{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9996178830722201,
  "eval_steps": 500,
  "global_step": 2289,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 35.73807457632996,
      "learning_rate": 1.4492753623188408e-07,
      "loss": 1.65,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 36.337838203557865,
      "learning_rate": 2.8985507246376816e-07,
      "loss": 1.6404,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 37.40920601574936,
      "learning_rate": 4.347826086956522e-07,
      "loss": 1.6548,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 36.345930321255196,
      "learning_rate": 5.797101449275363e-07,
      "loss": 1.6241,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 35.36180102079809,
      "learning_rate": 7.246376811594204e-07,
      "loss": 1.6356,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 34.97676930558001,
      "learning_rate": 8.695652173913044e-07,
      "loss": 1.5883,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 26.980930095844375,
      "learning_rate": 1.0144927536231885e-06,
      "loss": 1.4815,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 25.740765389926157,
      "learning_rate": 1.1594202898550726e-06,
      "loss": 1.4813,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 19.58275470800386,
      "learning_rate": 1.3043478260869566e-06,
      "loss": 1.2007,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 18.75921720001582,
      "learning_rate": 1.4492753623188408e-06,
      "loss": 1.1694,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.923914509942595,
      "learning_rate": 1.5942028985507246e-06,
      "loss": 1.0849,
      "step": 11
    },
    {
      "epoch": 0.01,
      "grad_norm": 14.1258693060245,
      "learning_rate": 1.7391304347826088e-06,
      "loss": 0.8442,
      "step": 12
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.943873967878744,
      "learning_rate": 1.884057971014493e-06,
      "loss": 0.8427,
      "step": 13
    },
    {
      "epoch": 0.01,
      "grad_norm": 13.78414049555475,
      "learning_rate": 2.028985507246377e-06,
      "loss": 0.7779,
      "step": 14
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.443861817360265,
      "learning_rate": 2.173913043478261e-06,
      "loss": 0.7213,
      "step": 15
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.811054682523029,
      "learning_rate": 2.3188405797101453e-06,
      "loss": 0.566,
      "step": 16
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.724937557035619,
      "learning_rate": 2.4637681159420295e-06,
      "loss": 0.485,
      "step": 17
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.060882330349075,
      "learning_rate": 2.6086956521739132e-06,
      "loss": 0.508,
      "step": 18
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.2037513587986712,
      "learning_rate": 2.7536231884057974e-06,
      "loss": 0.4902,
      "step": 19
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.769415773704144,
      "learning_rate": 2.8985507246376816e-06,
      "loss": 0.449,
      "step": 20
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2400247527906196,
      "learning_rate": 3.043478260869566e-06,
      "loss": 0.4604,
      "step": 21
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2274134582077108,
      "learning_rate": 3.188405797101449e-06,
      "loss": 0.4588,
      "step": 22
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.9545433583380665,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.45,
      "step": 23
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.421426340123597,
      "learning_rate": 3.4782608695652175e-06,
      "loss": 0.4395,
      "step": 24
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.025809557771624,
      "learning_rate": 3.6231884057971017e-06,
      "loss": 0.4206,
      "step": 25
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2485865900648068,
      "learning_rate": 3.768115942028986e-06,
      "loss": 0.3964,
      "step": 26
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8123560890988353,
      "learning_rate": 3.91304347826087e-06,
      "loss": 0.3891,
      "step": 27
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7504422299005393,
      "learning_rate": 4.057971014492754e-06,
      "loss": 0.4162,
      "step": 28
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6012829234694324,
      "learning_rate": 4.202898550724638e-06,
      "loss": 0.3812,
      "step": 29
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7854427225930178,
      "learning_rate": 4.347826086956522e-06,
      "loss": 0.3866,
      "step": 30
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7004553032578673,
      "learning_rate": 4.492753623188406e-06,
      "loss": 0.387,
      "step": 31
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5643525350373417,
      "learning_rate": 4.637681159420291e-06,
      "loss": 0.3918,
      "step": 32
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6302980149025115,
      "learning_rate": 4.782608695652174e-06,
      "loss": 0.3822,
      "step": 33
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5742286317258862,
      "learning_rate": 4.927536231884059e-06,
      "loss": 0.3799,
      "step": 34
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.569061106925404,
      "learning_rate": 5.072463768115943e-06,
      "loss": 0.3793,
      "step": 35
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5296888673614222,
      "learning_rate": 5.2173913043478265e-06,
      "loss": 0.3828,
      "step": 36
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6499699882084182,
      "learning_rate": 5.362318840579711e-06,
      "loss": 0.3719,
      "step": 37
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.530980966423564,
      "learning_rate": 5.507246376811595e-06,
      "loss": 0.3648,
      "step": 38
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4654717691149,
      "learning_rate": 5.652173913043479e-06,
      "loss": 0.3529,
      "step": 39
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5973238828162426,
      "learning_rate": 5.797101449275363e-06,
      "loss": 0.3659,
      "step": 40
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5304452631343155,
      "learning_rate": 5.942028985507247e-06,
      "loss": 0.3788,
      "step": 41
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5058752521559429,
      "learning_rate": 6.086956521739132e-06,
      "loss": 0.3444,
      "step": 42
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6034664922333661,
      "learning_rate": 6.2318840579710145e-06,
      "loss": 0.3507,
      "step": 43
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4569227502825812,
      "learning_rate": 6.376811594202898e-06,
      "loss": 0.3749,
      "step": 44
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.360109564327896,
      "learning_rate": 6.521739130434783e-06,
      "loss": 0.3187,
      "step": 45
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3263614205490715,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.3784,
      "step": 46
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7402651983226571,
      "learning_rate": 6.811594202898551e-06,
      "loss": 0.3515,
      "step": 47
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6039056405491847,
      "learning_rate": 6.956521739130435e-06,
      "loss": 0.3687,
      "step": 48
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6788895336498273,
      "learning_rate": 7.10144927536232e-06,
      "loss": 0.3518,
      "step": 49
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.431800407454397,
      "learning_rate": 7.246376811594203e-06,
      "loss": 0.3473,
      "step": 50
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4058704598073521,
      "learning_rate": 7.391304347826087e-06,
      "loss": 0.3404,
      "step": 51
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3972883252735324,
      "learning_rate": 7.536231884057972e-06,
      "loss": 0.3574,
      "step": 52
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.401614092901069,
      "learning_rate": 7.681159420289856e-06,
      "loss": 0.3561,
      "step": 53
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6201028893532599,
      "learning_rate": 7.82608695652174e-06,
      "loss": 0.3519,
      "step": 54
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3896353358331013,
      "learning_rate": 7.971014492753623e-06,
      "loss": 0.3383,
      "step": 55
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.511580079722605,
      "learning_rate": 8.115942028985508e-06,
      "loss": 0.3488,
      "step": 56
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6061504276079601,
      "learning_rate": 8.260869565217392e-06,
      "loss": 0.3307,
      "step": 57
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9531322741910215,
      "learning_rate": 8.405797101449275e-06,
      "loss": 0.3642,
      "step": 58
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4498249211305543,
      "learning_rate": 8.55072463768116e-06,
      "loss": 0.367,
      "step": 59
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4281361527225365,
      "learning_rate": 8.695652173913044e-06,
      "loss": 0.3391,
      "step": 60
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.318884496714559,
      "learning_rate": 8.840579710144929e-06,
      "loss": 0.3569,
      "step": 61
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3992157729863128,
      "learning_rate": 8.985507246376812e-06,
      "loss": 0.3445,
      "step": 62
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5105465751564662,
      "learning_rate": 9.130434782608697e-06,
      "loss": 0.3432,
      "step": 63
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6253813815875267,
      "learning_rate": 9.275362318840581e-06,
      "loss": 0.3589,
      "step": 64
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4386792665089194,
      "learning_rate": 9.420289855072464e-06,
      "loss": 0.3525,
      "step": 65
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5738489047666984,
      "learning_rate": 9.565217391304349e-06,
      "loss": 0.3498,
      "step": 66
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1744354435977558,
      "learning_rate": 9.710144927536233e-06,
      "loss": 0.3294,
      "step": 67
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2793995646787313,
      "learning_rate": 9.855072463768118e-06,
      "loss": 0.3244,
      "step": 68
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.299653455422727,
      "learning_rate": 1e-05,
      "loss": 0.3349,
      "step": 69
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1491549538590835,
      "learning_rate": 9.999994993505624e-06,
      "loss": 0.3426,
      "step": 70
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4016869607808637,
      "learning_rate": 9.99997997403252e-06,
      "loss": 0.3428,
      "step": 71
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2406322930912719,
      "learning_rate": 9.999954941610767e-06,
      "loss": 0.349,
      "step": 72
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4418809484188368,
      "learning_rate": 9.999919896290497e-06,
      "loss": 0.3561,
      "step": 73
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3425420312422616,
      "learning_rate": 9.999874838141888e-06,
      "loss": 0.3351,
      "step": 74
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3869384669175415,
      "learning_rate": 9.999819767255175e-06,
      "loss": 0.3363,
      "step": 75
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.308663748231825,
      "learning_rate": 9.999754683740643e-06,
      "loss": 0.37,
      "step": 76
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2415436747715651,
      "learning_rate": 9.999679587728626e-06,
      "loss": 0.3294,
      "step": 77
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6063953938270403,
      "learning_rate": 9.999594479369515e-06,
      "loss": 0.3613,
      "step": 78
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4417266399829523,
      "learning_rate": 9.999499358833745e-06,
      "loss": 0.3318,
      "step": 79
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4357974243662766,
      "learning_rate": 9.999394226311804e-06,
      "loss": 0.3579,
      "step": 80
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.378332235007548,
      "learning_rate": 9.999279082014233e-06,
      "loss": 0.3396,
      "step": 81
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6687379267618878,
      "learning_rate": 9.999153926171617e-06,
      "loss": 0.3074,
      "step": 82
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8821994746516488,
      "learning_rate": 9.999018759034594e-06,
      "loss": 0.3315,
      "step": 83
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2157156963433975,
      "learning_rate": 9.998873580873848e-06,
      "loss": 0.3505,
      "step": 84
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4973028564226396,
      "learning_rate": 9.998718391980113e-06,
      "loss": 0.3509,
      "step": 85
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.36418960811051,
      "learning_rate": 9.998553192664172e-06,
      "loss": 0.3435,
      "step": 86
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2243380658463527,
      "learning_rate": 9.998377983256851e-06,
      "loss": 0.307,
      "step": 87
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2844222361953224,
      "learning_rate": 9.998192764109023e-06,
      "loss": 0.3208,
      "step": 88
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4564438033022709,
      "learning_rate": 9.99799753559161e-06,
      "loss": 0.3545,
      "step": 89
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4146326999992611,
      "learning_rate": 9.997792298095572e-06,
      "loss": 0.3366,
      "step": 90
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2017589740522183,
      "learning_rate": 9.997577052031922e-06,
      "loss": 0.3512,
      "step": 91
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3221486553839878,
      "learning_rate": 9.99735179783171e-06,
      "loss": 0.3346,
      "step": 92
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.351574646644927,
      "learning_rate": 9.997116535946028e-06,
      "loss": 0.3562,
      "step": 93
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3549239868481333,
      "learning_rate": 9.99687126684601e-06,
      "loss": 0.327,
      "step": 94
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2607609867653238,
      "learning_rate": 9.996615991022835e-06,
      "loss": 0.3527,
      "step": 95
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2228108438845886,
      "learning_rate": 9.996350708987715e-06,
      "loss": 0.3675,
      "step": 96
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8135619182098155,
      "learning_rate": 9.996075421271905e-06,
      "loss": 0.3645,
      "step": 97
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2096253360547864,
      "learning_rate": 9.995790128426693e-06,
      "loss": 0.3351,
      "step": 98
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.0816914134946107,
      "learning_rate": 9.99549483102341e-06,
      "loss": 0.3484,
      "step": 99
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2158858999536524,
      "learning_rate": 9.995189529653412e-06,
      "loss": 0.3317,
      "step": 100
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1125148925670112,
      "learning_rate": 9.9948742249281e-06,
      "loss": 0.3353,
      "step": 101
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.154913958921298,
      "learning_rate": 9.994548917478899e-06,
      "loss": 0.3253,
      "step": 102
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1860503782277776,
      "learning_rate": 9.994213607957273e-06,
      "loss": 0.337,
      "step": 103
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6264640803226373,
      "learning_rate": 9.993868297034709e-06,
      "loss": 0.3357,
      "step": 104
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.246948629705014,
      "learning_rate": 9.993512985402724e-06,
      "loss": 0.3428,
      "step": 105
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0703511808456196,
      "learning_rate": 9.993147673772869e-06,
      "loss": 0.3284,
      "step": 106
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1092877850561624,
      "learning_rate": 9.992772362876714e-06,
      "loss": 0.3206,
      "step": 107
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1836535725680641,
      "learning_rate": 9.992387053465857e-06,
      "loss": 0.3278,
      "step": 108
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2155687316654749,
      "learning_rate": 9.991991746311916e-06,
      "loss": 0.3262,
      "step": 109
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.128729533282764,
      "learning_rate": 9.991586442206533e-06,
      "loss": 0.3518,
      "step": 110
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1940325845037203,
      "learning_rate": 9.99117114196137e-06,
      "loss": 0.3029,
      "step": 111
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1573292180716739,
      "learning_rate": 9.990745846408102e-06,
      "loss": 0.3513,
      "step": 112
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.190555373026399,
      "learning_rate": 9.99031055639843e-06,
      "loss": 0.3299,
      "step": 113
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2042949502789277,
      "learning_rate": 9.989865272804064e-06,
      "loss": 0.3399,
      "step": 114
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0101322541929851,
      "learning_rate": 9.989409996516726e-06,
      "loss": 0.3183,
      "step": 115
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3368920817346246,
      "learning_rate": 9.98894472844815e-06,
      "loss": 0.3186,
      "step": 116
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2769197671622792,
      "learning_rate": 9.988469469530086e-06,
      "loss": 0.3295,
      "step": 117
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1071746988087703,
      "learning_rate": 9.987984220714283e-06,
      "loss": 0.3532,
      "step": 118
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.136251931725166,
      "learning_rate": 9.9874889829725e-06,
      "loss": 0.3261,
      "step": 119
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2071492230174605,
      "learning_rate": 9.986983757296498e-06,
      "loss": 0.3503,
      "step": 120
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2669143339556903,
      "learning_rate": 9.986468544698041e-06,
      "loss": 0.3603,
      "step": 121
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1243318949690382,
      "learning_rate": 9.985943346208893e-06,
      "loss": 0.3021,
      "step": 122
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.348701897109546,
      "learning_rate": 9.985408162880813e-06,
      "loss": 0.341,
      "step": 123
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4325327046664946,
      "learning_rate": 9.984862995785564e-06,
      "loss": 0.3566,
      "step": 124
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1225396252604363,
      "learning_rate": 9.984307846014891e-06,
      "loss": 0.3131,
      "step": 125
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.162101660380597,
      "learning_rate": 9.983742714680538e-06,
      "loss": 0.3498,
      "step": 126
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.275529218936861,
      "learning_rate": 9.983167602914236e-06,
      "loss": 0.3332,
      "step": 127
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1417668079451533,
      "learning_rate": 9.9825825118677e-06,
      "loss": 0.3185,
      "step": 128
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1572048431151127,
      "learning_rate": 9.981987442712634e-06,
      "loss": 0.307,
      "step": 129
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3892892254072602,
      "learning_rate": 9.981382396640721e-06,
      "loss": 0.3453,
      "step": 130
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.291151446457912,
      "learning_rate": 9.980767374863627e-06,
      "loss": 0.3312,
      "step": 131
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2600061479485065,
      "learning_rate": 9.98014237861299e-06,
      "loss": 0.354,
      "step": 132
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.314697173072305,
      "learning_rate": 9.97950740914043e-06,
      "loss": 0.3233,
      "step": 133
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.235849465446764,
      "learning_rate": 9.978862467717532e-06,
      "loss": 0.3394,
      "step": 134
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0430512384101303,
      "learning_rate": 9.978207555635856e-06,
      "loss": 0.3479,
      "step": 135
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0766591046439202,
      "learning_rate": 9.977542674206927e-06,
      "loss": 0.3166,
      "step": 136
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0512086591785288,
      "learning_rate": 9.976867824762236e-06,
      "loss": 0.2992,
      "step": 137
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0705583330716162,
      "learning_rate": 9.976183008653232e-06,
      "loss": 0.3386,
      "step": 138
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0717263090282247,
      "learning_rate": 9.97548822725133e-06,
      "loss": 0.3327,
      "step": 139
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1053125088464186,
      "learning_rate": 9.974783481947896e-06,
      "loss": 0.3391,
      "step": 140
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.129577985171077,
      "learning_rate": 9.974068774154252e-06,
      "loss": 0.331,
      "step": 141
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1111561902801883,
      "learning_rate": 9.97334410530167e-06,
      "loss": 0.3239,
      "step": 142
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9931276585171439,
      "learning_rate": 9.972609476841368e-06,
      "loss": 0.2919,
      "step": 143
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0366421477482668,
      "learning_rate": 9.971864890244514e-06,
      "loss": 0.3432,
      "step": 144
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.04884839867951,
      "learning_rate": 9.971110347002213e-06,
      "loss": 0.3334,
      "step": 145
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1682889506405543,
      "learning_rate": 9.970345848625516e-06,
      "loss": 0.3698,
      "step": 146
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0491781614831779,
      "learning_rate": 9.9695713966454e-06,
      "loss": 0.3136,
      "step": 147
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.129998988562572,
      "learning_rate": 9.968786992612785e-06,
      "loss": 0.3474,
      "step": 148
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0915078113082295,
      "learning_rate": 9.967992638098517e-06,
      "loss": 0.3523,
      "step": 149
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.113082868685472,
      "learning_rate": 9.967188334693364e-06,
      "loss": 0.317,
      "step": 150
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0841890611576546,
      "learning_rate": 9.966374084008025e-06,
      "loss": 0.332,
      "step": 151
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0288849241728524,
      "learning_rate": 9.965549887673121e-06,
      "loss": 0.3456,
      "step": 152
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9954927797591521,
      "learning_rate": 9.964715747339178e-06,
      "loss": 0.3112,
      "step": 153
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0652907896211132,
      "learning_rate": 9.963871664676647e-06,
      "loss": 0.3327,
      "step": 154
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0867223398675476,
      "learning_rate": 9.963017641375888e-06,
      "loss": 0.3355,
      "step": 155
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0239760840306122,
      "learning_rate": 9.962153679147162e-06,
      "loss": 0.3031,
      "step": 156
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0131654194448114,
      "learning_rate": 9.961279779720641e-06,
      "loss": 0.3351,
      "step": 157
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9864846218903062,
      "learning_rate": 9.960395944846394e-06,
      "loss": 0.2967,
      "step": 158
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1436392644930293,
      "learning_rate": 9.959502176294384e-06,
      "loss": 0.3471,
      "step": 159
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0326273531768606,
      "learning_rate": 9.958598475854474e-06,
      "loss": 0.337,
      "step": 160
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1359187088372726,
      "learning_rate": 9.957684845336409e-06,
      "loss": 0.3295,
      "step": 161
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0394183085957711,
      "learning_rate": 9.956761286569824e-06,
      "loss": 0.3154,
      "step": 162
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0048121779089239,
      "learning_rate": 9.955827801404237e-06,
      "loss": 0.3156,
      "step": 163
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0059528301982212,
      "learning_rate": 9.954884391709043e-06,
      "loss": 0.3178,
      "step": 164
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9523496815823491,
      "learning_rate": 9.95393105937351e-06,
      "loss": 0.3203,
      "step": 165
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9951160932547103,
      "learning_rate": 9.952967806306784e-06,
      "loss": 0.3205,
      "step": 166
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9906236929463421,
      "learning_rate": 9.951994634437866e-06,
      "loss": 0.3395,
      "step": 167
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0455998677239253,
      "learning_rate": 9.951011545715637e-06,
      "loss": 0.3084,
      "step": 168
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.06717349512942,
      "learning_rate": 9.950018542108818e-06,
      "loss": 0.3544,
      "step": 169
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9620677053602605,
      "learning_rate": 9.949015625606007e-06,
      "loss": 0.3112,
      "step": 170
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.023489973783828,
      "learning_rate": 9.948002798215632e-06,
      "loss": 0.3394,
      "step": 171
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0640487190161578,
      "learning_rate": 9.946980061965985e-06,
      "loss": 0.323,
      "step": 172
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0562211937503259,
      "learning_rate": 9.945947418905193e-06,
      "loss": 0.3057,
      "step": 173
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1043272091288996,
      "learning_rate": 9.944904871101227e-06,
      "loss": 0.3374,
      "step": 174
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0365376944276505,
      "learning_rate": 9.943852420641888e-06,
      "loss": 0.3409,
      "step": 175
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0864399161152527,
      "learning_rate": 9.942790069634814e-06,
      "loss": 0.3129,
      "step": 176
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0417868696762218,
      "learning_rate": 9.941717820207461e-06,
      "loss": 0.3159,
      "step": 177
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9817959926295283,
      "learning_rate": 9.94063567450712e-06,
      "loss": 0.3327,
      "step": 178
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.010724880696758,
      "learning_rate": 9.939543634700891e-06,
      "loss": 0.3421,
      "step": 179
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1866010358105539,
      "learning_rate": 9.938441702975689e-06,
      "loss": 0.331,
      "step": 180
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.113758385327317,
      "learning_rate": 9.937329881538242e-06,
      "loss": 0.3385,
      "step": 181
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.100473796760047,
      "learning_rate": 9.936208172615081e-06,
      "loss": 0.3266,
      "step": 182
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0929665506869761,
      "learning_rate": 9.935076578452535e-06,
      "loss": 0.3417,
      "step": 183
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0480297127980043,
      "learning_rate": 9.933935101316735e-06,
      "loss": 0.3111,
      "step": 184
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0642973049167792,
      "learning_rate": 9.932783743493599e-06,
      "loss": 0.3252,
      "step": 185
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1396178883933634,
      "learning_rate": 9.931622507288835e-06,
      "loss": 0.2981,
      "step": 186
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9827494909807378,
      "learning_rate": 9.930451395027932e-06,
      "loss": 0.3359,
      "step": 187
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.090238961495594,
      "learning_rate": 9.929270409056154e-06,
      "loss": 0.3326,
      "step": 188
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1694584751671635,
      "learning_rate": 9.928079551738542e-06,
      "loss": 0.3071,
      "step": 189
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0043304320478834,
      "learning_rate": 9.926878825459907e-06,
      "loss": 0.3356,
      "step": 190
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0192780459967457,
      "learning_rate": 9.925668232624819e-06,
      "loss": 0.344,
      "step": 191
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0827156805854328,
      "learning_rate": 9.924447775657607e-06,
      "loss": 0.3405,
      "step": 192
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.046363441101504,
      "learning_rate": 9.923217457002356e-06,
      "loss": 0.3291,
      "step": 193
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1256618024138416,
      "learning_rate": 9.9219772791229e-06,
      "loss": 0.3353,
      "step": 194
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0719294501344858,
      "learning_rate": 9.92072724450282e-06,
      "loss": 0.326,
      "step": 195
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0642336512900712,
      "learning_rate": 9.919467355645425e-06,
      "loss": 0.3169,
      "step": 196
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9505286995407717,
      "learning_rate": 9.91819761507377e-06,
      "loss": 0.2866,
      "step": 197
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0202998342048124,
      "learning_rate": 9.916918025330635e-06,
      "loss": 0.3153,
      "step": 198
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9181109333852953,
      "learning_rate": 9.915628588978522e-06,
      "loss": 0.3,
      "step": 199
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9024166909880759,
      "learning_rate": 9.914329308599655e-06,
      "loss": 0.3013,
      "step": 200
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9679532244049025,
      "learning_rate": 9.913020186795967e-06,
      "loss": 0.3172,
      "step": 201
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9923582344949936,
      "learning_rate": 9.911701226189105e-06,
      "loss": 0.3355,
      "step": 202
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0669793811241335,
      "learning_rate": 9.910372429420416e-06,
      "loss": 0.3129,
      "step": 203
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0400198943209167,
      "learning_rate": 9.909033799150947e-06,
      "loss": 0.3066,
      "step": 204
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.931837981455371,
      "learning_rate": 9.907685338061433e-06,
      "loss": 0.3109,
      "step": 205
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0379083212594717,
      "learning_rate": 9.906327048852301e-06,
      "loss": 0.3196,
      "step": 206
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0297167009342894,
      "learning_rate": 9.904958934243655e-06,
      "loss": 0.3181,
      "step": 207
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0339059121320457,
      "learning_rate": 9.903580996975282e-06,
      "loss": 0.3361,
      "step": 208
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1069217555831534,
      "learning_rate": 9.902193239806634e-06,
      "loss": 0.3599,
      "step": 209
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.000512508699865,
      "learning_rate": 9.90079566551683e-06,
      "loss": 0.3417,
      "step": 210
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9336825776352402,
      "learning_rate": 9.899388276904653e-06,
      "loss": 0.3216,
      "step": 211
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0393119459958182,
      "learning_rate": 9.897971076788532e-06,
      "loss": 0.3182,
      "step": 212
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9439950500295985,
      "learning_rate": 9.89654406800655e-06,
      "loss": 0.297,
      "step": 213
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0599664543275802,
      "learning_rate": 9.895107253416434e-06,
      "loss": 0.3107,
      "step": 214
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.962506557867967,
      "learning_rate": 9.893660635895542e-06,
      "loss": 0.3174,
      "step": 215
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.024756733084169,
      "learning_rate": 9.892204218340866e-06,
      "loss": 0.3185,
      "step": 216
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9344620040606235,
      "learning_rate": 9.890738003669029e-06,
      "loss": 0.3326,
      "step": 217
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9533316275317056,
      "learning_rate": 9.889261994816266e-06,
      "loss": 0.3028,
      "step": 218
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9257576463446693,
      "learning_rate": 9.887776194738433e-06,
      "loss": 0.3277,
      "step": 219
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9688174925153652,
      "learning_rate": 9.886280606410984e-06,
      "loss": 0.3132,
      "step": 220
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9894414184301703,
      "learning_rate": 9.884775232828985e-06,
      "loss": 0.3233,
      "step": 221
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.011343193746138,
      "learning_rate": 9.883260077007092e-06,
      "loss": 0.3331,
      "step": 222
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8938874657236249,
      "learning_rate": 9.881735141979554e-06,
      "loss": 0.3004,
      "step": 223
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.146006427302524,
      "learning_rate": 9.8802004308002e-06,
      "loss": 0.3192,
      "step": 224
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.071587027147118,
      "learning_rate": 9.878655946542443e-06,
      "loss": 0.3241,
      "step": 225
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1062093290365642,
      "learning_rate": 9.87710169229926e-06,
      "loss": 0.3093,
      "step": 226
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0432093777002287,
      "learning_rate": 9.8755376711832e-06,
      "loss": 0.354,
      "step": 227
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1722671889293883,
      "learning_rate": 9.873963886326365e-06,
      "loss": 0.3023,
      "step": 228
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0616226361125578,
      "learning_rate": 9.872380340880416e-06,
      "loss": 0.327,
      "step": 229
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0681445244257646,
      "learning_rate": 9.870787038016556e-06,
      "loss": 0.3121,
      "step": 230
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.116548165014327,
      "learning_rate": 9.869183980925531e-06,
      "loss": 0.3382,
      "step": 231
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0248567210767583,
      "learning_rate": 9.867571172817617e-06,
      "loss": 0.3515,
      "step": 232
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0009786406494703,
      "learning_rate": 9.865948616922623e-06,
      "loss": 0.3209,
      "step": 233
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0295898148794254,
      "learning_rate": 9.864316316489873e-06,
      "loss": 0.3179,
      "step": 234
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.047565640527205,
      "learning_rate": 9.862674274788212e-06,
      "loss": 0.3122,
      "step": 235
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8668626426643352,
      "learning_rate": 9.861022495105983e-06,
      "loss": 0.302,
      "step": 236
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.031577240411254,
      "learning_rate": 9.85936098075104e-06,
      "loss": 0.3225,
      "step": 237
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9017329575208087,
      "learning_rate": 9.85768973505073e-06,
      "loss": 0.2916,
      "step": 238
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9794214343282003,
      "learning_rate": 9.856008761351882e-06,
      "loss": 0.3017,
      "step": 239
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0483331996384981,
      "learning_rate": 9.85431806302081e-06,
      "loss": 0.3266,
      "step": 240
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9137116839774575,
      "learning_rate": 9.85261764344331e-06,
      "loss": 0.321,
      "step": 241
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9240995941272654,
      "learning_rate": 9.850907506024627e-06,
      "loss": 0.3275,
      "step": 242
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9785767598833399,
      "learning_rate": 9.849187654189486e-06,
      "loss": 0.3348,
      "step": 243
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0254598805628692,
      "learning_rate": 9.847458091382057e-06,
      "loss": 0.3238,
      "step": 244
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9789954975865619,
      "learning_rate": 9.845718821065957e-06,
      "loss": 0.3385,
      "step": 245
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.997158151024336,
      "learning_rate": 9.843969846724247e-06,
      "loss": 0.3262,
      "step": 246
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9907284723441283,
      "learning_rate": 9.842211171859418e-06,
      "loss": 0.2987,
      "step": 247
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9004222343845464,
      "learning_rate": 9.840442799993389e-06,
      "loss": 0.3001,
      "step": 248
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9962295116550975,
      "learning_rate": 9.838664734667496e-06,
      "loss": 0.3175,
      "step": 249
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9832268420119982,
      "learning_rate": 9.83687697944249e-06,
      "loss": 0.3286,
      "step": 250
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0831675309800166,
      "learning_rate": 9.835079537898523e-06,
      "loss": 0.3186,
      "step": 251
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0350045444569569,
      "learning_rate": 9.833272413635151e-06,
      "loss": 0.3423,
      "step": 252
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9042509744057806,
      "learning_rate": 9.831455610271315e-06,
      "loss": 0.3118,
      "step": 253
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9856470160579057,
      "learning_rate": 9.829629131445342e-06,
      "loss": 0.2918,
      "step": 254
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9895637897399454,
      "learning_rate": 9.827792980814934e-06,
      "loss": 0.2967,
      "step": 255
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1382689109804724,
      "learning_rate": 9.825947162057163e-06,
      "loss": 0.3399,
      "step": 256
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9769027584759491,
      "learning_rate": 9.82409167886846e-06,
      "loss": 0.3247,
      "step": 257
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1103451259765424,
      "learning_rate": 9.822226534964613e-06,
      "loss": 0.3311,
      "step": 258
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1310859257818386,
      "learning_rate": 9.820351734080754e-06,
      "loss": 0.3219,
      "step": 259
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0796246383665198,
      "learning_rate": 9.818467279971356e-06,
      "loss": 0.3489,
      "step": 260
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9847947599842716,
      "learning_rate": 9.81657317641022e-06,
      "loss": 0.3283,
      "step": 261
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9496976838797854,
      "learning_rate": 9.814669427190477e-06,
      "loss": 0.3502,
      "step": 262
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0472770887281222,
      "learning_rate": 9.81275603612457e-06,
      "loss": 0.3394,
      "step": 263
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0070754117382343,
      "learning_rate": 9.810833007044247e-06,
      "loss": 0.3181,
      "step": 264
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0188471546966655,
      "learning_rate": 9.808900343800568e-06,
      "loss": 0.3287,
      "step": 265
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0197814120489197,
      "learning_rate": 9.80695805026388e-06,
      "loss": 0.3222,
      "step": 266
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0217941213476793,
      "learning_rate": 9.80500613032381e-06,
      "loss": 0.34,
      "step": 267
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9396521010248938,
      "learning_rate": 9.803044587889273e-06,
      "loss": 0.316,
      "step": 268
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9538783688300877,
      "learning_rate": 9.801073426888447e-06,
      "loss": 0.3007,
      "step": 269
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8670634540671506,
      "learning_rate": 9.799092651268778e-06,
      "loss": 0.3135,
      "step": 270
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9659798957000564,
      "learning_rate": 9.79710226499696e-06,
      "loss": 0.3152,
      "step": 271
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9198004755575426,
      "learning_rate": 9.795102272058937e-06,
      "loss": 0.311,
      "step": 272
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.980462331281668,
      "learning_rate": 9.79309267645989e-06,
      "loss": 0.3436,
      "step": 273
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9974269139168662,
      "learning_rate": 9.791073482224229e-06,
      "loss": 0.3294,
      "step": 274
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.934210468241927,
      "learning_rate": 9.789044693395593e-06,
      "loss": 0.3189,
      "step": 275
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8623292802196376,
      "learning_rate": 9.787006314036824e-06,
      "loss": 0.3095,
      "step": 276
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8946712986381768,
      "learning_rate": 9.784958348229978e-06,
      "loss": 0.3308,
      "step": 277
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9212331440061751,
      "learning_rate": 9.78290080007631e-06,
      "loss": 0.3153,
      "step": 278
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9485556430447161,
      "learning_rate": 9.780833673696255e-06,
      "loss": 0.3347,
      "step": 279
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.90415014153471,
      "learning_rate": 9.778756973229442e-06,
      "loss": 0.3129,
      "step": 280
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0026095874856258,
      "learning_rate": 9.776670702834662e-06,
      "loss": 0.3142,
      "step": 281
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9264822440275213,
      "learning_rate": 9.774574866689878e-06,
      "loss": 0.323,
      "step": 282
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8933797451503408,
      "learning_rate": 9.772469468992206e-06,
      "loss": 0.3284,
      "step": 283
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8716653148093717,
      "learning_rate": 9.77035451395791e-06,
      "loss": 0.3044,
      "step": 284
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9837074395439597,
      "learning_rate": 9.768230005822394e-06,
      "loss": 0.3585,
      "step": 285
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8711414563618449,
      "learning_rate": 9.766095948840195e-06,
      "loss": 0.3065,
      "step": 286
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9581948215961118,
      "learning_rate": 9.76395234728497e-06,
      "loss": 0.3408,
      "step": 287
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0486497013717402,
      "learning_rate": 9.761799205449492e-06,
      "loss": 0.3182,
      "step": 288
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9558738269559783,
      "learning_rate": 9.759636527645633e-06,
      "loss": 0.3375,
