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.012513337480798,
      "learning_rate": 1.999859372726941e-06,
      "loss": 1.2937,
      "step": 3170
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.373925924257122,
      "learning_rate": 1.9998592837103425e-06,
      "loss": 1.3488,
      "step": 3171
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.813452991567045,
      "learning_rate": 1.9998591946655812e-06,
      "loss": 1.4029,
      "step": 3172
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.354945741505938,
      "learning_rate": 1.9998591055926574e-06,
      "loss": 1.2901,
      "step": 3173
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.4499436163682775,
      "learning_rate": 1.999859016491571e-06,
      "loss": 1.3946,
      "step": 3174
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.212636696076956,
      "learning_rate": 1.9998589273623216e-06,
      "loss": 1.4967,
      "step": 3175
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.480145318055372,
      "learning_rate": 1.99985883820491e-06,
      "loss": 1.3128,
      "step": 3176
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.7962291969592865,
      "learning_rate": 1.999858749019335e-06,
      "loss": 1.3714,
      "step": 3177
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.699063147883259,
      "learning_rate": 1.999858659805598e-06,
      "loss": 1.3319,
      "step": 3178
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.026292120171505,
      "learning_rate": 1.999858570563698e-06,
      "loss": 1.2571,
      "step": 3179
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.409840960626129,
      "learning_rate": 1.999858481293635e-06,
      "loss": 1.2903,
      "step": 3180
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.625128875585861,
      "learning_rate": 1.9998583919954095e-06,
      "loss": 1.3866,
      "step": 3181
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5097657585652895,
      "learning_rate": 1.9998583026690216e-06,
      "loss": 1.3992,
      "step": 3182
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.614507760819113,
      "learning_rate": 1.9998582133144708e-06,
      "loss": 1.4418,
      "step": 3183
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5439263422269685,
      "learning_rate": 1.9998581239317573e-06,
      "loss": 1.4107,
      "step": 3184
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.919848478106863,
      "learning_rate": 1.9998580345208814e-06,
      "loss": 1.44,
      "step": 3185
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.270625301228678,
      "learning_rate": 1.9998579450818424e-06,
      "loss": 1.3487,
      "step": 3186
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.770245132363083,
      "learning_rate": 1.999857855614641e-06,
      "loss": 1.2209,
      "step": 3187
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.851859510913612,
      "learning_rate": 1.9998577661192766e-06,
      "loss": 1.5614,
      "step": 3188
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.675470634925794,
      "learning_rate": 1.99985767659575e-06,
      "loss": 1.3947,
      "step": 3189
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.468219419933821,
      "learning_rate": 1.9998575870440602e-06,
      "loss": 1.3805,
      "step": 3190
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.966019364108259,
      "learning_rate": 1.9998574974642083e-06,
      "loss": 1.5527,
      "step": 3191
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.403779222001704,
      "learning_rate": 1.9998574078561935e-06,
      "loss": 1.4134,
      "step": 3192
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.913592743917321,
      "learning_rate": 1.999857318220016e-06,
      "loss": 1.3894,
      "step": 3193
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.977596675633243,
      "learning_rate": 1.999857228555676e-06,
      "loss": 1.5042,
      "step": 3194
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.206567891787514,
      "learning_rate": 1.999857138863173e-06,
      "loss": 1.4468,
      "step": 3195
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.04153279402594,
      "learning_rate": 1.999857049142508e-06,
      "loss": 1.3358,
      "step": 3196
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.061753008719967,
      "learning_rate": 1.9998569593936797e-06,
      "loss": 1.2138,
      "step": 3197
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5006757680467615,
      "learning_rate": 1.999856869616689e-06,
      "loss": 1.4535,
      "step": 3198
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.783752399409675,
      "learning_rate": 1.9998567798115356e-06,
      "loss": 1.371,
      "step": 3199
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.316562075427887,
      "learning_rate": 1.9998566899782197e-06,
      "loss": 1.242,
      "step": 3200
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.77063522200799,
      "learning_rate": 1.999856600116741e-06,
      "loss": 1.3504,
      "step": 3201
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.811809642589763,
      "learning_rate": 1.9998565102271e-06,
      "loss": 1.244,
      "step": 3202
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.246394909798491,
      "learning_rate": 1.9998564203092958e-06,
      "loss": 1.2308,
      "step": 3203
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.515471870625349,
      "learning_rate": 1.999856330363329e-06,
      "loss": 1.4938,
      "step": 3204
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.4680360822544944,
      "learning_rate": 1.9998562403892e-06,
      "loss": 1.456,
      "step": 3205
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.505670670091151,
      "learning_rate": 1.999856150386908e-06,
      "loss": 1.3296,
      "step": 3206
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.709421641146748,
      "learning_rate": 1.999856060356454e-06,
      "loss": 1.5864,
      "step": 3207
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.147973376306806,
      "learning_rate": 1.999855970297837e-06,
      "loss": 1.5874,
      "step": 3208
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.382675576687693,
      "learning_rate": 1.999855880211057e-06,
      "loss": 1.3383,
      "step": 3209
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.6100363440541745,
      "learning_rate": 1.999855790096115e-06,
      "loss": 1.354,
      "step": 3210
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.6842429101328005,
      "learning_rate": 1.99985569995301e-06,
      "loss": 1.2987,
      "step": 3211
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.40109444884483,
      "learning_rate": 1.9998556097817423e-06,
      "loss": 1.2192,
      "step": 3212
    },
    {
      "epoch": 0.02,
      "eval_loss": 1.6133770942687988,
      "eval_runtime": 4.6291,
      "eval_samples_per_second": 1.944,
      "eval_steps_per_second": 1.08,
      "step": 3212
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.686660224921117,
      "learning_rate": 1.9998555195823125e-06,
      "loss": 1.3872,
      "step": 3213
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.0508756090986076,
      "learning_rate": 1.9998554293547197e-06,
      "loss": 1.3356,
      "step": 3214
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.52397968155206,
      "learning_rate": 1.9998553390989644e-06,
      "loss": 1.4718,
      "step": 3215
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.345795773897772,
      "learning_rate": 1.9998552488150464e-06,
      "loss": 1.3888,
      "step": 3216
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.4749772284400064,
      "learning_rate": 1.9998551585029663e-06,
      "loss": 1.4003,
      "step": 3217
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.894626866747002,
      "learning_rate": 1.9998550681627233e-06,
      "loss": 1.3879,
      "step": 3218
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.358238813293754,
      "learning_rate": 1.9998549777943176e-06,
      "loss": 1.3516,
      "step": 3219
    },
    {
      "epoch": 0.02,
      "grad_norm": 10.113707594899338,
      "learning_rate": 1.999854887397749e-06,
      "loss": 1.4023,
      "step": 3220
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.254325564929292,
      "learning_rate": 1.999854796973018e-06,
      "loss": 1.39,
      "step": 3221
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.960705725899118,
      "learning_rate": 1.9998547065201247e-06,
      "loss": 1.5043,
      "step": 3222
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.358280513258368,
      "learning_rate": 1.9998546160390688e-06,
      "loss": 1.334,
      "step": 3223
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.374786742690436,
      "learning_rate": 1.9998545255298502e-06,
      "loss": 1.4208,
      "step": 3224
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.11312037967647,
      "learning_rate": 1.999854434992469e-06,
      "loss": 1.3896,
      "step": 3225
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.056956101904353,
      "learning_rate": 1.9998543444269254e-06,
      "loss": 1.6125,
      "step": 3226
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.4444491440324345,
      "learning_rate": 1.999854253833219e-06,
      "loss": 1.4266,
      "step": 3227
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.562601418248393,
      "learning_rate": 1.99985416321135e-06,
      "loss": 1.3555,
      "step": 3228
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.494814980020431,
      "learning_rate": 1.9998540725613185e-06,
      "loss": 1.2719,
      "step": 3229
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.463077255851529,
      "learning_rate": 1.9998539818831245e-06,
      "loss": 1.4654,
      "step": 3230
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.6659227436502615,
      "learning_rate": 1.999853891176768e-06,
      "loss": 1.4081,
      "step": 3231
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.27576441173752,
      "learning_rate": 1.999853800442249e-06,
      "loss": 1.3004,
      "step": 3232
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.132500624849133,
      "learning_rate": 1.9998537096795676e-06,
      "loss": 1.4915,
      "step": 3233
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.3773789294030445,
      "learning_rate": 1.9998536188887233e-06,
      "loss": 1.2791,
      "step": 3234
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.608523278365858,
      "learning_rate": 1.999853528069716e-06,
      "loss": 1.5425,
      "step": 3235
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.707153438089712,
      "learning_rate": 1.999853437222547e-06,
      "loss": 1.46,
      "step": 3236
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.354798489424228,
      "learning_rate": 1.999853346347215e-06,
      "loss": 1.3055,
      "step": 3237
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.717858890133739,
      "learning_rate": 1.999853255443721e-06,
      "loss": 1.4242,
      "step": 3238
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.881285175892339,
      "learning_rate": 1.9998531645120637e-06,
      "loss": 1.3225,
      "step": 3239
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.351033621024971,
      "learning_rate": 1.999853073552244e-06,
      "loss": 1.4343,
      "step": 3240
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.092636219063135,
      "learning_rate": 1.999852982564262e-06,
      "loss": 1.4519,
      "step": 3241
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.523318551270846,
      "learning_rate": 1.9998528915481176e-06,
      "loss": 1.3111,
      "step": 3242
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.12499436718178,
      "learning_rate": 1.9998528005038105e-06,
      "loss": 1.4857,
      "step": 3243
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.379826186575379,
      "learning_rate": 1.999852709431341e-06,
      "loss": 1.5748,
      "step": 3244
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.845476579476725,
      "learning_rate": 1.9998526183307083e-06,
      "loss": 1.5415,
      "step": 3245
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5232667716691735,
      "learning_rate": 1.999852527201914e-06,
      "loss": 1.4959,
      "step": 3246
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.0382078094923335,
      "learning_rate": 1.9998524360449566e-06,
      "loss": 1.3551,
      "step": 3247
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.089730656857905,
      "learning_rate": 1.999852344859837e-06,
      "loss": 1.4391,
      "step": 3248
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.233017204929559,
      "learning_rate": 1.9998522536465547e-06,
      "loss": 1.3869,
      "step": 3249
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.927879269649831,
      "learning_rate": 1.99985216240511e-06,
      "loss": 1.3307,
      "step": 3250
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.8097509464845585,
      "learning_rate": 1.999852071135503e-06,
      "loss": 1.5221,
      "step": 3251
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.199061944873702,
      "learning_rate": 1.999851979837733e-06,
      "loss": 1.0908,
      "step": 3252
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.531291994897648,
      "learning_rate": 1.9998518885118007e-06,
      "loss": 1.3583,
      "step": 3253
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.903810401327965,
      "learning_rate": 1.999851797157706e-06,
      "loss": 1.4129,
      "step": 3254
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.132170562144993,
      "learning_rate": 1.9998517057754487e-06,
      "loss": 1.2676,
      "step": 3255
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.252451485945251,
      "learning_rate": 1.999851614365029e-06,
      "loss": 1.3305,
      "step": 3256
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.063511090876979,
      "learning_rate": 1.9998515229264468e-06,
      "loss": 1.5813,
      "step": 3257
