{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9999258581914342,
  "eval_steps": 500,
  "global_step": 10115,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 9.885574475446704e-05,
      "grad_norm": 30.475161450255904,
      "learning_rate": 6.578947368421052e-10,
      "loss": 1.3543,
      "step": 1
    },
    {
      "epoch": 0.00019771148950893408,
      "grad_norm": 25.63627551623113,
      "learning_rate": 1.3157894736842103e-09,
      "loss": 1.2477,
      "step": 2
    },
    {
      "epoch": 0.0002965672342634011,
      "grad_norm": 26.880189530771116,
      "learning_rate": 1.9736842105263157e-09,
      "loss": 1.2169,
      "step": 3
    },
    {
      "epoch": 0.00039542297901786817,
      "grad_norm": 29.204642281541645,
      "learning_rate": 2.6315789473684206e-09,
      "loss": 1.2682,
      "step": 4
    },
    {
      "epoch": 0.0004942787237723353,
      "grad_norm": 20.69177911235117,
      "learning_rate": 3.289473684210526e-09,
      "loss": 1.2487,
      "step": 5
    },
    {
      "epoch": 0.0005931344685268023,
      "grad_norm": 33.62965161068742,
      "learning_rate": 3.947368421052631e-09,
      "loss": 1.1952,
      "step": 6
    },
    {
      "epoch": 0.0006919902132812693,
      "grad_norm": 25.916269777673786,
      "learning_rate": 4.605263157894737e-09,
      "loss": 1.364,
      "step": 7
    },
    {
      "epoch": 0.0007908459580357363,
      "grad_norm": 27.53094707131892,
      "learning_rate": 5.263157894736841e-09,
      "loss": 1.4426,
      "step": 8
    },
    {
      "epoch": 0.0008897017027902034,
      "grad_norm": 22.741627054268797,
      "learning_rate": 5.921052631578947e-09,
      "loss": 1.258,
      "step": 9
    },
    {
      "epoch": 0.0009885574475446705,
      "grad_norm": 24.524525686982866,
      "learning_rate": 6.578947368421052e-09,
      "loss": 1.2817,
      "step": 10
    },
    {
      "epoch": 0.0010874131922991375,
      "grad_norm": 19.618340096500052,
      "learning_rate": 7.236842105263158e-09,
      "loss": 1.2174,
      "step": 11
    },
    {
      "epoch": 0.0011862689370536045,
      "grad_norm": 30.421092565807395,
      "learning_rate": 7.894736842105263e-09,
      "loss": 1.3101,
      "step": 12
    },
    {
      "epoch": 0.0012851246818080715,
      "grad_norm": 23.581419126592415,
      "learning_rate": 8.552631578947369e-09,
      "loss": 1.3247,
      "step": 13
    },
    {
      "epoch": 0.0013839804265625387,
      "grad_norm": 23.626275912522168,
      "learning_rate": 9.210526315789473e-09,
      "loss": 1.2552,
      "step": 14
    },
    {
      "epoch": 0.0014828361713170057,
      "grad_norm": 21.477523157020613,
      "learning_rate": 9.868421052631578e-09,
      "loss": 1.2507,
      "step": 15
    },
    {
      "epoch": 0.0015816919160714727,
      "grad_norm": 30.820945395036834,
      "learning_rate": 1.0526315789473683e-08,
      "loss": 1.2954,
      "step": 16
    },
    {
      "epoch": 0.0016805476608259397,
      "grad_norm": 23.824876326760428,
      "learning_rate": 1.1184210526315789e-08,
      "loss": 1.3248,
      "step": 17
    },
    {
      "epoch": 0.0017794034055804069,
      "grad_norm": 36.066258295720075,
      "learning_rate": 1.1842105263157893e-08,
      "loss": 1.233,
      "step": 18
    },
    {
      "epoch": 0.0018782591503348739,
      "grad_norm": 21.94762414426342,
      "learning_rate": 1.25e-08,
      "loss": 1.2854,
      "step": 19
    },
    {
      "epoch": 0.001977114895089341,
      "grad_norm": 21.681910498793197,
      "learning_rate": 1.3157894736842104e-08,
      "loss": 1.3429,
      "step": 20
    },
    {
      "epoch": 0.002075970639843808,
      "grad_norm": 16.688943321500336,
      "learning_rate": 1.3815789473684212e-08,
      "loss": 1.1988,
      "step": 21
    },
    {
      "epoch": 0.002174826384598275,
      "grad_norm": 36.416719109873455,
      "learning_rate": 1.4473684210526316e-08,
      "loss": 1.2642,
      "step": 22
    },
    {
      "epoch": 0.002273682129352742,
      "grad_norm": 34.54437953061398,
      "learning_rate": 1.513157894736842e-08,
      "loss": 1.3295,
      "step": 23
    },
    {
      "epoch": 0.002372537874107209,
      "grad_norm": 26.632002477591357,
      "learning_rate": 1.5789473684210525e-08,
      "loss": 1.3022,
      "step": 24
    },
    {
      "epoch": 0.002471393618861676,
      "grad_norm": 29.71648252728109,
      "learning_rate": 1.6447368421052633e-08,
      "loss": 1.3483,
      "step": 25
    },
    {
      "epoch": 0.002570249363616143,
      "grad_norm": 23.699256345176092,
      "learning_rate": 1.7105263157894738e-08,
      "loss": 1.2494,
      "step": 26
    },
    {
      "epoch": 0.00266910510837061,
      "grad_norm": 94.40521841909921,
      "learning_rate": 1.7763157894736842e-08,
      "loss": 1.3368,
      "step": 27
    },
    {
      "epoch": 0.0027679608531250774,
      "grad_norm": 22.203142185465968,
      "learning_rate": 1.8421052631578947e-08,
      "loss": 1.259,
      "step": 28
    },
    {
      "epoch": 0.0028668165978795444,
      "grad_norm": 19.614181402952656,
      "learning_rate": 1.907894736842105e-08,
      "loss": 1.1969,
      "step": 29
    },
    {
      "epoch": 0.0029656723426340114,
      "grad_norm": 26.376973408013956,
      "learning_rate": 1.9736842105263156e-08,
      "loss": 1.279,
      "step": 30
    },
    {
      "epoch": 0.0030645280873884784,
      "grad_norm": 21.422194350149937,
      "learning_rate": 2.039473684210526e-08,
      "loss": 1.301,
      "step": 31
    },
    {
      "epoch": 0.0031633838321429453,
      "grad_norm": 29.58679833523307,
      "learning_rate": 2.1052631578947365e-08,
      "loss": 1.2699,
      "step": 32
    },
    {
      "epoch": 0.0032622395768974123,
      "grad_norm": 25.45352992934416,
      "learning_rate": 2.1710526315789473e-08,
      "loss": 1.1979,
      "step": 33
    },
    {
      "epoch": 0.0033610953216518793,
      "grad_norm": 27.033596859445943,
      "learning_rate": 2.2368421052631577e-08,
      "loss": 1.1732,
      "step": 34
    },
    {
      "epoch": 0.0034599510664063467,
      "grad_norm": 23.589954718316708,
      "learning_rate": 2.3026315789473682e-08,
      "loss": 1.3077,
      "step": 35
    },
    {
      "epoch": 0.0035588068111608137,
      "grad_norm": 19.28129884303676,
      "learning_rate": 2.3684210526315786e-08,
      "loss": 1.2949,
      "step": 36
    },
    {
      "epoch": 0.0036576625559152807,
      "grad_norm": 30.998092460764106,
      "learning_rate": 2.4342105263157894e-08,
      "loss": 1.3412,
      "step": 37
    },
    {
      "epoch": 0.0037565183006697477,
      "grad_norm": 24.41733409673652,
      "learning_rate": 2.5e-08,
      "loss": 1.334,
      "step": 38
    },
    {
      "epoch": 0.0038553740454242147,
      "grad_norm": 18.129556875001967,
      "learning_rate": 2.5657894736842107e-08,
      "loss": 1.2206,
      "step": 39
    },
    {
      "epoch": 0.003954229790178682,
      "grad_norm": 27.09188453575059,
      "learning_rate": 2.6315789473684208e-08,
      "loss": 1.3589,
      "step": 40
    },
    {
      "epoch": 0.004053085534933149,
      "grad_norm": 23.938554293056164,
      "learning_rate": 2.6973684210526316e-08,
      "loss": 1.2469,
      "step": 41
    },
    {
      "epoch": 0.004151941279687616,
      "grad_norm": 38.17807070933581,
      "learning_rate": 2.7631578947368424e-08,
      "loss": 1.2433,
      "step": 42
    },
    {
      "epoch": 0.004250797024442083,
      "grad_norm": 18.963305645131506,
      "learning_rate": 2.8289473684210525e-08,
      "loss": 1.2614,
      "step": 43
    },
    {
      "epoch": 0.00434965276919655,
      "grad_norm": 25.80391305450501,
      "learning_rate": 2.8947368421052633e-08,
      "loss": 1.1906,
      "step": 44
    },
    {
      "epoch": 0.004448508513951017,
      "grad_norm": 28.059919956982725,
      "learning_rate": 2.9605263157894734e-08,
      "loss": 1.3289,
      "step": 45
    },
    {
      "epoch": 0.004547364258705484,
      "grad_norm": 30.02594751552414,
      "learning_rate": 3.026315789473684e-08,
      "loss": 1.2145,
      "step": 46
    },
    {
      "epoch": 0.0046462200034599515,
      "grad_norm": 34.83838131155323,
      "learning_rate": 3.0921052631578946e-08,
      "loss": 1.3216,
      "step": 47
    },
    {
      "epoch": 0.004745075748214418,
      "grad_norm": 27.90915268612838,
      "learning_rate": 3.157894736842105e-08,
      "loss": 1.1785,
      "step": 48
    },
    {
      "epoch": 0.0048439314929688854,
      "grad_norm": 24.931782308267127,
      "learning_rate": 3.2236842105263155e-08,
      "loss": 1.2273,
      "step": 49
    },
    {
      "epoch": 0.004942787237723352,
      "grad_norm": 32.21006520601201,
      "learning_rate": 3.2894736842105267e-08,
      "loss": 1.2628,
      "step": 50
    },
    {
      "epoch": 0.005041642982477819,
      "grad_norm": 26.164724322584906,
      "learning_rate": 3.3552631578947364e-08,
      "loss": 1.2355,
      "step": 51
    },
    {
      "epoch": 0.005140498727232286,
      "grad_norm": 28.95837805454712,
      "learning_rate": 3.4210526315789476e-08,
      "loss": 1.3627,
      "step": 52
    },
    {
      "epoch": 0.005239354471986753,
      "grad_norm": 20.649015924877492,
      "learning_rate": 3.4868421052631574e-08,
      "loss": 1.2143,
      "step": 53
    },
    {
      "epoch": 0.00533821021674122,
      "grad_norm": 17.220123821205775,
      "learning_rate": 3.5526315789473685e-08,
      "loss": 1.2758,
      "step": 54
    },
    {
      "epoch": 0.005437065961495687,
      "grad_norm": 36.53416030748279,
      "learning_rate": 3.618421052631579e-08,
      "loss": 1.2415,
      "step": 55
    },
    {
      "epoch": 0.005535921706250155,
      "grad_norm": 23.336546800404633,
      "learning_rate": 3.6842105263157894e-08,
      "loss": 1.3166,
      "step": 56
    },
    {
      "epoch": 0.005634777451004621,
      "grad_norm": 20.190739753364692,
      "learning_rate": 3.75e-08,
      "loss": 1.213,
      "step": 57
    },
    {
      "epoch": 0.005733633195759089,
      "grad_norm": 22.638034053855648,
      "learning_rate": 3.81578947368421e-08,
      "loss": 1.1663,
      "step": 58
    },
    {
      "epoch": 0.005832488940513555,
      "grad_norm": 28.52408410319144,
      "learning_rate": 3.881578947368421e-08,
      "loss": 1.1402,
      "step": 59
    },
    {
      "epoch": 0.005931344685268023,
      "grad_norm": 21.198308066537084,
      "learning_rate": 3.947368421052631e-08,
      "loss": 1.3457,
      "step": 60
    },
    {
      "epoch": 0.006030200430022489,
      "grad_norm": 27.24335295144261,
      "learning_rate": 4.013157894736842e-08,
      "loss": 1.3207,
      "step": 61
    },
    {
      "epoch": 0.006129056174776957,
      "grad_norm": 18.59161301136332,
      "learning_rate": 4.078947368421052e-08,
      "loss": 1.1665,
      "step": 62
    },
    {
      "epoch": 0.006227911919531424,
      "grad_norm": 15.292867095245674,
      "learning_rate": 4.144736842105263e-08,
      "loss": 1.173,
      "step": 63
    },
    {
      "epoch": 0.006326767664285891,
      "grad_norm": 22.11688269338932,
      "learning_rate": 4.210526315789473e-08,
      "loss": 1.2131,
      "step": 64
    },
    {
      "epoch": 0.006425623409040358,
      "grad_norm": 27.506214801818913,
      "learning_rate": 4.276315789473684e-08,
      "loss": 1.2134,
      "step": 65
    },
    {
      "epoch": 0.006524479153794825,
      "grad_norm": 26.107873774741005,
      "learning_rate": 4.3421052631578946e-08,
      "loss": 1.2446,
      "step": 66
    },
    {
      "epoch": 0.006623334898549292,
      "grad_norm": 22.452661046804067,
      "learning_rate": 4.407894736842105e-08,
      "loss": 1.2146,
      "step": 67
    },
    {
      "epoch": 0.006722190643303759,
      "grad_norm": 36.52498015327471,
      "learning_rate": 4.4736842105263155e-08,
      "loss": 1.2417,
      "step": 68
    },
    {
      "epoch": 0.006821046388058226,
      "grad_norm": 20.18679038264168,
      "learning_rate": 4.5394736842105266e-08,
      "loss": 1.0271,
      "step": 69
    },
    {
      "epoch": 0.0069199021328126935,
      "grad_norm": 33.1648312130604,
      "learning_rate": 4.6052631578947364e-08,
      "loss": 1.1842,
      "step": 70
    },
    {
      "epoch": 0.00701875787756716,
      "grad_norm": 32.75296826907222,
      "learning_rate": 4.6710526315789475e-08,
      "loss": 1.2243,
      "step": 71
    },
    {
      "epoch": 0.0071176136223216275,
      "grad_norm": 20.59350005334737,
      "learning_rate": 4.736842105263157e-08,
      "loss": 1.1486,
      "step": 72
    },
    {
      "epoch": 0.007216469367076094,
      "grad_norm": 21.834523901628476,
      "learning_rate": 4.8026315789473684e-08,
      "loss": 1.3099,
      "step": 73
    },
    {
      "epoch": 0.007315325111830561,
      "grad_norm": 28.44933168118671,
      "learning_rate": 4.868421052631579e-08,
      "loss": 1.1638,
      "step": 74
    },
    {
      "epoch": 0.007414180856585028,
      "grad_norm": 24.847112761036428,
      "learning_rate": 4.934210526315789e-08,
      "loss": 1.1478,
      "step": 75
    },
    {
      "epoch": 0.007513036601339495,
      "grad_norm": 21.335765633115127,
      "learning_rate": 5e-08,
      "loss": 1.0836,
      "step": 76
    },
    {
      "epoch": 0.007611892346093962,
      "grad_norm": 21.40429709802549,
      "learning_rate": 5.06578947368421e-08,
      "loss": 1.1951,
      "step": 77
    },
    {
      "epoch": 0.007710748090848429,
      "grad_norm": 21.6013249534994,
      "learning_rate": 5.1315789473684213e-08,
      "loss": 1.1567,
      "step": 78
    },
    {
      "epoch": 0.007809603835602897,
      "grad_norm": 19.529375441496942,
      "learning_rate": 5.197368421052631e-08,
      "loss": 1.2136,
      "step": 79
    },
    {
      "epoch": 0.007908459580357364,
      "grad_norm": 21.363164829302917,
      "learning_rate": 5.2631578947368416e-08,
      "loss": 1.2903,
      "step": 80
    },
    {
      "epoch": 0.00800731532511183,
      "grad_norm": 19.92070819296784,
      "learning_rate": 5.328947368421053e-08,
      "loss": 1.1887,
      "step": 81
    },
    {
      "epoch": 0.008106171069866297,
      "grad_norm": 22.045875357228017,
      "learning_rate": 5.394736842105263e-08,
      "loss": 1.1822,
      "step": 82
    },
    {
      "epoch": 0.008205026814620764,
      "grad_norm": 18.353684377040466,
      "learning_rate": 5.460526315789473e-08,
      "loss": 1.1455,
      "step": 83
    },
    {
      "epoch": 0.008303882559375232,
      "grad_norm": 20.06687356284966,
      "learning_rate": 5.526315789473685e-08,
      "loss": 1.1625,
      "step": 84
    },
    {
      "epoch": 0.008402738304129699,
      "grad_norm": 20.4056699355096,
      "learning_rate": 5.5921052631578945e-08,
      "loss": 1.1528,
      "step": 85
    },
    {
      "epoch": 0.008501594048884165,
      "grad_norm": 19.782015490514596,
      "learning_rate": 5.657894736842105e-08,
      "loss": 1.168,
      "step": 86
    },
    {
      "epoch": 0.008600449793638634,
      "grad_norm": 19.80399703853161,
      "learning_rate": 5.723684210526315e-08,
      "loss": 1.1928,
      "step": 87
    },
    {
      "epoch": 0.0086993055383931,
      "grad_norm": 19.14520180869539,
      "learning_rate": 5.7894736842105265e-08,
      "loss": 1.1129,
      "step": 88
    },
    {
      "epoch": 0.008798161283147567,
      "grad_norm": 25.11034976296009,
      "learning_rate": 5.8552631578947363e-08,
      "loss": 1.2428,
      "step": 89
    },
    {
      "epoch": 0.008897017027902033,
      "grad_norm": 22.554201467008006,
      "learning_rate": 5.921052631578947e-08,
      "loss": 1.1356,
      "step": 90
    },
    {
      "epoch": 0.008995872772656502,
      "grad_norm": 19.431948238616226,
      "learning_rate": 5.986842105263158e-08,
      "loss": 1.0743,
      "step": 91
    },
    {
      "epoch": 0.009094728517410968,
      "grad_norm": 32.468351431823194,
      "learning_rate": 6.052631578947368e-08,
      "loss": 1.1539,
      "step": 92
    },
    {
      "epoch": 0.009193584262165435,
      "grad_norm": 12.786904270549623,
      "learning_rate": 6.118421052631579e-08,
      "loss": 1.0592,
      "step": 93
    },
    {
      "epoch": 0.009292440006919903,
      "grad_norm": 18.29961861137409,
      "learning_rate": 6.184210526315789e-08,
      "loss": 1.1845,
      "step": 94
    },
    {
      "epoch": 0.00939129575167437,
      "grad_norm": 13.181401174381381,
      "learning_rate": 6.25e-08,
      "loss": 1.116,
      "step": 95
    },
    {
      "epoch": 0.009490151496428836,
      "grad_norm": 17.144426090808107,
      "learning_rate": 6.31578947368421e-08,
      "loss": 1.1364,
      "step": 96
    },
    {
      "epoch": 0.009589007241183303,
      "grad_norm": 25.228401337522772,
      "learning_rate": 6.38157894736842e-08,
      "loss": 1.1139,
      "step": 97
    },
    {
      "epoch": 0.009687862985937771,
      "grad_norm": 17.768356712511643,
      "learning_rate": 6.447368421052631e-08,
      "loss": 1.1066,
      "step": 98
    },
    {
      "epoch": 0.009786718730692237,
      "grad_norm": 16.849595142890006,
      "learning_rate": 6.513157894736842e-08,
      "loss": 1.071,
      "step": 99
    },
    {
      "epoch": 0.009885574475446704,
      "grad_norm": 12.668678548805174,
      "learning_rate": 6.578947368421053e-08,
      "loss": 1.0526,
      "step": 100
    },
    {
      "epoch": 0.009984430220201172,
      "grad_norm": 16.20538694127326,
      "learning_rate": 6.644736842105262e-08,
      "loss": 1.1871,
      "step": 101
    },
    {
      "epoch": 0.010083285964955639,
      "grad_norm": 38.05575626758174,
      "learning_rate": 6.710526315789473e-08,
      "loss": 1.1158,
      "step": 102
    },
    {
      "epoch": 0.010182141709710105,
      "grad_norm": 13.064750354178049,
      "learning_rate": 6.776315789473685e-08,
      "loss": 1.0432,
      "step": 103
    },
    {
      "epoch": 0.010280997454464572,
      "grad_norm": 14.03702608522195,
      "learning_rate": 6.842105263157895e-08,
      "loss": 1.1229,
      "step": 104
    },
    {
      "epoch": 0.01037985319921904,
      "grad_norm": 22.644509574103132,
      "learning_rate": 6.907894736842104e-08,
      "loss": 1.128,
      "step": 105
    },
    {
      "epoch": 0.010478708943973507,
      "grad_norm": 39.490779033555945,
      "learning_rate": 6.973684210526315e-08,
      "loss": 1.1456,
      "step": 106
    },
    {
      "epoch": 0.010577564688727973,
      "grad_norm": 12.934347664611838,
      "learning_rate": 7.039473684210526e-08,
      "loss": 1.0575,
      "step": 107
    },
    {
      "epoch": 0.01067642043348244,
      "grad_norm": 13.727672650380955,
      "learning_rate": 7.105263157894737e-08,
      "loss": 1.0456,
      "step": 108
    },
    {
      "epoch": 0.010775276178236908,
      "grad_norm": 18.287886771273506,
      "learning_rate": 7.171052631578946e-08,
      "loss": 1.0733,
      "step": 109
    },
    {
      "epoch": 0.010874131922991375,
      "grad_norm": 18.510818251501302,
      "learning_rate": 7.236842105263158e-08,
      "loss": 1.1613,
      "step": 110
    },
    {
      "epoch": 0.010972987667745841,
      "grad_norm": 25.108705441896266,
      "learning_rate": 7.302631578947368e-08,
      "loss": 1.0179,
      "step": 111
    },
    {
      "epoch": 0.01107184341250031,
      "grad_norm": 13.091098494263713,
      "learning_rate": 7.368421052631579e-08,
      "loss": 1.0004,
      "step": 112
    },
    {
      "epoch": 0.011170699157254776,
      "grad_norm": 10.93754505557004,
      "learning_rate": 7.434210526315789e-08,
      "loss": 1.0607,
      "step": 113
    },
    {
      "epoch": 0.011269554902009243,
      "grad_norm": 10.694229988651172,
      "learning_rate": 7.5e-08,
      "loss": 1.0775,
      "step": 114
    },
    {
      "epoch": 0.01136841064676371,
      "grad_norm": 11.588821552314744,
      "learning_rate": 7.56578947368421e-08,
      "loss": 1.1043,
      "step": 115
    },
    {
      "epoch": 0.011467266391518178,
      "grad_norm": 10.213282833720534,
      "learning_rate": 7.63157894736842e-08,
      "loss": 1.0575,
      "step": 116
    },
    {
      "epoch": 0.011566122136272644,
      "grad_norm": 24.529333519075678,
      "learning_rate": 7.697368421052631e-08,
      "loss": 1.0857,
      "step": 117
    },
    {
      "epoch": 0.01166497788102711,
      "grad_norm": 20.887484558964072,
      "learning_rate": 7.763157894736841e-08,
      "loss": 1.0521,
      "step": 118
    },
    {
      "epoch": 0.011763833625781579,
      "grad_norm": 13.900651669621821,
      "learning_rate": 7.828947368421053e-08,
      "loss": 1.0863,
      "step": 119
    },
    {
      "epoch": 0.011862689370536045,
      "grad_norm": 9.876241253694515,
      "learning_rate": 7.894736842105262e-08,
      "loss": 1.1065,
      "step": 120
    },
    {
      "epoch": 0.011961545115290512,
      "grad_norm": 17.44896731924779,
      "learning_rate": 7.960526315789473e-08,
      "loss": 1.0366,
      "step": 121
    },
    {
      "epoch": 0.012060400860044979,
      "grad_norm": 8.629138629726691,
      "learning_rate": 8.026315789473685e-08,
      "loss": 1.1312,
      "step": 122
    },
    {
      "epoch": 0.012159256604799447,
      "grad_norm": 8.70651247663794,
      "learning_rate": 8.092105263157895e-08,
      "loss": 1.0206,
      "step": 123
    },
    {
      "epoch": 0.012258112349553913,
      "grad_norm": 17.311173406077977,
      "learning_rate": 8.157894736842104e-08,
      "loss": 1.0678,
      "step": 124
    },
    {
      "epoch": 0.01235696809430838,
      "grad_norm": 13.257121528064598,
      "learning_rate": 8.223684210526315e-08,
      "loss": 0.9714,
      "step": 125
    },
    {
      "epoch": 0.012455823839062848,
      "grad_norm": 9.384450323703225,
      "learning_rate": 8.289473684210526e-08,
      "loss": 1.0012,
      "step": 126
    },
    {
      "epoch": 0.012554679583817315,
      "grad_norm": 8.932883850126371,
      "learning_rate": 8.355263157894737e-08,
      "loss": 0.9687,
      "step": 127
    },
    {
      "epoch": 0.012653535328571781,
      "grad_norm": 12.800751030963688,
      "learning_rate": 8.421052631578946e-08,
      "loss": 1.0067,
      "step": 128
    },
    {
      "epoch": 0.012752391073326248,
      "grad_norm": 9.220450048295941,
      "learning_rate": 8.486842105263158e-08,
      "loss": 1.0789,
      "step": 129
    },
    {
      "epoch": 0.012851246818080716,
      "grad_norm": 22.67630905748596,
      "learning_rate": 8.552631578947368e-08,
      "loss": 0.962,
      "step": 130
    },
    {
      "epoch": 0.012950102562835183,
      "grad_norm": 10.49845358818206,
      "learning_rate": 8.618421052631579e-08,
      "loss": 1.0665,
      "step": 131
    },
    {
      "epoch": 0.01304895830758965,
      "grad_norm": 15.38429172795839,
      "learning_rate": 8.684210526315789e-08,
      "loss": 0.9817,
      "step": 132
    },
    {
      "epoch": 0.013147814052344118,
      "grad_norm": 13.459637925702838,
      "learning_rate": 8.75e-08,
      "loss": 1.0242,
      "step": 133
    },
    {
      "epoch": 0.013246669797098584,
      "grad_norm": 8.105136238594314,
      "learning_rate": 8.81578947368421e-08,
      "loss": 0.9525,
      "step": 134
    },
    {
      "epoch": 0.01334552554185305,
      "grad_norm": 45.20154563248894,
      "learning_rate": 8.88157894736842e-08,
      "loss": 0.9923,
      "step": 135
    },
    {
      "epoch": 0.013444381286607517,
      "grad_norm": 9.795013178084414,
      "learning_rate": 8.947368421052631e-08,
      "loss": 1.0543,
      "step": 136
    },
    {
      "epoch": 0.013543237031361986,
      "grad_norm": 11.344645171644183,
      "learning_rate": 9.013157894736841e-08,
      "loss": 0.9959,
      "step": 137
    },
    {
      "epoch": 0.013642092776116452,
      "grad_norm": 30.49206720065832,
      "learning_rate": 9.078947368421053e-08,
      "loss": 1.015,
      "step": 138
    },
    {
      "epoch": 0.013740948520870919,
      "grad_norm": 10.73647581982021,
      "learning_rate": 9.144736842105262e-08,
      "loss": 1.0105,
      "step": 139
    },
    {
      "epoch": 0.013839804265625387,
      "grad_norm": 10.426677844741391,
      "learning_rate": 9.210526315789473e-08,
      "loss": 1.0127,
      "step": 140
    },
    {
      "epoch": 0.013938660010379854,
      "grad_norm": 9.93998020191762,
      "learning_rate": 9.276315789473685e-08,
      "loss": 1.0682,
      "step": 141
    },
    {
      "epoch": 0.01403751575513432,
      "grad_norm": 84.9191220115612,
      "learning_rate": 9.342105263157895e-08,
      "loss": 0.9885,
      "step": 142
    },
    {
      "epoch": 0.014136371499888787,
      "grad_norm": 9.60801124579524,
      "learning_rate": 9.407894736842104e-08,
      "loss": 1.0754,
      "step": 143
    },
    {
      "epoch": 0.014235227244643255,
      "grad_norm": 7.352246015543652,
      "learning_rate": 9.473684210526315e-08,
      "loss": 0.9727,
      "step": 144
    },
    {
      "epoch": 0.014334082989397721,
      "grad_norm": 37.867138735653675,
      "learning_rate": 9.539473684210526e-08,
      "loss": 1.0598,
      "step": 145
    },
    {
      "epoch": 0.014432938734152188,
      "grad_norm": 8.738463381307774,
      "learning_rate": 9.605263157894737e-08,
      "loss": 0.9525,
      "step": 146
    },
    {
      "epoch": 0.014531794478906656,
      "grad_norm": 154.06000804647067,
      "learning_rate": 9.671052631578946e-08,
      "loss": 0.989,
      "step": 147
    },
    {
      "epoch": 0.014630650223661123,
      "grad_norm": 18.822958567470394,
      "learning_rate": 9.736842105263158e-08,
      "loss": 1.0833,
      "step": 148
    },
    {
      "epoch": 0.01472950596841559,
      "grad_norm": 8.804414684626433,
      "learning_rate": 9.802631578947368e-08,
      "loss": 1.0046,
      "step": 149
    },
    {
      "epoch": 0.014828361713170056,
      "grad_norm": 15.858054663189922,
      "learning_rate": 9.868421052631579e-08,
      "loss": 1.0,
      "step": 150
    },
    {
      "epoch": 0.014927217457924524,
      "grad_norm": 8.002709504902622,
      "learning_rate": 9.934210526315789e-08,
      "loss": 0.9236,
      "step": 151
    },
    {
      "epoch": 0.01502607320267899,
      "grad_norm": 14.553093668522887,
      "learning_rate": 1e-07,
      "loss": 0.9369,
      "step": 152
    },
    {
      "epoch": 0.015124928947433457,
      "grad_norm": 9.28585115225019,
      "learning_rate": 1.006578947368421e-07,
      "loss": 1.117,
      "step": 153
    },
    {
      "epoch": 0.015223784692187924,
      "grad_norm": 8.54560759406047,
      "learning_rate": 1.013157894736842e-07,
      "loss": 1.0503,
      "step": 154
    },
    {
      "epoch": 0.015322640436942392,
      "grad_norm": 7.67773918625784,
      "learning_rate": 1.019736842105263e-07,
      "loss": 1.0919,
      "step": 155
    },
    {
      "epoch": 0.015421496181696859,
      "grad_norm": 10.42466209953423,
      "learning_rate": 1.0263157894736843e-07,
      "loss": 1.018,
      "step": 156
    },
    {
      "epoch": 0.015520351926451325,
      "grad_norm": 13.379321545328466,
      "learning_rate": 1.0328947368421053e-07,
      "loss": 0.9797,
      "step": 157
    },
    {
      "epoch": 0.015619207671205794,
      "grad_norm": 8.046068101481339,
      "learning_rate": 1.0394736842105262e-07,
      "loss": 0.9958,
      "step": 158
    },
    {
      "epoch": 0.01571806341596026,
      "grad_norm": 7.254624033301213,
      "learning_rate": 1.0460526315789473e-07,
      "loss": 1.1204,
      "step": 159
    },
    {
      "epoch": 0.01581691916071473,
      "grad_norm": 8.036053744601878,
      "learning_rate": 1.0526315789473683e-07,
      "loss": 1.078,
      "step": 160
    },
    {
      "epoch": 0.015915774905469193,
      "grad_norm": 10.96173017141793,
      "learning_rate": 1.0592105263157894e-07,
      "loss": 1.1073,
      "step": 161
    },
    {
      "epoch": 0.01601463065022366,
      "grad_norm": 6.227885279321283,
      "learning_rate": 1.0657894736842105e-07,
      "loss": 0.9296,
      "step": 162
    },
    {
      "epoch": 0.01611348639497813,
      "grad_norm": 6.87831428471293,
      "learning_rate": 1.0723684210526316e-07,
      "loss": 0.969,
      "step": 163
    },
    {
      "epoch": 0.016212342139732595,
      "grad_norm": 8.7835846647056,
      "learning_rate": 1.0789473684210526e-07,
      "loss": 0.9994,
      "step": 164
    },
    {
      "epoch": 0.016311197884487063,
      "grad_norm": 9.585784133415094,
      "learning_rate": 1.0855263157894737e-07,
      "loss": 1.0493,
      "step": 165
    },
    {
      "epoch": 0.016410053629241528,
      "grad_norm": 8.586580508648508,
      "learning_rate": 1.0921052631578946e-07,
      "loss": 1.0069,
      "step": 166
    },
    {
      "epoch": 0.016508909373995996,
      "grad_norm": 7.839914197423952,
      "learning_rate": 1.0986842105263156e-07,
      "loss": 1.0349,
      "step": 167
    },
    {
      "epoch": 0.016607765118750464,
      "grad_norm": 6.214837014620745,
      "learning_rate": 1.105263157894737e-07,
      "loss": 0.948,
      "step": 168
    },
    {
      "epoch": 0.01670662086350493,
      "grad_norm": 8.672882714510084,
      "learning_rate": 1.1118421052631579e-07,
      "loss": 0.9165,
      "step": 169
    },
    {
      "epoch": 0.016805476608259397,
      "grad_norm": 9.19483184084849,
      "learning_rate": 1.1184210526315789e-07,
      "loss": 0.9568,
      "step": 170
    },
    {
      "epoch": 0.016904332353013866,
      "grad_norm": 7.211727957865169,
      "learning_rate": 1.125e-07,
      "loss": 1.0107,
      "step": 171
    },
    {
      "epoch": 0.01700318809776833,
      "grad_norm": 7.000621559410127,
      "learning_rate": 1.131578947368421e-07,
      "loss": 0.9805,
      "step": 172
    },
    {
      "epoch": 0.0171020438425228,
      "grad_norm": 6.776582260245124,
      "learning_rate": 1.138157894736842e-07,
      "loss": 0.9136,
      "step": 173
    },
    {
      "epoch": 0.017200899587277267,
      "grad_norm": 48.80703781971459,
      "learning_rate": 1.144736842105263e-07,
      "loss": 1.0061,
      "step": 174
    },
    {
      "epoch": 0.017299755332031732,
      "grad_norm": 9.899510148202532,
      "learning_rate": 1.1513157894736843e-07,
      "loss": 0.9793,
      "step": 175
    },
    {
      "epoch": 0.0173986110767862,
      "grad_norm": 6.807484060976695,
      "learning_rate": 1.1578947368421053e-07,
      "loss": 0.9438,
      "step": 176
    },
    {
      "epoch": 0.017497466821540665,
      "grad_norm": 9.501054529202943,
      "learning_rate": 1.1644736842105262e-07,
      "loss": 0.9263,
      "step": 177
    },
    {
      "epoch": 0.017596322566295133,
      "grad_norm": 12.684440239429364,
      "learning_rate": 1.1710526315789473e-07,
      "loss": 1.0145,
      "step": 178
    },
    {
      "epoch": 0.0176951783110496,
      "grad_norm": 10.238266556413308,
      "learning_rate": 1.1776315789473683e-07,
      "loss": 0.9481,
      "step": 179
    },
    {
      "epoch": 0.017794034055804066,
      "grad_norm": 11.388249978188357,
      "learning_rate": 1.1842105263157894e-07,
      "loss": 1.0565,
      "step": 180
    },
    {
      "epoch": 0.017892889800558535,
      "grad_norm": 8.638444349599615,
      "learning_rate": 1.1907894736842105e-07,
      "loss": 0.9731,
      "step": 181
    },
    {
      "epoch": 0.017991745545313003,
      "grad_norm": 7.600743407147101,
      "learning_rate": 1.1973684210526316e-07,
      "loss": 0.9954,
      "step": 182
    },
    {
      "epoch": 0.018090601290067468,
      "grad_norm": 8.062910417564103,
      "learning_rate": 1.2039473684210526e-07,
      "loss": 1.0336,
      "step": 183
    },
    {
      "epoch": 0.018189457034821936,
      "grad_norm": 6.845585087062875,
      "learning_rate": 1.2105263157894737e-07,
      "loss": 0.8703,
      "step": 184
    },
    {
      "epoch": 0.018288312779576404,
      "grad_norm": 5.500591921602448,
      "learning_rate": 1.2171052631578947e-07,
      "loss": 0.946,
      "step": 185
    },
    {
      "epoch": 0.01838716852433087,
      "grad_norm": 5.792695121885709,
      "learning_rate": 1.2236842105263158e-07,
      "loss": 0.9055,
      "step": 186
    },
    {
      "epoch": 0.018486024269085338,
      "grad_norm": 6.750291369804461,
      "learning_rate": 1.2302631578947368e-07,
      "loss": 0.9828,
      "step": 187
    },
    {
      "epoch": 0.018584880013839806,
      "grad_norm": 5.66225015799794,
      "learning_rate": 1.2368421052631579e-07,
      "loss": 0.9379,
      "step": 188
    },
    {
      "epoch": 0.01868373575859427,
      "grad_norm": 7.627144143639098,
      "learning_rate": 1.243421052631579e-07,
      "loss": 0.9354,
      "step": 189
    },
    {
      "epoch": 0.01878259150334874,
      "grad_norm": 5.6705022383636265,
      "learning_rate": 1.25e-07,
      "loss": 0.9458,
      "step": 190
    },
    {
      "epoch": 0.018881447248103204,
      "grad_norm": 8.81749430986215,
      "learning_rate": 1.256578947368421e-07,
      "loss": 0.9417,
      "step": 191
    },
    {
      "epoch": 0.018980302992857672,
      "grad_norm": 8.304119562253083,
      "learning_rate": 1.263157894736842e-07,
      "loss": 0.8957,
      "step": 192
    },
    {
      "epoch": 0.01907915873761214,
      "grad_norm": 11.067849371349094,
      "learning_rate": 1.269736842105263e-07,
      "loss": 0.8583,
      "step": 193
    },
    {
      "epoch": 0.019178014482366605,
      "grad_norm": 14.691139380694542,
      "learning_rate": 1.276315789473684e-07,
      "loss": 0.9321,
      "step": 194
    },
    {
      "epoch": 0.019276870227121073,
      "grad_norm": 6.033635379255785,
      "learning_rate": 1.2828947368421052e-07,
      "loss": 0.942,
      "step": 195
    },
    {
      "epoch": 0.019375725971875542,
      "grad_norm": 44.40632926098656,
      "learning_rate": 1.2894736842105262e-07,
      "loss": 1.0129,
      "step": 196
    },
    {
      "epoch": 0.019474581716630007,
      "grad_norm": 5.032443622461142,
      "learning_rate": 1.2960526315789473e-07,
      "loss": 0.9297,
      "step": 197
    },
    {
      "epoch": 0.019573437461384475,
      "grad_norm": 6.505173085220018,
      "learning_rate": 1.3026315789473683e-07,
      "loss": 0.8746,
      "step": 198
    },
    {
      "epoch": 0.019672293206138943,
      "grad_norm": 5.349635522929612,
      "learning_rate": 1.3092105263157894e-07,
      "loss": 0.9155,
      "step": 199
    },
    {
      "epoch": 0.019771148950893408,
      "grad_norm": 50.60483552817676,
      "learning_rate": 1.3157894736842107e-07,
      "loss": 0.9024,
      "step": 200
    },
    {
      "epoch": 0.019870004695647876,
      "grad_norm": 5.695287837213799,
      "learning_rate": 1.3223684210526317e-07,
      "loss": 0.9828,
      "step": 201
    },
    {
      "epoch": 0.019968860440402345,
      "grad_norm": 6.222308626549295,
      "learning_rate": 1.3289473684210525e-07,
      "loss": 0.8506,
      "step": 202
    },
    {
      "epoch": 0.02006771618515681,
      "grad_norm": 6.975139667568593,
      "learning_rate": 1.3355263157894735e-07,
      "loss": 0.9418,
      "step": 203
    },
    {
      "epoch": 0.020166571929911278,
      "grad_norm": 13.365042956743633,
      "learning_rate": 1.3421052631578946e-07,
      "loss": 0.9018,
      "step": 204
    },
    {
      "epoch": 0.020265427674665742,
      "grad_norm": 7.788559187107898,
      "learning_rate": 1.3486842105263156e-07,
      "loss": 0.9469,
      "step": 205
    },
    {
      "epoch": 0.02036428341942021,
      "grad_norm": 4.8050129828736035,
      "learning_rate": 1.355263157894737e-07,
      "loss": 0.9741,
      "step": 206
    },
    {
      "epoch": 0.02046313916417468,
      "grad_norm": 7.160654119075421,
      "learning_rate": 1.361842105263158e-07,
      "loss": 0.9262,
      "step": 207
    },
    {
      "epoch": 0.020561994908929144,
      "grad_norm": 6.100421251572534,
      "learning_rate": 1.368421052631579e-07,
      "loss": 0.9218,
      "step": 208
    },
    {
      "epoch": 0.020660850653683612,
      "grad_norm": 6.737953849181826,
      "learning_rate": 1.375e-07,
      "loss": 0.918,
      "step": 209
    },
    {
      "epoch": 0.02075970639843808,
      "grad_norm": 7.92374117191971,
      "learning_rate": 1.3815789473684209e-07,
      "loss": 0.9205,
      "step": 210
    },
    {
      "epoch": 0.020858562143192545,
      "grad_norm": 5.4721842622805035,
      "learning_rate": 1.388157894736842e-07,
      "loss": 1.0177,
      "step": 211
    },
    {
      "epoch": 0.020957417887947014,
      "grad_norm": 14.430571833713174,
      "learning_rate": 1.394736842105263e-07,
      "loss": 0.8871,
      "step": 212
    },
    {
      "epoch": 0.021056273632701482,
      "grad_norm": 6.42911338349966,
      "learning_rate": 1.4013157894736843e-07,
      "loss": 0.876,
      "step": 213
    },
    {
      "epoch": 0.021155129377455947,
      "grad_norm": 5.88556656635028,
      "learning_rate": 1.4078947368421053e-07,
      "loss": 0.8746,
      "step": 214
    },
    {
      "epoch": 0.021253985122210415,
      "grad_norm": 14.045518583665917,
      "learning_rate": 1.4144736842105263e-07,
      "loss": 0.981,
      "step": 215
    },
    {
      "epoch": 0.02135284086696488,
      "grad_norm": 14.989803515257334,
      "learning_rate": 1.4210526315789474e-07,
      "loss": 0.8946,
      "step": 216
    },
    {
      "epoch": 0.021451696611719348,
      "grad_norm": 4.184925185507607,
      "learning_rate": 1.4276315789473682e-07,
      "loss": 0.9389,
      "step": 217
    },
    {
      "epoch": 0.021550552356473816,
      "grad_norm": 6.687260028808132,
      "learning_rate": 1.4342105263157892e-07,
      "loss": 0.883,
      "step": 218
    },
    {
      "epoch": 0.02164940810122828,
      "grad_norm": 5.630434184179349,
      "learning_rate": 1.4407894736842105e-07,
      "loss": 0.8768,
      "step": 219
    },
    {
      "epoch": 0.02174826384598275,
      "grad_norm": 5.19053635806534,
      "learning_rate": 1.4473684210526316e-07,
      "loss": 0.8423,
      "step": 220
    },
    {
      "epoch": 0.021847119590737218,
      "grad_norm": 5.861033794277367,
      "learning_rate": 1.4539473684210526e-07,
      "loss": 0.9897,
      "step": 221
    },
    {
      "epoch": 0.021945975335491683,
      "grad_norm": 6.0708170823934475,
      "learning_rate": 1.4605263157894737e-07,
      "loss": 0.9804,
      "step": 222
    },
    {
      "epoch": 0.02204483108024615,
      "grad_norm": 7.0080960560379255,
      "learning_rate": 1.4671052631578947e-07,
      "loss": 0.9715,
      "step": 223
    },
    {
      "epoch": 0.02214368682500062,
      "grad_norm": 5.600333528334201,
      "learning_rate": 1.4736842105263158e-07,
      "loss": 0.8283,
      "step": 224
    },
    {
      "epoch": 0.022242542569755084,
      "grad_norm": 8.34310915369233,
      "learning_rate": 1.4802631578947368e-07,
      "loss": 0.7616,
      "step": 225
    },
    {
      "epoch": 0.022341398314509552,
      "grad_norm": 5.157518897388644,
      "learning_rate": 1.4868421052631578e-07,
      "loss": 0.891,
      "step": 226
    },
    {
      "epoch": 0.02244025405926402,
      "grad_norm": 5.9524881787097215,
      "learning_rate": 1.493421052631579e-07,
      "loss": 0.8792,
      "step": 227
    },
    {
      "epoch": 0.022539109804018485,
      "grad_norm": 7.514775598803251,
      "learning_rate": 1.5e-07,
      "loss": 0.9301,
      "step": 228
    },
    {
      "epoch": 0.022637965548772954,
      "grad_norm": 14.046518614465036,
      "learning_rate": 1.506578947368421e-07,
      "loss": 0.9474,
      "step": 229
    },
    {
      "epoch": 0.02273682129352742,
      "grad_norm": 6.192233454523717,
      "learning_rate": 1.513157894736842e-07,
      "loss": 0.8696,
      "step": 230
    },
    {
      "epoch": 0.022835677038281887,
      "grad_norm": 5.978207230614067,
      "learning_rate": 1.519736842105263e-07,
      "loss": 0.9174,
      "step": 231
    },
    {
      "epoch": 0.022934532783036355,
      "grad_norm": 6.578420824460962,
      "learning_rate": 1.526315789473684e-07,
      "loss": 0.9253,
      "step": 232
    },
    {
      "epoch": 0.02303338852779082,
      "grad_norm": 7.009076585244014,
      "learning_rate": 1.5328947368421052e-07,
      "loss": 0.8757,
      "step": 233
    },
    {
      "epoch": 0.023132244272545288,
      "grad_norm": 4.127479188865312,
      "learning_rate": 1.5394736842105262e-07,
      "loss": 0.9192,
      "step": 234
    },
    {
      "epoch": 0.023231100017299756,
      "grad_norm": 17.260180936889498,
      "learning_rate": 1.5460526315789472e-07,
      "loss": 0.9029,
      "step": 235
    },
    {
      "epoch": 0.02332995576205422,
      "grad_norm": 5.665725666958535,
      "learning_rate": 1.5526315789473683e-07,
      "loss": 0.945,
      "step": 236
    },
    {
      "epoch": 0.02342881150680869,
      "grad_norm": 5.4796015900415584,
      "learning_rate": 1.5592105263157893e-07,
      "loss": 0.9806,
      "step": 237
    },
    {
      "epoch": 0.023527667251563158,
      "grad_norm": 6.387678925959827,
      "learning_rate": 1.5657894736842107e-07,
      "loss": 0.9551,
      "step": 238
    },
    {
      "epoch": 0.023626522996317623,
      "grad_norm": 5.534777241888097,
      "learning_rate": 1.5723684210526317e-07,
      "loss": 0.8345,
      "step": 239
    },
    {
      "epoch": 0.02372537874107209,
      "grad_norm": 5.412055751863232,
      "learning_rate": 1.5789473684210525e-07,
      "loss": 0.9603,
      "step": 240
    },
    {
      "epoch": 0.02382423448582656,
      "grad_norm": 4.743483162983015,
      "learning_rate": 1.5855263157894735e-07,
      "loss": 0.8826,
      "step": 241
    },
    {
      "epoch": 0.023923090230581024,
      "grad_norm": 5.762764154527077,
      "learning_rate": 1.5921052631578946e-07,
      "loss": 0.8683,
      "step": 242
    },
    {
      "epoch": 0.024021945975335492,
      "grad_norm": 8.730391086925994,
      "learning_rate": 1.5986842105263156e-07,
      "loss": 0.8318,
      "step": 243
    },
    {
      "epoch": 0.024120801720089957,
      "grad_norm": 5.742619182540114,
      "learning_rate": 1.605263157894737e-07,
      "loss": 0.8868,
      "step": 244
    },
    {
      "epoch": 0.024219657464844425,
      "grad_norm": 8.534223110897912,
      "learning_rate": 1.611842105263158e-07,
      "loss": 0.8559,
      "step": 245
    },
    {
      "epoch": 0.024318513209598894,
      "grad_norm": 32.46803214043959,
      "learning_rate": 1.618421052631579e-07,
      "loss": 0.9414,
      "step": 246
    },
    {
      "epoch": 0.02441736895435336,
      "grad_norm": 6.375318276638526,
      "learning_rate": 1.6249999999999998e-07,
      "loss": 0.9299,
      "step": 247
    },
    {
      "epoch": 0.024516224699107827,
      "grad_norm": 5.979823353380745,
      "learning_rate": 1.6315789473684208e-07,
      "loss": 0.8036,
      "step": 248
    },
    {
      "epoch": 0.024615080443862295,
      "grad_norm": 6.087671317981926,
      "learning_rate": 1.638157894736842e-07,
      "loss": 0.8693,
      "step": 249
    },
    {
      "epoch": 0.02471393618861676,
      "grad_norm": 6.915152298729674,
      "learning_rate": 1.644736842105263e-07,
      "loss": 0.8972,
      "step": 250
    },
    {
      "epoch": 0.024812791933371228,
      "grad_norm": 10.247492978180086,
      "learning_rate": 1.6513157894736842e-07,
      "loss": 0.8578,
      "step": 251
    },
    {
      "epoch": 0.024911647678125697,
      "grad_norm": 9.431763370350696,
      "learning_rate": 1.6578947368421053e-07,
      "loss": 0.8925,
      "step": 252
    },
    {
      "epoch": 0.02501050342288016,
      "grad_norm": 7.194837800136805,
      "learning_rate": 1.6644736842105263e-07,
      "loss": 0.9089,
      "step": 253
    },
    {
      "epoch": 0.02510935916763463,
      "grad_norm": 6.58907781938773,
      "learning_rate": 1.6710526315789474e-07,
      "loss": 0.898,
      "step": 254
    },
    {
      "epoch": 0.025208214912389098,
      "grad_norm": 4.292600996418018,
      "learning_rate": 1.6776315789473682e-07,
      "loss": 0.8772,
      "step": 255
    },
    {
      "epoch": 0.025307070657143563,
      "grad_norm": 11.905136070935857,
      "learning_rate": 1.6842105263157892e-07,
      "loss": 0.8715,
      "step": 256
    },
    {
      "epoch": 0.02540592640189803,
      "grad_norm": 4.699543582800261,
      "learning_rate": 1.6907894736842105e-07,
      "loss": 0.7948,
      "step": 257
    },
    {
      "epoch": 0.025504782146652496,
      "grad_norm": 6.130909306230232,
      "learning_rate": 1.6973684210526316e-07,
      "loss": 0.8896,
      "step": 258
    },
    {
      "epoch": 0.025603637891406964,
      "grad_norm": 7.395915028686098,
      "learning_rate": 1.7039473684210526e-07,
      "loss": 0.892,
      "step": 259
    },
    {
      "epoch": 0.025702493636161432,
      "grad_norm": 34.6930542988766,
      "learning_rate": 1.7105263157894736e-07,
      "loss": 0.8237,
      "step": 260
    },
    {
      "epoch": 0.025801349380915897,
      "grad_norm": 10.254149909100025,
      "learning_rate": 1.7171052631578947e-07,
      "loss": 0.8613,
      "step": 261
    },
    {
      "epoch": 0.025900205125670366,
      "grad_norm": 5.083506746674676,
      "learning_rate": 1.7236842105263157e-07,
      "loss": 0.8866,
      "step": 262
    },
    {
      "epoch": 0.025999060870424834,
      "grad_norm": 4.836952013765102,
      "learning_rate": 1.7302631578947368e-07,
      "loss": 0.8888,
      "step": 263
    },
    {
      "epoch": 0.0260979166151793,
      "grad_norm": 11.414818959540721,
      "learning_rate": 1.7368421052631578e-07,
      "loss": 0.8184,
      "step": 264
    },
    {
      "epoch": 0.026196772359933767,
      "grad_norm": 7.102826819721976,
      "learning_rate": 1.743421052631579e-07,
      "loss": 0.9272,
      "step": 265
    },
    {
      "epoch": 0.026295628104688235,
      "grad_norm": 4.305526545612905,
      "learning_rate": 1.75e-07,
      "loss": 0.9278,
      "step": 266
    },
    {
      "epoch": 0.0263944838494427,
      "grad_norm": 4.136733915711035,
      "learning_rate": 1.756578947368421e-07,
      "loss": 0.7844,
      "step": 267
    },
    {
      "epoch": 0.02649333959419717,
      "grad_norm": 4.096048932106808,
      "learning_rate": 1.763157894736842e-07,
      "loss": 0.9174,
      "step": 268
    },
    {
      "epoch": 0.026592195338951633,
      "grad_norm": 5.937952086656196,
      "learning_rate": 1.769736842105263e-07,
      "loss": 0.8535,
      "step": 269
    },
    {
      "epoch": 0.0266910510837061,
      "grad_norm": 14.045522971307724,
      "learning_rate": 1.776315789473684e-07,
      "loss": 0.8854,
      "step": 270
    },
    {
      "epoch": 0.02678990682846057,
      "grad_norm": 5.691906363431568,
      "learning_rate": 1.7828947368421051e-07,
      "loss": 0.9041,
      "step": 271
    },
    {
      "epoch": 0.026888762573215035,
      "grad_norm": 6.0728266785093155,
      "learning_rate": 1.7894736842105262e-07,
      "loss": 0.8814,
      "step": 272
    },
    {
      "epoch": 0.026987618317969503,
      "grad_norm": 7.196579720309739,
      "learning_rate": 1.7960526315789472e-07,
      "loss": 0.8354,
      "step": 273
    },
    {
      "epoch": 0.02708647406272397,
      "grad_norm": 6.841013900237215,
      "learning_rate": 1.8026315789473683e-07,
      "loss": 0.8896,
      "step": 274
    },
    {
      "epoch": 0.027185329807478436,
      "grad_norm": 4.664238625859023,
      "learning_rate": 1.8092105263157893e-07,
      "loss": 0.9236,
      "step": 275
    },
    {
      "epoch": 0.027284185552232904,
      "grad_norm": 5.890687750366183,
      "learning_rate": 1.8157894736842106e-07,
      "loss": 0.8317,
      "step": 276
    },
    {
      "epoch": 0.027383041296987373,
      "grad_norm": 4.162270080326169,
      "learning_rate": 1.8223684210526317e-07,
      "loss": 0.8367,
      "step": 277
    },
    {
      "epoch": 0.027481897041741837,
      "grad_norm": 6.1790375030473745,
      "learning_rate": 1.8289473684210525e-07,
      "loss": 0.9113,
      "step": 278
    },
    {
      "epoch": 0.027580752786496306,
      "grad_norm": 5.140351816938926,
      "learning_rate": 1.8355263157894735e-07,
      "loss": 0.8163,
      "step": 279
    },
    {
      "epoch": 0.027679608531250774,
      "grad_norm": 9.482714168289796,
      "learning_rate": 1.8421052631578946e-07,
      "loss": 0.9143,
      "step": 280
    },
    {
      "epoch": 0.02777846427600524,
      "grad_norm": 7.179291769102827,
      "learning_rate": 1.8486842105263156e-07,
      "loss": 0.9207,
      "step": 281
    },
    {
      "epoch": 0.027877320020759707,
      "grad_norm": 7.595006199026464,
      "learning_rate": 1.855263157894737e-07,
      "loss": 0.8438,
      "step": 282
    },
    {
      "epoch": 0.027976175765514172,
      "grad_norm": 4.479918478157686,
      "learning_rate": 1.861842105263158e-07,
      "loss": 0.8697,
      "step": 283
    },
    {
      "epoch": 0.02807503151026864,
      "grad_norm": 6.524122071899191,
      "learning_rate": 1.868421052631579e-07,
      "loss": 0.8941,
      "step": 284
    },
    {
      "epoch": 0.02817388725502311,
      "grad_norm": 7.300543531941686,
      "learning_rate": 1.875e-07,
      "loss": 0.7524,
      "step": 285
    },
    {
      "epoch": 0.028272742999777573,
      "grad_norm": 8.132327592430405,
      "learning_rate": 1.8815789473684208e-07,
      "loss": 0.8157,
      "step": 286
    },
    {
      "epoch": 0.02837159874453204,
      "grad_norm": 11.705346345672435,
      "learning_rate": 1.888157894736842e-07,
      "loss": 0.8662,
      "step": 287
    },
    {
      "epoch": 0.02847045448928651,
      "grad_norm": 5.3530130770076445,
      "learning_rate": 1.894736842105263e-07,
      "loss": 0.8341,
      "step": 288
    },
    {
      "epoch": 0.028569310234040975,
      "grad_norm": 6.5029793213155225,
      "learning_rate": 1.9013157894736842e-07,
      "loss": 0.8981,
      "step": 289
    },
    {
      "epoch": 0.028668165978795443,
      "grad_norm": 8.365708325740078,
      "learning_rate": 1.9078947368421053e-07,
      "loss": 0.8404,
      "step": 290
    },
    {
      "epoch": 0.02876702172354991,
      "grad_norm": 5.082720662307608,
      "learning_rate": 1.9144736842105263e-07,
      "loss": 0.8874,
      "step": 291
    },
    {
      "epoch": 0.028865877468304376,
      "grad_norm": 6.541338043696397,
      "learning_rate": 1.9210526315789474e-07,
      "loss": 0.9247,
      "step": 292
    },
    {
      "epoch": 0.028964733213058844,
      "grad_norm": 4.255655761637827,
      "learning_rate": 1.9276315789473681e-07,
      "loss": 0.8499,
      "step": 293
    },
    {
      "epoch": 0.029063588957813313,
      "grad_norm": 8.038414636102766,
      "learning_rate": 1.9342105263157892e-07,
      "loss": 0.7558,
      "step": 294
    },
    {
      "epoch": 0.029162444702567777,
      "grad_norm": 4.06354472740116,
      "learning_rate": 1.9407894736842105e-07,
      "loss": 0.8431,
      "step": 295
    },
    {
      "epoch": 0.029261300447322246,
      "grad_norm": 8.03754489269599,
      "learning_rate": 1.9473684210526315e-07,
      "loss": 0.8378,
      "step": 296
    },
    {
      "epoch": 0.02936015619207671,
      "grad_norm": 10.985117531903805,
      "learning_rate": 1.9539473684210526e-07,
      "loss": 0.9433,
      "step": 297
    },
    {
      "epoch": 0.02945901193683118,
      "grad_norm": 5.196650218876621,
      "learning_rate": 1.9605263157894736e-07,
      "loss": 0.901,
      "step": 298
    },
    {
      "epoch": 0.029557867681585647,
      "grad_norm": 6.141495419134231,
      "learning_rate": 1.9671052631578947e-07,
      "loss": 0.8812,
      "step": 299
    },
    {
      "epoch": 0.029656723426340112,
      "grad_norm": 4.7059109300415445,
      "learning_rate": 1.9736842105263157e-07,
      "loss": 0.8957,
      "step": 300
    },
    {
      "epoch": 0.02975557917109458,
      "grad_norm": 12.684093658926217,
      "learning_rate": 1.9802631578947368e-07,
      "loss": 0.8535,
      "step": 301
    },
    {
      "epoch": 0.02985443491584905,
      "grad_norm": 3.5528956101403706,
      "learning_rate": 1.9868421052631578e-07,
      "loss": 0.768,
      "step": 302
    },
    {
      "epoch": 0.029953290660603513,
      "grad_norm": 11.92314178068368,
      "learning_rate": 1.9934210526315789e-07,
      "loss": 0.8062,
      "step": 303
    },
    {
      "epoch": 0.03005214640535798,
      "grad_norm": 3.7151086410731793,
      "learning_rate": 2e-07,
      "loss": 0.8422,
      "step": 304
    },
    {
      "epoch": 0.03015100215011245,
      "grad_norm": 5.266364512168245,
      "learning_rate": 1.9999999487323755e-07,
      "loss": 0.9344,
      "step": 305
    },
    {
      "epoch": 0.030249857894866915,
      "grad_norm": 6.384002459161972,
      "learning_rate": 1.9999997949295072e-07,
      "loss": 0.832,
      "step": 306
    },
    {
      "epoch": 0.030348713639621383,
      "grad_norm": 4.672920472534386,
      "learning_rate": 1.9999995385914113e-07,
      "loss": 0.7971,
      "step": 307
    },
    {
      "epoch": 0.030447569384375848,
      "grad_norm": 4.788246342785124,
      "learning_rate": 1.9999991797181134e-07,
      "loss": 0.7648,
      "step": 308
    },
    {
      "epoch": 0.030546425129130316,
      "grad_norm": 5.249308298706671,
      "learning_rate": 1.9999987183096508e-07,
      "loss": 0.8745,
      "step": 309
    },
    {
      "epoch": 0.030645280873884784,
      "grad_norm": 26.80512696800146,
      "learning_rate": 1.9999981543660706e-07,
      "loss": 0.8147,
      "step": 310
    },
    {
      "epoch": 0.03074413661863925,
      "grad_norm": 3.4260569534034615,
      "learning_rate": 1.9999974878874305e-07,
      "loss": 0.8392,
      "step": 311
    },
    {
      "epoch": 0.030842992363393718,
      "grad_norm": 3.90712815438934,
      "learning_rate": 1.9999967188737995e-07,
      "loss": 0.8453,
      "step": 312
    },
    {
      "epoch": 0.030941848108148186,
      "grad_norm": 7.835493126132616,
      "learning_rate": 1.9999958473252557e-07,
      "loss": 0.7857,
      "step": 313
    },
    {
      "epoch": 0.03104070385290265,
      "grad_norm": 4.889129625411227,
      "learning_rate": 1.9999948732418885e-07,
      "loss": 0.7823,
      "step": 314
    },
    {
      "epoch": 0.03113955959765712,
      "grad_norm": 22.55418392914248,
      "learning_rate": 1.9999937966237983e-07,
      "loss": 0.8466,
      "step": 315
    },
    {
      "epoch": 0.031238415342411587,
      "grad_norm": 8.018760679336987,
      "learning_rate": 1.9999926174710954e-07,
      "loss": 0.7594,
      "step": 316
    },
    {
      "epoch": 0.03133727108716605,
      "grad_norm": 10.44961557573718,
      "learning_rate": 1.9999913357839e-07,
      "loss": 0.7775,
      "step": 317
    },
    {
      "epoch": 0.03143612683192052,
      "grad_norm": 4.3998926640318965,
      "learning_rate": 1.9999899515623443e-07,
      "loss": 0.8891,
      "step": 318
    },
    {
      "epoch": 0.03153498257667499,
      "grad_norm": 4.035369799976758,
      "learning_rate": 1.9999884648065699e-07,
      "loss": 0.881,
      "step": 319
    },
    {
      "epoch": 0.03163383832142946,
      "grad_norm": 5.334145767204725,
      "learning_rate": 1.9999868755167295e-07,
      "loss": 0.7735,
      "step": 320
    },
    {
      "epoch": 0.03173269406618392,
      "grad_norm": 4.308574675131147,
      "learning_rate": 1.9999851836929857e-07,
      "loss": 0.8619,
      "step": 321
    },
    {
      "epoch": 0.03183154981093839,
      "grad_norm": 5.610229127457844,
      "learning_rate": 1.999983389335512e-07,
      "loss": 0.7464,
      "step": 322
    },
    {
      "epoch": 0.031930405555692855,
      "grad_norm": 8.076606759530424,
      "learning_rate": 1.9999814924444926e-07,
      "loss": 0.831,
      "step": 323
    },
    {
      "epoch": 0.03202926130044732,
      "grad_norm": 3.9371867728901284,
      "learning_rate": 1.999979493020122e-07,
      "loss": 0.8347,
      "step": 324
    },
    {
      "epoch": 0.03212811704520179,
      "grad_norm": 6.7604657597178965,
      "learning_rate": 1.9999773910626052e-07,
      "loss": 0.8846,
      "step": 325
    },
    {
      "epoch": 0.03222697278995626,
      "grad_norm": 5.044381436362942,
      "learning_rate": 1.9999751865721574e-07,
      "loss": 0.9535,
      "step": 326
    },
    {
      "epoch": 0.03232582853471072,
      "grad_norm": 4.025716575067825,
      "learning_rate": 1.9999728795490052e-07,
      "loss": 0.8322,
      "step": 327
    },
    {
      "epoch": 0.03242468427946519,
      "grad_norm": 7.2894766020034805,
      "learning_rate": 1.9999704699933847e-07,
      "loss": 0.8434,
      "step": 328
    },
    {
      "epoch": 0.03252354002421966,
      "grad_norm": 5.177011522310207,
      "learning_rate": 1.999967957905543e-07,
      "loss": 0.8951,
      "step": 329
    },
    {
      "epoch": 0.032622395768974126,
      "grad_norm": 7.085312951981446,
      "learning_rate": 1.9999653432857379e-07,
      "loss": 0.8164,
      "step": 330
    },
    {
      "epoch": 0.032721251513728594,
      "grad_norm": 4.7538627531621165,
      "learning_rate": 1.9999626261342373e-07,
      "loss": 0.8931,
      "step": 331
    },
    {
      "epoch": 0.032820107258483056,
      "grad_norm": 5.408924907598474,
      "learning_rate": 1.99995980645132e-07,
      "loss": 0.841,
      "step": 332
    },
    {
      "epoch": 0.032918963003237524,
      "grad_norm": 14.449733296281659,
      "learning_rate": 1.9999568842372748e-07,
      "loss": 0.9255,
      "step": 333
    },
    {
      "epoch": 0.03301781874799199,
      "grad_norm": 4.422665063018142,
      "learning_rate": 1.9999538594924016e-07,
      "loss": 0.792,
      "step": 334
    },
    {
      "epoch": 0.03311667449274646,
      "grad_norm": 4.613792173053604,
      "learning_rate": 1.9999507322170104e-07,
      "loss": 0.864,
      "step": 335
    },
    {
      "epoch": 0.03321553023750093,
      "grad_norm": 6.887211411703311,
      "learning_rate": 1.9999475024114219e-07,
      "loss": 0.8341,
      "step": 336
    },
    {
      "epoch": 0.0333143859822554,
      "grad_norm": 7.7204027978785845,
      "learning_rate": 1.9999441700759676e-07,
      "loss": 0.9416,
      "step": 337
    },
    {
      "epoch": 0.03341324172700986,
      "grad_norm": 5.6965610314041815,
      "learning_rate": 1.9999407352109886e-07,
      "loss": 0.8506,
      "step": 338
    },
    {
      "epoch": 0.03351209747176433,
      "grad_norm": 11.934550181444044,
      "learning_rate": 1.9999371978168374e-07,
      "loss": 0.8064,
      "step": 339
    },
    {
      "epoch": 0.033610953216518795,
      "grad_norm": 5.789253230698797,
      "learning_rate": 1.9999335578938767e-07,
      "loss": 0.934,
      "step": 340
    },
    {
      "epoch": 0.03370980896127326,
      "grad_norm": 6.230886370004745,
      "learning_rate": 1.9999298154424796e-07,
      "loss": 0.8545,
      "step": 341
    },
    {
      "epoch": 0.03380866470602773,
      "grad_norm": 6.399403280360268,
      "learning_rate": 1.99992597046303e-07,
      "loss": 0.8393,
      "step": 342
    },
    {
      "epoch": 0.03390752045078219,
      "grad_norm": 3.611734081030316,
      "learning_rate": 1.999922022955922e-07,
      "loss": 0.8417,
      "step": 343
    },
    {
      "epoch": 0.03400637619553666,
      "grad_norm": 6.901492076279038,
      "learning_rate": 1.999917972921561e-07,
      "loss": 0.7153,
      "step": 344
    },
    {
      "epoch": 0.03410523194029113,
      "grad_norm": 5.667731123821929,
      "learning_rate": 1.999913820360361e-07,
      "loss": 0.771,
      "step": 345
    },
    {
      "epoch": 0.0342040876850456,
      "grad_norm": 4.086026152003282,
      "learning_rate": 1.9999095652727487e-07,
      "loss": 0.8265,
      "step": 346
    },
    {
      "epoch": 0.034302943429800066,
      "grad_norm": 5.691119881385631,
      "learning_rate": 1.9999052076591606e-07,
      "loss": 0.8483,
      "step": 347
    },
    {
      "epoch": 0.034401799174554534,
      "grad_norm": 12.438329093933309,
      "learning_rate": 1.9999007475200425e-07,
      "loss": 0.8122,
      "step": 348
    },
    {
      "epoch": 0.034500654919308996,
      "grad_norm": 4.057495729207768,
      "learning_rate": 1.9998961848558524e-07,
      "loss": 0.8218,
      "step": 349
    },
    {
      "epoch": 0.034599510664063464,
      "grad_norm": 4.4885159107755355,
      "learning_rate": 1.9998915196670584e-07,
      "loss": 0.8406,
      "step": 350
    },
    {
      "epoch": 0.03469836640881793,
      "grad_norm": 15.238497547228697,
      "learning_rate": 1.999886751954138e-07,
      "loss": 0.858,
      "step": 351
    },
    {
      "epoch": 0.0347972221535724,
      "grad_norm": 8.765867990730682,
      "learning_rate": 1.9998818817175808e-07,
      "loss": 0.7987,
      "step": 352
    },
    {
      "epoch": 0.03489607789832687,
      "grad_norm": 20.71816332235506,
      "learning_rate": 1.9998769089578863e-07,
      "loss": 0.7851,
      "step": 353
    },
    {
      "epoch": 0.03499493364308133,
      "grad_norm": 4.539548707016545,
      "learning_rate": 1.9998718336755635e-07,
      "loss": 0.8783,
      "step": 354
    },
    {
      "epoch": 0.0350937893878358,
      "grad_norm": 5.160358072525824,
      "learning_rate": 1.9998666558711338e-07,
      "loss": 0.8797,
      "step": 355
    },
    {
      "epoch": 0.03519264513259027,
      "grad_norm": 8.032174110957932,
      "learning_rate": 1.999861375545127e-07,
      "loss": 0.8274,
      "step": 356
    },
    {
      "epoch": 0.035291500877344735,
      "grad_norm": 5.719100658732963,
      "learning_rate": 1.9998559926980856e-07,
      "loss": 0.8271,
      "step": 357
    },
    {
      "epoch": 0.0353903566220992,
      "grad_norm": 3.5258946125174924,
      "learning_rate": 1.9998505073305606e-07,
      "loss": 0.7885,
      "step": 358
    },
    {
      "epoch": 0.03548921236685367,
      "grad_norm": 5.324561683955382,
      "learning_rate": 1.9998449194431154e-07,
      "loss": 0.85,
      "step": 359
    },
    {
      "epoch": 0.03558806811160813,
      "grad_norm": 5.679128420602625,
      "learning_rate": 1.9998392290363223e-07,
      "loss": 0.7947,
      "step": 360
    },
    {
      "epoch": 0.0356869238563626,
      "grad_norm": 4.67555845171299,
      "learning_rate": 1.9998334361107646e-07,
      "loss": 0.8389,
      "step": 361
    },
    {
      "epoch": 0.03578577960111707,
      "grad_norm": 4.068869759218887,
      "learning_rate": 1.9998275406670372e-07,
      "loss": 0.7432,
      "step": 362
    },
    {
      "epoch": 0.03588463534587154,
      "grad_norm": 3.9129955108062515,
      "learning_rate": 1.9998215427057435e-07,
      "loss": 0.7722,
      "step": 363
    },
    {
      "epoch": 0.035983491090626006,
      "grad_norm": 14.655470022543033,
      "learning_rate": 1.9998154422274993e-07,
      "loss": 0.8562,
      "step": 364
    },
    {
      "epoch": 0.036082346835380474,
      "grad_norm": 7.1942008439858265,
      "learning_rate": 1.9998092392329295e-07,
      "loss": 0.7988,
      "step": 365
    },
    {
      "epoch": 0.036181202580134936,
      "grad_norm": 5.111128699112608,
      "learning_rate": 1.9998029337226706e-07,
      "loss": 0.8902,
      "step": 366
    },
    {
      "epoch": 0.036280058324889404,
      "grad_norm": 5.630858305543143,
      "learning_rate": 1.999796525697369e-07,
      "loss": 0.8966,
      "step": 367
    },
    {
      "epoch": 0.03637891406964387,
      "grad_norm": 6.040433997532994,
      "learning_rate": 1.9997900151576818e-07,
      "loss": 0.7762,
      "step": 368
    },
    {
      "epoch": 0.03647776981439834,
      "grad_norm": 4.183838129731422,
      "learning_rate": 1.9997834021042764e-07,
      "loss": 0.8513,
      "step": 369
    },
    {
      "epoch": 0.03657662555915281,
      "grad_norm": 9.358697870269305,
      "learning_rate": 1.9997766865378313e-07,
      "loss": 0.8595,
      "step": 370
    },
    {
      "epoch": 0.03667548130390727,
      "grad_norm": 18.3748233760883,
      "learning_rate": 1.999769868459034e-07,
      "loss": 0.7567,
      "step": 371
    },
    {
      "epoch": 0.03677433704866174,
      "grad_norm": 6.182458126295749,
      "learning_rate": 1.9997629478685848e-07,
      "loss": 0.8599,
      "step": 372
    },
    {
      "epoch": 0.03687319279341621,
      "grad_norm": 5.99717192059847,
      "learning_rate": 1.999755924767193e-07,
      "loss": 0.7251,
      "step": 373
    },
    {
      "epoch": 0.036972048538170675,
      "grad_norm": 4.320412630204419,
      "learning_rate": 1.9997487991555783e-07,
      "loss": 0.868,
      "step": 374
    },
    {
      "epoch": 0.03707090428292514,
      "grad_norm": 5.80599906162185,
      "learning_rate": 1.9997415710344716e-07,
      "loss": 0.8735,
      "step": 375
    },
    {
      "epoch": 0.03716976002767961,
      "grad_norm": 9.266422696089464,
      "learning_rate": 1.999734240404614e-07,
      "loss": 0.8892,
      "step": 376
    },
    {
      "epoch": 0.03726861577243407,
      "grad_norm": 4.813529822223293,
      "learning_rate": 1.9997268072667572e-07,
      "loss": 0.7663,
      "step": 377
    },
    {
      "epoch": 0.03736747151718854,
      "grad_norm": 15.687710450264023,
      "learning_rate": 1.9997192716216635e-07,
      "loss": 0.8878,
      "step": 378
    },
    {
      "epoch": 0.03746632726194301,
      "grad_norm": 5.664425049239898,
      "learning_rate": 1.9997116334701052e-07,
      "loss": 0.8761,
      "step": 379
    },
    {
      "epoch": 0.03756518300669748,
      "grad_norm": 4.346012986911804,
      "learning_rate": 1.9997038928128656e-07,
      "loss": 0.8236,
      "step": 380
    },
    {
      "epoch": 0.037664038751451946,
      "grad_norm": 4.095818649520275,
      "learning_rate": 1.9996960496507387e-07,
      "loss": 0.6989,
      "step": 381
    },
    {
      "epoch": 0.03776289449620641,
      "grad_norm": 6.726619669774511,
      "learning_rate": 1.999688103984528e-07,
      "loss": 0.8213,
      "step": 382
    },
    {
      "epoch": 0.037861750240960876,
      "grad_norm": 4.923925961446826,
      "learning_rate": 1.999680055815049e-07,
      "loss": 0.8308,
      "step": 383
    },
    {
      "epoch": 0.037960605985715344,
      "grad_norm": 4.985612472471738,
      "learning_rate": 1.9996719051431267e-07,
      "loss": 0.8373,
      "step": 384
    },
    {
      "epoch": 0.03805946173046981,
      "grad_norm": 8.082388320822929,
      "learning_rate": 1.9996636519695965e-07,
      "loss": 0.7695,
      "step": 385
    },
    {
      "epoch": 0.03815831747522428,
      "grad_norm": 11.255138644968893,
      "learning_rate": 1.9996552962953052e-07,
      "loss": 0.8885,
      "step": 386
    },
    {
      "epoch": 0.03825717321997875,
      "grad_norm": 7.45857009927108,
      "learning_rate": 1.999646838121109e-07,
      "loss": 0.8612,
      "step": 387
    },
    {
      "epoch": 0.03835602896473321,
      "grad_norm": 4.660663038832222,
      "learning_rate": 1.9996382774478753e-07,
      "loss": 0.839,
      "step": 388
    },
    {
      "epoch": 0.03845488470948768,
      "grad_norm": 4.343319707601303,
      "learning_rate": 1.9996296142764821e-07,
      "loss": 0.7547,
      "step": 389
    },
    {
      "epoch": 0.03855374045424215,
      "grad_norm": 4.305769211448222,
      "learning_rate": 1.9996208486078174e-07,
      "loss": 0.8076,
      "step": 390
    },
    {
      "epoch": 0.038652596198996615,
      "grad_norm": 10.864278383580485,
      "learning_rate": 1.9996119804427802e-07,
      "loss": 0.7821,
      "step": 391
    },
    {
      "epoch": 0.038751451943751083,
      "grad_norm": 6.575855235990588,
      "learning_rate": 1.9996030097822798e-07,
      "loss": 0.8118,
      "step": 392
    },
    {
      "epoch": 0.038850307688505545,
      "grad_norm": 4.926962019448554,
      "learning_rate": 1.9995939366272357e-07,
      "loss": 0.7367,
      "step": 393
    },
    {
      "epoch": 0.03894916343326001,
      "grad_norm": 8.776714876815573,
      "learning_rate": 1.9995847609785788e-07,
      "loss": 0.7832,
      "step": 394
    },
    {
      "epoch": 0.03904801917801448,
      "grad_norm": 4.480226018772917,
      "learning_rate": 1.9995754828372493e-07,
      "loss": 0.7809,
      "step": 395
    },
    {
      "epoch": 0.03914687492276895,
      "grad_norm": 4.157006260320855,
      "learning_rate": 1.9995661022041987e-07,
      "loss": 0.779,
      "step": 396
    },
    {
      "epoch": 0.03924573066752342,
      "grad_norm": 3.378999916148288,
      "learning_rate": 1.9995566190803892e-07,
      "loss": 0.8089,
      "step": 397
    },
    {
      "epoch": 0.039344586412277886,
      "grad_norm": 4.710117957876718,
      "learning_rate": 1.9995470334667928e-07,
      "loss": 0.7844,
      "step": 398
    },
    {
      "epoch": 0.03944344215703235,
      "grad_norm": 3.26494476411337,
      "learning_rate": 1.9995373453643923e-07,
      "loss": 0.674,
      "step": 399
    },
    {
      "epoch": 0.039542297901786816,
      "grad_norm": 11.412128052415037,
      "learning_rate": 1.9995275547741816e-07,
      "loss": 0.763,
      "step": 400
    },
    {
      "epoch": 0.039641153646541284,
      "grad_norm": 4.450091174077025,
      "learning_rate": 1.999517661697164e-07,
      "loss": 0.8026,
      "step": 401
    },
    {
      "epoch": 0.03974000939129575,
      "grad_norm": 5.224552408444077,
      "learning_rate": 1.999507666134354e-07,
      "loss": 0.8021,
      "step": 402
    },
    {
      "epoch": 0.03983886513605022,
      "grad_norm": 10.230362602292555,
      "learning_rate": 1.9994975680867768e-07,
      "loss": 0.8656,
      "step": 403
    },
    {
      "epoch": 0.03993772088080469,
      "grad_norm": 15.460096400102703,
      "learning_rate": 1.999487367555468e-07,
      "loss": 0.716,
      "step": 404
    },
    {
      "epoch": 0.04003657662555915,
      "grad_norm": 3.5136394629567844,
      "learning_rate": 1.9994770645414728e-07,
      "loss": 0.8477,
      "step": 405
    },
    {
      "epoch": 0.04013543237031362,
      "grad_norm": 4.192756040070386,
      "learning_rate": 1.9994666590458477e-07,
      "loss": 0.721,
      "step": 406
    },
    {
      "epoch": 0.04023428811506809,
      "grad_norm": 4.2283514119369014,
      "learning_rate": 1.99945615106966e-07,
      "loss": 0.7956,
      "step": 407
    },
    {
      "epoch": 0.040333143859822555,
      "grad_norm": 6.312766155035111,
      "learning_rate": 1.9994455406139875e-07,
      "loss": 0.7579,
      "step": 408
    },
    {
      "epoch": 0.040431999604577024,
      "grad_norm": 5.525523972845212,
      "learning_rate": 1.9994348276799168e-07,
      "loss": 0.8388,
      "step": 409
    },
    {
      "epoch": 0.040530855349331485,
      "grad_norm": 5.507709095515701,
      "learning_rate": 1.999424012268548e-07,
      "loss": 0.7933,
      "step": 410
    },
    {
      "epoch": 0.04062971109408595,
      "grad_norm": 3.7403961158486907,
      "learning_rate": 1.9994130943809887e-07,
      "loss": 0.8482,
      "step": 411
    },
    {
      "epoch": 0.04072856683884042,
      "grad_norm": 5.611186069925248,
      "learning_rate": 1.9994020740183594e-07,
      "loss": 0.7101,
      "step": 412
    },
    {
      "epoch": 0.04082742258359489,
      "grad_norm": 9.28560206635102,
      "learning_rate": 1.9993909511817896e-07,
      "loss": 0.7582,
      "step": 413
    },
    {
      "epoch": 0.04092627832834936,
      "grad_norm": 17.711117570151426,
      "learning_rate": 1.9993797258724196e-07,
      "loss": 0.8654,
      "step": 414
    },
    {
      "epoch": 0.041025134073103826,
      "grad_norm": 5.5302807433352745,
      "learning_rate": 1.9993683980914008e-07,
      "loss": 0.7833,
      "step": 415
    },
    {
      "epoch": 0.04112398981785829,
      "grad_norm": 14.932613924003414,
      "learning_rate": 1.9993569678398943e-07,
      "loss": 0.7905,
      "step": 416
    },
    {
      "epoch": 0.041222845562612756,
      "grad_norm": 18.364873674608894,
      "learning_rate": 1.9993454351190723e-07,
      "loss": 0.8167,
      "step": 417
    },
    {
      "epoch": 0.041321701307367224,
      "grad_norm": 5.599083189015394,
      "learning_rate": 1.9993337999301176e-07,
      "loss": 0.9052,
      "step": 418
    },
    {
      "epoch": 0.04142055705212169,
      "grad_norm": 4.855817914066184,
      "learning_rate": 1.9993220622742226e-07,
      "loss": 0.781,
      "step": 419
    },
    {
      "epoch": 0.04151941279687616,
      "grad_norm": 4.396621048217355,
      "learning_rate": 1.9993102221525912e-07,
      "loss": 0.8649,
      "step": 420
    },
    {
      "epoch": 0.04161826854163062,
      "grad_norm": 4.303637387538735,
      "learning_rate": 1.9992982795664375e-07,
      "loss": 0.7832,
      "step": 421
    },
    {
      "epoch": 0.04171712428638509,
      "grad_norm": 13.425544054353267,
      "learning_rate": 1.9992862345169858e-07,
      "loss": 0.7953,
      "step": 422
    },
    {
      "epoch": 0.04181598003113956,
      "grad_norm": 9.003501221875005,
      "learning_rate": 1.9992740870054713e-07,
      "loss": 0.7927,
      "step": 423
    },
    {
      "epoch": 0.04191483577589403,
      "grad_norm": 15.221501817381794,
      "learning_rate": 1.9992618370331397e-07,
      "loss": 0.7623,
      "step": 424
    },
    {
      "epoch": 0.042013691520648495,
      "grad_norm": 6.174005704483763,
      "learning_rate": 1.9992494846012466e-07,
      "loss": 0.7659,
      "step": 425
    },
    {
      "epoch": 0.042112547265402964,
      "grad_norm": 32.25792275726504,
      "learning_rate": 1.999237029711059e-07,
      "loss": 0.7666,
      "step": 426
    },
    {
      "epoch": 0.042211403010157425,
      "grad_norm": 7.070717533882822,
      "learning_rate": 1.9992244723638536e-07,
      "loss": 0.8148,
      "step": 427
    },
    {
      "epoch": 0.04231025875491189,
      "grad_norm": 4.003858667861757,
      "learning_rate": 1.9992118125609184e-07,
      "loss": 0.8473,
      "step": 428
    },
    {
      "epoch": 0.04240911449966636,
      "grad_norm": 8.259202006210145,
      "learning_rate": 1.999199050303551e-07,
      "loss": 0.7927,
      "step": 429
    },
    {
      "epoch": 0.04250797024442083,
      "grad_norm": 3.8126595523371356,
      "learning_rate": 1.9991861855930604e-07,
      "loss": 0.8087,
      "step": 430
    },
    {
      "epoch": 0.0426068259891753,
      "grad_norm": 3.766792519005445,
      "learning_rate": 1.9991732184307654e-07,
      "loss": 0.833,
      "step": 431
    },
    {
      "epoch": 0.04270568173392976,
      "grad_norm": 9.38053135747968,
      "learning_rate": 1.9991601488179957e-07,
      "loss": 0.8239,
      "step": 432
    },
    {
      "epoch": 0.04280453747868423,
      "grad_norm": 3.357235436597659,
      "learning_rate": 1.9991469767560914e-07,
      "loss": 0.8866,
      "step": 433
    },
    {
      "epoch": 0.042903393223438696,
      "grad_norm": 5.944776713804741,
      "learning_rate": 1.9991337022464033e-07,
      "loss": 0.7594,
      "step": 434
    },
    {
      "epoch": 0.043002248968193164,
      "grad_norm": 3.9399995967574872,
      "learning_rate": 1.999120325290292e-07,
      "loss": 0.8278,
      "step": 435
    },
    {
      "epoch": 0.04310110471294763,
      "grad_norm": 3.7662243910024853,
      "learning_rate": 1.999106845889129e-07,
      "loss": 0.7511,
      "step": 436
    },
    {
      "epoch": 0.0431999604577021,
      "grad_norm": 5.9009578029556495,
      "learning_rate": 1.9990932640442976e-07,
      "loss": 0.7971,
      "step": 437
    },
    {
      "epoch": 0.04329881620245656,
      "grad_norm": 7.317625052744907,
      "learning_rate": 1.9990795797571893e-07,
      "loss": 0.7689,
      "step": 438
    },
    {
      "epoch": 0.04339767194721103,
      "grad_norm": 8.387711716615467,
      "learning_rate": 1.9990657930292077e-07,
      "loss": 0.718,
      "step": 439
    },
    {
      "epoch": 0.0434965276919655,
      "grad_norm": 4.055861802351913,
      "learning_rate": 1.999051903861766e-07,
      "loss": 0.8252,
      "step": 440
    },
    {
      "epoch": 0.04359538343671997,
      "grad_norm": 3.4134345369414887,
      "learning_rate": 1.9990379122562888e-07,
      "loss": 0.8018,
      "step": 441
    },
    {
      "epoch": 0.043694239181474435,
      "grad_norm": 5.1746524425174645,
      "learning_rate": 1.9990238182142107e-07,
      "loss": 0.7318,
      "step": 442
    },
    {
      "epoch": 0.043793094926228904,
      "grad_norm": 5.063514977037955,
      "learning_rate": 1.9990096217369766e-07,
      "loss": 0.9258,
      "step": 443
    },
    {
      "epoch": 0.043891950670983365,
      "grad_norm": 4.542862596562116,
      "learning_rate": 1.9989953228260418e-07,
      "loss": 0.7844,
      "step": 444
    },
    {
      "epoch": 0.04399080641573783,
      "grad_norm": 8.966816768548158,
      "learning_rate": 1.9989809214828736e-07,
      "loss": 0.7806,
      "step": 445
    },
    {
      "epoch": 0.0440896621604923,
      "grad_norm": 3.906974327784337,
      "learning_rate": 1.9989664177089472e-07,
      "loss": 0.8423,
      "step": 446
    },
    {
      "epoch": 0.04418851790524677,
      "grad_norm": 7.853685824907073,
      "learning_rate": 1.998951811505751e-07,
      "loss": 0.8135,
      "step": 447
    },
    {
      "epoch": 0.04428737365000124,
      "grad_norm": 5.384518855964276,
      "learning_rate": 1.9989371028747817e-07,
      "loss": 0.7961,
      "step": 448
    },
    {
      "epoch": 0.0443862293947557,
      "grad_norm": 6.336490624882754,
      "learning_rate": 1.9989222918175478e-07,
      "loss": 0.7797,
      "step": 449
    },
    {
      "epoch": 0.04448508513951017,
      "grad_norm": 5.738542241917747,
      "learning_rate": 1.9989073783355684e-07,
      "loss": 0.8043,
      "step": 450
    },
    {
      "epoch": 0.044583940884264636,
      "grad_norm": 12.707916032636668,
      "learning_rate": 1.9988923624303718e-07,
      "loss": 0.8174,
      "step": 451
    },
    {
      "epoch": 0.044682796629019104,
      "grad_norm": 5.321506103710058,
      "learning_rate": 1.9988772441034988e-07,
      "loss": 0.9339,
      "step": 452
    },
    {
      "epoch": 0.04478165237377357,
      "grad_norm": 5.861776095011742,
      "learning_rate": 1.9988620233564984e-07,
      "loss": 0.8142,
      "step": 453
    },
    {
      "epoch": 0.04488050811852804,
      "grad_norm": 4.721179738793206,
      "learning_rate": 1.998846700190932e-07,
      "loss": 0.745,
      "step": 454
    },
    {
      "epoch": 0.0449793638632825,
      "grad_norm": 4.2427011803511085,
      "learning_rate": 1.99883127460837e-07,
      "loss": 0.788,
      "step": 455
    },
    {
      "epoch": 0.04507821960803697,
      "grad_norm": 10.458101930384597,
      "learning_rate": 1.9988157466103953e-07,
      "loss": 0.8246,
      "step": 456
    },
    {
      "epoch": 0.04517707535279144,
      "grad_norm": 3.748182656422571,
      "learning_rate": 1.998800116198599e-07,
      "loss": 0.8637,
      "step": 457
    },
    {
      "epoch": 0.04527593109754591,
      "grad_norm": 6.658952763185419,
      "learning_rate": 1.9987843833745841e-07,
      "loss": 0.7919,
      "step": 458
    },
    {
      "epoch": 0.045374786842300376,
      "grad_norm": 6.900471611676066,
      "learning_rate": 1.9987685481399642e-07,
      "loss": 0.9194,
      "step": 459
    },
    {
      "epoch": 0.04547364258705484,
      "grad_norm": 4.901462337331058,
      "learning_rate": 1.9987526104963619e-07,
      "loss": 0.7935,
      "step": 460
    },
    {
      "epoch": 0.045572498331809305,
      "grad_norm": 15.655914151588954,
      "learning_rate": 1.9987365704454123e-07,
      "loss": 0.7235,
      "step": 461
    },
    {
      "epoch": 0.045671354076563773,
      "grad_norm": 4.131477921193643,
      "learning_rate": 1.99872042798876e-07,
      "loss": 0.6886,
      "step": 462
    },
    {
      "epoch": 0.04577020982131824,
      "grad_norm": 4.027212923721047,
      "learning_rate": 1.9987041831280597e-07,
      "loss": 0.7966,
      "step": 463
    },
    {
      "epoch": 0.04586906556607271,
      "grad_norm": 4.711726780055351,
      "learning_rate": 1.9986878358649774e-07,
      "loss": 0.8675,
      "step": 464
    },
    {
      "epoch": 0.04596792131082718,
      "grad_norm": 4.730429086413092,
      "learning_rate": 1.9986713862011896e-07,
      "loss": 0.776,
      "step": 465
    },
    {
      "epoch": 0.04606677705558164,
      "grad_norm": 4.0128621231391355,
      "learning_rate": 1.998654834138382e-07,
      "loss": 0.777,
      "step": 466
    },
    {
      "epoch": 0.04616563280033611,
      "grad_norm": 9.381545453598022,
      "learning_rate": 1.9986381796782527e-07,
      "loss": 0.7446,
      "step": 467
    },
    {
      "epoch": 0.046264488545090576,
      "grad_norm": 4.597355506639818,
      "learning_rate": 1.998621422822509e-07,
      "loss": 0.809,
      "step": 468
    },
    {
      "epoch": 0.046363344289845045,
      "grad_norm": 3.754386598963816,
      "learning_rate": 1.998604563572869e-07,
      "loss": 0.7365,
      "step": 469
    },
    {
      "epoch": 0.04646220003459951,
      "grad_norm": 5.058433019133744,
      "learning_rate": 1.9985876019310614e-07,
      "loss": 0.6917,
      "step": 470
    },
    {
      "epoch": 0.04656105577935398,
      "grad_norm": 3.919436101021462,
      "learning_rate": 1.9985705378988257e-07,
      "loss": 0.7461,
      "step": 471
    },
    {
      "epoch": 0.04665991152410844,
      "grad_norm": 4.434462610030953,
      "learning_rate": 1.998553371477911e-07,
      "loss": 0.7533,
      "step": 472
    },
    {
      "epoch": 0.04675876726886291,
      "grad_norm": 6.6046225316529075,
      "learning_rate": 1.9985361026700779e-07,
      "loss": 0.83,
      "step": 473
    },
    {
      "epoch": 0.04685762301361738,
      "grad_norm": 6.388539912069082,
      "learning_rate": 1.9985187314770968e-07,
      "loss": 0.8052,
      "step": 474
    },
    {
      "epoch": 0.04695647875837185,
      "grad_norm": 11.41393496393874,
      "learning_rate": 1.9985012579007491e-07,
      "loss": 0.7331,
      "step": 475
    },
    {
      "epoch": 0.047055334503126316,
      "grad_norm": 8.626758156377393,
      "learning_rate": 1.9984836819428264e-07,
      "loss": 0.6719,
      "step": 476
    },
    {
      "epoch": 0.04715419024788078,
      "grad_norm": 4.065204659167926,
      "learning_rate": 1.9984660036051306e-07,
      "loss": 0.7464,
      "step": 477
    },
    {
      "epoch": 0.047253045992635245,
      "grad_norm": 5.510067861728492,
      "learning_rate": 1.9984482228894747e-07,
      "loss": 0.8867,
      "step": 478
    },
    {
      "epoch": 0.047351901737389714,
      "grad_norm": 10.175246884586214,
      "learning_rate": 1.9984303397976816e-07,
      "loss": 0.8034,
      "step": 479
    },
    {
      "epoch": 0.04745075748214418,
      "grad_norm": 4.269998288611957,
      "learning_rate": 1.998412354331585e-07,
      "loss": 0.7339,
      "step": 480
    },
    {
      "epoch": 0.04754961322689865,
      "grad_norm": 5.267560544804125,
      "learning_rate": 1.998394266493029e-07,
      "loss": 0.7914,
      "step": 481
    },
    {
      "epoch": 0.04764846897165312,
      "grad_norm": 8.594420200713202,
      "learning_rate": 1.9983760762838687e-07,
      "loss": 0.7174,
      "step": 482
    },
    {
      "epoch": 0.04774732471640758,
      "grad_norm": 3.696091337560013,
      "learning_rate": 1.9983577837059685e-07,
      "loss": 0.7635,
      "step": 483
    },
    {
      "epoch": 0.04784618046116205,
      "grad_norm": 3.703476016689107,
      "learning_rate": 1.9983393887612046e-07,
      "loss": 0.817,
      "step": 484
    },
    {
      "epoch": 0.047945036205916516,
      "grad_norm": 7.503945564129945,
      "learning_rate": 1.9983208914514629e-07,
      "loss": 0.7699,
      "step": 485
    },
    {
      "epoch": 0.048043891950670985,
      "grad_norm": 3.698256953092467,
      "learning_rate": 1.99830229177864e-07,
      "loss": 0.7252,
      "step": 486
    },
    {
      "epoch": 0.04814274769542545,
      "grad_norm": 5.640795718727372,
      "learning_rate": 1.9982835897446427e-07,
      "loss": 0.7718,
      "step": 487
    },
    {
      "epoch": 0.048241603440179914,
      "grad_norm": 4.966435240807712,
      "learning_rate": 1.9982647853513894e-07,
      "loss": 0.8087,
      "step": 488
    },
    {
      "epoch": 0.04834045918493438,
      "grad_norm": 3.164752233697765,
      "learning_rate": 1.998245878600808e-07,
      "loss": 0.8634,
      "step": 489
    },
    {
      "epoch": 0.04843931492968885,
      "grad_norm": 5.0940183942065955,
      "learning_rate": 1.9982268694948365e-07,
      "loss": 0.7244,
      "step": 490
    },
    {
      "epoch": 0.04853817067444332,
      "grad_norm": 4.876365362960067,
      "learning_rate": 1.9982077580354247e-07,
      "loss": 0.7371,
      "step": 491
    },
    {
      "epoch": 0.04863702641919779,
      "grad_norm": 22.80252668595278,
      "learning_rate": 1.9981885442245315e-07,
      "loss": 0.7156,
      "step": 492
    },
    {
      "epoch": 0.048735882163952256,
      "grad_norm": 6.76950883906046,
      "learning_rate": 1.9981692280641277e-07,
      "loss": 0.8417,
      "step": 493
    },
    {
      "epoch": 0.04883473790870672,
      "grad_norm": 4.159066054549585,
      "learning_rate": 1.9981498095561937e-07,
      "loss": 0.849,
      "step": 494
    },
    {
      "epoch": 0.048933593653461185,
      "grad_norm": 3.9077094431015045,
      "learning_rate": 1.9981302887027205e-07,
      "loss": 0.7607,
      "step": 495
    },
    {
      "epoch": 0.049032449398215654,
      "grad_norm": 3.7492417530136146,
      "learning_rate": 1.9981106655057096e-07,
      "loss": 0.7815,
      "step": 496
    },
    {
      "epoch": 0.04913130514297012,
      "grad_norm": 7.488461103359557,
      "learning_rate": 1.9980909399671728e-07,
      "loss": 0.7452,
      "step": 497
    },
    {
      "epoch": 0.04923016088772459,
      "grad_norm": 9.058752920828953,
      "learning_rate": 1.9980711120891333e-07,
      "loss": 0.7781,
      "step": 498
    },
    {
      "epoch": 0.04932901663247905,
      "grad_norm": 6.51558292977401,
      "learning_rate": 1.9980511818736238e-07,
      "loss": 0.7415,
      "step": 499
    },
    {
      "epoch": 0.04942787237723352,
      "grad_norm": 3.9958327707302996,
      "learning_rate": 1.9980311493226878e-07,
      "loss": 0.9181,
      "step": 500
    },
    {
      "epoch": 0.04952672812198799,
      "grad_norm": 4.878511089224303,
      "learning_rate": 1.9980110144383796e-07,
      "loss": 0.8493,
      "step": 501
    },
    {
      "epoch": 0.049625583866742456,
      "grad_norm": 12.433887239823019,
      "learning_rate": 1.9979907772227634e-07,
      "loss": 0.7992,
      "step": 502
    },
    {
      "epoch": 0.049724439611496925,
      "grad_norm": 5.584226186237488,
      "learning_rate": 1.9979704376779142e-07,
      "loss": 0.7808,
      "step": 503
    },
    {
      "epoch": 0.04982329535625139,
      "grad_norm": 12.399411454924568,
      "learning_rate": 1.9979499958059182e-07,
      "loss": 0.8029,
      "step": 504
    },
    {
      "epoch": 0.049922151101005854,
      "grad_norm": 3.257168379805648,
      "learning_rate": 1.9979294516088706e-07,
      "loss": 0.744,
      "step": 505
    },
    {
      "epoch": 0.05002100684576032,
      "grad_norm": 4.134701304991991,
      "learning_rate": 1.9979088050888785e-07,
      "loss": 0.7814,
      "step": 506
    },
    {
      "epoch": 0.05011986259051479,
      "grad_norm": 11.931233682131015,
      "learning_rate": 1.9978880562480583e-07,
      "loss": 0.8345,
      "step": 507
    },
    {
      "epoch": 0.05021871833526926,
      "grad_norm": 8.86114964567862,
      "learning_rate": 1.9978672050885378e-07,
      "loss": 0.8853,
      "step": 508
    },
    {
      "epoch": 0.05031757408002373,
      "grad_norm": 9.954132493610551,
      "learning_rate": 1.9978462516124553e-07,
      "loss": 0.8393,
      "step": 509
    },
    {
      "epoch": 0.050416429824778196,
      "grad_norm": 5.291354925913319,
      "learning_rate": 1.9978251958219588e-07,
      "loss": 0.8716,
      "step": 510
    },
    {
      "epoch": 0.05051528556953266,
      "grad_norm": 6.747534177003873,
      "learning_rate": 1.9978040377192075e-07,
      "loss": 0.7181,
      "step": 511
    },
    {
      "epoch": 0.050614141314287125,
      "grad_norm": 8.322192074892827,
      "learning_rate": 1.9977827773063707e-07,
      "loss": 0.6897,
      "step": 512
    },
    {
      "epoch": 0.050712997059041594,
      "grad_norm": 4.2905864401934855,
      "learning_rate": 1.9977614145856284e-07,
      "loss": 0.7452,
      "step": 513
    },
    {
      "epoch": 0.05081185280379606,
      "grad_norm": 6.262494662534928,
      "learning_rate": 1.9977399495591713e-07,
      "loss": 0.8059,
      "step": 514
    },
    {
      "epoch": 0.05091070854855053,
      "grad_norm": 3.9009630828697293,
      "learning_rate": 1.9977183822291998e-07,
      "loss": 0.7179,
      "step": 515
    },
    {
      "epoch": 0.05100956429330499,
      "grad_norm": 2.9077842959893707,
      "learning_rate": 1.997696712597926e-07,
      "loss": 0.789,
      "step": 516
    },
    {
      "epoch": 0.05110842003805946,
      "grad_norm": 5.719037759261635,
      "learning_rate": 1.9976749406675709e-07,
      "loss": 0.8164,
      "step": 517
    },
    {
      "epoch": 0.05120727578281393,
      "grad_norm": 4.091389196901534,
      "learning_rate": 1.9976530664403675e-07,
      "loss": 0.7506,
      "step": 518
    },
    {
      "epoch": 0.0513061315275684,
      "grad_norm": 5.022957864443125,
      "learning_rate": 1.9976310899185584e-07,
      "loss": 0.8671,
      "step": 519
    },
    {
      "epoch": 0.051404987272322865,
      "grad_norm": 8.798367885087051,
      "learning_rate": 1.9976090111043973e-07,
      "loss": 0.7709,
      "step": 520
    },
    {
      "epoch": 0.05150384301707733,
      "grad_norm": 5.936090413495646,
      "learning_rate": 1.997586830000148e-07,
      "loss": 0.7173,
      "step": 521
    },
    {
      "epoch": 0.051602698761831794,
      "grad_norm": 6.21000128848077,
      "learning_rate": 1.9975645466080845e-07,
      "loss": 0.8219,
      "step": 522
    },
    {
      "epoch": 0.05170155450658626,
      "grad_norm": 9.538415774632162,
      "learning_rate": 1.997542160930492e-07,
      "loss": 0.8722,
      "step": 523
    },
    {
      "epoch": 0.05180041025134073,
      "grad_norm": 4.463319927114203,
      "learning_rate": 1.9975196729696656e-07,
      "loss": 0.7206,
      "step": 524
    },
    {
      "epoch": 0.0518992659960952,
      "grad_norm": 8.67454765119347,
      "learning_rate": 1.9974970827279111e-07,
      "loss": 0.8521,
      "step": 525
    },
    {
      "epoch": 0.05199812174084967,
      "grad_norm": 8.377243385090626,
      "learning_rate": 1.9974743902075453e-07,
      "loss": 0.8201,
      "step": 526
    },
    {
      "epoch": 0.05209697748560413,
      "grad_norm": 6.246276768881138,
      "learning_rate": 1.9974515954108943e-07,
      "loss": 0.6447,
      "step": 527
    },
    {
      "epoch": 0.0521958332303586,
      "grad_norm": 5.362805647620259,
      "learning_rate": 1.9974286983402955e-07,
      "loss": 0.7369,
      "step": 528
    },
    {
      "epoch": 0.052294688975113066,
      "grad_norm": 4.4514513739921115,
      "learning_rate": 1.9974056989980968e-07,
      "loss": 0.8543,
      "step": 529
    },
    {
      "epoch": 0.052393544719867534,
      "grad_norm": 4.770550658376816,
      "learning_rate": 1.9973825973866564e-07,
      "loss": 0.752,
      "step": 530
    },
    {
      "epoch": 0.052492400464622,
      "grad_norm": 8.835892173275344,
      "learning_rate": 1.9973593935083435e-07,
      "loss": 0.883,
      "step": 531
    },
    {
      "epoch": 0.05259125620937647,
      "grad_norm": 4.08375576083483,
      "learning_rate": 1.9973360873655366e-07,
      "loss": 0.7774,
      "step": 532
    },
    {
      "epoch": 0.05269011195413093,
      "grad_norm": 9.977815375993414,
      "learning_rate": 1.9973126789606254e-07,
      "loss": 0.7743,
      "step": 533
    },
    {
      "epoch": 0.0527889676988854,
      "grad_norm": 5.598171039920406,
      "learning_rate": 1.9972891682960108e-07,
      "loss": 0.8274,
      "step": 534
    },
    {
      "epoch": 0.05288782344363987,
      "grad_norm": 4.28862547125591,
      "learning_rate": 1.997265555374103e-07,
      "loss": 0.9384,
      "step": 535
    },
    {
      "epoch": 0.05298667918839434,
      "grad_norm": 5.834062887410753,
      "learning_rate": 1.997241840197323e-07,
      "loss": 0.752,
      "step": 536
    },
    {
      "epoch": 0.053085534933148805,
      "grad_norm": 3.9039154556378004,
      "learning_rate": 1.9972180227681025e-07,
      "loss": 0.7705,
      "step": 537
    },
    {
      "epoch": 0.053184390677903266,
      "grad_norm": 5.935253694093829,
      "learning_rate": 1.9971941030888843e-07,
      "loss": 0.8063,
      "step": 538
    },
    {
      "epoch": 0.053283246422657735,
      "grad_norm": 3.2227217592716646,
      "learning_rate": 1.99717008116212e-07,
      "loss": 0.8059,
      "step": 539
    },
    {
      "epoch": 0.0533821021674122,
      "grad_norm": 5.450449662583103,
      "learning_rate": 1.9971459569902733e-07,
      "loss": 0.7929,
      "step": 540
    },
    {
      "epoch": 0.05348095791216667,
      "grad_norm": 3.7246751094365202,
      "learning_rate": 1.9971217305758178e-07,
      "loss": 0.7557,
      "step": 541
    },
    {
      "epoch": 0.05357981365692114,
      "grad_norm": 6.453846399486815,
      "learning_rate": 1.997097401921237e-07,
      "loss": 0.8986,
      "step": 542
    },
    {
      "epoch": 0.05367866940167561,
      "grad_norm": 5.804207991789014,
      "learning_rate": 1.9970729710290264e-07,
      "loss": 0.7101,
      "step": 543
    },
    {
      "epoch": 0.05377752514643007,
      "grad_norm": 5.2838025283155385,
      "learning_rate": 1.9970484379016902e-07,
      "loss": 0.8211,
      "step": 544
    },
    {
      "epoch": 0.05387638089118454,
      "grad_norm": 3.9761968932275904,
      "learning_rate": 1.9970238025417442e-07,
      "loss": 0.7843,
      "step": 545
    },
    {
      "epoch": 0.053975236635939006,
      "grad_norm": 5.311875424897549,
      "learning_rate": 1.9969990649517144e-07,
      "loss": 0.7806,
      "step": 546
    },
    {
      "epoch": 0.054074092380693474,
      "grad_norm": 4.75912330094027,
      "learning_rate": 1.9969742251341373e-07,
      "loss": 0.7949,
      "step": 547
    },
    {
      "epoch": 0.05417294812544794,
      "grad_norm": 4.72231071871465,
      "learning_rate": 1.9969492830915598e-07,
      "loss": 0.7815,
      "step": 548
    },
    {
      "epoch": 0.05427180387020241,
      "grad_norm": 8.997717916815803,
      "learning_rate": 1.9969242388265392e-07,
      "loss": 0.8302,
      "step": 549
    },
    {
      "epoch": 0.05437065961495687,
      "grad_norm": 9.20612542983796,
      "learning_rate": 1.996899092341644e-07,
      "loss": 0.7585,
      "step": 550
    },
    {
      "epoch": 0.05446951535971134,
      "grad_norm": 6.463546073650438,
      "learning_rate": 1.996873843639452e-07,
      "loss": 0.8169,
      "step": 551
    },
    {
      "epoch": 0.05456837110446581,
      "grad_norm": 4.9679609855771725,
      "learning_rate": 1.996848492722552e-07,
      "loss": 0.8339,
      "step": 552
    },
    {
      "epoch": 0.05466722684922028,
      "grad_norm": 11.819676562227507,
      "learning_rate": 1.9968230395935437e-07,
      "loss": 0.7215,
      "step": 553
    },
    {
      "epoch": 0.054766082593974745,
      "grad_norm": 4.710563877393096,
      "learning_rate": 1.996797484255037e-07,
      "loss": 0.7768,
      "step": 554
    },
    {
      "epoch": 0.054864938338729206,
      "grad_norm": 3.627944342853949,
      "learning_rate": 1.996771826709652e-07,
      "loss": 0.7502,
      "step": 555
    },
    {
      "epoch": 0.054963794083483675,
      "grad_norm": 5.129998740293697,
      "learning_rate": 1.9967460669600196e-07,
      "loss": 0.8111,
      "step": 556
    },
    {
      "epoch": 0.05506264982823814,
      "grad_norm": 8.082798011288501,
      "learning_rate": 1.9967202050087812e-07,
      "loss": 0.793,
      "step": 557
    },
    {
      "epoch": 0.05516150557299261,
      "grad_norm": 5.60444285644719,
      "learning_rate": 1.9966942408585882e-07,
      "loss": 0.7875,
      "step": 558
    },
    {
      "epoch": 0.05526036131774708,
      "grad_norm": 4.67849293655465,
      "learning_rate": 1.9966681745121033e-07,
      "loss": 0.844,
      "step": 559
    },
    {
      "epoch": 0.05535921706250155,
      "grad_norm": 3.6781896715539504,
      "learning_rate": 1.9966420059719988e-07,
      "loss": 0.7517,
      "step": 560
    },
    {
      "epoch": 0.05545807280725601,
      "grad_norm": 4.06316282473506,
      "learning_rate": 1.9966157352409583e-07,
      "loss": 0.7607,
      "step": 561
    },
    {
      "epoch": 0.05555692855201048,
      "grad_norm": 4.967578424876608,
      "learning_rate": 1.996589362321675e-07,
      "loss": 0.7942,
      "step": 562
    },
    {
      "epoch": 0.055655784296764946,
      "grad_norm": 3.6258564455888327,
      "learning_rate": 1.9965628872168537e-07,
      "loss": 0.7543,
      "step": 563
    },
    {
      "epoch": 0.055754640041519414,
      "grad_norm": 3.849656379576122,
      "learning_rate": 1.9965363099292082e-07,
      "loss": 0.7587,
      "step": 564
    },
    {
      "epoch": 0.05585349578627388,
      "grad_norm": 3.6637604691149943,
      "learning_rate": 1.9965096304614642e-07,
      "loss": 0.6645,
      "step": 565
    },
    {
      "epoch": 0.055952351531028344,
      "grad_norm": 5.360051375440951,
      "learning_rate": 1.9964828488163574e-07,
      "loss": 0.7071,
      "step": 566
    },
    {
      "epoch": 0.05605120727578281,
      "grad_norm": 3.345687429320808,
      "learning_rate": 1.9964559649966334e-07,
      "loss": 0.8106,
      "step": 567
    },
    {
      "epoch": 0.05615006302053728,
      "grad_norm": 28.340580580250098,
      "learning_rate": 1.996428979005049e-07,
      "loss": 0.8606,
      "step": 568
    },
    {
      "epoch": 0.05624891876529175,
      "grad_norm": 8.118026925873485,
      "learning_rate": 1.9964018908443712e-07,
      "loss": 0.8248,
      "step": 569
    },
    {
      "epoch": 0.05634777451004622,
      "grad_norm": 8.775371762225875,
      "learning_rate": 1.9963747005173774e-07,
      "loss": 0.7204,
      "step": 570
    },
    {
      "epoch": 0.056446630254800685,
      "grad_norm": 4.654273518498881,
      "learning_rate": 1.9963474080268557e-07,
      "loss": 0.7796,
      "step": 571
    },
    {
      "epoch": 0.056545485999555146,
      "grad_norm": 4.629830796688502,
      "learning_rate": 1.9963200133756044e-07,
      "loss": 0.8423,
      "step": 572
    },
    {
      "epoch": 0.056644341744309615,
      "grad_norm": 7.188741104822034,
      "learning_rate": 1.9962925165664324e-07,
      "loss": 0.782,
      "step": 573
    },
    {
      "epoch": 0.05674319748906408,
      "grad_norm": 4.935874132525405,
      "learning_rate": 1.9962649176021593e-07,
      "loss": 0.7979,
      "step": 574
    },
    {
      "epoch": 0.05684205323381855,
      "grad_norm": 4.183425875542571,
      "learning_rate": 1.996237216485615e-07,
      "loss": 0.6732,
      "step": 575
    },
    {
      "epoch": 0.05694090897857302,
      "grad_norm": 4.163474122412592,
      "learning_rate": 1.9962094132196393e-07,
      "loss": 0.7941,
      "step": 576
    },
    {
      "epoch": 0.05703976472332748,
      "grad_norm": 3.627186731015763,
      "learning_rate": 1.9961815078070836e-07,
      "loss": 0.8357,
      "step": 577
    },
    {
      "epoch": 0.05713862046808195,
      "grad_norm": 9.303576093524484,
      "learning_rate": 1.996153500250809e-07,
      "loss": 0.79,
      "step": 578
    },
    {
      "epoch": 0.05723747621283642,
      "grad_norm": 8.328395885993935,
      "learning_rate": 1.9961253905536871e-07,
      "loss": 0.7688,
      "step": 579
    },
    {
      "epoch": 0.057336331957590886,
      "grad_norm": 5.186268338753882,
      "learning_rate": 1.9960971787186004e-07,
      "loss": 0.7931,
      "step": 580
    },
    {
      "epoch": 0.057435187702345354,
      "grad_norm": 3.8716703742571745,
      "learning_rate": 1.9960688647484413e-07,
      "loss": 0.7904,
      "step": 581
    },
    {
      "epoch": 0.05753404344709982,
      "grad_norm": 4.44536045499621,
      "learning_rate": 1.9960404486461136e-07,
      "loss": 0.718,
      "step": 582
    },
    {
      "epoch": 0.057632899191854284,
      "grad_norm": 8.810447328417162,
      "learning_rate": 1.9960119304145305e-07,
      "loss": 0.8017,
      "step": 583
    },
    {
      "epoch": 0.05773175493660875,
      "grad_norm": 6.3291666385324845,
      "learning_rate": 1.9959833100566156e-07,
      "loss": 0.7878,
      "step": 584
    },
    {
      "epoch": 0.05783061068136322,
      "grad_norm": 4.2180688597399865,
      "learning_rate": 1.9959545875753045e-07,
      "loss": 0.7297,
      "step": 585
    },
    {
      "epoch": 0.05792946642611769,
      "grad_norm": 14.23868236573787,
      "learning_rate": 1.9959257629735418e-07,
      "loss": 0.8153,
      "step": 586
    },
    {
      "epoch": 0.05802832217087216,
      "grad_norm": 6.302976070727713,
      "learning_rate": 1.9958968362542826e-07,
      "loss": 0.789,
      "step": 587
    },
    {
      "epoch": 0.058127177915626625,
      "grad_norm": 4.210520190750532,
      "learning_rate": 1.995867807420494e-07,
      "loss": 0.8542,
      "step": 588
    },
    {
      "epoch": 0.05822603366038109,
      "grad_norm": 3.1024841417657267,
      "learning_rate": 1.9958386764751515e-07,
      "loss": 0.6535,
      "step": 589
    },
    {
      "epoch": 0.058324889405135555,
      "grad_norm": 3.158443208099288,
      "learning_rate": 1.9958094434212423e-07,
      "loss": 0.7538,
      "step": 590
    },
    {
      "epoch": 0.05842374514989002,
      "grad_norm": 4.356934053596402,
      "learning_rate": 1.9957801082617637e-07,
      "loss": 0.8416,
      "step": 591
    },
    {
      "epoch": 0.05852260089464449,
      "grad_norm": 7.50093203784475,
      "learning_rate": 1.9957506709997243e-07,
      "loss": 0.8185,
      "step": 592
    },
    {
      "epoch": 0.05862145663939896,
      "grad_norm": 4.175917853831336,
      "learning_rate": 1.9957211316381418e-07,
      "loss": 0.8628,
      "step": 593
    },
    {
      "epoch": 0.05872031238415342,
      "grad_norm": 4.457940489279119,
      "learning_rate": 1.995691490180045e-07,
      "loss": 0.7124,
      "step": 594
    },
    {
      "epoch": 0.05881916812890789,
      "grad_norm": 14.728033479268223,
      "learning_rate": 1.9956617466284732e-07,
      "loss": 0.689,
      "step": 595
    },
    {
      "epoch": 0.05891802387366236,
      "grad_norm": 11.856245814360035,
      "learning_rate": 1.9956319009864768e-07,
      "loss": 0.713,
      "step": 596
    },
    {
      "epoch": 0.059016879618416826,
      "grad_norm": 5.017404840825539,
      "learning_rate": 1.9956019532571153e-07,
      "loss": 0.7474,
      "step": 597
    },
    {
      "epoch": 0.059115735363171294,
      "grad_norm": 13.030117726828792,
      "learning_rate": 1.9955719034434596e-07,
      "loss": 0.7832,
      "step": 598
    },
    {
      "epoch": 0.05921459110792576,
      "grad_norm": 6.4253291347651444,
      "learning_rate": 1.995541751548591e-07,
      "loss": 0.8043,
      "step": 599
    },
    {
      "epoch": 0.059313446852680224,
      "grad_norm": 5.10778417019394,
      "learning_rate": 1.995511497575601e-07,
      "loss": 0.7389,
      "step": 600
    },
    {
      "epoch": 0.05941230259743469,
      "grad_norm": 5.526231585134557,
      "learning_rate": 1.9954811415275921e-07,
      "loss": 0.8539,
      "step": 601
    },
    {
      "epoch": 0.05951115834218916,
      "grad_norm": 21.8768788559293,
      "learning_rate": 1.9954506834076761e-07,
      "loss": 0.8014,
      "step": 602
    },
    {
      "epoch": 0.05961001408694363,
      "grad_norm": 4.010757730135512,
      "learning_rate": 1.9954201232189766e-07,
      "loss": 0.8794,
      "step": 603
    },
    {
      "epoch": 0.0597088698316981,
      "grad_norm": 3.484538389165416,
      "learning_rate": 1.9953894609646272e-07,
      "loss": 0.7805,
      "step": 604
    },
    {
      "epoch": 0.05980772557645256,
      "grad_norm": 4.440352691583752,
      "learning_rate": 1.995358696647771e-07,
      "loss": 0.8571,
      "step": 605
    },
    {
      "epoch": 0.05990658132120703,
      "grad_norm": 7.032420462493198,
      "learning_rate": 1.9953278302715635e-07,
      "loss": 0.7065,
      "step": 606
    },
    {
      "epoch": 0.060005437065961495,
      "grad_norm": 3.812710806972093,
      "learning_rate": 1.995296861839169e-07,
      "loss": 0.7068,
      "step": 607
    },
    {
      "epoch": 0.06010429281071596,
      "grad_norm": 6.1595051747188,
      "learning_rate": 1.9952657913537632e-07,
      "loss": 0.7021,
      "step": 608
    },
    {
      "epoch": 0.06020314855547043,
      "grad_norm": 5.022461644219375,
      "learning_rate": 1.9952346188185313e-07,
      "loss": 0.7742,
      "step": 609
    },
    {
      "epoch": 0.0603020043002249,
      "grad_norm": 14.668415177089749,
      "learning_rate": 1.9952033442366703e-07,
      "loss": 0.6975,
      "step": 610
    },
    {
      "epoch": 0.06040086004497936,
      "grad_norm": 5.933759916083074,
      "learning_rate": 1.9951719676113864e-07,
      "loss": 0.7489,
      "step": 611
    },
    {
      "epoch": 0.06049971578973383,
      "grad_norm": 3.1866079560632445,
      "learning_rate": 1.995140488945897e-07,
      "loss": 0.6679,
      "step": 612
    },
    {
      "epoch": 0.0605985715344883,
      "grad_norm": 4.280786151924749,
      "learning_rate": 1.9951089082434302e-07,
      "loss": 0.9005,
      "step": 613
    },
    {
      "epoch": 0.060697427279242766,
      "grad_norm": 5.50939212040624,
      "learning_rate": 1.9950772255072233e-07,
      "loss": 0.8538,
      "step": 614
    },
    {
      "epoch": 0.060796283023997234,
      "grad_norm": 15.066566346118774,
      "learning_rate": 1.9950454407405254e-07,
      "loss": 0.8126,
      "step": 615
    },
    {
      "epoch": 0.060895138768751696,
      "grad_norm": 3.4439412127846047,
      "learning_rate": 1.9950135539465957e-07,
      "loss": 0.7629,
      "step": 616
    },
    {
      "epoch": 0.060993994513506164,
      "grad_norm": 5.825167031224391,
      "learning_rate": 1.9949815651287034e-07,
      "loss": 0.7718,
      "step": 617
    },
    {
      "epoch": 0.06109285025826063,
      "grad_norm": 3.90172213318952,
      "learning_rate": 1.994949474290129e-07,
      "loss": 0.7553,
      "step": 618
    },
    {
      "epoch": 0.0611917060030151,
      "grad_norm": 3.514512547153918,
      "learning_rate": 1.9949172814341622e-07,
      "loss": 0.7976,
      "step": 619
    },
    {
      "epoch": 0.06129056174776957,
      "grad_norm": 3.970479559516452,
      "learning_rate": 1.994884986564104e-07,
      "loss": 0.6856,
      "step": 620
    },
    {
      "epoch": 0.06138941749252404,
      "grad_norm": 70.4010028852345,
      "learning_rate": 1.9948525896832663e-07,
      "loss": 0.7019,
      "step": 621
    },
    {
      "epoch": 0.0614882732372785,
      "grad_norm": 12.414491369358108,
      "learning_rate": 1.9948200907949708e-07,
      "loss": 0.7817,
      "step": 622
    },
    {
      "epoch": 0.06158712898203297,
      "grad_norm": 5.184155447406912,
      "learning_rate": 1.9947874899025494e-07,
      "loss": 0.6618,
      "step": 623
    },
    {
      "epoch": 0.061685984726787435,
      "grad_norm": 4.6088433248204845,
      "learning_rate": 1.994754787009345e-07,
      "loss": 0.7924,
      "step": 624
    },
    {
      "epoch": 0.0617848404715419,
      "grad_norm": 3.9693478561502684,
      "learning_rate": 1.9947219821187108e-07,
      "loss": 0.7984,
      "step": 625
    },
    {
      "epoch": 0.06188369621629637,
      "grad_norm": 32.2803123742957,
      "learning_rate": 1.994689075234011e-07,
      "loss": 0.8223,
      "step": 626
    },
    {
      "epoch": 0.06198255196105084,
      "grad_norm": 3.1820784083814893,
      "learning_rate": 1.9946560663586188e-07,
      "loss": 0.7401,
      "step": 627
    },
    {
      "epoch": 0.0620814077058053,
      "grad_norm": 3.5986410593181137,
      "learning_rate": 1.9946229554959193e-07,
      "loss": 0.7495,
      "step": 628
    },
    {
      "epoch": 0.06218026345055977,
      "grad_norm": 4.687726427257642,
      "learning_rate": 1.9945897426493075e-07,
      "loss": 0.8244,
      "step": 629
    },
    {
      "epoch": 0.06227911919531424,
      "grad_norm": 3.0907456771802253,
      "learning_rate": 1.9945564278221887e-07,
      "loss": 0.7553,
      "step": 630
    },
    {
      "epoch": 0.062377974940068706,
      "grad_norm": 3.7255512052420983,
      "learning_rate": 1.994523011017979e-07,
      "loss": 0.7302,
      "step": 631
    },
    {
      "epoch": 0.062476830684823174,
      "grad_norm": 5.272383098590543,
      "learning_rate": 1.9944894922401046e-07,
      "loss": 0.7688,
      "step": 632
    },
    {
      "epoch": 0.06257568642957764,
      "grad_norm": 5.75235275058399,
      "learning_rate": 1.994455871492003e-07,
      "loss": 0.7722,
      "step": 633
    },
    {
      "epoch": 0.0626745421743321,
      "grad_norm": 4.55774155520791,
      "learning_rate": 1.9944221487771209e-07,
      "loss": 0.7168,
      "step": 634
    },
    {
      "epoch": 0.06277339791908658,
      "grad_norm": 3.4794790875591657,
      "learning_rate": 1.994388324098916e-07,
      "loss": 0.771,
      "step": 635
    },
    {
      "epoch": 0.06287225366384104,
      "grad_norm": 6.291936548838725,
      "learning_rate": 1.9943543974608564e-07,
      "loss": 0.7667,
      "step": 636
    },
    {
      "epoch": 0.0629711094085955,
      "grad_norm": 7.771698903118231,
      "learning_rate": 1.9943203688664217e-07,
      "loss": 0.6848,
      "step": 637
    },
    {
      "epoch": 0.06306996515334998,
      "grad_norm": 27.328515965799834,
      "learning_rate": 1.9942862383191e-07,
      "loss": 0.7659,
      "step": 638
    },
    {
      "epoch": 0.06316882089810444,
      "grad_norm": 4.1937092196964025,
      "learning_rate": 1.9942520058223918e-07,
      "loss": 0.7268,
      "step": 639
    },
    {
      "epoch": 0.06326767664285891,
      "grad_norm": 11.373640171137183,
      "learning_rate": 1.9942176713798063e-07,
      "loss": 0.7306,
      "step": 640
    },
    {
      "epoch": 0.06336653238761338,
      "grad_norm": 5.345114298656406,
      "learning_rate": 1.9941832349948643e-07,
      "loss": 0.6483,
      "step": 641
    },
    {
      "epoch": 0.06346538813236784,
      "grad_norm": 3.8407595679524307,
      "learning_rate": 1.9941486966710972e-07,
      "loss": 0.8045,
      "step": 642
    },
    {
      "epoch": 0.06356424387712231,
      "grad_norm": 4.30973568003605,
      "learning_rate": 1.9941140564120457e-07,
      "loss": 0.7329,
      "step": 643
    },
    {
      "epoch": 0.06366309962187677,
      "grad_norm": 3.66491375611255,
      "learning_rate": 1.994079314221262e-07,
      "loss": 0.7248,
      "step": 644
    },
    {
      "epoch": 0.06376195536663125,
      "grad_norm": 3.7657527531925097,
      "learning_rate": 1.994044470102308e-07,
      "loss": 0.7249,
      "step": 645
    },
    {
      "epoch": 0.06386081111138571,
      "grad_norm": 25.55088229929469,
      "learning_rate": 1.994009524058757e-07,
      "loss": 0.6385,
      "step": 646
    },
    {
      "epoch": 0.06395966685614017,
      "grad_norm": 5.531713167280419,
      "learning_rate": 1.9939744760941922e-07,
      "loss": 0.7285,
      "step": 647
    },
    {
      "epoch": 0.06405852260089465,
      "grad_norm": 4.400238289529198,
      "learning_rate": 1.9939393262122067e-07,
      "loss": 0.7689,
      "step": 648
    },
    {
      "epoch": 0.06415737834564911,
      "grad_norm": 4.858059268299443,
      "learning_rate": 1.993904074416405e-07,
      "loss": 0.7799,
      "step": 649
    },
    {
      "epoch": 0.06425623409040358,
      "grad_norm": 7.258924284412419,
      "learning_rate": 1.993868720710402e-07,
      "loss": 0.8285,
      "step": 650
    },
    {
      "epoch": 0.06435508983515804,
      "grad_norm": 5.060143218946353,
      "learning_rate": 1.9938332650978217e-07,
      "loss": 0.7457,
      "step": 651
    },
    {
      "epoch": 0.06445394557991252,
      "grad_norm": 5.72119992213571,
      "learning_rate": 1.9937977075823005e-07,
      "loss": 0.7443,
      "step": 652
    },
    {
      "epoch": 0.06455280132466698,
      "grad_norm": 4.890847574554586,
      "learning_rate": 1.9937620481674837e-07,
      "loss": 0.811,
      "step": 653
    },
    {
      "epoch": 0.06465165706942144,
      "grad_norm": 7.7869397889051095,
      "learning_rate": 1.9937262868570279e-07,
      "loss": 0.679,
      "step": 654
    },
    {
      "epoch": 0.06475051281417592,
      "grad_norm": 7.229676300629243,
      "learning_rate": 1.9936904236546002e-07,
      "loss": 0.7437,
      "step": 655
    },
    {
      "epoch": 0.06484936855893038,
      "grad_norm": 15.38591407667973,
      "learning_rate": 1.9936544585638772e-07,
      "loss": 0.7672,
      "step": 656
    },
    {
      "epoch": 0.06494822430368485,
      "grad_norm": 6.745946086128724,
      "learning_rate": 1.9936183915885468e-07,
      "loss": 0.7958,
      "step": 657
    },
    {
      "epoch": 0.06504708004843932,
      "grad_norm": 3.947889297936475,
      "learning_rate": 1.9935822227323077e-07,
      "loss": 0.8445,
      "step": 658
    },
    {
      "epoch": 0.06514593579319378,
      "grad_norm": 7.155655628276617,
      "learning_rate": 1.9935459519988677e-07,
      "loss": 0.7605,
      "step": 659
    },
    {
      "epoch": 0.06524479153794825,
      "grad_norm": 3.517338056352976,
      "learning_rate": 1.9935095793919463e-07,
      "loss": 0.7676,
      "step": 660
    },
    {
      "epoch": 0.06534364728270271,
      "grad_norm": 3.469106887177477,
      "learning_rate": 1.9934731049152727e-07,
      "loss": 0.7434,
      "step": 661
    },
    {
      "epoch": 0.06544250302745719,
      "grad_norm": 5.588840309280119,
      "learning_rate": 1.993436528572587e-07,
      "loss": 0.7828,
      "step": 662
    },
    {
      "epoch": 0.06554135877221165,
      "grad_norm": 3.922214188420278,
      "learning_rate": 1.9933998503676394e-07,
      "loss": 0.7857,
      "step": 663
    },
    {
      "epoch": 0.06564021451696611,
      "grad_norm": 10.07131497745927,
      "learning_rate": 1.993363070304191e-07,
      "loss": 0.7161,
      "step": 664
    },
    {
      "epoch": 0.06573907026172059,
      "grad_norm": 3.617634166245376,
      "learning_rate": 1.993326188386013e-07,
      "loss": 0.7259,
      "step": 665
    },
    {
      "epoch": 0.06583792600647505,
      "grad_norm": 9.172056109781693,
      "learning_rate": 1.9932892046168867e-07,
      "loss": 0.7009,
      "step": 666
    },
    {
      "epoch": 0.06593678175122952,
      "grad_norm": 5.047590614155565,
      "learning_rate": 1.9932521190006046e-07,
      "loss": 0.7824,
      "step": 667
    },
    {
      "epoch": 0.06603563749598398,
      "grad_norm": 3.10977448498011,
      "learning_rate": 1.9932149315409692e-07,
      "loss": 0.6725,
      "step": 668
    },
    {
      "epoch": 0.06613449324073845,
      "grad_norm": 24.761525998573045,
      "learning_rate": 1.9931776422417937e-07,
      "loss": 0.8204,
      "step": 669
    },
    {
      "epoch": 0.06623334898549292,
      "grad_norm": 3.942792193979351,
      "learning_rate": 1.993140251106901e-07,
      "loss": 0.7908,
      "step": 670
    },
    {
      "epoch": 0.06633220473024738,
      "grad_norm": 8.355570942070676,
      "learning_rate": 1.9931027581401255e-07,
      "loss": 0.7062,
      "step": 671
    },
    {
      "epoch": 0.06643106047500186,
      "grad_norm": 4.6033483075338895,
      "learning_rate": 1.993065163345312e-07,
      "loss": 0.7105,
      "step": 672
    },
    {
      "epoch": 0.06652991621975632,
      "grad_norm": 9.67939959813092,
      "learning_rate": 1.993027466726314e-07,
      "loss": 0.7591,
      "step": 673
    },
    {
      "epoch": 0.0666287719645108,
      "grad_norm": 7.083669208638802,
      "learning_rate": 1.9929896682869977e-07,
      "loss": 0.6727,
      "step": 674
    },
    {
      "epoch": 0.06672762770926526,
      "grad_norm": 3.058717258565138,
      "learning_rate": 1.9929517680312386e-07,
      "loss": 0.6755,
      "step": 675
    },
    {
      "epoch": 0.06682648345401972,
      "grad_norm": 4.516076094137312,
      "learning_rate": 1.992913765962923e-07,
      "loss": 0.6971,
      "step": 676
    },
    {
      "epoch": 0.06692533919877419,
      "grad_norm": 5.745370500179706,
      "learning_rate": 1.992875662085947e-07,
      "loss": 0.825,
      "step": 677
    },
    {
      "epoch": 0.06702419494352865,
      "grad_norm": 18.566326875176973,
      "learning_rate": 1.9928374564042176e-07,
      "loss": 0.8062,
      "step": 678
    },
    {
      "epoch": 0.06712305068828313,
      "grad_norm": 6.8621971288259935,
      "learning_rate": 1.992799148921653e-07,
      "loss": 0.8035,
      "step": 679
    },
    {
      "epoch": 0.06722190643303759,
      "grad_norm": 3.739600262633109,
      "learning_rate": 1.99276073964218e-07,
      "loss": 0.6959,
      "step": 680
    },
    {
      "epoch": 0.06732076217779205,
      "grad_norm": 3.663244751054349,
      "learning_rate": 1.9927222285697375e-07,
      "loss": 0.7894,
      "step": 681
    },
    {
      "epoch": 0.06741961792254653,
      "grad_norm": 4.417755463361492,
      "learning_rate": 1.992683615708274e-07,
      "loss": 0.7265,
      "step": 682
    },
    {
      "epoch": 0.06751847366730099,
      "grad_norm": 4.362436476373061,
      "learning_rate": 1.992644901061749e-07,
      "loss": 0.7141,
      "step": 683
    },
    {
      "epoch": 0.06761732941205546,
      "grad_norm": 8.292734662051044,
      "learning_rate": 1.992606084634132e-07,
      "loss": 0.7415,
      "step": 684
    },
    {
      "epoch": 0.06771618515680992,
      "grad_norm": 5.489698564883644,
      "learning_rate": 1.9925671664294028e-07,
      "loss": 0.8587,
      "step": 685
    },
    {
      "epoch": 0.06781504090156439,
      "grad_norm": 3.596805019634159,
      "learning_rate": 1.9925281464515522e-07,
      "loss": 0.7083,
      "step": 686
    },
    {
      "epoch": 0.06791389664631886,
      "grad_norm": 8.57288755803437,
      "learning_rate": 1.9924890247045811e-07,
      "loss": 0.7754,
      "step": 687
    },
    {
      "epoch": 0.06801275239107332,
      "grad_norm": 4.806516791011535,
      "learning_rate": 1.9924498011925008e-07,
      "loss": 0.7973,
      "step": 688
    },
    {
      "epoch": 0.0681116081358278,
      "grad_norm": 3.9078803043106527,
      "learning_rate": 1.9924104759193327e-07,
      "loss": 0.7071,
      "step": 689
    },
    {
      "epoch": 0.06821046388058226,
      "grad_norm": 7.114712903518572,
      "learning_rate": 1.9923710488891098e-07,
      "loss": 0.7833,
      "step": 690
    },
    {
      "epoch": 0.06830931962533673,
      "grad_norm": 4.053358353645886,
      "learning_rate": 1.992331520105874e-07,
      "loss": 0.8274,
      "step": 691
    },
    {
      "epoch": 0.0684081753700912,
      "grad_norm": 22.38140021424328,
      "learning_rate": 1.992291889573679e-07,
      "loss": 0.7256,
      "step": 692
    },
    {
      "epoch": 0.06850703111484566,
      "grad_norm": 3.2279270111388048,
      "learning_rate": 1.9922521572965877e-07,
      "loss": 0.7996,
      "step": 693
    },
    {
      "epoch": 0.06860588685960013,
      "grad_norm": 5.8828313406716095,
      "learning_rate": 1.9922123232786748e-07,
      "loss": 0.7702,
      "step": 694
    },
    {
      "epoch": 0.0687047426043546,
      "grad_norm": 3.952389997780548,
      "learning_rate": 1.992172387524024e-07,
      "loss": 0.7501,
      "step": 695
    },
    {
      "epoch": 0.06880359834910907,
      "grad_norm": 5.059293401018895,
      "learning_rate": 1.9921323500367306e-07,
      "loss": 0.7629,
      "step": 696
    },
    {
      "epoch": 0.06890245409386353,
      "grad_norm": 6.029913322129757,
      "learning_rate": 1.9920922108208996e-07,
      "loss": 0.8431,
      "step": 697
    },
    {
      "epoch": 0.06900130983861799,
      "grad_norm": 3.7737552386342594,
      "learning_rate": 1.9920519698806469e-07,
      "loss": 0.6893,
      "step": 698
    },
    {
      "epoch": 0.06910016558337247,
      "grad_norm": 4.231710088102548,
      "learning_rate": 1.9920116272200987e-07,
      "loss": 0.7673,
      "step": 699
    },
    {
      "epoch": 0.06919902132812693,
      "grad_norm": 14.754044814347832,
      "learning_rate": 1.991971182843391e-07,
      "loss": 0.7091,
      "step": 700
    },
    {
      "epoch": 0.0692978770728814,
      "grad_norm": 5.4321874193895345,
      "learning_rate": 1.991930636754671e-07,
      "loss": 0.7892,
      "step": 701
    },
    {
      "epoch": 0.06939673281763586,
      "grad_norm": 5.012149811902219,
      "learning_rate": 1.9918899889580962e-07,
      "loss": 0.769,
      "step": 702
    },
    {
      "epoch": 0.06949558856239033,
      "grad_norm": 3.4596001879756217,
      "learning_rate": 1.9918492394578347e-07,
      "loss": 0.6569,
      "step": 703
    },
    {
      "epoch": 0.0695944443071448,
      "grad_norm": 3.2515720722496755,
      "learning_rate": 1.9918083882580644e-07,
      "loss": 0.6982,
      "step": 704
    },
    {
      "epoch": 0.06969330005189926,
      "grad_norm": 3.1027587528774276,
      "learning_rate": 1.9917674353629743e-07,
      "loss": 0.7664,
      "step": 705
    },
    {
      "epoch": 0.06979215579665374,
      "grad_norm": 3.7072685611372616,
      "learning_rate": 1.9917263807767627e-07,
      "loss": 0.806,
      "step": 706
    },
    {
      "epoch": 0.0698910115414082,
      "grad_norm": 26.561335475692584,
      "learning_rate": 1.9916852245036404e-07,
      "loss": 0.8222,
      "step": 707
    },
    {
      "epoch": 0.06998986728616266,
      "grad_norm": 7.824107075299489,
      "learning_rate": 1.9916439665478265e-07,
      "loss": 0.783,
      "step": 708
    },
    {
      "epoch": 0.07008872303091714,
      "grad_norm": 7.776914185685562,
      "learning_rate": 1.9916026069135516e-07,
      "loss": 0.7119,
      "step": 709
    },
    {
      "epoch": 0.0701875787756716,
      "grad_norm": 3.622366111082512,
      "learning_rate": 1.9915611456050563e-07,
      "loss": 0.7765,
      "step": 710
    },
    {
      "epoch": 0.07028643452042607,
      "grad_norm": 3.7378901464724184,
      "learning_rate": 1.9915195826265924e-07,
      "loss": 0.7432,
      "step": 711
    },
    {
      "epoch": 0.07038529026518053,
      "grad_norm": 6.189146855791468,
      "learning_rate": 1.9914779179824212e-07,
      "loss": 0.6783,
      "step": 712
    },
    {
      "epoch": 0.07048414600993501,
      "grad_norm": 7.550242476675408,
      "learning_rate": 1.9914361516768145e-07,
      "loss": 0.6958,
      "step": 713
    },
    {
      "epoch": 0.07058300175468947,
      "grad_norm": 5.027182750871855,
      "learning_rate": 1.9913942837140554e-07,
      "loss": 0.848,
      "step": 714
    },
    {
      "epoch": 0.07068185749944393,
      "grad_norm": 3.8080429553909614,
      "learning_rate": 1.9913523140984367e-07,
      "loss": 0.7291,
      "step": 715
    },
    {
      "epoch": 0.0707807132441984,
      "grad_norm": 4.058155457610422,
      "learning_rate": 1.9913102428342614e-07,
      "loss": 0.8742,
      "step": 716
    },
    {
      "epoch": 0.07087956898895287,
      "grad_norm": 4.71909905231287,
      "learning_rate": 1.991268069925844e-07,
      "loss": 0.7738,
      "step": 717
    },
    {
      "epoch": 0.07097842473370734,
      "grad_norm": 7.7816740597923175,
      "learning_rate": 1.991225795377508e-07,
      "loss": 0.7115,
      "step": 718
    },
    {
      "epoch": 0.0710772804784618,
      "grad_norm": 3.594697127816907,
      "learning_rate": 1.991183419193588e-07,
      "loss": 0.7341,
      "step": 719
    },
    {
      "epoch": 0.07117613622321627,
      "grad_norm": 3.6824583745598085,
      "learning_rate": 1.9911409413784298e-07,
      "loss": 0.771,
      "step": 720
    },
    {
      "epoch": 0.07127499196797074,
      "grad_norm": 5.7014248291136225,
      "learning_rate": 1.991098361936388e-07,
      "loss": 0.8336,
      "step": 721
    },
    {
      "epoch": 0.0713738477127252,
      "grad_norm": 5.763057597012828,
      "learning_rate": 1.9910556808718293e-07,
      "loss": 0.795,
      "step": 722
    },
    {
      "epoch": 0.07147270345747968,
      "grad_norm": 24.317252156960127,
      "learning_rate": 1.9910128981891296e-07,
      "loss": 0.7262,
      "step": 723
    },
    {
      "epoch": 0.07157155920223414,
      "grad_norm": 4.145075407351751,
      "learning_rate": 1.9909700138926755e-07,
      "loss": 0.8862,
      "step": 724
    },
    {
      "epoch": 0.0716704149469886,
      "grad_norm": 4.640838633475026,
      "learning_rate": 1.9909270279868645e-07,
      "loss": 0.7328,
      "step": 725
    },
    {
      "epoch": 0.07176927069174308,
      "grad_norm": 10.418647182247435,
      "learning_rate": 1.990883940476104e-07,
      "loss": 0.7135,
      "step": 726
    },
    {
      "epoch": 0.07186812643649754,
      "grad_norm": 4.395141068248769,
      "learning_rate": 1.9908407513648117e-07,
      "loss": 0.7018,
      "step": 727
    },
    {
      "epoch": 0.07196698218125201,
      "grad_norm": 4.329061315465672,
      "learning_rate": 1.9907974606574163e-07,
      "loss": 0.7926,
      "step": 728
    },
    {
      "epoch": 0.07206583792600647,
      "grad_norm": 4.516061543655133,
      "learning_rate": 1.9907540683583567e-07,
      "loss": 0.8152,
      "step": 729
    },
    {
      "epoch": 0.07216469367076095,
      "grad_norm": 4.339204628915288,
      "learning_rate": 1.990710574472082e-07,
      "loss": 0.798,
      "step": 730
    },
    {
      "epoch": 0.07226354941551541,
      "grad_norm": 3.836271426299221,
      "learning_rate": 1.9906669790030516e-07,
      "loss": 0.7091,
      "step": 731
    },
    {
      "epoch": 0.07236240516026987,
      "grad_norm": 4.304068222738715,
      "learning_rate": 1.9906232819557362e-07,
      "loss": 0.7338,
      "step": 732
    },
    {
      "epoch": 0.07246126090502435,
      "grad_norm": 6.434619047274208,
      "learning_rate": 1.990579483334616e-07,
      "loss": 0.6505,
      "step": 733
    },
    {
      "epoch": 0.07256011664977881,
      "grad_norm": 7.030358176849566,
      "learning_rate": 1.990535583144182e-07,
      "loss": 0.7025,
      "step": 734
    },
    {
      "epoch": 0.07265897239453328,
      "grad_norm": 6.167883454262384,
      "learning_rate": 1.990491581388935e-07,
      "loss": 0.7471,
      "step": 735
    },
    {
      "epoch": 0.07275782813928774,
      "grad_norm": 6.2570900528319875,
      "learning_rate": 1.9904474780733874e-07,
      "loss": 0.7548,
      "step": 736
    },
    {
      "epoch": 0.0728566838840422,
      "grad_norm": 4.039410039137964,
      "learning_rate": 1.990403273202061e-07,
      "loss": 0.6529,
      "step": 737
    },
    {
      "epoch": 0.07295553962879668,
      "grad_norm": 3.795692911038379,
      "learning_rate": 1.9903589667794882e-07,
      "loss": 0.7442,
      "step": 738
    },
    {
      "epoch": 0.07305439537355114,
      "grad_norm": 4.080900896114658,
      "learning_rate": 1.9903145588102124e-07,
      "loss": 0.8195,
      "step": 739
    },
    {
      "epoch": 0.07315325111830562,
      "grad_norm": 3.171736275523173,
      "learning_rate": 1.9902700492987866e-07,
      "loss": 0.7191,
      "step": 740
    },
    {
      "epoch": 0.07325210686306008,
      "grad_norm": 4.305999994181826,
      "learning_rate": 1.9902254382497748e-07,
      "loss": 0.6409,
      "step": 741
    },
    {
      "epoch": 0.07335096260781454,
      "grad_norm": 5.608596448884323,
      "learning_rate": 1.9901807256677512e-07,
      "loss": 0.7246,
      "step": 742
    },
    {
      "epoch": 0.07344981835256902,
      "grad_norm": 4.501493468624949,
      "learning_rate": 1.9901359115573006e-07,
      "loss": 0.6875,
      "step": 743
    },
    {
      "epoch": 0.07354867409732348,
      "grad_norm": 10.209197672187054,
      "learning_rate": 1.9900909959230173e-07,
      "loss": 0.7652,
      "step": 744
    },
    {
      "epoch": 0.07364752984207795,
      "grad_norm": 3.0972966120161596,
      "learning_rate": 1.9900459787695078e-07,
      "loss": 0.6976,
      "step": 745
    },
    {
      "epoch": 0.07374638558683241,
      "grad_norm": 39.16403054221962,
      "learning_rate": 1.990000860101387e-07,
      "loss": 0.7273,
      "step": 746
    },
    {
      "epoch": 0.07384524133158687,
      "grad_norm": 7.2527880937637965,
      "learning_rate": 1.9899556399232816e-07,
      "loss": 0.6912,
      "step": 747
    },
    {
      "epoch": 0.07394409707634135,
      "grad_norm": 3.607023924978217,
      "learning_rate": 1.9899103182398282e-07,
      "loss": 0.7301,
      "step": 748
    },
    {
      "epoch": 0.07404295282109581,
      "grad_norm": 3.749955276091892,
      "learning_rate": 1.989864895055674e-07,
      "loss": 0.7225,
      "step": 749
    },
    {
      "epoch": 0.07414180856585029,
      "grad_norm": 12.680948495817658,
      "learning_rate": 1.9898193703754763e-07,
      "loss": 0.7612,
      "step": 750
    },
    {
      "epoch": 0.07424066431060475,
      "grad_norm": 7.764726194697529,
      "learning_rate": 1.989773744203903e-07,
      "loss": 0.8334,
      "step": 751
    },
    {
      "epoch": 0.07433952005535922,
      "grad_norm": 3.4362378962387345,
      "learning_rate": 1.9897280165456322e-07,
      "loss": 0.7719,
      "step": 752
    },
    {
      "epoch": 0.07443837580011368,
      "grad_norm": 3.5780340402541704,
      "learning_rate": 1.9896821874053532e-07,
      "loss": 0.7446,
      "step": 753
    },
    {
      "epoch": 0.07453723154486815,
      "grad_norm": 3.2078953717286662,
      "learning_rate": 1.9896362567877644e-07,
      "loss": 0.7293,
      "step": 754
    },
    {
      "epoch": 0.07463608728962262,
      "grad_norm": 7.78350898021283,
      "learning_rate": 1.9895902246975757e-07,
      "loss": 0.6874,
      "step": 755
    },
    {
      "epoch": 0.07473494303437708,
      "grad_norm": 5.352410268779308,
      "learning_rate": 1.9895440911395073e-07,
      "loss": 0.7125,
      "step": 756
    },
    {
      "epoch": 0.07483379877913156,
      "grad_norm": 3.7735431196452227,
      "learning_rate": 1.9894978561182887e-07,
      "loss": 0.862,
      "step": 757
    },
    {
      "epoch": 0.07493265452388602,
      "grad_norm": 8.809300514529598,
      "learning_rate": 1.989451519638661e-07,
      "loss": 0.7905,
      "step": 758
    },
    {
      "epoch": 0.07503151026864048,
      "grad_norm": 4.884794697963846,
      "learning_rate": 1.989405081705376e-07,
      "loss": 0.8184,
      "step": 759
    },
    {
      "epoch": 0.07513036601339496,
      "grad_norm": 8.456171736946942,
      "learning_rate": 1.9893585423231943e-07,
      "loss": 0.7935,
      "step": 760
    },
    {
      "epoch": 0.07522922175814942,
      "grad_norm": 8.555494271273227,
      "learning_rate": 1.989311901496888e-07,
      "loss": 0.7945,
      "step": 761
    },
    {
      "epoch": 0.07532807750290389,
      "grad_norm": 4.325461156044643,
      "learning_rate": 1.9892651592312396e-07,
      "loss": 0.8131,
      "step": 762
    },
    {
      "epoch": 0.07542693324765835,
      "grad_norm": 5.466562070992862,
      "learning_rate": 1.989218315531042e-07,
      "loss": 0.7042,
      "step": 763
    },
    {
      "epoch": 0.07552578899241282,
      "grad_norm": 3.6685325391509176,
      "learning_rate": 1.9891713704010982e-07,
      "loss": 0.727,
      "step": 764
    },
    {
      "epoch": 0.07562464473716729,
      "grad_norm": 7.232233865782656,
      "learning_rate": 1.9891243238462216e-07,
      "loss": 0.743,
      "step": 765
    },
    {
      "epoch": 0.07572350048192175,
      "grad_norm": 10.13390524854547,
      "learning_rate": 1.9890771758712362e-07,
      "loss": 0.6879,
      "step": 766
    },
    {
      "epoch": 0.07582235622667623,
      "grad_norm": 7.944449062723659,
      "learning_rate": 1.989029926480976e-07,
      "loss": 0.6923,
      "step": 767
    },
    {
      "epoch": 0.07592121197143069,
      "grad_norm": 3.829357682804349,
      "learning_rate": 1.9889825756802865e-07,
      "loss": 0.7916,
      "step": 768
    },
    {
      "epoch": 0.07602006771618516,
      "grad_norm": 3.851113104447836,
      "learning_rate": 1.988935123474022e-07,
      "loss": 0.7054,
      "step": 769
    },
    {
      "epoch": 0.07611892346093962,
      "grad_norm": 3.0377050449586904,
      "learning_rate": 1.9888875698670486e-07,
      "loss": 0.7536,
      "step": 770
    },
    {
      "epoch": 0.07621777920569409,
      "grad_norm": 3.5991429727313333,
      "learning_rate": 1.988839914864242e-07,
      "loss": 0.7309,
      "step": 771
    },
    {
      "epoch": 0.07631663495044856,
      "grad_norm": 4.521781137218234,
      "learning_rate": 1.9887921584704888e-07,
      "loss": 0.8344,
      "step": 772
    },
    {
      "epoch": 0.07641549069520302,
      "grad_norm": 7.279719438255686,
      "learning_rate": 1.9887443006906853e-07,
      "loss": 0.7766,
      "step": 773
    },
    {
      "epoch": 0.0765143464399575,
      "grad_norm": 9.11961298703795,
      "learning_rate": 1.9886963415297384e-07,
      "loss": 0.7536,
      "step": 774
    },
    {
      "epoch": 0.07661320218471196,
      "grad_norm": 5.225782649350661,
      "learning_rate": 1.9886482809925663e-07,
      "loss": 0.7133,
      "step": 775
    },
    {
      "epoch": 0.07671205792946642,
      "grad_norm": 3.484893864137296,
      "learning_rate": 1.9886001190840965e-07,
      "loss": 0.8302,
      "step": 776
    },
    {
      "epoch": 0.0768109136742209,
      "grad_norm": 3.5253487712773746,
      "learning_rate": 1.9885518558092675e-07,
      "loss": 0.7662,
      "step": 777
    },
    {
      "epoch": 0.07690976941897536,
      "grad_norm": 4.296758766405044,
      "learning_rate": 1.9885034911730277e-07,
      "loss": 0.7891,
      "step": 778
    },
    {
      "epoch": 0.07700862516372983,
      "grad_norm": 4.962800844289576,
      "learning_rate": 1.9884550251803365e-07,
      "loss": 0.8553,
      "step": 779
    },
    {
      "epoch": 0.0771074809084843,
      "grad_norm": 6.025076616658453,
      "learning_rate": 1.988406457836163e-07,
      "loss": 0.639,
      "step": 780
    },
    {
      "epoch": 0.07720633665323876,
      "grad_norm": 4.75313492543018,
      "learning_rate": 1.9883577891454874e-07,
      "loss": 0.8041,
      "step": 781
    },
    {
      "epoch": 0.07730519239799323,
      "grad_norm": 3.7474338993119094,
      "learning_rate": 1.9883090191132997e-07,
      "loss": 0.7953,
      "step": 782
    },
    {
      "epoch": 0.07740404814274769,
      "grad_norm": 4.852853186612177,
      "learning_rate": 1.9882601477446007e-07,
      "loss": 0.7452,
      "step": 783
    },
    {
      "epoch": 0.07750290388750217,
      "grad_norm": 4.413149669292896,
      "learning_rate": 1.9882111750444018e-07,
      "loss": 0.8106,
      "step": 784
    },
    {
      "epoch": 0.07760175963225663,
      "grad_norm": 3.721686318953266,
      "learning_rate": 1.9881621010177235e-07,
      "loss": 0.7316,
      "step": 785
    },
    {
      "epoch": 0.07770061537701109,
      "grad_norm": 5.2104866879326845,
      "learning_rate": 1.9881129256695983e-07,
      "loss": 0.8006,
      "step": 786
    },
    {
      "epoch": 0.07779947112176556,
      "grad_norm": 3.891899037836648,
      "learning_rate": 1.9880636490050686e-07,
      "loss": 0.7691,
      "step": 787
    },
    {
      "epoch": 0.07789832686652003,
      "grad_norm": 5.005587609141315,
      "learning_rate": 1.9880142710291863e-07,
      "loss": 0.7833,
      "step": 788
    },
    {
      "epoch": 0.0779971826112745,
      "grad_norm": 4.155246434173738,
      "learning_rate": 1.987964791747015e-07,
      "loss": 0.8192,
      "step": 789
    },
    {
      "epoch": 0.07809603835602896,
      "grad_norm": 4.185286011202474,
      "learning_rate": 1.9879152111636275e-07,
      "loss": 0.7309,
      "step": 790
    },
    {
      "epoch": 0.07819489410078344,
      "grad_norm": 4.662798597717031,
      "learning_rate": 1.987865529284108e-07,
      "loss": 0.7921,
      "step": 791
    },
    {
      "epoch": 0.0782937498455379,
      "grad_norm": 5.658409162286452,
      "learning_rate": 1.9878157461135509e-07,
      "loss": 0.7284,
      "step": 792
    },
    {
      "epoch": 0.07839260559029236,
      "grad_norm": 9.44546993410144,
      "learning_rate": 1.9877658616570597e-07,
      "loss": 0.7405,
      "step": 793
    },
    {
      "epoch": 0.07849146133504684,
      "grad_norm": 3.4992306426165904,
      "learning_rate": 1.98771587591975e-07,
      "loss": 0.6944,
      "step": 794
    },
    {
      "epoch": 0.0785903170798013,
      "grad_norm": 4.104978737836563,
      "learning_rate": 1.9876657889067476e-07,
      "loss": 0.7596,
      "step": 795
    },
    {
      "epoch": 0.07868917282455577,
      "grad_norm": 5.214548542970481,
      "learning_rate": 1.987615600623187e-07,
      "loss": 0.8185,
      "step": 796
    },
    {
      "epoch": 0.07878802856931023,
      "grad_norm": 17.965256797520407,
      "learning_rate": 1.9875653110742157e-07,
      "loss": 0.8118,
      "step": 797
    },
    {
      "epoch": 0.0788868843140647,
      "grad_norm": 4.829172213887559,
      "learning_rate": 1.9875149202649886e-07,
      "loss": 0.6847,
      "step": 798
    },
    {
      "epoch": 0.07898574005881917,
      "grad_norm": 4.105287199321695,
      "learning_rate": 1.9874644282006735e-07,
      "loss": 0.7341,
      "step": 799
    },
    {
      "epoch": 0.07908459580357363,
      "grad_norm": 4.226025231830703,
      "learning_rate": 1.9874138348864476e-07,
      "loss": 0.7034,
      "step": 800
    },
    {
      "epoch": 0.07918345154832811,
      "grad_norm": 3.630380700965607,
      "learning_rate": 1.9873631403274983e-07,
      "loss": 0.8017,
      "step": 801
    },
    {
      "epoch": 0.07928230729308257,
      "grad_norm": 6.659541247636427,
      "learning_rate": 1.9873123445290236e-07,
      "loss": 0.7424,
      "step": 802
    },
    {
      "epoch": 0.07938116303783703,
      "grad_norm": 19.23650281561679,
      "learning_rate": 1.9872614474962317e-07,
      "loss": 0.7158,
      "step": 803
    },
    {
      "epoch": 0.0794800187825915,
      "grad_norm": 3.929424724418909,
      "learning_rate": 1.9872104492343418e-07,
      "loss": 0.6706,
      "step": 804
    },
    {
      "epoch": 0.07957887452734597,
      "grad_norm": 6.905009982565782,
      "learning_rate": 1.9871593497485823e-07,
      "loss": 0.7558,
      "step": 805
    },
    {
      "epoch": 0.07967773027210044,
      "grad_norm": 12.255420033554843,
      "learning_rate": 1.9871081490441935e-07,
      "loss": 0.7736,
      "step": 806
    },
    {
      "epoch": 0.0797765860168549,
      "grad_norm": 5.250536386257707,
      "learning_rate": 1.9870568471264243e-07,
      "loss": 0.8035,
      "step": 807
    },
    {
      "epoch": 0.07987544176160938,
      "grad_norm": 5.354749942689716,
      "learning_rate": 1.9870054440005362e-07,
      "loss": 0.7937,
      "step": 808
    },
    {
      "epoch": 0.07997429750636384,
      "grad_norm": 8.837205278518203,
      "learning_rate": 1.9869539396717988e-07,
      "loss": 0.7456,
      "step": 809
    },
    {
      "epoch": 0.0800731532511183,
      "grad_norm": 5.3864153004375295,
      "learning_rate": 1.9869023341454935e-07,
      "loss": 0.8031,
      "step": 810
    },
    {
      "epoch": 0.08017200899587278,
      "grad_norm": 6.15435767117148,
      "learning_rate": 1.986850627426912e-07,
      "loss": 0.8313,
      "step": 811
    },
    {
      "epoch": 0.08027086474062724,
      "grad_norm": 4.245939297558537,
      "learning_rate": 1.9867988195213554e-07,
      "loss": 0.7846,
      "step": 812
    },
    {
      "epoch": 0.08036972048538171,
      "grad_norm": 6.239636917418423,
      "learning_rate": 1.986746910434136e-07,
      "loss": 0.8954,
      "step": 813
    },
    {
      "epoch": 0.08046857623013617,
      "grad_norm": 3.976873909313196,
      "learning_rate": 1.9866949001705768e-07,
      "loss": 0.8021,
      "step": 814
    },
    {
      "epoch": 0.08056743197489064,
      "grad_norm": 3.7468659372137556,
      "learning_rate": 1.9866427887360105e-07,
      "loss": 0.6576,
      "step": 815
    },
    {
      "epoch": 0.08066628771964511,
      "grad_norm": 30.690961391913937,
      "learning_rate": 1.98659057613578e-07,
      "loss": 0.6978,
      "step": 816
    },
    {
      "epoch": 0.08076514346439957,
      "grad_norm": 14.407851009278248,
      "learning_rate": 1.986538262375239e-07,
      "loss": 0.7896,
      "step": 817
    },
    {
      "epoch": 0.08086399920915405,
      "grad_norm": 3.2539214509653256,
      "learning_rate": 1.9864858474597516e-07,
      "loss": 0.6667,
      "step": 818
    },
    {
      "epoch": 0.08096285495390851,
      "grad_norm": 3.5579486011855552,
      "learning_rate": 1.9864333313946927e-07,
      "loss": 0.6281,
      "step": 819
    },
    {
      "epoch": 0.08106171069866297,
      "grad_norm": 7.384169343550695,
      "learning_rate": 1.9863807141854463e-07,
      "loss": 0.7666,
      "step": 820
    },
    {
      "epoch": 0.08116056644341745,
      "grad_norm": 3.8644177584333566,
      "learning_rate": 1.9863279958374076e-07,
      "loss": 0.6997,
      "step": 821
    },
    {
      "epoch": 0.0812594221881719,
      "grad_norm": 4.991519416288131,
      "learning_rate": 1.9862751763559827e-07,
      "loss": 0.7833,
      "step": 822
    },
    {
      "epoch": 0.08135827793292638,
      "grad_norm": 3.6610485188644057,
      "learning_rate": 1.9862222557465866e-07,
      "loss": 0.8443,
      "step": 823
    },
    {
      "epoch": 0.08145713367768084,
      "grad_norm": 4.922354990484463,
      "learning_rate": 1.9861692340146463e-07,
      "loss": 0.787,
      "step": 824
    },
    {
      "epoch": 0.0815559894224353,
      "grad_norm": 3.259339898108427,
      "learning_rate": 1.986116111165598e-07,
      "loss": 0.7704,
      "step": 825
    },
    {
      "epoch": 0.08165484516718978,
      "grad_norm": 5.328174756434093,
      "learning_rate": 1.9860628872048888e-07,
      "loss": 0.7362,
      "step": 826
    },
    {
      "epoch": 0.08175370091194424,
      "grad_norm": 4.372367691286817,
      "learning_rate": 1.9860095621379757e-07,
      "loss": 0.7859,
      "step": 827
    },
    {
      "epoch": 0.08185255665669872,
      "grad_norm": 10.38007253378599,
      "learning_rate": 1.9859561359703272e-07,
      "loss": 0.6434,
      "step": 828
    },
    {
      "epoch": 0.08195141240145318,
      "grad_norm": 4.415345800072433,
      "learning_rate": 1.9859026087074203e-07,
      "loss": 0.859,
      "step": 829
    },
    {
      "epoch": 0.08205026814620765,
      "grad_norm": 3.6221072347778955,
      "learning_rate": 1.9858489803547443e-07,
      "loss": 0.805,
      "step": 830
    },
    {
      "epoch": 0.08214912389096211,
      "grad_norm": 11.912724116335559,
      "learning_rate": 1.9857952509177974e-07,
      "loss": 0.7603,
      "step": 831
    },
    {
      "epoch": 0.08224797963571658,
      "grad_norm": 3.8282029364525885,
      "learning_rate": 1.985741420402089e-07,
      "loss": 0.8448,
      "step": 832
    },
    {
      "epoch": 0.08234683538047105,
      "grad_norm": 9.48714494587464,
      "learning_rate": 1.985687488813139e-07,
      "loss": 0.8396,
      "step": 833
    },
    {
      "epoch": 0.08244569112522551,
      "grad_norm": 4.27357580429121,
      "learning_rate": 1.9856334561564767e-07,
      "loss": 0.8055,
      "step": 834
    },
    {
      "epoch": 0.08254454686997999,
      "grad_norm": 4.294978182834366,
      "learning_rate": 1.9855793224376427e-07,
      "loss": 0.7707,
      "step": 835
    },
    {
      "epoch": 0.08264340261473445,
      "grad_norm": 3.87519284546518,
      "learning_rate": 1.985525087662187e-07,
      "loss": 0.7649,
      "step": 836
    },
    {
      "epoch": 0.08274225835948891,
      "grad_norm": 3.2009063024540763,
      "learning_rate": 1.9854707518356717e-07,
      "loss": 0.8032,
      "step": 837
    },
    {
      "epoch": 0.08284111410424339,
      "grad_norm": 5.419668170208262,
      "learning_rate": 1.9854163149636672e-07,
      "loss": 0.7129,
      "step": 838
    },
    {
      "epoch": 0.08293996984899785,
      "grad_norm": 5.322405729929826,
      "learning_rate": 1.9853617770517558e-07,
      "loss": 0.7338,
      "step": 839
    },
    {
      "epoch": 0.08303882559375232,
      "grad_norm": 3.522059614559362,
      "learning_rate": 1.9853071381055292e-07,
      "loss": 0.7693,
      "step": 840
    },
    {
      "epoch": 0.08313768133850678,
      "grad_norm": 4.8160918372272725,
      "learning_rate": 1.9852523981305896e-07,
      "loss": 0.737,
      "step": 841
    },
    {
      "epoch": 0.08323653708326124,
      "grad_norm": 3.610621735537937,
      "learning_rate": 1.9851975571325504e-07,
      "loss": 0.7458,
      "step": 842
    },
    {
      "epoch": 0.08333539282801572,
      "grad_norm": 4.988577800410283,
      "learning_rate": 1.9851426151170342e-07,
      "loss": 0.6967,
      "step": 843
    },
    {
      "epoch": 0.08343424857277018,
      "grad_norm": 2.89792044905663,
      "learning_rate": 1.9850875720896746e-07,
      "loss": 0.7069,
      "step": 844
    },
    {
      "epoch": 0.08353310431752466,
      "grad_norm": 6.99272429488204,
      "learning_rate": 1.985032428056116e-07,
      "loss": 0.7559,
      "step": 845
    },
    {
      "epoch": 0.08363196006227912,
      "grad_norm": 8.991531595643592,
      "learning_rate": 1.9849771830220118e-07,
      "loss": 0.8234,
      "step": 846
    },
    {
      "epoch": 0.08373081580703359,
      "grad_norm": 4.484469749634898,
      "learning_rate": 1.9849218369930268e-07,
      "loss": 0.7822,
      "step": 847
    },
    {
      "epoch": 0.08382967155178805,
      "grad_norm": 5.372239032979066,
      "learning_rate": 1.9848663899748363e-07,
      "loss": 0.6474,
      "step": 848
    },
    {
      "epoch": 0.08392852729654252,
      "grad_norm": 3.7097462411279647,
      "learning_rate": 1.9848108419731254e-07,
      "loss": 0.7021,
      "step": 849
    },
    {
      "epoch": 0.08402738304129699,
      "grad_norm": 3.7672815213882713,
      "learning_rate": 1.9847551929935893e-07,
      "loss": 0.7664,
      "step": 850
    },
    {
      "epoch": 0.08412623878605145,
      "grad_norm": 5.112768690470133,
      "learning_rate": 1.9846994430419347e-07,
      "loss": 0.6364,
      "step": 851
    },
    {
      "epoch": 0.08422509453080593,
      "grad_norm": 9.283184836367477,
      "learning_rate": 1.9846435921238772e-07,
      "loss": 0.8053,
      "step": 852
    },
    {
      "epoch": 0.08432395027556039,
      "grad_norm": 4.641899073408122,
      "learning_rate": 1.9845876402451439e-07,
      "loss": 0.7857,
      "step": 853
    },
    {
      "epoch": 0.08442280602031485,
      "grad_norm": 7.844723835520541,
      "learning_rate": 1.9845315874114719e-07,
      "loss": 0.6722,
      "step": 854
    },
    {
      "epoch": 0.08452166176506933,
      "grad_norm": 11.89001883872346,
      "learning_rate": 1.9844754336286083e-07,
      "loss": 0.7465,
      "step": 855
    },
    {
      "epoch": 0.08462051750982379,
      "grad_norm": 11.862602767462036,
      "learning_rate": 1.984419178902311e-07,
      "loss": 0.6613,
      "step": 856
    },
    {
      "epoch": 0.08471937325457826,
      "grad_norm": 4.486796269006515,
      "learning_rate": 1.9843628232383484e-07,
      "loss": 0.7362,
      "step": 857
    },
    {
      "epoch": 0.08481822899933272,
      "grad_norm": 4.14493865745818,
      "learning_rate": 1.9843063666424987e-07,
      "loss": 0.7703,
      "step": 858
    },
    {
      "epoch": 0.08491708474408718,
      "grad_norm": 52.992928550541514,
      "learning_rate": 1.9842498091205502e-07,
      "loss": 0.7132,
      "step": 859
    },
    {
      "epoch": 0.08501594048884166,
      "grad_norm": 6.535440644055109,
      "learning_rate": 1.9841931506783027e-07,
      "loss": 0.7032,
      "step": 860
    },
    {
      "epoch": 0.08511479623359612,
      "grad_norm": 5.624438777712978,
      "learning_rate": 1.9841363913215658e-07,
      "loss": 0.8004,
      "step": 861
    },
    {
      "epoch": 0.0852136519783506,
      "grad_norm": 6.287734676271625,
      "learning_rate": 1.9840795310561585e-07,
      "loss": 0.8073,
      "step": 862
    },
    {
      "epoch": 0.08531250772310506,
      "grad_norm": 3.930633126665939,
      "learning_rate": 1.9840225698879117e-07,
      "loss": 0.8182,
      "step": 863
    },
    {
      "epoch": 0.08541136346785952,
      "grad_norm": 3.5212968957296,
      "learning_rate": 1.9839655078226657e-07,
      "loss": 0.7972,
      "step": 864
    },
    {
      "epoch": 0.085510219212614,
      "grad_norm": 11.084579392991461,
      "learning_rate": 1.9839083448662712e-07,
      "loss": 0.7326,
      "step": 865
    },
    {
      "epoch": 0.08560907495736846,
      "grad_norm": 8.09834389690182,
      "learning_rate": 1.98385108102459e-07,
      "loss": 0.6885,
      "step": 866
    },
    {
      "epoch": 0.08570793070212293,
      "grad_norm": 9.462335183684687,
      "learning_rate": 1.9837937163034927e-07,
      "loss": 0.8104,
      "step": 867
    },
    {
      "epoch": 0.08580678644687739,
      "grad_norm": 3.8763047047789927,
      "learning_rate": 1.9837362507088624e-07,
      "loss": 0.7138,
      "step": 868
    },
    {
      "epoch": 0.08590564219163187,
      "grad_norm": 5.430271804166304,
      "learning_rate": 1.9836786842465903e-07,
      "loss": 0.678,
      "step": 869
    },
    {
      "epoch": 0.08600449793638633,
      "grad_norm": 4.0562925411616115,
      "learning_rate": 1.9836210169225796e-07,
      "loss": 0.6687,
      "step": 870
    },
    {
      "epoch": 0.08610335368114079,
      "grad_norm": 4.986835142943566,
      "learning_rate": 1.9835632487427428e-07,
      "loss": 0.7112,
      "step": 871
    },
    {
      "epoch": 0.08620220942589527,
      "grad_norm": 13.365202470400366,
      "learning_rate": 1.9835053797130037e-07,
      "loss": 0.7808,
      "step": 872
    },
    {
      "epoch": 0.08630106517064973,
      "grad_norm": 4.778659956489834,
      "learning_rate": 1.9834474098392957e-07,
      "loss": 0.7104,
      "step": 873
    },
    {
      "epoch": 0.0863999209154042,
      "grad_norm": 5.768771104939935,
      "learning_rate": 1.9833893391275625e-07,
      "loss": 0.7597,
      "step": 874
    },
    {
      "epoch": 0.08649877666015866,
      "grad_norm": 7.567129429976021,
      "learning_rate": 1.9833311675837584e-07,
      "loss": 0.7378,
      "step": 875
    },
    {
      "epoch": 0.08659763240491312,
      "grad_norm": 4.455226815361179,
      "learning_rate": 1.9832728952138487e-07,
      "loss": 0.7451,
      "step": 876
    },
    {
      "epoch": 0.0866964881496676,
      "grad_norm": 4.381278016031531,
      "learning_rate": 1.9832145220238075e-07,
      "loss": 0.7645,
      "step": 877
    },
    {
      "epoch": 0.08679534389442206,
      "grad_norm": 3.7943218022766114,
      "learning_rate": 1.9831560480196207e-07,
      "loss": 0.7358,
      "step": 878
    },
    {
      "epoch": 0.08689419963917654,
      "grad_norm": 3.696921007763004,
      "learning_rate": 1.9830974732072838e-07,
      "loss": 0.7855,
      "step": 879
    },
    {
      "epoch": 0.086993055383931,
      "grad_norm": 3.1371834363680806,
      "learning_rate": 1.9830387975928024e-07,
      "loss": 0.7547,
      "step": 880
    },
    {
      "epoch": 0.08709191112868546,
      "grad_norm": 4.64452160841938,
      "learning_rate": 1.9829800211821936e-07,
      "loss": 0.7795,
      "step": 881
    },
    {
      "epoch": 0.08719076687343993,
      "grad_norm": 4.0040460958056565,
      "learning_rate": 1.9829211439814832e-07,
      "loss": 0.8118,
      "step": 882
    },
    {
      "epoch": 0.0872896226181944,
      "grad_norm": 6.366119570437248,
      "learning_rate": 1.982862165996709e-07,
      "loss": 0.8181,
      "step": 883
    },
    {
      "epoch": 0.08738847836294887,
      "grad_norm": 4.515983980850584,
      "learning_rate": 1.9828030872339174e-07,
      "loss": 0.7759,
      "step": 884
    },
    {
      "epoch": 0.08748733410770333,
      "grad_norm": 3.902417988633249,
      "learning_rate": 1.982743907699167e-07,
      "loss": 0.6248,
      "step": 885
    },
    {
      "epoch": 0.08758618985245781,
      "grad_norm": 8.65243180095402,
      "learning_rate": 1.9826846273985252e-07,
      "loss": 0.8386,
      "step": 886
    },
    {
      "epoch": 0.08768504559721227,
      "grad_norm": 3.092683472698625,
      "learning_rate": 1.9826252463380704e-07,
      "loss": 0.8499,
      "step": 887
    },
    {
      "epoch": 0.08778390134196673,
      "grad_norm": 4.421673007825054,
      "learning_rate": 1.9825657645238912e-07,
      "loss": 0.7746,
      "step": 888
    },
    {
      "epoch": 0.0878827570867212,
      "grad_norm": 9.058862356024104,
      "learning_rate": 1.982506181962087e-07,
      "loss": 0.7161,
      "step": 889
    },
    {
      "epoch": 0.08798161283147567,
      "grad_norm": 5.978148399638466,
      "learning_rate": 1.9824464986587671e-07,
      "loss": 0.7142,
      "step": 890
    },
    {
      "epoch": 0.08808046857623014,
      "grad_norm": 7.241222338192638,
      "learning_rate": 1.9823867146200505e-07,
      "loss": 0.7484,
      "step": 891
    },
    {
      "epoch": 0.0881793243209846,
      "grad_norm": 5.250237427444657,
      "learning_rate": 1.9823268298520674e-07,
      "loss": 0.7677,
      "step": 892
    },
    {
      "epoch": 0.08827818006573906,
      "grad_norm": 15.90843219663121,
      "learning_rate": 1.9822668443609585e-07,
      "loss": 0.7517,
      "step": 893
    },
    {
      "epoch": 0.08837703581049354,
      "grad_norm": 11.376828907266571,
      "learning_rate": 1.982206758152874e-07,
      "loss": 0.6686,
      "step": 894
    },
    {
      "epoch": 0.088475891555248,
      "grad_norm": 5.809925732898762,
      "learning_rate": 1.982146571233975e-07,
      "loss": 0.7382,
      "step": 895
    },
    {
      "epoch": 0.08857474730000248,
      "grad_norm": 4.921938729652527,
      "learning_rate": 1.9820862836104332e-07,
      "loss": 0.7787,
      "step": 896
    },
    {
      "epoch": 0.08867360304475694,
      "grad_norm": 4.54734638410769,
      "learning_rate": 1.982025895288429e-07,
      "loss": 0.8104,
      "step": 897
    },
    {
      "epoch": 0.0887724587895114,
      "grad_norm": 5.196461481925448,
      "learning_rate": 1.9819654062741558e-07,
      "loss": 0.7759,
      "step": 898
    },
    {
      "epoch": 0.08887131453426587,
      "grad_norm": 3.6519244294956525,
      "learning_rate": 1.9819048165738153e-07,
      "loss": 0.8168,
      "step": 899
    },
    {
      "epoch": 0.08897017027902034,
      "grad_norm": 5.241362548827772,
      "learning_rate": 1.9818441261936196e-07,
      "loss": 0.6995,
      "step": 900
    },
    {
      "epoch": 0.08906902602377481,
      "grad_norm": 4.557603654576993,
      "learning_rate": 1.981783335139792e-07,
      "loss": 0.8512,
      "step": 901
    },
    {
      "epoch": 0.08916788176852927,
      "grad_norm": 3.208950772220537,
      "learning_rate": 1.981722443418566e-07,
      "loss": 0.7318,
      "step": 902
    },
    {
      "epoch": 0.08926673751328375,
      "grad_norm": 5.33583082029189,
      "learning_rate": 1.9816614510361844e-07,
      "loss": 0.7552,
      "step": 903
    },
    {
      "epoch": 0.08936559325803821,
      "grad_norm": 4.07701830814837,
      "learning_rate": 1.9816003579989018e-07,
      "loss": 0.8403,
      "step": 904
    },
    {
      "epoch": 0.08946444900279267,
      "grad_norm": 5.716591169589403,
      "learning_rate": 1.981539164312982e-07,
      "loss": 0.7406,
      "step": 905
    },
    {
      "epoch": 0.08956330474754715,
      "grad_norm": 5.552818519185844,
      "learning_rate": 1.9814778699846997e-07,
      "loss": 0.768,
      "step": 906
    },
    {
      "epoch": 0.0896621604923016,
      "grad_norm": 7.003601100450262,
      "learning_rate": 1.9814164750203396e-07,
      "loss": 0.7827,
      "step": 907
    },
    {
      "epoch": 0.08976101623705608,
      "grad_norm": 4.371753709971432,
      "learning_rate": 1.981354979426197e-07,
      "loss": 0.6607,
      "step": 908
    },
    {
      "epoch": 0.08985987198181054,
      "grad_norm": 3.8385239013421795,
      "learning_rate": 1.9812933832085772e-07,
      "loss": 0.8206,
      "step": 909
    },
    {
      "epoch": 0.089958727726565,
      "grad_norm": 7.402589332585354,
      "learning_rate": 1.9812316863737962e-07,
      "loss": 0.7459,
      "step": 910
    },
    {
      "epoch": 0.09005758347131948,
      "grad_norm": 2.707285399914594,
      "learning_rate": 1.9811698889281797e-07,
      "loss": 0.6618,
      "step": 911
    },
    {
      "epoch": 0.09015643921607394,
      "grad_norm": 6.550472888640813,
      "learning_rate": 1.9811079908780646e-07,
      "loss": 0.7571,
      "step": 912
    },
    {
      "epoch": 0.09025529496082842,
      "grad_norm": 7.373598063489139,
      "learning_rate": 1.9810459922297972e-07,
      "loss": 0.7319,
      "step": 913
    },
    {
      "epoch": 0.09035415070558288,
      "grad_norm": 4.341690460005673,
      "learning_rate": 1.9809838929897347e-07,
      "loss": 0.6778,
      "step": 914
    },
    {
      "epoch": 0.09045300645033734,
      "grad_norm": 5.548671858522059,
      "learning_rate": 1.9809216931642446e-07,
      "loss": 0.8277,
      "step": 915
    },
    {
      "epoch": 0.09055186219509181,
      "grad_norm": 4.761726693405643,
      "learning_rate": 1.9808593927597046e-07,
      "loss": 0.6672,
      "step": 916
    },
    {
      "epoch": 0.09065071793984628,
      "grad_norm": 3.916484533329558,
      "learning_rate": 1.9807969917825025e-07,
      "loss": 0.9155,
      "step": 917
    },
    {
      "epoch": 0.09074957368460075,
      "grad_norm": 4.3301571325649775,
      "learning_rate": 1.9807344902390366e-07,
      "loss": 0.7297,
      "step": 918
    },
    {
      "epoch": 0.09084842942935521,
      "grad_norm": 4.672323203137857,
      "learning_rate": 1.9806718881357156e-07,
      "loss": 0.7816,
      "step": 919
    },
    {
      "epoch": 0.09094728517410967,
      "grad_norm": 5.98096678247886,
      "learning_rate": 1.9806091854789583e-07,
      "loss": 0.731,
      "step": 920
    },
    {
      "epoch": 0.09104614091886415,
      "grad_norm": 3.7668782708425805,
      "learning_rate": 1.9805463822751942e-07,
      "loss": 0.6752,
      "step": 921
    },
    {
      "epoch": 0.09114499666361861,
      "grad_norm": 4.016819716439341,
      "learning_rate": 1.9804834785308626e-07,
      "loss": 0.8044,
      "step": 922
    },
    {
      "epoch": 0.09124385240837309,
      "grad_norm": 4.757797269891826,
      "learning_rate": 1.9804204742524133e-07,
      "loss": 0.7248,
      "step": 923
    },
    {
      "epoch": 0.09134270815312755,
      "grad_norm": 4.607458782010897,
      "learning_rate": 1.9803573694463069e-07,
      "loss": 0.6414,
      "step": 924
    },
    {
      "epoch": 0.09144156389788202,
      "grad_norm": 4.081323920089845,
      "learning_rate": 1.9802941641190133e-07,
      "loss": 0.7212,
      "step": 925
    },
    {
      "epoch": 0.09154041964263648,
      "grad_norm": 3.2899023716749776,
      "learning_rate": 1.9802308582770134e-07,
      "loss": 0.8306,
      "step": 926
    },
    {
      "epoch": 0.09163927538739094,
      "grad_norm": 4.8897356054291645,
      "learning_rate": 1.9801674519267987e-07,
      "loss": 0.8189,
      "step": 927
    },
    {
      "epoch": 0.09173813113214542,
      "grad_norm": 7.770108855179524,
      "learning_rate": 1.9801039450748704e-07,
      "loss": 0.754,
      "step": 928
    },
    {
      "epoch": 0.09183698687689988,
      "grad_norm": 5.238870306300288,
      "learning_rate": 1.9800403377277398e-07,
      "loss": 0.7439,
      "step": 929
    },
    {
      "epoch": 0.09193584262165436,
      "grad_norm": 20.473114733729616,
      "learning_rate": 1.9799766298919294e-07,
      "loss": 0.7405,
      "step": 930
    },
    {
      "epoch": 0.09203469836640882,
      "grad_norm": 3.010421290904927,
      "learning_rate": 1.979912821573971e-07,
      "loss": 0.7703,
      "step": 931
    },
    {
      "epoch": 0.09213355411116328,
      "grad_norm": 3.905348440801245,
      "learning_rate": 1.9798489127804077e-07,
      "loss": 0.7894,
      "step": 932
    },
    {
      "epoch": 0.09223240985591775,
      "grad_norm": 4.2854173947543535,
      "learning_rate": 1.979784903517792e-07,
      "loss": 0.7619,
      "step": 933
    },
    {
      "epoch": 0.09233126560067222,
      "grad_norm": 3.6045291253242695,
      "learning_rate": 1.9797207937926873e-07,
      "loss": 0.815,
      "step": 934
    },
    {
      "epoch": 0.09243012134542669,
      "grad_norm": 3.022906754604075,
      "learning_rate": 1.9796565836116671e-07,
      "loss": 0.7989,
      "step": 935
    },
    {
      "epoch": 0.09252897709018115,
      "grad_norm": 5.081533686869524,
      "learning_rate": 1.9795922729813155e-07,
      "loss": 0.7203,
      "step": 936
    },
    {
      "epoch": 0.09262783283493561,
      "grad_norm": 4.827018422233551,
      "learning_rate": 1.9795278619082262e-07,
      "loss": 0.7713,
      "step": 937
    },
    {
      "epoch": 0.09272668857969009,
      "grad_norm": 3.2845832350857243,
      "learning_rate": 1.9794633503990035e-07,
      "loss": 0.8243,
      "step": 938
    },
    {
      "epoch": 0.09282554432444455,
      "grad_norm": 6.175907729606399,
      "learning_rate": 1.9793987384602625e-07,
      "loss": 0.7173,
      "step": 939
    },
    {
      "epoch": 0.09292440006919903,
      "grad_norm": 3.198463070244492,
      "learning_rate": 1.9793340260986277e-07,
      "loss": 0.7715,
      "step": 940
    },
    {
      "epoch": 0.09302325581395349,
      "grad_norm": 4.192722546034901,
      "learning_rate": 1.9792692133207352e-07,
      "loss": 0.6863,
      "step": 941
    },
    {
      "epoch": 0.09312211155870796,
      "grad_norm": 4.108045458234193,
      "learning_rate": 1.97920430013323e-07,
      "loss": 0.7532,
      "step": 942
    },
    {
      "epoch": 0.09322096730346242,
      "grad_norm": 4.304530946797021,
      "learning_rate": 1.9791392865427677e-07,
      "loss": 0.7793,
      "step": 943
    },
    {
      "epoch": 0.09331982304821689,
      "grad_norm": 4.691091631388994,
      "learning_rate": 1.979074172556015e-07,
      "loss": 0.6875,
      "step": 944
    },
    {
      "epoch": 0.09341867879297136,
      "grad_norm": 3.6303162997585563,
      "learning_rate": 1.9790089581796488e-07,
      "loss": 0.6835,
      "step": 945
    },
    {
      "epoch": 0.09351753453772582,
      "grad_norm": 4.353557897553249,
      "learning_rate": 1.9789436434203549e-07,
      "loss": 0.7362,
      "step": 946
    },
    {
      "epoch": 0.0936163902824803,
      "grad_norm": 6.087437903762105,
      "learning_rate": 1.9788782282848308e-07,
      "loss": 0.9038,
      "step": 947
    },
    {
      "epoch": 0.09371524602723476,
      "grad_norm": 4.031163558251467,
      "learning_rate": 1.978812712779784e-07,
      "loss": 0.8409,
      "step": 948
    },
    {
      "epoch": 0.09381410177198922,
      "grad_norm": 5.43922746258757,
      "learning_rate": 1.9787470969119317e-07,
      "loss": 0.7912,
      "step": 949
    },
    {
      "epoch": 0.0939129575167437,
      "grad_norm": 3.621040438012115,
      "learning_rate": 1.9786813806880023e-07,
      "loss": 0.8181,
      "step": 950
    },
    {
      "epoch": 0.09401181326149816,
      "grad_norm": 4.241873340293866,
      "learning_rate": 1.978615564114734e-07,
      "loss": 0.8378,
      "step": 951
    },
    {
      "epoch": 0.09411066900625263,
      "grad_norm": 4.035819500055859,
      "learning_rate": 1.9785496471988755e-07,
      "loss": 0.7106,
      "step": 952
    },
    {
      "epoch": 0.09420952475100709,
      "grad_norm": 20.701732290784598,
      "learning_rate": 1.9784836299471846e-07,
      "loss": 0.7705,
      "step": 953
    },
    {
      "epoch": 0.09430838049576155,
      "grad_norm": 3.874751164266666,
      "learning_rate": 1.9784175123664314e-07,
      "loss": 0.6765,
      "step": 954
    },
    {
      "epoch": 0.09440723624051603,
      "grad_norm": 6.588364848570739,
      "learning_rate": 1.9783512944633953e-07,
      "loss": 0.7609,
      "step": 955
    },
    {
      "epoch": 0.09450609198527049,
      "grad_norm": 7.42533836140291,
      "learning_rate": 1.9782849762448652e-07,
      "loss": 0.6578,
      "step": 956
    },
    {
      "epoch": 0.09460494773002497,
      "grad_norm": 3.5094067020374844,
      "learning_rate": 1.9782185577176418e-07,
      "loss": 0.8069,
      "step": 957
    },
    {
      "epoch": 0.09470380347477943,
      "grad_norm": 4.346961361797709,
      "learning_rate": 1.9781520388885348e-07,
      "loss": 0.6762,
      "step": 958
    },
    {
      "epoch": 0.09480265921953389,
      "grad_norm": 9.331837964571536,
      "learning_rate": 1.978085419764365e-07,
      "loss": 0.7363,
      "step": 959
    },
    {
      "epoch": 0.09490151496428836,
      "grad_norm": 5.8697024580756265,
      "learning_rate": 1.9780187003519635e-07,
      "loss": 0.7576,
      "step": 960
    },
    {
      "epoch": 0.09500037070904283,
      "grad_norm": 3.72559139490969,
      "learning_rate": 1.9779518806581708e-07,
      "loss": 0.8156,
      "step": 961
    },
    {
      "epoch": 0.0950992264537973,
      "grad_norm": 7.395256131794705,
      "learning_rate": 1.9778849606898387e-07,
      "loss": 0.7383,
      "step": 962
    },
    {
      "epoch": 0.09519808219855176,
      "grad_norm": 4.948538900812303,
      "learning_rate": 1.9778179404538285e-07,
      "loss": 0.8069,
      "step": 963
    },
    {
      "epoch": 0.09529693794330624,
      "grad_norm": 17.95977751087989,
      "learning_rate": 1.9777508199570124e-07,
      "loss": 0.6726,
      "step": 964
    },
    {
      "epoch": 0.0953957936880607,
      "grad_norm": 5.094379940164772,
      "learning_rate": 1.9776835992062725e-07,
      "loss": 0.811,
      "step": 965
    },
    {
      "epoch": 0.09549464943281516,
      "grad_norm": 5.623995858805992,
      "learning_rate": 1.9776162782085014e-07,
      "loss": 0.726,
      "step": 966
    },
    {
      "epoch": 0.09559350517756963,
      "grad_norm": 4.590338505016363,
      "learning_rate": 1.977548856970602e-07,
      "loss": 0.662,
      "step": 967
    },
    {
      "epoch": 0.0956923609223241,
      "grad_norm": 6.486950103382593,
      "learning_rate": 1.9774813354994867e-07,
      "loss": 0.7609,
      "step": 968
    },
    {
      "epoch": 0.09579121666707857,
      "grad_norm": 3.589315737214571,
      "learning_rate": 1.97741371380208e-07,
      "loss": 0.8192,
      "step": 969
    },
    {
      "epoch": 0.09589007241183303,
      "grad_norm": 3.102030899765386,
      "learning_rate": 1.9773459918853145e-07,
      "loss": 0.7638,
      "step": 970
    },
    {
      "epoch": 0.0959889281565875,
      "grad_norm": 5.934393260945401,
      "learning_rate": 1.9772781697561343e-07,
      "loss": 0.7565,
      "step": 971
    },
    {
      "epoch": 0.09608778390134197,
      "grad_norm": 5.9666292665219505,
      "learning_rate": 1.9772102474214937e-07,
      "loss": 0.833,
      "step": 972
    },
    {
      "epoch": 0.09618663964609643,
      "grad_norm": 3.357070708740453,
      "learning_rate": 1.977142224888357e-07,
      "loss": 0.8521,
      "step": 973
    },
    {
      "epoch": 0.0962854953908509,
      "grad_norm": 10.075695851945303,
      "learning_rate": 1.9770741021636992e-07,
      "loss": 0.7401,
      "step": 974
    },
    {
      "epoch": 0.09638435113560537,
      "grad_norm": 4.373683497144296,
      "learning_rate": 1.9770058792545048e-07,
      "loss": 0.6434,
      "step": 975
    },
    {
      "epoch": 0.09648320688035983,
      "grad_norm": 3.15898742945996,
      "learning_rate": 1.9769375561677697e-07,
      "loss": 0.7243,
      "step": 976
    },
    {
      "epoch": 0.0965820626251143,
      "grad_norm": 4.372271071293006,
      "learning_rate": 1.9768691329104988e-07,
      "loss": 0.7819,
      "step": 977
    },
    {
      "epoch": 0.09668091836986877,
      "grad_norm": 3.739274707747004,
      "learning_rate": 1.9768006094897083e-07,
      "loss": 0.7167,
      "step": 978
    },
    {
      "epoch": 0.09677977411462324,
      "grad_norm": 23.963350151486566,
      "learning_rate": 1.9767319859124242e-07,
      "loss": 0.7779,
      "step": 979
    },
    {
      "epoch": 0.0968786298593777,
      "grad_norm": 4.450571302225443,
      "learning_rate": 1.9766632621856826e-07,
      "loss": 0.6823,
      "step": 980
    },
    {
      "epoch": 0.09697748560413218,
      "grad_norm": 13.873177812407274,
      "learning_rate": 1.97659443831653e-07,
      "loss": 0.8032,
      "step": 981
    },
    {
      "epoch": 0.09707634134888664,
      "grad_norm": 3.051291589998188,
      "learning_rate": 1.9765255143120239e-07,
      "loss": 0.7391,
      "step": 982
    },
    {
      "epoch": 0.0971751970936411,
      "grad_norm": 4.3085709875783555,
      "learning_rate": 1.976456490179231e-07,
      "loss": 0.7842,
      "step": 983
    },
    {
      "epoch": 0.09727405283839557,
      "grad_norm": 6.3181135205734655,
      "learning_rate": 1.9763873659252285e-07,
      "loss": 0.744,
      "step": 984
    },
    {
      "epoch": 0.09737290858315004,
      "grad_norm": 5.805975665942823,
      "learning_rate": 1.9763181415571046e-07,
      "loss": 0.6678,
      "step": 985
    },
    {
      "epoch": 0.09747176432790451,
      "grad_norm": 5.081865261638517,
      "learning_rate": 1.9762488170819571e-07,
      "loss": 0.6042,
      "step": 986
    },
    {
      "epoch": 0.09757062007265897,
      "grad_norm": 4.844748896886957,
      "learning_rate": 1.976179392506894e-07,
      "loss": 0.6653,
      "step": 987
    },
    {
      "epoch": 0.09766947581741343,
      "grad_norm": 9.529115173620644,
      "learning_rate": 1.976109867839034e-07,
      "loss": 0.8133,
      "step": 988
    },
    {
      "epoch": 0.09776833156216791,
      "grad_norm": 2.8890740282775558,
      "learning_rate": 1.9760402430855055e-07,
      "loss": 0.6894,
      "step": 989
    },
    {
      "epoch": 0.09786718730692237,
      "grad_norm": 4.971949484883432,
      "learning_rate": 1.9759705182534477e-07,
      "loss": 0.7774,
      "step": 990
    },
    {
      "epoch": 0.09796604305167685,
      "grad_norm": 4.667810418562867,
      "learning_rate": 1.9759006933500097e-07,
      "loss": 0.7886,
      "step": 991
    },
    {
      "epoch": 0.09806489879643131,
      "grad_norm": 6.427228375156375,
      "learning_rate": 1.9758307683823515e-07,
      "loss": 0.6729,
      "step": 992
    },
    {
      "epoch": 0.09816375454118577,
      "grad_norm": 3.6509465735128206,
      "learning_rate": 1.9757607433576423e-07,
      "loss": 0.7593,
      "step": 993
    },
    {
      "epoch": 0.09826261028594024,
      "grad_norm": 4.230309612816775,
      "learning_rate": 1.975690618283062e-07,
      "loss": 0.7646,
      "step": 994
    },
    {
      "epoch": 0.0983614660306947,
      "grad_norm": 5.906958817561689,
      "learning_rate": 1.9756203931658017e-07,
      "loss": 0.7371,
      "step": 995
    },
    {
      "epoch": 0.09846032177544918,
      "grad_norm": 3.379544638620219,
      "learning_rate": 1.9755500680130616e-07,
      "loss": 0.7032,
      "step": 996
    },
    {
      "epoch": 0.09855917752020364,
      "grad_norm": 3.4021128793021287,
      "learning_rate": 1.975479642832052e-07,
      "loss": 0.7351,
      "step": 997
    },
    {
      "epoch": 0.0986580332649581,
      "grad_norm": 5.234007072288429,
      "learning_rate": 1.9754091176299945e-07,
      "loss": 0.8133,
      "step": 998
    },
    {
      "epoch": 0.09875688900971258,
      "grad_norm": 6.317727173916338,
      "learning_rate": 1.9753384924141205e-07,
      "loss": 0.7669,
      "step": 999
    },
    {
      "epoch": 0.09885574475446704,
      "grad_norm": 6.791241934411815,
      "learning_rate": 1.975267767191671e-07,
      "loss": 0.7618,
      "step": 1000
    },
    {
      "epoch": 0.09895460049922152,
      "grad_norm": 4.089582814858532,
      "learning_rate": 1.9751969419698986e-07,
      "loss": 0.7951,
      "step": 1001
    },
    {
      "epoch": 0.09905345624397598,
      "grad_norm": 3.8297635481480015,
      "learning_rate": 1.975126016756065e-07,
      "loss": 0.715,
      "step": 1002
    },
    {
      "epoch": 0.09915231198873045,
      "grad_norm": 8.339206921821702,
      "learning_rate": 1.975054991557442e-07,
      "loss": 0.7943,
      "step": 1003
    },
    {
      "epoch": 0.09925116773348491,
      "grad_norm": 10.760578096980979,
      "learning_rate": 1.9749838663813129e-07,
      "loss": 0.7175,
      "step": 1004
    },
    {
      "epoch": 0.09935002347823937,
      "grad_norm": 5.979719600999058,
      "learning_rate": 1.9749126412349704e-07,
      "loss": 0.8145,
      "step": 1005
    },
    {
      "epoch": 0.09944887922299385,
      "grad_norm": 16.13788492016263,
      "learning_rate": 1.9748413161257176e-07,
      "loss": 0.7253,
      "step": 1006
    },
    {
      "epoch": 0.09954773496774831,
      "grad_norm": 3.5307258714803593,
      "learning_rate": 1.9747698910608675e-07,
      "loss": 0.7169,
      "step": 1007
    },
    {
      "epoch": 0.09964659071250279,
      "grad_norm": 4.481765994485093,
      "learning_rate": 1.974698366047744e-07,
      "loss": 0.7776,
      "step": 1008
    },
    {
      "epoch": 0.09974544645725725,
      "grad_norm": 8.083205743494124,
      "learning_rate": 1.9746267410936813e-07,
      "loss": 0.8545,
      "step": 1009
    },
    {
      "epoch": 0.09984430220201171,
      "grad_norm": 3.7992316756178535,
      "learning_rate": 1.9745550162060225e-07,
      "loss": 0.7355,
      "step": 1010
    },
    {
      "epoch": 0.09994315794676618,
      "grad_norm": 4.197308047337933,
      "learning_rate": 1.9744831913921227e-07,
      "loss": 0.7161,
      "step": 1011
    },
    {
      "epoch": 0.10004201369152065,
      "grad_norm": 3.783430405720231,
      "learning_rate": 1.974411266659346e-07,
      "loss": 0.7179,
      "step": 1012
    },
    {
      "epoch": 0.10014086943627512,
      "grad_norm": 3.7409822124565397,
      "learning_rate": 1.974339242015068e-07,
      "loss": 0.7114,
      "step": 1013
    },
    {
      "epoch": 0.10023972518102958,
      "grad_norm": 5.843017495574706,
      "learning_rate": 1.974267117466673e-07,
      "loss": 0.7258,
      "step": 1014
    },
    {
      "epoch": 0.10033858092578404,
      "grad_norm": 5.48415231388352,
      "learning_rate": 1.9741948930215563e-07,
      "loss": 0.7598,
      "step": 1015
    },
    {
      "epoch": 0.10043743667053852,
      "grad_norm": 8.916798361030231,
      "learning_rate": 1.974122568687124e-07,
      "loss": 0.6695,
      "step": 1016
    },
    {
      "epoch": 0.10053629241529298,
      "grad_norm": 4.187343714602581,
      "learning_rate": 1.9740501444707917e-07,
      "loss": 0.8011,
      "step": 1017
    },
    {
      "epoch": 0.10063514816004746,
      "grad_norm": 12.0113550947875,
      "learning_rate": 1.9739776203799852e-07,
      "loss": 0.7311,
      "step": 1018
    },
    {
      "epoch": 0.10073400390480192,
      "grad_norm": 12.237979607480492,
      "learning_rate": 1.973904996422141e-07,
      "loss": 0.7165,
      "step": 1019
    },
    {
      "epoch": 0.10083285964955639,
      "grad_norm": 4.433247551889724,
      "learning_rate": 1.973832272604705e-07,
      "loss": 0.795,
      "step": 1020
    },
    {
      "epoch": 0.10093171539431085,
      "grad_norm": 3.640821147666539,
      "learning_rate": 1.973759448935135e-07,
      "loss": 0.7325,
      "step": 1021
    },
    {
      "epoch": 0.10103057113906531,
      "grad_norm": 4.434911511956639,
      "learning_rate": 1.9736865254208976e-07,
      "loss": 0.7679,
      "step": 1022
    },
    {
      "epoch": 0.10112942688381979,
      "grad_norm": 8.195345823076845,
      "learning_rate": 1.9736135020694697e-07,
      "loss": 0.6896,
      "step": 1023
    },
    {
      "epoch": 0.10122828262857425,
      "grad_norm": 4.14381987341804,
      "learning_rate": 1.973540378888339e-07,
      "loss": 0.7249,
      "step": 1024
    },
    {
      "epoch": 0.10132713837332873,
      "grad_norm": 4.445419156526462,
      "learning_rate": 1.9734671558850028e-07,
      "loss": 0.838,
      "step": 1025
    },
    {
      "epoch": 0.10142599411808319,
      "grad_norm": 3.706931421041756,
      "learning_rate": 1.97339383306697e-07,
      "loss": 0.7931,
      "step": 1026
    },
    {
      "epoch": 0.10152484986283765,
      "grad_norm": 4.328332031171043,
      "learning_rate": 1.973320410441758e-07,
      "loss": 0.8096,
      "step": 1027
    },
    {
      "epoch": 0.10162370560759212,
      "grad_norm": 3.702947727662279,
      "learning_rate": 1.973246888016895e-07,
      "loss": 0.7362,
      "step": 1028
    },
    {
      "epoch": 0.10172256135234659,
      "grad_norm": 5.560187698580335,
      "learning_rate": 1.9731732657999208e-07,
      "loss": 0.7376,
      "step": 1029
    },
    {
      "epoch": 0.10182141709710106,
      "grad_norm": 4.331835544794006,
      "learning_rate": 1.973099543798383e-07,
      "loss": 0.7909,
      "step": 1030
    },
    {
      "epoch": 0.10192027284185552,
      "grad_norm": 4.2138603990037495,
      "learning_rate": 1.973025722019841e-07,
      "loss": 0.7804,
      "step": 1031
    },
    {
      "epoch": 0.10201912858660998,
      "grad_norm": 20.163209344267734,
      "learning_rate": 1.9729518004718647e-07,
      "loss": 0.8203,
      "step": 1032
    },
    {
      "epoch": 0.10211798433136446,
      "grad_norm": 5.862920688811912,
      "learning_rate": 1.9728777791620332e-07,
      "loss": 0.6981,
      "step": 1033
    },
    {
      "epoch": 0.10221684007611892,
      "grad_norm": 4.523560242993291,
      "learning_rate": 1.9728036580979362e-07,
      "loss": 0.7058,
      "step": 1034
    },
    {
      "epoch": 0.1023156958208734,
      "grad_norm": 2.980192314030771,
      "learning_rate": 1.9727294372871745e-07,
      "loss": 0.7241,
      "step": 1035
    },
    {
      "epoch": 0.10241455156562786,
      "grad_norm": 4.764962331364637,
      "learning_rate": 1.9726551167373576e-07,
      "loss": 0.7144,
      "step": 1036
    },
    {
      "epoch": 0.10251340731038232,
      "grad_norm": 27.796379601881302,
      "learning_rate": 1.9725806964561057e-07,
      "loss": 0.8143,
      "step": 1037
    },
    {
      "epoch": 0.1026122630551368,
      "grad_norm": 5.138934897207487,
      "learning_rate": 1.9725061764510502e-07,
      "loss": 0.7758,
      "step": 1038
    },
    {
      "epoch": 0.10271111879989125,
      "grad_norm": 9.310032774161082,
      "learning_rate": 1.972431556729832e-07,
      "loss": 0.7223,
      "step": 1039
    },
    {
      "epoch": 0.10280997454464573,
      "grad_norm": 6.657170197698489,
      "learning_rate": 1.9723568373001018e-07,
      "loss": 0.6629,
      "step": 1040
    },
    {
      "epoch": 0.10290883028940019,
      "grad_norm": 2.7418742863363406,
      "learning_rate": 1.9722820181695212e-07,
      "loss": 0.831,
      "step": 1041
    },
    {
      "epoch": 0.10300768603415467,
      "grad_norm": 2.8046324244086795,
      "learning_rate": 1.972207099345762e-07,
      "loss": 0.6286,
      "step": 1042
    },
    {
      "epoch": 0.10310654177890913,
      "grad_norm": 3.513262408718687,
      "learning_rate": 1.9721320808365056e-07,
      "loss": 0.6932,
      "step": 1043
    },
    {
      "epoch": 0.10320539752366359,
      "grad_norm": 3.1478488611088387,
      "learning_rate": 1.9720569626494442e-07,
      "loss": 0.6618,
      "step": 1044
    },
    {
      "epoch": 0.10330425326841806,
      "grad_norm": 6.71008368007407,
      "learning_rate": 1.9719817447922807e-07,
      "loss": 0.6924,
      "step": 1045
    },
    {
      "epoch": 0.10340310901317253,
      "grad_norm": 4.976275748455934,
      "learning_rate": 1.9719064272727266e-07,
      "loss": 0.7761,
      "step": 1046
    },
    {
      "epoch": 0.103501964757927,
      "grad_norm": 4.835436894878733,
      "learning_rate": 1.9718310100985049e-07,
      "loss": 0.7491,
      "step": 1047
    },
    {
      "epoch": 0.10360082050268146,
      "grad_norm": 5.505738276932341,
      "learning_rate": 1.9717554932773486e-07,
      "loss": 0.7947,
      "step": 1048
    },
    {
      "epoch": 0.10369967624743592,
      "grad_norm": 4.650673059137605,
      "learning_rate": 1.971679876817001e-07,
      "loss": 0.7621,
      "step": 1049
    },
    {
      "epoch": 0.1037985319921904,
      "grad_norm": 9.0132398437828,
      "learning_rate": 1.9716041607252153e-07,
      "loss": 0.6717,
      "step": 1050
    },
    {
      "epoch": 0.10389738773694486,
      "grad_norm": 7.210296106423745,
      "learning_rate": 1.9715283450097552e-07,
      "loss": 0.6264,
      "step": 1051
    },
    {
      "epoch": 0.10399624348169934,
      "grad_norm": 2.8208429835220326,
      "learning_rate": 1.9714524296783942e-07,
      "loss": 0.6776,
      "step": 1052
    },
    {
      "epoch": 0.1040950992264538,
      "grad_norm": 4.965030213933701,
      "learning_rate": 1.9713764147389166e-07,
      "loss": 0.692,
      "step": 1053
    },
    {
      "epoch": 0.10419395497120826,
      "grad_norm": 4.266548644806464,
      "learning_rate": 1.9713003001991168e-07,
      "loss": 0.7079,
      "step": 1054
    },
    {
      "epoch": 0.10429281071596273,
      "grad_norm": 3.3866909579638116,
      "learning_rate": 1.9712240860667984e-07,
      "loss": 0.701,
      "step": 1055
    },
    {
      "epoch": 0.1043916664607172,
      "grad_norm": 3.4095031223740775,
      "learning_rate": 1.9711477723497767e-07,
      "loss": 0.7018,
      "step": 1056
    },
    {
      "epoch": 0.10449052220547167,
      "grad_norm": 4.7434332735427605,
      "learning_rate": 1.9710713590558763e-07,
      "loss": 0.7954,
      "step": 1057
    },
    {
      "epoch": 0.10458937795022613,
      "grad_norm": 4.575346164681996,
      "learning_rate": 1.9709948461929326e-07,
      "loss": 0.6915,
      "step": 1058
    },
    {
      "epoch": 0.1046882336949806,
      "grad_norm": 5.294386608138551,
      "learning_rate": 1.9709182337687902e-07,
      "loss": 0.6664,
      "step": 1059
    },
    {
      "epoch": 0.10478708943973507,
      "grad_norm": 5.264900973190541,
      "learning_rate": 1.9708415217913052e-07,
      "loss": 0.7531,
      "step": 1060
    },
    {
      "epoch": 0.10488594518448953,
      "grad_norm": 7.294654766218854,
      "learning_rate": 1.9707647102683432e-07,
      "loss": 0.7676,
      "step": 1061
    },
    {
      "epoch": 0.104984800929244,
      "grad_norm": 3.372747187081616,
      "learning_rate": 1.9706877992077796e-07,
      "loss": 0.6362,
      "step": 1062
    },
    {
      "epoch": 0.10508365667399847,
      "grad_norm": 3.615211759187662,
      "learning_rate": 1.9706107886175013e-07,
      "loss": 0.7794,
      "step": 1063
    },
    {
      "epoch": 0.10518251241875294,
      "grad_norm": 4.326398793083993,
      "learning_rate": 1.9705336785054037e-07,
      "loss": 0.7747,
      "step": 1064
    },
    {
      "epoch": 0.1052813681635074,
      "grad_norm": 6.538676043835103,
      "learning_rate": 1.970456468879394e-07,
      "loss": 0.7274,
      "step": 1065
    },
    {
      "epoch": 0.10538022390826186,
      "grad_norm": 4.547453850034342,
      "learning_rate": 1.9703791597473884e-07,
      "loss": 0.6926,
      "step": 1066
    },
    {
      "epoch": 0.10547907965301634,
      "grad_norm": 3.1263192275847858,
      "learning_rate": 1.9703017511173144e-07,
      "loss": 0.7459,
      "step": 1067
    },
    {
      "epoch": 0.1055779353977708,
      "grad_norm": 3.881137195441936,
      "learning_rate": 1.9702242429971086e-07,
      "loss": 0.8003,
      "step": 1068
    },
    {
      "epoch": 0.10567679114252528,
      "grad_norm": 5.630079794462404,
      "learning_rate": 1.9701466353947186e-07,
      "loss": 0.6097,
      "step": 1069
    },
    {
      "epoch": 0.10577564688727974,
      "grad_norm": 5.949046138484859,
      "learning_rate": 1.9700689283181017e-07,
      "loss": 0.6775,
      "step": 1070
    },
    {
      "epoch": 0.1058745026320342,
      "grad_norm": 3.503929647219076,
      "learning_rate": 1.9699911217752255e-07,
      "loss": 0.7771,
      "step": 1071
    },
    {
      "epoch": 0.10597335837678867,
      "grad_norm": 5.682027771800958,
      "learning_rate": 1.9699132157740684e-07,
      "loss": 0.7099,
      "step": 1072
    },
    {
      "epoch": 0.10607221412154313,
      "grad_norm": 11.623976142033273,
      "learning_rate": 1.9698352103226187e-07,
      "loss": 0.6727,
      "step": 1073
    },
    {
      "epoch": 0.10617106986629761,
      "grad_norm": 6.787581016457171,
      "learning_rate": 1.9697571054288735e-07,
      "loss": 0.7595,
      "step": 1074
    },
    {
      "epoch": 0.10626992561105207,
      "grad_norm": 7.113529088133563,
      "learning_rate": 1.9696789011008423e-07,
      "loss": 0.8299,
      "step": 1075
    },
    {
      "epoch": 0.10636878135580653,
      "grad_norm": 4.674931834142927,
      "learning_rate": 1.969600597346544e-07,
      "loss": 0.7377,
      "step": 1076
    },
    {
      "epoch": 0.10646763710056101,
      "grad_norm": 26.37932469893689,
      "learning_rate": 1.9695221941740066e-07,
      "loss": 0.6845,
      "step": 1077
    },
    {
      "epoch": 0.10656649284531547,
      "grad_norm": 5.917148411684509,
      "learning_rate": 1.9694436915912695e-07,
      "loss": 0.8323,
      "step": 1078
    },
    {
      "epoch": 0.10666534859006994,
      "grad_norm": 3.29157486205495,
      "learning_rate": 1.9693650896063825e-07,
      "loss": 0.6998,
      "step": 1079
    },
    {
      "epoch": 0.1067642043348244,
      "grad_norm": 4.07341156003961,
      "learning_rate": 1.9692863882274048e-07,
      "loss": 0.8016,
      "step": 1080
    },
    {
      "epoch": 0.10686306007957888,
      "grad_norm": 10.674640491187617,
      "learning_rate": 1.9692075874624054e-07,
      "loss": 0.7617,
      "step": 1081
    },
    {
      "epoch": 0.10696191582433334,
      "grad_norm": 14.411437567503732,
      "learning_rate": 1.9691286873194654e-07,
      "loss": 0.6477,
      "step": 1082
    },
    {
      "epoch": 0.1070607715690878,
      "grad_norm": 3.817854280659467,
      "learning_rate": 1.9690496878066735e-07,
      "loss": 0.8388,
      "step": 1083
    },
    {
      "epoch": 0.10715962731384228,
      "grad_norm": 3.787193130907739,
      "learning_rate": 1.968970588932131e-07,
      "loss": 0.7989,
      "step": 1084
    },
    {
      "epoch": 0.10725848305859674,
      "grad_norm": 7.6332037165015,
      "learning_rate": 1.968891390703948e-07,
      "loss": 0.812,
      "step": 1085
    },
    {
      "epoch": 0.10735733880335122,
      "grad_norm": 6.063861616237909,
      "learning_rate": 1.9688120931302448e-07,
      "loss": 0.7984,
      "step": 1086
    },
    {
      "epoch": 0.10745619454810568,
      "grad_norm": 3.3226248446759685,
      "learning_rate": 1.9687326962191527e-07,
      "loss": 0.6695,
      "step": 1087
    },
    {
      "epoch": 0.10755505029286014,
      "grad_norm": 6.291851977946078,
      "learning_rate": 1.968653199978812e-07,
      "loss": 0.7984,
      "step": 1088
    },
    {
      "epoch": 0.10765390603761461,
      "grad_norm": 5.3831950239290824,
      "learning_rate": 1.9685736044173743e-07,
      "loss": 0.8153,
      "step": 1089
    },
    {
      "epoch": 0.10775276178236907,
      "grad_norm": 9.272983922789042,
      "learning_rate": 1.9684939095430014e-07,
      "loss": 0.7326,
      "step": 1090
    },
    {
      "epoch": 0.10785161752712355,
      "grad_norm": 4.655731506191009,
      "learning_rate": 1.9684141153638644e-07,
      "loss": 0.6039,
      "step": 1091
    },
    {
      "epoch": 0.10795047327187801,
      "grad_norm": 4.828917617121873,
      "learning_rate": 1.9683342218881444e-07,
      "loss": 0.805,
      "step": 1092
    },
    {
      "epoch": 0.10804932901663247,
      "grad_norm": 15.430383165763907,
      "learning_rate": 1.9682542291240342e-07,
      "loss": 0.6797,
      "step": 1093
    },
    {
      "epoch": 0.10814818476138695,
      "grad_norm": 4.571745453317567,
      "learning_rate": 1.9681741370797356e-07,
      "loss": 0.8237,
      "step": 1094
    },
    {
      "epoch": 0.10824704050614141,
      "grad_norm": 15.835594030467476,
      "learning_rate": 1.968093945763461e-07,
      "loss": 0.7219,
      "step": 1095
    },
    {
      "epoch": 0.10834589625089588,
      "grad_norm": 3.33590869730009,
      "learning_rate": 1.9680136551834323e-07,
      "loss": 0.8101,
      "step": 1096
    },
    {
      "epoch": 0.10844475199565035,
      "grad_norm": 3.5105968629894577,
      "learning_rate": 1.9679332653478825e-07,
      "loss": 0.6862,
      "step": 1097
    },
    {
      "epoch": 0.10854360774040482,
      "grad_norm": 2.9814034247680836,
      "learning_rate": 1.9678527762650545e-07,
      "loss": 0.6918,
      "step": 1098
    },
    {
      "epoch": 0.10864246348515928,
      "grad_norm": 4.016729235695684,
      "learning_rate": 1.967772187943201e-07,
      "loss": 0.692,
      "step": 1099
    },
    {
      "epoch": 0.10874131922991374,
      "grad_norm": 4.670344606243124,
      "learning_rate": 1.967691500390585e-07,
      "loss": 0.7523,
      "step": 1100
    },
    {
      "epoch": 0.10884017497466822,
      "grad_norm": 3.128074070781782,
      "learning_rate": 1.9676107136154806e-07,
      "loss": 0.6762,
      "step": 1101
    },
    {
      "epoch": 0.10893903071942268,
      "grad_norm": 3.4313740603166893,
      "learning_rate": 1.967529827626171e-07,
      "loss": 0.6412,
      "step": 1102
    },
    {
      "epoch": 0.10903788646417716,
      "grad_norm": 6.4972151871219985,
      "learning_rate": 1.967448842430949e-07,
      "loss": 0.6447,
      "step": 1103
    },
    {
      "epoch": 0.10913674220893162,
      "grad_norm": 4.54849499935191,
      "learning_rate": 1.9673677580381193e-07,
      "loss": 0.763,
      "step": 1104
    },
    {
      "epoch": 0.10923559795368608,
      "grad_norm": 2.867658574634454,
      "learning_rate": 1.9672865744559956e-07,
      "loss": 0.6693,
      "step": 1105
    },
    {
      "epoch": 0.10933445369844055,
      "grad_norm": 3.6273336915466596,
      "learning_rate": 1.9672052916929025e-07,
      "loss": 0.709,
      "step": 1106
    },
    {
      "epoch": 0.10943330944319501,
      "grad_norm": 5.5670507438997765,
      "learning_rate": 1.967123909757174e-07,
      "loss": 0.7317,
      "step": 1107
    },
    {
      "epoch": 0.10953216518794949,
      "grad_norm": 6.583102806759151,
      "learning_rate": 1.9670424286571544e-07,
      "loss": 0.7847,
      "step": 1108
    },
    {
      "epoch": 0.10963102093270395,
      "grad_norm": 3.8170210838420515,
      "learning_rate": 1.9669608484011985e-07,
      "loss": 0.7841,
      "step": 1109
    },
    {
      "epoch": 0.10972987667745841,
      "grad_norm": 4.934154269714187,
      "learning_rate": 1.9668791689976713e-07,
      "loss": 0.7749,
      "step": 1110
    },
    {
      "epoch": 0.10982873242221289,
      "grad_norm": 4.849229288240025,
      "learning_rate": 1.9667973904549482e-07,
      "loss": 0.6902,
      "step": 1111
    },
    {
      "epoch": 0.10992758816696735,
      "grad_norm": 4.461236967449306,
      "learning_rate": 1.9667155127814136e-07,
      "loss": 0.7947,
      "step": 1112
    },
    {
      "epoch": 0.11002644391172182,
      "grad_norm": 4.053317864905526,
      "learning_rate": 1.9666335359854633e-07,
      "loss": 0.8277,
      "step": 1113
    },
    {
      "epoch": 0.11012529965647629,
      "grad_norm": 7.934024393932383,
      "learning_rate": 1.9665514600755028e-07,
      "loss": 0.8569,
      "step": 1114
    },
    {
      "epoch": 0.11022415540123075,
      "grad_norm": 3.5147136229295883,
      "learning_rate": 1.9664692850599478e-07,
      "loss": 0.7587,
      "step": 1115
    },
    {
      "epoch": 0.11032301114598522,
      "grad_norm": 4.480598341900733,
      "learning_rate": 1.9663870109472242e-07,
      "loss": 0.723,
      "step": 1116
    },
    {
      "epoch": 0.11042186689073968,
      "grad_norm": 2.5140330503222277,
      "learning_rate": 1.9663046377457674e-07,
      "loss": 0.8137,
      "step": 1117
    },
    {
      "epoch": 0.11052072263549416,
      "grad_norm": 5.069882928382326,
      "learning_rate": 1.966222165464024e-07,
      "loss": 0.8157,
      "step": 1118
    },
    {
      "epoch": 0.11061957838024862,
      "grad_norm": 6.268607336288659,
      "learning_rate": 1.9661395941104505e-07,
      "loss": 0.7637,
      "step": 1119
    },
    {
      "epoch": 0.1107184341250031,
      "grad_norm": 3.5806802613882396,
      "learning_rate": 1.966056923693513e-07,
      "loss": 0.7799,
      "step": 1120
    },
    {
      "epoch": 0.11081728986975756,
      "grad_norm": 24.2998326666818,
      "learning_rate": 1.9659741542216886e-07,
      "loss": 0.753,
      "step": 1121
    },
    {
      "epoch": 0.11091614561451202,
      "grad_norm": 4.668212902193974,
      "learning_rate": 1.9658912857034635e-07,
      "loss": 0.7939,
      "step": 1122
    },
    {
      "epoch": 0.1110150013592665,
      "grad_norm": 6.077830709044338,
      "learning_rate": 1.9658083181473352e-07,
      "loss": 0.6345,
      "step": 1123
    },
    {
      "epoch": 0.11111385710402095,
      "grad_norm": 11.538505877466205,
      "learning_rate": 1.9657252515618106e-07,
      "loss": 0.8116,
      "step": 1124
    },
    {
      "epoch": 0.11121271284877543,
      "grad_norm": 3.2752053098944023,
      "learning_rate": 1.9656420859554066e-07,
      "loss": 0.6048,
      "step": 1125
    },
    {
      "epoch": 0.11131156859352989,
      "grad_norm": 26.945483711685743,
      "learning_rate": 1.965558821336651e-07,
      "loss": 0.7621,
      "step": 1126
    },
    {
      "epoch": 0.11141042433828435,
      "grad_norm": 3.923797973036838,
      "learning_rate": 1.9654754577140816e-07,
      "loss": 0.7232,
      "step": 1127
    },
    {
      "epoch": 0.11150928008303883,
      "grad_norm": 4.702446306488891,
      "learning_rate": 1.9653919950962454e-07,
      "loss": 0.5601,
      "step": 1128
    },
    {
      "epoch": 0.11160813582779329,
      "grad_norm": 4.083872239507477,
      "learning_rate": 1.965308433491701e-07,
      "loss": 0.6972,
      "step": 1129
    },
    {
      "epoch": 0.11170699157254776,
      "grad_norm": 3.9862494110315883,
      "learning_rate": 1.9652247729090156e-07,
      "loss": 0.7132,
      "step": 1130
    },
    {
      "epoch": 0.11180584731730223,
      "grad_norm": 6.173695704281635,
      "learning_rate": 1.965141013356768e-07,
      "loss": 0.8349,
      "step": 1131
    },
    {
      "epoch": 0.11190470306205669,
      "grad_norm": 12.834819820350553,
      "learning_rate": 1.9650571548435464e-07,
      "loss": 0.729,
      "step": 1132
    },
    {
      "epoch": 0.11200355880681116,
      "grad_norm": 3.7334168286670217,
      "learning_rate": 1.964973197377949e-07,
      "loss": 0.7109,
      "step": 1133
    },
    {
      "epoch": 0.11210241455156562,
      "grad_norm": 3.3256479558716654,
      "learning_rate": 1.9648891409685847e-07,
      "loss": 0.7381,
      "step": 1134
    },
    {
      "epoch": 0.1122012702963201,
      "grad_norm": 3.6595892573805955,
      "learning_rate": 1.964804985624072e-07,
      "loss": 0.5925,
      "step": 1135
    },
    {
      "epoch": 0.11230012604107456,
      "grad_norm": 6.761103959706202,
      "learning_rate": 1.96472073135304e-07,
      "loss": 0.7941,
      "step": 1136
    },
    {
      "epoch": 0.11239898178582904,
      "grad_norm": 4.059086064032219,
      "learning_rate": 1.9646363781641274e-07,
      "loss": 0.7364,
      "step": 1137
    },
    {
      "epoch": 0.1124978375305835,
      "grad_norm": 5.0497379196121726,
      "learning_rate": 1.9645519260659839e-07,
      "loss": 0.7695,
      "step": 1138
    },
    {
      "epoch": 0.11259669327533796,
      "grad_norm": 4.089581892064416,
      "learning_rate": 1.9644673750672686e-07,
      "loss": 0.7558,
      "step": 1139
    },
    {
      "epoch": 0.11269554902009243,
      "grad_norm": 8.131012161746074,
      "learning_rate": 1.9643827251766507e-07,
      "loss": 0.7793,
      "step": 1140
    },
    {
      "epoch": 0.1127944047648469,
      "grad_norm": 5.888324807609808,
      "learning_rate": 1.96429797640281e-07,
      "loss": 0.6362,
      "step": 1141
    },
    {
      "epoch": 0.11289326050960137,
      "grad_norm": 2.552170674412662,
      "learning_rate": 1.9642131287544363e-07,
      "loss": 0.6943,
      "step": 1142
    },
    {
      "epoch": 0.11299211625435583,
      "grad_norm": 6.760296543329767,
      "learning_rate": 1.964128182240229e-07,
      "loss": 0.732,
      "step": 1143
    },
    {
      "epoch": 0.11309097199911029,
      "grad_norm": 6.392655919170905,
      "learning_rate": 1.964043136868899e-07,
      "loss": 0.6789,
      "step": 1144
    },
    {
      "epoch": 0.11318982774386477,
      "grad_norm": 5.2410735783499165,
      "learning_rate": 1.9639579926491656e-07,
      "loss": 0.7987,
      "step": 1145
    },
    {
      "epoch": 0.11328868348861923,
      "grad_norm": 10.23231614927342,
      "learning_rate": 1.9638727495897599e-07,
      "loss": 0.8954,
      "step": 1146
    },
    {
      "epoch": 0.1133875392333737,
      "grad_norm": 3.2265123050826405,
      "learning_rate": 1.9637874076994215e-07,
      "loss": 0.7466,
      "step": 1147
    },
    {
      "epoch": 0.11348639497812817,
      "grad_norm": 5.351436761266222,
      "learning_rate": 1.9637019669869014e-07,
      "loss": 0.851,
      "step": 1148
    },
    {
      "epoch": 0.11358525072288263,
      "grad_norm": 3.2548096976373686,
      "learning_rate": 1.96361642746096e-07,
      "loss": 0.7805,
      "step": 1149
    },
    {
      "epoch": 0.1136841064676371,
      "grad_norm": 6.349500022495833,
      "learning_rate": 1.9635307891303685e-07,
      "loss": 0.6832,
      "step": 1150
    },
    {
      "epoch": 0.11378296221239156,
      "grad_norm": 3.3732504099341245,
      "learning_rate": 1.9634450520039074e-07,
      "loss": 0.8051,
      "step": 1151
    },
    {
      "epoch": 0.11388181795714604,
      "grad_norm": 8.431809111718307,
      "learning_rate": 1.9633592160903683e-07,
      "loss": 0.7565,
      "step": 1152
    },
    {
      "epoch": 0.1139806737019005,
      "grad_norm": 4.046361396728659,
      "learning_rate": 1.963273281398552e-07,
      "loss": 0.7323,
      "step": 1153
    },
    {
      "epoch": 0.11407952944665496,
      "grad_norm": 4.763758884306997,
      "learning_rate": 1.9631872479372702e-07,
      "loss": 0.8714,
      "step": 1154
    },
    {
      "epoch": 0.11417838519140944,
      "grad_norm": 4.297751085871059,
      "learning_rate": 1.963101115715344e-07,
      "loss": 0.7141,
      "step": 1155
    },
    {
      "epoch": 0.1142772409361639,
      "grad_norm": 4.020862265663989,
      "learning_rate": 1.963014884741605e-07,
      "loss": 0.781,
      "step": 1156
    },
    {
      "epoch": 0.11437609668091837,
      "grad_norm": 3.832556709664541,
      "learning_rate": 1.962928555024895e-07,
      "loss": 0.6506,
      "step": 1157
    },
    {
      "epoch": 0.11447495242567284,
      "grad_norm": 4.307343782368354,
      "learning_rate": 1.962842126574066e-07,
      "loss": 0.6564,
      "step": 1158
    },
    {
      "epoch": 0.11457380817042731,
      "grad_norm": 5.2540407056750675,
      "learning_rate": 1.96275559939798e-07,
      "loss": 0.7478,
      "step": 1159
    },
    {
      "epoch": 0.11467266391518177,
      "grad_norm": 15.500103201837033,
      "learning_rate": 1.9626689735055088e-07,
      "loss": 0.7886,
      "step": 1160
    },
    {
      "epoch": 0.11477151965993623,
      "grad_norm": 4.7581293093390356,
      "learning_rate": 1.9625822489055345e-07,
      "loss": 0.8076,
      "step": 1161
    },
    {
      "epoch": 0.11487037540469071,
      "grad_norm": 6.685121039232671,
      "learning_rate": 1.96249542560695e-07,
      "loss": 0.7979,
      "step": 1162
    },
    {
      "epoch": 0.11496923114944517,
      "grad_norm": 26.34103516111553,
      "learning_rate": 1.9624085036186574e-07,
      "loss": 0.7069,
      "step": 1163
    },
    {
      "epoch": 0.11506808689419964,
      "grad_norm": 5.135120025828368,
      "learning_rate": 1.9623214829495692e-07,
      "loss": 0.7518,
      "step": 1164
    },
    {
      "epoch": 0.1151669426389541,
      "grad_norm": 3.823081028889439,
      "learning_rate": 1.962234363608608e-07,
      "loss": 0.7241,
      "step": 1165
    },
    {
      "epoch": 0.11526579838370857,
      "grad_norm": 5.725249390062637,
      "learning_rate": 1.962147145604707e-07,
      "loss": 0.8034,
      "step": 1166
    },
    {
      "epoch": 0.11536465412846304,
      "grad_norm": 4.489484866435147,
      "learning_rate": 1.962059828946809e-07,
      "loss": 0.8066,
      "step": 1167
    },
    {
      "epoch": 0.1154635098732175,
      "grad_norm": 5.279773803730768,
      "learning_rate": 1.9619724136438663e-07,
      "loss": 0.7972,
      "step": 1168
    },
    {
      "epoch": 0.11556236561797198,
      "grad_norm": 7.9119799787410345,
      "learning_rate": 1.9618848997048433e-07,
      "loss": 0.8289,
      "step": 1169
    },
    {
      "epoch": 0.11566122136272644,
      "grad_norm": 5.289202047687702,
      "learning_rate": 1.9617972871387126e-07,
      "loss": 0.6869,
      "step": 1170
    },
    {
      "epoch": 0.1157600771074809,
      "grad_norm": 3.2877272079389375,
      "learning_rate": 1.9617095759544573e-07,
      "loss": 0.7334,
      "step": 1171
    },
    {
      "epoch": 0.11585893285223538,
      "grad_norm": 4.490103626492425,
      "learning_rate": 1.9616217661610713e-07,
      "loss": 0.8483,
      "step": 1172
    },
    {
      "epoch": 0.11595778859698984,
      "grad_norm": 4.656653462676062,
      "learning_rate": 1.961533857767558e-07,
      "loss": 0.6926,
      "step": 1173
    },
    {
      "epoch": 0.11605664434174431,
      "grad_norm": 8.61064428661033,
      "learning_rate": 1.9614458507829315e-07,
      "loss": 0.7868,
      "step": 1174
    },
    {
      "epoch": 0.11615550008649878,
      "grad_norm": 7.2963973902716734,
      "learning_rate": 1.961357745216215e-07,
      "loss": 0.7808,
      "step": 1175
    },
    {
      "epoch": 0.11625435583125325,
      "grad_norm": 5.755811897626709,
      "learning_rate": 1.961269541076443e-07,
      "loss": 0.6962,
      "step": 1176
    },
    {
      "epoch": 0.11635321157600771,
      "grad_norm": 7.9948885433264225,
      "learning_rate": 1.9611812383726592e-07,
      "loss": 0.8393,
      "step": 1177
    },
    {
      "epoch": 0.11645206732076217,
      "grad_norm": 12.525050946062107,
      "learning_rate": 1.9610928371139178e-07,
      "loss": 0.5462,
      "step": 1178
    },
    {
      "epoch": 0.11655092306551665,
      "grad_norm": 3.789658205770461,
      "learning_rate": 1.961004337309283e-07,
      "loss": 0.6946,
      "step": 1179
    },
    {
      "epoch": 0.11664977881027111,
      "grad_norm": 8.036777000989021,
      "learning_rate": 1.9609157389678294e-07,
      "loss": 0.7911,
      "step": 1180
    },
    {
      "epoch": 0.11674863455502558,
      "grad_norm": 4.109307745292568,
      "learning_rate": 1.960827042098641e-07,
      "loss": 0.7662,
      "step": 1181
    },
    {
      "epoch": 0.11684749029978005,
      "grad_norm": 3.4578480041785746,
      "learning_rate": 1.9607382467108132e-07,
      "loss": 0.6511,
      "step": 1182
    },
    {
      "epoch": 0.11694634604453451,
      "grad_norm": 5.209534382648805,
      "learning_rate": 1.9606493528134496e-07,
      "loss": 0.62,
      "step": 1183
    },
    {
      "epoch": 0.11704520178928898,
      "grad_norm": 4.572993872537506,
      "learning_rate": 1.9605603604156652e-07,
      "loss": 0.6923,
      "step": 1184
    },
    {
      "epoch": 0.11714405753404344,
      "grad_norm": 4.026154171424179,
      "learning_rate": 1.9604712695265858e-07,
      "loss": 0.7478,
      "step": 1185
    },
    {
      "epoch": 0.11724291327879792,
      "grad_norm": 5.319915188704291,
      "learning_rate": 1.9603820801553452e-07,
      "loss": 0.771,
      "step": 1186
    },
    {
      "epoch": 0.11734176902355238,
      "grad_norm": 11.436323995444692,
      "learning_rate": 1.960292792311089e-07,
      "loss": 0.8073,
      "step": 1187
    },
    {
      "epoch": 0.11744062476830684,
      "grad_norm": 6.571150718533046,
      "learning_rate": 1.9602034060029725e-07,
      "loss": 0.6192,
      "step": 1188
    },
    {
      "epoch": 0.11753948051306132,
      "grad_norm": 31.254039055780368,
      "learning_rate": 1.9601139212401604e-07,
      "loss": 0.8308,
      "step": 1189
    },
    {
      "epoch": 0.11763833625781578,
      "grad_norm": 3.143021111290879,
      "learning_rate": 1.9600243380318284e-07,
      "loss": 0.7133,
      "step": 1190
    },
    {
      "epoch": 0.11773719200257025,
      "grad_norm": 4.466601954082947,
      "learning_rate": 1.959934656387162e-07,
      "loss": 0.7804,
      "step": 1191
    },
    {
      "epoch": 0.11783604774732472,
      "grad_norm": 3.907789020401191,
      "learning_rate": 1.9598448763153565e-07,
      "loss": 0.7534,
      "step": 1192
    },
    {
      "epoch": 0.11793490349207918,
      "grad_norm": 5.873474795959918,
      "learning_rate": 1.959754997825618e-07,
      "loss": 0.7646,
      "step": 1193
    },
    {
      "epoch": 0.11803375923683365,
      "grad_norm": 5.639852245172317,
      "learning_rate": 1.959665020927161e-07,
      "loss": 0.7175,
      "step": 1194
    },
    {
      "epoch": 0.11813261498158811,
      "grad_norm": 7.206126669642875,
      "learning_rate": 1.959574945629213e-07,
      "loss": 0.7889,
      "step": 1195
    },
    {
      "epoch": 0.11823147072634259,
      "grad_norm": 8.967908879496926,
      "learning_rate": 1.959484771941009e-07,
      "loss": 0.7406,
      "step": 1196
    },
    {
      "epoch": 0.11833032647109705,
      "grad_norm": 11.540403948175356,
      "learning_rate": 1.9593944998717945e-07,
      "loss": 0.6326,
      "step": 1197
    },
    {
      "epoch": 0.11842918221585153,
      "grad_norm": 3.505811829692756,
      "learning_rate": 1.9593041294308264e-07,
      "loss": 0.674,
      "step": 1198
    },
    {
      "epoch": 0.11852803796060599,
      "grad_norm": 8.749859277628525,
      "learning_rate": 1.9592136606273706e-07,
      "loss": 0.8097,
      "step": 1199
    },
    {
      "epoch": 0.11862689370536045,
      "grad_norm": 3.882724256173352,
      "learning_rate": 1.9591230934707032e-07,
      "loss": 0.7141,
      "step": 1200
    },
    {
      "epoch": 0.11872574945011492,
      "grad_norm": 3.5060938007521707,
      "learning_rate": 1.9590324279701108e-07,
      "loss": 0.7364,
      "step": 1201
    },
    {
      "epoch": 0.11882460519486938,
      "grad_norm": 3.605655663471504,
      "learning_rate": 1.9589416641348894e-07,
      "loss": 0.7474,
      "step": 1202
    },
    {
      "epoch": 0.11892346093962386,
      "grad_norm": 6.681771021555283,
      "learning_rate": 1.958850801974346e-07,
      "loss": 0.7417,
      "step": 1203
    },
    {
      "epoch": 0.11902231668437832,
      "grad_norm": 9.673119432920508,
      "learning_rate": 1.958759841497797e-07,
      "loss": 0.7792,
      "step": 1204
    },
    {
      "epoch": 0.11912117242913278,
      "grad_norm": 3.3188697785959342,
      "learning_rate": 1.9586687827145685e-07,
      "loss": 0.6227,
      "step": 1205
    },
    {
      "epoch": 0.11922002817388726,
      "grad_norm": 5.3966291114283615,
      "learning_rate": 1.958577625633998e-07,
      "loss": 0.6442,
      "step": 1206
    },
    {
      "epoch": 0.11931888391864172,
      "grad_norm": 6.217554867578828,
      "learning_rate": 1.9584863702654318e-07,
      "loss": 0.8291,
      "step": 1207
    },
    {
      "epoch": 0.1194177396633962,
      "grad_norm": 13.672791203285747,
      "learning_rate": 1.958395016618227e-07,
      "loss": 0.7068,
      "step": 1208
    },
    {
      "epoch": 0.11951659540815066,
      "grad_norm": 5.09596439942259,
      "learning_rate": 1.9583035647017507e-07,
      "loss": 0.6982,
      "step": 1209
    },
    {
      "epoch": 0.11961545115290512,
      "grad_norm": 4.3332757811370755,
      "learning_rate": 1.9582120145253797e-07,
      "loss": 0.9263,
      "step": 1210
    },
    {
      "epoch": 0.11971430689765959,
      "grad_norm": 5.825728404093789,
      "learning_rate": 1.9581203660985012e-07,
      "loss": 0.7653,
      "step": 1211
    },
    {
      "epoch": 0.11981316264241405,
      "grad_norm": 4.000236758457947,
      "learning_rate": 1.9580286194305127e-07,
      "loss": 0.8053,
      "step": 1212
    },
    {
      "epoch": 0.11991201838716853,
      "grad_norm": 5.650188215307143,
      "learning_rate": 1.9579367745308214e-07,
      "loss": 0.6577,
      "step": 1213
    },
    {
      "epoch": 0.12001087413192299,
      "grad_norm": 5.737539255059414,
      "learning_rate": 1.9578448314088438e-07,
      "loss": 0.7967,
      "step": 1214
    },
    {
      "epoch": 0.12010972987667747,
      "grad_norm": 4.000045582858667,
      "learning_rate": 1.9577527900740082e-07,
      "loss": 0.7507,
      "step": 1215
    },
    {
      "epoch": 0.12020858562143193,
      "grad_norm": 4.333244652692537,
      "learning_rate": 1.957660650535752e-07,
      "loss": 0.7736,
      "step": 1216
    },
    {
      "epoch": 0.12030744136618639,
      "grad_norm": 18.244089493775345,
      "learning_rate": 1.957568412803522e-07,
      "loss": 0.7372,
      "step": 1217
    },
    {
      "epoch": 0.12040629711094086,
      "grad_norm": 4.9138681935681054,
      "learning_rate": 1.9574760768867772e-07,
      "loss": 0.783,
      "step": 1218
    },
    {
      "epoch": 0.12050515285569532,
      "grad_norm": 5.601676307151677,
      "learning_rate": 1.957383642794984e-07,
      "loss": 0.7929,
      "step": 1219
    },
    {
      "epoch": 0.1206040086004498,
      "grad_norm": 12.00338297048234,
      "learning_rate": 1.9572911105376208e-07,
      "loss": 0.6742,
      "step": 1220
    },
    {
      "epoch": 0.12070286434520426,
      "grad_norm": 3.6006908146041647,
      "learning_rate": 1.957198480124175e-07,
      "loss": 0.7406,
      "step": 1221
    },
    {
      "epoch": 0.12080172008995872,
      "grad_norm": 7.016600601553458,
      "learning_rate": 1.957105751564145e-07,
      "loss": 0.6779,
      "step": 1222
    },
    {
      "epoch": 0.1209005758347132,
      "grad_norm": 8.445968365442841,
      "learning_rate": 1.9570129248670385e-07,
      "loss": 0.7915,
      "step": 1223
    },
    {
      "epoch": 0.12099943157946766,
      "grad_norm": 7.743229078307025,
      "learning_rate": 1.9569200000423734e-07,
      "loss": 0.7078,
      "step": 1224
    },
    {
      "epoch": 0.12109828732422213,
      "grad_norm": 5.186175670859474,
      "learning_rate": 1.956826977099678e-07,
      "loss": 0.7497,
      "step": 1225
    },
    {
      "epoch": 0.1211971430689766,
      "grad_norm": 5.832923622225965,
      "learning_rate": 1.95673385604849e-07,
      "loss": 0.7432,
      "step": 1226
    },
    {
      "epoch": 0.12129599881373106,
      "grad_norm": 4.30809855952609,
      "learning_rate": 1.956640636898358e-07,
      "loss": 0.6764,
      "step": 1227
    },
    {
      "epoch": 0.12139485455848553,
      "grad_norm": 4.05714902381824,
      "learning_rate": 1.9565473196588402e-07,
      "loss": 0.766,
      "step": 1228
    },
    {
      "epoch": 0.12149371030324,
      "grad_norm": 16.056362468266308,
      "learning_rate": 1.9564539043395048e-07,
      "loss": 0.6746,
      "step": 1229
    },
    {
      "epoch": 0.12159256604799447,
      "grad_norm": 7.451086901454275,
      "learning_rate": 1.95636039094993e-07,
      "loss": 0.7156,
      "step": 1230
    },
    {
      "epoch": 0.12169142179274893,
      "grad_norm": 2.889269086294448,
      "learning_rate": 1.9562667794997045e-07,
      "loss": 0.6457,
      "step": 1231
    },
    {
      "epoch": 0.12179027753750339,
      "grad_norm": 10.493801474266144,
      "learning_rate": 1.956173069998427e-07,
      "loss": 0.754,
      "step": 1232
    },
    {
      "epoch": 0.12188913328225787,
      "grad_norm": 17.031949300848748,
      "learning_rate": 1.9560792624557053e-07,
      "loss": 0.6569,
      "step": 1233
    },
    {
      "epoch": 0.12198798902701233,
      "grad_norm": 43.709640459001356,
      "learning_rate": 1.955985356881159e-07,
      "loss": 0.6453,
      "step": 1234
    },
    {
      "epoch": 0.1220868447717668,
      "grad_norm": 3.901935284874946,
      "learning_rate": 1.9558913532844156e-07,
      "loss": 0.7836,
      "step": 1235
    },
    {
      "epoch": 0.12218570051652126,
      "grad_norm": 4.378586997239479,
      "learning_rate": 1.9557972516751142e-07,
      "loss": 0.7568,
      "step": 1236
    },
    {
      "epoch": 0.12228455626127574,
      "grad_norm": 4.803563065932561,
      "learning_rate": 1.955703052062904e-07,
      "loss": 0.7519,
      "step": 1237
    },
    {
      "epoch": 0.1223834120060302,
      "grad_norm": 4.247082092429813,
      "learning_rate": 1.9556087544574432e-07,
      "loss": 0.7779,
      "step": 1238
    },
    {
      "epoch": 0.12248226775078466,
      "grad_norm": 4.317158197230976,
      "learning_rate": 1.955514358868401e-07,
      "loss": 0.7735,
      "step": 1239
    },
    {
      "epoch": 0.12258112349553914,
      "grad_norm": 4.209967533409742,
      "learning_rate": 1.9554198653054562e-07,
      "loss": 0.8249,
      "step": 1240
    },
    {
      "epoch": 0.1226799792402936,
      "grad_norm": 6.411781425356489,
      "learning_rate": 1.9553252737782975e-07,
      "loss": 0.8235,
      "step": 1241
    },
    {
      "epoch": 0.12277883498504807,
      "grad_norm": 4.056401182675957,
      "learning_rate": 1.9552305842966238e-07,
      "loss": 0.6938,
      "step": 1242
    },
    {
      "epoch": 0.12287769072980254,
      "grad_norm": 4.901489205430493,
      "learning_rate": 1.9551357968701447e-07,
      "loss": 0.7586,
      "step": 1243
    },
    {
      "epoch": 0.122976546474557,
      "grad_norm": 4.929723099586568,
      "learning_rate": 1.9550409115085785e-07,
      "loss": 0.81,
      "step": 1244
    },
    {
      "epoch": 0.12307540221931147,
      "grad_norm": 4.387900679367985,
      "learning_rate": 1.954945928221655e-07,
      "loss": 0.8074,
      "step": 1245
    },
    {
      "epoch": 0.12317425796406593,
      "grad_norm": 4.701965355419206,
      "learning_rate": 1.954850847019113e-07,
      "loss": 0.6603,
      "step": 1246
    },
    {
      "epoch": 0.12327311370882041,
      "grad_norm": 4.446984597354013,
      "learning_rate": 1.9547556679107018e-07,
      "loss": 0.7001,
      "step": 1247
    },
    {
      "epoch": 0.12337196945357487,
      "grad_norm": 3.6657419025938442,
      "learning_rate": 1.9546603909061801e-07,
      "loss": 0.766,
      "step": 1248
    },
    {
      "epoch": 0.12347082519832933,
      "grad_norm": 4.166094618692891,
      "learning_rate": 1.9545650160153177e-07,
      "loss": 0.7549,
      "step": 1249
    },
    {
      "epoch": 0.1235696809430838,
      "grad_norm": 4.042033462675203,
      "learning_rate": 1.9544695432478937e-07,
      "loss": 0.7188,
      "step": 1250
    },
    {
      "epoch": 0.12366853668783827,
      "grad_norm": 7.930801012743697,
      "learning_rate": 1.9543739726136977e-07,
      "loss": 0.7216,
      "step": 1251
    },
    {
      "epoch": 0.12376739243259274,
      "grad_norm": 3.920476062203573,
      "learning_rate": 1.9542783041225285e-07,
      "loss": 0.6874,
      "step": 1252
    },
    {
      "epoch": 0.1238662481773472,
      "grad_norm": 3.5741685138491532,
      "learning_rate": 1.954182537784196e-07,
      "loss": 0.6991,
      "step": 1253
    },
    {
      "epoch": 0.12396510392210168,
      "grad_norm": 7.031236541691503,
      "learning_rate": 1.9540866736085193e-07,
      "loss": 0.7818,
      "step": 1254
    },
    {
      "epoch": 0.12406395966685614,
      "grad_norm": 5.1624735502316135,
      "learning_rate": 1.9539907116053282e-07,
      "loss": 0.6713,
      "step": 1255
    },
    {
      "epoch": 0.1241628154116106,
      "grad_norm": 3.528913892861073,
      "learning_rate": 1.953894651784462e-07,
      "loss": 0.7318,
      "step": 1256
    },
    {
      "epoch": 0.12426167115636508,
      "grad_norm": 4.1709411560180225,
      "learning_rate": 1.95379849415577e-07,
      "loss": 0.8074,
      "step": 1257
    },
    {
      "epoch": 0.12436052690111954,
      "grad_norm": 4.780207591888047,
      "learning_rate": 1.9537022387291118e-07,
      "loss": 0.7489,
      "step": 1258
    },
    {
      "epoch": 0.12445938264587401,
      "grad_norm": 3.6478405339856987,
      "learning_rate": 1.9536058855143575e-07,
      "loss": 0.6573,
      "step": 1259
    },
    {
      "epoch": 0.12455823839062848,
      "grad_norm": 7.2511622322238365,
      "learning_rate": 1.9535094345213858e-07,
      "loss": 0.7967,
      "step": 1260
    },
    {
      "epoch": 0.12465709413538294,
      "grad_norm": 4.827253777819403,
      "learning_rate": 1.953412885760087e-07,
      "loss": 0.6464,
      "step": 1261
    },
    {
      "epoch": 0.12475594988013741,
      "grad_norm": 5.420568613399631,
      "learning_rate": 1.953316239240361e-07,
      "loss": 0.7883,
      "step": 1262
    },
    {
      "epoch": 0.12485480562489187,
      "grad_norm": 5.02760004568445,
      "learning_rate": 1.9532194949721168e-07,