      "step": 289
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8807154291111594,
      "learning_rate": 9.757464318204374e-06,
      "loss": 0.2974,
      "step": 290
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8503009410121021,
      "learning_rate": 9.755282581475769e-06,
      "loss": 0.2867,
      "step": 291
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9267744472702208,
      "learning_rate": 9.753091321828964e-06,
      "loss": 0.3217,
      "step": 292
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9243147043407282,
      "learning_rate": 9.750890543652169e-06,
      "loss": 0.3222,
      "step": 293
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0113192854811468,
      "learning_rate": 9.74868025135266e-06,
      "loss": 0.3135,
      "step": 294
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9359470707084092,
      "learning_rate": 9.74646044935676e-06,
      "loss": 0.3313,
      "step": 295
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9195246840869694,
      "learning_rate": 9.744231142109842e-06,
      "loss": 0.3205,
      "step": 296
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8939206012032249,
      "learning_rate": 9.741992334076309e-06,
      "loss": 0.3129,
      "step": 297
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0277009933307621,
      "learning_rate": 9.739744029739596e-06,
      "loss": 0.3026,
      "step": 298
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9422042523836677,
      "learning_rate": 9.737486233602149e-06,
      "loss": 0.3337,
      "step": 299
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9300639179140263,
      "learning_rate": 9.735218950185428e-06,
      "loss": 0.334,
      "step": 300
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0469084498971397,
      "learning_rate": 9.73294218402989e-06,
      "loss": 0.2958,
      "step": 301
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.075290826100831,
      "learning_rate": 9.730655939694979e-06,
      "loss": 0.3005,
      "step": 302
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8819735389627644,
      "learning_rate": 9.728360221759125e-06,
      "loss": 0.3226,
      "step": 303
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0062675321359353,
      "learning_rate": 9.726055034819726e-06,
      "loss": 0.334,
      "step": 304
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.103564420886612,
      "learning_rate": 9.723740383493146e-06,
      "loss": 0.3225,
      "step": 305
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8794169380632961,
      "learning_rate": 9.721416272414698e-06,
      "loss": 0.3,
      "step": 306
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9446412618415204,
      "learning_rate": 9.719082706238645e-06,
      "loss": 0.3354,
      "step": 307
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8939035409014716,
      "learning_rate": 9.716739689638179e-06,
      "loss": 0.319,
      "step": 308
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9214282033765197,
      "learning_rate": 9.714387227305422e-06,
      "loss": 0.2959,
      "step": 309
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.946586774336498,
      "learning_rate": 9.712025323951406e-06,
      "loss": 0.297,
      "step": 310
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9311875204422642,
      "learning_rate": 9.709653984306077e-06,
      "loss": 0.338,
      "step": 311
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9828016144347893,
      "learning_rate": 9.70727321311827e-06,
      "loss": 0.3151,
      "step": 312
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8598618875563708,
      "learning_rate": 9.70488301515572e-06,
      "loss": 0.3014,
      "step": 313
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9619314866860545,
      "learning_rate": 9.702483395205023e-06,
      "loss": 0.3077,
      "step": 314
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9217099396966812,
      "learning_rate": 9.700074358071658e-06,
      "loss": 0.3228,
      "step": 315
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0097212015208115,
      "learning_rate": 9.697655908579958e-06,
      "loss": 0.3409,
      "step": 316
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8935738086422582,
      "learning_rate": 9.695228051573098e-06,
      "loss": 0.3054,
      "step": 317
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7987930403698044,
      "learning_rate": 9.692790791913107e-06,
      "loss": 0.303,
      "step": 318
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9780694964806677,
      "learning_rate": 9.69034413448083e-06,
      "loss": 0.3339,
      "step": 319
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8691262829593847,
      "learning_rate": 9.68788808417594e-06,
      "loss": 0.3126,
      "step": 320
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9559138464147898,
      "learning_rate": 9.68542264591692e-06,
      "loss": 0.3112,
      "step": 321
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9420747675280177,
      "learning_rate": 9.682947824641045e-06,
      "loss": 0.3228,
      "step": 322
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9398589435244177,
      "learning_rate": 9.680463625304393e-06,
      "loss": 0.2885,
      "step": 323
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9067272067524683,
      "learning_rate": 9.677970052881811e-06,
      "loss": 0.2855,
      "step": 324
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9076533520634805,
      "learning_rate": 9.675467112366925e-06,
      "loss": 0.3097,
      "step": 325
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8885207370107971,
      "learning_rate": 9.672954808772118e-06,
      "loss": 0.3347,
      "step": 326
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9959164279908957,
      "learning_rate": 9.670433147128522e-06,
      "loss": 0.3202,
      "step": 327
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0008517998861886,
      "learning_rate": 9.667902132486009e-06,
      "loss": 0.3095,
      "step": 328
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8682036084811254,
      "learning_rate": 9.665361769913187e-06,
      "loss": 0.3516,
      "step": 329
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.921987724740865,
      "learning_rate": 9.66281206449738e-06,
      "loss": 0.3053,
      "step": 330
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8558190966517871,
      "learning_rate": 9.660253021344618e-06,
      "loss": 0.2903,
      "step": 331
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9471627697819528,
      "learning_rate": 9.65768464557964e-06,
      "loss": 0.3029,
      "step": 332
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.057820123899647,
      "learning_rate": 9.65510694234587e-06,
      "loss": 0.3178,
      "step": 333
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9676868616711124,
      "learning_rate": 9.652519916805406e-06,
      "loss": 0.3071,
      "step": 334
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9971733654273649,
      "learning_rate": 9.649923574139023e-06,
      "loss": 0.3065,
      "step": 335
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9616833867639251,
      "learning_rate": 9.64731791954615e-06,
      "loss": 0.3285,
      "step": 336
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8522999435781541,
      "learning_rate": 9.644702958244865e-06,
      "loss": 0.3147,
      "step": 337
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9015718563535392,
      "learning_rate": 9.642078695471882e-06,
      "loss": 0.3011,
      "step": 338
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9540590738234375,
      "learning_rate": 9.639445136482549e-06,
      "loss": 0.3371,
      "step": 339
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9374772216035895,
      "learning_rate": 9.636802286550817e-06,
      "loss": 0.3227,
      "step": 340
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9464222600179373,
      "learning_rate": 9.634150150969259e-06,
      "loss": 0.3157,
      "step": 341
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8450263846668521,
      "learning_rate": 9.631488735049033e-06,
      "loss": 0.3097,
      "step": 342
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9074355402631343,
      "learning_rate": 9.628818044119884e-06,
      "loss": 0.3325,
      "step": 343
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9126751710464209,
      "learning_rate": 9.62613808353013e-06,
      "loss": 0.2958,
      "step": 344
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9183905621494797,
      "learning_rate": 9.623448858646658e-06,
      "loss": 0.3038,
      "step": 345
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.037677363620643,
      "learning_rate": 9.620750374854901e-06,
      "loss": 0.319,
      "step": 346
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9810815532267979,
      "learning_rate": 9.618042637558838e-06,
      "loss": 0.3163,
      "step": 347
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.036728587151219,
      "learning_rate": 9.615325652180976e-06,
      "loss": 0.3178,
      "step": 348
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9587487706385894,
      "learning_rate": 9.612599424162344e-06,
      "loss": 0.2966,
      "step": 349
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8679911279213723,
      "learning_rate": 9.609863958962483e-06,
      "loss": 0.3091,
      "step": 350
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8972868571666467,
      "learning_rate": 9.607119262059426e-06,
      "loss": 0.3197,
      "step": 351
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8381622808634421,
      "learning_rate": 9.604365338949697e-06,
      "loss": 0.3248,
      "step": 352
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9424737256212821,
      "learning_rate": 9.601602195148299e-06,
      "loss": 0.301,
      "step": 353
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.851688944869089,
      "learning_rate": 9.598829836188694e-06,
      "loss": 0.2983,
      "step": 354
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8181263204502468,
      "learning_rate": 9.596048267622806e-06,
      "loss": 0.2868,
      "step": 355
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.86828296499146,
      "learning_rate": 9.593257495020993e-06,
      "loss": 0.3302,
      "step": 356
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9179766453146265,
      "learning_rate": 9.590457523972055e-06,
      "loss": 0.3079,
      "step": 357
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8263474679986963,
      "learning_rate": 9.587648360083205e-06,
      "loss": 0.2998,
      "step": 358
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8486904908001796,
      "learning_rate": 9.584830008980068e-06,
      "loss": 0.3019,
      "step": 359
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8767035678377422,
      "learning_rate": 9.582002476306668e-06,
      "loss": 0.3042,
      "step": 360
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.913170998401528,
      "learning_rate": 9.579165767725416e-06,
      "loss": 0.3302,
      "step": 361
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8219170556382018,
      "learning_rate": 9.576319888917099e-06,
      "loss": 0.2984,
      "step": 362
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9074705618134189,
      "learning_rate": 9.573464845580864e-06,
      "loss": 0.2972,
      "step": 363
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8945623228454003,
      "learning_rate": 9.570600643434217e-06,
      "loss": 0.2916,
      "step": 364
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8789337173713913,
      "learning_rate": 9.567727288213005e-06,
      "loss": 0.2939,
      "step": 365
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9252097031185456,
      "learning_rate": 9.5648447856714e-06,
      "loss": 0.2853,
      "step": 366
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8806268680502126,
      "learning_rate": 9.561953141581895e-06,
      "loss": 0.3145,
      "step": 367
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8983089945587068,
      "learning_rate": 9.559052361735288e-06,
      "loss": 0.2885,
      "step": 368
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8844006801209731,
      "learning_rate": 9.55614245194068e-06,
      "loss": 0.3045,
      "step": 369
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9310091416061946,
      "learning_rate": 9.553223418025443e-06,
      "loss": 0.3238,
      "step": 370
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9324717519260864,
      "learning_rate": 9.550295265835235e-06,
      "loss": 0.3133,
      "step": 371
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9355732669776776,
      "learning_rate": 9.547358001233959e-06,
      "loss": 0.3122,
      "step": 372
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8544068487335935,
      "learning_rate": 9.544411630103782e-06,
      "loss": 0.315,
      "step": 373
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8512240344848098,
      "learning_rate": 9.541456158345094e-06,
      "loss": 0.3036,
      "step": 374
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9169456837785289,
      "learning_rate": 9.538491591876522e-06,
      "loss": 0.3244,
      "step": 375
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8808156831500024,
      "learning_rate": 9.535517936634895e-06,
      "loss": 0.3112,
      "step": 376
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9395773422746726,
      "learning_rate": 9.53253519857525e-06,
      "loss": 0.3181,
      "step": 377
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8753849848595671,
      "learning_rate": 9.529543383670814e-06,
      "loss": 0.3358,
      "step": 378
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8897094917257743,
      "learning_rate": 9.526542497912984e-06,
      "loss": 0.3128,
      "step": 379
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8431898603626422,
      "learning_rate": 9.523532547311332e-06,
      "loss": 0.3117,
      "step": 380
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0154687474747905,
      "learning_rate": 9.520513537893574e-06,
      "loss": 0.3359,
      "step": 381
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8207880208118807,
      "learning_rate": 9.517485475705574e-06,
      "loss": 0.2804,
      "step": 382
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8366353086574577,
      "learning_rate": 9.514448366811322e-06,
      "loss": 0.3286,
      "step": 383
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9458083635163719,
      "learning_rate": 9.511402217292927e-06,
      "loss": 0.3046,
      "step": 384
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9803958799515059,
      "learning_rate": 9.508347033250597e-06,
      "loss": 0.3174,
      "step": 385
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9793231380933154,
      "learning_rate": 9.505282820802642e-06,
      "loss": 0.3289,
      "step": 386
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8792898957251518,
      "learning_rate": 9.502209586085444e-06,
      "loss": 0.3197,
      "step": 387
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8818269656049114,
      "learning_rate": 9.499127335253456e-06,
      "loss": 0.2963,
      "step": 388
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8633066021877013,
      "learning_rate": 9.496036074479184e-06,
      "loss": 0.3264,
      "step": 389
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8299459760038647,
      "learning_rate": 9.492935809953185e-06,
      "loss": 0.3055,
      "step": 390
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.911947276006245,
      "learning_rate": 9.48982654788404e-06,
      "loss": 0.3444,
      "step": 391
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9246781743945012,
      "learning_rate": 9.486708294498348e-06,
      "loss": 0.3189,
      "step": 392
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9135965371658781,
      "learning_rate": 9.48358105604072e-06,
      "loss": 0.2909,
      "step": 393
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8601441979856773,
      "learning_rate": 9.480444838773753e-06,
      "loss": 0.335,
      "step": 394
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8779227508609755,
      "learning_rate": 9.47729964897803e-06,
      "loss": 0.3158,
      "step": 395
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8983702184601497,
      "learning_rate": 9.474145492952102e-06,
      "loss": 0.2956,
      "step": 396
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8830400002486968,
      "learning_rate": 9.470982377012475e-06,
      "loss": 0.3179,
      "step": 397
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8014074483472782,
      "learning_rate": 9.467810307493595e-06,
      "loss": 0.3058,
      "step": 398
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8618655786943608,
      "learning_rate": 9.464629290747844e-06,
      "loss": 0.3131,
      "step": 399
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9663778320902771,
      "learning_rate": 9.461439333145516e-06,
      "loss": 0.3274,
      "step": 400
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.929473290364399,
      "learning_rate": 9.458240441074818e-06,
      "loss": 0.3263,
      "step": 401
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8447229287878972,
      "learning_rate": 9.45503262094184e-06,
      "loss": 0.2992,
      "step": 402
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9167546604102287,
      "learning_rate": 9.451815879170556e-06,
      "loss": 0.3144,
      "step": 403
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9051539526188835,
      "learning_rate": 9.448590222202808e-06,
      "loss": 0.3269,
      "step": 404
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7920834539574064,
      "learning_rate": 9.445355656498284e-06,
      "loss": 0.287,
      "step": 405
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9569197604942764,
      "learning_rate": 9.442112188534526e-06,
      "loss": 0.3276,
      "step": 406
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7855433293363763,
      "learning_rate": 9.438859824806889e-06,
      "loss": 0.2845,
      "step": 407
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8970435894287223,
      "learning_rate": 9.435598571828552e-06,
      "loss": 0.2948,
      "step": 408
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.027221232184958,
      "learning_rate": 9.432328436130493e-06,
      "loss": 0.3092,
      "step": 409
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9000524446346089,
      "learning_rate": 9.429049424261478e-06,
      "loss": 0.3042,
      "step": 410
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8670829185035347,
      "learning_rate": 9.425761542788049e-06,
      "loss": 0.2947,
      "step": 411
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.885547700386459,
      "learning_rate": 9.422464798294509e-06,
      "loss": 0.3188,
      "step": 412
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8845214080856865,
      "learning_rate": 9.419159197382913e-06,
      "loss": 0.2889,
      "step": 413
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8539347231721706,
      "learning_rate": 9.415844746673047e-06,
      "loss": 0.3153,
      "step": 414
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8281540948276968,
      "learning_rate": 9.412521452802425e-06,
      "loss": 0.3154,
      "step": 415
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8838333675828067,
      "learning_rate": 9.409189322426267e-06,
      "loss": 0.3043,
      "step": 416
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9442860245620236,
      "learning_rate": 9.40584836221749e-06,
      "loss": 0.3228,
      "step": 417
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.928241048197894,
      "learning_rate": 9.402498578866695e-06,
      "loss": 0.3458,
      "step": 418
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8932331120871652,
      "learning_rate": 9.399139979082148e-06,
      "loss": 0.3199,
      "step": 419
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8828285376497922,
      "learning_rate": 9.395772569589774e-06,
      "loss": 0.2698,
      "step": 420
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.917564961215885,
      "learning_rate": 9.39239635713314e-06,
      "loss": 0.3291,
      "step": 421
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9004137989338258,
      "learning_rate": 9.389011348473442e-06,
      "loss": 0.3367,
      "step": 422
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9347015329110837,
      "learning_rate": 9.38561755038949e-06,
      "loss": 0.3261,
      "step": 423
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8688135800195382,
      "learning_rate": 9.382214969677697e-06,
      "loss": 0.3233,
      "step": 424
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8452726507806062,
      "learning_rate": 9.37880361315206e-06,
      "loss": 0.3129,
      "step": 425
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9352844102119053,
      "learning_rate": 9.375383487644161e-06,
      "loss": 0.3414,
      "step": 426
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8880205351400158,
      "learning_rate": 9.37195460000313e-06,
      "loss": 0.2933,
      "step": 427
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8853512727009014,
      "learning_rate": 9.368516957095652e-06,
      "loss": 0.3235,
      "step": 428
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8729399632592538,
      "learning_rate": 9.365070565805941e-06,
      "loss": 0.3301,
      "step": 429
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8560073614657919,
      "learning_rate": 9.361615433035736e-06,
      "loss": 0.3126,
      "step": 430
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8565008118616368,
      "learning_rate": 9.358151565704275e-06,
      "loss": 0.2776,
      "step": 431
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9211484980738996,
      "learning_rate": 9.35467897074829e-06,
      "loss": 0.3279,
      "step": 432
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8784300130804092,
      "learning_rate": 9.351197655121995e-06,
      "loss": 0.2884,
      "step": 433
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9740848949459224,
      "learning_rate": 9.347707625797062e-06,
      "loss": 0.3135,
      "step": 434
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8247162909763773,
      "learning_rate": 9.34420888976262e-06,
      "loss": 0.2847,
      "step": 435
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8363625539341427,
      "learning_rate": 9.340701454025224e-06,
      "loss": 0.3282,
      "step": 436
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0047197819011753,
      "learning_rate": 9.337185325608864e-06,
      "loss": 0.3221,
      "step": 437
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9172071054159138,
      "learning_rate": 9.333660511554926e-06,
      "loss": 0.3151,
      "step": 438
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7977217629335378,
      "learning_rate": 9.330127018922195e-06,
      "loss": 0.3061,
      "step": 439
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8387112803726957,
      "learning_rate": 9.326584854786835e-06,
      "loss": 0.3326,
      "step": 440
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8790163202803737,
      "learning_rate": 9.323034026242378e-06,
      "loss": 0.3374,
      "step": 441
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.002115439900629,
      "learning_rate": 9.319474540399704e-06,
      "loss": 0.3041,
      "step": 442
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.778905889651517,
      "learning_rate": 9.315906404387032e-06,
      "loss": 0.2905,
      "step": 443
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.788589250448959,
      "learning_rate": 9.312329625349903e-06,
      "loss": 0.2981,
      "step": 444
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8910431380615856,
      "learning_rate": 9.308744210451167e-06,
      "loss": 0.3168,
      "step": 445
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9240840518263157,
      "learning_rate": 9.305150166870966e-06,
      "loss": 0.308,
      "step": 446
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8528326026852335,
      "learning_rate": 9.301547501806725e-06,
      "loss": 0.306,
      "step": 447
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9303492007769486,
      "learning_rate": 9.297936222473134e-06,
      "loss": 0.3058,
      "step": 448
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9315910060259597,
      "learning_rate": 9.294316336102132e-06,
      "loss": 0.3065,
      "step": 449
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9349664575440197,
      "learning_rate": 9.290687849942893e-06,
      "loss": 0.3078,
      "step": 450
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8178289497161899,
      "learning_rate": 9.287050771261818e-06,
      "loss": 0.2918,
      "step": 451
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9453421630706376,
      "learning_rate": 9.283405107342512e-06,
      "loss": 0.2998,
      "step": 452
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8847335445080732,
      "learning_rate": 9.279750865485772e-06,
      "loss": 0.3153,
      "step": 453
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9025983943811766,
      "learning_rate": 9.276088053009578e-06,
      "loss": 0.3253,
      "step": 454
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9301100353457417,
      "learning_rate": 9.272416677249067e-06,
      "loss": 0.3193,
      "step": 455
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9120993008602606,
      "learning_rate": 9.268736745556528e-06,
      "loss": 0.3124,
      "step": 456
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9408970304364769,
      "learning_rate": 9.265048265301385e-06,
      "loss": 0.332,
      "step": 457
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.840559768388453,
      "learning_rate": 9.261351243870178e-06,
      "loss": 0.3066,
      "step": 458
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9419249145321315,
      "learning_rate": 9.257645688666557e-06,
      "loss": 0.3049,
      "step": 459
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8642824029099644,
      "learning_rate": 9.253931607111257e-06,
      "loss": 0.3074,
      "step": 460
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8577006648395137,
      "learning_rate": 9.25020900664209e-06,
      "loss": 0.2932,
      "step": 461
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8446394569038618,
      "learning_rate": 9.246477894713925e-06,
      "loss": 0.2872,
      "step": 462
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9422399610480132,
      "learning_rate": 9.242738278798682e-06,
      "loss": 0.3267,
      "step": 463
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8538920316540486,
      "learning_rate": 9.238990166385304e-06,
      "loss": 0.3107,
      "step": 464
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9073765783971015,
      "learning_rate": 9.235233564979756e-06,
      "loss": 0.2916,
      "step": 465
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9174691136515106,
      "learning_rate": 9.231468482104994e-06,
      "loss": 0.304,
      "step": 466
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9131656688422567,
      "learning_rate": 9.227694925300972e-06,
      "loss": 0.3122,
      "step": 467
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8426350652161437,
      "learning_rate": 9.223912902124601e-06,
      "loss": 0.3198,
      "step": 468
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.964749237860055,
      "learning_rate": 9.220122420149753e-06,
      "loss": 0.295,
      "step": 469
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8254006549679825,
      "learning_rate": 9.216323486967239e-06,
      "loss": 0.294,
      "step": 470
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9098730868260394,
      "learning_rate": 9.212516110184794e-06,
      "loss": 0.3188,
      "step": 471
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8979968693532835,
      "learning_rate": 9.208700297427063e-06,
      "loss": 0.2992,
      "step": 472
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8499979133081547,
      "learning_rate": 9.204876056335584e-06,
      "loss": 0.3091,
      "step": 473
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9372165714579413,
      "learning_rate": 9.201043394568773e-06,
      "loss": 0.3197,
      "step": 474
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8731997283861855,
      "learning_rate": 9.197202319801908e-06,
      "loss": 0.2923,
      "step": 475
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8187450928665471,
      "learning_rate": 9.193352839727122e-06,
      "loss": 0.3003,
      "step": 476
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8748478891154613,
      "learning_rate": 9.18949496205337e-06,
      "loss": 0.2909,
      "step": 477
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8904029555440679,
      "learning_rate": 9.18562869450643e-06,
      "loss": 0.2829,
      "step": 478
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8051855942505556,
      "learning_rate": 9.181754044828882e-06,
      "loss": 0.3031,
      "step": 479
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8140454424291693,
      "learning_rate": 9.17787102078009e-06,
      "loss": 0.3055,
      "step": 480
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8405469803661499,
      "learning_rate": 9.17397963013619e-06,
      "loss": 0.3071,
      "step": 481
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8417842777518616,
      "learning_rate": 9.170079880690072e-06,
      "loss": 0.3053,
      "step": 482
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7983052703708655,
      "learning_rate": 9.166171780251365e-06,
      "loss": 0.3079,
      "step": 483
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8586638307943795,
      "learning_rate": 9.162255336646422e-06,
      "loss": 0.3269,
      "step": 484
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8899037130990646,
      "learning_rate": 9.158330557718304e-06,
      "loss": 0.3096,
      "step": 485
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8110590401332586,
      "learning_rate": 9.154397451326766e-06,
      "loss": 0.2922,
      "step": 486
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.837172667759998,
      "learning_rate": 9.150456025348236e-06,
      "loss": 0.3134,
      "step": 487
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9398061464425186,
      "learning_rate": 9.146506287675805e-06,
      "loss": 0.3184,
      "step": 488
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8741103922343479,
      "learning_rate": 9.142548246219212e-06,
      "loss": 0.3126,
      "step": 489
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8438024830877229,
      "learning_rate": 9.138581908904818e-06,
      "loss": 0.3026,
      "step": 490
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8477742310775299,
      "learning_rate": 9.134607283675601e-06,
      "loss": 0.3031,
      "step": 491
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8103945267031356,
      "learning_rate": 9.13062437849114e-06,
      "loss": 0.2951,
      "step": 492
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9328728079752384,
      "learning_rate": 9.12663320132759e-06,
      "loss": 0.3065,
      "step": 493
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8173940230119922,
      "learning_rate": 9.122633760177674e-06,
      "loss": 0.2812,
      "step": 494
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8979161831110007,
      "learning_rate": 9.118626063050661e-06,
      "loss": 0.3354,
      "step": 495
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8424221922840908,
      "learning_rate": 9.114610117972362e-06,
      "loss": 0.293,
      "step": 496
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8729013998407127,
      "learning_rate": 9.110585932985095e-06,
      "loss": 0.3117,
      "step": 497
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8630979191663862,
      "learning_rate": 9.106553516147683e-06,
      "loss": 0.3051,
      "step": 498
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8812551568960534,
      "learning_rate": 9.102512875535439e-06,
      "loss": 0.3212,
      "step": 499
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8629451810854976,
      "learning_rate": 9.098464019240137e-06,
      "loss": 0.2955,
      "step": 500
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8927808369679009,
      "learning_rate": 9.09440695537001e-06,
      "loss": 0.3068,
      "step": 501
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9311616061957062,
      "learning_rate": 9.090341692049725e-06,
      "loss": 0.3236,
      "step": 502
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8325777039955035,
      "learning_rate": 9.086268237420368e-06,
      "loss": 0.3109,
      "step": 503
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8429547937591914,
      "learning_rate": 9.082186599639429e-06,
      "loss": 0.3189,
      "step": 504
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8033989686061983,
      "learning_rate": 9.07809678688079e-06,
      "loss": 0.276,
      "step": 505
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9273092043815968,
      "learning_rate": 9.073998807334698e-06,
      "loss": 0.2879,
      "step": 506
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8864779894336426,
      "learning_rate": 9.069892669207757e-06,
      "loss": 0.2852,
      "step": 507
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8070366853642637,
      "learning_rate": 9.065778380722914e-06,
      "loss": 0.3177,
      "step": 508
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7898091340558379,
      "learning_rate": 9.06165595011943e-06,
      "loss": 0.2894,
      "step": 509
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8488185731269791,
      "learning_rate": 9.057525385652877e-06,
      "loss": 0.3102,
      "step": 510
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9321456586746238,
      "learning_rate": 9.053386695595114e-06,
      "loss": 0.3123,
      "step": 511
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.848988300187308,
      "learning_rate": 9.049239888234271e-06,
      "loss": 0.3031,
      "step": 512
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9327264500843722,
      "learning_rate": 9.045084971874738e-06,
      "loss": 0.3145,
      "step": 513
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8136609174086732,
      "learning_rate": 9.040921954837139e-06,
      "loss": 0.2927,
      "step": 514
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9196585002829474,
      "learning_rate": 9.036750845458323e-06,
      "loss": 0.3101,
      "step": 515
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8782398445745344,
      "learning_rate": 9.032571652091341e-06,
      "loss": 0.2958,
      "step": 516
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.79330562988252,
      "learning_rate": 9.028384383105444e-06,
      "loss": 0.2765,
      "step": 517
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8498591219048203,
      "learning_rate": 9.02418904688604e-06,
      "loss": 0.3249,
      "step": 518
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8840303612588961,
      "learning_rate": 9.019985651834703e-06,
      "loss": 0.3058,
      "step": 519
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.858564845394782,
      "learning_rate": 9.015774206369144e-06,
      "loss": 0.2978,
      "step": 520
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.836339561119018,
      "learning_rate": 9.01155471892319e-06,
      "loss": 0.318,
      "step": 521
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8533588096441308,
      "learning_rate": 9.007327197946782e-06,
      "loss": 0.2714,
      "step": 522
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8542989681686293,
      "learning_rate": 9.00309165190594e-06,
      "loss": 0.3069,
      "step": 523
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8301681540807847,
      "learning_rate": 8.99884808928276e-06,
      "loss": 0.3066,
      "step": 524
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8922722213364641,
      "learning_rate": 8.994596518575393e-06,
      "loss": 0.3235,
      "step": 525
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8761047110857169,
      "learning_rate": 8.990336948298023e-06,
      "loss": 0.3204,
      "step": 526
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8503189815343855,
      "learning_rate": 8.986069386980855e-06,
      "loss": 0.319,
      "step": 527
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9010588779467374,
      "learning_rate": 8.9817938431701e-06,
      "loss": 0.3165,
      "step": 528
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9116558184471515,
      "learning_rate": 8.97751032542795e-06,
      "loss": 0.3068,
      "step": 529
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7934539728328021,
      "learning_rate": 8.97321884233257e-06,
      "loss": 0.282,
      "step": 530
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8109288498095296,
      "learning_rate": 8.968919402478076e-06,
      "loss": 0.2976,
      "step": 531
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9052767034798376,
      "learning_rate": 8.964612014474512e-06,
      "loss": 0.3033,
      "step": 532
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8435566575506597,
      "learning_rate": 8.960296686947846e-06,
      "loss": 0.295,
      "step": 533
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8718439527389112,
      "learning_rate": 8.955973428539943e-06,
      "loss": 0.2778,
      "step": 534
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8079180788866078,
      "learning_rate": 8.95164224790855e-06,
      "loss": 0.3054,
      "step": 535
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8393182214021081,
      "learning_rate": 8.947303153727283e-06,
      "loss": 0.2991,
      "step": 536
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.867678763690361,
      "learning_rate": 8.942956154685596e-06,
      "loss": 0.2812,
      "step": 537
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8345194951500529,
      "learning_rate": 8.938601259488784e-06,
      "loss": 0.3008,
      "step": 538
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8274542051466486,
      "learning_rate": 8.93423847685795e-06,
      "loss": 0.3076,
      "step": 539
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8360050490638713,
      "learning_rate": 8.929867815529992e-06,
      "loss": 0.3123,
      "step": 540
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9412519719142663,
      "learning_rate": 8.925489284257587e-06,
      "loss": 0.3245,
      "step": 541
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9697679514515116,
      "learning_rate": 8.921102891809171e-06,
      "loss": 0.3275,
      "step": 542
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8818878506674674,
      "learning_rate": 8.916708646968924e-06,
      "loss": 0.2712,
      "step": 543
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8218749869869766,
      "learning_rate": 8.91230655853675e-06,
      "loss": 0.3211,
      "step": 544
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8950835714445418,
      "learning_rate": 8.907896635328262e-06,
      "loss": 0.3057,
      "step": 545
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8984066580507666,
      "learning_rate": 8.903478886174763e-06,
      "loss": 0.294,
      "step": 546
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8631232004224514,
      "learning_rate": 8.899053319923228e-06,
      "loss": 0.2958,
      "step": 547
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7875900991252627,
      "learning_rate": 8.894619945436285e-06,
      "loss": 0.3161,
      "step": 548
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8591166232588537,
      "learning_rate": 8.890178771592198e-06,
      "loss": 0.3003,
      "step": 549
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8894619799550535,
      "learning_rate": 8.885729807284855e-06,
      "loss": 0.3244,
      "step": 550
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8401340163420987,
      "learning_rate": 8.881273061423741e-06,
      "loss": 0.3059,
      "step": 551
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8350505118028848,
      "learning_rate": 8.876808542933924e-06,
      "loss": 0.2753,
      "step": 552
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8589357768680091,
      "learning_rate": 8.87233626075604e-06,
      "loss": 0.2964,
      "step": 553
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8523378986009381,
      "learning_rate": 8.86785622384627e-06,
      "loss": 0.3128,
      "step": 554
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8494613347517852,
      "learning_rate": 8.863368441176326e-06,
      "loss": 0.3146,
      "step": 555
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8973104795516421,
      "learning_rate": 8.858872921733433e-06,
      "loss": 0.2956,
      "step": 556
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8050742113908046,
      "learning_rate": 8.854369674520306e-06,
      "loss": 0.3205,
      "step": 557
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8764563541353061,
      "learning_rate": 8.84985870855514e-06,
      "loss": 0.3053,
      "step": 558
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9012654811760539,
      "learning_rate": 8.845340032871584e-06,
      "loss": 0.3163,
      "step": 559
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8912271816476123,
      "learning_rate": 8.840813656518728e-06,
      "loss": 0.3242,
      "step": 560
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9050188411460058,
      "learning_rate": 8.836279588561084e-06,
      "loss": 0.3053,
      "step": 561
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8759855570231105,
      "learning_rate": 8.831737838078562e-06,
      "loss": 0.3093,
      "step": 562
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8898651193442678,
      "learning_rate": 8.827188414166468e-06,
      "loss": 0.339,
      "step": 563
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8276197929929642,
      "learning_rate": 8.822631325935463e-06,
      "loss": 0.3199,
      "step": 564
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9421968316357671,
      "learning_rate": 8.818066582511564e-06,
      "loss": 0.3172,
      "step": 565
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8626122903929899,
      "learning_rate": 8.813494193036115e-06,
      "loss": 0.3186,
      "step": 566
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8504558556084155,
      "learning_rate": 8.808914166665773e-06,
      "loss": 0.3419,
      "step": 567
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8334718951263743,
      "learning_rate": 8.80432651257249e-06,
      "loss": 0.3001,
      "step": 568
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.890776263573779,
      "learning_rate": 8.799731239943488e-06,
      "loss": 0.3178,
      "step": 569
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8989898976773554,
      "learning_rate": 8.795128357981253e-06,
      "loss": 0.2981,
      "step": 570
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8053169735014596,
      "learning_rate": 8.790517875903506e-06,
      "loss": 0.294,
      "step": 571
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8059017408986384,
      "learning_rate": 8.785899802943186e-06,
      "loss": 0.3056,
      "step": 572
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8603200586438344,
      "learning_rate": 8.781274148348438e-06,
      "loss": 0.3501,
      "step": 573
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.768055946741158,
      "learning_rate": 8.776640921382585e-06,
      "loss": 0.286,
      "step": 574
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8372962929884067,
      "learning_rate": 8.772000131324117e-06,
      "loss": 0.3161,
      "step": 575
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8447415599344804,
      "learning_rate": 8.767351787466674e-06,
      "loss": 0.3214,
      "step": 576
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8536274039323959,
      "learning_rate": 8.762695899119012e-06,
      "loss": 0.309,