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.235824119971203,
      "learning_rate": 1.999851431459702e-06,
      "loss": 1.186,
      "step": 3258
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.324167091287429,
      "learning_rate": 1.9998513399647946e-06,
      "loss": 1.3319,
      "step": 3259
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.418031249539854,
      "learning_rate": 1.999851248441725e-06,
      "loss": 1.3418,
      "step": 3260
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.669234391667805,
      "learning_rate": 1.9998511568904925e-06,
      "loss": 1.4365,
      "step": 3261
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.883978731284588,
      "learning_rate": 1.999851065311098e-06,
      "loss": 1.2247,
      "step": 3262
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.3548479130386,
      "learning_rate": 1.9998509737035405e-06,
      "loss": 1.3808,
      "step": 3263
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5380211493875,
      "learning_rate": 1.999850882067821e-06,
      "loss": 1.4246,
      "step": 3264
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.00105285779209,
      "learning_rate": 1.999850790403939e-06,
      "loss": 1.1821,
      "step": 3265
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.375788203793615,
      "learning_rate": 1.9998506987118946e-06,
      "loss": 1.2683,
      "step": 3266
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.435134668590281,
      "learning_rate": 1.9998506069916874e-06,
      "loss": 1.3995,
      "step": 3267
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.011365587056869,
      "learning_rate": 1.9998505152433177e-06,
      "loss": 1.4123,
      "step": 3268
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.3076628146390945,
      "learning_rate": 1.999850423466786e-06,
      "loss": 1.5271,
      "step": 3269
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.518196335665455,
      "learning_rate": 1.9998503316620914e-06,
      "loss": 1.4203,
      "step": 3270
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.862304791483059,
      "learning_rate": 1.9998502398292344e-06,
      "loss": 1.3987,
      "step": 3271
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.78170745299268,
      "learning_rate": 1.999850147968215e-06,
      "loss": 1.3068,
      "step": 3272
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.178380281277487,
      "learning_rate": 1.999850056079033e-06,
      "loss": 1.2724,
      "step": 3273
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.433887746386324,
      "learning_rate": 1.999849964161689e-06,
      "loss": 1.1814,
      "step": 3274
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.528452236866192,
      "learning_rate": 1.9998498722161823e-06,
      "loss": 1.3414,
      "step": 3275
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.177232219836514,
      "learning_rate": 1.999849780242513e-06,
      "loss": 1.4879,
      "step": 3276
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.066602473545536,
      "learning_rate": 1.9998496882406816e-06,
      "loss": 1.5043,
      "step": 3277
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.589177379962832,
      "learning_rate": 1.9998495962106874e-06,
      "loss": 1.4113,
      "step": 3278
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.298228051116629,
      "learning_rate": 1.999849504152531e-06,
      "loss": 1.5326,
      "step": 3279
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.762978250317543,
      "learning_rate": 1.999849412066212e-06,
      "loss": 1.3229,
      "step": 3280
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.719035320919961,
      "learning_rate": 1.9998493199517306e-06,
      "loss": 1.495,
      "step": 3281
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.522910759102749,
      "learning_rate": 1.999849227809087e-06,
      "loss": 1.4205,
      "step": 3282
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.837186220843632,
      "learning_rate": 1.9998491356382808e-06,
      "loss": 1.4704,
      "step": 3283
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.817799074894965,
      "learning_rate": 1.9998490434393124e-06,
      "loss": 1.3439,
      "step": 3284
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.763809137445679,
      "learning_rate": 1.999848951212181e-06,
      "loss": 1.3933,
      "step": 3285
    },
    {
      "epoch": 0.02,
      "eval_loss": 1.6162359714508057,
      "eval_runtime": 4.6209,
      "eval_samples_per_second": 1.948,
      "eval_steps_per_second": 1.082,
      "step": 3285
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.2310982403645685,
      "learning_rate": 1.9998488589568875e-06,
      "loss": 1.3159,
      "step": 3286
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.751228665002726,
      "learning_rate": 1.999848766673432e-06,
      "loss": 1.508,
      "step": 3287
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.287367204511815,
      "learning_rate": 1.9998486743618136e-06,
      "loss": 1.2603,
      "step": 3288
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.191777133783223,
      "learning_rate": 1.9998485820220328e-06,
      "loss": 1.4435,
      "step": 3289
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.963362551339361,
      "learning_rate": 1.9998484896540898e-06,
      "loss": 1.6377,
      "step": 3290
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.380914773426925,
      "learning_rate": 1.9998483972579842e-06,
      "loss": 1.3618,
      "step": 3291
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.881974168253751,
      "learning_rate": 1.999848304833716e-06,
      "loss": 1.4793,
      "step": 3292
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.35338471574215,
      "learning_rate": 1.999848212381286e-06,
      "loss": 1.3686,
      "step": 3293
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.647483695518025,
      "learning_rate": 1.999848119900693e-06,
      "loss": 1.3954,
      "step": 3294
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.374460008441114,
      "learning_rate": 1.999848027391938e-06,
      "loss": 1.3345,
      "step": 3295
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.811138428730455,
      "learning_rate": 1.9998479348550204e-06,
      "loss": 1.5838,
      "step": 3296
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.982938535210265,
      "learning_rate": 1.9998478422899407e-06,
      "loss": 1.3415,
      "step": 3297
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.428729068613124,
      "learning_rate": 1.9998477496966984e-06,
      "loss": 1.5737,
      "step": 3298
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.419973685315894,
      "learning_rate": 1.9998476570752935e-06,
      "loss": 1.4732,
      "step": 3299
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.763708028938937,
      "learning_rate": 1.9998475644257264e-06,
      "loss": 1.3344,
      "step": 3300
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.49987874526447,
      "learning_rate": 1.999847471747997e-06,
      "loss": 1.3842,
      "step": 3301
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.059868654394915,
      "learning_rate": 1.999847379042105e-06,
      "loss": 1.5563,
      "step": 3302
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.101031294475946,
      "learning_rate": 1.9998472863080507e-06,
      "loss": 1.3156,
      "step": 3303
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.04036355932879,
      "learning_rate": 1.9998471935458343e-06,
      "loss": 1.4731,
      "step": 3304
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.42245956153174,
      "learning_rate": 1.999847100755455e-06,
      "loss": 1.3902,
      "step": 3305
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.774733280681285,
      "learning_rate": 1.999847007936914e-06,
      "loss": 1.4806,
      "step": 3306
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.071391799173309,
      "learning_rate": 1.99984691509021e-06,
      "loss": 1.3596,
      "step": 3307
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.2995003638028235,
      "learning_rate": 1.9998468222153443e-06,
      "loss": 1.4474,
      "step": 3308
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.3213954454836445,
      "learning_rate": 1.999846729312316e-06,
      "loss": 1.4512,
      "step": 3309
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.334198218103061,
      "learning_rate": 1.999846636381125e-06,
      "loss": 1.5658,
      "step": 3310
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.974385468276713,
      "learning_rate": 1.9998465434217723e-06,
      "loss": 1.4068,
      "step": 3311
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.891182503439006,
      "learning_rate": 1.999846450434257e-06,
      "loss": 1.321,
      "step": 3312
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.359459633381728,
      "learning_rate": 1.999846357418579e-06,
      "loss": 1.428,
      "step": 3313
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.035895559789005,
      "learning_rate": 1.999846264374739e-06,
      "loss": 1.2147,
      "step": 3314
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.397355468592714,
      "learning_rate": 1.999846171302736e-06,
      "loss": 1.3722,
      "step": 3315
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.155669309774268,
      "learning_rate": 1.9998460782025713e-06,
      "loss": 1.3978,
      "step": 3316
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.757021430848576,
      "learning_rate": 1.9998459850742443e-06,
      "loss": 1.5465,
      "step": 3317
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.786149548595665,
      "learning_rate": 1.9998458919177547e-06,
      "loss": 1.41,
      "step": 3318
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.289680760432469,
      "learning_rate": 1.9998457987331026e-06,
      "loss": 1.3994,
      "step": 3319
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.633968114607555,
      "learning_rate": 1.9998457055202887e-06,
      "loss": 1.5452,
      "step": 3320
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.513494336305803,
      "learning_rate": 1.9998456122793123e-06,
      "loss": 1.3446,
      "step": 3321
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.396311638031696,
      "learning_rate": 1.9998455190101732e-06,
      "loss": 1.4307,
      "step": 3322
    },
    {
      "epoch": 0.02,
      "grad_norm": 12.224345682379019,
      "learning_rate": 1.999845425712872e-06,
      "loss": 1.3998,
      "step": 3323
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.037124480361044,
      "learning_rate": 1.9998453323874087e-06,
      "loss": 1.3464,
      "step": 3324
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.619878980217735,
      "learning_rate": 1.999845239033783e-06,
      "loss": 1.3604,
      "step": 3325
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.480008801845805,
      "learning_rate": 1.999845145651995e-06,
      "loss": 1.3031,
      "step": 3326
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.454286328620764,
      "learning_rate": 1.9998450522420446e-06,
      "loss": 1.3064,
      "step": 3327
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.57634704126347,
      "learning_rate": 1.999844958803932e-06,
      "loss": 1.4419,
      "step": 3328
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.525977006177297,
      "learning_rate": 1.9998448653376565e-06,
      "loss": 1.4284,
      "step": 3329
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.871951894152188,
      "learning_rate": 1.9998447718432194e-06,
      "loss": 1.5416,
      "step": 3330
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.355722208034374,
      "learning_rate": 1.9998446783206197e-06,
      "loss": 1.3012,
      "step": 3331
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.768559786700981,
      "learning_rate": 1.999844584769858e-06,
      "loss": 1.3515,
      "step": 3332
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.700472874326849,
      "learning_rate": 1.9998444911909335e-06,
      "loss": 1.406,
      "step": 3333
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.6108709253998414,
      "learning_rate": 1.999844397583847e-06,
      "loss": 1.2965,
      "step": 3334
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.119228316339279,
      "learning_rate": 1.9998443039485983e-06,
      "loss": 1.2358,
      "step": 3335
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.6431919347398525,
      "learning_rate": 1.999844210285187e-06,
      "loss": 1.3773,
      "step": 3336
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.036505345315098,
      "learning_rate": 1.9998441165936137e-06,
      "loss": 1.426,
      "step": 3337
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.916331951067264,
      "learning_rate": 1.999844022873878e-06,
      "loss": 1.4319,
      "step": 3338
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.502760853285271,
      "learning_rate": 1.99984392912598e-06,
      "loss": 1.3404,
      "step": 3339
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.674212137830531,
      "learning_rate": 1.9998438353499197e-06,
      "loss": 1.3524,
      "step": 3340
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.969002536880517,
      "learning_rate": 1.9998437415456972e-06,
      "loss": 1.4791,
      "step": 3341
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.336062177765582,
      "learning_rate": 1.9998436477133126e-06,
      "loss": 1.5062,
      "step": 3342
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.803443625291179,
      "learning_rate": 1.9998435538527655e-06,
      "loss": 1.5203,
      "step": 3343
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.519808932086315,
      "learning_rate": 1.999843459964056e-06,
      "loss": 1.3434,
      "step": 3344
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.916582018340864,
      "learning_rate": 1.9998433660471847e-06,
      "loss": 1.2779,
      "step": 3345