      "loss": 0.8018,
      "step": 1263
    },
    {
      "epoch": 0.12495366136964635,
      "grad_norm": 8.287705619050639,
      "learning_rate": 1.9531226529652742e-07,
      "loss": 0.6981,
      "step": 1264
    },
    {
      "epoch": 0.12505251711440082,
      "grad_norm": 7.298016156977976,
      "learning_rate": 1.9530257132297633e-07,
      "loss": 0.7505,
      "step": 1265
    },
    {
      "epoch": 0.12515137285915529,
      "grad_norm": 7.29132583981489,
      "learning_rate": 1.952928675775524e-07,
      "loss": 0.6865,
      "step": 1266
    },
    {
      "epoch": 0.12525022860390975,
      "grad_norm": 6.413961032623333,
      "learning_rate": 1.952831540612505e-07,
      "loss": 0.7335,
      "step": 1267
    },
    {
      "epoch": 0.1253490843486642,
      "grad_norm": 3.8741187117853695,
      "learning_rate": 1.952734307750667e-07,
      "loss": 0.7474,
      "step": 1268
    },
    {
      "epoch": 0.12544794009341867,
      "grad_norm": 8.743839635210694,
      "learning_rate": 1.9526369771999794e-07,
      "loss": 0.7789,
      "step": 1269
    },
    {
      "epoch": 0.12554679583817316,
      "grad_norm": 3.448467471337922,
      "learning_rate": 1.9525395489704225e-07,
      "loss": 0.8067,
      "step": 1270
    },
    {
      "epoch": 0.12564565158292762,
      "grad_norm": 3.3729542827377608,
      "learning_rate": 1.952442023071986e-07,
      "loss": 0.6878,
      "step": 1271
    },
    {
      "epoch": 0.12574450732768208,
      "grad_norm": 9.28311289390572,
      "learning_rate": 1.952344399514669e-07,
      "loss": 0.747,
      "step": 1272
    },
    {
      "epoch": 0.12584336307243654,
      "grad_norm": 4.224489912819047,
      "learning_rate": 1.9522466783084822e-07,
      "loss": 0.7407,
      "step": 1273
    },
    {
      "epoch": 0.125942218817191,
      "grad_norm": 7.233095731757951,
      "learning_rate": 1.952148859463445e-07,
      "loss": 0.772,
      "step": 1274
    },
    {
      "epoch": 0.1260410745619455,
      "grad_norm": 7.444100979964925,
      "learning_rate": 1.9520509429895875e-07,
      "loss": 0.6453,
      "step": 1275
    },
    {
      "epoch": 0.12613993030669995,
      "grad_norm": 4.580101184426206,
      "learning_rate": 1.9519529288969497e-07,
      "loss": 0.7189,
      "step": 1276
    },
    {
      "epoch": 0.12623878605145442,
      "grad_norm": 4.401794505660196,
      "learning_rate": 1.9518548171955812e-07,
      "loss": 0.7429,
      "step": 1277
    },
    {
      "epoch": 0.12633764179620888,
      "grad_norm": 4.358764896796247,
      "learning_rate": 1.951756607895542e-07,
      "loss": 0.7925,
      "step": 1278
    },
    {
      "epoch": 0.12643649754096334,
      "grad_norm": 13.740068262295237,
      "learning_rate": 1.951658301006902e-07,
      "loss": 0.7283,
      "step": 1279
    },
    {
      "epoch": 0.12653535328571783,
      "grad_norm": 8.23562103457597,
      "learning_rate": 1.9515598965397414e-07,
      "loss": 0.7603,
      "step": 1280
    },
    {
      "epoch": 0.1266342090304723,
      "grad_norm": 3.542544199280855,
      "learning_rate": 1.9514613945041498e-07,
      "loss": 0.7438,
      "step": 1281
    },
    {
      "epoch": 0.12673306477522675,
      "grad_norm": 9.481413761723882,
      "learning_rate": 1.9513627949102274e-07,
      "loss": 0.7024,
      "step": 1282
    },
    {
      "epoch": 0.1268319205199812,
      "grad_norm": 3.7241120862635064,
      "learning_rate": 1.9512640977680838e-07,
      "loss": 0.8008,
      "step": 1283
    },
    {
      "epoch": 0.12693077626473567,
      "grad_norm": 4.568118807965242,
      "learning_rate": 1.951165303087839e-07,
      "loss": 0.7146,
      "step": 1284
    },
    {
      "epoch": 0.12702963200949016,
      "grad_norm": 7.024757136288451,
      "learning_rate": 1.951066410879623e-07,
      "loss": 0.7448,
      "step": 1285
    },
    {
      "epoch": 0.12712848775424462,
      "grad_norm": 12.191197392329478,
      "learning_rate": 1.9509674211535764e-07,
      "loss": 0.738,
      "step": 1286
    },
    {
      "epoch": 0.12722734349899908,
      "grad_norm": 7.947132890685516,
      "learning_rate": 1.9508683339198478e-07,
      "loss": 0.6649,
      "step": 1287
    },
    {
      "epoch": 0.12732619924375355,
      "grad_norm": 7.606792519329367,
      "learning_rate": 1.950769149188598e-07,
      "loss": 0.8307,
      "step": 1288
    },
    {
      "epoch": 0.127425054988508,
      "grad_norm": 15.465219342300303,
      "learning_rate": 1.950669866969997e-07,
      "loss": 0.7658,
      "step": 1289
    },
    {
      "epoch": 0.1275239107332625,
      "grad_norm": 5.7248459454171785,
      "learning_rate": 1.9505704872742243e-07,
      "loss": 0.8787,
      "step": 1290
    },
    {
      "epoch": 0.12762276647801696,
      "grad_norm": 7.820474639484991,
      "learning_rate": 1.9504710101114702e-07,
      "loss": 0.788,
      "step": 1291
    },
    {
      "epoch": 0.12772162222277142,
      "grad_norm": 7.184047667138028,
      "learning_rate": 1.9503714354919344e-07,
      "loss": 0.6762,
      "step": 1292
    },
    {
      "epoch": 0.12782047796752588,
      "grad_norm": 14.44474925276693,
      "learning_rate": 1.9502717634258268e-07,
      "loss": 0.7012,
      "step": 1293
    },
    {
      "epoch": 0.12791933371228034,
      "grad_norm": 5.897747577191385,
      "learning_rate": 1.9501719939233673e-07,
      "loss": 0.7559,
      "step": 1294
    },
    {
      "epoch": 0.12801818945703483,
      "grad_norm": 4.667608603366718,
      "learning_rate": 1.9500721269947859e-07,
      "loss": 0.6939,
      "step": 1295
    },
    {
      "epoch": 0.1281170452017893,
      "grad_norm": 3.248058468807248,
      "learning_rate": 1.9499721626503224e-07,
      "loss": 0.6804,
      "step": 1296
    },
    {
      "epoch": 0.12821590094654375,
      "grad_norm": 3.934328935106518,
      "learning_rate": 1.9498721009002268e-07,
      "loss": 0.8116,
      "step": 1297
    },
    {
      "epoch": 0.12831475669129822,
      "grad_norm": 6.735125656589037,
      "learning_rate": 1.949771941754759e-07,
      "loss": 0.7175,
      "step": 1298
    },
    {
      "epoch": 0.12841361243605268,
      "grad_norm": 3.259338332639882,
      "learning_rate": 1.9496716852241882e-07,
      "loss": 0.7276,
      "step": 1299
    },
    {
      "epoch": 0.12851246818080717,
      "grad_norm": 6.795060294357565,
      "learning_rate": 1.949571331318795e-07,
      "loss": 0.785,
      "step": 1300
    },
    {
      "epoch": 0.12861132392556163,
      "grad_norm": 4.924199099151521,
      "learning_rate": 1.949470880048869e-07,
      "loss": 0.6504,
      "step": 1301
    },
    {
      "epoch": 0.1287101796703161,
      "grad_norm": 5.795101902559786,
      "learning_rate": 1.94937033142471e-07,
      "loss": 0.7422,
      "step": 1302
    },
    {
      "epoch": 0.12880903541507055,
      "grad_norm": 5.807133416286141,
      "learning_rate": 1.9492696854566275e-07,
      "loss": 0.8002,
      "step": 1303
    },
    {
      "epoch": 0.12890789115982504,
      "grad_norm": 3.088621678409238,
      "learning_rate": 1.9491689421549417e-07,
      "loss": 0.7295,
      "step": 1304
    },
    {
      "epoch": 0.1290067469045795,
      "grad_norm": 9.41964476544067,
      "learning_rate": 1.9490681015299822e-07,
      "loss": 0.7674,
      "step": 1305
    },
    {
      "epoch": 0.12910560264933396,
      "grad_norm": 3.5569244825065462,
      "learning_rate": 1.9489671635920884e-07,
      "loss": 0.675,
      "step": 1306
    },
    {
      "epoch": 0.12920445839408842,
      "grad_norm": 7.4171468871707775,
      "learning_rate": 1.94886612835161e-07,
      "loss": 0.7885,
      "step": 1307
    },
    {
      "epoch": 0.12930331413884288,
      "grad_norm": 9.286194158144225,
      "learning_rate": 1.948764995818907e-07,
      "loss": 0.7028,
      "step": 1308
    },
    {
      "epoch": 0.12940216988359737,
      "grad_norm": 4.175889275207011,
      "learning_rate": 1.9486637660043493e-07,
      "loss": 0.8338,
      "step": 1309
    },
    {
      "epoch": 0.12950102562835183,
      "grad_norm": 18.35857279811028,
      "learning_rate": 1.9485624389183158e-07,
      "loss": 0.8069,
      "step": 1310
    },
    {
      "epoch": 0.1295998813731063,
      "grad_norm": 4.5818825508176255,
      "learning_rate": 1.9484610145711967e-07,
      "loss": 0.7754,
      "step": 1311
    },
    {
      "epoch": 0.12969873711786076,
      "grad_norm": 4.697021104375583,
      "learning_rate": 1.9483594929733912e-07,
      "loss": 0.6822,
      "step": 1312
    },
    {
      "epoch": 0.12979759286261522,
      "grad_norm": 7.372009666775526,
      "learning_rate": 1.9482578741353093e-07,
      "loss": 0.7255,
      "step": 1313
    },
    {
      "epoch": 0.1298964486073697,
      "grad_norm": 5.221355706547645,
      "learning_rate": 1.9481561580673696e-07,
      "loss": 0.8832,
      "step": 1314
    },
    {
      "epoch": 0.12999530435212417,
      "grad_norm": 5.399428139333168,
      "learning_rate": 1.9480543447800027e-07,
      "loss": 0.7226,
      "step": 1315
    },
    {
      "epoch": 0.13009416009687863,
      "grad_norm": 4.513574027608461,
      "learning_rate": 1.9479524342836473e-07,
      "loss": 0.8458,
      "step": 1316
    },
    {
      "epoch": 0.1301930158416331,
      "grad_norm": 5.108713985120417,
      "learning_rate": 1.947850426588753e-07,
      "loss": 0.684,
      "step": 1317
    },
    {
      "epoch": 0.13029187158638755,
      "grad_norm": 5.775529312253205,
      "learning_rate": 1.9477483217057795e-07,
      "loss": 0.7664,
      "step": 1318
    },
    {
      "epoch": 0.13039072733114204,
      "grad_norm": 4.794664882056166,
      "learning_rate": 1.9476461196451957e-07,
      "loss": 0.7883,
      "step": 1319
    },
    {
      "epoch": 0.1304895830758965,
      "grad_norm": 3.8488207081371977,
      "learning_rate": 1.947543820417481e-07,
      "loss": 0.7113,
      "step": 1320
    },
    {
      "epoch": 0.13058843882065096,
      "grad_norm": 3.559217896659162,
      "learning_rate": 1.947441424033125e-07,
      "loss": 0.6441,
      "step": 1321
    },
    {
      "epoch": 0.13068729456540543,
      "grad_norm": 4.149228042199806,
      "learning_rate": 1.9473389305026265e-07,
      "loss": 0.7678,
      "step": 1322
    },
    {
      "epoch": 0.1307861503101599,
      "grad_norm": 4.012179470123882,
      "learning_rate": 1.947236339836495e-07,
      "loss": 0.743,
      "step": 1323
    },
    {
      "epoch": 0.13088500605491438,
      "grad_norm": 3.776544306964287,
      "learning_rate": 1.9471336520452495e-07,
      "loss": 0.774,
      "step": 1324
    },
    {
      "epoch": 0.13098386179966884,
      "grad_norm": 3.720322182859924,
      "learning_rate": 1.947030867139419e-07,
      "loss": 0.7474,
      "step": 1325
    },
    {
      "epoch": 0.1310827175444233,
      "grad_norm": 3.424130844307389,
      "learning_rate": 1.9469279851295433e-07,
      "loss": 0.7309,
      "step": 1326
    },
    {
      "epoch": 0.13118157328917776,
      "grad_norm": 3.0943503359180076,
      "learning_rate": 1.9468250060261707e-07,
      "loss": 0.8612,
      "step": 1327
    },
    {
      "epoch": 0.13128042903393222,
      "grad_norm": 4.377012185510415,
      "learning_rate": 1.9467219298398602e-07,
      "loss": 0.6473,
      "step": 1328
    },
    {
      "epoch": 0.1313792847786867,
      "grad_norm": 23.455664801101783,
      "learning_rate": 1.946618756581181e-07,
      "loss": 0.8614,
      "step": 1329
    },
    {
      "epoch": 0.13147814052344117,
      "grad_norm": 4.235341487808135,
      "learning_rate": 1.9465154862607118e-07,
      "loss": 0.819,
      "step": 1330
    },
    {
      "epoch": 0.13157699626819563,
      "grad_norm": 3.8407259304836727,
      "learning_rate": 1.9464121188890417e-07,
      "loss": 0.6649,
      "step": 1331
    },
    {
      "epoch": 0.1316758520129501,
      "grad_norm": 36.76002447001568,
      "learning_rate": 1.9463086544767693e-07,
      "loss": 0.7403,
      "step": 1332
    },
    {
      "epoch": 0.13177470775770456,
      "grad_norm": 4.624763226917082,
      "learning_rate": 1.9462050930345035e-07,
      "loss": 0.8033,
      "step": 1333
    },
    {
      "epoch": 0.13187356350245905,
      "grad_norm": 4.048662439547007,
      "learning_rate": 1.946101434572863e-07,
      "loss": 0.7541,
      "step": 1334
    },
    {
      "epoch": 0.1319724192472135,
      "grad_norm": 7.5842227498787524,
      "learning_rate": 1.945997679102476e-07,
      "loss": 0.7668,
      "step": 1335
    },
    {
      "epoch": 0.13207127499196797,
      "grad_norm": 3.587197425998407,
      "learning_rate": 1.945893826633982e-07,
      "loss": 0.8553,
      "step": 1336
    },
    {
      "epoch": 0.13217013073672243,
      "grad_norm": 3.7058427650000274,
      "learning_rate": 1.9457898771780287e-07,
      "loss": 0.7371,
      "step": 1337
    },
    {
      "epoch": 0.1322689864814769,
      "grad_norm": 7.291052153888168,
      "learning_rate": 1.945685830745275e-07,
      "loss": 0.7215,
      "step": 1338
    },
    {
      "epoch": 0.13236784222623138,
      "grad_norm": 5.482840831478433,
      "learning_rate": 1.9455816873463892e-07,
      "loss": 0.7551,
      "step": 1339
    },
    {
      "epoch": 0.13246669797098584,
      "grad_norm": 10.130599407742041,
      "learning_rate": 1.9454774469920495e-07,
      "loss": 0.7321,
      "step": 1340
    },
    {
      "epoch": 0.1325655537157403,
      "grad_norm": 4.930751449113531,
      "learning_rate": 1.9453731096929445e-07,
      "loss": 0.6529,
      "step": 1341
    },
    {
      "epoch": 0.13266440946049476,
      "grad_norm": 4.539557532683274,
      "learning_rate": 1.9452686754597721e-07,
      "loss": 0.7042,
      "step": 1342
    },
    {
      "epoch": 0.13276326520524925,
      "grad_norm": 4.280443475694957,
      "learning_rate": 1.945164144303241e-07,
      "loss": 0.8114,
      "step": 1343
    },
    {
      "epoch": 0.13286212095000371,
      "grad_norm": 3.6569330611824262,
      "learning_rate": 1.945059516234069e-07,
      "loss": 0.6973,
      "step": 1344
    },
    {
      "epoch": 0.13296097669475818,
      "grad_norm": 6.584574166707988,
      "learning_rate": 1.944954791262984e-07,
      "loss": 0.7838,
      "step": 1345
    },
    {
      "epoch": 0.13305983243951264,
      "grad_norm": 29.24873544355493,
      "learning_rate": 1.9448499694007245e-07,
      "loss": 0.6982,
      "step": 1346
    },
    {
      "epoch": 0.1331586881842671,
      "grad_norm": 4.75546331336948,
      "learning_rate": 1.944745050658038e-07,
      "loss": 0.7532,
      "step": 1347
    },
    {
      "epoch": 0.1332575439290216,
      "grad_norm": 7.023118645785252,
      "learning_rate": 1.9446400350456829e-07,
      "loss": 0.7718,
      "step": 1348
    },
    {
      "epoch": 0.13335639967377605,
      "grad_norm": 3.7200800511128365,
      "learning_rate": 1.944534922574426e-07,
      "loss": 0.7079,
      "step": 1349
    },
    {
      "epoch": 0.1334552554185305,
      "grad_norm": 3.0702870115940817,
      "learning_rate": 1.9444297132550461e-07,
      "loss": 0.6964,
      "step": 1350
    },
    {
      "epoch": 0.13355411116328497,
      "grad_norm": 4.1721816190056975,
      "learning_rate": 1.9443244070983305e-07,
      "loss": 0.7392,
      "step": 1351
    },
    {
      "epoch": 0.13365296690803943,
      "grad_norm": 3.777911982959137,
      "learning_rate": 1.9442190041150764e-07,
      "loss": 0.7297,
      "step": 1352
    },
    {
      "epoch": 0.13375182265279392,
      "grad_norm": 5.648955675901979,
      "learning_rate": 1.9441135043160916e-07,
      "loss": 0.7971,
      "step": 1353
    },
    {
      "epoch": 0.13385067839754838,
      "grad_norm": 6.236030497699051,
      "learning_rate": 1.9440079077121937e-07,
      "loss": 0.7811,
      "step": 1354
    },
    {
      "epoch": 0.13394953414230285,
      "grad_norm": 3.615837981281405,
      "learning_rate": 1.9439022143142103e-07,
      "loss": 0.6834,
      "step": 1355
    },
    {
      "epoch": 0.1340483898870573,
      "grad_norm": 5.560580172868369,
      "learning_rate": 1.9437964241329778e-07,
      "loss": 0.7275,
      "step": 1356
    },
    {
      "epoch": 0.13414724563181177,
      "grad_norm": 3.621762819330606,
      "learning_rate": 1.9436905371793445e-07,
      "loss": 0.7102,
      "step": 1357
    },
    {
      "epoch": 0.13424610137656626,
      "grad_norm": 4.660293085117622,
      "learning_rate": 1.9435845534641666e-07,
      "loss": 0.7777,
      "step": 1358
    },
    {
      "epoch": 0.13434495712132072,
      "grad_norm": 7.63569243960629,
      "learning_rate": 1.9434784729983118e-07,
      "loss": 0.779,
      "step": 1359
    },
    {
      "epoch": 0.13444381286607518,
      "grad_norm": 3.3268529024897826,
      "learning_rate": 1.9433722957926567e-07,
      "loss": 0.7834,
      "step": 1360
    },
    {
      "epoch": 0.13454266861082964,
      "grad_norm": 29.68526888014553,
      "learning_rate": 1.9432660218580886e-07,
      "loss": 0.793,
      "step": 1361
    },
    {
      "epoch": 0.1346415243555841,
      "grad_norm": 4.506316873453273,
      "learning_rate": 1.9431596512055043e-07,
      "loss": 0.7301,
      "step": 1362
    },
    {
      "epoch": 0.1347403801003386,
      "grad_norm": 6.193933803275632,
      "learning_rate": 1.94305318384581e-07,
      "loss": 0.7319,
      "step": 1363
    },
    {
      "epoch": 0.13483923584509305,
      "grad_norm": 9.47378067408924,
      "learning_rate": 1.9429466197899228e-07,
      "loss": 0.7472,
      "step": 1364
    },
    {
      "epoch": 0.13493809158984751,
      "grad_norm": 4.180190812584948,
      "learning_rate": 1.942839959048769e-07,
      "loss": 0.7279,
      "step": 1365
    },
    {
      "epoch": 0.13503694733460198,
      "grad_norm": 21.98513867260318,
      "learning_rate": 1.9427332016332855e-07,
      "loss": 0.6663,
      "step": 1366
    },
    {
      "epoch": 0.13513580307935644,
      "grad_norm": 11.708253059193414,
      "learning_rate": 1.9426263475544186e-07,
      "loss": 0.7114,
      "step": 1367
    },
    {
      "epoch": 0.13523465882411093,
      "grad_norm": 4.573183701005416,
      "learning_rate": 1.9425193968231244e-07,
      "loss": 0.7566,
      "step": 1368
    },
    {
      "epoch": 0.1353335145688654,
      "grad_norm": 3.8387281049553135,
      "learning_rate": 1.942412349450369e-07,
      "loss": 0.7859,
      "step": 1369
    },
    {
      "epoch": 0.13543237031361985,
      "grad_norm": 4.8122996739465815,
      "learning_rate": 1.9423052054471288e-07,
      "loss": 0.7044,
      "step": 1370
    },
    {
      "epoch": 0.1355312260583743,
      "grad_norm": 22.411899130703777,
      "learning_rate": 1.9421979648243897e-07,
      "loss": 0.8109,
      "step": 1371
    },
    {
      "epoch": 0.13563008180312877,
      "grad_norm": 4.3781717748279885,
      "learning_rate": 1.942090627593148e-07,
      "loss": 0.7535,
      "step": 1372
    },
    {
      "epoch": 0.13572893754788326,
      "grad_norm": 5.687653360861449,
      "learning_rate": 1.941983193764409e-07,
      "loss": 0.7418,
      "step": 1373
    },
    {
      "epoch": 0.13582779329263772,
      "grad_norm": 4.35297259949213,
      "learning_rate": 1.9418756633491887e-07,
      "loss": 0.7388,
      "step": 1374
    },
    {
      "epoch": 0.13592664903739218,
      "grad_norm": 6.152921087589322,
      "learning_rate": 1.9417680363585128e-07,
      "loss": 0.8501,
      "step": 1375
    },
    {
      "epoch": 0.13602550478214664,
      "grad_norm": 2.92710831127592,
      "learning_rate": 1.9416603128034167e-07,
      "loss": 0.864,
      "step": 1376
    },
    {
      "epoch": 0.1361243605269011,
      "grad_norm": 4.941607333415102,
      "learning_rate": 1.9415524926949463e-07,
      "loss": 0.7149,
      "step": 1377
    },
    {
      "epoch": 0.1362232162716556,
      "grad_norm": 3.9649950358114263,
      "learning_rate": 1.9414445760441564e-07,
      "loss": 0.6256,
      "step": 1378
    },
    {
      "epoch": 0.13632207201641006,
      "grad_norm": 4.650852239190141,
      "learning_rate": 1.9413365628621125e-07,
      "loss": 0.6982,
      "step": 1379
    },
    {
      "epoch": 0.13642092776116452,
      "grad_norm": 4.219773324319354,
      "learning_rate": 1.94122845315989e-07,
      "loss": 0.7725,
      "step": 1380
    },
    {
      "epoch": 0.13651978350591898,
      "grad_norm": 3.1555974278550107,
      "learning_rate": 1.9411202469485736e-07,
      "loss": 0.7742,
      "step": 1381
    },
    {
      "epoch": 0.13661863925067347,
      "grad_norm": 4.738023842130687,
      "learning_rate": 1.9410119442392582e-07,
      "loss": 0.7889,
      "step": 1382
    },
    {
      "epoch": 0.13671749499542793,
      "grad_norm": 3.944258295157563,
      "learning_rate": 1.9409035450430491e-07,
      "loss": 0.8146,
      "step": 1383
    },
    {
      "epoch": 0.1368163507401824,
      "grad_norm": 4.38271913121012,
      "learning_rate": 1.9407950493710606e-07,
      "loss": 0.7929,
      "step": 1384
    },
    {
      "epoch": 0.13691520648493685,
      "grad_norm": 4.2414662942654315,
      "learning_rate": 1.9406864572344176e-07,
      "loss": 0.8052,
      "step": 1385
    },
    {
      "epoch": 0.1370140622296913,
      "grad_norm": 4.252615987101041,
      "learning_rate": 1.9405777686442546e-07,
      "loss": 0.7845,
      "step": 1386
    },
    {
      "epoch": 0.1371129179744458,
      "grad_norm": 18.199539468330546,
      "learning_rate": 1.9404689836117156e-07,
      "loss": 0.7206,
      "step": 1387
    },
    {
      "epoch": 0.13721177371920026,
      "grad_norm": 6.907778440197796,
      "learning_rate": 1.9403601021479555e-07,
      "loss": 0.6843,
      "step": 1388
    },
    {
      "epoch": 0.13731062946395473,
      "grad_norm": 5.896600317169673,
      "learning_rate": 1.940251124264138e-07,
      "loss": 0.7344,
      "step": 1389
    },
    {
      "epoch": 0.1374094852087092,
      "grad_norm": 10.008726338069106,
      "learning_rate": 1.9401420499714376e-07,
      "loss": 0.821,
      "step": 1390
    },
    {
      "epoch": 0.13750834095346365,
      "grad_norm": 3.0136658479596257,
      "learning_rate": 1.940032879281038e-07,
      "loss": 0.7144,
      "step": 1391
    },
    {
      "epoch": 0.13760719669821814,
      "grad_norm": 5.456440668372255,
      "learning_rate": 1.939923612204133e-07,
      "loss": 0.7217,
      "step": 1392
    },
    {
      "epoch": 0.1377060524429726,
      "grad_norm": 6.020398618263541,
      "learning_rate": 1.9398142487519263e-07,
      "loss": 0.6423,
      "step": 1393
    },
    {
      "epoch": 0.13780490818772706,
      "grad_norm": 5.960336313331038,
      "learning_rate": 1.9397047889356316e-07,
      "loss": 0.7137,
      "step": 1394
    },
    {
      "epoch": 0.13790376393248152,
      "grad_norm": 17.50382780907154,
      "learning_rate": 1.9395952327664727e-07,
      "loss": 0.7425,
      "step": 1395
    },
    {
      "epoch": 0.13800261967723598,
      "grad_norm": 3.780528294448486,
      "learning_rate": 1.939485580255683e-07,
      "loss": 0.673,
      "step": 1396
    },
    {
      "epoch": 0.13810147542199047,
      "grad_norm": 13.140642357908101,
      "learning_rate": 1.9393758314145047e-07,
      "loss": 0.7024,
      "step": 1397
    },
    {
      "epoch": 0.13820033116674493,
      "grad_norm": 3.777844836239173,
      "learning_rate": 1.939265986254192e-07,
      "loss": 0.7478,
      "step": 1398
    },
    {
      "epoch": 0.1382991869114994,
      "grad_norm": 15.98678841525936,
      "learning_rate": 1.9391560447860077e-07,
      "loss": 0.74,
      "step": 1399
    },
    {
      "epoch": 0.13839804265625386,
      "grad_norm": 5.041189944991649,
      "learning_rate": 1.939046007021224e-07,
      "loss": 0.7324,
      "step": 1400
    },
    {
      "epoch": 0.13849689840100832,
      "grad_norm": 3.4559382821560436,
      "learning_rate": 1.9389358729711246e-07,
      "loss": 0.7291,
      "step": 1401
    },
    {
      "epoch": 0.1385957541457628,
      "grad_norm": 11.125760399271332,
      "learning_rate": 1.9388256426470017e-07,
      "loss": 0.7538,
      "step": 1402
    },
    {
      "epoch": 0.13869460989051727,
      "grad_norm": 3.430411631490619,
      "learning_rate": 1.9387153160601578e-07,
      "loss": 0.7266,
      "step": 1403
    },
    {
      "epoch": 0.13879346563527173,
      "grad_norm": 11.993985952886646,
      "learning_rate": 1.9386048932219052e-07,
      "loss": 0.8122,
      "step": 1404
    },
    {
      "epoch": 0.1388923213800262,
      "grad_norm": 15.290296896549313,
      "learning_rate": 1.938494374143566e-07,
      "loss": 0.8529,
      "step": 1405
    },
    {
      "epoch": 0.13899117712478065,
      "grad_norm": 19.109306753259112,
      "learning_rate": 1.938383758836473e-07,
      "loss": 0.7385,
      "step": 1406
    },
    {
      "epoch": 0.13909003286953514,
      "grad_norm": 3.615065507727475,
      "learning_rate": 1.9382730473119672e-07,
      "loss": 0.6506,
      "step": 1407
    },
    {
      "epoch": 0.1391888886142896,
      "grad_norm": 9.528671558582328,
      "learning_rate": 1.9381622395814012e-07,
      "loss": 0.6654,
      "step": 1408
    },
    {
      "epoch": 0.13928774435904406,
      "grad_norm": 13.771324013398806,
      "learning_rate": 1.9380513356561362e-07,
      "loss": 0.7969,
      "step": 1409
    },
    {
      "epoch": 0.13938660010379852,
      "grad_norm": 4.0698850042679195,
      "learning_rate": 1.9379403355475442e-07,
      "loss": 0.7935,
      "step": 1410
    },
    {
      "epoch": 0.13948545584855299,
      "grad_norm": 6.2220305786063035,
      "learning_rate": 1.9378292392670063e-07,
      "loss": 0.7663,
      "step": 1411
    },
    {
      "epoch": 0.13958431159330748,
      "grad_norm": 3.37158315674799,
      "learning_rate": 1.9377180468259135e-07,
      "loss": 0.8319,
      "step": 1412
    },
    {
      "epoch": 0.13968316733806194,
      "grad_norm": 13.23424261209968,
      "learning_rate": 1.9376067582356677e-07,
      "loss": 0.7455,
      "step": 1413
    },
    {
      "epoch": 0.1397820230828164,
      "grad_norm": 6.0262709041883875,
      "learning_rate": 1.9374953735076795e-07,
      "loss": 0.7564,
      "step": 1414
    },
    {
      "epoch": 0.13988087882757086,
      "grad_norm": 5.440792351084578,
      "learning_rate": 1.9373838926533695e-07,
      "loss": 0.6601,
      "step": 1415
    },
    {
      "epoch": 0.13997973457232532,
      "grad_norm": 15.160546657396278,
      "learning_rate": 1.937272315684169e-07,
      "loss": 0.6886,
      "step": 1416
    },
    {
      "epoch": 0.1400785903170798,
      "grad_norm": 3.340986150286348,
      "learning_rate": 1.937160642611518e-07,
      "loss": 0.6758,
      "step": 1417
    },
    {
      "epoch": 0.14017744606183427,
      "grad_norm": 6.371982665642472,
      "learning_rate": 1.9370488734468676e-07,
      "loss": 0.7167,
      "step": 1418
    },
    {
      "epoch": 0.14027630180658873,
      "grad_norm": 5.038146212194647,
      "learning_rate": 1.9369370082016775e-07,
      "loss": 0.7466,
      "step": 1419
    },
    {
      "epoch": 0.1403751575513432,
      "grad_norm": 4.100900260420734,
      "learning_rate": 1.9368250468874176e-07,
      "loss": 0.8302,
      "step": 1420
    },
    {
      "epoch": 0.14047401329609768,
      "grad_norm": 5.450668542794107,
      "learning_rate": 1.9367129895155685e-07,
      "loss": 0.8322,
      "step": 1421
    },
    {
      "epoch": 0.14057286904085214,
      "grad_norm": 4.662044682651495,
      "learning_rate": 1.93660083609762e-07,
      "loss": 0.7804,
      "step": 1422
    },
    {
      "epoch": 0.1406717247856066,
      "grad_norm": 4.505313939666244,
      "learning_rate": 1.9364885866450714e-07,
      "loss": 0.7709,
      "step": 1423
    },
    {
      "epoch": 0.14077058053036107,
      "grad_norm": 4.722296370589035,
      "learning_rate": 1.9363762411694326e-07,
      "loss": 0.8197,
      "step": 1424
    },
    {
      "epoch": 0.14086943627511553,
      "grad_norm": 5.457639767685148,
      "learning_rate": 1.9362637996822228e-07,
      "loss": 0.6688,
      "step": 1425
    },
    {
      "epoch": 0.14096829201987002,
      "grad_norm": 3.4443641149801154,
      "learning_rate": 1.936151262194971e-07,
      "loss": 0.736,
      "step": 1426
    },
    {
      "epoch": 0.14106714776462448,
      "grad_norm": 5.124536878684874,
      "learning_rate": 1.9360386287192165e-07,
      "loss": 0.7431,
      "step": 1427
    },
    {
      "epoch": 0.14116600350937894,
      "grad_norm": 7.542218355017893,
      "learning_rate": 1.9359258992665084e-07,
      "loss": 0.8147,
      "step": 1428
    },
    {
      "epoch": 0.1412648592541334,
      "grad_norm": 3.9881518022981357,
      "learning_rate": 1.9358130738484052e-07,
      "loss": 0.7174,
      "step": 1429
    },
    {
      "epoch": 0.14136371499888786,
      "grad_norm": 6.720187345525028,
      "learning_rate": 1.935700152476475e-07,
      "loss": 0.7843,
      "step": 1430
    },
    {
      "epoch": 0.14146257074364235,
      "grad_norm": 4.6374020964250375,
      "learning_rate": 1.9355871351622972e-07,
      "loss": 0.7268,
      "step": 1431
    },
    {
      "epoch": 0.1415614264883968,
      "grad_norm": 4.37549976521285,
      "learning_rate": 1.9354740219174595e-07,
      "loss": 0.7189,
      "step": 1432
    },
    {
      "epoch": 0.14166028223315127,
      "grad_norm": 5.254152100489806,
      "learning_rate": 1.93536081275356e-07,
      "loss": 0.6686,
      "step": 1433
    },
    {
      "epoch": 0.14175913797790574,
      "grad_norm": 4.319821875227706,
      "learning_rate": 1.935247507682207e-07,
      "loss": 0.7172,
      "step": 1434
    },
    {
      "epoch": 0.1418579937226602,
      "grad_norm": 4.358075376224312,
      "learning_rate": 1.9351341067150173e-07,
      "loss": 0.8069,
      "step": 1435
    },
    {
      "epoch": 0.1419568494674147,
      "grad_norm": 6.293810939166785,
      "learning_rate": 1.9350206098636194e-07,
      "loss": 0.7603,
      "step": 1436
    },
    {
      "epoch": 0.14205570521216915,
      "grad_norm": 11.54639304270749,
      "learning_rate": 1.9349070171396507e-07,
      "loss": 0.6856,
      "step": 1437
    },
    {
      "epoch": 0.1421545609569236,
      "grad_norm": 3.3547288186254423,
      "learning_rate": 1.934793328554758e-07,
      "loss": 0.6903,
      "step": 1438
    },
    {
      "epoch": 0.14225341670167807,
      "grad_norm": 4.728455184453007,
      "learning_rate": 1.9346795441205986e-07,
      "loss": 0.8374,
      "step": 1439
    },
    {
      "epoch": 0.14235227244643253,
      "grad_norm": 4.084892295794369,
      "learning_rate": 1.9345656638488399e-07,
      "loss": 0.6927,
      "step": 1440
    },
    {
      "epoch": 0.14245112819118702,
      "grad_norm": 8.311491093754027,
      "learning_rate": 1.934451687751158e-07,
      "loss": 0.7855,
      "step": 1441
    },
    {
      "epoch": 0.14254998393594148,
      "grad_norm": 8.293275287486674,
      "learning_rate": 1.9343376158392394e-07,
      "loss": 0.7438,
      "step": 1442
    },
    {
      "epoch": 0.14264883968069594,
      "grad_norm": 7.223429305970848,
      "learning_rate": 1.934223448124781e-07,
      "loss": 0.7357,
      "step": 1443
    },
    {
      "epoch": 0.1427476954254504,
      "grad_norm": 3.0730419976838195,
      "learning_rate": 1.9341091846194885e-07,
      "loss": 0.7995,
      "step": 1444
    },
    {
      "epoch": 0.14284655117020487,
      "grad_norm": 8.761012232056142,
      "learning_rate": 1.9339948253350783e-07,
      "loss": 0.7657,
      "step": 1445
    },
    {
      "epoch": 0.14294540691495936,
      "grad_norm": 3.6141645075230033,
      "learning_rate": 1.9338803702832763e-07,
      "loss": 0.755,
      "step": 1446
    },
    {
      "epoch": 0.14304426265971382,
      "grad_norm": 3.415139603183968,
      "learning_rate": 1.933765819475818e-07,
      "loss": 0.6291,
      "step": 1447
    },
    {
      "epoch": 0.14314311840446828,
      "grad_norm": 6.521962319653034,
      "learning_rate": 1.9336511729244486e-07,
      "loss": 0.7549,
      "step": 1448
    },
    {
      "epoch": 0.14324197414922274,
      "grad_norm": 4.339468239543043,
      "learning_rate": 1.9335364306409242e-07,
      "loss": 0.757,
      "step": 1449
    },
    {
      "epoch": 0.1433408298939772,
      "grad_norm": 4.558189990777735,
      "learning_rate": 1.9334215926370092e-07,
      "loss": 0.8108,
      "step": 1450
    },
    {
      "epoch": 0.1434396856387317,
      "grad_norm": 6.979782870282327,
      "learning_rate": 1.933306658924479e-07,
      "loss": 0.828,
      "step": 1451
    },
    {
      "epoch": 0.14353854138348615,
      "grad_norm": 3.9402975878148365,
      "learning_rate": 1.933191629515118e-07,
      "loss": 0.6948,
      "step": 1452
    },
    {
      "epoch": 0.1436373971282406,
      "grad_norm": 3.696961671807679,
      "learning_rate": 1.9330765044207208e-07,
      "loss": 0.724,
      "step": 1453
    },
    {
      "epoch": 0.14373625287299507,
      "grad_norm": 3.879991912958697,
      "learning_rate": 1.9329612836530922e-07,
      "loss": 0.8277,
      "step": 1454
    },
    {
      "epoch": 0.14383510861774954,
      "grad_norm": 4.932418438866404,
      "learning_rate": 1.932845967224046e-07,
      "loss": 0.7401,
      "step": 1455
    },
    {
      "epoch": 0.14393396436250402,
      "grad_norm": 5.101493830846892,
      "learning_rate": 1.9327305551454062e-07,
      "loss": 0.7159,
      "step": 1456
    },
    {
      "epoch": 0.14403282010725849,
      "grad_norm": 3.837683939972634,
      "learning_rate": 1.932615047429007e-07,
      "loss": 0.7344,
      "step": 1457
    },
    {
      "epoch": 0.14413167585201295,
      "grad_norm": 6.568224750773238,
      "learning_rate": 1.9324994440866915e-07,
      "loss": 0.7199,
      "step": 1458
    },
    {
      "epoch": 0.1442305315967674,
      "grad_norm": 8.17601519218967,
      "learning_rate": 1.9323837451303135e-07,
      "loss": 0.5969,
      "step": 1459
    },
    {
      "epoch": 0.1443293873415219,
      "grad_norm": 4.019112032465348,
      "learning_rate": 1.932267950571736e-07,
      "loss": 0.8197,
      "step": 1460
    },
    {
      "epoch": 0.14442824308627636,
      "grad_norm": 3.65502413426619,
      "learning_rate": 1.9321520604228318e-07,
      "loss": 0.7397,
      "step": 1461
    },
    {
      "epoch": 0.14452709883103082,
      "grad_norm": 3.3740961649192167,
      "learning_rate": 1.9320360746954844e-07,
      "loss": 0.7374,
      "step": 1462
    },
    {
      "epoch": 0.14462595457578528,
      "grad_norm": 3.3393584905667497,
      "learning_rate": 1.931919993401586e-07,
      "loss": 0.6669,
      "step": 1463
    },
    {
      "epoch": 0.14472481032053974,
      "grad_norm": 4.82517060544772,
      "learning_rate": 1.9318038165530388e-07,
      "loss": 0.6037,
      "step": 1464
    },
    {
      "epoch": 0.14482366606529423,
      "grad_norm": 2.9205808105180444,
      "learning_rate": 1.9316875441617554e-07,
      "loss": 0.7483,
      "step": 1465
    },
    {
      "epoch": 0.1449225218100487,
      "grad_norm": 3.1539630004781527,
      "learning_rate": 1.9315711762396578e-07,
      "loss": 0.8731,
      "step": 1466
    },
    {
      "epoch": 0.14502137755480315,
      "grad_norm": 5.034377152047869,
      "learning_rate": 1.9314547127986777e-07,
      "loss": 0.726,
      "step": 1467
    },
    {
      "epoch": 0.14512023329955762,
      "grad_norm": 13.836598060302233,
      "learning_rate": 1.9313381538507566e-07,
      "loss": 0.7733,
      "step": 1468
    },
    {
      "epoch": 0.14521908904431208,
      "grad_norm": 5.585324995373478,
      "learning_rate": 1.9312214994078463e-07,
      "loss": 0.6232,
      "step": 1469
    },
    {
      "epoch": 0.14531794478906657,
      "grad_norm": 3.383043496983211,
      "learning_rate": 1.9311047494819074e-07,
      "loss": 0.7256,
      "step": 1470
    },
    {
      "epoch": 0.14541680053382103,
      "grad_norm": 3.489694879061595,
      "learning_rate": 1.9309879040849115e-07,
      "loss": 0.8566,
      "step": 1471
    },
    {
      "epoch": 0.1455156562785755,
      "grad_norm": 3.6263275502866494,
      "learning_rate": 1.930870963228839e-07,
      "loss": 0.6735,
      "step": 1472
    },
    {
      "epoch": 0.14561451202332995,
      "grad_norm": 3.553057924550956,
      "learning_rate": 1.9307539269256804e-07,
      "loss": 0.7887,
      "step": 1473
    },
    {
      "epoch": 0.1457133677680844,
      "grad_norm": 123.46346679141459,
      "learning_rate": 1.9306367951874364e-07,
      "loss": 0.6818,
      "step": 1474
    },
    {
      "epoch": 0.1458122235128389,
      "grad_norm": 3.0945519996571327,
      "learning_rate": 1.9305195680261165e-07,
      "loss": 0.7427,
      "step": 1475
    },
    {
      "epoch": 0.14591107925759336,
      "grad_norm": 4.051772402279348,
      "learning_rate": 1.9304022454537413e-07,
      "loss": 0.7294,
      "step": 1476
    },
    {
      "epoch": 0.14600993500234782,
      "grad_norm": 9.35186747196895,
      "learning_rate": 1.9302848274823402e-07,
      "loss": 0.7408,
      "step": 1477
    },
    {
      "epoch": 0.14610879074710229,
      "grad_norm": 15.35706687457826,
      "learning_rate": 1.930167314123953e-07,
      "loss": 0.6714,
      "step": 1478
    },
    {
      "epoch": 0.14620764649185675,
      "grad_norm": 11.86036047702726,
      "learning_rate": 1.9300497053906283e-07,
      "loss": 0.6643,
      "step": 1479
    },
    {
      "epoch": 0.14630650223661124,
      "grad_norm": 6.749478052614757,
      "learning_rate": 1.9299320012944257e-07,
      "loss": 0.6805,
      "step": 1480
    },
    {
      "epoch": 0.1464053579813657,
      "grad_norm": 3.283058326494834,
      "learning_rate": 1.9298142018474137e-07,
      "loss": 0.7011,
      "step": 1481
    },
    {
      "epoch": 0.14650421372612016,
      "grad_norm": 3.412368370680551,
      "learning_rate": 1.929696307061671e-07,
      "loss": 0.8198,
      "step": 1482
    },
    {
      "epoch": 0.14660306947087462,
      "grad_norm": 4.289625409739477,
      "learning_rate": 1.9295783169492862e-07,
      "loss": 0.6152,
      "step": 1483
    },
    {
      "epoch": 0.14670192521562908,
      "grad_norm": 5.013028713005321,
      "learning_rate": 1.9294602315223569e-07,
      "loss": 0.6591,
      "step": 1484
    },
    {
      "epoch": 0.14680078096038357,
      "grad_norm": 4.962834750584621,
      "learning_rate": 1.9293420507929917e-07,
      "loss": 0.7536,
      "step": 1485
    },
    {
      "epoch": 0.14689963670513803,
      "grad_norm": 7.863441490118767,
      "learning_rate": 1.9292237747733076e-07,
      "loss": 0.7886,
      "step": 1486
    },
    {
      "epoch": 0.1469984924498925,
      "grad_norm": 4.451697087553442,
      "learning_rate": 1.9291054034754325e-07,
      "loss": 0.6923,
      "step": 1487
    },
    {
      "epoch": 0.14709734819464695,
      "grad_norm": 14.3244214071126,
      "learning_rate": 1.928986936911504e-07,
      "loss": 0.7917,
      "step": 1488
    },
    {
      "epoch": 0.14719620393940142,
      "grad_norm": 4.285114122989727,
      "learning_rate": 1.928868375093668e-07,
      "loss": 0.7854,
      "step": 1489
    },
    {
      "epoch": 0.1472950596841559,
      "grad_norm": 3.8766697100583647,
      "learning_rate": 1.9287497180340825e-07,
      "loss": 0.7583,
      "step": 1490
    },
    {
      "epoch": 0.14739391542891037,
      "grad_norm": 4.544588849791243,
      "learning_rate": 1.9286309657449128e-07,
      "loss": 0.7193,
      "step": 1491
    },
    {
      "epoch": 0.14749277117366483,
      "grad_norm": 3.51320475948321,
      "learning_rate": 1.9285121182383365e-07,
      "loss": 0.7396,
      "step": 1492
    },
    {
      "epoch": 0.1475916269184193,
      "grad_norm": 5.222921275181644,
      "learning_rate": 1.9283931755265383e-07,
      "loss": 0.6992,
      "step": 1493
    },
    {
      "epoch": 0.14769048266317375,
      "grad_norm": 5.5037023775055856,
      "learning_rate": 1.9282741376217152e-07,
      "loss": 0.6968,
      "step": 1494
    },
    {
      "epoch": 0.14778933840792824,
      "grad_norm": 4.752937492371211,
      "learning_rate": 1.9281550045360722e-07,
      "loss": 0.7051,
      "step": 1495
    },
    {
      "epoch": 0.1478881941526827,
      "grad_norm": 5.757977579453367,
      "learning_rate": 1.928035776281825e-07,
      "loss": 0.7373,
      "step": 1496
    },
    {
      "epoch": 0.14798704989743716,
      "grad_norm": 4.95884198986748,
      "learning_rate": 1.927916452871198e-07,
      "loss": 0.7491,
      "step": 1497
    },
    {
      "epoch": 0.14808590564219162,
      "grad_norm": 4.769487633549527,
      "learning_rate": 1.9277970343164266e-07,
      "loss": 0.7371,
      "step": 1498
    },
    {
      "epoch": 0.1481847613869461,
      "grad_norm": 7.233728705859902,
      "learning_rate": 1.9276775206297552e-07,
      "loss": 0.74,
      "step": 1499
    },
    {
      "epoch": 0.14828361713170057,
      "grad_norm": 5.446996892682441,
      "learning_rate": 1.9275579118234387e-07,
      "loss": 0.7054,
      "step": 1500
    },
    {
      "epoch": 0.14838247287645503,
      "grad_norm": 3.758225553458617,
      "learning_rate": 1.9274382079097403e-07,
      "loss": 0.69,
      "step": 1501
    },
    {
      "epoch": 0.1484813286212095,
      "grad_norm": 8.33796855242717,
      "learning_rate": 1.9273184089009344e-07,
      "loss": 0.6461,
      "step": 1502
    },
    {
      "epoch": 0.14858018436596396,
      "grad_norm": 4.742184176616951,
      "learning_rate": 1.927198514809305e-07,
      "loss": 0.7053,
      "step": 1503
    },
    {
      "epoch": 0.14867904011071845,
      "grad_norm": 4.817529328880994,
      "learning_rate": 1.9270785256471446e-07,
      "loss": 0.7569,
      "step": 1504
    },
    {
      "epoch": 0.1487778958554729,
      "grad_norm": 25.59893032923366,
      "learning_rate": 1.9269584414267567e-07,
      "loss": 0.6292,
      "step": 1505
    },
    {
      "epoch": 0.14887675160022737,
      "grad_norm": 6.168159052469597,
      "learning_rate": 1.9268382621604545e-07,
      "loss": 0.7788,
      "step": 1506
    },
    {
      "epoch": 0.14897560734498183,
      "grad_norm": 8.692739381097766,
      "learning_rate": 1.9267179878605603e-07,
      "loss": 0.7796,
      "step": 1507
    },
    {
      "epoch": 0.1490744630897363,
      "grad_norm": 3.4118822461261553,
      "learning_rate": 1.9265976185394067e-07,
      "loss": 0.856,
      "step": 1508
    },
    {
      "epoch": 0.14917331883449078,
      "grad_norm": 5.427039541739763,
      "learning_rate": 1.9264771542093354e-07,
      "loss": 0.8505,
      "step": 1509
    },
    {
      "epoch": 0.14927217457924524,
      "grad_norm": 4.782109353730399,
      "learning_rate": 1.9263565948826983e-07,
      "loss": 0.792,
      "step": 1510
    },
    {
      "epoch": 0.1493710303239997,
      "grad_norm": 10.940986123839895,
      "learning_rate": 1.9262359405718574e-07,
      "loss": 0.712,
      "step": 1511
    },
    {
      "epoch": 0.14946988606875417,
      "grad_norm": 3.3706384552712527,
      "learning_rate": 1.9261151912891834e-07,
      "loss": 0.7021,
      "step": 1512
    },
    {
      "epoch": 0.14956874181350863,
      "grad_norm": 4.445894250883976,
      "learning_rate": 1.925994347047058e-07,
      "loss": 0.7552,
      "step": 1513
    },
    {
      "epoch": 0.14966759755826312,
      "grad_norm": 4.126808956205197,
      "learning_rate": 1.9258734078578716e-07,
      "loss": 0.8587,
      "step": 1514
    },
    {
      "epoch": 0.14976645330301758,
      "grad_norm": 3.8608640421775666,
      "learning_rate": 1.9257523737340246e-07,
      "loss": 0.7259,
      "step": 1515
    },
    {
      "epoch": 0.14986530904777204,
      "grad_norm": 7.999588573156191,
      "learning_rate": 1.9256312446879277e-07,
      "loss": 0.6297,
      "step": 1516
    },
    {
      "epoch": 0.1499641647925265,
      "grad_norm": 3.728283848292913,
      "learning_rate": 1.9255100207320008e-07,
      "loss": 0.821,
      "step": 1517
    },
    {
      "epoch": 0.15006302053728096,
      "grad_norm": 6.998036941205037,
      "learning_rate": 1.925388701878673e-07,
      "loss": 0.592,
      "step": 1518
    },
    {
      "epoch": 0.15016187628203545,
      "grad_norm": 3.3605830413765383,
      "learning_rate": 1.925267288140385e-07,
      "loss": 0.7186,
      "step": 1519
    },
    {
      "epoch": 0.1502607320267899,
      "grad_norm": 30.00060491495836,
      "learning_rate": 1.925145779529585e-07,
      "loss": 0.6854,
      "step": 1520
    },
    {
      "epoch": 0.15035958777154437,
      "grad_norm": 6.970179356784423,
      "learning_rate": 1.925024176058732e-07,
      "loss": 0.8279,
      "step": 1521
    },
    {
      "epoch": 0.15045844351629883,
      "grad_norm": 4.775717585131605,
      "learning_rate": 1.9249024777402947e-07,
      "loss": 0.7929,
      "step": 1522
    },
    {
      "epoch": 0.1505572992610533,
      "grad_norm": 4.841735468332748,
      "learning_rate": 1.924780684586752e-07,
      "loss": 0.6897,
      "step": 1523
    },
    {
      "epoch": 0.15065615500580778,
      "grad_norm": 10.12166633802942,
      "learning_rate": 1.9246587966105912e-07,
      "loss": 0.8085,
      "step": 1524
    },
    {
      "epoch": 0.15075501075056225,
      "grad_norm": 3.076304912750501,
      "learning_rate": 1.9245368138243103e-07,
      "loss": 0.7482,
      "step": 1525
    },
    {
      "epoch": 0.1508538664953167,
      "grad_norm": 5.192137743877308,
      "learning_rate": 1.9244147362404175e-07,
      "loss": 0.6607,
      "step": 1526
    },
    {
      "epoch": 0.15095272224007117,
      "grad_norm": 3.714455601447199,
      "learning_rate": 1.9242925638714293e-07,
      "loss": 0.7035,
      "step": 1527
    },
    {
      "epoch": 0.15105157798482563,
      "grad_norm": 8.812304574468612,
      "learning_rate": 1.924170296729873e-07,
      "loss": 0.7764,
      "step": 1528
    },
    {
      "epoch": 0.15115043372958012,
      "grad_norm": 19.51234441335062,
      "learning_rate": 1.9240479348282852e-07,
      "loss": 0.6953,
      "step": 1529
    },
    {
      "epoch": 0.15124928947433458,
      "grad_norm": 9.573115054523573,
      "learning_rate": 1.9239254781792126e-07,
      "loss": 0.6851,
      "step": 1530
    },
    {
      "epoch": 0.15134814521908904,
      "grad_norm": 4.528486059217281,
      "learning_rate": 1.923802926795211e-07,
      "loss": 0.7348,
      "step": 1531
    },
    {
      "epoch": 0.1514470009638435,
      "grad_norm": 3.6997505126599233,
      "learning_rate": 1.923680280688846e-07,
      "loss": 0.6958,
      "step": 1532
    },
    {
      "epoch": 0.15154585670859796,
      "grad_norm": 6.591169756644878,
      "learning_rate": 1.9235575398726937e-07,
      "loss": 0.8258,
      "step": 1533
    },
    {
      "epoch": 0.15164471245335245,
      "grad_norm": 4.495197279001063,
      "learning_rate": 1.923434704359339e-07,
      "loss": 0.7023,
      "step": 1534
    },
    {
      "epoch": 0.15174356819810692,
      "grad_norm": 28.017614527403055,
      "learning_rate": 1.9233117741613772e-07,
      "loss": 0.8394,
      "step": 1535
    },
    {
      "epoch": 0.15184242394286138,
      "grad_norm": 3.4415132322601054,
      "learning_rate": 1.9231887492914126e-07,
      "loss": 0.8196,
      "step": 1536
    },
    {
      "epoch": 0.15194127968761584,
      "grad_norm": 3.739787890220457,
      "learning_rate": 1.9230656297620598e-07,
      "loss": 0.8061,
      "step": 1537
    },
    {
      "epoch": 0.15204013543237033,
      "grad_norm": 4.915846055808366,
      "learning_rate": 1.922942415585943e-07,
      "loss": 0.7406,
      "step": 1538
    },
    {
      "epoch": 0.1521389911771248,
      "grad_norm": 4.676451066090141,
      "learning_rate": 1.9228191067756955e-07,
      "loss": 0.8032,
      "step": 1539
    },
    {
      "epoch": 0.15223784692187925,
      "grad_norm": 3.2699820157528845,
      "learning_rate": 1.9226957033439613e-07,
      "loss": 0.6667,
      "step": 1540
    },
    {
      "epoch": 0.1523367026666337,
      "grad_norm": 8.0037158709571,
      "learning_rate": 1.9225722053033935e-07,
      "loss": 0.7365,
      "step": 1541
    },
    {
      "epoch": 0.15243555841138817,
      "grad_norm": 5.200785865474845,
      "learning_rate": 1.922448612666655e-07,
      "loss": 0.7661,
      "step": 1542
    },
    {
      "epoch": 0.15253441415614266,
      "grad_norm": 10.832139319705773,
      "learning_rate": 1.9223249254464182e-07,
      "loss": 0.6156,
      "step": 1543
    },
    {
      "epoch": 0.15263326990089712,
      "grad_norm": 26.746517272549674,
      "learning_rate": 1.9222011436553657e-07,
      "loss": 0.8328,
      "step": 1544
    },
    {
      "epoch": 0.15273212564565158,
      "grad_norm": 8.07201581679297,
      "learning_rate": 1.922077267306189e-07,
      "loss": 0.7071,
      "step": 1545
    },
    {
      "epoch": 0.15283098139040605,
      "grad_norm": 4.093842586030663,
      "learning_rate": 1.9219532964115903e-07,
      "loss": 0.7305,
      "step": 1546
    },
    {
      "epoch": 0.1529298371351605,
      "grad_norm": 4.04528849417612,
      "learning_rate": 1.921829230984281e-07,
      "loss": 0.7083,
      "step": 1547
    },
    {
      "epoch": 0.153028692879915,
      "grad_norm": 11.103611031027844,
      "learning_rate": 1.9217050710369819e-07,
      "loss": 0.5863,
      "step": 1548
    },
    {
      "epoch": 0.15312754862466946,
      "grad_norm": 4.510210195277667,
      "learning_rate": 1.921580816582424e-07,
      "loss": 0.8053,
      "step": 1549
    },
    {
      "epoch": 0.15322640436942392,
      "grad_norm": 3.6354685042425436,
      "learning_rate": 1.9214564676333473e-07,
      "loss": 0.7931,
      "step": 1550
    },
    {
      "epoch": 0.15332526011417838,
      "grad_norm": 5.2470931357245565,
      "learning_rate": 1.9213320242025024e-07,
      "loss": 0.7342,
      "step": 1551
    },
    {
      "epoch": 0.15342411585893284,
      "grad_norm": 4.4800205466208975,
      "learning_rate": 1.921207486302649e-07,
      "loss": 0.7264,
      "step": 1552
    },
    {
      "epoch": 0.15352297160368733,
      "grad_norm": 4.638358945962189,
      "learning_rate": 1.921082853946557e-07,
      "loss": 0.7836,
      "step": 1553
    },
    {
      "epoch": 0.1536218273484418,
      "grad_norm": 4.953240349748287,
      "learning_rate": 1.920958127147005e-07,
      "loss": 0.7612,
      "step": 1554
    },
    {
      "epoch": 0.15372068309319625,
      "grad_norm": 7.3873423954196245,
      "learning_rate": 1.920833305916782e-07,
      "loss": 0.7931,
      "step": 1555
    },
    {
      "epoch": 0.15381953883795071,
      "grad_norm": 6.4053236563326275,
      "learning_rate": 1.920708390268687e-07,
      "loss": 0.7211,
      "step": 1556
    },
    {
      "epoch": 0.15391839458270518,
      "grad_norm": 13.711925204296382,
      "learning_rate": 1.9205833802155277e-07,
      "loss": 0.7521,
      "step": 1557
    },
    {
      "epoch": 0.15401725032745966,
      "grad_norm": 3.8896622519805866,
      "learning_rate": 1.9204582757701227e-07,
      "loss": 0.7583,
      "step": 1558
    },
    {
      "epoch": 0.15411610607221413,
      "grad_norm": 3.9378413737509548,
      "learning_rate": 1.920333076945299e-07,
      "loss": 0.7857,
      "step": 1559
    },
    {
      "epoch": 0.1542149618169686,
      "grad_norm": 6.230453223577076,
      "learning_rate": 1.9202077837538942e-07,
      "loss": 0.7138,
      "step": 1560
    },
    {
      "epoch": 0.15431381756172305,
      "grad_norm": 5.595915633650135,
      "learning_rate": 1.920082396208755e-07,
      "loss": 0.8164,
      "step": 1561
    },
    {
      "epoch": 0.1544126733064775,
      "grad_norm": 4.335010447520082,
      "learning_rate": 1.9199569143227383e-07,
      "loss": 0.6981,
      "step": 1562
    },
    {
      "epoch": 0.154511529051232,
      "grad_norm": 4.837310489335801,
      "learning_rate": 1.9198313381087106e-07,
      "loss": 0.7022,
      "step": 1563
    },
    {
      "epoch": 0.15461038479598646,
      "grad_norm": 3.723694717252675,
      "learning_rate": 1.9197056675795473e-07,
      "loss": 0.8161,
      "step": 1564
    },
    {
      "epoch": 0.15470924054074092,
      "grad_norm": 3.0896086055418004,
      "learning_rate": 1.9195799027481345e-07,
      "loss": 0.6625,
      "step": 1565
    },
    {
      "epoch": 0.15480809628549538,
      "grad_norm": 5.706000179632502,
      "learning_rate": 1.9194540436273675e-07,
      "loss": 0.7478,
      "step": 1566
    },
    {
      "epoch": 0.15490695203024984,
      "grad_norm": 4.757753899528588,
      "learning_rate": 1.9193280902301515e-07,
      "loss": 0.6686,
      "step": 1567
    },
    {
      "epoch": 0.15500580777500433,
      "grad_norm": 7.555489303026925,
      "learning_rate": 1.9192020425694005e-07,
      "loss": 0.7547,
      "step": 1568
    },
    {
      "epoch": 0.1551046635197588,
      "grad_norm": 5.913143870141827,
      "learning_rate": 1.9190759006580392e-07,
      "loss": 0.6286,
      "step": 1569
    },
    {
      "epoch": 0.15520351926451326,
      "grad_norm": 3.9572313883594523,
      "learning_rate": 1.918949664509002e-07,
      "loss": 0.7156,
      "step": 1570
    },
    {
      "epoch": 0.15530237500926772,
      "grad_norm": 4.595135162380234,
      "learning_rate": 1.9188233341352317e-07,
      "loss": 0.6948,
      "step": 1571
    },
    {
      "epoch": 0.15540123075402218,
      "grad_norm": 3.5878831274198566,
      "learning_rate": 1.9186969095496824e-07,
      "loss": 0.7328,
      "step": 1572
    },
    {
      "epoch": 0.15550008649877667,
      "grad_norm": 3.5136528565110976,
      "learning_rate": 1.9185703907653165e-07,
      "loss": 0.7224,
      "step": 1573
    },
    {
      "epoch": 0.15559894224353113,
      "grad_norm": 6.66415885347148,
      "learning_rate": 1.9184437777951073e-07,
      "loss": 0.6755,
      "step": 1574
    },
    {
      "epoch": 0.1556977979882856,
      "grad_norm": 25.6448053053244,
      "learning_rate": 1.918317070652036e-07,
      "loss": 0.7521,
      "step": 1575
    },
    {
      "epoch": 0.15579665373304005,
      "grad_norm": 5.882500317835107,
      "learning_rate": 1.918190269349096e-07,
      "loss": 0.6761,
      "step": 1576
    },
    {
      "epoch": 0.15589550947779454,
      "grad_norm": 3.382166802221458,
      "learning_rate": 1.9180633738992878e-07,
      "loss": 0.7754,
      "step": 1577
    },
    {
      "epoch": 0.155994365222549,
      "grad_norm": 5.114046320581695,
      "learning_rate": 1.917936384315623e-07,
      "loss": 0.6556,
      "step": 1578
    },
    {
      "epoch": 0.15609322096730346,
      "grad_norm": 3.1270215643475088,
      "learning_rate": 1.9178093006111224e-07,
      "loss": 0.7537,
      "step": 1579
    },
    {
      "epoch": 0.15619207671205793,
      "grad_norm": 3.8883365808600017,
      "learning_rate": 1.9176821227988168e-07,
      "loss": 0.7954,
      "step": 1580
    },
    {
      "epoch": 0.1562909324568124,
      "grad_norm": 15.712779667378834,
      "learning_rate": 1.9175548508917462e-07,
      "loss": 0.6948,
      "step": 1581
    },
    {
      "epoch": 0.15638978820156688,
      "grad_norm": 6.563607788816714,
      "learning_rate": 1.9174274849029609e-07,
      "loss": 0.7526,
      "step": 1582
    },
    {
      "epoch": 0.15648864394632134,
      "grad_norm": 7.311172566957364,
      "learning_rate": 1.9173000248455194e-07,
      "loss": 0.7823,
      "step": 1583
    },
    {
      "epoch": 0.1565874996910758,
      "grad_norm": 3.7912846543923515,
      "learning_rate": 1.9171724707324918e-07,
      "loss": 0.7991,
      "step": 1584
    },
    {
      "epoch": 0.15668635543583026,
      "grad_norm": 6.994261486897858,
      "learning_rate": 1.9170448225769565e-07,
      "loss": 0.6608,
      "step": 1585
    },
    {
      "epoch": 0.15678521118058472,
      "grad_norm": 11.569582566961419,
      "learning_rate": 1.916917080392002e-07,
      "loss": 0.6759,
      "step": 1586
    },
    {
      "epoch": 0.1568840669253392,
      "grad_norm": 3.8445801879424657,
      "learning_rate": 1.9167892441907267e-07,
      "loss": 0.7497,
      "step": 1587
    },
    {
      "epoch": 0.15698292267009367,
      "grad_norm": 5.184131630721037,
      "learning_rate": 1.9166613139862378e-07,
      "loss": 0.735,
      "step": 1588
    },
    {
      "epoch": 0.15708177841484813,
      "grad_norm": 3.918117482127153,
      "learning_rate": 1.9165332897916527e-07,
      "loss": 0.7695,
      "step": 1589
    },
    {
      "epoch": 0.1571806341596026,
      "grad_norm": 3.8725308640037426,
      "learning_rate": 1.9164051716200986e-07,
      "loss": 0.7177,
      "step": 1590
    },
    {
      "epoch": 0.15727948990435706,
      "grad_norm": 8.465956009455192,
      "learning_rate": 1.916276959484712e-07,
      "loss": 0.7368,
      "step": 1591
    },
    {
      "epoch": 0.15737834564911155,
      "grad_norm": 10.21865027205561,
      "learning_rate": 1.9161486533986395e-07,
      "loss": 0.6694,
      "step": 1592
    },
    {
      "epoch": 0.157477201393866,
      "grad_norm": 3.5848187313447126,
      "learning_rate": 1.9160202533750365e-07,
      "loss": 0.6904,
      "step": 1593
    },
    {
      "epoch": 0.15757605713862047,
      "grad_norm": 4.13282462410451,
      "learning_rate": 1.915891759427069e-07,
      "loss": 0.6973,
      "step": 1594
    },
    {
      "epoch": 0.15767491288337493,
      "grad_norm": 3.2674453884883508,
      "learning_rate": 1.9157631715679118e-07,
      "loss": 0.6815,
      "step": 1595
    },
    {
      "epoch": 0.1577737686281294,
      "grad_norm": 9.801519868663762,
      "learning_rate": 1.91563448981075e-07,
      "loss": 0.6501,
      "step": 1596
    },
    {
      "epoch": 0.15787262437288388,
      "grad_norm": 12.151408104070974,
      "learning_rate": 1.9155057141687775e-07,
      "loss": 0.663,
      "step": 1597
    },
    {
      "epoch": 0.15797148011763834,
      "grad_norm": 4.795163950429591,
      "learning_rate": 1.9153768446551988e-07,
      "loss": 0.7831,
      "step": 1598
    },
    {
      "epoch": 0.1580703358623928,
      "grad_norm": 3.755113794429627,
      "learning_rate": 1.9152478812832275e-07,
      "loss": 0.6873,
      "step": 1599
    },
    {
      "epoch": 0.15816919160714726,
      "grad_norm": 3.2739515091938864,
      "learning_rate": 1.9151188240660866e-07,
      "loss": 0.7983,
      "step": 1600
    },
    {
      "epoch": 0.15826804735190173,
      "grad_norm": 3.902641688922389,
      "learning_rate": 1.9149896730170093e-07,
      "loss": 0.7201,
      "step": 1601
    },
    {
      "epoch": 0.15836690309665621,
      "grad_norm": 6.579513457620932,
      "learning_rate": 1.9148604281492385e-07,
      "loss": 0.6073,
      "step": 1602
    },
    {
      "epoch": 0.15846575884141068,
      "grad_norm": 4.172848937097197,
      "learning_rate": 1.9147310894760255e-07,
      "loss": 0.7022,
      "step": 1603
    },
    {
      "epoch": 0.15856461458616514,
      "grad_norm": 8.265687288687849,
      "learning_rate": 1.9146016570106327e-07,
      "loss": 0.6278,
      "step": 1604
    },
    {
      "epoch": 0.1586634703309196,
      "grad_norm": 5.051956218935498,
      "learning_rate": 1.9144721307663312e-07,
      "loss": 0.7233,
      "step": 1605
    },
    {
      "epoch": 0.15876232607567406,
      "grad_norm": 6.259930465096014,
      "learning_rate": 1.9143425107564023e-07,
      "loss": 0.7696,
      "step": 1606
    },
    {
      "epoch": 0.15886118182042855,
      "grad_norm": 2.7816100900528777,
      "learning_rate": 1.9142127969941362e-07,
      "loss": 0.6445,
      "step": 1607
    },
    {
      "epoch": 0.158960037565183,
      "grad_norm": 5.267354432560429,
      "learning_rate": 1.9140829894928333e-07,
      "loss": 0.79,
      "step": 1608
    },
    {
      "epoch": 0.15905889330993747,
      "grad_norm": 4.9039292217091575,
      "learning_rate": 1.9139530882658036e-07,
      "loss": 0.7976,
      "step": 1609
    },
    {
      "epoch": 0.15915774905469193,
      "grad_norm": 6.234482401868283,
      "learning_rate": 1.9138230933263666e-07,
      "loss": 0.8292,
      "step": 1610
    },
    {
      "epoch": 0.1592566047994464,
      "grad_norm": 9.876803248541771,
      "learning_rate": 1.9136930046878512e-07,
      "loss": 0.7732,
      "step": 1611
    },
    {
      "epoch": 0.15935546054420088,
      "grad_norm": 11.311176703503378,
      "learning_rate": 1.913562822363596e-07,
      "loss": 0.7693,
      "step": 1612
    },
    {
      "epoch": 0.15945431628895534,
      "grad_norm": 3.829842522592581,
      "learning_rate": 1.913432546366949e-07,
      "loss": 0.7361,
      "step": 1613
    },
    {
      "epoch": 0.1595531720337098,
      "grad_norm": 10.03421192101163,
      "learning_rate": 1.913302176711269e-07,
      "loss": 0.729,
      "step": 1614
    },
    {
      "epoch": 0.15965202777846427,
      "grad_norm": 3.3359775978053126,
      "learning_rate": 1.9131717134099228e-07,
      "loss": 0.7775,
      "step": 1615
    },
    {
      "epoch": 0.15975088352321876,
      "grad_norm": 3.4509300400139495,
      "learning_rate": 1.9130411564762876e-07,
      "loss": 0.7704,
      "step": 1616
    },
    {
      "epoch": 0.15984973926797322,
      "grad_norm": 24.308432247845968,
      "learning_rate": 1.91291050592375e-07,
      "loss": 0.6567,
      "step": 1617
    },
    {
      "epoch": 0.15994859501272768,
      "grad_norm": 4.877329188251267,
      "learning_rate": 1.9127797617657065e-07,
      "loss": 0.7461,
      "step": 1618
    },
    {
      "epoch": 0.16004745075748214,
      "grad_norm": 6.982641347396236,
      "learning_rate": 1.912648924015563e-07,
      "loss": 0.7273,
      "step": 1619
    },
    {
      "epoch": 0.1601463065022366,
      "grad_norm": 3.405945151404724,
      "learning_rate": 1.9125179926867346e-07,
      "loss": 0.6058,
      "step": 1620
    },
    {
      "epoch": 0.1602451622469911,
      "grad_norm": 4.023510961444016,
      "learning_rate": 1.9123869677926467e-07,
      "loss": 0.6563,
      "step": 1621
    },
    {
      "epoch": 0.16034401799174555,
      "grad_norm": 3.006902563397172,
      "learning_rate": 1.9122558493467338e-07,
      "loss": 0.7405,
      "step": 1622
    },
    {
      "epoch": 0.1604428737365,
      "grad_norm": 4.585723275294254,
      "learning_rate": 1.9121246373624405e-07,
      "loss": 0.8422,
      "step": 1623
    },
    {
      "epoch": 0.16054172948125447,
      "grad_norm": 4.109780986344828,
      "learning_rate": 1.9119933318532203e-07,
      "loss": 0.7552,
      "step": 1624
    },
    {
      "epoch": 0.16064058522600894,
      "grad_norm": 4.577524366477109,
      "learning_rate": 1.9118619328325368e-07,
      "loss": 0.6955,
      "step": 1625
    },
    {
      "epoch": 0.16073944097076343,
      "grad_norm": 2.989832816105517,
      "learning_rate": 1.911730440313863e-07,
      "loss": 0.7245,
      "step": 1626
    },
    {
      "epoch": 0.1608382967155179,
      "grad_norm": 5.093778944072761,
      "learning_rate": 1.9115988543106815e-07,
      "loss": 0.7401,
      "step": 1627
    },
    {
      "epoch": 0.16093715246027235,
      "grad_norm": 6.3798775989768455,
      "learning_rate": 1.9114671748364847e-07,
      "loss": 0.7389,
      "step": 1628
    },
    {
      "epoch": 0.1610360082050268,
      "grad_norm": 4.733875121126953,
      "learning_rate": 1.911335401904774e-07,
      "loss": 0.66,
      "step": 1629
    },
    {
      "epoch": 0.16113486394978127,
      "grad_norm": 5.829348903436602,
      "learning_rate": 1.911203535529061e-07,
      "loss": 0.6849,
      "step": 1630
    },
    {
      "epoch": 0.16123371969453576,
      "grad_norm": 3.404188709889018,
      "learning_rate": 1.9110715757228667e-07,
      "loss": 0.6923,
      "step": 1631
    },
    {
      "epoch": 0.16133257543929022,
      "grad_norm": 7.433723893558098,
      "learning_rate": 1.9109395224997217e-07,
      "loss": 0.653,
      "step": 1632
    },
    {
      "epoch": 0.16143143118404468,
      "grad_norm": 4.175206053479436,
      "learning_rate": 1.9108073758731657e-07,
      "loss": 0.6795,
      "step": 1633
    },
    {
      "epoch": 0.16153028692879914,
      "grad_norm": 3.5049837418927137,
      "learning_rate": 1.9106751358567488e-07,
      "loss": 0.8686,
      "step": 1634
    },
    {
      "epoch": 0.1616291426735536,
      "grad_norm": 3.54641786603234,
      "learning_rate": 1.9105428024640296e-07,
      "loss": 0.7185,
      "step": 1635
    },
    {
      "epoch": 0.1617279984183081,
      "grad_norm": 11.119653407264087,
      "learning_rate": 1.910410375708578e-07,
      "loss": 0.7349,
      "step": 1636
    },
    {
      "epoch": 0.16182685416306256,
      "grad_norm": 4.516333182072062,
      "learning_rate": 1.910277855603972e-07,
      "loss": 0.7188,
      "step": 1637
    },
    {
      "epoch": 0.16192570990781702,
      "grad_norm": 4.791356374985866,
      "learning_rate": 1.910145242163799e-07,
      "loss": 0.6664,
      "step": 1638
    },
    {
      "epoch": 0.16202456565257148,
      "grad_norm": 6.141424162927891,
      "learning_rate": 1.9100125354016577e-07,
      "loss": 0.7269,
      "step": 1639
    },
    {
      "epoch": 0.16212342139732594,
      "grad_norm": 3.910480494693417,
      "learning_rate": 1.909879735331154e-07,
      "loss": 0.7225,
      "step": 1640
    },
    {
      "epoch": 0.16222227714208043,
      "grad_norm": 3.367096064404229,
      "learning_rate": 1.9097468419659052e-07,
      "loss": 0.778,
      "step": 1641
    },
    {
      "epoch": 0.1623211328868349,
      "grad_norm": 5.080913062272009,
      "learning_rate": 1.9096138553195374e-07,
      "loss": 0.7466,
      "step": 1642
    },
    {
      "epoch": 0.16241998863158935,
      "grad_norm": 31.212206669511,
      "learning_rate": 1.9094807754056866e-07,
      "loss": 0.8658,
      "step": 1643
    },
    {
      "epoch": 0.1625188443763438,
      "grad_norm": 7.527090381762391,
      "learning_rate": 1.909347602237998e-07,
      "loss": 0.6421,
      "step": 1644
    },
    {
      "epoch": 0.16261770012109827,
      "grad_norm": 3.5186531049315066,
      "learning_rate": 1.9092143358301265e-07,
      "loss": 0.7286,
      "step": 1645
    },
    {
      "epoch": 0.16271655586585276,
      "grad_norm": 4.066301537608991,
      "learning_rate": 1.909080976195737e-07,
      "loss": 0.8669,
      "step": 1646
    },
    {
      "epoch": 0.16281541161060722,
      "grad_norm": 3.375247822900512,
      "learning_rate": 1.908947523348503e-07,
      "loss": 0.7377,
      "step": 1647
    },
    {
      "epoch": 0.16291426735536169,
      "grad_norm": 3.8126304271095477,
      "learning_rate": 1.9088139773021084e-07,
      "loss": 0.8071,
      "step": 1648
    },
    {
      "epoch": 0.16301312310011615,
      "grad_norm": 3.5703729009065226,
      "learning_rate": 1.9086803380702464e-07,
      "loss": 0.6445,
      "step": 1649
    },
    {
      "epoch": 0.1631119788448706,
      "grad_norm": 4.705917066318424,
      "learning_rate": 1.90854660566662e-07,
      "loss": 0.7642,
      "step": 1650
    },
    {
      "epoch": 0.1632108345896251,
      "grad_norm": 3.5353365352669788,
      "learning_rate": 1.908412780104941e-07,
      "loss": 0.7656,
      "step": 1651
    },
    {
      "epoch": 0.16330969033437956,
      "grad_norm": 4.916287267960891,
      "learning_rate": 1.9082788613989313e-07,
      "loss": 0.7828,
      "step": 1652
    },
    {
      "epoch": 0.16340854607913402,
      "grad_norm": 3.724085070875087,
      "learning_rate": 1.9081448495623226e-07,
      "loss": 0.6242,
      "step": 1653
    },
    {
      "epoch": 0.16350740182388848,
      "grad_norm": 4.4813730664795814,
      "learning_rate": 1.9080107446088555e-07,
      "loss": 0.7808,
      "step": 1654
    },
    {
      "epoch": 0.16360625756864297,
      "grad_norm": 3.8439357535012357,
      "learning_rate": 1.9078765465522807e-07,
      "loss": 0.6864,
      "step": 1655
    },
    {
      "epoch": 0.16370511331339743,
      "grad_norm": 3.1149160924169705,
      "learning_rate": 1.9077422554063586e-07,
      "loss": 0.7858,
      "step": 1656
    },
    {
      "epoch": 0.1638039690581519,
      "grad_norm": 53.43499734830409,
      "learning_rate": 1.9076078711848577e-07,
      "loss": 0.7309,
      "step": 1657
    },
    {
      "epoch": 0.16390282480290636,
      "grad_norm": 3.0515389376220394,
      "learning_rate": 1.9074733939015578e-07,
      "loss": 0.6818,
      "step": 1658
    },
    {
      "epoch": 0.16400168054766082,
      "grad_norm": 4.096741442756604,
      "learning_rate": 1.907338823570248e-07,
      "loss": 0.7046,
      "step": 1659
    },
    {
      "epoch": 0.1641005362924153,
      "grad_norm": 4.028177595552836,
      "learning_rate": 1.9072041602047258e-07,
      "loss": 0.6924,
      "step": 1660
    },
    {
      "epoch": 0.16419939203716977,
      "grad_norm": 3.8554340442166803,
      "learning_rate": 1.9070694038187992e-07,
      "loss": 0.8177,
      "step": 1661
    },
    {
      "epoch": 0.16429824778192423,
      "grad_norm": 4.496679320512058,
      "learning_rate": 1.9069345544262854e-07,
      "loss": 0.6496,
      "step": 1662
    },
    {
      "epoch": 0.1643971035266787,
      "grad_norm": 7.816519393586867,
      "learning_rate": 1.9067996120410111e-07,
      "loss": 0.7417,
      "step": 1663
    },
    {
      "epoch": 0.16449595927143315,
      "grad_norm": 10.384589476659524,
      "learning_rate": 1.906664576676813e-07,
      "loss": 0.7162,
      "step": 1664
    },
    {
      "epoch": 0.16459481501618764,
      "grad_norm": 4.361680122063604,
      "learning_rate": 1.906529448347537e-07,
      "loss": 0.7125,
      "step": 1665
    },
    {
      "epoch": 0.1646936707609421,
      "grad_norm": 4.043313096527939,
      "learning_rate": 1.9063942270670382e-07,
      "loss": 0.7197,
      "step": 1666
    },
    {
      "epoch": 0.16479252650569656,
      "grad_norm": 4.406479883673326,
      "learning_rate": 1.9062589128491814e-07,
      "loss": 0.7829,
      "step": 1667
    },
    {
      "epoch": 0.16489138225045102,
      "grad_norm": 2.871680622149349,
      "learning_rate": 1.9061235057078418e-07,
      "loss": 0.6711,
      "step": 1668
    },
    {
      "epoch": 0.16499023799520549,
      "grad_norm": 10.393191466386375,
      "learning_rate": 1.9059880056569025e-07,
      "loss": 0.7484,
      "step": 1669
    },
    {
      "epoch": 0.16508909373995997,
      "grad_norm": 2.981156093786772,
      "learning_rate": 1.9058524127102578e-07,
      "loss": 0.762,
      "step": 1670
    },
    {
      "epoch": 0.16518794948471444,
      "grad_norm": 7.815663874880071,
      "learning_rate": 1.9057167268818102e-07,
      "loss": 0.8046,
      "step": 1671
    },
    {
      "epoch": 0.1652868052294689,
      "grad_norm": 4.022937268972966,
      "learning_rate": 1.9055809481854728e-07,
      "loss": 0.7215,
      "step": 1672
    },
    {
      "epoch": 0.16538566097422336,
      "grad_norm": 6.090162475797414,
      "learning_rate": 1.9054450766351672e-07,
      "loss": 0.6999,
      "step": 1673
    },
    {
      "epoch": 0.16548451671897782,
      "grad_norm": 5.238496981247826,
      "learning_rate": 1.9053091122448252e-07,
      "loss": 0.7162,
      "step": 1674
    },
    {
      "epoch": 0.1655833724637323,
      "grad_norm": 7.481448856154265,
      "learning_rate": 1.905173055028388e-07,
      "loss": 0.7096,
      "step": 1675
    },
    {
      "epoch": 0.16568222820848677,
      "grad_norm": 4.129954325612517,
      "learning_rate": 1.9050369049998066e-07,
      "loss": 0.7445,
      "step": 1676
    },
    {
      "epoch": 0.16578108395324123,
      "grad_norm": 4.706673828854298,
      "learning_rate": 1.9049006621730406e-07,
      "loss": 0.7204,
      "step": 1677
    },
    {
      "epoch": 0.1658799396979957,
      "grad_norm": 3.065198480265075,
      "learning_rate": 1.9047643265620598e-07,
      "loss": 0.6656,
      "step": 1678
    },
    {
      "epoch": 0.16597879544275015,
      "grad_norm": 3.4783977923125278,
      "learning_rate": 1.9046278981808436e-07,
      "loss": 0.7648,
      "step": 1679
    },
    {
      "epoch": 0.16607765118750464,
      "grad_norm": 4.958889508187133,
      "learning_rate": 1.9044913770433808e-07,
      "loss": 0.8153,
      "step": 1680
    },
    {
      "epoch": 0.1661765069322591,
      "grad_norm": 3.614700163121188,
      "learning_rate": 1.9043547631636694e-07,
      "loss": 0.8014,
      "step": 1681
    },
    {
      "epoch": 0.16627536267701357,
      "grad_norm": 11.926934499565979,
      "learning_rate": 1.904218056555717e-07,
      "loss": 0.6898,
      "step": 1682
    },
    {
      "epoch": 0.16637421842176803,
      "grad_norm": 4.73634351378157,
      "learning_rate": 1.9040812572335412e-07,
      "loss": 0.6701,
      "step": 1683
    },
    {
      "epoch": 0.1664730741665225,
      "grad_norm": 6.727120221114989,
      "learning_rate": 1.9039443652111687e-07,
      "loss": 0.5657,
      "step": 1684
    },
    {
      "epoch": 0.16657192991127698,
      "grad_norm": 5.018045517313606,
      "learning_rate": 1.9038073805026356e-07,
      "loss": 0.7971,
      "step": 1685
    },
    {
      "epoch": 0.16667078565603144,
      "grad_norm": 14.667402363682845,
      "learning_rate": 1.9036703031219877e-07,
      "loss": 0.6563,
      "step": 1686
    },
    {
      "epoch": 0.1667696414007859,
      "grad_norm": 4.215522359262666,
      "learning_rate": 1.9035331330832802e-07,
      "loss": 0.6984,
      "step": 1687
    },
    {
      "epoch": 0.16686849714554036,
      "grad_norm": 4.5584260587053755,
      "learning_rate": 1.9033958704005778e-07,
      "loss": 0.718,
      "step": 1688
    },
    {
      "epoch": 0.16696735289029482,
      "grad_norm": 5.023904749410402,
      "learning_rate": 1.9032585150879553e-07,
      "loss": 0.7272,
      "step": 1689
    },
    {
      "epoch": 0.1670662086350493,
      "grad_norm": 6.0462743504777645,
      "learning_rate": 1.9031210671594963e-07,
      "loss": 0.6987,
      "step": 1690
    },
    {
      "epoch": 0.16716506437980377,
      "grad_norm": 10.444076777045005,
      "learning_rate": 1.9029835266292934e-07,
      "loss": 0.7779,
      "step": 1691
    },
    {
      "epoch": 0.16726392012455824,
      "grad_norm": 142.38434698405595,
      "learning_rate": 1.9028458935114498e-07,
      "loss": 0.7066,
      "step": 1692
    },
    {
      "epoch": 0.1673627758693127,
      "grad_norm": 3.706237944886247,
      "learning_rate": 1.9027081678200778e-07,
      "loss": 0.8256,
      "step": 1693
    },
    {
      "epoch": 0.16746163161406719,
      "grad_norm": 3.5426956297858885,
      "learning_rate": 1.902570349569299e-07,
      "loss": 0.6673,
      "step": 1694
    },
    {
      "epoch": 0.16756048735882165,
      "grad_norm": 5.418971558482785,
      "learning_rate": 1.9024324387732449e-07,
      "loss": 0.8023,
      "step": 1695
    },
    {
      "epoch": 0.1676593431035761,
      "grad_norm": 4.342970067889736,
      "learning_rate": 1.9022944354460557e-07,
      "loss": 0.7473,
      "step": 1696
    },
    {
      "epoch": 0.16775819884833057,
      "grad_norm": 2.876493019350703,
      "learning_rate": 1.9021563396018822e-07,
      "loss": 0.8146,
      "step": 1697
    },
    {
      "epoch": 0.16785705459308503,
      "grad_norm": 3.4083648875047876,
      "learning_rate": 1.9020181512548837e-07,
      "loss": 0.6976,
      "step": 1698
    },
    {
      "epoch": 0.16795591033783952,
      "grad_norm": 6.628650821173191,
      "learning_rate": 1.9018798704192296e-07,
      "loss": 0.6328,
      "step": 1699
    },
    {
      "epoch": 0.16805476608259398,
      "grad_norm": 3.643165201023311,
      "learning_rate": 1.9017414971090983e-07,
      "loss": 0.7643,
      "step": 1700
    },
    {
      "epoch": 0.16815362182734844,
      "grad_norm": 3.851485684197734,
      "learning_rate": 1.901603031338678e-07,
      "loss": 0.8556,
      "step": 1701
    },
    {
      "epoch": 0.1682524775721029,
      "grad_norm": 4.447041682571627,
      "learning_rate": 1.9014644731221668e-07,
      "loss": 0.7687,
      "step": 1702
    },
    {
      "epoch": 0.16835133331685737,
      "grad_norm": 4.1854428411498885,
      "learning_rate": 1.9013258224737713e-07,
      "loss": 0.702,
      "step": 1703
    },
    {
      "epoch": 0.16845018906161185,
      "grad_norm": 3.9803373773992363,
      "learning_rate": 1.901187079407708e-07,
      "loss": 0.6336,
      "step": 1704
    },
    {
      "epoch": 0.16854904480636632,
      "grad_norm": 4.2805448524660505,
      "learning_rate": 1.901048243938203e-07,
      "loss": 0.7376,
      "step": 1705
    },
    {
      "epoch": 0.16864790055112078,
      "grad_norm": 6.631879612240396,
      "learning_rate": 1.9009093160794923e-07,
      "loss": 0.7085,
      "step": 1706
    },
    {
      "epoch": 0.16874675629587524,
      "grad_norm": 6.679649169622515,
      "learning_rate": 1.9007702958458208e-07,
      "loss": 0.821,
      "step": 1707
    },
    {
      "epoch": 0.1688456120406297,
      "grad_norm": 4.168141591967967,
      "learning_rate": 1.9006311832514424e-07,
      "loss": 0.6918,
      "step": 1708
    },
    {
      "epoch": 0.1689444677853842,
      "grad_norm": 4.4554496357926086,
      "learning_rate": 1.9004919783106215e-07,
      "loss": 0.6644,
      "step": 1709
    },
    {
      "epoch": 0.16904332353013865,
      "grad_norm": 3.2773324556179135,
      "learning_rate": 1.9003526810376314e-07,
      "loss": 0.6859,
      "step": 1710
    },
    {
      "epoch": 0.1691421792748931,
      "grad_norm": 4.435898621621379,
      "learning_rate": 1.900213291446755e-07,
      "loss": 0.7286,
      "step": 1711
    },
    {
      "epoch": 0.16924103501964757,
      "grad_norm": 4.600776536740985,
      "learning_rate": 1.9000738095522845e-07,
      "loss": 0.7366,
      "step": 1712
    },
    {
      "epoch": 0.16933989076440203,
      "grad_norm": 3.000091884373794,
      "learning_rate": 1.8999342353685218e-07,
      "loss": 0.7813,
      "step": 1713
    },
    {
      "epoch": 0.16943874650915652,
      "grad_norm": 4.535331652119621,
      "learning_rate": 1.8997945689097789e-07,
      "loss": 0.8168,
      "step": 1714
    },
    {
      "epoch": 0.16953760225391099,
      "grad_norm": 7.068950893937681,
      "learning_rate": 1.899654810190375e-07,
      "loss": 0.6461,
      "step": 1715
    },
    {
      "epoch": 0.16963645799866545,
      "grad_norm": 3.120734608640814,
      "learning_rate": 1.8995149592246417e-07,
      "loss": 0.7773,
      "step": 1716
    },
    {
      "epoch": 0.1697353137434199,
      "grad_norm": 4.412998530041257,
      "learning_rate": 1.899375016026918e-07,
      "loss": 0.9191,
      "step": 1717
    },
    {
      "epoch": 0.16983416948817437,
      "grad_norm": 3.43073580285405,
      "learning_rate": 1.899234980611553e-07,
      "loss": 0.7317,
      "step": 1718
    },
    {
      "epoch": 0.16993302523292886,
      "grad_norm": 4.233550258650561,
      "learning_rate": 1.8990948529929054e-07,
      "loss": 0.6977,
      "step": 1719
    },
    {
      "epoch": 0.17003188097768332,
      "grad_norm": 4.907745067852664,
      "learning_rate": 1.8989546331853434e-07,
      "loss": 0.7754,
      "step": 1720
    },
    {
      "epoch": 0.17013073672243778,
      "grad_norm": 6.51290228630269,
      "learning_rate": 1.898814321203244e-07,
      "loss": 0.832,
      "step": 1721
    },
    {
      "epoch": 0.17022959246719224,
      "grad_norm": 3.839650625038715,
      "learning_rate": 1.8986739170609945e-07,
      "loss": 0.782,
      "step": 1722
    },
    {
      "epoch": 0.1703284482119467,
      "grad_norm": 4.930498419663581,
      "learning_rate": 1.8985334207729914e-07,
      "loss": 0.7026,
      "step": 1723
    },
    {
      "epoch": 0.1704273039567012,
      "grad_norm": 3.955689719599107,
      "learning_rate": 1.89839283235364e-07,
      "loss": 0.7171,
      "step": 1724
    },
    {
      "epoch": 0.17052615970145565,
      "grad_norm": 2.8878718030475246,
      "learning_rate": 1.898252151817356e-07,
      "loss": 0.6674,
      "step": 1725
    },
    {
      "epoch": 0.17062501544621012,
      "grad_norm": 5.489335130573389,
      "learning_rate": 1.8981113791785641e-07,
      "loss": 0.6934,
      "step": 1726
    },
    {
      "epoch": 0.17072387119096458,
      "grad_norm": 7.928674863071406,
      "learning_rate": 1.897970514451698e-07,
      "loss": 0.7331,
      "step": 1727
    },
    {
      "epoch": 0.17082272693571904,
      "grad_norm": 4.353054313942994,
      "learning_rate": 1.8978295576512017e-07,
      "loss": 0.7606,
      "step": 1728
    },
    {
      "epoch": 0.17092158268047353,
      "grad_norm": 5.006464359075778,
      "learning_rate": 1.8976885087915286e-07,
      "loss": 0.786,
      "step": 1729
    },
    {
      "epoch": 0.171020438425228,
      "grad_norm": 4.357417179087538,
      "learning_rate": 1.8975473678871406e-07,
      "loss": 0.7185,
      "step": 1730
    },
    {
      "epoch": 0.17111929416998245,
      "grad_norm": 10.81384220176505,
      "learning_rate": 1.8974061349525096e-07,
      "loss": 0.7901,
      "step": 1731
    },
    {
      "epoch": 0.1712181499147369,
      "grad_norm": 4.406026358238695,
      "learning_rate": 1.8972648100021172e-07,
      "loss": 0.7734,
      "step": 1732
    },
    {
      "epoch": 0.1713170056594914,
      "grad_norm": 16.202336011277236,
      "learning_rate": 1.897123393050454e-07,
      "loss": 0.8214,
      "step": 1733
    },
    {
      "epoch": 0.17141586140424586,
      "grad_norm": 4.080143971755174,
      "learning_rate": 1.8969818841120207e-07,
      "loss": 0.7314,
      "step": 1734
    },
    {
      "epoch": 0.17151471714900032,
      "grad_norm": 4.171856302880728,
      "learning_rate": 1.8968402832013265e-07,
      "loss": 0.6863,
      "step": 1735
    },
    {
      "epoch": 0.17161357289375478,
      "grad_norm": 4.84946915449895,
      "learning_rate": 1.8966985903328903e-07,
      "loss": 0.7238,
      "step": 1736
    },
    {
      "epoch": 0.17171242863850925,
      "grad_norm": 6.441182281391494,
      "learning_rate": 1.896556805521241e-07,
      "loss": 0.7789,
      "step": 1737
    },
    {
      "epoch": 0.17181128438326373,
      "grad_norm": 8.291743822086435,
      "learning_rate": 1.8964149287809167e-07,
      "loss": 0.8129,
      "step": 1738
    },
    {
      "epoch": 0.1719101401280182,
      "grad_norm": 4.70383052545174,
      "learning_rate": 1.896272960126464e-07,
      "loss": 0.7125,
      "step": 1739
    },
    {
      "epoch": 0.17200899587277266,
      "grad_norm": 5.9811051893954374,
      "learning_rate": 1.8961308995724405e-07,
      "loss": 0.7708,
      "step": 1740
    },
    {
      "epoch": 0.17210785161752712,
      "grad_norm": 13.069501563222797,
      "learning_rate": 1.895988747133412e-07,
      "loss": 0.6855,
      "step": 1741
    },
    {
      "epoch": 0.17220670736228158,
      "grad_norm": 3.4705187303408316,
      "learning_rate": 1.8958465028239545e-07,
      "loss": 0.7314,
      "step": 1742
    },
    {
      "epoch": 0.17230556310703607,
      "grad_norm": 4.901384492946249,
      "learning_rate": 1.8957041666586528e-07,
      "loss": 0.7463,
      "step": 1743
    },
    {
      "epoch": 0.17240441885179053,
      "grad_norm": 6.536802538689493,
      "learning_rate": 1.895561738652101e-07,
      "loss": 0.6924,
      "step": 1744
    },
    {
      "epoch": 0.172503274596545,
      "grad_norm": 3.974540928351649,
      "learning_rate": 1.8954192188189039e-07,
      "loss": 0.7854,
      "step": 1745
    },
    {
      "epoch": 0.17260213034129945,
      "grad_norm": 4.336895499611506,
      "learning_rate": 1.8952766071736736e-07,
      "loss": 0.6423,
      "step": 1746
    },
    {
      "epoch": 0.17270098608605391,
      "grad_norm": 6.373389861461393,
      "learning_rate": 1.895133903731034e-07,
      "loss": 0.6428,
      "step": 1747
    },
    {
      "epoch": 0.1727998418308084,
      "grad_norm": 8.206593870199931,
      "learning_rate": 1.8949911085056166e-07,
      "loss": 0.665,
      "step": 1748
    },
    {
      "epoch": 0.17289869757556287,
      "grad_norm": 3.165081106294251,
      "learning_rate": 1.8948482215120628e-07,
      "loss": 0.7252,
      "step": 1749
    },
    {
      "epoch": 0.17299755332031733,
      "grad_norm": 3.578501599162503,
      "learning_rate": 1.894705242765024e-07,
      "loss": 0.7664,
      "step": 1750
    },
    {
      "epoch": 0.1730964090650718,
      "grad_norm": 3.402998652649381,
      "learning_rate": 1.8945621722791607e-07,
      "loss": 0.7624,
      "step": 1751
    },
    {
      "epoch": 0.17319526480982625,
      "grad_norm": 4.638525594784221,
      "learning_rate": 1.894419010069142e-07,
      "loss": 0.7252,
      "step": 1752
    },
    {
      "epoch": 0.17329412055458074,
      "grad_norm": 3.042258569106757,
      "learning_rate": 1.8942757561496475e-07,
      "loss": 0.7808,
      "step": 1753
    },
    {
      "epoch": 0.1733929762993352,
      "grad_norm": 7.542922716157105,
      "learning_rate": 1.894132410535366e-07,
      "loss": 0.7284,
      "step": 1754
    },
    {
      "epoch": 0.17349183204408966,
      "grad_norm": 4.685950028445863,
      "learning_rate": 1.8939889732409949e-07,
      "loss": 0.7875,
      "step": 1755
    },
    {
      "epoch": 0.17359068778884412,
      "grad_norm": 5.658840237884323,
      "learning_rate": 1.8938454442812416e-07,
      "loss": 0.8001,
      "step": 1756
    },
    {
      "epoch": 0.17368954353359858,
      "grad_norm": 8.07060851496577,
      "learning_rate": 1.8937018236708235e-07,
      "loss": 0.6387,
      "step": 1757
    },
    {
      "epoch": 0.17378839927835307,
      "grad_norm": 8.830843705495276,
      "learning_rate": 1.8935581114244666e-07,
      "loss": 0.6546,
      "step": 1758
    },
    {
      "epoch": 0.17388725502310753,
      "grad_norm": 4.272025947799766,
      "learning_rate": 1.893414307556906e-07,
      "loss": 0.786,
      "step": 1759
    },
    {
      "epoch": 0.173986110767862,
      "grad_norm": 4.61647122642708,
      "learning_rate": 1.8932704120828869e-07,
      "loss": 0.7276,
      "step": 1760
    },
    {
      "epoch": 0.17408496651261646,
      "grad_norm": 3.9374874255778627,
      "learning_rate": 1.8931264250171638e-07,
      "loss": 0.6783,
      "step": 1761
    },
    {
      "epoch": 0.17418382225737092,
      "grad_norm": 4.048144042566202,
      "learning_rate": 1.8929823463745003e-07,
      "loss": 0.6552,
      "step": 1762
    },
    {
      "epoch": 0.1742826780021254,
      "grad_norm": 4.2539613749152885,
      "learning_rate": 1.8928381761696698e-07,
      "loss": 0.6568,
      "step": 1763
    },
    {
      "epoch": 0.17438153374687987,
      "grad_norm": 4.458867409695586,
      "learning_rate": 1.892693914417454e-07,
      "loss": 0.6735,
      "step": 1764
    },
    {
      "epoch": 0.17448038949163433,
      "grad_norm": 3.010595195350182,
      "learning_rate": 1.8925495611326462e-07,
      "loss": 0.6897,
      "step": 1765
    },
    {
      "epoch": 0.1745792452363888,
      "grad_norm": 22.852024670608735,
      "learning_rate": 1.8924051163300466e-07,
      "loss": 0.771,
      "step": 1766
    },
    {
      "epoch": 0.17467810098114328,
      "grad_norm": 4.016872586373533,
      "learning_rate": 1.892260580024466e-07,
      "loss": 0.8302,
      "step": 1767
    },
    {
      "epoch": 0.17477695672589774,
      "grad_norm": 4.500413307954438,
      "learning_rate": 1.892115952230725e-07,
      "loss": 0.7506,
      "step": 1768
    },
    {
      "epoch": 0.1748758124706522,
      "grad_norm": 3.8781098857454186,
      "learning_rate": 1.8919712329636528e-07,
      "loss": 0.6943,
      "step": 1769
    },
    {
      "epoch": 0.17497466821540666,
      "grad_norm": 5.629554095566065,
      "learning_rate": 1.891826422238088e-07,
      "loss": 0.7075,
      "step": 1770
    },
    {
      "epoch": 0.17507352396016113,
      "grad_norm": 11.310222685857072,
      "learning_rate": 1.891681520068879e-07,
      "loss": 0.712,
      "step": 1771
    },
    {
      "epoch": 0.17517237970491562,
      "grad_norm": 4.060853691527991,
      "learning_rate": 1.8915365264708831e-07,
      "loss": 0.7665,
      "step": 1772
    },
    {
      "epoch": 0.17527123544967008,
      "grad_norm": 4.62303655426697,
      "learning_rate": 1.891391441458968e-07,
      "loss": 0.6926,
      "step": 1773
    },
    {
      "epoch": 0.17537009119442454,
      "grad_norm": 3.8642200552047528,
      "learning_rate": 1.8912462650480092e-07,
      "loss": 0.7449,
      "step": 1774
    },
    {
      "epoch": 0.175468946939179,
      "grad_norm": 3.901746644457528,
      "learning_rate": 1.891100997252893e-07,
      "loss": 0.6979,
      "step": 1775
    },
    {
      "epoch": 0.17556780268393346,
      "grad_norm": 3.8068463817214844,
      "learning_rate": 1.890955638088514e-07,
      "loss": 0.8209,
      "step": 1776
    },
    {
      "epoch": 0.17566665842868795,
      "grad_norm": 3.2587877828065173,
      "learning_rate": 1.8908101875697766e-07,
      "loss": 0.7189,
      "step": 1777
    },
    {
      "epoch": 0.1757655141734424,
      "grad_norm": 3.9959476291139544,
      "learning_rate": 1.890664645711595e-07,
      "loss": 0.7277,
      "step": 1778
    },
    {
      "epoch": 0.17586436991819687,
      "grad_norm": 5.5307517045140155,
      "learning_rate": 1.8905190125288923e-07,
      "loss": 0.8285,
      "step": 1779
    },
    {
      "epoch": 0.17596322566295133,
      "grad_norm": 4.399177816817838,
      "learning_rate": 1.890373288036601e-07,
      "loss": 0.8217,
      "step": 1780
    },
    {
      "epoch": 0.1760620814077058,
      "grad_norm": 6.839882407150985,
      "learning_rate": 1.8902274722496628e-07,
      "loss": 0.8515,
      "step": 1781
    },
    {
      "epoch": 0.17616093715246028,
      "grad_norm": 3.836977304198438,
      "learning_rate": 1.890081565183029e-07,
      "loss": 0.7943,
      "step": 1782
    },
    {
      "epoch": 0.17625979289721475,
      "grad_norm": 3.555806399108246,
      "learning_rate": 1.8899355668516605e-07,
      "loss": 0.6454,
      "step": 1783
    },
    {
      "epoch": 0.1763586486419692,
      "grad_norm": 3.8563238025197486,
      "learning_rate": 1.8897894772705272e-07,
      "loss": 0.6754,
      "step": 1784
    },
    {
      "epoch": 0.17645750438672367,
      "grad_norm": 6.27192475326568,
      "learning_rate": 1.8896432964546084e-07,
      "loss": 0.7594,
      "step": 1785
    },
    {
      "epoch": 0.17655636013147813,
      "grad_norm": 3.8325504702393802,
      "learning_rate": 1.889497024418892e-07,
      "loss": 0.7975,
      "step": 1786
    },
    {
      "epoch": 0.17665521587623262,
      "grad_norm": 4.365220680776323,
      "learning_rate": 1.8893506611783776e-07,
      "loss": 0.8145,
      "step": 1787
    },
    {
      "epoch": 0.17675407162098708,
      "grad_norm": 7.20686772467932,
      "learning_rate": 1.8892042067480713e-07,
      "loss": 0.6858,
      "step": 1788
    },
    {
      "epoch": 0.17685292736574154,
      "grad_norm": 4.279943063461015,
      "learning_rate": 1.8890576611429906e-07,
      "loss": 0.7093,
      "step": 1789
    },
    {
      "epoch": 0.176951783110496,
      "grad_norm": 4.731685967918578,
      "learning_rate": 1.888911024378161e-07,
      "loss": 0.7318,
      "step": 1790
    },
    {
      "epoch": 0.17705063885525046,
      "grad_norm": 11.9445551856101,
      "learning_rate": 1.888764296468618e-07,
      "loss": 0.7626,
      "step": 1791
    },
    {
      "epoch": 0.17714949460000495,
      "grad_norm": 3.966015145645517,
      "learning_rate": 1.888617477429407e-07,
      "loss": 0.7392,
      "step": 1792
    },
    {
      "epoch": 0.17724835034475941,
      "grad_norm": 5.053992575307353,
      "learning_rate": 1.8884705672755817e-07,
      "loss": 0.863,
      "step": 1793
    },
    {
      "epoch": 0.17734720608951388,
      "grad_norm": 3.629841627651401,
      "learning_rate": 1.8883235660222052e-07,
      "loss": 0.7258,
      "step": 1794
    },
    {
      "epoch": 0.17744606183426834,
      "grad_norm": 4.023209819926416,
      "learning_rate": 1.8881764736843512e-07,
      "loss": 0.6932,
      "step": 1795
    },
    {
      "epoch": 0.1775449175790228,
      "grad_norm": 4.577415029841412,
      "learning_rate": 1.8880292902771008e-07,
      "loss": 0.7128,
      "step": 1796
    },
    {
      "epoch": 0.1776437733237773,
      "grad_norm": 3.451597529321051,
      "learning_rate": 1.8878820158155463e-07,
      "loss": 0.6791,
      "step": 1797
    },
    {
      "epoch": 0.17774262906853175,
      "grad_norm": 3.8990301245707992,
      "learning_rate": 1.8877346503147885e-07,
      "loss": 0.6368,
      "step": 1798
    },
    {
      "epoch": 0.1778414848132862,
      "grad_norm": 77.30844159490094,
      "learning_rate": 1.8875871937899374e-07,
      "loss": 0.8033,
      "step": 1799
    },
    {
      "epoch": 0.17794034055804067,
      "grad_norm": 6.176577443754129,
      "learning_rate": 1.887439646256112e-07,
      "loss": 0.7062,
      "step": 1800
    },
    {
      "epoch": 0.17803919630279513,
      "grad_norm": 4.475411286539406,
      "learning_rate": 1.8872920077284414e-07,
      "loss": 0.7526,
      "step": 1801
    },
    {
      "epoch": 0.17813805204754962,
      "grad_norm": 3.2811333797559197,
      "learning_rate": 1.8871442782220643e-07,
      "loss": 0.8072,
      "step": 1802
    },
    {
      "epoch": 0.17823690779230408,
      "grad_norm": 3.3281516515946965,
      "learning_rate": 1.8869964577521276e-07,
      "loss": 0.7376,
      "step": 1803
    },
    {
      "epoch": 0.17833576353705854,
      "grad_norm": 5.941068734113968,
      "learning_rate": 1.8868485463337884e-07,
      "loss": 0.7736,
      "step": 1804
    },
    {
      "epoch": 0.178434619281813,
      "grad_norm": 34.95799558733265,
      "learning_rate": 1.8867005439822126e-07,
      "loss": 0.7185,
      "step": 1805
    },
    {
      "epoch": 0.1785334750265675,
      "grad_norm": 4.76487394551764,
      "learning_rate": 1.8865524507125755e-07,
      "loss": 0.7055,
      "step": 1806
    },
    {
      "epoch": 0.17863233077132196,
      "grad_norm": 3.9363867184476793,
      "learning_rate": 1.8864042665400625e-07,
      "loss": 0.7474,
      "step": 1807
    },
    {
      "epoch": 0.17873118651607642,
      "grad_norm": 5.202205334626098,
      "learning_rate": 1.8862559914798673e-07,
      "loss": 0.6999,
      "step": 1808
    },
    {
      "epoch": 0.17883004226083088,
      "grad_norm": 4.5455299664929845,
      "learning_rate": 1.8861076255471934e-07,
      "loss": 0.6991,
      "step": 1809
    },
    {
      "epoch": 0.17892889800558534,
      "grad_norm": 4.643668243486312,
      "learning_rate": 1.8859591687572536e-07,
      "loss": 0.7752,
      "step": 1810
    },
    {
      "epoch": 0.17902775375033983,
      "grad_norm": 8.444676119069486,
      "learning_rate": 1.8858106211252695e-07,
      "loss": 0.7251,
      "step": 1811
    },
    {
      "epoch": 0.1791266094950943,
      "grad_norm": 3.82870412967721,
      "learning_rate": 1.8856619826664732e-07,
      "loss": 0.7802,
      "step": 1812
    },
    {
      "epoch": 0.17922546523984875,
      "grad_norm": 15.079608015041147,
      "learning_rate": 1.885513253396105e-07,
      "loss": 0.7686,
      "step": 1813
    },
    {
      "epoch": 0.1793243209846032,
      "grad_norm": 4.205803605995836,
      "learning_rate": 1.8853644333294145e-07,
      "loss": 0.7702,
      "step": 1814
    },
    {
      "epoch": 0.17942317672935768,
      "grad_norm": 2.743874180639262,
      "learning_rate": 1.8852155224816616e-07,
      "loss": 0.7473,
      "step": 1815
    },
    {
      "epoch": 0.17952203247411216,
      "grad_norm": 30.951166140010553,
      "learning_rate": 1.885066520868115e-07,
      "loss": 0.7655,
      "step": 1816
    },
    {
      "epoch": 0.17962088821886663,
      "grad_norm": 4.519911135954376,
      "learning_rate": 1.8849174285040518e-07,
      "loss": 0.7192,
      "step": 1817
    },
    {
      "epoch": 0.1797197439636211,
      "grad_norm": 4.323111743427903,
      "learning_rate": 1.8847682454047602e-07,
      "loss": 0.6746,
      "step": 1818
    },
    {
      "epoch": 0.17981859970837555,
      "grad_norm": 3.437457255062977,
      "learning_rate": 1.8846189715855362e-07,
      "loss": 0.6972,
      "step": 1819
    },
    {
      "epoch": 0.17991745545313,
      "grad_norm": 3.216207640147576,
      "learning_rate": 1.8844696070616857e-07,
      "loss": 0.7215,
      "step": 1820
    },
    {
      "epoch": 0.1800163111978845,
      "grad_norm": 4.650357954327682,
      "learning_rate": 1.8843201518485237e-07,
      "loss": 0.7258,
      "step": 1821
    },
    {
      "epoch": 0.18011516694263896,
      "grad_norm": 8.463705440406722,
      "learning_rate": 1.884170605961375e-07,
      "loss": 0.7728,
      "step": 1822
    },
    {
      "epoch": 0.18021402268739342,
      "grad_norm": 3.32009383004289,
      "learning_rate": 1.884020969415573e-07,
      "loss": 0.7168,
      "step": 1823
    },
    {
      "epoch": 0.18031287843214788,
      "grad_norm": 3.8647142381400856,
      "learning_rate": 1.8838712422264606e-07,
      "loss": 0.8023,
      "step": 1824
    },
    {
      "epoch": 0.18041173417690234,
      "grad_norm": 4.262500882953969,
      "learning_rate": 1.8837214244093907e-07,
      "loss": 0.6391,
      "step": 1825
    },
    {
      "epoch": 0.18051058992165683,
      "grad_norm": 4.545539652406747,
      "learning_rate": 1.883571515979724e-07,
      "loss": 0.6546,
      "step": 1826
    },
    {
      "epoch": 0.1806094456664113,
      "grad_norm": 4.215530451054043,
      "learning_rate": 1.8834215169528323e-07,
      "loss": 0.7225,
      "step": 1827
    },
    {
      "epoch": 0.18070830141116576,
      "grad_norm": 4.4946764209759404,
      "learning_rate": 1.8832714273440957e-07,
      "loss": 0.7314,
      "step": 1828
    },
    {
      "epoch": 0.18080715715592022,
      "grad_norm": 18.95783094880516,
      "learning_rate": 1.883121247168903e-07,
      "loss": 0.7093,
      "step": 1829
    },
    {
      "epoch": 0.18090601290067468,
      "grad_norm": 3.0818185579694557,
      "learning_rate": 1.8829709764426535e-07,
      "loss": 0.756,
      "step": 1830
    },
    {
      "epoch": 0.18100486864542917,
      "grad_norm": 3.9336023794880077,
      "learning_rate": 1.8828206151807547e-07,
      "loss": 0.7297,
      "step": 1831
    },
    {
      "epoch": 0.18110372439018363,
      "grad_norm": 4.6207629100820675,
      "learning_rate": 1.8826701633986247e-07,
      "loss": 0.7402,
      "step": 1832
    },
    {
      "epoch": 0.1812025801349381,
      "grad_norm": 5.9592450990500465,
      "learning_rate": 1.88251962111169e-07,
      "loss": 0.7302,
      "step": 1833
    },
    {
      "epoch": 0.18130143587969255,
      "grad_norm": 4.720060689042958,
      "learning_rate": 1.882368988335386e-07,
      "loss": 0.6564,
      "step": 1834
    },
    {
      "epoch": 0.181400291624447,
      "grad_norm": 15.209966017243241,
      "learning_rate": 1.8822182650851581e-07,
      "loss": 0.6643,
      "step": 1835
    },
    {
      "epoch": 0.1814991473692015,
      "grad_norm": 4.445402013094883,
      "learning_rate": 1.8820674513764607e-07,
      "loss": 0.6571,
      "step": 1836
    },
    {
      "epoch": 0.18159800311395596,
      "grad_norm": 3.8778550577600885,
      "learning_rate": 1.8819165472247576e-07,
      "loss": 0.7792,
      "step": 1837
    },
    {
      "epoch": 0.18169685885871042,
      "grad_norm": 4.03354195572501,
      "learning_rate": 1.881765552645522e-07,
      "loss": 0.6878,
      "step": 1838
    },
    {
      "epoch": 0.1817957146034649,
      "grad_norm": 3.900934242724237,
      "learning_rate": 1.8816144676542356e-07,
      "loss": 0.8178,
      "step": 1839
    },
    {
      "epoch": 0.18189457034821935,
      "grad_norm": 3.578923415300182,
      "learning_rate": 1.8814632922663902e-07,
      "loss": 0.7317,
      "step": 1840
    },
    {
      "epoch": 0.18199342609297384,
      "grad_norm": 6.598521219653272,
      "learning_rate": 1.881312026497487e-07,
      "loss": 0.639,
      "step": 1841
    },
    {
      "epoch": 0.1820922818377283,
      "grad_norm": 4.3452935955708165,
      "learning_rate": 1.8811606703630353e-07,
      "loss": 0.6151,
      "step": 1842
    },
    {
      "epoch": 0.18219113758248276,
      "grad_norm": 3.8910122840411434,
      "learning_rate": 1.8810092238785552e-07,
      "loss": 0.7168,
      "step": 1843
    },
    {
      "epoch": 0.18228999332723722,
      "grad_norm": 2.866194611743589,
      "learning_rate": 1.880857687059575e-07,
      "loss": 0.803,
      "step": 1844
    },
    {
      "epoch": 0.1823888490719917,
      "grad_norm": 4.802179268857267,
      "learning_rate": 1.8807060599216324e-07,
      "loss": 0.883,
      "step": 1845
    },
    {
      "epoch": 0.18248770481674617,
      "grad_norm": 3.6272125313623023,
      "learning_rate": 1.880554342480275e-07,
      "loss": 0.7024,
      "step": 1846
    },
    {
      "epoch": 0.18258656056150063,
      "grad_norm": 3.448906467400841,
      "learning_rate": 1.8804025347510586e-07,
      "loss": 0.658,
      "step": 1847
    },
    {
      "epoch": 0.1826854163062551,
      "grad_norm": 4.524931398519262,
      "learning_rate": 1.8802506367495493e-07,
      "loss": 0.7029,
      "step": 1848
    },
    {
      "epoch": 0.18278427205100956,
      "grad_norm": 4.5672967374315885,
      "learning_rate": 1.8800986484913217e-07,
      "loss": 0.8162,
      "step": 1849
    },
    {
      "epoch": 0.18288312779576404,
      "grad_norm": 4.574513430638343,
      "learning_rate": 1.8799465699919602e-07,
      "loss": 0.7142,
      "step": 1850
    },
    {
      "epoch": 0.1829819835405185,
      "grad_norm": 5.878244969011009,
      "learning_rate": 1.8797944012670578e-07,
      "loss": 0.8426,
      "step": 1851
    },
    {
      "epoch": 0.18308083928527297,
      "grad_norm": 4.231494737065659,
      "learning_rate": 1.8796421423322174e-07,
      "loss": 0.8209,
      "step": 1852
    },
    {
      "epoch": 0.18317969503002743,
      "grad_norm": 3.4652217495660533,
      "learning_rate": 1.8794897932030513e-07,
      "loss": 0.7012,
      "step": 1853
    },
    {
      "epoch": 0.1832785507747819,
      "grad_norm": 3.9181472581296686,
      "learning_rate": 1.87933735389518e-07,
      "loss": 0.6576,
      "step": 1854
    },
    {
      "epoch": 0.18337740651953638,
      "grad_norm": 4.050154772169212,
      "learning_rate": 1.8791848244242344e-07,
      "loss": 0.7323,
      "step": 1855
    },
    {
      "epoch": 0.18347626226429084,
      "grad_norm": 3.558667355033466,
      "learning_rate": 1.8790322048058541e-07,
      "loss": 0.7516,
      "step": 1856
    },
    {
      "epoch": 0.1835751180090453,
      "grad_norm": 4.117433311008633,
      "learning_rate": 1.8788794950556874e-07,
      "loss": 0.8548,
      "step": 1857
    },
    {
      "epoch": 0.18367397375379976,
      "grad_norm": 105.21866878898952,
      "learning_rate": 1.878726695189393e-07,
      "loss": 0.8643,
      "step": 1858
    },
    {
      "epoch": 0.18377282949855422,
      "grad_norm": 7.094828338186231,
      "learning_rate": 1.878573805222638e-07,
      "loss": 0.7679,
      "step": 1859
    },
    {
      "epoch": 0.1838716852433087,
      "grad_norm": 4.658572665766819,
      "learning_rate": 1.8784208251710996e-07,
      "loss": 0.7895,
      "step": 1860
    },
    {
      "epoch": 0.18397054098806317,
      "grad_norm": 4.947935419646001,
      "learning_rate": 1.878267755050463e-07,
      "loss": 0.7476,
      "step": 1861
    },
    {
      "epoch": 0.18406939673281764,
      "grad_norm": 5.916255601364905,
      "learning_rate": 1.8781145948764233e-07,
      "loss": 0.6732,
      "step": 1862
    },
    {
      "epoch": 0.1841682524775721,
      "grad_norm": 3.682682444582462,
      "learning_rate": 1.8779613446646851e-07,
      "loss": 0.7406,
      "step": 1863
    },
    {
      "epoch": 0.18426710822232656,
      "grad_norm": 4.45693514640527,
      "learning_rate": 1.877808004430962e-07,
      "loss": 0.7681,
      "step": 1864
    },
    {
      "epoch": 0.18436596396708105,
      "grad_norm": 8.042125278281219,
      "learning_rate": 1.8776545741909764e-07,
      "loss": 0.6658,
      "step": 1865
    },
    {
      "epoch": 0.1844648197118355,
      "grad_norm": 3.6591392297887224,
      "learning_rate": 1.8775010539604606e-07,
      "loss": 0.7218,
      "step": 1866
    },
    {
      "epoch": 0.18456367545658997,
      "grad_norm": 3.917209878717619,
      "learning_rate": 1.8773474437551557e-07,
      "loss": 0.7321,
      "step": 1867
    },
    {
      "epoch": 0.18466253120134443,
      "grad_norm": 6.442848237061841,
      "learning_rate": 1.8771937435908122e-07,
      "loss": 0.7925,
      "step": 1868
    },
    {
      "epoch": 0.1847613869460989,
      "grad_norm": 40.479942240295,
      "learning_rate": 1.87703995348319e-07,
      "loss": 0.7677,
      "step": 1869
    },
    {
      "epoch": 0.18486024269085338,
      "grad_norm": 4.848335356501039,
      "learning_rate": 1.876886073448058e-07,
      "loss": 0.7047,
      "step": 1870
    },
    {
      "epoch": 0.18495909843560784,
      "grad_norm": 6.818397165275487,
      "learning_rate": 1.8767321035011938e-07,
      "loss": 0.7149,
      "step": 1871
    },
    {
      "epoch": 0.1850579541803623,
      "grad_norm": 9.346820435549832,
      "learning_rate": 1.876578043658385e-07,
      "loss": 0.6335,
      "step": 1872
    },
    {
      "epoch": 0.18515680992511677,
      "grad_norm": 4.263900221463838,
      "learning_rate": 1.8764238939354288e-07,
      "loss": 0.734,
      "step": 1873
    },
    {
      "epoch": 0.18525566566987123,
      "grad_norm": 4.336061474534013,
      "learning_rate": 1.87626965434813e-07,
      "loss": 0.7549,
      "step": 1874
    },
    {
      "epoch": 0.18535452141462572,
      "grad_norm": 3.3656806316016934,
      "learning_rate": 1.876115324912304e-07,
      "loss": 0.757,
      "step": 1875
    },
    {
      "epoch": 0.18545337715938018,
      "grad_norm": 3.901975381835877,
      "learning_rate": 1.8759609056437751e-07,
      "loss": 0.647,
      "step": 1876
    },
    {
      "epoch": 0.18555223290413464,
      "grad_norm": 19.389664288223855,
      "learning_rate": 1.8758063965583766e-07,
      "loss": 0.7356,
      "step": 1877
    },
    {
      "epoch": 0.1856510886488891,
      "grad_norm": 6.770302235559567,
      "learning_rate": 1.8756517976719516e-07,
      "loss": 0.6859,
      "step": 1878
    },
    {
      "epoch": 0.18574994439364356,
      "grad_norm": 3.9133105150520837,
      "learning_rate": 1.875497109000351e-07,
      "loss": 0.7713,
      "step": 1879
    },
    {
      "epoch": 0.18584880013839805,
      "grad_norm": 10.78513007135395,
      "learning_rate": 1.8753423305594366e-07,
      "loss": 0.7126,
      "step": 1880
    },
    {
      "epoch": 0.1859476558831525,
      "grad_norm": 4.7028035085464674,
      "learning_rate": 1.8751874623650786e-07,
      "loss": 0.7267,
      "step": 1881
    },
    {
      "epoch": 0.18604651162790697,
      "grad_norm": 5.349512460822471,
      "learning_rate": 1.875032504433156e-07,
      "loss": 0.7985,
      "step": 1882
    },
    {
      "epoch": 0.18614536737266144,
      "grad_norm": 5.874298178455858,
      "learning_rate": 1.8748774567795577e-07,
      "loss": 0.7717,
      "step": 1883
    },
    {
      "epoch": 0.18624422311741592,
      "grad_norm": 3.742046928033859,
      "learning_rate": 1.8747223194201815e-07,
      "loss": 0.7402,
      "step": 1884
    },
    {
      "epoch": 0.18634307886217039,
      "grad_norm": 4.0533616968828525,
      "learning_rate": 1.8745670923709348e-07,
      "loss": 0.7952,
      "step": 1885
    },
    {
      "epoch": 0.18644193460692485,
      "grad_norm": 4.217258366941031,
      "learning_rate": 1.8744117756477332e-07,
      "loss": 0.7459,
      "step": 1886
    },
    {
      "epoch": 0.1865407903516793,
      "grad_norm": 5.344972462545019,
      "learning_rate": 1.874256369266503e-07,
      "loss": 0.7263,
      "step": 1887
    },
    {
      "epoch": 0.18663964609643377,
      "grad_norm": 8.574753898603118,
      "learning_rate": 1.8741008732431778e-07,
      "loss": 0.6801,
      "step": 1888
    },
    {
      "epoch": 0.18673850184118826,
      "grad_norm": 31.076856296542843,
      "learning_rate": 1.873945287593702e-07,
      "loss": 0.8197,
      "step": 1889
    },
    {
      "epoch": 0.18683735758594272,
      "grad_norm": 3.5117957311870778,
      "learning_rate": 1.8737896123340286e-07,
      "loss": 0.7142,
      "step": 1890
    },
    {
      "epoch": 0.18693621333069718,
      "grad_norm": 4.82712285754925,
      "learning_rate": 1.8736338474801201e-07,
      "loss": 0.8138,
      "step": 1891
    },
    {
      "epoch": 0.18703506907545164,
      "grad_norm": 6.744625155113649,
      "learning_rate": 1.8734779930479472e-07,
      "loss": 0.7457,
      "step": 1892
    },
    {
      "epoch": 0.1871339248202061,
      "grad_norm": 3.7288790158498593,
      "learning_rate": 1.873322049053491e-07,
      "loss": 0.719,
      "step": 1893
    },
    {
      "epoch": 0.1872327805649606,
      "grad_norm": 4.50784502630179,
      "learning_rate": 1.873166015512741e-07,
      "loss": 0.7124,
      "step": 1894
    },
    {
      "epoch": 0.18733163630971505,
      "grad_norm": 5.732874961730996,
      "learning_rate": 1.873009892441696e-07,
      "loss": 0.7532,
      "step": 1895
    },
    {
      "epoch": 0.18743049205446952,
      "grad_norm": 3.2969103847205723,
      "learning_rate": 1.8728536798563645e-07,
      "loss": 0.7395,
      "step": 1896
    },
    {
      "epoch": 0.18752934779922398,
      "grad_norm": 3.3422283968473323,
      "learning_rate": 1.8726973777727638e-07,
      "loss": 0.6764,
      "step": 1897
    },
    {
      "epoch": 0.18762820354397844,
      "grad_norm": 3.2836247088860167,
      "learning_rate": 1.8725409862069198e-07,
      "loss": 0.7615,
      "step": 1898
    },
    {
      "epoch": 0.18772705928873293,
      "grad_norm": 4.482358254849221,
      "learning_rate": 1.8723845051748693e-07,
      "loss": 0.6663,
      "step": 1899
    },
    {
      "epoch": 0.1878259150334874,
      "grad_norm": 4.500240968314746,
      "learning_rate": 1.8722279346926557e-07,
      "loss": 0.7412,
      "step": 1900
    },
    {
      "epoch": 0.18792477077824185,
      "grad_norm": 3.63285317766633,
      "learning_rate": 1.8720712747763338e-07,
      "loss": 0.7828,
      "step": 1901
    },
    {
      "epoch": 0.1880236265229963,
      "grad_norm": 3.0464095686219603,
      "learning_rate": 1.8719145254419667e-07,
      "loss": 0.6592,
      "step": 1902
    },
    {
      "epoch": 0.18812248226775077,
      "grad_norm": 4.991750217735482,
      "learning_rate": 1.8717576867056267e-07,
      "loss": 0.7257,
      "step": 1903
    },
    {
      "epoch": 0.18822133801250526,
      "grad_norm": 4.045140642123365,
      "learning_rate": 1.871600758583395e-07,
      "loss": 0.7731,
      "step": 1904
    },
    {
      "epoch": 0.18832019375725972,
      "grad_norm": 8.689155234185737,
      "learning_rate": 1.871443741091363e-07,
      "loss": 0.8706,
      "step": 1905
    },
    {
      "epoch": 0.18841904950201419,
      "grad_norm": 5.803533385050469,
      "learning_rate": 1.8712866342456296e-07,
      "loss": 0.74,
      "step": 1906
    },
    {
      "epoch": 0.18851790524676865,
      "grad_norm": 5.016204710358484,
      "learning_rate": 1.8711294380623044e-07,
      "loss": 0.7538,
      "step": 1907
    },
    {
      "epoch": 0.1886167609915231,
      "grad_norm": 3.6876257342722556,
      "learning_rate": 1.8709721525575054e-07,
      "loss": 0.8652,
      "step": 1908
    },
    {
      "epoch": 0.1887156167362776,
      "grad_norm": 3.483514718884918,
      "learning_rate": 1.87081477774736e-07,
      "loss": 0.746,
      "step": 1909
    },
    {
      "epoch": 0.18881447248103206,
      "grad_norm": 3.809410292563556,
      "learning_rate": 1.8706573136480043e-07,
      "loss": 0.7447,
      "step": 1910
    },
    {
      "epoch": 0.18891332822578652,
      "grad_norm": 3.620845125978442,
      "learning_rate": 1.8704997602755842e-07,
      "loss": 0.8005,
      "step": 1911
    },
    {
      "epoch": 0.18901218397054098,
      "grad_norm": 3.501943072052065,
      "learning_rate": 1.8703421176462549e-07,
      "loss": 0.6845,
      "step": 1912
    },
    {
      "epoch": 0.18911103971529544,
      "grad_norm": 4.899741223158862,
      "learning_rate": 1.8701843857761796e-07,
      "loss": 0.6851,
      "step": 1913
    },
    {
      "epoch": 0.18920989546004993,
      "grad_norm": 7.057631324938306,
      "learning_rate": 1.8700265646815317e-07,
      "loss": 0.848,
      "step": 1914
    },
    {
      "epoch": 0.1893087512048044,
      "grad_norm": 5.148111888918845,
      "learning_rate": 1.8698686543784935e-07,
      "loss": 0.7295,
      "step": 1915
    },
    {
      "epoch": 0.18940760694955885,
      "grad_norm": 4.108260700235145,
      "learning_rate": 1.869710654883256e-07,
      "loss": 0.6357,
      "step": 1916
    },
    {
      "epoch": 0.18950646269431332,
      "grad_norm": 7.051704098816246,
      "learning_rate": 1.86955256621202e-07,
      "loss": 0.7892,
      "step": 1917
    },
    {
      "epoch": 0.18960531843906778,
      "grad_norm": 4.202887880591141,
      "learning_rate": 1.8693943883809953e-07,
      "loss": 0.803,
      "step": 1918
    },
    {
      "epoch": 0.18970417418382227,
      "grad_norm": 7.799982007759803,
      "learning_rate": 1.8692361214064004e-07,
      "loss": 0.6871,
      "step": 1919
    },
    {
      "epoch": 0.18980302992857673,
      "grad_norm": 4.983445244764989,
      "learning_rate": 1.8690777653044634e-07,
      "loss": 0.6644,
      "step": 1920
    },
    {
      "epoch": 0.1899018856733312,
      "grad_norm": 3.5499396652532353,
      "learning_rate": 1.8689193200914215e-07,
      "loss": 0.731,
      "step": 1921
    },
    {
      "epoch": 0.19000074141808565,
      "grad_norm": 8.074545821666383,
      "learning_rate": 1.8687607857835206e-07,
      "loss": 0.7884,
      "step": 1922
    },
    {
      "epoch": 0.19009959716284014,
      "grad_norm": 21.766083737473622,
      "learning_rate": 1.8686021623970163e-07,
      "loss": 0.757,
      "step": 1923
    },
    {
      "epoch": 0.1901984529075946,
      "grad_norm": 12.684671453147544,
      "learning_rate": 1.868443449948173e-07,
      "loss": 0.7505,
      "step": 1924
    },
    {
      "epoch": 0.19029730865234906,
      "grad_norm": 19.674597938538255,
      "learning_rate": 1.8682846484532644e-07,
      "loss": 0.727,
      "step": 1925
    },
    {
      "epoch": 0.19039616439710352,
      "grad_norm": 6.1356617369851305,
      "learning_rate": 1.8681257579285734e-07,
      "loss": 0.7033,
      "step": 1926
    },
    {
      "epoch": 0.19049502014185798,
      "grad_norm": 3.341396678996001,
      "learning_rate": 1.8679667783903917e-07,
      "loss": 0.6992,
      "step": 1927
    },
    {
      "epoch": 0.19059387588661247,
      "grad_norm": 3.074389685482824,
      "learning_rate": 1.8678077098550197e-07,
      "loss": 0.6631,
      "step": 1928
    },
    {
      "epoch": 0.19069273163136694,
      "grad_norm": 4.238446523740354,
      "learning_rate": 1.8676485523387686e-07,
      "loss": 0.7505,
      "step": 1929
    },
    {
      "epoch": 0.1907915873761214,
      "grad_norm": 3.783729317151618,
      "learning_rate": 1.8674893058579572e-07,
      "loss": 0.6987,
      "step": 1930
    },
    {
      "epoch": 0.19089044312087586,
      "grad_norm": 3.0395851080727287,
      "learning_rate": 1.8673299704289136e-07,
      "loss": 0.6899,
      "step": 1931
    },
    {
      "epoch": 0.19098929886563032,
      "grad_norm": 6.138223006987396,
      "learning_rate": 1.867170546067976e-07,
      "loss": 0.7533,
      "step": 1932
    },
    {
      "epoch": 0.1910881546103848,
      "grad_norm": 4.2118464175427235,
      "learning_rate": 1.86701103279149e-07,
      "loss": 0.7461,
      "step": 1933
    },
    {
      "epoch": 0.19118701035513927,
      "grad_norm": 9.11331141470458,
      "learning_rate": 1.866851430615812e-07,
      "loss": 0.6615,
      "step": 1934
    },
    {
      "epoch": 0.19128586609989373,
      "grad_norm": 9.654146727942761,
      "learning_rate": 1.866691739557307e-07,
      "loss": 0.7581,
      "step": 1935
    },
    {
      "epoch": 0.1913847218446482,
      "grad_norm": 12.265088307108782,
      "learning_rate": 1.8665319596323487e-07,
      "loss": 0.7688,
      "step": 1936
    },
    {
      "epoch": 0.19148357758940265,
      "grad_norm": 4.849643320705063,
      "learning_rate": 1.8663720908573199e-07,
      "loss": 0.623,
      "step": 1937
    },
    {
      "epoch": 0.19158243333415714,
      "grad_norm": 36.85656350290554,
      "learning_rate": 1.8662121332486135e-07,
      "loss": 0.7656,
      "step": 1938
    },
    {
      "epoch": 0.1916812890789116,
      "grad_norm": 21.62825651259929,
      "learning_rate": 1.8660520868226303e-07,
      "loss": 0.7854,
      "step": 1939
    },
    {
      "epoch": 0.19178014482366607,
      "grad_norm": 5.372182705836245,
      "learning_rate": 1.8658919515957804e-07,
      "loss": 0.6756,
      "step": 1940
    },
    {
      "epoch": 0.19187900056842053,
      "grad_norm": 4.017895334617562,
      "learning_rate": 1.8657317275844836e-07,
      "loss": 0.7287,
      "step": 1941
    },
    {
      "epoch": 0.191977856313175,
      "grad_norm": 4.21043627975122,
      "learning_rate": 1.865571414805169e-07,
      "loss": 0.7624,
      "step": 1942
    },
    {
      "epoch": 0.19207671205792948,
      "grad_norm": 9.389874270421053,
      "learning_rate": 1.865411013274274e-07,
      "loss": 0.6689,
      "step": 1943
    },
    {
      "epoch": 0.19217556780268394,
      "grad_norm": 6.9484055105973175,
      "learning_rate": 1.8652505230082448e-07,
      "loss": 0.8289,
      "step": 1944
    },
    {
      "epoch": 0.1922744235474384,
      "grad_norm": 9.64788591882521,
      "learning_rate": 1.865089944023538e-07,
      "loss": 0.76,
      "step": 1945
    },
    {
      "epoch": 0.19237327929219286,
      "grad_norm": 3.780502316695749,
      "learning_rate": 1.8649292763366185e-07,
      "loss": 0.8278,
      "step": 1946
    },
    {
      "epoch": 0.19247213503694732,
      "grad_norm": 5.222946234151323,
      "learning_rate": 1.8647685199639604e-07,
      "loss": 0.7218,
      "step": 1947
    },
    {
      "epoch": 0.1925709907817018,
      "grad_norm": 4.0293218875811085,
      "learning_rate": 1.8646076749220464e-07,
      "loss": 0.7352,
      "step": 1948
    },
    {
      "epoch": 0.19266984652645627,
      "grad_norm": 4.187860621454855,
      "learning_rate": 1.8644467412273696e-07,
      "loss": 0.669,
      "step": 1949
    },
    {
      "epoch": 0.19276870227121073,
      "grad_norm": 3.7720366068084745,
      "learning_rate": 1.8642857188964306e-07,
      "loss": 0.6443,
      "step": 1950
    },
    {
      "epoch": 0.1928675580159652,
      "grad_norm": 5.777924010325393,
      "learning_rate": 1.8641246079457407e-07,
      "loss": 0.6816,
      "step": 1951
    },
    {
      "epoch": 0.19296641376071966,
      "grad_norm": 5.836200292135517,
      "learning_rate": 1.8639634083918186e-07,
      "loss": 0.6988,
      "step": 1952
    },
    {
      "epoch": 0.19306526950547415,
      "grad_norm": 3.756071286363221,
      "learning_rate": 1.8638021202511935e-07,
      "loss": 0.6478,
      "step": 1953
    },
    {
      "epoch": 0.1931641252502286,
      "grad_norm": 4.355592454079265,
      "learning_rate": 1.863640743540403e-07,
      "loss": 0.7816,
      "step": 1954
    },
    {
      "epoch": 0.19326298099498307,
      "grad_norm": 5.051697416123819,
      "learning_rate": 1.8634792782759939e-07,
      "loss": 0.6211,
      "step": 1955
    },
    {
      "epoch": 0.19336183673973753,
      "grad_norm": 3.2188495907151133,
      "learning_rate": 1.863317724474522e-07,
      "loss": 0.7853,
      "step": 1956
    },
    {
      "epoch": 0.193460692484492,
      "grad_norm": 4.156959307316423,
      "learning_rate": 1.8631560821525523e-07,
      "loss": 0.7482,
      "step": 1957
    },
    {
      "epoch": 0.19355954822924648,
      "grad_norm": 3.662095110274648,
      "learning_rate": 1.8629943513266586e-07,
      "loss": 0.6833,
      "step": 1958
    },
    {
      "epoch": 0.19365840397400094,
      "grad_norm": 4.177477282943034,
      "learning_rate": 1.8628325320134245e-07,
      "loss": 0.7474,
      "step": 1959
    },
    {
      "epoch": 0.1937572597187554,
      "grad_norm": 8.999719021418775,
      "learning_rate": 1.8626706242294414e-07,
      "loss": 0.7092,
      "step": 1960
    },
    {
      "epoch": 0.19385611546350986,
      "grad_norm": 3.863179878292873,
      "learning_rate": 1.862508627991312e-07,
      "loss": 0.694,
      "step": 1961
    },
    {
      "epoch": 0.19395497120826435,
      "grad_norm": 7.25389030221628,
      "learning_rate": 1.862346543315645e-07,
      "loss": 0.6656,
      "step": 1962
    },
    {
      "epoch": 0.19405382695301882,
      "grad_norm": 3.011368458648499,
      "learning_rate": 1.8621843702190605e-07,
      "loss": 0.808,
      "step": 1963
    },
    {
      "epoch": 0.19415268269777328,
      "grad_norm": 5.078148150117684,
      "learning_rate": 1.8620221087181868e-07,
      "loss": 0.6292,
      "step": 1964
    },
    {
      "epoch": 0.19425153844252774,
      "grad_norm": 3.6537949752011603,
      "learning_rate": 1.8618597588296621e-07,
      "loss": 0.7502,
      "step": 1965
    },
    {
      "epoch": 0.1943503941872822,
      "grad_norm": 3.5930459606781397,
      "learning_rate": 1.8616973205701322e-07,
      "loss": 0.7732,
      "step": 1966
    },
    {
      "epoch": 0.1944492499320367,
      "grad_norm": 3.6873448276760135,
      "learning_rate": 1.861534793956253e-07,
      "loss": 0.6476,
      "step": 1967
    },
    {
      "epoch": 0.19454810567679115,
      "grad_norm": 3.5309316161982793,
      "learning_rate": 1.8613721790046893e-07,
      "loss": 0.8254,
      "step": 1968
    },
    {
      "epoch": 0.1946469614215456,
      "grad_norm": 5.0617123642870885,
      "learning_rate": 1.8612094757321146e-07,
      "loss": 0.7454,
      "step": 1969
    },
    {
      "epoch": 0.19474581716630007,
      "grad_norm": 21.50718598654768,
      "learning_rate": 1.8610466841552122e-07,
      "loss": 0.762,
      "step": 1970
    },
    {
      "epoch": 0.19484467291105453,
      "grad_norm": 5.682328387896488,
      "learning_rate": 1.8608838042906734e-07,
      "loss": 0.7942,
      "step": 1971
    },
    {
      "epoch": 0.19494352865580902,
      "grad_norm": 3.4986147379014816,
      "learning_rate": 1.8607208361551993e-07,
      "loss": 0.7663,
      "step": 1972
    },
    {
      "epoch": 0.19504238440056348,
      "grad_norm": 5.411974138464323,
      "learning_rate": 1.8605577797655002e-07,
      "loss": 0.5895,
      "step": 1973
    },
    {
      "epoch": 0.19514124014531795,
      "grad_norm": 8.903525364175954,
      "learning_rate": 1.860394635138295e-07,
      "loss": 0.7109,
      "step": 1974
    },
    {
      "epoch": 0.1952400958900724,
      "grad_norm": 3.478387702105189,
      "learning_rate": 1.8602314022903117e-07,
      "loss": 0.7293,
      "step": 1975
    },
    {
      "epoch": 0.19533895163482687,
      "grad_norm": 4.086198379982533,
      "learning_rate": 1.8600680812382872e-07,
      "loss": 0.7607,
      "step": 1976
    },
    {
      "epoch": 0.19543780737958136,
      "grad_norm": 3.33171134362502,
      "learning_rate": 1.8599046719989677e-07,
      "loss": 0.718,
      "step": 1977
    },
    {
      "epoch": 0.19553666312433582,
      "grad_norm": 3.9085603799658375,
      "learning_rate": 1.859741174589109e-07,
      "loss": 0.7634,
      "step": 1978
    },
    {
      "epoch": 0.19563551886909028,
      "grad_norm": 3.222983510646508,
      "learning_rate": 1.8595775890254749e-07,
      "loss": 0.717,
      "step": 1979
    },
    {
      "epoch": 0.19573437461384474,
      "grad_norm": 9.33255233547826,
      "learning_rate": 1.8594139153248382e-07,
      "loss": 0.7527,
      "step": 1980
    },
    {
      "epoch": 0.1958332303585992,
      "grad_norm": 3.564222216714175,
      "learning_rate": 1.8592501535039823e-07,
      "loss": 0.7394,
      "step": 1981
    },
    {
      "epoch": 0.1959320861033537,
      "grad_norm": 3.118912220548284,
      "learning_rate": 1.8590863035796976e-07,
      "loss": 0.8251,
      "step": 1982
    },
    {
      "epoch": 0.19603094184810815,
      "grad_norm": 4.926386829837075,
      "learning_rate": 1.8589223655687852e-07,
      "loss": 0.7798,
      "step": 1983
    },
    {
      "epoch": 0.19612979759286261,
      "grad_norm": 3.875115983833439,
      "learning_rate": 1.858758339488054e-07,
      "loss": 0.7311,
      "step": 1984
    },
    {
      "epoch": 0.19622865333761708,
      "grad_norm": 4.277079821104643,
      "learning_rate": 1.8585942253543226e-07,
      "loss": 0.7129,
      "step": 1985
    },
    {
      "epoch": 0.19632750908237154,
      "grad_norm": 4.202060299357318,
      "learning_rate": 1.8584300231844186e-07,
      "loss": 0.6502,
      "step": 1986
    },
    {
      "epoch": 0.19642636482712603,
      "grad_norm": 4.421674473782117,
      "learning_rate": 1.8582657329951786e-07,
      "loss": 0.6394,
      "step": 1987
    },
    {
      "epoch": 0.1965252205718805,
      "grad_norm": 4.549215536605262,
      "learning_rate": 1.8581013548034477e-07,
      "loss": 0.7185,
      "step": 1988
    },
    {
      "epoch": 0.19662407631663495,
      "grad_norm": 5.123613862132663,
      "learning_rate": 1.857936888626081e-07,
      "loss": 0.7048,
      "step": 1989
    },
    {
      "epoch": 0.1967229320613894,
      "grad_norm": 4.129078078543203,
      "learning_rate": 1.8577723344799417e-07,
      "loss": 0.7498,
      "step": 1990
    },
    {
      "epoch": 0.19682178780614387,
      "grad_norm": 4.655640894864318,
      "learning_rate": 1.8576076923819027e-07,
      "loss": 0.6881,
      "step": 1991
    },
    {
      "epoch": 0.19692064355089836,
      "grad_norm": 3.3859907568126655,
      "learning_rate": 1.8574429623488453e-07,
      "loss": 0.7849,
      "step": 1992
    },
    {
      "epoch": 0.19701949929565282,
      "grad_norm": 8.159999385393213,
      "learning_rate": 1.8572781443976603e-07,
      "loss": 0.7758,
      "step": 1993
    },
    {
      "epoch": 0.19711835504040728,
      "grad_norm": 3.231858111683805,
      "learning_rate": 1.8571132385452474e-07,
      "loss": 0.692,
      "step": 1994
    },
    {
      "epoch": 0.19721721078516175,
      "grad_norm": 3.666922421322966,
      "learning_rate": 1.8569482448085152e-07,
      "loss": 0.6934,
      "step": 1995
    },
    {
      "epoch": 0.1973160665299162,
      "grad_norm": 5.7330239538606165,
      "learning_rate": 1.856783163204381e-07,
      "loss": 0.7089,
      "step": 1996
    },
    {
      "epoch": 0.1974149222746707,
      "grad_norm": 5.456843455235043,
      "learning_rate": 1.8566179937497725e-07,
      "loss": 0.8004,
      "step": 1997
    },
    {
      "epoch": 0.19751377801942516,
      "grad_norm": 7.596041030306734,
      "learning_rate": 1.8564527364616245e-07,
      "loss": 0.743,
      "step": 1998
    },
    {
      "epoch": 0.19761263376417962,
      "grad_norm": 4.39477424418152,
      "learning_rate": 1.8562873913568817e-07,
      "loss": 0.7342,
      "step": 1999
    },
    {
      "epoch": 0.19771148950893408,
      "grad_norm": 4.908385848226004,
      "learning_rate": 1.856121958452498e-07,
      "loss": 0.7569,
      "step": 2000
    },
    {
      "epoch": 0.19781034525368857,
      "grad_norm": 5.459821876791214,
      "learning_rate": 1.8559564377654365e-07,
      "loss": 0.7871,
      "step": 2001
    },
    {
      "epoch": 0.19790920099844303,
      "grad_norm": 3.7794298240922912,
      "learning_rate": 1.8557908293126686e-07,
      "loss": 0.7838,
      "step": 2002
    },
    {
      "epoch": 0.1980080567431975,
      "grad_norm": 5.536526902226849,
      "learning_rate": 1.8556251331111746e-07,
      "loss": 0.715,
      "step": 2003
    },
    {
      "epoch": 0.19810691248795195,
      "grad_norm": 3.7771766620918386,
      "learning_rate": 1.855459349177945e-07,
      "loss": 0.617,
      "step": 2004
    },
    {
      "epoch": 0.19820576823270641,
      "grad_norm": 3.158765925520094,
      "learning_rate": 1.8552934775299777e-07,
      "loss": 0.6715,
      "step": 2005
    },
    {
      "epoch": 0.1983046239774609,
      "grad_norm": 4.031318975815699,
      "learning_rate": 1.8551275181842806e-07,
      "loss": 0.7263,
      "step": 2006
    },
    {
      "epoch": 0.19840347972221536,
      "grad_norm": 5.807040972288272,
      "learning_rate": 1.854961471157871e-07,
      "loss": 0.6296,
      "step": 2007
    },
    {
      "epoch": 0.19850233546696983,
      "grad_norm": 4.744984302603516,
      "learning_rate": 1.8547953364677738e-07,
      "loss": 0.7239,
      "step": 2008
    },
    {
      "epoch": 0.1986011912117243,
      "grad_norm": 8.198867899753633,
      "learning_rate": 1.854629114131024e-07,
      "loss": 0.7271,
      "step": 2009
    },
    {
      "epoch": 0.19870004695647875,
      "grad_norm": 4.316456517769032,
      "learning_rate": 1.8544628041646653e-07,
      "loss": 0.7086,
      "step": 2010
    },
    {
      "epoch": 0.19879890270123324,
      "grad_norm": 6.89464877790718,
      "learning_rate": 1.85429640658575e-07,
      "loss": 0.7465,
      "step": 2011
    },
    {
      "epoch": 0.1988977584459877,
      "grad_norm": 4.854001987489522,
      "learning_rate": 1.8541299214113405e-07,
      "loss": 0.795,
      "step": 2012
    },
    {
      "epoch": 0.19899661419074216,
      "grad_norm": 7.521661854532481,
      "learning_rate": 1.8539633486585063e-07,
      "loss": 0.7936,
      "step": 2013
    },
    {
      "epoch": 0.19909546993549662,
      "grad_norm": 4.234939822727864,
      "learning_rate": 1.8537966883443277e-07,
      "loss": 0.6846,
      "step": 2014
    },
    {
      "epoch": 0.19919432568025108,
      "grad_norm": 115.80350118368003,
      "learning_rate": 1.8536299404858933e-07,
      "loss": 0.6982,
      "step": 2015
    },
    {
      "epoch": 0.19929318142500557,
      "grad_norm": 2.677720720621416,
      "learning_rate": 1.8534631051003002e-07,
      "loss": 0.6713,
      "step": 2016
    },
    {
      "epoch": 0.19939203716976003,
      "grad_norm": 5.747486046378795,
      "learning_rate": 1.8532961822046552e-07,
      "loss": 0.6139,
      "step": 2017
    },
    {
      "epoch": 0.1994908929145145,
      "grad_norm": 3.8757676307567266,
      "learning_rate": 1.853129171816074e-07,
      "loss": 0.7421,
      "step": 2018
    },
    {
      "epoch": 0.19958974865926896,
      "grad_norm": 17.221514102630028,
      "learning_rate": 1.8529620739516806e-07,
      "loss": 0.7648,
      "step": 2019
    },
    {
      "epoch": 0.19968860440402342,
      "grad_norm": 4.257730799713445,
      "learning_rate": 1.8527948886286081e-07,
      "loss": 0.7028,
      "step": 2020
    },
    {
      "epoch": 0.1997874601487779,
      "grad_norm": 4.773551643875445,
      "learning_rate": 1.8526276158639997e-07,
      "loss": 0.747,
      "step": 2021
    },
    {
      "epoch": 0.19988631589353237,
      "grad_norm": 5.458688489622943,
      "learning_rate": 1.8524602556750067e-07,
      "loss": 0.6962,
      "step": 2022
    },
    {
      "epoch": 0.19998517163828683,
      "grad_norm": 5.381793198970702,
      "learning_rate": 1.8522928080787888e-07,
      "loss": 0.7362,
      "step": 2023
    },
    {
      "epoch": 0.2000840273830413,
      "grad_norm": 4.048573073555702,
      "learning_rate": 1.852125273092516e-07,
      "loss": 0.6398,
      "step": 2024
    },
    {
      "epoch": 0.20018288312779575,
      "grad_norm": 14.49924567761937,
      "learning_rate": 1.851957650733366e-07,
      "loss": 0.683,
      "step": 2025
    },
    {
      "epoch": 0.20028173887255024,
      "grad_norm": 7.416574550730637,
      "learning_rate": 1.851789941018526e-07,
      "loss": 0.789,
      "step": 2026
    },
    {
      "epoch": 0.2003805946173047,
      "grad_norm": 3.630509307082584,
      "learning_rate": 1.8516221439651927e-07,
      "loss": 0.7117,
      "step": 2027
    },
    {
      "epoch": 0.20047945036205916,
      "grad_norm": 15.585682846142296,
      "learning_rate": 1.8514542595905707e-07,
      "loss": 0.792,
      "step": 2028
    },
    {
      "epoch": 0.20057830610681363,
      "grad_norm": 2.7644625792571107,
      "learning_rate": 1.8512862879118742e-07,
      "loss": 0.628,
      "step": 2029
    },
    {
      "epoch": 0.2006771618515681,
      "grad_norm": 3.9225088005558213,
      "learning_rate": 1.851118228946326e-07,
      "loss": 0.7364,
      "step": 2030
    },
    {
      "epoch": 0.20077601759632258,
      "grad_norm": 4.363391086571473,
      "learning_rate": 1.8509500827111585e-07,
      "loss": 0.7564,
      "step": 2031
    },
    {
      "epoch": 0.20087487334107704,
      "grad_norm": 3.6919457556012456,
      "learning_rate": 1.8507818492236125e-07,
      "loss": 0.6936,
      "step": 2032
    },
    {
      "epoch": 0.2009737290858315,
      "grad_norm": 3.3602267770233554,
      "learning_rate": 1.8506135285009378e-07,
      "loss": 0.6135,
      "step": 2033
    },
    {
      "epoch": 0.20107258483058596,
      "grad_norm": 7.385664038669656,
      "learning_rate": 1.8504451205603928e-07,
      "loss": 0.7465,
      "step": 2034
    },
    {
      "epoch": 0.20117144057534042,
      "grad_norm": 3.9188951468791826,
      "learning_rate": 1.850276625419246e-07,
      "loss": 0.6843,
      "step": 2035
    },
    {
      "epoch": 0.2012702963200949,
      "grad_norm": 5.017566962052129,
      "learning_rate": 1.8501080430947735e-07,
      "loss": 0.7245,
      "step": 2036
    },
    {
      "epoch": 0.20136915206484937,
      "grad_norm": 4.628907637624656,
      "learning_rate": 1.8499393736042616e-07,
      "loss": 0.7355,
      "step": 2037
    },
    {
      "epoch": 0.20146800780960383,
      "grad_norm": 4.530533025640185,
      "learning_rate": 1.8497706169650038e-07,
      "loss": 0.8286,
      "step": 2038
    },
    {
      "epoch": 0.2015668635543583,
      "grad_norm": 3.7062807705538923,
      "learning_rate": 1.849601773194305e-07,
      "loss": 0.7194,
      "step": 2039
    },
    {
      "epoch": 0.20166571929911278,
      "grad_norm": 3.5619644661959073,
      "learning_rate": 1.849432842309476e-07,
      "loss": 0.7154,
      "step": 2040
    },
    {
      "epoch": 0.20176457504386724,
      "grad_norm": 5.554081810371467,
      "learning_rate": 1.8492638243278395e-07,
      "loss": 0.7267,
      "step": 2041
    },
    {
      "epoch": 0.2018634307886217,
      "grad_norm": 3.3862874961756493,
      "learning_rate": 1.849094719266725e-07,
      "loss": 0.7968,
      "step": 2042
    },
    {
      "epoch": 0.20196228653337617,
      "grad_norm": 11.152754608173318,
      "learning_rate": 1.8489255271434728e-07,
      "loss": 0.7983,
      "step": 2043
    },
    {
      "epoch": 0.20206114227813063,
      "grad_norm": 3.4220570371602004,
      "learning_rate": 1.8487562479754296e-07,
      "loss": 0.726,
      "step": 2044
    },
    {
      "epoch": 0.20215999802288512,
      "grad_norm": 4.314969863895004,
      "learning_rate": 1.8485868817799535e-07,
      "loss": 0.6704,
      "step": 2045
    },
    {
      "epoch": 0.20225885376763958,
      "grad_norm": 4.071755654413249,
      "learning_rate": 1.8484174285744102e-07,
      "loss": 0.8868,
      "step": 2046
    },
    {
      "epoch": 0.20235770951239404,
      "grad_norm": 6.439666685979803,
      "learning_rate": 1.8482478883761748e-07,
      "loss": 0.8431,
      "step": 2047
    },
    {
      "epoch": 0.2024565652571485,
      "grad_norm": 9.751228204612607,
      "learning_rate": 1.8480782612026306e-07,
      "loss": 0.7837,
      "step": 2048
    },
    {
      "epoch": 0.20255542100190296,
      "grad_norm": 3.5880327492709276,
      "learning_rate": 1.8479085470711714e-07,
      "loss": 0.8187,
      "step": 2049
    },
    {
      "epoch": 0.20265427674665745,
      "grad_norm": 4.055449412623008,
      "learning_rate": 1.8477387459991974e-07,
      "loss": 0.7319,
      "step": 2050
    },
    {
      "epoch": 0.2027531324914119,
      "grad_norm": 6.833578040595688,
      "learning_rate": 1.8475688580041204e-07,
      "loss": 0.7508,
      "step": 2051
    },
    {
      "epoch": 0.20285198823616638,
      "grad_norm": 4.344836122301642,
      "learning_rate": 1.8473988831033595e-07,
      "loss": 0.796,
      "step": 2052
    },
    {
      "epoch": 0.20295084398092084,
      "grad_norm": 4.925981567297614,
      "learning_rate": 1.8472288213143434e-07,
      "loss": 0.6863,
      "step": 2053
    },
    {
      "epoch": 0.2030496997256753,
      "grad_norm": 3.5608217943702996,
      "learning_rate": 1.847058672654509e-07,
      "loss": 0.7233,
      "step": 2054
    },
    {
      "epoch": 0.2031485554704298,
      "grad_norm": 3.563368266088548,
      "learning_rate": 1.8468884371413026e-07,
      "loss": 0.6431,
      "step": 2055
    },
    {
      "epoch": 0.20324741121518425,
      "grad_norm": 3.592625770041143,
      "learning_rate": 1.8467181147921795e-07,
      "loss": 0.745,
      "step": 2056
    },
    {
      "epoch": 0.2033462669599387,
      "grad_norm": 4.33495726175727,
      "learning_rate": 1.8465477056246036e-07,
      "loss": 0.8441,
      "step": 2057
    },
    {
      "epoch": 0.20344512270469317,
      "grad_norm": 4.953149981510225,
      "learning_rate": 1.8463772096560483e-07,
      "loss": 0.7599,
      "step": 2058
    },
    {
      "epoch": 0.20354397844944763,
      "grad_norm": 8.463998441983403,
      "learning_rate": 1.846206626903995e-07,
      "loss": 0.6967,
      "step": 2059
    },
    {
      "epoch": 0.20364283419420212,
      "grad_norm": 3.405542817435583,
      "learning_rate": 1.846035957385934e-07,
      "loss": 0.7417,
      "step": 2060
    },
    {
      "epoch": 0.20374168993895658,
      "grad_norm": 7.838212717808324,
      "learning_rate": 1.8458652011193664e-07,
      "loss": 0.7302,
      "step": 2061
    },
    {
      "epoch": 0.20384054568371104,
      "grad_norm": 4.915235171303165,
      "learning_rate": 1.8456943581217993e-07,
      "loss": 0.8382,
      "step": 2062
    },
    {
      "epoch": 0.2039394014284655,
      "grad_norm": 3.767450173957523,
      "learning_rate": 1.8455234284107504e-07,
      "loss": 0.6793,
      "step": 2063
    },
    {
      "epoch": 0.20403825717321997,
      "grad_norm": 5.223157479598452,
      "learning_rate": 1.8453524120037467e-07,
      "loss": 0.6745,
      "step": 2064
    },
    {
      "epoch": 0.20413711291797446,
      "grad_norm": 3.774702897314655,
      "learning_rate": 1.8451813089183226e-07,
      "loss": 0.7461,
      "step": 2065
    },
    {
      "epoch": 0.20423596866272892,
      "grad_norm": 19.412836236720153,
      "learning_rate": 1.8450101191720227e-07,
      "loss": 0.7119,
      "step": 2066
    },
    {
      "epoch": 0.20433482440748338,
      "grad_norm": 3.402853657796023,
      "learning_rate": 1.8448388427824e-07,
      "loss": 0.7962,
      "step": 2067
    },
    {
      "epoch": 0.20443368015223784,
      "grad_norm": 11.967001619721819,
      "learning_rate": 1.844667479767016e-07,
      "loss": 0.7935,
      "step": 2068
    },
    {
      "epoch": 0.2045325358969923,
      "grad_norm": 3.0368270693781247,
      "learning_rate": 1.844496030143442e-07,
      "loss": 0.7266,
      "step": 2069
    },
    {
      "epoch": 0.2046313916417468,
      "grad_norm": 18.462100944743057,
      "learning_rate": 1.844324493929257e-07,
      "loss": 0.8014,
      "step": 2070
    },
    {
      "epoch": 0.20473024738650125,
      "grad_norm": 4.75736371013612,
      "learning_rate": 1.8441528711420499e-07,
      "loss": 0.8248,
      "step": 2071
    },
    {
      "epoch": 0.2048291031312557,
      "grad_norm": 3.618245171947734,
      "learning_rate": 1.843981161799418e-07,
      "loss": 0.6217,
      "step": 2072
    },
    {
      "epoch": 0.20492795887601017,
      "grad_norm": 3.487991315167392,
      "learning_rate": 1.8438093659189677e-07,
      "loss": 0.7709,
      "step": 2073
    },
    {
      "epoch": 0.20502681462076464,
      "grad_norm": 4.352386651774669,
      "learning_rate": 1.843637483518314e-07,
      "loss": 0.6945,
      "step": 2074
    },
    {
      "epoch": 0.20512567036551912,
      "grad_norm": 6.420154417782353,
      "learning_rate": 1.8434655146150806e-07,
      "loss": 0.6558,
      "step": 2075
    },
    {
      "epoch": 0.2052245261102736,
      "grad_norm": 4.7312805085286485,
      "learning_rate": 1.843293459226901e-07,
      "loss": 0.7636,
      "step": 2076
    },
    {
      "epoch": 0.20532338185502805,
      "grad_norm": 3.659408825577001,
      "learning_rate": 1.843121317371416e-07,
      "loss": 0.7459,
      "step": 2077
    },
    {
      "epoch": 0.2054222375997825,
      "grad_norm": 5.559727190030357,
      "learning_rate": 1.8429490890662776e-07,
      "loss": 0.707,
      "step": 2078
    },
    {
      "epoch": 0.205521093344537,
      "grad_norm": 3.5705899234134786,
      "learning_rate": 1.842776774329144e-07,
      "loss": 0.7974,
      "step": 2079
    },
    {
      "epoch": 0.20561994908929146,
      "grad_norm": 3.2588898234167933,
      "learning_rate": 1.8426043731776844e-07,
      "loss": 0.7841,
      "step": 2080
    },
    {
      "epoch": 0.20571880483404592,
      "grad_norm": 4.492173142495916,
      "learning_rate": 1.8424318856295754e-07,
      "loss": 0.8034,
      "step": 2081
    },
    {
      "epoch": 0.20581766057880038,
      "grad_norm": 4.3932208529625845,
      "learning_rate": 1.8422593117025033e-07,
      "loss": 0.7288,
      "step": 2082
    },
    {
      "epoch": 0.20591651632355484,
      "grad_norm": 5.21394630006988,
      "learning_rate": 1.8420866514141626e-07,
      "loss": 0.6397,
      "step": 2083
    },
    {
      "epoch": 0.20601537206830933,
      "grad_norm": 3.5215545693706574,
      "learning_rate": 1.8419139047822577e-07,
      "loss": 0.6099,
      "step": 2084
    },
    {
      "epoch": 0.2061142278130638,
      "grad_norm": 18.00847661944974,
      "learning_rate": 1.841741071824501e-07,
      "loss": 0.7183,
      "step": 2085
    },
    {
      "epoch": 0.20621308355781826,
      "grad_norm": 6.8528321739097775,
      "learning_rate": 1.8415681525586143e-07,
      "loss": 0.7877,
      "step": 2086
    },
    {
      "epoch": 0.20631193930257272,
      "grad_norm": 5.111669500790089,
      "learning_rate": 1.8413951470023273e-07,
      "loss": 0.7644,
      "step": 2087
    },
    {
      "epoch": 0.20641079504732718,
      "grad_norm": 6.258121625823441,
      "learning_rate": 1.8412220551733793e-07,
      "loss": 0.8078,
      "step": 2088
    },
    {
      "epoch": 0.20650965079208167,
      "grad_norm": 8.997569899170418,
      "learning_rate": 1.8410488770895184e-07,
      "loss": 0.7153,
      "step": 2089
    },
    {
      "epoch": 0.20660850653683613,
      "grad_norm": 10.729357812280155,
      "learning_rate": 1.840875612768502e-07,
      "loss": 0.7624,
      "step": 2090
    },
    {
      "epoch": 0.2067073622815906,
      "grad_norm": 6.566868326808714,
      "learning_rate": 1.8407022622280948e-07,
      "loss": 0.7515,
      "step": 2091
    },
    {
      "epoch": 0.20680621802634505,
      "grad_norm": 45.55189711159325,
      "learning_rate": 1.8405288254860718e-07,
      "loss": 0.6676,
      "step": 2092
    },
    {
      "epoch": 0.2069050737710995,
      "grad_norm": 3.793282011424986,
      "learning_rate": 1.8403553025602167e-07,
      "loss": 0.7879,
      "step": 2093
    },
    {
      "epoch": 0.207003929515854,
      "grad_norm": 5.457850851607358,
      "learning_rate": 1.8401816934683212e-07,
      "loss": 0.7482,
      "step": 2094
    },
    {
      "epoch": 0.20710278526060846,
      "grad_norm": 3.3799164583540238,
      "learning_rate": 1.8400079982281868e-07,
      "loss": 0.7373,
      "step": 2095
    },
    {
      "epoch": 0.20720164100536292,
      "grad_norm": 4.767220049395196,
      "learning_rate": 1.839834216857623e-07,
      "loss": 0.8869,
      "step": 2096
    },
    {
      "epoch": 0.20730049675011739,
      "grad_norm": 4.142270179894253,
      "learning_rate": 1.8396603493744489e-07,
      "loss": 0.7879,
      "step": 2097
    },
    {
      "epoch": 0.20739935249487185,
      "grad_norm": 5.714830123396406,
      "learning_rate": 1.8394863957964917e-07,
      "loss": 0.7503,
      "step": 2098
    },
    {
      "epoch": 0.20749820823962634,
      "grad_norm": 3.5871396891535703,
      "learning_rate": 1.8393123561415877e-07,
      "loss": 0.797,
      "step": 2099
    },
    {
      "epoch": 0.2075970639843808,
      "grad_norm": 6.308185656596125,
      "learning_rate": 1.8391382304275826e-07,
      "loss": 0.7072,
      "step": 2100
    },
    {
      "epoch": 0.20769591972913526,
      "grad_norm": 5.525101712979395,
      "learning_rate": 1.83896401867233e-07,
      "loss": 0.764,
      "step": 2101
    },
    {
      "epoch": 0.20779477547388972,
      "grad_norm": 11.75651874126839,
      "learning_rate": 1.8387897208936931e-07,
      "loss": 0.6815,
      "step": 2102
    },
    {
      "epoch": 0.20789363121864418,
      "grad_norm": 6.980919786922329,
      "learning_rate": 1.838615337109543e-07,
      "loss": 0.7162,
      "step": 2103
    },
    {
      "epoch": 0.20799248696339867,
      "grad_norm": 3.5346411460993683,
      "learning_rate": 1.838440867337761e-07,
      "loss": 0.6865,
      "step": 2104
    },
    {
      "epoch": 0.20809134270815313,
      "grad_norm": 9.921758683914227,
      "learning_rate": 1.8382663115962352e-07,
      "loss": 0.7827,
      "step": 2105
    },
    {
      "epoch": 0.2081901984529076,
      "grad_norm": 4.6396282860501925,
      "learning_rate": 1.838091669902865e-07,
      "loss": 0.6941,
      "step": 2106
    },
    {
      "epoch": 0.20828905419766205,
      "grad_norm": 3.1688828878373907,
      "learning_rate": 1.8379169422755568e-07,
      "loss": 0.673,
      "step": 2107
    },
    {
      "epoch": 0.20838790994241652,
      "grad_norm": 4.756508313635189,
      "learning_rate": 1.8377421287322261e-07,
      "loss": 0.6044,
      "step": 2108
    },
    {
      "epoch": 0.208486765687171,
      "grad_norm": 6.79528214883216,
      "learning_rate": 1.8375672292907978e-07,
      "loss": 0.8208,
      "step": 2109
    },
    {
      "epoch": 0.20858562143192547,
      "grad_norm": 3.7616754942296913,
      "learning_rate": 1.8373922439692048e-07,
      "loss": 0.6928,
      "step": 2110
    },
    {
      "epoch": 0.20868447717667993,
      "grad_norm": 20.056356484600997,
      "learning_rate": 1.83721717278539e-07,
      "loss": 0.7198,
      "step": 2111
    },
    {
      "epoch": 0.2087833329214344,
      "grad_norm": 6.185995567145533,
      "learning_rate": 1.8370420157573036e-07,
      "loss": 0.7831,
      "step": 2112
    },
    {
      "epoch": 0.20888218866618885,
      "grad_norm": 12.977941978979313,
      "learning_rate": 1.836866772902906e-07,
      "loss": 0.7699,
      "step": 2113
    },
    {
      "epoch": 0.20898104441094334,
      "grad_norm": 3.675452620697515,
      "learning_rate": 1.8366914442401654e-07,
      "loss": 0.6464,
      "step": 2114
    },
    {
      "epoch": 0.2090799001556978,
      "grad_norm": 15.794609978215982,
      "learning_rate": 1.8365160297870593e-07,
      "loss": 0.6497,
      "step": 2115
    },
    {
      "epoch": 0.20917875590045226,
      "grad_norm": 5.76524704391002,
      "learning_rate": 1.8363405295615736e-07,
      "loss": 0.792,
      "step": 2116
    },
    {
      "epoch": 0.20927761164520672,
      "grad_norm": 3.7204594931732387,
      "learning_rate": 1.8361649435817038e-07,
      "loss": 0.7349,
      "step": 2117
    },
    {
      "epoch": 0.2093764673899612,
      "grad_norm": 4.199165174262191,
      "learning_rate": 1.835989271865453e-07,
      "loss": 0.7948,
      "step": 2118
    },
    {
      "epoch": 0.20947532313471567,
      "grad_norm": 4.517813469477539,
      "learning_rate": 1.8358135144308345e-07,
      "loss": 0.7161,
      "step": 2119
    },
    {
      "epoch": 0.20957417887947014,
      "grad_norm": 4.444002612523335,
      "learning_rate": 1.8356376712958687e-07,
      "loss": 0.7794,
      "step": 2120
    },
    {
      "epoch": 0.2096730346242246,
      "grad_norm": 4.641434700604214,
      "learning_rate": 1.8354617424785867e-07,
      "loss": 0.5881,
      "step": 2121
    },
    {
      "epoch": 0.20977189036897906,
      "grad_norm": 3.340730291696587,
      "learning_rate": 1.835285727997027e-07,
      "loss": 0.5884,
      "step": 2122
    },
    {
      "epoch": 0.20987074611373355,
      "grad_norm": 4.6119961131275335,
      "learning_rate": 1.8351096278692365e-07,
      "loss": 0.7077,
      "step": 2123
    },
    {
      "epoch": 0.209969601858488,
      "grad_norm": 3.2498345588957993,
      "learning_rate": 1.834933442113273e-07,
      "loss": 0.7009,
      "step": 2124
    },
    {
      "epoch": 0.21006845760324247,
      "grad_norm": 3.9087821897893225,
      "learning_rate": 1.834757170747201e-07,
      "loss": 0.7216,
      "step": 2125
    },
    {
      "epoch": 0.21016731334799693,
      "grad_norm": 3.8096058853236943,
      "learning_rate": 1.8345808137890947e-07,
      "loss": 0.6644,
      "step": 2126
    },
    {
      "epoch": 0.2102661690927514,
      "grad_norm": 3.1156212801168297,
      "learning_rate": 1.834404371257037e-07,
      "loss": 0.6929,
      "step": 2127
    },
    {
      "epoch": 0.21036502483750588,
      "grad_norm": 5.424580706755206,
      "learning_rate": 1.8342278431691192e-07,
      "loss": 0.7093,
      "step": 2128
    },
    {
      "epoch": 0.21046388058226034,
      "grad_norm": 14.967898638424883,
      "learning_rate": 1.8340512295434417e-07,
      "loss": 0.8553,
      "step": 2129
    },
    {
      "epoch": 0.2105627363270148,
      "grad_norm": 4.258055042342696,
      "learning_rate": 1.833874530398114e-07,
      "loss": 0.7685,
      "step": 2130
    },
    {
      "epoch": 0.21066159207176927,
      "grad_norm": 3.2329403046462635,
      "learning_rate": 1.8336977457512535e-07,
      "loss": 0.807,
      "step": 2131
    },
    {
      "epoch": 0.21076044781652373,
      "grad_norm": 4.793296715070051,
      "learning_rate": 1.8335208756209872e-07,
      "loss": 0.7893,
      "step": 2132
    },
    {
      "epoch": 0.21085930356127822,
      "grad_norm": 4.758018451015908,
      "learning_rate": 1.8333439200254505e-07,
      "loss": 0.7211,
      "step": 2133
    },
    {
      "epoch": 0.21095815930603268,
      "grad_norm": 2.96568219236192,
      "learning_rate": 1.8331668789827873e-07,
      "loss": 0.6547,
      "step": 2134
    },
    {
      "epoch": 0.21105701505078714,
      "grad_norm": 7.804179668992488,
      "learning_rate": 1.832989752511151e-07,
      "loss": 0.7053,
      "step": 2135
    },
    {
      "epoch": 0.2111558707955416,
      "grad_norm": 5.008373272229768,
      "learning_rate": 1.8328125406287026e-07,
      "loss": 0.7609,
      "step": 2136
    },
    {
      "epoch": 0.21125472654029606,
      "grad_norm": 3.5562725664885133,
      "learning_rate": 1.8326352433536135e-07,
      "loss": 0.6547,
      "step": 2137
    },
    {
      "epoch": 0.21135358228505055,
      "grad_norm": 2.482367584789509,
      "learning_rate": 1.832457860704062e-07,
      "loss": 0.6352,
      "step": 2138
    },
    {
      "epoch": 0.211452438029805,
      "grad_norm": 3.842954393673078,
      "learning_rate": 1.8322803926982368e-07,
      "loss": 0.6497,
      "step": 2139
    },
    {
      "epoch": 0.21155129377455947,
      "grad_norm": 3.6948440040270722,
      "learning_rate": 1.8321028393543344e-07,
      "loss": 0.8204,
      "step": 2140
    },
    {
      "epoch": 0.21165014951931393,
      "grad_norm": 3.655054804571866,
      "learning_rate": 1.8319252006905597e-07,
      "loss": 0.7908,
      "step": 2141
    },
    {
      "epoch": 0.2117490052640684,
      "grad_norm": 4.41506955968452,
      "learning_rate": 1.8317474767251277e-07,
      "loss": 0.6334,
      "step": 2142
    },
    {
      "epoch": 0.21184786100882289,
      "grad_norm": 6.528677333345299,
      "learning_rate": 1.8315696674762613e-07,
      "loss": 0.6998,
      "step": 2143
    },
    {
      "epoch": 0.21194671675357735,
      "grad_norm": 3.6893863536195544,
      "learning_rate": 1.8313917729621916e-07,
      "loss": 0.7459,
      "step": 2144
    },
    {
      "epoch": 0.2120455724983318,
      "grad_norm": 6.0064499056825165,
      "learning_rate": 1.83121379320116e-07,
      "loss": 0.7912,
      "step": 2145
    },
    {
      "epoch": 0.21214442824308627,
      "grad_norm": 7.23661761547919,
      "learning_rate": 1.8310357282114148e-07,
      "loss": 0.7834,
      "step": 2146
    },
    {
      "epoch": 0.21224328398784073,
      "grad_norm": 6.3072089601577,
      "learning_rate": 1.8308575780112143e-07,
      "loss": 0.7302,
      "step": 2147
    },
    {
      "epoch": 0.21234213973259522,
      "grad_norm": 3.44486679680075,
      "learning_rate": 1.8306793426188256e-07,
      "loss": 0.7526,
      "step": 2148
    },
    {
      "epoch": 0.21244099547734968,
      "grad_norm": 10.936906715756319,
      "learning_rate": 1.8305010220525232e-07,
      "loss": 0.7555,
      "step": 2149
    },
    {
      "epoch": 0.21253985122210414,
      "grad_norm": 3.937312233048026,
      "learning_rate": 1.830322616330592e-07,
      "loss": 0.7109,
      "step": 2150
    },
    {
      "epoch": 0.2126387069668586,
      "grad_norm": 78.99223907164522,
      "learning_rate": 1.8301441254713245e-07,
      "loss": 0.6832,
      "step": 2151
    },
    {
      "epoch": 0.21273756271161307,
      "grad_norm": 5.174156483408895,
      "learning_rate": 1.8299655494930225e-07,
      "loss": 0.693,
      "step": 2152
    },
    {
      "epoch": 0.21283641845636755,
      "grad_norm": 7.239649263197708,
      "learning_rate": 1.8297868884139964e-07,
      "loss": 0.5962,
      "step": 2153
    },
    {
      "epoch": 0.21293527420112202,
      "grad_norm": 5.7272916451380365,
      "learning_rate": 1.8296081422525648e-07,
      "loss": 0.655,
      "step": 2154
    },
    {
      "epoch": 0.21303412994587648,
      "grad_norm": 3.177520083211639,
      "learning_rate": 1.829429311027056e-07,
      "loss": 0.8213,
      "step": 2155
    },
    {
      "epoch": 0.21313298569063094,
      "grad_norm": 3.770164547246557,
      "learning_rate": 1.829250394755806e-07,
      "loss": 0.8228,
      "step": 2156
    },
    {
      "epoch": 0.21323184143538543,
      "grad_norm": 61.419029841970634,
      "learning_rate": 1.8290713934571609e-07,
      "loss": 0.7918,
      "step": 2157
    },
    {
      "epoch": 0.2133306971801399,
      "grad_norm": 5.344440704361083,
      "learning_rate": 1.8288923071494735e-07,
      "loss": 0.7851,
      "step": 2158
    },
    {
      "epoch": 0.21342955292489435,
      "grad_norm": 3.3540363263139494,
      "learning_rate": 1.8287131358511074e-07,
      "loss": 0.7482,
      "step": 2159
    },
    {
      "epoch": 0.2135284086696488,
      "grad_norm": 3.7330411823188365,
      "learning_rate": 1.8285338795804334e-07,
      "loss": 0.7143,
      "step": 2160
    },
    {
      "epoch": 0.21362726441440327,
      "grad_norm": 4.4114012223554555,
      "learning_rate": 1.828354538355832e-07,
      "loss": 0.6923,
      "step": 2161
    },
    {
      "epoch": 0.21372612015915776,
      "grad_norm": 6.746280681019429,
      "learning_rate": 1.8281751121956917e-07,
      "loss": 0.7905,
      "step": 2162
    },
    {
      "epoch": 0.21382497590391222,
      "grad_norm": 4.020923849067699,
      "learning_rate": 1.8279956011184097e-07,
      "loss": 0.7712,
      "step": 2163
    },
    {
      "epoch": 0.21392383164866668,
      "grad_norm": 7.997368122310902,
      "learning_rate": 1.827816005142393e-07,
      "loss": 0.7001,
      "step": 2164
    },
    {
      "epoch": 0.21402268739342115,
      "grad_norm": 4.631155727425905,
      "learning_rate": 1.827636324286056e-07,
      "loss": 0.7746,
      "step": 2165
    },
    {
      "epoch": 0.2141215431381756,
      "grad_norm": 8.748761337423502,
      "learning_rate": 1.8274565585678222e-07,
      "loss": 0.7391,
      "step": 2166
    },
    {
      "epoch": 0.2142203988829301,
      "grad_norm": 12.734442089521497,
      "learning_rate": 1.8272767080061246e-07,
      "loss": 0.7347,
      "step": 2167
    },
    {
      "epoch": 0.21431925462768456,
      "grad_norm": 6.428052270470054,
      "learning_rate": 1.8270967726194035e-07,
      "loss": 0.7977,
      "step": 2168
    },
    {
      "epoch": 0.21441811037243902,
      "grad_norm": 8.618949323725866,
      "learning_rate": 1.826916752426109e-07,
      "loss": 0.696,
      "step": 2169
    },
    {
      "epoch": 0.21451696611719348,
      "grad_norm": 3.2692778231826822,
      "learning_rate": 1.8267366474446995e-07,
      "loss": 0.6723,
      "step": 2170
    },
    {
      "epoch": 0.21461582186194794,
      "grad_norm": 23.356634351924736,
      "learning_rate": 1.826556457693642e-07,
      "loss": 0.8346,
      "step": 2171
    },
    {
      "epoch": 0.21471467760670243,
      "grad_norm": 4.906812233683138,
      "learning_rate": 1.8263761831914125e-07,
      "loss": 0.7472,
      "step": 2172
    },
    {
      "epoch": 0.2148135333514569,
      "grad_norm": 6.023416865971916,
      "learning_rate": 1.8261958239564948e-07,
      "loss": 0.7463,
      "step": 2173
    },
    {
      "epoch": 0.21491238909621135,
      "grad_norm": 2.7618035799401945,
      "learning_rate": 1.826015380007383e-07,
      "loss": 0.6531,
      "step": 2174
    },
    {
      "epoch": 0.21501124484096582,
      "grad_norm": 3.490987219972063,
      "learning_rate": 1.8258348513625785e-07,
      "loss": 0.7879,
      "step": 2175
    },
    {
      "epoch": 0.21511010058572028,
      "grad_norm": 3.3025358742403643,
      "learning_rate": 1.825654238040592e-07,
      "loss": 0.6393,
      "step": 2176
    },
    {
      "epoch": 0.21520895633047477,
      "grad_norm": 3.587798195825661,
      "learning_rate": 1.8254735400599423e-07,
      "loss": 0.7019,
      "step": 2177
    },
    {
      "epoch": 0.21530781207522923,
      "grad_norm": 6.84285102318608,
      "learning_rate": 1.8252927574391582e-07,
      "loss": 0.8288,
      "step": 2178
    },
    {
      "epoch": 0.2154066678199837,
      "grad_norm": 4.911645001686645,
      "learning_rate": 1.825111890196775e-07,
      "loss": 0.6884,
      "step": 2179
    },
    {
      "epoch": 0.21550552356473815,
      "grad_norm": 7.494434164130699,
      "learning_rate": 1.8249309383513395e-07,
      "loss": 0.6932,
      "step": 2180
    },
    {
      "epoch": 0.2156043793094926,
      "grad_norm": 3.943736510728258,
      "learning_rate": 1.8247499019214044e-07,
      "loss": 0.7381,
      "step": 2181
    },
    {
      "epoch": 0.2157032350542471,
      "grad_norm": 3.529616431930521,
      "learning_rate": 1.824568780925533e-07,
      "loss": 0.7876,
      "step": 2182
    },
    {
      "epoch": 0.21580209079900156,
      "grad_norm": 3.431223497160719,
      "learning_rate": 1.8243875753822962e-07,
      "loss": 0.6867,
      "step": 2183
    },
    {
      "epoch": 0.21590094654375602,
      "grad_norm": 33.86237858007758,
      "learning_rate": 1.8242062853102742e-07,
      "loss": 0.8048,
      "step": 2184
    },
    {
      "epoch": 0.21599980228851048,
      "grad_norm": 4.823276676023819,
      "learning_rate": 1.8240249107280555e-07,
      "loss": 0.7602,
      "step": 2185
    },
    {
      "epoch": 0.21609865803326495,
      "grad_norm": 3.5494244317799404,
      "learning_rate": 1.8238434516542372e-07,
      "loss": 0.7636,
      "step": 2186
    },
    {
      "epoch": 0.21619751377801943,
      "grad_norm": 5.038376640663892,
      "learning_rate": 1.8236619081074257e-07,
      "loss": 0.7625,
      "step": 2187
    },
    {
      "epoch": 0.2162963695227739,
      "grad_norm": 5.1634319073152914,
      "learning_rate": 1.8234802801062355e-07,
      "loss": 0.6188,
      "step": 2188
    },
    {
      "epoch": 0.21639522526752836,
      "grad_norm": 4.884125998810155,
      "learning_rate": 1.8232985676692896e-07,
      "loss": 0.7342,
      "step": 2189
    },
    {
      "epoch": 0.21649408101228282,
      "grad_norm": 22.110463935697503,
      "learning_rate": 1.8231167708152202e-07,
      "loss": 0.7291,
      "step": 2190
    },
    {
      "epoch": 0.21659293675703728,
      "grad_norm": 3.863379214738135,
      "learning_rate": 1.8229348895626677e-07,
      "loss": 0.7727,
      "step": 2191
    },
    {
      "epoch": 0.21669179250179177,
      "grad_norm": 25.23031394306092,
      "learning_rate": 1.8227529239302814e-07,
      "loss": 0.7705,
      "step": 2192
    },
    {
      "epoch": 0.21679064824654623,
      "grad_norm": 4.4999397945867505,
      "learning_rate": 1.8225708739367194e-07,
      "loss": 0.7869,
      "step": 2193
    },
    {
      "epoch": 0.2168895039913007,
      "grad_norm": 4.108292662260369,
      "learning_rate": 1.822388739600648e-07,
      "loss": 0.7164,
      "step": 2194
    },
    {
      "epoch": 0.21698835973605515,
      "grad_norm": 6.043449906829299,
      "learning_rate": 1.8222065209407421e-07,
      "loss": 0.744,
      "step": 2195
    },
    {
      "epoch": 0.21708721548080964,
      "grad_norm": 22.235348832982133,
      "learning_rate": 1.8220242179756863e-07,
      "loss": 0.8166,
      "step": 2196
    },
    {
      "epoch": 0.2171860712255641,
      "grad_norm": 4.339269132697207,
      "learning_rate": 1.8218418307241725e-07,
      "loss": 0.8546,
      "step": 2197
    },
    {
      "epoch": 0.21728492697031856,
      "grad_norm": 5.526685920232019,
      "learning_rate": 1.821659359204902e-07,
      "loss": 0.7148,
      "step": 2198
    },
    {
      "epoch": 0.21738378271507303,
      "grad_norm": 4.018299684782868,
      "learning_rate": 1.8214768034365845e-07,
      "loss": 0.7311,
      "step": 2199
    },
    {
      "epoch": 0.2174826384598275,
      "grad_norm": 4.841124887835498,
      "learning_rate": 1.8212941634379384e-07,
      "loss": 0.6494,
      "step": 2200
    },
    {
      "epoch": 0.21758149420458198,
      "grad_norm": 8.70689107260504,
      "learning_rate": 1.8211114392276907e-07,
      "loss": 0.7098,
      "step": 2201
    },
    {
      "epoch": 0.21768034994933644,
      "grad_norm": 4.740666919833462,
      "learning_rate": 1.8209286308245775e-07,
      "loss": 0.7426,
      "step": 2202
    },
    {
      "epoch": 0.2177792056940909,
      "grad_norm": 8.29916457451471,
      "learning_rate": 1.8207457382473426e-07,
      "loss": 0.6981,
      "step": 2203
    },
    {
      "epoch": 0.21787806143884536,
      "grad_norm": 3.199727396806589,
      "learning_rate": 1.820562761514739e-07,
      "loss": 0.6757,
      "step": 2204
    },
    {
      "epoch": 0.21797691718359982,
      "grad_norm": 4.331528777962173,
      "learning_rate": 1.8203797006455283e-07,
      "loss": 0.8289,
      "step": 2205
    },
    {
      "epoch": 0.2180757729283543,
      "grad_norm": 4.58579796110199,
      "learning_rate": 1.820196555658481e-07,
      "loss": 0.7684,
      "step": 2206
    },
    {
      "epoch": 0.21817462867310877,
      "grad_norm": 72.18523980245836,
      "learning_rate": 1.8200133265723756e-07,
      "loss": 0.7199,
      "step": 2207
    },
    {
      "epoch": 0.21827348441786323,
      "grad_norm": 2.9385078096013064,
      "learning_rate": 1.8198300134059994e-07,
      "loss": 0.8161,
      "step": 2208
    },
    {
      "epoch": 0.2183723401626177,
      "grad_norm": 8.5177614265311,
      "learning_rate": 1.8196466161781486e-07,
      "loss": 0.686,
      "step": 2209
    },
    {
      "epoch": 0.21847119590737216,
      "grad_norm": 4.580667717352341,
      "learning_rate": 1.8194631349076285e-07,
      "loss": 0.695,
      "step": 2210
    },
    {
      "epoch": 0.21857005165212665,
      "grad_norm": 3.56322058790244,
      "learning_rate": 1.8192795696132515e-07,
      "loss": 0.7411,
      "step": 2211
    },
    {
      "epoch": 0.2186689073968811,
      "grad_norm": 3.6400923621150367,
      "learning_rate": 1.8190959203138397e-07,
      "loss": 0.6811,
      "step": 2212
    },
    {
      "epoch": 0.21876776314163557,
      "grad_norm": 4.218950695708814,
      "learning_rate": 1.818912187028224e-07,
      "loss": 0.7433,
      "step": 2213
    },
    {
      "epoch": 0.21886661888639003,
      "grad_norm": 4.851248346687532,
      "learning_rate": 1.8187283697752432e-07,
      "loss": 0.721,
      "step": 2214
    },
    {
      "epoch": 0.2189654746311445,
      "grad_norm": 5.813175843892063,
      "learning_rate": 1.8185444685737452e-07,
      "loss": 0.7779,
      "step": 2215
    },
    {
      "epoch": 0.21906433037589898,
      "grad_norm": 6.5804617842054425,
      "learning_rate": 1.8183604834425865e-07,
      "loss": 0.6459,
      "step": 2216
    },
    {
      "epoch": 0.21916318612065344,
      "grad_norm": 6.433011453456284,
      "learning_rate": 1.8181764144006317e-07,
      "loss": 0.7164,
      "step": 2217
    },
    {
      "epoch": 0.2192620418654079,
      "grad_norm": 3.162140485708682,
      "learning_rate": 1.8179922614667548e-07,
      "loss": 0.8184,
      "step": 2218
    },
    {
      "epoch": 0.21936089761016236,
      "grad_norm": 4.0400889202512555,
      "learning_rate": 1.8178080246598372e-07,
      "loss": 0.7182,
      "step": 2219
    },
    {
      "epoch": 0.21945975335491683,
      "grad_norm": 5.040201357782888,
      "learning_rate": 1.8176237039987706e-07,
      "loss": 0.7525,
      "step": 2220
    },
    {
      "epoch": 0.21955860909967131,
      "grad_norm": 2.843395922223714,
      "learning_rate": 1.817439299502454e-07,
      "loss": 0.6454,
      "step": 2221
    },
    {
      "epoch": 0.21965746484442578,
      "grad_norm": 3.7689775570715414,
      "learning_rate": 1.817254811189795e-07,
      "loss": 0.7218,
      "step": 2222
    },
    {
      "epoch": 0.21975632058918024,
      "grad_norm": 3.4067846840258413,
      "learning_rate": 1.8170702390797107e-07,
      "loss": 0.6857,
      "step": 2223
    },
    {
      "epoch": 0.2198551763339347,
      "grad_norm": 4.411899866147998,
      "learning_rate": 1.8168855831911261e-07,
      "loss": 0.7113,
      "step": 2224
    },
    {
      "epoch": 0.21995403207868916,
      "grad_norm": 4.294685321888325,
      "learning_rate": 1.8167008435429747e-07,
      "loss": 0.7546,
      "step": 2225
    },
    {
      "epoch": 0.22005288782344365,
      "grad_norm": 7.83924496395221,
      "learning_rate": 1.816516020154199e-07,
      "loss": 0.6298,
      "step": 2226
    },
    {
      "epoch": 0.2201517435681981,
      "grad_norm": 5.0500929246539785,
      "learning_rate": 1.8163311130437496e-07,
      "loss": 0.783,
      "step": 2227
    },
    {
      "epoch": 0.22025059931295257,
      "grad_norm": 4.414900404650592,
      "learning_rate": 1.8161461222305866e-07,
      "loss": 0.6177,
      "step": 2228
    },
    {
      "epoch": 0.22034945505770703,
      "grad_norm": 6.038395985833523,
      "learning_rate": 1.815961047733678e-07,
      "loss": 0.7146,
      "step": 2229
    },
    {
      "epoch": 0.2204483108024615,
      "grad_norm": 5.794213359705727,
      "learning_rate": 1.8157758895719999e-07,
      "loss": 0.6268,
      "step": 2230
    },
    {
      "epoch": 0.22054716654721598,
      "grad_norm": 4.667437999322638,
      "learning_rate": 1.815590647764538e-07,
      "loss": 0.8057,
      "step": 2231
    },
    {
      "epoch": 0.22064602229197045,
      "grad_norm": 7.6866328362622,
      "learning_rate": 1.815405322330286e-07,
      "loss": 0.7478,
      "step": 2232
    },
    {
      "epoch": 0.2207448780367249,
      "grad_norm": 7.046859814249418,
      "learning_rate": 1.8152199132882462e-07,
      "loss": 0.796,
      "step": 2233
    },
    {
      "epoch": 0.22084373378147937,
      "grad_norm": 5.13464041961843,
      "learning_rate": 1.8150344206574296e-07,
      "loss": 0.834,
      "step": 2234
    },
    {
      "epoch": 0.22094258952623386,
      "grad_norm": 3.6557481382908774,
      "learning_rate": 1.8148488444568557e-07,
      "loss": 0.6793,
      "step": 2235
    },
    {
      "epoch": 0.22104144527098832,
      "grad_norm": 4.943340127729641,
      "learning_rate": 1.814663184705553e-07,
      "loss": 0.758,
      "step": 2236
    },
    {
      "epoch": 0.22114030101574278,
      "grad_norm": 9.404996615559199,
      "learning_rate": 1.8144774414225574e-07,
      "loss": 0.8042,
      "step": 2237
    },
    {
      "epoch": 0.22123915676049724,
      "grad_norm": 3.4347086295708706,
      "learning_rate": 1.814291614626915e-07,
      "loss": 0.7619,
      "step": 2238
    },
    {
      "epoch": 0.2213380125052517,
      "grad_norm": 4.723037044603319,
      "learning_rate": 1.8141057043376789e-07,
      "loss": 0.7029,
      "step": 2239
    },
    {
      "epoch": 0.2214368682500062,
      "grad_norm": 3.9063199651559506,
      "learning_rate": 1.8139197105739118e-07,
      "loss": 0.6983,
      "step": 2240
    },
    {
      "epoch": 0.22153572399476065,
      "grad_norm": 3.3986448765300294,
      "learning_rate": 1.8137336333546842e-07,
      "loss": 0.8021,
      "step": 2241
    },
    {
      "epoch": 0.22163457973951511,
      "grad_norm": 4.430915242472711,
      "learning_rate": 1.8135474726990764e-07,
      "loss": 0.6895,
      "step": 2242
    },
    {
      "epoch": 0.22173343548426958,
      "grad_norm": 3.583820231953911,
      "learning_rate": 1.8133612286261758e-07,
      "loss": 0.7229,
      "step": 2243
    },
    {
      "epoch": 0.22183229122902404,
      "grad_norm": 5.420435602322517,
      "learning_rate": 1.813174901155079e-07,
      "loss": 0.6538,
      "step": 2244
    },
    {
      "epoch": 0.22193114697377853,
      "grad_norm": 7.461733121188919,
      "learning_rate": 1.8129884903048913e-07,
      "loss": 0.713,
      "step": 2245
    },
    {
      "epoch": 0.222030002718533,
      "grad_norm": 7.302142482531213,
      "learning_rate": 1.8128019960947266e-07,
      "loss": 0.6695,
      "step": 2246
    },
    {
      "epoch": 0.22212885846328745,
      "grad_norm": 7.512341595921433,
      "learning_rate": 1.8126154185437066e-07,
      "loss": 0.6403,
      "step": 2247
    },
    {
      "epoch": 0.2222277142080419,
      "grad_norm": 11.401539791865352,
      "learning_rate": 1.8124287576709622e-07,
      "loss": 0.7208,
      "step": 2248
    },
    {
      "epoch": 0.22232656995279637,
      "grad_norm": 4.602699245826749,
      "learning_rate": 1.8122420134956332e-07,
      "loss": 0.677,
      "step": 2249
    },
    {
      "epoch": 0.22242542569755086,
      "grad_norm": 6.5364598443905715,
      "learning_rate": 1.8120551860368672e-07,
      "loss": 0.6142,
      "step": 2250
    },
    {
      "epoch": 0.22252428144230532,
      "grad_norm": 4.855845394062281,
      "learning_rate": 1.8118682753138202e-07,
      "loss": 0.7401,
      "step": 2251
    },
    {
      "epoch": 0.22262313718705978,
      "grad_norm": 3.7295662275728017,
      "learning_rate": 1.8116812813456575e-07,
      "loss": 0.6516,
      "step": 2252
    },
    {
      "epoch": 0.22272199293181424,
      "grad_norm": 11.614116434831002,
      "learning_rate": 1.8114942041515527e-07,
      "loss": 0.7183,
      "step": 2253
    },
    {
      "epoch": 0.2228208486765687,
      "grad_norm": 4.0489110019320735,
      "learning_rate": 1.8113070437506878e-07,
      "loss": 0.7919,
      "step": 2254
    },
    {
      "epoch": 0.2229197044213232,
      "grad_norm": 16.538539406264523,
      "learning_rate": 1.811119800162253e-07,
      "loss": 0.7205,
      "step": 2255
    },
    {
      "epoch": 0.22301856016607766,
      "grad_norm": 3.908051445386847,
      "learning_rate": 1.8109324734054474e-07,
      "loss": 0.7822,
      "step": 2256
    },
    {
      "epoch": 0.22311741591083212,
      "grad_norm": 29.725503523381,
      "learning_rate": 1.8107450634994788e-07,
      "loss": 0.6996,
      "step": 2257
    },
    {
      "epoch": 0.22321627165558658,
      "grad_norm": 4.326472687580419,
      "learning_rate": 1.8105575704635634e-07,
      "loss": 0.7383,
      "step": 2258
    },
    {
      "epoch": 0.22331512740034104,
      "grad_norm": 3.4411171326858963,
      "learning_rate": 1.8103699943169257e-07,
      "loss": 0.7789,
      "step": 2259
    },
    {
      "epoch": 0.22341398314509553,
      "grad_norm": 4.066362488064526,
      "learning_rate": 1.8101823350787988e-07,
      "loss": 0.7288,
      "step": 2260
    },
    {
      "epoch": 0.22351283888985,
      "grad_norm": 6.469017007670019,
      "learning_rate": 1.8099945927684246e-07,
      "loss": 0.6703,
      "step": 2261
    },
    {
      "epoch": 0.22361169463460445,
      "grad_norm": 4.047195437170868,
      "learning_rate": 1.809806767405053e-07,
      "loss": 0.7092,
      "step": 2262
    },
    {
      "epoch": 0.2237105503793589,
      "grad_norm": 7.197360406271074,
      "learning_rate": 1.809618859007943e-07,
      "loss": 0.6652,
      "step": 2263
    },
    {
      "epoch": 0.22380940612411337,
      "grad_norm": 5.703925639337483,
      "learning_rate": 1.8094308675963617e-07,
      "loss": 0.7526,
      "step": 2264
    },
    {
      "epoch": 0.22390826186886786,
      "grad_norm": 6.763971887388284,
      "learning_rate": 1.8092427931895848e-07,
      "loss": 0.706,
      "step": 2265
    },
    {
      "epoch": 0.22400711761362233,
      "grad_norm": 3.6348701586371193,
      "learning_rate": 1.809054635806897e-07,
      "loss": 0.7149,
      "step": 2266
    },
    {
      "epoch": 0.2241059733583768,
      "grad_norm": 5.826371786519916,
      "learning_rate": 1.80886639546759e-07,
      "loss": 0.6991,
      "step": 2267
    },
    {
      "epoch": 0.22420482910313125,
      "grad_norm": 4.788298193014567,
      "learning_rate": 1.8086780721909663e-07,
      "loss": 0.6116,
      "step": 2268
    },
    {
      "epoch": 0.2243036848478857,
      "grad_norm": 3.3951348734656994,
      "learning_rate": 1.808489665996335e-07,
      "loss": 0.6954,
      "step": 2269
    },
    {
      "epoch": 0.2244025405926402,
      "grad_norm": 3.0269984561207397,
      "learning_rate": 1.8083011769030145e-07,
      "loss": 0.6668,
      "step": 2270
    },
    {
      "epoch": 0.22450139633739466,
      "grad_norm": 5.399258655302236,
      "learning_rate": 1.8081126049303314e-07,
      "loss": 0.742,
      "step": 2271
    },
    {
      "epoch": 0.22460025208214912,
      "grad_norm": 4.375264209829578,
      "learning_rate": 1.8079239500976216e-07,
      "loss": 0.6423,
      "step": 2272
    },
    {
      "epoch": 0.22469910782690358,
      "grad_norm": 4.838705580265416,
      "learning_rate": 1.8077352124242283e-07,
      "loss": 0.714,
      "step": 2273
    },
    {
      "epoch": 0.22479796357165807,
      "grad_norm": 3.165851657635201,
      "learning_rate": 1.8075463919295037e-07,
      "loss": 0.7254,
      "step": 2274
    },
    {
      "epoch": 0.22489681931641253,
      "grad_norm": 4.710963170588682,
      "learning_rate": 1.807357488632809e-07,
      "loss": 0.6338,
      "step": 2275
    },
    {
      "epoch": 0.224995675061167,
      "grad_norm": 5.178372708029737,
      "learning_rate": 1.807168502553513e-07,
      "loss": 0.701,
      "step": 2276
    },
    {
      "epoch": 0.22509453080592146,
      "grad_norm": 4.107801873224287,
      "learning_rate": 1.8069794337109935e-07,
      "loss": 0.7084,
      "step": 2277
    },
    {
      "epoch": 0.22519338655067592,
      "grad_norm": 5.679366561806041,
      "learning_rate": 1.8067902821246373e-07,
      "loss": 0.6844,
      "step": 2278
    },
    {
      "epoch": 0.2252922422954304,
      "grad_norm": 6.59816162929532,
      "learning_rate": 1.8066010478138383e-07,
      "loss": 0.8134,
      "step": 2279
    },
    {
      "epoch": 0.22539109804018487,
      "grad_norm": 3.3273584325408856,
      "learning_rate": 1.806411730798e-07,
      "loss": 0.6035,
      "step": 2280
    },
    {
      "epoch": 0.22548995378493933,
      "grad_norm": 4.929975644030452,
      "learning_rate": 1.8062223310965346e-07,
      "loss": 0.7126,
      "step": 2281
    },
    {
      "epoch": 0.2255888095296938,
      "grad_norm": 4.837395406386643,
      "learning_rate": 1.8060328487288613e-07,
      "loss": 0.7736,
      "step": 2282
    },
    {
      "epoch": 0.22568766527444825,
      "grad_norm": 6.744421903709185,
      "learning_rate": 1.8058432837144094e-07,
      "loss": 0.7378,
      "step": 2283
    },
    {
      "epoch": 0.22578652101920274,
      "grad_norm": 11.331873234553592,
      "learning_rate": 1.8056536360726155e-07,
      "loss": 0.836,
      "step": 2284
    },
    {
      "epoch": 0.2258853767639572,
      "grad_norm": 7.034215524006916,
      "learning_rate": 1.8054639058229258e-07,
      "loss": 0.6975,
      "step": 2285
    },
    {
      "epoch": 0.22598423250871166,
      "grad_norm": 3.697899115301812,
      "learning_rate": 1.805274092984794e-07,
      "loss": 0.7234,
      "step": 2286
    },
    {
      "epoch": 0.22608308825346612,
      "grad_norm": 4.423837218013055,
      "learning_rate": 1.8050841975776822e-07,
      "loss": 0.7686,
      "step": 2287
    },
    {
      "epoch": 0.22618194399822059,
      "grad_norm": 17.753626327998095,
      "learning_rate": 1.804894219621062e-07,
      "loss": 0.7287,
      "step": 2288
    },
    {
      "epoch": 0.22628079974297508,
      "grad_norm": 3.9442940828468473,
      "learning_rate": 1.8047041591344128e-07,
      "loss": 0.8078,
      "step": 2289
    },
    {
      "epoch": 0.22637965548772954,
      "grad_norm": 6.0976288445848,
      "learning_rate": 1.804514016137222e-07,
      "loss": 0.6417,
      "step": 2290
    },
    {
      "epoch": 0.226478511232484,
      "grad_norm": 6.562205855032461,
      "learning_rate": 1.8043237906489864e-07,
      "loss": 0.7282,
      "step": 2291
    },
    {
      "epoch": 0.22657736697723846,
      "grad_norm": 17.434091864506833,
      "learning_rate": 1.8041334826892106e-07,
      "loss": 0.8158,
      "step": 2292
    },
    {
      "epoch": 0.22667622272199292,
      "grad_norm": 4.908122533680318,
      "learning_rate": 1.803943092277408e-07,
      "loss": 0.7187,
      "step": 2293
    },
    {
      "epoch": 0.2267750784667474,
      "grad_norm": 3.482336274221947,
      "learning_rate": 1.8037526194331001e-07,
      "loss": 0.7715,
      "step": 2294
    },
    {
      "epoch": 0.22687393421150187,
      "grad_norm": 3.9402111758658602,
      "learning_rate": 1.8035620641758175e-07,
      "loss": 0.6528,
      "step": 2295
    },
    {
      "epoch": 0.22697278995625633,
      "grad_norm": 5.392592682095414,
      "learning_rate": 1.8033714265250987e-07,
      "loss": 0.7167,
      "step": 2296
    },
    {
      "epoch": 0.2270716457010108,
      "grad_norm": 4.812434834058206,
      "learning_rate": 1.8031807065004906e-07,
      "loss": 0.7223,
      "step": 2297
    },
    {
      "epoch": 0.22717050144576525,
      "grad_norm": 5.080801626880967,
      "learning_rate": 1.8029899041215483e-07,
      "loss": 0.6515,
      "step": 2298
    },
    {
      "epoch": 0.22726935719051974,
      "grad_norm": 3.462825250797019,
      "learning_rate": 1.8027990194078367e-07,
      "loss": 0.7081,
      "step": 2299
    },
    {
      "epoch": 0.2273682129352742,
      "grad_norm": 3.671139786738081,
      "learning_rate": 1.8026080523789279e-07,
      "loss": 0.8255,
      "step": 2300
    },
    {
      "epoch": 0.22746706868002867,
      "grad_norm": 33.89805121425542,
      "learning_rate": 1.8024170030544027e-07,
      "loss": 0.5625,
      "step": 2301
    },
    {
      "epoch": 0.22756592442478313,
      "grad_norm": 3.526006402270299,
      "learning_rate": 1.80222587145385e-07,
      "loss": 0.7136,
      "step": 2302
    },
    {
      "epoch": 0.2276647801695376,
      "grad_norm": 3.3260083601149515,
      "learning_rate": 1.8020346575968679e-07,
      "loss": 0.6723,
      "step": 2303
    },
    {
      "epoch": 0.22776363591429208,
      "grad_norm": 3.14714515282056,
      "learning_rate": 1.8018433615030626e-07,
      "loss": 0.6868,
      "step": 2304
    },
    {
      "epoch": 0.22786249165904654,
      "grad_norm": 3.5529878995590622,
      "learning_rate": 1.8016519831920485e-07,
      "loss": 0.6026,
      "step": 2305
    },
    {
      "epoch": 0.227961347403801,
      "grad_norm": 5.320884115235474,
      "learning_rate": 1.801460522683449e-07,
      "loss": 0.7869,
      "step": 2306
    },
    {
      "epoch": 0.22806020314855546,
      "grad_norm": 4.165275900460274,
      "learning_rate": 1.801268979996895e-07,
      "loss": 0.7834,
      "step": 2307
    },
    {
      "epoch": 0.22815905889330992,
      "grad_norm": 3.2649217933485604,
      "learning_rate": 1.8010773551520266e-07,
      "loss": 0.8007,
      "step": 2308
    },
    {
      "epoch": 0.2282579146380644,
      "grad_norm": 2.8281277234448914,
      "learning_rate": 1.8008856481684922e-07,
      "loss": 0.8039,
      "step": 2309
    },
    {
      "epoch": 0.22835677038281887,
      "grad_norm": 3.7238205726417104,
      "learning_rate": 1.8006938590659484e-07,
      "loss": 0.7633,
      "step": 2310
    },
    {
      "epoch": 0.22845562612757334,
      "grad_norm": 3.256446732220314,
      "learning_rate": 1.800501987864061e-07,
      "loss": 0.7141,
      "step": 2311
    },
    {
      "epoch": 0.2285544818723278,
      "grad_norm": 5.506007748468183,
      "learning_rate": 1.8003100345825021e-07,
      "loss": 0.7768,
      "step": 2312
    },
    {
      "epoch": 0.2286533376170823,
      "grad_norm": 11.065627113870244,
      "learning_rate": 1.8001179992409552e-07,
      "loss": 0.7111,
      "step": 2313
    },
    {
      "epoch": 0.22875219336183675,
      "grad_norm": 31.425612251565877,
      "learning_rate": 1.7999258818591098e-07,
      "loss": 0.8612,
      "step": 2314
    },
    {
      "epoch": 0.2288510491065912,
      "grad_norm": 4.775154158123451,
      "learning_rate": 1.7997336824566648e-07,
      "loss": 0.7068,
      "step": 2315
    },
    {
      "epoch": 0.22894990485134567,
      "grad_norm": 6.043512788077851,
      "learning_rate": 1.7995414010533279e-07,
      "loss": 0.6667,
      "step": 2316
    },
    {
      "epoch": 0.22904876059610013,
      "grad_norm": 3.2202504544162602,
      "learning_rate": 1.7993490376688142e-07,
      "loss": 0.7319,
      "step": 2317
    },
    {
      "epoch": 0.22914761634085462,
      "grad_norm": 3.1033158319487675,
      "learning_rate": 1.799156592322848e-07,
      "loss": 0.6512,
      "step": 2318
    },
    {
      "epoch": 0.22924647208560908,
      "grad_norm": 2.7598582404283043,
      "learning_rate": 1.7989640650351617e-07,
      "loss": 0.6186,
      "step": 2319
    },
    {
      "epoch": 0.22934532783036354,
      "grad_norm": 4.966225277743665,
      "learning_rate": 1.798771455825496e-07,
      "loss": 0.7385,
      "step": 2320
    },
    {
      "epoch": 0.229444183575118,
      "grad_norm": 3.9072878009258236,
      "learning_rate": 1.7985787647135999e-07,
      "loss": 0.703,
      "step": 2321
    },
    {
      "epoch": 0.22954303931987247,
      "grad_norm": 3.6469386914853854,
      "learning_rate": 1.7983859917192318e-07,
      "loss": 0.71,
      "step": 2322
    },
    {
      "epoch": 0.22964189506462696,
      "grad_norm": 3.6043384478092997,
      "learning_rate": 1.798193136862157e-07,
      "loss": 0.724,
      "step": 2323
    },
    {
      "epoch": 0.22974075080938142,
      "grad_norm": 3.0602221544184474,
      "learning_rate": 1.7980002001621504e-07,
      "loss": 0.7766,
      "step": 2324
    },
    {
      "epoch": 0.22983960655413588,
      "grad_norm": 4.01702872145419,
      "learning_rate": 1.7978071816389947e-07,
      "loss": 0.6897,
      "step": 2325
    },
    {
      "epoch": 0.22993846229889034,
      "grad_norm": 13.66378925802652,
      "learning_rate": 1.7976140813124806e-07,
      "loss": 0.7134,
      "step": 2326
    },
    {
      "epoch": 0.2300373180436448,
      "grad_norm": 4.6438545677081455,
      "learning_rate": 1.7974208992024081e-07,
      "loss": 0.8275,
      "step": 2327
    },
    {
      "epoch": 0.2301361737883993,
      "grad_norm": 6.7419453227795385,
      "learning_rate": 1.7972276353285857e-07,
      "loss": 0.7333,
      "step": 2328
    },
    {
      "epoch": 0.23023502953315375,
      "grad_norm": 18.36546642927546,
      "learning_rate": 1.797034289710829e-07,
      "loss": 0.8672,
      "step": 2329
    },
    {
      "epoch": 0.2303338852779082,
      "grad_norm": 4.956906007056394,
      "learning_rate": 1.7968408623689627e-07,
      "loss": 0.7237,
      "step": 2330
    },
    {
      "epoch": 0.23043274102266267,
      "grad_norm": 3.0788188220679324,
      "learning_rate": 1.7966473533228203e-07,
      "loss": 0.6961,
      "step": 2331
    },
    {
      "epoch": 0.23053159676741714,
      "grad_norm": 7.094401340209455,
      "learning_rate": 1.7964537625922433e-07,
      "loss": 0.6539,
      "step": 2332
    },
    {
      "epoch": 0.23063045251217162,
      "grad_norm": 7.459589007777191,
      "learning_rate": 1.7962600901970816e-07,
      "loss": 0.7682,
      "step": 2333
    },
    {
      "epoch": 0.23072930825692609,
      "grad_norm": 3.3939079431698427,
      "learning_rate": 1.7960663361571933e-07,
      "loss": 0.7493,
      "step": 2334
    },
    {
      "epoch": 0.23082816400168055,
      "grad_norm": 4.226424166004542,
      "learning_rate": 1.795872500492445e-07,
      "loss": 0.7842,
      "step": 2335
    },
    {
      "epoch": 0.230927019746435,
      "grad_norm": 6.6913751353105635,
      "learning_rate": 1.7956785832227113e-07,
      "loss": 0.7799,
      "step": 2336
    },
    {
      "epoch": 0.23102587549118947,
      "grad_norm": 4.635867380265187,
      "learning_rate": 1.7954845843678763e-07,
      "loss": 0.6994,
      "step": 2337
    },
    {
      "epoch": 0.23112473123594396,
      "grad_norm": 3.504023533105126,
      "learning_rate": 1.7952905039478315e-07,
      "loss": 0.7593,
      "step": 2338
    },
    {
      "epoch": 0.23122358698069842,
      "grad_norm": 4.7814818653627045,
      "learning_rate": 1.7950963419824768e-07,
      "loss": 0.7856,
      "step": 2339
    },
    {
      "epoch": 0.23132244272545288,
      "grad_norm": 5.507026970149726,
      "learning_rate": 1.7949020984917204e-07,
      "loss": 0.7714,
      "step": 2340
    },
    {
      "epoch": 0.23142129847020734,
      "grad_norm": 4.654546938506628,
      "learning_rate": 1.79470777349548e-07,
      "loss": 0.7142,
      "step": 2341
    },
    {
      "epoch": 0.2315201542149618,
      "grad_norm": 4.898484545939503,
      "learning_rate": 1.7945133670136796e-07,
      "loss": 0.8338,
      "step": 2342
    },
    {
      "epoch": 0.2316190099597163,
      "grad_norm": 3.658022149572809,
      "learning_rate": 1.7943188790662537e-07,
      "loss": 0.7558,
      "step": 2343
    },
    {
      "epoch": 0.23171786570447075,
      "grad_norm": 3.9307864321389063,
      "learning_rate": 1.7941243096731437e-07,
      "loss": 0.6032,
      "step": 2344
    },
    {
      "epoch": 0.23181672144922522,
      "grad_norm": 7.731591997601774,
      "learning_rate": 1.7939296588542999e-07,
      "loss": 0.8005,
      "step": 2345
    },
    {
      "epoch": 0.23191557719397968,
      "grad_norm": 3.85219343279815,
      "learning_rate": 1.793734926629681e-07,
      "loss": 0.6628,
      "step": 2346
    },
    {
      "epoch": 0.23201443293873414,
      "grad_norm": 6.633923508841117,
      "learning_rate": 1.7935401130192536e-07,
      "loss": 0.7895,
      "step": 2347
    },
    {
      "epoch": 0.23211328868348863,
      "grad_norm": 5.36892087512545,
      "learning_rate": 1.7933452180429933e-07,
      "loss": 0.7296,
      "step": 2348
    },
    {
      "epoch": 0.2322121444282431,
      "grad_norm": 37.54816536924243,
      "learning_rate": 1.7931502417208834e-07,
      "loss": 0.6484,
      "step": 2349
    },
    {
      "epoch": 0.23231100017299755,
      "grad_norm": 5.0667391259538945,
      "learning_rate": 1.792955184072916e-07,
      "loss": 0.7387,
      "step": 2350
    },
    {
      "epoch": 0.232409855917752,
      "grad_norm": 3.1678264686625455,
      "learning_rate": 1.7927600451190914e-07,
      "loss": 0.7885,
      "step": 2351
    },
    {
      "epoch": 0.2325087116625065,
      "grad_norm": 3.7693286044633396,
      "learning_rate": 1.7925648248794185e-07,
      "loss": 0.7022,
      "step": 2352
    },
    {
      "epoch": 0.23260756740726096,
      "grad_norm": 4.289063185685198,
      "learning_rate": 1.7923695233739134e-07,
      "loss": 0.731,
      "step": 2353
    },
    {
      "epoch": 0.23270642315201542,
      "grad_norm": 3.906410943401662,
      "learning_rate": 1.7921741406226026e-07,
      "loss": 0.7678,
      "step": 2354
    },
    {
      "epoch": 0.23280527889676988,
      "grad_norm": 23.97322630909973,
      "learning_rate": 1.7919786766455186e-07,
      "loss": 0.6156,
      "step": 2355
    },
    {
      "epoch": 0.23290413464152435,
      "grad_norm": 8.939167568433776,
      "learning_rate": 1.7917831314627038e-07,
      "loss": 0.6808,
      "step": 2356
    },
    {
      "epoch": 0.23300299038627884,
      "grad_norm": 3.589132537500845,
      "learning_rate": 1.7915875050942088e-07,
      "loss": 0.6517,
      "step": 2357
    },
    {
      "epoch": 0.2331018461310333,
      "grad_norm": 4.162590781641654,
      "learning_rate": 1.7913917975600916e-07,
      "loss": 0.7593,
      "step": 2358
    },
    {
      "epoch": 0.23320070187578776,
      "grad_norm": 3.153506456844686,
      "learning_rate": 1.7911960088804198e-07,
      "loss": 0.7614,
      "step": 2359
    },
    {
      "epoch": 0.23329955762054222,
      "grad_norm": 4.068563217201816,
      "learning_rate": 1.791000139075268e-07,
      "loss": 0.7418,
      "step": 2360
    },
    {
      "epoch": 0.23339841336529668,
      "grad_norm": 11.211266619277687,
      "learning_rate": 1.7908041881647197e-07,
      "loss": 0.7118,
      "step": 2361
    },
    {
      "epoch": 0.23349726911005117,
      "grad_norm": 5.808212732113187,
      "learning_rate": 1.7906081561688675e-07,
      "loss": 0.7331,
      "step": 2362
    },
    {
      "epoch": 0.23359612485480563,
      "grad_norm": 3.6482394240581324,
      "learning_rate": 1.7904120431078112e-07,
      "loss": 0.6094,
      "step": 2363
    },
    {
      "epoch": 0.2336949805995601,
      "grad_norm": 5.829953011390809,
      "learning_rate": 1.790215849001659e-07,
      "loss": 0.7617,
      "step": 2364
    },
    {
      "epoch": 0.23379383634431455,
      "grad_norm": 3.731916947125293,
      "learning_rate": 1.7900195738705285e-07,
      "loss": 0.8435,
      "step": 2365
    },
    {
      "epoch": 0.23389269208906902,
      "grad_norm": 4.369194243787675,
      "learning_rate": 1.7898232177345438e-07,
      "loss": 0.7434,
      "step": 2366
    },
    {
      "epoch": 0.2339915478338235,
      "grad_norm": 4.052027796294193,
      "learning_rate": 1.789626780613839e-07,
      "loss": 0.6862,
      "step": 2367
    },
    {
      "epoch": 0.23409040357857797,
      "grad_norm": 3.01400142277909,
      "learning_rate": 1.7894302625285557e-07,
      "loss": 0.6983,
      "step": 2368
    },
    {
      "epoch": 0.23418925932333243,
      "grad_norm": 6.284158009203555,
      "learning_rate": 1.7892336634988437e-07,
      "loss": 0.6354,
      "step": 2369
    },
    {
      "epoch": 0.2342881150680869,
      "grad_norm": 7.339184832533638,
      "learning_rate": 1.789036983544862e-07,
      "loss": 0.762,
      "step": 2370
    },
    {
      "epoch": 0.23438697081284135,
      "grad_norm": 3.8723245608623165,
      "learning_rate": 1.788840222686776e-07,
      "loss": 0.78,
      "step": 2371
    },
    {
      "epoch": 0.23448582655759584,
      "grad_norm": 3.7307895594635796,
      "learning_rate": 1.788643380944762e-07,
      "loss": 0.736,
      "step": 2372
    },
    {
      "epoch": 0.2345846823023503,
      "grad_norm": 3.687147335894984,
      "learning_rate": 1.7884464583390022e-07,
      "loss": 0.8017,
      "step": 2373
    },
    {
      "epoch": 0.23468353804710476,
      "grad_norm": 13.830887503214711,
      "learning_rate": 1.7882494548896888e-07,
      "loss": 0.6776,
      "step": 2374
    },
    {
      "epoch": 0.23478239379185922,
      "grad_norm": 5.661337264896834,
      "learning_rate": 1.788052370617021e-07,
      "loss": 0.7208,
      "step": 2375
    },
    {
      "epoch": 0.23488124953661368,
      "grad_norm": 5.197337976405989,
      "learning_rate": 1.7878552055412075e-07,
      "loss": 0.6372,
      "step": 2376
    },
    {
      "epoch": 0.23498010528136817,
      "grad_norm": 3.9043732366999406,
      "learning_rate": 1.787657959682464e-07,
      "loss": 0.6534,
      "step": 2377
    },
    {
      "epoch": 0.23507896102612263,
      "grad_norm": 4.0907893114550005,
      "learning_rate": 1.7874606330610158e-07,
      "loss": 0.6868,
      "step": 2378
    },
    {
      "epoch": 0.2351778167708771,
      "grad_norm": 3.6462265429114384,
      "learning_rate": 1.7872632256970955e-07,
      "loss": 0.5791,
      "step": 2379
    },
    {
      "epoch": 0.23527667251563156,
      "grad_norm": 4.951975658320117,
      "learning_rate": 1.7870657376109444e-07,
      "loss": 0.6944,
      "step": 2380
    },
    {
      "epoch": 0.23537552826038602,
      "grad_norm": 2.942556561569832,
      "learning_rate": 1.7868681688228118e-07,
      "loss": 0.6583,
      "step": 2381
    },
    {
      "epoch": 0.2354743840051405,
      "grad_norm": 3.431049890053801,
      "learning_rate": 1.7866705193529557e-07,
      "loss": 0.7448,
      "step": 2382
    },
    {
      "epoch": 0.23557323974989497,
      "grad_norm": 9.564637596281061,
      "learning_rate": 1.7864727892216421e-07,
      "loss": 0.7544,
      "step": 2383
    },
    {
      "epoch": 0.23567209549464943,
      "grad_norm": 3.0651565406088457,
      "learning_rate": 1.7862749784491452e-07,
      "loss": 0.6529,
      "step": 2384
    },
    {
      "epoch": 0.2357709512394039,
      "grad_norm": 4.581361277514506,
      "learning_rate": 1.7860770870557475e-07,
      "loss": 0.7145,
      "step": 2385
    },
    {
      "epoch": 0.23586980698415835,
      "grad_norm": 6.469513367778448,
      "learning_rate": 1.78587911506174e-07,
      "loss": 0.7533,
      "step": 2386
    },
    {
      "epoch": 0.23596866272891284,
      "grad_norm": 3.850491056571681,
      "learning_rate": 1.7856810624874219e-07,
      "loss": 0.8474,
      "step": 2387
    },
    {
      "epoch": 0.2360675184736673,
      "grad_norm": 11.584442039658784,
      "learning_rate": 1.7854829293531005e-07,
      "loss": 0.7001,
      "step": 2388
    },
    {
      "epoch": 0.23616637421842177,
      "grad_norm": 11.027745221703094,
      "learning_rate": 1.7852847156790913e-07,
      "loss": 0.6828,
      "step": 2389
    },
    {
      "epoch": 0.23626522996317623,
      "grad_norm": 4.64465382324378,
      "learning_rate": 1.7850864214857183e-07,
      "loss": 0.6506,
      "step": 2390
    },
    {
      "epoch": 0.23636408570793072,
      "grad_norm": 3.1749441463210393,
      "learning_rate": 1.7848880467933136e-07,
      "loss": 0.722,
      "step": 2391
    },
    {
      "epoch": 0.23646294145268518,
      "grad_norm": 61.646353164000786,
      "learning_rate": 1.7846895916222176e-07,
      "loss": 0.767,
      "step": 2392
    },
    {
      "epoch": 0.23656179719743964,
      "grad_norm": 11.018486115030829,
      "learning_rate": 1.7844910559927787e-07,
      "loss": 0.7252,
      "step": 2393
    },
    {
      "epoch": 0.2366606529421941,
      "grad_norm": 4.284640556283989,
      "learning_rate": 1.7842924399253544e-07,
      "loss": 0.6773,
      "step": 2394
    },
    {
      "epoch": 0.23675950868694856,
      "grad_norm": 7.13147733163538,
      "learning_rate": 1.7840937434403095e-07,
      "loss": 0.721,
      "step": 2395
    },
    {
      "epoch": 0.23685836443170305,
      "grad_norm": 5.642915635528565,
      "learning_rate": 1.7838949665580176e-07,
      "loss": 0.6723,
      "step": 2396
    },
    {
      "epoch": 0.2369572201764575,
      "grad_norm": 4.1914069601372725,
      "learning_rate": 1.78369610929886e-07,
      "loss": 0.6639,
      "step": 2397
    },
    {
      "epoch": 0.23705607592121197,
      "grad_norm": 9.1931913207487,
      "learning_rate": 1.7834971716832264e-07,
      "loss": 0.5895,
      "step": 2398
    },
    {
      "epoch": 0.23715493166596643,
      "grad_norm": 9.85802401041135,
      "learning_rate": 1.7832981537315155e-07,
      "loss": 0.7286,
      "step": 2399
    },
    {
      "epoch": 0.2372537874107209,
      "grad_norm": 4.0438282666943595,
      "learning_rate": 1.7830990554641332e-07,
      "loss": 0.7264,
      "step": 2400
    },
    {
      "epoch": 0.23735264315547538,
      "grad_norm": 4.586859428760961,
      "learning_rate": 1.7828998769014944e-07,
      "loss": 0.7917,
      "step": 2401
    },
    {
      "epoch": 0.23745149890022985,
      "grad_norm": 3.58116537807087,
      "learning_rate": 1.7827006180640218e-07,
      "loss": 0.698,
      "step": 2402
    },
    {
      "epoch": 0.2375503546449843,
      "grad_norm": 18.235971692408402,
      "learning_rate": 1.7825012789721468e-07,
      "loss": 0.7142,
      "step": 2403
    },
    {
      "epoch": 0.23764921038973877,
      "grad_norm": 6.129467137853977,
      "learning_rate": 1.7823018596463083e-07,
      "loss": 0.7401,
      "step": 2404
    },
    {
      "epoch": 0.23774806613449323,
      "grad_norm": 6.113230941695234,
      "learning_rate": 1.7821023601069534e-07,
      "loss": 0.6702,
      "step": 2405
    },
    {
      "epoch": 0.23784692187924772,
      "grad_norm": 3.593330461486022,
      "learning_rate": 1.7819027803745382e-07,
      "loss": 0.6313,
      "step": 2406
    },
    {
      "epoch": 0.23794577762400218,
      "grad_norm": 4.848387348189897,
      "learning_rate": 1.7817031204695274e-07,
      "loss": 0.7,
      "step": 2407
    },
    {
      "epoch": 0.23804463336875664,
      "grad_norm": 4.165376422912614,
      "learning_rate": 1.781503380412392e-07,
      "loss": 0.7617,
      "step": 2408
    },
    {
      "epoch": 0.2381434891135111,
      "grad_norm": 3.207873659842536,
      "learning_rate": 1.7813035602236134e-07,
      "loss": 0.6821,
      "step": 2409
    },
    {
      "epoch": 0.23824234485826556,
      "grad_norm": 9.297991517936739,
      "learning_rate": 1.7811036599236792e-07,
      "loss": 0.7983,
      "step": 2410
    },
    {
      "epoch": 0.23834120060302005,
      "grad_norm": 9.434388126478582,
      "learning_rate": 1.780903679533087e-07,
      "loss": 0.6749,
      "step": 2411
    },
    {
      "epoch": 0.23844005634777451,
      "grad_norm": 4.578553844885686,
      "learning_rate": 1.7807036190723416e-07,
      "loss": 0.7396,
      "step": 2412
    },
    {
      "epoch": 0.23853891209252898,
      "grad_norm": 4.770548581920049,
      "learning_rate": 1.7805034785619567e-07,
      "loss": 0.7183,
      "step": 2413
    },
    {
      "epoch": 0.23863776783728344,
      "grad_norm": 5.508829134442424,
      "learning_rate": 1.7803032580224526e-07,
      "loss": 0.6187,
      "step": 2414
    },
    {
      "epoch": 0.2387366235820379,
      "grad_norm": 5.452310724083021,
      "learning_rate": 1.78010295747436e-07,
      "loss": 0.6974,
      "step": 2415
    },
    {
      "epoch": 0.2388354793267924,
      "grad_norm": 4.433225706780586,
      "learning_rate": 1.7799025769382163e-07,
      "loss": 0.7491,
      "step": 2416
    },
    {
      "epoch": 0.23893433507154685,
      "grad_norm": 4.895244287104808,
      "learning_rate": 1.779702116434568e-07,
      "loss": 0.8103,
      "step": 2417
    },
    {
      "epoch": 0.2390331908163013,
      "grad_norm": 3.317456281218132,
      "learning_rate": 1.7795015759839687e-07,
      "loss": 0.7614,
      "step": 2418
    },
    {
      "epoch": 0.23913204656105577,
      "grad_norm": 5.929703900740601,
      "learning_rate": 1.7793009556069815e-07,
      "loss": 0.7896,
      "step": 2419
    },
    {
      "epoch": 0.23923090230581023,
      "grad_norm": 3.748422486996989,
      "learning_rate": 1.7791002553241764e-07,
      "loss": 0.8026,
      "step": 2420
    },
    {
      "epoch": 0.23932975805056472,
      "grad_norm": 5.378306172576823,
      "learning_rate": 1.7788994751561331e-07,
      "loss": 0.8009,
      "step": 2421
    },
    {
      "epoch": 0.23942861379531918,
      "grad_norm": 3.143932157198784,
      "learning_rate": 1.7786986151234383e-07,
      "loss": 0.6877,
      "step": 2422
    },
    {
      "epoch": 0.23952746954007365,
      "grad_norm": 4.719441353736738,
      "learning_rate": 1.778497675246687e-07,
      "loss": 0.7498,
      "step": 2423
    },
    {
      "epoch": 0.2396263252848281,
      "grad_norm": 4.461989297266671,
      "learning_rate": 1.7782966555464824e-07,
      "loss": 0.7741,
      "step": 2424
    },
    {
      "epoch": 0.23972518102958257,
      "grad_norm": 2.7239360153355197,
      "learning_rate": 1.778095556043437e-07,
      "loss": 0.7333,
      "step": 2425
    },
    {
      "epoch": 0.23982403677433706,
      "grad_norm": 8.343779775782885,
      "learning_rate": 1.7778943767581696e-07,
      "loss": 0.7103,
      "step": 2426
    },
    {
      "epoch": 0.23992289251909152,
      "grad_norm": 3.494465602430989,
      "learning_rate": 1.7776931177113088e-07,
      "loss": 0.7738,
      "step": 2427
    },
    {
      "epoch": 0.24002174826384598,
      "grad_norm": 4.910352088982755,
      "learning_rate": 1.7774917789234904e-07,
      "loss": 0.7845,
      "step": 2428
    },
    {
      "epoch": 0.24012060400860044,
      "grad_norm": 6.657978956847763,
      "learning_rate": 1.7772903604153592e-07,
      "loss": 0.6877,
      "step": 2429
    },
    {
      "epoch": 0.24021945975335493,
      "grad_norm": 9.401235725437902,
      "learning_rate": 1.777088862207567e-07,
      "loss": 0.7322,
      "step": 2430
    },
    {
      "epoch": 0.2403183154981094,
      "grad_norm": 6.20288001572692,
      "learning_rate": 1.776887284320775e-07,
      "loss": 0.7323,
      "step": 2431
    },
    {
      "epoch": 0.24041717124286385,
      "grad_norm": 8.846944423796538,
      "learning_rate": 1.776685626775652e-07,
      "loss": 0.6497,
      "step": 2432
    },
    {
      "epoch": 0.24051602698761831,
      "grad_norm": 8.760498628095192,
      "learning_rate": 1.7764838895928748e-07,
      "loss": 0.6188,
      "step": 2433
    },
    {
      "epoch": 0.24061488273237278,
      "grad_norm": 5.550861172900912,
      "learning_rate": 1.776282072793129e-07,
      "loss": 0.7082,
      "step": 2434
    },
    {
      "epoch": 0.24071373847712726,
      "grad_norm": 12.843868488854449,
      "learning_rate": 1.7760801763971072e-07,
      "loss": 0.6888,
      "step": 2435
    },
    {
      "epoch": 0.24081259422188173,
      "grad_norm": 6.212730071033395,
      "learning_rate": 1.7758782004255112e-07,
      "loss": 0.8031,
      "step": 2436
    },
    {
      "epoch": 0.2409114499666362,
      "grad_norm": 3.68952862099971,
      "learning_rate": 1.775676144899051e-07,
      "loss": 0.6585,
      "step": 2437
    },
    {
      "epoch": 0.24101030571139065,
      "grad_norm": 4.278970502269807,
      "learning_rate": 1.7754740098384443e-07,
      "loss": 0.7585,
      "step": 2438
    },
    {
      "epoch": 0.2411091614561451,
      "grad_norm": 3.6180778406406846,
      "learning_rate": 1.7752717952644166e-07,
      "loss": 0.7455,
      "step": 2439
    },
    {
      "epoch": 0.2412080172008996,
      "grad_norm": 5.74139252973786,
      "learning_rate": 1.7750695011977025e-07,
      "loss": 0.657,
      "step": 2440
    },
    {
      "epoch": 0.24130687294565406,
      "grad_norm": 4.296857508256599,
      "learning_rate": 1.7748671276590442e-07,
      "loss": 0.6487,
      "step": 2441
    },
    {
      "epoch": 0.24140572869040852,
      "grad_norm": 3.3868148381179894,
      "learning_rate": 1.774664674669192e-07,
      "loss": 0.802,
      "step": 2442
    },
    {
      "epoch": 0.24150458443516298,
      "grad_norm": 3.414264655151745,
      "learning_rate": 1.7744621422489045e-07,
      "loss": 0.7315,
      "step": 2443
    },
    {
      "epoch": 0.24160344017991744,
      "grad_norm": 4.118589148267608,
      "learning_rate": 1.7742595304189484e-07,
      "loss": 0.6053,
      "step": 2444
    },
    {
      "epoch": 0.24170229592467193,
      "grad_norm": 3.138085187201764,
      "learning_rate": 1.7740568392000987e-07,
      "loss": 0.6863,
      "step": 2445
    },
    {
      "epoch": 0.2418011516694264,
      "grad_norm": 4.705837508606869,
      "learning_rate": 1.7738540686131383e-07,
      "loss": 0.6122,
      "step": 2446
    },
    {
      "epoch": 0.24190000741418086,
      "grad_norm": 3.897090698443874,
      "learning_rate": 1.773651218678858e-07,
      "loss": 0.6448,
      "step": 2447
    },
    {
      "epoch": 0.24199886315893532,
      "grad_norm": 3.878911662684798,
      "learning_rate": 1.7734482894180579e-07,
      "loss": 0.6736,
      "step": 2448
    },
    {
      "epoch": 0.24209771890368978,
      "grad_norm": 3.995836069155526,
      "learning_rate": 1.7732452808515444e-07,
      "loss": 0.7111,
      "step": 2449
    },
    {
      "epoch": 0.24219657464844427,
      "grad_norm": 6.906842004693163,
      "learning_rate": 1.773042193000134e-07,
      "loss": 0.7113,
      "step": 2450
    },
    {
      "epoch": 0.24229543039319873,
      "grad_norm": 7.173488825434043,
      "learning_rate": 1.7728390258846492e-07,
      "loss": 0.6916,
      "step": 2451
    },
    {
      "epoch": 0.2423942861379532,
      "grad_norm": 8.342601355042884,
      "learning_rate": 1.772635779525923e-07,
      "loss": 0.7219,
      "step": 2452
    },
    {
      "epoch": 0.24249314188270765,
      "grad_norm": 3.724555406583542,
      "learning_rate": 1.7724324539447948e-07,
      "loss": 0.8007,
      "step": 2453
    },
    {
      "epoch": 0.2425919976274621,
      "grad_norm": 3.7456644729109363,
      "learning_rate": 1.7722290491621123e-07,
      "loss": 0.6959,
      "step": 2454
    },
    {
      "epoch": 0.2426908533722166,
      "grad_norm": 5.078708952891566,
      "learning_rate": 1.7720255651987323e-07,
      "loss": 0.6996,
      "step": 2455
    },
    {
      "epoch": 0.24278970911697106,
      "grad_norm": 4.744083328166485,
      "learning_rate": 1.7718220020755186e-07,
      "loss": 0.6636,
      "step": 2456
    },
    {
      "epoch": 0.24288856486172553,
      "grad_norm": 4.978765425272639,
      "learning_rate": 1.7716183598133438e-07,
      "loss": 0.7324,
      "step": 2457
    },
    {
      "epoch": 0.24298742060648,
      "grad_norm": 2.944275704610409,
      "learning_rate": 1.7714146384330886e-07,
      "loss": 0.6906,
      "step": 2458
    },
    {
      "epoch": 0.24308627635123445,
      "grad_norm": 3.8385708481468694,
      "learning_rate": 1.771210837955641e-07,
      "loss": 0.6235,
      "step": 2459
    },
    {
      "epoch": 0.24318513209598894,
      "grad_norm": 4.758588526650373,
      "learning_rate": 1.7710069584018982e-07,
      "loss": 0.7568,
      "step": 2460
    },
    {
      "epoch": 0.2432839878407434,
      "grad_norm": 3.4259530743689446,
      "learning_rate": 1.7708029997927652e-07,
      "loss": 0.8023,
      "step": 2461
    },
    {
      "epoch": 0.24338284358549786,
      "grad_norm": 4.061480294838327,
      "learning_rate": 1.7705989621491546e-07,
      "loss": 0.7084,
      "step": 2462
    },
    {
      "epoch": 0.24348169933025232,
      "grad_norm": 8.376766382646403,
      "learning_rate": 1.7703948454919876e-07,
      "loss": 0.7411,
      "step": 2463
    },
    {
      "epoch": 0.24358055507500678,
      "grad_norm": 3.65852939562364,
      "learning_rate": 1.7701906498421933e-07,
      "loss": 0.6834,
      "step": 2464
    },
    {
      "epoch": 0.24367941081976127,
      "grad_norm": 6.140092843402329,
      "learning_rate": 1.769986375220709e-07,
      "loss": 0.6963,
      "step": 2465
    },
    {
      "epoch": 0.24377826656451573,
      "grad_norm": 4.230968968577057,
      "learning_rate": 1.7697820216484797e-07,
      "loss": 0.6845,
      "step": 2466
    },
    {
      "epoch": 0.2438771223092702,
      "grad_norm": 5.737229580318362,
      "learning_rate": 1.7695775891464596e-07,
      "loss": 0.7439,
      "step": 2467
    },
    {
      "epoch": 0.24397597805402466,
      "grad_norm": 3.982041588088474,
      "learning_rate": 1.7693730777356096e-07,
      "loss": 0.8045,
      "step": 2468
    },
    {
      "epoch": 0.24407483379877914,
      "grad_norm": 4.329869812743968,
      "learning_rate": 1.7691684874368995e-07,
      "loss": 0.7624,
      "step": 2469
    },
    {
      "epoch": 0.2441736895435336,
      "grad_norm": 9.156272479026446,
      "learning_rate": 1.7689638182713072e-07,
      "loss": 0.6536,
      "step": 2470
    },
    {
      "epoch": 0.24427254528828807,
      "grad_norm": 4.581996162099595,
      "learning_rate": 1.7687590702598182e-07,
      "loss": 0.7165,
      "step": 2471
    },
    {
      "epoch": 0.24437140103304253,
      "grad_norm": 3.4468341883897367,
      "learning_rate": 1.7685542434234267e-07,
      "loss": 0.6788,
      "step": 2472
    },
    {
      "epoch": 0.244470256777797,
      "grad_norm": 3.612944149748399,
      "learning_rate": 1.7683493377831343e-07,
      "loss": 0.7927,
      "step": 2473
    },
    {
      "epoch": 0.24456911252255148,
      "grad_norm": 3.2185652520044927,
      "learning_rate": 1.7681443533599513e-07,
      "loss": 0.7547,
      "step": 2474
    },
    {
      "epoch": 0.24466796826730594,
      "grad_norm": 3.602920615066137,
      "learning_rate": 1.7679392901748958e-07,
      "loss": 0.8225,
      "step": 2475
    },
    {
      "epoch": 0.2447668240120604,
      "grad_norm": 4.282765241008579,
      "learning_rate": 1.7677341482489943e-07,
      "loss": 0.6793,
      "step": 2476
    },
    {
      "epoch": 0.24486567975681486,
      "grad_norm": 4.398788648639304,
      "learning_rate": 1.7675289276032803e-07,
      "loss": 0.7702,
      "step": 2477
    },
    {
      "epoch": 0.24496453550156932,
      "grad_norm": 7.839821873873161,
      "learning_rate": 1.7673236282587972e-07,
      "loss": 0.7816,
      "step": 2478
    },
    {
      "epoch": 0.24506339124632381,
      "grad_norm": 10.692733943743454,
      "learning_rate": 1.7671182502365943e-07,
      "loss": 0.6693,
      "step": 2479
    },
    {
      "epoch": 0.24516224699107828,
      "grad_norm": 3.4116748290941445,
      "learning_rate": 1.766912793557731e-07,
      "loss": 0.7076,
      "step": 2480
    },
    {
      "epoch": 0.24526110273583274,
      "grad_norm": 19.053769537014283,
      "learning_rate": 1.7667072582432732e-07,
      "loss": 0.7422,
      "step": 2481
    },
    {
      "epoch": 0.2453599584805872,
      "grad_norm": 3.4981970208898945,
      "learning_rate": 1.7665016443142958e-07,
      "loss": 0.6751,
      "step": 2482
    },
    {
      "epoch": 0.24545881422534166,
      "grad_norm": 3.1412536625789573,
      "learning_rate": 1.7662959517918815e-07,
      "loss": 0.6647,
      "step": 2483
    },
    {
      "epoch": 0.24555766997009615,
      "grad_norm": 3.2647934372260257,
      "learning_rate": 1.766090180697121e-07,
      "loss": 0.8186,
      "step": 2484
    },
    {
      "epoch": 0.2456565257148506,
      "grad_norm": 6.946925242548534,
      "learning_rate": 1.7658843310511134e-07,
      "loss": 0.7262,
      "step": 2485
    },
    {
      "epoch": 0.24575538145960507,
      "grad_norm": 8.889455891869993,
      "learning_rate": 1.7656784028749648e-07,
      "loss": 0.6679,
      "step": 2486
    },
    {
      "epoch": 0.24585423720435953,
      "grad_norm": 11.410057643375843,
      "learning_rate": 1.7654723961897909e-07,
      "loss": 0.715,
      "step": 2487
    },
    {
      "epoch": 0.245953092949114,
      "grad_norm": 3.320585726937451,
      "learning_rate": 1.7652663110167142e-07,
      "loss": 0.6114,
      "step": 2488
    },
    {
      "epoch": 0.24605194869386848,
      "grad_norm": 5.625502555218986,
      "learning_rate": 1.7650601473768653e-07,
      "loss": 0.6662,
      "step": 2489
    },
    {
      "epoch": 0.24615080443862294,
      "grad_norm": 4.336177650344499,
      "learning_rate": 1.7648539052913843e-07,
      "loss": 0.738,
      "step": 2490
    },
    {
      "epoch": 0.2462496601833774,
      "grad_norm": 3.671404396096694,
      "learning_rate": 1.7646475847814174e-07,
      "loss": 0.777,
      "step": 2491
    },
    {
      "epoch": 0.24634851592813187,
      "grad_norm": 4.7663528149013255,
      "learning_rate": 1.7644411858681203e-07,
      "loss": 0.7318,
      "step": 2492
    },
    {
      "epoch": 0.24644737167288633,
      "grad_norm": 3.5098593616931124,
      "learning_rate": 1.7642347085726556e-07,
      "loss": 0.7514,
      "step": 2493
    },
    {
      "epoch": 0.24654622741764082,
      "grad_norm": 7.9922739981760955,
      "learning_rate": 1.7640281529161948e-07,
      "loss": 0.6086,
      "step": 2494
    },
    {
      "epoch": 0.24664508316239528,
      "grad_norm": 5.38458925551755,
      "learning_rate": 1.7638215189199175e-07,
      "loss": 0.5735,
      "step": 2495
    },
    {
      "epoch": 0.24674393890714974,
      "grad_norm": 4.183484775840916,
      "learning_rate": 1.7636148066050104e-07,
      "loss": 0.68,
      "step": 2496
    },
    {
      "epoch": 0.2468427946519042,
      "grad_norm": 8.62483509107993,
      "learning_rate": 1.7634080159926689e-07,
      "loss": 0.7281,
      "step": 2497
    },
    {
      "epoch": 0.24694165039665866,
      "grad_norm": 5.1621187931689985,
      "learning_rate": 1.7632011471040965e-07,
      "loss": 0.7217,
      "step": 2498
    },
    {
      "epoch": 0.24704050614141315,
      "grad_norm": 9.356829092901002,
      "learning_rate": 1.7629941999605044e-07,
      "loss": 0.7168,
      "step": 2499
    },
    {
      "epoch": 0.2471393618861676,
      "grad_norm": 4.5730549708997295,
      "learning_rate": 1.7627871745831123e-07,
      "loss": 0.7306,
      "step": 2500
    },
    {
      "epoch": 0.24723821763092207,
      "grad_norm": 4.2724531875057545,
      "learning_rate": 1.7625800709931468e-07,
      "loss": 0.6544,
      "step": 2501
    },
    {
      "epoch": 0.24733707337567654,
      "grad_norm": 3.794592541830703,
      "learning_rate": 1.7623728892118444e-07,
      "loss": 0.696,
      "step": 2502
    },
    {
      "epoch": 0.247435929120431,
      "grad_norm": 3.498996806633802,
      "learning_rate": 1.7621656292604476e-07,
      "loss": 0.7138,
      "step": 2503
    },
    {
      "epoch": 0.2475347848651855,
      "grad_norm": 10.443949698391775,
      "learning_rate": 1.7619582911602087e-07,
      "loss": 0.7766,
      "step": 2504
    },
    {
      "epoch": 0.24763364060993995,
      "grad_norm": 3.4102472791754033,
      "learning_rate": 1.7617508749323864e-07,
      "loss": 0.6973,
      "step": 2505
    },
    {
      "epoch": 0.2477324963546944,
      "grad_norm": 3.2018984068622625,
      "learning_rate": 1.7615433805982484e-07,
      "loss": 0.6959,
      "step": 2506
    },
    {
      "epoch": 0.24783135209944887,
      "grad_norm": 4.892942300996585,
      "learning_rate": 1.7613358081790705e-07,
      "loss": 0.7664,
      "step": 2507
    },
    {
      "epoch": 0.24793020784420336,
      "grad_norm": 9.93184581581607,
      "learning_rate": 1.7611281576961358e-07,
      "loss": 0.7035,
      "step": 2508
    },
    {
      "epoch": 0.24802906358895782,
      "grad_norm": 4.7808847259358584,
      "learning_rate": 1.7609204291707363e-07,
      "loss": 0.6357,
      "step": 2509
    },
    {
      "epoch": 0.24812791933371228,
      "grad_norm": 8.089359093380752,
      "learning_rate": 1.760712622624171e-07,
      "loss": 0.724,
      "step": 2510
    },
    {
      "epoch": 0.24822677507846674,
      "grad_norm": 4.168674566110212,
      "learning_rate": 1.7605047380777473e-07,
      "loss": 0.7836,
      "step": 2511
    },
    {
      "epoch": 0.2483256308232212,
      "grad_norm": 9.360843840944087,
      "learning_rate": 1.760296775552781e-07,
      "loss": 0.6153,
      "step": 2512
    },
    {
      "epoch": 0.2484244865679757,
      "grad_norm": 5.454715932020585,
      "learning_rate": 1.7600887350705959e-07,
      "loss": 0.7085,
      "step": 2513
    },
    {
      "epoch": 0.24852334231273016,
      "grad_norm": 3.748058744017857,
      "learning_rate": 1.7598806166525228e-07,
      "loss": 0.6329,
      "step": 2514
    },
    {
      "epoch": 0.24862219805748462,
      "grad_norm": 3.636136424326512,
      "learning_rate": 1.7596724203199017e-07,
      "loss": 0.6148,
      "step": 2515
    },
    {
      "epoch": 0.24872105380223908,
      "grad_norm": 13.715925881131213,
      "learning_rate": 1.7594641460940794e-07,
      "loss": 0.7218,
      "step": 2516
    },
    {
      "epoch": 0.24881990954699354,
      "grad_norm": 8.988378303470881,
      "learning_rate": 1.759255793996412e-07,
      "loss": 0.7762,
      "step": 2517
    },
    {
      "epoch": 0.24891876529174803,
      "grad_norm": 4.13650552473767,
      "learning_rate": 1.7590473640482633e-07,
      "loss": 0.6962,
      "step": 2518
    },
    {
      "epoch": 0.2490176210365025,
      "grad_norm": 3.8377917263110746,
      "learning_rate": 1.7588388562710036e-07,
      "loss": 0.7977,
      "step": 2519
    },
    {
      "epoch": 0.24911647678125695,
      "grad_norm": 3.793851245889237,
      "learning_rate": 1.758630270686013e-07,
      "loss": 0.6946,
      "step": 2520
    },
    {
      "epoch": 0.2492153325260114,
      "grad_norm": 3.6573226902860996,
      "learning_rate": 1.7584216073146784e-07,
      "loss": 0.7414,
      "step": 2521
    },
    {
      "epoch": 0.24931418827076587,
      "grad_norm": 9.162345359558287,
      "learning_rate": 1.758212866178396e-07,
      "loss": 0.8288,
      "step": 2522
    },
    {
      "epoch": 0.24941304401552036,