      "step": 577
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8531940538666637,
      "learning_rate": 8.75803247560501e-06,
      "loss": 0.31,
      "step": 578
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8662946524507412,
      "learning_rate": 8.753361526263622e-06,
      "loss": 0.3084,
      "step": 579
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7922940962680398,
      "learning_rate": 8.748683060448886e-06,
      "loss": 0.2847,
      "step": 580
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7852429201736567,
      "learning_rate": 8.743997087529886e-06,
      "loss": 0.2857,
      "step": 581
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7877383004155214,
      "learning_rate": 8.739303616890739e-06,
      "loss": 0.3134,
      "step": 582
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8252443261770144,
      "learning_rate": 8.734602657930582e-06,
      "loss": 0.3141,
      "step": 583
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8491166378006001,
      "learning_rate": 8.729894220063542e-06,
      "loss": 0.3121,
      "step": 584
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9289222047326974,
      "learning_rate": 8.725178312718727e-06,
      "loss": 0.3379,
      "step": 585
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8473475575702363,
      "learning_rate": 8.720454945340202e-06,
      "loss": 0.2958,
      "step": 586
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9050673316444299,
      "learning_rate": 8.715724127386971e-06,
      "loss": 0.3161,
      "step": 587
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8154046735157956,
      "learning_rate": 8.710985868332964e-06,
      "loss": 0.2958,
      "step": 588
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8566215742604768,
      "learning_rate": 8.706240177667003e-06,
      "loss": 0.2941,
      "step": 589
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.946830121192962,
      "learning_rate": 8.701487064892797e-06,
      "loss": 0.3135,
      "step": 590
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8929525627112201,
      "learning_rate": 8.696726539528924e-06,
      "loss": 0.3045,
      "step": 591
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8366518332547225,
      "learning_rate": 8.691958611108795e-06,
      "loss": 0.2949,
      "step": 592
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7913726028660641,
      "learning_rate": 8.687183289180659e-06,
      "loss": 0.2914,
      "step": 593
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.811048696270156,
      "learning_rate": 8.682400583307562e-06,
      "loss": 0.2906,
      "step": 594
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7907687160866751,
      "learning_rate": 8.67761050306734e-06,
      "loss": 0.2946,
      "step": 595
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9186645435646583,
      "learning_rate": 8.672813058052596e-06,
      "loss": 0.327,
      "step": 596
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8012929568134569,
      "learning_rate": 8.668008257870684e-06,
      "loss": 0.28,
      "step": 597
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8542961148094255,
      "learning_rate": 8.663196112143685e-06,
      "loss": 0.3135,
      "step": 598
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.006158101326816,
      "learning_rate": 8.658376630508391e-06,
      "loss": 0.3425,
      "step": 599
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8047329071672816,
      "learning_rate": 8.65354982261629e-06,
      "loss": 0.2813,
      "step": 600
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8465068764844832,
      "learning_rate": 8.648715698133529e-06,
      "loss": 0.2841,
      "step": 601
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7888736714933827,
      "learning_rate": 8.64387426674092e-06,
      "loss": 0.2882,
      "step": 602
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8568662562882384,
      "learning_rate": 8.639025538133899e-06,
      "loss": 0.3112,
      "step": 603
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8587024917794117,
      "learning_rate": 8.634169522022522e-06,
      "loss": 0.3041,
      "step": 604
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7800991233549786,
      "learning_rate": 8.629306228131435e-06,
      "loss": 0.2727,
      "step": 605
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8737794874611873,
      "learning_rate": 8.62443566619986e-06,
      "loss": 0.3237,
      "step": 606
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9175629776969463,
      "learning_rate": 8.619557845981571e-06,
      "loss": 0.3169,
      "step": 607
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8128855849766639,
      "learning_rate": 8.614672777244885e-06,
      "loss": 0.3116,
      "step": 608
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8066384372369907,
      "learning_rate": 8.609780469772623e-06,
      "loss": 0.2962,
      "step": 609
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8484013335491392,
      "learning_rate": 8.604880933362113e-06,
      "loss": 0.3267,
      "step": 610
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8388958660138932,
      "learning_rate": 8.599974177825155e-06,
      "loss": 0.285,
      "step": 611
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7860495961058528,
      "learning_rate": 8.595060212988006e-06,
      "loss": 0.2822,
      "step": 612
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8276012340886084,
      "learning_rate": 8.590139048691361e-06,
      "loss": 0.2996,
      "step": 613
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7764812584863781,
      "learning_rate": 8.585210694790333e-06,
      "loss": 0.2819,
      "step": 614
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7607112854990693,
      "learning_rate": 8.580275161154432e-06,
      "loss": 0.2893,
      "step": 615
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7979951678453457,
      "learning_rate": 8.575332457667546e-06,
      "loss": 0.2831,
      "step": 616
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7887780763673536,
      "learning_rate": 8.570382594227923e-06,
      "loss": 0.2845,
      "step": 617
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7725595968740181,
      "learning_rate": 8.56542558074815e-06,
      "loss": 0.2594,
      "step": 618
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8398130909443955,
      "learning_rate": 8.56046142715513e-06,
      "loss": 0.2899,
      "step": 619
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7512449946334305,
      "learning_rate": 8.555490143390061e-06,
      "loss": 0.2762,
      "step": 620
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9027530903129523,
      "learning_rate": 8.550511739408428e-06,
      "loss": 0.311,
      "step": 621
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9145886866717408,
      "learning_rate": 8.545526225179976e-06,
      "loss": 0.3324,
      "step": 622
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7767903534667611,
      "learning_rate": 8.540533610688679e-06,
      "loss": 0.2921,
      "step": 623
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8185066976205962,
      "learning_rate": 8.535533905932739e-06,
      "loss": 0.3154,
      "step": 624
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8341146865308158,
      "learning_rate": 8.53052712092455e-06,
      "loss": 0.3051,
      "step": 625
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9125699484031415,
      "learning_rate": 8.52551326569069e-06,
      "loss": 0.3017,
      "step": 626
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8249695562862555,
      "learning_rate": 8.520492350271895e-06,
      "loss": 0.3286,
      "step": 627
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8436100809695265,
      "learning_rate": 8.51546438472304e-06,
      "loss": 0.2915,
      "step": 628
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7705893355360133,
      "learning_rate": 8.510429379113114e-06,
      "loss": 0.301,
      "step": 629
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7953145267483371,
      "learning_rate": 8.50538734352521e-06,
      "loss": 0.2882,
      "step": 630
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7850541012170598,
      "learning_rate": 8.500338288056496e-06,
      "loss": 0.3133,
      "step": 631
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8282372529102404,
      "learning_rate": 8.495282222818202e-06,
      "loss": 0.3096,
      "step": 632
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8433293237976868,
      "learning_rate": 8.490219157935589e-06,
      "loss": 0.3207,
      "step": 633
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7835065958095416,
      "learning_rate": 8.485149103547943e-06,
      "loss": 0.316,
      "step": 634
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8009895244225264,
      "learning_rate": 8.480072069808541e-06,
      "loss": 0.2874,
      "step": 635
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8275586950923594,
      "learning_rate": 8.47498806688464e-06,
      "loss": 0.3139,
      "step": 636
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8122427423865404,
      "learning_rate": 8.469897104957454e-06,
      "loss": 0.301,
      "step": 637
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8774920319509736,
      "learning_rate": 8.464799194222129e-06,
      "loss": 0.3292,
      "step": 638
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8716853364843633,
      "learning_rate": 8.459694344887732e-06,
      "loss": 0.3286,
      "step": 639
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8486382683871918,
      "learning_rate": 8.454582567177223e-06,
      "loss": 0.3193,
      "step": 640
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7891851898838985,
      "learning_rate": 8.449463871327434e-06,
      "loss": 0.3014,
      "step": 641
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8187050956226559,
      "learning_rate": 8.444338267589057e-06,
      "loss": 0.2905,
      "step": 642
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.788570001181129,
      "learning_rate": 8.439205766226613e-06,
      "loss": 0.2938,
      "step": 643
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8261409867515237,
      "learning_rate": 8.434066377518437e-06,
      "loss": 0.3055,
      "step": 644
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7900670608257222,
      "learning_rate": 8.428920111756658e-06,
      "loss": 0.3004,
      "step": 645
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8672153837097204,
      "learning_rate": 8.423766979247176e-06,
      "loss": 0.2851,
      "step": 646
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8244074923670143,
      "learning_rate": 8.418606990309645e-06,
      "loss": 0.3019,
      "step": 647
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7526680732541623,
      "learning_rate": 8.413440155277444e-06,
      "loss": 0.2952,
      "step": 648
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8259402940318475,
      "learning_rate": 8.408266484497664e-06,
      "loss": 0.2954,
      "step": 649
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8777715958804737,
      "learning_rate": 8.403085988331092e-06,
      "loss": 0.2912,
      "step": 650
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.792386055354803,
      "learning_rate": 8.397898677152173e-06,
      "loss": 0.2968,
      "step": 651
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7710904853424877,
      "learning_rate": 8.392704561349006e-06,
      "loss": 0.2848,
      "step": 652
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8182030036394279,
      "learning_rate": 8.387503651323317e-06,
      "loss": 0.321,
      "step": 653
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7888800281096944,
      "learning_rate": 8.382295957490435e-06,
      "loss": 0.2742,
      "step": 654
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8120585400558087,
      "learning_rate": 8.377081490279278e-06,
      "loss": 0.3094,
      "step": 655
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8408988231610814,
      "learning_rate": 8.371860260132323e-06,
      "loss": 0.3215,
      "step": 656
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.863184461907884,
      "learning_rate": 8.366632277505598e-06,
      "loss": 0.3268,
      "step": 657
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9802528591908519,
      "learning_rate": 8.361397552868645e-06,
      "loss": 0.316,
      "step": 658
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.821204338072159,
      "learning_rate": 8.356156096704516e-06,
      "loss": 0.3024,
      "step": 659
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.801029714848551,
      "learning_rate": 8.350907919509734e-06,
      "loss": 0.2889,
      "step": 660
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8298279350286875,
      "learning_rate": 8.345653031794292e-06,
      "loss": 0.2865,
      "step": 661
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7695219701636958,
      "learning_rate": 8.340391444081615e-06,
      "loss": 0.2855,
      "step": 662
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8489687103352507,
      "learning_rate": 8.335123166908544e-06,
      "loss": 0.3,
      "step": 663
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9162243578308146,
      "learning_rate": 8.329848210825322e-06,
      "loss": 0.3252,
      "step": 664
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8493189274547335,
      "learning_rate": 8.324566586395563e-06,
      "loss": 0.3038,
      "step": 665
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8056394007714314,
      "learning_rate": 8.319278304196237e-06,
      "loss": 0.2842,
      "step": 666
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.808994633080053,
      "learning_rate": 8.313983374817643e-06,
      "loss": 0.2953,
      "step": 667
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.779178095741086,
      "learning_rate": 8.3086818088634e-06,
      "loss": 0.3096,
      "step": 668
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8217280990658212,
      "learning_rate": 8.303373616950408e-06,
      "loss": 0.269,
      "step": 669
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.90568726604773,
      "learning_rate": 8.298058809708841e-06,
      "loss": 0.3376,
      "step": 670
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8512066723106371,
      "learning_rate": 8.29273739778212e-06,
      "loss": 0.328,
      "step": 671
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.848501033171717,
      "learning_rate": 8.287409391826895e-06,
      "loss": 0.2943,
      "step": 672
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7950956801134167,
      "learning_rate": 8.282074802513016e-06,
      "loss": 0.2768,
      "step": 673
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8450211410983335,
      "learning_rate": 8.27673364052352e-06,
      "loss": 0.3101,
      "step": 674
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7744437814145162,
      "learning_rate": 8.271385916554605e-06,
      "loss": 0.2816,
      "step": 675
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.840907203964868,
      "learning_rate": 8.266031641315612e-06,
      "loss": 0.3015,
      "step": 676
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.77645316561974,
      "learning_rate": 8.260670825529002e-06,
      "loss": 0.2889,
      "step": 677
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8588881985298106,
      "learning_rate": 8.255303479930332e-06,
      "loss": 0.2898,
      "step": 678
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9105751029355986,
      "learning_rate": 8.249929615268234e-06,
      "loss": 0.3358,
      "step": 679
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7986159562507521,
      "learning_rate": 8.2445492423044e-06,
      "loss": 0.3005,
      "step": 680
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8803594105957724,
      "learning_rate": 8.239162371813552e-06,
      "loss": 0.2914,
      "step": 681
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9142724389745248,
      "learning_rate": 8.233769014583423e-06,
      "loss": 0.3087,
      "step": 682
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7909796158308176,
      "learning_rate": 8.22836918141474e-06,
      "loss": 0.2795,
      "step": 683
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7986808265671965,
      "learning_rate": 8.222962883121196e-06,
      "loss": 0.2884,
      "step": 684
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8654808566764097,
      "learning_rate": 8.217550130529433e-06,
      "loss": 0.3394,
      "step": 685
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7861170051542049,
      "learning_rate": 8.212130934479015e-06,
      "loss": 0.2778,
      "step": 686
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.764095410872562,
      "learning_rate": 8.206705305822414e-06,
      "loss": 0.3023,
      "step": 687
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7987209867031574,
      "learning_rate": 8.201273255424979e-06,
      "loss": 0.322,
      "step": 688
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8719605905672777,
      "learning_rate": 8.195834794164925e-06,
      "loss": 0.3027,
      "step": 689
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8104033886118766,
      "learning_rate": 8.190389932933301e-06,
      "loss": 0.3089,
      "step": 690
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8633193530592153,
      "learning_rate": 8.184938682633973e-06,
      "loss": 0.3159,
      "step": 691
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7727874321988651,
      "learning_rate": 8.179481054183605e-06,
      "loss": 0.2893,
      "step": 692
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7983605835741124,
      "learning_rate": 8.17401705851163e-06,
      "loss": 0.3483,
      "step": 693
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7970561188276698,
      "learning_rate": 8.168546706560231e-06,
      "loss": 0.2831,
      "step": 694
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7968171896353036,
      "learning_rate": 8.163070009284326e-06,
      "loss": 0.3097,
      "step": 695
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8509686431100105,
      "learning_rate": 8.157586977651534e-06,
      "loss": 0.3172,
      "step": 696
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8344036744644164,
      "learning_rate": 8.152097622642167e-06,
      "loss": 0.298,
      "step": 697
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8703049030129351,
      "learning_rate": 8.146601955249187e-06,
      "loss": 0.2991,
      "step": 698
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7849422791247518,
      "learning_rate": 8.141099986478212e-06,
      "loss": 0.291,
      "step": 699
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8430524323895718,
      "learning_rate": 8.135591727347469e-06,
      "loss": 0.3257,
      "step": 700
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7822834769836519,
      "learning_rate": 8.130077188887786e-06,
      "loss": 0.3008,
      "step": 701
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8607358083540139,
      "learning_rate": 8.124556382142566e-06,
      "loss": 0.2989,
      "step": 702
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7577135594796796,
      "learning_rate": 8.119029318167763e-06,
      "loss": 0.2667,
      "step": 703
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8152972554194636,
      "learning_rate": 8.113496008031863e-06,
      "loss": 0.3065,
      "step": 704
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7529408872055527,
      "learning_rate": 8.107956462815862e-06,
      "loss": 0.2949,
      "step": 705
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8412148586088325,
      "learning_rate": 8.102410693613239e-06,
      "loss": 0.2661,
      "step": 706
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8735160193672382,
      "learning_rate": 8.096858711529938e-06,
      "loss": 0.317,
      "step": 707
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7878222208254866,
      "learning_rate": 8.09130052768435e-06,
      "loss": 0.2959,
      "step": 708
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7820946173885982,
      "learning_rate": 8.085736153207277e-06,
      "loss": 0.2895,
      "step": 709
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8220493842780991,
      "learning_rate": 8.080165599241924e-06,
      "loss": 0.3051,
      "step": 710
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.856207130636198,
      "learning_rate": 8.074588876943872e-06,
      "loss": 0.3,
      "step": 711
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8593067107348992,
      "learning_rate": 8.069005997481053e-06,
      "loss": 0.2986,
      "step": 712
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.812195032991693,
      "learning_rate": 8.063416972033725e-06,
      "loss": 0.2792,
      "step": 713
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8485504876610058,
      "learning_rate": 8.057821811794457e-06,
      "loss": 0.2884,
      "step": 714
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8331444073699428,
      "learning_rate": 8.05222052796811e-06,
      "loss": 0.2864,
      "step": 715
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8324128475082347,
      "learning_rate": 8.046613131771798e-06,
      "loss": 0.3087,
      "step": 716
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7465185431377486,
      "learning_rate": 8.040999634434883e-06,
      "loss": 0.282,
      "step": 717
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8746592967120925,
      "learning_rate": 8.03538004719894e-06,
      "loss": 0.309,
      "step": 718
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8318988586921525,
      "learning_rate": 8.029754381317741e-06,
      "loss": 0.311,
      "step": 719
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8054544228823451,
      "learning_rate": 8.024122648057234e-06,
      "loss": 0.2888,
      "step": 720
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.845312336304049,
      "learning_rate": 8.018484858695514e-06,
      "loss": 0.3001,
      "step": 721
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7524823682859203,
      "learning_rate": 8.012841024522808e-06,
      "loss": 0.266,
      "step": 722
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.823655106444496,
      "learning_rate": 8.00719115684144e-06,
      "loss": 0.2869,
      "step": 723
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8993242140326638,
      "learning_rate": 8.001535266965829e-06,
      "loss": 0.2933,
      "step": 724
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7953566280454175,
      "learning_rate": 7.995873366222442e-06,
      "loss": 0.2748,
      "step": 725
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8271920284565865,
      "learning_rate": 7.990205465949792e-06,
      "loss": 0.3,
      "step": 726
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.814326243372176,
      "learning_rate": 7.9845315774984e-06,
      "loss": 0.2921,
      "step": 727
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7579982248334637,
      "learning_rate": 7.978851712230786e-06,
      "loss": 0.3075,
      "step": 728
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8044184060624087,
      "learning_rate": 7.973165881521435e-06,
      "loss": 0.3095,
      "step": 729
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.769643106909649,
      "learning_rate": 7.967474096756775e-06,
      "loss": 0.2788,
      "step": 730
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8248985183596695,
      "learning_rate": 7.961776369335164e-06,
      "loss": 0.287,
      "step": 731
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.819210256500425,
      "learning_rate": 7.95607271066686e-06,
      "loss": 0.306,
      "step": 732
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8017078802726885,
      "learning_rate": 7.950363132173992e-06,
      "loss": 0.2969,
      "step": 733
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.84527053781287,
      "learning_rate": 7.944647645290555e-06,
      "loss": 0.3045,
      "step": 734
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7982376140430907,
      "learning_rate": 7.938926261462366e-06,
      "loss": 0.2901,
      "step": 735
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7657998409347052,
      "learning_rate": 7.933198992147057e-06,
      "loss": 0.251,
      "step": 736
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8008237807458931,
      "learning_rate": 7.927465848814045e-06,
      "loss": 0.2947,
      "step": 737
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7836203469437216,
      "learning_rate": 7.921726842944507e-06,
      "loss": 0.2819,
      "step": 738
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.806581450560192,
      "learning_rate": 7.915981986031367e-06,
      "loss": 0.3149,
      "step": 739
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8715307404539555,
      "learning_rate": 7.910231289579263e-06,
      "loss": 0.3254,
      "step": 740
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8324025651873598,
      "learning_rate": 7.90447476510452e-06,
      "loss": 0.2828,
      "step": 741
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8281055201390866,
      "learning_rate": 7.898712424135149e-06,
      "loss": 0.3218,
      "step": 742
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7863377392447002,
      "learning_rate": 7.892944278210796e-06,
      "loss": 0.2736,
      "step": 743
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7960539753232007,
      "learning_rate": 7.887170338882742e-06,
      "loss": 0.3046,
      "step": 744
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8382828746702077,
      "learning_rate": 7.88139061771386e-06,
      "loss": 0.2951,
      "step": 745
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7552275680892166,
      "learning_rate": 7.875605126278608e-06,
      "loss": 0.2848,
      "step": 746
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.775880139252017,
      "learning_rate": 7.869813876162997e-06,
      "loss": 0.2879,
      "step": 747
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8749397353253353,
      "learning_rate": 7.864016878964575e-06,
      "loss": 0.2904,
      "step": 748
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9437448755561537,
      "learning_rate": 7.858214146292394e-06,
      "loss": 0.3127,
      "step": 749
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7818325000503746,
      "learning_rate": 7.852405689766993e-06,
      "loss": 0.3072,
      "step": 750
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7676172653620015,
      "learning_rate": 7.846591521020373e-06,
      "loss": 0.2723,
      "step": 751
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8472949949670019,
      "learning_rate": 7.840771651695976e-06,
      "loss": 0.295,
      "step": 752
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8173646604183937,
      "learning_rate": 7.834946093448658e-06,
      "loss": 0.2859,
      "step": 753
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8870881426934443,
      "learning_rate": 7.829114857944672e-06,
      "loss": 0.3026,
      "step": 754
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8755992883212927,
      "learning_rate": 7.823277956861635e-06,
      "loss": 0.29,
      "step": 755
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8675431683401105,
      "learning_rate": 7.817435401888514e-06,
      "loss": 0.3047,
      "step": 756
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8811705899893023,
      "learning_rate": 7.811587204725593e-06,
      "loss": 0.3238,
      "step": 757
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8439162999681662,
      "learning_rate": 7.80573337708446e-06,
      "loss": 0.2972,
      "step": 758
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.868304944475457,
      "learning_rate": 7.799873930687979e-06,
      "loss": 0.2933,
      "step": 759
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8150063157227956,
      "learning_rate": 7.79400887727026e-06,
      "loss": 0.31,
      "step": 760
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8366467718289208,
      "learning_rate": 7.78813822857665e-06,
      "loss": 0.309,
      "step": 761
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8601108252991556,
      "learning_rate": 7.782261996363694e-06,
      "loss": 0.2913,
      "step": 762
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8361849835786183,
      "learning_rate": 7.776380192399122e-06,
      "loss": 0.3401,
      "step": 763
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8928054848592836,
      "learning_rate": 7.770492828461824e-06,
      "loss": 0.3291,
      "step": 764
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7396633463012768,
      "learning_rate": 7.764599916341817e-06,
      "loss": 0.282,
      "step": 765
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8347350298441228,
      "learning_rate": 7.75870146784024e-06,
      "loss": 0.2864,
      "step": 766
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8631920540149018,
      "learning_rate": 7.752797494769307e-06,
      "loss": 0.2908,
      "step": 767
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8133284445247652,
      "learning_rate": 7.7468880089523e-06,
      "loss": 0.3042,
      "step": 768
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8450143200641446,
      "learning_rate": 7.74097302222355e-06,
      "loss": 0.2969,
      "step": 769
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7503339430290001,
      "learning_rate": 7.73505254642839e-06,
      "loss": 0.2891,
      "step": 770
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8565836159640364,
      "learning_rate": 7.729126593423151e-06,
      "loss": 0.2928,
      "step": 771
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.810739752508949,
      "learning_rate": 7.723195175075136e-06,
      "loss": 0.3198,
      "step": 772
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8631216564816397,
      "learning_rate": 7.71725830326259e-06,
      "loss": 0.2872,
      "step": 773
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8542645445469651,
      "learning_rate": 7.711315989874677e-06,
      "loss": 0.3169,
      "step": 774
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.810684921553861,
      "learning_rate": 7.70536824681146e-06,
      "loss": 0.3132,
      "step": 775
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.816751017669758,
      "learning_rate": 7.69941508598388e-06,
      "loss": 0.2932,
      "step": 776
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8595509303582775,
      "learning_rate": 7.69345651931372e-06,
      "loss": 0.292,
      "step": 777
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7656653461913022,
      "learning_rate": 7.687492558733593e-06,
      "loss": 0.3107,
      "step": 778
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.849053948371894,
      "learning_rate": 7.681523216186912e-06,
      "loss": 0.2901,
      "step": 779
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7943239126378621,
      "learning_rate": 7.675548503627872e-06,
      "loss": 0.2821,
      "step": 780
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8571099983224592,
      "learning_rate": 7.669568433021418e-06,
      "loss": 0.3296,
      "step": 781
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8025200779656888,
      "learning_rate": 7.663583016343224e-06,
      "loss": 0.277,
      "step": 782
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8407088981263721,
      "learning_rate": 7.65759226557967e-06,
      "loss": 0.302,
      "step": 783
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8073426185315818,
      "learning_rate": 7.651596192727826e-06,
      "loss": 0.2984,
      "step": 784
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7840272274443647,
      "learning_rate": 7.64559480979541e-06,
      "loss": 0.2922,
      "step": 785
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8052947193828663,
      "learning_rate": 7.63958812880078e-06,
      "loss": 0.2788,
      "step": 786
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.857125831719273,
      "learning_rate": 7.633576161772898e-06,
      "loss": 0.2721,
      "step": 787
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7819382010926632,
      "learning_rate": 7.627558920751321e-06,
      "loss": 0.2929,
      "step": 788
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.804543547844341,
      "learning_rate": 7.621536417786159e-06,
      "loss": 0.2954,
      "step": 789
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7813260910187387,
      "learning_rate": 7.615508664938065e-06,
      "loss": 0.3088,
      "step": 790
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7309051252427843,
      "learning_rate": 7.6094756742782e-06,
      "loss": 0.2839,
      "step": 791
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7673945684714921,
      "learning_rate": 7.6034374578882205e-06,
      "loss": 0.2984,
      "step": 792
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8427536811812298,
      "learning_rate": 7.597394027860243e-06,
      "loss": 0.3119,
      "step": 793
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8541302734781536,
      "learning_rate": 7.5913453962968296e-06,
      "loss": 0.2989,
      "step": 794
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.821114775177324,
      "learning_rate": 7.585291575310952e-06,
      "loss": 0.2883,
      "step": 795
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7872607368639121,
      "learning_rate": 7.579232577025984e-06,
      "loss": 0.2964,
      "step": 796
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7761950406720638,
      "learning_rate": 7.5731684135756566e-06,
      "loss": 0.295,
      "step": 797
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8052821077867923,
      "learning_rate": 7.567099097104053e-06,
      "loss": 0.325,
      "step": 798
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8522354191195298,
      "learning_rate": 7.5610246397655715e-06,
      "loss": 0.3214,
      "step": 799
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8145796196813088,
      "learning_rate": 7.5549450537249096e-06,
      "loss": 0.3027,
      "step": 800
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8032917337458791,
      "learning_rate": 7.548860351157028e-06,
      "loss": 0.2965,
      "step": 801
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.810868535072896,
      "learning_rate": 7.542770544247141e-06,
      "loss": 0.311,
      "step": 802
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7948960432490044,
      "learning_rate": 7.536675645190682e-06,
      "loss": 0.2964,
      "step": 803
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7797765399749693,
      "learning_rate": 7.530575666193283e-06,
      "loss": 0.2937,
      "step": 804
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8136980743604897,
      "learning_rate": 7.524470619470745e-06,
      "loss": 0.3117,
      "step": 805
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8572319708003505,
      "learning_rate": 7.518360517249024e-06,
      "loss": 0.3258,
      "step": 806
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.751746890429509,
      "learning_rate": 7.512245371764197e-06,
      "loss": 0.2923,
      "step": 807
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8124388878002045,
      "learning_rate": 7.506125195262437e-06,
      "loss": 0.3163,
      "step": 808
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7858034546633371,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.3091,
      "step": 809
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8613171493095833,
      "learning_rate": 7.493869798243188e-06,
      "loss": 0.3053,
      "step": 810
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8946159380192973,
      "learning_rate": 7.487734602268326e-06,
      "loss": 0.3132,
      "step": 811
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.730802899540197,
      "learning_rate": 7.4815944243617455e-06,
      "loss": 0.2375,
      "step": 812
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7614681555391482,
      "learning_rate": 7.475449276819753e-06,
      "loss": 0.2879,
      "step": 813
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8480451627190042,
      "learning_rate": 7.469299171948608e-06,
      "loss": 0.28,
      "step": 814
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8268700748423344,
      "learning_rate": 7.463144122064494e-06,
      "loss": 0.3223,
      "step": 815
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8281156618181602,
      "learning_rate": 7.456984139493502e-06,
      "loss": 0.3055,
      "step": 816
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9074044331955596,
      "learning_rate": 7.4508192365716005e-06,
      "loss": 0.3164,
      "step": 817
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7460284175822396,
      "learning_rate": 7.444649425644606e-06,
      "loss": 0.2896,
      "step": 818
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7904538240093412,
      "learning_rate": 7.438474719068174e-06,
      "loss": 0.2757,
      "step": 819
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8082813222022442,
      "learning_rate": 7.432295129207753e-06,
      "loss": 0.3039,
      "step": 820
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7932524827394453,
      "learning_rate": 7.426110668438578e-06,
      "loss": 0.2997,
      "step": 821
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8524688168457281,
      "learning_rate": 7.419921349145635e-06,
      "loss": 0.295,
      "step": 822
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8441274805844985,
      "learning_rate": 7.413727183723642e-06,
      "loss": 0.2943,
      "step": 823
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8412813493604678,
      "learning_rate": 7.4075281845770196e-06,
      "loss": 0.3116,
      "step": 824
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7822392918597183,
      "learning_rate": 7.401324364119872e-06,
      "loss": 0.2931,
      "step": 825
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7717565499393549,
      "learning_rate": 7.395115734775953e-06,
      "loss": 0.2922,
      "step": 826
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8726958171376579,
      "learning_rate": 7.388902308978652e-06,
      "loss": 0.2898,
      "step": 827
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8446576065777153,
      "learning_rate": 7.382684099170959e-06,
      "loss": 0.3122,
      "step": 828
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7917459121844254,
      "learning_rate": 7.37646111780545e-06,
      "loss": 0.3095,
      "step": 829
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.835196112560585,
      "learning_rate": 7.370233377344252e-06,
      "loss": 0.3214,
      "step": 830
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8102393259064526,
      "learning_rate": 7.364000890259024e-06,
      "loss": 0.2796,
      "step": 831
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9508004127357791,
      "learning_rate": 7.357763669030931e-06,
      "loss": 0.3094,
      "step": 832
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8354473500467093,
      "learning_rate": 7.351521726150619e-06,
      "loss": 0.3129,
      "step": 833
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7559013149079604,
      "learning_rate": 7.3452750741181855e-06,
      "loss": 0.2947,
      "step": 834
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8605827515561323,
      "learning_rate": 7.339023725443166e-06,
      "loss": 0.3132,
      "step": 835
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8696799172664447,
      "learning_rate": 7.332767692644493e-06,
      "loss": 0.3036,
      "step": 836
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8108762854227414,
      "learning_rate": 7.326506988250488e-06,
      "loss": 0.285,
      "step": 837
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8411616039461777,
      "learning_rate": 7.320241624798821e-06,
      "loss": 0.2801,
      "step": 838
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.785611635055323,
      "learning_rate": 7.313971614836496e-06,
      "loss": 0.2894,
      "step": 839
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8419616342156286,
      "learning_rate": 7.307696970919819e-06,
      "loss": 0.3001,
      "step": 840
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8609574609850862,
      "learning_rate": 7.3014177056143795e-06,
      "loss": 0.2984,
      "step": 841
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8138301649874357,
      "learning_rate": 7.295133831495021e-06,
      "loss": 0.2981,
      "step": 842
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8219653559545645,
      "learning_rate": 7.288845361145812e-06,
      "loss": 0.2977,
      "step": 843
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.804541301007275,
      "learning_rate": 7.282552307160033e-06,
      "loss": 0.2933,
      "step": 844
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8223164520422939,
      "learning_rate": 7.276254682140137e-06,
      "loss": 0.3045,
      "step": 845
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7553225199653282,
      "learning_rate": 7.269952498697734e-06,
      "loss": 0.281,
      "step": 846
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7847478922399703,
      "learning_rate": 7.2636457694535655e-06,
      "loss": 0.2972,
      "step": 847
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.758301494619438,
      "learning_rate": 7.2573345070374675e-06,
      "loss": 0.2892,
      "step": 848
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7872172694202422,
      "learning_rate": 7.251018724088367e-06,
      "loss": 0.2977,
      "step": 849
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.803504926423266,
      "learning_rate": 7.244698433254231e-06,
      "loss": 0.2759,
      "step": 850
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8023724889638798,
      "learning_rate": 7.238373647192063e-06,
      "loss": 0.278,
      "step": 851
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8101696893982147,
      "learning_rate": 7.232044378567864e-06,
      "loss": 0.3021,
      "step": 852
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8054212308745915,
      "learning_rate": 7.225710640056614e-06,
      "loss": 0.2989,
      "step": 853
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8126684587985833,
      "learning_rate": 7.2193724443422405e-06,
      "loss": 0.2986,
      "step": 854
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7821929297236117,
      "learning_rate": 7.213029804117604e-06,
      "loss": 0.2788,
      "step": 855
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8101074962968023,
      "learning_rate": 7.206682732084458e-06,
      "loss": 0.3165,
      "step": 856
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7592712546442367,
      "learning_rate": 7.200331240953439e-06,
      "loss": 0.2898,
      "step": 857
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7890940178145244,
      "learning_rate": 7.193975343444024e-06,
      "loss": 0.3106,
      "step": 858
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8225455925338874,
      "learning_rate": 7.187615052284522e-06,
      "loss": 0.2696,
      "step": 859
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8135149432487249,
      "learning_rate": 7.181250380212037e-06,
      "loss": 0.2842,
      "step": 860
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7460941019293909,
      "learning_rate": 7.174881339972448e-06,
      "loss": 0.265,
      "step": 861
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7694160364558165,
      "learning_rate": 7.1685079443203784e-06,
      "loss": 0.2845,
      "step": 862
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8083754949502184,
      "learning_rate": 7.16213020601918e-06,
      "loss": 0.3167,
      "step": 863
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8328898564058709,
      "learning_rate": 7.155748137840892e-06,
      "loss": 0.282,
      "step": 864
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8138443413782529,
      "learning_rate": 7.149361752566233e-06,