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.597694435049481,
      "learning_rate": 1.9998432721021506e-06,
      "loss": 1.4044,
      "step": 3346
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.780005071775639,
      "learning_rate": 1.9998431781289544e-06,
      "loss": 1.1947,
      "step": 3347
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.419392700972962,
      "learning_rate": 1.999843084127596e-06,
      "loss": 1.3743,
      "step": 3348
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.791430996135701,
      "learning_rate": 1.9998429900980756e-06,
      "loss": 1.2246,
      "step": 3349
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.482748079898622,
      "learning_rate": 1.9998428960403925e-06,
      "loss": 1.2955,
      "step": 3350
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.958560367082578,
      "learning_rate": 1.9998428019545477e-06,
      "loss": 1.3907,
      "step": 3351
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.686365686539982,
      "learning_rate": 1.9998427078405404e-06,
      "loss": 1.4705,
      "step": 3352
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.865648897158208,
      "learning_rate": 1.9998426136983704e-06,
      "loss": 1.4142,
      "step": 3353
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.720528487925224,
      "learning_rate": 1.9998425195280387e-06,
      "loss": 1.3621,
      "step": 3354
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.23799082246452,
      "learning_rate": 1.999842425329545e-06,
      "loss": 1.4965,
      "step": 3355
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.338394874264906,
      "learning_rate": 1.999842331102888e-06,
      "loss": 1.4025,
      "step": 3356
    },
    {
      "epoch": 0.02,
      "grad_norm": 22.872377507742062,
      "learning_rate": 1.99984223684807e-06,
      "loss": 1.2324,
      "step": 3357
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.052623717517635,
      "learning_rate": 1.9998421425650892e-06,
      "loss": 1.1698,
      "step": 3358
    },
    {
      "epoch": 0.02,
      "eval_loss": 1.615498661994934,
      "eval_runtime": 4.6186,
      "eval_samples_per_second": 1.949,
      "eval_steps_per_second": 1.083,
      "step": 3358
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.403452284708548,
      "learning_rate": 1.999842048253946e-06,
      "loss": 1.4041,
      "step": 3359
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.612691456341564,
      "learning_rate": 1.999841953914641e-06,
      "loss": 1.4215,
      "step": 3360
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5286749608866055,
      "learning_rate": 1.9998418595471733e-06,
      "loss": 1.2596,
      "step": 3361
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.675637414315521,
      "learning_rate": 1.9998417651515436e-06,
      "loss": 1.4302,
      "step": 3362
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.99356723093498,
      "learning_rate": 1.9998416707277517e-06,
      "loss": 1.5043,
      "step": 3363
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.346376765762083,
      "learning_rate": 1.9998415762757977e-06,
      "loss": 1.4411,
      "step": 3364
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.465445688844873,
      "learning_rate": 1.9998414817956815e-06,
      "loss": 1.3665,
      "step": 3365
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.947373911092748,
      "learning_rate": 1.9998413872874027e-06,
      "loss": 1.4684,
      "step": 3366
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.336576397977949,
      "learning_rate": 1.9998412927509622e-06,
      "loss": 1.4559,
      "step": 3367
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.554379287052272,
      "learning_rate": 1.999841198186359e-06,
      "loss": 1.3741,
      "step": 3368
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.691693057261446,
      "learning_rate": 1.999841103593594e-06,
      "loss": 1.3809,
      "step": 3369
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.841766853561151,
      "learning_rate": 1.9998410089726666e-06,
      "loss": 1.4358,
      "step": 3370
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.445225429854286,
      "learning_rate": 1.999840914323577e-06,
      "loss": 1.2202,
      "step": 3371
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.587434031150207,
      "learning_rate": 1.9998408196463254e-06,
      "loss": 1.3254,
      "step": 3372
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.635518865319935,
      "learning_rate": 1.9998407249409115e-06,
      "loss": 1.4297,
      "step": 3373
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.693662361831059,
      "learning_rate": 1.9998406302073356e-06,
      "loss": 1.5567,
      "step": 3374
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.86578821971193,
      "learning_rate": 1.999840535445597e-06,
      "loss": 1.5298,
      "step": 3375
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.196254708579967,
      "learning_rate": 1.9998404406556967e-06,
      "loss": 1.4366,
      "step": 3376
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.704264956948557,
      "learning_rate": 1.999840345837634e-06,
      "loss": 1.2585,
      "step": 3377
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.123480430806445,
      "learning_rate": 1.9998402509914093e-06,
      "loss": 1.4561,
      "step": 3378
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.651652795610573,
      "learning_rate": 1.999840156117022e-06,
      "loss": 1.422,
      "step": 3379
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.895531346419424,
      "learning_rate": 1.999840061214473e-06,
      "loss": 1.39,
      "step": 3380
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.431768446645135,
      "learning_rate": 1.9998399662837614e-06,
      "loss": 1.2118,
      "step": 3381
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.73997458177594,
      "learning_rate": 1.999839871324888e-06,
      "loss": 1.3665,
      "step": 3382
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.995671187040337,
      "learning_rate": 1.9998397763378524e-06,
      "loss": 1.3142,
      "step": 3383
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.272798024995659,
      "learning_rate": 1.9998396813226545e-06,
      "loss": 1.3689,
      "step": 3384
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.4670178221368655,
      "learning_rate": 1.9998395862792944e-06,
      "loss": 1.4755,
      "step": 3385
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.832806175109676,
      "learning_rate": 1.999839491207772e-06,
      "loss": 1.4044,
      "step": 3386
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.27660795435882,
      "learning_rate": 1.9998393961080883e-06,
      "loss": 1.3548,
      "step": 3387
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.819442397568239,
      "learning_rate": 1.9998393009802417e-06,
      "loss": 1.4295,
      "step": 3388
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.911584590161657,
      "learning_rate": 1.999839205824233e-06,
      "loss": 1.5393,
      "step": 3389
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.255900800064722,
      "learning_rate": 1.9998391106400622e-06,
      "loss": 1.4227,
      "step": 3390
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.46927240762758,
      "learning_rate": 1.9998390154277293e-06,
      "loss": 1.3794,
      "step": 3391
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.445886168494522,
      "learning_rate": 1.999838920187234e-06,
      "loss": 1.3245,
      "step": 3392
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.605358505461649,
      "learning_rate": 1.9998388249185773e-06,
      "loss": 1.4412,
      "step": 3393
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.535610642306297,
      "learning_rate": 1.999838729621758e-06,
      "loss": 1.5244,
      "step": 3394
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.61293667721256,
      "learning_rate": 1.9998386342967767e-06,
      "loss": 1.3762,
      "step": 3395
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.576615477919866,
      "learning_rate": 1.999838538943633e-06,
      "loss": 1.1744,
      "step": 3396
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.982158065266377,
      "learning_rate": 1.999838443562327e-06,
      "loss": 1.4918,
      "step": 3397
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.380167900793968,
      "learning_rate": 1.9998383481528595e-06,
      "loss": 1.4074,
      "step": 3398
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.950258411072331,
      "learning_rate": 1.9998382527152297e-06,
      "loss": 1.4923,
      "step": 3399
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.592714907688114,
      "learning_rate": 1.9998381572494374e-06,
      "loss": 1.3628,
      "step": 3400
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.424967211816432,
      "learning_rate": 1.9998380617554833e-06,
      "loss": 1.4329,
      "step": 3401
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.525704926126062,
      "learning_rate": 1.999837966233367e-06,
      "loss": 1.412,
      "step": 3402
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.7381010367112735,
      "learning_rate": 1.9998378706830887e-06,
      "loss": 1.4563,
      "step": 3403
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.357820102817708,
      "learning_rate": 1.999837775104648e-06,
      "loss": 1.392,
      "step": 3404
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.877459475025764,
      "learning_rate": 1.9998376794980455e-06,
      "loss": 1.446,
      "step": 3405
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.736389873828967,
      "learning_rate": 1.999837583863281e-06,
      "loss": 1.3899,
      "step": 3406
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.6136258426518655,
      "learning_rate": 1.999837488200354e-06,
      "loss": 1.4183,
      "step": 3407
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.573629869371029,
      "learning_rate": 1.9998373925092654e-06,
      "loss": 1.4148,
      "step": 3408
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.331117554718104,
      "learning_rate": 1.999837296790014e-06,
      "loss": 1.4761,
      "step": 3409
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.167559123390286,
      "learning_rate": 1.999837201042601e-06,
      "loss": 1.374,
      "step": 3410
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.348496307914284,
      "learning_rate": 1.999837105267026e-06,
      "loss": 1.3786,
      "step": 3411
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.130052691622517,
      "learning_rate": 1.9998370094632887e-06,
      "loss": 1.4301,
      "step": 3412
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.497336313396663,
      "learning_rate": 1.9998369136313892e-06,
      "loss": 1.24,
      "step": 3413
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.813744620601688,
      "learning_rate": 1.999836817771328e-06,
      "loss": 1.4909,
      "step": 3414
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.591174887700467,
      "learning_rate": 1.9998367218831043e-06,
      "loss": 1.4375,
      "step": 3415
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.522216046172172,
      "learning_rate": 1.9998366259667188e-06,
      "loss": 1.3572,
      "step": 3416
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.717345593583763,
      "learning_rate": 1.999836530022171e-06,
      "loss": 1.4897,
      "step": 3417
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.321027044092839,
      "learning_rate": 1.9998364340494614e-06,
      "loss": 1.4912,
      "step": 3418
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.7605877409336195,
      "learning_rate": 1.99983633804859e-06,
      "loss": 1.4961,
      "step": 3419
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.4178839583225775,
      "learning_rate": 1.999836242019556e-06,
      "loss": 1.2966,
      "step": 3420
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.056171291505301,
      "learning_rate": 1.99983614596236e-06,
      "loss": 1.3807,
      "step": 3421
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.4340753688988075,
      "learning_rate": 1.9998360498770024e-06,
      "loss": 1.3203,
      "step": 3422
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.057972934945415,
      "learning_rate": 1.9998359537634822e-06,
      "loss": 1.4758,
      "step": 3423
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.829632088605847,
      "learning_rate": 1.9998358576218004e-06,
      "loss": 1.517,
      "step": 3424
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.85482687780751,
      "learning_rate": 1.999835761451956e-06,
      "loss": 1.4677,
      "step": 3425
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.922500655311902,
      "learning_rate": 1.9998356652539498e-06,
      "loss": 1.3043,
      "step": 3426
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.588250525378909,
      "learning_rate": 1.999835569027782e-06,
      "loss": 1.3587,
      "step": 3427
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.382966019888421,
      "learning_rate": 1.9998354727734514e-06,
      "loss": 1.255,
      "step": 3428
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.920889632741727,
      "learning_rate": 1.999835376490959e-06,
      "loss": 1.3713,
      "step": 3429
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.80574963169274,
      "learning_rate": 1.9998352801803045e-06,
      "loss": 1.6275,
      "step": 3430
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.028070166963233,
      "learning_rate": 1.999835183841488e-06,
      "loss": 1.4875,
      "step": 3431
    },
    {
      "epoch": 0.02,
      "eval_loss": 1.6133036613464355,
      "eval_runtime": 4.6357,
      "eval_samples_per_second": 1.941,
      "eval_steps_per_second": 1.079,
      "step": 3431
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.434318252171852,
      "learning_rate": 1.9998350874745098e-06,