      "grad_norm": 6.657945839598378,
      "learning_rate": 1.7580040472985683e-07,
      "loss": 0.7862,
      "step": 2523
    },
    {
      "epoch": 0.24951189976027482,
      "grad_norm": 4.942357749035868,
      "learning_rate": 1.757795150696607e-07,
      "loss": 0.5859,
      "step": 2524
    },
    {
      "epoch": 0.24961075550502929,
      "grad_norm": 5.394180287800419,
      "learning_rate": 1.757586176393931e-07,
      "loss": 0.7195,
      "step": 2525
    },
    {
      "epoch": 0.24970961124978375,
      "grad_norm": 43.9773010585701,
      "learning_rate": 1.7573771244119683e-07,
      "loss": 0.7369,
      "step": 2526
    },
    {
      "epoch": 0.2498084669945382,
      "grad_norm": 4.2205964718940825,
      "learning_rate": 1.757167994772153e-07,
      "loss": 0.7298,
      "step": 2527
    },
    {
      "epoch": 0.2499073227392927,
      "grad_norm": 5.70351166541157,
      "learning_rate": 1.7569587874959292e-07,
      "loss": 0.7665,
      "step": 2528
    },
    {
      "epoch": 0.25000617848404716,
      "grad_norm": 3.23737141804287,
      "learning_rate": 1.7567495026047477e-07,
      "loss": 0.7298,
      "step": 2529
    },
    {
      "epoch": 0.25010503422880165,
      "grad_norm": 3.3806513715060884,
      "learning_rate": 1.7565401401200674e-07,
      "loss": 0.704,
      "step": 2530
    },
    {
      "epoch": 0.2502038899735561,
      "grad_norm": 3.3885555044601947,
      "learning_rate": 1.7563307000633555e-07,
      "loss": 0.7181,
      "step": 2531
    },
    {
      "epoch": 0.25030274571831057,
      "grad_norm": 3.792444764732508,
      "learning_rate": 1.756121182456087e-07,
      "loss": 0.6499,
      "step": 2532
    },
    {
      "epoch": 0.250401601463065,
      "grad_norm": 4.856788120683121,
      "learning_rate": 1.755911587319745e-07,
      "loss": 0.725,
      "step": 2533
    },
    {
      "epoch": 0.2505004572078195,
      "grad_norm": 8.579585706017015,
      "learning_rate": 1.75570191467582e-07,
      "loss": 0.714,
      "step": 2534
    },
    {
      "epoch": 0.250599312952574,
      "grad_norm": 5.535268583082957,
      "learning_rate": 1.7554921645458108e-07,
      "loss": 0.6537,
      "step": 2535
    },
    {
      "epoch": 0.2506981686973284,
      "grad_norm": 3.5267650147920886,
      "learning_rate": 1.755282336951225e-07,
      "loss": 0.8029,
      "step": 2536
    },
    {
      "epoch": 0.2507970244420829,
      "grad_norm": 5.424893068152502,
      "learning_rate": 1.755072431913576e-07,
      "loss": 0.64,
      "step": 2537
    },
    {
      "epoch": 0.25089588018683734,
      "grad_norm": 11.56611810511707,
      "learning_rate": 1.754862449454388e-07,
      "loss": 0.7468,
      "step": 2538
    },
    {
      "epoch": 0.25099473593159183,
      "grad_norm": 25.88718748593189,
      "learning_rate": 1.7546523895951903e-07,
      "loss": 0.6794,
      "step": 2539
    },
    {
      "epoch": 0.2510935916763463,
      "grad_norm": 3.9695773040182134,
      "learning_rate": 1.754442252357522e-07,
      "loss": 0.8229,
      "step": 2540
    },
    {
      "epoch": 0.25119244742110075,
      "grad_norm": 3.5143341011768867,
      "learning_rate": 1.7542320377629295e-07,
      "loss": 0.6254,
      "step": 2541
    },
    {
      "epoch": 0.25129130316585524,
      "grad_norm": 3.8495303661309634,
      "learning_rate": 1.7540217458329677e-07,
      "loss": 0.6262,
      "step": 2542
    },
    {
      "epoch": 0.2513901589106097,
      "grad_norm": 3.5593099644802355,
      "learning_rate": 1.753811376589198e-07,
      "loss": 0.6988,
      "step": 2543
    },
    {
      "epoch": 0.25148901465536416,
      "grad_norm": 4.158512862002003,
      "learning_rate": 1.7536009300531913e-07,
      "loss": 0.7698,
      "step": 2544
    },
    {
      "epoch": 0.25158787040011865,
      "grad_norm": 14.136582755812698,
      "learning_rate": 1.7533904062465257e-07,
      "loss": 0.734,
      "step": 2545
    },
    {
      "epoch": 0.2516867261448731,
      "grad_norm": 5.514154460949664,
      "learning_rate": 1.7531798051907872e-07,
      "loss": 0.7633,
      "step": 2546
    },
    {
      "epoch": 0.2517855818896276,
      "grad_norm": 5.115326242639329,
      "learning_rate": 1.7529691269075694e-07,
      "loss": 0.7784,
      "step": 2547
    },
    {
      "epoch": 0.251884437634382,
      "grad_norm": 6.145716270484772,
      "learning_rate": 1.7527583714184753e-07,
      "loss": 0.7014,
      "step": 2548
    },
    {
      "epoch": 0.2519832933791365,
      "grad_norm": 3.047543441126006,
      "learning_rate": 1.752547538745114e-07,
      "loss": 0.7166,
      "step": 2549
    },
    {
      "epoch": 0.252082149123891,
      "grad_norm": 5.586113701172232,
      "learning_rate": 1.7523366289091035e-07,
      "loss": 0.6718,
      "step": 2550
    },
    {
      "epoch": 0.2521810048686454,
      "grad_norm": 4.885472839868687,
      "learning_rate": 1.7521256419320692e-07,
      "loss": 0.6802,
      "step": 2551
    },
    {
      "epoch": 0.2522798606133999,
      "grad_norm": 7.826482265874667,
      "learning_rate": 1.7519145778356452e-07,
      "loss": 0.7708,
      "step": 2552
    },
    {
      "epoch": 0.25237871635815434,
      "grad_norm": 5.958178735698945,
      "learning_rate": 1.7517034366414726e-07,
      "loss": 0.7668,
      "step": 2553
    },
    {
      "epoch": 0.25247757210290883,
      "grad_norm": 4.086636375152883,
      "learning_rate": 1.751492218371201e-07,
      "loss": 0.7085,
      "step": 2554
    },
    {
      "epoch": 0.2525764278476633,
      "grad_norm": 22.792613738618396,
      "learning_rate": 1.751280923046488e-07,
      "loss": 0.827,
      "step": 2555
    },
    {
      "epoch": 0.25267528359241775,
      "grad_norm": 3.6925920407315105,
      "learning_rate": 1.751069550688998e-07,
      "loss": 0.5968,
      "step": 2556
    },
    {
      "epoch": 0.25277413933717224,
      "grad_norm": 6.553517637441141,
      "learning_rate": 1.750858101320405e-07,
      "loss": 0.8419,
      "step": 2557
    },
    {
      "epoch": 0.2528729950819267,
      "grad_norm": 4.432554748077511,
      "learning_rate": 1.7506465749623898e-07,
      "loss": 0.709,
      "step": 2558
    },
    {
      "epoch": 0.25297185082668117,
      "grad_norm": 5.3852153749394605,
      "learning_rate": 1.7504349716366407e-07,
      "loss": 0.7895,
      "step": 2559
    },
    {
      "epoch": 0.25307070657143566,
      "grad_norm": 14.443490714717976,
      "learning_rate": 1.7502232913648554e-07,
      "loss": 0.7514,
      "step": 2560
    },
    {
      "epoch": 0.2531695623161901,
      "grad_norm": 5.555020684815613,
      "learning_rate": 1.7500115341687377e-07,
      "loss": 0.7208,
      "step": 2561
    },
    {
      "epoch": 0.2532684180609446,
      "grad_norm": 7.225553851688573,
      "learning_rate": 1.7497997000700008e-07,
      "loss": 0.7861,
      "step": 2562
    },
    {
      "epoch": 0.253367273805699,
      "grad_norm": 4.8137115689304775,
      "learning_rate": 1.749587789090365e-07,
      "loss": 0.787,
      "step": 2563
    },
    {
      "epoch": 0.2534661295504535,
      "grad_norm": 4.82780296394149,
      "learning_rate": 1.7493758012515588e-07,
      "loss": 0.6047,
      "step": 2564
    },
    {
      "epoch": 0.253564985295208,
      "grad_norm": 4.258589274278392,
      "learning_rate": 1.7491637365753178e-07,
      "loss": 0.6346,
      "step": 2565
    },
    {
      "epoch": 0.2536638410399624,
      "grad_norm": 3.5468784879298054,
      "learning_rate": 1.748951595083387e-07,
      "loss": 0.6608,
      "step": 2566
    },
    {
      "epoch": 0.2537626967847169,
      "grad_norm": 5.477811882757876,
      "learning_rate": 1.748739376797518e-07,
      "loss": 0.8272,
      "step": 2567
    },
    {
      "epoch": 0.25386155252947135,
      "grad_norm": 3.501439849147315,
      "learning_rate": 1.7485270817394704e-07,
      "loss": 0.8043,
      "step": 2568
    },
    {
      "epoch": 0.25396040827422584,
      "grad_norm": 6.967249219877242,
      "learning_rate": 1.7483147099310118e-07,
      "loss": 0.7524,
      "step": 2569
    },
    {
      "epoch": 0.2540592640189803,
      "grad_norm": 3.3410928137378457,
      "learning_rate": 1.7481022613939186e-07,
      "loss": 0.5104,
      "step": 2570
    },
    {
      "epoch": 0.25415811976373476,
      "grad_norm": 4.11969303504438,
      "learning_rate": 1.7478897361499735e-07,
      "loss": 0.6687,
      "step": 2571
    },
    {
      "epoch": 0.25425697550848925,
      "grad_norm": 4.878248876566138,
      "learning_rate": 1.7476771342209683e-07,
      "loss": 0.6857,
      "step": 2572
    },
    {
      "epoch": 0.2543558312532437,
      "grad_norm": 5.395236963535395,
      "learning_rate": 1.7474644556287017e-07,
      "loss": 0.7044,
      "step": 2573
    },
    {
      "epoch": 0.25445468699799817,
      "grad_norm": 13.660917410263362,
      "learning_rate": 1.7472517003949813e-07,
      "loss": 0.7021,
      "step": 2574
    },
    {
      "epoch": 0.25455354274275266,
      "grad_norm": 3.809978207545114,
      "learning_rate": 1.7470388685416217e-07,
      "loss": 0.6392,
      "step": 2575
    },
    {
      "epoch": 0.2546523984875071,
      "grad_norm": 3.1759207254770936,
      "learning_rate": 1.7468259600904458e-07,
      "loss": 0.6506,
      "step": 2576
    },
    {
      "epoch": 0.2547512542322616,
      "grad_norm": 4.394458490649577,
      "learning_rate": 1.7466129750632842e-07,
      "loss": 0.71,
      "step": 2577
    },
    {
      "epoch": 0.254850109977016,
      "grad_norm": 5.326604839835025,
      "learning_rate": 1.746399913481975e-07,
      "loss": 0.7149,
      "step": 2578
    },
    {
      "epoch": 0.2549489657217705,
      "grad_norm": 3.7356530750302666,
      "learning_rate": 1.7461867753683651e-07,
      "loss": 0.8131,
      "step": 2579
    },
    {
      "epoch": 0.255047821466525,
      "grad_norm": 5.103292922619978,
      "learning_rate": 1.7459735607443086e-07,
      "loss": 0.6932,
      "step": 2580
    },
    {
      "epoch": 0.2551466772112794,
      "grad_norm": 4.593642833846963,
      "learning_rate": 1.745760269631667e-07,
      "loss": 0.621,
      "step": 2581
    },
    {
      "epoch": 0.2552455329560339,
      "grad_norm": 3.9308917513932067,
      "learning_rate": 1.7455469020523106e-07,
      "loss": 0.7009,
      "step": 2582
    },
    {
      "epoch": 0.25534438870078835,
      "grad_norm": 4.853226896317163,
      "learning_rate": 1.7453334580281166e-07,
      "loss": 0.7247,
      "step": 2583
    },
    {
      "epoch": 0.25544324444554284,
      "grad_norm": 4.841370552271901,
      "learning_rate": 1.7451199375809718e-07,
      "loss": 0.7723,
      "step": 2584
    },
    {
      "epoch": 0.2555421001902973,
      "grad_norm": 4.448589158346291,
      "learning_rate": 1.744906340732768e-07,
      "loss": 0.7668,
      "step": 2585
    },
    {
      "epoch": 0.25564095593505176,
      "grad_norm": 3.624675363661294,
      "learning_rate": 1.7446926675054074e-07,
      "loss": 0.7008,
      "step": 2586
    },
    {
      "epoch": 0.25573981167980625,
      "grad_norm": 3.5013149745444596,
      "learning_rate": 1.7444789179207983e-07,
      "loss": 0.6981,
      "step": 2587
    },
    {
      "epoch": 0.2558386674245607,
      "grad_norm": 10.351704041559495,
      "learning_rate": 1.7442650920008585e-07,
      "loss": 0.7115,
      "step": 2588
    },
    {
      "epoch": 0.2559375231693152,
      "grad_norm": 8.483091007225129,
      "learning_rate": 1.744051189767512e-07,
      "loss": 0.634,
      "step": 2589
    },
    {
      "epoch": 0.25603637891406966,
      "grad_norm": 3.5885496473003955,
      "learning_rate": 1.7438372112426919e-07,
      "loss": 0.5632,
      "step": 2590
    },
    {
      "epoch": 0.2561352346588241,
      "grad_norm": 3.497578399064761,
      "learning_rate": 1.7436231564483376e-07,
      "loss": 0.7095,
      "step": 2591
    },
    {
      "epoch": 0.2562340904035786,
      "grad_norm": 3.688603429035982,
      "learning_rate": 1.743409025406398e-07,
      "loss": 0.7295,
      "step": 2592
    },
    {
      "epoch": 0.256332946148333,
      "grad_norm": 3.768977578229503,
      "learning_rate": 1.743194818138829e-07,
      "loss": 0.7285,
      "step": 2593
    },
    {
      "epoch": 0.2564318018930875,
      "grad_norm": 4.712294050693658,
      "learning_rate": 1.742980534667594e-07,
      "loss": 0.7723,
      "step": 2594
    },
    {
      "epoch": 0.256530657637842,
      "grad_norm": 4.797464312145532,
      "learning_rate": 1.7427661750146654e-07,
      "loss": 0.6319,
      "step": 2595
    },
    {
      "epoch": 0.25662951338259643,
      "grad_norm": 3.6908362441407268,
      "learning_rate": 1.7425517392020215e-07,
      "loss": 0.7309,
      "step": 2596
    },
    {
      "epoch": 0.2567283691273509,
      "grad_norm": 3.8501315625658,
      "learning_rate": 1.7423372272516507e-07,
      "loss": 0.736,
      "step": 2597
    },
    {
      "epoch": 0.25682722487210535,
      "grad_norm": 8.697286336874848,
      "learning_rate": 1.7421226391855471e-07,
      "loss": 0.7217,
      "step": 2598
    },
    {
      "epoch": 0.25692608061685984,
      "grad_norm": 2.8978683470043536,
      "learning_rate": 1.7419079750257141e-07,
      "loss": 0.6761,
      "step": 2599
    },
    {
      "epoch": 0.25702493636161433,
      "grad_norm": 4.322022387896223,
      "learning_rate": 1.7416932347941624e-07,
      "loss": 0.6336,
      "step": 2600
    },
    {
      "epoch": 0.25712379210636876,
      "grad_norm": 3.6066343237292067,
      "learning_rate": 1.74147841851291e-07,
      "loss": 0.8463,
      "step": 2601
    },
    {
      "epoch": 0.25722264785112325,
      "grad_norm": 4.135812261543081,
      "learning_rate": 1.7412635262039834e-07,
      "loss": 0.7642,
      "step": 2602
    },
    {
      "epoch": 0.25732150359587774,
      "grad_norm": 3.1280831529841957,
      "learning_rate": 1.7410485578894164e-07,
      "loss": 0.7085,
      "step": 2603
    },
    {
      "epoch": 0.2574203593406322,
      "grad_norm": 6.8858715409845965,
      "learning_rate": 1.7408335135912512e-07,
      "loss": 0.7356,
      "step": 2604
    },
    {
      "epoch": 0.25751921508538667,
      "grad_norm": 3.7029362919694084,
      "learning_rate": 1.7406183933315371e-07,
      "loss": 0.6994,
      "step": 2605
    },
    {
      "epoch": 0.2576180708301411,
      "grad_norm": 6.182826578957922,
      "learning_rate": 1.740403197132332e-07,
      "loss": 0.6721,
      "step": 2606
    },
    {
      "epoch": 0.2577169265748956,
      "grad_norm": 8.894863279889158,
      "learning_rate": 1.7401879250157006e-07,
      "loss": 0.7417,
      "step": 2607
    },
    {
      "epoch": 0.2578157823196501,
      "grad_norm": 3.62953261263379,
      "learning_rate": 1.739972577003716e-07,
      "loss": 0.7378,
      "step": 2608
    },
    {
      "epoch": 0.2579146380644045,
      "grad_norm": 4.807859443073986,
      "learning_rate": 1.7397571531184587e-07,
      "loss": 0.73,
      "step": 2609
    },
    {
      "epoch": 0.258013493809159,
      "grad_norm": 12.985849348791023,
      "learning_rate": 1.739541653382018e-07,
      "loss": 0.6791,
      "step": 2610
    },
    {
      "epoch": 0.25811234955391343,
      "grad_norm": 2.9348041243124516,
      "learning_rate": 1.7393260778164894e-07,
      "loss": 0.655,
      "step": 2611
    },
    {
      "epoch": 0.2582112052986679,
      "grad_norm": 3.8368424121236084,
      "learning_rate": 1.7391104264439775e-07,
      "loss": 0.746,
      "step": 2612
    },
    {
      "epoch": 0.2583100610434224,
      "grad_norm": 4.97911318533631,
      "learning_rate": 1.738894699286594e-07,
      "loss": 0.6371,
      "step": 2613
    },
    {
      "epoch": 0.25840891678817685,
      "grad_norm": 8.896766517343393,
      "learning_rate": 1.7386788963664587e-07,
      "loss": 0.7509,
      "step": 2614
    },
    {
      "epoch": 0.25850777253293133,
      "grad_norm": 5.48965729056475,
      "learning_rate": 1.738463017705699e-07,
      "loss": 0.735,
      "step": 2615
    },
    {
      "epoch": 0.25860662827768577,
      "grad_norm": 4.970798072265461,
      "learning_rate": 1.7382470633264493e-07,
      "loss": 0.7343,
      "step": 2616
    },
    {
      "epoch": 0.25870548402244026,
      "grad_norm": 5.136390429075031,
      "learning_rate": 1.7380310332508533e-07,
      "loss": 0.7656,
      "step": 2617
    },
    {
      "epoch": 0.25880433976719475,
      "grad_norm": 5.822043653378741,
      "learning_rate": 1.737814927501062e-07,
      "loss": 0.6614,
      "step": 2618
    },
    {
      "epoch": 0.2589031955119492,
      "grad_norm": 26.764213329308824,
      "learning_rate": 1.7375987460992331e-07,
      "loss": 0.7416,
      "step": 2619
    },
    {
      "epoch": 0.25900205125670367,
      "grad_norm": 4.994614069200326,
      "learning_rate": 1.7373824890675334e-07,
      "loss": 0.7057,
      "step": 2620
    },
    {
      "epoch": 0.2591009070014581,
      "grad_norm": 6.737518394462079,
      "learning_rate": 1.7371661564281365e-07,
      "loss": 0.8052,
      "step": 2621
    },
    {
      "epoch": 0.2591997627462126,
      "grad_norm": 2.6138975694348408,
      "learning_rate": 1.736949748203224e-07,
      "loss": 0.6464,
      "step": 2622
    },
    {
      "epoch": 0.2592986184909671,
      "grad_norm": 3.550778200999612,
      "learning_rate": 1.736733264414986e-07,
      "loss": 0.7328,
      "step": 2623
    },
    {
      "epoch": 0.2593974742357215,
      "grad_norm": 17.70772412089997,
      "learning_rate": 1.736516705085619e-07,
      "loss": 0.6884,
      "step": 2624
    },
    {
      "epoch": 0.259496329980476,
      "grad_norm": 4.434915052820346,
      "learning_rate": 1.7363000702373283e-07,
      "loss": 0.7044,
      "step": 2625
    },
    {
      "epoch": 0.25959518572523044,
      "grad_norm": 15.345038158908627,
      "learning_rate": 1.7360833598923265e-07,
      "loss": 0.7497,
      "step": 2626
    },
    {
      "epoch": 0.2596940414699849,
      "grad_norm": 14.807401051540316,
      "learning_rate": 1.7358665740728344e-07,
      "loss": 0.6828,
      "step": 2627
    },
    {
      "epoch": 0.2597928972147394,
      "grad_norm": 4.620533412149072,
      "learning_rate": 1.7356497128010798e-07,
      "loss": 0.7361,
      "step": 2628
    },
    {
      "epoch": 0.25989175295949385,
      "grad_norm": 5.021193712104873,
      "learning_rate": 1.7354327760992988e-07,
      "loss": 0.6963,
      "step": 2629
    },
    {
      "epoch": 0.25999060870424834,
      "grad_norm": 4.7564961143060245,
      "learning_rate": 1.735215763989735e-07,
      "loss": 0.7376,
      "step": 2630
    },
    {
      "epoch": 0.26008946444900277,
      "grad_norm": 4.112645269630967,
      "learning_rate": 1.7349986764946395e-07,
      "loss": 0.7682,
      "step": 2631
    },
    {
      "epoch": 0.26018832019375726,
      "grad_norm": 3.7346349649161863,
      "learning_rate": 1.734781513636272e-07,
      "loss": 0.6742,
      "step": 2632
    },
    {
      "epoch": 0.26028717593851175,
      "grad_norm": 4.606703462831743,
      "learning_rate": 1.734564275436899e-07,
      "loss": 0.713,
      "step": 2633
    },
    {
      "epoch": 0.2603860316832662,
      "grad_norm": 4.7472223618483005,
      "learning_rate": 1.7343469619187954e-07,
      "loss": 0.6362,
      "step": 2634
    },
    {
      "epoch": 0.2604848874280207,
      "grad_norm": 4.018166403523397,
      "learning_rate": 1.7341295731042428e-07,
      "loss": 0.7657,
      "step": 2635
    },
    {
      "epoch": 0.2605837431727751,
      "grad_norm": 4.759837265957269,
      "learning_rate": 1.733912109015532e-07,
      "loss": 0.6945,
      "step": 2636
    },
    {
      "epoch": 0.2606825989175296,
      "grad_norm": 3.4492680476832036,
      "learning_rate": 1.7336945696749602e-07,
      "loss": 0.6996,
      "step": 2637
    },
    {
      "epoch": 0.2607814546622841,
      "grad_norm": 3.5237982642190198,
      "learning_rate": 1.7334769551048328e-07,
      "loss": 0.7095,
      "step": 2638
    },
    {
      "epoch": 0.2608803104070385,
      "grad_norm": 5.601059903563019,
      "learning_rate": 1.7332592653274637e-07,
      "loss": 0.7558,
      "step": 2639
    },
    {
      "epoch": 0.260979166151793,
      "grad_norm": 3.6636456101104904,
      "learning_rate": 1.7330415003651733e-07,
      "loss": 0.735,
      "step": 2640
    },
    {
      "epoch": 0.26107802189654744,
      "grad_norm": 12.803390046774275,
      "learning_rate": 1.7328236602402898e-07,
      "loss": 0.6236,
      "step": 2641
    },
    {
      "epoch": 0.26117687764130193,
      "grad_norm": 3.66787765414936,
      "learning_rate": 1.73260574497515e-07,
      "loss": 0.7887,
      "step": 2642
    },
    {
      "epoch": 0.2612757333860564,
      "grad_norm": 3.970944152972017,
      "learning_rate": 1.7323877545920977e-07,
      "loss": 0.8116,
      "step": 2643
    },
    {
      "epoch": 0.26137458913081085,
      "grad_norm": 14.763060745929776,
      "learning_rate": 1.7321696891134847e-07,
      "loss": 0.7117,
      "step": 2644
    },
    {
      "epoch": 0.26147344487556534,
      "grad_norm": 9.749440150341105,
      "learning_rate": 1.7319515485616702e-07,
      "loss": 0.6928,
      "step": 2645
    },
    {
      "epoch": 0.2615723006203198,
      "grad_norm": 4.180740344376574,
      "learning_rate": 1.7317333329590217e-07,
      "loss": 0.726,
      "step": 2646
    },
    {
      "epoch": 0.26167115636507426,
      "grad_norm": 3.7262429250804003,
      "learning_rate": 1.7315150423279135e-07,
      "loss": 0.8468,
      "step": 2647
    },
    {
      "epoch": 0.26177001210982875,
      "grad_norm": 4.88127956969713,
      "learning_rate": 1.7312966766907283e-07,
      "loss": 0.858,
      "step": 2648
    },
    {
      "epoch": 0.2618688678545832,
      "grad_norm": 7.888114437666803,
      "learning_rate": 1.7310782360698565e-07,
      "loss": 0.7806,
      "step": 2649
    },
    {
      "epoch": 0.2619677235993377,
      "grad_norm": 10.189201481307917,
      "learning_rate": 1.7308597204876959e-07,
      "loss": 0.7432,
      "step": 2650
    },
    {
      "epoch": 0.2620665793440921,
      "grad_norm": 5.402701603631917,
      "learning_rate": 1.7306411299666516e-07,
      "loss": 0.7283,
      "step": 2651
    },
    {
      "epoch": 0.2621654350888466,
      "grad_norm": 3.4802971113431536,
      "learning_rate": 1.730422464529137e-07,
      "loss": 0.71,
      "step": 2652
    },
    {
      "epoch": 0.2622642908336011,
      "grad_norm": 2.8613607328180253,
      "learning_rate": 1.7302037241975734e-07,
      "loss": 0.7194,
      "step": 2653
    },
    {
      "epoch": 0.2623631465783555,
      "grad_norm": 4.208893593328062,
      "learning_rate": 1.729984908994389e-07,
      "loss": 0.7518,
      "step": 2654
    },
    {
      "epoch": 0.26246200232311,
      "grad_norm": 4.4884253532834855,
      "learning_rate": 1.7297660189420205e-07,
      "loss": 0.74,
      "step": 2655
    },
    {
      "epoch": 0.26256085806786444,
      "grad_norm": 16.546122331381905,
      "learning_rate": 1.7295470540629112e-07,
      "loss": 0.7648,
      "step": 2656
    },
    {
      "epoch": 0.26265971381261893,
      "grad_norm": 3.341909227031189,
      "learning_rate": 1.7293280143795131e-07,
      "loss": 0.6008,
      "step": 2657
    },
    {
      "epoch": 0.2627585695573734,
      "grad_norm": 3.1300278122615843,
      "learning_rate": 1.7291088999142859e-07,
      "loss": 0.8031,
      "step": 2658
    },
    {
      "epoch": 0.26285742530212786,
      "grad_norm": 3.709792149743311,
      "learning_rate": 1.7288897106896956e-07,
      "loss": 0.7415,
      "step": 2659
    },
    {
      "epoch": 0.26295628104688235,
      "grad_norm": 3.2591827875433976,
      "learning_rate": 1.7286704467282176e-07,
      "loss": 0.7641,
      "step": 2660
    },
    {
      "epoch": 0.2630551367916368,
      "grad_norm": 3.7786186986847654,
      "learning_rate": 1.728451108052334e-07,
      "loss": 0.7612,
      "step": 2661
    },
    {
      "epoch": 0.26315399253639127,
      "grad_norm": 3.6271066329749453,
      "learning_rate": 1.7282316946845347e-07,
      "loss": 0.819,
      "step": 2662
    },
    {
      "epoch": 0.26325284828114576,
      "grad_norm": 3.414955300272365,
      "learning_rate": 1.728012206647317e-07,
      "loss": 0.7077,
      "step": 2663
    },
    {
      "epoch": 0.2633517040259002,
      "grad_norm": 319.9984134019195,
      "learning_rate": 1.7277926439631866e-07,
      "loss": 0.6744,
      "step": 2664
    },
    {
      "epoch": 0.2634505597706547,
      "grad_norm": 5.03937928907127,
      "learning_rate": 1.7275730066546563e-07,
      "loss": 0.6655,
      "step": 2665
    },
    {
      "epoch": 0.2635494155154091,
      "grad_norm": 4.248578237555021,
      "learning_rate": 1.7273532947442466e-07,
      "loss": 0.6928,
      "step": 2666
    },
    {
      "epoch": 0.2636482712601636,
      "grad_norm": 5.068226798374296,
      "learning_rate": 1.7271335082544858e-07,
      "loss": 0.824,
      "step": 2667
    },
    {
      "epoch": 0.2637471270049181,
      "grad_norm": 3.96359665467839,
      "learning_rate": 1.72691364720791e-07,
      "loss": 0.6795,
      "step": 2668
    },
    {
      "epoch": 0.2638459827496725,
      "grad_norm": 3.324405264743262,
      "learning_rate": 1.7266937116270618e-07,
      "loss": 0.7028,
      "step": 2669
    },
    {
      "epoch": 0.263944838494427,
      "grad_norm": 4.795453918704021,
      "learning_rate": 1.7264737015344933e-07,
      "loss": 0.7729,
      "step": 2670
    },
    {
      "epoch": 0.26404369423918145,
      "grad_norm": 3.372399072594103,
      "learning_rate": 1.7262536169527629e-07,
      "loss": 0.8507,
      "step": 2671
    },
    {
      "epoch": 0.26414254998393594,
      "grad_norm": 5.518100165847274,
      "learning_rate": 1.7260334579044372e-07,
      "loss": 0.8385,
      "step": 2672
    },
    {
      "epoch": 0.2642414057286904,
      "grad_norm": 5.157689787915596,
      "learning_rate": 1.7258132244120899e-07,
      "loss": 0.7875,
      "step": 2673
    },
    {
      "epoch": 0.26434026147344486,
      "grad_norm": 3.7256774351640862,
      "learning_rate": 1.7255929164983028e-07,
      "loss": 0.7103,
      "step": 2674
    },
    {
      "epoch": 0.26443911721819935,
      "grad_norm": 3.6069213424878686,
      "learning_rate": 1.7253725341856654e-07,
      "loss": 0.7952,
      "step": 2675
    },
    {
      "epoch": 0.2645379729629538,
      "grad_norm": 2.7879353602583112,
      "learning_rate": 1.725152077496775e-07,
      "loss": 0.7253,
      "step": 2676
    },
    {
      "epoch": 0.26463682870770827,
      "grad_norm": 5.9389715641056755,
      "learning_rate": 1.7249315464542358e-07,
      "loss": 0.7899,
      "step": 2677
    },
    {
      "epoch": 0.26473568445246276,
      "grad_norm": 3.5535183317217145,
      "learning_rate": 1.7247109410806598e-07,
      "loss": 0.6564,
      "step": 2678
    },
    {
      "epoch": 0.2648345401972172,
      "grad_norm": 4.00066731254155,
      "learning_rate": 1.7244902613986669e-07,
      "loss": 0.6602,
      "step": 2679
    },
    {
      "epoch": 0.2649333959419717,
      "grad_norm": 4.532286930600595,
      "learning_rate": 1.724269507430885e-07,
      "loss": 0.8024,
      "step": 2680
    },
    {
      "epoch": 0.2650322516867262,
      "grad_norm": 14.237854914680526,
      "learning_rate": 1.7240486791999484e-07,
      "loss": 0.6932,
      "step": 2681
    },
    {
      "epoch": 0.2651311074314806,
      "grad_norm": 8.737687547522533,
      "learning_rate": 1.7238277767285005e-07,
      "loss": 0.7213,
      "step": 2682
    },
    {
      "epoch": 0.2652299631762351,
      "grad_norm": 5.268082798280748,
      "learning_rate": 1.7236068000391914e-07,
      "loss": 0.7229,
      "step": 2683
    },
    {
      "epoch": 0.26532881892098953,
      "grad_norm": 3.5678429355501318,
      "learning_rate": 1.7233857491546787e-07,
      "loss": 0.7374,
      "step": 2684
    },
    {
      "epoch": 0.265427674665744,
      "grad_norm": 7.414483735275459,
      "learning_rate": 1.7231646240976283e-07,
      "loss": 0.6713,
      "step": 2685
    },
    {
      "epoch": 0.2655265304104985,
      "grad_norm": 7.787999242025072,
      "learning_rate": 1.722943424890713e-07,
      "loss": 0.7745,
      "step": 2686
    },
    {
      "epoch": 0.26562538615525294,
      "grad_norm": 4.523955830252212,
      "learning_rate": 1.7227221515566137e-07,
      "loss": 0.6894,
      "step": 2687
    },
    {
      "epoch": 0.26572424190000743,
      "grad_norm": 7.0225782833062365,
      "learning_rate": 1.7225008041180185e-07,
      "loss": 0.7893,
      "step": 2688
    },
    {
      "epoch": 0.26582309764476186,
      "grad_norm": 6.60583685843071,
      "learning_rate": 1.7222793825976236e-07,
      "loss": 0.7677,
      "step": 2689
    },
    {
      "epoch": 0.26592195338951635,
      "grad_norm": 5.930441967518548,
      "learning_rate": 1.7220578870181325e-07,
      "loss": 0.7012,
      "step": 2690
    },
    {
      "epoch": 0.26602080913427084,
      "grad_norm": 6.587046904087079,
      "learning_rate": 1.721836317402256e-07,
      "loss": 0.6668,
      "step": 2691
    },
    {
      "epoch": 0.2661196648790253,
      "grad_norm": 5.96238644071406,
      "learning_rate": 1.721614673772713e-07,
      "loss": 0.6208,
      "step": 2692
    },
    {
      "epoch": 0.26621852062377976,
      "grad_norm": 3.8429411676577603,
      "learning_rate": 1.72139295615223e-07,
      "loss": 0.7585,
      "step": 2693
    },
    {
      "epoch": 0.2663173763685342,
      "grad_norm": 3.126805205616945,
      "learning_rate": 1.7211711645635404e-07,
      "loss": 0.6213,
      "step": 2694
    },
    {
      "epoch": 0.2664162321132887,
      "grad_norm": 4.2934639786529685,
      "learning_rate": 1.7209492990293862e-07,
      "loss": 0.6095,
      "step": 2695
    },
    {
      "epoch": 0.2665150878580432,
      "grad_norm": 4.520292397005222,
      "learning_rate": 1.7207273595725157e-07,
      "loss": 0.7099,
      "step": 2696
    },
    {
      "epoch": 0.2666139436027976,
      "grad_norm": 3.8172081669623465,
      "learning_rate": 1.720505346215686e-07,
      "loss": 0.6163,
      "step": 2697
    },
    {
      "epoch": 0.2667127993475521,
      "grad_norm": 3.88547183495742,
      "learning_rate": 1.7202832589816615e-07,
      "loss": 0.7033,
      "step": 2698
    },
    {
      "epoch": 0.26681165509230653,
      "grad_norm": 4.973127253909407,
      "learning_rate": 1.7200610978932137e-07,
      "loss": 0.8837,
      "step": 2699
    },
    {
      "epoch": 0.266910510837061,
      "grad_norm": 13.515075718147092,
      "learning_rate": 1.719838862973122e-07,
      "loss": 0.779,
      "step": 2700
    },
    {
      "epoch": 0.2670093665818155,
      "grad_norm": 3.3211256723621294,
      "learning_rate": 1.719616554244173e-07,
      "loss": 0.7111,
      "step": 2701
    },
    {
      "epoch": 0.26710822232656994,
      "grad_norm": 4.38572314928293,
      "learning_rate": 1.7193941717291618e-07,
      "loss": 0.7166,
      "step": 2702
    },
    {
      "epoch": 0.26720707807132443,
      "grad_norm": 5.186600974771468,
      "learning_rate": 1.7191717154508897e-07,
      "loss": 0.8215,
      "step": 2703
    },
    {
      "epoch": 0.26730593381607887,
      "grad_norm": 3.72848570156967,
      "learning_rate": 1.7189491854321672e-07,
      "loss": 0.7429,
      "step": 2704
    },
    {
      "epoch": 0.26740478956083336,
      "grad_norm": 3.6077111490470184,
      "learning_rate": 1.7187265816958105e-07,
      "loss": 0.5842,
      "step": 2705
    },
    {
      "epoch": 0.26750364530558784,
      "grad_norm": 5.057305919298859,
      "learning_rate": 1.718503904264645e-07,
      "loss": 0.7476,
      "step": 2706
    },
    {
      "epoch": 0.2676025010503423,
      "grad_norm": 3.4203649368439204,
      "learning_rate": 1.718281153161503e-07,
      "loss": 0.6563,
      "step": 2707
    },
    {
      "epoch": 0.26770135679509677,
      "grad_norm": 3.7845360973386364,
      "learning_rate": 1.718058328409224e-07,
      "loss": 0.7985,
      "step": 2708
    },
    {
      "epoch": 0.2678002125398512,
      "grad_norm": 2.7978239614305207,
      "learning_rate": 1.7178354300306555e-07,
      "loss": 0.742,
      "step": 2709
    },
    {
      "epoch": 0.2678990682846057,
      "grad_norm": 3.2886283354133634,
      "learning_rate": 1.7176124580486523e-07,
      "loss": 0.7314,
      "step": 2710
    },
    {
      "epoch": 0.2679979240293602,
      "grad_norm": 5.0870432459155515,
      "learning_rate": 1.7173894124860771e-07,
      "loss": 0.7277,
      "step": 2711
    },
    {
      "epoch": 0.2680967797741146,
      "grad_norm": 5.1474967420976085,
      "learning_rate": 1.7171662933658e-07,
      "loss": 0.8155,
      "step": 2712
    },
    {
      "epoch": 0.2681956355188691,
      "grad_norm": 6.411846710363167,
      "learning_rate": 1.7169431007106984e-07,
      "loss": 0.7462,
      "step": 2713
    },
    {
      "epoch": 0.26829449126362354,
      "grad_norm": 4.373310845647385,
      "learning_rate": 1.7167198345436575e-07,
      "loss": 0.7847,
      "step": 2714
    },
    {
      "epoch": 0.268393347008378,
      "grad_norm": 4.406336033052503,
      "learning_rate": 1.71649649488757e-07,
      "loss": 0.6869,
      "step": 2715
    },
    {
      "epoch": 0.2684922027531325,
      "grad_norm": 15.459789719604203,
      "learning_rate": 1.716273081765336e-07,
      "loss": 0.6823,
      "step": 2716
    },
    {
      "epoch": 0.26859105849788695,
      "grad_norm": 5.114592238506369,
      "learning_rate": 1.716049595199863e-07,
      "loss": 0.6557,
      "step": 2717
    },
    {
      "epoch": 0.26868991424264144,
      "grad_norm": 4.276011660410829,
      "learning_rate": 1.715826035214067e-07,
      "loss": 0.792,
      "step": 2718
    },
    {
      "epoch": 0.26878876998739587,
      "grad_norm": 6.317679090142235,
      "learning_rate": 1.71560240183087e-07,
      "loss": 0.7087,
      "step": 2719
    },
    {
      "epoch": 0.26888762573215036,
      "grad_norm": 11.149777039212214,
      "learning_rate": 1.7153786950732023e-07,
      "loss": 0.7259,
      "step": 2720
    },
    {
      "epoch": 0.26898648147690485,
      "grad_norm": 5.640768744686316,
      "learning_rate": 1.7151549149640024e-07,
      "loss": 0.7417,
      "step": 2721
    },
    {
      "epoch": 0.2690853372216593,
      "grad_norm": 3.804775695626346,
      "learning_rate": 1.714931061526215e-07,
      "loss": 0.6923,
      "step": 2722
    },
    {
      "epoch": 0.26918419296641377,
      "grad_norm": 4.174490303535939,
      "learning_rate": 1.7147071347827933e-07,
      "loss": 0.738,
      "step": 2723
    },
    {
      "epoch": 0.2692830487111682,
      "grad_norm": 3.8418018878860845,
      "learning_rate": 1.7144831347566976e-07,
      "loss": 0.7336,
      "step": 2724
    },
    {
      "epoch": 0.2693819044559227,
      "grad_norm": 5.695029858028046,
      "learning_rate": 1.7142590614708957e-07,
      "loss": 0.7512,
      "step": 2725
    },
    {
      "epoch": 0.2694807602006772,
      "grad_norm": 3.231158971327436,
      "learning_rate": 1.714034914948363e-07,
      "loss": 0.7184,
      "step": 2726
    },
    {
      "epoch": 0.2695796159454316,
      "grad_norm": 7.91830619774103,
      "learning_rate": 1.713810695212083e-07,
      "loss": 0.7815,
      "step": 2727
    },
    {
      "epoch": 0.2696784716901861,
      "grad_norm": 4.0589960880831875,
      "learning_rate": 1.713586402285045e-07,
      "loss": 0.7903,
      "step": 2728
    },
    {
      "epoch": 0.26977732743494054,
      "grad_norm": 3.9100993896595972,
      "learning_rate": 1.7133620361902482e-07,
      "loss": 0.7514,
      "step": 2729
    },
    {
      "epoch": 0.26987618317969503,
      "grad_norm": 8.026697077426341,
      "learning_rate": 1.7131375969506972e-07,
      "loss": 0.6745,
      "step": 2730
    },
    {
      "epoch": 0.2699750389244495,
      "grad_norm": 3.4843343589437725,
      "learning_rate": 1.7129130845894053e-07,
      "loss": 0.7383,
      "step": 2731
    },
    {
      "epoch": 0.27007389466920395,
      "grad_norm": 6.677193416839262,
      "learning_rate": 1.7126884991293924e-07,
      "loss": 0.6941,
      "step": 2732
    },
    {
      "epoch": 0.27017275041395844,
      "grad_norm": 6.579620412921381,
      "learning_rate": 1.712463840593687e-07,
      "loss": 0.7634,
      "step": 2733
    },
    {
      "epoch": 0.2702716061587129,
      "grad_norm": 3.372900121240285,
      "learning_rate": 1.7122391090053246e-07,
      "loss": 0.6974,
      "step": 2734
    },
    {
      "epoch": 0.27037046190346736,
      "grad_norm": 3.849039439915618,
      "learning_rate": 1.7120143043873477e-07,
      "loss": 0.6682,
      "step": 2735
    },
    {
      "epoch": 0.27046931764822185,
      "grad_norm": 14.236431164391774,
      "learning_rate": 1.7117894267628066e-07,
      "loss": 0.7373,
      "step": 2736
    },
    {
      "epoch": 0.2705681733929763,
      "grad_norm": 4.983920512019094,
      "learning_rate": 1.7115644761547596e-07,
      "loss": 0.7421,
      "step": 2737
    },
    {
      "epoch": 0.2706670291377308,
      "grad_norm": 5.766959578690897,
      "learning_rate": 1.7113394525862718e-07,
      "loss": 0.7103,
      "step": 2738
    },
    {
      "epoch": 0.2707658848824852,
      "grad_norm": 3.5123996283938177,
      "learning_rate": 1.711114356080416e-07,
      "loss": 0.6569,
      "step": 2739
    },
    {
      "epoch": 0.2708647406272397,
      "grad_norm": 3.641046838805623,
      "learning_rate": 1.7108891866602727e-07,
      "loss": 0.6919,
      "step": 2740
    },
    {
      "epoch": 0.2709635963719942,
      "grad_norm": 7.379567825751909,
      "learning_rate": 1.71066394434893e-07,
      "loss": 0.7081,
      "step": 2741
    },
    {
      "epoch": 0.2710624521167486,
      "grad_norm": 5.721743710596797,
      "learning_rate": 1.7104386291694827e-07,
      "loss": 0.6656,
      "step": 2742
    },
    {
      "epoch": 0.2711613078615031,
      "grad_norm": 11.809436160498016,
      "learning_rate": 1.7102132411450336e-07,
      "loss": 0.6572,
      "step": 2743
    },
    {
      "epoch": 0.27126016360625754,
      "grad_norm": 3.6298938671609124,
      "learning_rate": 1.7099877802986932e-07,
      "loss": 0.5259,
      "step": 2744
    },
    {
      "epoch": 0.27135901935101203,
      "grad_norm": 3.4756043981456175,
      "learning_rate": 1.7097622466535785e-07,
      "loss": 0.6858,
      "step": 2745
    },
    {
      "epoch": 0.2714578750957665,
      "grad_norm": 4.268714582796012,
      "learning_rate": 1.7095366402328156e-07,
      "loss": 0.7996,
      "step": 2746
    },
    {
      "epoch": 0.27155673084052095,
      "grad_norm": 5.110685425511531,
      "learning_rate": 1.7093109610595365e-07,
      "loss": 0.7577,
      "step": 2747
    },
    {
      "epoch": 0.27165558658527544,
      "grad_norm": 3.180075781983478,
      "learning_rate": 1.7090852091568814e-07,
      "loss": 0.7166,
      "step": 2748
    },
    {
      "epoch": 0.2717544423300299,
      "grad_norm": 4.555187870547729,
      "learning_rate": 1.7088593845479978e-07,
      "loss": 0.7601,
      "step": 2749
    },
    {
      "epoch": 0.27185329807478437,
      "grad_norm": 3.8379359357874585,
      "learning_rate": 1.7086334872560406e-07,
      "loss": 0.6451,
      "step": 2750
    },
    {
      "epoch": 0.27195215381953886,
      "grad_norm": 5.588082387607627,
      "learning_rate": 1.7084075173041727e-07,
      "loss": 0.7403,
      "step": 2751
    },
    {
      "epoch": 0.2720510095642933,
      "grad_norm": 11.662015857809356,
      "learning_rate": 1.7081814747155634e-07,
      "loss": 0.6834,
      "step": 2752
    },
    {
      "epoch": 0.2721498653090478,
      "grad_norm": 3.0972144542479514,
      "learning_rate": 1.7079553595133904e-07,
      "loss": 0.7007,
      "step": 2753
    },
    {
      "epoch": 0.2722487210538022,
      "grad_norm": 6.484561990930114,
      "learning_rate": 1.7077291717208383e-07,
      "loss": 0.6974,
      "step": 2754
    },
    {
      "epoch": 0.2723475767985567,
      "grad_norm": 3.069365584711649,
      "learning_rate": 1.7075029113610992e-07,
      "loss": 0.7119,
      "step": 2755
    },
    {
      "epoch": 0.2724464325433112,
      "grad_norm": 6.326860751064532,
      "learning_rate": 1.7072765784573732e-07,
      "loss": 0.7133,
      "step": 2756
    },
    {
      "epoch": 0.2725452882880656,
      "grad_norm": 19.095871113892567,
      "learning_rate": 1.707050173032867e-07,
      "loss": 0.6668,
      "step": 2757
    },
    {
      "epoch": 0.2726441440328201,
      "grad_norm": 3.265472457129888,
      "learning_rate": 1.7068236951107954e-07,
      "loss": 0.7157,
      "step": 2758
    },
    {
      "epoch": 0.2727429997775746,
      "grad_norm": 4.878611144433243,
      "learning_rate": 1.7065971447143804e-07,
      "loss": 0.785,
      "step": 2759
    },
    {
      "epoch": 0.27284185552232904,
      "grad_norm": 3.6253477542031542,
      "learning_rate": 1.706370521866851e-07,
      "loss": 0.7549,
      "step": 2760
    },
    {
      "epoch": 0.2729407112670835,
      "grad_norm": 4.715531098562006,
      "learning_rate": 1.7061438265914445e-07,
      "loss": 0.6667,
      "step": 2761
    },
    {
      "epoch": 0.27303956701183796,
      "grad_norm": 8.055872437049178,
      "learning_rate": 1.7059170589114048e-07,
      "loss": 0.704,
      "step": 2762
    },
    {
      "epoch": 0.27313842275659245,
      "grad_norm": 7.276697355399356,
      "learning_rate": 1.7056902188499836e-07,
      "loss": 0.666,
      "step": 2763
    },
    {
      "epoch": 0.27323727850134694,
      "grad_norm": 4.139905557877823,
      "learning_rate": 1.70546330643044e-07,
      "loss": 0.6967,
      "step": 2764
    },
    {
      "epoch": 0.27333613424610137,
      "grad_norm": 3.1560459948303916,
      "learning_rate": 1.705236321676041e-07,
      "loss": 0.6864,
      "step": 2765
    },
    {
      "epoch": 0.27343498999085586,
      "grad_norm": 4.122840556752035,
      "learning_rate": 1.70500926461006e-07,
      "loss": 0.6452,
      "step": 2766
    },
    {
      "epoch": 0.2735338457356103,
      "grad_norm": 12.754842171734808,
      "learning_rate": 1.704782135255779e-07,
      "loss": 0.8194,
      "step": 2767
    },
    {
      "epoch": 0.2736327014803648,
      "grad_norm": 3.667547230244823,
      "learning_rate": 1.7045549336364854e-07,
      "loss": 0.7546,
      "step": 2768
    },
    {
      "epoch": 0.27373155722511927,
      "grad_norm": 4.768286744570924,
      "learning_rate": 1.704327659775477e-07,
      "loss": 0.6623,
      "step": 2769
    },
    {
      "epoch": 0.2738304129698737,
      "grad_norm": 3.407922886015219,
      "learning_rate": 1.7041003136960564e-07,
      "loss": 0.7402,
      "step": 2770
    },
    {
      "epoch": 0.2739292687146282,
      "grad_norm": 3.8010381477705804,
      "learning_rate": 1.7038728954215346e-07,
      "loss": 0.7771,
      "step": 2771
    },
    {
      "epoch": 0.2740281244593826,
      "grad_norm": 10.353262258327643,
      "learning_rate": 1.7036454049752304e-07,
      "loss": 0.7542,
      "step": 2772
    },
    {
      "epoch": 0.2741269802041371,
      "grad_norm": 3.4756256338326987,
      "learning_rate": 1.7034178423804698e-07,
      "loss": 0.7522,
      "step": 2773
    },
    {
      "epoch": 0.2742258359488916,
      "grad_norm": 8.270971235605833,
      "learning_rate": 1.7031902076605853e-07,
      "loss": 0.6332,
      "step": 2774
    },
    {
      "epoch": 0.27432469169364604,
      "grad_norm": 9.11170275799054,
      "learning_rate": 1.7029625008389177e-07,
      "loss": 0.7218,
      "step": 2775
    },
    {
      "epoch": 0.27442354743840053,
      "grad_norm": 4.026906649270887,
      "learning_rate": 1.702734721938815e-07,
      "loss": 0.5586,
      "step": 2776
    },
    {
      "epoch": 0.27452240318315496,
      "grad_norm": 4.070619481102734,
      "learning_rate": 1.7025068709836327e-07,
      "loss": 0.7469,
      "step": 2777
    },
    {
      "epoch": 0.27462125892790945,
      "grad_norm": 3.738105023272579,
      "learning_rate": 1.7022789479967337e-07,
      "loss": 0.6428,
      "step": 2778
    },
    {
      "epoch": 0.27472011467266394,
      "grad_norm": 3.870121556443452,
      "learning_rate": 1.7020509530014878e-07,
      "loss": 0.798,
      "step": 2779
    },
    {
      "epoch": 0.2748189704174184,
      "grad_norm": 3.6017344301138077,
      "learning_rate": 1.7018228860212726e-07,
      "loss": 0.5284,
      "step": 2780
    },
    {
      "epoch": 0.27491782616217286,
      "grad_norm": 5.704918520965602,
      "learning_rate": 1.701594747079473e-07,
      "loss": 0.7337,
      "step": 2781
    },
    {
      "epoch": 0.2750166819069273,
      "grad_norm": 9.983047148869666,
      "learning_rate": 1.7013665361994815e-07,
      "loss": 0.7453,
      "step": 2782
    },
    {
      "epoch": 0.2751155376516818,
      "grad_norm": 3.373031994405885,
      "learning_rate": 1.7011382534046977e-07,
      "loss": 0.8295,
      "step": 2783
    },
    {
      "epoch": 0.2752143933964363,
      "grad_norm": 4.48474411181987,
      "learning_rate": 1.700909898718528e-07,
      "loss": 0.7813,
      "step": 2784
    },
    {
      "epoch": 0.2753132491411907,
      "grad_norm": 4.138888399143323,
      "learning_rate": 1.700681472164388e-07,
      "loss": 0.6591,
      "step": 2785
    },
    {
      "epoch": 0.2754121048859452,
      "grad_norm": 7.719928914748751,
      "learning_rate": 1.7004529737656986e-07,
      "loss": 0.6932,
      "step": 2786
    },
    {
      "epoch": 0.27551096063069963,
      "grad_norm": 3.5853207966446896,
      "learning_rate": 1.700224403545889e-07,
      "loss": 0.6036,
      "step": 2787
    },
    {
      "epoch": 0.2756098163754541,
      "grad_norm": 4.70423189644711,
      "learning_rate": 1.6999957615283956e-07,
      "loss": 0.6995,
      "step": 2788
    },
    {
      "epoch": 0.2757086721202086,
      "grad_norm": 5.756835304811161,
      "learning_rate": 1.6997670477366631e-07,
      "loss": 0.7856,
      "step": 2789
    },
    {
      "epoch": 0.27580752786496304,
      "grad_norm": 3.1137238895000956,
      "learning_rate": 1.699538262194142e-07,
      "loss": 0.7606,
      "step": 2790
    },
    {
      "epoch": 0.27590638360971753,
      "grad_norm": 4.016064938163057,
      "learning_rate": 1.699309404924291e-07,
      "loss": 0.7817,
      "step": 2791
    },
    {
      "epoch": 0.27600523935447197,
      "grad_norm": 7.78352844587789,
      "learning_rate": 1.6990804759505756e-07,
      "loss": 0.7192,
      "step": 2792
    },
    {
      "epoch": 0.27610409509922645,
      "grad_norm": 5.059689749454026,
      "learning_rate": 1.69885147529647e-07,
      "loss": 0.8114,
      "step": 2793
    },
    {
      "epoch": 0.27620295084398094,
      "grad_norm": 3.9391395393723148,
      "learning_rate": 1.6986224029854543e-07,
      "loss": 0.7772,
      "step": 2794
    },
    {
      "epoch": 0.2763018065887354,
      "grad_norm": 3.4095566878137404,
      "learning_rate": 1.698393259041017e-07,
      "loss": 0.7723,
      "step": 2795
    },
    {
      "epoch": 0.27640066233348987,
      "grad_norm": 3.2045375296487695,
      "learning_rate": 1.6981640434866523e-07,
      "loss": 0.741,
      "step": 2796
    },
    {
      "epoch": 0.2764995180782443,
      "grad_norm": 3.5179637813241955,
      "learning_rate": 1.6979347563458636e-07,
      "loss": 0.6722,
      "step": 2797
    },
    {
      "epoch": 0.2765983738229988,
      "grad_norm": 5.311132005149992,
      "learning_rate": 1.6977053976421614e-07,
      "loss": 0.6788,
      "step": 2798
    },
    {
      "epoch": 0.2766972295677533,
      "grad_norm": 4.868701827805657,
      "learning_rate": 1.6974759673990624e-07,
      "loss": 0.6363,
      "step": 2799
    },
    {
      "epoch": 0.2767960853125077,
      "grad_norm": 9.703872098721078,
      "learning_rate": 1.6972464656400912e-07,
      "loss": 0.7397,
      "step": 2800
    },
    {
      "epoch": 0.2768949410572622,
      "grad_norm": 4.732192263472022,
      "learning_rate": 1.69701689238878e-07,
      "loss": 0.5856,
      "step": 2801
    },
    {
      "epoch": 0.27699379680201663,
      "grad_norm": 13.997146616948957,
      "learning_rate": 1.6967872476686686e-07,
      "loss": 0.6972,
      "step": 2802
    },
    {
      "epoch": 0.2770926525467711,
      "grad_norm": 12.40888459050881,
      "learning_rate": 1.6965575315033035e-07,
      "loss": 0.7141,
      "step": 2803
    },
    {
      "epoch": 0.2771915082915256,
      "grad_norm": 6.9588597584764935,
      "learning_rate": 1.696327743916238e-07,
      "loss": 0.8439,
      "step": 2804
    },
    {
      "epoch": 0.27729036403628005,
      "grad_norm": 4.189039861994579,
      "learning_rate": 1.696097884931034e-07,
      "loss": 0.7188,
      "step": 2805
    },
    {
      "epoch": 0.27738921978103454,
      "grad_norm": 3.051323762707604,
      "learning_rate": 1.69586795457126e-07,
      "loss": 0.7341,
      "step": 2806
    },
    {
      "epoch": 0.27748807552578897,
      "grad_norm": 3.4573748698540574,
      "learning_rate": 1.6956379528604924e-07,
      "loss": 0.6919,
      "step": 2807
    },
    {
      "epoch": 0.27758693127054346,
      "grad_norm": 3.228416792717548,
      "learning_rate": 1.6954078798223136e-07,
      "loss": 0.753,
      "step": 2808
    },
    {
      "epoch": 0.27768578701529795,
      "grad_norm": 4.282133470970379,
      "learning_rate": 1.695177735480315e-07,
      "loss": 0.7471,
      "step": 2809
    },
    {
      "epoch": 0.2777846427600524,
      "grad_norm": 4.6729284497532495,
      "learning_rate": 1.6949475198580942e-07,
      "loss": 0.7314,
      "step": 2810
    },
    {
      "epoch": 0.27788349850480687,
      "grad_norm": 3.167353788249365,
      "learning_rate": 1.6947172329792565e-07,
      "loss": 0.7288,
      "step": 2811
    },
    {
      "epoch": 0.2779823542495613,
      "grad_norm": 4.23408905054495,
      "learning_rate": 1.6944868748674142e-07,
      "loss": 0.6659,
      "step": 2812
    },
    {
      "epoch": 0.2780812099943158,
      "grad_norm": 3.7231378919294595,
      "learning_rate": 1.6942564455461873e-07,
      "loss": 0.6045,
      "step": 2813
    },
    {
      "epoch": 0.2781800657390703,
      "grad_norm": 13.448174180378265,
      "learning_rate": 1.6940259450392032e-07,
      "loss": 0.777,
      "step": 2814
    },
    {
      "epoch": 0.2782789214838247,
      "grad_norm": 4.873261974364326,
      "learning_rate": 1.6937953733700954e-07,
      "loss": 0.7609,
      "step": 2815
    },
    {
      "epoch": 0.2783777772285792,
      "grad_norm": 5.907146252544327,
      "learning_rate": 1.6935647305625068e-07,
      "loss": 0.7629,
      "step": 2816
    },
    {
      "epoch": 0.27847663297333364,
      "grad_norm": 8.734817881861092,
      "learning_rate": 1.6933340166400858e-07,
      "loss": 0.7878,
      "step": 2817
    },
    {
      "epoch": 0.2785754887180881,
      "grad_norm": 4.158507642036664,
      "learning_rate": 1.693103231626489e-07,
      "loss": 0.7645,
      "step": 2818
    },
    {
      "epoch": 0.2786743444628426,
      "grad_norm": 4.200664521405948,
      "learning_rate": 1.692872375545379e-07,
      "loss": 0.7476,
      "step": 2819
    },
    {
      "epoch": 0.27877320020759705,
      "grad_norm": 3.5485701523357136,
      "learning_rate": 1.6926414484204283e-07,
      "loss": 0.6651,
      "step": 2820
    },
    {
      "epoch": 0.27887205595235154,
      "grad_norm": 4.324023948982937,
      "learning_rate": 1.692410450275314e-07,
      "loss": 0.6624,
      "step": 2821
    },
    {
      "epoch": 0.27897091169710597,
      "grad_norm": 4.333034138510861,
      "learning_rate": 1.6921793811337217e-07,
      "loss": 0.6685,
      "step": 2822
    },
    {
      "epoch": 0.27906976744186046,
      "grad_norm": 9.068404695842725,
      "learning_rate": 1.6919482410193443e-07,
      "loss": 0.6782,
      "step": 2823
    },
    {
      "epoch": 0.27916862318661495,
      "grad_norm": 3.2363748983731937,
      "learning_rate": 1.6917170299558818e-07,
      "loss": 0.5884,
      "step": 2824
    },
    {
      "epoch": 0.2792674789313694,
      "grad_norm": 3.665437285190474,
      "learning_rate": 1.6914857479670412e-07,
      "loss": 0.6785,
      "step": 2825
    },
    {
      "epoch": 0.2793663346761239,
      "grad_norm": 4.855405481919817,
      "learning_rate": 1.6912543950765374e-07,
      "loss": 0.7524,
      "step": 2826
    },
    {
      "epoch": 0.2794651904208783,
      "grad_norm": 6.211779238103908,
      "learning_rate": 1.6910229713080922e-07,
      "loss": 0.7208,
      "step": 2827
    },
    {
      "epoch": 0.2795640461656328,
      "grad_norm": 4.155043109512756,
      "learning_rate": 1.6907914766854344e-07,
      "loss": 0.7273,
      "step": 2828
    },
    {
      "epoch": 0.2796629019103873,
      "grad_norm": 3.969927702184816,
      "learning_rate": 1.6905599112323008e-07,
      "loss": 0.7038,
      "step": 2829
    },
    {
      "epoch": 0.2797617576551417,
      "grad_norm": 4.833828568473684,
      "learning_rate": 1.6903282749724345e-07,
      "loss": 0.7982,
      "step": 2830
    },
    {
      "epoch": 0.2798606133998962,
      "grad_norm": 5.781415210383675,
      "learning_rate": 1.6900965679295868e-07,
      "loss": 0.6931,
      "step": 2831
    },
    {
      "epoch": 0.27995946914465064,
      "grad_norm": 4.383609742721973,
      "learning_rate": 1.6898647901275157e-07,
      "loss": 0.7466,
      "step": 2832
    },
    {
      "epoch": 0.28005832488940513,
      "grad_norm": 5.016381818609073,
      "learning_rate": 1.6896329415899868e-07,
      "loss": 0.7556,
      "step": 2833
    },
    {
      "epoch": 0.2801571806341596,
      "grad_norm": 3.682293338678717,
      "learning_rate": 1.6894010223407724e-07,
      "loss": 0.7194,
      "step": 2834
    },
    {
      "epoch": 0.28025603637891405,
      "grad_norm": 17.37940688854949,
      "learning_rate": 1.6891690324036526e-07,
      "loss": 0.7496,
      "step": 2835
    },
    {
      "epoch": 0.28035489212366854,
      "grad_norm": 5.225918800480516,
      "learning_rate": 1.6889369718024143e-07,
      "loss": 0.7189,
      "step": 2836
    },
    {
      "epoch": 0.28045374786842303,
      "grad_norm": 2.971087388548522,
      "learning_rate": 1.6887048405608519e-07,
      "loss": 0.7428,
      "step": 2837
    },
    {
      "epoch": 0.28055260361317746,
      "grad_norm": 3.7216510252264103,
      "learning_rate": 1.6884726387027678e-07,
      "loss": 0.8743,
      "step": 2838
    },
    {
      "epoch": 0.28065145935793195,
      "grad_norm": 3.237759571123034,
      "learning_rate": 1.68824036625197e-07,
      "loss": 0.6934,
      "step": 2839
    },
    {
      "epoch": 0.2807503151026864,
      "grad_norm": 4.211631839018718,
      "learning_rate": 1.6880080232322746e-07,
      "loss": 0.6126,
      "step": 2840
    },
    {
      "epoch": 0.2808491708474409,
      "grad_norm": 6.430570881994642,
      "learning_rate": 1.6877756096675054e-07,
      "loss": 0.6701,
      "step": 2841
    },
    {
      "epoch": 0.28094802659219537,
      "grad_norm": 4.086316188650863,
      "learning_rate": 1.687543125581493e-07,
      "loss": 0.6913,
      "step": 2842
    },
    {
      "epoch": 0.2810468823369498,
      "grad_norm": 4.714319876249902,
      "learning_rate": 1.687310570998075e-07,
      "loss": 0.7544,
      "step": 2843
    },
    {
      "epoch": 0.2811457380817043,
      "grad_norm": 10.870422498915262,
      "learning_rate": 1.6870779459410965e-07,
      "loss": 0.6762,
      "step": 2844
    },
    {
      "epoch": 0.2812445938264587,
      "grad_norm": 5.722773257808273,
      "learning_rate": 1.6868452504344096e-07,
      "loss": 0.6675,
      "step": 2845
    },
    {
      "epoch": 0.2813434495712132,
      "grad_norm": 5.084053881943481,
      "learning_rate": 1.6866124845018743e-07,
      "loss": 0.8293,
      "step": 2846
    },
    {
      "epoch": 0.2814423053159677,
      "grad_norm": 3.616338770554812,
      "learning_rate": 1.6863796481673567e-07,
      "loss": 0.7257,
      "step": 2847
    },
    {
      "epoch": 0.28154116106072213,
      "grad_norm": 6.414030303976801,
      "learning_rate": 1.686146741454731e-07,
      "loss": 0.7652,
      "step": 2848
    },
    {
      "epoch": 0.2816400168054766,
      "grad_norm": 6.649939923045427,
      "learning_rate": 1.6859137643878783e-07,
      "loss": 0.7246,
      "step": 2849
    },
    {
      "epoch": 0.28173887255023106,
      "grad_norm": 3.516095853527276,
      "learning_rate": 1.6856807169906873e-07,
      "loss": 0.7709,
      "step": 2850
    },
    {
      "epoch": 0.28183772829498555,
      "grad_norm": 3.9424212774849736,
      "learning_rate": 1.685447599287053e-07,
      "loss": 0.7738,
      "step": 2851
    },
    {
      "epoch": 0.28193658403974003,
      "grad_norm": 11.92224343241575,
      "learning_rate": 1.6852144113008784e-07,
      "loss": 0.6406,
      "step": 2852
    },
    {
      "epoch": 0.28203543978449447,
      "grad_norm": 3.196653867262117,
      "learning_rate": 1.6849811530560737e-07,
      "loss": 0.8113,
      "step": 2853
    },
    {
      "epoch": 0.28213429552924896,
      "grad_norm": 4.951119183837967,
      "learning_rate": 1.6847478245765562e-07,
      "loss": 0.6373,
      "step": 2854
    },
    {
      "epoch": 0.2822331512740034,
      "grad_norm": 5.919404959664721,
      "learning_rate": 1.6845144258862495e-07,
      "loss": 0.7213,
      "step": 2855
    },
    {
      "epoch": 0.2823320070187579,
      "grad_norm": 3.2095734786175063,
      "learning_rate": 1.684280957009086e-07,
      "loss": 0.6944,
      "step": 2856
    },
    {
      "epoch": 0.28243086276351237,
      "grad_norm": 4.069845856582812,
      "learning_rate": 1.684047417969004e-07,
      "loss": 0.6189,
      "step": 2857
    },
    {
      "epoch": 0.2825297185082668,
      "grad_norm": 11.518702082261028,
      "learning_rate": 1.6838138087899497e-07,
      "loss": 0.6965,
      "step": 2858
    },
    {
      "epoch": 0.2826285742530213,
      "grad_norm": 4.766719028162733,
      "learning_rate": 1.6835801294958764e-07,
      "loss": 0.7674,
      "step": 2859
    },
    {
      "epoch": 0.2827274299977757,
      "grad_norm": 5.515020459840256,
      "learning_rate": 1.6833463801107447e-07,
      "loss": 0.6982,
      "step": 2860
    },
    {
      "epoch": 0.2828262857425302,
      "grad_norm": 6.928467974207027,
      "learning_rate": 1.6831125606585213e-07,
      "loss": 0.7151,
      "step": 2861
    },
    {
      "epoch": 0.2829251414872847,
      "grad_norm": 5.051392298917339,
      "learning_rate": 1.6828786711631816e-07,
      "loss": 0.7204,
      "step": 2862
    },
    {
      "epoch": 0.28302399723203914,
      "grad_norm": 4.862576416082069,
      "learning_rate": 1.6826447116487072e-07,
      "loss": 0.7921,
      "step": 2863
    },
    {
      "epoch": 0.2831228529767936,
      "grad_norm": 3.4006399713670428,
      "learning_rate": 1.6824106821390873e-07,
      "loss": 0.7188,
      "step": 2864
    },
    {
      "epoch": 0.28322170872154806,
      "grad_norm": 4.348481050486068,
      "learning_rate": 1.6821765826583185e-07,
      "loss": 0.6761,
      "step": 2865
    },
    {
      "epoch": 0.28332056446630255,
      "grad_norm": 3.636123668405624,
      "learning_rate": 1.681942413230404e-07,
      "loss": 0.6679,
      "step": 2866
    },
    {
      "epoch": 0.28341942021105704,
      "grad_norm": 3.996002290216521,
      "learning_rate": 1.681708173879354e-07,
      "loss": 0.6948,
      "step": 2867
    },
    {
      "epoch": 0.28351827595581147,
      "grad_norm": 4.59692825259003,
      "learning_rate": 1.6814738646291867e-07,
      "loss": 0.759,
      "step": 2868
    },
    {
      "epoch": 0.28361713170056596,
      "grad_norm": 3.6556817858709327,
      "learning_rate": 1.6812394855039275e-07,
      "loss": 0.7179,
      "step": 2869
    },
    {
      "epoch": 0.2837159874453204,
      "grad_norm": 18.985045924265723,
      "learning_rate": 1.6810050365276078e-07,
      "loss": 0.6925,
      "step": 2870
    },
    {
      "epoch": 0.2838148431900749,
      "grad_norm": 3.6442241566462314,
      "learning_rate": 1.680770517724267e-07,
      "loss": 0.7531,
      "step": 2871
    },
    {
      "epoch": 0.2839136989348294,
      "grad_norm": 5.341634637513418,
      "learning_rate": 1.6805359291179516e-07,
      "loss": 0.6976,
      "step": 2872
    },
    {
      "epoch": 0.2840125546795838,
      "grad_norm": 3.9254379925815375,
      "learning_rate": 1.6803012707327157e-07,
      "loss": 0.7132,
      "step": 2873
    },
    {
      "epoch": 0.2841114104243383,
      "grad_norm": 6.264937793549122,
      "learning_rate": 1.6800665425926193e-07,
      "loss": 0.8026,
      "step": 2874
    },
    {
      "epoch": 0.28421026616909273,
      "grad_norm": 4.232157567778203,
      "learning_rate": 1.6798317447217306e-07,
      "loss": 0.6403,
      "step": 2875
    },
    {
      "epoch": 0.2843091219138472,
      "grad_norm": 3.9715828648934934,
      "learning_rate": 1.679596877144125e-07,
      "loss": 0.7674,
      "step": 2876
    },
    {
      "epoch": 0.2844079776586017,
      "grad_norm": 10.06476920334696,
      "learning_rate": 1.6793619398838841e-07,
      "loss": 0.6966,
      "step": 2877
    },
    {
      "epoch": 0.28450683340335614,
      "grad_norm": 4.683602055792839,
      "learning_rate": 1.679126932965098e-07,
      "loss": 0.7873,
      "step": 2878
    },
    {
      "epoch": 0.28460568914811063,
      "grad_norm": 21.737212338572675,
      "learning_rate": 1.6788918564118624e-07,
      "loss": 0.7248,
      "step": 2879
    },
    {
      "epoch": 0.28470454489286506,
      "grad_norm": 3.6208041225978596,
      "learning_rate": 1.6786567102482812e-07,
      "loss": 0.7503,
      "step": 2880
    },
    {
      "epoch": 0.28480340063761955,
      "grad_norm": 4.063916175889555,
      "learning_rate": 1.6784214944984653e-07,
      "loss": 0.7492,
      "step": 2881
    },
    {
      "epoch": 0.28490225638237404,
      "grad_norm": 4.771983223129476,
      "learning_rate": 1.678186209186533e-07,
      "loss": 0.7795,
      "step": 2882
    },
    {
      "epoch": 0.2850011121271285,
      "grad_norm": 3.478213936421685,
      "learning_rate": 1.6779508543366085e-07,
      "loss": 0.7681,
      "step": 2883
    },
    {
      "epoch": 0.28509996787188296,
      "grad_norm": 5.214815118640428,
      "learning_rate": 1.6777154299728244e-07,
      "loss": 0.8241,
      "step": 2884
    },
    {
      "epoch": 0.2851988236166374,
      "grad_norm": 5.3560928952892555,
      "learning_rate": 1.6774799361193197e-07,
      "loss": 0.7107,
      "step": 2885
    },
    {
      "epoch": 0.2852976793613919,
      "grad_norm": 4.816028933181752,
      "learning_rate": 1.6772443728002414e-07,
      "loss": 0.6983,
      "step": 2886
    },
    {
      "epoch": 0.2853965351061464,
      "grad_norm": 6.814466816749082,
      "learning_rate": 1.677008740039743e-07,
      "loss": 0.6199,
      "step": 2887
    },
    {
      "epoch": 0.2854953908509008,
      "grad_norm": 10.708775733874663,
      "learning_rate": 1.6767730378619845e-07,
      "loss": 0.6948,
      "step": 2888
    },
    {
      "epoch": 0.2855942465956553,
      "grad_norm": 3.3777068158072825,
      "learning_rate": 1.676537266291134e-07,
      "loss": 0.7592,
      "step": 2889
    },
    {
      "epoch": 0.28569310234040973,
      "grad_norm": 4.280452404553901,
      "learning_rate": 1.6763014253513666e-07,
      "loss": 0.6916,
      "step": 2890
    },
    {
      "epoch": 0.2857919580851642,
      "grad_norm": 3.6163916616914062,
      "learning_rate": 1.6760655150668644e-07,
      "loss": 0.6249,
      "step": 2891
    },
    {
      "epoch": 0.2858908138299187,
      "grad_norm": 4.932592597327992,
      "learning_rate": 1.675829535461816e-07,
      "loss": 0.806,
      "step": 2892
    },
    {
      "epoch": 0.28598966957467314,
      "grad_norm": 5.705966311411102,
      "learning_rate": 1.675593486560418e-07,
      "loss": 0.6412,
      "step": 2893
    },
    {
      "epoch": 0.28608852531942763,
      "grad_norm": 4.431458251651551,
      "learning_rate": 1.675357368386874e-07,
      "loss": 0.7001,
      "step": 2894
    },
    {
      "epoch": 0.28618738106418207,
      "grad_norm": 3.1674056628654914,
      "learning_rate": 1.6751211809653937e-07,
      "loss": 0.7109,
      "step": 2895
    },
    {
      "epoch": 0.28628623680893656,
      "grad_norm": 3.350369820753518,
      "learning_rate": 1.6748849243201948e-07,
      "loss": 0.7188,
      "step": 2896
    },
    {
      "epoch": 0.28638509255369105,
      "grad_norm": 4.254320612478171,
      "learning_rate": 1.6746485984755026e-07,
      "loss": 0.7023,
      "step": 2897
    },
    {
      "epoch": 0.2864839482984455,
      "grad_norm": 3.7109889419325794,
      "learning_rate": 1.6744122034555482e-07,
      "loss": 0.8068,
      "step": 2898
    },
    {
      "epoch": 0.28658280404319997,
      "grad_norm": 5.706799763509911,
      "learning_rate": 1.6741757392845707e-07,
      "loss": 0.7974,
      "step": 2899
    },
    {
      "epoch": 0.2866816597879544,
      "grad_norm": 5.869183973296711,
      "learning_rate": 1.673939205986816e-07,
      "loss": 0.7768,
      "step": 2900
    },
    {
      "epoch": 0.2867805155327089,
      "grad_norm": 3.2361292670326733,
      "learning_rate": 1.6737026035865367e-07,
      "loss": 0.6772,
      "step": 2901
    },
    {
      "epoch": 0.2868793712774634,
      "grad_norm": 3.462386269100872,
      "learning_rate": 1.6734659321079937e-07,
      "loss": 0.6761,
      "step": 2902
    },
    {
      "epoch": 0.2869782270222178,
      "grad_norm": 4.243407352340233,
      "learning_rate": 1.6732291915754532e-07,
      "loss": 0.7636,
      "step": 2903
    },
    {
      "epoch": 0.2870770827669723,
      "grad_norm": 4.082366359224146,
      "learning_rate": 1.6729923820131902e-07,
      "loss": 0.6824,
      "step": 2904
    },
    {
      "epoch": 0.28717593851172674,
      "grad_norm": 4.0235471054253305,
      "learning_rate": 1.6727555034454858e-07,
      "loss": 0.6207,
      "step": 2905
    },
    {
      "epoch": 0.2872747942564812,
      "grad_norm": 50.42877646547694,
      "learning_rate": 1.6725185558966283e-07,
      "loss": 0.6896,
      "step": 2906
    },
    {
      "epoch": 0.2873736500012357,
      "grad_norm": 24.766436414665236,
      "learning_rate": 1.672281539390913e-07,
      "loss": 0.7342,
      "step": 2907
    },
    {
      "epoch": 0.28747250574599015,
      "grad_norm": 6.078040851964941,
      "learning_rate": 1.672044453952643e-07,
      "loss": 0.7116,
      "step": 2908
    },
    {
      "epoch": 0.28757136149074464,
      "grad_norm": 3.697732134158651,
      "learning_rate": 1.6718072996061278e-07,
      "loss": 0.6656,
      "step": 2909
    },
    {
      "epoch": 0.28767021723549907,
      "grad_norm": 3.7794213993260284,
      "learning_rate": 1.6715700763756833e-07,
      "loss": 0.7212,
      "step": 2910
    },
    {
      "epoch": 0.28776907298025356,
      "grad_norm": 4.452822381337555,
      "learning_rate": 1.671332784285634e-07,
      "loss": 0.7014,
      "step": 2911
    },
    {
      "epoch": 0.28786792872500805,
      "grad_norm": 5.330463295720958,
      "learning_rate": 1.671095423360311e-07,
      "loss": 0.7317,
      "step": 2912
    },
    {
      "epoch": 0.2879667844697625,
      "grad_norm": 5.5598647708252535,
      "learning_rate": 1.6708579936240511e-07,
      "loss": 0.6969,
      "step": 2913
    },
    {
      "epoch": 0.28806564021451697,
      "grad_norm": 14.046334380035546,
      "learning_rate": 1.6706204951012002e-07,
      "loss": 0.6786,
      "step": 2914
    },
    {
      "epoch": 0.28816449595927146,
      "grad_norm": 6.977130181158292,
      "learning_rate": 1.6703829278161097e-07,
      "loss": 0.6661,
      "step": 2915
    },
    {
      "epoch": 0.2882633517040259,
      "grad_norm": 6.275243500853885,
      "learning_rate": 1.6701452917931389e-07,
      "loss": 0.68,
      "step": 2916
    },
    {
      "epoch": 0.2883622074487804,
      "grad_norm": 3.3461338033201344,
      "learning_rate": 1.6699075870566536e-07,
      "loss": 0.6708,
      "step": 2917
    },
    {
      "epoch": 0.2884610631935348,
      "grad_norm": 4.741526704915874,
      "learning_rate": 1.6696698136310272e-07,
      "loss": 0.6326,
      "step": 2918
    },
    {
      "epoch": 0.2885599189382893,
      "grad_norm": 2.9948196367244,
      "learning_rate": 1.6694319715406398e-07,
      "loss": 0.7366,
      "step": 2919
    },
    {
      "epoch": 0.2886587746830438,
      "grad_norm": 4.477561455645616,
      "learning_rate": 1.6691940608098784e-07,
      "loss": 0.7147,
      "step": 2920
    },
    {
      "epoch": 0.28875763042779823,
      "grad_norm": 4.009977902128466,
      "learning_rate": 1.6689560814631374e-07,
      "loss": 0.749,
      "step": 2921
    },
    {
      "epoch": 0.2888564861725527,
      "grad_norm": 9.988636877376875,
      "learning_rate": 1.6687180335248182e-07,
      "loss": 0.7416,
      "step": 2922
    },
    {
      "epoch": 0.28895534191730715,
      "grad_norm": 8.075965488731708,
      "learning_rate": 1.6684799170193287e-07,
      "loss": 0.66,
      "step": 2923
    },
    {
      "epoch": 0.28905419766206164,
      "grad_norm": 4.660852732184505,
      "learning_rate": 1.6682417319710846e-07,
      "loss": 0.7228,
      "step": 2924
    },
    {
      "epoch": 0.28915305340681613,
      "grad_norm": 3.568649443570103,
      "learning_rate": 1.668003478404508e-07,
      "loss": 0.7191,
      "step": 2925
    },
    {
      "epoch": 0.28925190915157056,
      "grad_norm": 6.416431315859174,
      "learning_rate": 1.6677651563440285e-07,
      "loss": 0.7531,
      "step": 2926
    },
    {
      "epoch": 0.28935076489632505,
      "grad_norm": 10.93267692091047,
      "learning_rate": 1.6675267658140826e-07,
      "loss": 0.6913,
      "step": 2927
    },
    {
      "epoch": 0.2894496206410795,
      "grad_norm": 3.0879162301742618,
      "learning_rate": 1.6672883068391132e-07,
      "loss": 0.7046,
      "step": 2928
    },
    {
      "epoch": 0.289548476385834,
      "grad_norm": 5.983675047586937,
      "learning_rate": 1.6670497794435717e-07,
      "loss": 0.8554,
      "step": 2929
    },
    {
      "epoch": 0.28964733213058846,
      "grad_norm": 4.349238384685525,
      "learning_rate": 1.6668111836519147e-07,
      "loss": 0.686,
      "step": 2930
    },
    {
      "epoch": 0.2897461878753429,
      "grad_norm": 4.511126269322746,
      "learning_rate": 1.666572519488607e-07,
      "loss": 0.7194,
      "step": 2931
    },
    {
      "epoch": 0.2898450436200974,
      "grad_norm": 7.16640082927601,
      "learning_rate": 1.6663337869781198e-07,
      "loss": 0.5864,
      "step": 2932
    },
    {
      "epoch": 0.2899438993648518,
      "grad_norm": 4.966485168665288,
      "learning_rate": 1.6660949861449322e-07,
      "loss": 0.6871,
      "step": 2933
    },
    {
      "epoch": 0.2900427551096063,
      "grad_norm": 5.7887026562655395,
      "learning_rate": 1.6658561170135291e-07,
      "loss": 0.6512,
      "step": 2934
    },
    {
      "epoch": 0.2901416108543608,
      "grad_norm": 5.166783202317499,
      "learning_rate": 1.6656171796084039e-07,
      "loss": 0.7067,
      "step": 2935
    },
    {
      "epoch": 0.29024046659911523,
      "grad_norm": 3.9151767244014373,
      "learning_rate": 1.6653781739540547e-07,
      "loss": 0.7325,
      "step": 2936
    },
    {
      "epoch": 0.2903393223438697,
      "grad_norm": 3.8487066115986357,
      "learning_rate": 1.665139100074989e-07,
      "loss": 0.6924,
      "step": 2937
    },
    {
      "epoch": 0.29043817808862415,
      "grad_norm": 3.7980176192593036,
      "learning_rate": 1.6648999579957204e-07,
      "loss": 0.8204,
      "step": 2938
    },
    {
      "epoch": 0.29053703383337864,
      "grad_norm": 3.200030341089195,
      "learning_rate": 1.6646607477407687e-07,
      "loss": 0.747,
      "step": 2939
    },
    {
      "epoch": 0.29063588957813313,
      "grad_norm": 6.182722955579262,
      "learning_rate": 1.664421469334662e-07,
      "loss": 0.7084,
      "step": 2940
    },
    {
      "epoch": 0.29073474532288757,
      "grad_norm": 3.8641692473348304,
      "learning_rate": 1.6641821228019345e-07,
      "loss": 0.7844,
      "step": 2941
    },
    {
      "epoch": 0.29083360106764206,
      "grad_norm": 4.942923102788017,
      "learning_rate": 1.6639427081671277e-07,
      "loss": 0.6582,
      "step": 2942
    },
    {
      "epoch": 0.2909324568123965,
      "grad_norm": 8.762598659639725,
      "learning_rate": 1.6637032254547898e-07,
      "loss": 0.6555,
      "step": 2943
    },
    {
      "epoch": 0.291031312557151,
      "grad_norm": 6.508385031944667,
      "learning_rate": 1.6634636746894762e-07,
      "loss": 0.6483,
      "step": 2944
    },
    {
      "epoch": 0.29113016830190547,
      "grad_norm": 8.7970375321157,
      "learning_rate": 1.6632240558957502e-07,
      "loss": 0.7103,
      "step": 2945
    },
    {
      "epoch": 0.2912290240466599,
      "grad_norm": 5.671693045505708,
      "learning_rate": 1.66298436909818e-07,
      "loss": 0.8462,
      "step": 2946
    },
    {
      "epoch": 0.2913278797914144,
      "grad_norm": 3.6232265698574357,
      "learning_rate": 1.6627446143213427e-07,
      "loss": 0.7414,
      "step": 2947
    },
    {
      "epoch": 0.2914267355361688,
      "grad_norm": 8.020449977608214,
      "learning_rate": 1.6625047915898214e-07,
      "loss": 0.7268,
      "step": 2948
    },
    {
      "epoch": 0.2915255912809233,
      "grad_norm": 7.2283264676661325,
      "learning_rate": 1.6622649009282065e-07,
      "loss": 0.6989,
      "step": 2949
    },
    {
      "epoch": 0.2916244470256778,
      "grad_norm": 5.220455960744127,
      "learning_rate": 1.6620249423610947e-07,
      "loss": 0.7514,
      "step": 2950
    },
    {
      "epoch": 0.29172330277043224,
      "grad_norm": 3.7334273700753213,
      "learning_rate": 1.661784915913091e-07,
      "loss": 0.758,
      "step": 2951
    },
    {
      "epoch": 0.2918221585151867,
      "grad_norm": 18.19610169147317,
      "learning_rate": 1.661544821608806e-07,
      "loss": 0.7377,
      "step": 2952
    },
    {
      "epoch": 0.29192101425994116,
      "grad_norm": 4.36827086675971,
      "learning_rate": 1.661304659472858e-07,
      "loss": 0.6868,
      "step": 2953
    },
    {
      "epoch": 0.29201987000469565,
      "grad_norm": 48.63254532072366,
      "learning_rate": 1.6610644295298722e-07,
      "loss": 0.6601,
      "step": 2954
    },
    {
      "epoch": 0.29211872574945014,
      "grad_norm": 4.443718576682958,
      "learning_rate": 1.6608241318044804e-07,
      "loss": 0.7658,
      "step": 2955
    },
    {
      "epoch": 0.29221758149420457,
      "grad_norm": 4.005970576124661,
      "learning_rate": 1.6605837663213217e-07,
      "loss": 0.695,
      "step": 2956
    },
    {
      "epoch": 0.29231643723895906,
      "grad_norm": 4.598474590093278,
      "learning_rate": 1.6603433331050423e-07,
      "loss": 0.7404,
      "step": 2957
    },
    {
      "epoch": 0.2924152929837135,
      "grad_norm": 7.327568675665927,
      "learning_rate": 1.6601028321802947e-07,
      "loss": 0.6888,
      "step": 2958
    },
    {
      "epoch": 0.292514148728468,
      "grad_norm": 4.587068098689788,
      "learning_rate": 1.6598622635717388e-07,
      "loss": 0.7501,
      "step": 2959
    },
    {
      "epoch": 0.29261300447322247,
      "grad_norm": 4.333076312326571,
      "learning_rate": 1.6596216273040415e-07,
      "loss": 0.6223,
      "step": 2960
    },
    {
      "epoch": 0.2927118602179769,
      "grad_norm": 8.160101619169971,
      "learning_rate": 1.6593809234018763e-07,
      "loss": 0.6468,
      "step": 2961
    },
    {
      "epoch": 0.2928107159627314,
      "grad_norm": 4.433412237394138,
      "learning_rate": 1.659140151889924e-07,
      "loss": 0.7401,
      "step": 2962
    },
    {
      "epoch": 0.2929095717074858,
      "grad_norm": 5.372283018107874,
      "learning_rate": 1.6588993127928722e-07,
      "loss": 0.6307,
      "step": 2963
    },
    {
      "epoch": 0.2930084274522403,
      "grad_norm": 3.127936750932502,
      "learning_rate": 1.6586584061354155e-07,
      "loss": 0.8461,
      "step": 2964
    },
    {
      "epoch": 0.2931072831969948,
      "grad_norm": 3.194941310910834,
      "learning_rate": 1.6584174319422548e-07,
      "loss": 0.6365,
      "step": 2965
    },
    {
      "epoch": 0.29320613894174924,
      "grad_norm": 5.56381485861516,
      "learning_rate": 1.658176390238099e-07,
      "loss": 0.7335,
      "step": 2966
    },
    {
      "epoch": 0.29330499468650373,
      "grad_norm": 19.961263715217665,
      "learning_rate": 1.657935281047663e-07,
      "loss": 0.6592,
      "step": 2967
    },
    {
      "epoch": 0.29340385043125816,
      "grad_norm": 9.051526679705852,
      "learning_rate": 1.6576941043956694e-07,
      "loss": 0.8002,
      "step": 2968
    },
    {
      "epoch": 0.29350270617601265,
      "grad_norm": 3.152604692712811,
      "learning_rate": 1.6574528603068468e-07,
      "loss": 0.6581,
      "step": 2969
    },
    {
      "epoch": 0.29360156192076714,
      "grad_norm": 5.272157188295804,
      "learning_rate": 1.6572115488059317e-07,
      "loss": 0.6278,
      "step": 2970
    },
    {
      "epoch": 0.2937004176655216,
      "grad_norm": 7.168307922641315,
      "learning_rate": 1.6569701699176667e-07,
      "loss": 0.7023,
      "step": 2971
    },
    {
      "epoch": 0.29379927341027606,
      "grad_norm": 4.093847027078792,
      "learning_rate": 1.6567287236668018e-07,
      "loss": 0.7866,
      "step": 2972
    },
    {
      "epoch": 0.2938981291550305,
      "grad_norm": 3.5160967723592917,
      "learning_rate": 1.6564872100780938e-07,
      "loss": 0.705,
      "step": 2973
    },
    {
      "epoch": 0.293996984899785,
      "grad_norm": 3.9093644109658814,
      "learning_rate": 1.656245629176306e-07,
      "loss": 0.7184,
      "step": 2974
    },
    {
      "epoch": 0.2940958406445395,
      "grad_norm": 3.4994053117086885,
      "learning_rate": 1.6560039809862096e-07,
      "loss": 0.7201,
      "step": 2975
    },
    {
      "epoch": 0.2941946963892939,
      "grad_norm": 5.46941639257022,
      "learning_rate": 1.6557622655325812e-07,
      "loss": 0.7074,
      "step": 2976
    },
    {
      "epoch": 0.2942935521340484,
      "grad_norm": 8.625950544010237,
      "learning_rate": 1.6555204828402062e-07,
      "loss": 0.7551,
      "step": 2977
    },
    {
      "epoch": 0.29439240787880283,
      "grad_norm": 3.44053630922227,
      "learning_rate": 1.655278632933875e-07,
      "loss": 0.7024,
      "step": 2978
    },
    {
      "epoch": 0.2944912636235573,
      "grad_norm": 3.2769510141931586,
      "learning_rate": 1.6550367158383863e-07,
      "loss": 0.6996,
      "step": 2979
    },
    {
      "epoch": 0.2945901193683118,
      "grad_norm": 5.087418971183405,
      "learning_rate": 1.6547947315785445e-07,
      "loss": 0.7046,
      "step": 2980
    },
    {
      "epoch": 0.29468897511306624,
      "grad_norm": 4.603317339518199,
      "learning_rate": 1.6545526801791622e-07,
      "loss": 0.6772,
      "step": 2981
    },
    {
      "epoch": 0.29478783085782073,
      "grad_norm": 4.200853302529868,
      "learning_rate": 1.6543105616650578e-07,
      "loss": 0.7141,
      "step": 2982
    },
    {
      "epoch": 0.29488668660257517,
      "grad_norm": 4.358310546584248,
      "learning_rate": 1.6540683760610569e-07,
      "loss": 0.681,
      "step": 2983
    },
    {
      "epoch": 0.29498554234732965,
      "grad_norm": 3.443361249799756,
      "learning_rate": 1.653826123391992e-07,
      "loss": 0.6801,
      "step": 2984
    },
    {
      "epoch": 0.29508439809208414,
      "grad_norm": 4.0707045879363655,
      "learning_rate": 1.6535838036827033e-07,
      "loss": 0.7006,
      "step": 2985
    },
    {
      "epoch": 0.2951832538368386,
      "grad_norm": 3.4866961306645767,
      "learning_rate": 1.6533414169580363e-07,
      "loss": 0.6722,
      "step": 2986
    },
    {
      "epoch": 0.29528210958159307,
      "grad_norm": 4.334466789087875,
      "learning_rate": 1.6530989632428446e-07,
      "loss": 0.72,
      "step": 2987
    },
    {
      "epoch": 0.2953809653263475,
      "grad_norm": 5.9272484422229645,
      "learning_rate": 1.6528564425619877e-07,
      "loss": 0.7018,
      "step": 2988
    },
    {
      "epoch": 0.295479821071102,
      "grad_norm": 3.35716496417885,
      "learning_rate": 1.6526138549403332e-07,
      "loss": 0.6778,
      "step": 2989
    },
    {
      "epoch": 0.2955786768158565,
      "grad_norm": 4.141959897175426,
      "learning_rate": 1.652371200402755e-07,
      "loss": 0.6012,
      "step": 2990
    },
    {
      "epoch": 0.2956775325606109,
      "grad_norm": 3.8420179003946893,
      "learning_rate": 1.652128478974133e-07,
      "loss": 0.7247,
      "step": 2991
    },
    {
      "epoch": 0.2957763883053654,
      "grad_norm": 3.475362912219321,
      "learning_rate": 1.6518856906793547e-07,
      "loss": 0.6828,
      "step": 2992
    },
    {
      "epoch": 0.2958752440501199,
      "grad_norm": 6.585532581813327,
      "learning_rate": 1.6516428355433153e-07,
      "loss": 0.7764,
      "step": 2993
    },
    {
      "epoch": 0.2959740997948743,
      "grad_norm": 4.389184830521142,
      "learning_rate": 1.6513999135909152e-07,
      "loss": 0.7403,
      "step": 2994
    },
    {
      "epoch": 0.2960729555396288,
      "grad_norm": 5.106313007504171,
      "learning_rate": 1.651156924847063e-07,
      "loss": 0.7402,
      "step": 2995
    },
    {
      "epoch": 0.29617181128438325,
      "grad_norm": 3.877936240751328,
      "learning_rate": 1.6509138693366732e-07,
      "loss": 0.761,
      "step": 2996
    },
    {
      "epoch": 0.29627066702913774,
      "grad_norm": 3.078411657447238,
      "learning_rate": 1.650670747084668e-07,
      "loss": 0.7418,
      "step": 2997
    },
    {
      "epoch": 0.2963695227738922,
      "grad_norm": 3.9771299924636927,
      "learning_rate": 1.6504275581159758e-07,
      "loss": 0.7953,
      "step": 2998
    },
    {
      "epoch": 0.29646837851864666,
      "grad_norm": 6.894003844094074,
      "learning_rate": 1.6501843024555318e-07,
      "loss": 0.6874,
      "step": 2999
    },
    {
      "epoch": 0.29656723426340115,
      "grad_norm": 3.5467093215905994,
      "learning_rate": 1.6499409801282785e-07,
      "loss": 0.7513,
      "step": 3000
    },
    {
      "epoch": 0.2966660900081556,
      "grad_norm": 12.18314006526318,
      "learning_rate": 1.649697591159165e-07,
      "loss": 0.7785,
      "step": 3001
    },
    {
      "epoch": 0.29676494575291007,
      "grad_norm": 4.301698889518849,
      "learning_rate": 1.6494541355731474e-07,
      "loss": 0.7476,
      "step": 3002
    },
    {
      "epoch": 0.29686380149766456,
      "grad_norm": 3.096021783492918,
      "learning_rate": 1.649210613395188e-07,
      "loss": 0.6774,
      "step": 3003
    },
    {
      "epoch": 0.296962657242419,
      "grad_norm": 13.242118369689427,
      "learning_rate": 1.648967024650257e-07,
      "loss": 0.8341,
      "step": 3004
    },
    {
      "epoch": 0.2970615129871735,
      "grad_norm": 3.1352313424120792,
      "learning_rate": 1.6487233693633303e-07,
      "loss": 0.706,
      "step": 3005
    },
    {
      "epoch": 0.2971603687319279,
      "grad_norm": 3.8988976560114414,
      "learning_rate": 1.6484796475593918e-07,
      "loss": 0.6745,
      "step": 3006
    },
    {
      "epoch": 0.2972592244766824,
      "grad_norm": 3.9653422286768474,
      "learning_rate": 1.648235859263431e-07,
      "loss": 0.678,
      "step": 3007
    },
    {
      "epoch": 0.2973580802214369,
      "grad_norm": 6.022121126067576,
      "learning_rate": 1.6479920045004446e-07,
      "loss": 0.8432,
      "step": 3008
    },
    {
      "epoch": 0.2974569359661913,
      "grad_norm": 4.468648450264459,
      "learning_rate": 1.647748083295437e-07,
      "loss": 0.8277,
      "step": 3009
    },
    {
      "epoch": 0.2975557917109458,
      "grad_norm": 3.6150772843579895,
      "learning_rate": 1.6475040956734186e-07,
      "loss": 0.7898,
      "step": 3010
    },
    {
      "epoch": 0.29765464745570025,
      "grad_norm": 3.276258235841093,
      "learning_rate": 1.6472600416594067e-07,
      "loss": 0.7221,
      "step": 3011
    },
    {
      "epoch": 0.29775350320045474,
      "grad_norm": 3.5035569586161253,
      "learning_rate": 1.647015921278425e-07,
      "loss": 0.7533,
      "step": 3012
    },
    {
      "epoch": 0.29785235894520923,
      "grad_norm": 4.3866409792970416,
      "learning_rate": 1.6467717345555044e-07,
      "loss": 0.7357,
      "step": 3013
    },
    {
      "epoch": 0.29795121468996366,
      "grad_norm": 3.2447351712563095,
      "learning_rate": 1.6465274815156836e-07,
      "loss": 0.6759,
      "step": 3014
    },
    {
      "epoch": 0.29805007043471815,
      "grad_norm": 3.276743809408426,
      "learning_rate": 1.646283162184006e-07,
      "loss": 0.7715,
      "step": 3015
    },
    {
      "epoch": 0.2981489261794726,
      "grad_norm": 8.38757623437377,
      "learning_rate": 1.6460387765855238e-07,
      "loss": 0.7909,
      "step": 3016
    },
    {
      "epoch": 0.2982477819242271,
      "grad_norm": 20.162705488559553,
      "learning_rate": 1.6457943247452944e-07,
      "loss": 0.8305,
      "step": 3017
    },
    {
      "epoch": 0.29834663766898156,
      "grad_norm": 3.832009024872091,
      "learning_rate": 1.6455498066883838e-07,
      "loss": 0.6396,
      "step": 3018
    },
    {
      "epoch": 0.298445493413736,
      "grad_norm": 2.8746847241118942,
      "learning_rate": 1.6453052224398624e-07,
      "loss": 0.8247,
      "step": 3019
    },
    {
      "epoch": 0.2985443491584905,
      "grad_norm": 7.952288280138214,
      "learning_rate": 1.6450605720248098e-07,
      "loss": 0.6961,
      "step": 3020
    },
    {
      "epoch": 0.2986432049032449,
      "grad_norm": 4.3646281982968365,
      "learning_rate": 1.6448158554683106e-07,
      "loss": 0.8318,
      "step": 3021
    },
    {
      "epoch": 0.2987420606479994,
      "grad_norm": 4.191030397469099,
      "learning_rate": 1.6445710727954572e-07,
      "loss": 0.6939,
      "step": 3022
    },
    {
      "epoch": 0.2988409163927539,
      "grad_norm": 19.31873245798296,
      "learning_rate": 1.6443262240313486e-07,
      "loss": 0.7921,
      "step": 3023
    },
    {
      "epoch": 0.29893977213750833,
      "grad_norm": 4.7396718496914625,
      "learning_rate": 1.64408130920109e-07,
      "loss": 0.6267,
      "step": 3024
    },
    {
      "epoch": 0.2990386278822628,
      "grad_norm": 4.417926287717722,
      "learning_rate": 1.643836328329794e-07,
      "loss": 0.7204,
      "step": 3025
    },
    {
      "epoch": 0.29913748362701725,
      "grad_norm": 3.367857407363134,
      "learning_rate": 1.6435912814425795e-07,
      "loss": 0.796,
      "step": 3026
    },
    {
      "epoch": 0.29923633937177174,
      "grad_norm": 4.088557351052837,
      "learning_rate": 1.643346168564573e-07,
      "loss": 0.8114,
      "step": 3027
    },
    {
      "epoch": 0.29933519511652623,
      "grad_norm": 4.122775506905657,
      "learning_rate": 1.6431009897209072e-07,
      "loss": 0.66,
      "step": 3028
    },
    {
      "epoch": 0.29943405086128067,
      "grad_norm": 4.55890112487925,
      "learning_rate": 1.6428557449367212e-07,
      "loss": 0.7867,
      "step": 3029
    },
    {
      "epoch": 0.29953290660603515,
      "grad_norm": 3.2382730956525854,
      "learning_rate": 1.6426104342371612e-07,
      "loss": 0.6183,
      "step": 3030
    },
    {
      "epoch": 0.2996317623507896,
      "grad_norm": 2.872709031471928,
      "learning_rate": 1.6423650576473803e-07,
      "loss": 0.7287,
      "step": 3031
    },
    {
      "epoch": 0.2997306180955441,
      "grad_norm": 5.044942029246941,
      "learning_rate": 1.6421196151925383e-07,
      "loss": 0.7432,
      "step": 3032
    },
    {
      "epoch": 0.29982947384029857,
      "grad_norm": 3.2838876070869003,
      "learning_rate": 1.6418741068978016e-07,
      "loss": 0.7317,
      "step": 3033
    },
    {
      "epoch": 0.299928329585053,
      "grad_norm": 7.504296292192675,
      "learning_rate": 1.641628532788344e-07,
      "loss": 0.643,
      "step": 3034
    },
    {
      "epoch": 0.3000271853298075,
      "grad_norm": 3.883738129807063,
      "learning_rate": 1.6413828928893445e-07,
      "loss": 0.826,
      "step": 3035
    },
    {
      "epoch": 0.3001260410745619,
      "grad_norm": 3.599532546755983,
      "learning_rate": 1.6411371872259906e-07,
      "loss": 0.7351,
      "step": 3036
    },
    {
      "epoch": 0.3002248968193164,
      "grad_norm": 3.2939757214087373,
      "learning_rate": 1.6408914158234756e-07,
      "loss": 0.7792,
      "step": 3037
    },
    {
      "epoch": 0.3003237525640709,
      "grad_norm": 5.85486341694874,
      "learning_rate": 1.6406455787069996e-07,
      "loss": 0.711,
      "step": 3038
    },
    {
      "epoch": 0.30042260830882533,
      "grad_norm": 31.16229070700125,
      "learning_rate": 1.64039967590177e-07,
      "loss": 0.6939,
      "step": 3039
    },
    {
      "epoch": 0.3005214640535798,
      "grad_norm": 4.099422928038333,
      "learning_rate": 1.6401537074330002e-07,
      "loss": 0.7602,
      "step": 3040
    },
    {
      "epoch": 0.30062031979833426,
      "grad_norm": 3.8823670954870972,
      "learning_rate": 1.6399076733259104e-07,
      "loss": 0.7163,
      "step": 3041
    },
    {
      "epoch": 0.30071917554308875,
      "grad_norm": 4.018243214502019,
      "learning_rate": 1.639661573605728e-07,
      "loss": 0.7302,
      "step": 3042
    },
    {
      "epoch": 0.30081803128784323,
      "grad_norm": 10.349509187919443,
      "learning_rate": 1.6394154082976867e-07,
      "loss": 0.675,
      "step": 3043
    },
    {
      "epoch": 0.30091688703259767,
      "grad_norm": 4.411477420919039,
      "learning_rate": 1.639169177427028e-07,
      "loss": 0.6924,
      "step": 3044
    },
    {
      "epoch": 0.30101574277735216,
      "grad_norm": 6.011067600226859,
      "learning_rate": 1.6389228810189982e-07,
      "loss": 0.6971,
      "step": 3045
    },
    {
      "epoch": 0.3011145985221066,
      "grad_norm": 46.06342986631917,
      "learning_rate": 1.6386765190988515e-07,
      "loss": 0.7748,
      "step": 3046
    },
    {
      "epoch": 0.3012134542668611,
      "grad_norm": 3.438218271289163,
      "learning_rate": 1.638430091691849e-07,
      "loss": 0.6768,
      "step": 3047
    },
    {
      "epoch": 0.30131231001161557,
      "grad_norm": 5.623255561522635,
      "learning_rate": 1.6381835988232582e-07,
      "loss": 0.6231,
      "step": 3048
    },
    {
      "epoch": 0.30141116575637,
      "grad_norm": 5.505622036068598,
      "learning_rate": 1.637937040518353e-07,
      "loss": 0.7659,
      "step": 3049
    },
    {
      "epoch": 0.3015100215011245,
      "grad_norm": 3.8456712975785012,
      "learning_rate": 1.6376904168024147e-07,
      "loss": 0.6824,
      "step": 3050
    },
    {
      "epoch": 0.3016088772458789,
      "grad_norm": 4.789170039776482,
      "learning_rate": 1.6374437277007308e-07,
      "loss": 0.7449,
      "step": 3051
    },
    {
      "epoch": 0.3017077329906334,
      "grad_norm": 3.634712860281757,
      "learning_rate": 1.6371969732385958e-07,
      "loss": 0.7803,
      "step": 3052
    },
    {
      "epoch": 0.3018065887353879,
      "grad_norm": 3.8623462783507336,
      "learning_rate": 1.6369501534413105e-07,
      "loss": 0.7092,
      "step": 3053
    },
    {
      "epoch": 0.30190544448014234,
      "grad_norm": 4.554782506176385,
      "learning_rate": 1.6367032683341822e-07,
      "loss": 0.7383,
      "step": 3054
    },
    {
      "epoch": 0.3020043002248968,
      "grad_norm": 4.317877561165182,
      "learning_rate": 1.636456317942526e-07,
      "loss": 0.6486,
      "step": 3055
    },
    {
      "epoch": 0.30210315596965126,
      "grad_norm": 5.7475177264748245,
      "learning_rate": 1.6362093022916626e-07,
      "loss": 0.6804,
      "step": 3056
    },
    {
      "epoch": 0.30220201171440575,
      "grad_norm": 4.181151380699082,
      "learning_rate": 1.6359622214069203e-07,
      "loss": 0.7304,
      "step": 3057
    },
    {
      "epoch": 0.30230086745916024,
      "grad_norm": 3.813265967235769,
      "learning_rate": 1.6357150753136332e-07,
      "loss": 0.7328,
      "step": 3058
    },
    {
      "epoch": 0.30239972320391467,
      "grad_norm": 2.900921414808093,
      "learning_rate": 1.6354678640371424e-07,
      "loss": 0.644,
      "step": 3059
    },
    {
      "epoch": 0.30249857894866916,
      "grad_norm": 11.882840495746054,
      "learning_rate": 1.635220587602796e-07,
      "loss": 0.6538,
      "step": 3060
    },
    {
      "epoch": 0.3025974346934236,
      "grad_norm": 3.812887930455186,
      "learning_rate": 1.6349732460359487e-07,
      "loss": 0.7588,
      "step": 3061
    },
    {
      "epoch": 0.3026962904381781,
      "grad_norm": 3.1695359462393453,
      "learning_rate": 1.6347258393619616e-07,
      "loss": 0.6601,
      "step": 3062
    },
    {
      "epoch": 0.3027951461829326,
      "grad_norm": 3.3679458251671837,
      "learning_rate": 1.6344783676062024e-07,
      "loss": 0.7461,
      "step": 3063
    },
    {
      "epoch": 0.302894001927687,
      "grad_norm": 3.4548765309086904,
      "learning_rate": 1.6342308307940455e-07,
      "loss": 0.706,
      "step": 3064
    },
    {
      "epoch": 0.3029928576724415,
      "grad_norm": 3.7249640849836463,
      "learning_rate": 1.633983228950873e-07,
      "loss": 0.6068,
      "step": 3065
    },
    {
      "epoch": 0.30309171341719593,
      "grad_norm": 3.545096209465632,
      "learning_rate": 1.633735562102072e-07,
      "loss": 0.8371,
      "step": 3066
    },
    {
      "epoch": 0.3031905691619504,
      "grad_norm": 4.059800417703488,
      "learning_rate": 1.6334878302730376e-07,
      "loss": 0.6302,
      "step": 3067
    },
    {
      "epoch": 0.3032894249067049,
      "grad_norm": 3.997277477922249,
      "learning_rate": 1.6332400334891704e-07,
      "loss": 0.705,
      "step": 3068
    },
    {
      "epoch": 0.30338828065145934,
      "grad_norm": 3.8249612035917617,
      "learning_rate": 1.632992171775879e-07,
      "loss": 0.6465,
      "step": 3069
    },
    {
      "epoch": 0.30348713639621383,
      "grad_norm": 8.228644771238056,
      "learning_rate": 1.6327442451585777e-07,
      "loss": 0.7486,
      "step": 3070
    },
    {
      "epoch": 0.3035859921409683,
      "grad_norm": 3.9682034763635508,
      "learning_rate": 1.6324962536626874e-07,
      "loss": 0.7942,
      "step": 3071
    },
    {
      "epoch": 0.30368484788572275,
      "grad_norm": 6.419239084606046,
      "learning_rate": 1.6322481973136365e-07,
      "loss": 0.7425,
      "step": 3072
    },
    {
      "epoch": 0.30378370363047724,
      "grad_norm": 3.218656513466064,
      "learning_rate": 1.632000076136859e-07,
      "loss": 0.7533,
      "step": 3073
    },
    {
      "epoch": 0.3038825593752317,
      "grad_norm": 16.17444489663314,
      "learning_rate": 1.6317518901577967e-07,
      "loss": 0.7352,
      "step": 3074
    },
    {
      "epoch": 0.30398141511998616,
      "grad_norm": 3.7105015264543235,
      "learning_rate": 1.631503639401897e-07,
      "loss": 0.7001,
      "step": 3075
    },
    {
      "epoch": 0.30408027086474065,
      "grad_norm": 3.0039992783569485,
      "learning_rate": 1.6312553238946143e-07,
      "loss": 0.7638,
      "step": 3076
    },
    {
      "epoch": 0.3041791266094951,
      "grad_norm": 3.686136821310322,
      "learning_rate": 1.63100694366141e-07,
      "loss": 0.6989,
      "step": 3077
    },
    {
      "epoch": 0.3042779823542496,
      "grad_norm": 7.693469946487675,
      "learning_rate": 1.6307584987277513e-07,
      "loss": 0.7557,
      "step": 3078
    },
    {
      "epoch": 0.304376838099004,
      "grad_norm": 3.383620416517909,
      "learning_rate": 1.6305099891191132e-07,
      "loss": 0.6734,
      "step": 3079
    },
    {
      "epoch": 0.3044756938437585,
      "grad_norm": 3.9735014332275944,
      "learning_rate": 1.6302614148609763e-07,
      "loss": 0.6798,
      "step": 3080
    },
    {
      "epoch": 0.304574549588513,
      "grad_norm": 4.7230406496389215,
      "learning_rate": 1.6300127759788286e-07,
      "loss": 0.6551,
      "step": 3081
    },
    {
      "epoch": 0.3046734053332674,
      "grad_norm": 3.0836621663492108,
      "learning_rate": 1.6297640724981634e-07,
      "loss": 0.7666,
      "step": 3082
    },
    {
      "epoch": 0.3047722610780219,
      "grad_norm": 2.8621892965253704,
      "learning_rate": 1.6295153044444825e-07,
      "loss": 0.7399,
      "step": 3083
    },
    {
      "epoch": 0.30487111682277634,
      "grad_norm": 9.902262882864472,
      "learning_rate": 1.6292664718432935e-07,
      "loss": 0.6928,
      "step": 3084
    },
    {
      "epoch": 0.30496997256753083,
      "grad_norm": 4.29046327545885,
      "learning_rate": 1.62901757472011e-07,
      "loss": 0.7393,
      "step": 3085
    },
    {
      "epoch": 0.3050688283122853,
      "grad_norm": 6.681251786043108,
      "learning_rate": 1.6287686131004525e-07,
      "loss": 0.7935,
      "step": 3086
    },
    {
      "epoch": 0.30516768405703976,
      "grad_norm": 43.60897253496351,
      "learning_rate": 1.628519587009849e-07,
      "loss": 0.6452,
      "step": 3087
    },
    {
      "epoch": 0.30526653980179425,
      "grad_norm": 4.655354265437068,
      "learning_rate": 1.628270496473833e-07,
      "loss": 0.7226,
      "step": 3088
    },
    {
      "epoch": 0.3053653955465487,
      "grad_norm": 6.792481195253978,
      "learning_rate": 1.628021341517946e-07,
      "loss": 0.7551,
      "step": 3089
    },
    {
      "epoch": 0.30546425129130317,
      "grad_norm": 6.477207765613272,
      "learning_rate": 1.6277721221677334e-07,
      "loss": 0.7089,
      "step": 3090
    },
    {
      "epoch": 0.30556310703605766,
      "grad_norm": 17.36732032557593,
      "learning_rate": 1.6275228384487502e-07,
      "loss": 0.6438,
      "step": 3091
    },
    {
      "epoch": 0.3056619627808121,
      "grad_norm": 3.3424784109244765,
      "learning_rate": 1.627273490386557e-07,
      "loss": 0.6121,
      "step": 3092
    },
    {
      "epoch": 0.3057608185255666,
      "grad_norm": 4.817878930465507,
      "learning_rate": 1.6270240780067197e-07,
      "loss": 0.7918,
      "step": 3093
    },
    {
      "epoch": 0.305859674270321,
      "grad_norm": 4.006770884183399,
      "learning_rate": 1.6267746013348127e-07,
      "loss": 0.6872,
      "step": 3094
    },
    {
      "epoch": 0.3059585300150755,
      "grad_norm": 11.105386747893618,
      "learning_rate": 1.6265250603964157e-07,
      "loss": 0.7505,
      "step": 3095
    },
    {
      "epoch": 0.30605738575983,
      "grad_norm": 3.883662354681814,
      "learning_rate": 1.626275455217116e-07,
      "loss": 0.7667,
      "step": 3096
    },
    {
      "epoch": 0.3061562415045844,
      "grad_norm": 3.322560467894458,
      "learning_rate": 1.6260257858225065e-07,
      "loss": 0.6633,
      "step": 3097
    },
    {
      "epoch": 0.3062550972493389,
      "grad_norm": 4.041459531337794,
      "learning_rate": 1.6257760522381873e-07,
      "loss": 0.6675,
      "step": 3098
    },
    {
      "epoch": 0.30635395299409335,
      "grad_norm": 4.081085214255873,
      "learning_rate": 1.6255262544897646e-07,
      "loss": 0.6213,
      "step": 3099
    },
    {
      "epoch": 0.30645280873884784,
      "grad_norm": 3.999896342127683,
      "learning_rate": 1.6252763926028518e-07,
      "loss": 0.659,
      "step": 3100
    },
    {
      "epoch": 0.3065516644836023,
      "grad_norm": 4.975146389722468,
      "learning_rate": 1.625026466603068e-07,
      "loss": 0.7932,
      "step": 3101
    },
    {
      "epoch": 0.30665052022835676,
      "grad_norm": 3.2615672270887925,
      "learning_rate": 1.6247764765160405e-07,
      "loss": 0.7203,
      "step": 3102
    },
    {
      "epoch": 0.30674937597311125,
      "grad_norm": 6.504723545341909,
      "learning_rate": 1.624526422367401e-07,
      "loss": 0.6436,
      "step": 3103
    },
    {
      "epoch": 0.3068482317178657,
      "grad_norm": 5.686391892395115,
      "learning_rate": 1.624276304182789e-07,
      "loss": 0.735,
      "step": 3104
    },
    {
      "epoch": 0.30694708746262017,
      "grad_norm": 77.54356413436516,
      "learning_rate": 1.6240261219878516e-07,
      "loss": 0.6867,
      "step": 3105
    },
    {
      "epoch": 0.30704594320737466,
      "grad_norm": 3.6204672124544923,
      "learning_rate": 1.6237758758082398e-07,
      "loss": 0.722,
      "step": 3106
    },
    {
      "epoch": 0.3071447989521291,
      "grad_norm": 4.405275220901767,
      "learning_rate": 1.6235255656696137e-07,
      "loss": 0.7698,
      "step": 3107
    },
    {
      "epoch": 0.3072436546968836,
      "grad_norm": 4.885301823578195,
      "learning_rate": 1.623275191597638e-07,
      "loss": 0.7335,
      "step": 3108
    },
    {
      "epoch": 0.307342510441638,
      "grad_norm": 4.046505686855976,
      "learning_rate": 1.6230247536179855e-07,
      "loss": 0.8159,
      "step": 3109
    },
    {
      "epoch": 0.3074413661863925,
      "grad_norm": 4.04904326947199,
      "learning_rate": 1.622774251756335e-07,
      "loss": 0.7644,
      "step": 3110
    },
    {
      "epoch": 0.307540221931147,
      "grad_norm": 3.722910512330455,
      "learning_rate": 1.6225236860383714e-07,
      "loss": 0.7631,
      "step": 3111
    },
    {
      "epoch": 0.30763907767590143,
      "grad_norm": 3.7322877934288488,
      "learning_rate": 1.6222730564897865e-07,
      "loss": 0.7862,
      "step": 3112
    },
    {
      "epoch": 0.3077379334206559,
      "grad_norm": 4.688069516978413,
      "learning_rate": 1.622022363136279e-07,
      "loss": 0.8245,
      "step": 3113
    },
    {
      "epoch": 0.30783678916541035,
      "grad_norm": 3.1744215706657553,
      "learning_rate": 1.621771606003554e-07,
      "loss": 0.667,
      "step": 3114
    },
    {
      "epoch": 0.30793564491016484,
      "grad_norm": 4.946242855402599,
      "learning_rate": 1.6215207851173218e-07,
      "loss": 0.6667,
      "step": 3115
    },
    {
      "epoch": 0.30803450065491933,
      "grad_norm": 4.019576444901938,
      "learning_rate": 1.6212699005033014e-07,
      "loss": 0.6986,
      "step": 3116
    },
    {
      "epoch": 0.30813335639967376,
      "grad_norm": 5.148462238495362,
      "learning_rate": 1.6210189521872174e-07,
      "loss": 0.7098,
      "step": 3117
    },
    {
      "epoch": 0.30823221214442825,
      "grad_norm": 6.276412413498243,
      "learning_rate": 1.6207679401948003e-07,
      "loss": 0.7543,
      "step": 3118
    },
    {
      "epoch": 0.3083310678891827,
      "grad_norm": 9.429717909344458,
      "learning_rate": 1.620516864551788e-07,
      "loss": 0.7389,
      "step": 3119
    },
    {
      "epoch": 0.3084299236339372,
      "grad_norm": 3.1539515089912125,
      "learning_rate": 1.6202657252839247e-07,
      "loss": 0.7858,
      "step": 3120
    },
    {
      "epoch": 0.30852877937869166,
      "grad_norm": 7.346566015854159,
      "learning_rate": 1.6200145224169605e-07,
      "loss": 0.7377,
      "step": 3121
    },
    {
      "epoch": 0.3086276351234461,
      "grad_norm": 51.557128284741296,
      "learning_rate": 1.6197632559766532e-07,
      "loss": 0.7272,
      "step": 3122
    },
    {
      "epoch": 0.3087264908682006,
      "grad_norm": 5.632638187754077,
      "learning_rate": 1.6195119259887657e-07,
      "loss": 0.7416,
      "step": 3123
    },
    {
      "epoch": 0.308825346612955,
      "grad_norm": 4.479740172583247,
      "learning_rate": 1.619260532479069e-07,
      "loss": 0.7225,
      "step": 3124
    },
    {
      "epoch": 0.3089242023577095,
      "grad_norm": 4.182762574714818,
      "learning_rate": 1.6190090754733395e-07,
      "loss": 0.8104,
      "step": 3125
    },
    {
      "epoch": 0.309023058102464,
      "grad_norm": 6.075811160977129,
      "learning_rate": 1.61875755499736e-07,
      "loss": 0.6996,
      "step": 3126
    },
    {
      "epoch": 0.30912191384721843,
      "grad_norm": 5.055769736571222,
      "learning_rate": 1.6185059710769207e-07,
      "loss": 0.8077,
      "step": 3127
    },
    {
      "epoch": 0.3092207695919729,
      "grad_norm": 6.81880936309117,
      "learning_rate": 1.6182543237378176e-07,
      "loss": 0.6308,
      "step": 3128
    },
    {
      "epoch": 0.30931962533672736,
      "grad_norm": 3.5561933807617674,
      "learning_rate": 1.6180026130058536e-07,
      "loss": 0.6644,
      "step": 3129
    },
    {
      "epoch": 0.30941848108148184,
      "grad_norm": 14.142626665939046,
      "learning_rate": 1.6177508389068377e-07,
      "loss": 0.7415,
      "step": 3130
    },
    {
      "epoch": 0.30951733682623633,
      "grad_norm": 2.8932617346199563,
      "learning_rate": 1.617499001466586e-07,
      "loss": 0.7474,
      "step": 3131
    },
    {
      "epoch": 0.30961619257099077,
      "grad_norm": 3.6265659075909618,
      "learning_rate": 1.6172471007109204e-07,
      "loss": 0.7566,
      "step": 3132
    },
    {
      "epoch": 0.30971504831574526,
      "grad_norm": 5.347121636340912,
      "learning_rate": 1.6169951366656695e-07,
      "loss": 0.7944,
      "step": 3133
    },
    {
      "epoch": 0.3098139040604997,
      "grad_norm": 7.489733646882159,
      "learning_rate": 1.6167431093566685e-07,
      "loss": 0.7771,
      "step": 3134
    },
    {
      "epoch": 0.3099127598052542,
      "grad_norm": 5.625544578481589,
      "learning_rate": 1.6164910188097596e-07,
      "loss": 0.725,
      "step": 3135
    },
    {
      "epoch": 0.31001161555000867,
      "grad_norm": 3.676137599360571,
      "learning_rate": 1.6162388650507905e-07,
      "loss": 0.668,
      "step": 3136
    },
    {
      "epoch": 0.3101104712947631,
      "grad_norm": 3.8171910189358997,
      "learning_rate": 1.6159866481056158e-07,
      "loss": 0.8045,
      "step": 3137
    },
    {
      "epoch": 0.3102093270395176,
      "grad_norm": 3.5032258381961574,
      "learning_rate": 1.6157343680000968e-07,
      "loss": 0.7261,
      "step": 3138
    },
    {
      "epoch": 0.310308182784272,
      "grad_norm": 9.32078367004573,
      "learning_rate": 1.615482024760101e-07,
      "loss": 0.6973,
      "step": 3139
    },
    {
      "epoch": 0.3104070385290265,
      "grad_norm": 2.8226795804959965,
      "learning_rate": 1.615229618411503e-07,
      "loss": 0.687,
      "step": 3140
    },
    {
      "epoch": 0.310505894273781,
      "grad_norm": 2.7367426174195026,
      "learning_rate": 1.6149771489801825e-07,
      "loss": 0.7427,
      "step": 3141
    },
    {
      "epoch": 0.31060475001853544,
      "grad_norm": 3.620513083135988,
      "learning_rate": 1.614724616492027e-07,
      "loss": 0.731,
      "step": 3142
    },
    {
      "epoch": 0.3107036057632899,
      "grad_norm": 4.577007027789433,
      "learning_rate": 1.61447202097293e-07,
      "loss": 0.7957,
      "step": 3143
    },
    {
      "epoch": 0.31080246150804436,
      "grad_norm": 7.64932389687859,
      "learning_rate": 1.6142193624487914e-07,
      "loss": 0.6617,
      "step": 3144
    },
    {
      "epoch": 0.31090131725279885,
      "grad_norm": 4.23692058370569,
      "learning_rate": 1.6139666409455174e-07,
      "loss": 0.6345,
      "step": 3145
    },
    {
      "epoch": 0.31100017299755334,
      "grad_norm": 3.7380716683853303,
      "learning_rate": 1.613713856489021e-07,
      "loss": 0.7873,
      "step": 3146
    },
    {
      "epoch": 0.31109902874230777,
      "grad_norm": 6.32370188144456,
      "learning_rate": 1.6134610091052216e-07,
      "loss": 0.686,
      "step": 3147
    },
    {
      "epoch": 0.31119788448706226,
      "grad_norm": 3.6400341999642416,
      "learning_rate": 1.6132080988200446e-07,
      "loss": 0.7636,
      "step": 3148
    },
    {
      "epoch": 0.31129674023181675,
      "grad_norm": 6.864143868059582,
      "learning_rate": 1.612955125659423e-07,
      "loss": 0.7901,
      "step": 3149
    },
    {
      "epoch": 0.3113955959765712,
      "grad_norm": 8.77562304276397,
      "learning_rate": 1.612702089649295e-07,
      "loss": 0.6709,
      "step": 3150
    },
    {
      "epoch": 0.31149445172132567,
      "grad_norm": 4.004634208967606,
      "learning_rate": 1.6124489908156055e-07,
      "loss": 0.7254,
      "step": 3151
    },
    {
      "epoch": 0.3115933074660801,
      "grad_norm": 8.349117394932485,
      "learning_rate": 1.6121958291843063e-07,
      "loss": 0.727,
      "step": 3152
    },
    {
      "epoch": 0.3116921632108346,
      "grad_norm": 5.683772574907308,
      "learning_rate": 1.611942604781355e-07,
      "loss": 0.7883,
      "step": 3153
    },
    {
      "epoch": 0.3117910189555891,
      "grad_norm": 5.062255380590402,
      "learning_rate": 1.6116893176327168e-07,
      "loss": 0.6857,
      "step": 3154
    },
    {
      "epoch": 0.3118898747003435,
      "grad_norm": 4.029030996424644,
      "learning_rate": 1.611435967764362e-07,
      "loss": 0.7107,
      "step": 3155
    },
    {
      "epoch": 0.311988730445098,
      "grad_norm": 5.071430499035809,
      "learning_rate": 1.611182555202268e-07,
      "loss": 0.6461,
      "step": 3156
    },
    {
      "epoch": 0.31208758618985244,
      "grad_norm": 3.9898575992657093,
      "learning_rate": 1.6109290799724186e-07,
      "loss": 0.704,
      "step": 3157
    },
    {
      "epoch": 0.31218644193460693,
      "grad_norm": 4.454628765161279,
      "learning_rate": 1.610675542100804e-07,
      "loss": 0.7211,
      "step": 3158
    },
    {
      "epoch": 0.3122852976793614,
      "grad_norm": 5.112426171889325,
      "learning_rate": 1.6104219416134202e-07,
      "loss": 0.5837,
      "step": 3159
    },
    {
      "epoch": 0.31238415342411585,
      "grad_norm": 3.2648024811619045,
      "learning_rate": 1.6101682785362712e-07,
      "loss": 0.6798,
      "step": 3160
    },
    {
      "epoch": 0.31248300916887034,
      "grad_norm": 3.0869118010867376,
      "learning_rate": 1.6099145528953656e-07,
      "loss": 0.653,
      "step": 3161
    },
    {
      "epoch": 0.3125818649136248,
      "grad_norm": 6.661095180353757,
      "learning_rate": 1.6096607647167196e-07,
      "loss": 0.6696,
      "step": 3162
    },
    {
      "epoch": 0.31268072065837926,
      "grad_norm": 4.197494382504308,
      "learning_rate": 1.609406914026355e-07,
      "loss": 0.6642,
      "step": 3163
    },
    {
      "epoch": 0.31277957640313375,
      "grad_norm": 3.5932311508629686,
      "learning_rate": 1.609153000850301e-07,
      "loss": 0.7463,
      "step": 3164
    },
    {
      "epoch": 0.3128784321478882,
      "grad_norm": 3.30962981900364,
      "learning_rate": 1.6088990252145923e-07,
      "loss": 0.8362,
      "step": 3165
    },
    {
      "epoch": 0.3129772878926427,
      "grad_norm": 10.861086236325606,
      "learning_rate": 1.6086449871452704e-07,
      "loss": 0.6519,
      "step": 3166
    },
    {
      "epoch": 0.3130761436373971,
      "grad_norm": 4.303069842395143,
      "learning_rate": 1.6083908866683834e-07,
      "loss": 0.6781,
      "step": 3167
    },
    {
      "epoch": 0.3131749993821516,
      "grad_norm": 5.554174316808517,
      "learning_rate": 1.6081367238099852e-07,
      "loss": 0.7323,
      "step": 3168
    },
    {
      "epoch": 0.3132738551269061,
      "grad_norm": 3.2988579472245783,
      "learning_rate": 1.607882498596137e-07,
      "loss": 0.8069,
      "step": 3169
    },
    {
      "epoch": 0.3133727108716605,
      "grad_norm": 3.640981945624258,
      "learning_rate": 1.6076282110529046e-07,
      "loss": 0.7426,
      "step": 3170
    },
    {
      "epoch": 0.313471566616415,
      "grad_norm": 19.283408134741215,
      "learning_rate": 1.607373861206363e-07,
      "loss": 0.6706,
      "step": 3171
    },
    {
      "epoch": 0.31357042236116944,
      "grad_norm": 3.967406664133984,
      "learning_rate": 1.6071194490825913e-07,
      "loss": 0.7237,
      "step": 3172
    },
    {
      "epoch": 0.31366927810592393,
      "grad_norm": 3.041059016278564,
      "learning_rate": 1.6068649747076756e-07,
      "loss": 0.7485,
      "step": 3173
    },
    {
      "epoch": 0.3137681338506784,
      "grad_norm": 3.975934224978301,
      "learning_rate": 1.6066104381077087e-07,
      "loss": 0.6835,
      "step": 3174
    },
    {
      "epoch": 0.31386698959543285,
      "grad_norm": 214.0147667857848,
      "learning_rate": 1.6063558393087892e-07,
      "loss": 0.7995,
      "step": 3175
    },
    {
      "epoch": 0.31396584534018734,
      "grad_norm": 3.7810348249133727,
      "learning_rate": 1.606101178337023e-07,
      "loss": 0.7785,
      "step": 3176
    },
    {
      "epoch": 0.3140647010849418,
      "grad_norm": 3.1995701341288103,
      "learning_rate": 1.6058464552185217e-07,
      "loss": 0.7558,
      "step": 3177
    },
    {
      "epoch": 0.31416355682969627,
      "grad_norm": 3.2938762659084846,
      "learning_rate": 1.6055916699794033e-07,
      "loss": 0.7111,
      "step": 3178
    },
    {
      "epoch": 0.31426241257445076,
      "grad_norm": 5.170885538008125,
      "learning_rate": 1.6053368226457922e-07,
      "loss": 0.7419,
      "step": 3179
    },
    {
      "epoch": 0.3143612683192052,
      "grad_norm": 3.4035493663473035,
      "learning_rate": 1.6050819132438194e-07,
      "loss": 0.64,
      "step": 3180
    },
    {
      "epoch": 0.3144601240639597,
      "grad_norm": 2.9431462433655846,
      "learning_rate": 1.6048269417996215e-07,
      "loss": 0.6461,
      "step": 3181
    },
    {
      "epoch": 0.3145589798087141,
      "grad_norm": 5.697668733984508,
      "learning_rate": 1.604571908339343e-07,
      "loss": 0.6602,
      "step": 3182
    },
    {
      "epoch": 0.3146578355534686,
      "grad_norm": 3.8022249590138246,
      "learning_rate": 1.604316812889133e-07,
      "loss": 0.7232,
      "step": 3183
    },
    {
      "epoch": 0.3147566912982231,
      "grad_norm": 3.2408982735915655,
      "learning_rate": 1.6040616554751485e-07,
      "loss": 0.7702,
      "step": 3184
    },
    {
      "epoch": 0.3148555470429775,
      "grad_norm": 3.7245093889576193,
      "learning_rate": 1.6038064361235515e-07,
      "loss": 0.6727,
      "step": 3185
    },
    {
      "epoch": 0.314954402787732,
      "grad_norm": 3.455059421803336,
      "learning_rate": 1.6035511548605115e-07,
      "loss": 0.7233,
      "step": 3186
    },
    {
      "epoch": 0.31505325853248645,
      "grad_norm": 4.12028273821115,
      "learning_rate": 1.6032958117122032e-07,
      "loss": 0.726,
      "step": 3187
    },
    {
      "epoch": 0.31515211427724094,
      "grad_norm": 2.839390399656588,
      "learning_rate": 1.6030404067048088e-07,
      "loss": 0.7496,
      "step": 3188
    },
    {
      "epoch": 0.3152509700219954,
      "grad_norm": 4.994014576264601,
      "learning_rate": 1.6027849398645165e-07,
      "loss": 0.7332,
      "step": 3189
    },
    {
      "epoch": 0.31534982576674986,
      "grad_norm": 5.308066043736463,
      "learning_rate": 1.60252941121752e-07,
      "loss": 0.643,
      "step": 3190
    },
    {
      "epoch": 0.31544868151150435,
      "grad_norm": 4.011855770291665,
      "learning_rate": 1.6022738207900205e-07,
      "loss": 0.7381,
      "step": 3191
    },
    {
      "epoch": 0.3155475372562588,
      "grad_norm": 3.87209508073509,
      "learning_rate": 1.6020181686082247e-07,
      "loss": 0.7361,
      "step": 3192
    },
    {
      "epoch": 0.31564639300101327,
      "grad_norm": 3.828731423154275,
      "learning_rate": 1.6017624546983462e-07,
      "loss": 0.7446,
      "step": 3193
    },
    {
      "epoch": 0.31574524874576776,
      "grad_norm": 4.913120092669052,
      "learning_rate": 1.6015066790866044e-07,
      "loss": 0.6323,
      "step": 3194
    },
    {
      "epoch": 0.3158441044905222,
      "grad_norm": 4.241669540612553,
      "learning_rate": 1.6012508417992256e-07,
      "loss": 0.6474,
      "step": 3195
    },
    {
      "epoch": 0.3159429602352767,
      "grad_norm": 4.249215223461972,
      "learning_rate": 1.600994942862442e-07,
      "loss": 0.8365,
      "step": 3196
    },
    {
      "epoch": 0.3160418159800311,
      "grad_norm": 8.692851959569614,
      "learning_rate": 1.6007389823024924e-07,
      "loss": 0.6972,
      "step": 3197
    },
    {
      "epoch": 0.3161406717247856,
      "grad_norm": 5.420239597274601,
      "learning_rate": 1.6004829601456218e-07,
      "loss": 0.7164,
      "step": 3198
    },
    {
      "epoch": 0.3162395274695401,
      "grad_norm": 5.987386444863469,
      "learning_rate": 1.6002268764180812e-07,
      "loss": 0.7295,
      "step": 3199
    },
    {
      "epoch": 0.3163383832142945,
      "grad_norm": 3.7468435316284725,
      "learning_rate": 1.5999707311461282e-07,
      "loss": 0.7183,
      "step": 3200
    },
    {
      "epoch": 0.316437238959049,
      "grad_norm": 3.6774283248548794,
      "learning_rate": 1.5997145243560276e-07,
      "loss": 0.7605,
      "step": 3201
    },
    {
      "epoch": 0.31653609470380345,
      "grad_norm": 10.208711565209999,
      "learning_rate": 1.5994582560740483e-07,
      "loss": 0.7468,
      "step": 3202
    },
    {
      "epoch": 0.31663495044855794,
      "grad_norm": 6.474217852897381,
      "learning_rate": 1.5992019263264676e-07,
      "loss": 0.7062,
      "step": 3203
    },
    {
      "epoch": 0.31673380619331243,
      "grad_norm": 4.438314121705886,
      "learning_rate": 1.5989455351395684e-07,
      "loss": 0.73,
      "step": 3204
    },
    {
      "epoch": 0.31683266193806686,
      "grad_norm": 4.278386419944885,
      "learning_rate": 1.5986890825396395e-07,
      "loss": 0.7483,
      "step": 3205
    },
    {
      "epoch": 0.31693151768282135,
      "grad_norm": 11.298225774965173,
      "learning_rate": 1.5984325685529765e-07,
      "loss": 0.6724,
      "step": 3206
    },
    {
      "epoch": 0.3170303734275758,
      "grad_norm": 6.541117063538867,
      "learning_rate": 1.598175993205881e-07,
      "loss": 0.6766,
      "step": 3207
    },
    {
      "epoch": 0.3171292291723303,
      "grad_norm": 4.680971859262622,
      "learning_rate": 1.5979193565246613e-07,
      "loss": 0.7143,
      "step": 3208
    },
    {
      "epoch": 0.31722808491708476,
      "grad_norm": 5.675154195106667,
      "learning_rate": 1.5976626585356313e-07,
      "loss": 0.7018,
      "step": 3209
    },
    {
      "epoch": 0.3173269406618392,
      "grad_norm": 3.250745011627665,
      "learning_rate": 1.597405899265112e-07,
      "loss": 0.7742,
      "step": 3210
    },
    {
      "epoch": 0.3174257964065937,
      "grad_norm": 5.294920687347966,
      "learning_rate": 1.59714907873943e-07,
      "loss": 0.7247,
      "step": 3211
    },
    {
      "epoch": 0.3175246521513481,
      "grad_norm": 4.691377611805262,
      "learning_rate": 1.5968921969849187e-07,
      "loss": 0.6947,
      "step": 3212
    },
    {
      "epoch": 0.3176235078961026,
      "grad_norm": 6.137665697664341,
      "learning_rate": 1.5966352540279174e-07,
      "loss": 0.7765,
      "step": 3213
    },
    {
      "epoch": 0.3177223636408571,
      "grad_norm": 10.424550195467136,
      "learning_rate": 1.5963782498947714e-07,
      "loss": 0.7624,
      "step": 3214
    },
    {
      "epoch": 0.31782121938561153,
      "grad_norm": 5.743434311174949,
      "learning_rate": 1.5961211846118333e-07,
      "loss": 0.6861,
      "step": 3215
    },
    {
      "epoch": 0.317920075130366,
      "grad_norm": 5.264398509385489,
      "learning_rate": 1.5958640582054612e-07,
      "loss": 0.7471,
      "step": 3216
    },
    {
      "epoch": 0.31801893087512045,
      "grad_norm": 3.2766981444127965,
      "learning_rate": 1.5956068707020194e-07,
      "loss": 0.6773,
      "step": 3217
    },
    {
      "epoch": 0.31811778661987494,
      "grad_norm": 17.080313206362586,
      "learning_rate": 1.5953496221278788e-07,
      "loss": 0.7384,
      "step": 3218
    },
    {
      "epoch": 0.31821664236462943,
      "grad_norm": 3.0119809076017527,
      "learning_rate": 1.5950923125094166e-07,
      "loss": 0.6913,
      "step": 3219
    },
    {
      "epoch": 0.31831549810938387,
      "grad_norm": 10.366661359735955,
      "learning_rate": 1.594834941873016e-07,
      "loss": 0.7666,
      "step": 3220
    },
    {
      "epoch": 0.31841435385413835,
      "grad_norm": 9.04798457571723,
      "learning_rate": 1.5945775102450664e-07,
      "loss": 0.6769,
      "step": 3221
    },
    {
      "epoch": 0.3185132095988928,
      "grad_norm": 3.5476084306984017,
      "learning_rate": 1.594320017651964e-07,
      "loss": 0.7274,
      "step": 3222
    },
    {
      "epoch": 0.3186120653436473,
      "grad_norm": 8.263748107072734,
      "learning_rate": 1.5940624641201102e-07,
      "loss": 0.7301,
      "step": 3223
    },
    {
      "epoch": 0.31871092108840177,
      "grad_norm": 3.8743575849753946,
      "learning_rate": 1.5938048496759144e-07,
      "loss": 0.7405,
      "step": 3224
    },
    {
      "epoch": 0.3188097768331562,
      "grad_norm": 4.351028533895859,
      "learning_rate": 1.59354717434579e-07,
      "loss": 0.7204,
      "step": 3225
    },
    {
      "epoch": 0.3189086325779107,
      "grad_norm": 3.428073248116463,
      "learning_rate": 1.5932894381561582e-07,
      "loss": 0.6701,
      "step": 3226
    },
    {
      "epoch": 0.3190074883226652,
      "grad_norm": 18.939481114517278,
      "learning_rate": 1.5930316411334465e-07,
      "loss": 0.7979,
      "step": 3227
    },
    {
      "epoch": 0.3191063440674196,
      "grad_norm": 9.179237422879316,
      "learning_rate": 1.5927737833040878e-07,
      "loss": 0.5994,
      "step": 3228
    },
    {
      "epoch": 0.3192051998121741,
      "grad_norm": 4.430295974747974,
      "learning_rate": 1.5925158646945216e-07,
      "loss": 0.7063,
      "step": 3229
    },
    {
      "epoch": 0.31930405555692853,
      "grad_norm": 8.350042120720248,
      "learning_rate": 1.5922578853311936e-07,
      "loss": 0.7313,
      "step": 3230
    },
    {
      "epoch": 0.319402911301683,
      "grad_norm": 4.084326850136773,
      "learning_rate": 1.591999845240556e-07,
      "loss": 0.7653,
      "step": 3231
    },
    {
      "epoch": 0.3195017670464375,
      "grad_norm": 5.1566576173428125,
      "learning_rate": 1.591741744449067e-07,
      "loss": 0.6125,
      "step": 3232
    },
    {
      "epoch": 0.31960062279119195,
      "grad_norm": 9.298098450828299,
      "learning_rate": 1.5914835829831907e-07,
      "loss": 0.7074,
      "step": 3233
    },
    {
      "epoch": 0.31969947853594644,
      "grad_norm": 3.4232915237369417,
      "learning_rate": 1.591225360869398e-07,
      "loss": 0.6838,
      "step": 3234
    },
    {
      "epoch": 0.31979833428070087,
      "grad_norm": 4.075727554046275,
      "learning_rate": 1.5909670781341657e-07,
      "loss": 0.6962,
      "step": 3235
    },
    {
      "epoch": 0.31989719002545536,
      "grad_norm": 13.650805214901233,
      "learning_rate": 1.590708734803977e-07,
      "loss": 0.779,
      "step": 3236
    },
    {
      "epoch": 0.31999604577020985,
      "grad_norm": 2.992145289536845,
      "learning_rate": 1.590450330905321e-07,
      "loss": 0.7561,
      "step": 3237
    },
    {
      "epoch": 0.3200949015149643,
      "grad_norm": 7.815708447200665,
      "learning_rate": 1.5901918664646935e-07,
      "loss": 0.6734,
      "step": 3238
    },
    {
      "epoch": 0.32019375725971877,
      "grad_norm": 3.693271731527079,
      "learning_rate": 1.5899333415085958e-07,
      "loss": 0.706,
      "step": 3239
    },
    {
      "epoch": 0.3202926130044732,
      "grad_norm": 45.133257041910824,
      "learning_rate": 1.5896747560635362e-07,
      "loss": 0.7468,
      "step": 3240
    },
    {
      "epoch": 0.3203914687492277,
      "grad_norm": 5.853807340624508,
      "learning_rate": 1.5894161101560286e-07,
      "loss": 0.6782,
      "step": 3241
    },
    {
      "epoch": 0.3204903244939822,
      "grad_norm": 7.918295518645881,
      "learning_rate": 1.5891574038125938e-07,
      "loss": 0.7622,
      "step": 3242
    },
    {
      "epoch": 0.3205891802387366,
      "grad_norm": 4.527723605429583,
      "learning_rate": 1.5888986370597572e-07,
      "loss": 0.7506,
      "step": 3243
    },
    {
      "epoch": 0.3206880359834911,
      "grad_norm": 34.794240690429476,
      "learning_rate": 1.588639809924053e-07,
      "loss": 0.748,
      "step": 3244
    },
    {
      "epoch": 0.32078689172824554,
      "grad_norm": 3.9132682569057837,
      "learning_rate": 1.5883809224320187e-07,
      "loss": 0.6817,
      "step": 3245
    },
    {
      "epoch": 0.320885747473,
      "grad_norm": 2.9667834096065255,
      "learning_rate": 1.5881219746102003e-07,
      "loss": 0.7089,
      "step": 3246
    },
    {
      "epoch": 0.3209846032177545,
      "grad_norm": 4.501879590115043,
      "learning_rate": 1.5878629664851486e-07,
      "loss": 0.7265,
      "step": 3247
    },
    {
      "epoch": 0.32108345896250895,
      "grad_norm": 3.9920481731631745,
      "learning_rate": 1.5876038980834216e-07,
      "loss": 0.7419,
      "step": 3248
    },
    {
      "epoch": 0.32118231470726344,
      "grad_norm": 3.371668959683131,
      "learning_rate": 1.5873447694315827e-07,
      "loss": 0.6052,
      "step": 3249
    },
    {
      "epoch": 0.3212811704520179,
      "grad_norm": 4.948979880715994,
      "learning_rate": 1.5870855805562013e-07,
      "loss": 0.6153,
      "step": 3250
    },
    {
      "epoch": 0.32138002619677236,
      "grad_norm": 3.413678509643568,
      "learning_rate": 1.5868263314838539e-07,
      "loss": 0.722,
      "step": 3251
    },
    {
      "epoch": 0.32147888194152685,
      "grad_norm": 6.6200596374525995,
      "learning_rate": 1.5865670222411226e-07,
      "loss": 0.7094,
      "step": 3252
    },
    {
      "epoch": 0.3215777376862813,
      "grad_norm": 4.92235774706373,
      "learning_rate": 1.5863076528545953e-07,
      "loss": 0.7893,
      "step": 3253
    },
    {
      "epoch": 0.3216765934310358,
      "grad_norm": 6.77376879340371,
      "learning_rate": 1.586048223350867e-07,
      "loss": 0.6305,
      "step": 3254
    },
    {
      "epoch": 0.3217754491757902,
      "grad_norm": 3.844167344818066,
      "learning_rate": 1.5857887337565387e-07,
      "loss": 0.782,
      "step": 3255
    },
    {
      "epoch": 0.3218743049205447,
      "grad_norm": 3.339732617141909,
      "learning_rate": 1.5855291840982162e-07,
      "loss": 0.7342,
      "step": 3256
    },
    {
      "epoch": 0.3219731606652992,
      "grad_norm": 4.0052517159322605,
      "learning_rate": 1.5852695744025133e-07,
      "loss": 0.6825,
      "step": 3257
    },
    {
      "epoch": 0.3220720164100536,
      "grad_norm": 3.8381130444890617,
      "learning_rate": 1.5850099046960485e-07,
      "loss": 0.6352,
      "step": 3258
    },
    {
      "epoch": 0.3221708721548081,
      "grad_norm": 4.678706163791063,
      "learning_rate": 1.584750175005448e-07,
      "loss": 0.7168,
      "step": 3259
    },
    {
      "epoch": 0.32226972789956254,
      "grad_norm": 4.211278075144235,
      "learning_rate": 1.584490385357342e-07,
      "loss": 0.7934,
      "step": 3260
    },
    {
      "epoch": 0.32236858364431703,
      "grad_norm": 5.609145869275169,
      "learning_rate": 1.5842305357783695e-07,
      "loss": 0.8129,
      "step": 3261
    },
    {
      "epoch": 0.3224674393890715,
      "grad_norm": 6.813974164403523,
      "learning_rate": 1.5839706262951732e-07,
      "loss": 0.7061,
      "step": 3262
    },
    {
      "epoch": 0.32256629513382595,
      "grad_norm": 10.777347512710701,
      "learning_rate": 1.5837106569344036e-07,
      "loss": 0.6095,
      "step": 3263
    },
    {
      "epoch": 0.32266515087858044,
      "grad_norm": 3.73424080498466,
      "learning_rate": 1.5834506277227164e-07,
      "loss": 0.8041,
      "step": 3264
    },
    {
      "epoch": 0.3227640066233349,
      "grad_norm": 5.383686860916343,
      "learning_rate": 1.5831905386867738e-07,
      "loss": 0.6672,
      "step": 3265
    },
    {
      "epoch": 0.32286286236808937,
      "grad_norm": 12.684281288637656,
      "learning_rate": 1.582930389853244e-07,
      "loss": 0.7279,
      "step": 3266
    },
    {
      "epoch": 0.32296171811284385,
      "grad_norm": 4.164795099679318,
      "learning_rate": 1.5826701812488014e-07,
      "loss": 0.7224,
      "step": 3267
    },
    {
      "epoch": 0.3230605738575983,
      "grad_norm": 4.923132750305054,
      "learning_rate": 1.5824099129001271e-07,
      "loss": 0.7151,
      "step": 3268
    },
    {
      "epoch": 0.3231594296023528,
      "grad_norm": 8.919002025372901,
      "learning_rate": 1.5821495848339072e-07,
      "loss": 0.6535,
      "step": 3269
    },
    {
      "epoch": 0.3232582853471072,
      "grad_norm": 3.408671584960044,
      "learning_rate": 1.581889197076835e-07,
      "loss": 0.6788,
      "step": 3270
    },
    {
      "epoch": 0.3233571410918617,
      "grad_norm": 3.41354058805903,
      "learning_rate": 1.5816287496556087e-07,
      "loss": 0.6361,
      "step": 3271
    },
    {
      "epoch": 0.3234559968366162,
      "grad_norm": 3.6212506895110166,
      "learning_rate": 1.581368242596934e-07,
      "loss": 0.7694,
      "step": 3272
    },
    {
      "epoch": 0.3235548525813706,
      "grad_norm": 5.401243850504713,
      "learning_rate": 1.5811076759275215e-07,
      "loss": 0.7908,
      "step": 3273
    },
    {
      "epoch": 0.3236537083261251,
      "grad_norm": 2.935439136367023,
      "learning_rate": 1.5808470496740892e-07,
      "loss": 0.6568,
      "step": 3274
    },
    {
      "epoch": 0.32375256407087954,
      "grad_norm": 3.2935266094445583,
      "learning_rate": 1.5805863638633596e-07,
      "loss": 0.729,
      "step": 3275
    },
    {
      "epoch": 0.32385141981563403,
      "grad_norm": 4.276882910631688,
      "learning_rate": 1.5803256185220627e-07,
      "loss": 0.6386,
      "step": 3276
    },
    {
      "epoch": 0.3239502755603885,
      "grad_norm": 10.732901115070991,
      "learning_rate": 1.5800648136769345e-07,
      "loss": 0.7099,
      "step": 3277
    },
    {
      "epoch": 0.32404913130514296,
      "grad_norm": 4.786979018600005,
      "learning_rate": 1.5798039493547157e-07,
      "loss": 0.7182,
      "step": 3278
    },
    {
      "epoch": 0.32414798704989745,
      "grad_norm": 3.5086606362182544,
      "learning_rate": 1.579543025582155e-07,
      "loss": 0.7293,
      "step": 3279
    },
    {
      "epoch": 0.3242468427946519,
      "grad_norm": 4.193260455146662,
      "learning_rate": 1.5792820423860053e-07,
      "loss": 0.7239,
      "step": 3280
    },
    {
      "epoch": 0.32434569853940637,
      "grad_norm": 3.5910158506782506,
      "learning_rate": 1.5790209997930278e-07,
      "loss": 0.6155,
      "step": 3281
    },
    {
      "epoch": 0.32444455428416086,
      "grad_norm": 3.664837721649474,
      "learning_rate": 1.5787598978299878e-07,
      "loss": 0.7887,
      "step": 3282
    },
    {
      "epoch": 0.3245434100289153,
      "grad_norm": 3.29893564642089,
      "learning_rate": 1.5784987365236573e-07,
      "loss": 0.6183,
      "step": 3283
    },
    {
      "epoch": 0.3246422657736698,
      "grad_norm": 6.707336026665019,
      "learning_rate": 1.578237515900815e-07,
      "loss": 0.7238,
      "step": 3284
    },
    {
      "epoch": 0.3247411215184242,
      "grad_norm": 3.2866442687654933,
      "learning_rate": 1.577976235988245e-07,
      "loss": 0.7048,
      "step": 3285
    },
    {
      "epoch": 0.3248399772631787,
      "grad_norm": 6.158798360283989,
      "learning_rate": 1.577714896812738e-07,
      "loss": 0.6621,
      "step": 3286
    },
    {
      "epoch": 0.3249388330079332,
      "grad_norm": 5.5317353590887555,
      "learning_rate": 1.57745349840109e-07,
      "loss": 0.6863,
      "step": 3287
    },
    {
      "epoch": 0.3250376887526876,
      "grad_norm": 3.354995744646344,
      "learning_rate": 1.5771920407801037e-07,
      "loss": 0.7176,
      "step": 3288
    },
    {
      "epoch": 0.3251365444974421,
      "grad_norm": 16.49388902397358,
      "learning_rate": 1.5769305239765877e-07,
      "loss": 0.7032,
      "step": 3289
    },
    {
      "epoch": 0.32523540024219655,
      "grad_norm": 5.388823537841306,
      "learning_rate": 1.576668948017357e-07,
      "loss": 0.7279,
      "step": 3290
    },
    {
      "epoch": 0.32533425598695104,
      "grad_norm": 11.610477465415848,
      "learning_rate": 1.5764073129292324e-07,
      "loss": 0.6574,
      "step": 3291
    },
    {
      "epoch": 0.3254331117317055,
      "grad_norm": 3.152045175576492,
      "learning_rate": 1.57614561873904e-07,
      "loss": 0.6332,
      "step": 3292
    },
    {
      "epoch": 0.32553196747645996,
      "grad_norm": 3.8582105554225543,
      "learning_rate": 1.5758838654736134e-07,
      "loss": 0.7964,
      "step": 3293
    },
    {
      "epoch": 0.32563082322121445,
      "grad_norm": 3.75121100409366,
      "learning_rate": 1.5756220531597912e-07,
      "loss": 0.7699,
      "step": 3294
    },
    {
      "epoch": 0.3257296789659689,
      "grad_norm": 16.809456365538537,
      "learning_rate": 1.5753601818244185e-07,
      "loss": 0.6738,
      "step": 3295
    },
    {
      "epoch": 0.32582853471072337,
      "grad_norm": 3.645619644794589,
      "learning_rate": 1.5750982514943462e-07,
      "loss": 0.7102,
      "step": 3296
    },
    {
      "epoch": 0.32592739045547786,
      "grad_norm": 3.869170248644768,
      "learning_rate": 1.5748362621964316e-07,
      "loss": 0.7707,
      "step": 3297
    },
    {
      "epoch": 0.3260262462002323,
      "grad_norm": 8.414091394110764,
      "learning_rate": 1.574574213957538e-07,
      "loss": 0.6492,
      "step": 3298
    },
    {
      "epoch": 0.3261251019449868,
      "grad_norm": 10.214319721148106,
      "learning_rate": 1.574312106804534e-07,
      "loss": 0.6722,
      "step": 3299
    },
    {
      "epoch": 0.3262239576897412,
      "grad_norm": 3.2351268944392486,
      "learning_rate": 1.5740499407642953e-07,
      "loss": 0.8297,
      "step": 3300
    },
    {
      "epoch": 0.3263228134344957,
      "grad_norm": 3.904766714083447,
      "learning_rate": 1.5737877158637027e-07,
      "loss": 0.6326,
      "step": 3301
    },
    {
      "epoch": 0.3264216691792502,
      "grad_norm": 3.563684441492177,
      "learning_rate": 1.5735254321296441e-07,
      "loss": 0.7174,
      "step": 3302
    },
    {
      "epoch": 0.32652052492400463,
      "grad_norm": 16.596705530665872,
      "learning_rate": 1.5732630895890124e-07,
      "loss": 0.7618,
      "step": 3303
    },
    {
      "epoch": 0.3266193806687591,
      "grad_norm": 3.1715554333618186,
      "learning_rate": 1.5730006882687073e-07,
      "loss": 0.7184,
      "step": 3304
    },
    {
      "epoch": 0.3267182364135136,
      "grad_norm": 19.282022588893298,
      "learning_rate": 1.5727382281956338e-07,
      "loss": 0.7401,
      "step": 3305
    },
    {
      "epoch": 0.32681709215826804,
      "grad_norm": 3.0952677153969073,
      "learning_rate": 1.5724757093967033e-07,
      "loss": 0.6612,
      "step": 3306
    },
    {
      "epoch": 0.32691594790302253,
      "grad_norm": 4.269383011691742,
      "learning_rate": 1.5722131318988337e-07,
      "loss": 0.6776,
      "step": 3307
    },
    {
      "epoch": 0.32701480364777696,
      "grad_norm": 6.140797151154471,
      "learning_rate": 1.5719504957289483e-07,
      "loss": 0.7865,
      "step": 3308
    },
    {
      "epoch": 0.32711365939253145,
      "grad_norm": 3.4068476244268378,
      "learning_rate": 1.571687800913976e-07,
      "loss": 0.7308,
      "step": 3309
    },
    {
      "epoch": 0.32721251513728594,
      "grad_norm": 3.4679012689203925,
      "learning_rate": 1.571425047480853e-07,
      "loss": 0.6744,
      "step": 3310
    },
    {
      "epoch": 0.3273113708820404,
      "grad_norm": 4.014563913942624,
      "learning_rate": 1.5711622354565204e-07,
      "loss": 0.6831,
      "step": 3311
    },
    {
      "epoch": 0.32741022662679486,
      "grad_norm": 4.379640409102196,
      "learning_rate": 1.5708993648679258e-07,
      "loss": 0.7193,
      "step": 3312
    },
    {
      "epoch": 0.3275090823715493,
      "grad_norm": 4.797074808143279,
      "learning_rate": 1.5706364357420227e-07,
      "loss": 0.667,
      "step": 3313
    },
    {
      "epoch": 0.3276079381163038,
      "grad_norm": 4.311187677523124,
      "learning_rate": 1.5703734481057706e-07,
      "loss": 0.759,
      "step": 3314
    },
    {
      "epoch": 0.3277067938610583,
      "grad_norm": 3.921918732401354,
      "learning_rate": 1.5701104019861352e-07,
      "loss": 0.7953,
      "step": 3315
    },
    {
      "epoch": 0.3278056496058127,
      "grad_norm": 3.704703585499704,
      "learning_rate": 1.5698472974100874e-07,
      "loss": 0.6367,
      "step": 3316
    },
    {
      "epoch": 0.3279045053505672,
      "grad_norm": 4.6074218029380045,
      "learning_rate": 1.5695841344046053e-07,
      "loss": 0.6563,
      "step": 3317
    },
    {
      "epoch": 0.32800336109532163,
      "grad_norm": 4.11326626040705,
      "learning_rate": 1.569320912996672e-07,
      "loss": 0.8535,
      "step": 3318
    },
    {
      "epoch": 0.3281022168400761,
      "grad_norm": 5.848390995911136,
      "learning_rate": 1.5690576332132775e-07,
      "loss": 0.6387,
      "step": 3319
    },
    {
      "epoch": 0.3282010725848306,
      "grad_norm": 9.145338698934648,
      "learning_rate": 1.5687942950814166e-07,
      "loss": 0.7215,
      "step": 3320
    },
    {
      "epoch": 0.32829992832958504,
      "grad_norm": 6.218805856468651,
      "learning_rate": 1.568530898628091e-07,
      "loss": 0.6596,
      "step": 3321
    },
    {
      "epoch": 0.32839878407433953,
      "grad_norm": 3.369089538713358,
      "learning_rate": 1.5682674438803085e-07,
      "loss": 0.7627,
      "step": 3322
    },
    {
      "epoch": 0.32849763981909397,
      "grad_norm": 4.632724212475782,
      "learning_rate": 1.568003930865082e-07,
      "loss": 0.7297,
      "step": 3323
    },
    {
      "epoch": 0.32859649556384846,
      "grad_norm": 3.6062146484630855,
      "learning_rate": 1.5677403596094307e-07,
      "loss": 0.781,
      "step": 3324
    },
    {
      "epoch": 0.32869535130860295,
      "grad_norm": 4.9251460654501935,
      "learning_rate": 1.5674767301403806e-07,
      "loss": 0.651,
      "step": 3325
    },
    {
      "epoch": 0.3287942070533574,
      "grad_norm": 5.022244771229125,
      "learning_rate": 1.567213042484963e-07,
      "loss": 0.6499,
      "step": 3326
    },
    {
      "epoch": 0.32889306279811187,
      "grad_norm": 14.560985645520223,
      "learning_rate": 1.5669492966702145e-07,
      "loss": 0.7457,
      "step": 3327
    },
    {
      "epoch": 0.3289919185428663,
      "grad_norm": 4.025316919491248,
      "learning_rate": 1.5666854927231788e-07,
      "loss": 0.7219,
      "step": 3328
    },
    {
      "epoch": 0.3290907742876208,
      "grad_norm": 16.4329466844937,
      "learning_rate": 1.5664216306709048e-07,
      "loss": 0.6901,
      "step": 3329
    },
    {
      "epoch": 0.3291896300323753,
      "grad_norm": 3.508676997279705,
      "learning_rate": 1.5661577105404483e-07,
      "loss": 0.7182,
      "step": 3330
    },
    {
      "epoch": 0.3292884857771297,
      "grad_norm": 4.111932589610013,
      "learning_rate": 1.5658937323588695e-07,
      "loss": 0.6978,
      "step": 3331
    },
    {
      "epoch": 0.3293873415218842,
      "grad_norm": 3.830670009301702,
      "learning_rate": 1.5656296961532363e-07,
      "loss": 0.7398,
      "step": 3332
    },
    {
      "epoch": 0.32948619726663864,
      "grad_norm": 7.116548265214187,
      "learning_rate": 1.5653656019506214e-07,
      "loss": 0.7183,
      "step": 3333
    },
    {
      "epoch": 0.3295850530113931,
      "grad_norm": 2.5596687783418974,
      "learning_rate": 1.5651014497781037e-07,
      "loss": 0.6528,
      "step": 3334
    },
    {
      "epoch": 0.3296839087561476,
      "grad_norm": 17.294416088881754,
      "learning_rate": 1.564837239662768e-07,
      "loss": 0.6425,
      "step": 3335
    },
    {
      "epoch": 0.32978276450090205,
      "grad_norm": 5.220942969197681,
      "learning_rate": 1.5645729716317052e-07,
      "loss": 0.7277,
      "step": 3336
    },
    {
      "epoch": 0.32988162024565654,
      "grad_norm": 5.24058718233169,
      "learning_rate": 1.5643086457120126e-07,
      "loss": 0.673,
      "step": 3337
    },
    {
      "epoch": 0.32998047599041097,
      "grad_norm": 3.6452757121169452,
      "learning_rate": 1.5640442619307924e-07,
      "loss": 0.7989,
      "step": 3338
    },
    {
      "epoch": 0.33007933173516546,
      "grad_norm": 3.955271879529515,
      "learning_rate": 1.5637798203151533e-07,
      "loss": 0.7437,
      "step": 3339
    },
    {
      "epoch": 0.33017818747991995,
      "grad_norm": 5.512147896226781,
      "learning_rate": 1.5635153208922102e-07,
      "loss": 0.7089,
      "step": 3340
    },
    {
      "epoch": 0.3302770432246744,
      "grad_norm": 4.413564604184577,
      "learning_rate": 1.5632507636890832e-07,
      "loss": 0.7037,
      "step": 3341
    },
    {
      "epoch": 0.33037589896942887,
      "grad_norm": 4.401560364734806,
      "learning_rate": 1.5629861487328987e-07,
      "loss": 0.6526,
      "step": 3342
    },
    {
      "epoch": 0.3304747547141833,
      "grad_norm": 7.835097899917708,
      "learning_rate": 1.5627214760507895e-07,
      "loss": 0.7652,
      "step": 3343
    },
    {
      "epoch": 0.3305736104589378,
      "grad_norm": 2.881351065295805,
      "learning_rate": 1.562456745669894e-07,
      "loss": 0.7648,
      "step": 3344
    },
    {
      "epoch": 0.3306724662036923,
      "grad_norm": 5.084410762411646,
      "learning_rate": 1.5621919576173554e-07,
      "loss": 0.7608,
      "step": 3345
    },
    {
      "epoch": 0.3307713219484467,
      "grad_norm": 4.727347757804229,
      "learning_rate": 1.5619271119203247e-07,
      "loss": 0.5968,
      "step": 3346
    },
    {
      "epoch": 0.3308701776932012,
      "grad_norm": 5.320998479591818,
      "learning_rate": 1.561662208605958e-07,
      "loss": 0.6748,
      "step": 3347
    },
    {
      "epoch": 0.33096903343795564,
      "grad_norm": 3.9777929377082866,
      "learning_rate": 1.5613972477014165e-07,
      "loss": 0.6999,
      "step": 3348
    },
    {
      "epoch": 0.33106788918271013,
      "grad_norm": 3.734666444773565,
      "learning_rate": 1.5611322292338685e-07,
      "loss": 0.7181,
      "step": 3349
    },
    {
      "epoch": 0.3311667449274646,
      "grad_norm": 5.725695504267395,
      "learning_rate": 1.5608671532304877e-07,
      "loss": 0.6503,
      "step": 3350
    },
    {
      "epoch": 0.33126560067221905,
      "grad_norm": 4.495864913159523,
      "learning_rate": 1.5606020197184537e-07,
      "loss": 0.7584,
      "step": 3351
    },
    {
      "epoch": 0.33136445641697354,
      "grad_norm": 5.249674158532663,
      "learning_rate": 1.560336828724952e-07,
      "loss": 0.6859,
      "step": 3352
    },
    {
      "epoch": 0.331463312161728,
      "grad_norm": 3.4608185232923034,
      "learning_rate": 1.5600715802771743e-07,
      "loss": 0.7523,
      "step": 3353
    },
    {
      "epoch": 0.33156216790648246,
      "grad_norm": 3.0839283700561286,
      "learning_rate": 1.5598062744023174e-07,
      "loss": 0.7432,
      "step": 3354
    },
    {
      "epoch": 0.33166102365123695,
      "grad_norm": 3.2511545819493275,
      "learning_rate": 1.559540911127585e-07,
      "loss": 0.6602,
      "step": 3355
    },
    {
      "epoch": 0.3317598793959914,
      "grad_norm": 7.362585979341072,
      "learning_rate": 1.5592754904801859e-07,
      "loss": 0.7257,
      "step": 3356
    },
    {
      "epoch": 0.3318587351407459,
      "grad_norm": 3.9068456498679915,
      "learning_rate": 1.5590100124873352e-07,
      "loss": 0.7225,
      "step": 3357
    },
    {
      "epoch": 0.3319575908855003,
      "grad_norm": 3.008765401939737,
      "learning_rate": 1.5587444771762533e-07,
      "loss": 0.7161,
      "step": 3358
    },
    {
      "epoch": 0.3320564466302548,
      "grad_norm": 5.055740983924254,
      "learning_rate": 1.5584788845741682e-07,
      "loss": 0.708,
      "step": 3359
    },
    {
      "epoch": 0.3321553023750093,
      "grad_norm": 4.177986259959426,
      "learning_rate": 1.5582132347083108e-07,
      "loss": 0.8116,
      "step": 3360
    },
    {
      "epoch": 0.3322541581197637,
      "grad_norm": 3.470784503939456,
      "learning_rate": 1.5579475276059207e-07,
      "loss": 0.6615,
      "step": 3361
    },
    {
      "epoch": 0.3323530138645182,
      "grad_norm": 3.7690855699685253,
      "learning_rate": 1.5576817632942421e-07,
      "loss": 0.6938,
      "step": 3362
    },
    {
      "epoch": 0.33245186960927264,
      "grad_norm": 2.92040859835785,
      "learning_rate": 1.5574159418005247e-07,
      "loss": 0.7825,
      "step": 3363
    },
    {
      "epoch": 0.33255072535402713,
      "grad_norm": 3.481749690836404,
      "learning_rate": 1.5571500631520254e-07,
      "loss": 0.6949,
      "step": 3364
    },
    {
      "epoch": 0.3326495810987816,
      "grad_norm": 2.6802031859849444,
      "learning_rate": 1.5568841273760055e-07,
      "loss": 0.6565,
      "step": 3365
    },
    {
      "epoch": 0.33274843684353606,
      "grad_norm": 5.830619743517246,
      "learning_rate": 1.5566181344997327e-07,
      "loss": 0.6615,
      "step": 3366
    },
    {
      "epoch": 0.33284729258829054,
      "grad_norm": 3.708599753443995,
      "learning_rate": 1.556352084550481e-07,
      "loss": 0.7361,
      "step": 3367
    },
    {
      "epoch": 0.332946148333045,
      "grad_norm": 7.950403461506017,
      "learning_rate": 1.5560859775555296e-07,
      "loss": 0.7878,
      "step": 3368
    },
    {
      "epoch": 0.33304500407779947,
      "grad_norm": 4.100171477492787,
      "learning_rate": 1.5558198135421648e-07,
      "loss": 0.8139,
      "step": 3369
    },
    {
      "epoch": 0.33314385982255396,
      "grad_norm": 4.501477811299288,
      "learning_rate": 1.5555535925376763e-07,
      "loss": 0.7256,
      "step": 3370
    },
    {
      "epoch": 0.3332427155673084,
      "grad_norm": 4.826852806293034,
      "learning_rate": 1.5552873145693623e-07,
      "loss": 0.7287,
      "step": 3371
    },
    {
      "epoch": 0.3333415713120629,
      "grad_norm": 9.384279243863439,
      "learning_rate": 1.555020979664525e-07,
      "loss": 0.7215,
      "step": 3372
    },
    {
      "epoch": 0.3334404270568173,
      "grad_norm": 13.36268725850793,
      "learning_rate": 1.5547545878504734e-07,
      "loss": 0.6717,
      "step": 3373
    },
    {
      "epoch": 0.3335392828015718,
      "grad_norm": 14.993033765111912,
      "learning_rate": 1.554488139154522e-07,
      "loss": 0.6099,
      "step": 3374
    },
    {
      "epoch": 0.3336381385463263,
      "grad_norm": 3.355874761285684,
      "learning_rate": 1.5542216336039913e-07,
      "loss": 0.7319,
      "step": 3375
    },
    {
      "epoch": 0.3337369942910807,
      "grad_norm": 3.441567396634726,
      "learning_rate": 1.5539550712262073e-07,
      "loss": 0.5852,
      "step": 3376
    },
    {
      "epoch": 0.3338358500358352,
      "grad_norm": 7.304128548857031,
      "learning_rate": 1.5536884520485022e-07,
      "loss": 0.7953,
      "step": 3377
    },
    {
      "epoch": 0.33393470578058965,
      "grad_norm": 6.372795311992978,
      "learning_rate": 1.5534217760982142e-07,
      "loss": 0.7305,
      "step": 3378
    },
    {
      "epoch": 0.33403356152534414,
      "grad_norm": 3.169429001516498,
      "learning_rate": 1.553155043402686e-07,
      "loss": 0.6968,
      "step": 3379
    },
    {
      "epoch": 0.3341324172700986,
      "grad_norm": 3.2082653451911254,
      "learning_rate": 1.552888253989268e-07,
      "loss": 0.7367,
      "step": 3380
    },
    {
      "epoch": 0.33423127301485306,
      "grad_norm": 4.14001201681523,
      "learning_rate": 1.5526214078853154e-07,
      "loss": 0.7292,
      "step": 3381
    },
    {
      "epoch": 0.33433012875960755,
      "grad_norm": 4.748658978370739,
      "learning_rate": 1.552354505118189e-07,
      "loss": 0.7058,
      "step": 3382
    },
    {
      "epoch": 0.33442898450436204,
      "grad_norm": 2.675636335241841,
      "learning_rate": 1.5520875457152559e-07,
      "loss": 0.7534,
      "step": 3383
    },
    {
      "epoch": 0.33452784024911647,
      "grad_norm": 3.8858371288736384,
      "learning_rate": 1.551820529703889e-07,
      "loss": 0.6956,
      "step": 3384
    },
    {
      "epoch": 0.33462669599387096,
      "grad_norm": 7.193825715201345,
      "learning_rate": 1.5515534571114662e-07,
      "loss": 0.7028,
      "step": 3385
    },
    {
      "epoch": 0.3347255517386254,
      "grad_norm": 3.20656162922396,
      "learning_rate": 1.5512863279653726e-07,
      "loss": 0.6846,
      "step": 3386
    },
    {
      "epoch": 0.3348244074833799,
      "grad_norm": 3.284387709445011,
      "learning_rate": 1.5510191422929983e-07,
      "loss": 0.6777,
      "step": 3387
    },
    {
      "epoch": 0.33492326322813437,
      "grad_norm": 3.7232391760940535,
      "learning_rate": 1.5507519001217393e-07,
      "loss": 0.7124,
      "step": 3388
    },
    {
      "epoch": 0.3350221189728888,
      "grad_norm": 6.312293890642153,
      "learning_rate": 1.5504846014789968e-07,
      "loss": 0.7225,
      "step": 3389
    },
    {
      "epoch": 0.3351209747176433,
      "grad_norm": 3.7344802949567684,
      "learning_rate": 1.5502172463921784e-07,
      "loss": 0.6989,
      "step": 3390
    },
    {
      "epoch": 0.3352198304623977,
      "grad_norm": 4.1657204438288975,
      "learning_rate": 1.549949834888698e-07,
      "loss": 0.7075,
      "step": 3391
    },
    {
      "epoch": 0.3353186862071522,
      "grad_norm": 9.246200845789376,
      "learning_rate": 1.549682366995974e-07,
      "loss": 0.7064,
      "step": 3392
    },
    {
      "epoch": 0.3354175419519067,
      "grad_norm": 3.4690785631806187,
      "learning_rate": 1.549414842741432e-07,
      "loss": 0.7355,
      "step": 3393
    },
    {
      "epoch": 0.33551639769666114,
      "grad_norm": 6.219510268129855,
      "learning_rate": 1.5491472621525023e-07,
      "loss": 0.664,
      "step": 3394
    },
    {
      "epoch": 0.33561525344141563,
      "grad_norm": 16.236229142668194,
      "learning_rate": 1.548879625256621e-07,
      "loss": 0.8051,
      "step": 3395
    },
    {
      "epoch": 0.33571410918617006,
      "grad_norm": 3.3754943253426317,
      "learning_rate": 1.5486119320812315e-07,
      "loss": 0.7154,
      "step": 3396
    },
    {
      "epoch": 0.33581296493092455,
      "grad_norm": 8.104192442493066,
      "learning_rate": 1.54834418265378e-07,
      "loss": 0.6722,
      "step": 3397
    },
    {
      "epoch": 0.33591182067567904,
      "grad_norm": 4.351751319896158,
      "learning_rate": 1.5480763770017218e-07,
      "loss": 0.7202,
      "step": 3398
    },
    {
      "epoch": 0.3360106764204335,
      "grad_norm": 3.1659015905809085,
      "learning_rate": 1.5478085151525155e-07,
      "loss": 0.7809,
      "step": 3399
    },
    {
      "epoch": 0.33610953216518796,
      "grad_norm": 22.931191725149997,
      "learning_rate": 1.5475405971336272e-07,
      "loss": 0.6823,
      "step": 3400
    },
    {
      "epoch": 0.3362083879099424,
      "grad_norm": 4.179182046622785,
      "learning_rate": 1.547272622972527e-07,
      "loss": 0.7477,
      "step": 3401
    },
    {
      "epoch": 0.3363072436546969,
      "grad_norm": 10.377681879359452,
      "learning_rate": 1.547004592696692e-07,
      "loss": 0.6915,
      "step": 3402
    },
    {
      "epoch": 0.3364060993994514,
      "grad_norm": 4.181482891849487,
      "learning_rate": 1.5467365063336052e-07,
      "loss": 0.7485,
      "step": 3403
    },
    {
      "epoch": 0.3365049551442058,
      "grad_norm": 6.205950005754268,
      "learning_rate": 1.5464683639107546e-07,
      "loss": 0.7404,
      "step": 3404
    },
    {
      "epoch": 0.3366038108889603,
      "grad_norm": 5.089735939931679,
      "learning_rate": 1.5462001654556344e-07,
      "loss": 0.7644,
      "step": 3405
    },
    {
      "epoch": 0.33670266663371473,
      "grad_norm": 9.920190067126605,
      "learning_rate": 1.545931910995744e-07,
      "loss": 0.7,
      "step": 3406
    },
    {
      "epoch": 0.3368015223784692,
      "grad_norm": 4.378507390051743,
      "learning_rate": 1.5456636005585893e-07,
      "loss": 0.7498,
      "step": 3407
    },
    {
      "epoch": 0.3369003781232237,
      "grad_norm": 5.287062516677671,
      "learning_rate": 1.5453952341716814e-07,
      "loss": 0.6652,
      "step": 3408
    },
    {
      "epoch": 0.33699923386797814,
      "grad_norm": 3.125418452623746,
      "learning_rate": 1.5451268118625374e-07,
      "loss": 0.7545,
      "step": 3409
    },
    {
      "epoch": 0.33709808961273263,
      "grad_norm": 3.7213114229095403,
      "learning_rate": 1.5448583336586802e-07,
      "loss": 0.8438,
      "step": 3410
    },
    {
      "epoch": 0.33719694535748707,
      "grad_norm": 4.6221567691655885,
      "learning_rate": 1.544589799587638e-07,
      "loss": 0.7401,
      "step": 3411
    },
    {
      "epoch": 0.33729580110224155,
      "grad_norm": 4.407400476487539,
      "learning_rate": 1.5443212096769449e-07,
      "loss": 0.8883,
      "step": 3412
    },
    {
      "epoch": 0.33739465684699604,
      "grad_norm": 3.5633952549105166,
      "learning_rate": 1.5440525639541413e-07,
      "loss": 0.6873,
      "step": 3413
    },
    {
      "epoch": 0.3374935125917505,
      "grad_norm": 3.374152459307575,
      "learning_rate": 1.5437838624467725e-07,
      "loss": 0.5983,
      "step": 3414
    },
    {
      "epoch": 0.33759236833650497,
      "grad_norm": 5.815609012281859,
      "learning_rate": 1.54351510518239e-07,
      "loss": 0.672,
      "step": 3415
    },
    {
      "epoch": 0.3376912240812594,
      "grad_norm": 3.9096887089050862,
      "learning_rate": 1.543246292188551e-07,
      "loss": 0.7057,
      "step": 3416
    },
    {
      "epoch": 0.3377900798260139,
      "grad_norm": 7.252330308690125,
      "learning_rate": 1.542977423492818e-07,
      "loss": 0.7466,
      "step": 3417
    },
    {
      "epoch": 0.3378889355707684,
      "grad_norm": 3.741137031911654,
      "learning_rate": 1.5427084991227598e-07,
      "loss": 0.7057,
      "step": 3418
    },
    {
      "epoch": 0.3379877913155228,
      "grad_norm": 5.403564310987992,
      "learning_rate": 1.5424395191059504e-07,
      "loss": 0.7762,
      "step": 3419
    },
    {
      "epoch": 0.3380866470602773,
      "grad_norm": 4.6238689582853345,
      "learning_rate": 1.5421704834699703e-07,
      "loss": 0.7073,
      "step": 3420
    },
    {
      "epoch": 0.33818550280503173,
      "grad_norm": 5.574016534854883,
      "learning_rate": 1.541901392242404e-07,
      "loss": 0.7049,
      "step": 3421
    },
    {
      "epoch": 0.3382843585497862,
      "grad_norm": 7.423189065037691,
      "learning_rate": 1.541632245450844e-07,
      "loss": 0.7122,
      "step": 3422
    },
    {
      "epoch": 0.3383832142945407,
      "grad_norm": 3.93400199944179,
      "learning_rate": 1.541363043122887e-07,
      "loss": 0.6675,
      "step": 3423
    },
    {
      "epoch": 0.33848207003929515,
      "grad_norm": 4.505967003032372,
      "learning_rate": 1.5410937852861353e-07,
      "loss": 0.8308,
      "step": 3424
    },
    {
      "epoch": 0.33858092578404964,
      "grad_norm": 4.142049694910113,
      "learning_rate": 1.5408244719681975e-07,
      "loss": 0.7395,
      "step": 3425
    },
    {
      "epoch": 0.33867978152880407,
      "grad_norm": 2.952409713569996,
      "learning_rate": 1.5405551031966884e-07,
      "loss": 0.5934,
      "step": 3426
    },
    {
      "epoch": 0.33877863727355856,
      "grad_norm": 3.494110302691782,
      "learning_rate": 1.5402856789992264e-07,
      "loss": 0.8214,
      "step": 3427
    },
    {
      "epoch": 0.33887749301831305,
      "grad_norm": 4.049408079991959,
      "learning_rate": 1.5400161994034384e-07,
      "loss": 0.7321,
      "step": 3428
    },
    {
      "epoch": 0.3389763487630675,
      "grad_norm": 14.900677526181031,
      "learning_rate": 1.5397466644369548e-07,
      "loss": 0.7159,
      "step": 3429
    },
    {
      "epoch": 0.33907520450782197,
      "grad_norm": 3.7496756464084626,
      "learning_rate": 1.5394770741274123e-07,
      "loss": 0.7717,
      "step": 3430
    },
    {
      "epoch": 0.3391740602525764,
      "grad_norm": 10.113740079840348,
      "learning_rate": 1.539207428502454e-07,
      "loss": 0.7211,
      "step": 3431
    },
    {
      "epoch": 0.3392729159973309,
      "grad_norm": 6.786960486767994,
      "learning_rate": 1.5389377275897276e-07,
      "loss": 0.7201,
      "step": 3432
    },
    {
      "epoch": 0.3393717717420854,
      "grad_norm": 7.521776711038403,
      "learning_rate": 1.5386679714168875e-07,
      "loss": 0.7717,
      "step": 3433
    },
    {
      "epoch": 0.3394706274868398,
      "grad_norm": 6.281606183573816,
      "learning_rate": 1.5383981600115927e-07,
      "loss": 0.7144,
      "step": 3434
    },
    {
      "epoch": 0.3395694832315943,
      "grad_norm": 6.7710367670826574,
      "learning_rate": 1.5381282934015082e-07,
      "loss": 0.7436,
      "step": 3435
    },
    {
      "epoch": 0.33966833897634874,
      "grad_norm": 3.840386609343554,
      "learning_rate": 1.5378583716143055e-07,
      "loss": 0.7122,
      "step": 3436
    },
    {
      "epoch": 0.3397671947211032,
      "grad_norm": 2.903791529528234,
      "learning_rate": 1.5375883946776608e-07,
      "loss": 0.72,
      "step": 3437
    },
    {
      "epoch": 0.3398660504658577,
      "grad_norm": 3.347402257706789,
      "learning_rate": 1.537318362619256e-07,
      "loss": 0.8048,
      "step": 3438
    },
    {
      "epoch": 0.33996490621061215,
      "grad_norm": 4.65261636066201,
      "learning_rate": 1.5370482754667792e-07,
      "loss": 0.7432,
      "step": 3439
    },
    {
      "epoch": 0.34006376195536664,
      "grad_norm": 3.73523971825845,
      "learning_rate": 1.5367781332479237e-07,
      "loss": 0.6098,
      "step": 3440
    },
    {
      "epoch": 0.3401626177001211,
      "grad_norm": 4.52246404304519,
      "learning_rate": 1.5365079359903885e-07,
      "loss": 0.7002,
      "step": 3441
    },
    {
      "epoch": 0.34026147344487556,
      "grad_norm": 3.1328148000937754,
      "learning_rate": 1.536237683721879e-07,
      "loss": 0.8115,
      "step": 3442
    },
    {
      "epoch": 0.34036032918963005,
      "grad_norm": 5.167569637842267,
      "learning_rate": 1.5359673764701048e-07,
      "loss": 0.7556,
      "step": 3443
    },
    {
      "epoch": 0.3404591849343845,
      "grad_norm": 3.450659395525906,
      "learning_rate": 1.5356970142627824e-07,
      "loss": 0.6362,
      "step": 3444
    },
    {
      "epoch": 0.340558040679139,
      "grad_norm": 20.160413183637818,
      "learning_rate": 1.535426597127633e-07,
      "loss": 0.7412,
      "step": 3445
    },
    {
      "epoch": 0.3406568964238934,
      "grad_norm": 3.618393746075291,
      "learning_rate": 1.535156125092384e-07,
      "loss": 0.6917,
      "step": 3446
    },
    {
      "epoch": 0.3407557521686479,
      "grad_norm": 3.8795819479009883,
      "learning_rate": 1.534885598184769e-07,
      "loss": 0.7324,
      "step": 3447
    },
    {
      "epoch": 0.3408546079134024,
      "grad_norm": 5.715136243513716,
      "learning_rate": 1.5346150164325256e-07,
      "loss": 0.7664,
      "step": 3448
    },
    {
      "epoch": 0.3409534636581568,
      "grad_norm": 6.8743957814530985,
      "learning_rate": 1.5343443798633984e-07,
      "loss": 0.7779,
      "step": 3449
    },
    {
      "epoch": 0.3410523194029113,
      "grad_norm": 3.4703042988085278,
      "learning_rate": 1.5340736885051375e-07,
      "loss": 0.7263,
      "step": 3450
    },
    {
      "epoch": 0.34115117514766574,
      "grad_norm": 3.596781608477162,
      "learning_rate": 1.533802942385498e-07,
      "loss": 0.7726,
      "step": 3451
    },
    {
      "epoch": 0.34125003089242023,
      "grad_norm": 3.0185521258379326,
      "learning_rate": 1.5335321415322407e-07,
      "loss": 0.6822,
      "step": 3452
    },
    {
      "epoch": 0.3413488866371747,
      "grad_norm": 3.1745446662853043,
      "learning_rate": 1.5332612859731323e-07,
      "loss": 0.7563,
      "step": 3453
    },
    {
      "epoch": 0.34144774238192915,
      "grad_norm": 2.918064530234702,
      "learning_rate": 1.5329903757359454e-07,
      "loss": 0.7457,
      "step": 3454
    },
    {
      "epoch": 0.34154659812668364,
      "grad_norm": 4.173601190597856,
      "learning_rate": 1.5327194108484575e-07,
      "loss": 0.6952,
      "step": 3455
    },
    {
      "epoch": 0.3416454538714381,
      "grad_norm": 17.88578398999503,
      "learning_rate": 1.5324483913384523e-07,
      "loss": 0.7426,
      "step": 3456
    },
    {
      "epoch": 0.34174430961619257,
      "grad_norm": 6.984622317110158,
      "learning_rate": 1.5321773172337187e-07,
      "loss": 0.8306,
      "step": 3457
    },
    {
      "epoch": 0.34184316536094705,
      "grad_norm": 13.027294432977849,
      "learning_rate": 1.5319061885620515e-07,
      "loss": 0.685,
      "step": 3458
    },
    {
      "epoch": 0.3419420211057015,
      "grad_norm": 5.351129861616586,
      "learning_rate": 1.5316350053512505e-07,
      "loss": 0.6603,
      "step": 3459
    },
    {
      "epoch": 0.342040876850456,
      "grad_norm": 3.5787038712614767,
      "learning_rate": 1.5313637676291222e-07,
      "loss": 0.7466,
      "step": 3460
    },
    {
      "epoch": 0.34213973259521047,
      "grad_norm": 3.711529990152059,
      "learning_rate": 1.5310924754234775e-07,
      "loss": 0.6843,
      "step": 3461
    },
    {
      "epoch": 0.3422385883399649,
      "grad_norm": 3.935626977595704,
      "learning_rate": 1.5308211287621337e-07,
      "loss": 0.7082,
      "step": 3462
    },
    {
      "epoch": 0.3423374440847194,
      "grad_norm": 3.0005123599210655,
      "learning_rate": 1.5305497276729133e-07,
      "loss": 0.6987,
      "step": 3463
    },
    {
      "epoch": 0.3424362998294738,
      "grad_norm": 9.946484957189817,
      "learning_rate": 1.5302782721836445e-07,
      "loss": 0.6379,
      "step": 3464
    },
    {
      "epoch": 0.3425351555742283,
      "grad_norm": 4.642564100319465,
      "learning_rate": 1.5300067623221612e-07,
      "loss": 0.7444,
      "step": 3465
    },
    {
      "epoch": 0.3426340113189828,
      "grad_norm": 5.371474910118832,
      "learning_rate": 1.5297351981163026e-07,
      "loss": 0.6839,
      "step": 3466
    },
    {
      "epoch": 0.34273286706373723,
      "grad_norm": 8.444037158874908,
      "learning_rate": 1.529463579593913e-07,
      "loss": 0.7859,
      "step": 3467
    },
    {
      "epoch": 0.3428317228084917,
      "grad_norm": 7.626211949736986,
      "learning_rate": 1.529191906782844e-07,
      "loss": 0.6871,
      "step": 3468
    },
    {
      "epoch": 0.34293057855324616,
      "grad_norm": 2.806900203121799,
      "learning_rate": 1.528920179710951e-07,
      "loss": 0.692,
      "step": 3469
    },
    {
      "epoch": 0.34302943429800065,
      "grad_norm": 3.0707822878338944,
      "learning_rate": 1.5286483984060956e-07,
      "loss": 0.6969,
      "step": 3470
    },
    {
      "epoch": 0.34312829004275514,
      "grad_norm": 3.13576791570386,
      "learning_rate": 1.528376562896145e-07,
      "loss": 0.6532,
      "step": 3471
    },
    {
      "epoch": 0.34322714578750957,
      "grad_norm": 7.816720306346298,
      "learning_rate": 1.528104673208972e-07,
      "loss": 0.7698,
      "step": 3472
    },
    {
      "epoch": 0.34332600153226406,
      "grad_norm": 4.096292936747229,
      "learning_rate": 1.5278327293724548e-07,
      "loss": 0.6498,
      "step": 3473
    },
    {
      "epoch": 0.3434248572770185,
      "grad_norm": 3.5978119615729685,
      "learning_rate": 1.5275607314144775e-07,
      "loss": 0.6859,
      "step": 3474
    },
    {
      "epoch": 0.343523713021773,
      "grad_norm": 3.672440759945241,
      "learning_rate": 1.527288679362929e-07,
      "loss": 0.7381,
      "step": 3475
    },
    {
      "epoch": 0.34362256876652747,
      "grad_norm": 6.602873425372868,
      "learning_rate": 1.5270165732457047e-07,
      "loss": 0.6881,
      "step": 3476
    },
    {
      "epoch": 0.3437214245112819,
      "grad_norm": 3.8893793149300824,
      "learning_rate": 1.526744413090705e-07,
      "loss": 0.734,
      "step": 3477
    },
    {
      "epoch": 0.3438202802560364,
      "grad_norm": 5.657276115624566,
      "learning_rate": 1.5264721989258352e-07,
      "loss": 0.7321,
      "step": 3478
    },
    {
      "epoch": 0.3439191360007908,
      "grad_norm": 3.1849025446724135,
      "learning_rate": 1.526199930779008e-07,
      "loss": 0.724,
      "step": 3479
    },
    {
      "epoch": 0.3440179917455453,
      "grad_norm": 2.983018431230655,
      "learning_rate": 1.5259276086781397e-07,
      "loss": 0.695,
      "step": 3480
    },
    {
      "epoch": 0.3441168474902998,
      "grad_norm": 7.763802298955474,
      "learning_rate": 1.5256552326511529e-07,
      "loss": 0.6063,
      "step": 3481
    },
    {
      "epoch": 0.34421570323505424,
      "grad_norm": 11.148054589894054,
      "learning_rate": 1.5253828027259762e-07,
      "loss": 0.7037,
      "step": 3482
    },
    {
      "epoch": 0.3443145589798087,
      "grad_norm": 4.375930007180887,
      "learning_rate": 1.525110318930543e-07,
      "loss": 0.662,
      "step": 3483
    },
    {
      "epoch": 0.34441341472456316,
      "grad_norm": 20.892056465673637,
      "learning_rate": 1.5248377812927924e-07,
      "loss": 0.7612,
      "step": 3484
    },
    {
      "epoch": 0.34451227046931765,
      "grad_norm": 4.434538545730614,
      "learning_rate": 1.5245651898406694e-07,
      "loss": 0.6684,
      "step": 3485
    },
    {
      "epoch": 0.34461112621407214,
      "grad_norm": 4.4008864879589975,
      "learning_rate": 1.524292544602124e-07,
      "loss": 0.7273,
      "step": 3486
    },
    {
      "epoch": 0.3447099819588266,
      "grad_norm": 17.81289332047644,
      "learning_rate": 1.524019845605112e-07,
      "loss": 0.603,
      "step": 3487
    },
    {
      "epoch": 0.34480883770358106,
      "grad_norm": 3.7967736071444858,
      "learning_rate": 1.5237470928775948e-07,
      "loss": 0.739,
      "step": 3488
    },
    {
      "epoch": 0.3449076934483355,
      "grad_norm": 4.037451805521626,
      "learning_rate": 1.5234742864475385e-07,
      "loss": 0.7925,
      "step": 3489
    },
    {
      "epoch": 0.34500654919309,
      "grad_norm": 7.085736869079253,
      "learning_rate": 1.5232014263429164e-07,
      "loss": 0.7605,
      "step": 3490
    },
    {
      "epoch": 0.3451054049378445,
      "grad_norm": 4.289994202369924,
      "learning_rate": 1.5229285125917058e-07,
      "loss": 0.8461,
      "step": 3491
    },
    {
      "epoch": 0.3452042606825989,
      "grad_norm": 5.667258014621529,
      "learning_rate": 1.5226555452218897e-07,
      "loss": 0.7336,
      "step": 3492
    },
    {
      "epoch": 0.3453031164273534,
      "grad_norm": 5.8837231558996015,
      "learning_rate": 1.5223825242614572e-07,
      "loss": 0.7481,
      "step": 3493
    },
    {
      "epoch": 0.34540197217210783,
      "grad_norm": 3.5363562203119914,
      "learning_rate": 1.5221094497384026e-07,
      "loss": 0.6776,
      "step": 3494
    },
    {
      "epoch": 0.3455008279168623,
      "grad_norm": 2.9308737999293206,
      "learning_rate": 1.521836321680726e-07,
      "loss": 0.7538,
      "step": 3495
    },
    {
      "epoch": 0.3455996836616168,
      "grad_norm": 4.943695216104501,
      "learning_rate": 1.5215631401164316e-07,
      "loss": 0.8277,
      "step": 3496
    },
    {
      "epoch": 0.34569853940637124,
      "grad_norm": 4.152638068581815,
      "learning_rate": 1.521289905073531e-07,
      "loss": 0.6888,
      "step": 3497
    },
    {
      "epoch": 0.34579739515112573,
      "grad_norm": 5.131385668278917,
      "learning_rate": 1.5210166165800403e-07,
      "loss": 0.7036,
      "step": 3498
    },
    {
      "epoch": 0.34589625089588016,
      "grad_norm": 9.210620189593723,
      "learning_rate": 1.520743274663981e-07,
      "loss": 0.6877,
      "step": 3499
    },
    {
      "epoch": 0.34599510664063465,
      "grad_norm": 5.50061700114168,
      "learning_rate": 1.52046987935338e-07,
      "loss": 0.7255,
      "step": 3500
    },
    {
      "epoch": 0.34609396238538914,
      "grad_norm": 4.846706679376976,
      "learning_rate": 1.520196430676271e-07,
      "loss": 0.748,
      "step": 3501
    },
    {
      "epoch": 0.3461928181301436,
      "grad_norm": 8.168753102690593,
      "learning_rate": 1.5199229286606912e-07,
      "loss": 0.698,
      "step": 3502
    },
    {
      "epoch": 0.34629167387489806,
      "grad_norm": 4.581332839265094,
      "learning_rate": 1.5196493733346846e-07,
      "loss": 0.702,
      "step": 3503
    },
    {
      "epoch": 0.3463905296196525,
      "grad_norm": 22.286580967054995,
      "learning_rate": 1.5193757647263e-07,
      "loss": 0.7059,
      "step": 3504
    },
    {
      "epoch": 0.346489385364407,
      "grad_norm": 5.617006624423454,
      "learning_rate": 1.5191021028635922e-07,
      "loss": 0.7186,
      "step": 3505
    },
    {
      "epoch": 0.3465882411091615,
      "grad_norm": 3.8745316417796087,
      "learning_rate": 1.518828387774621e-07,
      "loss": 0.6037,
      "step": 3506
    },
    {
      "epoch": 0.3466870968539159,
      "grad_norm": 5.876061715600381,
      "learning_rate": 1.5185546194874518e-07,
      "loss": 0.7474,
      "step": 3507
    },
    {
      "epoch": 0.3467859525986704,
      "grad_norm": 3.196815890671076,
      "learning_rate": 1.5182807980301555e-07,
      "loss": 0.7624,
      "step": 3508
    },
    {
      "epoch": 0.34688480834342483,
      "grad_norm": 3.1725137931540943,
      "learning_rate": 1.518006923430809e-07,
      "loss": 0.7217,
      "step": 3509
    },
    {
      "epoch": 0.3469836640881793,
      "grad_norm": 2.7157228586467146,
      "learning_rate": 1.517732995717493e-07,
      "loss": 0.6387,
      "step": 3510
    },
    {
      "epoch": 0.3470825198329338,
      "grad_norm": 9.923734573065127,
      "learning_rate": 1.5174590149182957e-07,
      "loss": 0.7095,
      "step": 3511
    },
    {
      "epoch": 0.34718137557768824,
      "grad_norm": 4.365618426693073,
      "learning_rate": 1.5171849810613097e-07,
      "loss": 0.7301,
      "step": 3512
    },
    {
      "epoch": 0.34728023132244273,
      "grad_norm": 4.514788388920906,
      "learning_rate": 1.5169108941746327e-07,
      "loss": 0.6812,
      "step": 3513
    },
    {
      "epoch": 0.34737908706719717,
      "grad_norm": 3.384827947043089,
      "learning_rate": 1.516636754286368e-07,
      "loss": 0.7315,
      "step": 3514
    },
    {
      "epoch": 0.34747794281195166,
      "grad_norm": 3.8093610004456173,
      "learning_rate": 1.5163625614246257e-07,
      "loss": 0.7798,
      "step": 3515
    },
    {
      "epoch": 0.34757679855670615,
      "grad_norm": 5.027058402465436,
      "learning_rate": 1.5160883156175195e-07,
      "loss": 0.675,
      "step": 3516
    },
    {
      "epoch": 0.3476756543014606,
      "grad_norm": 11.575420165225765,
      "learning_rate": 1.5158140168931693e-07,
      "loss": 0.6189,
      "step": 3517
    },
    {
      "epoch": 0.34777451004621507,
      "grad_norm": 3.3807262487667677,
      "learning_rate": 1.5155396652797003e-07,
      "loss": 0.683,
      "step": 3518
    },
    {
      "epoch": 0.3478733657909695,
      "grad_norm": 4.426345777794725,
      "learning_rate": 1.5152652608052433e-07,
      "loss": 0.6542,
      "step": 3519
    },
    {
      "epoch": 0.347972221535724,
      "grad_norm": 3.957028460298404,
      "learning_rate": 1.514990803497935e-07,
      "loss": 0.6858,
      "step": 3520
    },
    {
      "epoch": 0.3480710772804785,
      "grad_norm": 3.1117063771388302,
      "learning_rate": 1.5147162933859163e-07,
      "loss": 0.7622,
      "step": 3521
    },
    {
      "epoch": 0.3481699330252329,
      "grad_norm": 3.9932157247200095,
      "learning_rate": 1.5144417304973342e-07,
      "loss": 0.7966,
      "step": 3522
    },
    {
      "epoch": 0.3482687887699874,
      "grad_norm": 5.009970884138738,
      "learning_rate": 1.5141671148603412e-07,
      "loss": 0.7224,
      "step": 3523
    },
    {
      "epoch": 0.34836764451474184,
      "grad_norm": 3.3642018627306913,
      "learning_rate": 1.513892446503095e-07,
      "loss": 0.7312,
      "step": 3524
    },
    {
      "epoch": 0.3484665002594963,
      "grad_norm": 3.1270127627541346,
      "learning_rate": 1.5136177254537587e-07,
      "loss": 0.739,
      "step": 3525
    },
    {
      "epoch": 0.3485653560042508,
      "grad_norm": 3.432870835645691,
      "learning_rate": 1.5133429517405016e-07,
      "loss": 0.6256,
      "step": 3526
    },
    {
      "epoch": 0.34866421174900525,
      "grad_norm": 7.695780904259153,
      "learning_rate": 1.5130681253914968e-07,
      "loss": 0.7474,
      "step": 3527
    },
    {
      "epoch": 0.34876306749375974,
      "grad_norm": 3.604866464597523,
      "learning_rate": 1.512793246434924e-07,
      "loss": 0.8163,
      "step": 3528
    },
    {
      "epoch": 0.34886192323851417,
      "grad_norm": 3.244802464975086,
      "learning_rate": 1.512518314898968e-07,
      "loss": 0.7144,
      "step": 3529
    },
    {
      "epoch": 0.34896077898326866,
      "grad_norm": 3.3156638485057526,
      "learning_rate": 1.512243330811819e-07,
      "loss": 0.6428,
      "step": 3530
    },
    {
      "epoch": 0.34905963472802315,
      "grad_norm": 3.0231793072423607,
      "learning_rate": 1.5119682942016727e-07,
      "loss": 0.5858,
      "step": 3531
    },
    {
      "epoch": 0.3491584904727776,
      "grad_norm": 5.298337297851877,
      "learning_rate": 1.51169320509673e-07,
      "loss": 0.7473,
      "step": 3532
    },
    {
      "epoch": 0.34925734621753207,
      "grad_norm": 10.946866392807316,
      "learning_rate": 1.5114180635251966e-07,
      "loss": 0.7283,
      "step": 3533
    },
    {
      "epoch": 0.34935620196228656,
      "grad_norm": 4.311552878663955,
      "learning_rate": 1.511142869515285e-07,
      "loss": 0.6813,
      "step": 3534
    },
    {
      "epoch": 0.349455057707041,
      "grad_norm": 3.83280635864606,
      "learning_rate": 1.510867623095212e-07,
      "loss": 0.6871,
      "step": 3535
    },
    {
      "epoch": 0.3495539134517955,
      "grad_norm": 3.4529997755190625,
      "learning_rate": 1.5105923242932002e-07,
      "loss": 0.7064,
      "step": 3536
    },
    {
      "epoch": 0.3496527691965499,
      "grad_norm": 6.53175765617147,
      "learning_rate": 1.510316973137477e-07,
      "loss": 0.6994,
      "step": 3537
    },
    {
      "epoch": 0.3497516249413044,
      "grad_norm": 3.5075250024517857,
      "learning_rate": 1.5100415696562765e-07,
      "loss": 0.6663,
      "step": 3538
    },
    {
      "epoch": 0.3498504806860589,
      "grad_norm": 2.7801210002071963,
      "learning_rate": 1.5097661138778363e-07,
      "loss": 0.8319,
      "step": 3539
    },
    {
      "epoch": 0.34994933643081333,
      "grad_norm": 3.8992182146229912,
      "learning_rate": 1.5094906058304003e-07,
      "loss": 0.8037,
      "step": 3540
    },
    {
      "epoch": 0.3500481921755678,
      "grad_norm": 3.189224845181829,
      "learning_rate": 1.509215045542219e-07,
      "loss": 0.6482,
      "step": 3541
    },
    {
      "epoch": 0.35014704792032225,
      "grad_norm": 3.9117892633324836,
      "learning_rate": 1.5089394330415458e-07,
      "loss": 0.7433,
      "step": 3542
    },
    {
      "epoch": 0.35024590366507674,
      "grad_norm": 6.787832456304181,
      "learning_rate": 1.5086637683566413e-07,
      "loss": 0.6908,
      "step": 3543
    },
    {
      "epoch": 0.35034475940983123,
      "grad_norm": 7.889532236288772,
      "learning_rate": 1.5083880515157702e-07,
      "loss": 0.755,
      "step": 3544
    },
    {
      "epoch": 0.35044361515458566,
      "grad_norm": 4.702098459409698,
      "learning_rate": 1.508112282547204e-07,
      "loss": 0.7293,
      "step": 3545
    },
    {
      "epoch": 0.35054247089934015,
      "grad_norm": 3.7583379708936158,
      "learning_rate": 1.5078364614792186e-07,
      "loss": 0.7062,
      "step": 3546
    },
    {
      "epoch": 0.3506413266440946,
      "grad_norm": 3.0226732365585796,
      "learning_rate": 1.507560588340095e-07,
      "loss": 0.6973,
      "step": 3547
    },
    {
      "epoch": 0.3507401823888491,
      "grad_norm": 6.457467073729727,
      "learning_rate": 1.5072846631581202e-07,
      "loss": 0.7191,
      "step": 3548
    },
    {
      "epoch": 0.35083903813360356,
      "grad_norm": 7.927066686907352,
      "learning_rate": 1.5070086859615861e-07,
      "loss": 0.7585,
      "step": 3549
    },
    {
      "epoch": 0.350937893878358,
      "grad_norm": 5.723467655921544,
      "learning_rate": 1.50673265677879e-07,
      "loss": 0.6529,
      "step": 3550
    },
    {
      "epoch": 0.3510367496231125,
      "grad_norm": 3.743349590358212,
      "learning_rate": 1.5064565756380348e-07,
      "loss": 0.6308,
      "step": 3551
    },
    {
      "epoch": 0.3511356053678669,
      "grad_norm": 7.322723653496503,
      "learning_rate": 1.5061804425676288e-07,
      "loss": 0.7316,
      "step": 3552
    },
    {
      "epoch": 0.3512344611126214,
      "grad_norm": 12.300069279967863,
      "learning_rate": 1.505904257595885e-07,
      "loss": 0.7123,
      "step": 3553
    },
    {
      "epoch": 0.3513333168573759,
      "grad_norm": 3.2817336495573994,
      "learning_rate": 1.5056280207511223e-07,
      "loss": 0.6803,
      "step": 3554
    },
    {
      "epoch": 0.35143217260213033,
      "grad_norm": 5.343286822319338,
      "learning_rate": 1.5053517320616643e-07,
      "loss": 0.6529,
      "step": 3555
    },
    {
      "epoch": 0.3515310283468848,
      "grad_norm": 12.70158932641803,
      "learning_rate": 1.5050753915558408e-07,
      "loss": 0.7863,
      "step": 3556
    },
    {
      "epoch": 0.35162988409163926,
      "grad_norm": 3.126182924564837,
      "learning_rate": 1.5047989992619864e-07,
      "loss": 0.7702,
      "step": 3557
    },
    {
      "epoch": 0.35172873983639374,
      "grad_norm": 3.885258177897647,
      "learning_rate": 1.5045225552084405e-07,
      "loss": 0.6827,
      "step": 3558
    },
    {
      "epoch": 0.35182759558114823,
      "grad_norm": 4.142647420107966,
      "learning_rate": 1.5042460594235493e-07,
      "loss": 0.7477,
      "step": 3559
    },
    {
      "epoch": 0.35192645132590267,
      "grad_norm": 3.0291495262740726,
      "learning_rate": 1.5039695119356627e-07,
      "loss": 0.721,
      "step": 3560
    },
    {
      "epoch": 0.35202530707065716,
      "grad_norm": 4.539818305723451,
      "learning_rate": 1.5036929127731365e-07,
      "loss": 0.7561,
      "step": 3561
    },
    {
      "epoch": 0.3521241628154116,
      "grad_norm": 4.503891756808077,
      "learning_rate": 1.5034162619643324e-07,
      "loss": 0.7512,
      "step": 3562
    },
    {
      "epoch": 0.3522230185601661,
      "grad_norm": 3.7661209936815596,
      "learning_rate": 1.5031395595376164e-07,
      "loss": 0.7763,
      "step": 3563
    },
    {
      "epoch": 0.35232187430492057,
      "grad_norm": 3.615777300922611,
      "learning_rate": 1.5028628055213605e-07,
      "loss": 0.7896,
      "step": 3564
    },
    {
      "epoch": 0.352420730049675,
      "grad_norm": 4.328058368408435,
      "learning_rate": 1.5025859999439415e-07,
      "loss": 0.6687,
      "step": 3565
    },
    {
      "epoch": 0.3525195857944295,
      "grad_norm": 7.636062885849978,
      "learning_rate": 1.502309142833742e-07,
      "loss": 0.7688,
      "step": 3566
    },
    {
      "epoch": 0.3526184415391839,
      "grad_norm": 7.227419613475218,
      "learning_rate": 1.5020322342191497e-07,
      "loss": 0.672,
      "step": 3567
    },
    {
      "epoch": 0.3527172972839384,
      "grad_norm": 6.459968792122518,
      "learning_rate": 1.5017552741285568e-07,
      "loss": 0.6789,
      "step": 3568
    },
    {
      "epoch": 0.3528161530286929,
      "grad_norm": 4.3900211430987905,
      "learning_rate": 1.5014782625903618e-07,
      "loss": 0.6491,
      "step": 3569
    },
    {
      "epoch": 0.35291500877344734,
      "grad_norm": 3.7753090841119445,
      "learning_rate": 1.5012011996329688e-07,
      "loss": 0.6839,
      "step": 3570
    },
    {
      "epoch": 0.3530138645182018,
      "grad_norm": 3.094466311507995,
      "learning_rate": 1.500924085284786e-07,
      "loss": 0.7132,
      "step": 3571
    },
    {
      "epoch": 0.35311272026295626,
      "grad_norm": 3.0592766818150476,
      "learning_rate": 1.5006469195742273e-07,
      "loss": 0.6905,
      "step": 3572
    },
    {
      "epoch": 0.35321157600771075,
      "grad_norm": 2.997323221332699,
      "learning_rate": 1.500369702529712e-07,
      "loss": 0.6528,
      "step": 3573
    },
    {
      "epoch": 0.35331043175246524,
      "grad_norm": 2.7041979859823897,
      "learning_rate": 1.500092434179665e-07,
      "loss": 0.7827,
      "step": 3574
    },
    {
      "epoch": 0.35340928749721967,
      "grad_norm": 7.9111499071354325,
      "learning_rate": 1.4998151145525154e-07,
      "loss": 0.7306,
      "step": 3575
    },
    {
      "epoch": 0.35350814324197416,
      "grad_norm": 5.228559768965661,
      "learning_rate": 1.4995377436766984e-07,
      "loss": 0.7739,
      "step": 3576
    },
    {
      "epoch": 0.3536069989867286,
      "grad_norm": 3.559793581806373,
      "learning_rate": 1.4992603215806548e-07,
      "loss": 0.7239,
      "step": 3577
    },
    {
      "epoch": 0.3537058547314831,
      "grad_norm": 10.761271739465293,
      "learning_rate": 1.4989828482928298e-07,
      "loss": 0.6949,
      "step": 3578
    },
    {
      "epoch": 0.35380471047623757,
      "grad_norm": 4.962133754698089,
      "learning_rate": 1.4987053238416743e-07,
      "loss": 0.6694,
      "step": 3579
    },
    {
      "epoch": 0.353903566220992,
      "grad_norm": 6.990741266969024,
      "learning_rate": 1.498427748255644e-07,
      "loss": 0.7174,
      "step": 3580
    },
    {
      "epoch": 0.3540024219657465,
      "grad_norm": 15.97666795643509,
      "learning_rate": 1.4981501215632008e-07,
      "loss": 0.7755,
      "step": 3581
    },
    {
      "epoch": 0.35410127771050093,
      "grad_norm": 2.852960808113818,
      "learning_rate": 1.4978724437928105e-07,
      "loss": 0.6581,
      "step": 3582
    },
    {
      "epoch": 0.3542001334552554,
      "grad_norm": 3.1520382918086525,
      "learning_rate": 1.4975947149729455e-07,
      "loss": 0.7381,
      "step": 3583
    },
    {
      "epoch": 0.3542989892000099,
      "grad_norm": 10.162958080614407,
      "learning_rate": 1.4973169351320824e-07,
      "loss": 0.6536,
      "step": 3584
    },
    {
      "epoch": 0.35439784494476434,
      "grad_norm": 4.038614161194025,
      "learning_rate": 1.4970391042987035e-07,
      "loss": 0.7599,
      "step": 3585
    },
    {
      "epoch": 0.35449670068951883,
      "grad_norm": 4.26496554030443,
      "learning_rate": 1.4967612225012964e-07,
      "loss": 0.7883,
      "step": 3586
    },
    {
      "epoch": 0.35459555643427326,
      "grad_norm": 18.720833719002044,
      "learning_rate": 1.4964832897683536e-07,
      "loss": 0.6689,
      "step": 3587
    },
    {
      "epoch": 0.35469441217902775,
      "grad_norm": 4.880920090188901,
      "learning_rate": 1.496205306128373e-07,
      "loss": 0.7435,
      "step": 3588
    },
    {
      "epoch": 0.35479326792378224,
      "grad_norm": 4.396380744804757,
      "learning_rate": 1.4959272716098578e-07,
      "loss": 0.6598,
      "step": 3589
    },
    {
      "epoch": 0.3548921236685367,
      "grad_norm": 3.9117014656584606,
      "learning_rate": 1.4956491862413164e-07,
      "loss": 0.8134,
      "step": 3590
    },
    {
      "epoch": 0.35499097941329116,
      "grad_norm": 6.494599286210102,
      "learning_rate": 1.4953710500512625e-07,
      "loss": 0.7639,
      "step": 3591
    },
    {
      "epoch": 0.3550898351580456,
      "grad_norm": 5.1758771497641645,
      "learning_rate": 1.4950928630682146e-07,
      "loss": 0.7392,
      "step": 3592
    },
    {
      "epoch": 0.3551886909028001,
      "grad_norm": 4.621941366646833,
      "learning_rate": 1.4948146253206967e-07,
      "loss": 0.7206,
      "step": 3593
    },
    {
      "epoch": 0.3552875466475546,
      "grad_norm": 57.392796566108395,
      "learning_rate": 1.4945363368372378e-07,
      "loss": 0.7321,
      "step": 3594
    },
    {
      "epoch": 0.355386402392309,
      "grad_norm": 4.906146898594353,
      "learning_rate": 1.4942579976463725e-07,
      "loss": 0.6247,
      "step": 3595
    },
    {
      "epoch": 0.3554852581370635,
      "grad_norm": 3.181409189999283,
      "learning_rate": 1.4939796077766403e-07,
      "loss": 0.6626,
      "step": 3596
    },
    {
      "epoch": 0.35558411388181793,
      "grad_norm": 2.9463108480031392,
      "learning_rate": 1.4937011672565865e-07,
      "loss": 0.7061,
      "step": 3597
    },
    {
      "epoch": 0.3556829696265724,
      "grad_norm": 5.391663339219531,
      "learning_rate": 1.4934226761147603e-07,
      "loss": 0.7289,
      "step": 3598
    },
    {
      "epoch": 0.3557818253713269,
      "grad_norm": 6.034349354070085,
      "learning_rate": 1.493144134379717e-07,
      "loss": 0.7978,
      "step": 3599
    },
    {
      "epoch": 0.35588068111608134,
      "grad_norm": 3.533853007792903,
      "learning_rate": 1.4928655420800173e-07,
      "loss": 0.6773,
      "step": 3600
    },
    {
      "epoch": 0.35597953686083583,
      "grad_norm": 6.2053060906735675,
      "learning_rate": 1.4925868992442268e-07,
      "loss": 0.4986,
      "step": 3601
    },
    {
      "epoch": 0.35607839260559027,
      "grad_norm": 5.323765181127638,
      "learning_rate": 1.4923082059009154e-07,
      "loss": 0.7335,
      "step": 3602
    },
    {
      "epoch": 0.35617724835034476,
      "grad_norm": 5.529427976407117,
      "learning_rate": 1.4920294620786597e-07,
      "loss": 0.6965,
      "step": 3603
    },
    {
      "epoch": 0.35627610409509924,
      "grad_norm": 3.4663870057435733,
      "learning_rate": 1.4917506678060408e-07,
      "loss": 0.8293,
      "step": 3604
    },
    {
      "epoch": 0.3563749598398537,
      "grad_norm": 4.533449667026421,
      "learning_rate": 1.4914718231116447e-07,
      "loss": 0.6456,
      "step": 3605
    },
    {
      "epoch": 0.35647381558460817,
      "grad_norm": 7.3341347480888555,
      "learning_rate": 1.4911929280240628e-07,
      "loss": 0.67,
      "step": 3606
    },
    {
      "epoch": 0.3565726713293626,
      "grad_norm": 3.0837285792238145,
      "learning_rate": 1.4909139825718918e-07,
      "loss": 0.7892,
      "step": 3607
    },
    {
      "epoch": 0.3566715270741171,
      "grad_norm": 3.9440466081806087,
      "learning_rate": 1.490634986783733e-07,
      "loss": 0.7695,
      "step": 3608
    },
    {
      "epoch": 0.3567703828188716,
      "grad_norm": 3.5951081335379684,
      "learning_rate": 1.4903559406881944e-07,
      "loss": 0.7185,
      "step": 3609
    },
    {
      "epoch": 0.356869238563626,
      "grad_norm": 16.749967233475115,
      "learning_rate": 1.490076844313887e-07,
      "loss": 0.7642,
      "step": 3610
    },
    {
      "epoch": 0.3569680943083805,
      "grad_norm": 4.691976232644765,
      "learning_rate": 1.4897976976894282e-07,
      "loss": 0.699,
      "step": 3611
    },
    {
      "epoch": 0.357066950053135,
      "grad_norm": 3.6628060708962615,
      "learning_rate": 1.4895185008434405e-07,
      "loss": 0.6847,
      "step": 3612
    },
    {
      "epoch": 0.3571658057978894,
      "grad_norm": 12.632188543529255,
      "learning_rate": 1.4892392538045515e-07,
      "loss": 0.716,
      "step": 3613
    },
    {
      "epoch": 0.3572646615426439,
      "grad_norm": 4.06672354863936,
      "learning_rate": 1.4889599566013938e-07,
      "loss": 0.6801,
      "step": 3614
    },
    {
      "epoch": 0.35736351728739835,
      "grad_norm": 4.294167648926604,
      "learning_rate": 1.4886806092626053e-07,
      "loss": 0.695,
      "step": 3615
    },
    {
      "epoch": 0.35746237303215284,
      "grad_norm": 6.809606100436298,
      "learning_rate": 1.488401211816829e-07,
      "loss": 0.7019,
      "step": 3616
    },
    {
      "epoch": 0.3575612287769073,
      "grad_norm": 3.498010553802556,
      "learning_rate": 1.488121764292713e-07,
      "loss": 0.7476,
      "step": 3617
    },
    {
      "epoch": 0.35766008452166176,
      "grad_norm": 3.538001316796065,
      "learning_rate": 1.4878422667189098e-07,
      "loss": 0.6762,
      "step": 3618
    },
    {
      "epoch": 0.35775894026641625,
      "grad_norm": 9.350832690884783,
      "learning_rate": 1.487562719124079e-07,
      "loss": 0.6522,
      "step": 3619
    },
    {
      "epoch": 0.3578577960111707,
      "grad_norm": 3.265537145861484,
      "learning_rate": 1.4872831215368828e-07,
      "loss": 0.7538,
      "step": 3620
    },
    {
      "epoch": 0.35795665175592517,
      "grad_norm": 2.7792225725585746,
      "learning_rate": 1.4870034739859906e-07,
      "loss": 0.737,
      "step": 3621
    },
    {
      "epoch": 0.35805550750067966,
      "grad_norm": 6.896852167330569,
      "learning_rate": 1.486723776500076e-07,
      "loss": 0.7553,
      "step": 3622
    },
    {
      "epoch": 0.3581543632454341,
      "grad_norm": 3.842412874140129,
      "learning_rate": 1.486444029107818e-07,
      "loss": 0.7743,
      "step": 3623
    },
    {
      "epoch": 0.3582532189901886,
      "grad_norm": 3.8920428978117547,
      "learning_rate": 1.4861642318379e-07,
      "loss": 0.6526,
      "step": 3624
    },
    {
      "epoch": 0.358352074734943,
      "grad_norm": 2.843723893493013,
      "learning_rate": 1.4858843847190115e-07,
      "loss": 0.6369,
      "step": 3625
    },
    {
      "epoch": 0.3584509304796975,
      "grad_norm": 3.9943138418972746,
      "learning_rate": 1.485604487779847e-07,
      "loss": 0.6621,
      "step": 3626
    },
    {
      "epoch": 0.358549786224452,
      "grad_norm": 23.574058654771576,
      "learning_rate": 1.4853245410491054e-07,
      "loss": 0.6664,
      "step": 3627
    },
    {
      "epoch": 0.3586486419692064,
      "grad_norm": 3.5105090727833472,
      "learning_rate": 1.4850445445554906e-07,
      "loss": 0.736,
      "step": 3628
    },
    {
      "epoch": 0.3587474977139609,
      "grad_norm": 4.52796256428317,
      "learning_rate": 1.484764498327713e-07,
      "loss": 0.8814,
      "step": 3629
    },
    {
      "epoch": 0.35884635345871535,
      "grad_norm": 3.61307472157643,
      "learning_rate": 1.484484402394487e-07,
      "loss": 0.7433,
      "step": 3630
    },
    {
      "epoch": 0.35894520920346984,
      "grad_norm": 3.1323685544251885,
      "learning_rate": 1.4842042567845323e-07,
      "loss": 0.6954,
      "step": 3631
    },
    {
      "epoch": 0.35904406494822433,
      "grad_norm": 4.778258149592983,
      "learning_rate": 1.4839240615265732e-07,
      "loss": 0.8673,
      "step": 3632
    },
    {
      "epoch": 0.35914292069297876,
      "grad_norm": 3.1459659546138,
      "learning_rate": 1.4836438166493402e-07,
      "loss": 0.7535,
      "step": 3633
    },
    {
      "epoch": 0.35924177643773325,
      "grad_norm": 3.5388816162076795,
      "learning_rate": 1.483363522181568e-07,
      "loss": 0.631,
      "step": 3634
    },
    {
      "epoch": 0.3593406321824877,
      "grad_norm": 2.709089793298291,
      "learning_rate": 1.4830831781519967e-07,
      "loss": 0.7561,
      "step": 3635
    },
    {
      "epoch": 0.3594394879272422,
      "grad_norm": 4.585001470495791,
      "learning_rate": 1.4828027845893715e-07,
      "loss": 0.7911,
      "step": 3636
    },
    {
      "epoch": 0.35953834367199666,
      "grad_norm": 3.010205364624134,
      "learning_rate": 1.4825223415224427e-07,
      "loss": 0.6452,
      "step": 3637
    },
    {
      "epoch": 0.3596371994167511,
      "grad_norm": 2.9008046211295153,
      "learning_rate": 1.4822418489799653e-07,
      "loss": 0.7033,
      "step": 3638
    },
    {
      "epoch": 0.3597360551615056,
      "grad_norm": 3.250226240587601,
      "learning_rate": 1.4819613069906998e-07,
      "loss": 0.8759,
      "step": 3639
    },
    {
      "epoch": 0.35983491090626,
      "grad_norm": 7.43852164708562,
      "learning_rate": 1.4816807155834117e-07,
      "loss": 0.6842,
      "step": 3640
    },
    {
      "epoch": 0.3599337666510145,
      "grad_norm": 4.768673449359528,
      "learning_rate": 1.4814000747868714e-07,
      "loss": 0.7781,
      "step": 3641
    },
    {
      "epoch": 0.360032622395769,
      "grad_norm": 3.2487134246053677,
      "learning_rate": 1.4811193846298552e-07,
      "loss": 0.6069,
      "step": 3642
    },
    {
      "epoch": 0.36013147814052343,
      "grad_norm": 3.810051852484412,
      "learning_rate": 1.4808386451411424e-07,
      "loss": 0.8084,
      "step": 3643
    },
    {
      "epoch": 0.3602303338852779,
      "grad_norm": 2.8724611653101144,
      "learning_rate": 1.48055785634952e-07,
      "loss": 0.7148,
      "step": 3644
    },
    {
      "epoch": 0.36032918963003235,
      "grad_norm": 3.41696190279382,
      "learning_rate": 1.480277018283778e-07,
      "loss": 0.7433,
      "step": 3645
    },
    {
      "epoch": 0.36042804537478684,
      "grad_norm": 4.0214063949870775,
      "learning_rate": 1.479996130972712e-07,
      "loss": 0.6876,
      "step": 3646
    },
    {
      "epoch": 0.36052690111954133,
      "grad_norm": 3.4810808439349312,
      "learning_rate": 1.4797151944451233e-07,
      "loss": 0.7872,
      "step": 3647
    },
    {
      "epoch": 0.36062575686429577,
      "grad_norm": 5.881220253017299,
      "learning_rate": 1.479434208729818e-07,
      "loss": 0.6805,
      "step": 3648
    },
    {
      "epoch": 0.36072461260905025,
      "grad_norm": 4.042773929620973,
      "learning_rate": 1.4791531738556066e-07,
      "loss": 0.7676,
      "step": 3649
    },
    {
      "epoch": 0.3608234683538047,
      "grad_norm": 35.35422784286995,
      "learning_rate": 1.478872089851305e-07,
      "loss": 0.721,
      "step": 3650
    },
    {
      "epoch": 0.3609223240985592,
      "grad_norm": 5.753701013298731,
      "learning_rate": 1.4785909567457347e-07,
      "loss": 0.6904,
      "step": 3651
    },
    {
      "epoch": 0.36102117984331367,
      "grad_norm": 4.242589158252777,
      "learning_rate": 1.4783097745677214e-07,
      "loss": 0.7583,
      "step": 3652
    },
    {
      "epoch": 0.3611200355880681,
      "grad_norm": 3.3178323967603536,
      "learning_rate": 1.4780285433460966e-07,
      "loss": 0.762,
      "step": 3653
    },
    {
      "epoch": 0.3612188913328226,
      "grad_norm": 5.109321238883726,
      "learning_rate": 1.477747263109696e-07,
      "loss": 0.7061,
      "step": 3654
    },
    {
      "epoch": 0.361317747077577,
      "grad_norm": 3.694369670782072,
      "learning_rate": 1.4774659338873604e-07,
      "loss": 0.6455,
      "step": 3655
    },
    {
      "epoch": 0.3614166028223315,
      "grad_norm": 9.00655024315941,
      "learning_rate": 1.4771845557079367e-07,
      "loss": 0.6764,
      "step": 3656
    },
    {
      "epoch": 0.361515458567086,
      "grad_norm": 4.252860320051604,
      "learning_rate": 1.4769031286002759e-07,
      "loss": 0.7111,
      "step": 3657
    },
    {
      "epoch": 0.36161431431184043,
      "grad_norm": 3.0528342298512583,
      "learning_rate": 1.4766216525932338e-07,
      "loss": 0.6207,
      "step": 3658
    },
    {
      "epoch": 0.3617131700565949,
      "grad_norm": 3.49128515906311,
      "learning_rate": 1.476340127715672e-07,
      "loss": 0.6417,
      "step": 3659
    },
    {
      "epoch": 0.36181202580134936,
      "grad_norm": 10.130568883581747,
      "learning_rate": 1.4760585539964566e-07,
      "loss": 0.6862,
      "step": 3660
    },
    {
      "epoch": 0.36191088154610385,
      "grad_norm": 3.7979188734732765,
      "learning_rate": 1.475776931464459e-07,
      "loss": 0.6852,
      "step": 3661
    },
    {
      "epoch": 0.36200973729085834,
      "grad_norm": 4.826450125747866,
      "learning_rate": 1.4754952601485552e-07,
      "loss": 0.7537,
      "step": 3662
    },
    {
      "epoch": 0.36210859303561277,
      "grad_norm": 4.73201016657891,
      "learning_rate": 1.4752135400776266e-07,
      "loss": 0.7248,
      "step": 3663
    },
    {
      "epoch": 0.36220744878036726,
      "grad_norm": 4.769561134201383,
      "learning_rate": 1.474931771280559e-07,
      "loss": 0.811,
      "step": 3664
    },
    {
      "epoch": 0.3623063045251217,
      "grad_norm": 4.2894055857568905,
      "learning_rate": 1.4746499537862445e-07,
      "loss": 0.7021,
      "step": 3665
    },
    {
      "epoch": 0.3624051602698762,
      "grad_norm": 3.8654516305296367,
      "learning_rate": 1.4743680876235783e-07,
      "loss": 0.8171,
      "step": 3666
    },
    {
      "epoch": 0.36250401601463067,
      "grad_norm": 3.222910537016421,
      "learning_rate": 1.4740861728214626e-07,
      "loss": 0.7587,
      "step": 3667
    },
    {
      "epoch": 0.3626028717593851,
      "grad_norm": 6.647395269469236,
      "learning_rate": 1.4738042094088028e-07,
      "loss": 0.7387,
      "step": 3668
    },
    {
      "epoch": 0.3627017275041396,
      "grad_norm": 5.24899423135289,
      "learning_rate": 1.4735221974145104e-07,
      "loss": 0.719,
      "step": 3669
    },
    {
      "epoch": 0.362800583248894,
      "grad_norm": 4.552162839509807,
      "learning_rate": 1.473240136867502e-07,
      "loss": 0.6894,
      "step": 3670
    },
    {
      "epoch": 0.3628994389936485,
      "grad_norm": 3.706143857815008,
      "learning_rate": 1.4729580277966981e-07,
      "loss": 0.6868,
      "step": 3671
    },
    {
      "epoch": 0.362998294738403,
      "grad_norm": 3.767087111052779,
      "learning_rate": 1.472675870231025e-07,
      "loss": 0.6068,
      "step": 3672
    },
    {
      "epoch": 0.36309715048315744,
      "grad_norm": 4.055451608309152,
      "learning_rate": 1.472393664199414e-07,
      "loss": 0.7999,
      "step": 3673
    },
    {
      "epoch": 0.3631960062279119,
      "grad_norm": 3.9857492064063615,
      "learning_rate": 1.4721114097308012e-07,
      "loss": 0.7349,
      "step": 3674
    },
    {
      "epoch": 0.36329486197266636,
      "grad_norm": 15.498486021687006,
      "learning_rate": 1.471829106854127e-07,
      "loss": 0.7611,
      "step": 3675
    },
    {
      "epoch": 0.36339371771742085,
      "grad_norm": 3.3546242230637873,
      "learning_rate": 1.471546755598338e-07,
      "loss": 0.7916,
      "step": 3676
    },
    {
      "epoch": 0.36349257346217534,
      "grad_norm": 4.32823231744246,
      "learning_rate": 1.4712643559923852e-07,
      "loss": 0.7383,
      "step": 3677
    },
    {
      "epoch": 0.3635914292069298,
      "grad_norm": 3.942507469591325,
      "learning_rate": 1.4709819080652244e-07,
      "loss": 0.7639,
      "step": 3678
    },
    {
      "epoch": 0.36369028495168426,
      "grad_norm": 6.20192126794547,
      "learning_rate": 1.470699411845816e-07,
      "loss": 0.7336,
      "step": 3679
    },
    {
      "epoch": 0.3637891406964387,
      "grad_norm": 2.847975375953498,
      "learning_rate": 1.4704168673631266e-07,
      "loss": 0.7267,
      "step": 3680
    },
    {
      "epoch": 0.3638879964411932,
      "grad_norm": 3.3847220707475967,
      "learning_rate": 1.4701342746461264e-07,
      "loss": 0.6306,
      "step": 3681
    },
    {
      "epoch": 0.3639868521859477,
      "grad_norm": 5.227086473188239,
      "learning_rate": 1.4698516337237916e-07,
      "loss": 0.7493,
      "step": 3682
    },
    {
      "epoch": 0.3640857079307021,
      "grad_norm": 4.0903294212781365,
      "learning_rate": 1.4695689446251024e-07,
      "loss": 0.7549,
      "step": 3683
    },
    {
      "epoch": 0.3641845636754566,
      "grad_norm": 6.391138634465423,
      "learning_rate": 1.4692862073790443e-07,
      "loss": 0.7116,
      "step": 3684
    },
    {
      "epoch": 0.36428341942021103,
      "grad_norm": 5.962367896957634,
      "learning_rate": 1.4690034220146083e-07,
      "loss": 0.6957,
      "step": 3685
    },
    {
      "epoch": 0.3643822751649655,
      "grad_norm": 3.2352943494835076,
      "learning_rate": 1.4687205885607896e-07,
      "loss": 0.6195,
      "step": 3686
    },
    {
      "epoch": 0.36448113090972,
      "grad_norm": 6.05635398614733,
      "learning_rate": 1.4684377070465889e-07,
      "loss": 0.7303,
      "step": 3687
    },
    {
      "epoch": 0.36457998665447444,
      "grad_norm": 5.8495066657352,
      "learning_rate": 1.468154777501011e-07,
      "loss": 0.6651,
      "step": 3688
    },
    {
      "epoch": 0.36467884239922893,
      "grad_norm": 3.8648590556467637,
      "learning_rate": 1.4678717999530666e-07,
      "loss": 0.688,
      "step": 3689
    },
    {
      "epoch": 0.3647776981439834,
      "grad_norm": 9.445869636338283,
      "learning_rate": 1.4675887744317706e-07,
      "loss": 0.7142,
      "step": 3690
    },
    {
      "epoch": 0.36487655388873785,
      "grad_norm": 2.979750569128963,
      "learning_rate": 1.4673057009661432e-07,
      "loss": 0.7772,
      "step": 3691
    },
    {
      "epoch": 0.36497540963349234,
      "grad_norm": 3.1053265448806355,
      "learning_rate": 1.4670225795852092e-07,
      "loss": 0.8139,
      "step": 3692
    },
    {
      "epoch": 0.3650742653782468,
      "grad_norm": 3.3995116526538998,
      "learning_rate": 1.466739410317999e-07,
      "loss": 0.644,
      "step": 3693
    },
    {
      "epoch": 0.36517312112300127,
      "grad_norm": 4.452139069508782,
      "learning_rate": 1.4664561931935474e-07,
      "loss": 0.7543,
      "step": 3694
    },
    {
      "epoch": 0.36527197686775575,
      "grad_norm": 4.643532483875229,
      "learning_rate": 1.4661729282408934e-07,
      "loss": 0.62,
      "step": 3695
    },
    {
      "epoch": 0.3653708326125102,
      "grad_norm": 3.7135638795007195,
      "learning_rate": 1.4658896154890825e-07,
      "loss": 0.7669,
      "step": 3696
    },
    {
      "epoch": 0.3654696883572647,
      "grad_norm": 5.290097389846457,
      "learning_rate": 1.465606254967164e-07,
      "loss": 0.7396,
      "step": 3697
    },
    {
      "epoch": 0.3655685441020191,
      "grad_norm": 3.2385668684521938,
      "learning_rate": 1.4653228467041917e-07,
      "loss": 0.6711,
      "step": 3698
    },
    {
      "epoch": 0.3656673998467736,
      "grad_norm": 3.9086768518953248,
      "learning_rate": 1.4650393907292257e-07,
      "loss": 0.6945,
      "step": 3699
    },
    {
      "epoch": 0.3657662555915281,
      "grad_norm": 3.592808297693197,
      "learning_rate": 1.4647558870713302e-07,
      "loss": 0.6037,
      "step": 3700
    },
    {
      "epoch": 0.3658651113362825,
      "grad_norm": 3.881343659392955,
      "learning_rate": 1.4644723357595738e-07,
      "loss": 0.6651,
      "step": 3701
    },
    {
      "epoch": 0.365963967081037,
      "grad_norm": 3.5552836221358506,
      "learning_rate": 1.464188736823031e-07,
      "loss": 0.6501,
      "step": 3702
    },
    {
      "epoch": 0.36606282282579145,
      "grad_norm": 4.663740398167636,
      "learning_rate": 1.4639050902907805e-07,
      "loss": 0.8298,
      "step": 3703
    },
    {
      "epoch": 0.36616167857054593,
      "grad_norm": 4.59516646973444,
      "learning_rate": 1.463621396191906e-07,
      "loss": 0.7064,
      "step": 3704
    },
    {
      "epoch": 0.3662605343153004,
      "grad_norm": 13.197558070435921,
      "learning_rate": 1.4633376545554966e-07,
      "loss": 0.5933,
      "step": 3705
    },
    {
      "epoch": 0.36635939006005486,
      "grad_norm": 5.528350042398596,
      "learning_rate": 1.463053865410645e-07,
      "loss": 0.7566,
      "step": 3706
    },
    {
      "epoch": 0.36645824580480935,
      "grad_norm": 3.628773762901117,
      "learning_rate": 1.4627700287864502e-07,
      "loss": 0.6146,
      "step": 3707
    },
    {
      "epoch": 0.3665571015495638,
      "grad_norm": 26.980812548083414,
      "learning_rate": 1.4624861447120155e-07,
      "loss": 0.7897,
      "step": 3708
    },
    {
      "epoch": 0.36665595729431827,
      "grad_norm": 4.226615883760169,
      "learning_rate": 1.4622022132164483e-07,
      "loss": 0.7244,
      "step": 3709
    },
    {
      "epoch": 0.36675481303907276,
      "grad_norm": 2.6952402148837833,
      "learning_rate": 1.4619182343288623e-07,
      "loss": 0.6694,
      "step": 3710
    },
    {
      "epoch": 0.3668536687838272,
      "grad_norm": 3.0884267528749385,
      "learning_rate": 1.461634208078375e-07,
      "loss": 0.8529,
      "step": 3711
    },
    {
      "epoch": 0.3669525245285817,
      "grad_norm": 5.629629300356241,
      "learning_rate": 1.4613501344941096e-07,
      "loss": 0.7982,
      "step": 3712
    },
    {
      "epoch": 0.3670513802733361,
      "grad_norm": 6.421780163941448,
      "learning_rate": 1.4610660136051928e-07,
      "loss": 0.7947,
      "step": 3713
    },
    {
      "epoch": 0.3671502360180906,
      "grad_norm": 3.436041828849022,
      "learning_rate": 1.460781845440758e-07,
      "loss": 0.7406,
      "step": 3714
    },
    {
      "epoch": 0.3672490917628451,
      "grad_norm": 4.642129919224962,
      "learning_rate": 1.460497630029942e-07,
      "loss": 0.6823,
      "step": 3715
    },
    {
      "epoch": 0.3673479475075995,
      "grad_norm": 3.9816058868343105,
      "learning_rate": 1.4602133674018867e-07,
      "loss": 0.736,
      "step": 3716
    },
    {
      "epoch": 0.367446803252354,
      "grad_norm": 4.643835763120923,
      "learning_rate": 1.459929057585739e-07,
      "loss": 0.819,
      "step": 3717
    },
    {
      "epoch": 0.36754565899710845,
      "grad_norm": 5.540647424418481,
      "learning_rate": 1.4596447006106513e-07,
      "loss": 0.7242,
      "step": 3718
    },
    {
      "epoch": 0.36764451474186294,
      "grad_norm": 4.668644764956431,
      "learning_rate": 1.459360296505779e-07,
      "loss": 0.6353,
      "step": 3719
    },
    {
      "epoch": 0.3677433704866174,
      "grad_norm": 15.938455990022064,
      "learning_rate": 1.459075845300285e-07,
      "loss": 0.6798,
      "step": 3720
    },
    {
      "epoch": 0.36784222623137186,
      "grad_norm": 6.259922330575343,
      "learning_rate": 1.4587913470233348e-07,
      "loss": 0.7163,
      "step": 3721
    },
    {
      "epoch": 0.36794108197612635,
      "grad_norm": 5.611970573918138,
      "learning_rate": 1.4585068017040996e-07,
      "loss": 0.7232,
      "step": 3722
    },
    {
      "epoch": 0.3680399377208808,
      "grad_norm": 6.892965548170632,
      "learning_rate": 1.4582222093717554e-07,
      "loss": 0.7947,
      "step": 3723
    },
    {
      "epoch": 0.3681387934656353,
      "grad_norm": 3.0431060098310576,
      "learning_rate": 1.4579375700554828e-07,
      "loss": 0.6657,
      "step": 3724
    },
    {
      "epoch": 0.36823764921038976,
      "grad_norm": 6.403457597244105,
      "learning_rate": 1.4576528837844674e-07,
      "loss": 0.7812,
      "step": 3725
    },
    {
      "epoch": 0.3683365049551442,
      "grad_norm": 4.5561572090310785,
      "learning_rate": 1.4573681505878996e-07,
      "loss": 0.6348,
      "step": 3726
    },
    {
      "epoch": 0.3684353606998987,
      "grad_norm": 8.894566503408681,
      "learning_rate": 1.4570833704949746e-07,
      "loss": 0.8652,
      "step": 3727
    },
    {
      "epoch": 0.3685342164446531,
      "grad_norm": 22.63536318870003,
      "learning_rate": 1.4567985435348926e-07,
      "loss": 0.7191,
      "step": 3728
    },
    {
      "epoch": 0.3686330721894076,
      "grad_norm": 3.477222386098271,
      "learning_rate": 1.456513669736858e-07,
      "loss": 0.8078,
      "step": 3729
    },
    {
      "epoch": 0.3687319279341621,
      "grad_norm": 4.324272999738785,
      "learning_rate": 1.4562287491300808e-07,
      "loss": 0.7225,
      "step": 3730
    },
    {
      "epoch": 0.36883078367891653,
      "grad_norm": 3.174075100210025,
      "learning_rate": 1.4559437817437753e-07,
      "loss": 0.6384,
      "step": 3731
    },
    {
      "epoch": 0.368929639423671,
      "grad_norm": 4.764392974913229,
      "learning_rate": 1.4556587676071602e-07,
      "loss": 0.7169,
      "step": 3732
    },
    {
      "epoch": 0.36902849516842545,
      "grad_norm": 4.448229127269209,
      "learning_rate": 1.4553737067494603e-07,
      "loss": 0.6438,
      "step": 3733
    },
    {
      "epoch": 0.36912735091317994,
      "grad_norm": 8.154185467828954,
      "learning_rate": 1.4550885991999038e-07,
      "loss": 0.6426,
      "step": 3734
    },
    {
      "epoch": 0.36922620665793443,
      "grad_norm": 6.012363866815673,
      "learning_rate": 1.4548034449877245e-07,
      "loss": 0.6403,
      "step": 3735
    },
    {
      "epoch": 0.36932506240268886,
      "grad_norm": 3.452356229150554,
      "learning_rate": 1.4545182441421607e-07,
      "loss": 0.782,
      "step": 3736
    },
    {
      "epoch": 0.36942391814744335,
      "grad_norm": 4.698745747368808,
      "learning_rate": 1.454232996692456e-07,
      "loss": 0.6974,
      "step": 3737
    },
    {
      "epoch": 0.3695227738921978,
      "grad_norm": 4.141339333448688,
      "learning_rate": 1.4539477026678576e-07,
      "loss": 0.7588,
      "step": 3738
    },
    {
      "epoch": 0.3696216296369523,
      "grad_norm": 7.588497593051641,
      "learning_rate": 1.4536623620976182e-07,
      "loss": 0.7688,
      "step": 3739
    },
    {
      "epoch": 0.36972048538170676,
      "grad_norm": 4.881478311060784,
      "learning_rate": 1.453376975010996e-07,
      "loss": 0.6645,
      "step": 3740
    },
    {
      "epoch": 0.3698193411264612,
      "grad_norm": 10.351039531887343,
      "learning_rate": 1.4530915414372526e-07,
      "loss": 0.7843,
      "step": 3741
    },
    {
      "epoch": 0.3699181968712157,
      "grad_norm": 8.734386296133778,
      "learning_rate": 1.4528060614056553e-07,
      "loss": 0.6563,
      "step": 3742
    },
    {
      "epoch": 0.3700170526159701,
      "grad_norm": 4.506233399890511,
      "learning_rate": 1.4525205349454756e-07,
      "loss": 0.6733,
      "step": 3743
    },
    {
      "epoch": 0.3701159083607246,
      "grad_norm": 3.8233482002636396,
      "learning_rate": 1.4522349620859904e-07,
      "loss": 0.7501,
      "step": 3744
    },
    {
      "epoch": 0.3702147641054791,
      "grad_norm": 3.058382078745008,
      "learning_rate": 1.4519493428564804e-07,
      "loss": 0.8108,
      "step": 3745
    },
    {
      "epoch": 0.37031361985023353,
      "grad_norm": 5.844453116367016,
      "learning_rate": 1.4516636772862322e-07,
      "loss": 0.7017,
      "step": 3746
    },
    {
      "epoch": 0.370412475594988,
      "grad_norm": 5.347106019110983,
      "learning_rate": 1.4513779654045367e-07,
      "loss": 0.7757,
      "step": 3747
    },
    {
      "epoch": 0.37051133133974246,
      "grad_norm": 9.375756622696139,
      "learning_rate": 1.451092207240689e-07,
      "loss": 0.695,
      "step": 3748
    },
    {
      "epoch": 0.37061018708449694,
      "grad_norm": 3.6499124050116927,
      "learning_rate": 1.4508064028239892e-07,
      "loss": 0.7319,
      "step": 3749
    },
    {
      "epoch": 0.37070904282925143,
      "grad_norm": 3.341110004299616,
      "learning_rate": 1.4505205521837431e-07,
      "loss": 0.6793,
      "step": 3750
    },
    {
      "epoch": 0.37080789857400587,
      "grad_norm": 3.3624991550569128,
      "learning_rate": 1.4502346553492598e-07,
      "loss": 0.6767,
      "step": 3751
    },
    {
      "epoch": 0.37090675431876036,
      "grad_norm": 15.549667202913941,
      "learning_rate": 1.449948712349854e-07,
      "loss": 0.6741,
      "step": 3752
    },
    {
      "epoch": 0.3710056100635148,
      "grad_norm": 2.93119167222858,
      "learning_rate": 1.4496627232148448e-07,
      "loss": 0.7107,
      "step": 3753
    },
    {
      "epoch": 0.3711044658082693,
      "grad_norm": 3.910226743062934,
      "learning_rate": 1.4493766879735567e-07,
      "loss": 0.8058,
      "step": 3754
    },
    {
      "epoch": 0.37120332155302377,
      "grad_norm": 4.6729889080985085,
      "learning_rate": 1.4490906066553178e-07,
      "loss": 0.6704,
      "step": 3755
    },
    {
      "epoch": 0.3713021772977782,
      "grad_norm": 4.693994670370568,
      "learning_rate": 1.4488044792894616e-07,
      "loss": 0.7837,
      "step": 3756
    },
    {
      "epoch": 0.3714010330425327,
      "grad_norm": 7.645498796116748,
      "learning_rate": 1.4485183059053268e-07,
      "loss": 0.7126,
      "step": 3757
    },
    {
      "epoch": 0.3714998887872871,
      "grad_norm": 3.7355431191151087,
      "learning_rate": 1.4482320865322555e-07,
      "loss": 0.8309,
      "step": 3758
    },
    {
      "epoch": 0.3715987445320416,
      "grad_norm": 4.263149997398231,
      "learning_rate": 1.447945821199596e-07,
      "loss": 0.7383,
      "step": 3759
    },
    {
      "epoch": 0.3716976002767961,
      "grad_norm": 4.945275260844403,
      "learning_rate": 1.4476595099366996e-07,
      "loss": 0.6729,
      "step": 3760
    },
    {
      "epoch": 0.37179645602155054,
      "grad_norm": 2.965647778574568,
      "learning_rate": 1.4473731527729242e-07,
      "loss": 0.7692,
      "step": 3761
    },
    {
      "epoch": 0.371895311766305,
      "grad_norm": 3.589069200099517,
      "learning_rate": 1.4470867497376314e-07,
      "loss": 0.6518,
      "step": 3762
    },
    {
      "epoch": 0.37199416751105946,
      "grad_norm": 2.8176045052626906,
      "learning_rate": 1.446800300860187e-07,
      "loss": 0.6101,
      "step": 3763
    },
    {
      "epoch": 0.37209302325581395,
      "grad_norm": 4.775632209173637,
      "learning_rate": 1.4465138061699626e-07,
      "loss": 0.679,
      "step": 3764
    },
    {
      "epoch": 0.37219187900056844,
      "grad_norm": 6.11156094451087,
      "learning_rate": 1.446227265696334e-07,
      "loss": 0.7105,
      "step": 3765
    },
    {
      "epoch": 0.37229073474532287,
      "grad_norm": 3.681850942029265,
      "learning_rate": 1.445940679468682e-07,
      "loss": 0.6859,
      "step": 3766
    },
    {
      "epoch": 0.37238959049007736,
      "grad_norm": 8.294940893741206,
      "learning_rate": 1.445654047516391e-07,
      "loss": 0.8124,
      "step": 3767
    },
    {
      "epoch": 0.37248844623483185,
      "grad_norm": 8.448794220248097,
      "learning_rate": 1.4453673698688514e-07,
      "loss": 0.667,
      "step": 3768
    },
    {
      "epoch": 0.3725873019795863,
      "grad_norm": 3.5678598371590344,
      "learning_rate": 1.4450806465554578e-07,
      "loss": 0.8021,
      "step": 3769
    },
    {
      "epoch": 0.37268615772434077,
      "grad_norm": 12.27858298265465,
      "learning_rate": 1.444793877605609e-07,
      "loss": 0.738,
      "step": 3770
    },
    {
      "epoch": 0.3727850134690952,
      "grad_norm": 4.403874511710134,
      "learning_rate": 1.444507063048709e-07,
      "loss": 0.8104,
      "step": 3771
    },
    {
      "epoch": 0.3728838692138497,
      "grad_norm": 4.882687293341145,
      "learning_rate": 1.4442202029141672e-07,
      "loss": 0.706,
      "step": 3772
    },
    {
      "epoch": 0.3729827249586042,
      "grad_norm": 6.787625511297325,
      "learning_rate": 1.443933297231396e-07,
      "loss": 0.7099,
      "step": 3773
    },
    {
      "epoch": 0.3730815807033586,
      "grad_norm": 7.967015031593697,
      "learning_rate": 1.4436463460298137e-07,
      "loss": 0.6684,
      "step": 3774
    },
    {
      "epoch": 0.3731804364481131,
      "grad_norm": 4.069003265275342,
      "learning_rate": 1.443359349338843e-07,
      "loss": 0.5601,
      "step": 3775
    },
    {
      "epoch": 0.37327929219286754,
      "grad_norm": 4.097953672965826,
      "learning_rate": 1.4430723071879108e-07,
      "loss": 0.6401,
      "step": 3776
    },
    {
      "epoch": 0.37337814793762203,
      "grad_norm": 4.791118977843389,
      "learning_rate": 1.4427852196064497e-07,
      "loss": 0.6876,
      "step": 3777
    },
    {
      "epoch": 0.3734770036823765,
      "grad_norm": 4.290826416435919,
      "learning_rate": 1.442498086623895e-07,
      "loss": 0.7287,
      "step": 3778
    },
    {
      "epoch": 0.37357585942713095,
      "grad_norm": 5.940333148719334,
      "learning_rate": 1.4422109082696897e-07,
      "loss": 0.7537,
      "step": 3779
    },
    {
      "epoch": 0.37367471517188544,
      "grad_norm": 3.094113599599272,
      "learning_rate": 1.4419236845732784e-07,
      "loss": 0.6326,
      "step": 3780
    },
    {
      "epoch": 0.3737735709166399,
      "grad_norm": 5.330764104455922,
      "learning_rate": 1.4416364155641122e-07,
      "loss": 0.7224,
      "step": 3781
    },
    {
      "epoch": 0.37387242666139436,
      "grad_norm": 3.5990684127346055,
      "learning_rate": 1.4413491012716458e-07,
      "loss": 0.6924,
      "step": 3782
    },
    {
      "epoch": 0.37397128240614885,
      "grad_norm": 5.406752879912292,
      "learning_rate": 1.4410617417253402e-07,
      "loss": 0.6825,
      "step": 3783
    },
    {
      "epoch": 0.3740701381509033,
      "grad_norm": 5.420243330807034,
      "learning_rate": 1.440774336954659e-07,