      "loss": 0.2974,
      "step": 865
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7557300104058424,
      "learning_rate": 7.142971062984563e-06,
      "loss": 0.277,
      "step": 866
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8220204946421954,
      "learning_rate": 7.136576081893863e-06,
      "loss": 0.2889,
      "step": 867
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7906162689483117,
      "learning_rate": 7.130176822100707e-06,
      "loss": 0.2888,
      "step": 868
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.885459971233032,
      "learning_rate": 7.12377329642024e-06,
      "loss": 0.3058,
      "step": 869
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7628957413799972,
      "learning_rate": 7.117365517676146e-06,
      "loss": 0.2886,
      "step": 870
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8225167161834194,
      "learning_rate": 7.110953498700629e-06,
      "loss": 0.3037,
      "step": 871
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.77196010588948,
      "learning_rate": 7.104537252334383e-06,
      "loss": 0.2922,
      "step": 872
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7409320398811431,
      "learning_rate": 7.09811679142657e-06,
      "loss": 0.2825,
      "step": 873
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8552635026960944,
      "learning_rate": 7.09169212883479e-06,
      "loss": 0.3044,
      "step": 874
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7554853838331839,
      "learning_rate": 7.085263277425058e-06,
      "loss": 0.2799,
      "step": 875
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8267595597155777,
      "learning_rate": 7.078830250071778e-06,
      "loss": 0.2881,
      "step": 876
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7805409026090043,
      "learning_rate": 7.072393059657716e-06,
      "loss": 0.3152,
      "step": 877
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8986153219840282,
      "learning_rate": 7.065951719073973e-06,
      "loss": 0.3242,
      "step": 878
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.78889261550404,
      "learning_rate": 7.059506241219964e-06,
      "loss": 0.2877,
      "step": 879
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7477031896956824,
      "learning_rate": 7.053056639003389e-06,
      "loss": 0.2814,
      "step": 880
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7873968525239828,
      "learning_rate": 7.046602925340209e-06,
      "loss": 0.2921,
      "step": 881
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7617260544448323,
      "learning_rate": 7.0401451131546125e-06,
      "loss": 0.2784,
      "step": 882
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8061565452062496,
      "learning_rate": 7.033683215379002e-06,
      "loss": 0.3014,
      "step": 883
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7832781677908607,
      "learning_rate": 7.027217244953958e-06,
      "loss": 0.2807,
      "step": 884
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8571225289952341,
      "learning_rate": 7.020747214828221e-06,
      "loss": 0.2962,
      "step": 885
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7976428579225935,
      "learning_rate": 7.014273137958656e-06,
      "loss": 0.2903,
      "step": 886
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7920243543664983,
      "learning_rate": 7.007795027310236e-06,
      "loss": 0.3019,
      "step": 887
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7554069779905803,
      "learning_rate": 7.001312895856011e-06,
      "loss": 0.2667,
      "step": 888
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7426890467099893,
      "learning_rate": 6.994826756577082e-06,
      "loss": 0.2547,
      "step": 889
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.840658556478749,
      "learning_rate": 6.988336622462578e-06,
      "loss": 0.3024,
      "step": 890
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7875101686167535,
      "learning_rate": 6.981842506509626e-06,
      "loss": 0.2938,
      "step": 891
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.839482769368259,
      "learning_rate": 6.975344421723329e-06,
      "loss": 0.2987,
      "step": 892
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7716276826545896,
      "learning_rate": 6.968842381116736e-06,
      "loss": 0.2855,
      "step": 893
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.828310267468954,
      "learning_rate": 6.962336397710819e-06,
      "loss": 0.2971,
      "step": 894
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8013787634548576,
      "learning_rate": 6.955826484534447e-06,
      "loss": 0.2897,
      "step": 895
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7871205514215948,
      "learning_rate": 6.949312654624355e-06,
      "loss": 0.2726,
      "step": 896
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8482881621953591,
      "learning_rate": 6.942794921025127e-06,
      "loss": 0.298,
      "step": 897
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.839410317264949,
      "learning_rate": 6.93627329678916e-06,
      "loss": 0.3006,
      "step": 898
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7560673952642657,
      "learning_rate": 6.9297477949766445e-06,
      "loss": 0.2855,
      "step": 899
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8057811958104958,
      "learning_rate": 6.9232184286555345e-06,
      "loss": 0.2792,
      "step": 900
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7954287126743045,
      "learning_rate": 6.916685210901527e-06,
      "loss": 0.3138,
      "step": 901
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8124494264570531,
      "learning_rate": 6.910148154798026e-06,
      "loss": 0.3137,
      "step": 902
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.776591741754379,
      "learning_rate": 6.903607273436128e-06,
      "loss": 0.3232,
      "step": 903
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7642555777827827,
      "learning_rate": 6.897062579914587e-06,
      "loss": 0.269,
      "step": 904
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8217776480759984,
      "learning_rate": 6.890514087339789e-06,
      "loss": 0.2873,
      "step": 905
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7775566185341836,
      "learning_rate": 6.883961808825732e-06,
      "loss": 0.29,
      "step": 906
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7697052175003833,
      "learning_rate": 6.8774057574939956e-06,
      "loss": 0.2934,
      "step": 907
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7609529169612607,
      "learning_rate": 6.870845946473711e-06,
      "loss": 0.2566,
      "step": 908
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8014677179100446,
      "learning_rate": 6.864282388901544e-06,
      "loss": 0.2589,
      "step": 909
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7915524053631409,
      "learning_rate": 6.857715097921657e-06,
      "loss": 0.2977,
      "step": 910
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7753054706231047,
      "learning_rate": 6.851144086685695e-06,
      "loss": 0.2946,
      "step": 911
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7720753114149705,
      "learning_rate": 6.844569368352748e-06,
      "loss": 0.2989,
      "step": 912
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7630362235281248,
      "learning_rate": 6.837990956089334e-06,
      "loss": 0.2885,
      "step": 913
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8429618140488413,
      "learning_rate": 6.831408863069364e-06,
      "loss": 0.3034,
      "step": 914
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7819956182825385,
      "learning_rate": 6.824823102474127e-06,
      "loss": 0.2821,
      "step": 915
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8009383061872573,
      "learning_rate": 6.818233687492248e-06,
      "loss": 0.307,
      "step": 916
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8015705204703861,
      "learning_rate": 6.811640631319677e-06,
      "loss": 0.3018,
      "step": 917
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7376970443972016,
      "learning_rate": 6.805043947159652e-06,
      "loss": 0.2887,
      "step": 918
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8192233196591263,
      "learning_rate": 6.79844364822268e-06,
      "loss": 0.2979,
      "step": 919
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.74315147548249,
      "learning_rate": 6.7918397477265e-06,
      "loss": 0.3064,
      "step": 920
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7881541391885561,
      "learning_rate": 6.785232258896078e-06,
      "loss": 0.2877,
      "step": 921
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7656367369961653,
      "learning_rate": 6.778621194963545e-06,
      "loss": 0.2724,
      "step": 922
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7374167189055814,
      "learning_rate": 6.77200656916821e-06,
      "loss": 0.2754,
      "step": 923
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7315824873793291,
      "learning_rate": 6.765388394756504e-06,
      "loss": 0.2791,
      "step": 924
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7660765496506612,
      "learning_rate": 6.758766684981972e-06,
      "loss": 0.2932,
      "step": 925
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8248546613116938,
      "learning_rate": 6.752141453105231e-06,
      "loss": 0.2892,
      "step": 926
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7418383712001643,
      "learning_rate": 6.745512712393958e-06,
      "loss": 0.2774,
      "step": 927
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8749600449466967,
      "learning_rate": 6.7388804761228535e-06,
      "loss": 0.3258,
      "step": 928
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8439163291102914,
      "learning_rate": 6.732244757573619e-06,
      "loss": 0.2844,
      "step": 929
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8482709544859836,
      "learning_rate": 6.725605570034928e-06,
      "loss": 0.3182,
      "step": 930
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7996473221793843,
      "learning_rate": 6.718962926802407e-06,
      "loss": 0.2911,
      "step": 931
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8099431880496363,
      "learning_rate": 6.712316841178592e-06,
      "loss": 0.2944,
      "step": 932
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8532315364398908,
      "learning_rate": 6.705667326472926e-06,
      "loss": 0.2909,
      "step": 933
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.861181977445455,
      "learning_rate": 6.699014396001707e-06,
      "loss": 0.2948,
      "step": 934
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.77079241149733,
      "learning_rate": 6.69235806308808e-06,
      "loss": 0.2792,
      "step": 935
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8750464110366833,
      "learning_rate": 6.685698341062003e-06,
      "loss": 0.291,
      "step": 936
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8020733214196163,
      "learning_rate": 6.67903524326022e-06,
      "loss": 0.2952,
      "step": 937
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8345156774687199,
      "learning_rate": 6.672368783026235e-06,
      "loss": 0.2704,
      "step": 938
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.857052482510153,
      "learning_rate": 6.665698973710289e-06,
      "loss": 0.2881,
      "step": 939
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8290273989111514,
      "learning_rate": 6.659025828669323e-06,
      "loss": 0.2882,
      "step": 940
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8773192497576445,
      "learning_rate": 6.652349361266967e-06,
      "loss": 0.3078,
      "step": 941
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7878153354816635,
      "learning_rate": 6.645669584873495e-06,
      "loss": 0.2797,
      "step": 942
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8297937087700786,
      "learning_rate": 6.638986512865814e-06,
      "loss": 0.2818,
      "step": 943
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7923672191615014,
      "learning_rate": 6.632300158627427e-06,
      "loss": 0.2816,
      "step": 944
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.735505530371253,
      "learning_rate": 6.625610535548418e-06,
      "loss": 0.2683,
      "step": 945
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7874616603932696,
      "learning_rate": 6.6189176570254045e-06,
      "loss": 0.2598,
      "step": 946
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7648970531528292,
      "learning_rate": 6.612221536461534e-06,
      "loss": 0.282,
      "step": 947
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7910307387428833,
      "learning_rate": 6.605522187266441e-06,
      "loss": 0.2801,
      "step": 948
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8048440397323523,
      "learning_rate": 6.598819622856227e-06,
      "loss": 0.2968,
      "step": 949
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8074636491368827,
      "learning_rate": 6.592113856653433e-06,
      "loss": 0.2915,
      "step": 950
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8347022696306284,
      "learning_rate": 6.585404902087011e-06,
      "loss": 0.3028,
      "step": 951
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7848991831193056,
      "learning_rate": 6.578692772592298e-06,
      "loss": 0.2865,
      "step": 952
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7865375175565809,
      "learning_rate": 6.571977481610989e-06,
      "loss": 0.3,
      "step": 953
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.762741962854006,
      "learning_rate": 6.565259042591112e-06,
      "loss": 0.2718,
      "step": 954
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7534604575812777,
      "learning_rate": 6.558537468986998e-06,
      "loss": 0.2987,
      "step": 955
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8250158865972146,
      "learning_rate": 6.551812774259253e-06,
      "loss": 0.3099,
      "step": 956
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7290946517048582,
      "learning_rate": 6.545084971874738e-06,
      "loss": 0.2731,
      "step": 957
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8273599463378737,
      "learning_rate": 6.538354075306532e-06,
      "loss": 0.2947,
      "step": 958
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8462127218768327,
      "learning_rate": 6.531620098033919e-06,
      "loss": 0.2972,
      "step": 959
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8791760535132391,
      "learning_rate": 6.524883053542339e-06,
      "loss": 0.2955,
      "step": 960
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7680582746348493,
      "learning_rate": 6.518142955323389e-06,
      "loss": 0.27,
      "step": 961
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8408954081180605,
      "learning_rate": 6.511399816874769e-06,
      "loss": 0.2986,
      "step": 962
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7685386229299469,
      "learning_rate": 6.504653651700278e-06,
      "loss": 0.3075,
      "step": 963
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7686394827864159,
      "learning_rate": 6.497904473309766e-06,
      "loss": 0.2941,
      "step": 964
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.861662108568695,
      "learning_rate": 6.4911522952191275e-06,
      "loss": 0.2823,
      "step": 965
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7837264726483254,
      "learning_rate": 6.4843971309502554e-06,
      "loss": 0.2774,
      "step": 966
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8307003886382289,
      "learning_rate": 6.477638994031028e-06,
      "loss": 0.2885,
      "step": 967
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.776864976638076,
      "learning_rate": 6.470877897995275e-06,
      "loss": 0.3112,
      "step": 968
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7479481068157341,
      "learning_rate": 6.464113856382752e-06,
      "loss": 0.2991,
      "step": 969
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8204582000652352,
      "learning_rate": 6.457346882739114e-06,
      "loss": 0.2776,
      "step": 970
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8271652327666897,
      "learning_rate": 6.450576990615888e-06,
      "loss": 0.3013,
      "step": 971
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.869084742837244,
      "learning_rate": 6.443804193570441e-06,
      "loss": 0.2907,
      "step": 972
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7911854889177932,
      "learning_rate": 6.437028505165965e-06,
      "loss": 0.2901,
      "step": 973
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7343887903019277,
      "learning_rate": 6.430249938971438e-06,
      "loss": 0.2966,
      "step": 974
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8415264453577354,
      "learning_rate": 6.423468508561599e-06,
      "loss": 0.2963,
      "step": 975
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7964924010092707,
      "learning_rate": 6.416684227516927e-06,
      "loss": 0.2809,
      "step": 976
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.888819035323612,
      "learning_rate": 6.409897109423607e-06,
      "loss": 0.2825,
      "step": 977
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7619231734234677,
      "learning_rate": 6.403107167873509e-06,
      "loss": 0.2954,
      "step": 978
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.832173710286822,
      "learning_rate": 6.396314416464151e-06,
      "loss": 0.2819,
      "step": 979
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7757552084659244,
      "learning_rate": 6.389518868798683e-06,
      "loss": 0.2915,
      "step": 980
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8312402509535516,
      "learning_rate": 6.382720538485856e-06,
      "loss": 0.2964,
      "step": 981
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8049732529596779,
      "learning_rate": 6.375919439139988e-06,
      "loss": 0.2959,
      "step": 982
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7730739215528822,
      "learning_rate": 6.3691155843809475e-06,
      "loss": 0.3022,
      "step": 983
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8155878697544217,
      "learning_rate": 6.3623089878341146e-06,
      "loss": 0.2724,
      "step": 984
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7888968742788925,
      "learning_rate": 6.355499663130369e-06,
      "loss": 0.3045,
      "step": 985
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8095818419158981,
      "learning_rate": 6.348687623906047e-06,
      "loss": 0.2844,
      "step": 986
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8256450236780357,
      "learning_rate": 6.341872883802923e-06,
      "loss": 0.2922,
      "step": 987
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.80349272832681,
      "learning_rate": 6.33505545646818e-06,
      "loss": 0.3005,
      "step": 988
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8786220311609493,
      "learning_rate": 6.328235355554382e-06,
      "loss": 0.3136,
      "step": 989
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8514156584098225,
      "learning_rate": 6.32141259471945e-06,
      "loss": 0.2984,
      "step": 990
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8217827339032813,
      "learning_rate": 6.314587187626628e-06,
      "loss": 0.2945,
      "step": 991
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8756808112266792,
      "learning_rate": 6.307759147944461e-06,
      "loss": 0.3125,
      "step": 992
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7357246251883435,
      "learning_rate": 6.3009284893467655e-06,
      "loss": 0.2848,
      "step": 993
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7837260930270211,
      "learning_rate": 6.294095225512604e-06,
      "loss": 0.2938,
      "step": 994
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8285394389785709,
      "learning_rate": 6.287259370126255e-06,
      "loss": 0.3113,
      "step": 995
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8453328611908368,
      "learning_rate": 6.280420936877187e-06,
      "loss": 0.3017,
      "step": 996
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7733567925642197,
      "learning_rate": 6.273579939460032e-06,
      "loss": 0.2902,
      "step": 997
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.757587209830131,
      "learning_rate": 6.266736391574553e-06,
      "loss": 0.2557,
      "step": 998
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8993664464198563,
      "learning_rate": 6.259890306925627e-06,
      "loss": 0.2893,
      "step": 999
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7504825678073168,
      "learning_rate": 6.253041699223206e-06,
      "loss": 0.2807,
      "step": 1000
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7835280265435107,
      "learning_rate": 6.246190582182296e-06,
      "loss": 0.2928,
      "step": 1001
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8471492320212818,
      "learning_rate": 6.239336969522931e-06,
      "loss": 0.2817,
      "step": 1002
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8058939810532947,
      "learning_rate": 6.232480874970137e-06,
      "loss": 0.3401,
      "step": 1003
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7528498257482347,
      "learning_rate": 6.225622312253916e-06,
      "loss": 0.2692,
      "step": 1004
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8525904270603767,
      "learning_rate": 6.218761295109209e-06,
      "loss": 0.2908,
      "step": 1005
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8543511849994663,
      "learning_rate": 6.211897837275873e-06,
      "loss": 0.3203,
      "step": 1006
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8180877087503405,
      "learning_rate": 6.205031952498656e-06,
      "loss": 0.3058,
      "step": 1007
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8648541208591511,
      "learning_rate": 6.198163654527161e-06,
      "loss": 0.3067,
      "step": 1008
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7494023826842601,
      "learning_rate": 6.191292957115825e-06,
      "loss": 0.2784,
      "step": 1009
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7706360501266472,
      "learning_rate": 6.184419874023895e-06,
      "loss": 0.2744,
      "step": 1010
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7399826072781285,
      "learning_rate": 6.177544419015388e-06,
      "loss": 0.298,
      "step": 1011
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7496919150590914,
      "learning_rate": 6.170666605859077e-06,
      "loss": 0.2873,
      "step": 1012
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8283557787528354,
      "learning_rate": 6.163786448328455e-06,
      "loss": 0.2843,
      "step": 1013
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.748437778160926,
      "learning_rate": 6.156903960201709e-06,
      "loss": 0.2651,
      "step": 1014
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7943728304718941,
      "learning_rate": 6.150019155261694e-06,
      "loss": 0.3049,
      "step": 1015
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7720661722563843,
      "learning_rate": 6.1431320472959054e-06,
      "loss": 0.2801,
      "step": 1016
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.773473832773557,
      "learning_rate": 6.136242650096451e-06,
      "loss": 0.2991,
      "step": 1017
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8018292753637557,
      "learning_rate": 6.12935097746002e-06,
      "loss": 0.2896,
      "step": 1018
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8045866858172642,
      "learning_rate": 6.122457043187863e-06,
      "loss": 0.2863,
      "step": 1019
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8536608442980604,
      "learning_rate": 6.115560861085757e-06,
      "loss": 0.3087,
      "step": 1020
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.785728418296924,
      "learning_rate": 6.108662444963979e-06,
      "loss": 0.2963,
      "step": 1021
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7588293302301273,
      "learning_rate": 6.101761808637283e-06,
      "loss": 0.2804,
      "step": 1022
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7509084874784008,
      "learning_rate": 6.094858965924866e-06,
      "loss": 0.2953,
      "step": 1023
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8951433414228627,
      "learning_rate": 6.087953930650349e-06,
      "loss": 0.2898,
      "step": 1024
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8554604180914017,
      "learning_rate": 6.081046716641735e-06,
      "loss": 0.3088,
      "step": 1025
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7549040525554626,
      "learning_rate": 6.0741373377314005e-06,
      "loss": 0.283,
      "step": 1026
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.783077750774575,
      "learning_rate": 6.0672258077560475e-06,
      "loss": 0.2947,
      "step": 1027
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8280729217946468,
      "learning_rate": 6.060312140556694e-06,
      "loss": 0.311,
      "step": 1028
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.822932438815351,
      "learning_rate": 6.053396349978632e-06,
      "loss": 0.2846,
      "step": 1029
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8200290800756612,
      "learning_rate": 6.046478449871409e-06,
      "loss": 0.2905,
      "step": 1030
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8012951554380421,
      "learning_rate": 6.039558454088796e-06,
      "loss": 0.3005,
      "step": 1031
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7759534044761558,
      "learning_rate": 6.032636376488763e-06,
      "loss": 0.3067,
      "step": 1032
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7779347988765138,
      "learning_rate": 6.025712230933443e-06,
      "loss": 0.3063,
      "step": 1033
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7835291392931276,
      "learning_rate": 6.018786031289119e-06,
      "loss": 0.2933,
      "step": 1034
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7412453852624011,
      "learning_rate": 6.0118577914261786e-06,
      "loss": 0.282,
      "step": 1035
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7091276744933369,
      "learning_rate": 6.004927525219103e-06,
      "loss": 0.2756,
      "step": 1036
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7962419557759635,
      "learning_rate": 5.997995246546424e-06,
      "loss": 0.282,
      "step": 1037
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7835383203839851,
      "learning_rate": 5.9910609692907105e-06,
      "loss": 0.2847,
      "step": 1038
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7640861962360703,
      "learning_rate": 5.984124707338528e-06,
      "loss": 0.282,
      "step": 1039
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7570454020458116,
      "learning_rate": 5.977186474580421e-06,
      "loss": 0.2789,
      "step": 1040
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.794171972069631,
      "learning_rate": 5.970246284910877e-06,
      "loss": 0.2795,
      "step": 1041
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7495651439287019,
      "learning_rate": 5.963304152228307e-06,
      "loss": 0.2935,
      "step": 1042
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7776164719164649,
      "learning_rate": 5.9563600904350074e-06,
      "loss": 0.2988,
      "step": 1043
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7936168885987579,
      "learning_rate": 5.949414113437142e-06,
      "loss": 0.3232,
      "step": 1044
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7145854762809316,
      "learning_rate": 5.942466235144708e-06,
      "loss": 0.2521,
      "step": 1045
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7746580569027581,
      "learning_rate": 5.935516469471514e-06,
      "loss": 0.2836,
      "step": 1046
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7757218368003066,
      "learning_rate": 5.9285648303351404e-06,
      "loss": 0.287,
      "step": 1047
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7898112171962735,
      "learning_rate": 5.921611331656928e-06,
      "loss": 0.3109,
      "step": 1048
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7578780814973718,
      "learning_rate": 5.914655987361934e-06,
      "loss": 0.2946,
      "step": 1049
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7379691507288241,
      "learning_rate": 5.907698811378919e-06,
      "loss": 0.2693,
      "step": 1050
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7877311781565212,
      "learning_rate": 5.900739817640304e-06,
      "loss": 0.2961,
      "step": 1051
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7557597108274106,
      "learning_rate": 5.893779020082159e-06,
      "loss": 0.2796,
      "step": 1052
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8095634039058751,
      "learning_rate": 5.886816432644155e-06,
      "loss": 0.3005,
      "step": 1053
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7591529212463982,
      "learning_rate": 5.8798520692695605e-06,
      "loss": 0.284,
      "step": 1054
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7534791458165836,
      "learning_rate": 5.872885943905191e-06,
      "loss": 0.2619,
      "step": 1055
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8058579322234414,
      "learning_rate": 5.865918070501394e-06,
      "loss": 0.2998,
      "step": 1056
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8433293948861006,
      "learning_rate": 5.858948463012015e-06,
      "loss": 0.2802,
      "step": 1057
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8162325576692708,
      "learning_rate": 5.851977135394379e-06,
      "loss": 0.2986,
      "step": 1058
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7603486249780477,
      "learning_rate": 5.8450041016092465e-06,
      "loss": 0.2767,
      "step": 1059
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7948178965513,
      "learning_rate": 5.838029375620801e-06,
      "loss": 0.2897,
      "step": 1060
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7587209206157945,
      "learning_rate": 5.831052971396613e-06,
      "loss": 0.264,
      "step": 1061
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8025218131456192,
      "learning_rate": 5.824074902907614e-06,
      "loss": 0.2867,
      "step": 1062
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8189309565316873,
      "learning_rate": 5.817095184128067e-06,
      "loss": 0.3032,
      "step": 1063
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8174440816000124,
      "learning_rate": 5.810113829035544e-06,
      "loss": 0.311,
      "step": 1064
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8487380408705193,
      "learning_rate": 5.803130851610887e-06,
      "loss": 0.3046,
      "step": 1065
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7670367935381363,
      "learning_rate": 5.796146265838195e-06,
      "loss": 0.276,
      "step": 1066
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7918266710882845,
      "learning_rate": 5.78916008570478e-06,
      "loss": 0.2783,
      "step": 1067
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.782258362116417,
      "learning_rate": 5.782172325201155e-06,
      "loss": 0.2933,
      "step": 1068
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7739619252484187,
      "learning_rate": 5.77518299832099e-06,
      "loss": 0.2919,
      "step": 1069
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.823706532178144,
      "learning_rate": 5.768192119061096e-06,
      "loss": 0.3119,
      "step": 1070
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.781690162978914,
      "learning_rate": 5.761199701421392e-06,
      "loss": 0.2757,
      "step": 1071
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7822854071331564,
      "learning_rate": 5.754205759404879e-06,
      "loss": 0.283,
      "step": 1072
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7370878731952462,
      "learning_rate": 5.747210307017609e-06,
      "loss": 0.2783,
      "step": 1073
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8120110203299638,
      "learning_rate": 5.740213358268658e-06,
      "loss": 0.3017,
      "step": 1074
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.748086480658812,
      "learning_rate": 5.733214927170101e-06,
      "loss": 0.2677,
      "step": 1075
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.753160684054174,
      "learning_rate": 5.726215027736981e-06,
      "loss": 0.3047,
      "step": 1076
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7178337309596344,
      "learning_rate": 5.719213673987277e-06,
      "loss": 0.2885,
      "step": 1077
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7854732013912585,
      "learning_rate": 5.712210879941891e-06,
      "loss": 0.2915,
      "step": 1078
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8068514524168943,
      "learning_rate": 5.705206659624597e-06,
      "loss": 0.2841,
      "step": 1079
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.746011768199683,
      "learning_rate": 5.698201027062034e-06,
      "loss": 0.2961,
      "step": 1080
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8243585121400852,
      "learning_rate": 5.6911939962836625e-06,
      "loss": 0.3176,
      "step": 1081
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8397163806245905,
      "learning_rate": 5.684185581321752e-06,
      "loss": 0.2897,
      "step": 1082
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7654244375294023,
      "learning_rate": 5.6771757962113325e-06,
      "loss": 0.2884,
      "step": 1083
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.746882730145918,
      "learning_rate": 5.670164654990189e-06,
      "loss": 0.2702,
      "step": 1084
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7614785613871563,
      "learning_rate": 5.663152171698813e-06,
      "loss": 0.3083,
      "step": 1085
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.816870897358134,
      "learning_rate": 5.656138360380391e-06,
      "loss": 0.2934,
      "step": 1086
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8737643468734537,
      "learning_rate": 5.649123235080762e-06,
      "loss": 0.2927,
      "step": 1087
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9303031737992061,
      "learning_rate": 5.642106809848405e-06,
      "loss": 0.2818,
      "step": 1088
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7436365036478807,
      "learning_rate": 5.635089098734394e-06,
      "loss": 0.2787,
      "step": 1089
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7968783018498391,
      "learning_rate": 5.628070115792384e-06,
      "loss": 0.3147,
      "step": 1090
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7802091789462073,
      "learning_rate": 5.621049875078571e-06,
      "loss": 0.298,
      "step": 1091
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7832965946378883,
      "learning_rate": 5.614028390651676e-06,
      "loss": 0.3026,
      "step": 1092
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8263159123177249,
      "learning_rate": 5.6070056765729065e-06,
      "loss": 0.277,
      "step": 1093
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7398232572259636,
      "learning_rate": 5.599981746905935e-06,
      "loss": 0.2802,
      "step": 1094
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7781784240058431,
      "learning_rate": 5.592956615716867e-06,
      "loss": 0.2881,
      "step": 1095
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7683601391234414,
      "learning_rate": 5.585930297074214e-06,
      "loss": 0.2919,
      "step": 1096
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7945218647067406,
      "learning_rate": 5.578902805048866e-06,
      "loss": 0.2658,
      "step": 1097
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7924099255613128,
      "learning_rate": 5.571874153714063e-06,
      "loss": 0.2724,
      "step": 1098
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8032906954145927,
      "learning_rate": 5.564844357145365e-06,
      "loss": 0.2685,
      "step": 1099
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7960903376695436,
      "learning_rate": 5.55781342942063e-06,
      "loss": 0.3017,
      "step": 1100
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.755613999339923,
      "learning_rate": 5.550781384619974e-06,
      "loss": 0.2818,
      "step": 1101
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7941355784513522,
      "learning_rate": 5.543748236825756e-06,
      "loss": 0.2929,
      "step": 1102
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.748903238170741,
      "learning_rate": 5.536714000122543e-06,
      "loss": 0.2847,
      "step": 1103
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7880004779740841,
      "learning_rate": 5.529678688597081e-06,
      "loss": 0.3114,
      "step": 1104
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7918105662720527,
      "learning_rate": 5.522642316338268e-06,
      "loss": 0.2891,
      "step": 1105
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8090174080748075,
      "learning_rate": 5.515604897437129e-06,
      "loss": 0.2945,
      "step": 1106
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7843883399853673,
      "learning_rate": 5.50856644598678e-06,
      "loss": 0.2824,
      "step": 1107
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7686599983829168,
      "learning_rate": 5.501526976082413e-06,
      "loss": 0.2935,
      "step": 1108
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7446525815679835,
      "learning_rate": 5.49448650182125e-06,
      "loss": 0.2779,
      "step": 1109
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7326731070797221,
      "learning_rate": 5.487445037302531e-06,
      "loss": 0.2923,
      "step": 1110
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6919051655959592,
      "learning_rate": 5.480402596627477e-06,
      "loss": 0.2528,
      "step": 1111
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7841914436935139,
      "learning_rate": 5.473359193899265e-06,
      "loss": 0.3022,
      "step": 1112
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7566490193583224,
      "learning_rate": 5.466314843222993e-06,
      "loss": 0.2855,
      "step": 1113
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8030947957119701,
      "learning_rate": 5.459269558705667e-06,
      "loss": 0.2752,
      "step": 1114
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7879627474379077,
      "learning_rate": 5.452223354456154e-06,
      "loss": 0.2732,
      "step": 1115
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.863054344097621,
      "learning_rate": 5.445176244585171e-06,
      "loss": 0.3117,
      "step": 1116
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8160475961158206,
      "learning_rate": 5.438128243205241e-06,
      "loss": 0.2985,
      "step": 1117
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7956596402448185,
      "learning_rate": 5.431079364430677e-06,
      "loss": 0.2962,
      "step": 1118
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7970144102338533,
      "learning_rate": 5.4240296223775465e-06,
      "loss": 0.3023,
      "step": 1119
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8767245232453328,
      "learning_rate": 5.416979031163649e-06,
      "loss": 0.3081,
      "step": 1120
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8024218984204665,
      "learning_rate": 5.409927604908481e-06,
      "loss": 0.2842,
      "step": 1121
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7721847639460926,
      "learning_rate": 5.402875357733215e-06,
      "loss": 0.2817,
      "step": 1122
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.791961687879552,
      "learning_rate": 5.395822303760662e-06,
      "loss": 0.2912,
      "step": 1123
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.793383215323616,
      "learning_rate": 5.388768457115254e-06,
      "loss": 0.2771,
      "step": 1124
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8237351615359557,
      "learning_rate": 5.381713831923008e-06,
      "loss": 0.2917,
      "step": 1125
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7887382289711645,
      "learning_rate": 5.3746584423115025e-06,
      "loss": 0.2819,
      "step": 1126
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7387674130310357,
      "learning_rate": 5.367602302409841e-06,
      "loss": 0.2626,
      "step": 1127
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7064352530937822,
      "learning_rate": 5.360545426348638e-06,
      "loss": 0.2681,
      "step": 1128
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7533920953465262,
      "learning_rate": 5.353487828259973e-06,
      "loss": 0.2752,
      "step": 1129
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7617165409690563,
      "learning_rate": 5.346429522277381e-06,
      "loss": 0.269,
      "step": 1130
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8262196692302946,
      "learning_rate": 5.339370522535805e-06,
      "loss": 0.2953,
      "step": 1131
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7620019074306038,
      "learning_rate": 5.3323108431715845e-06,
      "loss": 0.2846,
      "step": 1132
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8289544667819924,
      "learning_rate": 5.325250498322417e-06,
      "loss": 0.2999,
      "step": 1133
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7694419357293719,
      "learning_rate": 5.318189502127332e-06,
      "loss": 0.2929,
      "step": 1134
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8078576344006587,
      "learning_rate": 5.311127868726667e-06,
      "loss": 0.3142,
      "step": 1135
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7905364898452981,
      "learning_rate": 5.304065612262033e-06,
      "loss": 0.2956,
      "step": 1136
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.791699943287134,
      "learning_rate": 5.297002746876284e-06,
      "loss": 0.3029,
      "step": 1137
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7771182754836248,
      "learning_rate": 5.289939286713505e-06,
      "loss": 0.2891,
      "step": 1138
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.702084949783083,
      "learning_rate": 5.282875245918963e-06,
      "loss": 0.2604,
      "step": 1139
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7519354595037325,
      "learning_rate": 5.275810638639088e-06,
      "loss": 0.3045,
      "step": 1140
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7640004842755324,
      "learning_rate": 5.26874547902145e-06,
      "loss": 0.2845,
      "step": 1141
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7593896191664194,
      "learning_rate": 5.2616797812147205e-06,
      "loss": 0.2914,
      "step": 1142
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7519550014873447,
      "learning_rate": 5.2546135593686484e-06,
      "loss": 0.2729,
      "step": 1143
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7761156790560342,
      "learning_rate": 5.247546827634035e-06,
      "loss": 0.2712,
      "step": 1144
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7155390984024538,
      "learning_rate": 5.240479600162704e-06,
      "loss": 0.2768,
      "step": 1145
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7831462256842718,
      "learning_rate": 5.233411891107464e-06,
      "loss": 0.282,
      "step": 1146
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7604036144919483,
      "learning_rate": 5.226343714622096e-06,
      "loss": 0.2662,
      "step": 1147
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8247285947951177,
      "learning_rate": 5.219275084861315e-06,
      "loss": 0.3117,
      "step": 1148
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7883599507605149,
      "learning_rate": 5.212206015980742e-06,
      "loss": 0.2769,
      "step": 1149
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7594303069395828,
      "learning_rate": 5.2051365221368785e-06,
      "loss": 0.2679,
      "step": 1150
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.723646641347664,
      "learning_rate": 5.198066617487078e-06,
      "loss": 0.2787,
      "step": 1151
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7616523245578691,