      "loss": 1.363,
      "step": 3432
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.623393165669105,
      "learning_rate": 1.9998349910793694e-06,
      "loss": 1.5814,
      "step": 3433
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.456113027521206,
      "learning_rate": 1.999834894656067e-06,
      "loss": 1.2424,
      "step": 3434
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.519733967210898,
      "learning_rate": 1.9998347982046026e-06,
      "loss": 1.4818,
      "step": 3435
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.648419524976711,
      "learning_rate": 1.9998347017249762e-06,
      "loss": 1.2958,
      "step": 3436
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.479293176004785,
      "learning_rate": 1.9998346052171877e-06,
      "loss": 1.3048,
      "step": 3437
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.723612084510464,
      "learning_rate": 1.999834508681237e-06,
      "loss": 1.4803,
      "step": 3438
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.379602061766586,
      "learning_rate": 1.9998344121171245e-06,
      "loss": 1.3367,
      "step": 3439
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.763559064357727,
      "learning_rate": 1.99983431552485e-06,
      "loss": 1.5105,
      "step": 3440
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.923117534883834,
      "learning_rate": 1.9998342189044136e-06,
      "loss": 1.4361,
      "step": 3441
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.714745220308912,
      "learning_rate": 1.9998341222558148e-06,
      "loss": 1.5172,
      "step": 3442
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.225844294787685,
      "learning_rate": 1.999834025579054e-06,
      "loss": 1.4366,
      "step": 3443
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.3720897698041625,
      "learning_rate": 1.999833928874132e-06,
      "loss": 1.3114,
      "step": 3444
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.468815321569939,
      "learning_rate": 1.9998338321410473e-06,
      "loss": 1.3577,
      "step": 3445
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.369771276792973,
      "learning_rate": 1.9998337353798007e-06,
      "loss": 1.4633,
      "step": 3446
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.373295136247841,
      "learning_rate": 1.9998336385903923e-06,
      "loss": 1.3569,
      "step": 3447
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.775224501462429,
      "learning_rate": 1.9998335417728218e-06,
      "loss": 1.3241,
      "step": 3448
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.063351567591072,
      "learning_rate": 1.999833444927089e-06,
      "loss": 1.4984,
      "step": 3449
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.460324129913872,
      "learning_rate": 1.9998333480531947e-06,
      "loss": 1.4443,
      "step": 3450
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.49405386456319,
      "learning_rate": 1.999833251151138e-06,
      "loss": 1.3851,
      "step": 3451
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.561632118793946,
      "learning_rate": 1.99983315422092e-06,
      "loss": 1.3756,
      "step": 3452
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.787428115911711,
      "learning_rate": 1.9998330572625394e-06,
      "loss": 1.2516,
      "step": 3453
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.2654374946244955,
      "learning_rate": 1.999832960275997e-06,
      "loss": 1.3323,
      "step": 3454
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.943112051244451,
      "learning_rate": 1.9998328632612925e-06,
      "loss": 1.433,
      "step": 3455
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.64758931877515,
      "learning_rate": 1.999832766218426e-06,
      "loss": 1.3929,
      "step": 3456
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.004313093362644,
      "learning_rate": 1.999832669147398e-06,
      "loss": 1.3248,
      "step": 3457
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.974142973723598,
      "learning_rate": 1.9998325720482075e-06,
      "loss": 1.4423,
      "step": 3458
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.115923629555795,
      "learning_rate": 1.9998324749208554e-06,
      "loss": 1.3141,
      "step": 3459
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.29889351882071,
      "learning_rate": 1.999832377765341e-06,
      "loss": 1.5085,
      "step": 3460
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.933579769870536,
      "learning_rate": 1.999832280581665e-06,
      "loss": 1.3622,
      "step": 3461
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.54785822775587,
      "learning_rate": 1.9998321833698267e-06,
      "loss": 1.4151,
      "step": 3462
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.464340495701981,
      "learning_rate": 1.999832086129827e-06,
      "loss": 1.3599,
      "step": 3463
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.73588324623857,
      "learning_rate": 1.999831988861665e-06,
      "loss": 1.6014,
      "step": 3464
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.4591831342702966,
      "learning_rate": 1.9998318915653408e-06,
      "loss": 1.322,
      "step": 3465
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.723127658638915,
      "learning_rate": 1.9998317942408553e-06,
      "loss": 1.5451,
      "step": 3466
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.852467128721474,
      "learning_rate": 1.9998316968882073e-06,
      "loss": 1.301,
      "step": 3467
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.810154485536803,
      "learning_rate": 1.9998315995073976e-06,
      "loss": 1.421,
      "step": 3468
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.355440555741922,
      "learning_rate": 1.999831502098426e-06,
      "loss": 1.3608,
      "step": 3469
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.2600779975771434,
      "learning_rate": 1.9998314046612925e-06,
      "loss": 1.3282,
      "step": 3470
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.713840798819233,
      "learning_rate": 1.9998313071959967e-06,
      "loss": 1.469,
      "step": 3471
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.40703476488466,
      "learning_rate": 1.9998312097025396e-06,
      "loss": 1.195,
      "step": 3472
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.707652605308421,
      "learning_rate": 1.99983111218092e-06,
      "loss": 1.401,
      "step": 3473
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.9351410023359135,
      "learning_rate": 1.9998310146311386e-06,
      "loss": 1.5518,
      "step": 3474
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.835268364237103,
      "learning_rate": 1.999830917053196e-06,
      "loss": 1.4963,
      "step": 3475
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.461769170842469,
      "learning_rate": 1.9998308194470907e-06,
      "loss": 1.3913,
      "step": 3476
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.986732482048744,
      "learning_rate": 1.9998307218128237e-06,
      "loss": 1.4655,
      "step": 3477
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.103030642318946,
      "learning_rate": 1.999830624150395e-06,
      "loss": 1.4183,
      "step": 3478
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.347945323080289,
      "learning_rate": 1.999830526459804e-06,
      "loss": 1.5263,
      "step": 3479
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.836255099815674,
      "learning_rate": 1.999830428741051e-06,
      "loss": 1.3183,
      "step": 3480
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.36280200083679,
      "learning_rate": 1.9998303309941367e-06,
      "loss": 1.4045,
      "step": 3481
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.277029361572978,
      "learning_rate": 1.9998302332190603e-06,
      "loss": 1.2107,
      "step": 3482
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.367758928198617,
      "learning_rate": 1.9998301354158217e-06,
      "loss": 1.3907,
      "step": 3483
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.39073373168116,
      "learning_rate": 1.9998300375844213e-06,
      "loss": 1.2528,
      "step": 3484
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.626910873986676,
      "learning_rate": 1.9998299397248592e-06,
      "loss": 1.3984,
      "step": 3485
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.690894198671245,
      "learning_rate": 1.9998298418371354e-06,
      "loss": 1.463,
      "step": 3486
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.2521261877601555,
      "learning_rate": 1.9998297439212494e-06,
      "loss": 1.2599,
      "step": 3487
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.954516216479742,
      "learning_rate": 1.9998296459772013e-06,
      "loss": 1.3263,
      "step": 3488
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.64293902683026,
      "learning_rate": 1.999829548004992e-06,
      "loss": 1.4007,
      "step": 3489
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.7871583563034035,
      "learning_rate": 1.9998294500046204e-06,
      "loss": 1.4429,
      "step": 3490
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.615463851562626,
      "learning_rate": 1.999829351976087e-06,
      "loss": 1.5247,
      "step": 3491
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.704943808755736,
      "learning_rate": 1.9998292539193916e-06,
      "loss": 1.576,
      "step": 3492
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.7783079479655655,
      "learning_rate": 1.9998291558345344e-06,
      "loss": 1.4055,
      "step": 3493
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.441145271123304,
      "learning_rate": 1.9998290577215155e-06,
      "loss": 1.5482,
      "step": 3494
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.36043045218442,
      "learning_rate": 1.9998289595803345e-06,
      "loss": 1.3262,
      "step": 3495
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.745604478290225,
      "learning_rate": 1.9998288614109917e-06,
      "loss": 1.5021,
      "step": 3496
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.4255945297536385,
      "learning_rate": 1.999828763213487e-06,
      "loss": 1.4826,
      "step": 3497
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.268280892701323,
      "learning_rate": 1.999828664987821e-06,
      "loss": 1.3277,
      "step": 3498
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.925308919808803,
      "learning_rate": 1.9998285667339926e-06,
      "loss": 1.3895,
      "step": 3499
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.388627815177559,
      "learning_rate": 1.9998284684520024e-06,
      "loss": 1.1406,
      "step": 3500
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.7990704395075205,
      "learning_rate": 1.99982837014185e-06,
      "loss": 1.5664,
      "step": 3501
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.069150584999445,
      "learning_rate": 1.9998282718035366e-06,
      "loss": 1.5177,
      "step": 3502
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.792701471795762,
      "learning_rate": 1.999828173437061e-06,
      "loss": 1.4161,
      "step": 3503
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.535667453487885,
      "learning_rate": 1.999828075042424e-06,
      "loss": 1.3286,
      "step": 3504
    },
    {
      "epoch": 0.02,
      "eval_loss": 1.612628698348999,
      "eval_runtime": 4.6365,
      "eval_samples_per_second": 1.941,
      "eval_steps_per_second": 1.078,
      "step": 3504
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.836032937210056,
      "learning_rate": 1.9998279766196242e-06,
      "loss": 1.4098,
      "step": 3505
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.780070229048976,
      "learning_rate": 1.999827878168663e-06,
      "loss": 1.4433,
      "step": 3506
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.2390642232027655,
      "learning_rate": 1.9998277796895403e-06,
      "loss": 1.3572,
      "step": 3507
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.973359117911899,
      "learning_rate": 1.999827681182255e-06,
      "loss": 1.3459,
      "step": 3508
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.378947132862404,
      "learning_rate": 1.9998275826468085e-06,
      "loss": 1.4072,
      "step": 3509
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.58519122967338,
      "learning_rate": 1.9998274840832003e-06,
      "loss": 1.4547,
      "step": 3510
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.718146444105029,
      "learning_rate": 1.99982738549143e-06,
      "loss": 1.3933,
      "step": 3511
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.433816849751808,
      "learning_rate": 1.9998272868714978e-06,
      "loss": 1.5081,
      "step": 3512
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.668728068958253,
      "learning_rate": 1.999827188223404e-06,
      "loss": 1.4535,
      "step": 3513
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.205098278485999,
      "learning_rate": 1.9998270895471484e-06,
      "loss": 1.3971,
      "step": 3514
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.2812240461470905,
      "learning_rate": 1.999826990842731e-06,
      "loss": 1.2847,
      "step": 3515
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.421468940513477,
      "learning_rate": 1.9998268921101516e-06,
      "loss": 1.2344,
      "step": 3516
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.240697700729821,
      "learning_rate": 1.999826793349411e-06,
      "loss": 1.3846,
      "step": 3517
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.624113047494641,
      "learning_rate": 1.999826694560508e-06,
      "loss": 1.5729,
      "step": 3518
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.125813300538199,
      "learning_rate": 1.999826595743443e-06,
      "loss": 1.3067,
      "step": 3519
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.72485279895408,
      "learning_rate": 1.9998264968982165e-06,