      "loss": 0.6311,
      "step": 3784
    },
    {
      "epoch": 0.3741689938956578,
      "grad_norm": 5.5675126859551085,
      "learning_rate": 1.4404868869890708e-07,
      "loss": 0.7887,
      "step": 3785
    },
    {
      "epoch": 0.3742678496404122,
      "grad_norm": 2.9830545145334755,
      "learning_rate": 1.4401993918580502e-07,
      "loss": 0.6898,
      "step": 3786
    },
    {
      "epoch": 0.3743667053851667,
      "grad_norm": 7.870631468791329,
      "learning_rate": 1.4399118515910756e-07,
      "loss": 0.6379,
      "step": 3787
    },
    {
      "epoch": 0.3744655611299212,
      "grad_norm": 4.431501225348336,
      "learning_rate": 1.4396242662176293e-07,
      "loss": 0.616,
      "step": 3788
    },
    {
      "epoch": 0.3745644168746756,
      "grad_norm": 5.975140565413247,
      "learning_rate": 1.4393366357671998e-07,
      "loss": 0.6488,
      "step": 3789
    },
    {
      "epoch": 0.3746632726194301,
      "grad_norm": 63.10675229816202,
      "learning_rate": 1.439048960269279e-07,
      "loss": 0.6558,
      "step": 3790
    },
    {
      "epoch": 0.37476212836418454,
      "grad_norm": 5.286250300182555,
      "learning_rate": 1.4387612397533637e-07,
      "loss": 0.7036,
      "step": 3791
    },
    {
      "epoch": 0.37486098410893903,
      "grad_norm": 2.909815980190188,
      "learning_rate": 1.4384734742489552e-07,
      "loss": 0.7135,
      "step": 3792
    },
    {
      "epoch": 0.3749598398536935,
      "grad_norm": 4.425359406861969,
      "learning_rate": 1.4381856637855603e-07,
      "loss": 0.6967,
      "step": 3793
    },
    {
      "epoch": 0.37505869559844796,
      "grad_norm": 5.1839098216375685,
      "learning_rate": 1.4378978083926893e-07,
      "loss": 0.6665,
      "step": 3794
    },
    {
      "epoch": 0.37515755134320244,
      "grad_norm": 3.3778017450646924,
      "learning_rate": 1.437609908099857e-07,
      "loss": 0.712,
      "step": 3795
    },
    {
      "epoch": 0.3752564070879569,
      "grad_norm": 3.8425475184270863,
      "learning_rate": 1.437321962936584e-07,
      "loss": 0.7607,
      "step": 3796
    },
    {
      "epoch": 0.37535526283271137,
      "grad_norm": 4.01863783061278,
      "learning_rate": 1.4370339729323946e-07,
      "loss": 0.7005,
      "step": 3797
    },
    {
      "epoch": 0.37545411857746586,
      "grad_norm": 7.890354412861929,
      "learning_rate": 1.4367459381168182e-07,
      "loss": 0.8143,
      "step": 3798
    },
    {
      "epoch": 0.3755529743222203,
      "grad_norm": 4.110045165196139,
      "learning_rate": 1.4364578585193883e-07,
      "loss": 0.812,
      "step": 3799
    },
    {
      "epoch": 0.3756518300669748,
      "grad_norm": 27.235236490852454,
      "learning_rate": 1.436169734169643e-07,
      "loss": 0.7559,
      "step": 3800
    },
    {
      "epoch": 0.3757506858117292,
      "grad_norm": 3.6101353361572754,
      "learning_rate": 1.4358815650971257e-07,
      "loss": 0.6582,
      "step": 3801
    },
    {
      "epoch": 0.3758495415564837,
      "grad_norm": 3.391699666442935,
      "learning_rate": 1.4355933513313833e-07,
      "loss": 0.6901,
      "step": 3802
    },
    {
      "epoch": 0.3759483973012382,
      "grad_norm": 4.048960003955002,
      "learning_rate": 1.4353050929019683e-07,
      "loss": 0.7612,
      "step": 3803
    },
    {
      "epoch": 0.3760472530459926,
      "grad_norm": 15.161078818077417,
      "learning_rate": 1.435016789838437e-07,
      "loss": 0.6033,
      "step": 3804
    },
    {
      "epoch": 0.3761461087907471,
      "grad_norm": 5.775847519567743,
      "learning_rate": 1.4347284421703511e-07,
      "loss": 0.771,
      "step": 3805
    },
    {
      "epoch": 0.37624496453550155,
      "grad_norm": 4.092839562617768,
      "learning_rate": 1.4344400499272763e-07,
      "loss": 0.728,
      "step": 3806
    },
    {
      "epoch": 0.37634382028025604,
      "grad_norm": 5.619419211931012,
      "learning_rate": 1.4341516131387824e-07,
      "loss": 0.7355,
      "step": 3807
    },
    {
      "epoch": 0.3764426760250105,
      "grad_norm": 3.5534246430558625,
      "learning_rate": 1.4338631318344453e-07,
      "loss": 0.6202,
      "step": 3808
    },
    {
      "epoch": 0.37654153176976496,
      "grad_norm": 3.858523324137024,
      "learning_rate": 1.4335746060438435e-07,
      "loss": 0.6827,
      "step": 3809
    },
    {
      "epoch": 0.37664038751451945,
      "grad_norm": 4.460448716039743,
      "learning_rate": 1.4332860357965619e-07,
      "loss": 0.7385,
      "step": 3810
    },
    {
      "epoch": 0.3767392432592739,
      "grad_norm": 3.4059069343345425,
      "learning_rate": 1.4329974211221885e-07,
      "loss": 0.7945,
      "step": 3811
    },
    {
      "epoch": 0.37683809900402837,
      "grad_norm": 3.1883169024925575,
      "learning_rate": 1.432708762050317e-07,
      "loss": 0.7677,
      "step": 3812
    },
    {
      "epoch": 0.37693695474878286,
      "grad_norm": 13.597756300761619,
      "learning_rate": 1.4324200586105444e-07,
      "loss": 0.7181,
      "step": 3813
    },
    {
      "epoch": 0.3770358104935373,
      "grad_norm": 7.344170018859479,
      "learning_rate": 1.4321313108324738e-07,
      "loss": 0.7277,
      "step": 3814
    },
    {
      "epoch": 0.3771346662382918,
      "grad_norm": 3.8690125318804376,
      "learning_rate": 1.4318425187457113e-07,
      "loss": 0.8051,
      "step": 3815
    },
    {
      "epoch": 0.3772335219830462,
      "grad_norm": 6.8656182320992825,
      "learning_rate": 1.4315536823798688e-07,
      "loss": 0.645,
      "step": 3816
    },
    {
      "epoch": 0.3773323777278007,
      "grad_norm": 6.167892756188442,
      "learning_rate": 1.4312648017645623e-07,
      "loss": 0.7771,
      "step": 3817
    },
    {
      "epoch": 0.3774312334725552,
      "grad_norm": 2.9291366154317644,
      "learning_rate": 1.4309758769294118e-07,
      "loss": 0.8183,
      "step": 3818
    },
    {
      "epoch": 0.37753008921730963,
      "grad_norm": 4.940510717736193,
      "learning_rate": 1.4306869079040425e-07,
      "loss": 0.7444,
      "step": 3819
    },
    {
      "epoch": 0.3776289449620641,
      "grad_norm": 3.909903448154188,
      "learning_rate": 1.4303978947180838e-07,
      "loss": 0.8454,
      "step": 3820
    },
    {
      "epoch": 0.37772780070681855,
      "grad_norm": 5.798104209610484,
      "learning_rate": 1.43010883740117e-07,
      "loss": 0.6903,
      "step": 3821
    },
    {
      "epoch": 0.37782665645157304,
      "grad_norm": 5.0191492644571,
      "learning_rate": 1.429819735982939e-07,
      "loss": 0.7647,
      "step": 3822
    },
    {
      "epoch": 0.37792551219632753,
      "grad_norm": 5.567695285222633,
      "learning_rate": 1.4295305904930346e-07,
      "loss": 0.743,
      "step": 3823
    },
    {
      "epoch": 0.37802436794108196,
      "grad_norm": 13.61106204393925,
      "learning_rate": 1.4292414009611041e-07,
      "loss": 0.7008,
      "step": 3824
    },
    {
      "epoch": 0.37812322368583645,
      "grad_norm": 4.848277654323795,
      "learning_rate": 1.4289521674168002e-07,
      "loss": 0.7036,
      "step": 3825
    },
    {
      "epoch": 0.3782220794305909,
      "grad_norm": 5.559587479072044,
      "learning_rate": 1.4286628898897784e-07,
      "loss": 0.8188,
      "step": 3826
    },
    {
      "epoch": 0.3783209351753454,
      "grad_norm": 5.93350931400336,
      "learning_rate": 1.4283735684097006e-07,
      "loss": 0.6989,
      "step": 3827
    },
    {
      "epoch": 0.37841979092009986,
      "grad_norm": 5.167998164283585,
      "learning_rate": 1.4280842030062327e-07,
      "loss": 0.6228,
      "step": 3828
    },
    {
      "epoch": 0.3785186466648543,
      "grad_norm": 7.604535306903161,
      "learning_rate": 1.427794793709044e-07,
      "loss": 0.6167,
      "step": 3829
    },
    {
      "epoch": 0.3786175024096088,
      "grad_norm": 3.8850834831101686,
      "learning_rate": 1.42750534054781e-07,
      "loss": 0.6653,
      "step": 3830
    },
    {
      "epoch": 0.3787163581543632,
      "grad_norm": 4.94632752981876,
      "learning_rate": 1.427215843552209e-07,
      "loss": 0.8099,
      "step": 3831
    },
    {
      "epoch": 0.3788152138991177,
      "grad_norm": 5.148228607709109,
      "learning_rate": 1.4269263027519256e-07,
      "loss": 0.7997,
      "step": 3832
    },
    {
      "epoch": 0.3789140696438722,
      "grad_norm": 3.4266682944272144,
      "learning_rate": 1.426636718176647e-07,
      "loss": 0.6893,
      "step": 3833
    },
    {
      "epoch": 0.37901292538862663,
      "grad_norm": 3.578283355351983,
      "learning_rate": 1.4263470898560664e-07,
      "loss": 0.7252,
      "step": 3834
    },
    {
      "epoch": 0.3791117811333811,
      "grad_norm": 4.165948236262646,
      "learning_rate": 1.426057417819881e-07,
      "loss": 0.7352,
      "step": 3835
    },
    {
      "epoch": 0.37921063687813555,
      "grad_norm": 6.488945808433723,
      "learning_rate": 1.4257677020977923e-07,
      "loss": 0.6822,
      "step": 3836
    },
    {
      "epoch": 0.37930949262289004,
      "grad_norm": 5.869491630806906,
      "learning_rate": 1.4254779427195063e-07,
      "loss": 0.7428,
      "step": 3837
    },
    {
      "epoch": 0.37940834836764453,
      "grad_norm": 6.442818607011678,
      "learning_rate": 1.4251881397147333e-07,
      "loss": 0.6585,
      "step": 3838
    },
    {
      "epoch": 0.37950720411239897,
      "grad_norm": 9.751769721421384,
      "learning_rate": 1.4248982931131885e-07,
      "loss": 0.6731,
      "step": 3839
    },
    {
      "epoch": 0.37960605985715346,
      "grad_norm": 4.618944956068225,
      "learning_rate": 1.4246084029445912e-07,
      "loss": 0.7327,
      "step": 3840
    },
    {
      "epoch": 0.3797049156019079,
      "grad_norm": 3.41188228045909,
      "learning_rate": 1.424318469238666e-07,
      "loss": 0.626,
      "step": 3841
    },
    {
      "epoch": 0.3798037713466624,
      "grad_norm": 3.852348390250874,
      "learning_rate": 1.4240284920251408e-07,
      "loss": 0.7286,
      "step": 3842
    },
    {
      "epoch": 0.37990262709141687,
      "grad_norm": 2.6976706310354697,
      "learning_rate": 1.4237384713337487e-07,
      "loss": 0.6587,
      "step": 3843
    },
    {
      "epoch": 0.3800014828361713,
      "grad_norm": 9.913093535044265,
      "learning_rate": 1.4234484071942267e-07,
      "loss": 0.8326,
      "step": 3844
    },
    {
      "epoch": 0.3801003385809258,
      "grad_norm": 4.670423523445183,
      "learning_rate": 1.4231582996363172e-07,
      "loss": 0.7012,
      "step": 3845
    },
    {
      "epoch": 0.3801991943256803,
      "grad_norm": 4.175514023360776,
      "learning_rate": 1.422868148689766e-07,
      "loss": 0.8026,
      "step": 3846
    },
    {
      "epoch": 0.3802980500704347,
      "grad_norm": 7.0416949971028835,
      "learning_rate": 1.4225779543843238e-07,
      "loss": 0.6571,
      "step": 3847
    },
    {
      "epoch": 0.3803969058151892,
      "grad_norm": 12.12374785406397,
      "learning_rate": 1.4222877167497455e-07,
      "loss": 0.7351,
      "step": 3848
    },
    {
      "epoch": 0.38049576155994363,
      "grad_norm": 4.7629493375642635,
      "learning_rate": 1.4219974358157916e-07,
      "loss": 0.7598,
      "step": 3849
    },
    {
      "epoch": 0.3805946173046981,
      "grad_norm": 4.5845521145546435,
      "learning_rate": 1.4217071116122252e-07,
      "loss": 0.7479,
      "step": 3850
    },
    {
      "epoch": 0.3806934730494526,
      "grad_norm": 5.154982854945801,
      "learning_rate": 1.421416744168815e-07,
      "loss": 0.6295,
      "step": 3851
    },
    {
      "epoch": 0.38079232879420705,
      "grad_norm": 6.2234535929846295,
      "learning_rate": 1.4211263335153343e-07,
      "loss": 0.7651,
      "step": 3852
    },
    {
      "epoch": 0.38089118453896154,
      "grad_norm": 4.994264675027666,
      "learning_rate": 1.4208358796815604e-07,
      "loss": 0.7437,
      "step": 3853
    },
    {
      "epoch": 0.38099004028371597,
      "grad_norm": 4.264777765053465,
      "learning_rate": 1.4205453826972745e-07,
      "loss": 0.7204,
      "step": 3854
    },
    {
      "epoch": 0.38108889602847046,
      "grad_norm": 3.687976008759733,
      "learning_rate": 1.420254842592263e-07,
      "loss": 0.5996,
      "step": 3855
    },
    {
      "epoch": 0.38118775177322495,
      "grad_norm": 4.029233603643152,
      "learning_rate": 1.4199642593963165e-07,
      "loss": 0.6611,
      "step": 3856
    },
    {
      "epoch": 0.3812866075179794,
      "grad_norm": 3.7569657525175657,
      "learning_rate": 1.4196736331392305e-07,
      "loss": 0.7859,
      "step": 3857
    },
    {
      "epoch": 0.38138546326273387,
      "grad_norm": 6.142873067158661,
      "learning_rate": 1.4193829638508037e-07,
      "loss": 0.7227,
      "step": 3858
    },
    {
      "epoch": 0.3814843190074883,
      "grad_norm": 5.03968909949774,
      "learning_rate": 1.4190922515608403e-07,
      "loss": 0.7299,
      "step": 3859
    },
    {
      "epoch": 0.3815831747522428,
      "grad_norm": 3.088592517274486,
      "learning_rate": 1.4188014962991488e-07,
      "loss": 0.7936,
      "step": 3860
    },
    {
      "epoch": 0.3816820304969973,
      "grad_norm": 3.6823715092175604,
      "learning_rate": 1.4185106980955415e-07,
      "loss": 0.6891,
      "step": 3861
    },
    {
      "epoch": 0.3817808862417517,
      "grad_norm": 4.523899212077057,
      "learning_rate": 1.4182198569798356e-07,
      "loss": 0.6506,
      "step": 3862
    },
    {
      "epoch": 0.3818797419865062,
      "grad_norm": 16.59667579842722,
      "learning_rate": 1.4179289729818525e-07,
      "loss": 0.7576,
      "step": 3863
    },
    {
      "epoch": 0.38197859773126064,
      "grad_norm": 3.825091570713382,
      "learning_rate": 1.4176380461314182e-07,
      "loss": 0.8613,
      "step": 3864
    },
    {
      "epoch": 0.3820774534760151,
      "grad_norm": 4.744488221725254,
      "learning_rate": 1.4173470764583627e-07,
      "loss": 0.7042,
      "step": 3865
    },
    {
      "epoch": 0.3821763092207696,
      "grad_norm": 6.452028072293079,
      "learning_rate": 1.4170560639925206e-07,
      "loss": 0.795,
      "step": 3866
    },
    {
      "epoch": 0.38227516496552405,
      "grad_norm": 4.170030550702709,
      "learning_rate": 1.4167650087637315e-07,
      "loss": 0.6372,
      "step": 3867
    },
    {
      "epoch": 0.38237402071027854,
      "grad_norm": 3.485443253154045,
      "learning_rate": 1.4164739108018386e-07,
      "loss": 0.6965,
      "step": 3868
    },
    {
      "epoch": 0.382472876455033,
      "grad_norm": 4.390355101053253,
      "learning_rate": 1.4161827701366893e-07,
      "loss": 0.7278,
      "step": 3869
    },
    {
      "epoch": 0.38257173219978746,
      "grad_norm": 5.295728553849626,
      "learning_rate": 1.415891586798136e-07,
      "loss": 0.688,
      "step": 3870
    },
    {
      "epoch": 0.38267058794454195,
      "grad_norm": 4.361576217936964,
      "learning_rate": 1.4156003608160353e-07,
      "loss": 0.8103,
      "step": 3871
    },
    {
      "epoch": 0.3827694436892964,
      "grad_norm": 4.679320363031833,
      "learning_rate": 1.4153090922202487e-07,
      "loss": 0.6994,
      "step": 3872
    },
    {
      "epoch": 0.3828682994340509,
      "grad_norm": 4.571448121916102,
      "learning_rate": 1.4150177810406404e-07,
      "loss": 0.7441,
      "step": 3873
    },
    {
      "epoch": 0.3829671551788053,
      "grad_norm": 7.111254056398886,
      "learning_rate": 1.4147264273070807e-07,
      "loss": 0.6669,
      "step": 3874
    },
    {
      "epoch": 0.3830660109235598,
      "grad_norm": 7.424086766704459,
      "learning_rate": 1.4144350310494434e-07,
      "loss": 0.7524,
      "step": 3875
    },
    {
      "epoch": 0.3831648666683143,
      "grad_norm": 3.439820533549137,
      "learning_rate": 1.414143592297607e-07,
      "loss": 0.8642,
      "step": 3876
    },
    {
      "epoch": 0.3832637224130687,
      "grad_norm": 6.693289072750498,
      "learning_rate": 1.4138521110814542e-07,
      "loss": 0.7834,
      "step": 3877
    },
    {
      "epoch": 0.3833625781578232,
      "grad_norm": 6.924196556687613,
      "learning_rate": 1.4135605874308725e-07,
      "loss": 0.7548,
      "step": 3878
    },
    {
      "epoch": 0.38346143390257764,
      "grad_norm": 2.672586463706325,
      "learning_rate": 1.413269021375753e-07,
      "loss": 0.6576,
      "step": 3879
    },
    {
      "epoch": 0.38356028964733213,
      "grad_norm": 5.085894950379209,
      "learning_rate": 1.4129774129459913e-07,
      "loss": 0.6437,
      "step": 3880
    },
    {
      "epoch": 0.3836591453920866,
      "grad_norm": 5.232923287911695,
      "learning_rate": 1.4126857621714877e-07,
      "loss": 0.7026,
      "step": 3881
    },
    {
      "epoch": 0.38375800113684105,
      "grad_norm": 3.6845415196924596,
      "learning_rate": 1.412394069082147e-07,
      "loss": 0.694,
      "step": 3882
    },
    {
      "epoch": 0.38385685688159554,
      "grad_norm": 4.140800181908264,
      "learning_rate": 1.4121023337078778e-07,
      "loss": 0.707,
      "step": 3883
    },
    {
      "epoch": 0.38395571262635,
      "grad_norm": 3.2287612543673543,
      "learning_rate": 1.4118105560785926e-07,
      "loss": 0.6521,
      "step": 3884
    },
    {
      "epoch": 0.38405456837110447,
      "grad_norm": 3.4739187737755106,
      "learning_rate": 1.41151873622421e-07,
      "loss": 0.7371,
      "step": 3885
    },
    {
      "epoch": 0.38415342411585895,
      "grad_norm": 3.84977434803961,
      "learning_rate": 1.4112268741746513e-07,
      "loss": 0.7405,
      "step": 3886
    },
    {
      "epoch": 0.3842522798606134,
      "grad_norm": 3.9681373614406827,
      "learning_rate": 1.4109349699598428e-07,
      "loss": 0.6825,
      "step": 3887
    },
    {
      "epoch": 0.3843511356053679,
      "grad_norm": 4.884763885652891,
      "learning_rate": 1.4106430236097144e-07,
      "loss": 0.7312,
      "step": 3888
    },
    {
      "epoch": 0.3844499913501223,
      "grad_norm": 3.5926853310490996,
      "learning_rate": 1.4103510351542018e-07,
      "loss": 0.7544,
      "step": 3889
    },
    {
      "epoch": 0.3845488470948768,
      "grad_norm": 4.677172552829473,
      "learning_rate": 1.4100590046232434e-07,
      "loss": 0.68,
      "step": 3890
    },
    {
      "epoch": 0.3846477028396313,
      "grad_norm": 7.634700466142064,
      "learning_rate": 1.4097669320467832e-07,
      "loss": 0.8012,
      "step": 3891
    },
    {
      "epoch": 0.3847465585843857,
      "grad_norm": 3.0767658583021844,
      "learning_rate": 1.409474817454768e-07,
      "loss": 0.6991,
      "step": 3892
    },
    {
      "epoch": 0.3848454143291402,
      "grad_norm": 6.974124381910248,
      "learning_rate": 1.4091826608771506e-07,
      "loss": 0.8201,
      "step": 3893
    },
    {
      "epoch": 0.38494427007389465,
      "grad_norm": 6.349902884995338,
      "learning_rate": 1.4088904623438875e-07,
      "loss": 0.6555,
      "step": 3894
    },
    {
      "epoch": 0.38504312581864913,
      "grad_norm": 9.851481321510652,
      "learning_rate": 1.408598221884939e-07,
      "loss": 0.7219,
      "step": 3895
    },
    {
      "epoch": 0.3851419815634036,
      "grad_norm": 4.702174022580679,
      "learning_rate": 1.4083059395302698e-07,
      "loss": 0.7198,
      "step": 3896
    },
    {
      "epoch": 0.38524083730815806,
      "grad_norm": 4.7163923415536155,
      "learning_rate": 1.4080136153098497e-07,
      "loss": 0.7643,
      "step": 3897
    },
    {
      "epoch": 0.38533969305291255,
      "grad_norm": 5.507335081019142,
      "learning_rate": 1.4077212492536522e-07,
      "loss": 0.7426,
      "step": 3898
    },
    {
      "epoch": 0.385438548797667,
      "grad_norm": 4.56634335411123,
      "learning_rate": 1.4074288413916543e-07,
      "loss": 0.7708,
      "step": 3899
    },
    {
      "epoch": 0.38553740454242147,
      "grad_norm": 3.125313607395516,
      "learning_rate": 1.4071363917538393e-07,
      "loss": 0.6577,
      "step": 3900
    },
    {
      "epoch": 0.38563626028717596,
      "grad_norm": 3.445990557425034,
      "learning_rate": 1.4068439003701929e-07,
      "loss": 0.7158,
      "step": 3901
    },
    {
      "epoch": 0.3857351160319304,
      "grad_norm": 4.546979720419309,
      "learning_rate": 1.4065513672707056e-07,
      "loss": 0.7741,
      "step": 3902
    },
    {
      "epoch": 0.3858339717766849,
      "grad_norm": 3.8098091416924875,
      "learning_rate": 1.4062587924853727e-07,
      "loss": 0.7902,
      "step": 3903
    },
    {
      "epoch": 0.3859328275214393,
      "grad_norm": 3.7955486472633813,
      "learning_rate": 1.4059661760441935e-07,
      "loss": 0.7642,
      "step": 3904
    },
    {
      "epoch": 0.3860316832661938,
      "grad_norm": 3.762734986400821,
      "learning_rate": 1.4056735179771712e-07,
      "loss": 0.6855,
      "step": 3905
    },
    {
      "epoch": 0.3861305390109483,
      "grad_norm": 3.5483592976486995,
      "learning_rate": 1.405380818314314e-07,
      "loss": 0.7162,
      "step": 3906
    },
    {
      "epoch": 0.3862293947557027,
      "grad_norm": 3.3902630681461012,
      "learning_rate": 1.4050880770856336e-07,
      "loss": 0.6429,
      "step": 3907
    },
    {
      "epoch": 0.3863282505004572,
      "grad_norm": 5.675720849112474,
      "learning_rate": 1.4047952943211464e-07,
      "loss": 0.6679,
      "step": 3908
    },
    {
      "epoch": 0.38642710624521165,
      "grad_norm": 6.069390012547412,
      "learning_rate": 1.4045024700508727e-07,
      "loss": 0.7184,
      "step": 3909
    },
    {
      "epoch": 0.38652596198996614,
      "grad_norm": 12.039207879403753,
      "learning_rate": 1.4042096043048374e-07,
      "loss": 0.7248,
      "step": 3910
    },
    {
      "epoch": 0.3866248177347206,
      "grad_norm": 3.9312315834379707,
      "learning_rate": 1.40391669711307e-07,
      "loss": 0.7291,
      "step": 3911
    },
    {
      "epoch": 0.38672367347947506,
      "grad_norm": 6.990248611373263,
      "learning_rate": 1.4036237485056034e-07,
      "loss": 0.6484,
      "step": 3912
    },
    {
      "epoch": 0.38682252922422955,
      "grad_norm": 3.3698788327852003,
      "learning_rate": 1.4033307585124754e-07,
      "loss": 0.7032,
      "step": 3913
    },
    {
      "epoch": 0.386921384968984,
      "grad_norm": 3.7472316664794145,
      "learning_rate": 1.4030377271637272e-07,
      "loss": 0.7491,
      "step": 3914
    },
    {
      "epoch": 0.3870202407137385,
      "grad_norm": 3.8988200847559122,
      "learning_rate": 1.4027446544894059e-07,
      "loss": 0.7178,
      "step": 3915
    },
    {
      "epoch": 0.38711909645849296,
      "grad_norm": 4.032716587987853,
      "learning_rate": 1.4024515405195613e-07,
      "loss": 0.7225,
      "step": 3916
    },
    {
      "epoch": 0.3872179522032474,
      "grad_norm": 9.150499973970511,
      "learning_rate": 1.4021583852842472e-07,
      "loss": 0.7027,
      "step": 3917
    },
    {
      "epoch": 0.3873168079480019,
      "grad_norm": 7.444709494176636,
      "learning_rate": 1.4018651888135232e-07,
      "loss": 0.7695,
      "step": 3918
    },
    {
      "epoch": 0.3874156636927563,
      "grad_norm": 4.799775726204708,
      "learning_rate": 1.401571951137452e-07,
      "loss": 0.7079,
      "step": 3919
    },
    {
      "epoch": 0.3875145194375108,
      "grad_norm": 3.3752479496614063,
      "learning_rate": 1.4012786722861008e-07,
      "loss": 0.6128,
      "step": 3920
    },
    {
      "epoch": 0.3876133751822653,
      "grad_norm": 5.675649348640341,
      "learning_rate": 1.4009853522895406e-07,
      "loss": 0.6008,
      "step": 3921
    },
    {
      "epoch": 0.38771223092701973,
      "grad_norm": 4.001744964228146,
      "learning_rate": 1.4006919911778482e-07,
      "loss": 0.7433,
      "step": 3922
    },
    {
      "epoch": 0.3878110866717742,
      "grad_norm": 4.183878413131544,
      "learning_rate": 1.4003985889811024e-07,
      "loss": 0.7659,
      "step": 3923
    },
    {
      "epoch": 0.3879099424165287,
      "grad_norm": 3.687098778421048,
      "learning_rate": 1.4001051457293878e-07,
      "loss": 0.7972,
      "step": 3924
    },
    {
      "epoch": 0.38800879816128314,
      "grad_norm": 4.35359288643428,
      "learning_rate": 1.399811661452792e-07,
      "loss": 0.7175,
      "step": 3925
    },
    {
      "epoch": 0.38810765390603763,
      "grad_norm": 3.953724201032734,
      "learning_rate": 1.3995181361814087e-07,
      "loss": 0.6272,
      "step": 3926
    },
    {
      "epoch": 0.38820650965079206,
      "grad_norm": 5.584745759158036,
      "learning_rate": 1.3992245699453335e-07,
      "loss": 0.7085,
      "step": 3927
    },
    {
      "epoch": 0.38830536539554655,
      "grad_norm": 4.516091129561202,
      "learning_rate": 1.398930962774667e-07,
      "loss": 0.7604,
      "step": 3928
    },
    {
      "epoch": 0.38840422114030104,
      "grad_norm": 100.33700115759795,
      "learning_rate": 1.3986373146995155e-07,
      "loss": 0.8592,
      "step": 3929
    },
    {
      "epoch": 0.3885030768850555,
      "grad_norm": 3.7630701700223685,
      "learning_rate": 1.3983436257499878e-07,
      "loss": 0.7146,
      "step": 3930
    },
    {
      "epoch": 0.38860193262980997,
      "grad_norm": 11.40197766691148,
      "learning_rate": 1.398049895956197e-07,
      "loss": 0.8082,
      "step": 3931
    },
    {
      "epoch": 0.3887007883745644,
      "grad_norm": 4.502894994983002,
      "learning_rate": 1.397756125348261e-07,
      "loss": 0.5928,
      "step": 3932
    },
    {
      "epoch": 0.3887996441193189,
      "grad_norm": 3.0854630212166447,
      "learning_rate": 1.397462313956302e-07,
      "loss": 0.7407,
      "step": 3933
    },
    {
      "epoch": 0.3888984998640734,
      "grad_norm": 3.536696016464864,
      "learning_rate": 1.3971684618104456e-07,
      "loss": 0.8516,
      "step": 3934
    },
    {
      "epoch": 0.3889973556088278,
      "grad_norm": 3.448537403137319,
      "learning_rate": 1.396874568940822e-07,
      "loss": 0.7414,
      "step": 3935
    },
    {
      "epoch": 0.3890962113535823,
      "grad_norm": 3.614189885706666,
      "learning_rate": 1.3965806353775653e-07,
      "loss": 0.6138,
      "step": 3936
    },
    {
      "epoch": 0.38919506709833673,
      "grad_norm": 22.539773619085196,
      "learning_rate": 1.396286661150815e-07,
      "loss": 0.8686,
      "step": 3937
    },
    {
      "epoch": 0.3892939228430912,
      "grad_norm": 3.4897583064533597,
      "learning_rate": 1.395992646290713e-07,
      "loss": 0.7075,
      "step": 3938
    },
    {
      "epoch": 0.3893927785878457,
      "grad_norm": 5.3200450108196975,
      "learning_rate": 1.395698590827406e-07,
      "loss": 0.6451,
      "step": 3939
    },
    {
      "epoch": 0.38949163433260015,
      "grad_norm": 4.036642057361084,
      "learning_rate": 1.395404494791046e-07,
      "loss": 0.7224,
      "step": 3940
    },
    {
      "epoch": 0.38959049007735463,
      "grad_norm": 16.502682297018513,
      "learning_rate": 1.3951103582117875e-07,
      "loss": 0.8234,
      "step": 3941
    },
    {
      "epoch": 0.38968934582210907,
      "grad_norm": 3.5072238205148256,
      "learning_rate": 1.3948161811197902e-07,
      "loss": 0.7998,
      "step": 3942
    },
    {
      "epoch": 0.38978820156686356,
      "grad_norm": 3.9304980286823037,
      "learning_rate": 1.3945219635452174e-07,
      "loss": 0.6395,
      "step": 3943
    },
    {
      "epoch": 0.38988705731161805,
      "grad_norm": 3.3061010208115356,
      "learning_rate": 1.394227705518237e-07,
      "loss": 0.7203,
      "step": 3944
    },
    {
      "epoch": 0.3899859130563725,
      "grad_norm": 3.365491894419096,
      "learning_rate": 1.3939334070690204e-07,
      "loss": 0.7977,
      "step": 3945
    },
    {
      "epoch": 0.39008476880112697,
      "grad_norm": 3.5060786997470266,
      "learning_rate": 1.393639068227744e-07,
      "loss": 0.657,
      "step": 3946
    },
    {
      "epoch": 0.3901836245458814,
      "grad_norm": 4.078831332801902,
      "learning_rate": 1.3933446890245876e-07,
      "loss": 0.6437,
      "step": 3947
    },
    {
      "epoch": 0.3902824802906359,
      "grad_norm": 10.377797885712024,
      "learning_rate": 1.3930502694897358e-07,
      "loss": 0.7615,
      "step": 3948
    },
    {
      "epoch": 0.3903813360353904,
      "grad_norm": 8.11413733549631,
      "learning_rate": 1.3927558096533764e-07,
      "loss": 0.7592,
      "step": 3949
    },
    {
      "epoch": 0.3904801917801448,
      "grad_norm": 9.152498073936858,
      "learning_rate": 1.3924613095457026e-07,
      "loss": 0.6687,
      "step": 3950
    },
    {
      "epoch": 0.3905790475248993,
      "grad_norm": 2.9845396965966198,
      "learning_rate": 1.3921667691969105e-07,
      "loss": 0.7113,
      "step": 3951
    },
    {
      "epoch": 0.39067790326965374,
      "grad_norm": 42.97942356262272,
      "learning_rate": 1.3918721886372014e-07,
      "loss": 0.7648,
      "step": 3952
    },
    {
      "epoch": 0.3907767590144082,
      "grad_norm": 3.0252759197155648,
      "learning_rate": 1.3915775678967796e-07,
      "loss": 0.7353,
      "step": 3953
    },
    {
      "epoch": 0.3908756147591627,
      "grad_norm": 3.236379253890259,
      "learning_rate": 1.3912829070058544e-07,
      "loss": 0.7718,
      "step": 3954
    },
    {
      "epoch": 0.39097447050391715,
      "grad_norm": 4.935773505157943,
      "learning_rate": 1.390988205994639e-07,
      "loss": 0.7087,
      "step": 3955
    },
    {
      "epoch": 0.39107332624867164,
      "grad_norm": 3.5016825679238086,
      "learning_rate": 1.3906934648933506e-07,
      "loss": 0.7452,
      "step": 3956
    },
    {
      "epoch": 0.39117218199342607,
      "grad_norm": 10.021434670579003,
      "learning_rate": 1.3903986837322106e-07,
      "loss": 0.6824,
      "step": 3957
    },
    {
      "epoch": 0.39127103773818056,
      "grad_norm": 3.6357160874283,
      "learning_rate": 1.390103862541444e-07,
      "loss": 0.7224,
      "step": 3958
    },
    {
      "epoch": 0.39136989348293505,
      "grad_norm": 6.162404678703686,
      "learning_rate": 1.389809001351281e-07,
      "loss": 0.6583,
      "step": 3959
    },
    {
      "epoch": 0.3914687492276895,
      "grad_norm": 3.4819435228178945,
      "learning_rate": 1.389514100191955e-07,
      "loss": 0.7436,
      "step": 3960
    },
    {
      "epoch": 0.39156760497244397,
      "grad_norm": 3.260804084695177,
      "learning_rate": 1.3892191590937038e-07,
      "loss": 0.6883,
      "step": 3961
    },
    {
      "epoch": 0.3916664607171984,
      "grad_norm": 4.5596200940155605,
      "learning_rate": 1.3889241780867691e-07,
      "loss": 0.6893,
      "step": 3962
    },
    {
      "epoch": 0.3917653164619529,
      "grad_norm": 3.87831445921358,
      "learning_rate": 1.388629157201397e-07,
      "loss": 0.6883,
      "step": 3963
    },
    {
      "epoch": 0.3918641722067074,
      "grad_norm": 7.221312059468625,
      "learning_rate": 1.3883340964678376e-07,
      "loss": 0.6418,
      "step": 3964
    },
    {
      "epoch": 0.3919630279514618,
      "grad_norm": 3.5948307170376257,
      "learning_rate": 1.3880389959163444e-07,
      "loss": 0.7604,
      "step": 3965
    },
    {
      "epoch": 0.3920618836962163,
      "grad_norm": 3.3372034685319263,
      "learning_rate": 1.387743855577177e-07,
      "loss": 0.8553,
      "step": 3966
    },
    {
      "epoch": 0.39216073944097074,
      "grad_norm": 2.843255974140833,
      "learning_rate": 1.3874486754805963e-07,
      "loss": 0.7611,
      "step": 3967
    },
    {
      "epoch": 0.39225959518572523,
      "grad_norm": 3.7761235293642277,
      "learning_rate": 1.3871534556568695e-07,
      "loss": 0.8032,
      "step": 3968
    },
    {
      "epoch": 0.3923584509304797,
      "grad_norm": 14.213313193347721,
      "learning_rate": 1.3868581961362665e-07,
      "loss": 0.6968,
      "step": 3969
    },
    {
      "epoch": 0.39245730667523415,
      "grad_norm": 6.757362479401202,
      "learning_rate": 1.386562896949062e-07,
      "loss": 0.6831,
      "step": 3970
    },
    {
      "epoch": 0.39255616241998864,
      "grad_norm": 15.427731334849755,
      "learning_rate": 1.3862675581255346e-07,
      "loss": 0.7974,
      "step": 3971
    },
    {
      "epoch": 0.3926550181647431,
      "grad_norm": 6.753785678257378,
      "learning_rate": 1.385972179695967e-07,
      "loss": 0.6699,
      "step": 3972
    },
    {
      "epoch": 0.39275387390949756,
      "grad_norm": 3.118284784028652,
      "learning_rate": 1.3856767616906462e-07,
      "loss": 0.7338,
      "step": 3973
    },
    {
      "epoch": 0.39285272965425205,
      "grad_norm": 8.566998372639103,
      "learning_rate": 1.3853813041398623e-07,
      "loss": 0.7152,
      "step": 3974
    },
    {
      "epoch": 0.3929515853990065,
      "grad_norm": 3.3351316661012604,
      "learning_rate": 1.3850858070739107e-07,
      "loss": 0.8157,
      "step": 3975
    },
    {
      "epoch": 0.393050441143761,
      "grad_norm": 5.219607904892859,
      "learning_rate": 1.38479027052309e-07,
      "loss": 0.6135,
      "step": 3976
    },
    {
      "epoch": 0.3931492968885154,
      "grad_norm": 3.604949472845347,
      "learning_rate": 1.3844946945177027e-07,
      "loss": 0.6143,
      "step": 3977
    },
    {
      "epoch": 0.3932481526332699,
      "grad_norm": 8.020503338586627,
      "learning_rate": 1.3841990790880566e-07,
      "loss": 0.6776,
      "step": 3978
    },
    {
      "epoch": 0.3933470083780244,
      "grad_norm": 3.630927745504876,
      "learning_rate": 1.383903424264462e-07,
      "loss": 0.7986,
      "step": 3979
    },
    {
      "epoch": 0.3934458641227788,
      "grad_norm": 3.1814130525572066,
      "learning_rate": 1.383607730077234e-07,
      "loss": 0.7367,
      "step": 3980
    },
    {
      "epoch": 0.3935447198675333,
      "grad_norm": 3.3592782351311667,
      "learning_rate": 1.3833119965566923e-07,
      "loss": 0.6911,
      "step": 3981
    },
    {
      "epoch": 0.39364357561228774,
      "grad_norm": 9.981549653414024,
      "learning_rate": 1.3830162237331595e-07,
      "loss": 0.7729,
      "step": 3982
    },
    {
      "epoch": 0.39374243135704223,
      "grad_norm": 4.355782260320361,
      "learning_rate": 1.382720411636963e-07,
      "loss": 0.7506,
      "step": 3983
    },
    {
      "epoch": 0.3938412871017967,
      "grad_norm": 6.914409616613802,
      "learning_rate": 1.3824245602984334e-07,
      "loss": 0.784,
      "step": 3984
    },
    {
      "epoch": 0.39394014284655116,
      "grad_norm": 12.074656431495715,
      "learning_rate": 1.3821286697479067e-07,
      "loss": 0.7456,
      "step": 3985
    },
    {
      "epoch": 0.39403899859130564,
      "grad_norm": 25.201140030482893,
      "learning_rate": 1.3818327400157215e-07,
      "loss": 0.7682,
      "step": 3986
    },
    {
      "epoch": 0.3941378543360601,
      "grad_norm": 3.4357751250433433,
      "learning_rate": 1.3815367711322215e-07,
      "loss": 0.7077,
      "step": 3987
    },
    {
      "epoch": 0.39423671008081457,
      "grad_norm": 20.46404228847062,
      "learning_rate": 1.3812407631277534e-07,
      "loss": 0.7113,
      "step": 3988
    },
    {
      "epoch": 0.39433556582556906,
      "grad_norm": 4.257612710271797,
      "learning_rate": 1.3809447160326688e-07,
      "loss": 0.6631,
      "step": 3989
    },
    {
      "epoch": 0.3944344215703235,
      "grad_norm": 3.8656963509398987,
      "learning_rate": 1.3806486298773228e-07,
      "loss": 0.653,
      "step": 3990
    },
    {
      "epoch": 0.394533277315078,
      "grad_norm": 3.751104890093559,
      "learning_rate": 1.380352504692075e-07,
      "loss": 0.7484,
      "step": 3991
    },
    {
      "epoch": 0.3946321330598324,
      "grad_norm": 4.740090446677816,
      "learning_rate": 1.3800563405072882e-07,
      "loss": 0.6134,
      "step": 3992
    },
    {
      "epoch": 0.3947309888045869,
      "grad_norm": 4.869731284667213,
      "learning_rate": 1.37976013735333e-07,
      "loss": 0.7987,
      "step": 3993
    },
    {
      "epoch": 0.3948298445493414,
      "grad_norm": 10.245722370497834,
      "learning_rate": 1.3794638952605716e-07,
      "loss": 0.8081,
      "step": 3994
    },
    {
      "epoch": 0.3949287002940958,
      "grad_norm": 3.6480969213401706,
      "learning_rate": 1.3791676142593884e-07,
      "loss": 0.753,
      "step": 3995
    },
    {
      "epoch": 0.3950275560388503,
      "grad_norm": 3.45310251226282,
      "learning_rate": 1.3788712943801594e-07,
      "loss": 0.7191,
      "step": 3996
    },
    {
      "epoch": 0.39512641178360475,
      "grad_norm": 7.080359996086882,
      "learning_rate": 1.3785749356532679e-07,
      "loss": 0.7597,
      "step": 3997
    },
    {
      "epoch": 0.39522526752835924,
      "grad_norm": 5.7720015767131425,
      "learning_rate": 1.378278538109101e-07,
      "loss": 0.8329,
      "step": 3998
    },
    {
      "epoch": 0.3953241232731137,
      "grad_norm": 3.8972610892052835,
      "learning_rate": 1.37798210177805e-07,
      "loss": 0.7919,
      "step": 3999
    },
    {
      "epoch": 0.39542297901786816,
      "grad_norm": 3.7600148325388005,
      "learning_rate": 1.3776856266905102e-07,
      "loss": 0.7614,
      "step": 4000
    },
    {
      "epoch": 0.39552183476262265,
      "grad_norm": 3.83611771663307,
      "learning_rate": 1.3773891128768803e-07,
      "loss": 0.7161,
      "step": 4001
    },
    {
      "epoch": 0.39562069050737714,
      "grad_norm": 5.4779545353004915,
      "learning_rate": 1.377092560367564e-07,
      "loss": 0.7545,
      "step": 4002
    },
    {
      "epoch": 0.39571954625213157,
      "grad_norm": 3.9287786967075475,
      "learning_rate": 1.3767959691929684e-07,
      "loss": 0.7248,
      "step": 4003
    },
    {
      "epoch": 0.39581840199688606,
      "grad_norm": 2.474771019161882,
      "learning_rate": 1.3764993393835042e-07,
      "loss": 0.7199,
      "step": 4004
    },
    {
      "epoch": 0.3959172577416405,
      "grad_norm": 3.782543037957009,
      "learning_rate": 1.3762026709695862e-07,
      "loss": 0.6962,
      "step": 4005
    },
    {
      "epoch": 0.396016113486395,
      "grad_norm": 7.107575656085507,
      "learning_rate": 1.375905963981634e-07,
      "loss": 0.6905,
      "step": 4006
    },
    {
      "epoch": 0.39611496923114947,
      "grad_norm": 4.404630015438051,
      "learning_rate": 1.3756092184500698e-07,
      "loss": 0.7111,
      "step": 4007
    },
    {
      "epoch": 0.3962138249759039,
      "grad_norm": 3.4650878363545603,
      "learning_rate": 1.3753124344053212e-07,
      "loss": 0.6405,
      "step": 4008
    },
    {
      "epoch": 0.3963126807206584,
      "grad_norm": 7.227359937431192,
      "learning_rate": 1.3750156118778184e-07,
      "loss": 0.7142,
      "step": 4009
    },
    {
      "epoch": 0.39641153646541283,
      "grad_norm": 5.7591457990245285,
      "learning_rate": 1.3747187508979968e-07,
      "loss": 0.7305,
      "step": 4010
    },
    {
      "epoch": 0.3965103922101673,
      "grad_norm": 3.6305039026454726,
      "learning_rate": 1.3744218514962946e-07,
      "loss": 0.6313,
      "step": 4011
    },
    {
      "epoch": 0.3966092479549218,
      "grad_norm": 4.206663973658014,
      "learning_rate": 1.374124913703155e-07,
      "loss": 0.7875,
      "step": 4012
    },
    {
      "epoch": 0.39670810369967624,
      "grad_norm": 5.397436446944782,
      "learning_rate": 1.373827937549024e-07,
      "loss": 0.7624,
      "step": 4013
    },
    {
      "epoch": 0.39680695944443073,
      "grad_norm": 3.321113422746335,
      "learning_rate": 1.3735309230643523e-07,
      "loss": 0.7111,
      "step": 4014
    },
    {
      "epoch": 0.39690581518918516,
      "grad_norm": 3.1702087209170395,
      "learning_rate": 1.3732338702795944e-07,
      "loss": 0.7555,
      "step": 4015
    },
    {
      "epoch": 0.39700467093393965,
      "grad_norm": 3.8134528210713254,
      "learning_rate": 1.3729367792252085e-07,
      "loss": 0.73,
      "step": 4016
    },
    {
      "epoch": 0.39710352667869414,
      "grad_norm": 2.960023740250979,
      "learning_rate": 1.3726396499316572e-07,
      "loss": 0.7241,
      "step": 4017
    },
    {
      "epoch": 0.3972023824234486,
      "grad_norm": 11.6207427374832,
      "learning_rate": 1.3723424824294067e-07,
      "loss": 0.7334,
      "step": 4018
    },
    {
      "epoch": 0.39730123816820306,
      "grad_norm": 4.626121052753708,
      "learning_rate": 1.372045276748927e-07,
      "loss": 0.6166,
      "step": 4019
    },
    {
      "epoch": 0.3974000939129575,
      "grad_norm": 8.614831745943947,
      "learning_rate": 1.3717480329206925e-07,
      "loss": 0.6984,
      "step": 4020
    },
    {
      "epoch": 0.397498949657712,
      "grad_norm": 10.262874150053017,
      "learning_rate": 1.371450750975181e-07,
      "loss": 0.7176,
      "step": 4021
    },
    {
      "epoch": 0.3975978054024665,
      "grad_norm": 4.137836423560795,
      "learning_rate": 1.371153430942874e-07,
      "loss": 0.7064,
      "step": 4022
    },
    {
      "epoch": 0.3976966611472209,
      "grad_norm": 4.588145825341124,
      "learning_rate": 1.3708560728542572e-07,
      "loss": 0.7565,
      "step": 4023
    },
    {
      "epoch": 0.3977955168919754,
      "grad_norm": 4.0464427444575835,
      "learning_rate": 1.370558676739821e-07,
      "loss": 0.706,
      "step": 4024
    },
    {
      "epoch": 0.39789437263672983,
      "grad_norm": 4.732773842949324,
      "learning_rate": 1.3702612426300587e-07,
      "loss": 0.748,
      "step": 4025
    },
    {
      "epoch": 0.3979932283814843,
      "grad_norm": 7.278629800390711,
      "learning_rate": 1.3699637705554674e-07,
      "loss": 0.6705,
      "step": 4026
    },
    {
      "epoch": 0.3980920841262388,
      "grad_norm": 4.246837129154639,
      "learning_rate": 1.3696662605465489e-07,
      "loss": 0.7201,
      "step": 4027
    },
    {
      "epoch": 0.39819093987099324,
      "grad_norm": 4.344085865751359,
      "learning_rate": 1.3693687126338086e-07,
      "loss": 0.7068,
      "step": 4028
    },
    {
      "epoch": 0.39828979561574773,
      "grad_norm": 3.1712231808989078,
      "learning_rate": 1.369071126847755e-07,
      "loss": 0.5849,
      "step": 4029
    },
    {
      "epoch": 0.39838865136050217,
      "grad_norm": 6.578610152910971,
      "learning_rate": 1.3687735032189022e-07,
      "loss": 0.5786,
      "step": 4030
    },
    {
      "epoch": 0.39848750710525666,
      "grad_norm": 8.284407053384417,
      "learning_rate": 1.3684758417777655e-07,
      "loss": 0.6598,
      "step": 4031
    },
    {
      "epoch": 0.39858636285001114,
      "grad_norm": 4.523601075982932,
      "learning_rate": 1.3681781425548672e-07,
      "loss": 0.6711,
      "step": 4032
    },
    {
      "epoch": 0.3986852185947656,
      "grad_norm": 3.914255274973054,
      "learning_rate": 1.3678804055807314e-07,
      "loss": 0.7347,
      "step": 4033
    },
    {
      "epoch": 0.39878407433952007,
      "grad_norm": 5.629800474897567,
      "learning_rate": 1.3675826308858862e-07,
      "loss": 0.6336,
      "step": 4034
    },
    {
      "epoch": 0.3988829300842745,
      "grad_norm": 3.6712606184518557,
      "learning_rate": 1.367284818500865e-07,
      "loss": 0.7386,
      "step": 4035
    },
    {
      "epoch": 0.398981785829029,
      "grad_norm": 3.64098127686032,
      "learning_rate": 1.3669869684562032e-07,
      "loss": 0.6257,
      "step": 4036
    },
    {
      "epoch": 0.3990806415737835,
      "grad_norm": 4.869767129871034,
      "learning_rate": 1.3666890807824411e-07,
      "loss": 0.6784,
      "step": 4037
    },
    {
      "epoch": 0.3991794973185379,
      "grad_norm": 5.796957005290569,
      "learning_rate": 1.366391155510123e-07,
      "loss": 0.7015,
      "step": 4038
    },
    {
      "epoch": 0.3992783530632924,
      "grad_norm": 2.8894908798086756,
      "learning_rate": 1.3660931926697967e-07,
      "loss": 0.643,
      "step": 4039
    },
    {
      "epoch": 0.39937720880804684,
      "grad_norm": 10.104397118841547,
      "learning_rate": 1.3657951922920137e-07,
      "loss": 0.761,
      "step": 4040
    },
    {
      "epoch": 0.3994760645528013,
      "grad_norm": 3.339828871181091,
      "learning_rate": 1.3654971544073293e-07,
      "loss": 0.6703,
      "step": 4041
    },
    {
      "epoch": 0.3995749202975558,
      "grad_norm": 3.4697652639366945,
      "learning_rate": 1.3651990790463033e-07,
      "loss": 0.7497,
      "step": 4042
    },
    {
      "epoch": 0.39967377604231025,
      "grad_norm": 3.915813820660208,
      "learning_rate": 1.3649009662394988e-07,
      "loss": 0.6328,
      "step": 4043
    },
    {
      "epoch": 0.39977263178706474,
      "grad_norm": 7.040194932655014,
      "learning_rate": 1.364602816017483e-07,
      "loss": 0.6508,
      "step": 4044
    },
    {
      "epoch": 0.39987148753181917,
      "grad_norm": 4.980836509271625,
      "learning_rate": 1.3643046284108264e-07,
      "loss": 0.6762,
      "step": 4045
    },
    {
      "epoch": 0.39997034327657366,
      "grad_norm": 2.8818179378940694,
      "learning_rate": 1.364006403450104e-07,
      "loss": 0.5983,
      "step": 4046
    },
    {
      "epoch": 0.40006919902132815,
      "grad_norm": 23.09596715590435,
      "learning_rate": 1.3637081411658952e-07,
      "loss": 0.7155,
      "step": 4047
    },
    {
      "epoch": 0.4001680547660826,
      "grad_norm": 8.286993696904812,
      "learning_rate": 1.363409841588781e-07,
      "loss": 0.8009,
      "step": 4048
    },
    {
      "epoch": 0.40026691051083707,
      "grad_norm": 4.652889838141533,
      "learning_rate": 1.363111504749348e-07,
      "loss": 0.7487,
      "step": 4049
    },
    {
      "epoch": 0.4003657662555915,
      "grad_norm": 6.903087202743209,
      "learning_rate": 1.3628131306781867e-07,
      "loss": 0.7661,
      "step": 4050
    },
    {
      "epoch": 0.400464622000346,
      "grad_norm": 5.200728879720409,
      "learning_rate": 1.3625147194058907e-07,
      "loss": 0.6334,
      "step": 4051
    },
    {
      "epoch": 0.4005634777451005,
      "grad_norm": 3.757789092384292,
      "learning_rate": 1.3622162709630578e-07,
      "loss": 0.7543,
      "step": 4052
    },
    {
      "epoch": 0.4006623334898549,
      "grad_norm": 3.2770684079798107,
      "learning_rate": 1.3619177853802892e-07,
      "loss": 0.7857,
      "step": 4053
    },
    {
      "epoch": 0.4007611892346094,
      "grad_norm": 7.21133612661471,
      "learning_rate": 1.3616192626881903e-07,
      "loss": 0.6842,
      "step": 4054
    },
    {
      "epoch": 0.40086004497936384,
      "grad_norm": 4.8346287612410315,
      "learning_rate": 1.3613207029173708e-07,
      "loss": 0.8077,
      "step": 4055
    },
    {
      "epoch": 0.40095890072411833,
      "grad_norm": 3.8399678086675073,
      "learning_rate": 1.3610221060984427e-07,
      "loss": 0.6734,
      "step": 4056
    },
    {
      "epoch": 0.4010577564688728,
      "grad_norm": 4.1244589780258964,
      "learning_rate": 1.360723472262023e-07,
      "loss": 0.7141,
      "step": 4057
    },
    {
      "epoch": 0.40115661221362725,
      "grad_norm": 5.697798411548719,
      "learning_rate": 1.3604248014387325e-07,
      "loss": 0.6845,
      "step": 4058
    },
    {
      "epoch": 0.40125546795838174,
      "grad_norm": 4.19376393613447,
      "learning_rate": 1.3601260936591953e-07,
      "loss": 0.8519,
      "step": 4059
    },
    {
      "epoch": 0.4013543237031362,
      "grad_norm": 7.1413700997838125,
      "learning_rate": 1.3598273489540387e-07,
      "loss": 0.7024,
      "step": 4060
    },
    {
      "epoch": 0.40145317944789066,
      "grad_norm": 3.5292365905340652,
      "learning_rate": 1.359528567353896e-07,
      "loss": 0.6432,
      "step": 4061
    },
    {
      "epoch": 0.40155203519264515,
      "grad_norm": 3.639278164546402,
      "learning_rate": 1.359229748889402e-07,
      "loss": 0.646,
      "step": 4062
    },
    {
      "epoch": 0.4016508909373996,
      "grad_norm": 4.058273282652866,
      "learning_rate": 1.3589308935911962e-07,
      "loss": 0.7576,
      "step": 4063
    },
    {
      "epoch": 0.4017497466821541,
      "grad_norm": 5.452381484054536,
      "learning_rate": 1.3586320014899218e-07,
      "loss": 0.7471,
      "step": 4064
    },
    {
      "epoch": 0.4018486024269085,
      "grad_norm": 3.3475648135181864,
      "learning_rate": 1.3583330726162261e-07,
      "loss": 0.7348,
      "step": 4065
    },
    {
      "epoch": 0.401947458171663,
      "grad_norm": 3.4670129777800502,
      "learning_rate": 1.3580341070007595e-07,
      "loss": 0.6516,
      "step": 4066
    },
    {
      "epoch": 0.4020463139164175,
      "grad_norm": 4.129596210117518,
      "learning_rate": 1.3577351046741764e-07,
      "loss": 0.7515,
      "step": 4067
    },
    {
      "epoch": 0.4021451696611719,
      "grad_norm": 6.543429753887918,
      "learning_rate": 1.3574360656671354e-07,
      "loss": 0.7206,
      "step": 4068
    },
    {
      "epoch": 0.4022440254059264,
      "grad_norm": 8.77206435341458,
      "learning_rate": 1.3571369900102985e-07,
      "loss": 0.6811,
      "step": 4069
    },
    {
      "epoch": 0.40234288115068084,
      "grad_norm": 7.86506050713617,
      "learning_rate": 1.3568378777343312e-07,
      "loss": 0.6892,
      "step": 4070
    },
    {
      "epoch": 0.40244173689543533,
      "grad_norm": 4.201738300196104,
      "learning_rate": 1.3565387288699034e-07,
      "loss": 0.8339,
      "step": 4071
    },
    {
      "epoch": 0.4025405926401898,
      "grad_norm": 16.745038618492913,
      "learning_rate": 1.3562395434476882e-07,
      "loss": 0.6584,
      "step": 4072
    },
    {
      "epoch": 0.40263944838494425,
      "grad_norm": 4.952864354211317,
      "learning_rate": 1.355940321498363e-07,
      "loss": 0.8108,
      "step": 4073
    },
    {
      "epoch": 0.40273830412969874,
      "grad_norm": 7.873730708754862,
      "learning_rate": 1.3556410630526082e-07,
      "loss": 0.8158,
      "step": 4074
    },
    {
      "epoch": 0.4028371598744532,
      "grad_norm": 4.270952692994384,
      "learning_rate": 1.3553417681411082e-07,
      "loss": 0.6584,
      "step": 4075
    },
    {
      "epoch": 0.40293601561920767,
      "grad_norm": 6.700923925057391,
      "learning_rate": 1.3550424367945516e-07,
      "loss": 0.7669,
      "step": 4076
    },
    {
      "epoch": 0.40303487136396215,
      "grad_norm": 3.61847847510114,
      "learning_rate": 1.3547430690436306e-07,
      "loss": 0.8079,
      "step": 4077
    },
    {
      "epoch": 0.4031337271087166,
      "grad_norm": 4.238468069781882,
      "learning_rate": 1.3544436649190404e-07,
      "loss": 0.7151,
      "step": 4078
    },
    {
      "epoch": 0.4032325828534711,
      "grad_norm": 9.920337783018105,
      "learning_rate": 1.354144224451481e-07,
      "loss": 0.7161,
      "step": 4079
    },
    {
      "epoch": 0.40333143859822557,
      "grad_norm": 5.006999764593267,
      "learning_rate": 1.353844747671655e-07,
      "loss": 0.6878,
      "step": 4080
    },
    {
      "epoch": 0.40343029434298,
      "grad_norm": 3.707862341029652,
      "learning_rate": 1.3535452346102704e-07,
      "loss": 0.634,
      "step": 4081
    },
    {
      "epoch": 0.4035291500877345,
      "grad_norm": 4.544735110547847,
      "learning_rate": 1.3532456852980367e-07,
      "loss": 0.7088,
      "step": 4082
    },
    {
      "epoch": 0.4036280058324889,
      "grad_norm": 3.684116725939991,
      "learning_rate": 1.3529460997656687e-07,
      "loss": 0.6868,
      "step": 4083
    },
    {
      "epoch": 0.4037268615772434,
      "grad_norm": 4.8156656307100585,
      "learning_rate": 1.3526464780438847e-07,
      "loss": 0.6944,
      "step": 4084
    },
    {
      "epoch": 0.4038257173219979,
      "grad_norm": 4.524529139159837,
      "learning_rate": 1.3523468201634063e-07,
      "loss": 0.7991,
      "step": 4085
    },
    {
      "epoch": 0.40392457306675233,
      "grad_norm": 4.479873756069666,
      "learning_rate": 1.3520471261549587e-07,
      "loss": 0.6985,
      "step": 4086
    },
    {
      "epoch": 0.4040234288115068,
      "grad_norm": 3.8440799451718974,
      "learning_rate": 1.3517473960492713e-07,
      "loss": 0.7253,
      "step": 4087
    },
    {
      "epoch": 0.40412228455626126,
      "grad_norm": 6.466647349400415,
      "learning_rate": 1.3514476298770775e-07,
      "loss": 0.7151,
      "step": 4088
    },
    {
      "epoch": 0.40422114030101575,
      "grad_norm": 4.40723970182832,
      "learning_rate": 1.3511478276691129e-07,
      "loss": 0.7045,
      "step": 4089
    },
    {
      "epoch": 0.40431999604577024,
      "grad_norm": 6.855370722068972,
      "learning_rate": 1.3508479894561186e-07,
      "loss": 0.6716,
      "step": 4090
    },
    {
      "epoch": 0.40441885179052467,
      "grad_norm": 3.817123356586155,
      "learning_rate": 1.350548115268839e-07,
      "loss": 0.6711,
      "step": 4091
    },
    {
      "epoch": 0.40451770753527916,
      "grad_norm": 5.378601781530161,
      "learning_rate": 1.3502482051380205e-07,
      "loss": 0.7868,
      "step": 4092
    },
    {
      "epoch": 0.4046165632800336,
      "grad_norm": 4.0811280150210125,
      "learning_rate": 1.3499482590944151e-07,
      "loss": 0.7287,
      "step": 4093
    },
    {
      "epoch": 0.4047154190247881,
      "grad_norm": 3.0062025758898603,
      "learning_rate": 1.3496482771687778e-07,
      "loss": 0.7027,
      "step": 4094
    },
    {
      "epoch": 0.40481427476954257,
      "grad_norm": 6.154310808808824,
      "learning_rate": 1.3493482593918675e-07,
      "loss": 0.723,
      "step": 4095
    },
    {
      "epoch": 0.404913130514297,
      "grad_norm": 10.423995926458753,
      "learning_rate": 1.3490482057944466e-07,
      "loss": 0.6942,
      "step": 4096
    },
    {
      "epoch": 0.4050119862590515,
      "grad_norm": 4.048513288377183,
      "learning_rate": 1.3487481164072804e-07,
      "loss": 0.6855,
      "step": 4097
    },
    {
      "epoch": 0.4051108420038059,
      "grad_norm": 11.404783134408174,
      "learning_rate": 1.34844799126114e-07,
      "loss": 0.7447,
      "step": 4098
    },
    {
      "epoch": 0.4052096977485604,
      "grad_norm": 3.392279360141884,
      "learning_rate": 1.3481478303867977e-07,
      "loss": 0.8255,
      "step": 4099
    },
    {
      "epoch": 0.4053085534933149,
      "grad_norm": 8.122762073996684,
      "learning_rate": 1.347847633815031e-07,
      "loss": 0.7659,
      "step": 4100
    },
    {
      "epoch": 0.40540740923806934,
      "grad_norm": 5.649219548964488,
      "learning_rate": 1.3475474015766203e-07,
      "loss": 0.8197,
      "step": 4101
    },
    {
      "epoch": 0.4055062649828238,
      "grad_norm": 13.352487300658492,
      "learning_rate": 1.3472471337023509e-07,
      "loss": 0.678,
      "step": 4102
    },
    {
      "epoch": 0.40560512072757826,
      "grad_norm": 4.240576012296398,
      "learning_rate": 1.34694683022301e-07,
      "loss": 0.7466,
      "step": 4103
    },
    {
      "epoch": 0.40570397647233275,
      "grad_norm": 4.677013368174527,
      "learning_rate": 1.346646491169389e-07,
      "loss": 0.6441,
      "step": 4104
    },
    {
      "epoch": 0.40580283221708724,
      "grad_norm": 7.575260837194892,
      "learning_rate": 1.3463461165722838e-07,
      "loss": 0.6424,
      "step": 4105
    },
    {
      "epoch": 0.4059016879618417,
      "grad_norm": 6.4835821572479375,
      "learning_rate": 1.3460457064624935e-07,
      "loss": 0.8052,
      "step": 4106
    },
    {
      "epoch": 0.40600054370659616,
      "grad_norm": 5.130605481987227,
      "learning_rate": 1.3457452608708205e-07,
      "loss": 0.8062,
      "step": 4107
    },
    {
      "epoch": 0.4060993994513506,
      "grad_norm": 4.810238829632351,
      "learning_rate": 1.3454447798280714e-07,
      "loss": 0.6795,
      "step": 4108
    },
    {
      "epoch": 0.4061982551961051,
      "grad_norm": 5.136336157373226,
      "learning_rate": 1.3451442633650556e-07,
      "loss": 0.6812,
      "step": 4109
    },
    {
      "epoch": 0.4062971109408596,
      "grad_norm": 4.271319149052712,
      "learning_rate": 1.3448437115125867e-07,
      "loss": 0.663,
      "step": 4110
    },
    {
      "epoch": 0.406395966685614,
      "grad_norm": 5.700904386598441,
      "learning_rate": 1.344543124301482e-07,
      "loss": 0.7088,
      "step": 4111
    },
    {
      "epoch": 0.4064948224303685,
      "grad_norm": 4.362687832835221,
      "learning_rate": 1.3442425017625625e-07,
      "loss": 0.6827,
      "step": 4112
    },
    {
      "epoch": 0.40659367817512293,
      "grad_norm": 5.08607591473625,
      "learning_rate": 1.3439418439266522e-07,
      "loss": 0.7235,
      "step": 4113
    },
    {
      "epoch": 0.4066925339198774,
      "grad_norm": 4.155751201111846,
      "learning_rate": 1.3436411508245793e-07,
      "loss": 0.6815,
      "step": 4114
    },
    {
      "epoch": 0.4067913896646319,
      "grad_norm": 4.563187309179184,
      "learning_rate": 1.3433404224871755e-07,
      "loss": 0.7036,
      "step": 4115
    },
    {
      "epoch": 0.40689024540938634,
      "grad_norm": 4.577381662375172,
      "learning_rate": 1.3430396589452758e-07,
      "loss": 0.658,
      "step": 4116
    },
    {
      "epoch": 0.40698910115414083,
      "grad_norm": 11.547868482356213,
      "learning_rate": 1.3427388602297194e-07,
      "loss": 0.7641,
      "step": 4117
    },
    {
      "epoch": 0.40708795689889526,
      "grad_norm": 4.662895221904019,
      "learning_rate": 1.3424380263713492e-07,
      "loss": 0.6339,
      "step": 4118
    },
    {
      "epoch": 0.40718681264364975,
      "grad_norm": 6.004413533624086,
      "learning_rate": 1.3421371574010095e-07,
      "loss": 0.6937,
      "step": 4119
    },
    {
      "epoch": 0.40728566838840424,
      "grad_norm": 4.788561945038308,
      "learning_rate": 1.3418362533495523e-07,
      "loss": 0.8257,
      "step": 4120
    },
    {
      "epoch": 0.4073845241331587,
      "grad_norm": 7.3666903479737655,
      "learning_rate": 1.341535314247829e-07,
      "loss": 0.7065,
      "step": 4121
    },
    {
      "epoch": 0.40748337987791317,
      "grad_norm": 3.315968921654255,
      "learning_rate": 1.3412343401266976e-07,
      "loss": 0.7846,
      "step": 4122
    },
    {
      "epoch": 0.4075822356226676,
      "grad_norm": 4.154920525651394,
      "learning_rate": 1.3409333310170182e-07,
      "loss": 0.7613,
      "step": 4123
    },
    {
      "epoch": 0.4076810913674221,
      "grad_norm": 5.482760758365575,
      "learning_rate": 1.340632286949655e-07,
      "loss": 0.7424,
      "step": 4124
    },
    {
      "epoch": 0.4077799471121766,
      "grad_norm": 3.2306197677240664,
      "learning_rate": 1.3403312079554748e-07,
      "loss": 0.644,
      "step": 4125
    },
    {
      "epoch": 0.407878802856931,
      "grad_norm": 3.657253134504103,
      "learning_rate": 1.3400300940653502e-07,
      "loss": 0.6532,
      "step": 4126
    },
    {
      "epoch": 0.4079776586016855,
      "grad_norm": 3.7194324967367365,
      "learning_rate": 1.339728945310155e-07,
      "loss": 0.6315,
      "step": 4127
    },
    {
      "epoch": 0.40807651434643993,
      "grad_norm": 3.416240958479376,
      "learning_rate": 1.3394277617207676e-07,
      "loss": 0.6678,
      "step": 4128
    },
    {
      "epoch": 0.4081753700911944,
      "grad_norm": 4.311140956824454,
      "learning_rate": 1.3391265433280703e-07,
      "loss": 0.7883,
      "step": 4129
    },
    {
      "epoch": 0.4082742258359489,
      "grad_norm": 4.033071796817395,
      "learning_rate": 1.3388252901629484e-07,
      "loss": 0.6322,
      "step": 4130
    },
    {
      "epoch": 0.40837308158070335,
      "grad_norm": 4.828479300138106,
      "learning_rate": 1.338524002256291e-07,
      "loss": 0.7148,
      "step": 4131
    },
    {
      "epoch": 0.40847193732545783,
      "grad_norm": 5.0262323223595855,
      "learning_rate": 1.3382226796389907e-07,
      "loss": 0.6838,
      "step": 4132
    },
    {
      "epoch": 0.40857079307021227,
      "grad_norm": 10.372692847096229,
      "learning_rate": 1.3379213223419438e-07,
      "loss": 0.7517,
      "step": 4133
    },
    {
      "epoch": 0.40866964881496676,
      "grad_norm": 4.7193611174939845,
      "learning_rate": 1.3376199303960502e-07,
      "loss": 0.7886,
      "step": 4134
    },
    {
      "epoch": 0.40876850455972125,
      "grad_norm": 9.388815139161405,
      "learning_rate": 1.337318503832213e-07,
      "loss": 0.6944,
      "step": 4135
    },
    {
      "epoch": 0.4088673603044757,
      "grad_norm": 5.296882716483853,
      "learning_rate": 1.3370170426813386e-07,
      "loss": 0.7117,
      "step": 4136
    },
    {
      "epoch": 0.40896621604923017,
      "grad_norm": 3.7576753793180364,
      "learning_rate": 1.336715546974338e-07,
      "loss": 0.6974,
      "step": 4137
    },
    {
      "epoch": 0.4090650717939846,
      "grad_norm": 3.4600104326380148,
      "learning_rate": 1.3364140167421252e-07,
      "loss": 0.6996,
      "step": 4138
    },
    {
      "epoch": 0.4091639275387391,
      "grad_norm": 3.578149951697119,
      "learning_rate": 1.3361124520156174e-07,
      "loss": 0.6494,
      "step": 4139
    },
    {
      "epoch": 0.4092627832834936,
      "grad_norm": 4.124512486794874,
      "learning_rate": 1.3358108528257355e-07,
      "loss": 0.7094,
      "step": 4140
    },
    {
      "epoch": 0.409361639028248,
      "grad_norm": 3.7909296040839564,
      "learning_rate": 1.3355092192034037e-07,
      "loss": 0.6693,
      "step": 4141
    },
    {
      "epoch": 0.4094604947730025,
      "grad_norm": 10.027111051379237,
      "learning_rate": 1.3352075511795514e-07,
      "loss": 0.672,
      "step": 4142
    },
    {
      "epoch": 0.40955935051775694,
      "grad_norm": 4.184900457188137,
      "learning_rate": 1.3349058487851093e-07,
      "loss": 0.6926,
      "step": 4143
    },
    {
      "epoch": 0.4096582062625114,
      "grad_norm": 3.816461379806068,
      "learning_rate": 1.3346041120510123e-07,
      "loss": 0.6854,
      "step": 4144
    },
    {
      "epoch": 0.4097570620072659,
      "grad_norm": 3.138289445685196,
      "learning_rate": 1.3343023410081995e-07,
      "loss": 0.7398,
      "step": 4145
    },
    {
      "epoch": 0.40985591775202035,
      "grad_norm": 6.667259519513072,
      "learning_rate": 1.3340005356876129e-07,
      "loss": 0.6979,
      "step": 4146
    },
    {
      "epoch": 0.40995477349677484,
      "grad_norm": 5.063314824366961,
      "learning_rate": 1.3336986961201983e-07,
      "loss": 0.7192,
      "step": 4147
    },
    {
      "epoch": 0.41005362924152927,
      "grad_norm": 6.25006471242061,
      "learning_rate": 1.3333968223369045e-07,
      "loss": 0.7995,
      "step": 4148
    },
    {
      "epoch": 0.41015248498628376,
      "grad_norm": 4.682739822101903,
      "learning_rate": 1.333094914368685e-07,
      "loss": 0.6172,
      "step": 4149
    },
    {
      "epoch": 0.41025134073103825,
      "grad_norm": 4.904357566647226,
      "learning_rate": 1.3327929722464953e-07,
      "loss": 0.6698,
      "step": 4150
    },
    {
      "epoch": 0.4103501964757927,
      "grad_norm": 3.7512662160052304,
      "learning_rate": 1.332490996001295e-07,
      "loss": 0.7246,
      "step": 4151
    },
    {
      "epoch": 0.4104490522205472,
      "grad_norm": 4.1059405588388245,
      "learning_rate": 1.3321889856640481e-07,
      "loss": 0.7276,
      "step": 4152
    },
    {
      "epoch": 0.4105479079653016,
      "grad_norm": 5.289667843108652,
      "learning_rate": 1.3318869412657207e-07,
      "loss": 0.7082,
      "step": 4153
    },
    {
      "epoch": 0.4106467637100561,
      "grad_norm": 4.42016271758641,
      "learning_rate": 1.3315848628372835e-07,
      "loss": 0.7207,
      "step": 4154
    },
    {
      "epoch": 0.4107456194548106,
      "grad_norm": 4.226248768050254,
      "learning_rate": 1.3312827504097092e-07,
      "loss": 0.732,
      "step": 4155
    },
    {
      "epoch": 0.410844475199565,
      "grad_norm": 16.689637415270777,
      "learning_rate": 1.3309806040139758e-07,
      "loss": 0.6785,
      "step": 4156
    },
    {
      "epoch": 0.4109433309443195,
      "grad_norm": 3.528625719977437,
      "learning_rate": 1.3306784236810643e-07,
      "loss": 0.666,
      "step": 4157
    },
    {
      "epoch": 0.411042186689074,
      "grad_norm": 4.289495554823182,
      "learning_rate": 1.3303762094419578e-07,
      "loss": 0.7018,
      "step": 4158
    },
    {
      "epoch": 0.41114104243382843,
      "grad_norm": 3.4444168190781594,
      "learning_rate": 1.3300739613276442e-07,
      "loss": 0.7227,
      "step": 4159
    },
    {
      "epoch": 0.4112398981785829,
      "grad_norm": 8.51425251109211,
      "learning_rate": 1.3297716793691152e-07,
      "loss": 0.6491,
      "step": 4160
    },
    {
      "epoch": 0.41133875392333735,
      "grad_norm": 3.8218078989572,
      "learning_rate": 1.329469363597365e-07,
      "loss": 0.6747,
      "step": 4161
    },
    {
      "epoch": 0.41143760966809184,
      "grad_norm": 5.8392546526973375,
      "learning_rate": 1.3291670140433914e-07,
      "loss": 0.7654,
      "step": 4162
    },
    {
      "epoch": 0.41153646541284633,
      "grad_norm": 4.05451166130992,
      "learning_rate": 1.328864630738196e-07,
      "loss": 0.685,
      "step": 4163
    },
    {
      "epoch": 0.41163532115760076,
      "grad_norm": 2.8385978683656403,
      "learning_rate": 1.328562213712784e-07,
      "loss": 0.8128,
      "step": 4164
    },
    {
      "epoch": 0.41173417690235525,
      "grad_norm": 5.305264485637704,
      "learning_rate": 1.3282597629981636e-07,
      "loss": 0.6463,
      "step": 4165
    },
    {
      "epoch": 0.4118330326471097,
      "grad_norm": 6.0807564828619025,
      "learning_rate": 1.3279572786253464e-07,
      "loss": 0.6528,
      "step": 4166
    },
    {
      "epoch": 0.4119318883918642,
      "grad_norm": 4.411072547377337,
      "learning_rate": 1.3276547606253482e-07,
      "loss": 0.6391,
      "step": 4167
    },
    {
      "epoch": 0.41203074413661867,
      "grad_norm": 8.576258871781818,
      "learning_rate": 1.3273522090291876e-07,
      "loss": 0.6986,
      "step": 4168
    },
    {
      "epoch": 0.4121295998813731,
      "grad_norm": 6.326283823466673,
      "learning_rate": 1.3270496238678867e-07,
      "loss": 0.7625,
      "step": 4169
    },
    {
      "epoch": 0.4122284556261276,
      "grad_norm": 4.0330553043248125,
      "learning_rate": 1.3267470051724714e-07,
      "loss": 0.8122,
      "step": 4170
    },
    {
      "epoch": 0.412327311370882,
      "grad_norm": 5.547031900172805,
      "learning_rate": 1.3264443529739705e-07,
      "loss": 0.7777,
      "step": 4171
    },
    {
      "epoch": 0.4124261671156365,
      "grad_norm": 6.1020691703148024,
      "learning_rate": 1.3261416673034163e-07,
      "loss": 0.6888,
      "step": 4172
    },
    {
      "epoch": 0.412525022860391,
      "grad_norm": 4.762960742516245,
      "learning_rate": 1.325838948191845e-07,
      "loss": 0.7273,
      "step": 4173
    },
    {
      "epoch": 0.41262387860514543,
      "grad_norm": 3.468085628334079,
      "learning_rate": 1.3255361956702961e-07,
      "loss": 0.7207,
      "step": 4174
    },
    {
      "epoch": 0.4127227343498999,
      "grad_norm": 4.139633296415921,
      "learning_rate": 1.3252334097698127e-07,
      "loss": 0.7633,
      "step": 4175
    },
    {
      "epoch": 0.41282159009465436,
      "grad_norm": 3.1265741365353943,
      "learning_rate": 1.3249305905214403e-07,
      "loss": 0.6937,
      "step": 4176
    },
    {
      "epoch": 0.41292044583940885,
      "grad_norm": 3.663736771192603,
      "learning_rate": 1.324627737956229e-07,
      "loss": 0.7116,
      "step": 4177
    },
    {
      "epoch": 0.41301930158416333,
      "grad_norm": 2.8907151168411334,
      "learning_rate": 1.3243248521052317e-07,
      "loss": 0.6855,
      "step": 4178
    },
    {
      "epoch": 0.41311815732891777,
      "grad_norm": 5.49325597981901,
      "learning_rate": 1.3240219329995052e-07,
      "loss": 0.8764,
      "step": 4179
    },
    {
      "epoch": 0.41321701307367226,
      "grad_norm": 5.914429364586734,
      "learning_rate": 1.323718980670109e-07,
      "loss": 0.7913,
      "step": 4180
    },
    {
      "epoch": 0.4133158688184267,
      "grad_norm": 7.4166513009622,
      "learning_rate": 1.3234159951481062e-07,
      "loss": 0.6933,
      "step": 4181
    },
    {
      "epoch": 0.4134147245631812,
      "grad_norm": 10.497114907352046,
      "learning_rate": 1.3231129764645644e-07,
      "loss": 0.7808,
      "step": 4182
    },
    {
      "epoch": 0.41351358030793567,
      "grad_norm": 26.73464326126239,
      "learning_rate": 1.3228099246505528e-07,
      "loss": 0.6892,
      "step": 4183
    },
    {
      "epoch": 0.4136124360526901,
      "grad_norm": 6.454655450237664,
      "learning_rate": 1.3225068397371453e-07,
      "loss": 0.8097,
      "step": 4184
    },
    {
      "epoch": 0.4137112917974446,
      "grad_norm": 4.089267807025319,
      "learning_rate": 1.3222037217554186e-07,
      "loss": 0.8105,
      "step": 4185
    },
    {
      "epoch": 0.413810147542199,
      "grad_norm": 7.137514667883646,
      "learning_rate": 1.321900570736453e-07,
      "loss": 0.6659,
      "step": 4186
    },
    {
      "epoch": 0.4139090032869535,
      "grad_norm": 4.535880225065229,
      "learning_rate": 1.3215973867113323e-07,
      "loss": 0.6687,
      "step": 4187
    },
    {
      "epoch": 0.414007859031708,
      "grad_norm": 13.072613652095777,
      "learning_rate": 1.3212941697111436e-07,
      "loss": 0.8001,
      "step": 4188
    },
    {
      "epoch": 0.41410671477646244,
      "grad_norm": 3.8115259848286596,
      "learning_rate": 1.3209909197669774e-07,
      "loss": 0.7345,
      "step": 4189
    },
    {
      "epoch": 0.4142055705212169,
      "grad_norm": 4.1744050896259415,
      "learning_rate": 1.320687636909927e-07,
      "loss": 0.7406,
      "step": 4190
    },
    {
      "epoch": 0.41430442626597136,
      "grad_norm": 5.0448130177705375,
      "learning_rate": 1.32038432117109e-07,
      "loss": 0.6781,
      "step": 4191
    },
    {
      "epoch": 0.41440328201072585,
      "grad_norm": 3.712726308212724,
      "learning_rate": 1.3200809725815667e-07,
      "loss": 0.7302,
      "step": 4192
    },
    {
      "epoch": 0.41450213775548034,
      "grad_norm": 4.775312234267417,
      "learning_rate": 1.3197775911724613e-07,
      "loss": 0.8619,
      "step": 4193
    },
    {
      "epoch": 0.41460099350023477,
      "grad_norm": 7.283231885923636,
      "learning_rate": 1.319474176974881e-07,
      "loss": 0.7221,
      "step": 4194
    },
    {
      "epoch": 0.41469984924498926,
      "grad_norm": 3.9150880884735115,
      "learning_rate": 1.3191707300199365e-07,
      "loss": 0.7346,
      "step": 4195
    },
    {
      "epoch": 0.4147987049897437,
      "grad_norm": 4.254786955188688,
      "learning_rate": 1.3188672503387417e-07,
      "loss": 0.639,
      "step": 4196
    },
    {
      "epoch": 0.4148975607344982,
      "grad_norm": 7.187899887693538,
      "learning_rate": 1.318563737962414e-07,
      "loss": 0.6822,
      "step": 4197
    },
    {
      "epoch": 0.41499641647925267,
      "grad_norm": 6.00280896675638,
      "learning_rate": 1.3182601929220744e-07,
      "loss": 0.6223,
      "step": 4198
    },
    {
      "epoch": 0.4150952722240071,
      "grad_norm": 9.567787542558415,
      "learning_rate": 1.317956615248846e-07,
      "loss": 0.5771,
      "step": 4199
    },
    {
      "epoch": 0.4151941279687616,
      "grad_norm": 2.943451369839274,
      "learning_rate": 1.3176530049738573e-07,
      "loss": 0.6585,
      "step": 4200
    },
    {
      "epoch": 0.41529298371351603,
      "grad_norm": 6.951195446865218,
      "learning_rate": 1.3173493621282384e-07,
      "loss": 0.645,
      "step": 4201
    },
    {
      "epoch": 0.4153918394582705,
      "grad_norm": 5.100460019764596,
      "learning_rate": 1.3170456867431242e-07,
      "loss": 0.748,
      "step": 4202
    },
    {
      "epoch": 0.415490695203025,
      "grad_norm": 3.815386210307499,
      "learning_rate": 1.3167419788496509e-07,
      "loss": 0.6644,
      "step": 4203
    },
    {
      "epoch": 0.41558955094777944,
      "grad_norm": 4.762921805977451,
      "learning_rate": 1.3164382384789602e-07,
      "loss": 0.6948,
      "step": 4204
    },
    {
      "epoch": 0.41568840669253393,
      "grad_norm": 4.655185678526731,
      "learning_rate": 1.3161344656621961e-07,
      "loss": 0.766,
      "step": 4205
    },
    {
      "epoch": 0.41578726243728836,
      "grad_norm": 4.379285576381063,
      "learning_rate": 1.3158306604305063e-07,
      "loss": 0.626,
      "step": 4206
    },
    {
      "epoch": 0.41588611818204285,
      "grad_norm": 3.985568600716228,
      "learning_rate": 1.31552682281504e-07,
      "loss": 0.6875,
      "step": 4207
    },
    {
      "epoch": 0.41598497392679734,
      "grad_norm": 2.8127696626830323,
      "learning_rate": 1.315222952846953e-07,
      "loss": 0.7688,
      "step": 4208
    },
    {
      "epoch": 0.4160838296715518,
      "grad_norm": 7.922990145200747,
      "learning_rate": 1.3149190505574024e-07,
      "loss": 0.7248,
      "step": 4209
    },
    {
      "epoch": 0.41618268541630626,
      "grad_norm": 26.79635756871759,
      "learning_rate": 1.314615115977548e-07,
      "loss": 0.8011,
      "step": 4210
    },
    {
      "epoch": 0.4162815411610607,
      "grad_norm": 4.453661912482545,
      "learning_rate": 1.3143111491385543e-07,
      "loss": 0.8224,
      "step": 4211
    },
    {
      "epoch": 0.4163803969058152,
      "grad_norm": 3.743084456260384,
      "learning_rate": 1.314007150071589e-07,
      "loss": 0.6336,
      "step": 4212
    },
    {
      "epoch": 0.4164792526505697,
      "grad_norm": 11.380289602556273,
      "learning_rate": 1.3137031188078224e-07,
      "loss": 0.7087,
      "step": 4213
    },
    {
      "epoch": 0.4165781083953241,
      "grad_norm": 3.4164025087991585,
      "learning_rate": 1.3133990553784285e-07,
      "loss": 0.7264,
      "step": 4214
    },
    {
      "epoch": 0.4166769641400786,
      "grad_norm": 4.627037006155441,
      "learning_rate": 1.3130949598145842e-07,
      "loss": 0.7336,
      "step": 4215
    },
    {
      "epoch": 0.41677581988483303,
      "grad_norm": 5.799516705035995,
      "learning_rate": 1.3127908321474702e-07,
      "loss": 0.6855,
      "step": 4216
    },
    {
      "epoch": 0.4168746756295875,
      "grad_norm": 3.150824890690916,
      "learning_rate": 1.3124866724082703e-07,
      "loss": 0.6744,
      "step": 4217
    },
    {
      "epoch": 0.416973531374342,
      "grad_norm": 3.598199714599413,
      "learning_rate": 1.3121824806281718e-07,
      "loss": 0.7392,
      "step": 4218
    },
    {
      "epoch": 0.41707238711909644,
      "grad_norm": 4.851393469064133,
      "learning_rate": 1.3118782568383648e-07,
      "loss": 0.7758,
      "step": 4219
    },
    {
      "epoch": 0.41717124286385093,
      "grad_norm": 3.8413086362031366,
      "learning_rate": 1.3115740010700433e-07,
      "loss": 0.705,
      "step": 4220
    },
    {
      "epoch": 0.41727009860860537,
      "grad_norm": 4.322342856411626,
      "learning_rate": 1.3112697133544038e-07,
      "loss": 0.6818,
      "step": 4221
    },
    {
      "epoch": 0.41736895435335986,
      "grad_norm": 4.343681241028514,
      "learning_rate": 1.310965393722647e-07,
      "loss": 0.7412,
      "step": 4222
    },
    {
      "epoch": 0.41746781009811434,
      "grad_norm": 6.512687985129534,
      "learning_rate": 1.310661042205976e-07,
      "loss": 0.7373,
      "step": 4223
    },
    {
      "epoch": 0.4175666658428688,
      "grad_norm": 3.1981218064096093,
      "learning_rate": 1.3103566588355977e-07,
      "loss": 0.6552,
      "step": 4224
    },
    {
      "epoch": 0.41766552158762327,
      "grad_norm": 3.7922130021152864,
      "learning_rate": 1.310052243642722e-07,
      "loss": 0.7402,
      "step": 4225
    },
    {
      "epoch": 0.4177643773323777,
      "grad_norm": 3.9471658645217627,
      "learning_rate": 1.3097477966585624e-07,
      "loss": 0.6651,
      "step": 4226
    },
    {
      "epoch": 0.4178632330771322,
      "grad_norm": 3.957191536892536,
      "learning_rate": 1.3094433179143357e-07,
      "loss": 0.7377,
      "step": 4227
    },
    {
      "epoch": 0.4179620888218867,
      "grad_norm": 3.1708347473949985,
      "learning_rate": 1.3091388074412609e-07,
      "loss": 0.7678,
      "step": 4228
    },
    {
      "epoch": 0.4180609445666411,
      "grad_norm": 4.013879059795465,
      "learning_rate": 1.3088342652705614e-07,
      "loss": 0.646,
      "step": 4229
    },
    {
      "epoch": 0.4181598003113956,
      "grad_norm": 16.187526344087836,
      "learning_rate": 1.3085296914334638e-07,
      "loss": 0.6569,
      "step": 4230
    },
    {
      "epoch": 0.41825865605615004,
      "grad_norm": 10.896850463452301,
      "learning_rate": 1.3082250859611976e-07,
      "loss": 0.7035,
      "step": 4231
    },
    {
      "epoch": 0.4183575118009045,
      "grad_norm": 5.075156988991124,
      "learning_rate": 1.3079204488849955e-07,
      "loss": 0.6554,
      "step": 4232
    },
    {
      "epoch": 0.418456367545659,
      "grad_norm": 4.2394394907333135,
      "learning_rate": 1.3076157802360934e-07,
      "loss": 0.7127,
      "step": 4233
    },
    {
      "epoch": 0.41855522329041345,
      "grad_norm": 3.182869860010819,
      "learning_rate": 1.3073110800457308e-07,
      "loss": 0.7491,
      "step": 4234
    },
    {
      "epoch": 0.41865407903516794,
      "grad_norm": 3.018938095120049,
      "learning_rate": 1.3070063483451501e-07,
      "loss": 0.6439,
      "step": 4235
    },
    {
      "epoch": 0.4187529347799224,
      "grad_norm": 7.716857285602844,
      "learning_rate": 1.3067015851655966e-07,
      "loss": 0.7135,
      "step": 4236
    },
    {
      "epoch": 0.41885179052467686,
      "grad_norm": 4.135321382879602,
      "learning_rate": 1.3063967905383205e-07,
      "loss": 0.7429,
      "step": 4237
    },
    {
      "epoch": 0.41895064626943135,
      "grad_norm": 5.635047715864612,
      "learning_rate": 1.3060919644945725e-07,
      "loss": 0.6875,
      "step": 4238
    },
    {
      "epoch": 0.4190495020141858,
      "grad_norm": 3.117496205847984,
      "learning_rate": 1.3057871070656092e-07,
      "loss": 0.7285,
      "step": 4239
    },
    {
      "epoch": 0.41914835775894027,
      "grad_norm": 3.4137128878124012,
      "learning_rate": 1.3054822182826886e-07,
      "loss": 0.7565,
      "step": 4240
    },
    {
      "epoch": 0.41924721350369476,
      "grad_norm": 3.110118256946252,
      "learning_rate": 1.3051772981770726e-07,
      "loss": 0.6797,
      "step": 4241
    },
    {
      "epoch": 0.4193460692484492,
      "grad_norm": 3.854698494167551,
      "learning_rate": 1.3048723467800265e-07,
      "loss": 0.6703,
      "step": 4242
    },
    {
      "epoch": 0.4194449249932037,
      "grad_norm": 4.602089899332918,
      "learning_rate": 1.304567364122818e-07,
      "loss": 0.7728,
      "step": 4243
    },
    {
      "epoch": 0.4195437807379581,
      "grad_norm": 3.6437942450514775,
      "learning_rate": 1.3042623502367194e-07,
      "loss": 0.7124,
      "step": 4244
    },
    {
      "epoch": 0.4196426364827126,
      "grad_norm": 10.005784764682327,
      "learning_rate": 1.3039573051530049e-07,
      "loss": 0.6175,
      "step": 4245
    },
    {
      "epoch": 0.4197414922274671,
      "grad_norm": 3.7561039271630148,
      "learning_rate": 1.3036522289029524e-07,
      "loss": 0.7024,
      "step": 4246
    },
    {
      "epoch": 0.41984034797222153,
      "grad_norm": 5.623028762994784,
      "learning_rate": 1.3033471215178428e-07,
      "loss": 0.7756,
      "step": 4247
    },
    {
      "epoch": 0.419939203716976,
      "grad_norm": 7.165027583210982,
      "learning_rate": 1.3030419830289607e-07,
      "loss": 0.7428,
      "step": 4248
    },
    {
      "epoch": 0.42003805946173045,
      "grad_norm": 11.345036131321374,
      "learning_rate": 1.3027368134675935e-07,
      "loss": 0.8126,
      "step": 4249
    },
    {
      "epoch": 0.42013691520648494,
      "grad_norm": 3.419486302995425,
      "learning_rate": 1.3024316128650322e-07,
      "loss": 0.7802,
      "step": 4250
    },
    {
      "epoch": 0.42023577095123943,
      "grad_norm": 3.148351811230768,
      "learning_rate": 1.3021263812525697e-07,
      "loss": 0.6944,
      "step": 4251
    },
    {
      "epoch": 0.42033462669599386,
      "grad_norm": 3.41944531883536,
      "learning_rate": 1.3018211186615035e-07,
      "loss": 0.6684,
      "step": 4252
    },
    {
      "epoch": 0.42043348244074835,
      "grad_norm": 5.054559649598608,
      "learning_rate": 1.3015158251231338e-07,
      "loss": 0.6941,
      "step": 4253
    },
    {
      "epoch": 0.4205323381855028,
      "grad_norm": 10.627621477921144,
      "learning_rate": 1.3012105006687637e-07,
      "loss": 0.7439,
      "step": 4254
    },
    {
      "epoch": 0.4206311939302573,
      "grad_norm": 3.5446811718602493,
      "learning_rate": 1.3009051453297003e-07,
      "loss": 0.7525,
      "step": 4255
    },
    {
      "epoch": 0.42073004967501176,
      "grad_norm": 3.7799938017053294,
      "learning_rate": 1.3005997591372525e-07,
      "loss": 0.7815,
      "step": 4256
    },
    {
      "epoch": 0.4208289054197662,
      "grad_norm": 3.8022865208711343,
      "learning_rate": 1.3002943421227338e-07,
      "loss": 0.6318,
      "step": 4257
    },
    {
      "epoch": 0.4209277611645207,
      "grad_norm": 2.805054642441229,
      "learning_rate": 1.2999888943174603e-07,
      "loss": 0.6555,
      "step": 4258
    },
    {
      "epoch": 0.4210266169092751,
      "grad_norm": 4.98767407367027,
      "learning_rate": 1.2996834157527503e-07,
      "loss": 0.7475,
      "step": 4259
    },
    {
      "epoch": 0.4211254726540296,
      "grad_norm": 5.867577277486626,
      "learning_rate": 1.2993779064599268e-07,
      "loss": 0.7701,
      "step": 4260
    },
    {
      "epoch": 0.4212243283987841,
      "grad_norm": 3.2108946998775796,
      "learning_rate": 1.299072366470315e-07,
      "loss": 0.6929,
      "step": 4261
    },
    {
      "epoch": 0.42132318414353853,
      "grad_norm": 11.397387089225367,
      "learning_rate": 1.298766795815244e-07,
      "loss": 0.7414,
      "step": 4262
    },
    {
      "epoch": 0.421422039888293,
      "grad_norm": 6.792582028031758,
      "learning_rate": 1.298461194526045e-07,
      "loss": 0.7658,
      "step": 4263
    },
    {
      "epoch": 0.42152089563304745,
      "grad_norm": 3.739186657137695,
      "learning_rate": 1.2981555626340534e-07,
      "loss": 0.7977,
      "step": 4264
    },
    {
      "epoch": 0.42161975137780194,
      "grad_norm": 9.2855546379833,
      "learning_rate": 1.2978499001706063e-07,
      "loss": 0.7081,
      "step": 4265
    },
    {
      "epoch": 0.42171860712255643,
      "grad_norm": 2.9010126193510684,
      "learning_rate": 1.2975442071670465e-07,
      "loss": 0.7139,
      "step": 4266
    },
    {
      "epoch": 0.42181746286731087,
      "grad_norm": 2.889527640368366,
      "learning_rate": 1.2972384836547164e-07,
      "loss": 0.6306,
      "step": 4267
    },
    {
      "epoch": 0.42191631861206536,
      "grad_norm": 5.050864523488164,
      "learning_rate": 1.296932729664965e-07,
      "loss": 0.645,
      "step": 4268
    },
    {
      "epoch": 0.4220151743568198,
      "grad_norm": 6.054781258281637,
      "learning_rate": 1.296626945229142e-07,
      "loss": 0.7705,
      "step": 4269
    },
    {
      "epoch": 0.4221140301015743,
      "grad_norm": 4.603335207205996,
      "learning_rate": 1.2963211303786011e-07,
      "loss": 0.7142,
      "step": 4270
    },
    {
      "epoch": 0.42221288584632877,
      "grad_norm": 3.428583261872119,
      "learning_rate": 1.2960152851446995e-07,
      "loss": 0.7827,
      "step": 4271
    },
    {
      "epoch": 0.4223117415910832,
      "grad_norm": 5.626756198225381,
      "learning_rate": 1.2957094095587968e-07,
      "loss": 0.747,
      "step": 4272
    },
    {
      "epoch": 0.4224105973358377,
      "grad_norm": 4.277068149133427,
      "learning_rate": 1.2954035036522566e-07,
      "loss": 0.7015,
      "step": 4273
    },
    {
      "epoch": 0.4225094530805921,
      "grad_norm": 6.513191653666133,
      "learning_rate": 1.2950975674564442e-07,
      "loss": 0.6498,
      "step": 4274
    },
    {
      "epoch": 0.4226083088253466,
      "grad_norm": 6.080438836459712,
      "learning_rate": 1.2947916010027297e-07,
      "loss": 0.6813,
      "step": 4275
    },
    {
      "epoch": 0.4227071645701011,
      "grad_norm": 6.485488214474651,
      "learning_rate": 1.2944856043224846e-07,
      "loss": 0.7228,
      "step": 4276
    },
    {
      "epoch": 0.42280602031485554,
      "grad_norm": 9.171123004234422,
      "learning_rate": 1.2941795774470849e-07,
      "loss": 0.7065,
      "step": 4277
    },
    {
      "epoch": 0.42290487605961,
      "grad_norm": 6.462563499714309,
      "learning_rate": 1.293873520407909e-07,
      "loss": 0.7655,
      "step": 4278
    },
    {
      "epoch": 0.42300373180436446,
      "grad_norm": 3.9263814117833156,
      "learning_rate": 1.2935674332363385e-07,
      "loss": 0.7119,
      "step": 4279
    },
    {
      "epoch": 0.42310258754911895,
      "grad_norm": 3.2757300384786188,
      "learning_rate": 1.293261315963758e-07,
      "loss": 0.6252,
      "step": 4280
    },
    {
      "epoch": 0.42320144329387344,
      "grad_norm": 5.2941345996013025,
      "learning_rate": 1.2929551686215554e-07,
      "loss": 0.7082,
      "step": 4281
    },
    {
      "epoch": 0.42330029903862787,
      "grad_norm": 3.8906088967964045,
      "learning_rate": 1.2926489912411218e-07,
      "loss": 0.6788,
      "step": 4282
    },
    {
      "epoch": 0.42339915478338236,
      "grad_norm": 9.789232333710155,
      "learning_rate": 1.2923427838538512e-07,
      "loss": 0.6802,
      "step": 4283
    },
    {
      "epoch": 0.4234980105281368,
      "grad_norm": 4.632933159698568,
      "learning_rate": 1.2920365464911402e-07,
      "loss": 0.7385,
      "step": 4284
    },
    {
      "epoch": 0.4235968662728913,
      "grad_norm": 3.2194541455200154,
      "learning_rate": 1.2917302791843897e-07,
      "loss": 0.5911,
      "step": 4285
    },
    {
      "epoch": 0.42369572201764577,
      "grad_norm": 3.300277627698636,
      "learning_rate": 1.291423981965002e-07,
      "loss": 0.6389,
      "step": 4286
    },
    {
      "epoch": 0.4237945777624002,
      "grad_norm": 6.781442155054954,
      "learning_rate": 1.2911176548643834e-07,
      "loss": 0.7463,
      "step": 4287
    },
    {
      "epoch": 0.4238934335071547,
      "grad_norm": 9.184575955142646,
      "learning_rate": 1.290811297913944e-07,
      "loss": 0.7388,
      "step": 4288
    },
    {
      "epoch": 0.4239922892519091,
      "grad_norm": 4.476161662953344,
      "learning_rate": 1.2905049111450958e-07,
      "loss": 0.7657,
      "step": 4289
    },
    {
      "epoch": 0.4240911449966636,
      "grad_norm": 6.979539947746087,
      "learning_rate": 1.2901984945892542e-07,
      "loss": 0.6962,
      "step": 4290
    },
    {
      "epoch": 0.4241900007414181,
      "grad_norm": 3.993502749854961,
      "learning_rate": 1.2898920482778372e-07,
      "loss": 0.7106,
      "step": 4291
    },
    {
      "epoch": 0.42428885648617254,
      "grad_norm": 3.8241512412751395,
      "learning_rate": 1.2895855722422674e-07,
      "loss": 0.7487,
      "step": 4292
    },
    {
      "epoch": 0.42438771223092703,
      "grad_norm": 3.115888402122379,
      "learning_rate": 1.289279066513969e-07,
      "loss": 0.5497,
      "step": 4293
    },
    {
      "epoch": 0.42448656797568146,
      "grad_norm": 2.988733158512786,
      "learning_rate": 1.288972531124369e-07,
      "loss": 0.6752,
      "step": 4294
    },
    {
      "epoch": 0.42458542372043595,
      "grad_norm": 4.960556726980964,
      "learning_rate": 1.2886659661048992e-07,
      "loss": 0.7067,
      "step": 4295
    },
    {
      "epoch": 0.42468427946519044,
      "grad_norm": 4.871374304429121,
      "learning_rate": 1.2883593714869921e-07,
      "loss": 0.7291,
      "step": 4296
    },
    {
      "epoch": 0.4247831352099449,
      "grad_norm": 2.6948418375720182,
      "learning_rate": 1.2880527473020856e-07,
      "loss": 0.7158,
      "step": 4297
    },
    {
      "epoch": 0.42488199095469936,
      "grad_norm": 7.66664006996376,
      "learning_rate": 1.2877460935816182e-07,
      "loss": 0.7468,
      "step": 4298
    },
    {
      "epoch": 0.4249808466994538,
      "grad_norm": 3.2013752995905507,
      "learning_rate": 1.2874394103570342e-07,
      "loss": 0.619,
      "step": 4299
    },
    {
      "epoch": 0.4250797024442083,
      "grad_norm": 5.552470757576194,
      "learning_rate": 1.2871326976597782e-07,
      "loss": 0.8107,
      "step": 4300
    },
    {
      "epoch": 0.4251785581889628,
      "grad_norm": 6.7049379654740315,
      "learning_rate": 1.2868259555212996e-07,
      "loss": 0.8039,
      "step": 4301
    },
    {
      "epoch": 0.4252774139337172,
      "grad_norm": 3.6763348531618134,
      "learning_rate": 1.2865191839730504e-07,
      "loss": 0.7498,
      "step": 4302
    },
    {
      "epoch": 0.4253762696784717,
      "grad_norm": 3.5952447053413583,
      "learning_rate": 1.2862123830464854e-07,
      "loss": 0.7337,
      "step": 4303
    },
    {
      "epoch": 0.42547512542322613,
      "grad_norm": 3.4833497833747322,
      "learning_rate": 1.2859055527730624e-07,
      "loss": 0.8011,
      "step": 4304
    },
    {
      "epoch": 0.4255739811679806,
      "grad_norm": 4.136442942795607,
      "learning_rate": 1.2855986931842423e-07,
      "loss": 0.635,
      "step": 4305
    },
    {
      "epoch": 0.4256728369127351,
      "grad_norm": 3.2665450327074006,
      "learning_rate": 1.2852918043114892e-07,
      "loss": 0.6927,
      "step": 4306
    },
    {
      "epoch": 0.42577169265748954,
      "grad_norm": 4.4937892486060536,
      "learning_rate": 1.2849848861862698e-07,
      "loss": 0.6611,
      "step": 4307
    },
    {
      "epoch": 0.42587054840224403,
      "grad_norm": 3.1221564948742877,
      "learning_rate": 1.284677938840054e-07,
      "loss": 0.7629,
      "step": 4308
    },
    {
      "epoch": 0.42596940414699846,
      "grad_norm": 11.67751252323262,
      "learning_rate": 1.284370962304315e-07,
      "loss": 0.698,
      "step": 4309
    },
    {
      "epoch": 0.42606825989175295,
      "grad_norm": 8.077474830286286,
      "learning_rate": 1.2840639566105288e-07,
      "loss": 0.7271,
      "step": 4310
    },
    {
      "epoch": 0.42616711563650744,
      "grad_norm": 11.984633035072333,
      "learning_rate": 1.283756921790174e-07,
      "loss": 0.7015,
      "step": 4311
    },
    {
      "epoch": 0.4262659713812619,
      "grad_norm": 3.7213064225905312,
      "learning_rate": 1.2834498578747327e-07,
      "loss": 0.6355,
      "step": 4312
    },
    {
      "epoch": 0.42636482712601637,
      "grad_norm": 5.351540137863119,
      "learning_rate": 1.283142764895689e-07,
      "loss": 0.6761,
      "step": 4313
    },
    {
      "epoch": 0.42646368287077085,
      "grad_norm": 8.474271054067831,
      "learning_rate": 1.2828356428845322e-07,
      "loss": 0.707,
      "step": 4314
    },
    {
      "epoch": 0.4265625386155253,
      "grad_norm": 4.421087814022167,
      "learning_rate": 1.2825284918727518e-07,
      "loss": 0.7386,
      "step": 4315
    },
    {
      "epoch": 0.4266613943602798,
      "grad_norm": 6.509729067225355,
      "learning_rate": 1.2822213118918425e-07,
      "loss": 0.7923,
      "step": 4316
    },
    {
      "epoch": 0.4267602501050342,
      "grad_norm": 4.603889426587649,
      "learning_rate": 1.2819141029733005e-07,
      "loss": 0.6768,
      "step": 4317
    },
    {
      "epoch": 0.4268591058497887,
      "grad_norm": 37.99671174094154,
      "learning_rate": 1.2816068651486255e-07,
      "loss": 0.7193,
      "step": 4318
    },
    {
      "epoch": 0.4269579615945432,
      "grad_norm": 5.808324892064933,
      "learning_rate": 1.281299598449321e-07,
      "loss": 0.6644,
      "step": 4319
    },
    {
      "epoch": 0.4270568173392976,
      "grad_norm": 3.610514385721219,
      "learning_rate": 1.2809923029068922e-07,
      "loss": 0.7383,
      "step": 4320
    },
    {
      "epoch": 0.4271556730840521,
      "grad_norm": 3.8690291386311797,
      "learning_rate": 1.2806849785528472e-07,
      "loss": 0.6994,
      "step": 4321
    },
    {
      "epoch": 0.42725452882880655,
      "grad_norm": 4.9720881066044775,
      "learning_rate": 1.2803776254186985e-07,
      "loss": 0.696,
      "step": 4322
    },
    {
      "epoch": 0.42735338457356103,
      "grad_norm": 10.489705384118094,
      "learning_rate": 1.28007024353596e-07,
      "loss": 0.7063,
      "step": 4323
    },
    {
      "epoch": 0.4274522403183155,
      "grad_norm": 10.156889144823271,
      "learning_rate": 1.279762832936149e-07,
      "loss": 0.7111,
      "step": 4324
    },
    {
      "epoch": 0.42755109606306996,
      "grad_norm": 4.064546666181333,
      "learning_rate": 1.2794553936507866e-07,
      "loss": 0.6213,
      "step": 4325
    },
    {
      "epoch": 0.42764995180782445,
      "grad_norm": 4.5495291288938455,
      "learning_rate": 1.2791479257113958e-07,
      "loss": 0.7411,
      "step": 4326
    },
    {
      "epoch": 0.4277488075525789,
      "grad_norm": 6.008356395336652,
      "learning_rate": 1.278840429149503e-07,
      "loss": 0.6886,
      "step": 4327
    },
    {
      "epoch": 0.42784766329733337,
      "grad_norm": 4.897801670254142,
      "learning_rate": 1.2785329039966373e-07,
      "loss": 0.715,
      "step": 4328
    },
    {
      "epoch": 0.42794651904208786,
      "grad_norm": 3.0442160985529068,
      "learning_rate": 1.278225350284331e-07,
      "loss": 0.642,
      "step": 4329
    },
    {
      "epoch": 0.4280453747868423,
      "grad_norm": 4.057452925170052,
      "learning_rate": 1.277917768044119e-07,
      "loss": 0.7051,
      "step": 4330
    },
    {
      "epoch": 0.4281442305315968,
      "grad_norm": 3.444549949648897,
      "learning_rate": 1.2776101573075396e-07,
      "loss": 0.6496,
      "step": 4331
    },
    {
      "epoch": 0.4282430862763512,
      "grad_norm": 2.9525505451994336,
      "learning_rate": 1.2773025181061337e-07,
      "loss": 0.6661,
      "step": 4332
    },
    {
      "epoch": 0.4283419420211057,
      "grad_norm": 5.04625991089581,
      "learning_rate": 1.276994850471445e-07,
      "loss": 0.6904,
      "step": 4333
    },
    {
      "epoch": 0.4284407977658602,
      "grad_norm": 3.103642275491389,
      "learning_rate": 1.27668715443502e-07,
      "loss": 0.6139,
      "step": 4334
    },
    {
      "epoch": 0.4285396535106146,
      "grad_norm": 3.2801239862414975,
      "learning_rate": 1.276379430028409e-07,
      "loss": 0.8062,
      "step": 4335
    },
    {
      "epoch": 0.4286385092553691,
      "grad_norm": 6.89741587957365,
      "learning_rate": 1.2760716772831645e-07,
      "loss": 0.781,
      "step": 4336
    },
    {
      "epoch": 0.42873736500012355,
      "grad_norm": 5.124431694699376,
      "learning_rate": 1.275763896230842e-07,
      "loss": 0.753,
      "step": 4337
    },
    {
      "epoch": 0.42883622074487804,
      "grad_norm": 3.5043506639887028,
      "learning_rate": 1.275456086902999e-07,
      "loss": 0.7129,
      "step": 4338
    },
    {
      "epoch": 0.4289350764896325,
      "grad_norm": 6.725192764441534,
      "learning_rate": 1.275148249331198e-07,
      "loss": 0.7289,
      "step": 4339
    },
    {
      "epoch": 0.42903393223438696,
      "grad_norm": 4.289084601167294,
      "learning_rate": 1.2748403835470027e-07,
      "loss": 0.6991,
      "step": 4340
    },
    {
      "epoch": 0.42913278797914145,
      "grad_norm": 3.412954652627056,
      "learning_rate": 1.2745324895819805e-07,
      "loss": 0.6345,
      "step": 4341
    },
    {
      "epoch": 0.4292316437238959,
      "grad_norm": 4.359660170908633,
      "learning_rate": 1.2742245674677008e-07,
      "loss": 0.6353,
      "step": 4342
    },
    {
      "epoch": 0.4293304994686504,
      "grad_norm": 6.004726049690692,
      "learning_rate": 1.273916617235737e-07,
      "loss": 0.7129,
      "step": 4343
    },
    {
      "epoch": 0.42942935521340486,
      "grad_norm": 5.30434329102087,
      "learning_rate": 1.2736086389176646e-07,
      "loss": 0.7029,
      "step": 4344
    },
    {
      "epoch": 0.4295282109581593,
      "grad_norm": 3.3467146890743518,
      "learning_rate": 1.2733006325450622e-07,
      "loss": 0.6204,
      "step": 4345
    },
    {
      "epoch": 0.4296270667029138,
      "grad_norm": 6.308991652462591,
      "learning_rate": 1.2729925981495114e-07,
      "loss": 0.6439,
      "step": 4346
    },
    {
      "epoch": 0.4297259224476682,
      "grad_norm": 4.22619870188735,
      "learning_rate": 1.2726845357625968e-07,
      "loss": 0.7821,
      "step": 4347
    },
    {
      "epoch": 0.4298247781924227,
      "grad_norm": 18.602160604572163,
      "learning_rate": 1.2723764454159053e-07,
      "loss": 0.6534,
      "step": 4348
    },
    {
      "epoch": 0.4299236339371772,
      "grad_norm": 4.644661111120811,
      "learning_rate": 1.272068327141027e-07,
      "loss": 0.711,
      "step": 4349
    },
    {
      "epoch": 0.43002248968193163,
      "grad_norm": 4.660188510926421,
      "learning_rate": 1.2717601809695553e-07,
      "loss": 0.6934,
      "step": 4350
    },
    {
      "epoch": 0.4301213454266861,
      "grad_norm": 5.3953222748973975,
      "learning_rate": 1.2714520069330857e-07,
      "loss": 0.7623,
      "step": 4351
    },
    {
      "epoch": 0.43022020117144055,
      "grad_norm": 4.22370307234915,
      "learning_rate": 1.2711438050632168e-07,
      "loss": 0.6401,
      "step": 4352
    },
    {
      "epoch": 0.43031905691619504,
      "grad_norm": 7.588591804898984,
      "learning_rate": 1.2708355753915506e-07,
      "loss": 0.7086,
      "step": 4353
    },
    {
      "epoch": 0.43041791266094953,
      "grad_norm": 3.0145203162230825,
      "learning_rate": 1.2705273179496914e-07,
      "loss": 0.6347,
      "step": 4354
    },
    {
      "epoch": 0.43051676840570396,
      "grad_norm": 3.4919977204765695,
      "learning_rate": 1.270219032769246e-07,
      "loss": 0.7106,
      "step": 4355
    },
    {
      "epoch": 0.43061562415045845,
      "grad_norm": 5.285295345279206,
      "learning_rate": 1.2699107198818249e-07,
      "loss": 0.7076,
      "step": 4356
    },
    {
      "epoch": 0.4307144798952129,
      "grad_norm": 4.604014830179218,
      "learning_rate": 1.2696023793190407e-07,
      "loss": 0.6514,
      "step": 4357
    },
    {
      "epoch": 0.4308133356399674,
      "grad_norm": 6.4999279368531955,
      "learning_rate": 1.2692940111125096e-07,
      "loss": 0.7201,
      "step": 4358
    },
    {
      "epoch": 0.43091219138472187,
      "grad_norm": 30.077677883730775,
      "learning_rate": 1.2689856152938503e-07,
      "loss": 0.7752,
      "step": 4359
    },
    {
      "epoch": 0.4310110471294763,
      "grad_norm": 17.40373359824735,
      "learning_rate": 1.2686771918946833e-07,
      "loss": 0.7012,
      "step": 4360
    },
    {
      "epoch": 0.4311099028742308,
      "grad_norm": 2.7744416624335084,
      "learning_rate": 1.2683687409466343e-07,
      "loss": 0.6608,
      "step": 4361
    },
    {
      "epoch": 0.4312087586189852,
      "grad_norm": 4.951734798684461,
      "learning_rate": 1.268060262481329e-07,
      "loss": 0.7203,
      "step": 4362
    },
    {
      "epoch": 0.4313076143637397,
      "grad_norm": 3.7613180023067767,
      "learning_rate": 1.2677517565303984e-07,
      "loss": 0.6602,
      "step": 4363
    },
    {
      "epoch": 0.4314064701084942,
      "grad_norm": 4.799613210065259,
      "learning_rate": 1.2674432231254744e-07,
      "loss": 0.8099,
      "step": 4364
    },
    {
      "epoch": 0.43150532585324863,
      "grad_norm": 6.769777888943804,
      "learning_rate": 1.2671346622981936e-07,
      "loss": 0.6769,
      "step": 4365
    },
    {
      "epoch": 0.4316041815980031,
      "grad_norm": 4.394752711791347,
      "learning_rate": 1.266826074080193e-07,
      "loss": 0.6921,
      "step": 4366
    },
    {
      "epoch": 0.43170303734275756,
      "grad_norm": 7.29462833045304,
      "learning_rate": 1.2665174585031148e-07,
      "loss": 0.6879,
      "step": 4367
    },
    {
      "epoch": 0.43180189308751205,
      "grad_norm": 5.386676072898909,
      "learning_rate": 1.2662088155986024e-07,
      "loss": 0.6425,
      "step": 4368
    },
    {
      "epoch": 0.43190074883226653,
      "grad_norm": 5.091814726763433,
      "learning_rate": 1.265900145398303e-07,
      "loss": 0.7103,
      "step": 4369
    },
    {
      "epoch": 0.43199960457702097,
      "grad_norm": 8.75232455834141,
      "learning_rate": 1.265591447933866e-07,
      "loss": 0.649,
      "step": 4370
    },
    {
      "epoch": 0.43209846032177546,
      "grad_norm": 4.667663277345671,
      "learning_rate": 1.265282723236944e-07,
      "loss": 0.7344,
      "step": 4371
    },
    {
      "epoch": 0.4321973160665299,
      "grad_norm": 3.4294750830664427,
      "learning_rate": 1.2649739713391915e-07,
      "loss": 0.7673,
      "step": 4372
    },
    {
      "epoch": 0.4322961718112844,
      "grad_norm": 8.262375070432439,
      "learning_rate": 1.2646651922722672e-07,
      "loss": 0.7281,
      "step": 4373
    },
    {
      "epoch": 0.43239502755603887,
      "grad_norm": 4.5168429854814125,
      "learning_rate": 1.2643563860678313e-07,
      "loss": 0.7319,
      "step": 4374
    },
    {
      "epoch": 0.4324938833007933,
      "grad_norm": 3.5130406942244248,
      "learning_rate": 1.2640475527575474e-07,
      "loss": 0.7099,
      "step": 4375
    },
    {
      "epoch": 0.4325927390455478,
      "grad_norm": 4.997817930470844,
      "learning_rate": 1.263738692373082e-07,
      "loss": 0.8305,
      "step": 4376
    },
    {
      "epoch": 0.4326915947903022,
      "grad_norm": 4.400517902086195,
      "learning_rate": 1.2634298049461043e-07,
      "loss": 0.6829,
      "step": 4377
    },
    {
      "epoch": 0.4327904505350567,
      "grad_norm": 5.3044053964218225,
      "learning_rate": 1.263120890508286e-07,
      "loss": 0.7341,
      "step": 4378
    },
    {
      "epoch": 0.4328893062798112,
      "grad_norm": 5.849384098437953,
      "learning_rate": 1.2628119490913013e-07,
      "loss": 0.6987,
      "step": 4379
    },
    {
      "epoch": 0.43298816202456564,
      "grad_norm": 5.492677717683169,
      "learning_rate": 1.2625029807268287e-07,
      "loss": 0.6919,
      "step": 4380
    },
    {
      "epoch": 0.4330870177693201,
      "grad_norm": 6.8367239311903605,
      "learning_rate": 1.2621939854465472e-07,
      "loss": 0.6621,
      "step": 4381
    },
    {
      "epoch": 0.43318587351407456,
      "grad_norm": 4.466951493405875,
      "learning_rate": 1.26188496328214e-07,
      "loss": 0.6858,
      "step": 4382
    },
    {
      "epoch": 0.43328472925882905,
      "grad_norm": 4.538675015767456,
      "learning_rate": 1.2615759142652927e-07,
      "loss": 0.7062,
      "step": 4383
    },
    {
      "epoch": 0.43338358500358354,
      "grad_norm": 8.83538273287972,
      "learning_rate": 1.2612668384276945e-07,
      "loss": 0.5815,
      "step": 4384
    },
    {
      "epoch": 0.43348244074833797,
      "grad_norm": 2.744002988402256,
      "learning_rate": 1.2609577358010358e-07,
      "loss": 0.7475,
      "step": 4385
    },
    {
      "epoch": 0.43358129649309246,
      "grad_norm": 6.486717477469218,
      "learning_rate": 1.2606486064170102e-07,
      "loss": 0.7697,
      "step": 4386
    },
    {
      "epoch": 0.4336801522378469,
      "grad_norm": 4.070926237751014,
      "learning_rate": 1.2603394503073156e-07,
      "loss": 0.8002,
      "step": 4387
    },
    {
      "epoch": 0.4337790079826014,
      "grad_norm": 4.206049778302832,
      "learning_rate": 1.26003026750365e-07,
      "loss": 0.6518,
      "step": 4388
    },
    {
      "epoch": 0.4338778637273559,
      "grad_norm": 3.865201676393821,
      "learning_rate": 1.2597210580377166e-07,
      "loss": 0.6699,
      "step": 4389
    },
    {
      "epoch": 0.4339767194721103,
      "grad_norm": 3.8879430037055176,
      "learning_rate": 1.2594118219412198e-07,
      "loss": 0.7246,
      "step": 4390
    },
    {
      "epoch": 0.4340755752168648,
      "grad_norm": 11.197237523245033,
      "learning_rate": 1.2591025592458672e-07,
      "loss": 0.7116,
      "step": 4391
    },
    {
      "epoch": 0.4341744309616193,
      "grad_norm": 4.605759831314139,
      "learning_rate": 1.258793269983369e-07,
      "loss": 0.7276,
      "step": 4392
    },
    {
      "epoch": 0.4342732867063737,
      "grad_norm": 3.5941489036446708,
      "learning_rate": 1.2584839541854387e-07,
      "loss": 0.7143,
      "step": 4393
    },
    {
      "epoch": 0.4343721424511282,
      "grad_norm": 5.184211638820875,
      "learning_rate": 1.2581746118837915e-07,
      "loss": 0.8074,
      "step": 4394
    },
    {
      "epoch": 0.43447099819588264,
      "grad_norm": 6.022482001726597,
      "learning_rate": 1.2578652431101465e-07,
      "loss": 0.7228,
      "step": 4395
    },
    {
      "epoch": 0.43456985394063713,
      "grad_norm": 2.749775118282979,
      "learning_rate": 1.2575558478962246e-07,
      "loss": 0.7878,
      "step": 4396
    },
    {
      "epoch": 0.4346687096853916,
      "grad_norm": 3.283533919143582,
      "learning_rate": 1.2572464262737493e-07,
      "loss": 0.7669,
      "step": 4397
    },
    {
      "epoch": 0.43476756543014605,
      "grad_norm": 4.422052847420881,
      "learning_rate": 1.256936978274448e-07,
      "loss": 0.622,
      "step": 4398
    },
    {
      "epoch": 0.43486642117490054,
      "grad_norm": 3.458444463914515,
      "learning_rate": 1.2566275039300497e-07,
      "loss": 0.6919,
      "step": 4399
    },
    {
      "epoch": 0.434965276919655,
      "grad_norm": 3.2184056055563612,
      "learning_rate": 1.2563180032722864e-07,
      "loss": 0.7163,
      "step": 4400
    },
    {
      "epoch": 0.43506413266440946,
      "grad_norm": 5.577293245462467,
      "learning_rate": 1.2560084763328925e-07,
      "loss": 0.7796,
      "step": 4401
    },
    {
      "epoch": 0.43516298840916395,
      "grad_norm": 4.979559465331439,
      "learning_rate": 1.2556989231436062e-07,
      "loss": 0.7381,
      "step": 4402
    },
    {
      "epoch": 0.4352618441539184,
      "grad_norm": 4.8979629830041445,
      "learning_rate": 1.255389343736167e-07,
      "loss": 0.773,
      "step": 4403
    },
    {
      "epoch": 0.4353606998986729,
      "grad_norm": 4.96966890020378,
      "learning_rate": 1.2550797381423175e-07,
      "loss": 0.7652,
      "step": 4404
    },
    {
      "epoch": 0.4354595556434273,
      "grad_norm": 10.278301396615241,
      "learning_rate": 1.2547701063938037e-07,
      "loss": 0.7334,
      "step": 4405
    },
    {
      "epoch": 0.4355584113881818,
      "grad_norm": 3.0367457589605977,
      "learning_rate": 1.2544604485223738e-07,
      "loss": 0.6099,
      "step": 4406
    },
    {
      "epoch": 0.4356572671329363,
      "grad_norm": 4.239197505564717,
      "learning_rate": 1.2541507645597785e-07,
      "loss": 0.734,
      "step": 4407
    },
    {
      "epoch": 0.4357561228776907,
      "grad_norm": 8.290412218083139,
      "learning_rate": 1.253841054537771e-07,
      "loss": 0.6962,
      "step": 4408
    },
    {
      "epoch": 0.4358549786224452,
      "grad_norm": 3.9414910788371187,
      "learning_rate": 1.253531318488108e-07,
      "loss": 0.7629,
      "step": 4409
    },
    {
      "epoch": 0.43595383436719964,
      "grad_norm": 4.228695858547385,
      "learning_rate": 1.2532215564425484e-07,
      "loss": 0.6836,
      "step": 4410
    },
    {
      "epoch": 0.43605269011195413,
      "grad_norm": 4.342870926769546,
      "learning_rate": 1.2529117684328529e-07,
      "loss": 0.6812,
      "step": 4411
    },
    {
      "epoch": 0.4361515458567086,
      "grad_norm": 8.621313403882251,
      "learning_rate": 1.2526019544907865e-07,
      "loss": 0.6475,
      "step": 4412
    },
    {
      "epoch": 0.43625040160146306,
      "grad_norm": 3.2162120902575064,
      "learning_rate": 1.2522921146481157e-07,
      "loss": 0.7607,
      "step": 4413
    },
    {
      "epoch": 0.43634925734621755,
      "grad_norm": 24.510036779593257,
      "learning_rate": 1.2519822489366102e-07,
      "loss": 0.6977,
      "step": 4414
    },
    {
      "epoch": 0.436448113090972,
      "grad_norm": 4.100377435496341,
      "learning_rate": 1.2516723573880424e-07,
      "loss": 0.6785,
      "step": 4415
    },
    {
      "epoch": 0.43654696883572647,
      "grad_norm": 3.640739682116871,
      "learning_rate": 1.2513624400341865e-07,
      "loss": 0.7524,
      "step": 4416
    },
    {
      "epoch": 0.43664582458048096,
      "grad_norm": 3.110513398577446,
      "learning_rate": 1.25105249690682e-07,
      "loss": 0.6609,
      "step": 4417
    },
    {
      "epoch": 0.4367446803252354,
      "grad_norm": 8.1246429261824,
      "learning_rate": 1.2507425280377237e-07,
      "loss": 0.6467,
      "step": 4418
    },
    {
      "epoch": 0.4368435360699899,
      "grad_norm": 3.8303794823742066,
      "learning_rate": 1.2504325334586794e-07,
      "loss": 0.6186,
      "step": 4419
    },
    {
      "epoch": 0.4369423918147443,
      "grad_norm": 5.064371279810979,
      "learning_rate": 1.2501225132014732e-07,
      "loss": 0.7728,
      "step": 4420
    },
    {
      "epoch": 0.4370412475594988,
      "grad_norm": 4.406913377961118,
      "learning_rate": 1.2498124672978926e-07,
      "loss": 0.694,
      "step": 4421
    },
    {
      "epoch": 0.4371401033042533,
      "grad_norm": 8.664113637523895,
      "learning_rate": 1.249502395779729e-07,
      "loss": 0.7107,
      "step": 4422
    },
    {
      "epoch": 0.4372389590490077,
      "grad_norm": 3.413037520587641,
      "learning_rate": 1.2491922986787744e-07,
      "loss": 0.6398,
      "step": 4423
    },
    {
      "epoch": 0.4373378147937622,
      "grad_norm": 3.1169176984800404,
      "learning_rate": 1.2488821760268257e-07,
      "loss": 0.7062,
      "step": 4424
    },
    {
      "epoch": 0.43743667053851665,
      "grad_norm": 7.6481769376041955,
      "learning_rate": 1.2485720278556817e-07,
      "loss": 0.7161,
      "step": 4425
    },
    {
      "epoch": 0.43753552628327114,
      "grad_norm": 3.7604710815942957,
      "learning_rate": 1.2482618541971421e-07,
      "loss": 0.6928,
      "step": 4426
    },
    {
      "epoch": 0.4376343820280256,
      "grad_norm": 4.666300995906099,
      "learning_rate": 1.247951655083012e-07,
      "loss": 0.7924,
      "step": 4427
    },
    {
      "epoch": 0.43773323777278006,
      "grad_norm": 5.155820412847689,
      "learning_rate": 1.247641430545097e-07,
      "loss": 0.6049,
      "step": 4428
    },
    {
      "epoch": 0.43783209351753455,
      "grad_norm": 4.287211360862099,
      "learning_rate": 1.2473311806152063e-07,
      "loss": 0.8135,
      "step": 4429
    },
    {
      "epoch": 0.437930949262289,
      "grad_norm": 5.266649361322913,
      "learning_rate": 1.2470209053251512e-07,
      "loss": 0.7076,
      "step": 4430
    },
    {
      "epoch": 0.43802980500704347,
      "grad_norm": 15.456491604262355,
      "learning_rate": 1.2467106047067463e-07,
      "loss": 0.7349,
      "step": 4431
    },
    {
      "epoch": 0.43812866075179796,
      "grad_norm": 6.295204589479301,
      "learning_rate": 1.2464002787918081e-07,
      "loss": 0.746,
      "step": 4432
    },
    {
      "epoch": 0.4382275164965524,
      "grad_norm": 4.7418785713357785,
      "learning_rate": 1.2460899276121562e-07,
      "loss": 0.7396,
      "step": 4433
    },
    {
      "epoch": 0.4383263722413069,
      "grad_norm": 3.419410270782434,
      "learning_rate": 1.2457795511996121e-07,
      "loss": 0.66,
      "step": 4434
    },
    {
      "epoch": 0.4384252279860613,
      "grad_norm": 3.9940326336933087,
      "learning_rate": 1.2454691495860007e-07,
      "loss": 0.7172,
      "step": 4435
    },
    {
      "epoch": 0.4385240837308158,
      "grad_norm": 5.622727285437912,
      "learning_rate": 1.2451587228031487e-07,
      "loss": 0.6655,
      "step": 4436
    },
    {
      "epoch": 0.4386229394755703,
      "grad_norm": 5.53886105315408,
      "learning_rate": 1.244848270882886e-07,
      "loss": 0.6859,
      "step": 4437
    },
    {
      "epoch": 0.43872179522032473,
      "grad_norm": 3.9429385985391225,
      "learning_rate": 1.244537793857045e-07,
      "loss": 0.6745,
      "step": 4438
    },
    {
      "epoch": 0.4388206509650792,
      "grad_norm": 5.9124074971423966,
      "learning_rate": 1.2442272917574602e-07,
      "loss": 0.784,
      "step": 4439
    },
    {
      "epoch": 0.43891950670983365,
      "grad_norm": 4.870215060491875,
      "learning_rate": 1.2439167646159697e-07,
      "loss": 0.6839,
      "step": 4440
    },
    {
      "epoch": 0.43901836245458814,
      "grad_norm": 4.816482647965505,
      "learning_rate": 1.2436062124644126e-07,
      "loss": 0.6599,
      "step": 4441
    },
    {
      "epoch": 0.43911721819934263,
      "grad_norm": 8.201739083063268,
      "learning_rate": 1.2432956353346323e-07,
      "loss": 0.6821,
      "step": 4442
    },
    {
      "epoch": 0.43921607394409706,
      "grad_norm": 4.299313184498597,
      "learning_rate": 1.242985033258473e-07,
      "loss": 0.731,
      "step": 4443
    },
    {
      "epoch": 0.43931492968885155,
      "grad_norm": 5.105167828321615,
      "learning_rate": 1.242674406267783e-07,
      "loss": 0.7029,
      "step": 4444
    },
    {
      "epoch": 0.439413785433606,
      "grad_norm": 7.920638463598698,
      "learning_rate": 1.242363754394412e-07,
      "loss": 0.7742,
      "step": 4445
    },
    {
      "epoch": 0.4395126411783605,
      "grad_norm": 9.59679063558055,
      "learning_rate": 1.2420530776702136e-07,
      "loss": 0.6812,
      "step": 4446
    },
    {
      "epoch": 0.43961149692311496,
      "grad_norm": 8.216441295033869,
      "learning_rate": 1.2417423761270424e-07,
      "loss": 0.8131,
      "step": 4447
    },
    {
      "epoch": 0.4397103526678694,
      "grad_norm": 6.585632555768356,
      "learning_rate": 1.2414316497967565e-07,
      "loss": 0.7246,
      "step": 4448
    },
    {
      "epoch": 0.4398092084126239,
      "grad_norm": 17.450940592278886,
      "learning_rate": 1.2411208987112164e-07,
      "loss": 0.7396,
      "step": 4449
    },
    {
      "epoch": 0.4399080641573783,
      "grad_norm": 3.6426803671272485,
      "learning_rate": 1.2408101229022848e-07,
      "loss": 0.7118,
      "step": 4450
    },
    {
      "epoch": 0.4400069199021328,
      "grad_norm": 7.878755847522421,
      "learning_rate": 1.2404993224018272e-07,
      "loss": 0.7433,
      "step": 4451
    },
    {
      "epoch": 0.4401057756468873,
      "grad_norm": 4.024591844987845,
      "learning_rate": 1.2401884972417122e-07,
      "loss": 0.755,
      "step": 4452
    },
    {
      "epoch": 0.44020463139164173,
      "grad_norm": 4.380316553528367,
      "learning_rate": 1.2398776474538092e-07,
      "loss": 0.6621,
      "step": 4453
    },
    {
      "epoch": 0.4403034871363962,
      "grad_norm": 3.4616190007493297,
      "learning_rate": 1.2395667730699923e-07,
      "loss": 0.5833,
      "step": 4454
    },
    {
      "epoch": 0.44040234288115065,
      "grad_norm": 5.212395189633108,
      "learning_rate": 1.2392558741221364e-07,
      "loss": 0.7293,
      "step": 4455
    },
    {
      "epoch": 0.44050119862590514,
      "grad_norm": 4.823604756138332,
      "learning_rate": 1.2389449506421199e-07,
      "loss": 0.696,
      "step": 4456
    },
    {
      "epoch": 0.44060005437065963,
      "grad_norm": 3.7280142651962924,
      "learning_rate": 1.2386340026618235e-07,
      "loss": 0.7675,
      "step": 4457
    },
    {
      "epoch": 0.44069891011541407,
      "grad_norm": 6.458838267961823,
      "learning_rate": 1.2383230302131305e-07,
      "loss": 0.6742,
      "step": 4458
    },
    {
      "epoch": 0.44079776586016856,
      "grad_norm": 4.45332170936143,
      "learning_rate": 1.2380120333279258e-07,
      "loss": 0.6878,
      "step": 4459
    },
    {
      "epoch": 0.440896621604923,
      "grad_norm": 6.764293359217026,
      "learning_rate": 1.2377010120380984e-07,
      "loss": 0.78,
      "step": 4460
    },
    {
      "epoch": 0.4409954773496775,
      "grad_norm": 4.89660273082407,
      "learning_rate": 1.2373899663755382e-07,
      "loss": 0.7385,
      "step": 4461
    },
    {
      "epoch": 0.44109433309443197,
      "grad_norm": 16.090459728839996,
      "learning_rate": 1.237078896372139e-07,
      "loss": 0.6731,
      "step": 4462
    },
    {
      "epoch": 0.4411931888391864,
      "grad_norm": 5.6724518325023965,
      "learning_rate": 1.236767802059796e-07,
      "loss": 0.6869,
      "step": 4463
    },
    {
      "epoch": 0.4412920445839409,
      "grad_norm": 5.453777922367036,
      "learning_rate": 1.2364566834704074e-07,
      "loss": 0.7756,
      "step": 4464
    },
    {
      "epoch": 0.4413909003286953,
      "grad_norm": 2.9047569178020227,
      "learning_rate": 1.2361455406358738e-07,
      "loss": 0.6542,
      "step": 4465
    },
    {
      "epoch": 0.4414897560734498,
      "grad_norm": 4.968953442224472,
      "learning_rate": 1.2358343735880988e-07,
      "loss": 0.7105,
      "step": 4466
    },
    {
      "epoch": 0.4415886118182043,
      "grad_norm": 10.090350201393843,
      "learning_rate": 1.235523182358987e-07,
      "loss": 0.7127,
      "step": 4467
    },
    {
      "epoch": 0.44168746756295874,
      "grad_norm": 4.329821831988494,
      "learning_rate": 1.2352119669804476e-07,
      "loss": 0.7074,
      "step": 4468
    },
    {
      "epoch": 0.4417863233077132,
      "grad_norm": 7.269967251474778,
      "learning_rate": 1.2349007274843905e-07,
      "loss": 0.7071,
      "step": 4469
    },
    {
      "epoch": 0.4418851790524677,
      "grad_norm": 4.239048571706389,
      "learning_rate": 1.2345894639027285e-07,
      "loss": 0.6525,
      "step": 4470
    },
    {
      "epoch": 0.44198403479722215,
      "grad_norm": 5.445068191429476,
      "learning_rate": 1.2342781762673777e-07,
      "loss": 0.8434,
      "step": 4471
    },
    {
      "epoch": 0.44208289054197664,
      "grad_norm": 8.43378628827867,
      "learning_rate": 1.2339668646102558e-07,
      "loss": 0.6469,
      "step": 4472
    },
    {
      "epoch": 0.44218174628673107,
      "grad_norm": 4.462925830105942,
      "learning_rate": 1.2336555289632832e-07,
      "loss": 0.7454,
      "step": 4473
    },
    {
      "epoch": 0.44228060203148556,
      "grad_norm": 4.1501502401287915,
      "learning_rate": 1.2333441693583825e-07,
      "loss": 0.712,
      "step": 4474
    },
    {
      "epoch": 0.44237945777624005,
      "grad_norm": 5.080971815573156,
      "learning_rate": 1.2330327858274795e-07,
      "loss": 0.6216,
      "step": 4475
    },
    {
      "epoch": 0.4424783135209945,
      "grad_norm": 3.5525067456010584,
      "learning_rate": 1.2327213784025019e-07,
      "loss": 0.7316,
      "step": 4476
    },
    {
      "epoch": 0.44257716926574897,
      "grad_norm": 3.7901045466999794,
      "learning_rate": 1.2324099471153797e-07,
      "loss": 0.8385,
      "step": 4477
    },
    {
      "epoch": 0.4426760250105034,
      "grad_norm": 4.829630274722525,
      "learning_rate": 1.2320984919980457e-07,
      "loss": 0.7073,
      "step": 4478
    },
    {
      "epoch": 0.4427748807552579,
      "grad_norm": 6.838625149691489,
      "learning_rate": 1.2317870130824352e-07,
      "loss": 0.6746,
      "step": 4479
    },
    {
      "epoch": 0.4428737365000124,
      "grad_norm": 3.752519521585151,
      "learning_rate": 1.2314755104004854e-07,
      "loss": 0.7264,
      "step": 4480
    },
    {
      "epoch": 0.4429725922447668,
      "grad_norm": 5.3986834053287005,
      "learning_rate": 1.2311639839841364e-07,
      "loss": 0.6827,
      "step": 4481
    },
    {
      "epoch": 0.4430714479895213,
      "grad_norm": 4.752032787902562,
      "learning_rate": 1.230852433865331e-07,
      "loss": 0.7504,
      "step": 4482
    },
    {
      "epoch": 0.44317030373427574,
      "grad_norm": 3.233549496532395,
      "learning_rate": 1.2305408600760136e-07,
      "loss": 0.6811,
      "step": 4483
    },
    {
      "epoch": 0.44326915947903023,
      "grad_norm": 3.3578525709958003,
      "learning_rate": 1.230229262648132e-07,
      "loss": 0.816,
      "step": 4484
    },
    {
      "epoch": 0.4433680152237847,
      "grad_norm": 4.017067039483065,
      "learning_rate": 1.2299176416136355e-07,
      "loss": 0.6925,
      "step": 4485
    },
    {
      "epoch": 0.44346687096853915,
      "grad_norm": 4.069350476945201,
      "learning_rate": 1.2296059970044766e-07,
      "loss": 0.6968,
      "step": 4486
    },
    {
      "epoch": 0.44356572671329364,
      "grad_norm": 3.103115312144985,
      "learning_rate": 1.2292943288526093e-07,
      "loss": 0.7901,
      "step": 4487
    },
    {
      "epoch": 0.4436645824580481,
      "grad_norm": 4.215047374448014,
      "learning_rate": 1.228982637189991e-07,
      "loss": 0.6818,
      "step": 4488
    },
    {
      "epoch": 0.44376343820280256,
      "grad_norm": 6.348230034396298,
      "learning_rate": 1.228670922048581e-07,
      "loss": 0.7571,
      "step": 4489
    },
    {
      "epoch": 0.44386229394755705,
      "grad_norm": 5.590526277076746,
      "learning_rate": 1.2283591834603408e-07,
      "loss": 0.7182,
      "step": 4490
    },
    {
      "epoch": 0.4439611496923115,
      "grad_norm": 4.273379647594977,
      "learning_rate": 1.2280474214572353e-07,
      "loss": 0.569,
      "step": 4491
    },
    {
      "epoch": 0.444060005437066,
      "grad_norm": 3.428113293724736,
      "learning_rate": 1.2277356360712303e-07,
      "loss": 0.6703,
      "step": 4492
    },
    {
      "epoch": 0.4441588611818204,
      "grad_norm": 5.759277166479194,
      "learning_rate": 1.2274238273342953e-07,
      "loss": 0.7263,
      "step": 4493
    },
    {
      "epoch": 0.4442577169265749,
      "grad_norm": 3.9582515686031257,
      "learning_rate": 1.2271119952784017e-07,
      "loss": 0.6925,
      "step": 4494
    },
    {
      "epoch": 0.4443565726713294,
      "grad_norm": 5.324509422002788,
      "learning_rate": 1.2268001399355231e-07,
      "loss": 0.6739,
      "step": 4495
    },
    {
      "epoch": 0.4444554284160838,
      "grad_norm": 4.695072901320314,
      "learning_rate": 1.2264882613376358e-07,
      "loss": 0.7827,
      "step": 4496
    },
    {
      "epoch": 0.4445542841608383,
      "grad_norm": 5.140840930603065,
      "learning_rate": 1.2261763595167182e-07,
      "loss": 0.7352,
      "step": 4497
    },
    {
      "epoch": 0.44465313990559274,
      "grad_norm": 6.9231287155565076,
      "learning_rate": 1.2258644345047507e-07,
      "loss": 0.6598,
      "step": 4498
    },
    {
      "epoch": 0.44475199565034723,
      "grad_norm": 3.56066444947844,
      "learning_rate": 1.225552486333718e-07,
      "loss": 0.7522,
      "step": 4499
    },
    {
      "epoch": 0.4448508513951017,
      "grad_norm": 5.901764905112503,
      "learning_rate": 1.2252405150356042e-07,
      "loss": 0.701,
      "step": 4500
    },
    {
      "epoch": 0.44494970713985615,
      "grad_norm": 4.074530756891591,
      "learning_rate": 1.2249285206423986e-07,
      "loss": 0.7118,
      "step": 4501
    },
    {
      "epoch": 0.44504856288461064,
      "grad_norm": 3.0701122737352353,
      "learning_rate": 1.2246165031860912e-07,
      "loss": 0.7684,
      "step": 4502
    },
    {
      "epoch": 0.4451474186293651,
      "grad_norm": 4.945021383246274,
      "learning_rate": 1.2243044626986744e-07,
      "loss": 0.7291,
      "step": 4503
    },
    {
      "epoch": 0.44524627437411957,
      "grad_norm": 3.6497231508580543,
      "learning_rate": 1.223992399212144e-07,
      "loss": 0.7627,
      "step": 4504
    },
    {
      "epoch": 0.44534513011887406,
      "grad_norm": 25.357354651331907,
      "learning_rate": 1.2236803127584968e-07,
      "loss": 0.7924,
      "step": 4505
    },
    {
      "epoch": 0.4454439858636285,
      "grad_norm": 4.18909074711267,
      "learning_rate": 1.2233682033697332e-07,
      "loss": 0.6177,
      "step": 4506
    },
    {
      "epoch": 0.445542841608383,
      "grad_norm": 3.660823984173389,
      "learning_rate": 1.223056071077855e-07,
      "loss": 0.6863,
      "step": 4507
    },
    {
      "epoch": 0.4456416973531374,
      "grad_norm": 3.5894453950617353,
      "learning_rate": 1.2227439159148675e-07,
      "loss": 0.6828,
      "step": 4508
    },
    {
      "epoch": 0.4457405530978919,
      "grad_norm": 6.312791248000489,
      "learning_rate": 1.222431737912777e-07,
      "loss": 0.6865,
      "step": 4509
    },
    {
      "epoch": 0.4458394088426464,
      "grad_norm": 11.299661389221923,
      "learning_rate": 1.2221195371035928e-07,
      "loss": 0.6815,
      "step": 4510
    },
    {
      "epoch": 0.4459382645874008,
      "grad_norm": 3.0943843287373647,
      "learning_rate": 1.2218073135193266e-07,
      "loss": 0.7197,
      "step": 4511
    },
    {
      "epoch": 0.4460371203321553,
      "grad_norm": 11.010272172576105,
      "learning_rate": 1.2214950671919924e-07,
      "loss": 0.6673,
      "step": 4512
    },
    {
      "epoch": 0.44613597607690975,
      "grad_norm": 4.146332121678589,
      "learning_rate": 1.2211827981536063e-07,
      "loss": 0.6509,
      "step": 4513
    },
    {
      "epoch": 0.44623483182166424,
      "grad_norm": 6.116564283184518,
      "learning_rate": 1.2208705064361867e-07,
      "loss": 0.6971,
      "step": 4514
    },
    {
      "epoch": 0.4463336875664187,
      "grad_norm": 3.906039076014608,
      "learning_rate": 1.220558192071755e-07,
      "loss": 0.6471,
      "step": 4515
    },
    {
      "epoch": 0.44643254331117316,
      "grad_norm": 3.4786839706551143,
      "learning_rate": 1.2202458550923341e-07,
      "loss": 0.737,
      "step": 4516
    },
    {
      "epoch": 0.44653139905592765,
      "grad_norm": 8.249372936557851,
      "learning_rate": 1.2199334955299493e-07,
      "loss": 0.7722,
      "step": 4517
    },
    {
      "epoch": 0.4466302548006821,
      "grad_norm": 16.98135624043647,
      "learning_rate": 1.219621113416629e-07,
      "loss": 0.7517,
      "step": 4518
    },
    {
      "epoch": 0.44672911054543657,
      "grad_norm": 5.345593157287908,
      "learning_rate": 1.219308708784403e-07,
      "loss": 0.7465,
      "step": 4519
    },
    {
      "epoch": 0.44682796629019106,
      "grad_norm": 3.6997305097740543,
      "learning_rate": 1.2189962816653044e-07,
      "loss": 0.7379,
      "step": 4520
    },
    {
      "epoch": 0.4469268220349455,
      "grad_norm": 5.59564178855951,
      "learning_rate": 1.2186838320913673e-07,
      "loss": 0.7738,
      "step": 4521
    },
    {
      "epoch": 0.4470256777797,
      "grad_norm": 4.913077706077855,
      "learning_rate": 1.218371360094629e-07,
      "loss": 0.6352,
      "step": 4522
    },
    {
      "epoch": 0.4471245335244544,
      "grad_norm": 4.751800492601646,
      "learning_rate": 1.218058865707129e-07,
      "loss": 0.7764,
      "step": 4523
    },
    {
      "epoch": 0.4472233892692089,
      "grad_norm": 4.640952698944985,
      "learning_rate": 1.2177463489609088e-07,
      "loss": 0.7188,
      "step": 4524
    },
    {
      "epoch": 0.4473222450139634,
      "grad_norm": 5.94554685699758,
      "learning_rate": 1.2174338098880123e-07,
      "loss": 0.7682,
      "step": 4525
    },
    {
      "epoch": 0.4474211007587178,
      "grad_norm": 3.2151806624893844,
      "learning_rate": 1.2171212485204862e-07,
      "loss": 0.7857,
      "step": 4526
    },
    {
      "epoch": 0.4475199565034723,
      "grad_norm": 4.497718292386858,
      "learning_rate": 1.216808664890379e-07,
      "loss": 0.7493,
      "step": 4527
    },
    {
      "epoch": 0.44761881224822675,
      "grad_norm": 4.801821268629588,
      "learning_rate": 1.2164960590297413e-07,
      "loss": 0.6905,
      "step": 4528
    },
    {
      "epoch": 0.44771766799298124,
      "grad_norm": 2.7608439931811604,
      "learning_rate": 1.2161834309706261e-07,
      "loss": 0.6767,
      "step": 4529
    },
    {
      "epoch": 0.4478165237377357,
      "grad_norm": 3.2188912957496405,
      "learning_rate": 1.2158707807450892e-07,
      "loss": 0.6571,
      "step": 4530
    },
    {
      "epoch": 0.44791537948249016,
      "grad_norm": 14.21015777898158,
      "learning_rate": 1.215558108385188e-07,
      "loss": 0.6582,
      "step": 4531
    },
    {
      "epoch": 0.44801423522724465,
      "grad_norm": 41.42054385999666,
      "learning_rate": 1.2152454139229823e-07,
      "loss": 0.6809,
      "step": 4532
    },
    {
      "epoch": 0.4481130909719991,
      "grad_norm": 2.881791183891332,
      "learning_rate": 1.2149326973905346e-07,
      "loss": 0.6467,
      "step": 4533
    },
    {
      "epoch": 0.4482119467167536,
      "grad_norm": 2.9942151798850047,
      "learning_rate": 1.2146199588199094e-07,
      "loss": 0.6783,
      "step": 4534
    },
    {
      "epoch": 0.44831080246150806,
      "grad_norm": 4.9079634483442,
      "learning_rate": 1.214307198243173e-07,
      "loss": 0.715,
      "step": 4535
    },
    {
      "epoch": 0.4484096582062625,
      "grad_norm": 14.526484974452057,
      "learning_rate": 1.2139944156923946e-07,
      "loss": 0.685,
      "step": 4536
    },
    {
      "epoch": 0.448508513951017,
      "grad_norm": 4.034172284609909,
      "learning_rate": 1.213681611199646e-07,
      "loss": 0.7095,
      "step": 4537
    },
    {
      "epoch": 0.4486073696957714,
      "grad_norm": 3.6479634060739947,
      "learning_rate": 1.213368784797e-07,
      "loss": 0.6842,
      "step": 4538
    },
    {
      "epoch": 0.4487062254405259,
      "grad_norm": 9.189295409900012,
      "learning_rate": 1.2130559365165328e-07,
      "loss": 0.696,
      "step": 4539
    },
    {
      "epoch": 0.4488050811852804,
      "grad_norm": 5.892721199056226,
      "learning_rate": 1.2127430663903215e-07,
      "loss": 0.7006,
      "step": 4540
    },
    {
      "epoch": 0.44890393693003483,
      "grad_norm": 4.371711302071998,
      "learning_rate": 1.2124301744504472e-07,
      "loss": 0.6916,
      "step": 4541
    },
    {
      "epoch": 0.4490027926747893,
      "grad_norm": 3.4784343338603314,
      "learning_rate": 1.2121172607289923e-07,
      "loss": 0.6284,
      "step": 4542
    },
    {
      "epoch": 0.44910164841954375,
      "grad_norm": 3.8110529640568935,
      "learning_rate": 1.2118043252580408e-07,
      "loss": 0.6914,
      "step": 4543
    },
    {
      "epoch": 0.44920050416429824,
      "grad_norm": 3.172560418074783,
      "learning_rate": 1.2114913680696804e-07,
      "loss": 0.7162,
      "step": 4544
    },
    {
      "epoch": 0.44929935990905273,
      "grad_norm": 5.610711696262875,
      "learning_rate": 1.2111783891959996e-07,
      "loss": 0.7484,
      "step": 4545
    },
    {
      "epoch": 0.44939821565380716,
      "grad_norm": 5.254378421284161,
      "learning_rate": 1.2108653886690905e-07,
      "loss": 0.6584,
      "step": 4546
    },
    {
      "epoch": 0.44949707139856165,
      "grad_norm": 6.136010070680098,
      "learning_rate": 1.2105523665210462e-07,
      "loss": 0.8016,
      "step": 4547
    },
    {
      "epoch": 0.44959592714331614,
      "grad_norm": 5.130816913143164,
      "learning_rate": 1.2102393227839625e-07,
      "loss": 0.7674,
      "step": 4548
    },
    {
      "epoch": 0.4496947828880706,
      "grad_norm": 3.228551260776028,
      "learning_rate": 1.2099262574899375e-07,
      "loss": 0.7629,
      "step": 4549
    },
    {
      "epoch": 0.44979363863282507,
      "grad_norm": 3.408691378054981,
      "learning_rate": 1.2096131706710714e-07,
      "loss": 0.6306,
      "step": 4550
    },
    {
      "epoch": 0.4498924943775795,
      "grad_norm": 4.5897149865060465,
      "learning_rate": 1.2093000623594667e-07,
      "loss": 0.8141,
      "step": 4551
    },
    {
      "epoch": 0.449991350122334,
      "grad_norm": 17.836618983725316,
      "learning_rate": 1.208986932587228e-07,
      "loss": 0.7886,
      "step": 4552
    },
    {
      "epoch": 0.4500902058670885,
      "grad_norm": 7.362429932192727,
      "learning_rate": 1.208673781386462e-07,
      "loss": 0.7546,
      "step": 4553
    },
    {
      "epoch": 0.4501890616118429,
      "grad_norm": 8.47689853549441,
      "learning_rate": 1.208360608789278e-07,
      "loss": 0.8327,
      "step": 4554
    },
    {
      "epoch": 0.4502879173565974,
      "grad_norm": 3.49563267163135,
      "learning_rate": 1.208047414827787e-07,
      "loss": 0.7024,
      "step": 4555
    },
    {
      "epoch": 0.45038677310135183,
      "grad_norm": 5.937133594674317,
      "learning_rate": 1.2077341995341028e-07,
      "loss": 0.7006,
      "step": 4556
    },
    {
      "epoch": 0.4504856288461063,
      "grad_norm": 5.930105528725069,
      "learning_rate": 1.2074209629403406e-07,
      "loss": 0.6942,
      "step": 4557
    },
    {
      "epoch": 0.4505844845908608,
      "grad_norm": 5.976429960786998,
      "learning_rate": 1.207107705078618e-07,
      "loss": 0.6788,
      "step": 4558
    },
    {
      "epoch": 0.45068334033561525,
      "grad_norm": 3.4803268648109347,
      "learning_rate": 1.2067944259810556e-07,
      "loss": 0.6649,
      "step": 4559
    },
    {
      "epoch": 0.45078219608036973,
      "grad_norm": 4.685447139675545,
      "learning_rate": 1.2064811256797754e-07,
      "loss": 0.7955,
      "step": 4560
    },
    {
      "epoch": 0.45088105182512417,
      "grad_norm": 3.6730342355862935,
      "learning_rate": 1.2061678042069013e-07,
      "loss": 0.7432,
      "step": 4561
    },
    {
      "epoch": 0.45097990756987866,
      "grad_norm": 6.016444148595582,
      "learning_rate": 1.20585446159456e-07,
      "loss": 0.7102,
      "step": 4562
    },
    {
      "epoch": 0.45107876331463315,
      "grad_norm": 3.7047181913074168,
      "learning_rate": 1.2055410978748805e-07,
      "loss": 0.5793,
      "step": 4563
    },
    {
      "epoch": 0.4511776190593876,
      "grad_norm": 5.275003338565461,
      "learning_rate": 1.2052277130799933e-07,
      "loss": 0.693,
      "step": 4564
    },
    {
      "epoch": 0.45127647480414207,
      "grad_norm": 6.697750758089831,
      "learning_rate": 1.2049143072420313e-07,
      "loss": 0.7331,
      "step": 4565
    },
    {
      "epoch": 0.4513753305488965,
      "grad_norm": 8.751222439336527,
      "learning_rate": 1.20460088039313e-07,
      "loss": 0.8122,
      "step": 4566
    },
    {
      "epoch": 0.451474186293651,
      "grad_norm": 14.816459236312584,
      "learning_rate": 1.2042874325654264e-07,
      "loss": 0.6501,
      "step": 4567
    },
    {
      "epoch": 0.4515730420384055,
      "grad_norm": 3.2785204461265542,
      "learning_rate": 1.2039739637910603e-07,
      "loss": 0.5907,
      "step": 4568
    },
    {
      "epoch": 0.4516718977831599,
      "grad_norm": 3.4539430209643864,
      "learning_rate": 1.2036604741021723e-07,
      "loss": 0.7262,
      "step": 4569
    },
    {
      "epoch": 0.4517707535279144,
      "grad_norm": 10.550135756384824,
      "learning_rate": 1.2033469635309075e-07,
      "loss": 0.7679,
      "step": 4570
    },
    {
      "epoch": 0.45186960927266884,
      "grad_norm": 3.915627998302014,
      "learning_rate": 1.2030334321094108e-07,
      "loss": 0.8026,
      "step": 4571
    },
    {
      "epoch": 0.4519684650174233,
      "grad_norm": 3.4131574715448774,
      "learning_rate": 1.202719879869831e-07,
      "loss": 0.6477,
      "step": 4572
    },
    {
      "epoch": 0.4520673207621778,
      "grad_norm": 3.141244801962473,
      "learning_rate": 1.2024063068443177e-07,
      "loss": 0.674,
      "step": 4573
    },
    {
      "epoch": 0.45216617650693225,
      "grad_norm": 3.614573880719883,
      "learning_rate": 1.2020927130650233e-07,
      "loss": 0.6978,
      "step": 4574
    },
    {
      "epoch": 0.45226503225168674,
      "grad_norm": 9.724336633908523,
      "learning_rate": 1.2017790985641023e-07,
      "loss": 0.8219,
      "step": 4575
    },
    {
      "epoch": 0.45236388799644117,
      "grad_norm": 6.601043032691873,
      "learning_rate": 1.201465463373711e-07,
      "loss": 0.8092,
      "step": 4576
    },
    {
      "epoch": 0.45246274374119566,
      "grad_norm": 8.635980617435617,
      "learning_rate": 1.2011518075260082e-07,
      "loss": 0.6653,
      "step": 4577
    },
    {
      "epoch": 0.45256159948595015,
      "grad_norm": 11.299400332656445,
      "learning_rate": 1.200838131053155e-07,
      "loss": 0.6089,
      "step": 4578
    },
    {
      "epoch": 0.4526604552307046,
      "grad_norm": 3.9505272191270815,
      "learning_rate": 1.2005244339873143e-07,
      "loss": 0.6578,
      "step": 4579
    },
    {
      "epoch": 0.4527593109754591,
      "grad_norm": 3.7730488992274687,
      "learning_rate": 1.20021071636065e-07,
      "loss": 0.6204,
      "step": 4580
    },
    {
      "epoch": 0.4528581667202135,
      "grad_norm": 28.647370673967924,
      "learning_rate": 1.199896978205331e-07,
      "loss": 0.7644,
      "step": 4581
    },
    {
      "epoch": 0.452957022464968,
      "grad_norm": 3.5440750881373666,
      "learning_rate": 1.199583219553525e-07,
      "loss": 0.7213,
      "step": 4582
    },
    {
      "epoch": 0.4530558782097225,
      "grad_norm": 7.096377642374201,
      "learning_rate": 1.1992694404374043e-07,
      "loss": 0.7145,
      "step": 4583
    },
    {
      "epoch": 0.4531547339544769,
      "grad_norm": 3.6802297303889686,
      "learning_rate": 1.1989556408891412e-07,
      "loss": 0.7604,
      "step": 4584
    },
    {
      "epoch": 0.4532535896992314,
      "grad_norm": 4.0278272454309105,
      "learning_rate": 1.1986418209409124e-07,
      "loss": 0.752,
      "step": 4585
    },
    {
      "epoch": 0.45335244544398584,
      "grad_norm": 3.8418144310224744,
      "learning_rate": 1.198327980624895e-07,
      "loss": 0.7673,
      "step": 4586
    },
    {
      "epoch": 0.45345130118874033,
      "grad_norm": 3.92102209897504,
      "learning_rate": 1.1980141199732688e-07,
      "loss": 0.7214,
      "step": 4587
    },
    {
      "epoch": 0.4535501569334948,
      "grad_norm": 3.9674124813499976,
      "learning_rate": 1.1977002390182155e-07,
      "loss": 0.7375,
      "step": 4588
    },
    {
      "epoch": 0.45364901267824925,
      "grad_norm": 3.528700138324538,
      "learning_rate": 1.1973863377919186e-07,
      "loss": 0.6958,
      "step": 4589
    },
    {
      "epoch": 0.45374786842300374,
      "grad_norm": 3.1044076872644277,
      "learning_rate": 1.1970724163265648e-07,
      "loss": 0.6193,
      "step": 4590
    },
    {
      "epoch": 0.4538467241677582,
      "grad_norm": 27.6871821716826,
      "learning_rate": 1.1967584746543417e-07,
      "loss": 0.6438,
      "step": 4591
    },
    {
      "epoch": 0.45394557991251266,
      "grad_norm": 14.786972188457668,
      "learning_rate": 1.1964445128074394e-07,
      "loss": 0.6226,
      "step": 4592
    },
    {
      "epoch": 0.45404443565726715,
      "grad_norm": 5.2853616059093165,
      "learning_rate": 1.19613053081805e-07,
      "loss": 0.6831,
      "step": 4593
    },
    {
      "epoch": 0.4541432914020216,
      "grad_norm": 4.845051184201071,
      "learning_rate": 1.195816528718368e-07,
      "loss": 0.743,
      "step": 4594
    },
    {
      "epoch": 0.4542421471467761,
      "grad_norm": 5.012754119941879,
      "learning_rate": 1.195502506540589e-07,
      "loss": 0.6853,
      "step": 4595
    },
    {
      "epoch": 0.4543410028915305,
      "grad_norm": 12.764780735693048,
      "learning_rate": 1.1951884643169124e-07,
      "loss": 0.7892,
      "step": 4596
    },
    {
      "epoch": 0.454439858636285,
      "grad_norm": 6.09271412284658,
      "learning_rate": 1.1948744020795376e-07,
      "loss": 0.6677,
      "step": 4597
    },
    {
      "epoch": 0.4545387143810395,
      "grad_norm": 3.989347893709507,
      "learning_rate": 1.1945603198606675e-07,
      "loss": 0.7573,
      "step": 4598
    },
    {
      "epoch": 0.4546375701257939,
      "grad_norm": 3.4152608338685235,
      "learning_rate": 1.1942462176925068e-07,
      "loss": 0.6939,
      "step": 4599
    },
    {
      "epoch": 0.4547364258705484,
      "grad_norm": 5.498366042000261,
      "learning_rate": 1.1939320956072618e-07,
      "loss": 0.7263,
      "step": 4600
    },
    {
      "epoch": 0.45483528161530284,
      "grad_norm": 3.3929465033737736,
      "learning_rate": 1.193617953637141e-07,
      "loss": 0.768,
      "step": 4601
    },
    {
      "epoch": 0.45493413736005733,
      "grad_norm": 3.778416689658618,
      "learning_rate": 1.1933037918143549e-07,
      "loss": 0.7386,
      "step": 4602
    },
    {
      "epoch": 0.4550329931048118,
      "grad_norm": 4.082904268829284,
      "learning_rate": 1.1929896101711166e-07,
      "loss": 0.6219,
      "step": 4603
    },
    {
      "epoch": 0.45513184884956626,
      "grad_norm": 3.2315080196937855,
      "learning_rate": 1.1926754087396407e-07,
      "loss": 0.6956,
      "step": 4604
    },
    {
      "epoch": 0.45523070459432075,
      "grad_norm": 5.851598970621131,
      "learning_rate": 1.1923611875521437e-07,
      "loss": 0.6969,
      "step": 4605
    },
    {
      "epoch": 0.4553295603390752,
      "grad_norm": 3.4232129752488762,
      "learning_rate": 1.1920469466408441e-07,
      "loss": 0.7825,
      "step": 4606
    },
    {
      "epoch": 0.45542841608382967,
      "grad_norm": 3.590420034995068,
      "learning_rate": 1.1917326860379633e-07,
      "loss": 0.6751,
      "step": 4607
    },
    {
      "epoch": 0.45552727182858416,
      "grad_norm": 5.7855401702308225,
      "learning_rate": 1.1914184057757239e-07,
      "loss": 0.6886,
      "step": 4608
    },
    {
      "epoch": 0.4556261275733386,
      "grad_norm": 4.2818653453716795,
      "learning_rate": 1.1911041058863504e-07,
      "loss": 0.7755,
      "step": 4609
    },
    {
      "epoch": 0.4557249833180931,
      "grad_norm": 3.9186367535425735,
      "learning_rate": 1.1907897864020701e-07,
      "loss": 0.7954,
      "step": 4610
    },
    {
      "epoch": 0.4558238390628475,
      "grad_norm": 3.808290157360293,
      "learning_rate": 1.1904754473551116e-07,
      "loss": 0.7262,
      "step": 4611
    },
    {
      "epoch": 0.455922694807602,
      "grad_norm": 6.996252744032158,
      "learning_rate": 1.1901610887777054e-07,
      "loss": 0.8001,
      "step": 4612
    },
    {
      "epoch": 0.4560215505523565,
      "grad_norm": 3.681890634758645,
      "learning_rate": 1.1898467107020844e-07,
      "loss": 0.6396,
      "step": 4613
    },
    {
      "epoch": 0.4561204062971109,
      "grad_norm": 4.325822456743447,
      "learning_rate": 1.189532313160484e-07,
      "loss": 0.7736,
      "step": 4614
    },
    {
      "epoch": 0.4562192620418654,
      "grad_norm": 10.453009376026047,
      "learning_rate": 1.1892178961851407e-07,
      "loss": 0.7022,
      "step": 4615
    },
    {
      "epoch": 0.45631811778661985,
      "grad_norm": 2.898088069437076,
      "learning_rate": 1.1889034598082933e-07,
      "loss": 0.6342,
      "step": 4616
    },
    {
      "epoch": 0.45641697353137434,
      "grad_norm": 3.7583073972783856,
      "learning_rate": 1.1885890040621826e-07,
      "loss": 0.6708,
      "step": 4617
    },
    {
      "epoch": 0.4565158292761288,
      "grad_norm": 3.955420137703679,
      "learning_rate": 1.1882745289790514e-07,
      "loss": 0.654,
      "step": 4618
    },
    {
      "epoch": 0.45661468502088326,
      "grad_norm": 3.5779500549970877,
      "learning_rate": 1.1879600345911445e-07,
      "loss": 0.6767,
      "step": 4619
    },
    {
      "epoch": 0.45671354076563775,
      "grad_norm": 3.916301513297369,
      "learning_rate": 1.1876455209307086e-07,
      "loss": 0.7384,
      "step": 4620
    },
    {
      "epoch": 0.4568123965103922,
      "grad_norm": 6.507337440728105,
      "learning_rate": 1.1873309880299924e-07,
      "loss": 0.6897,
      "step": 4621
    },
    {
      "epoch": 0.45691125225514667,
      "grad_norm": 10.425992368475319,
      "learning_rate": 1.1870164359212468e-07,
      "loss": 0.7366,
      "step": 4622
    },
    {
      "epoch": 0.45701010799990116,
      "grad_norm": 4.663593769406385,
      "learning_rate": 1.1867018646367246e-07,
      "loss": 0.7409,
      "step": 4623
    },
    {
      "epoch": 0.4571089637446556,
      "grad_norm": 4.373052096860973,
      "learning_rate": 1.1863872742086798e-07,
      "loss": 0.711,
      "step": 4624
    },
    {
      "epoch": 0.4572078194894101,
      "grad_norm": 3.8927789876540717,
      "learning_rate": 1.1860726646693697e-07,
      "loss": 0.6951,
      "step": 4625
    },
    {
      "epoch": 0.4573066752341646,
      "grad_norm": 3.2092514675880235,
      "learning_rate": 1.1857580360510527e-07,
      "loss": 0.7611,
      "step": 4626
    },
    {
      "epoch": 0.457405530978919,
      "grad_norm": 4.0947723905829605,
      "learning_rate": 1.1854433883859894e-07,
      "loss": 0.734,
      "step": 4627
    },
    {
      "epoch": 0.4575043867236735,
      "grad_norm": 3.503449877220396,
      "learning_rate": 1.1851287217064414e-07,
      "loss": 0.6434,
      "step": 4628
    },
    {
      "epoch": 0.45760324246842793,
      "grad_norm": 20.035457869414834,
      "learning_rate": 1.1848140360446743e-07,
      "loss": 0.6888,
      "step": 4629
    },
    {
      "epoch": 0.4577020982131824,
      "grad_norm": 6.236922508272508,
      "learning_rate": 1.1844993314329541e-07,
      "loss": 0.6366,
      "step": 4630
    },
    {
      "epoch": 0.4578009539579369,
      "grad_norm": 14.72188144394107,
      "learning_rate": 1.1841846079035487e-07,
      "loss": 0.6729,
      "step": 4631
    },
    {
      "epoch": 0.45789980970269134,
      "grad_norm": 3.8204088593688055,
      "learning_rate": 1.1838698654887287e-07,
      "loss": 0.6545,
      "step": 4632
    },
    {
      "epoch": 0.45799866544744583,
      "grad_norm": 2.795099865429774,
      "learning_rate": 1.1835551042207665e-07,
      "loss": 0.6779,
      "step": 4633
    },
    {
      "epoch": 0.45809752119220026,
      "grad_norm": 10.711559894723669,
      "learning_rate": 1.183240324131936e-07,
      "loss": 0.6823,
      "step": 4634
    },
    {
      "epoch": 0.45819637693695475,
      "grad_norm": 3.5246733324598747,
      "learning_rate": 1.1829255252545132e-07,
      "loss": 0.6329,
      "step": 4635
    },
    {
      "epoch": 0.45829523268170924,
      "grad_norm": 5.157053081345313,
      "learning_rate": 1.1826107076207761e-07,
      "loss": 0.6977,
      "step": 4636
    },
    {
      "epoch": 0.4583940884264637,
      "grad_norm": 8.491817674752408,
      "learning_rate": 1.1822958712630047e-07,
      "loss": 0.6184,
      "step": 4637
    },
    {
      "epoch": 0.45849294417121816,
      "grad_norm": 8.31485096379298,
      "learning_rate": 1.1819810162134804e-07,
      "loss": 0.7409,
      "step": 4638
    },
    {
      "epoch": 0.4585917999159726,
      "grad_norm": 4.201103259407058,
      "learning_rate": 1.1816661425044876e-07,
      "loss": 0.6651,
      "step": 4639
    },
    {
      "epoch": 0.4586906556607271,
      "grad_norm": 3.0060358503292166,
      "learning_rate": 1.1813512501683116e-07,
      "loss": 0.728,
      "step": 4640
    },
    {
      "epoch": 0.4587895114054816,
      "grad_norm": 8.804458737932189,
      "learning_rate": 1.18103633923724e-07,
      "loss": 0.714,
      "step": 4641
    },
    {
      "epoch": 0.458888367150236,
      "grad_norm": 6.9604376842802855,
      "learning_rate": 1.180721409743562e-07,
      "loss": 0.6234,
      "step": 4642
    },
    {
      "epoch": 0.4589872228949905,
      "grad_norm": 11.346181740783173,
      "learning_rate": 1.1804064617195697e-07,
      "loss": 0.6969,
      "step": 4643
    },
    {
      "epoch": 0.45908607863974493,
      "grad_norm": 7.896040164038983,
      "learning_rate": 1.180091495197556e-07,
      "loss": 0.6288,
      "step": 4644
    },
    {
      "epoch": 0.4591849343844994,
      "grad_norm": 4.432438131744009,
      "learning_rate": 1.1797765102098158e-07,
      "loss": 0.7576,
      "step": 4645
    },
    {
      "epoch": 0.4592837901292539,
      "grad_norm": 3.816758736867726,
      "learning_rate": 1.1794615067886462e-07,
      "loss": 0.6632,
      "step": 4646
    },
    {
      "epoch": 0.45938264587400834,
      "grad_norm": 2.267710691897828,
      "learning_rate": 1.1791464849663466e-07,
      "loss": 0.5286,
      "step": 4647
    },
    {
      "epoch": 0.45948150161876283,
      "grad_norm": 5.1641920899183384,
      "learning_rate": 1.1788314447752175e-07,
      "loss": 0.7829,
      "step": 4648
    },
    {
      "epoch": 0.45958035736351727,
      "grad_norm": 3.7744862640296892,
      "learning_rate": 1.1785163862475616e-07,
      "loss": 0.6891,
      "step": 4649
    },
    {
      "epoch": 0.45967921310827176,
      "grad_norm": 3.0755400356752034,
      "learning_rate": 1.1782013094156835e-07,
      "loss": 0.7477,
      "step": 4650
    },
    {
      "epoch": 0.45977806885302624,
      "grad_norm": 9.077743933110929,
      "learning_rate": 1.1778862143118899e-07,
      "loss": 0.7725,
      "step": 4651
    },
    {
      "epoch": 0.4598769245977807,
      "grad_norm": 2.912147389523053,
      "learning_rate": 1.177571100968489e-07,
      "loss": 0.6833,
      "step": 4652
    },
    {
      "epoch": 0.45997578034253517,
      "grad_norm": 3.4453638112579648,
      "learning_rate": 1.1772559694177911e-07,
      "loss": 0.6483,
      "step": 4653
    },
    {
      "epoch": 0.4600746360872896,
      "grad_norm": 4.293927106449199,
      "learning_rate": 1.1769408196921082e-07,
      "loss": 0.6469,
      "step": 4654
    },
    {
      "epoch": 0.4601734918320441,
      "grad_norm": 12.590819462184028,
      "learning_rate": 1.1766256518237545e-07,
      "loss": 0.7062,
      "step": 4655
    },
    {
      "epoch": 0.4602723475767986,
      "grad_norm": 4.039522575073549,
      "learning_rate": 1.1763104658450453e-07,
      "loss": 0.6753,
      "step": 4656
    },
    {
      "epoch": 0.460371203321553,
      "grad_norm": 4.950722352279942,
      "learning_rate": 1.1759952617882987e-07,
      "loss": 0.7893,
      "step": 4657
    },
    {
      "epoch": 0.4604700590663075,
      "grad_norm": 4.2828131086538175,
      "learning_rate": 1.1756800396858341e-07,
      "loss": 0.6674,
      "step": 4658
    },
    {
      "epoch": 0.46056891481106194,
      "grad_norm": 6.606696125613661,
      "learning_rate": 1.1753647995699731e-07,
      "loss": 0.6599,
      "step": 4659
    },
    {
      "epoch": 0.4606677705558164,
      "grad_norm": 3.203459937806646,
      "learning_rate": 1.1750495414730386e-07,
      "loss": 0.7089,
      "step": 4660
    },
    {
      "epoch": 0.4607666263005709,
      "grad_norm": 8.774166201620362,
      "learning_rate": 1.1747342654273557e-07,
      "loss": 0.6957,
      "step": 4661
    },
    {
      "epoch": 0.46086548204532535,
      "grad_norm": 4.495767098772353,
      "learning_rate": 1.1744189714652514e-07,
      "loss": 0.678,
      "step": 4662
    },
    {
      "epoch": 0.46096433779007984,
      "grad_norm": 5.635033539579608,
      "learning_rate": 1.1741036596190546e-07,
      "loss": 0.7815,
      "step": 4663
    },
    {
      "epoch": 0.46106319353483427,
      "grad_norm": 5.042411270691449,
      "learning_rate": 1.1737883299210953e-07,
      "loss": 0.682,
      "step": 4664
    },
    {
      "epoch": 0.46116204927958876,
      "grad_norm": 3.692571613269367,
      "learning_rate": 1.1734729824037069e-07,
      "loss": 0.7163,
      "step": 4665
    },
    {
      "epoch": 0.46126090502434325,
      "grad_norm": 3.8647548789410897,
      "learning_rate": 1.1731576170992225e-07,
      "loss": 0.6965,
      "step": 4666
    },
    {
      "epoch": 0.4613597607690977,
      "grad_norm": 7.039948623611418,
      "learning_rate": 1.1728422340399793e-07,
      "loss": 0.6622,
      "step": 4667
    },
    {
      "epoch": 0.46145861651385217,
      "grad_norm": 4.252241589524265,
      "learning_rate": 1.172526833258314e-07,
      "loss": 0.6806,
      "step": 4668
    },
    {
      "epoch": 0.4615574722586066,
      "grad_norm": 6.611354818948827,
      "learning_rate": 1.1722114147865671e-07,
      "loss": 0.7299,
      "step": 4669
    },
    {
      "epoch": 0.4616563280033611,
      "grad_norm": 9.058539266225598,
      "learning_rate": 1.1718959786570801e-07,
      "loss": 0.7538,
      "step": 4670
    },
    {
      "epoch": 0.4617551837481156,
      "grad_norm": 3.378186138624088,
      "learning_rate": 1.1715805249021964e-07,
      "loss": 0.6563,
      "step": 4671
    },
    {
      "epoch": 0.46185403949287,
      "grad_norm": 3.565069229451203,
      "learning_rate": 1.1712650535542603e-07,
      "loss": 0.6354,
      "step": 4672
    },
    {
      "epoch": 0.4619528952376245,
      "grad_norm": 4.844047598409716,
      "learning_rate": 1.1709495646456196e-07,
      "loss": 0.7444,
      "step": 4673
    },
    {
      "epoch": 0.46205175098237894,
      "grad_norm": 3.2531833346651844,
      "learning_rate": 1.1706340582086227e-07,
      "loss": 0.7827,
      "step": 4674
    },
    {
      "epoch": 0.46215060672713343,
      "grad_norm": 10.568667644395836,
      "learning_rate": 1.1703185342756199e-07,
      "loss": 0.7442,
      "step": 4675
    },
    {
      "epoch": 0.4622494624718879,
      "grad_norm": 3.463765321618339,
      "learning_rate": 1.1700029928789639e-07,
      "loss": 0.6823,
      "step": 4676
    },
    {
      "epoch": 0.46234831821664235,
      "grad_norm": 3.9990672789499135,
      "learning_rate": 1.1696874340510088e-07,
      "loss": 0.7682,
      "step": 4677
    },
    {
      "epoch": 0.46244717396139684,
      "grad_norm": 3.536414998583463,
      "learning_rate": 1.1693718578241105e-07,
      "loss": 0.6907,
      "step": 4678
    },
    {
      "epoch": 0.4625460297061513,
      "grad_norm": 3.290131956037447,
      "learning_rate": 1.1690562642306266e-07,
      "loss": 0.7729,
      "step": 4679
    },
    {
      "epoch": 0.46264488545090576,
      "grad_norm": 5.149613395297949,
      "learning_rate": 1.1687406533029163e-07,
      "loss": 0.6595,
      "step": 4680
    },
    {
      "epoch": 0.46274374119566025,
      "grad_norm": 3.128533817498692,
      "learning_rate": 1.1684250250733412e-07,
      "loss": 0.7791,
      "step": 4681
    },
    {
      "epoch": 0.4628425969404147,
      "grad_norm": 3.570379796147385,
      "learning_rate": 1.1681093795742644e-07,
      "loss": 0.769,
      "step": 4682
    },
    {
      "epoch": 0.4629414526851692,
      "grad_norm": 4.088388059245153,
      "learning_rate": 1.1677937168380503e-07,
      "loss": 0.6796,
      "step": 4683
    },
    {
      "epoch": 0.4630403084299236,
      "grad_norm": 3.6741413775561185,
      "learning_rate": 1.1674780368970656e-07,
      "loss": 0.6659,
      "step": 4684
    },
    {
      "epoch": 0.4631391641746781,
      "grad_norm": 3.735292716456063,
      "learning_rate": 1.1671623397836789e-07,
      "loss": 0.791,
      "step": 4685
    },
    {
      "epoch": 0.4632380199194326,
      "grad_norm": 4.008793165795117,
      "learning_rate": 1.1668466255302598e-07,
      "loss": 0.5466,
      "step": 4686
    },
    {
      "epoch": 0.463336875664187,
      "grad_norm": 3.1534999885384485,
      "learning_rate": 1.1665308941691807e-07,
      "loss": 0.7017,
      "step": 4687
    },
    {
      "epoch": 0.4634357314089415,
      "grad_norm": 4.242041194185755,
      "learning_rate": 1.1662151457328151e-07,
      "loss": 0.8476,
      "step": 4688
    },
    {
      "epoch": 0.46353458715369594,
      "grad_norm": 5.349901967760987,
      "learning_rate": 1.165899380253538e-07,
      "loss": 0.6427,
      "step": 4689
    },
    {
      "epoch": 0.46363344289845043,
      "grad_norm": 3.740536799551845,
      "learning_rate": 1.1655835977637263e-07,
      "loss": 0.7943,
      "step": 4690
    },
    {
      "epoch": 0.4637322986432049,
      "grad_norm": 5.445941783149553,
      "learning_rate": 1.1652677982957593e-07,
      "loss": 0.6684,
      "step": 4691
    },
    {
      "epoch": 0.46383115438795935,
      "grad_norm": 3.3083071740962686,
      "learning_rate": 1.1649519818820175e-07,
      "loss": 0.687,
      "step": 4692
    },
    {
      "epoch": 0.46393001013271384,
      "grad_norm": 3.4590813114977434,
      "learning_rate": 1.1646361485548833e-07,
      "loss": 0.8196,
      "step": 4693
    },
    {
      "epoch": 0.4640288658774683,
      "grad_norm": 4.365700732306801,
      "learning_rate": 1.1643202983467404e-07,
      "loss": 0.6632,
      "step": 4694
    },
    {
      "epoch": 0.46412772162222277,
      "grad_norm": 5.749723334242027,
      "learning_rate": 1.1640044312899751e-07,
      "loss": 0.7067,
      "step": 4695
    },
    {
      "epoch": 0.46422657736697726,
      "grad_norm": 11.519415353869658,
      "learning_rate": 1.1636885474169744e-07,
      "loss": 0.7793,
      "step": 4696
    },
    {
      "epoch": 0.4643254331117317,
      "grad_norm": 9.678608984391213,
      "learning_rate": 1.1633726467601277e-07,
      "loss": 0.6656,
      "step": 4697
    },
    {
      "epoch": 0.4644242888564862,
      "grad_norm": 2.7417267428040075,
      "learning_rate": 1.163056729351826e-07,
      "loss": 0.679,
      "step": 4698
    },
    {
      "epoch": 0.46452314460124067,
      "grad_norm": 5.982984539032504,
      "learning_rate": 1.1627407952244624e-07,
      "loss": 0.6859,
      "step": 4699
    },
    {
      "epoch": 0.4646220003459951,
      "grad_norm": 5.555793290449128,
      "learning_rate": 1.1624248444104303e-07,
      "loss": 0.7489,
      "step": 4700
    },
    {
      "epoch": 0.4647208560907496,
      "grad_norm": 9.621755606373215,
      "learning_rate": 1.1621088769421263e-07,
      "loss": 0.669,
      "step": 4701
    },
    {
      "epoch": 0.464819711835504,
      "grad_norm": 5.7329061174706455,
      "learning_rate": 1.1617928928519487e-07,
      "loss": 0.752,
      "step": 4702
    },
    {
      "epoch": 0.4649185675802585,
      "grad_norm": 3.6604994378405573,
      "learning_rate": 1.1614768921722965e-07,
      "loss": 0.6597,
      "step": 4703
    },
    {
      "epoch": 0.465017423325013,
      "grad_norm": 8.609836965523133,
      "learning_rate": 1.1611608749355706e-07,
      "loss": 0.7116,
      "step": 4704
    },
    {
      "epoch": 0.46511627906976744,
      "grad_norm": 17.111637428732216,
      "learning_rate": 1.1608448411741745e-07,
      "loss": 0.5876,
      "step": 4705
    },
    {
      "epoch": 0.4652151348145219,
      "grad_norm": 4.060235593766756,
      "learning_rate": 1.1605287909205126e-07,
      "loss": 0.6775,
      "step": 4706
    },
    {
      "epoch": 0.46531399055927636,
      "grad_norm": 3.2910948919967082,
      "learning_rate": 1.1602127242069911e-07,
      "loss": 0.6543,
      "step": 4707
    },
    {
      "epoch": 0.46541284630403085,
      "grad_norm": 3.7808516791336397,
      "learning_rate": 1.1598966410660178e-07,
      "loss": 0.7264,
      "step": 4708
    },
    {
      "epoch": 0.46551170204878534,
      "grad_norm": 9.427986524593207,
      "learning_rate": 1.1595805415300029e-07,
      "loss": 0.5736,
      "step": 4709
    },
    {
      "epoch": 0.46561055779353977,
      "grad_norm": 5.976630673309605,
      "learning_rate": 1.1592644256313572e-07,
      "loss": 0.7481,
      "step": 4710
    },
    {
      "epoch": 0.46570941353829426,
      "grad_norm": 4.624927364809713,
      "learning_rate": 1.1589482934024941e-07,
      "loss": 0.7149,
      "step": 4711
    },
    {
      "epoch": 0.4658082692830487,
      "grad_norm": 3.4419574358635963,
      "learning_rate": 1.1586321448758281e-07,
      "loss": 0.7398,
      "step": 4712
    },
    {
      "epoch": 0.4659071250278032,
      "grad_norm": 3.120717223096803,
      "learning_rate": 1.1583159800837758e-07,
      "loss": 0.5969,
      "step": 4713
    },
    {
      "epoch": 0.46600598077255767,
      "grad_norm": 4.8453948273489305,
      "learning_rate": 1.1579997990587549e-07,
      "loss": 0.6516,
      "step": 4714
    },
    {
      "epoch": 0.4661048365173121,
      "grad_norm": 3.0943879641097056,
      "learning_rate": 1.1576836018331857e-07,
      "loss": 0.729,
      "step": 4715
    },
    {
      "epoch": 0.4662036922620666,
      "grad_norm": 3.9193846330208775,
      "learning_rate": 1.1573673884394885e-07,
      "loss": 0.875,
      "step": 4716
    },
    {
      "epoch": 0.466302548006821,
      "grad_norm": 6.834003181684879,
      "learning_rate": 1.1570511589100874e-07,
      "loss": 0.7031,
      "step": 4717
    },
    {
      "epoch": 0.4664014037515755,
      "grad_norm": 5.252000842016032,
      "learning_rate": 1.1567349132774062e-07,
      "loss": 0.7122,
      "step": 4718
    },
    {
      "epoch": 0.46650025949633,
      "grad_norm": 4.3698348109257905,
      "learning_rate": 1.1564186515738719e-07,
      "loss": 0.678,
      "step": 4719
    },
    {
      "epoch": 0.46659911524108444,
      "grad_norm": 4.544744341755708,
      "learning_rate": 1.1561023738319122e-07,
      "loss": 0.6869,
      "step": 4720
    },
    {
      "epoch": 0.46669797098583893,
      "grad_norm": 3.9455358074527456,
      "learning_rate": 1.1557860800839567e-07,
      "loss": 0.7238,
      "step": 4721
    },
    {
      "epoch": 0.46679682673059336,
      "grad_norm": 3.0692528455587156,
      "learning_rate": 1.155469770362437e-07,
      "loss": 0.67,
      "step": 4722
    },
    {
      "epoch": 0.46689568247534785,
      "grad_norm": 4.3596911248609445,
      "learning_rate": 1.1551534446997852e-07,
      "loss": 0.6617,
      "step": 4723
    },
    {
      "epoch": 0.46699453822010234,
      "grad_norm": 3.8898571289552053,
      "learning_rate": 1.1548371031284368e-07,
      "loss": 0.7197,
      "step": 4724
    },
    {
      "epoch": 0.4670933939648568,
      "grad_norm": 3.687045836318085,
      "learning_rate": 1.1545207456808272e-07,
      "loss": 0.6699,
      "step": 4725
    },
    {
      "epoch": 0.46719224970961126,
      "grad_norm": 15.0945003114756,
      "learning_rate": 1.1542043723893944e-07,
      "loss": 0.8255,
      "step": 4726
    },
    {
      "epoch": 0.4672911054543657,
      "grad_norm": 6.63233313877866,
      "learning_rate": 1.153887983286578e-07,
      "loss": 0.6651,
      "step": 4727
    },
    {
      "epoch": 0.4673899611991202,
      "grad_norm": 6.088459862920595,
      "learning_rate": 1.1535715784048188e-07,
      "loss": 0.5999,
      "step": 4728
    },
    {
      "epoch": 0.4674888169438747,
      "grad_norm": 46.09721081772401,
      "learning_rate": 1.1532551577765597e-07,
      "loss": 0.6397,
      "step": 4729
    },
    {
      "epoch": 0.4675876726886291,
      "grad_norm": 3.4992431270471203,
      "learning_rate": 1.1529387214342447e-07,
      "loss": 0.8203,
      "step": 4730
    },
    {
      "epoch": 0.4676865284333836,
      "grad_norm": 5.759420848194705,
      "learning_rate": 1.1526222694103202e-07,
      "loss": 0.6325,
      "step": 4731
    },
    {
      "epoch": 0.46778538417813803,
      "grad_norm": 5.698135216003437,
      "learning_rate": 1.1523058017372331e-07,
      "loss": 0.6558,
      "step": 4732
    },
    {
      "epoch": 0.4678842399228925,
      "grad_norm": 4.41892910247054,
      "learning_rate": 1.1519893184474327e-07,
      "loss": 0.5974,
      "step": 4733
    },
    {
      "epoch": 0.467983095667647,
      "grad_norm": 4.127738154456137,
      "learning_rate": 1.1516728195733695e-07,
      "loss": 0.6639,
      "step": 4734
    },
    {
      "epoch": 0.46808195141240144,
      "grad_norm": 4.589122011658872,
      "learning_rate": 1.1513563051474961e-07,
      "loss": 0.6775,
      "step": 4735
    },
    {
      "epoch": 0.46818080715715593,
      "grad_norm": 12.503323495320583,
      "learning_rate": 1.1510397752022665e-07,
      "loss": 0.7309,
      "step": 4736
    },
    {
      "epoch": 0.46827966290191037,
      "grad_norm": 4.046731133438157,
      "learning_rate": 1.1507232297701357e-07,
      "loss": 0.6989,
      "step": 4737
    },
    {
      "epoch": 0.46837851864666485,
      "grad_norm": 4.116497819421549,
      "learning_rate": 1.1504066688835607e-07,
      "loss": 0.7187,
      "step": 4738
    },
    {
      "epoch": 0.46847737439141934,
      "grad_norm": 3.3660008058564177,
      "learning_rate": 1.150090092575001e-07,
      "loss": 0.7176,
      "step": 4739
    },
    {
      "epoch": 0.4685762301361738,
      "grad_norm": 4.967269189763686,
      "learning_rate": 1.1497735008769161e-07,
      "loss": 0.7224,
      "step": 4740
    },
    {
      "epoch": 0.46867508588092827,
      "grad_norm": 4.115200445846632,
      "learning_rate": 1.1494568938217682e-07,
      "loss": 0.7072,
      "step": 4741
    },
    {
      "epoch": 0.4687739416256827,
      "grad_norm": 3.5369347910871918,
      "learning_rate": 1.1491402714420202e-07,
      "loss": 0.7283,
      "step": 4742
    },
    {
      "epoch": 0.4688727973704372,
      "grad_norm": 3.7275759695320736,
      "learning_rate": 1.1488236337701376e-07,
      "loss": 0.6797,
      "step": 4743
    },
    {
      "epoch": 0.4689716531151917,
      "grad_norm": 11.993252242219121,
      "learning_rate": 1.1485069808385866e-07,
      "loss": 0.7723,
      "step": 4744
    },
    {
      "epoch": 0.4690705088599461,
      "grad_norm": 4.145050439710796,
      "learning_rate": 1.1481903126798349e-07,
      "loss": 0.7302,
      "step": 4745
    },
    {
      "epoch": 0.4691693646047006,
      "grad_norm": 7.082357114593416,
      "learning_rate": 1.1478736293263529e-07,
      "loss": 0.6925,
      "step": 4746
    },
    {
      "epoch": 0.46926822034945503,
      "grad_norm": 3.8466547753478246,
      "learning_rate": 1.1475569308106117e-07,
      "loss": 0.6449,
      "step": 4747
    },
    {
      "epoch": 0.4693670760942095,
      "grad_norm": 4.029119842199248,
      "learning_rate": 1.1472402171650837e-07,
      "loss": 0.7002,
      "step": 4748
    },
    {
      "epoch": 0.469465931838964,
      "grad_norm": 3.4612435528820535,
      "learning_rate": 1.1469234884222434e-07,
      "loss": 0.6831,
      "step": 4749
    },
    {
      "epoch": 0.46956478758371845,
      "grad_norm": 3.0351195562618227,
      "learning_rate": 1.1466067446145666e-07,
      "loss": 0.6727,
      "step": 4750
    },
    {
      "epoch": 0.46966364332847294,
      "grad_norm": 6.237232336046948,
      "learning_rate": 1.1462899857745307e-07,
      "loss": 0.8575,
      "step": 4751
    },
    {
      "epoch": 0.46976249907322737,
      "grad_norm": 3.656110778328478,
      "learning_rate": 1.1459732119346146e-07,
      "loss": 0.7421,
      "step": 4752
    },
    {
      "epoch": 0.46986135481798186,
      "grad_norm": 6.0778065533405385,
      "learning_rate": 1.1456564231272992e-07,
      "loss": 0.7809,
      "step": 4753
    },
    {
      "epoch": 0.46996021056273635,
      "grad_norm": 3.2879756432519347,
      "learning_rate": 1.1453396193850659e-07,
      "loss": 0.6001,
      "step": 4754
    },
    {
      "epoch": 0.4700590663074908,
      "grad_norm": 4.345443234668562,
      "learning_rate": 1.1450228007403988e-07,
      "loss": 0.7303,
      "step": 4755
    },
    {
      "epoch": 0.47015792205224527,
      "grad_norm": 5.888325107309177,
      "learning_rate": 1.1447059672257823e-07,
      "loss": 0.6582,
      "step": 4756
    },
    {
      "epoch": 0.4702567777969997,
      "grad_norm": 8.653022346998403,
      "learning_rate": 1.1443891188737038e-07,
      "loss": 0.7547,
      "step": 4757
    },
    {
      "epoch": 0.4703556335417542,
      "grad_norm": 3.6413584568839705,
      "learning_rate": 1.1440722557166511e-07,
      "loss": 0.727,
      "step": 4758
    },
    {
      "epoch": 0.4704544892865087,
      "grad_norm": 5.0862550364480965,
      "learning_rate": 1.1437553777871138e-07,
      "loss": 0.6694,
      "step": 4759
    },
    {
      "epoch": 0.4705533450312631,
      "grad_norm": 4.686032588845638,
      "learning_rate": 1.1434384851175826e-07,
      "loss": 0.6295,
      "step": 4760
    },
    {
      "epoch": 0.4706522007760176,
      "grad_norm": 12.294816050316133,
      "learning_rate": 1.143121577740551e-07,
      "loss": 0.6007,
      "step": 4761
    },
    {
      "epoch": 0.47075105652077204,
      "grad_norm": 5.013159575825405,
      "learning_rate": 1.1428046556885127e-07,
      "loss": 0.7146,
      "step": 4762
    },
    {
      "epoch": 0.4708499122655265,
      "grad_norm": 3.8525977166211627,
      "learning_rate": 1.1424877189939632e-07,
      "loss": 0.6922,
      "step": 4763
    },
    {
      "epoch": 0.470948768010281,
      "grad_norm": 4.642034119818238,
      "learning_rate": 1.1421707676894004e-07,
      "loss": 0.8061,
      "step": 4764
    },
    {
      "epoch": 0.47104762375503545,
      "grad_norm": 3.8282134631909974,
      "learning_rate": 1.1418538018073226e-07,
      "loss": 0.6999,
      "step": 4765
    },
    {
      "epoch": 0.47114647949978994,
      "grad_norm": 4.173515301928192,
      "learning_rate": 1.1415368213802295e-07,
      "loss": 0.6954,
      "step": 4766
    },
    {
      "epoch": 0.47124533524454437,
      "grad_norm": 3.796953713890127,
      "learning_rate": 1.1412198264406234e-07,
      "loss": 0.6877,
      "step": 4767
    },
    {
      "epoch": 0.47134419098929886,
      "grad_norm": 4.250623861562799,
      "learning_rate": 1.1409028170210074e-07,
      "loss": 0.7364,
      "step": 4768
    },
    {
      "epoch": 0.47144304673405335,
      "grad_norm": 8.128875259125175,
      "learning_rate": 1.1405857931538856e-07,
      "loss": 0.8536,
      "step": 4769
    },
    {
      "epoch": 0.4715419024788078,
      "grad_norm": 4.421817449054516,
      "learning_rate": 1.140268754871765e-07,
      "loss": 0.6887,
      "step": 4770
    },
    {
      "epoch": 0.4716407582235623,
      "grad_norm": 3.521137018769241,
      "learning_rate": 1.1399517022071519e-07,
      "loss": 0.721,
      "step": 4771
    },
    {
      "epoch": 0.4717396139683167,
      "grad_norm": 2.8038154966074096,
      "learning_rate": 1.1396346351925568e-07,
      "loss": 0.6235,
      "step": 4772
    },
    {
      "epoch": 0.4718384697130712,
      "grad_norm": 4.066169650778574,
      "learning_rate": 1.1393175538604895e-07,
      "loss": 0.7841,
      "step": 4773
    },
    {
      "epoch": 0.4719373254578257,
      "grad_norm": 3.6796767458230772,
      "learning_rate": 1.1390004582434617e-07,
      "loss": 0.7959,
      "step": 4774
    },
    {
      "epoch": 0.4720361812025801,
      "grad_norm": 3.912029722415188,
      "learning_rate": 1.1386833483739881e-07,
      "loss": 0.6601,
      "step": 4775
    },
    {
      "epoch": 0.4721350369473346,
      "grad_norm": 6.775346766966058,
      "learning_rate": 1.1383662242845826e-07,
      "loss": 0.6692,
      "step": 4776
    },
    {
      "epoch": 0.4722338926920891,
      "grad_norm": 3.1135649797503215,
      "learning_rate": 1.1380490860077616e-07,
      "loss": 0.6993,
      "step": 4777
    },
    {
      "epoch": 0.47233274843684353,
      "grad_norm": 4.684035762143831,
      "learning_rate": 1.1377319335760431e-07,
      "loss": 0.7153,
      "step": 4778
    },
    {
      "epoch": 0.472431604181598,
      "grad_norm": 3.5865089801424945,
      "learning_rate": 1.1374147670219467e-07,
      "loss": 0.7265,
      "step": 4779
    },
    {
      "epoch": 0.47253045992635245,
      "grad_norm": 4.831593982873998,
      "learning_rate": 1.137097586377993e-07,
      "loss": 0.7061,
      "step": 4780
    },
    {
      "epoch": 0.47262931567110694,
      "grad_norm": 3.8743394390008654,
      "learning_rate": 1.1367803916767043e-07,
      "loss": 0.6044,
      "step": 4781
    },
    {
      "epoch": 0.47272817141586143,
      "grad_norm": 6.090812780367724,
      "learning_rate": 1.1364631829506037e-07,
      "loss": 0.6497,
      "step": 4782
    },
    {
      "epoch": 0.47282702716061586,
      "grad_norm": 5.346108340936592,
      "learning_rate": 1.1361459602322168e-07,
      "loss": 0.7767,
      "step": 4783
    },
    {
      "epoch": 0.47292588290537035,
      "grad_norm": 4.37422065164614,
      "learning_rate": 1.13582872355407e-07,
      "loss": 0.7884,
      "step": 4784
    },
    {
      "epoch": 0.4730247386501248,
      "grad_norm": 3.6019420207804043,
      "learning_rate": 1.1355114729486915e-07,
      "loss": 0.734,
      "step": 4785
    },
    {
      "epoch": 0.4731235943948793,
      "grad_norm": 4.540844869309253,
      "learning_rate": 1.13519420844861e-07,
      "loss": 0.7172,
      "step": 4786
    },
    {
      "epoch": 0.47322245013963377,
      "grad_norm": 4.464854434756222,
      "learning_rate": 1.1348769300863567e-07,
      "loss": 0.67,
      "step": 4787
    },
    {
      "epoch": 0.4733213058843882,
      "grad_norm": 4.534591024449965,
      "learning_rate": 1.134559637894464e-07,
      "loss": 0.7273,
      "step": 4788
    },
    {
      "epoch": 0.4734201616291427,
      "grad_norm": 4.97740350991135,
      "learning_rate": 1.134242331905465e-07,
      "loss": 0.7346,
      "step": 4789
    },
    {
      "epoch": 0.4735190173738971,
      "grad_norm": 3.9399808379737777,
      "learning_rate": 1.1339250121518955e-07,
      "loss": 0.6016,
      "step": 4790
    },
    {
      "epoch": 0.4736178731186516,
      "grad_norm": 7.115401611468567,
      "learning_rate": 1.1336076786662914e-07,
      "loss": 0.6615,
      "step": 4791
    },
    {
      "epoch": 0.4737167288634061,
      "grad_norm": 4.217901948350837,
      "learning_rate": 1.1332903314811906e-07,
      "loss": 0.6223,
      "step": 4792
    },
    {
      "epoch": 0.47381558460816053,
      "grad_norm": 6.5004319458787165,
      "learning_rate": 1.1329729706291324e-07,
      "loss": 0.7242,
      "step": 4793
    },
    {
      "epoch": 0.473914440352915,
      "grad_norm": 7.48493753883289,
      "learning_rate": 1.1326555961426578e-07,
      "loss": 0.6623,
      "step": 4794
    },
    {
      "epoch": 0.47401329609766946,
      "grad_norm": 6.371928638039179,
      "learning_rate": 1.1323382080543085e-07,
      "loss": 0.693,
      "step": 4795
    },
    {
      "epoch": 0.47411215184242395,
      "grad_norm": 6.076585785333862,
      "learning_rate": 1.1320208063966279e-07,
      "loss": 0.717,
      "step": 4796
    },
    {
      "epoch": 0.47421100758717843,
      "grad_norm": 3.3954887025293403,
      "learning_rate": 1.1317033912021612e-07,
      "loss": 0.749,
      "step": 4797
    },
    {
      "epoch": 0.47430986333193287,
      "grad_norm": 5.799272460384484,
      "learning_rate": 1.1313859625034545e-07,
      "loss": 0.6789,
      "step": 4798
    },
    {
      "epoch": 0.47440871907668736,
      "grad_norm": 4.21886273307908,
      "learning_rate": 1.1310685203330556e-07,
      "loss": 0.8423,
      "step": 4799
    },
    {
      "epoch": 0.4745075748214418,
      "grad_norm": 12.337340627004302,
      "learning_rate": 1.130751064723513e-07,
      "loss": 0.7518,
      "step": 4800
    },
    {
      "epoch": 0.4746064305661963,
      "grad_norm": 5.262698770432406,
      "learning_rate": 1.1304335957073776e-07,
      "loss": 0.7164,
      "step": 4801
    },
    {
      "epoch": 0.47470528631095077,
      "grad_norm": 4.26689992209808,
      "learning_rate": 1.1301161133172012e-07,
      "loss": 0.6662,
      "step": 4802
    },
    {
      "epoch": 0.4748041420557052,
      "grad_norm": 7.039276245213629,
      "learning_rate": 1.1297986175855365e-07,
      "loss": 0.6439,
      "step": 4803
    },
    {
      "epoch": 0.4749029978004597,
      "grad_norm": 7.910952725458966,
      "learning_rate": 1.1294811085449383e-07,
      "loss": 0.7488,
      "step": 4804
    },
    {
      "epoch": 0.4750018535452141,
      "grad_norm": 4.585138380142986,
      "learning_rate": 1.1291635862279623e-07,
      "loss": 0.6983,
      "step": 4805
    },
    {
      "epoch": 0.4751007092899686,
      "grad_norm": 4.774946028813517,
      "learning_rate": 1.128846050667166e-07,
      "loss": 0.6798,
      "step": 4806
    },
    {
      "epoch": 0.4751995650347231,
      "grad_norm": 5.0624522952712345,
      "learning_rate": 1.1285285018951074e-07,
      "loss": 0.8252,
      "step": 4807
    },
    {
      "epoch": 0.47529842077947754,
      "grad_norm": 4.956299902408893,
      "learning_rate": 1.1282109399443474e-07,
      "loss": 0.6432,
      "step": 4808
    },
    {
      "epoch": 0.475397276524232,
      "grad_norm": 5.3385094409519445,
      "learning_rate": 1.1278933648474464e-07,
      "loss": 0.6993,
      "step": 4809
    },
    {
      "epoch": 0.47549613226898646,
      "grad_norm": 3.2067952421989583,
      "learning_rate": 1.1275757766369675e-07,
      "loss": 0.5856,
      "step": 4810
    },
    {
      "epoch": 0.47559498801374095,
      "grad_norm": 7.636609818464929,
      "learning_rate": 1.1272581753454747e-07,
      "loss": 0.7802,
      "step": 4811
    },
    {
      "epoch": 0.47569384375849544,
      "grad_norm": 4.388299091644086,
      "learning_rate": 1.1269405610055331e-07,
      "loss": 0.7627,
      "step": 4812
    },
    {
      "epoch": 0.47579269950324987,
      "grad_norm": 3.4479448005176576,
      "learning_rate": 1.1266229336497096e-07,
      "loss": 0.6996,
      "step": 4813
    },
    {
      "epoch": 0.47589155524800436,
      "grad_norm": 4.3080875408764685,
      "learning_rate": 1.1263052933105717e-07,
      "loss": 0.8197,
      "step": 4814
    },
    {
      "epoch": 0.4759904109927588,
      "grad_norm": 4.471056154693649,
      "learning_rate": 1.1259876400206894e-07,
      "loss": 0.7308,
      "step": 4815
    },
    {
      "epoch": 0.4760892667375133,
      "grad_norm": 3.848082640712577,
      "learning_rate": 1.125669973812633e-07,
      "loss": 0.7682,
      "step": 4816
    },
    {
      "epoch": 0.4761881224822678,
      "grad_norm": 3.147038818538059,
      "learning_rate": 1.1253522947189747e-07,
      "loss": 0.6963,
      "step": 4817
    },
    {
      "epoch": 0.4762869782270222,
      "grad_norm": 3.535078839851762,
      "learning_rate": 1.1250346027722872e-07,
      "loss": 0.7354,
      "step": 4818
    },
    {
      "epoch": 0.4763858339717767,
      "grad_norm": 9.890132115370285,
      "learning_rate": 1.1247168980051463e-07,
      "loss": 0.6767,
      "step": 4819
    },
    {
      "epoch": 0.47648468971653113,
      "grad_norm": 3.2037953418363094,
      "learning_rate": 1.1243991804501267e-07,
      "loss": 0.7104,
      "step": 4820
    },
    {
      "epoch": 0.4765835454612856,
      "grad_norm": 5.202070579947606,
      "learning_rate": 1.1240814501398064e-07,
      "loss": 0.8042,
      "step": 4821
    },
    {
      "epoch": 0.4766824012060401,
      "grad_norm": 23.71745865829343,
      "learning_rate": 1.1237637071067634e-07,
      "loss": 0.7019,
      "step": 4822
    },
    {
      "epoch": 0.47678125695079454,
      "grad_norm": 9.399607428764302,
      "learning_rate": 1.1234459513835781e-07,
      "loss": 0.6735,
      "step": 4823
    },
    {
      "epoch": 0.47688011269554903,
      "grad_norm": 3.582615826921775,
      "learning_rate": 1.1231281830028314e-07,
      "loss": 0.731,
      "step": 4824
    },
    {
      "epoch": 0.47697896844030346,
      "grad_norm": 3.677347538629893,
      "learning_rate": 1.1228104019971057e-07,
      "loss": 0.6394,
      "step": 4825
    },
    {
      "epoch": 0.47707782418505795,
      "grad_norm": 3.1797647087728604,
      "learning_rate": 1.122492608398985e-07,
      "loss": 0.6892,
      "step": 4826
    },
    {
      "epoch": 0.47717667992981244,
      "grad_norm": 4.035150629111296,
      "learning_rate": 1.1221748022410542e-07,
      "loss": 0.8077,
      "step": 4827
    },
    {
      "epoch": 0.4772755356745669,
      "grad_norm": 3.680488951032811,
      "learning_rate": 1.1218569835558995e-07,
      "loss": 0.7462,
      "step": 4828
    },
    {
      "epoch": 0.47737439141932136,
      "grad_norm": 4.935144470291025,
      "learning_rate": 1.1215391523761088e-07,
      "loss": 0.7609,
      "step": 4829
    },
    {
      "epoch": 0.4774732471640758,
      "grad_norm": 3.6819595425598637,
      "learning_rate": 1.121221308734271e-07,
      "loss": 0.6772,
      "step": 4830
    },
    {
      "epoch": 0.4775721029088303,
      "grad_norm": 4.1167397740147615,
      "learning_rate": 1.1209034526629759e-07,
      "loss": 0.6039,
      "step": 4831
    },
    {
      "epoch": 0.4776709586535848,
      "grad_norm": 25.84147200352015,
      "learning_rate": 1.1205855841948154e-07,
      "loss": 0.809,
      "step": 4832
    },
    {
      "epoch": 0.4777698143983392,
      "grad_norm": 3.5539630088529615,
      "learning_rate": 1.1202677033623815e-07,
      "loss": 0.7654,
      "step": 4833
    },
    {
      "epoch": 0.4778686701430937,
      "grad_norm": 3.5436509469854256,
      "learning_rate": 1.119949810198269e-07,
      "loss": 0.7418,
      "step": 4834
    },
    {
      "epoch": 0.47796752588784813,
      "grad_norm": 4.482298063598885,
      "learning_rate": 1.1196319047350728e-07,
      "loss": 0.7581,
      "step": 4835
    },
    {
      "epoch": 0.4780663816326026,
      "grad_norm": 3.6674359756353145,
      "learning_rate": 1.1193139870053898e-07,
      "loss": 0.6866,
      "step": 4836
    },
    {
      "epoch": 0.4781652373773571,
      "grad_norm": 6.962563511880813,
      "learning_rate": 1.1189960570418171e-07,
      "loss": 0.8009,
      "step": 4837
    },
    {
      "epoch": 0.47826409312211154,
      "grad_norm": 4.785340720588053,
      "learning_rate": 1.1186781148769541e-07,
      "loss": 0.7426,
      "step": 4838
    },
    {
      "epoch": 0.47836294886686603,
      "grad_norm": 3.9149436740723944,
      "learning_rate": 1.1183601605434012e-07,
      "loss": 0.6507,
      "step": 4839
    },
    {
      "epoch": 0.47846180461162047,
      "grad_norm": 4.985419092151062,
      "learning_rate": 1.1180421940737594e-07,
      "loss": 0.7343,
      "step": 4840
    },
    {
      "epoch": 0.47856066035637496,
      "grad_norm": 3.2885346449293382,
      "learning_rate": 1.1177242155006322e-07,
      "loss": 0.6948,
      "step": 4841
    },
    {
      "epoch": 0.47865951610112945,
      "grad_norm": 4.268916446902121,
      "learning_rate": 1.1174062248566231e-07,
      "loss": 0.7457,
      "step": 4842
    },
    {
      "epoch": 0.4787583718458839,
      "grad_norm": 3.6470551769381734,
      "learning_rate": 1.1170882221743377e-07,
      "loss": 0.6651,
      "step": 4843
    },
    {
      "epoch": 0.47885722759063837,
      "grad_norm": 8.10325163583803,
      "learning_rate": 1.1167702074863818e-07,
      "loss": 0.7827,
      "step": 4844
    },
    {
      "epoch": 0.4789560833353928,
      "grad_norm": 4.138027673662975,
      "learning_rate": 1.1164521808253641e-07,
      "loss": 0.6729,
      "step": 4845
    },
    {
      "epoch": 0.4790549390801473,
      "grad_norm": 4.856152200566382,
      "learning_rate": 1.1161341422238932e-07,
      "loss": 0.7171,
      "step": 4846
    },
    {
      "epoch": 0.4791537948249018,
      "grad_norm": 4.264553747395361,
      "learning_rate": 1.1158160917145785e-07,
      "loss": 0.7344,
      "step": 4847
    },
    {
      "epoch": 0.4792526505696562,
      "grad_norm": 7.44916434482811,
      "learning_rate": 1.1154980293300324e-07,
      "loss": 0.7674,
      "step": 4848
    },
    {
      "epoch": 0.4793515063144107,
      "grad_norm": 6.61599031701135,
      "learning_rate": 1.1151799551028672e-07,
      "loss": 0.7685,
      "step": 4849
    },
    {
      "epoch": 0.47945036205916514,
      "grad_norm": 4.780726837919409,
      "learning_rate": 1.1148618690656963e-07,
      "loss": 0.7175,
      "step": 4850
    },
    {
      "epoch": 0.4795492178039196,
      "grad_norm": 7.378049036465865,
      "learning_rate": 1.114543771251135e-07,
      "loss": 0.7261,
      "step": 4851
    },
    {
      "epoch": 0.4796480735486741,
      "grad_norm": 4.116607169199086,
      "learning_rate": 1.1142256616917997e-07,
      "loss": 0.7532,
      "step": 4852
    },
    {
      "epoch": 0.47974692929342855,
      "grad_norm": 3.4624542346270544,
      "learning_rate": 1.113907540420308e-07,
      "loss": 0.7046,
      "step": 4853
    },
    {
      "epoch": 0.47984578503818304,
      "grad_norm": 2.8371761074191064,
      "learning_rate": 1.1135894074692782e-07,
      "loss": 0.6423,
      "step": 4854
    },
    {
      "epoch": 0.4799446407829375,
      "grad_norm": 4.8682443234018145,
      "learning_rate": 1.1132712628713301e-07,
      "loss": 0.6299,
      "step": 4855
    },
    {
      "epoch": 0.48004349652769196,
      "grad_norm": 6.1767399923675805,
      "learning_rate": 1.1129531066590851e-07,
      "loss": 0.7411,
      "step": 4856
    },
    {
      "epoch": 0.48014235227244645,
      "grad_norm": 16.89503162892244,
      "learning_rate": 1.112634938865165e-07,
      "loss": 0.7584,
      "step": 4857
    },
    {
      "epoch": 0.4802412080172009,
      "grad_norm": 4.792935880981809,
      "learning_rate": 1.1123167595221931e-07,
      "loss": 0.6388,
      "step": 4858
    },
    {
      "epoch": 0.48034006376195537,
      "grad_norm": 5.501212260840952,
      "learning_rate": 1.1119985686627948e-07,
      "loss": 0.7277,
      "step": 4859
    },
    {
      "epoch": 0.48043891950670986,
      "grad_norm": 5.092174238310617,
      "learning_rate": 1.1116803663195951e-07,
      "loss": 0.6602,
      "step": 4860
    },
    {
      "epoch": 0.4805377752514643,
      "grad_norm": 4.843152833414339,
      "learning_rate": 1.1113621525252214e-07,
      "loss": 0.6915,
      "step": 4861
    },
    {
      "epoch": 0.4806366309962188,
      "grad_norm": 6.012568691979855,
      "learning_rate": 1.1110439273123013e-07,
      "loss": 0.7156,
      "step": 4862
    },
    {
      "epoch": 0.4807354867409732,
      "grad_norm": 4.347330702886589,
      "learning_rate": 1.110725690713465e-07,
      "loss": 0.7467,
      "step": 4863
    },
    {
      "epoch": 0.4808343424857277,
      "grad_norm": 3.585039754199605,
      "learning_rate": 1.1104074427613423e-07,
      "loss": 0.6921,
      "step": 4864
    },
    {
      "epoch": 0.4809331982304822,
      "grad_norm": 6.447538620949876,
      "learning_rate": 1.1100891834885648e-07,
      "loss": 0.6029,
      "step": 4865
    },
    {
      "epoch": 0.48103205397523663,
      "grad_norm": 4.635661748257955,
      "learning_rate": 1.1097709129277654e-07,
      "loss": 0.7376,
      "step": 4866
    },
    {
      "epoch": 0.4811309097199911,
      "grad_norm": 4.696447370792104,
      "learning_rate": 1.1094526311115783e-07,
      "loss": 0.6939,
      "step": 4867
    },
    {
      "epoch": 0.48122976546474555,
      "grad_norm": 5.107637324728658,
      "learning_rate": 1.1091343380726383e-07,
      "loss": 0.7985,
      "step": 4868
    },
    {
      "epoch": 0.48132862120950004,
      "grad_norm": 5.544975864518503,
      "learning_rate": 1.1088160338435817e-07,
      "loss": 0.6573,
      "step": 4869
    },
    {
      "epoch": 0.48142747695425453,
      "grad_norm": 5.431725763529287,
      "learning_rate": 1.1084977184570461e-07,
      "loss": 0.6657,
      "step": 4870
    },
    {
      "epoch": 0.48152633269900896,
      "grad_norm": 3.4985533351354148,
      "learning_rate": 1.1081793919456701e-07,
      "loss": 0.7258,
      "step": 4871
    },
    {
      "epoch": 0.48162518844376345,
      "grad_norm": 5.023255670594426,
      "learning_rate": 1.1078610543420929e-07,
      "loss": 0.683,
      "step": 4872
    },
    {
      "epoch": 0.4817240441885179,
      "grad_norm": 3.586634101102196,
      "learning_rate": 1.107542705678956e-07,
      "loss": 0.6842,
      "step": 4873
    },
    {
      "epoch": 0.4818228999332724,
      "grad_norm": 3.6448472291203573,
      "learning_rate": 1.1072243459889008e-07,
      "loss": 0.7298,
      "step": 4874
    },
    {
      "epoch": 0.48192175567802686,
      "grad_norm": 4.829019537947762,
      "learning_rate": 1.1069059753045706e-07,
      "loss": 0.719,
      "step": 4875
    },
    {
      "epoch": 0.4820206114227813,
      "grad_norm": 3.1138484941934124,
      "learning_rate": 1.1065875936586095e-07,
      "loss": 0.7371,
      "step": 4876
    },
    {
      "epoch": 0.4821194671675358,
      "grad_norm": 11.40634640585387,
      "learning_rate": 1.106269201083663e-07,
      "loss": 0.7791,
      "step": 4877
    },
    {
      "epoch": 0.4822183229122902,
      "grad_norm": 2.9205769620197928,
      "learning_rate": 1.1059507976123774e-07,
      "loss": 0.6838,
      "step": 4878
    },
    {
      "epoch": 0.4823171786570447,
      "grad_norm": 3.6816004166090917,
      "learning_rate": 1.1056323832774006e-07,
      "loss": 0.6252,
      "step": 4879
    },
    {
      "epoch": 0.4824160344017992,
      "grad_norm": 4.735532988650914,
      "learning_rate": 1.105313958111381e-07,
      "loss": 0.7852,
      "step": 4880
    },
    {
      "epoch": 0.48251489014655363,
      "grad_norm": 3.1617779835221924,
      "learning_rate": 1.1049955221469687e-07,
      "loss": 0.6897,
      "step": 4881
    },
    {
      "epoch": 0.4826137458913081,
      "grad_norm": 3.046302848193647,
      "learning_rate": 1.1046770754168142e-07,
      "loss": 0.663,
      "step": 4882
    },
    {
      "epoch": 0.48271260163606255,
      "grad_norm": 6.099039869597704,
      "learning_rate": 1.1043586179535697e-07,
      "loss": 0.6886,
      "step": 4883
    },
    {
      "epoch": 0.48281145738081704,
      "grad_norm": 4.337379858521869,
      "learning_rate": 1.1040401497898881e-07,
      "loss": 0.5963,
      "step": 4884
    },
    {
      "epoch": 0.48291031312557153,
      "grad_norm": 3.6174672942509276,
      "learning_rate": 1.103721670958424e-07,
      "loss": 0.6593,
      "step": 4885
    },
    {
      "epoch": 0.48300916887032597,
      "grad_norm": 4.224917200277799,
      "learning_rate": 1.1034031814918328e-07,
      "loss": 0.7961,
      "step": 4886
    },
    {
      "epoch": 0.48310802461508046,
      "grad_norm": 4.492040764180408,
      "learning_rate": 1.1030846814227705e-07,
      "loss": 0.7349,
      "step": 4887
    },
    {
      "epoch": 0.4832068803598349,
      "grad_norm": 3.447323592006722,
      "learning_rate": 1.1027661707838945e-07,
      "loss": 0.632,
      "step": 4888
    },
    {
      "epoch": 0.4833057361045894,
      "grad_norm": 12.614376681060467,
      "learning_rate": 1.1024476496078638e-07,
      "loss": 0.7729,
      "step": 4889
    },
    {
      "epoch": 0.48340459184934387,
      "grad_norm": 12.601489905707208,
      "learning_rate": 1.102129117927338e-07,
      "loss": 0.6169,
      "step": 4890
    },
    {
      "epoch": 0.4835034475940983,
      "grad_norm": 3.570453380771824,
      "learning_rate": 1.1018105757749774e-07,
      "loss": 0.682,
      "step": 4891
    },
    {
      "epoch": 0.4836023033388528,
      "grad_norm": 4.990232075839659,
      "learning_rate": 1.1014920231834441e-07,
      "loss": 0.6476,
      "step": 4892
    },
    {
      "epoch": 0.4837011590836072,
      "grad_norm": 4.725634761681587,
      "learning_rate": 1.101173460185401e-07,
      "loss": 0.684,
      "step": 4893
    },
    {
      "epoch": 0.4838000148283617,
      "grad_norm": 3.3958394553069198,
      "learning_rate": 1.1008548868135122e-07,
      "loss": 0.7597,
      "step": 4894
    },
    {
      "epoch": 0.4838988705731162,
      "grad_norm": 4.3408337071728305,
      "learning_rate": 1.100536303100442e-07,
      "loss": 0.6889,
      "step": 4895
    },
    {
      "epoch": 0.48399772631787064,
      "grad_norm": 14.485135727220992,
      "learning_rate": 1.1002177090788572e-07,
      "loss": 0.7882,
      "step": 4896
    },
    {
      "epoch": 0.4840965820626251,
      "grad_norm": 6.648612373067033,
      "learning_rate": 1.0998991047814246e-07,
      "loss": 0.6048,
      "step": 4897
    },
    {
      "epoch": 0.48419543780737956,
      "grad_norm": 3.432258795261783,
      "learning_rate": 1.0995804902408123e-07,
      "loss": 0.7526,
      "step": 4898
    },
    {
      "epoch": 0.48429429355213405,
      "grad_norm": 8.12721884864471,
      "learning_rate": 1.0992618654896899e-07,
      "loss": 0.7138,
      "step": 4899
    },
    {
      "epoch": 0.48439314929688854,
      "grad_norm": 4.333881865905653,
      "learning_rate": 1.0989432305607273e-07,
      "loss": 0.6797,
      "step": 4900
    },
    {
      "epoch": 0.48449200504164297,
      "grad_norm": 5.115473361220506,
      "learning_rate": 1.0986245854865961e-07,
      "loss": 0.6705,
      "step": 4901
    },
    {
      "epoch": 0.48459086078639746,
      "grad_norm": 3.3935865984658773,
      "learning_rate": 1.098305930299968e-07,
      "loss": 0.5724,
      "step": 4902
    },
    {
      "epoch": 0.4846897165311519,
      "grad_norm": 4.504836598243756,
      "learning_rate": 1.0979872650335171e-07,
      "loss": 0.7025,
      "step": 4903
    },
    {
      "epoch": 0.4847885722759064,
      "grad_norm": 4.511563342818421,
      "learning_rate": 1.0976685897199175e-07,
      "loss": 0.6407,
      "step": 4904
    },
    {
      "epoch": 0.48488742802066087,
      "grad_norm": 2.9988238088928827,
      "learning_rate": 1.097349904391845e-07,
      "loss": 0.6047,
      "step": 4905
    },
    {
      "epoch": 0.4849862837654153,
      "grad_norm": 8.927154429048263,
      "learning_rate": 1.0970312090819755e-07,
      "loss": 0.6352,
      "step": 4906
    },
    {
      "epoch": 0.4850851395101698,
      "grad_norm": 8.711643319759613,
      "learning_rate": 1.0967125038229874e-07,
      "loss": 0.6844,
      "step": 4907
    },
    {
      "epoch": 0.4851839952549242,
      "grad_norm": 5.782719561135157,
      "learning_rate": 1.0963937886475583e-07,
      "loss": 0.7042,
      "step": 4908
    },
    {
      "epoch": 0.4852828509996787,
      "grad_norm": 2.8923411606330376,
      "learning_rate": 1.0960750635883679e-07,
      "loss": 0.7207,
      "step": 4909
    },
    {
      "epoch": 0.4853817067444332,
      "grad_norm": 3.3605990073025356,
      "learning_rate": 1.0957563286780969e-07,
      "loss": 0.6807,
      "step": 4910
    },
    {
      "epoch": 0.48548056248918764,
      "grad_norm": 4.218151862823981,
      "learning_rate": 1.0954375839494272e-07,
      "loss": 0.7573,
      "step": 4911
    },
    {
      "epoch": 0.48557941823394213,
      "grad_norm": 3.3294122183202997,
      "learning_rate": 1.0951188294350409e-07,
      "loss": 0.7068,
      "step": 4912
    },
    {
      "epoch": 0.48567827397869656,
      "grad_norm": 8.178559215236952,
      "learning_rate": 1.0948000651676217e-07,
      "loss": 0.7101,
      "step": 4913
    },
    {
      "epoch": 0.48577712972345105,
      "grad_norm": 7.862273245276272,
      "learning_rate": 1.0944812911798542e-07,
      "loss": 0.7458,
      "step": 4914
    },
    {
      "epoch": 0.48587598546820554,
      "grad_norm": 25.986679876859643,
      "learning_rate": 1.094162507504424e-07,
      "loss": 0.776,
      "step": 4915
    },
    {
      "epoch": 0.48597484121296,
      "grad_norm": 3.5897489126669315,
      "learning_rate": 1.0938437141740177e-07,
      "loss": 0.6998,
      "step": 4916
    },
    {
      "epoch": 0.48607369695771446,
      "grad_norm": 3.2189081744016774,
      "learning_rate": 1.093524911221323e-07,
      "loss": 0.6713,
      "step": 4917
    },
    {
      "epoch": 0.4861725527024689,
      "grad_norm": 3.574269575948743,
      "learning_rate": 1.0932060986790281e-07,
      "loss": 0.6786,
      "step": 4918
    },
    {
      "epoch": 0.4862714084472234,
      "grad_norm": 3.459697969171859,
      "learning_rate": 1.0928872765798227e-07,
      "loss": 0.726,
      "step": 4919
    },
    {
      "epoch": 0.4863702641919779,
      "grad_norm": 4.520662088588456,
      "learning_rate": 1.0925684449563973e-07,
      "loss": 0.6619,
      "step": 4920
    },
    {
      "epoch": 0.4864691199367323,
      "grad_norm": 6.164768573204754,
      "learning_rate": 1.092249603841443e-07,
      "loss": 0.7914,
      "step": 4921
    },
    {
      "epoch": 0.4865679756814868,
      "grad_norm": 6.43478289522857,
      "learning_rate": 1.091930753267653e-07,
      "loss": 0.6005,
      "step": 4922
    },
    {
      "epoch": 0.48666683142624123,
      "grad_norm": 5.958186682462018,
      "learning_rate": 1.09161189326772e-07,
      "loss": 0.6968,
      "step": 4923
    },
    {
      "epoch": 0.4867656871709957,
      "grad_norm": 7.180773082882422,
      "learning_rate": 1.0912930238743392e-07,
      "loss": 0.7897,
      "step": 4924
    },
    {
      "epoch": 0.4868645429157502,
      "grad_norm": 6.2747857603568935,
      "learning_rate": 1.090974145120205e-07,
      "loss": 0.6621,
      "step": 4925
    },
    {
      "epoch": 0.48696339866050464,
      "grad_norm": 3.021712135056737,
      "learning_rate": 1.0906552570380145e-07,
      "loss": 0.8071,
      "step": 4926
    },
    {
      "epoch": 0.48706225440525913,
      "grad_norm": 3.9489475081622256,
      "learning_rate": 1.0903363596604644e-07,
      "loss": 0.6973,
      "step": 4927
    },
    {
      "epoch": 0.48716111015001357,
      "grad_norm": 4.760470168740779,
      "learning_rate": 1.0900174530202533e-07,
      "loss": 0.6806,
      "step": 4928
    },
    {
      "epoch": 0.48725996589476805,
      "grad_norm": 3.454604103633635,
      "learning_rate": 1.0896985371500802e-07,
      "loss": 0.7747,
      "step": 4929
    },
    {
      "epoch": 0.48735882163952254,
      "grad_norm": 3.8421558969954632,
      "learning_rate": 1.0893796120826453e-07,
      "loss": 0.6332,
      "step": 4930
    },
    {
      "epoch": 0.487457677384277,
      "grad_norm": 15.913924841022988,
      "learning_rate": 1.0890606778506497e-07,
      "loss": 0.6284,
      "step": 4931
    },
    {
      "epoch": 0.48755653312903147,
      "grad_norm": 7.6035743143298165,
      "learning_rate": 1.0887417344867951e-07,
      "loss": 0.5953,
      "step": 4932
    },
    {
      "epoch": 0.48765538887378596,
      "grad_norm": 3.2006596639042377,
      "learning_rate": 1.0884227820237848e-07,
      "loss": 0.735,
      "step": 4933
    },
    {
      "epoch": 0.4877542446185404,
      "grad_norm": 9.258333040284953,
      "learning_rate": 1.0881038204943228e-07,
      "loss": 0.7165,
      "step": 4934
    },
    {
      "epoch": 0.4878531003632949,
      "grad_norm": 5.556418887789064,
      "learning_rate": 1.087784849931113e-07,
      "loss": 0.7434,
      "step": 4935
    },
    {
      "epoch": 0.4879519561080493,
      "grad_norm": 3.6102378113474303,
      "learning_rate": 1.0874658703668623e-07,
      "loss": 0.5925,
      "step": 4936
    },
    {
      "epoch": 0.4880508118528038,
      "grad_norm": 3.1599544907409,
      "learning_rate": 1.0871468818342766e-07,
      "loss": 0.8631,
      "step": 4937
    },
    {
      "epoch": 0.4881496675975583,
      "grad_norm": 3.598948492907296,
      "learning_rate": 1.0868278843660637e-07,
      "loss": 0.6245,
      "step": 4938
    },
    {
      "epoch": 0.4882485233423127,
      "grad_norm": 3.145963938068849,
      "learning_rate": 1.0865088779949318e-07,
      "loss": 0.6839,
      "step": 4939
    },
    {
      "epoch": 0.4883473790870672,
      "grad_norm": 5.163289995708715,
      "learning_rate": 1.0861898627535909e-07,
      "loss": 0.7454,
      "step": 4940
    },
    {
      "epoch": 0.48844623483182165,
      "grad_norm": 3.6404858164882836,
      "learning_rate": 1.0858708386747508e-07,
      "loss": 0.6903,
      "step": 4941
    },
    {
      "epoch": 0.48854509057657614,
      "grad_norm": 5.276805873085752,
      "learning_rate": 1.0855518057911231e-07,
      "loss": 0.7433,
      "step": 4942
    },
    {
      "epoch": 0.4886439463213306,
      "grad_norm": 6.819631877960541,
      "learning_rate": 1.0852327641354195e-07,
      "loss": 0.6382,
      "step": 4943
    },
    {
      "epoch": 0.48874280206608506,
      "grad_norm": 5.518998993843801,
      "learning_rate": 1.0849137137403532e-07,
      "loss": 0.6651,
      "step": 4944
    },
    {
      "epoch": 0.48884165781083955,
      "grad_norm": 3.4046267506747077,
      "learning_rate": 1.0845946546386384e-07,
      "loss": 0.6908,
      "step": 4945
    },
    {
      "epoch": 0.488940513555594,
      "grad_norm": 3.6975439907321097,
      "learning_rate": 1.0842755868629891e-07,
      "loss": 0.6344,
      "step": 4946
    },
    {
      "epoch": 0.48903936930034847,
      "grad_norm": 4.329024447910644,
      "learning_rate": 1.0839565104461218e-07,
      "loss": 0.5819,
      "step": 4947
    },
    {
      "epoch": 0.48913822504510296,
      "grad_norm": 4.392438971097348,
      "learning_rate": 1.083637425420753e-07,
      "loss": 0.7464,
      "step": 4948
    },
    {
      "epoch": 0.4892370807898574,
      "grad_norm": 3.599142782017634,
      "learning_rate": 1.0833183318195997e-07,
      "loss": 0.6971,
      "step": 4949
    },
    {
      "epoch": 0.4893359365346119,
      "grad_norm": 6.727671288477943,
      "learning_rate": 1.0829992296753807e-07,
      "loss": 0.7277,
      "step": 4950
    },
    {
      "epoch": 0.4894347922793663,
      "grad_norm": 5.2901096652897115,
      "learning_rate": 1.082680119020815e-07,
      "loss": 0.6519,
      "step": 4951
    },
    {
      "epoch": 0.4895336480241208,
      "grad_norm": 5.102324251878221,
      "learning_rate": 1.0823609998886225e-07,
      "loss": 0.7061,
      "step": 4952
    },
    {
      "epoch": 0.4896325037688753,
      "grad_norm": 4.148032543498175,
      "learning_rate": 1.0820418723115246e-07,
      "loss": 0.657,
      "step": 4953
    },
    {
      "epoch": 0.4897313595136297,
      "grad_norm": 9.660494305659068,
      "learning_rate": 1.0817227363222425e-07,
      "loss": 0.6937,
      "step": 4954
    },
    {
      "epoch": 0.4898302152583842,
      "grad_norm": 9.145800112237929,
      "learning_rate": 1.0814035919534998e-07,
      "loss": 0.6701,
      "step": 4955
    },
    {
      "epoch": 0.48992907100313865,
      "grad_norm": 3.2385170266232337,
      "learning_rate": 1.0810844392380192e-07,
      "loss": 0.6434,
      "step": 4956
    },
    {
      "epoch": 0.49002792674789314,
      "grad_norm": 6.687435237480198,
      "learning_rate": 1.0807652782085255e-07,
      "loss": 0.5718,
      "step": 4957
    },
    {
      "epoch": 0.49012678249264763,
      "grad_norm": 3.975918914329794,
      "learning_rate": 1.0804461088977439e-07,
      "loss": 0.7477,
      "step": 4958
    },
    {
      "epoch": 0.49022563823740206,
      "grad_norm": 5.3922727118795875,
      "learning_rate": 1.0801269313384005e-07,
      "loss": 0.6425,
      "step": 4959
    },
    {
      "epoch": 0.49032449398215655,
      "grad_norm": 7.26258367513739,
      "learning_rate": 1.0798077455632225e-07,
      "loss": 0.8142,
      "step": 4960
    },
    {
      "epoch": 0.490423349726911,
      "grad_norm": 4.520364987380895,
      "learning_rate": 1.079488551604937e-07,
      "loss": 0.664,
      "step": 4961
    },
    {
      "epoch": 0.4905222054716655,
      "grad_norm": 3.8095858960602627,
      "learning_rate": 1.0791693494962735e-07,
      "loss": 0.7486,
      "step": 4962
    },
    {
      "epoch": 0.49062106121641996,
      "grad_norm": 3.477944933816795,
      "learning_rate": 1.0788501392699607e-07,
      "loss": 0.7324,
      "step": 4963
    },
    {
      "epoch": 0.4907199169611744,
      "grad_norm": 3.841160323580473,
      "learning_rate": 1.0785309209587294e-07,
      "loss": 0.7805,
      "step": 4964
    },
    {
      "epoch": 0.4908187727059289,
      "grad_norm": 4.137282273502954,
      "learning_rate": 1.0782116945953103e-07,
      "loss": 0.7728,
      "step": 4965
    },
    {
      "epoch": 0.4909176284506833,
      "grad_norm": 4.690748766502305,
      "learning_rate": 1.0778924602124359e-07,
      "loss": 0.815,
      "step": 4966
    },
    {
      "epoch": 0.4910164841954378,
      "grad_norm": 4.143948678023414,
      "learning_rate": 1.0775732178428388e-07,
      "loss": 0.6759,
      "step": 4967
    },
    {
      "epoch": 0.4911153399401923,
      "grad_norm": 3.897769804714911,
      "learning_rate": 1.0772539675192524e-07,
      "loss": 0.7087,
      "step": 4968
    },
    {
      "epoch": 0.49121419568494673,
      "grad_norm": 37.72190598504822,
      "learning_rate": 1.0769347092744112e-07,
      "loss": 0.6965,
      "step": 4969
    },
    {
      "epoch": 0.4913130514297012,
      "grad_norm": 6.371711871081384,
      "learning_rate": 1.0766154431410503e-07,
      "loss": 0.6356,
      "step": 4970
    },
    {
      "epoch": 0.49141190717445565,
      "grad_norm": 6.903172273037336,
      "learning_rate": 1.0762961691519062e-07,
      "loss": 0.6844,
      "step": 4971
    },
    {
      "epoch": 0.49151076291921014,
      "grad_norm": 3.691618512246146,
      "learning_rate": 1.0759768873397148e-07,
      "loss": 0.7027,
      "step": 4972
    },
    {
      "epoch": 0.49160961866396463,
      "grad_norm": 4.573614105269382,
      "learning_rate": 1.0756575977372149e-07,
      "loss": 0.6994,
      "step": 4973
    },
    {
      "epoch": 0.49170847440871907,
      "grad_norm": 7.11723595424024,
      "learning_rate": 1.0753383003771441e-07,
      "loss": 0.7093,
      "step": 4974
    },
    {
      "epoch": 0.49180733015347355,
      "grad_norm": 8.43079941025531,
      "learning_rate": 1.075018995292242e-07,
      "loss": 0.631,
      "step": 4975
    },
    {
      "epoch": 0.491906185898228,
      "grad_norm": 5.5401386696437935,
      "learning_rate": 1.0746996825152483e-07,
      "loss": 0.7699,
      "step": 4976
    },
    {
      "epoch": 0.4920050416429825,
      "grad_norm": 3.298834216814719,
      "learning_rate": 1.0743803620789047e-07,
      "loss": 0.7692,
      "step": 4977
    },
    {
      "epoch": 0.49210389738773697,
      "grad_norm": 3.585881000433193,
      "learning_rate": 1.0740610340159517e-07,
      "loss": 0.7808,
      "step": 4978
    },
    {
      "epoch": 0.4922027531324914,
      "grad_norm": 3.696715570267291,
      "learning_rate": 1.0737416983591321e-07,
      "loss": 0.6954,
      "step": 4979
    },
    {
      "epoch": 0.4923016088772459,
      "grad_norm": 9.625967380537853,
      "learning_rate": 1.0734223551411892e-07,
      "loss": 0.6648,
      "step": 4980
    },
    {
      "epoch": 0.4924004646220003,
      "grad_norm": 6.30308593212552,
      "learning_rate": 1.0731030043948667e-07,
      "loss": 0.6922,
      "step": 4981
    },
    {
      "epoch": 0.4924993203667548,
      "grad_norm": 14.112451421758925,
      "learning_rate": 1.0727836461529097e-07,
      "loss": 0.7646,
      "step": 4982
    },
    {
      "epoch": 0.4925981761115093,
      "grad_norm": 4.582849062488963,
      "learning_rate": 1.072464280448063e-07,
      "loss": 0.7249,
      "step": 4983
    },
    {
      "epoch": 0.49269703185626373,
      "grad_norm": 7.677843819211311,
      "learning_rate": 1.0721449073130738e-07,
      "loss": 0.7603,
      "step": 4984
    },
    {
      "epoch": 0.4927958876010182,
      "grad_norm": 3.190630291649785,
      "learning_rate": 1.0718255267806882e-07,
      "loss": 0.7758,
      "step": 4985
    },
    {
      "epoch": 0.49289474334577266,
      "grad_norm": 3.7498270934065485,
      "learning_rate": 1.0715061388836546e-07,
      "loss": 0.66,
      "step": 4986
    },
    {
      "epoch": 0.49299359909052715,
      "grad_norm": 3.735043621174924,
      "learning_rate": 1.071186743654721e-07,
      "loss": 0.681,
      "step": 4987
    },
    {
      "epoch": 0.49309245483528164,
      "grad_norm": 2.888198110448533,
      "learning_rate": 1.070867341126637e-07,
      "loss": 0.6898,
      "step": 4988
    },
    {
      "epoch": 0.49319131058003607,
      "grad_norm": 3.730553329203329,
      "learning_rate": 1.0705479313321526e-07,
      "loss": 0.7512,
      "step": 4989
    },
    {
      "epoch": 0.49329016632479056,
      "grad_norm": 4.905389476979263,
      "learning_rate": 1.0702285143040181e-07,
      "loss": 0.641,
      "step": 4990
    },
    {
      "epoch": 0.493389022069545,
      "grad_norm": 3.470931005840382,
      "learning_rate": 1.0699090900749858e-07,
      "loss": 0.6645,
      "step": 4991
    },
    {
      "epoch": 0.4934878778142995,
      "grad_norm": 4.595086793982326,
      "learning_rate": 1.0695896586778077e-07,
      "loss": 0.8125,
      "step": 4992
    },
    {
      "epoch": 0.49358673355905397,
      "grad_norm": 3.823470447683234,
      "learning_rate": 1.0692702201452364e-07,
      "loss": 0.8686,
      "step": 4993
    },
    {
      "epoch": 0.4936855893038084,
      "grad_norm": 2.9844375428344243,
      "learning_rate": 1.068950774510026e-07,
      "loss": 0.7339,
      "step": 4994
    },
    {
      "epoch": 0.4937844450485629,
      "grad_norm": 8.880984788533603,
      "learning_rate": 1.0686313218049306e-07,
      "loss": 0.7127,
      "step": 4995
    },
    {
      "epoch": 0.4938833007933173,
      "grad_norm": 5.8180761791677105,
      "learning_rate": 1.0683118620627056e-07,
      "loss": 0.6894,
      "step": 4996
    },
    {
      "epoch": 0.4939821565380718,
      "grad_norm": 4.591844241287601,
      "learning_rate": 1.0679923953161067e-07,
      "loss": 0.7175,
      "step": 4997
    },
    {
      "epoch": 0.4940810122828263,
      "grad_norm": 6.530912476683678,
      "learning_rate": 1.0676729215978905e-07,
      "loss": 0.7191,
      "step": 4998
    },
    {
      "epoch": 0.49417986802758074,
      "grad_norm": 3.9683884036815313,
      "learning_rate": 1.0673534409408147e-07,
      "loss": 0.7803,
      "step": 4999
    },
    {
      "epoch": 0.4942787237723352,
      "grad_norm": 4.197272000757149,
      "learning_rate": 1.067033953377637e-07,
      "loss": 0.5637,
      "step": 5000
    },
    {
      "epoch": 0.49437757951708966,
      "grad_norm": 2.684990338410304,
      "learning_rate": 1.0667144589411158e-07,
      "loss": 0.6712,
      "step": 5001
    },
    {
      "epoch": 0.49447643526184415,
      "grad_norm": 7.762974912945271,
      "learning_rate": 1.0663949576640114e-07,
      "loss": 0.7853,
      "step": 5002
    },
    {
      "epoch": 0.49457529100659864,
      "grad_norm": 3.394081531813035,
      "learning_rate": 1.0660754495790835e-07,
      "loss": 0.7058,
      "step": 5003
    },
    {
      "epoch": 0.49467414675135307,
      "grad_norm": 7.196613795039451,
      "learning_rate": 1.065755934719093e-07,
      "loss": 0.7531,
      "step": 5004
    },
    {
      "epoch": 0.49477300249610756,
      "grad_norm": 54.37138027758901,
      "learning_rate": 1.0654364131168012e-07,
      "loss": 0.7367,
      "step": 5005
    },
    {
      "epoch": 0.494871858240862,
      "grad_norm": 4.5279112488389295,
      "learning_rate": 1.0651168848049706e-07,
      "loss": 0.6614,
      "step": 5006
    },
    {
      "epoch": 0.4949707139856165,
      "grad_norm": 4.9012278128282905,
      "learning_rate": 1.0647973498163638e-07,
      "loss": 0.7591,
      "step": 5007
    },
    {
      "epoch": 0.495069569730371,
      "grad_norm": 3.007142921183577,
      "learning_rate": 1.0644778081837447e-07,
      "loss": 0.7288,
      "step": 5008
    },
    {
      "epoch": 0.4951684254751254,
      "grad_norm": 4.391830645113759,
      "learning_rate": 1.0641582599398773e-07,
      "loss": 0.7173,
      "step": 5009
    },
    {
      "epoch": 0.4952672812198799,
      "grad_norm": 4.309477122914594,
      "learning_rate": 1.063838705117527e-07,
      "loss": 0.6826,
      "step": 5010
    },
    {
      "epoch": 0.4953661369646344,
      "grad_norm": 3.1310574498744463,
      "learning_rate": 1.063519143749459e-07,
      "loss": 0.6513,
      "step": 5011
    },
    {
      "epoch": 0.4954649927093888,
      "grad_norm": 3.3234414419344906,
      "learning_rate": 1.06319957586844e-07,
      "loss": 0.7001,
      "step": 5012
    },
    {
      "epoch": 0.4955638484541433,
      "grad_norm": 4.0428888167506445,
      "learning_rate": 1.0628800015072368e-07,
      "loss": 0.6596,
      "step": 5013
    },
    {
      "epoch": 0.49566270419889774,
      "grad_norm": 3.3302362107812153,
      "learning_rate": 1.0625604206986168e-07,
      "loss": 0.642,
      "step": 5014
    },
    {
      "epoch": 0.49576155994365223,
      "grad_norm": 4.496326321937332,
      "learning_rate": 1.0622408334753485e-07,
      "loss": 0.6662,
      "step": 5015
    },
    {
      "epoch": 0.4958604156884067,
      "grad_norm": 3.070089463176683,
      "learning_rate": 1.0619212398702007e-07,
      "loss": 0.6216,
      "step": 5016
    },
    {
      "epoch": 0.49595927143316115,
      "grad_norm": 3.020486485168461,
      "learning_rate": 1.0616016399159433e-07,
      "loss": 0.62,
      "step": 5017
    },
    {
      "epoch": 0.49605812717791564,
      "grad_norm": 20.22838156412443,
      "learning_rate": 1.0612820336453465e-07,
      "loss": 0.677,
      "step": 5018
    },
    {
      "epoch": 0.4961569829226701,
      "grad_norm": 3.9063623596569204,
      "learning_rate": 1.0609624210911812e-07,
      "loss": 0.7841,
      "step": 5019
    },
    {
      "epoch": 0.49625583866742456,
      "grad_norm": 4.011213146817881,
      "learning_rate": 1.0606428022862184e-07,
      "loss": 0.5877,
      "step": 5020
    },
    {
      "epoch": 0.49635469441217905,
      "grad_norm": 3.703255820278696,
      "learning_rate": 1.0603231772632315e-07,
      "loss": 0.706,
      "step": 5021
    },
    {
      "epoch": 0.4964535501569335,
      "grad_norm": 7.28321698606604,
      "learning_rate": 1.0600035460549921e-07,
      "loss": 0.6038,
      "step": 5022
    },
    {
      "epoch": 0.496552405901688,
      "grad_norm": 4.195299232698053,
      "learning_rate": 1.0596839086942741e-07,
      "loss": 0.7447,
      "step": 5023
    },
    {
      "epoch": 0.4966512616464424,
      "grad_norm": 6.76027746362631,
      "learning_rate": 1.0593642652138519e-07,
      "loss": 0.6998,
      "step": 5024
    },
    {
      "epoch": 0.4967501173911969,
      "grad_norm": 4.100007129441382,
      "learning_rate": 1.0590446156465e-07,
      "loss": 0.6783,
      "step": 5025
    },
    {
      "epoch": 0.4968489731359514,
      "grad_norm": 4.6506920134143,
      "learning_rate": 1.0587249600249937e-07,
      "loss": 0.7802,
      "step": 5026
    },
    {
      "epoch": 0.4969478288807058,
      "grad_norm": 7.231279095645681,
      "learning_rate": 1.0584052983821087e-07,
      "loss": 0.7239,
      "step": 5027
    },
    {
      "epoch": 0.4970466846254603,
      "grad_norm": 4.560547802296419,
      "learning_rate": 1.0580856307506223e-07,
      "loss": 0.692,
      "step": 5028
    },
    {
      "epoch": 0.49714554037021474,
      "grad_norm": 4.038279041149098,
      "learning_rate": 1.0577659571633111e-07,
      "loss": 0.6925,
      "step": 5029
    },
    {
      "epoch": 0.49724439611496923,
      "grad_norm": 3.7659357234343815,
      "learning_rate": 1.0574462776529531e-07,
      "loss": 0.7079,
      "step": 5030
    },
    {
      "epoch": 0.4973432518597237,
      "grad_norm": 3.639560907065725,
      "learning_rate": 1.0571265922523269e-07,
      "loss": 0.6966,
      "step": 5031
    },
    {
      "epoch": 0.49744210760447816,
      "grad_norm": 6.320479606522326,
      "learning_rate": 1.056806900994211e-07,
      "loss": 0.7409,
      "step": 5032
    },
    {
      "epoch": 0.49754096334923265,
      "grad_norm": 3.357759264460438,
      "learning_rate": 1.0564872039113857e-07,
      "loss": 0.7769,
      "step": 5033
    },
    {
      "epoch": 0.4976398190939871,
      "grad_norm": 3.493563400968946,
      "learning_rate": 1.0561675010366306e-07,
      "loss": 0.8383,
      "step": 5034
    },
    {
      "epoch": 0.49773867483874157,
      "grad_norm": 4.596751370131817,
      "learning_rate": 1.055847792402727e-07,
      "loss": 0.7593,
      "step": 5035
    },
    {
      "epoch": 0.49783753058349606,
      "grad_norm": 3.261957459812754,
      "learning_rate": 1.055528078042456e-07,
      "loss": 0.6537,
      "step": 5036
    },
    {
      "epoch": 0.4979363863282505,
      "grad_norm": 3.8465278989835383,
      "learning_rate": 1.0552083579885998e-07,
      "loss": 0.6029,
      "step": 5037
    },
    {
      "epoch": 0.498035242073005,
      "grad_norm": 3.2451251957141123,
      "learning_rate": 1.054888632273941e-07,
      "loss": 0.7566,
      "step": 5038
    },
    {
      "epoch": 0.4981340978177594,
      "grad_norm": 4.768612960692695,
      "learning_rate": 1.0545689009312624e-07,
      "loss": 0.7796,
      "step": 5039
    },
    {
      "epoch": 0.4982329535625139,
      "grad_norm": 3.8758635074619776,
      "learning_rate": 1.054249163993348e-07,
      "loss": 0.7251,
      "step": 5040
    },
    {
      "epoch": 0.4983318093072684,
      "grad_norm": 3.7518217612703477,
      "learning_rate": 1.0539294214929821e-07,
      "loss": 0.6249,
      "step": 5041
    },
    {
      "epoch": 0.4984306650520228,
      "grad_norm": 6.265295092139124,
      "learning_rate": 1.0536096734629494e-07,
      "loss": 0.7092,
      "step": 5042
    },
    {
      "epoch": 0.4985295207967773,
      "grad_norm": 4.052293808920071,
      "learning_rate": 1.0532899199360354e-07,
      "loss": 0.6427,
      "step": 5043
    },
    {
      "epoch": 0.49862837654153175,
      "grad_norm": 2.5715405623678405,
      "learning_rate": 1.0529701609450266e-07,
      "loss": 0.663,
      "step": 5044
    },
    {
      "epoch": 0.49872723228628624,
      "grad_norm": 4.38825780774269,
      "learning_rate": 1.0526503965227087e-07,