      "learning_rate": 5.190996316189515e-06,
      "loss": 0.3003,
      "step": 1152
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7700452096437183,
      "learning_rate": 5.1839256324031575e-06,
      "loss": 0.2724,
      "step": 1153
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8437413615659389,
      "learning_rate": 5.176854580287744e-06,
      "loss": 0.3135,
      "step": 1154
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.761742485881136,
      "learning_rate": 5.169783174003744e-06,
      "loss": 0.2619,
      "step": 1155
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7566382311600068,
      "learning_rate": 5.1627114277123435e-06,
      "loss": 0.2659,
      "step": 1156
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7737890293709694,
      "learning_rate": 5.155639355575402e-06,
      "loss": 0.2771,
      "step": 1157
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7698235080417508,
      "learning_rate": 5.14856697175544e-06,
      "loss": 0.2841,
      "step": 1158
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8290447510228104,
      "learning_rate": 5.141494290415592e-06,
      "loss": 0.271,
      "step": 1159
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8038427971939314,
      "learning_rate": 5.1344213257195965e-06,
      "loss": 0.3053,
      "step": 1160
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7725647911649238,
      "learning_rate": 5.127348091831755e-06,
      "loss": 0.2936,
      "step": 1161
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7545183277407386,
      "learning_rate": 5.120274602916915e-06,
      "loss": 0.2849,
      "step": 1162
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7784944153716037,
      "learning_rate": 5.1132008731404225e-06,
      "loss": 0.2993,
      "step": 1163
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7717247776542023,
      "learning_rate": 5.106126916668118e-06,
      "loss": 0.2936,
      "step": 1164
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7533579650251199,
      "learning_rate": 5.099052747666289e-06,
      "loss": 0.2905,
      "step": 1165
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7622170858269731,
      "learning_rate": 5.09197838030165e-06,
      "loss": 0.2879,
      "step": 1166
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8157259622030317,
      "learning_rate": 5.084903828741312e-06,
      "loss": 0.3097,
      "step": 1167
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7550930780832696,
      "learning_rate": 5.0778291071527596e-06,
      "loss": 0.2913,
      "step": 1168
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7938034127009114,
      "learning_rate": 5.070754229703811e-06,
      "loss": 0.277,
      "step": 1169
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7750475054050129,
      "learning_rate": 5.063679210562602e-06,
      "loss": 0.2857,
      "step": 1170
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8213378532764167,
      "learning_rate": 5.056604063897549e-06,
      "loss": 0.3028,
      "step": 1171
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7890518604135954,
      "learning_rate": 5.049528803877324e-06,
      "loss": 0.27,
      "step": 1172
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7842815707819177,
      "learning_rate": 5.042453444670829e-06,
      "loss": 0.2934,
      "step": 1173
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7001384125381409,
      "learning_rate": 5.0353780004471605e-06,
      "loss": 0.2635,
      "step": 1174
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8042052600107539,
      "learning_rate": 5.028302485375588e-06,
      "loss": 0.2766,
      "step": 1175
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.853731150769343,
      "learning_rate": 5.021226913625522e-06,
      "loss": 0.2978,
      "step": 1176
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8171484998627504,
      "learning_rate": 5.0141512993664864e-06,
      "loss": 0.2816,
      "step": 1177
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7745860019829481,
      "learning_rate": 5.007075656768091e-06,
      "loss": 0.2834,
      "step": 1178
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7512088941562156,
      "learning_rate": 5e-06,
      "loss": 0.2917,
      "step": 1179
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7881164253964926,
      "learning_rate": 4.99292434323191e-06,
      "loss": 0.3089,
      "step": 1180
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7569657991672306,
      "learning_rate": 4.985848700633515e-06,
      "loss": 0.2743,
      "step": 1181
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.796761128733197,
      "learning_rate": 4.97877308637448e-06,
      "loss": 0.3002,
      "step": 1182
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7666808981790826,
      "learning_rate": 4.971697514624414e-06,
      "loss": 0.3004,
      "step": 1183
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8287139502801465,
      "learning_rate": 4.964621999552841e-06,
      "loss": 0.2981,
      "step": 1184
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7994338828143851,
      "learning_rate": 4.957546555329174e-06,
      "loss": 0.2831,
      "step": 1185
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7574839775987559,
      "learning_rate": 4.950471196122677e-06,
      "loss": 0.2791,
      "step": 1186
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7340149596435569,
      "learning_rate": 4.943395936102453e-06,
      "loss": 0.2883,
      "step": 1187
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8200001114406699,
      "learning_rate": 4.9363207894374e-06,
      "loss": 0.286,
      "step": 1188
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7668564356958615,
      "learning_rate": 4.929245770296191e-06,
      "loss": 0.3029,
      "step": 1189
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.834913925357583,
      "learning_rate": 4.922170892847242e-06,
      "loss": 0.292,
      "step": 1190
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7774761636602814,
      "learning_rate": 4.91509617125869e-06,
      "loss": 0.286,
      "step": 1191
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7549615815768301,
      "learning_rate": 4.908021619698352e-06,
      "loss": 0.2752,
      "step": 1192
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8075422483064455,
      "learning_rate": 4.900947252333712e-06,
      "loss": 0.2961,
      "step": 1193
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8193148672888736,
      "learning_rate": 4.8938730833318825e-06,
      "loss": 0.2931,
      "step": 1194
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7766615732099263,
      "learning_rate": 4.8867991268595775e-06,
      "loss": 0.287,
      "step": 1195
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7534727267167107,
      "learning_rate": 4.879725397083087e-06,
      "loss": 0.2698,
      "step": 1196
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7830632643519113,
      "learning_rate": 4.872651908168244e-06,
      "loss": 0.2949,
      "step": 1197
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7846361367239916,
      "learning_rate": 4.865578674280406e-06,
      "loss": 0.2856,
      "step": 1198
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7488044851311444,
      "learning_rate": 4.85850570958441e-06,
      "loss": 0.2917,
      "step": 1199
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7619535565895358,
      "learning_rate": 4.851433028244562e-06,
      "loss": 0.2948,
      "step": 1200
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7666243522548212,
      "learning_rate": 4.844360644424599e-06,
      "loss": 0.3014,
      "step": 1201
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7516168306882631,
      "learning_rate": 4.837288572287658e-06,
      "loss": 0.2641,
      "step": 1202
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7610028402217457,
      "learning_rate": 4.830216825996257e-06,
      "loss": 0.2698,
      "step": 1203
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7155941075888851,
      "learning_rate": 4.8231454197122575e-06,
      "loss": 0.2603,
      "step": 1204
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8044626656070201,
      "learning_rate": 4.816074367596843e-06,
      "loss": 0.2753,
      "step": 1205
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.762059344456115,
      "learning_rate": 4.809003683810487e-06,
      "loss": 0.2735,
      "step": 1206
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7705871324957114,
      "learning_rate": 4.801933382512923e-06,
      "loss": 0.2757,
      "step": 1207
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8479965168917547,
      "learning_rate": 4.794863477863122e-06,
      "loss": 0.302,
      "step": 1208
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7408716290147157,
      "learning_rate": 4.78779398401926e-06,
      "loss": 0.2684,
      "step": 1209
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7800057850828923,
      "learning_rate": 4.780724915138686e-06,
      "loss": 0.2928,
      "step": 1210
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.800854488285206,
      "learning_rate": 4.7736562853779066e-06,
      "loss": 0.2931,
      "step": 1211
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7635214091361583,
      "learning_rate": 4.766588108892538e-06,
      "loss": 0.2694,
      "step": 1212
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7741626623493758,
      "learning_rate": 4.7595203998373e-06,
      "loss": 0.2866,
      "step": 1213
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.796275596041006,
      "learning_rate": 4.752453172365966e-06,
      "loss": 0.2893,
      "step": 1214
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8135850217874707,
      "learning_rate": 4.745386440631354e-06,
      "loss": 0.2645,
      "step": 1215
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7304812672336712,
      "learning_rate": 4.738320218785281e-06,
      "loss": 0.2761,
      "step": 1216
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7760859189425744,
      "learning_rate": 4.73125452097855e-06,
      "loss": 0.2765,
      "step": 1217
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7822746272154665,
      "learning_rate": 4.7241893613609126e-06,
      "loss": 0.3227,
      "step": 1218
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8481685537611515,
      "learning_rate": 4.717124754081038e-06,
      "loss": 0.3341,
      "step": 1219
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8163082557159872,
      "learning_rate": 4.710060713286496e-06,
      "loss": 0.3225,
      "step": 1220
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8133514730121715,
      "learning_rate": 4.702997253123716e-06,
      "loss": 0.2847,
      "step": 1221
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8098666434943971,
      "learning_rate": 4.6959343877379695e-06,
      "loss": 0.3089,
      "step": 1222
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7451799617476838,
      "learning_rate": 4.688872131273334e-06,
      "loss": 0.3,
      "step": 1223
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8520230703695089,
      "learning_rate": 4.6818104978726685e-06,
      "loss": 0.3056,
      "step": 1224
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6983998132159116,
      "learning_rate": 4.674749501677585e-06,
      "loss": 0.2581,
      "step": 1225
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.766688401310411,
      "learning_rate": 4.667689156828417e-06,
      "loss": 0.32,
      "step": 1226
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8269649146837612,
      "learning_rate": 4.660629477464197e-06,
      "loss": 0.3337,
      "step": 1227
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7319909695830631,
      "learning_rate": 4.653570477722621e-06,
      "loss": 0.2572,
      "step": 1228
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.775347155351386,
      "learning_rate": 4.646512171740028e-06,
      "loss": 0.2829,
      "step": 1229
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7426128519076995,
      "learning_rate": 4.639454573651364e-06,
      "loss": 0.2634,
      "step": 1230
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7591026354749015,
      "learning_rate": 4.6323976975901595e-06,
      "loss": 0.2692,
      "step": 1231
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7476044702955544,
      "learning_rate": 4.625341557688499e-06,
      "loss": 0.2784,
      "step": 1232
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7256444427495207,
      "learning_rate": 4.618286168076993e-06,
      "loss": 0.2911,
      "step": 1233
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8175919252872501,
      "learning_rate": 4.611231542884747e-06,
      "loss": 0.2807,
      "step": 1234
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7776916907070021,
      "learning_rate": 4.6041776962393405e-06,
      "loss": 0.2702,
      "step": 1235
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7672353658275362,
      "learning_rate": 4.597124642266788e-06,
      "loss": 0.2813,
      "step": 1236
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7556289913588801,
      "learning_rate": 4.590072395091521e-06,
      "loss": 0.2782,
      "step": 1237
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8224055264884026,
      "learning_rate": 4.583020968836353e-06,
      "loss": 0.3132,
      "step": 1238
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7669936996179963,
      "learning_rate": 4.575970377622456e-06,
      "loss": 0.2957,
      "step": 1239
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7668638596682423,
      "learning_rate": 4.568920635569325e-06,
      "loss": 0.2608,
      "step": 1240
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8525175060906292,
      "learning_rate": 4.56187175679476e-06,
      "loss": 0.3004,
      "step": 1241
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7390079191318651,
      "learning_rate": 4.554823755414829e-06,
      "loss": 0.2934,
      "step": 1242
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7311690791099713,
      "learning_rate": 4.547776645543846e-06,
      "loss": 0.2506,
      "step": 1243
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.790126318576123,
      "learning_rate": 4.540730441294334e-06,
      "loss": 0.2858,
      "step": 1244
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7302250199467034,
      "learning_rate": 4.533685156777008e-06,
      "loss": 0.2519,
      "step": 1245
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7427037249198536,
      "learning_rate": 4.526640806100737e-06,
      "loss": 0.2874,
      "step": 1246
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7708562598159626,
      "learning_rate": 4.519597403372524e-06,
      "loss": 0.3055,
      "step": 1247
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7921984655593431,
      "learning_rate": 4.5125549626974695e-06,
      "loss": 0.2894,
      "step": 1248
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.775308186051875,
      "learning_rate": 4.505513498178752e-06,
      "loss": 0.2769,
      "step": 1249
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8833996881169215,
      "learning_rate": 4.4984730239175885e-06,
      "loss": 0.297,
      "step": 1250
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8369037241238041,
      "learning_rate": 4.491433554013221e-06,
      "loss": 0.3015,
      "step": 1251
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7526080752570474,
      "learning_rate": 4.484395102562873e-06,
      "loss": 0.2861,
      "step": 1252
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8051923161619936,
      "learning_rate": 4.477357683661734e-06,
      "loss": 0.2735,
      "step": 1253
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7513935879095343,
      "learning_rate": 4.47032131140292e-06,
      "loss": 0.2919,
      "step": 1254
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.75784563216117,
      "learning_rate": 4.463285999877458e-06,
      "loss": 0.2718,
      "step": 1255
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7618086444482179,
      "learning_rate": 4.456251763174245e-06,
      "loss": 0.2986,
      "step": 1256
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7276768770693584,
      "learning_rate": 4.449218615380029e-06,
      "loss": 0.2547,
      "step": 1257
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7698190640507314,
      "learning_rate": 4.442186570579373e-06,
      "loss": 0.3075,
      "step": 1258
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7801554853958343,
      "learning_rate": 4.435155642854637e-06,
      "loss": 0.2844,
      "step": 1259
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8088200657764824,
      "learning_rate": 4.42812584628594e-06,
      "loss": 0.3117,
      "step": 1260
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7912254364404534,
      "learning_rate": 4.421097194951137e-06,
      "loss": 0.2829,
      "step": 1261
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8036337749633977,
      "learning_rate": 4.414069702925789e-06,
      "loss": 0.299,
      "step": 1262
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8400590974328768,
      "learning_rate": 4.407043384283136e-06,
      "loss": 0.272,
      "step": 1263
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7300685499152774,
      "learning_rate": 4.400018253094065e-06,
      "loss": 0.2667,
      "step": 1264
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8060638670739841,
      "learning_rate": 4.3929943234270935e-06,
      "loss": 0.281,
      "step": 1265
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7972950584939378,
      "learning_rate": 4.385971609348324e-06,
      "loss": 0.2983,
      "step": 1266
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7507803319593184,
      "learning_rate": 4.37895012492143e-06,
      "loss": 0.2909,
      "step": 1267
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7618412103236987,
      "learning_rate": 4.371929884207617e-06,
      "loss": 0.2752,
      "step": 1268
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7790336633895281,
      "learning_rate": 4.364910901265607e-06,
      "loss": 0.2855,
      "step": 1269
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7756925194523558,
      "learning_rate": 4.3578931901515955e-06,
      "loss": 0.261,
      "step": 1270
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7315361902531536,
      "learning_rate": 4.350876764919238e-06,
      "loss": 0.2495,
      "step": 1271
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7719988659865565,
      "learning_rate": 4.343861639619611e-06,
      "loss": 0.2717,
      "step": 1272
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8100150870883477,
      "learning_rate": 4.336847828301189e-06,
      "loss": 0.2824,
      "step": 1273
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.823682073465986,
      "learning_rate": 4.329835345009813e-06,
      "loss": 0.3037,
      "step": 1274
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7460762702869246,
      "learning_rate": 4.322824203788669e-06,
      "loss": 0.2917,
      "step": 1275
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7353169942921559,
      "learning_rate": 4.3158144186782504e-06,
      "loss": 0.2811,
      "step": 1276
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7603828964796858,
      "learning_rate": 4.308806003716338e-06,
      "loss": 0.2863,
      "step": 1277
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7724520272921733,
      "learning_rate": 4.301798972937968e-06,
      "loss": 0.2818,
      "step": 1278
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7984084364647598,
      "learning_rate": 4.294793340375405e-06,
      "loss": 0.2837,
      "step": 1279
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7170114524161252,
      "learning_rate": 4.287789120058111e-06,
      "loss": 0.2747,
      "step": 1280
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8142555311195558,
      "learning_rate": 4.280786326012724e-06,
      "loss": 0.2773,
      "step": 1281
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8021298135395517,
      "learning_rate": 4.273784972263021e-06,
      "loss": 0.2709,
      "step": 1282
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7764967904249219,
      "learning_rate": 4.266785072829901e-06,
      "loss": 0.2749,
      "step": 1283
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7838344332917934,
      "learning_rate": 4.259786641731344e-06,
      "loss": 0.3014,
      "step": 1284
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7947785094523545,
      "learning_rate": 4.252789692982394e-06,
      "loss": 0.2839,
      "step": 1285
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7631497051286774,
      "learning_rate": 4.245794240595123e-06,
      "loss": 0.2687,
      "step": 1286
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6999894189173538,
      "learning_rate": 4.23880029857861e-06,
      "loss": 0.2749,
      "step": 1287
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7818475549695977,
      "learning_rate": 4.231807880938904e-06,
      "loss": 0.303,
      "step": 1288
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7557704162066594,
      "learning_rate": 4.224817001679011e-06,
      "loss": 0.285,
      "step": 1289
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7698994840897398,
      "learning_rate": 4.217827674798845e-06,
      "loss": 0.3063,
      "step": 1290
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7792464463928961,
      "learning_rate": 4.210839914295219e-06,
      "loss": 0.2897,
      "step": 1291
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7955822117184155,
      "learning_rate": 4.203853734161806e-06,
      "loss": 0.2987,
      "step": 1292
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7430287002666323,
      "learning_rate": 4.196869148389114e-06,
      "loss": 0.2767,
      "step": 1293
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7768834121536915,
      "learning_rate": 4.189886170964458e-06,
      "loss": 0.2789,
      "step": 1294
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7454272772667635,
      "learning_rate": 4.1829048158719344e-06,
      "loss": 0.2818,
      "step": 1295
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7438218072906642,
      "learning_rate": 4.1759250970923876e-06,
      "loss": 0.2764,
      "step": 1296
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7415659780804209,
      "learning_rate": 4.168947028603389e-06,
      "loss": 0.2736,
      "step": 1297
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.719307170169241,
      "learning_rate": 4.161970624379201e-06,
      "loss": 0.2663,
      "step": 1298
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.758340994759609,
      "learning_rate": 4.154995898390756e-06,
      "loss": 0.2733,
      "step": 1299
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8022637083163398,
      "learning_rate": 4.1480228646056225e-06,
      "loss": 0.2785,
      "step": 1300
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7182821438078845,
      "learning_rate": 4.141051536987986e-06,
      "loss": 0.2788,
      "step": 1301
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7663891624371467,
      "learning_rate": 4.1340819294986076e-06,
      "loss": 0.2778,
      "step": 1302
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7208431351706666,
      "learning_rate": 4.127114056094811e-06,
      "loss": 0.2633,
      "step": 1303
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7974082195481736,
      "learning_rate": 4.12014793073044e-06,
      "loss": 0.3,
      "step": 1304
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8222205016613726,
      "learning_rate": 4.113183567355846e-06,
      "loss": 0.2805,
      "step": 1305
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7399086737286078,
      "learning_rate": 4.106220979917845e-06,
      "loss": 0.2661,
      "step": 1306
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7808395203278649,
      "learning_rate": 4.099260182359698e-06,
      "loss": 0.2911,
      "step": 1307
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7494320478150415,
      "learning_rate": 4.092301188621084e-06,
      "loss": 0.2794,
      "step": 1308
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7427337276701809,
      "learning_rate": 4.085344012638067e-06,
      "loss": 0.2779,
      "step": 1309
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8034930679203175,
      "learning_rate": 4.078388668343075e-06,
      "loss": 0.319,
      "step": 1310
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7332268944082129,
      "learning_rate": 4.071435169664861e-06,
      "loss": 0.2768,
      "step": 1311
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7517340267954221,
      "learning_rate": 4.064483530528487e-06,
      "loss": 0.2894,
      "step": 1312
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7338521582756717,
      "learning_rate": 4.057533764855291e-06,
      "loss": 0.2711,
      "step": 1313
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7210372416603034,
      "learning_rate": 4.050585886562858e-06,
      "loss": 0.2595,
      "step": 1314
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7825697380328936,
      "learning_rate": 4.043639909564993e-06,
      "loss": 0.3043,
      "step": 1315
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8048418036780111,
      "learning_rate": 4.036695847771694e-06,
      "loss": 0.3025,
      "step": 1316
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7573161812332089,
      "learning_rate": 4.029753715089124e-06,
      "loss": 0.289,
      "step": 1317
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7705025731814703,
      "learning_rate": 4.02281352541958e-06,
      "loss": 0.2819,
      "step": 1318
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8187891502418783,
      "learning_rate": 4.015875292661474e-06,
      "loss": 0.2872,
      "step": 1319
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7392375035581149,
      "learning_rate": 4.008939030709291e-06,
      "loss": 0.2769,
      "step": 1320
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7874965122169427,
      "learning_rate": 4.002004753453577e-06,
      "loss": 0.2744,
      "step": 1321
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8084721228388309,
      "learning_rate": 3.995072474780899e-06,
      "loss": 0.299,
      "step": 1322
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8219790463423629,
      "learning_rate": 3.988142208573822e-06,
      "loss": 0.2862,
      "step": 1323
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8433253296573074,
      "learning_rate": 3.981213968710882e-06,
      "loss": 0.2639,
      "step": 1324
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.810957333088122,
      "learning_rate": 3.9742877690665575e-06,
      "loss": 0.278,
      "step": 1325
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7810251939825757,
      "learning_rate": 3.967363623511238e-06,
      "loss": 0.286,
      "step": 1326
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7581984467867272,
      "learning_rate": 3.960441545911205e-06,
      "loss": 0.276,
      "step": 1327
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8585447551084022,
      "learning_rate": 3.953521550128591e-06,
      "loss": 0.306,
      "step": 1328
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7983329800301929,
      "learning_rate": 3.94660365002137e-06,
      "loss": 0.2817,
      "step": 1329
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8012999589789032,
      "learning_rate": 3.939687859443308e-06,
      "loss": 0.286,
      "step": 1330
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7989905600081587,
      "learning_rate": 3.932774192243953e-06,
      "loss": 0.2901,
      "step": 1331
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7796563094205847,
      "learning_rate": 3.925862662268602e-06,
      "loss": 0.3068,
      "step": 1332
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7695199644442456,
      "learning_rate": 3.918953283358266e-06,
      "loss": 0.2817,
      "step": 1333
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8808322456406279,
      "learning_rate": 3.912046069349654e-06,
      "loss": 0.2796,
      "step": 1334
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7600835244123556,
      "learning_rate": 3.905141034075135e-06,
      "loss": 0.2974,
      "step": 1335
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8082901662601609,
      "learning_rate": 3.898238191362718e-06,
      "loss": 0.2867,
      "step": 1336
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8014728421832985,
      "learning_rate": 3.891337555036022e-06,
      "loss": 0.2932,
      "step": 1337
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7638007793673576,
      "learning_rate": 3.884439138914243e-06,
      "loss": 0.2876,
      "step": 1338
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7641813101747438,
      "learning_rate": 3.877542956812137e-06,
      "loss": 0.2962,
      "step": 1339
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7614068045737236,
      "learning_rate": 3.870649022539979e-06,
      "loss": 0.278,
      "step": 1340
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7789555251533775,
      "learning_rate": 3.863757349903551e-06,
      "loss": 0.2717,
      "step": 1341
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8402972800143939,
      "learning_rate": 3.856867952704095e-06,
      "loss": 0.2831,
      "step": 1342
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7586240743769034,
      "learning_rate": 3.849980844738308e-06,
      "loss": 0.2704,
      "step": 1343
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7964940612159159,
      "learning_rate": 3.843096039798293e-06,
      "loss": 0.3061,
      "step": 1344
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7344932953471209,
      "learning_rate": 3.836213551671547e-06,
      "loss": 0.2887,
      "step": 1345
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7714212158290606,
      "learning_rate": 3.829333394140924e-06,
      "loss": 0.2727,
      "step": 1346
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7562681736013256,
      "learning_rate": 3.822455580984613e-06,
      "loss": 0.2943,
      "step": 1347
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.815911212809822,
      "learning_rate": 3.8155801259761065e-06,
      "loss": 0.2837,
      "step": 1348
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7635803455543975,
      "learning_rate": 3.808707042884176e-06,
      "loss": 0.2598,
      "step": 1349
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7434867748368242,
      "learning_rate": 3.8018363454728415e-06,
      "loss": 0.2728,
      "step": 1350
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.766528373133731,
      "learning_rate": 3.794968047501345e-06,
      "loss": 0.3032,
      "step": 1351
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7888489270022023,
      "learning_rate": 3.7881021627241283e-06,
      "loss": 0.2806,
      "step": 1352
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7502656429105663,
      "learning_rate": 3.781238704890793e-06,
      "loss": 0.2764,
      "step": 1353
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7122013801019099,
      "learning_rate": 3.7743776877460864e-06,
      "loss": 0.2701,
      "step": 1354
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7678243460082275,
      "learning_rate": 3.767519125029865e-06,
      "loss": 0.2812,
      "step": 1355
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7834176837984048,
      "learning_rate": 3.760663030477072e-06,
      "loss": 0.2765,
      "step": 1356
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.787364682648952,
      "learning_rate": 3.7538094178177054e-06,
      "loss": 0.2885,
      "step": 1357
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8438388255234791,
      "learning_rate": 3.7469583007767964e-06,
      "loss": 0.267,
      "step": 1358
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7842786647769401,
      "learning_rate": 3.7401096930743753e-06,
      "loss": 0.2847,
      "step": 1359
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7348477602112988,
      "learning_rate": 3.733263608425447e-06,
      "loss": 0.2585,
      "step": 1360
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7447236894108619,
      "learning_rate": 3.72642006053997e-06,
      "loss": 0.2734,
      "step": 1361
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8686019841432686,
      "learning_rate": 3.7195790631228133e-06,
      "loss": 0.2985,
      "step": 1362
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.770815989855171,
      "learning_rate": 3.7127406298737458e-06,
      "loss": 0.2684,
      "step": 1363
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7908761113644293,
      "learning_rate": 3.705904774487396e-06,
      "loss": 0.28,
      "step": 1364
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8237445588721097,
      "learning_rate": 3.6990715106532353e-06,
      "loss": 0.2963,
      "step": 1365
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8025330404556588,
      "learning_rate": 3.69224085205554e-06,
      "loss": 0.2832,
      "step": 1366
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7828270051909172,
      "learning_rate": 3.6854128123733736e-06,
      "loss": 0.3106,
      "step": 1367
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.735955608532665,
      "learning_rate": 3.6785874052805515e-06,
      "loss": 0.3002,
      "step": 1368
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7758611098432651,
      "learning_rate": 3.6717646444456196e-06,
      "loss": 0.2889,
      "step": 1369
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8324886624582384,
      "learning_rate": 3.664944543531822e-06,
      "loss": 0.2913,
      "step": 1370
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8049678130134647,
      "learning_rate": 3.658127116197079e-06,
      "loss": 0.2677,
      "step": 1371
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7595775595477665,
      "learning_rate": 3.651312376093955e-06,
      "loss": 0.2742,
      "step": 1372
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7237651462801606,
      "learning_rate": 3.6445003368696317e-06,
      "loss": 0.2768,
      "step": 1373
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7710929317209538,
      "learning_rate": 3.6376910121658867e-06,
      "loss": 0.2743,
      "step": 1374
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7537876747484944,
      "learning_rate": 3.6308844156190546e-06,
      "loss": 0.2868,
      "step": 1375
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7641796818638373,
      "learning_rate": 3.6240805608600137e-06,
      "loss": 0.2754,
      "step": 1376
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7544180260481639,
      "learning_rate": 3.6172794615141448e-06,
      "loss": 0.265,
      "step": 1377
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7710532721415672,
      "learning_rate": 3.6104811312013175e-06,
      "loss": 0.2735,
      "step": 1378
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7564284389083008,
      "learning_rate": 3.60368558353585e-06,
      "loss": 0.2907,
      "step": 1379
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.781241199417252,
      "learning_rate": 3.596892832126494e-06,
      "loss": 0.2965,
      "step": 1380
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8274007774813885,
      "learning_rate": 3.590102890576394e-06,
      "loss": 0.3006,
      "step": 1381
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7798626990612841,
      "learning_rate": 3.5833157724830757e-06,
      "loss": 0.2844,
      "step": 1382
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8263186000400004,
      "learning_rate": 3.576531491438403e-06,
      "loss": 0.2763,
      "step": 1383
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8158241132090143,
      "learning_rate": 3.569750061028565e-06,
      "loss": 0.2764,
      "step": 1384
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7951284840063413,
      "learning_rate": 3.5629714948340354e-06,
      "loss": 0.2667,
      "step": 1385
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.763609028214548,
      "learning_rate": 3.5561958064295585e-06,
      "loss": 0.277,
      "step": 1386
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7129397694351979,
      "learning_rate": 3.5494230093841144e-06,
      "loss": 0.2599,
      "step": 1387
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8264161900452021,
      "learning_rate": 3.5426531172608858e-06,
      "loss": 0.2805,
      "step": 1388
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8075841554963025,
      "learning_rate": 3.5358861436172487e-06,
      "loss": 0.3119,
      "step": 1389
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8103059493412832,
      "learning_rate": 3.5291221020047257e-06,
      "loss": 0.2822,
      "step": 1390
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8174274559956919,
      "learning_rate": 3.5223610059689733e-06,
      "loss": 0.2584,
      "step": 1391
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7182919896957126,
      "learning_rate": 3.515602869049746e-06,
      "loss": 0.2735,
      "step": 1392
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7454960939166745,
      "learning_rate": 3.5088477047808738e-06,
      "loss": 0.2801,
      "step": 1393
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7836740560459942,
      "learning_rate": 3.5020955266902344e-06,
      "loss": 0.2879,
      "step": 1394
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.726190256143591,
      "learning_rate": 3.495346348299724e-06,
      "loss": 0.2666,
      "step": 1395
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8431236579990274,
      "learning_rate": 3.4886001831252324e-06,
      "loss": 0.2986,
      "step": 1396
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.748941065876904,
      "learning_rate": 3.481857044676613e-06,
      "loss": 0.2749,
      "step": 1397
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7416022433533628,
      "learning_rate": 3.475116946457662e-06,
      "loss": 0.2729,
      "step": 1398
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7626500047935744,
      "learning_rate": 3.4683799019660834e-06,
      "loss": 0.2811,
      "step": 1399
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7590606975476356,
      "learning_rate": 3.4616459246934686e-06,
      "loss": 0.2831,
      "step": 1400
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7661716577948693,
      "learning_rate": 3.4549150281252635e-06,
      "loss": 0.277,
      "step": 1401
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7516675691947509,
      "learning_rate": 3.448187225740749e-06,
      "loss": 0.2839,
      "step": 1402
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7263576253344189,
      "learning_rate": 3.4414625310130038e-06,
      "loss": 0.2734,
      "step": 1403
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8241545025168745,
      "learning_rate": 3.4347409574088896e-06,
      "loss": 0.3144,
      "step": 1404
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8340834781584392,
      "learning_rate": 3.4280225183890124e-06,
      "loss": 0.292,
      "step": 1405
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8190969034855755,
      "learning_rate": 3.4213072274077047e-06,
      "loss": 0.2989,
      "step": 1406
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8151547032195056,
      "learning_rate": 3.4145950979129916e-06,
      "loss": 0.2767,
      "step": 1407
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7717566216137728,
      "learning_rate": 3.4078861433465693e-06,
      "loss": 0.2848,
      "step": 1408
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8115393033973809,
      "learning_rate": 3.401180377143774e-06,
      "loss": 0.2928,
      "step": 1409
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7938098806120854,
      "learning_rate": 3.39447781273356e-06,
      "loss": 0.2762,
      "step": 1410
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7541564963127007,
      "learning_rate": 3.3877784635384668e-06,
      "loss": 0.2816,
      "step": 1411
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7334771419839684,
      "learning_rate": 3.3810823429745955e-06,
      "loss": 0.2954,
      "step": 1412
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8010339997727411,
      "learning_rate": 3.374389464451583e-06,
      "loss": 0.2782,
      "step": 1413
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7961388954433084,
      "learning_rate": 3.3676998413725726e-06,
      "loss": 0.2659,
      "step": 1414
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7486760508730766,
      "learning_rate": 3.361013487134188e-06,
      "loss": 0.2799,
      "step": 1415
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7626649257403206,
      "learning_rate": 3.354330415126508e-06,
      "loss": 0.2689,
      "step": 1416
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8371777383779784,
      "learning_rate": 3.3476506387330355e-06,
      "loss": 0.2828,
      "step": 1417
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8257285770605151,
      "learning_rate": 3.3409741713306777e-06,
      "loss": 0.2913,
      "step": 1418
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7392851456255574,
      "learning_rate": 3.3343010262897125e-06,
      "loss": 0.2643,
      "step": 1419
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.761407203268617,
      "learning_rate": 3.327631216973766e-06,
      "loss": 0.2806,
      "step": 1420
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7796427469978933,
      "learning_rate": 3.320964756739781e-06,
      "loss": 0.2908,
      "step": 1421
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7898480255189366,
      "learning_rate": 3.3143016589379984e-06,
      "loss": 0.2948,
      "step": 1422
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7681704085632984,
      "learning_rate": 3.3076419369119216e-06,
      "loss": 0.2845,
      "step": 1423
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8297888984936387,
      "learning_rate": 3.300985603998296e-06,
      "loss": 0.3058,
      "step": 1424
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7893715356308656,
      "learning_rate": 3.2943326735270766e-06,
      "loss": 0.2717,
      "step": 1425
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7739696576870105,
      "learning_rate": 3.2876831588214093e-06,
      "loss": 0.2697,
      "step": 1426
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7545987450809305,
      "learning_rate": 3.2810370731975956e-06,
      "loss": 0.2953,
      "step": 1427
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7754957405543996,
      "learning_rate": 3.2743944299650733e-06,
      "loss": 0.2585,
      "step": 1428
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7468553517885084,
      "learning_rate": 3.2677552424263836e-06,
      "loss": 0.2859,
      "step": 1429
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7538488053282489,
      "learning_rate": 3.2611195238771494e-06,
      "loss": 0.2784,
      "step": 1430
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7750311858623725,
      "learning_rate": 3.254487287606044e-06,
      "loss": 0.2637,
      "step": 1431
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.781107069099668,
      "learning_rate": 3.247858546894772e-06,
      "loss": 0.2694,
      "step": 1432
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7288059624081348,
      "learning_rate": 3.2412333150180298e-06,
      "loss": 0.2645,
      "step": 1433
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7833960854502732,
      "learning_rate": 3.234611605243496e-06,
      "loss": 0.2983,
      "step": 1434
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7438106341904023,
      "learning_rate": 3.2279934308317905e-06,
      "loss": 0.2878,
      "step": 1435
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7631141277617292,
      "learning_rate": 3.2213788050364557e-06,
      "loss": 0.2513,
      "step": 1436
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7714580356451604,
      "learning_rate": 3.2147677411039236e-06,
      "loss": 0.3255,