      "loss": 1.4624,
      "step": 3520
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.440580236377528,
      "learning_rate": 1.9998263980248284e-06,
      "loss": 1.3921,
      "step": 3521
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.506465574269989,
      "learning_rate": 1.9998262991232786e-06,
      "loss": 1.3429,
      "step": 3522
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.892687291895052,
      "learning_rate": 1.9998262001935666e-06,
      "loss": 1.4849,
      "step": 3523
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.714503092123628,
      "learning_rate": 1.999826101235693e-06,
      "loss": 1.3495,
      "step": 3524
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.080297383574053,
      "learning_rate": 1.9998260022496574e-06,
      "loss": 1.4098,
      "step": 3525
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.672363930477558,
      "learning_rate": 1.9998259032354607e-06,
      "loss": 1.1808,
      "step": 3526
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.762254121900801,
      "learning_rate": 1.999825804193102e-06,
      "loss": 1.2784,
      "step": 3527
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.0090886444132465,
      "learning_rate": 1.999825705122581e-06,
      "loss": 1.3983,
      "step": 3528
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.541188950258432,
      "learning_rate": 1.9998256060238984e-06,
      "loss": 1.4821,
      "step": 3529
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.508252601462831,
      "learning_rate": 1.9998255068970544e-06,
      "loss": 1.4049,
      "step": 3530
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.384733739344799,
      "learning_rate": 1.9998254077420486e-06,
      "loss": 1.3762,
      "step": 3531
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.91582726748242,
      "learning_rate": 1.9998253085588806e-06,
      "loss": 1.4418,
      "step": 3532
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.228789208224146,
      "learning_rate": 1.9998252093475513e-06,
      "loss": 1.2963,
      "step": 3533
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.697321495816711,
      "learning_rate": 1.99982511010806e-06,
      "loss": 1.4528,
      "step": 3534
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.803901064568725,
      "learning_rate": 1.9998250108404072e-06,
      "loss": 1.569,
      "step": 3535
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.353829814405607,
      "learning_rate": 1.9998249115445924e-06,
      "loss": 1.3503,
      "step": 3536
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.469444434313929,
      "learning_rate": 1.9998248122206158e-06,
      "loss": 1.485,
      "step": 3537
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.135382837905591,
      "learning_rate": 1.999824712868478e-06,
      "loss": 1.3355,
      "step": 3538
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.198880866777614,
      "learning_rate": 1.999824613488178e-06,
      "loss": 1.469,
      "step": 3539
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.2331303310285415,
      "learning_rate": 1.999824514079716e-06,
      "loss": 1.3518,
      "step": 3540
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.8429386253121125,
      "learning_rate": 1.9998244146430926e-06,
      "loss": 1.4942,
      "step": 3541
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.574611939532659,
      "learning_rate": 1.9998243151783077e-06,
      "loss": 1.4268,
      "step": 3542
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.94932612125061,
      "learning_rate": 1.9998242156853608e-06,
      "loss": 1.6209,
      "step": 3543
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.006103951828245,
      "learning_rate": 1.999824116164252e-06,
      "loss": 1.2351,
      "step": 3544
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.166042167548124,
      "learning_rate": 1.999824016614982e-06,
      "loss": 1.142,
      "step": 3545
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.0620007916305925,
      "learning_rate": 1.99982391703755e-06,
      "loss": 1.4374,
      "step": 3546
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.971313439844006,
      "learning_rate": 1.999823817431956e-06,
      "loss": 1.4932,
      "step": 3547
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.948526977664145,
      "learning_rate": 1.999823717798201e-06,
      "loss": 1.3942,
      "step": 3548
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.433000571492981,
      "learning_rate": 1.9998236181362836e-06,
      "loss": 1.6458,
      "step": 3549
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.7172630835592475,
      "learning_rate": 1.9998235184462045e-06,
      "loss": 1.3787,
      "step": 3550
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.835524125983266,
      "learning_rate": 1.999823418727964e-06,
      "loss": 1.4862,
      "step": 3551
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.538989738684982,
      "learning_rate": 1.999823318981562e-06,
      "loss": 1.2833,
      "step": 3552
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.900851370893367,
      "learning_rate": 1.999823219206998e-06,
      "loss": 1.4289,
      "step": 3553
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.742076041128345,
      "learning_rate": 1.9998231194042723e-06,
      "loss": 1.5084,
      "step": 3554
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.130400288886862,
      "learning_rate": 1.999823019573385e-06,
      "loss": 1.251,
      "step": 3555
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.2268449306817,
      "learning_rate": 1.999822919714336e-06,
      "loss": 1.2141,
      "step": 3556
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5252607523373936,
      "learning_rate": 1.9998228198271253e-06,
      "loss": 1.3066,
      "step": 3557
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.351860568361683,
      "learning_rate": 1.999822719911753e-06,
      "loss": 1.3784,
      "step": 3558
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.3402345801839415,
      "learning_rate": 1.9998226199682187e-06,
      "loss": 1.375,
      "step": 3559
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.810672354782998,
      "learning_rate": 1.999822519996523e-06,
      "loss": 1.4548,
      "step": 3560
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.856779921376611,
      "learning_rate": 1.999822419996666e-06,
      "loss": 1.3662,
      "step": 3561
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.495518456829636,
      "learning_rate": 1.9998223199686466e-06,
      "loss": 1.3658,
      "step": 3562
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.972175534549607,
      "learning_rate": 1.999822219912466e-06,
      "loss": 1.4305,
      "step": 3563
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5993830494584484,
      "learning_rate": 1.9998221198281236e-06,
      "loss": 1.3633,
      "step": 3564
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.464312906911071,
      "learning_rate": 1.9998220197156194e-06,
      "loss": 1.3512,
      "step": 3565
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.666030982227595,
      "learning_rate": 1.9998219195749536e-06,
      "loss": 1.4271,
      "step": 3566
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.5017841565719365,
      "learning_rate": 1.9998218194061264e-06,
      "loss": 1.2379,
      "step": 3567
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.160263531672295,
      "learning_rate": 1.9998217192091375e-06,
      "loss": 1.4269,
      "step": 3568
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.394350494166971,
      "learning_rate": 1.9998216189839865e-06,
      "loss": 1.5847,
      "step": 3569
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.543610113197201,
      "learning_rate": 1.999821518730674e-06,
      "loss": 1.5249,
      "step": 3570
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.524606361318028,
      "learning_rate": 1.9998214184492e-06,
      "loss": 1.4545,
      "step": 3571
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.8729914641558425,
      "learning_rate": 1.9998213181395643e-06,
      "loss": 1.4856,
      "step": 3572
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.712818698671254,
      "learning_rate": 1.999821217801767e-06,
      "loss": 1.4254,
      "step": 3573
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.129380023262798,
      "learning_rate": 1.9998211174358083e-06,
      "loss": 1.5263,
      "step": 3574
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.632118655345206,
      "learning_rate": 1.999821017041688e-06,
      "loss": 1.2905,
      "step": 3575
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.713339231139657,
      "learning_rate": 1.9998209166194055e-06,
      "loss": 1.4079,
      "step": 3576
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.761666895797936,
      "learning_rate": 1.9998208161689615e-06,
      "loss": 1.348,
      "step": 3577
    },
    {
      "epoch": 0.02,
      "eval_loss": 1.6088365316390991,
      "eval_runtime": 4.6298,
      "eval_samples_per_second": 1.944,
      "eval_steps_per_second": 1.08,
      "step": 3577
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.053550759411962,
      "learning_rate": 1.999820715690356e-06,
      "loss": 1.4378,
      "step": 3578
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.656417963233415,
      "learning_rate": 1.999820615183589e-06,
      "loss": 1.4196,
      "step": 3579
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.720114969577187,
      "learning_rate": 1.9998205146486604e-06,
      "loss": 1.4177,
      "step": 3580
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.1590061020053435,
      "learning_rate": 1.99982041408557e-06,
      "loss": 1.3961,
      "step": 3581
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.609450714516466,
      "learning_rate": 1.999820313494318e-06,
      "loss": 1.3561,
      "step": 3582
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.864818427122029,
      "learning_rate": 1.9998202128749045e-06,
      "loss": 1.3521,
      "step": 3583
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.532234217131842,
      "learning_rate": 1.9998201122273292e-06,
      "loss": 1.2275,
      "step": 3584
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.083685514923316,
      "learning_rate": 1.9998200115515927e-06,
      "loss": 1.7183,
      "step": 3585
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.1648857831860004,
      "learning_rate": 1.9998199108476944e-06,
      "loss": 1.4292,
      "step": 3586
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.404570069932921,
      "learning_rate": 1.9998198101156344e-06,
      "loss": 1.4524,
      "step": 3587
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.4636298910229435,
      "learning_rate": 1.9998197093554126e-06,
      "loss": 1.3022,
      "step": 3588
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.698700968868405,
      "learning_rate": 1.9998196085670296e-06,
      "loss": 1.3204,
      "step": 3589
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.210913435197878,
      "learning_rate": 1.9998195077504848e-06,
      "loss": 1.3195,
      "step": 3590
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.072601657781737,
      "learning_rate": 1.9998194069057783e-06,
      "loss": 1.4633,
      "step": 3591
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.947711709512535,
      "learning_rate": 1.9998193060329105e-06,
      "loss": 1.3934,
      "step": 3592
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.489367907366415,
      "learning_rate": 1.999819205131881e-06,
      "loss": 1.3327,
      "step": 3593
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.046442232090646,
      "learning_rate": 1.99981910420269e-06,
      "loss": 1.519,
      "step": 3594
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.4848572461168015,
      "learning_rate": 1.999819003245337e-06,
      "loss": 1.4837,
      "step": 3595
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.189445025381679,
      "learning_rate": 1.999818902259823e-06,
      "loss": 1.322,
      "step": 3596
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.052080974977652,
      "learning_rate": 1.999818801246147e-06,
      "loss": 1.4127,
      "step": 3597
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.364751238319127,
      "learning_rate": 1.99981870020431e-06,
      "loss": 1.2812,
      "step": 3598
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.0230864070326104,
      "learning_rate": 1.9998185991343108e-06,
      "loss": 1.1783,
      "step": 3599
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.459589611896564,
      "learning_rate": 1.9998184980361504e-06,
      "loss": 1.2804,
      "step": 3600
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.424443515802799,
      "learning_rate": 1.9998183969098283e-06,
      "loss": 1.2931,
      "step": 3601
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.769365934369103,
      "learning_rate": 1.9998182957553445e-06,
      "loss": 1.3411,
      "step": 3602
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.944616559405836,
      "learning_rate": 1.9998181945726993e-06,
      "loss": 1.2028,
      "step": 3603
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.852238644907209,
      "learning_rate": 1.9998180933618925e-06,
      "loss": 1.3328,
      "step": 3604
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.822059635157921,
      "learning_rate": 1.9998179921229243e-06,
      "loss": 1.2309,
      "step": 3605
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.495892687998826,
      "learning_rate": 1.9998178908557944e-06,
      "loss": 1.4956,
      "step": 3606
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.355235617835983,
      "learning_rate": 1.999817789560503e-06,
      "loss": 1.181,
      "step": 3607
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.220700935420473,
      "learning_rate": 1.99981768823705e-06,