      "loss": 0.726,
      "step": 5045
    },
    {
      "epoch": 0.4988260880310407,
      "grad_norm": 2.9601955626616734,
      "learning_rate": 1.0523306267018695e-07,
      "loss": 0.7776,
      "step": 5046
    },
    {
      "epoch": 0.49892494377579516,
      "grad_norm": 6.246606028877477,
      "learning_rate": 1.0520108515152966e-07,
      "loss": 0.6923,
      "step": 5047
    },
    {
      "epoch": 0.49902379952054965,
      "grad_norm": 5.14330628594759,
      "learning_rate": 1.051691070995778e-07,
      "loss": 0.7717,
      "step": 5048
    },
    {
      "epoch": 0.4991226552653041,
      "grad_norm": 3.4823069337695833,
      "learning_rate": 1.0513712851761025e-07,
      "loss": 0.7207,
      "step": 5049
    },
    {
      "epoch": 0.49922151101005857,
      "grad_norm": 3.26302499917406,
      "learning_rate": 1.0510514940890596e-07,
      "loss": 0.7716,
      "step": 5050
    },
    {
      "epoch": 0.49932036675481306,
      "grad_norm": 4.3728703416083015,
      "learning_rate": 1.050731697767439e-07,
      "loss": 0.706,
      "step": 5051
    },
    {
      "epoch": 0.4994192224995675,
      "grad_norm": 5.4408352585657855,
      "learning_rate": 1.0504118962440311e-07,
      "loss": 0.6704,
      "step": 5052
    },
    {
      "epoch": 0.499518078244322,
      "grad_norm": 3.4812510528833402,
      "learning_rate": 1.0500920895516266e-07,
      "loss": 0.7085,
      "step": 5053
    },
    {
      "epoch": 0.4996169339890764,
      "grad_norm": 3.232223468575376,
      "learning_rate": 1.0497722777230176e-07,
      "loss": 0.7111,
      "step": 5054
    },
    {
      "epoch": 0.4997157897338309,
      "grad_norm": 5.548592464449253,
      "learning_rate": 1.0494524607909956e-07,
      "loss": 0.7049,
      "step": 5055
    },
    {
      "epoch": 0.4998146454785854,
      "grad_norm": 3.785803594530723,
      "learning_rate": 1.0491326387883532e-07,
      "loss": 0.6959,
      "step": 5056
    },
    {
      "epoch": 0.49991350122333983,
      "grad_norm": 3.2230721962357,
      "learning_rate": 1.0488128117478836e-07,
      "loss": 0.79,
      "step": 5057
    },
    {
      "epoch": 0.5000123569680943,
      "grad_norm": 23.531308969988828,
      "learning_rate": 1.04849297970238e-07,
      "loss": 0.5968,
      "step": 5058
    },
    {
      "epoch": 0.5001112127128488,
      "grad_norm": 4.85752578815118,
      "learning_rate": 1.048173142684637e-07,
      "loss": 0.6882,
      "step": 5059
    },
    {
      "epoch": 0.5002100684576033,
      "grad_norm": 3.3964328554239462,
      "learning_rate": 1.0478533007274482e-07,
      "loss": 0.7655,
      "step": 5060
    },
    {
      "epoch": 0.5003089242023577,
      "grad_norm": 4.600841991442908,
      "learning_rate": 1.0475334538636098e-07,
      "loss": 0.6631,
      "step": 5061
    },
    {
      "epoch": 0.5004077799471122,
      "grad_norm": 5.802837513349761,
      "learning_rate": 1.0472136021259166e-07,
      "loss": 0.7694,
      "step": 5062
    },
    {
      "epoch": 0.5005066356918666,
      "grad_norm": 3.483054735334268,
      "learning_rate": 1.0468937455471652e-07,
      "loss": 0.6191,
      "step": 5063
    },
    {
      "epoch": 0.5006054914366211,
      "grad_norm": 3.777220123432716,
      "learning_rate": 1.0465738841601516e-07,
      "loss": 0.6588,
      "step": 5064
    },
    {
      "epoch": 0.5007043471813756,
      "grad_norm": 3.259472870371891,
      "learning_rate": 1.0462540179976735e-07,
      "loss": 0.742,
      "step": 5065
    },
    {
      "epoch": 0.50080320292613,
      "grad_norm": 3.6318339877154178,
      "learning_rate": 1.0459341470925283e-07,
      "loss": 0.7148,
      "step": 5066
    },
    {
      "epoch": 0.5009020586708846,
      "grad_norm": 7.768127885232397,
      "learning_rate": 1.0456142714775134e-07,
      "loss": 0.7385,
      "step": 5067
    },
    {
      "epoch": 0.501000914415639,
      "grad_norm": 2.7609873819795667,
      "learning_rate": 1.0452943911854279e-07,
      "loss": 0.7288,
      "step": 5068
    },
    {
      "epoch": 0.5010997701603934,
      "grad_norm": 4.685291127054411,
      "learning_rate": 1.0449745062490711e-07,
      "loss": 0.6148,
      "step": 5069
    },
    {
      "epoch": 0.501198625905148,
      "grad_norm": 5.067410968390232,
      "learning_rate": 1.0446546167012418e-07,
      "loss": 0.8727,
      "step": 5070
    },
    {
      "epoch": 0.5012974816499024,
      "grad_norm": 4.6172892146910804,
      "learning_rate": 1.0443347225747403e-07,
      "loss": 0.6946,
      "step": 5071
    },
    {
      "epoch": 0.5013963373946568,
      "grad_norm": 3.327521992282824,
      "learning_rate": 1.0440148239023669e-07,
      "loss": 0.7325,
      "step": 5072
    },
    {
      "epoch": 0.5014951931394113,
      "grad_norm": 5.040498064472465,
      "learning_rate": 1.0436949207169228e-07,
      "loss": 0.6842,
      "step": 5073
    },
    {
      "epoch": 0.5015940488841658,
      "grad_norm": 6.435603453996979,
      "learning_rate": 1.0433750130512092e-07,
      "loss": 0.737,
      "step": 5074
    },
    {
      "epoch": 0.5016929046289202,
      "grad_norm": 3.864804608093426,
      "learning_rate": 1.0430551009380275e-07,
      "loss": 0.7834,
      "step": 5075
    },
    {
      "epoch": 0.5017917603736747,
      "grad_norm": 3.7715096531100034,
      "learning_rate": 1.0427351844101806e-07,
      "loss": 0.6934,
      "step": 5076
    },
    {
      "epoch": 0.5018906161184292,
      "grad_norm": 28.95598901390921,
      "learning_rate": 1.0424152635004709e-07,
      "loss": 0.7093,
      "step": 5077
    },
    {
      "epoch": 0.5019894718631837,
      "grad_norm": 3.022504717229382,
      "learning_rate": 1.0420953382417013e-07,
      "loss": 0.7055,
      "step": 5078
    },
    {
      "epoch": 0.5020883276079381,
      "grad_norm": 3.7343294773832527,
      "learning_rate": 1.0417754086666759e-07,
      "loss": 0.7308,
      "step": 5079
    },
    {
      "epoch": 0.5021871833526926,
      "grad_norm": 3.367538565609187,
      "learning_rate": 1.0414554748081985e-07,
      "loss": 0.7023,
      "step": 5080
    },
    {
      "epoch": 0.5022860390974471,
      "grad_norm": 5.035334503354703,
      "learning_rate": 1.0411355366990737e-07,
      "loss": 0.6035,
      "step": 5081
    },
    {
      "epoch": 0.5023848948422015,
      "grad_norm": 13.296338822586444,
      "learning_rate": 1.0408155943721066e-07,
      "loss": 0.6405,
      "step": 5082
    },
    {
      "epoch": 0.5024837505869559,
      "grad_norm": 4.454165977119614,
      "learning_rate": 1.0404956478601021e-07,
      "loss": 0.7537,
      "step": 5083
    },
    {
      "epoch": 0.5025826063317105,
      "grad_norm": 5.273263540296186,
      "learning_rate": 1.0401756971958666e-07,
      "loss": 0.6619,
      "step": 5084
    },
    {
      "epoch": 0.5026814620764649,
      "grad_norm": 4.803211778154168,
      "learning_rate": 1.0398557424122054e-07,
      "loss": 0.7319,
      "step": 5085
    },
    {
      "epoch": 0.5027803178212193,
      "grad_norm": 5.5100420831309025,
      "learning_rate": 1.0395357835419263e-07,
      "loss": 0.6834,
      "step": 5086
    },
    {
      "epoch": 0.5028791735659739,
      "grad_norm": 3.3279158132159896,
      "learning_rate": 1.0392158206178356e-07,
      "loss": 0.6641,
      "step": 5087
    },
    {
      "epoch": 0.5029780293107283,
      "grad_norm": 3.408262926431073,
      "learning_rate": 1.038895853672741e-07,
      "loss": 0.5861,
      "step": 5088
    },
    {
      "epoch": 0.5030768850554828,
      "grad_norm": 5.997607109953472,
      "learning_rate": 1.0385758827394502e-07,
      "loss": 0.7022,
      "step": 5089
    },
    {
      "epoch": 0.5031757408002373,
      "grad_norm": 8.552085199645681,
      "learning_rate": 1.0382559078507717e-07,
      "loss": 0.7537,
      "step": 5090
    },
    {
      "epoch": 0.5032745965449917,
      "grad_norm": 18.24595993812914,
      "learning_rate": 1.0379359290395142e-07,
      "loss": 0.7635,
      "step": 5091
    },
    {
      "epoch": 0.5033734522897462,
      "grad_norm": 4.829693449294128,
      "learning_rate": 1.037615946338487e-07,
      "loss": 0.6548,
      "step": 5092
    },
    {
      "epoch": 0.5034723080345006,
      "grad_norm": 3.16054147862091,
      "learning_rate": 1.0372959597804992e-07,
      "loss": 0.739,
      "step": 5093
    },
    {
      "epoch": 0.5035711637792551,
      "grad_norm": 4.8708693283791495,
      "learning_rate": 1.0369759693983608e-07,
      "loss": 0.6941,
      "step": 5094
    },
    {
      "epoch": 0.5036700195240096,
      "grad_norm": 3.4371241105043517,
      "learning_rate": 1.0366559752248823e-07,
      "loss": 0.7047,
      "step": 5095
    },
    {
      "epoch": 0.503768875268764,
      "grad_norm": 3.494896385057394,
      "learning_rate": 1.036335977292874e-07,
      "loss": 0.6932,
      "step": 5096
    },
    {
      "epoch": 0.5038677310135186,
      "grad_norm": 8.062785604954087,
      "learning_rate": 1.0360159756351474e-07,
      "loss": 0.7593,
      "step": 5097
    },
    {
      "epoch": 0.503966586758273,
      "grad_norm": 3.6887712257974026,
      "learning_rate": 1.0356959702845139e-07,
      "loss": 0.7505,
      "step": 5098
    },
    {
      "epoch": 0.5040654425030274,
      "grad_norm": 4.493360868755047,
      "learning_rate": 1.0353759612737851e-07,
      "loss": 0.7897,
      "step": 5099
    },
    {
      "epoch": 0.504164298247782,
      "grad_norm": 3.2044107262557455,
      "learning_rate": 1.0350559486357732e-07,
      "loss": 0.7438,
      "step": 5100
    },
    {
      "epoch": 0.5042631539925364,
      "grad_norm": 2.7455519238809543,
      "learning_rate": 1.0347359324032912e-07,
      "loss": 0.6502,
      "step": 5101
    },
    {
      "epoch": 0.5043620097372908,
      "grad_norm": 3.223588495610112,
      "learning_rate": 1.0344159126091514e-07,
      "loss": 0.5932,
      "step": 5102
    },
    {
      "epoch": 0.5044608654820453,
      "grad_norm": 8.64157431519327,
      "learning_rate": 1.0340958892861675e-07,
      "loss": 0.6908,
      "step": 5103
    },
    {
      "epoch": 0.5045597212267998,
      "grad_norm": 3.918983925152668,
      "learning_rate": 1.0337758624671531e-07,
      "loss": 0.6875,
      "step": 5104
    },
    {
      "epoch": 0.5046585769715543,
      "grad_norm": 3.8080882943896426,
      "learning_rate": 1.0334558321849223e-07,
      "loss": 0.6938,
      "step": 5105
    },
    {
      "epoch": 0.5047574327163087,
      "grad_norm": 3.4184332150044354,
      "learning_rate": 1.0331357984722893e-07,
      "loss": 0.7184,
      "step": 5106
    },
    {
      "epoch": 0.5048562884610632,
      "grad_norm": 3.056512970246195,
      "learning_rate": 1.0328157613620689e-07,
      "loss": 0.7281,
      "step": 5107
    },
    {
      "epoch": 0.5049551442058177,
      "grad_norm": 14.220366318296442,
      "learning_rate": 1.0324957208870764e-07,
      "loss": 0.7566,
      "step": 5108
    },
    {
      "epoch": 0.5050539999505721,
      "grad_norm": 9.071063719628482,
      "learning_rate": 1.0321756770801273e-07,
      "loss": 0.8569,
      "step": 5109
    },
    {
      "epoch": 0.5051528556953266,
      "grad_norm": 22.874070004169347,
      "learning_rate": 1.0318556299740368e-07,
      "loss": 0.6875,
      "step": 5110
    },
    {
      "epoch": 0.5052517114400811,
      "grad_norm": 21.982008593108702,
      "learning_rate": 1.0315355796016214e-07,
      "loss": 0.7047,
      "step": 5111
    },
    {
      "epoch": 0.5053505671848355,
      "grad_norm": 5.534161950524919,
      "learning_rate": 1.0312155259956976e-07,
      "loss": 0.7331,
      "step": 5112
    },
    {
      "epoch": 0.5054494229295899,
      "grad_norm": 5.679967048738447,
      "learning_rate": 1.0308954691890819e-07,
      "loss": 0.7543,
      "step": 5113
    },
    {
      "epoch": 0.5055482786743445,
      "grad_norm": 3.8671769368750684,
      "learning_rate": 1.0305754092145917e-07,
      "loss": 0.7414,
      "step": 5114
    },
    {
      "epoch": 0.5056471344190989,
      "grad_norm": 5.2003656977615895,
      "learning_rate": 1.030255346105044e-07,
      "loss": 0.6228,
      "step": 5115
    },
    {
      "epoch": 0.5057459901638534,
      "grad_norm": 5.218802482180568,
      "learning_rate": 1.0299352798932573e-07,
      "loss": 0.7447,
      "step": 5116
    },
    {
      "epoch": 0.5058448459086079,
      "grad_norm": 3.2587440320237966,
      "learning_rate": 1.0296152106120491e-07,
      "loss": 0.7435,
      "step": 5117
    },
    {
      "epoch": 0.5059437016533623,
      "grad_norm": 6.088887244130914,
      "learning_rate": 1.0292951382942378e-07,
      "loss": 0.7588,
      "step": 5118
    },
    {
      "epoch": 0.5060425573981168,
      "grad_norm": 4.5291697676512666,
      "learning_rate": 1.0289750629726423e-07,
      "loss": 0.799,
      "step": 5119
    },
    {
      "epoch": 0.5061414131428713,
      "grad_norm": 4.163832501718261,
      "learning_rate": 1.0286549846800816e-07,
      "loss": 0.7405,
      "step": 5120
    },
    {
      "epoch": 0.5062402688876257,
      "grad_norm": 3.3796633520404247,
      "learning_rate": 1.0283349034493749e-07,
      "loss": 0.7056,
      "step": 5121
    },
    {
      "epoch": 0.5063391246323802,
      "grad_norm": 16.73013739419581,
      "learning_rate": 1.0280148193133413e-07,
      "loss": 0.7281,
      "step": 5122
    },
    {
      "epoch": 0.5064379803771346,
      "grad_norm": 9.975423516502135,
      "learning_rate": 1.0276947323048017e-07,
      "loss": 0.6973,
      "step": 5123
    },
    {
      "epoch": 0.5065368361218892,
      "grad_norm": 3.836962076531748,
      "learning_rate": 1.0273746424565758e-07,
      "loss": 0.7428,
      "step": 5124
    },
    {
      "epoch": 0.5066356918666436,
      "grad_norm": 3.8179497709529238,
      "learning_rate": 1.0270545498014842e-07,
      "loss": 0.7078,
      "step": 5125
    },
    {
      "epoch": 0.506734547611398,
      "grad_norm": 4.762315556206525,
      "learning_rate": 1.0267344543723475e-07,
      "loss": 0.7376,
      "step": 5126
    },
    {
      "epoch": 0.5068334033561526,
      "grad_norm": 23.804401027496468,
      "learning_rate": 1.0264143562019869e-07,
      "loss": 0.755,
      "step": 5127
    },
    {
      "epoch": 0.506932259100907,
      "grad_norm": 3.1423667171901974,
      "learning_rate": 1.0260942553232236e-07,
      "loss": 0.8391,
      "step": 5128
    },
    {
      "epoch": 0.5070311148456614,
      "grad_norm": 3.448777551216671,
      "learning_rate": 1.0257741517688792e-07,
      "loss": 0.7023,
      "step": 5129
    },
    {
      "epoch": 0.507129970590416,
      "grad_norm": 5.8102254002219995,
      "learning_rate": 1.0254540455717759e-07,
      "loss": 0.7307,
      "step": 5130
    },
    {
      "epoch": 0.5072288263351704,
      "grad_norm": 4.713276967394988,
      "learning_rate": 1.0251339367647358e-07,
      "loss": 0.7176,
      "step": 5131
    },
    {
      "epoch": 0.5073276820799248,
      "grad_norm": 4.993035647270986,
      "learning_rate": 1.0248138253805812e-07,
      "loss": 0.718,
      "step": 5132
    },
    {
      "epoch": 0.5074265378246794,
      "grad_norm": 5.434236201759911,
      "learning_rate": 1.0244937114521346e-07,
      "loss": 0.5812,
      "step": 5133
    },
    {
      "epoch": 0.5075253935694338,
      "grad_norm": 6.219552462818052,
      "learning_rate": 1.0241735950122194e-07,
      "loss": 0.6319,
      "step": 5134
    },
    {
      "epoch": 0.5076242493141883,
      "grad_norm": 2.852682538350435,
      "learning_rate": 1.0238534760936588e-07,
      "loss": 0.6421,
      "step": 5135
    },
    {
      "epoch": 0.5077231050589427,
      "grad_norm": 5.097951143512171,
      "learning_rate": 1.0235333547292761e-07,
      "loss": 0.6886,
      "step": 5136
    },
    {
      "epoch": 0.5078219608036972,
      "grad_norm": 3.772337278373112,
      "learning_rate": 1.0232132309518946e-07,
      "loss": 0.6044,
      "step": 5137
    },
    {
      "epoch": 0.5079208165484517,
      "grad_norm": 3.9054236547256176,
      "learning_rate": 1.0228931047943389e-07,
      "loss": 0.6452,
      "step": 5138
    },
    {
      "epoch": 0.5080196722932061,
      "grad_norm": 3.8508755939247865,
      "learning_rate": 1.022572976289433e-07,
      "loss": 0.61,
      "step": 5139
    },
    {
      "epoch": 0.5081185280379606,
      "grad_norm": 8.499286326453216,
      "learning_rate": 1.0222528454700013e-07,
      "loss": 0.7948,
      "step": 5140
    },
    {
      "epoch": 0.5082173837827151,
      "grad_norm": 4.164683368315368,
      "learning_rate": 1.0219327123688686e-07,
      "loss": 0.8173,
      "step": 5141
    },
    {
      "epoch": 0.5083162395274695,
      "grad_norm": 4.8170643666537,
      "learning_rate": 1.0216125770188597e-07,
      "loss": 0.7723,
      "step": 5142
    },
    {
      "epoch": 0.5084150952722241,
      "grad_norm": 7.572281133703752,
      "learning_rate": 1.0212924394528e-07,
      "loss": 0.6941,
      "step": 5143
    },
    {
      "epoch": 0.5085139510169785,
      "grad_norm": 4.555360599937588,
      "learning_rate": 1.0209722997035146e-07,
      "loss": 0.6208,
      "step": 5144
    },
    {
      "epoch": 0.5086128067617329,
      "grad_norm": 5.179980797423835,
      "learning_rate": 1.0206521578038294e-07,
      "loss": 0.6981,
      "step": 5145
    },
    {
      "epoch": 0.5087116625064874,
      "grad_norm": 3.1993187337342985,
      "learning_rate": 1.0203320137865699e-07,
      "loss": 0.6813,
      "step": 5146
    },
    {
      "epoch": 0.5088105182512419,
      "grad_norm": 3.936348817172037,
      "learning_rate": 1.0200118676845624e-07,
      "loss": 0.7492,
      "step": 5147
    },
    {
      "epoch": 0.5089093739959963,
      "grad_norm": 12.927839544668123,
      "learning_rate": 1.0196917195306328e-07,
      "loss": 0.686,
      "step": 5148
    },
    {
      "epoch": 0.5090082297407508,
      "grad_norm": 4.195562827286109,
      "learning_rate": 1.0193715693576079e-07,
      "loss": 0.8012,
      "step": 5149
    },
    {
      "epoch": 0.5091070854855053,
      "grad_norm": 5.18228485381953,
      "learning_rate": 1.0190514171983146e-07,
      "loss": 0.7079,
      "step": 5150
    },
    {
      "epoch": 0.5092059412302598,
      "grad_norm": 4.3051513160659765,
      "learning_rate": 1.0187312630855791e-07,
      "loss": 0.6558,
      "step": 5151
    },
    {
      "epoch": 0.5093047969750142,
      "grad_norm": 3.669228064002108,
      "learning_rate": 1.0184111070522292e-07,
      "loss": 0.712,
      "step": 5152
    },
    {
      "epoch": 0.5094036527197687,
      "grad_norm": 2.8312284064662188,
      "learning_rate": 1.0180909491310921e-07,
      "loss": 0.7698,
      "step": 5153
    },
    {
      "epoch": 0.5095025084645232,
      "grad_norm": 4.678167712805343,
      "learning_rate": 1.0177707893549947e-07,
      "loss": 0.6712,
      "step": 5154
    },
    {
      "epoch": 0.5096013642092776,
      "grad_norm": 3.9397437605349435,
      "learning_rate": 1.0174506277567647e-07,
      "loss": 0.7273,
      "step": 5155
    },
    {
      "epoch": 0.509700219954032,
      "grad_norm": 5.550400268314377,
      "learning_rate": 1.0171304643692305e-07,
      "loss": 0.686,
      "step": 5156
    },
    {
      "epoch": 0.5097990756987866,
      "grad_norm": 5.976739558639031,
      "learning_rate": 1.0168102992252199e-07,
      "loss": 0.748,
      "step": 5157
    },
    {
      "epoch": 0.509897931443541,
      "grad_norm": 142.2819580157784,
      "learning_rate": 1.0164901323575612e-07,
      "loss": 0.712,
      "step": 5158
    },
    {
      "epoch": 0.5099967871882954,
      "grad_norm": 3.6344098239381113,
      "learning_rate": 1.0161699637990824e-07,
      "loss": 0.7775,
      "step": 5159
    },
    {
      "epoch": 0.51009564293305,
      "grad_norm": 3.8497019485257304,
      "learning_rate": 1.0158497935826125e-07,
      "loss": 0.7661,
      "step": 5160
    },
    {
      "epoch": 0.5101944986778044,
      "grad_norm": 5.204497622541665,
      "learning_rate": 1.0155296217409802e-07,
      "loss": 0.6897,
      "step": 5161
    },
    {
      "epoch": 0.5102933544225589,
      "grad_norm": 4.673098396692905,
      "learning_rate": 1.0152094483070139e-07,
      "loss": 0.7807,
      "step": 5162
    },
    {
      "epoch": 0.5103922101673134,
      "grad_norm": 5.693224136089495,
      "learning_rate": 1.0148892733135434e-07,
      "loss": 0.7649,
      "step": 5163
    },
    {
      "epoch": 0.5104910659120678,
      "grad_norm": 4.24875032997248,
      "learning_rate": 1.0145690967933973e-07,
      "loss": 0.6796,
      "step": 5164
    },
    {
      "epoch": 0.5105899216568223,
      "grad_norm": 3.7453503495465967,
      "learning_rate": 1.0142489187794054e-07,
      "loss": 0.8031,
      "step": 5165
    },
    {
      "epoch": 0.5106887774015767,
      "grad_norm": 9.061420243019416,
      "learning_rate": 1.0139287393043969e-07,
      "loss": 0.7696,
      "step": 5166
    },
    {
      "epoch": 0.5107876331463312,
      "grad_norm": 3.8839209691581367,
      "learning_rate": 1.0136085584012017e-07,
      "loss": 0.9186,
      "step": 5167
    },
    {
      "epoch": 0.5108864888910857,
      "grad_norm": 4.920805585373029,
      "learning_rate": 1.0132883761026499e-07,
      "loss": 0.8102,
      "step": 5168
    },
    {
      "epoch": 0.5109853446358401,
      "grad_norm": 3.476681553549399,
      "learning_rate": 1.0129681924415708e-07,
      "loss": 0.747,
      "step": 5169
    },
    {
      "epoch": 0.5110842003805947,
      "grad_norm": 3.164388232449787,
      "learning_rate": 1.0126480074507949e-07,
      "loss": 0.8058,
      "step": 5170
    },
    {
      "epoch": 0.5111830561253491,
      "grad_norm": 4.306732297698357,
      "learning_rate": 1.0123278211631525e-07,
      "loss": 0.636,
      "step": 5171
    },
    {
      "epoch": 0.5112819118701035,
      "grad_norm": 3.370488024708446,
      "learning_rate": 1.0120076336114738e-07,
      "loss": 0.6589,
      "step": 5172
    },
    {
      "epoch": 0.5113807676148581,
      "grad_norm": 3.415411654343872,
      "learning_rate": 1.0116874448285892e-07,
      "loss": 0.6243,
      "step": 5173
    },
    {
      "epoch": 0.5114796233596125,
      "grad_norm": 3.5450332726854112,
      "learning_rate": 1.0113672548473297e-07,
      "loss": 0.7661,
      "step": 5174
    },
    {
      "epoch": 0.5115784791043669,
      "grad_norm": 5.171405128140053,
      "learning_rate": 1.0110470637005261e-07,
      "loss": 0.6753,
      "step": 5175
    },
    {
      "epoch": 0.5116773348491214,
      "grad_norm": 4.166783663456934,
      "learning_rate": 1.010726871421009e-07,
      "loss": 0.6717,
      "step": 5176
    },
    {
      "epoch": 0.5117761905938759,
      "grad_norm": 6.567135829836197,
      "learning_rate": 1.0104066780416091e-07,
      "loss": 0.6553,
      "step": 5177
    },
    {
      "epoch": 0.5118750463386303,
      "grad_norm": 3.57172831700378,
      "learning_rate": 1.0100864835951584e-07,
      "loss": 0.7242,
      "step": 5178
    },
    {
      "epoch": 0.5119739020833848,
      "grad_norm": 3.8908196605736904,
      "learning_rate": 1.0097662881144873e-07,
      "loss": 0.7761,
      "step": 5179
    },
    {
      "epoch": 0.5120727578281393,
      "grad_norm": 3.9559701276201094,
      "learning_rate": 1.0094460916324279e-07,
      "loss": 0.5855,
      "step": 5180
    },
    {
      "epoch": 0.5121716135728938,
      "grad_norm": 3.237798191440199,
      "learning_rate": 1.0091258941818104e-07,
      "loss": 0.7348,
      "step": 5181
    },
    {
      "epoch": 0.5122704693176482,
      "grad_norm": 3.3036582043503078,
      "learning_rate": 1.0088056957954673e-07,
      "loss": 0.7364,
      "step": 5182
    },
    {
      "epoch": 0.5123693250624027,
      "grad_norm": 4.089944748045106,
      "learning_rate": 1.0084854965062301e-07,
      "loss": 0.6559,
      "step": 5183
    },
    {
      "epoch": 0.5124681808071572,
      "grad_norm": 7.548530594243388,
      "learning_rate": 1.0081652963469303e-07,
      "loss": 0.8397,
      "step": 5184
    },
    {
      "epoch": 0.5125670365519116,
      "grad_norm": 4.623602792953504,
      "learning_rate": 1.0078450953504001e-07,
      "loss": 0.7855,
      "step": 5185
    },
    {
      "epoch": 0.512665892296666,
      "grad_norm": 2.907186252271077,
      "learning_rate": 1.007524893549471e-07,
      "loss": 0.7495,
      "step": 5186
    },
    {
      "epoch": 0.5127647480414206,
      "grad_norm": 9.920193790916882,
      "learning_rate": 1.0072046909769747e-07,
      "loss": 0.638,
      "step": 5187
    },
    {
      "epoch": 0.512863603786175,
      "grad_norm": 3.9355850675224477,
      "learning_rate": 1.006884487665744e-07,
      "loss": 0.774,
      "step": 5188
    },
    {
      "epoch": 0.5129624595309294,
      "grad_norm": 3.448045786761459,
      "learning_rate": 1.0065642836486106e-07,
      "loss": 0.6504,
      "step": 5189
    },
    {
      "epoch": 0.513061315275684,
      "grad_norm": 6.552323919328628,
      "learning_rate": 1.0062440789584066e-07,
      "loss": 0.6768,
      "step": 5190
    },
    {
      "epoch": 0.5131601710204384,
      "grad_norm": 6.083481870527904,
      "learning_rate": 1.0059238736279645e-07,
      "loss": 0.6908,
      "step": 5191
    },
    {
      "epoch": 0.5132590267651929,
      "grad_norm": 5.792277355776187,
      "learning_rate": 1.0056036676901163e-07,
      "loss": 0.6785,
      "step": 5192
    },
    {
      "epoch": 0.5133578825099474,
      "grad_norm": 7.41000081424419,
      "learning_rate": 1.005283461177695e-07,
      "loss": 0.676,
      "step": 5193
    },
    {
      "epoch": 0.5134567382547018,
      "grad_norm": 3.8747432307773657,
      "learning_rate": 1.0049632541235324e-07,
      "loss": 0.7829,
      "step": 5194
    },
    {
      "epoch": 0.5135555939994563,
      "grad_norm": 4.7933927384914545,
      "learning_rate": 1.0046430465604614e-07,
      "loss": 0.5304,
      "step": 5195
    },
    {
      "epoch": 0.5136544497442107,
      "grad_norm": 3.3367437924634484,
      "learning_rate": 1.0043228385213147e-07,
      "loss": 0.7934,
      "step": 5196
    },
    {
      "epoch": 0.5137533054889653,
      "grad_norm": 3.8873580859976236,
      "learning_rate": 1.0040026300389246e-07,
      "loss": 0.6835,
      "step": 5197
    },
    {
      "epoch": 0.5138521612337197,
      "grad_norm": 4.4507919105265525,
      "learning_rate": 1.0036824211461237e-07,
      "loss": 0.6893,
      "step": 5198
    },
    {
      "epoch": 0.5139510169784741,
      "grad_norm": 4.017058438683735,
      "learning_rate": 1.0033622118757448e-07,
      "loss": 0.7206,
      "step": 5199
    },
    {
      "epoch": 0.5140498727232287,
      "grad_norm": 3.9534355430569166,
      "learning_rate": 1.0030420022606205e-07,
      "loss": 0.6959,
      "step": 5200
    },
    {
      "epoch": 0.5141487284679831,
      "grad_norm": 4.616934756735788,
      "learning_rate": 1.002721792333584e-07,
      "loss": 0.6958,
      "step": 5201
    },
    {
      "epoch": 0.5142475842127375,
      "grad_norm": 4.861821462753415,
      "learning_rate": 1.0024015821274678e-07,
      "loss": 0.754,
      "step": 5202
    },
    {
      "epoch": 0.5143464399574921,
      "grad_norm": 6.612771220690631,
      "learning_rate": 1.0020813716751047e-07,
      "loss": 0.6958,
      "step": 5203
    },
    {
      "epoch": 0.5144452957022465,
      "grad_norm": 8.715288422021418,
      "learning_rate": 1.0017611610093277e-07,
      "loss": 0.7077,
      "step": 5204
    },
    {
      "epoch": 0.5145441514470009,
      "grad_norm": 4.112484682583029,
      "learning_rate": 1.0014409501629697e-07,
      "loss": 0.6087,
      "step": 5205
    },
    {
      "epoch": 0.5146430071917555,
      "grad_norm": 4.520008954717863,
      "learning_rate": 1.0011207391688637e-07,
      "loss": 0.644,
      "step": 5206
    },
    {
      "epoch": 0.5147418629365099,
      "grad_norm": 3.2701410613526014,
      "learning_rate": 1.0008005280598419e-07,
      "loss": 0.6765,
      "step": 5207
    },
    {
      "epoch": 0.5148407186812644,
      "grad_norm": 6.455400598492269,
      "learning_rate": 1.0004803168687382e-07,
      "loss": 0.7496,
      "step": 5208
    },
    {
      "epoch": 0.5149395744260188,
      "grad_norm": 4.809831205254909,
      "learning_rate": 1.000160105628385e-07,
      "loss": 0.8063,
      "step": 5209
    },
    {
      "epoch": 0.5150384301707733,
      "grad_norm": 3.3885573019784947,
      "learning_rate": 9.99839894371615e-08,
      "loss": 0.6612,
      "step": 5210
    },
    {
      "epoch": 0.5151372859155278,
      "grad_norm": 6.7911551954643254,
      "learning_rate": 9.99519683131262e-08,
      "loss": 0.615,
      "step": 5211
    },
    {
      "epoch": 0.5152361416602822,
      "grad_norm": 6.176796459895119,
      "learning_rate": 9.99199471940158e-08,
      "loss": 0.7829,
      "step": 5212
    },
    {
      "epoch": 0.5153349974050367,
      "grad_norm": 4.372200004356536,
      "learning_rate": 9.988792608311365e-08,
      "loss": 0.7035,
      "step": 5213
    },
    {
      "epoch": 0.5154338531497912,
      "grad_norm": 6.069113498501078,
      "learning_rate": 9.985590498370304e-08,
      "loss": 0.709,
      "step": 5214
    },
    {
      "epoch": 0.5155327088945456,
      "grad_norm": 4.298654150422346,
      "learning_rate": 9.982388389906723e-08,
      "loss": 0.6624,
      "step": 5215
    },
    {
      "epoch": 0.5156315646393002,
      "grad_norm": 3.2326061885799406,
      "learning_rate": 9.97918628324895e-08,
      "loss": 0.7043,
      "step": 5216
    },
    {
      "epoch": 0.5157304203840546,
      "grad_norm": 5.146232945356326,
      "learning_rate": 9.975984178725323e-08,
      "loss": 0.7432,
      "step": 5217
    },
    {
      "epoch": 0.515829276128809,
      "grad_norm": 3.864810953749978,
      "learning_rate": 9.972782076664159e-08,
      "loss": 0.7482,
      "step": 5218
    },
    {
      "epoch": 0.5159281318735635,
      "grad_norm": 4.9190639591151895,
      "learning_rate": 9.969579977393792e-08,
      "loss": 0.7433,
      "step": 5219
    },
    {
      "epoch": 0.516026987618318,
      "grad_norm": 3.8908218099868113,
      "learning_rate": 9.966377881242554e-08,
      "loss": 0.7312,
      "step": 5220
    },
    {
      "epoch": 0.5161258433630724,
      "grad_norm": 3.2843813635241794,
      "learning_rate": 9.963175788538764e-08,
      "loss": 0.726,
      "step": 5221
    },
    {
      "epoch": 0.5162246991078269,
      "grad_norm": 11.448661674785312,
      "learning_rate": 9.959973699610757e-08,
      "loss": 0.6648,
      "step": 5222
    },
    {
      "epoch": 0.5163235548525814,
      "grad_norm": 13.606729043089802,
      "learning_rate": 9.956771614786852e-08,
      "loss": 0.8351,
      "step": 5223
    },
    {
      "epoch": 0.5164224105973358,
      "grad_norm": 25.953744511506006,
      "learning_rate": 9.953569534395384e-08,
      "loss": 0.8038,
      "step": 5224
    },
    {
      "epoch": 0.5165212663420903,
      "grad_norm": 3.514073206219885,
      "learning_rate": 9.950367458764677e-08,
      "loss": 0.6976,
      "step": 5225
    },
    {
      "epoch": 0.5166201220868448,
      "grad_norm": 3.601173306142156,
      "learning_rate": 9.94716538822305e-08,
      "loss": 0.7966,
      "step": 5226
    },
    {
      "epoch": 0.5167189778315993,
      "grad_norm": 4.123896239181222,
      "learning_rate": 9.943963323098833e-08,
      "loss": 0.765,
      "step": 5227
    },
    {
      "epoch": 0.5168178335763537,
      "grad_norm": 3.8432940011626617,
      "learning_rate": 9.940761263720357e-08,
      "loss": 0.5997,
      "step": 5228
    },
    {
      "epoch": 0.5169166893211081,
      "grad_norm": 30.719505042538515,
      "learning_rate": 9.937559210415933e-08,
      "loss": 0.6984,
      "step": 5229
    },
    {
      "epoch": 0.5170155450658627,
      "grad_norm": 4.0367666101635225,
      "learning_rate": 9.934357163513895e-08,
      "loss": 0.6536,
      "step": 5230
    },
    {
      "epoch": 0.5171144008106171,
      "grad_norm": 4.176012707811109,
      "learning_rate": 9.93115512334256e-08,
      "loss": 0.6907,
      "step": 5231
    },
    {
      "epoch": 0.5172132565553715,
      "grad_norm": 3.4289638096813775,
      "learning_rate": 9.927953090230252e-08,
      "loss": 0.614,
      "step": 5232
    },
    {
      "epoch": 0.5173121123001261,
      "grad_norm": 4.333031644724551,
      "learning_rate": 9.924751064505292e-08,
      "loss": 0.6901,
      "step": 5233
    },
    {
      "epoch": 0.5174109680448805,
      "grad_norm": 13.055056372965046,
      "learning_rate": 9.921549046496e-08,
      "loss": 0.6229,
      "step": 5234
    },
    {
      "epoch": 0.517509823789635,
      "grad_norm": 5.498379447409048,
      "learning_rate": 9.918347036530695e-08,
      "loss": 0.6551,
      "step": 5235
    },
    {
      "epoch": 0.5176086795343895,
      "grad_norm": 6.089428710759652,
      "learning_rate": 9.9151450349377e-08,
      "loss": 0.782,
      "step": 5236
    },
    {
      "epoch": 0.5177075352791439,
      "grad_norm": 4.303092302017949,
      "learning_rate": 9.911943042045326e-08,
      "loss": 0.7237,
      "step": 5237
    },
    {
      "epoch": 0.5178063910238984,
      "grad_norm": 3.372905109299711,
      "learning_rate": 9.908741058181895e-08,
      "loss": 0.5261,
      "step": 5238
    },
    {
      "epoch": 0.5179052467686528,
      "grad_norm": 5.554612226759997,
      "learning_rate": 9.905539083675724e-08,
      "loss": 0.6823,
      "step": 5239
    },
    {
      "epoch": 0.5180041025134073,
      "grad_norm": 3.174415449306823,
      "learning_rate": 9.902337118855128e-08,
      "loss": 0.7487,
      "step": 5240
    },
    {
      "epoch": 0.5181029582581618,
      "grad_norm": 3.8473932083703644,
      "learning_rate": 9.899135164048414e-08,
      "loss": 0.7666,
      "step": 5241
    },
    {
      "epoch": 0.5182018140029162,
      "grad_norm": 3.6183568430264894,
      "learning_rate": 9.895933219583908e-08,
      "loss": 0.6741,
      "step": 5242
    },
    {
      "epoch": 0.5183006697476708,
      "grad_norm": 4.673341645996973,
      "learning_rate": 9.89273128578991e-08,
      "loss": 0.6127,
      "step": 5243
    },
    {
      "epoch": 0.5183995254924252,
      "grad_norm": 2.951573839659583,
      "learning_rate": 9.88952936299474e-08,
      "loss": 0.7249,
      "step": 5244
    },
    {
      "epoch": 0.5184983812371796,
      "grad_norm": 5.778532143835413,
      "learning_rate": 9.886327451526702e-08,
      "loss": 0.6712,
      "step": 5245
    },
    {
      "epoch": 0.5185972369819342,
      "grad_norm": 3.3155600664817926,
      "learning_rate": 9.883125551714106e-08,
      "loss": 0.7173,
      "step": 5246
    },
    {
      "epoch": 0.5186960927266886,
      "grad_norm": 7.4333772082145035,
      "learning_rate": 9.879923663885265e-08,
      "loss": 0.7007,
      "step": 5247
    },
    {
      "epoch": 0.518794948471443,
      "grad_norm": 11.2828630746386,
      "learning_rate": 9.876721788368474e-08,
      "loss": 0.6314,
      "step": 5248
    },
    {
      "epoch": 0.5188938042161975,
      "grad_norm": 3.412410516030076,
      "learning_rate": 9.87351992549205e-08,
      "loss": 0.801,
      "step": 5249
    },
    {
      "epoch": 0.518992659960952,
      "grad_norm": 3.890896643775121,
      "learning_rate": 9.870318075584293e-08,
      "loss": 0.7304,
      "step": 5250
    },
    {
      "epoch": 0.5190915157057064,
      "grad_norm": 3.1257103693668484,
      "learning_rate": 9.867116238973503e-08,
      "loss": 0.6252,
      "step": 5251
    },
    {
      "epoch": 0.5191903714504609,
      "grad_norm": 21.35795565687115,
      "learning_rate": 9.86391441598798e-08,
      "loss": 0.6884,
      "step": 5252
    },
    {
      "epoch": 0.5192892271952154,
      "grad_norm": 3.1865135867723864,
      "learning_rate": 9.860712606956032e-08,
      "loss": 0.6601,
      "step": 5253
    },
    {
      "epoch": 0.5193880829399699,
      "grad_norm": 5.038067798009723,
      "learning_rate": 9.857510812205946e-08,
      "loss": 0.7356,
      "step": 5254
    },
    {
      "epoch": 0.5194869386847243,
      "grad_norm": 7.09118023591621,
      "learning_rate": 9.854309032066028e-08,
      "loss": 0.7758,
      "step": 5255
    },
    {
      "epoch": 0.5195857944294788,
      "grad_norm": 4.395564396480461,
      "learning_rate": 9.851107266864566e-08,
      "loss": 0.7167,
      "step": 5256
    },
    {
      "epoch": 0.5196846501742333,
      "grad_norm": 4.67886310070672,
      "learning_rate": 9.847905516929861e-08,
      "loss": 0.6517,
      "step": 5257
    },
    {
      "epoch": 0.5197835059189877,
      "grad_norm": 3.150371172776044,
      "learning_rate": 9.8447037825902e-08,
      "loss": 0.703,
      "step": 5258
    },
    {
      "epoch": 0.5198823616637421,
      "grad_norm": 4.065840212229686,
      "learning_rate": 9.841502064173875e-08,
      "loss": 0.5704,
      "step": 5259
    },
    {
      "epoch": 0.5199812174084967,
      "grad_norm": 4.7053417205395025,
      "learning_rate": 9.838300362009175e-08,
      "loss": 0.6417,
      "step": 5260
    },
    {
      "epoch": 0.5200800731532511,
      "grad_norm": 4.811618252193097,
      "learning_rate": 9.83509867642439e-08,
      "loss": 0.7635,
      "step": 5261
    },
    {
      "epoch": 0.5201789288980055,
      "grad_norm": 3.5557017912260878,
      "learning_rate": 9.8318970077478e-08,
      "loss": 0.7389,
      "step": 5262
    },
    {
      "epoch": 0.5202777846427601,
      "grad_norm": 4.852927795474712,
      "learning_rate": 9.828695356307692e-08,
      "loss": 0.6832,
      "step": 5263
    },
    {
      "epoch": 0.5203766403875145,
      "grad_norm": 3.8331197446808134,
      "learning_rate": 9.825493722432353e-08,
      "loss": 0.7171,
      "step": 5264
    },
    {
      "epoch": 0.520475496132269,
      "grad_norm": 3.2829345013563924,
      "learning_rate": 9.822292106450054e-08,
      "loss": 0.6038,
      "step": 5265
    },
    {
      "epoch": 0.5205743518770235,
      "grad_norm": 3.7848737299129644,
      "learning_rate": 9.819090508689083e-08,
      "loss": 0.7253,
      "step": 5266
    },
    {
      "epoch": 0.5206732076217779,
      "grad_norm": 5.299228562629602,
      "learning_rate": 9.81588892947771e-08,
      "loss": 0.7639,
      "step": 5267
    },
    {
      "epoch": 0.5207720633665324,
      "grad_norm": 10.091681620437388,
      "learning_rate": 9.812687369144207e-08,
      "loss": 0.6627,
      "step": 5268
    },
    {
      "epoch": 0.5208709191112868,
      "grad_norm": 4.346398286078482,
      "learning_rate": 9.809485828016857e-08,
      "loss": 0.6428,
      "step": 5269
    },
    {
      "epoch": 0.5209697748560413,
      "grad_norm": 3.1023145932430767,
      "learning_rate": 9.80628430642392e-08,
      "loss": 0.7347,
      "step": 5270
    },
    {
      "epoch": 0.5210686306007958,
      "grad_norm": 3.5102422033259137,
      "learning_rate": 9.803082804693671e-08,
      "loss": 0.7415,
      "step": 5271
    },
    {
      "epoch": 0.5211674863455502,
      "grad_norm": 5.01719246158629,
      "learning_rate": 9.799881323154379e-08,
      "loss": 0.7559,
      "step": 5272
    },
    {
      "epoch": 0.5212663420903048,
      "grad_norm": 3.983272583258675,
      "learning_rate": 9.796679862134302e-08,
      "loss": 0.6663,
      "step": 5273
    },
    {
      "epoch": 0.5213651978350592,
      "grad_norm": 6.311605848936584,
      "learning_rate": 9.793478421961707e-08,
      "loss": 0.6654,
      "step": 5274
    },
    {
      "epoch": 0.5214640535798136,
      "grad_norm": 4.762220336186127,
      "learning_rate": 9.790277002964853e-08,
      "loss": 0.707,
      "step": 5275
    },
    {
      "epoch": 0.5215629093245682,
      "grad_norm": 5.2960363403934645,
      "learning_rate": 9.787075605472e-08,
      "loss": 0.7728,
      "step": 5276
    },
    {
      "epoch": 0.5216617650693226,
      "grad_norm": 13.739321268757857,
      "learning_rate": 9.783874229811402e-08,
      "loss": 0.7873,
      "step": 5277
    },
    {
      "epoch": 0.521760620814077,
      "grad_norm": 9.501204247564678,
      "learning_rate": 9.780672876311316e-08,
      "loss": 0.7855,
      "step": 5278
    },
    {
      "epoch": 0.5218594765588315,
      "grad_norm": 3.5818092975818843,
      "learning_rate": 9.777471545299986e-08,
      "loss": 0.6771,
      "step": 5279
    },
    {
      "epoch": 0.521958332303586,
      "grad_norm": 3.8212807758113945,
      "learning_rate": 9.774270237105671e-08,
      "loss": 0.7305,
      "step": 5280
    },
    {
      "epoch": 0.5220571880483404,
      "grad_norm": 3.6429981389132666,
      "learning_rate": 9.771068952056611e-08,
      "loss": 0.7702,
      "step": 5281
    },
    {
      "epoch": 0.5221560437930949,
      "grad_norm": 7.985525678494868,
      "learning_rate": 9.767867690481052e-08,
      "loss": 0.7942,
      "step": 5282
    },
    {
      "epoch": 0.5222548995378494,
      "grad_norm": 32.33185405477215,
      "learning_rate": 9.764666452707242e-08,
      "loss": 0.8347,
      "step": 5283
    },
    {
      "epoch": 0.5223537552826039,
      "grad_norm": 3.347470362218499,
      "learning_rate": 9.761465239063414e-08,
      "loss": 0.6238,
      "step": 5284
    },
    {
      "epoch": 0.5224526110273583,
      "grad_norm": 3.4686541078437,
      "learning_rate": 9.758264049877804e-08,
      "loss": 0.7496,
      "step": 5285
    },
    {
      "epoch": 0.5225514667721128,
      "grad_norm": 5.162626535625287,
      "learning_rate": 9.755062885478654e-08,
      "loss": 0.6892,
      "step": 5286
    },
    {
      "epoch": 0.5226503225168673,
      "grad_norm": 3.8243612211473073,
      "learning_rate": 9.751861746194187e-08,
      "loss": 0.716,
      "step": 5287
    },
    {
      "epoch": 0.5227491782616217,
      "grad_norm": 4.526422719290918,
      "learning_rate": 9.748660632352645e-08,
      "loss": 0.6562,
      "step": 5288
    },
    {
      "epoch": 0.5228480340063762,
      "grad_norm": 7.206696790624657,
      "learning_rate": 9.74545954428224e-08,
      "loss": 0.6695,
      "step": 5289
    },
    {
      "epoch": 0.5229468897511307,
      "grad_norm": 4.386047485729767,
      "learning_rate": 9.742258482311205e-08,
      "loss": 0.7809,
      "step": 5290
    },
    {
      "epoch": 0.5230457454958851,
      "grad_norm": 3.3481502561278758,
      "learning_rate": 9.739057446767765e-08,
      "loss": 0.7167,
      "step": 5291
    },
    {
      "epoch": 0.5231446012406396,
      "grad_norm": 3.686013103652321,
      "learning_rate": 9.735856437980131e-08,
      "loss": 0.6786,
      "step": 5292
    },
    {
      "epoch": 0.5232434569853941,
      "grad_norm": 9.876831332956012,
      "learning_rate": 9.732655456276525e-08,
      "loss": 0.6447,
      "step": 5293
    },
    {
      "epoch": 0.5233423127301485,
      "grad_norm": 5.773988367260863,
      "learning_rate": 9.729454501985161e-08,
      "loss": 0.6395,
      "step": 5294
    },
    {
      "epoch": 0.523441168474903,
      "grad_norm": 4.647334979150752,
      "learning_rate": 9.72625357543424e-08,
      "loss": 0.772,
      "step": 5295
    },
    {
      "epoch": 0.5235400242196575,
      "grad_norm": 5.804408970521308,
      "learning_rate": 9.72305267695198e-08,
      "loss": 0.7193,
      "step": 5296
    },
    {
      "epoch": 0.5236388799644119,
      "grad_norm": 5.482655754507211,
      "learning_rate": 9.719851806866586e-08,
      "loss": 0.7366,
      "step": 5297
    },
    {
      "epoch": 0.5237377357091664,
      "grad_norm": 4.152915387587098,
      "learning_rate": 9.716650965506252e-08,
      "loss": 0.7396,
      "step": 5298
    },
    {
      "epoch": 0.5238365914539209,
      "grad_norm": 4.221233859165618,
      "learning_rate": 9.713450153199186e-08,
      "loss": 0.6486,
      "step": 5299
    },
    {
      "epoch": 0.5239354471986754,
      "grad_norm": 5.080527180483363,
      "learning_rate": 9.710249370273577e-08,
      "loss": 0.7544,
      "step": 5300
    },
    {
      "epoch": 0.5240343029434298,
      "grad_norm": 3.0696626920100094,
      "learning_rate": 9.707048617057622e-08,
      "loss": 0.6822,
      "step": 5301
    },
    {
      "epoch": 0.5241331586881842,
      "grad_norm": 5.208476985391087,
      "learning_rate": 9.703847893879508e-08,
      "loss": 0.725,
      "step": 5302
    },
    {
      "epoch": 0.5242320144329388,
      "grad_norm": 3.808271160397599,
      "learning_rate": 9.700647201067426e-08,
      "loss": 0.6428,
      "step": 5303
    },
    {
      "epoch": 0.5243308701776932,
      "grad_norm": 4.631353678306793,
      "learning_rate": 9.697446538949557e-08,
      "loss": 0.6309,
      "step": 5304
    },
    {
      "epoch": 0.5244297259224476,
      "grad_norm": 2.9671363513437994,
      "learning_rate": 9.694245907854085e-08,
      "loss": 0.6725,
      "step": 5305
    },
    {
      "epoch": 0.5245285816672022,
      "grad_norm": 3.5522472313030002,
      "learning_rate": 9.69104530810918e-08,
      "loss": 0.7153,
      "step": 5306
    },
    {
      "epoch": 0.5246274374119566,
      "grad_norm": 3.7366827826358993,
      "learning_rate": 9.687844740043023e-08,
      "loss": 0.6499,
      "step": 5307
    },
    {
      "epoch": 0.524726293156711,
      "grad_norm": 10.523144344247834,
      "learning_rate": 9.684644203983786e-08,
      "loss": 0.8197,
      "step": 5308
    },
    {
      "epoch": 0.5248251489014656,
      "grad_norm": 3.816403970613758,
      "learning_rate": 9.681443700259631e-08,
      "loss": 0.6736,
      "step": 5309
    },
    {
      "epoch": 0.52492400464622,
      "grad_norm": 4.069107452649294,
      "learning_rate": 9.67824322919873e-08,
      "loss": 0.7402,
      "step": 5310
    },
    {
      "epoch": 0.5250228603909745,
      "grad_norm": 3.56526041250723,
      "learning_rate": 9.675042791129236e-08,
      "loss": 0.7701,
      "step": 5311
    },
    {
      "epoch": 0.5251217161357289,
      "grad_norm": 3.3363081306379314,
      "learning_rate": 9.671842386379309e-08,
      "loss": 0.659,
      "step": 5312
    },
    {
      "epoch": 0.5252205718804834,
      "grad_norm": 7.488569722620966,
      "learning_rate": 9.668642015277109e-08,
      "loss": 0.7315,
      "step": 5313
    },
    {
      "epoch": 0.5253194276252379,
      "grad_norm": 3.6500107143571987,
      "learning_rate": 9.665441678150778e-08,
      "loss": 0.737,
      "step": 5314
    },
    {
      "epoch": 0.5254182833699923,
      "grad_norm": 3.056116706714036,
      "learning_rate": 9.662241375328467e-08,
      "loss": 0.6721,
      "step": 5315
    },
    {
      "epoch": 0.5255171391147468,
      "grad_norm": 3.5935793310225765,
      "learning_rate": 9.659041107138325e-08,
      "loss": 0.6698,
      "step": 5316
    },
    {
      "epoch": 0.5256159948595013,
      "grad_norm": 3.279032468034479,
      "learning_rate": 9.655840873908485e-08,
      "loss": 0.7844,
      "step": 5317
    },
    {
      "epoch": 0.5257148506042557,
      "grad_norm": 2.932466634038758,
      "learning_rate": 9.652640675967089e-08,
      "loss": 0.786,
      "step": 5318
    },
    {
      "epoch": 0.5258137063490103,
      "grad_norm": 4.398059859207706,
      "learning_rate": 9.649440513642268e-08,
      "loss": 0.6293,
      "step": 5319
    },
    {
      "epoch": 0.5259125620937647,
      "grad_norm": 4.238561609352924,
      "learning_rate": 9.64624038726215e-08,
      "loss": 0.6738,
      "step": 5320
    },
    {
      "epoch": 0.5260114178385191,
      "grad_norm": 2.6896932666634266,
      "learning_rate": 9.643040297154864e-08,
      "loss": 0.7658,
      "step": 5321
    },
    {
      "epoch": 0.5261102735832736,
      "grad_norm": 2.9443598200672048,
      "learning_rate": 9.639840243648527e-08,
      "loss": 0.6289,
      "step": 5322
    },
    {
      "epoch": 0.5262091293280281,
      "grad_norm": 2.8673342998474127,
      "learning_rate": 9.636640227071257e-08,
      "loss": 0.7656,
      "step": 5323
    },
    {
      "epoch": 0.5263079850727825,
      "grad_norm": 4.451367431285134,
      "learning_rate": 9.633440247751179e-08,
      "loss": 0.6611,
      "step": 5324
    },
    {
      "epoch": 0.526406840817537,
      "grad_norm": 6.502169529009352,
      "learning_rate": 9.630240306016392e-08,
      "loss": 0.6667,
      "step": 5325
    },
    {
      "epoch": 0.5265056965622915,
      "grad_norm": 2.971161455676223,
      "learning_rate": 9.627040402195008e-08,
      "loss": 0.645,
      "step": 5326
    },
    {
      "epoch": 0.526604552307046,
      "grad_norm": 13.147015131273006,
      "learning_rate": 9.62384053661513e-08,
      "loss": 0.747,
      "step": 5327
    },
    {
      "epoch": 0.5267034080518004,
      "grad_norm": 6.820267516720407,
      "learning_rate": 9.620640709604858e-08,
      "loss": 0.7424,
      "step": 5328
    },
    {
      "epoch": 0.5268022637965549,
      "grad_norm": 3.6567919476206927,
      "learning_rate": 9.617440921492282e-08,
      "loss": 0.7003,
      "step": 5329
    },
    {
      "epoch": 0.5269011195413094,
      "grad_norm": 3.6609980853752484,
      "learning_rate": 9.6142411726055e-08,
      "loss": 0.7127,
      "step": 5330
    },
    {
      "epoch": 0.5269999752860638,
      "grad_norm": 9.665047500114843,
      "learning_rate": 9.611041463272591e-08,
      "loss": 0.6106,
      "step": 5331
    },
    {
      "epoch": 0.5270988310308182,
      "grad_norm": 3.6079352534226716,
      "learning_rate": 9.607841793821647e-08,
      "loss": 0.6217,
      "step": 5332
    },
    {
      "epoch": 0.5271976867755728,
      "grad_norm": 6.495209991188831,
      "learning_rate": 9.604642164580738e-08,
      "loss": 0.7479,
      "step": 5333
    },
    {
      "epoch": 0.5272965425203272,
      "grad_norm": 4.488740246574695,
      "learning_rate": 9.601442575877943e-08,
      "loss": 0.7018,
      "step": 5334
    },
    {
      "epoch": 0.5273953982650816,
      "grad_norm": 3.192376243232391,
      "learning_rate": 9.598243028041337e-08,
      "loss": 0.6869,
      "step": 5335
    },
    {
      "epoch": 0.5274942540098362,
      "grad_norm": 3.87773430843339,
      "learning_rate": 9.595043521398978e-08,
      "loss": 0.6924,
      "step": 5336
    },
    {
      "epoch": 0.5275931097545906,
      "grad_norm": 3.523983480639423,
      "learning_rate": 9.591844056278934e-08,
      "loss": 0.6664,
      "step": 5337
    },
    {
      "epoch": 0.527691965499345,
      "grad_norm": 4.091210219873373,
      "learning_rate": 9.588644633009265e-08,
      "loss": 0.7058,
      "step": 5338
    },
    {
      "epoch": 0.5277908212440996,
      "grad_norm": 5.088197778296859,
      "learning_rate": 9.585445251918015e-08,
      "loss": 0.6311,
      "step": 5339
    },
    {
      "epoch": 0.527889676988854,
      "grad_norm": 5.5225793835332215,
      "learning_rate": 9.58224591333324e-08,
      "loss": 0.7114,
      "step": 5340
    },
    {
      "epoch": 0.5279885327336085,
      "grad_norm": 3.396246553776375,
      "learning_rate": 9.579046617582989e-08,
      "loss": 0.7227,
      "step": 5341
    },
    {
      "epoch": 0.5280873884783629,
      "grad_norm": 3.400920482041242,
      "learning_rate": 9.575847364995293e-08,
      "loss": 0.7585,
      "step": 5342
    },
    {
      "epoch": 0.5281862442231174,
      "grad_norm": 3.828138069714206,
      "learning_rate": 9.572648155898198e-08,
      "loss": 0.7595,
      "step": 5343
    },
    {
      "epoch": 0.5282850999678719,
      "grad_norm": 3.490815359043999,
      "learning_rate": 9.569448990619726e-08,
      "loss": 0.7322,
      "step": 5344
    },
    {
      "epoch": 0.5283839557126263,
      "grad_norm": 8.799359519922534,
      "learning_rate": 9.56624986948791e-08,
      "loss": 0.7737,
      "step": 5345
    },
    {
      "epoch": 0.5284828114573809,
      "grad_norm": 2.683491622840961,
      "learning_rate": 9.563050792830772e-08,
      "loss": 0.6959,
      "step": 5346
    },
    {
      "epoch": 0.5285816672021353,
      "grad_norm": 5.416065061226504,
      "learning_rate": 9.559851760976331e-08,
      "loss": 0.672,
      "step": 5347
    },
    {
      "epoch": 0.5286805229468897,
      "grad_norm": 3.9718374698671264,
      "learning_rate": 9.556652774252596e-08,
      "loss": 0.7409,
      "step": 5348
    },
    {
      "epoch": 0.5287793786916443,
      "grad_norm": 2.9909945049759994,
      "learning_rate": 9.553453832987584e-08,
      "loss": 0.7111,
      "step": 5349
    },
    {
      "epoch": 0.5288782344363987,
      "grad_norm": 3.9146887178079526,
      "learning_rate": 9.55025493750929e-08,
      "loss": 0.7451,
      "step": 5350
    },
    {
      "epoch": 0.5289770901811531,
      "grad_norm": 3.8022655853677447,
      "learning_rate": 9.547056088145717e-08,
      "loss": 0.6056,
      "step": 5351
    },
    {
      "epoch": 0.5290759459259076,
      "grad_norm": 12.85163268573417,
      "learning_rate": 9.543857285224867e-08,
      "loss": 0.6606,
      "step": 5352
    },
    {
      "epoch": 0.5291748016706621,
      "grad_norm": 3.3205172076614917,
      "learning_rate": 9.540658529074717e-08,
      "loss": 0.7463,
      "step": 5353
    },
    {
      "epoch": 0.5292736574154165,
      "grad_norm": 3.806466081490698,
      "learning_rate": 9.537459820023266e-08,
      "loss": 0.7112,
      "step": 5354
    },
    {
      "epoch": 0.529372513160171,
      "grad_norm": 3.629168844336948,
      "learning_rate": 9.534261158398484e-08,
      "loss": 0.7028,
      "step": 5355
    },
    {
      "epoch": 0.5294713689049255,
      "grad_norm": 10.64505156612048,
      "learning_rate": 9.531062544528349e-08,
      "loss": 0.6611,
      "step": 5356
    },
    {
      "epoch": 0.52957022464968,
      "grad_norm": 3.9796107803835605,
      "learning_rate": 9.527863978740836e-08,
      "loss": 0.6597,
      "step": 5357
    },
    {
      "epoch": 0.5296690803944344,
      "grad_norm": 7.0406201582893075,
      "learning_rate": 9.524665461363904e-08,
      "loss": 0.6836,
      "step": 5358
    },
    {
      "epoch": 0.5297679361391889,
      "grad_norm": 4.166047574829191,
      "learning_rate": 9.521466992725516e-08,
      "loss": 0.6966,
      "step": 5359
    },
    {
      "epoch": 0.5298667918839434,
      "grad_norm": 3.5387040508142804,
      "learning_rate": 9.518268573153633e-08,
      "loss": 0.6876,
      "step": 5360
    },
    {
      "epoch": 0.5299656476286978,
      "grad_norm": 4.384376731333225,
      "learning_rate": 9.515070202976199e-08,
      "loss": 0.9271,
      "step": 5361
    },
    {
      "epoch": 0.5300645033734523,
      "grad_norm": 6.250510697432791,
      "learning_rate": 9.511871882521165e-08,
      "loss": 0.717,
      "step": 5362
    },
    {
      "epoch": 0.5301633591182068,
      "grad_norm": 4.170966517180185,
      "learning_rate": 9.508673612116468e-08,
      "loss": 0.6382,
      "step": 5363
    },
    {
      "epoch": 0.5302622148629612,
      "grad_norm": 4.51495853240118,
      "learning_rate": 9.505475392090044e-08,
      "loss": 0.5835,
      "step": 5364
    },
    {
      "epoch": 0.5303610706077156,
      "grad_norm": 6.691991298913472,
      "learning_rate": 9.502277222769827e-08,
      "loss": 0.7499,
      "step": 5365
    },
    {
      "epoch": 0.5304599263524702,
      "grad_norm": 13.227388407564307,
      "learning_rate": 9.499079104483733e-08,
      "loss": 0.6764,
      "step": 5366
    },
    {
      "epoch": 0.5305587820972246,
      "grad_norm": 3.166354705811398,
      "learning_rate": 9.49588103755969e-08,
      "loss": 0.7161,
      "step": 5367
    },
    {
      "epoch": 0.5306576378419791,
      "grad_norm": 4.999643311901486,
      "learning_rate": 9.492683022325613e-08,
      "loss": 0.723,
      "step": 5368
    },
    {
      "epoch": 0.5307564935867336,
      "grad_norm": 3.3255525415319993,
      "learning_rate": 9.489485059109406e-08,
      "loss": 0.7341,
      "step": 5369
    },
    {
      "epoch": 0.530855349331488,
      "grad_norm": 6.821401561891237,
      "learning_rate": 9.486287148238974e-08,
      "loss": 0.724,
      "step": 5370
    },
    {
      "epoch": 0.5309542050762425,
      "grad_norm": 3.877730353188762,
      "learning_rate": 9.48308929004222e-08,
      "loss": 0.787,
      "step": 5371
    },
    {
      "epoch": 0.531053060820997,
      "grad_norm": 5.531131715160006,
      "learning_rate": 9.479891484847036e-08,
      "loss": 0.6524,
      "step": 5372
    },
    {
      "epoch": 0.5311519165657514,
      "grad_norm": 5.4062351054504,
      "learning_rate": 9.476693732981303e-08,
      "loss": 0.7314,
      "step": 5373
    },
    {
      "epoch": 0.5312507723105059,
      "grad_norm": 6.544957477039126,
      "learning_rate": 9.473496034772915e-08,
      "loss": 0.6593,
      "step": 5374
    },
    {
      "epoch": 0.5313496280552603,
      "grad_norm": 4.549033763162557,
      "learning_rate": 9.470298390549736e-08,
      "loss": 0.7478,
      "step": 5375
    },
    {
      "epoch": 0.5314484838000149,
      "grad_norm": 3.276611138826865,
      "learning_rate": 9.467100800639648e-08,
      "loss": 0.6976,
      "step": 5376
    },
    {
      "epoch": 0.5315473395447693,
      "grad_norm": 24.242688782387553,
      "learning_rate": 9.463903265370508e-08,
      "loss": 0.6755,
      "step": 5377
    },
    {
      "epoch": 0.5316461952895237,
      "grad_norm": 5.967936369543118,
      "learning_rate": 9.46070578507018e-08,
      "loss": 0.7069,
      "step": 5378
    },
    {
      "epoch": 0.5317450510342783,
      "grad_norm": 4.629084668988543,
      "learning_rate": 9.457508360066523e-08,
      "loss": 0.7378,
      "step": 5379
    },
    {
      "epoch": 0.5318439067790327,
      "grad_norm": 4.667877970336129,
      "learning_rate": 9.454310990687376e-08,
      "loss": 0.7371,
      "step": 5380
    },
    {
      "epoch": 0.5319427625237871,
      "grad_norm": 5.647855461205118,
      "learning_rate": 9.451113677260592e-08,
      "loss": 0.6785,
      "step": 5381
    },
    {
      "epoch": 0.5320416182685417,
      "grad_norm": 4.824479466797406,
      "learning_rate": 9.447916420114004e-08,
      "loss": 0.8157,
      "step": 5382
    },
    {
      "epoch": 0.5321404740132961,
      "grad_norm": 3.4861476964037412,
      "learning_rate": 9.44471921957544e-08,
      "loss": 0.7541,
      "step": 5383
    },
    {
      "epoch": 0.5322393297580506,
      "grad_norm": 3.820436937681865,
      "learning_rate": 9.441522075972728e-08,
      "loss": 0.7067,
      "step": 5384
    },
    {
      "epoch": 0.532338185502805,
      "grad_norm": 2.8852400013278325,
      "learning_rate": 9.438324989633693e-08,
      "loss": 0.8021,
      "step": 5385
    },
    {
      "epoch": 0.5324370412475595,
      "grad_norm": 4.405682455622722,
      "learning_rate": 9.435127960886142e-08,
      "loss": 0.791,
      "step": 5386
    },
    {
      "epoch": 0.532535896992314,
      "grad_norm": 3.755991563757569,
      "learning_rate": 9.43193099005789e-08,
      "loss": 0.7159,
      "step": 5387
    },
    {
      "epoch": 0.5326347527370684,
      "grad_norm": 44.06448313599135,
      "learning_rate": 9.428734077476732e-08,
      "loss": 0.7759,
      "step": 5388
    },
    {
      "epoch": 0.5327336084818229,
      "grad_norm": 6.108174380713787,
      "learning_rate": 9.425537223470468e-08,
      "loss": 0.6942,
      "step": 5389
    },
    {
      "epoch": 0.5328324642265774,
      "grad_norm": 4.336660140179777,
      "learning_rate": 9.422340428366889e-08,
      "loss": 0.7215,
      "step": 5390
    },
    {
      "epoch": 0.5329313199713318,
      "grad_norm": 122.95643173275614,
      "learning_rate": 9.419143692493779e-08,
      "loss": 0.7261,
      "step": 5391
    },
    {
      "epoch": 0.5330301757160864,
      "grad_norm": 3.477675484616408,
      "learning_rate": 9.415947016178912e-08,
      "loss": 0.7731,
      "step": 5392
    },
    {
      "epoch": 0.5331290314608408,
      "grad_norm": 6.479670018244905,
      "learning_rate": 9.412750399750065e-08,
      "loss": 0.7065,
      "step": 5393
    },
    {
      "epoch": 0.5332278872055952,
      "grad_norm": 5.24269137261282,
      "learning_rate": 9.409553843535e-08,
      "loss": 0.752,
      "step": 5394
    },
    {
      "epoch": 0.5333267429503497,
      "grad_norm": 18.366372080502995,
      "learning_rate": 9.406357347861479e-08,
      "loss": 0.7551,
      "step": 5395
    },
    {
      "epoch": 0.5334255986951042,
      "grad_norm": 10.86944982050237,
      "learning_rate": 9.403160913057259e-08,
      "loss": 0.7366,
      "step": 5396
    },
    {
      "epoch": 0.5335244544398586,
      "grad_norm": 3.221393003541163,
      "learning_rate": 9.399964539450078e-08,
      "loss": 0.7223,
      "step": 5397
    },
    {
      "epoch": 0.5336233101846131,
      "grad_norm": 3.7905873330595337,
      "learning_rate": 9.396768227367688e-08,
      "loss": 0.8052,
      "step": 5398
    },
    {
      "epoch": 0.5337221659293676,
      "grad_norm": 3.5299465325264157,
      "learning_rate": 9.393571977137815e-08,
      "loss": 0.6819,
      "step": 5399
    },
    {
      "epoch": 0.533821021674122,
      "grad_norm": 3.045356334802583,
      "learning_rate": 9.39037578908819e-08,
      "loss": 0.7397,
      "step": 5400
    },
    {
      "epoch": 0.5339198774188765,
      "grad_norm": 4.1317070515040974,
      "learning_rate": 9.387179663546537e-08,
      "loss": 0.6599,
      "step": 5401
    },
    {
      "epoch": 0.534018733163631,
      "grad_norm": 3.635136931127015,
      "learning_rate": 9.383983600840568e-08,
      "loss": 0.7313,
      "step": 5402
    },
    {
      "epoch": 0.5341175889083855,
      "grad_norm": 2.9470844160740346,
      "learning_rate": 9.380787601297991e-08,
      "loss": 0.7115,
      "step": 5403
    },
    {
      "epoch": 0.5342164446531399,
      "grad_norm": 6.4388871085024135,
      "learning_rate": 9.377591665246517e-08,
      "loss": 0.719,
      "step": 5404
    },
    {
      "epoch": 0.5343153003978943,
      "grad_norm": 3.7349838835342273,
      "learning_rate": 9.374395793013833e-08,
      "loss": 0.6901,
      "step": 5405
    },
    {
      "epoch": 0.5344141561426489,
      "grad_norm": 3.465854549560405,
      "learning_rate": 9.371199984927634e-08,
      "loss": 0.6989,
      "step": 5406
    },
    {
      "epoch": 0.5345130118874033,
      "grad_norm": 6.287125722855138,
      "learning_rate": 9.3680042413156e-08,
      "loss": 0.6392,
      "step": 5407
    },
    {
      "epoch": 0.5346118676321577,
      "grad_norm": 3.799814547584937,
      "learning_rate": 9.364808562505409e-08,
      "loss": 0.8046,
      "step": 5408
    },
    {
      "epoch": 0.5347107233769123,
      "grad_norm": 7.259538871202132,
      "learning_rate": 9.361612948824733e-08,
      "loss": 0.7523,
      "step": 5409
    },
    {
      "epoch": 0.5348095791216667,
      "grad_norm": 3.4053315275678164,
      "learning_rate": 9.358417400601227e-08,
      "loss": 0.8123,
      "step": 5410
    },
    {
      "epoch": 0.5349084348664211,
      "grad_norm": 3.9450130142257303,
      "learning_rate": 9.355221918162552e-08,
      "loss": 0.7188,
      "step": 5411
    },
    {
      "epoch": 0.5350072906111757,
      "grad_norm": 5.60294983752288,
      "learning_rate": 9.352026501836364e-08,
      "loss": 0.7964,
      "step": 5412
    },
    {
      "epoch": 0.5351061463559301,
      "grad_norm": 9.129685295911456,
      "learning_rate": 9.348831151950296e-08,
      "loss": 0.6908,
      "step": 5413
    },
    {
      "epoch": 0.5352050021006846,
      "grad_norm": 3.461508842941914,
      "learning_rate": 9.345635868831987e-08,
      "loss": 0.6712,
      "step": 5414
    },
    {
      "epoch": 0.535303857845439,
      "grad_norm": 11.427661546302343,
      "learning_rate": 9.342440652809072e-08,
      "loss": 0.7372,
      "step": 5415
    },
    {
      "epoch": 0.5354027135901935,
      "grad_norm": 6.558547358740465,
      "learning_rate": 9.339245504209165e-08,
      "loss": 0.7489,
      "step": 5416
    },
    {
      "epoch": 0.535501569334948,
      "grad_norm": 3.4982618005043267,
      "learning_rate": 9.336050423359882e-08,
      "loss": 0.6131,
      "step": 5417
    },
    {
      "epoch": 0.5356004250797024,
      "grad_norm": 4.141737327688224,
      "learning_rate": 9.332855410588841e-08,
      "loss": 0.6798,
      "step": 5418
    },
    {
      "epoch": 0.535699280824457,
      "grad_norm": 15.541883029369082,
      "learning_rate": 9.329660466223631e-08,
      "loss": 0.6631,
      "step": 5419
    },
    {
      "epoch": 0.5357981365692114,
      "grad_norm": 2.7629508110540577,
      "learning_rate": 9.326465590591856e-08,
      "loss": 0.6797,
      "step": 5420
    },
    {
      "epoch": 0.5358969923139658,
      "grad_norm": 6.015384766635216,
      "learning_rate": 9.323270784021095e-08,
      "loss": 0.6836,
      "step": 5421
    },
    {
      "epoch": 0.5359958480587204,
      "grad_norm": 3.77724641498081,
      "learning_rate": 9.320076046838933e-08,
      "loss": 0.6129,
      "step": 5422
    },
    {
      "epoch": 0.5360947038034748,
      "grad_norm": 3.92501865516041,
      "learning_rate": 9.316881379372947e-08,
      "loss": 0.6538,
      "step": 5423
    },
    {
      "epoch": 0.5361935595482292,
      "grad_norm": 24.14731406779913,
      "learning_rate": 9.313686781950695e-08,
      "loss": 0.8144,
      "step": 5424
    },
    {
      "epoch": 0.5362924152929837,
      "grad_norm": 4.640915148908676,
      "learning_rate": 9.310492254899741e-08,
      "loss": 0.7327,
      "step": 5425
    },
    {
      "epoch": 0.5363912710377382,
      "grad_norm": 3.7917096698758592,
      "learning_rate": 9.307297798547639e-08,
      "loss": 0.6747,
      "step": 5426
    },
    {
      "epoch": 0.5364901267824926,
      "grad_norm": 7.3582079381203025,
      "learning_rate": 9.304103413221923e-08,
      "loss": 0.7534,
      "step": 5427
    },
    {
      "epoch": 0.5365889825272471,
      "grad_norm": 5.352069519658265,
      "learning_rate": 9.300909099250138e-08,
      "loss": 0.6636,
      "step": 5428
    },
    {
      "epoch": 0.5366878382720016,
      "grad_norm": 3.4019803496724914,
      "learning_rate": 9.297714856959818e-08,
      "loss": 0.619,
      "step": 5429
    },
    {
      "epoch": 0.536786694016756,
      "grad_norm": 6.646057306170499,
      "learning_rate": 9.294520686678475e-08,
      "loss": 0.7193,
      "step": 5430
    },
    {
      "epoch": 0.5368855497615105,
      "grad_norm": 4.84817283366918,
      "learning_rate": 9.291326588733632e-08,
      "loss": 0.6845,
      "step": 5431
    },
    {
      "epoch": 0.536984405506265,
      "grad_norm": 4.195807621039848,
      "learning_rate": 9.28813256345279e-08,
      "loss": 0.6944,
      "step": 5432
    },
    {
      "epoch": 0.5370832612510195,
      "grad_norm": 3.7680529711326063,
      "learning_rate": 9.284938611163456e-08,
      "loss": 0.8095,
      "step": 5433
    },
    {
      "epoch": 0.5371821169957739,
      "grad_norm": 3.0677126365669762,
      "learning_rate": 9.281744732193119e-08,
      "loss": 0.8169,
      "step": 5434
    },
    {
      "epoch": 0.5372809727405283,
      "grad_norm": 4.979035118053513,
      "learning_rate": 9.278550926869264e-08,
      "loss": 0.7716,
      "step": 5435
    },
    {
      "epoch": 0.5373798284852829,
      "grad_norm": 4.3320888027646145,
      "learning_rate": 9.275357195519367e-08,
      "loss": 0.7856,
      "step": 5436
    },
    {
      "epoch": 0.5374786842300373,
      "grad_norm": 2.906186518219438,
      "learning_rate": 9.272163538470905e-08,
      "loss": 0.7361,
      "step": 5437
    },
    {
      "epoch": 0.5375775399747917,
      "grad_norm": 5.425923335394132,
      "learning_rate": 9.268969956051332e-08,
      "loss": 0.7046,
      "step": 5438
    },
    {
      "epoch": 0.5376763957195463,
      "grad_norm": 3.7786532735895935,
      "learning_rate": 9.265776448588107e-08,
      "loss": 0.6129,
      "step": 5439
    },
    {
      "epoch": 0.5377752514643007,
      "grad_norm": 6.146759373051676,
      "learning_rate": 9.26258301640868e-08,
      "loss": 0.5892,
      "step": 5440
    },
    {
      "epoch": 0.5378741072090552,
      "grad_norm": 6.016780851816374,
      "learning_rate": 9.259389659840483e-08,
      "loss": 0.7555,
      "step": 5441
    },
    {
      "epoch": 0.5379729629538097,
      "grad_norm": 4.369671957464884,
      "learning_rate": 9.256196379210956e-08,
      "loss": 0.6824,
      "step": 5442
    },
    {
      "epoch": 0.5380718186985641,
      "grad_norm": 4.180072298792183,
      "learning_rate": 9.253003174847517e-08,
      "loss": 0.7196,
      "step": 5443
    },
    {
      "epoch": 0.5381706744433186,
      "grad_norm": 3.172646302371047,
      "learning_rate": 9.24981004707758e-08,
      "loss": 0.7795,
      "step": 5444
    },
    {
      "epoch": 0.5382695301880731,
      "grad_norm": 3.8121429345468942,
      "learning_rate": 9.246616996228562e-08,
      "loss": 0.7335,
      "step": 5445
    },
    {
      "epoch": 0.5383683859328275,
      "grad_norm": 5.814330052843449,
      "learning_rate": 9.243424022627853e-08,
      "loss": 0.682,
      "step": 5446
    },
    {
      "epoch": 0.538467241677582,
      "grad_norm": 3.135967788104405,
      "learning_rate": 9.240231126602848e-08,
      "loss": 0.7809,
      "step": 5447
    },
    {
      "epoch": 0.5385660974223364,
      "grad_norm": 3.831745689131576,
      "learning_rate": 9.237038308480941e-08,
      "loss": 0.6559,
      "step": 5448
    },
    {
      "epoch": 0.538664953167091,
      "grad_norm": 3.379669930861098,
      "learning_rate": 9.233845568589496e-08,
      "loss": 0.6808,
      "step": 5449
    },
    {
      "epoch": 0.5387638089118454,
      "grad_norm": 5.838964225784685,
      "learning_rate": 9.230652907255889e-08,
      "loss": 0.7132,
      "step": 5450
    },
    {
      "epoch": 0.5388626646565998,
      "grad_norm": 10.812413360891448,
      "learning_rate": 9.227460324807476e-08,
      "loss": 0.713,
      "step": 5451
    },
    {
      "epoch": 0.5389615204013544,
      "grad_norm": 4.4399873177177165,
      "learning_rate": 9.224267821571611e-08,
      "loss": 0.6447,
      "step": 5452
    },
    {
      "epoch": 0.5390603761461088,
      "grad_norm": 10.238154447067211,
      "learning_rate": 9.221075397875641e-08,
      "loss": 0.8021,
      "step": 5453
    },
    {
      "epoch": 0.5391592318908632,
      "grad_norm": 11.71938049989512,
      "learning_rate": 9.217883054046896e-08,
      "loss": 0.6848,
      "step": 5454
    },
    {
      "epoch": 0.5392580876356178,
      "grad_norm": 3.4484475280824403,
      "learning_rate": 9.214690790412706e-08,
      "loss": 0.6501,
      "step": 5455
    },
    {
      "epoch": 0.5393569433803722,
      "grad_norm": 3.4700852686455685,
      "learning_rate": 9.211498607300395e-08,
      "loss": 0.7701,
      "step": 5456
    },
    {
      "epoch": 0.5394557991251266,
      "grad_norm": 3.877856330953515,
      "learning_rate": 9.208306505037267e-08,
      "loss": 0.6394,
      "step": 5457
    },
    {
      "epoch": 0.5395546548698811,
      "grad_norm": 3.8812918534125274,
      "learning_rate": 9.205114483950627e-08,
      "loss": 0.7444,
      "step": 5458
    },
    {
      "epoch": 0.5396535106146356,
      "grad_norm": 4.722495655866669,
      "learning_rate": 9.201922544367777e-08,
      "loss": 0.6711,
      "step": 5459
    },
    {
      "epoch": 0.5397523663593901,
      "grad_norm": 5.468362658950029,
      "learning_rate": 9.198730686615996e-08,
      "loss": 0.6901,
      "step": 5460
    },
    {
      "epoch": 0.5398512221041445,
      "grad_norm": 4.140087006798514,
      "learning_rate": 9.195538911022559e-08,
      "loss": 0.71,
      "step": 5461
    },
    {
      "epoch": 0.539950077848899,
      "grad_norm": 4.9502626964559395,
      "learning_rate": 9.192347217914747e-08,
      "loss": 0.7086,
      "step": 5462
    },
    {
      "epoch": 0.5400489335936535,
      "grad_norm": 28.13863413661896,
      "learning_rate": 9.189155607619807e-08,
      "loss": 0.6439,
      "step": 5463
    },
    {
      "epoch": 0.5401477893384079,
      "grad_norm": 3.5687479489289764,
      "learning_rate": 9.185964080465005e-08,
      "loss": 0.6022,
      "step": 5464
    },
    {
      "epoch": 0.5402466450831624,
      "grad_norm": 4.469302045759572,
      "learning_rate": 9.182772636777574e-08,
      "loss": 0.8433,
      "step": 5465
    },
    {
      "epoch": 0.5403455008279169,
      "grad_norm": 3.6061423738737255,
      "learning_rate": 9.179581276884753e-08,
      "loss": 0.6161,
      "step": 5466
    },
    {
      "epoch": 0.5404443565726713,
      "grad_norm": 4.217293793177038,
      "learning_rate": 9.176390001113776e-08,
      "loss": 0.7618,
      "step": 5467
    },
    {
      "epoch": 0.5405432123174257,
      "grad_norm": 3.4447405742853854,
      "learning_rate": 9.173198809791852e-08,
      "loss": 0.6941,
      "step": 5468
    },
    {
      "epoch": 0.5406420680621803,
      "grad_norm": 3.984047793844125,
      "learning_rate": 9.170007703246194e-08,
      "loss": 0.7831,
      "step": 5469
    },
    {
      "epoch": 0.5407409238069347,
      "grad_norm": 3.625239199551596,
      "learning_rate": 9.166816681804005e-08,
      "loss": 0.6621,
      "step": 5470
    },
    {
      "epoch": 0.5408397795516892,
      "grad_norm": 3.9580308841676186,
      "learning_rate": 9.16362574579247e-08,
      "loss": 0.7031,
      "step": 5471
    },
    {
      "epoch": 0.5409386352964437,
      "grad_norm": 77.87778527285352,
      "learning_rate": 9.16043489553878e-08,
      "loss": 0.6442,
      "step": 5472
    },
    {
      "epoch": 0.5410374910411981,
      "grad_norm": 6.147148953454198,
      "learning_rate": 9.157244131370109e-08,
      "loss": 0.7195,
      "step": 5473
    },
    {
      "epoch": 0.5411363467859526,
      "grad_norm": 3.387613009112442,
      "learning_rate": 9.154053453613618e-08,
      "loss": 0.6763,
      "step": 5474
    },
    {
      "epoch": 0.5412352025307071,
      "grad_norm": 10.765708384847265,
      "learning_rate": 9.15086286259647e-08,
      "loss": 0.7472,
      "step": 5475
    },
    {
      "epoch": 0.5413340582754615,
      "grad_norm": 11.90628827032786,
      "learning_rate": 9.147672358645805e-08,
      "loss": 0.5985,
      "step": 5476
    },
    {
      "epoch": 0.541432914020216,
      "grad_norm": 17.701121105499567,
      "learning_rate": 9.14448194208877e-08,
      "loss": 0.614,
      "step": 5477
    },
    {
      "epoch": 0.5415317697649704,
      "grad_norm": 5.5408404872502866,
      "learning_rate": 9.141291613252492e-08,
      "loss": 0.6497,
      "step": 5478
    },
    {
      "epoch": 0.541630625509725,
      "grad_norm": 6.456424132930539,
      "learning_rate": 9.138101372464092e-08,
      "loss": 0.7164,
      "step": 5479
    },
    {
      "epoch": 0.5417294812544794,
      "grad_norm": 2.9932764938827376,
      "learning_rate": 9.13491122005068e-08,
      "loss": 0.7791,
      "step": 5480
    },
    {
      "epoch": 0.5418283369992338,
      "grad_norm": 3.4287656798764634,
      "learning_rate": 9.131721156339366e-08,
      "loss": 0.6824,
      "step": 5481
    },
    {
      "epoch": 0.5419271927439884,
      "grad_norm": 3.7522588328017665,
      "learning_rate": 9.128531181657235e-08,
      "loss": 0.8041,
      "step": 5482
    },
    {
      "epoch": 0.5420260484887428,
      "grad_norm": 3.785335328560544,
      "learning_rate": 9.125341296331376e-08,
      "loss": 0.7039,
      "step": 5483
    },
    {
      "epoch": 0.5421249042334972,
      "grad_norm": 3.3990959331531987,
      "learning_rate": 9.12215150068887e-08,
      "loss": 0.6522,
      "step": 5484
    },
    {
      "epoch": 0.5422237599782518,
      "grad_norm": 3.561372601686652,
      "learning_rate": 9.118961795056773e-08,
      "loss": 0.6945,
      "step": 5485
    },
    {
      "epoch": 0.5423226157230062,
      "grad_norm": 4.370453426565204,
      "learning_rate": 9.115772179762151e-08,
      "loss": 0.6528,
      "step": 5486
    },
    {
      "epoch": 0.5424214714677607,
      "grad_norm": 4.1211165041085085,
      "learning_rate": 9.11258265513205e-08,
      "loss": 0.7766,
      "step": 5487
    },
    {
      "epoch": 0.5425203272125151,
      "grad_norm": 4.395433877439904,
      "learning_rate": 9.109393221493505e-08,
      "loss": 0.6875,
      "step": 5488
    },
    {
      "epoch": 0.5426191829572696,
      "grad_norm": 12.252475542161267,
      "learning_rate": 9.106203879173549e-08,
      "loss": 0.7579,
      "step": 5489
    },
    {
      "epoch": 0.5427180387020241,
      "grad_norm": 4.291697436872045,
      "learning_rate": 9.103014628499198e-08,
      "loss": 0.5939,
      "step": 5490
    },
    {
      "epoch": 0.5428168944467785,
      "grad_norm": 3.929743092804351,
      "learning_rate": 9.099825469797466e-08,
      "loss": 0.7201,
      "step": 5491
    },
    {
      "epoch": 0.542915750191533,
      "grad_norm": 6.218055650145727,
      "learning_rate": 9.096636403395357e-08,
      "loss": 0.5193,
      "step": 5492
    },
    {
      "epoch": 0.5430146059362875,
      "grad_norm": 11.65905241281121,
      "learning_rate": 9.093447429619857e-08,
      "loss": 0.7929,
      "step": 5493
    },
    {
      "epoch": 0.5431134616810419,
      "grad_norm": 2.9171501520951364,
      "learning_rate": 9.090258548797949e-08,
      "loss": 0.793,
      "step": 5494
    },
    {
      "epoch": 0.5432123174257965,
      "grad_norm": 4.024531347082412,
      "learning_rate": 9.08706976125661e-08,
      "loss": 0.6122,
      "step": 5495
    },
    {
      "epoch": 0.5433111731705509,
      "grad_norm": 4.500235280890116,
      "learning_rate": 9.083881067322799e-08,
      "loss": 0.7868,
      "step": 5496
    },
    {
      "epoch": 0.5434100289153053,
      "grad_norm": 3.451843386523784,
      "learning_rate": 9.080692467323468e-08,
      "loss": 0.7211,
      "step": 5497
    },
    {
      "epoch": 0.5435088846600598,
      "grad_norm": 4.258854199488862,
      "learning_rate": 9.07750396158557e-08,
      "loss": 0.6672,
      "step": 5498
    },
    {
      "epoch": 0.5436077404048143,
      "grad_norm": 6.527749091116677,
      "learning_rate": 9.074315550436028e-08,
      "loss": 0.6746,
      "step": 5499
    },
    {
      "epoch": 0.5437065961495687,
      "grad_norm": 5.213404965435421,
      "learning_rate": 9.071127234201775e-08,
      "loss": 0.7045,
      "step": 5500
    },
    {
      "epoch": 0.5438054518943232,
      "grad_norm": 3.3700677049826706,
      "learning_rate": 9.067939013209718e-08,
      "loss": 0.5861,
      "step": 5501
    },
    {
      "epoch": 0.5439043076390777,
      "grad_norm": 4.0810181874671745,
      "learning_rate": 9.064750887786767e-08,
      "loss": 0.5966,
      "step": 5502
    },
    {
      "epoch": 0.5440031633838321,
      "grad_norm": 6.726793792058383,
      "learning_rate": 9.061562858259822e-08,
      "loss": 0.6797,
      "step": 5503
    },
    {
      "epoch": 0.5441020191285866,
      "grad_norm": 3.717451241545791,
      "learning_rate": 9.05837492495576e-08,
      "loss": 0.8343,
      "step": 5504
    },
    {
      "epoch": 0.5442008748733411,
      "grad_norm": 4.697313225424097,
      "learning_rate": 9.055187088201456e-08,
      "loss": 0.7736,
      "step": 5505
    },
    {
      "epoch": 0.5442997306180956,
      "grad_norm": 4.770680363697563,
      "learning_rate": 9.051999348323785e-08,
      "loss": 0.7161,
      "step": 5506
    },
    {
      "epoch": 0.54439858636285,
      "grad_norm": 5.429225776648173,
      "learning_rate": 9.048811705649592e-08,
      "loss": 0.6712,
      "step": 5507
    },
    {
      "epoch": 0.5444974421076044,
      "grad_norm": 3.2051581223127035,
      "learning_rate": 9.045624160505727e-08,
      "loss": 0.6766,
      "step": 5508
    },
    {
      "epoch": 0.544596297852359,
      "grad_norm": 4.510738233401222,
      "learning_rate": 9.04243671321903e-08,
      "loss": 0.6538,
      "step": 5509
    },
    {
      "epoch": 0.5446951535971134,
      "grad_norm": 6.197542378045775,
      "learning_rate": 9.03924936411632e-08,
      "loss": 0.6705,
      "step": 5510
    },
    {
      "epoch": 0.5447940093418678,
      "grad_norm": 3.7752484665053005,
      "learning_rate": 9.036062113524421e-08,
      "loss": 0.6498,
      "step": 5511
    },
    {
      "epoch": 0.5448928650866224,
      "grad_norm": 4.497341250314914,
      "learning_rate": 9.032874961770127e-08,
      "loss": 0.6511,
      "step": 5512
    },
    {
      "epoch": 0.5449917208313768,
      "grad_norm": 4.064564426114251,
      "learning_rate": 9.029687909180243e-08,
      "loss": 0.699,
      "step": 5513
    },
    {
      "epoch": 0.5450905765761312,
      "grad_norm": 4.754041374336784,
      "learning_rate": 9.026500956081553e-08,
      "loss": 0.6613,
      "step": 5514
    },
    {
      "epoch": 0.5451894323208858,
      "grad_norm": 3.489472220578984,
      "learning_rate": 9.023314102800824e-08,
      "loss": 0.6853,
      "step": 5515
    },
    {
      "epoch": 0.5452882880656402,
      "grad_norm": 7.891031451872883,
      "learning_rate": 9.020127349664827e-08,
      "loss": 0.6302,
      "step": 5516
    },
    {
      "epoch": 0.5453871438103947,
      "grad_norm": 7.625379786793269,
      "learning_rate": 9.01694069700032e-08,
      "loss": 0.7476,
      "step": 5517
    },
    {
      "epoch": 0.5454859995551492,
      "grad_norm": 3.107728661132936,
      "learning_rate": 9.01375414513404e-08,
      "loss": 0.6054,
      "step": 5518
    },
    {
      "epoch": 0.5455848552999036,
      "grad_norm": 3.562197078015887,
      "learning_rate": 9.010567694392725e-08,
      "loss": 0.7907,
      "step": 5519
    },
    {
      "epoch": 0.5456837110446581,
      "grad_norm": 5.872470714060215,
      "learning_rate": 9.007381345103102e-08,
      "loss": 0.7386,
      "step": 5520
    },
    {
      "epoch": 0.5457825667894125,
      "grad_norm": 3.7441498914686613,
      "learning_rate": 9.004195097591876e-08,
      "loss": 0.8025,
      "step": 5521
    },
    {
      "epoch": 0.545881422534167,
      "grad_norm": 6.930792501001246,
      "learning_rate": 9.001008952185754e-08,
      "loss": 0.7006,
      "step": 5522
    },
    {
      "epoch": 0.5459802782789215,
      "grad_norm": 4.022299129630744,
      "learning_rate": 8.997822909211429e-08,
      "loss": 0.6925,
      "step": 5523
    },
    {
      "epoch": 0.5460791340236759,
      "grad_norm": 3.5553617555203125,
      "learning_rate": 8.994636968995579e-08,
      "loss": 0.7307,
      "step": 5524
    },
    {
      "epoch": 0.5461779897684305,
      "grad_norm": 4.561391687189453,
      "learning_rate": 8.991451131864883e-08,
      "loss": 0.801,
      "step": 5525
    },
    {
      "epoch": 0.5462768455131849,
      "grad_norm": 3.364673516426641,
      "learning_rate": 8.98826539814599e-08,
      "loss": 0.6042,
      "step": 5526
    },
    {
      "epoch": 0.5463757012579393,
      "grad_norm": 10.16617145769568,
      "learning_rate": 8.985079768165557e-08,
      "loss": 0.7708,
      "step": 5527
    },
    {
      "epoch": 0.5464745570026939,
      "grad_norm": 4.269358608322347,
      "learning_rate": 8.981894242250226e-08,
      "loss": 0.7701,
      "step": 5528
    },
    {
      "epoch": 0.5465734127474483,
      "grad_norm": 4.359602381881918,
      "learning_rate": 8.978708820726619e-08,
      "loss": 0.6446,
      "step": 5529
    },
    {
      "epoch": 0.5466722684922027,
      "grad_norm": 10.462931881144845,
      "learning_rate": 8.97552350392136e-08,
      "loss": 0.7056,
      "step": 5530
    },
    {
      "epoch": 0.5467711242369572,
      "grad_norm": 6.047998358609549,
      "learning_rate": 8.972338292161056e-08,
      "loss": 0.7066,
      "step": 5531
    },
    {
      "epoch": 0.5468699799817117,
      "grad_norm": 4.826188245403201,
      "learning_rate": 8.969153185772295e-08,
      "loss": 0.7207,
      "step": 5532
    },
    {
      "epoch": 0.5469688357264662,
      "grad_norm": 4.095565506216744,
      "learning_rate": 8.965968185081675e-08,
      "loss": 0.7136,
      "step": 5533
    },
    {
      "epoch": 0.5470676914712206,
      "grad_norm": 4.02221765645325,
      "learning_rate": 8.962783290415759e-08,
      "loss": 0.6305,
      "step": 5534
    },
    {
      "epoch": 0.5471665472159751,
      "grad_norm": 7.824484349227778,
      "learning_rate": 8.959598502101117e-08,
      "loss": 0.7115,
      "step": 5535
    },
    {
      "epoch": 0.5472654029607296,
      "grad_norm": 4.256196260529801,
      "learning_rate": 8.956413820464306e-08,
      "loss": 0.782,
      "step": 5536
    },
    {
      "epoch": 0.547364258705484,
      "grad_norm": 5.898916078685272,
      "learning_rate": 8.95322924583186e-08,
      "loss": 0.6099,
      "step": 5537
    },
    {
      "epoch": 0.5474631144502385,
      "grad_norm": 3.6753305870723922,
      "learning_rate": 8.950044778530315e-08,
      "loss": 0.6328,
      "step": 5538
    },
    {
      "epoch": 0.547561970194993,
      "grad_norm": 4.015635708726488,
      "learning_rate": 8.94686041888619e-08,
      "loss": 0.7294,
      "step": 5539
    },
    {
      "epoch": 0.5476608259397474,
      "grad_norm": 5.091307193355761,
      "learning_rate": 8.943676167225994e-08,
      "loss": 0.6792,
      "step": 5540
    },
    {
      "epoch": 0.5477596816845018,
      "grad_norm": 4.520374372522441,
      "learning_rate": 8.940492023876223e-08,
      "loss": 0.7164,
      "step": 5541
    },
    {
      "epoch": 0.5478585374292564,
      "grad_norm": 3.320176358254382,
      "learning_rate": 8.93730798916337e-08,
      "loss": 0.6684,
      "step": 5542
    },
    {
      "epoch": 0.5479573931740108,
      "grad_norm": 2.9887273669938574,
      "learning_rate": 8.934124063413905e-08,
      "loss": 0.6916,
      "step": 5543
    },
    {
      "epoch": 0.5480562489187653,
      "grad_norm": 7.2280716962757126,
      "learning_rate": 8.930940246954297e-08,
      "loss": 0.689,
      "step": 5544
    },
    {
      "epoch": 0.5481551046635198,
      "grad_norm": 5.390700883764027,
      "learning_rate": 8.927756540110993e-08,
      "loss": 0.7536,
      "step": 5545
    },
    {
      "epoch": 0.5482539604082742,
      "grad_norm": 3.4384781030555343,
      "learning_rate": 8.924572943210438e-08,
      "loss": 0.7548,
      "step": 5546
    },
    {
      "epoch": 0.5483528161530287,
      "grad_norm": 4.7911688066834675,
      "learning_rate": 8.92138945657907e-08,
      "loss": 0.6943,
      "step": 5547
    },
    {
      "epoch": 0.5484516718977832,
      "grad_norm": 3.883674084984668,
      "learning_rate": 8.9182060805433e-08,
      "loss": 0.6815,
      "step": 5548
    },
    {
      "epoch": 0.5485505276425376,
      "grad_norm": 4.04316114344802,
      "learning_rate": 8.915022815429536e-08,
      "loss": 0.6381,
      "step": 5549
    },
    {
      "epoch": 0.5486493833872921,
      "grad_norm": 3.380245677679543,
      "learning_rate": 8.911839661564183e-08,
      "loss": 0.677,
      "step": 5550
    },
    {
      "epoch": 0.5487482391320465,
      "grad_norm": 9.0081302799177,
      "learning_rate": 8.908656619273616e-08,
      "loss": 0.6371,
      "step": 5551
    },
    {
      "epoch": 0.5488470948768011,
      "grad_norm": 3.678860272685736,
      "learning_rate": 8.905473688884216e-08,
      "loss": 0.7552,
      "step": 5552
    },
    {
      "epoch": 0.5489459506215555,
      "grad_norm": 3.0605748564990107,
      "learning_rate": 8.902290870722346e-08,
      "loss": 0.7265,
      "step": 5553
    },
    {
      "epoch": 0.5490448063663099,
      "grad_norm": 5.365106057964961,
      "learning_rate": 8.899108165114351e-08,
      "loss": 0.6148,
      "step": 5554
    },
    {
      "epoch": 0.5491436621110645,
      "grad_norm": 4.145758757317412,
      "learning_rate": 8.895925572386579e-08,
      "loss": 0.7616,
      "step": 5555
    },
    {
      "epoch": 0.5492425178558189,
      "grad_norm": 4.936720118853012,
      "learning_rate": 8.892743092865349e-08,
      "loss": 0.6904,
      "step": 5556
    },
    {
      "epoch": 0.5493413736005733,
      "grad_norm": 8.430388842468727,
      "learning_rate": 8.889560726876985e-08,
      "loss": 0.7691,
      "step": 5557
    },
    {
      "epoch": 0.5494402293453279,
      "grad_norm": 10.548933633059514,
      "learning_rate": 8.886378474747789e-08,
      "loss": 0.5702,
      "step": 5558
    },
    {
      "epoch": 0.5495390850900823,
      "grad_norm": 5.319103864985203,
      "learning_rate": 8.883196336804048e-08,
      "loss": 0.7897,
      "step": 5559
    },
    {
      "epoch": 0.5496379408348367,
      "grad_norm": 4.704144139448874,
      "learning_rate": 8.880014313372051e-08,
      "loss": 0.6232,
      "step": 5560
    },
    {
      "epoch": 0.5497367965795912,
      "grad_norm": 4.2014624318402225,
      "learning_rate": 8.876832404778068e-08,
      "loss": 0.715,
      "step": 5561
    },
    {
      "epoch": 0.5498356523243457,
      "grad_norm": 4.022355809340892,
      "learning_rate": 8.87365061134835e-08,
      "loss": 0.7218,
      "step": 5562
    },
    {
      "epoch": 0.5499345080691002,
      "grad_norm": 4.022430124720475,
      "learning_rate": 8.870468933409148e-08,
      "loss": 0.6734,
      "step": 5563
    },
    {
      "epoch": 0.5500333638138546,
      "grad_norm": 12.845749564657876,
      "learning_rate": 8.867287371286698e-08,
      "loss": 0.7575,
      "step": 5564
    },
    {
      "epoch": 0.5501322195586091,
      "grad_norm": 4.482773976108219,
      "learning_rate": 8.864105925307218e-08,
      "loss": 0.6879,
      "step": 5565
    },
    {
      "epoch": 0.5502310753033636,
      "grad_norm": 3.3647568782221255,
      "learning_rate": 8.86092459579692e-08,
      "loss": 0.7875,
      "step": 5566
    },
    {
      "epoch": 0.550329931048118,
      "grad_norm": 4.405467519958067,
      "learning_rate": 8.857743383082002e-08,
      "loss": 0.6725,
      "step": 5567
    },
    {
      "epoch": 0.5504287867928725,
      "grad_norm": 3.5587018102698527,
      "learning_rate": 8.854562287488647e-08,
      "loss": 0.6429,
      "step": 5568
    },
    {
      "epoch": 0.550527642537627,
      "grad_norm": 9.293468479140465,
      "learning_rate": 8.851381309343039e-08,
      "loss": 0.6273,
      "step": 5569
    },
    {
      "epoch": 0.5506264982823814,
      "grad_norm": 3.6197614265710714,
      "learning_rate": 8.84820044897133e-08,
      "loss": 0.7144,
      "step": 5570
    },
    {
      "epoch": 0.5507253540271358,
      "grad_norm": 13.700183084632403,
      "learning_rate": 8.845019706699675e-08,
      "loss": 0.73,
      "step": 5571
    },
    {
      "epoch": 0.5508242097718904,
      "grad_norm": 4.024269740711445,
      "learning_rate": 8.841839082854214e-08,
      "loss": 0.6517,
      "step": 5572
    },
    {
      "epoch": 0.5509230655166448,
      "grad_norm": 4.036703423631276,
      "learning_rate": 8.83865857776107e-08,
      "loss": 0.734,
      "step": 5573
    },
    {
      "epoch": 0.5510219212613993,
      "grad_norm": 3.7322799783295735,
      "learning_rate": 8.835478191746358e-08,
      "loss": 0.6152,
      "step": 5574
    },
    {
      "epoch": 0.5511207770061538,
      "grad_norm": 5.093130662247495,
      "learning_rate": 8.832297925136181e-08,
      "loss": 0.8138,
      "step": 5575
    },
    {
      "epoch": 0.5512196327509082,
      "grad_norm": 4.07743722182365,
      "learning_rate": 8.829117778256624e-08,
      "loss": 0.6701,
      "step": 5576
    },
    {
      "epoch": 0.5513184884956627,
      "grad_norm": 3.9983810804591964,
      "learning_rate": 8.82593775143377e-08,
      "loss": 0.7474,
      "step": 5577
    },
    {
      "epoch": 0.5514173442404172,
      "grad_norm": 3.689715046446761,
      "learning_rate": 8.822757844993677e-08,
      "loss": 0.7203,
      "step": 5578
    },
    {
      "epoch": 0.5515161999851717,
      "grad_norm": 16.619696254095807,
      "learning_rate": 8.819578059262403e-08,
      "loss": 0.7204,
      "step": 5579
    },
    {
      "epoch": 0.5516150557299261,
      "grad_norm": 11.475064704466122,
      "learning_rate": 8.81639839456599e-08,
      "loss": 0.6753,
      "step": 5580
    },
    {
      "epoch": 0.5517139114746805,
      "grad_norm": 4.429756993004492,
      "learning_rate": 8.813218851230459e-08,
      "loss": 0.7233,
      "step": 5581
    },
    {
      "epoch": 0.5518127672194351,
      "grad_norm": 11.026386988486845,
      "learning_rate": 8.810039429581829e-08,
      "loss": 0.7162,
      "step": 5582
    },
    {
      "epoch": 0.5519116229641895,
      "grad_norm": 5.766134192222263,
      "learning_rate": 8.806860129946104e-08,
      "loss": 0.7425,
      "step": 5583
    },
    {
      "epoch": 0.5520104787089439,
      "grad_norm": 5.229851152812705,
      "learning_rate": 8.803680952649271e-08,
      "loss": 0.6962,
      "step": 5584
    },
    {
      "epoch": 0.5521093344536985,
      "grad_norm": 3.325035193411012,
      "learning_rate": 8.800501898017308e-08,
      "loss": 0.6467,
      "step": 5585
    },
    {
      "epoch": 0.5522081901984529,
      "grad_norm": 3.5516167228520223,
      "learning_rate": 8.797322966376184e-08,
      "loss": 0.7307,
      "step": 5586
    },
    {
      "epoch": 0.5523070459432073,
      "grad_norm": 4.160764597801965,
      "learning_rate": 8.794144158051848e-08,
      "loss": 0.6707,
      "step": 5587
    },
    {
      "epoch": 0.5524059016879619,
      "grad_norm": 4.0513193892289365,
      "learning_rate": 8.790965473370243e-08,
      "loss": 0.6727,
      "step": 5588
    },
    {
      "epoch": 0.5525047574327163,
      "grad_norm": 5.078600711242933,
      "learning_rate": 8.787786912657292e-08,
      "loss": 0.7337,
      "step": 5589
    },
    {
      "epoch": 0.5526036131774708,
      "grad_norm": 7.488991274585162,
      "learning_rate": 8.78460847623891e-08,
      "loss": 0.7325,
      "step": 5590
    },
    {
      "epoch": 0.5527024689222252,
      "grad_norm": 5.001535437570028,
      "learning_rate": 8.781430164441004e-08,
      "loss": 0.6524,
      "step": 5591
    },
    {
      "epoch": 0.5528013246669797,
      "grad_norm": 4.075585777545015,
      "learning_rate": 8.77825197758946e-08,
      "loss": 0.6832,
      "step": 5592
    },
    {
      "epoch": 0.5529001804117342,
      "grad_norm": 3.915397765113006,
      "learning_rate": 8.775073916010148e-08,
      "loss": 0.6551,
      "step": 5593
    },
    {
      "epoch": 0.5529990361564886,
      "grad_norm": 5.247256671257444,
      "learning_rate": 8.771895980028944e-08,
      "loss": 0.8286,
      "step": 5594
    },
    {
      "epoch": 0.5530978919012431,
      "grad_norm": 2.8912981759923704,
      "learning_rate": 8.768718169971686e-08,
      "loss": 0.6929,
      "step": 5595
    },
    {
      "epoch": 0.5531967476459976,
      "grad_norm": 4.713692562528132,
      "learning_rate": 8.765540486164217e-08,
      "loss": 0.7746,
      "step": 5596
    },
    {
      "epoch": 0.553295603390752,
      "grad_norm": 5.517764174430145,
      "learning_rate": 8.762362928932366e-08,
      "loss": 0.7283,
      "step": 5597
    },
    {
      "epoch": 0.5533944591355066,
      "grad_norm": 3.9426679883651867,
      "learning_rate": 8.759185498601936e-08,
      "loss": 0.8107,
      "step": 5598
    },
    {
      "epoch": 0.553493314880261,
      "grad_norm": 3.339600969238521,
      "learning_rate": 8.756008195498734e-08,
      "loss": 0.6555,
      "step": 5599
    },
    {
      "epoch": 0.5535921706250154,
      "grad_norm": 3.681021496188008,
      "learning_rate": 8.752831019948538e-08,
      "loss": 0.7137,
      "step": 5600
    },
    {
      "epoch": 0.55369102636977,
      "grad_norm": 4.035826165918055,
      "learning_rate": 8.749653972277126e-08,
      "loss": 0.6833,
      "step": 5601
    },
    {
      "epoch": 0.5537898821145244,
      "grad_norm": 2.8805914260434404,
      "learning_rate": 8.746477052810257e-08,
      "loss": 0.719,
      "step": 5602
    },
    {
      "epoch": 0.5538887378592788,
      "grad_norm": 10.526796026594154,
      "learning_rate": 8.743300261873669e-08,
      "loss": 0.6611,
      "step": 5603
    },
    {
      "epoch": 0.5539875936040333,
      "grad_norm": 5.205519395114601,
      "learning_rate": 8.740123599793104e-08,
      "loss": 0.7559,
      "step": 5604
    },
    {
      "epoch": 0.5540864493487878,
      "grad_norm": 4.776950210953872,
      "learning_rate": 8.736947066894283e-08,
      "loss": 0.6861,
      "step": 5605
    },
    {
      "epoch": 0.5541853050935422,
      "grad_norm": 40.714080449565955,
      "learning_rate": 8.733770663502905e-08,
      "loss": 0.6564,
      "step": 5606
    },
    {
      "epoch": 0.5542841608382967,
      "grad_norm": 4.9489954288138325,
      "learning_rate": 8.730594389944669e-08,
      "loss": 0.6733,
      "step": 5607
    },
    {
      "epoch": 0.5543830165830512,
      "grad_norm": 5.882334733277827,
      "learning_rate": 8.727418246545252e-08,
      "loss": 0.6653,
      "step": 5608
    },
    {
      "epoch": 0.5544818723278057,
      "grad_norm": 8.268548555348994,
      "learning_rate": 8.724242233630325e-08,
      "loss": 0.7213,
      "step": 5609
    },
    {
      "epoch": 0.5545807280725601,
      "grad_norm": 3.159330222643178,
      "learning_rate": 8.721066351525536e-08,
      "loss": 0.6389,
      "step": 5610
    },
    {
      "epoch": 0.5546795838173146,
      "grad_norm": 4.09652386713119,
      "learning_rate": 8.717890600556528e-08,
      "loss": 0.6606,
      "step": 5611
    },
    {
      "epoch": 0.5547784395620691,
      "grad_norm": 2.9026272103396407,
      "learning_rate": 8.714714981048922e-08,
      "loss": 0.5965,
      "step": 5612
    },
    {
      "epoch": 0.5548772953068235,
      "grad_norm": 3.2673117426611644,
      "learning_rate": 8.711539493328342e-08,
      "loss": 0.6459,
      "step": 5613
    },
    {
      "epoch": 0.5549761510515779,
      "grad_norm": 4.044456456768989,
      "learning_rate": 8.708364137720376e-08,
      "loss": 0.6917,
      "step": 5614
    },
    {
      "epoch": 0.5550750067963325,
      "grad_norm": 3.8506962865232865,
      "learning_rate": 8.705188914550616e-08,
      "loss": 0.6235,
      "step": 5615
    },
    {
      "epoch": 0.5551738625410869,
      "grad_norm": 3.686930830953817,
      "learning_rate": 8.702013824144636e-08,
      "loss": 0.7465,
      "step": 5616
    },
    {
      "epoch": 0.5552727182858413,
      "grad_norm": 5.660576549895734,
      "learning_rate": 8.698838866827987e-08,
      "loss": 0.7145,
      "step": 5617
    },
    {
      "epoch": 0.5553715740305959,
      "grad_norm": 5.497481043501608,
      "learning_rate": 8.695664042926222e-08,
      "loss": 0.6305,
      "step": 5618
    },
    {
      "epoch": 0.5554704297753503,
      "grad_norm": 2.996520461814899,
      "learning_rate": 8.692489352764871e-08,
      "loss": 0.6817,
      "step": 5619
    },
    {
      "epoch": 0.5555692855201048,
      "grad_norm": 3.1167523001211666,
      "learning_rate": 8.689314796669444e-08,
      "loss": 0.7615,
      "step": 5620
    },
    {
      "epoch": 0.5556681412648593,
      "grad_norm": 3.794812792088693,
      "learning_rate": 8.686140374965456e-08,
      "loss": 0.6658,
      "step": 5621
    },
    {
      "epoch": 0.5557669970096137,
      "grad_norm": 4.892222622837848,
      "learning_rate": 8.682966087978387e-08,
      "loss": 0.7006,
      "step": 5622
    },
    {
      "epoch": 0.5558658527543682,
      "grad_norm": 3.7294096728945525,
      "learning_rate": 8.679791936033719e-08,
      "loss": 0.6517,
      "step": 5623
    },
    {
      "epoch": 0.5559647084991226,
      "grad_norm": 9.319425853681834,
      "learning_rate": 8.676617919456917e-08,
      "loss": 0.7767,
      "step": 5624
    },
    {
      "epoch": 0.5560635642438772,
      "grad_norm": 3.7417975030093347,
      "learning_rate": 8.673444038573422e-08,
      "loss": 0.6976,
      "step": 5625
    },
    {
      "epoch": 0.5561624199886316,
      "grad_norm": 3.8820702178797655,
      "learning_rate": 8.670270293708675e-08,
      "loss": 0.6207,
      "step": 5626
    },
    {
      "epoch": 0.556261275733386,
      "grad_norm": 4.5674981519659745,
      "learning_rate": 8.667096685188095e-08,
      "loss": 0.6688,
      "step": 5627
    },
    {
      "epoch": 0.5563601314781406,
      "grad_norm": 2.682461763165966,
      "learning_rate": 8.663923213337088e-08,
      "loss": 0.7809,
      "step": 5628
    },
    {
      "epoch": 0.556458987222895,
      "grad_norm": 17.547632814610964,
      "learning_rate": 8.660749878481045e-08,
      "loss": 0.7059,
      "step": 5629
    },
    {
      "epoch": 0.5565578429676494,
      "grad_norm": 9.21040078190634,
      "learning_rate": 8.65757668094535e-08,
      "loss": 0.7484,
      "step": 5630
    },
    {
      "epoch": 0.556656698712404,
      "grad_norm": 3.081651563735899,
      "learning_rate": 8.654403621055359e-08,
      "loss": 0.729,
      "step": 5631
    },
    {
      "epoch": 0.5567555544571584,
      "grad_norm": 9.043719464648992,
      "learning_rate": 8.651230699136434e-08,
      "loss": 0.6884,
      "step": 5632
    },
    {
      "epoch": 0.5568544102019128,
      "grad_norm": 3.5039319159445577,
      "learning_rate": 8.6480579155139e-08,
      "loss": 0.7293,
      "step": 5633
    },
    {
      "epoch": 0.5569532659466673,
      "grad_norm": 5.382759753518157,
      "learning_rate": 8.644885270513084e-08,
      "loss": 0.6456,
      "step": 5634
    },
    {
      "epoch": 0.5570521216914218,
      "grad_norm": 3.542371790811809,
      "learning_rate": 8.6417127644593e-08,
      "loss": 0.6355,
      "step": 5635
    },
    {
      "epoch": 0.5571509774361763,
      "grad_norm": 3.074498441481339,
      "learning_rate": 8.638540397677831e-08,
      "loss": 0.6593,
      "step": 5636
    },
    {
      "epoch": 0.5572498331809307,
      "grad_norm": 5.912201559428354,
      "learning_rate": 8.635368170493962e-08,
      "loss": 0.682,
      "step": 5637
    },
    {
      "epoch": 0.5573486889256852,
      "grad_norm": 4.7916428868133165,
      "learning_rate": 8.63219608323296e-08,
      "loss": 0.7132,
      "step": 5638
    },
    {
      "epoch": 0.5574475446704397,
      "grad_norm": 4.282830867407417,
      "learning_rate": 8.629024136220069e-08,
      "loss": 0.7362,
      "step": 5639
    },
    {
      "epoch": 0.5575464004151941,
      "grad_norm": 3.7155949810912245,
      "learning_rate": 8.625852329780529e-08,
      "loss": 0.7362,
      "step": 5640
    },
    {
      "epoch": 0.5576452561599486,
      "grad_norm": 6.522620965312519,
      "learning_rate": 8.622680664239568e-08,
      "loss": 0.7041,
      "step": 5641
    },
    {
      "epoch": 0.5577441119047031,
      "grad_norm": 3.1046785464144238,
      "learning_rate": 8.619509139922383e-08,
      "loss": 0.7153,
      "step": 5642
    },
    {
      "epoch": 0.5578429676494575,
      "grad_norm": 4.477754923460198,
      "learning_rate": 8.616337757154177e-08,
      "loss": 0.7843,
      "step": 5643
    },
    {
      "epoch": 0.5579418233942119,
      "grad_norm": 4.294438785967592,
      "learning_rate": 8.613166516260119e-08,
      "loss": 0.5985,
      "step": 5644
    },
    {
      "epoch": 0.5580406791389665,
      "grad_norm": 2.783312081115992,
      "learning_rate": 8.609995417565379e-08,
      "loss": 0.7085,
      "step": 5645
    },
    {
      "epoch": 0.5581395348837209,
      "grad_norm": 3.4675851284163444,
      "learning_rate": 8.606824461395108e-08,
      "loss": 0.7965,
      "step": 5646
    },
    {
      "epoch": 0.5582383906284754,
      "grad_norm": 7.998306474484964,
      "learning_rate": 8.603653648074432e-08,
      "loss": 0.6634,
      "step": 5647
    },
    {
      "epoch": 0.5583372463732299,
      "grad_norm": 3.948815768332269,
      "learning_rate": 8.600482977928479e-08,
      "loss": 0.7197,
      "step": 5648
    },
    {
      "epoch": 0.5584361021179843,
      "grad_norm": 5.056535199583191,
      "learning_rate": 8.597312451282355e-08,
      "loss": 0.6733,
      "step": 5649
    },
    {
      "epoch": 0.5585349578627388,
      "grad_norm": 3.4944567647291787,
      "learning_rate": 8.594142068461143e-08,
      "loss": 0.7506,
      "step": 5650
    },
    {
      "epoch": 0.5586338136074933,
      "grad_norm": 4.201141738962651,
      "learning_rate": 8.590971829789928e-08,
      "loss": 0.6408,
      "step": 5651
    },
    {
      "epoch": 0.5587326693522477,
      "grad_norm": 6.6069574961053545,
      "learning_rate": 8.587801735593766e-08,
      "loss": 0.7145,
      "step": 5652
    },
    {
      "epoch": 0.5588315250970022,
      "grad_norm": 6.730359927484027,
      "learning_rate": 8.584631786197704e-08,
      "loss": 0.6921,
      "step": 5653
    },
    {
      "epoch": 0.5589303808417566,
      "grad_norm": 4.192976022981044,
      "learning_rate": 8.581461981926776e-08,
      "loss": 0.681,
      "step": 5654
    },
    {
      "epoch": 0.5590292365865112,
      "grad_norm": 3.993661664637694,
      "learning_rate": 8.578292323105997e-08,
      "loss": 0.6607,
      "step": 5655
    },
    {
      "epoch": 0.5591280923312656,
      "grad_norm": 6.485782892393973,
      "learning_rate": 8.575122810060364e-08,
      "loss": 0.6805,
      "step": 5656
    },
    {
      "epoch": 0.55922694807602,
      "grad_norm": 7.279940174237071,
      "learning_rate": 8.571953443114875e-08,
      "loss": 0.7109,
      "step": 5657
    },
    {
      "epoch": 0.5593258038207746,
      "grad_norm": 3.6874575802399803,
      "learning_rate": 8.56878422259449e-08,
      "loss": 0.6594,
      "step": 5658
    },
    {
      "epoch": 0.559424659565529,
      "grad_norm": 4.284744610763881,
      "learning_rate": 8.565615148824172e-08,
      "loss": 0.7074,
      "step": 5659
    },
    {
      "epoch": 0.5595235153102834,
      "grad_norm": 2.981343561633971,
      "learning_rate": 8.562446222128865e-08,
      "loss": 0.7233,
      "step": 5660
    },
    {
      "epoch": 0.559622371055038,
      "grad_norm": 6.281605853069698,
      "learning_rate": 8.559277442833488e-08,
      "loss": 0.6764,
      "step": 5661
    },
    {
      "epoch": 0.5597212267997924,
      "grad_norm": 4.253184774471703,
      "learning_rate": 8.556108811262961e-08,
      "loss": 0.7146,
      "step": 5662
    },
    {
      "epoch": 0.5598200825445468,
      "grad_norm": 3.782897517519226,
      "learning_rate": 8.552940327742177e-08,
      "loss": 0.6333,
      "step": 5663
    },
    {
      "epoch": 0.5599189382893013,
      "grad_norm": 7.478670004556808,
      "learning_rate": 8.549771992596013e-08,
      "loss": 0.5912,
      "step": 5664
    },
    {
      "epoch": 0.5600177940340558,
      "grad_norm": 7.153195481384384,
      "learning_rate": 8.546603806149343e-08,
      "loss": 0.7423,
      "step": 5665
    },
    {
      "epoch": 0.5601166497788103,
      "grad_norm": 5.870151004078986,
      "learning_rate": 8.543435768727009e-08,
      "loss": 0.7131,
      "step": 5666
    },
    {
      "epoch": 0.5602155055235647,
      "grad_norm": 3.753953592829209,
      "learning_rate": 8.540267880653851e-08,
      "loss": 0.6095,
      "step": 5667
    },
    {
      "epoch": 0.5603143612683192,
      "grad_norm": 3.5666826515263876,
      "learning_rate": 8.537100142254695e-08,
      "loss": 0.7677,
      "step": 5668
    },
    {
      "epoch": 0.5604132170130737,
      "grad_norm": 10.378312931503434,
      "learning_rate": 8.533932553854334e-08,
      "loss": 0.7615,
      "step": 5669
    },
    {
      "epoch": 0.5605120727578281,
      "grad_norm": 5.759047416319832,
      "learning_rate": 8.530765115777567e-08,
      "loss": 0.6372,
      "step": 5670
    },
    {
      "epoch": 0.5606109285025827,
      "grad_norm": 3.218960966161888,
      "learning_rate": 8.527597828349163e-08,
      "loss": 0.7038,
      "step": 5671
    },
    {
      "epoch": 0.5607097842473371,
      "grad_norm": 4.676239404422223,
      "learning_rate": 8.524430691893885e-08,
      "loss": 0.709,
      "step": 5672
    },
    {
      "epoch": 0.5608086399920915,
      "grad_norm": 4.628160729360212,
      "learning_rate": 8.521263706736468e-08,
      "loss": 0.8096,
      "step": 5673
    },
    {
      "epoch": 0.5609074957368461,
      "grad_norm": 4.676249055778843,
      "learning_rate": 8.518096873201652e-08,
      "loss": 0.7634,
      "step": 5674
    },
    {
      "epoch": 0.5610063514816005,
      "grad_norm": 4.363687119587278,
      "learning_rate": 8.514930191614136e-08,
      "loss": 0.7115,
      "step": 5675
    },
    {
      "epoch": 0.5611052072263549,
      "grad_norm": 9.275169337031349,
      "learning_rate": 8.511763662298627e-08,
      "loss": 0.8771,
      "step": 5676
    },
    {
      "epoch": 0.5612040629711094,
      "grad_norm": 8.89765068140308,
      "learning_rate": 8.508597285579798e-08,
      "loss": 0.7237,
      "step": 5677
    },
    {
      "epoch": 0.5613029187158639,
      "grad_norm": 4.685988277965803,
      "learning_rate": 8.50543106178232e-08,
      "loss": 0.684,
      "step": 5678
    },
    {
      "epoch": 0.5614017744606183,
      "grad_norm": 4.008633401180562,
      "learning_rate": 8.502264991230838e-08,
      "loss": 0.6854,
      "step": 5679
    },
    {
      "epoch": 0.5615006302053728,
      "grad_norm": 3.9393636152286007,
      "learning_rate": 8.49909907424999e-08,
      "loss": 0.6368,
      "step": 5680
    },
    {
      "epoch": 0.5615994859501273,
      "grad_norm": 3.7254231563734903,
      "learning_rate": 8.49593331116439e-08,
      "loss": 0.6293,
      "step": 5681
    },
    {
      "epoch": 0.5616983416948818,
      "grad_norm": 12.626517413198266,
      "learning_rate": 8.492767702298646e-08,
      "loss": 0.6836,
      "step": 5682
    },
    {
      "epoch": 0.5617971974396362,
      "grad_norm": 3.699984331521919,
      "learning_rate": 8.489602247977337e-08,
      "loss": 0.6336,
      "step": 5683
    },
    {
      "epoch": 0.5618960531843907,
      "grad_norm": 3.2258209229275545,
      "learning_rate": 8.486436948525037e-08,
      "loss": 0.7019,
      "step": 5684
    },
    {
      "epoch": 0.5619949089291452,
      "grad_norm": 4.064638538781205,
      "learning_rate": 8.483271804266305e-08,
      "loss": 0.6548,
      "step": 5685
    },
    {
      "epoch": 0.5620937646738996,
      "grad_norm": 3.5857249936384044,
      "learning_rate": 8.480106815525673e-08,
      "loss": 0.7256,
      "step": 5686
    },
    {
      "epoch": 0.562192620418654,
      "grad_norm": 4.556952758334905,
      "learning_rate": 8.476941982627672e-08,
      "loss": 0.7208,
      "step": 5687
    },
    {
      "epoch": 0.5622914761634086,
      "grad_norm": 6.709768272261749,
      "learning_rate": 8.473777305896798e-08,
      "loss": 0.7914,
      "step": 5688
    },
    {
      "epoch": 0.562390331908163,
      "grad_norm": 4.151950919762593,
      "learning_rate": 8.470612785657552e-08,
      "loss": 0.6732,
      "step": 5689
    },
    {
      "epoch": 0.5624891876529174,
      "grad_norm": 16.41292721404762,
      "learning_rate": 8.467448422234405e-08,
      "loss": 0.7063,
      "step": 5690
    },
    {
      "epoch": 0.562588043397672,
      "grad_norm": 3.723090133650851,
      "learning_rate": 8.464284215951811e-08,
      "loss": 0.7595,
      "step": 5691
    },
    {
      "epoch": 0.5626868991424264,
      "grad_norm": 3.9791937756092346,
      "learning_rate": 8.46112016713422e-08,
      "loss": 0.6581,
      "step": 5692
    },
    {
      "epoch": 0.5627857548871809,
      "grad_norm": 7.8539308560782874,
      "learning_rate": 8.457956276106058e-08,
      "loss": 0.6756,
      "step": 5693
    },
    {
      "epoch": 0.5628846106319354,
      "grad_norm": 3.627789598143021,
      "learning_rate": 8.454792543191728e-08,
      "loss": 0.7719,
      "step": 5694
    },
    {
      "epoch": 0.5629834663766898,
      "grad_norm": 3.809734508939976,
      "learning_rate": 8.451628968715634e-08,
      "loss": 0.6491,
      "step": 5695
    },
    {
      "epoch": 0.5630823221214443,
      "grad_norm": 3.2717409487970355,
      "learning_rate": 8.448465553002147e-08,
      "loss": 0.6866,
      "step": 5696
    },
    {
      "epoch": 0.5631811778661987,
      "grad_norm": 4.703541519248384,
      "learning_rate": 8.445302296375632e-08,
      "loss": 0.7219,
      "step": 5697
    },
    {
      "epoch": 0.5632800336109532,
      "grad_norm": 4.549821166626662,
      "learning_rate": 8.442139199160432e-08,
      "loss": 0.6065,
      "step": 5698
    },
    {
      "epoch": 0.5633788893557077,
      "grad_norm": 80.68119080243555,
      "learning_rate": 8.438976261680878e-08,
      "loss": 0.6911,
      "step": 5699
    },
    {
      "epoch": 0.5634777451004621,
      "grad_norm": 7.1312155505191335,
      "learning_rate": 8.435813484261279e-08,
      "loss": 0.6474,
      "step": 5700
    },
    {
      "epoch": 0.5635766008452167,
      "grad_norm": 6.724437770527778,
      "learning_rate": 8.432650867225938e-08,
      "loss": 0.6257,
      "step": 5701
    },
    {
      "epoch": 0.5636754565899711,
      "grad_norm": 2.5008685272942572,
      "learning_rate": 8.429488410899127e-08,
      "loss": 0.7585,
      "step": 5702
    },
    {
      "epoch": 0.5637743123347255,
      "grad_norm": 9.666956877987843,
      "learning_rate": 8.426326115605113e-08,
      "loss": 0.7435,
      "step": 5703
    },
    {
      "epoch": 0.5638731680794801,
      "grad_norm": 8.832196501962256,
      "learning_rate": 8.423163981668146e-08,
      "loss": 0.7361,
      "step": 5704
    },
    {
      "epoch": 0.5639720238242345,
      "grad_norm": 7.140635472763243,
      "learning_rate": 8.420002009412452e-08,
      "loss": 0.7347,
      "step": 5705
    },
    {
      "epoch": 0.5640708795689889,
      "grad_norm": 5.229333175956623,
      "learning_rate": 8.416840199162241e-08,
      "loss": 0.763,
      "step": 5706
    },
    {
      "epoch": 0.5641697353137434,
      "grad_norm": 8.153786396146925,
      "learning_rate": 8.41367855124172e-08,
      "loss": 0.7684,
      "step": 5707
    },
    {
      "epoch": 0.5642685910584979,
      "grad_norm": 4.713407812171167,
      "learning_rate": 8.410517065975058e-08,
      "loss": 0.8002,
      "step": 5708
    },
    {
      "epoch": 0.5643674468032523,
      "grad_norm": 3.5788265525199647,
      "learning_rate": 8.407355743686431e-08,
      "loss": 0.6956,
      "step": 5709
    },
    {
      "epoch": 0.5644663025480068,
      "grad_norm": 4.099951119870306,
      "learning_rate": 8.404194584699973e-08,
      "loss": 0.6819,
      "step": 5710
    },
    {
      "epoch": 0.5645651582927613,
      "grad_norm": 3.768054094521578,
      "learning_rate": 8.401033589339821e-08,
      "loss": 0.74,
      "step": 5711
    },
    {
      "epoch": 0.5646640140375158,
      "grad_norm": 7.508807184604626,
      "learning_rate": 8.397872757930092e-08,
      "loss": 0.7271,
      "step": 5712
    },
    {
      "epoch": 0.5647628697822702,
      "grad_norm": 4.956585813835168,
      "learning_rate": 8.394712090794875e-08,
      "loss": 0.8154,
      "step": 5713
    },
    {
      "epoch": 0.5648617255270247,
      "grad_norm": 4.822708481057138,
      "learning_rate": 8.391551588258256e-08,
      "loss": 0.7151,
      "step": 5714
    },
    {
      "epoch": 0.5649605812717792,
      "grad_norm": 3.939179407295999,
      "learning_rate": 8.388391250644294e-08,
      "loss": 0.7422,
      "step": 5715
    },
    {
      "epoch": 0.5650594370165336,
      "grad_norm": 4.315366935907443,
      "learning_rate": 8.385231078277037e-08,
      "loss": 0.7205,
      "step": 5716
    },
    {
      "epoch": 0.565158292761288,
      "grad_norm": 2.972167877981754,
      "learning_rate": 8.382071071480512e-08,
      "loss": 0.6552,
      "step": 5717
    },
    {
      "epoch": 0.5652571485060426,
      "grad_norm": 3.4979336182699026,
      "learning_rate": 8.378911230578736e-08,
      "loss": 0.6368,
      "step": 5718
    },
    {
      "epoch": 0.565356004250797,
      "grad_norm": 4.041805627082899,
      "learning_rate": 8.375751555895696e-08,
      "loss": 0.6603,
      "step": 5719
    },
    {
      "epoch": 0.5654548599955515,
      "grad_norm": 2.761764977345551,
      "learning_rate": 8.37259204775538e-08,
      "loss": 0.6331,
      "step": 5720
    },
    {
      "epoch": 0.565553715740306,
      "grad_norm": 9.010982556321155,
      "learning_rate": 8.369432706481739e-08,
      "loss": 0.7272,
      "step": 5721
    },
    {
      "epoch": 0.5656525714850604,
      "grad_norm": 3.5572630199836808,
      "learning_rate": 8.366273532398722e-08,
      "loss": 0.6002,
      "step": 5722
    },
    {
      "epoch": 0.5657514272298149,
      "grad_norm": 3.7754466083540943,
      "learning_rate": 8.363114525830258e-08,
      "loss": 0.7224,
      "step": 5723
    },
    {
      "epoch": 0.5658502829745694,
      "grad_norm": 4.098463553745817,
      "learning_rate": 8.359955687100251e-08,
      "loss": 0.69,
      "step": 5724
    },
    {
      "epoch": 0.5659491387193238,
      "grad_norm": 3.0640059259359265,
      "learning_rate": 8.356797016532594e-08,
      "loss": 0.719,
      "step": 5725
    },
    {
      "epoch": 0.5660479944640783,
      "grad_norm": 5.282919925044142,
      "learning_rate": 8.353638514451169e-08,
      "loss": 0.7294,
      "step": 5726
    },
    {
      "epoch": 0.5661468502088327,
      "grad_norm": 4.544609554361328,
      "learning_rate": 8.350480181179824e-08,
      "loss": 0.7844,
      "step": 5727
    },
    {
      "epoch": 0.5662457059535873,
      "grad_norm": 4.862583421981987,
      "learning_rate": 8.347322017042405e-08,
      "loss": 0.6996,
      "step": 5728
    },
    {
      "epoch": 0.5663445616983417,
      "grad_norm": 3.8002182466109384,
      "learning_rate": 8.344164022362738e-08,
      "loss": 0.6518,
      "step": 5729
    },
    {
      "epoch": 0.5664434174430961,
      "grad_norm": 6.825539751138623,
      "learning_rate": 8.341006197464622e-08,
      "loss": 0.6999,
      "step": 5730
    },
    {
      "epoch": 0.5665422731878507,
      "grad_norm": 18.932724092348842,
      "learning_rate": 8.337848542671853e-08,
      "loss": 0.6666,
      "step": 5731
    },
    {
      "epoch": 0.5666411289326051,
      "grad_norm": 4.7699768253783095,
      "learning_rate": 8.334691058308195e-08,
      "loss": 0.7087,
      "step": 5732
    },
    {
      "epoch": 0.5667399846773595,
      "grad_norm": 4.756790664870883,
      "learning_rate": 8.3315337446974e-08,
      "loss": 0.7634,
      "step": 5733
    },
    {
      "epoch": 0.5668388404221141,
      "grad_norm": 6.588416923534919,
      "learning_rate": 8.328376602163214e-08,
      "loss": 0.73,
      "step": 5734
    },
    {
      "epoch": 0.5669376961668685,
      "grad_norm": 3.371678969811359,
      "learning_rate": 8.325219631029344e-08,
      "loss": 0.6783,
      "step": 5735
    },
    {
      "epoch": 0.5670365519116229,
      "grad_norm": 22.220386229484344,
      "learning_rate": 8.322062831619497e-08,
      "loss": 0.6524,
      "step": 5736
    },
    {
      "epoch": 0.5671354076563774,
      "grad_norm": 8.536178910381503,
      "learning_rate": 8.318906204257359e-08,
      "loss": 0.6484,
      "step": 5737
    },
    {
      "epoch": 0.5672342634011319,
      "grad_norm": 11.011969830819462,
      "learning_rate": 8.315749749266587e-08,
      "loss": 0.7472,
      "step": 5738
    },
    {
      "epoch": 0.5673331191458864,
      "grad_norm": 5.710677824791938,
      "learning_rate": 8.312593466970836e-08,
      "loss": 0.7241,
      "step": 5739
    },
    {
      "epoch": 0.5674319748906408,
      "grad_norm": 4.4132678849852445,
      "learning_rate": 8.309437357693736e-08,
      "loss": 0.6213,
      "step": 5740
    },
    {
      "epoch": 0.5675308306353953,
      "grad_norm": 3.1534352488636395,
      "learning_rate": 8.306281421758897e-08,
      "loss": 0.7506,
      "step": 5741
    },
    {
      "epoch": 0.5676296863801498,
      "grad_norm": 4.918406098672621,
      "learning_rate": 8.303125659489911e-08,
      "loss": 0.7598,
      "step": 5742
    },
    {
      "epoch": 0.5677285421249042,
      "grad_norm": 7.631878667265885,
      "learning_rate": 8.29997007121036e-08,
      "loss": 0.8237,
      "step": 5743
    },
    {
      "epoch": 0.5678273978696587,
      "grad_norm": 4.14068433601075,
      "learning_rate": 8.2968146572438e-08,
      "loss": 0.7765,
      "step": 5744
    },
    {
      "epoch": 0.5679262536144132,
      "grad_norm": 45.05423416723282,
      "learning_rate": 8.293659417913776e-08,
      "loss": 0.6744,
      "step": 5745
    },
    {
      "epoch": 0.5680251093591676,
      "grad_norm": 4.7279335330810275,
      "learning_rate": 8.290504353543804e-08,
      "loss": 0.6541,
      "step": 5746
    },
    {
      "epoch": 0.568123965103922,
      "grad_norm": 3.390051848068928,
      "learning_rate": 8.287349464457395e-08,
      "loss": 0.6931,
      "step": 5747
    },
    {
      "epoch": 0.5682228208486766,
      "grad_norm": 2.9129505963200684,
      "learning_rate": 8.284194750978039e-08,
      "loss": 0.6718,
      "step": 5748
    },
    {
      "epoch": 0.568321676593431,
      "grad_norm": 3.6593708627000523,
      "learning_rate": 8.281040213429198e-08,
      "loss": 0.6695,
      "step": 5749
    },
    {
      "epoch": 0.5684205323381855,
      "grad_norm": 2.8688822870472324,
      "learning_rate": 8.277885852134325e-08,
      "loss": 0.7183,
      "step": 5750
    },
    {
      "epoch": 0.56851938808294,
      "grad_norm": 16.833099758020104,
      "learning_rate": 8.27473166741686e-08,
      "loss": 0.7057,
      "step": 5751
    },
    {
      "epoch": 0.5686182438276944,
      "grad_norm": 2.6682612779261,
      "learning_rate": 8.271577659600209e-08,
      "loss": 0.6512,
      "step": 5752
    },
    {
      "epoch": 0.5687170995724489,
      "grad_norm": 3.52474110491925,
      "learning_rate": 8.268423829007775e-08,
      "loss": 0.7473,
      "step": 5753
    },
    {
      "epoch": 0.5688159553172034,
      "grad_norm": 4.986351505263934,
      "learning_rate": 8.265270175962933e-08,
      "loss": 0.6192,
      "step": 5754
    },
    {
      "epoch": 0.5689148110619578,
      "grad_norm": 5.1967259930388385,
      "learning_rate": 8.262116700789044e-08,
      "loss": 0.6308,
      "step": 5755
    },
    {
      "epoch": 0.5690136668067123,
      "grad_norm": 2.9213192393116567,
      "learning_rate": 8.258963403809456e-08,
      "loss": 0.5351,
      "step": 5756
    },
    {
      "epoch": 0.5691125225514668,
      "grad_norm": 4.262134562506784,
      "learning_rate": 8.255810285347485e-08,
      "loss": 0.7139,
      "step": 5757
    },
    {
      "epoch": 0.5692113782962213,
      "grad_norm": 7.068833415909573,
      "learning_rate": 8.252657345726443e-08,
      "loss": 0.8013,
      "step": 5758
    },
    {
      "epoch": 0.5693102340409757,
      "grad_norm": 3.9343458423149316,
      "learning_rate": 8.249504585269617e-08,
      "loss": 0.6747,
      "step": 5759
    },
    {
      "epoch": 0.5694090897857301,
      "grad_norm": 2.763257187458842,
      "learning_rate": 8.24635200430027e-08,
      "loss": 0.6999,
      "step": 5760
    },
    {
      "epoch": 0.5695079455304847,
      "grad_norm": 5.76451974249321,
      "learning_rate": 8.243199603141656e-08,
      "loss": 0.732,
      "step": 5761
    },
    {
      "epoch": 0.5696068012752391,
      "grad_norm": 3.4904799917522493,
      "learning_rate": 8.240047382117014e-08,
      "loss": 0.7845,
      "step": 5762
    },
    {
      "epoch": 0.5697056570199935,
      "grad_norm": 8.26485778888831,
      "learning_rate": 8.236895341549546e-08,
      "loss": 0.7382,
      "step": 5763
    },
    {
      "epoch": 0.5698045127647481,
      "grad_norm": 3.6040753777628094,
      "learning_rate": 8.233743481762459e-08,
      "loss": 0.6768,
      "step": 5764
    },
    {
      "epoch": 0.5699033685095025,
      "grad_norm": 5.468959215410337,
      "learning_rate": 8.230591803078917e-08,
      "loss": 0.7257,
      "step": 5765
    },
    {
      "epoch": 0.570002224254257,
      "grad_norm": 8.053436255268032,
      "learning_rate": 8.227440305822088e-08,
      "loss": 0.816,
      "step": 5766
    },
    {
      "epoch": 0.5701010799990115,
      "grad_norm": 6.11959781746441,
      "learning_rate": 8.22428899031511e-08,
      "loss": 0.7844,
      "step": 5767
    },
    {
      "epoch": 0.5701999357437659,
      "grad_norm": 4.089909421988143,
      "learning_rate": 8.221137856881102e-08,
      "loss": 0.6849,
      "step": 5768
    },
    {
      "epoch": 0.5702987914885204,
      "grad_norm": 3.4137107573537637,
      "learning_rate": 8.217986905843164e-08,
      "loss": 0.7379,
      "step": 5769
    },
    {
      "epoch": 0.5703976472332748,
      "grad_norm": 4.551681195752091,
      "learning_rate": 8.214836137524387e-08,
      "loss": 0.7564,
      "step": 5770
    },
    {
      "epoch": 0.5704965029780293,
      "grad_norm": 3.433775326095088,
      "learning_rate": 8.211685552247827e-08,
      "loss": 0.6907,
      "step": 5771
    },
    {
      "epoch": 0.5705953587227838,
      "grad_norm": 2.6980100066663906,
      "learning_rate": 8.208535150336533e-08,
      "loss": 0.6782,
      "step": 5772
    },
    {
      "epoch": 0.5706942144675382,
      "grad_norm": 5.468379290895837,
      "learning_rate": 8.205384932113539e-08,
      "loss": 0.733,
      "step": 5773
    },
    {
      "epoch": 0.5707930702122928,
      "grad_norm": 12.673819014864709,
      "learning_rate": 8.202234897901843e-08,
      "loss": 0.72,
      "step": 5774
    },
    {
      "epoch": 0.5708919259570472,
      "grad_norm": 4.269312301786664,
      "learning_rate": 8.199085048024442e-08,
      "loss": 0.7221,
      "step": 5775
    },
    {
      "epoch": 0.5709907817018016,
      "grad_norm": 5.233617616599987,
      "learning_rate": 8.195935382804303e-08,
      "loss": 0.73,
      "step": 5776
    },
    {
      "epoch": 0.5710896374465562,
      "grad_norm": 3.6306908923904055,
      "learning_rate": 8.192785902564377e-08,
      "loss": 0.7306,
      "step": 5777
    },
    {
      "epoch": 0.5711884931913106,
      "grad_norm": 3.667700951061173,
      "learning_rate": 8.189636607627603e-08,
      "loss": 0.8471,
      "step": 5778
    },
    {
      "epoch": 0.571287348936065,
      "grad_norm": 7.601872591540931,
      "learning_rate": 8.186487498316886e-08,
      "loss": 0.7407,
      "step": 5779
    },
    {
      "epoch": 0.5713862046808195,
      "grad_norm": 4.052302064846763,
      "learning_rate": 8.183338574955123e-08,
      "loss": 0.7702,
      "step": 5780
    },
    {
      "epoch": 0.571485060425574,
      "grad_norm": 7.850648924733983,
      "learning_rate": 8.180189837865196e-08,
      "loss": 0.83,
      "step": 5781
    },
    {
      "epoch": 0.5715839161703284,
      "grad_norm": 5.900913768119201,
      "learning_rate": 8.177041287369955e-08,
      "loss": 0.6382,
      "step": 5782
    },
    {
      "epoch": 0.5716827719150829,
      "grad_norm": 10.103702306148817,
      "learning_rate": 8.17389292379224e-08,
      "loss": 0.699,
      "step": 5783
    },
    {
      "epoch": 0.5717816276598374,
      "grad_norm": 2.9188908710032555,
      "learning_rate": 8.17074474745487e-08,
      "loss": 0.8353,
      "step": 5784
    },
    {
      "epoch": 0.5718804834045919,
      "grad_norm": 4.598168268928201,
      "learning_rate": 8.167596758680641e-08,
      "loss": 0.6587,
      "step": 5785
    },
    {
      "epoch": 0.5719793391493463,
      "grad_norm": 6.095519085801619,
      "learning_rate": 8.164448957792335e-08,
      "loss": 0.7825,
      "step": 5786
    },
    {
      "epoch": 0.5720781948941008,
      "grad_norm": 3.1011564605333715,
      "learning_rate": 8.161301345112713e-08,
      "loss": 0.5837,
      "step": 5787
    },
    {
      "epoch": 0.5721770506388553,
      "grad_norm": 4.480807429685703,
      "learning_rate": 8.158153920964512e-08,
      "loss": 0.8083,
      "step": 5788
    },
    {
      "epoch": 0.5722759063836097,
      "grad_norm": 75.89713442947463,
      "learning_rate": 8.155006685670462e-08,
      "loss": 0.7313,
      "step": 5789
    },
    {
      "epoch": 0.5723747621283641,
      "grad_norm": 3.380624603865087,
      "learning_rate": 8.151859639553256e-08,
      "loss": 0.6522,
      "step": 5790
    },
    {
      "epoch": 0.5724736178731187,
      "grad_norm": 19.833349387044496,
      "learning_rate": 8.148712782935582e-08,
      "loss": 0.72,
      "step": 5791
    },
    {
      "epoch": 0.5725724736178731,
      "grad_norm": 4.63398151607536,
      "learning_rate": 8.14556611614011e-08,
      "loss": 0.6154,
      "step": 5792
    },
    {
      "epoch": 0.5726713293626275,
      "grad_norm": 4.866717584760495,
      "learning_rate": 8.142419639489474e-08,
      "loss": 0.7425,
      "step": 5793
    },
    {
      "epoch": 0.5727701851073821,
      "grad_norm": 6.663186184438289,
      "learning_rate": 8.139273353306301e-08,
      "loss": 0.6653,
      "step": 5794
    },
    {
      "epoch": 0.5728690408521365,
      "grad_norm": 12.075860888574542,
      "learning_rate": 8.136127257913202e-08,
      "loss": 0.7437,
      "step": 5795
    },
    {
      "epoch": 0.572967896596891,
      "grad_norm": 10.802707858363302,
      "learning_rate": 8.132981353632755e-08,
      "loss": 0.7488,
      "step": 5796
    },
    {
      "epoch": 0.5730667523416455,
      "grad_norm": 8.520515363367952,
      "learning_rate": 8.129835640787533e-08,
      "loss": 0.8038,
      "step": 5797
    },
    {
      "epoch": 0.5731656080863999,
      "grad_norm": 4.511429937538754,
      "learning_rate": 8.126690119700076e-08,
      "loss": 0.6427,
      "step": 5798
    },
    {
      "epoch": 0.5732644638311544,
      "grad_norm": 3.059444873125183,
      "learning_rate": 8.123544790692913e-08,
      "loss": 0.7931,
      "step": 5799
    },
    {
      "epoch": 0.5733633195759088,
      "grad_norm": 3.505868420706949,
      "learning_rate": 8.120399654088558e-08,
      "loss": 0.61,
      "step": 5800
    },
    {
      "epoch": 0.5734621753206633,
      "grad_norm": 3.3058881131662763,
      "learning_rate": 8.117254710209487e-08,
      "loss": 0.6278,
      "step": 5801
    },
    {
      "epoch": 0.5735610310654178,
      "grad_norm": 5.045298905863333,
      "learning_rate": 8.114109959378174e-08,
      "loss": 0.8382,
      "step": 5802
    },
    {
      "epoch": 0.5736598868101722,
      "grad_norm": 13.815477797537483,
      "learning_rate": 8.110965401917069e-08,
      "loss": 0.7029,
      "step": 5803
    },
    {
      "epoch": 0.5737587425549268,
      "grad_norm": 3.705503463916965,
      "learning_rate": 8.107821038148593e-08,
      "loss": 0.6941,
      "step": 5804
    },
    {
      "epoch": 0.5738575982996812,
      "grad_norm": 6.449322587320779,
      "learning_rate": 8.104676868395157e-08,
      "loss": 0.7907,
      "step": 5805
    },
    {
      "epoch": 0.5739564540444356,
      "grad_norm": 4.0116850747215125,
      "learning_rate": 8.101532892979156e-08,
      "loss": 0.6839,
      "step": 5806
    },
    {
      "epoch": 0.5740553097891902,
      "grad_norm": 3.7561531816151077,
      "learning_rate": 8.098389112222947e-08,
      "loss": 0.704,
      "step": 5807
    },
    {
      "epoch": 0.5741541655339446,
      "grad_norm": 3.6036409982736486,
      "learning_rate": 8.095245526448889e-08,
      "loss": 0.557,
      "step": 5808
    },
    {
      "epoch": 0.574253021278699,
      "grad_norm": 3.0827418430911724,
      "learning_rate": 8.092102135979299e-08,
      "loss": 0.6956,
      "step": 5809
    },
    {
      "epoch": 0.5743518770234535,
      "grad_norm": 3.566026886930881,
      "learning_rate": 8.088958941136495e-08,
      "loss": 0.7153,
      "step": 5810
    },
    {
      "epoch": 0.574450732768208,
      "grad_norm": 4.732488932605444,
      "learning_rate": 8.085815942242762e-08,
      "loss": 0.7856,
      "step": 5811
    },
    {
      "epoch": 0.5745495885129625,
      "grad_norm": 3.724780661292933,
      "learning_rate": 8.082673139620366e-08,
      "loss": 0.7896,
      "step": 5812
    },
    {
      "epoch": 0.5746484442577169,
      "grad_norm": 3.900739667497971,
      "learning_rate": 8.079530533591557e-08,
      "loss": 0.6779,
      "step": 5813
    },
    {
      "epoch": 0.5747473000024714,
      "grad_norm": 7.730751549577519,
      "learning_rate": 8.076388124478565e-08,
      "loss": 0.799,
      "step": 5814
    },
    {
      "epoch": 0.5748461557472259,
      "grad_norm": 4.622447362041101,
      "learning_rate": 8.073245912603594e-08,
      "loss": 0.6887,
      "step": 5815
    },
    {
      "epoch": 0.5749450114919803,
      "grad_norm": 9.943640137080605,
      "learning_rate": 8.070103898288832e-08,
      "loss": 0.7301,
      "step": 5816
    },
    {
      "epoch": 0.5750438672367348,
      "grad_norm": 4.490399283427174,
      "learning_rate": 8.06696208185645e-08,
      "loss": 0.6978,
      "step": 5817
    },
    {
      "epoch": 0.5751427229814893,
      "grad_norm": 3.5603857530213685,
      "learning_rate": 8.06382046362859e-08,
      "loss": 0.7077,
      "step": 5818
    },
    {
      "epoch": 0.5752415787262437,
      "grad_norm": 5.877840407618834,
      "learning_rate": 8.060679043927385e-08,
      "loss": 0.6192,
      "step": 5819
    },
    {
      "epoch": 0.5753404344709981,
      "grad_norm": 8.995545527066994,
      "learning_rate": 8.057537823074934e-08,
      "loss": 0.7211,
      "step": 5820
    },
    {
      "epoch": 0.5754392902157527,
      "grad_norm": 4.781175743001937,
      "learning_rate": 8.054396801393324e-08,
      "loss": 0.6954,
      "step": 5821
    },
    {
      "epoch": 0.5755381459605071,
      "grad_norm": 4.674748784270626,
      "learning_rate": 8.051255979204626e-08,
      "loss": 0.7397,
      "step": 5822
    },
    {
      "epoch": 0.5756370017052616,
      "grad_norm": 3.9335288319339066,
      "learning_rate": 8.048115356830878e-08,
      "loss": 0.682,
      "step": 5823
    },
    {
      "epoch": 0.5757358574500161,
      "grad_norm": 4.799503056442607,
      "learning_rate": 8.044974934594107e-08,
      "loss": 0.7865,
      "step": 5824
    },
    {
      "epoch": 0.5758347131947705,
      "grad_norm": 3.4969583882819197,
      "learning_rate": 8.041834712816322e-08,
      "loss": 0.7952,
      "step": 5825
    },
    {
      "epoch": 0.575933568939525,
      "grad_norm": 26.713203978176885,
      "learning_rate": 8.038694691819498e-08,
      "loss": 0.7202,
      "step": 5826
    },
    {
      "epoch": 0.5760324246842795,
      "grad_norm": 7.925842485413404,
      "learning_rate": 8.035554871925606e-08,
      "loss": 0.6686,
      "step": 5827
    },
    {
      "epoch": 0.5761312804290339,
      "grad_norm": 5.293275657336656,
      "learning_rate": 8.032415253456582e-08,
      "loss": 0.7347,
      "step": 5828
    },
    {
      "epoch": 0.5762301361737884,
      "grad_norm": 19.569068067883787,
      "learning_rate": 8.029275836734351e-08,
      "loss": 0.6881,
      "step": 5829
    },
    {
      "epoch": 0.5763289919185429,
      "grad_norm": 19.479621137583376,
      "learning_rate": 8.026136622080815e-08,
      "loss": 0.7339,
      "step": 5830
    },
    {
      "epoch": 0.5764278476632974,
      "grad_norm": 4.993737931599202,
      "learning_rate": 8.022997609817848e-08,
      "loss": 0.8138,
      "step": 5831
    },
    {
      "epoch": 0.5765267034080518,
      "grad_norm": 3.626587976341799,
      "learning_rate": 8.019858800267312e-08,
      "loss": 0.7079,
      "step": 5832
    },
    {
      "epoch": 0.5766255591528062,
      "grad_norm": 6.351841017192694,
      "learning_rate": 8.016720193751053e-08,
      "loss": 0.7097,
      "step": 5833
    },
    {
      "epoch": 0.5767244148975608,
      "grad_norm": 3.107258168791699,
      "learning_rate": 8.013581790590876e-08,
      "loss": 0.7429,
      "step": 5834
    },
    {
      "epoch": 0.5768232706423152,
      "grad_norm": 4.587197958447509,
      "learning_rate": 8.010443591108586e-08,
      "loss": 0.6762,
      "step": 5835
    },
    {
      "epoch": 0.5769221263870696,
      "grad_norm": 4.023453381044657,
      "learning_rate": 8.007305595625961e-08,
      "loss": 0.7179,
      "step": 5836
    },
    {
      "epoch": 0.5770209821318242,
      "grad_norm": 4.089297316147371,
      "learning_rate": 8.004167804464752e-08,
      "loss": 0.6711,
      "step": 5837
    },
    {
      "epoch": 0.5771198378765786,
      "grad_norm": 41.539799620816986,
      "learning_rate": 8.001030217946692e-08,
      "loss": 0.7855,
      "step": 5838
    },
    {
      "epoch": 0.577218693621333,
      "grad_norm": 7.122310685310781,
      "learning_rate": 7.997892836393499e-08,
      "loss": 0.695,
      "step": 5839
    },
    {
      "epoch": 0.5773175493660876,
      "grad_norm": 11.18500330772825,
      "learning_rate": 7.99475566012686e-08,
      "loss": 0.7644,
      "step": 5840
    },
    {
      "epoch": 0.577416405110842,
      "grad_norm": 3.9949963628443492,
      "learning_rate": 7.991618689468452e-08,
      "loss": 0.7114,
      "step": 5841
    },
    {
      "epoch": 0.5775152608555965,
      "grad_norm": 5.279655696806209,
      "learning_rate": 7.988481924739916e-08,
      "loss": 0.7425,
      "step": 5842
    },
    {
      "epoch": 0.5776141166003509,
      "grad_norm": 3.439545550586066,
      "learning_rate": 7.98534536626289e-08,
      "loss": 0.7286,
      "step": 5843
    },
    {
      "epoch": 0.5777129723451054,
      "grad_norm": 5.635034799693416,
      "learning_rate": 7.982209014358979e-08,
      "loss": 0.6863,
      "step": 5844
    },
    {
      "epoch": 0.5778118280898599,
      "grad_norm": 6.931549644456772,
      "learning_rate": 7.979072869349767e-08,
      "loss": 0.6724,
      "step": 5845
    },
    {
      "epoch": 0.5779106838346143,
      "grad_norm": 3.499675700286754,
      "learning_rate": 7.975936931556823e-08,
      "loss": 0.6907,
      "step": 5846
    },
    {
      "epoch": 0.5780095395793688,
      "grad_norm": 4.2641770854892735,
      "learning_rate": 7.972801201301693e-08,
      "loss": 0.7617,
      "step": 5847
    },
    {
      "epoch": 0.5781083953241233,
      "grad_norm": 4.429592683293635,
      "learning_rate": 7.96966567890589e-08,
      "loss": 0.6937,
      "step": 5848
    },
    {
      "epoch": 0.5782072510688777,
      "grad_norm": 4.8511688524800585,
      "learning_rate": 7.966530364690924e-08,
      "loss": 0.6573,
      "step": 5849
    },
    {
      "epoch": 0.5783061068136323,
      "grad_norm": 3.625076061371386,
      "learning_rate": 7.963395258978276e-08,
      "loss": 0.6143,
      "step": 5850
    },
    {
      "epoch": 0.5784049625583867,
      "grad_norm": 4.067307136870742,
      "learning_rate": 7.960260362089399e-08,
      "loss": 0.6956,
      "step": 5851
    },
    {
      "epoch": 0.5785038183031411,
      "grad_norm": 2.9773974338626017,
      "learning_rate": 7.957125674345738e-08,
      "loss": 0.6935,
      "step": 5852
    },
    {
      "epoch": 0.5786026740478956,
      "grad_norm": 3.788201303457232,
      "learning_rate": 7.9539911960687e-08,
      "loss": 0.6761,
      "step": 5853
    },
    {
      "epoch": 0.5787015297926501,
      "grad_norm": 4.9975764811263215,
      "learning_rate": 7.950856927579685e-08,
      "loss": 0.7872,
      "step": 5854
    },
    {
      "epoch": 0.5788003855374045,
      "grad_norm": 3.489363421165812,
      "learning_rate": 7.947722869200068e-08,
      "loss": 0.7368,
      "step": 5855
    },
    {
      "epoch": 0.578899241282159,
      "grad_norm": 9.156578215396106,
      "learning_rate": 7.944589021251196e-08,
      "loss": 0.652,
      "step": 5856
    },
    {
      "epoch": 0.5789980970269135,
      "grad_norm": 3.4162520693209815,
      "learning_rate": 7.941455384054398e-08,
      "loss": 0.6855,
      "step": 5857
    },
    {
      "epoch": 0.579096952771668,
      "grad_norm": 3.6104453864047676,
      "learning_rate": 7.938321957930988e-08,
      "loss": 0.7325,
      "step": 5858
    },
    {
      "epoch": 0.5791958085164224,
      "grad_norm": 3.0774551108812074,
      "learning_rate": 7.935188743202247e-08,
      "loss": 0.7036,
      "step": 5859
    },
    {
      "epoch": 0.5792946642611769,
      "grad_norm": 3.9428916589763525,
      "learning_rate": 7.93205574018944e-08,
      "loss": 0.7453,
      "step": 5860
    },
    {
      "epoch": 0.5793935200059314,
      "grad_norm": 4.487331526368109,
      "learning_rate": 7.928922949213819e-08,
      "loss": 0.7228,
      "step": 5861
    },
    {
      "epoch": 0.5794923757506858,
      "grad_norm": 6.491521534954926,
      "learning_rate": 7.925790370596595e-08,
      "loss": 0.7891,
      "step": 5862
    },
    {
      "epoch": 0.5795912314954402,
      "grad_norm": 3.1847051536956634,
      "learning_rate": 7.922658004658975e-08,
      "loss": 0.6549,
      "step": 5863
    },
    {
      "epoch": 0.5796900872401948,
      "grad_norm": 6.213625417149403,
      "learning_rate": 7.91952585172213e-08,
      "loss": 0.7271,
      "step": 5864
    },
    {
      "epoch": 0.5797889429849492,
      "grad_norm": 4.920221040352242,
      "learning_rate": 7.91639391210722e-08,
      "loss": 0.6466,
      "step": 5865
    },
    {
      "epoch": 0.5798877987297036,
      "grad_norm": 4.365108961695499,
      "learning_rate": 7.913262186135383e-08,
      "loss": 0.6079,
      "step": 5866
    },
    {
      "epoch": 0.5799866544744582,
      "grad_norm": 3.1329846597671778,
      "learning_rate": 7.910130674127722e-08,
      "loss": 0.7357,
      "step": 5867
    },
    {
      "epoch": 0.5800855102192126,
      "grad_norm": 3.7368679480368963,
      "learning_rate": 7.906999376405332e-08,
      "loss": 0.7441,
      "step": 5868
    },
    {
      "epoch": 0.580184365963967,
      "grad_norm": 15.680182285561171,
      "learning_rate": 7.903868293289287e-08,
      "loss": 0.6966,
      "step": 5869
    },
    {
      "epoch": 0.5802832217087216,
      "grad_norm": 6.748491855622522,
      "learning_rate": 7.900737425100626e-08,
      "loss": 0.5918,
      "step": 5870
    },
    {
      "epoch": 0.580382077453476,
      "grad_norm": 5.61974014949513,
      "learning_rate": 7.897606772160375e-08,
      "loss": 0.6816,
      "step": 5871
    },
    {
      "epoch": 0.5804809331982305,
      "grad_norm": 4.893340026241116,
      "learning_rate": 7.894476334789538e-08,
      "loss": 0.7071,
      "step": 5872
    },
    {
      "epoch": 0.5805797889429849,
      "grad_norm": 3.071633172245995,
      "learning_rate": 7.891346113309094e-08,
      "loss": 0.7655,
      "step": 5873
    },
    {
      "epoch": 0.5806786446877394,
      "grad_norm": 14.960560787808406,
      "learning_rate": 7.888216108040004e-08,
      "loss": 0.6689,
      "step": 5874
    },
    {
      "epoch": 0.5807775004324939,
      "grad_norm": 3.2063728889731915,
      "learning_rate": 7.885086319303198e-08,
      "loss": 0.6905,
      "step": 5875
    },
    {
      "epoch": 0.5808763561772483,
      "grad_norm": 4.1815904743522,
      "learning_rate": 7.881956747419592e-08,
      "loss": 0.675,
      "step": 5876
    },
    {
      "epoch": 0.5809752119220029,
      "grad_norm": 4.179430979779995,
      "learning_rate": 7.878827392710082e-08,
      "loss": 0.5979,
      "step": 5877
    },
    {
      "epoch": 0.5810740676667573,
      "grad_norm": 4.509452329597974,
      "learning_rate": 7.875698255495528e-08,
      "loss": 0.8314,
      "step": 5878
    },
    {
      "epoch": 0.5811729234115117,
      "grad_norm": 3.5921645477455244,
      "learning_rate": 7.872569336096784e-08,
      "loss": 0.6134,
      "step": 5879
    },
    {
      "epoch": 0.5812717791562663,
      "grad_norm": 3.8676706377328394,
      "learning_rate": 7.869440634834677e-08,
      "loss": 0.7895,
      "step": 5880
    },
    {
      "epoch": 0.5813706349010207,
      "grad_norm": 4.287131517670594,
      "learning_rate": 7.866312152030002e-08,
      "loss": 0.7106,
      "step": 5881
    },
    {
      "epoch": 0.5814694906457751,
      "grad_norm": 4.550275263469946,
      "learning_rate": 7.863183888003538e-08,
      "loss": 0.6836,
      "step": 5882
    },
    {
      "epoch": 0.5815683463905296,
      "grad_norm": 6.071150373206913,
      "learning_rate": 7.860055843076053e-08,
      "loss": 0.6145,
      "step": 5883
    },
    {
      "epoch": 0.5816672021352841,
      "grad_norm": 5.009452125649452,
      "learning_rate": 7.85692801756827e-08,
      "loss": 0.7332,
      "step": 5884
    },
    {
      "epoch": 0.5817660578800385,
      "grad_norm": 190.7721169486765,
      "learning_rate": 7.85380041180091e-08,
      "loss": 0.792,
      "step": 5885
    },
    {
      "epoch": 0.581864913624793,
      "grad_norm": 3.574231089733683,
      "learning_rate": 7.850673026094655e-08,
      "loss": 0.6461,
      "step": 5886
    },
    {
      "epoch": 0.5819637693695475,
      "grad_norm": 3.3558854893815253,
      "learning_rate": 7.847545860770176e-08,
      "loss": 0.6365,
      "step": 5887
    },
    {
      "epoch": 0.582062625114302,
      "grad_norm": 6.134268333001915,
      "learning_rate": 7.844418916148123e-08,
      "loss": 0.7239,
      "step": 5888
    },
    {
      "epoch": 0.5821614808590564,
      "grad_norm": 3.625876845667662,
      "learning_rate": 7.841292192549109e-08,
      "loss": 0.718,
      "step": 5889
    },
    {
      "epoch": 0.5822603366038109,
      "grad_norm": 3.3105921229242603,
      "learning_rate": 7.838165690293739e-08,
      "loss": 0.7009,
      "step": 5890
    },
    {
      "epoch": 0.5823591923485654,
      "grad_norm": 4.693272614967511,
      "learning_rate": 7.835039409702591e-08,
      "loss": 0.7295,
      "step": 5891
    },
    {
      "epoch": 0.5824580480933198,
      "grad_norm": 10.730359160486268,
      "learning_rate": 7.831913351096211e-08,
      "loss": 0.7511,
      "step": 5892
    },
    {
      "epoch": 0.5825569038380742,
      "grad_norm": 9.4897000996809,
      "learning_rate": 7.828787514795135e-08,
      "loss": 0.7829,
      "step": 5893
    },
    {
      "epoch": 0.5826557595828288,
      "grad_norm": 4.585429833122537,
      "learning_rate": 7.825661901119876e-08,
      "loss": 0.711,
      "step": 5894
    },
    {
      "epoch": 0.5827546153275832,
      "grad_norm": 7.602459208086165,
      "learning_rate": 7.822536510390912e-08,
      "loss": 0.7149,
      "step": 5895
    },
    {
      "epoch": 0.5828534710723376,
      "grad_norm": 3.068522981919128,
      "learning_rate": 7.819411342928713e-08,
      "loss": 0.7085,
      "step": 5896
    },
    {
      "epoch": 0.5829523268170922,
      "grad_norm": 3.177925283313242,
      "learning_rate": 7.816286399053711e-08,
      "loss": 0.664,
      "step": 5897
    },
    {
      "epoch": 0.5830511825618466,
      "grad_norm": 3.318800879096164,
      "learning_rate": 7.813161679086329e-08,
      "loss": 0.7627,
      "step": 5898
    },
    {
      "epoch": 0.5831500383066011,
      "grad_norm": 3.2592079472103577,
      "learning_rate": 7.810037183346957e-08,
      "loss": 0.6802,
      "step": 5899
    },
    {
      "epoch": 0.5832488940513556,
      "grad_norm": 9.289225150229415,
      "learning_rate": 7.806912912155968e-08,
      "loss": 0.6414,
      "step": 5900
    },
    {
      "epoch": 0.58334774979611,
      "grad_norm": 5.1661077381618945,
      "learning_rate": 7.803788865833709e-08,
      "loss": 0.6592,
      "step": 5901
    },
    {
      "epoch": 0.5834466055408645,
      "grad_norm": 7.17108156046824,
      "learning_rate": 7.800665044700509e-08,
      "loss": 0.6081,
      "step": 5902
    },
    {
      "epoch": 0.583545461285619,
      "grad_norm": 5.175224019261036,
      "learning_rate": 7.797541449076662e-08,
      "loss": 0.7604,
      "step": 5903
    },
    {
      "epoch": 0.5836443170303734,
      "grad_norm": 5.05016768437379,
      "learning_rate": 7.79441807928245e-08,
      "loss": 0.7028,
      "step": 5904
    },
    {
      "epoch": 0.5837431727751279,
      "grad_norm": 3.970636555410633,
      "learning_rate": 7.791294935638133e-08,
      "loss": 0.7006,
      "step": 5905
    },
    {
      "epoch": 0.5838420285198823,
      "grad_norm": 2.9235381406080476,
      "learning_rate": 7.788172018463938e-08,
      "loss": 0.7101,
      "step": 5906
    },
    {
      "epoch": 0.5839408842646369,
      "grad_norm": 2.9442897939573704,
      "learning_rate": 7.785049328080078e-08,
      "loss": 0.6299,
      "step": 5907
    },
    {
      "epoch": 0.5840397400093913,
      "grad_norm": 8.562840833666016,
      "learning_rate": 7.781926864806735e-08,
      "loss": 0.7022,
      "step": 5908
    },
    {
      "epoch": 0.5841385957541457,
      "grad_norm": 2.9099019011260703,
      "learning_rate": 7.778804628964071e-08,
      "loss": 0.684,
      "step": 5909
    },
    {
      "epoch": 0.5842374514989003,
      "grad_norm": 9.192077885254628,
      "learning_rate": 7.775682620872232e-08,
      "loss": 0.8509,
      "step": 5910
    },
    {
      "epoch": 0.5843363072436547,
      "grad_norm": 4.02258664859726,
      "learning_rate": 7.772560840851324e-08,
      "loss": 0.7837,
      "step": 5911
    },
    {
      "epoch": 0.5844351629884091,
      "grad_norm": 4.561139569023554,
      "learning_rate": 7.769439289221445e-08,
      "loss": 0.6759,
      "step": 5912
    },
    {
      "epoch": 0.5845340187331637,
      "grad_norm": 4.0637088517959175,
      "learning_rate": 7.766317966302669e-08,
      "loss": 0.643,
      "step": 5913
    },
    {
      "epoch": 0.5846328744779181,
      "grad_norm": 3.277485881564006,
      "learning_rate": 7.763196872415032e-08,
      "loss": 0.7116,
      "step": 5914
    },
    {
      "epoch": 0.5847317302226726,
      "grad_norm": 3.92710865429683,
      "learning_rate": 7.760076007878562e-08,
      "loss": 0.8483,
      "step": 5915
    },
    {
      "epoch": 0.584830585967427,
      "grad_norm": 5.559333763087896,
      "learning_rate": 7.756955373013256e-08,
      "loss": 0.7733,
      "step": 5916
    },
    {
      "epoch": 0.5849294417121815,
      "grad_norm": 4.496787787577906,
      "learning_rate": 7.75383496813909e-08,
      "loss": 0.7259,
      "step": 5917
    },
    {
      "epoch": 0.585028297456936,
      "grad_norm": 10.503370124217925,
      "learning_rate": 7.750714793576016e-08,
      "loss": 0.6947,
      "step": 5918
    },
    {
      "epoch": 0.5851271532016904,
      "grad_norm": 4.026571300324471,
      "learning_rate": 7.747594849643957e-08,
      "loss": 0.741,
      "step": 5919
    },
    {
      "epoch": 0.5852260089464449,
      "grad_norm": 4.415241206868575,
      "learning_rate": 7.744475136662822e-08,
      "loss": 0.6282,
      "step": 5920
    },
    {
      "epoch": 0.5853248646911994,
      "grad_norm": 5.982679706718438,
      "learning_rate": 7.741355654952493e-08,
      "loss": 0.7073,
      "step": 5921
    },
    {
      "epoch": 0.5854237204359538,
      "grad_norm": 11.96229749384239,
      "learning_rate": 7.738236404832821e-08,
      "loss": 0.6712,
      "step": 5922
    },
    {
      "epoch": 0.5855225761807084,
      "grad_norm": 5.387699713455965,
      "learning_rate": 7.735117386623641e-08,
      "loss": 0.7243,
      "step": 5923
    },
    {
      "epoch": 0.5856214319254628,
      "grad_norm": 3.536856985786053,
      "learning_rate": 7.731998600644769e-08,
      "loss": 0.6683,
      "step": 5924
    },
    {
      "epoch": 0.5857202876702172,
      "grad_norm": 4.384844992152021,
      "learning_rate": 7.728880047215982e-08,
      "loss": 0.8397,
      "step": 5925
    },
    {
      "epoch": 0.5858191434149717,
      "grad_norm": 4.496139682364623,
      "learning_rate": 7.725761726657044e-08,
      "loss": 0.7454,
      "step": 5926
    },
    {
      "epoch": 0.5859179991597262,
      "grad_norm": 4.725576037748162,
      "learning_rate": 7.722643639287698e-08,
      "loss": 0.7541,
      "step": 5927
    },
    {
      "epoch": 0.5860168549044806,
      "grad_norm": 3.679826199351501,
      "learning_rate": 7.719525785427647e-08,
      "loss": 0.7234,
      "step": 5928
    },
    {
      "epoch": 0.5861157106492351,
      "grad_norm": 4.880051339244408,
      "learning_rate": 7.716408165396593e-08,
      "loss": 0.7361,
      "step": 5929
    },
    {
      "epoch": 0.5862145663939896,
      "grad_norm": 3.063055984799968,
      "learning_rate": 7.713290779514192e-08,
      "loss": 0.68,
      "step": 5930
    },
    {
      "epoch": 0.586313422138744,
      "grad_norm": 10.723015931617267,
      "learning_rate": 7.71017362810009e-08,
      "loss": 0.6356,
      "step": 5931
    },
    {
      "epoch": 0.5864122778834985,
      "grad_norm": 4.330015750058091,
      "learning_rate": 7.707056711473909e-08,
      "loss": 0.6989,
      "step": 5932
    },
    {
      "epoch": 0.586511133628253,
      "grad_norm": 4.733451645488777,
      "learning_rate": 7.703940029955236e-08,
      "loss": 0.6727,
      "step": 5933
    },
    {
      "epoch": 0.5866099893730075,
      "grad_norm": 3.3419036738634706,
      "learning_rate": 7.700823583863644e-08,
      "loss": 0.6886,
      "step": 5934
    },
    {
      "epoch": 0.5867088451177619,
      "grad_norm": 4.831015775667677,
      "learning_rate": 7.697707373518683e-08,
      "loss": 0.8237,
      "step": 5935
    },
    {
      "epoch": 0.5868077008625163,
      "grad_norm": 2.641889115851129,
      "learning_rate": 7.694591399239862e-08,
      "loss": 0.6404,
      "step": 5936
    },
    {
      "epoch": 0.5869065566072709,
      "grad_norm": 3.9799969432940308,
      "learning_rate": 7.69147566134669e-08,
      "loss": 0.7062,
      "step": 5937
    },
    {
      "epoch": 0.5870054123520253,
      "grad_norm": 4.045138190323958,
      "learning_rate": 7.688360160158635e-08,
      "loss": 0.7893,
      "step": 5938
    },
    {
      "epoch": 0.5871042680967797,
      "grad_norm": 5.702411303256432,
      "learning_rate": 7.685244895995147e-08,
      "loss": 0.7391,
      "step": 5939
    },
    {
      "epoch": 0.5872031238415343,
      "grad_norm": 6.700297380896003,
      "learning_rate": 7.682129869175651e-08,
      "loss": 0.7123,
      "step": 5940
    },
    {
      "epoch": 0.5873019795862887,
      "grad_norm": 9.020802825149772,
      "learning_rate": 7.679015080019543e-08,
      "loss": 0.7272,
      "step": 5941
    },
    {
      "epoch": 0.5874008353310431,
      "grad_norm": 4.300161118822472,
      "learning_rate": 7.675900528846203e-08,
      "loss": 0.7019,
      "step": 5942
    },
    {
      "epoch": 0.5874996910757977,
      "grad_norm": 3.105683134889722,
      "learning_rate": 7.672786215974982e-08,
      "loss": 0.6221,
      "step": 5943
    },
    {
      "epoch": 0.5875985468205521,
      "grad_norm": 4.028657347614808,
      "learning_rate": 7.669672141725205e-08,
      "loss": 0.6677,
      "step": 5944
    },
    {
      "epoch": 0.5876974025653066,
      "grad_norm": 5.894485685237364,
      "learning_rate": 7.666558306416173e-08,
      "loss": 0.6233,
      "step": 5945
    },
    {
      "epoch": 0.587796258310061,
      "grad_norm": 6.225182708962404,
      "learning_rate": 7.66344471036717e-08,
      "loss": 0.7306,
      "step": 5946
    },
    {
      "epoch": 0.5878951140548155,
      "grad_norm": 3.196215573795415,
      "learning_rate": 7.660331353897441e-08,
      "loss": 0.6812,
      "step": 5947
    },
    {
      "epoch": 0.58799396979957,
      "grad_norm": 2.933671636110435,
      "learning_rate": 7.65721823732622e-08,
      "loss": 0.6708,
      "step": 5948
    },
    {
      "epoch": 0.5880928255443244,
      "grad_norm": 3.508713392903397,
      "learning_rate": 7.654105360972714e-08,
      "loss": 0.8536,
      "step": 5949
    },
    {
      "epoch": 0.588191681289079,
      "grad_norm": 3.4772505466392833,
      "learning_rate": 7.650992725156095e-08,
      "loss": 0.7136,
      "step": 5950
    },
    {
      "epoch": 0.5882905370338334,
      "grad_norm": 3.1537476026561944,
      "learning_rate": 7.647880330195525e-08,
      "loss": 0.6939,
      "step": 5951
    },
    {
      "epoch": 0.5883893927785878,
      "grad_norm": 3.1256520660546063,
      "learning_rate": 7.64476817641013e-08,
      "loss": 0.7236,
      "step": 5952
    },
    {
      "epoch": 0.5884882485233424,
      "grad_norm": 4.7457617222063515,
      "learning_rate": 7.641656264119013e-08,
      "loss": 0.5965,
      "step": 5953
    },
    {
      "epoch": 0.5885871042680968,
      "grad_norm": 3.5011957111357597,
      "learning_rate": 7.638544593641262e-08,
      "loss": 0.6852,
      "step": 5954
    },
    {
      "epoch": 0.5886859600128512,
      "grad_norm": 6.842915958176698,
      "learning_rate": 7.635433165295927e-08,
      "loss": 0.5828,
      "step": 5955
    },
    {
      "epoch": 0.5887848157576057,
      "grad_norm": 3.143601467512425,
      "learning_rate": 7.63232197940204e-08,
      "loss": 0.72,
      "step": 5956
    },
    {
      "epoch": 0.5888836715023602,
      "grad_norm": 9.6878131069622,
      "learning_rate": 7.629211036278612e-08,
      "loss": 0.7056,
      "step": 5957
    },
    {
      "epoch": 0.5889825272471146,
      "grad_norm": 21.119867018598832,
      "learning_rate": 7.626100336244617e-08,
      "loss": 0.7668,
      "step": 5958
    },
    {
      "epoch": 0.5890813829918691,
      "grad_norm": 4.86522905840004,
      "learning_rate": 7.622989879619017e-08,
      "loss": 0.7268,
      "step": 5959
    },
    {
      "epoch": 0.5891802387366236,
      "grad_norm": 3.746260723606131,
      "learning_rate": 7.619879666720741e-08,
      "loss": 0.8577,
      "step": 5960
    },
    {
      "epoch": 0.589279094481378,
      "grad_norm": 4.538439354328951,
      "learning_rate": 7.616769697868697e-08,
      "loss": 0.7946,
      "step": 5961
    },
    {
      "epoch": 0.5893779502261325,
      "grad_norm": 3.7909250546936484,
      "learning_rate": 7.613659973381768e-08,
      "loss": 0.709,
      "step": 5962
    },
    {
      "epoch": 0.589476805970887,
      "grad_norm": 3.727142886436844,
      "learning_rate": 7.610550493578802e-08,
      "loss": 0.7223,
      "step": 5963
    },
    {
      "epoch": 0.5895756617156415,
      "grad_norm": 3.7323233324712755,
      "learning_rate": 7.607441258778635e-08,
      "loss": 0.7481,
      "step": 5964
    },
    {
      "epoch": 0.5896745174603959,
      "grad_norm": 6.502522920904786,
      "learning_rate": 7.60433226930008e-08,
      "loss": 0.6534,
      "step": 5965
    },
    {
      "epoch": 0.5897733732051503,
      "grad_norm": 6.205588024023494,
      "learning_rate": 7.601223525461908e-08,
      "loss": 0.6618,
      "step": 5966
    },
    {
      "epoch": 0.5898722289499049,
      "grad_norm": 3.613633261360325,
      "learning_rate": 7.598115027582879e-08,
      "loss": 0.6338,
      "step": 5967
    },
    {
      "epoch": 0.5899710846946593,
      "grad_norm": 5.074091465929371,
      "learning_rate": 7.595006775981727e-08,
      "loss": 0.7512,
      "step": 5968
    },
    {
      "epoch": 0.5900699404394137,
      "grad_norm": 4.354136987998412,
      "learning_rate": 7.591898770977153e-08,
      "loss": 0.6967,
      "step": 5969
    },
    {
      "epoch": 0.5901687961841683,
      "grad_norm": 4.053908132522949,
      "learning_rate": 7.588791012887835e-08,
      "loss": 0.771,
      "step": 5970
    },
    {
      "epoch": 0.5902676519289227,
      "grad_norm": 3.2175591927237868,
      "learning_rate": 7.585683502032436e-08,
      "loss": 0.6931,
      "step": 5971
    },
    {
      "epoch": 0.5903665076736772,
      "grad_norm": 6.484221934247272,
      "learning_rate": 7.582576238729576e-08,
      "loss": 0.6761,
      "step": 5972
    },
    {
      "epoch": 0.5904653634184317,
      "grad_norm": 4.258239134911527,
      "learning_rate": 7.579469223297867e-08,
      "loss": 0.747,
      "step": 5973
    },
    {
      "epoch": 0.5905642191631861,
      "grad_norm": 4.72034245351533,
      "learning_rate": 7.576362456055878e-08,
      "loss": 0.7118,
      "step": 5974
    },
    {
      "epoch": 0.5906630749079406,
      "grad_norm": 6.336388888693745,
      "learning_rate": 7.573255937322169e-08,
      "loss": 0.814,
      "step": 5975
    },
    {
      "epoch": 0.590761930652695,
      "grad_norm": 5.02855401409818,
      "learning_rate": 7.570149667415273e-08,
      "loss": 0.6643,
      "step": 5976
    },
    {
      "epoch": 0.5908607863974495,
      "grad_norm": 4.39796329716832,
      "learning_rate": 7.56704364665368e-08,
      "loss": 0.6095,
      "step": 5977
    },
    {
      "epoch": 0.590959642142204,
      "grad_norm": 5.608328426287897,
      "learning_rate": 7.563937875355873e-08,
      "loss": 0.7317,
      "step": 5978
    },
    {
      "epoch": 0.5910584978869584,
      "grad_norm": 3.469008190547643,
      "learning_rate": 7.560832353840306e-08,
      "loss": 0.7053,
      "step": 5979
    },
    {
      "epoch": 0.591157353631713,
      "grad_norm": 5.366534365916716,
      "learning_rate": 7.557727082425397e-08,
      "loss": 0.7029,
      "step": 5980
    },
    {
      "epoch": 0.5912562093764674,
      "grad_norm": 4.610105854821648,
      "learning_rate": 7.55462206142955e-08,
      "loss": 0.7367,
      "step": 5981
    },
    {
      "epoch": 0.5913550651212218,
      "grad_norm": 3.8365163859550773,
      "learning_rate": 7.551517291171142e-08,
      "loss": 0.7192,
      "step": 5982
    },
    {
      "epoch": 0.5914539208659764,
      "grad_norm": 4.366245778155662,
      "learning_rate": 7.548412771968513e-08,
      "loss": 0.7876,
      "step": 5983
    },
    {
      "epoch": 0.5915527766107308,
      "grad_norm": 4.0856823760413326,
      "learning_rate": 7.545308504139996e-08,
      "loss": 0.7759,
      "step": 5984
    },
    {
      "epoch": 0.5916516323554852,
      "grad_norm": 4.780251662975842,
      "learning_rate": 7.542204488003878e-08,
      "loss": 0.754,
      "step": 5985
    },
    {
      "epoch": 0.5917504881002398,
      "grad_norm": 2.6386679171344887,
      "learning_rate": 7.539100723878437e-08,
      "loss": 0.6029,
      "step": 5986
    },
    {
      "epoch": 0.5918493438449942,
      "grad_norm": 3.1868580309186982,
      "learning_rate": 7.535997212081918e-08,
      "loss": 0.6794,
      "step": 5987
    },
    {
      "epoch": 0.5919481995897486,
      "grad_norm": 6.70439307677369,
      "learning_rate": 7.532893952932536e-08,
      "loss": 0.7758,
      "step": 5988
    },
    {
      "epoch": 0.5920470553345031,
      "grad_norm": 4.567660245960072,
      "learning_rate": 7.529790946748486e-08,
      "loss": 0.6809,
      "step": 5989
    },
    {
      "epoch": 0.5921459110792576,
      "grad_norm": 3.892820659011896,
      "learning_rate": 7.526688193847939e-08,
      "loss": 0.6416,
      "step": 5990
    },
    {
      "epoch": 0.5922447668240121,
      "grad_norm": 4.293167158478321,
      "learning_rate": 7.52358569454903e-08,
      "loss": 0.7329,
      "step": 5991
    },
    {
      "epoch": 0.5923436225687665,
      "grad_norm": 3.960476994924039,
      "learning_rate": 7.520483449169879e-08,
      "loss": 0.7759,
      "step": 5992
    },
    {
      "epoch": 0.592442478313521,
      "grad_norm": 40.77267531544019,
      "learning_rate": 7.517381458028578e-08,
      "loss": 0.6951,
      "step": 5993
    },
    {
      "epoch": 0.5925413340582755,
      "grad_norm": 11.454351494669874,
      "learning_rate": 7.514279721443183e-08,
      "loss": 0.7291,
      "step": 5994
    },
    {
      "epoch": 0.5926401898030299,
      "grad_norm": 3.3031246225380806,
      "learning_rate": 7.511178239731741e-08,
      "loss": 0.6393,
      "step": 5995
    },
    {
      "epoch": 0.5927390455477844,
      "grad_norm": 3.6377100536063467,
      "learning_rate": 7.508077013212256e-08,
      "loss": 0.7855,
      "step": 5996
    },
    {
      "epoch": 0.5928379012925389,
      "grad_norm": 7.980435991727292,
      "learning_rate": 7.504976042202712e-08,
      "loss": 0.6373,
      "step": 5997
    },
    {
      "epoch": 0.5929367570372933,
      "grad_norm": 3.992715987286459,
      "learning_rate": 7.501875327021075e-08,
      "loss": 0.6206,
      "step": 5998
    },
    {
      "epoch": 0.5930356127820477,
      "grad_norm": 13.002390362681936,
      "learning_rate": 7.498774867985269e-08,
      "loss": 0.7444,
      "step": 5999
    },
    {
      "epoch": 0.5931344685268023,
      "grad_norm": 3.678034102743296,
      "learning_rate": 7.495674665413204e-08,
      "loss": 0.7966,
      "step": 6000
    },
    {
      "epoch": 0.5932333242715567,
      "grad_norm": 2.958795622155242,
      "learning_rate": 7.492574719622766e-08,
      "loss": 0.7263,
      "step": 6001
    },
    {
      "epoch": 0.5933321800163112,
      "grad_norm": 6.6265387940336575,
      "learning_rate": 7.489475030931797e-08,
      "loss": 0.7132,
      "step": 6002
    },
    {
      "epoch": 0.5934310357610657,
      "grad_norm": 5.179298312076587,
      "learning_rate": 7.486375599658135e-08,
      "loss": 0.6492,
      "step": 6003
    },
    {
      "epoch": 0.5935298915058201,
      "grad_norm": 5.077244939220825,
      "learning_rate": 7.483276426119577e-08,
      "loss": 0.7077,
      "step": 6004
    },
    {
      "epoch": 0.5936287472505746,
      "grad_norm": 3.634273306199684,
      "learning_rate": 7.480177510633897e-08,
      "loss": 0.6865,
      "step": 6005
    },
    {
      "epoch": 0.5937276029953291,
      "grad_norm": 3.9029252169169286,
      "learning_rate": 7.477078853518844e-08,
      "loss": 0.6881,
      "step": 6006
    },
    {
      "epoch": 0.5938264587400836,
      "grad_norm": 4.843816493424362,
      "learning_rate": 7.473980455092134e-08,
      "loss": 0.69,
      "step": 6007
    },
    {
      "epoch": 0.593925314484838,
      "grad_norm": 3.124920780123082,
      "learning_rate": 7.470882315671469e-08,
      "loss": 0.7757,
      "step": 6008
    },
    {
      "epoch": 0.5940241702295924,
      "grad_norm": 3.6662147510433343,
      "learning_rate": 7.46778443557452e-08,
      "loss": 0.6135,
      "step": 6009
    },
    {
      "epoch": 0.594123025974347,
      "grad_norm": 4.831140780655563,
      "learning_rate": 7.464686815118919e-08,
      "loss": 0.6818,
      "step": 6010
    },
    {
      "epoch": 0.5942218817191014,
      "grad_norm": 8.16578223054908,
      "learning_rate": 7.461589454622285e-08,
      "loss": 0.6626,
      "step": 6011
    },
    {
      "epoch": 0.5943207374638558,
      "grad_norm": 2.702353838837334,
      "learning_rate": 7.458492354402214e-08,
      "loss": 0.6677,
      "step": 6012
    },
    {
      "epoch": 0.5944195932086104,
      "grad_norm": 5.101901862370919,
      "learning_rate": 7.455395514776261e-08,
      "loss": 0.7578,
      "step": 6013
    },
    {
      "epoch": 0.5945184489533648,
      "grad_norm": 2.9987383287034324,
      "learning_rate": 7.45229893606196e-08,
      "loss": 0.6367,
      "step": 6014
    },
    {
      "epoch": 0.5946173046981192,
      "grad_norm": 4.0480747926596115,
      "learning_rate": 7.449202618576827e-08,
      "loss": 0.6138,
      "step": 6015
    },
    {
      "epoch": 0.5947161604428738,
      "grad_norm": 4.593445423636252,
      "learning_rate": 7.446106562638332e-08,
      "loss": 0.6688,
      "step": 6016
    },
    {
      "epoch": 0.5948150161876282,
      "grad_norm": 3.0311974093111504,
      "learning_rate": 7.443010768563942e-08,
      "loss": 0.7457,
      "step": 6017
    },
    {
      "epoch": 0.5949138719323827,
      "grad_norm": 17.06387839503038,
      "learning_rate": 7.439915236671075e-08,
      "loss": 0.6977,
      "step": 6018
    },
    {
      "epoch": 0.5950127276771371,
      "grad_norm": 3.5923556848685863,
      "learning_rate": 7.436819967277135e-08,
      "loss": 0.7306,
      "step": 6019
    },
    {
      "epoch": 0.5951115834218916,
      "grad_norm": 3.574045115704137,
      "learning_rate": 7.433724960699504e-08,
      "loss": 0.6723,
      "step": 6020
    },
    {
      "epoch": 0.5952104391666461,
      "grad_norm": 5.1793874012375625,
      "learning_rate": 7.430630217255518e-08,
      "loss": 0.6239,
      "step": 6021
    },
    {
      "epoch": 0.5953092949114005,
      "grad_norm": 3.355606893562129,
      "learning_rate": 7.427535737262505e-08,
      "loss": 0.7383,
      "step": 6022
    },
    {
      "epoch": 0.595408150656155,
      "grad_norm": 2.9126303612879925,
      "learning_rate": 7.424441521037757e-08,
      "loss": 0.7774,
      "step": 6023
    },
    {
      "epoch": 0.5955070064009095,
      "grad_norm": 4.088861054070277,
      "learning_rate": 7.421347568898535e-08,
      "loss": 0.7289,
      "step": 6024
    },
    {
      "epoch": 0.5956058621456639,
      "grad_norm": 4.261131713838084,
      "learning_rate": 7.418253881162082e-08,
      "loss": 0.6978,
      "step": 6025
    },
    {
      "epoch": 0.5957047178904185,
      "grad_norm": 3.8383997988007614,
      "learning_rate": 7.415160458145615e-08,
      "loss": 0.5824,
      "step": 6026
    },
    {
      "epoch": 0.5958035736351729,
      "grad_norm": 3.632572858001046,
      "learning_rate": 7.41206730016631e-08,
      "loss": 0.772,
      "step": 6027
    },
    {
      "epoch": 0.5959024293799273,
      "grad_norm": 4.073657314247237,
      "learning_rate": 7.408974407541326e-08,
      "loss": 0.7002,
      "step": 6028
    },
    {
      "epoch": 0.5960012851246818,
      "grad_norm": 3.486117115901315,
      "learning_rate": 7.405881780587802e-08,
      "loss": 0.6797,
      "step": 6029
    },
    {
      "epoch": 0.5961001408694363,
      "grad_norm": 31.092973767989847,
      "learning_rate": 7.402789419622834e-08,
      "loss": 0.7706,
      "step": 6030
    },
    {
      "epoch": 0.5961989966141907,
      "grad_norm": 3.332056725205519,
      "learning_rate": 7.399697324963499e-08,
      "loss": 0.6799,
      "step": 6031
    },
    {
      "epoch": 0.5962978523589452,
      "grad_norm": 6.689025615370689,
      "learning_rate": 7.396605496926847e-08,
      "loss": 0.6381,
      "step": 6032
    },
    {
      "epoch": 0.5963967081036997,
      "grad_norm": 4.318722212419483,
      "learning_rate": 7.393513935829894e-08,
      "loss": 0.633,
      "step": 6033
    },
    {
      "epoch": 0.5964955638484541,
      "grad_norm": 3.796785721419977,
      "learning_rate": 7.390422641989644e-08,
      "loss": 0.7607,
      "step": 6034
    },
    {
      "epoch": 0.5965944195932086,
      "grad_norm": 3.8894743573572494,
      "learning_rate": 7.387331615723054e-08,
      "loss": 0.7246,
      "step": 6035
    },
    {
      "epoch": 0.5966932753379631,
      "grad_norm": 5.40320474344557,
      "learning_rate": 7.384240857347068e-08,
      "loss": 0.6774,
      "step": 6036
    },
    {
      "epoch": 0.5967921310827176,
      "grad_norm": 6.688019270873335,
      "learning_rate": 7.381150367178601e-08,
      "loss": 0.632,
      "step": 6037
    },
    {
      "epoch": 0.596890986827472,
      "grad_norm": 8.552557628329735,
      "learning_rate": 7.378060145534528e-08,
      "loss": 0.659,
      "step": 6038
    },
    {
      "epoch": 0.5969898425722264,
      "grad_norm": 3.2469217460299755,
      "learning_rate": 7.374970192731713e-08,
      "loss": 0.7284,
      "step": 6039
    },
    {
      "epoch": 0.597088698316981,
      "grad_norm": 6.832405342256464,
      "learning_rate": 7.371880509086985e-08,
      "loss": 0.6967,
      "step": 6040
    },
    {
      "epoch": 0.5971875540617354,
      "grad_norm": 5.025383214870353,
      "learning_rate": 7.368791094917139e-08,
      "loss": 0.7521,
      "step": 6041
    },
    {
      "epoch": 0.5972864098064898,
      "grad_norm": 4.842876624402265,
      "learning_rate": 7.365701950538958e-08,
      "loss": 0.719,
      "step": 6042
    },
    {
      "epoch": 0.5973852655512444,
      "grad_norm": 3.7835024575051865,
      "learning_rate": 7.36261307626918e-08,
      "loss": 0.5807,
      "step": 6043
    },
    {
      "epoch": 0.5974841212959988,
      "grad_norm": 3.899651666319529,
      "learning_rate": 7.359524472424525e-08,
      "loss": 0.7377,
      "step": 6044
    },
    {
      "epoch": 0.5975829770407532,
      "grad_norm": 4.61916050442917,
      "learning_rate": 7.35643613932169e-08,
      "loss": 0.6896,
      "step": 6045
    },
    {
      "epoch": 0.5976818327855078,
      "grad_norm": 6.6308882872472745,
      "learning_rate": 7.35334807727733e-08,
      "loss": 0.7051,
      "step": 6046
    },
    {
      "epoch": 0.5977806885302622,
      "grad_norm": 3.1473779658040635,
      "learning_rate": 7.350260286608086e-08,
      "loss": 0.7668,
      "step": 6047
    },
    {
      "epoch": 0.5978795442750167,
      "grad_norm": 4.984498897589573,
      "learning_rate": 7.347172767630562e-08,
      "loss": 0.6651,
      "step": 6048
    },
    {
      "epoch": 0.5979784000197711,
      "grad_norm": 7.775846736170326,
      "learning_rate": 7.34408552066134e-08,
      "loss": 0.6327,
      "step": 6049
    },
    {
      "epoch": 0.5980772557645256,
      "grad_norm": 7.0297045816479615,
      "learning_rate": 7.340998546016967e-08,
      "loss": 0.6666,
      "step": 6050
    },
    {
      "epoch": 0.5981761115092801,
      "grad_norm": 2.892086370735846,
      "learning_rate": 7.337911844013975e-08,
      "loss": 0.7016,
      "step": 6051
    },
    {
      "epoch": 0.5982749672540345,
      "grad_norm": 7.5094717632948536,
      "learning_rate": 7.334825414968852e-08,
      "loss": 0.8374,
      "step": 6052
    },
    {
      "epoch": 0.598373822998789,
      "grad_norm": 5.726785843702308,
      "learning_rate": 7.331739259198071e-08,
      "loss": 0.7381,
      "step": 6053
    },
    {
      "epoch": 0.5984726787435435,
      "grad_norm": 3.930157864628389,
      "learning_rate": 7.328653377018066e-08,
      "loss": 0.7291,
      "step": 6054
    },
    {
      "epoch": 0.5985715344882979,
      "grad_norm": 6.819135986973994,
      "learning_rate": 7.32556776874525e-08,
      "loss": 0.7123,
      "step": 6055
    },
    {
      "epoch": 0.5986703902330525,
      "grad_norm": 3.227003923682198,
      "learning_rate": 7.322482434696017e-08,
      "loss": 0.6966,
      "step": 6056
    },
    {
      "epoch": 0.5987692459778069,
      "grad_norm": 3.265060805047526,
      "learning_rate": 7.319397375186708e-08,
      "loss": 0.6709,
      "step": 6057
    },
    {
      "epoch": 0.5988681017225613,
      "grad_norm": 5.405244423477094,
      "learning_rate": 7.316312590533658e-08,
      "loss": 0.6975,
      "step": 6058
    },
    {
      "epoch": 0.5989669574673159,
      "grad_norm": 2.920350681249926,
      "learning_rate": 7.313228081053167e-08,
      "loss": 0.7587,
      "step": 6059
    },
    {
      "epoch": 0.5990658132120703,
      "grad_norm": 3.438678371445836,
      "learning_rate": 7.310143847061499e-08,
      "loss": 0.6443,
      "step": 6060
    },
    {
      "epoch": 0.5991646689568247,
      "grad_norm": 6.315888802415208,
      "learning_rate": 7.307059888874902e-08,
      "loss": 0.7457,
      "step": 6061
    },
    {
      "epoch": 0.5992635247015792,
      "grad_norm": 3.3929897080859712,
      "learning_rate": 7.303976206809594e-08,
      "loss": 0.7628,
      "step": 6062
    },
    {
      "epoch": 0.5993623804463337,
      "grad_norm": 4.98229535966313,
      "learning_rate": 7.30089280118175e-08,
      "loss": 0.6978,
      "step": 6063
    },
    {
      "epoch": 0.5994612361910882,
      "grad_norm": 3.240583751948194,
      "learning_rate": 7.297809672307542e-08,
      "loss": 0.7849,
      "step": 6064
    },
    {
      "epoch": 0.5995600919358426,
      "grad_norm": 4.195874643109401,
      "learning_rate": 7.294726820503088e-08,
      "loss": 0.7559,
      "step": 6065
    },
    {
      "epoch": 0.5996589476805971,
      "grad_norm": 12.692466151883247,
      "learning_rate": 7.291644246084494e-08,
      "loss": 0.6588,
      "step": 6066
    },
    {
      "epoch": 0.5997578034253516,
      "grad_norm": 3.4396118437224246,
      "learning_rate": 7.288561949367833e-08,
      "loss": 0.7803,
      "step": 6067
    },
    {
      "epoch": 0.599856659170106,
      "grad_norm": 5.256114037161938,
      "learning_rate": 7.285479930669144e-08,
      "loss": 0.8087,
      "step": 6068
    },
    {
      "epoch": 0.5999555149148605,
      "grad_norm": 3.5339806913882286,
      "learning_rate": 7.282398190304446e-08,
      "loss": 0.7731,
      "step": 6069
    },
    {
      "epoch": 0.600054370659615,
      "grad_norm": 3.285794909785628,
      "learning_rate": 7.27931672858973e-08,
      "loss": 0.6149,
      "step": 6070
    },
    {
      "epoch": 0.6001532264043694,
      "grad_norm": 3.726336648155763,
      "learning_rate": 7.276235545840947e-08,
      "loss": 0.7443,
      "step": 6071
    },
    {
      "epoch": 0.6002520821491238,
      "grad_norm": 2.8162038205518853,
      "learning_rate": 7.273154642374032e-08,
      "loss": 0.7624,
      "step": 6072
    },
    {
      "epoch": 0.6003509378938784,
      "grad_norm": 4.067303009874905,
      "learning_rate": 7.270074018504886e-08,
      "loss": 0.6704,
      "step": 6073
    },
    {
      "epoch": 0.6004497936386328,
      "grad_norm": 3.49053605977276,
      "learning_rate": 7.266993674549378e-08,
      "loss": 0.6122,
      "step": 6074
    },
    {
      "epoch": 0.6005486493833873,
      "grad_norm": 3.3485262962771953,
      "learning_rate": 7.263913610823356e-08,
      "loss": 0.6824,
      "step": 6075
    },
    {
      "epoch": 0.6006475051281418,
      "grad_norm": 3.654108547114296,
      "learning_rate": 7.260833827642632e-08,
      "loss": 0.7043,
      "step": 6076
    },
    {
      "epoch": 0.6007463608728962,
      "grad_norm": 3.7112913621750745,
      "learning_rate": 7.257754325322991e-08,
      "loss": 0.6876,
      "step": 6077
    },
    {
      "epoch": 0.6008452166176507,
      "grad_norm": 2.912053091827616,
      "learning_rate": 7.254675104180197e-08,
      "loss": 0.6353,
      "step": 6078
    },
    {
      "epoch": 0.6009440723624052,
      "grad_norm": 3.5679808672419577,
      "learning_rate": 7.251596164529972e-08,
      "loss": 0.6586,
      "step": 6079
    },
    {
      "epoch": 0.6010429281071596,
      "grad_norm": 3.2678992817610526,
      "learning_rate": 7.248517506688016e-08,
      "loss": 0.6901,
      "step": 6080
    },
    {
      "epoch": 0.6011417838519141,
      "grad_norm": 4.405056440417408,
      "learning_rate": 7.24543913097001e-08,
      "loss": 0.7062,
      "step": 6081
    },
    {
      "epoch": 0.6012406395966685,
      "grad_norm": 7.323425730250339,
      "learning_rate": 7.242361037691582e-08,
      "loss": 0.7787,
      "step": 6082
    },
    {
      "epoch": 0.6013394953414231,
      "grad_norm": 6.79452911100599,
      "learning_rate": 7.239283227168355e-08,
      "loss": 0.6341,
      "step": 6083
    },
    {
      "epoch": 0.6014383510861775,
      "grad_norm": 5.561736958860742,
      "learning_rate": 7.236205699715911e-08,
      "loss": 0.5904,
      "step": 6084
    },
    {
      "epoch": 0.6015372068309319,
      "grad_norm": 3.22113669895406,
      "learning_rate": 7.2331284556498e-08,
      "loss": 0.6263,
      "step": 6085
    },
    {
      "epoch": 0.6016360625756865,
      "grad_norm": 4.40371186059166,
      "learning_rate": 7.230051495285554e-08,
      "loss": 0.6635,
      "step": 6086
    },
    {
      "epoch": 0.6017349183204409,
      "grad_norm": 6.9142193971938335,
      "learning_rate": 7.226974818938665e-08,
      "loss": 0.8103,
      "step": 6087
    },
    {
      "epoch": 0.6018337740651953,
      "grad_norm": 4.3490551901122645,
      "learning_rate": 7.223898426924603e-08,
      "loss": 0.7195,
      "step": 6088
    },
    {
      "epoch": 0.6019326298099499,
      "grad_norm": 5.581394427827968,
      "learning_rate": 7.220822319558811e-08,
      "loss": 0.6964,
      "step": 6089
    },
    {
      "epoch": 0.6020314855547043,
      "grad_norm": 3.408776865121762,
      "learning_rate": 7.21774649715669e-08,
      "loss": 0.5861,
      "step": 6090
    },
    {
      "epoch": 0.6021303412994587,
      "grad_norm": 4.858221083729485,
      "learning_rate": 7.214670960033626e-08,
      "loss": 0.7265,
      "step": 6091
    },
    {
      "epoch": 0.6022291970442132,
      "grad_norm": 3.143802770360347,
      "learning_rate": 7.21159570850497e-08,
      "loss": 0.749,
      "step": 6092
    },
    {
      "epoch": 0.6023280527889677,
      "grad_norm": 6.191496082532079,
      "learning_rate": 7.208520742886041e-08,
      "loss": 0.6746,
      "step": 6093
    },
    {
      "epoch": 0.6024269085337222,
      "grad_norm": 3.6751360032162403,
      "learning_rate": 7.205446063492133e-08,
      "loss": 0.6748,
      "step": 6094
    },
    {
      "epoch": 0.6025257642784766,
      "grad_norm": 7.100666803967072,
      "learning_rate": 7.202371670638511e-08,
      "loss": 0.7121,
      "step": 6095
    },
    {
      "epoch": 0.6026246200232311,
      "grad_norm": 4.749101269619705,
      "learning_rate": 7.199297564640401e-08,
      "loss": 0.6581,
      "step": 6096
    },
    {
      "epoch": 0.6027234757679856,
      "grad_norm": 3.8125619985215,
      "learning_rate": 7.196223745813018e-08,
      "loss": 0.6857,
      "step": 6097
    },
    {
      "epoch": 0.60282233151274,
      "grad_norm": 6.779419157432558,
      "learning_rate": 7.193150214471527e-08,
      "loss": 0.6956,
      "step": 6098
    },
    {
      "epoch": 0.6029211872574946,
      "grad_norm": 12.142110300253409,
      "learning_rate": 7.190076970931079e-08,
      "loss": 0.6761,
      "step": 6099
    },
    {
      "epoch": 0.603020043002249,
      "grad_norm": 4.36426237816225,
      "learning_rate": 7.18700401550679e-08,
      "loss": 0.7272,
      "step": 6100
    },
    {
      "epoch": 0.6031188987470034,
      "grad_norm": 4.564992063919095,
      "learning_rate": 7.183931348513743e-08,
      "loss": 0.7627,
      "step": 6101
    },
    {
      "epoch": 0.6032177544917579,
      "grad_norm": 4.93146704427161,
      "learning_rate": 7.180858970266994e-08,
      "loss": 0.7398,
      "step": 6102
    },
    {
      "epoch": 0.6033166102365124,
      "grad_norm": 3.550226324272571,
      "learning_rate": 7.177786881081578e-08,
      "loss": 0.707,
      "step": 6103
    },
    {
      "epoch": 0.6034154659812668,
      "grad_norm": 3.183475386004774,
      "learning_rate": 7.174715081272482e-08,
      "loss": 0.7346,
      "step": 6104
    },
    {
      "epoch": 0.6035143217260213,
      "grad_norm": 4.779806380283616,
      "learning_rate": 7.171643571154678e-08,
      "loss": 0.7834,
      "step": 6105
    },
    {
      "epoch": 0.6036131774707758,
      "grad_norm": 4.951433934543384,
      "learning_rate": 7.168572351043109e-08,
      "loss": 0.6414,
      "step": 6106
    },
    {
      "epoch": 0.6037120332155302,
      "grad_norm": 10.327166515549669,
      "learning_rate": 7.165501421252675e-08,
      "loss": 0.7062,
      "step": 6107
    },
    {
      "epoch": 0.6038108889602847,
      "grad_norm": 4.056658359667362,
      "learning_rate": 7.162430782098262e-08,
      "loss": 0.6627,
      "step": 6108
    },
    {
      "epoch": 0.6039097447050392,
      "grad_norm": 7.9760866068268825,
      "learning_rate": 7.159360433894711e-08,
      "loss": 0.7664,
      "step": 6109
    },
    {
      "epoch": 0.6040086004497937,
      "grad_norm": 7.309268465093565,
      "learning_rate": 7.156290376956848e-08,
      "loss": 0.7712,
      "step": 6110
    },
    {
      "epoch": 0.6041074561945481,
      "grad_norm": 4.1349689919889,
      "learning_rate": 7.153220611599461e-08,
      "loss": 0.6894,
      "step": 6111
    },
    {
      "epoch": 0.6042063119393025,
      "grad_norm": 8.005353554299786,
      "learning_rate": 7.150151138137304e-08,
      "loss": 0.7137,
      "step": 6112
    },
    {
      "epoch": 0.6043051676840571,
      "grad_norm": 3.6781717545585058,
      "learning_rate": 7.147081956885109e-08,
      "loss": 0.7527,
      "step": 6113
    },
    {
      "epoch": 0.6044040234288115,
      "grad_norm": 3.638303787436359,
      "learning_rate": 7.144013068157579e-08,
      "loss": 0.6572,
      "step": 6114
    },
    {
      "epoch": 0.6045028791735659,
      "grad_norm": 5.377356417554445,
      "learning_rate": 7.140944472269376e-08,
      "loss": 0.842,
      "step": 6115
    },
    {
      "epoch": 0.6046017349183205,
      "grad_norm": 7.240689908400418,
      "learning_rate": 7.137876169535146e-08,
      "loss": 0.7056,
      "step": 6116
    },
    {
      "epoch": 0.6047005906630749,
      "grad_norm": 13.217449244704335,
      "learning_rate": 7.134808160269494e-08,
      "loss": 0.7177,
      "step": 6117
    },
    {
      "epoch": 0.6047994464078293,
      "grad_norm": 6.293442975708147,
      "learning_rate": 7.131740444787003e-08,
      "loss": 0.6685,
      "step": 6118
    },
    {
      "epoch": 0.6048983021525839,
      "grad_norm": 3.9279272260711604,
      "learning_rate": 7.128673023402219e-08,
      "loss": 0.74,
      "step": 6119
    },
    {
      "epoch": 0.6049971578973383,
      "grad_norm": 5.7837658422224925,
      "learning_rate": 7.125605896429661e-08,
      "loss": 0.7253,
      "step": 6120
    },
    {
      "epoch": 0.6050960136420928,
      "grad_norm": 7.679579766788226,
      "learning_rate": 7.122539064183816e-08,
      "loss": 0.7588,
      "step": 6121
    },
    {
      "epoch": 0.6051948693868472,
      "grad_norm": 4.3001770986102335,
      "learning_rate": 7.119472526979149e-08,
      "loss": 0.6279,
      "step": 6122
    },
    {
      "epoch": 0.6052937251316017,
      "grad_norm": 3.014875805161474,
      "learning_rate": 7.116406285130079e-08,
      "loss": 0.6052,
      "step": 6123
    },
    {
      "epoch": 0.6053925808763562,
      "grad_norm": 7.9074998916701595,
      "learning_rate": 7.113340338951008e-08,
      "loss": 0.7348,
      "step": 6124
    },
    {
      "epoch": 0.6054914366211106,
      "grad_norm": 8.708431587075943,
      "learning_rate": 7.110274688756308e-08,
      "loss": 0.6498,
      "step": 6125
    },
    {
      "epoch": 0.6055902923658651,
      "grad_norm": 2.8975018779483053,
      "learning_rate": 7.107209334860309e-08,
      "loss": 0.7577,
      "step": 6126
    },
    {
      "epoch": 0.6056891481106196,
      "grad_norm": 4.482538609304724,
      "learning_rate": 7.104144277577323e-08,
      "loss": 0.765,
      "step": 6127
    },
    {
      "epoch": 0.605788003855374,
      "grad_norm": 3.6435330986647565,
      "learning_rate": 7.101079517221626e-08,
      "loss": 0.8837,
      "step": 6128
    },
    {
      "epoch": 0.6058868596001286,
      "grad_norm": 3.446152494247545,
      "learning_rate": 7.09801505410746e-08,
      "loss": 0.7771,
      "step": 6129
    },
    {
      "epoch": 0.605985715344883,
      "grad_norm": 6.3843022772507485,
      "learning_rate": 7.094950888549045e-08,
      "loss": 0.7145,
      "step": 6130
    },
    {
      "epoch": 0.6060845710896374,
      "grad_norm": 8.279078926085116,
      "learning_rate": 7.091887020860561e-08,
      "loss": 0.6372,
      "step": 6131
    },
    {
      "epoch": 0.6061834268343919,
      "grad_norm": 41.07874655386906,
      "learning_rate": 7.088823451356163e-08,
      "loss": 0.7807,
      "step": 6132
    },
    {
      "epoch": 0.6062822825791464,
      "grad_norm": 16.05306619691092,
      "learning_rate": 7.085760180349984e-08,
      "loss": 0.7045,
      "step": 6133
    },
    {
      "epoch": 0.6063811383239008,
      "grad_norm": 2.9571142301152604,
      "learning_rate": 7.082697208156105e-08,
      "loss": 0.7697,
      "step": 6134
    },
    {
      "epoch": 0.6064799940686553,
      "grad_norm": 8.846272447427229,
      "learning_rate": 7.079634535088597e-08,
      "loss": 0.78,
      "step": 6135
    },
    {
      "epoch": 0.6065788498134098,
      "grad_norm": 4.289469846473578,
      "learning_rate": 7.076572161461488e-08,
      "loss": 0.7307,
      "step": 6136
    },
    {
      "epoch": 0.6066777055581642,
      "grad_norm": 6.136987673804083,
      "learning_rate": 7.073510087588781e-08,
      "loss": 0.7275,
      "step": 6137
    },
    {
      "epoch": 0.6067765613029187,
      "grad_norm": 4.437671437956001,
      "learning_rate": 7.070448313784442e-08,
      "loss": 0.8118,
      "step": 6138
    },
    {
      "epoch": 0.6068754170476732,
      "grad_norm": 4.0954426429409985,
      "learning_rate": 7.067386840362422e-08,
      "loss": 0.6747,
      "step": 6139
    },
    {
      "epoch": 0.6069742727924277,
      "grad_norm": 6.663765971038575,
      "learning_rate": 7.064325667636615e-08,
      "loss": 0.7144,
      "step": 6140
    },
    {
      "epoch": 0.6070731285371821,
      "grad_norm": 7.032556211999101,
      "learning_rate": 7.061264795920914e-08,
      "loss": 0.6316,
      "step": 6141
    },
    {
      "epoch": 0.6071719842819366,
      "grad_norm": 2.8222037034359424,
      "learning_rate": 7.058204225529152e-08,
      "loss": 0.6705,
      "step": 6142
    },
    {
      "epoch": 0.6072708400266911,
      "grad_norm": 5.29213207767586,
      "learning_rate": 7.055143956775155e-08,
      "loss": 0.5493,
      "step": 6143
    },
    {
      "epoch": 0.6073696957714455,
      "grad_norm": 4.443996388712738,
      "learning_rate": 7.052083989972706e-08,
      "loss": 0.7234,
      "step": 6144
    },
    {
      "epoch": 0.6074685515161999,
      "grad_norm": 6.38134241668575,
      "learning_rate": 7.049024325435558e-08,
      "loss": 0.7197,
      "step": 6145
    },
    {
      "epoch": 0.6075674072609545,
      "grad_norm": 4.125585818492529,
      "learning_rate": 7.045964963477433e-08,
      "loss": 0.6818,
      "step": 6146
    },
    {
      "epoch": 0.6076662630057089,
      "grad_norm": 5.460499115585014,
      "learning_rate": 7.042905904412031e-08,
      "loss": 0.6702,
      "step": 6147
    },
    {
      "epoch": 0.6077651187504634,
      "grad_norm": 3.185490521620197,
      "learning_rate": 7.039847148553004e-08,
      "loss": 0.6157,
      "step": 6148
    },
    {
      "epoch": 0.6078639744952179,
      "grad_norm": 8.996580345814403,
      "learning_rate": 7.036788696213986e-08,
      "loss": 0.5816,
      "step": 6149
    },
    {
      "epoch": 0.6079628302399723,
      "grad_norm": 7.563929909720708,
      "learning_rate": 7.033730547708583e-08,
      "loss": 0.8294,
      "step": 6150
    },
    {
      "epoch": 0.6080616859847268,
      "grad_norm": 5.334148136496572,
      "learning_rate": 7.030672703350351e-08,
      "loss": 0.6902,
      "step": 6151
    },
    {
      "epoch": 0.6081605417294813,
      "grad_norm": 3.2007011358107382,
      "learning_rate": 7.027615163452837e-08,
      "loss": 0.6602,
      "step": 6152
    },
    {
      "epoch": 0.6082593974742357,
      "grad_norm": 5.706528930385931,
      "learning_rate": 7.024557928329537e-08,
      "loss": 0.6738,
      "step": 6153
    },
    {
      "epoch": 0.6083582532189902,
      "grad_norm": 3.4503841791674277,
      "learning_rate": 7.021500998293935e-08,
      "loss": 0.6479,
      "step": 6154
    },
    {
      "epoch": 0.6084571089637446,
      "grad_norm": 2.8930884688935015,
      "learning_rate": 7.01844437365947e-08,
      "loss": 0.6789,
      "step": 6155
    },
    {
      "epoch": 0.6085559647084992,
      "grad_norm": 5.0808361526697015,
      "learning_rate": 7.015388054739549e-08,
      "loss": 0.7366,
      "step": 6156
    },
    {
      "epoch": 0.6086548204532536,
      "grad_norm": 3.5930953616493264,
      "learning_rate": 7.012332041847558e-08,
      "loss": 0.6205,
      "step": 6157
    },
    {
      "epoch": 0.608753676198008,
      "grad_norm": 4.217386192704726,
      "learning_rate": 7.009276335296848e-08,
      "loss": 0.618,
      "step": 6158
    },
    {
      "epoch": 0.6088525319427626,
      "grad_norm": 3.637088119307671,
      "learning_rate": 7.006220935400731e-08,
      "loss": 0.6864,
      "step": 6159
    },
    {
      "epoch": 0.608951387687517,
      "grad_norm": 3.208677652873969,
      "learning_rate": 7.003165842472496e-08,
      "loss": 0.7245,
      "step": 6160
    },
    {
      "epoch": 0.6090502434322714,
      "grad_norm": 3.3939628477862995,
      "learning_rate": 7.000111056825399e-08,
      "loss": 0.7726,
      "step": 6161
    },
    {
      "epoch": 0.609149099177026,
      "grad_norm": 8.322265139819734,
      "learning_rate": 6.997056578772661e-08,
      "loss": 0.6964,
      "step": 6162
    },
    {
      "epoch": 0.6092479549217804,
      "grad_norm": 2.990147228259665,
      "learning_rate": 6.994002408627474e-08,
      "loss": 0.5972,
      "step": 6163
    },
    {
      "epoch": 0.6093468106665348,
      "grad_norm": 16.36466765194963,
      "learning_rate": 6.990948546702999e-08,
      "loss": 0.7732,
      "step": 6164
    },
    {
      "epoch": 0.6094456664112893,
      "grad_norm": 5.656507031910142,
      "learning_rate": 6.987894993312362e-08,
      "loss": 0.7719,
      "step": 6165
    },
    {
      "epoch": 0.6095445221560438,
      "grad_norm": 4.387993127505781,
      "learning_rate": 6.984841748768664e-08,
      "loss": 0.7302,
      "step": 6166
    },
    {
      "epoch": 0.6096433779007983,
      "grad_norm": 7.039956825767253,
      "learning_rate": 6.981788813384966e-08,
      "loss": 0.6351,
      "step": 6167
    },
    {
      "epoch": 0.6097422336455527,
      "grad_norm": 4.659087742014038,
      "learning_rate": 6.978736187474302e-08,
      "loss": 0.6614,
      "step": 6168
    },
    {
      "epoch": 0.6098410893903072,
      "grad_norm": 13.172710758951316,
      "learning_rate": 6.97568387134968e-08,
      "loss": 0.6577,
      "step": 6169
    },
    {
      "epoch": 0.6099399451350617,
      "grad_norm": 5.538785065813369,
      "learning_rate": 6.972631865324064e-08,
      "loss": 0.7592,
      "step": 6170
    },
    {
      "epoch": 0.6100388008798161,
      "grad_norm": 2.98933881866553,
      "learning_rate": 6.969580169710389e-08,
      "loss": 0.5631,
      "step": 6171
    },
    {
      "epoch": 0.6101376566245706,
      "grad_norm": 2.8057294563027684,
      "learning_rate": 6.966528784821573e-08,
      "loss": 0.6713,
      "step": 6172
    },
    {
      "epoch": 0.6102365123693251,
      "grad_norm": 4.184486291097229,
      "learning_rate": 6.963477710970477e-08,
      "loss": 0.7819,
      "step": 6173
    },
    {
      "epoch": 0.6103353681140795,
      "grad_norm": 4.14598973494095,
      "learning_rate": 6.960426948469954e-08,
      "loss": 0.6659,
      "step": 6174
    },
    {
      "epoch": 0.610434223858834,
      "grad_norm": 4.442005608966007,
      "learning_rate": 6.957376497632807e-08,
      "loss": 0.7784,
      "step": 6175
    },
    {
      "epoch": 0.6105330796035885,
      "grad_norm": 3.7976578954273865,
      "learning_rate": 6.954326358771818e-08,
      "loss": 0.7583,
      "step": 6176
    },
    {
      "epoch": 0.6106319353483429,
      "grad_norm": 4.50317746368112,
      "learning_rate": 6.95127653219974e-08,
      "loss": 0.7364,
      "step": 6177
    },
    {
      "epoch": 0.6107307910930974,
      "grad_norm": 4.856367276441941,
      "learning_rate": 6.948227018229274e-08,
      "loss": 0.7946,
      "step": 6178
    },
    {
      "epoch": 0.6108296468378519,
      "grad_norm": 3.723820114024291,
      "learning_rate": 6.945177817173116e-08,
      "loss": 0.643,
      "step": 6179
    },
    {
      "epoch": 0.6109285025826063,
      "grad_norm": 6.012898570669482,
      "learning_rate": 6.942128929343909e-08,
      "loss": 0.7586,
      "step": 6180
    },
    {
      "epoch": 0.6110273583273608,
      "grad_norm": 20.252771162776323,
      "learning_rate": 6.939080355054274e-08,
      "loss": 0.651,
      "step": 6181
    },
    {
      "epoch": 0.6111262140721153,
      "grad_norm": 3.7952188613666586,
      "learning_rate": 6.936032094616795e-08,
      "loss": 0.7832,
      "step": 6182
    },
    {
      "epoch": 0.6112250698168697,
      "grad_norm": 4.178351858036433,
      "learning_rate": 6.932984148344033e-08,
      "loss": 0.6827,
      "step": 6183
    },
    {
      "epoch": 0.6113239255616242,
      "grad_norm": 3.0291990160402147,
      "learning_rate": 6.929936516548499e-08,
      "loss": 0.7678,
      "step": 6184
    },
    {
      "epoch": 0.6114227813063786,
      "grad_norm": 4.12973257057786,
      "learning_rate": 6.926889199542694e-08,
      "loss": 0.6995,
      "step": 6185
    },
    {
      "epoch": 0.6115216370511332,
      "grad_norm": 5.107059444909494,
      "learning_rate": 6.923842197639065e-08,
      "loss": 0.6293,
      "step": 6186
    },
    {
      "epoch": 0.6116204927958876,
      "grad_norm": 2.8863484693149295,
      "learning_rate": 6.920795511150045e-08,
      "loss": 0.7457,
      "step": 6187
    },
    {
      "epoch": 0.611719348540642,
      "grad_norm": 7.862364312192056,
      "learning_rate": 6.917749140388023e-08,
      "loss": 0.6754,
      "step": 6188
    },
    {
      "epoch": 0.6118182042853966,
      "grad_norm": 6.156064191160388,
      "learning_rate": 6.914703085665361e-08,
      "loss": 0.6481,
      "step": 6189
    },
    {
      "epoch": 0.611917060030151,
      "grad_norm": 5.336506090112736,
      "learning_rate": 6.911657347294385e-08,
      "loss": 0.6923,
      "step": 6190
    },
    {
      "epoch": 0.6120159157749054,
      "grad_norm": 2.944909938351846,
      "learning_rate": 6.908611925587395e-08,
      "loss": 0.6998,
      "step": 6191
    },
    {
      "epoch": 0.61211477151966,
      "grad_norm": 4.419652773412835,
      "learning_rate": 6.905566820856645e-08,
      "loss": 0.7413,
      "step": 6192
    },
    {
      "epoch": 0.6122136272644144,
      "grad_norm": 3.241371241130809,
      "learning_rate": 6.902522033414373e-08,
      "loss": 0.632,
      "step": 6193
    },
    {
      "epoch": 0.6123124830091689,
      "grad_norm": 4.996806842911084,
      "learning_rate": 6.89947756357278e-08,
      "loss": 0.7204,
      "step": 6194
    },
    {
      "epoch": 0.6124113387539233,
      "grad_norm": 5.205737629638206,
      "learning_rate": 6.896433411644023e-08,
      "loss": 0.7839,
      "step": 6195
    },
    {
      "epoch": 0.6125101944986778,
      "grad_norm": 4.554328849292869,
      "learning_rate": 6.893389577940241e-08,
      "loss": 0.7235,
      "step": 6196
    },
    {
      "epoch": 0.6126090502434323,
      "grad_norm": 9.007461067052628,
      "learning_rate": 6.89034606277353e-08,
      "loss": 0.7391,
      "step": 6197
    },
    {
      "epoch": 0.6127079059881867,
      "grad_norm": 6.962756218713341,
      "learning_rate": 6.887302866455962e-08,
      "loss": 0.6661,
      "step": 6198
    },
    {
      "epoch": 0.6128067617329412,
      "grad_norm": 4.274171958089775,
      "learning_rate": 6.88425998929957e-08,
      "loss": 0.6883,
      "step": 6199
    },
    {
      "epoch": 0.6129056174776957,
      "grad_norm": 11.931845265985965,
      "learning_rate": 6.881217431616352e-08,
      "loss": 0.7206,
      "step": 6200
    },
    {
      "epoch": 0.6130044732224501,
      "grad_norm": 3.092277314735298,
      "learning_rate": 6.87817519371828e-08,
      "loss": 0.7209,
      "step": 6201
    },
    {
      "epoch": 0.6131033289672047,
      "grad_norm": 3.0687625419703397,
      "learning_rate": 6.875133275917298e-08,
      "loss": 0.6177,
      "step": 6202
    },
    {
      "epoch": 0.6132021847119591,
      "grad_norm": 4.062303595073752,
      "learning_rate": 6.872091678525298e-08,
      "loss": 0.7045,
      "step": 6203
    },
    {
      "epoch": 0.6133010404567135,
      "grad_norm": 4.891724917850189,
      "learning_rate": 6.86905040185416e-08,
      "loss": 0.7055,
      "step": 6204
    },
    {
      "epoch": 0.613399896201468,
      "grad_norm": 3.883859304661967,
      "learning_rate": 6.866009446215717e-08,
      "loss": 0.7309,
      "step": 6205
    },
    {
      "epoch": 0.6134987519462225,
      "grad_norm": 3.1274452959711163,
      "learning_rate": 6.862968811921776e-08,
      "loss": 0.7119,
      "step": 6206
    },
    {
      "epoch": 0.6135976076909769,
      "grad_norm": 6.631919120302921,
      "learning_rate": 6.85992849928411e-08,
      "loss": 0.6683,
      "step": 6207
    },
    {
      "epoch": 0.6136964634357314,
      "grad_norm": 3.9032229404521264,
      "learning_rate": 6.856888508614455e-08,
      "loss": 0.6638,
      "step": 6208
    },
    {
      "epoch": 0.6137953191804859,
      "grad_norm": 5.370073232677507,
      "learning_rate": 6.853848840224518e-08,
      "loss": 0.7109,
      "step": 6209
    },
    {
      "epoch": 0.6138941749252403,
      "grad_norm": 4.847396970437173,
      "learning_rate": 6.850809494425979e-08,
      "loss": 0.7659,
      "step": 6210
    },
    {
      "epoch": 0.6139930306699948,
      "grad_norm": 6.606907931608661,
      "learning_rate": 6.847770471530467e-08,
      "loss": 0.6095,
      "step": 6211
    },
    {
      "epoch": 0.6140918864147493,
      "grad_norm": 4.5997332259545844,
      "learning_rate": 6.844731771849595e-08,
      "loss": 0.7539,
      "step": 6212
    },
    {
      "epoch": 0.6141907421595038,
      "grad_norm": 3.5505427503314193,
      "learning_rate": 6.841693395694941e-08,
      "loss": 0.6893,
      "step": 6213
    },
    {
      "epoch": 0.6142895979042582,
      "grad_norm": 4.644674789489748,
      "learning_rate": 6.838655343378036e-08,
      "loss": 0.7729,
      "step": 6214
    },
    {
      "epoch": 0.6143884536490127,
      "grad_norm": 4.92344174143465,
      "learning_rate": 6.835617615210394e-08,
      "loss": 0.7081,
      "step": 6215
    },
    {
      "epoch": 0.6144873093937672,
      "grad_norm": 5.23042078824137,
      "learning_rate": 6.83258021150349e-08,
      "loss": 0.8037,
      "step": 6216
    },
    {
      "epoch": 0.6145861651385216,
      "grad_norm": 9.242251788918047,
      "learning_rate": 6.829543132568759e-08,
      "loss": 0.784,
      "step": 6217
    },
    {
      "epoch": 0.614685020883276,
      "grad_norm": 8.69679550091551,
      "learning_rate": 6.826506378717616e-08,
      "loss": 0.7367,
      "step": 6218
    },
    {
      "epoch": 0.6147838766280306,
      "grad_norm": 3.153364076718097,
      "learning_rate": 6.823469950261428e-08,
      "loss": 0.6835,
      "step": 6219
    },
    {
      "epoch": 0.614882732372785,
      "grad_norm": 2.9668624931413228,
      "learning_rate": 6.820433847511538e-08,
      "loss": 0.7645,
      "step": 6220
    },
    {
      "epoch": 0.6149815881175394,
      "grad_norm": 10.935259864044317,
      "learning_rate": 6.817398070779261e-08,
      "loss": 0.8057,
      "step": 6221
    },
    {
      "epoch": 0.615080443862294,
      "grad_norm": 7.093704448137936,
      "learning_rate": 6.814362620375859e-08,
      "loss": 0.7025,
      "step": 6222
    },
    {
      "epoch": 0.6151792996070484,
      "grad_norm": 4.384723473528429,
      "learning_rate": 6.811327496612583e-08,
      "loss": 0.8146,
      "step": 6223
    },
    {
      "epoch": 0.6152781553518029,
      "grad_norm": 3.0471806316304217,
      "learning_rate": 6.808292699800635e-08,
      "loss": 0.7135,
      "step": 6224
    },
    {
      "epoch": 0.6153770110965574,
      "grad_norm": 7.279625831306129,
      "learning_rate": 6.80525823025119e-08,
      "loss": 0.63,
      "step": 6225
    },
    {
      "epoch": 0.6154758668413118,
      "grad_norm": 3.3237403938810837,
      "learning_rate": 6.802224088275386e-08,
      "loss": 0.7492,
      "step": 6226
    },
    {
      "epoch": 0.6155747225860663,
      "grad_norm": 8.18870448427928,
      "learning_rate": 6.799190274184335e-08,
      "loss": 0.7847,
      "step": 6227
    },
    {
      "epoch": 0.6156735783308207,
      "grad_norm": 3.7910888733726735,
      "learning_rate": 6.796156788289102e-08,
      "loss": 0.6959,
      "step": 6228
    },
    {
      "epoch": 0.6157724340755752,
      "grad_norm": 4.294180527062127,
      "learning_rate": 6.793123630900733e-08,
      "loss": 0.7288,
      "step": 6229
    },
    {
      "epoch": 0.6158712898203297,
      "grad_norm": 3.6922329874747533,
      "learning_rate": 6.790090802330228e-08,
      "loss": 0.7438,
      "step": 6230
    },
    {
      "epoch": 0.6159701455650841,
      "grad_norm": 4.81162728027057,
      "learning_rate": 6.787058302888565e-08,
      "loss": 0.7136,
      "step": 6231
    },
    {
      "epoch": 0.6160690013098387,
      "grad_norm": 3.274706760392854,
      "learning_rate": 6.784026132886676e-08,
      "loss": 0.7057,
      "step": 6232
    },
    {
      "epoch": 0.6161678570545931,
      "grad_norm": 4.751856226945673,
      "learning_rate": 6.78099429263547e-08,
      "loss": 0.5817,
      "step": 6233
    },
    {
      "epoch": 0.6162667127993475,
      "grad_norm": 4.193783336666513,
      "learning_rate": 6.777962782445814e-08,
      "loss": 0.6874,
      "step": 6234
    },
    {
      "epoch": 0.6163655685441021,
      "grad_norm": 4.896208182100557,
      "learning_rate": 6.774931602628549e-08,
      "loss": 0.7392,
      "step": 6235
    },
    {
      "epoch": 0.6164644242888565,
      "grad_norm": 3.378470669279497,
      "learning_rate": 6.771900753494473e-08,
      "loss": 0.7553,
      "step": 6236
    },
    {
      "epoch": 0.6165632800336109,
      "grad_norm": 11.927256860100345,
      "learning_rate": 6.768870235354354e-08,
      "loss": 0.6117,
      "step": 6237
    },
    {
      "epoch": 0.6166621357783654,
      "grad_norm": 3.8377416024393622,
      "learning_rate": 6.765840048518936e-08,
      "loss": 0.5667,
      "step": 6238
    },
    {
      "epoch": 0.6167609915231199,
      "grad_norm": 4.535338977268666,
      "learning_rate": 6.76281019329891e-08,
      "loss": 0.6744,
      "step": 6239
    },
    {
      "epoch": 0.6168598472678744,
      "grad_norm": 8.272872411950482,
      "learning_rate": 6.759780670004948e-08,
      "loss": 0.7106,
      "step": 6240
    },
    {
      "epoch": 0.6169587030126288,
      "grad_norm": 3.6726678398962593,
      "learning_rate": 6.756751478947682e-08,
      "loss": 0.7435,
      "step": 6241
    },
    {
      "epoch": 0.6170575587573833,
      "grad_norm": 5.960883474467487,
      "learning_rate": 6.753722620437709e-08,
      "loss": 0.758,
      "step": 6242
    },
    {
      "epoch": 0.6171564145021378,
      "grad_norm": 20.128041102672576,
      "learning_rate": 6.750694094785598e-08,
      "loss": 0.785,
      "step": 6243
    },
    {
      "epoch": 0.6172552702468922,
      "grad_norm": 3.138104728846168,
      "learning_rate": 6.747665902301874e-08,
      "loss": 0.8189,
      "step": 6244
    },
    {
      "epoch": 0.6173541259916467,
      "grad_norm": 13.82120287358253,
      "learning_rate": 6.744638043297035e-08,
      "loss": 0.7481,
      "step": 6245
    },
    {
      "epoch": 0.6174529817364012,
      "grad_norm": 3.575192087922712,
      "learning_rate": 6.741610518081551e-08,
      "loss": 0.7439,
      "step": 6246
    },
    {
      "epoch": 0.6175518374811556,
      "grad_norm": 19.427179705203343,
      "learning_rate": 6.738583326965837e-08,
      "loss": 0.7524,
      "step": 6247
    },
    {
      "epoch": 0.61765069322591,
      "grad_norm": 3.625597361531161,
      "learning_rate": 6.735556470260297e-08,
      "loss": 0.7847,
      "step": 6248
    },
    {
      "epoch": 0.6177495489706646,
      "grad_norm": 2.9780986087858965,
      "learning_rate": 6.732529948275288e-08,
      "loss": 0.5668,
      "step": 6249
    },
    {
      "epoch": 0.617848404715419,
      "grad_norm": 13.471357455358863,
      "learning_rate": 6.729503761321133e-08,
      "loss": 0.6893,
      "step": 6250
    },
    {
      "epoch": 0.6179472604601735,
      "grad_norm": 3.3916893170123283,
      "learning_rate": 6.726477909708123e-08,
      "loss": 0.6807,
      "step": 6251
    },
    {
      "epoch": 0.618046116204928,
      "grad_norm": 3.6340040080200624,
      "learning_rate": 6.723452393746519e-08,
      "loss": 0.6653,
      "step": 6252
    },
    {
      "epoch": 0.6181449719496824,
      "grad_norm": 4.0448200679631885,
      "learning_rate": 6.720427213746534e-08,
      "loss": 0.6453,
      "step": 6253
    },
    {
      "epoch": 0.6182438276944369,
      "grad_norm": 10.226989720994846,
      "learning_rate": 6.717402370018366e-08,
      "loss": 0.6821,
      "step": 6254
    },
    {
      "epoch": 0.6183426834391914,
      "grad_norm": 3.957763001026486,
      "learning_rate": 6.71437786287216e-08,
      "loss": 0.7745,
      "step": 6255
    },
    {
      "epoch": 0.6184415391839458,
      "grad_norm": 4.731305437730604,
      "learning_rate": 6.711353692618037e-08,
      "loss": 0.8482,
      "step": 6256
    },
    {
      "epoch": 0.6185403949287003,
      "grad_norm": 7.4669956353077875,
      "learning_rate": 6.708329859566086e-08,
      "loss": 0.7425,
      "step": 6257
    },
    {
      "epoch": 0.6186392506734547,
      "grad_norm": 3.6020710698928804,
      "learning_rate": 6.705306364026351e-08,
      "loss": 0.6738,
      "step": 6258
    },
    {
      "epoch": 0.6187381064182093,
      "grad_norm": 5.910326640846438,
      "learning_rate": 6.702283206308846e-08,
      "loss": 0.6654,
      "step": 6259
    },
    {
      "epoch": 0.6188369621629637,
      "grad_norm": 4.933082495467321,
      "learning_rate": 6.699260386723558e-08,
      "loss": 0.6579,
      "step": 6260
    },
    {
      "epoch": 0.6189358179077181,
      "grad_norm": 4.607648434896453,
      "learning_rate": 6.696237905580422e-08,
      "loss": 0.6754,
      "step": 6261
    },
    {
      "epoch": 0.6190346736524727,
      "grad_norm": 4.896115761004341,
      "learning_rate": 6.693215763189361e-08,
      "loss": 0.783,
      "step": 6262
    },
    {
      "epoch": 0.6191335293972271,
      "grad_norm": 4.181545759128898,
      "learning_rate": 6.69019395986024e-08,
      "loss": 0.7463,
      "step": 6263
    },
    {
      "epoch": 0.6192323851419815,
      "grad_norm": 3.5392519789266585,
      "learning_rate": 6.687172495902907e-08,
      "loss": 0.673,
      "step": 6264
    },
    {
      "epoch": 0.6193312408867361,
      "grad_norm": 3.1205880188763606,
      "learning_rate": 6.684151371627168e-08,
      "loss": 0.6422,
      "step": 6265
    },
    {
      "epoch": 0.6194300966314905,
      "grad_norm": 3.779961040697167,
      "learning_rate": 6.681130587342792e-08,
      "loss": 0.7022,
      "step": 6266
    },
    {
      "epoch": 0.619528952376245,
      "grad_norm": 4.222559511640989,
      "learning_rate": 6.678110143359518e-08,
      "loss": 0.6806,
      "step": 6267
    },
    {
      "epoch": 0.6196278081209994,
      "grad_norm": 4.2906768866299325,
      "learning_rate": 6.67509003998705e-08,
      "loss": 0.6306,
      "step": 6268
    },
    {
      "epoch": 0.6197266638657539,
      "grad_norm": 17.117394372937067,
      "learning_rate": 6.672070277535049e-08,
      "loss": 0.6796,
      "step": 6269
    },
    {
      "epoch": 0.6198255196105084,
      "grad_norm": 3.655166234031778,
      "learning_rate": 6.66905085631315e-08,
      "loss": 0.7317,
      "step": 6270
    },
    {
      "epoch": 0.6199243753552628,
      "grad_norm": 15.254431853727558,
      "learning_rate": 6.666031776630954e-08,
      "loss": 0.5562,
      "step": 6271
    },
    {
      "epoch": 0.6200232311000173,
      "grad_norm": 6.5981761334403,
      "learning_rate": 6.663013038798018e-08,
      "loss": 0.7478,
      "step": 6272
    },
    {
      "epoch": 0.6201220868447718,
      "grad_norm": 3.696566535617549,
      "learning_rate": 6.659994643123873e-08,
      "loss": 0.7006,
      "step": 6273
    },
    {
      "epoch": 0.6202209425895262,
      "grad_norm": 2.8523028331358944,
      "learning_rate": 6.656976589918006e-08,
      "loss": 0.6843,
      "step": 6274
    },
    {
      "epoch": 0.6203197983342807,
      "grad_norm": 3.1620233517000274,
      "learning_rate": 6.653958879489877e-08,
      "loss": 0.6463,
      "step": 6275
    },
    {
      "epoch": 0.6204186540790352,
      "grad_norm": 6.043312077238663,
      "learning_rate": 6.650941512148909e-08,
      "loss": 0.665,
      "step": 6276
    },
    {
      "epoch": 0.6205175098237896,
      "grad_norm": 7.3017135629872865,
      "learning_rate": 6.647924488204486e-08,
      "loss": 0.6789,
      "step": 6277
    },
    {
      "epoch": 0.620616365568544,
      "grad_norm": 3.208461430279885,
      "learning_rate": 6.644907807965958e-08,
      "loss": 0.6787,
      "step": 6278
    },
    {
      "epoch": 0.6207152213132986,
      "grad_norm": 6.249949642126891,
      "learning_rate": 6.641891471742648e-08,
      "loss": 0.7854,
      "step": 6279
    },
    {
      "epoch": 0.620814077058053,
      "grad_norm": 3.8993490003725872,
      "learning_rate": 6.638875479843828e-08,
      "loss": 0.7629,
      "step": 6280
    },
    {
      "epoch": 0.6209129328028075,
      "grad_norm": 7.663960225524915,
      "learning_rate": 6.635859832578746e-08,
      "loss": 0.716,
      "step": 6281
    },
    {
      "epoch": 0.621011788547562,
      "grad_norm": 8.208653431460247,
      "learning_rate": 6.632844530256619e-08,
      "loss": 0.6555,
      "step": 6282
    },
    {
      "epoch": 0.6211106442923164,
      "grad_norm": 5.85834194043953,
      "learning_rate": 6.629829573186611e-08,
      "loss": 0.6702,
      "step": 6283
    },
    {
      "epoch": 0.6212095000370709,
      "grad_norm": 3.2591961981432402,
      "learning_rate": 6.626814961677872e-08,
      "loss": 0.74,
      "step": 6284
    },
    {
      "epoch": 0.6213083557818254,
      "grad_norm": 4.632411144714355,
      "learning_rate": 6.623800696039499e-08,
      "loss": 0.6905,
      "step": 6285
    },
    {
      "epoch": 0.6214072115265799,
      "grad_norm": 14.725198427679178,
      "learning_rate": 6.62078677658056e-08,
      "loss": 0.6295,
      "step": 6286
    },
    {
      "epoch": 0.6215060672713343,
      "grad_norm": 4.433767577658281,
      "learning_rate": 6.617773203610093e-08,
      "loss": 0.705,
      "step": 6287
    },
    {
      "epoch": 0.6216049230160887,
      "grad_norm": 3.635158821672549,
      "learning_rate": 6.614759977437089e-08,
      "loss": 0.8058,
      "step": 6288
    },
    {
      "epoch": 0.6217037787608433,
      "grad_norm": 3.353846745683276,
      "learning_rate": 6.611747098370515e-08,
      "loss": 0.6826,
      "step": 6289
    },
    {
      "epoch": 0.6218026345055977,
      "grad_norm": 3.030404340737694,
      "learning_rate": 6.608734566719299e-08,
      "loss": 0.7408,
      "step": 6290
    },
    {
      "epoch": 0.6219014902503521,
      "grad_norm": 3.2809743341731634,
      "learning_rate": 6.605722382792323e-08,
      "loss": 0.6824,
      "step": 6291
    },
    {
      "epoch": 0.6220003459951067,
      "grad_norm": 3.870936100330101,
      "learning_rate": 6.602710546898451e-08,
      "loss": 0.7945,
      "step": 6292
    },
    {
      "epoch": 0.6220992017398611,
      "grad_norm": 8.799497247674406,
      "learning_rate": 6.5996990593465e-08,
      "loss": 0.7427,
      "step": 6293
    },
    {
      "epoch": 0.6221980574846155,
      "grad_norm": 4.875790614137491,
      "learning_rate": 6.59668792044525e-08,
      "loss": 0.712,
      "step": 6294
    },
    {
      "epoch": 0.6222969132293701,
      "grad_norm": 6.046370302826926,
      "learning_rate": 6.593677130503454e-08,
      "loss": 0.7385,
      "step": 6295
    },
    {
      "epoch": 0.6223957689741245,
      "grad_norm": 9.058245861545682,
      "learning_rate": 6.590666689829818e-08,
      "loss": 0.5422,
      "step": 6296
    },
    {
      "epoch": 0.622494624718879,
      "grad_norm": 6.798626020353305,
      "learning_rate": 6.587656598733023e-08,
      "loss": 0.7778,
      "step": 6297
    },
    {
      "epoch": 0.6225934804636335,
      "grad_norm": 3.7924407395037307,
      "learning_rate": 6.584646857521709e-08,
      "loss": 0.7461,
      "step": 6298
    },
    {
      "epoch": 0.6226923362083879,
      "grad_norm": 7.079361794716562,
      "learning_rate": 6.581637466504479e-08,
      "loss": 0.7278,
      "step": 6299
    },
    {
      "epoch": 0.6227911919531424,
      "grad_norm": 3.916469769110348,
      "learning_rate": 6.5786284259899e-08,
      "loss": 0.7994,
      "step": 6300
    },
    {
      "epoch": 0.6228900476978968,
      "grad_norm": 3.462061171856143,
      "learning_rate": 6.575619736286513e-08,
      "loss": 0.7018,
      "step": 6301
    },
    {
      "epoch": 0.6229889034426513,
      "grad_norm": 10.5058302339052,
      "learning_rate": 6.572611397702805e-08,
      "loss": 0.7438,
      "step": 6302
    },
    {
      "epoch": 0.6230877591874058,
      "grad_norm": 10.79715971253462,
      "learning_rate": 6.56960341054724e-08,
      "loss": 0.7847,
      "step": 6303
    },
    {
      "epoch": 0.6231866149321602,
      "grad_norm": 5.775058360699875,
      "learning_rate": 6.566595775128248e-08,
      "loss": 0.6894,
      "step": 6304
    },
    {
      "epoch": 0.6232854706769148,
      "grad_norm": 3.44836020208474,
      "learning_rate": 6.563588491754208e-08,
      "loss": 0.6476,
      "step": 6305
    },
    {
      "epoch": 0.6233843264216692,
      "grad_norm": 15.55859993490643,
      "learning_rate": 6.560581560733482e-08,
      "loss": 0.7717,
      "step": 6306
    },
    {
      "epoch": 0.6234831821664236,
      "grad_norm": 3.7422368606297396,
      "learning_rate": 6.557574982374377e-08,
      "loss": 0.6557,
      "step": 6307
    },
    {
      "epoch": 0.6235820379111782,
      "grad_norm": 2.9501495090671117,
      "learning_rate": 6.554568756985178e-08,
      "loss": 0.5925,
      "step": 6308
    },
    {
      "epoch": 0.6236808936559326,
      "grad_norm": 3.7825402619566737,
      "learning_rate": 6.551562884874135e-08,
      "loss": 0.8043,
      "step": 6309
    },
    {
      "epoch": 0.623779749400687,
      "grad_norm": 7.968455970047168,
      "learning_rate": 6.548557366349445e-08,
      "loss": 0.7606,
      "step": 6310
    },
    {
      "epoch": 0.6238786051454415,
      "grad_norm": 4.46391660858993,
      "learning_rate": 6.545552201719286e-08,
      "loss": 0.7418,
      "step": 6311
    },
    {
      "epoch": 0.623977460890196,
      "grad_norm": 8.49460961185926,
      "learning_rate": 6.542547391291795e-08,
      "loss": 0.6551,
      "step": 6312
    },
    {
      "epoch": 0.6240763166349504,
      "grad_norm": 2.9851707034607324,
      "learning_rate": 6.539542935375064e-08,
      "loss": 0.6861,
      "step": 6313
    },
    {
      "epoch": 0.6241751723797049,
      "grad_norm": 3.696139506804978,
      "learning_rate": 6.53653883427716e-08,
      "loss": 0.632,
      "step": 6314
    },
    {
      "epoch": 0.6242740281244594,
      "grad_norm": 5.781119761688448,
      "learning_rate": 6.533535088306112e-08,
      "loss": 0.6851,
      "step": 6315