      "step": 1437
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7616161664700041,
      "learning_rate": 3.2081602522734987e-06,
      "loss": 0.2632,
      "step": 1438
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7668454078560527,
      "learning_rate": 3.2015563517773214e-06,
      "loss": 0.2881,
      "step": 1439
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7379101719413148,
      "learning_rate": 3.1949560528403487e-06,
      "loss": 0.2724,
      "step": 1440
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.816901850688223,
      "learning_rate": 3.1883593686803244e-06,
      "loss": 0.2874,
      "step": 1441
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7506813218463502,
      "learning_rate": 3.1817663125077534e-06,
      "loss": 0.2776,
      "step": 1442
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7533311888667901,
      "learning_rate": 3.1751768975258745e-06,
      "loss": 0.27,
      "step": 1443
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7929206115792868,
      "learning_rate": 3.1685911369306364e-06,
      "loss": 0.3212,
      "step": 1444
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.751814822281902,
      "learning_rate": 3.162009043910668e-06,
      "loss": 0.2965,
      "step": 1445
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7588005759172233,
      "learning_rate": 3.155430631647254e-06,
      "loss": 0.2964,
      "step": 1446
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7472991820163527,
      "learning_rate": 3.1488559133143065e-06,
      "loss": 0.2728,
      "step": 1447
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.713811625413155,
      "learning_rate": 3.1422849020783442e-06,
      "loss": 0.2756,
      "step": 1448
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7693991757753348,
      "learning_rate": 3.1357176110984578e-06,
      "loss": 0.3057,
      "step": 1449
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7406727650486468,
      "learning_rate": 3.1291540535262905e-06,
      "loss": 0.2586,
      "step": 1450
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7594525720821216,
      "learning_rate": 3.122594242506006e-06,
      "loss": 0.2744,
      "step": 1451
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.81978484599436,
      "learning_rate": 3.11603819117427e-06,
      "loss": 0.2854,
      "step": 1452
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7361554080805606,
      "learning_rate": 3.1094859126602132e-06,
      "loss": 0.255,
      "step": 1453
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8110841784411963,
      "learning_rate": 3.1029374200854167e-06,
      "loss": 0.3009,
      "step": 1454
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7507700441812762,
      "learning_rate": 3.096392726563874e-06,
      "loss": 0.254,
      "step": 1455
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.774721271394725,
      "learning_rate": 3.089851845201976e-06,
      "loss": 0.297,
      "step": 1456
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8070014836561342,
      "learning_rate": 3.0833147890984735e-06,
      "loss": 0.2722,
      "step": 1457
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7527903487202354,
      "learning_rate": 3.0767815713444655e-06,
      "loss": 0.2646,
      "step": 1458
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7443587789205496,
      "learning_rate": 3.070252205023356e-06,
      "loss": 0.2794,
      "step": 1459
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7542765317779386,
      "learning_rate": 3.063726703210841e-06,
      "loss": 0.2541,
      "step": 1460
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7960067947569965,
      "learning_rate": 3.0572050789748732e-06,
      "loss": 0.2886,
      "step": 1461
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7630752438741694,
      "learning_rate": 3.050687345375645e-06,
      "loss": 0.2685,
      "step": 1462
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7523323663335508,
      "learning_rate": 3.044173515465554e-06,
      "loss": 0.2574,
      "step": 1463
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7426650343634806,
      "learning_rate": 3.0376636022891813e-06,
      "loss": 0.2959,
      "step": 1464
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7137300330696793,
      "learning_rate": 3.031157618883265e-06,
      "loss": 0.2338,
      "step": 1465
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7320151678619478,
      "learning_rate": 3.0246555782766723e-06,
      "loss": 0.2933,
      "step": 1466
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7874245157729821,
      "learning_rate": 3.0181574934903745e-06,
      "loss": 0.2708,
      "step": 1467
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7940763416428638,
      "learning_rate": 3.0116633775374237e-06,
      "loss": 0.2963,
      "step": 1468
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7975105812548973,
      "learning_rate": 3.0051732434229185e-06,
      "loss": 0.3151,
      "step": 1469
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.757668679151663,
      "learning_rate": 2.998687104143991e-06,
      "loss": 0.2717,
      "step": 1470
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7883798397355029,
      "learning_rate": 2.992204972689765e-06,
      "loss": 0.3015,
      "step": 1471
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.736557528278319,
      "learning_rate": 2.985726862041346e-06,
      "loss": 0.2849,
      "step": 1472
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8427901893446652,
      "learning_rate": 2.9792527851717805e-06,
      "loss": 0.2724,
      "step": 1473
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7523137255542257,
      "learning_rate": 2.972782755046043e-06,
      "loss": 0.2847,
      "step": 1474
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7868910139484604,
      "learning_rate": 2.966316784621e-06,
      "loss": 0.2768,
      "step": 1475
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7316607044784842,
      "learning_rate": 2.9598548868453904e-06,
      "loss": 0.2807,
      "step": 1476
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7702823847780866,
      "learning_rate": 2.9533970746597935e-06,
      "loss": 0.3103,
      "step": 1477
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.802957526569481,
      "learning_rate": 2.9469433609966125e-06,
      "loss": 0.267,
      "step": 1478
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7693523121734552,
      "learning_rate": 2.9404937587800374e-06,
      "loss": 0.293,
      "step": 1479
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7438060154347975,
      "learning_rate": 2.9340482809260294e-06,
      "loss": 0.2623,
      "step": 1480
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.767856947123154,
      "learning_rate": 2.9276069403422846e-06,
      "loss": 0.2676,
      "step": 1481
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7725787714135133,
      "learning_rate": 2.9211697499282223e-06,
      "loss": 0.2817,
      "step": 1482
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7541336355929656,
      "learning_rate": 2.9147367225749425e-06,
      "loss": 0.2727,
      "step": 1483
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.810694538696133,
      "learning_rate": 2.90830787116521e-06,
      "loss": 0.2762,
      "step": 1484
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7844840626591026,
      "learning_rate": 2.90188320857343e-06,
      "loss": 0.2985,
      "step": 1485
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7780702146558434,
      "learning_rate": 2.895462747665617e-06,
      "loss": 0.2669,
      "step": 1486
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.876115832807721,
      "learning_rate": 2.889046501299373e-06,
      "loss": 0.2955,
      "step": 1487
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.794550476071244,
      "learning_rate": 2.882634482323856e-06,
      "loss": 0.2669,
      "step": 1488
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7614265452542902,
      "learning_rate": 2.8762267035797607e-06,
      "loss": 0.2903,
      "step": 1489
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8071779009663055,
      "learning_rate": 2.8698231778992925e-06,
      "loss": 0.2609,
      "step": 1490
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7324088091034475,
      "learning_rate": 2.8634239181061384e-06,
      "loss": 0.2727,
      "step": 1491
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.827775370979292,
      "learning_rate": 2.857028937015438e-06,
      "loss": 0.2915,
      "step": 1492
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8382374232015088,
      "learning_rate": 2.850638247433768e-06,
      "loss": 0.3077,
      "step": 1493
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8053639499843955,
      "learning_rate": 2.8442518621591085e-06,
      "loss": 0.2777,
      "step": 1494
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7065394589470531,
      "learning_rate": 2.837869793980823e-06,
      "loss": 0.2432,
      "step": 1495
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7502783886397508,
      "learning_rate": 2.831492055679622e-06,
      "loss": 0.3007,
      "step": 1496
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7936465025738403,
      "learning_rate": 2.8251186600275533e-06,
      "loss": 0.2638,
      "step": 1497
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7696281493273076,
      "learning_rate": 2.8187496197879637e-06,
      "loss": 0.2974,
      "step": 1498
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7597574577957912,
      "learning_rate": 2.8123849477154808e-06,
      "loss": 0.2592,
      "step": 1499
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7330754455804657,
      "learning_rate": 2.806024656555978e-06,
      "loss": 0.276,
      "step": 1500
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7612924431260562,
      "learning_rate": 2.799668759046564e-06,
      "loss": 0.2875,
      "step": 1501
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7548953749239738,
      "learning_rate": 2.7933172679155427e-06,
      "loss": 0.2654,
      "step": 1502
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7578693701916017,
      "learning_rate": 2.786970195882398e-06,
      "loss": 0.2855,
      "step": 1503
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7368215718287692,
      "learning_rate": 2.7806275556577624e-06,
      "loss": 0.2701,
      "step": 1504
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7526900511576837,
      "learning_rate": 2.7742893599433875e-06,
      "loss": 0.2896,
      "step": 1505
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7731041928050081,
      "learning_rate": 2.7679556214321357e-06,
      "loss": 0.2684,
      "step": 1506
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7615060365701503,
      "learning_rate": 2.761626352807938e-06,
      "loss": 0.2594,
      "step": 1507
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8000685966808864,
      "learning_rate": 2.7553015667457694e-06,
      "loss": 0.2827,
      "step": 1508
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7910030861892312,
      "learning_rate": 2.748981275911633e-06,
      "loss": 0.2919,
      "step": 1509
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7715867923641077,
      "learning_rate": 2.7426654929625307e-06,
      "loss": 0.2744,
      "step": 1510
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7647056801228633,
      "learning_rate": 2.736354230546437e-06,
      "loss": 0.2658,
      "step": 1511
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7854176457880635,
      "learning_rate": 2.7300475013022666e-06,
      "loss": 0.273,
      "step": 1512
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8864982183272028,
      "learning_rate": 2.7237453178598645e-06,
      "loss": 0.2989,
      "step": 1513
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8581362427714041,
      "learning_rate": 2.7174476928399685e-06,
      "loss": 0.2976,
      "step": 1514
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7668970548724269,
      "learning_rate": 2.71115463885419e-06,
      "loss": 0.2741,
      "step": 1515
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7476175477147188,
      "learning_rate": 2.7048661685049814e-06,
      "loss": 0.2846,
      "step": 1516
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7623079279701871,
      "learning_rate": 2.698582294385621e-06,
      "loss": 0.2494,
      "step": 1517
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7644175440483688,
      "learning_rate": 2.6923030290801817e-06,
      "loss": 0.2874,
      "step": 1518
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8016057578441708,
      "learning_rate": 2.6860283851635067e-06,
      "loss": 0.2773,
      "step": 1519
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8216704953862317,
      "learning_rate": 2.679758375201181e-06,
      "loss": 0.2957,
      "step": 1520
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7895168375844376,
      "learning_rate": 2.6734930117495133e-06,
      "loss": 0.2724,
      "step": 1521
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7841743457920944,
      "learning_rate": 2.6672323073555074e-06,
      "loss": 0.2922,
      "step": 1522
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7454968451843106,
      "learning_rate": 2.6609762745568358e-06,
      "loss": 0.2575,
      "step": 1523
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7625677090145697,
      "learning_rate": 2.6547249258818162e-06,
      "loss": 0.2856,
      "step": 1524
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7678226295399453,
      "learning_rate": 2.6484782738493843e-06,
      "loss": 0.2669,
      "step": 1525
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7564943988592681,
      "learning_rate": 2.6422363309690702e-06,
      "loss": 0.2732,
      "step": 1526
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.758808076156311,
      "learning_rate": 2.6359991097409767e-06,
      "loss": 0.2783,
      "step": 1527
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7924188667452507,
      "learning_rate": 2.6297666226557505e-06,
      "loss": 0.3171,
      "step": 1528
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7751543623283452,
      "learning_rate": 2.6235388821945497e-06,
      "loss": 0.2743,
      "step": 1529
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7629271501105163,
      "learning_rate": 2.6173159008290407e-06,
      "loss": 0.269,
      "step": 1530
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7343317699884891,
      "learning_rate": 2.61109769102135e-06,
      "loss": 0.2617,
      "step": 1531
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7641537171799322,
      "learning_rate": 2.604884265224048e-06,
      "loss": 0.2708,
      "step": 1532
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7736126048801671,
      "learning_rate": 2.598675635880129e-06,
      "loss": 0.2832,
      "step": 1533
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8316895759919394,
      "learning_rate": 2.59247181542298e-06,
      "loss": 0.2939,
      "step": 1534
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7979517747788513,
      "learning_rate": 2.5862728162763595e-06,
      "loss": 0.2949,
      "step": 1535
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7769076622227137,
      "learning_rate": 2.580078650854366e-06,
      "loss": 0.2598,
      "step": 1536
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7739767247894968,
      "learning_rate": 2.573889331561423e-06,
      "loss": 0.2701,
      "step": 1537
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7902247200858955,
      "learning_rate": 2.5677048707922482e-06,
      "loss": 0.2678,
      "step": 1538
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7938139914589235,
      "learning_rate": 2.5615252809318287e-06,
      "loss": 0.2929,
      "step": 1539
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7582359686373125,
      "learning_rate": 2.555350574355395e-06,
      "loss": 0.2758,
      "step": 1540
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7613801640780713,
      "learning_rate": 2.5491807634284016e-06,
      "loss": 0.2655,
      "step": 1541
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8324400469125435,
      "learning_rate": 2.543015860506499e-06,
      "loss": 0.3079,
      "step": 1542
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7924073966596439,
      "learning_rate": 2.536855877935507e-06,
      "loss": 0.2769,
      "step": 1543
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7737722492813376,
      "learning_rate": 2.5307008280513956e-06,
      "loss": 0.2659,
      "step": 1544
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7486080695320196,
      "learning_rate": 2.524550723180249e-06,
      "loss": 0.2809,
      "step": 1545
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8174989154615325,
      "learning_rate": 2.5184055756382566e-06,
      "loss": 0.2882,
      "step": 1546
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8207184405720027,
      "learning_rate": 2.5122653977316758e-06,
      "loss": 0.2881,
      "step": 1547
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7370464905611824,
      "learning_rate": 2.5061302017568153e-06,
      "loss": 0.2627,
      "step": 1548
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7135921515343453,
      "learning_rate": 2.5000000000000015e-06,
      "loss": 0.2467,
      "step": 1549
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7363509389053651,
      "learning_rate": 2.4938748047375637e-06,
      "loss": 0.2729,
      "step": 1550
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7530626815866382,
      "learning_rate": 2.487754628235805e-06,
      "loss": 0.2975,
      "step": 1551
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7905014397419904,
      "learning_rate": 2.481639482750979e-06,
      "loss": 0.2784,
      "step": 1552
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.801534345069752,
      "learning_rate": 2.4755293805292572e-06,
      "loss": 0.2819,
      "step": 1553
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7325452268439524,
      "learning_rate": 2.469424333806718e-06,
      "loss": 0.2548,
      "step": 1554
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7781300580429674,
      "learning_rate": 2.4633243548093195e-06,
      "loss": 0.2843,
      "step": 1555
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7654916622144602,
      "learning_rate": 2.4572294557528604e-06,
      "loss": 0.2716,
      "step": 1556
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8129760556636138,
      "learning_rate": 2.4511396488429727e-06,
      "loss": 0.3115,
      "step": 1557
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7405923320495612,
      "learning_rate": 2.4450549462750917e-06,
      "loss": 0.2792,
      "step": 1558
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.784212285256923,
      "learning_rate": 2.4389753602344298e-06,
      "loss": 0.2784,
      "step": 1559
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7510334652811116,
      "learning_rate": 2.4329009028959484e-06,
      "loss": 0.2756,
      "step": 1560
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8174612547152958,
      "learning_rate": 2.4268315864243447e-06,
      "loss": 0.3048,
      "step": 1561
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8007258869311775,
      "learning_rate": 2.420767422974018e-06,
      "loss": 0.2822,
      "step": 1562
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7438795677734192,
      "learning_rate": 2.4147084246890478e-06,
      "loss": 0.2775,
      "step": 1563
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7632070193208693,
      "learning_rate": 2.4086546037031734e-06,
      "loss": 0.2871,
      "step": 1564
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.801194749646044,
      "learning_rate": 2.402605972139759e-06,
      "loss": 0.2957,
      "step": 1565
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7763852742864334,
      "learning_rate": 2.396562542111781e-06,
      "loss": 0.2913,
      "step": 1566
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7517205713605628,
      "learning_rate": 2.3905243257218007e-06,
      "loss": 0.2749,
      "step": 1567
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8387789596135893,
      "learning_rate": 2.3844913350619382e-06,
      "loss": 0.2861,
      "step": 1568
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7741470327043375,
      "learning_rate": 2.3784635822138424e-06,
      "loss": 0.2964,
      "step": 1569
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7593140216018964,
      "learning_rate": 2.37244107924868e-06,
      "loss": 0.2878,
      "step": 1570
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8215883025630957,
      "learning_rate": 2.366423838227102e-06,
      "loss": 0.3092,
      "step": 1571
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7599301075049097,
      "learning_rate": 2.3604118711992237e-06,
      "loss": 0.2469,
      "step": 1572
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7615188690717493,
      "learning_rate": 2.354405190204592e-06,
      "loss": 0.2815,
      "step": 1573
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7658222167634471,
      "learning_rate": 2.348403807272176e-06,
      "loss": 0.2581,
      "step": 1574
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7647244853240786,
      "learning_rate": 2.342407734420331e-06,
      "loss": 0.2709,
      "step": 1575
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7577846337527439,
      "learning_rate": 2.3364169836567805e-06,
      "loss": 0.28,
      "step": 1576
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7726492096034634,
      "learning_rate": 2.3304315669785853e-06,
      "loss": 0.2757,
      "step": 1577
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7755981865649162,
      "learning_rate": 2.3244514963721276e-06,
      "loss": 0.2764,
      "step": 1578
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8308102580625603,
      "learning_rate": 2.318476783813088e-06,
      "loss": 0.2972,
      "step": 1579
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7708906652044489,
      "learning_rate": 2.3125074412664085e-06,
      "loss": 0.2989,
      "step": 1580
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7090014616330524,
      "learning_rate": 2.306543480686281e-06,
      "loss": 0.252,
      "step": 1581
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7047339846163643,
      "learning_rate": 2.30058491401612e-06,
      "loss": 0.26,
      "step": 1582
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7763430772749095,
      "learning_rate": 2.294631753188541e-06,
      "loss": 0.2662,
      "step": 1583
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8137922540471979,
      "learning_rate": 2.288684010125325e-06,
      "loss": 0.2736,
      "step": 1584
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.765425069148542,
      "learning_rate": 2.2827416967374115e-06,
      "loss": 0.2838,
      "step": 1585
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7136507109651902,
      "learning_rate": 2.2768048249248648e-06,
      "loss": 0.2577,
      "step": 1586
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7975324999064268,
      "learning_rate": 2.270873406576849e-06,
      "loss": 0.2784,
      "step": 1587
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8198055998745547,
      "learning_rate": 2.2649474535716128e-06,
      "loss": 0.3106,
      "step": 1588
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7488167585555587,
      "learning_rate": 2.2590269777764516e-06,
      "loss": 0.2783,
      "step": 1589
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7068011653992272,
      "learning_rate": 2.2531119910476994e-06,
      "loss": 0.2654,
      "step": 1590
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.76219217545605,
      "learning_rate": 2.247202505230695e-06,
      "loss": 0.2765,
      "step": 1591
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8386016142100573,
      "learning_rate": 2.241298532159763e-06,
      "loss": 0.2901,
      "step": 1592
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7849373041580059,
      "learning_rate": 2.2354000836581834e-06,
      "loss": 0.2708,
      "step": 1593
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.780300317072106,
      "learning_rate": 2.229507171538178e-06,
      "loss": 0.2547,
      "step": 1594
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8192450329232029,
      "learning_rate": 2.223619807600878e-06,
      "loss": 0.2664,
      "step": 1595
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7399928288044658,
      "learning_rate": 2.217738003636309e-06,
      "loss": 0.2558,
      "step": 1596
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7616699230722795,
      "learning_rate": 2.2118617714233524e-06,
      "loss": 0.2849,
      "step": 1597
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7189002535184756,
      "learning_rate": 2.205991122729742e-06,
      "loss": 0.2678,
      "step": 1598
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7423741855714601,
      "learning_rate": 2.2001260693120236e-06,
      "loss": 0.2611,
      "step": 1599
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7662702130896291,
      "learning_rate": 2.1942666229155424e-06,
      "loss": 0.2617,
      "step": 1600
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7650610568061111,
      "learning_rate": 2.1884127952744095e-06,
      "loss": 0.2713,
      "step": 1601
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.731983738001859,
      "learning_rate": 2.1825645981114862e-06,
      "loss": 0.2761,
      "step": 1602
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6828562569992563,
      "learning_rate": 2.1767220431383652e-06,
      "loss": 0.2264,
      "step": 1603
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8119068043721541,
      "learning_rate": 2.1708851420553277e-06,
      "loss": 0.2835,
      "step": 1604
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7818409341718692,
      "learning_rate": 2.1650539065513415e-06,
      "loss": 0.2825,
      "step": 1605
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7580855064008305,
      "learning_rate": 2.159228348304025e-06,
      "loss": 0.2727,
      "step": 1606
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7632401186516631,
      "learning_rate": 2.153408478979628e-06,
      "loss": 0.2782,
      "step": 1607
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7721377735284007,
      "learning_rate": 2.147594310233009e-06,
      "loss": 0.2639,
      "step": 1608
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7049322455024499,
      "learning_rate": 2.141785853707607e-06,
      "loss": 0.2604,
      "step": 1609
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.797115488479798,
      "learning_rate": 2.1359831210354247e-06,
      "loss": 0.2965,
      "step": 1610
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7640559982381862,
      "learning_rate": 2.130186123837002e-06,
      "loss": 0.2725,
      "step": 1611
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7539724850186016,
      "learning_rate": 2.124394873721394e-06,
      "loss": 0.2845,
      "step": 1612
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8042624514094923,
      "learning_rate": 2.118609382286142e-06,
      "loss": 0.2636,
      "step": 1613
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7931114095963266,
      "learning_rate": 2.1128296611172593e-06,
      "loss": 0.2916,
      "step": 1614
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7595796432227941,
      "learning_rate": 2.107055721789203e-06,
      "loss": 0.2779,
      "step": 1615
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7566378296581672,
      "learning_rate": 2.101287575864853e-06,
      "loss": 0.2804,
      "step": 1616
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.749155042921936,
      "learning_rate": 2.0955252348954806e-06,
      "loss": 0.2509,
      "step": 1617
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7619302355811478,
      "learning_rate": 2.08976871042074e-06,
      "loss": 0.2788,
      "step": 1618
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8242944033463034,
      "learning_rate": 2.0840180139686333e-06,
      "loss": 0.2866,
      "step": 1619
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7615145352767886,
      "learning_rate": 2.0782731570554946e-06,
      "loss": 0.2668,
      "step": 1620
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7912165171778326,
      "learning_rate": 2.0725341511859576e-06,
      "loss": 0.2784,
      "step": 1621
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8298862388881647,
      "learning_rate": 2.066801007852944e-06,
      "loss": 0.2808,
      "step": 1622
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7714944003028738,
      "learning_rate": 2.061073738537635e-06,
      "loss": 0.2686,
      "step": 1623
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7842307356652318,
      "learning_rate": 2.0553523547094473e-06,
      "loss": 0.2817,
      "step": 1624
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7892733752048692,
      "learning_rate": 2.0496368678260094e-06,
      "loss": 0.2891,
      "step": 1625
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8509095842440997,
      "learning_rate": 2.043927289333141e-06,
      "loss": 0.2918,
      "step": 1626
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7748708260314136,
      "learning_rate": 2.038223630664835e-06,
      "loss": 0.2708,
      "step": 1627
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7993865118393632,
      "learning_rate": 2.0325259032432264e-06,
      "loss": 0.2961,
      "step": 1628
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7513179323181158,
      "learning_rate": 2.0268341184785674e-06,
      "loss": 0.2818,
      "step": 1629
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7732210265496705,
      "learning_rate": 2.021148287769214e-06,
      "loss": 0.2692,
      "step": 1630
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8007760796925063,
      "learning_rate": 2.015468422501599e-06,
      "loss": 0.2957,
      "step": 1631
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7721658648166554,
      "learning_rate": 2.00979453405021e-06,
      "loss": 0.277,
      "step": 1632
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7662265596720798,
      "learning_rate": 2.004126633777559e-06,
      "loss": 0.2689,
      "step": 1633
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7733465173520869,
      "learning_rate": 1.998464733034172e-06,
      "loss": 0.2889,
      "step": 1634
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7917863072309378,
      "learning_rate": 1.992808843158559e-06,
      "loss": 0.2921,
      "step": 1635
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8377765365967087,
      "learning_rate": 1.987158975477194e-06,
      "loss": 0.2826,
      "step": 1636
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8061481334016037,
      "learning_rate": 1.9815151413044865e-06,
      "loss": 0.2808,
      "step": 1637
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7592828376245938,
      "learning_rate": 1.9758773519427664e-06,
      "loss": 0.2936,
      "step": 1638
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7385249194636658,
      "learning_rate": 1.9702456186822595e-06,
      "loss": 0.2434,
      "step": 1639
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.768560888935936,
      "learning_rate": 1.9646199528010625e-06,
      "loss": 0.2707,
      "step": 1640
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7056633103614284,
      "learning_rate": 1.959000365565119e-06,
      "loss": 0.2582,
      "step": 1641
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7945928977034089,
      "learning_rate": 1.9533868682282026e-06,
      "loss": 0.2857,
      "step": 1642
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8026050165466926,
      "learning_rate": 1.947779472031891e-06,
      "loss": 0.2877,
      "step": 1643
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7447641262464846,
      "learning_rate": 1.9421781882055447e-06,
      "loss": 0.2674,
      "step": 1644
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9075271366905479,
      "learning_rate": 1.936583027966279e-06,
      "loss": 0.2679,
      "step": 1645
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7966042133184701,
      "learning_rate": 1.93099400251895e-06,
      "loss": 0.2775,
      "step": 1646
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7969361301546982,
      "learning_rate": 1.9254111230561282e-06,
      "loss": 0.2898,
      "step": 1647
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7641724625995733,
      "learning_rate": 1.9198344007580776e-06,
      "loss": 0.2579,
      "step": 1648
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7874846826162981,
      "learning_rate": 1.9142638467927254e-06,
      "loss": 0.3003,
      "step": 1649
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7371742689426259,
      "learning_rate": 1.908699472315651e-06,
      "loss": 0.2671,
      "step": 1650
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7601630446866584,
      "learning_rate": 1.9031412884700612e-06,
      "loss": 0.2765,
      "step": 1651
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7355083405399186,
      "learning_rate": 1.8975893063867635e-06,
      "loss": 0.2747,
      "step": 1652
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6898761094140462,
      "learning_rate": 1.8920435371841394e-06,
      "loss": 0.2506,
      "step": 1653
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.749213902886905,
      "learning_rate": 1.8865039919681377e-06,
      "loss": 0.2855,
      "step": 1654
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.772513428298872,
      "learning_rate": 1.8809706818322375e-06,
      "loss": 0.2737,
      "step": 1655
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7574035686541581,
      "learning_rate": 1.8754436178574358e-06,
      "loss": 0.263,
      "step": 1656
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7710047997468473,
      "learning_rate": 1.8699228111122146e-06,
      "loss": 0.2819,
      "step": 1657
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7459999511163288,
      "learning_rate": 1.8644082726525313e-06,
      "loss": 0.2675,
      "step": 1658
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7854474111111044,
      "learning_rate": 1.8589000135217882e-06,
      "loss": 0.2921,
      "step": 1659
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8055502281605319,
      "learning_rate": 1.8533980447508138e-06,
      "loss": 0.2763,
      "step": 1660
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7195657381181403,
      "learning_rate": 1.8479023773578354e-06,
      "loss": 0.2728,
      "step": 1661
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7703528631894458,
      "learning_rate": 1.8424130223484655e-06,
      "loss": 0.2773,
      "step": 1662
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7950103026336739,
      "learning_rate": 1.8369299907156745e-06,
      "loss": 0.2828,
      "step": 1663
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.75983959377857,
      "learning_rate": 1.831453293439771e-06,
      "loss": 0.2807,
      "step": 1664
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.800578378535595,
      "learning_rate": 1.8259829414883728e-06,
      "loss": 0.2987,
      "step": 1665
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7721190561570801,
      "learning_rate": 1.8205189458163964e-06,
      "loss": 0.2815,
      "step": 1666
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8081314887340981,
      "learning_rate": 1.815061317366027e-06,
      "loss": 0.2712,
      "step": 1667
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7506433210686816,
      "learning_rate": 1.809610067066701e-06,
      "loss": 0.273,
      "step": 1668
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7819211641899507,
      "learning_rate": 1.8041652058350768e-06,
      "loss": 0.2869,
      "step": 1669
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7865319781411564,
      "learning_rate": 1.7987267445750224e-06,
      "loss": 0.2906,
      "step": 1670
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.747636966658803,
      "learning_rate": 1.7932946941775882e-06,
      "loss": 0.2757,
      "step": 1671
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.754759376883786,
      "learning_rate": 1.7878690655209862e-06,
      "loss": 0.2785,
      "step": 1672
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7782458350652882,
      "learning_rate": 1.78244986947057e-06,
      "loss": 0.2782,
      "step": 1673
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7474471347025646,
      "learning_rate": 1.7770371168788042e-06,
      "loss": 0.2866,
      "step": 1674
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7562094526586637,
      "learning_rate": 1.7716308185852605e-06,
      "loss": 0.2708,
      "step": 1675
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7525057127324669,
      "learning_rate": 1.7662309854165788e-06,
      "loss": 0.2635,
      "step": 1676
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8547932575690257,
      "learning_rate": 1.7608376281864502e-06,
      "loss": 0.3067,
      "step": 1677
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7688190103129865,
      "learning_rate": 1.755450757695601e-06,
      "loss": 0.279,
      "step": 1678
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7765635007599195,
      "learning_rate": 1.7500703847317663e-06,
      "loss": 0.2507,
      "step": 1679
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7641228906347997,
      "learning_rate": 1.7446965200696702e-06,
      "loss": 0.2827,
      "step": 1680
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7995791747179738,
      "learning_rate": 1.739329174470999e-06,
      "loss": 0.2871,
      "step": 1681
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7715273096213876,
      "learning_rate": 1.733968358684388e-06,
      "loss": 0.2773,
      "step": 1682
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7568544890988609,
      "learning_rate": 1.7286140834453958e-06,
      "loss": 0.282,
      "step": 1683
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7693940309186054,
      "learning_rate": 1.723266359476483e-06,
      "loss": 0.2916,
      "step": 1684
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7814719968926227,
      "learning_rate": 1.7179251974869858e-06,
      "loss": 0.2733,
      "step": 1685
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7701518939552389,
      "learning_rate": 1.7125906081731064e-06,
      "loss": 0.2704,
      "step": 1686
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7952301895273249,
      "learning_rate": 1.7072626022178796e-06,
      "loss": 0.2666,
      "step": 1687
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7990016459629282,
      "learning_rate": 1.7019411902911609e-06,
      "loss": 0.2783,
      "step": 1688
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7468894341678111,
      "learning_rate": 1.6966263830495939e-06,
      "loss": 0.2723,
      "step": 1689
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7609561186335311,
      "learning_rate": 1.6913181911366017e-06,
      "loss": 0.2788,
      "step": 1690
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7762866001322529,
      "learning_rate": 1.6860166251823574e-06,
      "loss": 0.2556,
      "step": 1691
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8275761012587025,
      "learning_rate": 1.6807216958037664e-06,
      "loss": 0.2856,
      "step": 1692
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8097847479859916,
      "learning_rate": 1.6754334136044398e-06,
      "loss": 0.2882,
      "step": 1693
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7971684458259619,
      "learning_rate": 1.6701517891746805e-06,
      "loss": 0.3046,
      "step": 1694
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7214751826365752,
      "learning_rate": 1.6648768330914578e-06,
      "loss": 0.2644,
      "step": 1695
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7909528600399285,
      "learning_rate": 1.6596085559183872e-06,
      "loss": 0.2655,
      "step": 1696
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7634927812185448,
      "learning_rate": 1.6543469682057105e-06,
      "loss": 0.2817,
      "step": 1697
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7398179034452677,
      "learning_rate": 1.649092080490266e-06,
      "loss": 0.2575,
      "step": 1698
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7596936532539,
      "learning_rate": 1.6438439032954857e-06,
      "loss": 0.2886,
      "step": 1699
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7249997033266135,
      "learning_rate": 1.6386024471313561e-06,
      "loss": 0.2627,
      "step": 1700
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8207976303168232,
      "learning_rate": 1.633367722494404e-06,
      "loss": 0.2789,
      "step": 1701
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.73980059086777,
      "learning_rate": 1.6281397398676774e-06,
      "loss": 0.2779,
      "step": 1702
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7276381936179099,
      "learning_rate": 1.6229185097207228e-06,
      "loss": 0.2757,
      "step": 1703
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7914434826080745,
      "learning_rate": 1.6177040425095664e-06,
      "loss": 0.2895,
      "step": 1704
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7547582711809696,
      "learning_rate": 1.6124963486766844e-06,
      "loss": 0.2876,
      "step": 1705
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7312704604505884,
      "learning_rate": 1.6072954386509948e-06,
      "loss": 0.2778,
      "step": 1706
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7434682794191925,
      "learning_rate": 1.6021013228478277e-06,
      "loss": 0.2644,
      "step": 1707
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8364926343715718,
      "learning_rate": 1.5969140116689102e-06,
      "loss": 0.2842,
      "step": 1708
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7560250009854409,
      "learning_rate": 1.5917335155023368e-06,
      "loss": 0.2666,
      "step": 1709
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7402204283098273,
      "learning_rate": 1.5865598447225576e-06,
      "loss": 0.2918,
      "step": 1710
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7546376469987476,
      "learning_rate": 1.5813930096903562e-06,
      "loss": 0.2597,
      "step": 1711
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7772036661210319,
      "learning_rate": 1.5762330207528249e-06,
      "loss": 0.2765,
      "step": 1712
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7460151313407648,
      "learning_rate": 1.5710798882433432e-06,
      "loss": 0.271,
      "step": 1713
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7366893242735868,
      "learning_rate": 1.5659336224815642e-06,
      "loss": 0.2705,
      "step": 1714
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7954042820283579,
      "learning_rate": 1.5607942337733884e-06,
      "loss": 0.277,
      "step": 1715
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7343650799356555,
      "learning_rate": 1.555661732410944e-06,
      "loss": 0.2568,
      "step": 1716
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7608126470204042,
      "learning_rate": 1.5505361286725679e-06,
      "loss": 0.2764,
      "step": 1717
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8299809412791321,
      "learning_rate": 1.5454174328227794e-06,
      "loss": 0.2932,
      "step": 1718
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7496110423027263,
      "learning_rate": 1.5403056551122697e-06,
      "loss": 0.255,
      "step": 1719
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8074827944887215,
      "learning_rate": 1.5352008057778723e-06,
      "loss": 0.2945,
      "step": 1720
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7310519480888766,
      "learning_rate": 1.5301028950425494e-06,
      "loss": 0.2543,
      "step": 1721
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7468625611309215,