      "loss": 1.2949,
      "step": 3608
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.93767906248555,
      "learning_rate": 1.9998175868854353e-06,
      "loss": 1.4398,
      "step": 3609
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.976911864648587,
      "learning_rate": 1.9998174855056593e-06,
      "loss": 1.3841,
      "step": 3610
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.915768062533412,
      "learning_rate": 1.999817384097722e-06,
      "loss": 1.3776,
      "step": 3611
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.400452386799031,
      "learning_rate": 1.999817282661623e-06,
      "loss": 1.5242,
      "step": 3612
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.596731304469015,
      "learning_rate": 1.999817181197362e-06,
      "loss": 1.4432,
      "step": 3613
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.417819290463723,
      "learning_rate": 1.99981707970494e-06,
      "loss": 1.3012,
      "step": 3614
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.908570364549861,
      "learning_rate": 1.9998169781843566e-06,
      "loss": 1.4815,
      "step": 3615
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.538384996779538,
      "learning_rate": 1.999816876635611e-06,
      "loss": 1.4016,
      "step": 3616
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.527777934939574,
      "learning_rate": 1.9998167750587048e-06,
      "loss": 1.3748,
      "step": 3617
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.9438933831368645,
      "learning_rate": 1.9998166734536362e-06,
      "loss": 1.3516,
      "step": 3618
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.111197603829205,
      "learning_rate": 1.9998165718204064e-06,
      "loss": 1.4535,
      "step": 3619
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.3741990248604035,
      "learning_rate": 1.9998164701590153e-06,
      "loss": 1.3127,
      "step": 3620
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.416015259175785,
      "learning_rate": 1.999816368469463e-06,
      "loss": 1.2723,
      "step": 3621
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.045256087161643,
      "learning_rate": 1.9998162667517488e-06,
      "loss": 1.4302,
      "step": 3622
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.026141735704103,
      "learning_rate": 1.999816165005873e-06,
      "loss": 1.5235,
      "step": 3623
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.817827993597373,
      "learning_rate": 1.9998160632318357e-06,
      "loss": 1.3734,
      "step": 3624
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.13290587555347,
      "learning_rate": 1.999815961429637e-06,
      "loss": 1.6108,
      "step": 3625
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.988580619133651,
      "learning_rate": 1.9998158595992766e-06,
      "loss": 1.1573,
      "step": 3626
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.66265731554959,
      "learning_rate": 1.9998157577407546e-06,
      "loss": 1.1933,
      "step": 3627
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.648371974483903,
      "learning_rate": 1.999815655854072e-06,
      "loss": 1.4983,
      "step": 3628
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.62998311734592,
      "learning_rate": 1.9998155539392273e-06,
      "loss": 1.3605,
      "step": 3629
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.650232836970276,
      "learning_rate": 1.999815451996221e-06,
      "loss": 1.4764,
      "step": 3630
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.871641037949958,
      "learning_rate": 1.9998153500250534e-06,
      "loss": 1.2777,
      "step": 3631
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.617802934649518,
      "learning_rate": 1.999815248025724e-06,
      "loss": 1.4119,
      "step": 3632
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.411468513060949,
      "learning_rate": 1.999815145998234e-06,
      "loss": 1.2856,
      "step": 3633
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.4136554607658995,
      "learning_rate": 1.999815043942582e-06,
      "loss": 1.3036,
      "step": 3634
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.696213561907659,
      "learning_rate": 1.9998149418587684e-06,
      "loss": 1.4445,
      "step": 3635
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.707575975813092,
      "learning_rate": 1.9998148397467934e-06,
      "loss": 1.4956,
      "step": 3636
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.477661718215679,
      "learning_rate": 1.9998147376066567e-06,
      "loss": 1.4149,
      "step": 3637
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.242955704509717,
      "learning_rate": 1.9998146354383588e-06,
      "loss": 1.3746,
      "step": 3638
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.407436531750599,
      "learning_rate": 1.9998145332418995e-06,
      "loss": 1.3222,
      "step": 3639
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.4015116251774575,
      "learning_rate": 1.999814431017279e-06,
      "loss": 1.4696,
      "step": 3640
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.230060461672011,
      "learning_rate": 1.9998143287644966e-06,
      "loss": 1.1936,
      "step": 3641
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.832174281020922,
      "learning_rate": 1.999814226483553e-06,
      "loss": 1.5139,
      "step": 3642
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.458935294035869,
      "learning_rate": 1.999814124174448e-06,
      "loss": 1.366,
      "step": 3643
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.858967822250987,
      "learning_rate": 1.9998140218371814e-06,
      "loss": 1.4256,
      "step": 3644
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.384171006945276,
      "learning_rate": 1.9998139194717534e-06,
      "loss": 1.5396,
      "step": 3645
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.362217816203799,
      "learning_rate": 1.999813817078164e-06,
      "loss": 1.4846,
      "step": 3646
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.829886703693458,
      "learning_rate": 1.999813714656413e-06,
      "loss": 1.465,
      "step": 3647
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.300928121079802,
      "learning_rate": 1.999813612206501e-06,
      "loss": 1.3776,
      "step": 3648
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.186443596966576,
      "learning_rate": 1.9998135097284272e-06,
      "loss": 1.3409,
      "step": 3649
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.883455465101393,
      "learning_rate": 1.9998134072221923e-06,
      "loss": 1.41,
      "step": 3650
    },
    {
      "epoch": 0.02,
      "eval_loss": 1.6074836254119873,
      "eval_runtime": 4.6299,
      "eval_samples_per_second": 1.944,
      "eval_steps_per_second": 1.08,
      "step": 3650
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.160544967304233,
      "learning_rate": 1.9998133046877957e-06,
      "loss": 1.4479,
      "step": 3651
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.591999173432497,
      "learning_rate": 1.9998132021252378e-06,
      "loss": 1.424,
      "step": 3652
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.334725366745728,
      "learning_rate": 1.9998130995345185e-06,
      "loss": 1.3303,
      "step": 3653
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.667679588938856,
      "learning_rate": 1.9998129969156376e-06,
      "loss": 1.3497,
      "step": 3654
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.648593704664988,
      "learning_rate": 1.9998128942685953e-06,
      "loss": 1.5196,
      "step": 3655
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.228106530732631,
      "learning_rate": 1.999812791593392e-06,
      "loss": 1.5034,
      "step": 3656
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.825079154423852,
      "learning_rate": 1.9998126888900272e-06,
      "loss": 1.5114,
      "step": 3657
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.588839443240046,
      "learning_rate": 1.9998125861585006e-06,
      "loss": 1.224,
      "step": 3658
    },
    {
      "epoch": 0.02,
      "grad_norm": 11.115206705835469,
      "learning_rate": 1.9998124833988127e-06,
      "loss": 1.402,
      "step": 3659
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.028407264356555,
      "learning_rate": 1.9998123806109635e-06,
      "loss": 1.3366,
      "step": 3660
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.883233407550293,
      "learning_rate": 1.999812277794953e-06,
      "loss": 1.3908,
      "step": 3661
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.849560264366864,
      "learning_rate": 1.999812174950781e-06,
      "loss": 1.4642,
      "step": 3662
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.134509693028477,
      "learning_rate": 1.999812072078448e-06,
      "loss": 1.338,
      "step": 3663
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.64736551563639,
      "learning_rate": 1.999811969177953e-06,
      "loss": 1.3188,
      "step": 3664
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.202295612938597,
      "learning_rate": 1.999811866249297e-06,
      "loss": 1.5156,
      "step": 3665
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.864330255262217,
      "learning_rate": 1.9998117632924795e-06,
      "loss": 1.4044,
      "step": 3666
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.095975352302671,
      "learning_rate": 1.9998116603075008e-06,
      "loss": 1.4834,
      "step": 3667
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.401897149251593,
      "learning_rate": 1.9998115572943607e-06,
      "loss": 1.3598,
      "step": 3668
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.516321121878343,
      "learning_rate": 1.9998114542530593e-06,
      "loss": 1.3812,
      "step": 3669
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.582276547112316,
      "learning_rate": 1.999811351183596e-06,
      "loss": 1.4511,
      "step": 3670
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.268406465536788,
      "learning_rate": 1.9998112480859718e-06,
      "loss": 1.3777,
      "step": 3671
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.414725079724578,
      "learning_rate": 1.999811144960186e-06,
      "loss": 1.3594,
      "step": 3672
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.036395550915229,
      "learning_rate": 1.9998110418062394e-06,
      "loss": 1.5366,
      "step": 3673
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.811922967477642,
      "learning_rate": 1.9998109386241307e-06,
      "loss": 1.489,
      "step": 3674
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.909204659000984,
      "learning_rate": 1.999810835413861e-06,
      "loss": 1.4438,
      "step": 3675
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.933064342481729,
      "learning_rate": 1.99981073217543e-06,
      "loss": 1.3926,
      "step": 3676
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.172865172666342,
      "learning_rate": 1.999810628908838e-06,
      "loss": 1.3739,
      "step": 3677
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.581844754398152,
      "learning_rate": 1.999810525614084e-06,
      "loss": 1.3825,
      "step": 3678
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.002706815163878,
      "learning_rate": 1.999810422291169e-06,
      "loss": 1.2023,
      "step": 3679
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.529339089739914,
      "learning_rate": 1.9998103189400925e-06,
      "loss": 1.3956,
      "step": 3680
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.371778302872117,
      "learning_rate": 1.999810215560855e-06,
      "loss": 1.0882,
      "step": 3681
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.604473618690761,
      "learning_rate": 1.999810112153456e-06,
      "loss": 1.3797,
      "step": 3682
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.229913176526649,
      "learning_rate": 1.9998100087178954e-06,
      "loss": 1.3024,
      "step": 3683
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.892279631388166,
      "learning_rate": 1.9998099052541736e-06,
      "loss": 1.4261,
      "step": 3684
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.375988978498727,
      "learning_rate": 1.9998098017622905e-06,
      "loss": 1.426,
      "step": 3685
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.517183513297791,
      "learning_rate": 1.9998096982422465e-06,
      "loss": 1.4715,
      "step": 3686
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.715497920821096,
      "learning_rate": 1.9998095946940408e-06,
      "loss": 1.2492,
      "step": 3687
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.4409337214742814,
      "learning_rate": 1.9998094911176738e-06,
      "loss": 1.3908,
      "step": 3688
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.768774973994264,
      "learning_rate": 1.9998093875131454e-06,
      "loss": 1.3043,
      "step": 3689
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.488443359293641,
      "learning_rate": 1.999809283880456e-06,
      "loss": 1.4254,
      "step": 3690
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.946907576554227,
      "learning_rate": 1.999809180219605e-06,
      "loss": 1.3815,
      "step": 3691
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.024643467089974,
      "learning_rate": 1.9998090765305927e-06,
      "loss": 1.3476,
      "step": 3692
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.375622450304601,
      "learning_rate": 1.9998089728134195e-06,
      "loss": 1.575,
      "step": 3693
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.7741580232890755,
      "learning_rate": 1.9998088690680847e-06,
      "loss": 1.3576,
      "step": 3694
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.690371879763435,
      "learning_rate": 1.9998087652945886e-06,
      "loss": 1.3921,
      "step": 3695
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.895129039414259,
      "learning_rate": 1.999808661492931e-06,