    },
    {
      "epoch": 0.6243728838692139,
      "grad_norm": 11.880790847848965,
      "learning_rate": 6.530531697769904e-08,
      "loss": 0.8123,
      "step": 6316
    },
    {
      "epoch": 0.6244717396139683,
      "grad_norm": 25.15244698735989,
      "learning_rate": 6.527528662976495e-08,
      "loss": 0.7507,
      "step": 6317
    },
    {
      "epoch": 0.6245705953587228,
      "grad_norm": 4.565999530092545,
      "learning_rate": 6.524525984233795e-08,
      "loss": 0.636,
      "step": 6318
    },
    {
      "epoch": 0.6246694511034773,
      "grad_norm": 3.9391003907729507,
      "learning_rate": 6.521523661849691e-08,
      "loss": 0.674,
      "step": 6319
    },
    {
      "epoch": 0.6247683068482317,
      "grad_norm": 4.3503919391850925,
      "learning_rate": 6.518521696132024e-08,
      "loss": 0.7646,
      "step": 6320
    },
    {
      "epoch": 0.6248671625929861,
      "grad_norm": 5.039283039436438,
      "learning_rate": 6.515520087388602e-08,
      "loss": 0.6825,
      "step": 6321
    },
    {
      "epoch": 0.6249660183377407,
      "grad_norm": 4.712927891922407,
      "learning_rate": 6.512518835927193e-08,
      "loss": 0.683,
      "step": 6322
    },
    {
      "epoch": 0.6250648740824951,
      "grad_norm": 3.2027899931866592,
      "learning_rate": 6.509517942055538e-08,
      "loss": 0.7349,
      "step": 6323
    },
    {
      "epoch": 0.6251637298272495,
      "grad_norm": 3.4141505945147643,
      "learning_rate": 6.506517406081325e-08,
      "loss": 0.5947,
      "step": 6324
    },
    {
      "epoch": 0.6252625855720041,
      "grad_norm": 5.469997325423968,
      "learning_rate": 6.50351722831222e-08,
      "loss": 0.6565,
      "step": 6325
    },
    {
      "epoch": 0.6253614413167585,
      "grad_norm": 8.439113255546191,
      "learning_rate": 6.50051740905585e-08,
      "loss": 0.7422,
      "step": 6326
    },
    {
      "epoch": 0.625460297061513,
      "grad_norm": 4.177085746007796,
      "learning_rate": 6.497517948619796e-08,
      "loss": 0.8066,
      "step": 6327
    },
    {
      "epoch": 0.6255591528062675,
      "grad_norm": 2.9306928778316848,
      "learning_rate": 6.494518847311614e-08,
      "loss": 0.7816,
      "step": 6328
    },
    {
      "epoch": 0.6256580085510219,
      "grad_norm": 7.959703243777141,
      "learning_rate": 6.491520105438813e-08,
      "loss": 0.6277,
      "step": 6329
    },
    {
      "epoch": 0.6257568642957764,
      "grad_norm": 4.765365558257992,
      "learning_rate": 6.488521723308868e-08,
      "loss": 0.7469,
      "step": 6330
    },
    {
      "epoch": 0.6258557200405308,
      "grad_norm": 5.055629753000996,
      "learning_rate": 6.485523701229228e-08,
      "loss": 0.6932,
      "step": 6331
    },
    {
      "epoch": 0.6259545757852853,
      "grad_norm": 5.8648072800843085,
      "learning_rate": 6.482526039507287e-08,
      "loss": 0.786,
      "step": 6332
    },
    {
      "epoch": 0.6260534315300398,
      "grad_norm": 4.024240802144581,
      "learning_rate": 6.479528738450412e-08,
      "loss": 0.7225,
      "step": 6333
    },
    {
      "epoch": 0.6261522872747942,
      "grad_norm": 3.1805352633292174,
      "learning_rate": 6.476531798365941e-08,
      "loss": 0.6877,
      "step": 6334
    },
    {
      "epoch": 0.6262511430195488,
      "grad_norm": 20.35932820289275,
      "learning_rate": 6.473535219561152e-08,
      "loss": 0.7368,
      "step": 6335
    },
    {
      "epoch": 0.6263499987643032,
      "grad_norm": 5.55151726059079,
      "learning_rate": 6.470539002343312e-08,
      "loss": 0.666,
      "step": 6336
    },
    {
      "epoch": 0.6264488545090576,
      "grad_norm": 2.736333447523643,
      "learning_rate": 6.467543147019633e-08,
      "loss": 0.5854,
      "step": 6337
    },
    {
      "epoch": 0.6265477102538122,
      "grad_norm": 3.7030818840743946,
      "learning_rate": 6.464547653897297e-08,
      "loss": 0.6504,
      "step": 6338
    },
    {
      "epoch": 0.6266465659985666,
      "grad_norm": 4.260082788578561,
      "learning_rate": 6.461552523283449e-08,
      "loss": 0.7211,
      "step": 6339
    },
    {
      "epoch": 0.626745421743321,
      "grad_norm": 4.071906303752496,
      "learning_rate": 6.458557755485192e-08,
      "loss": 0.7423,
      "step": 6340
    },
    {
      "epoch": 0.6268442774880755,
      "grad_norm": 5.564647488815131,
      "learning_rate": 6.455563350809595e-08,
      "loss": 0.6378,
      "step": 6341
    },
    {
      "epoch": 0.62694313323283,
      "grad_norm": 3.0921984573223344,
      "learning_rate": 6.452569309563696e-08,
      "loss": 0.6712,
      "step": 6342
    },
    {
      "epoch": 0.6270419889775845,
      "grad_norm": 6.707406854142768,
      "learning_rate": 6.449575632054482e-08,
      "loss": 0.6994,
      "step": 6343
    },
    {
      "epoch": 0.6271408447223389,
      "grad_norm": 4.036187087791546,
      "learning_rate": 6.446582318588917e-08,
      "loss": 0.7527,
      "step": 6344
    },
    {
      "epoch": 0.6272397004670934,
      "grad_norm": 3.362909720957736,
      "learning_rate": 6.443589369473921e-08,
      "loss": 0.8145,
      "step": 6345
    },
    {
      "epoch": 0.6273385562118479,
      "grad_norm": 7.006676537878311,
      "learning_rate": 6.440596785016373e-08,
      "loss": 0.6488,
      "step": 6346
    },
    {
      "epoch": 0.6274374119566023,
      "grad_norm": 4.836704986110418,
      "learning_rate": 6.437604565523116e-08,
      "loss": 0.7151,
      "step": 6347
    },
    {
      "epoch": 0.6275362677013568,
      "grad_norm": 3.4880233746656923,
      "learning_rate": 6.434612711300968e-08,
      "loss": 0.6539,
      "step": 6348
    },
    {
      "epoch": 0.6276351234461113,
      "grad_norm": 7.272032443269752,
      "learning_rate": 6.431621222656688e-08,
      "loss": 0.7508,
      "step": 6349
    },
    {
      "epoch": 0.6277339791908657,
      "grad_norm": 4.801087082220287,
      "learning_rate": 6.428630099897018e-08,
      "loss": 0.7032,
      "step": 6350
    },
    {
      "epoch": 0.6278328349356201,
      "grad_norm": 6.652784198425823,
      "learning_rate": 6.425639343328647e-08,
      "loss": 0.6764,
      "step": 6351
    },
    {
      "epoch": 0.6279316906803747,
      "grad_norm": 3.515211984210982,
      "learning_rate": 6.422648953258235e-08,
      "loss": 0.6086,
      "step": 6352
    },
    {
      "epoch": 0.6280305464251291,
      "grad_norm": 3.269088046660099,
      "learning_rate": 6.419658929992408e-08,
      "loss": 0.6704,
      "step": 6353
    },
    {
      "epoch": 0.6281294021698836,
      "grad_norm": 3.777496552214907,
      "learning_rate": 6.416669273837739e-08,
      "loss": 0.7838,
      "step": 6354
    },
    {
      "epoch": 0.6282282579146381,
      "grad_norm": 4.148701788678798,
      "learning_rate": 6.413679985100781e-08,
      "loss": 0.7843,
      "step": 6355
    },
    {
      "epoch": 0.6283271136593925,
      "grad_norm": 3.995997001977764,
      "learning_rate": 6.410691064088041e-08,
      "loss": 0.7326,
      "step": 6356
    },
    {
      "epoch": 0.628425969404147,
      "grad_norm": 5.184056174537193,
      "learning_rate": 6.407702511105981e-08,
      "loss": 0.6243,
      "step": 6357
    },
    {
      "epoch": 0.6285248251489015,
      "grad_norm": 6.789385749779301,
      "learning_rate": 6.404714326461039e-08,
      "loss": 0.741,
      "step": 6358
    },
    {
      "epoch": 0.628623680893656,
      "grad_norm": 3.4951273803051324,
      "learning_rate": 6.401726510459612e-08,
      "loss": 0.6736,
      "step": 6359
    },
    {
      "epoch": 0.6287225366384104,
      "grad_norm": 5.177298856205879,
      "learning_rate": 6.39873906340805e-08,
      "loss": 0.5941,
      "step": 6360
    },
    {
      "epoch": 0.6288213923831648,
      "grad_norm": 7.255165100817646,
      "learning_rate": 6.395751985612678e-08,
      "loss": 0.7812,
      "step": 6361
    },
    {
      "epoch": 0.6289202481279194,
      "grad_norm": 3.74722379532626,
      "learning_rate": 6.392765277379771e-08,
      "loss": 0.5796,
      "step": 6362
    },
    {
      "epoch": 0.6290191038726738,
      "grad_norm": 3.533143545489133,
      "learning_rate": 6.389778939015573e-08,
      "loss": 0.7553,
      "step": 6363
    },
    {
      "epoch": 0.6291179596174282,
      "grad_norm": 3.718738738194319,
      "learning_rate": 6.386792970826294e-08,
      "loss": 0.651,
      "step": 6364
    },
    {
      "epoch": 0.6292168153621828,
      "grad_norm": 4.095190293126384,
      "learning_rate": 6.383807373118095e-08,
      "loss": 0.7629,
      "step": 6365
    },
    {
      "epoch": 0.6293156711069372,
      "grad_norm": 3.8849259685379414,
      "learning_rate": 6.380822146197107e-08,
      "loss": 0.6974,
      "step": 6366
    },
    {
      "epoch": 0.6294145268516916,
      "grad_norm": 4.993614323674291,
      "learning_rate": 6.377837290369423e-08,
      "loss": 0.6708,
      "step": 6367
    },
    {
      "epoch": 0.6295133825964462,
      "grad_norm": 3.3156951811996147,
      "learning_rate": 6.374852805941091e-08,
      "loss": 0.6292,
      "step": 6368
    },
    {
      "epoch": 0.6296122383412006,
      "grad_norm": 4.4837953452362465,
      "learning_rate": 6.37186869321813e-08,
      "loss": 0.7018,
      "step": 6369
    },
    {
      "epoch": 0.629711094085955,
      "grad_norm": 9.254188847604313,
      "learning_rate": 6.368884952506519e-08,
      "loss": 0.7675,
      "step": 6370
    },
    {
      "epoch": 0.6298099498307096,
      "grad_norm": 3.7962807462777555,
      "learning_rate": 6.36590158411219e-08,
      "loss": 0.6598,
      "step": 6371
    },
    {
      "epoch": 0.629908805575464,
      "grad_norm": 3.1013880379869287,
      "learning_rate": 6.362918588341051e-08,
      "loss": 0.6387,
      "step": 6372
    },
    {
      "epoch": 0.6300076613202185,
      "grad_norm": 6.406959164884269,
      "learning_rate": 6.359935965498958e-08,
      "loss": 0.69,
      "step": 6373
    },
    {
      "epoch": 0.6301065170649729,
      "grad_norm": 4.261266028022718,
      "learning_rate": 6.356953715891735e-08,
      "loss": 0.6752,
      "step": 6374
    },
    {
      "epoch": 0.6302053728097274,
      "grad_norm": 3.3171578548134293,
      "learning_rate": 6.353971839825173e-08,
      "loss": 0.7442,
      "step": 6375
    },
    {
      "epoch": 0.6303042285544819,
      "grad_norm": 7.413550029893226,
      "learning_rate": 6.350990337605012e-08,
      "loss": 0.6789,
      "step": 6376
    },
    {
      "epoch": 0.6304030842992363,
      "grad_norm": 3.2517768343000277,
      "learning_rate": 6.348009209536966e-08,
      "loss": 0.6849,
      "step": 6377
    },
    {
      "epoch": 0.6305019400439908,
      "grad_norm": 3.9850479709004536,
      "learning_rate": 6.345028455926708e-08,
      "loss": 0.6859,
      "step": 6378
    },
    {
      "epoch": 0.6306007957887453,
      "grad_norm": 3.8055348206382047,
      "learning_rate": 6.342048077079865e-08,
      "loss": 0.6039,
      "step": 6379
    },
    {
      "epoch": 0.6306996515334997,
      "grad_norm": 4.340607776792594,
      "learning_rate": 6.339068073302035e-08,
      "loss": 0.672,
      "step": 6380
    },
    {
      "epoch": 0.6307985072782543,
      "grad_norm": 3.366983103331821,
      "learning_rate": 6.336088444898768e-08,
      "loss": 0.6622,
      "step": 6381
    },
    {
      "epoch": 0.6308973630230087,
      "grad_norm": 3.885755580649514,
      "learning_rate": 6.333109192175588e-08,
      "loss": 0.7064,
      "step": 6382
    },
    {
      "epoch": 0.6309962187677631,
      "grad_norm": 3.6295313569291126,
      "learning_rate": 6.330130315437971e-08,
      "loss": 0.6758,
      "step": 6383
    },
    {
      "epoch": 0.6310950745125176,
      "grad_norm": 3.386109506699371,
      "learning_rate": 6.327151814991351e-08,
      "loss": 0.6923,
      "step": 6384
    },
    {
      "epoch": 0.6311939302572721,
      "grad_norm": 5.482291727613971,
      "learning_rate": 6.324173691141134e-08,
      "loss": 0.6564,
      "step": 6385
    },
    {
      "epoch": 0.6312927860020265,
      "grad_norm": 4.65870897632115,
      "learning_rate": 6.321195944192689e-08,
      "loss": 0.5785,
      "step": 6386
    },
    {
      "epoch": 0.631391641746781,
      "grad_norm": 3.5783784494065483,
      "learning_rate": 6.318218574451328e-08,
      "loss": 0.7012,
      "step": 6387
    },
    {
      "epoch": 0.6314904974915355,
      "grad_norm": 10.067256954978951,
      "learning_rate": 6.315241582222341e-08,
      "loss": 0.7222,
      "step": 6388
    },
    {
      "epoch": 0.63158935323629,
      "grad_norm": 5.305422470266035,
      "learning_rate": 6.312264967810981e-08,
      "loss": 0.6525,
      "step": 6389
    },
    {
      "epoch": 0.6316882089810444,
      "grad_norm": 3.4693765037419038,
      "learning_rate": 6.30928873152245e-08,
      "loss": 0.6519,
      "step": 6390
    },
    {
      "epoch": 0.6317870647257989,
      "grad_norm": 3.741554464418401,
      "learning_rate": 6.306312873661915e-08,
      "loss": 0.6987,
      "step": 6391
    },
    {
      "epoch": 0.6318859204705534,
      "grad_norm": 4.373886036227352,
      "learning_rate": 6.303337394534512e-08,
      "loss": 0.7349,
      "step": 6392
    },
    {
      "epoch": 0.6319847762153078,
      "grad_norm": 5.803830460322083,
      "learning_rate": 6.300362294445326e-08,
      "loss": 0.7379,
      "step": 6393
    },
    {
      "epoch": 0.6320836319600622,
      "grad_norm": 5.166470407993537,
      "learning_rate": 6.297387573699417e-08,
      "loss": 0.6896,
      "step": 6394
    },
    {
      "epoch": 0.6321824877048168,
      "grad_norm": 3.710213265058296,
      "learning_rate": 6.294413232601791e-08,
      "loss": 0.7322,
      "step": 6395
    },
    {
      "epoch": 0.6322813434495712,
      "grad_norm": 9.519209874902531,
      "learning_rate": 6.291439271457426e-08,
      "loss": 0.7498,
      "step": 6396
    },
    {
      "epoch": 0.6323801991943256,
      "grad_norm": 7.2123473719816955,
      "learning_rate": 6.288465690571266e-08,
      "loss": 0.5397,
      "step": 6397
    },
    {
      "epoch": 0.6324790549390802,
      "grad_norm": 4.699541872063325,
      "learning_rate": 6.285492490248193e-08,
      "loss": 0.6603,
      "step": 6398
    },
    {
      "epoch": 0.6325779106838346,
      "grad_norm": 3.050316220011718,
      "learning_rate": 6.282519670793074e-08,
      "loss": 0.7436,
      "step": 6399
    },
    {
      "epoch": 0.632676766428589,
      "grad_norm": 3.932108235035159,
      "learning_rate": 6.27954723251073e-08,
      "loss": 0.7477,
      "step": 6400
    },
    {
      "epoch": 0.6327756221733436,
      "grad_norm": 13.247102237347239,
      "learning_rate": 6.276575175705931e-08,
      "loss": 0.7172,
      "step": 6401
    },
    {
      "epoch": 0.632874477918098,
      "grad_norm": 16.52936903615708,
      "learning_rate": 6.273603500683425e-08,
      "loss": 0.8389,
      "step": 6402
    },
    {
      "epoch": 0.6329733336628525,
      "grad_norm": 4.329529652296655,
      "learning_rate": 6.270632207747916e-08,
      "loss": 0.6625,
      "step": 6403
    },
    {
      "epoch": 0.6330721894076069,
      "grad_norm": 4.329015821616079,
      "learning_rate": 6.267661297204057e-08,
      "loss": 0.765,
      "step": 6404
    },
    {
      "epoch": 0.6331710451523614,
      "grad_norm": 4.683682862613541,
      "learning_rate": 6.26469076935648e-08,
      "loss": 0.733,
      "step": 6405
    },
    {
      "epoch": 0.6332699008971159,
      "grad_norm": 4.455716016356109,
      "learning_rate": 6.26172062450976e-08,
      "loss": 0.8027,
      "step": 6406
    },
    {
      "epoch": 0.6333687566418703,
      "grad_norm": 3.922959500703809,
      "learning_rate": 6.258750862968451e-08,
      "loss": 0.7216,
      "step": 6407
    },
    {
      "epoch": 0.6334676123866249,
      "grad_norm": 3.8248469162398226,
      "learning_rate": 6.255781485037053e-08,
      "loss": 0.6948,
      "step": 6408
    },
    {
      "epoch": 0.6335664681313793,
      "grad_norm": 3.3977091392623864,
      "learning_rate": 6.252812491020033e-08,
      "loss": 0.6763,
      "step": 6409
    },
    {
      "epoch": 0.6336653238761337,
      "grad_norm": 3.425729098705427,
      "learning_rate": 6.249843881221813e-08,
      "loss": 0.6409,
      "step": 6410
    },
    {
      "epoch": 0.6337641796208883,
      "grad_norm": 4.909279605984304,
      "learning_rate": 6.24687565594679e-08,
      "loss": 0.6834,
      "step": 6411
    },
    {
      "epoch": 0.6338630353656427,
      "grad_norm": 4.274723882313248,
      "learning_rate": 6.243907815499301e-08,
      "loss": 0.6851,
      "step": 6412
    },
    {
      "epoch": 0.6339618911103971,
      "grad_norm": 5.517001292376823,
      "learning_rate": 6.24094036018366e-08,
      "loss": 0.6527,
      "step": 6413
    },
    {
      "epoch": 0.6340607468551516,
      "grad_norm": 4.561284733199471,
      "learning_rate": 6.237973290304139e-08,
      "loss": 0.7897,
      "step": 6414
    },
    {
      "epoch": 0.6341596025999061,
      "grad_norm": 6.563418355972411,
      "learning_rate": 6.235006606164959e-08,
      "loss": 0.5021,
      "step": 6415
    },
    {
      "epoch": 0.6342584583446605,
      "grad_norm": 4.995332333435272,
      "learning_rate": 6.232040308070318e-08,
      "loss": 0.6667,
      "step": 6416
    },
    {
      "epoch": 0.634357314089415,
      "grad_norm": 4.959069253611251,
      "learning_rate": 6.22907439632436e-08,
      "loss": 0.7101,
      "step": 6417
    },
    {
      "epoch": 0.6344561698341695,
      "grad_norm": 183.99518658654262,
      "learning_rate": 6.226108871231195e-08,
      "loss": 0.6665,
      "step": 6418
    },
    {
      "epoch": 0.634555025578924,
      "grad_norm": 5.285171832611314,
      "learning_rate": 6.223143733094903e-08,
      "loss": 0.6145,
      "step": 6419
    },
    {
      "epoch": 0.6346538813236784,
      "grad_norm": 4.2811709182488205,
      "learning_rate": 6.220178982219502e-08,
      "loss": 0.698,
      "step": 6420
    },
    {
      "epoch": 0.6347527370684329,
      "grad_norm": 4.779629064138447,
      "learning_rate": 6.21721461890899e-08,
      "loss": 0.7109,
      "step": 6421
    },
    {
      "epoch": 0.6348515928131874,
      "grad_norm": 4.804619680648377,
      "learning_rate": 6.214250643467323e-08,
      "loss": 0.5505,
      "step": 6422
    },
    {
      "epoch": 0.6349504485579418,
      "grad_norm": 3.055946114153866,
      "learning_rate": 6.211287056198406e-08,
      "loss": 0.726,
      "step": 6423
    },
    {
      "epoch": 0.6350493043026962,
      "grad_norm": 3.798059534515849,
      "learning_rate": 6.208323857406117e-08,
      "loss": 0.6324,
      "step": 6424
    },
    {
      "epoch": 0.6351481600474508,
      "grad_norm": 4.675098904405104,
      "learning_rate": 6.205361047394282e-08,
      "loss": 0.7247,
      "step": 6425
    },
    {
      "epoch": 0.6352470157922052,
      "grad_norm": 3.4660828922508014,
      "learning_rate": 6.202398626466699e-08,
      "loss": 0.7267,
      "step": 6426
    },
    {
      "epoch": 0.6353458715369596,
      "grad_norm": 4.606541723310948,
      "learning_rate": 6.19943659492712e-08,
      "loss": 0.7622,
      "step": 6427
    },
    {
      "epoch": 0.6354447272817142,
      "grad_norm": 3.281979557746655,
      "learning_rate": 6.196474953079251e-08,
      "loss": 0.6551,
      "step": 6428
    },
    {
      "epoch": 0.6355435830264686,
      "grad_norm": 5.807771707080433,
      "learning_rate": 6.19351370122677e-08,
      "loss": 0.7426,
      "step": 6429
    },
    {
      "epoch": 0.6356424387712231,
      "grad_norm": 4.099257708226999,
      "learning_rate": 6.190552839673313e-08,
      "loss": 0.6929,
      "step": 6430
    },
    {
      "epoch": 0.6357412945159776,
      "grad_norm": 2.9157953122739,
      "learning_rate": 6.187592368722466e-08,
      "loss": 0.767,
      "step": 6431
    },
    {
      "epoch": 0.635840150260732,
      "grad_norm": 5.601730544955432,
      "learning_rate": 6.184632288677785e-08,
      "loss": 0.7772,
      "step": 6432
    },
    {
      "epoch": 0.6359390060054865,
      "grad_norm": 4.233916674747636,
      "learning_rate": 6.181672599842784e-08,
      "loss": 0.641,
      "step": 6433
    },
    {
      "epoch": 0.6360378617502409,
      "grad_norm": 6.359670868870838,
      "learning_rate": 6.178713302520933e-08,
      "loss": 0.7219,
      "step": 6434
    },
    {
      "epoch": 0.6361367174949955,
      "grad_norm": 4.1221599136619,
      "learning_rate": 6.175754397015664e-08,
      "loss": 0.5986,
      "step": 6435
    },
    {
      "epoch": 0.6362355732397499,
      "grad_norm": 4.207712853628502,
      "learning_rate": 6.172795883630373e-08,
      "loss": 0.6706,
      "step": 6436
    },
    {
      "epoch": 0.6363344289845043,
      "grad_norm": 3.6549585745719773,
      "learning_rate": 6.169837762668404e-08,
      "loss": 0.6968,
      "step": 6437
    },
    {
      "epoch": 0.6364332847292589,
      "grad_norm": 9.903965092876502,
      "learning_rate": 6.166880034433079e-08,
      "loss": 0.7994,
      "step": 6438
    },
    {
      "epoch": 0.6365321404740133,
      "grad_norm": 4.715480007295255,
      "learning_rate": 6.163922699227658e-08,
      "loss": 0.6873,
      "step": 6439
    },
    {
      "epoch": 0.6366309962187677,
      "grad_norm": 4.074777482781056,
      "learning_rate": 6.16096575735538e-08,
      "loss": 0.6592,
      "step": 6440
    },
    {
      "epoch": 0.6367298519635223,
      "grad_norm": 3.780192160664811,
      "learning_rate": 6.158009209119438e-08,
      "loss": 0.7361,
      "step": 6441
    },
    {
      "epoch": 0.6368287077082767,
      "grad_norm": 4.180092635925593,
      "learning_rate": 6.155053054822972e-08,
      "loss": 0.633,
      "step": 6442
    },
    {
      "epoch": 0.6369275634530311,
      "grad_norm": 14.138339763712994,
      "learning_rate": 6.152097294769102e-08,
      "loss": 0.6339,
      "step": 6443
    },
    {
      "epoch": 0.6370264191977856,
      "grad_norm": 4.106240941027661,
      "learning_rate": 6.149141929260895e-08,
      "loss": 0.7474,
      "step": 6444
    },
    {
      "epoch": 0.6371252749425401,
      "grad_norm": 4.371595411096042,
      "learning_rate": 6.146186958601376e-08,
      "loss": 0.6846,
      "step": 6445
    },
    {
      "epoch": 0.6372241306872946,
      "grad_norm": 3.3410156213849374,
      "learning_rate": 6.143232383093536e-08,
      "loss": 0.7447,
      "step": 6446
    },
    {
      "epoch": 0.637322986432049,
      "grad_norm": 3.9305835910831837,
      "learning_rate": 6.140278203040328e-08,
      "loss": 0.7485,
      "step": 6447
    },
    {
      "epoch": 0.6374218421768035,
      "grad_norm": 6.24582263324381,
      "learning_rate": 6.137324418744651e-08,
      "loss": 0.708,
      "step": 6448
    },
    {
      "epoch": 0.637520697921558,
      "grad_norm": 3.2660072546939802,
      "learning_rate": 6.134371030509382e-08,
      "loss": 0.6815,
      "step": 6449
    },
    {
      "epoch": 0.6376195536663124,
      "grad_norm": 4.7776864307728735,
      "learning_rate": 6.131418038637336e-08,
      "loss": 0.7963,
      "step": 6450
    },
    {
      "epoch": 0.6377184094110669,
      "grad_norm": 6.793380576282161,
      "learning_rate": 6.128465443431307e-08,
      "loss": 0.7273,
      "step": 6451
    },
    {
      "epoch": 0.6378172651558214,
      "grad_norm": 3.005588763310597,
      "learning_rate": 6.125513245194039e-08,
      "loss": 0.6509,
      "step": 6452
    },
    {
      "epoch": 0.6379161209005758,
      "grad_norm": 4.337393023650969,
      "learning_rate": 6.122561444228233e-08,
      "loss": 0.7434,
      "step": 6453
    },
    {
      "epoch": 0.6380149766453304,
      "grad_norm": 18.213656547481868,
      "learning_rate": 6.119610040836552e-08,
      "loss": 0.6535,
      "step": 6454
    },
    {
      "epoch": 0.6381138323900848,
      "grad_norm": 6.635999922179362,
      "learning_rate": 6.116659035321628e-08,
      "loss": 0.8318,
      "step": 6455
    },
    {
      "epoch": 0.6382126881348392,
      "grad_norm": 4.836673146338819,
      "learning_rate": 6.113708427986031e-08,
      "loss": 0.7117,
      "step": 6456
    },
    {
      "epoch": 0.6383115438795937,
      "grad_norm": 3.4886505931882454,
      "learning_rate": 6.110758219132308e-08,
      "loss": 0.6077,
      "step": 6457
    },
    {
      "epoch": 0.6384103996243482,
      "grad_norm": 5.538933251306919,
      "learning_rate": 6.107808409062964e-08,
      "loss": 0.7678,
      "step": 6458
    },
    {
      "epoch": 0.6385092553691026,
      "grad_norm": 5.037187018889392,
      "learning_rate": 6.104858998080449e-08,
      "loss": 0.7508,
      "step": 6459
    },
    {
      "epoch": 0.6386081111138571,
      "grad_norm": 3.086058778695983,
      "learning_rate": 6.101909986487191e-08,
      "loss": 0.746,
      "step": 6460
    },
    {
      "epoch": 0.6387069668586116,
      "grad_norm": 8.02715026300322,
      "learning_rate": 6.098961374585561e-08,
      "loss": 0.7226,
      "step": 6461
    },
    {
      "epoch": 0.638805822603366,
      "grad_norm": 3.9477957911240065,
      "learning_rate": 6.096013162677895e-08,
      "loss": 0.7607,
      "step": 6462
    },
    {
      "epoch": 0.6389046783481205,
      "grad_norm": 4.916301663619931,
      "learning_rate": 6.093065351066497e-08,
      "loss": 0.6259,
      "step": 6463
    },
    {
      "epoch": 0.639003534092875,
      "grad_norm": 5.160944499534483,
      "learning_rate": 6.09011794005361e-08,
      "loss": 0.6998,
      "step": 6464
    },
    {
      "epoch": 0.6391023898376295,
      "grad_norm": 3.514360932095559,
      "learning_rate": 6.087170929941455e-08,
      "loss": 0.7765,
      "step": 6465
    },
    {
      "epoch": 0.6392012455823839,
      "grad_norm": 6.059806639276277,
      "learning_rate": 6.084224321032206e-08,
      "loss": 0.8323,
      "step": 6466
    },
    {
      "epoch": 0.6393001013271383,
      "grad_norm": 3.915271063468134,
      "learning_rate": 6.081278113627986e-08,
      "loss": 0.6702,
      "step": 6467
    },
    {
      "epoch": 0.6393989570718929,
      "grad_norm": 6.659257243700594,
      "learning_rate": 6.078332308030893e-08,
      "loss": 0.7459,
      "step": 6468
    },
    {
      "epoch": 0.6394978128166473,
      "grad_norm": 5.965686232431402,
      "learning_rate": 6.075386904542974e-08,
      "loss": 0.7705,
      "step": 6469
    },
    {
      "epoch": 0.6395966685614017,
      "grad_norm": 14.274941194643295,
      "learning_rate": 6.072441903466234e-08,
      "loss": 0.6254,
      "step": 6470
    },
    {
      "epoch": 0.6396955243061563,
      "grad_norm": 6.571029157661409,
      "learning_rate": 6.069497305102647e-08,
      "loss": 0.6683,
      "step": 6471
    },
    {
      "epoch": 0.6397943800509107,
      "grad_norm": 4.225812972082855,
      "learning_rate": 6.066553109754125e-08,
      "loss": 0.7721,
      "step": 6472
    },
    {
      "epoch": 0.6398932357956651,
      "grad_norm": 6.747128004536368,
      "learning_rate": 6.06360931772256e-08,
      "loss": 0.6024,
      "step": 6473
    },
    {
      "epoch": 0.6399920915404197,
      "grad_norm": 20.37159778891613,
      "learning_rate": 6.060665929309798e-08,
      "loss": 0.7785,
      "step": 6474
    },
    {
      "epoch": 0.6400909472851741,
      "grad_norm": 9.771142085052215,
      "learning_rate": 6.057722944817631e-08,
      "loss": 0.6946,
      "step": 6475
    },
    {
      "epoch": 0.6401898030299286,
      "grad_norm": 4.658965509952915,
      "learning_rate": 6.054780364547824e-08,
      "loss": 0.7596,
      "step": 6476
    },
    {
      "epoch": 0.640288658774683,
      "grad_norm": 3.503737103939055,
      "learning_rate": 6.051838188802097e-08,
      "loss": 0.5524,
      "step": 6477
    },
    {
      "epoch": 0.6403875145194375,
      "grad_norm": 3.809439751149928,
      "learning_rate": 6.048896417882124e-08,
      "loss": 0.8028,
      "step": 6478
    },
    {
      "epoch": 0.640486370264192,
      "grad_norm": 3.464375024342769,
      "learning_rate": 6.045955052089537e-08,
      "loss": 0.8145,
      "step": 6479
    },
    {
      "epoch": 0.6405852260089464,
      "grad_norm": 6.737056168960275,
      "learning_rate": 6.04301409172594e-08,
      "loss": 0.7215,
      "step": 6480
    },
    {
      "epoch": 0.640684081753701,
      "grad_norm": 4.361605899929606,
      "learning_rate": 6.040073537092872e-08,
      "loss": 0.5987,
      "step": 6481
    },
    {
      "epoch": 0.6407829374984554,
      "grad_norm": 4.652411574008578,
      "learning_rate": 6.037133388491853e-08,
      "loss": 0.7768,
      "step": 6482
    },
    {
      "epoch": 0.6408817932432098,
      "grad_norm": 5.189576475388783,
      "learning_rate": 6.034193646224346e-08,
      "loss": 0.7198,
      "step": 6483
    },
    {
      "epoch": 0.6409806489879644,
      "grad_norm": 3.235423135241254,
      "learning_rate": 6.03125431059178e-08,
      "loss": 0.6668,
      "step": 6484
    },
    {
      "epoch": 0.6410795047327188,
      "grad_norm": 2.900350989080107,
      "learning_rate": 6.028315381895547e-08,
      "loss": 0.6931,
      "step": 6485
    },
    {
      "epoch": 0.6411783604774732,
      "grad_norm": 9.048006058078702,
      "learning_rate": 6.02537686043698e-08,
      "loss": 0.7022,
      "step": 6486
    },
    {
      "epoch": 0.6412772162222277,
      "grad_norm": 5.937724555154702,
      "learning_rate": 6.022438746517387e-08,
      "loss": 0.6618,
      "step": 6487
    },
    {
      "epoch": 0.6413760719669822,
      "grad_norm": 3.457013827258602,
      "learning_rate": 6.01950104043803e-08,
      "loss": 0.7273,
      "step": 6488
    },
    {
      "epoch": 0.6414749277117366,
      "grad_norm": 3.0295083835818284,
      "learning_rate": 6.016563742500122e-08,
      "loss": 0.6576,
      "step": 6489
    },
    {
      "epoch": 0.6415737834564911,
      "grad_norm": 3.0521374704073594,
      "learning_rate": 6.013626853004843e-08,
      "loss": 0.7102,
      "step": 6490
    },
    {
      "epoch": 0.6416726392012456,
      "grad_norm": 3.5952714126726097,
      "learning_rate": 6.010690372253328e-08,
      "loss": 0.7697,
      "step": 6491
    },
    {
      "epoch": 0.641771494946,
      "grad_norm": 4.312572611664246,
      "learning_rate": 6.007754300546668e-08,
      "loss": 0.7521,
      "step": 6492
    },
    {
      "epoch": 0.6418703506907545,
      "grad_norm": 6.4130060161791365,
      "learning_rate": 6.004818638185916e-08,
      "loss": 0.6598,
      "step": 6493
    },
    {
      "epoch": 0.641969206435509,
      "grad_norm": 4.060788060503483,
      "learning_rate": 6.001883385472077e-08,
      "loss": 0.6304,
      "step": 6494
    },
    {
      "epoch": 0.6420680621802635,
      "grad_norm": 4.392635727553592,
      "learning_rate": 5.998948542706123e-08,
      "loss": 0.618,
      "step": 6495
    },
    {
      "epoch": 0.6421669179250179,
      "grad_norm": 6.186133045217443,
      "learning_rate": 5.996014110188976e-08,
      "loss": 0.6552,
      "step": 6496
    },
    {
      "epoch": 0.6422657736697723,
      "grad_norm": 5.237087279533955,
      "learning_rate": 5.993080088221519e-08,
      "loss": 0.6513,
      "step": 6497
    },
    {
      "epoch": 0.6423646294145269,
      "grad_norm": 4.562725888709792,
      "learning_rate": 5.990146477104589e-08,
      "loss": 0.7975,
      "step": 6498
    },
    {
      "epoch": 0.6424634851592813,
      "grad_norm": 24.965414716769097,
      "learning_rate": 5.987213277138995e-08,
      "loss": 0.9008,
      "step": 6499
    },
    {
      "epoch": 0.6425623409040357,
      "grad_norm": 4.763959095270247,
      "learning_rate": 5.984280488625481e-08,
      "loss": 0.6259,
      "step": 6500
    },
    {
      "epoch": 0.6426611966487903,
      "grad_norm": 4.183606661318391,
      "learning_rate": 5.981348111864768e-08,
      "loss": 0.7465,
      "step": 6501
    },
    {
      "epoch": 0.6427600523935447,
      "grad_norm": 5.08239547546349,
      "learning_rate": 5.97841614715753e-08,
      "loss": 0.7009,
      "step": 6502
    },
    {
      "epoch": 0.6428589081382992,
      "grad_norm": 3.659542967389106,
      "learning_rate": 5.975484594804389e-08,
      "loss": 0.6974,
      "step": 6503
    },
    {
      "epoch": 0.6429577638830537,
      "grad_norm": 3.2933307662050715,
      "learning_rate": 5.972553455105942e-08,
      "loss": 0.7659,
      "step": 6504
    },
    {
      "epoch": 0.6430566196278081,
      "grad_norm": 4.055417574415319,
      "learning_rate": 5.969622728362726e-08,
      "loss": 0.7187,
      "step": 6505
    },
    {
      "epoch": 0.6431554753725626,
      "grad_norm": 4.277061550373825,
      "learning_rate": 5.966692414875246e-08,
      "loss": 0.6716,
      "step": 6506
    },
    {
      "epoch": 0.643254331117317,
      "grad_norm": 10.626899589241658,
      "learning_rate": 5.963762514943968e-08,
      "loss": 0.6978,
      "step": 6507
    },
    {
      "epoch": 0.6433531868620715,
      "grad_norm": 4.27106916509404,
      "learning_rate": 5.960833028869301e-08,
      "loss": 0.7462,
      "step": 6508
    },
    {
      "epoch": 0.643452042606826,
      "grad_norm": 3.497387967647567,
      "learning_rate": 5.9579039569516243e-08,
      "loss": 0.7545,
      "step": 6509
    },
    {
      "epoch": 0.6435508983515804,
      "grad_norm": 3.4922645794280784,
      "learning_rate": 5.954975299491276e-08,
      "loss": 0.7219,
      "step": 6510
    },
    {
      "epoch": 0.643649754096335,
      "grad_norm": 3.7741248696686407,
      "learning_rate": 5.952047056788538e-08,
      "loss": 0.6097,
      "step": 6511
    },
    {
      "epoch": 0.6437486098410894,
      "grad_norm": 3.5478487575501068,
      "learning_rate": 5.949119229143664e-08,
      "loss": 0.6645,
      "step": 6512
    },
    {
      "epoch": 0.6438474655858438,
      "grad_norm": 5.167513937746826,
      "learning_rate": 5.94619181685686e-08,
      "loss": 0.7129,
      "step": 6513
    },
    {
      "epoch": 0.6439463213305984,
      "grad_norm": 4.896478760074855,
      "learning_rate": 5.943264820228288e-08,
      "loss": 0.7461,
      "step": 6514
    },
    {
      "epoch": 0.6440451770753528,
      "grad_norm": 6.903734947605831,
      "learning_rate": 5.940338239558068e-08,
      "loss": 0.778,
      "step": 6515
    },
    {
      "epoch": 0.6441440328201072,
      "grad_norm": 6.167444476321172,
      "learning_rate": 5.937412075146273e-08,
      "loss": 0.6209,
      "step": 6516
    },
    {
      "epoch": 0.6442428885648617,
      "grad_norm": 9.17797202093982,
      "learning_rate": 5.9344863272929447e-08,
      "loss": 0.6478,
      "step": 6517
    },
    {
      "epoch": 0.6443417443096162,
      "grad_norm": 3.106626199676358,
      "learning_rate": 5.931560996298075e-08,
      "loss": 0.6689,
      "step": 6518
    },
    {
      "epoch": 0.6444406000543706,
      "grad_norm": 4.881725634296227,
      "learning_rate": 5.928636082461607e-08,
      "loss": 0.6944,
      "step": 6519
    },
    {
      "epoch": 0.6445394557991251,
      "grad_norm": 42.438642246419235,
      "learning_rate": 5.925711586083453e-08,
      "loss": 0.6626,
      "step": 6520
    },
    {
      "epoch": 0.6446383115438796,
      "grad_norm": 5.772256418439717,
      "learning_rate": 5.92278750746348e-08,
      "loss": 0.733,
      "step": 6521
    },
    {
      "epoch": 0.6447371672886341,
      "grad_norm": 4.882392707672826,
      "learning_rate": 5.919863846901503e-08,
      "loss": 0.6341,
      "step": 6522
    },
    {
      "epoch": 0.6448360230333885,
      "grad_norm": 4.843882959753123,
      "learning_rate": 5.916940604697299e-08,
      "loss": 0.7826,
      "step": 6523
    },
    {
      "epoch": 0.644934878778143,
      "grad_norm": 3.6826392875528233,
      "learning_rate": 5.914017781150611e-08,
      "loss": 0.5958,
      "step": 6524
    },
    {
      "epoch": 0.6450337345228975,
      "grad_norm": 3.539009557564371,
      "learning_rate": 5.911095376561124e-08,
      "loss": 0.6038,
      "step": 6525
    },
    {
      "epoch": 0.6451325902676519,
      "grad_norm": 3.8883727627772564,
      "learning_rate": 5.908173391228495e-08,
      "loss": 0.6077,
      "step": 6526
    },
    {
      "epoch": 0.6452314460124065,
      "grad_norm": 4.442433504110512,
      "learning_rate": 5.90525182545232e-08,
      "loss": 0.7237,
      "step": 6527
    },
    {
      "epoch": 0.6453303017571609,
      "grad_norm": 11.230025241999561,
      "learning_rate": 5.90233067953217e-08,
      "loss": 0.7279,
      "step": 6528
    },
    {
      "epoch": 0.6454291575019153,
      "grad_norm": 3.2253951532499125,
      "learning_rate": 5.899409953767568e-08,
      "loss": 0.827,
      "step": 6529
    },
    {
      "epoch": 0.6455280132466698,
      "grad_norm": 4.734233169503264,
      "learning_rate": 5.896489648457983e-08,
      "loss": 0.8131,
      "step": 6530
    },
    {
      "epoch": 0.6456268689914243,
      "grad_norm": 5.070455986855912,
      "learning_rate": 5.893569763902855e-08,
      "loss": 0.7049,
      "step": 6531
    },
    {
      "epoch": 0.6457257247361787,
      "grad_norm": 10.658132984072129,
      "learning_rate": 5.890650300401576e-08,
      "loss": 0.6156,
      "step": 6532
    },
    {
      "epoch": 0.6458245804809332,
      "grad_norm": 3.5558353668687714,
      "learning_rate": 5.887731258253488e-08,
      "loss": 0.5295,
      "step": 6533
    },
    {
      "epoch": 0.6459234362256877,
      "grad_norm": 9.435107515436814,
      "learning_rate": 5.884812637757899e-08,
      "loss": 0.7724,
      "step": 6534
    },
    {
      "epoch": 0.6460222919704421,
      "grad_norm": 5.105519594396979,
      "learning_rate": 5.881894439214073e-08,
      "loss": 0.6887,
      "step": 6535
    },
    {
      "epoch": 0.6461211477151966,
      "grad_norm": 4.399163402350806,
      "learning_rate": 5.878976662921225e-08,
      "loss": 0.7257,
      "step": 6536
    },
    {
      "epoch": 0.6462200034599511,
      "grad_norm": 6.00198354905455,
      "learning_rate": 5.8760593091785316e-08,
      "loss": 0.8015,
      "step": 6537
    },
    {
      "epoch": 0.6463188592047056,
      "grad_norm": 5.6869742446663745,
      "learning_rate": 5.873142378285122e-08,
      "loss": 0.6633,
      "step": 6538
    },
    {
      "epoch": 0.64641771494946,
      "grad_norm": 3.6159910542217824,
      "learning_rate": 5.870225870540087e-08,
      "loss": 0.7019,
      "step": 6539
    },
    {
      "epoch": 0.6465165706942144,
      "grad_norm": 3.57621084615235,
      "learning_rate": 5.8673097862424714e-08,
      "loss": 0.7272,
      "step": 6540
    },
    {
      "epoch": 0.646615426438969,
      "grad_norm": 3.84838087065922,
      "learning_rate": 5.8643941256912746e-08,
      "loss": 0.7566,
      "step": 6541
    },
    {
      "epoch": 0.6467142821837234,
      "grad_norm": 5.464650255363838,
      "learning_rate": 5.861478889185454e-08,
      "loss": 0.7003,
      "step": 6542
    },
    {
      "epoch": 0.6468131379284778,
      "grad_norm": 4.020733908662698,
      "learning_rate": 5.858564077023931e-08,
      "loss": 0.7082,
      "step": 6543
    },
    {
      "epoch": 0.6469119936732324,
      "grad_norm": 4.413259127028339,
      "learning_rate": 5.855649689505566e-08,
      "loss": 0.6288,
      "step": 6544
    },
    {
      "epoch": 0.6470108494179868,
      "grad_norm": 3.90848317509244,
      "learning_rate": 5.852735726929192e-08,
      "loss": 0.7784,
      "step": 6545
    },
    {
      "epoch": 0.6471097051627412,
      "grad_norm": 10.152985217822845,
      "learning_rate": 5.8498221895935984e-08,
      "loss": 0.7694,
      "step": 6546
    },
    {
      "epoch": 0.6472085609074958,
      "grad_norm": 9.866717954247566,
      "learning_rate": 5.846909077797514e-08,
      "loss": 0.69,
      "step": 6547
    },
    {
      "epoch": 0.6473074166522502,
      "grad_norm": 3.2543750685513713,
      "learning_rate": 5.843996391839646e-08,
      "loss": 0.6884,
      "step": 6548
    },
    {
      "epoch": 0.6474062723970047,
      "grad_norm": 3.447756892105157,
      "learning_rate": 5.841084132018641e-08,
      "loss": 0.6745,
      "step": 6549
    },
    {
      "epoch": 0.6475051281417591,
      "grad_norm": 2.967498192277288,
      "learning_rate": 5.838172298633107e-08,
      "loss": 0.7136,
      "step": 6550
    },
    {
      "epoch": 0.6476039838865136,
      "grad_norm": 5.7043387246543436,
      "learning_rate": 5.835260891981617e-08,
      "loss": 0.7134,
      "step": 6551
    },
    {
      "epoch": 0.6477028396312681,
      "grad_norm": 9.005153262854412,
      "learning_rate": 5.8323499123626843e-08,
      "loss": 0.8361,
      "step": 6552
    },
    {
      "epoch": 0.6478016953760225,
      "grad_norm": 4.4971677920183675,
      "learning_rate": 5.82943936007479e-08,
      "loss": 0.5962,
      "step": 6553
    },
    {
      "epoch": 0.647900551120777,
      "grad_norm": 3.224244434024139,
      "learning_rate": 5.826529235416375e-08,
      "loss": 0.8004,
      "step": 6554
    },
    {
      "epoch": 0.6479994068655315,
      "grad_norm": 5.05232567846715,
      "learning_rate": 5.823619538685819e-08,
      "loss": 0.8029,
      "step": 6555
    },
    {
      "epoch": 0.6480982626102859,
      "grad_norm": 3.8761261469191397,
      "learning_rate": 5.820710270181474e-08,
      "loss": 0.7354,
      "step": 6556
    },
    {
      "epoch": 0.6481971183550405,
      "grad_norm": 3.079727984980904,
      "learning_rate": 5.817801430201645e-08,
      "loss": 0.7491,
      "step": 6557
    },
    {
      "epoch": 0.6482959740997949,
      "grad_norm": 4.485350384868977,
      "learning_rate": 5.8148930190445847e-08,
      "loss": 0.738,
      "step": 6558
    },
    {
      "epoch": 0.6483948298445493,
      "grad_norm": 3.569395701197582,
      "learning_rate": 5.811985037008513e-08,
      "loss": 0.6325,
      "step": 6559
    },
    {
      "epoch": 0.6484936855893038,
      "grad_norm": 3.740286026969137,
      "learning_rate": 5.8090774843915947e-08,
      "loss": 0.7882,
      "step": 6560
    },
    {
      "epoch": 0.6485925413340583,
      "grad_norm": 5.038231161119399,
      "learning_rate": 5.8061703614919644e-08,
      "loss": 0.6525,
      "step": 6561
    },
    {
      "epoch": 0.6486913970788127,
      "grad_norm": 7.707653065973685,
      "learning_rate": 5.8032636686076965e-08,
      "loss": 0.7401,
      "step": 6562
    },
    {
      "epoch": 0.6487902528235672,
      "grad_norm": 3.5317569171205982,
      "learning_rate": 5.8003574060368353e-08,
      "loss": 0.6512,
      "step": 6563
    },
    {
      "epoch": 0.6488891085683217,
      "grad_norm": 3.1759835404646166,
      "learning_rate": 5.797451574077371e-08,
      "loss": 0.5716,
      "step": 6564
    },
    {
      "epoch": 0.6489879643130761,
      "grad_norm": 4.844690624423004,
      "learning_rate": 5.794546173027259e-08,
      "loss": 0.6852,
      "step": 6565
    },
    {
      "epoch": 0.6490868200578306,
      "grad_norm": 8.126028900642929,
      "learning_rate": 5.7916412031843985e-08,
      "loss": 0.7283,
      "step": 6566
    },
    {
      "epoch": 0.6491856758025851,
      "grad_norm": 3.24610222691711,
      "learning_rate": 5.788736664846655e-08,
      "loss": 0.7402,
      "step": 6567
    },
    {
      "epoch": 0.6492845315473396,
      "grad_norm": 7.17288530206866,
      "learning_rate": 5.785832558311849e-08,
      "loss": 0.7901,
      "step": 6568
    },
    {
      "epoch": 0.649383387292094,
      "grad_norm": 6.420706113658787,
      "learning_rate": 5.782928883877749e-08,
      "loss": 0.6779,
      "step": 6569
    },
    {
      "epoch": 0.6494822430368484,
      "grad_norm": 6.252471719352112,
      "learning_rate": 5.780025641842088e-08,
      "loss": 0.6122,
      "step": 6570
    },
    {
      "epoch": 0.649581098781603,
      "grad_norm": 9.832590061246103,
      "learning_rate": 5.777122832502544e-08,
      "loss": 0.7687,
      "step": 6571
    },
    {
      "epoch": 0.6496799545263574,
      "grad_norm": 2.807221496512729,
      "learning_rate": 5.774220456156763e-08,
      "loss": 0.7633,
      "step": 6572
    },
    {
      "epoch": 0.6497788102711118,
      "grad_norm": 4.062038456568383,
      "learning_rate": 5.7713185131023435e-08,
      "loss": 0.7225,
      "step": 6573
    },
    {
      "epoch": 0.6498776660158664,
      "grad_norm": 5.7771650079473496,
      "learning_rate": 5.768417003636828e-08,
      "loss": 0.675,
      "step": 6574
    },
    {
      "epoch": 0.6499765217606208,
      "grad_norm": 4.266881192923484,
      "learning_rate": 5.765515928057729e-08,
      "loss": 0.651,
      "step": 6575
    },
    {
      "epoch": 0.6500753775053753,
      "grad_norm": 3.664575109830586,
      "learning_rate": 5.7626152866625135e-08,
      "loss": 0.7167,
      "step": 6576
    },
    {
      "epoch": 0.6501742332501298,
      "grad_norm": 4.110501624296853,
      "learning_rate": 5.7597150797485905e-08,
      "loss": 0.7971,
      "step": 6577
    },
    {
      "epoch": 0.6502730889948842,
      "grad_norm": 4.378608983609914,
      "learning_rate": 5.756815307613341e-08,
      "loss": 0.642,
      "step": 6578
    },
    {
      "epoch": 0.6503719447396387,
      "grad_norm": 4.819508230011718,
      "learning_rate": 5.753915970554085e-08,
      "loss": 0.7512,
      "step": 6579
    },
    {
      "epoch": 0.6504708004843931,
      "grad_norm": 3.182412232648986,
      "learning_rate": 5.751017068868118e-08,
      "loss": 0.7639,
      "step": 6580
    },
    {
      "epoch": 0.6505696562291476,
      "grad_norm": 8.213687734841196,
      "learning_rate": 5.748118602852668e-08,
      "loss": 0.759,
      "step": 6581
    },
    {
      "epoch": 0.6506685119739021,
      "grad_norm": 3.7369547214854157,
      "learning_rate": 5.745220572804941e-08,
      "loss": 0.6639,
      "step": 6582
    },
    {
      "epoch": 0.6507673677186565,
      "grad_norm": 6.041745222547543,
      "learning_rate": 5.742322979022076e-08,
      "loss": 0.7029,
      "step": 6583
    },
    {
      "epoch": 0.650866223463411,
      "grad_norm": 3.0137967526597707,
      "learning_rate": 5.73942582180119e-08,
      "loss": 0.6633,
      "step": 6584
    },
    {
      "epoch": 0.6509650792081655,
      "grad_norm": 3.695073792941132,
      "learning_rate": 5.7365291014393334e-08,
      "loss": 0.693,
      "step": 6585
    },
    {
      "epoch": 0.6510639349529199,
      "grad_norm": 4.540839714056835,
      "learning_rate": 5.733632818233528e-08,
      "loss": 0.7535,
      "step": 6586
    },
    {
      "epoch": 0.6511627906976745,
      "grad_norm": 5.7605482579052545,
      "learning_rate": 5.7307369724807476e-08,
      "loss": 0.6559,
      "step": 6587
    },
    {
      "epoch": 0.6512616464424289,
      "grad_norm": 3.1298386030881393,
      "learning_rate": 5.727841564477909e-08,
      "loss": 0.7257,
      "step": 6588
    },
    {
      "epoch": 0.6513605021871833,
      "grad_norm": 13.857883985502385,
      "learning_rate": 5.7249465945219e-08,
      "loss": 0.7683,
      "step": 6589
    },
    {
      "epoch": 0.6514593579319378,
      "grad_norm": 3.9873735287582246,
      "learning_rate": 5.7220520629095604e-08,
      "loss": 0.7648,
      "step": 6590
    },
    {
      "epoch": 0.6515582136766923,
      "grad_norm": 3.6633109353701303,
      "learning_rate": 5.7191579699376736e-08,
      "loss": 0.73,
      "step": 6591
    },
    {
      "epoch": 0.6516570694214467,
      "grad_norm": 4.07605565177372,
      "learning_rate": 5.71626431590299e-08,
      "loss": 0.7466,
      "step": 6592
    },
    {
      "epoch": 0.6517559251662012,
      "grad_norm": 2.8545320572611503,
      "learning_rate": 5.713371101102216e-08,
      "loss": 0.6904,
      "step": 6593
    },
    {
      "epoch": 0.6518547809109557,
      "grad_norm": 3.927951716471381,
      "learning_rate": 5.7104783258319977e-08,
      "loss": 0.6475,
      "step": 6594
    },
    {
      "epoch": 0.6519536366557102,
      "grad_norm": 4.179723301917675,
      "learning_rate": 5.707585990388957e-08,
      "loss": 0.6155,
      "step": 6595
    },
    {
      "epoch": 0.6520524924004646,
      "grad_norm": 3.140564880290087,
      "learning_rate": 5.704694095069652e-08,
      "loss": 0.7917,
      "step": 6596
    },
    {
      "epoch": 0.6521513481452191,
      "grad_norm": 3.4615098731525826,
      "learning_rate": 5.701802640170611e-08,
      "loss": 0.6747,
      "step": 6597
    },
    {
      "epoch": 0.6522502038899736,
      "grad_norm": 24.18937743896222,
      "learning_rate": 5.698911625988306e-08,
      "loss": 0.7949,
      "step": 6598
    },
    {
      "epoch": 0.652349059634728,
      "grad_norm": 5.800519546772479,
      "learning_rate": 5.696021052819164e-08,
      "loss": 0.6025,
      "step": 6599
    },
    {
      "epoch": 0.6524479153794824,
      "grad_norm": 3.8920093315256983,
      "learning_rate": 5.6931309209595766e-08,
      "loss": 0.7015,
      "step": 6600
    },
    {
      "epoch": 0.652546771124237,
      "grad_norm": 6.350962104727854,
      "learning_rate": 5.6902412307058855e-08,
      "loss": 0.7892,
      "step": 6601
    },
    {
      "epoch": 0.6526456268689914,
      "grad_norm": 3.755972410226422,
      "learning_rate": 5.687351982354378e-08,
      "loss": 0.7197,
      "step": 6602
    },
    {
      "epoch": 0.6527444826137458,
      "grad_norm": 3.423339335265188,
      "learning_rate": 5.684463176201308e-08,
      "loss": 0.599,
      "step": 6603
    },
    {
      "epoch": 0.6528433383585004,
      "grad_norm": 4.290524552626333,
      "learning_rate": 5.681574812542887e-08,
      "loss": 0.6615,
      "step": 6604
    },
    {
      "epoch": 0.6529421941032548,
      "grad_norm": 3.1246216251968852,
      "learning_rate": 5.6786868916752615e-08,
      "loss": 0.712,
      "step": 6605
    },
    {
      "epoch": 0.6530410498480093,
      "grad_norm": 4.6768083002398075,
      "learning_rate": 5.675799413894557e-08,
      "loss": 0.7698,
      "step": 6606
    },
    {
      "epoch": 0.6531399055927638,
      "grad_norm": 3.8541974917265636,
      "learning_rate": 5.672912379496833e-08,
      "loss": 0.7396,
      "step": 6607
    },
    {
      "epoch": 0.6532387613375182,
      "grad_norm": 3.9902344358404096,
      "learning_rate": 5.670025788778113e-08,
      "loss": 0.7225,
      "step": 6608
    },
    {
      "epoch": 0.6533376170822727,
      "grad_norm": 5.675479736375696,
      "learning_rate": 5.667139642034382e-08,
      "loss": 0.672,
      "step": 6609
    },
    {
      "epoch": 0.6534364728270272,
      "grad_norm": 5.507322172615904,
      "learning_rate": 5.664253939561563e-08,
      "loss": 0.6401,
      "step": 6610
    },
    {
      "epoch": 0.6535353285717816,
      "grad_norm": 6.681148729906989,
      "learning_rate": 5.661368681655545e-08,
      "loss": 0.8024,
      "step": 6611
    },
    {
      "epoch": 0.6536341843165361,
      "grad_norm": 4.241971000295251,
      "learning_rate": 5.658483868612173e-08,
      "loss": 0.7631,
      "step": 6612
    },
    {
      "epoch": 0.6537330400612905,
      "grad_norm": 3.6680550140206827,
      "learning_rate": 5.655599500727235e-08,
      "loss": 0.7187,
      "step": 6613
    },
    {
      "epoch": 0.6538318958060451,
      "grad_norm": 6.675880275448662,
      "learning_rate": 5.652715578296488e-08,
      "loss": 0.7161,
      "step": 6614
    },
    {
      "epoch": 0.6539307515507995,
      "grad_norm": 5.001396819240162,
      "learning_rate": 5.64983210161563e-08,
      "loss": 0.6874,
      "step": 6615
    },
    {
      "epoch": 0.6540296072955539,
      "grad_norm": 4.294958081435149,
      "learning_rate": 5.646949070980318e-08,
      "loss": 0.8579,
      "step": 6616
    },
    {
      "epoch": 0.6541284630403085,
      "grad_norm": 4.409366906265974,
      "learning_rate": 5.64406648668617e-08,
      "loss": 0.6468,
      "step": 6617
    },
    {
      "epoch": 0.6542273187850629,
      "grad_norm": 3.368601110768223,
      "learning_rate": 5.641184349028747e-08,
      "loss": 0.6216,
      "step": 6618
    },
    {
      "epoch": 0.6543261745298173,
      "grad_norm": 3.205988444732904,
      "learning_rate": 5.63830265830357e-08,
      "loss": 0.7164,
      "step": 6619
    },
    {
      "epoch": 0.6544250302745719,
      "grad_norm": 3.6998914405323635,
      "learning_rate": 5.6354214148061205e-08,
      "loss": 0.7293,
      "step": 6620
    },
    {
      "epoch": 0.6545238860193263,
      "grad_norm": 4.328171058643871,
      "learning_rate": 5.632540618831818e-08,
      "loss": 0.7744,
      "step": 6621
    },
    {
      "epoch": 0.6546227417640808,
      "grad_norm": 3.3848451908408737,
      "learning_rate": 5.629660270676051e-08,
      "loss": 0.709,
      "step": 6622
    },
    {
      "epoch": 0.6547215975088352,
      "grad_norm": 3.804742458976016,
      "learning_rate": 5.626780370634161e-08,
      "loss": 0.7115,
      "step": 6623
    },
    {
      "epoch": 0.6548204532535897,
      "grad_norm": 4.635032443572569,
      "learning_rate": 5.6239009190014297e-08,
      "loss": 0.6999,
      "step": 6624
    },
    {
      "epoch": 0.6549193089983442,
      "grad_norm": 7.269119753247966,
      "learning_rate": 5.621021916073108e-08,
      "loss": 0.8301,
      "step": 6625
    },
    {
      "epoch": 0.6550181647430986,
      "grad_norm": 5.094045449840713,
      "learning_rate": 5.6181433621443974e-08,
      "loss": 0.7268,
      "step": 6626
    },
    {
      "epoch": 0.6551170204878531,
      "grad_norm": 3.859522329814548,
      "learning_rate": 5.615265257510444e-08,
      "loss": 0.7998,
      "step": 6627
    },
    {
      "epoch": 0.6552158762326076,
      "grad_norm": 12.222383852521206,
      "learning_rate": 5.612387602466364e-08,
      "loss": 0.7243,
      "step": 6628
    },
    {
      "epoch": 0.655314731977362,
      "grad_norm": 4.6575014951615525,
      "learning_rate": 5.609510397307209e-08,
      "loss": 0.6645,
      "step": 6629
    },
    {
      "epoch": 0.6554135877221166,
      "grad_norm": 4.7088107413095095,
      "learning_rate": 5.606633642328003e-08,
      "loss": 0.6557,
      "step": 6630
    },
    {
      "epoch": 0.655512443466871,
      "grad_norm": 3.6332271986886187,
      "learning_rate": 5.6037573378237045e-08,
      "loss": 0.6921,
      "step": 6631
    },
    {
      "epoch": 0.6556112992116254,
      "grad_norm": 3.509381485555218,
      "learning_rate": 5.6008814840892484e-08,
      "loss": 0.6068,
      "step": 6632
    },
    {
      "epoch": 0.6557101549563799,
      "grad_norm": 6.80605661812052,
      "learning_rate": 5.598006081419498e-08,
      "loss": 0.7931,
      "step": 6633
    },
    {
      "epoch": 0.6558090107011344,
      "grad_norm": 4.34009183505453,
      "learning_rate": 5.5951311301092955e-08,
      "loss": 0.629,
      "step": 6634
    },
    {
      "epoch": 0.6559078664458888,
      "grad_norm": 145.63758619303948,
      "learning_rate": 5.5922566304534144e-08,
      "loss": 0.7253,
      "step": 6635
    },
    {
      "epoch": 0.6560067221906433,
      "grad_norm": 3.079590707019999,
      "learning_rate": 5.589382582746598e-08,
      "loss": 0.666,
      "step": 6636
    },
    {
      "epoch": 0.6561055779353978,
      "grad_norm": 3.92505507503517,
      "learning_rate": 5.58650898728354e-08,
      "loss": 0.6275,
      "step": 6637
    },
    {
      "epoch": 0.6562044336801522,
      "grad_norm": 4.020464408465863,
      "learning_rate": 5.5836358443588785e-08,
      "loss": 0.7323,
      "step": 6638
    },
    {
      "epoch": 0.6563032894249067,
      "grad_norm": 5.572547388924002,
      "learning_rate": 5.580763154267218e-08,
      "loss": 0.7013,
      "step": 6639
    },
    {
      "epoch": 0.6564021451696612,
      "grad_norm": 5.0633553194692755,
      "learning_rate": 5.5778909173031044e-08,
      "loss": 0.6215,
      "step": 6640
    },
    {
      "epoch": 0.6565010009144157,
      "grad_norm": 7.348870561947552,
      "learning_rate": 5.575019133761046e-08,
      "loss": 0.7301,
      "step": 6641
    },
    {
      "epoch": 0.6565998566591701,
      "grad_norm": 3.679799119860612,
      "learning_rate": 5.5721478039355076e-08,
      "loss": 0.7391,
      "step": 6642
    },
    {
      "epoch": 0.6566987124039245,
      "grad_norm": 3.500588295984293,
      "learning_rate": 5.569276928120892e-08,
      "loss": 0.7143,
      "step": 6643
    },
    {
      "epoch": 0.6567975681486791,
      "grad_norm": 4.1070943119419105,
      "learning_rate": 5.5664065066115685e-08,
      "loss": 0.6597,
      "step": 6644
    },
    {
      "epoch": 0.6568964238934335,
      "grad_norm": 3.3475519987480165,
      "learning_rate": 5.563536539701863e-08,
      "loss": 0.5269,
      "step": 6645
    },
    {
      "epoch": 0.6569952796381879,
      "grad_norm": 5.339517316091903,
      "learning_rate": 5.560667027686038e-08,
      "loss": 0.7679,
      "step": 6646
    },
    {
      "epoch": 0.6570941353829425,
      "grad_norm": 4.645189692176482,
      "learning_rate": 5.557797970858329e-08,
      "loss": 0.7332,
      "step": 6647
    },
    {
      "epoch": 0.6571929911276969,
      "grad_norm": 4.685681982763314,
      "learning_rate": 5.5549293695129064e-08,
      "loss": 0.7926,
      "step": 6648
    },
    {
      "epoch": 0.6572918468724513,
      "grad_norm": 4.637700469772344,
      "learning_rate": 5.552061223943912e-08,
      "loss": 0.8,
      "step": 6649
    },
    {
      "epoch": 0.6573907026172059,
      "grad_norm": 7.454173026709327,
      "learning_rate": 5.549193534445425e-08,
      "loss": 0.5874,
      "step": 6650
    },
    {
      "epoch": 0.6574895583619603,
      "grad_norm": 3.361228536934839,
      "learning_rate": 5.546326301311488e-08,
      "loss": 0.8492,
      "step": 6651
    },
    {
      "epoch": 0.6575884141067148,
      "grad_norm": 9.171976025493011,
      "learning_rate": 5.5434595248360906e-08,
      "loss": 0.6859,
      "step": 6652
    },
    {
      "epoch": 0.6576872698514692,
      "grad_norm": 3.8410310624189696,
      "learning_rate": 5.5405932053131834e-08,
      "loss": 0.6456,
      "step": 6653
    },
    {
      "epoch": 0.6577861255962237,
      "grad_norm": 3.049508060402196,
      "learning_rate": 5.537727343036658e-08,
      "loss": 0.6938,
      "step": 6654
    },
    {
      "epoch": 0.6578849813409782,
      "grad_norm": 4.837043752267119,
      "learning_rate": 5.534861938300371e-08,
      "loss": 0.7803,
      "step": 6655
    },
    {
      "epoch": 0.6579838370857326,
      "grad_norm": 3.652155587438829,
      "learning_rate": 5.5319969913981315e-08,
      "loss": 0.7641,
      "step": 6656
    },
    {
      "epoch": 0.6580826928304871,
      "grad_norm": 4.507283987901102,
      "learning_rate": 5.529132502623688e-08,
      "loss": 0.5167,
      "step": 6657
    },
    {
      "epoch": 0.6581815485752416,
      "grad_norm": 3.630754632681864,
      "learning_rate": 5.5262684722707564e-08,
      "loss": 0.6465,
      "step": 6658
    },
    {
      "epoch": 0.658280404319996,
      "grad_norm": 4.415447095278395,
      "learning_rate": 5.523404900633004e-08,
      "loss": 0.718,
      "step": 6659
    },
    {
      "epoch": 0.6583792600647506,
      "grad_norm": 3.8189958706613614,
      "learning_rate": 5.520541788004042e-08,
      "loss": 0.6904,
      "step": 6660
    },
    {
      "epoch": 0.658478115809505,
      "grad_norm": 4.534146437697531,
      "learning_rate": 5.517679134677446e-08,
      "loss": 0.7339,
      "step": 6661
    },
    {
      "epoch": 0.6585769715542594,
      "grad_norm": 3.6673094293964223,
      "learning_rate": 5.5148169409467317e-08,
      "loss": 0.7711,
      "step": 6662
    },
    {
      "epoch": 0.6586758272990139,
      "grad_norm": 8.827378519442334,
      "learning_rate": 5.51195520710538e-08,
      "loss": 0.6305,
      "step": 6663
    },
    {
      "epoch": 0.6587746830437684,
      "grad_norm": 5.465121191151299,
      "learning_rate": 5.509093933446822e-08,
      "loss": 0.7247,
      "step": 6664
    },
    {
      "epoch": 0.6588735387885228,
      "grad_norm": 3.1212084137016882,
      "learning_rate": 5.506233120264432e-08,
      "loss": 0.7097,
      "step": 6665
    },
    {
      "epoch": 0.6589723945332773,
      "grad_norm": 4.2926051055695345,
      "learning_rate": 5.503372767851551e-08,
      "loss": 0.7539,
      "step": 6666
    },
    {
      "epoch": 0.6590712502780318,
      "grad_norm": 13.87567208627923,
      "learning_rate": 5.50051287650146e-08,
      "loss": 0.605,
      "step": 6667
    },
    {
      "epoch": 0.6591701060227863,
      "grad_norm": 5.046959270333449,
      "learning_rate": 5.497653446507404e-08,
      "loss": 0.8018,
      "step": 6668
    },
    {
      "epoch": 0.6592689617675407,
      "grad_norm": 4.281588619876789,
      "learning_rate": 5.49479447816257e-08,
      "loss": 0.707,
      "step": 6669
    },
    {
      "epoch": 0.6593678175122952,
      "grad_norm": 3.5107722721360424,
      "learning_rate": 5.491935971760109e-08,
      "loss": 0.7047,
      "step": 6670
    },
    {
      "epoch": 0.6594666732570497,
      "grad_norm": 5.721236485242851,
      "learning_rate": 5.489077927593111e-08,
      "loss": 0.719,
      "step": 6671
    },
    {
      "epoch": 0.6595655290018041,
      "grad_norm": 5.280059719979246,
      "learning_rate": 5.486220345954635e-08,
      "loss": 0.6649,
      "step": 6672
    },
    {
      "epoch": 0.6596643847465585,
      "grad_norm": 6.331957336630566,
      "learning_rate": 5.4833632271376765e-08,
      "loss": 0.6673,
      "step": 6673
    },
    {
      "epoch": 0.6597632404913131,
      "grad_norm": 7.689295036137141,
      "learning_rate": 5.480506571435193e-08,
      "loss": 0.6193,
      "step": 6674
    },
    {
      "epoch": 0.6598620962360675,
      "grad_norm": 4.000480239150988,
      "learning_rate": 5.477650379140097e-08,
      "loss": 0.7635,
      "step": 6675
    },
    {
      "epoch": 0.6599609519808219,
      "grad_norm": 17.081719789805117,
      "learning_rate": 5.4747946505452437e-08,
      "loss": 0.648,
      "step": 6676
    },
    {
      "epoch": 0.6600598077255765,
      "grad_norm": 2.9995817272118956,
      "learning_rate": 5.471939385943445e-08,
      "loss": 0.6871,
      "step": 6677
    },
    {
      "epoch": 0.6601586634703309,
      "grad_norm": 3.610094409180275,
      "learning_rate": 5.469084585627475e-08,
      "loss": 0.7708,
      "step": 6678
    },
    {
      "epoch": 0.6602575192150854,
      "grad_norm": 4.722887626359667,
      "learning_rate": 5.466230249890038e-08,
      "loss": 0.6968,
      "step": 6679
    },
    {
      "epoch": 0.6603563749598399,
      "grad_norm": 3.3820850579170694,
      "learning_rate": 5.463376379023814e-08,
      "loss": 0.7486,
      "step": 6680
    },
    {
      "epoch": 0.6604552307045943,
      "grad_norm": 4.779615838499907,
      "learning_rate": 5.460522973321424e-08,
      "loss": 0.733,
      "step": 6681
    },
    {
      "epoch": 0.6605540864493488,
      "grad_norm": 4.673893277640158,
      "learning_rate": 5.457670033075439e-08,
      "loss": 0.7158,
      "step": 6682
    },
    {
      "epoch": 0.6606529421941033,
      "grad_norm": 2.831506343490153,
      "learning_rate": 5.4548175585783905e-08,
      "loss": 0.6226,
      "step": 6683
    },
    {
      "epoch": 0.6607517979388577,
      "grad_norm": 4.223456994983366,
      "learning_rate": 5.4519655501227525e-08,
      "loss": 0.6282,
      "step": 6684
    },
    {
      "epoch": 0.6608506536836122,
      "grad_norm": 2.927438881574786,
      "learning_rate": 5.449114008000962e-08,
      "loss": 0.6375,
      "step": 6685
    },
    {
      "epoch": 0.6609495094283666,
      "grad_norm": 3.6075153838735092,
      "learning_rate": 5.446262932505401e-08,
      "loss": 0.5984,
      "step": 6686
    },
    {
      "epoch": 0.6610483651731212,
      "grad_norm": 5.765833421856698,
      "learning_rate": 5.443412323928398e-08,
      "loss": 0.7899,
      "step": 6687
    },
    {
      "epoch": 0.6611472209178756,
      "grad_norm": 11.710287352056758,
      "learning_rate": 5.4405621825622495e-08,
      "loss": 0.5728,
      "step": 6688
    },
    {
      "epoch": 0.66124607666263,
      "grad_norm": 5.1658009294261085,
      "learning_rate": 5.4377125086991946e-08,
      "loss": 0.825,
      "step": 6689
    },
    {
      "epoch": 0.6613449324073846,
      "grad_norm": 4.186356871553078,
      "learning_rate": 5.434863302631421e-08,
      "loss": 0.7078,
      "step": 6690
    },
    {
      "epoch": 0.661443788152139,
      "grad_norm": 3.9542103824062655,
      "learning_rate": 5.432014564651073e-08,
      "loss": 0.7571,
      "step": 6691
    },
    {
      "epoch": 0.6615426438968934,
      "grad_norm": 5.3962978519305835,
      "learning_rate": 5.4291662950502534e-08,
      "loss": 0.7389,
      "step": 6692
    },
    {
      "epoch": 0.661641499641648,
      "grad_norm": 3.204008098484777,
      "learning_rate": 5.426318494121003e-08,
      "loss": 0.6519,
      "step": 6693
    },
    {
      "epoch": 0.6617403553864024,
      "grad_norm": 5.37393861788962,
      "learning_rate": 5.4234711621553286e-08,
      "loss": 0.7114,
      "step": 6694
    },
    {
      "epoch": 0.6618392111311568,
      "grad_norm": 5.178694733917559,
      "learning_rate": 5.420624299445173e-08,
      "loss": 0.687,
      "step": 6695
    },
    {
      "epoch": 0.6619380668759113,
      "grad_norm": 3.6982257106986474,
      "learning_rate": 5.417777906282445e-08,
      "loss": 0.6679,
      "step": 6696
    },
    {
      "epoch": 0.6620369226206658,
      "grad_norm": 2.936303312010156,
      "learning_rate": 5.414931982959005e-08,
      "loss": 0.6741,
      "step": 6697
    },
    {
      "epoch": 0.6621357783654203,
      "grad_norm": 8.295375689509793,
      "learning_rate": 5.412086529766651e-08,
      "loss": 0.7356,
      "step": 6698
    },
    {
      "epoch": 0.6622346341101747,
      "grad_norm": 8.230119346607147,
      "learning_rate": 5.409241546997146e-08,
      "loss": 0.6095,
      "step": 6699
    },
    {
      "epoch": 0.6623334898549292,
      "grad_norm": 4.7780774120361045,
      "learning_rate": 5.406397034942207e-08,
      "loss": 0.6397,
      "step": 6700
    },
    {
      "epoch": 0.6624323455996837,
      "grad_norm": 10.90626277701858,
      "learning_rate": 5.4035529938934924e-08,
      "loss": 0.7011,
      "step": 6701
    },
    {
      "epoch": 0.6625312013444381,
      "grad_norm": 13.079605774736509,
      "learning_rate": 5.40070942414261e-08,
      "loss": 0.7079,
      "step": 6702
    },
    {
      "epoch": 0.6626300570891926,
      "grad_norm": 3.2431121305150006,
      "learning_rate": 5.3978663259811374e-08,
      "loss": 0.7387,
      "step": 6703
    },
    {
      "epoch": 0.6627289128339471,
      "grad_norm": 5.103224978675987,
      "learning_rate": 5.3950236997005814e-08,
      "loss": 0.7099,
      "step": 6704
    },
    {
      "epoch": 0.6628277685787015,
      "grad_norm": 6.245254496651759,
      "learning_rate": 5.392181545592423e-08,
      "loss": 0.7672,
      "step": 6705
    },
    {
      "epoch": 0.662926624323456,
      "grad_norm": 3.7899643123841336,
      "learning_rate": 5.3893398639480705e-08,
      "loss": 0.7761,
      "step": 6706
    },
    {
      "epoch": 0.6630254800682105,
      "grad_norm": 3.923130698650747,
      "learning_rate": 5.386498655058904e-08,
      "loss": 0.6512,
      "step": 6707
    },
    {
      "epoch": 0.6631243358129649,
      "grad_norm": 5.271545825101287,
      "learning_rate": 5.383657919216251e-08,
      "loss": 0.6748,
      "step": 6708
    },
    {
      "epoch": 0.6632231915577194,
      "grad_norm": 6.51764594875741,
      "learning_rate": 5.380817656711378e-08,
      "loss": 0.7825,
      "step": 6709
    },
    {
      "epoch": 0.6633220473024739,
      "grad_norm": 7.103833863614909,
      "learning_rate": 5.377977867835517e-08,
      "loss": 0.6775,
      "step": 6710
    },
    {
      "epoch": 0.6634209030472283,
      "grad_norm": 5.294886270756204,
      "learning_rate": 5.3751385528798496e-08,
      "loss": 0.6566,
      "step": 6711
    },
    {
      "epoch": 0.6635197587919828,
      "grad_norm": 7.625777983818856,
      "learning_rate": 5.372299712135497e-08,
      "loss": 0.6932,
      "step": 6712
    },
    {
      "epoch": 0.6636186145367373,
      "grad_norm": 3.900991956430983,
      "learning_rate": 5.369461345893548e-08,
      "loss": 0.6964,
      "step": 6713
    },
    {
      "epoch": 0.6637174702814918,
      "grad_norm": 4.832581465427084,
      "learning_rate": 5.366623454445035e-08,
      "loss": 0.6388,
      "step": 6714
    },
    {
      "epoch": 0.6638163260262462,
      "grad_norm": 3.0584650737084154,
      "learning_rate": 5.3637860380809366e-08,
      "loss": 0.7081,
      "step": 6715
    },
    {
      "epoch": 0.6639151817710006,
      "grad_norm": 3.2095551979055537,
      "learning_rate": 5.3609490970921936e-08,
      "loss": 0.695,
      "step": 6716
    },
    {
      "epoch": 0.6640140375157552,
      "grad_norm": 7.956383656781262,
      "learning_rate": 5.358112631769687e-08,
      "loss": 0.7015,
      "step": 6717
    },
    {
      "epoch": 0.6641128932605096,
      "grad_norm": 3.2993593768481793,
      "learning_rate": 5.3552766424042615e-08,
      "loss": 0.751,
      "step": 6718
    },
    {
      "epoch": 0.664211749005264,
      "grad_norm": 9.22959414390837,
      "learning_rate": 5.352441129286698e-08,
      "loss": 0.6258,
      "step": 6719
    },
    {
      "epoch": 0.6643106047500186,
      "grad_norm": 7.462663106592477,
      "learning_rate": 5.349606092707743e-08,
      "loss": 0.7183,
      "step": 6720
    },
    {
      "epoch": 0.664409460494773,
      "grad_norm": 3.4798346352741154,
      "learning_rate": 5.346771532958083e-08,
      "loss": 0.6721,
      "step": 6721
    },
    {
      "epoch": 0.6645083162395274,
      "grad_norm": 3.9204322294949647,
      "learning_rate": 5.3439374503283644e-08,
      "loss": 0.7589,
      "step": 6722
    },
    {
      "epoch": 0.664607171984282,
      "grad_norm": 7.778321217831362,
      "learning_rate": 5.3411038451091763e-08,
      "loss": 0.7063,
      "step": 6723
    },
    {
      "epoch": 0.6647060277290364,
      "grad_norm": 23.901329207195484,
      "learning_rate": 5.3382707175910636e-08,
      "loss": 0.6886,
      "step": 6724
    },
    {
      "epoch": 0.6648048834737909,
      "grad_norm": 5.58819900601528,
      "learning_rate": 5.335438068064528e-08,
      "loss": 0.6985,
      "step": 6725
    },
    {
      "epoch": 0.6649037392185453,
      "grad_norm": 4.606073810574221,
      "learning_rate": 5.332605896820008e-08,
      "loss": 0.7136,
      "step": 6726
    },
    {
      "epoch": 0.6650025949632998,
      "grad_norm": 3.5518976583201183,
      "learning_rate": 5.329774204147908e-08,
      "loss": 0.6674,
      "step": 6727
    },
    {
      "epoch": 0.6651014507080543,
      "grad_norm": 4.448883084877687,
      "learning_rate": 5.3269429903385685e-08,
      "loss": 0.6509,
      "step": 6728
    },
    {
      "epoch": 0.6652003064528087,
      "grad_norm": 4.757073410811336,
      "learning_rate": 5.3241122556822935e-08,
      "loss": 0.643,
      "step": 6729
    },
    {
      "epoch": 0.6652991621975632,
      "grad_norm": 6.115019030258822,
      "learning_rate": 5.321282000469336e-08,
      "loss": 0.6429,
      "step": 6730
    },
    {
      "epoch": 0.6653980179423177,
      "grad_norm": 4.405555246533661,
      "learning_rate": 5.318452224989889e-08,
      "loss": 0.6952,
      "step": 6731
    },
    {
      "epoch": 0.6654968736870721,
      "grad_norm": 2.8612379345927823,
      "learning_rate": 5.315622929534109e-08,
      "loss": 0.7867,
      "step": 6732
    },
    {
      "epoch": 0.6655957294318267,
      "grad_norm": 5.794354467530149,
      "learning_rate": 5.312794114392103e-08,
      "loss": 0.5582,
      "step": 6733
    },
    {
      "epoch": 0.6656945851765811,
      "grad_norm": 6.819754001104054,
      "learning_rate": 5.309965779853915e-08,
      "loss": 0.6935,
      "step": 6734
    },
    {
      "epoch": 0.6657934409213355,
      "grad_norm": 7.585053969541475,
      "learning_rate": 5.307137926209557e-08,
      "loss": 0.7151,
      "step": 6735
    },
    {
      "epoch": 0.66589229666609,
      "grad_norm": 3.383240652383676,
      "learning_rate": 5.3043105537489764e-08,
      "loss": 0.635,
      "step": 6736
    },
    {
      "epoch": 0.6659911524108445,
      "grad_norm": 4.958498991872869,
      "learning_rate": 5.301483662762085e-08,
      "loss": 0.6732,
      "step": 6737
    },
    {
      "epoch": 0.6660900081555989,
      "grad_norm": 5.470041455548959,
      "learning_rate": 5.2986572535387343e-08,
      "loss": 0.8272,
      "step": 6738
    },
    {
      "epoch": 0.6661888639003534,
      "grad_norm": 35.15042824935367,
      "learning_rate": 5.2958313263687346e-08,
      "loss": 0.5697,
      "step": 6739
    },
    {
      "epoch": 0.6662877196451079,
      "grad_norm": 8.500426293830461,
      "learning_rate": 5.2930058815418386e-08,
      "loss": 0.6636,
      "step": 6740
    },
    {
      "epoch": 0.6663865753898623,
      "grad_norm": 3.3516527176877906,
      "learning_rate": 5.2901809193477586e-08,
      "loss": 0.7141,
      "step": 6741
    },
    {
      "epoch": 0.6664854311346168,
      "grad_norm": 4.9408431966948525,
      "learning_rate": 5.287356440076147e-08,
      "loss": 0.724,
      "step": 6742
    },
    {
      "epoch": 0.6665842868793713,
      "grad_norm": 5.552864779419335,
      "learning_rate": 5.2845324440166175e-08,
      "loss": 0.7145,
      "step": 6743
    },
    {
      "epoch": 0.6666831426241258,
      "grad_norm": 5.832433254921617,
      "learning_rate": 5.28170893145873e-08,
      "loss": 0.7185,
      "step": 6744
    },
    {
      "epoch": 0.6667819983688802,
      "grad_norm": 56.326360388161504,
      "learning_rate": 5.2788859026919893e-08,
      "loss": 0.6774,
      "step": 6745
    },
    {
      "epoch": 0.6668808541136346,
      "grad_norm": 4.149143964798494,
      "learning_rate": 5.2760633580058575e-08,
      "loss": 0.6959,
      "step": 6746
    },
    {
      "epoch": 0.6669797098583892,
      "grad_norm": 3.888407298099877,
      "learning_rate": 5.27324129768975e-08,
      "loss": 0.6002,
      "step": 6747
    },
    {
      "epoch": 0.6670785656031436,
      "grad_norm": 6.129426366986402,
      "learning_rate": 5.270419722033017e-08,
      "loss": 0.7233,
      "step": 6748
    },
    {
      "epoch": 0.667177421347898,
      "grad_norm": 3.058165519691487,
      "learning_rate": 5.267598631324981e-08,
      "loss": 0.66,
      "step": 6749
    },
    {
      "epoch": 0.6672762770926526,
      "grad_norm": 3.5812140948908184,
      "learning_rate": 5.2647780258548924e-08,
      "loss": 0.7314,
      "step": 6750
    },
    {
      "epoch": 0.667375132837407,
      "grad_norm": 6.030730302087922,
      "learning_rate": 5.261957905911969e-08,
      "loss": 0.7036,
      "step": 6751
    },
    {
      "epoch": 0.6674739885821614,
      "grad_norm": 5.032620753796471,
      "learning_rate": 5.259138271785375e-08,
      "loss": 0.715,
      "step": 6752
    },
    {
      "epoch": 0.667572844326916,
      "grad_norm": 6.398874442337878,
      "learning_rate": 5.2563191237642144e-08,
      "loss": 0.6756,
      "step": 6753
    },
    {
      "epoch": 0.6676717000716704,
      "grad_norm": 3.3258611670214577,
      "learning_rate": 5.253500462137557e-08,
      "loss": 0.6802,
      "step": 6754
    },
    {
      "epoch": 0.6677705558164249,
      "grad_norm": 3.054905637578774,
      "learning_rate": 5.2506822871944124e-08,
      "loss": 0.6874,
      "step": 6755
    },
    {
      "epoch": 0.6678694115611793,
      "grad_norm": 6.411033781430989,
      "learning_rate": 5.247864599223737e-08,
      "loss": 0.6923,
      "step": 6756
    },
    {
      "epoch": 0.6679682673059338,
      "grad_norm": 3.491075919522455,
      "learning_rate": 5.245047398514448e-08,
      "loss": 0.7605,
      "step": 6757
    },
    {
      "epoch": 0.6680671230506883,
      "grad_norm": 3.732582222289584,
      "learning_rate": 5.242230685355411e-08,
      "loss": 0.6821,
      "step": 6758
    },
    {
      "epoch": 0.6681659787954427,
      "grad_norm": 4.185982750832962,
      "learning_rate": 5.239414460035433e-08,
      "loss": 0.7959,
      "step": 6759
    },
    {
      "epoch": 0.6682648345401973,
      "grad_norm": 6.68517367613169,
      "learning_rate": 5.2365987228432815e-08,
      "loss": 0.7958,
      "step": 6760
    },
    {
      "epoch": 0.6683636902849517,
      "grad_norm": 3.4601949386836526,
      "learning_rate": 5.233783474067662e-08,
      "loss": 0.7638,
      "step": 6761
    },
    {
      "epoch": 0.6684625460297061,
      "grad_norm": 2.964957852677751,
      "learning_rate": 5.230968713997241e-08,
      "loss": 0.7627,
      "step": 6762
    },
    {
      "epoch": 0.6685614017744607,
      "grad_norm": 14.975749292150468,
      "learning_rate": 5.2281544429206336e-08,
      "loss": 0.7089,
      "step": 6763
    },
    {
      "epoch": 0.6686602575192151,
      "grad_norm": 4.170970821570884,
      "learning_rate": 5.225340661126395e-08,
      "loss": 0.6983,
      "step": 6764
    },
    {
      "epoch": 0.6687591132639695,
      "grad_norm": 7.5475294264630985,
      "learning_rate": 5.22252736890304e-08,
      "loss": 0.6863,
      "step": 6765
    },
    {
      "epoch": 0.6688579690087241,
      "grad_norm": 3.2996723429130395,
      "learning_rate": 5.219714566539035e-08,
      "loss": 0.6533,
      "step": 6766
    },
    {
      "epoch": 0.6689568247534785,
      "grad_norm": 3.9458950498699434,
      "learning_rate": 5.216902254322783e-08,
      "loss": 0.6915,
      "step": 6767
    },
    {
      "epoch": 0.6690556804982329,
      "grad_norm": 3.100522704483637,
      "learning_rate": 5.2140904325426484e-08,
      "loss": 0.5747,
      "step": 6768
    },
    {
      "epoch": 0.6691545362429874,
      "grad_norm": 3.8668015605349835,
      "learning_rate": 5.211279101486947e-08,
      "loss": 0.7475,
      "step": 6769
    },
    {
      "epoch": 0.6692533919877419,
      "grad_norm": 3.2739451753962068,
      "learning_rate": 5.2084682614439325e-08,
      "loss": 0.6408,
      "step": 6770
    },
    {
      "epoch": 0.6693522477324964,
      "grad_norm": 4.89288785563659,
      "learning_rate": 5.20565791270182e-08,
      "loss": 0.8086,
      "step": 6771
    },
    {
      "epoch": 0.6694511034772508,
      "grad_norm": 6.072262667347912,
      "learning_rate": 5.2028480555487676e-08,
      "loss": 0.7817,
      "step": 6772
    },
    {
      "epoch": 0.6695499592220053,
      "grad_norm": 3.409823859514897,
      "learning_rate": 5.200038690272881e-08,
      "loss": 0.7101,
      "step": 6773
    },
    {
      "epoch": 0.6696488149667598,
      "grad_norm": 4.6192354139885685,
      "learning_rate": 5.1972298171622255e-08,
      "loss": 0.7608,
      "step": 6774
    },
    {
      "epoch": 0.6697476707115142,
      "grad_norm": 8.771578910975975,
      "learning_rate": 5.194421436504802e-08,
      "loss": 0.6094,
      "step": 6775
    },
    {
      "epoch": 0.6698465264562687,
      "grad_norm": 6.997581682260295,
      "learning_rate": 5.191613548588574e-08,
      "loss": 0.6895,
      "step": 6776
    },
    {
      "epoch": 0.6699453822010232,
      "grad_norm": 4.909476160013606,
      "learning_rate": 5.1888061537014505e-08,
      "loss": 0.6297,
      "step": 6777
    },
    {
      "epoch": 0.6700442379457776,
      "grad_norm": 2.8885665044087894,
      "learning_rate": 5.185999252131283e-08,
      "loss": 0.674,
      "step": 6778
    },
    {
      "epoch": 0.670143093690532,
      "grad_norm": 11.126686014382956,
      "learning_rate": 5.18319284416588e-08,
      "loss": 0.7662,
      "step": 6779
    },
    {
      "epoch": 0.6702419494352866,
      "grad_norm": 9.39467778636712,
      "learning_rate": 5.1803869300930034e-08,
      "loss": 0.6784,
      "step": 6780
    },
    {
      "epoch": 0.670340805180041,
      "grad_norm": 4.989638768165759,
      "learning_rate": 5.177581510200347e-08,
      "loss": 0.6454,
      "step": 6781
    },
    {
      "epoch": 0.6704396609247955,
      "grad_norm": 5.263844535098252,
      "learning_rate": 5.1747765847755753e-08,
      "loss": 0.74,
      "step": 6782
    },
    {
      "epoch": 0.67053851666955,
      "grad_norm": 16.203609667606223,
      "learning_rate": 5.171972154106286e-08,
      "loss": 0.6408,
      "step": 6783
    },
    {
      "epoch": 0.6706373724143044,
      "grad_norm": 3.6255576136962,
      "learning_rate": 5.169168218480031e-08,
      "loss": 0.7453,
      "step": 6784
    },
    {
      "epoch": 0.6707362281590589,
      "grad_norm": 4.615145572903434,
      "learning_rate": 5.16636477818432e-08,
      "loss": 0.7234,
      "step": 6785
    },
    {
      "epoch": 0.6708350839038134,
      "grad_norm": 5.333011558170849,
      "learning_rate": 5.1635618335065975e-08,
      "loss": 0.721,
      "step": 6786
    },
    {
      "epoch": 0.6709339396485678,
      "grad_norm": 5.451128806865163,
      "learning_rate": 5.160759384734266e-08,
      "loss": 0.7154,
      "step": 6787
    },
    {
      "epoch": 0.6710327953933223,
      "grad_norm": 3.60497325291686,
      "learning_rate": 5.157957432154678e-08,
      "loss": 0.7974,
      "step": 6788
    },
    {
      "epoch": 0.6711316511380767,
      "grad_norm": 6.451600684100504,
      "learning_rate": 5.155155976055131e-08,
      "loss": 0.6748,
      "step": 6789
    },
    {
      "epoch": 0.6712305068828313,
      "grad_norm": 4.0501398428670115,
      "learning_rate": 5.152355016722869e-08,
      "loss": 0.7143,
      "step": 6790
    },
    {
      "epoch": 0.6713293626275857,
      "grad_norm": 5.334725346163968,
      "learning_rate": 5.149554554445096e-08,
      "loss": 0.6662,
      "step": 6791
    },
    {
      "epoch": 0.6714282183723401,
      "grad_norm": 15.989212688595666,
      "learning_rate": 5.1467545895089506e-08,
      "loss": 0.776,
      "step": 6792
    },
    {
      "epoch": 0.6715270741170947,
      "grad_norm": 2.9152107773792846,
      "learning_rate": 5.1439551222015344e-08,
      "loss": 0.6135,
      "step": 6793
    },
    {
      "epoch": 0.6716259298618491,
      "grad_norm": 3.098640885898533,
      "learning_rate": 5.1411561528098844e-08,
      "loss": 0.7281,
      "step": 6794
    },
    {
      "epoch": 0.6717247856066035,
      "grad_norm": 5.056906080072996,
      "learning_rate": 5.1383576816209997e-08,
      "loss": 0.5936,
      "step": 6795
    },
    {
      "epoch": 0.6718236413513581,
      "grad_norm": 6.999161458538162,
      "learning_rate": 5.135559708921824e-08,
      "loss": 0.707,
      "step": 6796
    },
    {
      "epoch": 0.6719224970961125,
      "grad_norm": 4.71023108554262,
      "learning_rate": 5.132762234999239e-08,
      "loss": 0.7689,
      "step": 6797
    },
    {
      "epoch": 0.672021352840867,
      "grad_norm": 3.8900876620592997,
      "learning_rate": 5.129965260140092e-08,
      "loss": 0.6388,
      "step": 6798
    },
    {
      "epoch": 0.6721202085856214,
      "grad_norm": 3.2319729226906646,
      "learning_rate": 5.1271687846311725e-08,
      "loss": 0.7253,
      "step": 6799
    },
    {
      "epoch": 0.6722190643303759,
      "grad_norm": 4.028619592893613,
      "learning_rate": 5.124372808759212e-08,
      "loss": 0.7008,
      "step": 6800
    },
    {
      "epoch": 0.6723179200751304,
      "grad_norm": 4.284638804453745,
      "learning_rate": 5.121577332810898e-08,
      "loss": 0.7569,
      "step": 6801
    },
    {
      "epoch": 0.6724167758198848,
      "grad_norm": 10.763042515320947,
      "learning_rate": 5.118782357072873e-08,
      "loss": 0.7668,
      "step": 6802
    },
    {
      "epoch": 0.6725156315646393,
      "grad_norm": 7.555716307197169,
      "learning_rate": 5.115987881831707e-08,
      "loss": 0.7564,
      "step": 6803
    },
    {
      "epoch": 0.6726144873093938,
      "grad_norm": 11.384482112843473,
      "learning_rate": 5.113193907373945e-08,
      "loss": 0.7009,
      "step": 6804
    },
    {
      "epoch": 0.6727133430541482,
      "grad_norm": 3.5196919143119922,
      "learning_rate": 5.1104004339860595e-08,
      "loss": 0.654,
      "step": 6805
    },
    {
      "epoch": 0.6728121987989027,
      "grad_norm": 4.616694375059566,
      "learning_rate": 5.107607461954485e-08,
      "loss": 0.8057,
      "step": 6806
    },
    {
      "epoch": 0.6729110545436572,
      "grad_norm": 3.613801782890328,
      "learning_rate": 5.1048149915655936e-08,
      "loss": 0.7178,
      "step": 6807
    },
    {
      "epoch": 0.6730099102884116,
      "grad_norm": 3.0903112702303352,
      "learning_rate": 5.102023023105722e-08,
      "loss": 0.6991,
      "step": 6808
    },
    {
      "epoch": 0.673108766033166,
      "grad_norm": 6.471821239669036,
      "learning_rate": 5.099231556861132e-08,
      "loss": 0.6626,
      "step": 6809
    },
    {
      "epoch": 0.6732076217779206,
      "grad_norm": 5.539924235755165,
      "learning_rate": 5.09644059311806e-08,
      "loss": 0.7433,
      "step": 6810
    },
    {
      "epoch": 0.673306477522675,
      "grad_norm": 8.132154354317967,
      "learning_rate": 5.093650132162668e-08,
      "loss": 0.7217,
      "step": 6811
    },
    {
      "epoch": 0.6734053332674295,
      "grad_norm": 2.979673044484443,
      "learning_rate": 5.090860174281082e-08,
      "loss": 0.7369,
      "step": 6812
    },
    {
      "epoch": 0.673504189012184,
      "grad_norm": 6.274236372168946,
      "learning_rate": 5.088070719759374e-08,
      "loss": 0.7522,
      "step": 6813
    },
    {
      "epoch": 0.6736030447569384,
      "grad_norm": 4.3796850184779315,
      "learning_rate": 5.085281768883554e-08,
      "loss": 0.7813,
      "step": 6814
    },
    {
      "epoch": 0.6737019005016929,
      "grad_norm": 12.85910083790196,
      "learning_rate": 5.0824933219395935e-08,
      "loss": 0.7435,
      "step": 6815
    },
    {
      "epoch": 0.6738007562464474,
      "grad_norm": 4.8747433467524655,
      "learning_rate": 5.0797053792134024e-08,
      "loss": 0.7218,
      "step": 6816
    },
    {
      "epoch": 0.6738996119912019,
      "grad_norm": 8.19871153019696,
      "learning_rate": 5.076917940990845e-08,
      "loss": 0.6886,
      "step": 6817
    },
    {
      "epoch": 0.6739984677359563,
      "grad_norm": 8.566417385688627,
      "learning_rate": 5.074131007557736e-08,
      "loss": 0.6285,
      "step": 6818
    },
    {
      "epoch": 0.6740973234807107,
      "grad_norm": 19.29585175460254,
      "learning_rate": 5.071344579199826e-08,
      "loss": 0.6677,
      "step": 6819
    },
    {
      "epoch": 0.6741961792254653,
      "grad_norm": 61.89689824451084,
      "learning_rate": 5.068558656202827e-08,
      "loss": 0.6092,
      "step": 6820
    },
    {
      "epoch": 0.6742950349702197,
      "grad_norm": 3.040544692548938,
      "learning_rate": 5.065773238852398e-08,
      "loss": 0.6731,
      "step": 6821
    },
    {
      "epoch": 0.6743938907149741,
      "grad_norm": 3.469502390996233,
      "learning_rate": 5.062988327434134e-08,
      "loss": 0.5938,
      "step": 6822
    },
    {
      "epoch": 0.6744927464597287,
      "grad_norm": 5.000464681759872,
      "learning_rate": 5.0602039222335946e-08,
      "loss": 0.6663,
      "step": 6823
    },
    {
      "epoch": 0.6745916022044831,
      "grad_norm": 3.759747599126241,
      "learning_rate": 5.0574200235362727e-08,
      "loss": 0.7535,
      "step": 6824
    },
    {
      "epoch": 0.6746904579492375,
      "grad_norm": 3.9344217141345372,
      "learning_rate": 5.054636631627623e-08,
      "loss": 0.6653,
      "step": 6825
    },
    {
      "epoch": 0.6747893136939921,
      "grad_norm": 6.182891817215699,
      "learning_rate": 5.051853746793038e-08,
      "loss": 0.6951,
      "step": 6826
    },
    {
      "epoch": 0.6748881694387465,
      "grad_norm": 4.167358944270966,
      "learning_rate": 5.049071369317857e-08,
      "loss": 0.6887,
      "step": 6827
    },
    {
      "epoch": 0.674987025183501,
      "grad_norm": 3.298263388817007,
      "learning_rate": 5.046289499487375e-08,
      "loss": 0.692,
      "step": 6828
    },
    {
      "epoch": 0.6750858809282554,
      "grad_norm": 10.923487124478394,
      "learning_rate": 5.043508137586836e-08,
      "loss": 0.6413,
      "step": 6829
    },
    {
      "epoch": 0.6751847366730099,
      "grad_norm": 4.48850383240122,
      "learning_rate": 5.04072728390142e-08,
      "loss": 0.6433,
      "step": 6830
    },
    {
      "epoch": 0.6752835924177644,
      "grad_norm": 4.073508453341519,
      "learning_rate": 5.0379469387162684e-08,
      "loss": 0.7421,
      "step": 6831
    },
    {
      "epoch": 0.6753824481625188,
      "grad_norm": 3.2358847113009737,
      "learning_rate": 5.0351671023164655e-08,
      "loss": 0.6871,
      "step": 6832
    },
    {
      "epoch": 0.6754813039072733,
      "grad_norm": 8.50330815808087,
      "learning_rate": 5.032387774987037e-08,
      "loss": 0.7038,
      "step": 6833
    },
    {
      "epoch": 0.6755801596520278,
      "grad_norm": 3.767945333046059,
      "learning_rate": 5.029608957012963e-08,
      "loss": 0.722,
      "step": 6834
    },
    {
      "epoch": 0.6756790153967822,
      "grad_norm": 8.67454522483573,
      "learning_rate": 5.026830648679177e-08,
      "loss": 0.7402,
      "step": 6835
    },
    {
      "epoch": 0.6757778711415368,
      "grad_norm": 10.761263278714258,
      "learning_rate": 5.024052850270544e-08,
      "loss": 0.7885,
      "step": 6836
    },
    {
      "epoch": 0.6758767268862912,
      "grad_norm": 30.00726554487649,
      "learning_rate": 5.021275562071896e-08,
      "loss": 0.6368,
      "step": 6837
    },
    {
      "epoch": 0.6759755826310456,
      "grad_norm": 3.5222781661145404,
      "learning_rate": 5.018498784367993e-08,
      "loss": 0.6619,
      "step": 6838
    },
    {
      "epoch": 0.6760744383758002,
      "grad_norm": 6.598657633384732,
      "learning_rate": 5.015722517443557e-08,
      "loss": 0.6796,
      "step": 6839
    },
    {
      "epoch": 0.6761732941205546,
      "grad_norm": 7.257881222039123,
      "learning_rate": 5.0129467615832576e-08,
      "loss": 0.8054,
      "step": 6840
    },
    {
      "epoch": 0.676272149865309,
      "grad_norm": 4.410335598514433,
      "learning_rate": 5.0101715170716995e-08,
      "loss": 0.7087,
      "step": 6841
    },
    {
      "epoch": 0.6763710056100635,
      "grad_norm": 5.229515771855248,
      "learning_rate": 5.007396784193452e-08,
      "loss": 0.5995,
      "step": 6842
    },
    {
      "epoch": 0.676469861354818,
      "grad_norm": 6.7163826864507845,
      "learning_rate": 5.0046225632330185e-08,
      "loss": 0.7135,
      "step": 6843
    },
    {
      "epoch": 0.6765687170995724,
      "grad_norm": 3.6096788154150077,
      "learning_rate": 5.001848854474848e-08,
      "loss": 0.6781,
      "step": 6844
    },
    {
      "epoch": 0.6766675728443269,
      "grad_norm": 5.144818572672796,
      "learning_rate": 4.9990756582033524e-08,
      "loss": 0.646,
      "step": 6845
    },
    {
      "epoch": 0.6767664285890814,
      "grad_norm": 45.91528018723797,
      "learning_rate": 4.996302974702882e-08,
      "loss": 0.6881,
      "step": 6846
    },
    {
      "epoch": 0.6768652843338359,
      "grad_norm": 6.214543403062526,
      "learning_rate": 4.993530804257727e-08,
      "loss": 0.7315,
      "step": 6847
    },
    {
      "epoch": 0.6769641400785903,
      "grad_norm": 8.519939838457285,
      "learning_rate": 4.9907591471521415e-08,
      "loss": 0.7692,
      "step": 6848
    },
    {
      "epoch": 0.6770629958233448,
      "grad_norm": 6.852349806020006,
      "learning_rate": 4.987988003670312e-08,
      "loss": 0.7436,
      "step": 6849
    },
    {
      "epoch": 0.6771618515680993,
      "grad_norm": 3.5236694116561154,
      "learning_rate": 4.985217374096378e-08,
      "loss": 0.6579,
      "step": 6850
    },
    {
      "epoch": 0.6772607073128537,
      "grad_norm": 7.860086317894921,
      "learning_rate": 4.982447258714434e-08,
      "loss": 0.8831,
      "step": 6851
    },
    {
      "epoch": 0.6773595630576081,
      "grad_norm": 10.290153666143267,
      "learning_rate": 4.979677657808505e-08,
      "loss": 0.7464,
      "step": 6852
    },
    {
      "epoch": 0.6774584188023627,
      "grad_norm": 2.9129561228148875,
      "learning_rate": 4.976908571662578e-08,
      "loss": 0.6691,
      "step": 6853
    },
    {
      "epoch": 0.6775572745471171,
      "grad_norm": 3.6517710707466478,
      "learning_rate": 4.9741400005605846e-08,
      "loss": 0.7698,
      "step": 6854
    },
    {
      "epoch": 0.6776561302918716,
      "grad_norm": 43.883086607895045,
      "learning_rate": 4.971371944786393e-08,
      "loss": 0.676,
      "step": 6855
    },
    {
      "epoch": 0.6777549860366261,
      "grad_norm": 5.053330528651903,
      "learning_rate": 4.968604404623832e-08,
      "loss": 0.5802,
      "step": 6856
    },
    {
      "epoch": 0.6778538417813805,
      "grad_norm": 24.160997486178267,
      "learning_rate": 4.965837380356675e-08,
      "loss": 0.7803,
      "step": 6857
    },
    {
      "epoch": 0.677952697526135,
      "grad_norm": 4.055121047593691,
      "learning_rate": 4.963070872268631e-08,
      "loss": 0.6842,
      "step": 6858
    },
    {
      "epoch": 0.6780515532708895,
      "grad_norm": 6.456773685263812,
      "learning_rate": 4.960304880643373e-08,
      "loss": 0.6539,
      "step": 6859
    },
    {
      "epoch": 0.6781504090156439,
      "grad_norm": 3.9703818043639645,
      "learning_rate": 4.957539405764509e-08,
      "loss": 0.7285,
      "step": 6860
    },
    {
      "epoch": 0.6782492647603984,
      "grad_norm": 3.7865751470777367,
      "learning_rate": 4.9547744479155936e-08,
      "loss": 0.7777,
      "step": 6861
    },
    {
      "epoch": 0.6783481205051528,
      "grad_norm": 10.317836805555624,
      "learning_rate": 4.952010007380139e-08,
      "loss": 0.7056,
      "step": 6862
    },
    {
      "epoch": 0.6784469762499074,
      "grad_norm": 4.328398420605287,
      "learning_rate": 4.949246084441593e-08,
      "loss": 0.7625,
      "step": 6863
    },
    {
      "epoch": 0.6785458319946618,
      "grad_norm": 4.581452250896427,
      "learning_rate": 4.946482679383357e-08,
      "loss": 0.7496,
      "step": 6864
    },
    {
      "epoch": 0.6786446877394162,
      "grad_norm": 3.584586257395639,
      "learning_rate": 4.943719792488781e-08,
      "loss": 0.5998,
      "step": 6865
    },
    {
      "epoch": 0.6787435434841708,
      "grad_norm": 9.224589967636696,
      "learning_rate": 4.94095742404115e-08,
      "loss": 0.6916,
      "step": 6866
    },
    {
      "epoch": 0.6788423992289252,
      "grad_norm": 4.975142700250369,
      "learning_rate": 4.93819557432371e-08,
      "loss": 0.675,
      "step": 6867
    },
    {
      "epoch": 0.6789412549736796,
      "grad_norm": 4.143003162179376,
      "learning_rate": 4.935434243619651e-08,
      "loss": 0.7365,
      "step": 6868
    },
    {
      "epoch": 0.6790401107184342,
      "grad_norm": 3.6014852945241134,
      "learning_rate": 4.932673432212099e-08,
      "loss": 0.6873,
      "step": 6869
    },
    {
      "epoch": 0.6791389664631886,
      "grad_norm": 4.119182194047706,
      "learning_rate": 4.929913140384141e-08,
      "loss": 0.7649,
      "step": 6870
    },
    {
      "epoch": 0.679237822207943,
      "grad_norm": 3.01901960597528,
      "learning_rate": 4.9271533684187984e-08,
      "loss": 0.642,
      "step": 6871
    },
    {
      "epoch": 0.6793366779526975,
      "grad_norm": 7.008676264886851,
      "learning_rate": 4.924394116599048e-08,
      "loss": 0.6502,
      "step": 6872
    },
    {
      "epoch": 0.679435533697452,
      "grad_norm": 28.44345425358551,
      "learning_rate": 4.921635385207814e-08,
      "loss": 0.6515,
      "step": 6873
    },
    {
      "epoch": 0.6795343894422065,
      "grad_norm": 7.773255624415793,
      "learning_rate": 4.918877174527957e-08,
      "loss": 0.6238,
      "step": 6874
    },
    {
      "epoch": 0.6796332451869609,
      "grad_norm": 5.709703784199952,
      "learning_rate": 4.9161194848422937e-08,
      "loss": 0.6795,
      "step": 6875
    },
    {
      "epoch": 0.6797321009317154,
      "grad_norm": 3.6449057839784182,
      "learning_rate": 4.913362316433588e-08,
      "loss": 0.7266,
      "step": 6876
    },
    {
      "epoch": 0.6798309566764699,
      "grad_norm": 4.54525678008131,
      "learning_rate": 4.910605669584544e-08,
      "loss": 0.642,
      "step": 6877
    },
    {
      "epoch": 0.6799298124212243,
      "grad_norm": 6.453322957820563,
      "learning_rate": 4.9078495445778114e-08,
      "loss": 0.7973,
      "step": 6878
    },
    {
      "epoch": 0.6800286681659788,
      "grad_norm": 4.505215899254946,
      "learning_rate": 4.905093941695996e-08,
      "loss": 0.7725,
      "step": 6879
    },
    {
      "epoch": 0.6801275239107333,
      "grad_norm": 3.2175590221620216,
      "learning_rate": 4.902338861221639e-08,
      "loss": 0.5864,
      "step": 6880
    },
    {
      "epoch": 0.6802263796554877,
      "grad_norm": 3.674000116246416,
      "learning_rate": 4.899584303437239e-08,
      "loss": 0.6796,
      "step": 6881
    },
    {
      "epoch": 0.6803252354002421,
      "grad_norm": 5.223100096392552,
      "learning_rate": 4.896830268625229e-08,
      "loss": 0.6873,
      "step": 6882
    },
    {
      "epoch": 0.6804240911449967,
      "grad_norm": 9.847794039806638,
      "learning_rate": 4.8940767570679974e-08,
      "loss": 0.698,
      "step": 6883
    },
    {
      "epoch": 0.6805229468897511,
      "grad_norm": 3.4539713799061236,
      "learning_rate": 4.8913237690478814e-08,
      "loss": 0.6823,
      "step": 6884
    },
    {
      "epoch": 0.6806218026345056,
      "grad_norm": 2.68701786093155,
      "learning_rate": 4.8885713048471496e-08,
      "loss": 0.6888,
      "step": 6885
    },
    {
      "epoch": 0.6807206583792601,
      "grad_norm": 5.38268592247178,
      "learning_rate": 4.885819364748033e-08,
      "loss": 0.6765,
      "step": 6886
    },
    {
      "epoch": 0.6808195141240145,
      "grad_norm": 5.199286414193881,
      "learning_rate": 4.883067949032704e-08,
      "loss": 0.702,
      "step": 6887
    },
    {
      "epoch": 0.680918369868769,
      "grad_norm": 4.7331600785469625,
      "learning_rate": 4.880317057983273e-08,
      "loss": 0.7339,
      "step": 6888
    },
    {
      "epoch": 0.6810172256135235,
      "grad_norm": 3.925974233253034,
      "learning_rate": 4.877566691881807e-08,
      "loss": 0.6814,
      "step": 6889
    },
    {
      "epoch": 0.681116081358278,
      "grad_norm": 3.4475610866722883,
      "learning_rate": 4.8748168510103194e-08,
      "loss": 0.7374,
      "step": 6890
    },
    {
      "epoch": 0.6812149371030324,
      "grad_norm": 8.18605286398184,
      "learning_rate": 4.872067535650758e-08,
      "loss": 0.7174,
      "step": 6891
    },
    {
      "epoch": 0.6813137928477868,
      "grad_norm": 3.4315461713930357,
      "learning_rate": 4.8693187460850326e-08,
      "loss": 0.7225,
      "step": 6892
    },
    {
      "epoch": 0.6814126485925414,
      "grad_norm": 2.652718045116078,
      "learning_rate": 4.866570482594984e-08,
      "loss": 0.6098,
      "step": 6893
    },
    {
      "epoch": 0.6815115043372958,
      "grad_norm": 4.700895924114779,
      "learning_rate": 4.8638227454624114e-08,
      "loss": 0.6696,
      "step": 6894
    },
    {
      "epoch": 0.6816103600820502,
      "grad_norm": 2.9915036690600734,
      "learning_rate": 4.8610755349690494e-08,
      "loss": 0.7632,
      "step": 6895
    },
    {
      "epoch": 0.6817092158268048,
      "grad_norm": 9.432653009692261,
      "learning_rate": 4.858328851396589e-08,
      "loss": 0.6526,
      "step": 6896
    },
    {
      "epoch": 0.6818080715715592,
      "grad_norm": 4.269756703680442,
      "learning_rate": 4.8555826950266576e-08,
      "loss": 0.7509,
      "step": 6897
    },
    {
      "epoch": 0.6819069273163136,
      "grad_norm": 10.586998629231845,
      "learning_rate": 4.852837066140839e-08,
      "loss": 0.68,
      "step": 6898
    },
    {
      "epoch": 0.6820057830610682,
      "grad_norm": 5.633736726901227,
      "learning_rate": 4.8500919650206494e-08,
      "loss": 0.7223,
      "step": 6899
    },
    {
      "epoch": 0.6821046388058226,
      "grad_norm": 4.672784187019227,
      "learning_rate": 4.847347391947563e-08,
      "loss": 0.787,
      "step": 6900
    },
    {
      "epoch": 0.682203494550577,
      "grad_norm": 3.698741175343985,
      "learning_rate": 4.844603347202997e-08,
      "loss": 0.7541,
      "step": 6901
    },
    {
      "epoch": 0.6823023502953315,
      "grad_norm": 3.5927796368160463,
      "learning_rate": 4.8418598310683065e-08,
      "loss": 0.7384,
      "step": 6902
    },
    {
      "epoch": 0.682401206040086,
      "grad_norm": 3.27263871494874,
      "learning_rate": 4.839116843824808e-08,
      "loss": 0.6882,
      "step": 6903
    },
    {
      "epoch": 0.6825000617848405,
      "grad_norm": 14.331237130589473,
      "learning_rate": 4.836374385753743e-08,
      "loss": 0.7054,
      "step": 6904
    },
    {
      "epoch": 0.6825989175295949,
      "grad_norm": 4.226825488965418,
      "learning_rate": 4.8336324571363164e-08,
      "loss": 0.7509,
      "step": 6905
    },
    {
      "epoch": 0.6826977732743494,
      "grad_norm": 6.912528144207288,
      "learning_rate": 4.830891058253677e-08,
      "loss": 0.6692,
      "step": 6906
    },
    {
      "epoch": 0.6827966290191039,
      "grad_norm": 3.647287556415683,
      "learning_rate": 4.828150189386905e-08,
      "loss": 0.7148,
      "step": 6907
    },
    {
      "epoch": 0.6828954847638583,
      "grad_norm": 14.819176302837613,
      "learning_rate": 4.8254098508170404e-08,
      "loss": 0.7197,
      "step": 6908
    },
    {
      "epoch": 0.6829943405086129,
      "grad_norm": 3.0249221226345964,
      "learning_rate": 4.82267004282507e-08,
      "loss": 0.6955,
      "step": 6909
    },
    {
      "epoch": 0.6830931962533673,
      "grad_norm": 3.721584503499986,
      "learning_rate": 4.819930765691911e-08,
      "loss": 0.7759,
      "step": 6910
    },
    {
      "epoch": 0.6831920519981217,
      "grad_norm": 3.3445865519529927,
      "learning_rate": 4.817192019698445e-08,
      "loss": 0.6797,
      "step": 6911
    },
    {
      "epoch": 0.6832909077428762,
      "grad_norm": 4.898585075805994,
      "learning_rate": 4.8144538051254814e-08,
      "loss": 0.7505,
      "step": 6912
    },
    {
      "epoch": 0.6833897634876307,
      "grad_norm": 3.4087692565303898,
      "learning_rate": 4.811716122253793e-08,
      "loss": 0.7699,
      "step": 6913
    },
    {
      "epoch": 0.6834886192323851,
      "grad_norm": 6.097220545003893,
      "learning_rate": 4.8089789713640814e-08,
      "loss": 0.7399,
      "step": 6914
    },
    {
      "epoch": 0.6835874749771396,
      "grad_norm": 3.6922918135559613,
      "learning_rate": 4.806242352737001e-08,
      "loss": 0.6545,
      "step": 6915
    },
    {
      "epoch": 0.6836863307218941,
      "grad_norm": 4.044641687068706,
      "learning_rate": 4.8035062666531536e-08,
      "loss": 0.742,
      "step": 6916
    },
    {
      "epoch": 0.6837851864666485,
      "grad_norm": 2.6762695537413697,
      "learning_rate": 4.800770713393089e-08,
      "loss": 0.6043,
      "step": 6917
    },
    {
      "epoch": 0.683884042211403,
      "grad_norm": 4.883396393599119,
      "learning_rate": 4.79803569323729e-08,
      "loss": 0.7531,
      "step": 6918
    },
    {
      "epoch": 0.6839828979561575,
      "grad_norm": 17.44980841798662,
      "learning_rate": 4.7953012064661956e-08,
      "loss": 0.704,
      "step": 6919
    },
    {
      "epoch": 0.684081753700912,
      "grad_norm": 18.456757305404814,
      "learning_rate": 4.792567253360192e-08,
      "loss": 0.7214,
      "step": 6920
    },
    {
      "epoch": 0.6841806094456664,
      "grad_norm": 6.103534579673057,
      "learning_rate": 4.789833834199598e-08,
      "loss": 0.7503,
      "step": 6921
    },
    {
      "epoch": 0.6842794651904209,
      "grad_norm": 10.77700498482542,
      "learning_rate": 4.787100949264689e-08,
      "loss": 0.5429,
      "step": 6922
    },
    {
      "epoch": 0.6843783209351754,
      "grad_norm": 3.857301112120819,
      "learning_rate": 4.784368598835685e-08,
      "loss": 0.7199,
      "step": 6923
    },
    {
      "epoch": 0.6844771766799298,
      "grad_norm": 3.776227179824211,
      "learning_rate": 4.7816367831927416e-08,
      "loss": 0.7694,
      "step": 6924
    },
    {
      "epoch": 0.6845760324246842,
      "grad_norm": 3.75230882991863,
      "learning_rate": 4.778905502615974e-08,
      "loss": 0.6948,
      "step": 6925
    },
    {
      "epoch": 0.6846748881694388,
      "grad_norm": 7.558730876775452,
      "learning_rate": 4.776174757385427e-08,
      "loss": 0.7334,
      "step": 6926
    },
    {
      "epoch": 0.6847737439141932,
      "grad_norm": 4.984377059430879,
      "learning_rate": 4.7734445477811016e-08,
      "loss": 0.8301,
      "step": 6927
    },
    {
      "epoch": 0.6848725996589476,
      "grad_norm": 4.796051509063145,
      "learning_rate": 4.770714874082944e-08,
      "loss": 0.7197,
      "step": 6928
    },
    {
      "epoch": 0.6849714554037022,
      "grad_norm": 2.9130885477803625,
      "learning_rate": 4.767985736570835e-08,
      "loss": 0.6725,
      "step": 6929
    },
    {
      "epoch": 0.6850703111484566,
      "grad_norm": 2.8242580284319945,
      "learning_rate": 4.765257135524615e-08,
      "loss": 0.6228,
      "step": 6930
    },
    {
      "epoch": 0.6851691668932111,
      "grad_norm": 3.1362185780204217,
      "learning_rate": 4.7625290712240575e-08,
      "loss": 0.6882,
      "step": 6931
    },
    {
      "epoch": 0.6852680226379656,
      "grad_norm": 3.925733239970925,
      "learning_rate": 4.7598015439488824e-08,
      "loss": 0.7749,
      "step": 6932
    },
    {
      "epoch": 0.68536687838272,
      "grad_norm": 4.190017270568966,
      "learning_rate": 4.7570745539787605e-08,
      "loss": 0.7653,
      "step": 6933
    },
    {
      "epoch": 0.6854657341274745,
      "grad_norm": 7.197463900046511,
      "learning_rate": 4.754348101593308e-08,
      "loss": 0.6995,
      "step": 6934
    },
    {
      "epoch": 0.6855645898722289,
      "grad_norm": 3.0534948004950624,
      "learning_rate": 4.7516221870720754e-08,
      "loss": 0.7125,
      "step": 6935
    },
    {
      "epoch": 0.6856634456169834,
      "grad_norm": 5.702806688323771,
      "learning_rate": 4.7488968106945724e-08,
      "loss": 0.7914,
      "step": 6936
    },
    {
      "epoch": 0.6857623013617379,
      "grad_norm": 33.78031274474961,
      "learning_rate": 4.7461719727402386e-08,
      "loss": 0.7491,
      "step": 6937
    },
    {
      "epoch": 0.6858611571064923,
      "grad_norm": 5.539549152347562,
      "learning_rate": 4.7434476734884686e-08,
      "loss": 0.6893,
      "step": 6938
    },
    {
      "epoch": 0.6859600128512469,
      "grad_norm": 6.615388740482905,
      "learning_rate": 4.740723913218605e-08,
      "loss": 0.6089,
      "step": 6939
    },
    {
      "epoch": 0.6860588685960013,
      "grad_norm": 6.924555318259605,
      "learning_rate": 4.738000692209919e-08,
      "loss": 0.6385,
      "step": 6940
    },
    {
      "epoch": 0.6861577243407557,
      "grad_norm": 3.138866659763953,
      "learning_rate": 4.7352780107416433e-08,
      "loss": 0.6036,
      "step": 6941
    },
    {
      "epoch": 0.6862565800855103,
      "grad_norm": 3.4513439957517504,
      "learning_rate": 4.7325558690929514e-08,
      "loss": 0.6677,
      "step": 6942
    },
    {
      "epoch": 0.6863554358302647,
      "grad_norm": 3.253709566020095,
      "learning_rate": 4.7298342675429506e-08,
      "loss": 0.6209,
      "step": 6943
    },
    {
      "epoch": 0.6864542915750191,
      "grad_norm": 5.799598788322272,
      "learning_rate": 4.727113206370705e-08,
      "loss": 0.7176,
      "step": 6944
    },
    {
      "epoch": 0.6865531473197736,
      "grad_norm": 5.365056362103453,
      "learning_rate": 4.7243926858552244e-08,
      "loss": 0.7836,
      "step": 6945
    },
    {
      "epoch": 0.6866520030645281,
      "grad_norm": 4.778812516991259,
      "learning_rate": 4.721672706275448e-08,
      "loss": 0.7239,
      "step": 6946
    },
    {
      "epoch": 0.6867508588092825,
      "grad_norm": 4.629921746670702,
      "learning_rate": 4.718953267910279e-08,
      "loss": 0.6773,
      "step": 6947
    },
    {
      "epoch": 0.686849714554037,
      "grad_norm": 5.579109019554846,
      "learning_rate": 4.716234371038552e-08,
      "loss": 0.7541,
      "step": 6948
    },
    {
      "epoch": 0.6869485702987915,
      "grad_norm": 4.758957705585035,
      "learning_rate": 4.7135160159390454e-08,
      "loss": 0.6683,
      "step": 6949
    },
    {
      "epoch": 0.687047426043546,
      "grad_norm": 4.195006991169889,
      "learning_rate": 4.7107982028904935e-08,
      "loss": 0.7253,
      "step": 6950
    },
    {
      "epoch": 0.6871462817883004,
      "grad_norm": 5.32580584084987,
      "learning_rate": 4.708080932171562e-08,
      "loss": 0.6512,
      "step": 6951
    },
    {
      "epoch": 0.6872451375330549,
      "grad_norm": 4.272632542991704,
      "learning_rate": 4.705364204060869e-08,
      "loss": 0.6579,
      "step": 6952
    },
    {
      "epoch": 0.6873439932778094,
      "grad_norm": 10.81736615931638,
      "learning_rate": 4.7026480188369786e-08,
      "loss": 0.6936,
      "step": 6953
    },
    {
      "epoch": 0.6874428490225638,
      "grad_norm": 4.152721692515138,
      "learning_rate": 4.699932376778388e-08,
      "loss": 0.5855,
      "step": 6954
    },
    {
      "epoch": 0.6875417047673182,
      "grad_norm": 3.0762782592733324,
      "learning_rate": 4.697217278163552e-08,
      "loss": 0.7149,
      "step": 6955
    },
    {
      "epoch": 0.6876405605120728,
      "grad_norm": 9.30964837927263,
      "learning_rate": 4.694502723270867e-08,
      "loss": 0.6871,
      "step": 6956
    },
    {
      "epoch": 0.6877394162568272,
      "grad_norm": 4.990785873297677,
      "learning_rate": 4.6917887123786617e-08,
      "loss": 0.7463,
      "step": 6957
    },
    {
      "epoch": 0.6878382720015817,
      "grad_norm": 6.7067265457758065,
      "learning_rate": 4.689075245765225e-08,
      "loss": 0.6029,
      "step": 6958
    },
    {
      "epoch": 0.6879371277463362,
      "grad_norm": 3.311523944426087,
      "learning_rate": 4.686362323708778e-08,
      "loss": 0.7121,
      "step": 6959
    },
    {
      "epoch": 0.6880359834910906,
      "grad_norm": 6.001790907195006,
      "learning_rate": 4.6836499464874916e-08,
      "loss": 0.7343,
      "step": 6960
    },
    {
      "epoch": 0.6881348392358451,
      "grad_norm": 4.1790906378052215,
      "learning_rate": 4.680938114379486e-08,
      "loss": 0.6271,
      "step": 6961
    },
    {
      "epoch": 0.6882336949805996,
      "grad_norm": 3.7741752379456477,
      "learning_rate": 4.678226827662811e-08,
      "loss": 0.7145,
      "step": 6962
    },
    {
      "epoch": 0.688332550725354,
      "grad_norm": 2.7902857485756316,
      "learning_rate": 4.6755160866154745e-08,
      "loss": 0.5493,
      "step": 6963
    },
    {
      "epoch": 0.6884314064701085,
      "grad_norm": 3.2281486018893077,
      "learning_rate": 4.6728058915154234e-08,
      "loss": 0.6889,
      "step": 6964
    },
    {
      "epoch": 0.6885302622148629,
      "grad_norm": 15.02017488018892,
      "learning_rate": 4.6700962426405475e-08,
      "loss": 0.534,
      "step": 6965
    },
    {
      "epoch": 0.6886291179596175,
      "grad_norm": 3.6835119299155443,
      "learning_rate": 4.667387140268677e-08,
      "loss": 0.6278,
      "step": 6966
    },
    {
      "epoch": 0.6887279737043719,
      "grad_norm": 21.539915563172084,
      "learning_rate": 4.6646785846775974e-08,
      "loss": 0.7097,
      "step": 6967
    },
    {
      "epoch": 0.6888268294491263,
      "grad_norm": 3.670108027498525,
      "learning_rate": 4.6619705761450226e-08,
      "loss": 0.6639,
      "step": 6968
    },
    {
      "epoch": 0.6889256851938809,
      "grad_norm": 10.321352344782213,
      "learning_rate": 4.6592631149486284e-08,
      "loss": 0.696,
      "step": 6969
    },
    {
      "epoch": 0.6890245409386353,
      "grad_norm": 3.0426274862385534,
      "learning_rate": 4.6565562013660154e-08,
      "loss": 0.6781,
      "step": 6970
    },
    {
      "epoch": 0.6891233966833897,
      "grad_norm": 7.990407110684896,
      "learning_rate": 4.653849835674744e-08,
      "loss": 0.6438,
      "step": 6971
    },
    {
      "epoch": 0.6892222524281443,
      "grad_norm": 3.7343883732729393,
      "learning_rate": 4.6511440181523144e-08,
      "loss": 0.664,
      "step": 6972
    },
    {
      "epoch": 0.6893211081728987,
      "grad_norm": 22.007635310269958,
      "learning_rate": 4.64843874907616e-08,
      "loss": 0.7313,
      "step": 6973
    },
    {
      "epoch": 0.6894199639176531,
      "grad_norm": 18.079633471455384,
      "learning_rate": 4.6457340287236714e-08,
      "loss": 0.7012,
      "step": 6974
    },
    {
      "epoch": 0.6895188196624076,
      "grad_norm": 2.9944582516664386,
      "learning_rate": 4.643029857372181e-08,
      "loss": 0.6674,
      "step": 6975
    },
    {
      "epoch": 0.6896176754071621,
      "grad_norm": 4.500741074965972,
      "learning_rate": 4.6403262352989525e-08,
      "loss": 0.6701,
      "step": 6976
    },
    {
      "epoch": 0.6897165311519166,
      "grad_norm": 2.548954140982369,
      "learning_rate": 4.637623162781209e-08,
      "loss": 0.6563,
      "step": 6977
    },
    {
      "epoch": 0.689815386896671,
      "grad_norm": 3.413288798767727,
      "learning_rate": 4.634920640096114e-08,
      "loss": 0.6268,
      "step": 6978
    },
    {
      "epoch": 0.6899142426414255,
      "grad_norm": 3.860836915974056,
      "learning_rate": 4.632218667520762e-08,
      "loss": 0.6269,
      "step": 6979
    },
    {
      "epoch": 0.69001309838618,
      "grad_norm": 3.910412860339552,
      "learning_rate": 4.6295172453322094e-08,
      "loss": 0.672,
      "step": 6980
    },
    {
      "epoch": 0.6901119541309344,
      "grad_norm": 2.9692310329764164,
      "learning_rate": 4.62681637380744e-08,
      "loss": 0.7331,
      "step": 6981
    },
    {
      "epoch": 0.690210809875689,
      "grad_norm": 4.051498392425163,
      "learning_rate": 4.624116053223395e-08,
      "loss": 0.7156,
      "step": 6982
    },
    {
      "epoch": 0.6903096656204434,
      "grad_norm": 5.926826218201472,
      "learning_rate": 4.6214162838569445e-08,
      "loss": 0.6409,
      "step": 6983
    },
    {
      "epoch": 0.6904085213651978,
      "grad_norm": 3.6555988559722015,
      "learning_rate": 4.6187170659849184e-08,
      "loss": 0.7848,
      "step": 6984
    },
    {
      "epoch": 0.6905073771099522,
      "grad_norm": 4.442348514567032,
      "learning_rate": 4.616018399884074e-08,
      "loss": 0.668,
      "step": 6985
    },
    {
      "epoch": 0.6906062328547068,
      "grad_norm": 5.600797610747027,
      "learning_rate": 4.613320285831127e-08,
      "loss": 0.7223,
      "step": 6986
    },
    {
      "epoch": 0.6907050885994612,
      "grad_norm": 2.3467663777227483,
      "learning_rate": 4.610622724102722e-08,
      "loss": 0.6415,
      "step": 6987
    },
    {
      "epoch": 0.6908039443442157,
      "grad_norm": 5.62466816179407,
      "learning_rate": 4.607925714975458e-08,
      "loss": 0.6499,
      "step": 6988
    },
    {
      "epoch": 0.6909028000889702,
      "grad_norm": 9.752393621424634,
      "learning_rate": 4.605229258725876e-08,
      "loss": 0.6266,
      "step": 6989
    },
    {
      "epoch": 0.6910016558337246,
      "grad_norm": 4.653689361003493,
      "learning_rate": 4.6025333556304526e-08,
      "loss": 0.7372,
      "step": 6990
    },
    {
      "epoch": 0.6911005115784791,
      "grad_norm": 3.877412010196789,
      "learning_rate": 4.599838005965618e-08,
      "loss": 0.8346,
      "step": 6991
    },
    {
      "epoch": 0.6911993673232336,
      "grad_norm": 3.405936658631345,
      "learning_rate": 4.597143210007735e-08,
      "loss": 0.7511,
      "step": 6992
    },
    {
      "epoch": 0.691298223067988,
      "grad_norm": 4.112167731098671,
      "learning_rate": 4.5944489680331164e-08,
      "loss": 0.6474,
      "step": 6993
    },
    {
      "epoch": 0.6913970788127425,
      "grad_norm": 3.55792547605359,
      "learning_rate": 4.591755280318025e-08,
      "loss": 0.7022,
      "step": 6994
    },
    {
      "epoch": 0.691495934557497,
      "grad_norm": 3.783431799249201,
      "learning_rate": 4.589062147138647e-08,
      "loss": 0.7833,
      "step": 6995
    },
    {
      "epoch": 0.6915947903022515,
      "grad_norm": 4.399061423112457,
      "learning_rate": 4.586369568771129e-08,
      "loss": 0.7456,
      "step": 6996
    },
    {
      "epoch": 0.6916936460470059,
      "grad_norm": 4.172590673669197,
      "learning_rate": 4.583677545491559e-08,
      "loss": 0.6749,
      "step": 6997
    },
    {
      "epoch": 0.6917925017917603,
      "grad_norm": 3.170618759000232,
      "learning_rate": 4.580986077575956e-08,
      "loss": 0.6892,
      "step": 6998
    },
    {
      "epoch": 0.6918913575365149,
      "grad_norm": 4.082652316355774,
      "learning_rate": 4.5782951653002986e-08,
      "loss": 0.706,
      "step": 6999
    },
    {
      "epoch": 0.6919902132812693,
      "grad_norm": 4.147069141668596,
      "learning_rate": 4.575604808940493e-08,
      "loss": 0.6585,
      "step": 7000
    },
    {
      "epoch": 0.6920890690260237,
      "grad_norm": 3.4024669367827287,
      "learning_rate": 4.572915008772402e-08,
      "loss": 0.7437,
      "step": 7001
    },
    {
      "epoch": 0.6921879247707783,
      "grad_norm": 9.821484566344866,
      "learning_rate": 4.5702257650718226e-08,
      "loss": 0.7625,
      "step": 7002
    },
    {
      "epoch": 0.6922867805155327,
      "grad_norm": 2.8417534691742126,
      "learning_rate": 4.567537078114492e-08,
      "loss": 0.6546,
      "step": 7003
    },
    {
      "epoch": 0.6923856362602872,
      "grad_norm": 3.3909536373141513,
      "learning_rate": 4.564848948176099e-08,
      "loss": 0.6391,
      "step": 7004
    },
    {
      "epoch": 0.6924844920050417,
      "grad_norm": 3.662450905195977,
      "learning_rate": 4.562161375532276e-08,
      "loss": 0.6067,
      "step": 7005
    },
    {
      "epoch": 0.6925833477497961,
      "grad_norm": 5.217868580566663,
      "learning_rate": 4.559474360458587e-08,
      "loss": 0.7441,
      "step": 7006
    },
    {
      "epoch": 0.6926822034945506,
      "grad_norm": 3.797889082697075,
      "learning_rate": 4.556787903230549e-08,
      "loss": 0.7772,
      "step": 7007
    },
    {
      "epoch": 0.692781059239305,
      "grad_norm": 5.981689063556762,
      "learning_rate": 4.554102004123622e-08,
      "loss": 0.7744,
      "step": 7008
    },
    {
      "epoch": 0.6928799149840595,
      "grad_norm": 14.40533530994454,
      "learning_rate": 4.551416663413199e-08,
      "loss": 0.6073,
      "step": 7009
    },
    {
      "epoch": 0.692978770728814,
      "grad_norm": 4.811375091166377,
      "learning_rate": 4.548731881374623e-08,
      "loss": 0.7217,
      "step": 7010
    },
    {
      "epoch": 0.6930776264735684,
      "grad_norm": 13.01851273427871,
      "learning_rate": 4.546047658283186e-08,
      "loss": 0.6341,
      "step": 7011
    },
    {
      "epoch": 0.693176482218323,
      "grad_norm": 2.8729040704522197,
      "learning_rate": 4.5433639944141055e-08,
      "loss": 0.6061,
      "step": 7012
    },
    {
      "epoch": 0.6932753379630774,
      "grad_norm": 4.240744459015942,
      "learning_rate": 4.5406808900425605e-08,
      "loss": 0.7378,
      "step": 7013
    },
    {
      "epoch": 0.6933741937078318,
      "grad_norm": 5.6521093090528485,
      "learning_rate": 4.537998345443657e-08,
      "loss": 0.613,
      "step": 7014
    },
    {
      "epoch": 0.6934730494525864,
      "grad_norm": 4.485684184463186,
      "learning_rate": 4.535316360892451e-08,
      "loss": 0.7241,
      "step": 7015
    },
    {
      "epoch": 0.6935719051973408,
      "grad_norm": 3.3234221227069014,
      "learning_rate": 4.532634936663947e-08,
      "loss": 0.6685,
      "step": 7016
    },
    {
      "epoch": 0.6936707609420952,
      "grad_norm": 4.984493110485975,
      "learning_rate": 4.5299540730330774e-08,
      "loss": 0.6965,
      "step": 7017
    },
    {
      "epoch": 0.6937696166868497,
      "grad_norm": 4.515608463647462,
      "learning_rate": 4.527273770274732e-08,
      "loss": 0.7289,
      "step": 7018
    },
    {
      "epoch": 0.6938684724316042,
      "grad_norm": 5.292856165379726,
      "learning_rate": 4.524594028663734e-08,
      "loss": 0.6921,
      "step": 7019
    },
    {
      "epoch": 0.6939673281763586,
      "grad_norm": 3.0354016105813764,
      "learning_rate": 4.521914848474846e-08,
      "loss": 0.779,
      "step": 7020
    },
    {
      "epoch": 0.6940661839211131,
      "grad_norm": 14.324095130019321,
      "learning_rate": 4.519236229982783e-08,
      "loss": 0.7788,
      "step": 7021
    },
    {
      "epoch": 0.6941650396658676,
      "grad_norm": 4.204408069559874,
      "learning_rate": 4.516558173462202e-08,
      "loss": 0.6728,
      "step": 7022
    },
    {
      "epoch": 0.6942638954106221,
      "grad_norm": 7.110813126564625,
      "learning_rate": 4.5138806791876884e-08,
      "loss": 0.6098,
      "step": 7023
    },
    {
      "epoch": 0.6943627511553765,
      "grad_norm": 4.146005771157704,
      "learning_rate": 4.5112037474337904e-08,
      "loss": 0.7019,
      "step": 7024
    },
    {
      "epoch": 0.694461606900131,
      "grad_norm": 5.935757855069707,
      "learning_rate": 4.5085273784749775e-08,
      "loss": 0.745,
      "step": 7025
    },
    {
      "epoch": 0.6945604626448855,
      "grad_norm": 11.45314793580284,
      "learning_rate": 4.5058515725856785e-08,
      "loss": 0.7465,
      "step": 7026
    },
    {
      "epoch": 0.6946593183896399,
      "grad_norm": 3.9245967108397006,
      "learning_rate": 4.5031763300402594e-08,
      "loss": 0.6676,
      "step": 7027
    },
    {
      "epoch": 0.6947581741343943,
      "grad_norm": 3.8409359679008572,
      "learning_rate": 4.5005016511130203e-08,
      "loss": 0.6195,
      "step": 7028
    },
    {
      "epoch": 0.6948570298791489,
      "grad_norm": 5.150819451275333,
      "learning_rate": 4.4978275360782136e-08,
      "loss": 0.7264,
      "step": 7029
    },
    {
      "epoch": 0.6949558856239033,
      "grad_norm": 3.16579865386107,
      "learning_rate": 4.495153985210034e-08,
      "loss": 0.68,
      "step": 7030
    },
    {
      "epoch": 0.6950547413686577,
      "grad_norm": 3.421796146852053,
      "learning_rate": 4.492480998782607e-08,
      "loss": 0.7373,
      "step": 7031
    },
    {
      "epoch": 0.6951535971134123,
      "grad_norm": 3.666440588925384,
      "learning_rate": 4.4898085770700125e-08,
      "loss": 0.7259,
      "step": 7032
    },
    {
      "epoch": 0.6952524528581667,
      "grad_norm": 4.465128853745658,
      "learning_rate": 4.487136720346271e-08,
      "loss": 0.7244,
      "step": 7033
    },
    {
      "epoch": 0.6953513086029212,
      "grad_norm": 5.78350564710398,
      "learning_rate": 4.484465428885335e-08,
      "loss": 0.7129,
      "step": 7034
    },
    {
      "epoch": 0.6954501643476757,
      "grad_norm": 3.7295906491059756,
      "learning_rate": 4.4817947029611125e-08,
      "loss": 0.7558,
      "step": 7035
    },
    {
      "epoch": 0.6955490200924301,
      "grad_norm": 3.7811358501318884,
      "learning_rate": 4.479124542847443e-08,
      "loss": 0.6949,
      "step": 7036
    },
    {
      "epoch": 0.6956478758371846,
      "grad_norm": 4.639442056557572,
      "learning_rate": 4.4764549488181116e-08,
      "loss": 0.7043,
      "step": 7037
    },
    {
      "epoch": 0.695746731581939,
      "grad_norm": 4.075285133289127,
      "learning_rate": 4.473785921146849e-08,
      "loss": 0.7531,
      "step": 7038
    },
    {
      "epoch": 0.6958455873266935,
      "grad_norm": 5.4340791258996015,
      "learning_rate": 4.47111746010732e-08,
      "loss": 0.6943,
      "step": 7039
    },
    {
      "epoch": 0.695944443071448,
      "grad_norm": 4.074630723378779,
      "learning_rate": 4.468449565973138e-08,
      "loss": 0.5914,
      "step": 7040
    },
    {
      "epoch": 0.6960432988162024,
      "grad_norm": 3.8061809138914238,
      "learning_rate": 4.465782239017861e-08,
      "loss": 0.7755,
      "step": 7041
    },
    {
      "epoch": 0.696142154560957,
      "grad_norm": 5.691660826362966,
      "learning_rate": 4.4631154795149764e-08,
      "loss": 0.7417,
      "step": 7042
    },
    {
      "epoch": 0.6962410103057114,
      "grad_norm": 15.537355902641137,
      "learning_rate": 4.460449287737924e-08,
      "loss": 0.772,
      "step": 7043
    },
    {
      "epoch": 0.6963398660504658,
      "grad_norm": 4.206928426433293,
      "learning_rate": 4.4577836639600876e-08,
      "loss": 0.7213,
      "step": 7044
    },
    {
      "epoch": 0.6964387217952204,
      "grad_norm": 4.336555186801764,
      "learning_rate": 4.4551186084547784e-08,
      "loss": 0.6845,
      "step": 7045
    },
    {
      "epoch": 0.6965375775399748,
      "grad_norm": 3.2330924777576646,
      "learning_rate": 4.452454121495267e-08,
      "loss": 0.7612,
      "step": 7046
    },
    {
      "epoch": 0.6966364332847292,
      "grad_norm": 136.36339910817512,
      "learning_rate": 4.449790203354751e-08,
      "loss": 0.6485,
      "step": 7047
    },
    {
      "epoch": 0.6967352890294837,
      "grad_norm": 2.977048798206112,
      "learning_rate": 4.447126854306377e-08,
      "loss": 0.6409,
      "step": 7048
    },
    {
      "epoch": 0.6968341447742382,
      "grad_norm": 4.937909238594574,
      "learning_rate": 4.4444640746232377e-08,
      "loss": 0.7296,
      "step": 7049
    },
    {
      "epoch": 0.6969330005189927,
      "grad_norm": 3.516311597379479,
      "learning_rate": 4.441801864578353e-08,
      "loss": 0.6764,
      "step": 7050
    },
    {
      "epoch": 0.6970318562637471,
      "grad_norm": 11.572678323382982,
      "learning_rate": 4.4391402244446985e-08,
      "loss": 0.8202,
      "step": 7051
    },
    {
      "epoch": 0.6971307120085016,
      "grad_norm": 22.22791551057962,
      "learning_rate": 4.43647915449519e-08,
      "loss": 0.6876,
      "step": 7052
    },
    {
      "epoch": 0.6972295677532561,
      "grad_norm": 4.723386750588535,
      "learning_rate": 4.4338186550026756e-08,
      "loss": 0.6855,
      "step": 7053
    },
    {
      "epoch": 0.6973284234980105,
      "grad_norm": 3.9965873336255804,
      "learning_rate": 4.4311587262399475e-08,
      "loss": 0.7287,
      "step": 7054
    },
    {
      "epoch": 0.697427279242765,
      "grad_norm": 5.216397754384862,
      "learning_rate": 4.42849936847975e-08,
      "loss": 0.7498,
      "step": 7055
    },
    {
      "epoch": 0.6975261349875195,
      "grad_norm": 5.551847374625398,
      "learning_rate": 4.425840581994752e-08,
      "loss": 0.715,
      "step": 7056
    },
    {
      "epoch": 0.6976249907322739,
      "grad_norm": 3.7092672429705607,
      "learning_rate": 4.423182367057583e-08,
      "loss": 0.7424,
      "step": 7057
    },
    {
      "epoch": 0.6977238464770283,
      "grad_norm": 4.080197852458536,
      "learning_rate": 4.420524723940794e-08,
      "loss": 0.795,
      "step": 7058
    },
    {
      "epoch": 0.6978227022217829,
      "grad_norm": 3.923023988000037,
      "learning_rate": 4.4178676529168915e-08,
      "loss": 0.7831,
      "step": 7059
    },
    {
      "epoch": 0.6979215579665373,
      "grad_norm": 3.3750621450988905,
      "learning_rate": 4.4152111542583216e-08,
      "loss": 0.7331,
      "step": 7060
    },
    {
      "epoch": 0.6980204137112918,
      "grad_norm": 4.972138888702306,
      "learning_rate": 4.412555228237464e-08,
      "loss": 0.6147,
      "step": 7061
    },
    {
      "epoch": 0.6981192694560463,
      "grad_norm": 3.4778845510959324,
      "learning_rate": 4.409899875126647e-08,
      "loss": 0.7695,
      "step": 7062
    },
    {
      "epoch": 0.6982181252008007,
      "grad_norm": 5.190071433585254,
      "learning_rate": 4.407245095198141e-08,
      "loss": 0.6722,
      "step": 7063
    },
    {
      "epoch": 0.6983169809455552,
      "grad_norm": 5.0893367855025184,
      "learning_rate": 4.404590888724149e-08,
      "loss": 0.6719,
      "step": 7064
    },
    {
      "epoch": 0.6984158366903097,
      "grad_norm": 13.689314178992209,
      "learning_rate": 4.401937255976822e-08,
      "loss": 0.7426,
      "step": 7065
    },
    {
      "epoch": 0.6985146924350641,
      "grad_norm": 3.886432452230138,
      "learning_rate": 4.3992841972282566e-08,
      "loss": 0.7488,
      "step": 7066
    },
    {
      "epoch": 0.6986135481798186,
      "grad_norm": 4.29462273567633,
      "learning_rate": 4.396631712750477e-08,
      "loss": 0.6614,
      "step": 7067
    },
    {
      "epoch": 0.6987124039245731,
      "grad_norm": 3.6147096058032897,
      "learning_rate": 4.393979802815463e-08,
      "loss": 0.7249,
      "step": 7068
    },
    {
      "epoch": 0.6988112596693276,
      "grad_norm": 13.362529953846478,
      "learning_rate": 4.391328467695122e-08,
      "loss": 0.6867,
      "step": 7069
    },
    {
      "epoch": 0.698910115414082,
      "grad_norm": 4.805445614567799,
      "learning_rate": 4.388677707661317e-08,
      "loss": 0.7476,
      "step": 7070
    },
    {
      "epoch": 0.6990089711588364,
      "grad_norm": 4.002959406047479,
      "learning_rate": 4.3860275229858353e-08,
      "loss": 0.694,
      "step": 7071
    },
    {
      "epoch": 0.699107826903591,
      "grad_norm": 3.634920035259162,
      "learning_rate": 4.3833779139404245e-08,
      "loss": 0.718,
      "step": 7072
    },
    {
      "epoch": 0.6992066826483454,
      "grad_norm": 4.463016932780265,
      "learning_rate": 4.3807288807967525e-08,
      "loss": 0.8144,
      "step": 7073
    },
    {
      "epoch": 0.6993055383930998,
      "grad_norm": 4.443265671485834,
      "learning_rate": 4.378080423826449e-08,
      "loss": 0.7437,
      "step": 7074
    },
    {
      "epoch": 0.6994043941378544,
      "grad_norm": 5.115215769795058,
      "learning_rate": 4.3754325433010645e-08,
      "loss": 0.8109,
      "step": 7075
    },
    {
      "epoch": 0.6995032498826088,
      "grad_norm": 6.609490572834012,
      "learning_rate": 4.372785239492104e-08,
      "loss": 0.5913,
      "step": 7076
    },
    {
      "epoch": 0.6996021056273632,
      "grad_norm": 3.556834520991442,
      "learning_rate": 4.3701385126710136e-08,
      "loss": 0.6607,
      "step": 7077
    },
    {
      "epoch": 0.6997009613721178,
      "grad_norm": 3.503267377482586,
      "learning_rate": 4.367492363109169e-08,
      "loss": 0.7208,
      "step": 7078
    },
    {
      "epoch": 0.6997998171168722,
      "grad_norm": 3.1412317102400085,
      "learning_rate": 4.364846791077901e-08,
      "loss": 0.6978,
      "step": 7079
    },
    {
      "epoch": 0.6998986728616267,
      "grad_norm": 4.63413506192923,
      "learning_rate": 4.362201796848466e-08,
      "loss": 0.7937,
      "step": 7080
    },
    {
      "epoch": 0.6999975286063811,
      "grad_norm": 4.597505101723923,
      "learning_rate": 4.359557380692074e-08,
      "loss": 0.6991,
      "step": 7081
    },
    {
      "epoch": 0.7000963843511356,
      "grad_norm": 3.365196526002669,
      "learning_rate": 4.3569135428798734e-08,
      "loss": 0.6017,
      "step": 7082
    },
    {
      "epoch": 0.7001952400958901,
      "grad_norm": 9.478094752797892,
      "learning_rate": 4.354270283682945e-08,
      "loss": 0.7241,
      "step": 7083
    },
    {
      "epoch": 0.7002940958406445,
      "grad_norm": 6.457823557372031,
      "learning_rate": 4.3516276033723174e-08,
      "loss": 0.6037,
      "step": 7084
    },
    {
      "epoch": 0.700392951585399,
      "grad_norm": 3.9380422882965562,
      "learning_rate": 4.3489855022189634e-08,
      "loss": 0.5678,
      "step": 7085
    },
    {
      "epoch": 0.7004918073301535,
      "grad_norm": 3.367728219530565,
      "learning_rate": 4.346343980493785e-08,
      "loss": 0.6664,
      "step": 7086
    },
    {
      "epoch": 0.7005906630749079,
      "grad_norm": 4.40610125053866,
      "learning_rate": 4.343703038467638e-08,
      "loss": 0.7944,
      "step": 7087
    },
    {
      "epoch": 0.7006895188196625,
      "grad_norm": 3.1311042868791215,
      "learning_rate": 4.341062676411303e-08,
      "loss": 0.725,
      "step": 7088
    },
    {
      "epoch": 0.7007883745644169,
      "grad_norm": 3.485293479105385,
      "learning_rate": 4.3384228945955205e-08,
      "loss": 0.6667,
      "step": 7089
    },
    {
      "epoch": 0.7008872303091713,
      "grad_norm": 20.987759454480397,
      "learning_rate": 4.335783693290955e-08,
      "loss": 0.7054,
      "step": 7090
    },
    {
      "epoch": 0.7009860860539258,
      "grad_norm": 5.051812844773426,
      "learning_rate": 4.333145072768215e-08,
      "loss": 0.7161,
      "step": 7091
    },
    {
      "epoch": 0.7010849417986803,
      "grad_norm": 4.8723660761958945,
      "learning_rate": 4.3305070332978563e-08,
      "loss": 0.7809,
      "step": 7092
    },
    {
      "epoch": 0.7011837975434347,
      "grad_norm": 6.653228636101896,
      "learning_rate": 4.3278695751503736e-08,
      "loss": 0.6777,
      "step": 7093
    },
    {
      "epoch": 0.7012826532881892,
      "grad_norm": 3.7535631666686946,
      "learning_rate": 4.325232698596193e-08,
      "loss": 0.7351,
      "step": 7094
    },
    {
      "epoch": 0.7013815090329437,
      "grad_norm": 3.75088931722641,
      "learning_rate": 4.322596403905691e-08,
      "loss": 0.7055,
      "step": 7095
    },
    {
      "epoch": 0.7014803647776982,
      "grad_norm": 3.232997524104842,
      "learning_rate": 4.3199606913491824e-08,
      "loss": 0.6897,
      "step": 7096
    },
    {
      "epoch": 0.7015792205224526,
      "grad_norm": 12.092553304005346,
      "learning_rate": 4.317325561196916e-08,
      "loss": 0.689,
      "step": 7097
    },
    {
      "epoch": 0.7016780762672071,
      "grad_norm": 5.726537049770308,
      "learning_rate": 4.3146910137190876e-08,
      "loss": 0.7249,
      "step": 7098
    },
    {
      "epoch": 0.7017769320119616,
      "grad_norm": 3.555417878662625,
      "learning_rate": 4.312057049185834e-08,
      "loss": 0.7228,
      "step": 7099
    },
    {
      "epoch": 0.701875787756716,
      "grad_norm": 9.247607530788189,
      "learning_rate": 4.3094236678672256e-08,
      "loss": 0.7404,
      "step": 7100
    },
    {
      "epoch": 0.7019746435014704,
      "grad_norm": 2.933228533227474,
      "learning_rate": 4.30679087003328e-08,
      "loss": 0.6708,
      "step": 7101
    },
    {
      "epoch": 0.702073499246225,
      "grad_norm": 3.245974214667151,
      "learning_rate": 4.3041586559539466e-08,
      "loss": 0.6951,
      "step": 7102
    },
    {
      "epoch": 0.7021723549909794,
      "grad_norm": 3.961343344308368,
      "learning_rate": 4.301527025899124e-08,
      "loss": 0.7106,
      "step": 7103
    },
    {
      "epoch": 0.7022712107357338,
      "grad_norm": 3.4102726668129093,
      "learning_rate": 4.2988959801386506e-08,
      "loss": 0.6328,
      "step": 7104
    },
    {
      "epoch": 0.7023700664804884,
      "grad_norm": 3.0778986151463923,
      "learning_rate": 4.296265518942292e-08,
      "loss": 0.6228,
      "step": 7105
    },
    {
      "epoch": 0.7024689222252428,
      "grad_norm": 6.765650497131367,
      "learning_rate": 4.2936356425797734e-08,
      "loss": 0.6811,
      "step": 7106
    },
    {
      "epoch": 0.7025677779699973,
      "grad_norm": 13.977110989176442,
      "learning_rate": 4.291006351320744e-08,
      "loss": 0.7232,
      "step": 7107
    },
    {
      "epoch": 0.7026666337147518,
      "grad_norm": 57.596676762599046,
      "learning_rate": 4.288377645434797e-08,
      "loss": 0.7356,
      "step": 7108
    },
    {
      "epoch": 0.7027654894595062,
      "grad_norm": 4.290878052886105,
      "learning_rate": 4.2857495251914686e-08,
      "loss": 0.7631,
      "step": 7109
    },
    {
      "epoch": 0.7028643452042607,
      "grad_norm": 5.53437944069947,
      "learning_rate": 4.28312199086024e-08,
      "loss": 0.6355,
      "step": 7110
    },
    {
      "epoch": 0.7029632009490151,
      "grad_norm": 3.616164757006522,
      "learning_rate": 4.280495042710518e-08,
      "loss": 0.6466,
      "step": 7111
    },
    {
      "epoch": 0.7030620566937696,
      "grad_norm": 2.946669126898014,
      "learning_rate": 4.277868681011664e-08,
      "loss": 0.6664,
      "step": 7112
    },
    {
      "epoch": 0.7031609124385241,
      "grad_norm": 6.7022489805688314,
      "learning_rate": 4.2752429060329656e-08,
      "loss": 0.7145,
      "step": 7113
    },
    {
      "epoch": 0.7032597681832785,
      "grad_norm": 12.275474720841915,
      "learning_rate": 4.272617718043661e-08,
      "loss": 0.7119,
      "step": 7114
    },
    {
      "epoch": 0.703358623928033,
      "grad_norm": 4.513289535796356,
      "learning_rate": 4.2699931173129286e-08,
      "loss": 0.6515,
      "step": 7115
    },
    {
      "epoch": 0.7034574796727875,
      "grad_norm": 40.80113905035267,
      "learning_rate": 4.2673691041098746e-08,
      "loss": 0.6651,
      "step": 7116
    },
    {
      "epoch": 0.7035563354175419,
      "grad_norm": 4.552578260540014,
      "learning_rate": 4.264745678703556e-08,
      "loss": 0.7741,
      "step": 7117
    },
    {
      "epoch": 0.7036551911622965,
      "grad_norm": 4.001574193412799,
      "learning_rate": 4.2621228413629715e-08,
      "loss": 0.7352,
      "step": 7118
    },
    {
      "epoch": 0.7037540469070509,
      "grad_norm": 12.000232701614044,
      "learning_rate": 4.259500592357047e-08,
      "loss": 0.7251,
      "step": 7119
    },
    {
      "epoch": 0.7038529026518053,
      "grad_norm": 7.09439459186691,
      "learning_rate": 4.256878931954657e-08,
      "loss": 0.617,
      "step": 7120
    },
    {
      "epoch": 0.7039517583965598,
      "grad_norm": 4.089826996412454,
      "learning_rate": 4.25425786042462e-08,
      "loss": 0.8402,
      "step": 7121
    },
    {
      "epoch": 0.7040506141413143,
      "grad_norm": 23.906156185720793,
      "learning_rate": 4.251637378035681e-08,
      "loss": 0.7931,
      "step": 7122
    },
    {
      "epoch": 0.7041494698860687,
      "grad_norm": 3.1789710913045193,
      "learning_rate": 4.249017485056536e-08,
      "loss": 0.7957,
      "step": 7123
    },
    {
      "epoch": 0.7042483256308232,
      "grad_norm": 11.08731434494615,
      "learning_rate": 4.246398181755817e-08,
      "loss": 0.7068,
      "step": 7124
    },
    {
      "epoch": 0.7043471813755777,
      "grad_norm": 10.92260393985129,
      "learning_rate": 4.243779468402089e-08,
      "loss": 0.7054,
      "step": 7125
    },
    {
      "epoch": 0.7044460371203322,
      "grad_norm": 5.098222294923528,
      "learning_rate": 4.2411613452638686e-08,
      "loss": 0.7201,
      "step": 7126
    },
    {
      "epoch": 0.7045448928650866,
      "grad_norm": 4.23680631877255,
      "learning_rate": 4.238543812609601e-08,
      "loss": 0.7509,
      "step": 7127
    },
    {
      "epoch": 0.7046437486098411,
      "grad_norm": 4.27494069813554,
      "learning_rate": 4.235926870707677e-08,
      "loss": 0.7784,
      "step": 7128
    },
    {
      "epoch": 0.7047426043545956,
      "grad_norm": 5.23603896608559,
      "learning_rate": 4.233310519826431e-08,
      "loss": 0.686,
      "step": 7129
    },
    {
      "epoch": 0.70484146009935,
      "grad_norm": 3.176026712239949,
      "learning_rate": 4.230694760234121e-08,
      "loss": 0.6351,
      "step": 7130
    },
    {
      "epoch": 0.7049403158441044,
      "grad_norm": 3.3142590296602465,
      "learning_rate": 4.228079592198962e-08,
      "loss": 0.6233,
      "step": 7131
    },
    {
      "epoch": 0.705039171588859,
      "grad_norm": 4.9303057926105325,
      "learning_rate": 4.2254650159891026e-08,
      "loss": 0.7797,
      "step": 7132
    },
    {
      "epoch": 0.7051380273336134,
      "grad_norm": 4.5423075325528846,
      "learning_rate": 4.2228510318726206e-08,
      "loss": 0.7655,
      "step": 7133
    },
    {
      "epoch": 0.7052368830783678,
      "grad_norm": 3.7352274381409383,
      "learning_rate": 4.2202376401175474e-08,
      "loss": 0.719,
      "step": 7134
    },
    {
      "epoch": 0.7053357388231224,
      "grad_norm": 4.5461994146137465,
      "learning_rate": 4.2176248409918504e-08,
      "loss": 0.6398,
      "step": 7135
    },
    {
      "epoch": 0.7054345945678768,
      "grad_norm": 3.79267438194552,
      "learning_rate": 4.215012634763425e-08,
      "loss": 0.6724,
      "step": 7136
    },
    {
      "epoch": 0.7055334503126313,
      "grad_norm": 6.9236159618297775,
      "learning_rate": 4.2124010217001236e-08,
      "loss": 0.6627,
      "step": 7137
    },
    {
      "epoch": 0.7056323060573858,
      "grad_norm": 7.424118220359549,
      "learning_rate": 4.209790002069721e-08,
      "loss": 0.7533,
      "step": 7138
    },
    {
      "epoch": 0.7057311618021402,
      "grad_norm": 3.3115339338439447,
      "learning_rate": 4.207179576139945e-08,
      "loss": 0.7038,
      "step": 7139
    },
    {
      "epoch": 0.7058300175468947,
      "grad_norm": 4.885036892760342,
      "learning_rate": 4.204569744178451e-08,
      "loss": 0.7131,
      "step": 7140
    },
    {
      "epoch": 0.7059288732916491,
      "grad_norm": 3.3687687444852186,
      "learning_rate": 4.2019605064528444e-08,
      "loss": 0.631,
      "step": 7141
    },
    {
      "epoch": 0.7060277290364037,
      "grad_norm": 3.949452937226992,
      "learning_rate": 4.1993518632306565e-08,
      "loss": 0.8222,
      "step": 7142
    },
    {
      "epoch": 0.7061265847811581,
      "grad_norm": 4.061439592842218,
      "learning_rate": 4.196743814779373e-08,
      "loss": 0.7126,
      "step": 7143
    },
    {
      "epoch": 0.7062254405259125,
      "grad_norm": 3.3878597093092684,
      "learning_rate": 4.1941363613664046e-08,
      "loss": 0.7683,
      "step": 7144
    },
    {
      "epoch": 0.7063242962706671,
      "grad_norm": 4.233334140149833,
      "learning_rate": 4.1915295032591095e-08,
      "loss": 0.783,
      "step": 7145
    },
    {
      "epoch": 0.7064231520154215,
      "grad_norm": 6.098846126414787,
      "learning_rate": 4.1889232407247856e-08,
      "loss": 0.6928,
      "step": 7146
    },
    {
      "epoch": 0.7065220077601759,
      "grad_norm": 3.6644696303036803,
      "learning_rate": 4.186317574030661e-08,
      "loss": 0.7197,
      "step": 7147
    },
    {
      "epoch": 0.7066208635049305,
      "grad_norm": 3.5804933897021014,
      "learning_rate": 4.1837125034439145e-08,
      "loss": 0.6973,
      "step": 7148
    },
    {
      "epoch": 0.7067197192496849,
      "grad_norm": 3.666759106170043,
      "learning_rate": 4.181108029231651e-08,
      "loss": 0.6663,
      "step": 7149
    },
    {
      "epoch": 0.7068185749944393,
      "grad_norm": 3.761212372800903,
      "learning_rate": 4.1785041516609256e-08,
      "loss": 0.7063,
      "step": 7150
    },
    {
      "epoch": 0.7069174307391939,
      "grad_norm": 6.367274296889707,
      "learning_rate": 4.1759008709987284e-08,
      "loss": 0.6299,
      "step": 7151
    },
    {
      "epoch": 0.7070162864839483,
      "grad_norm": 3.5149682323889277,
      "learning_rate": 4.173298187511983e-08,
      "loss": 0.686,
      "step": 7152
    },
    {
      "epoch": 0.7071151422287028,
      "grad_norm": 6.376683677952883,
      "learning_rate": 4.170696101467558e-08,
      "loss": 0.697,
      "step": 7153
    },
    {
      "epoch": 0.7072139979734572,
      "grad_norm": 3.960716441720981,
      "learning_rate": 4.168094613132262e-08,
      "loss": 0.6653,
      "step": 7154
    },
    {
      "epoch": 0.7073128537182117,
      "grad_norm": 4.92844524538202,
      "learning_rate": 4.165493722772835e-08,
      "loss": 0.7083,
      "step": 7155
    },
    {
      "epoch": 0.7074117094629662,
      "grad_norm": 8.287568434528591,
      "learning_rate": 4.162893430655965e-08,
      "loss": 0.8366,
      "step": 7156
    },
    {
      "epoch": 0.7075105652077206,
      "grad_norm": 3.304248141586816,
      "learning_rate": 4.160293737048266e-08,
      "loss": 0.7186,
      "step": 7157
    },
    {
      "epoch": 0.7076094209524751,
      "grad_norm": 8.129703731912658,
      "learning_rate": 4.157694642216306e-08,
      "loss": 0.6908,
      "step": 7158
    },
    {
      "epoch": 0.7077082766972296,
      "grad_norm": 3.4565816665807527,
      "learning_rate": 4.1550961464265776e-08,
      "loss": 0.8496,
      "step": 7159
    },
    {
      "epoch": 0.707807132441984,
      "grad_norm": 3.2320154504039333,
      "learning_rate": 4.1524982499455244e-08,
      "loss": 0.6824,
      "step": 7160
    },
    {
      "epoch": 0.7079059881867386,
      "grad_norm": 4.812802317050291,
      "learning_rate": 4.1499009530395143e-08,
      "loss": 0.738,
      "step": 7161
    },
    {
      "epoch": 0.708004843931493,
      "grad_norm": 5.513297795468437,
      "learning_rate": 4.147304255974872e-08,
      "loss": 0.6456,
      "step": 7162
    },
    {
      "epoch": 0.7081036996762474,
      "grad_norm": 3.8000332682859796,
      "learning_rate": 4.144708159017839e-08,
      "loss": 0.8356,
      "step": 7163
    },
    {
      "epoch": 0.7082025554210019,
      "grad_norm": 5.1156434128363415,
      "learning_rate": 4.142112662434615e-08,
      "loss": 0.6468,
      "step": 7164
    },
    {
      "epoch": 0.7083014111657564,
      "grad_norm": 6.3081443432214375,
      "learning_rate": 4.1395177664913294e-08,
      "loss": 0.7103,
      "step": 7165
    },
    {
      "epoch": 0.7084002669105108,
      "grad_norm": 3.5735351468605776,
      "learning_rate": 4.136923471454046e-08,
      "loss": 0.7784,
      "step": 7166
    },
    {
      "epoch": 0.7084991226552653,
      "grad_norm": 2.8034085704790086,
      "learning_rate": 4.134329777588773e-08,
      "loss": 0.6378,
      "step": 7167
    },
    {
      "epoch": 0.7085979784000198,
      "grad_norm": 7.397070278415529,
      "learning_rate": 4.131736685161461e-08,
      "loss": 0.7487,
      "step": 7168
    },
    {
      "epoch": 0.7086968341447742,
      "grad_norm": 10.016487416425305,
      "learning_rate": 4.129144194437986e-08,
      "loss": 0.6969,
      "step": 7169
    },
    {
      "epoch": 0.7087956898895287,
      "grad_norm": 4.4050566382880865,
      "learning_rate": 4.1265523056841745e-08,
      "loss": 0.837,
      "step": 7170
    },
    {
      "epoch": 0.7088945456342832,
      "grad_norm": 3.4584459064716753,
      "learning_rate": 4.1239610191657826e-08,
      "loss": 0.6528,
      "step": 7171
    },
    {
      "epoch": 0.7089934013790377,
      "grad_norm": 3.4440750620595773,
      "learning_rate": 4.12137033514851e-08,
      "loss": 0.6651,
      "step": 7172
    },
    {
      "epoch": 0.7090922571237921,
      "grad_norm": 7.10471008497355,
      "learning_rate": 4.118780253897996e-08,
      "loss": 0.7433,
      "step": 7173
    },
    {
      "epoch": 0.7091911128685465,
      "grad_norm": 3.1352884422694647,
      "learning_rate": 4.1161907756798106e-08,
      "loss": 0.71,
      "step": 7174
    },
    {
      "epoch": 0.7092899686133011,
      "grad_norm": 5.843182262736252,
      "learning_rate": 4.113601900759474e-08,
      "loss": 0.6737,
      "step": 7175
    },
    {
      "epoch": 0.7093888243580555,
      "grad_norm": 6.110978461766818,
      "learning_rate": 4.111013629402425e-08,
      "loss": 0.6147,
      "step": 7176
    },
    {
      "epoch": 0.7094876801028099,
      "grad_norm": 4.253386340574145,
      "learning_rate": 4.108425961874066e-08,
      "loss": 0.7472,
      "step": 7177
    },
    {
      "epoch": 0.7095865358475645,
      "grad_norm": 6.872792306168577,
      "learning_rate": 4.105838898439713e-08,
      "loss": 0.7651,
      "step": 7178
    },
    {
      "epoch": 0.7096853915923189,
      "grad_norm": 2.7194275676275987,
      "learning_rate": 4.1032524393646394e-08,
      "loss": 0.6783,
      "step": 7179
    },
    {
      "epoch": 0.7097842473370733,
      "grad_norm": 8.709779837841921,
      "learning_rate": 4.100666584914042e-08,
      "loss": 0.8754,
      "step": 7180
    },
    {
      "epoch": 0.7098831030818279,
      "grad_norm": 9.511570800402716,
      "learning_rate": 4.0980813353530686e-08,
      "loss": 0.8333,
      "step": 7181
    },
    {
      "epoch": 0.7099819588265823,
      "grad_norm": 4.9829555573695545,
      "learning_rate": 4.0954966909467905e-08,
      "loss": 0.738,
      "step": 7182
    },
    {
      "epoch": 0.7100808145713368,
      "grad_norm": 5.479746071182573,
      "learning_rate": 4.09291265196023e-08,
      "loss": 0.701,
      "step": 7183
    },
    {
      "epoch": 0.7101796703160912,
      "grad_norm": 3.8583318393224286,
      "learning_rate": 4.090329218658345e-08,
      "loss": 0.7691,
      "step": 7184
    },
    {
      "epoch": 0.7102785260608457,
      "grad_norm": 3.4396114594089515,
      "learning_rate": 4.0877463913060206e-08,
      "loss": 0.6874,
      "step": 7185
    },
    {
      "epoch": 0.7103773818056002,
      "grad_norm": 11.375985972930724,
      "learning_rate": 4.085164170168092e-08,
      "loss": 0.6774,
      "step": 7186
    },
    {
      "epoch": 0.7104762375503546,
      "grad_norm": 3.7087432308194836,
      "learning_rate": 4.082582555509332e-08,
      "loss": 0.7291,
      "step": 7187
    },
    {
      "epoch": 0.7105750932951092,
      "grad_norm": 7.933637303930125,
      "learning_rate": 4.080001547594438e-08,
      "loss": 0.7768,
      "step": 7188
    },
    {
      "epoch": 0.7106739490398636,
      "grad_norm": 7.417532054182681,
      "learning_rate": 4.0774211466880606e-08,
      "loss": 0.7672,
      "step": 7189
    },
    {
      "epoch": 0.710772804784618,
      "grad_norm": 5.807702054391788,
      "learning_rate": 4.074841353054783e-08,
      "loss": 0.83,
      "step": 7190
    },
    {
      "epoch": 0.7108716605293726,
      "grad_norm": 3.9388409534754185,
      "learning_rate": 4.072262166959119e-08,
      "loss": 0.7362,
      "step": 7191
    },
    {
      "epoch": 0.710970516274127,
      "grad_norm": 4.336892855348351,
      "learning_rate": 4.0696835886655336e-08,
      "loss": 0.7666,
      "step": 7192
    },
    {
      "epoch": 0.7110693720188814,
      "grad_norm": 2.884561757033427,
      "learning_rate": 4.0671056184384175e-08,
      "loss": 0.6788,
      "step": 7193
    },
    {
      "epoch": 0.7111682277636359,
      "grad_norm": 8.111279396184418,
      "learning_rate": 4.0645282565421003e-08,
      "loss": 0.763,
      "step": 7194
    },
    {
      "epoch": 0.7112670835083904,
      "grad_norm": 3.69549837187306,
      "learning_rate": 4.06195150324086e-08,
      "loss": 0.7813,
      "step": 7195
    },
    {
      "epoch": 0.7113659392531448,
      "grad_norm": 4.090766688691093,
      "learning_rate": 4.059375358798897e-08,
      "loss": 0.6716,
      "step": 7196
    },
    {
      "epoch": 0.7114647949978993,
      "grad_norm": 3.3916770785395047,
      "learning_rate": 4.0567998234803604e-08,
      "loss": 0.7262,
      "step": 7197
    },
    {
      "epoch": 0.7115636507426538,
      "grad_norm": 4.453203661399148,
      "learning_rate": 4.054224897549337e-08,
      "loss": 0.6416,
      "step": 7198
    },
    {
      "epoch": 0.7116625064874083,
      "grad_norm": 3.6957868711703417,
      "learning_rate": 4.0516505812698396e-08,
      "loss": 0.6856,
      "step": 7199
    },
    {
      "epoch": 0.7117613622321627,
      "grad_norm": 4.741333338158835,
      "learning_rate": 4.049076874905832e-08,
      "loss": 0.715,
      "step": 7200
    },
    {
      "epoch": 0.7118602179769172,
      "grad_norm": 6.061234757732551,
      "learning_rate": 4.046503778721212e-08,
      "loss": 0.7746,
      "step": 7201
    },
    {
      "epoch": 0.7119590737216717,
      "grad_norm": 3.4095660697350634,
      "learning_rate": 4.0439312929798054e-08,
      "loss": 0.7549,
      "step": 7202
    },
    {
      "epoch": 0.7120579294664261,
      "grad_norm": 9.39283891751759,
      "learning_rate": 4.0413594179453906e-08,
      "loss": 0.6824,
      "step": 7203
    },
    {
      "epoch": 0.7121567852111805,
      "grad_norm": 3.7628073693443937,
      "learning_rate": 4.038788153881667e-08,
      "loss": 0.6742,
      "step": 7204
    },
    {
      "epoch": 0.7122556409559351,
      "grad_norm": 12.731873638290553,
      "learning_rate": 4.0362175010522836e-08,
      "loss": 0.7373,
      "step": 7205
    },
    {
      "epoch": 0.7123544967006895,
      "grad_norm": 4.279034018972368,
      "learning_rate": 4.033647459720828e-08,
      "loss": 0.6737,
      "step": 7206
    },
    {
      "epoch": 0.7124533524454439,
      "grad_norm": 3.8841529272756734,
      "learning_rate": 4.031078030150812e-08,
      "loss": 0.6266,
      "step": 7207
    },
    {
      "epoch": 0.7125522081901985,
      "grad_norm": 3.6332887966725425,
      "learning_rate": 4.028509212605696e-08,
      "loss": 0.6009,
      "step": 7208
    },
    {
      "epoch": 0.7126510639349529,
      "grad_norm": 4.313504831501446,
      "learning_rate": 4.0259410073488785e-08,
      "loss": 0.7247,
      "step": 7209
    },
    {
      "epoch": 0.7127499196797074,
      "grad_norm": 7.394639836883826,
      "learning_rate": 4.023373414643687e-08,
      "loss": 0.6785,
      "step": 7210
    },
    {
      "epoch": 0.7128487754244619,
      "grad_norm": 6.622793036804556,
      "learning_rate": 4.0208064347533865e-08,
      "loss": 0.6954,
      "step": 7211
    },
    {
      "epoch": 0.7129476311692163,
      "grad_norm": 49.12174230278148,
      "learning_rate": 4.018240067941191e-08,
      "loss": 0.8061,
      "step": 7212
    },
    {
      "epoch": 0.7130464869139708,
      "grad_norm": 5.4962449605388555,
      "learning_rate": 4.015674314470235e-08,
      "loss": 0.6688,
      "step": 7213
    },
    {
      "epoch": 0.7131453426587252,
      "grad_norm": 4.100387572989624,
      "learning_rate": 4.013109174603607e-08,
      "loss": 0.6952,
      "step": 7214
    },
    {
      "epoch": 0.7132441984034797,
      "grad_norm": 4.714319862904531,
      "learning_rate": 4.010544648604317e-08,
      "loss": 0.6099,
      "step": 7215
    },
    {
      "epoch": 0.7133430541482342,
      "grad_norm": 9.418752542828472,
      "learning_rate": 4.0079807367353226e-08,
      "loss": 0.6704,
      "step": 7216
    },
    {
      "epoch": 0.7134419098929886,
      "grad_norm": 7.441707733515792,
      "learning_rate": 4.005417439259518e-08,
      "loss": 0.679,
      "step": 7217
    },
    {
      "epoch": 0.7135407656377432,
      "grad_norm": 4.550048217578439,
      "learning_rate": 4.0028547564397255e-08,
      "loss": 0.7649,
      "step": 7218
    },
    {
      "epoch": 0.7136396213824976,
      "grad_norm": 3.902056909234297,
      "learning_rate": 4.0002926885387136e-08,
      "loss": 0.8116,
      "step": 7219
    },
    {
      "epoch": 0.713738477127252,
      "grad_norm": 5.2151751478334285,
      "learning_rate": 3.9977312358191885e-08,
      "loss": 0.6541,
      "step": 7220
    },
    {
      "epoch": 0.7138373328720066,
      "grad_norm": 4.385114318994977,
      "learning_rate": 3.9951703985437814e-08,
      "loss": 0.69,
      "step": 7221
    },
    {
      "epoch": 0.713936188616761,
      "grad_norm": 2.9428628789085147,
      "learning_rate": 3.992610176975072e-08,
      "loss": 0.6998,
      "step": 7222
    },
    {
      "epoch": 0.7140350443615154,
      "grad_norm": 3.8243811699042607,
      "learning_rate": 3.9900505713755786e-08,
      "loss": 0.7899,
      "step": 7223
    },
    {
      "epoch": 0.71413390010627,
      "grad_norm": 5.232589690573017,
      "learning_rate": 3.9874915820077415e-08,
      "loss": 0.6005,
      "step": 7224
    },
    {
      "epoch": 0.7142327558510244,
      "grad_norm": 4.500795495086504,
      "learning_rate": 3.984933209133956e-08,
      "loss": 0.7156,
      "step": 7225
    },
    {
      "epoch": 0.7143316115957788,
      "grad_norm": 5.752989649198968,
      "learning_rate": 3.9823754530165374e-08,
      "loss": 0.6963,
      "step": 7226
    },
    {
      "epoch": 0.7144304673405333,
      "grad_norm": 3.6456077733059917,
      "learning_rate": 3.979818313917753e-08,
      "loss": 0.7257,
      "step": 7227
    },
    {
      "epoch": 0.7145293230852878,
      "grad_norm": 3.275052817329671,
      "learning_rate": 3.9772617920997944e-08,
      "loss": 0.783,
      "step": 7228
    },
    {
      "epoch": 0.7146281788300423,
      "grad_norm": 6.041836661263248,
      "learning_rate": 3.974705887824801e-08,
      "loss": 0.8157,
      "step": 7229
    },
    {
      "epoch": 0.7147270345747967,
      "grad_norm": 2.757069223990188,
      "learning_rate": 3.972150601354834e-08,
      "loss": 0.7483,
      "step": 7230
    },
    {
      "epoch": 0.7148258903195512,
      "grad_norm": 4.801815665471596,
      "learning_rate": 3.9695959329519114e-08,
      "loss": 0.7638,
      "step": 7231
    },
    {
      "epoch": 0.7149247460643057,
      "grad_norm": 4.803006410234423,
      "learning_rate": 3.967041882877966e-08,
      "loss": 0.8053,
      "step": 7232
    },
    {
      "epoch": 0.7150236018090601,
      "grad_norm": 3.9527041072027846,
      "learning_rate": 3.964488451394884e-08,
      "loss": 0.7576,
      "step": 7233
    },
    {
      "epoch": 0.7151224575538146,
      "grad_norm": 5.069433693207704,
      "learning_rate": 3.961935638764485e-08,
      "loss": 0.7557,
      "step": 7234
    },
    {
      "epoch": 0.7152213132985691,
      "grad_norm": 4.603975333410448,
      "learning_rate": 3.959383445248514e-08,
      "loss": 0.6804,
      "step": 7235
    },
    {
      "epoch": 0.7153201690433235,
      "grad_norm": 5.1960754221607655,
      "learning_rate": 3.95683187110867e-08,
      "loss": 0.6072,
      "step": 7236
    },
    {
      "epoch": 0.715419024788078,
      "grad_norm": 6.190205584124171,
      "learning_rate": 3.954280916606571e-08,
      "loss": 0.6431,
      "step": 7237
    },
    {
      "epoch": 0.7155178805328325,
      "grad_norm": 4.574662210280579,
      "learning_rate": 3.9517305820037826e-08,
      "loss": 0.6776,
      "step": 7238
    },
    {
      "epoch": 0.7156167362775869,
      "grad_norm": 7.189538630389273,
      "learning_rate": 3.949180867561809e-08,
      "loss": 0.6903,
      "step": 7239
    },
    {
      "epoch": 0.7157155920223414,
      "grad_norm": 6.259109202330865,
      "learning_rate": 3.9466317735420775e-08,
      "loss": 0.6476,
      "step": 7240
    },
    {
      "epoch": 0.7158144477670959,
      "grad_norm": 3.4771908472219413,
      "learning_rate": 3.9440833002059645e-08,
      "loss": 0.668,
      "step": 7241
    },
    {
      "epoch": 0.7159133035118503,
      "grad_norm": 7.219237989135649,
      "learning_rate": 3.941535447814781e-08,
      "loss": 0.6529,
      "step": 7242
    },
    {
      "epoch": 0.7160121592566048,
      "grad_norm": 4.1864913838083115,
      "learning_rate": 3.938988216629766e-08,
      "loss": 0.7093,
      "step": 7243
    },
    {
      "epoch": 0.7161110150013593,
      "grad_norm": 4.968633008795248,
      "learning_rate": 3.9364416069121066e-08,
      "loss": 0.6016,
      "step": 7244
    },
    {
      "epoch": 0.7162098707461138,
      "grad_norm": 10.059162178520529,
      "learning_rate": 3.9338956189229124e-08,
      "loss": 0.5587,
      "step": 7245
    },
    {
      "epoch": 0.7163087264908682,
      "grad_norm": 8.016012225073071,
      "learning_rate": 3.931350252923244e-08,
      "loss": 0.7628,
      "step": 7246
    },
    {
      "epoch": 0.7164075822356226,
      "grad_norm": 6.153936816751572,
      "learning_rate": 3.928805509174087e-08,
      "loss": 0.6703,
      "step": 7247
    },
    {
      "epoch": 0.7165064379803772,
      "grad_norm": 9.743808958443637,
      "learning_rate": 3.92626138793637e-08,
      "loss": 0.6951,
      "step": 7248
    },
    {
      "epoch": 0.7166052937251316,
      "grad_norm": 3.3218129792986484,
      "learning_rate": 3.923717889470952e-08,
      "loss": 0.6526,
      "step": 7249
    },
    {
      "epoch": 0.716704149469886,
      "grad_norm": 3.828314231317914,
      "learning_rate": 3.9211750140386354e-08,
      "loss": 0.7478,
      "step": 7250
    },
    {
      "epoch": 0.7168030052146406,
      "grad_norm": 4.622625743351136,
      "learning_rate": 3.918632761900149e-08,
      "loss": 0.7768,
      "step": 7251
    },
    {
      "epoch": 0.716901860959395,
      "grad_norm": 4.032919517063897,
      "learning_rate": 3.916091133316166e-08,
      "loss": 0.7106,
      "step": 7252
    },
    {
      "epoch": 0.7170007167041494,
      "grad_norm": 3.614769658136436,
      "learning_rate": 3.913550128547297e-08,
      "loss": 0.6679,
      "step": 7253
    },
    {
      "epoch": 0.717099572448904,
      "grad_norm": 4.9103931899585636,
      "learning_rate": 3.911009747854077e-08,
      "loss": 0.8231,
      "step": 7254
    },
    {
      "epoch": 0.7171984281936584,
      "grad_norm": 4.415993719952776,
      "learning_rate": 3.908469991496989e-08,
      "loss": 0.7668,
      "step": 7255
    },
    {
      "epoch": 0.7172972839384129,
      "grad_norm": 4.1812657452159305,
      "learning_rate": 3.9059308597364505e-08,
      "loss": 0.8071,
      "step": 7256
    },
    {
      "epoch": 0.7173961396831673,
      "grad_norm": 7.876578653183208,
      "learning_rate": 3.903392352832805e-08,
      "loss": 0.6671,
      "step": 7257
    },
    {
      "epoch": 0.7174949954279218,
      "grad_norm": 3.4212707603754158,
      "learning_rate": 3.900854471046345e-08,
      "loss": 0.709,
      "step": 7258
    },
    {
      "epoch": 0.7175938511726763,
      "grad_norm": 10.48144188768974,
      "learning_rate": 3.898317214637288e-08,
      "loss": 0.6852,
      "step": 7259
    },
    {
      "epoch": 0.7176927069174307,
      "grad_norm": 4.191491797066778,
      "learning_rate": 3.895780583865794e-08,
      "loss": 0.62,
      "step": 7260
    },
    {
      "epoch": 0.7177915626621852,
      "grad_norm": 3.154361399329377,
      "learning_rate": 3.893244578991961e-08,
      "loss": 0.7197,
      "step": 7261
    },
    {
      "epoch": 0.7178904184069397,
      "grad_norm": 3.8207507775030733,
      "learning_rate": 3.8907092002758126e-08,
      "loss": 0.7889,
      "step": 7262
    },
    {
      "epoch": 0.7179892741516941,
      "grad_norm": 4.809156188299764,
      "learning_rate": 3.8881744479773194e-08,
      "loss": 0.7377,
      "step": 7263
    },
    {
      "epoch": 0.7180881298964487,
      "grad_norm": 7.4247787976058754,
      "learning_rate": 3.885640322356382e-08,
      "loss": 0.7052,
      "step": 7264
    },
    {
      "epoch": 0.7181869856412031,
      "grad_norm": 6.740803316011307,
      "learning_rate": 3.8831068236728325e-08,
      "loss": 0.6873,
      "step": 7265
    },
    {
      "epoch": 0.7182858413859575,
      "grad_norm": 3.517649558469712,
      "learning_rate": 3.880573952186448e-08,
      "loss": 0.7278,
      "step": 7266
    },
    {
      "epoch": 0.718384697130712,
      "grad_norm": 5.684897771899062,
      "learning_rate": 3.8780417081569396e-08,
      "loss": 0.6787,
      "step": 7267
    },
    {
      "epoch": 0.7184835528754665,
      "grad_norm": 4.044140096900547,
      "learning_rate": 3.8755100918439457e-08,
      "loss": 0.6703,
      "step": 7268
    },
    {
      "epoch": 0.7185824086202209,
      "grad_norm": 3.534216382410133,
      "learning_rate": 3.8729791035070525e-08,
      "loss": 0.5907,
      "step": 7269
    },
    {
      "epoch": 0.7186812643649754,
      "grad_norm": 4.564932700707634,
      "learning_rate": 3.87044874340577e-08,
      "loss": 0.6249,
      "step": 7270
    },
    {
      "epoch": 0.7187801201097299,
      "grad_norm": 4.1463939580754,
      "learning_rate": 3.8679190117995494e-08,
      "loss": 0.757,
      "step": 7271
    },
    {
      "epoch": 0.7188789758544843,
      "grad_norm": 8.963493258081316,
      "learning_rate": 3.865389908947785e-08,
      "loss": 0.6969,
      "step": 7272
    },
    {
      "epoch": 0.7189778315992388,
      "grad_norm": 7.96203486952326,
      "learning_rate": 3.86286143510979e-08,
      "loss": 0.7054,
      "step": 7273
    },
    {
      "epoch": 0.7190766873439933,
      "grad_norm": 4.992801524848335,
      "learning_rate": 3.860333590544824e-08,
      "loss": 0.702,
      "step": 7274
    },
    {
      "epoch": 0.7191755430887478,
      "grad_norm": 7.7715004477342555,
      "learning_rate": 3.857806375512087e-08,
      "loss": 0.7474,
      "step": 7275
    },
    {
      "epoch": 0.7192743988335022,
      "grad_norm": 2.758760578435779,
      "learning_rate": 3.855279790270699e-08,
      "loss": 0.696,
      "step": 7276
    },
    {
      "epoch": 0.7193732545782566,
      "grad_norm": 4.572787141803022,
      "learning_rate": 3.8527538350797263e-08,
      "loss": 0.7434,
      "step": 7277
    },
    {
      "epoch": 0.7194721103230112,
      "grad_norm": 4.158001076278279,
      "learning_rate": 3.850228510198173e-08,
      "loss": 0.7634,
      "step": 7278
    },
    {
      "epoch": 0.7195709660677656,
      "grad_norm": 3.4832393659395984,
      "learning_rate": 3.847703815884969e-08,
      "loss": 0.5925,
      "step": 7279
    },
    {
      "epoch": 0.71966982181252,
      "grad_norm": 6.049605336973558,
      "learning_rate": 3.8451797523989885e-08,
      "loss": 0.6403,
      "step": 7280
    },
    {
      "epoch": 0.7197686775572746,
      "grad_norm": 4.602179710988769,
      "learning_rate": 3.8426563199990336e-08,
      "loss": 0.7629,
      "step": 7281
    },
    {
      "epoch": 0.719867533302029,
      "grad_norm": 3.1811931328295073,
      "learning_rate": 3.840133518943842e-08,
      "loss": 0.5972,
      "step": 7282
    },
    {
      "epoch": 0.7199663890467835,
      "grad_norm": 3.1099179976933313,
      "learning_rate": 3.837611349492098e-08,
      "loss": 0.6779,
      "step": 7283
    },
    {
      "epoch": 0.720065244791538,
      "grad_norm": 15.871684100894305,
      "learning_rate": 3.8350898119024055e-08,
      "loss": 0.6108,
      "step": 7284
    },
    {
      "epoch": 0.7201641005362924,
      "grad_norm": 3.002651036758441,
      "learning_rate": 3.832568906433313e-08,
      "loss": 0.6653,
      "step": 7285
    },
    {
      "epoch": 0.7202629562810469,
      "grad_norm": 19.20528807672984,
      "learning_rate": 3.8300486333433065e-08,
      "loss": 0.7197,
      "step": 7286
    },
    {
      "epoch": 0.7203618120258013,
      "grad_norm": 6.091391702347325,
      "learning_rate": 3.827528992890797e-08,
      "loss": 0.7082,
      "step": 7287
    },
    {
      "epoch": 0.7204606677705558,
      "grad_norm": 8.07608116927132,
      "learning_rate": 3.825009985334138e-08,
      "loss": 0.6642,
      "step": 7288
    },
    {
      "epoch": 0.7205595235153103,
      "grad_norm": 3.2803409873650478,
      "learning_rate": 3.822491610931622e-08,
      "loss": 0.8123,
      "step": 7289
    },
    {
      "epoch": 0.7206583792600647,
      "grad_norm": 4.622622419804162,
      "learning_rate": 3.819973869941462e-08,
      "loss": 0.6422,
      "step": 7290
    },
    {
      "epoch": 0.7207572350048193,
      "grad_norm": 6.058913313544786,
      "learning_rate": 3.8174567626218234e-08,
      "loss": 0.6294,
      "step": 7291
    },
    {
      "epoch": 0.7208560907495737,
      "grad_norm": 6.164019313859351,
      "learning_rate": 3.8149402892307926e-08,
      "loss": 0.7133,
      "step": 7292
    },
    {
      "epoch": 0.7209549464943281,
      "grad_norm": 8.406567617267314,
      "learning_rate": 3.812424450026398e-08,
      "loss": 0.7779,
      "step": 7293
    },
    {
      "epoch": 0.7210538022390827,
      "grad_norm": 4.179730782252114,
      "learning_rate": 3.809909245266607e-08,
      "loss": 0.7808,
      "step": 7294
    },
    {
      "epoch": 0.7211526579838371,
      "grad_norm": 14.40445112674945,
      "learning_rate": 3.807394675209309e-08,
      "loss": 0.6324,
      "step": 7295
    },
    {
      "epoch": 0.7212515137285915,
      "grad_norm": 3.118803719832981,
      "learning_rate": 3.80488074011234e-08,
      "loss": 0.7631,
      "step": 7296
    },
    {
      "epoch": 0.721350369473346,
      "grad_norm": 3.068769986251991,
      "learning_rate": 3.802367440233468e-08,
      "loss": 0.7202,
      "step": 7297
    },
    {
      "epoch": 0.7214492252181005,
      "grad_norm": 3.607349233601858,
      "learning_rate": 3.7998547758303956e-08,
      "loss": 0.818,
      "step": 7298
    },
    {
      "epoch": 0.7215480809628549,
      "grad_norm": 4.375728144249166,
      "learning_rate": 3.7973427471607536e-08,
      "loss": 0.7065,
      "step": 7299
    },
    {
      "epoch": 0.7216469367076094,
      "grad_norm": 4.188191294007096,
      "learning_rate": 3.794831354482121e-08,
      "loss": 0.5747,
      "step": 7300
    },
    {
      "epoch": 0.7217457924523639,
      "grad_norm": 3.746743819838766,
      "learning_rate": 3.792320598051996e-08,
      "loss": 0.6863,
      "step": 7301
    },
    {
      "epoch": 0.7218446481971184,
      "grad_norm": 4.007815714558189,
      "learning_rate": 3.7898104781278276e-08,
      "loss": 0.6992,
      "step": 7302
    },
    {
      "epoch": 0.7219435039418728,
      "grad_norm": 3.7910130943118197,
      "learning_rate": 3.7873009949669846e-08,
      "loss": 0.7794,
      "step": 7303
    },
    {
      "epoch": 0.7220423596866273,
      "grad_norm": 4.762449606599511,
      "learning_rate": 3.784792148826781e-08,
      "loss": 0.6732,
      "step": 7304
    },
    {
      "epoch": 0.7221412154313818,
      "grad_norm": 3.4203276834292975,
      "learning_rate": 3.7822839399644655e-08,
      "loss": 0.6884,
      "step": 7305
    },
    {
      "epoch": 0.7222400711761362,
      "grad_norm": 3.9950114656588585,
      "learning_rate": 3.77977636863721e-08,
      "loss": 0.8591,
      "step": 7306
    },
    {
      "epoch": 0.7223389269208907,
      "grad_norm": 3.686630045279255,
      "learning_rate": 3.777269435102134e-08,
      "loss": 0.8105,
      "step": 7307
    },
    {
      "epoch": 0.7224377826656452,
      "grad_norm": 3.130114398638946,
      "learning_rate": 3.774763139616288e-08,
      "loss": 0.621,
      "step": 7308
    },
    {
      "epoch": 0.7225366384103996,
      "grad_norm": 5.383661646167647,
      "learning_rate": 3.772257482436651e-08,
      "loss": 0.7834,
      "step": 7309
    },
    {
      "epoch": 0.722635494155154,
      "grad_norm": 3.532396032832338,
      "learning_rate": 3.7697524638201426e-08,
      "loss": 0.7656,
      "step": 7310
    },
    {
      "epoch": 0.7227343498999086,
      "grad_norm": 3.1655790533912977,
      "learning_rate": 3.767248084023621e-08,
      "loss": 0.7025,
      "step": 7311
    },
    {
      "epoch": 0.722833205644663,
      "grad_norm": 7.743606568238336,
      "learning_rate": 3.764744343303864e-08,
      "loss": 0.6332,
      "step": 7312
    },
    {
      "epoch": 0.7229320613894175,
      "grad_norm": 5.695455196302267,
      "learning_rate": 3.762241241917602e-08,
      "loss": 0.7623,
      "step": 7313
    },
    {
      "epoch": 0.723030917134172,
      "grad_norm": 3.3090528178229404,
      "learning_rate": 3.759738780121484e-08,
      "loss": 0.6543,
      "step": 7314
    },
    {
      "epoch": 0.7231297728789264,
      "grad_norm": 3.282117749879745,
      "learning_rate": 3.7572369581721075e-08,
      "loss": 0.7321,
      "step": 7315
    },
    {
      "epoch": 0.7232286286236809,
      "grad_norm": 5.968436571668573,
      "learning_rate": 3.75473577632599e-08,
      "loss": 0.7813,
      "step": 7316
    },
    {
      "epoch": 0.7233274843684354,
      "grad_norm": 5.6898750542813605,
      "learning_rate": 3.7522352348395983e-08,
      "loss": 0.6998,
      "step": 7317
    },
    {
      "epoch": 0.7234263401131898,
      "grad_norm": 8.636242442216014,
      "learning_rate": 3.7497353339693185e-08,
      "loss": 0.6337,
      "step": 7318
    },
    {
      "epoch": 0.7235251958579443,
      "grad_norm": 5.018692664621684,
      "learning_rate": 3.747236073971485e-08,
      "loss": 0.7249,
      "step": 7319
    },
    {
      "epoch": 0.7236240516026987,
      "grad_norm": 6.990117444758741,
      "learning_rate": 3.744737455102355e-08,
      "loss": 0.6772,
      "step": 7320
    },
    {
      "epoch": 0.7237229073474533,
      "grad_norm": 3.029985714054099,
      "learning_rate": 3.7422394776181264e-08,
      "loss": 0.629,
      "step": 7321
    },
    {
      "epoch": 0.7238217630922077,
      "grad_norm": 4.519916165996968,
      "learning_rate": 3.739742141774935e-08,
      "loss": 0.7856,
      "step": 7322
    },
    {
      "epoch": 0.7239206188369621,
      "grad_norm": 12.572969578201336,
      "learning_rate": 3.737245447828838e-08,
      "loss": 0.6027,
      "step": 7323
    },
    {
      "epoch": 0.7240194745817167,
      "grad_norm": 3.050312762739581,
      "learning_rate": 3.7347493960358423e-08,
      "loss": 0.7497,
      "step": 7324
    },
    {
      "epoch": 0.7241183303264711,
      "grad_norm": 5.674091492504297,
      "learning_rate": 3.732253986651873e-08,
      "loss": 0.6783,
      "step": 7325
    },
    {
      "epoch": 0.7242171860712255,
      "grad_norm": 4.170333137661759,
      "learning_rate": 3.729759219932802e-08,
      "loss": 0.6514,
      "step": 7326
    },
    {
      "epoch": 0.7243160418159801,
      "grad_norm": 4.364500947319158,
      "learning_rate": 3.727265096134433e-08,
      "loss": 0.6797,
      "step": 7327
    },
    {
      "epoch": 0.7244148975607345,
      "grad_norm": 5.255068144104749,
      "learning_rate": 3.7247716155124965e-08,
      "loss": 0.748,
      "step": 7328
    },
    {
      "epoch": 0.724513753305489,
      "grad_norm": 4.051218545819326,
      "learning_rate": 3.722278778322664e-08,
      "loss": 0.7683,
      "step": 7329
    },
    {
      "epoch": 0.7246126090502434,
      "grad_norm": 10.835198070045639,
      "learning_rate": 3.7197865848205446e-08,
      "loss": 0.8365,
      "step": 7330
    },
    {
      "epoch": 0.7247114647949979,
      "grad_norm": 4.4655424090683224,
      "learning_rate": 3.717295035261666e-08,
      "loss": 0.7806,
      "step": 7331
    },
    {
      "epoch": 0.7248103205397524,
      "grad_norm": 4.532474818486525,
      "learning_rate": 3.71480412990151e-08,
      "loss": 0.7371,
      "step": 7332
    },
    {
      "epoch": 0.7249091762845068,
      "grad_norm": 3.5373599779345843,
      "learning_rate": 3.7123138689954736e-08,
      "loss": 0.5911,
      "step": 7333
    },
    {
      "epoch": 0.7250080320292613,
      "grad_norm": 2.959868965400439,
      "learning_rate": 3.709824252798903e-08,
      "loss": 0.576,
      "step": 7334
    },
    {
      "epoch": 0.7251068877740158,
      "grad_norm": 8.083983988183228,
      "learning_rate": 3.707335281567069e-08,
      "loss": 0.7413,
      "step": 7335
    },
    {
      "epoch": 0.7252057435187702,
      "grad_norm": 4.220084962807107,
      "learning_rate": 3.7048469555551754e-08,
      "loss": 0.7713,
      "step": 7336
    },
    {
      "epoch": 0.7253045992635248,
      "grad_norm": 3.7259260405457337,
      "learning_rate": 3.7023592750183654e-08,
      "loss": 0.6555,
      "step": 7337
    },
    {
      "epoch": 0.7254034550082792,
      "grad_norm": 6.19648662530911,
      "learning_rate": 3.699872240211719e-08,
      "loss": 0.6701,
      "step": 7338
    },
    {
      "epoch": 0.7255023107530336,
      "grad_norm": 3.8911637241846724,
      "learning_rate": 3.697385851390238e-08,
      "loss": 0.7364,
      "step": 7339
    },
    {
      "epoch": 0.725601166497788,
      "grad_norm": 3.6743817295025574,
      "learning_rate": 3.694900108808867e-08,
      "loss": 0.7429,
      "step": 7340
    },
    {
      "epoch": 0.7257000222425426,
      "grad_norm": 6.531392739973485,
      "learning_rate": 3.692415012722487e-08,
      "loss": 0.7291,
      "step": 7341
    },
    {
      "epoch": 0.725798877987297,
      "grad_norm": 5.183875581091765,
      "learning_rate": 3.6899305633859014e-08,
      "loss": 0.7612,
      "step": 7342
    },
    {
      "epoch": 0.7258977337320515,
      "grad_norm": 3.6868071948126078,
      "learning_rate": 3.687446761053855e-08,
      "loss": 0.6757,
      "step": 7343
    },
    {
      "epoch": 0.725996589476806,
      "grad_norm": 5.74727703121191,
      "learning_rate": 3.6849636059810306e-08,
      "loss": 0.7899,
      "step": 7344
    },
    {
      "epoch": 0.7260954452215604,
      "grad_norm": 6.185501433262192,
      "learning_rate": 3.682481098422031e-08,
      "loss": 0.7072,
      "step": 7345
    },
    {
      "epoch": 0.7261943009663149,
      "grad_norm": 3.2652266885916497,
      "learning_rate": 3.6799992386314094e-08,
      "loss": 0.7171,
      "step": 7346
    },
    {
      "epoch": 0.7262931567110694,
      "grad_norm": 4.067004127937666,
      "learning_rate": 3.6775180268636355e-08,
      "loss": 0.7374,
      "step": 7347
    },
    {
      "epoch": 0.7263920124558239,
      "grad_norm": 10.161158229326492,
      "learning_rate": 3.675037463373124e-08,
      "loss": 0.7398,
      "step": 7348
    },
    {
      "epoch": 0.7264908682005783,
      "grad_norm": 5.42528044095739,
      "learning_rate": 3.6725575484142256e-08,
      "loss": 0.7313,
      "step": 7349
    },
    {
      "epoch": 0.7265897239453327,
      "grad_norm": 5.748117432053352,
      "learning_rate": 3.6700782822412104e-08,
      "loss": 0.6771,
      "step": 7350
    },
    {
      "epoch": 0.7266885796900873,
      "grad_norm": 6.430398489218253,
      "learning_rate": 3.6675996651082975e-08,
      "loss": 0.7826,
      "step": 7351
    },
    {
      "epoch": 0.7267874354348417,
      "grad_norm": 3.197896845969707,
      "learning_rate": 3.6651216972696285e-08,
      "loss": 0.6759,
      "step": 7352
    },
    {
      "epoch": 0.7268862911795961,
      "grad_norm": 5.170561135574947,
      "learning_rate": 3.6626443789792815e-08,
      "loss": 0.7392,
      "step": 7353
    },
    {
      "epoch": 0.7269851469243507,
      "grad_norm": 3.9338373131172957,
      "learning_rate": 3.660167710491271e-08,
      "loss": 0.7352,
      "step": 7354
    },
    {
      "epoch": 0.7270840026691051,
      "grad_norm": 5.266194256581534,
      "learning_rate": 3.6576916920595446e-08,
      "loss": 0.8256,
      "step": 7355
    },
    {
      "epoch": 0.7271828584138595,
      "grad_norm": 3.4700814323499625,
      "learning_rate": 3.6552163239379774e-08,
      "loss": 0.5898,
      "step": 7356
    },
    {
      "epoch": 0.7272817141586141,
      "grad_norm": 3.428373436054677,
      "learning_rate": 3.6527416063803875e-08,
      "loss": 0.7144,
      "step": 7357
    },
    {
      "epoch": 0.7273805699033685,
      "grad_norm": 3.444372794521405,
      "learning_rate": 3.650267539640514e-08,
      "loss": 0.7893,
      "step": 7358
    },
    {
      "epoch": 0.727479425648123,
      "grad_norm": 2.834926209772839,
      "learning_rate": 3.647794123972038e-08,
      "loss": 0.7409,
      "step": 7359
    },
    {
      "epoch": 0.7275782813928774,
      "grad_norm": 5.69860784710094,
      "learning_rate": 3.645321359628577e-08,
      "loss": 0.719,
      "step": 7360
    },
    {
      "epoch": 0.7276771371376319,
      "grad_norm": 3.9545425434718733,
      "learning_rate": 3.6428492468636686e-08,
      "loss": 0.6888,
      "step": 7361
    },
    {
      "epoch": 0.7277759928823864,
      "grad_norm": 5.823664045510305,
      "learning_rate": 3.640377785930796e-08,
      "loss": 0.6834,
      "step": 7362
    },
    {
      "epoch": 0.7278748486271408,
      "grad_norm": 3.107355404784063,
      "learning_rate": 3.6379069770833734e-08,
      "loss": 0.6758,
      "step": 7363
    },
    {
      "epoch": 0.7279737043718953,
      "grad_norm": 12.19219174770848,
      "learning_rate": 3.63543682057474e-08,
      "loss": 0.7382,
      "step": 7364
    },
    {
      "epoch": 0.7280725601166498,
      "grad_norm": 3.7154345634182415,
      "learning_rate": 3.6329673166581754e-08,
      "loss": 0.6769,
      "step": 7365
    },
    {
      "epoch": 0.7281714158614042,
      "grad_norm": 13.191955904067434,
      "learning_rate": 3.630498465586896e-08,
      "loss": 0.7036,
      "step": 7366
    },
    {
      "epoch": 0.7282702716061588,
      "grad_norm": 2.5622003678921765,
      "learning_rate": 3.6280302676140406e-08,
      "loss": 0.5699,
      "step": 7367
    },
    {
      "epoch": 0.7283691273509132,
      "grad_norm": 2.8140690293596475,
      "learning_rate": 3.62556272299269e-08,
      "loss": 0.7317,
      "step": 7368
    },
    {
      "epoch": 0.7284679830956676,
      "grad_norm": 5.962688836187505,
      "learning_rate": 3.623095831975852e-08,
      "loss": 0.7149,
      "step": 7369
    },
    {
      "epoch": 0.7285668388404221,
      "grad_norm": 3.521524220954424,
      "learning_rate": 3.6206295948164676e-08,
      "loss": 0.7064,
      "step": 7370
    },
    {
      "epoch": 0.7286656945851766,
      "grad_norm": 3.885849768376188,
      "learning_rate": 3.618164011767421e-08,
      "loss": 0.7111,
      "step": 7371
    },
    {
      "epoch": 0.728764550329931,
      "grad_norm": 3.608310023303222,
      "learning_rate": 3.6156990830815106e-08,
      "loss": 0.7299,
      "step": 7372
    },
    {
      "epoch": 0.7288634060746855,
      "grad_norm": 3.879972226997316,
      "learning_rate": 3.613234809011485e-08,
      "loss": 0.7228,
      "step": 7373
    },
    {
      "epoch": 0.72896226181944,
      "grad_norm": 4.750971679774283,
      "learning_rate": 3.610771189810021e-08,
      "loss": 0.7265,
      "step": 7374
    },
    {
      "epoch": 0.7290611175641944,
      "grad_norm": 77.13859105497136,
      "learning_rate": 3.6083082257297215e-08,
      "loss": 0.8042,
      "step": 7375
    },
    {
      "epoch": 0.7291599733089489,
      "grad_norm": 4.988003077417005,
      "learning_rate": 3.605845917023129e-08,
      "loss": 0.6589,
      "step": 7376
    },
    {
      "epoch": 0.7292588290537034,
      "grad_norm": 7.849302359150328,
      "learning_rate": 3.60338426394272e-08,
      "loss": 0.7589,
      "step": 7377
    },
    {
      "epoch": 0.7293576847984579,
      "grad_norm": 3.61602888774024,
      "learning_rate": 3.600923266740896e-08,
      "loss": 0.6609,
      "step": 7378
    },
    {
      "epoch": 0.7294565405432123,
      "grad_norm": 3.8838186209892296,
      "learning_rate": 3.5984629256700004e-08,
      "loss": 0.7452,
      "step": 7379
    },
    {
      "epoch": 0.7295553962879668,
      "grad_norm": 3.3779892351436933,
      "learning_rate": 3.596003240982299e-08,
      "loss": 0.6061,
      "step": 7380
    },
    {
      "epoch": 0.7296542520327213,
      "grad_norm": 4.66135140867551,
      "learning_rate": 3.593544212930001e-08,
      "loss": 0.745,
      "step": 7381
    },
    {
      "epoch": 0.7297531077774757,
      "grad_norm": 7.2549223743340185,
      "learning_rate": 3.5910858417652436e-08,
      "loss": 0.6487,
      "step": 7382
    },
    {
      "epoch": 0.7298519635222301,
      "grad_norm": 10.862855724991713,
      "learning_rate": 3.588628127740092e-08,
      "loss": 0.6936,
      "step": 7383
    },
    {
      "epoch": 0.7299508192669847,
      "grad_norm": 4.45758219101402,
      "learning_rate": 3.586171071106553e-08,
      "loss": 0.636,
      "step": 7384
    },
    {
      "epoch": 0.7300496750117391,
      "grad_norm": 3.144296742316466,
      "learning_rate": 3.583714672116561e-08,
      "loss": 0.578,
      "step": 7385
    },
    {
      "epoch": 0.7301485307564936,
      "grad_norm": 16.03708348386177,
      "learning_rate": 3.5812589310219844e-08,
      "loss": 0.7896,
      "step": 7386
    },
    {
      "epoch": 0.7302473865012481,
      "grad_norm": 6.05445394716261,
      "learning_rate": 3.578803848074618e-08,
      "loss": 0.7093,
      "step": 7387
    },
    {
      "epoch": 0.7303462422460025,
      "grad_norm": 3.238085886840995,
      "learning_rate": 3.5763494235262004e-08,
      "loss": 0.7693,
      "step": 7388
    },
    {
      "epoch": 0.730445097990757,
      "grad_norm": 3.3500755218596883,
      "learning_rate": 3.5738956576283894e-08,
      "loss": 0.718,
      "step": 7389
    },
    {
      "epoch": 0.7305439537355115,
      "grad_norm": 4.309545326748535,
      "learning_rate": 3.571442550632792e-08,
      "loss": 0.6553,
      "step": 7390
    },
    {
      "epoch": 0.7306428094802659,
      "grad_norm": 3.837182821225649,
      "learning_rate": 3.5689901027909294e-08,
      "loss": 0.7566,
      "step": 7391
    },
    {
      "epoch": 0.7307416652250204,
      "grad_norm": 10.343614384707289,
      "learning_rate": 3.566538314354267e-08,
      "loss": 0.6249,
      "step": 7392
    },
    {
      "epoch": 0.7308405209697748,
      "grad_norm": 3.169657567828645,
      "learning_rate": 3.564087185574205e-08,
      "loss": 0.749,
      "step": 7393
    },
    {
      "epoch": 0.7309393767145294,
      "grad_norm": 12.508530284009275,
      "learning_rate": 3.561636716702061e-08,
      "loss": 0.6439,
      "step": 7394
    },
    {
      "epoch": 0.7310382324592838,
      "grad_norm": 3.9293743800319487,
      "learning_rate": 3.5591869079891e-08,
      "loss": 0.7854,
      "step": 7395
    },
    {
      "epoch": 0.7311370882040382,
      "grad_norm": 3.048484397765153,
      "learning_rate": 3.5567377596865156e-08,
      "loss": 0.6835,
      "step": 7396
    },
    {
      "epoch": 0.7312359439487928,
      "grad_norm": 5.1085240418533,
      "learning_rate": 3.5542892720454264e-08,
      "loss": 0.7112,
      "step": 7397
    },
    {
      "epoch": 0.7313347996935472,
      "grad_norm": 3.53750653493388,
      "learning_rate": 3.551841445316891e-08,
      "loss": 0.7539,
      "step": 7398
    },
    {
      "epoch": 0.7314336554383016,
      "grad_norm": 5.569889898845101,
      "learning_rate": 3.5493942797519016e-08,
      "loss": 0.6989,
      "step": 7399
    },
    {
      "epoch": 0.7315325111830562,
      "grad_norm": 3.5302841587831693,
      "learning_rate": 3.546947775601373e-08,
      "loss": 0.7302,
      "step": 7400
    },
    {
      "epoch": 0.7316313669278106,
      "grad_norm": 4.282807525550946,
      "learning_rate": 3.544501933116164e-08,
      "loss": 0.6692,
      "step": 7401
    },
    {
      "epoch": 0.731730222672565,
      "grad_norm": 5.222107790246693,
      "learning_rate": 3.542056752547052e-08,
      "loss": 0.6724,
      "step": 7402
    },
    {
      "epoch": 0.7318290784173195,
      "grad_norm": 7.618750843075519,
      "learning_rate": 3.539612234144763e-08,
      "loss": 0.7275,
      "step": 7403
    },
    {
      "epoch": 0.731927934162074,
      "grad_norm": 4.33156286328052,
      "learning_rate": 3.537168378159938e-08,
      "loss": 0.7415,
      "step": 7404
    },
    {
      "epoch": 0.7320267899068285,
      "grad_norm": 10.578897364390052,
      "learning_rate": 3.534725184843167e-08,
      "loss": 0.6608,
      "step": 7405
    },
    {
      "epoch": 0.7321256456515829,
      "grad_norm": 4.071904674769014,
      "learning_rate": 3.532282654444955e-08,
      "loss": 0.693,
      "step": 7406
    },
    {
      "epoch": 0.7322245013963374,
      "grad_norm": 3.7362948624904964,
      "learning_rate": 3.529840787215753e-08,
      "loss": 0.683,
      "step": 7407
    },
    {
      "epoch": 0.7323233571410919,
      "grad_norm": 5.20605071246516,
      "learning_rate": 3.527399583405936e-08,
      "loss": 0.6591,
      "step": 7408
    },
    {
      "epoch": 0.7324222128858463,
      "grad_norm": 4.077748064387112,
      "learning_rate": 3.524959043265812e-08,
      "loss": 0.6578,
      "step": 7409
    },
    {
      "epoch": 0.7325210686306008,
      "grad_norm": 5.207410011729742,
      "learning_rate": 3.522519167045629e-08,
      "loss": 0.7288,
      "step": 7410
    },
    {
      "epoch": 0.7326199243753553,
      "grad_norm": 3.130241474950181,
      "learning_rate": 3.520079954995552e-08,
      "loss": 0.6408,
      "step": 7411
    },
    {
      "epoch": 0.7327187801201097,
      "grad_norm": 4.657718302513017,
      "learning_rate": 3.5176414073656936e-08,
      "loss": 0.7478,
      "step": 7412
    },
    {
      "epoch": 0.7328176358648641,
      "grad_norm": 5.2845170041554566,
      "learning_rate": 3.515203524406083e-08,
      "loss": 0.7429,
      "step": 7413
    },
    {
      "epoch": 0.7329164916096187,
      "grad_norm": 3.660652662868905,
      "learning_rate": 3.5127663063666944e-08,
      "loss": 0.6832,
      "step": 7414
    },
    {
      "epoch": 0.7330153473543731,
      "grad_norm": 3.1643111940130297,
      "learning_rate": 3.510329753497431e-08,
      "loss": 0.7629,
      "step": 7415
    },
    {
      "epoch": 0.7331142030991276,
      "grad_norm": 3.688840719368677,
      "learning_rate": 3.507893866048119e-08,
      "loss": 0.6935,
      "step": 7416
    },
    {
      "epoch": 0.7332130588438821,
      "grad_norm": 4.5977069067342375,
      "learning_rate": 3.505458644268525e-08,
      "loss": 0.7176,
      "step": 7417
    },
    {
      "epoch": 0.7333119145886365,
      "grad_norm": 4.171768732989724,
      "learning_rate": 3.503024088408349e-08,
      "loss": 0.5957,
      "step": 7418
    },
    {
      "epoch": 0.733410770333391,
      "grad_norm": 3.4564311213355543,
      "learning_rate": 3.500590198717214e-08,
      "loss": 0.6309,
      "step": 7419
    },
    {
      "epoch": 0.7335096260781455,
      "grad_norm": 9.668442513248033,
      "learning_rate": 3.4981569754446827e-08,
      "loss": 0.7048,
      "step": 7420
    },
    {
      "epoch": 0.7336084818229,
      "grad_norm": 6.564653996910882,
      "learning_rate": 3.4957244188402427e-08,
      "loss": 0.7521,
      "step": 7421
    },
    {
      "epoch": 0.7337073375676544,
      "grad_norm": 3.4707177873445847,
      "learning_rate": 3.49329252915332e-08,
      "loss": 0.6263,
      "step": 7422
    },
    {
      "epoch": 0.7338061933124088,
      "grad_norm": 6.089389960152548,
      "learning_rate": 3.490861306633269e-08,
      "loss": 0.597,
      "step": 7423
    },
    {
      "epoch": 0.7339050490571634,
      "grad_norm": 9.698114871013251,
      "learning_rate": 3.488430751529371e-08,
      "loss": 0.6695,
      "step": 7424
    },
    {
      "epoch": 0.7340039048019178,
      "grad_norm": 3.2133813969935896,
      "learning_rate": 3.486000864090848e-08,
      "loss": 0.8143,
      "step": 7425
    },
    {
      "epoch": 0.7341027605466722,
      "grad_norm": 3.746452946329229,
      "learning_rate": 3.4835716445668493e-08,
      "loss": 0.7585,
      "step": 7426
    },
    {
      "epoch": 0.7342016162914268,
      "grad_norm": 15.15664327618828,
      "learning_rate": 3.481143093206453e-08,
      "loss": 0.7051,
      "step": 7427
    },
    {
      "epoch": 0.7343004720361812,
      "grad_norm": 3.9686172649529046,
      "learning_rate": 3.4787152102586716e-08,
      "loss": 0.6238,
      "step": 7428
    },
    {
      "epoch": 0.7343993277809356,
      "grad_norm": 5.288245161563879,
      "learning_rate": 3.4762879959724536e-08,
      "loss": 0.6759,
      "step": 7429
    },
    {
      "epoch": 0.7344981835256902,
      "grad_norm": 6.26369992211701,
      "learning_rate": 3.473861450596667e-08,
      "loss": 0.7862,
      "step": 7430
    },
    {
      "epoch": 0.7345970392704446,
      "grad_norm": 5.926038086944689,
      "learning_rate": 3.4714355743801204e-08,
      "loss": 0.6769,
      "step": 7431
    },
    {
      "epoch": 0.734695895015199,
      "grad_norm": 3.0493275912835607,
      "learning_rate": 3.4690103675715555e-08,
      "loss": 0.7187,
      "step": 7432
    },
    {
      "epoch": 0.7347947507599535,
      "grad_norm": 5.966557963654479,
      "learning_rate": 3.466585830419636e-08,
      "loss": 0.844,
      "step": 7433
    },
    {
      "epoch": 0.734893606504708,
      "grad_norm": 2.527838323379806,
      "learning_rate": 3.464161963172968e-08,
      "loss": 0.7475,
      "step": 7434
    },
    {
      "epoch": 0.7349924622494625,
      "grad_norm": 2.9538887941305774,
      "learning_rate": 3.4617387660800775e-08,
      "loss": 0.5617,
      "step": 7435
    },
    {
      "epoch": 0.7350913179942169,
      "grad_norm": 3.65972430758589,
      "learning_rate": 3.4593162393894305e-08,
      "loss": 0.6494,
      "step": 7436
    },
    {
      "epoch": 0.7351901737389714,
      "grad_norm": 3.9812011265417993,
      "learning_rate": 3.456894383349423e-08,
      "loss": 0.7168,
      "step": 7437
    },
    {
      "epoch": 0.7352890294837259,
      "grad_norm": 6.408497090075123,
      "learning_rate": 3.4544731982083774e-08,
      "loss": 0.6426,
      "step": 7438
    },
    {
      "epoch": 0.7353878852284803,
      "grad_norm": 3.348056741846032,
      "learning_rate": 3.452052684214555e-08,
      "loss": 0.775,
      "step": 7439
    },
    {
      "epoch": 0.7354867409732349,
      "grad_norm": 3.351691308474385,
      "learning_rate": 3.44963284161614e-08,
      "loss": 0.8251,
      "step": 7440
    },
    {
      "epoch": 0.7355855967179893,
      "grad_norm": 4.718423311176304,
      "learning_rate": 3.4472136706612497e-08,
      "loss": 0.7707,
      "step": 7441
    },
    {
      "epoch": 0.7356844524627437,
      "grad_norm": 9.656976219531279,
      "learning_rate": 3.4447951715979374e-08,
      "loss": 0.7533,
      "step": 7442
    },
    {
      "epoch": 0.7357833082074982,
      "grad_norm": 3.649301639696371,
      "learning_rate": 3.442377344674187e-08,
      "loss": 0.6623,
      "step": 7443
    },
    {
      "epoch": 0.7358821639522527,
      "grad_norm": 5.216764482873255,
      "learning_rate": 3.4399601901379037e-08,
      "loss": 0.691,
      "step": 7444
    },
    {
      "epoch": 0.7359810196970071,
      "grad_norm": 3.7328444233246794,
      "learning_rate": 3.437543708236941e-08,
      "loss": 0.8226,
      "step": 7445
    },
    {
      "epoch": 0.7360798754417616,
      "grad_norm": 10.616242155327802,
      "learning_rate": 3.4351278992190635e-08,
      "loss": 0.7679,
      "step": 7446
    },
    {
      "epoch": 0.7361787311865161,
      "grad_norm": 4.326893341252115,
      "learning_rate": 3.432712763331981e-08,
      "loss": 0.6688,
      "step": 7447
    },
    {
      "epoch": 0.7362775869312705,
      "grad_norm": 3.402816403531725,
      "learning_rate": 3.430298300823334e-08,
      "loss": 0.8084,
      "step": 7448
    },
    {
      "epoch": 0.736376442676025,
      "grad_norm": 14.04830186478802,
      "learning_rate": 3.427884511940683e-08,
      "loss": 0.6927,
      "step": 7449
    },
    {
      "epoch": 0.7364752984207795,
      "grad_norm": 3.820032192165113,
      "learning_rate": 3.42547139693153e-08,
      "loss": 0.707,
      "step": 7450
    },
    {
      "epoch": 0.736574154165534,
      "grad_norm": 14.497299173102338,
      "learning_rate": 3.423058956043308e-08,
      "loss": 0.7596,
      "step": 7451
    },
    {
      "epoch": 0.7366730099102884,
      "grad_norm": 4.79596473455302,
      "learning_rate": 3.420647189523368e-08,
      "loss": 0.643,
      "step": 7452
    },
    {
      "epoch": 0.7367718656550428,
      "grad_norm": 15.291532850769846,
      "learning_rate": 3.418236097619008e-08,
      "loss": 0.7619,
      "step": 7453
    },
    {
      "epoch": 0.7368707213997974,
      "grad_norm": 4.535128543046381,
      "learning_rate": 3.415825680577451e-08,
      "loss": 0.6996,
      "step": 7454
    },
    {
      "epoch": 0.7369695771445518,
      "grad_norm": 4.8829535995249795,
      "learning_rate": 3.413415938645845e-08,
      "loss": 0.6912,
      "step": 7455
    },
    {
      "epoch": 0.7370684328893062,
      "grad_norm": 4.8063321433293815,
      "learning_rate": 3.411006872071278e-08,
      "loss": 0.681,
      "step": 7456
    },
    {
      "epoch": 0.7371672886340608,
      "grad_norm": 4.283904851799877,
      "learning_rate": 3.408598481100762e-08,
      "loss": 0.675,
      "step": 7457
    },
    {
      "epoch": 0.7372661443788152,
      "grad_norm": 10.732677706087673,
      "learning_rate": 3.406190765981237e-08,
      "loss": 0.7358,
      "step": 7458
    },
    {
      "epoch": 0.7373650001235696,
      "grad_norm": 3.467159315191169,
      "learning_rate": 3.4037837269595884e-08,
      "loss": 0.6777,
      "step": 7459
    },
    {
      "epoch": 0.7374638558683242,
      "grad_norm": 4.612059522722652,
      "learning_rate": 3.4013773642826135e-08,
      "loss": 0.6609,
      "step": 7460
    },
    {
      "epoch": 0.7375627116130786,
      "grad_norm": 6.4345756656274915,
      "learning_rate": 3.398971678197054e-08,
      "loss": 0.5928,
      "step": 7461
    },
    {
      "epoch": 0.7376615673578331,
      "grad_norm": 7.0267489669080945,
      "learning_rate": 3.39656666894958e-08,
      "loss": 0.7737,
      "step": 7462
    },
    {
      "epoch": 0.7377604231025876,
      "grad_norm": 3.711157073551067,
      "learning_rate": 3.3941623367867834e-08,
      "loss": 0.7443,
      "step": 7463
    },
    {
      "epoch": 0.737859278847342,
      "grad_norm": 4.039456920421458,
      "learning_rate": 3.3917586819551956e-08,
      "loss": 0.7938,
      "step": 7464
    },
    {
      "epoch": 0.7379581345920965,
      "grad_norm": 3.9185570716532485,
      "learning_rate": 3.38935570470128e-08,
      "loss": 0.7398,
      "step": 7465
    },
    {
      "epoch": 0.7380569903368509,
      "grad_norm": 3.5457364592839062,
      "learning_rate": 3.38695340527142e-08,
      "loss": 0.6958,
      "step": 7466
    },
    {
      "epoch": 0.7381558460816054,
      "grad_norm": 3.985931689054105,
      "learning_rate": 3.384551783911942e-08,
      "loss": 0.7548,
      "step": 7467
    },
    {
      "epoch": 0.7382547018263599,
      "grad_norm": 4.195000795328834,
      "learning_rate": 3.38215084086909e-08,
      "loss": 0.8091,
      "step": 7468
    },
    {
      "epoch": 0.7383535575711143,
      "grad_norm": 3.6047676468813283,
      "learning_rate": 3.3797505763890496e-08,
      "loss": 0.7372,
      "step": 7469
    },
    {
      "epoch": 0.7384524133158689,
      "grad_norm": 3.884104293650104,
      "learning_rate": 3.377350990717937e-08,
      "loss": 0.7076,
      "step": 7470
    },
    {
      "epoch": 0.7385512690606233,
      "grad_norm": 3.4601903397084315,
      "learning_rate": 3.374952084101784e-08,
      "loss": 0.6649,
      "step": 7471
    },
    {
      "epoch": 0.7386501248053777,
      "grad_norm": 3.246253025659364,
      "learning_rate": 3.372553856786568e-08,
      "loss": 0.7519,
      "step": 7472
    },
    {
      "epoch": 0.7387489805501323,
      "grad_norm": 6.48670276579539,
      "learning_rate": 3.370156309018197e-08,
      "loss": 0.7127,
      "step": 7473
    },
    {
      "epoch": 0.7388478362948867,
      "grad_norm": 3.692430004284158,
      "learning_rate": 3.367759441042499e-08,
      "loss": 0.702,
      "step": 7474
    },
    {
      "epoch": 0.7389466920396411,
      "grad_norm": 2.938270482676433,
      "learning_rate": 3.3653632531052355e-08,
      "loss": 0.6557,
      "step": 7475
    },
    {
      "epoch": 0.7390455477843956,
      "grad_norm": 6.671302130464745,
      "learning_rate": 3.3629677454521046e-08,
      "loss": 0.687,
      "step": 7476
    },
    {
      "epoch": 0.7391444035291501,
      "grad_norm": 3.4246277615526513,
      "learning_rate": 3.3605729183287255e-08,
      "loss": 0.7015,
      "step": 7477
    },
    {
      "epoch": 0.7392432592739046,
      "grad_norm": 4.843294902284229,
      "learning_rate": 3.3581787719806584e-08,
      "loss": 0.7273,
      "step": 7478
    },
    {
      "epoch": 0.739342115018659,
      "grad_norm": 4.4962345853495345,
      "learning_rate": 3.3557853066533814e-08,
      "loss": 0.6026,
      "step": 7479
    },
    {
      "epoch": 0.7394409707634135,
      "grad_norm": 5.194759417673226,
      "learning_rate": 3.353392522592312e-08,
      "loss": 0.7779,
      "step": 7480
    },
    {
      "epoch": 0.739539826508168,
      "grad_norm": 4.150761894124218,
      "learning_rate": 3.351000420042799e-08,
      "loss": 0.7449,
      "step": 7481
    },
    {
      "epoch": 0.7396386822529224,
      "grad_norm": 4.247644289883783,
      "learning_rate": 3.348608999250109e-08,
      "loss": 0.7355,
      "step": 7482
    },
    {
      "epoch": 0.7397375379976769,
      "grad_norm": 6.114185266738917,
      "learning_rate": 3.346218260459451e-08,
      "loss": 0.6938,
      "step": 7483
    },
    {
      "epoch": 0.7398363937424314,
      "grad_norm": 6.024452805623244,
      "learning_rate": 3.343828203915964e-08,
      "loss": 0.8236,
      "step": 7484
    },
    {
      "epoch": 0.7399352494871858,
      "grad_norm": 4.277147396842752,
      "learning_rate": 3.341438829864706e-08,
      "loss": 0.676,
      "step": 7485
    },
    {
      "epoch": 0.7400341052319402,
      "grad_norm": 4.358411680979199,
      "learning_rate": 3.339050138550675e-08,
      "loss": 0.7116,
      "step": 7486
    },
    {
      "epoch": 0.7401329609766948,
      "grad_norm": 3.131069958100941,
      "learning_rate": 3.3366621302188e-08,
      "loss": 0.7787,
      "step": 7487
    },
    {
      "epoch": 0.7402318167214492,
      "grad_norm": 4.219166293295592,
      "learning_rate": 3.334274805113929e-08,
      "loss": 0.6615,
      "step": 7488
    },
    {
      "epoch": 0.7403306724662037,
      "grad_norm": 11.76015559344868,
      "learning_rate": 3.331888163480854e-08,
      "loss": 0.6778,
      "step": 7489
    },
    {
      "epoch": 0.7404295282109582,
      "grad_norm": 2.940301211066793,
      "learning_rate": 3.329502205564283e-08,
      "loss": 0.6576,
      "step": 7490
    },
    {
      "epoch": 0.7405283839557126,
      "grad_norm": 5.9557824125201755,
      "learning_rate": 3.3271169316088666e-08,
      "loss": 0.6629,
      "step": 7491
    },
    {
      "epoch": 0.7406272397004671,
      "grad_norm": 7.479078663073239,
      "learning_rate": 3.3247323418591745e-08,
      "loss": 0.6543,
      "step": 7492
    },
    {
      "epoch": 0.7407260954452216,
      "grad_norm": 3.871716133540677,
      "learning_rate": 3.3223484365597156e-08,
      "loss": 0.6554,
      "step": 7493
    },
    {
      "epoch": 0.740824951189976,
      "grad_norm": 3.3047005620098164,
      "learning_rate": 3.31996521595492e-08,
      "loss": 0.7767,
      "step": 7494
    },
    {
      "epoch": 0.7409238069347305,
      "grad_norm": 4.863619856391789,
      "learning_rate": 3.317582680289157e-08,
      "loss": 0.7587,
      "step": 7495
    },
    {
      "epoch": 0.7410226626794849,
      "grad_norm": 4.656195908166571,
      "learning_rate": 3.315200829806715e-08,
      "loss": 0.7242,
      "step": 7496
    },
    {
      "epoch": 0.7411215184242395,
      "grad_norm": 5.823183182204489,
      "learning_rate": 3.3128196647518195e-08,
      "loss": 0.7826,
      "step": 7497
    },
    {
      "epoch": 0.7412203741689939,
      "grad_norm": 3.3017556920923252,
      "learning_rate": 3.310439185368628e-08,
      "loss": 0.6362,
      "step": 7498
    },
    {
      "epoch": 0.7413192299137483,
      "grad_norm": 4.996396511974805,
      "learning_rate": 3.308059391901217e-08,
      "loss": 0.7026,
      "step": 7499
    },
    {
      "epoch": 0.7414180856585029,
      "grad_norm": 3.4512218373346766,
      "learning_rate": 3.305680284593605e-08,
      "loss": 0.6185,
      "step": 7500
    },
    {
      "epoch": 0.7415169414032573,
      "grad_norm": 3.491391912611703,
      "learning_rate": 3.303301863689728e-08,
      "loss": 0.7414,
      "step": 7501
    },
    {
      "epoch": 0.7416157971480117,
      "grad_norm": 2.6936380913724474,
      "learning_rate": 3.300924129433462e-08,
      "loss": 0.6916,
      "step": 7502
    },
    {
      "epoch": 0.7417146528927663,
      "grad_norm": 2.9639088650564815,
      "learning_rate": 3.2985470820686125e-08,
      "loss": 0.6918,
      "step": 7503
    },
    {
      "epoch": 0.7418135086375207,
      "grad_norm": 3.7812181474771873,
      "learning_rate": 3.2961707218389026e-08,
      "loss": 0.6846,
      "step": 7504
    },
    {
      "epoch": 0.7419123643822751,
      "grad_norm": 3.124146906163775,
      "learning_rate": 3.2937950489879954e-08,
      "loss": 0.7388,
      "step": 7505
    },
    {
      "epoch": 0.7420112201270296,
      "grad_norm": 3.8928821379137095,
      "learning_rate": 3.291420063759487e-08,
      "loss": 0.6494,
      "step": 7506
    },
    {
      "epoch": 0.7421100758717841,
      "grad_norm": 3.244671882801789,
      "learning_rate": 3.289045766396889e-08,
      "loss": 0.7179,
      "step": 7507
    },
    {
      "epoch": 0.7422089316165386,
      "grad_norm": 5.647353298111852,
      "learning_rate": 3.2866721571436576e-08,
      "loss": 0.6881,
      "step": 7508
    },
    {
      "epoch": 0.742307787361293,
      "grad_norm": 5.0777622618652165,
      "learning_rate": 3.284299236243164e-08,
      "loss": 0.6931,
      "step": 7509
    },
    {
      "epoch": 0.7424066431060475,
      "grad_norm": 3.669598681610694,
      "learning_rate": 3.2819270039387235e-08,
      "loss": 0.672,
      "step": 7510
    },
    {
      "epoch": 0.742505498850802,
      "grad_norm": 3.8498272856680824,
      "learning_rate": 3.2795554604735707e-08,
      "loss": 0.7126,
      "step": 7511
    },
    {
      "epoch": 0.7426043545955564,
      "grad_norm": 4.974522607844476,
      "learning_rate": 3.277184606090869e-08,
      "loss": 0.7628,
      "step": 7512
    },
    {
      "epoch": 0.742703210340311,
      "grad_norm": 3.4197673591410176,
      "learning_rate": 3.274814441033717e-08,
      "loss": 0.723,
      "step": 7513
    },
    {
      "epoch": 0.7428020660850654,
      "grad_norm": 6.6841552348849556,
      "learning_rate": 3.2724449655451436e-08,
      "loss": 0.6452,
      "step": 7514
    },
    {
      "epoch": 0.7429009218298198,
      "grad_norm": 37.50461298870969,
      "learning_rate": 3.2700761798680974e-08,
      "loss": 0.677,
      "step": 7515
    },
    {
      "epoch": 0.7429997775745742,
      "grad_norm": 3.336381493861423,
      "learning_rate": 3.267708084245466e-08,
      "loss": 0.5924,
      "step": 7516
    },
    {
      "epoch": 0.7430986333193288,
      "grad_norm": 8.175023120990426,
      "learning_rate": 3.265340678920065e-08,
      "loss": 0.6593,
      "step": 7517
    },
    {
      "epoch": 0.7431974890640832,
      "grad_norm": 3.1124807729926323,
      "learning_rate": 3.262973964134631e-08,
      "loss": 0.6896,
      "step": 7518
    },
    {
      "epoch": 0.7432963448088377,
      "grad_norm": 4.715274375769304,
      "learning_rate": 3.2606079401318397e-08,
      "loss": 0.7812,
      "step": 7519
    },
    {
      "epoch": 0.7433952005535922,
      "grad_norm": 5.563966947058339,
      "learning_rate": 3.2582426071542934e-08,
      "loss": 0.6724,
      "step": 7520
    },
    {
      "epoch": 0.7434940562983466,
      "grad_norm": 3.0916158573643746,
      "learning_rate": 3.255877965444517e-08,
      "loss": 0.6848,
      "step": 7521
    },
    {
      "epoch": 0.7435929120431011,
      "grad_norm": 3.4498607062025597,
      "learning_rate": 3.2535140152449756e-08,
      "loss": 0.7962,
      "step": 7522
    },
    {
      "epoch": 0.7436917677878556,
      "grad_norm": 4.251054653977877,
      "learning_rate": 3.251150756798051e-08,
      "loss": 0.7039,
      "step": 7523
    },
    {
      "epoch": 0.74379062353261,
      "grad_norm": 3.820494400976635,
      "learning_rate": 3.248788190346063e-08,
      "loss": 0.7117,
      "step": 7524
    },
    {
      "epoch": 0.7438894792773645,
      "grad_norm": 10.158271157058255,
      "learning_rate": 3.2464263161312634e-08,
      "loss": 0.6874,
      "step": 7525
    },
    {
      "epoch": 0.7439883350221189,
      "grad_norm": 3.4339986996974763,
      "learning_rate": 3.244065134395818e-08,
      "loss": 0.6403,
      "step": 7526
    },
    {
      "epoch": 0.7440871907668735,
      "grad_norm": 6.250392913773462,
      "learning_rate": 3.241704645381841e-08,
      "loss": 0.6694,
      "step": 7527
    },
    {
      "epoch": 0.7441860465116279,
      "grad_norm": 40.069697539147015,
      "learning_rate": 3.239344849331359e-08,
      "loss": 0.7268,
      "step": 7528
    },
    {
      "epoch": 0.7442849022563823,
      "grad_norm": 4.5001561972213056,
      "learning_rate": 3.236985746486334e-08,
      "loss": 0.7044,
      "step": 7529
    },
    {
      "epoch": 0.7443837580011369,
      "grad_norm": 4.765233090866179,
      "learning_rate": 3.234627337088659e-08,
      "loss": 0.7201,
      "step": 7530
    },
    {
      "epoch": 0.7444826137458913,
      "grad_norm": 3.6030402177594008,
      "learning_rate": 3.232269621380157e-08,
      "loss": 0.7252,
      "step": 7531
    },
    {
      "epoch": 0.7445814694906457,
      "grad_norm": 2.885407503996708,
      "learning_rate": 3.2299125996025725e-08,
      "loss": 0.6703,
      "step": 7532
    },
    {
      "epoch": 0.7446803252354003,
      "grad_norm": 3.93632964298045,
      "learning_rate": 3.227556271997587e-08,
      "loss": 0.6251,
      "step": 7533
    },
    {
      "epoch": 0.7447791809801547,
      "grad_norm": 15.506032478083085,
      "learning_rate": 3.225200638806802e-08,
      "loss": 0.7991,
      "step": 7534
    },
    {
      "epoch": 0.7448780367249092,
      "grad_norm": 3.1995950453074706,
      "learning_rate": 3.2228457002717565e-08,
      "loss": 0.69,
      "step": 7535
    },
    {
      "epoch": 0.7449768924696637,
      "grad_norm": 6.094519671929715,
      "learning_rate": 3.220491456633918e-08,
      "loss": 0.7143,
      "step": 7536
    },
    {
      "epoch": 0.7450757482144181,
      "grad_norm": 4.411457028449421,
      "learning_rate": 3.2181379081346715e-08,
      "loss": 0.6915,
      "step": 7537
    },
    {
      "epoch": 0.7451746039591726,
      "grad_norm": 4.198057732484937,
      "learning_rate": 3.2157850550153446e-08,
      "loss": 0.7126,
      "step": 7538
    },
    {
      "epoch": 0.745273459703927,
      "grad_norm": 2.642248402645812,
      "learning_rate": 3.2134328975171874e-08,
      "loss": 0.6747,
      "step": 7539
    },
    {
      "epoch": 0.7453723154486815,
      "grad_norm": 3.466041020297305,
      "learning_rate": 3.2110814358813753e-08,
      "loss": 0.6811,
      "step": 7540
    },
    {
      "epoch": 0.745471171193436,
      "grad_norm": 4.225773658899789,
      "learning_rate": 3.208730670349019e-08,
      "loss": 0.6641,
      "step": 7541
    },
    {
      "epoch": 0.7455700269381904,
      "grad_norm": 5.087014704836251,
      "learning_rate": 3.206380601161157e-08,
      "loss": 0.7423,
      "step": 7542
    },
    {
      "epoch": 0.745668882682945,
      "grad_norm": 3.194945851842056,
      "learning_rate": 3.2040312285587476e-08,
      "loss": 0.6523,
      "step": 7543
    },
    {
      "epoch": 0.7457677384276994,
      "grad_norm": 6.646152331621697,
      "learning_rate": 3.2016825527826915e-08,
      "loss": 0.7354,
      "step": 7544
    },
    {
      "epoch": 0.7458665941724538,
      "grad_norm": 4.9553211840018205,
      "learning_rate": 3.199334574073809e-08,
      "loss": 0.7793,
      "step": 7545
    },
    {
      "epoch": 0.7459654499172084,
      "grad_norm": 3.649095304650544,
      "learning_rate": 3.196987292672844e-08,
      "loss": 0.7187,
      "step": 7546
    },
    {
      "epoch": 0.7460643056619628,
      "grad_norm": 4.42804683747053,
      "learning_rate": 3.1946407088204844e-08,
      "loss": 0.7069,
      "step": 7547
    },
    {
      "epoch": 0.7461631614067172,
      "grad_norm": 3.744736765242146,
      "learning_rate": 3.192294822757331e-08,
      "loss": 0.6693,
      "step": 7548
    },
    {
      "epoch": 0.7462620171514717,
      "grad_norm": 11.85350000871913,
      "learning_rate": 3.189949634723922e-08,
      "loss": 0.7628,
      "step": 7549
    },
    {
      "epoch": 0.7463608728962262,
      "grad_norm": 2.8669724743242027,
      "learning_rate": 3.1876051449607266e-08,
      "loss": 0.6829,
      "step": 7550
    },
    {
      "epoch": 0.7464597286409806,
      "grad_norm": 2.9606963134347577,
      "learning_rate": 3.185261353708131e-08,
      "loss": 0.6824,
      "step": 7551
    },
    {
      "epoch": 0.7465585843857351,
      "grad_norm": 4.811613672546457,
      "learning_rate": 3.182918261206458e-08,
      "loss": 0.7793,
      "step": 7552
    },
    {
      "epoch": 0.7466574401304896,
      "grad_norm": 10.425931725719128,
      "learning_rate": 3.180575867695961e-08,
      "loss": 0.6692,
      "step": 7553
    },
    {
      "epoch": 0.7467562958752441,
      "grad_norm": 4.944996106518451,
      "learning_rate": 3.1782341734168136e-08,
      "loss": 0.7202,
      "step": 7554
    },
    {
      "epoch": 0.7468551516199985,
      "grad_norm": 3.7307109629518727,
      "learning_rate": 3.175893178609126e-08,
      "loss": 0.7451,
      "step": 7555
    },
    {
      "epoch": 0.746954007364753,
      "grad_norm": 5.704717656699929,
      "learning_rate": 3.173552883512928e-08,
      "loss": 0.6037,
      "step": 7556
    },
    {
      "epoch": 0.7470528631095075,
      "grad_norm": 3.5100716183645404,
      "learning_rate": 3.171213288368183e-08,
      "loss": 0.6567,
      "step": 7557
    },
    {
      "epoch": 0.7471517188542619,
      "grad_norm": 4.211413326169119,
      "learning_rate": 3.168874393414788e-08,
      "loss": 0.7525,
      "step": 7558
    },
    {
      "epoch": 0.7472505745990163,
      "grad_norm": 3.2815990606841803,
      "learning_rate": 3.166536198892554e-08,
      "loss": 0.7684,
      "step": 7559
    },
    {
      "epoch": 0.7473494303437709,
      "grad_norm": 4.549860088121821,
      "learning_rate": 3.164198705041232e-08,
      "loss": 0.7612,
      "step": 7560
    },
    {
      "epoch": 0.7474482860885253,
      "grad_norm": 5.013148713135209,
      "learning_rate": 3.161861912100501e-08,
      "loss": 0.6231,
      "step": 7561
    },
    {
      "epoch": 0.7475471418332797,
      "grad_norm": 10.557824861092357,
      "learning_rate": 3.159525820309962e-08,
      "loss": 0.7189,
      "step": 7562
    },
    {
      "epoch": 0.7476459975780343,
      "grad_norm": 4.547605941498288,
      "learning_rate": 3.157190429909141e-08,
      "loss": 0.651,
      "step": 7563
    },
    {
      "epoch": 0.7477448533227887,
      "grad_norm": 5.051960179199116,
      "learning_rate": 3.154855741137508e-08,
      "loss": 0.579,
      "step": 7564
    },
    {
      "epoch": 0.7478437090675432,
      "grad_norm": 8.816172932591353,
      "learning_rate": 3.1525217542344415e-08,
      "loss": 0.7034,
      "step": 7565
    },
    {
      "epoch": 0.7479425648122977,
      "grad_norm": 3.5677867270234063,
      "learning_rate": 3.1501884694392656e-08,
      "loss": 0.7279,
      "step": 7566
    },
    {
      "epoch": 0.7480414205570521,
      "grad_norm": 4.26761128960305,
      "learning_rate": 3.1478558869912166e-08,
      "loss": 0.7695,
      "step": 7567
    },
    {
      "epoch": 0.7481402763018066,
      "grad_norm": 6.388390131092705,
      "learning_rate": 3.145524007129471e-08,
      "loss": 0.7196,
      "step": 7568
    },
    {
      "epoch": 0.748239132046561,
      "grad_norm": 3.8785166414070575,
      "learning_rate": 3.14319283009313e-08,
      "loss": 0.7601,
      "step": 7569
    },
    {
      "epoch": 0.7483379877913156,
      "grad_norm": 4.298214663975996,
      "learning_rate": 3.140862356121216e-08,
      "loss": 0.7499,
      "step": 7570
    },
    {
      "epoch": 0.74843684353607,
      "grad_norm": 6.564641383143187,
      "learning_rate": 3.1385325854526886e-08,
      "loss": 0.7464,
      "step": 7571
    },
    {
      "epoch": 0.7485356992808244,
      "grad_norm": 4.861326696109598,
      "learning_rate": 3.1362035183264334e-08,
      "loss": 0.7472,
      "step": 7572
    },
    {
      "epoch": 0.748634555025579,
      "grad_norm": 5.177711266123443,
      "learning_rate": 3.133875154981257e-08,
      "loss": 0.6286,
      "step": 7573
    },
    {
      "epoch": 0.7487334107703334,
      "grad_norm": 4.927023145411169,
      "learning_rate": 3.1315474956559e-08,
      "loss": 0.7271,
      "step": 7574
    },
    {
      "epoch": 0.7488322665150878,
      "grad_norm": 3.102062848522213,
      "learning_rate": 3.1292205405890336e-08,
      "loss": 0.6931,
      "step": 7575
    },
    {
      "epoch": 0.7489311222598424,
      "grad_norm": 5.25024901896606,
      "learning_rate": 3.126894290019247e-08,
      "loss": 0.794,
      "step": 7576
    },
    {
      "epoch": 0.7490299780045968,
      "grad_norm": 4.330657898888768,
      "learning_rate": 3.124568744185069e-08,
      "loss": 0.7135,
      "step": 7577
    },
    {
      "epoch": 0.7491288337493512,
      "grad_norm": 8.417444250963834,
      "learning_rate": 3.1222439033249436e-08,
      "loss": 0.7461,
      "step": 7578
    },
    {
      "epoch": 0.7492276894941057,
      "grad_norm": 4.110628120461992,
      "learning_rate": 3.1199197676772536e-08,
      "loss": 0.7517,
      "step": 7579
    },
    {
      "epoch": 0.7493265452388602,
      "grad_norm": 3.5259042857512846,
      "learning_rate": 3.117596337480302e-08,
      "loss": 0.8562,
      "step": 7580
    },
    {
      "epoch": 0.7494254009836147,
      "grad_norm": 3.8598780458136197,
      "learning_rate": 3.1152736129723244e-08,
      "loss": 0.6969,
      "step": 7581
    },
    {
      "epoch": 0.7495242567283691,
      "grad_norm": 9.601406180114708,
      "learning_rate": 3.1129515943914785e-08,
      "loss": 0.7234,
      "step": 7582
    },
    {
      "epoch": 0.7496231124731236,
      "grad_norm": 3.701599179813755,
      "learning_rate": 3.110630281975859e-08,
      "loss": 0.7948,
      "step": 7583
    },
    {
      "epoch": 0.7497219682178781,
      "grad_norm": 4.7212248590396255,
      "learning_rate": 3.108309675963476e-08,