      "learning_rate": 1.5250119331153606e-06,
      "loss": 0.2709,
      "step": 1722
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7585058846217572,
      "learning_rate": 1.5199279301914592e-06,
      "loss": 0.2605,
      "step": 1723
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.763723038023221,
      "learning_rate": 1.5148508964520586e-06,
      "loss": 0.2899,
      "step": 1724
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7402697501126515,
      "learning_rate": 1.5097808420644117e-06,
      "loss": 0.277,
      "step": 1725
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7743701756398587,
      "learning_rate": 1.5047177771817989e-06,
      "loss": 0.2761,
      "step": 1726
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.772911965807274,
      "learning_rate": 1.4996617119435036e-06,
      "loss": 0.2813,
      "step": 1727
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7325685759325268,
      "learning_rate": 1.494612656474792e-06,
      "loss": 0.2535,
      "step": 1728
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7524733440656881,
      "learning_rate": 1.4895706208868876e-06,
      "loss": 0.2737,
      "step": 1729
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7315664621360706,
      "learning_rate": 1.4845356152769618e-06,
      "loss": 0.2759,
      "step": 1730
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8095413108511555,
      "learning_rate": 1.4795076497281052e-06,
      "loss": 0.2629,
      "step": 1731
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7969525143777143,
      "learning_rate": 1.4744867343093112e-06,
      "loss": 0.2669,
      "step": 1732
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8074325632982164,
      "learning_rate": 1.4694728790754514e-06,
      "loss": 0.2924,
      "step": 1733
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.78822110317636,
      "learning_rate": 1.4644660940672628e-06,
      "loss": 0.26,
      "step": 1734
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7241675400057549,
      "learning_rate": 1.459466389311321e-06,
      "loss": 0.2687,
      "step": 1735
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7768027466831612,
      "learning_rate": 1.4544737748200238e-06,
      "loss": 0.2649,
      "step": 1736
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7726248653314839,
      "learning_rate": 1.4494882605915717e-06,
      "loss": 0.2736,
      "step": 1737
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7715041242666287,
      "learning_rate": 1.4445098566099409e-06,
      "loss": 0.2738,
      "step": 1738
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7658473533537563,
      "learning_rate": 1.439538572844873e-06,
      "loss": 0.2808,
      "step": 1739
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7593957679539316,
      "learning_rate": 1.4345744192518507e-06,
      "loss": 0.2746,
      "step": 1740
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7711493294738938,
      "learning_rate": 1.4296174057720775e-06,
      "loss": 0.2654,
      "step": 1741
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7305658315047187,
      "learning_rate": 1.4246675423324546e-06,
      "loss": 0.252,
      "step": 1742
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8029496088874257,
      "learning_rate": 1.4197248388455693e-06,
      "loss": 0.296,
      "step": 1743
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7290998922219774,
      "learning_rate": 1.4147893052096684e-06,
      "loss": 0.2791,
      "step": 1744
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7446791133937168,
      "learning_rate": 1.4098609513086414e-06,
      "loss": 0.2709,
      "step": 1745
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7525045019673389,
      "learning_rate": 1.4049397870119962e-06,
      "loss": 0.2552,
      "step": 1746
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7484006922241091,
      "learning_rate": 1.4000258221748452e-06,
      "loss": 0.2679,
      "step": 1747
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7305223449084324,
      "learning_rate": 1.3951190666378884e-06,
      "loss": 0.2734,
      "step": 1748
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7416989966203232,
      "learning_rate": 1.390219530227378e-06,
      "loss": 0.2539,
      "step": 1749
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.731922226958195,
      "learning_rate": 1.3853272227551167e-06,
      "loss": 0.2657,
      "step": 1750
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7456253703750415,
      "learning_rate": 1.3804421540184282e-06,
      "loss": 0.2614,
      "step": 1751
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7348127930295889,
      "learning_rate": 1.3755643338001411e-06,
      "loss": 0.2648,
      "step": 1752
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7072117644013031,
      "learning_rate": 1.3706937718685654e-06,
      "loss": 0.2551,
      "step": 1753
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7493178842833728,
      "learning_rate": 1.3658304779774784e-06,
      "loss": 0.2988,
      "step": 1754
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7338397725846229,
      "learning_rate": 1.3609744618661013e-06,
      "loss": 0.2565,
      "step": 1755
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7743655771747295,
      "learning_rate": 1.3561257332590816e-06,
      "loss": 0.2737,
      "step": 1756
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.76529824683644,
      "learning_rate": 1.3512843018664728e-06,
      "loss": 0.2702,
      "step": 1757
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8100455423318919,
      "learning_rate": 1.3464501773837124e-06,
      "loss": 0.2856,
      "step": 1758
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7427354578125572,
      "learning_rate": 1.3416233694916086e-06,
      "loss": 0.2709,
      "step": 1759
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7629529417092723,
      "learning_rate": 1.3368038878563154e-06,
      "loss": 0.2837,
      "step": 1760
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7043626458936962,
      "learning_rate": 1.3319917421293182e-06,
      "loss": 0.2435,
      "step": 1761
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7105179424918625,
      "learning_rate": 1.3271869419474059e-06,
      "loss": 0.2587,
      "step": 1762
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7606389028313847,
      "learning_rate": 1.3223894969326623e-06,
      "loss": 0.2681,
      "step": 1763
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.736902110633843,
      "learning_rate": 1.3175994166924394e-06,
      "loss": 0.2652,
      "step": 1764
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7390421273508612,
      "learning_rate": 1.312816710819343e-06,
      "loss": 0.2693,
      "step": 1765
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7663831652577159,
      "learning_rate": 1.3080413888912057e-06,
      "loss": 0.266,
      "step": 1766
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7532800545580999,
      "learning_rate": 1.3032734604710784e-06,
      "loss": 0.2506,
      "step": 1767
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7671390786561625,
      "learning_rate": 1.298512935107204e-06,
      "loss": 0.277,
      "step": 1768
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7869918864487332,
      "learning_rate": 1.2937598223330006e-06,
      "loss": 0.2779,
      "step": 1769
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7566229598521557,
      "learning_rate": 1.289014131667039e-06,
      "loss": 0.2839,
      "step": 1770
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7918953495491338,
      "learning_rate": 1.2842758726130283e-06,
      "loss": 0.2924,
      "step": 1771
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7493724005829686,
      "learning_rate": 1.2795450546597999e-06,
      "loss": 0.2773,
      "step": 1772
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7455453741035084,
      "learning_rate": 1.2748216872812747e-06,
      "loss": 0.2598,
      "step": 1773
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7597336749278539,
      "learning_rate": 1.2701057799364591e-06,
      "loss": 0.278,
      "step": 1774
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7945452318474064,
      "learning_rate": 1.2653973420694182e-06,
      "loss": 0.2816,
      "step": 1775
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7502266454941833,
      "learning_rate": 1.26069638310926e-06,
      "loss": 0.271,
      "step": 1776
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.815499713754788,
      "learning_rate": 1.256002912470115e-06,
      "loss": 0.2884,
      "step": 1777
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7503011370324022,
      "learning_rate": 1.2513169395511143e-06,
      "loss": 0.2658,
      "step": 1778
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7496604785944369,
      "learning_rate": 1.246638473736378e-06,
      "loss": 0.2645,
      "step": 1779
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7894862576649267,
      "learning_rate": 1.2419675243949914e-06,
      "loss": 0.2741,
      "step": 1780
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7625409038883164,
      "learning_rate": 1.2373041008809888e-06,
      "loss": 0.2863,
      "step": 1781
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7028467892508579,
      "learning_rate": 1.2326482125333284e-06,
      "loss": 0.2584,
      "step": 1782
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7580508402576086,
      "learning_rate": 1.2279998686758833e-06,
      "loss": 0.2895,
      "step": 1783
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7002197129952317,
      "learning_rate": 1.223359078617416e-06,
      "loss": 0.2659,
      "step": 1784
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7340537240110445,
      "learning_rate": 1.2187258516515644e-06,
      "loss": 0.2607,
      "step": 1785
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7561394679461831,
      "learning_rate": 1.214100197056815e-06,
      "loss": 0.2719,
      "step": 1786
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.759118818592083,
      "learning_rate": 1.2094821240964955e-06,
      "loss": 0.2603,
      "step": 1787
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7771607337150733,
      "learning_rate": 1.2048716420187473e-06,
      "loss": 0.286,
      "step": 1788
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7173667164823075,
      "learning_rate": 1.2002687600565138e-06,
      "loss": 0.2651,
      "step": 1789
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.762402296585614,
      "learning_rate": 1.1956734874275127e-06,
      "loss": 0.273,
      "step": 1790
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7452789968287707,
      "learning_rate": 1.1910858333342279e-06,
      "loss": 0.2616,
      "step": 1791
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8744675337154127,
      "learning_rate": 1.1865058069638857e-06,
      "loss": 0.3144,
      "step": 1792
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7862423486858641,
      "learning_rate": 1.1819334174884378e-06,
      "loss": 0.303,
      "step": 1793
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7546126666527287,
      "learning_rate": 1.1773686740645384e-06,
      "loss": 0.2772,
      "step": 1794
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8018971965343563,
      "learning_rate": 1.172811585833532e-06,
      "loss": 0.2652,
      "step": 1795
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7809292948861885,
      "learning_rate": 1.168262161921438e-06,
      "loss": 0.2982,
      "step": 1796
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7682089188385242,
      "learning_rate": 1.1637204114389179e-06,
      "loss": 0.2775,
      "step": 1797
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7160005414061417,
      "learning_rate": 1.1591863434812722e-06,
      "loss": 0.249,
      "step": 1798
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7599628857615511,
      "learning_rate": 1.1546599671284158e-06,
      "loss": 0.2885,
      "step": 1799
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7167644710246258,
      "learning_rate": 1.1501412914448596e-06,
      "loss": 0.2492,
      "step": 1800
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7278424372232234,
      "learning_rate": 1.1456303254796941e-06,
      "loss": 0.2493,
      "step": 1801
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7770169252757541,
      "learning_rate": 1.141127078266568e-06,
      "loss": 0.293,
      "step": 1802
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7614541314078548,
      "learning_rate": 1.1366315588236743e-06,
      "loss": 0.2673,
      "step": 1803
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.738452592810774,
      "learning_rate": 1.1321437761537307e-06,
      "loss": 0.2656,
      "step": 1804
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7463835757668856,
      "learning_rate": 1.127663739243962e-06,
      "loss": 0.2741,
      "step": 1805
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8081535734792142,
      "learning_rate": 1.1231914570660774e-06,
      "loss": 0.2747,
      "step": 1806
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7946476061316474,
      "learning_rate": 1.1187269385762605e-06,
      "loss": 0.2861,
      "step": 1807
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7405587702146419,
      "learning_rate": 1.1142701927151456e-06,
      "loss": 0.2716,
      "step": 1808
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7430038773013604,
      "learning_rate": 1.1098212284078037e-06,
      "loss": 0.2668,
      "step": 1809
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7623104295804228,
      "learning_rate": 1.1053800545637173e-06,
      "loss": 0.2944,
      "step": 1810
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7396643934867233,
      "learning_rate": 1.1009466800767738e-06,
      "loss": 0.2839,
      "step": 1811
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7405916768788684,
      "learning_rate": 1.0965211138252373e-06,
      "loss": 0.2745,
      "step": 1812
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7456206892298256,
      "learning_rate": 1.09210336467174e-06,
      "loss": 0.2715,
      "step": 1813
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7021284053553314,
      "learning_rate": 1.0876934414632523e-06,
      "loss": 0.2577,
      "step": 1814
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7434910887211985,
      "learning_rate": 1.0832913530310784e-06,
      "loss": 0.2614,
      "step": 1815
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7472976521590565,
      "learning_rate": 1.0788971081908306e-06,
      "loss": 0.2723,
      "step": 1816
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7612407759133846,
      "learning_rate": 1.0745107157424155e-06,
      "loss": 0.2774,
      "step": 1817
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7743269315850725,
      "learning_rate": 1.0701321844700097e-06,
      "loss": 0.2939,
      "step": 1818
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7289258134290334,
      "learning_rate": 1.0657615231420492e-06,
      "loss": 0.2619,
      "step": 1819
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8110722370526543,
      "learning_rate": 1.0613987405112153e-06,
      "loss": 0.2782,
      "step": 1820
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7395672807279592,
      "learning_rate": 1.0570438453144044e-06,
      "loss": 0.2816,
      "step": 1821
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.727529574723766,
      "learning_rate": 1.0526968462727189e-06,
      "loss": 0.2526,
      "step": 1822
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7941749603743686,
      "learning_rate": 1.0483577520914495e-06,
      "loss": 0.2704,
      "step": 1823
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7888564213973864,
      "learning_rate": 1.0440265714600573e-06,
      "loss": 0.2767,
      "step": 1824
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7459167084720597,
      "learning_rate": 1.0397033130521554e-06,
      "loss": 0.272,
      "step": 1825
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.756353470827699,
      "learning_rate": 1.0353879855254895e-06,
      "loss": 0.2898,
      "step": 1826
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7598718815361815,
      "learning_rate": 1.0310805975219256e-06,
      "loss": 0.27,
      "step": 1827
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7438830697231386,
      "learning_rate": 1.0267811576674298e-06,
      "loss": 0.2627,
      "step": 1828
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7649028391416501,
      "learning_rate": 1.0224896745720513e-06,
      "loss": 0.2586,
      "step": 1829
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7768880348667921,
      "learning_rate": 1.0182061568299019e-06,
      "loss": 0.2715,
      "step": 1830
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7438423754027685,
      "learning_rate": 1.0139306130191463e-06,
      "loss": 0.2542,
      "step": 1831
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7556671018996225,
      "learning_rate": 1.009663051701979e-06,
      "loss": 0.2621,
      "step": 1832
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7735206765964004,
      "learning_rate": 1.0054034814246093e-06,
      "loss": 0.303,
      "step": 1833
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7750549537069624,
      "learning_rate": 1.0011519107172413e-06,
      "loss": 0.2707,
      "step": 1834
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.783834633505104,
      "learning_rate": 9.969083480940617e-07,
      "loss": 0.2754,
      "step": 1835
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7344089099875092,
      "learning_rate": 9.926728020532196e-07,
      "loss": 0.2631,
      "step": 1836
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7499799991744321,
      "learning_rate": 9.884452810768115e-07,
      "loss": 0.267,
      "step": 1837
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7515112169565904,
      "learning_rate": 9.842257936308586e-07,
      "loss": 0.2705,
      "step": 1838
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.816947119821366,
      "learning_rate": 9.80014348165298e-07,
      "loss": 0.2657,
      "step": 1839
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8391438778274399,
      "learning_rate": 9.758109531139614e-07,
      "loss": 0.2782,
      "step": 1840
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7680378474434831,
      "learning_rate": 9.71615616894559e-07,
      "loss": 0.2787,
      "step": 1841
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.751140977072147,
      "learning_rate": 9.6742834790866e-07,
      "loss": 0.2588,
      "step": 1842
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7729712508401576,
      "learning_rate": 9.632491545416788e-07,
      "loss": 0.2843,
      "step": 1843
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.743254350192822,
      "learning_rate": 9.590780451628617e-07,
      "loss": 0.2721,
      "step": 1844
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7550710651729868,
      "learning_rate": 9.549150281252633e-07,
      "loss": 0.2978,
      "step": 1845
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7467489994564781,
      "learning_rate": 9.507601117657289e-07,
      "loss": 0.2629,
      "step": 1846
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8021717717723761,
      "learning_rate": 9.466133044048865e-07,
      "loss": 0.2714,
      "step": 1847
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7629933845019029,
      "learning_rate": 9.424746143471231e-07,
      "loss": 0.2652,
      "step": 1848
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7671410970567217,
      "learning_rate": 9.383440498805712e-07,
      "loss": 0.2891,
      "step": 1849
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7951739276213825,
      "learning_rate": 9.34221619277087e-07,
      "loss": 0.286,
      "step": 1850
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.762146981487799,
      "learning_rate": 9.301073307922432e-07,
      "loss": 0.2492,
      "step": 1851
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8265481707155987,
      "learning_rate": 9.260011926653034e-07,
      "loss": 0.2808,
      "step": 1852
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7598046276870877,
      "learning_rate": 9.219032131192124e-07,
      "loss": 0.2662,
      "step": 1853
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7791095143931884,
      "learning_rate": 9.178134003605721e-07,
      "loss": 0.2676,
      "step": 1854
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7597119230282594,
      "learning_rate": 9.137317625796338e-07,
      "loss": 0.2788,
      "step": 1855
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8082447373639543,
      "learning_rate": 9.096583079502763e-07,
      "loss": 0.2868,
      "step": 1856
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7294748334711889,
      "learning_rate": 9.055930446299916e-07,
      "loss": 0.2769,
      "step": 1857
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.721870329156041,
      "learning_rate": 9.015359807598645e-07,
      "loss": 0.2449,
      "step": 1858
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7747204221888708,
      "learning_rate": 8.974871244645628e-07,
      "loss": 0.27,
      "step": 1859
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7533677155994495,
      "learning_rate": 8.934464838523183e-07,
      "loss": 0.2665,
      "step": 1860
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7728648217497609,
      "learning_rate": 8.894140670149082e-07,
      "loss": 0.2805,
      "step": 1861
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7395247705837903,
      "learning_rate": 8.853898820276402e-07,
      "loss": 0.2614,
      "step": 1862
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7816292911342424,
      "learning_rate": 8.813739369493395e-07,
      "loss": 0.2827,
      "step": 1863
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.751941048649901,
      "learning_rate": 8.773662398223276e-07,
      "loss": 0.2957,
      "step": 1864
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7590928447481631,
      "learning_rate": 8.733667986724109e-07,
      "loss": 0.28,
      "step": 1865
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7796234882398546,
      "learning_rate": 8.693756215088617e-07,
      "loss": 0.2751,
      "step": 1866
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7365864417131243,
      "learning_rate": 8.653927163243986e-07,
      "loss": 0.262,
      "step": 1867
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7415670756219633,
      "learning_rate": 8.61418091095183e-07,
      "loss": 0.2617,
      "step": 1868
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7417658553717682,
      "learning_rate": 8.574517537807897e-07,
      "loss": 0.2527,
      "step": 1869
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7698036605797287,
      "learning_rate": 8.534937123241949e-07,
      "loss": 0.2854,
      "step": 1870
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.740427765582935,
      "learning_rate": 8.495439746517648e-07,
      "loss": 0.2735,
      "step": 1871
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7833842744418559,
      "learning_rate": 8.45602548673235e-07,
      "loss": 0.2939,
      "step": 1872
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.770878098317987,
      "learning_rate": 8.416694422816974e-07,
      "loss": 0.2772,
      "step": 1873
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7587084153651091,
      "learning_rate": 8.377446633535797e-07,
      "loss": 0.2727,
      "step": 1874
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7576270359085742,
      "learning_rate": 8.338282197486364e-07,
      "loss": 0.2702,
      "step": 1875
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7648792531822077,
      "learning_rate": 8.299201193099287e-07,
      "loss": 0.2624,
      "step": 1876
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7707828122602025,
      "learning_rate": 8.260203698638114e-07,
      "loss": 0.2693,
      "step": 1877
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7858871133135622,
      "learning_rate": 8.221289792199117e-07,
      "loss": 0.2984,
      "step": 1878
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7763025023518362,
      "learning_rate": 8.182459551711197e-07,
      "loss": 0.265,
      "step": 1879
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7549915237889264,
      "learning_rate": 8.143713054935715e-07,
      "loss": 0.2637,
      "step": 1880
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8040949548274046,
      "learning_rate": 8.105050379466334e-07,
      "loss": 0.2781,
      "step": 1881
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7688309345276367,
      "learning_rate": 8.066471602728804e-07,
      "loss": 0.2883,
      "step": 1882
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.78158787843222,
      "learning_rate": 8.027976801980924e-07,
      "loss": 0.2729,
      "step": 1883
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8053544729291938,
      "learning_rate": 7.989566054312286e-07,
      "loss": 0.2816,
      "step": 1884
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7334633942866413,
      "learning_rate": 7.95123943664417e-07,
      "loss": 0.2645,
      "step": 1885
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6853083262423729,
      "learning_rate": 7.91299702572938e-07,
      "loss": 0.2505,
      "step": 1886
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8193678247871004,
      "learning_rate": 7.87483889815207e-07,
      "loss": 0.2995,
      "step": 1887
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7905250303703207,
      "learning_rate": 7.836765130327622e-07,
      "loss": 0.288,
      "step": 1888
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7442139211629281,
      "learning_rate": 7.798775798502484e-07,
      "loss": 0.3096,
      "step": 1889
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7386721744734464,
      "learning_rate": 7.760870978754015e-07,
      "loss": 0.2544,
      "step": 1890
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7758063892538867,
      "learning_rate": 7.723050746990291e-07,
      "loss": 0.2691,
      "step": 1891
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7095398603245021,
      "learning_rate": 7.685315178950048e-07,
      "loss": 0.2496,
      "step": 1892
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7580259666576418,
      "learning_rate": 7.647664350202461e-07,
      "loss": 0.2759,
      "step": 1893
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8141978662586711,
      "learning_rate": 7.610098336146965e-07,
      "loss": 0.3066,
      "step": 1894
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.734120045736112,
      "learning_rate": 7.572617212013189e-07,
      "loss": 0.276,
      "step": 1895
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.789834133822412,
      "learning_rate": 7.535221052860747e-07,
      "loss": 0.2853,
      "step": 1896
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7996829773836711,
      "learning_rate": 7.497909933579117e-07,
      "loss": 0.2899,
      "step": 1897
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7379062913071022,
      "learning_rate": 7.460683928887441e-07,
      "loss": 0.2591,
      "step": 1898
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7601397052112312,
      "learning_rate": 7.423543113334436e-07,
      "loss": 0.2554,
      "step": 1899
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7513567808645648,
      "learning_rate": 7.386487561298223e-07,
      "loss": 0.2615,
      "step": 1900
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7700340854069232,
      "learning_rate": 7.349517346986173e-07,
      "loss": 0.2508,
      "step": 1901
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7164266883072792,
      "learning_rate": 7.312632544434739e-07,
      "loss": 0.248,
      "step": 1902
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7565753985181748,
      "learning_rate": 7.275833227509344e-07,
      "loss": 0.2691,
      "step": 1903
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7278650134664856,
      "learning_rate": 7.239119469904227e-07,
      "loss": 0.2687,
      "step": 1904
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7839258087046345,
      "learning_rate": 7.202491345142288e-07,
      "loss": 0.2834,
      "step": 1905
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7330906841215518,
      "learning_rate": 7.165948926574895e-07,
      "loss": 0.2726,
      "step": 1906
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7722452885360467,
      "learning_rate": 7.129492287381828e-07,
      "loss": 0.2705,
      "step": 1907
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7695838390909011,
      "learning_rate": 7.093121500571082e-07,
      "loss": 0.275,
      "step": 1908
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7519869857842357,
      "learning_rate": 7.056836638978698e-07,
      "loss": 0.2675,
      "step": 1909
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7401266012256532,
      "learning_rate": 7.020637775268674e-07,
      "loss": 0.2446,
      "step": 1910
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7471103054583882,
      "learning_rate": 6.984524981932756e-07,
      "loss": 0.2714,
      "step": 1911
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8152337289044853,
      "learning_rate": 6.948498331290349e-07,
      "loss": 0.2993,
      "step": 1912
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7348697043074609,
      "learning_rate": 6.912557895488342e-07,
      "loss": 0.2609,
      "step": 1913
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7375016939145235,
      "learning_rate": 6.876703746500984e-07,
      "loss": 0.2707,
      "step": 1914
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7415444217002594,
      "learning_rate": 6.84093595612968e-07,
      "loss": 0.2865,
      "step": 1915
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7154156511143559,
      "learning_rate": 6.805254596002959e-07,
      "loss": 0.2545,
      "step": 1916
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7724470119069486,
      "learning_rate": 6.769659737576229e-07,
      "loss": 0.2658,
      "step": 1917
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7287532318875162,
      "learning_rate": 6.73415145213166e-07,
      "loss": 0.2554,
      "step": 1918
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7346521666682647,
      "learning_rate": 6.698729810778065e-07,
      "loss": 0.2671,
      "step": 1919
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.789971599500364,
      "learning_rate": 6.663394884450752e-07,
      "loss": 0.2652,
      "step": 1920
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.797647482046774,
      "learning_rate": 6.628146743911374e-07,
      "loss": 0.2521,
      "step": 1921
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7625485452602314,
      "learning_rate": 6.592985459747753e-07,
      "loss": 0.2542,
      "step": 1922
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7127270718170019,
      "learning_rate": 6.55791110237381e-07,
      "loss": 0.256,
      "step": 1923
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7731473679080779,
      "learning_rate": 6.522923742029374e-07,
      "loss": 0.2754,
      "step": 1924
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.767963028340167,
      "learning_rate": 6.488023448780068e-07,
      "loss": 0.2781,
      "step": 1925
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7470912305400985,
      "learning_rate": 6.453210292517114e-07,
      "loss": 0.2724,
      "step": 1926
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7771898009728954,
      "learning_rate": 6.418484342957277e-07,
      "loss": 0.2658,
      "step": 1927
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7904730933730023,
      "learning_rate": 6.383845669642657e-07,
      "loss": 0.2747,
      "step": 1928
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.819929775937833,
      "learning_rate": 6.349294341940593e-07,
      "loss": 0.2646,
      "step": 1929
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7331775931466258,
      "learning_rate": 6.3148304290435e-07,
      "loss": 0.2589,
      "step": 1930
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7579376020710628,
      "learning_rate": 6.280453999968711e-07,
      "loss": 0.2749,
      "step": 1931
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7330957561138618,
      "learning_rate": 6.246165123558401e-07,
      "loss": 0.2759,
      "step": 1932
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7729765649183341,
      "learning_rate": 6.211963868479393e-07,
      "loss": 0.2655,
      "step": 1933
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8234114293992821,
      "learning_rate": 6.177850303223059e-07,
      "loss": 0.2892,
      "step": 1934
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7724921627805719,
      "learning_rate": 6.143824496105122e-07,
      "loss": 0.2907,
      "step": 1935
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8343210419422149,
      "learning_rate": 6.109886515265595e-07,
      "loss": 0.2902,
      "step": 1936
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7636052036568655,
      "learning_rate": 6.076036428668613e-07,
      "loss": 0.2612,
      "step": 1937
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7520709002225863,
      "learning_rate": 6.042274304102281e-07,
      "loss": 0.2942,
      "step": 1938
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7594658144284699,
      "learning_rate": 6.008600209178539e-07,
      "loss": 0.2636,
      "step": 1939
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7275305000031295,
      "learning_rate": 5.975014211333053e-07,
      "loss": 0.2613,
      "step": 1940
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7019736381265406,
      "learning_rate": 5.941516377825102e-07,
      "loss": 0.2459,
      "step": 1941
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7944534126272603,
      "learning_rate": 5.908106775737332e-07,
      "loss": 0.2897,
      "step": 1942
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.738809808282589,
      "learning_rate": 5.874785471975753e-07,
      "loss": 0.2749,
      "step": 1943
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7968520849968763,
      "learning_rate": 5.841552533269534e-07,
      "loss": 0.2728,
      "step": 1944
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7724108924574359,
      "learning_rate": 5.808408026170892e-07,
      "loss": 0.2822,
      "step": 1945
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7408264976619694,
      "learning_rate": 5.775352017054919e-07,
      "loss": 0.2769,
      "step": 1946
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7184954329570868,
      "learning_rate": 5.742384572119519e-07,
      "loss": 0.2653,
      "step": 1947
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7665401679494428,
      "learning_rate": 5.70950575738522e-07,
      "loss": 0.2604,
      "step": 1948
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7604230748978644,
      "learning_rate": 5.676715638695063e-07,
      "loss": 0.2675,
      "step": 1949
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7540783478904938,
      "learning_rate": 5.644014281714483e-07,
      "loss": 0.2674,
      "step": 1950
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7364776057523307,
      "learning_rate": 5.611401751931112e-07,
      "loss": 0.2752,
      "step": 1951
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7272042251510548,
      "learning_rate": 5.578878114654746e-07,
      "loss": 0.2393,
      "step": 1952
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7427468677498608,
      "learning_rate": 5.546443435017146e-07,
      "loss": 0.2499,
      "step": 1953
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.743319830080549,
      "learning_rate": 5.514097777971939e-07,
      "loss": 0.2648,
      "step": 1954
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8014108696932148,
      "learning_rate": 5.481841208294447e-07,
      "loss": 0.3105,
      "step": 1955
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7577995829957135,
      "learning_rate": 5.449673790581611e-07,
      "loss": 0.2686,
      "step": 1956
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7112072031061208,
      "learning_rate": 5.417595589251828e-07,
      "loss": 0.2586,
      "step": 1957
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7405860001577842,
      "learning_rate": 5.385606668544846e-07,
      "loss": 0.2672,
      "step": 1958
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7998924240566275,
      "learning_rate": 5.353707092521581e-07,
      "loss": 0.2804,
      "step": 1959
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7632115249437038,
      "learning_rate": 5.321896925064069e-07,
      "loss": 0.2841,
      "step": 1960
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7381625398196124,
      "learning_rate": 5.290176229875271e-07,
      "loss": 0.2856,
      "step": 1961
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7614395442801815,
      "learning_rate": 5.258545070478999e-07,
      "loss": 0.2908,
      "step": 1962
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7613781471158051,
      "learning_rate": 5.227003510219714e-07,
      "loss": 0.2637,
      "step": 1963
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7835111292675196,
      "learning_rate": 5.195551612262478e-07,
      "loss": 0.2789,
      "step": 1964
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7764135901132732,
      "learning_rate": 5.16418943959282e-07,
      "loss": 0.2777,
      "step": 1965
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7760361356949298,
      "learning_rate": 5.132917055016523e-07,
      "loss": 0.2819,
      "step": 1966
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.779395353092655,
      "learning_rate": 5.101734521159613e-07,
      "loss": 0.2884,
      "step": 1967
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.741769920816033,
      "learning_rate": 5.070641900468149e-07,
      "loss": 0.2698,
      "step": 1968
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7262609277956771,
      "learning_rate": 5.039639255208156e-07,
      "loss": 0.2675,
      "step": 1969
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7727638135079957,
      "learning_rate": 5.008726647465468e-07,
      "loss": 0.2852,
      "step": 1970
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7346852278546467,
      "learning_rate": 4.977904139145579e-07,
      "loss": 0.2619,
      "step": 1971
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7916497177604424,
      "learning_rate": 4.947171791973587e-07,
      "loss": 0.284,
      "step": 1972
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7628185225580226,
      "learning_rate": 4.916529667494024e-07,
      "loss": 0.2425,
      "step": 1973
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7732688466472237,
      "learning_rate": 4.885977827070748e-07,
      "loss": 0.2605,
      "step": 1974
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.774930298930306,
      "learning_rate": 4.855516331886783e-07,
      "loss": 0.2883,
      "step": 1975
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7795859101522271,
      "learning_rate": 4.825145242944262e-07,
      "loss": 0.2998,
      "step": 1976
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7448018070032864,
      "learning_rate": 4.794864621064266e-07,
      "loss": 0.2521,
      "step": 1977
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7795933965421548,
      "learning_rate": 4.7646745268867034e-07,
      "loss": 0.3035,
      "step": 1978
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7355908752983016,
      "learning_rate": 4.734575020870169e-07,
      "loss": 0.2577,
      "step": 1979
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7644465981718811,
      "learning_rate": 4.704566163291879e-07,
      "loss": 0.2617,
      "step": 1980
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8108283784725004,
      "learning_rate": 4.6746480142475015e-07,
      "loss": 0.2795,
      "step": 1981
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7312922936596545,
      "learning_rate": 4.6448206336510646e-07,
      "loss": 0.2602,
      "step": 1982
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.77006406855866,
      "learning_rate": 4.6150840812348e-07,
      "loss": 0.2942,
      "step": 1983
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.771835163760772,
      "learning_rate": 4.5854384165490596e-07,
      "loss": 0.2747,
      "step": 1984
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7675479417112172,
      "learning_rate": 4.5558836989621933e-07,
      "loss": 0.2726,
      "step": 1985
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7623956921113128,
      "learning_rate": 4.526419987660419e-07,
      "loss": 0.2811,
      "step": 1986
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7010053939339441,
      "learning_rate": 4.497047341647676e-07,
      "loss": 0.2428,
      "step": 1987
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7938120741143672,
      "learning_rate": 4.467765819745562e-07,
      "loss": 0.2762,
      "step": 1988
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7528378082350491,
      "learning_rate": 4.43857548059321e-07,
      "loss": 0.2805,
      "step": 1989
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7709593997157546,
      "learning_rate": 4.409476382647121e-07,
      "loss": 0.2868,
      "step": 1990
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7075952788884045,
      "learning_rate": 4.380468584181069e-07,
      "loss": 0.2569,
      "step": 1991
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7556627213579192,
      "learning_rate": 4.351552143286014e-07,
      "loss": 0.2928,
      "step": 1992
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7973438966837143,
      "learning_rate": 4.322727117869951e-07,
      "loss": 0.2662,
      "step": 1993
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7654184439435956,
      "learning_rate": 4.293993565657828e-07,
      "loss": 0.2889,
      "step": 1994
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7305198604522679,
      "learning_rate": 4.265351544191365e-07,
      "loss": 0.2558,
      "step": 1995
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7977250980417679,
      "learning_rate": 4.2368011108290306e-07,
      "loss": 0.2757,
      "step": 1996
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7035308138193134,
      "learning_rate": 4.208342322745845e-07,
      "loss": 0.2508,
      "step": 1997
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.791805098542658,
      "learning_rate": 4.1799752369333313e-07,
      "loss": 0.2533,
      "step": 1998
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7813493420754078,
      "learning_rate": 4.151699910199336e-07,
      "loss": 0.2625,
      "step": 1999
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.728103843615513,
      "learning_rate": 4.123516399167965e-07,
      "loss": 0.2599,
      "step": 2000
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7434113622998794,
      "learning_rate": 4.095424760279454e-07,
      "loss": 0.2652,
      "step": 2001
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7654120401224813,
      "learning_rate": 4.067425049790075e-07,
      "loss": 0.2619,
      "step": 2002
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7838569817604232,
      "learning_rate": 4.039517323771958e-07,
      "loss": 0.2777,
      "step": 2003
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7090597303115252,
      "learning_rate": 4.0117016381130636e-07,
      "loss": 0.2514,
      "step": 2004
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7643155819356673,
      "learning_rate": 3.983978048517029e-07,
      "loss": 0.292,
      "step": 2005
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7970298753667525,
      "learning_rate": 3.956346610503048e-07,
      "loss": 0.2715,
      "step": 2006
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.75347522706858,
      "learning_rate": 3.9288073794057634e-07,