      "loss": 1.5504,
      "step": 3696
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.841292889636168,
      "learning_rate": 1.9998085576631128e-06,
      "loss": 1.6215,
      "step": 3697
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.335373503405417,
      "learning_rate": 1.9998084538051327e-06,
      "loss": 1.4034,
      "step": 3698
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.892579988645074,
      "learning_rate": 1.999808349918992e-06,
      "loss": 1.4615,
      "step": 3699
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.80405901674261,
      "learning_rate": 1.999808246004689e-06,
      "loss": 1.3834,
      "step": 3700
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.015073387547917,
      "learning_rate": 1.9998081420622256e-06,
      "loss": 1.5382,
      "step": 3701
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.340605580327248,
      "learning_rate": 1.9998080380916003e-06,
      "loss": 1.5587,
      "step": 3702
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.6548885345585145,
      "learning_rate": 1.999807934092814e-06,
      "loss": 1.4507,
      "step": 3703
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.988246529171995,
      "learning_rate": 1.9998078300658667e-06,
      "loss": 1.2863,
      "step": 3704
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.366479726955373,
      "learning_rate": 1.999807726010758e-06,
      "loss": 1.3532,
      "step": 3705
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.954787351869969,
      "learning_rate": 1.999807621927488e-06,
      "loss": 1.4679,
      "step": 3706
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.106594076376741,
      "learning_rate": 1.9998075178160565e-06,
      "loss": 1.4912,
      "step": 3707
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.53644272240156,
      "learning_rate": 1.999807413676464e-06,
      "loss": 1.3928,
      "step": 3708
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.3898086275317265,
      "learning_rate": 1.9998073095087102e-06,
      "loss": 1.2247,
      "step": 3709
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.129858409048548,
      "learning_rate": 1.9998072053127954e-06,
      "loss": 1.2888,
      "step": 3710
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.411528618345709,
      "learning_rate": 1.999807101088719e-06,
      "loss": 1.4598,
      "step": 3711
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.245071088827053,
      "learning_rate": 1.9998069968364813e-06,
      "loss": 1.3139,
      "step": 3712
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.001994498250823,
      "learning_rate": 1.9998068925560825e-06,
      "loss": 1.4319,
      "step": 3713
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.5711686332255725,
      "learning_rate": 1.9998067882475225e-06,
      "loss": 1.4286,
      "step": 3714
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.494680754322679,
      "learning_rate": 1.999806683910801e-06,
      "loss": 1.3624,
      "step": 3715
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.920443475726351,
      "learning_rate": 1.999806579545919e-06,
      "loss": 1.4834,
      "step": 3716
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.09972634364239,
      "learning_rate": 1.9998064751528752e-06,
      "loss": 1.3838,
      "step": 3717
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.696129650401296,
      "learning_rate": 1.99980637073167e-06,
      "loss": 1.5004,
      "step": 3718
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.3037260313404415,
      "learning_rate": 1.999806266282304e-06,
      "loss": 1.5459,
      "step": 3719
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.88221476190435,
      "learning_rate": 1.9998061618047767e-06,
      "loss": 1.4734,
      "step": 3720
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.956709505619386,
      "learning_rate": 1.999806057299088e-06,
      "loss": 1.3683,
      "step": 3721
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.1834233457849495,
      "learning_rate": 1.9998059527652382e-06,
      "loss": 1.3125,
      "step": 3722
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.492850793982879,
      "learning_rate": 1.9998058482032273e-06,
      "loss": 1.2745,
      "step": 3723
    },
    {
      "epoch": 0.03,
      "eval_loss": 1.6084189414978027,
      "eval_runtime": 4.642,
      "eval_samples_per_second": 1.939,
      "eval_steps_per_second": 1.077,
      "step": 3723
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.713923973024249,
      "learning_rate": 1.999805743613055e-06,
      "loss": 1.4203,
      "step": 3724
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.441921382155817,
      "learning_rate": 1.999805638994722e-06,
      "loss": 1.4163,
      "step": 3725
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.164638367038176,
      "learning_rate": 1.9998055343482274e-06,
      "loss": 1.4703,
      "step": 3726
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.707717776966889,
      "learning_rate": 1.999805429673571e-06,
      "loss": 1.4389,
      "step": 3727
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.404527912836125,
      "learning_rate": 1.9998053249707545e-06,
      "loss": 1.4697,
      "step": 3728
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.582127757876631,
      "learning_rate": 1.999805220239776e-06,
      "loss": 1.4003,
      "step": 3729
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.82572183924319,
      "learning_rate": 1.999805115480637e-06,
      "loss": 1.3832,
      "step": 3730
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.099262606984633,
      "learning_rate": 1.9998050106933363e-06,
      "loss": 1.4593,
      "step": 3731
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.705269403296674,
      "learning_rate": 1.9998049058778745e-06,
      "loss": 1.4539,
      "step": 3732
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.343650845199192,
      "learning_rate": 1.9998048010342517e-06,
      "loss": 1.4299,
      "step": 3733
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.761090749890984,
      "learning_rate": 1.9998046961624677e-06,
      "loss": 1.376,
      "step": 3734
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.47848246597197,
      "learning_rate": 1.9998045912625223e-06,
      "loss": 1.3492,
      "step": 3735
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.553841135265006,
      "learning_rate": 1.999804486334416e-06,
      "loss": 1.4707,
      "step": 3736
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.112011524717292,
      "learning_rate": 1.999804381378148e-06,
      "loss": 1.3998,
      "step": 3737
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.053186099424353,
      "learning_rate": 1.9998042763937197e-06,
      "loss": 1.3719,
      "step": 3738
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.9280058328899665,
      "learning_rate": 1.9998041713811296e-06,
      "loss": 1.4414,
      "step": 3739
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.927412166137148,
      "learning_rate": 1.9998040663403785e-06,
      "loss": 1.3497,
      "step": 3740
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.868129380898875,
      "learning_rate": 1.999803961271466e-06,
      "loss": 1.4854,
      "step": 3741
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.083726937764924,
      "learning_rate": 1.999803856174393e-06,
      "loss": 1.3262,
      "step": 3742
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.680945628502999,
      "learning_rate": 1.9998037510491585e-06,
      "loss": 1.4529,
      "step": 3743
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.213903521376912,
      "learning_rate": 1.9998036458957626e-06,
      "loss": 1.5463,
      "step": 3744
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.604742089204025,
      "learning_rate": 1.999803540714206e-06,
      "loss": 1.4577,
      "step": 3745
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.36229884657475,
      "learning_rate": 1.999803435504488e-06,
      "loss": 1.4056,
      "step": 3746
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.600285403993327,
      "learning_rate": 1.9998033302666086e-06,
      "loss": 1.4862,
      "step": 3747
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.265400100794315,
      "learning_rate": 1.9998032250005684e-06,
      "loss": 1.5089,
      "step": 3748
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.1969214189492305,
      "learning_rate": 1.9998031197063673e-06,
      "loss": 1.4962,
      "step": 3749
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.320190989350664,
      "learning_rate": 1.9998030143840045e-06,
      "loss": 1.3034,
      "step": 3750
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.579075184163126,
      "learning_rate": 1.9998029090334813e-06,
      "loss": 1.4736,
      "step": 3751
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.007083829002728,
      "learning_rate": 1.9998028036547963e-06,
      "loss": 1.3576,
      "step": 3752
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.20578658851182,
      "learning_rate": 1.9998026982479504e-06,
      "loss": 1.3413,
      "step": 3753
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.940850610449721,
      "learning_rate": 1.9998025928129432e-06,
      "loss": 1.5133,
      "step": 3754
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.822761183175982,
      "learning_rate": 1.9998024873497756e-06,
      "loss": 1.2899,
      "step": 3755
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.545241760698208,
      "learning_rate": 1.9998023818584462e-06,
      "loss": 1.3797,
      "step": 3756
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.742084979843826,
      "learning_rate": 1.999802276338956e-06,
      "loss": 1.4692,
      "step": 3757
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.658419562131754,
      "learning_rate": 1.9998021707913045e-06,
      "loss": 1.3793,
      "step": 3758
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.183021678979285,
      "learning_rate": 1.999802065215492e-06,
      "loss": 1.4525,
      "step": 3759
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.736088379622611,
      "learning_rate": 1.9998019596115183e-06,
      "loss": 1.2959,
      "step": 3760
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.656062094359612,
      "learning_rate": 1.9998018539793837e-06,
      "loss": 1.3658,
      "step": 3761
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.616745476907714,
      "learning_rate": 1.9998017483190878e-06,
      "loss": 1.4136,
      "step": 3762
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.731566326102406,
      "learning_rate": 1.999801642630631e-06,
      "loss": 1.3751,
      "step": 3763
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.432514416044134,
      "learning_rate": 1.999801536914013e-06,
      "loss": 1.3943,
      "step": 3764
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.866033462628277,
      "learning_rate": 1.999801431169234e-06,
      "loss": 1.3415,
      "step": 3765
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.468340002269182,
      "learning_rate": 1.9998013253962936e-06,
      "loss": 1.3156,
      "step": 3766
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.647820771341247,
      "learning_rate": 1.9998012195951925e-06,
      "loss": 1.3947,
      "step": 3767
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.657410377037526,
      "learning_rate": 1.99980111376593e-06,
      "loss": 1.4256,
      "step": 3768
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.8464614615398505,
      "learning_rate": 1.9998010079085066e-06,
      "loss": 1.3297,
      "step": 3769
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.332318170076133,
      "learning_rate": 1.9998009020229224e-06,
      "loss": 1.2867,
      "step": 3770
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.878338350222502,
      "learning_rate": 1.999800796109177e-06,
      "loss": 1.38,
      "step": 3771
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.493111995241049,
      "learning_rate": 1.9998006901672704e-06,
      "loss": 1.4075,
      "step": 3772
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.469116344826484,
      "learning_rate": 1.9998005841972027e-06,
      "loss": 1.4129,
      "step": 3773
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.5213045347853535,
      "learning_rate": 1.999800478198974e-06,
      "loss": 1.3233,
      "step": 3774
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.65101410292212,
      "learning_rate": 1.999800372172584e-06,
      "loss": 1.3441,
      "step": 3775
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.216969674240255,
      "learning_rate": 1.9998002661180334e-06,
      "loss": 1.2946,
      "step": 3776
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.3767982950299755,
      "learning_rate": 1.9998001600353217e-06,
      "loss": 1.4547,
      "step": 3777
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.118813225734277,
      "learning_rate": 1.9998000539244488e-06,
      "loss": 1.4563,
      "step": 3778
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.720418776141597,
      "learning_rate": 1.999799947785415e-06,
      "loss": 1.305,
      "step": 3779
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.516048109136921,
      "learning_rate": 1.9997998416182198e-06,
      "loss": 1.3212,
      "step": 3780
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.62003964568539,
      "learning_rate": 1.9997997354228637e-06,
      "loss": 1.3838,
      "step": 3781
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.0358945130998976,
      "learning_rate": 1.999799629199347e-06,
      "loss": 1.4947,
      "step": 3782
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.5824414800490185,
      "learning_rate": 1.9997995229476686e-06,
      "loss": 1.3435,
      "step": 3783
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.75725254853182,