      "loss": 0.712,
      "step": 7584
    },
    {
      "epoch": 0.7498208239626325,
      "grad_norm": 3.160289454822699,
      "learning_rate": 3.105989776592276e-08,
      "loss": 0.6953,
      "step": 7585
    },
    {
      "epoch": 0.749919679707387,
      "grad_norm": 6.448333417557734,
      "learning_rate": 3.103670584100133e-08,
      "loss": 0.678,
      "step": 7586
    },
    {
      "epoch": 0.7500185354521415,
      "grad_norm": 19.64414399660531,
      "learning_rate": 3.101352098724841e-08,
      "loss": 0.7792,
      "step": 7587
    },
    {
      "epoch": 0.7501173911968959,
      "grad_norm": 3.8198719607161125,
      "learning_rate": 3.099034320704132e-08,
      "loss": 0.7195,
      "step": 7588
    },
    {
      "epoch": 0.7502162469416503,
      "grad_norm": 4.54128178353477,
      "learning_rate": 3.0967172502756535e-08,
      "loss": 0.583,
      "step": 7589
    },
    {
      "epoch": 0.7503151026864049,
      "grad_norm": 3.8614357225200626,
      "learning_rate": 3.094400887676991e-08,
      "loss": 0.8195,
      "step": 7590
    },
    {
      "epoch": 0.7504139584311593,
      "grad_norm": 5.997000531300967,
      "learning_rate": 3.092085233145655e-08,
      "loss": 0.728,
      "step": 7591
    },
    {
      "epoch": 0.7505128141759138,
      "grad_norm": 3.5223468355689542,
      "learning_rate": 3.089770286919077e-08,
      "loss": 0.7709,
      "step": 7592
    },
    {
      "epoch": 0.7506116699206683,
      "grad_norm": 3.7654595803767754,
      "learning_rate": 3.0874560492346234e-08,
      "loss": 0.6903,
      "step": 7593
    },
    {
      "epoch": 0.7507105256654227,
      "grad_norm": 5.49042060768461,
      "learning_rate": 3.0851425203295864e-08,
      "loss": 0.748,
      "step": 7594
    },
    {
      "epoch": 0.7508093814101772,
      "grad_norm": 6.357842269584003,
      "learning_rate": 3.0828297004411806e-08,
      "loss": 0.6789,
      "step": 7595
    },
    {
      "epoch": 0.7509082371549317,
      "grad_norm": 2.762026000679715,
      "learning_rate": 3.0805175898065574e-08,
      "loss": 0.5489,
      "step": 7596
    },
    {
      "epoch": 0.7510070928996861,
      "grad_norm": 2.7650324838509266,
      "learning_rate": 3.078206188662782e-08,
      "loss": 0.7846,
      "step": 7597
    },
    {
      "epoch": 0.7511059486444406,
      "grad_norm": 6.731361765768484,
      "learning_rate": 3.075895497246861e-08,
      "loss": 0.676,
      "step": 7598
    },
    {
      "epoch": 0.751204804389195,
      "grad_norm": 3.252938789585685,
      "learning_rate": 3.0735855157957194e-08,
      "loss": 0.6772,
      "step": 7599
    },
    {
      "epoch": 0.7513036601339496,
      "grad_norm": 3.8466541096028264,
      "learning_rate": 3.071276244546208e-08,
      "loss": 0.7268,
      "step": 7600
    },
    {
      "epoch": 0.751402515878704,
      "grad_norm": 8.876294300129304,
      "learning_rate": 3.068967683735112e-08,
      "loss": 0.768,
      "step": 7601
    },
    {
      "epoch": 0.7515013716234584,
      "grad_norm": 5.133172219064097,
      "learning_rate": 3.066659833599143e-08,
      "loss": 0.7081,
      "step": 7602
    },
    {
      "epoch": 0.751600227368213,
      "grad_norm": 3.6446865683356964,
      "learning_rate": 3.064352694374932e-08,
      "loss": 0.714,
      "step": 7603
    },
    {
      "epoch": 0.7516990831129674,
      "grad_norm": 6.7785281047169175,
      "learning_rate": 3.062046266299043e-08,
      "loss": 0.7539,
      "step": 7604
    },
    {
      "epoch": 0.7517979388577218,
      "grad_norm": 5.226082016453731,
      "learning_rate": 3.059740549607971e-08,
      "loss": 0.7072,
      "step": 7605
    },
    {
      "epoch": 0.7518967946024764,
      "grad_norm": 4.428488219731953,
      "learning_rate": 3.057435544538126e-08,
      "loss": 0.6951,
      "step": 7606
    },
    {
      "epoch": 0.7519956503472308,
      "grad_norm": 4.0093504163253355,
      "learning_rate": 3.055131251325856e-08,
      "loss": 0.8544,
      "step": 7607
    },
    {
      "epoch": 0.7520945060919852,
      "grad_norm": 3.144548631767095,
      "learning_rate": 3.0528276702074365e-08,
      "loss": 0.7219,
      "step": 7608
    },
    {
      "epoch": 0.7521933618367397,
      "grad_norm": 21.654758165657338,
      "learning_rate": 3.0505248014190575e-08,
      "loss": 0.6313,
      "step": 7609
    },
    {
      "epoch": 0.7522922175814942,
      "grad_norm": 4.854278205297361,
      "learning_rate": 3.048222645196851e-08,
      "loss": 0.7703,
      "step": 7610
    },
    {
      "epoch": 0.7523910733262487,
      "grad_norm": 4.4911386655841845,
      "learning_rate": 3.045921201776863e-08,
      "loss": 0.7155,
      "step": 7611
    },
    {
      "epoch": 0.7524899290710031,
      "grad_norm": 3.9972892338599864,
      "learning_rate": 3.0436204713950764e-08,
      "loss": 0.5707,
      "step": 7612
    },
    {
      "epoch": 0.7525887848157576,
      "grad_norm": 3.1527256262220815,
      "learning_rate": 3.041320454287399e-08,
      "loss": 0.7749,
      "step": 7613
    },
    {
      "epoch": 0.7526876405605121,
      "grad_norm": 4.281559184225426,
      "learning_rate": 3.039021150689659e-08,
      "loss": 0.5895,
      "step": 7614
    },
    {
      "epoch": 0.7527864963052665,
      "grad_norm": 4.513850337351555,
      "learning_rate": 3.036722560837621e-08,
      "loss": 0.7441,
      "step": 7615
    },
    {
      "epoch": 0.752885352050021,
      "grad_norm": 3.5447537714549884,
      "learning_rate": 3.03442468496697e-08,
      "loss": 0.6761,
      "step": 7616
    },
    {
      "epoch": 0.7529842077947755,
      "grad_norm": 5.981874218367335,
      "learning_rate": 3.0321275233133135e-08,
      "loss": 0.7139,
      "step": 7617
    },
    {
      "epoch": 0.7530830635395299,
      "grad_norm": 4.053591214227134,
      "learning_rate": 3.029831076112196e-08,
      "loss": 0.6669,
      "step": 7618
    },
    {
      "epoch": 0.7531819192842845,
      "grad_norm": 4.291712301310414,
      "learning_rate": 3.027535343599088e-08,
      "loss": 0.6544,
      "step": 7619
    },
    {
      "epoch": 0.7532807750290389,
      "grad_norm": 8.08695496618772,
      "learning_rate": 3.025240326009377e-08,
      "loss": 0.6541,
      "step": 7620
    },
    {
      "epoch": 0.7533796307737933,
      "grad_norm": 4.578253429428077,
      "learning_rate": 3.0229460235783876e-08,
      "loss": 0.7135,
      "step": 7621
    },
    {
      "epoch": 0.7534784865185478,
      "grad_norm": 3.300532165733539,
      "learning_rate": 3.020652436541362e-08,
      "loss": 0.6129,
      "step": 7622
    },
    {
      "epoch": 0.7535773422633023,
      "grad_norm": 4.632658334756227,
      "learning_rate": 3.018359565133477e-08,
      "loss": 0.7979,
      "step": 7623
    },
    {
      "epoch": 0.7536761980080567,
      "grad_norm": 3.822684754234831,
      "learning_rate": 3.0160674095898354e-08,
      "loss": 0.7521,
      "step": 7624
    },
    {
      "epoch": 0.7537750537528112,
      "grad_norm": 5.332499679257379,
      "learning_rate": 3.013775970145457e-08,
      "loss": 0.8523,
      "step": 7625
    },
    {
      "epoch": 0.7538739094975657,
      "grad_norm": 7.9448143301130285,
      "learning_rate": 3.011485247035298e-08,
      "loss": 0.7366,
      "step": 7626
    },
    {
      "epoch": 0.7539727652423202,
      "grad_norm": 2.9517272271484947,
      "learning_rate": 3.009195240494243e-08,
      "loss": 0.6801,
      "step": 7627
    },
    {
      "epoch": 0.7540716209870746,
      "grad_norm": 13.574500234877924,
      "learning_rate": 3.006905950757091e-08,
      "loss": 0.7818,
      "step": 7628
    },
    {
      "epoch": 0.7541704767318291,
      "grad_norm": 4.749344133539396,
      "learning_rate": 3.0046173780585795e-08,
      "loss": 0.7214,
      "step": 7629
    },
    {
      "epoch": 0.7542693324765836,
      "grad_norm": 4.819825525615457,
      "learning_rate": 3.002329522633369e-08,
      "loss": 0.805,
      "step": 7630
    },
    {
      "epoch": 0.754368188221338,
      "grad_norm": 3.172837068096982,
      "learning_rate": 3.0000423847160396e-08,
      "loss": 0.8271,
      "step": 7631
    },
    {
      "epoch": 0.7544670439660924,
      "grad_norm": 3.9827485512455967,
      "learning_rate": 2.997755964541111e-08,
      "loss": 0.7309,
      "step": 7632
    },
    {
      "epoch": 0.754565899710847,
      "grad_norm": 11.141780044594084,
      "learning_rate": 2.9954702623430176e-08,
      "loss": 0.661,
      "step": 7633
    },
    {
      "epoch": 0.7546647554556014,
      "grad_norm": 14.177583971520994,
      "learning_rate": 2.993185278356121e-08,
      "loss": 0.6946,
      "step": 7634
    },
    {
      "epoch": 0.7547636112003558,
      "grad_norm": 6.964024633322112,
      "learning_rate": 2.9909010128147194e-08,
      "loss": 0.685,
      "step": 7635
    },
    {
      "epoch": 0.7548624669451104,
      "grad_norm": 3.5164858403286448,
      "learning_rate": 2.988617465953025e-08,
      "loss": 0.6585,
      "step": 7636
    },
    {
      "epoch": 0.7549613226898648,
      "grad_norm": 4.6922376332229305,
      "learning_rate": 2.986334638005184e-08,
      "loss": 0.5989,
      "step": 7637
    },
    {
      "epoch": 0.7550601784346193,
      "grad_norm": 3.1940524416039793,
      "learning_rate": 2.98405252920527e-08,
      "loss": 0.6475,
      "step": 7638
    },
    {
      "epoch": 0.7551590341793738,
      "grad_norm": 5.863109394148172,
      "learning_rate": 2.9817711397872736e-08,
      "loss": 0.6708,
      "step": 7639
    },
    {
      "epoch": 0.7552578899241282,
      "grad_norm": 6.466243476237676,
      "learning_rate": 2.9794904699851208e-08,
      "loss": 0.7649,
      "step": 7640
    },
    {
      "epoch": 0.7553567456688827,
      "grad_norm": 17.631525824941374,
      "learning_rate": 2.9772105200326634e-08,
      "loss": 0.6843,
      "step": 7641
    },
    {
      "epoch": 0.7554556014136371,
      "grad_norm": 3.0845093219429778,
      "learning_rate": 2.9749312901636702e-08,
      "loss": 0.7494,
      "step": 7642
    },
    {
      "epoch": 0.7555544571583916,
      "grad_norm": 13.56661224639578,
      "learning_rate": 2.9726527806118507e-08,
      "loss": 0.7162,
      "step": 7643
    },
    {
      "epoch": 0.7556533129031461,
      "grad_norm": 4.098600995888041,
      "learning_rate": 2.9703749916108235e-08,
      "loss": 0.6318,
      "step": 7644
    },
    {
      "epoch": 0.7557521686479005,
      "grad_norm": 3.161626137802655,
      "learning_rate": 2.9680979233941473e-08,
      "loss": 0.6732,
      "step": 7645
    },
    {
      "epoch": 0.7558510243926551,
      "grad_norm": 3.401617237544605,
      "learning_rate": 2.9658215761953043e-08,
      "loss": 0.7721,
      "step": 7646
    },
    {
      "epoch": 0.7559498801374095,
      "grad_norm": 5.497506938042624,
      "learning_rate": 2.9635459502476932e-08,
      "loss": 0.6904,
      "step": 7647
    },
    {
      "epoch": 0.7560487358821639,
      "grad_norm": 8.079116392493239,
      "learning_rate": 2.9612710457846536e-08,
      "loss": 0.7648,
      "step": 7648
    },
    {
      "epoch": 0.7561475916269185,
      "grad_norm": 3.2601357850847954,
      "learning_rate": 2.958996863039437e-08,
      "loss": 0.7132,
      "step": 7649
    },
    {
      "epoch": 0.7562464473716729,
      "grad_norm": 3.938109496194402,
      "learning_rate": 2.9567234022452327e-08,
      "loss": 0.7422,
      "step": 7650
    },
    {
      "epoch": 0.7563453031164273,
      "grad_norm": 3.0883033466711893,
      "learning_rate": 2.954450663635145e-08,
      "loss": 0.6978,
      "step": 7651
    },
    {
      "epoch": 0.7564441588611818,
      "grad_norm": 6.157603532709804,
      "learning_rate": 2.952178647442215e-08,
      "loss": 0.752,
      "step": 7652
    },
    {
      "epoch": 0.7565430146059363,
      "grad_norm": 6.392721749118564,
      "learning_rate": 2.949907353899399e-08,
      "loss": 0.786,
      "step": 7653
    },
    {
      "epoch": 0.7566418703506907,
      "grad_norm": 4.533814007253662,
      "learning_rate": 2.947636783239592e-08,
      "loss": 0.5945,
      "step": 7654
    },
    {
      "epoch": 0.7567407260954452,
      "grad_norm": 3.8136378819051746,
      "learning_rate": 2.945366935695599e-08,
      "loss": 0.7114,
      "step": 7655
    },
    {
      "epoch": 0.7568395818401997,
      "grad_norm": 4.931013072461832,
      "learning_rate": 2.9430978115001637e-08,
      "loss": 0.7068,
      "step": 7656
    },
    {
      "epoch": 0.7569384375849542,
      "grad_norm": 3.398335400953236,
      "learning_rate": 2.9408294108859545e-08,
      "loss": 0.6603,
      "step": 7657
    },
    {
      "epoch": 0.7570372933297086,
      "grad_norm": 4.651788099368189,
      "learning_rate": 2.9385617340855564e-08,
      "loss": 0.6241,
      "step": 7658
    },
    {
      "epoch": 0.7571361490744631,
      "grad_norm": 8.723901212436036,
      "learning_rate": 2.9362947813314888e-08,
      "loss": 0.7441,
      "step": 7659
    },
    {
      "epoch": 0.7572350048192176,
      "grad_norm": 4.495480156561079,
      "learning_rate": 2.934028552856197e-08,
      "loss": 0.7671,
      "step": 7660
    },
    {
      "epoch": 0.757333860563972,
      "grad_norm": 3.4667212737553226,
      "learning_rate": 2.931763048892043e-08,
      "loss": 0.7624,
      "step": 7661
    },
    {
      "epoch": 0.7574327163087264,
      "grad_norm": 14.171305990381178,
      "learning_rate": 2.9294982696713255e-08,
      "loss": 0.7268,
      "step": 7662
    },
    {
      "epoch": 0.757531572053481,
      "grad_norm": 3.9784887737511805,
      "learning_rate": 2.9272342154262665e-08,
      "loss": 0.7863,
      "step": 7663
    },
    {
      "epoch": 0.7576304277982354,
      "grad_norm": 5.041139432814118,
      "learning_rate": 2.9249708863890045e-08,
      "loss": 0.7224,
      "step": 7664
    },
    {
      "epoch": 0.7577292835429899,
      "grad_norm": 3.7086496032860947,
      "learning_rate": 2.9227082827916172e-08,
      "loss": 0.7458,
      "step": 7665
    },
    {
      "epoch": 0.7578281392877444,
      "grad_norm": 3.4328869468934315,
      "learning_rate": 2.9204464048660948e-08,
      "loss": 0.7127,
      "step": 7666
    },
    {
      "epoch": 0.7579269950324988,
      "grad_norm": 4.413863503587194,
      "learning_rate": 2.918185252844366e-08,
      "loss": 0.7592,
      "step": 7667
    },
    {
      "epoch": 0.7580258507772533,
      "grad_norm": 4.260350919773249,
      "learning_rate": 2.9159248269582727e-08,
      "loss": 0.6613,
      "step": 7668
    },
    {
      "epoch": 0.7581247065220078,
      "grad_norm": 3.695428028317312,
      "learning_rate": 2.913665127439594e-08,
      "loss": 0.7319,
      "step": 7669
    },
    {
      "epoch": 0.7582235622667622,
      "grad_norm": 3.2041656363461577,
      "learning_rate": 2.911406154520022e-08,
      "loss": 0.6522,
      "step": 7670
    },
    {
      "epoch": 0.7583224180115167,
      "grad_norm": 4.058804338982078,
      "learning_rate": 2.909147908431189e-08,
      "loss": 0.6691,
      "step": 7671
    },
    {
      "epoch": 0.7584212737562711,
      "grad_norm": 4.646514625418928,
      "learning_rate": 2.906890389404636e-08,
      "loss": 0.7118,
      "step": 7672
    },
    {
      "epoch": 0.7585201295010257,
      "grad_norm": 4.2605858989752035,
      "learning_rate": 2.9046335976718438e-08,
      "loss": 0.6933,
      "step": 7673
    },
    {
      "epoch": 0.7586189852457801,
      "grad_norm": 3.970395726046271,
      "learning_rate": 2.9023775334642154e-08,
      "loss": 0.6264,
      "step": 7674
    },
    {
      "epoch": 0.7587178409905345,
      "grad_norm": 7.0836151930196465,
      "learning_rate": 2.9001221970130697e-08,
      "loss": 0.742,
      "step": 7675
    },
    {
      "epoch": 0.7588166967352891,
      "grad_norm": 10.376454545053726,
      "learning_rate": 2.8978675885496627e-08,
      "loss": 0.7233,
      "step": 7676
    },
    {
      "epoch": 0.7589155524800435,
      "grad_norm": 7.494289143163814,
      "learning_rate": 2.8956137083051735e-08,
      "loss": 0.6788,
      "step": 7677
    },
    {
      "epoch": 0.7590144082247979,
      "grad_norm": 3.2950742928235286,
      "learning_rate": 2.893360556510699e-08,
      "loss": 0.716,
      "step": 7678
    },
    {
      "epoch": 0.7591132639695525,
      "grad_norm": 9.203779922109412,
      "learning_rate": 2.891108133397271e-08,
      "loss": 0.7455,
      "step": 7679
    },
    {
      "epoch": 0.7592121197143069,
      "grad_norm": 2.8546459484307123,
      "learning_rate": 2.888856439195838e-08,
      "loss": 0.6707,
      "step": 7680
    },
    {
      "epoch": 0.7593109754590613,
      "grad_norm": 5.634179656389038,
      "learning_rate": 2.8866054741372803e-08,
      "loss": 0.8216,
      "step": 7681
    },
    {
      "epoch": 0.7594098312038158,
      "grad_norm": 4.339289165860522,
      "learning_rate": 2.884355238452405e-08,
      "loss": 0.5192,
      "step": 7682
    },
    {
      "epoch": 0.7595086869485703,
      "grad_norm": 3.3165759561772665,
      "learning_rate": 2.8821057323719322e-08,
      "loss": 0.7386,
      "step": 7683
    },
    {
      "epoch": 0.7596075426933248,
      "grad_norm": 2.9066172590634536,
      "learning_rate": 2.879856956126525e-08,
      "loss": 0.7795,
      "step": 7684
    },
    {
      "epoch": 0.7597063984380792,
      "grad_norm": 5.719469561115347,
      "learning_rate": 2.8776089099467537e-08,
      "loss": 0.7622,
      "step": 7685
    },
    {
      "epoch": 0.7598052541828337,
      "grad_norm": 3.4361212754804447,
      "learning_rate": 2.8753615940631293e-08,
      "loss": 0.7718,
      "step": 7686
    },
    {
      "epoch": 0.7599041099275882,
      "grad_norm": 3.854583665844336,
      "learning_rate": 2.873115008706074e-08,
      "loss": 0.7329,
      "step": 7687
    },
    {
      "epoch": 0.7600029656723426,
      "grad_norm": 5.670550314712351,
      "learning_rate": 2.8708691541059494e-08,
      "loss": 0.7531,
      "step": 7688
    },
    {
      "epoch": 0.7601018214170971,
      "grad_norm": 3.768510966317342,
      "learning_rate": 2.8686240304930266e-08,
      "loss": 0.7482,
      "step": 7689
    },
    {
      "epoch": 0.7602006771618516,
      "grad_norm": 4.165613402830636,
      "learning_rate": 2.8663796380975192e-08,
      "loss": 0.7258,
      "step": 7690
    },
    {
      "epoch": 0.760299532906606,
      "grad_norm": 4.886643853622551,
      "learning_rate": 2.8641359771495466e-08,
      "loss": 0.7764,
      "step": 7691
    },
    {
      "epoch": 0.7603983886513606,
      "grad_norm": 3.470613586958174,
      "learning_rate": 2.861893047879169e-08,
      "loss": 0.787,
      "step": 7692
    },
    {
      "epoch": 0.760497244396115,
      "grad_norm": 5.013998272962857,
      "learning_rate": 2.8596508505163687e-08,
      "loss": 0.7366,
      "step": 7693
    },
    {
      "epoch": 0.7605961001408694,
      "grad_norm": 5.866121311604234,
      "learning_rate": 2.8574093852910418e-08,
      "loss": 0.7451,
      "step": 7694
    },
    {
      "epoch": 0.7606949558856239,
      "grad_norm": 4.099549992024438,
      "learning_rate": 2.855168652433023e-08,
      "loss": 0.7777,
      "step": 7695
    },
    {
      "epoch": 0.7607938116303784,
      "grad_norm": 3.786202661499008,
      "learning_rate": 2.8529286521720675e-08,
      "loss": 0.8161,
      "step": 7696
    },
    {
      "epoch": 0.7608926673751328,
      "grad_norm": 3.638618476916516,
      "learning_rate": 2.8506893847378488e-08,
      "loss": 0.6963,
      "step": 7697
    },
    {
      "epoch": 0.7609915231198873,
      "grad_norm": 3.260468197090366,
      "learning_rate": 2.8484508503599748e-08,
      "loss": 0.6518,
      "step": 7698
    },
    {
      "epoch": 0.7610903788646418,
      "grad_norm": 3.9428210456253376,
      "learning_rate": 2.846213049267976e-08,
      "loss": 0.7618,
      "step": 7699
    },
    {
      "epoch": 0.7611892346093962,
      "grad_norm": 4.729381769570667,
      "learning_rate": 2.8439759816913e-08,
      "loss": 0.592,
      "step": 7700
    },
    {
      "epoch": 0.7612880903541507,
      "grad_norm": 6.319555541027446,
      "learning_rate": 2.8417396478593315e-08,
      "loss": 0.6484,
      "step": 7701
    },
    {
      "epoch": 0.7613869460989052,
      "grad_norm": 5.42687326584172,
      "learning_rate": 2.83950404800137e-08,
      "loss": 0.6854,
      "step": 7702
    },
    {
      "epoch": 0.7614858018436597,
      "grad_norm": 6.967365256549571,
      "learning_rate": 2.837269182346641e-08,
      "loss": 0.7738,
      "step": 7703
    },
    {
      "epoch": 0.7615846575884141,
      "grad_norm": 4.149131907356359,
      "learning_rate": 2.8350350511243016e-08,
      "loss": 0.6562,
      "step": 7704
    },
    {
      "epoch": 0.7616835133331685,
      "grad_norm": 4.822271234033115,
      "learning_rate": 2.832801654563425e-08,
      "loss": 0.7774,
      "step": 7705
    },
    {
      "epoch": 0.7617823690779231,
      "grad_norm": 4.9508556882021155,
      "learning_rate": 2.8305689928930153e-08,
      "loss": 0.7201,
      "step": 7706
    },
    {
      "epoch": 0.7618812248226775,
      "grad_norm": 3.493834358485,
      "learning_rate": 2.8283370663420015e-08,
      "loss": 0.6808,
      "step": 7707
    },
    {
      "epoch": 0.7619800805674319,
      "grad_norm": 3.316985897985715,
      "learning_rate": 2.8261058751392286e-08,
      "loss": 0.6998,
      "step": 7708
    },
    {
      "epoch": 0.7620789363121865,
      "grad_norm": 3.643452157574719,
      "learning_rate": 2.8238754195134762e-08,
      "loss": 0.7914,
      "step": 7709
    },
    {
      "epoch": 0.7621777920569409,
      "grad_norm": 4.771487245857684,
      "learning_rate": 2.8216456996934478e-08,
      "loss": 0.713,
      "step": 7710
    },
    {
      "epoch": 0.7622766478016954,
      "grad_norm": 26.200209005438687,
      "learning_rate": 2.819416715907761e-08,
      "loss": 0.7206,
      "step": 7711
    },
    {
      "epoch": 0.7623755035464499,
      "grad_norm": 4.13997449967982,
      "learning_rate": 2.8171884683849722e-08,
      "loss": 0.7292,
      "step": 7712
    },
    {
      "epoch": 0.7624743592912043,
      "grad_norm": 3.095377883871744,
      "learning_rate": 2.8149609573535493e-08,
      "loss": 0.6719,
      "step": 7713
    },
    {
      "epoch": 0.7625732150359588,
      "grad_norm": 3.1089972088474718,
      "learning_rate": 2.8127341830418926e-08,
      "loss": 0.6477,
      "step": 7714
    },
    {
      "epoch": 0.7626720707807132,
      "grad_norm": 8.412194103305481,
      "learning_rate": 2.8105081456783297e-08,
      "loss": 0.6128,
      "step": 7715
    },
    {
      "epoch": 0.7627709265254677,
      "grad_norm": 3.984004545033115,
      "learning_rate": 2.808282845491101e-08,
      "loss": 0.6586,
      "step": 7716
    },
    {
      "epoch": 0.7628697822702222,
      "grad_norm": 3.3132265995412555,
      "learning_rate": 2.806058282708381e-08,
      "loss": 0.7717,
      "step": 7717
    },
    {
      "epoch": 0.7629686380149766,
      "grad_norm": 5.061407620579918,
      "learning_rate": 2.8038344575582684e-08,
      "loss": 0.6649,
      "step": 7718
    },
    {
      "epoch": 0.7630674937597312,
      "grad_norm": 3.3754566312456973,
      "learning_rate": 2.8016113702687817e-08,
      "loss": 0.8219,
      "step": 7719
    },
    {
      "epoch": 0.7631663495044856,
      "grad_norm": 3.598952798592271,
      "learning_rate": 2.7993890210678628e-08,
      "loss": 0.7785,
      "step": 7720
    },
    {
      "epoch": 0.76326520524924,
      "grad_norm": 10.817164405416433,
      "learning_rate": 2.7971674101833854e-08,
      "loss": 0.7852,
      "step": 7721
    },
    {
      "epoch": 0.7633640609939946,
      "grad_norm": 12.316263732399053,
      "learning_rate": 2.7949465378431382e-08,
      "loss": 0.6152,
      "step": 7722
    },
    {
      "epoch": 0.763462916738749,
      "grad_norm": 15.199118233567981,
      "learning_rate": 2.7927264042748443e-08,
      "loss": 0.7112,
      "step": 7723
    },
    {
      "epoch": 0.7635617724835034,
      "grad_norm": 4.061778410765547,
      "learning_rate": 2.7905070097061413e-08,
      "loss": 0.6377,
      "step": 7724
    },
    {
      "epoch": 0.7636606282282579,
      "grad_norm": 11.745589284977505,
      "learning_rate": 2.7882883543645953e-08,
      "loss": 0.7218,
      "step": 7725
    },
    {
      "epoch": 0.7637594839730124,
      "grad_norm": 3.477080117477137,
      "learning_rate": 2.7860704384777023e-08,
      "loss": 0.7437,
      "step": 7726
    },
    {
      "epoch": 0.7638583397177668,
      "grad_norm": 4.552923198138568,
      "learning_rate": 2.7838532622728682e-08,
      "loss": 0.7012,
      "step": 7727
    },
    {
      "epoch": 0.7639571954625213,
      "grad_norm": 3.8808959990857024,
      "learning_rate": 2.781636825977438e-08,
      "loss": 0.669,
      "step": 7728
    },
    {
      "epoch": 0.7640560512072758,
      "grad_norm": 5.650939051106397,
      "learning_rate": 2.7794211298186754e-08,
      "loss": 0.6271,
      "step": 7729
    },
    {
      "epoch": 0.7641549069520303,
      "grad_norm": 4.180632579747479,
      "learning_rate": 2.777206174023763e-08,
      "loss": 0.6882,
      "step": 7730
    },
    {
      "epoch": 0.7642537626967847,
      "grad_norm": 3.0241712401143235,
      "learning_rate": 2.7749919588198122e-08,
      "loss": 0.6835,
      "step": 7731
    },
    {
      "epoch": 0.7643526184415392,
      "grad_norm": 3.993316534072559,
      "learning_rate": 2.772778484433863e-08,
      "loss": 0.798,
      "step": 7732
    },
    {
      "epoch": 0.7644514741862937,
      "grad_norm": 6.436699875066889,
      "learning_rate": 2.7705657510928692e-08,
      "loss": 0.6836,
      "step": 7733
    },
    {
      "epoch": 0.7645503299310481,
      "grad_norm": 16.933401219670642,
      "learning_rate": 2.7683537590237184e-08,
      "loss": 0.6698,
      "step": 7734
    },
    {
      "epoch": 0.7646491856758025,
      "grad_norm": 3.8586673930188837,
      "learning_rate": 2.766142508453212e-08,
      "loss": 0.6222,
      "step": 7735
    },
    {
      "epoch": 0.7647480414205571,
      "grad_norm": 5.91160829565221,
      "learning_rate": 2.7639319996080868e-08,
      "loss": 0.6491,
      "step": 7736
    },
    {
      "epoch": 0.7648468971653115,
      "grad_norm": 11.119945612413623,
      "learning_rate": 2.7617222327149935e-08,
      "loss": 0.6415,
      "step": 7737
    },
    {
      "epoch": 0.764945752910066,
      "grad_norm": 4.704230677614593,
      "learning_rate": 2.759513208000517e-08,
      "loss": 0.7527,
      "step": 7738
    },
    {
      "epoch": 0.7650446086548205,
      "grad_norm": 3.5558337515050247,
      "learning_rate": 2.757304925691152e-08,
      "loss": 0.6404,
      "step": 7739
    },
    {
      "epoch": 0.7651434643995749,
      "grad_norm": 4.0060590444869995,
      "learning_rate": 2.755097386013332e-08,
      "loss": 0.5807,
      "step": 7740
    },
    {
      "epoch": 0.7652423201443294,
      "grad_norm": 5.81216463833762,
      "learning_rate": 2.752890589193404e-08,
      "loss": 0.5835,
      "step": 7741
    },
    {
      "epoch": 0.7653411758890839,
      "grad_norm": 4.7544752237139,
      "learning_rate": 2.7506845354576425e-08,
      "loss": 0.697,
      "step": 7742
    },
    {
      "epoch": 0.7654400316338383,
      "grad_norm": 3.4941906818371975,
      "learning_rate": 2.7484792250322496e-08,
      "loss": 0.8299,
      "step": 7743
    },
    {
      "epoch": 0.7655388873785928,
      "grad_norm": 13.917772410335527,
      "learning_rate": 2.7462746581433426e-08,
      "loss": 0.7745,
      "step": 7744
    },
    {
      "epoch": 0.7656377431233472,
      "grad_norm": 3.249852155137751,
      "learning_rate": 2.7440708350169727e-08,
      "loss": 0.6978,
      "step": 7745
    },
    {
      "epoch": 0.7657365988681017,
      "grad_norm": 11.314630652254577,
      "learning_rate": 2.7418677558791025e-08,
      "loss": 0.6757,
      "step": 7746
    },
    {
      "epoch": 0.7658354546128562,
      "grad_norm": 13.68896732251147,
      "learning_rate": 2.739665420955629e-08,
      "loss": 0.7617,
      "step": 7747
    },
    {
      "epoch": 0.7659343103576106,
      "grad_norm": 3.4474684717881168,
      "learning_rate": 2.7374638304723728e-08,
      "loss": 0.7689,
      "step": 7748
    },
    {
      "epoch": 0.7660331661023652,
      "grad_norm": 4.263329517579261,
      "learning_rate": 2.735262984655067e-08,
      "loss": 0.8261,
      "step": 7749
    },
    {
      "epoch": 0.7661320218471196,
      "grad_norm": 2.775954664006627,
      "learning_rate": 2.73306288372938e-08,
      "loss": 0.6442,
      "step": 7750
    },
    {
      "epoch": 0.766230877591874,
      "grad_norm": 5.419677290077693,
      "learning_rate": 2.7308635279209025e-08,
      "loss": 0.7991,
      "step": 7751
    },
    {
      "epoch": 0.7663297333366286,
      "grad_norm": 3.798322933612664,
      "learning_rate": 2.72866491745514e-08,
      "loss": 0.6729,
      "step": 7752
    },
    {
      "epoch": 0.766428589081383,
      "grad_norm": 5.185541687952851,
      "learning_rate": 2.7264670525575328e-08,
      "loss": 0.7384,
      "step": 7753
    },
    {
      "epoch": 0.7665274448261374,
      "grad_norm": 3.1205196068877954,
      "learning_rate": 2.7242699334534347e-08,
      "loss": 0.6821,
      "step": 7754
    },
    {
      "epoch": 0.7666263005708919,
      "grad_norm": 4.056640492876918,
      "learning_rate": 2.7220735603681332e-08,
      "loss": 0.5666,
      "step": 7755
    },
    {
      "epoch": 0.7667251563156464,
      "grad_norm": 22.334397871861015,
      "learning_rate": 2.719877933526832e-08,
      "loss": 0.5935,
      "step": 7756
    },
    {
      "epoch": 0.7668240120604009,
      "grad_norm": 5.949645476294345,
      "learning_rate": 2.7176830531546557e-08,
      "loss": 0.6794,
      "step": 7757
    },
    {
      "epoch": 0.7669228678051553,
      "grad_norm": 3.9239446342427033,
      "learning_rate": 2.7154889194766595e-08,
      "loss": 0.6975,
      "step": 7758
    },
    {
      "epoch": 0.7670217235499098,
      "grad_norm": 5.492743512117824,
      "learning_rate": 2.7132955327178253e-08,
      "loss": 0.7964,
      "step": 7759
    },
    {
      "epoch": 0.7671205792946643,
      "grad_norm": 3.8698749328447675,
      "learning_rate": 2.7111028931030435e-08,
      "loss": 0.8151,
      "step": 7760
    },
    {
      "epoch": 0.7672194350394187,
      "grad_norm": 3.362204455027206,
      "learning_rate": 2.7089110008571415e-08,
      "loss": 0.6523,
      "step": 7761
    },
    {
      "epoch": 0.7673182907841732,
      "grad_norm": 10.792713286716948,
      "learning_rate": 2.7067198562048686e-08,
      "loss": 0.7384,
      "step": 7762
    },
    {
      "epoch": 0.7674171465289277,
      "grad_norm": 4.352181853603256,
      "learning_rate": 2.7045294593708877e-08,
      "loss": 0.719,
      "step": 7763
    },
    {
      "epoch": 0.7675160022736821,
      "grad_norm": 4.010174820326327,
      "learning_rate": 2.7023398105797956e-08,
      "loss": 0.7616,
      "step": 7764
    },
    {
      "epoch": 0.7676148580184365,
      "grad_norm": 3.1718757532126567,
      "learning_rate": 2.7001509100561103e-08,
      "loss": 0.6704,
      "step": 7765
    },
    {
      "epoch": 0.7677137137631911,
      "grad_norm": 3.899917485517286,
      "learning_rate": 2.697962758024266e-08,
      "loss": 0.7817,
      "step": 7766
    },
    {
      "epoch": 0.7678125695079455,
      "grad_norm": 6.5566046730775245,
      "learning_rate": 2.6957753547086305e-08,
      "loss": 0.6248,
      "step": 7767
    },
    {
      "epoch": 0.7679114252527,
      "grad_norm": 33.42092510007329,
      "learning_rate": 2.6935887003334856e-08,
      "loss": 0.6785,
      "step": 7768
    },
    {
      "epoch": 0.7680102809974545,
      "grad_norm": 4.17861467745345,
      "learning_rate": 2.691402795123041e-08,
      "loss": 0.7551,
      "step": 7769
    },
    {
      "epoch": 0.7681091367422089,
      "grad_norm": 4.48541131935433,
      "learning_rate": 2.6892176393014344e-08,
      "loss": 0.6944,
      "step": 7770
    },
    {
      "epoch": 0.7682079924869634,
      "grad_norm": 8.49018939585132,
      "learning_rate": 2.6870332330927148e-08,
      "loss": 0.697,
      "step": 7771
    },
    {
      "epoch": 0.7683068482317179,
      "grad_norm": 4.699940903760272,
      "learning_rate": 2.6848495767208657e-08,
      "loss": 0.6344,
      "step": 7772
    },
    {
      "epoch": 0.7684057039764723,
      "grad_norm": 7.60990929239871,
      "learning_rate": 2.6826666704097857e-08,
      "loss": 0.7217,
      "step": 7773
    },
    {
      "epoch": 0.7685045597212268,
      "grad_norm": 3.9679978372658513,
      "learning_rate": 2.680484514383298e-08,
      "loss": 0.7022,
      "step": 7774
    },
    {
      "epoch": 0.7686034154659813,
      "grad_norm": 3.340331177863032,
      "learning_rate": 2.678303108865153e-08,
      "loss": 0.6926,
      "step": 7775
    },
    {
      "epoch": 0.7687022712107358,
      "grad_norm": 4.818799757456171,
      "learning_rate": 2.676122454079024e-08,
      "loss": 0.7474,
      "step": 7776
    },
    {
      "epoch": 0.7688011269554902,
      "grad_norm": 5.010739229758358,
      "learning_rate": 2.6739425502485003e-08,
      "loss": 0.6385,
      "step": 7777
    },
    {
      "epoch": 0.7688999827002446,
      "grad_norm": 4.231616556863292,
      "learning_rate": 2.6717633975971032e-08,
      "loss": 0.6535,
      "step": 7778
    },
    {
      "epoch": 0.7689988384449992,
      "grad_norm": 8.299982756642985,
      "learning_rate": 2.6695849963482687e-08,
      "loss": 0.7788,
      "step": 7779
    },
    {
      "epoch": 0.7690976941897536,
      "grad_norm": 5.255707822625191,
      "learning_rate": 2.667407346725361e-08,
      "loss": 0.7507,
      "step": 7780
    },
    {
      "epoch": 0.769196549934508,
      "grad_norm": 4.1920709379414145,
      "learning_rate": 2.6652304489516696e-08,
      "loss": 0.7026,
      "step": 7781
    },
    {
      "epoch": 0.7692954056792626,
      "grad_norm": 3.824001847060637,
      "learning_rate": 2.663054303250397e-08,
      "loss": 0.6327,
      "step": 7782
    },
    {
      "epoch": 0.769394261424017,
      "grad_norm": 10.250700256341972,
      "learning_rate": 2.660878909844678e-08,
      "loss": 0.6404,
      "step": 7783
    },
    {
      "epoch": 0.7694931171687714,
      "grad_norm": 73.13139297712725,
      "learning_rate": 2.658704268957571e-08,
      "loss": 0.6649,
      "step": 7784
    },
    {
      "epoch": 0.769591972913526,
      "grad_norm": 3.99899882767821,
      "learning_rate": 2.6565303808120453e-08,
      "loss": 0.6891,
      "step": 7785
    },
    {
      "epoch": 0.7696908286582804,
      "grad_norm": 3.3974851859886357,
      "learning_rate": 2.654357245631006e-08,
      "loss": 0.6823,
      "step": 7786
    },
    {
      "epoch": 0.7697896844030349,
      "grad_norm": 8.404689328926327,
      "learning_rate": 2.6521848636372778e-08,
      "loss": 0.7519,
      "step": 7787
    },
    {
      "epoch": 0.7698885401477893,
      "grad_norm": 5.973949684267329,
      "learning_rate": 2.650013235053602e-08,
      "loss": 0.6163,
      "step": 7788
    },
    {
      "epoch": 0.7699873958925438,
      "grad_norm": 5.972775361704657,
      "learning_rate": 2.647842360102651e-08,
      "loss": 0.6249,
      "step": 7789
    },
    {
      "epoch": 0.7700862516372983,
      "grad_norm": 3.2950315646023762,
      "learning_rate": 2.645672239007014e-08,
      "loss": 0.7241,
      "step": 7790
    },
    {
      "epoch": 0.7701851073820527,
      "grad_norm": 8.36952578944059,
      "learning_rate": 2.643502871989203e-08,
      "loss": 0.7098,
      "step": 7791
    },
    {
      "epoch": 0.7702839631268072,
      "grad_norm": 3.4740483916749896,
      "learning_rate": 2.6413342592716582e-08,
      "loss": 0.7469,
      "step": 7792
    },
    {
      "epoch": 0.7703828188715617,
      "grad_norm": 6.768872898735627,
      "learning_rate": 2.6391664010767345e-08,
      "loss": 0.7554,
      "step": 7793
    },
    {
      "epoch": 0.7704816746163161,
      "grad_norm": 11.111822640368167,
      "learning_rate": 2.6369992976267164e-08,
      "loss": 0.8389,
      "step": 7794
    },
    {
      "epoch": 0.7705805303610707,
      "grad_norm": 40.11663348099736,
      "learning_rate": 2.6348329491438124e-08,
      "loss": 0.6296,
      "step": 7795
    },
    {
      "epoch": 0.7706793861058251,
      "grad_norm": 4.008511875473146,
      "learning_rate": 2.632667355850141e-08,
      "loss": 0.7142,
      "step": 7796
    },
    {
      "epoch": 0.7707782418505795,
      "grad_norm": 3.7868138865566947,
      "learning_rate": 2.6305025179677576e-08,
      "loss": 0.7762,
      "step": 7797
    },
    {
      "epoch": 0.770877097595334,
      "grad_norm": 3.56291595561482,
      "learning_rate": 2.6283384357186367e-08,
      "loss": 0.6486,
      "step": 7798
    },
    {
      "epoch": 0.7709759533400885,
      "grad_norm": 3.780496904388255,
      "learning_rate": 2.6261751093246655e-08,
      "loss": 0.6145,
      "step": 7799
    },
    {
      "epoch": 0.7710748090848429,
      "grad_norm": 3.967801893055563,
      "learning_rate": 2.6240125390076684e-08,
      "loss": 0.7541,
      "step": 7800
    },
    {
      "epoch": 0.7711736648295974,
      "grad_norm": 4.972618896221208,
      "learning_rate": 2.6218507249893794e-08,
      "loss": 0.6587,
      "step": 7801
    },
    {
      "epoch": 0.7712725205743519,
      "grad_norm": 5.47735140435406,
      "learning_rate": 2.6196896674914624e-08,
      "loss": 0.6386,
      "step": 7802
    },
    {
      "epoch": 0.7713713763191063,
      "grad_norm": 4.7975108484846825,
      "learning_rate": 2.617529366735507e-08,
      "loss": 0.7187,
      "step": 7803
    },
    {
      "epoch": 0.7714702320638608,
      "grad_norm": 5.87551946125205,
      "learning_rate": 2.615369822943012e-08,
      "loss": 0.8296,
      "step": 7804
    },
    {
      "epoch": 0.7715690878086153,
      "grad_norm": 3.4354087194927767,
      "learning_rate": 2.613211036335411e-08,
      "loss": 0.6193,
      "step": 7805
    },
    {
      "epoch": 0.7716679435533698,
      "grad_norm": 9.134694197069546,
      "learning_rate": 2.6110530071340576e-08,
      "loss": 0.6854,
      "step": 7806
    },
    {
      "epoch": 0.7717667992981242,
      "grad_norm": 3.116167539522937,
      "learning_rate": 2.6088957355602248e-08,
      "loss": 0.6728,
      "step": 7807
    },
    {
      "epoch": 0.7718656550428786,
      "grad_norm": 3.104692206613377,
      "learning_rate": 2.606739221835105e-08,
      "loss": 0.6906,
      "step": 7808
    },
    {
      "epoch": 0.7719645107876332,
      "grad_norm": 13.449777594554032,
      "learning_rate": 2.6045834661798218e-08,
      "loss": 0.6655,
      "step": 7809
    },
    {
      "epoch": 0.7720633665323876,
      "grad_norm": 5.294268431202817,
      "learning_rate": 2.6024284688154107e-08,
      "loss": 0.7741,
      "step": 7810
    },
    {
      "epoch": 0.772162222277142,
      "grad_norm": 4.120823294699228,
      "learning_rate": 2.600274229962842e-08,
      "loss": 0.8111,
      "step": 7811
    },
    {
      "epoch": 0.7722610780218966,
      "grad_norm": 3.698124093794932,
      "learning_rate": 2.5981207498429946e-08,
      "loss": 0.7976,
      "step": 7812
    },
    {
      "epoch": 0.772359933766651,
      "grad_norm": 7.690019920759459,
      "learning_rate": 2.5959680286766796e-08,
      "loss": 0.5981,
      "step": 7813
    },
    {
      "epoch": 0.7724587895114055,
      "grad_norm": 5.278427886910685,
      "learning_rate": 2.5938160666846277e-08,
      "loss": 0.7954,
      "step": 7814
    },
    {
      "epoch": 0.77255764525616,
      "grad_norm": 3.7137211111589794,
      "learning_rate": 2.5916648640874872e-08,
      "loss": 0.5925,
      "step": 7815
    },
    {
      "epoch": 0.7726565010009144,
      "grad_norm": 3.5673840619424166,
      "learning_rate": 2.5895144211058327e-08,
      "loss": 0.6599,
      "step": 7816
    },
    {
      "epoch": 0.7727553567456689,
      "grad_norm": 4.756944285952017,
      "learning_rate": 2.587364737960167e-08,
      "loss": 0.7658,
      "step": 7817
    },
    {
      "epoch": 0.7728542124904233,
      "grad_norm": 12.920488899905608,
      "learning_rate": 2.585215814870899e-08,
      "loss": 0.6735,
      "step": 7818
    },
    {
      "epoch": 0.7729530682351778,
      "grad_norm": 3.193698076410193,
      "learning_rate": 2.5830676520583737e-08,
      "loss": 0.6738,
      "step": 7819
    },
    {
      "epoch": 0.7730519239799323,
      "grad_norm": 15.33694145620271,
      "learning_rate": 2.5809202497428573e-08,
      "loss": 0.706,
      "step": 7820
    },
    {
      "epoch": 0.7731507797246867,
      "grad_norm": 3.4167809834507694,
      "learning_rate": 2.5787736081445256e-08,
      "loss": 0.794,
      "step": 7821
    },
    {
      "epoch": 0.7732496354694413,
      "grad_norm": 3.62114800737765,
      "learning_rate": 2.5766277274834937e-08,
      "loss": 0.603,
      "step": 7822
    },
    {
      "epoch": 0.7733484912141957,
      "grad_norm": 3.7659756095915027,
      "learning_rate": 2.574482607979782e-08,
      "loss": 0.7573,
      "step": 7823
    },
    {
      "epoch": 0.7734473469589501,
      "grad_norm": 4.037636062084781,
      "learning_rate": 2.5723382498533484e-08,
      "loss": 0.7389,
      "step": 7824
    },
    {
      "epoch": 0.7735462027037047,
      "grad_norm": 4.555404502891533,
      "learning_rate": 2.5701946533240582e-08,
      "loss": 0.7518,
      "step": 7825
    },
    {
      "epoch": 0.7736450584484591,
      "grad_norm": 9.05611203869871,
      "learning_rate": 2.568051818611712e-08,
      "loss": 0.7707,
      "step": 7826
    },
    {
      "epoch": 0.7737439141932135,
      "grad_norm": 12.019151131639116,
      "learning_rate": 2.56590974593602e-08,
      "loss": 0.7421,
      "step": 7827
    },
    {
      "epoch": 0.773842769937968,
      "grad_norm": 3.6326228063213075,
      "learning_rate": 2.5637684355166268e-08,
      "loss": 0.6279,
      "step": 7828
    },
    {
      "epoch": 0.7739416256827225,
      "grad_norm": 4.687242757218988,
      "learning_rate": 2.5616278875730846e-08,
      "loss": 0.6011,
      "step": 7829
    },
    {
      "epoch": 0.774040481427477,
      "grad_norm": 3.108509908321275,
      "learning_rate": 2.559488102324878e-08,
      "loss": 0.6954,
      "step": 7830
    },
    {
      "epoch": 0.7741393371722314,
      "grad_norm": 3.368255466405175,
      "learning_rate": 2.5573490799914156e-08,
      "loss": 0.7082,
      "step": 7831
    },
    {
      "epoch": 0.7742381929169859,
      "grad_norm": 3.2315958395553506,
      "learning_rate": 2.5552108207920143e-08,
      "loss": 0.7823,
      "step": 7832
    },
    {
      "epoch": 0.7743370486617404,
      "grad_norm": 4.050550965030356,
      "learning_rate": 2.553073324945928e-08,
      "loss": 0.7997,
      "step": 7833
    },
    {
      "epoch": 0.7744359044064948,
      "grad_norm": 6.618567190019175,
      "learning_rate": 2.5509365926723204e-08,
      "loss": 0.6491,
      "step": 7834
    },
    {
      "epoch": 0.7745347601512493,
      "grad_norm": 3.5003407542799936,
      "learning_rate": 2.5488006241902827e-08,
      "loss": 0.6545,
      "step": 7835
    },
    {
      "epoch": 0.7746336158960038,
      "grad_norm": 4.949596426640748,
      "learning_rate": 2.546665419718832e-08,
      "loss": 0.7222,
      "step": 7836
    },
    {
      "epoch": 0.7747324716407582,
      "grad_norm": 3.600510288365261,
      "learning_rate": 2.5445309794768943e-08,
      "loss": 0.7661,
      "step": 7837
    },
    {
      "epoch": 0.7748313273855126,
      "grad_norm": 4.065068672310413,
      "learning_rate": 2.5423973036833292e-08,
      "loss": 0.7256,
      "step": 7838
    },
    {
      "epoch": 0.7749301831302672,
      "grad_norm": 7.8984641923595085,
      "learning_rate": 2.5402643925569155e-08,
      "loss": 0.6948,
      "step": 7839
    },
    {
      "epoch": 0.7750290388750216,
      "grad_norm": 4.067442692286579,
      "learning_rate": 2.538132246316347e-08,
      "loss": 0.6678,
      "step": 7840
    },
    {
      "epoch": 0.775127894619776,
      "grad_norm": 3.280842068791394,
      "learning_rate": 2.536000865180249e-08,
      "loss": 0.657,
      "step": 7841
    },
    {
      "epoch": 0.7752267503645306,
      "grad_norm": 4.582771439075677,
      "learning_rate": 2.533870249367158e-08,
      "loss": 0.8016,
      "step": 7842
    },
    {
      "epoch": 0.775325606109285,
      "grad_norm": 4.107048169349126,
      "learning_rate": 2.5317403990955433e-08,
      "loss": 0.6851,
      "step": 7843
    },
    {
      "epoch": 0.7754244618540395,
      "grad_norm": 15.914086903796616,
      "learning_rate": 2.5296113145837848e-08,
      "loss": 0.6078,
      "step": 7844
    },
    {
      "epoch": 0.775523317598794,
      "grad_norm": 3.1379085250894208,
      "learning_rate": 2.5274829960501876e-08,
      "loss": 0.807,
      "step": 7845
    },
    {
      "epoch": 0.7756221733435484,
      "grad_norm": 5.689082901435306,
      "learning_rate": 2.525355443712982e-08,
      "loss": 0.6493,
      "step": 7846
    },
    {
      "epoch": 0.7757210290883029,
      "grad_norm": 3.1499860768384913,
      "learning_rate": 2.5232286577903196e-08,
      "loss": 0.7222,
      "step": 7847
    },
    {
      "epoch": 0.7758198848330574,
      "grad_norm": 4.823590016474512,
      "learning_rate": 2.5211026385002655e-08,
      "loss": 0.6952,
      "step": 7848
    },
    {
      "epoch": 0.7759187405778118,
      "grad_norm": 5.027458951369968,
      "learning_rate": 2.5189773860608142e-08,
      "loss": 0.6623,
      "step": 7849
    },
    {
      "epoch": 0.7760175963225663,
      "grad_norm": 11.202331662735812,
      "learning_rate": 2.516852900689882e-08,
      "loss": 0.7407,
      "step": 7850
    },
    {
      "epoch": 0.7761164520673207,
      "grad_norm": 9.87498330880613,
      "learning_rate": 2.5147291826052974e-08,
      "loss": 0.6608,
      "step": 7851
    },
    {
      "epoch": 0.7762153078120753,
      "grad_norm": 4.304316812843773,
      "learning_rate": 2.5126062320248198e-08,
      "loss": 0.6151,
      "step": 7852
    },
    {
      "epoch": 0.7763141635568297,
      "grad_norm": 4.678076872263316,
      "learning_rate": 2.5104840491661294e-08,
      "loss": 0.679,
      "step": 7853
    },
    {
      "epoch": 0.7764130193015841,
      "grad_norm": 3.1820867182128465,
      "learning_rate": 2.508362634246819e-08,
      "loss": 0.57,
      "step": 7854
    },
    {
      "epoch": 0.7765118750463387,
      "grad_norm": 4.893697264970483,
      "learning_rate": 2.506241987484413e-08,
      "loss": 0.6696,
      "step": 7855
    },
    {
      "epoch": 0.7766107307910931,
      "grad_norm": 4.721493917207291,
      "learning_rate": 2.5041221090963484e-08,
      "loss": 0.6965,
      "step": 7856
    },
    {
      "epoch": 0.7767095865358475,
      "grad_norm": 4.609969200912756,
      "learning_rate": 2.5020029992999893e-08,
      "loss": 0.7247,
      "step": 7857
    },
    {
      "epoch": 0.7768084422806021,
      "grad_norm": 20.289897832421474,
      "learning_rate": 2.4998846583126222e-08,
      "loss": 0.5889,
      "step": 7858
    },
    {
      "epoch": 0.7769072980253565,
      "grad_norm": 6.718793825307793,
      "learning_rate": 2.4977670863514466e-08,
      "loss": 0.5897,
      "step": 7859
    },
    {
      "epoch": 0.777006153770111,
      "grad_norm": 5.03242359189194,
      "learning_rate": 2.4956502836335923e-08,
      "loss": 0.8359,
      "step": 7860
    },
    {
      "epoch": 0.7771050095148654,
      "grad_norm": 5.320287356015045,
      "learning_rate": 2.4935342503761057e-08,
      "loss": 0.8093,
      "step": 7861
    },
    {
      "epoch": 0.7772038652596199,
      "grad_norm": 5.123208578651246,
      "learning_rate": 2.49141898679595e-08,
      "loss": 0.8577,
      "step": 7862
    },
    {
      "epoch": 0.7773027210043744,
      "grad_norm": 3.576465095909544,
      "learning_rate": 2.4893044931100184e-08,
      "loss": 0.6591,
      "step": 7863
    },
    {
      "epoch": 0.7774015767491288,
      "grad_norm": 3.161508179896915,
      "learning_rate": 2.4871907695351223e-08,
      "loss": 0.7164,
      "step": 7864
    },
    {
      "epoch": 0.7775004324938833,
      "grad_norm": 5.63194555365071,
      "learning_rate": 2.485077816287988e-08,
      "loss": 0.8067,
      "step": 7865
    },
    {
      "epoch": 0.7775992882386378,
      "grad_norm": 5.0081160182814015,
      "learning_rate": 2.4829656335852757e-08,
      "loss": 0.7283,
      "step": 7866
    },
    {
      "epoch": 0.7776981439833922,
      "grad_norm": 4.922390743381536,
      "learning_rate": 2.480854221643549e-08,
      "loss": 0.7226,
      "step": 7867
    },
    {
      "epoch": 0.7777969997281468,
      "grad_norm": 3.921028153008956,
      "learning_rate": 2.4787435806793067e-08,
      "loss": 0.6976,
      "step": 7868
    },
    {
      "epoch": 0.7778958554729012,
      "grad_norm": 6.405015568662818,
      "learning_rate": 2.4766337109089674e-08,
      "loss": 0.7753,
      "step": 7869
    },
    {
      "epoch": 0.7779947112176556,
      "grad_norm": 5.548867722173285,
      "learning_rate": 2.4745246125488606e-08,
      "loss": 0.7417,
      "step": 7870
    },
    {
      "epoch": 0.77809356696241,
      "grad_norm": 5.175124487425344,
      "learning_rate": 2.4724162858152454e-08,
      "loss": 0.6572,
      "step": 7871
    },
    {
      "epoch": 0.7781924227071646,
      "grad_norm": 5.088730738339677,
      "learning_rate": 2.470308730924304e-08,
      "loss": 0.8145,
      "step": 7872
    },
    {
      "epoch": 0.778291278451919,
      "grad_norm": 5.3897524131874235,
      "learning_rate": 2.468201948092129e-08,
      "loss": 0.6558,
      "step": 7873
    },
    {
      "epoch": 0.7783901341966735,
      "grad_norm": 4.354893623262532,
      "learning_rate": 2.4660959375347412e-08,
      "loss": 0.6953,
      "step": 7874
    },
    {
      "epoch": 0.778488989941428,
      "grad_norm": 4.959656743126704,
      "learning_rate": 2.463990699468086e-08,
      "loss": 0.7365,
      "step": 7875
    },
    {
      "epoch": 0.7785878456861824,
      "grad_norm": 5.729973811774531,
      "learning_rate": 2.4618862341080173e-08,
      "loss": 0.6843,
      "step": 7876
    },
    {
      "epoch": 0.7786867014309369,
      "grad_norm": 5.1103458409325535,
      "learning_rate": 2.4597825416703232e-08,
      "loss": 0.6905,
      "step": 7877
    },
    {
      "epoch": 0.7787855571756914,
      "grad_norm": 4.329918194066023,
      "learning_rate": 2.4576796223707043e-08,
      "loss": 0.833,
      "step": 7878
    },
    {
      "epoch": 0.7788844129204459,
      "grad_norm": 4.415346141883247,
      "learning_rate": 2.4555774764247794e-08,
      "loss": 0.647,
      "step": 7879
    },
    {
      "epoch": 0.7789832686652003,
      "grad_norm": 4.373319218030685,
      "learning_rate": 2.453476104048099e-08,
      "loss": 0.7205,
      "step": 7880
    },
    {
      "epoch": 0.7790821244099547,
      "grad_norm": 4.895082542938074,
      "learning_rate": 2.4513755054561215e-08,
      "loss": 0.6694,
      "step": 7881
    },
    {
      "epoch": 0.7791809801547093,
      "grad_norm": 3.5007784331611718,
      "learning_rate": 2.4492756808642368e-08,
      "loss": 0.7823,
      "step": 7882
    },
    {
      "epoch": 0.7792798358994637,
      "grad_norm": 12.033646083348511,
      "learning_rate": 2.447176630487753e-08,
      "loss": 0.7201,
      "step": 7883
    },
    {
      "epoch": 0.7793786916442181,
      "grad_norm": 4.189358711577295,
      "learning_rate": 2.4450783545418906e-08,
      "loss": 0.7126,
      "step": 7884
    },
    {
      "epoch": 0.7794775473889727,
      "grad_norm": 15.371167473047045,
      "learning_rate": 2.4429808532418007e-08,
      "loss": 0.6917,
      "step": 7885
    },
    {
      "epoch": 0.7795764031337271,
      "grad_norm": 4.091382019648271,
      "learning_rate": 2.4408841268025516e-08,
      "loss": 0.6573,
      "step": 7886
    },
    {
      "epoch": 0.7796752588784815,
      "grad_norm": 3.7722525361623194,
      "learning_rate": 2.4387881754391294e-08,
      "loss": 0.6972,
      "step": 7887
    },
    {
      "epoch": 0.7797741146232361,
      "grad_norm": 6.886480454037972,
      "learning_rate": 2.4366929993664455e-08,
      "loss": 0.7049,
      "step": 7888
    },
    {
      "epoch": 0.7798729703679905,
      "grad_norm": 14.724240358170256,
      "learning_rate": 2.4345985987993256e-08,
      "loss": 0.7445,
      "step": 7889
    },
    {
      "epoch": 0.779971826112745,
      "grad_norm": 3.5660519230763787,
      "learning_rate": 2.4325049739525226e-08,
      "loss": 0.7818,
      "step": 7890
    },
    {
      "epoch": 0.7800706818574994,
      "grad_norm": 3.285684590489847,
      "learning_rate": 2.4304121250407082e-08,
      "loss": 0.5649,
      "step": 7891
    },
    {
      "epoch": 0.7801695376022539,
      "grad_norm": 3.3685643436292048,
      "learning_rate": 2.4283200522784686e-08,
      "loss": 0.5813,
      "step": 7892
    },
    {
      "epoch": 0.7802683933470084,
      "grad_norm": 5.510563386249811,
      "learning_rate": 2.4262287558803174e-08,
      "loss": 0.6165,
      "step": 7893
    },
    {
      "epoch": 0.7803672490917628,
      "grad_norm": 3.995089416160853,
      "learning_rate": 2.4241382360606888e-08,
      "loss": 0.7725,
      "step": 7894
    },
    {
      "epoch": 0.7804661048365173,
      "grad_norm": 4.774022732811174,
      "learning_rate": 2.4220484930339326e-08,
      "loss": 0.7543,
      "step": 7895
    },
    {
      "epoch": 0.7805649605812718,
      "grad_norm": 3.8650770983945653,
      "learning_rate": 2.419959527014318e-08,
      "loss": 0.7446,
      "step": 7896
    },
    {
      "epoch": 0.7806638163260262,
      "grad_norm": 5.476286244833428,
      "learning_rate": 2.4178713382160433e-08,
      "loss": 0.6131,
      "step": 7897
    },
    {
      "epoch": 0.7807626720707808,
      "grad_norm": 5.0621497585986415,
      "learning_rate": 2.4157839268532154e-08,
      "loss": 0.703,
      "step": 7898
    },
    {
      "epoch": 0.7808615278155352,
      "grad_norm": 5.741148338724125,
      "learning_rate": 2.4136972931398736e-08,
      "loss": 0.8045,
      "step": 7899
    },
    {
      "epoch": 0.7809603835602896,
      "grad_norm": 4.458241664363763,
      "learning_rate": 2.4116114372899666e-08,
      "loss": 0.6747,
      "step": 7900
    },
    {
      "epoch": 0.7810592393050441,
      "grad_norm": 3.5361974923975277,
      "learning_rate": 2.409526359517369e-08,
      "loss": 0.7751,
      "step": 7901
    },
    {
      "epoch": 0.7811580950497986,
      "grad_norm": 6.824568871400273,
      "learning_rate": 2.407442060035879e-08,
      "loss": 0.6868,
      "step": 7902
    },
    {
      "epoch": 0.781256950794553,
      "grad_norm": 8.140043075304199,
      "learning_rate": 2.4053585390592046e-08,
      "loss": 0.609,
      "step": 7903
    },
    {
      "epoch": 0.7813558065393075,
      "grad_norm": 3.039565769862241,
      "learning_rate": 2.4032757968009842e-08,
      "loss": 0.6627,
      "step": 7904
    },
    {
      "epoch": 0.781454662284062,
      "grad_norm": 3.434873646152947,
      "learning_rate": 2.401193833474773e-08,
      "loss": 0.7198,
      "step": 7905
    },
    {
      "epoch": 0.7815535180288165,
      "grad_norm": 3.4310713585531722,
      "learning_rate": 2.3991126492940416e-08,
      "loss": 0.7228,
      "step": 7906
    },
    {
      "epoch": 0.7816523737735709,
      "grad_norm": 4.63224542081652,
      "learning_rate": 2.397032244472187e-08,
      "loss": 0.7034,
      "step": 7907
    },
    {
      "epoch": 0.7817512295183254,
      "grad_norm": 5.193377121493787,
      "learning_rate": 2.3949526192225267e-08,
      "loss": 0.6743,
      "step": 7908
    },
    {
      "epoch": 0.7818500852630799,
      "grad_norm": 15.873124967105285,
      "learning_rate": 2.3928737737582904e-08,
      "loss": 0.6583,
      "step": 7909
    },
    {
      "epoch": 0.7819489410078343,
      "grad_norm": 28.859846596895796,
      "learning_rate": 2.390795708292638e-08,
      "loss": 0.6525,
      "step": 7910
    },
    {
      "epoch": 0.7820477967525887,
      "grad_norm": 2.5298926352566906,
      "learning_rate": 2.3887184230386394e-08,
      "loss": 0.6051,
      "step": 7911
    },
    {
      "epoch": 0.7821466524973433,
      "grad_norm": 3.3718660496167887,
      "learning_rate": 2.3866419182092944e-08,
      "loss": 0.6883,
      "step": 7912
    },
    {
      "epoch": 0.7822455082420977,
      "grad_norm": 3.934728351151892,
      "learning_rate": 2.3845661940175144e-08,
      "loss": 0.6053,
      "step": 7913
    },
    {
      "epoch": 0.7823443639868521,
      "grad_norm": 3.8666302131467436,
      "learning_rate": 2.3824912506761374e-08,
      "loss": 0.6473,
      "step": 7914
    },
    {
      "epoch": 0.7824432197316067,
      "grad_norm": 7.342190123868332,
      "learning_rate": 2.3804170883979135e-08,
      "loss": 0.7657,
      "step": 7915
    },
    {
      "epoch": 0.7825420754763611,
      "grad_norm": 2.9422973762010156,
      "learning_rate": 2.378343707395525e-08,
      "loss": 0.7167,
      "step": 7916
    },
    {
      "epoch": 0.7826409312211156,
      "grad_norm": 3.8855545171358656,
      "learning_rate": 2.3762711078815568e-08,
      "loss": 0.777,
      "step": 7917
    },
    {
      "epoch": 0.7827397869658701,
      "grad_norm": 6.730717492546894,
      "learning_rate": 2.3741992900685304e-08,
      "loss": 0.7544,
      "step": 7918
    },
    {
      "epoch": 0.7828386427106245,
      "grad_norm": 4.851802296442411,
      "learning_rate": 2.3721282541688805e-08,
      "loss": 0.8233,
      "step": 7919
    },
    {
      "epoch": 0.782937498455379,
      "grad_norm": 2.75529166195502,
      "learning_rate": 2.3700580003949556e-08,
      "loss": 0.7013,
      "step": 7920
    },
    {
      "epoch": 0.7830363542001334,
      "grad_norm": 4.3353416403159715,
      "learning_rate": 2.367988528959036e-08,
      "loss": 0.7325,
      "step": 7921
    },
    {
      "epoch": 0.7831352099448879,
      "grad_norm": 11.16760617367665,
      "learning_rate": 2.3659198400733114e-08,
      "loss": 0.7297,
      "step": 7922
    },
    {
      "epoch": 0.7832340656896424,
      "grad_norm": 4.722837574271411,
      "learning_rate": 2.363851933949895e-08,
      "loss": 0.7152,
      "step": 7923
    },
    {
      "epoch": 0.7833329214343968,
      "grad_norm": 5.543837318559596,
      "learning_rate": 2.3617848108008253e-08,
      "loss": 0.6866,
      "step": 7924
    },
    {
      "epoch": 0.7834317771791514,
      "grad_norm": 3.4161797423835827,
      "learning_rate": 2.359718470838049e-08,
      "loss": 0.6271,
      "step": 7925
    },
    {
      "epoch": 0.7835306329239058,
      "grad_norm": 5.888002783240124,
      "learning_rate": 2.3576529142734404e-08,
      "loss": 0.8256,
      "step": 7926
    },
    {
      "epoch": 0.7836294886686602,
      "grad_norm": 4.587234413420677,
      "learning_rate": 2.3555881413187973e-08,
      "loss": 0.6735,
      "step": 7927
    },
    {
      "epoch": 0.7837283444134148,
      "grad_norm": 3.8163570026212597,
      "learning_rate": 2.353524152185823e-08,
      "loss": 0.6536,
      "step": 7928
    },
    {
      "epoch": 0.7838272001581692,
      "grad_norm": 4.120597087554894,
      "learning_rate": 2.351460947086157e-08,
      "loss": 0.6932,
      "step": 7929
    },
    {
      "epoch": 0.7839260559029236,
      "grad_norm": 4.141673464485594,
      "learning_rate": 2.349398526231344e-08,
      "loss": 0.7183,
      "step": 7930
    },
    {
      "epoch": 0.7840249116476782,
      "grad_norm": 5.32607780005227,
      "learning_rate": 2.3473368898328603e-08,
      "loss": 0.8215,
      "step": 7931
    },
    {
      "epoch": 0.7841237673924326,
      "grad_norm": 5.842244627963679,
      "learning_rate": 2.345276038102094e-08,
      "loss": 0.6964,
      "step": 7932
    },
    {
      "epoch": 0.784222623137187,
      "grad_norm": 7.742630230415506,
      "learning_rate": 2.343215971250353e-08,
      "loss": 0.7151,
      "step": 7933
    },
    {
      "epoch": 0.7843214788819415,
      "grad_norm": 68.40132664667709,
      "learning_rate": 2.3411566894888666e-08,
      "loss": 0.8121,
      "step": 7934
    },
    {
      "epoch": 0.784420334626696,
      "grad_norm": 4.456467583694892,
      "learning_rate": 2.33909819302879e-08,
      "loss": 0.7447,
      "step": 7935
    },
    {
      "epoch": 0.7845191903714505,
      "grad_norm": 10.864087984207856,
      "learning_rate": 2.337040482081185e-08,
      "loss": 0.6607,
      "step": 7936
    },
    {
      "epoch": 0.7846180461162049,
      "grad_norm": 6.510730612787214,
      "learning_rate": 2.334983556857042e-08,
      "loss": 0.7296,
      "step": 7937
    },
    {
      "epoch": 0.7847169018609594,
      "grad_norm": 4.307067143286345,
      "learning_rate": 2.33292741756727e-08,
      "loss": 0.7123,
      "step": 7938
    },
    {
      "epoch": 0.7848157576057139,
      "grad_norm": 3.9713333485682383,
      "learning_rate": 2.330872064422692e-08,
      "loss": 0.6539,
      "step": 7939
    },
    {
      "epoch": 0.7849146133504683,
      "grad_norm": 4.574166985472579,
      "learning_rate": 2.328817497634056e-08,
      "loss": 0.6858,
      "step": 7940
    },
    {
      "epoch": 0.7850134690952228,
      "grad_norm": 3.5062007763974705,
      "learning_rate": 2.3267637174120302e-08,
      "loss": 0.6881,
      "step": 7941
    },
    {
      "epoch": 0.7851123248399773,
      "grad_norm": 3.834903599709114,
      "learning_rate": 2.3247107239671947e-08,
      "loss": 0.6369,
      "step": 7942
    },
    {
      "epoch": 0.7852111805847317,
      "grad_norm": 3.334683755610701,
      "learning_rate": 2.3226585175100587e-08,
      "loss": 0.6692,
      "step": 7943
    },
    {
      "epoch": 0.7853100363294861,
      "grad_norm": 10.88139317589456,
      "learning_rate": 2.3206070982510396e-08,
      "loss": 0.7643,
      "step": 7944
    },
    {
      "epoch": 0.7854088920742407,
      "grad_norm": 6.373125928747028,
      "learning_rate": 2.318556466400484e-08,
      "loss": 0.6868,
      "step": 7945
    },
    {
      "epoch": 0.7855077478189951,
      "grad_norm": 3.408507971062486,
      "learning_rate": 2.316506622168657e-08,
      "loss": 0.7236,
      "step": 7946
    },
    {
      "epoch": 0.7856066035637496,
      "grad_norm": 5.928232301649691,
      "learning_rate": 2.3144575657657328e-08,
      "loss": 0.6597,
      "step": 7947
    },
    {
      "epoch": 0.7857054593085041,
      "grad_norm": 3.514562112733798,
      "learning_rate": 2.3124092974018183e-08,
      "loss": 0.7336,
      "step": 7948
    },
    {
      "epoch": 0.7858043150532585,
      "grad_norm": 3.6303565368409,
      "learning_rate": 2.31036181728693e-08,
      "loss": 0.8211,
      "step": 7949
    },
    {
      "epoch": 0.785903170798013,
      "grad_norm": 3.4470937067307714,
      "learning_rate": 2.3083151256310053e-08,
      "loss": 0.7089,
      "step": 7950
    },
    {
      "epoch": 0.7860020265427675,
      "grad_norm": 4.236232256129904,
      "learning_rate": 2.306269222643904e-08,
      "loss": 0.6852,
      "step": 7951
    },
    {
      "epoch": 0.786100882287522,
      "grad_norm": 5.342598569692525,
      "learning_rate": 2.3042241085354054e-08,
      "loss": 0.7738,
      "step": 7952
    },
    {
      "epoch": 0.7861997380322764,
      "grad_norm": 3.998224217778554,
      "learning_rate": 2.3021797835152014e-08,
      "loss": 0.7597,
      "step": 7953
    },
    {
      "epoch": 0.7862985937770308,
      "grad_norm": 2.9681612838234797,
      "learning_rate": 2.300136247792913e-08,
      "loss": 0.5823,
      "step": 7954
    },
    {
      "epoch": 0.7863974495217854,
      "grad_norm": 5.625524409358926,
      "learning_rate": 2.298093501578068e-08,
      "loss": 0.6348,
      "step": 7955
    },
    {
      "epoch": 0.7864963052665398,
      "grad_norm": 4.000476782769655,
      "learning_rate": 2.2960515450801242e-08,
      "loss": 0.67,
      "step": 7956
    },
    {
      "epoch": 0.7865951610112942,
      "grad_norm": 27.433428244556023,
      "learning_rate": 2.2940103785084553e-08,
      "loss": 0.7078,
      "step": 7957
    },
    {
      "epoch": 0.7866940167560488,
      "grad_norm": 4.41186366678544,
      "learning_rate": 2.2919700020723475e-08,
      "loss": 0.8276,
      "step": 7958
    },
    {
      "epoch": 0.7867928725008032,
      "grad_norm": 4.341287447159085,
      "learning_rate": 2.2899304159810153e-08,
      "loss": 0.7159,
      "step": 7959
    },
    {
      "epoch": 0.7868917282455576,
      "grad_norm": 10.948784268806152,
      "learning_rate": 2.2878916204435904e-08,
      "loss": 0.7844,
      "step": 7960
    },
    {
      "epoch": 0.7869905839903122,
      "grad_norm": 3.2389869778020453,
      "learning_rate": 2.2858536156691143e-08,
      "loss": 0.7001,
      "step": 7961
    },
    {
      "epoch": 0.7870894397350666,
      "grad_norm": 3.33751696309188,
      "learning_rate": 2.2838164018665597e-08,
      "loss": 0.6745,
      "step": 7962
    },
    {
      "epoch": 0.787188295479821,
      "grad_norm": 4.115244278942007,
      "learning_rate": 2.2817799792448134e-08,
      "loss": 0.5759,
      "step": 7963
    },
    {
      "epoch": 0.7872871512245755,
      "grad_norm": 3.1788366146536347,
      "learning_rate": 2.2797443480126754e-08,
      "loss": 0.7119,
      "step": 7964
    },
    {
      "epoch": 0.78738600696933,
      "grad_norm": 16.756997493328615,
      "learning_rate": 2.2777095083788754e-08,
      "loss": 0.6871,
      "step": 7965
    },
    {
      "epoch": 0.7874848627140845,
      "grad_norm": 3.188214031370651,
      "learning_rate": 2.2756754605520545e-08,
      "loss": 0.6604,
      "step": 7966
    },
    {
      "epoch": 0.7875837184588389,
      "grad_norm": 3.8184620351545555,
      "learning_rate": 2.27364220474077e-08,
      "loss": 0.732,
      "step": 7967
    },
    {
      "epoch": 0.7876825742035934,
      "grad_norm": 3.855626198529875,
      "learning_rate": 2.2716097411535084e-08,
      "loss": 0.7148,
      "step": 7968
    },
    {
      "epoch": 0.7877814299483479,
      "grad_norm": 4.078361034788952,
      "learning_rate": 2.2695780699986634e-08,
      "loss": 0.5817,
      "step": 7969
    },
    {
      "epoch": 0.7878802856931023,
      "grad_norm": 8.010814655795262,
      "learning_rate": 2.2675471914845555e-08,
      "loss": 0.7712,
      "step": 7970
    },
    {
      "epoch": 0.7879791414378569,
      "grad_norm": 3.758006897735093,
      "learning_rate": 2.2655171058194232e-08,
      "loss": 0.715,
      "step": 7971
    },
    {
      "epoch": 0.7880779971826113,
      "grad_norm": 6.823619081569287,
      "learning_rate": 2.263487813211419e-08,
      "loss": 0.6903,
      "step": 7972
    },
    {
      "epoch": 0.7881768529273657,
      "grad_norm": 4.611349601855551,
      "learning_rate": 2.2614593138686167e-08,
      "loss": 0.7746,
      "step": 7973
    },
    {
      "epoch": 0.7882757086721202,
      "grad_norm": 5.571814339614893,
      "learning_rate": 2.259431607999014e-08,
      "loss": 0.6791,
      "step": 7974
    },
    {
      "epoch": 0.7883745644168747,
      "grad_norm": 7.199250654338662,
      "learning_rate": 2.2574046958105154e-08,
      "loss": 0.6552,
      "step": 7975
    },
    {
      "epoch": 0.7884734201616291,
      "grad_norm": 3.6545987502297104,
      "learning_rate": 2.255378577510956e-08,
      "loss": 0.7622,
      "step": 7976
    },
    {
      "epoch": 0.7885722759063836,
      "grad_norm": 4.811301527382787,
      "learning_rate": 2.2533532533080802e-08,
      "loss": 0.6054,
      "step": 7977
    },
    {
      "epoch": 0.7886711316511381,
      "grad_norm": 4.8765034076708575,
      "learning_rate": 2.2513287234095568e-08,
      "loss": 0.6655,
      "step": 7978
    },
    {
      "epoch": 0.7887699873958925,
      "grad_norm": 3.741553624909421,
      "learning_rate": 2.2493049880229754e-08,
      "loss": 0.6305,
      "step": 7979
    },
    {
      "epoch": 0.788868843140647,
      "grad_norm": 15.733397774190555,
      "learning_rate": 2.247282047355833e-08,
      "loss": 0.7941,
      "step": 7980
    },
    {
      "epoch": 0.7889676988854015,
      "grad_norm": 20.5054392399232,
      "learning_rate": 2.245259901615556e-08,
      "loss": 0.6917,
      "step": 7981
    },
    {
      "epoch": 0.789066554630156,
      "grad_norm": 2.923224560262284,
      "learning_rate": 2.2432385510094886e-08,
      "loss": 0.6864,
      "step": 7982
    },
    {
      "epoch": 0.7891654103749104,
      "grad_norm": 4.319365478203326,
      "learning_rate": 2.2412179957448873e-08,
      "loss": 0.7587,
      "step": 7983
    },
    {
      "epoch": 0.7892642661196648,
      "grad_norm": 4.0825688442557055,
      "learning_rate": 2.2391982360289286e-08,
      "loss": 0.6798,
      "step": 7984
    },
    {
      "epoch": 0.7893631218644194,
      "grad_norm": 3.7058223886823964,
      "learning_rate": 2.2371792720687134e-08,
      "loss": 0.6221,
      "step": 7985
    },
    {
      "epoch": 0.7894619776091738,
      "grad_norm": 4.697623782339271,
      "learning_rate": 2.2351611040712502e-08,
      "loss": 0.7228,
      "step": 7986
    },
    {
      "epoch": 0.7895608333539282,
      "grad_norm": 30.67986443614289,
      "learning_rate": 2.233143732243481e-08,
      "loss": 0.5827,
      "step": 7987
    },
    {
      "epoch": 0.7896596890986828,
      "grad_norm": 5.224985591619126,
      "learning_rate": 2.2311271567922486e-08,
      "loss": 0.7818,
      "step": 7988
    },
    {
      "epoch": 0.7897585448434372,
      "grad_norm": 3.8320025054831097,
      "learning_rate": 2.2291113779243276e-08,
      "loss": 0.671,
      "step": 7989
    },
    {
      "epoch": 0.7898574005881916,
      "grad_norm": 37.16870983121721,
      "learning_rate": 2.2270963958464095e-08,
      "loss": 0.7273,
      "step": 7990
    },
    {
      "epoch": 0.7899562563329462,
      "grad_norm": 4.2490256535518265,
      "learning_rate": 2.2250822107650947e-08,
      "loss": 0.7054,
      "step": 7991
    },
    {
      "epoch": 0.7900551120777006,
      "grad_norm": 5.355763337700417,
      "learning_rate": 2.223068822886911e-08,
      "loss": 0.7593,
      "step": 7992
    },
    {
      "epoch": 0.7901539678224551,
      "grad_norm": 8.449043213625671,
      "learning_rate": 2.2210562324183046e-08,
      "loss": 0.8008,
      "step": 7993
    },
    {
      "epoch": 0.7902528235672095,
      "grad_norm": 3.1969202069638922,
      "learning_rate": 2.2190444395656304e-08,
      "loss": 0.7921,
      "step": 7994
    },
    {
      "epoch": 0.790351679311964,
      "grad_norm": 3.560845240894026,
      "learning_rate": 2.2170334445351722e-08,
      "loss": 0.8001,
      "step": 7995
    },
    {
      "epoch": 0.7904505350567185,
      "grad_norm": 6.1252900465398525,
      "learning_rate": 2.215023247533131e-08,
      "loss": 0.7569,
      "step": 7996
    },
    {
      "epoch": 0.7905493908014729,
      "grad_norm": 2.90522990305727,
      "learning_rate": 2.213013848765616e-08,
      "loss": 0.7044,
      "step": 7997
    },
    {
      "epoch": 0.7906482465462275,
      "grad_norm": 4.984670803789874,
      "learning_rate": 2.2110052484386677e-08,
      "loss": 0.7538,
      "step": 7998
    },
    {
      "epoch": 0.7907471022909819,
      "grad_norm": 11.744281145178316,
      "learning_rate": 2.2089974467582317e-08,
      "loss": 0.6442,
      "step": 7999
    },
    {
      "epoch": 0.7908459580357363,
      "grad_norm": 4.175585473705624,
      "learning_rate": 2.2069904439301857e-08,
      "loss": 0.6857,
      "step": 8000
    },
    {
      "epoch": 0.7909448137804909,
      "grad_norm": 6.671109255896544,
      "learning_rate": 2.2049842401603114e-08,
      "loss": 0.6932,
      "step": 8001
    },
    {
      "epoch": 0.7910436695252453,
      "grad_norm": 5.018828823924566,
      "learning_rate": 2.202978835654322e-08,
      "loss": 0.7712,
      "step": 8002
    },
    {
      "epoch": 0.7911425252699997,
      "grad_norm": 4.374935048770241,
      "learning_rate": 2.200974230617836e-08,
      "loss": 0.6862,
      "step": 8003
    },
    {
      "epoch": 0.7912413810147543,
      "grad_norm": 4.683133690992875,
      "learning_rate": 2.198970425256401e-08,
      "loss": 0.6377,
      "step": 8004
    },
    {
      "epoch": 0.7913402367595087,
      "grad_norm": 6.027326559440247,
      "learning_rate": 2.196967419775474e-08,
      "loss": 0.6484,
      "step": 8005
    },
    {
      "epoch": 0.7914390925042631,
      "grad_norm": 13.081244321430484,
      "learning_rate": 2.1949652143804353e-08,
      "loss": 0.7065,
      "step": 8006
    },
    {
      "epoch": 0.7915379482490176,
      "grad_norm": 3.6743067014600075,
      "learning_rate": 2.1929638092765833e-08,
      "loss": 0.6428,
      "step": 8007
    },
    {
      "epoch": 0.7916368039937721,
      "grad_norm": 6.662122603116459,
      "learning_rate": 2.1909632046691286e-08,
      "loss": 0.7678,
      "step": 8008
    },
    {
      "epoch": 0.7917356597385266,
      "grad_norm": 4.004078300757308,
      "learning_rate": 2.1889634007632086e-08,
      "loss": 0.6403,
      "step": 8009
    },
    {
      "epoch": 0.791834515483281,
      "grad_norm": 7.335553535180043,
      "learning_rate": 2.1869643977638673e-08,
      "loss": 0.6983,
      "step": 8010
    },
    {
      "epoch": 0.7919333712280355,
      "grad_norm": 6.744684942020077,
      "learning_rate": 2.184966195876078e-08,
      "loss": 0.6596,
      "step": 8011
    },
    {
      "epoch": 0.79203222697279,
      "grad_norm": 4.89132616212019,
      "learning_rate": 2.182968795304727e-08,
      "loss": 0.7001,
      "step": 8012
    },
    {
      "epoch": 0.7921310827175444,
      "grad_norm": 4.884987268045727,
      "learning_rate": 2.1809721962546147e-08,
      "loss": 0.6913,
      "step": 8013
    },
    {
      "epoch": 0.7922299384622989,
      "grad_norm": 10.45320160068239,
      "learning_rate": 2.1789763989304643e-08,
      "loss": 0.7488,
      "step": 8014
    },
    {
      "epoch": 0.7923287942070534,
      "grad_norm": 6.647304119100279,
      "learning_rate": 2.1769814035369193e-08,
      "loss": 0.7222,
      "step": 8015
    },
    {
      "epoch": 0.7924276499518078,
      "grad_norm": 6.219794983380087,
      "learning_rate": 2.1749872102785316e-08,
      "loss": 0.5973,
      "step": 8016
    },
    {
      "epoch": 0.7925265056965622,
      "grad_norm": 9.390846267667538,
      "learning_rate": 2.1729938193597798e-08,
      "loss": 0.6158,
      "step": 8017
    },
    {
      "epoch": 0.7926253614413168,
      "grad_norm": 4.23259132486889,
      "learning_rate": 2.1710012309850533e-08,
      "loss": 0.5986,
      "step": 8018
    },
    {
      "epoch": 0.7927242171860712,
      "grad_norm": 4.428209106563973,
      "learning_rate": 2.169009445358667e-08,
      "loss": 0.6677,
      "step": 8019
    },
    {
      "epoch": 0.7928230729308257,
      "grad_norm": 3.782525782249729,
      "learning_rate": 2.1670184626848477e-08,
      "loss": 0.7392,
      "step": 8020
    },
    {
      "epoch": 0.7929219286755802,
      "grad_norm": 3.3784302891183473,
      "learning_rate": 2.1650282831677367e-08,
      "loss": 0.6603,
      "step": 8021
    },
    {
      "epoch": 0.7930207844203346,
      "grad_norm": 5.772061568954741,
      "learning_rate": 2.163038907011402e-08,
      "loss": 0.5832,
      "step": 8022
    },
    {
      "epoch": 0.7931196401650891,
      "grad_norm": 5.244454350414224,
      "learning_rate": 2.161050334419827e-08,
      "loss": 0.5931,
      "step": 8023
    },
    {
      "epoch": 0.7932184959098436,
      "grad_norm": 5.773776029106545,
      "learning_rate": 2.1590625655969043e-08,
      "loss": 0.7465,
      "step": 8024
    },
    {
      "epoch": 0.793317351654598,
      "grad_norm": 3.2557435661991287,
      "learning_rate": 2.157075600746453e-08,
      "loss": 0.6085,
      "step": 8025
    },
    {
      "epoch": 0.7934162073993525,
      "grad_norm": 5.225229002193835,
      "learning_rate": 2.155089440072212e-08,
      "loss": 0.6335,
      "step": 8026
    },
    {
      "epoch": 0.7935150631441069,
      "grad_norm": 4.152833505454176,
      "learning_rate": 2.1531040837778234e-08,
      "loss": 0.7487,
      "step": 8027
    },
    {
      "epoch": 0.7936139188888615,
      "grad_norm": 6.424456164882012,
      "learning_rate": 2.151119532066863e-08,
      "loss": 0.7194,
      "step": 8028
    },
    {
      "epoch": 0.7937127746336159,
      "grad_norm": 4.008157264099073,
      "learning_rate": 2.149135785142817e-08,
      "loss": 0.6901,
      "step": 8029
    },
    {
      "epoch": 0.7938116303783703,
      "grad_norm": 2.9733637686769843,
      "learning_rate": 2.1471528432090856e-08,
      "loss": 0.7102,
      "step": 8030
    },
    {
      "epoch": 0.7939104861231249,
      "grad_norm": 3.875086120823478,
      "learning_rate": 2.1451707064689963e-08,
      "loss": 0.7404,
      "step": 8031
    },
    {
      "epoch": 0.7940093418678793,
      "grad_norm": 3.309873499276737,
      "learning_rate": 2.1431893751257802e-08,
      "loss": 0.7221,
      "step": 8032
    },
    {
      "epoch": 0.7941081976126337,
      "grad_norm": 5.727981786135744,
      "learning_rate": 2.1412088493825974e-08,
      "loss": 0.7539,
      "step": 8033
    },
    {
      "epoch": 0.7942070533573883,
      "grad_norm": 38.26110321206332,
      "learning_rate": 2.1392291294425245e-08,
      "loss": 0.6302,
      "step": 8034
    },
    {
      "epoch": 0.7943059091021427,
      "grad_norm": 3.9731071033035574,
      "learning_rate": 2.137250215508548e-08,
      "loss": 0.6897,
      "step": 8035
    },
    {
      "epoch": 0.7944047648468971,
      "grad_norm": 2.901708803360247,
      "learning_rate": 2.13527210778358e-08,
      "loss": 0.6879,
      "step": 8036
    },
    {
      "epoch": 0.7945036205916516,
      "grad_norm": 3.530732500312674,
      "learning_rate": 2.133294806470445e-08,
      "loss": 0.6599,
      "step": 8037
    },
    {
      "epoch": 0.7946024763364061,
      "grad_norm": 4.444139872083399,
      "learning_rate": 2.1313183117718825e-08,
      "loss": 0.8117,
      "step": 8038
    },
    {
      "epoch": 0.7947013320811606,
      "grad_norm": 5.049486215810018,
      "learning_rate": 2.1293426238905554e-08,
      "loss": 0.793,
      "step": 8039
    },
    {
      "epoch": 0.794800187825915,
      "grad_norm": 3.3138905766336193,
      "learning_rate": 2.127367743029046e-08,
      "loss": 0.7655,
      "step": 8040
    },
    {
      "epoch": 0.7948990435706695,
      "grad_norm": 3.9109465963994645,
      "learning_rate": 2.1253936693898412e-08,
      "loss": 0.7563,
      "step": 8041
    },
    {
      "epoch": 0.794997899315424,
      "grad_norm": 67.3712578921352,
      "learning_rate": 2.12342040317536e-08,
      "loss": 0.653,
      "step": 8042
    },
    {
      "epoch": 0.7950967550601784,
      "grad_norm": 14.243264535549752,
      "learning_rate": 2.1214479445879263e-08,
      "loss": 0.7728,
      "step": 8043
    },
    {
      "epoch": 0.795195610804933,
      "grad_norm": 6.049394722691088,
      "learning_rate": 2.1194762938297884e-08,
      "loss": 0.8142,
      "step": 8044
    },
    {
      "epoch": 0.7952944665496874,
      "grad_norm": 9.18411710478427,
      "learning_rate": 2.1175054511031144e-08,
      "loss": 0.7855,
      "step": 8045
    },
    {
      "epoch": 0.7953933222944418,
      "grad_norm": 3.5867285444547616,
      "learning_rate": 2.1155354166099783e-08,
      "loss": 0.7274,
      "step": 8046
    },
    {
      "epoch": 0.7954921780391963,
      "grad_norm": 5.828503104307514,
      "learning_rate": 2.1135661905523804e-08,
      "loss": 0.7112,
      "step": 8047
    },
    {
      "epoch": 0.7955910337839508,
      "grad_norm": 6.116947221754191,
      "learning_rate": 2.1115977731322398e-08,
      "loss": 0.6881,
      "step": 8048
    },
    {
      "epoch": 0.7956898895287052,
      "grad_norm": 4.662552406288064,
      "learning_rate": 2.1096301645513826e-08,
      "loss": 0.7266,
      "step": 8049
    },
    {
      "epoch": 0.7957887452734597,
      "grad_norm": 4.314948749707543,
      "learning_rate": 2.1076633650115617e-08,
      "loss": 0.5832,
      "step": 8050
    },
    {
      "epoch": 0.7958876010182142,
      "grad_norm": 3.2581541910745324,
      "learning_rate": 2.1056973747144445e-08,
      "loss": 0.6591,
      "step": 8051
    },
    {
      "epoch": 0.7959864567629686,
      "grad_norm": 19.010837937080357,
      "learning_rate": 2.1037321938616093e-08,
      "loss": 0.8007,
      "step": 8052
    },
    {
      "epoch": 0.7960853125077231,
      "grad_norm": 12.149160451127367,
      "learning_rate": 2.1017678226545632e-08,
      "loss": 0.6793,
      "step": 8053
    },
    {
      "epoch": 0.7961841682524776,
      "grad_norm": 3.595750101527138,
      "learning_rate": 2.0998042612947187e-08,
      "loss": 0.6897,
      "step": 8054
    },
    {
      "epoch": 0.796283023997232,
      "grad_norm": 3.5330965799129967,
      "learning_rate": 2.097841509983409e-08,
      "loss": 0.7404,
      "step": 8055
    },
    {
      "epoch": 0.7963818797419865,
      "grad_norm": 3.6692030257104937,
      "learning_rate": 2.09587956892189e-08,
      "loss": 0.7822,
      "step": 8056
    },
    {
      "epoch": 0.7964807354867409,
      "grad_norm": 8.942758370329562,
      "learning_rate": 2.0939184383113252e-08,
      "loss": 0.726,
      "step": 8057
    },
    {
      "epoch": 0.7965795912314955,
      "grad_norm": 11.722889513971337,
      "learning_rate": 2.0919581183528012e-08,
      "loss": 0.7415,
      "step": 8058
    },
    {
      "epoch": 0.7966784469762499,
      "grad_norm": 34.11785209571026,
      "learning_rate": 2.089998609247322e-08,
      "loss": 0.7142,
      "step": 8059
    },
    {
      "epoch": 0.7967773027210043,
      "grad_norm": 4.408576341054688,
      "learning_rate": 2.0880399111958036e-08,
      "loss": 0.666,
      "step": 8060
    },
    {
      "epoch": 0.7968761584657589,
      "grad_norm": 3.3745661865778627,
      "learning_rate": 2.0860820243990818e-08,
      "loss": 0.6888,
      "step": 8061
    },
    {
      "epoch": 0.7969750142105133,
      "grad_norm": 4.319421893654931,
      "learning_rate": 2.0841249490579127e-08,
      "loss": 0.73,
      "step": 8062
    },
    {
      "epoch": 0.7970738699552677,
      "grad_norm": 4.41272352931148,
      "learning_rate": 2.08216868537296e-08,
      "loss": 0.7473,
      "step": 8063
    },
    {
      "epoch": 0.7971727257000223,
      "grad_norm": 13.401357850818037,
      "learning_rate": 2.0802132335448154e-08,
      "loss": 0.6473,
      "step": 8064
    },
    {
      "epoch": 0.7972715814447767,
      "grad_norm": 3.929366975149527,
      "learning_rate": 2.078258593773975e-08,
      "loss": 0.5866,
      "step": 8065
    },
    {
      "epoch": 0.7973704371895312,
      "grad_norm": 3.722015365338939,
      "learning_rate": 2.0763047662608623e-08,
      "loss": 0.6099,
      "step": 8066
    },
    {
      "epoch": 0.7974692929342856,
      "grad_norm": 5.199326734944557,
      "learning_rate": 2.0743517512058163e-08,
      "loss": 0.7046,
      "step": 8067
    },
    {
      "epoch": 0.7975681486790401,
      "grad_norm": 3.6675757830585813,
      "learning_rate": 2.072399548809084e-08,
      "loss": 0.6662,
      "step": 8068
    },
    {
      "epoch": 0.7976670044237946,
      "grad_norm": 8.062979898629724,
      "learning_rate": 2.070448159270838e-08,
      "loss": 0.6765,
      "step": 8069
    },
    {
      "epoch": 0.797765860168549,
      "grad_norm": 5.0576434118092655,
      "learning_rate": 2.0684975827911656e-08,
      "loss": 0.6662,
      "step": 8070
    },
    {
      "epoch": 0.7978647159133035,
      "grad_norm": 5.180636850437206,
      "learning_rate": 2.0665478195700693e-08,
      "loss": 0.8115,
      "step": 8071
    },
    {
      "epoch": 0.797963571658058,
      "grad_norm": 4.4458296156617525,
      "learning_rate": 2.0645988698074634e-08,
      "loss": 0.7213,
      "step": 8072
    },
    {
      "epoch": 0.7980624274028124,
      "grad_norm": 3.528763287094208,
      "learning_rate": 2.0626507337031917e-08,
      "loss": 0.6119,
      "step": 8073
    },
    {
      "epoch": 0.798161283147567,
      "grad_norm": 3.274028460644096,
      "learning_rate": 2.060703411457001e-08,
      "loss": 0.6698,
      "step": 8074
    },
    {
      "epoch": 0.7982601388923214,
      "grad_norm": 30.180095009728458,
      "learning_rate": 2.0587569032685635e-08,
      "loss": 0.6384,
      "step": 8075
    },
    {
      "epoch": 0.7983589946370758,
      "grad_norm": 3.670248011518941,
      "learning_rate": 2.0568112093374634e-08,
      "loss": 0.8001,
      "step": 8076
    },
    {
      "epoch": 0.7984578503818303,
      "grad_norm": 2.7472876690875454,
      "learning_rate": 2.054866329863202e-08,
      "loss": 0.7215,
      "step": 8077
    },
    {
      "epoch": 0.7985567061265848,
      "grad_norm": 2.9660745758187725,
      "learning_rate": 2.0529222650452028e-08,
      "loss": 0.7297,
      "step": 8078
    },
    {
      "epoch": 0.7986555618713392,
      "grad_norm": 4.195381394223327,
      "learning_rate": 2.0509790150827943e-08,
      "loss": 0.7175,
      "step": 8079
    },
    {
      "epoch": 0.7987544176160937,
      "grad_norm": 4.17238074415259,
      "learning_rate": 2.0490365801752317e-08,
      "loss": 0.7221,
      "step": 8080
    },
    {
      "epoch": 0.7988532733608482,
      "grad_norm": 5.37700508777497,
      "learning_rate": 2.0470949605216868e-08,
      "loss": 0.7411,
      "step": 8081
    },
    {
      "epoch": 0.7989521291056026,
      "grad_norm": 4.118258419236742,
      "learning_rate": 2.045154156321236e-08,
      "loss": 0.7819,
      "step": 8082
    },
    {
      "epoch": 0.7990509848503571,
      "grad_norm": 3.643008844472692,
      "learning_rate": 2.0432141677728855e-08,
      "loss": 0.6822,
      "step": 8083
    },
    {
      "epoch": 0.7991498405951116,
      "grad_norm": 4.274867072538961,
      "learning_rate": 2.0412749950755525e-08,
      "loss": 0.8026,
      "step": 8084
    },
    {
      "epoch": 0.7992486963398661,
      "grad_norm": 3.544116648031486,
      "learning_rate": 2.0393366384280674e-08,
      "loss": 0.7935,
      "step": 8085
    },
    {
      "epoch": 0.7993475520846205,
      "grad_norm": 4.304452531980691,
      "learning_rate": 2.0373990980291844e-08,
      "loss": 0.6773,
      "step": 8086
    },
    {
      "epoch": 0.799446407829375,
      "grad_norm": 4.01816693830789,
      "learning_rate": 2.0354623740775643e-08,
      "loss": 0.644,
      "step": 8087
    },
    {
      "epoch": 0.7995452635741295,
      "grad_norm": 3.555294405147759,
      "learning_rate": 2.0335264667717956e-08,
      "loss": 0.7897,
      "step": 8088
    },
    {
      "epoch": 0.7996441193188839,
      "grad_norm": 3.279926600650593,
      "learning_rate": 2.0315913763103722e-08,
      "loss": 0.7821,
      "step": 8089
    },
    {
      "epoch": 0.7997429750636383,
      "grad_norm": 6.233369017568878,
      "learning_rate": 2.029657102891712e-08,
      "loss": 0.6589,
      "step": 8090
    },
    {
      "epoch": 0.7998418308083929,
      "grad_norm": 19.2216354020714,
      "learning_rate": 2.0277236467141436e-08,
      "loss": 0.6563,
      "step": 8091
    },
    {
      "epoch": 0.7999406865531473,
      "grad_norm": 6.511785824731293,
      "learning_rate": 2.0257910079759176e-08,
      "loss": 0.7167,
      "step": 8092
    },
    {
      "epoch": 0.8000395422979018,
      "grad_norm": 6.030238248282697,
      "learning_rate": 2.023859186875193e-08,
      "loss": 0.6585,
      "step": 8093
    },
    {
      "epoch": 0.8001383980426563,
      "grad_norm": 7.466910954072572,
      "learning_rate": 2.0219281836100532e-08,
      "loss": 0.7082,
      "step": 8094
    },
    {
      "epoch": 0.8002372537874107,
      "grad_norm": 3.5697836165587757,
      "learning_rate": 2.0199979983784966e-08,
      "loss": 0.7237,
      "step": 8095
    },
    {
      "epoch": 0.8003361095321652,
      "grad_norm": 4.426020902102568,
      "learning_rate": 2.0180686313784278e-08,
      "loss": 0.6752,
      "step": 8096
    },
    {
      "epoch": 0.8004349652769197,
      "grad_norm": 3.5160122410106163,
      "learning_rate": 2.0161400828076826e-08,
      "loss": 0.5937,
      "step": 8097
    },
    {
      "epoch": 0.8005338210216741,
      "grad_norm": 4.762634656978886,
      "learning_rate": 2.0142123528639988e-08,
      "loss": 0.7138,
      "step": 8098
    },
    {
      "epoch": 0.8006326767664286,
      "grad_norm": 4.09055842820674,
      "learning_rate": 2.0122854417450398e-08,
      "loss": 0.747,
      "step": 8099
    },
    {
      "epoch": 0.800731532511183,
      "grad_norm": 4.189456518534083,
      "learning_rate": 2.010359349648384e-08,
      "loss": 0.7289,
      "step": 8100
    },
    {
      "epoch": 0.8008303882559376,
      "grad_norm": 3.857438340912207,
      "learning_rate": 2.008434076771519e-08,
      "loss": 0.6649,
      "step": 8101
    },
    {
      "epoch": 0.800929244000692,
      "grad_norm": 4.625002840569425,
      "learning_rate": 2.006509623311856e-08,
      "loss": 0.6313,
      "step": 8102
    },
    {
      "epoch": 0.8010280997454464,
      "grad_norm": 3.108123766184425,
      "learning_rate": 2.0045859894667206e-08,
      "loss": 0.6706,
      "step": 8103
    },
    {
      "epoch": 0.801126955490201,
      "grad_norm": 3.005111633933565,
      "learning_rate": 2.002663175433349e-08,
      "loss": 0.6397,
      "step": 8104
    },
    {
      "epoch": 0.8012258112349554,
      "grad_norm": 3.9238905838901523,
      "learning_rate": 2.0007411814089027e-08,
      "loss": 0.756,
      "step": 8105
    },
    {
      "epoch": 0.8013246669797098,
      "grad_norm": 4.996409609996933,
      "learning_rate": 1.9988200075904483e-08,
      "loss": 0.7001,
      "step": 8106
    },
    {
      "epoch": 0.8014235227244644,
      "grad_norm": 2.6206739923307087,
      "learning_rate": 1.9968996541749783e-08,
      "loss": 0.6713,
      "step": 8107
    },
    {
      "epoch": 0.8015223784692188,
      "grad_norm": 5.005501281655616,
      "learning_rate": 1.9949801213593952e-08,
      "loss": 0.8405,
      "step": 8108
    },
    {
      "epoch": 0.8016212342139732,
      "grad_norm": 5.714842396372804,
      "learning_rate": 1.9930614093405153e-08,
      "loss": 0.6252,
      "step": 8109
    },
    {
      "epoch": 0.8017200899587277,
      "grad_norm": 3.2964230152860847,
      "learning_rate": 1.991143518315077e-08,
      "loss": 0.796,
      "step": 8110
    },
    {
      "epoch": 0.8018189457034822,
      "grad_norm": 4.509164692423234,
      "learning_rate": 1.989226448479736e-08,
      "loss": 0.7527,
      "step": 8111
    },
    {
      "epoch": 0.8019178014482367,
      "grad_norm": 17.58630934442817,
      "learning_rate": 1.9873102000310515e-08,
      "loss": 0.6999,
      "step": 8112
    },
    {
      "epoch": 0.8020166571929911,
      "grad_norm": 2.89555627043851,
      "learning_rate": 1.985394773165511e-08,
      "loss": 0.7395,
      "step": 8113
    },
    {
      "epoch": 0.8021155129377456,
      "grad_norm": 5.208982929334602,
      "learning_rate": 1.9834801680795156e-08,
      "loss": 0.8391,
      "step": 8114
    },
    {
      "epoch": 0.8022143686825001,
      "grad_norm": 6.857070521786657,
      "learning_rate": 1.981566384969374e-08,
      "loss": 0.8851,
      "step": 8115
    },
    {
      "epoch": 0.8023132244272545,
      "grad_norm": 4.747328370063339,
      "learning_rate": 1.979653424031319e-08,
      "loss": 0.7525,
      "step": 8116
    },
    {
      "epoch": 0.802412080172009,
      "grad_norm": 3.166068649909162,
      "learning_rate": 1.9777412854615005e-08,
      "loss": 0.6254,
      "step": 8117
    },
    {
      "epoch": 0.8025109359167635,
      "grad_norm": 3.6454930191087884,
      "learning_rate": 1.975829969455973e-08,
      "loss": 0.7109,
      "step": 8118
    },
    {
      "epoch": 0.8026097916615179,
      "grad_norm": 6.320274982021155,
      "learning_rate": 1.9739194762107202e-08,
      "loss": 0.6239,
      "step": 8119
    },
    {
      "epoch": 0.8027086474062723,
      "grad_norm": 8.968933846069138,
      "learning_rate": 1.9720098059216307e-08,
      "loss": 0.6738,
      "step": 8120
    },
    {
      "epoch": 0.8028075031510269,
      "grad_norm": 5.536111403605802,
      "learning_rate": 1.970100958784513e-08,
      "loss": 0.7811,
      "step": 8121
    },
    {
      "epoch": 0.8029063588957813,
      "grad_norm": 3.3590779422663264,
      "learning_rate": 1.9681929349950955e-08,
      "loss": 0.676,
      "step": 8122
    },
    {
      "epoch": 0.8030052146405358,
      "grad_norm": 4.470301323728995,
      "learning_rate": 1.9662857347490124e-08,
      "loss": 0.6292,
      "step": 8123
    },
    {
      "epoch": 0.8031040703852903,
      "grad_norm": 4.597226970557335,
      "learning_rate": 1.9643793582418244e-08,
      "loss": 0.7593,
      "step": 8124
    },
    {
      "epoch": 0.8032029261300447,
      "grad_norm": 3.2212151211836693,
      "learning_rate": 1.9624738056689994e-08,
      "loss": 0.757,
      "step": 8125
    },
    {
      "epoch": 0.8033017818747992,
      "grad_norm": 9.927715332984322,
      "learning_rate": 1.960569077225921e-08,
      "loss": 0.7206,
      "step": 8126
    },
    {
      "epoch": 0.8034006376195537,
      "grad_norm": 10.386715619595421,
      "learning_rate": 1.958665173107894e-08,
      "loss": 0.6704,
      "step": 8127
    },
    {
      "epoch": 0.8034994933643081,
      "grad_norm": 4.490359632699761,
      "learning_rate": 1.956762093510138e-08,
      "loss": 0.6756,
      "step": 8128
    },
    {
      "epoch": 0.8035983491090626,
      "grad_norm": 5.533786778150065,
      "learning_rate": 1.9548598386277804e-08,
      "loss": 0.7621,
      "step": 8129
    },
    {
      "epoch": 0.803697204853817,
      "grad_norm": 4.937049909336295,
      "learning_rate": 1.9529584086558747e-08,
      "loss": 0.6913,
      "step": 8130
    },
    {
      "epoch": 0.8037960605985716,
      "grad_norm": 4.284610871927281,
      "learning_rate": 1.95105780378938e-08,
      "loss": 0.777,
      "step": 8131
    },
    {
      "epoch": 0.803894916343326,
      "grad_norm": 2.8599050318870924,
      "learning_rate": 1.9491580242231764e-08,
      "loss": 0.7054,
      "step": 8132
    },
    {
      "epoch": 0.8039937720880804,
      "grad_norm": 5.659455328997459,
      "learning_rate": 1.947259070152063e-08,
      "loss": 0.7568,
      "step": 8133
    },
    {
      "epoch": 0.804092627832835,
      "grad_norm": 3.6799102402691752,
      "learning_rate": 1.9453609417707417e-08,
      "loss": 0.6872,
      "step": 8134
    },
    {
      "epoch": 0.8041914835775894,
      "grad_norm": 3.49143131373834,
      "learning_rate": 1.9434636392738424e-08,
      "loss": 0.6436,
      "step": 8135
    },
    {
      "epoch": 0.8042903393223438,
      "grad_norm": 3.13834011693585,
      "learning_rate": 1.941567162855907e-08,
      "loss": 0.6957,
      "step": 8136
    },
    {
      "epoch": 0.8043891950670984,
      "grad_norm": 4.853966902434128,
      "learning_rate": 1.939671512711386e-08,
      "loss": 0.6408,
      "step": 8137
    },
    {
      "epoch": 0.8044880508118528,
      "grad_norm": 4.791482849456619,
      "learning_rate": 1.9377766890346538e-08,
      "loss": 0.7154,
      "step": 8138
    },
    {
      "epoch": 0.8045869065566073,
      "grad_norm": 4.126254003317854,
      "learning_rate": 1.9358826920199976e-08,
      "loss": 0.6496,
      "step": 8139
    },
    {
      "epoch": 0.8046857623013617,
      "grad_norm": 13.911987775063885,
      "learning_rate": 1.9339895218616186e-08,
      "loss": 0.6047,
      "step": 8140
    },
    {
      "epoch": 0.8047846180461162,
      "grad_norm": 3.4689454434307425,
      "learning_rate": 1.9320971787536276e-08,
      "loss": 0.8412,
      "step": 8141
    },
    {
      "epoch": 0.8048834737908707,
      "grad_norm": 3.345980641525808,
      "learning_rate": 1.9302056628900653e-08,
      "loss": 0.6865,
      "step": 8142
    },
    {
      "epoch": 0.8049823295356251,
      "grad_norm": 3.0140554749876562,
      "learning_rate": 1.928314974464871e-08,
      "loss": 0.7132,
      "step": 8143
    },
    {
      "epoch": 0.8050811852803796,
      "grad_norm": 3.450757274384056,
      "learning_rate": 1.9264251136719135e-08,
      "loss": 0.6339,
      "step": 8144
    },
    {
      "epoch": 0.8051800410251341,
      "grad_norm": 3.91145694498434,
      "learning_rate": 1.9245360807049637e-08,
      "loss": 0.747,
      "step": 8145
    },
    {
      "epoch": 0.8052788967698885,
      "grad_norm": 4.077156308594188,
      "learning_rate": 1.922647875757718e-08,
      "loss": 0.6584,
      "step": 8146
    },
    {
      "epoch": 0.805377752514643,
      "grad_norm": 3.435731444071695,
      "learning_rate": 1.9207604990237858e-08,
      "loss": 0.6942,
      "step": 8147
    },
    {
      "epoch": 0.8054766082593975,
      "grad_norm": 4.03687652878411,
      "learning_rate": 1.9188739506966835e-08,
      "loss": 0.7265,
      "step": 8148
    },
    {
      "epoch": 0.8055754640041519,
      "grad_norm": 3.9164971357718,
      "learning_rate": 1.9169882309698537e-08,
      "loss": 0.7718,
      "step": 8149
    },
    {
      "epoch": 0.8056743197489064,
      "grad_norm": 2.6830858016985486,
      "learning_rate": 1.9151033400366502e-08,
      "loss": 0.6243,
      "step": 8150
    },
    {
      "epoch": 0.8057731754936609,
      "grad_norm": 5.916403410851465,
      "learning_rate": 1.913219278090337e-08,
      "loss": 0.7133,
      "step": 8151
    },
    {
      "epoch": 0.8058720312384153,
      "grad_norm": 3.6043494264536546,
      "learning_rate": 1.9113360453241e-08,
      "loss": 0.7657,
      "step": 8152
    },
    {
      "epoch": 0.8059708869831698,
      "grad_norm": 4.320696767191986,
      "learning_rate": 1.9094536419310324e-08,
      "loss": 0.6846,
      "step": 8153
    },
    {
      "epoch": 0.8060697427279243,
      "grad_norm": 3.1294090047000545,
      "learning_rate": 1.9075720681041507e-08,
      "loss": 0.6783,
      "step": 8154
    },
    {
      "epoch": 0.8061685984726787,
      "grad_norm": 4.020975441216155,
      "learning_rate": 1.9056913240363835e-08,
      "loss": 0.6656,
      "step": 8155
    },
    {
      "epoch": 0.8062674542174332,
      "grad_norm": 4.885802947011346,
      "learning_rate": 1.9038114099205695e-08,
      "loss": 0.7113,
      "step": 8156
    },
    {
      "epoch": 0.8063663099621877,
      "grad_norm": 8.555111694530508,
      "learning_rate": 1.9019323259494714e-08,
      "loss": 0.6834,
      "step": 8157
    },
    {
      "epoch": 0.8064651657069422,
      "grad_norm": 4.856333314508178,
      "learning_rate": 1.9000540723157543e-08,
      "loss": 0.7044,
      "step": 8158
    },
    {
      "epoch": 0.8065640214516966,
      "grad_norm": 4.113343164492327,
      "learning_rate": 1.8981766492120133e-08,
      "loss": 0.6448,
      "step": 8159
    },
    {
      "epoch": 0.8066628771964511,
      "grad_norm": 18.911948160779964,
      "learning_rate": 1.896300056830744e-08,
      "loss": 0.7058,
      "step": 8160
    },
    {
      "epoch": 0.8067617329412056,
      "grad_norm": 9.360276347502339,
      "learning_rate": 1.8944242953643676e-08,
      "loss": 0.6714,
      "step": 8161
    },
    {
      "epoch": 0.80686058868596,
      "grad_norm": 3.8378025395492132,
      "learning_rate": 1.8925493650052127e-08,
      "loss": 0.7446,
      "step": 8162
    },
    {
      "epoch": 0.8069594444307144,
      "grad_norm": 3.577522924752322,
      "learning_rate": 1.8906752659455283e-08,
      "loss": 0.7303,
      "step": 8163
    },
    {
      "epoch": 0.807058300175469,
      "grad_norm": 3.2082284026510948,
      "learning_rate": 1.8888019983774726e-08,
      "loss": 0.6839,
      "step": 8164
    },
    {
      "epoch": 0.8071571559202234,
      "grad_norm": 5.704733146418734,
      "learning_rate": 1.8869295624931237e-08,
      "loss": 0.7363,
      "step": 8165
    },
    {
      "epoch": 0.8072560116649778,
      "grad_norm": 4.334902989990931,
      "learning_rate": 1.885057958484474e-08,
      "loss": 0.7036,
      "step": 8166
    },
    {
      "epoch": 0.8073548674097324,
      "grad_norm": 4.75450110571059,
      "learning_rate": 1.8831871865434246e-08,
      "loss": 0.7063,
      "step": 8167
    },
    {
      "epoch": 0.8074537231544868,
      "grad_norm": 4.324332530902224,
      "learning_rate": 1.8813172468617967e-08,
      "loss": 0.8492,
      "step": 8168
    },
    {
      "epoch": 0.8075525788992413,
      "grad_norm": 8.002785807368168,
      "learning_rate": 1.87944813963133e-08,
      "loss": 0.7189,
      "step": 8169
    },
    {
      "epoch": 0.8076514346439958,
      "grad_norm": 3.896823401794053,
      "learning_rate": 1.877579865043667e-08,
      "loss": 0.7236,
      "step": 8170
    },
    {
      "epoch": 0.8077502903887502,
      "grad_norm": 3.4020044430335328,
      "learning_rate": 1.8757124232903743e-08,
      "loss": 0.7651,
      "step": 8171
    },
    {
      "epoch": 0.8078491461335047,
      "grad_norm": 3.2813559967400243,
      "learning_rate": 1.8738458145629345e-08,
      "loss": 0.6374,
      "step": 8172
    },
    {
      "epoch": 0.8079480018782591,
      "grad_norm": 24.987478659834885,
      "learning_rate": 1.871980039052734e-08,
      "loss": 0.7074,
      "step": 8173
    },
    {
      "epoch": 0.8080468576230136,
      "grad_norm": 9.452432362472065,
      "learning_rate": 1.870115096951086e-08,
      "loss": 0.8397,
      "step": 8174
    },
    {
      "epoch": 0.8081457133677681,
      "grad_norm": 4.693398145572609,
      "learning_rate": 1.868250988449209e-08,
      "loss": 0.6917,
      "step": 8175
    },
    {
      "epoch": 0.8082445691125225,
      "grad_norm": 27.419923313480933,
      "learning_rate": 1.8663877137382432e-08,
      "loss": 0.7055,
      "step": 8176
    },
    {
      "epoch": 0.8083434248572771,
      "grad_norm": 5.093839624193209,
      "learning_rate": 1.8645252730092353e-08,
      "loss": 0.7145,
      "step": 8177
    },
    {
      "epoch": 0.8084422806020315,
      "grad_norm": 5.76758467169519,
      "learning_rate": 1.862663666453157e-08,
      "loss": 0.7011,
      "step": 8178
    },
    {
      "epoch": 0.8085411363467859,
      "grad_norm": 7.1236026205168725,
      "learning_rate": 1.8608028942608832e-08,
      "loss": 0.7308,
      "step": 8179
    },
    {
      "epoch": 0.8086399920915405,
      "grad_norm": 4.601693834473134,
      "learning_rate": 1.858942956623214e-08,
      "loss": 0.8292,
      "step": 8180
    },
    {
      "epoch": 0.8087388478362949,
      "grad_norm": 4.170198767104453,
      "learning_rate": 1.857083853730852e-08,
      "loss": 0.7412,
      "step": 8181
    },
    {
      "epoch": 0.8088377035810493,
      "grad_norm": 4.783223108454432,
      "learning_rate": 1.855225585774425e-08,
      "loss": 0.7081,
      "step": 8182
    },
    {
      "epoch": 0.8089365593258038,
      "grad_norm": 10.260101586026657,
      "learning_rate": 1.8533681529444723e-08,
      "loss": 0.7088,
      "step": 8183
    },
    {
      "epoch": 0.8090354150705583,
      "grad_norm": 4.012702342966552,
      "learning_rate": 1.851511555431442e-08,
      "loss": 0.6145,
      "step": 8184
    },
    {
      "epoch": 0.8091342708153128,
      "grad_norm": 4.776286707360679,
      "learning_rate": 1.849655793425705e-08,
      "loss": 0.6851,
      "step": 8185
    },
    {
      "epoch": 0.8092331265600672,
      "grad_norm": 4.2249096739713945,
      "learning_rate": 1.8478008671175382e-08,
      "loss": 0.7273,
      "step": 8186
    },
    {
      "epoch": 0.8093319823048217,
      "grad_norm": 8.850385820498555,
      "learning_rate": 1.8459467766971393e-08,
      "loss": 0.6715,
      "step": 8187
    },
    {
      "epoch": 0.8094308380495762,
      "grad_norm": 3.9476191124175593,
      "learning_rate": 1.8440935223546194e-08,
      "loss": 0.6839,
      "step": 8188
    },
    {
      "epoch": 0.8095296937943306,
      "grad_norm": 3.1798306506150094,
      "learning_rate": 1.842241104279999e-08,
      "loss": 0.865,
      "step": 8189
    },
    {
      "epoch": 0.8096285495390851,
      "grad_norm": 3.0677553873650503,
      "learning_rate": 1.8403895226632182e-08,
      "loss": 0.6936,
      "step": 8190
    },
    {
      "epoch": 0.8097274052838396,
      "grad_norm": 4.473809637646809,
      "learning_rate": 1.8385387776941308e-08,
      "loss": 0.6968,
      "step": 8191
    },
    {
      "epoch": 0.809826261028594,
      "grad_norm": 8.203366040248381,
      "learning_rate": 1.8366888695625006e-08,
      "loss": 0.7318,
      "step": 8192
    },
    {
      "epoch": 0.8099251167733484,
      "grad_norm": 4.907774632496555,
      "learning_rate": 1.834839798458011e-08,
      "loss": 0.7082,
      "step": 8193
    },
    {
      "epoch": 0.810023972518103,
      "grad_norm": 3.413933087269006,
      "learning_rate": 1.8329915645702522e-08,
      "loss": 0.6947,
      "step": 8194
    },
    {
      "epoch": 0.8101228282628574,
      "grad_norm": 4.627868986948393,
      "learning_rate": 1.8311441680887407e-08,
      "loss": 0.8207,
      "step": 8195
    },
    {
      "epoch": 0.8102216840076119,
      "grad_norm": 3.548140764944573,
      "learning_rate": 1.8292976092028945e-08,
      "loss": 0.6936,
      "step": 8196
    },
    {
      "epoch": 0.8103205397523664,
      "grad_norm": 4.238567579393971,
      "learning_rate": 1.8274518881020507e-08,
      "loss": 0.6912,
      "step": 8197
    },
    {
      "epoch": 0.8104193954971208,
      "grad_norm": 5.016609537819754,
      "learning_rate": 1.8256070049754613e-08,
      "loss": 0.7587,
      "step": 8198
    },
    {
      "epoch": 0.8105182512418753,
      "grad_norm": 3.4734301664061333,
      "learning_rate": 1.823762960012295e-08,
      "loss": 0.718,
      "step": 8199
    },
    {
      "epoch": 0.8106171069866298,
      "grad_norm": 3.0597498150441313,
      "learning_rate": 1.8219197534016272e-08,
      "loss": 0.6586,
      "step": 8200
    },
    {
      "epoch": 0.8107159627313842,
      "grad_norm": 5.027853165180677,
      "learning_rate": 1.8200773853324535e-08,
      "loss": 0.6562,
      "step": 8201
    },
    {
      "epoch": 0.8108148184761387,
      "grad_norm": 2.920335786429377,
      "learning_rate": 1.818235855993684e-08,
      "loss": 0.6463,
      "step": 8202
    },
    {
      "epoch": 0.8109136742208931,
      "grad_norm": 10.040753312854289,
      "learning_rate": 1.816395165574136e-08,
      "loss": 0.6952,
      "step": 8203
    },
    {
      "epoch": 0.8110125299656477,
      "grad_norm": 3.18723536190724,
      "learning_rate": 1.8145553142625458e-08,
      "loss": 0.7675,
      "step": 8204
    },
    {
      "epoch": 0.8111113857104021,
      "grad_norm": 3.1496356897801183,
      "learning_rate": 1.8127163022475678e-08,
      "loss": 0.7491,
      "step": 8205
    },
    {
      "epoch": 0.8112102414551565,
      "grad_norm": 3.5359247867680454,
      "learning_rate": 1.81087812971776e-08,
      "loss": 0.6448,
      "step": 8206
    },
    {
      "epoch": 0.8113090971999111,
      "grad_norm": 4.708420359075111,
      "learning_rate": 1.809040796861604e-08,
      "loss": 0.7781,
      "step": 8207
    },
    {
      "epoch": 0.8114079529446655,
      "grad_norm": 3.415514179578994,
      "learning_rate": 1.8072043038674868e-08,
      "loss": 0.6977,
      "step": 8208
    },
    {
      "epoch": 0.8115068086894199,
      "grad_norm": 6.925374243937216,
      "learning_rate": 1.8053686509237154e-08,
      "loss": 0.7203,
      "step": 8209
    },
    {
      "epoch": 0.8116056644341745,
      "grad_norm": 6.911373662620026,
      "learning_rate": 1.8035338382185118e-08,
      "loss": 0.7213,
      "step": 8210
    },
    {
      "epoch": 0.8117045201789289,
      "grad_norm": 7.645235528449383,
      "learning_rate": 1.8016998659400073e-08,
      "loss": 0.629,
      "step": 8211
    },
    {
      "epoch": 0.8118033759236833,
      "grad_norm": 3.4842729849002185,
      "learning_rate": 1.7998667342762464e-08,
      "loss": 0.6894,
      "step": 8212
    },
    {
      "epoch": 0.8119022316684378,
      "grad_norm": 2.6477565026968275,
      "learning_rate": 1.798034443415193e-08,
      "loss": 0.6941,
      "step": 8213
    },
    {
      "epoch": 0.8120010874131923,
      "grad_norm": 4.224898656359294,
      "learning_rate": 1.7962029935447164e-08,
      "loss": 0.7863,
      "step": 8214
    },
    {
      "epoch": 0.8120999431579468,
      "grad_norm": 3.859186241181576,
      "learning_rate": 1.7943723848526104e-08,
      "loss": 0.6992,
      "step": 8215
    },
    {
      "epoch": 0.8121987989027012,
      "grad_norm": 4.863523930984556,
      "learning_rate": 1.7925426175265758e-08,
      "loss": 0.674,
      "step": 8216
    },
    {
      "epoch": 0.8122976546474557,
      "grad_norm": 3.3832295739100022,
      "learning_rate": 1.7907136917542253e-08,
      "loss": 0.6082,
      "step": 8217
    },
    {
      "epoch": 0.8123965103922102,
      "grad_norm": 11.726975414823167,
      "learning_rate": 1.788885607723091e-08,
      "loss": 0.577,
      "step": 8218
    },
    {
      "epoch": 0.8124953661369646,
      "grad_norm": 4.252033247864316,
      "learning_rate": 1.7870583656206163e-08,
      "loss": 0.6609,
      "step": 8219
    },
    {
      "epoch": 0.8125942218817191,
      "grad_norm": 6.70813638768911,
      "learning_rate": 1.785231965634154e-08,
      "loss": 0.6596,
      "step": 8220
    },
    {
      "epoch": 0.8126930776264736,
      "grad_norm": 4.11505250816376,
      "learning_rate": 1.7834064079509815e-08,
      "loss": 0.7548,
      "step": 8221
    },
    {
      "epoch": 0.812791933371228,
      "grad_norm": 4.520543684170736,
      "learning_rate": 1.781581692758275e-08,
      "loss": 0.7039,
      "step": 8222
    },
    {
      "epoch": 0.8128907891159824,
      "grad_norm": 2.9931530564415216,
      "learning_rate": 1.779757820243136e-08,
      "loss": 0.6816,
      "step": 8223
    },
    {
      "epoch": 0.812989644860737,
      "grad_norm": 3.009717395528903,
      "learning_rate": 1.777934790592578e-08,
      "loss": 0.7389,
      "step": 8224
    },
    {
      "epoch": 0.8130885006054914,
      "grad_norm": 6.244645921569853,
      "learning_rate": 1.7761126039935204e-08,
      "loss": 0.6736,
      "step": 8225
    },
    {
      "epoch": 0.8131873563502459,
      "grad_norm": 4.75447451498345,
      "learning_rate": 1.774291260632804e-08,
      "loss": 0.7427,
      "step": 8226
    },
    {
      "epoch": 0.8132862120950004,
      "grad_norm": 5.636699321971508,
      "learning_rate": 1.7724707606971846e-08,
      "loss": 0.7537,
      "step": 8227
    },
    {
      "epoch": 0.8133850678397548,
      "grad_norm": 3.122017304945608,
      "learning_rate": 1.770651104373323e-08,
      "loss": 0.6136,
      "step": 8228
    },
    {
      "epoch": 0.8134839235845093,
      "grad_norm": 5.078289691533819,
      "learning_rate": 1.7688322918477982e-08,
      "loss": 0.7013,
      "step": 8229
    },
    {
      "epoch": 0.8135827793292638,
      "grad_norm": 3.686192421778246,
      "learning_rate": 1.7670143233071043e-08,
      "loss": 0.6481,
      "step": 8230
    },
    {
      "epoch": 0.8136816350740183,
      "grad_norm": 4.731440608192188,
      "learning_rate": 1.765197198937646e-08,
      "loss": 0.7546,
      "step": 8231
    },
    {
      "epoch": 0.8137804908187727,
      "grad_norm": 4.4472733813265,
      "learning_rate": 1.763380918925744e-08,
      "loss": 0.605,
      "step": 8232
    },
    {
      "epoch": 0.8138793465635272,
      "grad_norm": 3.4940563734216274,
      "learning_rate": 1.7615654834576278e-08,
      "loss": 0.6619,
      "step": 8233
    },
    {
      "epoch": 0.8139782023082817,
      "grad_norm": 7.900678883149771,
      "learning_rate": 1.759750892719446e-08,
      "loss": 0.6459,
      "step": 8234
    },
    {
      "epoch": 0.8140770580530361,
      "grad_norm": 21.397274551094846,
      "learning_rate": 1.7579371468972603e-08,
      "loss": 0.6797,
      "step": 8235
    },
    {
      "epoch": 0.8141759137977905,
      "grad_norm": 7.528602648478857,
      "learning_rate": 1.7561242461770386e-08,
      "loss": 0.7786,
      "step": 8236
    },
    {
      "epoch": 0.8142747695425451,
      "grad_norm": 3.9842856462379084,
      "learning_rate": 1.75431219074467e-08,
      "loss": 0.6509,
      "step": 8237
    },
    {
      "epoch": 0.8143736252872995,
      "grad_norm": 3.201332350645333,
      "learning_rate": 1.752500980785957e-08,
      "loss": 0.7668,
      "step": 8238
    },
    {
      "epoch": 0.8144724810320539,
      "grad_norm": 4.524756083342483,
      "learning_rate": 1.7506906164866052e-08,
      "loss": 0.7083,
      "step": 8239
    },
    {
      "epoch": 0.8145713367768085,
      "grad_norm": 3.9634568385791717,
      "learning_rate": 1.7488810980322466e-08,
      "loss": 0.7308,
      "step": 8240
    },
    {
      "epoch": 0.8146701925215629,
      "grad_norm": 6.133669712075953,
      "learning_rate": 1.74707242560842e-08,
      "loss": 0.8066,
      "step": 8241
    },
    {
      "epoch": 0.8147690482663174,
      "grad_norm": 4.47751430175296,
      "learning_rate": 1.7452645994005744e-08,
      "loss": 0.7421,
      "step": 8242
    },
    {
      "epoch": 0.8148679040110719,
      "grad_norm": 5.1936843210131425,
      "learning_rate": 1.743457619594082e-08,
      "loss": 0.8076,
      "step": 8243
    },
    {
      "epoch": 0.8149667597558263,
      "grad_norm": 4.031384021215409,
      "learning_rate": 1.7416514863742138e-08,
      "loss": 0.756,
      "step": 8244
    },
    {
      "epoch": 0.8150656155005808,
      "grad_norm": 11.59836546382563,
      "learning_rate": 1.7398461999261705e-08,
      "loss": 0.8364,
      "step": 8245
    },
    {
      "epoch": 0.8151644712453352,
      "grad_norm": 4.29904940252144,
      "learning_rate": 1.7380417604350505e-08,
      "loss": 0.6336,
      "step": 8246
    },
    {
      "epoch": 0.8152633269900897,
      "grad_norm": 3.8169772336844954,
      "learning_rate": 1.7362381680858774e-08,
      "loss": 0.739,
      "step": 8247
    },
    {
      "epoch": 0.8153621827348442,
      "grad_norm": 6.61068990424553,
      "learning_rate": 1.7344354230635792e-08,
      "loss": 0.706,
      "step": 8248
    },
    {
      "epoch": 0.8154610384795986,
      "grad_norm": 4.646341179504866,
      "learning_rate": 1.7326335255530057e-08,
      "loss": 0.7144,
      "step": 8249
    },
    {
      "epoch": 0.8155598942243532,
      "grad_norm": 5.111739598268035,
      "learning_rate": 1.730832475738908e-08,
      "loss": 0.625,
      "step": 8250
    },
    {
      "epoch": 0.8156587499691076,
      "grad_norm": 4.674203153356449,
      "learning_rate": 1.7290322738059625e-08,
      "loss": 0.7047,
      "step": 8251
    },
    {
      "epoch": 0.815757605713862,
      "grad_norm": 2.7935455080211127,
      "learning_rate": 1.7272329199387535e-08,
      "loss": 0.5972,
      "step": 8252
    },
    {
      "epoch": 0.8158564614586166,
      "grad_norm": 3.3819912286115494,
      "learning_rate": 1.7254344143217748e-08,
      "loss": 0.7222,
      "step": 8253
    },
    {
      "epoch": 0.815955317203371,
      "grad_norm": 4.593922459927797,
      "learning_rate": 1.7236367571394418e-08,
      "loss": 0.656,
      "step": 8254
    },
    {
      "epoch": 0.8160541729481254,
      "grad_norm": 6.222984830531296,
      "learning_rate": 1.7218399485760704e-08,
      "loss": 0.7111,
      "step": 8255
    },
    {
      "epoch": 0.8161530286928799,
      "grad_norm": 3.58629320365369,
      "learning_rate": 1.7200439888159014e-08,
      "loss": 0.6331,
      "step": 8256
    },
    {
      "epoch": 0.8162518844376344,
      "grad_norm": 5.075534190745787,
      "learning_rate": 1.718248878043086e-08,
      "loss": 0.7241,
      "step": 8257
    },
    {
      "epoch": 0.8163507401823888,
      "grad_norm": 5.4473328675178205,
      "learning_rate": 1.7164546164416804e-08,
      "loss": 0.7723,
      "step": 8258
    },
    {
      "epoch": 0.8164495959271433,
      "grad_norm": 3.0122881662567416,
      "learning_rate": 1.7146612041956642e-08,
      "loss": 0.7104,
      "step": 8259
    },
    {
      "epoch": 0.8165484516718978,
      "grad_norm": 3.5677235976378423,
      "learning_rate": 1.712868641488926e-08,
      "loss": 0.7709,
      "step": 8260
    },
    {
      "epoch": 0.8166473074166523,
      "grad_norm": 4.699049151007746,
      "learning_rate": 1.711076928505262e-08,
      "loss": 0.5824,
      "step": 8261
    },
    {
      "epoch": 0.8167461631614067,
      "grad_norm": 8.601399317553106,
      "learning_rate": 1.7092860654283914e-08,
      "loss": 0.7879,
      "step": 8262
    },
    {
      "epoch": 0.8168450189061612,
      "grad_norm": 2.897728248092048,
      "learning_rate": 1.707496052441937e-08,
      "loss": 0.5682,
      "step": 8263
    },
    {
      "epoch": 0.8169438746509157,
      "grad_norm": 4.059608378273966,
      "learning_rate": 1.7057068897294404e-08,
      "loss": 0.691,
      "step": 8264
    },
    {
      "epoch": 0.8170427303956701,
      "grad_norm": 3.254323209476913,
      "learning_rate": 1.7039185774743536e-08,
      "loss": 0.7245,
      "step": 8265
    },
    {
      "epoch": 0.8171415861404245,
      "grad_norm": 4.164441407328137,
      "learning_rate": 1.7021311158600372e-08,
      "loss": 0.6555,
      "step": 8266
    },
    {
      "epoch": 0.8172404418851791,
      "grad_norm": 4.523652875761072,
      "learning_rate": 1.700344505069774e-08,
      "loss": 0.7107,
      "step": 8267
    },
    {
      "epoch": 0.8173392976299335,
      "grad_norm": 5.598328129388018,
      "learning_rate": 1.6985587452867556e-08,
      "loss": 0.6664,
      "step": 8268
    },
    {
      "epoch": 0.817438153374688,
      "grad_norm": 6.140245733243421,
      "learning_rate": 1.6967738366940797e-08,
      "loss": 0.7225,
      "step": 8269
    },
    {
      "epoch": 0.8175370091194425,
      "grad_norm": 4.567549527795139,
      "learning_rate": 1.6949897794747658e-08,
      "loss": 0.7086,
      "step": 8270
    },
    {
      "epoch": 0.8176358648641969,
      "grad_norm": 23.811263962604535,
      "learning_rate": 1.693206573811745e-08,
      "loss": 0.7385,
      "step": 8271
    },
    {
      "epoch": 0.8177347206089514,
      "grad_norm": 5.5288924747831425,
      "learning_rate": 1.6914242198878548e-08,
      "loss": 0.6963,
      "step": 8272
    },
    {
      "epoch": 0.8178335763537059,
      "grad_norm": 3.6813253789228373,
      "learning_rate": 1.6896427178858496e-08,
      "loss": 0.7195,
      "step": 8273
    },
    {
      "epoch": 0.8179324320984603,
      "grad_norm": 2.7909725424180114,
      "learning_rate": 1.6878620679884005e-08,
      "loss": 0.7113,
      "step": 8274
    },
    {
      "epoch": 0.8180312878432148,
      "grad_norm": 3.45831358542465,
      "learning_rate": 1.6860822703780807e-08,
      "loss": 0.6751,
      "step": 8275
    },
    {
      "epoch": 0.8181301435879692,
      "grad_norm": 4.054286922209515,
      "learning_rate": 1.6843033252373885e-08,
      "loss": 0.7092,
      "step": 8276
    },
    {
      "epoch": 0.8182289993327237,
      "grad_norm": 4.825193486260379,
      "learning_rate": 1.6825252327487215e-08,
      "loss": 0.7521,
      "step": 8277
    },
    {
      "epoch": 0.8183278550774782,
      "grad_norm": 4.001320099516522,
      "learning_rate": 1.680747993094401e-08,
      "loss": 0.7399,
      "step": 8278
    },
    {
      "epoch": 0.8184267108222326,
      "grad_norm": 5.0033345084824195,
      "learning_rate": 1.6789716064566584e-08,
      "loss": 0.7666,
      "step": 8279
    },
    {
      "epoch": 0.8185255665669872,
      "grad_norm": 5.559744975744832,
      "learning_rate": 1.6771960730176315e-08,
      "loss": 0.7571,
      "step": 8280
    },
    {
      "epoch": 0.8186244223117416,
      "grad_norm": 3.3275998935342797,
      "learning_rate": 1.6754213929593795e-08,
      "loss": 0.6216,
      "step": 8281
    },
    {
      "epoch": 0.818723278056496,
      "grad_norm": 5.6785801638946385,
      "learning_rate": 1.6736475664638683e-08,
      "loss": 0.6617,
      "step": 8282
    },
    {
      "epoch": 0.8188221338012506,
      "grad_norm": 4.92166843824104,
      "learning_rate": 1.671874593712973e-08,
      "loss": 0.6288,
      "step": 8283
    },
    {
      "epoch": 0.818920989546005,
      "grad_norm": 4.089705936751735,
      "learning_rate": 1.6701024748884907e-08,
      "loss": 0.5966,
      "step": 8284
    },
    {
      "epoch": 0.8190198452907594,
      "grad_norm": 5.712583100205691,
      "learning_rate": 1.668331210172127e-08,
      "loss": 0.7065,
      "step": 8285
    },
    {
      "epoch": 0.8191187010355139,
      "grad_norm": 7.664507508837457,
      "learning_rate": 1.6665607997454956e-08,
      "loss": 0.6599,
      "step": 8286
    },
    {
      "epoch": 0.8192175567802684,
      "grad_norm": 4.503768432042703,
      "learning_rate": 1.664791243790128e-08,
      "loss": 0.6524,
      "step": 8287
    },
    {
      "epoch": 0.8193164125250229,
      "grad_norm": 3.49595098592785,
      "learning_rate": 1.6630225424874644e-08,
      "loss": 0.7348,
      "step": 8288
    },
    {
      "epoch": 0.8194152682697773,
      "grad_norm": 11.670657499005333,
      "learning_rate": 1.661254696018859e-08,
      "loss": 0.661,
      "step": 8289
    },
    {
      "epoch": 0.8195141240145318,
      "grad_norm": 4.888578689794759,
      "learning_rate": 1.6594877045655818e-08,
      "loss": 0.8014,
      "step": 8290
    },
    {
      "epoch": 0.8196129797592863,
      "grad_norm": 8.49400272653561,
      "learning_rate": 1.6577215683088075e-08,
      "loss": 0.7111,
      "step": 8291
    },
    {
      "epoch": 0.8197118355040407,
      "grad_norm": 3.593352216267742,
      "learning_rate": 1.6559562874296284e-08,
      "loss": 0.7232,
      "step": 8292
    },
    {
      "epoch": 0.8198106912487952,
      "grad_norm": 5.004505436285176,
      "learning_rate": 1.6541918621090522e-08,
      "loss": 0.7659,
      "step": 8293
    },
    {
      "epoch": 0.8199095469935497,
      "grad_norm": 6.1196477879901465,
      "learning_rate": 1.6524282925279876e-08,
      "loss": 0.7334,
      "step": 8294
    },
    {
      "epoch": 0.8200084027383041,
      "grad_norm": 4.677397523924521,
      "learning_rate": 1.6506655788672674e-08,
      "loss": 0.613,
      "step": 8295
    },
    {
      "epoch": 0.8201072584830585,
      "grad_norm": 4.078897000627077,
      "learning_rate": 1.6489037213076318e-08,
      "loss": 0.6557,
      "step": 8296
    },
    {
      "epoch": 0.8202061142278131,
      "grad_norm": 4.7253345790109025,
      "learning_rate": 1.6471427200297305e-08,
      "loss": 0.7185,
      "step": 8297
    },
    {
      "epoch": 0.8203049699725675,
      "grad_norm": 3.7938464536301946,
      "learning_rate": 1.6453825752141325e-08,
      "loss": 0.6978,
      "step": 8298
    },
    {
      "epoch": 0.820403825717322,
      "grad_norm": 3.214738810396406,
      "learning_rate": 1.6436232870413125e-08,
      "loss": 0.6793,
      "step": 8299
    },
    {
      "epoch": 0.8205026814620765,
      "grad_norm": 4.968759587456315,
      "learning_rate": 1.641864855691656e-08,
      "loss": 0.6695,
      "step": 8300
    },
    {
      "epoch": 0.8206015372068309,
      "grad_norm": 2.928517898900211,
      "learning_rate": 1.6401072813454697e-08,
      "loss": 0.5499,
      "step": 8301
    },
    {
      "epoch": 0.8207003929515854,
      "grad_norm": 2.821924426880065,
      "learning_rate": 1.638350564182963e-08,
      "loss": 0.7541,
      "step": 8302
    },
    {
      "epoch": 0.8207992486963399,
      "grad_norm": 6.00140054162403,
      "learning_rate": 1.636594704384262e-08,
      "loss": 0.6658,
      "step": 8303
    },
    {
      "epoch": 0.8208981044410943,
      "grad_norm": 4.035540182030818,
      "learning_rate": 1.634839702129409e-08,
      "loss": 0.703,
      "step": 8304
    },
    {
      "epoch": 0.8209969601858488,
      "grad_norm": 2.605790675300495,
      "learning_rate": 1.633085557598346e-08,
      "loss": 0.6488,
      "step": 8305
    },
    {
      "epoch": 0.8210958159306032,
      "grad_norm": 4.763399059015804,
      "learning_rate": 1.6313322709709387e-08,
      "loss": 0.6941,
      "step": 8306
    },
    {
      "epoch": 0.8211946716753578,
      "grad_norm": 3.077832850955196,
      "learning_rate": 1.6295798424269636e-08,
      "loss": 0.7497,
      "step": 8307
    },
    {
      "epoch": 0.8212935274201122,
      "grad_norm": 4.63172512895805,
      "learning_rate": 1.6278282721461e-08,
      "loss": 0.8086,
      "step": 8308
    },
    {
      "epoch": 0.8213923831648666,
      "grad_norm": 3.491385239515395,
      "learning_rate": 1.6260775603079514e-08,
      "loss": 0.77,
      "step": 8309
    },
    {
      "epoch": 0.8214912389096212,
      "grad_norm": 3.0345941601615403,
      "learning_rate": 1.6243277070920223e-08,
      "loss": 0.6368,
      "step": 8310
    },
    {
      "epoch": 0.8215900946543756,
      "grad_norm": 5.550387968562197,
      "learning_rate": 1.622578712677738e-08,
      "loss": 0.7537,
      "step": 8311
    },
    {
      "epoch": 0.82168895039913,
      "grad_norm": 7.594189685548211,
      "learning_rate": 1.620830577244433e-08,
      "loss": 0.7361,
      "step": 8312
    },
    {
      "epoch": 0.8217878061438846,
      "grad_norm": 9.036354056799702,
      "learning_rate": 1.6190833009713478e-08,
      "loss": 0.6386,
      "step": 8313
    },
    {
      "epoch": 0.821886661888639,
      "grad_norm": 16.87266436104744,
      "learning_rate": 1.6173368840376433e-08,
      "loss": 0.6561,
      "step": 8314
    },
    {
      "epoch": 0.8219855176333934,
      "grad_norm": 4.265718735573408,
      "learning_rate": 1.615591326622391e-08,
      "loss": 0.6674,
      "step": 8315
    },
    {
      "epoch": 0.822084373378148,
      "grad_norm": 4.045967832081505,
      "learning_rate": 1.6138466289045694e-08,
      "loss": 0.7716,
      "step": 8316
    },
    {
      "epoch": 0.8221832291229024,
      "grad_norm": 6.411548971302622,
      "learning_rate": 1.612102791063069e-08,
      "loss": 0.6994,
      "step": 8317
    },
    {
      "epoch": 0.8222820848676569,
      "grad_norm": 4.005056481892624,
      "learning_rate": 1.6103598132766992e-08,
      "loss": 0.7419,
      "step": 8318
    },
    {
      "epoch": 0.8223809406124113,
      "grad_norm": 4.6074434271142435,
      "learning_rate": 1.6086176957241727e-08,
      "loss": 0.7675,
      "step": 8319
    },
    {
      "epoch": 0.8224797963571658,
      "grad_norm": 4.451022912902485,
      "learning_rate": 1.6068764385841226e-08,
      "loss": 0.6247,
      "step": 8320
    },
    {
      "epoch": 0.8225786521019203,
      "grad_norm": 4.578728126771792,
      "learning_rate": 1.6051360420350835e-08,
      "loss": 0.813,
      "step": 8321
    },
    {
      "epoch": 0.8226775078466747,
      "grad_norm": 4.320378280614996,
      "learning_rate": 1.603396506255511e-08,
      "loss": 0.6343,
      "step": 8322
    },
    {
      "epoch": 0.8227763635914292,
      "grad_norm": 7.475883019237751,
      "learning_rate": 1.6016578314237694e-08,
      "loss": 0.712,
      "step": 8323
    },
    {
      "epoch": 0.8228752193361837,
      "grad_norm": 3.4114281323850046,
      "learning_rate": 1.5999200177181317e-08,
      "loss": 0.7037,
      "step": 8324
    },
    {
      "epoch": 0.8229740750809381,
      "grad_norm": 4.527512916740641,
      "learning_rate": 1.598183065316785e-08,
      "loss": 0.6333,
      "step": 8325
    },
    {
      "epoch": 0.8230729308256927,
      "grad_norm": 17.07600038139287,
      "learning_rate": 1.5964469743978327e-08,
      "loss": 0.5808,
      "step": 8326
    },
    {
      "epoch": 0.8231717865704471,
      "grad_norm": 10.155003574549776,
      "learning_rate": 1.59471174513928e-08,
      "loss": 0.7144,
      "step": 8327
    },
    {
      "epoch": 0.8232706423152015,
      "grad_norm": 5.575450733186684,
      "learning_rate": 1.5929773777190503e-08,
      "loss": 0.7074,
      "step": 8328
    },
    {
      "epoch": 0.823369498059956,
      "grad_norm": 4.374997845633731,
      "learning_rate": 1.5912438723149814e-08,
      "loss": 0.6833,
      "step": 8329
    },
    {
      "epoch": 0.8234683538047105,
      "grad_norm": 3.7089454696365203,
      "learning_rate": 1.589511229104813e-08,
      "loss": 0.7028,
      "step": 8330
    },
    {
      "epoch": 0.8235672095494649,
      "grad_norm": 4.477289867045588,
      "learning_rate": 1.5877794482662067e-08,
      "loss": 0.6475,
      "step": 8331
    },
    {
      "epoch": 0.8236660652942194,
      "grad_norm": 19.03225158055497,
      "learning_rate": 1.5860485299767256e-08,
      "loss": 0.7136,
      "step": 8332
    },
    {
      "epoch": 0.8237649210389739,
      "grad_norm": 30.791059398983332,
      "learning_rate": 1.5843184744138572e-08,
      "loss": 0.7807,
      "step": 8333
    },
    {
      "epoch": 0.8238637767837284,
      "grad_norm": 5.085049640488362,
      "learning_rate": 1.582589281754986e-08,
      "loss": 0.6963,
      "step": 8334
    },
    {
      "epoch": 0.8239626325284828,
      "grad_norm": 3.545174176970526,
      "learning_rate": 1.580860952177421e-08,
      "loss": 0.7309,
      "step": 8335
    },
    {
      "epoch": 0.8240614882732373,
      "grad_norm": 4.005614185889951,
      "learning_rate": 1.5791334858583727e-08,
      "loss": 0.7004,
      "step": 8336
    },
    {
      "epoch": 0.8241603440179918,
      "grad_norm": 9.263872204438943,
      "learning_rate": 1.5774068829749698e-08,
      "loss": 0.6254,
      "step": 8337
    },
    {
      "epoch": 0.8242591997627462,
      "grad_norm": 12.32613124659327,
      "learning_rate": 1.575681143704247e-08,
      "loss": 0.7532,
      "step": 8338
    },
    {
      "epoch": 0.8243580555075006,
      "grad_norm": 7.958827132984919,
      "learning_rate": 1.5739562682231567e-08,
      "loss": 0.6851,
      "step": 8339
    },
    {
      "epoch": 0.8244569112522552,
      "grad_norm": 3.544651006025782,
      "learning_rate": 1.57223225670856e-08,
      "loss": 0.758,
      "step": 8340
    },
    {
      "epoch": 0.8245557669970096,
      "grad_norm": 3.289423408132008,
      "learning_rate": 1.5705091093372237e-08,
      "loss": 0.7944,
      "step": 8341
    },
    {
      "epoch": 0.824654622741764,
      "grad_norm": 2.700903622355619,
      "learning_rate": 1.5687868262858383e-08,
      "loss": 0.7648,
      "step": 8342
    },
    {
      "epoch": 0.8247534784865186,
      "grad_norm": 10.058877745505415,
      "learning_rate": 1.567065407730991e-08,
      "loss": 0.6756,
      "step": 8343
    },
    {
      "epoch": 0.824852334231273,
      "grad_norm": 3.944522090652046,
      "learning_rate": 1.565344853849193e-08,
      "loss": 0.6761,
      "step": 8344
    },
    {
      "epoch": 0.8249511899760275,
      "grad_norm": 3.1750631698464358,
      "learning_rate": 1.563625164816862e-08,
      "loss": 0.7694,
      "step": 8345
    },
    {
      "epoch": 0.825050045720782,
      "grad_norm": 3.686615443054929,
      "learning_rate": 1.561906340810323e-08,
      "loss": 0.6833,
      "step": 8346
    },
    {
      "epoch": 0.8251489014655364,
      "grad_norm": 4.448221986604397,
      "learning_rate": 1.560188382005817e-08,
      "loss": 0.5606,
      "step": 8347
    },
    {
      "epoch": 0.8252477572102909,
      "grad_norm": 17.426064647210772,
      "learning_rate": 1.5584712885794993e-08,
      "loss": 0.8086,
      "step": 8348
    },
    {
      "epoch": 0.8253466129550453,
      "grad_norm": 73.01560356175342,
      "learning_rate": 1.5567550607074277e-08,
      "loss": 0.7518,
      "step": 8349
    },
    {
      "epoch": 0.8254454686997998,
      "grad_norm": 9.990590851631255,
      "learning_rate": 1.55503969856558e-08,
      "loss": 0.7298,
      "step": 8350
    },
    {
      "epoch": 0.8255443244445543,
      "grad_norm": 5.0362370024237775,
      "learning_rate": 1.553325202329837e-08,
      "loss": 0.8066,
      "step": 8351
    },
    {
      "epoch": 0.8256431801893087,
      "grad_norm": 4.229847563330237,
      "learning_rate": 1.5516115721760002e-08,
      "loss": 0.7195,
      "step": 8352
    },
    {
      "epoch": 0.8257420359340633,
      "grad_norm": 3.9793701011856935,
      "learning_rate": 1.5498988082797737e-08,
      "loss": 0.7396,
      "step": 8353
    },
    {
      "epoch": 0.8258408916788177,
      "grad_norm": 3.4198261592247463,
      "learning_rate": 1.5481869108167746e-08,
      "loss": 0.7162,
      "step": 8354
    },
    {
      "epoch": 0.8259397474235721,
      "grad_norm": 7.051106559053553,
      "learning_rate": 1.5464758799625342e-08,
      "loss": 0.7242,
      "step": 8355
    },
    {
      "epoch": 0.8260386031683267,
      "grad_norm": 4.325617709016498,
      "learning_rate": 1.5447657158924964e-08,
      "loss": 0.6406,
      "step": 8356
    },
    {
      "epoch": 0.8261374589130811,
      "grad_norm": 3.0296782343908575,
      "learning_rate": 1.5430564187820094e-08,
      "loss": 0.7492,
      "step": 8357
    },
    {
      "epoch": 0.8262363146578355,
      "grad_norm": 4.138735893466102,
      "learning_rate": 1.5413479888063373e-08,
      "loss": 0.6108,
      "step": 8358
    },
    {
      "epoch": 0.82633517040259,
      "grad_norm": 6.1708453314380245,
      "learning_rate": 1.539640426140658e-08,
      "loss": 0.6979,
      "step": 8359
    },
    {
      "epoch": 0.8264340261473445,
      "grad_norm": 2.663752452815474,
      "learning_rate": 1.5379337309600515e-08,
      "loss": 0.5624,
      "step": 8360
    },
    {
      "epoch": 0.826532881892099,
      "grad_norm": 10.83706099551059,
      "learning_rate": 1.5362279034395154e-08,
      "loss": 0.7461,
      "step": 8361
    },
    {
      "epoch": 0.8266317376368534,
      "grad_norm": 5.688496479767749,
      "learning_rate": 1.534522943753962e-08,
      "loss": 0.7326,
      "step": 8362
    },
    {
      "epoch": 0.8267305933816079,
      "grad_norm": 3.7716473451494035,
      "learning_rate": 1.5328188520782038e-08,
      "loss": 0.7161,
      "step": 8363
    },
    {
      "epoch": 0.8268294491263624,
      "grad_norm": 6.747578791013639,
      "learning_rate": 1.5311156285869743e-08,
      "loss": 0.7032,
      "step": 8364
    },
    {
      "epoch": 0.8269283048711168,
      "grad_norm": 2.9329721875188666,
      "learning_rate": 1.5294132734549103e-08,
      "loss": 0.6501,
      "step": 8365
    },
    {
      "epoch": 0.8270271606158713,
      "grad_norm": 5.787881881999662,
      "learning_rate": 1.527711786856565e-08,
      "loss": 0.6189,
      "step": 8366
    },
    {
      "epoch": 0.8271260163606258,
      "grad_norm": 3.0423283381581516,
      "learning_rate": 1.5260111689664034e-08,
      "loss": 0.6611,
      "step": 8367
    },
    {
      "epoch": 0.8272248721053802,
      "grad_norm": 4.186371233228861,
      "learning_rate": 1.5243114199587938e-08,
      "loss": 0.7501,
      "step": 8368
    },
    {
      "epoch": 0.8273237278501346,
      "grad_norm": 4.020584042988217,
      "learning_rate": 1.5226125400080247e-08,
      "loss": 0.6927,
      "step": 8369
    },
    {
      "epoch": 0.8274225835948892,
      "grad_norm": 5.2396084939466,
      "learning_rate": 1.5209145292882907e-08,
      "loss": 0.6241,
      "step": 8370
    },
    {
      "epoch": 0.8275214393396436,
      "grad_norm": 3.176471290028604,
      "learning_rate": 1.5192173879736926e-08,
      "loss": 0.7076,
      "step": 8371
    },
    {
      "epoch": 0.827620295084398,
      "grad_norm": 2.983634030241156,
      "learning_rate": 1.517521116238252e-08,
      "loss": 0.5683,
      "step": 8372
    },
    {
      "epoch": 0.8277191508291526,
      "grad_norm": 3.6476545727788388,
      "learning_rate": 1.5158257142558972e-08,
      "loss": 0.7798,
      "step": 8373
    },
    {
      "epoch": 0.827818006573907,
      "grad_norm": 4.42617828954565,
      "learning_rate": 1.5141311822004632e-08,
      "loss": 0.7395,
      "step": 8374
    },
    {
      "epoch": 0.8279168623186615,
      "grad_norm": 7.300808652890452,
      "learning_rate": 1.512437520245704e-08,
      "loss": 0.7628,
      "step": 8375
    },
    {
      "epoch": 0.828015718063416,
      "grad_norm": 5.76394435579745,
      "learning_rate": 1.510744728565273e-08,
      "loss": 0.6916,
      "step": 8376
    },
    {
      "epoch": 0.8281145738081704,
      "grad_norm": 7.771273040953953,
      "learning_rate": 1.5090528073327458e-08,
      "loss": 0.7155,
      "step": 8377
    },
    {
      "epoch": 0.8282134295529249,
      "grad_norm": 6.975691809674111,
      "learning_rate": 1.5073617567216047e-08,
      "loss": 0.7576,
      "step": 8378
    },
    {
      "epoch": 0.8283122852976793,
      "grad_norm": 3.872725970533153,
      "learning_rate": 1.5056715769052387e-08,
      "loss": 0.8019,
      "step": 8379
    },
    {
      "epoch": 0.8284111410424339,
      "grad_norm": 3.9969505814254487,
      "learning_rate": 1.5039822680569513e-08,
      "loss": 0.6189,
      "step": 8380
    },
    {
      "epoch": 0.8285099967871883,
      "grad_norm": 3.2316004106672915,
      "learning_rate": 1.50229383034996e-08,
      "loss": 0.8106,
      "step": 8381
    },
    {
      "epoch": 0.8286088525319427,
      "grad_norm": 5.038566813300291,
      "learning_rate": 1.5006062639573848e-08,
      "loss": 0.6772,
      "step": 8382
    },
    {
      "epoch": 0.8287077082766973,
      "grad_norm": 5.589243653836192,
      "learning_rate": 1.4989195690522615e-08,
      "loss": 0.7383,
      "step": 8383
    },
    {
      "epoch": 0.8288065640214517,
      "grad_norm": 4.963385631749689,
      "learning_rate": 1.497233745807539e-08,
      "loss": 0.5963,
      "step": 8384
    },
    {
      "epoch": 0.8289054197662061,
      "grad_norm": 6.666860866649288,
      "learning_rate": 1.4955487943960688e-08,
      "loss": 0.5939,
      "step": 8385
    },
    {
      "epoch": 0.8290042755109607,
      "grad_norm": 6.69735952552464,
      "learning_rate": 1.4938647149906226e-08,
      "loss": 0.715,
      "step": 8386
    },
    {
      "epoch": 0.8291031312557151,
      "grad_norm": 3.9429314085133567,
      "learning_rate": 1.4921815077638756e-08,
      "loss": 0.6739,
      "step": 8387
    },
    {
      "epoch": 0.8292019870004695,
      "grad_norm": 17.843405463186627,
      "learning_rate": 1.4904991728884144e-08,
      "loss": 0.7374,
      "step": 8388
    },
    {
      "epoch": 0.8293008427452241,
      "grad_norm": 4.110535600608065,
      "learning_rate": 1.4888177105367406e-08,
      "loss": 0.6628,
      "step": 8389
    },
    {
      "epoch": 0.8293996984899785,
      "grad_norm": 3.6985991626488466,
      "learning_rate": 1.487137120881259e-08,
      "loss": 0.7026,
      "step": 8390
    },
    {
      "epoch": 0.829498554234733,
      "grad_norm": 7.590234269957879,
      "learning_rate": 1.4854574040942924e-08,
      "loss": 0.6785,
      "step": 8391
    },
    {
      "epoch": 0.8295974099794874,
      "grad_norm": 3.891130443134417,
      "learning_rate": 1.4837785603480734e-08,
      "loss": 0.6326,
      "step": 8392
    },
    {
      "epoch": 0.8296962657242419,
      "grad_norm": 4.210015559180781,
      "learning_rate": 1.4821005898147376e-08,
      "loss": 0.7574,
      "step": 8393
    },
    {
      "epoch": 0.8297951214689964,
      "grad_norm": 4.519118160840881,
      "learning_rate": 1.4804234926663383e-08,
      "loss": 0.6916,
      "step": 8394
    },
    {
      "epoch": 0.8298939772137508,
      "grad_norm": 3.2651750564988906,
      "learning_rate": 1.4787472690748392e-08,
      "loss": 0.7135,
      "step": 8395
    },
    {
      "epoch": 0.8299928329585053,
      "grad_norm": 13.11160032545962,
      "learning_rate": 1.4770719192121083e-08,
      "loss": 0.6197,
      "step": 8396
    },
    {
      "epoch": 0.8300916887032598,
      "grad_norm": 5.80355515277913,
      "learning_rate": 1.4753974432499327e-08,
      "loss": 0.6078,
      "step": 8397
    },
    {
      "epoch": 0.8301905444480142,
      "grad_norm": 3.4331403665267994,
      "learning_rate": 1.4737238413600005e-08,
      "loss": 0.7504,
      "step": 8398
    },
    {
      "epoch": 0.8302894001927688,
      "grad_norm": 6.265011323639912,
      "learning_rate": 1.4720511137139158e-08,
      "loss": 0.7032,
      "step": 8399
    },
    {
      "epoch": 0.8303882559375232,
      "grad_norm": 4.022749769966276,
      "learning_rate": 1.4703792604831966e-08,
      "loss": 0.7627,
      "step": 8400
    },
    {
      "epoch": 0.8304871116822776,
      "grad_norm": 4.426030682826308,
      "learning_rate": 1.4687082818392605e-08,
      "loss": 0.747,
      "step": 8401
    },
    {
      "epoch": 0.8305859674270321,
      "grad_norm": 3.3110957601359727,
      "learning_rate": 1.4670381779534446e-08,
      "loss": 0.6878,
      "step": 8402
    },
    {
      "epoch": 0.8306848231717866,
      "grad_norm": 5.862627453235098,
      "learning_rate": 1.4653689489969967e-08,
      "loss": 0.748,
      "step": 8403
    },
    {
      "epoch": 0.830783678916541,
      "grad_norm": 9.109290056583887,
      "learning_rate": 1.4637005951410676e-08,
      "loss": 0.7425,
      "step": 8404
    },
    {
      "epoch": 0.8308825346612955,
      "grad_norm": 6.373231684124402,
      "learning_rate": 1.4620331165567212e-08,
      "loss": 0.7299,
      "step": 8405
    },
    {
      "epoch": 0.83098139040605,
      "grad_norm": 5.007916551345476,
      "learning_rate": 1.460366513414938e-08,
      "loss": 0.7297,
      "step": 8406
    },
    {
      "epoch": 0.8310802461508044,
      "grad_norm": 3.4168897557232074,
      "learning_rate": 1.4587007858865964e-08,
      "loss": 0.6993,
      "step": 8407
    },
    {
      "epoch": 0.8311791018955589,
      "grad_norm": 3.24849974299769,
      "learning_rate": 1.4570359341424998e-08,
      "loss": 0.6877,
      "step": 8408
    },
    {
      "epoch": 0.8312779576403134,
      "grad_norm": 4.243389647969769,
      "learning_rate": 1.4553719583533475e-08,
      "loss": 0.76,
      "step": 8409
    },
    {
      "epoch": 0.8313768133850679,
      "grad_norm": 7.769668518136011,
      "learning_rate": 1.4537088586897583e-08,
      "loss": 0.7191,
      "step": 8410
    },
    {
      "epoch": 0.8314756691298223,
      "grad_norm": 3.3502197490534975,
      "learning_rate": 1.4520466353222627e-08,
      "loss": 0.674,
      "step": 8411
    },
    {
      "epoch": 0.8315745248745767,
      "grad_norm": 3.8990861851489402,
      "learning_rate": 1.4503852884212897e-08,
      "loss": 0.738,
      "step": 8412
    },
    {
      "epoch": 0.8316733806193313,
      "grad_norm": 6.440867400141936,
      "learning_rate": 1.4487248181571909e-08,
      "loss": 0.7199,
      "step": 8413
    },
    {
      "epoch": 0.8317722363640857,
      "grad_norm": 5.954434549071798,
      "learning_rate": 1.4470652247002235e-08,
      "loss": 0.5838,
      "step": 8414
    },
    {
      "epoch": 0.8318710921088401,
      "grad_norm": 9.354101418231958,
      "learning_rate": 1.4454065082205503e-08,
      "loss": 0.6712,
      "step": 8415
    },
    {
      "epoch": 0.8319699478535947,
      "grad_norm": 3.253206329678569,
      "learning_rate": 1.4437486688882506e-08,
      "loss": 0.7023,
      "step": 8416
    },
    {
      "epoch": 0.8320688035983491,
      "grad_norm": 4.183063972718322,
      "learning_rate": 1.4420917068733141e-08,
      "loss": 0.6843,
      "step": 8417
    },
    {
      "epoch": 0.8321676593431035,
      "grad_norm": 19.711341637973977,
      "learning_rate": 1.4404356223456326e-08,
      "loss": 0.7626,
      "step": 8418
    },
    {
      "epoch": 0.8322665150878581,
      "grad_norm": 5.2312665384497,
      "learning_rate": 1.4387804154750171e-08,
      "loss": 0.6991,
      "step": 8419
    },
    {
      "epoch": 0.8323653708326125,
      "grad_norm": 4.398129961166469,
      "learning_rate": 1.4371260864311817e-08,
      "loss": 0.6639,
      "step": 8420
    },
    {
      "epoch": 0.832464226577367,
      "grad_norm": 3.4987046129129,
      "learning_rate": 1.4354726353837565e-08,
      "loss": 0.6492,
      "step": 8421
    },
    {
      "epoch": 0.8325630823221214,
      "grad_norm": 4.032073529190712,
      "learning_rate": 1.4338200625022756e-08,
      "loss": 0.7109,
      "step": 8422
    },
    {
      "epoch": 0.8326619380668759,
      "grad_norm": 2.772261643102233,
      "learning_rate": 1.432168367956188e-08,
      "loss": 0.6592,
      "step": 8423
    },
    {
      "epoch": 0.8327607938116304,
      "grad_norm": 5.289960931090339,
      "learning_rate": 1.4305175519148483e-08,
      "loss": 0.7121,
      "step": 8424
    },
    {
      "epoch": 0.8328596495563848,
      "grad_norm": 4.8570080365121475,
      "learning_rate": 1.4288676145475276e-08,
      "loss": 0.7978,
      "step": 8425
    },
    {
      "epoch": 0.8329585053011394,
      "grad_norm": 5.5494524855144975,
      "learning_rate": 1.4272185560233974e-08,
      "loss": 0.7546,
      "step": 8426
    },
    {
      "epoch": 0.8330573610458938,
      "grad_norm": 3.4205557381827205,
      "learning_rate": 1.4255703765115468e-08,
      "loss": 0.6435,
      "step": 8427
    },
    {
      "epoch": 0.8331562167906482,
      "grad_norm": 3.435215908598894,
      "learning_rate": 1.423923076180974e-08,
      "loss": 0.7006,
      "step": 8428
    },
    {
      "epoch": 0.8332550725354028,
      "grad_norm": 5.20846584385242,
      "learning_rate": 1.4222766552005827e-08,
      "loss": 0.6863,
      "step": 8429
    },
    {
      "epoch": 0.8333539282801572,
      "grad_norm": 3.0200734504686393,
      "learning_rate": 1.4206311137391914e-08,
      "loss": 0.5699,
      "step": 8430
    },
    {
      "epoch": 0.8334527840249116,
      "grad_norm": 4.579210645914143,
      "learning_rate": 1.4189864519655226e-08,
      "loss": 0.7581,
      "step": 8431
    },
    {
      "epoch": 0.8335516397696661,
      "grad_norm": 3.166561061995313,
      "learning_rate": 1.417342670048215e-08,
      "loss": 0.5814,
      "step": 8432
    },
    {
      "epoch": 0.8336504955144206,
      "grad_norm": 8.14588502728149,
      "learning_rate": 1.4156997681558147e-08,
      "loss": 0.7271,
      "step": 8433
    },
    {
      "epoch": 0.833749351259175,
      "grad_norm": 3.1790204380794087,
      "learning_rate": 1.4140577464567738e-08,
      "loss": 0.6831,
      "step": 8434
    },
    {
      "epoch": 0.8338482070039295,
      "grad_norm": 4.382773083521978,
      "learning_rate": 1.4124166051194608e-08,
      "loss": 0.6821,
      "step": 8435
    },
    {
      "epoch": 0.833947062748684,
      "grad_norm": 4.101192551030484,
      "learning_rate": 1.4107763443121501e-08,
      "loss": 0.5987,
      "step": 8436
    },
    {
      "epoch": 0.8340459184934385,
      "grad_norm": 3.8061989433986,
      "learning_rate": 1.4091369642030227e-08,
      "loss": 0.728,
      "step": 8437
    },
    {
      "epoch": 0.8341447742381929,
      "grad_norm": 22.40776214194871,
      "learning_rate": 1.4074984649601785e-08,
      "loss": 0.6833,
      "step": 8438
    },
    {
      "epoch": 0.8342436299829474,
      "grad_norm": 3.9745329802884553,
      "learning_rate": 1.4058608467516164e-08,
      "loss": 0.6765,
      "step": 8439
    },
    {
      "epoch": 0.8343424857277019,
      "grad_norm": 4.499406959795461,
      "learning_rate": 1.4042241097452534e-08,
      "loss": 0.6447,
      "step": 8440
    },
    {
      "epoch": 0.8344413414724563,
      "grad_norm": 2.6459490105881374,
      "learning_rate": 1.4025882541089118e-08,
      "loss": 0.7376,
      "step": 8441
    },
    {
      "epoch": 0.8345401972172107,
      "grad_norm": 3.4076972058650727,
      "learning_rate": 1.4009532800103218e-08,
      "loss": 0.6952,
      "step": 8442
    },
    {
      "epoch": 0.8346390529619653,
      "grad_norm": 2.70093654578028,
      "learning_rate": 1.3993191876171284e-08,
      "loss": 0.6357,
      "step": 8443
    },
    {
      "epoch": 0.8347379087067197,
      "grad_norm": 3.2404982251436962,
      "learning_rate": 1.3976859770968852e-08,
      "loss": 0.662,
      "step": 8444
    },
    {
      "epoch": 0.8348367644514741,
      "grad_norm": 3.5799945216846085,
      "learning_rate": 1.3960536486170493e-08,
      "loss": 0.6198,
      "step": 8445
    },
    {
      "epoch": 0.8349356201962287,
      "grad_norm": 4.776214668550081,
      "learning_rate": 1.3944222023449959e-08,
      "loss": 0.6956,
      "step": 8446
    },
    {
      "epoch": 0.8350344759409831,
      "grad_norm": 4.0648306536869265,
      "learning_rate": 1.3927916384480054e-08,
      "loss": 0.7457,
      "step": 8447
    },
    {
      "epoch": 0.8351333316857376,
      "grad_norm": 5.449152846135534,
      "learning_rate": 1.3911619570932654e-08,
      "loss": 0.6345,
      "step": 8448
    },
    {
      "epoch": 0.8352321874304921,
      "grad_norm": 7.983470980996166,
      "learning_rate": 1.3895331584478776e-08,
      "loss": 0.5954,
      "step": 8449
    },
    {
      "epoch": 0.8353310431752465,
      "grad_norm": 6.242531433773275,
      "learning_rate": 1.3879052426788529e-08,
      "loss": 0.6594,
      "step": 8450
    },
    {
      "epoch": 0.835429898920001,
      "grad_norm": 4.095958505138432,
      "learning_rate": 1.3862782099531056e-08,
      "loss": 0.7421,
      "step": 8451
    },
    {
      "epoch": 0.8355287546647554,
      "grad_norm": 6.12624439355586,
      "learning_rate": 1.3846520604374701e-08,
      "loss": 0.8037,
      "step": 8452
    },
    {
      "epoch": 0.83562761040951,
      "grad_norm": 3.674892353793679,
      "learning_rate": 1.3830267942986773e-08,
      "loss": 0.7862,
      "step": 8453
    },
    {
      "epoch": 0.8357264661542644,
      "grad_norm": 2.662325493976984,
      "learning_rate": 1.3814024117033773e-08,
      "loss": 0.6265,
      "step": 8454
    },
    {
      "epoch": 0.8358253218990188,
      "grad_norm": 4.339811366529225,
      "learning_rate": 1.3797789128181292e-08,
      "loss": 0.6834,
      "step": 8455
    },
    {
      "epoch": 0.8359241776437734,
      "grad_norm": 6.49488273797302,
      "learning_rate": 1.3781562978093942e-08,
      "loss": 0.6891,
      "step": 8456
    },
    {
      "epoch": 0.8360230333885278,
      "grad_norm": 3.3915200866585966,
      "learning_rate": 1.3765345668435525e-08,
      "loss": 0.657,
      "step": 8457
    },
    {
      "epoch": 0.8361218891332822,
      "grad_norm": 2.9120639006544327,
      "learning_rate": 1.3749137200868843e-08,
      "loss": 0.6626,
      "step": 8458
    },
    {
      "epoch": 0.8362207448780368,
      "grad_norm": 4.769168923177625,
      "learning_rate": 1.3732937577055847e-08,
      "loss": 0.8156,
      "step": 8459
    },
    {
      "epoch": 0.8363196006227912,
      "grad_norm": 3.625457161607391,
      "learning_rate": 1.3716746798657564e-08,
      "loss": 0.6153,
      "step": 8460
    },
    {
      "epoch": 0.8364184563675456,
      "grad_norm": 6.365248432688433,
      "learning_rate": 1.3700564867334153e-08,
      "loss": 0.7382,
      "step": 8461
    },
    {
      "epoch": 0.8365173121123001,
      "grad_norm": 3.3760188542107685,
      "learning_rate": 1.3684391784744787e-08,
      "loss": 0.6985,
      "step": 8462
    },
    {
      "epoch": 0.8366161678570546,
      "grad_norm": 3.343502248958111,
      "learning_rate": 1.3668227552547818e-08,
      "loss": 0.661,
      "step": 8463
    },
    {
      "epoch": 0.836715023601809,
      "grad_norm": 3.6543215439529813,
      "learning_rate": 1.3652072172400608e-08,
      "loss": 0.6541,
      "step": 8464
    },
    {
      "epoch": 0.8368138793465635,
      "grad_norm": 5.145635721245211,
      "learning_rate": 1.3635925645959678e-08,
      "loss": 0.6433,
      "step": 8465
    },
    {
      "epoch": 0.836912735091318,
      "grad_norm": 3.1973631932779685,
      "learning_rate": 1.3619787974880637e-08,
      "loss": 0.7136,
      "step": 8466
    },
    {
      "epoch": 0.8370115908360725,
      "grad_norm": 4.698718547369085,
      "learning_rate": 1.3603659160818126e-08,
      "loss": 0.8074,
      "step": 8467
    },
    {
      "epoch": 0.8371104465808269,
      "grad_norm": 3.8499600304409385,
      "learning_rate": 1.3587539205425924e-08,
      "loss": 0.733,
      "step": 8468
    },
    {
      "epoch": 0.8372093023255814,
      "grad_norm": 4.45241522097495,
      "learning_rate": 1.3571428110356919e-08,
      "loss": 0.652,
      "step": 8469
    },
    {
      "epoch": 0.8373081580703359,
      "grad_norm": 3.506235294495939,
      "learning_rate": 1.3555325877263036e-08,
      "loss": 0.7324,
      "step": 8470
    },
    {
      "epoch": 0.8374070138150903,
      "grad_norm": 4.830960847165661,
      "learning_rate": 1.3539232507795328e-08,
      "loss": 0.6316,
      "step": 8471
    },
    {
      "epoch": 0.8375058695598449,
      "grad_norm": 3.8574745584025547,
      "learning_rate": 1.3523148003603979e-08,
      "loss": 0.5847,
      "step": 8472
    },
    {
      "epoch": 0.8376047253045993,
      "grad_norm": 5.205343142494865,
      "learning_rate": 1.3507072366338146e-08,
      "loss": 0.7532,
      "step": 8473
    },
    {
      "epoch": 0.8377035810493537,
      "grad_norm": 3.9034086923723796,
      "learning_rate": 1.34910055976462e-08,
      "loss": 0.7088,
      "step": 8474
    },
    {
      "epoch": 0.8378024367941082,
      "grad_norm": 17.015037699502475,
      "learning_rate": 1.3474947699175532e-08,
      "loss": 0.6922,
      "step": 8475
    },
    {
      "epoch": 0.8379012925388627,
      "grad_norm": 3.5478827209781145,
      "learning_rate": 1.3458898672572627e-08,
      "loss": 0.7484,
      "step": 8476
    },
    {
      "epoch": 0.8380001482836171,
      "grad_norm": 6.2084265076394765,
      "learning_rate": 1.3442858519483114e-08,
      "loss": 0.8831,
      "step": 8477
    },
    {
      "epoch": 0.8380990040283716,
      "grad_norm": 4.606604779187436,
      "learning_rate": 1.3426827241551631e-08,
      "loss": 0.712,
      "step": 8478
    },
    {
      "epoch": 0.8381978597731261,
      "grad_norm": 7.05882036168716,
      "learning_rate": 1.3410804840421963e-08,
      "loss": 0.7525,
      "step": 8479
    },
    {
      "epoch": 0.8382967155178805,
      "grad_norm": 3.7082707181826007,
      "learning_rate": 1.3394791317737009e-08,
      "loss": 0.7107,
      "step": 8480
    },
    {
      "epoch": 0.838395571262635,
      "grad_norm": 3.939137037292204,
      "learning_rate": 1.3378786675138653e-08,
      "loss": 0.6796,
      "step": 8481
    },
    {
      "epoch": 0.8384944270073895,
      "grad_norm": 3.379979743229636,
      "learning_rate": 1.3362790914267985e-08,
      "loss": 0.7316,
      "step": 8482
    },
    {
      "epoch": 0.838593282752144,
      "grad_norm": 3.4161665274230186,
      "learning_rate": 1.3346804036765137e-08,
      "loss": 0.739,
      "step": 8483
    },
    {
      "epoch": 0.8386921384968984,
      "grad_norm": 5.53846879180623,
      "learning_rate": 1.3330826044269295e-08,
      "loss": 0.6174,
      "step": 8484
    },
    {
      "epoch": 0.8387909942416528,
      "grad_norm": 2.9454603068462815,
      "learning_rate": 1.3314856938418795e-08,
      "loss": 0.7308,
      "step": 8485
    },
    {
      "epoch": 0.8388898499864074,
      "grad_norm": 5.081424152232738,
      "learning_rate": 1.3298896720850994e-08,
      "loss": 0.7607,
      "step": 8486
    },
    {
      "epoch": 0.8389887057311618,
      "grad_norm": 3.9489857938144985,
      "learning_rate": 1.3282945393202416e-08,
      "loss": 0.6531,
      "step": 8487
    },
    {
      "epoch": 0.8390875614759162,
      "grad_norm": 4.270815222393693,
      "learning_rate": 1.326700295710863e-08,
      "loss": 0.7123,
      "step": 8488
    },
    {
      "epoch": 0.8391864172206708,
      "grad_norm": 4.275710869665043,
      "learning_rate": 1.3251069414204274e-08,
      "loss": 0.7499,
      "step": 8489
    },
    {
      "epoch": 0.8392852729654252,
      "grad_norm": 4.538386486686956,
      "learning_rate": 1.3235144766123118e-08,
      "loss": 0.7852,
      "step": 8490
    },
    {
      "epoch": 0.8393841287101796,
      "grad_norm": 8.379906984987116,
      "learning_rate": 1.3219229014498012e-08,
      "loss": 0.7224,
      "step": 8491
    },
    {
      "epoch": 0.8394829844549342,
      "grad_norm": 5.512766459256944,
      "learning_rate": 1.3203322160960861e-08,
      "loss": 0.6629,
      "step": 8492
    },
    {
      "epoch": 0.8395818401996886,
      "grad_norm": 4.580272921466746,
      "learning_rate": 1.318742420714266e-08,
      "loss": 0.8011,
      "step": 8493
    },
    {
      "epoch": 0.8396806959444431,
      "grad_norm": 3.270600316469669,
      "learning_rate": 1.3171535154673553e-08,
      "loss": 0.6766,
      "step": 8494
    },
    {
      "epoch": 0.8397795516891975,
      "grad_norm": 4.590350668459391,
      "learning_rate": 1.3155655005182687e-08,
      "loss": 0.7174,
      "step": 8495
    },
    {
      "epoch": 0.839878407433952,
      "grad_norm": 4.540651697909307,
      "learning_rate": 1.3139783760298372e-08,
      "loss": 0.752,
      "step": 8496
    },
    {
      "epoch": 0.8399772631787065,
      "grad_norm": 4.947158340172083,
      "learning_rate": 1.3123921421647944e-08,
      "loss": 0.7698,
      "step": 8497
    },
    {
      "epoch": 0.8400761189234609,
      "grad_norm": 3.8723219195540612,
      "learning_rate": 1.3108067990857852e-08,
      "loss": 0.6403,
      "step": 8498
    },
    {
      "epoch": 0.8401749746682154,
      "grad_norm": 7.316772989313302,
      "learning_rate": 1.3092223469553665e-08,
      "loss": 0.7771,
      "step": 8499
    },
    {
      "epoch": 0.8402738304129699,
      "grad_norm": 4.793960901709731,
      "learning_rate": 1.307638785935996e-08,
      "loss": 0.7454,
      "step": 8500
    },
    {
      "epoch": 0.8403726861577243,
      "grad_norm": 4.996272945539198,
      "learning_rate": 1.3060561161900462e-08,
      "loss": 0.7175,
      "step": 8501
    },
    {
      "epoch": 0.8404715419024789,
      "grad_norm": 5.955794121651958,
      "learning_rate": 1.3044743378798007e-08,
      "loss": 0.7072,
      "step": 8502
    },
    {
      "epoch": 0.8405703976472333,
      "grad_norm": 5.220372225578878,
      "learning_rate": 1.3028934511674405e-08,
      "loss": 0.5918,
      "step": 8503
    },
    {
      "epoch": 0.8406692533919877,
      "grad_norm": 5.654852020065491,
      "learning_rate": 1.3013134562150652e-08,
      "loss": 0.7096,
      "step": 8504
    },
    {
      "epoch": 0.8407681091367422,
      "grad_norm": 3.3924819294309105,
      "learning_rate": 1.2997343531846827e-08,
      "loss": 0.7123,
      "step": 8505
    },
    {
      "epoch": 0.8408669648814967,
      "grad_norm": 4.1706080685615765,
      "learning_rate": 1.2981561422382026e-08,
      "loss": 0.6254,
      "step": 8506
    },
    {
      "epoch": 0.8409658206262511,
      "grad_norm": 4.823449455314663,
      "learning_rate": 1.2965788235374508e-08,
      "loss": 0.7037,
      "step": 8507
    },
    {
      "epoch": 0.8410646763710056,
      "grad_norm": 3.764854631961377,
      "learning_rate": 1.2950023972441538e-08,
      "loss": 0.7639,
      "step": 8508
    },
    {
      "epoch": 0.8411635321157601,
      "grad_norm": 4.163932951807543,
      "learning_rate": 1.2934268635199552e-08,
      "loss": 0.5572,
      "step": 8509
    },
    {
      "epoch": 0.8412623878605145,
      "grad_norm": 3.1377234721187564,
      "learning_rate": 1.2918522225263995e-08,
      "loss": 0.8255,
      "step": 8510
    },
    {
      "epoch": 0.841361243605269,
      "grad_norm": 4.194937521687677,
      "learning_rate": 1.2902784744249462e-08,
      "loss": 0.7952,
      "step": 8511
    },
    {
      "epoch": 0.8414600993500235,
      "grad_norm": 2.8859037235084193,
      "learning_rate": 1.2887056193769553e-08,
      "loss": 0.7061,
      "step": 8512
    },
    {
      "epoch": 0.841558955094778,
      "grad_norm": 4.0230617530772,
      "learning_rate": 1.2871336575437042e-08,
      "loss": 0.634,
      "step": 8513
    },
    {
      "epoch": 0.8416578108395324,
      "grad_norm": 4.350890707939826,
      "learning_rate": 1.2855625890863708e-08,
      "loss": 0.7305,
      "step": 8514
    },
    {
      "epoch": 0.8417566665842868,
      "grad_norm": 4.279562089389925,
      "learning_rate": 1.2839924141660474e-08,
      "loss": 0.7091,
      "step": 8515
    },
    {
      "epoch": 0.8418555223290414,
      "grad_norm": 3.094083782710582,
      "learning_rate": 1.282423132943734e-08,
      "loss": 0.7007,
      "step": 8516
    },
    {
      "epoch": 0.8419543780737958,
      "grad_norm": 10.00209648917891,
      "learning_rate": 1.280854745580332e-08,
      "loss": 0.7592,
      "step": 8517
    },
    {
      "epoch": 0.8420532338185502,
      "grad_norm": 4.691680850694479,
      "learning_rate": 1.2792872522366626e-08,
      "loss": 0.6592,
      "step": 8518
    },
    {
      "epoch": 0.8421520895633048,
      "grad_norm": 4.4568290333283835,
      "learning_rate": 1.277720653073443e-08,
      "loss": 0.8316,
      "step": 8519
    },
    {
      "epoch": 0.8422509453080592,
      "grad_norm": 4.839182414328947,
      "learning_rate": 1.2761549482513078e-08,
      "loss": 0.7016,
      "step": 8520
    },
    {
      "epoch": 0.8423498010528137,
      "grad_norm": 3.4799222506666845,
      "learning_rate": 1.2745901379307988e-08,
      "loss": 0.5869,
      "step": 8521
    },
    {
      "epoch": 0.8424486567975682,
      "grad_norm": 57.20596102085202,
      "learning_rate": 1.2730262222723609e-08,
      "loss": 0.7077,
      "step": 8522
    },
    {
      "epoch": 0.8425475125423226,
      "grad_norm": 7.221651336927589,
      "learning_rate": 1.2714632014363513e-08,
      "loss": 0.696,
      "step": 8523
    },
    {
      "epoch": 0.8426463682870771,
      "grad_norm": 4.122182628922068,
      "learning_rate": 1.2699010755830385e-08,
      "loss": 0.648,
      "step": 8524
    },
    {
      "epoch": 0.8427452240318315,
      "grad_norm": 4.925281220369655,
      "learning_rate": 1.2683398448725891e-08,
      "loss": 0.7535,
      "step": 8525
    },
    {
      "epoch": 0.842844079776586,
      "grad_norm": 6.640910939568281,
      "learning_rate": 1.2667795094650901e-08,
      "loss": 0.6975,
      "step": 8526
    },
    {
      "epoch": 0.8429429355213405,
      "grad_norm": 5.006651811746319,
      "learning_rate": 1.2652200695205273e-08,
      "loss": 0.7749,
      "step": 8527
    },
    {
      "epoch": 0.8430417912660949,
      "grad_norm": 5.701928785931822,
      "learning_rate": 1.2636615251988003e-08,
      "loss": 0.793,
      "step": 8528
    },
    {
      "epoch": 0.8431406470108495,
      "grad_norm": 4.999523507117406,
      "learning_rate": 1.2621038766597137e-08,
      "loss": 0.7035,
      "step": 8529
    },
    {
      "epoch": 0.8432395027556039,
      "grad_norm": 3.2234117566514855,
      "learning_rate": 1.2605471240629796e-08,
      "loss": 0.6736,
      "step": 8530
    },
    {
      "epoch": 0.8433383585003583,
      "grad_norm": 4.620812787634261,
      "learning_rate": 1.2589912675682224e-08,
      "loss": 0.6973,
      "step": 8531
    },
    {
      "epoch": 0.8434372142451129,
      "grad_norm": 4.466336685202424,
      "learning_rate": 1.2574363073349737e-08,
      "loss": 0.6481,
      "step": 8532
    },
    {
      "epoch": 0.8435360699898673,
      "grad_norm": 5.167082200396749,
      "learning_rate": 1.2558822435226667e-08,
      "loss": 0.7195,
      "step": 8533
    },
    {
      "epoch": 0.8436349257346217,
      "grad_norm": 4.798241891567873,
      "learning_rate": 1.2543290762906522e-08,
      "loss": 0.6207,
      "step": 8534
    },
    {
      "epoch": 0.8437337814793762,
      "grad_norm": 3.373334173747111,
      "learning_rate": 1.2527768057981847e-08,
      "loss": 0.7724,
      "step": 8535
    },
    {
      "epoch": 0.8438326372241307,
      "grad_norm": 5.255831231422743,
      "learning_rate": 1.2512254322044224e-08,
      "loss": 0.6442,
      "step": 8536
    },
    {
      "epoch": 0.8439314929688851,
      "grad_norm": 4.3807112490341735,
      "learning_rate": 1.2496749556684393e-08,
      "loss": 0.6671,
      "step": 8537
    },
    {
      "epoch": 0.8440303487136396,
      "grad_norm": 3.6376313653245447,
      "learning_rate": 1.248125376349215e-08,
      "loss": 0.6761,
      "step": 8538
    },
    {
      "epoch": 0.8441292044583941,
      "grad_norm": 8.431564354772107,
      "learning_rate": 1.2465766944056322e-08,
      "loss": 0.6509,
      "step": 8539
    },
    {
      "epoch": 0.8442280602031486,
      "grad_norm": 3.3603642987362212,
      "learning_rate": 1.2450289099964895e-08,
      "loss": 0.715,
      "step": 8540
    },
    {
      "epoch": 0.844326915947903,
      "grad_norm": 3.804290592171711,
      "learning_rate": 1.2434820232804844e-08,
      "loss": 0.71,
      "step": 8541
    },
    {
      "epoch": 0.8444257716926575,
      "grad_norm": 4.205808905865243,
      "learning_rate": 1.241936034416231e-08,
      "loss": 0.7089,
      "step": 8542
    },
    {
      "epoch": 0.844524627437412,
      "grad_norm": 3.031134677061193,
      "learning_rate": 1.2403909435622484e-08,
      "loss": 0.7164,
      "step": 8543
    },
    {
      "epoch": 0.8446234831821664,
      "grad_norm": 49.90387611291004,
      "learning_rate": 1.2388467508769595e-08,
      "loss": 0.6565,
      "step": 8544
    },
    {
      "epoch": 0.844722338926921,
      "grad_norm": 13.074588240419608,
      "learning_rate": 1.2373034565187013e-08,
      "loss": 0.7075,
      "step": 8545
    },
    {
      "epoch": 0.8448211946716754,
      "grad_norm": 3.321670878155703,
      "learning_rate": 1.2357610606457158e-08,
      "loss": 0.6846,
      "step": 8546
    },
    {
      "epoch": 0.8449200504164298,
      "grad_norm": 3.6414581273701243,
      "learning_rate": 1.2342195634161489e-08,
      "loss": 0.7506,
      "step": 8547
    },
    {
      "epoch": 0.8450189061611842,
      "grad_norm": 2.9716924480151454,
      "learning_rate": 1.232678964988062e-08,
      "loss": 0.7168,
      "step": 8548
    },
    {
      "epoch": 0.8451177619059388,
      "grad_norm": 6.9095744220983555,
      "learning_rate": 1.231139265519422e-08,
      "loss": 0.7318,
      "step": 8549
    },
    {
      "epoch": 0.8452166176506932,
      "grad_norm": 7.8671277712383025,
      "learning_rate": 1.2296004651680992e-08,
      "loss": 0.6489,
      "step": 8550
    },
    {
      "epoch": 0.8453154733954477,
      "grad_norm": 13.662821910573875,
      "learning_rate": 1.2280625640918774e-08,
      "loss": 0.6608,
      "step": 8551
    },
    {
      "epoch": 0.8454143291402022,
      "grad_norm": 15.794476588832703,
      "learning_rate": 1.2265255624484427e-08,
      "loss": 0.677,
      "step": 8552
    },
    {
      "epoch": 0.8455131848849566,
      "grad_norm": 6.246189058887154,
      "learning_rate": 1.2249894603953937e-08,
      "loss": 0.7408,
      "step": 8553
    },
    {
      "epoch": 0.8456120406297111,
      "grad_norm": 3.553400025242204,
      "learning_rate": 1.2234542580902373e-08,
      "loss": 0.7701,
      "step": 8554
    },
    {
      "epoch": 0.8457108963744656,
      "grad_norm": 3.4843861516162917,
      "learning_rate": 1.2219199556903815e-08,
      "loss": 0.6451,
      "step": 8555
    },
    {
      "epoch": 0.84580975211922,
      "grad_norm": 3.5882770897965024,
      "learning_rate": 1.2203865533531477e-08,
      "loss": 0.6034,
      "step": 8556
    },
    {
      "epoch": 0.8459086078639745,
      "grad_norm": 3.4719375369650294,
      "learning_rate": 1.218854051235767e-08,
      "loss": 0.7745,
      "step": 8557
    },
    {
      "epoch": 0.8460074636087289,
      "grad_norm": 3.962421733725962,
      "learning_rate": 1.2173224494953705e-08,
      "loss": 0.7876,
      "step": 8558
    },
    {
      "epoch": 0.8461063193534835,
      "grad_norm": 8.44009361247056,
      "learning_rate": 1.2157917482890023e-08,
      "loss": 0.6846,
      "step": 8559
    },
    {
      "epoch": 0.8462051750982379,
      "grad_norm": 3.4554090703811045,
      "learning_rate": 1.2142619477736172e-08,
      "loss": 0.6856,
      "step": 8560
    },
    {
      "epoch": 0.8463040308429923,
      "grad_norm": 3.931842630883542,
      "learning_rate": 1.2127330481060682e-08,
      "loss": 0.767,
      "step": 8561
    },
    {
      "epoch": 0.8464028865877469,
      "grad_norm": 3.0103973945070504,
      "learning_rate": 1.2112050494431259e-08,
      "loss": 0.7861,
      "step": 8562
    },
    {
      "epoch": 0.8465017423325013,
      "grad_norm": 3.782932433127624,
      "learning_rate": 1.2096779519414624e-08,
      "loss": 0.8176,
      "step": 8563
    },
    {
      "epoch": 0.8466005980772557,
      "grad_norm": 8.897153196433312,
      "learning_rate": 1.2081517557576559e-08,
      "loss": 0.6339,
      "step": 8564
    },
    {
      "epoch": 0.8466994538220103,
      "grad_norm": 20.42593798197291,
      "learning_rate": 1.2066264610482002e-08,
      "loss": 0.7768,
      "step": 8565
    },
    {
      "epoch": 0.8467983095667647,
      "grad_norm": 14.78339119240879,
      "learning_rate": 1.2051020679694868e-08,
      "loss": 0.6685,
      "step": 8566
    },
    {
      "epoch": 0.8468971653115192,
      "grad_norm": 4.5203314724947194,
      "learning_rate": 1.2035785766778238e-08,
      "loss": 0.7724,
      "step": 8567
    },
    {
      "epoch": 0.8469960210562736,
      "grad_norm": 3.473537978596026,
      "learning_rate": 1.2020559873294223e-08,
      "loss": 0.6616,
      "step": 8568
    },
    {
      "epoch": 0.8470948768010281,
      "grad_norm": 7.097008051439177,
      "learning_rate": 1.200534300080399e-08,
      "loss": 0.7305,
      "step": 8569
    },
    {
      "epoch": 0.8471937325457826,
      "grad_norm": 14.791652840034502,
      "learning_rate": 1.199013515086782e-08,
      "loss": 0.7885,
      "step": 8570
    },
    {
      "epoch": 0.847292588290537,
      "grad_norm": 3.176442937947911,
      "learning_rate": 1.1974936325045071e-08,
      "loss": 0.73,
      "step": 8571
    },
    {
      "epoch": 0.8473914440352915,
      "grad_norm": 5.630353787687965,
      "learning_rate": 1.195974652489412e-08,
      "loss": 0.7074,
      "step": 8572
    },
    {
      "epoch": 0.847490299780046,
      "grad_norm": 4.857311683132986,
      "learning_rate": 1.1944565751972502e-08,
      "loss": 0.6683,
      "step": 8573
    },
    {
      "epoch": 0.8475891555248004,
      "grad_norm": 3.730590729990339,
      "learning_rate": 1.1929394007836734e-08,
      "loss": 0.6062,
      "step": 8574
    },
    {
      "epoch": 0.847688011269555,
      "grad_norm": 5.1433800661819165,
      "learning_rate": 1.1914231294042476e-08,
      "loss": 0.7404,
      "step": 8575
    },
    {
      "epoch": 0.8477868670143094,
      "grad_norm": 3.644935348443862,
      "learning_rate": 1.189907761214447e-08,
      "loss": 0.6446,
      "step": 8576
    },
    {
      "epoch": 0.8478857227590638,
      "grad_norm": 3.828507428914757,
      "learning_rate": 1.1883932963696452e-08,
      "loss": 0.6128,
      "step": 8577
    },
    {
      "epoch": 0.8479845785038183,
      "grad_norm": 3.8150320860946976,
      "learning_rate": 1.1868797350251313e-08,
      "loss": 0.7264,
      "step": 8578
    },
    {
      "epoch": 0.8480834342485728,
      "grad_norm": 3.1862804013382777,
      "learning_rate": 1.1853670773360968e-08,
      "loss": 0.7476,
      "step": 8579
    },
    {
      "epoch": 0.8481822899933272,
      "grad_norm": 2.9545089345263453,
      "learning_rate": 1.1838553234576465e-08,
      "loss": 0.743,
      "step": 8580
    },
    {
      "epoch": 0.8482811457380817,
      "grad_norm": 3.72739591912932,
      "learning_rate": 1.1823444735447818e-08,
      "loss": 0.5657,
      "step": 8581
    },
    {
      "epoch": 0.8483800014828362,
      "grad_norm": 3.220837083445289,
      "learning_rate": 1.1808345277524257e-08,
      "loss": 0.6392,
      "step": 8582
    },
    {
      "epoch": 0.8484788572275906,
      "grad_norm": 3.773286984151342,
      "learning_rate": 1.179325486235393e-08,
      "loss": 0.6494,
      "step": 8583
    },
    {
      "epoch": 0.8485777129723451,
      "grad_norm": 4.925270975620759,
      "learning_rate": 1.177817349148421e-08,
      "loss": 0.6951,
      "step": 8584
    },
    {
      "epoch": 0.8486765687170996,
      "grad_norm": 3.0739840171724744,
      "learning_rate": 1.176310116646141e-08,
      "loss": 0.6527,
      "step": 8585
    },
    {
      "epoch": 0.8487754244618541,
      "grad_norm": 4.418622425277918,
      "learning_rate": 1.1748037888830998e-08,
      "loss": 0.6535,
      "step": 8586
    },
    {
      "epoch": 0.8488742802066085,
      "grad_norm": 3.5878343045202077,
      "learning_rate": 1.1732983660137508e-08,
      "loss": 0.7181,
      "step": 8587
    },
    {
      "epoch": 0.8489731359513629,
      "grad_norm": 3.324750742044803,
      "learning_rate": 1.1717938481924494e-08,
      "loss": 0.8038,
      "step": 8588
    },
    {
      "epoch": 0.8490719916961175,
      "grad_norm": 6.758118662821149,
      "learning_rate": 1.1702902355734645e-08,
      "loss": 0.7342,
      "step": 8589
    },
    {
      "epoch": 0.8491708474408719,
      "grad_norm": 4.233508942378558,
      "learning_rate": 1.1687875283109705e-08,
      "loss": 0.6805,
      "step": 8590
    },
    {
      "epoch": 0.8492697031856263,
      "grad_norm": 4.846403369794062,
      "learning_rate": 1.167285726559043e-08,
      "loss": 0.6969,
      "step": 8591
    },
    {
      "epoch": 0.8493685589303809,
      "grad_norm": 6.805190738379136,
      "learning_rate": 1.1657848304716733e-08,
      "loss": 0.6977,
      "step": 8592
    },
    {
      "epoch": 0.8494674146751353,
      "grad_norm": 4.513462623217717,
      "learning_rate": 1.1642848402027572e-08,
      "loss": 0.7345,
      "step": 8593
    },
    {
      "epoch": 0.8495662704198897,
      "grad_norm": 3.129360204356596,
      "learning_rate": 1.1627857559060927e-08,
      "loss": 0.7411,
      "step": 8594
    },
    {
      "epoch": 0.8496651261646443,
      "grad_norm": 5.513847916834339,
      "learning_rate": 1.1612875777353925e-08,
      "loss": 0.7225,
      "step": 8595
    },
    {
      "epoch": 0.8497639819093987,
      "grad_norm": 4.173027093027577,
      "learning_rate": 1.1597903058442704e-08,
      "loss": 0.6668,
      "step": 8596
    },
    {
      "epoch": 0.8498628376541532,
      "grad_norm": 5.69921067371925,
      "learning_rate": 1.1582939403862512e-08,
      "loss": 0.7004,
      "step": 8597
    },
    {
      "epoch": 0.8499616933989076,
      "grad_norm": 3.7406242782703405,
      "learning_rate": 1.1567984815147614e-08,
      "loss": 0.7045,
      "step": 8598
    },
    {
      "epoch": 0.8500605491436621,
      "grad_norm": 3.530349804704448,
      "learning_rate": 1.1553039293831435e-08,
      "loss": 0.7452,
      "step": 8599
    },
    {
      "epoch": 0.8501594048884166,
      "grad_norm": 16.55668336937637,
      "learning_rate": 1.1538102841446373e-08,
      "loss": 0.6558,
      "step": 8600
    },
    {
      "epoch": 0.850258260633171,
      "grad_norm": 3.2796493944414578,
      "learning_rate": 1.1523175459523981e-08,
      "loss": 0.7255,
      "step": 8601
    },
    {
      "epoch": 0.8503571163779255,
      "grad_norm": 3.90340212671732,
      "learning_rate": 1.1508257149594802e-08,
      "loss": 0.7148,
      "step": 8602
    },
    {
      "epoch": 0.85045597212268,
      "grad_norm": 5.59878708680492,
      "learning_rate": 1.14933479131885e-08,
      "loss": 0.6661,
      "step": 8603
    },
    {
      "epoch": 0.8505548278674344,
      "grad_norm": 3.428531348637977,
      "learning_rate": 1.1478447751833831e-08,
      "loss": 0.6818,
      "step": 8604
    },
    {
      "epoch": 0.850653683612189,
      "grad_norm": 5.738782412316438,
      "learning_rate": 1.1463556667058538e-08,
      "loss": 0.6314,
      "step": 8605
    },
    {
      "epoch": 0.8507525393569434,
      "grad_norm": 5.637009291767376,
      "learning_rate": 1.1448674660389523e-08,
      "loss": 0.767,
      "step": 8606
    },
    {
      "epoch": 0.8508513951016978,
      "grad_norm": 6.404340845261244,
      "learning_rate": 1.1433801733352688e-08,
      "loss": 0.6912,
      "step": 8607
    },
    {
      "epoch": 0.8509502508464523,
      "grad_norm": 8.829665197552877,
      "learning_rate": 1.1418937887473024e-08,
      "loss": 0.6911,
      "step": 8608
    },
    {
      "epoch": 0.8510491065912068,
      "grad_norm": 3.4425234094167108,
      "learning_rate": 1.1404083124274655e-08,
      "loss": 0.7076,
      "step": 8609
    },
    {
      "epoch": 0.8511479623359612,
      "grad_norm": 3.904077261974068,
      "learning_rate": 1.1389237445280652e-08,
      "loss": 0.7606,
      "step": 8610
    },
    {
      "epoch": 0.8512468180807157,
      "grad_norm": 3.2837413601800676,
      "learning_rate": 1.1374400852013244e-08,
      "loss": 0.8117,
      "step": 8611
    },
    {
      "epoch": 0.8513456738254702,
      "grad_norm": 4.116250777135752,
      "learning_rate": 1.1359573345993734e-08,
      "loss": 0.6509,
      "step": 8612
    },
    {
      "epoch": 0.8514445295702247,
      "grad_norm": 9.60842227449942,
      "learning_rate": 1.1344754928742417e-08,
      "loss": 0.6632,
      "step": 8613
    },
    {
      "epoch": 0.8515433853149791,
      "grad_norm": 11.44916089302077,
      "learning_rate": 1.1329945601778745e-08,
      "loss": 0.7044,
      "step": 8614
    },
    {
      "epoch": 0.8516422410597336,
      "grad_norm": 4.6621759953591795,
      "learning_rate": 1.1315145366621159e-08,
      "loss": 0.6589,
      "step": 8615
    },
    {
      "epoch": 0.8517410968044881,
      "grad_norm": 3.939092411665556,
      "learning_rate": 1.1300354224787234e-08,
      "loss": 0.7315,
      "step": 8616
    },
    {
      "epoch": 0.8518399525492425,
      "grad_norm": 3.993345063548697,
      "learning_rate": 1.1285572177793579e-08,
      "loss": 0.6842,
      "step": 8617
    },
    {
      "epoch": 0.8519388082939969,
      "grad_norm": 17.178478847529327,
      "learning_rate": 1.1270799227155847e-08,
      "loss": 0.7377,
      "step": 8618
    },
    {
      "epoch": 0.8520376640387515,
      "grad_norm": 4.166605608921774,
      "learning_rate": 1.1256035374388805e-08,
      "loss": 0.6778,
      "step": 8619
    },
    {
      "epoch": 0.8521365197835059,
      "grad_norm": 3.0084052009049773,
      "learning_rate": 1.1241280621006288e-08,
      "loss": 0.6333,
      "step": 8620
    },
    {
      "epoch": 0.8522353755282603,
      "grad_norm": 7.911728728347471,
      "learning_rate": 1.122653496852114e-08,
      "loss": 0.6219,
      "step": 8621
    },
    {
      "epoch": 0.8523342312730149,
      "grad_norm": 4.438244012375608,
      "learning_rate": 1.121179841844534e-08,
      "loss": 0.7634,
      "step": 8622
    },
    {
      "epoch": 0.8524330870177693,
      "grad_norm": 5.170051539577789,
      "learning_rate": 1.1197070972289902e-08,
      "loss": 0.7558,
      "step": 8623
    },
    {
      "epoch": 0.8525319427625238,
      "grad_norm": 4.841628033527822,
      "learning_rate": 1.1182352631564885e-08,
      "loss": 0.8701,
      "step": 8624
    },
    {
      "epoch": 0.8526307985072783,
      "grad_norm": 2.7671423091890093,
      "learning_rate": 1.1167643397779458e-08,
      "loss": 0.7993,
      "step": 8625
    },
    {
      "epoch": 0.8527296542520327,
      "grad_norm": 4.93382937269188,
      "learning_rate": 1.115294327244184e-08,
      "loss": 0.753,
      "step": 8626
    },
    {
      "epoch": 0.8528285099967872,
      "grad_norm": 2.9870770133848277,
      "learning_rate": 1.1138252257059289e-08,
      "loss": 0.7439,
      "step": 8627
    },
    {
      "epoch": 0.8529273657415417,
      "grad_norm": 4.878601645057894,
      "learning_rate": 1.112357035313819e-08,
      "loss": 0.8117,
      "step": 8628
    },
    {
      "epoch": 0.8530262214862961,
      "grad_norm": 9.527334246902315,
      "learning_rate": 1.1108897562183906e-08,
      "loss": 0.6226,
      "step": 8629
    },
    {
      "epoch": 0.8531250772310506,
      "grad_norm": 29.186506763431137,
      "learning_rate": 1.1094233885700942e-08,
      "loss": 0.749,
      "step": 8630
    },
    {
      "epoch": 0.853223932975805,
      "grad_norm": 2.9975181593125733,
      "learning_rate": 1.1079579325192856e-08,
      "loss": 0.668,
      "step": 8631
    },
    {
      "epoch": 0.8533227887205596,
      "grad_norm": 3.1361771833850334,
      "learning_rate": 1.1064933882162231e-08,
      "loss": 0.7028,
      "step": 8632
    },
    {
      "epoch": 0.853421644465314,
      "grad_norm": 3.7025041790769886,
      "learning_rate": 1.1050297558110767e-08,
      "loss": 0.7708,
      "step": 8633
    },
    {
      "epoch": 0.8535205002100684,
      "grad_norm": 9.237076297921618,
      "learning_rate": 1.1035670354539195e-08,
      "loss": 0.7108,
      "step": 8634
    },
    {
      "epoch": 0.853619355954823,
      "grad_norm": 2.8953670503357842,
      "learning_rate": 1.1021052272947285e-08,
      "loss": 0.7656,
      "step": 8635
    },
    {
      "epoch": 0.8537182116995774,
      "grad_norm": 4.671584040102387,
      "learning_rate": 1.1006443314833936e-08,
      "loss": 0.6948,
      "step": 8636
    },
    {
      "epoch": 0.8538170674443318,
      "grad_norm": 3.840536118442229,
      "learning_rate": 1.0991843481697094e-08,
      "loss": 0.7001,
      "step": 8637
    },
    {
      "epoch": 0.8539159231890864,
      "grad_norm": 3.1399487003276088,
      "learning_rate": 1.0977252775033718e-08,
      "loss": 0.6816,
      "step": 8638
    },
    {
      "epoch": 0.8540147789338408,
      "grad_norm": 6.4489357880989955,
      "learning_rate": 1.0962671196339912e-08,
      "loss": 0.6689,
      "step": 8639
    },
    {
      "epoch": 0.8541136346785952,
      "grad_norm": 3.658538642877313,
      "learning_rate": 1.0948098747110768e-08,
      "loss": 0.7454,
      "step": 8640
    },
    {
      "epoch": 0.8542124904233497,
      "grad_norm": 6.049345703170962,
      "learning_rate": 1.093353542884049e-08,
      "loss": 0.7826,
      "step": 8641
    },
    {
      "epoch": 0.8543113461681042,
      "grad_norm": 12.073461188219037,
      "learning_rate": 1.0918981243022341e-08,
      "loss": 0.7132,
      "step": 8642
    },
    {
      "epoch": 0.8544102019128587,
      "grad_norm": 4.2754238788651415,
      "learning_rate": 1.0904436191148615e-08,
      "loss": 0.815,
      "step": 8643
    },
    {
      "epoch": 0.8545090576576131,
      "grad_norm": 4.469116543362615,
      "learning_rate": 1.0889900274710706e-08,
      "loss": 0.5443,
      "step": 8644
    },
    {
      "epoch": 0.8546079134023676,
      "grad_norm": 4.206560662009468,
      "learning_rate": 1.087537349519908e-08,
      "loss": 0.73,
      "step": 8645
    },
    {
      "epoch": 0.8547067691471221,
      "grad_norm": 3.743486051482117,
      "learning_rate": 1.0860855854103201e-08,
      "loss": 0.6457,
      "step": 8646
    },
    {
      "epoch": 0.8548056248918765,
      "grad_norm": 27.120450452041894,
      "learning_rate": 1.0846347352911655e-08,
      "loss": 0.6371,
      "step": 8647
    },
    {
      "epoch": 0.854904480636631,
      "grad_norm": 3.410331616972023,
      "learning_rate": 1.0831847993112097e-08,
      "loss": 0.6486,
      "step": 8648
    },
    {
      "epoch": 0.8550033363813855,
      "grad_norm": 21.247585100504924,
      "learning_rate": 1.0817357776191204e-08,
      "loss": 0.6255,
      "step": 8649
    },
    {
      "epoch": 0.8551021921261399,
      "grad_norm": 3.074339590690305,
      "learning_rate": 1.0802876703634722e-08,
      "loss": 0.6622,
      "step": 8650
    },
    {
      "epoch": 0.8552010478708943,
      "grad_norm": 5.387176652551241,
      "learning_rate": 1.0788404776927496e-08,
      "loss": 0.7492,
      "step": 8651
    },
    {
      "epoch": 0.8552999036156489,
      "grad_norm": 4.401914340613718,
      "learning_rate": 1.0773941997553382e-08,
      "loss": 0.6447,
      "step": 8652
    },
    {
      "epoch": 0.8553987593604033,
      "grad_norm": 3.0100601385974866,
      "learning_rate": 1.0759488366995362e-08,
      "loss": 0.7657,
      "step": 8653
    },
    {
      "epoch": 0.8554976151051578,
      "grad_norm": 3.486497545314362,
      "learning_rate": 1.0745043886735395e-08,
      "loss": 0.6473,
      "step": 8654
    },
    {
      "epoch": 0.8555964708499123,
      "grad_norm": 7.4827516360350925,
      "learning_rate": 1.0730608558254573e-08,
      "loss": 0.6849,
      "step": 8655
    },
    {
      "epoch": 0.8556953265946667,
      "grad_norm": 7.014862449870262,
      "learning_rate": 1.0716182383033045e-08,
      "loss": 0.6562,
      "step": 8656
    },
    {
      "epoch": 0.8557941823394212,
      "grad_norm": 3.2979276119402154,
      "learning_rate": 1.0701765362549975e-08,
      "loss": 0.7241,
      "step": 8657
    },
    {
      "epoch": 0.8558930380841757,
      "grad_norm": 14.119133956554663,
      "learning_rate": 1.0687357498283622e-08,
      "loss": 0.7519,
      "step": 8658
    },
    {
      "epoch": 0.8559918938289302,
      "grad_norm": 4.025196643932552,
      "learning_rate": 1.0672958791711317e-08,
      "loss": 0.7066,
      "step": 8659
    },
    {
      "epoch": 0.8560907495736846,
      "grad_norm": 3.872896239619561,
      "learning_rate": 1.0658569244309402e-08,
      "loss": 0.7395,
      "step": 8660
    },
    {
      "epoch": 0.856189605318439,
      "grad_norm": 2.913936092828878,
      "learning_rate": 1.064418885755336e-08,
      "loss": 0.735,
      "step": 8661
    },
    {
      "epoch": 0.8562884610631936,
      "grad_norm": 13.34846610700837,
      "learning_rate": 1.0629817632917637e-08,
      "loss": 0.6949,
      "step": 8662
    },
    {
      "epoch": 0.856387316807948,
      "grad_norm": 4.825349082215248,
      "learning_rate": 1.0615455571875809e-08,
      "loss": 0.6531,
      "step": 8663
    },
    {
      "epoch": 0.8564861725527024,
      "grad_norm": 4.243511534169478,
      "learning_rate": 1.060110267590052e-08,
      "loss": 0.7663,
      "step": 8664
    },
    {
      "epoch": 0.856585028297457,
      "grad_norm": 4.667084369448501,
      "learning_rate": 1.0586758946463403e-08,
      "loss": 0.7193,
      "step": 8665
    },
    {
      "epoch": 0.8566838840422114,
      "grad_norm": 4.682010805104338,
      "learning_rate": 1.0572424385035239e-08,
      "loss": 0.7389,
      "step": 8666
    },
    {
      "epoch": 0.8567827397869658,
      "grad_norm": 5.5331189247847625,
      "learning_rate": 1.0558098993085784e-08,
      "loss": 0.6884,
      "step": 8667
    },
    {
      "epoch": 0.8568815955317204,
      "grad_norm": 6.784267804794987,
      "learning_rate": 1.0543782772083932e-08,
      "loss": 0.7238,
      "step": 8668
    },
    {
      "epoch": 0.8569804512764748,
      "grad_norm": 4.22099583991144,
      "learning_rate": 1.0529475723497571e-08,
      "loss": 0.6956,
      "step": 8669
    },
    {
      "epoch": 0.8570793070212293,
      "grad_norm": 37.992527271641315,
      "learning_rate": 1.051517784879371e-08,
      "loss": 0.6799,
      "step": 8670
    },
    {
      "epoch": 0.8571781627659837,
      "grad_norm": 13.041609321636171,
      "learning_rate": 1.050088914943834e-08,
      "loss": 0.5856,
      "step": 8671
    },
    {
      "epoch": 0.8572770185107382,
      "grad_norm": 3.5033801180941797,
      "learning_rate": 1.0486609626896614e-08,
      "loss": 0.6064,
      "step": 8672
    },
    {
      "epoch": 0.8573758742554927,
      "grad_norm": 4.435807298808124,
      "learning_rate": 1.0472339282632637e-08,
      "loss": 0.8445,
      "step": 8673
    },
    {
      "epoch": 0.8574747300002471,
      "grad_norm": 4.469625416832599,
      "learning_rate": 1.0458078118109626e-08,
      "loss": 0.719,
      "step": 8674
    },
    {
      "epoch": 0.8575735857450016,
      "grad_norm": 4.4191596656962835,
      "learning_rate": 1.0443826134789901e-08,
      "loss": 0.7365,
      "step": 8675
    },
    {
      "epoch": 0.8576724414897561,
      "grad_norm": 3.3422166818409393,
      "learning_rate": 1.0429583334134739e-08,
      "loss": 0.7232,
      "step": 8676
    },
    {
      "epoch": 0.8577712972345105,
      "grad_norm": 3.414458561839941,
      "learning_rate": 1.0415349717604537e-08,
      "loss": 0.6365,
      "step": 8677
    },
    {
      "epoch": 0.857870152979265,
      "grad_norm": 2.718274859849102,
      "learning_rate": 1.0401125286658785e-08,
      "loss": 0.6888,
      "step": 8678
    },
    {
      "epoch": 0.8579690087240195,
      "grad_norm": 5.551944598949903,
      "learning_rate": 1.0386910042755937e-08,
      "loss": 0.7565,
      "step": 8679
    },
    {
      "epoch": 0.8580678644687739,
      "grad_norm": 4.4692023378354815,
      "learning_rate": 1.0372703987353581e-08,
      "loss": 0.8306,
      "step": 8680
    },
    {
      "epoch": 0.8581667202135284,
      "grad_norm": 6.454068231664784,
      "learning_rate": 1.0358507121908344e-08,
      "loss": 0.8514,
      "step": 8681
    },
    {
      "epoch": 0.8582655759582829,
      "grad_norm": 3.3796120106747316,
      "learning_rate": 1.0344319447875893e-08,
      "loss": 0.7532,
      "step": 8682
    },
    {
      "epoch": 0.8583644317030373,
      "grad_norm": 7.750928948459797,
      "learning_rate": 1.0330140966710976e-08,
      "loss": 0.6494,
      "step": 8683
    },
    {
      "epoch": 0.8584632874477918,
      "grad_norm": 8.252272232864279,
      "learning_rate": 1.0315971679867364e-08,
      "loss": 0.6477,
      "step": 8684
    },
    {
      "epoch": 0.8585621431925463,
      "grad_norm": 3.3608185956131633,
      "learning_rate": 1.0301811588797938e-08,
      "loss": 0.696,
      "step": 8685
    },
    {
      "epoch": 0.8586609989373007,
      "grad_norm": 3.8909509331029137,
      "learning_rate": 1.028766069495458e-08,
      "loss": 0.6406,
      "step": 8686
    },
    {
      "epoch": 0.8587598546820552,
      "grad_norm": 4.039903810264723,
      "learning_rate": 1.0273518999788288e-08,
      "loss": 0.6789,
      "step": 8687
    },
    {
      "epoch": 0.8588587104268097,
      "grad_norm": 5.380218236606968,
      "learning_rate": 1.0259386504749034e-08,
      "loss": 0.7093,
      "step": 8688
    },
    {
      "epoch": 0.8589575661715642,
      "grad_norm": 3.616318053299666,
      "learning_rate": 1.024526321128596e-08,
      "loss": 0.6764,
      "step": 8689
    },
    {
      "epoch": 0.8590564219163186,
      "grad_norm": 5.862041209264524,
      "learning_rate": 1.0231149120847137e-08,
      "loss": 0.7559,
      "step": 8690
    },
    {
      "epoch": 0.859155277661073,
      "grad_norm": 3.9265757929371294,
      "learning_rate": 1.021704423487979e-08,
      "loss": 0.7172,
      "step": 8691
    },
    {
      "epoch": 0.8592541334058276,
      "grad_norm": 5.3528392899950665,
      "learning_rate": 1.0202948554830194e-08,
      "loss": 0.7398,
      "step": 8692
    },
    {
      "epoch": 0.859352989150582,
      "grad_norm": 3.679831300628519,
      "learning_rate": 1.018886208214359e-08,
      "loss": 0.6784,
      "step": 8693
    },
    {
      "epoch": 0.8594518448953364,
      "grad_norm": 3.396465730842341,
      "learning_rate": 1.0174784818264404e-08,
      "loss": 0.6711,
      "step": 8694
    },
    {
      "epoch": 0.859550700640091,
      "grad_norm": 4.3053866972303,
      "learning_rate": 1.016071676463599e-08,
      "loss": 0.7732,
      "step": 8695
    },
    {
      "epoch": 0.8596495563848454,
      "grad_norm": 5.0484321464408,
      "learning_rate": 1.0146657922700863e-08,
      "loss": 0.7396,
      "step": 8696
    },
    {
      "epoch": 0.8597484121295998,
      "grad_norm": 3.8234309458579894,
      "learning_rate": 1.0132608293900547e-08,
      "loss": 0.6966,
      "step": 8697
    },
    {
      "epoch": 0.8598472678743544,
      "grad_norm": 3.5043980519445923,
      "learning_rate": 1.0118567879675587e-08,
      "loss": 0.722,
      "step": 8698
    },
    {
      "epoch": 0.8599461236191088,
      "grad_norm": 4.184632423052862,
      "learning_rate": 1.0104536681465658e-08,
      "loss": 0.6837,
      "step": 8699
    },
    {
      "epoch": 0.8600449793638633,
      "grad_norm": 11.263220299322613,
      "learning_rate": 1.0090514700709451e-08,
      "loss": 0.6881,
      "step": 8700
    },
    {
      "epoch": 0.8601438351086178,
      "grad_norm": 4.123417074001255,
      "learning_rate": 1.0076501938844694e-08,
      "loss": 0.6434,
      "step": 8701
    },
    {
      "epoch": 0.8602426908533722,
      "grad_norm": 3.9585163142615323,
      "learning_rate": 1.0062498397308205e-08,
      "loss": 0.7401,
      "step": 8702
    },
    {
      "epoch": 0.8603415465981267,
      "grad_norm": 3.415464844862139,
      "learning_rate": 1.0048504077535846e-08,
      "loss": 0.6605,
      "step": 8703
    },
    {
      "epoch": 0.8604404023428811,
      "grad_norm": 4.31684379522864,
      "learning_rate": 1.0034518980962492e-08,
      "loss": 0.8306,
      "step": 8704
    },
    {
      "epoch": 0.8605392580876357,
      "grad_norm": 3.352121276016661,
      "learning_rate": 1.002054310902215e-08,
      "loss": 0.7975,
      "step": 8705
    },
    {
      "epoch": 0.8606381138323901,
      "grad_norm": 3.7630931849961566,
      "learning_rate": 1.0006576463147797e-08,
      "loss": 0.6384,
      "step": 8706
    },
    {
      "epoch": 0.8607369695771445,
      "grad_norm": 3.5353133748254275,
      "learning_rate": 9.992619044771544e-09,
      "loss": 0.6089,
      "step": 8707
    },
    {
      "epoch": 0.8608358253218991,
      "grad_norm": 6.099731053276534,
      "learning_rate": 9.978670855324512e-09,
      "loss": 0.645,
      "step": 8708
    },
    {
      "epoch": 0.8609346810666535,
      "grad_norm": 4.116758070430057,
      "learning_rate": 9.964731896236867e-09,
      "loss": 0.6998,
      "step": 8709
    },
    {
      "epoch": 0.8610335368114079,
      "grad_norm": 9.020869106518614,
      "learning_rate": 9.950802168937843e-09,
      "loss": 0.6967,
      "step": 8710
    },
    {
      "epoch": 0.8611323925561625,
      "grad_norm": 10.97234212201642,
      "learning_rate": 9.936881674855768e-09,
      "loss": 0.7295,
      "step": 8711
    },
    {
      "epoch": 0.8612312483009169,
      "grad_norm": 2.9971556706620093,
      "learning_rate": 9.922970415417931e-09,
      "loss": 0.6494,
      "step": 8712
    },
    {
      "epoch": 0.8613301040456713,
      "grad_norm": 4.444234511067111,
      "learning_rate": 9.909068392050735e-09,
      "loss": 0.6235,
      "step": 8713
    },
    {
      "epoch": 0.8614289597904258,
      "grad_norm": 3.25232576475982,
      "learning_rate": 9.895175606179673e-09,
      "loss": 0.5969,
      "step": 8714
    },
    {
      "epoch": 0.8615278155351803,
      "grad_norm": 4.011636697464486,
      "learning_rate": 9.881292059229195e-09,
      "loss": 0.6862,
      "step": 8715
    },
    {
      "epoch": 0.8616266712799348,
      "grad_norm": 2.861597406889946,
      "learning_rate": 9.867417752622887e-09,
      "loss": 0.7134,
      "step": 8716
    },
    {
      "epoch": 0.8617255270246892,
      "grad_norm": 3.3820591773412403,
      "learning_rate": 9.853552687783318e-09,
      "loss": 0.8082,
      "step": 8717
    },
    {
      "epoch": 0.8618243827694437,
      "grad_norm": 9.800371565050407,
      "learning_rate": 9.839696866132164e-09,
      "loss": 0.6929,
      "step": 8718
    },
    {
      "epoch": 0.8619232385141982,
      "grad_norm": 5.255403067722871,
      "learning_rate": 9.825850289090164e-09,
      "loss": 0.6419,
      "step": 8719
    },
    {
      "epoch": 0.8620220942589526,
      "grad_norm": 7.219805833879519,
      "learning_rate": 9.812012958077054e-09,
      "loss": 0.7689,
      "step": 8720
    },
    {
      "epoch": 0.8621209500037071,
      "grad_norm": 3.7362987953898084,
      "learning_rate": 9.798184874511628e-09,
      "loss": 0.7278,
      "step": 8721
    },
    {
      "epoch": 0.8622198057484616,
      "grad_norm": 4.359118244053497,
      "learning_rate": 9.784366039811787e-09,
      "loss": 0.6545,
      "step": 8722
    },
    {
      "epoch": 0.862318661493216,
      "grad_norm": 2.958740673628993,
      "learning_rate": 9.770556455394429e-09,
      "loss": 0.645,
      "step": 8723
    },
    {
      "epoch": 0.8624175172379704,
      "grad_norm": 18.027789762760783,
      "learning_rate": 9.756756122675513e-09,
      "loss": 0.646,
      "step": 8724
    },
    {
      "epoch": 0.862516372982725,
      "grad_norm": 6.308876288170628,
      "learning_rate": 9.742965043070105e-09,
      "loss": 0.7124,
      "step": 8725
    },
    {
      "epoch": 0.8626152287274794,
      "grad_norm": 15.19638233938541,
      "learning_rate": 9.72918321799222e-09,
      "loss": 0.7611,
      "step": 8726
    },
    {
      "epoch": 0.8627140844722339,
      "grad_norm": 7.5366789564379415,
      "learning_rate": 9.715410648855039e-09,
      "loss": 0.6626,
      "step": 8727
    },
    {
      "epoch": 0.8628129402169884,
      "grad_norm": 4.096083467022603,
      "learning_rate": 9.701647337070673e-09,
      "loss": 0.743,
      "step": 8728
    },
    {
      "epoch": 0.8629117959617428,
      "grad_norm": 9.282828992279972,
      "learning_rate": 9.687893284050386e-09,
      "loss": 0.707,
      "step": 8729
    },
    {
      "epoch": 0.8630106517064973,
      "grad_norm": 4.587995607161568,
      "learning_rate": 9.674148491204458e-09,
      "loss": 0.7394,
      "step": 8730
    },
    {
      "epoch": 0.8631095074512518,
      "grad_norm": 4.646003949284046,
      "learning_rate": 9.660412959942188e-09,
      "loss": 0.7486,
      "step": 8731
    },
    {
      "epoch": 0.8632083631960062,
      "grad_norm": 2.7706079324301793,
      "learning_rate": 9.646686691671957e-09,
      "loss": 0.7123,
      "step": 8732
    },
    {
      "epoch": 0.8633072189407607,
      "grad_norm": 3.9719214145392843,
      "learning_rate": 9.63296968780123e-09,
      "loss": 0.7309,
      "step": 8733
    },
    {
      "epoch": 0.8634060746855151,
      "grad_norm": 40.58037114601579,
      "learning_rate": 9.619261949736435e-09,
      "loss": 0.6698,
      "step": 8734
    },
    {
      "epoch": 0.8635049304302697,
      "grad_norm": 4.139113532335888,
      "learning_rate": 9.605563478883116e-09,
      "loss": 0.7145,
      "step": 8735
    },
    {
      "epoch": 0.8636037861750241,
      "grad_norm": 4.337579719788171,
      "learning_rate": 9.591874276645861e-09,
      "loss": 0.6127,
      "step": 8736
    },
    {
      "epoch": 0.8637026419197785,
      "grad_norm": 4.020608605895523,
      "learning_rate": 9.578194344428291e-09,
      "loss": 0.8,
      "step": 8737
    },
    {
      "epoch": 0.8638014976645331,
      "grad_norm": 4.017405657528764,
      "learning_rate": 9.564523683633064e-09,
      "loss": 0.65,
      "step": 8738
    },
    {
      "epoch": 0.8639003534092875,
      "grad_norm": 8.535704077831664,
      "learning_rate": 9.550862295661932e-09,
      "loss": 0.6012,
      "step": 8739
    },
    {
      "epoch": 0.8639992091540419,
      "grad_norm": 3.1060104318065824,
      "learning_rate": 9.537210181915622e-09,
      "loss": 0.7005,
      "step": 8740
    },
    {
      "epoch": 0.8640980648987965,
      "grad_norm": 4.11184168751312,
      "learning_rate": 9.523567343794015e-09,
      "loss": 0.6689,
      "step": 8741
    },
    {
      "epoch": 0.8641969206435509,
      "grad_norm": 9.750668818116093,
      "learning_rate": 9.509933782695945e-09,
      "loss": 0.7788,
      "step": 8742
    },
    {
      "epoch": 0.8642957763883053,
      "grad_norm": 2.6941895629981114,
      "learning_rate": 9.496309500019328e-09,
      "loss": 0.8048,
      "step": 8743
    },
    {
      "epoch": 0.8643946321330598,
      "grad_norm": 3.438409769153508,
      "learning_rate": 9.48269449716118e-09,
      "loss": 0.5649,
      "step": 8744
    },
    {
      "epoch": 0.8644934878778143,
      "grad_norm": 2.8700570042368088,
      "learning_rate": 9.469088775517464e-09,
      "loss": 0.774,
      "step": 8745
    },
    {
      "epoch": 0.8645923436225688,
      "grad_norm": 3.5363338756664673,
      "learning_rate": 9.455492336483262e-09,
      "loss": 0.7024,
      "step": 8746
    },
    {
      "epoch": 0.8646911993673232,
      "grad_norm": 4.178623315967932,
      "learning_rate": 9.441905181452725e-09,
      "loss": 0.6993,
      "step": 8747
    },
    {
      "epoch": 0.8647900551120777,
      "grad_norm": 7.943577629810506,
      "learning_rate": 9.428327311818962e-09,
      "loss": 0.7559,
      "step": 8748
    },
    {
      "epoch": 0.8648889108568322,
      "grad_norm": 7.170393042105103,
      "learning_rate": 9.414758728974226e-09,
      "loss": 0.6667,
      "step": 8749
    },
    {
      "epoch": 0.8649877666015866,
      "grad_norm": 4.066462510351933,
      "learning_rate": 9.401199434309736e-09,
      "loss": 0.6335,
      "step": 8750
    },
    {
      "epoch": 0.8650866223463411,
      "grad_norm": 7.663538720962882,
      "learning_rate": 9.387649429215816e-09,
      "loss": 0.6389,
      "step": 8751
    },
    {
      "epoch": 0.8651854780910956,
      "grad_norm": 3.7810132512284307,
      "learning_rate": 9.37410871508184e-09,
      "loss": 0.7214,
      "step": 8752
    },
    {
      "epoch": 0.86528433383585,
      "grad_norm": 4.758669795015115,
      "learning_rate": 9.360577293296179e-09,
      "loss": 0.6519,
      "step": 8753
    },
    {
      "epoch": 0.8653831895806045,
      "grad_norm": 4.694440930957621,
      "learning_rate": 9.347055165246298e-09,
      "loss": 0.6909,
      "step": 8754
    },
    {
      "epoch": 0.865482045325359,
      "grad_norm": 4.284354105466366,
      "learning_rate": 9.333542332318679e-09,
      "loss": 0.7918,
      "step": 8755
    },
    {
      "epoch": 0.8655809010701134,
      "grad_norm": 5.37075965527498,
      "learning_rate": 9.320038795898877e-09,
      "loss": 0.7313,
      "step": 8756
    },
    {
      "epoch": 0.8656797568148679,
      "grad_norm": 4.96768697370037,
      "learning_rate": 9.306544557371455e-09,
      "loss": 0.6409,
      "step": 8757
    },
    {
      "epoch": 0.8657786125596224,
      "grad_norm": 4.314959971111847,
      "learning_rate": 9.293059618120091e-09,
      "loss": 0.6652,
      "step": 8758
    },
    {
      "epoch": 0.8658774683043768,
      "grad_norm": 3.103779569760233,
      "learning_rate": 9.279583979527416e-09,
      "loss": 0.6904,
      "step": 8759
    },
    {
      "epoch": 0.8659763240491313,
      "grad_norm": 6.039707627980897,
      "learning_rate": 9.26611764297519e-09,
      "loss": 0.7365,
      "step": 8760
    },
    {
      "epoch": 0.8660751797938858,
      "grad_norm": 4.6155515016381825,
      "learning_rate": 9.252660609844198e-09,
      "loss": 0.6787,
      "step": 8761
    },
    {
      "epoch": 0.8661740355386403,
      "grad_norm": 6.401217728674372,
      "learning_rate": 9.239212881514225e-09,
      "loss": 0.6253,
      "step": 8762
    },
    {
      "epoch": 0.8662728912833947,
      "grad_norm": 2.575455841388774,
      "learning_rate": 9.225774459364165e-09,
      "loss": 0.6793,
      "step": 8763
    },
    {
      "epoch": 0.8663717470281491,
      "grad_norm": 4.02595379304599,
      "learning_rate": 9.212345344771911e-09,
      "loss": 0.5624,
      "step": 8764
    },
    {
      "epoch": 0.8664706027729037,
      "grad_norm": 3.6220420957585606,
      "learning_rate": 9.198925539114433e-09,
      "loss": 0.7972,
      "step": 8765
    },
    {
      "epoch": 0.8665694585176581,
      "grad_norm": 3.5239043292179253,
      "learning_rate": 9.185515043767744e-09,
      "loss": 0.8094,
      "step": 8766
    },
    {
      "epoch": 0.8666683142624125,
      "grad_norm": 3.140265580951762,
      "learning_rate": 9.172113860106857e-09,
      "loss": 0.7566,
      "step": 8767
    },
    {
      "epoch": 0.8667671700071671,
      "grad_norm": 3.8531861516271753,
      "learning_rate": 9.15872198950589e-09,
      "loss": 0.7778,
      "step": 8768
    },
    {
      "epoch": 0.8668660257519215,
      "grad_norm": 5.4854655676393795,
      "learning_rate": 9.145339433338007e-09,
      "loss": 0.6816,
      "step": 8769
    },
    {
      "epoch": 0.8669648814966759,
      "grad_norm": 3.5775275515620515,
      "learning_rate": 9.131966192975339e-09,
      "loss": 0.6705,
      "step": 8770
    },
    {
      "epoch": 0.8670637372414305,
      "grad_norm": 6.231422615973247,
      "learning_rate": 9.118602269789155e-09,
      "loss": 0.6907,
      "step": 8771
    },
    {
      "epoch": 0.8671625929861849,
      "grad_norm": 30.713183053254085,
      "learning_rate": 9.105247665149695e-09,
      "loss": 0.7279,
      "step": 8772
    },
    {
      "epoch": 0.8672614487309394,
      "grad_norm": 3.578011624569336,
      "learning_rate": 9.09190238042632e-09,
      "loss": 0.7667,
      "step": 8773
    },
    {
      "epoch": 0.8673603044756938,
      "grad_norm": 4.217074998894786,
      "learning_rate": 9.078566416987365e-09,
      "loss": 0.6238,
      "step": 8774
    },
    {
      "epoch": 0.8674591602204483,
      "grad_norm": 3.493544210134473,
      "learning_rate": 9.065239776200218e-09,
      "loss": 0.6417,
      "step": 8775
    },
    {
      "epoch": 0.8675580159652028,
      "grad_norm": 3.9647127310128765,
      "learning_rate": 9.05192245943135e-09,
      "loss": 0.7099,
      "step": 8776
    },
    {
      "epoch": 0.8676568717099572,
      "grad_norm": 3.2106871687970835,
      "learning_rate": 9.03861446804628e-09,
      "loss": 0.7429,
      "step": 8777
    },
    {
      "epoch": 0.8677557274547117,
      "grad_norm": 3.802958822444391,
      "learning_rate": 9.025315803409494e-09,
      "loss": 0.6912,
      "step": 8778
    },
    {
      "epoch": 0.8678545831994662,
      "grad_norm": 4.008206496407164,
      "learning_rate": 9.012026466884603e-09,
      "loss": 0.741,
      "step": 8779
    },
    {
      "epoch": 0.8679534389442206,
      "grad_norm": 13.166805462948453,
      "learning_rate": 8.99874645983425e-09,
      "loss": 0.6348,
      "step": 8780
    },
    {
      "epoch": 0.8680522946889752,
      "grad_norm": 3.9159703691045418,
      "learning_rate": 8.985475783620067e-09,
      "loss": 0.7127,
      "step": 8781
    },
    {
      "epoch": 0.8681511504337296,
      "grad_norm": 6.970481654423302,
      "learning_rate": 8.972214439602787e-09,
      "loss": 0.7168,
      "step": 8782
    },
    {
      "epoch": 0.868250006178484,
      "grad_norm": 4.086676083428383,
      "learning_rate": 8.95896242914218e-09,
      "loss": 0.7148,
      "step": 8783
    },
    {
      "epoch": 0.8683488619232386,
      "grad_norm": 9.614858089711998,
      "learning_rate": 8.945719753597003e-09,
      "loss": 0.7437,
      "step": 8784
    },
    {
      "epoch": 0.868447717667993,
      "grad_norm": 4.843403893866844,
      "learning_rate": 8.932486414325135e-09,
      "loss": 0.7095,
      "step": 8785
    },
    {
      "epoch": 0.8685465734127474,
      "grad_norm": 3.1394924937619413,
      "learning_rate": 8.919262412683438e-09,
      "loss": 0.7096,
      "step": 8786
    },
    {
      "epoch": 0.8686454291575019,
      "grad_norm": 4.70879146619149,
      "learning_rate": 8.906047750027835e-09,
      "loss": 0.7117,
      "step": 8787
    },
    {
      "epoch": 0.8687442849022564,
      "grad_norm": 6.613484872043874,
      "learning_rate": 8.892842427713332e-09,
      "loss": 0.7425,
      "step": 8788
    },
    {
      "epoch": 0.8688431406470108,
      "grad_norm": 4.6207079801743784,
      "learning_rate": 8.879646447093891e-09,
      "loss": 0.7303,
      "step": 8789
    },
    {
      "epoch": 0.8689419963917653,
      "grad_norm": 2.9109905623665164,
      "learning_rate": 8.866459809522597e-09,
      "loss": 0.7092,
      "step": 8790
    },
    {
      "epoch": 0.8690408521365198,
      "grad_norm": 5.817113843715692,
      "learning_rate": 8.853282516351545e-09,
      "loss": 0.694,
      "step": 8791
    },
    {
      "epoch": 0.8691397078812743,
      "grad_norm": 4.708245533546307,
      "learning_rate": 8.840114568931844e-09,
      "loss": 0.6874,
      "step": 8792
    },
    {
      "epoch": 0.8692385636260287,
      "grad_norm": 12.40603062572837,
      "learning_rate": 8.82695596861368e-09,
      "loss": 0.7788,
      "step": 8793
    },
    {
      "epoch": 0.8693374193707832,
      "grad_norm": 7.334026383500333,
      "learning_rate": 8.81380671674632e-09,
      "loss": 0.746,
      "step": 8794
    },
    {
      "epoch": 0.8694362751155377,
      "grad_norm": 4.459729483034317,
      "learning_rate": 8.80066681467796e-09,
      "loss": 0.7191,
      "step": 8795
    },
    {
      "epoch": 0.8695351308602921,
      "grad_norm": 3.4785068130384103,
      "learning_rate": 8.787536263755957e-09,
      "loss": 0.5721,
      "step": 8796
    },
    {
      "epoch": 0.8696339866050465,
      "grad_norm": 5.286976065430261,
      "learning_rate": 8.774415065326612e-09,
      "loss": 0.6168,
      "step": 8797
    },
    {
      "epoch": 0.8697328423498011,
      "grad_norm": 4.672510348038304,
      "learning_rate": 8.761303220735327e-09,
      "loss": 0.8147,
      "step": 8798
    },
    {
      "epoch": 0.8698316980945555,
      "grad_norm": 6.841114483944898,
      "learning_rate": 8.74820073132655e-09,
      "loss": 0.6292,
      "step": 8799
    },
    {
      "epoch": 0.86993055383931,
      "grad_norm": 4.175597029087264,
      "learning_rate": 8.735107598443714e-09,
      "loss": 0.6357,
      "step": 8800
    },
    {
      "epoch": 0.8700294095840645,
      "grad_norm": 4.0422014245309414,
      "learning_rate": 8.722023823429336e-09,
      "loss": 0.7369,
      "step": 8801
    },
    {
      "epoch": 0.8701282653288189,
      "grad_norm": 6.833759323875495,
      "learning_rate": 8.708949407625e-09,
      "loss": 0.78,
      "step": 8802
    },
    {
      "epoch": 0.8702271210735734,
      "grad_norm": 4.334623353241719,
      "learning_rate": 8.69588435237123e-09,
      "loss": 0.6466,
      "step": 8803
    },
    {
      "epoch": 0.8703259768183279,
      "grad_norm": 4.46321018833137,
      "learning_rate": 8.682828659007702e-09,
      "loss": 0.742,
      "step": 8804
    },
    {
      "epoch": 0.8704248325630823,
      "grad_norm": 4.333773582324982,
      "learning_rate": 8.669782328873088e-09,
      "loss": 0.6972,
      "step": 8805
    },
    {
      "epoch": 0.8705236883078368,
      "grad_norm": 18.070184541070912,
      "learning_rate": 8.656745363305063e-09,
      "loss": 0.7177,
      "step": 8806
    },
    {
      "epoch": 0.8706225440525912,
      "grad_norm": 5.405760863997132,
      "learning_rate": 8.643717763640402e-09,
      "loss": 0.6293,
      "step": 8807
    },
    {
      "epoch": 0.8707213997973458,
      "grad_norm": 5.4032829592342795,
      "learning_rate": 8.630699531214891e-09,
      "loss": 0.7003,
      "step": 8808
    },
    {
      "epoch": 0.8708202555421002,
      "grad_norm": 4.5019953875097904,
      "learning_rate": 8.617690667363341e-09,
      "loss": 0.7776,
      "step": 8809
    },
    {
      "epoch": 0.8709191112868546,
      "grad_norm": 3.0542023268082583,
      "learning_rate": 8.604691173419632e-09,
      "loss": 0.7032,
      "step": 8810
    },
    {
      "epoch": 0.8710179670316092,
      "grad_norm": 5.945890298010638,
      "learning_rate": 8.59170105071666e-09,
      "loss": 0.7741,
      "step": 8811
    },
    {
      "epoch": 0.8711168227763636,
      "grad_norm": 2.599215169875295,
      "learning_rate": 8.578720300586372e-09,
      "loss": 0.6619,
      "step": 8812
    },
    {
      "epoch": 0.871215678521118,
      "grad_norm": 5.07575178367205,
      "learning_rate": 8.565748924359784e-09,
      "loss": 0.686,
      "step": 8813
    },
    {
      "epoch": 0.8713145342658726,
      "grad_norm": 3.427103039030657,
      "learning_rate": 8.552786923366873e-09,
      "loss": 0.6469,
      "step": 8814
    },
    {
      "epoch": 0.871413390010627,
      "grad_norm": 5.07544055264377,
      "learning_rate": 8.539834298936721e-09,
      "loss": 0.7119,
      "step": 8815
    },
    {
      "epoch": 0.8715122457553814,
      "grad_norm": 4.190692101773349,
      "learning_rate": 8.526891052397444e-09,
      "loss": 0.7857,
      "step": 8816
    },
    {
      "epoch": 0.8716111015001359,
      "grad_norm": 4.297389385715255,
      "learning_rate": 8.513957185076149e-09,
      "loss": 0.6742,
      "step": 8817
    },
    {
      "epoch": 0.8717099572448904,
      "grad_norm": 14.78482327539833,
      "learning_rate": 8.501032698299049e-09,
      "loss": 0.5984,
      "step": 8818
    },
    {
      "epoch": 0.8718088129896449,
      "grad_norm": 5.639983779835924,
      "learning_rate": 8.48811759339133e-09,
      "loss": 0.7374,
      "step": 8819
    },
    {
      "epoch": 0.8719076687343993,
      "grad_norm": 9.064821884091982,
      "learning_rate": 8.475211871677246e-09,
      "loss": 0.754,
      "step": 8820
    },
    {
      "epoch": 0.8720065244791538,
      "grad_norm": 3.3134792391944394,
      "learning_rate": 8.462315534480125e-09,
      "loss": 0.8118,
      "step": 8821
    },
    {
      "epoch": 0.8721053802239083,
      "grad_norm": 4.230887399798083,
      "learning_rate": 8.449428583122243e-09,
      "loss": 0.7354,
      "step": 8822
    },
    {
      "epoch": 0.8722042359686627,
      "grad_norm": 5.971152371658918,
      "learning_rate": 8.436551018924998e-09,
      "loss": 0.6762,
      "step": 8823
    },
    {
      "epoch": 0.8723030917134172,
      "grad_norm": 3.2149714448155673,
      "learning_rate": 8.423682843208813e-09,
      "loss": 0.6183,
      "step": 8824
    },
    {
      "epoch": 0.8724019474581717,
      "grad_norm": 4.817781222030279,
      "learning_rate": 8.410824057293109e-09,
      "loss": 0.6362,
      "step": 8825
    },
    {
      "epoch": 0.8725008032029261,
      "grad_norm": 5.632104365569159,
      "learning_rate": 8.397974662496333e-09,
      "loss": 0.7215,
      "step": 8826
    },
    {
      "epoch": 0.8725996589476805,
      "grad_norm": 4.130796738869813,
      "learning_rate": 8.38513466013605e-09,
      "loss": 0.7495,
      "step": 8827
    },
    {
      "epoch": 0.8726985146924351,
      "grad_norm": 3.0788245405419286,
      "learning_rate": 8.37230405152879e-09,
      "loss": 0.7063,
      "step": 8828
    },
    {
      "epoch": 0.8727973704371895,
      "grad_norm": 4.771919826041508,
      "learning_rate": 8.359482837990151e-09,
      "loss": 0.7861,
      "step": 8829
    },
    {
      "epoch": 0.872896226181944,
      "grad_norm": 5.979634236704562,
      "learning_rate": 8.346671020834739e-09,
      "loss": 0.7692,
      "step": 8830
    },
    {
      "epoch": 0.8729950819266985,
      "grad_norm": 4.587234776925188,
      "learning_rate": 8.333868601376227e-09,
      "loss": 0.7559,
      "step": 8831
    },
    {
      "epoch": 0.8730939376714529,
      "grad_norm": 3.421663615815811,
      "learning_rate": 8.321075580927339e-09,
      "loss": 0.7465,
      "step": 8832
    },
    {
      "epoch": 0.8731927934162074,
      "grad_norm": 3.4528025059520826,
      "learning_rate": 8.308291960799784e-09,
      "loss": 0.6328,
      "step": 8833
    },
    {
      "epoch": 0.8732916491609619,
      "grad_norm": 4.182213352395917,
      "learning_rate": 8.295517742304325e-09,
      "loss": 0.6173,
      "step": 8834
    },
    {
      "epoch": 0.8733905049057163,
      "grad_norm": 4.472295396157893,
      "learning_rate": 8.282752926750814e-09,
      "loss": 0.7132,
      "step": 8835
    },
    {
      "epoch": 0.8734893606504708,
      "grad_norm": 9.341332686639797,
      "learning_rate": 8.269997515448046e-09,
      "loss": 0.7109,
      "step": 8836
    },
    {
      "epoch": 0.8735882163952252,
      "grad_norm": 8.870826630065606,
      "learning_rate": 8.257251509703922e-09,
      "loss": 0.776,
      "step": 8837
    },
    {
      "epoch": 0.8736870721399798,
      "grad_norm": 4.056565299027339,
      "learning_rate": 8.244514910825362e-09,
      "loss": 0.8221,
      "step": 8838
    },
    {
      "epoch": 0.8737859278847342,
      "grad_norm": 2.6749054789888103,
      "learning_rate": 8.231787720118299e-09,
      "loss": 0.668,
      "step": 8839
    },
    {
      "epoch": 0.8738847836294886,
      "grad_norm": 3.877747239903266,
      "learning_rate": 8.219069938887746e-09,
      "loss": 0.7595,
      "step": 8840
    },
    {
      "epoch": 0.8739836393742432,
      "grad_norm": 3.7617029893790703,
      "learning_rate": 8.206361568437692e-09,
      "loss": 0.7245,
      "step": 8841
    },
    {
      "epoch": 0.8740824951189976,
      "grad_norm": 3.5178296382624006,
      "learning_rate": 8.193662610071217e-09,
      "loss": 0.7328,
      "step": 8842
    },
    {
      "epoch": 0.874181350863752,
      "grad_norm": 4.481060663685702,
      "learning_rate": 8.180973065090391e-09,
      "loss": 0.6896,
      "step": 8843
    },
    {
      "epoch": 0.8742802066085066,
      "grad_norm": 7.922486321279734,
      "learning_rate": 8.168292934796373e-09,
      "loss": 0.7297,
      "step": 8844
    },
    {
      "epoch": 0.874379062353261,
      "grad_norm": 5.932750578757979,
      "learning_rate": 8.155622220489278e-09,
      "loss": 0.6879,
      "step": 8845
    },
    {
      "epoch": 0.8744779180980154,
      "grad_norm": 3.9820155846097856,
      "learning_rate": 8.142960923468344e-09,
      "loss": 0.7177,
      "step": 8846
    },
    {
      "epoch": 0.8745767738427699,
      "grad_norm": 9.324340365018982,
      "learning_rate": 8.130309045031769e-09,
      "loss": 0.7442,
      "step": 8847
    },
    {
      "epoch": 0.8746756295875244,
      "grad_norm": 4.733139769251487,
      "learning_rate": 8.117666586476813e-09,
      "loss": 0.6973,
      "step": 8848
    },
    {
      "epoch": 0.8747744853322789,
      "grad_norm": 6.63913694688246,
      "learning_rate": 8.105033549099816e-09,
      "loss": 0.739,
      "step": 8849
    },
    {
      "epoch": 0.8748733410770333,
      "grad_norm": 15.872995709645496,
      "learning_rate": 8.092409934196066e-09,
      "loss": 0.7236,