      "loss": 0.2864,
      "step": 2007
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8022553983995037,
      "learning_rate": 3.9013604103751933e-07,
      "loss": 0.2748,
      "step": 2008
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7431568423360158,
      "learning_rate": 3.87400575837657e-07,
      "loss": 0.285,
      "step": 2009
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.807741308386567,
      "learning_rate": 3.8467434781902624e-07,
      "loss": 0.268,
      "step": 2010
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7857918098872572,
      "learning_rate": 3.819573624411643e-07,
      "loss": 0.2849,
      "step": 2011
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7606414010725052,
      "learning_rate": 3.792496251451e-07,
      "loss": 0.2911,
      "step": 2012
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7590931681482301,
      "learning_rate": 3.765511413533429e-07,
      "loss": 0.2697,
      "step": 2013
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7759848286545509,
      "learning_rate": 3.7386191646987094e-07,
      "loss": 0.2758,
      "step": 2014
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7906813097447062,
      "learning_rate": 3.711819558801183e-07,
      "loss": 0.2921,
      "step": 2015
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7523489569201227,
      "learning_rate": 3.685112649509681e-07,
      "loss": 0.2663,
      "step": 2016
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7378207313818843,
      "learning_rate": 3.6584984903074104e-07,
      "loss": 0.246,
      "step": 2017
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7774856590880864,
      "learning_rate": 3.631977134491832e-07,
      "loss": 0.2932,
      "step": 2018
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.756950322750425,
      "learning_rate": 3.6055486351745327e-07,
      "loss": 0.2787,
      "step": 2019
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7480792957340122,
      "learning_rate": 3.5792130452811825e-07,
      "loss": 0.2513,
      "step": 2020
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7926123264129371,
      "learning_rate": 3.5529704175513604e-07,
      "loss": 0.2753,
      "step": 2021
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7564644850499262,
      "learning_rate": 3.5268208045385167e-07,
      "loss": 0.2849,
      "step": 2022
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7673174283214625,
      "learning_rate": 3.500764258609779e-07,
      "loss": 0.2924,
      "step": 2023
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7514983461151122,
      "learning_rate": 3.4748008319459457e-07,
      "loss": 0.2921,
      "step": 2024
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7916989678609273,
      "learning_rate": 3.4489305765413097e-07,
      "loss": 0.2914,
      "step": 2025
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7777088712009528,
      "learning_rate": 3.4231535442035947e-07,
      "loss": 0.2565,
      "step": 2026
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7630414559932442,
      "learning_rate": 3.3974697865538195e-07,
      "loss": 0.2513,
      "step": 2027
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7679925814183014,
      "learning_rate": 3.37187935502622e-07,
      "loss": 0.2723,
      "step": 2028
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7293866281409724,
      "learning_rate": 3.346382300868134e-07,
      "loss": 0.2539,
      "step": 2029
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8131875749692871,
      "learning_rate": 3.320978675139919e-07,
      "loss": 0.258,
      "step": 2030
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7373596354833456,
      "learning_rate": 3.2956685287148014e-07,
      "loss": 0.251,
      "step": 2031
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7582707365450034,
      "learning_rate": 3.2704519122788324e-07,
      "loss": 0.2729,
      "step": 2032
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7497408801179702,
      "learning_rate": 3.24532887633075e-07,
      "loss": 0.2655,
      "step": 2033
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7152748140664134,
      "learning_rate": 3.220299471181898e-07,
      "loss": 0.26,
      "step": 2034
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7855953706511497,
      "learning_rate": 3.195363746956087e-07,
      "loss": 0.2951,
      "step": 2035
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7816020696983322,
      "learning_rate": 3.1705217535895504e-07,
      "loss": 0.3022,
      "step": 2036
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7486743754524493,
      "learning_rate": 3.1457735408308153e-07,
      "loss": 0.2675,
      "step": 2037
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7145072635913473,
      "learning_rate": 3.121119158240604e-07,
      "loss": 0.2571,
      "step": 2038
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7815188357002163,
      "learning_rate": 3.096558655191706e-07,
      "loss": 0.264,
      "step": 2039
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7722050304719482,
      "learning_rate": 3.0720920808689435e-07,
      "loss": 0.2749,
      "step": 2040
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7700552927758477,
      "learning_rate": 3.047719484269018e-07,
      "loss": 0.2689,
      "step": 2041
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7143198975652422,
      "learning_rate": 3.023440914200448e-07,
      "loss": 0.2548,
      "step": 2042
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7830677611155217,
      "learning_rate": 2.9992564192834253e-07,
      "loss": 0.2729,
      "step": 2043
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7140546478074725,
      "learning_rate": 2.9751660479497737e-07,
      "loss": 0.2447,
      "step": 2044
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7404714450544222,
      "learning_rate": 2.9511698484428144e-07,
      "loss": 0.2732,
      "step": 2045
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7459646955902111,
      "learning_rate": 2.9272678688172905e-07,
      "loss": 0.2867,
      "step": 2046
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7696067259912243,
      "learning_rate": 2.9034601569392417e-07,
      "loss": 0.2474,
      "step": 2047
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7377234844652334,
      "learning_rate": 2.879746760485946e-07,
      "loss": 0.264,
      "step": 2048
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.847698268284129,
      "learning_rate": 2.85612772694579e-07,
      "loss": 0.2987,
      "step": 2049
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7182728273818337,
      "learning_rate": 2.832603103618214e-07,
      "loss": 0.2588,
      "step": 2050
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8240882362495338,
      "learning_rate": 2.809172937613552e-07,
      "loss": 0.2673,
      "step": 2051
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7218130387508205,
      "learning_rate": 2.7858372758530206e-07,
      "loss": 0.2649,
      "step": 2052
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7713470069755515,
      "learning_rate": 2.762596165068548e-07,
      "loss": 0.2706,
      "step": 2053
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7979605196942585,
      "learning_rate": 2.739449651802756e-07,
      "loss": 0.2771,
      "step": 2054
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7413915496727176,
      "learning_rate": 2.7163977824087696e-07,
      "loss": 0.2555,
      "step": 2055
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8305521363476968,
      "learning_rate": 2.6934406030502235e-07,
      "loss": 0.2564,
      "step": 2056
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.758272644435761,
      "learning_rate": 2.6705781597011147e-07,
      "loss": 0.2512,
      "step": 2057
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7450959885778398,
      "learning_rate": 2.6478104981457254e-07,
      "loss": 0.2655,
      "step": 2058
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7638810331793243,
      "learning_rate": 2.6251376639785163e-07,
      "loss": 0.2637,
      "step": 2059
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7629940920537911,
      "learning_rate": 2.6025597026040463e-07,
      "loss": 0.2661,
      "step": 2060
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7730732112353788,
      "learning_rate": 2.5800766592369077e-07,
      "loss": 0.267,
      "step": 2061
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7652230037589127,
      "learning_rate": 2.5576885789015905e-07,
      "loss": 0.2662,
      "step": 2062
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7906860958800764,
      "learning_rate": 2.535395506432403e-07,
      "loss": 0.2973,
      "step": 2063
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7697604437800446,
      "learning_rate": 2.5131974864734063e-07,
      "loss": 0.2823,
      "step": 2064
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7074402557592219,
      "learning_rate": 2.4910945634783024e-07,
      "loss": 0.238,
      "step": 2065
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7257885814055683,
      "learning_rate": 2.469086781710367e-07,
      "loss": 0.2648,
      "step": 2066
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7985114845244932,
      "learning_rate": 2.447174185242324e-07,
      "loss": 0.2793,
      "step": 2067
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7858093309591058,
      "learning_rate": 2.425356817956287e-07,
      "loss": 0.2717,
      "step": 2068
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7734867881627874,
      "learning_rate": 2.403634723543674e-07,
      "loss": 0.2773,
      "step": 2069
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7173948750822898,
      "learning_rate": 2.3820079455051048e-07,
      "loss": 0.2561,
      "step": 2070
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7578665059438355,
      "learning_rate": 2.3604765271503073e-07,
      "loss": 0.2606,
      "step": 2071
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7379282125541793,
      "learning_rate": 2.3390405115980563e-07,
      "loss": 0.2593,
      "step": 2072
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7751750370455964,
      "learning_rate": 2.3176999417760637e-07,
      "loss": 0.2531,
      "step": 2073
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7894359495541736,
      "learning_rate": 2.2964548604209214e-07,
      "loss": 0.2811,
      "step": 2074
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7491697768185932,
      "learning_rate": 2.2753053100779632e-07,
      "loss": 0.2601,
      "step": 2075
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7271093453954678,
      "learning_rate": 2.2542513331012372e-07,
      "loss": 0.2521,
      "step": 2076
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8181824853227404,
      "learning_rate": 2.2332929716533947e-07,
      "loss": 0.3176,
      "step": 2077
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7896263635985499,
      "learning_rate": 2.2124302677055954e-07,
      "loss": 0.2687,
      "step": 2078
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7920648541198535,
      "learning_rate": 2.1916632630374579e-07,
      "loss": 0.2626,
      "step": 2079
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8081540694386906,
      "learning_rate": 2.1709919992369211e-07,
      "loss": 0.3126,
      "step": 2080
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7571009708979625,
      "learning_rate": 2.1504165177002212e-07,
      "loss": 0.2728,
      "step": 2081
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7813587899123894,
      "learning_rate": 2.1299368596317705e-07,
      "loss": 0.2683,
      "step": 2082
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7344771202787193,
      "learning_rate": 2.1095530660440954e-07,
      "loss": 0.2419,
      "step": 2083
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7608164255981397,
      "learning_rate": 2.0892651777577045e-07,
      "loss": 0.255,
      "step": 2084
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7528056545362743,
      "learning_rate": 2.069073235401109e-07,
      "loss": 0.2681,
      "step": 2085
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7639256244729793,
      "learning_rate": 2.0489772794106411e-07,
      "loss": 0.2608,
      "step": 2086
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8463452420782958,
      "learning_rate": 2.028977350030409e-07,
      "loss": 0.2865,
      "step": 2087
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7385188416715235,
      "learning_rate": 2.0090734873122243e-07,
      "loss": 0.2539,
      "step": 2088
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.770902532099322,
      "learning_rate": 1.989265731115525e-07,
      "loss": 0.2616,
      "step": 2089
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7713880996877653,
      "learning_rate": 1.969554121107281e-07,
      "loss": 0.2949,
      "step": 2090
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7309255907438141,
      "learning_rate": 1.9499386967619104e-07,
      "loss": 0.2683,
      "step": 2091
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7785506609601652,
      "learning_rate": 1.9304194973612244e-07,
      "loss": 0.2854,
      "step": 2092
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7720226609646021,
      "learning_rate": 1.910996561994316e-07,
      "loss": 0.2909,
      "step": 2093
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7967394377143134,
      "learning_rate": 1.8916699295575324e-07,
      "loss": 0.2536,
      "step": 2094
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8032806596978133,
      "learning_rate": 1.872439638754331e-07,
      "loss": 0.2739,
      "step": 2095
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.75777239817011,
      "learning_rate": 1.853305728095245e-07,
      "loss": 0.2663,
      "step": 2096
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7918809670741389,
      "learning_rate": 1.834268235897807e-07,
      "loss": 0.2888,
      "step": 2097
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7909572574991177,
      "learning_rate": 1.8153272002864542e-07,
      "loss": 0.2877,
      "step": 2098
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7540733914352655,
      "learning_rate": 1.7964826591924722e-07,
      "loss": 0.2589,
      "step": 2099
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7699012176557264,
      "learning_rate": 1.7777346503538796e-07,
      "loss": 0.2766,
      "step": 2100
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7224943992899954,
      "learning_rate": 1.7590832113154045e-07,
      "loss": 0.2361,
      "step": 2101
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6790602396940318,
      "learning_rate": 1.7405283794283745e-07,
      "loss": 0.23,
      "step": 2102
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7261825143435299,
      "learning_rate": 1.7220701918506666e-07,
      "loss": 0.2486,
      "step": 2103
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8067111168185477,
      "learning_rate": 1.7037086855465902e-07,
      "loss": 0.3023,
      "step": 2104
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7745703639442444,
      "learning_rate": 1.685443897286859e-07,
      "loss": 0.2828,
      "step": 2105
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7688548893990761,
      "learning_rate": 1.6672758636485043e-07,
      "loss": 0.2661,
      "step": 2106
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7392018206659917,
      "learning_rate": 1.6492046210147884e-07,
      "loss": 0.2638,
      "step": 2107
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7727961513389019,
      "learning_rate": 1.6312302055751295e-07,
      "loss": 0.2693,
      "step": 2108
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7500126553404481,
      "learning_rate": 1.6133526533250566e-07,
      "loss": 0.2704,
      "step": 2109
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7897474568332211,
      "learning_rate": 1.5955720000661312e-07,
      "loss": 0.2788,
      "step": 2110
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.742250390900984,
      "learning_rate": 1.577888281405826e-07,
      "loss": 0.2601,
      "step": 2111
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7642764377536019,
      "learning_rate": 1.5603015327575354e-07,
      "loss": 0.2531,
      "step": 2112
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7912238472821711,
      "learning_rate": 1.5428117893404315e-07,
      "loss": 0.2796,
      "step": 2113
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7615803351510887,
      "learning_rate": 1.5254190861794415e-07,
      "loss": 0.2734,
      "step": 2114
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7496448486054121,
      "learning_rate": 1.5081234581051484e-07,
      "loss": 0.2676,
      "step": 2115
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7689891595328037,
      "learning_rate": 1.490924939753735e-07,
      "loss": 0.2779,
      "step": 2116
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7712259755340914,
      "learning_rate": 1.473823565566923e-07,
      "loss": 0.2783,
      "step": 2117
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7401730879631578,
      "learning_rate": 1.4568193697918843e-07,
      "loss": 0.2698,
      "step": 2118
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7411800972649775,
      "learning_rate": 1.4399123864811904e-07,
      "loss": 0.273,
      "step": 2119
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7444249855321009,
      "learning_rate": 1.4231026494927082e-07,
      "loss": 0.2738,
      "step": 2120
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7368039956882756,
      "learning_rate": 1.4063901924895985e-07,
      "loss": 0.2642,
      "step": 2121
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7830711853757314,
      "learning_rate": 1.3897750489401785e-07,
      "loss": 0.2661,
      "step": 2122
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7610981617872303,
      "learning_rate": 1.3732572521179043e-07,
      "loss": 0.2745,
      "step": 2123
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7238220966935012,
      "learning_rate": 1.3568368351012718e-07,
      "loss": 0.2712,
      "step": 2124
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7439369288579287,
      "learning_rate": 1.3405138307737765e-07,
      "loss": 0.2622,
      "step": 2125
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7380277582761108,
      "learning_rate": 1.324288271823826e-07,
      "loss": 0.2682,
      "step": 2126
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7144322769054927,
      "learning_rate": 1.3081601907447007e-07,
      "loss": 0.2456,
      "step": 2127
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7760167880781901,
      "learning_rate": 1.2921296198344424e-07,
      "loss": 0.2803,
      "step": 2128
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7508512045214063,
      "learning_rate": 1.2761965911958385e-07,
      "loss": 0.266,
      "step": 2129
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7624591281086401,
      "learning_rate": 1.260361136736349e-07,
      "loss": 0.2637,
      "step": 2130
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7698859079803505,
      "learning_rate": 1.244623288168012e-07,
      "loss": 0.2699,
      "step": 2131
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8190151409537498,
      "learning_rate": 1.2289830770074064e-07,
      "loss": 0.2627,
      "step": 2132
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8155151449503344,
      "learning_rate": 1.2134405345755773e-07,
      "loss": 0.2925,
      "step": 2133
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.795802718873389,
      "learning_rate": 1.1979956919979996e-07,
      "loss": 0.2765,
      "step": 2134
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8292986354578453,
      "learning_rate": 1.1826485802044707e-07,
      "loss": 0.2717,
      "step": 2135
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.732603876833168,
      "learning_rate": 1.1673992299290893e-07,
      "loss": 0.2599,
      "step": 2136
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7639439151441669,
      "learning_rate": 1.1522476717101605e-07,
      "loss": 0.2795,
      "step": 2137
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6994943394945053,
      "learning_rate": 1.137193935890174e-07,
      "loss": 0.2605,
      "step": 2138
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7910861861837787,
      "learning_rate": 1.1222380526156929e-07,
      "loss": 0.2701,
      "step": 2139
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.738163237460845,
      "learning_rate": 1.1073800518373423e-07,
      "loss": 0.2704,
      "step": 2140
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7617186452658012,
      "learning_rate": 1.0926199633097156e-07,
      "loss": 0.2811,
      "step": 2141
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7274037091566434,
      "learning_rate": 1.0779578165913463e-07,
      "loss": 0.259,
      "step": 2142
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7798932265847865,
      "learning_rate": 1.0633936410446077e-07,
      "loss": 0.2637,
      "step": 2143
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7449944337147956,
      "learning_rate": 1.0489274658356808e-07,
      "loss": 0.2652,
      "step": 2144
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7781044754165846,
      "learning_rate": 1.0345593199344972e-07,
      "loss": 0.2789,
      "step": 2145
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7727901043973068,
      "learning_rate": 1.0202892321146796e-07,
      "loss": 0.2421,
      "step": 2146
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7042845220988333,
      "learning_rate": 1.0061172309534739e-07,
      "loss": 0.2697,
      "step": 2147
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.804320539855615,
      "learning_rate": 9.92043344831689e-08,
      "loss": 0.2618,
      "step": 2148
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7586429590701782,
      "learning_rate": 9.780676019336632e-08,
      "loss": 0.2623,
      "step": 2149
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7522897299298875,
      "learning_rate": 9.641900302471918e-08,
      "loss": 0.2816,
      "step": 2150
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7695492760237153,
      "learning_rate": 9.504106575634664e-08,
      "loss": 0.2511,
      "step": 2151
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7508843791378448,
      "learning_rate": 9.367295114770191e-08,
      "loss": 0.263,
      "step": 2152
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7875084403367537,
      "learning_rate": 9.231466193856897e-08,
      "loss": 0.2769,
      "step": 2153
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.757297024315734,
      "learning_rate": 9.096620084905472e-08,
      "loss": 0.2621,
      "step": 2154
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7529309831742007,
      "learning_rate": 8.96275705795846e-08,
      "loss": 0.2664,
      "step": 2155
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.781798773906386,
      "learning_rate": 8.829877381089536e-08,
      "loss": 0.2794,
      "step": 2156
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7943631242224873,
      "learning_rate": 8.697981320403337e-08,
      "loss": 0.2871,
      "step": 2157
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.73553607148722,
      "learning_rate": 8.56706914003469e-08,
      "loss": 0.2727,
      "step": 2158
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7359455291409597,
      "learning_rate": 8.437141102147883e-08,
      "loss": 0.2724,
      "step": 2159
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7618266823434364,
      "learning_rate": 8.308197466936619e-08,
      "loss": 0.2593,
      "step": 2160
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7811017840845587,
      "learning_rate": 8.18023849262306e-08,
      "loss": 0.2705,
      "step": 2161
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7772385070650162,
      "learning_rate": 8.05326443545762e-08,
      "loss": 0.2946,
      "step": 2162
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7631776218940284,
      "learning_rate": 7.927275549718228e-08,
      "loss": 0.2836,
      "step": 2163
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7844444940103564,
      "learning_rate": 7.802272087709951e-08,
      "loss": 0.2851,
      "step": 2164
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7713481600527503,
      "learning_rate": 7.678254299764431e-08,
      "loss": 0.2693,
      "step": 2165
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7519794451230432,
      "learning_rate": 7.555222434239395e-08,
      "loss": 0.2614,
      "step": 2166
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.770191305726789,
      "learning_rate": 7.433176737518255e-08,
      "loss": 0.2539,
      "step": 2167
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7553601213845197,
      "learning_rate": 7.312117454009337e-08,
      "loss": 0.2693,
      "step": 2168
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7125398215424915,
      "learning_rate": 7.192044826145772e-08,
      "loss": 0.2416,
      "step": 2169
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7854638912968984,
      "learning_rate": 7.072959094384713e-08,
      "loss": 0.2936,
      "step": 2170
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7917934219807625,
      "learning_rate": 6.954860497207006e-08,
      "loss": 0.2696,
      "step": 2171
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7878749775919384,
      "learning_rate": 6.837749271116578e-08,
      "loss": 0.3026,
      "step": 2172
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.767727856107273,
      "learning_rate": 6.721625650640163e-08,
      "loss": 0.2853,
      "step": 2173
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7858825895317805,
      "learning_rate": 6.606489868326571e-08,
      "loss": 0.2893,
      "step": 2174
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7927714665650804,
      "learning_rate": 6.49234215474659e-08,
      "loss": 0.276,
      "step": 2175
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7756081941479313,
      "learning_rate": 6.379182738492085e-08,
      "loss": 0.2716,
      "step": 2176
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7521352411797542,
      "learning_rate": 6.267011846175896e-08,
      "loss": 0.2725,
      "step": 2177
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7505234700652894,
      "learning_rate": 6.15582970243117e-08,
      "loss": 0.276,
      "step": 2178
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7506803524755093,
      "learning_rate": 6.045636529911025e-08,
      "loss": 0.269,
      "step": 2179
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7600728339573537,
      "learning_rate": 5.9364325492880514e-08,
      "loss": 0.2709,
      "step": 2180
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7559229996299667,
      "learning_rate": 5.82821797925387e-08,
      "loss": 0.2869,
      "step": 2181
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7851991316173891,
      "learning_rate": 5.720993036518796e-08,
      "loss": 0.2929,
      "step": 2182
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7319986626030169,
      "learning_rate": 5.6147579358112324e-08,
      "loss": 0.2567,
      "step": 2183
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7383747418309984,
      "learning_rate": 5.509512889877333e-08,
      "loss": 0.2734,
      "step": 2184
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7604089991815375,
      "learning_rate": 5.4052581094806713e-08,
      "loss": 0.2667,
      "step": 2185
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.75688514903944,
      "learning_rate": 5.301993803401517e-08,
      "loss": 0.2737,
      "step": 2186
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7343593022602145,
      "learning_rate": 5.19972017843684e-08,
      "loss": 0.2714,
      "step": 2187
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.779849964600701,
      "learning_rate": 5.098437439399473e-08,
      "loss": 0.2677,
      "step": 2188
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7744840767123657,
      "learning_rate": 4.998145789118114e-08,
      "loss": 0.2622,
      "step": 2189
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.770181423792915,
      "learning_rate": 4.898845428436494e-08,
      "loss": 0.2837,
      "step": 2190
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7450049436097146,
      "learning_rate": 4.800536556213375e-08,
      "loss": 0.2554,
      "step": 2191
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7889032194922797,
      "learning_rate": 4.7032193693217765e-08,
      "loss": 0.2597,
      "step": 2192
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7644540363052238,
      "learning_rate": 4.60689406264897e-08,
      "loss": 0.2801,
      "step": 2193
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7337513937895791,
      "learning_rate": 4.511560829095818e-08,
      "loss": 0.2607,
      "step": 2194
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6676061982006292,
      "learning_rate": 4.417219859576383e-08,
      "loss": 0.2323,
      "step": 2195
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6846811521073093,
      "learning_rate": 4.3238713430176494e-08,
      "loss": 0.2514,
      "step": 2196
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7724438643357927,
      "learning_rate": 4.231515466359193e-08,
      "loss": 0.2724,
      "step": 2197
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7986199922128123,
      "learning_rate": 4.140152414552678e-08,
      "loss": 0.2965,
      "step": 2198
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7500316780270709,
      "learning_rate": 4.0497823705615836e-08,
      "loss": 0.262,
      "step": 2199
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7654101648165158,
      "learning_rate": 3.960405515360699e-08,
      "loss": 0.2583,
      "step": 2200
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7318335562192915,
      "learning_rate": 3.8720220279359065e-08,
      "loss": 0.2469,
      "step": 2201
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7709101737839337,
      "learning_rate": 3.784632085283846e-08,
      "loss": 0.2816,
      "step": 2202
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8156018763847249,
      "learning_rate": 3.698235862411359e-08,
      "loss": 0.2706,
      "step": 2203
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7223510654625391,
      "learning_rate": 3.6128335323353804e-08,
      "loss": 0.2754,
      "step": 2204
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8318369887216122,
      "learning_rate": 3.528425266082325e-08,
      "loss": 0.3001,
      "step": 2205
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7592616679487475,
      "learning_rate": 3.445011232688089e-08,
      "loss": 0.2687,
      "step": 2206
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8000863393356934,
      "learning_rate": 3.36259159919744e-08,
      "loss": 0.2924,
      "step": 2207
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7759820690236486,
      "learning_rate": 3.281166530663682e-08,
      "loss": 0.2819,
      "step": 2208
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.78172890561744,
      "learning_rate": 3.2007361901485455e-08,
      "loss": 0.2589,
      "step": 2209
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7844129994768493,
      "learning_rate": 3.121300738721578e-08,
      "loss": 0.2734,
      "step": 2210
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7278784238520013,
      "learning_rate": 3.042860335460085e-08,
      "loss": 0.2633,
      "step": 2211
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8283183263246772,
      "learning_rate": 2.9654151374485797e-08,
      "loss": 0.296,
      "step": 2212
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.791695025516109,
      "learning_rate": 2.8889652997787233e-08,
      "loss": 0.2784,
      "step": 2213
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.754242217070193,
      "learning_rate": 2.8135109755487723e-08,
      "loss": 0.2682,
      "step": 2214
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7723774482730715,
      "learning_rate": 2.7390523158633552e-08,
      "loss": 0.2805,
      "step": 2215
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7846382055645278,
      "learning_rate": 2.6655894698331962e-08,
      "loss": 0.2658,
      "step": 2216
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7661382539479455,
      "learning_rate": 2.593122584574892e-08,
      "loss": 0.2515,
      "step": 2217
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7088811449485795,
      "learning_rate": 2.5216518052104122e-08,
      "loss": 0.2492,
      "step": 2218
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7664152792575851,
      "learning_rate": 2.4511772748669894e-08,
      "loss": 0.2747,
      "step": 2219
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7871137305471669,
      "learning_rate": 2.381699134676785e-08,
      "loss": 0.2791,
      "step": 2220
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.720180465486174,
      "learning_rate": 2.3132175237765564e-08,
      "loss": 0.2506,
      "step": 2221
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7257921506182616,
      "learning_rate": 2.2457325793073803e-08,
      "loss": 0.249,
      "step": 2222
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7142509823588122,
      "learning_rate": 2.179244436414485e-08,
      "loss": 0.2287,
      "step": 2223
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8117718753450942,
      "learning_rate": 2.1137532282469176e-08,
      "loss": 0.2741,
      "step": 2224
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7476903713490182,
      "learning_rate": 2.0492590859571558e-08,
      "loss": 0.2715,
      "step": 2225
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.778176909615104,
      "learning_rate": 1.9857621387010528e-08,
      "loss": 0.2754,
      "step": 2226
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7505953007137998,
      "learning_rate": 1.9232625136374472e-08,
      "loss": 0.2591,
      "step": 2227
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7703930773460543,
      "learning_rate": 1.8617603359279977e-08,
      "loss": 0.2661,
      "step": 2228
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7920906922727621,
      "learning_rate": 1.8012557287367394e-08,
      "loss": 0.2868,
      "step": 2229
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7218622088729432,
      "learning_rate": 1.7417488132301375e-08,
      "loss": 0.2563,
      "step": 2230
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7111514008758689,
      "learning_rate": 1.6832397085765893e-08,
      "loss": 0.2536,
      "step": 2231
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7750668024467428,
      "learning_rate": 1.6257285319462578e-08,
      "loss": 0.2612,
      "step": 2232
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7469380801445711,
      "learning_rate": 1.5692153985109037e-08,
      "loss": 0.2794,
      "step": 2233
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7677528493049346,
      "learning_rate": 1.513700421443609e-08,
      "loss": 0.2807,
      "step": 2234
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8014768836126841,
      "learning_rate": 1.4591837119186104e-08,
      "loss": 0.2795,
      "step": 2235
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7557075500586822,
      "learning_rate": 1.4056653791108543e-08,
      "loss": 0.2824,
      "step": 2236
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.760084773244274,
      "learning_rate": 1.3531455301960539e-08,
      "loss": 0.2751,
      "step": 2237
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7887349951909567,
      "learning_rate": 1.3016242703503545e-08,
      "loss": 0.2908,
      "step": 2238
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8145405440642419,
      "learning_rate": 1.2511017027501682e-08,
      "loss": 0.2833,
      "step": 2239
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8098023324377155,
      "learning_rate": 1.2015779285717844e-08,
      "loss": 0.2998,
      "step": 2240
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7251800713361966,
      "learning_rate": 1.1530530469914258e-08,
      "loss": 0.2474,
      "step": 2241
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7361851622835689,
      "learning_rate": 1.1055271551849156e-08,
      "loss": 0.2492,
      "step": 2242
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7385899352791996,
      "learning_rate": 1.0590003483275657e-08,
      "loss": 0.2561,
      "step": 2243
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7412725723443051,
      "learning_rate": 1.0134727195937332e-08,
      "loss": 0.2579,
      "step": 2244
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7763271216130152,
      "learning_rate": 9.689443601570425e-09,
      "loss": 0.279,
      "step": 2245
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8026954627412234,
      "learning_rate": 9.254153591898852e-09,
      "loss": 0.2839,
      "step": 2246
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7402635865991887,
      "learning_rate": 8.828858038632538e-09,
      "loss": 0.2755,
      "step": 2247
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7634625349849409,
      "learning_rate": 8.413557793467974e-09,
      "loss": 0.2702,
      "step": 2248
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7429349515623473,
      "learning_rate": 8.008253688084888e-09,
      "loss": 0.2587,
      "step": 2249
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8059818060370135,
      "learning_rate": 7.612946534143461e-09,
      "loss": 0.2743,
      "step": 2250
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7535555934417596,
      "learning_rate": 7.227637123285447e-09,
      "loss": 0.2594,
      "step": 2251
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7650665721014813,
      "learning_rate": 6.852326227130835e-09,
      "loss": 0.2569,
      "step": 2252
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7617701040007823,
      "learning_rate": 6.487014597275631e-09,
      "loss": 0.28,
      "step": 2253
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6927683087245518,
      "learning_rate": 6.1317029652929734e-09,
      "loss": 0.2375,
      "step": 2254
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7472043763072015,
      "learning_rate": 5.7863920427286834e-09,
      "loss": 0.2587,
      "step": 2255
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7274142525881988,
      "learning_rate": 5.45108252110127e-09,
      "loss": 0.2766,
      "step": 2256
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7488554051835796,
      "learning_rate": 5.125775071901373e-09,
      "loss": 0.2651,
      "step": 2257
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7506049167317456,
      "learning_rate": 4.810470346588992e-09,
      "loss": 0.2611,
      "step": 2258
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7560597477541602,
      "learning_rate": 4.505168976592922e-09,
      "loss": 0.254,
      "step": 2259
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7541918236553519,
      "learning_rate": 4.209871573307989e-09,
      "loss": 0.2551,
      "step": 2260
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7292129254888725,
      "learning_rate": 3.924578728096707e-09,
      "loss": 0.2678,
      "step": 2261
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.73086531016034,
      "learning_rate": 3.6492910122859493e-09,
      "loss": 0.2591,
      "step": 2262
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7339666152236763,
      "learning_rate": 3.3840089771658424e-09,
      "loss": 0.2538,
      "step": 2263
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7776345262336998,
      "learning_rate": 3.1287331539903155e-09,
      "loss": 0.2707,
      "step": 2264
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7711390910494883,
      "learning_rate": 2.8834640539737723e-09,
      "loss": 0.2743,
      "step": 2265
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8062718429898951,
      "learning_rate": 2.648202168291092e-09,
      "loss": 0.3111,
      "step": 2266
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8009711840737762,
      "learning_rate": 2.422947968078182e-09,
      "loss": 0.276,
      "step": 2267
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7428555121180189,
      "learning_rate": 2.2077019044280946e-09,
      "loss": 0.2839,
      "step": 2268
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7547826622093031,
      "learning_rate": 2.002464408392135e-09,
      "loss": 0.2692,
      "step": 2269
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7414608060501388,
      "learning_rate": 1.8072358909781983e-09,
      "loss": 0.2573,
      "step": 2270
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7610011660135121,
      "learning_rate": 1.622016743150212e-09,
      "loss": 0.2668,
      "step": 2271
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7473336084560271,
      "learning_rate": 1.4468073358286926e-09,
      "loss": 0.2549,
      "step": 2272
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7295567649374716,
      "learning_rate": 1.2816080198868596e-09,
      "loss": 0.2454,
      "step": 2273
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7661127524928036,
      "learning_rate": 1.1264191261528557e-09,
      "loss": 0.2857,
      "step": 2274
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.76847932597934,
      "learning_rate": 9.812409654075261e-10,
      "loss": 0.2857,
      "step": 2275
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7237312118877681,
      "learning_rate": 8.460738283838643e-10,
      "loss": 0.2685,
      "step": 2276
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8176233907062571,
      "learning_rate": 7.209179857675664e-10,
      "loss": 0.3186,
      "step": 2277
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7756874528542166,
      "learning_rate": 6.057736881959209e-10,
      "loss": 0.2756,
      "step": 2278
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7213375717115983,
      "learning_rate": 5.006411662555888e-10,
      "loss": 0.2718,
      "step": 2279
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7449685566847628,
      "learning_rate": 4.0552063048593384e-10,
      "loss": 0.2678,
      "step": 2280
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8102459024164149,
      "learning_rate": 3.204122713740265e-10,
      "loss": 0.2984,
      "step": 2281
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7640182085150641,
      "learning_rate": 2.453162593585301e-10,
      "loss": 0.2616,
      "step": 2282
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7373330972319527,
      "learning_rate": 1.8023274482636965e-10,
      "loss": 0.2657,
      "step": 2283
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6999058074255728,
      "learning_rate": 1.251618581127323e-10,
      "loss": 0.2708,
      "step": 2284
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7470331640039006,
      "learning_rate": 8.01037095038426e-11,
      "loss": 0.2618,
      "step": 2285
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7974929379651826,
      "learning_rate": 4.505838923252181e-11,
      "loss": 0.2694,
      "step": 2286
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7530250623015052,
      "learning_rate": 2.0025967479853082e-11,
      "loss": 0.2487,
      "step": 2287
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7435145249858726,
      "learning_rate": 5.006494376291748e-12,
      "loss": 0.2523,
      "step": 2288
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7941585673691149,
      "learning_rate": 0.0,
      "loss": 0.2499,
      "step": 2289
    },
    {
      "epoch": 1.0,
      "step": 2289,
      "total_flos": 3620033072070656.0,
      "train_loss": 0.299031489512034,
      "train_runtime": 34916.5013,
      "train_samples_per_second": 8.394,
      "train_steps_per_second": 0.066
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 2289,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "total_flos": 3620033072070656.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}