      "learning_rate": 1.99979941666783e-06,
      "loss": 1.3627,
      "step": 3784
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.14007088469601,
      "learning_rate": 1.9997993103598295e-06,
      "loss": 1.4974,
      "step": 3785
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.60496266899906,
      "learning_rate": 1.9997992040236686e-06,
      "loss": 1.4381,
      "step": 3786
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.287067707408165,
      "learning_rate": 1.999799097659346e-06,
      "loss": 1.4361,
      "step": 3787
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.864411077361251,
      "learning_rate": 1.999798991266863e-06,
      "loss": 1.4978,
      "step": 3788
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.476884055738115,
      "learning_rate": 1.999798884846219e-06,
      "loss": 1.5751,
      "step": 3789
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.607204170305541,
      "learning_rate": 1.999798778397414e-06,
      "loss": 1.2613,
      "step": 3790
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.513574436603101,
      "learning_rate": 1.9997986719204477e-06,
      "loss": 1.2507,
      "step": 3791
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.397374539254503,
      "learning_rate": 1.9997985654153202e-06,
      "loss": 1.3785,
      "step": 3792
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.7647312609589285,
      "learning_rate": 1.999798458882032e-06,
      "loss": 1.398,
      "step": 3793
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.532986119411458,
      "learning_rate": 1.999798352320583e-06,
      "loss": 1.473,
      "step": 3794
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.289540776605135,
      "learning_rate": 1.9997982457309727e-06,
      "loss": 1.254,
      "step": 3795
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.650435124831928,
      "learning_rate": 1.9997981391132013e-06,
      "loss": 1.3785,
      "step": 3796
    },
    {
      "epoch": 0.03,
      "eval_loss": 1.6124725341796875,
      "eval_runtime": 4.6361,
      "eval_samples_per_second": 1.941,
      "eval_steps_per_second": 1.078,
      "step": 3796
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.606995994308827,
      "learning_rate": 1.9997980324672695e-06,
      "loss": 1.4598,
      "step": 3797
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.628886714284138,
      "learning_rate": 1.9997979257931764e-06,
      "loss": 1.4204,
      "step": 3798
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.472096608209409,
      "learning_rate": 1.9997978190909223e-06,
      "loss": 1.2646,
      "step": 3799
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.899108769896611,
      "learning_rate": 1.999797712360507e-06,
      "loss": 1.7302,
      "step": 3800
    },
    {
      "epoch": 0.03,
      "grad_norm": 11.698902110094348,
      "learning_rate": 1.999797605601931e-06,
      "loss": 1.2991,
      "step": 3801
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.678694191733831,
      "learning_rate": 1.999797498815194e-06,
      "loss": 1.3871,
      "step": 3802
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.623073739366768,
      "learning_rate": 1.999797392000296e-06,
      "loss": 1.4762,
      "step": 3803
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.528916340724794,
      "learning_rate": 1.999797285157237e-06,
      "loss": 1.2994,
      "step": 3804
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.120009553724364,
      "learning_rate": 1.9997971782860172e-06,
      "loss": 1.4481,
      "step": 3805
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.646312172610378,
      "learning_rate": 1.9997970713866358e-06,
      "loss": 1.4237,
      "step": 3806
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.824187645447148,
      "learning_rate": 1.9997969644590943e-06,
      "loss": 1.3742,
      "step": 3807
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.707453765528019,
      "learning_rate": 1.999796857503391e-06,
      "loss": 1.358,
      "step": 3808
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.874962378800376,
      "learning_rate": 1.9997967505195274e-06,
      "loss": 1.4034,
      "step": 3809
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.728054711081159,
      "learning_rate": 1.999796643507503e-06,
      "loss": 1.4522,
      "step": 3810
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.488435565656918,
      "learning_rate": 1.999796536467317e-06,
      "loss": 1.3188,
      "step": 3811
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.309632018627234,
      "learning_rate": 1.9997964293989703e-06,
      "loss": 1.4794,
      "step": 3812
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.21029005693669,
      "learning_rate": 1.9997963223024627e-06,
      "loss": 1.4637,
      "step": 3813
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.719975810384763,
      "learning_rate": 1.999796215177794e-06,
      "loss": 1.4294,
      "step": 3814
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.947061511680672,
      "learning_rate": 1.9997961080249648e-06,
      "loss": 1.2268,
      "step": 3815
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.717301549581428,
      "learning_rate": 1.9997960008439745e-06,
      "loss": 1.5564,
      "step": 3816
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.337568360997204,
      "learning_rate": 1.999795893634823e-06,
      "loss": 1.3617,
      "step": 3817
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.714909910098067,
      "learning_rate": 1.999795786397511e-06,
      "loss": 1.3707,
      "step": 3818
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.863348635773201,
      "learning_rate": 1.999795679132038e-06,
      "loss": 1.5601,
      "step": 3819
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.725409416597695,
      "learning_rate": 1.999795571838404e-06,
      "loss": 1.5155,
      "step": 3820
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.390907850020624,
      "learning_rate": 1.9997954645166087e-06,
      "loss": 1.3048,
      "step": 3821
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.204362569456592,
      "learning_rate": 1.9997953571666528e-06,
      "loss": 1.4248,
      "step": 3822
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.373634118732567,
      "learning_rate": 1.9997952497885363e-06,
      "loss": 1.4316,
      "step": 3823
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.830427217746328,
      "learning_rate": 1.999795142382258e-06,
      "loss": 1.5032,
      "step": 3824
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.35064585778059,
      "learning_rate": 1.99979503494782e-06,
      "loss": 1.3917,
      "step": 3825
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.621090262533992,
      "learning_rate": 1.99979492748522e-06,
      "loss": 1.4374,
      "step": 3826
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.486560370946347,
      "learning_rate": 1.9997948199944597e-06,
      "loss": 1.3844,
      "step": 3827
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.89173032023379,
      "learning_rate": 1.9997947124755385e-06,
      "loss": 1.4796,
      "step": 3828
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.968544330306352,
      "learning_rate": 1.9997946049284563e-06,
      "loss": 1.5391,
      "step": 3829
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.432121217331672,
      "learning_rate": 1.9997944973532133e-06,
      "loss": 1.3526,
      "step": 3830
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.433372309371036,
      "learning_rate": 1.9997943897498094e-06,
      "loss": 1.3612,
      "step": 3831
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.388748973282565,
      "learning_rate": 1.9997942821182442e-06,
      "loss": 1.2879,
      "step": 3832
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.522001370681306,
      "learning_rate": 1.9997941744585186e-06,
      "loss": 1.3845,
      "step": 3833
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.064642846647261,
      "learning_rate": 1.999794066770632e-06,
      "loss": 1.4568,
      "step": 3834
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.99394851245948,
      "learning_rate": 1.9997939590545846e-06,
      "loss": 1.3925,
      "step": 3835
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.180590276962394,
      "learning_rate": 1.9997938513103763e-06,
      "loss": 1.4198,
      "step": 3836
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.625626685100316,
      "learning_rate": 1.999793743538007e-06,
      "loss": 1.4152,
      "step": 3837
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.7309361359320725,
      "learning_rate": 1.999793635737477e-06,
      "loss": 1.448,
      "step": 3838
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.700706207946969,
      "learning_rate": 1.9997935279087857e-06,
      "loss": 1.4112,
      "step": 3839
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.344275675891425,
      "learning_rate": 1.9997934200519343e-06,
      "loss": 1.3493,
      "step": 3840
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.112163685600387,
      "learning_rate": 1.9997933121669216e-06,
      "loss": 1.4048,
      "step": 3841
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.620444881119614,
      "learning_rate": 1.999793204253748e-06,
      "loss": 1.48,
      "step": 3842
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.318406868845656,
      "learning_rate": 1.9997930963124135e-06,
      "loss": 1.32,
      "step": 3843
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.786834196466122,
      "learning_rate": 1.999792988342918e-06,
      "loss": 1.4818,
      "step": 3844
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.616685881971367,
      "learning_rate": 1.9997928803452624e-06,
      "loss": 1.3424,
      "step": 3845
    },
    {
      "epoch": 0.03,
      "grad_norm": 10.976292065630325,
      "learning_rate": 1.9997927723194453e-06,
      "loss": 1.529,
      "step": 3846
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.479056535258812,
      "learning_rate": 1.9997926642654677e-06,
      "loss": 1.3713,
      "step": 3847
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.29815155050577,
      "learning_rate": 1.9997925561833293e-06,
      "loss": 1.4513,
      "step": 3848
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.642297981539935,
      "learning_rate": 1.99979244807303e-06,
      "loss": 1.2959,
      "step": 3849
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.1046013534719465,
      "learning_rate": 1.9997923399345697e-06,
      "loss": 1.4979,
      "step": 3850
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.646559943022004,
      "learning_rate": 1.9997922317679486e-06,
      "loss": 1.444,
      "step": 3851
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.21849530959606,
      "learning_rate": 1.9997921235731667e-06,
      "loss": 1.5293,
      "step": 3852
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.091788114990917,
      "learning_rate": 1.999792015350224e-06,
      "loss": 1.3336,
      "step": 3853
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.027116597468565,
      "learning_rate": 1.9997919070991205e-06,
      "loss": 1.4953,
      "step": 3854
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.351333006720373,
      "learning_rate": 1.999791798819856e-06,
      "loss": 1.2278,
      "step": 3855
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.065186193436848,
      "learning_rate": 1.999791690512431e-06,
      "loss": 1.5342,
      "step": 3856
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.055266043001361,
      "learning_rate": 1.9997915821768453e-06,
      "loss": 1.4839,
      "step": 3857
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.4172809483128495,
      "learning_rate": 1.9997914738130985e-06,
      "loss": 1.3977,
      "step": 3858
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.314641655648586,
      "learning_rate": 1.9997913654211908e-06,
      "loss": 1.452,
      "step": 3859
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.308280921675756,
      "learning_rate": 1.9997912570011226e-06,
      "loss": 1.4482,
      "step": 3860
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.735078171685458,
      "learning_rate": 1.9997911485528935e-06,
      "loss": 1.3584,
      "step": 3861
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.8743091116457995,
      "learning_rate": 1.9997910400765036e-06,
      "loss": 1.4825,
      "step": 3862
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.46078710955025,
      "learning_rate": 1.9997909315719528e-06,
      "loss": 1.2225,
      "step": 3863
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.185001019529193,
      "learning_rate": 1.999790823039241e-06,
      "loss": 1.1979,
      "step": 3864
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.476108668212412,
      "learning_rate": 1.999790714478369e-06,
      "loss": 1.5096,
      "step": 3865
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.333965154940147,
      "learning_rate": 1.999790605889336e-06,
      "loss": 1.4091,
      "step": 3866
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.971115470731839,
      "learning_rate": 1.999790497272142e-06,
      "loss": 1.3178,
      "step": 3867
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.595571343395057,
      "learning_rate": 1.9997903886267876e-06,
      "loss": 1.4777,
      "step": 3868
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.804248226375533,
      "learning_rate": 1.9997902799532724e-06,
      "loss": 1.5475,
      "step": 3869
    },
    {
      "epoch": 0.03,
      "eval_loss": 1.609044075012207,
      "eval_runtime": 4.6325,
      "eval_samples_per_second": 1.943,
      "eval_steps_per_second": 1.079,
      "step": 3869
    }
  ],
  "logging_steps": 1,
  "max_steps": 591956,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 73,
  "total_flos": 405018771456000.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}