      "step": 8850
    },
    {
      "epoch": 0.8749721968217878,
      "grad_norm": 7.104618523080822,
      "learning_rate": 8.079795743059958e-09,
      "loss": 0.6462,
      "step": 8851
    },
    {
      "epoch": 0.8750710525665423,
      "grad_norm": 3.8142001969236294,
      "learning_rate": 8.067190976984861e-09,
      "loss": 0.7069,
      "step": 8852
    },
    {
      "epoch": 0.8751699083112967,
      "grad_norm": 4.801325547490255,
      "learning_rate": 8.054595637263228e-09,
      "loss": 0.6321,
      "step": 8853
    },
    {
      "epoch": 0.8752687640560513,
      "grad_norm": 6.89401319557989,
      "learning_rate": 8.042009725186538e-09,
      "loss": 0.5614,
      "step": 8854
    },
    {
      "epoch": 0.8753676198008057,
      "grad_norm": 3.2565245071384714,
      "learning_rate": 8.029433242045257e-09,
      "loss": 0.7698,
      "step": 8855
    },
    {
      "epoch": 0.8754664755455601,
      "grad_norm": 4.728283309120825,
      "learning_rate": 8.016866189128946e-09,
      "loss": 0.7417,
      "step": 8856
    },
    {
      "epoch": 0.8755653312903147,
      "grad_norm": 4.618824904378138,
      "learning_rate": 8.00430856772617e-09,
      "loss": 0.6428,
      "step": 8857
    },
    {
      "epoch": 0.8756641870350691,
      "grad_norm": 8.181547616326991,
      "learning_rate": 7.99176037912449e-09,
      "loss": 0.7087,
      "step": 8858
    },
    {
      "epoch": 0.8757630427798235,
      "grad_norm": 7.740776331666418,
      "learning_rate": 7.9792216246106e-09,
      "loss": 0.8545,
      "step": 8859
    },
    {
      "epoch": 0.875861898524578,
      "grad_norm": 3.714420205282046,
      "learning_rate": 7.966692305470101e-09,
      "loss": 0.6253,
      "step": 8860
    },
    {
      "epoch": 0.8759607542693325,
      "grad_norm": 3.5638677312526332,
      "learning_rate": 7.954172422987748e-09,
      "loss": 0.7765,
      "step": 8861
    },
    {
      "epoch": 0.8760596100140869,
      "grad_norm": 3.6526212870591768,
      "learning_rate": 7.941661978447234e-09,
      "loss": 0.6246,
      "step": 8862
    },
    {
      "epoch": 0.8761584657588414,
      "grad_norm": 4.997595734378032,
      "learning_rate": 7.92916097313131e-09,
      "loss": 0.6886,
      "step": 8863
    },
    {
      "epoch": 0.8762573215035959,
      "grad_norm": 4.045775710049452,
      "learning_rate": 7.916669408321797e-09,
      "loss": 0.7328,
      "step": 8864
    },
    {
      "epoch": 0.8763561772483504,
      "grad_norm": 11.916422211000354,
      "learning_rate": 7.904187285299513e-09,
      "loss": 0.7616,
      "step": 8865
    },
    {
      "epoch": 0.8764550329931048,
      "grad_norm": 9.565784346361918,
      "learning_rate": 7.891714605344313e-09,
      "loss": 0.7001,
      "step": 8866
    },
    {
      "epoch": 0.8765538887378593,
      "grad_norm": 4.284172212468677,
      "learning_rate": 7.879251369735074e-09,
      "loss": 0.6539,
      "step": 8867
    },
    {
      "epoch": 0.8766527444826138,
      "grad_norm": 4.3152017623025944,
      "learning_rate": 7.86679757974975e-09,
      "loss": 0.7438,
      "step": 8868
    },
    {
      "epoch": 0.8767516002273682,
      "grad_norm": 3.1837607463527053,
      "learning_rate": 7.854353236665268e-09,
      "loss": 0.6375,
      "step": 8869
    },
    {
      "epoch": 0.8768504559721226,
      "grad_norm": 4.223069809184284,
      "learning_rate": 7.841918341757603e-09,
      "loss": 0.6695,
      "step": 8870
    },
    {
      "epoch": 0.8769493117168772,
      "grad_norm": 2.6906068177400377,
      "learning_rate": 7.829492896301815e-09,
      "loss": 0.7215,
      "step": 8871
    },
    {
      "epoch": 0.8770481674616316,
      "grad_norm": 4.1433598538754355,
      "learning_rate": 7.817076901571896e-09,
      "loss": 0.6574,
      "step": 8872
    },
    {
      "epoch": 0.877147023206386,
      "grad_norm": 3.243791163915268,
      "learning_rate": 7.804670358840959e-09,
      "loss": 0.6408,
      "step": 8873
    },
    {
      "epoch": 0.8772458789511406,
      "grad_norm": 3.6627357929379225,
      "learning_rate": 7.792273269381088e-09,
      "loss": 0.693,
      "step": 8874
    },
    {
      "epoch": 0.877344734695895,
      "grad_norm": 5.880631597536817,
      "learning_rate": 7.779885634463434e-09,
      "loss": 0.7728,
      "step": 8875
    },
    {
      "epoch": 0.8774435904406495,
      "grad_norm": 3.85653198442103,
      "learning_rate": 7.767507455358191e-09,
      "loss": 0.631,
      "step": 8876
    },
    {
      "epoch": 0.877542446185404,
      "grad_norm": 5.012050768160655,
      "learning_rate": 7.755138733334499e-09,
      "loss": 0.6924,
      "step": 8877
    },
    {
      "epoch": 0.8776413019301584,
      "grad_norm": 3.0414707581509233,
      "learning_rate": 7.742779469660654e-09,
      "loss": 0.7231,
      "step": 8878
    },
    {
      "epoch": 0.8777401576749129,
      "grad_norm": 4.464666771081131,
      "learning_rate": 7.730429665603876e-09,
      "loss": 0.6047,
      "step": 8879
    },
    {
      "epoch": 0.8778390134196673,
      "grad_norm": 3.3877946695342755,
      "learning_rate": 7.718089322430455e-09,
      "loss": 0.6815,
      "step": 8880
    },
    {
      "epoch": 0.8779378691644218,
      "grad_norm": 3.1018432624247194,
      "learning_rate": 7.705758441405707e-09,
      "loss": 0.6,
      "step": 8881
    },
    {
      "epoch": 0.8780367249091763,
      "grad_norm": 7.129200808710328,
      "learning_rate": 7.693437023794024e-09,
      "loss": 0.7385,
      "step": 8882
    },
    {
      "epoch": 0.8781355806539307,
      "grad_norm": 3.833043605505092,
      "learning_rate": 7.681125070858741e-09,
      "loss": 0.6696,
      "step": 8883
    },
    {
      "epoch": 0.8782344363986853,
      "grad_norm": 6.289033461064958,
      "learning_rate": 7.668822583862289e-09,
      "loss": 0.6499,
      "step": 8884
    },
    {
      "epoch": 0.8783332921434397,
      "grad_norm": 3.6999890812253944,
      "learning_rate": 7.656529564066094e-09,
      "loss": 0.6633,
      "step": 8885
    },
    {
      "epoch": 0.8784321478881941,
      "grad_norm": 3.692471935404897,
      "learning_rate": 7.644246012730627e-09,
      "loss": 0.6271,
      "step": 8886
    },
    {
      "epoch": 0.8785310036329487,
      "grad_norm": 4.496493740058102,
      "learning_rate": 7.631971931115399e-09,
      "loss": 0.7407,
      "step": 8887
    },
    {
      "epoch": 0.8786298593777031,
      "grad_norm": 8.970234708943824,
      "learning_rate": 7.619707320478908e-09,
      "loss": 0.6495,
      "step": 8888
    },
    {
      "epoch": 0.8787287151224575,
      "grad_norm": 6.107617586757581,
      "learning_rate": 7.607452182078733e-09,
      "loss": 0.6601,
      "step": 8889
    },
    {
      "epoch": 0.878827570867212,
      "grad_norm": 3.8579464745080894,
      "learning_rate": 7.59520651717146e-09,
      "loss": 0.6758,
      "step": 8890
    },
    {
      "epoch": 0.8789264266119665,
      "grad_norm": 4.037929991638243,
      "learning_rate": 7.582970327012683e-09,
      "loss": 0.5317,
      "step": 8891
    },
    {
      "epoch": 0.879025282356721,
      "grad_norm": 14.817822721398237,
      "learning_rate": 7.570743612857055e-09,
      "loss": 0.661,
      "step": 8892
    },
    {
      "epoch": 0.8791241381014754,
      "grad_norm": 7.892104073234358,
      "learning_rate": 7.558526375958252e-09,
      "loss": 0.6404,
      "step": 8893
    },
    {
      "epoch": 0.8792229938462299,
      "grad_norm": 3.3529088777209926,
      "learning_rate": 7.546318617568937e-09,
      "loss": 0.6172,
      "step": 8894
    },
    {
      "epoch": 0.8793218495909844,
      "grad_norm": 4.4804534628956665,
      "learning_rate": 7.534120338940886e-09,
      "loss": 0.7499,
      "step": 8895
    },
    {
      "epoch": 0.8794207053357388,
      "grad_norm": 3.098713483706549,
      "learning_rate": 7.521931541324832e-09,
      "loss": 0.6397,
      "step": 8896
    },
    {
      "epoch": 0.8795195610804933,
      "grad_norm": 5.165933630827799,
      "learning_rate": 7.509752225970523e-09,
      "loss": 0.698,
      "step": 8897
    },
    {
      "epoch": 0.8796184168252478,
      "grad_norm": 3.664141066295289,
      "learning_rate": 7.497582394126822e-09,
      "loss": 0.8364,
      "step": 8898
    },
    {
      "epoch": 0.8797172725700022,
      "grad_norm": 3.1545925551739296,
      "learning_rate": 7.485422047041523e-09,
      "loss": 0.7857,
      "step": 8899
    },
    {
      "epoch": 0.8798161283147566,
      "grad_norm": 3.5645049183033537,
      "learning_rate": 7.473271185961493e-09,
      "loss": 0.6608,
      "step": 8900
    },
    {
      "epoch": 0.8799149840595112,
      "grad_norm": 8.179113785884658,
      "learning_rate": 7.461129812132671e-09,
      "loss": 0.6741,
      "step": 8901
    },
    {
      "epoch": 0.8800138398042656,
      "grad_norm": 4.375366211842156,
      "learning_rate": 7.448997926799916e-09,
      "loss": 0.7284,
      "step": 8902
    },
    {
      "epoch": 0.88011269554902,
      "grad_norm": 4.269435602402366,
      "learning_rate": 7.43687553120721e-09,
      "loss": 0.6676,
      "step": 8903
    },
    {
      "epoch": 0.8802115512937746,
      "grad_norm": 5.253248961095636,
      "learning_rate": 7.424762626597525e-09,
      "loss": 0.7035,
      "step": 8904
    },
    {
      "epoch": 0.880310407038529,
      "grad_norm": 3.1922609650759366,
      "learning_rate": 7.412659214212835e-09,
      "loss": 0.7131,
      "step": 8905
    },
    {
      "epoch": 0.8804092627832835,
      "grad_norm": 5.234084422524008,
      "learning_rate": 7.400565295294192e-09,
      "loss": 0.7327,
      "step": 8906
    },
    {
      "epoch": 0.880508118528038,
      "grad_norm": 3.595250690560991,
      "learning_rate": 7.388480871081637e-09,
      "loss": 0.6601,
      "step": 8907
    },
    {
      "epoch": 0.8806069742727924,
      "grad_norm": 3.8163446480149834,
      "learning_rate": 7.376405942814245e-09,
      "loss": 0.6738,
      "step": 8908
    },
    {
      "epoch": 0.8807058300175469,
      "grad_norm": 6.73393045959171,
      "learning_rate": 7.364340511730149e-09,
      "loss": 0.6926,
      "step": 8909
    },
    {
      "epoch": 0.8808046857623013,
      "grad_norm": 3.9960969394956964,
      "learning_rate": 7.3522845790664476e-09,
      "loss": 0.6187,
      "step": 8910
    },
    {
      "epoch": 0.8809035415070559,
      "grad_norm": 2.8333849529140975,
      "learning_rate": 7.340238146059308e-09,
      "loss": 0.7299,
      "step": 8911
    },
    {
      "epoch": 0.8810023972518103,
      "grad_norm": 4.222306028682343,
      "learning_rate": 7.328201213943941e-09,
      "loss": 0.7028,
      "step": 8912
    },
    {
      "epoch": 0.8811012529965647,
      "grad_norm": 4.455751639541688,
      "learning_rate": 7.316173783954527e-09,
      "loss": 0.6933,
      "step": 8913
    },
    {
      "epoch": 0.8812001087413193,
      "grad_norm": 5.735798550190392,
      "learning_rate": 7.304155857324301e-09,
      "loss": 0.5745,
      "step": 8914
    },
    {
      "epoch": 0.8812989644860737,
      "grad_norm": 3.551166599457233,
      "learning_rate": 7.292147435285545e-09,
      "loss": 0.7322,
      "step": 8915
    },
    {
      "epoch": 0.8813978202308281,
      "grad_norm": 2.9066843139899388,
      "learning_rate": 7.2801485190695066e-09,
      "loss": 0.717,
      "step": 8916
    },
    {
      "epoch": 0.8814966759755827,
      "grad_norm": 3.4969125259322653,
      "learning_rate": 7.268159109906558e-09,
      "loss": 0.7309,
      "step": 8917
    },
    {
      "epoch": 0.8815955317203371,
      "grad_norm": 6.171568072620089,
      "learning_rate": 7.25617920902597e-09,
      "loss": 0.7495,
      "step": 8918
    },
    {
      "epoch": 0.8816943874650915,
      "grad_norm": 4.238434861586912,
      "learning_rate": 7.2442088176561385e-09,
      "loss": 0.7496,
      "step": 8919
    },
    {
      "epoch": 0.881793243209846,
      "grad_norm": 4.438327534461717,
      "learning_rate": 7.232247937024471e-09,
      "loss": 0.6449,
      "step": 8920
    },
    {
      "epoch": 0.8818920989546005,
      "grad_norm": 3.583685052209322,
      "learning_rate": 7.220296568357343e-09,
      "loss": 0.7508,
      "step": 8921
    },
    {
      "epoch": 0.881990954699355,
      "grad_norm": 3.43386795687596,
      "learning_rate": 7.208354712880194e-09,
      "loss": 0.7178,
      "step": 8922
    },
    {
      "epoch": 0.8820898104441094,
      "grad_norm": 3.4553004608937985,
      "learning_rate": 7.196422371817523e-09,
      "loss": 0.7438,
      "step": 8923
    },
    {
      "epoch": 0.8821886661888639,
      "grad_norm": 4.683864535543013,
      "learning_rate": 7.184499546392763e-09,
      "loss": 0.607,
      "step": 8924
    },
    {
      "epoch": 0.8822875219336184,
      "grad_norm": 5.333158442826294,
      "learning_rate": 7.1725862378284595e-09,
      "loss": 0.5755,
      "step": 8925
    },
    {
      "epoch": 0.8823863776783728,
      "grad_norm": 3.2599067773094395,
      "learning_rate": 7.160682447346145e-09,
      "loss": 0.6524,
      "step": 8926
    },
    {
      "epoch": 0.8824852334231273,
      "grad_norm": 3.477099238836639,
      "learning_rate": 7.1487881761663535e-09,
      "loss": 0.6713,
      "step": 8927
    },
    {
      "epoch": 0.8825840891678818,
      "grad_norm": 18.36335269819047,
      "learning_rate": 7.1369034255087e-09,
      "loss": 0.6216,
      "step": 8928
    },
    {
      "epoch": 0.8826829449126362,
      "grad_norm": 5.069234897230882,
      "learning_rate": 7.125028196591754e-09,
      "loss": 0.6981,
      "step": 8929
    },
    {
      "epoch": 0.8827818006573906,
      "grad_norm": 3.133220326217673,
      "learning_rate": 7.113162490633184e-09,
      "loss": 0.7839,
      "step": 8930
    },
    {
      "epoch": 0.8828806564021452,
      "grad_norm": 12.668323792089929,
      "learning_rate": 7.1013063088496085e-09,
      "loss": 0.5464,
      "step": 8931
    },
    {
      "epoch": 0.8829795121468996,
      "grad_norm": 5.748200553569565,
      "learning_rate": 7.089459652456742e-09,
      "loss": 0.5639,
      "step": 8932
    },
    {
      "epoch": 0.8830783678916541,
      "grad_norm": 9.33867054769068,
      "learning_rate": 7.077622522669236e-09,
      "loss": 0.567,
      "step": 8933
    },
    {
      "epoch": 0.8831772236364086,
      "grad_norm": 3.01929452863069,
      "learning_rate": 7.065794920700852e-09,
      "loss": 0.5769,
      "step": 8934
    },
    {
      "epoch": 0.883276079381163,
      "grad_norm": 7.602731212368465,
      "learning_rate": 7.053976847764309e-09,
      "loss": 0.8101,
      "step": 8935
    },
    {
      "epoch": 0.8833749351259175,
      "grad_norm": 6.366955070378368,
      "learning_rate": 7.042168305071394e-09,
      "loss": 0.7681,
      "step": 8936
    },
    {
      "epoch": 0.883473790870672,
      "grad_norm": 3.396994110657782,
      "learning_rate": 7.0303692938329055e-09,
      "loss": 0.859,
      "step": 8937
    },
    {
      "epoch": 0.8835726466154264,
      "grad_norm": 10.030898648430062,
      "learning_rate": 7.018579815258641e-09,
      "loss": 0.6341,
      "step": 8938
    },
    {
      "epoch": 0.8836715023601809,
      "grad_norm": 5.473360144159395,
      "learning_rate": 7.006799870557445e-09,
      "loss": 0.775,
      "step": 8939
    },
    {
      "epoch": 0.8837703581049354,
      "grad_norm": 2.9508525553181184,
      "learning_rate": 6.995029460937174e-09,
      "loss": 0.6195,
      "step": 8940
    },
    {
      "epoch": 0.8838692138496899,
      "grad_norm": 4.171083004447715,
      "learning_rate": 6.983268587604707e-09,
      "loss": 0.7373,
      "step": 8941
    },
    {
      "epoch": 0.8839680695944443,
      "grad_norm": 4.778084407363342,
      "learning_rate": 6.971517251765968e-09,
      "loss": 0.7034,
      "step": 8942
    },
    {
      "epoch": 0.8840669253391987,
      "grad_norm": 4.516531278754341,
      "learning_rate": 6.959775454625849e-09,
      "loss": 0.7477,
      "step": 8943
    },
    {
      "epoch": 0.8841657810839533,
      "grad_norm": 8.842039742015581,
      "learning_rate": 6.9480431973883204e-09,
      "loss": 0.7059,
      "step": 8944
    },
    {
      "epoch": 0.8842646368287077,
      "grad_norm": 4.285733672363656,
      "learning_rate": 6.936320481256363e-09,
      "loss": 0.6366,
      "step": 8945
    },
    {
      "epoch": 0.8843634925734621,
      "grad_norm": 9.926416001866258,
      "learning_rate": 6.92460730743194e-09,
      "loss": 0.7045,
      "step": 8946
    },
    {
      "epoch": 0.8844623483182167,
      "grad_norm": 4.147065608976252,
      "learning_rate": 6.9129036771161e-09,
      "loss": 0.6409,
      "step": 8947
    },
    {
      "epoch": 0.8845612040629711,
      "grad_norm": 2.520575679660517,
      "learning_rate": 6.901209591508839e-09,
      "loss": 0.7045,
      "step": 8948
    },
    {
      "epoch": 0.8846600598077256,
      "grad_norm": 3.7903656249601734,
      "learning_rate": 6.889525051809242e-09,
      "loss": 0.7247,
      "step": 8949
    },
    {
      "epoch": 0.8847589155524801,
      "grad_norm": 3.1128611611619186,
      "learning_rate": 6.877850059215373e-09,
      "loss": 0.6842,
      "step": 8950
    },
    {
      "epoch": 0.8848577712972345,
      "grad_norm": 8.087260785410317,
      "learning_rate": 6.8661846149243195e-09,
      "loss": 0.6359,
      "step": 8951
    },
    {
      "epoch": 0.884956627041989,
      "grad_norm": 3.2153643687804934,
      "learning_rate": 6.854528720132213e-09,
      "loss": 0.6939,
      "step": 8952
    },
    {
      "epoch": 0.8850554827867434,
      "grad_norm": 2.908186717435241,
      "learning_rate": 6.842882376034209e-09,
      "loss": 0.745,
      "step": 8953
    },
    {
      "epoch": 0.8851543385314979,
      "grad_norm": 2.956046452326921,
      "learning_rate": 6.831245583824441e-09,
      "loss": 0.7485,
      "step": 8954
    },
    {
      "epoch": 0.8852531942762524,
      "grad_norm": 3.2760740178402186,
      "learning_rate": 6.819618344696099e-09,
      "loss": 0.7091,
      "step": 8955
    },
    {
      "epoch": 0.8853520500210068,
      "grad_norm": 16.015330113239326,
      "learning_rate": 6.808000659841406e-09,
      "loss": 0.7451,
      "step": 8956
    },
    {
      "epoch": 0.8854509057657614,
      "grad_norm": 5.6723996434187525,
      "learning_rate": 6.796392530451556e-09,
      "loss": 0.654,
      "step": 8957
    },
    {
      "epoch": 0.8855497615105158,
      "grad_norm": 4.974980975513896,
      "learning_rate": 6.784793957716795e-09,
      "loss": 0.6556,
      "step": 8958
    },
    {
      "epoch": 0.8856486172552702,
      "grad_norm": 6.157163165419612,
      "learning_rate": 6.773204942826405e-09,
      "loss": 0.8166,
      "step": 8959
    },
    {
      "epoch": 0.8857474730000248,
      "grad_norm": 4.180858284192593,
      "learning_rate": 6.761625486968647e-09,
      "loss": 0.779,
      "step": 8960
    },
    {
      "epoch": 0.8858463287447792,
      "grad_norm": 4.3450298005889945,
      "learning_rate": 6.75005559133085e-09,
      "loss": 0.766,
      "step": 8961
    },
    {
      "epoch": 0.8859451844895336,
      "grad_norm": 3.1553343552652136,
      "learning_rate": 6.738495257099297e-09,
      "loss": 0.7613,
      "step": 8962
    },
    {
      "epoch": 0.8860440402342881,
      "grad_norm": 4.350488321705772,
      "learning_rate": 6.726944485459352e-09,
      "loss": 0.6,
      "step": 8963
    },
    {
      "epoch": 0.8861428959790426,
      "grad_norm": 4.16114426122731,
      "learning_rate": 6.715403277595399e-09,
      "loss": 0.6613,
      "step": 8964
    },
    {
      "epoch": 0.886241751723797,
      "grad_norm": 5.7414927995208584,
      "learning_rate": 6.703871634690772e-09,
      "loss": 0.744,
      "step": 8965
    },
    {
      "epoch": 0.8863406074685515,
      "grad_norm": 4.28154950492256,
      "learning_rate": 6.692349557927912e-09,
      "loss": 0.6286,
      "step": 8966
    },
    {
      "epoch": 0.886439463213306,
      "grad_norm": 9.682879403320705,
      "learning_rate": 6.68083704848822e-09,
      "loss": 0.8129,
      "step": 8967
    },
    {
      "epoch": 0.8865383189580605,
      "grad_norm": 8.135506802537659,
      "learning_rate": 6.669334107552116e-09,
      "loss": 0.7295,
      "step": 8968
    },
    {
      "epoch": 0.8866371747028149,
      "grad_norm": 6.071801768121945,
      "learning_rate": 6.65784073629907e-09,
      "loss": 0.6906,
      "step": 8969
    },
    {
      "epoch": 0.8867360304475694,
      "grad_norm": 5.094894620707953,
      "learning_rate": 6.646356935907594e-09,
      "loss": 0.7817,
      "step": 8970
    },
    {
      "epoch": 0.8868348861923239,
      "grad_norm": 3.19983089758021,
      "learning_rate": 6.634882707555123e-09,
      "loss": 0.6953,
      "step": 8971
    },
    {
      "epoch": 0.8869337419370783,
      "grad_norm": 6.822290209337686,
      "learning_rate": 6.623418052418217e-09,
      "loss": 0.8176,
      "step": 8972
    },
    {
      "epoch": 0.8870325976818327,
      "grad_norm": 4.147547261845449,
      "learning_rate": 6.611962971672369e-09,
      "loss": 0.9011,
      "step": 8973
    },
    {
      "epoch": 0.8871314534265873,
      "grad_norm": 7.535114354216261,
      "learning_rate": 6.600517466492151e-09,
      "loss": 0.644,
      "step": 8974
    },
    {
      "epoch": 0.8872303091713417,
      "grad_norm": 4.79003518255959,
      "learning_rate": 6.589081538051144e-09,
      "loss": 0.7822,
      "step": 8975
    },
    {
      "epoch": 0.8873291649160961,
      "grad_norm": 5.562793045303385,
      "learning_rate": 6.577655187521902e-09,
      "loss": 0.6674,
      "step": 8976
    },
    {
      "epoch": 0.8874280206608507,
      "grad_norm": 3.09788781676729,
      "learning_rate": 6.566238416076053e-09,
      "loss": 0.6766,
      "step": 8977
    },
    {
      "epoch": 0.8875268764056051,
      "grad_norm": 2.968956097039168,
      "learning_rate": 6.554831224884206e-09,
      "loss": 0.702,
      "step": 8978
    },
    {
      "epoch": 0.8876257321503596,
      "grad_norm": 6.9303927536382925,
      "learning_rate": 6.543433615116001e-09,
      "loss": 0.7222,
      "step": 8979
    },
    {
      "epoch": 0.8877245878951141,
      "grad_norm": 7.8917198025548485,
      "learning_rate": 6.532045587940094e-09,
      "loss": 0.7229,
      "step": 8980
    },
    {
      "epoch": 0.8878234436398685,
      "grad_norm": 4.835954142192795,
      "learning_rate": 6.520667144524183e-09,
      "loss": 0.7248,
      "step": 8981
    },
    {
      "epoch": 0.887922299384623,
      "grad_norm": 40.49474806392048,
      "learning_rate": 6.5092982860349146e-09,
      "loss": 0.5968,
      "step": 8982
    },
    {
      "epoch": 0.8880211551293774,
      "grad_norm": 4.190641662101274,
      "learning_rate": 6.497939013638043e-09,
      "loss": 0.7244,
      "step": 8983
    },
    {
      "epoch": 0.888120010874132,
      "grad_norm": 4.533030723035096,
      "learning_rate": 6.486589328498271e-09,
      "loss": 0.7445,
      "step": 8984
    },
    {
      "epoch": 0.8882188666188864,
      "grad_norm": 2.7907242597095587,
      "learning_rate": 6.475249231779334e-09,
      "loss": 0.6589,
      "step": 8985
    },
    {
      "epoch": 0.8883177223636408,
      "grad_norm": 3.190521678225836,
      "learning_rate": 6.463918724644013e-09,
      "loss": 0.6019,
      "step": 8986
    },
    {
      "epoch": 0.8884165781083954,
      "grad_norm": 3.8617659041462575,
      "learning_rate": 6.452597808254046e-09,
      "loss": 0.6229,
      "step": 8987
    },
    {
      "epoch": 0.8885154338531498,
      "grad_norm": 4.428409455016847,
      "learning_rate": 6.44128648377027e-09,
      "loss": 0.6376,
      "step": 8988
    },
    {
      "epoch": 0.8886142895979042,
      "grad_norm": 4.540746245508736,
      "learning_rate": 6.42998475235248e-09,
      "loss": 0.6683,
      "step": 8989
    },
    {
      "epoch": 0.8887131453426588,
      "grad_norm": 3.6543412219064075,
      "learning_rate": 6.418692615159493e-09,
      "loss": 0.7449,
      "step": 8990
    },
    {
      "epoch": 0.8888120010874132,
      "grad_norm": 3.893454205360025,
      "learning_rate": 6.407410073349151e-09,
      "loss": 0.6743,
      "step": 8991
    },
    {
      "epoch": 0.8889108568321676,
      "grad_norm": 14.483955770154605,
      "learning_rate": 6.396137128078327e-09,
      "loss": 0.6984,
      "step": 8992
    },
    {
      "epoch": 0.8890097125769221,
      "grad_norm": 3.7805706005587876,
      "learning_rate": 6.3848737805028865e-09,
      "loss": 0.703,
      "step": 8993
    },
    {
      "epoch": 0.8891085683216766,
      "grad_norm": 5.058099376722726,
      "learning_rate": 6.373620031777727e-09,
      "loss": 0.6807,
      "step": 8994
    },
    {
      "epoch": 0.889207424066431,
      "grad_norm": 4.647019801950886,
      "learning_rate": 6.3623758830567256e-09,
      "loss": 0.622,
      "step": 8995
    },
    {
      "epoch": 0.8893062798111855,
      "grad_norm": 6.4047763710172525,
      "learning_rate": 6.351141335492827e-09,
      "loss": 0.6561,
      "step": 8996
    },
    {
      "epoch": 0.88940513555594,
      "grad_norm": 2.7613180098149552,
      "learning_rate": 6.339916390237987e-09,
      "loss": 0.6959,
      "step": 8997
    },
    {
      "epoch": 0.8895039913006945,
      "grad_norm": 4.957547271908797,
      "learning_rate": 6.32870104844313e-09,
      "loss": 0.686,
      "step": 8998
    },
    {
      "epoch": 0.8896028470454489,
      "grad_norm": 3.8640531531698707,
      "learning_rate": 6.317495311258214e-09,
      "loss": 0.6848,
      "step": 8999
    },
    {
      "epoch": 0.8897017027902034,
      "grad_norm": 13.329761460473481,
      "learning_rate": 6.3062991798322644e-09,
      "loss": 0.6856,
      "step": 9000
    },
    {
      "epoch": 0.8898005585349579,
      "grad_norm": 5.135860137562225,
      "learning_rate": 6.295112655313261e-09,
      "loss": 0.666,
      "step": 9001
    },
    {
      "epoch": 0.8898994142797123,
      "grad_norm": 4.181638158252477,
      "learning_rate": 6.28393573884819e-09,
      "loss": 0.6913,
      "step": 9002
    },
    {
      "epoch": 0.8899982700244667,
      "grad_norm": 2.946539925146132,
      "learning_rate": 6.272768431583109e-09,
      "loss": 0.6394,
      "step": 9003
    },
    {
      "epoch": 0.8900971257692213,
      "grad_norm": 7.307136132147955,
      "learning_rate": 6.261610734663037e-09,
      "loss": 0.7523,
      "step": 9004
    },
    {
      "epoch": 0.8901959815139757,
      "grad_norm": 2.96374021312273,
      "learning_rate": 6.250462649232069e-09,
      "loss": 0.7002,
      "step": 9005
    },
    {
      "epoch": 0.8902948372587302,
      "grad_norm": 6.976320102927197,
      "learning_rate": 6.239324176433236e-09,
      "loss": 0.7786,
      "step": 9006
    },
    {
      "epoch": 0.8903936930034847,
      "grad_norm": 4.312248087019854,
      "learning_rate": 6.228195317408635e-09,
      "loss": 0.7042,
      "step": 9007
    },
    {
      "epoch": 0.8904925487482391,
      "grad_norm": 3.4778481871968325,
      "learning_rate": 6.217076073299388e-09,
      "loss": 0.6438,
      "step": 9008
    },
    {
      "epoch": 0.8905914044929936,
      "grad_norm": 3.8717620898674387,
      "learning_rate": 6.2059664452455804e-09,
      "loss": 0.7497,
      "step": 9009
    },
    {
      "epoch": 0.8906902602377481,
      "grad_norm": 2.9582186991739685,
      "learning_rate": 6.194866434386359e-09,
      "loss": 0.7707,
      "step": 9010
    },
    {
      "epoch": 0.8907891159825025,
      "grad_norm": 3.414150571634223,
      "learning_rate": 6.1837760418598784e-09,
      "loss": 0.713,
      "step": 9011
    },
    {
      "epoch": 0.890887971727257,
      "grad_norm": 12.10026033582786,
      "learning_rate": 6.172695268803263e-09,
      "loss": 0.7788,
      "step": 9012
    },
    {
      "epoch": 0.8909868274720115,
      "grad_norm": 4.218361531726069,
      "learning_rate": 6.1616241163526925e-09,
      "loss": 0.6538,
      "step": 9013
    },
    {
      "epoch": 0.891085683216766,
      "grad_norm": 3.0889679845656173,
      "learning_rate": 6.150562585643371e-09,
      "loss": 0.7331,
      "step": 9014
    },
    {
      "epoch": 0.8911845389615204,
      "grad_norm": 3.490195572766684,
      "learning_rate": 6.1395106778094675e-09,
      "loss": 0.7725,
      "step": 9015
    },
    {
      "epoch": 0.8912833947062748,
      "grad_norm": 3.3485301183015426,
      "learning_rate": 6.12846839398421e-09,
      "loss": 0.6797,
      "step": 9016
    },
    {
      "epoch": 0.8913822504510294,
      "grad_norm": 4.212954671576331,
      "learning_rate": 6.117435735299814e-09,
      "loss": 0.752,
      "step": 9017
    },
    {
      "epoch": 0.8914811061957838,
      "grad_norm": 6.929240985820221,
      "learning_rate": 6.10641270288752e-09,
      "loss": 0.8735,
      "step": 9018
    },
    {
      "epoch": 0.8915799619405382,
      "grad_norm": 5.083297109324743,
      "learning_rate": 6.095399297877568e-09,
      "loss": 0.7884,
      "step": 9019
    },
    {
      "epoch": 0.8916788176852928,
      "grad_norm": 6.989054297516516,
      "learning_rate": 6.084395521399244e-09,
      "loss": 0.7774,
      "step": 9020
    },
    {
      "epoch": 0.8917776734300472,
      "grad_norm": 4.653964383788346,
      "learning_rate": 6.073401374580789e-09,
      "loss": 0.6881,
      "step": 9021
    },
    {
      "epoch": 0.8918765291748016,
      "grad_norm": 4.165552676218666,
      "learning_rate": 6.062416858549524e-09,
      "loss": 0.7449,
      "step": 9022
    },
    {
      "epoch": 0.8919753849195562,
      "grad_norm": 11.479109517702655,
      "learning_rate": 6.051441974431726e-09,
      "loss": 0.7049,
      "step": 9023
    },
    {
      "epoch": 0.8920742406643106,
      "grad_norm": 3.4793859662421665,
      "learning_rate": 6.040476723352705e-09,
      "loss": 0.8017,
      "step": 9024
    },
    {
      "epoch": 0.8921730964090651,
      "grad_norm": 3.4095329537012713,
      "learning_rate": 6.029521106436819e-09,
      "loss": 0.6643,
      "step": 9025
    },
    {
      "epoch": 0.8922719521538195,
      "grad_norm": 5.068537630488904,
      "learning_rate": 6.018575124807357e-09,
      "loss": 0.6505,
      "step": 9026
    },
    {
      "epoch": 0.892370807898574,
      "grad_norm": 3.805180777231529,
      "learning_rate": 6.00763877958671e-09,
      "loss": 0.6463,
      "step": 9027
    },
    {
      "epoch": 0.8924696636433285,
      "grad_norm": 8.495149118695764,
      "learning_rate": 5.996712071896204e-09,
      "loss": 0.709,
      "step": 9028
    },
    {
      "epoch": 0.8925685193880829,
      "grad_norm": 5.255575198595314,
      "learning_rate": 5.985795002856231e-09,
      "loss": 0.7306,
      "step": 9029
    },
    {
      "epoch": 0.8926673751328374,
      "grad_norm": 13.868492499827381,
      "learning_rate": 5.974887573586196e-09,
      "loss": 0.6884,
      "step": 9030
    },
    {
      "epoch": 0.8927662308775919,
      "grad_norm": 4.086260077727874,
      "learning_rate": 5.963989785204448e-09,
      "loss": 0.6899,
      "step": 9031
    },
    {
      "epoch": 0.8928650866223463,
      "grad_norm": 4.83524296422827,
      "learning_rate": 5.953101638828417e-09,
      "loss": 0.6506,
      "step": 9032
    },
    {
      "epoch": 0.8929639423671009,
      "grad_norm": 3.524309093097231,
      "learning_rate": 5.942223135574554e-09,
      "loss": 0.6956,
      "step": 9033
    },
    {
      "epoch": 0.8930627981118553,
      "grad_norm": 3.061242048582986,
      "learning_rate": 5.9313542765582316e-09,
      "loss": 0.8005,
      "step": 9034
    },
    {
      "epoch": 0.8931616538566097,
      "grad_norm": 5.033840740380672,
      "learning_rate": 5.920495062893927e-09,
      "loss": 0.6631,
      "step": 9035
    },
    {
      "epoch": 0.8932605096013642,
      "grad_norm": 7.325986510286813,
      "learning_rate": 5.9096454956950816e-09,
      "loss": 0.7263,
      "step": 9036
    },
    {
      "epoch": 0.8933593653461187,
      "grad_norm": 2.916857839480505,
      "learning_rate": 5.898805576074162e-09,
      "loss": 0.6897,
      "step": 9037
    },
    {
      "epoch": 0.8934582210908731,
      "grad_norm": 3.9232604715027417,
      "learning_rate": 5.887975305142656e-09,
      "loss": 0.6747,
      "step": 9038
    },
    {
      "epoch": 0.8935570768356276,
      "grad_norm": 5.587357331909816,
      "learning_rate": 5.877154684010999e-09,
      "loss": 0.6924,
      "step": 9039
    },
    {
      "epoch": 0.8936559325803821,
      "grad_norm": 3.416957362285262,
      "learning_rate": 5.866343713788735e-09,
      "loss": 0.6609,
      "step": 9040
    },
    {
      "epoch": 0.8937547883251366,
      "grad_norm": 3.599630724364669,
      "learning_rate": 5.855542395584368e-09,
      "loss": 0.563,
      "step": 9041
    },
    {
      "epoch": 0.893853644069891,
      "grad_norm": 6.825961874497692,
      "learning_rate": 5.844750730505377e-09,
      "loss": 0.7298,
      "step": 9042
    },
    {
      "epoch": 0.8939524998146455,
      "grad_norm": 6.6110321370393,
      "learning_rate": 5.833968719658311e-09,
      "loss": 0.6394,
      "step": 9043
    },
    {
      "epoch": 0.8940513555594,
      "grad_norm": 11.720121781431233,
      "learning_rate": 5.82319636414873e-09,
      "loss": 0.6136,
      "step": 9044
    },
    {
      "epoch": 0.8941502113041544,
      "grad_norm": 9.3804149128615,
      "learning_rate": 5.812433665081129e-09,
      "loss": 0.7575,
      "step": 9045
    },
    {
      "epoch": 0.8942490670489088,
      "grad_norm": 4.823477261196422,
      "learning_rate": 5.801680623559102e-09,
      "loss": 0.7094,
      "step": 9046
    },
    {
      "epoch": 0.8943479227936634,
      "grad_norm": 8.46788510396276,
      "learning_rate": 5.790937240685212e-09,
      "loss": 0.6917,
      "step": 9047
    },
    {
      "epoch": 0.8944467785384178,
      "grad_norm": 11.54979699663987,
      "learning_rate": 5.780203517561011e-09,
      "loss": 0.7065,
      "step": 9048
    },
    {
      "epoch": 0.8945456342831722,
      "grad_norm": 2.912727532031011,
      "learning_rate": 5.7694794552871185e-09,
      "loss": 0.6422,
      "step": 9049
    },
    {
      "epoch": 0.8946444900279268,
      "grad_norm": 5.304295995793127,
      "learning_rate": 5.7587650549630994e-09,
      "loss": 0.733,
      "step": 9050
    },
    {
      "epoch": 0.8947433457726812,
      "grad_norm": 2.661527394536616,
      "learning_rate": 5.748060317687564e-09,
      "loss": 0.5528,
      "step": 9051
    },
    {
      "epoch": 0.8948422015174357,
      "grad_norm": 3.183297272318141,
      "learning_rate": 5.737365244558145e-09,
      "loss": 0.7162,
      "step": 9052
    },
    {
      "epoch": 0.8949410572621902,
      "grad_norm": 4.6871734251818795,
      "learning_rate": 5.726679836671422e-09,
      "loss": 0.8039,
      "step": 9053
    },
    {
      "epoch": 0.8950399130069446,
      "grad_norm": 3.3112091256687264,
      "learning_rate": 5.716004095123084e-09,
      "loss": 0.6498,
      "step": 9054
    },
    {
      "epoch": 0.8951387687516991,
      "grad_norm": 19.79664047289547,
      "learning_rate": 5.705338021007733e-09,
      "loss": 0.7042,
      "step": 9055
    },
    {
      "epoch": 0.8952376244964535,
      "grad_norm": 3.1102504157905333,
      "learning_rate": 5.694681615419006e-09,
      "loss": 0.6338,
      "step": 9056
    },
    {
      "epoch": 0.895336480241208,
      "grad_norm": 6.993121660563989,
      "learning_rate": 5.684034879449573e-09,
      "loss": 0.699,
      "step": 9057
    },
    {
      "epoch": 0.8954353359859625,
      "grad_norm": 5.430977169890235,
      "learning_rate": 5.673397814191128e-09,
      "loss": 0.5812,
      "step": 9058
    },
    {
      "epoch": 0.8955341917307169,
      "grad_norm": 4.482412218566147,
      "learning_rate": 5.662770420734308e-09,
      "loss": 0.7535,
      "step": 9059
    },
    {
      "epoch": 0.8956330474754715,
      "grad_norm": 5.1549890919692025,
      "learning_rate": 5.6521527001688216e-09,
      "loss": 0.7584,
      "step": 9060
    },
    {
      "epoch": 0.8957319032202259,
      "grad_norm": 5.1675340776152,
      "learning_rate": 5.64154465358333e-09,
      "loss": 0.8233,
      "step": 9061
    },
    {
      "epoch": 0.8958307589649803,
      "grad_norm": 3.850055254095426,
      "learning_rate": 5.630946282065552e-09,
      "loss": 0.764,
      "step": 9062
    },
    {
      "epoch": 0.8959296147097349,
      "grad_norm": 5.703398948245286,
      "learning_rate": 5.620357586702207e-09,
      "loss": 0.6532,
      "step": 9063
    },
    {
      "epoch": 0.8960284704544893,
      "grad_norm": 5.740486459681621,
      "learning_rate": 5.609778568578982e-09,
      "loss": 0.6895,
      "step": 9064
    },
    {
      "epoch": 0.8961273261992437,
      "grad_norm": 4.112036684430276,
      "learning_rate": 5.5992092287805995e-09,
      "loss": 0.7801,
      "step": 9065
    },
    {
      "epoch": 0.8962261819439982,
      "grad_norm": 5.867129017590787,
      "learning_rate": 5.5886495683908244e-09,
      "loss": 0.6395,
      "step": 9066
    },
    {
      "epoch": 0.8963250376887527,
      "grad_norm": 3.9431844818610378,
      "learning_rate": 5.5780995884923575e-09,
      "loss": 0.6978,
      "step": 9067
    },
    {
      "epoch": 0.8964238934335071,
      "grad_norm": 3.4360856451310138,
      "learning_rate": 5.5675592901669566e-09,
      "loss": 0.7497,
      "step": 9068
    },
    {
      "epoch": 0.8965227491782616,
      "grad_norm": 9.984321583756216,
      "learning_rate": 5.557028674495379e-09,
      "loss": 0.7033,
      "step": 9069
    },
    {
      "epoch": 0.8966216049230161,
      "grad_norm": 21.866246961997962,
      "learning_rate": 5.5465077425573715e-09,
      "loss": 0.6281,
      "step": 9070
    },
    {
      "epoch": 0.8967204606677706,
      "grad_norm": 3.349559942110378,
      "learning_rate": 5.535996495431728e-09,
      "loss": 0.7403,
      "step": 9071
    },
    {
      "epoch": 0.896819316412525,
      "grad_norm": 4.8741040817045596,
      "learning_rate": 5.525494934196195e-09,
      "loss": 0.7504,
      "step": 9072
    },
    {
      "epoch": 0.8969181721572795,
      "grad_norm": 8.522772128530024,
      "learning_rate": 5.5150030599275366e-09,
      "loss": 0.7928,
      "step": 9073
    },
    {
      "epoch": 0.897017027902034,
      "grad_norm": 5.691304670056707,
      "learning_rate": 5.5045208737015905e-09,
      "loss": 0.6849,
      "step": 9074
    },
    {
      "epoch": 0.8971158836467884,
      "grad_norm": 4.883295442914156,
      "learning_rate": 5.494048376593108e-09,
      "loss": 0.7323,
      "step": 9075
    },
    {
      "epoch": 0.8972147393915428,
      "grad_norm": 3.6902118230138177,
      "learning_rate": 5.483585569675897e-09,
      "loss": 0.7778,
      "step": 9076
    },
    {
      "epoch": 0.8973135951362974,
      "grad_norm": 3.3539238296993075,
      "learning_rate": 5.473132454022788e-09,
      "loss": 0.675,
      "step": 9077
    },
    {
      "epoch": 0.8974124508810518,
      "grad_norm": 3.119658709639869,
      "learning_rate": 5.462689030705558e-09,
      "loss": 0.6749,
      "step": 9078
    },
    {
      "epoch": 0.8975113066258062,
      "grad_norm": 2.999884114519488,
      "learning_rate": 5.45225530079505e-09,
      "loss": 0.648,
      "step": 9079
    },
    {
      "epoch": 0.8976101623705608,
      "grad_norm": 4.160503732036003,
      "learning_rate": 5.441831265361096e-09,
      "loss": 0.7173,
      "step": 9080
    },
    {
      "epoch": 0.8977090181153152,
      "grad_norm": 4.167825264338345,
      "learning_rate": 5.431416925472498e-09,
      "loss": 0.6897,
      "step": 9081
    },
    {
      "epoch": 0.8978078738600697,
      "grad_norm": 7.392268336783591,
      "learning_rate": 5.4210122821971235e-09,
      "loss": 0.6878,
      "step": 9082
    },
    {
      "epoch": 0.8979067296048242,
      "grad_norm": 3.4941309848222066,
      "learning_rate": 5.410617336601797e-09,
      "loss": 0.7539,
      "step": 9083
    },
    {
      "epoch": 0.8980055853495786,
      "grad_norm": 8.375166878139579,
      "learning_rate": 5.400232089752355e-09,
      "loss": 0.6155,
      "step": 9084
    },
    {
      "epoch": 0.8981044410943331,
      "grad_norm": 3.1230967650450787,
      "learning_rate": 5.389856542713689e-09,
      "loss": 0.6892,
      "step": 9085
    },
    {
      "epoch": 0.8982032968390875,
      "grad_norm": 3.995060930152873,
      "learning_rate": 5.379490696549627e-09,
      "loss": 0.7315,
      "step": 9086
    },
    {
      "epoch": 0.898302152583842,
      "grad_norm": 3.639293663231565,
      "learning_rate": 5.369134552323051e-09,
      "loss": 0.7155,
      "step": 9087
    },
    {
      "epoch": 0.8984010083285965,
      "grad_norm": 3.610226268734547,
      "learning_rate": 5.358788111095814e-09,
      "loss": 0.754,
      "step": 9088
    },
    {
      "epoch": 0.8984998640733509,
      "grad_norm": 3.57608861158042,
      "learning_rate": 5.348451373928808e-09,
      "loss": 0.7054,
      "step": 9089
    },
    {
      "epoch": 0.8985987198181055,
      "grad_norm": 5.857492662603637,
      "learning_rate": 5.3381243418819e-09,
      "loss": 0.787,
      "step": 9090
    },
    {
      "epoch": 0.8986975755628599,
      "grad_norm": 16.150956912896042,
      "learning_rate": 5.327807016013985e-09,
      "loss": 0.659,
      "step": 9091
    },
    {
      "epoch": 0.8987964313076143,
      "grad_norm": 6.72392288990043,
      "learning_rate": 5.3174993973829404e-09,
      "loss": 0.8339,
      "step": 9092
    },
    {
      "epoch": 0.8988952870523689,
      "grad_norm": 8.20118051089745,
      "learning_rate": 5.307201487045676e-09,
      "loss": 0.6369,
      "step": 9093
    },
    {
      "epoch": 0.8989941427971233,
      "grad_norm": 8.282513268639477,
      "learning_rate": 5.296913286058069e-09,
      "loss": 0.6453,
      "step": 9094
    },
    {
      "epoch": 0.8990929985418777,
      "grad_norm": 3.348955257899067,
      "learning_rate": 5.286634795475042e-09,
      "loss": 0.6654,
      "step": 9095
    },
    {
      "epoch": 0.8991918542866323,
      "grad_norm": 6.222665655154139,
      "learning_rate": 5.276366016350508e-09,
      "loss": 0.7257,
      "step": 9096
    },
    {
      "epoch": 0.8992907100313867,
      "grad_norm": 4.21566590862493,
      "learning_rate": 5.266106949737348e-09,
      "loss": 0.7653,
      "step": 9097
    },
    {
      "epoch": 0.8993895657761412,
      "grad_norm": 4.900619409465956,
      "learning_rate": 5.255857596687497e-09,
      "loss": 0.6568,
      "step": 9098
    },
    {
      "epoch": 0.8994884215208956,
      "grad_norm": 3.6187592824639436,
      "learning_rate": 5.245617958251891e-09,
      "loss": 0.6318,
      "step": 9099
    },
    {
      "epoch": 0.8995872772656501,
      "grad_norm": 10.901697593603277,
      "learning_rate": 5.235388035480426e-09,
      "loss": 0.782,
      "step": 9100
    },
    {
      "epoch": 0.8996861330104046,
      "grad_norm": 5.750625274174728,
      "learning_rate": 5.2251678294220395e-09,
      "loss": 0.6619,
      "step": 9101
    },
    {
      "epoch": 0.899784988755159,
      "grad_norm": 5.183019381626044,
      "learning_rate": 5.214957341124682e-09,
      "loss": 0.5876,
      "step": 9102
    },
    {
      "epoch": 0.8998838444999135,
      "grad_norm": 3.0465757329157794,
      "learning_rate": 5.204756571635261e-09,
      "loss": 0.7354,
      "step": 9103
    },
    {
      "epoch": 0.899982700244668,
      "grad_norm": 3.4343915947460513,
      "learning_rate": 5.194565521999728e-09,
      "loss": 0.7344,
      "step": 9104
    },
    {
      "epoch": 0.9000815559894224,
      "grad_norm": 9.604479466289362,
      "learning_rate": 5.184384193263014e-09,
      "loss": 0.6272,
      "step": 9105
    },
    {
      "epoch": 0.900180411734177,
      "grad_norm": 5.078502020966605,
      "learning_rate": 5.174212586469084e-09,
      "loss": 0.6087,
      "step": 9106
    },
    {
      "epoch": 0.9002792674789314,
      "grad_norm": 8.839361932451537,
      "learning_rate": 5.16405070266086e-09,
      "loss": 0.7852,
      "step": 9107
    },
    {
      "epoch": 0.9003781232236858,
      "grad_norm": 16.291408825058742,
      "learning_rate": 5.153898542880331e-09,
      "loss": 0.6661,
      "step": 9108
    },
    {
      "epoch": 0.9004769789684403,
      "grad_norm": 9.988033833899815,
      "learning_rate": 5.1437561081684065e-09,
      "loss": 0.6419,
      "step": 9109
    },
    {
      "epoch": 0.9005758347131948,
      "grad_norm": 8.327939792923141,
      "learning_rate": 5.133623399565079e-09,
      "loss": 0.6527,
      "step": 9110
    },
    {
      "epoch": 0.9006746904579492,
      "grad_norm": 3.5837198660212928,
      "learning_rate": 5.123500418109272e-09,
      "loss": 0.658,
      "step": 9111
    },
    {
      "epoch": 0.9007735462027037,
      "grad_norm": 12.63504232241865,
      "learning_rate": 5.113387164838989e-09,
      "loss": 0.7589,
      "step": 9112
    },
    {
      "epoch": 0.9008724019474582,
      "grad_norm": 7.094367936253062,
      "learning_rate": 5.103283640791178e-09,
      "loss": 0.6842,
      "step": 9113
    },
    {
      "epoch": 0.9009712576922126,
      "grad_norm": 3.267531718388435,
      "learning_rate": 5.0931898470017886e-09,
      "loss": 0.662,
      "step": 9114
    },
    {
      "epoch": 0.9010701134369671,
      "grad_norm": 4.368424453260849,
      "learning_rate": 5.083105784505826e-09,
      "loss": 0.7322,
      "step": 9115
    },
    {
      "epoch": 0.9011689691817216,
      "grad_norm": 4.0387402343416205,
      "learning_rate": 5.073031454337229e-09,
      "loss": 0.5428,
      "step": 9116
    },
    {
      "epoch": 0.9012678249264761,
      "grad_norm": 5.515160745670795,
      "learning_rate": 5.062966857528983e-09,
      "loss": 0.6834,
      "step": 9117
    },
    {
      "epoch": 0.9013666806712305,
      "grad_norm": 7.6455174936224815,
      "learning_rate": 5.052911995113074e-09,
      "loss": 0.6217,
      "step": 9118
    },
    {
      "epoch": 0.9014655364159849,
      "grad_norm": 4.54992581097053,
      "learning_rate": 5.042866868120466e-09,
      "loss": 0.7917,
      "step": 9119
    },
    {
      "epoch": 0.9015643921607395,
      "grad_norm": 4.1676280825882,
      "learning_rate": 5.032831477581145e-09,
      "loss": 0.7805,
      "step": 9120
    },
    {
      "epoch": 0.9016632479054939,
      "grad_norm": 2.726208934793631,
      "learning_rate": 5.022805824524101e-09,
      "loss": 0.5852,
      "step": 9121
    },
    {
      "epoch": 0.9017621036502483,
      "grad_norm": 3.3024583625912673,
      "learning_rate": 5.0127899099772995e-09,
      "loss": 0.7409,
      "step": 9122
    },
    {
      "epoch": 0.9018609593950029,
      "grad_norm": 3.0840779595691874,
      "learning_rate": 5.002783734967742e-09,
      "loss": 0.5984,
      "step": 9123
    },
    {
      "epoch": 0.9019598151397573,
      "grad_norm": 3.952418716489158,
      "learning_rate": 4.992787300521395e-09,
      "loss": 0.7413,
      "step": 9124
    },
    {
      "epoch": 0.9020586708845117,
      "grad_norm": 5.984436501574509,
      "learning_rate": 4.9828006076632626e-09,
      "loss": 0.6936,
      "step": 9125
    },
    {
      "epoch": 0.9021575266292663,
      "grad_norm": 6.678080411569986,
      "learning_rate": 4.9728236574173245e-09,
      "loss": 0.6778,
      "step": 9126
    },
    {
      "epoch": 0.9022563823740207,
      "grad_norm": 4.137278296645323,
      "learning_rate": 4.962856450806563e-09,
      "loss": 0.7594,
      "step": 9127
    },
    {
      "epoch": 0.9023552381187752,
      "grad_norm": 3.09541510390098,
      "learning_rate": 4.952898988852972e-09,
      "loss": 0.5969,
      "step": 9128
    },
    {
      "epoch": 0.9024540938635296,
      "grad_norm": 3.298622031148859,
      "learning_rate": 4.942951272577567e-09,
      "loss": 0.6166,
      "step": 9129
    },
    {
      "epoch": 0.9025529496082841,
      "grad_norm": 2.9604188938755382,
      "learning_rate": 4.933013303000299e-09,
      "loss": 0.7895,
      "step": 9130
    },
    {
      "epoch": 0.9026518053530386,
      "grad_norm": 3.3278858436309444,
      "learning_rate": 4.923085081140188e-09,
      "loss": 0.7391,
      "step": 9131
    },
    {
      "epoch": 0.902750661097793,
      "grad_norm": 4.693182347030312,
      "learning_rate": 4.913166608015218e-09,
      "loss": 0.7931,
      "step": 9132
    },
    {
      "epoch": 0.9028495168425476,
      "grad_norm": 3.40290267741298,
      "learning_rate": 4.9032578846423865e-09,
      "loss": 0.7497,
      "step": 9133
    },
    {
      "epoch": 0.902948372587302,
      "grad_norm": 4.445475895764255,
      "learning_rate": 4.8933589120376706e-09,
      "loss": 0.7445,
      "step": 9134
    },
    {
      "epoch": 0.9030472283320564,
      "grad_norm": 27.10899936477761,
      "learning_rate": 4.883469691216091e-09,
      "loss": 0.7498,
      "step": 9135
    },
    {
      "epoch": 0.903146084076811,
      "grad_norm": 4.828366005176653,
      "learning_rate": 4.8735902231916144e-09,
      "loss": 0.7948,
      "step": 9136
    },
    {
      "epoch": 0.9032449398215654,
      "grad_norm": 4.520922345392458,
      "learning_rate": 4.863720508977265e-09,
      "loss": 0.7133,
      "step": 9137
    },
    {
      "epoch": 0.9033437955663198,
      "grad_norm": 3.9366300668740357,
      "learning_rate": 4.853860549584998e-09,
      "loss": 0.6592,
      "step": 9138
    },
    {
      "epoch": 0.9034426513110743,
      "grad_norm": 5.082017815111666,
      "learning_rate": 4.84401034602584e-09,
      "loss": 0.6666,
      "step": 9139
    },
    {
      "epoch": 0.9035415070558288,
      "grad_norm": 3.659387955031945,
      "learning_rate": 4.83416989930977e-09,
      "loss": 0.7275,
      "step": 9140
    },
    {
      "epoch": 0.9036403628005832,
      "grad_norm": 6.527314847258016,
      "learning_rate": 4.824339210445794e-09,
      "loss": 0.6639,
      "step": 9141
    },
    {
      "epoch": 0.9037392185453377,
      "grad_norm": 6.170357481234975,
      "learning_rate": 4.81451828044187e-09,
      "loss": 0.666,
      "step": 9142
    },
    {
      "epoch": 0.9038380742900922,
      "grad_norm": 4.15418579377171,
      "learning_rate": 4.8047071103050284e-09,
      "loss": 0.672,
      "step": 9143
    },
    {
      "epoch": 0.9039369300348467,
      "grad_norm": 5.700091675504564,
      "learning_rate": 4.79490570104123e-09,
      "loss": 0.7578,
      "step": 9144
    },
    {
      "epoch": 0.9040357857796011,
      "grad_norm": 6.306017738963321,
      "learning_rate": 4.785114053655481e-09,
      "loss": 0.7114,
      "step": 9145
    },
    {
      "epoch": 0.9041346415243556,
      "grad_norm": 4.353419037464043,
      "learning_rate": 4.77533216915178e-09,
      "loss": 0.649,
      "step": 9146
    },
    {
      "epoch": 0.9042334972691101,
      "grad_norm": 5.196233454332332,
      "learning_rate": 4.765560048533079e-09,
      "loss": 0.7439,
      "step": 9147
    },
    {
      "epoch": 0.9043323530138645,
      "grad_norm": 3.229132138747151,
      "learning_rate": 4.755797692801411e-09,
      "loss": 0.732,
      "step": 9148
    },
    {
      "epoch": 0.9044312087586189,
      "grad_norm": 3.430926671494994,
      "learning_rate": 4.74604510295773e-09,
      "loss": 0.7313,
      "step": 9149
    },
    {
      "epoch": 0.9045300645033735,
      "grad_norm": 4.993021636451876,
      "learning_rate": 4.736302280002025e-09,
      "loss": 0.6791,
      "step": 9150
    },
    {
      "epoch": 0.9046289202481279,
      "grad_norm": 3.7545931791677902,
      "learning_rate": 4.726569224933286e-09,
      "loss": 0.7077,
      "step": 9151
    },
    {
      "epoch": 0.9047277759928823,
      "grad_norm": 4.492517168270949,
      "learning_rate": 4.7168459387494915e-09,
      "loss": 0.7858,
      "step": 9152
    },
    {
      "epoch": 0.9048266317376369,
      "grad_norm": 4.8936173155249865,
      "learning_rate": 4.707132422447613e-09,
      "loss": 0.6661,
      "step": 9153
    },
    {
      "epoch": 0.9049254874823913,
      "grad_norm": 8.0483465117594,
      "learning_rate": 4.6974286770236514e-09,
      "loss": 0.7807,
      "step": 9154
    },
    {
      "epoch": 0.9050243432271458,
      "grad_norm": 5.0692161784939,
      "learning_rate": 4.687734703472546e-09,
      "loss": 0.6535,
      "step": 9155
    },
    {
      "epoch": 0.9051231989719003,
      "grad_norm": 7.857257734350887,
      "learning_rate": 4.6780505027883e-09,
      "loss": 0.6921,
      "step": 9156
    },
    {
      "epoch": 0.9052220547166547,
      "grad_norm": 3.641741634881134,
      "learning_rate": 4.668376075963898e-09,
      "loss": 0.7254,
      "step": 9157
    },
    {
      "epoch": 0.9053209104614092,
      "grad_norm": 7.366780785575918,
      "learning_rate": 4.65871142399128e-09,
      "loss": 0.8213,
      "step": 9158
    },
    {
      "epoch": 0.9054197662061636,
      "grad_norm": 3.667250024952076,
      "learning_rate": 4.649056547861407e-09,
      "loss": 0.6443,
      "step": 9159
    },
    {
      "epoch": 0.9055186219509181,
      "grad_norm": 3.8927745258362236,
      "learning_rate": 4.639411448564279e-09,
      "loss": 0.7763,
      "step": 9160
    },
    {
      "epoch": 0.9056174776956726,
      "grad_norm": 6.286606537532318,
      "learning_rate": 4.629776127088814e-09,
      "loss": 0.761,
      "step": 9161
    },
    {
      "epoch": 0.905716333440427,
      "grad_norm": 4.462630520304676,
      "learning_rate": 4.620150584423021e-09,
      "loss": 0.7915,
      "step": 9162
    },
    {
      "epoch": 0.9058151891851816,
      "grad_norm": 12.323341772066613,
      "learning_rate": 4.610534821553824e-09,
      "loss": 0.6968,
      "step": 9163
    },
    {
      "epoch": 0.905914044929936,
      "grad_norm": 3.3455902280067558,
      "learning_rate": 4.600928839467177e-09,
      "loss": 0.7132,
      "step": 9164
    },
    {
      "epoch": 0.9060129006746904,
      "grad_norm": 8.660503840222322,
      "learning_rate": 4.59133263914806e-09,
      "loss": 0.723,
      "step": 9165
    },
    {
      "epoch": 0.906111756419445,
      "grad_norm": 7.178735333852051,
      "learning_rate": 4.581746221580396e-09,
      "loss": 0.6066,
      "step": 9166
    },
    {
      "epoch": 0.9062106121641994,
      "grad_norm": 4.793942003078857,
      "learning_rate": 4.572169587747132e-09,
      "loss": 0.8233,
      "step": 9167
    },
    {
      "epoch": 0.9063094679089538,
      "grad_norm": 4.1479795756526805,
      "learning_rate": 4.562602738630239e-09,
      "loss": 0.6846,
      "step": 9168
    },
    {
      "epoch": 0.9064083236537084,
      "grad_norm": 12.328306075205111,
      "learning_rate": 4.55304567521061e-09,
      "loss": 0.6102,
      "step": 9169
    },
    {
      "epoch": 0.9065071793984628,
      "grad_norm": 7.18462582334617,
      "learning_rate": 4.543498398468226e-09,
      "loss": 0.5514,
      "step": 9170
    },
    {
      "epoch": 0.9066060351432172,
      "grad_norm": 4.7173500363385505,
      "learning_rate": 4.5339609093819845e-09,
      "loss": 0.8319,
      "step": 9171
    },
    {
      "epoch": 0.9067048908879717,
      "grad_norm": 3.4431375693037576,
      "learning_rate": 4.524433208929823e-09,
      "loss": 0.7881,
      "step": 9172
    },
    {
      "epoch": 0.9068037466327262,
      "grad_norm": 8.525410044459592,
      "learning_rate": 4.5149152980886955e-09,
      "loss": 0.7845,
      "step": 9173
    },
    {
      "epoch": 0.9069026023774807,
      "grad_norm": 17.56127226046748,
      "learning_rate": 4.505407177834475e-09,
      "loss": 0.7543,
      "step": 9174
    },
    {
      "epoch": 0.9070014581222351,
      "grad_norm": 6.332380267674635,
      "learning_rate": 4.495908849142127e-09,
      "loss": 0.7574,
      "step": 9175
    },
    {
      "epoch": 0.9071003138669896,
      "grad_norm": 4.970590289637656,
      "learning_rate": 4.486420312985528e-09,
      "loss": 0.6779,
      "step": 9176
    },
    {
      "epoch": 0.9071991696117441,
      "grad_norm": 3.3821020093764353,
      "learning_rate": 4.4769415703376115e-09,
      "loss": 0.7232,
      "step": 9177
    },
    {
      "epoch": 0.9072980253564985,
      "grad_norm": 8.77429492204542,
      "learning_rate": 4.467472622170254e-09,
      "loss": 0.7148,
      "step": 9178
    },
    {
      "epoch": 0.907396881101253,
      "grad_norm": 3.300041622122075,
      "learning_rate": 4.458013469454392e-09,
      "loss": 0.7613,
      "step": 9179
    },
    {
      "epoch": 0.9074957368460075,
      "grad_norm": 3.1294584695264915,
      "learning_rate": 4.4485641131599024e-09,
      "loss": 0.6238,
      "step": 9180
    },
    {
      "epoch": 0.9075945925907619,
      "grad_norm": 5.842354043094671,
      "learning_rate": 4.43912455425568e-09,
      "loss": 0.7426,
      "step": 9181
    },
    {
      "epoch": 0.9076934483355164,
      "grad_norm": 3.448425872612245,
      "learning_rate": 4.429694793709604e-09,
      "loss": 0.7205,
      "step": 9182
    },
    {
      "epoch": 0.9077923040802709,
      "grad_norm": 3.6422889285752795,
      "learning_rate": 4.420274832488568e-09,
      "loss": 0.7602,
      "step": 9183
    },
    {
      "epoch": 0.9078911598250253,
      "grad_norm": 5.1073963048987245,
      "learning_rate": 4.410864671558456e-09,
      "loss": 0.7249,
      "step": 9184
    },
    {
      "epoch": 0.9079900155697798,
      "grad_norm": 7.05371822605644,
      "learning_rate": 4.401464311884129e-09,
      "loss": 0.6958,
      "step": 9185
    },
    {
      "epoch": 0.9080888713145343,
      "grad_norm": 8.144070996780728,
      "learning_rate": 4.392073754429448e-09,
      "loss": 0.7487,
      "step": 9186
    },
    {
      "epoch": 0.9081877270592887,
      "grad_norm": 3.3683058358430435,
      "learning_rate": 4.3826930001573116e-09,
      "loss": 0.6726,
      "step": 9187
    },
    {
      "epoch": 0.9082865828040432,
      "grad_norm": 3.746071205742553,
      "learning_rate": 4.373322050029527e-09,
      "loss": 0.6821,
      "step": 9188
    },
    {
      "epoch": 0.9083854385487977,
      "grad_norm": 4.462880768512415,
      "learning_rate": 4.363960905006981e-09,
      "loss": 0.711,
      "step": 9189
    },
    {
      "epoch": 0.9084842942935522,
      "grad_norm": 3.510071019385119,
      "learning_rate": 4.354609566049528e-09,
      "loss": 0.6632,
      "step": 9190
    },
    {
      "epoch": 0.9085831500383066,
      "grad_norm": 3.3664182204481996,
      "learning_rate": 4.345268034115979e-09,
      "loss": 0.6679,
      "step": 9191
    },
    {
      "epoch": 0.908682005783061,
      "grad_norm": 3.6209106998243064,
      "learning_rate": 4.335936310164201e-09,
      "loss": 0.6491,
      "step": 9192
    },
    {
      "epoch": 0.9087808615278156,
      "grad_norm": 5.080519302283076,
      "learning_rate": 4.326614395150996e-09,
      "loss": 0.6247,
      "step": 9193
    },
    {
      "epoch": 0.90887971727257,
      "grad_norm": 5.930758502776924,
      "learning_rate": 4.317302290032221e-09,
      "loss": 0.676,
      "step": 9194
    },
    {
      "epoch": 0.9089785730173244,
      "grad_norm": 2.969010412379885,
      "learning_rate": 4.307999995762657e-09,
      "loss": 0.7604,
      "step": 9195
    },
    {
      "epoch": 0.909077428762079,
      "grad_norm": 5.204428468281225,
      "learning_rate": 4.298707513296151e-09,
      "loss": 0.729,
      "step": 9196
    },
    {
      "epoch": 0.9091762845068334,
      "grad_norm": 3.792727533705053,
      "learning_rate": 4.2894248435854875e-09,
      "loss": 0.6376,
      "step": 9197
    },
    {
      "epoch": 0.9092751402515878,
      "grad_norm": 15.76575439831711,
      "learning_rate": 4.280151987582492e-09,
      "loss": 0.7183,
      "step": 9198
    },
    {
      "epoch": 0.9093739959963424,
      "grad_norm": 11.475887921892777,
      "learning_rate": 4.270888946237927e-09,
      "loss": 0.7504,
      "step": 9199
    },
    {
      "epoch": 0.9094728517410968,
      "grad_norm": 4.171933796855026,
      "learning_rate": 4.261635720501589e-09,
      "loss": 0.682,
      "step": 9200
    },
    {
      "epoch": 0.9095717074858513,
      "grad_norm": 3.0870335314420028,
      "learning_rate": 4.2523923113222836e-09,
      "loss": 0.7382,
      "step": 9201
    },
    {
      "epoch": 0.9096705632306057,
      "grad_norm": 2.869006747061063,
      "learning_rate": 4.243158719647766e-09,
      "loss": 0.7723,
      "step": 9202
    },
    {
      "epoch": 0.9097694189753602,
      "grad_norm": 6.873986863749269,
      "learning_rate": 4.233934946424811e-09,
      "loss": 0.7515,
      "step": 9203
    },
    {
      "epoch": 0.9098682747201147,
      "grad_norm": 8.666341949724863,
      "learning_rate": 4.224720992599173e-09,
      "loss": 0.6836,
      "step": 9204
    },
    {
      "epoch": 0.9099671304648691,
      "grad_norm": 4.382439665689499,
      "learning_rate": 4.215516859115609e-09,
      "loss": 0.6563,
      "step": 9205
    },
    {
      "epoch": 0.9100659862096236,
      "grad_norm": 4.114693030871112,
      "learning_rate": 4.2063225469178845e-09,
      "loss": 0.4812,
      "step": 9206
    },
    {
      "epoch": 0.9101648419543781,
      "grad_norm": 4.053021948548916,
      "learning_rate": 4.197138056948712e-09,
      "loss": 0.7028,
      "step": 9207
    },
    {
      "epoch": 0.9102636976991325,
      "grad_norm": 3.735670011177251,
      "learning_rate": 4.18796339014984e-09,
      "loss": 0.7408,
      "step": 9208
    },
    {
      "epoch": 0.9103625534438871,
      "grad_norm": 3.4970588299049457,
      "learning_rate": 4.178798547462015e-09,
      "loss": 0.7466,
      "step": 9209
    },
    {
      "epoch": 0.9104614091886415,
      "grad_norm": 3.865972687879547,
      "learning_rate": 4.169643529824918e-09,
      "loss": 0.624,
      "step": 9210
    },
    {
      "epoch": 0.9105602649333959,
      "grad_norm": 2.974082807789901,
      "learning_rate": 4.1604983381772876e-09,
      "loss": 0.6793,
      "step": 9211
    },
    {
      "epoch": 0.9106591206781504,
      "grad_norm": 10.04376664598997,
      "learning_rate": 4.151362973456829e-09,
      "loss": 0.8244,
      "step": 9212
    },
    {
      "epoch": 0.9107579764229049,
      "grad_norm": 5.208614615696838,
      "learning_rate": 4.1422374366002154e-09,
      "loss": 0.8731,
      "step": 9213
    },
    {
      "epoch": 0.9108568321676593,
      "grad_norm": 8.014865117176841,
      "learning_rate": 4.133121728543165e-09,
      "loss": 0.7424,
      "step": 9214
    },
    {
      "epoch": 0.9109556879124138,
      "grad_norm": 3.6980153682093166,
      "learning_rate": 4.124015850220331e-09,
      "loss": 0.6738,
      "step": 9215
    },
    {
      "epoch": 0.9110545436571683,
      "grad_norm": 2.665152801681194,
      "learning_rate": 4.114919802565397e-09,
      "loss": 0.7255,
      "step": 9216
    },
    {
      "epoch": 0.9111533994019227,
      "grad_norm": 6.702429372608779,
      "learning_rate": 4.105833586511054e-09,
      "loss": 0.68,
      "step": 9217
    },
    {
      "epoch": 0.9112522551466772,
      "grad_norm": 6.196124464351072,
      "learning_rate": 4.096757202988921e-09,
      "loss": 0.6501,
      "step": 9218
    },
    {
      "epoch": 0.9113511108914317,
      "grad_norm": 19.54025269125227,
      "learning_rate": 4.087690652929665e-09,
      "loss": 0.6889,
      "step": 9219
    },
    {
      "epoch": 0.9114499666361862,
      "grad_norm": 3.7811333890956678,
      "learning_rate": 4.078633937262932e-09,
      "loss": 0.6077,
      "step": 9220
    },
    {
      "epoch": 0.9115488223809406,
      "grad_norm": 3.02598678349501,
      "learning_rate": 4.069587056917345e-09,
      "loss": 0.6386,
      "step": 9221
    },
    {
      "epoch": 0.911647678125695,
      "grad_norm": 9.812229757611037,
      "learning_rate": 4.06055001282053e-09,
      "loss": 0.6498,
      "step": 9222
    },
    {
      "epoch": 0.9117465338704496,
      "grad_norm": 3.435322394449184,
      "learning_rate": 4.051522805899121e-09,
      "loss": 0.6909,
      "step": 9223
    },
    {
      "epoch": 0.911845389615204,
      "grad_norm": 5.093478416736397,
      "learning_rate": 4.042505437078692e-09,
      "loss": 0.7712,
      "step": 9224
    },
    {
      "epoch": 0.9119442453599584,
      "grad_norm": 4.229582456985856,
      "learning_rate": 4.0334979072838674e-09,
      "loss": 0.753,
      "step": 9225
    },
    {
      "epoch": 0.912043101104713,
      "grad_norm": 3.469554935949177,
      "learning_rate": 4.0245002174382205e-09,
      "loss": 0.6988,
      "step": 9226
    },
    {
      "epoch": 0.9121419568494674,
      "grad_norm": 7.365487132079871,
      "learning_rate": 4.015512368464336e-09,
      "loss": 0.6537,
      "step": 9227
    },
    {
      "epoch": 0.9122408125942219,
      "grad_norm": 13.479612707147696,
      "learning_rate": 4.006534361283798e-09,
      "loss": 0.6855,
      "step": 9228
    },
    {
      "epoch": 0.9123396683389764,
      "grad_norm": 6.415872931415979,
      "learning_rate": 3.997566196817159e-09,
      "loss": 0.7538,
      "step": 9229
    },
    {
      "epoch": 0.9124385240837308,
      "grad_norm": 4.3242681933872245,
      "learning_rate": 3.9886078759839624e-09,
      "loss": 0.7651,
      "step": 9230
    },
    {
      "epoch": 0.9125373798284853,
      "grad_norm": 3.531653815952403,
      "learning_rate": 3.979659399702773e-09,
      "loss": 0.6288,
      "step": 9231
    },
    {
      "epoch": 0.9126362355732397,
      "grad_norm": 3.702377417090328,
      "learning_rate": 3.9707207688910896e-09,
      "loss": 0.7389,
      "step": 9232
    },
    {
      "epoch": 0.9127350913179942,
      "grad_norm": 5.076857106069224,
      "learning_rate": 3.961791984465468e-09,
      "loss": 0.6269,
      "step": 9233
    },
    {
      "epoch": 0.9128339470627487,
      "grad_norm": 4.478393465335068,
      "learning_rate": 3.9528730473414315e-09,
      "loss": 0.6887,
      "step": 9234
    },
    {
      "epoch": 0.9129328028075031,
      "grad_norm": 4.443941183243416,
      "learning_rate": 3.943963958433449e-09,
      "loss": 0.6495,
      "step": 9235
    },
    {
      "epoch": 0.9130316585522577,
      "grad_norm": 4.3111805380406585,
      "learning_rate": 3.9350647186550455e-09,
      "loss": 0.7296,
      "step": 9236
    },
    {
      "epoch": 0.9131305142970121,
      "grad_norm": 3.4652639181472518,
      "learning_rate": 3.92617532891869e-09,
      "loss": 0.6418,
      "step": 9237
    },
    {
      "epoch": 0.9132293700417665,
      "grad_norm": 3.2594762134861095,
      "learning_rate": 3.917295790135877e-09,
      "loss": 0.615,
      "step": 9238
    },
    {
      "epoch": 0.9133282257865211,
      "grad_norm": 4.002180559581622,
      "learning_rate": 3.9084261032170554e-09,
      "loss": 0.6916,
      "step": 9239
    },
    {
      "epoch": 0.9134270815312755,
      "grad_norm": 3.754785391911426,
      "learning_rate": 3.8995662690716855e-09,
      "loss": 0.648,
      "step": 9240
    },
    {
      "epoch": 0.9135259372760299,
      "grad_norm": 4.723355654959425,
      "learning_rate": 3.890716288608209e-09,
      "loss": 0.639,
      "step": 9241
    },
    {
      "epoch": 0.9136247930207844,
      "grad_norm": 6.7448813757071715,
      "learning_rate": 3.8818761627340765e-09,
      "loss": 0.6397,
      "step": 9242
    },
    {
      "epoch": 0.9137236487655389,
      "grad_norm": 5.5422574450015905,
      "learning_rate": 3.873045892355697e-09,
      "loss": 0.6734,
      "step": 9243
    },
    {
      "epoch": 0.9138225045102933,
      "grad_norm": 4.323750224883757,
      "learning_rate": 3.86422547837848e-09,
      "loss": 0.7015,
      "step": 9244
    },
    {
      "epoch": 0.9139213602550478,
      "grad_norm": 13.249319446498447,
      "learning_rate": 3.855414921706857e-09,
      "loss": 0.6403,
      "step": 9245
    },
    {
      "epoch": 0.9140202159998023,
      "grad_norm": 4.686824255956182,
      "learning_rate": 3.846614223244194e-09,
      "loss": 0.6921,
      "step": 9246
    },
    {
      "epoch": 0.9141190717445568,
      "grad_norm": 14.813762574906415,
      "learning_rate": 3.837823383892869e-09,
      "loss": 0.5978,
      "step": 9247
    },
    {
      "epoch": 0.9142179274893112,
      "grad_norm": 10.254013874212204,
      "learning_rate": 3.829042404554284e-09,
      "loss": 0.6234,
      "step": 9248
    },
    {
      "epoch": 0.9143167832340657,
      "grad_norm": 3.0245895592387373,
      "learning_rate": 3.820271286128762e-09,
      "loss": 0.6068,
      "step": 9249
    },
    {
      "epoch": 0.9144156389788202,
      "grad_norm": 4.931793303757395,
      "learning_rate": 3.811510029515674e-09,
      "loss": 0.6695,
      "step": 9250
    },
    {
      "epoch": 0.9145144947235746,
      "grad_norm": 3.4863823098720927,
      "learning_rate": 3.8027586356133565e-09,
      "loss": 0.7004,
      "step": 9251
    },
    {
      "epoch": 0.9146133504683291,
      "grad_norm": 18.34838853790831,
      "learning_rate": 3.794017105319114e-09,
      "loss": 0.735,
      "step": 9252
    },
    {
      "epoch": 0.9147122062130836,
      "grad_norm": 3.631121291380736,
      "learning_rate": 3.785285439529307e-09,
      "loss": 0.8391,
      "step": 9253
    },
    {
      "epoch": 0.914811061957838,
      "grad_norm": 6.177329566689876,
      "learning_rate": 3.776563639139185e-09,
      "loss": 0.6358,
      "step": 9254
    },
    {
      "epoch": 0.9149099177025924,
      "grad_norm": 5.002852930401152,
      "learning_rate": 3.767851705043079e-09,
      "loss": 0.8186,
      "step": 9255
    },
    {
      "epoch": 0.915008773447347,
      "grad_norm": 14.268079960323641,
      "learning_rate": 3.7591496381342625e-09,
      "loss": 0.6095,
      "step": 9256
    },
    {
      "epoch": 0.9151076291921014,
      "grad_norm": 8.84783672554617,
      "learning_rate": 3.750457439304977e-09,
      "loss": 0.7628,
      "step": 9257
    },
    {
      "epoch": 0.9152064849368559,
      "grad_norm": 3.7536427377578168,
      "learning_rate": 3.7417751094465324e-09,
      "loss": 0.6243,
      "step": 9258
    },
    {
      "epoch": 0.9153053406816104,
      "grad_norm": 3.4165679123794037,
      "learning_rate": 3.733102649449116e-09,
      "loss": 0.6619,
      "step": 9259
    },
    {
      "epoch": 0.9154041964263648,
      "grad_norm": 10.790409931795745,
      "learning_rate": 3.7244400602019833e-09,
      "loss": 0.6562,
      "step": 9260
    },
    {
      "epoch": 0.9155030521711193,
      "grad_norm": 9.511954699671238,
      "learning_rate": 3.7157873425933793e-09,
      "loss": 0.692,
      "step": 9261
    },
    {
      "epoch": 0.9156019079158738,
      "grad_norm": 4.746753216219439,
      "learning_rate": 3.707144497510473e-09,
      "loss": 0.6563,
      "step": 9262
    },
    {
      "epoch": 0.9157007636606282,
      "grad_norm": 3.1887944054248973,
      "learning_rate": 3.6985115258394985e-09,
      "loss": 0.714,
      "step": 9263
    },
    {
      "epoch": 0.9157996194053827,
      "grad_norm": 3.5806343645462015,
      "learning_rate": 3.689888428465604e-09,
      "loss": 0.6365,
      "step": 9264
    },
    {
      "epoch": 0.9158984751501371,
      "grad_norm": 2.8613683524194844,
      "learning_rate": 3.6812752062729825e-09,
      "loss": 0.6933,
      "step": 9265
    },
    {
      "epoch": 0.9159973308948917,
      "grad_norm": 14.147453600226305,
      "learning_rate": 3.6726718601447826e-09,
      "loss": 0.6575,
      "step": 9266
    },
    {
      "epoch": 0.9160961866396461,
      "grad_norm": 4.932456020074218,
      "learning_rate": 3.6640783909631655e-09,
      "loss": 0.7341,
      "step": 9267
    },
    {
      "epoch": 0.9161950423844005,
      "grad_norm": 14.08405230592423,
      "learning_rate": 3.655494799609249e-09,
      "loss": 0.7562,
      "step": 9268
    },
    {
      "epoch": 0.9162938981291551,
      "grad_norm": 2.5207725685209077,
      "learning_rate": 3.6469210869631616e-09,
      "loss": 0.6925,
      "step": 9269
    },
    {
      "epoch": 0.9163927538739095,
      "grad_norm": 2.7895886588558927,
      "learning_rate": 3.6383572539040007e-09,
      "loss": 0.7789,
      "step": 9270
    },
    {
      "epoch": 0.9164916096186639,
      "grad_norm": 4.180408280119502,
      "learning_rate": 3.629803301309864e-09,
      "loss": 0.6589,
      "step": 9271
    },
    {
      "epoch": 0.9165904653634185,
      "grad_norm": 3.279077684369699,
      "learning_rate": 3.6212592300578605e-09,
      "loss": 0.6125,
      "step": 9272
    },
    {
      "epoch": 0.9166893211081729,
      "grad_norm": 4.348384887735285,
      "learning_rate": 3.6127250410240227e-09,
      "loss": 0.7766,
      "step": 9273
    },
    {
      "epoch": 0.9167881768529274,
      "grad_norm": 7.174700801978917,
      "learning_rate": 3.6042007350834068e-09,
      "loss": 0.8456,
      "step": 9274
    },
    {
      "epoch": 0.9168870325976818,
      "grad_norm": 5.065336606947412,
      "learning_rate": 3.595686313110091e-09,
      "loss": 0.7408,
      "step": 9275
    },
    {
      "epoch": 0.9169858883424363,
      "grad_norm": 7.297817717048018,
      "learning_rate": 3.5871817759770663e-09,
      "loss": 0.6759,
      "step": 9276
    },
    {
      "epoch": 0.9170847440871908,
      "grad_norm": 12.877861501682393,
      "learning_rate": 3.578687124556368e-09,
      "loss": 0.6697,
      "step": 9277
    },
    {
      "epoch": 0.9171835998319452,
      "grad_norm": 3.863183044266501,
      "learning_rate": 3.570202359718988e-09,
      "loss": 0.6834,
      "step": 9278
    },
    {
      "epoch": 0.9172824555766997,
      "grad_norm": 3.297513831922445,
      "learning_rate": 3.5617274823349197e-09,
      "loss": 0.6968,
      "step": 9279
    },
    {
      "epoch": 0.9173813113214542,
      "grad_norm": 5.283775161918684,
      "learning_rate": 3.5532624932731346e-09,
      "loss": 0.762,
      "step": 9280
    },
    {
      "epoch": 0.9174801670662086,
      "grad_norm": 3.8312405740651867,
      "learning_rate": 3.544807393401583e-09,
      "loss": 0.7967,
      "step": 9281
    },
    {
      "epoch": 0.9175790228109632,
      "grad_norm": 3.846415732086695,
      "learning_rate": 3.536362183587227e-09,
      "loss": 0.7381,
      "step": 9282
    },
    {
      "epoch": 0.9176778785557176,
      "grad_norm": 3.142275481552717,
      "learning_rate": 3.527926864695996e-09,
      "loss": 0.7532,
      "step": 9283
    },
    {
      "epoch": 0.917776734300472,
      "grad_norm": 8.977727317151844,
      "learning_rate": 3.519501437592787e-09,
      "loss": 0.7018,
      "step": 9284
    },
    {
      "epoch": 0.9178755900452265,
      "grad_norm": 7.712045329907277,
      "learning_rate": 3.5110859031415195e-09,
      "loss": 0.696,
      "step": 9285
    },
    {
      "epoch": 0.917974445789981,
      "grad_norm": 4.026643944497605,
      "learning_rate": 3.502680262205093e-09,
      "loss": 0.7642,
      "step": 9286
    },
    {
      "epoch": 0.9180733015347354,
      "grad_norm": 5.265587045453004,
      "learning_rate": 3.4942845156453514e-09,
      "loss": 0.6893,
      "step": 9287
    },
    {
      "epoch": 0.9181721572794899,
      "grad_norm": 4.614930117669901,
      "learning_rate": 3.4858986643231837e-09,
      "loss": 0.7295,
      "step": 9288
    },
    {
      "epoch": 0.9182710130242444,
      "grad_norm": 3.5308943032486058,
      "learning_rate": 3.477522709098435e-09,
      "loss": 0.6546,
      "step": 9289
    },
    {
      "epoch": 0.9183698687689988,
      "grad_norm": 4.1031178338106375,
      "learning_rate": 3.4691566508299076e-09,
      "loss": 0.7398,
      "step": 9290
    },
    {
      "epoch": 0.9184687245137533,
      "grad_norm": 7.906841691014189,
      "learning_rate": 3.4608004903754484e-09,
      "loss": 0.6606,
      "step": 9291
    },
    {
      "epoch": 0.9185675802585078,
      "grad_norm": 5.015895906866056,
      "learning_rate": 3.452454228591839e-09,
      "loss": 0.6061,
      "step": 9292
    },
    {
      "epoch": 0.9186664360032623,
      "grad_norm": 4.205587771748279,
      "learning_rate": 3.4441178663348724e-09,
      "loss": 0.7299,
      "step": 9293
    },
    {
      "epoch": 0.9187652917480167,
      "grad_norm": 4.064622495677099,
      "learning_rate": 3.4357914044593206e-09,
      "loss": 0.6653,
      "step": 9294
    },
    {
      "epoch": 0.9188641474927711,
      "grad_norm": 4.1130542331187,
      "learning_rate": 3.4274748438189337e-09,
      "loss": 0.7347,
      "step": 9295
    },
    {
      "epoch": 0.9189630032375257,
      "grad_norm": 6.944756316850829,
      "learning_rate": 3.4191681852664634e-09,
      "loss": 0.6149,
      "step": 9296
    },
    {
      "epoch": 0.9190618589822801,
      "grad_norm": 3.5775546432048198,
      "learning_rate": 3.410871429653628e-09,
      "loss": 0.7713,
      "step": 9297
    },
    {
      "epoch": 0.9191607147270345,
      "grad_norm": 7.149491108958584,
      "learning_rate": 3.402584577831136e-09,
      "loss": 0.7458,
      "step": 9298
    },
    {
      "epoch": 0.9192595704717891,
      "grad_norm": 3.9132493925469567,
      "learning_rate": 3.394307630648685e-09,
      "loss": 0.7426,
      "step": 9299
    },
    {
      "epoch": 0.9193584262165435,
      "grad_norm": 4.485575432264101,
      "learning_rate": 3.3860405889549524e-09,
      "loss": 0.6547,
      "step": 9300
    },
    {
      "epoch": 0.919457281961298,
      "grad_norm": 4.073806245277227,
      "learning_rate": 3.3777834535975935e-09,
      "loss": 0.643,
      "step": 9301
    },
    {
      "epoch": 0.9195561377060525,
      "grad_norm": 6.01288467258024,
      "learning_rate": 3.369536225423264e-09,
      "loss": 0.7406,
      "step": 9302
    },
    {
      "epoch": 0.9196549934508069,
      "grad_norm": 3.403492438776547,
      "learning_rate": 3.3612989052776096e-09,
      "loss": 0.7126,
      "step": 9303
    },
    {
      "epoch": 0.9197538491955614,
      "grad_norm": 10.38213259386531,
      "learning_rate": 3.353071494005222e-09,
      "loss": 0.7638,
      "step": 9304
    },
    {
      "epoch": 0.9198527049403158,
      "grad_norm": 5.153302318722467,
      "learning_rate": 3.3448539924497143e-09,
      "loss": 0.6701,
      "step": 9305
    },
    {
      "epoch": 0.9199515606850703,
      "grad_norm": 4.142047486665847,
      "learning_rate": 3.3366464014536576e-09,
      "loss": 0.6411,
      "step": 9306
    },
    {
      "epoch": 0.9200504164298248,
      "grad_norm": 3.187386041508717,
      "learning_rate": 3.3284487218586232e-09,
      "loss": 0.7027,
      "step": 9307
    },
    {
      "epoch": 0.9201492721745792,
      "grad_norm": 36.98745708881937,
      "learning_rate": 3.320260954505183e-09,
      "loss": 0.7075,
      "step": 9308
    },
    {
      "epoch": 0.9202481279193337,
      "grad_norm": 4.86778079152197,
      "learning_rate": 3.3120831002328432e-09,
      "loss": 0.6838,
      "step": 9309
    },
    {
      "epoch": 0.9203469836640882,
      "grad_norm": 4.233662728046249,
      "learning_rate": 3.303915159880133e-09,
      "loss": 0.7888,
      "step": 9310
    },
    {
      "epoch": 0.9204458394088426,
      "grad_norm": 2.912088899226901,
      "learning_rate": 3.2957571342845714e-09,
      "loss": 0.6928,
      "step": 9311
    },
    {
      "epoch": 0.9205446951535972,
      "grad_norm": 3.459586873641062,
      "learning_rate": 3.2876090242826116e-09,
      "loss": 0.685,
      "step": 9312
    },
    {
      "epoch": 0.9206435508983516,
      "grad_norm": 2.6919002240260395,
      "learning_rate": 3.2794708307097295e-09,
      "loss": 0.7568,
      "step": 9313
    },
    {
      "epoch": 0.920742406643106,
      "grad_norm": 6.695894651811346,
      "learning_rate": 3.2713425544004135e-09,
      "loss": 0.7455,
      "step": 9314
    },
    {
      "epoch": 0.9208412623878605,
      "grad_norm": 3.558689756961042,
      "learning_rate": 3.2632241961880526e-09,
      "loss": 0.6676,
      "step": 9315
    },
    {
      "epoch": 0.920940118132615,
      "grad_norm": 4.113663625824068,
      "learning_rate": 3.2551157569050913e-09,
      "loss": 0.6648,
      "step": 9316
    },
    {
      "epoch": 0.9210389738773694,
      "grad_norm": 4.989929516170186,
      "learning_rate": 3.247017237382921e-09,
      "loss": 0.6709,
      "step": 9317
    },
    {
      "epoch": 0.9211378296221239,
      "grad_norm": 3.4873563938638585,
      "learning_rate": 3.238928638451921e-09,
      "loss": 0.602,
      "step": 9318
    },
    {
      "epoch": 0.9212366853668784,
      "grad_norm": 4.648467174141105,
      "learning_rate": 3.230849960941473e-09,
      "loss": 0.6664,
      "step": 9319
    },
    {
      "epoch": 0.9213355411116328,
      "grad_norm": 3.8844632783056845,
      "learning_rate": 3.222781205679903e-09,
      "loss": 0.7318,
      "step": 9320
    },
    {
      "epoch": 0.9214343968563873,
      "grad_norm": 6.634637809478282,
      "learning_rate": 3.2147223734945492e-09,
      "loss": 0.6645,
      "step": 9321
    },
    {
      "epoch": 0.9215332526011418,
      "grad_norm": 3.580990785175267,
      "learning_rate": 3.2066734652117513e-09,
      "loss": 0.728,
      "step": 9322
    },
    {
      "epoch": 0.9216321083458963,
      "grad_norm": 4.7338847431403925,
      "learning_rate": 3.1986344816567702e-09,
      "loss": 0.7033,
      "step": 9323
    },
    {
      "epoch": 0.9217309640906507,
      "grad_norm": 3.7239724216181918,
      "learning_rate": 3.190605423653914e-09,
      "loss": 0.6349,
      "step": 9324
    },
    {
      "epoch": 0.9218298198354052,
      "grad_norm": 5.966533523860844,
      "learning_rate": 3.1825862920264346e-09,
      "loss": 0.6941,
      "step": 9325
    },
    {
      "epoch": 0.9219286755801597,
      "grad_norm": 3.9106414412488557,
      "learning_rate": 3.174577087596564e-09,
      "loss": 0.7158,
      "step": 9326
    },
    {
      "epoch": 0.9220275313249141,
      "grad_norm": 10.846930399777913,
      "learning_rate": 3.1665778111855558e-09,
      "loss": 0.7787,
      "step": 9327
    },
    {
      "epoch": 0.9221263870696685,
      "grad_norm": 16.080221713526157,
      "learning_rate": 3.1585884636135763e-09,
      "loss": 0.6454,
      "step": 9328
    },
    {
      "epoch": 0.9222252428144231,
      "grad_norm": 18.17945257134747,
      "learning_rate": 3.150609045699848e-09,
      "loss": 0.7735,
      "step": 9329
    },
    {
      "epoch": 0.9223240985591775,
      "grad_norm": 3.043931480291416,
      "learning_rate": 3.142639558262539e-09,
      "loss": 0.8213,
      "step": 9330
    },
    {
      "epoch": 0.922422954303932,
      "grad_norm": 2.9904184356574537,
      "learning_rate": 3.1346800021187836e-09,
      "loss": 0.7201,
      "step": 9331
    },
    {
      "epoch": 0.9225218100486865,
      "grad_norm": 4.199251100845723,
      "learning_rate": 3.126730378084741e-09,
      "loss": 0.6277,
      "step": 9332
    },
    {
      "epoch": 0.9226206657934409,
      "grad_norm": 6.393834087256833,
      "learning_rate": 3.1187906869755143e-09,
      "loss": 0.7687,
      "step": 9333
    },
    {
      "epoch": 0.9227195215381954,
      "grad_norm": 5.01027992516318,
      "learning_rate": 3.110860929605208e-09,
      "loss": 0.709,
      "step": 9334
    },
    {
      "epoch": 0.9228183772829499,
      "grad_norm": 4.306987487303191,
      "learning_rate": 3.1029411067868937e-09,
      "loss": 0.8013,
      "step": 9335
    },
    {
      "epoch": 0.9229172330277043,
      "grad_norm": 3.318404859315426,
      "learning_rate": 3.0950312193326446e-09,
      "loss": 0.6495,
      "step": 9336
    },
    {
      "epoch": 0.9230160887724588,
      "grad_norm": 3.682240879878511,
      "learning_rate": 3.0871312680534777e-09,
      "loss": 0.5861,
      "step": 9337
    },
    {
      "epoch": 0.9231149445172132,
      "grad_norm": 3.7775644234870556,
      "learning_rate": 3.079241253759457e-09,
      "loss": 0.6343,
      "step": 9338
    },
    {
      "epoch": 0.9232138002619678,
      "grad_norm": 4.566987191967971,
      "learning_rate": 3.071361177259546e-09,
      "loss": 0.7209,
      "step": 9339
    },
    {
      "epoch": 0.9233126560067222,
      "grad_norm": 3.9411048211406445,
      "learning_rate": 3.063491039361743e-09,
      "loss": 0.7872,
      "step": 9340
    },
    {
      "epoch": 0.9234115117514766,
      "grad_norm": 3.106358780521815,
      "learning_rate": 3.055630840873036e-09,
      "loss": 0.7232,
      "step": 9341
    },
    {
      "epoch": 0.9235103674962312,
      "grad_norm": 4.540832367721259,
      "learning_rate": 3.0477805825993465e-09,
      "loss": 0.6803,
      "step": 9342
    },
    {
      "epoch": 0.9236092232409856,
      "grad_norm": 3.8892334200676006,
      "learning_rate": 3.0399402653456085e-09,
      "loss": 0.7845,
      "step": 9343
    },
    {
      "epoch": 0.92370807898574,
      "grad_norm": 2.939295721859762,
      "learning_rate": 3.032109889915757e-09,
      "loss": 0.6812,
      "step": 9344
    },
    {
      "epoch": 0.9238069347304946,
      "grad_norm": 3.838101453770588,
      "learning_rate": 3.0242894571126386e-09,
      "loss": 0.7119,
      "step": 9345
    },
    {
      "epoch": 0.923905790475249,
      "grad_norm": 4.554007451484778,
      "learning_rate": 3.0164789677381454e-09,
      "loss": 0.753,
      "step": 9346
    },
    {
      "epoch": 0.9240046462200034,
      "grad_norm": 5.435994435554091,
      "learning_rate": 3.008678422593136e-09,
      "loss": 0.6185,
      "step": 9347
    },
    {
      "epoch": 0.9241035019647579,
      "grad_norm": 4.689923960756403,
      "learning_rate": 3.000887822477416e-09,
      "loss": 0.6833,
      "step": 9348
    },
    {
      "epoch": 0.9242023577095124,
      "grad_norm": 3.2971144935308963,
      "learning_rate": 2.9931071681898346e-09,
      "loss": 0.7211,
      "step": 9349
    },
    {
      "epoch": 0.9243012134542669,
      "grad_norm": 3.664468757470984,
      "learning_rate": 2.9853364605281427e-09,
      "loss": 0.6896,
      "step": 9350
    },
    {
      "epoch": 0.9244000691990213,
      "grad_norm": 4.992906990496116,
      "learning_rate": 2.9775757002891367e-09,
      "loss": 0.7195,
      "step": 9351
    },
    {
      "epoch": 0.9244989249437758,
      "grad_norm": 9.08647283858197,
      "learning_rate": 2.969824888268557e-09,
      "loss": 0.5418,
      "step": 9352
    },
    {
      "epoch": 0.9245977806885303,
      "grad_norm": 21.20081088746383,
      "learning_rate": 2.962084025261147e-09,
      "loss": 0.6779,
      "step": 9353
    },
    {
      "epoch": 0.9246966364332847,
      "grad_norm": 2.8015583577199306,
      "learning_rate": 2.9543531120606035e-09,
      "loss": 0.689,
      "step": 9354
    },
    {
      "epoch": 0.9247954921780392,
      "grad_norm": 3.9555241200022992,
      "learning_rate": 2.946632149459638e-09,
      "loss": 0.6692,
      "step": 9355
    },
    {
      "epoch": 0.9248943479227937,
      "grad_norm": 4.786341344169829,
      "learning_rate": 2.9389211382498837e-09,
      "loss": 0.6918,
      "step": 9356
    },
    {
      "epoch": 0.9249932036675481,
      "grad_norm": 12.028999663525427,
      "learning_rate": 2.9312200792220297e-09,
      "loss": 0.7418,
      "step": 9357
    },
    {
      "epoch": 0.9250920594123025,
      "grad_norm": 4.395978551505031,
      "learning_rate": 2.92352897316569e-09,
      "loss": 0.8512,
      "step": 9358
    },
    {
      "epoch": 0.9251909151570571,
      "grad_norm": 9.657148888111719,
      "learning_rate": 2.915847820869477e-09,
      "loss": 0.742,
      "step": 9359
    },
    {
      "epoch": 0.9252897709018115,
      "grad_norm": 4.035575254472802,
      "learning_rate": 2.9081766231209835e-09,
      "loss": 0.6243,
      "step": 9360
    },
    {
      "epoch": 0.925388626646566,
      "grad_norm": 7.507516496453883,
      "learning_rate": 2.9005153807067583e-09,
      "loss": 0.7906,
      "step": 9361
    },
    {
      "epoch": 0.9254874823913205,
      "grad_norm": 6.68664104615038,
      "learning_rate": 2.892864094412362e-09,
      "loss": 0.8031,
      "step": 9362
    },
    {
      "epoch": 0.9255863381360749,
      "grad_norm": 6.469033671826198,
      "learning_rate": 2.8852227650223325e-09,
      "loss": 0.6852,
      "step": 9363
    },
    {
      "epoch": 0.9256851938808294,
      "grad_norm": 3.6380734967598047,
      "learning_rate": 2.8775913933201557e-09,
      "loss": 0.7174,
      "step": 9364
    },
    {
      "epoch": 0.9257840496255839,
      "grad_norm": 6.380407976344252,
      "learning_rate": 2.869969980088327e-09,
      "loss": 0.6802,
      "step": 9365
    },
    {
      "epoch": 0.9258829053703383,
      "grad_norm": 3.406803227240417,
      "learning_rate": 2.8623585261083215e-09,
      "loss": 0.7045,
      "step": 9366
    },
    {
      "epoch": 0.9259817611150928,
      "grad_norm": 6.78639386395157,
      "learning_rate": 2.8547570321605486e-09,
      "loss": 0.6455,
      "step": 9367
    },
    {
      "epoch": 0.9260806168598472,
      "grad_norm": 11.109920539495475,
      "learning_rate": 2.847165499024462e-09,
      "loss": 0.7462,
      "step": 9368
    },
    {
      "epoch": 0.9261794726046018,
      "grad_norm": 5.3357597227266425,
      "learning_rate": 2.8395839274784393e-09,
      "loss": 0.7119,
      "step": 9369
    },
    {
      "epoch": 0.9262783283493562,
      "grad_norm": 3.997242949195111,
      "learning_rate": 2.8320123182998812e-09,
      "loss": 0.7481,
      "step": 9370
    },
    {
      "epoch": 0.9263771840941106,
      "grad_norm": 8.486579571912982,
      "learning_rate": 2.8244506722651328e-09,
      "loss": 0.7081,
      "step": 9371
    },
    {
      "epoch": 0.9264760398388652,
      "grad_norm": 3.0740515071455365,
      "learning_rate": 2.816898990149508e-09,
      "loss": 0.79,
      "step": 9372
    },
    {
      "epoch": 0.9265748955836196,
      "grad_norm": 3.025121469832653,
      "learning_rate": 2.809357272727342e-09,
      "loss": 0.6885,
      "step": 9373
    },
    {
      "epoch": 0.926673751328374,
      "grad_norm": 8.704721371587047,
      "learning_rate": 2.801825520771939e-09,
      "loss": 0.7993,
      "step": 9374
    },
    {
      "epoch": 0.9267726070731286,
      "grad_norm": 2.879266407771542,
      "learning_rate": 2.7943037350555476e-09,
      "loss": 0.7086,
      "step": 9375
    },
    {
      "epoch": 0.926871462817883,
      "grad_norm": 3.705636804522886,
      "learning_rate": 2.786791916349418e-09,
      "loss": 0.6692,
      "step": 9376
    },
    {
      "epoch": 0.9269703185626375,
      "grad_norm": 5.6892899932397105,
      "learning_rate": 2.7792900654238006e-09,
      "loss": 0.6528,
      "step": 9377
    },
    {
      "epoch": 0.9270691743073919,
      "grad_norm": 7.07633172841689,
      "learning_rate": 2.7717981830478574e-09,
      "loss": 0.7401,
      "step": 9378
    },
    {
      "epoch": 0.9271680300521464,
      "grad_norm": 4.585924105168714,
      "learning_rate": 2.764316269989797e-09,
      "loss": 0.6124,
      "step": 9379
    },
    {
      "epoch": 0.9272668857969009,
      "grad_norm": 5.097629093145771,
      "learning_rate": 2.756844327016794e-09,
      "loss": 0.6841,
      "step": 9380
    },
    {
      "epoch": 0.9273657415416553,
      "grad_norm": 5.296407382228247,
      "learning_rate": 2.749382354894958e-09,
      "loss": 0.6634,
      "step": 9381
    },
    {
      "epoch": 0.9274645972864098,
      "grad_norm": 6.918524000070982,
      "learning_rate": 2.7419303543894103e-09,
      "loss": 0.7557,
      "step": 9382
    },
    {
      "epoch": 0.9275634530311643,
      "grad_norm": 4.796107143404165,
      "learning_rate": 2.7344883262642505e-09,
      "loss": 0.7556,
      "step": 9383
    },
    {
      "epoch": 0.9276623087759187,
      "grad_norm": 7.707362434429936,
      "learning_rate": 2.7270562712825353e-09,
      "loss": 0.7298,
      "step": 9384
    },
    {
      "epoch": 0.9277611645206733,
      "grad_norm": 4.960323823213689,
      "learning_rate": 2.7196341902063434e-09,
      "loss": 0.6123,
      "step": 9385
    },
    {
      "epoch": 0.9278600202654277,
      "grad_norm": 5.596142570356802,
      "learning_rate": 2.712222083796667e-09,
      "loss": 0.7159,
      "step": 9386
    },
    {
      "epoch": 0.9279588760101821,
      "grad_norm": 3.413153455118787,
      "learning_rate": 2.7048199528135198e-09,
      "loss": 0.6095,
      "step": 9387
    },
    {
      "epoch": 0.9280577317549366,
      "grad_norm": 4.661464675753066,
      "learning_rate": 2.6974277980158944e-09,
      "loss": 0.7721,
      "step": 9388
    },
    {
      "epoch": 0.9281565874996911,
      "grad_norm": 6.486070713051215,
      "learning_rate": 2.6900456201617185e-09,
      "loss": 0.6431,
      "step": 9389
    },
    {
      "epoch": 0.9282554432444455,
      "grad_norm": 3.4431390329539657,
      "learning_rate": 2.6826734200079415e-09,
      "loss": 0.7744,
      "step": 9390
    },
    {
      "epoch": 0.9283542989892,
      "grad_norm": 3.326627314887195,
      "learning_rate": 2.6753111983104814e-09,
      "loss": 0.6834,
      "step": 9391
    },
    {
      "epoch": 0.9284531547339545,
      "grad_norm": 4.262504116837925,
      "learning_rate": 2.6679589558242122e-09,
      "loss": 0.7311,
      "step": 9392
    },
    {
      "epoch": 0.928552010478709,
      "grad_norm": 4.600521171275992,
      "learning_rate": 2.6606166933030085e-09,
      "loss": 0.6625,
      "step": 9393
    },
    {
      "epoch": 0.9286508662234634,
      "grad_norm": 5.304312392399939,
      "learning_rate": 2.653284411499701e-09,
      "loss": 0.6693,
      "step": 9394
    },
    {
      "epoch": 0.9287497219682179,
      "grad_norm": 3.245661315694154,
      "learning_rate": 2.6459621111661e-09,
      "loss": 0.6952,
      "step": 9395
    },
    {
      "epoch": 0.9288485777129724,
      "grad_norm": 13.093851123708832,
      "learning_rate": 2.6386497930530383e-09,
      "loss": 0.6293,
      "step": 9396
    },
    {
      "epoch": 0.9289474334577268,
      "grad_norm": 4.271081861875373,
      "learning_rate": 2.631347457910238e-09,
      "loss": 0.7339,
      "step": 9397
    },
    {
      "epoch": 0.9290462892024813,
      "grad_norm": 4.241356741790104,
      "learning_rate": 2.624055106486467e-09,
      "loss": 0.5985,
      "step": 9398
    },
    {
      "epoch": 0.9291451449472358,
      "grad_norm": 3.7755902256872886,
      "learning_rate": 2.6167727395294716e-09,
      "loss": 0.6597,
      "step": 9399
    },
    {
      "epoch": 0.9292440006919902,
      "grad_norm": 15.788884203214172,
      "learning_rate": 2.6095003577859097e-09,
      "loss": 0.6631,
      "step": 9400
    },
    {
      "epoch": 0.9293428564367446,
      "grad_norm": 4.545762585552154,
      "learning_rate": 2.6022379620014747e-09,
      "loss": 0.6923,
      "step": 9401
    },
    {
      "epoch": 0.9294417121814992,
      "grad_norm": 3.231907348676443,
      "learning_rate": 2.5949855529208254e-09,
      "loss": 0.7057,
      "step": 9402
    },
    {
      "epoch": 0.9295405679262536,
      "grad_norm": 5.349823730124508,
      "learning_rate": 2.5877431312875787e-09,
      "loss": 0.7413,
      "step": 9403
    },
    {
      "epoch": 0.929639423671008,
      "grad_norm": 14.676549099703486,
      "learning_rate": 2.5805106978443405e-09,
      "loss": 0.7496,
      "step": 9404
    },
    {
      "epoch": 0.9297382794157626,
      "grad_norm": 6.219080022335714,
      "learning_rate": 2.573288253332706e-09,
      "loss": 0.7594,
      "step": 9405
    },
    {
      "epoch": 0.929837135160517,
      "grad_norm": 4.993296810225713,
      "learning_rate": 2.5660757984931948e-09,
      "loss": 0.7607,
      "step": 9406
    },
    {
      "epoch": 0.9299359909052715,
      "grad_norm": 6.145423765878271,
      "learning_rate": 2.5588733340653814e-09,
      "loss": 0.7466,
      "step": 9407
    },
    {
      "epoch": 0.930034846650026,
      "grad_norm": 3.531956482823787,
      "learning_rate": 2.5516808607877304e-09,
      "loss": 0.7864,
      "step": 9408
    },
    {
      "epoch": 0.9301337023947804,
      "grad_norm": 13.951159133353551,
      "learning_rate": 2.5444983793977417e-09,
      "loss": 0.7719,
      "step": 9409
    },
    {
      "epoch": 0.9302325581395349,
      "grad_norm": 3.210603597837471,
      "learning_rate": 2.5373258906318916e-09,
      "loss": 0.647,
      "step": 9410
    },
    {
      "epoch": 0.9303314138842893,
      "grad_norm": 5.210853408432664,
      "learning_rate": 2.530163395225582e-09,
      "loss": 0.7094,
      "step": 9411
    },
    {
      "epoch": 0.9304302696290438,
      "grad_norm": 4.279214675589479,
      "learning_rate": 2.5230108939132355e-09,
      "loss": 0.5599,
      "step": 9412
    },
    {
      "epoch": 0.9305291253737983,
      "grad_norm": 3.456512921085227,
      "learning_rate": 2.5158683874282438e-09,
      "loss": 0.676,
      "step": 9413
    },
    {
      "epoch": 0.9306279811185527,
      "grad_norm": 13.73664052500378,
      "learning_rate": 2.508735876502954e-09,
      "loss": 0.7074,
      "step": 9414
    },
    {
      "epoch": 0.9307268368633073,
      "grad_norm": 10.144969870262862,
      "learning_rate": 2.501613361868704e-09,
      "loss": 0.688,
      "step": 9415
    },
    {
      "epoch": 0.9308256926080617,
      "grad_norm": 3.5550203266616207,
      "learning_rate": 2.494500844255798e-09,
      "loss": 0.691,
      "step": 9416
    },
    {
      "epoch": 0.9309245483528161,
      "grad_norm": 5.279824921322023,
      "learning_rate": 2.4873983243935192e-09,
      "loss": 0.7186,
      "step": 9417
    },
    {
      "epoch": 0.9310234040975707,
      "grad_norm": 4.17567267809548,
      "learning_rate": 2.480305803010141e-09,
      "loss": 0.8118,
      "step": 9418
    },
    {
      "epoch": 0.9311222598423251,
      "grad_norm": 3.5491342312346714,
      "learning_rate": 2.47322328083287e-09,
      "loss": 0.7583,
      "step": 9419
    },
    {
      "epoch": 0.9312211155870795,
      "grad_norm": 3.154140956805501,
      "learning_rate": 2.4661507585879372e-09,
      "loss": 0.8024,
      "step": 9420
    },
    {
      "epoch": 0.931319971331834,
      "grad_norm": 4.680718601170543,
      "learning_rate": 2.459088237000528e-09,
      "loss": 0.6817,
      "step": 9421
    },
    {
      "epoch": 0.9314188270765885,
      "grad_norm": 5.34507780494192,
      "learning_rate": 2.452035716794787e-09,
      "loss": 0.8004,
      "step": 9422
    },
    {
      "epoch": 0.931517682821343,
      "grad_norm": 5.795434304280505,
      "learning_rate": 2.4449931986938455e-09,
      "loss": 0.6597,
      "step": 9423
    },
    {
      "epoch": 0.9316165385660974,
      "grad_norm": 3.0892500217355683,
      "learning_rate": 2.437960683419815e-09,
      "loss": 0.6936,
      "step": 9424
    },
    {
      "epoch": 0.9317153943108519,
      "grad_norm": 4.2714302165973095,
      "learning_rate": 2.430938171693775e-09,
      "loss": 0.7193,
      "step": 9425
    },
    {
      "epoch": 0.9318142500556064,
      "grad_norm": 5.4945809116354445,
      "learning_rate": 2.423925664235782e-09,
      "loss": 0.795,
      "step": 9426
    },
    {
      "epoch": 0.9319131058003608,
      "grad_norm": 3.860428218279135,
      "learning_rate": 2.416923161764861e-09,
      "loss": 0.6972,
      "step": 9427
    },
    {
      "epoch": 0.9320119615451153,
      "grad_norm": 3.238415999273486,
      "learning_rate": 2.409930664999016e-09,
      "loss": 0.7454,
      "step": 9428
    },
    {
      "epoch": 0.9321108172898698,
      "grad_norm": 3.2989982802315074,
      "learning_rate": 2.4029481746552283e-09,
      "loss": 0.7824,
      "step": 9429
    },
    {
      "epoch": 0.9322096730346242,
      "grad_norm": 16.82300292630578,
      "learning_rate": 2.3959756914494478e-09,
      "loss": 0.706,
      "step": 9430
    },
    {
      "epoch": 0.9323085287793786,
      "grad_norm": 5.19254925230508,
      "learning_rate": 2.3890132160966026e-09,
      "loss": 0.7129,
      "step": 9431
    },
    {
      "epoch": 0.9324073845241332,
      "grad_norm": 4.059691334429038,
      "learning_rate": 2.382060749310588e-09,
      "loss": 0.6811,
      "step": 9432
    },
    {
      "epoch": 0.9325062402688876,
      "grad_norm": 4.615312574408365,
      "learning_rate": 2.375118291804268e-09,
      "loss": 0.7357,
      "step": 9433
    },
    {
      "epoch": 0.932605096013642,
      "grad_norm": 4.445173247207517,
      "learning_rate": 2.3681858442895055e-09,
      "loss": 0.6976,
      "step": 9434
    },
    {
      "epoch": 0.9327039517583966,
      "grad_norm": 3.9776128280467233,
      "learning_rate": 2.361263407477121e-09,
      "loss": 0.6747,
      "step": 9435
    },
    {
      "epoch": 0.932802807503151,
      "grad_norm": 22.62467826351968,
      "learning_rate": 2.3543509820768914e-09,
      "loss": 0.686,
      "step": 9436
    },
    {
      "epoch": 0.9329016632479055,
      "grad_norm": 11.893786216687706,
      "learning_rate": 2.347448568797605e-09,
      "loss": 0.655,
      "step": 9437
    },
    {
      "epoch": 0.93300051899266,
      "grad_norm": 4.768994886656905,
      "learning_rate": 2.3405561683469856e-09,
      "loss": 0.6388,
      "step": 9438
    },
    {
      "epoch": 0.9330993747374144,
      "grad_norm": 4.447149780193799,
      "learning_rate": 2.3336737814317555e-09,
      "loss": 0.706,
      "step": 9439
    },
    {
      "epoch": 0.9331982304821689,
      "grad_norm": 4.381167304650915,
      "learning_rate": 2.3268014087575947e-09,
      "loss": 0.707,
      "step": 9440
    },
    {
      "epoch": 0.9332970862269233,
      "grad_norm": 3.9319400823924786,
      "learning_rate": 2.319939051029174e-09,
      "loss": 0.6648,
      "step": 9441
    },
    {
      "epoch": 0.9333959419716779,
      "grad_norm": 3.061105258824254,
      "learning_rate": 2.313086708950107e-09,
      "loss": 0.6602,
      "step": 9442
    },
    {
      "epoch": 0.9334947977164323,
      "grad_norm": 4.094735733124143,
      "learning_rate": 2.306244383223033e-09,
      "loss": 0.7352,
      "step": 9443
    },
    {
      "epoch": 0.9335936534611867,
      "grad_norm": 4.390781015786918,
      "learning_rate": 2.2994120745495003e-09,
      "loss": 0.7472,
      "step": 9444
    },
    {
      "epoch": 0.9336925092059413,
      "grad_norm": 3.6443007264815517,
      "learning_rate": 2.2925897836300723e-09,
      "loss": 0.6338,
      "step": 9445
    },
    {
      "epoch": 0.9337913649506957,
      "grad_norm": 11.324004293916774,
      "learning_rate": 2.2857775111642884e-09,
      "loss": 0.6953,
      "step": 9446
    },
    {
      "epoch": 0.9338902206954501,
      "grad_norm": 5.210198916517,
      "learning_rate": 2.278975257850624e-09,
      "loss": 0.6849,
      "step": 9447
    },
    {
      "epoch": 0.9339890764402047,
      "grad_norm": 2.859410714836566,
      "learning_rate": 2.2721830243865757e-09,
      "loss": 0.7012,
      "step": 9448
    },
    {
      "epoch": 0.9340879321849591,
      "grad_norm": 3.906393462896743,
      "learning_rate": 2.2654008114685543e-09,
      "loss": 0.68,
      "step": 9449
    },
    {
      "epoch": 0.9341867879297135,
      "grad_norm": 9.746065070084937,
      "learning_rate": 2.2586286197920023e-09,
      "loss": 0.64,
      "step": 9450
    },
    {
      "epoch": 0.934285643674468,
      "grad_norm": 2.9001646340466585,
      "learning_rate": 2.2518664500512985e-09,
      "loss": 0.6532,
      "step": 9451
    },
    {
      "epoch": 0.9343844994192225,
      "grad_norm": 6.082909802346714,
      "learning_rate": 2.245114302939799e-09,
      "loss": 0.7301,
      "step": 9452
    },
    {
      "epoch": 0.934483355163977,
      "grad_norm": 3.8992260906642433,
      "learning_rate": 2.238372179149839e-09,
      "loss": 0.7729,
      "step": 9453
    },
    {
      "epoch": 0.9345822109087314,
      "grad_norm": 5.362699412546064,
      "learning_rate": 2.231640079372732e-09,
      "loss": 0.7257,
      "step": 9454
    },
    {
      "epoch": 0.9346810666534859,
      "grad_norm": 4.344396929479076,
      "learning_rate": 2.224918004298748e-09,
      "loss": 0.693,
      "step": 9455
    },
    {
      "epoch": 0.9347799223982404,
      "grad_norm": 3.9623458891185765,
      "learning_rate": 2.218205954617147e-09,
      "loss": 0.6778,
      "step": 9456
    },
    {
      "epoch": 0.9348787781429948,
      "grad_norm": 7.584680015343473,
      "learning_rate": 2.2115039310161343e-09,
      "loss": 0.7909,
      "step": 9457
    },
    {
      "epoch": 0.9349776338877493,
      "grad_norm": 4.915494994874523,
      "learning_rate": 2.204811934182915e-09,
      "loss": 0.6907,
      "step": 9458
    },
    {
      "epoch": 0.9350764896325038,
      "grad_norm": 8.704286000576328,
      "learning_rate": 2.198129964803652e-09,
      "loss": 0.6431,
      "step": 9459
    },
    {
      "epoch": 0.9351753453772582,
      "grad_norm": 4.6482714561707965,
      "learning_rate": 2.191458023563486e-09,
      "loss": 0.7197,
      "step": 9460
    },
    {
      "epoch": 0.9352742011220126,
      "grad_norm": 3.32249016507946,
      "learning_rate": 2.184796111146503e-09,
      "loss": 0.6231,
      "step": 9461
    },
    {
      "epoch": 0.9353730568667672,
      "grad_norm": 5.774286974729944,
      "learning_rate": 2.1781442282358232e-09,
      "loss": 0.7401,
      "step": 9462
    },
    {
      "epoch": 0.9354719126115216,
      "grad_norm": 3.8058705511329145,
      "learning_rate": 2.1715023755134675e-09,
      "loss": 0.7252,
      "step": 9463
    },
    {
      "epoch": 0.9355707683562761,
      "grad_norm": 5.637458940336325,
      "learning_rate": 2.1648705536604694e-09,
      "loss": 0.6851,
      "step": 9464
    },
    {
      "epoch": 0.9356696241010306,
      "grad_norm": 3.518999157990596,
      "learning_rate": 2.1582487633568403e-09,
      "loss": 0.7118,
      "step": 9465
    },
    {
      "epoch": 0.935768479845785,
      "grad_norm": 8.463993457413432,
      "learning_rate": 2.151637005281526e-09,
      "loss": 0.7213,
      "step": 9466
    },
    {
      "epoch": 0.9358673355905395,
      "grad_norm": 5.340538822994558,
      "learning_rate": 2.1450352801124618e-09,
      "loss": 0.7097,
      "step": 9467
    },
    {
      "epoch": 0.935966191335294,
      "grad_norm": 7.484612627412143,
      "learning_rate": 2.138443588526584e-09,
      "loss": 0.7609,
      "step": 9468
    },
    {
      "epoch": 0.9360650470800485,
      "grad_norm": 5.200703103508244,
      "learning_rate": 2.131861931199741e-09,
      "loss": 0.5201,
      "step": 9469
    },
    {
      "epoch": 0.9361639028248029,
      "grad_norm": 4.491616172682409,
      "learning_rate": 2.1252903088068153e-09,
      "loss": 0.6586,
      "step": 9470
    },
    {
      "epoch": 0.9362627585695573,
      "grad_norm": 3.5065925640953335,
      "learning_rate": 2.1187287220216123e-09,
      "loss": 0.6327,
      "step": 9471
    },
    {
      "epoch": 0.9363616143143119,
      "grad_norm": 4.803224605484602,
      "learning_rate": 2.1121771715169157e-09,
      "loss": 0.7321,
      "step": 9472
    },
    {
      "epoch": 0.9364604700590663,
      "grad_norm": 3.0880079005376926,
      "learning_rate": 2.105635657964522e-09,
      "loss": 0.7694,
      "step": 9473
    },
    {
      "epoch": 0.9365593258038207,
      "grad_norm": 47.38564803886269,
      "learning_rate": 2.0991041820351273e-09,
      "loss": 0.7044,
      "step": 9474
    },
    {
      "epoch": 0.9366581815485753,
      "grad_norm": 7.223594611681838,
      "learning_rate": 2.0925827443984746e-09,
      "loss": 0.7684,
      "step": 9475
    },
    {
      "epoch": 0.9367570372933297,
      "grad_norm": 4.30390666061517,
      "learning_rate": 2.086071345723228e-09,
      "loss": 0.7593,
      "step": 9476
    },
    {
      "epoch": 0.9368558930380841,
      "grad_norm": 4.937063302113026,
      "learning_rate": 2.079569986677021e-09,
      "loss": 0.7374,
      "step": 9477
    },
    {
      "epoch": 0.9369547487828387,
      "grad_norm": 3.3452245412954342,
      "learning_rate": 2.0730786679264866e-09,
      "loss": 0.6575,
      "step": 9478
    },
    {
      "epoch": 0.9370536045275931,
      "grad_norm": 4.054223306944417,
      "learning_rate": 2.066597390137215e-09,
      "loss": 0.7345,
      "step": 9479
    },
    {
      "epoch": 0.9371524602723476,
      "grad_norm": 8.150040826733962,
      "learning_rate": 2.0601261539737513e-09,
      "loss": 0.6282,
      "step": 9480
    },
    {
      "epoch": 0.9372513160171021,
      "grad_norm": 3.4134266091742,
      "learning_rate": 2.053664960099655e-09,
      "loss": 0.7141,
      "step": 9481
    },
    {
      "epoch": 0.9373501717618565,
      "grad_norm": 3.344018582080203,
      "learning_rate": 2.047213809177384e-09,
      "loss": 0.6747,
      "step": 9482
    },
    {
      "epoch": 0.937449027506611,
      "grad_norm": 6.082404859024422,
      "learning_rate": 2.040772701868443e-09,
      "loss": 0.6916,
      "step": 9483
    },
    {
      "epoch": 0.9375478832513654,
      "grad_norm": 5.579251691797863,
      "learning_rate": 2.03434163883327e-09,
      "loss": 0.6345,
      "step": 9484
    },
    {
      "epoch": 0.9376467389961199,
      "grad_norm": 5.099846693445276,
      "learning_rate": 2.027920620731249e-09,
      "loss": 0.7108,
      "step": 9485
    },
    {
      "epoch": 0.9377455947408744,
      "grad_norm": 6.8962614783778635,
      "learning_rate": 2.0215096482207873e-09,
      "loss": 0.7121,
      "step": 9486
    },
    {
      "epoch": 0.9378444504856288,
      "grad_norm": 3.0314534851749824,
      "learning_rate": 2.0151087219592354e-09,
      "loss": 0.6382,
      "step": 9487
    },
    {
      "epoch": 0.9379433062303834,
      "grad_norm": 11.116323960196492,
      "learning_rate": 2.0087178426028917e-09,
      "loss": 0.7046,
      "step": 9488
    },
    {
      "epoch": 0.9380421619751378,
      "grad_norm": 6.770631732536594,
      "learning_rate": 2.0023370108070537e-09,
      "loss": 0.6302,
      "step": 9489
    },
    {
      "epoch": 0.9381410177198922,
      "grad_norm": 3.7757503215476764,
      "learning_rate": 1.9959662272260093e-09,
      "loss": 0.6815,
      "step": 9490
    },
    {
      "epoch": 0.9382398734646468,
      "grad_norm": 5.637119343984888,
      "learning_rate": 1.989605492512958e-09,
      "loss": 0.6814,
      "step": 9491
    },
    {
      "epoch": 0.9383387292094012,
      "grad_norm": 5.000635018735602,
      "learning_rate": 1.9832548073201117e-09,
      "loss": 0.7011,
      "step": 9492
    },
    {
      "epoch": 0.9384375849541556,
      "grad_norm": 7.513236565366982,
      "learning_rate": 1.9769141722986383e-09,
      "loss": 0.6583,
      "step": 9493
    },
    {
      "epoch": 0.9385364406989101,
      "grad_norm": 3.3954197170877314,
      "learning_rate": 1.970583588098662e-09,
      "loss": 0.6947,
      "step": 9494
    },
    {
      "epoch": 0.9386352964436646,
      "grad_norm": 4.261247997682572,
      "learning_rate": 1.9642630553693085e-09,
      "loss": 0.6283,
      "step": 9495
    },
    {
      "epoch": 0.938734152188419,
      "grad_norm": 6.607808117113459,
      "learning_rate": 1.957952574758648e-09,
      "loss": 0.8332,
      "step": 9496
    },
    {
      "epoch": 0.9388330079331735,
      "grad_norm": 3.3222110248573795,
      "learning_rate": 1.9516521469137294e-09,
      "loss": 0.6823,
      "step": 9497
    },
    {
      "epoch": 0.938931863677928,
      "grad_norm": 3.303349851751872,
      "learning_rate": 1.9453617724805805e-09,
      "loss": 0.7286,
      "step": 9498
    },
    {
      "epoch": 0.9390307194226825,
      "grad_norm": 3.442410999705263,
      "learning_rate": 1.939081452104152e-09,
      "loss": 0.6639,
      "step": 9499
    },
    {
      "epoch": 0.9391295751674369,
      "grad_norm": 3.258662389114907,
      "learning_rate": 1.932811186428429e-09,
      "loss": 0.6113,
      "step": 9500
    },
    {
      "epoch": 0.9392284309121914,
      "grad_norm": 4.453182285188058,
      "learning_rate": 1.9265509760963415e-09,
      "loss": 0.7437,
      "step": 9501
    },
    {
      "epoch": 0.9393272866569459,
      "grad_norm": 3.2506802505446637,
      "learning_rate": 1.920300821749743e-09,
      "loss": 0.697,
      "step": 9502
    },
    {
      "epoch": 0.9394261424017003,
      "grad_norm": 4.884906440355357,
      "learning_rate": 1.9140607240295427e-09,
      "loss": 0.8005,
      "step": 9503
    },
    {
      "epoch": 0.9395249981464547,
      "grad_norm": 9.32451080478913,
      "learning_rate": 1.907830683575529e-09,
      "loss": 0.7036,
      "step": 9504
    },
    {
      "epoch": 0.9396238538912093,
      "grad_norm": 5.610917000034162,
      "learning_rate": 1.9016107010265124e-09,
      "loss": 0.6402,
      "step": 9505
    },
    {
      "epoch": 0.9397227096359637,
      "grad_norm": 5.61151799539278,
      "learning_rate": 1.895400777020273e-09,
      "loss": 0.7526,
      "step": 9506
    },
    {
      "epoch": 0.9398215653807181,
      "grad_norm": 4.2011025372962925,
      "learning_rate": 1.889200912193545e-09,
      "loss": 0.602,
      "step": 9507
    },
    {
      "epoch": 0.9399204211254727,
      "grad_norm": 4.505273502431266,
      "learning_rate": 1.8830111071820085e-09,
      "loss": 0.8132,
      "step": 9508
    },
    {
      "epoch": 0.9400192768702271,
      "grad_norm": 4.485959205093944,
      "learning_rate": 1.876831362620379e-09,
      "loss": 0.6671,
      "step": 9509
    },
    {
      "epoch": 0.9401181326149816,
      "grad_norm": 3.7240502530218462,
      "learning_rate": 1.8706616791422715e-09,
      "loss": 0.6899,
      "step": 9510
    },
    {
      "epoch": 0.9402169883597361,
      "grad_norm": 3.7184910265387696,
      "learning_rate": 1.8645020573802905e-09,
      "loss": 0.5886,
      "step": 9511
    },
    {
      "epoch": 0.9403158441044905,
      "grad_norm": 4.818279616103093,
      "learning_rate": 1.8583524979660314e-09,
      "loss": 0.7816,
      "step": 9512
    },
    {
      "epoch": 0.940414699849245,
      "grad_norm": 4.442129241017797,
      "learning_rate": 1.8522130015300119e-09,
      "loss": 0.8389,
      "step": 9513
    },
    {
      "epoch": 0.9405135555939994,
      "grad_norm": 6.303731532760152,
      "learning_rate": 1.8460835687017951e-09,
      "loss": 0.6037,
      "step": 9514
    },
    {
      "epoch": 0.940612411338754,
      "grad_norm": 7.707398075323764,
      "learning_rate": 1.8399642001098116e-09,
      "loss": 0.7242,
      "step": 9515
    },
    {
      "epoch": 0.9407112670835084,
      "grad_norm": 4.490662999858469,
      "learning_rate": 1.8338548963815482e-09,
      "loss": 0.6895,
      "step": 9516
    },
    {
      "epoch": 0.9408101228282628,
      "grad_norm": 9.035661648967672,
      "learning_rate": 1.827755658143415e-09,
      "loss": 0.7628,
      "step": 9517
    },
    {
      "epoch": 0.9409089785730174,
      "grad_norm": 3.324822260282589,
      "learning_rate": 1.8216664860207898e-09,
      "loss": 0.7586,
      "step": 9518
    },
    {
      "epoch": 0.9410078343177718,
      "grad_norm": 2.883821582969836,
      "learning_rate": 1.8155873806380396e-09,
      "loss": 0.7082,
      "step": 9519
    },
    {
      "epoch": 0.9411066900625262,
      "grad_norm": 3.2885143907397887,
      "learning_rate": 1.8095183426184878e-09,
      "loss": 0.7967,
      "step": 9520
    },
    {
      "epoch": 0.9412055458072808,
      "grad_norm": 4.016651746323613,
      "learning_rate": 1.8034593725844038e-09,
      "loss": 0.761,
      "step": 9521
    },
    {
      "epoch": 0.9413044015520352,
      "grad_norm": 37.94754153041741,
      "learning_rate": 1.797410471157057e-09,
      "loss": 0.7407,
      "step": 9522
    },
    {
      "epoch": 0.9414032572967896,
      "grad_norm": 4.781137485705265,
      "learning_rate": 1.7913716389566958e-09,
      "loss": 0.702,
      "step": 9523
    },
    {
      "epoch": 0.9415021130415441,
      "grad_norm": 3.675469303494617,
      "learning_rate": 1.7853428766024803e-09,
      "loss": 0.7073,
      "step": 9524
    },
    {
      "epoch": 0.9416009687862986,
      "grad_norm": 4.928609094733691,
      "learning_rate": 1.779324184712594e-09,
      "loss": 0.5802,
      "step": 9525
    },
    {
      "epoch": 0.941699824531053,
      "grad_norm": 3.996639579107412,
      "learning_rate": 1.773315563904143e-09,
      "loss": 0.8325,
      "step": 9526
    },
    {
      "epoch": 0.9417986802758075,
      "grad_norm": 3.637963527474457,
      "learning_rate": 1.767317014793257e-09,
      "loss": 0.7845,
      "step": 9527
    },
    {
      "epoch": 0.941897536020562,
      "grad_norm": 4.180321532776557,
      "learning_rate": 1.761328537994955e-09,
      "loss": 0.7203,
      "step": 9528
    },
    {
      "epoch": 0.9419963917653165,
      "grad_norm": 3.4498714140904005,
      "learning_rate": 1.7553501341233013e-09,
      "loss": 0.7422,
      "step": 9529
    },
    {
      "epoch": 0.9420952475100709,
      "grad_norm": 3.1662044692711513,
      "learning_rate": 1.7493818037912723e-09,
      "loss": 0.6721,
      "step": 9530
    },
    {
      "epoch": 0.9421941032548254,
      "grad_norm": 4.096140524567607,
      "learning_rate": 1.743423547610845e-09,
      "loss": 0.6697,
      "step": 9531
    },
    {
      "epoch": 0.9422929589995799,
      "grad_norm": 5.641133485488107,
      "learning_rate": 1.737475366192953e-09,
      "loss": 0.6811,
      "step": 9532
    },
    {
      "epoch": 0.9423918147443343,
      "grad_norm": 3.2898073327490946,
      "learning_rate": 1.7315372601474753e-09,
      "loss": 0.6675,
      "step": 9533
    },
    {
      "epoch": 0.9424906704890887,
      "grad_norm": 3.546898110060117,
      "learning_rate": 1.7256092300833025e-09,
      "loss": 0.5832,
      "step": 9534
    },
    {
      "epoch": 0.9425895262338433,
      "grad_norm": 10.202019965684796,
      "learning_rate": 1.7196912766082373e-09,
      "loss": 0.6834,
      "step": 9535
    },
    {
      "epoch": 0.9426883819785977,
      "grad_norm": 3.4272911735767244,
      "learning_rate": 1.7137834003291162e-09,
      "loss": 0.6664,
      "step": 9536
    },
    {
      "epoch": 0.9427872377233522,
      "grad_norm": 2.7459119684544544,
      "learning_rate": 1.7078856018516664e-09,
      "loss": 0.7507,
      "step": 9537
    },
    {
      "epoch": 0.9428860934681067,
      "grad_norm": 3.8630611789509635,
      "learning_rate": 1.701997881780637e-09,
      "loss": 0.7495,
      "step": 9538
    },
    {
      "epoch": 0.9429849492128611,
      "grad_norm": 5.384312523499959,
      "learning_rate": 1.696120240719734e-09,
      "loss": 0.7234,
      "step": 9539
    },
    {
      "epoch": 0.9430838049576156,
      "grad_norm": 2.889369447358657,
      "learning_rate": 1.6902526792716199e-09,
      "loss": 0.6785,
      "step": 9540
    },
    {
      "epoch": 0.9431826607023701,
      "grad_norm": 4.834195663686083,
      "learning_rate": 1.6843951980379133e-09,
      "loss": 0.6519,
      "step": 9541
    },
    {
      "epoch": 0.9432815164471245,
      "grad_norm": 4.055440550535125,
      "learning_rate": 1.678547797619234e-09,
      "loss": 0.6583,
      "step": 9542
    },
    {
      "epoch": 0.943380372191879,
      "grad_norm": 4.554359241228796,
      "learning_rate": 1.6727104786151247e-09,
      "loss": 0.7572,
      "step": 9543
    },
    {
      "epoch": 0.9434792279366334,
      "grad_norm": 4.543998028730801,
      "learning_rate": 1.6668832416241285e-09,
      "loss": 0.6764,
      "step": 9544
    },
    {
      "epoch": 0.943578083681388,
      "grad_norm": 6.375545131358971,
      "learning_rate": 1.6610660872437454e-09,
      "loss": 0.732,
      "step": 9545
    },
    {
      "epoch": 0.9436769394261424,
      "grad_norm": 4.5446417771247205,
      "learning_rate": 1.6552590160704317e-09,
      "loss": 0.6697,
      "step": 9546
    },
    {
      "epoch": 0.9437757951708968,
      "grad_norm": 2.8446867103316693,
      "learning_rate": 1.649462028699622e-09,
      "loss": 0.7065,
      "step": 9547
    },
    {
      "epoch": 0.9438746509156514,
      "grad_norm": 14.12016688933478,
      "learning_rate": 1.6436751257256965e-09,
      "loss": 0.7156,
      "step": 9548
    },
    {
      "epoch": 0.9439735066604058,
      "grad_norm": 4.9885228255671175,
      "learning_rate": 1.637898307742036e-09,
      "loss": 0.7318,
      "step": 9549
    },
    {
      "epoch": 0.9440723624051602,
      "grad_norm": 4.921791159261515,
      "learning_rate": 1.632131575340967e-09,
      "loss": 0.6344,
      "step": 9550
    },
    {
      "epoch": 0.9441712181499148,
      "grad_norm": 5.358949474243263,
      "learning_rate": 1.6263749291137606e-09,
      "loss": 0.7132,
      "step": 9551
    },
    {
      "epoch": 0.9442700738946692,
      "grad_norm": 10.314000415825845,
      "learning_rate": 1.6206283696507006e-09,
      "loss": 0.7748,
      "step": 9552
    },
    {
      "epoch": 0.9443689296394236,
      "grad_norm": 2.9639729766388063,
      "learning_rate": 1.6148918975410154e-09,
      "loss": 0.6851,
      "step": 9553
    },
    {
      "epoch": 0.9444677853841782,
      "grad_norm": 13.78915318982032,
      "learning_rate": 1.6091655133728676e-09,
      "loss": 0.7048,
      "step": 9554
    },
    {
      "epoch": 0.9445666411289326,
      "grad_norm": 3.1420632635632955,
      "learning_rate": 1.6034492177334324e-09,
      "loss": 0.5917,
      "step": 9555
    },
    {
      "epoch": 0.9446654968736871,
      "grad_norm": 4.228712703902489,
      "learning_rate": 1.5977430112088296e-09,
      "loss": 0.7274,
      "step": 9556
    },
    {
      "epoch": 0.9447643526184415,
      "grad_norm": 3.6848647723822903,
      "learning_rate": 1.5920468943841469e-09,
      "loss": 0.6859,
      "step": 9557
    },
    {
      "epoch": 0.944863208363196,
      "grad_norm": 2.8463442592894563,
      "learning_rate": 1.5863608678434393e-09,
      "loss": 0.6775,
      "step": 9558
    },
    {
      "epoch": 0.9449620641079505,
      "grad_norm": 14.16641554404359,
      "learning_rate": 1.580684932169718e-09,
      "loss": 0.7362,
      "step": 9559
    },
    {
      "epoch": 0.9450609198527049,
      "grad_norm": 5.7874088781158415,
      "learning_rate": 1.575019087944962e-09,
      "loss": 0.7285,
      "step": 9560
    },
    {
      "epoch": 0.9451597755974595,
      "grad_norm": 5.8754845106660945,
      "learning_rate": 1.56936333575014e-09,
      "loss": 0.6576,
      "step": 9561
    },
    {
      "epoch": 0.9452586313422139,
      "grad_norm": 4.368819259455327,
      "learning_rate": 1.563717676165155e-09,
      "loss": 0.667,
      "step": 9562
    },
    {
      "epoch": 0.9453574870869683,
      "grad_norm": 5.533860586705769,
      "learning_rate": 1.5580821097688768e-09,
      "loss": 0.6598,
      "step": 9563
    },
    {
      "epoch": 0.9454563428317229,
      "grad_norm": 5.076910941421249,
      "learning_rate": 1.5524566371391656e-09,
      "loss": 0.6482,
      "step": 9564
    },
    {
      "epoch": 0.9455551985764773,
      "grad_norm": 3.461274933528676,
      "learning_rate": 1.5468412588528157e-09,
      "loss": 0.7712,
      "step": 9565
    },
    {
      "epoch": 0.9456540543212317,
      "grad_norm": 3.2905133723351443,
      "learning_rate": 1.5412359754856108e-09,
      "loss": 0.7098,
      "step": 9566
    },
    {
      "epoch": 0.9457529100659862,
      "grad_norm": 11.135119342609995,
      "learning_rate": 1.5356407876122911e-09,
      "loss": 0.7017,
      "step": 9567
    },
    {
      "epoch": 0.9458517658107407,
      "grad_norm": 4.6191519086631985,
      "learning_rate": 1.5300556958065536e-09,
      "loss": 0.7163,
      "step": 9568
    },
    {
      "epoch": 0.9459506215554951,
      "grad_norm": 4.370946682008554,
      "learning_rate": 1.5244807006410732e-09,
      "loss": 0.675,
      "step": 9569
    },
    {
      "epoch": 0.9460494773002496,
      "grad_norm": 4.436988320092492,
      "learning_rate": 1.5189158026874704e-09,
      "loss": 0.7369,
      "step": 9570
    },
    {
      "epoch": 0.9461483330450041,
      "grad_norm": 18.427922074618525,
      "learning_rate": 1.5133610025163557e-09,
      "loss": 0.7057,
      "step": 9571
    },
    {
      "epoch": 0.9462471887897586,
      "grad_norm": 6.36627222240923,
      "learning_rate": 1.5078163006973066e-09,
      "loss": 0.6836,
      "step": 9572
    },
    {
      "epoch": 0.946346044534513,
      "grad_norm": 10.70946063942642,
      "learning_rate": 1.502281697798824e-09,
      "loss": 0.6454,
      "step": 9573
    },
    {
      "epoch": 0.9464449002792675,
      "grad_norm": 7.818420657876119,
      "learning_rate": 1.4967571943883983e-09,
      "loss": 0.8043,
      "step": 9574
    },
    {
      "epoch": 0.946543756024022,
      "grad_norm": 3.406175339000333,
      "learning_rate": 1.4912427910325209e-09,
      "loss": 0.7003,
      "step": 9575
    },
    {
      "epoch": 0.9466426117687764,
      "grad_norm": 6.243815321159405,
      "learning_rate": 1.4857384882965729e-09,
      "loss": 0.7077,
      "step": 9576
    },
    {
      "epoch": 0.9467414675135308,
      "grad_norm": 8.738423005550182,
      "learning_rate": 1.4802442867449472e-09,
      "loss": 0.5686,
      "step": 9577
    },
    {
      "epoch": 0.9468403232582854,
      "grad_norm": 9.123164796795818,
      "learning_rate": 1.4747601869410154e-09,
      "loss": 0.7058,
      "step": 9578
    },
    {
      "epoch": 0.9469391790030398,
      "grad_norm": 5.809920538056159,
      "learning_rate": 1.4692861894470721e-09,
      "loss": 0.7757,
      "step": 9579
    },
    {
      "epoch": 0.9470380347477942,
      "grad_norm": 7.965921667950453,
      "learning_rate": 1.4638222948244127e-09,
      "loss": 0.6984,
      "step": 9580
    },
    {
      "epoch": 0.9471368904925488,
      "grad_norm": 13.969528876775147,
      "learning_rate": 1.458368503633256e-09,
      "loss": 0.781,
      "step": 9581
    },
    {
      "epoch": 0.9472357462373032,
      "grad_norm": 4.957091867756254,
      "learning_rate": 1.45292481643281e-09,
      "loss": 0.644,
      "step": 9582
    },
    {
      "epoch": 0.9473346019820577,
      "grad_norm": 3.4013317612039975,
      "learning_rate": 1.4474912337812616e-09,
      "loss": 0.6661,
      "step": 9583
    },
    {
      "epoch": 0.9474334577268122,
      "grad_norm": 6.493526394811664,
      "learning_rate": 1.442067756235732e-09,
      "loss": 0.6709,
      "step": 9584
    },
    {
      "epoch": 0.9475323134715666,
      "grad_norm": 2.9247587221022617,
      "learning_rate": 1.4366543843523205e-09,
      "loss": 0.7275,
      "step": 9585
    },
    {
      "epoch": 0.9476311692163211,
      "grad_norm": 3.7720568998272364,
      "learning_rate": 1.4312511186861053e-09,
      "loss": 0.7638,
      "step": 9586
    },
    {
      "epoch": 0.9477300249610755,
      "grad_norm": 4.0993808269323235,
      "learning_rate": 1.425857959791077e-09,
      "loss": 0.7151,
      "step": 9587
    },
    {
      "epoch": 0.94782888070583,
      "grad_norm": 3.995536861087634,
      "learning_rate": 1.4204749082202482e-09,
      "loss": 0.7474,
      "step": 9588
    },
    {
      "epoch": 0.9479277364505845,
      "grad_norm": 3.3457523059123266,
      "learning_rate": 1.415101964525578e-09,
      "loss": 0.6676,
      "step": 9589
    },
    {
      "epoch": 0.9480265921953389,
      "grad_norm": 5.458175249248427,
      "learning_rate": 1.4097391292579586e-09,
      "loss": 0.6722,
      "step": 9590
    },
    {
      "epoch": 0.9481254479400935,
      "grad_norm": 15.708911263237425,
      "learning_rate": 1.4043864029672948e-09,
      "loss": 0.6667,
      "step": 9591
    },
    {
      "epoch": 0.9482243036848479,
      "grad_norm": 2.731033598470992,
      "learning_rate": 1.399043786202403e-09,
      "loss": 0.7365,
      "step": 9592
    },
    {
      "epoch": 0.9483231594296023,
      "grad_norm": 5.586102167529366,
      "learning_rate": 1.3937112795111117e-09,
      "loss": 0.6179,
      "step": 9593
    },
    {
      "epoch": 0.9484220151743569,
      "grad_norm": 3.190846058589971,
      "learning_rate": 1.3883888834401836e-09,
      "loss": 0.7002,
      "step": 9594
    },
    {
      "epoch": 0.9485208709191113,
      "grad_norm": 7.143162335583241,
      "learning_rate": 1.3830765985353488e-09,
      "loss": 0.6477,
      "step": 9595
    },
    {
      "epoch": 0.9486197266638657,
      "grad_norm": 4.003028787628363,
      "learning_rate": 1.3777744253413159e-09,
      "loss": 0.7631,
      "step": 9596
    },
    {
      "epoch": 0.9487185824086202,
      "grad_norm": 13.211707019449607,
      "learning_rate": 1.372482364401728e-09,
      "loss": 0.6653,
      "step": 9597
    },
    {
      "epoch": 0.9488174381533747,
      "grad_norm": 3.266825494737795,
      "learning_rate": 1.3672004162592287e-09,
      "loss": 0.7338,
      "step": 9598
    },
    {
      "epoch": 0.9489162938981291,
      "grad_norm": 3.202834326360478,
      "learning_rate": 1.3619285814553738e-09,
      "loss": 0.7086,
      "step": 9599
    },
    {
      "epoch": 0.9490151496428836,
      "grad_norm": 4.083126667197546,
      "learning_rate": 1.356666860530742e-09,
      "loss": 0.8052,
      "step": 9600
    },
    {
      "epoch": 0.9491140053876381,
      "grad_norm": 5.352618424847311,
      "learning_rate": 1.3514152540248237e-09,
      "loss": 0.7569,
      "step": 9601
    },
    {
      "epoch": 0.9492128611323926,
      "grad_norm": 3.451007196577972,
      "learning_rate": 1.3461737624760993e-09,
      "loss": 0.6259,
      "step": 9602
    },
    {
      "epoch": 0.949311716877147,
      "grad_norm": 3.8648539121877,
      "learning_rate": 1.340942386422017e-09,
      "loss": 0.7411,
      "step": 9603
    },
    {
      "epoch": 0.9494105726219015,
      "grad_norm": 3.8074213702915505,
      "learning_rate": 1.3357211263989588e-09,
      "loss": 0.7252,
      "step": 9604
    },
    {
      "epoch": 0.949509428366656,
      "grad_norm": 13.916537197939082,
      "learning_rate": 1.3305099829423072e-09,
      "loss": 0.6702,
      "step": 9605
    },
    {
      "epoch": 0.9496082841114104,
      "grad_norm": 3.822858963432504,
      "learning_rate": 1.3253089565863794e-09,
      "loss": 0.6091,
      "step": 9606
    },
    {
      "epoch": 0.9497071398561648,
      "grad_norm": 4.665566657321777,
      "learning_rate": 1.3201180478644601e-09,
      "loss": 0.7278,
      "step": 9607
    },
    {
      "epoch": 0.9498059956009194,
      "grad_norm": 5.033409044133929,
      "learning_rate": 1.3149372573088123e-09,
      "loss": 0.7568,
      "step": 9608
    },
    {
      "epoch": 0.9499048513456738,
      "grad_norm": 4.320330014231467,
      "learning_rate": 1.3097665854506334e-09,
      "loss": 0.6726,
      "step": 9609
    },
    {
      "epoch": 0.9500037070904283,
      "grad_norm": 4.935807217177442,
      "learning_rate": 1.3046060328200992e-09,
      "loss": 0.8174,
      "step": 9610
    },
    {
      "epoch": 0.9501025628351828,
      "grad_norm": 3.720533022372357,
      "learning_rate": 1.2994555999463751e-09,
      "loss": 0.723,
      "step": 9611
    },
    {
      "epoch": 0.9502014185799372,
      "grad_norm": 8.625374400364402,
      "learning_rate": 1.2943152873575391e-09,
      "loss": 0.6938,
      "step": 9612
    },
    {
      "epoch": 0.9503002743246917,
      "grad_norm": 4.266800230286825,
      "learning_rate": 1.2891850955806582e-09,
      "loss": 0.6431,
      "step": 9613
    },
    {
      "epoch": 0.9503991300694462,
      "grad_norm": 5.13796154773431,
      "learning_rate": 1.2840650251417562e-09,
      "loss": 0.6069,
      "step": 9614
    },
    {
      "epoch": 0.9504979858142006,
      "grad_norm": 3.8725358868236706,
      "learning_rate": 1.2789550765658241e-09,
      "loss": 0.6704,
      "step": 9615
    },
    {
      "epoch": 0.9505968415589551,
      "grad_norm": 4.105359328341502,
      "learning_rate": 1.2738552503768096e-09,
      "loss": 0.7355,
      "step": 9616
    },
    {
      "epoch": 0.9506956973037095,
      "grad_norm": 3.088569221513314,
      "learning_rate": 1.2687655470976277e-09,
      "loss": 0.7426,
      "step": 9617
    },
    {
      "epoch": 0.950794553048464,
      "grad_norm": 4.319713397637834,
      "learning_rate": 1.2636859672501498e-09,
      "loss": 0.7625,
      "step": 9618
    },
    {
      "epoch": 0.9508934087932185,
      "grad_norm": 3.821154319114971,
      "learning_rate": 1.258616511355215e-09,
      "loss": 0.6724,
      "step": 9619
    },
    {
      "epoch": 0.9509922645379729,
      "grad_norm": 4.032760426388844,
      "learning_rate": 1.2535571799326183e-09,
      "loss": 0.7275,
      "step": 9620
    },
    {
      "epoch": 0.9510911202827275,
      "grad_norm": 4.198799957790205,
      "learning_rate": 1.2485079735011227e-09,
      "loss": 0.6757,
      "step": 9621
    },
    {
      "epoch": 0.9511899760274819,
      "grad_norm": 2.918365892761069,
      "learning_rate": 1.2434688925784476e-09,
      "loss": 0.6602,
      "step": 9622
    },
    {
      "epoch": 0.9512888317722363,
      "grad_norm": 5.697747485084655,
      "learning_rate": 1.2384399376812683e-09,
      "loss": 0.6641,
      "step": 9623
    },
    {
      "epoch": 0.9513876875169909,
      "grad_norm": 3.4429000935753216,
      "learning_rate": 1.23342110932525e-09,
      "loss": 0.7265,
      "step": 9624
    },
    {
      "epoch": 0.9514865432617453,
      "grad_norm": 3.382926476464799,
      "learning_rate": 1.2284124080249703e-09,
      "loss": 0.654,
      "step": 9625
    },
    {
      "epoch": 0.9515853990064997,
      "grad_norm": 3.7131804057062703,
      "learning_rate": 1.2234138342940293e-09,
      "loss": 0.5713,
      "step": 9626
    },
    {
      "epoch": 0.9516842547512542,
      "grad_norm": 7.390561081347685,
      "learning_rate": 1.2184253886449392e-09,
      "loss": 0.6979,
      "step": 9627
    },
    {
      "epoch": 0.9517831104960087,
      "grad_norm": 5.243705524115942,
      "learning_rate": 1.213447071589191e-09,
      "loss": 0.7429,
      "step": 9628
    },
    {
      "epoch": 0.9518819662407632,
      "grad_norm": 3.55169440754598,
      "learning_rate": 1.2084788836372318e-09,
      "loss": 0.754,
      "step": 9629
    },
    {
      "epoch": 0.9519808219855176,
      "grad_norm": 7.556704159227265,
      "learning_rate": 1.2035208252985097e-09,
      "loss": 0.7159,
      "step": 9630
    },
    {
      "epoch": 0.9520796777302721,
      "grad_norm": 4.888665254885543,
      "learning_rate": 1.1985728970813625e-09,
      "loss": 0.6682,
      "step": 9631
    },
    {
      "epoch": 0.9521785334750266,
      "grad_norm": 10.85766643603529,
      "learning_rate": 1.1936350994931398e-09,
      "loss": 0.6057,
      "step": 9632
    },
    {
      "epoch": 0.952277389219781,
      "grad_norm": 7.950734072601201,
      "learning_rate": 1.1887074330401481e-09,
      "loss": 0.7753,
      "step": 9633
    },
    {
      "epoch": 0.9523762449645355,
      "grad_norm": 20.72581477301955,
      "learning_rate": 1.1837898982276384e-09,
      "loss": 0.7647,
      "step": 9634
    },
    {
      "epoch": 0.95247510070929,
      "grad_norm": 4.554731291558747,
      "learning_rate": 1.1788824955598299e-09,
      "loss": 0.6497,
      "step": 9635
    },
    {
      "epoch": 0.9525739564540444,
      "grad_norm": 4.52733011973087,
      "learning_rate": 1.1739852255399086e-09,
      "loss": 0.7165,
      "step": 9636
    },
    {
      "epoch": 0.952672812198799,
      "grad_norm": 3.9655019303795602,
      "learning_rate": 1.1690980886700063e-09,
      "loss": 0.6843,
      "step": 9637
    },
    {
      "epoch": 0.9527716679435534,
      "grad_norm": 5.967078803704163,
      "learning_rate": 1.1642210854512557e-09,
      "loss": 0.7442,
      "step": 9638
    },
    {
      "epoch": 0.9528705236883078,
      "grad_norm": 2.9626798270821686,
      "learning_rate": 1.15935421638369e-09,
      "loss": 0.6834,
      "step": 9639
    },
    {
      "epoch": 0.9529693794330623,
      "grad_norm": 4.304890334308182,
      "learning_rate": 1.1544974819663434e-09,
      "loss": 0.7134,
      "step": 9640
    },
    {
      "epoch": 0.9530682351778168,
      "grad_norm": 4.210812933699883,
      "learning_rate": 1.1496508826972172e-09,
      "loss": 0.7285,
      "step": 9641
    },
    {
      "epoch": 0.9531670909225712,
      "grad_norm": 22.28098723576354,
      "learning_rate": 1.1448144190732479e-09,
      "loss": 0.6296,
      "step": 9642
    },
    {
      "epoch": 0.9532659466673257,
      "grad_norm": 3.6553961273250697,
      "learning_rate": 1.139988091590327e-09,
      "loss": 0.5938,
      "step": 9643
    },
    {
      "epoch": 0.9533648024120802,
      "grad_norm": 3.498547217589503,
      "learning_rate": 1.1351719007433592e-09,
      "loss": 0.5882,
      "step": 9644
    },
    {
      "epoch": 0.9534636581568346,
      "grad_norm": 4.03815831675684,
      "learning_rate": 1.130365847026149e-09,
      "loss": 0.7056,
      "step": 9645
    },
    {
      "epoch": 0.9535625139015891,
      "grad_norm": 3.5073943887801713,
      "learning_rate": 1.1255699309314914e-09,
      "loss": 0.7356,
      "step": 9646
    },
    {
      "epoch": 0.9536613696463436,
      "grad_norm": 2.9851538930781296,
      "learning_rate": 1.1207841529511263e-09,
      "loss": 0.6836,
      "step": 9647
    },
    {
      "epoch": 0.9537602253910981,
      "grad_norm": 20.80705722156873,
      "learning_rate": 1.1160085135757834e-09,
      "loss": 0.7815,
      "step": 9648
    },
    {
      "epoch": 0.9538590811358525,
      "grad_norm": 7.188801675477845,
      "learning_rate": 1.1112430132951266e-09,
      "loss": 0.6002,
      "step": 9649
    },
    {
      "epoch": 0.9539579368806069,
      "grad_norm": 4.615957562552767,
      "learning_rate": 1.1064876525977872e-09,
      "loss": 0.6842,
      "step": 9650
    },
    {
      "epoch": 0.9540567926253615,
      "grad_norm": 4.029867573310436,
      "learning_rate": 1.101742431971353e-09,
      "loss": 0.6653,
      "step": 9651
    },
    {
      "epoch": 0.9541556483701159,
      "grad_norm": 8.724206220828663,
      "learning_rate": 1.0970073519023904e-09,
      "loss": 0.7537,
      "step": 9652
    },
    {
      "epoch": 0.9542545041148703,
      "grad_norm": 3.6550534375922896,
      "learning_rate": 1.0922824128763997e-09,
      "loss": 0.7431,
      "step": 9653
    },
    {
      "epoch": 0.9543533598596249,
      "grad_norm": 4.142880988521536,
      "learning_rate": 1.087567615377849e-09,
      "loss": 0.7114,
      "step": 9654
    },
    {
      "epoch": 0.9544522156043793,
      "grad_norm": 3.4525522389574332,
      "learning_rate": 1.0828629598901739e-09,
      "loss": 0.6682,
      "step": 9655
    },
    {
      "epoch": 0.9545510713491338,
      "grad_norm": 32.59812766707639,
      "learning_rate": 1.0781684468957774e-09,
      "loss": 0.71,
      "step": 9656
    },
    {
      "epoch": 0.9546499270938883,
      "grad_norm": 13.05438644738811,
      "learning_rate": 1.073484076876019e-09,
      "loss": 0.6863,
      "step": 9657
    },
    {
      "epoch": 0.9547487828386427,
      "grad_norm": 3.7740075508938067,
      "learning_rate": 1.068809850311181e-09,
      "loss": 0.6884,
      "step": 9658
    },
    {
      "epoch": 0.9548476385833972,
      "grad_norm": 2.911363290074344,
      "learning_rate": 1.0641457676805577e-09,
      "loss": 0.6908,
      "step": 9659
    },
    {
      "epoch": 0.9549464943281516,
      "grad_norm": 4.5614308785149555,
      "learning_rate": 1.059491829462389e-09,
      "loss": 0.7117,
      "step": 9660
    },
    {
      "epoch": 0.9550453500729061,
      "grad_norm": 4.1285527033913985,
      "learning_rate": 1.0548480361338597e-09,
      "loss": 0.7176,
      "step": 9661
    },
    {
      "epoch": 0.9551442058176606,
      "grad_norm": 26.878823688206424,
      "learning_rate": 1.0502143881711113e-09,
      "loss": 0.5754,
      "step": 9662
    },
    {
      "epoch": 0.955243061562415,
      "grad_norm": 4.536673748063369,
      "learning_rate": 1.0455908860492745e-09,
      "loss": 0.6253,
      "step": 9663
    },
    {
      "epoch": 0.9553419173071696,
      "grad_norm": 5.417885611029824,
      "learning_rate": 1.040977530242404e-09,
      "loss": 0.7872,
      "step": 9664
    },
    {
      "epoch": 0.955440773051924,
      "grad_norm": 5.711958789984218,
      "learning_rate": 1.036374321223532e-09,
      "loss": 0.6848,
      "step": 9665
    },
    {
      "epoch": 0.9555396287966784,
      "grad_norm": 3.1785964713292216,
      "learning_rate": 1.0317812594646702e-09,
      "loss": 0.6607,
      "step": 9666
    },
    {
      "epoch": 0.955638484541433,
      "grad_norm": 3.3837373478797166,
      "learning_rate": 1.0271983454367639e-09,
      "loss": 0.6939,
      "step": 9667
    },
    {
      "epoch": 0.9557373402861874,
      "grad_norm": 18.033654552987084,
      "learning_rate": 1.0226255796096927e-09,
      "loss": 0.7108,
      "step": 9668
    },
    {
      "epoch": 0.9558361960309418,
      "grad_norm": 3.420800981305091,
      "learning_rate": 1.0180629624523707e-09,
      "loss": 0.7955,
      "step": 9669
    },
    {
      "epoch": 0.9559350517756963,
      "grad_norm": 8.328212262374233,
      "learning_rate": 1.01351049443259e-09,
      "loss": 0.7239,
      "step": 9670
    },
    {
      "epoch": 0.9560339075204508,
      "grad_norm": 5.201204712180959,
      "learning_rate": 1.0089681760171663e-09,
      "loss": 0.7447,
      "step": 9671
    },
    {
      "epoch": 0.9561327632652052,
      "grad_norm": 4.387677888756311,
      "learning_rate": 1.0044360076718273e-09,
      "loss": 0.7292,
      "step": 9672
    },
    {
      "epoch": 0.9562316190099597,
      "grad_norm": 4.500534657618348,
      "learning_rate": 9.999139898612896e-10,
      "loss": 0.7185,
      "step": 9673
    },
    {
      "epoch": 0.9563304747547142,
      "grad_norm": 3.9323838181360937,
      "learning_rate": 9.95402123049227e-10,
      "loss": 0.8698,
      "step": 9674
    },
    {
      "epoch": 0.9564293304994687,
      "grad_norm": 3.0976638983995906,
      "learning_rate": 9.909004076982474e-10,
      "loss": 0.6414,
      "step": 9675
    },
    {
      "epoch": 0.9565281862442231,
      "grad_norm": 4.022168224939847,
      "learning_rate": 9.86408844269937e-10,
      "loss": 0.7131,
      "step": 9676
    },
    {
      "epoch": 0.9566270419889776,
      "grad_norm": 2.7652593911582346,
      "learning_rate": 9.819274332248607e-10,
      "loss": 0.663,
      "step": 9677
    },
    {
      "epoch": 0.9567258977337321,
      "grad_norm": 3.585534024343479,
      "learning_rate": 9.774561750225064e-10,
      "loss": 0.7385,
      "step": 9678
    },
    {
      "epoch": 0.9568247534784865,
      "grad_norm": 3.856384147511755,
      "learning_rate": 9.729950701213296e-10,
      "loss": 0.655,
      "step": 9679
    },
    {
      "epoch": 0.9569236092232409,
      "grad_norm": 5.169840256261319,
      "learning_rate": 9.685441189787535e-10,
      "loss": 0.5754,
      "step": 9680
    },
    {
      "epoch": 0.9570224649679955,
      "grad_norm": 8.898940549930296,
      "learning_rate": 9.641033220511686e-10,
      "loss": 0.7557,
      "step": 9681
    },
    {
      "epoch": 0.9571213207127499,
      "grad_norm": 3.256256487517087,
      "learning_rate": 9.596726797939102e-10,
      "loss": 0.7442,
      "step": 9682
    },
    {
      "epoch": 0.9572201764575043,
      "grad_norm": 3.9139738065965393,
      "learning_rate": 9.5525219266126e-10,
      "loss": 0.6913,
      "step": 9683
    },
    {
      "epoch": 0.9573190322022589,
      "grad_norm": 3.5272098876913303,
      "learning_rate": 9.508418611064883e-10,
      "loss": 0.8209,
      "step": 9684
    },
    {
      "epoch": 0.9574178879470133,
      "grad_norm": 5.083151727401354,
      "learning_rate": 9.464416855818002e-10,
      "loss": 0.6329,
      "step": 9685
    },
    {
      "epoch": 0.9575167436917678,
      "grad_norm": 9.721560612665419,
      "learning_rate": 9.420516665383904e-10,
      "loss": 0.7256,
      "step": 9686
    },
    {
      "epoch": 0.9576155994365223,
      "grad_norm": 4.267563678627701,
      "learning_rate": 9.376718044263653e-10,
      "loss": 0.7852,
      "step": 9687
    },
    {
      "epoch": 0.9577144551812767,
      "grad_norm": 6.773626415891948,
      "learning_rate": 9.333020996948216e-10,
      "loss": 0.5887,
      "step": 9688
    },
    {
      "epoch": 0.9578133109260312,
      "grad_norm": 4.236723764927924,
      "learning_rate": 9.289425527918116e-10,
      "loss": 0.7548,
      "step": 9689
    },
    {
      "epoch": 0.9579121666707856,
      "grad_norm": 5.593399423076282,
      "learning_rate": 9.245931641643446e-10,
      "loss": 0.681,
      "step": 9690
    },
    {
      "epoch": 0.9580110224155401,
      "grad_norm": 3.3661947112741206,
      "learning_rate": 9.202539342583859e-10,
      "loss": 0.7419,
      "step": 9691
    },
    {
      "epoch": 0.9581098781602946,
      "grad_norm": 3.1598945524679385,
      "learning_rate": 9.159248635188466e-10,
      "loss": 0.6697,
      "step": 9692
    },
    {
      "epoch": 0.958208733905049,
      "grad_norm": 4.269866843534518,
      "learning_rate": 9.116059523896269e-10,
      "loss": 0.7978,
      "step": 9693
    },
    {
      "epoch": 0.9583075896498036,
      "grad_norm": 4.3204954154925765,
      "learning_rate": 9.072972013135616e-10,
      "loss": 0.7042,
      "step": 9694
    },
    {
      "epoch": 0.958406445394558,
      "grad_norm": 3.158918891628624,
      "learning_rate": 9.029986107324416e-10,
      "loss": 0.66,
      "step": 9695
    },
    {
      "epoch": 0.9585053011393124,
      "grad_norm": 3.0862266905690876,
      "learning_rate": 8.987101810870478e-10,
      "loss": 0.5958,
      "step": 9696
    },
    {
      "epoch": 0.958604156884067,
      "grad_norm": 3.9873507596323416,
      "learning_rate": 8.944319128170619e-10,
      "loss": 0.6888,
      "step": 9697
    },
    {
      "epoch": 0.9587030126288214,
      "grad_norm": 3.0080664566583724,
      "learning_rate": 8.901638063611661e-10,
      "loss": 0.6494,
      "step": 9698
    },
    {
      "epoch": 0.9588018683735758,
      "grad_norm": 3.4987754836462437,
      "learning_rate": 8.859058621570215e-10,
      "loss": 0.6369,
      "step": 9699
    },
    {
      "epoch": 0.9589007241183303,
      "grad_norm": 12.494324803650857,
      "learning_rate": 8.816580806411789e-10,
      "loss": 0.7379,
      "step": 9700
    },
    {
      "epoch": 0.9589995798630848,
      "grad_norm": 8.995062488748955,
      "learning_rate": 8.774204622492121e-10,
      "loss": 0.7173,
      "step": 9701
    },
    {
      "epoch": 0.9590984356078393,
      "grad_norm": 11.053461650189186,
      "learning_rate": 8.731930074156069e-10,
      "loss": 0.6636,
      "step": 9702
    },
    {
      "epoch": 0.9591972913525937,
      "grad_norm": 4.026287265933042,
      "learning_rate": 8.689757165738387e-10,
      "loss": 0.6949,
      "step": 9703
    },
    {
      "epoch": 0.9592961470973482,
      "grad_norm": 5.2627006179694575,
      "learning_rate": 8.647685901563284e-10,
      "loss": 0.6509,
      "step": 9704
    },
    {
      "epoch": 0.9593950028421027,
      "grad_norm": 3.651827585355856,
      "learning_rate": 8.60571628594442e-10,
      "loss": 0.7442,
      "step": 9705
    },
    {
      "epoch": 0.9594938585868571,
      "grad_norm": 4.194697476578764,
      "learning_rate": 8.563848323185353e-10,
      "loss": 0.7904,
      "step": 9706
    },
    {
      "epoch": 0.9595927143316116,
      "grad_norm": 13.33652523740163,
      "learning_rate": 8.522082017578869e-10,
      "loss": 0.8419,
      "step": 9707
    },
    {
      "epoch": 0.9596915700763661,
      "grad_norm": 4.820101658623336,
      "learning_rate": 8.480417373407656e-10,
      "loss": 0.6705,
      "step": 9708
    },
    {
      "epoch": 0.9597904258211205,
      "grad_norm": 6.728002230520101,
      "learning_rate": 8.438854394943518e-10,
      "loss": 0.7492,
      "step": 9709
    },
    {
      "epoch": 0.959889281565875,
      "grad_norm": 3.5587468298612297,
      "learning_rate": 8.39739308644849e-10,
      "loss": 0.7615,
      "step": 9710
    },
    {
      "epoch": 0.9599881373106295,
      "grad_norm": 3.1144952089965647,
      "learning_rate": 8.356033452173505e-10,
      "loss": 0.7201,
      "step": 9711
    },
    {
      "epoch": 0.9600869930553839,
      "grad_norm": 6.799729312686852,
      "learning_rate": 8.314775496359616e-10,
      "loss": 0.7051,
      "step": 9712
    },
    {
      "epoch": 0.9601858488001384,
      "grad_norm": 3.1806485460535963,
      "learning_rate": 8.273619223236994e-10,
      "loss": 0.684,
      "step": 9713
    },
    {
      "epoch": 0.9602847045448929,
      "grad_norm": 4.980913827814384,
      "learning_rate": 8.232564637025707e-10,
      "loss": 0.7499,
      "step": 9714
    },
    {
      "epoch": 0.9603835602896473,
      "grad_norm": 5.738533061745901,
      "learning_rate": 8.1916117419355e-10,
      "loss": 0.6237,
      "step": 9715
    },
    {
      "epoch": 0.9604824160344018,
      "grad_norm": 4.332679892084157,
      "learning_rate": 8.150760542165125e-10,
      "loss": 0.7221,
      "step": 9716
    },
    {
      "epoch": 0.9605812717791563,
      "grad_norm": 4.473401907434002,
      "learning_rate": 8.110011041903564e-10,
      "loss": 0.6932,
      "step": 9717
    },
    {
      "epoch": 0.9606801275239107,
      "grad_norm": 7.415137561464604,
      "learning_rate": 8.069363245328919e-10,
      "loss": 0.7354,
      "step": 9718
    },
    {
      "epoch": 0.9607789832686652,
      "grad_norm": 4.90114092837828,
      "learning_rate": 8.02881715660908e-10,
      "loss": 0.7159,
      "step": 9719
    },
    {
      "epoch": 0.9608778390134197,
      "grad_norm": 4.186600454680899,
      "learning_rate": 7.988372779901387e-10,
      "loss": 0.7927,
      "step": 9720
    },
    {
      "epoch": 0.9609766947581742,
      "grad_norm": 3.7126562603143123,
      "learning_rate": 7.948030119352966e-10,
      "loss": 0.7742,
      "step": 9721
    },
    {
      "epoch": 0.9610755505029286,
      "grad_norm": 3.3351796314899347,
      "learning_rate": 7.907789179100177e-10,
      "loss": 0.7884,
      "step": 9722
    },
    {
      "epoch": 0.961174406247683,
      "grad_norm": 3.612729035356766,
      "learning_rate": 7.867649963269274e-10,
      "loss": 0.6505,
      "step": 9723
    },
    {
      "epoch": 0.9612732619924376,
      "grad_norm": 5.102026893976018,
      "learning_rate": 7.827612475975853e-10,
      "loss": 0.6761,
      "step": 9724
    },
    {
      "epoch": 0.961372117737192,
      "grad_norm": 3.240145263389014,
      "learning_rate": 7.787676721325187e-10,
      "loss": 0.6334,
      "step": 9725
    },
    {
      "epoch": 0.9614709734819464,
      "grad_norm": 2.8857992231216625,
      "learning_rate": 7.747842703412111e-10,
      "loss": 0.6471,
      "step": 9726
    },
    {
      "epoch": 0.961569829226701,
      "grad_norm": 11.087193659402446,
      "learning_rate": 7.708110426321024e-10,
      "loss": 0.7482,
      "step": 9727
    },
    {
      "epoch": 0.9616686849714554,
      "grad_norm": 4.395452736301807,
      "learning_rate": 7.66847989412589e-10,
      "loss": 0.6584,
      "step": 9728
    },
    {
      "epoch": 0.9617675407162098,
      "grad_norm": 5.473843599944854,
      "learning_rate": 7.628951110890236e-10,
      "loss": 0.7508,
      "step": 9729
    },
    {
      "epoch": 0.9618663964609644,
      "grad_norm": 4.3477825167187705,
      "learning_rate": 7.589524080667154e-10,
      "loss": 0.677,
      "step": 9730
    },
    {
      "epoch": 0.9619652522057188,
      "grad_norm": 3.6123872421783236,
      "learning_rate": 7.550198807499187e-10,
      "loss": 0.6595,
      "step": 9731
    },
    {
      "epoch": 0.9620641079504733,
      "grad_norm": 2.6982750489933647,
      "learning_rate": 7.510975295418776e-10,
      "loss": 0.7506,
      "step": 9732
    },
    {
      "epoch": 0.9621629636952277,
      "grad_norm": 5.267785490279878,
      "learning_rate": 7.471853548447593e-10,
      "loss": 0.6672,
      "step": 9733
    },
    {
      "epoch": 0.9622618194399822,
      "grad_norm": 4.360093215967715,
      "learning_rate": 7.432833570596985e-10,
      "loss": 0.6338,
      "step": 9734
    },
    {
      "epoch": 0.9623606751847367,
      "grad_norm": 3.1019696241552093,
      "learning_rate": 7.393915365867864e-10,
      "loss": 0.6757,
      "step": 9735
    },
    {
      "epoch": 0.9624595309294911,
      "grad_norm": 3.3082399215520937,
      "learning_rate": 7.355098938250815e-10,
      "loss": 0.7658,
      "step": 9736
    },
    {
      "epoch": 0.9625583866742456,
      "grad_norm": 5.742720731818882,
      "learning_rate": 7.316384291725763e-10,
      "loss": 0.7226,
      "step": 9737
    },
    {
      "epoch": 0.9626572424190001,
      "grad_norm": 5.126850206228592,
      "learning_rate": 7.277771430262536e-10,
      "loss": 0.6803,
      "step": 9738
    },
    {
      "epoch": 0.9627560981637545,
      "grad_norm": 9.714562543482561,
      "learning_rate": 7.239260357819965e-10,
      "loss": 0.6812,
      "step": 9739
    },
    {
      "epoch": 0.9628549539085091,
      "grad_norm": 3.8644242410869176,
      "learning_rate": 7.200851078347114e-10,
      "loss": 0.7827,
      "step": 9740
    },
    {
      "epoch": 0.9629538096532635,
      "grad_norm": 4.071164485270337,
      "learning_rate": 7.162543595782166e-10,
      "loss": 0.7898,
      "step": 9741
    },
    {
      "epoch": 0.9630526653980179,
      "grad_norm": 6.625017485617194,
      "learning_rate": 7.124337914052869e-10,
      "loss": 0.7369,
      "step": 9742
    },
    {
      "epoch": 0.9631515211427724,
      "grad_norm": 4.758917132208141,
      "learning_rate": 7.086234037076976e-10,
      "loss": 0.6858,
      "step": 9743
    },
    {
      "epoch": 0.9632503768875269,
      "grad_norm": 3.725849334856043,
      "learning_rate": 7.048231968761142e-10,
      "loss": 0.6787,
      "step": 9744
    },
    {
      "epoch": 0.9633492326322813,
      "grad_norm": 3.1488583994301824,
      "learning_rate": 7.010331713002027e-10,
      "loss": 0.6303,
      "step": 9745
    },
    {
      "epoch": 0.9634480883770358,
      "grad_norm": 8.119405740304217,
      "learning_rate": 6.972533273685854e-10,
      "loss": 0.7227,
      "step": 9746
    },
    {
      "epoch": 0.9635469441217903,
      "grad_norm": 4.186050938117485,
      "learning_rate": 6.934836654688081e-10,
      "loss": 0.7752,
      "step": 9747
    },
    {
      "epoch": 0.9636457998665447,
      "grad_norm": 4.115467115259784,
      "learning_rate": 6.89724185987417e-10,
      "loss": 0.6861,
      "step": 9748
    },
    {
      "epoch": 0.9637446556112992,
      "grad_norm": 6.16089539141914,
      "learning_rate": 6.859748893098816e-10,
      "loss": 0.615,
      "step": 9749
    },
    {
      "epoch": 0.9638435113560537,
      "grad_norm": 4.725833263509995,
      "learning_rate": 6.822357758206276e-10,
      "loss": 0.6372,
      "step": 9750
    },
    {
      "epoch": 0.9639423671008082,
      "grad_norm": 4.5118571425752485,
      "learning_rate": 6.785068459030707e-10,
      "loss": 0.7269,
      "step": 9751
    },
    {
      "epoch": 0.9640412228455626,
      "grad_norm": 10.53675692264601,
      "learning_rate": 6.747880999395272e-10,
      "loss": 0.7397,
      "step": 9752
    },
    {
      "epoch": 0.964140078590317,
      "grad_norm": 23.27990148023194,
      "learning_rate": 6.710795383113143e-10,
      "loss": 0.6905,
      "step": 9753
    },
    {
      "epoch": 0.9642389343350716,
      "grad_norm": 4.432709126082438,
      "learning_rate": 6.673811613986946e-10,
      "loss": 0.7574,
      "step": 9754
    },
    {
      "epoch": 0.964337790079826,
      "grad_norm": 4.841921083451247,
      "learning_rate": 6.636929695808868e-10,
      "loss": 0.7212,
      "step": 9755
    },
    {
      "epoch": 0.9644366458245804,
      "grad_norm": 5.069606632487082,
      "learning_rate": 6.60014963236033e-10,
      "loss": 0.6933,
      "step": 9756
    },
    {
      "epoch": 0.964535501569335,
      "grad_norm": 6.3082976381566365,
      "learning_rate": 6.56347142741298e-10,
      "loss": 0.6713,
      "step": 9757
    },
    {
      "epoch": 0.9646343573140894,
      "grad_norm": 3.3573668685547418,
      "learning_rate": 6.526895084727257e-10,
      "loss": 0.7103,
      "step": 9758
    },
    {
      "epoch": 0.9647332130588439,
      "grad_norm": 3.076213164864413,
      "learning_rate": 6.490420608053715e-10,
      "loss": 0.7103,
      "step": 9759
    },
    {
      "epoch": 0.9648320688035984,
      "grad_norm": 3.1470329026729127,
      "learning_rate": 6.45404800113225e-10,
      "loss": 0.7622,
      "step": 9760
    },
    {
      "epoch": 0.9649309245483528,
      "grad_norm": 3.419263046341245,
      "learning_rate": 6.417777267692326e-10,
      "loss": 0.7406,
      "step": 9761
    },
    {
      "epoch": 0.9650297802931073,
      "grad_norm": 8.01218112882822,
      "learning_rate": 6.381608411452965e-10,
      "loss": 0.6728,
      "step": 9762
    },
    {
      "epoch": 0.9651286360378617,
      "grad_norm": 4.410672966082493,
      "learning_rate": 6.345541436122759e-10,
      "loss": 0.7442,
      "step": 9763
    },
    {
      "epoch": 0.9652274917826162,
      "grad_norm": 4.696448536849743,
      "learning_rate": 6.309576345399858e-10,
      "loss": 0.6381,
      "step": 9764
    },
    {
      "epoch": 0.9653263475273707,
      "grad_norm": 3.170348971763446,
      "learning_rate": 6.273713142971871e-10,
      "loss": 0.7322,
      "step": 9765
    },
    {
      "epoch": 0.9654252032721251,
      "grad_norm": 7.433237395538701,
      "learning_rate": 6.237951832516186e-10,
      "loss": 0.68,
      "step": 9766
    },
    {
      "epoch": 0.9655240590168797,
      "grad_norm": 3.875400779386129,
      "learning_rate": 6.20229241769954e-10,
      "loss": 0.7307,
      "step": 9767
    },
    {
      "epoch": 0.9656229147616341,
      "grad_norm": 16.586924224126516,
      "learning_rate": 6.166734902178228e-10,
      "loss": 0.7675,
      "step": 9768
    },
    {
      "epoch": 0.9657217705063885,
      "grad_norm": 2.814432549697897,
      "learning_rate": 6.131279289598113e-10,
      "loss": 0.6566,
      "step": 9769
    },
    {
      "epoch": 0.9658206262511431,
      "grad_norm": 3.524509665070472,
      "learning_rate": 6.095925583594841e-10,
      "loss": 0.7798,
      "step": 9770
    },
    {
      "epoch": 0.9659194819958975,
      "grad_norm": 3.6488139135331203,
      "learning_rate": 6.060673787793181e-10,
      "loss": 0.7088,
      "step": 9771
    },
    {
      "epoch": 0.9660183377406519,
      "grad_norm": 5.01039292669606,
      "learning_rate": 6.025523905807795e-10,
      "loss": 0.672,
      "step": 9772
    },
    {
      "epoch": 0.9661171934854064,
      "grad_norm": 8.93765463919214,
      "learning_rate": 5.990475941242801e-10,
      "loss": 0.8412,
      "step": 9773
    },
    {
      "epoch": 0.9662160492301609,
      "grad_norm": 3.0111872846821623,
      "learning_rate": 5.95552989769188e-10,
      "loss": 0.7112,
      "step": 9774
    },
    {
      "epoch": 0.9663149049749153,
      "grad_norm": 4.980707458375923,
      "learning_rate": 5.920685778738055e-10,
      "loss": 0.7043,
      "step": 9775
    },
    {
      "epoch": 0.9664137607196698,
      "grad_norm": 6.119793841855568,
      "learning_rate": 5.885943587954356e-10,
      "loss": 0.7859,
      "step": 9776
    },
    {
      "epoch": 0.9665126164644243,
      "grad_norm": 3.584889669981746,
      "learning_rate": 5.851303328902823e-10,
      "loss": 0.6501,
      "step": 9777
    },
    {
      "epoch": 0.9666114722091788,
      "grad_norm": 4.7611567316222985,
      "learning_rate": 5.816765005135393e-10,
      "loss": 0.6904,
      "step": 9778
    },
    {
      "epoch": 0.9667103279539332,
      "grad_norm": 3.280101140946791,
      "learning_rate": 5.782328620193566e-10,
      "loss": 0.7019,
      "step": 9779
    },
    {
      "epoch": 0.9668091836986877,
      "grad_norm": 3.4807381079820012,
      "learning_rate": 5.747994177608073e-10,
      "loss": 0.7081,
      "step": 9780
    },
    {
      "epoch": 0.9669080394434422,
      "grad_norm": 5.463363369625085,
      "learning_rate": 5.713761680899654e-10,
      "loss": 0.5903,
      "step": 9781
    },
    {
      "epoch": 0.9670068951881966,
      "grad_norm": 6.110272995259872,
      "learning_rate": 5.679631133578166e-10,
      "loss": 0.807,
      "step": 9782
    },
    {
      "epoch": 0.967105750932951,
      "grad_norm": 6.232554247048842,
      "learning_rate": 5.645602539143257e-10,
      "loss": 0.7717,
      "step": 9783
    },
    {
      "epoch": 0.9672046066777056,
      "grad_norm": 4.768974985616243,
      "learning_rate": 5.611675901084023e-10,
      "loss": 0.677,
      "step": 9784
    },
    {
      "epoch": 0.96730346242246,
      "grad_norm": 3.6904399764117266,
      "learning_rate": 5.577851222879127e-10,
      "loss": 0.7649,
      "step": 9785
    },
    {
      "epoch": 0.9674023181672144,
      "grad_norm": 3.227719385733215,
      "learning_rate": 5.544128507996903e-10,
      "loss": 0.6948,
      "step": 9786
    },
    {
      "epoch": 0.967501173911969,
      "grad_norm": 3.5532518998844864,
      "learning_rate": 5.510507759895033e-10,
      "loss": 0.7744,
      "step": 9787
    },
    {
      "epoch": 0.9676000296567234,
      "grad_norm": 7.312322990578629,
      "learning_rate": 5.476988982020869e-10,
      "loss": 0.7747,
      "step": 9788
    },
    {
      "epoch": 0.9676988854014779,
      "grad_norm": 3.9065111260856957,
      "learning_rate": 5.443572177811217e-10,
      "loss": 0.6107,
      "step": 9789
    },
    {
      "epoch": 0.9677977411462324,
      "grad_norm": 5.849672187547683,
      "learning_rate": 5.410257350692449e-10,
      "loss": 0.7302,
      "step": 9790
    },
    {
      "epoch": 0.9678965968909868,
      "grad_norm": 2.8837299740660804,
      "learning_rate": 5.377044504080608e-10,
      "loss": 0.612,
      "step": 9791
    },
    {
      "epoch": 0.9679954526357413,
      "grad_norm": 3.8240600878818936,
      "learning_rate": 5.343933641381193e-10,
      "loss": 0.6592,
      "step": 9792
    },
    {
      "epoch": 0.9680943083804958,
      "grad_norm": 6.130189209584632,
      "learning_rate": 5.310924765989044e-10,
      "loss": 0.6922,
      "step": 9793
    },
    {
      "epoch": 0.9681931641252502,
      "grad_norm": 3.81312987172568,
      "learning_rate": 5.278017881288898e-10,
      "loss": 0.6982,
      "step": 9794
    },
    {
      "epoch": 0.9682920198700047,
      "grad_norm": 43.36402181810251,
      "learning_rate": 5.245212990654835e-10,
      "loss": 0.6802,
      "step": 9795
    },
    {
      "epoch": 0.9683908756147591,
      "grad_norm": 11.284023388337026,
      "learning_rate": 5.212510097450606e-10,
      "loss": 0.5642,
      "step": 9796
    },
    {
      "epoch": 0.9684897313595137,
      "grad_norm": 5.984420379265393,
      "learning_rate": 5.179909205029198e-10,
      "loss": 0.7121,
      "step": 9797
    },
    {
      "epoch": 0.9685885871042681,
      "grad_norm": 5.519638998020378,
      "learning_rate": 5.147410316733491e-10,
      "loss": 0.6762,
      "step": 9798
    },
    {
      "epoch": 0.9686874428490225,
      "grad_norm": 3.9828941993472444,
      "learning_rate": 5.115013435895821e-10,
      "loss": 0.679,
      "step": 9799
    },
    {
      "epoch": 0.9687862985937771,
      "grad_norm": 7.862648797220195,
      "learning_rate": 5.082718565837862e-10,
      "loss": 0.7183,
      "step": 9800
    },
    {
      "epoch": 0.9688851543385315,
      "grad_norm": 5.496884303112868,
      "learning_rate": 5.050525709871078e-10,
      "loss": 0.7439,
      "step": 9801
    },
    {
      "epoch": 0.9689840100832859,
      "grad_norm": 3.3413601803040383,
      "learning_rate": 5.018434871296384e-10,
      "loss": 0.5929,
      "step": 9802
    },
    {
      "epoch": 0.9690828658280405,
      "grad_norm": 3.602073932968947,
      "learning_rate": 4.986446053404148e-10,
      "loss": 0.7776,
      "step": 9803
    },
    {
      "epoch": 0.9691817215727949,
      "grad_norm": 21.958791351568347,
      "learning_rate": 4.954559259474411e-10,
      "loss": 0.6789,
      "step": 9804
    },
    {
      "epoch": 0.9692805773175494,
      "grad_norm": 6.777421932263544,
      "learning_rate": 4.92277449277656e-10,
      "loss": 0.7388,
      "step": 9805
    },
    {
      "epoch": 0.9693794330623038,
      "grad_norm": 3.893636829701767,
      "learning_rate": 4.891091756569876e-10,
      "loss": 0.6374,
      "step": 9806
    },
    {
      "epoch": 0.9694782888070583,
      "grad_norm": 12.890525529240557,
      "learning_rate": 4.859511054102872e-10,
      "loss": 0.77,
      "step": 9807
    },
    {
      "epoch": 0.9695771445518128,
      "grad_norm": 3.466698809403156,
      "learning_rate": 4.828032388613623e-10,
      "loss": 0.6431,
      "step": 9808
    },
    {
      "epoch": 0.9696760002965672,
      "grad_norm": 3.5365022998460818,
      "learning_rate": 4.796655763329771e-10,
      "loss": 0.6638,
      "step": 9809
    },
    {
      "epoch": 0.9697748560413217,
      "grad_norm": 4.105668325236833,
      "learning_rate": 4.76538118146863e-10,
      "loss": 0.7199,
      "step": 9810
    },
    {
      "epoch": 0.9698737117860762,
      "grad_norm": 2.736198003256814,
      "learning_rate": 4.734208646236971e-10,
      "loss": 0.6361,
      "step": 9811
    },
    {
      "epoch": 0.9699725675308306,
      "grad_norm": 4.3015621516780636,
      "learning_rate": 4.7031381608309e-10,
      "loss": 0.7504,
      "step": 9812
    },
    {
      "epoch": 0.9700714232755852,
      "grad_norm": 5.431033522245993,
      "learning_rate": 4.672169728436425e-10,
      "loss": 0.684,
      "step": 9813
    },
    {
      "epoch": 0.9701702790203396,
      "grad_norm": 4.297093631581839,
      "learning_rate": 4.641303352228787e-10,
      "loss": 0.6638,
      "step": 9814
    },
    {
      "epoch": 0.970269134765094,
      "grad_norm": 4.18150798110243,
      "learning_rate": 4.6105390353728954e-10,
      "loss": 0.783,
      "step": 9815
    },
    {
      "epoch": 0.9703679905098485,
      "grad_norm": 7.59770805003625,
      "learning_rate": 4.5798767810232287e-10,
      "loss": 0.6932,
      "step": 9816
    },
    {
      "epoch": 0.970466846254603,
      "grad_norm": 14.334301412268056,
      "learning_rate": 4.5493165923237153e-10,
      "loss": 0.7922,
      "step": 9817
    },
    {
      "epoch": 0.9705657019993574,
      "grad_norm": 3.868625049813854,
      "learning_rate": 4.5188584724078495e-10,
      "loss": 0.6097,
      "step": 9818
    },
    {
      "epoch": 0.9706645577441119,
      "grad_norm": 3.1787901625067962,
      "learning_rate": 4.488502424398688e-10,
      "loss": 0.781,
      "step": 9819
    },
    {
      "epoch": 0.9707634134888664,
      "grad_norm": 3.989400451892565,
      "learning_rate": 4.4582484514087417e-10,
      "loss": 0.7462,
      "step": 9820
    },
    {
      "epoch": 0.9708622692336208,
      "grad_norm": 6.539807280750128,
      "learning_rate": 4.428096556540084e-10,
      "loss": 0.7481,
      "step": 9821
    },
    {
      "epoch": 0.9709611249783753,
      "grad_norm": 3.1922652785589998,
      "learning_rate": 4.398046742884576e-10,
      "loss": 0.7372,
      "step": 9822
    },
    {
      "epoch": 0.9710599807231298,
      "grad_norm": 4.6887046279928635,
      "learning_rate": 4.3680990135230856e-10,
      "loss": 0.7521,
      "step": 9823
    },
    {
      "epoch": 0.9711588364678843,
      "grad_norm": 4.39018015083259,
      "learning_rate": 4.3382533715264903e-10,
      "loss": 0.6379,
      "step": 9824
    },
    {
      "epoch": 0.9712576922126387,
      "grad_norm": 3.5213994865775917,
      "learning_rate": 4.3085098199548977e-10,
      "loss": 0.8004,
      "step": 9825
    },
    {
      "epoch": 0.9713565479573931,
      "grad_norm": 3.5237488431662536,
      "learning_rate": 4.278868361858201e-10,
      "loss": 0.7018,
      "step": 9826
    },
    {
      "epoch": 0.9714554037021477,
      "grad_norm": 9.717385674846772,
      "learning_rate": 4.2493290002756367e-10,
      "loss": 0.7299,
      "step": 9827
    },
    {
      "epoch": 0.9715542594469021,
      "grad_norm": 3.9083619720909426,
      "learning_rate": 4.2198917382360033e-10,
      "loss": 0.6253,
      "step": 9828
    },
    {
      "epoch": 0.9716531151916565,
      "grad_norm": 5.013508086628578,
      "learning_rate": 4.190556578757665e-10,
      "loss": 0.7451,
      "step": 9829
    },
    {
      "epoch": 0.9717519709364111,
      "grad_norm": 4.524913204463662,
      "learning_rate": 4.1613235248485476e-10,
      "loss": 0.881,
      "step": 9830
    },
    {
      "epoch": 0.9718508266811655,
      "grad_norm": 3.432922706559564,
      "learning_rate": 4.132192579506144e-10,
      "loss": 0.7332,
      "step": 9831
    },
    {
      "epoch": 0.97194968242592,
      "grad_norm": 6.726804365772089,
      "learning_rate": 4.1031637457171754e-10,
      "loss": 0.7111,
      "step": 9832
    },
    {
      "epoch": 0.9720485381706745,
      "grad_norm": 4.431726548872343,
      "learning_rate": 4.07423702645826e-10,
      "loss": 0.6306,
      "step": 9833
    },
    {
      "epoch": 0.9721473939154289,
      "grad_norm": 5.290738578075236,
      "learning_rate": 4.045412424695471e-10,
      "loss": 0.7735,
      "step": 9834
    },
    {
      "epoch": 0.9722462496601834,
      "grad_norm": 3.7088054856265247,
      "learning_rate": 4.016689943384222e-10,
      "loss": 0.7313,
      "step": 9835
    },
    {
      "epoch": 0.9723451054049378,
      "grad_norm": 3.057922273226034,
      "learning_rate": 3.988069585469711e-10,
      "loss": 0.652,
      "step": 9836
    },
    {
      "epoch": 0.9724439611496923,
      "grad_norm": 10.725433282676205,
      "learning_rate": 3.95955135388637e-10,
      "loss": 0.6984,
      "step": 9837
    },
    {
      "epoch": 0.9725428168944468,
      "grad_norm": 3.873926827278017,
      "learning_rate": 3.931135251558415e-10,
      "loss": 0.7975,
      "step": 9838
    },
    {
      "epoch": 0.9726416726392012,
      "grad_norm": 10.654237845121067,
      "learning_rate": 3.902821281399515e-10,
      "loss": 0.6359,
      "step": 9839
    },
    {
      "epoch": 0.9727405283839557,
      "grad_norm": 8.382597783813386,
      "learning_rate": 3.874609446312793e-10,
      "loss": 0.8106,
      "step": 9840
    },
    {
      "epoch": 0.9728393841287102,
      "grad_norm": 2.993188768524057,
      "learning_rate": 3.8464997491909347e-10,
      "loss": 0.6845,
      "step": 9841
    },
    {
      "epoch": 0.9729382398734646,
      "grad_norm": 8.047100018416605,
      "learning_rate": 3.8184921929163005e-10,
      "loss": 0.7647,
      "step": 9842
    },
    {
      "epoch": 0.9730370956182192,
      "grad_norm": 3.5632866677879442,
      "learning_rate": 3.790586780360705e-10,
      "loss": 0.6946,
      "step": 9843
    },
    {
      "epoch": 0.9731359513629736,
      "grad_norm": 4.01233934497106,
      "learning_rate": 3.7627835143850814e-10,
      "loss": 0.6746,
      "step": 9844
    },
    {
      "epoch": 0.973234807107728,
      "grad_norm": 9.718228466762687,
      "learning_rate": 3.7350823978405944e-10,
      "loss": 0.7411,
      "step": 9845
    },
    {
      "epoch": 0.9733336628524825,
      "grad_norm": 9.445175571161732,
      "learning_rate": 3.707483433567526e-10,
      "loss": 0.7534,
      "step": 9846
    },
    {
      "epoch": 0.973432518597237,
      "grad_norm": 3.293166124551131,
      "learning_rate": 3.6799866243956144e-10,
      "loss": 0.6839,
      "step": 9847
    },
    {
      "epoch": 0.9735313743419914,
      "grad_norm": 3.7910276177122264,
      "learning_rate": 3.65259197314427e-10,
      "loss": 0.7231,
      "step": 9848
    },
    {
      "epoch": 0.9736302300867459,
      "grad_norm": 4.087673153428777,
      "learning_rate": 3.625299482622468e-10,
      "loss": 0.7625,
      "step": 9849
    },
    {
      "epoch": 0.9737290858315004,
      "grad_norm": 3.600363712852235,
      "learning_rate": 3.598109155628748e-10,
      "loss": 0.732,
      "step": 9850
    },
    {
      "epoch": 0.9738279415762549,
      "grad_norm": 2.866469683521575,
      "learning_rate": 3.5710209949508796e-10,
      "loss": 0.7544,
      "step": 9851
    },
    {
      "epoch": 0.9739267973210093,
      "grad_norm": 3.0487963334165307,
      "learning_rate": 3.544035003366419e-10,
      "loss": 0.6315,
      "step": 9852
    },
    {
      "epoch": 0.9740256530657638,
      "grad_norm": 4.156559151766531,
      "learning_rate": 3.517151183642486e-10,
      "loss": 0.8085,
      "step": 9853
    },
    {
      "epoch": 0.9741245088105183,
      "grad_norm": 22.842367884742874,
      "learning_rate": 3.4903695385355424e-10,
      "loss": 0.6553,
      "step": 9854
    },
    {
      "epoch": 0.9742233645552727,
      "grad_norm": 7.214943387830145,
      "learning_rate": 3.463690070791614e-10,
      "loss": 0.7119,
      "step": 9855
    },
    {
      "epoch": 0.9743222203000271,
      "grad_norm": 3.7418582447586535,
      "learning_rate": 3.437112783146401e-10,
      "loss": 0.7107,
      "step": 9856
    },
    {
      "epoch": 0.9744210760447817,
      "grad_norm": 4.302395281430418,
      "learning_rate": 3.410637678324835e-10,
      "loss": 0.7328,
      "step": 9857
    },
    {
      "epoch": 0.9745199317895361,
      "grad_norm": 5.366360172707314,
      "learning_rate": 3.384264759041744e-10,
      "loss": 0.6542,
      "step": 9858
    },
    {
      "epoch": 0.9746187875342905,
      "grad_norm": 4.336013585816708,
      "learning_rate": 3.3579940280010763e-10,
      "loss": 0.6488,
      "step": 9859
    },
    {
      "epoch": 0.9747176432790451,
      "grad_norm": 3.3833596872229617,
      "learning_rate": 3.331825487896678e-10,
      "loss": 0.6197,
      "step": 9860
    },
    {
      "epoch": 0.9748164990237995,
      "grad_norm": 2.9435571400162948,
      "learning_rate": 3.3057591414116237e-10,
      "loss": 0.6591,
      "step": 9861
    },
    {
      "epoch": 0.974915354768554,
      "grad_norm": 3.4856193449371244,
      "learning_rate": 3.279794991218776e-10,
      "loss": 0.7431,
      "step": 9862
    },
    {
      "epoch": 0.9750142105133085,
      "grad_norm": 4.65297764388433,
      "learning_rate": 3.253933039980339e-10,
      "loss": 0.8518,
      "step": 9863
    },
    {
      "epoch": 0.9751130662580629,
      "grad_norm": 3.611510734468384,
      "learning_rate": 3.228173290347969e-10,
      "loss": 0.6759,
      "step": 9864
    },
    {
      "epoch": 0.9752119220028174,
      "grad_norm": 4.396228649727912,
      "learning_rate": 3.202515744962886e-10,
      "loss": 0.7519,
      "step": 9865
    },
    {
      "epoch": 0.9753107777475719,
      "grad_norm": 14.82489325607934,
      "learning_rate": 3.176960406456208e-10,
      "loss": 0.741,
      "step": 9866
    },
    {
      "epoch": 0.9754096334923263,
      "grad_norm": 5.341015611270225,
      "learning_rate": 3.1515072774479513e-10,
      "loss": 0.7356,
      "step": 9867
    },
    {
      "epoch": 0.9755084892370808,
      "grad_norm": 3.5268333061095847,
      "learning_rate": 3.126156360548138e-10,
      "loss": 0.7241,
      "step": 9868
    },
    {
      "epoch": 0.9756073449818352,
      "grad_norm": 4.405675457532796,
      "learning_rate": 3.1009076583560225e-10,
      "loss": 0.7032,
      "step": 9869
    },
    {
      "epoch": 0.9757062007265898,
      "grad_norm": 3.2748381471452204,
      "learning_rate": 3.075761173460534e-10,
      "loss": 0.6133,
      "step": 9870
    },
    {
      "epoch": 0.9758050564713442,
      "grad_norm": 4.147903530018896,
      "learning_rate": 3.050716908440054e-10,
      "loss": 0.7325,
      "step": 9871
    },
    {
      "epoch": 0.9759039122160986,
      "grad_norm": 4.956861501530702,
      "learning_rate": 3.0257748658625295e-10,
      "loss": 0.7907,
      "step": 9872
    },
    {
      "epoch": 0.9760027679608532,
      "grad_norm": 3.986120343335268,
      "learning_rate": 3.000935048285469e-10,
      "loss": 0.6754,
      "step": 9873
    },
    {
      "epoch": 0.9761016237056076,
      "grad_norm": 6.364793656683157,
      "learning_rate": 2.9761974582556136e-10,
      "loss": 0.7633,
      "step": 9874
    },
    {
      "epoch": 0.976200479450362,
      "grad_norm": 3.6333559581309927,
      "learning_rate": 2.9515620983097124e-10,
      "loss": 0.7138,
      "step": 9875
    },
    {
      "epoch": 0.9762993351951166,
      "grad_norm": 3.491369699818869,
      "learning_rate": 2.9270289709735217e-10,
      "loss": 0.6814,
      "step": 9876
    },
    {
      "epoch": 0.976398190939871,
      "grad_norm": 9.03366985906941,
      "learning_rate": 2.902598078762697e-10,
      "loss": 0.7035,
      "step": 9877
    },
    {
      "epoch": 0.9764970466846254,
      "grad_norm": 5.319529687433717,
      "learning_rate": 2.8782694241821226e-10,
      "loss": 0.7387,
      "step": 9878
    },
    {
      "epoch": 0.9765959024293799,
      "grad_norm": 3.2612937519412237,
      "learning_rate": 2.854043009726581e-10,
      "loss": 0.7056,
      "step": 9879
    },
    {
      "epoch": 0.9766947581741344,
      "grad_norm": 4.177403945793153,
      "learning_rate": 2.829918837879863e-10,
      "loss": 0.6132,
      "step": 9880
    },
    {
      "epoch": 0.9767936139188889,
      "grad_norm": 2.6820373346316226,
      "learning_rate": 2.805896911115768e-10,
      "loss": 0.6527,
      "step": 9881
    },
    {
      "epoch": 0.9768924696636433,
      "grad_norm": 32.13629786149372,
      "learning_rate": 2.7819772318972146e-10,
      "loss": 0.7369,
      "step": 9882
    },
    {
      "epoch": 0.9769913254083978,
      "grad_norm": 3.899100037701286,
      "learning_rate": 2.758159802677018e-10,
      "loss": 0.7851,
      "step": 9883
    },
    {
      "epoch": 0.9770901811531523,
      "grad_norm": 4.07674298736393,
      "learning_rate": 2.734444625897003e-10,
      "loss": 0.8357,
      "step": 9884
    },
    {
      "epoch": 0.9771890368979067,
      "grad_norm": 3.430923453811778,
      "learning_rate": 2.710831703989114e-10,
      "loss": 0.6811,
      "step": 9885
    },
    {
      "epoch": 0.9772878926426612,
      "grad_norm": 8.666058120189678,
      "learning_rate": 2.687321039374413e-10,
      "loss": 0.7351,
      "step": 9886
    },
    {
      "epoch": 0.9773867483874157,
      "grad_norm": 4.4659607270315504,
      "learning_rate": 2.6639126344634165e-10,
      "loss": 0.6445,
      "step": 9887
    },
    {
      "epoch": 0.9774856041321701,
      "grad_norm": 3.6939422184739454,
      "learning_rate": 2.6406064916565385e-10,
      "loss": 0.6638,
      "step": 9888
    },
    {
      "epoch": 0.9775844598769245,
      "grad_norm": 3.813327268858609,
      "learning_rate": 2.6174026133433113e-10,
      "loss": 0.7485,
      "step": 9889
    },
    {
      "epoch": 0.9776833156216791,
      "grad_norm": 3.728782927727464,
      "learning_rate": 2.5943010019030545e-10,
      "loss": 0.7076,
      "step": 9890
    },
    {
      "epoch": 0.9777821713664335,
      "grad_norm": 2.958500823953866,
      "learning_rate": 2.571301659704539e-10,
      "loss": 0.6435,
      "step": 9891
    },
    {
      "epoch": 0.977881027111188,
      "grad_norm": 3.013869448883406,
      "learning_rate": 2.548404589105768e-10,
      "loss": 0.5983,
      "step": 9892
    },
    {
      "epoch": 0.9779798828559425,
      "grad_norm": 5.302610464484311,
      "learning_rate": 2.5256097924547524e-10,
      "loss": 0.7132,
      "step": 9893
    },
    {
      "epoch": 0.9780787386006969,
      "grad_norm": 5.380830364862367,
      "learning_rate": 2.502917272088734e-10,
      "loss": 0.6642,
      "step": 9894
    },
    {
      "epoch": 0.9781775943454514,
      "grad_norm": 3.0368158527641045,
      "learning_rate": 2.480327030334295e-10,
      "loss": 0.6186,
      "step": 9895
    },
    {
      "epoch": 0.9782764500902059,
      "grad_norm": 4.9952298386964795,
      "learning_rate": 2.4578390695079166e-10,
      "loss": 0.7029,
      "step": 9896
    },
    {
      "epoch": 0.9783753058349604,
      "grad_norm": 4.742092684310476,
      "learning_rate": 2.435453391915421e-10,
      "loss": 0.7646,
      "step": 9897
    },
    {
      "epoch": 0.9784741615797148,
      "grad_norm": 5.886400038443044,
      "learning_rate": 2.413169999852083e-10,
      "loss": 0.7026,
      "step": 9898
    },
    {
      "epoch": 0.9785730173244692,
      "grad_norm": 4.797896750842494,
      "learning_rate": 2.3909888956026303e-10,
      "loss": 0.679,
      "step": 9899
    },
    {
      "epoch": 0.9786718730692238,
      "grad_norm": 3.348508879087131,
      "learning_rate": 2.368910081441466e-10,
      "loss": 0.6838,
      "step": 9900
    },
    {
      "epoch": 0.9787707288139782,
      "grad_norm": 4.609385267644916,
      "learning_rate": 2.346933559632558e-10,
      "loss": 0.7948,
      "step": 9901
    },
    {
      "epoch": 0.9788695845587326,
      "grad_norm": 4.118060742094916,
      "learning_rate": 2.3250593324292134e-10,
      "loss": 0.7459,
      "step": 9902
    },
    {
      "epoch": 0.9789684403034872,
      "grad_norm": 3.017253387567331,
      "learning_rate": 2.3032874020741943e-10,
      "loss": 0.6872,
      "step": 9903
    },
    {
      "epoch": 0.9790672960482416,
      "grad_norm": 3.6385323344191933,
      "learning_rate": 2.2816177708000483e-10,
      "loss": 0.6439,
      "step": 9904
    },
    {
      "epoch": 0.979166151792996,
      "grad_norm": 12.847527852372831,
      "learning_rate": 2.2600504408286647e-10,
      "loss": 0.7226,
      "step": 9905
    },
    {
      "epoch": 0.9792650075377506,
      "grad_norm": 3.5140447449746777,
      "learning_rate": 2.2385854143713856e-10,
      "loss": 0.7603,
      "step": 9906
    },
    {
      "epoch": 0.979363863282505,
      "grad_norm": 40.77871241287877,
      "learning_rate": 2.217222693629117e-10,
      "loss": 0.7148,
      "step": 9907
    },
    {
      "epoch": 0.9794627190272595,
      "grad_norm": 29.265393620751098,
      "learning_rate": 2.1959622807923296e-10,
      "loss": 0.6797,
      "step": 9908
    },
    {
      "epoch": 0.9795615747720139,
      "grad_norm": 4.524540230881037,
      "learning_rate": 2.1748041780409453e-10,
      "loss": 0.6265,
      "step": 9909
    },
    {
      "epoch": 0.9796604305167684,
      "grad_norm": 3.708832330757567,
      "learning_rate": 2.1537483875445628e-10,
      "loss": 0.6664,
      "step": 9910
    },
    {
      "epoch": 0.9797592862615229,
      "grad_norm": 3.8988257818651904,
      "learning_rate": 2.1327949114618994e-10,
      "loss": 0.7057,
      "step": 9911
    },
    {
      "epoch": 0.9798581420062773,
      "grad_norm": 3.321139168944259,
      "learning_rate": 2.1119437519415694e-10,
      "loss": 0.6951,
      "step": 9912
    },
    {
      "epoch": 0.9799569977510318,
      "grad_norm": 4.609928489862205,
      "learning_rate": 2.09119491112153e-10,
      "loss": 0.6407,
      "step": 9913
    },
    {
      "epoch": 0.9800558534957863,
      "grad_norm": 5.1431567885094065,
      "learning_rate": 2.070548391129301e-10,
      "loss": 0.678,
      "step": 9914
    },
    {
      "epoch": 0.9801547092405407,
      "grad_norm": 16.883590195279165,
      "learning_rate": 2.0500041940817448e-10,
      "loss": 0.7148,
      "step": 9915
    },
    {
      "epoch": 0.9802535649852953,
      "grad_norm": 3.552474609872056,
      "learning_rate": 2.0295623220855096e-10,
      "loss": 0.6009,
      "step": 9916
    },
    {
      "epoch": 0.9803524207300497,
      "grad_norm": 3.2013360023016935,
      "learning_rate": 2.0092227772365855e-10,
      "loss": 0.7687,
      "step": 9917
    },
    {
      "epoch": 0.9804512764748041,
      "grad_norm": 4.368264705397335,
      "learning_rate": 1.9889855616204155e-10,
      "loss": 0.7012,
      "step": 9918
    },
    {
      "epoch": 0.9805501322195586,
      "grad_norm": 3.20721544737697,
      "learning_rate": 1.9688506773121172e-10,
      "loss": 0.7571,
      "step": 9919
    },
    {
      "epoch": 0.9806489879643131,
      "grad_norm": 3.917994691789507,
      "learning_rate": 1.9488181263761505e-10,
      "loss": 0.7051,
      "step": 9920
    },
    {
      "epoch": 0.9807478437090675,
      "grad_norm": 4.62219505336901,
      "learning_rate": 1.92888791086665e-10,
      "loss": 0.665,
      "step": 9921
    },
    {
      "epoch": 0.980846699453822,
      "grad_norm": 3.5945281060062126,
      "learning_rate": 1.9090600328269811e-10,
      "loss": 0.6604,
      "step": 9922
    },
    {
      "epoch": 0.9809455551985765,
      "grad_norm": 3.5046324714024997,
      "learning_rate": 1.8893344942904065e-10,
      "loss": 0.7153,
      "step": 9923
    },
    {
      "epoch": 0.981044410943331,
      "grad_norm": 4.4433401946259234,
      "learning_rate": 1.8697112972795304e-10,
      "loss": 0.7825,
      "step": 9924
    },
    {
      "epoch": 0.9811432666880854,
      "grad_norm": 5.968315767392347,
      "learning_rate": 1.8501904438061877e-10,
      "loss": 0.699,
      "step": 9925
    },
    {
      "epoch": 0.9812421224328399,
      "grad_norm": 2.9966015408695634,
      "learning_rate": 1.8307719358719999e-10,
      "loss": 0.5807,
      "step": 9926
    },
    {
      "epoch": 0.9813409781775944,
      "grad_norm": 4.1649764387096955,
      "learning_rate": 1.8114557754682624e-10,
      "loss": 0.6301,
      "step": 9927
    },
    {
      "epoch": 0.9814398339223488,
      "grad_norm": 4.373360869171952,
      "learning_rate": 1.7922419645752807e-10,
      "loss": 0.6505,
      "step": 9928
    },
    {
      "epoch": 0.9815386896671032,
      "grad_norm": 4.449797143040474,
      "learning_rate": 1.773130505163367e-10,
      "loss": 0.7694,
      "step": 9929
    },
    {
      "epoch": 0.9816375454118578,
      "grad_norm": 4.832628783887782,
      "learning_rate": 1.7541213991920657e-10,
      "loss": 0.6792,
      "step": 9930
    },
    {
      "epoch": 0.9817364011566122,
      "grad_norm": 3.4562825537883204,
      "learning_rate": 1.7352146486103725e-10,
      "loss": 0.6455,
      "step": 9931
    },
    {
      "epoch": 0.9818352569013666,
      "grad_norm": 6.4940467465641385,
      "learning_rate": 1.7164102553570704e-10,
      "loss": 0.6946,
      "step": 9932
    },
    {
      "epoch": 0.9819341126461212,
      "grad_norm": 3.118820495656103,
      "learning_rate": 1.6977082213601723e-10,
      "loss": 0.7063,
      "step": 9933
    },
    {
      "epoch": 0.9820329683908756,
      "grad_norm": 3.0809508490163076,
      "learning_rate": 1.6791085485371448e-10,
      "loss": 0.689,
      "step": 9934
    },
    {
      "epoch": 0.98213182413563,
      "grad_norm": 2.707978513673834,
      "learning_rate": 1.660611238795462e-10,
      "loss": 0.5611,
      "step": 9935
    },
    {
      "epoch": 0.9822306798803846,
      "grad_norm": 2.8474787613651498,
      "learning_rate": 1.6422162940313845e-10,
      "loss": 0.754,
      "step": 9936
    },
    {
      "epoch": 0.982329535625139,
      "grad_norm": 4.952379964250739,
      "learning_rate": 1.6239237161312925e-10,
      "loss": 0.6998,
      "step": 9937
    },
    {
      "epoch": 0.9824283913698935,
      "grad_norm": 6.069667226078943,
      "learning_rate": 1.6057335069707966e-10,
      "loss": 0.6542,
      "step": 9938
    },
    {
      "epoch": 0.9825272471146479,
      "grad_norm": 5.39797377651625,
      "learning_rate": 1.5876456684149607e-10,
      "loss": 0.7748,
      "step": 9939
    },
    {
      "epoch": 0.9826261028594024,
      "grad_norm": 8.567652336502583,
      "learning_rate": 1.5696602023183014e-10,
      "loss": 0.6973,
      "step": 9940
    },
    {
      "epoch": 0.9827249586041569,
      "grad_norm": 4.129489446355094,
      "learning_rate": 1.551777110525232e-10,
      "loss": 0.844,
      "step": 9941
    },
    {
      "epoch": 0.9828238143489113,
      "grad_norm": 3.6931308247752823,
      "learning_rate": 1.5339963948692857e-10,
      "loss": 0.6902,
      "step": 9942
    },
    {
      "epoch": 0.9829226700936659,
      "grad_norm": 16.615405599373148,
      "learning_rate": 1.51631805717356e-10,
      "loss": 0.6141,
      "step": 9943
    },
    {
      "epoch": 0.9830215258384203,
      "grad_norm": 4.009538573329675,
      "learning_rate": 1.498742099250716e-10,
      "loss": 0.6958,
      "step": 9944
    },
    {
      "epoch": 0.9831203815831747,
      "grad_norm": 18.807016268334888,
      "learning_rate": 1.4812685229029787e-10,
      "loss": 0.6796,
      "step": 9945
    },
    {
      "epoch": 0.9832192373279293,
      "grad_norm": 4.706884180195693,
      "learning_rate": 1.4638973299220257e-10,
      "loss": 0.6463,
      "step": 9946
    },
    {
      "epoch": 0.9833180930726837,
      "grad_norm": 3.858215772783094,
      "learning_rate": 1.4466285220888775e-10,
      "loss": 0.7486,
      "step": 9947
    },
    {
      "epoch": 0.9834169488174381,
      "grad_norm": 18.819162273409713,
      "learning_rate": 1.4294621011742282e-10,
      "loss": 0.6732,
      "step": 9948
    },
    {
      "epoch": 0.9835158045621927,
      "grad_norm": 8.341471659853797,
      "learning_rate": 1.4123980689383364e-10,
      "loss": 0.7147,
      "step": 9949
    },
    {
      "epoch": 0.9836146603069471,
      "grad_norm": 5.144863771086624,
      "learning_rate": 1.3954364271309138e-10,
      "loss": 0.7498,
      "step": 9950
    },
    {
      "epoch": 0.9837135160517015,
      "grad_norm": 6.454512955081075,
      "learning_rate": 1.3785771774909026e-10,
      "loss": 0.8006,
      "step": 9951
    },
    {
      "epoch": 0.983812371796456,
      "grad_norm": 4.9022639051854435,
      "learning_rate": 1.3618203217471424e-10,
      "loss": 0.5668,
      "step": 9952
    },
    {
      "epoch": 0.9839112275412105,
      "grad_norm": 3.801050549529694,
      "learning_rate": 1.345165861617814e-10,
      "loss": 0.6802,
      "step": 9953
    },
    {
      "epoch": 0.984010083285965,
      "grad_norm": 4.90729300365936,
      "learning_rate": 1.3286137988104408e-10,
      "loss": 0.7377,
      "step": 9954
    },
    {
      "epoch": 0.9841089390307194,
      "grad_norm": 10.32943859126779,
      "learning_rate": 1.312164135022331e-10,
      "loss": 0.6448,
      "step": 9955
    },
    {
      "epoch": 0.9842077947754739,
      "grad_norm": 20.4578840912686,
      "learning_rate": 1.2958168719401364e-10,
      "loss": 0.6545,
      "step": 9956
    },
    {
      "epoch": 0.9843066505202284,
      "grad_norm": 5.851436624348358,
      "learning_rate": 1.27957201123996e-10,
      "loss": 0.6392,
      "step": 9957
    },
    {
      "epoch": 0.9844055062649828,
      "grad_norm": 4.133461163069233,
      "learning_rate": 1.2634295545874696e-10,
      "loss": 0.736,
      "step": 9958
    },
    {
      "epoch": 0.9845043620097373,
      "grad_norm": 27.950272901779933,
      "learning_rate": 1.2473895036378967e-10,
      "loss": 0.7065,
      "step": 9959
    },
    {
      "epoch": 0.9846032177544918,
      "grad_norm": 3.8626218321452694,
      "learning_rate": 1.2314518600359258e-10,
      "loss": 0.6702,
      "step": 9960
    },
    {
      "epoch": 0.9847020734992462,
      "grad_norm": 4.084255439715281,
      "learning_rate": 1.215616625415694e-10,
      "loss": 0.7354,
      "step": 9961
    },
    {
      "epoch": 0.9848009292440006,
      "grad_norm": 3.803808906054133,
      "learning_rate": 1.1998838014009026e-10,
      "loss": 0.6672,
      "step": 9962
    },
    {
      "epoch": 0.9848997849887552,
      "grad_norm": 4.534919876011162,
      "learning_rate": 1.1842533896045948e-10,
      "loss": 0.5804,
      "step": 9963
    },
    {
      "epoch": 0.9849986407335096,
      "grad_norm": 4.409387728588799,
      "learning_rate": 1.168725391629599e-10,
      "loss": 0.6469,
      "step": 9964
    },
    {
      "epoch": 0.9850974964782641,
      "grad_norm": 4.047139214508962,
      "learning_rate": 1.1532998090679758e-10,
      "loss": 0.7143,
      "step": 9965
    },
    {
      "epoch": 0.9851963522230186,
      "grad_norm": 4.1717455182270635,
      "learning_rate": 1.1379766435014593e-10,
      "loss": 0.6903,
      "step": 9966
    },
    {
      "epoch": 0.985295207967773,
      "grad_norm": 3.1196441973007065,
      "learning_rate": 1.1227558965012373e-10,
      "loss": 0.7123,
      "step": 9967
    },
    {
      "epoch": 0.9853940637125275,
      "grad_norm": 6.041824512948329,
      "learning_rate": 1.1076375696278395e-10,
      "loss": 0.6363,
      "step": 9968
    },
    {
      "epoch": 0.985492919457282,
      "grad_norm": 5.7478154367933785,
      "learning_rate": 1.0926216644315811e-10,
      "loss": 0.6774,
      "step": 9969
    },
    {
      "epoch": 0.9855917752020364,
      "grad_norm": 3.9755313559566954,
      "learning_rate": 1.0777081824520084e-10,
      "loss": 0.6642,
      "step": 9970
    },
    {
      "epoch": 0.9856906309467909,
      "grad_norm": 3.6597063824236042,
      "learning_rate": 1.0628971252183427e-10,
      "loss": 0.7258,
      "step": 9971
    },
    {
      "epoch": 0.9857894866915453,
      "grad_norm": 4.409509577091096,
      "learning_rate": 1.048188494249147e-10,
      "loss": 0.7138,
      "step": 9972
    },
    {
      "epoch": 0.9858883424362999,
      "grad_norm": 2.714569549268208,
      "learning_rate": 1.0335822910526593e-10,
      "loss": 0.6178,
      "step": 9973
    },
    {
      "epoch": 0.9859871981810543,
      "grad_norm": 4.201549001128713,
      "learning_rate": 1.0190785171265704e-10,
      "loss": 0.6367,
      "step": 9974
    },
    {
      "epoch": 0.9860860539258087,
      "grad_norm": 4.91811784438016,
      "learning_rate": 1.004677173958024e-10,
      "loss": 0.6393,
      "step": 9975
    },
    {
      "epoch": 0.9861849096705633,
      "grad_norm": 4.810910429479335,
      "learning_rate": 9.903782630235058e-11,
      "loss": 0.6492,
      "step": 9976
    },
    {
      "epoch": 0.9862837654153177,
      "grad_norm": 4.2504483467825525,
      "learning_rate": 9.761817857892873e-11,
      "loss": 0.6977,
      "step": 9977
    },
    {
      "epoch": 0.9863826211600721,
      "grad_norm": 4.2561207971180375,
      "learning_rate": 9.620877437110931e-11,
      "loss": 0.7868,
      "step": 9978
    },
    {
      "epoch": 0.9864814769048267,
      "grad_norm": 4.426917421579262,
      "learning_rate": 9.480961382338782e-11,
      "loss": 0.695,
      "step": 9979
    },
    {
      "epoch": 0.9865803326495811,
      "grad_norm": 5.324591250320784,
      "learning_rate": 9.342069707922729e-11,
      "loss": 0.636,
      "step": 9980
    },
    {
      "epoch": 0.9866791883943355,
      "grad_norm": 4.289974497887985,
      "learning_rate": 9.204202428105823e-11,
      "loss": 0.7109,
      "step": 9981
    },
    {
      "epoch": 0.98677804413909,
      "grad_norm": 4.1888420754500615,
      "learning_rate": 9.067359557023424e-11,
      "loss": 0.7514,
      "step": 9982
    },
    {
      "epoch": 0.9868768998838445,
      "grad_norm": 3.892203932230894,
      "learning_rate": 8.931541108705421e-11,
      "loss": 0.7324,
      "step": 9983
    },
    {
      "epoch": 0.986975755628599,
      "grad_norm": 5.3225450905975595,
      "learning_rate": 8.79674709708067e-11,
      "loss": 0.8322,
      "step": 9984
    },
    {
      "epoch": 0.9870746113733534,
      "grad_norm": 4.09339809361668,
      "learning_rate": 8.66297753596923e-11,
      "loss": 0.6508,
      "step": 9985
    },
    {
      "epoch": 0.9871734671181079,
      "grad_norm": 4.353615897795182,
      "learning_rate": 8.530232439085683e-11,
      "loss": 0.7173,
      "step": 9986
    },
    {
      "epoch": 0.9872723228628624,
      "grad_norm": 4.877910742704973,
      "learning_rate": 8.398511820042475e-11,
      "loss": 0.6991,
      "step": 9987
    },
    {
      "epoch": 0.9873711786076168,
      "grad_norm": 3.6874270013839983,
      "learning_rate": 8.26781569234547e-11,
      "loss": 0.6511,
      "step": 9988
    },
    {
      "epoch": 0.9874700343523714,
      "grad_norm": 5.122228699353762,
      "learning_rate": 8.138144069395058e-11,
      "loss": 0.7521,
      "step": 9989
    },
    {
      "epoch": 0.9875688900971258,
      "grad_norm": 3.9592800079501003,
      "learning_rate": 8.009496964488382e-11,
      "loss": 0.6783,
      "step": 9990
    },
    {
      "epoch": 0.9876677458418802,
      "grad_norm": 8.847261761070852,
      "learning_rate": 7.881874390816001e-11,
      "loss": 0.6572,
      "step": 9991
    },
    {
      "epoch": 0.9877666015866347,
      "grad_norm": 7.203053987586899,
      "learning_rate": 7.755276361461893e-11,
      "loss": 0.7016,
      "step": 9992
    },
    {
      "epoch": 0.9878654573313892,
      "grad_norm": 3.08811316611876,
      "learning_rate": 7.629702889409006e-11,
      "loss": 0.7255,
      "step": 9993
    },
    {
      "epoch": 0.9879643130761436,
      "grad_norm": 4.080404630571386,
      "learning_rate": 7.505153987532598e-11,
      "loss": 0.6619,
      "step": 9994
    },
    {
      "epoch": 0.9880631688208981,
      "grad_norm": 4.251324219728391,
      "learning_rate": 7.381629668602452e-11,
      "loss": 0.6056,
      "step": 9995
    },
    {
      "epoch": 0.9881620245656526,
      "grad_norm": 4.383310435659365,
      "learning_rate": 7.259129945285102e-11,
      "loss": 0.8156,
      "step": 9996
    },
    {
      "epoch": 0.988260880310407,
      "grad_norm": 3.8436919945017998,
      "learning_rate": 7.137654830140505e-11,
      "loss": 0.699,
      "step": 9997
    },
    {
      "epoch": 0.9883597360551615,
      "grad_norm": 5.947776600003691,
      "learning_rate": 7.017204335623139e-11,
      "loss": 0.6669,
      "step": 9998
    },
    {
      "epoch": 0.988458591799916,
      "grad_norm": 6.194668081288551,
      "learning_rate": 6.897778474086458e-11,
      "loss": 0.7522,
      "step": 9999
    },
    {
      "epoch": 0.9885574475446705,
      "grad_norm": 3.7058306150967733,
      "learning_rate": 6.779377257772889e-11,
      "loss": 0.6414,
      "step": 10000
    },
    {
      "epoch": 0.9886563032894249,
      "grad_norm": 8.590374700074914,
      "learning_rate": 6.662000698823833e-11,
      "loss": 0.732,
      "step": 10001
    },
    {
      "epoch": 0.9887551590341793,
      "grad_norm": 3.1665087349288323,
      "learning_rate": 6.545648809275218e-11,
      "loss": 0.7691,
      "step": 10002
    },
    {
      "epoch": 0.9888540147789339,
      "grad_norm": 6.200114195771567,
      "learning_rate": 6.430321601056388e-11,
      "loss": 0.7826,
      "step": 10003
    },
    {
      "epoch": 0.9889528705236883,
      "grad_norm": 6.266211800291764,
      "learning_rate": 6.31601908599233e-11,
      "loss": 0.656,
      "step": 10004
    },
    {
      "epoch": 0.9890517262684427,
      "grad_norm": 4.1992588520632905,
      "learning_rate": 6.202741275802558e-11,
      "loss": 0.8186,
      "step": 10005
    },
    {
      "epoch": 0.9891505820131973,
      "grad_norm": 8.553364970265939,
      "learning_rate": 6.090488182103337e-11,
      "loss": 0.7621,
      "step": 10006
    },
    {
      "epoch": 0.9892494377579517,
      "grad_norm": 18.116611480455088,
      "learning_rate": 5.979259816404347e-11,
      "loss": 0.7442,
      "step": 10007
    },
    {
      "epoch": 0.9893482935027061,
      "grad_norm": 3.6112898399444835,
      "learning_rate": 5.8690561901098e-11,
      "loss": 0.7449,
      "step": 10008
    },
    {
      "epoch": 0.9894471492474607,
      "grad_norm": 3.309982523423621,
      "learning_rate": 5.7598773145195455e-11,
      "loss": 0.6635,
      "step": 10009
    },
    {
      "epoch": 0.9895460049922151,
      "grad_norm": 3.7701211608509007,
      "learning_rate": 5.651723200829073e-11,
      "loss": 0.687,
      "step": 10010
    },
    {
      "epoch": 0.9896448607369696,
      "grad_norm": 8.601007196658097,
      "learning_rate": 5.544593860126179e-11,
      "loss": 0.7536,
      "step": 10011
    },
    {
      "epoch": 0.989743716481724,
      "grad_norm": 3.8821972748491054,
      "learning_rate": 5.438489303397631e-11,
      "loss": 0.6934,
      "step": 10012
    },
    {
      "epoch": 0.9898425722264785,
      "grad_norm": 3.4428153403931008,
      "learning_rate": 5.333409541521394e-11,
      "loss": 0.8656,
      "step": 10013
    },
    {
      "epoch": 0.989941427971233,
      "grad_norm": 5.096028362861823,
      "learning_rate": 5.229354585272183e-11,
      "loss": 0.7541,
      "step": 10014
    },
    {
      "epoch": 0.9900402837159874,
      "grad_norm": 2.610724580438784,
      "learning_rate": 5.126324445320351e-11,
      "loss": 0.6679,
      "step": 10015
    },
    {
      "epoch": 0.990139139460742,
      "grad_norm": 3.8456089108832354,
      "learning_rate": 5.0243191322285606e-11,
      "loss": 0.7534,
      "step": 10016
    },
    {
      "epoch": 0.9902379952054964,
      "grad_norm": 5.777160598567235,
      "learning_rate": 4.9233386564562217e-11,
      "loss": 0.6506,
      "step": 10017
    },
    {
      "epoch": 0.9903368509502508,
      "grad_norm": 4.076652240473739,
      "learning_rate": 4.823383028358385e-11,
      "loss": 0.682,
      "step": 10018
    },
    {
      "epoch": 0.9904357066950054,
      "grad_norm": 13.07678329562823,
      "learning_rate": 4.72445225818352e-11,
      "loss": 0.672,
      "step": 10019
    },
    {
      "epoch": 0.9905345624397598,
      "grad_norm": 6.286169184971684,
      "learning_rate": 4.626546356074623e-11,
      "loss": 0.6964,
      "step": 10020
    },
    {
      "epoch": 0.9906334181845142,
      "grad_norm": 33.62452340618132,
      "learning_rate": 4.529665332071442e-11,
      "loss": 0.6862,
      "step": 10021
    },
    {
      "epoch": 0.9907322739292688,
      "grad_norm": 3.2429492739181898,
      "learning_rate": 4.433809196108251e-11,
      "loss": 0.7006,
      "step": 10022
    },
    {
      "epoch": 0.9908311296740232,
      "grad_norm": 4.955199825057955,
      "learning_rate": 4.338977958011636e-11,
      "loss": 0.6715,
      "step": 10023
    },
    {
      "epoch": 0.9909299854187776,
      "grad_norm": 9.725605010256109,
      "learning_rate": 4.245171627507149e-11,
      "loss": 0.6936,
      "step": 10024
    },
    {
      "epoch": 0.9910288411635321,
      "grad_norm": 3.1505703113627836,
      "learning_rate": 4.1523902142126534e-11,
      "loss": 0.6118,
      "step": 10025
    },
    {
      "epoch": 0.9911276969082866,
      "grad_norm": 3.822181808685845,
      "learning_rate": 4.060633727641649e-11,
      "loss": 0.7161,
      "step": 10026
    },
    {
      "epoch": 0.991226552653041,
      "grad_norm": 4.049983523293202,
      "learning_rate": 3.969902177202167e-11,
      "loss": 0.6425,
      "step": 10027
    },
    {
      "epoch": 0.9913254083977955,
      "grad_norm": 5.796519034576253,
      "learning_rate": 3.880195572196765e-11,
      "loss": 0.7112,
      "step": 10028
    },
    {
      "epoch": 0.99142426414255,
      "grad_norm": 3.7939990623302458,
      "learning_rate": 3.7915139218247516e-11,
      "loss": 0.7745,
      "step": 10029
    },
    {
      "epoch": 0.9915231198873045,
      "grad_norm": 3.9943597395753736,
      "learning_rate": 3.703857235178853e-11,
      "loss": 0.6666,
      "step": 10030
    },
    {
      "epoch": 0.9916219756320589,
      "grad_norm": 6.259165740555087,
      "learning_rate": 3.617225521245215e-11,
      "loss": 0.6515,
      "step": 10031
    },
    {
      "epoch": 0.9917208313768134,
      "grad_norm": 7.834823501481987,
      "learning_rate": 3.531618788908952e-11,
      "loss": 0.7266,
      "step": 10032
    },
    {
      "epoch": 0.9918196871215679,
      "grad_norm": 5.04823730878093,
      "learning_rate": 3.447037046947487e-11,
      "loss": 0.6575,
      "step": 10033
    },
    {
      "epoch": 0.9919185428663223,
      "grad_norm": 3.881578798452637,
      "learning_rate": 3.363480304032773e-11,
      "loss": 0.7325,
      "step": 10034
    },
    {
      "epoch": 0.9920173986110767,
      "grad_norm": 6.95826052971369,
      "learning_rate": 3.2809485687312896e-11,
      "loss": 0.7633,
      "step": 10035
    },
    {
      "epoch": 0.9921162543558313,
      "grad_norm": 3.670076387614518,
      "learning_rate": 3.199441849507378e-11,
      "loss": 0.7828,
      "step": 10036
    },
    {
      "epoch": 0.9922151101005857,
      "grad_norm": 6.925535944950219,
      "learning_rate": 3.118960154717687e-11,
      "loss": 0.7452,
      "step": 10037
    },
    {
      "epoch": 0.9923139658453402,
      "grad_norm": 3.9562502499797585,
      "learning_rate": 3.039503492613393e-11,
      "loss": 0.6999,
      "step": 10038
    },
    {
      "epoch": 0.9924128215900947,
      "grad_norm": 3.0754543441414866,
      "learning_rate": 2.961071871342424e-11,
      "loss": 0.7571,
      "step": 10039
    },
    {
      "epoch": 0.9925116773348491,
      "grad_norm": 6.442723774692946,
      "learning_rate": 2.8836652989483456e-11,
      "loss": 0.7953,
      "step": 10040
    },
    {
      "epoch": 0.9926105330796036,
      "grad_norm": 3.595260920921651,
      "learning_rate": 2.807283783364811e-11,
      "loss": 0.7118,
      "step": 10041
    },
    {
      "epoch": 0.9927093888243581,
      "grad_norm": 3.653731524595311,
      "learning_rate": 2.7319273324266645e-11,
      "loss": 0.6934,
      "step": 10042
    },
    {
      "epoch": 0.9928082445691125,
      "grad_norm": 4.914944002414366,
      "learning_rate": 2.6575959538588377e-11,
      "loss": 0.7095,
      "step": 10043
    },
    {
      "epoch": 0.992907100313867,
      "grad_norm": 4.652113137789867,
      "learning_rate": 2.584289655283012e-11,
      "loss": 0.7061,
      "step": 10044
    },
    {
      "epoch": 0.9930059560586214,
      "grad_norm": 4.246194880014553,
      "learning_rate": 2.512008444216507e-11,
      "loss": 0.723,
      "step": 10045
    },
    {
      "epoch": 0.993104811803376,
      "grad_norm": 3.662838257210592,
      "learning_rate": 2.4407523280700614e-11,
      "loss": 0.7043,
      "step": 10046
    },
    {
      "epoch": 0.9932036675481304,
      "grad_norm": 28.292698575198894,
      "learning_rate": 2.3705213141500534e-11,
      "loss": 0.7052,
      "step": 10047
    },
    {
      "epoch": 0.9933025232928848,
      "grad_norm": 6.059770002362138,
      "learning_rate": 2.3013154096573894e-11,
      "loss": 0.6367,
      "step": 10048
    },
    {
      "epoch": 0.9934013790376394,
      "grad_norm": 4.661791045854803,
      "learning_rate": 2.233134621688615e-11,
      "loss": 0.7564,
      "step": 10049
    },
    {
      "epoch": 0.9935002347823938,
      "grad_norm": 5.262218364186442,
      "learning_rate": 2.165978957234804e-11,
      "loss": 0.6881,
      "step": 10050
    },
    {
      "epoch": 0.9935990905271482,
      "grad_norm": 4.525379456955892,
      "learning_rate": 2.0998484231804502e-11,
      "loss": 0.7056,
      "step": 10051
    },
    {
      "epoch": 0.9936979462719028,
      "grad_norm": 3.3615027974112497,
      "learning_rate": 2.0347430263079058e-11,
      "loss": 0.6282,
      "step": 10052
    },
    {
      "epoch": 0.9937968020166572,
      "grad_norm": 4.6494461083073375,
      "learning_rate": 1.9706627732918312e-11,
      "loss": 0.768,
      "step": 10053
    },
    {
      "epoch": 0.9938956577614116,
      "grad_norm": 3.4094258522168444,
      "learning_rate": 1.907607670703637e-11,
      "loss": 0.764,
      "step": 10054
    },
    {
      "epoch": 0.9939945135061661,
      "grad_norm": 6.886558040421821,
      "learning_rate": 1.8455777250070414e-11,
      "loss": 0.7426,
      "step": 10055
    },
    {
      "epoch": 0.9940933692509206,
      "grad_norm": 4.9332961533608755,
      "learning_rate": 1.7845729425647328e-11,
      "loss": 0.6931,
      "step": 10056
    },
    {
      "epoch": 0.9941922249956751,
      "grad_norm": 3.865952656555199,
      "learning_rate": 1.7245933296294867e-11,
      "loss": 0.7055,
      "step": 10057
    },
    {
      "epoch": 0.9942910807404295,
      "grad_norm": 4.175999167513154,
      "learning_rate": 1.6656388923519394e-11,
      "loss": 0.7318,
      "step": 10058
    },
    {
      "epoch": 0.994389936485184,
      "grad_norm": 4.8090703829565635,
      "learning_rate": 1.607709636778365e-11,
      "loss": 0.6848,
      "step": 10059
    },
    {
      "epoch": 0.9944887922299385,
      "grad_norm": 3.7310018250751,
      "learning_rate": 1.5508055688462364e-11,
      "loss": 0.7862,
      "step": 10060
    },
    {
      "epoch": 0.9945876479746929,
      "grad_norm": 5.503969161740724,
      "learning_rate": 1.4949266943919957e-11,
      "loss": 0.5905,
      "step": 10061
    },
    {
      "epoch": 0.9946865037194474,
      "grad_norm": 3.570452962276279,
      "learning_rate": 1.4400730191443944e-11,
      "loss": 0.7825,
      "step": 10062
    },
    {
      "epoch": 0.9947853594642019,
      "grad_norm": 3.5266760251606604,
      "learning_rate": 1.386244548728932e-11,
      "loss": 0.7146,
      "step": 10063
    },
    {
      "epoch": 0.9948842152089563,
      "grad_norm": 3.881392244544323,
      "learning_rate": 1.3334412886634172e-11,
      "loss": 0.6663,
      "step": 10064
    },
    {
      "epoch": 0.9949830709537107,
      "grad_norm": 19.405826166838725,
      "learning_rate": 1.2816632443635178e-11,
      "loss": 0.792,
      "step": 10065
    },
    {
      "epoch": 0.9950819266984653,
      "grad_norm": 5.983361769928404,
      "learning_rate": 1.23091042113721e-11,
      "loss": 0.6076,
      "step": 10066
    },
    {
      "epoch": 0.9951807824432197,
      "grad_norm": 5.24729335323364,
      "learning_rate": 1.1811828241892197e-11,
      "loss": 0.647,
      "step": 10067
    },
    {
      "epoch": 0.9952796381879742,
      "grad_norm": 4.179667870302554,
      "learning_rate": 1.1324804586176906e-11,
      "loss": 0.6401,
      "step": 10068
    },
    {
      "epoch": 0.9953784939327287,
      "grad_norm": 4.452015655357441,
      "learning_rate": 1.084803329416406e-11,
      "loss": 0.7658,
      "step": 10069
    },
    {
      "epoch": 0.9954773496774831,
      "grad_norm": 5.543867896527644,
      "learning_rate": 1.038151441473678e-11,
      "loss": 0.8326,
      "step": 10070
    },
    {
      "epoch": 0.9955762054222376,
      "grad_norm": 4.445258506951427,
      "learning_rate": 9.925247995745678e-12,
      "loss": 0.6965,
      "step": 10071
    },
    {
      "epoch": 0.9956750611669921,
      "grad_norm": 2.798483195999004,
      "learning_rate": 9.479234083953347e-12,
      "loss": 0.7135,
      "step": 10072
    },
    {
      "epoch": 0.9957739169117465,
      "grad_norm": 7.796110985897896,
      "learning_rate": 9.043472725112077e-12,
      "loss": 0.6553,
      "step": 10073
    },
    {
      "epoch": 0.995872772656501,
      "grad_norm": 4.024172767242262,
      "learning_rate": 8.617963963886143e-12,
      "loss": 0.6722,
      "step": 10074
    },
    {
      "epoch": 0.9959716284012554,
      "grad_norm": 4.213195866364077,
      "learning_rate": 8.202707843907308e-12,
      "loss": 0.7463,
      "step": 10075
    },
    {
      "epoch": 0.99607048414601,
      "grad_norm": 3.32949548312114,
      "learning_rate": 7.797704407763728e-12,
      "loss": 0.6087,
      "step": 10076
    },
    {
      "epoch": 0.9961693398907644,
      "grad_norm": 3.4689819195532703,
      "learning_rate": 7.402953696977743e-12,
      "loss": 0.6622,
      "step": 10077
    },
    {
      "epoch": 0.9962681956355188,
      "grad_norm": 2.941661083220348,
      "learning_rate": 7.018455752016983e-12,
      "loss": 0.6532,
      "step": 10078
    },
    {
      "epoch": 0.9963670513802734,
      "grad_norm": 2.9509758907018853,
      "learning_rate": 6.64421061231657e-12,
      "loss": 0.6302,
      "step": 10079
    },
    {
      "epoch": 0.9964659071250278,
      "grad_norm": 2.8195316907000225,
      "learning_rate": 6.280218316245811e-12,
      "loss": 0.7757,
      "step": 10080
    },
    {
      "epoch": 0.9965647628697822,
      "grad_norm": 5.756266518651886,
      "learning_rate": 5.926478901130405e-12,
      "loss": 0.6931,
      "step": 10081
    },
    {
      "epoch": 0.9966636186145368,
      "grad_norm": 3.8854233089112142,
      "learning_rate": 5.5829924032413375e-12,
      "loss": 0.7207,
      "step": 10082
    },
    {
      "epoch": 0.9967624743592912,
      "grad_norm": 3.7524955750325533,
      "learning_rate": 5.249758857794883e-12,
      "loss": 0.6798,
      "step": 10083
    },
    {
      "epoch": 0.9968613301040457,
      "grad_norm": 6.772098783919843,
      "learning_rate": 4.926778298952605e-12,
      "loss": 0.7572,
      "step": 10084
    },
    {
      "epoch": 0.9969601858488001,
      "grad_norm": 5.271145250423485,
      "learning_rate": 4.614050759843557e-12,
      "loss": 0.7143,
      "step": 10085
    },
    {
      "epoch": 0.9970590415935546,
      "grad_norm": 4.530367238002224,
      "learning_rate": 4.311576272519879e-12,
      "loss": 0.7126,
      "step": 10086
    },
    {
      "epoch": 0.9971578973383091,
      "grad_norm": 3.155896177827421,
      "learning_rate": 4.019354868012303e-12,
      "loss": 0.7533,
      "step": 10087
    },
    {
      "epoch": 0.9972567530830635,
      "grad_norm": 3.610583272152024,
      "learning_rate": 3.737386576274648e-12,
      "loss": 0.6648,
      "step": 10088
    },
    {
      "epoch": 0.997355608827818,
      "grad_norm": 5.296881977324083,
      "learning_rate": 3.4656714262171205e-12,
      "loss": 0.6955,
      "step": 10089
    },
    {
      "epoch": 0.9974544645725725,
      "grad_norm": 4.1446852602063915,
      "learning_rate": 3.2042094456952162e-12,
      "loss": 0.7763,
      "step": 10090
    },
    {
      "epoch": 0.9975533203173269,
      "grad_norm": 7.070971169380547,
      "learning_rate": 2.9530006615319236e-12,
      "loss": 0.6391,
      "step": 10091
    },
    {
      "epoch": 0.9976521760620815,
      "grad_norm": 6.875305478952838,
      "learning_rate": 2.712045099484417e-12,
      "loss": 0.8553,
      "step": 10092
    },
    {
      "epoch": 0.9977510318068359,
      "grad_norm": 3.9182128481002474,
      "learning_rate": 2.481342784244056e-12,
      "loss": 0.8486,
      "step": 10093
    },
    {
      "epoch": 0.9978498875515903,
      "grad_norm": 7.209460973752709,
      "learning_rate": 2.260893739480796e-12,
      "loss": 0.6622,
      "step": 10094
    },
    {
      "epoch": 0.9979487432963448,
      "grad_norm": 8.157860160842715,
      "learning_rate": 2.0506979877876753e-12,
      "loss": 0.7175,
      "step": 10095
    },
    {
      "epoch": 0.9980475990410993,
      "grad_norm": 4.104448229580695,
      "learning_rate": 1.850755550725225e-12,
      "loss": 0.6578,
      "step": 10096
    },
    {
      "epoch": 0.9981464547858537,
      "grad_norm": 4.552771378402265,
      "learning_rate": 1.6610664487881621e-12,
      "loss": 0.712,
      "step": 10097
    },
    {
      "epoch": 0.9982453105306082,
      "grad_norm": 4.508228463095843,
      "learning_rate": 1.4816307014275942e-12,
      "loss": 0.6569,
      "step": 10098
    },
    {
      "epoch": 0.9983441662753627,
      "grad_norm": 4.524135856727574,
      "learning_rate": 1.312448327051019e-12,
      "loss": 0.7496,
      "step": 10099
    },
    {
      "epoch": 0.9984430220201171,
      "grad_norm": 3.605315444377779,
      "learning_rate": 1.15351934300012e-12,
      "loss": 0.7303,
      "step": 10100
    },
    {
      "epoch": 0.9985418777648716,
      "grad_norm": 2.722257441947089,
      "learning_rate": 1.0048437655618692e-12,
      "loss": 0.6223,
      "step": 10101
    },
    {
      "epoch": 0.9986407335096261,
      "grad_norm": 89.37083624230522,
      "learning_rate": 8.664216099907306e-13,
      "loss": 0.6111,
      "step": 10102
    },
    {
      "epoch": 0.9987395892543806,
      "grad_norm": 4.616934768141887,
      "learning_rate": 7.382528904753549e-13,
      "loss": 0.7225,
      "step": 10103
    },
    {
      "epoch": 0.998838444999135,
      "grad_norm": 4.765481648396158,
      "learning_rate": 6.203376201607824e-13,
      "loss": 0.7538,
      "step": 10104
    },
    {
      "epoch": 0.9989373007438895,
      "grad_norm": 4.0134690215814715,
      "learning_rate": 5.126758111262397e-13,
      "loss": 0.7196,
      "step": 10105
    },
    {
      "epoch": 0.999036156488644,
      "grad_norm": 3.353449523485935,
      "learning_rate": 4.152674744295481e-13,
      "loss": 0.6381,
      "step": 10106
    },
    {
      "epoch": 0.9991350122333984,
      "grad_norm": 6.8539268312854364,
      "learning_rate": 3.2811262005161267e-13,
      "loss": 0.6561,
      "step": 10107
    },
    {
      "epoch": 0.9992338679781528,
      "grad_norm": 22.667845996164242,
      "learning_rate": 2.5121125692972867e-13,
      "loss": 0.5774,
      "step": 10108
    },
    {
      "epoch": 0.9993327237229074,
      "grad_norm": 5.17299561125041,
      "learning_rate": 1.8456339293537736e-13,
      "loss": 0.6462,
      "step": 10109
    },
    {
      "epoch": 0.9994315794676618,
      "grad_norm": 5.270354985489154,
      "learning_rate": 1.2816903491863484e-13,
      "loss": 0.7205,
      "step": 10110
    },
    {
      "epoch": 0.9995304352124162,
      "grad_norm": 5.855207465036062,
      "learning_rate": 8.202818865266082e-14,
      "loss": 0.5983,
      "step": 10111
    },
    {
      "epoch": 0.9996292909571708,
      "grad_norm": 4.619373120439512,
      "learning_rate": 4.614085887810759e-14,
      "loss": 0.6319,
      "step": 10112
    },
    {
      "epoch": 0.9997281467019252,
      "grad_norm": 4.850602544743707,
      "learning_rate": 2.0507049269813392e-14,
      "loss": 0.7138,
      "step": 10113
    },
    {
      "epoch": 0.9998270024466797,
      "grad_norm": 4.650663586725914,
      "learning_rate": 5.1267624479045536e-15,
      "loss": 0.7126,
      "step": 10114
    },
    {
      "epoch": 0.9999258581914342,
      "grad_norm": 5.069951464751154,
      "learning_rate": 0.0,
      "loss": 0.672,
      "step": 10115
    },
    {
      "epoch": 0.9999258581914342,
      "step": 10115,
      "total_flos": 4197411658629120.0,
      "train_loss": 0.7261561949301403,
      "train_runtime": 64868.5677,
      "train_samples_per_second": 19.96,
      "train_steps_per_second": 0.156
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 10115,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4197411658629120.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}